218 files changed, 36661 insertions, 89500 deletions

diff --git a/core/io/aes256.cpp b/core/io/aes256.cpp
index 69a5091f1d..e7f465dcc6 100644
--- a/core/io/aes256.cpp
+++ b/core/io/aes256.cpp
@@ -1,359 +1,359 @@
-/*  

-*   Byte-oriented AES-256 implementation.

-*   All lookup tables replaced with 'on the fly' calculations. 

-*

-*   Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com

-*   Other contributors: Hal Finney

-*

-*   Permission to use, copy, modify, and distribute this software for any

-*   purpose with or without fee is hereby granted, provided that the above

-*   copyright notice and this permission notice appear in all copies.

-*

-*   THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES

-*   WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF

-*   MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR

-*   ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES

-*   WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN

-*   ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF

-*   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

-*/

-#include "aes256.h"

-

-#define F(x)   (((x)<<1) ^ ((((x)>>7) & 1) * 0x1b))

-#define FD(x)  (((x) >> 1) ^ (((x) & 1) ? 0x8d : 0))

-

-// #define BACK_TO_TABLES

-#ifdef BACK_TO_TABLES

-

-const uint8_t sbox[256] = {

-    0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,

-    0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,

-    0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,

-    0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,

-    0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,

-    0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,

-    0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,

-    0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,

-    0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,

-    0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,

-    0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,

-    0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,

-    0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,

-    0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,

-    0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,

-    0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,

-    0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,

-    0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,

-    0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,

-    0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,

-    0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,

-    0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,

-    0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,

-    0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,

-    0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,

-    0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,

-    0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,

-    0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,

-    0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,

-    0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,

-    0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,

-    0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16

-};

-const uint8_t sboxinv[256] = {

-    0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38,

-    0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,

-    0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87,

-    0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,

-    0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d,

-    0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,

-    0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2,

-    0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,

-    0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16,

-    0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,

-    0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda,

-    0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,

-    0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a,

-    0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,

-    0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02,

-    0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,

-    0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea,

-    0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,

-    0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85,

-    0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,

-    0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89,

-    0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,

-    0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20,

-    0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,

-    0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31,

-    0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,

-    0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d,

-    0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,

-    0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0,

-    0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,

-    0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26,

-    0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d

-};

-

-#define rj_sbox(x)     sbox[(x)]

-#define rj_sbox_inv(x) sboxinv[(x)]

-

-#else /* tableless subroutines */

-

-/* -------------------------------------------------------------------------- */

-uint8_t gf_alog(uint8_t x) // calculate anti-logarithm gen 3

-{

-    uint8_t atb = 1, z;

-

-    while (x--) {z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;}

-

-    return atb;

-} /* gf_alog */

-

-/* -------------------------------------------------------------------------- */

-uint8_t gf_log(uint8_t x) // calculate logarithm gen 3

-{

-    uint8_t atb = 1, i = 0, z;

-

-    do {

-        if (atb == x) break;

-        z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;

-    } while (++i > 0);

-

-    return i;

-} /* gf_log */

-

-

-/* -------------------------------------------------------------------------- */

-uint8_t gf_mulinv(uint8_t x) // calculate multiplicative inverse

-{

-    return (x) ? gf_alog(255 - gf_log(x)) : 0;

-} /* gf_mulinv */

-

-/* -------------------------------------------------------------------------- */

-uint8_t rj_sbox(uint8_t x)

-{

-    uint8_t y, sb;

-

-    sb = y = gf_mulinv(x);

-    y = (y<<1)|(y>>7); sb ^= y;  y = (y<<1)|(y>>7); sb ^= y; 

-    y = (y<<1)|(y>>7); sb ^= y;  y = (y<<1)|(y>>7); sb ^= y;

-

-    return (sb ^ 0x63);

-} /* rj_sbox */

-

-/* -------------------------------------------------------------------------- */

-uint8_t rj_sbox_inv(uint8_t x)

-{

-    uint8_t y, sb;

-

-    y = x ^ 0x63;

-    sb = y = (y<<1)|(y>>7);

-    y = (y<<2)|(y>>6); sb ^= y; y = (y<<3)|(y>>5); sb ^= y;

-

-    return gf_mulinv(sb);

-} /* rj_sbox_inv */

-

-#endif

-

-/* -------------------------------------------------------------------------- */

-uint8_t rj_xtime(uint8_t x) 

-{

-    return (x & 0x80) ? ((x << 1) ^ 0x1b) : (x << 1);

-} /* rj_xtime */

-

-/* -------------------------------------------------------------------------- */

-void aes_subBytes(uint8_t *buf)

-{

-    register uint8_t i = 16;

-

-    while (i--) buf[i] = rj_sbox(buf[i]);

-} /* aes_subBytes */

-

-/* -------------------------------------------------------------------------- */

-void aes_subBytes_inv(uint8_t *buf)

-{

-    register uint8_t i = 16;

-

-    while (i--) buf[i] = rj_sbox_inv(buf[i]);

-} /* aes_subBytes_inv */

-

-/* -------------------------------------------------------------------------- */

-void aes_addRoundKey(uint8_t *buf, uint8_t *key)

-{

-    register uint8_t i = 16;

-

-    while (i--) buf[i] ^= key[i];

-} /* aes_addRoundKey */

-

-/* -------------------------------------------------------------------------- */

-void aes_addRoundKey_cpy(uint8_t *buf, uint8_t *key, uint8_t *cpk)

-{

-    register uint8_t i = 16;

-

-    while (i--)  buf[i] ^= (cpk[i] = key[i]), cpk[16+i] = key[16 + i];

-} /* aes_addRoundKey_cpy */

-

-

-/* -------------------------------------------------------------------------- */

-void aes_shiftRows(uint8_t *buf)

-{

-    register uint8_t i, j; /* to make it potentially parallelable :) */

-

-    i = buf[1]; buf[1] = buf[5]; buf[5] = buf[9]; buf[9] = buf[13]; buf[13] = i;

-    i = buf[10]; buf[10] = buf[2]; buf[2] = i;

-    j = buf[3]; buf[3] = buf[15]; buf[15] = buf[11]; buf[11] = buf[7]; buf[7] = j;

-    j = buf[14]; buf[14] = buf[6]; buf[6]  = j;

-

-} /* aes_shiftRows */

-

-/* -------------------------------------------------------------------------- */

-void aes_shiftRows_inv(uint8_t *buf)

-{

-    register uint8_t i, j; /* same as above :) */

-

-    i = buf[1]; buf[1] = buf[13]; buf[13] = buf[9]; buf[9] = buf[5]; buf[5] = i;

-    i = buf[2]; buf[2] = buf[10]; buf[10] = i;

-    j = buf[3]; buf[3] = buf[7]; buf[7] = buf[11]; buf[11] = buf[15]; buf[15] = j;

-    j = buf[6]; buf[6] = buf[14]; buf[14] = j;

-

-} /* aes_shiftRows_inv */

-

-/* -------------------------------------------------------------------------- */

-void aes_mixColumns(uint8_t *buf)

-{

-    register uint8_t i, a, b, c, d, e;

-

-    for (i = 0; i < 16; i += 4)

-    {

-        a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];

-        e = a ^ b ^ c ^ d;

-        buf[i] ^= e ^ rj_xtime(a^b);   buf[i+1] ^= e ^ rj_xtime(b^c);

-        buf[i+2] ^= e ^ rj_xtime(c^d); buf[i+3] ^= e ^ rj_xtime(d^a);

-    }

-} /* aes_mixColumns */

-

-/* -------------------------------------------------------------------------- */

-void aes_mixColumns_inv(uint8_t *buf)

-{

-    register uint8_t i, a, b, c, d, e, x, y, z;

-

-    for (i = 0; i < 16; i += 4)

-    {

-        a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];

-        e = a ^ b ^ c ^ d;

-        z = rj_xtime(e);

-        x = e ^ rj_xtime(rj_xtime(z^a^c));  y = e ^ rj_xtime(rj_xtime(z^b^d));

-        buf[i] ^= x ^ rj_xtime(a^b);   buf[i+1] ^= y ^ rj_xtime(b^c);

-        buf[i+2] ^= x ^ rj_xtime(c^d); buf[i+3] ^= y ^ rj_xtime(d^a);

-    }

-} /* aes_mixColumns_inv */

-

-/* -------------------------------------------------------------------------- */

-void aes_expandEncKey(uint8_t *k, uint8_t *rc) 

-{

-    register uint8_t i;

-

-    k[0] ^= rj_sbox(k[29]) ^ (*rc);

-    k[1] ^= rj_sbox(k[30]);

-    k[2] ^= rj_sbox(k[31]);

-    k[3] ^= rj_sbox(k[28]);

-    *rc = F( *rc);

-

-    for(i = 4; i < 16; i += 4)  k[i] ^= k[i-4],   k[i+1] ^= k[i-3],

-        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];

-    k[16] ^= rj_sbox(k[12]);

-    k[17] ^= rj_sbox(k[13]);

-    k[18] ^= rj_sbox(k[14]);

-    k[19] ^= rj_sbox(k[15]);

-

-    for(i = 20; i < 32; i += 4) k[i] ^= k[i-4],   k[i+1] ^= k[i-3],

-        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];

-

-} /* aes_expandEncKey */

-

-/* -------------------------------------------------------------------------- */

-void aes_expandDecKey(uint8_t *k, uint8_t *rc) 

-{

-    uint8_t i;

-

-    for(i = 28; i > 16; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3], 

-        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];

-

-    k[16] ^= rj_sbox(k[12]);

-    k[17] ^= rj_sbox(k[13]);

-    k[18] ^= rj_sbox(k[14]);

-    k[19] ^= rj_sbox(k[15]);

-

-    for(i = 12; i > 0; i -= 4)  k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],

-        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];

-

-    *rc = FD(*rc);

-    k[0] ^= rj_sbox(k[29]) ^ (*rc);

-    k[1] ^= rj_sbox(k[30]);

-    k[2] ^= rj_sbox(k[31]);

-    k[3] ^= rj_sbox(k[28]);

-} /* aes_expandDecKey */

-

-

-/* -------------------------------------------------------------------------- */

-void aes256_init(aes256_context *ctx, uint8_t *k)

-{

-    uint8_t rcon = 1;

-    register uint8_t i;

-

-    for (i = 0; i < sizeof(ctx->key); i++) ctx->enckey[i] = ctx->deckey[i] = k[i];

-    for (i = 8;--i;) aes_expandEncKey(ctx->deckey, &rcon);

-} /* aes256_init */

-

-/* -------------------------------------------------------------------------- */

-void aes256_done(aes256_context *ctx)

-{

-    register uint8_t i;

-

-    for (i = 0; i < sizeof(ctx->key); i++) 

-        ctx->key[i] = ctx->enckey[i] = ctx->deckey[i] = 0;

-} /* aes256_done */

-

-/* -------------------------------------------------------------------------- */

-void aes256_encrypt_ecb(aes256_context *ctx, uint8_t *buf)

-{

-    uint8_t i, rcon;

-

-    aes_addRoundKey_cpy(buf, ctx->enckey, ctx->key);

-    for(i = 1, rcon = 1; i < 14; ++i)

-    {

-        aes_subBytes(buf);

-        aes_shiftRows(buf);

-        aes_mixColumns(buf);

-        if( i & 1 ) aes_addRoundKey( buf, &ctx->key[16]);

-        else aes_expandEncKey(ctx->key, &rcon), aes_addRoundKey(buf, ctx->key);

-    }

-    aes_subBytes(buf);

-    aes_shiftRows(buf);

-    aes_expandEncKey(ctx->key, &rcon); 

-    aes_addRoundKey(buf, ctx->key);

-} /* aes256_encrypt */

-

-/* -------------------------------------------------------------------------- */

-void aes256_decrypt_ecb(aes256_context *ctx, uint8_t *buf)

-{

-    uint8_t i, rcon;

-

-    aes_addRoundKey_cpy(buf, ctx->deckey, ctx->key);

-    aes_shiftRows_inv(buf);

-    aes_subBytes_inv(buf);

-

-    for (i = 14, rcon = 0x80; --i;)

-    {

-        if( ( i & 1 ) )           

-        {

-            aes_expandDecKey(ctx->key, &rcon);

-            aes_addRoundKey(buf, &ctx->key[16]);

-        }

-        else aes_addRoundKey(buf, ctx->key);

-        aes_mixColumns_inv(buf);

-        aes_shiftRows_inv(buf);

-        aes_subBytes_inv(buf);

-    }

-    aes_addRoundKey( buf, ctx->key); 

-} /* aes256_decrypt */

+/*  
+*   Byte-oriented AES-256 implementation.
+*   All lookup tables replaced with 'on the fly' calculations. 
+*
+*   Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
+*   Other contributors: Hal Finney
+*
+*   Permission to use, copy, modify, and distribute this software for any
+*   purpose with or without fee is hereby granted, provided that the above
+*   copyright notice and this permission notice appear in all copies.
+*
+*   THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+*   WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+*   MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+*   ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+*   WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+*   ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+*   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+*/
+#include "aes256.h"
+
+#define F(x)   (((x)<<1) ^ ((((x)>>7) & 1) * 0x1b))
+#define FD(x)  (((x) >> 1) ^ (((x) & 1) ? 0x8d : 0))
+
+// #define BACK_TO_TABLES
+#ifdef BACK_TO_TABLES
+
+const uint8_t sbox[256] = {
+    0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,
+    0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,
+    0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
+    0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,
+    0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
+    0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
+    0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,
+    0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,
+    0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
+    0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
+    0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,
+    0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
+    0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,
+    0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,
+    0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
+    0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,
+    0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,
+    0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
+    0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,
+    0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
+    0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
+    0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,
+    0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,
+    0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
+    0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
+    0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,
+    0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
+    0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,
+    0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,
+    0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
+    0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,
+    0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16
+};
+const uint8_t sboxinv[256] = {
+    0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38,
+    0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,
+    0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87,
+    0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,
+    0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d,
+    0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,
+    0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2,
+    0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,
+    0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16,
+    0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,
+    0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda,
+    0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,
+    0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a,
+    0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,
+    0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02,
+    0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,
+    0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea,
+    0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,
+    0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85,
+    0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,
+    0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89,
+    0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,
+    0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20,
+    0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,
+    0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31,
+    0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,
+    0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d,
+    0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,
+    0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0,
+    0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,
+    0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26,
+    0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d
+};
+
+#define rj_sbox(x)     sbox[(x)]
+#define rj_sbox_inv(x) sboxinv[(x)]
+
+#else /* tableless subroutines */
+
+/* -------------------------------------------------------------------------- */
+uint8_t gf_alog(uint8_t x) // calculate anti-logarithm gen 3
+{
+    uint8_t atb = 1, z;
+
+    while (x--) {z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;}
+
+    return atb;
+} /* gf_alog */
+
+/* -------------------------------------------------------------------------- */
+uint8_t gf_log(uint8_t x) // calculate logarithm gen 3
+{
+    uint8_t atb = 1, i = 0, z;
+
+    do {
+        if (atb == x) break;
+        z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;
+    } while (++i > 0);
+
+    return i;
+} /* gf_log */
+
+
+/* -------------------------------------------------------------------------- */
+uint8_t gf_mulinv(uint8_t x) // calculate multiplicative inverse
+{
+    return (x) ? gf_alog(255 - gf_log(x)) : 0;
+} /* gf_mulinv */
+
+/* -------------------------------------------------------------------------- */
+uint8_t rj_sbox(uint8_t x)
+{
+    uint8_t y, sb;
+
+    sb = y = gf_mulinv(x);
+    y = (y<<1)|(y>>7); sb ^= y;  y = (y<<1)|(y>>7); sb ^= y; 
+    y = (y<<1)|(y>>7); sb ^= y;  y = (y<<1)|(y>>7); sb ^= y;
+
+    return (sb ^ 0x63);
+} /* rj_sbox */
+
+/* -------------------------------------------------------------------------- */
+uint8_t rj_sbox_inv(uint8_t x)
+{
+    uint8_t y, sb;
+
+    y = x ^ 0x63;
+    sb = y = (y<<1)|(y>>7);
+    y = (y<<2)|(y>>6); sb ^= y; y = (y<<3)|(y>>5); sb ^= y;
+
+    return gf_mulinv(sb);
+} /* rj_sbox_inv */
+
+#endif
+
+/* -------------------------------------------------------------------------- */
+uint8_t rj_xtime(uint8_t x) 
+{
+    return (x & 0x80) ? ((x << 1) ^ 0x1b) : (x << 1);
+} /* rj_xtime */
+
+/* -------------------------------------------------------------------------- */
+void aes_subBytes(uint8_t *buf)
+{
+    register uint8_t i = 16;
+
+    while (i--) buf[i] = rj_sbox(buf[i]);
+} /* aes_subBytes */
+
+/* -------------------------------------------------------------------------- */
+void aes_subBytes_inv(uint8_t *buf)
+{
+    register uint8_t i = 16;
+
+    while (i--) buf[i] = rj_sbox_inv(buf[i]);
+} /* aes_subBytes_inv */
+
+/* -------------------------------------------------------------------------- */
+void aes_addRoundKey(uint8_t *buf, uint8_t *key)
+{
+    register uint8_t i = 16;
+
+    while (i--) buf[i] ^= key[i];
+} /* aes_addRoundKey */
+
+/* -------------------------------------------------------------------------- */
+void aes_addRoundKey_cpy(uint8_t *buf, uint8_t *key, uint8_t *cpk)
+{
+    register uint8_t i = 16;
+
+    while (i--)  buf[i] ^= (cpk[i] = key[i]), cpk[16+i] = key[16 + i];
+} /* aes_addRoundKey_cpy */
+
+
+/* -------------------------------------------------------------------------- */
+void aes_shiftRows(uint8_t *buf)
+{
+    register uint8_t i, j; /* to make it potentially parallelable :) */
+
+    i = buf[1]; buf[1] = buf[5]; buf[5] = buf[9]; buf[9] = buf[13]; buf[13] = i;
+    i = buf[10]; buf[10] = buf[2]; buf[2] = i;
+    j = buf[3]; buf[3] = buf[15]; buf[15] = buf[11]; buf[11] = buf[7]; buf[7] = j;
+    j = buf[14]; buf[14] = buf[6]; buf[6]  = j;
+
+} /* aes_shiftRows */
+
+/* -------------------------------------------------------------------------- */
+void aes_shiftRows_inv(uint8_t *buf)
+{
+    register uint8_t i, j; /* same as above :) */
+
+    i = buf[1]; buf[1] = buf[13]; buf[13] = buf[9]; buf[9] = buf[5]; buf[5] = i;
+    i = buf[2]; buf[2] = buf[10]; buf[10] = i;
+    j = buf[3]; buf[3] = buf[7]; buf[7] = buf[11]; buf[11] = buf[15]; buf[15] = j;
+    j = buf[6]; buf[6] = buf[14]; buf[14] = j;
+
+} /* aes_shiftRows_inv */
+
+/* -------------------------------------------------------------------------- */
+void aes_mixColumns(uint8_t *buf)
+{
+    register uint8_t i, a, b, c, d, e;
+
+    for (i = 0; i < 16; i += 4)
+    {
+        a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
+        e = a ^ b ^ c ^ d;
+        buf[i] ^= e ^ rj_xtime(a^b);   buf[i+1] ^= e ^ rj_xtime(b^c);
+        buf[i+2] ^= e ^ rj_xtime(c^d); buf[i+3] ^= e ^ rj_xtime(d^a);
+    }
+} /* aes_mixColumns */
+
+/* -------------------------------------------------------------------------- */
+void aes_mixColumns_inv(uint8_t *buf)
+{
+    register uint8_t i, a, b, c, d, e, x, y, z;
+
+    for (i = 0; i < 16; i += 4)
+    {
+        a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
+        e = a ^ b ^ c ^ d;
+        z = rj_xtime(e);
+        x = e ^ rj_xtime(rj_xtime(z^a^c));  y = e ^ rj_xtime(rj_xtime(z^b^d));
+        buf[i] ^= x ^ rj_xtime(a^b);   buf[i+1] ^= y ^ rj_xtime(b^c);
+        buf[i+2] ^= x ^ rj_xtime(c^d); buf[i+3] ^= y ^ rj_xtime(d^a);
+    }
+} /* aes_mixColumns_inv */
+
+/* -------------------------------------------------------------------------- */
+void aes_expandEncKey(uint8_t *k, uint8_t *rc) 
+{
+    register uint8_t i;
+
+    k[0] ^= rj_sbox(k[29]) ^ (*rc);
+    k[1] ^= rj_sbox(k[30]);
+    k[2] ^= rj_sbox(k[31]);
+    k[3] ^= rj_sbox(k[28]);
+    *rc = F( *rc);
+
+    for(i = 4; i < 16; i += 4)  k[i] ^= k[i-4],   k[i+1] ^= k[i-3],
+        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
+    k[16] ^= rj_sbox(k[12]);
+    k[17] ^= rj_sbox(k[13]);
+    k[18] ^= rj_sbox(k[14]);
+    k[19] ^= rj_sbox(k[15]);
+
+    for(i = 20; i < 32; i += 4) k[i] ^= k[i-4],   k[i+1] ^= k[i-3],
+        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
+
+} /* aes_expandEncKey */
+
+/* -------------------------------------------------------------------------- */
+void aes_expandDecKey(uint8_t *k, uint8_t *rc) 
+{
+    uint8_t i;
+
+    for(i = 28; i > 16; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3], 
+        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
+
+    k[16] ^= rj_sbox(k[12]);
+    k[17] ^= rj_sbox(k[13]);
+    k[18] ^= rj_sbox(k[14]);
+    k[19] ^= rj_sbox(k[15]);
+
+    for(i = 12; i > 0; i -= 4)  k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],
+        k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
+
+    *rc = FD(*rc);
+    k[0] ^= rj_sbox(k[29]) ^ (*rc);
+    k[1] ^= rj_sbox(k[30]);
+    k[2] ^= rj_sbox(k[31]);
+    k[3] ^= rj_sbox(k[28]);
+} /* aes_expandDecKey */
+
+
+/* -------------------------------------------------------------------------- */
+void aes256_init(aes256_context *ctx, uint8_t *k)
+{
+    uint8_t rcon = 1;
+    register uint8_t i;
+
+    for (i = 0; i < sizeof(ctx->key); i++) ctx->enckey[i] = ctx->deckey[i] = k[i];
+    for (i = 8;--i;) aes_expandEncKey(ctx->deckey, &rcon);
+} /* aes256_init */
+
+/* -------------------------------------------------------------------------- */
+void aes256_done(aes256_context *ctx)
+{
+    register uint8_t i;
+
+    for (i = 0; i < sizeof(ctx->key); i++) 
+        ctx->key[i] = ctx->enckey[i] = ctx->deckey[i] = 0;
+} /* aes256_done */
+
+/* -------------------------------------------------------------------------- */
+void aes256_encrypt_ecb(aes256_context *ctx, uint8_t *buf)
+{
+    uint8_t i, rcon;
+
+    aes_addRoundKey_cpy(buf, ctx->enckey, ctx->key);
+    for(i = 1, rcon = 1; i < 14; ++i)
+    {
+        aes_subBytes(buf);
+        aes_shiftRows(buf);
+        aes_mixColumns(buf);
+        if( i & 1 ) aes_addRoundKey( buf, &ctx->key[16]);
+        else aes_expandEncKey(ctx->key, &rcon), aes_addRoundKey(buf, ctx->key);
+    }
+    aes_subBytes(buf);
+    aes_shiftRows(buf);
+    aes_expandEncKey(ctx->key, &rcon); 
+    aes_addRoundKey(buf, ctx->key);
+} /* aes256_encrypt */
+
+/* -------------------------------------------------------------------------- */
+void aes256_decrypt_ecb(aes256_context *ctx, uint8_t *buf)
+{
+    uint8_t i, rcon;
+
+    aes_addRoundKey_cpy(buf, ctx->deckey, ctx->key);
+    aes_shiftRows_inv(buf);
+    aes_subBytes_inv(buf);
+
+    for (i = 14, rcon = 0x80; --i;)
+    {
+        if( ( i & 1 ) )           
+        {
+            aes_expandDecKey(ctx->key, &rcon);
+            aes_addRoundKey(buf, &ctx->key[16]);
+        }
+        else aes_addRoundKey(buf, ctx->key);
+        aes_mixColumns_inv(buf);
+        aes_shiftRows_inv(buf);
+        aes_subBytes_inv(buf);
+    }
+    aes_addRoundKey( buf, ctx->key); 
+} /* aes256_decrypt */
diff --git a/core/io/aes256.h b/core/io/aes256.h
index 180352e970..fabbcf1968 100644
--- a/core/io/aes256.h
+++ b/core/io/aes256.h
@@ -1,46 +1,46 @@
-/*  

-*   Byte-oriented AES-256 implementation.

-*   All lookup tables replaced with 'on the fly' calculations. 

-*

-*   Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com

-*   Other contributors: Hal Finney

-*

-*   Permission to use, copy, modify, and distribute this software for any

-*   purpose with or without fee is hereby granted, provided that the above

-*   copyright notice and this permission notice appear in all copies.

-*

-*   THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES

-*   WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF

-*   MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR

-*   ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES

-*   WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN

-*   ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF

-*   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

-*/

-

-#ifndef AES_256_H

-#define AES_256_H

-

-#include "typedefs.h"

-

-#ifdef __cplusplus

-extern "C" { 

-#endif

-

-    typedef struct {

-        uint8_t key[32]; 

-        uint8_t enckey[32]; 

-        uint8_t deckey[32];

-    } aes256_context; 

-

-

-    void aes256_init(aes256_context *, uint8_t * /* key */);

-    void aes256_done(aes256_context *);

-    void aes256_encrypt_ecb(aes256_context *, uint8_t * /* plaintext */);

-    void aes256_decrypt_ecb(aes256_context *, uint8_t * /* cipertext */);

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif

+/*  
+*   Byte-oriented AES-256 implementation.
+*   All lookup tables replaced with 'on the fly' calculations. 
+*
+*   Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
+*   Other contributors: Hal Finney
+*
+*   Permission to use, copy, modify, and distribute this software for any
+*   purpose with or without fee is hereby granted, provided that the above
+*   copyright notice and this permission notice appear in all copies.
+*
+*   THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+*   WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+*   MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+*   ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+*   WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+*   ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+*   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+*/
+
+#ifndef AES_256_H
+#define AES_256_H
+
+#include "typedefs.h"
+
+#ifdef __cplusplus
+extern "C" { 
+#endif
+
+    typedef struct {
+        uint8_t key[32]; 
+        uint8_t enckey[32]; 
+        uint8_t deckey[32];
+    } aes256_context; 
+
+
+    void aes256_init(aes256_context *, uint8_t * /* key */);
+    void aes256_done(aes256_context *);
+    void aes256_encrypt_ecb(aes256_context *, uint8_t * /* plaintext */);
+    void aes256_decrypt_ecb(aes256_context *, uint8_t * /* cipertext */);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/core/io/unzip.c b/core/io/unzip.c
index 0cd975211e..b438021ad7 100644
--- a/core/io/unzip.c
+++ b/core/io/unzip.c
@@ -1788,7 +1788,7 @@ extern int ZEXPORT unzReadCurrentFile  (unzFile file, voidp buf, unsigned len)
         return UNZ_PARAMERROR;
 
 
-    if ((pfile_in_zip_read_info->read_buffer == NULL))
+    if (pfile_in_zip_read_info->read_buffer==NULL)
         return UNZ_END_OF_LIST_OF_FILE;
     if (len==0)
         return 0;
diff --git a/core/io/zip.c b/core/io/zip.c
index 8f6aeb922f..c4ab93ab81 100644
--- a/core/io/zip.c
+++ b/core/io/zip.c
@@ -1114,9 +1114,9 @@ extern int ZEXPORT zipOpenNewFileInZip4_64 (zipFile file, const char* filename,
     zi->ci.flag = flagBase;
     if ((level==8) || (level==9))
       zi->ci.flag |= 2;
-    if ((level==2))
+    if (level==2)
       zi->ci.flag |= 4;
-    if ((level==1))
+    if (level==1)
       zi->ci.flag |= 6;
     if (password != NULL)
       zi->ci.flag |= 1;
diff --git a/core/os/os.cpp b/core/os/os.cpp
index 2db926e556..ee9f12b79d 100644
--- a/core/os/os.cpp
+++ b/core/os/os.cpp
@@ -516,6 +516,7 @@ OS::OS() {
 	_target_fps=0;
 	_render_thread_mode=RENDER_THREAD_SAFE;
 	_time_scale=1.0;
+	_pixel_snap=false;
 	Math::seed(1234567);
 }
 
diff --git a/core/os/os.h b/core/os/os.h
index d89734d7d3..e5338b4a02 100644
--- a/core/os/os.h
+++ b/core/os/os.h
@@ -58,6 +58,7 @@ class OS {
 	float _fps;
 	int _target_fps;
 	float _time_scale;
+	bool _pixel_snap;
 
 	char *last_error;
 
@@ -393,7 +394,7 @@ public:
 	void set_time_scale(float p_scale);
 	float get_time_scale() const;
 
-
+	_FORCE_INLINE_ bool get_use_pixel_snap() const { return _pixel_snap; }
 
 	OS();	
 	virtual ~OS();
diff --git a/core/path_db.cpp b/core/path_db.cpp
index d3dc3aceb8..c6ea25d966 100644
--- a/core/path_db.cpp
+++ b/core/path_db.cpp
@@ -286,6 +286,37 @@ NodePath::NodePath(const Vector<StringName>& p_path,const Vector<StringName>& p_
 	data->property=p_property;
 }
 
+
+void NodePath::simplify() {
+
+	if (!data)
+		return;
+	for(int i=0;i<data->path.size();i++) {
+		if (data->path.size()==1)
+			break;
+		if (data->path[i].operator String()==".") {
+			data->path.remove(i);
+			i--;
+		} else if (data->path[i].operator String()==".." && i>0 && data->path[i-1].operator String()!="." && data->path[i-1].operator String()!="..") {
+			//remove both
+			data->path.remove(i-1);
+			data->path.remove(i-1);
+			i-=2;
+			if (data->path.size()==0) {
+				data->path.push_back(".");
+				break;
+			}
+		}
+	}
+}
+
+NodePath NodePath::simplified() const {
+
+	NodePath np=*this;
+	np.simplify();
+	return np;
+}
+
 NodePath::NodePath(const String& p_path) {
 
 	data=NULL;
diff --git a/core/path_db.h b/core/path_db.h
index b4f13d50be..de84216006 100644
--- a/core/path_db.h
+++ b/core/path_db.h
@@ -84,7 +84,10 @@ public:
 	bool operator==(const NodePath& p_path) const;
 	bool operator!=(const NodePath& p_path) const;
 	void operator=(const NodePath& p_path);
-	
+
+	void simplify();
+	NodePath simplified() const;
+
 	NodePath(const Vector<StringName>& p_path,bool p_absolute,const String& p_property="");	
 	NodePath(const Vector<StringName>& p_path,const Vector<StringName>& p_subpath,bool p_absolute,const String& p_property="");
 	NodePath(const NodePath& p_path);
diff --git a/core/resource.h b/core/resource.h
index 9d9c445e1d..3596abe673 100644
--- a/core/resource.h
+++ b/core/resource.h
@@ -130,7 +130,7 @@ public:
 	void set_name(const String& p_name);
 	String get_name() const;
 
-	void set_path(const String& p_path,bool p_take_over=false);
+	virtual void set_path(const String& p_path,bool p_take_over=false);
 	String get_path() const;
 
 	void set_subindex(int p_sub_index);
diff --git a/demos/2d/platformer/coin.gd b/demos/2d/platformer/coin.gd
index 3e31a395ae..983cd46d88 100644
--- a/demos/2d/platformer/coin.gd
+++ b/demos/2d/platformer/coin.gd
@@ -18,3 +18,11 @@ func _ready():
 	# Initalization here
 	pass
 
+
+
+func _on_coin_area_enter( area ):
+	pass # replace with function body
+
+
+func _on_coin_area_enter_shape( area_id, area, area_shape, area_shape ):
+	pass # replace with function body
diff --git a/doc/base/classes.xml b/doc/base/classes.xml
index d83997ad8a..157d49fbc8 100644
--- a/doc/base/classes.xml
+++ b/doc/base/classes.xml
@@ -5905,7 +5905,7 @@
 	<description>
 	Base class of anything 2D. Canvas items are laid out in a tree and children inherit and extend the transform of their parent. CanvasItem is extended by [Control], for anything GUI related, and by [Node2D] for anything 2D engine related.
 	Any CanvasItem can draw. For this, the "update" function must be called, then NOTIFICATION_DRAW will be received on idle time to request redraw. Because of this, canvas items don't need to be redraw on every frame, improving the performance significan'tly. Several functions for drawing on the CanvasItem are provided (see draw_* functions). They can only be used inside the notification, signal or _draw() overrided function, though.
-	Canvas items are draw in tree order. By default, children are on top of their parents so a root CanvasItem will be drawn behind everything (this can be changed per item though). 
+	Canvas items are draw in tree order. By default, children are on top of their parents so a root CanvasItem will be drawn behind everything (this can be changed per item though).
 	Canvas items can also be hidden (hiding also their subtree). They provide many means for changing standard parameters such as opacity (for it and the subtree) and self opacity, blend mode.
 	Ultimately, a transform notification can be requested, which will notify the node that its global position changed in case the parent tree changed.
 	</description>
@@ -11556,8 +11556,11 @@
 </class>
 <class name="GraphEdit" inherits="Control" category="Core">
 	<brief_description>
+		GraphEdit is an area capable of showing various GraphNodes. It manages connection events between them.
 	</brief_description>
 	<description>
+		GraphEdit manages the showing of GraphNodes it contains, as well as connections an disconnections between them. Signals are sent for each of these two events. Disconnection between GraphNodes slots is disabled by default.
+		It is greatly advised to enable low processor usage mode [OS.set_low_processor_usage_mode()] when using GraphEdits.
 	</description>
 	<methods>
 		<method name="connect_node">
@@ -11572,6 +11575,7 @@
 			<argument index="3" name="to_port" type="int">
 			</argument>
 			<description>
+				Create a connection between 'from_port' slot of 'from' GraphNode and 'to_port' slot of 'to' GraphNode. If the connection already exists, no connection is created.
 			</description>
 		</method>
 		<method name="is_node_connected">
@@ -11586,6 +11590,7 @@
 			<argument index="3" name="to_port" type="int">
 			</argument>
 			<description>
+				Return true if the 'from_port' slot of 'from' GraphNode is connected to the 'to_port' slot of 'to' GraphNode.
 			</description>
 		</method>
 		<method name="disconnect_node">
@@ -11598,24 +11603,28 @@
 			<argument index="3" name="to_port" type="int">
 			</argument>
 			<description>
+				Remove the connection between 'from_port' slot of 'from' GraphNode and 'to_port' slot of 'to' GraphNode, if connection exists.
 			</description>
 		</method>
 		<method name="get_connection_list" qualifiers="const">
 			<return type="Array">
 			</return>
 			<description>
+				Return an Array containing the list of connections. A connection consists in a structure of the form {from_slot: 0, from: "GraphNode name 0", to_slot: 1, to: "GraphNode name 1" }
 			</description>
 		</method>
 		<method name="set_right_disconnects">
 			<argument index="0" name="enable" type="bool">
 			</argument>
 			<description>
+				Enable the disconnection of existing connections in the visual GraphEdit by left-clicking a connection and releasing into the void.
 			</description>
 		</method>
 		<method name="is_right_disconnects_enabled" qualifiers="const">
 			<return type="bool">
 			</return>
 			<description>
+				Return true is the disconnection of connections is enable in the visual GraphEdit. False otherwise.
 			</description>
 		</method>
 	</methods>
@@ -11630,6 +11639,7 @@
 			<argument index="3" name="to_slot" type="int">
 			</argument>
 			<description>
+				Signal sent to the GraphEdit when the connection between 'from_slot' slot of 'from' GraphNode and 'to_slot' slot of 'to' GraphNode is attempted to be removed.
 			</description>
 		</signal>
 		<signal name="connection_request">
@@ -11642,6 +11652,7 @@
 			<argument index="3" name="to_slot" type="int">
 			</argument>
 			<description>
+				Signal sent to the GraphEdit when the connection between 'from_slot' slot of 'from' GraphNode and 'to_slot' slot of 'to' GraphNode is attempted to be created.
 			</description>
 		</signal>
 	</signals>
@@ -11650,20 +11661,24 @@
 </class>
 <class name="GraphNode" inherits="Container" category="Core">
 	<brief_description>
+		A GraphNode is a container with several input and output slots allowing connections between GraphNodes. Slots can have different, incompatible types.
 	</brief_description>
 	<description>
+		A GraphNode is a container defined by a title. It can have 1 or more input and output slots, which can be enabled (shown) or disabled (not shown) and have different (incompatible) types. Colors can also be assigned to slots. A tuple of input and output slots is defined for each GUI element included in the GraphNode. Input and output connections are left and right slots, but only enabled slots are counted as connections.
 	</description>
 	<methods>
 		<method name="set_title">
 			<argument index="0" name="title" type="String">
 			</argument>
 			<description>
+				Set the title of the GraphNode.
 			</description>
 		</method>
 		<method name="get_title" qualifiers="const">
 			<return type="String">
 			</return>
 			<description>
+				Return the title of the GraphNode.
 			</description>
 		</method>
 		<method name="set_slot">
@@ -11682,16 +11697,19 @@
 			<argument index="6" name="color_right" type="Color">
 			</argument>
 			<description>
+				Set the tuple of input/output slots defined by 'idx' ID. 'left' slots are input, 'right' are output. 'type' is an integer defining the type of the slot. Refer to description for the compatibility between slot types.
 			</description>
 		</method>
 		<method name="clear_slot">
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Disable input and ouput slot whose index is 'idx'.
 			</description>
 		</method>
 		<method name="clear_all_slots">
 			<description>
+				Disable all input and output slots of the GraphNode.
 			</description>
 		</method>
 		<method name="is_slot_enabled_left" qualifiers="const">
@@ -11700,6 +11718,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return true if left (input) slot 'idx' is enabled. False otherwise.
 			</description>
 		</method>
 		<method name="get_slot_type_left" qualifiers="const">
@@ -11708,6 +11727,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the (integer) type of left (input) 'idx' slot.
 			</description>
 		</method>
 		<method name="get_slot_color_left" qualifiers="const">
@@ -11716,6 +11736,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the color set to 'idx' left (input) slot.
 			</description>
 		</method>
 		<method name="is_slot_enabled_right" qualifiers="const">
@@ -11724,6 +11745,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return true if right (output) slot 'idx' is enabled. False otherwise.
 			</description>
 		</method>
 		<method name="get_slot_type_right" qualifiers="const">
@@ -11732,6 +11754,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the (integer) type of right (output) 'idx' slot.
 			</description>
 		</method>
 		<method name="get_slot_color_right" qualifiers="const">
@@ -11740,30 +11763,35 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the color set to 'idx' right (output) slot.
 			</description>
 		</method>
 		<method name="set_offset">
 			<argument index="0" name="offset" type="Vector2">
 			</argument>
 			<description>
+				Set the offset of the GraphNode.
 			</description>
 		</method>
 		<method name="get_offset" qualifiers="const">
 			<return type="Vector2">
 			</return>
 			<description>
+				Return the offset of the GraphNode.
 			</description>
 		</method>
 		<method name="get_connection_output_count">
 			<return type="int">
 			</return>
 			<description>
+				Return the number of enabled output slots (connections) of the GraphNode.
 			</description>
 		</method>
 		<method name="get_connection_input_count">
 			<return type="int">
 			</return>
 			<description>
+				Return the number of enabled input slots (connections) to the GraphNode.
 			</description>
 		</method>
 		<method name="get_connection_output_pos">
@@ -11772,6 +11800,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the position of the output connection 'idx'.
 			</description>
 		</method>
 		<method name="get_connection_output_type">
@@ -11780,6 +11809,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the type of the output connection 'idx'.
 			</description>
 		</method>
 		<method name="get_connection_output_color">
@@ -11788,6 +11818,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the color of the output connection 'idx'.
 			</description>
 		</method>
 		<method name="get_connection_input_pos">
@@ -11796,6 +11827,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the position of the input connection 'idx'.
 			</description>
 		</method>
 		<method name="get_connection_input_type">
@@ -11804,6 +11836,7 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the type of the input connection 'idx'.
 			</description>
 		</method>
 		<method name="get_connection_input_color">
@@ -11812,28 +11845,33 @@
 			<argument index="0" name="idx" type="int">
 			</argument>
 			<description>
+				Return the color of the input connection 'idx'.
 			</description>
 		</method>
 		<method name="set_show_close_button">
 			<argument index="0" name="show" type="bool">
 			</argument>
 			<description>
+				Show the close button on the GraphNode if 'show' is true (disabled by default). If enabled, a connection on the signal close_request is needed for the close button to work.
 			</description>
 		</method>
 		<method name="is_close_button_visible" qualifiers="const">
 			<return type="bool">
 			</return>
 			<description>
+				Returns true if the close button is shown. False otherwise.
 			</description>
 		</method>
 	</methods>
 	<signals>
 		<signal name="raise_request">
 			<description>
+				Signal sent when the GraphNode is requested to be displayed over other ones. Happens on focusing (clicking into) the GraphNode.
 			</description>
 		</signal>
 		<signal name="close_request">
 			<description>
+				Signal sent on closing the GraphNode.
 			</description>
 		</signal>
 		<signal name="dragged">
@@ -11842,10 +11880,12 @@
 			<argument index="1" name="to" type="Vector2">
 			</argument>
 			<description>
+				Signal sent when the GraphNode is dragged.
 			</description>
 		</signal>
 		<signal name="offset_changed">
 			<description>
+				Signal sent when the GraphNode is moved.
 			</description>
 		</signal>
 	</signals>
@@ -17934,7 +17974,7 @@
 	</brief_description>
 	<description>
 	Nodes can be set as children of other nodes, resulting in a tree arrangement. Any tree of nodes is called a "Scene".
-	Scenes can be saved to disk, and then instanced into other scenes. This allows for very high flexibility in the architecture and data model of the projects. 
+	Scenes can be saved to disk, and then instanced into other scenes. This allows for very high flexibility in the architecture and data model of the projects.
 	[SceneMainLoop] contains the "active" tree of nodes, and a node becomes active (receinving NOTIFICATION_ENTER_SCENE) when added to that tree.
 	A node can contain any number of nodes as a children (but there is only one tree root) with the requirement that no two childrens with the same name can exist.
 	Nodes can, optionally, be added to groups. This makes it easy to reach a number of nodes from the code (for example an "enemies" group).
@@ -18684,11 +18724,11 @@
 	<description>
 	Operating System functions. OS Wraps the most common functionality to communicate with the host Operating System, such as:
  -Mouse Grabbing
- -Mouse Cursors 
+ -Mouse Cursors
  -Clipboard
  -Video Mode
  -Date " Time
- -Timers 
+ -Timers
  -Environment Variables
  -Execution of Binaries
  -Command Line
@@ -27567,7 +27607,7 @@
 	Rigid body 2D node.
 	</brief_description>
 	<description>
-	Rigid body 2D node. This node is used for placing rigid bodies in the scene. It can contain a number of shapes, and also shift state between regular Rigid Body to Character or even Static. 
+	Rigid body 2D node. This node is used for placing rigid bodies in the scene. It can contain a number of shapes, and also shift state between regular Rigid Body to Character or even Static.
 	Character mode forbids the node from being rotated. This node can have a custom force integrator function, for writing complex physics motion behavior per node.
 	As a warning, don't change this node position every frame or very often. Sporadic changes work fine, but physics runs at a different granularity (fixed hz) than usual rendering (process callback) and maybe even in a separate thread, so changing this from a process loop will yield strange behavior.
 	</description>
@@ -37719,6 +37759,18 @@
 			<description>
 			</description>
 		</method>
+		<method name="set_audio_track">
+			<argument index="0" name="track" type="int">
+			</argument>
+			<description>
+			</description>
+		</method>
+		<method name="get_audio_track" qualifiers="const">
+			<return type="int">
+			</return>
+			<description>
+			</description>
+		</method>
 		<method name="get_stream_name" qualifiers="const">
 			<return type="String">
 			</return>
@@ -37755,6 +37807,18 @@
 			<description>
 			</description>
 		</method>
+		<method name="set_buffering_msec">
+			<argument index="0" name="msec" type="int">
+			</argument>
+			<description>
+			</description>
+		</method>
+		<method name="get_buffering_msec" qualifiers="const">
+			<return type="int">
+			</return>
+			<description>
+			</description>
+		</method>
 	</methods>
 	<constants>
 	</constants>
@@ -37765,30 +37829,16 @@
 	<description>
 	</description>
 	<methods>
-		<method name="get_pending_frame_count" qualifiers="const">
-			<return type="int">
-			</return>
-			<description>
-			</description>
-		</method>
-		<method name="pop_frame">
-			<argument index="0" name="arg0" type="Object">
-			</argument>
-			<description>
-			</description>
-		</method>
-		<method name="peek_frame" qualifiers="const">
-			<return type="Image">
-			</return>
-			<description>
-			</description>
-		</method>
-		<method name="set_audio_track">
-			<argument index="0" name="idx" type="int">
-			</argument>
-			<description>
-			</description>
-		</method>
+	</methods>
+	<constants>
+	</constants>
+</class>
+<class name="VideoStreamTheora" inherits="VideoStream" category="Core">
+	<brief_description>
+	</brief_description>
+	<description>
+	</description>
+	<methods>
 	</methods>
 	<constants>
 	</constants>
diff --git a/drivers/SCsub b/drivers/SCsub
index 7b96052f0b..bc46bf2cec 100644
--- a/drivers/SCsub
+++ b/drivers/SCsub
@@ -40,8 +40,8 @@ if (env["opus"]=="yes"):
 if (env["tools"]=="yes"):
 	SConscript("convex_decomp/SCsub");
 
-if env["theora"]=="yes":
-	SConscript("theoraplayer/SCsub")
+#if env["theora"]=="yes":
+#	SConscript("theoraplayer/SCsub")
 if (env["theora"]=="yes"):
 	SConscript("theora/SCsub");
 if (env['speex']=='yes'):
diff --git a/drivers/etc1/rg_etc1.cpp b/drivers/etc1/rg_etc1.cpp
index fd109f003c..47dcb57e6b 100644
--- a/drivers/etc1/rg_etc1.cpp
+++ b/drivers/etc1/rg_etc1.cpp
@@ -1,2454 +1,2454 @@
-// File: rg_etc1.cpp - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>

-// Please see ZLIB license at the end of rg_etc1.h.

-//

-// For more information Ericsson Texture Compression (ETC/ETC1), see:

-// http://www.khronos.org/registry/gles/extensions/OES/OES_compressed_ETC1_RGB8_texture.txt

-//

-// v1.03 - 5/12/13 - Initial public release

-#include "rg_etc1.h"

-

-#include <stdlib.h>

-#include <string.h>

-#include <assert.h>

-//#include <stdio.h>

-#include <math.h>

-#include <stdio.h>

-#pragma warning (disable: 4201) //  nonstandard extension used : nameless struct/union

-

-#if defined(_DEBUG) || defined(DEBUG)

-#define RG_ETC1_BUILD_DEBUG

-#endif

-

-#define RG_ETC1_ASSERT assert

-

-namespace rg_etc1

-{

-

-   inline long labs(long val) {

-        return val < 0 ? -val : val;

-   }

-

-   inline int intabs(int val) {

-

-       return val<0?-val:val;

-   }

-

-   typedef unsigned char uint8;

-   typedef unsigned short uint16;

-   typedef unsigned int uint;

-   typedef unsigned int uint32;

-   typedef long long int64;

-   typedef unsigned long long uint64;

-

-   const uint32 cUINT32_MAX = 0xFFFFFFFFU;

-   const uint64 cUINT64_MAX = 0xFFFFFFFFFFFFFFFFULL; //0xFFFFFFFFFFFFFFFFui64;

-   

-   template<typename T> inline T minimum(T a, T b) { return (a < b) ? a : b; }

-   template<typename T> inline T minimum(T a, T b, T c) { return minimum(minimum(a, b), c); }

-   template<typename T> inline T maximum(T a, T b) { return (a > b) ? a : b; }

-   template<typename T> inline T maximum(T a, T b, T c) { return maximum(maximum(a, b), c); }

-   template<typename T> inline T clamp(T value, T low, T high) { return (value < low) ? low : ((value > high) ? high : value); }

-   template<typename T> inline T square(T value) { return value * value; }

-   template<typename T> inline void zero_object(T& obj) { memset((void*)&obj, 0, sizeof(obj)); }

-   template<typename T> inline void zero_this(T* pObj) { memset((void*)pObj, 0, sizeof(*pObj)); }

-

-   template<class T, size_t N> T decay_array_to_subtype(T (&a)[N]);   

-

-#define RG_ETC1_ARRAY_SIZE(X) (sizeof(X) / sizeof(decay_array_to_subtype(X)))

-

-   enum eNoClamp { cNoClamp };

-

-   struct color_quad_u8

-   {

-      static inline int clamp(int v) { if (v & 0xFFFFFF00U) v = (~(static_cast<int>(v) >> 31)) & 0xFF; return v; }

-

-      struct component_traits { enum { cSigned = false, cFloat = false, cMin = 0U, cMax = 255U }; };

-

-   public:

-      typedef unsigned char component_t;

-      typedef int parameter_t;

-

-      enum { cNumComps = 4 };

-

-      union

-      {

-         struct

-         {

-            component_t r;

-            component_t g;

-            component_t b;

-            component_t a;

-         };

-

-         component_t c[cNumComps];

-

-         uint32 m_u32;

-      };

-

-      inline color_quad_u8()

-      {

-      }

-

-      inline color_quad_u8(const color_quad_u8& other) : m_u32(other.m_u32)

-      {

-      }

-

-      explicit inline color_quad_u8(parameter_t y, parameter_t alpha = component_traits::cMax)

-      {

-         set(y, alpha);

-      }

-

-      inline color_quad_u8(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)

-      {

-         set(red, green, blue, alpha);

-      }

-

-      explicit inline color_quad_u8(eNoClamp, parameter_t y, parameter_t alpha = component_traits::cMax)

-      {

-         set_noclamp_y_alpha(y, alpha);

-      }

-

-      inline color_quad_u8(eNoClamp, parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)

-      {

-         set_noclamp_rgba(red, green, blue, alpha);

-      }

-

-      inline void clear()

-      {

-         m_u32 = 0;

-      }

-

-      inline color_quad_u8& operator= (const color_quad_u8& other)

-      {

-         m_u32 = other.m_u32;

-         return *this;

-      }

-

-      inline color_quad_u8& set_rgb(const color_quad_u8& other)

-      {

-         r = other.r;

-         g = other.g;

-         b = other.b;

-         return *this;

-      }

-

-      inline color_quad_u8& operator= (parameter_t y)

-      {

-         set(y, component_traits::cMax);

-         return *this;

-      }

-

-      inline color_quad_u8& set(parameter_t y, parameter_t alpha = component_traits::cMax)

-      {

-         y = clamp(y);

-         alpha = clamp(alpha);

-         r = static_cast<component_t>(y);

-         g = static_cast<component_t>(y);

-         b = static_cast<component_t>(y);

-         a = static_cast<component_t>(alpha);

-         return *this;

-      }

-

-      inline color_quad_u8& set_noclamp_y_alpha(parameter_t y, parameter_t alpha = component_traits::cMax)

-      {

-         RG_ETC1_ASSERT( (y >= component_traits::cMin) && (y <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (alpha >= component_traits::cMin) && (alpha <= component_traits::cMax) );

-

-         r = static_cast<component_t>(y);

-         g = static_cast<component_t>(y);

-         b = static_cast<component_t>(y);

-         a = static_cast<component_t>(alpha);

-         return *this;

-      }

-

-      inline color_quad_u8& set(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)

-      {

-         r = static_cast<component_t>(clamp(red));

-         g = static_cast<component_t>(clamp(green));

-         b = static_cast<component_t>(clamp(blue));

-         a = static_cast<component_t>(clamp(alpha));

-         return *this;

-      }

-

-      inline color_quad_u8& set_noclamp_rgba(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha)

-      {

-         RG_ETC1_ASSERT( (red >= component_traits::cMin) && (red <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (green >= component_traits::cMin) && (green <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (blue >= component_traits::cMin) && (blue <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (alpha >= component_traits::cMin) && (alpha <= component_traits::cMax) );

-

-         r = static_cast<component_t>(red);

-         g = static_cast<component_t>(green);

-         b = static_cast<component_t>(blue);

-         a = static_cast<component_t>(alpha);

-         return *this;

-      }

-

-      inline color_quad_u8& set_noclamp_rgb(parameter_t red, parameter_t green, parameter_t blue)

-      {

-         RG_ETC1_ASSERT( (red >= component_traits::cMin) && (red <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (green >= component_traits::cMin) && (green <= component_traits::cMax) );

-         RG_ETC1_ASSERT( (blue >= component_traits::cMin) && (blue <= component_traits::cMax) );

-

-         r = static_cast<component_t>(red);

-         g = static_cast<component_t>(green);

-         b = static_cast<component_t>(blue);

-         return *this;

-      }

-

-      static inline parameter_t get_min_comp() { return component_traits::cMin; }

-      static inline parameter_t get_max_comp() { return component_traits::cMax; }

-      static inline bool get_comps_are_signed() { return component_traits::cSigned; }

-

-      inline component_t operator[] (uint i) const { RG_ETC1_ASSERT(i < cNumComps); return c[i]; }

-      inline component_t& operator[] (uint i) { RG_ETC1_ASSERT(i < cNumComps); return c[i]; }

-

-      inline color_quad_u8& set_component(uint i, parameter_t f)

-      {

-         RG_ETC1_ASSERT(i < cNumComps);

-

-         c[i] = static_cast<component_t>(clamp(f));

-

-         return *this;

-      }

-

-      inline color_quad_u8& set_grayscale(parameter_t l)

-      {

-         component_t x = static_cast<component_t>(clamp(l));

-         c[0] = x;

-         c[1] = x;

-         c[2] = x;

-         return *this;

-      }

-

-      inline color_quad_u8& clamp(const color_quad_u8& l, const color_quad_u8& h)

-      {

-         for (uint i = 0; i < cNumComps; i++)

-            c[i] = static_cast<component_t>(rg_etc1::clamp<parameter_t>(c[i], l[i], h[i]));

-         return *this;

-      }

-

-      inline color_quad_u8& clamp(parameter_t l, parameter_t h)

-      {

-         for (uint i = 0; i < cNumComps; i++)

-            c[i] = static_cast<component_t>(rg_etc1::clamp<parameter_t>(c[i], l, h));

-         return *this;

-      }

-

-      // Returns CCIR 601 luma (consistent with color_utils::RGB_To_Y).

-      inline parameter_t get_luma() const

-      {

-         return static_cast<parameter_t>((19595U * r + 38470U * g + 7471U * b + 32768U) >> 16U);

-      }

-

-      // Returns REC 709 luma.

-      inline parameter_t get_luma_rec709() const

-      {

-         return static_cast<parameter_t>((13938U * r + 46869U * g + 4729U * b + 32768U) >> 16U);

-      }

-

-      inline uint squared_distance_rgb(const color_quad_u8& c) const

-      {

-         return rg_etc1::square(r - c.r) + rg_etc1::square(g - c.g) + rg_etc1::square(b - c.b);

-      }

-

-      inline uint squared_distance_rgba(const color_quad_u8& c) const

-      {

-         return rg_etc1::square(r - c.r) + rg_etc1::square(g - c.g) + rg_etc1::square(b - c.b) + rg_etc1::square(a - c.a);

-      }

-

-      inline bool rgb_equals(const color_quad_u8& rhs) const

-      {

-         return (r == rhs.r) && (g == rhs.g) && (b == rhs.b);

-      }

-

-      inline bool operator== (const color_quad_u8& rhs) const

-      {

-         return m_u32 == rhs.m_u32;

-      }

-

-      color_quad_u8& operator+= (const color_quad_u8& other)

-      {

-         for (uint i = 0; i < 4; i++)

-            c[i] = static_cast<component_t>(clamp(c[i] + other.c[i]));

-         return *this;

-      }

-

-      color_quad_u8& operator-= (const color_quad_u8& other)

-      {

-         for (uint i = 0; i < 4; i++)

-            c[i] = static_cast<component_t>(clamp(c[i] - other.c[i]));

-         return *this;

-      }

-

-      friend color_quad_u8 operator+ (const color_quad_u8& lhs, const color_quad_u8& rhs)

-      {

-         color_quad_u8 result(lhs);

-         result += rhs;

-         return result;

-      }

-

-      friend color_quad_u8 operator- (const color_quad_u8& lhs, const color_quad_u8& rhs)

-      {

-         color_quad_u8 result(lhs);

-         result -= rhs;

-         return result;

-      }

-   }; // class color_quad_u8

-

-   struct vec3F

-   {

-      float m_s[3];

-      

-      inline vec3F() { }

-      inline vec3F(float s) { m_s[0] = s; m_s[1] = s; m_s[2] = s; }

-      inline vec3F(float x, float y, float z) { m_s[0] = x; m_s[1] = y; m_s[2] = z; }

-      

-      inline float operator[] (uint i) const { RG_ETC1_ASSERT(i < 3); return m_s[i]; }

-

-      inline vec3F& operator += (const vec3F& other) { for (uint i = 0; i < 3; i++) m_s[i] += other.m_s[i]; return *this; }

-

-      inline vec3F& operator *= (float s) { for (uint i = 0; i < 3; i++) m_s[i] *= s; return *this; }

-   };

-     

-   enum etc_constants

-   {

-      cETC1BytesPerBlock = 8U,

-

-      cETC1SelectorBits = 2U,

-      cETC1SelectorValues = 1U << cETC1SelectorBits,

-      cETC1SelectorMask = cETC1SelectorValues - 1U,

-

-      cETC1BlockShift = 2U,

-      cETC1BlockSize = 1U << cETC1BlockShift,

-

-      cETC1LSBSelectorIndicesBitOffset = 0,

-      cETC1MSBSelectorIndicesBitOffset = 16,

-

-      cETC1FlipBitOffset = 32,

-      cETC1DiffBitOffset = 33,

-

-      cETC1IntenModifierNumBits = 3,

-      cETC1IntenModifierValues = 1 << cETC1IntenModifierNumBits,

-      cETC1RightIntenModifierTableBitOffset = 34,

-      cETC1LeftIntenModifierTableBitOffset = 37,

-

-      // Base+Delta encoding (5 bit bases, 3 bit delta)

-      cETC1BaseColorCompNumBits = 5,

-      cETC1BaseColorCompMax = 1 << cETC1BaseColorCompNumBits,

-

-      cETC1DeltaColorCompNumBits = 3,

-      cETC1DeltaColorComp = 1 << cETC1DeltaColorCompNumBits,

-      cETC1DeltaColorCompMax = 1 << cETC1DeltaColorCompNumBits,

-

-      cETC1BaseColor5RBitOffset = 59,

-      cETC1BaseColor5GBitOffset = 51,

-      cETC1BaseColor5BBitOffset = 43,

-

-      cETC1DeltaColor3RBitOffset = 56,

-      cETC1DeltaColor3GBitOffset = 48,

-      cETC1DeltaColor3BBitOffset = 40,

-

-      // Absolute (non-delta) encoding (two 4-bit per component bases)

-      cETC1AbsColorCompNumBits = 4,

-      cETC1AbsColorCompMax = 1 << cETC1AbsColorCompNumBits,

-

-      cETC1AbsColor4R1BitOffset = 60,

-      cETC1AbsColor4G1BitOffset = 52,

-      cETC1AbsColor4B1BitOffset = 44,

-

-      cETC1AbsColor4R2BitOffset = 56,

-      cETC1AbsColor4G2BitOffset = 48,

-      cETC1AbsColor4B2BitOffset = 40,

-

-      cETC1ColorDeltaMin = -4,

-      cETC1ColorDeltaMax = 3,

-

-      // Delta3:

-      // 0   1   2   3   4   5   6   7

-      // 000 001 010 011 100 101 110 111

-      // 0   1   2   3   -4  -3  -2  -1

-   };

-   

-   static uint8 g_quant5_tab[256+16];

-

-

-   static const int g_etc1_inten_tables[cETC1IntenModifierValues][cETC1SelectorValues] = 

-   { 

-      { -8,  -2,   2,   8 }, { -17,  -5,  5,  17 }, { -29,  -9,   9,  29 }, {  -42, -13, 13,  42 }, 

-      { -60, -18, 18,  60 }, { -80, -24, 24,  80 }, { -106, -33, 33, 106 }, { -183, -47, 47, 183 } 

-   };

-

-   static const uint8 g_etc1_to_selector_index[cETC1SelectorValues] = { 2, 3, 1, 0 };

-   static const uint8 g_selector_index_to_etc1[cETC1SelectorValues] = { 3, 2, 0, 1 };

-      

-   // Given an ETC1 diff/inten_table/selector, and an 8-bit desired color, this table encodes the best packed_color in the low byte, and the abs error in the high byte.

-   static uint16 g_etc1_inverse_lookup[2*8*4][256];      // [diff/inten_table/selector][desired_color]

-

-   // g_color8_to_etc_block_config[color][table_index] = Supplies for each 8-bit color value a list of packed ETC1 diff/intensity table/selectors/packed_colors that map to that color.

-   // To pack: diff | (inten << 1) | (selector << 4) | (packed_c << 8)

-   static const uint16 g_color8_to_etc_block_config_0_255[2][33] =

-   {

-      { 0x0000,  0x0010,  0x0002,  0x0012,  0x0004,  0x0014,  0x0006,  0x0016,  0x0008,  0x0018,  0x000A,  0x001A,  0x000C,  0x001C,  0x000E,  0x001E,

-        0x0001,  0x0011,  0x0003,  0x0013,  0x0005,  0x0015,  0x0007,  0x0017,  0x0009,  0x0019,  0x000B,  0x001B,  0x000D,  0x001D,  0x000F,  0x001F, 0xFFFF },

-      { 0x0F20,  0x0F30,  0x0E32,  0x0F22,  0x0E34,  0x0F24,  0x0D36,  0x0F26,  0x0C38,  0x0E28,  0x0B3A,  0x0E2A,  0x093C,  0x0E2C,  0x053E,  0x0D2E,

-        0x1E31,  0x1F21,  0x1D33,  0x1F23,  0x1C35,  0x1E25,  0x1A37,  0x1E27,  0x1839,  0x1D29,  0x163B,  0x1C2B,  0x133D,  0x1B2D,  0x093F,  0x1A2F, 0xFFFF },

-   };

-

-   // Really only [254][11].

-   static const uint16 g_color8_to_etc_block_config_1_to_254[254][12] = 

-   {

-      { 0x021C, 0x0D0D, 0xFFFF }, { 0x0020, 0x0021, 0x0A0B, 0x061F, 0xFFFF }, { 0x0113, 0x0217, 0xFFFF }, { 0x0116, 0x031E,

-      0x0B0E, 0x0405, 0xFFFF }, { 0x0022, 0x0204, 0x050A, 0x0023, 0xFFFF }, { 0x0111, 0x0319, 0x0809, 0x170F, 0xFFFF }, {

-      0x0303, 0x0215, 0x0607, 0xFFFF }, { 0x0030, 0x0114, 0x0408, 0x0031, 0x0201, 0x051D, 0xFFFF }, { 0x0100, 0x0024, 0x0306,

-      0x0025, 0x041B, 0x0E0D, 0xFFFF }, { 0x021A, 0x0121, 0x0B0B, 0x071F, 0xFFFF }, { 0x0213, 0x0317, 0xFFFF }, { 0x0112,

-      0x0505, 0xFFFF }, { 0x0026, 0x070C, 0x0123, 0x0027, 0xFFFF }, { 0x0211, 0x0909, 0xFFFF }, { 0x0110, 0x0315, 0x0707,

-      0x0419, 0x180F, 0xFFFF }, { 0x0218, 0x0131, 0x0301, 0x0403, 0x061D, 0xFFFF }, { 0x0032, 0x0202, 0x0033, 0x0125, 0x051B,

-      0x0F0D, 0xFFFF }, { 0x0028, 0x031C, 0x0221, 0x0029, 0xFFFF }, { 0x0120, 0x0313, 0x0C0B, 0x081F, 0xFFFF }, { 0x0605,

-      0x0417, 0xFFFF }, { 0x0216, 0x041E, 0x0C0E, 0x0223, 0x0127, 0xFFFF }, { 0x0122, 0x0304, 0x060A, 0x0311, 0x0A09, 0xFFFF

-      }, { 0x0519, 0x190F, 0xFFFF }, { 0x002A, 0x0231, 0x0503, 0x0415, 0x0807, 0x002B, 0x071D, 0xFFFF }, { 0x0130, 0x0214,

-      0x0508, 0x0401, 0x0133, 0x0225, 0x061B, 0xFFFF }, { 0x0200, 0x0124, 0x0406, 0x0321, 0x0129, 0x100D, 0xFFFF }, { 0x031A,

-      0x0D0B, 0x091F, 0xFFFF }, { 0x0413, 0x0705, 0x0517, 0xFFFF }, { 0x0212, 0x0034, 0x0323, 0x0035, 0x0227, 0xFFFF }, {

-      0x0126, 0x080C, 0x0B09, 0xFFFF }, { 0x0411, 0x0619, 0x1A0F, 0xFFFF }, { 0x0210, 0x0331, 0x0603, 0x0515, 0x0907, 0x012B,

-      0xFFFF }, { 0x0318, 0x002C, 0x0501, 0x0233, 0x0325, 0x071B, 0x002D, 0x081D, 0xFFFF }, { 0x0132, 0x0302, 0x0229, 0x110D,

-      0xFFFF }, { 0x0128, 0x041C, 0x0421, 0x0E0B, 0x0A1F, 0xFFFF }, { 0x0220, 0x0513, 0x0617, 0xFFFF }, { 0x0135, 0x0805,

-      0x0327, 0xFFFF }, { 0x0316, 0x051E, 0x0D0E, 0x0423, 0xFFFF }, { 0x0222, 0x0404, 0x070A, 0x0511, 0x0719, 0x0C09, 0x1B0F,

-      0xFFFF }, { 0x0703, 0x0615, 0x0A07, 0x022B, 0xFFFF }, { 0x012A, 0x0431, 0x0601, 0x0333, 0x012D, 0x091D, 0xFFFF }, {

-      0x0230, 0x0314, 0x0036, 0x0608, 0x0425, 0x0037, 0x0329, 0x081B, 0x120D, 0xFFFF }, { 0x0300, 0x0224, 0x0506, 0x0521,

-      0x0F0B, 0x0B1F, 0xFFFF }, { 0x041A, 0x0613, 0x0717, 0xFFFF }, { 0x0235, 0x0905, 0xFFFF }, { 0x0312, 0x0134, 0x0523,

-      0x0427, 0xFFFF }, { 0x0226, 0x090C, 0x002E, 0x0611, 0x0D09, 0x002F, 0xFFFF }, { 0x0715, 0x0B07, 0x0819, 0x032B, 0x1C0F,

-      0xFFFF }, { 0x0310, 0x0531, 0x0701, 0x0803, 0x022D, 0x0A1D, 0xFFFF }, { 0x0418, 0x012C, 0x0433, 0x0525, 0x0137, 0x091B,

-      0x130D, 0xFFFF }, { 0x0232, 0x0402, 0x0621, 0x0429, 0xFFFF }, { 0x0228, 0x051C, 0x0713, 0x100B, 0x0C1F, 0xFFFF }, {

-      0x0320, 0x0335, 0x0A05, 0x0817, 0xFFFF }, { 0x0623, 0x0527, 0xFFFF }, { 0x0416, 0x061E, 0x0E0E, 0x0711, 0x0E09, 0x012F,

-      0xFFFF }, { 0x0322, 0x0504, 0x080A, 0x0919, 0x1D0F, 0xFFFF }, { 0x0631, 0x0903, 0x0815, 0x0C07, 0x042B, 0x032D, 0x0B1D,

-      0xFFFF }, { 0x022A, 0x0801, 0x0533, 0x0625, 0x0237, 0x0A1B, 0xFFFF }, { 0x0330, 0x0414, 0x0136, 0x0708, 0x0721, 0x0529,

-      0x140D, 0xFFFF }, { 0x0400, 0x0324, 0x0606, 0x0038, 0x0039, 0x110B, 0x0D1F, 0xFFFF }, { 0x051A, 0x0813, 0x0B05, 0x0917,

-      0xFFFF }, { 0x0723, 0x0435, 0x0627, 0xFFFF }, { 0x0412, 0x0234, 0x0F09, 0x022F, 0xFFFF }, { 0x0326, 0x0A0C, 0x012E,

-      0x0811, 0x0A19, 0x1E0F, 0xFFFF }, { 0x0731, 0x0A03, 0x0915, 0x0D07, 0x052B, 0xFFFF }, { 0x0410, 0x0901, 0x0633, 0x0725,

-      0x0337, 0x0B1B, 0x042D, 0x0C1D, 0xFFFF }, { 0x0518, 0x022C, 0x0629, 0x150D, 0xFFFF }, { 0x0332, 0x0502, 0x0821, 0x0139,

-      0x120B, 0x0E1F, 0xFFFF }, { 0x0328, 0x061C, 0x0913, 0x0A17, 0xFFFF }, { 0x0420, 0x0535, 0x0C05, 0x0727, 0xFFFF }, {

-      0x0823, 0x032F, 0xFFFF }, { 0x0516, 0x071E, 0x0F0E, 0x0911, 0x0B19, 0x1009, 0x1F0F, 0xFFFF }, { 0x0422, 0x0604, 0x090A,

-      0x0B03, 0x0A15, 0x0E07, 0x062B, 0xFFFF }, { 0x0831, 0x0A01, 0x0733, 0x052D, 0x0D1D, 0xFFFF }, { 0x032A, 0x0825, 0x0437,

-      0x0729, 0x0C1B, 0x160D, 0xFFFF }, { 0x0430, 0x0514, 0x0236, 0x0808, 0x0921, 0x0239, 0x130B, 0x0F1F, 0xFFFF }, { 0x0500,

-      0x0424, 0x0706, 0x0138, 0x0A13, 0x0B17, 0xFFFF }, { 0x061A, 0x0635, 0x0D05, 0xFFFF }, { 0x0923, 0x0827, 0xFFFF }, {

-      0x0512, 0x0334, 0x003A, 0x0A11, 0x1109, 0x003B, 0x042F, 0xFFFF }, { 0x0426, 0x0B0C, 0x022E, 0x0B15, 0x0F07, 0x0C19,

-      0x072B, 0xFFFF }, { 0x0931, 0x0B01, 0x0C03, 0x062D, 0x0E1D, 0xFFFF }, { 0x0510, 0x0833, 0x0925, 0x0537, 0x0D1B, 0x170D,

-      0xFFFF }, { 0x0618, 0x032C, 0x0A21, 0x0339, 0x0829, 0xFFFF }, { 0x0432, 0x0602, 0x0B13, 0x140B, 0x101F, 0xFFFF }, {

-      0x0428, 0x071C, 0x0735, 0x0E05, 0x0C17, 0xFFFF }, { 0x0520, 0x0A23, 0x0927, 0xFFFF }, { 0x0B11, 0x1209, 0x013B, 0x052F,

-      0xFFFF }, { 0x0616, 0x081E, 0x0D19, 0xFFFF }, { 0x0522, 0x0704, 0x0A0A, 0x0A31, 0x0D03, 0x0C15, 0x1007, 0x082B, 0x072D,

-      0x0F1D, 0xFFFF }, { 0x0C01, 0x0933, 0x0A25, 0x0637, 0x0E1B, 0xFFFF }, { 0x042A, 0x0B21, 0x0929, 0x180D, 0xFFFF }, {

-	      0x0530, 0x0614, 0x0336, 0x0908, 0x0439, 0x150B, 0x111F, 0xFFFF }, { 0x0600, 0x0524, 0x0806, 0x0238, 0x0C13, 0x0F05,

-      0x0D17, 0xFFFF }, { 0x071A, 0x0B23, 0x0835, 0x0A27, 0xFFFF }, { 0x1309, 0x023B, 0x062F, 0xFFFF }, { 0x0612, 0x0434,

-      0x013A, 0x0C11, 0x0E19, 0xFFFF }, { 0x0526, 0x0C0C, 0x032E, 0x0B31, 0x0E03, 0x0D15, 0x1107, 0x092B, 0xFFFF }, { 0x0D01,

-      0x0A33, 0x0B25, 0x0737, 0x0F1B, 0x082D, 0x101D, 0xFFFF }, { 0x0610, 0x0A29, 0x190D, 0xFFFF }, { 0x0718, 0x042C, 0x0C21,

-      0x0539, 0x160B, 0x121F, 0xFFFF }, { 0x0532, 0x0702, 0x0D13, 0x0E17, 0xFFFF }, { 0x0528, 0x081C, 0x0935, 0x1005, 0x0B27,

-      0xFFFF }, { 0x0620, 0x0C23, 0x033B, 0x072F, 0xFFFF }, { 0x0D11, 0x0F19, 0x1409, 0xFFFF }, { 0x0716, 0x003C, 0x091E,

-      0x0F03, 0x0E15, 0x1207, 0x0A2B, 0x003D, 0xFFFF }, { 0x0622, 0x0804, 0x0B0A, 0x0C31, 0x0E01, 0x0B33, 0x092D, 0x111D,

-      0xFFFF }, { 0x0C25, 0x0837, 0x0B29, 0x101B, 0x1A0D, 0xFFFF }, { 0x052A, 0x0D21, 0x0639, 0x170B, 0x131F, 0xFFFF }, {

-      0x0630, 0x0714, 0x0436, 0x0A08, 0x0E13, 0x0F17, 0xFFFF }, { 0x0700, 0x0624, 0x0906, 0x0338, 0x0A35, 0x1105, 0xFFFF }, {

-      0x081A, 0x0D23, 0x0C27, 0xFFFF }, { 0x0E11, 0x1509, 0x043B, 0x082F, 0xFFFF }, { 0x0712, 0x0534, 0x023A, 0x0F15, 0x1307,

-      0x1019, 0x0B2B, 0x013D, 0xFFFF }, { 0x0626, 0x0D0C, 0x042E, 0x0D31, 0x0F01, 0x1003, 0x0A2D, 0x121D, 0xFFFF }, { 0x0C33,

-      0x0D25, 0x0937, 0x111B, 0x1B0D, 0xFFFF }, { 0x0710, 0x0E21, 0x0739, 0x0C29, 0xFFFF }, { 0x0818, 0x052C, 0x0F13, 0x180B,

-      0x141F, 0xFFFF }, { 0x0632, 0x0802, 0x0B35, 0x1205, 0x1017, 0xFFFF }, { 0x0628, 0x091C, 0x0E23, 0x0D27, 0xFFFF }, {

-      0x0720, 0x0F11, 0x1609, 0x053B, 0x092F, 0xFFFF }, { 0x1119, 0x023D, 0xFFFF }, { 0x0816, 0x013C, 0x0A1E, 0x0E31, 0x1103,

-      0x1015, 0x1407, 0x0C2B, 0x0B2D, 0x131D, 0xFFFF }, { 0x0722, 0x0904, 0x0C0A, 0x1001, 0x0D33, 0x0E25, 0x0A37, 0x121B,

-      0xFFFF }, { 0x0F21, 0x0D29, 0x1C0D, 0xFFFF }, { 0x062A, 0x0839, 0x190B, 0x151F, 0xFFFF }, { 0x0730, 0x0814, 0x0536,

-      0x0B08, 0x1013, 0x1305, 0x1117, 0xFFFF }, { 0x0800, 0x0724, 0x0A06, 0x0438, 0x0F23, 0x0C35, 0x0E27, 0xFFFF }, { 0x091A,

-      0x1709, 0x063B, 0x0A2F, 0xFFFF }, { 0x1011, 0x1219, 0x033D, 0xFFFF }, { 0x0812, 0x0634, 0x033A, 0x0F31, 0x1203, 0x1115,

-      0x1507, 0x0D2B, 0xFFFF }, { 0x0726, 0x0E0C, 0x052E, 0x1101, 0x0E33, 0x0F25, 0x0B37, 0x131B, 0x0C2D, 0x141D, 0xFFFF }, {

-      0x0E29, 0x1D0D, 0xFFFF }, { 0x0810, 0x1021, 0x0939, 0x1A0B, 0x161F, 0xFFFF }, { 0x0918, 0x062C, 0x1113, 0x1217, 0xFFFF

-      }, { 0x0732, 0x0902, 0x0D35, 0x1405, 0x0F27, 0xFFFF }, { 0x0728, 0x0A1C, 0x1023, 0x073B, 0x0B2F, 0xFFFF }, { 0x0820,

-      0x1111, 0x1319, 0x1809, 0xFFFF }, { 0x1303, 0x1215, 0x1607, 0x0E2B, 0x043D, 0xFFFF }, { 0x0916, 0x023C, 0x0B1E, 0x1031,

-      0x1201, 0x0F33, 0x0D2D, 0x151D, 0xFFFF }, { 0x0822, 0x0A04, 0x0D0A, 0x1025, 0x0C37, 0x0F29, 0x141B, 0x1E0D, 0xFFFF }, {

-      0x1121, 0x0A39, 0x1B0B, 0x171F, 0xFFFF }, { 0x072A, 0x1213, 0x1317, 0xFFFF }, { 0x0830, 0x0914, 0x0636, 0x0C08, 0x0E35,

-      0x1505, 0xFFFF }, { 0x0900, 0x0824, 0x0B06, 0x0538, 0x1123, 0x1027, 0xFFFF }, { 0x0A1A, 0x1211, 0x1909, 0x083B, 0x0C2F,

-      0xFFFF }, { 0x1315, 0x1707, 0x1419, 0x0F2B, 0x053D, 0xFFFF }, { 0x0912, 0x0734, 0x043A, 0x1131, 0x1301, 0x1403, 0x0E2D,

-      0x161D, 0xFFFF }, { 0x0826, 0x0F0C, 0x062E, 0x1033, 0x1125, 0x0D37, 0x151B, 0x1F0D, 0xFFFF }, { 0x1221, 0x0B39, 0x1029,

-      0xFFFF }, { 0x0910, 0x1313, 0x1C0B, 0x181F, 0xFFFF }, { 0x0A18, 0x072C, 0x0F35, 0x1605, 0x1417, 0xFFFF }, { 0x0832,

-      0x0A02, 0x1223, 0x1127, 0xFFFF }, { 0x0828, 0x0B1C, 0x1311, 0x1A09, 0x093B, 0x0D2F, 0xFFFF }, { 0x0920, 0x1519, 0x063D,

-      0xFFFF }, { 0x1231, 0x1503, 0x1415, 0x1807, 0x102B, 0x0F2D, 0x171D, 0xFFFF }, { 0x0A16, 0x033C, 0x0C1E, 0x1401, 0x1133,

-      0x1225, 0x0E37, 0x161B, 0xFFFF }, { 0x0922, 0x0B04, 0x0E0A, 0x1321, 0x1129, 0xFFFF }, { 0x0C39, 0x1D0B, 0x191F, 0xFFFF

-      }, { 0x082A, 0x1413, 0x1705, 0x1517, 0xFFFF }, { 0x0930, 0x0A14, 0x0736, 0x0D08, 0x1323, 0x1035, 0x1227, 0xFFFF }, {

-      0x0A00, 0x0924, 0x0C06, 0x0638, 0x1B09, 0x0A3B, 0x0E2F, 0xFFFF }, { 0x0B1A, 0x1411, 0x1619, 0x073D, 0xFFFF }, { 0x1331,

-      0x1603, 0x1515, 0x1907, 0x112B, 0xFFFF }, { 0x0A12, 0x0834, 0x053A, 0x1501, 0x1233, 0x1325, 0x0F37, 0x171B, 0x102D,

-      0x181D, 0xFFFF }, { 0x0926, 0x072E, 0x1229, 0xFFFF }, { 0x1421, 0x0D39, 0x1E0B, 0x1A1F, 0xFFFF }, { 0x0A10, 0x1513,

-      0x1617, 0xFFFF }, { 0x0B18, 0x082C, 0x1135, 0x1805, 0x1327, 0xFFFF }, { 0x0932, 0x0B02, 0x1423, 0x0B3B, 0x0F2F, 0xFFFF

-      }, { 0x0928, 0x0C1C, 0x1511, 0x1719, 0x1C09, 0xFFFF }, { 0x0A20, 0x1703, 0x1615, 0x1A07, 0x122B, 0x083D, 0xFFFF }, {

-      0x1431, 0x1601, 0x1333, 0x112D, 0x191D, 0xFFFF }, { 0x0B16, 0x043C, 0x0D1E, 0x1425, 0x1037, 0x1329, 0x181B, 0xFFFF }, {

-      0x0A22, 0x0C04, 0x0F0A, 0x1521, 0x0E39, 0x1F0B, 0x1B1F, 0xFFFF }, { 0x1613, 0x1717, 0xFFFF }, { 0x092A, 0x1235, 0x1905,

-      0xFFFF }, { 0x0A30, 0x0B14, 0x0836, 0x0E08, 0x1523, 0x1427, 0xFFFF }, { 0x0B00, 0x0A24, 0x0D06, 0x0738, 0x1611, 0x1D09,

-      0x0C3B, 0x102F, 0xFFFF }, { 0x0C1A, 0x1715, 0x1B07, 0x1819, 0x132B, 0x093D, 0xFFFF }, { 0x1531, 0x1701, 0x1803, 0x122D,

-      0x1A1D, 0xFFFF }, { 0x0B12, 0x0934, 0x063A, 0x1433, 0x1525, 0x1137, 0x191B, 0xFFFF }, { 0x0A26, 0x003E, 0x082E, 0x1621,

-      0x0F39, 0x1429, 0x003F, 0xFFFF }, { 0x1713, 0x1C1F, 0xFFFF }, { 0x0B10, 0x1335, 0x1A05, 0x1817, 0xFFFF }, { 0x0C18,

-      0x092C, 0x1623, 0x1527, 0xFFFF }, { 0x0A32, 0x0C02, 0x1711, 0x1E09, 0x0D3B, 0x112F, 0xFFFF }, { 0x0A28, 0x0D1C, 0x1919,

-      0x0A3D, 0xFFFF }, { 0x0B20, 0x1631, 0x1903, 0x1815, 0x1C07, 0x142B, 0x132D, 0x1B1D, 0xFFFF }, { 0x1801, 0x1533, 0x1625,

-      0x1237, 0x1A1B, 0xFFFF }, { 0x0C16, 0x053C, 0x0E1E, 0x1721, 0x1529, 0x013F, 0xFFFF }, { 0x0B22, 0x0D04, 0x1039, 0x1D1F,

-      0xFFFF }, { 0x1813, 0x1B05, 0x1917, 0xFFFF }, { 0x0A2A, 0x1723, 0x1435, 0x1627, 0xFFFF }, { 0x0B30, 0x0C14, 0x0936,

-      0x0F08, 0x1F09, 0x0E3B, 0x122F, 0xFFFF }, { 0x0C00, 0x0B24, 0x0E06, 0x0838, 0x1811, 0x1A19, 0x0B3D, 0xFFFF }, { 0x0D1A,

-      0x1731, 0x1A03, 0x1915, 0x1D07, 0x152B, 0xFFFF }, { 0x1901, 0x1633, 0x1725, 0x1337, 0x1B1B, 0x142D, 0x1C1D, 0xFFFF }, {

-      0x0C12, 0x0A34, 0x073A, 0x1629, 0x023F, 0xFFFF }, { 0x0B26, 0x013E, 0x092E, 0x1821, 0x1139, 0x1E1F, 0xFFFF }, { 0x1913,

-      0x1A17, 0xFFFF }, { 0x0C10, 0x1535, 0x1C05, 0x1727, 0xFFFF }, { 0x0D18, 0x0A2C, 0x1823, 0x0F3B, 0x132F, 0xFFFF }, {

-      0x0B32, 0x0D02, 0x1911, 0x1B19, 0xFFFF }, { 0x0B28, 0x0E1C, 0x1B03, 0x1A15, 0x1E07, 0x162B, 0x0C3D, 0xFFFF }, { 0x0C20,

-      0x1831, 0x1A01, 0x1733, 0x152D, 0x1D1D, 0xFFFF }, { 0x1825, 0x1437, 0x1729, 0x1C1B, 0x033F, 0xFFFF }, { 0x0D16, 0x063C,

-      0x0F1E, 0x1921, 0x1239, 0x1F1F, 0xFFFF }, { 0x0C22, 0x0E04, 0x1A13, 0x1B17, 0xFFFF }, { 0x1635, 0x1D05, 0xFFFF }, {

-      0x0B2A, 0x1923, 0x1827, 0xFFFF }, { 0x0C30, 0x0D14, 0x0A36, 0x1A11, 0x103B, 0x142F, 0xFFFF }, { 0x0D00, 0x0C24, 0x0F06,

-      0x0938, 0x1B15, 0x1F07, 0x1C19, 0x172B, 0x0D3D, 0xFFFF }, { 0x0E1A, 0x1931, 0x1B01, 0x1C03, 0x162D, 0x1E1D, 0xFFFF }, {

-      0x1833, 0x1925, 0x1537, 0x1D1B, 0xFFFF }, { 0x0D12, 0x0B34, 0x083A, 0x1A21, 0x1339, 0x1829, 0x043F, 0xFFFF }, { 0x0C26,

-      0x023E, 0x0A2E, 0x1B13, 0xFFFF }, { 0x1735, 0x1E05, 0x1C17, 0xFFFF }, { 0x0D10, 0x1A23, 0x1927, 0xFFFF }, { 0x0E18,

-      0x0B2C, 0x1B11, 0x113B, 0x152F, 0xFFFF }, { 0x0C32, 0x0E02, 0x1D19, 0x0E3D, 0xFFFF }, { 0x0C28, 0x0F1C, 0x1A31, 0x1D03,

-      0x1C15, 0x182B, 0x172D, 0x1F1D, 0xFFFF }, { 0x0D20, 0x1C01, 0x1933, 0x1A25, 0x1637, 0x1E1B, 0xFFFF }, { 0x1B21, 0x1929,

-      0x053F, 0xFFFF }, { 0x0E16, 0x073C, 0x1439, 0xFFFF }, { 0x0D22, 0x0F04, 0x1C13, 0x1F05, 0x1D17, 0xFFFF }, { 0x1B23,

-      0x1835, 0x1A27, 0xFFFF }, { 0x0C2A, 0x123B, 0x162F, 0xFFFF }, { 0x0D30, 0x0E14, 0x0B36, 0x1C11, 0x1E19, 0x0F3D, 0xFFFF

-      }, { 0x0E00, 0x0D24, 0x0A38, 0x1B31, 0x1E03, 0x1D15, 0x192B, 0xFFFF }, { 0x0F1A, 0x1D01, 0x1A33, 0x1B25, 0x1737, 0x1F1B,

-      0x182D, 0xFFFF }, { 0x1A29, 0x063F, 0xFFFF }, { 0x0E12, 0x0C34, 0x093A, 0x1C21, 0x1539, 0xFFFF }, { 0x0D26, 0x033E,

-      0x0B2E, 0x1D13, 0x1E17, 0xFFFF }, { 0x1935, 0x1B27, 0xFFFF }, { 0x0E10, 0x1C23, 0x133B, 0x172F, 0xFFFF }, { 0x0F18,

-      0x0C2C, 0x1D11, 0x1F19, 0xFFFF }, { 0x0D32, 0x0F02, 0x1F03, 0x1E15, 0x1A2B, 0x103D, 0xFFFF }, { 0x0D28, 0x1C31, 0x1E01,

-      0x1B33, 0x192D, 0xFFFF }, { 0x0E20, 0x1C25, 0x1837, 0x1B29, 0x073F, 0xFFFF }, { 0x1D21, 0x1639, 0xFFFF }, { 0x0F16,

-      0x083C, 0x1E13, 0x1F17, 0xFFFF }, { 0x0E22, 0x1A35, 0xFFFF }, { 0x1D23, 0x1C27, 0xFFFF }, { 0x0D2A, 0x1E11, 0x143B,

-      0x182F, 0xFFFF }, { 0x0E30, 0x0F14, 0x0C36, 0x1F15, 0x1B2B, 0x113D, 0xFFFF }, { 0x0F00, 0x0E24, 0x0B38, 0x1D31, 0x1F01,

-      0x1A2D, 0xFFFF }, { 0x1C33, 0x1D25, 0x1937, 0xFFFF }, { 0x1E21, 0x1739, 0x1C29, 0x083F, 0xFFFF }, { 0x0F12, 0x0D34,

-      0x0A3A, 0x1F13, 0xFFFF }, { 0x0E26, 0x043E, 0x0C2E, 0x1B35, 0xFFFF }, { 0x1E23, 0x1D27, 0xFFFF }, { 0x0F10, 0x1F11,

-      0x153B, 0x192F, 0xFFFF }, { 0x0D2C, 0x123D, 0xFFFF },

-   };

-

-   struct etc1_block

-   {

-      // big endian uint64:

-      // bit ofs:  56  48  40  32  24  16   8   0

-      // byte ofs: b0, b1, b2, b3, b4, b5, b6, b7 

-      union 

-      {

-         uint64 m_uint64;

-         uint8 m_bytes[8];

-      };

-

-      uint8 m_low_color[2];

-      uint8 m_high_color[2];

-

-      enum { cNumSelectorBytes = 4 };

-      uint8 m_selectors[cNumSelectorBytes];

-

-      inline void clear()

-      {

-         zero_this(this);

-      }

-

-      inline uint get_byte_bits(uint ofs, uint num) const

-      {

-         RG_ETC1_ASSERT((ofs + num) <= 64U);

-         RG_ETC1_ASSERT(num && (num <= 8U));

-         RG_ETC1_ASSERT((ofs >> 3) == ((ofs + num - 1) >> 3));

-         const uint byte_ofs = 7 - (ofs >> 3);

-         const uint byte_bit_ofs = ofs & 7;

-         return (m_bytes[byte_ofs] >> byte_bit_ofs) & ((1 << num) - 1);

-      }

-

-      inline void set_byte_bits(uint ofs, uint num, uint bits)

-      {

-         RG_ETC1_ASSERT((ofs + num) <= 64U);

-         RG_ETC1_ASSERT(num && (num < 32U));

-         RG_ETC1_ASSERT((ofs >> 3) == ((ofs + num - 1) >> 3));

-         RG_ETC1_ASSERT(bits < (1U << num));

-         const uint byte_ofs = 7 - (ofs >> 3);

-         const uint byte_bit_ofs = ofs & 7;

-         const uint mask = (1 << num) - 1;

-         m_bytes[byte_ofs] &= ~(mask << byte_bit_ofs);

-         m_bytes[byte_ofs] |= (bits << byte_bit_ofs);

-      }

-

-      // false = left/right subblocks

-      // true = upper/lower subblocks

-      inline bool get_flip_bit() const 

-      {

-         return (m_bytes[3] & 1) != 0;

-      }   

-

-      inline void set_flip_bit(bool flip)

-      {

-         m_bytes[3] &= ~1;

-         m_bytes[3] |= static_cast<uint8>(flip);

-      }

-

-      inline bool get_diff_bit() const

-      {

-         return (m_bytes[3] & 2) != 0;

-      }

-

-      inline void set_diff_bit(bool diff)

-      {

-         m_bytes[3] &= ~2;

-         m_bytes[3] |= (static_cast<uint>(diff) << 1);

-      }

-

-      // Returns intensity modifier table (0-7) used by subblock subblock_id.

-      // subblock_id=0 left/top (CW 1), 1=right/bottom (CW 2)

-      inline uint get_inten_table(uint subblock_id) const

-      {

-         RG_ETC1_ASSERT(subblock_id < 2);

-         const uint ofs = subblock_id ? 2 : 5;

-         return (m_bytes[3] >> ofs) & 7;

-      }

-

-      // Sets intensity modifier table (0-7) used by subblock subblock_id (0 or 1)

-      inline void set_inten_table(uint subblock_id, uint t)

-      {

-         RG_ETC1_ASSERT(subblock_id < 2);

-         RG_ETC1_ASSERT(t < 8);

-         const uint ofs = subblock_id ? 2 : 5;

-         m_bytes[3] &= ~(7 << ofs);

-         m_bytes[3] |= (t << ofs);

-      }

-

-      // Returned selector value ranges from 0-3 and is a direct index into g_etc1_inten_tables.

-      inline uint get_selector(uint x, uint y) const

-      {

-         RG_ETC1_ASSERT((x | y) < 4);

-

-         const uint bit_index = x * 4 + y;

-         const uint byte_bit_ofs = bit_index & 7;

-         const uint8 *p = &m_bytes[7 - (bit_index >> 3)];

-         const uint lsb = (p[0] >> byte_bit_ofs) & 1;

-         const uint msb = (p[-2] >> byte_bit_ofs) & 1;

-         const uint val = lsb | (msb << 1);

-

-         return g_etc1_to_selector_index[val];

-      }

-

-      // Selector "val" ranges from 0-3 and is a direct index into g_etc1_inten_tables.

-      inline void set_selector(uint x, uint y, uint val)

-      {

-         RG_ETC1_ASSERT((x | y | val) < 4);

-         const uint bit_index = x * 4 + y;

-

-         uint8 *p = &m_bytes[7 - (bit_index >> 3)];

-

-         const uint byte_bit_ofs = bit_index & 7;

-         const uint mask = 1 << byte_bit_ofs;

-

-         const uint etc1_val = g_selector_index_to_etc1[val];

-

-         const uint lsb = etc1_val & 1;

-         const uint msb = etc1_val >> 1;

-

-         p[0] &= ~mask;

-         p[0] |= (lsb << byte_bit_ofs);

-

-         p[-2] &= ~mask;

-         p[-2] |= (msb << byte_bit_ofs);

-      }

-

-      inline void set_base4_color(uint idx, uint16 c)

-      {

-         if (idx)

-         {

-            set_byte_bits(cETC1AbsColor4R2BitOffset, 4, (c >> 8) & 15);

-            set_byte_bits(cETC1AbsColor4G2BitOffset, 4, (c >> 4) & 15);

-            set_byte_bits(cETC1AbsColor4B2BitOffset, 4, c & 15);

-         }

-         else

-         {

-            set_byte_bits(cETC1AbsColor4R1BitOffset, 4, (c >> 8) & 15);

-            set_byte_bits(cETC1AbsColor4G1BitOffset, 4, (c >> 4) & 15);

-            set_byte_bits(cETC1AbsColor4B1BitOffset, 4, c & 15);

-         }

-      }

-

-      inline uint16 get_base4_color(uint idx) const

-      {

-         uint r, g, b;

-         if (idx)

-         {

-            r = get_byte_bits(cETC1AbsColor4R2BitOffset, 4);

-            g = get_byte_bits(cETC1AbsColor4G2BitOffset, 4);

-            b = get_byte_bits(cETC1AbsColor4B2BitOffset, 4);

-         }

-         else

-         {

-            r = get_byte_bits(cETC1AbsColor4R1BitOffset, 4);

-            g = get_byte_bits(cETC1AbsColor4G1BitOffset, 4);

-            b = get_byte_bits(cETC1AbsColor4B1BitOffset, 4);

-         }

-         return static_cast<uint16>(b | (g << 4U) | (r << 8U));

-      }

-

-      inline void set_base5_color(uint16 c)

-      {

-         set_byte_bits(cETC1BaseColor5RBitOffset, 5, (c >> 10) & 31);

-         set_byte_bits(cETC1BaseColor5GBitOffset, 5, (c >> 5) & 31);

-         set_byte_bits(cETC1BaseColor5BBitOffset, 5, c & 31);

-      }

-

-      inline uint16 get_base5_color() const

-      {

-         const uint r = get_byte_bits(cETC1BaseColor5RBitOffset, 5);

-         const uint g = get_byte_bits(cETC1BaseColor5GBitOffset, 5);

-         const uint b = get_byte_bits(cETC1BaseColor5BBitOffset, 5);

-         return static_cast<uint16>(b | (g << 5U) | (r << 10U));

-      }

-

-      void set_delta3_color(uint16 c)

-      {

-         set_byte_bits(cETC1DeltaColor3RBitOffset, 3, (c >> 6) & 7);

-         set_byte_bits(cETC1DeltaColor3GBitOffset, 3, (c >> 3) & 7);

-         set_byte_bits(cETC1DeltaColor3BBitOffset, 3, c & 7);

-      }

-

-      inline uint16 get_delta3_color() const

-      {

-         const uint r = get_byte_bits(cETC1DeltaColor3RBitOffset, 3);

-         const uint g = get_byte_bits(cETC1DeltaColor3GBitOffset, 3);

-         const uint b = get_byte_bits(cETC1DeltaColor3BBitOffset, 3);

-         return static_cast<uint16>(b | (g << 3U) | (r << 6U));

-      }

-

-      // Base color 5

-      static uint16 pack_color5(const color_quad_u8& color, bool scaled, uint bias = 127U);

-      static uint16 pack_color5(uint r, uint g, uint b, bool scaled, uint bias = 127U);

-

-      static color_quad_u8 unpack_color5(uint16 packed_color5, bool scaled, uint alpha = 255U);

-      static void unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color, bool scaled);

-

-      static bool unpack_color5(color_quad_u8& result, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha = 255U);

-      static bool unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha = 255U);

-

-      // Delta color 3

-      // Inputs range from -4 to 3 (cETC1ColorDeltaMin to cETC1ColorDeltaMax)

-      static uint16 pack_delta3(int r, int g, int b);

-

-      // Results range from -4 to 3 (cETC1ColorDeltaMin to cETC1ColorDeltaMax)

-      static void unpack_delta3(int& r, int& g, int& b, uint16 packed_delta3);

-

-      // Abs color 4

-      static uint16 pack_color4(const color_quad_u8& color, bool scaled, uint bias = 127U);

-      static uint16 pack_color4(uint r, uint g, uint b, bool scaled, uint bias = 127U);

-

-      static color_quad_u8 unpack_color4(uint16 packed_color4, bool scaled, uint alpha = 255U);

-      static void unpack_color4(uint& r, uint& g, uint& b, uint16 packed_color4, bool scaled);

-

-      // subblock colors

-      static void get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint table_idx);

-      static bool get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint16 packed_delta3, uint table_idx);

-      static void get_abs_subblock_colors(color_quad_u8* pDst, uint16 packed_color4, uint table_idx);

-

-      static inline void unscaled_to_scaled_color(color_quad_u8& dst, const color_quad_u8& src, bool color4)

-      {

-         if (color4)

-         {

-            dst.r = src.r | (src.r << 4);

-            dst.g = src.g | (src.g << 4);

-            dst.b = src.b | (src.b << 4);

-         }

-         else

-         {

-            dst.r = (src.r >> 2) | (src.r << 3);

-            dst.g = (src.g >> 2) | (src.g << 3);

-            dst.b = (src.b >> 2) | (src.b << 3);

-         }

-         dst.a = src.a;

-      }

-   };

-

-   // Returns pointer to sorted array.

-   template<typename T, typename Q>

-   T* indirect_radix_sort(uint num_indices, T* pIndices0, T* pIndices1, const Q* pKeys, uint key_ofs, uint key_size, bool init_indices)

-   {  

-      RG_ETC1_ASSERT((key_ofs >= 0) && (key_ofs < sizeof(T)));

-      RG_ETC1_ASSERT((key_size >= 1) && (key_size <= 4));

-

-      if (init_indices)

-      {

-         T* p = pIndices0;

-         T* q = pIndices0 + (num_indices >> 1) * 2;

-         uint i;

-         for (i = 0; p != q; p += 2, i += 2)

-         {

-            p[0] = static_cast<T>(i);

-            p[1] = static_cast<T>(i + 1); 

-         }

-

-         if (num_indices & 1)

-            *p = static_cast<T>(i);

-      }

-

-      uint hist[256 * 4];

-

-      memset(hist, 0, sizeof(hist[0]) * 256 * key_size);

-

-#define RG_ETC1_GET_KEY(p) (*(const uint*)((const uint8*)(pKeys + *(p)) + key_ofs))

-#define RG_ETC1_GET_KEY_FROM_INDEX(i) (*(const uint*)((const uint8*)(pKeys + (i)) + key_ofs))

-

-      if (key_size == 4)

-      {

-         T* p = pIndices0;

-         T* q = pIndices0 + num_indices;

-         for ( ; p != q; p++)

-         {

-            const uint key = RG_ETC1_GET_KEY(p);

-

-            hist[        key        & 0xFF]++;

-            hist[256 + ((key >>  8) & 0xFF)]++;

-            hist[512 + ((key >> 16) & 0xFF)]++;

-            hist[768 + ((key >> 24) & 0xFF)]++;

-         }

-      }

-      else if (key_size == 3)

-      {

-         T* p = pIndices0;

-         T* q = pIndices0 + num_indices;

-         for ( ; p != q; p++)

-         {

-            const uint key = RG_ETC1_GET_KEY(p);

-

-            hist[        key        & 0xFF]++;

-            hist[256 + ((key >>  8) & 0xFF)]++;

-            hist[512 + ((key >> 16) & 0xFF)]++;

-         }

-      }   

-      else if (key_size == 2)

-      {

-         T* p = pIndices0;

-         T* q = pIndices0 + (num_indices >> 1) * 2;

-

-         for ( ; p != q; p += 2)

-         {

-            const uint key0 = RG_ETC1_GET_KEY(p);

-            const uint key1 = RG_ETC1_GET_KEY(p+1);

-

-            hist[        key0         & 0xFF]++;

-            hist[256 + ((key0 >>  8) & 0xFF)]++;

-

-            hist[        key1        & 0xFF]++;

-            hist[256 + ((key1 >>  8) & 0xFF)]++;

-         }

-

-         if (num_indices & 1)

-         {

-            const uint key = RG_ETC1_GET_KEY(p);

-

-            hist[        key        & 0xFF]++;

-            hist[256 + ((key >>  8) & 0xFF)]++;

-         }

-      }      

-      else

-      {

-         RG_ETC1_ASSERT(key_size == 1);

-         if (key_size != 1)

-            return NULL;

-

-         T* p = pIndices0;

-         T* q = pIndices0 + (num_indices >> 1) * 2;

-

-         for ( ; p != q; p += 2)

-         {

-            const uint key0 = RG_ETC1_GET_KEY(p);

-            const uint key1 = RG_ETC1_GET_KEY(p+1);

-

-            hist[key0 & 0xFF]++;

-            hist[key1 & 0xFF]++;

-         }

-

-         if (num_indices & 1)

-         {

-            const uint key = RG_ETC1_GET_KEY(p);

-

-            hist[key & 0xFF]++;

-         }

-      }      

-

-      T* pCur = pIndices0;

-      T* pNew = pIndices1;

-

-      for (uint pass = 0; pass < key_size; pass++)

-      {

-         const uint* pHist = &hist[pass << 8];

-

-         uint offsets[256];

-

-         uint cur_ofs = 0;

-         for (uint i = 0; i < 256; i += 2)

-         {

-            offsets[i] = cur_ofs;

-            cur_ofs += pHist[i];

-

-            offsets[i+1] = cur_ofs;

-            cur_ofs += pHist[i+1];

-         }

-

-         const uint pass_shift = pass << 3;

-

-         T* p = pCur;

-         T* q = pCur + (num_indices >> 1) * 2;

-

-         for ( ; p != q; p += 2)

-         {

-            uint index0 = p[0];

-            uint index1 = p[1];

-

-            uint c0 = (RG_ETC1_GET_KEY_FROM_INDEX(index0) >> pass_shift) & 0xFF;

-            uint c1 = (RG_ETC1_GET_KEY_FROM_INDEX(index1) >> pass_shift) & 0xFF;

-

-            if (c0 == c1)

-            {

-               uint dst_offset0 = offsets[c0];

-

-               offsets[c0] = dst_offset0 + 2;

-

-               pNew[dst_offset0] = static_cast<T>(index0);

-               pNew[dst_offset0 + 1] = static_cast<T>(index1);

-            }

-            else

-            {

-               uint dst_offset0 = offsets[c0]++;

-               uint dst_offset1 = offsets[c1]++;

-

-               pNew[dst_offset0] = static_cast<T>(index0);

-               pNew[dst_offset1] = static_cast<T>(index1);

-            }

-         }

-

-         if (num_indices & 1)

-         {

-            uint index = *p;

-            uint c = (RG_ETC1_GET_KEY_FROM_INDEX(index) >> pass_shift) & 0xFF;

-

-            uint dst_offset = offsets[c];

-            offsets[c] = dst_offset + 1;

-

-            pNew[dst_offset] = static_cast<T>(index);

-         }

-

-         T* t = pCur;

-         pCur = pNew;

-         pNew = t;

-      }            

-

-      return pCur;

-   }

-

-#undef RG_ETC1_GET_KEY

-#undef RG_ETC1_GET_KEY_FROM_INDEX

-

-   uint16 etc1_block::pack_color5(const color_quad_u8& color, bool scaled, uint bias)

-   {

-      return pack_color5(color.r, color.g, color.b, scaled, bias);

-   }

-   

-   uint16 etc1_block::pack_color5(uint r, uint g, uint b, bool scaled, uint bias)

-   {

-      if (scaled)

-      {

-         r = (r * 31U + bias) / 255U;

-         g = (g * 31U + bias) / 255U;

-         b = (b * 31U + bias) / 255U;

-      }

-

-      r = rg_etc1::minimum(r, 31U);

-      g = rg_etc1::minimum(g, 31U);

-      b = rg_etc1::minimum(b, 31U);

-

-      return static_cast<uint16>(b | (g << 5U) | (r << 10U));

-   }

-

-   color_quad_u8 etc1_block::unpack_color5(uint16 packed_color5, bool scaled, uint alpha)

-   {

-      uint b = packed_color5 & 31U;

-      uint g = (packed_color5 >> 5U) & 31U;

-      uint r = (packed_color5 >> 10U) & 31U;

-

-      if (scaled)

-      {

-         b = (b << 3U) | (b >> 2U);

-         g = (g << 3U) | (g >> 2U);

-         r = (r << 3U) | (r >> 2U);

-      }

-

-      return color_quad_u8(cNoClamp, r, g, b, rg_etc1::minimum(alpha, 255U));

-   }

-

-   void etc1_block::unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, bool scaled)

-   {

-      color_quad_u8 c(unpack_color5(packed_color5, scaled, 0));

-      r = c.r;

-      g = c.g;

-      b = c.b;

-   }

-

-   bool etc1_block::unpack_color5(color_quad_u8& result, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha)

-   {

-      int dc_r, dc_g, dc_b;

-      unpack_delta3(dc_r, dc_g, dc_b, packed_delta3);

-      

-      int b = (packed_color5 & 31U) + dc_b;

-      int g = ((packed_color5 >> 5U) & 31U) + dc_g;

-      int r = ((packed_color5 >> 10U) & 31U) + dc_r;

-

-      bool success = true;

-      if (static_cast<uint>(r | g | b) > 31U)

-      {

-         success = false;

-         r = rg_etc1::clamp<int>(r, 0, 31);

-         g = rg_etc1::clamp<int>(g, 0, 31);

-         b = rg_etc1::clamp<int>(b, 0, 31);

-      }

-

-      if (scaled)

-      {

-         b = (b << 3U) | (b >> 2U);

-         g = (g << 3U) | (g >> 2U);

-         r = (r << 3U) | (r >> 2U);

-      }

-

-      result.set_noclamp_rgba(r, g, b, rg_etc1::minimum(alpha, 255U));

-      return success;

-   }

-

-   bool etc1_block::unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha)

-   {

-      color_quad_u8 result;

-      const bool success = unpack_color5(result, packed_color5, packed_delta3, scaled, alpha);

-      r = result.r;

-      g = result.g;

-      b = result.b;

-      return success;

-   }

-     

-   uint16 etc1_block::pack_delta3(int r, int g, int b)

-   {

-      RG_ETC1_ASSERT((r >= cETC1ColorDeltaMin) && (r <= cETC1ColorDeltaMax));

-      RG_ETC1_ASSERT((g >= cETC1ColorDeltaMin) && (g <= cETC1ColorDeltaMax));

-      RG_ETC1_ASSERT((b >= cETC1ColorDeltaMin) && (b <= cETC1ColorDeltaMax));

-      if (r < 0) r += 8;

-      if (g < 0) g += 8;

-      if (b < 0) b += 8;

-      return static_cast<uint16>(b | (g << 3) | (r << 6));

-   }

-   

-   void etc1_block::unpack_delta3(int& r, int& g, int& b, uint16 packed_delta3)

-   {

-      r = (packed_delta3 >> 6) & 7;

-      g = (packed_delta3 >> 3) & 7;

-      b = packed_delta3 & 7;

-      if (r >= 4) r -= 8;

-      if (g >= 4) g -= 8;

-      if (b >= 4) b -= 8;

-   }

-

-   uint16 etc1_block::pack_color4(const color_quad_u8& color, bool scaled, uint bias)

-   {

-      return pack_color4(color.r, color.g, color.b, scaled, bias);

-   }

-   

-   uint16 etc1_block::pack_color4(uint r, uint g, uint b, bool scaled, uint bias)

-   {

-      if (scaled)

-      {

-         r = (r * 15U + bias) / 255U;

-         g = (g * 15U + bias) / 255U;

-         b = (b * 15U + bias) / 255U;

-      }

-

-      r = rg_etc1::minimum(r, 15U);

-      g = rg_etc1::minimum(g, 15U);

-      b = rg_etc1::minimum(b, 15U);

-

-      return static_cast<uint16>(b | (g << 4U) | (r << 8U));

-   }

-

-   color_quad_u8 etc1_block::unpack_color4(uint16 packed_color4, bool scaled, uint alpha)

-   {

-      uint b = packed_color4 & 15U;

-      uint g = (packed_color4 >> 4U) & 15U;

-      uint r = (packed_color4 >> 8U) & 15U;

-

-      if (scaled)

-      {

-         b = (b << 4U) | b;

-         g = (g << 4U) | g;

-         r = (r << 4U) | r;

-      }

-

-      return color_quad_u8(cNoClamp, r, g, b, rg_etc1::minimum(alpha, 255U));

-   }

-   

-   void etc1_block::unpack_color4(uint& r, uint& g, uint& b, uint16 packed_color4, bool scaled)

-   {

-      color_quad_u8 c(unpack_color4(packed_color4, scaled, 0));

-      r = c.r;

-      g = c.g;

-      b = c.b;

-   }

-

-   void etc1_block::get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint table_idx)

-   {

-      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);

-      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];

-

-      uint r, g, b;

-      unpack_color5(r, g, b, packed_color5, true);

-

-      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);

-

-      const int y0 = pInten_modifer_table[0];

-      pDst[0].set(ir + y0, ig + y0, ib + y0);

-

-      const int y1 = pInten_modifer_table[1];

-      pDst[1].set(ir + y1, ig + y1, ib + y1);

-

-      const int y2 = pInten_modifer_table[2];

-      pDst[2].set(ir + y2, ig + y2, ib + y2);

-

-      const int y3 = pInten_modifer_table[3];

-      pDst[3].set(ir + y3, ig + y3, ib + y3);

-   }

-   

-   bool etc1_block::get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint16 packed_delta3, uint table_idx)

-   {

-      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);

-      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];

-

-      uint r, g, b;

-      bool success = unpack_color5(r, g, b, packed_color5, packed_delta3, true);

-

-      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);

-

-      const int y0 = pInten_modifer_table[0];

-      pDst[0].set(ir + y0, ig + y0, ib + y0);

-

-      const int y1 = pInten_modifer_table[1];

-      pDst[1].set(ir + y1, ig + y1, ib + y1);

-

-      const int y2 = pInten_modifer_table[2];

-      pDst[2].set(ir + y2, ig + y2, ib + y2);

-

-      const int y3 = pInten_modifer_table[3];

-      pDst[3].set(ir + y3, ig + y3, ib + y3);

-

-      return success;

-   }

-   

-   void etc1_block::get_abs_subblock_colors(color_quad_u8* pDst, uint16 packed_color4, uint table_idx)

-   {

-      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);

-      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];

-

-      uint r, g, b;

-      unpack_color4(r, g, b, packed_color4, true);

-      

-      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);

-

-      const int y0 = pInten_modifer_table[0];

-      pDst[0].set(ir + y0, ig + y0, ib + y0);

-      

-      const int y1 = pInten_modifer_table[1];

-      pDst[1].set(ir + y1, ig + y1, ib + y1);

-

-      const int y2 = pInten_modifer_table[2];

-      pDst[2].set(ir + y2, ig + y2, ib + y2);

-

-      const int y3 = pInten_modifer_table[3];

-      pDst[3].set(ir + y3, ig + y3, ib + y3);

-   }

-      

-   bool unpack_etc1_block(const void* pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha)

-   {

-      color_quad_u8* pDst = reinterpret_cast<color_quad_u8*>(pDst_pixels_rgba);

-      const etc1_block& block = *static_cast<const etc1_block*>(pETC1_block);

-

-      const bool diff_flag = block.get_diff_bit();

-      const bool flip_flag = block.get_flip_bit();

-      const uint table_index0 = block.get_inten_table(0);

-      const uint table_index1 = block.get_inten_table(1);

-

-      color_quad_u8 subblock_colors0[4];

-      color_quad_u8 subblock_colors1[4];

-      bool success = true;

-

-      if (diff_flag)

-      {

-         const uint16 base_color5 = block.get_base5_color();

-         const uint16 delta_color3 = block.get_delta3_color();

-         etc1_block::get_diff_subblock_colors(subblock_colors0, base_color5, table_index0);

-            

-         if (!etc1_block::get_diff_subblock_colors(subblock_colors1, base_color5, delta_color3, table_index1))

-            success = false;

-      }

-      else

-      {

-         const uint16 base_color4_0 = block.get_base4_color(0);

-         etc1_block::get_abs_subblock_colors(subblock_colors0, base_color4_0, table_index0);

-

-         const uint16 base_color4_1 = block.get_base4_color(1);

-         etc1_block::get_abs_subblock_colors(subblock_colors1, base_color4_1, table_index1);

-      }

-

-      if (preserve_alpha)

-      {

-         if (flip_flag)

-         {

-            for (uint y = 0; y < 2; y++)

-            {

-               pDst[0].set_rgb(subblock_colors0[block.get_selector(0, y)]);

-               pDst[1].set_rgb(subblock_colors0[block.get_selector(1, y)]);

-               pDst[2].set_rgb(subblock_colors0[block.get_selector(2, y)]);

-               pDst[3].set_rgb(subblock_colors0[block.get_selector(3, y)]);

-               pDst += 4;

-            }

-

-            for (uint y = 2; y < 4; y++)

-            {

-               pDst[0].set_rgb(subblock_colors1[block.get_selector(0, y)]);

-               pDst[1].set_rgb(subblock_colors1[block.get_selector(1, y)]);

-               pDst[2].set_rgb(subblock_colors1[block.get_selector(2, y)]);

-               pDst[3].set_rgb(subblock_colors1[block.get_selector(3, y)]);

-               pDst += 4;

-            }

-         }

-         else

-         {

-            for (uint y = 0; y < 4; y++)

-            {

-               pDst[0].set_rgb(subblock_colors0[block.get_selector(0, y)]);

-               pDst[1].set_rgb(subblock_colors0[block.get_selector(1, y)]);

-               pDst[2].set_rgb(subblock_colors1[block.get_selector(2, y)]);

-               pDst[3].set_rgb(subblock_colors1[block.get_selector(3, y)]);

-               pDst += 4;

-            }

-         }

-      }

-      else 

-      {

-         if (flip_flag)

-         {

-            // 0000

-            // 0000

-            // 1111

-            // 1111

-            for (uint y = 0; y < 2; y++)

-            {

-               pDst[0] = subblock_colors0[block.get_selector(0, y)];

-               pDst[1] = subblock_colors0[block.get_selector(1, y)];

-               pDst[2] = subblock_colors0[block.get_selector(2, y)];

-               pDst[3] = subblock_colors0[block.get_selector(3, y)];

-               pDst += 4;

-            }

-

-            for (uint y = 2; y < 4; y++)

-            {

-               pDst[0] = subblock_colors1[block.get_selector(0, y)];

-               pDst[1] = subblock_colors1[block.get_selector(1, y)];

-               pDst[2] = subblock_colors1[block.get_selector(2, y)];

-               pDst[3] = subblock_colors1[block.get_selector(3, y)];

-               pDst += 4;

-            }

-         }

-         else

-         {

-            // 0011

-            // 0011

-            // 0011

-            // 0011

-            for (uint y = 0; y < 4; y++)

-            {

-               pDst[0] = subblock_colors0[block.get_selector(0, y)];

-               pDst[1] = subblock_colors0[block.get_selector(1, y)];

-               pDst[2] = subblock_colors1[block.get_selector(2, y)];

-               pDst[3] = subblock_colors1[block.get_selector(3, y)];

-               pDst += 4;

-            }

-         }

-      }

-      

-      return success;

-   }

-

-   struct etc1_solution_coordinates

-   {

-      inline etc1_solution_coordinates() :

-      m_unscaled_color(0, 0, 0, 0),

-         m_inten_table(0),

-         m_color4(false)

-      {

-      }

-

-      inline etc1_solution_coordinates(uint r, uint g, uint b, uint inten_table, bool color4) : 

-      m_unscaled_color(r, g, b, 255),

-         m_inten_table(inten_table),

-         m_color4(color4)

-      {

-      }

-

-      inline etc1_solution_coordinates(const color_quad_u8& c, uint inten_table, bool color4) : 

-      m_unscaled_color(c),

-         m_inten_table(inten_table),

-         m_color4(color4)

-      {

-      }

-

-      inline etc1_solution_coordinates(const etc1_solution_coordinates& other)

-      {

-         *this = other;

-      }

-

-      inline etc1_solution_coordinates& operator= (const etc1_solution_coordinates& rhs)

-      {

-         m_unscaled_color = rhs.m_unscaled_color;

-         m_inten_table = rhs.m_inten_table;

-         m_color4 = rhs.m_color4;

-         return *this;

-      }

-

-      inline void clear()

-      {

-         m_unscaled_color.clear();

-         m_inten_table = 0;

-         m_color4 = false;

-      }

-

-      inline color_quad_u8 get_scaled_color() const

-      {

-         int br, bg, bb;

-         if (m_color4)

-         {

-            br = m_unscaled_color.r | (m_unscaled_color.r << 4);

-            bg = m_unscaled_color.g | (m_unscaled_color.g << 4);

-            bb = m_unscaled_color.b | (m_unscaled_color.b << 4);

-         }

-         else

-         {

-            br = (m_unscaled_color.r >> 2) | (m_unscaled_color.r << 3);

-            bg = (m_unscaled_color.g >> 2) | (m_unscaled_color.g << 3);

-            bb = (m_unscaled_color.b >> 2) | (m_unscaled_color.b << 3);

-         }

-         return color_quad_u8(br, bg, bb);

-      }

-

-      inline void get_block_colors(color_quad_u8* pBlock_colors)

-      {

-         int br, bg, bb;

-         if (m_color4)

-         {

-            br = m_unscaled_color.r | (m_unscaled_color.r << 4);

-            bg = m_unscaled_color.g | (m_unscaled_color.g << 4);

-            bb = m_unscaled_color.b | (m_unscaled_color.b << 4);

-         }

-         else

-         {

-            br = (m_unscaled_color.r >> 2) | (m_unscaled_color.r << 3);

-            bg = (m_unscaled_color.g >> 2) | (m_unscaled_color.g << 3);

-            bb = (m_unscaled_color.b >> 2) | (m_unscaled_color.b << 3);

-         }

-         const int* pInten_table = g_etc1_inten_tables[m_inten_table];

-         pBlock_colors[0].set(br + pInten_table[0], bg + pInten_table[0], bb + pInten_table[0]);

-         pBlock_colors[1].set(br + pInten_table[1], bg + pInten_table[1], bb + pInten_table[1]);

-         pBlock_colors[2].set(br + pInten_table[2], bg + pInten_table[2], bb + pInten_table[2]);

-         pBlock_colors[3].set(br + pInten_table[3], bg + pInten_table[3], bb + pInten_table[3]);

-      }

-

-      color_quad_u8 m_unscaled_color;

-      uint m_inten_table;

-      bool m_color4;

-   };

-

-   class etc1_optimizer

-   {

-      etc1_optimizer(const etc1_optimizer&);

-      etc1_optimizer& operator= (const etc1_optimizer&);

-

-   public:

-      etc1_optimizer()

-      {

-         clear();

-      }

-

-      void clear()

-      {

-         m_pParams = NULL;

-         m_pResult = NULL;

-         m_pSorted_luma = NULL;

-         m_pSorted_luma_indices = NULL;

-      }

-

-      struct params : etc1_pack_params

-      {

-         params()

-         {

-            clear();

-         }

-

-         params(const etc1_pack_params& base_params) : 

-         etc1_pack_params(base_params)

-         {

-            clear_optimizer_params();

-         }

-

-         void clear()

-         {

-            etc1_pack_params::clear();

-            clear_optimizer_params();

-         }

-

-         void clear_optimizer_params()

-         {

-            m_num_src_pixels = 0;

-            m_pSrc_pixels = 0;

-

-            m_use_color4 = false;

-            static const int s_default_scan_delta[] = { 0 };

-            m_pScan_deltas = s_default_scan_delta;

-            m_scan_delta_size = 1;

-

-            m_base_color5.clear();

-            m_constrain_against_base_color5 = false;

-         }

-

-         uint m_num_src_pixels;

-         const color_quad_u8* m_pSrc_pixels;

-

-         bool m_use_color4;

-         const int* m_pScan_deltas;

-         uint m_scan_delta_size;

-

-         color_quad_u8 m_base_color5;

-         bool m_constrain_against_base_color5;

-      };

-

-      struct results

-      {

-         uint64 m_error;

-         color_quad_u8 m_block_color_unscaled;

-         uint m_block_inten_table;

-         uint m_n;

-         uint8* m_pSelectors;

-         bool m_block_color4;

-

-         inline results& operator= (const results& rhs)

-         {

-            m_block_color_unscaled = rhs.m_block_color_unscaled;

-            m_block_color4 = rhs.m_block_color4;

-            m_block_inten_table = rhs.m_block_inten_table;

-            m_error = rhs.m_error;

-            RG_ETC1_ASSERT(m_n == rhs.m_n);

-            memcpy(m_pSelectors, rhs.m_pSelectors, rhs.m_n);

-            return *this;

-         }

-      };

-

-      void init(const params& params, results& result);

-      bool compute();

-

-   private:      

-      struct potential_solution

-      {

-         potential_solution() : m_coords(), m_error(cUINT64_MAX), m_valid(false)

-         {

-         }

-

-         etc1_solution_coordinates  m_coords;

-         uint8                      m_selectors[8];

-         uint64                     m_error;

-         bool                       m_valid;

-

-         void clear()

-         {

-            m_coords.clear();

-            m_error = cUINT64_MAX;

-            m_valid = false;

-         }

-      };

-

-      const params* m_pParams;

-      results* m_pResult;

-

-      int m_limit;

-

-      vec3F m_avg_color;

-      int m_br, m_bg, m_bb;

-      uint16 m_luma[8];

-      uint32 m_sorted_luma[2][8];

-      const uint32* m_pSorted_luma_indices;

-      uint32* m_pSorted_luma;

-

-      uint8 m_selectors[8];

-      uint8 m_best_selectors[8];

-

-      potential_solution m_best_solution;

-      potential_solution m_trial_solution;

-      uint8 m_temp_selectors[8];

-

-      bool evaluate_solution(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution);

-      bool evaluate_solution_fast(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution);

-   };

-      

-   bool etc1_optimizer::compute()

-   {

-      const uint n = m_pParams->m_num_src_pixels;

-      const int scan_delta_size = m_pParams->m_scan_delta_size;

-      

-      // Scan through a subset of the 3D lattice centered around the avg block color trying each 3D (555 or 444) lattice point as a potential block color.

-      // Each time a better solution is found try to refine the current solution's block color based of the current selectors and intensity table index.

-      for (int zdi = 0; zdi < scan_delta_size; zdi++)

-      {

-         const int zd = m_pParams->m_pScan_deltas[zdi];

-         const int mbb = m_bb + zd;

-         if (mbb < 0) continue; else if (mbb > m_limit) break;

-         

-         for (int ydi = 0; ydi < scan_delta_size; ydi++)

-         {

-            const int yd = m_pParams->m_pScan_deltas[ydi];

-            const int mbg = m_bg + yd;

-            if (mbg < 0) continue; else if (mbg > m_limit) break;

-

-            for (int xdi = 0; xdi < scan_delta_size; xdi++)

-            {

-               const int xd = m_pParams->m_pScan_deltas[xdi];

-               const int mbr = m_br + xd;

-               if (mbr < 0) continue; else if (mbr > m_limit) break;

-      

-               etc1_solution_coordinates coords(mbr, mbg, mbb, 0, m_pParams->m_use_color4);

-               if (m_pParams->m_quality == cHighQuality)

-               {

-                  if (!evaluate_solution(coords, m_trial_solution, &m_best_solution))

-                     continue;

-               }

-               else

-               {

-                  if (!evaluate_solution_fast(coords, m_trial_solution, &m_best_solution))

-                     continue;

-               }

-               

-               // Now we have the input block, the avg. color of the input pixels, a set of trial selector indices, and the block color+intensity index.

-               // Now, for each component, attempt to refine the current solution by solving a simple linear equation. For example, for 4 colors:

-               // The goal is:

-               // pixel0 - (block_color+inten_table[selector0]) + pixel1 - (block_color+inten_table[selector1]) + pixel2 - (block_color+inten_table[selector2]) + pixel3 - (block_color+inten_table[selector3]) = 0

-               // Rearranging this:

-               // (pixel0 + pixel1 + pixel2 + pixel3) - (block_color+inten_table[selector0]) - (block_color+inten_table[selector1]) - (block_color+inten_table[selector2]) - (block_color+inten_table[selector3]) = 0

-               // (pixel0 + pixel1 + pixel2 + pixel3) - block_color - inten_table[selector0] - block_color-inten_table[selector1] - block_color-inten_table[selector2] - block_color-inten_table[selector3] = 0

-               // (pixel0 + pixel1 + pixel2 + pixel3) - 4*block_color - inten_table[selector0] - inten_table[selector1] - inten_table[selector2] - inten_table[selector3] = 0

-               // (pixel0 + pixel1 + pixel2 + pixel3) - 4*block_color - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3]) = 0

-               // (pixel0 + pixel1 + pixel2 + pixel3)/4 - block_color - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3])/4 = 0

-               // block_color = (pixel0 + pixel1 + pixel2 + pixel3)/4 - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3])/4

-               // So what this means:

-               // optimal_block_color = avg_input - avg_inten_delta

-               // So the optimal block color can be computed by taking the average block color and subtracting the current average of the intensity delta.

-               // Unfortunately, optimal_block_color must then be quantized to 555 or 444 so it's not always possible to improve matters using this formula.

-               // Also, the above formula is for unclamped intensity deltas. The actual implementation takes into account clamping.

-

-               const uint max_refinement_trials = (m_pParams->m_quality == cLowQuality) ? 2 : (((xd | yd | zd) == 0) ? 4 : 2);

-               for (uint refinement_trial = 0; refinement_trial < max_refinement_trials; refinement_trial++)

-               {

-                  const uint8* pSelectors = m_best_solution.m_selectors;

-                  const int* pInten_table = g_etc1_inten_tables[m_best_solution.m_coords.m_inten_table];

-

-                  int delta_sum_r = 0, delta_sum_g = 0, delta_sum_b = 0;

-                  const color_quad_u8 base_color(m_best_solution.m_coords.get_scaled_color());

-                  for (uint r = 0; r < n; r++)

-                  {

-                     const uint s = *pSelectors++;

-                     const int yd = pInten_table[s];

-                     // Compute actual delta being applied to each pixel, taking into account clamping.

-                     delta_sum_r += rg_etc1::clamp<int>(base_color.r + yd, 0, 255) - base_color.r;

-                     delta_sum_g += rg_etc1::clamp<int>(base_color.g + yd, 0, 255) - base_color.g;

-                     delta_sum_b += rg_etc1::clamp<int>(base_color.b + yd, 0, 255) - base_color.b;

-                  }

-                  if ((!delta_sum_r) && (!delta_sum_g) && (!delta_sum_b))

-                     break;

-                  const float avg_delta_r_f = static_cast<float>(delta_sum_r) / n;

-                  const float avg_delta_g_f = static_cast<float>(delta_sum_g) / n;

-                  const float avg_delta_b_f = static_cast<float>(delta_sum_b) / n;

-                  const int br1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[0] - avg_delta_r_f) * m_limit / 255.0f + .5f), 0, m_limit);

-                  const int bg1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[1] - avg_delta_g_f) * m_limit / 255.0f + .5f), 0, m_limit);

-                  const int bb1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[2] - avg_delta_b_f) * m_limit / 255.0f + .5f), 0, m_limit);

-                  

-                  bool skip = false;

-                  

-                  if ((mbr == br1) && (mbg == bg1) && (mbb == bb1))

-                     skip = true;

-                  else if ((br1 == m_best_solution.m_coords.m_unscaled_color.r) && (bg1 == m_best_solution.m_coords.m_unscaled_color.g) && (bb1 == m_best_solution.m_coords.m_unscaled_color.b))

-                     skip = true;

-                  else if ((m_br == br1) && (m_bg == bg1) && (m_bb == bb1))

-                     skip = true;

-

-                  if (skip)

-                     break;

-

-                  etc1_solution_coordinates coords1(br1, bg1, bb1, 0, m_pParams->m_use_color4);

-                  if (m_pParams->m_quality == cHighQuality)

-                  {

-                     if (!evaluate_solution(coords1, m_trial_solution, &m_best_solution)) 

-                        break;

-                  }

-                  else

-                  {

-                     if (!evaluate_solution_fast(coords1, m_trial_solution, &m_best_solution))

-                        break;

-                  }

-

-               }  // refinement_trial

-

-            } // xdi

-         } // ydi

-      } // zdi

-

-      if (!m_best_solution.m_valid)

-      {

-         m_pResult->m_error = cUINT32_MAX;

-         return false;

-      }

-      

-      const uint8* pSelectors = m_best_solution.m_selectors;

-

-#ifdef RG_ETC1_BUILD_DEBUG

-      {

-         color_quad_u8 block_colors[4];

-         m_best_solution.m_coords.get_block_colors(block_colors);

-

-         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;

-         uint64 actual_error = 0;

-         for (uint i = 0; i < n; i++)

-            actual_error += pSrc_pixels[i].squared_distance_rgb(block_colors[pSelectors[i]]);

-         

-         RG_ETC1_ASSERT(actual_error == m_best_solution.m_error);

-      }

-#endif      

-      

-      m_pResult->m_error = m_best_solution.m_error;

-

-      m_pResult->m_block_color_unscaled = m_best_solution.m_coords.m_unscaled_color;

-      m_pResult->m_block_color4 = m_best_solution.m_coords.m_color4;

-      

-      m_pResult->m_block_inten_table = m_best_solution.m_coords.m_inten_table;

-      memcpy(m_pResult->m_pSelectors, pSelectors, n);

-      m_pResult->m_n = n;

-

-      return true;

-   }

-

-   void etc1_optimizer::init(const params& p, results& r)

-   {

-      // This version is hardcoded for 8 pixel subblocks.

-      RG_ETC1_ASSERT(p.m_num_src_pixels == 8);

-      

-      m_pParams = &p;

-      m_pResult = &r;

-                  

-      const uint n = 8;

-      

-      m_limit = m_pParams->m_use_color4 ? 15 : 31;

-

-      vec3F avg_color(0.0f);

-

-      for (uint i = 0; i < n; i++)

-      {

-         const color_quad_u8& c = m_pParams->m_pSrc_pixels[i];

-         const vec3F fc(c.r, c.g, c.b);

-

-         avg_color += fc;

-

-         m_luma[i] = static_cast<uint16>(c.r + c.g + c.b);

-         m_sorted_luma[0][i] = i;

-      }

-      avg_color *= (1.0f / static_cast<float>(n));

-      m_avg_color = avg_color;

-

-      m_br = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[0] * m_limit / 255.0f + .5f), 0, m_limit);

-      m_bg = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[1] * m_limit / 255.0f + .5f), 0, m_limit);

-      m_bb = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[2] * m_limit / 255.0f + .5f), 0, m_limit);

-

-      if (m_pParams->m_quality <= cMediumQuality)

-      {

-         m_pSorted_luma_indices = indirect_radix_sort(n, m_sorted_luma[0], m_sorted_luma[1], m_luma, 0, sizeof(m_luma[0]), false);

-         m_pSorted_luma = m_sorted_luma[0];

-         if (m_pSorted_luma_indices == m_sorted_luma[0])

-            m_pSorted_luma = m_sorted_luma[1];

-      

-         for (uint i = 0; i < n; i++)

-            m_pSorted_luma[i] = m_luma[m_pSorted_luma_indices[i]];

-      }

-      

-      m_best_solution.m_coords.clear();

-      m_best_solution.m_valid = false;

-      m_best_solution.m_error = cUINT64_MAX;

-   }

-

-   bool etc1_optimizer::evaluate_solution(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution)

-   {

-      trial_solution.m_valid = false;

-

-      if (m_pParams->m_constrain_against_base_color5)

-      {

-         const int dr = coords.m_unscaled_color.r - m_pParams->m_base_color5.r;

-         const int dg = coords.m_unscaled_color.g - m_pParams->m_base_color5.g;

-         const int db = coords.m_unscaled_color.b - m_pParams->m_base_color5.b;

-

-         if ((rg_etc1::minimum(dr, dg, db) < cETC1ColorDeltaMin) || (rg_etc1::maximum(dr, dg, db) > cETC1ColorDeltaMax))

-            return false;

-      }

-

-      const color_quad_u8 base_color(coords.get_scaled_color());

-      

-      const uint n = 8;

-            

-      trial_solution.m_error = cUINT64_MAX;

-            

-      for (uint inten_table = 0; inten_table < cETC1IntenModifierValues; inten_table++)

-      {

-         const int* pInten_table = g_etc1_inten_tables[inten_table];

-

-         color_quad_u8 block_colors[4];

-         for (uint s = 0; s < 4; s++)

-         {

-            const int yd = pInten_table[s];

-            block_colors[s].set(base_color.r + yd, base_color.g + yd, base_color.b + yd, 0);

-         }

-         

-         uint64 total_error = 0;

-         

-         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;

-         for (uint c = 0; c < n; c++)

-         {

-            const color_quad_u8& src_pixel = *pSrc_pixels++;

-            

-            uint best_selector_index = 0;

-            uint best_error = rg_etc1::square(src_pixel.r - block_colors[0].r) + rg_etc1::square(src_pixel.g - block_colors[0].g) + rg_etc1::square(src_pixel.b - block_colors[0].b);

-

-            uint trial_error = rg_etc1::square(src_pixel.r - block_colors[1].r) + rg_etc1::square(src_pixel.g - block_colors[1].g) + rg_etc1::square(src_pixel.b - block_colors[1].b);

-            if (trial_error < best_error)

-            {

-               best_error = trial_error;

-               best_selector_index = 1;

-            }

-

-            trial_error = rg_etc1::square(src_pixel.r - block_colors[2].r) + rg_etc1::square(src_pixel.g - block_colors[2].g) + rg_etc1::square(src_pixel.b - block_colors[2].b);

-            if (trial_error < best_error)

-            {

-               best_error = trial_error;

-               best_selector_index = 2;

-            }

-

-            trial_error = rg_etc1::square(src_pixel.r - block_colors[3].r) + rg_etc1::square(src_pixel.g - block_colors[3].g) + rg_etc1::square(src_pixel.b - block_colors[3].b);

-            if (trial_error < best_error)

-            {

-               best_error = trial_error;

-               best_selector_index = 3;

-            }

-

-            m_temp_selectors[c] = static_cast<uint8>(best_selector_index);

-

-            total_error += best_error;

-            if (total_error >= trial_solution.m_error)

-               break;

-         }

-         

-         if (total_error < trial_solution.m_error)

-         {

-            trial_solution.m_error = total_error;

-            trial_solution.m_coords.m_inten_table = inten_table;

-            memcpy(trial_solution.m_selectors, m_temp_selectors, 8);

-            trial_solution.m_valid = true;

-         }

-      }

-      trial_solution.m_coords.m_unscaled_color = coords.m_unscaled_color;

-      trial_solution.m_coords.m_color4 = m_pParams->m_use_color4;

-

-      bool success = false;

-      if (pBest_solution)

-      {

-         if (trial_solution.m_error < pBest_solution->m_error)

-         {

-            *pBest_solution = trial_solution;

-            success = true;

-         }

-      }

-

-      return success;

-   }

-

-   bool etc1_optimizer::evaluate_solution_fast(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution)

-   {

-      if (m_pParams->m_constrain_against_base_color5)

-      {

-         const int dr = coords.m_unscaled_color.r - m_pParams->m_base_color5.r;

-         const int dg = coords.m_unscaled_color.g - m_pParams->m_base_color5.g;

-         const int db = coords.m_unscaled_color.b - m_pParams->m_base_color5.b;

-

-         if ((rg_etc1::minimum(dr, dg, db) < cETC1ColorDeltaMin) || (rg_etc1::maximum(dr, dg, db) > cETC1ColorDeltaMax))

-         {

-            trial_solution.m_valid = false;

-            return false;

-         }

-      }

-

-      const color_quad_u8 base_color(coords.get_scaled_color());

-

-      const uint n = 8;

-      

-      trial_solution.m_error = cUINT64_MAX;

-

-      for (int inten_table = cETC1IntenModifierValues - 1; inten_table >= 0; --inten_table)

-      {

-         const int* pInten_table = g_etc1_inten_tables[inten_table];

-

-         uint block_inten[4];

-         color_quad_u8 block_colors[4];

-         for (uint s = 0; s < 4; s++)

-         {

-            const int yd = pInten_table[s];

-            color_quad_u8 block_color(base_color.r + yd, base_color.g + yd, base_color.b + yd, 0);

-            block_colors[s] = block_color;

-            block_inten[s] = block_color.r + block_color.g + block_color.b;

-         }

-

-         // evaluate_solution_fast() enforces/assumesd a total ordering of the input colors along the intensity (1,1,1) axis to more quickly classify the inputs to selectors.

-         // The inputs colors have been presorted along the projection onto this axis, and ETC1 block colors are always ordered along the intensity axis, so this classification is fast.

-         // 0   1   2   3

-         //   01  12  23

-         const uint block_inten_midpoints[3] = { block_inten[0] + block_inten[1], block_inten[1] + block_inten[2], block_inten[2] + block_inten[3] };

-

-         uint64 total_error = 0;

-         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;

-         if ((m_pSorted_luma[n - 1] * 2) < block_inten_midpoints[0])

-         {

-            if (block_inten[0] > m_pSorted_luma[n - 1])

-            {

-           const uint min_error = intabs(block_inten[0] - m_pSorted_luma[n - 1]);

-               if (min_error >= trial_solution.m_error)

-                  continue;

-            }

-

-            memset(&m_temp_selectors[0], 0, n);

-

-            for (uint c = 0; c < n; c++)

-               total_error += block_colors[0].squared_distance_rgb(pSrc_pixels[c]);

-         }

-         else if ((m_pSorted_luma[0] * 2) >= block_inten_midpoints[2])

-         {

-            if (m_pSorted_luma[0] > block_inten[3])

-            {

-           const uint min_error = intabs(m_pSorted_luma[0] - block_inten[3]);

-               if (min_error >= trial_solution.m_error)

-                  continue;

-            }

-

-            memset(&m_temp_selectors[0], 3, n);

-

-            for (uint c = 0; c < n; c++)

-               total_error += block_colors[3].squared_distance_rgb(pSrc_pixels[c]);

-         }

-         else

-         {

-            uint cur_selector = 0, c;

-            for (c = 0; c < n; c++)

-            {

-               const uint y = m_pSorted_luma[c];

-               while ((y * 2) >= block_inten_midpoints[cur_selector])

-                  if (++cur_selector > 2)

-                     goto done;

-               const uint sorted_pixel_index = m_pSorted_luma_indices[c];

-               m_temp_selectors[sorted_pixel_index] = static_cast<uint8>(cur_selector);

-               total_error += block_colors[cur_selector].squared_distance_rgb(pSrc_pixels[sorted_pixel_index]);

-            }

-done:

-            while (c < n)

-            {

-               const uint sorted_pixel_index = m_pSorted_luma_indices[c];

-               m_temp_selectors[sorted_pixel_index] = 3;

-               total_error += block_colors[3].squared_distance_rgb(pSrc_pixels[sorted_pixel_index]);

-               ++c;

-            }

-         }

-

-         if (total_error < trial_solution.m_error)

-         {

-            trial_solution.m_error = total_error;

-            trial_solution.m_coords.m_inten_table = inten_table;

-            memcpy(trial_solution.m_selectors, m_temp_selectors, n);

-            trial_solution.m_valid = true;

-            if (!total_error)

-               break;

-         }

-      }

-      trial_solution.m_coords.m_unscaled_color = coords.m_unscaled_color;

-      trial_solution.m_coords.m_color4 = m_pParams->m_use_color4;

-      

-      bool success = false;

-      if (pBest_solution)

-      {

-         if (trial_solution.m_error < pBest_solution->m_error)

-         {

-            *pBest_solution = trial_solution;

-            success = true;

-         }

-      }

-

-      return success;

-   }

-         

-   static uint etc1_decode_value(uint diff, uint inten, uint selector, uint packed_c)

-   {

-      const uint limit = diff ? 32 : 16; limit;

-      RG_ETC1_ASSERT((diff < 2) && (inten < 8) && (selector < 4) && (packed_c < limit));

-      int c;

-      if (diff)

-         c = (packed_c >> 2) | (packed_c << 3);

-      else 

-         c = packed_c | (packed_c << 4);

-      c += g_etc1_inten_tables[inten][selector];

-      c = rg_etc1::clamp<int>(c, 0, 255);

-      return c;

-   }

-

-   static inline int mul_8bit(int a, int b) { int t = a*b + 128; return (t + (t >> 8)) >> 8; }

-

-   void pack_etc1_block_init()

-   {

-      for (uint diff = 0; diff < 2; diff++)

-      {

-         const uint limit = diff ? 32 : 16;

-

-         for (uint inten = 0; inten < 8; inten++)

-         {

-            for (uint selector = 0; selector < 4; selector++)

-            {

-               const uint inverse_table_index = diff + (inten << 1) + (selector << 4);

-               for (uint color = 0; color < 256; color++)

-               {

-                  uint best_error = cUINT32_MAX, best_packed_c = 0;

-                  for (uint packed_c = 0; packed_c < limit; packed_c++)

-                  {

-                     int v = etc1_decode_value(diff, inten, selector, packed_c);

-                     uint err = labs(v - static_cast<int>(color));

-		     //printf("err: %d - %u = %u\n",v,color,err);

-                     if (err < best_error)

-                     {

-                        best_error = err;

-                        best_packed_c = packed_c;

-                        if (!best_error) 

-                           break;

-                     }

-                  }

-                  RG_ETC1_ASSERT(best_error <= 255);

-                  g_etc1_inverse_lookup[inverse_table_index][color] = static_cast<uint16>(best_packed_c | (best_error << 8));

-               }

-            }

-         }

-      }

-      

-      uint expand5[32];

-      for(int i = 0; i < 32; i++)

-         expand5[i] = (i << 3) | (i >> 2);

-

-      for(int i = 0; i < 256 + 16; i++)

-      {

-         int v = clamp<int>(i - 8, 0, 255);

-         g_quant5_tab[i] = static_cast<uint8>(expand5[mul_8bit(v,31)]);

-      }

-   }

-

-   // Packs solid color blocks efficiently using a set of small precomputed tables.

-   // For random 888 inputs, MSE results are better than Erricson's ETC1 packer in "slow" mode ~9.5% of the time, is slightly worse only ~.01% of the time, and is equal the rest of the time.

-   static uint64 pack_etc1_block_solid_color(etc1_block& block, const uint8* pColor, etc1_pack_params& pack_params)

-   {

-      pack_params;

-      RG_ETC1_ASSERT(g_etc1_inverse_lookup[0][255]);

-            

-      static uint s_next_comp[4] = { 1, 2, 0, 1 };

-            

-      uint best_error = cUINT32_MAX, best_i = 0;

-      int best_x = 0, best_packed_c1 = 0, best_packed_c2 = 0;

-

-      // For each possible 8-bit value, there is a precomputed list of diff/inten/selector configurations that allow that 8-bit value to be encoded with no error.

-      for (uint i = 0; i < 3; i++)

-      {

-         const uint c1 = pColor[s_next_comp[i]], c2 = pColor[s_next_comp[i + 1]];

-

-         const int delta_range = 1;

-         for (int delta = -delta_range; delta <= delta_range; delta++)

-         {

-            const int c_plus_delta = rg_etc1::clamp<int>(pColor[i] + delta, 0, 255);

-

-            const uint16* pTable;

-            if (!c_plus_delta)

-               pTable = g_color8_to_etc_block_config_0_255[0];

-            else if (c_plus_delta == 255)

-               pTable = g_color8_to_etc_block_config_0_255[1];

-            else

-               pTable = g_color8_to_etc_block_config_1_to_254[c_plus_delta - 1];

-

-            do

-            {

-               const uint x = *pTable++;

-

-#ifdef RG_ETC1_BUILD_DEBUG

-               const uint diff = x & 1;

-               const uint inten = (x >> 1) & 7;

-               const uint selector = (x >> 4) & 3;

-               const uint p0 = (x >> 8) & 255;

-               RG_ETC1_ASSERT(etc1_decode_value(diff, inten, selector, p0) == (uint)c_plus_delta);

-#endif

-

-               const uint16* pInverse_table = g_etc1_inverse_lookup[x & 0xFF];

-               uint16 p1 = pInverse_table[c1];

-               uint16 p2 = pInverse_table[c2];

-               const uint trial_error = rg_etc1::square(c_plus_delta - pColor[i]) + rg_etc1::square(p1 >> 8) + rg_etc1::square(p2 >> 8);

-               if (trial_error < best_error)

-               {

-                  best_error = trial_error;

-                  best_x = x;

-                  best_packed_c1 = p1 & 0xFF;

-                  best_packed_c2 = p2 & 0xFF;

-                  best_i = i;

-                  if (!best_error)

-                     goto found_perfect_match;

-               }

-            } while (*pTable != 0xFFFF);

-         }

-      }

-found_perfect_match:

-

-      const uint diff = best_x & 1;

-      const uint inten = (best_x >> 1) & 7;

-

-      block.m_bytes[3] = static_cast<uint8>(((inten | (inten << 3)) << 2) | (diff << 1));

-                        

-      const uint etc1_selector = g_selector_index_to_etc1[(best_x >> 4) & 3];

-      *reinterpret_cast<uint16*>(&block.m_bytes[4]) = (etc1_selector & 2) ? 0xFFFF : 0;

-      *reinterpret_cast<uint16*>(&block.m_bytes[6]) = (etc1_selector & 1) ? 0xFFFF : 0;

-

-      const uint best_packed_c0 = (best_x >> 8) & 255;

-      if (diff)

-      {

-         block.m_bytes[best_i] = static_cast<uint8>(best_packed_c0 << 3);

-         block.m_bytes[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1 << 3);

-         block.m_bytes[s_next_comp[best_i+1]] = static_cast<uint8>(best_packed_c2 << 3);

-      }

-      else

-      {

-         block.m_bytes[best_i] = static_cast<uint8>(best_packed_c0 | (best_packed_c0 << 4));

-         block.m_bytes[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1 | (best_packed_c1 << 4));

-         block.m_bytes[s_next_comp[best_i+1]] = static_cast<uint8>(best_packed_c2 | (best_packed_c2 << 4));

-      }

-

-      return best_error;

-   }

-      

-   static uint pack_etc1_block_solid_color_constrained(

-      etc1_optimizer::results& results, 

-      uint num_colors, const uint8* pColor, 

-      etc1_pack_params& pack_params, 

-      bool use_diff,

-      const color_quad_u8* pBase_color5_unscaled)

-   {

-      RG_ETC1_ASSERT(g_etc1_inverse_lookup[0][255]);

-

-      pack_params;

-      static uint s_next_comp[4] = { 1, 2, 0, 1 };

-

-      uint best_error = cUINT32_MAX, best_i = 0;

-      int best_x = 0, best_packed_c1 = 0, best_packed_c2 = 0;

-

-      // For each possible 8-bit value, there is a precomputed list of diff/inten/selector configurations that allow that 8-bit value to be encoded with no error.

-      for (uint i = 0; i < 3; i++)

-      {

-         const uint c1 = pColor[s_next_comp[i]], c2 = pColor[s_next_comp[i + 1]];

-

-         const int delta_range = 1;

-         for (int delta = -delta_range; delta <= delta_range; delta++)

-         {

-            const int c_plus_delta = rg_etc1::clamp<int>(pColor[i] + delta, 0, 255);

-

-            const uint16* pTable;

-            if (!c_plus_delta)

-               pTable = g_color8_to_etc_block_config_0_255[0];

-            else if (c_plus_delta == 255)

-               pTable = g_color8_to_etc_block_config_0_255[1];

-            else

-               pTable = g_color8_to_etc_block_config_1_to_254[c_plus_delta - 1];

-

-            do

-            {

-               const uint x = *pTable++;

-               const uint diff = x & 1;

-               if (static_cast<uint>(use_diff) != diff)

-               {

-                  if (*pTable == 0xFFFF)

-                     break;

-                  continue;

-               }

-

-               if ((diff) && (pBase_color5_unscaled))

-               {

-                  const int p0 = (x >> 8) & 255;

-                  int delta = p0 - static_cast<int>(pBase_color5_unscaled->c[i]);

-                  if ((delta < cETC1ColorDeltaMin) || (delta > cETC1ColorDeltaMax))

-                  {

-                     if (*pTable == 0xFFFF)

-                        break;

-                     continue;

-                  }

-               }

-

-#ifdef RG_ETC1_BUILD_DEBUG

-               {

-                  const uint inten = (x >> 1) & 7;

-                  const uint selector = (x >> 4) & 3;

-                  const uint p0 = (x >> 8) & 255;

-                  RG_ETC1_ASSERT(etc1_decode_value(diff, inten, selector, p0) == (uint)c_plus_delta);

-               }

-#endif

-

-               const uint16* pInverse_table = g_etc1_inverse_lookup[x & 0xFF];

-               uint16 p1 = pInverse_table[c1];

-               uint16 p2 = pInverse_table[c2];

-

-               if ((diff) && (pBase_color5_unscaled))

-               {

-                  int delta1 = (p1 & 0xFF) - static_cast<int>(pBase_color5_unscaled->c[s_next_comp[i]]);

-                  int delta2 = (p2 & 0xFF) - static_cast<int>(pBase_color5_unscaled->c[s_next_comp[i + 1]]);

-                  if ((delta1 < cETC1ColorDeltaMin) || (delta1 > cETC1ColorDeltaMax) || (delta2 < cETC1ColorDeltaMin) || (delta2 > cETC1ColorDeltaMax))

-                  {

-                     if (*pTable == 0xFFFF)

-                        break;

-                     continue;

-                  }

-               }

-

-               const uint trial_error = rg_etc1::square(c_plus_delta - pColor[i]) + rg_etc1::square(p1 >> 8) + rg_etc1::square(p2 >> 8);

-               if (trial_error < best_error)

-               {

-                  best_error = trial_error;

-                  best_x = x;

-                  best_packed_c1 = p1 & 0xFF;

-                  best_packed_c2 = p2 & 0xFF;

-                  best_i = i;

-                  if (!best_error)

-                     goto found_perfect_match;

-               }

-            } while (*pTable != 0xFFFF);

-         }

-      }

-found_perfect_match:

-

-      if (best_error == cUINT32_MAX)

-         return best_error;

-

-      best_error *= num_colors;

-

-      results.m_n = num_colors;

-      results.m_block_color4 = !(best_x & 1);

-      results.m_block_inten_table = (best_x >> 1) & 7;

-      memset(results.m_pSelectors, (best_x >> 4) & 3, num_colors);

-

-      const uint best_packed_c0 = (best_x >> 8) & 255;

-      results.m_block_color_unscaled[best_i] = static_cast<uint8>(best_packed_c0);

-      results.m_block_color_unscaled[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1);

-      results.m_block_color_unscaled[s_next_comp[best_i + 1]] = static_cast<uint8>(best_packed_c2);

-      results.m_error = best_error;

-      

-      return best_error;

-   }

-

-   // Function originally from RYG's public domain real-time DXT1 compressor, modified for 555.

-   static void dither_block_555(color_quad_u8* dest, const color_quad_u8* block)

-   {

-      int err[8],*ep1 = err,*ep2 = err+4;

-      uint8 *quant = g_quant5_tab+8;

-

-      memset(dest, 0xFF, sizeof(color_quad_u8)*16);

-

-      // process channels seperately

-      for(int ch=0;ch<3;ch++)

-      {

-         uint8* bp = (uint8*)block;

-         uint8* dp = (uint8*)dest;

-

-         bp += ch; dp += ch;

-

-         memset(err,0, sizeof(err));

-         for(int y = 0; y < 4; y++)

-         {

-            // pixel 0

-            dp[ 0] = quant[bp[ 0] + ((3*ep2[1] + 5*ep2[0]) >> 4)];

-            ep1[0] = bp[ 0] - dp[ 0];

-

-            // pixel 1

-            dp[ 4] = quant[bp[ 4] + ((7*ep1[0] + 3*ep2[2] + 5*ep2[1] + ep2[0]) >> 4)];

-            ep1[1] = bp[ 4] - dp[ 4];

-

-            // pixel 2

-            dp[ 8] = quant[bp[ 8] + ((7*ep1[1] + 3*ep2[3] + 5*ep2[2] + ep2[1]) >> 4)];

-            ep1[2] = bp[ 8] - dp[ 8];

-

-            // pixel 3

-            dp[12] = quant[bp[12] + ((7*ep1[2] + 5*ep2[3] + ep2[2]) >> 4)];

-            ep1[3] = bp[12] - dp[12];

-

-            // advance to next line

-            int* tmp = ep1; ep1 = ep2; ep2 = tmp;

-            bp += 16;

-            dp += 16;

-         }

-      }

-   }

-

-   unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params)

-   {

-      const color_quad_u8* pSrc_pixels = reinterpret_cast<const color_quad_u8*>(pSrc_pixels_rgba);

-      etc1_block& dst_block = *static_cast<etc1_block*>(pETC1_block);

-

-#ifdef RG_ETC1_BUILD_DEBUG

-      // Ensure all alpha values are 0xFF.

-      for (uint i = 0; i < 16; i++)

-      {

-         RG_ETC1_ASSERT(pSrc_pixels[i].a == 255);

-      }

-#endif

-

-      color_quad_u8 src_pixel0(pSrc_pixels[0]);

-

-      // Check for solid block.

-      const uint32 first_pixel_u32 = pSrc_pixels->m_u32;

-      int r;

-      for (r = 15; r >= 1; --r)

-         if (pSrc_pixels[r].m_u32 != first_pixel_u32)

-            break;

-      if (!r)

-         return static_cast<unsigned int>(16 * pack_etc1_block_solid_color(dst_block, &pSrc_pixels[0].r, pack_params));

-      

-      color_quad_u8 dithered_pixels[16];

-      if (pack_params.m_dithering)

-      {

-         dither_block_555(dithered_pixels, pSrc_pixels);

-         pSrc_pixels = dithered_pixels;

-      }

-

-      etc1_optimizer optimizer;

-

-      uint64 best_error = cUINT64_MAX;

-      uint best_flip = false, best_use_color4 = false;

-      

-      uint8 best_selectors[2][8];

-      etc1_optimizer::results best_results[2];

-      for (uint i = 0; i < 2; i++)

-      {

-         best_results[i].m_n = 8;

-         best_results[i].m_pSelectors = best_selectors[i];

-      }

-      

-      uint8 selectors[3][8];

-      etc1_optimizer::results results[3];

-      

-      for (uint i = 0; i < 3; i++)

-      {

-         results[i].m_n = 8;

-         results[i].m_pSelectors = selectors[i];

-      }

-            

-      color_quad_u8 subblock_pixels[8];

-

-      etc1_optimizer::params params(pack_params);

-      params.m_num_src_pixels = 8;

-      params.m_pSrc_pixels = subblock_pixels;

-

-      for (uint flip = 0; flip < 2; flip++)

-      {

-         for (uint use_color4 = 0; use_color4 < 2; use_color4++)

-         {

-            uint64 trial_error = 0;

-

-            uint subblock;

-            for (subblock = 0; subblock < 2; subblock++)

-            {

-               if (flip)

-                  memcpy(subblock_pixels, pSrc_pixels + subblock * 8, sizeof(color_quad_u8) * 8);

-               else

-               {

-                  const color_quad_u8* pSrc_col = pSrc_pixels + subblock * 2;

-                  subblock_pixels[0] = pSrc_col[0]; subblock_pixels[1] = pSrc_col[4]; subblock_pixels[2] = pSrc_col[8]; subblock_pixels[3] = pSrc_col[12];

-                  subblock_pixels[4] = pSrc_col[1]; subblock_pixels[5] = pSrc_col[5]; subblock_pixels[6] = pSrc_col[9]; subblock_pixels[7] = pSrc_col[13];

-               }

-

-               results[2].m_error = cUINT64_MAX;

-               if ((params.m_quality >= cMediumQuality) && ((subblock) || (use_color4)))

-               {

-                  const uint32 subblock_pixel0_u32 = subblock_pixels[0].m_u32;

-                  for (r = 7; r >= 1; --r)

-                     if (subblock_pixels[r].m_u32 != subblock_pixel0_u32)

-                        break;

-                  if (!r)

-                  {

-                     pack_etc1_block_solid_color_constrained(results[2], 8, &subblock_pixels[0].r, pack_params, !use_color4, (subblock && !use_color4) ? &results[0].m_block_color_unscaled : NULL);

-                  }

-               }

-

-               params.m_use_color4 = (use_color4 != 0);

-               params.m_constrain_against_base_color5 = false;

-

-               if ((!use_color4) && (subblock))

-               {

-                  params.m_constrain_against_base_color5 = true;

-                  params.m_base_color5 = results[0].m_block_color_unscaled;

-               }

-                              

-               if (params.m_quality == cHighQuality)

-               {

-                  static const int s_scan_delta_0_to_4[] = { -4, -3, -2, -1, 0, 1, 2, 3, 4 };

-                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0_to_4);

-                  params.m_pScan_deltas = s_scan_delta_0_to_4;

-               }

-               else if (params.m_quality == cMediumQuality)

-               {

-                  static const int s_scan_delta_0_to_1[] = { -1, 0, 1 };

-                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0_to_1);

-                  params.m_pScan_deltas = s_scan_delta_0_to_1;

-               }

-               else

-               {

-                  static const int s_scan_delta_0[] = { 0 };

-                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0);

-                  params.m_pScan_deltas = s_scan_delta_0;

-               }

-               

-               optimizer.init(params, results[subblock]);

-               if (!optimizer.compute())

-                  break;

-                              

-               if (params.m_quality >= cMediumQuality)

-               {

-                  // TODO: Fix fairly arbitrary/unrefined thresholds that control how far away to scan for potentially better solutions.

-                  const uint refinement_error_thresh0 = 3000;

-                  const uint refinement_error_thresh1 = 6000;

-                  if (results[subblock].m_error > refinement_error_thresh0)

-                  {

-                     if (params.m_quality == cMediumQuality)

-                     {

-                        static const int s_scan_delta_2_to_3[] = { -3, -2, 2, 3 };

-                        params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_2_to_3);

-                        params.m_pScan_deltas = s_scan_delta_2_to_3;

-                     }

-                     else

-                     {

-                        static const int s_scan_delta_5_to_5[] = { -5, 5 };

-                        static const int s_scan_delta_5_to_8[] = { -8, -7, -6, -5, 5, 6, 7, 8 };

-                        if (results[subblock].m_error > refinement_error_thresh1)

-                        {

-                           params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_5_to_8);

-                           params.m_pScan_deltas = s_scan_delta_5_to_8;

-                        }

-                        else

-                        {

-                           params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_5_to_5);

-                           params.m_pScan_deltas = s_scan_delta_5_to_5;

-                        }

-                     }

-

-                     if (!optimizer.compute())

-                        break;

-                  }

-

-                  if (results[2].m_error < results[subblock].m_error)

-                     results[subblock] = results[2];

-               }

-                            

-               trial_error += results[subblock].m_error;

-               if (trial_error >= best_error)

-                  break;

-            }

-

-            if (subblock < 2)

-               continue;

-

-            best_error = trial_error;

-            best_results[0] = results[0];

-            best_results[1] = results[1];

-            best_flip = flip;

-            best_use_color4 = use_color4;

-            

-         } // use_color4

-

-      } // flip

-

-      int dr = best_results[1].m_block_color_unscaled.r - best_results[0].m_block_color_unscaled.r;

-      int dg = best_results[1].m_block_color_unscaled.g - best_results[0].m_block_color_unscaled.g;

-      int db = best_results[1].m_block_color_unscaled.b - best_results[0].m_block_color_unscaled.b;

-      RG_ETC1_ASSERT(best_use_color4 || ((rg_etc1::minimum(dr, dg, db) >= cETC1ColorDeltaMin) && (rg_etc1::maximum(dr, dg, db) <= cETC1ColorDeltaMax)));

-           

-      if (best_use_color4)

-      {

-         dst_block.m_bytes[0] = static_cast<uint8>(best_results[1].m_block_color_unscaled.r | (best_results[0].m_block_color_unscaled.r << 4));

-         dst_block.m_bytes[1] = static_cast<uint8>(best_results[1].m_block_color_unscaled.g | (best_results[0].m_block_color_unscaled.g << 4));

-         dst_block.m_bytes[2] = static_cast<uint8>(best_results[1].m_block_color_unscaled.b | (best_results[0].m_block_color_unscaled.b << 4));

-      }

-      else

-      {

-         if (dr < 0) dr += 8; dst_block.m_bytes[0] = static_cast<uint8>((best_results[0].m_block_color_unscaled.r << 3) | dr);

-         if (dg < 0) dg += 8; dst_block.m_bytes[1] = static_cast<uint8>((best_results[0].m_block_color_unscaled.g << 3) | dg);

-         if (db < 0) db += 8; dst_block.m_bytes[2] = static_cast<uint8>((best_results[0].m_block_color_unscaled.b << 3) | db);

-      }

-      

-      dst_block.m_bytes[3] = static_cast<uint8>( (best_results[1].m_block_inten_table << 2) | (best_results[0].m_block_inten_table << 5) | ((~best_use_color4 & 1) << 1) | best_flip );

-      

-      uint selector0 = 0, selector1 = 0;

-      if (best_flip)

-      {

-         // flipped:

-         // { 0, 0 }, { 1, 0 }, { 2, 0 }, { 3, 0 },               

-         // { 0, 1 }, { 1, 1 }, { 2, 1 }, { 3, 1 } 

-         //

-         // { 0, 2 }, { 1, 2 }, { 2, 2 }, { 3, 2 },

-         // { 0, 3 }, { 1, 3 }, { 2, 3 }, { 3, 3 }

-         const uint8* pSelectors0 = best_results[0].m_pSelectors;

-         const uint8* pSelectors1 = best_results[1].m_pSelectors;

-         for (int x = 3; x >= 0; --x)

-         {

-            uint b;

-            b = g_selector_index_to_etc1[pSelectors1[4 + x]];

-            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-            b = g_selector_index_to_etc1[pSelectors1[x]];

-            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-            b = g_selector_index_to_etc1[pSelectors0[4 + x]];

-            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-            b = g_selector_index_to_etc1[pSelectors0[x]];

-            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-         }

-      }

-      else

-      {

-         // non-flipped:

-         // { 0, 0 }, { 0, 1 }, { 0, 2 }, { 0, 3 },

-         // { 1, 0 }, { 1, 1 }, { 1, 2 }, { 1, 3 }

-         //

-         // { 2, 0 }, { 2, 1 }, { 2, 2 }, { 2, 3 },

-         // { 3, 0 }, { 3, 1 }, { 3, 2 }, { 3, 3 }

-         for (int subblock = 1; subblock >= 0; --subblock)

-         {

-            const uint8* pSelectors = best_results[subblock].m_pSelectors + 4;

-            for (uint i = 0; i < 2; i++)

-            {

-               uint b;

-               b = g_selector_index_to_etc1[pSelectors[3]];

-               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-               b = g_selector_index_to_etc1[pSelectors[2]];

-               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-               b = g_selector_index_to_etc1[pSelectors[1]];

-               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);

-

-               b = g_selector_index_to_etc1[pSelectors[0]];

-               selector0 = (selector0 << 1) | (b & 1);selector1 = (selector1 << 1) | (b >> 1);

-

-               pSelectors -= 4;

-            }

-         }

-      }

-                  

-      dst_block.m_bytes[4] = static_cast<uint8>(selector1 >> 8); dst_block.m_bytes[5] = static_cast<uint8>(selector1 & 0xFF);

-      dst_block.m_bytes[6] = static_cast<uint8>(selector0 >> 8); dst_block.m_bytes[7] = static_cast<uint8>(selector0 & 0xFF);

-

-      return static_cast<unsigned int>(best_error);

-   }

-

-} // namespace rg_etc1

+// File: rg_etc1.cpp - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>
+// Please see ZLIB license at the end of rg_etc1.h.
+//
+// For more information Ericsson Texture Compression (ETC/ETC1), see:
+// http://www.khronos.org/registry/gles/extensions/OES/OES_compressed_ETC1_RGB8_texture.txt
+//
+// v1.03 - 5/12/13 - Initial public release
+#include "rg_etc1.h"
+
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+//#include <stdio.h>
+#include <math.h>
+#include <stdio.h>
+#pragma warning (disable: 4201) //  nonstandard extension used : nameless struct/union
+
+#if defined(_DEBUG) || defined(DEBUG)
+#define RG_ETC1_BUILD_DEBUG
+#endif
+
+#define RG_ETC1_ASSERT assert
+
+namespace rg_etc1
+{
+
+   inline long labs(long val) {
+        return val < 0 ? -val : val;
+   }
+
+   inline int intabs(int val) {
+
+       return val<0?-val:val;
+   }
+
+   typedef unsigned char uint8;
+   typedef unsigned short uint16;
+   typedef unsigned int uint;
+   typedef unsigned int uint32;
+   typedef long long int64;
+   typedef unsigned long long uint64;
+
+   const uint32 cUINT32_MAX = 0xFFFFFFFFU;
+   const uint64 cUINT64_MAX = 0xFFFFFFFFFFFFFFFFULL; //0xFFFFFFFFFFFFFFFFui64;
+   
+   template<typename T> inline T minimum(T a, T b) { return (a < b) ? a : b; }
+   template<typename T> inline T minimum(T a, T b, T c) { return minimum(minimum(a, b), c); }
+   template<typename T> inline T maximum(T a, T b) { return (a > b) ? a : b; }
+   template<typename T> inline T maximum(T a, T b, T c) { return maximum(maximum(a, b), c); }
+   template<typename T> inline T clamp(T value, T low, T high) { return (value < low) ? low : ((value > high) ? high : value); }
+   template<typename T> inline T square(T value) { return value * value; }
+   template<typename T> inline void zero_object(T& obj) { memset((void*)&obj, 0, sizeof(obj)); }
+   template<typename T> inline void zero_this(T* pObj) { memset((void*)pObj, 0, sizeof(*pObj)); }
+
+   template<class T, size_t N> T decay_array_to_subtype(T (&a)[N]);   
+
+#define RG_ETC1_ARRAY_SIZE(X) (sizeof(X) / sizeof(decay_array_to_subtype(X)))
+
+   enum eNoClamp { cNoClamp };
+
+   struct color_quad_u8
+   {
+      static inline int clamp(int v) { if (v & 0xFFFFFF00U) v = (~(static_cast<int>(v) >> 31)) & 0xFF; return v; }
+
+      struct component_traits { enum { cSigned = false, cFloat = false, cMin = 0U, cMax = 255U }; };
+
+   public:
+      typedef unsigned char component_t;
+      typedef int parameter_t;
+
+      enum { cNumComps = 4 };
+
+      union
+      {
+         struct
+         {
+            component_t r;
+            component_t g;
+            component_t b;
+            component_t a;
+         };
+
+         component_t c[cNumComps];
+
+         uint32 m_u32;
+      };
+
+      inline color_quad_u8()
+      {
+      }
+
+      inline color_quad_u8(const color_quad_u8& other) : m_u32(other.m_u32)
+      {
+      }
+
+      explicit inline color_quad_u8(parameter_t y, parameter_t alpha = component_traits::cMax)
+      {
+         set(y, alpha);
+      }
+
+      inline color_quad_u8(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)
+      {
+         set(red, green, blue, alpha);
+      }
+
+      explicit inline color_quad_u8(eNoClamp, parameter_t y, parameter_t alpha = component_traits::cMax)
+      {
+         set_noclamp_y_alpha(y, alpha);
+      }
+
+      inline color_quad_u8(eNoClamp, parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)
+      {
+         set_noclamp_rgba(red, green, blue, alpha);
+      }
+
+      inline void clear()
+      {
+         m_u32 = 0;
+      }
+
+      inline color_quad_u8& operator= (const color_quad_u8& other)
+      {
+         m_u32 = other.m_u32;
+         return *this;
+      }
+
+      inline color_quad_u8& set_rgb(const color_quad_u8& other)
+      {
+         r = other.r;
+         g = other.g;
+         b = other.b;
+         return *this;
+      }
+
+      inline color_quad_u8& operator= (parameter_t y)
+      {
+         set(y, component_traits::cMax);
+         return *this;
+      }
+
+      inline color_quad_u8& set(parameter_t y, parameter_t alpha = component_traits::cMax)
+      {
+         y = clamp(y);
+         alpha = clamp(alpha);
+         r = static_cast<component_t>(y);
+         g = static_cast<component_t>(y);
+         b = static_cast<component_t>(y);
+         a = static_cast<component_t>(alpha);
+         return *this;
+      }
+
+      inline color_quad_u8& set_noclamp_y_alpha(parameter_t y, parameter_t alpha = component_traits::cMax)
+      {
+         RG_ETC1_ASSERT( (y >= component_traits::cMin) && (y <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (alpha >= component_traits::cMin) && (alpha <= component_traits::cMax) );
+
+         r = static_cast<component_t>(y);
+         g = static_cast<component_t>(y);
+         b = static_cast<component_t>(y);
+         a = static_cast<component_t>(alpha);
+         return *this;
+      }
+
+      inline color_quad_u8& set(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha = component_traits::cMax)
+      {
+         r = static_cast<component_t>(clamp(red));
+         g = static_cast<component_t>(clamp(green));
+         b = static_cast<component_t>(clamp(blue));
+         a = static_cast<component_t>(clamp(alpha));
+         return *this;
+      }
+
+      inline color_quad_u8& set_noclamp_rgba(parameter_t red, parameter_t green, parameter_t blue, parameter_t alpha)
+      {
+         RG_ETC1_ASSERT( (red >= component_traits::cMin) && (red <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (green >= component_traits::cMin) && (green <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (blue >= component_traits::cMin) && (blue <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (alpha >= component_traits::cMin) && (alpha <= component_traits::cMax) );
+
+         r = static_cast<component_t>(red);
+         g = static_cast<component_t>(green);
+         b = static_cast<component_t>(blue);
+         a = static_cast<component_t>(alpha);
+         return *this;
+      }
+
+      inline color_quad_u8& set_noclamp_rgb(parameter_t red, parameter_t green, parameter_t blue)
+      {
+         RG_ETC1_ASSERT( (red >= component_traits::cMin) && (red <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (green >= component_traits::cMin) && (green <= component_traits::cMax) );
+         RG_ETC1_ASSERT( (blue >= component_traits::cMin) && (blue <= component_traits::cMax) );
+
+         r = static_cast<component_t>(red);
+         g = static_cast<component_t>(green);
+         b = static_cast<component_t>(blue);
+         return *this;
+      }
+
+      static inline parameter_t get_min_comp() { return component_traits::cMin; }
+      static inline parameter_t get_max_comp() { return component_traits::cMax; }
+      static inline bool get_comps_are_signed() { return component_traits::cSigned; }
+
+      inline component_t operator[] (uint i) const { RG_ETC1_ASSERT(i < cNumComps); return c[i]; }
+      inline component_t& operator[] (uint i) { RG_ETC1_ASSERT(i < cNumComps); return c[i]; }
+
+      inline color_quad_u8& set_component(uint i, parameter_t f)
+      {
+         RG_ETC1_ASSERT(i < cNumComps);
+
+         c[i] = static_cast<component_t>(clamp(f));
+
+         return *this;
+      }
+
+      inline color_quad_u8& set_grayscale(parameter_t l)
+      {
+         component_t x = static_cast<component_t>(clamp(l));
+         c[0] = x;
+         c[1] = x;
+         c[2] = x;
+         return *this;
+      }
+
+      inline color_quad_u8& clamp(const color_quad_u8& l, const color_quad_u8& h)
+      {
+         for (uint i = 0; i < cNumComps; i++)
+            c[i] = static_cast<component_t>(rg_etc1::clamp<parameter_t>(c[i], l[i], h[i]));
+         return *this;
+      }
+
+      inline color_quad_u8& clamp(parameter_t l, parameter_t h)
+      {
+         for (uint i = 0; i < cNumComps; i++)
+            c[i] = static_cast<component_t>(rg_etc1::clamp<parameter_t>(c[i], l, h));
+         return *this;
+      }
+
+      // Returns CCIR 601 luma (consistent with color_utils::RGB_To_Y).
+      inline parameter_t get_luma() const
+      {
+         return static_cast<parameter_t>((19595U * r + 38470U * g + 7471U * b + 32768U) >> 16U);
+      }
+
+      // Returns REC 709 luma.
+      inline parameter_t get_luma_rec709() const
+      {
+         return static_cast<parameter_t>((13938U * r + 46869U * g + 4729U * b + 32768U) >> 16U);
+      }
+
+      inline uint squared_distance_rgb(const color_quad_u8& c) const
+      {
+         return rg_etc1::square(r - c.r) + rg_etc1::square(g - c.g) + rg_etc1::square(b - c.b);
+      }
+
+      inline uint squared_distance_rgba(const color_quad_u8& c) const
+      {
+         return rg_etc1::square(r - c.r) + rg_etc1::square(g - c.g) + rg_etc1::square(b - c.b) + rg_etc1::square(a - c.a);
+      }
+
+      inline bool rgb_equals(const color_quad_u8& rhs) const
+      {
+         return (r == rhs.r) && (g == rhs.g) && (b == rhs.b);
+      }
+
+      inline bool operator== (const color_quad_u8& rhs) const
+      {
+         return m_u32 == rhs.m_u32;
+      }
+
+      color_quad_u8& operator+= (const color_quad_u8& other)
+      {
+         for (uint i = 0; i < 4; i++)
+            c[i] = static_cast<component_t>(clamp(c[i] + other.c[i]));
+         return *this;
+      }
+
+      color_quad_u8& operator-= (const color_quad_u8& other)
+      {
+         for (uint i = 0; i < 4; i++)
+            c[i] = static_cast<component_t>(clamp(c[i] - other.c[i]));
+         return *this;
+      }
+
+      friend color_quad_u8 operator+ (const color_quad_u8& lhs, const color_quad_u8& rhs)
+      {
+         color_quad_u8 result(lhs);
+         result += rhs;
+         return result;
+      }
+
+      friend color_quad_u8 operator- (const color_quad_u8& lhs, const color_quad_u8& rhs)
+      {
+         color_quad_u8 result(lhs);
+         result -= rhs;
+         return result;
+      }
+   }; // class color_quad_u8
+
+   struct vec3F
+   {
+      float m_s[3];
+      
+      inline vec3F() { }
+      inline vec3F(float s) { m_s[0] = s; m_s[1] = s; m_s[2] = s; }
+      inline vec3F(float x, float y, float z) { m_s[0] = x; m_s[1] = y; m_s[2] = z; }
+      
+      inline float operator[] (uint i) const { RG_ETC1_ASSERT(i < 3); return m_s[i]; }
+
+      inline vec3F& operator += (const vec3F& other) { for (uint i = 0; i < 3; i++) m_s[i] += other.m_s[i]; return *this; }
+
+      inline vec3F& operator *= (float s) { for (uint i = 0; i < 3; i++) m_s[i] *= s; return *this; }
+   };
+     
+   enum etc_constants
+   {
+      cETC1BytesPerBlock = 8U,
+
+      cETC1SelectorBits = 2U,
+      cETC1SelectorValues = 1U << cETC1SelectorBits,
+      cETC1SelectorMask = cETC1SelectorValues - 1U,
+
+      cETC1BlockShift = 2U,
+      cETC1BlockSize = 1U << cETC1BlockShift,
+
+      cETC1LSBSelectorIndicesBitOffset = 0,
+      cETC1MSBSelectorIndicesBitOffset = 16,
+
+      cETC1FlipBitOffset = 32,
+      cETC1DiffBitOffset = 33,
+
+      cETC1IntenModifierNumBits = 3,
+      cETC1IntenModifierValues = 1 << cETC1IntenModifierNumBits,
+      cETC1RightIntenModifierTableBitOffset = 34,
+      cETC1LeftIntenModifierTableBitOffset = 37,
+
+      // Base+Delta encoding (5 bit bases, 3 bit delta)
+      cETC1BaseColorCompNumBits = 5,
+      cETC1BaseColorCompMax = 1 << cETC1BaseColorCompNumBits,
+
+      cETC1DeltaColorCompNumBits = 3,
+      cETC1DeltaColorComp = 1 << cETC1DeltaColorCompNumBits,
+      cETC1DeltaColorCompMax = 1 << cETC1DeltaColorCompNumBits,
+
+      cETC1BaseColor5RBitOffset = 59,
+      cETC1BaseColor5GBitOffset = 51,
+      cETC1BaseColor5BBitOffset = 43,
+
+      cETC1DeltaColor3RBitOffset = 56,
+      cETC1DeltaColor3GBitOffset = 48,
+      cETC1DeltaColor3BBitOffset = 40,
+
+      // Absolute (non-delta) encoding (two 4-bit per component bases)
+      cETC1AbsColorCompNumBits = 4,
+      cETC1AbsColorCompMax = 1 << cETC1AbsColorCompNumBits,
+
+      cETC1AbsColor4R1BitOffset = 60,
+      cETC1AbsColor4G1BitOffset = 52,
+      cETC1AbsColor4B1BitOffset = 44,
+
+      cETC1AbsColor4R2BitOffset = 56,
+      cETC1AbsColor4G2BitOffset = 48,
+      cETC1AbsColor4B2BitOffset = 40,
+
+      cETC1ColorDeltaMin = -4,
+      cETC1ColorDeltaMax = 3,
+
+      // Delta3:
+      // 0   1   2   3   4   5   6   7
+      // 000 001 010 011 100 101 110 111
+      // 0   1   2   3   -4  -3  -2  -1
+   };
+   
+   static uint8 g_quant5_tab[256+16];
+
+
+   static const int g_etc1_inten_tables[cETC1IntenModifierValues][cETC1SelectorValues] = 
+   { 
+      { -8,  -2,   2,   8 }, { -17,  -5,  5,  17 }, { -29,  -9,   9,  29 }, {  -42, -13, 13,  42 }, 
+      { -60, -18, 18,  60 }, { -80, -24, 24,  80 }, { -106, -33, 33, 106 }, { -183, -47, 47, 183 } 
+   };
+
+   static const uint8 g_etc1_to_selector_index[cETC1SelectorValues] = { 2, 3, 1, 0 };
+   static const uint8 g_selector_index_to_etc1[cETC1SelectorValues] = { 3, 2, 0, 1 };
+      
+   // Given an ETC1 diff/inten_table/selector, and an 8-bit desired color, this table encodes the best packed_color in the low byte, and the abs error in the high byte.
+   static uint16 g_etc1_inverse_lookup[2*8*4][256];      // [diff/inten_table/selector][desired_color]
+
+   // g_color8_to_etc_block_config[color][table_index] = Supplies for each 8-bit color value a list of packed ETC1 diff/intensity table/selectors/packed_colors that map to that color.
+   // To pack: diff | (inten << 1) | (selector << 4) | (packed_c << 8)
+   static const uint16 g_color8_to_etc_block_config_0_255[2][33] =
+   {
+      { 0x0000,  0x0010,  0x0002,  0x0012,  0x0004,  0x0014,  0x0006,  0x0016,  0x0008,  0x0018,  0x000A,  0x001A,  0x000C,  0x001C,  0x000E,  0x001E,
+        0x0001,  0x0011,  0x0003,  0x0013,  0x0005,  0x0015,  0x0007,  0x0017,  0x0009,  0x0019,  0x000B,  0x001B,  0x000D,  0x001D,  0x000F,  0x001F, 0xFFFF },
+      { 0x0F20,  0x0F30,  0x0E32,  0x0F22,  0x0E34,  0x0F24,  0x0D36,  0x0F26,  0x0C38,  0x0E28,  0x0B3A,  0x0E2A,  0x093C,  0x0E2C,  0x053E,  0x0D2E,
+        0x1E31,  0x1F21,  0x1D33,  0x1F23,  0x1C35,  0x1E25,  0x1A37,  0x1E27,  0x1839,  0x1D29,  0x163B,  0x1C2B,  0x133D,  0x1B2D,  0x093F,  0x1A2F, 0xFFFF },
+   };
+
+   // Really only [254][11].
+   static const uint16 g_color8_to_etc_block_config_1_to_254[254][12] = 
+   {
+      { 0x021C, 0x0D0D, 0xFFFF }, { 0x0020, 0x0021, 0x0A0B, 0x061F, 0xFFFF }, { 0x0113, 0x0217, 0xFFFF }, { 0x0116, 0x031E,
+      0x0B0E, 0x0405, 0xFFFF }, { 0x0022, 0x0204, 0x050A, 0x0023, 0xFFFF }, { 0x0111, 0x0319, 0x0809, 0x170F, 0xFFFF }, {
+      0x0303, 0x0215, 0x0607, 0xFFFF }, { 0x0030, 0x0114, 0x0408, 0x0031, 0x0201, 0x051D, 0xFFFF }, { 0x0100, 0x0024, 0x0306,
+      0x0025, 0x041B, 0x0E0D, 0xFFFF }, { 0x021A, 0x0121, 0x0B0B, 0x071F, 0xFFFF }, { 0x0213, 0x0317, 0xFFFF }, { 0x0112,
+      0x0505, 0xFFFF }, { 0x0026, 0x070C, 0x0123, 0x0027, 0xFFFF }, { 0x0211, 0x0909, 0xFFFF }, { 0x0110, 0x0315, 0x0707,
+      0x0419, 0x180F, 0xFFFF }, { 0x0218, 0x0131, 0x0301, 0x0403, 0x061D, 0xFFFF }, { 0x0032, 0x0202, 0x0033, 0x0125, 0x051B,
+      0x0F0D, 0xFFFF }, { 0x0028, 0x031C, 0x0221, 0x0029, 0xFFFF }, { 0x0120, 0x0313, 0x0C0B, 0x081F, 0xFFFF }, { 0x0605,
+      0x0417, 0xFFFF }, { 0x0216, 0x041E, 0x0C0E, 0x0223, 0x0127, 0xFFFF }, { 0x0122, 0x0304, 0x060A, 0x0311, 0x0A09, 0xFFFF
+      }, { 0x0519, 0x190F, 0xFFFF }, { 0x002A, 0x0231, 0x0503, 0x0415, 0x0807, 0x002B, 0x071D, 0xFFFF }, { 0x0130, 0x0214,
+      0x0508, 0x0401, 0x0133, 0x0225, 0x061B, 0xFFFF }, { 0x0200, 0x0124, 0x0406, 0x0321, 0x0129, 0x100D, 0xFFFF }, { 0x031A,
+      0x0D0B, 0x091F, 0xFFFF }, { 0x0413, 0x0705, 0x0517, 0xFFFF }, { 0x0212, 0x0034, 0x0323, 0x0035, 0x0227, 0xFFFF }, {
+      0x0126, 0x080C, 0x0B09, 0xFFFF }, { 0x0411, 0x0619, 0x1A0F, 0xFFFF }, { 0x0210, 0x0331, 0x0603, 0x0515, 0x0907, 0x012B,
+      0xFFFF }, { 0x0318, 0x002C, 0x0501, 0x0233, 0x0325, 0x071B, 0x002D, 0x081D, 0xFFFF }, { 0x0132, 0x0302, 0x0229, 0x110D,
+      0xFFFF }, { 0x0128, 0x041C, 0x0421, 0x0E0B, 0x0A1F, 0xFFFF }, { 0x0220, 0x0513, 0x0617, 0xFFFF }, { 0x0135, 0x0805,
+      0x0327, 0xFFFF }, { 0x0316, 0x051E, 0x0D0E, 0x0423, 0xFFFF }, { 0x0222, 0x0404, 0x070A, 0x0511, 0x0719, 0x0C09, 0x1B0F,
+      0xFFFF }, { 0x0703, 0x0615, 0x0A07, 0x022B, 0xFFFF }, { 0x012A, 0x0431, 0x0601, 0x0333, 0x012D, 0x091D, 0xFFFF }, {
+      0x0230, 0x0314, 0x0036, 0x0608, 0x0425, 0x0037, 0x0329, 0x081B, 0x120D, 0xFFFF }, { 0x0300, 0x0224, 0x0506, 0x0521,
+      0x0F0B, 0x0B1F, 0xFFFF }, { 0x041A, 0x0613, 0x0717, 0xFFFF }, { 0x0235, 0x0905, 0xFFFF }, { 0x0312, 0x0134, 0x0523,
+      0x0427, 0xFFFF }, { 0x0226, 0x090C, 0x002E, 0x0611, 0x0D09, 0x002F, 0xFFFF }, { 0x0715, 0x0B07, 0x0819, 0x032B, 0x1C0F,
+      0xFFFF }, { 0x0310, 0x0531, 0x0701, 0x0803, 0x022D, 0x0A1D, 0xFFFF }, { 0x0418, 0x012C, 0x0433, 0x0525, 0x0137, 0x091B,
+      0x130D, 0xFFFF }, { 0x0232, 0x0402, 0x0621, 0x0429, 0xFFFF }, { 0x0228, 0x051C, 0x0713, 0x100B, 0x0C1F, 0xFFFF }, {
+      0x0320, 0x0335, 0x0A05, 0x0817, 0xFFFF }, { 0x0623, 0x0527, 0xFFFF }, { 0x0416, 0x061E, 0x0E0E, 0x0711, 0x0E09, 0x012F,
+      0xFFFF }, { 0x0322, 0x0504, 0x080A, 0x0919, 0x1D0F, 0xFFFF }, { 0x0631, 0x0903, 0x0815, 0x0C07, 0x042B, 0x032D, 0x0B1D,
+      0xFFFF }, { 0x022A, 0x0801, 0x0533, 0x0625, 0x0237, 0x0A1B, 0xFFFF }, { 0x0330, 0x0414, 0x0136, 0x0708, 0x0721, 0x0529,
+      0x140D, 0xFFFF }, { 0x0400, 0x0324, 0x0606, 0x0038, 0x0039, 0x110B, 0x0D1F, 0xFFFF }, { 0x051A, 0x0813, 0x0B05, 0x0917,
+      0xFFFF }, { 0x0723, 0x0435, 0x0627, 0xFFFF }, { 0x0412, 0x0234, 0x0F09, 0x022F, 0xFFFF }, { 0x0326, 0x0A0C, 0x012E,
+      0x0811, 0x0A19, 0x1E0F, 0xFFFF }, { 0x0731, 0x0A03, 0x0915, 0x0D07, 0x052B, 0xFFFF }, { 0x0410, 0x0901, 0x0633, 0x0725,
+      0x0337, 0x0B1B, 0x042D, 0x0C1D, 0xFFFF }, { 0x0518, 0x022C, 0x0629, 0x150D, 0xFFFF }, { 0x0332, 0x0502, 0x0821, 0x0139,
+      0x120B, 0x0E1F, 0xFFFF }, { 0x0328, 0x061C, 0x0913, 0x0A17, 0xFFFF }, { 0x0420, 0x0535, 0x0C05, 0x0727, 0xFFFF }, {
+      0x0823, 0x032F, 0xFFFF }, { 0x0516, 0x071E, 0x0F0E, 0x0911, 0x0B19, 0x1009, 0x1F0F, 0xFFFF }, { 0x0422, 0x0604, 0x090A,
+      0x0B03, 0x0A15, 0x0E07, 0x062B, 0xFFFF }, { 0x0831, 0x0A01, 0x0733, 0x052D, 0x0D1D, 0xFFFF }, { 0x032A, 0x0825, 0x0437,
+      0x0729, 0x0C1B, 0x160D, 0xFFFF }, { 0x0430, 0x0514, 0x0236, 0x0808, 0x0921, 0x0239, 0x130B, 0x0F1F, 0xFFFF }, { 0x0500,
+      0x0424, 0x0706, 0x0138, 0x0A13, 0x0B17, 0xFFFF }, { 0x061A, 0x0635, 0x0D05, 0xFFFF }, { 0x0923, 0x0827, 0xFFFF }, {
+      0x0512, 0x0334, 0x003A, 0x0A11, 0x1109, 0x003B, 0x042F, 0xFFFF }, { 0x0426, 0x0B0C, 0x022E, 0x0B15, 0x0F07, 0x0C19,
+      0x072B, 0xFFFF }, { 0x0931, 0x0B01, 0x0C03, 0x062D, 0x0E1D, 0xFFFF }, { 0x0510, 0x0833, 0x0925, 0x0537, 0x0D1B, 0x170D,
+      0xFFFF }, { 0x0618, 0x032C, 0x0A21, 0x0339, 0x0829, 0xFFFF }, { 0x0432, 0x0602, 0x0B13, 0x140B, 0x101F, 0xFFFF }, {
+      0x0428, 0x071C, 0x0735, 0x0E05, 0x0C17, 0xFFFF }, { 0x0520, 0x0A23, 0x0927, 0xFFFF }, { 0x0B11, 0x1209, 0x013B, 0x052F,
+      0xFFFF }, { 0x0616, 0x081E, 0x0D19, 0xFFFF }, { 0x0522, 0x0704, 0x0A0A, 0x0A31, 0x0D03, 0x0C15, 0x1007, 0x082B, 0x072D,
+      0x0F1D, 0xFFFF }, { 0x0C01, 0x0933, 0x0A25, 0x0637, 0x0E1B, 0xFFFF }, { 0x042A, 0x0B21, 0x0929, 0x180D, 0xFFFF }, {
+	      0x0530, 0x0614, 0x0336, 0x0908, 0x0439, 0x150B, 0x111F, 0xFFFF }, { 0x0600, 0x0524, 0x0806, 0x0238, 0x0C13, 0x0F05,
+      0x0D17, 0xFFFF }, { 0x071A, 0x0B23, 0x0835, 0x0A27, 0xFFFF }, { 0x1309, 0x023B, 0x062F, 0xFFFF }, { 0x0612, 0x0434,
+      0x013A, 0x0C11, 0x0E19, 0xFFFF }, { 0x0526, 0x0C0C, 0x032E, 0x0B31, 0x0E03, 0x0D15, 0x1107, 0x092B, 0xFFFF }, { 0x0D01,
+      0x0A33, 0x0B25, 0x0737, 0x0F1B, 0x082D, 0x101D, 0xFFFF }, { 0x0610, 0x0A29, 0x190D, 0xFFFF }, { 0x0718, 0x042C, 0x0C21,
+      0x0539, 0x160B, 0x121F, 0xFFFF }, { 0x0532, 0x0702, 0x0D13, 0x0E17, 0xFFFF }, { 0x0528, 0x081C, 0x0935, 0x1005, 0x0B27,
+      0xFFFF }, { 0x0620, 0x0C23, 0x033B, 0x072F, 0xFFFF }, { 0x0D11, 0x0F19, 0x1409, 0xFFFF }, { 0x0716, 0x003C, 0x091E,
+      0x0F03, 0x0E15, 0x1207, 0x0A2B, 0x003D, 0xFFFF }, { 0x0622, 0x0804, 0x0B0A, 0x0C31, 0x0E01, 0x0B33, 0x092D, 0x111D,
+      0xFFFF }, { 0x0C25, 0x0837, 0x0B29, 0x101B, 0x1A0D, 0xFFFF }, { 0x052A, 0x0D21, 0x0639, 0x170B, 0x131F, 0xFFFF }, {
+      0x0630, 0x0714, 0x0436, 0x0A08, 0x0E13, 0x0F17, 0xFFFF }, { 0x0700, 0x0624, 0x0906, 0x0338, 0x0A35, 0x1105, 0xFFFF }, {
+      0x081A, 0x0D23, 0x0C27, 0xFFFF }, { 0x0E11, 0x1509, 0x043B, 0x082F, 0xFFFF }, { 0x0712, 0x0534, 0x023A, 0x0F15, 0x1307,
+      0x1019, 0x0B2B, 0x013D, 0xFFFF }, { 0x0626, 0x0D0C, 0x042E, 0x0D31, 0x0F01, 0x1003, 0x0A2D, 0x121D, 0xFFFF }, { 0x0C33,
+      0x0D25, 0x0937, 0x111B, 0x1B0D, 0xFFFF }, { 0x0710, 0x0E21, 0x0739, 0x0C29, 0xFFFF }, { 0x0818, 0x052C, 0x0F13, 0x180B,
+      0x141F, 0xFFFF }, { 0x0632, 0x0802, 0x0B35, 0x1205, 0x1017, 0xFFFF }, { 0x0628, 0x091C, 0x0E23, 0x0D27, 0xFFFF }, {
+      0x0720, 0x0F11, 0x1609, 0x053B, 0x092F, 0xFFFF }, { 0x1119, 0x023D, 0xFFFF }, { 0x0816, 0x013C, 0x0A1E, 0x0E31, 0x1103,
+      0x1015, 0x1407, 0x0C2B, 0x0B2D, 0x131D, 0xFFFF }, { 0x0722, 0x0904, 0x0C0A, 0x1001, 0x0D33, 0x0E25, 0x0A37, 0x121B,
+      0xFFFF }, { 0x0F21, 0x0D29, 0x1C0D, 0xFFFF }, { 0x062A, 0x0839, 0x190B, 0x151F, 0xFFFF }, { 0x0730, 0x0814, 0x0536,
+      0x0B08, 0x1013, 0x1305, 0x1117, 0xFFFF }, { 0x0800, 0x0724, 0x0A06, 0x0438, 0x0F23, 0x0C35, 0x0E27, 0xFFFF }, { 0x091A,
+      0x1709, 0x063B, 0x0A2F, 0xFFFF }, { 0x1011, 0x1219, 0x033D, 0xFFFF }, { 0x0812, 0x0634, 0x033A, 0x0F31, 0x1203, 0x1115,
+      0x1507, 0x0D2B, 0xFFFF }, { 0x0726, 0x0E0C, 0x052E, 0x1101, 0x0E33, 0x0F25, 0x0B37, 0x131B, 0x0C2D, 0x141D, 0xFFFF }, {
+      0x0E29, 0x1D0D, 0xFFFF }, { 0x0810, 0x1021, 0x0939, 0x1A0B, 0x161F, 0xFFFF }, { 0x0918, 0x062C, 0x1113, 0x1217, 0xFFFF
+      }, { 0x0732, 0x0902, 0x0D35, 0x1405, 0x0F27, 0xFFFF }, { 0x0728, 0x0A1C, 0x1023, 0x073B, 0x0B2F, 0xFFFF }, { 0x0820,
+      0x1111, 0x1319, 0x1809, 0xFFFF }, { 0x1303, 0x1215, 0x1607, 0x0E2B, 0x043D, 0xFFFF }, { 0x0916, 0x023C, 0x0B1E, 0x1031,
+      0x1201, 0x0F33, 0x0D2D, 0x151D, 0xFFFF }, { 0x0822, 0x0A04, 0x0D0A, 0x1025, 0x0C37, 0x0F29, 0x141B, 0x1E0D, 0xFFFF }, {
+      0x1121, 0x0A39, 0x1B0B, 0x171F, 0xFFFF }, { 0x072A, 0x1213, 0x1317, 0xFFFF }, { 0x0830, 0x0914, 0x0636, 0x0C08, 0x0E35,
+      0x1505, 0xFFFF }, { 0x0900, 0x0824, 0x0B06, 0x0538, 0x1123, 0x1027, 0xFFFF }, { 0x0A1A, 0x1211, 0x1909, 0x083B, 0x0C2F,
+      0xFFFF }, { 0x1315, 0x1707, 0x1419, 0x0F2B, 0x053D, 0xFFFF }, { 0x0912, 0x0734, 0x043A, 0x1131, 0x1301, 0x1403, 0x0E2D,
+      0x161D, 0xFFFF }, { 0x0826, 0x0F0C, 0x062E, 0x1033, 0x1125, 0x0D37, 0x151B, 0x1F0D, 0xFFFF }, { 0x1221, 0x0B39, 0x1029,
+      0xFFFF }, { 0x0910, 0x1313, 0x1C0B, 0x181F, 0xFFFF }, { 0x0A18, 0x072C, 0x0F35, 0x1605, 0x1417, 0xFFFF }, { 0x0832,
+      0x0A02, 0x1223, 0x1127, 0xFFFF }, { 0x0828, 0x0B1C, 0x1311, 0x1A09, 0x093B, 0x0D2F, 0xFFFF }, { 0x0920, 0x1519, 0x063D,
+      0xFFFF }, { 0x1231, 0x1503, 0x1415, 0x1807, 0x102B, 0x0F2D, 0x171D, 0xFFFF }, { 0x0A16, 0x033C, 0x0C1E, 0x1401, 0x1133,
+      0x1225, 0x0E37, 0x161B, 0xFFFF }, { 0x0922, 0x0B04, 0x0E0A, 0x1321, 0x1129, 0xFFFF }, { 0x0C39, 0x1D0B, 0x191F, 0xFFFF
+      }, { 0x082A, 0x1413, 0x1705, 0x1517, 0xFFFF }, { 0x0930, 0x0A14, 0x0736, 0x0D08, 0x1323, 0x1035, 0x1227, 0xFFFF }, {
+      0x0A00, 0x0924, 0x0C06, 0x0638, 0x1B09, 0x0A3B, 0x0E2F, 0xFFFF }, { 0x0B1A, 0x1411, 0x1619, 0x073D, 0xFFFF }, { 0x1331,
+      0x1603, 0x1515, 0x1907, 0x112B, 0xFFFF }, { 0x0A12, 0x0834, 0x053A, 0x1501, 0x1233, 0x1325, 0x0F37, 0x171B, 0x102D,
+      0x181D, 0xFFFF }, { 0x0926, 0x072E, 0x1229, 0xFFFF }, { 0x1421, 0x0D39, 0x1E0B, 0x1A1F, 0xFFFF }, { 0x0A10, 0x1513,
+      0x1617, 0xFFFF }, { 0x0B18, 0x082C, 0x1135, 0x1805, 0x1327, 0xFFFF }, { 0x0932, 0x0B02, 0x1423, 0x0B3B, 0x0F2F, 0xFFFF
+      }, { 0x0928, 0x0C1C, 0x1511, 0x1719, 0x1C09, 0xFFFF }, { 0x0A20, 0x1703, 0x1615, 0x1A07, 0x122B, 0x083D, 0xFFFF }, {
+      0x1431, 0x1601, 0x1333, 0x112D, 0x191D, 0xFFFF }, { 0x0B16, 0x043C, 0x0D1E, 0x1425, 0x1037, 0x1329, 0x181B, 0xFFFF }, {
+      0x0A22, 0x0C04, 0x0F0A, 0x1521, 0x0E39, 0x1F0B, 0x1B1F, 0xFFFF }, { 0x1613, 0x1717, 0xFFFF }, { 0x092A, 0x1235, 0x1905,
+      0xFFFF }, { 0x0A30, 0x0B14, 0x0836, 0x0E08, 0x1523, 0x1427, 0xFFFF }, { 0x0B00, 0x0A24, 0x0D06, 0x0738, 0x1611, 0x1D09,
+      0x0C3B, 0x102F, 0xFFFF }, { 0x0C1A, 0x1715, 0x1B07, 0x1819, 0x132B, 0x093D, 0xFFFF }, { 0x1531, 0x1701, 0x1803, 0x122D,
+      0x1A1D, 0xFFFF }, { 0x0B12, 0x0934, 0x063A, 0x1433, 0x1525, 0x1137, 0x191B, 0xFFFF }, { 0x0A26, 0x003E, 0x082E, 0x1621,
+      0x0F39, 0x1429, 0x003F, 0xFFFF }, { 0x1713, 0x1C1F, 0xFFFF }, { 0x0B10, 0x1335, 0x1A05, 0x1817, 0xFFFF }, { 0x0C18,
+      0x092C, 0x1623, 0x1527, 0xFFFF }, { 0x0A32, 0x0C02, 0x1711, 0x1E09, 0x0D3B, 0x112F, 0xFFFF }, { 0x0A28, 0x0D1C, 0x1919,
+      0x0A3D, 0xFFFF }, { 0x0B20, 0x1631, 0x1903, 0x1815, 0x1C07, 0x142B, 0x132D, 0x1B1D, 0xFFFF }, { 0x1801, 0x1533, 0x1625,
+      0x1237, 0x1A1B, 0xFFFF }, { 0x0C16, 0x053C, 0x0E1E, 0x1721, 0x1529, 0x013F, 0xFFFF }, { 0x0B22, 0x0D04, 0x1039, 0x1D1F,
+      0xFFFF }, { 0x1813, 0x1B05, 0x1917, 0xFFFF }, { 0x0A2A, 0x1723, 0x1435, 0x1627, 0xFFFF }, { 0x0B30, 0x0C14, 0x0936,
+      0x0F08, 0x1F09, 0x0E3B, 0x122F, 0xFFFF }, { 0x0C00, 0x0B24, 0x0E06, 0x0838, 0x1811, 0x1A19, 0x0B3D, 0xFFFF }, { 0x0D1A,
+      0x1731, 0x1A03, 0x1915, 0x1D07, 0x152B, 0xFFFF }, { 0x1901, 0x1633, 0x1725, 0x1337, 0x1B1B, 0x142D, 0x1C1D, 0xFFFF }, {
+      0x0C12, 0x0A34, 0x073A, 0x1629, 0x023F, 0xFFFF }, { 0x0B26, 0x013E, 0x092E, 0x1821, 0x1139, 0x1E1F, 0xFFFF }, { 0x1913,
+      0x1A17, 0xFFFF }, { 0x0C10, 0x1535, 0x1C05, 0x1727, 0xFFFF }, { 0x0D18, 0x0A2C, 0x1823, 0x0F3B, 0x132F, 0xFFFF }, {
+      0x0B32, 0x0D02, 0x1911, 0x1B19, 0xFFFF }, { 0x0B28, 0x0E1C, 0x1B03, 0x1A15, 0x1E07, 0x162B, 0x0C3D, 0xFFFF }, { 0x0C20,
+      0x1831, 0x1A01, 0x1733, 0x152D, 0x1D1D, 0xFFFF }, { 0x1825, 0x1437, 0x1729, 0x1C1B, 0x033F, 0xFFFF }, { 0x0D16, 0x063C,
+      0x0F1E, 0x1921, 0x1239, 0x1F1F, 0xFFFF }, { 0x0C22, 0x0E04, 0x1A13, 0x1B17, 0xFFFF }, { 0x1635, 0x1D05, 0xFFFF }, {
+      0x0B2A, 0x1923, 0x1827, 0xFFFF }, { 0x0C30, 0x0D14, 0x0A36, 0x1A11, 0x103B, 0x142F, 0xFFFF }, { 0x0D00, 0x0C24, 0x0F06,
+      0x0938, 0x1B15, 0x1F07, 0x1C19, 0x172B, 0x0D3D, 0xFFFF }, { 0x0E1A, 0x1931, 0x1B01, 0x1C03, 0x162D, 0x1E1D, 0xFFFF }, {
+      0x1833, 0x1925, 0x1537, 0x1D1B, 0xFFFF }, { 0x0D12, 0x0B34, 0x083A, 0x1A21, 0x1339, 0x1829, 0x043F, 0xFFFF }, { 0x0C26,
+      0x023E, 0x0A2E, 0x1B13, 0xFFFF }, { 0x1735, 0x1E05, 0x1C17, 0xFFFF }, { 0x0D10, 0x1A23, 0x1927, 0xFFFF }, { 0x0E18,
+      0x0B2C, 0x1B11, 0x113B, 0x152F, 0xFFFF }, { 0x0C32, 0x0E02, 0x1D19, 0x0E3D, 0xFFFF }, { 0x0C28, 0x0F1C, 0x1A31, 0x1D03,
+      0x1C15, 0x182B, 0x172D, 0x1F1D, 0xFFFF }, { 0x0D20, 0x1C01, 0x1933, 0x1A25, 0x1637, 0x1E1B, 0xFFFF }, { 0x1B21, 0x1929,
+      0x053F, 0xFFFF }, { 0x0E16, 0x073C, 0x1439, 0xFFFF }, { 0x0D22, 0x0F04, 0x1C13, 0x1F05, 0x1D17, 0xFFFF }, { 0x1B23,
+      0x1835, 0x1A27, 0xFFFF }, { 0x0C2A, 0x123B, 0x162F, 0xFFFF }, { 0x0D30, 0x0E14, 0x0B36, 0x1C11, 0x1E19, 0x0F3D, 0xFFFF
+      }, { 0x0E00, 0x0D24, 0x0A38, 0x1B31, 0x1E03, 0x1D15, 0x192B, 0xFFFF }, { 0x0F1A, 0x1D01, 0x1A33, 0x1B25, 0x1737, 0x1F1B,
+      0x182D, 0xFFFF }, { 0x1A29, 0x063F, 0xFFFF }, { 0x0E12, 0x0C34, 0x093A, 0x1C21, 0x1539, 0xFFFF }, { 0x0D26, 0x033E,
+      0x0B2E, 0x1D13, 0x1E17, 0xFFFF }, { 0x1935, 0x1B27, 0xFFFF }, { 0x0E10, 0x1C23, 0x133B, 0x172F, 0xFFFF }, { 0x0F18,
+      0x0C2C, 0x1D11, 0x1F19, 0xFFFF }, { 0x0D32, 0x0F02, 0x1F03, 0x1E15, 0x1A2B, 0x103D, 0xFFFF }, { 0x0D28, 0x1C31, 0x1E01,
+      0x1B33, 0x192D, 0xFFFF }, { 0x0E20, 0x1C25, 0x1837, 0x1B29, 0x073F, 0xFFFF }, { 0x1D21, 0x1639, 0xFFFF }, { 0x0F16,
+      0x083C, 0x1E13, 0x1F17, 0xFFFF }, { 0x0E22, 0x1A35, 0xFFFF }, { 0x1D23, 0x1C27, 0xFFFF }, { 0x0D2A, 0x1E11, 0x143B,
+      0x182F, 0xFFFF }, { 0x0E30, 0x0F14, 0x0C36, 0x1F15, 0x1B2B, 0x113D, 0xFFFF }, { 0x0F00, 0x0E24, 0x0B38, 0x1D31, 0x1F01,
+      0x1A2D, 0xFFFF }, { 0x1C33, 0x1D25, 0x1937, 0xFFFF }, { 0x1E21, 0x1739, 0x1C29, 0x083F, 0xFFFF }, { 0x0F12, 0x0D34,
+      0x0A3A, 0x1F13, 0xFFFF }, { 0x0E26, 0x043E, 0x0C2E, 0x1B35, 0xFFFF }, { 0x1E23, 0x1D27, 0xFFFF }, { 0x0F10, 0x1F11,
+      0x153B, 0x192F, 0xFFFF }, { 0x0D2C, 0x123D, 0xFFFF },
+   };
+
+   struct etc1_block
+   {
+      // big endian uint64:
+      // bit ofs:  56  48  40  32  24  16   8   0
+      // byte ofs: b0, b1, b2, b3, b4, b5, b6, b7 
+      union 
+      {
+         uint64 m_uint64;
+         uint8 m_bytes[8];
+      };
+
+      uint8 m_low_color[2];
+      uint8 m_high_color[2];
+
+      enum { cNumSelectorBytes = 4 };
+      uint8 m_selectors[cNumSelectorBytes];
+
+      inline void clear()
+      {
+         zero_this(this);
+      }
+
+      inline uint get_byte_bits(uint ofs, uint num) const
+      {
+         RG_ETC1_ASSERT((ofs + num) <= 64U);
+         RG_ETC1_ASSERT(num && (num <= 8U));
+         RG_ETC1_ASSERT((ofs >> 3) == ((ofs + num - 1) >> 3));
+         const uint byte_ofs = 7 - (ofs >> 3);
+         const uint byte_bit_ofs = ofs & 7;
+         return (m_bytes[byte_ofs] >> byte_bit_ofs) & ((1 << num) - 1);
+      }
+
+      inline void set_byte_bits(uint ofs, uint num, uint bits)
+      {
+         RG_ETC1_ASSERT((ofs + num) <= 64U);
+         RG_ETC1_ASSERT(num && (num < 32U));
+         RG_ETC1_ASSERT((ofs >> 3) == ((ofs + num - 1) >> 3));
+         RG_ETC1_ASSERT(bits < (1U << num));
+         const uint byte_ofs = 7 - (ofs >> 3);
+         const uint byte_bit_ofs = ofs & 7;
+         const uint mask = (1 << num) - 1;
+         m_bytes[byte_ofs] &= ~(mask << byte_bit_ofs);
+         m_bytes[byte_ofs] |= (bits << byte_bit_ofs);
+      }
+
+      // false = left/right subblocks
+      // true = upper/lower subblocks
+      inline bool get_flip_bit() const 
+      {
+         return (m_bytes[3] & 1) != 0;
+      }   
+
+      inline void set_flip_bit(bool flip)
+      {
+         m_bytes[3] &= ~1;
+         m_bytes[3] |= static_cast<uint8>(flip);
+      }
+
+      inline bool get_diff_bit() const
+      {
+         return (m_bytes[3] & 2) != 0;
+      }
+
+      inline void set_diff_bit(bool diff)
+      {
+         m_bytes[3] &= ~2;
+         m_bytes[3] |= (static_cast<uint>(diff) << 1);
+      }
+
+      // Returns intensity modifier table (0-7) used by subblock subblock_id.
+      // subblock_id=0 left/top (CW 1), 1=right/bottom (CW 2)
+      inline uint get_inten_table(uint subblock_id) const
+      {
+         RG_ETC1_ASSERT(subblock_id < 2);
+         const uint ofs = subblock_id ? 2 : 5;
+         return (m_bytes[3] >> ofs) & 7;
+      }
+
+      // Sets intensity modifier table (0-7) used by subblock subblock_id (0 or 1)
+      inline void set_inten_table(uint subblock_id, uint t)
+      {
+         RG_ETC1_ASSERT(subblock_id < 2);
+         RG_ETC1_ASSERT(t < 8);
+         const uint ofs = subblock_id ? 2 : 5;
+         m_bytes[3] &= ~(7 << ofs);
+         m_bytes[3] |= (t << ofs);
+      }
+
+      // Returned selector value ranges from 0-3 and is a direct index into g_etc1_inten_tables.
+      inline uint get_selector(uint x, uint y) const
+      {
+         RG_ETC1_ASSERT((x | y) < 4);
+
+         const uint bit_index = x * 4 + y;
+         const uint byte_bit_ofs = bit_index & 7;
+         const uint8 *p = &m_bytes[7 - (bit_index >> 3)];
+         const uint lsb = (p[0] >> byte_bit_ofs) & 1;
+         const uint msb = (p[-2] >> byte_bit_ofs) & 1;
+         const uint val = lsb | (msb << 1);
+
+         return g_etc1_to_selector_index[val];
+      }
+
+      // Selector "val" ranges from 0-3 and is a direct index into g_etc1_inten_tables.
+      inline void set_selector(uint x, uint y, uint val)
+      {
+         RG_ETC1_ASSERT((x | y | val) < 4);
+         const uint bit_index = x * 4 + y;
+
+         uint8 *p = &m_bytes[7 - (bit_index >> 3)];
+
+         const uint byte_bit_ofs = bit_index & 7;
+         const uint mask = 1 << byte_bit_ofs;
+
+         const uint etc1_val = g_selector_index_to_etc1[val];
+
+         const uint lsb = etc1_val & 1;
+         const uint msb = etc1_val >> 1;
+
+         p[0] &= ~mask;
+         p[0] |= (lsb << byte_bit_ofs);
+
+         p[-2] &= ~mask;
+         p[-2] |= (msb << byte_bit_ofs);
+      }
+
+      inline void set_base4_color(uint idx, uint16 c)
+      {
+         if (idx)
+         {
+            set_byte_bits(cETC1AbsColor4R2BitOffset, 4, (c >> 8) & 15);
+            set_byte_bits(cETC1AbsColor4G2BitOffset, 4, (c >> 4) & 15);
+            set_byte_bits(cETC1AbsColor4B2BitOffset, 4, c & 15);
+         }
+         else
+         {
+            set_byte_bits(cETC1AbsColor4R1BitOffset, 4, (c >> 8) & 15);
+            set_byte_bits(cETC1AbsColor4G1BitOffset, 4, (c >> 4) & 15);
+            set_byte_bits(cETC1AbsColor4B1BitOffset, 4, c & 15);
+         }
+      }
+
+      inline uint16 get_base4_color(uint idx) const
+      {
+         uint r, g, b;
+         if (idx)
+         {
+            r = get_byte_bits(cETC1AbsColor4R2BitOffset, 4);
+            g = get_byte_bits(cETC1AbsColor4G2BitOffset, 4);
+            b = get_byte_bits(cETC1AbsColor4B2BitOffset, 4);
+         }
+         else
+         {
+            r = get_byte_bits(cETC1AbsColor4R1BitOffset, 4);
+            g = get_byte_bits(cETC1AbsColor4G1BitOffset, 4);
+            b = get_byte_bits(cETC1AbsColor4B1BitOffset, 4);
+         }
+         return static_cast<uint16>(b | (g << 4U) | (r << 8U));
+      }
+
+      inline void set_base5_color(uint16 c)
+      {
+         set_byte_bits(cETC1BaseColor5RBitOffset, 5, (c >> 10) & 31);
+         set_byte_bits(cETC1BaseColor5GBitOffset, 5, (c >> 5) & 31);
+         set_byte_bits(cETC1BaseColor5BBitOffset, 5, c & 31);
+      }
+
+      inline uint16 get_base5_color() const
+      {
+         const uint r = get_byte_bits(cETC1BaseColor5RBitOffset, 5);
+         const uint g = get_byte_bits(cETC1BaseColor5GBitOffset, 5);
+         const uint b = get_byte_bits(cETC1BaseColor5BBitOffset, 5);
+         return static_cast<uint16>(b | (g << 5U) | (r << 10U));
+      }
+
+      void set_delta3_color(uint16 c)
+      {
+         set_byte_bits(cETC1DeltaColor3RBitOffset, 3, (c >> 6) & 7);
+         set_byte_bits(cETC1DeltaColor3GBitOffset, 3, (c >> 3) & 7);
+         set_byte_bits(cETC1DeltaColor3BBitOffset, 3, c & 7);
+      }
+
+      inline uint16 get_delta3_color() const
+      {
+         const uint r = get_byte_bits(cETC1DeltaColor3RBitOffset, 3);
+         const uint g = get_byte_bits(cETC1DeltaColor3GBitOffset, 3);
+         const uint b = get_byte_bits(cETC1DeltaColor3BBitOffset, 3);
+         return static_cast<uint16>(b | (g << 3U) | (r << 6U));
+      }
+
+      // Base color 5
+      static uint16 pack_color5(const color_quad_u8& color, bool scaled, uint bias = 127U);
+      static uint16 pack_color5(uint r, uint g, uint b, bool scaled, uint bias = 127U);
+
+      static color_quad_u8 unpack_color5(uint16 packed_color5, bool scaled, uint alpha = 255U);
+      static void unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color, bool scaled);
+
+      static bool unpack_color5(color_quad_u8& result, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha = 255U);
+      static bool unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha = 255U);
+
+      // Delta color 3
+      // Inputs range from -4 to 3 (cETC1ColorDeltaMin to cETC1ColorDeltaMax)
+      static uint16 pack_delta3(int r, int g, int b);
+
+      // Results range from -4 to 3 (cETC1ColorDeltaMin to cETC1ColorDeltaMax)
+      static void unpack_delta3(int& r, int& g, int& b, uint16 packed_delta3);
+
+      // Abs color 4
+      static uint16 pack_color4(const color_quad_u8& color, bool scaled, uint bias = 127U);
+      static uint16 pack_color4(uint r, uint g, uint b, bool scaled, uint bias = 127U);
+
+      static color_quad_u8 unpack_color4(uint16 packed_color4, bool scaled, uint alpha = 255U);
+      static void unpack_color4(uint& r, uint& g, uint& b, uint16 packed_color4, bool scaled);
+
+      // subblock colors
+      static void get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint table_idx);
+      static bool get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint16 packed_delta3, uint table_idx);
+      static void get_abs_subblock_colors(color_quad_u8* pDst, uint16 packed_color4, uint table_idx);
+
+      static inline void unscaled_to_scaled_color(color_quad_u8& dst, const color_quad_u8& src, bool color4)
+      {
+         if (color4)
+         {
+            dst.r = src.r | (src.r << 4);
+            dst.g = src.g | (src.g << 4);
+            dst.b = src.b | (src.b << 4);
+         }
+         else
+         {
+            dst.r = (src.r >> 2) | (src.r << 3);
+            dst.g = (src.g >> 2) | (src.g << 3);
+            dst.b = (src.b >> 2) | (src.b << 3);
+         }
+         dst.a = src.a;
+      }
+   };
+
+   // Returns pointer to sorted array.
+   template<typename T, typename Q>
+   T* indirect_radix_sort(uint num_indices, T* pIndices0, T* pIndices1, const Q* pKeys, uint key_ofs, uint key_size, bool init_indices)
+   {  
+      RG_ETC1_ASSERT((key_ofs >= 0) && (key_ofs < sizeof(T)));
+      RG_ETC1_ASSERT((key_size >= 1) && (key_size <= 4));
+
+      if (init_indices)
+      {
+         T* p = pIndices0;
+         T* q = pIndices0 + (num_indices >> 1) * 2;
+         uint i;
+         for (i = 0; p != q; p += 2, i += 2)
+         {
+            p[0] = static_cast<T>(i);
+            p[1] = static_cast<T>(i + 1); 
+         }
+
+         if (num_indices & 1)
+            *p = static_cast<T>(i);
+      }
+
+      uint hist[256 * 4];
+
+      memset(hist, 0, sizeof(hist[0]) * 256 * key_size);
+
+#define RG_ETC1_GET_KEY(p) (*(const uint*)((const uint8*)(pKeys + *(p)) + key_ofs))
+#define RG_ETC1_GET_KEY_FROM_INDEX(i) (*(const uint*)((const uint8*)(pKeys + (i)) + key_ofs))
+
+      if (key_size == 4)
+      {
+         T* p = pIndices0;
+         T* q = pIndices0 + num_indices;
+         for ( ; p != q; p++)
+         {
+            const uint key = RG_ETC1_GET_KEY(p);
+
+            hist[        key        & 0xFF]++;
+            hist[256 + ((key >>  8) & 0xFF)]++;
+            hist[512 + ((key >> 16) & 0xFF)]++;
+            hist[768 + ((key >> 24) & 0xFF)]++;
+         }
+      }
+      else if (key_size == 3)
+      {
+         T* p = pIndices0;
+         T* q = pIndices0 + num_indices;
+         for ( ; p != q; p++)
+         {
+            const uint key = RG_ETC1_GET_KEY(p);
+
+            hist[        key        & 0xFF]++;
+            hist[256 + ((key >>  8) & 0xFF)]++;
+            hist[512 + ((key >> 16) & 0xFF)]++;
+         }
+      }   
+      else if (key_size == 2)
+      {
+         T* p = pIndices0;
+         T* q = pIndices0 + (num_indices >> 1) * 2;
+
+         for ( ; p != q; p += 2)
+         {
+            const uint key0 = RG_ETC1_GET_KEY(p);
+            const uint key1 = RG_ETC1_GET_KEY(p+1);
+
+            hist[        key0         & 0xFF]++;
+            hist[256 + ((key0 >>  8) & 0xFF)]++;
+
+            hist[        key1        & 0xFF]++;
+            hist[256 + ((key1 >>  8) & 0xFF)]++;
+         }
+
+         if (num_indices & 1)
+         {
+            const uint key = RG_ETC1_GET_KEY(p);
+
+            hist[        key        & 0xFF]++;
+            hist[256 + ((key >>  8) & 0xFF)]++;
+         }
+      }      
+      else
+      {
+         RG_ETC1_ASSERT(key_size == 1);
+         if (key_size != 1)
+            return NULL;
+
+         T* p = pIndices0;
+         T* q = pIndices0 + (num_indices >> 1) * 2;
+
+         for ( ; p != q; p += 2)
+         {
+            const uint key0 = RG_ETC1_GET_KEY(p);
+            const uint key1 = RG_ETC1_GET_KEY(p+1);
+
+            hist[key0 & 0xFF]++;
+            hist[key1 & 0xFF]++;
+         }
+
+         if (num_indices & 1)
+         {
+            const uint key = RG_ETC1_GET_KEY(p);
+
+            hist[key & 0xFF]++;
+         }
+      }      
+
+      T* pCur = pIndices0;
+      T* pNew = pIndices1;
+
+      for (uint pass = 0; pass < key_size; pass++)
+      {
+         const uint* pHist = &hist[pass << 8];
+
+         uint offsets[256];
+
+         uint cur_ofs = 0;
+         for (uint i = 0; i < 256; i += 2)
+         {
+            offsets[i] = cur_ofs;
+            cur_ofs += pHist[i];
+
+            offsets[i+1] = cur_ofs;
+            cur_ofs += pHist[i+1];
+         }
+
+         const uint pass_shift = pass << 3;
+
+         T* p = pCur;
+         T* q = pCur + (num_indices >> 1) * 2;
+
+         for ( ; p != q; p += 2)
+         {
+            uint index0 = p[0];
+            uint index1 = p[1];
+
+            uint c0 = (RG_ETC1_GET_KEY_FROM_INDEX(index0) >> pass_shift) & 0xFF;
+            uint c1 = (RG_ETC1_GET_KEY_FROM_INDEX(index1) >> pass_shift) & 0xFF;
+
+            if (c0 == c1)
+            {
+               uint dst_offset0 = offsets[c0];
+
+               offsets[c0] = dst_offset0 + 2;
+
+               pNew[dst_offset0] = static_cast<T>(index0);
+               pNew[dst_offset0 + 1] = static_cast<T>(index1);
+            }
+            else
+            {
+               uint dst_offset0 = offsets[c0]++;
+               uint dst_offset1 = offsets[c1]++;
+
+               pNew[dst_offset0] = static_cast<T>(index0);
+               pNew[dst_offset1] = static_cast<T>(index1);
+            }
+         }
+
+         if (num_indices & 1)
+         {
+            uint index = *p;
+            uint c = (RG_ETC1_GET_KEY_FROM_INDEX(index) >> pass_shift) & 0xFF;
+
+            uint dst_offset = offsets[c];
+            offsets[c] = dst_offset + 1;
+
+            pNew[dst_offset] = static_cast<T>(index);
+         }
+
+         T* t = pCur;
+         pCur = pNew;
+         pNew = t;
+      }            
+
+      return pCur;
+   }
+
+#undef RG_ETC1_GET_KEY
+#undef RG_ETC1_GET_KEY_FROM_INDEX
+
+   uint16 etc1_block::pack_color5(const color_quad_u8& color, bool scaled, uint bias)
+   {
+      return pack_color5(color.r, color.g, color.b, scaled, bias);
+   }
+   
+   uint16 etc1_block::pack_color5(uint r, uint g, uint b, bool scaled, uint bias)
+   {
+      if (scaled)
+      {
+         r = (r * 31U + bias) / 255U;
+         g = (g * 31U + bias) / 255U;
+         b = (b * 31U + bias) / 255U;
+      }
+
+      r = rg_etc1::minimum(r, 31U);
+      g = rg_etc1::minimum(g, 31U);
+      b = rg_etc1::minimum(b, 31U);
+
+      return static_cast<uint16>(b | (g << 5U) | (r << 10U));
+   }
+
+   color_quad_u8 etc1_block::unpack_color5(uint16 packed_color5, bool scaled, uint alpha)
+   {
+      uint b = packed_color5 & 31U;
+      uint g = (packed_color5 >> 5U) & 31U;
+      uint r = (packed_color5 >> 10U) & 31U;
+
+      if (scaled)
+      {
+         b = (b << 3U) | (b >> 2U);
+         g = (g << 3U) | (g >> 2U);
+         r = (r << 3U) | (r >> 2U);
+      }
+
+      return color_quad_u8(cNoClamp, r, g, b, rg_etc1::minimum(alpha, 255U));
+   }
+
+   void etc1_block::unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, bool scaled)
+   {
+      color_quad_u8 c(unpack_color5(packed_color5, scaled, 0));
+      r = c.r;
+      g = c.g;
+      b = c.b;
+   }
+
+   bool etc1_block::unpack_color5(color_quad_u8& result, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha)
+   {
+      int dc_r, dc_g, dc_b;
+      unpack_delta3(dc_r, dc_g, dc_b, packed_delta3);
+      
+      int b = (packed_color5 & 31U) + dc_b;
+      int g = ((packed_color5 >> 5U) & 31U) + dc_g;
+      int r = ((packed_color5 >> 10U) & 31U) + dc_r;
+
+      bool success = true;
+      if (static_cast<uint>(r | g | b) > 31U)
+      {
+         success = false;
+         r = rg_etc1::clamp<int>(r, 0, 31);
+         g = rg_etc1::clamp<int>(g, 0, 31);
+         b = rg_etc1::clamp<int>(b, 0, 31);
+      }
+
+      if (scaled)
+      {
+         b = (b << 3U) | (b >> 2U);
+         g = (g << 3U) | (g >> 2U);
+         r = (r << 3U) | (r >> 2U);
+      }
+
+      result.set_noclamp_rgba(r, g, b, rg_etc1::minimum(alpha, 255U));
+      return success;
+   }
+
+   bool etc1_block::unpack_color5(uint& r, uint& g, uint& b, uint16 packed_color5, uint16 packed_delta3, bool scaled, uint alpha)
+   {
+      color_quad_u8 result;
+      const bool success = unpack_color5(result, packed_color5, packed_delta3, scaled, alpha);
+      r = result.r;
+      g = result.g;
+      b = result.b;
+      return success;
+   }
+     
+   uint16 etc1_block::pack_delta3(int r, int g, int b)
+   {
+      RG_ETC1_ASSERT((r >= cETC1ColorDeltaMin) && (r <= cETC1ColorDeltaMax));
+      RG_ETC1_ASSERT((g >= cETC1ColorDeltaMin) && (g <= cETC1ColorDeltaMax));
+      RG_ETC1_ASSERT((b >= cETC1ColorDeltaMin) && (b <= cETC1ColorDeltaMax));
+      if (r < 0) r += 8;
+      if (g < 0) g += 8;
+      if (b < 0) b += 8;
+      return static_cast<uint16>(b | (g << 3) | (r << 6));
+   }
+   
+   void etc1_block::unpack_delta3(int& r, int& g, int& b, uint16 packed_delta3)
+   {
+      r = (packed_delta3 >> 6) & 7;
+      g = (packed_delta3 >> 3) & 7;
+      b = packed_delta3 & 7;
+      if (r >= 4) r -= 8;
+      if (g >= 4) g -= 8;
+      if (b >= 4) b -= 8;
+   }
+
+   uint16 etc1_block::pack_color4(const color_quad_u8& color, bool scaled, uint bias)
+   {
+      return pack_color4(color.r, color.g, color.b, scaled, bias);
+   }
+   
+   uint16 etc1_block::pack_color4(uint r, uint g, uint b, bool scaled, uint bias)
+   {
+      if (scaled)
+      {
+         r = (r * 15U + bias) / 255U;
+         g = (g * 15U + bias) / 255U;
+         b = (b * 15U + bias) / 255U;
+      }
+
+      r = rg_etc1::minimum(r, 15U);
+      g = rg_etc1::minimum(g, 15U);
+      b = rg_etc1::minimum(b, 15U);
+
+      return static_cast<uint16>(b | (g << 4U) | (r << 8U));
+   }
+
+   color_quad_u8 etc1_block::unpack_color4(uint16 packed_color4, bool scaled, uint alpha)
+   {
+      uint b = packed_color4 & 15U;
+      uint g = (packed_color4 >> 4U) & 15U;
+      uint r = (packed_color4 >> 8U) & 15U;
+
+      if (scaled)
+      {
+         b = (b << 4U) | b;
+         g = (g << 4U) | g;
+         r = (r << 4U) | r;
+      }
+
+      return color_quad_u8(cNoClamp, r, g, b, rg_etc1::minimum(alpha, 255U));
+   }
+   
+   void etc1_block::unpack_color4(uint& r, uint& g, uint& b, uint16 packed_color4, bool scaled)
+   {
+      color_quad_u8 c(unpack_color4(packed_color4, scaled, 0));
+      r = c.r;
+      g = c.g;
+      b = c.b;
+   }
+
+   void etc1_block::get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint table_idx)
+   {
+      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);
+      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];
+
+      uint r, g, b;
+      unpack_color5(r, g, b, packed_color5, true);
+
+      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);
+
+      const int y0 = pInten_modifer_table[0];
+      pDst[0].set(ir + y0, ig + y0, ib + y0);
+
+      const int y1 = pInten_modifer_table[1];
+      pDst[1].set(ir + y1, ig + y1, ib + y1);
+
+      const int y2 = pInten_modifer_table[2];
+      pDst[2].set(ir + y2, ig + y2, ib + y2);
+
+      const int y3 = pInten_modifer_table[3];
+      pDst[3].set(ir + y3, ig + y3, ib + y3);
+   }
+   
+   bool etc1_block::get_diff_subblock_colors(color_quad_u8* pDst, uint16 packed_color5, uint16 packed_delta3, uint table_idx)
+   {
+      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);
+      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];
+
+      uint r, g, b;
+      bool success = unpack_color5(r, g, b, packed_color5, packed_delta3, true);
+
+      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);
+
+      const int y0 = pInten_modifer_table[0];
+      pDst[0].set(ir + y0, ig + y0, ib + y0);
+
+      const int y1 = pInten_modifer_table[1];
+      pDst[1].set(ir + y1, ig + y1, ib + y1);
+
+      const int y2 = pInten_modifer_table[2];
+      pDst[2].set(ir + y2, ig + y2, ib + y2);
+
+      const int y3 = pInten_modifer_table[3];
+      pDst[3].set(ir + y3, ig + y3, ib + y3);
+
+      return success;
+   }
+   
+   void etc1_block::get_abs_subblock_colors(color_quad_u8* pDst, uint16 packed_color4, uint table_idx)
+   {
+      RG_ETC1_ASSERT(table_idx < cETC1IntenModifierValues);
+      const int *pInten_modifer_table = &g_etc1_inten_tables[table_idx][0];
+
+      uint r, g, b;
+      unpack_color4(r, g, b, packed_color4, true);
+      
+      const int ir = static_cast<int>(r), ig = static_cast<int>(g), ib = static_cast<int>(b);
+
+      const int y0 = pInten_modifer_table[0];
+      pDst[0].set(ir + y0, ig + y0, ib + y0);
+      
+      const int y1 = pInten_modifer_table[1];
+      pDst[1].set(ir + y1, ig + y1, ib + y1);
+
+      const int y2 = pInten_modifer_table[2];
+      pDst[2].set(ir + y2, ig + y2, ib + y2);
+
+      const int y3 = pInten_modifer_table[3];
+      pDst[3].set(ir + y3, ig + y3, ib + y3);
+   }
+      
+   bool unpack_etc1_block(const void* pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha)
+   {
+      color_quad_u8* pDst = reinterpret_cast<color_quad_u8*>(pDst_pixels_rgba);
+      const etc1_block& block = *static_cast<const etc1_block*>(pETC1_block);
+
+      const bool diff_flag = block.get_diff_bit();
+      const bool flip_flag = block.get_flip_bit();
+      const uint table_index0 = block.get_inten_table(0);
+      const uint table_index1 = block.get_inten_table(1);
+
+      color_quad_u8 subblock_colors0[4];
+      color_quad_u8 subblock_colors1[4];
+      bool success = true;
+
+      if (diff_flag)
+      {
+         const uint16 base_color5 = block.get_base5_color();
+         const uint16 delta_color3 = block.get_delta3_color();
+         etc1_block::get_diff_subblock_colors(subblock_colors0, base_color5, table_index0);
+            
+         if (!etc1_block::get_diff_subblock_colors(subblock_colors1, base_color5, delta_color3, table_index1))
+            success = false;
+      }
+      else
+      {
+         const uint16 base_color4_0 = block.get_base4_color(0);
+         etc1_block::get_abs_subblock_colors(subblock_colors0, base_color4_0, table_index0);
+
+         const uint16 base_color4_1 = block.get_base4_color(1);
+         etc1_block::get_abs_subblock_colors(subblock_colors1, base_color4_1, table_index1);
+      }
+
+      if (preserve_alpha)
+      {
+         if (flip_flag)
+         {
+            for (uint y = 0; y < 2; y++)
+            {
+               pDst[0].set_rgb(subblock_colors0[block.get_selector(0, y)]);
+               pDst[1].set_rgb(subblock_colors0[block.get_selector(1, y)]);
+               pDst[2].set_rgb(subblock_colors0[block.get_selector(2, y)]);
+               pDst[3].set_rgb(subblock_colors0[block.get_selector(3, y)]);
+               pDst += 4;
+            }
+
+            for (uint y = 2; y < 4; y++)
+            {
+               pDst[0].set_rgb(subblock_colors1[block.get_selector(0, y)]);
+               pDst[1].set_rgb(subblock_colors1[block.get_selector(1, y)]);
+               pDst[2].set_rgb(subblock_colors1[block.get_selector(2, y)]);
+               pDst[3].set_rgb(subblock_colors1[block.get_selector(3, y)]);
+               pDst += 4;
+            }
+         }
+         else
+         {
+            for (uint y = 0; y < 4; y++)
+            {
+               pDst[0].set_rgb(subblock_colors0[block.get_selector(0, y)]);
+               pDst[1].set_rgb(subblock_colors0[block.get_selector(1, y)]);
+               pDst[2].set_rgb(subblock_colors1[block.get_selector(2, y)]);
+               pDst[3].set_rgb(subblock_colors1[block.get_selector(3, y)]);
+               pDst += 4;
+            }
+         }
+      }
+      else 
+      {
+         if (flip_flag)
+         {
+            // 0000
+            // 0000
+            // 1111
+            // 1111
+            for (uint y = 0; y < 2; y++)
+            {
+               pDst[0] = subblock_colors0[block.get_selector(0, y)];
+               pDst[1] = subblock_colors0[block.get_selector(1, y)];
+               pDst[2] = subblock_colors0[block.get_selector(2, y)];
+               pDst[3] = subblock_colors0[block.get_selector(3, y)];
+               pDst += 4;
+            }
+
+            for (uint y = 2; y < 4; y++)
+            {
+               pDst[0] = subblock_colors1[block.get_selector(0, y)];
+               pDst[1] = subblock_colors1[block.get_selector(1, y)];
+               pDst[2] = subblock_colors1[block.get_selector(2, y)];
+               pDst[3] = subblock_colors1[block.get_selector(3, y)];
+               pDst += 4;
+            }
+         }
+         else
+         {
+            // 0011
+            // 0011
+            // 0011
+            // 0011
+            for (uint y = 0; y < 4; y++)
+            {
+               pDst[0] = subblock_colors0[block.get_selector(0, y)];
+               pDst[1] = subblock_colors0[block.get_selector(1, y)];
+               pDst[2] = subblock_colors1[block.get_selector(2, y)];
+               pDst[3] = subblock_colors1[block.get_selector(3, y)];
+               pDst += 4;
+            }
+         }
+      }
+      
+      return success;
+   }
+
+   struct etc1_solution_coordinates
+   {
+      inline etc1_solution_coordinates() :
+      m_unscaled_color(0, 0, 0, 0),
+         m_inten_table(0),
+         m_color4(false)
+      {
+      }
+
+      inline etc1_solution_coordinates(uint r, uint g, uint b, uint inten_table, bool color4) : 
+      m_unscaled_color(r, g, b, 255),
+         m_inten_table(inten_table),
+         m_color4(color4)
+      {
+      }
+
+      inline etc1_solution_coordinates(const color_quad_u8& c, uint inten_table, bool color4) : 
+      m_unscaled_color(c),
+         m_inten_table(inten_table),
+         m_color4(color4)
+      {
+      }
+
+      inline etc1_solution_coordinates(const etc1_solution_coordinates& other)
+      {
+         *this = other;
+      }
+
+      inline etc1_solution_coordinates& operator= (const etc1_solution_coordinates& rhs)
+      {
+         m_unscaled_color = rhs.m_unscaled_color;
+         m_inten_table = rhs.m_inten_table;
+         m_color4 = rhs.m_color4;
+         return *this;
+      }
+
+      inline void clear()
+      {
+         m_unscaled_color.clear();
+         m_inten_table = 0;
+         m_color4 = false;
+      }
+
+      inline color_quad_u8 get_scaled_color() const
+      {
+         int br, bg, bb;
+         if (m_color4)
+         {
+            br = m_unscaled_color.r | (m_unscaled_color.r << 4);
+            bg = m_unscaled_color.g | (m_unscaled_color.g << 4);
+            bb = m_unscaled_color.b | (m_unscaled_color.b << 4);
+         }
+         else
+         {
+            br = (m_unscaled_color.r >> 2) | (m_unscaled_color.r << 3);
+            bg = (m_unscaled_color.g >> 2) | (m_unscaled_color.g << 3);
+            bb = (m_unscaled_color.b >> 2) | (m_unscaled_color.b << 3);
+         }
+         return color_quad_u8(br, bg, bb);
+      }
+
+      inline void get_block_colors(color_quad_u8* pBlock_colors)
+      {
+         int br, bg, bb;
+         if (m_color4)
+         {
+            br = m_unscaled_color.r | (m_unscaled_color.r << 4);
+            bg = m_unscaled_color.g | (m_unscaled_color.g << 4);
+            bb = m_unscaled_color.b | (m_unscaled_color.b << 4);
+         }
+         else
+         {
+            br = (m_unscaled_color.r >> 2) | (m_unscaled_color.r << 3);
+            bg = (m_unscaled_color.g >> 2) | (m_unscaled_color.g << 3);
+            bb = (m_unscaled_color.b >> 2) | (m_unscaled_color.b << 3);
+         }
+         const int* pInten_table = g_etc1_inten_tables[m_inten_table];
+         pBlock_colors[0].set(br + pInten_table[0], bg + pInten_table[0], bb + pInten_table[0]);
+         pBlock_colors[1].set(br + pInten_table[1], bg + pInten_table[1], bb + pInten_table[1]);
+         pBlock_colors[2].set(br + pInten_table[2], bg + pInten_table[2], bb + pInten_table[2]);
+         pBlock_colors[3].set(br + pInten_table[3], bg + pInten_table[3], bb + pInten_table[3]);
+      }
+
+      color_quad_u8 m_unscaled_color;
+      uint m_inten_table;
+      bool m_color4;
+   };
+
+   class etc1_optimizer
+   {
+      etc1_optimizer(const etc1_optimizer&);
+      etc1_optimizer& operator= (const etc1_optimizer&);
+
+   public:
+      etc1_optimizer()
+      {
+         clear();
+      }
+
+      void clear()
+      {
+         m_pParams = NULL;
+         m_pResult = NULL;
+         m_pSorted_luma = NULL;
+         m_pSorted_luma_indices = NULL;
+      }
+
+      struct params : etc1_pack_params
+      {
+         params()
+         {
+            clear();
+         }
+
+         params(const etc1_pack_params& base_params) : 
+         etc1_pack_params(base_params)
+         {
+            clear_optimizer_params();
+         }
+
+         void clear()
+         {
+            etc1_pack_params::clear();
+            clear_optimizer_params();
+         }
+
+         void clear_optimizer_params()
+         {
+            m_num_src_pixels = 0;
+            m_pSrc_pixels = 0;
+
+            m_use_color4 = false;
+            static const int s_default_scan_delta[] = { 0 };
+            m_pScan_deltas = s_default_scan_delta;
+            m_scan_delta_size = 1;
+
+            m_base_color5.clear();
+            m_constrain_against_base_color5 = false;
+         }
+
+         uint m_num_src_pixels;
+         const color_quad_u8* m_pSrc_pixels;
+
+         bool m_use_color4;
+         const int* m_pScan_deltas;
+         uint m_scan_delta_size;
+
+         color_quad_u8 m_base_color5;
+         bool m_constrain_against_base_color5;
+      };
+
+      struct results
+      {
+         uint64 m_error;
+         color_quad_u8 m_block_color_unscaled;
+         uint m_block_inten_table;
+         uint m_n;
+         uint8* m_pSelectors;
+         bool m_block_color4;
+
+         inline results& operator= (const results& rhs)
+         {
+            m_block_color_unscaled = rhs.m_block_color_unscaled;
+            m_block_color4 = rhs.m_block_color4;
+            m_block_inten_table = rhs.m_block_inten_table;
+            m_error = rhs.m_error;
+            RG_ETC1_ASSERT(m_n == rhs.m_n);
+            memcpy(m_pSelectors, rhs.m_pSelectors, rhs.m_n);
+            return *this;
+         }
+      };
+
+      void init(const params& params, results& result);
+      bool compute();
+
+   private:      
+      struct potential_solution
+      {
+         potential_solution() : m_coords(), m_error(cUINT64_MAX), m_valid(false)
+         {
+         }
+
+         etc1_solution_coordinates  m_coords;
+         uint8                      m_selectors[8];
+         uint64                     m_error;
+         bool                       m_valid;
+
+         void clear()
+         {
+            m_coords.clear();
+            m_error = cUINT64_MAX;
+            m_valid = false;
+         }
+      };
+
+      const params* m_pParams;
+      results* m_pResult;
+
+      int m_limit;
+
+      vec3F m_avg_color;
+      int m_br, m_bg, m_bb;
+      uint16 m_luma[8];
+      uint32 m_sorted_luma[2][8];
+      const uint32* m_pSorted_luma_indices;
+      uint32* m_pSorted_luma;
+
+      uint8 m_selectors[8];
+      uint8 m_best_selectors[8];
+
+      potential_solution m_best_solution;
+      potential_solution m_trial_solution;
+      uint8 m_temp_selectors[8];
+
+      bool evaluate_solution(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution);
+      bool evaluate_solution_fast(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution);
+   };
+      
+   bool etc1_optimizer::compute()
+   {
+      const uint n = m_pParams->m_num_src_pixels;
+      const int scan_delta_size = m_pParams->m_scan_delta_size;
+      
+      // Scan through a subset of the 3D lattice centered around the avg block color trying each 3D (555 or 444) lattice point as a potential block color.
+      // Each time a better solution is found try to refine the current solution's block color based of the current selectors and intensity table index.
+      for (int zdi = 0; zdi < scan_delta_size; zdi++)
+      {
+         const int zd = m_pParams->m_pScan_deltas[zdi];
+         const int mbb = m_bb + zd;
+         if (mbb < 0) continue; else if (mbb > m_limit) break;
+         
+         for (int ydi = 0; ydi < scan_delta_size; ydi++)
+         {
+            const int yd = m_pParams->m_pScan_deltas[ydi];
+            const int mbg = m_bg + yd;
+            if (mbg < 0) continue; else if (mbg > m_limit) break;
+
+            for (int xdi = 0; xdi < scan_delta_size; xdi++)
+            {
+               const int xd = m_pParams->m_pScan_deltas[xdi];
+               const int mbr = m_br + xd;
+               if (mbr < 0) continue; else if (mbr > m_limit) break;
+      
+               etc1_solution_coordinates coords(mbr, mbg, mbb, 0, m_pParams->m_use_color4);
+               if (m_pParams->m_quality == cHighQuality)
+               {
+                  if (!evaluate_solution(coords, m_trial_solution, &m_best_solution))
+                     continue;
+               }
+               else
+               {
+                  if (!evaluate_solution_fast(coords, m_trial_solution, &m_best_solution))
+                     continue;
+               }
+               
+               // Now we have the input block, the avg. color of the input pixels, a set of trial selector indices, and the block color+intensity index.
+               // Now, for each component, attempt to refine the current solution by solving a simple linear equation. For example, for 4 colors:
+               // The goal is:
+               // pixel0 - (block_color+inten_table[selector0]) + pixel1 - (block_color+inten_table[selector1]) + pixel2 - (block_color+inten_table[selector2]) + pixel3 - (block_color+inten_table[selector3]) = 0
+               // Rearranging this:
+               // (pixel0 + pixel1 + pixel2 + pixel3) - (block_color+inten_table[selector0]) - (block_color+inten_table[selector1]) - (block_color+inten_table[selector2]) - (block_color+inten_table[selector3]) = 0
+               // (pixel0 + pixel1 + pixel2 + pixel3) - block_color - inten_table[selector0] - block_color-inten_table[selector1] - block_color-inten_table[selector2] - block_color-inten_table[selector3] = 0
+               // (pixel0 + pixel1 + pixel2 + pixel3) - 4*block_color - inten_table[selector0] - inten_table[selector1] - inten_table[selector2] - inten_table[selector3] = 0
+               // (pixel0 + pixel1 + pixel2 + pixel3) - 4*block_color - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3]) = 0
+               // (pixel0 + pixel1 + pixel2 + pixel3)/4 - block_color - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3])/4 = 0
+               // block_color = (pixel0 + pixel1 + pixel2 + pixel3)/4 - (inten_table[selector0] + inten_table[selector1] + inten_table[selector2] + inten_table[selector3])/4
+               // So what this means:
+               // optimal_block_color = avg_input - avg_inten_delta
+               // So the optimal block color can be computed by taking the average block color and subtracting the current average of the intensity delta.
+               // Unfortunately, optimal_block_color must then be quantized to 555 or 444 so it's not always possible to improve matters using this formula.
+               // Also, the above formula is for unclamped intensity deltas. The actual implementation takes into account clamping.
+
+               const uint max_refinement_trials = (m_pParams->m_quality == cLowQuality) ? 2 : (((xd | yd | zd) == 0) ? 4 : 2);
+               for (uint refinement_trial = 0; refinement_trial < max_refinement_trials; refinement_trial++)
+               {
+                  const uint8* pSelectors = m_best_solution.m_selectors;
+                  const int* pInten_table = g_etc1_inten_tables[m_best_solution.m_coords.m_inten_table];
+
+                  int delta_sum_r = 0, delta_sum_g = 0, delta_sum_b = 0;
+                  const color_quad_u8 base_color(m_best_solution.m_coords.get_scaled_color());
+                  for (uint r = 0; r < n; r++)
+                  {
+                     const uint s = *pSelectors++;
+                     const int yd = pInten_table[s];
+                     // Compute actual delta being applied to each pixel, taking into account clamping.
+                     delta_sum_r += rg_etc1::clamp<int>(base_color.r + yd, 0, 255) - base_color.r;
+                     delta_sum_g += rg_etc1::clamp<int>(base_color.g + yd, 0, 255) - base_color.g;
+                     delta_sum_b += rg_etc1::clamp<int>(base_color.b + yd, 0, 255) - base_color.b;
+                  }
+                  if ((!delta_sum_r) && (!delta_sum_g) && (!delta_sum_b))
+                     break;
+                  const float avg_delta_r_f = static_cast<float>(delta_sum_r) / n;
+                  const float avg_delta_g_f = static_cast<float>(delta_sum_g) / n;
+                  const float avg_delta_b_f = static_cast<float>(delta_sum_b) / n;
+                  const int br1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[0] - avg_delta_r_f) * m_limit / 255.0f + .5f), 0, m_limit);
+                  const int bg1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[1] - avg_delta_g_f) * m_limit / 255.0f + .5f), 0, m_limit);
+                  const int bb1 = rg_etc1::clamp<int>(static_cast<uint>((m_avg_color[2] - avg_delta_b_f) * m_limit / 255.0f + .5f), 0, m_limit);
+                  
+                  bool skip = false;
+                  
+                  if ((mbr == br1) && (mbg == bg1) && (mbb == bb1))
+                     skip = true;
+                  else if ((br1 == m_best_solution.m_coords.m_unscaled_color.r) && (bg1 == m_best_solution.m_coords.m_unscaled_color.g) && (bb1 == m_best_solution.m_coords.m_unscaled_color.b))
+                     skip = true;
+                  else if ((m_br == br1) && (m_bg == bg1) && (m_bb == bb1))
+                     skip = true;
+
+                  if (skip)
+                     break;
+
+                  etc1_solution_coordinates coords1(br1, bg1, bb1, 0, m_pParams->m_use_color4);
+                  if (m_pParams->m_quality == cHighQuality)
+                  {
+                     if (!evaluate_solution(coords1, m_trial_solution, &m_best_solution)) 
+                        break;
+                  }
+                  else
+                  {
+                     if (!evaluate_solution_fast(coords1, m_trial_solution, &m_best_solution))
+                        break;
+                  }
+
+               }  // refinement_trial
+
+            } // xdi
+         } // ydi
+      } // zdi
+
+      if (!m_best_solution.m_valid)
+      {
+         m_pResult->m_error = cUINT32_MAX;
+         return false;
+      }
+      
+      const uint8* pSelectors = m_best_solution.m_selectors;
+
+#ifdef RG_ETC1_BUILD_DEBUG
+      {
+         color_quad_u8 block_colors[4];
+         m_best_solution.m_coords.get_block_colors(block_colors);
+
+         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;
+         uint64 actual_error = 0;
+         for (uint i = 0; i < n; i++)
+            actual_error += pSrc_pixels[i].squared_distance_rgb(block_colors[pSelectors[i]]);
+         
+         RG_ETC1_ASSERT(actual_error == m_best_solution.m_error);
+      }
+#endif      
+      
+      m_pResult->m_error = m_best_solution.m_error;
+
+      m_pResult->m_block_color_unscaled = m_best_solution.m_coords.m_unscaled_color;
+      m_pResult->m_block_color4 = m_best_solution.m_coords.m_color4;
+      
+      m_pResult->m_block_inten_table = m_best_solution.m_coords.m_inten_table;
+      memcpy(m_pResult->m_pSelectors, pSelectors, n);
+      m_pResult->m_n = n;
+
+      return true;
+   }
+
+   void etc1_optimizer::init(const params& p, results& r)
+   {
+      // This version is hardcoded for 8 pixel subblocks.
+      RG_ETC1_ASSERT(p.m_num_src_pixels == 8);
+      
+      m_pParams = &p;
+      m_pResult = &r;
+                  
+      const uint n = 8;
+      
+      m_limit = m_pParams->m_use_color4 ? 15 : 31;
+
+      vec3F avg_color(0.0f);
+
+      for (uint i = 0; i < n; i++)
+      {
+         const color_quad_u8& c = m_pParams->m_pSrc_pixels[i];
+         const vec3F fc(c.r, c.g, c.b);
+
+         avg_color += fc;
+
+         m_luma[i] = static_cast<uint16>(c.r + c.g + c.b);
+         m_sorted_luma[0][i] = i;
+      }
+      avg_color *= (1.0f / static_cast<float>(n));
+      m_avg_color = avg_color;
+
+      m_br = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[0] * m_limit / 255.0f + .5f), 0, m_limit);
+      m_bg = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[1] * m_limit / 255.0f + .5f), 0, m_limit);
+      m_bb = rg_etc1::clamp<int>(static_cast<uint>(m_avg_color[2] * m_limit / 255.0f + .5f), 0, m_limit);
+
+      if (m_pParams->m_quality <= cMediumQuality)
+      {
+         m_pSorted_luma_indices = indirect_radix_sort(n, m_sorted_luma[0], m_sorted_luma[1], m_luma, 0, sizeof(m_luma[0]), false);
+         m_pSorted_luma = m_sorted_luma[0];
+         if (m_pSorted_luma_indices == m_sorted_luma[0])
+            m_pSorted_luma = m_sorted_luma[1];
+      
+         for (uint i = 0; i < n; i++)
+            m_pSorted_luma[i] = m_luma[m_pSorted_luma_indices[i]];
+      }
+      
+      m_best_solution.m_coords.clear();
+      m_best_solution.m_valid = false;
+      m_best_solution.m_error = cUINT64_MAX;
+   }
+
+   bool etc1_optimizer::evaluate_solution(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution)
+   {
+      trial_solution.m_valid = false;
+
+      if (m_pParams->m_constrain_against_base_color5)
+      {
+         const int dr = coords.m_unscaled_color.r - m_pParams->m_base_color5.r;
+         const int dg = coords.m_unscaled_color.g - m_pParams->m_base_color5.g;
+         const int db = coords.m_unscaled_color.b - m_pParams->m_base_color5.b;
+
+         if ((rg_etc1::minimum(dr, dg, db) < cETC1ColorDeltaMin) || (rg_etc1::maximum(dr, dg, db) > cETC1ColorDeltaMax))
+            return false;
+      }
+
+      const color_quad_u8 base_color(coords.get_scaled_color());
+      
+      const uint n = 8;
+            
+      trial_solution.m_error = cUINT64_MAX;
+            
+      for (uint inten_table = 0; inten_table < cETC1IntenModifierValues; inten_table++)
+      {
+         const int* pInten_table = g_etc1_inten_tables[inten_table];
+
+         color_quad_u8 block_colors[4];
+         for (uint s = 0; s < 4; s++)
+         {
+            const int yd = pInten_table[s];
+            block_colors[s].set(base_color.r + yd, base_color.g + yd, base_color.b + yd, 0);
+         }
+         
+         uint64 total_error = 0;
+         
+         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;
+         for (uint c = 0; c < n; c++)
+         {
+            const color_quad_u8& src_pixel = *pSrc_pixels++;
+            
+            uint best_selector_index = 0;
+            uint best_error = rg_etc1::square(src_pixel.r - block_colors[0].r) + rg_etc1::square(src_pixel.g - block_colors[0].g) + rg_etc1::square(src_pixel.b - block_colors[0].b);
+
+            uint trial_error = rg_etc1::square(src_pixel.r - block_colors[1].r) + rg_etc1::square(src_pixel.g - block_colors[1].g) + rg_etc1::square(src_pixel.b - block_colors[1].b);
+            if (trial_error < best_error)
+            {
+               best_error = trial_error;
+               best_selector_index = 1;
+            }
+
+            trial_error = rg_etc1::square(src_pixel.r - block_colors[2].r) + rg_etc1::square(src_pixel.g - block_colors[2].g) + rg_etc1::square(src_pixel.b - block_colors[2].b);
+            if (trial_error < best_error)
+            {
+               best_error = trial_error;
+               best_selector_index = 2;
+            }
+
+            trial_error = rg_etc1::square(src_pixel.r - block_colors[3].r) + rg_etc1::square(src_pixel.g - block_colors[3].g) + rg_etc1::square(src_pixel.b - block_colors[3].b);
+            if (trial_error < best_error)
+            {
+               best_error = trial_error;
+               best_selector_index = 3;
+            }
+
+            m_temp_selectors[c] = static_cast<uint8>(best_selector_index);
+
+            total_error += best_error;
+            if (total_error >= trial_solution.m_error)
+               break;
+         }
+         
+         if (total_error < trial_solution.m_error)
+         {
+            trial_solution.m_error = total_error;
+            trial_solution.m_coords.m_inten_table = inten_table;
+            memcpy(trial_solution.m_selectors, m_temp_selectors, 8);
+            trial_solution.m_valid = true;
+         }
+      }
+      trial_solution.m_coords.m_unscaled_color = coords.m_unscaled_color;
+      trial_solution.m_coords.m_color4 = m_pParams->m_use_color4;
+
+      bool success = false;
+      if (pBest_solution)
+      {
+         if (trial_solution.m_error < pBest_solution->m_error)
+         {
+            *pBest_solution = trial_solution;
+            success = true;
+         }
+      }
+
+      return success;
+   }
+
+   bool etc1_optimizer::evaluate_solution_fast(const etc1_solution_coordinates& coords, potential_solution& trial_solution, potential_solution* pBest_solution)
+   {
+      if (m_pParams->m_constrain_against_base_color5)
+      {
+         const int dr = coords.m_unscaled_color.r - m_pParams->m_base_color5.r;
+         const int dg = coords.m_unscaled_color.g - m_pParams->m_base_color5.g;
+         const int db = coords.m_unscaled_color.b - m_pParams->m_base_color5.b;
+
+         if ((rg_etc1::minimum(dr, dg, db) < cETC1ColorDeltaMin) || (rg_etc1::maximum(dr, dg, db) > cETC1ColorDeltaMax))
+         {
+            trial_solution.m_valid = false;
+            return false;
+         }
+      }
+
+      const color_quad_u8 base_color(coords.get_scaled_color());
+
+      const uint n = 8;
+      
+      trial_solution.m_error = cUINT64_MAX;
+
+      for (int inten_table = cETC1IntenModifierValues - 1; inten_table >= 0; --inten_table)
+      {
+         const int* pInten_table = g_etc1_inten_tables[inten_table];
+
+         uint block_inten[4];
+         color_quad_u8 block_colors[4];
+         for (uint s = 0; s < 4; s++)
+         {
+            const int yd = pInten_table[s];
+            color_quad_u8 block_color(base_color.r + yd, base_color.g + yd, base_color.b + yd, 0);
+            block_colors[s] = block_color;
+            block_inten[s] = block_color.r + block_color.g + block_color.b;
+         }
+
+         // evaluate_solution_fast() enforces/assumesd a total ordering of the input colors along the intensity (1,1,1) axis to more quickly classify the inputs to selectors.
+         // The inputs colors have been presorted along the projection onto this axis, and ETC1 block colors are always ordered along the intensity axis, so this classification is fast.
+         // 0   1   2   3
+         //   01  12  23
+         const uint block_inten_midpoints[3] = { block_inten[0] + block_inten[1], block_inten[1] + block_inten[2], block_inten[2] + block_inten[3] };
+
+         uint64 total_error = 0;
+         const color_quad_u8* pSrc_pixels = m_pParams->m_pSrc_pixels;
+         if ((m_pSorted_luma[n - 1] * 2) < block_inten_midpoints[0])
+         {
+            if (block_inten[0] > m_pSorted_luma[n - 1])
+            {
+           const uint min_error = intabs(block_inten[0] - m_pSorted_luma[n - 1]);
+               if (min_error >= trial_solution.m_error)
+                  continue;
+            }
+
+            memset(&m_temp_selectors[0], 0, n);
+
+            for (uint c = 0; c < n; c++)
+               total_error += block_colors[0].squared_distance_rgb(pSrc_pixels[c]);
+         }
+         else if ((m_pSorted_luma[0] * 2) >= block_inten_midpoints[2])
+         {
+            if (m_pSorted_luma[0] > block_inten[3])
+            {
+           const uint min_error = intabs(m_pSorted_luma[0] - block_inten[3]);
+               if (min_error >= trial_solution.m_error)
+                  continue;
+            }
+
+            memset(&m_temp_selectors[0], 3, n);
+
+            for (uint c = 0; c < n; c++)
+               total_error += block_colors[3].squared_distance_rgb(pSrc_pixels[c]);
+         }
+         else
+         {
+            uint cur_selector = 0, c;
+            for (c = 0; c < n; c++)
+            {
+               const uint y = m_pSorted_luma[c];
+               while ((y * 2) >= block_inten_midpoints[cur_selector])
+                  if (++cur_selector > 2)
+                     goto done;
+               const uint sorted_pixel_index = m_pSorted_luma_indices[c];
+               m_temp_selectors[sorted_pixel_index] = static_cast<uint8>(cur_selector);
+               total_error += block_colors[cur_selector].squared_distance_rgb(pSrc_pixels[sorted_pixel_index]);
+            }
+done:
+            while (c < n)
+            {
+               const uint sorted_pixel_index = m_pSorted_luma_indices[c];
+               m_temp_selectors[sorted_pixel_index] = 3;
+               total_error += block_colors[3].squared_distance_rgb(pSrc_pixels[sorted_pixel_index]);
+               ++c;
+            }
+         }
+
+         if (total_error < trial_solution.m_error)
+         {
+            trial_solution.m_error = total_error;
+            trial_solution.m_coords.m_inten_table = inten_table;
+            memcpy(trial_solution.m_selectors, m_temp_selectors, n);
+            trial_solution.m_valid = true;
+            if (!total_error)
+               break;
+         }
+      }
+      trial_solution.m_coords.m_unscaled_color = coords.m_unscaled_color;
+      trial_solution.m_coords.m_color4 = m_pParams->m_use_color4;
+      
+      bool success = false;
+      if (pBest_solution)
+      {
+         if (trial_solution.m_error < pBest_solution->m_error)
+         {
+            *pBest_solution = trial_solution;
+            success = true;
+         }
+      }
+
+      return success;
+   }
+         
+   static uint etc1_decode_value(uint diff, uint inten, uint selector, uint packed_c)
+   {
+      const uint limit = diff ? 32 : 16; limit;
+      RG_ETC1_ASSERT((diff < 2) && (inten < 8) && (selector < 4) && (packed_c < limit));
+      int c;
+      if (diff)
+         c = (packed_c >> 2) | (packed_c << 3);
+      else 
+         c = packed_c | (packed_c << 4);
+      c += g_etc1_inten_tables[inten][selector];
+      c = rg_etc1::clamp<int>(c, 0, 255);
+      return c;
+   }
+
+   static inline int mul_8bit(int a, int b) { int t = a*b + 128; return (t + (t >> 8)) >> 8; }
+
+   void pack_etc1_block_init()
+   {
+      for (uint diff = 0; diff < 2; diff++)
+      {
+         const uint limit = diff ? 32 : 16;
+
+         for (uint inten = 0; inten < 8; inten++)
+         {
+            for (uint selector = 0; selector < 4; selector++)
+            {
+               const uint inverse_table_index = diff + (inten << 1) + (selector << 4);
+               for (uint color = 0; color < 256; color++)
+               {
+                  uint best_error = cUINT32_MAX, best_packed_c = 0;
+                  for (uint packed_c = 0; packed_c < limit; packed_c++)
+                  {
+                     int v = etc1_decode_value(diff, inten, selector, packed_c);
+                     uint err = labs(v - static_cast<int>(color));
+		     //printf("err: %d - %u = %u\n",v,color,err);
+                     if (err < best_error)
+                     {
+                        best_error = err;
+                        best_packed_c = packed_c;
+                        if (!best_error) 
+                           break;
+                     }
+                  }
+                  RG_ETC1_ASSERT(best_error <= 255);
+                  g_etc1_inverse_lookup[inverse_table_index][color] = static_cast<uint16>(best_packed_c | (best_error << 8));
+               }
+            }
+         }
+      }
+      
+      uint expand5[32];
+      for(int i = 0; i < 32; i++)
+         expand5[i] = (i << 3) | (i >> 2);
+
+      for(int i = 0; i < 256 + 16; i++)
+      {
+         int v = clamp<int>(i - 8, 0, 255);
+         g_quant5_tab[i] = static_cast<uint8>(expand5[mul_8bit(v,31)]);
+      }
+   }
+
+   // Packs solid color blocks efficiently using a set of small precomputed tables.
+   // For random 888 inputs, MSE results are better than Erricson's ETC1 packer in "slow" mode ~9.5% of the time, is slightly worse only ~.01% of the time, and is equal the rest of the time.
+   static uint64 pack_etc1_block_solid_color(etc1_block& block, const uint8* pColor, etc1_pack_params& pack_params)
+   {
+      pack_params;
+      RG_ETC1_ASSERT(g_etc1_inverse_lookup[0][255]);
+            
+      static uint s_next_comp[4] = { 1, 2, 0, 1 };
+            
+      uint best_error = cUINT32_MAX, best_i = 0;
+      int best_x = 0, best_packed_c1 = 0, best_packed_c2 = 0;
+
+      // For each possible 8-bit value, there is a precomputed list of diff/inten/selector configurations that allow that 8-bit value to be encoded with no error.
+      for (uint i = 0; i < 3; i++)
+      {
+         const uint c1 = pColor[s_next_comp[i]], c2 = pColor[s_next_comp[i + 1]];
+
+         const int delta_range = 1;
+         for (int delta = -delta_range; delta <= delta_range; delta++)
+         {
+            const int c_plus_delta = rg_etc1::clamp<int>(pColor[i] + delta, 0, 255);
+
+            const uint16* pTable;
+            if (!c_plus_delta)
+               pTable = g_color8_to_etc_block_config_0_255[0];
+            else if (c_plus_delta == 255)
+               pTable = g_color8_to_etc_block_config_0_255[1];
+            else
+               pTable = g_color8_to_etc_block_config_1_to_254[c_plus_delta - 1];
+
+            do
+            {
+               const uint x = *pTable++;
+
+#ifdef RG_ETC1_BUILD_DEBUG
+               const uint diff = x & 1;
+               const uint inten = (x >> 1) & 7;
+               const uint selector = (x >> 4) & 3;
+               const uint p0 = (x >> 8) & 255;
+               RG_ETC1_ASSERT(etc1_decode_value(diff, inten, selector, p0) == (uint)c_plus_delta);
+#endif
+
+               const uint16* pInverse_table = g_etc1_inverse_lookup[x & 0xFF];
+               uint16 p1 = pInverse_table[c1];
+               uint16 p2 = pInverse_table[c2];
+               const uint trial_error = rg_etc1::square(c_plus_delta - pColor[i]) + rg_etc1::square(p1 >> 8) + rg_etc1::square(p2 >> 8);
+               if (trial_error < best_error)
+               {
+                  best_error = trial_error;
+                  best_x = x;
+                  best_packed_c1 = p1 & 0xFF;
+                  best_packed_c2 = p2 & 0xFF;
+                  best_i = i;
+                  if (!best_error)
+                     goto found_perfect_match;
+               }
+            } while (*pTable != 0xFFFF);
+         }
+      }
+found_perfect_match:
+
+      const uint diff = best_x & 1;
+      const uint inten = (best_x >> 1) & 7;
+
+      block.m_bytes[3] = static_cast<uint8>(((inten | (inten << 3)) << 2) | (diff << 1));
+                        
+      const uint etc1_selector = g_selector_index_to_etc1[(best_x >> 4) & 3];
+      *reinterpret_cast<uint16*>(&block.m_bytes[4]) = (etc1_selector & 2) ? 0xFFFF : 0;
+      *reinterpret_cast<uint16*>(&block.m_bytes[6]) = (etc1_selector & 1) ? 0xFFFF : 0;
+
+      const uint best_packed_c0 = (best_x >> 8) & 255;
+      if (diff)
+      {
+         block.m_bytes[best_i] = static_cast<uint8>(best_packed_c0 << 3);
+         block.m_bytes[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1 << 3);
+         block.m_bytes[s_next_comp[best_i+1]] = static_cast<uint8>(best_packed_c2 << 3);
+      }
+      else
+      {
+         block.m_bytes[best_i] = static_cast<uint8>(best_packed_c0 | (best_packed_c0 << 4));
+         block.m_bytes[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1 | (best_packed_c1 << 4));
+         block.m_bytes[s_next_comp[best_i+1]] = static_cast<uint8>(best_packed_c2 | (best_packed_c2 << 4));
+      }
+
+      return best_error;
+   }
+      
+   static uint pack_etc1_block_solid_color_constrained(
+      etc1_optimizer::results& results, 
+      uint num_colors, const uint8* pColor, 
+      etc1_pack_params& pack_params, 
+      bool use_diff,
+      const color_quad_u8* pBase_color5_unscaled)
+   {
+      RG_ETC1_ASSERT(g_etc1_inverse_lookup[0][255]);
+
+      pack_params;
+      static uint s_next_comp[4] = { 1, 2, 0, 1 };
+
+      uint best_error = cUINT32_MAX, best_i = 0;
+      int best_x = 0, best_packed_c1 = 0, best_packed_c2 = 0;
+
+      // For each possible 8-bit value, there is a precomputed list of diff/inten/selector configurations that allow that 8-bit value to be encoded with no error.
+      for (uint i = 0; i < 3; i++)
+      {
+         const uint c1 = pColor[s_next_comp[i]], c2 = pColor[s_next_comp[i + 1]];
+
+         const int delta_range = 1;
+         for (int delta = -delta_range; delta <= delta_range; delta++)
+         {
+            const int c_plus_delta = rg_etc1::clamp<int>(pColor[i] + delta, 0, 255);
+
+            const uint16* pTable;
+            if (!c_plus_delta)
+               pTable = g_color8_to_etc_block_config_0_255[0];
+            else if (c_plus_delta == 255)
+               pTable = g_color8_to_etc_block_config_0_255[1];
+            else
+               pTable = g_color8_to_etc_block_config_1_to_254[c_plus_delta - 1];
+
+            do
+            {
+               const uint x = *pTable++;
+               const uint diff = x & 1;
+               if (static_cast<uint>(use_diff) != diff)
+               {
+                  if (*pTable == 0xFFFF)
+                     break;
+                  continue;
+               }
+
+               if ((diff) && (pBase_color5_unscaled))
+               {
+                  const int p0 = (x >> 8) & 255;
+                  int delta = p0 - static_cast<int>(pBase_color5_unscaled->c[i]);
+                  if ((delta < cETC1ColorDeltaMin) || (delta > cETC1ColorDeltaMax))
+                  {
+                     if (*pTable == 0xFFFF)
+                        break;
+                     continue;
+                  }
+               }
+
+#ifdef RG_ETC1_BUILD_DEBUG
+               {
+                  const uint inten = (x >> 1) & 7;
+                  const uint selector = (x >> 4) & 3;
+                  const uint p0 = (x >> 8) & 255;
+                  RG_ETC1_ASSERT(etc1_decode_value(diff, inten, selector, p0) == (uint)c_plus_delta);
+               }
+#endif
+
+               const uint16* pInverse_table = g_etc1_inverse_lookup[x & 0xFF];
+               uint16 p1 = pInverse_table[c1];
+               uint16 p2 = pInverse_table[c2];
+
+               if ((diff) && (pBase_color5_unscaled))
+               {
+                  int delta1 = (p1 & 0xFF) - static_cast<int>(pBase_color5_unscaled->c[s_next_comp[i]]);
+                  int delta2 = (p2 & 0xFF) - static_cast<int>(pBase_color5_unscaled->c[s_next_comp[i + 1]]);
+                  if ((delta1 < cETC1ColorDeltaMin) || (delta1 > cETC1ColorDeltaMax) || (delta2 < cETC1ColorDeltaMin) || (delta2 > cETC1ColorDeltaMax))
+                  {
+                     if (*pTable == 0xFFFF)
+                        break;
+                     continue;
+                  }
+               }
+
+               const uint trial_error = rg_etc1::square(c_plus_delta - pColor[i]) + rg_etc1::square(p1 >> 8) + rg_etc1::square(p2 >> 8);
+               if (trial_error < best_error)
+               {
+                  best_error = trial_error;
+                  best_x = x;
+                  best_packed_c1 = p1 & 0xFF;
+                  best_packed_c2 = p2 & 0xFF;
+                  best_i = i;
+                  if (!best_error)
+                     goto found_perfect_match;
+               }
+            } while (*pTable != 0xFFFF);
+         }
+      }
+found_perfect_match:
+
+      if (best_error == cUINT32_MAX)
+         return best_error;
+
+      best_error *= num_colors;
+
+      results.m_n = num_colors;
+      results.m_block_color4 = !(best_x & 1);
+      results.m_block_inten_table = (best_x >> 1) & 7;
+      memset(results.m_pSelectors, (best_x >> 4) & 3, num_colors);
+
+      const uint best_packed_c0 = (best_x >> 8) & 255;
+      results.m_block_color_unscaled[best_i] = static_cast<uint8>(best_packed_c0);
+      results.m_block_color_unscaled[s_next_comp[best_i]] = static_cast<uint8>(best_packed_c1);
+      results.m_block_color_unscaled[s_next_comp[best_i + 1]] = static_cast<uint8>(best_packed_c2);
+      results.m_error = best_error;
+      
+      return best_error;
+   }
+
+   // Function originally from RYG's public domain real-time DXT1 compressor, modified for 555.
+   static void dither_block_555(color_quad_u8* dest, const color_quad_u8* block)
+   {
+      int err[8],*ep1 = err,*ep2 = err+4;
+      uint8 *quant = g_quant5_tab+8;
+
+      memset(dest, 0xFF, sizeof(color_quad_u8)*16);
+
+      // process channels seperately
+      for(int ch=0;ch<3;ch++)
+      {
+         uint8* bp = (uint8*)block;
+         uint8* dp = (uint8*)dest;
+
+         bp += ch; dp += ch;
+
+         memset(err,0, sizeof(err));
+         for(int y = 0; y < 4; y++)
+         {
+            // pixel 0
+            dp[ 0] = quant[bp[ 0] + ((3*ep2[1] + 5*ep2[0]) >> 4)];
+            ep1[0] = bp[ 0] - dp[ 0];
+
+            // pixel 1
+            dp[ 4] = quant[bp[ 4] + ((7*ep1[0] + 3*ep2[2] + 5*ep2[1] + ep2[0]) >> 4)];
+            ep1[1] = bp[ 4] - dp[ 4];
+
+            // pixel 2
+            dp[ 8] = quant[bp[ 8] + ((7*ep1[1] + 3*ep2[3] + 5*ep2[2] + ep2[1]) >> 4)];
+            ep1[2] = bp[ 8] - dp[ 8];
+
+            // pixel 3
+            dp[12] = quant[bp[12] + ((7*ep1[2] + 5*ep2[3] + ep2[2]) >> 4)];
+            ep1[3] = bp[12] - dp[12];
+
+            // advance to next line
+            int* tmp = ep1; ep1 = ep2; ep2 = tmp;
+            bp += 16;
+            dp += 16;
+         }
+      }
+   }
+
+   unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params)
+   {
+      const color_quad_u8* pSrc_pixels = reinterpret_cast<const color_quad_u8*>(pSrc_pixels_rgba);
+      etc1_block& dst_block = *static_cast<etc1_block*>(pETC1_block);
+
+#ifdef RG_ETC1_BUILD_DEBUG
+      // Ensure all alpha values are 0xFF.
+      for (uint i = 0; i < 16; i++)
+      {
+         RG_ETC1_ASSERT(pSrc_pixels[i].a == 255);
+      }
+#endif
+
+      color_quad_u8 src_pixel0(pSrc_pixels[0]);
+
+      // Check for solid block.
+      const uint32 first_pixel_u32 = pSrc_pixels->m_u32;
+      int r;
+      for (r = 15; r >= 1; --r)
+         if (pSrc_pixels[r].m_u32 != first_pixel_u32)
+            break;
+      if (!r)
+         return static_cast<unsigned int>(16 * pack_etc1_block_solid_color(dst_block, &pSrc_pixels[0].r, pack_params));
+      
+      color_quad_u8 dithered_pixels[16];
+      if (pack_params.m_dithering)
+      {
+         dither_block_555(dithered_pixels, pSrc_pixels);
+         pSrc_pixels = dithered_pixels;
+      }
+
+      etc1_optimizer optimizer;
+
+      uint64 best_error = cUINT64_MAX;
+      uint best_flip = false, best_use_color4 = false;
+      
+      uint8 best_selectors[2][8];
+      etc1_optimizer::results best_results[2];
+      for (uint i = 0; i < 2; i++)
+      {
+         best_results[i].m_n = 8;
+         best_results[i].m_pSelectors = best_selectors[i];
+      }
+      
+      uint8 selectors[3][8];
+      etc1_optimizer::results results[3];
+      
+      for (uint i = 0; i < 3; i++)
+      {
+         results[i].m_n = 8;
+         results[i].m_pSelectors = selectors[i];
+      }
+            
+      color_quad_u8 subblock_pixels[8];
+
+      etc1_optimizer::params params(pack_params);
+      params.m_num_src_pixels = 8;
+      params.m_pSrc_pixels = subblock_pixels;
+
+      for (uint flip = 0; flip < 2; flip++)
+      {
+         for (uint use_color4 = 0; use_color4 < 2; use_color4++)
+         {
+            uint64 trial_error = 0;
+
+            uint subblock;
+            for (subblock = 0; subblock < 2; subblock++)
+            {
+               if (flip)
+                  memcpy(subblock_pixels, pSrc_pixels + subblock * 8, sizeof(color_quad_u8) * 8);
+               else
+               {
+                  const color_quad_u8* pSrc_col = pSrc_pixels + subblock * 2;
+                  subblock_pixels[0] = pSrc_col[0]; subblock_pixels[1] = pSrc_col[4]; subblock_pixels[2] = pSrc_col[8]; subblock_pixels[3] = pSrc_col[12];
+                  subblock_pixels[4] = pSrc_col[1]; subblock_pixels[5] = pSrc_col[5]; subblock_pixels[6] = pSrc_col[9]; subblock_pixels[7] = pSrc_col[13];
+               }
+
+               results[2].m_error = cUINT64_MAX;
+               if ((params.m_quality >= cMediumQuality) && ((subblock) || (use_color4)))
+               {
+                  const uint32 subblock_pixel0_u32 = subblock_pixels[0].m_u32;
+                  for (r = 7; r >= 1; --r)
+                     if (subblock_pixels[r].m_u32 != subblock_pixel0_u32)
+                        break;
+                  if (!r)
+                  {
+                     pack_etc1_block_solid_color_constrained(results[2], 8, &subblock_pixels[0].r, pack_params, !use_color4, (subblock && !use_color4) ? &results[0].m_block_color_unscaled : NULL);
+                  }
+               }
+
+               params.m_use_color4 = (use_color4 != 0);
+               params.m_constrain_against_base_color5 = false;
+
+               if ((!use_color4) && (subblock))
+               {
+                  params.m_constrain_against_base_color5 = true;
+                  params.m_base_color5 = results[0].m_block_color_unscaled;
+               }
+                              
+               if (params.m_quality == cHighQuality)
+               {
+                  static const int s_scan_delta_0_to_4[] = { -4, -3, -2, -1, 0, 1, 2, 3, 4 };
+                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0_to_4);
+                  params.m_pScan_deltas = s_scan_delta_0_to_4;
+               }
+               else if (params.m_quality == cMediumQuality)
+               {
+                  static const int s_scan_delta_0_to_1[] = { -1, 0, 1 };
+                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0_to_1);
+                  params.m_pScan_deltas = s_scan_delta_0_to_1;
+               }
+               else
+               {
+                  static const int s_scan_delta_0[] = { 0 };
+                  params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_0);
+                  params.m_pScan_deltas = s_scan_delta_0;
+               }
+               
+               optimizer.init(params, results[subblock]);
+               if (!optimizer.compute())
+                  break;
+                              
+               if (params.m_quality >= cMediumQuality)
+               {
+                  // TODO: Fix fairly arbitrary/unrefined thresholds that control how far away to scan for potentially better solutions.
+                  const uint refinement_error_thresh0 = 3000;
+                  const uint refinement_error_thresh1 = 6000;
+                  if (results[subblock].m_error > refinement_error_thresh0)
+                  {
+                     if (params.m_quality == cMediumQuality)
+                     {
+                        static const int s_scan_delta_2_to_3[] = { -3, -2, 2, 3 };
+                        params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_2_to_3);
+                        params.m_pScan_deltas = s_scan_delta_2_to_3;
+                     }
+                     else
+                     {
+                        static const int s_scan_delta_5_to_5[] = { -5, 5 };
+                        static const int s_scan_delta_5_to_8[] = { -8, -7, -6, -5, 5, 6, 7, 8 };
+                        if (results[subblock].m_error > refinement_error_thresh1)
+                        {
+                           params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_5_to_8);
+                           params.m_pScan_deltas = s_scan_delta_5_to_8;
+                        }
+                        else
+                        {
+                           params.m_scan_delta_size = RG_ETC1_ARRAY_SIZE(s_scan_delta_5_to_5);
+                           params.m_pScan_deltas = s_scan_delta_5_to_5;
+                        }
+                     }
+
+                     if (!optimizer.compute())
+                        break;
+                  }
+
+                  if (results[2].m_error < results[subblock].m_error)
+                     results[subblock] = results[2];
+               }
+                            
+               trial_error += results[subblock].m_error;
+               if (trial_error >= best_error)
+                  break;
+            }
+
+            if (subblock < 2)
+               continue;
+
+            best_error = trial_error;
+            best_results[0] = results[0];
+            best_results[1] = results[1];
+            best_flip = flip;
+            best_use_color4 = use_color4;
+            
+         } // use_color4
+
+      } // flip
+
+      int dr = best_results[1].m_block_color_unscaled.r - best_results[0].m_block_color_unscaled.r;
+      int dg = best_results[1].m_block_color_unscaled.g - best_results[0].m_block_color_unscaled.g;
+      int db = best_results[1].m_block_color_unscaled.b - best_results[0].m_block_color_unscaled.b;
+      RG_ETC1_ASSERT(best_use_color4 || ((rg_etc1::minimum(dr, dg, db) >= cETC1ColorDeltaMin) && (rg_etc1::maximum(dr, dg, db) <= cETC1ColorDeltaMax)));
+           
+      if (best_use_color4)
+      {
+         dst_block.m_bytes[0] = static_cast<uint8>(best_results[1].m_block_color_unscaled.r | (best_results[0].m_block_color_unscaled.r << 4));
+         dst_block.m_bytes[1] = static_cast<uint8>(best_results[1].m_block_color_unscaled.g | (best_results[0].m_block_color_unscaled.g << 4));
+         dst_block.m_bytes[2] = static_cast<uint8>(best_results[1].m_block_color_unscaled.b | (best_results[0].m_block_color_unscaled.b << 4));
+      }
+      else
+      {
+         if (dr < 0) dr += 8; dst_block.m_bytes[0] = static_cast<uint8>((best_results[0].m_block_color_unscaled.r << 3) | dr);
+         if (dg < 0) dg += 8; dst_block.m_bytes[1] = static_cast<uint8>((best_results[0].m_block_color_unscaled.g << 3) | dg);
+         if (db < 0) db += 8; dst_block.m_bytes[2] = static_cast<uint8>((best_results[0].m_block_color_unscaled.b << 3) | db);
+      }
+      
+      dst_block.m_bytes[3] = static_cast<uint8>( (best_results[1].m_block_inten_table << 2) | (best_results[0].m_block_inten_table << 5) | ((~best_use_color4 & 1) << 1) | best_flip );
+      
+      uint selector0 = 0, selector1 = 0;
+      if (best_flip)
+      {
+         // flipped:
+         // { 0, 0 }, { 1, 0 }, { 2, 0 }, { 3, 0 },               
+         // { 0, 1 }, { 1, 1 }, { 2, 1 }, { 3, 1 } 
+         //
+         // { 0, 2 }, { 1, 2 }, { 2, 2 }, { 3, 2 },
+         // { 0, 3 }, { 1, 3 }, { 2, 3 }, { 3, 3 }
+         const uint8* pSelectors0 = best_results[0].m_pSelectors;
+         const uint8* pSelectors1 = best_results[1].m_pSelectors;
+         for (int x = 3; x >= 0; --x)
+         {
+            uint b;
+            b = g_selector_index_to_etc1[pSelectors1[4 + x]];
+            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+            b = g_selector_index_to_etc1[pSelectors1[x]];
+            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+            b = g_selector_index_to_etc1[pSelectors0[4 + x]];
+            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+            b = g_selector_index_to_etc1[pSelectors0[x]];
+            selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+         }
+      }
+      else
+      {
+         // non-flipped:
+         // { 0, 0 }, { 0, 1 }, { 0, 2 }, { 0, 3 },
+         // { 1, 0 }, { 1, 1 }, { 1, 2 }, { 1, 3 }
+         //
+         // { 2, 0 }, { 2, 1 }, { 2, 2 }, { 2, 3 },
+         // { 3, 0 }, { 3, 1 }, { 3, 2 }, { 3, 3 }
+         for (int subblock = 1; subblock >= 0; --subblock)
+         {
+            const uint8* pSelectors = best_results[subblock].m_pSelectors + 4;
+            for (uint i = 0; i < 2; i++)
+            {
+               uint b;
+               b = g_selector_index_to_etc1[pSelectors[3]];
+               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+               b = g_selector_index_to_etc1[pSelectors[2]];
+               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+               b = g_selector_index_to_etc1[pSelectors[1]];
+               selector0 = (selector0 << 1) | (b & 1); selector1 = (selector1 << 1) | (b >> 1);
+
+               b = g_selector_index_to_etc1[pSelectors[0]];
+               selector0 = (selector0 << 1) | (b & 1);selector1 = (selector1 << 1) | (b >> 1);
+
+               pSelectors -= 4;
+            }
+         }
+      }
+                  
+      dst_block.m_bytes[4] = static_cast<uint8>(selector1 >> 8); dst_block.m_bytes[5] = static_cast<uint8>(selector1 & 0xFF);
+      dst_block.m_bytes[6] = static_cast<uint8>(selector0 >> 8); dst_block.m_bytes[7] = static_cast<uint8>(selector0 & 0xFF);
+
+      return static_cast<unsigned int>(best_error);
+   }
+
+} // namespace rg_etc1
diff --git a/drivers/etc1/rg_etc1.h b/drivers/etc1/rg_etc1.h
index 9a701506fd..9ce89a6cc6 100644
--- a/drivers/etc1/rg_etc1.h
+++ b/drivers/etc1/rg_etc1.h
@@ -1,76 +1,76 @@
-// File: rg_etc1.h - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>

-// Please see ZLIB license at the end of this file.

-#pragma once

-

-namespace rg_etc1

-{

-   // Unpacks an 8-byte ETC1 compressed block to a block of 4x4 32bpp RGBA pixels.

-   // Returns false if the block is invalid. Invalid blocks will still be unpacked with clamping.

-   // This function is thread safe, and does not dynamically allocate any memory.

-   // If preserve_alpha is true, the alpha channel of the destination pixels will not be overwritten. Otherwise, alpha will be set to 255.

-   bool unpack_etc1_block(const void *pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha = false);

-

-   // Quality setting = the higher the quality, the slower. 

-   // To pack large textures, it is highly recommended to call pack_etc1_block() in parallel, on different blocks, from multiple threads (particularly when using cHighQuality).

-   enum etc1_quality

-   { 

-      cLowQuality,

-      cMediumQuality,

-      cHighQuality,

-   };

-      

-   struct etc1_pack_params

-   {

-      etc1_quality m_quality;

-      bool m_dithering;

-                              

-      inline etc1_pack_params() 

-      {

-         clear();

-      }

-

-      void clear()

-      {

-         m_quality = cHighQuality;

-         m_dithering = false;

-      }

-   };

-

-   // Important: pack_etc1_block_init() must be called before calling pack_etc1_block().

-   void pack_etc1_block_init();

-

-   // Packs a 4x4 block of 32bpp RGBA pixels to an 8-byte ETC1 block.

-   // 32-bit RGBA pixels must always be arranged as (R,G,B,A) (R first, A last) in memory, independent of platform endianness. A should always be 255.

-   // Returns squared error of result.

-   // This function is thread safe, and does not dynamically allocate any memory.

-   // pack_etc1_block() does not currently support "perceptual" colorspace metrics - it primarily optimizes for RGB RMSE.

-   unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params);

-            

-} // namespace rg_etc1

-

-//------------------------------------------------------------------------------

-//

-// rg_etc1 uses the ZLIB license:

-// http://opensource.org/licenses/Zlib

-//

-// Copyright (c) 2012 Rich Geldreich

-//

-// This software is provided 'as-is', without any express or implied

-// warranty.  In no event will the authors be held liable for any damages

-// arising from the use of this software.

-//

-// Permission is granted to anyone to use this software for any purpose,

-// including commercial applications, and to alter it and redistribute it

-// freely, subject to the following restrictions:

-//

-// 1. The origin of this software must not be misrepresented; you must not

-// claim that you wrote the original software. If you use this software

-// in a product, an acknowledgment in the product documentation would be

-// appreciated but is not required.

-//

-// 2. Altered source versions must be plainly marked as such, and must not be

-// misrepresented as being the original software.

-//

-// 3. This notice may not be removed or altered from any source distribution.

-//

-//------------------------------------------------------------------------------

+// File: rg_etc1.h - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>
+// Please see ZLIB license at the end of this file.
+#pragma once
+
+namespace rg_etc1
+{
+   // Unpacks an 8-byte ETC1 compressed block to a block of 4x4 32bpp RGBA pixels.
+   // Returns false if the block is invalid. Invalid blocks will still be unpacked with clamping.
+   // This function is thread safe, and does not dynamically allocate any memory.
+   // If preserve_alpha is true, the alpha channel of the destination pixels will not be overwritten. Otherwise, alpha will be set to 255.
+   bool unpack_etc1_block(const void *pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha = false);
+
+   // Quality setting = the higher the quality, the slower. 
+   // To pack large textures, it is highly recommended to call pack_etc1_block() in parallel, on different blocks, from multiple threads (particularly when using cHighQuality).
+   enum etc1_quality
+   { 
+      cLowQuality,
+      cMediumQuality,
+      cHighQuality,
+   };
+      
+   struct etc1_pack_params
+   {
+      etc1_quality m_quality;
+      bool m_dithering;
+                              
+      inline etc1_pack_params() 
+      {
+         clear();
+      }
+
+      void clear()
+      {
+         m_quality = cHighQuality;
+         m_dithering = false;
+      }
+   };
+
+   // Important: pack_etc1_block_init() must be called before calling pack_etc1_block().
+   void pack_etc1_block_init();
+
+   // Packs a 4x4 block of 32bpp RGBA pixels to an 8-byte ETC1 block.
+   // 32-bit RGBA pixels must always be arranged as (R,G,B,A) (R first, A last) in memory, independent of platform endianness. A should always be 255.
+   // Returns squared error of result.
+   // This function is thread safe, and does not dynamically allocate any memory.
+   // pack_etc1_block() does not currently support "perceptual" colorspace metrics - it primarily optimizes for RGB RMSE.
+   unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params);
+            
+} // namespace rg_etc1
+
+//------------------------------------------------------------------------------
+//
+// rg_etc1 uses the ZLIB license:
+// http://opensource.org/licenses/Zlib
+//
+// Copyright (c) 2012 Rich Geldreich
+//
+// This software is provided 'as-is', without any express or implied
+// warranty.  In no event will the authors be held liable for any damages
+// arising from the use of this software.
+//
+// Permission is granted to anyone to use this software for any purpose,
+// including commercial applications, and to alter it and redistribute it
+// freely, subject to the following restrictions:
+//
+// 1. The origin of this software must not be misrepresented; you must not
+// claim that you wrote the original software. If you use this software
+// in a product, an acknowledgment in the product documentation would be
+// appreciated but is not required.
+//
+// 2. Altered source versions must be plainly marked as such, and must not be
+// misrepresented as being the original software.
+//
+// 3. This notice may not be removed or altered from any source distribution.
+//
+//------------------------------------------------------------------------------
diff --git a/drivers/gles2/rasterizer_gles2.cpp b/drivers/gles2/rasterizer_gles2.cpp
index a3c742577a..3d75ed29f3 100644
--- a/drivers/gles2/rasterizer_gles2.cpp
+++ b/drivers/gles2/rasterizer_gles2.cpp
@@ -4145,7 +4145,7 @@ void RasterizerGLES2::begin_frame() {
 
 	//fragment_lighting=Globals::get_singleton()->get("rasterizer/use_fragment_lighting");
 #ifdef TOOLS_ENABLED
-	canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("rasterizer/use_pixel_snap",false));
+	canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("display/use_2d_pixel_snap",false));
 	shadow_filter=ShadowFilterTechnique(int(Globals::get_singleton()->get("rasterizer/shadow_filter")));
 #endif
 
@@ -10807,7 +10807,7 @@ void RasterizerGLES2::init() {
 	copy_shader.set_conditional(CopyShaderGLES2::USE_8BIT_HDR,!use_fp16_fb);
 	canvas_shader.set_conditional(CanvasShaderGLES2::USE_DEPTH_SHADOWS,read_depth_supported);
 
-	canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("rasterizer/use_pixel_snap",false));
+	canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("display/use_2d_pixel_snap",false));
 
 	npo2_textures_available=true;
 	//fragment_lighting=false;
diff --git a/drivers/nedmalloc/malloc.c.h b/drivers/nedmalloc/malloc.c.h
index b9e65637d5..4fec5cc9d4 100644
--- a/drivers/nedmalloc/malloc.c.h
+++ b/drivers/nedmalloc/malloc.c.h
@@ -1,5814 +1,5814 @@
-#ifdef NEDMALLOC_ENABLED

-/*

-  This is a version (aka dlmalloc) of malloc/free/realloc written by

-  Doug Lea and released to the public domain, as explained at

-  http://creativecommons.org/licenses/publicdomain.  Send questions,

-  comments, complaints, performance data, etc to dl@cs.oswego.edu

-

-* Version 2.8.4 Wed May 27 09:56:23 2009  Doug Lea  (dl at gee)

-

-   Note: There may be an updated version of this malloc obtainable at

-           ftp://gee.cs.oswego.edu/pub/misc/malloc.c

-         Check before installing!

-

-* Quickstart

-

-  This library is all in one file to simplify the most common usage:

-  ftp it, compile it (-O3), and link it into another program. All of

-  the compile-time options default to reasonable values for use on

-  most platforms.  You might later want to step through various

-  compile-time and dynamic tuning options.

-

-  For convenience, an include file for code using this malloc is at:

-     ftp://gee.cs.oswego.edu/pub/misc/malloc-2.8.4.h

-  You don't really need this .h file unless you call functions not

-  defined in your system include files.  The .h file contains only the

-  excerpts from this file needed for using this malloc on ANSI C/C++

-  systems, so long as you haven't changed compile-time options about

-  naming and tuning parameters.  If you do, then you can create your

-  own malloc.h that does include all settings by cutting at the point

-  indicated below. Note that you may already by default be using a C

-  library containing a malloc that is based on some version of this

-  malloc (for example in linux). You might still want to use the one

-  in this file to customize settings or to avoid overheads associated

-  with library versions.

-

-* Vital statistics:

-

-  Supported pointer/size_t representation:       4 or 8 bytes

-       size_t MUST be an unsigned type of the same width as

-       pointers. (If you are using an ancient system that declares

-       size_t as a signed type, or need it to be a different width

-       than pointers, you can use a previous release of this malloc

-       (e.g. 2.7.2) supporting these.)

-

-  Alignment:                                     8 bytes (default)

-       This suffices for nearly all current machines and C compilers.

-       However, you can define MALLOC_ALIGNMENT to be wider than this

-       if necessary (up to 128bytes), at the expense of using more space.

-

-  Minimum overhead per allocated chunk:   4 or  8 bytes (if 4byte sizes)

-                                          8 or 16 bytes (if 8byte sizes)

-       Each malloced chunk has a hidden word of overhead holding size

-       and status information, and additional cross-check word

-       if FOOTERS is defined.

-

-  Minimum allocated size: 4-byte ptrs:  16 bytes    (including overhead)

-                          8-byte ptrs:  32 bytes    (including overhead)

-

-       Even a request for zero bytes (i.e., malloc(0)) returns a

-       pointer to something of the minimum allocatable size.

-       The maximum overhead wastage (i.e., number of extra bytes

-       allocated than were requested in malloc) is less than or equal

-       to the minimum size, except for requests >= mmap_threshold that

-       are serviced via mmap(), where the worst case wastage is about

-       32 bytes plus the remainder from a system page (the minimal

-       mmap unit); typically 4096 or 8192 bytes.

-

-  Security: static-safe; optionally more or less

-       The "security" of malloc refers to the ability of malicious

-       code to accentuate the effects of errors (for example, freeing

-       space that is not currently malloc'ed or overwriting past the

-       ends of chunks) in code that calls malloc.  This malloc

-       guarantees not to modify any memory locations below the base of

-       heap, i.e., static variables, even in the presence of usage

-       errors.  The routines additionally detect most improper frees

-       and reallocs.  All this holds as long as the static bookkeeping

-       for malloc itself is not corrupted by some other means.  This

-       is only one aspect of security -- these checks do not, and

-       cannot, detect all possible programming errors.

-

-       If FOOTERS is defined nonzero, then each allocated chunk

-       carries an additional check word to verify that it was malloced

-       from its space.  These check words are the same within each

-       execution of a program using malloc, but differ across

-       executions, so externally crafted fake chunks cannot be

-       freed. This improves security by rejecting frees/reallocs that

-       could corrupt heap memory, in addition to the checks preventing

-       writes to statics that are always on.  This may further improve

-       security at the expense of time and space overhead.  (Note that

-       FOOTERS may also be worth using with MSPACES.)

-

-       By default detected errors cause the program to abort (calling

-       "abort()"). You can override this to instead proceed past

-       errors by defining PROCEED_ON_ERROR.  In this case, a bad free

-       has no effect, and a malloc that encounters a bad address

-       caused by user overwrites will ignore the bad address by

-       dropping pointers and indices to all known memory. This may

-       be appropriate for programs that should continue if at all

-       possible in the face of programming errors, although they may

-       run out of memory because dropped memory is never reclaimed.

-

-       If you don't like either of these options, you can define

-       CORRUPTION_ERROR_ACTION and USAGE_ERROR_ACTION to do anything

-       else. And if if you are sure that your program using malloc has

-       no errors or vulnerabilities, you can define INSECURE to 1,

-       which might (or might not) provide a small performance improvement.

-

-  Thread-safety: NOT thread-safe unless USE_LOCKS defined

-       When USE_LOCKS is defined, each public call to malloc, free,

-       etc is surrounded with either a pthread mutex or a win32

-       spinlock (depending on WIN32). This is not especially fast, and

-       can be a major bottleneck.  It is designed only to provide

-       minimal protection in concurrent environments, and to provide a

-       basis for extensions.  If you are using malloc in a concurrent

-       program, consider instead using nedmalloc

-       (http://www.nedprod.com/programs/portable/nedmalloc/) or

-       ptmalloc (See http://www.malloc.de), which are derived

-       from versions of this malloc.

-

-  System requirements: Any combination of MORECORE and/or MMAP/MUNMAP

-       This malloc can use unix sbrk or any emulation (invoked using

-       the CALL_MORECORE macro) and/or mmap/munmap or any emulation

-       (invoked using CALL_MMAP/CALL_MUNMAP) to get and release system

-       memory.  On most unix systems, it tends to work best if both

-       MORECORE and MMAP are enabled.  On Win32, it uses emulations

-       based on VirtualAlloc. It also uses common C library functions

-       like memset.

-

-  Compliance: I believe it is compliant with the Single Unix Specification

-       (See http://www.unix.org). Also SVID/XPG, ANSI C, and probably

-       others as well.

-

-* Overview of algorithms

-

-  This is not the fastest, most space-conserving, most portable, or

-  most tunable malloc ever written. However it is among the fastest

-  while also being among the most space-conserving, portable and

-  tunable.  Consistent balance across these factors results in a good

-  general-purpose allocator for malloc-intensive programs.

-

-  In most ways, this malloc is a best-fit allocator. Generally, it

-  chooses the best-fitting existing chunk for a request, with ties

-  broken in approximately least-recently-used order. (This strategy

-  normally maintains low fragmentation.) However, for requests less

-  than 256bytes, it deviates from best-fit when there is not an

-  exactly fitting available chunk by preferring to use space adjacent

-  to that used for the previous small request, as well as by breaking

-  ties in approximately most-recently-used order. (These enhance

-  locality of series of small allocations.)  And for very large requests

-  (>= 256Kb by default), it relies on system memory mapping

-  facilities, if supported.  (This helps avoid carrying around and

-  possibly fragmenting memory used only for large chunks.)

-

-  All operations (except malloc_stats and mallinfo) have execution

-  times that are bounded by a constant factor of the number of bits in

-  a size_t, not counting any clearing in calloc or copying in realloc,

-  or actions surrounding MORECORE and MMAP that have times

-  proportional to the number of non-contiguous regions returned by

-  system allocation routines, which is often just 1. In real-time

-  applications, you can optionally suppress segment traversals using

-  NO_SEGMENT_TRAVERSAL, which assures bounded execution even when

-  system allocators return non-contiguous spaces, at the typical

-  expense of carrying around more memory and increased fragmentation.

-

-  The implementation is not very modular and seriously overuses

-  macros. Perhaps someday all C compilers will do as good a job

-  inlining modular code as can now be done by brute-force expansion,

-  but now, enough of them seem not to.

-

-  Some compilers issue a lot of warnings about code that is

-  dead/unreachable only on some platforms, and also about intentional

-  uses of negation on unsigned types. All known cases of each can be

-  ignored.

-

-  For a longer but out of date high-level description, see

-     http://gee.cs.oswego.edu/dl/html/malloc.html

-

-* MSPACES

-  If MSPACES is defined, then in addition to malloc, free, etc.,

-  this file also defines mspace_malloc, mspace_free, etc. These

-  are versions of malloc routines that take an "mspace" argument

-  obtained using create_mspace, to control all internal bookkeeping.

-  If ONLY_MSPACES is defined, only these versions are compiled.

-  So if you would like to use this allocator for only some allocations,

-  and your system malloc for others, you can compile with

-  ONLY_MSPACES and then do something like...

-    static mspace mymspace = create_mspace(0,0); // for example

-    #define mymalloc(bytes)  mspace_malloc(mymspace, bytes)

-

-  (Note: If you only need one instance of an mspace, you can instead

-  use "USE_DL_PREFIX" to relabel the global malloc.)

-

-  You can similarly create thread-local allocators by storing

-  mspaces as thread-locals. For example:

-    static __thread mspace tlms = 0;

-    void*  tlmalloc(size_t bytes) {

-      if (tlms == 0) tlms = create_mspace(0, 0);

-      return mspace_malloc(tlms, bytes);

-    }

-    void  tlfree(void* mem) { mspace_free(tlms, mem); }

-

-  Unless FOOTERS is defined, each mspace is completely independent.

-  You cannot allocate from one and free to another (although

-  conformance is only weakly checked, so usage errors are not always

-  caught). If FOOTERS is defined, then each chunk carries around a tag

-  indicating its originating mspace, and frees are directed to their

-  originating spaces.

-

- -------------------------  Compile-time options ---------------------------

-

-Be careful in setting #define values for numerical constants of type

-size_t. On some systems, literal values are not automatically extended

-to size_t precision unless they are explicitly casted. You can also

-use the symbolic values MAX_SIZE_T, SIZE_T_ONE, etc below.

-

-WIN32                    default: defined if _WIN32 defined

-  Defining WIN32 sets up defaults for MS environment and compilers.

-  Otherwise defaults are for unix. Beware that there seem to be some

-  cases where this malloc might not be a pure drop-in replacement for

-  Win32 malloc: Random-looking failures from Win32 GDI API's (eg;

-  SetDIBits()) may be due to bugs in some video driver implementations

-  when pixel buffers are malloc()ed, and the region spans more than

-  one VirtualAlloc()ed region. Because dlmalloc uses a small (64Kb)

-  default granularity, pixel buffers may straddle virtual allocation

-  regions more often than when using the Microsoft allocator.  You can

-  avoid this by using VirtualAlloc() and VirtualFree() for all pixel

-  buffers rather than using malloc().  If this is not possible,

-  recompile this malloc with a larger DEFAULT_GRANULARITY.

-

-MALLOC_ALIGNMENT         default: (size_t)8

-  Controls the minimum alignment for malloc'ed chunks.  It must be a

-  power of two and at least 8, even on machines for which smaller

-  alignments would suffice. It may be defined as larger than this

-  though. Note however that code and data structures are optimized for

-  the case of 8-byte alignment.

-

-MSPACES                  default: 0 (false)

-  If true, compile in support for independent allocation spaces.

-  This is only supported if HAVE_MMAP is true.

-

-ONLY_MSPACES             default: 0 (false)

-  If true, only compile in mspace versions, not regular versions.

-

-USE_LOCKS                default: 0 (false)

-  Causes each call to each public routine to be surrounded with

-  pthread or WIN32 mutex lock/unlock. (If set true, this can be

-  overridden on a per-mspace basis for mspace versions.) If set to a

-  non-zero value other than 1, locks are used, but their

-  implementation is left out, so lock functions must be supplied manually,

-  as described below.

-

-USE_SPIN_LOCKS           default: 1 iff USE_LOCKS and on x86 using gcc or MSC

-  If true, uses custom spin locks for locking. This is currently

-  supported only for x86 platforms using gcc or recent MS compilers.

-  Otherwise, posix locks or win32 critical sections are used.

-

-FOOTERS                  default: 0

-  If true, provide extra checking and dispatching by placing

-  information in the footers of allocated chunks. This adds

-  space and time overhead.

-

-INSECURE                 default: 0

-  If true, omit checks for usage errors and heap space overwrites.

-

-USE_DL_PREFIX            default: NOT defined

-  Causes compiler to prefix all public routines with the string 'dl'.

-  This can be useful when you only want to use this malloc in one part

-  of a program, using your regular system malloc elsewhere.

-

-ABORT                    default: defined as abort()

-  Defines how to abort on failed checks.  On most systems, a failed

-  check cannot die with an "assert" or even print an informative

-  message, because the underlying print routines in turn call malloc,

-  which will fail again.  Generally, the best policy is to simply call

-  abort(). It's not very useful to do more than this because many

-  errors due to overwriting will show up as address faults (null, odd

-  addresses etc) rather than malloc-triggered checks, so will also

-  abort.  Also, most compilers know that abort() does not return, so

-  can better optimize code conditionally calling it.

-

-PROCEED_ON_ERROR           default: defined as 0 (false)

-  Controls whether detected bad addresses cause them to bypassed

-  rather than aborting. If set, detected bad arguments to free and

-  realloc are ignored. And all bookkeeping information is zeroed out

-  upon a detected overwrite of freed heap space, thus losing the

-  ability to ever return it from malloc again, but enabling the

-  application to proceed. If PROCEED_ON_ERROR is defined, the

-  static variable malloc_corruption_error_count is compiled in

-  and can be examined to see if errors have occurred. This option

-  generates slower code than the default abort policy.

-

-DEBUG                    default: NOT defined

-  The DEBUG setting is mainly intended for people trying to modify

-  this code or diagnose problems when porting to new platforms.

-  However, it may also be able to better isolate user errors than just

-  using runtime checks.  The assertions in the check routines spell

-  out in more detail the assumptions and invariants underlying the

-  algorithms.  The checking is fairly extensive, and will slow down

-  execution noticeably. Calling malloc_stats or mallinfo with DEBUG

-  set will attempt to check every non-mmapped allocated and free chunk

-  in the course of computing the summaries.

-

-ABORT_ON_ASSERT_FAILURE   default: defined as 1 (true)

-  Debugging assertion failures can be nearly impossible if your

-  version of the assert macro causes malloc to be called, which will

-  lead to a cascade of further failures, blowing the runtime stack.

-  ABORT_ON_ASSERT_FAILURE cause assertions failures to call abort(),

-  which will usually make debugging easier.

-

-MALLOC_FAILURE_ACTION     default: sets errno to ENOMEM, or no-op on win32

-  The action to take before "return 0" when malloc fails to be able to

-  return memory because there is none available.

-

-HAVE_MORECORE             default: 1 (true) unless win32 or ONLY_MSPACES

-  True if this system supports sbrk or an emulation of it.

-

-MORECORE                  default: sbrk

-  The name of the sbrk-style system routine to call to obtain more

-  memory.  See below for guidance on writing custom MORECORE

-  functions. The type of the argument to sbrk/MORECORE varies across

-  systems.  It cannot be size_t, because it supports negative

-  arguments, so it is normally the signed type of the same width as

-  size_t (sometimes declared as "intptr_t").  It doesn't much matter

-  though. Internally, we only call it with arguments less than half

-  the max value of a size_t, which should work across all reasonable

-  possibilities, although sometimes generating compiler warnings.

-

-MORECORE_CONTIGUOUS       default: 1 (true) if HAVE_MORECORE

-  If true, take advantage of fact that consecutive calls to MORECORE

-  with positive arguments always return contiguous increasing

-  addresses.  This is true of unix sbrk. It does not hurt too much to

-  set it true anyway, since malloc copes with non-contiguities.

-  Setting it false when definitely non-contiguous saves time

-  and possibly wasted space it would take to discover this though.

-

-MORECORE_CANNOT_TRIM      default: NOT defined

-  True if MORECORE cannot release space back to the system when given

-  negative arguments. This is generally necessary only if you are

-  using a hand-crafted MORECORE function that cannot handle negative

-  arguments.

-

-NO_SEGMENT_TRAVERSAL       default: 0

-  If non-zero, suppresses traversals of memory segments

-  returned by either MORECORE or CALL_MMAP. This disables

-  merging of segments that are contiguous, and selectively

-  releasing them to the OS if unused, but bounds execution times.

-

-HAVE_MMAP                 default: 1 (true)

-  True if this system supports mmap or an emulation of it.  If so, and

-  HAVE_MORECORE is not true, MMAP is used for all system

-  allocation. If set and HAVE_MORECORE is true as well, MMAP is

-  primarily used to directly allocate very large blocks. It is also

-  used as a backup strategy in cases where MORECORE fails to provide

-  space from system. Note: A single call to MUNMAP is assumed to be

-  able to unmap memory that may have be allocated using multiple calls

-  to MMAP, so long as they are adjacent.

-

-HAVE_MREMAP               default: 1 on linux, else 0

-  If true realloc() uses mremap() to re-allocate large blocks and

-  extend or shrink allocation spaces.

-

-MMAP_CLEARS               default: 1 except on WINCE.

-  True if mmap clears memory so calloc doesn't need to. This is true

-  for standard unix mmap using /dev/zero and on WIN32 except for WINCE.

-

-USE_BUILTIN_FFS            default: 0 (i.e., not used)

-  Causes malloc to use the builtin ffs() function to compute indices.

-  Some compilers may recognize and intrinsify ffs to be faster than the

-  supplied C version. Also, the case of x86 using gcc is special-cased

-  to an asm instruction, so is already as fast as it can be, and so

-  this setting has no effect. Similarly for Win32 under recent MS compilers.

-  (On most x86s, the asm version is only slightly faster than the C version.)

-

-malloc_getpagesize         default: derive from system includes, or 4096.

-  The system page size. To the extent possible, this malloc manages

-  memory from the system in page-size units.  This may be (and

-  usually is) a function rather than a constant. This is ignored

-  if WIN32, where page size is determined using getSystemInfo during

-  initialization. This may be several megabytes if ENABLE_LARGE_PAGES

-  is enabled.

-

-ENABLE_LARGE_PAGES         default: NOT defined

-  Causes the system page size to be the value of GetLargePageMinimum()

-  if that function is available (Windows Server 2003/Vista or later).

-  This allows the use of large page entries in the MMU which can

-  significantly improve performance in large working set applications

-  as TLB cache load is reduced by a factor of three. Note that enabling

-  this option is equal to locking the process' memory in current

-  implementations of Windows and requires the SE_LOCK_MEMORY_PRIVILEGE

-  to be held by the process in order to succeed.

-

-USE_DEV_RANDOM             default: 0 (i.e., not used)

-  Causes malloc to use /dev/random to initialize secure magic seed for

-  stamping footers. Otherwise, the current time is used.

-

-NO_MALLINFO                default: 0

-  If defined, don't compile "mallinfo". This can be a simple way

-  of dealing with mismatches between system declarations and

-  those in this file.

-

-MALLINFO_FIELD_TYPE        default: size_t

-  The type of the fields in the mallinfo struct. This was originally

-  defined as "int" in SVID etc, but is more usefully defined as

-  size_t. The value is used only if  HAVE_USR_INCLUDE_MALLOC_H is not set

-

-REALLOC_ZERO_BYTES_FREES    default: not defined

-  This should be set if a call to realloc with zero bytes should

-  be the same as a call to free. Some people think it should. Otherwise,

-  since this malloc returns a unique pointer for malloc(0), so does

-  realloc(p, 0).

-

-LACKS_UNISTD_H, LACKS_FCNTL_H, LACKS_SYS_PARAM_H, LACKS_SYS_MMAN_H

-LACKS_STRINGS_H, LACKS_STRING_H, LACKS_SYS_TYPES_H,  LACKS_ERRNO_H

-LACKS_STDLIB_H                default: NOT defined unless on WIN32

-  Define these if your system does not have these header files.

-  You might need to manually insert some of the declarations they provide.

-

-DEFAULT_GRANULARITY        default: page size if MORECORE_CONTIGUOUS,

-                                system_info.dwAllocationGranularity in WIN32,

-                                GetLargePageMinimum() if ENABLE_LARGE_PAGES,

-                                otherwise 64K.

-      Also settable using mallopt(M_GRANULARITY, x)

-  The unit for allocating and deallocating memory from the system.  On

-  most systems with contiguous MORECORE, there is no reason to

-  make this more than a page. However, systems with MMAP tend to

-  either require or encourage larger granularities.  You can increase

-  this value to prevent system allocation functions to be called so

-  often, especially if they are slow.  The value must be at least one

-  page and must be a power of two.  Setting to 0 causes initialization

-  to either page size or win32 region size.  (Note: In previous

-  versions of malloc, the equivalent of this option was called

-  "TOP_PAD")

-

-DEFAULT_GRANULARITY_ALIGNED default: undefined (which means page size)

-  Whether to enforce alignment when allocating and deallocating memory

-  from the system i.e. the base address of all allocations will be

-  aligned to DEFAULT_GRANULARITY if it is set. Note that enabling this carries

-  some overhead as multiple calls must now be made when probing for a valid

-  aligned value, however it does greatly ease the checking for whether

-  a given memory pointer was allocated by this allocator rather than

-  some other.

-

-DEFAULT_TRIM_THRESHOLD    default: 2MB

-      Also settable using mallopt(M_TRIM_THRESHOLD, x)

-  The maximum amount of unused top-most memory to keep before

-  releasing via malloc_trim in free().  Automatic trimming is mainly

-  useful in long-lived programs using contiguous MORECORE.  Because

-  trimming via sbrk can be slow on some systems, and can sometimes be

-  wasteful (in cases where programs immediately afterward allocate

-  more large chunks) the value should be high enough so that your

-  overall system performance would improve by releasing this much

-  memory.  As a rough guide, you might set to a value close to the

-  average size of a process (program) running on your system.

-  Releasing this much memory would allow such a process to run in

-  memory.  Generally, it is worth tuning trim thresholds when a

-  program undergoes phases where several large chunks are allocated

-  and released in ways that can reuse each other's storage, perhaps

-  mixed with phases where there are no such chunks at all. The trim

-  value must be greater than page size to have any useful effect.  To

-  disable trimming completely, you can set to MAX_SIZE_T. Note that the trick

-  some people use of mallocing a huge space and then freeing it at

-  program startup, in an attempt to reserve system memory, doesn't

-  have the intended effect under automatic trimming, since that memory

-  will immediately be returned to the system.

-

-DEFAULT_MMAP_THRESHOLD       default: 256K

-      Also settable using mallopt(M_MMAP_THRESHOLD, x)

-  The request size threshold for using MMAP to directly service a

-  request. Requests of at least this size that cannot be allocated

-  using already-existing space will be serviced via mmap.  (If enough

-  normal freed space already exists it is used instead.)  Using mmap

-  segregates relatively large chunks of memory so that they can be

-  individually obtained and released from the host system. A request

-  serviced through mmap is never reused by any other request (at least

-  not directly; the system may just so happen to remap successive

-  requests to the same locations).  Segregating space in this way has

-  the benefits that: Mmapped space can always be individually released

-  back to the system, which helps keep the system level memory demands

-  of a long-lived program low.  Also, mapped memory doesn't become

-  `locked' between other chunks, as can happen with normally allocated

-  chunks, which means that even trimming via malloc_trim would not

-  release them.  However, it has the disadvantage that the space

-  cannot be reclaimed, consolidated, and then used to service later

-  requests, as happens with normal chunks.  The advantages of mmap

-  nearly always outweigh disadvantages for "large" chunks, but the

-  value of "large" may vary across systems.  The default is an

-  empirically derived value that works well in most systems. You can

-  disable mmap by setting to MAX_SIZE_T.

-

-MAX_RELEASE_CHECK_RATE   default: 4095 unless not HAVE_MMAP

-  The number of consolidated frees between checks to release

-  unused segments when freeing. When using non-contiguous segments,

-  especially with multiple mspaces, checking only for topmost space

-  doesn't always suffice to trigger trimming. To compensate for this,

-  free() will, with a period of MAX_RELEASE_CHECK_RATE (or the

-  current number of segments, if greater) try to release unused

-  segments to the OS when freeing chunks that result in

-  consolidation. The best value for this parameter is a compromise

-  between slowing down frees with relatively costly checks that

-  rarely trigger versus holding on to unused memory. To effectively

-  disable, set to MAX_SIZE_T. This may lead to a very slight speed

-  improvement at the expense of carrying around more memory.

-*/

-

-/* Version identifier to allow people to support multiple versions */

-#ifndef DLMALLOC_VERSION

-#define DLMALLOC_VERSION 20804

-#endif /* DLMALLOC_VERSION */

-

-#ifndef WIN32

-#ifdef _WIN32

-#define WIN32 1

-#endif  /* _WIN32 */

-#ifdef _WIN32_WCE

-#define LACKS_FCNTL_H

-#define WIN32 1

-#endif /* _WIN32_WCE */

-#endif  /* WIN32 */

-#ifdef WIN32

-#define WIN32_LEAN_AND_MEAN

-#include <windows.h>

-#include <tchar.h>

-#define HAVE_MMAP 1

-#define HAVE_MORECORE 0

-#define LACKS_UNISTD_H

-#define LACKS_SYS_PARAM_H

-#define LACKS_SYS_MMAN_H

-#define LACKS_STRING_H

-#define LACKS_STRINGS_H

-#define LACKS_SYS_TYPES_H

-#define LACKS_ERRNO_H

-#ifndef MALLOC_FAILURE_ACTION

-#define MALLOC_FAILURE_ACTION

-#endif /* MALLOC_FAILURE_ACTION */

-#ifdef _WIN32_WCE /* WINCE reportedly does not clear */

-#define MMAP_CLEARS 0

-#else

-#define MMAP_CLEARS 1

-#endif /* _WIN32_WCE */

-#endif  /* WIN32 */

-

-#if defined(DARWIN) || defined(_DARWIN)

-/* Mac OSX docs advise not to use sbrk; it seems better to use mmap */

-#ifndef HAVE_MORECORE

-#define HAVE_MORECORE 0

-#define HAVE_MMAP 1

-/* OSX allocators provide 16 byte alignment */

-#ifndef MALLOC_ALIGNMENT

-#define MALLOC_ALIGNMENT ((size_t)16U)

-#endif

-#endif  /* HAVE_MORECORE */

-#endif  /* DARWIN */

-

-#ifndef LACKS_SYS_TYPES_H

-#include <sys/types.h>  /* For size_t */

-#endif  /* LACKS_SYS_TYPES_H */

-

-#if (defined(__GNUC__) && ((defined(__i386__) || defined(__x86_64__)))) || (defined(_MSC_VER) && _MSC_VER>=1310)

-#define SPIN_LOCKS_AVAILABLE 1

-#else

-#define SPIN_LOCKS_AVAILABLE 0

-#endif

-

-/* The maximum possible size_t value has all bits set */

-#define MAX_SIZE_T           (~(size_t)0)

-

-#ifndef ONLY_MSPACES

-#define ONLY_MSPACES 0     /* define to a value */

-#else

-#define ONLY_MSPACES 1

-#endif  /* ONLY_MSPACES */

-#ifndef MSPACES

-#if ONLY_MSPACES

-#define MSPACES 1

-#else   /* ONLY_MSPACES */

-#define MSPACES 0

-#endif  /* ONLY_MSPACES */

-#endif  /* MSPACES */

-#ifndef MALLOC_ALIGNMENT

-#define MALLOC_ALIGNMENT ((size_t)8U)

-#endif  /* MALLOC_ALIGNMENT */

-#ifndef FOOTERS

-#define FOOTERS 0

-#endif  /* FOOTERS */

-#ifndef ABORT

-#define ABORT  abort()

-#endif  /* ABORT */

-#ifndef ABORT_ON_ASSERT_FAILURE

-#define ABORT_ON_ASSERT_FAILURE 1

-#endif  /* ABORT_ON_ASSERT_FAILURE */

-#ifndef PROCEED_ON_ERROR

-#define PROCEED_ON_ERROR 0

-#endif  /* PROCEED_ON_ERROR */

-#ifndef USE_LOCKS

-#define USE_LOCKS 0

-#endif  /* USE_LOCKS */

-#ifndef USE_SPIN_LOCKS

-#if USE_LOCKS && SPIN_LOCKS_AVAILABLE

-#define USE_SPIN_LOCKS 1

-#else

-#define USE_SPIN_LOCKS 0

-#endif /* USE_LOCKS && SPIN_LOCKS_AVAILABLE. */

-#endif /* USE_SPIN_LOCKS */

-#ifndef INSECURE

-#define INSECURE 0

-#endif  /* INSECURE */

-#ifndef HAVE_MMAP

-#define HAVE_MMAP 1

-#endif  /* HAVE_MMAP */

-#ifndef MMAP_CLEARS

-#define MMAP_CLEARS 1

-#endif  /* MMAP_CLEARS */

-#ifndef HAVE_MREMAP

-#ifdef linux

-#define HAVE_MREMAP 1

-#else   /* linux */

-#define HAVE_MREMAP 0

-#endif  /* linux */

-#endif  /* HAVE_MREMAP */

-#ifndef MALLOC_FAILURE_ACTION

-#define MALLOC_FAILURE_ACTION  errno = ENOMEM;

-#endif  /* MALLOC_FAILURE_ACTION */

-#ifndef HAVE_MORECORE

-#if ONLY_MSPACES

-#define HAVE_MORECORE 0

-#else   /* ONLY_MSPACES */

-#define HAVE_MORECORE 1

-#endif  /* ONLY_MSPACES */

-#endif  /* HAVE_MORECORE */

-#if !HAVE_MORECORE

-#define MORECORE_CONTIGUOUS 0

-#else   /* !HAVE_MORECORE */

-#define MORECORE_DEFAULT sbrk

-#ifndef MORECORE_CONTIGUOUS

-#define MORECORE_CONTIGUOUS 1

-#endif  /* MORECORE_CONTIGUOUS */

-#endif  /* HAVE_MORECORE */

-#ifndef DEFAULT_GRANULARITY

-#if (MORECORE_CONTIGUOUS || defined(WIN32))

-#define DEFAULT_GRANULARITY (0)  /* 0 means to compute in init_mparams */

-#else   /* MORECORE_CONTIGUOUS */

-#define DEFAULT_GRANULARITY ((size_t)64U * (size_t)1024U)

-#endif  /* MORECORE_CONTIGUOUS */

-#endif  /* DEFAULT_GRANULARITY */

-#ifndef DEFAULT_TRIM_THRESHOLD

-#ifndef MORECORE_CANNOT_TRIM

-#define DEFAULT_TRIM_THRESHOLD ((size_t)2U * (size_t)1024U * (size_t)1024U)

-#else   /* MORECORE_CANNOT_TRIM */

-#define DEFAULT_TRIM_THRESHOLD MAX_SIZE_T

-#endif  /* MORECORE_CANNOT_TRIM */

-#endif  /* DEFAULT_TRIM_THRESHOLD */

-#ifndef DEFAULT_MMAP_THRESHOLD

-#if HAVE_MMAP

-#define DEFAULT_MMAP_THRESHOLD ((size_t)256U * (size_t)1024U)

-#else   /* HAVE_MMAP */

-#define DEFAULT_MMAP_THRESHOLD MAX_SIZE_T

-#endif  /* HAVE_MMAP */

-#endif  /* DEFAULT_MMAP_THRESHOLD */

-#ifndef MAX_RELEASE_CHECK_RATE

-#if HAVE_MMAP

-#define MAX_RELEASE_CHECK_RATE 4095

-#else

-#define MAX_RELEASE_CHECK_RATE MAX_SIZE_T

-#endif /* HAVE_MMAP */

-#endif /* MAX_RELEASE_CHECK_RATE */

-#ifndef USE_BUILTIN_FFS

-#define USE_BUILTIN_FFS 0

-#endif  /* USE_BUILTIN_FFS */

-#ifndef USE_DEV_RANDOM

-#define USE_DEV_RANDOM 0

-#endif  /* USE_DEV_RANDOM */

-#ifndef NO_MALLINFO

-#define NO_MALLINFO 0

-#endif  /* NO_MALLINFO */

-#ifndef MALLINFO_FIELD_TYPE

-#define MALLINFO_FIELD_TYPE size_t

-#endif  /* MALLINFO_FIELD_TYPE */

-#ifndef NO_SEGMENT_TRAVERSAL

-#define NO_SEGMENT_TRAVERSAL 0

-#endif /* NO_SEGMENT_TRAVERSAL */

-

-/*

-  mallopt tuning options.  SVID/XPG defines four standard parameter

-  numbers for mallopt, normally defined in malloc.h.  None of these

-  are used in this malloc, so setting them has no effect. But this

-  malloc does support the following options.

-*/

-

-#define M_TRIM_THRESHOLD     (-1)

-#define M_GRANULARITY        (-2)

-#define M_MMAP_THRESHOLD     (-3)

-

-/* ------------------------ Mallinfo declarations ------------------------ */

-

-#if !NO_MALLINFO

-/*

-  This version of malloc supports the standard SVID/XPG mallinfo

-  routine that returns a struct containing usage properties and

-  statistics. It should work on any system that has a

-  /usr/include/malloc.h defining struct mallinfo.  The main

-  declaration needed is the mallinfo struct that is returned (by-copy)

-  by mallinfo().  The malloinfo struct contains a bunch of fields that

-  are not even meaningful in this version of malloc.  These fields are

-  are instead filled by mallinfo() with other numbers that might be of

-  interest.

-

-  HAVE_USR_INCLUDE_MALLOC_H should be set if you have a

-  /usr/include/malloc.h file that includes a declaration of struct

-  mallinfo.  If so, it is included; else a compliant version is

-  declared below.  These must be precisely the same for mallinfo() to

-  work.  The original SVID version of this struct, defined on most

-  systems with mallinfo, declares all fields as ints. But some others

-  define as unsigned long. If your system defines the fields using a

-  type of different width than listed here, you MUST #include your

-  system version and #define HAVE_USR_INCLUDE_MALLOC_H.

-*/

-

-/* #define HAVE_USR_INCLUDE_MALLOC_H */

-

-#ifdef HAVE_USR_INCLUDE_MALLOC_H

-#include "/usr/include/malloc.h"

-#else /* HAVE_USR_INCLUDE_MALLOC_H */

-#ifndef STRUCT_MALLINFO_DECLARED

-#define STRUCT_MALLINFO_DECLARED 1

-struct mallinfo {

-  MALLINFO_FIELD_TYPE arena;    /* non-mmapped space allocated from system */

-  MALLINFO_FIELD_TYPE ordblks;  /* number of free chunks */

-  MALLINFO_FIELD_TYPE smblks;   /* always 0 */

-  MALLINFO_FIELD_TYPE hblks;    /* always 0 */

-  MALLINFO_FIELD_TYPE hblkhd;   /* space in mmapped regions */

-  MALLINFO_FIELD_TYPE usmblks;  /* maximum total allocated space */

-  MALLINFO_FIELD_TYPE fsmblks;  /* always 0 */

-  MALLINFO_FIELD_TYPE uordblks; /* total allocated space */

-  MALLINFO_FIELD_TYPE fordblks; /* total free space */

-  MALLINFO_FIELD_TYPE keepcost; /* releasable (via malloc_trim) space */

-};

-#endif /* STRUCT_MALLINFO_DECLARED */

-#endif /* HAVE_USR_INCLUDE_MALLOC_H */

-#endif /* NO_MALLINFO */

-

-/*

-  Try to persuade compilers to inline. The most critical functions for

-  inlining are defined as macros, so these aren't used for them.

-*/

-

-#ifndef FORCEINLINE

-  #if defined(__GNUC__)

-#define FORCEINLINE __inline __attribute__ ((always_inline))

-  #elif defined(_MSC_VER)

-    #define FORCEINLINE __forceinline

-  #endif

-#endif

-#ifndef NOINLINE

-  #if defined(__GNUC__)

-    #define NOINLINE __attribute__ ((noinline))

-  #elif defined(_MSC_VER)

-    #define NOINLINE __declspec(noinline)

-  #else

-    #define NOINLINE

-  #endif

-#endif

-

-#ifdef __cplusplus

-extern "C" {

-#ifndef FORCEINLINE

- #define FORCEINLINE inline

-#endif

-#endif /* __cplusplus */

-#ifndef FORCEINLINE

- #define FORCEINLINE

-#endif

-

-#if !ONLY_MSPACES

-

-/* ------------------- Declarations of public routines ------------------- */

-

-#ifndef USE_DL_PREFIX

-#define dlcalloc               calloc

-#define dlfree                 free

-#define dlmalloc               malloc

-#define dlmemalign             memalign

-#define dlrealloc              realloc

-#define dlvalloc               valloc

-#define dlpvalloc              pvalloc

-#define dlmallinfo             mallinfo

-#define dlmallopt              mallopt

-#define dlmalloc_trim          malloc_trim

-#define dlmalloc_stats         malloc_stats

-#define dlmalloc_usable_size   malloc_usable_size

-#define dlmalloc_footprint     malloc_footprint

-#define dlmalloc_max_footprint malloc_max_footprint

-#define dlindependent_calloc   independent_calloc

-#define dlindependent_comalloc independent_comalloc

-#endif /* USE_DL_PREFIX */

-

-

-/*

-  malloc(size_t n)

-  Returns a pointer to a newly allocated chunk of at least n bytes, or

-  null if no space is available, in which case errno is set to ENOMEM

-  on ANSI C systems.

-

-  If n is zero, malloc returns a minimum-sized chunk. (The minimum

-  size is 16 bytes on most 32bit systems, and 32 bytes on 64bit

-  systems.)  Note that size_t is an unsigned type, so calls with

-  arguments that would be negative if signed are interpreted as

-  requests for huge amounts of space, which will often fail. The

-  maximum supported value of n differs across systems, but is in all

-  cases less than the maximum representable value of a size_t.

-*/

-void* dlmalloc(size_t);

-

-/*

-  free(void* p)

-  Releases the chunk of memory pointed to by p, that had been previously

-  allocated using malloc or a related routine such as realloc.

-  It has no effect if p is null. If p was not malloced or already

-  freed, free(p) will by default cause the current program to abort.

-*/

-void  dlfree(void*);

-

-/*

-  calloc(size_t n_elements, size_t element_size);

-  Returns a pointer to n_elements * element_size bytes, with all locations

-  set to zero.

-*/

-void* dlcalloc(size_t, size_t);

-

-/*

-  realloc(void* p, size_t n)

-  Returns a pointer to a chunk of size n that contains the same data

-  as does chunk p up to the minimum of (n, p's size) bytes, or null

-  if no space is available.

-

-  The returned pointer may or may not be the same as p. The algorithm

-  prefers extending p in most cases when possible, otherwise it

-  employs the equivalent of a malloc-copy-free sequence.

-

-  If p is null, realloc is equivalent to malloc.

-

-  If space is not available, realloc returns null, errno is set (if on

-  ANSI) and p is NOT freed.

-

-  if n is for fewer bytes than already held by p, the newly unused

-  space is lopped off and freed if possible.  realloc with a size

-  argument of zero (re)allocates a minimum-sized chunk.

-

-  The old unix realloc convention of allowing the last-free'd chunk

-  to be used as an argument to realloc is not supported.

-*/

-

-void* dlrealloc(void*, size_t);

-

-/*

-  memalign(size_t alignment, size_t n);

-  Returns a pointer to a newly allocated chunk of n bytes, aligned

-  in accord with the alignment argument.

-

-  The alignment argument should be a power of two. If the argument is

-  not a power of two, the nearest greater power is used.

-  8-byte alignment is guaranteed by normal malloc calls, so don't

-  bother calling memalign with an argument of 8 or less.

-

-  Overreliance on memalign is a sure way to fragment space.

-*/

-void* dlmemalign(size_t, size_t);

-

-/*

-  valloc(size_t n);

-  Equivalent to memalign(pagesize, n), where pagesize is the page

-  size of the system. If the pagesize is unknown, 4096 is used.

-*/

-void* dlvalloc(size_t);

-

-/*

-  mallopt(int parameter_number, int parameter_value)

-  Sets tunable parameters The format is to provide a

-  (parameter-number, parameter-value) pair.  mallopt then sets the

-  corresponding parameter to the argument value if it can (i.e., so

-  long as the value is meaningful), and returns 1 if successful else

-  0.  To workaround the fact that mallopt is specified to use int,

-  not size_t parameters, the value -1 is specially treated as the

-  maximum unsigned size_t value.

-

-  SVID/XPG/ANSI defines four standard param numbers for mallopt,

-  normally defined in malloc.h.  None of these are use in this malloc,

-  so setting them has no effect. But this malloc also supports other

-  options in mallopt. See below for details.  Briefly, supported

-  parameters are as follows (listed defaults are for "typical"

-  configurations).

-

-  Symbol            param #  default    allowed param values

-  M_TRIM_THRESHOLD     -1   2*1024*1024   any   (-1 disables)

-  M_GRANULARITY        -2     page size   any power of 2 >= page size

-  M_MMAP_THRESHOLD     -3      256*1024   any   (or 0 if no MMAP support)

-*/

-int dlmallopt(int, int);

-

-/*

-  malloc_footprint();

-  Returns the number of bytes obtained from the system.  The total

-  number of bytes allocated by malloc, realloc etc., is less than this

-  value. Unlike mallinfo, this function returns only a precomputed

-  result, so can be called frequently to monitor memory consumption.

-  Even if locks are otherwise defined, this function does not use them,

-  so results might not be up to date.

-*/

-size_t dlmalloc_footprint(void);

-

-/*

-  malloc_max_footprint();

-  Returns the maximum number of bytes obtained from the system. This

-  value will be greater than current footprint if deallocated space

-  has been reclaimed by the system. The peak number of bytes allocated

-  by malloc, realloc etc., is less than this value. Unlike mallinfo,

-  this function returns only a precomputed result, so can be called

-  frequently to monitor memory consumption.  Even if locks are

-  otherwise defined, this function does not use them, so results might

-  not be up to date.

-*/

-size_t dlmalloc_max_footprint(void);

-

-#if !NO_MALLINFO

-/*

-  mallinfo()

-  Returns (by copy) a struct containing various summary statistics:

-

-  arena:     current total non-mmapped bytes allocated from system

-  ordblks:   the number of free chunks

-  smblks:    always zero.

-  hblks:     current number of mmapped regions

-  hblkhd:    total bytes held in mmapped regions

-  usmblks:   the maximum total allocated space. This will be greater

-                than current total if trimming has occurred.

-  fsmblks:   always zero

-  uordblks:  current total allocated space (normal or mmapped)

-  fordblks:  total free space

-  keepcost:  the maximum number of bytes that could ideally be released

-               back to system via malloc_trim. ("ideally" means that

-               it ignores page restrictions etc.)

-

-  Because these fields are ints, but internal bookkeeping may

-  be kept as longs, the reported values may wrap around zero and

-  thus be inaccurate.

-*/

-struct mallinfo dlmallinfo(void);

-#endif /* NO_MALLINFO */

-

-/*

-  independent_calloc(size_t n_elements, size_t element_size, void* chunks[]);

-

-  independent_calloc is similar to calloc, but instead of returning a

-  single cleared space, it returns an array of pointers to n_elements

-  independent elements that can hold contents of size elem_size, each

-  of which starts out cleared, and can be independently freed,

-  realloc'ed etc. The elements are guaranteed to be adjacently

-  allocated (this is not guaranteed to occur with multiple callocs or

-  mallocs), which may also improve cache locality in some

-  applications.

-

-  The "chunks" argument is optional (i.e., may be null, which is

-  probably the most typical usage). If it is null, the returned array

-  is itself dynamically allocated and should also be freed when it is

-  no longer needed. Otherwise, the chunks array must be of at least

-  n_elements in length. It is filled in with the pointers to the

-  chunks.

-

-  In either case, independent_calloc returns this pointer array, or

-  null if the allocation failed.  If n_elements is zero and "chunks"

-  is null, it returns a chunk representing an array with zero elements

-  (which should be freed if not wanted).

-

-  Each element must be individually freed when it is no longer

-  needed. If you'd like to instead be able to free all at once, you

-  should instead use regular calloc and assign pointers into this

-  space to represent elements.  (In this case though, you cannot

-  independently free elements.)

-

-  independent_calloc simplifies and speeds up implementations of many

-  kinds of pools.  It may also be useful when constructing large data

-  structures that initially have a fixed number of fixed-sized nodes,

-  but the number is not known at compile time, and some of the nodes

-  may later need to be freed. For example:

-

-  struct Node { int item; struct Node* next; };

-

-  struct Node* build_list() {

-    struct Node** pool;

-    int n = read_number_of_nodes_needed();

-    if (n <= 0) return 0;

-    pool = (struct Node**)(independent_calloc(n, sizeof(struct Node), 0);

-    if (pool == 0) die();

-    // organize into a linked list...

-    struct Node* first = pool[0];

-    for (i = 0; i < n-1; ++i)

-      pool[i]->next = pool[i+1];

-    free(pool);     // Can now free the array (or not, if it is needed later)

-    return first;

-  }

-*/

-void** dlindependent_calloc(size_t, size_t, void**);

-

-/*

-  independent_comalloc(size_t n_elements, size_t sizes[], void* chunks[]);

-

-  independent_comalloc allocates, all at once, a set of n_elements

-  chunks with sizes indicated in the "sizes" array.    It returns

-  an array of pointers to these elements, each of which can be

-  independently freed, realloc'ed etc. The elements are guaranteed to

-  be adjacently allocated (this is not guaranteed to occur with

-  multiple callocs or mallocs), which may also improve cache locality

-  in some applications.

-

-  The "chunks" argument is optional (i.e., may be null). If it is null

-  the returned array is itself dynamically allocated and should also

-  be freed when it is no longer needed. Otherwise, the chunks array

-  must be of at least n_elements in length. It is filled in with the

-  pointers to the chunks.

-

-  In either case, independent_comalloc returns this pointer array, or

-  null if the allocation failed.  If n_elements is zero and chunks is

-  null, it returns a chunk representing an array with zero elements

-  (which should be freed if not wanted).

-

-  Each element must be individually freed when it is no longer

-  needed. If you'd like to instead be able to free all at once, you

-  should instead use a single regular malloc, and assign pointers at

-  particular offsets in the aggregate space. (In this case though, you

-  cannot independently free elements.)

-

-  independent_comallac differs from independent_calloc in that each

-  element may have a different size, and also that it does not

-  automatically clear elements.

-

-  independent_comalloc can be used to speed up allocation in cases

-  where several structs or objects must always be allocated at the

-  same time.  For example:

-

-  struct Head { ... }

-  struct Foot { ... }

-

-  void send_message(char* msg) {

-    int msglen = strlen(msg);

-    size_t sizes[3] = { sizeof(struct Head), msglen, sizeof(struct Foot) };

-    void* chunks[3];

-    if (independent_comalloc(3, sizes, chunks) == 0)

-      die();

-    struct Head* head = (struct Head*)(chunks[0]);

-    char*        body = (char*)(chunks[1]);

-    struct Foot* foot = (struct Foot*)(chunks[2]);

-    // ...

-  }

-

-  In general though, independent_comalloc is worth using only for

-  larger values of n_elements. For small values, you probably won't

-  detect enough difference from series of malloc calls to bother.

-

-  Overuse of independent_comalloc can increase overall memory usage,

-  since it cannot reuse existing noncontiguous small chunks that

-  might be available for some of the elements.

-*/

-void** dlindependent_comalloc(size_t, size_t*, void**);

-

-

-/*

-  pvalloc(size_t n);

-  Equivalent to valloc(minimum-page-that-holds(n)), that is,

-  round up n to nearest pagesize.

- */

-void*  dlpvalloc(size_t);

-

-/*

-  malloc_trim(size_t pad);

-

-  If possible, gives memory back to the system (via negative arguments

-  to sbrk) if there is unused memory at the `high' end of the malloc

-  pool or in unused MMAP segments. You can call this after freeing

-  large blocks of memory to potentially reduce the system-level memory

-  requirements of a program. However, it cannot guarantee to reduce

-  memory. Under some allocation patterns, some large free blocks of

-  memory will be locked between two used chunks, so they cannot be

-  given back to the system.

-

-  The `pad' argument to malloc_trim represents the amount of free

-  trailing space to leave untrimmed. If this argument is zero, only

-  the minimum amount of memory to maintain internal data structures

-  will be left. Non-zero arguments can be supplied to maintain enough

-  trailing space to service future expected allocations without having

-  to re-obtain memory from the system.

-

-  Malloc_trim returns 1 if it actually released any memory, else 0.

-*/

-int  dlmalloc_trim(size_t);

-

-/*

-  malloc_stats();

-  Prints on stderr the amount of space obtained from the system (both

-  via sbrk and mmap), the maximum amount (which may be more than

-  current if malloc_trim and/or munmap got called), and the current

-  number of bytes allocated via malloc (or realloc, etc) but not yet

-  freed. Note that this is the number of bytes allocated, not the

-  number requested. It will be larger than the number requested

-  because of alignment and bookkeeping overhead. Because it includes

-  alignment wastage as being in use, this figure may be greater than

-  zero even when no user-level chunks are allocated.

-

-  The reported current and maximum system memory can be inaccurate if

-  a program makes other calls to system memory allocation functions

-  (normally sbrk) outside of malloc.

-

-  malloc_stats prints only the most commonly interesting statistics.

-  More information can be obtained by calling mallinfo.

-*/

-void  dlmalloc_stats(void);

-

-#endif /* ONLY_MSPACES */

-

-/*

-  malloc_usable_size(void* p);

-

-  Returns the number of bytes you can actually use in

-  an allocated chunk, which may be more than you requested (although

-  often not) due to alignment and minimum size constraints.

-  You can use this many bytes without worrying about

-  overwriting other allocated objects. This is not a particularly great

-  programming practice. malloc_usable_size can be more useful in

-  debugging and assertions, for example:

-

-  p = malloc(n);

-  assert(malloc_usable_size(p) >= 256);

-*/

-size_t dlmalloc_usable_size(void*);

-

-

-#if MSPACES

-

-/*

-  mspace is an opaque type representing an independent

-  region of space that supports mspace_malloc, etc.

-*/

-typedef void* mspace;

-

-/*

-  create_mspace creates and returns a new independent space with the

-  given initial capacity, or, if 0, the default granularity size.  It

-  returns null if there is no system memory available to create the

-  space.  If argument locked is non-zero, the space uses a separate

-  lock to control access. The capacity of the space will grow

-  dynamically as needed to service mspace_malloc requests.  You can

-  control the sizes of incremental increases of this space by

-  compiling with a different DEFAULT_GRANULARITY or dynamically

-  setting with mallopt(M_GRANULARITY, value).

-*/

-mspace create_mspace(size_t capacity, int locked);

-

-/*

-  destroy_mspace destroys the given space, and attempts to return all

-  of its memory back to the system, returning the total number of

-  bytes freed. After destruction, the results of access to all memory

-  used by the space become undefined.

-*/

-size_t destroy_mspace(mspace msp);

-

-/*

-  create_mspace_with_base uses the memory supplied as the initial base

-  of a new mspace. Part (less than 128*sizeof(size_t) bytes) of this

-  space is used for bookkeeping, so the capacity must be at least this

-  large. (Otherwise 0 is returned.) When this initial space is

-  exhausted, additional memory will be obtained from the system.

-  Destroying this space will deallocate all additionally allocated

-  space (if possible) but not the initial base.

-*/

-mspace create_mspace_with_base(void* base, size_t capacity, int locked);

-

-/*

-  mspace_track_large_chunks controls whether requests for large chunks

-  are allocated in their own untracked mmapped regions, separate from

-  others in this mspace. By default large chunks are not tracked,

-  which reduces fragmentation. However, such chunks are not

-  necessarily released to the system upon destroy_mspace.  Enabling

-  tracking by setting to true may increase fragmentation, but avoids

-  leakage when relying on destroy_mspace to release all memory

-  allocated using this space.  The function returns the previous

-  setting.

-*/

-int mspace_track_large_chunks(mspace msp, int enable);

-

-

-/*

-  mspace_malloc behaves as malloc, but operates within

-  the given space.

-*/

-void* mspace_malloc(mspace msp, size_t bytes);

-

-/*

-  mspace_free behaves as free, but operates within

-  the given space.

-

-  If compiled with FOOTERS==1, mspace_free is not actually needed.

-  free may be called instead of mspace_free because freed chunks from

-  any space are handled by their originating spaces.

-*/

-void mspace_free(mspace msp, void* mem);

-

-/*

-  mspace_realloc behaves as realloc, but operates within

-  the given space.

-

-  If compiled with FOOTERS==1, mspace_realloc is not actually

-  needed.  realloc may be called instead of mspace_realloc because

-  realloced chunks from any space are handled by their originating

-  spaces.

-*/

-void* mspace_realloc(mspace msp, void* mem, size_t newsize);

-

-/*

-  mspace_calloc behaves as calloc, but operates within

-  the given space.

-*/

-void* mspace_calloc(mspace msp, size_t n_elements, size_t elem_size);

-

-/*

-  mspace_memalign behaves as memalign, but operates within

-  the given space.

-*/

-void* mspace_memalign(mspace msp, size_t alignment, size_t bytes);

-

-/*

-  mspace_independent_calloc behaves as independent_calloc, but

-  operates within the given space.

-*/

-void** mspace_independent_calloc(mspace msp, size_t n_elements,

-                                 size_t elem_size, void* chunks[]);

-

-/*

-  mspace_independent_comalloc behaves as independent_comalloc, but

-  operates within the given space.

-*/

-void** mspace_independent_comalloc(mspace msp, size_t n_elements,

-                                   size_t sizes[], void* chunks[]);

-

-/*

-  mspace_footprint() returns the number of bytes obtained from the

-  system for this space.

-*/

-size_t mspace_footprint(mspace msp);

-

-/*

-  mspace_max_footprint() returns the peak number of bytes obtained from the

-  system for this space.

-*/

-size_t mspace_max_footprint(mspace msp);

-

-

-#if !NO_MALLINFO

-/*

-  mspace_mallinfo behaves as mallinfo, but reports properties of

-  the given space.

-*/

-struct mallinfo mspace_mallinfo(mspace msp);

-#endif /* NO_MALLINFO */

-

-/*

-  malloc_usable_size(void* p) behaves the same as malloc_usable_size;

-*/

-  size_t mspace_usable_size(void* mem);

-

-/*

-  mspace_malloc_stats behaves as malloc_stats, but reports

-  properties of the given space.

-*/

-void mspace_malloc_stats(mspace msp);

-

-/*

-  mspace_trim behaves as malloc_trim, but

-  operates within the given space.

-*/

-int mspace_trim(mspace msp, size_t pad);

-

-/*

-  An alias for mallopt.

-*/

-int mspace_mallopt(int, int);

-

-#endif /* MSPACES */

-

-#ifdef __cplusplus

-}  /* end of extern "C" */

-#endif /* __cplusplus */

-

-/*

-  ========================================================================

-  To make a fully customizable malloc.h header file, cut everything

-  above this line, put into file malloc.h, edit to suit, and #include it

-  on the next line, as well as in programs that use this malloc.

-  ========================================================================

-*/

-

-/* #include "malloc.h" */

-

-/*------------------------------ internal #includes ---------------------- */

-

-#ifdef WIN32

-#pragma warning( disable : 4146 ) /* no "unsigned" warnings */

-#endif /* WIN32 */

-

-#include <stdio.h>       /* for printing in malloc_stats */

-

-#ifndef LACKS_ERRNO_H

-#include <errno.h>       /* for MALLOC_FAILURE_ACTION */

-#endif /* LACKS_ERRNO_H */

-#if FOOTERS || DEBUG

-#include <time.h>        /* for magic initialization */

-#endif /* FOOTERS */

-#ifndef LACKS_STDLIB_H

-#include <stdlib.h>      /* for abort() */

-#endif /* LACKS_STDLIB_H */

-#ifdef DEBUG

-#if ABORT_ON_ASSERT_FAILURE

-#undef assert

-#define assert(x) if(!(x)) ABORT

-#else /* ABORT_ON_ASSERT_FAILURE */

-#include <assert.h>

-#endif /* ABORT_ON_ASSERT_FAILURE */

-#else  /* DEBUG */

-#ifndef assert

-#define assert(x)

-#endif

-#define DEBUG 0

-#endif /* DEBUG */

-#ifndef LACKS_STRING_H

-#include <string.h>      /* for memset etc */

-#endif  /* LACKS_STRING_H */

-#if USE_BUILTIN_FFS

-#ifndef LACKS_STRINGS_H

-#include <strings.h>     /* for ffs */

-#endif /* LACKS_STRINGS_H */

-#endif /* USE_BUILTIN_FFS */

-#if HAVE_MMAP

-#ifndef LACKS_SYS_MMAN_H

-/* On some versions of linux, mremap decl in mman.h needs __USE_GNU set */

-#if (defined(linux) && !defined(__USE_GNU))

-#define __USE_GNU 1

-#include <sys/mman.h>    /* for mmap */

-#undef __USE_GNU

-#else

-#include <sys/mman.h>    /* for mmap */

-#endif /* linux */

-#endif /* LACKS_SYS_MMAN_H */

-#ifndef LACKS_FCNTL_H

-#include <fcntl.h>

-#endif /* LACKS_FCNTL_H */

-#endif /* HAVE_MMAP */

-#ifndef LACKS_UNISTD_H

-#include <unistd.h>     /* for sbrk, sysconf */

-#else /* LACKS_UNISTD_H */

-#if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)

-extern void*     sbrk(ptrdiff_t);

-#endif /* FreeBSD etc */

-#endif /* LACKS_UNISTD_H */

-

-/* Declarations for locking */

-#if USE_LOCKS

-#ifndef WIN32

-#include <pthread.h>

-#if defined (__SVR4) && defined (__sun)  /* solaris */

-#include <thread.h>

-#endif /* solaris */

-#else

-#ifndef _M_AMD64

-/* These are already defined on AMD64 builds */

-#ifdef __cplusplus

-extern "C" {

-#endif /* __cplusplus */

-LONG __cdecl _InterlockedCompareExchange(LONG volatile *Dest, LONG Exchange, LONG Comp);

-LONG __cdecl _InterlockedExchange(LONG volatile *Target, LONG Value);

-#ifdef __cplusplus

-}

-#endif /* __cplusplus */

-#endif /* _M_AMD64 */

-#pragma intrinsic (_InterlockedCompareExchange)

-#pragma intrinsic (_InterlockedExchange)

-#define interlockedcompareexchange _InterlockedCompareExchange

-#define interlockedexchange _InterlockedExchange

-#endif /* Win32 */

-#endif /* USE_LOCKS */

-

-/* Declarations for bit scanning on win32 */

-#if defined(_MSC_VER) && _MSC_VER>=1300

-#ifndef BitScanForward	/* Try to avoid pulling in WinNT.h */

-#ifdef __cplusplus

-extern "C" {

-#endif /* __cplusplus */

-unsigned char _BitScanForward(unsigned long *index, unsigned long mask);

-unsigned char _BitScanReverse(unsigned long *index, unsigned long mask);

-#ifdef __cplusplus

-}

-#endif /* __cplusplus */

-

-#define BitScanForward _BitScanForward

-#define BitScanReverse _BitScanReverse

-#pragma intrinsic(_BitScanForward)

-#pragma intrinsic(_BitScanReverse)

-#endif /* BitScanForward */

-#endif /* defined(_MSC_VER) && _MSC_VER>=1300 */

-

-#ifndef WIN32

-#ifndef malloc_getpagesize

-#  ifdef _SC_PAGESIZE         /* some SVR4 systems omit an underscore */

-#    ifndef _SC_PAGE_SIZE

-#      define _SC_PAGE_SIZE _SC_PAGESIZE

-#    endif

-#  endif

-#  ifdef _SC_PAGE_SIZE

-#    define malloc_getpagesize sysconf(_SC_PAGE_SIZE)

-#  else

-#    if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)

-       extern size_t getpagesize();

-#      define malloc_getpagesize getpagesize()

-#    else

-#      ifdef WIN32 /* use supplied emulation of getpagesize */

-#        define malloc_getpagesize getpagesize()

-#      else

-#        ifndef LACKS_SYS_PARAM_H

-#          include <sys/param.h>

-#        endif

-#        ifdef EXEC_PAGESIZE

-#          define malloc_getpagesize EXEC_PAGESIZE

-#        else

-#          ifdef NBPG

-#            ifndef CLSIZE

-#              define malloc_getpagesize NBPG

-#            else

-#              define malloc_getpagesize (NBPG * CLSIZE)

-#            endif

-#          else

-#            ifdef NBPC

-#              define malloc_getpagesize NBPC

-#            else

-#              ifdef PAGESIZE

-#                define malloc_getpagesize PAGESIZE

-#              else /* just guess */

-#                define malloc_getpagesize ((size_t)4096U)

-#              endif

-#            endif

-#          endif

-#        endif

-#      endif

-#    endif

-#  endif

-#endif

-#endif

-

-

-

-/* ------------------- size_t and alignment properties -------------------- */

-

-/* The byte and bit size of a size_t */

-#define SIZE_T_SIZE         (sizeof(size_t))

-#define SIZE_T_BITSIZE      (sizeof(size_t) << 3)

-

-/* Some constants coerced to size_t */

-/* Annoying but necessary to avoid errors on some platforms */

-#define SIZE_T_ZERO         ((size_t)0)

-#define SIZE_T_ONE          ((size_t)1)

-#define SIZE_T_TWO          ((size_t)2)

-#define SIZE_T_FOUR         ((size_t)4)

-#define TWO_SIZE_T_SIZES    (SIZE_T_SIZE<<1)

-#define FOUR_SIZE_T_SIZES   (SIZE_T_SIZE<<2)

-#define SIX_SIZE_T_SIZES    (FOUR_SIZE_T_SIZES+TWO_SIZE_T_SIZES)

-#define HALF_MAX_SIZE_T     (MAX_SIZE_T / 2U)

-

-/* The bit mask value corresponding to MALLOC_ALIGNMENT */

-#define CHUNK_ALIGN_MASK    (MALLOC_ALIGNMENT - SIZE_T_ONE)

-

-/* True if address a has acceptable alignment */

-#define is_aligned(A)       (((size_t)((A)) & (CHUNK_ALIGN_MASK)) == 0)

-

-/* the number of bytes to offset an address to align it */

-#define align_offset(A)\

- ((((size_t)(A) & CHUNK_ALIGN_MASK) == 0)? 0 :\

-  ((MALLOC_ALIGNMENT - ((size_t)(A) & CHUNK_ALIGN_MASK)) & CHUNK_ALIGN_MASK))

-

-/*

-  malloc_params holds global properties, including those that can be

-  dynamically set using mallopt. There is a single instance, mparams,

-  initialized in init_mparams. Note that the non-zeroness of "magic"

-  also serves as an initialization flag.

-*/

-typedef unsigned int flag_t;

-struct malloc_params {

-  volatile size_t magic;

-  size_t page_size;

-  size_t granularity;

-  size_t mmap_threshold;

-  size_t trim_threshold;

-  flag_t default_mflags;

-};

-

-static struct malloc_params mparams;

-

-/* Ensure mparams initialized */

-#define ensure_initialization() (void)(mparams.magic != 0 || init_mparams())

-

-/* -------------------------- MMAP preliminaries ------------------------- */

-

-/*

-   If HAVE_MORECORE or HAVE_MMAP are false, we just define calls and

-   checks to fail so compiler optimizer can delete code rather than

-   using so many "#if"s.

-*/

-

-

-/* MORECORE and MMAP must return MFAIL on failure */

-#define MFAIL                ((void*)(MAX_SIZE_T))

-#define CMFAIL               ((char*)(MFAIL)) /* defined for convenience */

-

-#if HAVE_MMAP

-

-#ifndef WIN32

-#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)

-#define MAP_ANONYMOUS        MAP_ANON

-#endif /* MAP_ANON */

-#ifdef DEFAULT_GRANULARITY_ALIGNED

-#define MMAP_IMPL mmap_aligned

-static void* lastAlignedmmap; /* Used as a hint */

-static void* mmap_aligned(void *start, size_t length, int prot, int flags, int fd, off_t offset) {

-  void* baseaddress = 0;

-  void* ptr = 0;

-  if(!start) {

-    baseaddress = lastAlignedmmap;

-    for(;;) {

-      if(baseaddress) flags|=MAP_FIXED;

-      ptr = mmap(baseaddress, length, prot, flags, fd, offset);

-      if(!ptr)

-        baseaddress = (void*)((size_t)baseaddress + mparams.granularity);

-      else if((size_t)ptr & (mparams.granularity - SIZE_T_ONE)) {

-        munmap(ptr, length);

-        baseaddress = (void*)(((size_t)ptr + mparams.granularity) & ~(mparams.granularity - SIZE_T_ONE));

-      }

-      else break;

-    }

-  }

-  else ptr = mmap(start, length, prot, flags, fd, offset);

-  if(ptr) lastAlignedmmap = (void*)((size_t) ptr + mparams.granularity);

-  return ptr;

-}

-#else

-#define MMAP_IMPL mmap

-#endif /* DEFAULT_GRANULARITY_ALIGNED */

-#define MUNMAP_DEFAULT(a, s)  munmap((a), (s))

-#define MMAP_PROT            (PROT_READ|PROT_WRITE)

-#ifdef MAP_ANONYMOUS

-#define MMAP_FLAGS           (MAP_PRIVATE|MAP_ANONYMOUS)

-#define MMAP_DEFAULT(s)       MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, -1, 0)

-#else /* MAP_ANONYMOUS */

-/*

-   Nearly all versions of mmap support MAP_ANONYMOUS, so the following

-   is unlikely to be needed, but is supplied just in case.

-*/

-#define MMAP_FLAGS           (MAP_PRIVATE)

-static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */

-#define MMAP_DEFAULT(s) ((dev_zero_fd < 0) ? \

-           (dev_zero_fd = open("/dev/zero", O_RDWR), \

-            MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0)) : \

-            MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0))

-#endif /* MAP_ANONYMOUS */

-

-#define DIRECT_MMAP_DEFAULT(s) MMAP_DEFAULT(s)

-

-#else /* WIN32 */

-

-/* Win32 MMAP via VirtualAlloc */

-#ifdef DEFAULT_GRANULARITY_ALIGNED

-static void* lastWin32mmap; /* Used as a hint */

-#endif /* DEFAULT_GRANULARITY_ALIGNED */

-#ifdef ENABLE_LARGE_PAGES

-static int largepagesavailable = 1;

-#endif /* ENABLE_LARGE_PAGES */

-static FORCEINLINE void* win32mmap(size_t size) {

-  void* baseaddress = 0;

-  void* ptr = 0;

-#ifdef ENABLE_LARGE_PAGES

-  /* Note that large pages are *always* allocated on a large page boundary.

-  If however granularity is small then don't waste a kernel call if size

-  isn't around the size of a large page */

-  if(largepagesavailable && size >= 1*1024*1024) {

-    ptr = VirtualAlloc(baseaddress, size, MEM_RESERVE|MEM_COMMIT|MEM_LARGE_PAGES, PAGE_READWRITE);

-    if(!ptr && ERROR_PRIVILEGE_NOT_HELD==GetLastError()) largepagesavailable=0;

-  }

-#endif

-  if(!ptr) {

-#ifdef DEFAULT_GRANULARITY_ALIGNED

-    /* We try to avoid overhead by speculatively reserving at aligned

-    addresses until we succeed */

-    baseaddress = lastWin32mmap;

-    for(;;) {

-      void* reserveaddr = VirtualAlloc(baseaddress, size, MEM_RESERVE, PAGE_READWRITE);

-      if(!reserveaddr)

-        baseaddress = (void*)((size_t)baseaddress + mparams.granularity);

-      else if((size_t)reserveaddr & (mparams.granularity - SIZE_T_ONE)) {

-        VirtualFree(reserveaddr, 0, MEM_RELEASE);

-        baseaddress = (void*)(((size_t)reserveaddr + mparams.granularity) & ~(mparams.granularity - SIZE_T_ONE));

-      }

-      else break;

-    }

-#endif

-    if(!ptr) ptr = VirtualAlloc(baseaddress, size, baseaddress ? MEM_COMMIT : MEM_RESERVE|MEM_COMMIT, PAGE_READWRITE);

-#if DEBUG

-    if(lastWin32mmap && ptr!=lastWin32mmap) printf("Non-contiguous VirtualAlloc between %p and %p\n", ptr, lastWin32mmap);

-#endif

-#ifdef DEFAULT_GRANULARITY_ALIGNED

-    if(ptr) lastWin32mmap = (void*)((size_t) ptr + mparams.granularity);

-#endif

-  }

-#if DEBUG

-#ifdef ENABLE_LARGE_PAGES

-  printf("VirtualAlloc returns %p size %u. LargePagesAvailable=%d\n", ptr, size, largepagesavailable);

-#else

-  printf("VirtualAlloc returns %p size %u\n", ptr, size);

-#endif

-#endif

-  return (ptr != 0)? ptr: MFAIL;

-}

-

-/* For direct MMAP, use MEM_TOP_DOWN to minimize interference */

-static FORCEINLINE void* win32direct_mmap(size_t size) {

-  void* ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT|MEM_TOP_DOWN,

-                           PAGE_READWRITE);

-  return (ptr != 0)? ptr: MFAIL;

-}

-

-/* This function supports releasing coalesed segments */

-static FORCEINLINE int win32munmap(void* ptr, size_t size) {

-  MEMORY_BASIC_INFORMATION minfo;

-  char* cptr = (char*)ptr;

-  while (size) {

-    if (VirtualQuery(cptr, &minfo, sizeof(minfo)) == 0)

-      return -1;

-    if (minfo.BaseAddress != cptr || minfo.AllocationBase != cptr ||

-        minfo.State != MEM_COMMIT || minfo.RegionSize > size)

-      return -1;

-    if (VirtualFree(cptr, 0, MEM_RELEASE) == 0)

-      return -1;

-    cptr += minfo.RegionSize;

-    size -= minfo.RegionSize;

-  }

-  return 0;

-}

-

-#define MMAP_DEFAULT(s)             win32mmap(s)

-#define MUNMAP_DEFAULT(a, s)        win32munmap((a), (s))

-#define DIRECT_MMAP_DEFAULT(s)      win32direct_mmap(s)

-#endif /* WIN32 */

-#endif /* HAVE_MMAP */

-

-#if HAVE_MREMAP

-#ifndef WIN32

-#define MREMAP_DEFAULT(addr, osz, nsz, mv) mremap((addr), (osz), (nsz), (mv))

-#endif /* WIN32 */

-#endif /* HAVE_MREMAP */

-

-

-/**

- * Define CALL_MORECORE

- */

-#if HAVE_MORECORE

-    #ifdef MORECORE

-        #define CALL_MORECORE(S)    MORECORE(S)

-    #else  /* MORECORE */

-        #define CALL_MORECORE(S)    MORECORE_DEFAULT(S)

-    #endif /* MORECORE */

-#else  /* HAVE_MORECORE */

-    #define CALL_MORECORE(S)        MFAIL

-#endif /* HAVE_MORECORE */

-

-/**

- * Define CALL_MMAP/CALL_MUNMAP/CALL_DIRECT_MMAP

- */

-#if HAVE_MMAP

-    #define USE_MMAP_BIT            (SIZE_T_ONE)

-

-    #ifdef MMAP

-        #define CALL_MMAP(s)        MMAP(s)

-    #else /* MMAP */

-        #define CALL_MMAP(s)        MMAP_DEFAULT(s)

-    #endif /* MMAP */

-    #ifdef MUNMAP

-        #define CALL_MUNMAP(a, s)   MUNMAP((a), (s))

-    #else /* MUNMAP */

-        #define CALL_MUNMAP(a, s)   MUNMAP_DEFAULT((a), (s))

-    #endif /* MUNMAP */

-    #ifdef DIRECT_MMAP

-        #define CALL_DIRECT_MMAP(s) DIRECT_MMAP(s)

-    #else /* DIRECT_MMAP */

-        #define CALL_DIRECT_MMAP(s) DIRECT_MMAP_DEFAULT(s)

-    #endif /* DIRECT_MMAP */

-#else  /* HAVE_MMAP */

-    #define USE_MMAP_BIT            (SIZE_T_ZERO)

-

-    #define MMAP(s)                 MFAIL

-    #define MUNMAP(a, s)            (-1)

-    #define DIRECT_MMAP(s)          MFAIL

-    #define CALL_DIRECT_MMAP(s)     DIRECT_MMAP(s)

-    #define CALL_MMAP(s)            MMAP(s)

-    #define CALL_MUNMAP(a, s)       MUNMAP((a), (s))

-#endif /* HAVE_MMAP */

-

-/**

- * Define CALL_MREMAP

- */

-#if HAVE_MMAP && HAVE_MREMAP

-    #ifdef MREMAP

-        #define CALL_MREMAP(addr, osz, nsz, mv) MREMAP((addr), (osz), (nsz), (mv))

-    #else /* MREMAP */

-        #define CALL_MREMAP(addr, osz, nsz, mv) MREMAP_DEFAULT((addr), (osz), (nsz), (mv))

-    #endif /* MREMAP */

-#else  /* HAVE_MMAP && HAVE_MREMAP */

-    #define CALL_MREMAP(addr, osz, nsz, mv)     MFAIL

-#endif /* HAVE_MMAP && HAVE_MREMAP */

-

-/* mstate bit set if continguous morecore disabled or failed */

-#define USE_NONCONTIGUOUS_BIT (4U)

-

-/* segment bit set in create_mspace_with_base */

-#define EXTERN_BIT            (8U)

-

-

-/* --------------------------- Lock preliminaries ------------------------ */

-

-/*

-  When locks are defined, there is one global lock, plus

-  one per-mspace lock.

-

-  The global lock_ensures that mparams.magic and other unique

-  mparams values are initialized only once. It also protects

-  sequences of calls to MORECORE.  In many cases sys_alloc requires

-  two calls, that should not be interleaved with calls by other

-  threads.  This does not protect against direct calls to MORECORE

-  by other threads not using this lock, so there is still code to

-  cope the best we can on interference.

-

-  Per-mspace locks surround calls to malloc, free, etc.  To enable use

-  in layered extensions, per-mspace locks are reentrant.

-

-  Because lock-protected regions generally have bounded times, it is

-  OK to use the supplied simple spinlocks in the custom versions for

-  x86. Spinlocks are likely to improve performance for lightly

-  contended applications, but worsen performance under heavy

-  contention.

-

-  If USE_LOCKS is > 1, the definitions of lock routines here are

-  bypassed, in which case you will need to define the type MLOCK_T,

-  and at least INITIAL_LOCK, ACQUIRE_LOCK, RELEASE_LOCK and possibly

-  TRY_LOCK (which is not used in this malloc, but commonly needed in

-  extensions.)  You must also declare a

-    static MLOCK_T malloc_global_mutex = { initialization values };.

-

-*/

-

-#if USE_LOCKS == 1

-

-#if USE_SPIN_LOCKS && SPIN_LOCKS_AVAILABLE

-#ifndef WIN32

-

-/* Custom pthread-style spin locks on x86 and x64 for gcc */

-struct pthread_mlock_t {

-  volatile unsigned int l;

-  char cachelinepadding[64];

-  unsigned int c;

-  pthread_t threadid;

-};

-#define MLOCK_T               struct pthread_mlock_t

-#define CURRENT_THREAD        pthread_self()

-#define INITIAL_LOCK(sl)      ((sl)->threadid = 0, (sl)->l = (sl)->c = 0, 0)

-#define ACQUIRE_LOCK(sl)      pthread_acquire_lock(sl)

-#define RELEASE_LOCK(sl)      pthread_release_lock(sl)

-#define TRY_LOCK(sl)          pthread_try_lock(sl)

-#define SPINS_PER_YIELD       63

-

-static MLOCK_T malloc_global_mutex = { 0, "", 0, 0};

-

-static FORCEINLINE int pthread_acquire_lock (MLOCK_T *sl) {

-  int spins = 0;

-  volatile unsigned int* lp = &sl->l;

-  for (;;) {

-    if (*lp != 0) {

-      if (sl->threadid == CURRENT_THREAD) {

-        ++sl->c;

-        return 0;

-      }

-    }

-    else {

-      /* place args to cmpxchgl in locals to evade oddities in some gccs */

-      int cmp = 0;

-      int val = 1;

-      int ret;

-      __asm__ __volatile__  ("lock; cmpxchgl %1, %2"

-                             : "=a" (ret)

-                             : "r" (val), "m" (*(lp)), "0"(cmp)

-                             : "memory", "cc");

-      if (!ret) {

-        assert(!sl->threadid);

-        sl->threadid = CURRENT_THREAD;

-        sl->c = 1;

-        return 0;

-      }

-    }

-    if ((++spins & SPINS_PER_YIELD) == 0) {

-#if defined (__SVR4) && defined (__sun) /* solaris */

-      thr_yield();

-#else

-#if defined(__linux__) || defined(__FreeBSD__) || defined(__APPLE__)

-      sched_yield();

-#else  /* no-op yield on unknown systems */

-      ;

-#endif /* __linux__ || __FreeBSD__ || __APPLE__ */

-#endif /* solaris */

-    }

-  }

-}

-

-static FORCEINLINE void pthread_release_lock (MLOCK_T *sl) {

-  volatile unsigned int* lp = &sl->l;

-  assert(*lp != 0);

-  assert(sl->threadid == CURRENT_THREAD);

-  if (--sl->c == 0) {

-    sl->threadid = 0;

-    int prev = 0;

-    int ret;

-    __asm__ __volatile__ ("lock; xchgl %0, %1"

-                          : "=r" (ret)

-                          : "m" (*(lp)), "0"(prev)

-                          : "memory");

-  }

-}

-

-static FORCEINLINE int pthread_try_lock (MLOCK_T *sl) {

-  volatile unsigned int* lp = &sl->l;

-  if (*lp != 0) {

-    if (sl->threadid == CURRENT_THREAD) {

-      ++sl->c;

-      return 1;

-    }

-  }

-  else {

-    int cmp = 0;

-    int val = 1;

-    int ret;

-    __asm__ __volatile__  ("lock; cmpxchgl %1, %2"

-                           : "=a" (ret)

-                           : "r" (val), "m" (*(lp)), "0"(cmp)

-                           : "memory", "cc");

-    if (!ret) {

-      assert(!sl->threadid);

-      sl->threadid = CURRENT_THREAD;

-      sl->c = 1;

-      return 1;

-    }

-  }

-  return 0;

-}

-

-

-#else /* WIN32 */

-/* Custom win32-style spin locks on x86 and x64 for MSC */

-struct win32_mlock_t {

-  volatile long l;

-  char cachelinepadding[64];

-  unsigned int c;

-  long threadid;

-};

-

-#define MLOCK_T               struct win32_mlock_t

-#define CURRENT_THREAD        ((long)GetCurrentThreadId())

-#define INITIAL_LOCK(sl)      ((sl)->threadid = 0, (sl)->l = (sl)->c = 0, 0)

-#define ACQUIRE_LOCK(sl)      win32_acquire_lock(sl)

-#define RELEASE_LOCK(sl)      win32_release_lock(sl)

-#define TRY_LOCK(sl)          win32_try_lock(sl)

-#define SPINS_PER_YIELD       63

-

-static MLOCK_T malloc_global_mutex = { 0, 0, 0};

-

-static FORCEINLINE int win32_acquire_lock (MLOCK_T *sl) {

-  int spins = 0;

-  for (;;) {

-    if (sl->l != 0) {

-      if (sl->threadid == CURRENT_THREAD) {

-        ++sl->c;

-        return 0;

-      }

-    }

-    else {

-      if (!interlockedexchange(&sl->l, 1)) {

-        assert(!sl->threadid);

-        sl->threadid = CURRENT_THREAD;

-        sl->c = 1;

-        return 0;

-      }

-    }

-    if ((++spins & SPINS_PER_YIELD) == 0)

-      SleepEx(0, FALSE);

-  }

-}

-

-static FORCEINLINE void win32_release_lock (MLOCK_T *sl) {

-  assert(sl->threadid == CURRENT_THREAD);

-  assert(sl->l != 0);

-  if (--sl->c == 0) {

-    sl->threadid = 0;

-    interlockedexchange (&sl->l, 0);

-  }

-}

-

-static FORCEINLINE int win32_try_lock (MLOCK_T *sl) {

-  if (sl->l != 0) {

-    if (sl->threadid == CURRENT_THREAD) {

-      ++sl->c;

-      return 1;

-    }

-  }

-  else {

-    if (!interlockedexchange(&sl->l, 1)){

-      assert(!sl->threadid);

-      sl->threadid = CURRENT_THREAD;

-      sl->c = 1;

-      return 1;

-    }

-  }

-  return 0;

-}

-

-#endif /* WIN32 */

-#else /* USE_SPIN_LOCKS */

-

-#ifndef WIN32

-/* pthreads-based locks */

-

-#define MLOCK_T               pthread_mutex_t

-#define CURRENT_THREAD        pthread_self()

-#define INITIAL_LOCK(sl)      pthread_init_lock(sl)

-#define ACQUIRE_LOCK(sl)      pthread_mutex_lock(sl)

-#define RELEASE_LOCK(sl)      pthread_mutex_unlock(sl)

-#define TRY_LOCK(sl)          (!pthread_mutex_trylock(sl))

-

-static MLOCK_T malloc_global_mutex = PTHREAD_MUTEX_INITIALIZER;

-

-/* Cope with old-style linux recursive lock initialization by adding */

-/* skipped internal declaration from pthread.h */

-#ifdef linux

-#ifndef PTHREAD_MUTEX_RECURSIVE

-extern int pthread_mutexattr_setkind_np __P ((pthread_mutexattr_t *__attr,

-					   int __kind));

-#define PTHREAD_MUTEX_RECURSIVE PTHREAD_MUTEX_RECURSIVE_NP

-#define pthread_mutexattr_settype(x,y) pthread_mutexattr_setkind_np(x,y)

-#endif

-#endif

-

-static int pthread_init_lock (MLOCK_T *sl) {

-  pthread_mutexattr_t attr;

-  if (pthread_mutexattr_init(&attr)) return 1;

-  if (pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_RECURSIVE)) return 1;

-  if (pthread_mutex_init(sl, &attr)) return 1;

-  if (pthread_mutexattr_destroy(&attr)) return 1;

-  return 0;

-}

-

-#else /* WIN32 */

-/* Win32 critical sections */

-#define MLOCK_T               CRITICAL_SECTION

-#define CURRENT_THREAD        GetCurrentThreadId()

-#define INITIAL_LOCK(s)       (!InitializeCriticalSectionAndSpinCount((s), 0x80000000|4000))

-#define ACQUIRE_LOCK(s)       (EnterCriticalSection(sl), 0)

-#define RELEASE_LOCK(s)       LeaveCriticalSection(sl)

-#define TRY_LOCK(s)           TryEnterCriticalSection(sl)

-#define NEED_GLOBAL_LOCK_INIT

-

-static MLOCK_T malloc_global_mutex;

-static volatile long malloc_global_mutex_status;

-

-/* Use spin loop to initialize global lock */

-static void init_malloc_global_mutex() {

-  for (;;) {

-    long stat = malloc_global_mutex_status;

-    if (stat > 0)

-      return;

-    /* transition to < 0 while initializing, then to > 0) */

-    if (stat == 0 &&

-        interlockedcompareexchange(&malloc_global_mutex_status, -1, 0) == 0) {

-      InitializeCriticalSection(&malloc_global_mutex);

-      interlockedexchange(&malloc_global_mutex_status,1);

-      return;

-    }

-    SleepEx(0, FALSE);

-  }

-}

-

-#endif /* WIN32 */

-#endif /* USE_SPIN_LOCKS */

-#endif /* USE_LOCKS == 1 */

-

-/* -----------------------  User-defined locks ------------------------ */

-

-#if USE_LOCKS > 1

-/* Define your own lock implementation here */

-/* #define INITIAL_LOCK(sl)  ... */

-/* #define ACQUIRE_LOCK(sl)  ... */

-/* #define RELEASE_LOCK(sl)  ... */

-/* #define TRY_LOCK(sl) ... */

-/* static MLOCK_T malloc_global_mutex = ... */

-#endif /* USE_LOCKS > 1 */

-

-/* -----------------------  Lock-based state ------------------------ */

-

-#if USE_LOCKS

-#define USE_LOCK_BIT               (2U)

-#else  /* USE_LOCKS */

-#define USE_LOCK_BIT               (0U)

-#define INITIAL_LOCK(l)

-#endif /* USE_LOCKS */

-

-#if USE_LOCKS

-#ifndef ACQUIRE_MALLOC_GLOBAL_LOCK

-#define ACQUIRE_MALLOC_GLOBAL_LOCK()  ACQUIRE_LOCK(&malloc_global_mutex);

-#endif

-#ifndef RELEASE_MALLOC_GLOBAL_LOCK

-#define RELEASE_MALLOC_GLOBAL_LOCK()  RELEASE_LOCK(&malloc_global_mutex);

-#endif

-#else  /* USE_LOCKS */

-#define ACQUIRE_MALLOC_GLOBAL_LOCK()

-#define RELEASE_MALLOC_GLOBAL_LOCK()

-#endif /* USE_LOCKS */

-

-

-/* -----------------------  Chunk representations ------------------------ */

-

-/*

-  (The following includes lightly edited explanations by Colin Plumb.)

-

-  The malloc_chunk declaration below is misleading (but accurate and

-  necessary).  It declares a "view" into memory allowing access to

-  necessary fields at known offsets from a given base.

-

-  Chunks of memory are maintained using a `boundary tag' method as

-  originally described by Knuth.  (See the paper by Paul Wilson

-  ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a survey of such

-  techniques.)  Sizes of free chunks are stored both in the front of

-  each chunk and at the end.  This makes consolidating fragmented

-  chunks into bigger chunks fast.  The head fields also hold bits

-  representing whether chunks are free or in use.

-

-  Here are some pictures to make it clearer.  They are "exploded" to

-  show that the state of a chunk can be thought of as extending from

-  the high 31 bits of the head field of its header through the

-  prev_foot and PINUSE_BIT bit of the following chunk header.

-

-  A chunk that's in use looks like:

-

-   chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-           | Size of previous chunk (if P = 0)                             |

-           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|

-         | Size of this chunk                                         1| +-+

-   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         |                                                               |

-         +-                                                             -+

-         |                                                               |

-         +-                                                             -+

-         |                                                               :

-         +-      size - sizeof(size_t) available payload bytes          -+

-         :                                                               |

- chunk-> +-                                                             -+

-         |                                                               |

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|

-       | Size of next chunk (may or may not be in use)               | +-+

- mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-

-    And if it's free, it looks like this:

-

-   chunk-> +-                                                             -+

-           | User payload (must be in use, or we would have merged!)       |

-           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|

-         | Size of this chunk                                         0| +-+

-   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         | Next pointer                                                  |

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         | Prev pointer                                                  |

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         |                                                               :

-         +-      size - sizeof(struct chunk) unused bytes               -+

-         :                                                               |

- chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-         | Size of this chunk                                            |

-         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0|

-       | Size of next chunk (must be in use, or we would have merged)| +-+

- mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-       |                                                               :

-       +- User payload                                                -+

-       :                                                               |

-       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-                                                                     |0|

-                                                                     +-+

-  Note that since we always merge adjacent free chunks, the chunks

-  adjacent to a free chunk must be in use.

-

-  Given a pointer to a chunk (which can be derived trivially from the

-  payload pointer) we can, in O(1) time, find out whether the adjacent

-  chunks are free, and if so, unlink them from the lists that they

-  are on and merge them with the current chunk.

-

-  Chunks always begin on even word boundaries, so the mem portion

-  (which is returned to the user) is also on an even word boundary, and

-  thus at least double-word aligned.

-

-  The P (PINUSE_BIT) bit, stored in the unused low-order bit of the

-  chunk size (which is always a multiple of two words), is an in-use

-  bit for the *previous* chunk.  If that bit is *clear*, then the

-  word before the current chunk size contains the previous chunk

-  size, and can be used to find the front of the previous chunk.

-  The very first chunk allocated always has this bit set, preventing

-  access to non-existent (or non-owned) memory. If pinuse is set for

-  any given chunk, then you CANNOT determine the size of the

-  previous chunk, and might even get a memory addressing fault when

-  trying to do so.

-

-  The C (CINUSE_BIT) bit, stored in the unused second-lowest bit of

-  the chunk size redundantly records whether the current chunk is

-  inuse (unless the chunk is mmapped). This redundancy enables usage

-  checks within free and realloc, and reduces indirection when freeing

-  and consolidating chunks.

-

-  Each freshly allocated chunk must have both cinuse and pinuse set.

-  That is, each allocated chunk borders either a previously allocated

-  and still in-use chunk, or the base of its memory arena. This is

-  ensured by making all allocations from the the `lowest' part of any

-  found chunk.  Further, no free chunk physically borders another one,

-  so each free chunk is known to be preceded and followed by either

-  inuse chunks or the ends of memory.

-

-  Note that the `foot' of the current chunk is actually represented

-  as the prev_foot of the NEXT chunk. This makes it easier to

-  deal with alignments etc but can be very confusing when trying

-  to extend or adapt this code.

-

-  The exceptions to all this are

-

-     1. The special chunk `top' is the top-most available chunk (i.e.,

-        the one bordering the end of available memory). It is treated

-        specially.  Top is never included in any bin, is used only if

-        no other chunk is available, and is released back to the

-        system if it is very large (see M_TRIM_THRESHOLD).  In effect,

-        the top chunk is treated as larger (and thus less well

-        fitting) than any other available chunk.  The top chunk

-        doesn't update its trailing size field since there is no next

-        contiguous chunk that would have to index off it. However,

-        space is still allocated for it (TOP_FOOT_SIZE) to enable

-        separation or merging when space is extended.

-

-     3. Chunks allocated via mmap, have both cinuse and pinuse bits

-        cleared in their head fields.  Because they are allocated

-        one-by-one, each must carry its own prev_foot field, which is

-        also used to hold the offset this chunk has within its mmapped

-        region, which is needed to preserve alignment. Each mmapped

-        chunk is trailed by the first two fields of a fake next-chunk

-        for sake of usage checks.

-

-*/

-

-struct malloc_chunk {

-  size_t               prev_foot;  /* Size of previous chunk (if free).  */

-  size_t               head;       /* Size and inuse bits. */

-  struct malloc_chunk* fd;         /* double links -- used only if free. */

-  struct malloc_chunk* bk;

-};

-

-typedef struct malloc_chunk  mchunk;

-typedef struct malloc_chunk* mchunkptr;

-typedef struct malloc_chunk* sbinptr;  /* The type of bins of chunks */

-typedef unsigned int bindex_t;         /* Described below */

-typedef unsigned int binmap_t;         /* Described below */

-

-/* ------------------- Chunks sizes and alignments ----------------------- */

-

-#define MCHUNK_SIZE         (sizeof(mchunk))

-

-#if FOOTERS

-#define CHUNK_OVERHEAD      (TWO_SIZE_T_SIZES)

-#else /* FOOTERS */

-#define CHUNK_OVERHEAD      (SIZE_T_SIZE)

-#endif /* FOOTERS */

-

-/* MMapped chunks need a second word of overhead ... */

-#define MMAP_CHUNK_OVERHEAD (TWO_SIZE_T_SIZES)

-/* ... and additional padding for fake next-chunk at foot */

-#define MMAP_FOOT_PAD       (FOUR_SIZE_T_SIZES)

-

-/* The smallest size we can malloc is an aligned minimal chunk */

-#define MIN_CHUNK_SIZE\

-  ((MCHUNK_SIZE + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)

-

-/* conversion from malloc headers to user pointers, and back */

-#define chunk2mem(p)        ((void*)((char*)(p)       + TWO_SIZE_T_SIZES))

-#define mem2chunk(mem)      ((mchunkptr)((char*)(mem) - TWO_SIZE_T_SIZES))

-/* chunk associated with aligned address A */

-#define align_as_chunk(A)   (mchunkptr)((A) + align_offset(chunk2mem(A)))

-

-/* Bounds on request (not chunk) sizes. */

-#define MAX_REQUEST         ((-MIN_CHUNK_SIZE) << 2)

-#define MIN_REQUEST         (MIN_CHUNK_SIZE - CHUNK_OVERHEAD - SIZE_T_ONE)

-

-/* pad request bytes into a usable size */

-#define pad_request(req) \

-   (((req) + CHUNK_OVERHEAD + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)

-

-/* pad request, checking for minimum (but not maximum) */

-#define request2size(req) \

-  (((req) < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(req))

-

-

-/* ------------------ Operations on head and foot fields ----------------- */

-

-/*

-  The head field of a chunk is or'ed with PINUSE_BIT when previous

-  adjacent chunk in use, and or'ed with CINUSE_BIT if this chunk is in

-  use, unless mmapped, in which case both bits are cleared.

-

-  FLAG4_BIT is not used by this malloc, but might be useful in extensions.

-*/

-

-#define PINUSE_BIT          (SIZE_T_ONE)

-#define CINUSE_BIT          (SIZE_T_TWO)

-#define FLAG4_BIT           (SIZE_T_FOUR)

-#define INUSE_BITS          (PINUSE_BIT|CINUSE_BIT)

-#define FLAG_BITS           (PINUSE_BIT|CINUSE_BIT|FLAG4_BIT)

-

-/* Head value for fenceposts */

-#define FENCEPOST_HEAD      (INUSE_BITS|SIZE_T_SIZE)

-

-/* extraction of fields from head words */

-#define cinuse(p)           ((p)->head & CINUSE_BIT)

-#define pinuse(p)           ((p)->head & PINUSE_BIT)

-#define is_inuse(p)         (((p)->head & INUSE_BITS) != PINUSE_BIT)

-#define is_mmapped(p)       (((p)->head & INUSE_BITS) == 0)

-

-#define chunksize(p)        ((p)->head & ~(FLAG_BITS))

-

-#define clear_pinuse(p)     ((p)->head &= ~PINUSE_BIT)

-

-/* Treat space at ptr +/- offset as a chunk */

-#define chunk_plus_offset(p, s)  ((mchunkptr)(((char*)(p)) + (s)))

-#define chunk_minus_offset(p, s) ((mchunkptr)(((char*)(p)) - (s)))

-

-/* Ptr to next or previous physical malloc_chunk. */

-#define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->head & ~FLAG_BITS)))

-#define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_foot) ))

-

-/* extract next chunk's pinuse bit */

-#define next_pinuse(p)  ((next_chunk(p)->head) & PINUSE_BIT)

-

-/* Get/set size at footer */

-#define get_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot)

-#define set_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot = (s))

-

-/* Set size, pinuse bit, and foot */

-#define set_size_and_pinuse_of_free_chunk(p, s)\

-  ((p)->head = (s|PINUSE_BIT), set_foot(p, s))

-

-/* Set size, pinuse bit, foot, and clear next pinuse */

-#define set_free_with_pinuse(p, s, n)\

-  (clear_pinuse(n), set_size_and_pinuse_of_free_chunk(p, s))

-

-/* Get the internal overhead associated with chunk p */

-#define overhead_for(p)\

- (is_mmapped(p)? MMAP_CHUNK_OVERHEAD : CHUNK_OVERHEAD)

-

-/* Return true if malloced space is not necessarily cleared */

-#if MMAP_CLEARS

-#define calloc_must_clear(p) (!is_mmapped(p))

-#else /* MMAP_CLEARS */

-#define calloc_must_clear(p) (1)

-#endif /* MMAP_CLEARS */

-

-/* ---------------------- Overlaid data structures ----------------------- */

-

-/*

-  When chunks are not in use, they are treated as nodes of either

-  lists or trees.

-

-  "Small"  chunks are stored in circular doubly-linked lists, and look

-  like this:

-

-    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Size of previous chunk                            |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-    `head:' |             Size of chunk, in bytes                         |P|

-      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Forward pointer to next chunk in list             |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Back pointer to previous chunk in list            |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Unused space (may be 0 bytes long)                .

-            .                                                               .

-            .                                                               |

-nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-    `foot:' |             Size of chunk, in bytes                           |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-

-  Larger chunks are kept in a form of bitwise digital trees (aka

-  tries) keyed on chunksizes.  Because malloc_tree_chunks are only for

-  free chunks greater than 256 bytes, their size doesn't impose any

-  constraints on user chunk sizes.  Each node looks like:

-

-    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Size of previous chunk                            |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-    `head:' |             Size of chunk, in bytes                         |P|

-      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Forward pointer to next chunk of same size        |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Back pointer to previous chunk of same size       |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Pointer to left child (child[0])                  |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Pointer to right child (child[1])                 |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Pointer to parent                                 |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             bin index of this chunk                           |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-            |             Unused space                                      .

-            .                                                               |

-nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-    `foot:' |             Size of chunk, in bytes                           |

-            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

-

-  Each tree holding treenodes is a tree of unique chunk sizes.  Chunks

-  of the same size are arranged in a circularly-linked list, with only

-  the oldest chunk (the next to be used, in our FIFO ordering)

-  actually in the tree.  (Tree members are distinguished by a non-null

-  parent pointer.)  If a chunk with the same size an an existing node

-  is inserted, it is linked off the existing node using pointers that

-  work in the same way as fd/bk pointers of small chunks.

-

-  Each tree contains a power of 2 sized range of chunk sizes (the

-  smallest is 0x100 <= x < 0x180), which is is divided in half at each

-  tree level, with the chunks in the smaller half of the range (0x100

-  <= x < 0x140 for the top nose) in the left subtree and the larger

-  half (0x140 <= x < 0x180) in the right subtree.  This is, of course,

-  done by inspecting individual bits.

-

-  Using these rules, each node's left subtree contains all smaller

-  sizes than its right subtree.  However, the node at the root of each

-  subtree has no particular ordering relationship to either.  (The

-  dividing line between the subtree sizes is based on trie relation.)

-  If we remove the last chunk of a given size from the interior of the

-  tree, we need to replace it with a leaf node.  The tree ordering

-  rules permit a node to be replaced by any leaf below it.

-

-  The smallest chunk in a tree (a common operation in a best-fit

-  allocator) can be found by walking a path to the leftmost leaf in

-  the tree.  Unlike a usual binary tree, where we follow left child

-  pointers until we reach a null, here we follow the right child

-  pointer any time the left one is null, until we reach a leaf with

-  both child pointers null. The smallest chunk in the tree will be

-  somewhere along that path.

-

-  The worst case number of steps to add, find, or remove a node is

-  bounded by the number of bits differentiating chunks within

-  bins. Under current bin calculations, this ranges from 6 up to 21

-  (for 32 bit sizes) or up to 53 (for 64 bit sizes). The typical case

-  is of course much better.

-*/

-

-struct malloc_tree_chunk {

-  /* The first four fields must be compatible with malloc_chunk */

-  size_t                    prev_foot;

-  size_t                    head;

-  struct malloc_tree_chunk* fd;

-  struct malloc_tree_chunk* bk;

-

-  struct malloc_tree_chunk* child[2];

-  struct malloc_tree_chunk* parent;

-  bindex_t                  index;

-};

-

-typedef struct malloc_tree_chunk  tchunk;

-typedef struct malloc_tree_chunk* tchunkptr;

-typedef struct malloc_tree_chunk* tbinptr; /* The type of bins of trees */

-

-/* A little helper macro for trees */

-#define leftmost_child(t) ((t)->child[0] != 0? (t)->child[0] : (t)->child[1])

-

-/* ----------------------------- Segments -------------------------------- */

-

-/*

-  Each malloc space may include non-contiguous segments, held in a

-  list headed by an embedded malloc_segment record representing the

-  top-most space. Segments also include flags holding properties of

-  the space. Large chunks that are directly allocated by mmap are not

-  included in this list. They are instead independently created and

-  destroyed without otherwise keeping track of them.

-

-  Segment management mainly comes into play for spaces allocated by

-  MMAP.  Any call to MMAP might or might not return memory that is

-  adjacent to an existing segment.  MORECORE normally contiguously

-  extends the current space, so this space is almost always adjacent,

-  which is simpler and faster to deal with. (This is why MORECORE is

-  used preferentially to MMAP when both are available -- see

-  sys_alloc.)  When allocating using MMAP, we don't use any of the

-  hinting mechanisms (inconsistently) supported in various

-  implementations of unix mmap, or distinguish reserving from

-  committing memory. Instead, we just ask for space, and exploit

-  contiguity when we get it.  It is probably possible to do

-  better than this on some systems, but no general scheme seems

-  to be significantly better.

-

-  Management entails a simpler variant of the consolidation scheme

-  used for chunks to reduce fragmentation -- new adjacent memory is

-  normally prepended or appended to an existing segment. However,

-  there are limitations compared to chunk consolidation that mostly

-  reflect the fact that segment processing is relatively infrequent

-  (occurring only when getting memory from system) and that we

-  don't expect to have huge numbers of segments:

-

-  * Segments are not indexed, so traversal requires linear scans.  (It

-    would be possible to index these, but is not worth the extra

-    overhead and complexity for most programs on most platforms.)

-  * New segments are only appended to old ones when holding top-most

-    memory; if they cannot be prepended to others, they are held in

-    different segments.

-

-  Except for the top-most segment of an mstate, each segment record

-  is kept at the tail of its segment. Segments are added by pushing

-  segment records onto the list headed by &mstate.seg for the

-  containing mstate.

-

-  Segment flags control allocation/merge/deallocation policies:

-  * If EXTERN_BIT set, then we did not allocate this segment,

-    and so should not try to deallocate or merge with others.

-    (This currently holds only for the initial segment passed

-    into create_mspace_with_base.)

-  * If USE_MMAP_BIT set, the segment may be merged with

-    other surrounding mmapped segments and trimmed/de-allocated

-    using munmap.

-  * If neither bit is set, then the segment was obtained using

-    MORECORE so can be merged with surrounding MORECORE'd segments

-    and deallocated/trimmed using MORECORE with negative arguments.

-*/

-

-struct malloc_segment {

-  char*        base;             /* base address */

-  size_t       size;             /* allocated size */

-  struct malloc_segment* next;   /* ptr to next segment */

-  flag_t       sflags;           /* mmap and extern flag */

-};

-

-#define is_mmapped_segment(S)  ((S)->sflags & USE_MMAP_BIT)

-#define is_extern_segment(S)   ((S)->sflags & EXTERN_BIT)

-

-typedef struct malloc_segment  msegment;

-typedef struct malloc_segment* msegmentptr;

-

-/* ---------------------------- malloc_state ----------------------------- */

-

-/*

-   A malloc_state holds all of the bookkeeping for a space.

-   The main fields are:

-

-  Top

-    The topmost chunk of the currently active segment. Its size is

-    cached in topsize.  The actual size of topmost space is

-    topsize+TOP_FOOT_SIZE, which includes space reserved for adding

-    fenceposts and segment records if necessary when getting more

-    space from the system.  The size at which to autotrim top is

-    cached from mparams in trim_check, except that it is disabled if

-    an autotrim fails.

-

-  Designated victim (dv)

-    This is the preferred chunk for servicing small requests that

-    don't have exact fits.  It is normally the chunk split off most

-    recently to service another small request.  Its size is cached in

-    dvsize. The link fields of this chunk are not maintained since it

-    is not kept in a bin.

-

-  SmallBins

-    An array of bin headers for free chunks.  These bins hold chunks

-    with sizes less than MIN_LARGE_SIZE bytes. Each bin contains

-    chunks of all the same size, spaced 8 bytes apart.  To simplify

-    use in double-linked lists, each bin header acts as a malloc_chunk

-    pointing to the real first node, if it exists (else pointing to

-    itself).  This avoids special-casing for headers.  But to avoid

-    waste, we allocate only the fd/bk pointers of bins, and then use

-    repositioning tricks to treat these as the fields of a chunk.

-

-  TreeBins

-    Treebins are pointers to the roots of trees holding a range of

-    sizes. There are 2 equally spaced treebins for each power of two

-    from TREE_SHIFT to TREE_SHIFT+16. The last bin holds anything

-    larger.

-

-  Bin maps

-    There is one bit map for small bins ("smallmap") and one for

-    treebins ("treemap).  Each bin sets its bit when non-empty, and

-    clears the bit when empty.  Bit operations are then used to avoid

-    bin-by-bin searching -- nearly all "search" is done without ever

-    looking at bins that won't be selected.  The bit maps

-    conservatively use 32 bits per map word, even if on 64bit system.

-    For a good description of some of the bit-based techniques used

-    here, see Henry S. Warren Jr's book "Hacker's Delight" (and

-    supplement at http://hackersdelight.org/). Many of these are

-    intended to reduce the branchiness of paths through malloc etc, as

-    well as to reduce the number of memory locations read or written.

-

-  Segments

-    A list of segments headed by an embedded malloc_segment record

-    representing the initial space.

-

-  Address check support

-    The least_addr field is the least address ever obtained from

-    MORECORE or MMAP. Attempted frees and reallocs of any address less

-    than this are trapped (unless INSECURE is defined).

-

-  Magic tag

-    A cross-check field that should always hold same value as mparams.magic.

-

-  Flags

-    Bits recording whether to use MMAP, locks, or contiguous MORECORE

-

-  Statistics

-    Each space keeps track of current and maximum system memory

-    obtained via MORECORE or MMAP.

-

-  Trim support

-    Fields holding the amount of unused topmost memory that should trigger

-    timming, and a counter to force periodic scanning to release unused

-    non-topmost segments.

-

-  Locking

-    If USE_LOCKS is defined, the "mutex" lock is acquired and released

-    around every public call using this mspace.

-

-  Extension support

-    A void* pointer and a size_t field that can be used to help implement

-    extensions to this malloc.

-*/

-

-/* Bin types, widths and sizes */

-#define NSMALLBINS        (32U)

-#define NTREEBINS         (32U)

-#define SMALLBIN_SHIFT    (3U)

-#define SMALLBIN_WIDTH    (SIZE_T_ONE << SMALLBIN_SHIFT)

-#define TREEBIN_SHIFT     (8U)

-#define MIN_LARGE_SIZE    (SIZE_T_ONE << TREEBIN_SHIFT)

-#define MAX_SMALL_SIZE    (MIN_LARGE_SIZE - SIZE_T_ONE)

-#define MAX_SMALL_REQUEST (MAX_SMALL_SIZE - CHUNK_ALIGN_MASK - CHUNK_OVERHEAD)

-

-struct malloc_state {

-  binmap_t   smallmap;

-  binmap_t   treemap;

-  size_t     dvsize;

-  size_t     topsize;

-  char*      least_addr;

-  mchunkptr  dv;

-  mchunkptr  top;

-  size_t     trim_check;

-  size_t     release_checks;

-  size_t     magic;

-  mchunkptr  smallbins[(NSMALLBINS+1)*2];

-  tbinptr    treebins[NTREEBINS];

-  size_t     footprint;

-  size_t     max_footprint;

-  flag_t     mflags;

-  msegment   seg;

-#if USE_LOCKS

-  MLOCK_T    mutex;     /* locate lock among fields that rarely change */

-#endif /* USE_LOCKS */

-  void*      extp;      /* Unused but available for extensions */

-  size_t     exts;

-};

-

-typedef struct malloc_state*    mstate;

-

-/* ------------- Global malloc_state and malloc_params ------------------- */

-

-#if !ONLY_MSPACES

-

-/* The global malloc_state used for all non-"mspace" calls */

-static struct malloc_state _gm_;

-#define gm                 (&_gm_)

-#define is_global(M)       ((M) == &_gm_)

-

-#endif /* !ONLY_MSPACES */

-

-#define is_initialized(M)  ((M)->top != 0)

-

-/* -------------------------- system alloc setup ------------------------- */

-

-/* Operations on mflags */

-

-#define use_lock(M)           ((M)->mflags &   USE_LOCK_BIT)

-#define enable_lock(M)        ((M)->mflags |=  USE_LOCK_BIT)

-#define disable_lock(M)       ((M)->mflags &= ~USE_LOCK_BIT)

-

-#define use_mmap(M)           ((M)->mflags &   USE_MMAP_BIT)

-#define enable_mmap(M)        ((M)->mflags |=  USE_MMAP_BIT)

-#define disable_mmap(M)       ((M)->mflags &= ~USE_MMAP_BIT)

-

-#define use_noncontiguous(M)  ((M)->mflags &   USE_NONCONTIGUOUS_BIT)

-#define disable_contiguous(M) ((M)->mflags |=  USE_NONCONTIGUOUS_BIT)

-

-#define set_lock(M,L)\

- ((M)->mflags = (L)?\

-  ((M)->mflags | USE_LOCK_BIT) :\

-  ((M)->mflags & ~USE_LOCK_BIT))

-

-/* page-align a size */

-#define page_align(S)\

- (((S) + (mparams.page_size - SIZE_T_ONE)) & ~(mparams.page_size - SIZE_T_ONE))

-

-/* granularity-align a size */

-#define granularity_align(S)\

-  (((S) + (mparams.granularity - SIZE_T_ONE))\

-   & ~(mparams.granularity - SIZE_T_ONE))

-

-

-/* For mmap, use granularity alignment on windows, else page-align */

-#ifdef WIN32

-#define mmap_align(S) granularity_align(S)

-#else

-#define mmap_align(S) page_align(S)

-#endif

-

-/* For sys_alloc, enough padding to ensure can malloc request on success */

-#define SYS_ALLOC_PADDING (TOP_FOOT_SIZE + MALLOC_ALIGNMENT)

-

-#define is_page_aligned(S)\

-   (((size_t)(S) & (mparams.page_size - SIZE_T_ONE)) == 0)

-#define is_granularity_aligned(S)\

-   (((size_t)(S) & (mparams.granularity - SIZE_T_ONE)) == 0)

-

-/*  True if segment S holds address A */

-#define segment_holds(S, A)\

-  ((char*)(A) >= S->base && (char*)(A) < S->base + S->size)

-

-/* Return segment holding given address */

-static msegmentptr segment_holding(mstate m, char* addr) {

-  msegmentptr sp = &m->seg;

-  for (;;) {

-    if (addr >= sp->base && addr < sp->base + sp->size)

-      return sp;

-    if ((sp = sp->next) == 0)

-      return 0;

-  }

-}

-

-/* Return true if segment contains a segment link */

-static int has_segment_link(mstate m, msegmentptr ss) {

-  msegmentptr sp = &m->seg;

-  for (;;) {

-    if ((char*)sp >= ss->base && (char*)sp < ss->base + ss->size)

-      return 1;

-    if ((sp = sp->next) == 0)

-      return 0;

-  }

-}

-

-#ifndef MORECORE_CANNOT_TRIM

-#define should_trim(M,s)  ((s) > (M)->trim_check)

-#else  /* MORECORE_CANNOT_TRIM */

-#define should_trim(M,s)  (0)

-#endif /* MORECORE_CANNOT_TRIM */

-

-/*

-  TOP_FOOT_SIZE is padding at the end of a segment, including space

-  that may be needed to place segment records and fenceposts when new

-  noncontiguous segments are added.

-*/

-#define TOP_FOOT_SIZE\

-  (align_offset(chunk2mem(0))+pad_request(sizeof(struct malloc_segment))+MIN_CHUNK_SIZE)

-

-

-/* -------------------------------  Hooks -------------------------------- */

-

-/*

-  PREACTION should be defined to return 0 on success, and nonzero on

-  failure. If you are not using locking, you can redefine these to do

-  anything you like.

-*/

-

-#if USE_LOCKS

-

-#define PREACTION(M)  ((use_lock(M))? ACQUIRE_LOCK(&(M)->mutex) : 0)

-#define POSTACTION(M) { if (use_lock(M)) RELEASE_LOCK(&(M)->mutex); }

-#else /* USE_LOCKS */

-

-#ifndef PREACTION

-#define PREACTION(M) (0)

-#endif  /* PREACTION */

-

-#ifndef POSTACTION

-#define POSTACTION(M)

-#endif  /* POSTACTION */

-

-#endif /* USE_LOCKS */

-

-/*

-  CORRUPTION_ERROR_ACTION is triggered upon detected bad addresses.

-  USAGE_ERROR_ACTION is triggered on detected bad frees and

-  reallocs. The argument p is an address that might have triggered the

-  fault. It is ignored by the two predefined actions, but might be

-  useful in custom actions that try to help diagnose errors.

-*/

-

-#if PROCEED_ON_ERROR

-

-/* A count of the number of corruption errors causing resets */

-int malloc_corruption_error_count;

-

-/* default corruption action */

-static void reset_on_error(mstate m);

-

-#define CORRUPTION_ERROR_ACTION(m)  reset_on_error(m)

-#define USAGE_ERROR_ACTION(m, p)

-

-#else /* PROCEED_ON_ERROR */

-

-#ifndef CORRUPTION_ERROR_ACTION

-#define CORRUPTION_ERROR_ACTION(m) ABORT

-#endif /* CORRUPTION_ERROR_ACTION */

-

-#ifndef USAGE_ERROR_ACTION

-#define USAGE_ERROR_ACTION(m,p) ABORT

-#endif /* USAGE_ERROR_ACTION */

-

-#endif /* PROCEED_ON_ERROR */

-

-/* -------------------------- Debugging setup ---------------------------- */

-

-#if ! DEBUG

-

-#define check_free_chunk(M,P)

-#define check_inuse_chunk(M,P)

-#define check_malloced_chunk(M,P,N)

-#define check_mmapped_chunk(M,P)

-#define check_malloc_state(M)

-#define check_top_chunk(M,P)

-

-#else /* DEBUG */

-#define check_free_chunk(M,P)       do_check_free_chunk(M,P)

-#define check_inuse_chunk(M,P)      do_check_inuse_chunk(M,P)

-#define check_top_chunk(M,P)        do_check_top_chunk(M,P)

-#define check_malloced_chunk(M,P,N) do_check_malloced_chunk(M,P,N)

-#define check_mmapped_chunk(M,P)    do_check_mmapped_chunk(M,P)

-#define check_malloc_state(M)       do_check_malloc_state(M)

-

-static void   do_check_any_chunk(mstate m, mchunkptr p);

-static void   do_check_top_chunk(mstate m, mchunkptr p);

-static void   do_check_mmapped_chunk(mstate m, mchunkptr p);

-static void   do_check_inuse_chunk(mstate m, mchunkptr p);

-static void   do_check_free_chunk(mstate m, mchunkptr p);

-static void   do_check_malloced_chunk(mstate m, void* mem, size_t s);

-static void   do_check_tree(mstate m, tchunkptr t);

-static void   do_check_treebin(mstate m, bindex_t i);

-static void   do_check_smallbin(mstate m, bindex_t i);

-static void   do_check_malloc_state(mstate m);

-static int    bin_find(mstate m, mchunkptr x);

-static size_t traverse_and_check(mstate m);

-#endif /* DEBUG */

-

-/* ---------------------------- Indexing Bins ---------------------------- */

-

-#define is_small(s)         (((s) >> SMALLBIN_SHIFT) < NSMALLBINS)

-#define small_index(s)      (bindex_t)((s)  >> SMALLBIN_SHIFT)

-#define small_index2size(i) ((i)  << SMALLBIN_SHIFT)

-#define MIN_SMALL_INDEX     (small_index(MIN_CHUNK_SIZE))

-

-/* addressing by index. See above about smallbin repositioning */

-#define smallbin_at(M, i)   ((sbinptr)((char*)&((M)->smallbins[(i)<<1])))

-#define treebin_at(M,i)     (&((M)->treebins[i]))

-

-/* assign tree index for size S to variable I. Use x86 asm if possible  */

-#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))

-#define compute_tree_index(S, I)\

-{\

-  unsigned int X = S >> TREEBIN_SHIFT;\

-  if (X == 0)\

-    I = 0;\

-  else if (X > 0xFFFF)\

-    I = NTREEBINS-1;\

-  else {\

-    unsigned int K;\

-    __asm__("bsrl\t%1, %0\n\t" : "=r" (K) : "g"  (X));\

-    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\

-  }\

-}

-

-#elif defined (__INTEL_COMPILER)

-#define compute_tree_index(S, I)\

-{\

-  size_t X = S >> TREEBIN_SHIFT;\

-  if (X == 0)\

-    I = 0;\

-  else if (X > 0xFFFF)\

-    I = NTREEBINS-1;\

-  else {\

-    unsigned int K = _bit_scan_reverse (X); \

-    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\

-  }\

-}

-

-#elif defined(_MSC_VER) && _MSC_VER>=1300

-#define compute_tree_index(S, I)\

-{\

-  size_t X = S >> TREEBIN_SHIFT;\

-  if (X == 0)\

-    I = 0;\

-  else if (X > 0xFFFF)\

-    I = NTREEBINS-1;\

-  else {\

-    unsigned int K;\

-    _BitScanReverse((DWORD *) &K, (DWORD) X);\

-    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\

-  }\

-}

-

-#else /* GNUC */

-#define compute_tree_index(S, I)\

-{\

-  size_t X = S >> TREEBIN_SHIFT;\

-  if (X == 0)\

-    I = 0;\

-  else if (X > 0xFFFF)\

-    I = NTREEBINS-1;\

-  else {\

-    unsigned int Y = (unsigned int)X;\

-    unsigned int N = ((Y - 0x100) >> 16) & 8;\

-    unsigned int K = (((Y <<= N) - 0x1000) >> 16) & 4;\

-    N += K;\

-    N += K = (((Y <<= K) - 0x4000) >> 16) & 2;\

-    K = 14 - N + ((Y <<= K) >> 15);\

-    I = (K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1));\

-  }\

-}

-#endif /* GNUC */

-

-/* Bit representing maximum resolved size in a treebin at i */

-#define bit_for_tree_index(i) \

-   (i == NTREEBINS-1)? (SIZE_T_BITSIZE-1) : (((i) >> 1) + TREEBIN_SHIFT - 2)

-

-/* Shift placing maximum resolved bit in a treebin at i as sign bit */

-#define leftshift_for_tree_index(i) \

-   ((i == NTREEBINS-1)? 0 : \

-    ((SIZE_T_BITSIZE-SIZE_T_ONE) - (((i) >> 1) + TREEBIN_SHIFT - 2)))

-

-/* The size of the smallest chunk held in bin with index i */

-#define minsize_for_tree_index(i) \

-   ((SIZE_T_ONE << (((i) >> 1) + TREEBIN_SHIFT)) |  \

-   (((size_t)((i) & SIZE_T_ONE)) << (((i) >> 1) + TREEBIN_SHIFT - 1)))

-

-

-/* ------------------------ Operations on bin maps ----------------------- */

-

-/* bit corresponding to given index */

-#define idx2bit(i)              ((binmap_t)(1) << (i))

-

-/* Mark/Clear bits with given index */

-#define mark_smallmap(M,i)      ((M)->smallmap |=  idx2bit(i))

-#define clear_smallmap(M,i)     ((M)->smallmap &= ~idx2bit(i))

-#define smallmap_is_marked(M,i) ((M)->smallmap &   idx2bit(i))

-

-#define mark_treemap(M,i)       ((M)->treemap  |=  idx2bit(i))

-#define clear_treemap(M,i)      ((M)->treemap  &= ~idx2bit(i))

-#define treemap_is_marked(M,i)  ((M)->treemap  &   idx2bit(i))

-

-/* isolate the least set bit of a bitmap */

-#define least_bit(x)         ((x) & -(x))

-

-/* mask with all bits to left of least bit of x on */

-#define left_bits(x)         ((x<<1) | -(x<<1))

-

-/* mask with all bits to left of or equal to least bit of x on */

-#define same_or_left_bits(x) ((x) | -(x))

-

-/* index corresponding to given bit. Use x86 asm if possible */

-

-#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))

-#define compute_bit2idx(X, I)\

-{\

-  unsigned int J;\

-  __asm__("bsfl\t%1, %0\n\t" : "=r" (J) : "g" (X));\

-  I = (bindex_t)J;\

-}

-

-#elif defined (__INTEL_COMPILER)

-#define compute_bit2idx(X, I)\

-{\

-  unsigned int J;\

-  J = _bit_scan_forward (X); \

-  I = (bindex_t)J;\

-}

-

-#elif defined(_MSC_VER) && _MSC_VER>=1300

-#define compute_bit2idx(X, I)\

-{\

-  unsigned int J;\

-  _BitScanForward((DWORD *) &J, X);\

-  I = (bindex_t)J;\

-}

-

-#elif USE_BUILTIN_FFS

-#define compute_bit2idx(X, I) I = ffs(X)-1

-

-#else

-#define compute_bit2idx(X, I)\

-{\

-  unsigned int Y = X - 1;\

-  unsigned int K = Y >> (16-4) & 16;\

-  unsigned int N = K;        Y >>= K;\

-  N += K = Y >> (8-3) &  8;  Y >>= K;\

-  N += K = Y >> (4-2) &  4;  Y >>= K;\

-  N += K = Y >> (2-1) &  2;  Y >>= K;\

-  N += K = Y >> (1-0) &  1;  Y >>= K;\

-  I = (bindex_t)(N + Y);\

-}

-#endif /* GNUC */

-

-

-/* ----------------------- Runtime Check Support ------------------------- */

-

-/*

-  For security, the main invariant is that malloc/free/etc never

-  writes to a static address other than malloc_state, unless static

-  malloc_state itself has been corrupted, which cannot occur via

-  malloc (because of these checks). In essence this means that we

-  believe all pointers, sizes, maps etc held in malloc_state, but

-  check all of those linked or offsetted from other embedded data

-  structures.  These checks are interspersed with main code in a way

-  that tends to minimize their run-time cost.

-

-  When FOOTERS is defined, in addition to range checking, we also

-  verify footer fields of inuse chunks, which can be used guarantee

-  that the mstate controlling malloc/free is intact.  This is a

-  streamlined version of the approach described by William Robertson

-  et al in "Run-time Detection of Heap-based Overflows" LISA'03

-  http://www.usenix.org/events/lisa03/tech/robertson.html The footer

-  of an inuse chunk holds the xor of its mstate and a random seed,

-  that is checked upon calls to free() and realloc().  This is

-  (probablistically) unguessable from outside the program, but can be

-  computed by any code successfully malloc'ing any chunk, so does not

-  itself provide protection against code that has already broken

-  security through some other means.  Unlike Robertson et al, we

-  always dynamically check addresses of all offset chunks (previous,

-  next, etc). This turns out to be cheaper than relying on hashes.

-*/

-

-#if !INSECURE

-/* Check if address a is at least as high as any from MORECORE or MMAP */

-#define ok_address(M, a) ((char*)(a) >= (M)->least_addr)

-/* Check if address of next chunk n is higher than base chunk p */

-#define ok_next(p, n)    ((char*)(p) < (char*)(n))

-/* Check if p has inuse status */

-#define ok_inuse(p)     is_inuse(p)

-/* Check if p has its pinuse bit on */

-#define ok_pinuse(p)     pinuse(p)

-

-#else /* !INSECURE */

-#define ok_address(M, a) (1)

-#define ok_next(b, n)    (1)

-#define ok_inuse(p)      (1)

-#define ok_pinuse(p)     (1)

-#endif /* !INSECURE */

-

-#if (FOOTERS && !INSECURE)

-/* Check if (alleged) mstate m has expected magic field */

-#define ok_magic(M)      ((M)->magic == mparams.magic)

-#else  /* (FOOTERS && !INSECURE) */

-#define ok_magic(M)      (1)

-#endif /* (FOOTERS && !INSECURE) */

-

-

-/* In gcc, use __builtin_expect to minimize impact of checks */

-#if !INSECURE

-#if defined(__GNUC__) && __GNUC__ >= 3

-#define RTCHECK(e)  __builtin_expect(e, 1)

-#else /* GNUC */

-#define RTCHECK(e)  (e)

-#endif /* GNUC */

-#else /* !INSECURE */

-#define RTCHECK(e)  (1)

-#endif /* !INSECURE */

-

-/* macros to set up inuse chunks with or without footers */

-

-#if !FOOTERS

-

-#define mark_inuse_foot(M,p,s)

-

-/* Macros for setting head/foot of non-mmapped chunks */

-

-/* Set cinuse bit and pinuse bit of next chunk */

-#define set_inuse(M,p,s)\

-  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\

-  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)

-

-/* Set cinuse and pinuse of this chunk and pinuse of next chunk */

-#define set_inuse_and_pinuse(M,p,s)\

-  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\

-  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)

-

-/* Set size, cinuse and pinuse bit of this chunk */

-#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\

-  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT))

-

-#else /* FOOTERS */

-

-/* Set foot of inuse chunk to be xor of mstate and seed */

-#define mark_inuse_foot(M,p,s)\

-  (((mchunkptr)((char*)(p) + (s)))->prev_foot = ((size_t)(M) ^ mparams.magic))

-

-#define get_mstate_for(p)\

-  ((mstate)(((mchunkptr)((char*)(p) +\

-    (chunksize(p))))->prev_foot ^ mparams.magic))

-

-#define set_inuse(M,p,s)\

-  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\

-  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT), \

-  mark_inuse_foot(M,p,s))

-

-#define set_inuse_and_pinuse(M,p,s)\

-  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\

-  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT),\

- mark_inuse_foot(M,p,s))

-

-#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\

-  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\

-  mark_inuse_foot(M, p, s))

-

-#endif /* !FOOTERS */

-

-/* ---------------------------- setting mparams -------------------------- */

-

-#ifdef ENABLE_LARGE_PAGES

-typedef size_t (WINAPI *GetLargePageMinimum_t)(void);

-#endif

-

-/* Initialize mparams */

-static int init_mparams(void) {

-#ifdef NEED_GLOBAL_LOCK_INIT

-  if (malloc_global_mutex_status <= 0)

-    init_malloc_global_mutex();

-#endif

-

-  ACQUIRE_MALLOC_GLOBAL_LOCK();

-  if (mparams.magic == 0) {

-    size_t magic;

-    size_t psize;

-    size_t gsize;

-

-#ifndef WIN32

-    psize = malloc_getpagesize;

-    gsize = ((DEFAULT_GRANULARITY != 0)? DEFAULT_GRANULARITY : psize);

-#else /* WIN32 */

-    {

-      SYSTEM_INFO system_info;

-      GetSystemInfo(&system_info);

-      psize = system_info.dwPageSize;

-      gsize = ((DEFAULT_GRANULARITY != 0)?

-               DEFAULT_GRANULARITY : system_info.dwAllocationGranularity);

-#ifdef ENABLE_LARGE_PAGES

-      { 

-          GetLargePageMinimum_t GetLargePageMinimum_ = (GetLargePageMinimum_t) GetProcAddress(GetModuleHandle(__T("kernel32.dll")), "GetLargePageMinimum");

-          if(GetLargePageMinimum_) {

-              size_t largepagesize = GetLargePageMinimum_();

-              if(largepagesize) {

-                  psize = largepagesize;

-                  gsize = ((DEFAULT_GRANULARITY != 0)?

-                           DEFAULT_GRANULARITY : largepagesize);

-                  if(gsize < largepagesize) gsize = largepagesize;

-              }

-          }

-      }

-#endif

-    }

-#endif /* WIN32 */

-

-    /* Sanity-check configuration:

-       size_t must be unsigned and as wide as pointer type.

-       ints must be at least 4 bytes.

-       alignment must be at least 8.

-       Alignment, min chunk size, and page size must all be powers of 2.

-    */

-    if ((sizeof(size_t) != sizeof(char*)) ||

-        (MAX_SIZE_T < MIN_CHUNK_SIZE)  ||

-        (sizeof(int) < 4)  ||

-        (MALLOC_ALIGNMENT < (size_t)8U) ||

-        ((MALLOC_ALIGNMENT & (MALLOC_ALIGNMENT-SIZE_T_ONE)) != 0) ||

-        ((MCHUNK_SIZE      & (MCHUNK_SIZE-SIZE_T_ONE))      != 0) ||

-        ((gsize            & (gsize-SIZE_T_ONE))            != 0) ||

-        ((psize            & (psize-SIZE_T_ONE))            != 0))

-      ABORT;

-

-    mparams.granularity = gsize;

-    mparams.page_size = psize;

-    mparams.mmap_threshold = DEFAULT_MMAP_THRESHOLD;

-    mparams.trim_threshold = DEFAULT_TRIM_THRESHOLD;

-#if MORECORE_CONTIGUOUS

-    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT;

-#else  /* MORECORE_CONTIGUOUS */

-    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT|USE_NONCONTIGUOUS_BIT;

-#endif /* MORECORE_CONTIGUOUS */

-

-#if !ONLY_MSPACES

-    /* Set up lock for main malloc area */

-    gm->mflags = mparams.default_mflags;

-    INITIAL_LOCK(&gm->mutex);

-#endif

-

-    {

-#if USE_DEV_RANDOM

-      int fd;

-      unsigned char buf[sizeof(size_t)];

-      /* Try to use /dev/urandom, else fall back on using time */

-      if ((fd = open("/dev/urandom", O_RDONLY)) >= 0 &&

-          read(fd, buf, sizeof(buf)) == sizeof(buf)) {

-        magic = *((size_t *) buf);

-        close(fd);

-      }

-      else

-#endif /* USE_DEV_RANDOM */

-#ifdef WIN32

-        magic = (size_t)(GetTickCount() ^ (size_t)0x55555555U);

-#else

-        magic = (size_t)(time(0) ^ (size_t)0x55555555U);

-#endif

-      magic |= (size_t)8U;    /* ensure nonzero */

-      magic &= ~(size_t)7U;   /* improve chances of fault for bad values */

-      mparams.magic = magic;

-    }

-  }

-

-  RELEASE_MALLOC_GLOBAL_LOCK();

-  return 1;

-}

-

-/* support for mallopt */

-static int change_mparam(int param_number, int value) {

-  size_t val;

-  ensure_initialization();

-  val = (value == -1)? MAX_SIZE_T : (size_t)value;

-  switch(param_number) {

-  case M_TRIM_THRESHOLD:

-    mparams.trim_threshold = val;

-    return 1;

-  case M_GRANULARITY:

-    if (val >= mparams.page_size && ((val & (val-1)) == 0)) {

-      mparams.granularity = val;

-      return 1;

-    }

-    else

-      return 0;

-  case M_MMAP_THRESHOLD:

-    mparams.mmap_threshold = val;

-    return 1;

-  default:

-    return 0;

-  }

-}

-

-#if DEBUG

-/* ------------------------- Debugging Support --------------------------- */

-

-/* Check properties of any chunk, whether free, inuse, mmapped etc  */

-static void do_check_any_chunk(mstate m, mchunkptr p) {

-  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));

-  assert(ok_address(m, p));

-}

-

-/* Check properties of top chunk */

-static void do_check_top_chunk(mstate m, mchunkptr p) {

-  msegmentptr sp = segment_holding(m, (char*)p);

-  size_t  sz = p->head & ~INUSE_BITS; /* third-lowest bit can be set! */

-  assert(sp != 0);

-  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));

-  assert(ok_address(m, p));

-  assert(sz == m->topsize);

-  assert(sz > 0);

-  assert(sz == ((sp->base + sp->size) - (char*)p) - TOP_FOOT_SIZE);

-  assert(pinuse(p));

-  assert(!pinuse(chunk_plus_offset(p, sz)));

-}

-

-/* Check properties of (inuse) mmapped chunks */

-static void do_check_mmapped_chunk(mstate m, mchunkptr p) {

-  size_t  sz = chunksize(p);

-  size_t len = (sz + (p->prev_foot) + MMAP_FOOT_PAD);

-  assert(is_mmapped(p));

-  assert(use_mmap(m));

-  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));

-  assert(ok_address(m, p));

-  assert(!is_small(sz));

-  assert((len & (mparams.page_size-SIZE_T_ONE)) == 0);

-  assert(chunk_plus_offset(p, sz)->head == FENCEPOST_HEAD);

-  assert(chunk_plus_offset(p, sz+SIZE_T_SIZE)->head == 0);

-}

-

-/* Check properties of inuse chunks */

-static void do_check_inuse_chunk(mstate m, mchunkptr p) {

-  do_check_any_chunk(m, p);

-  assert(is_inuse(p));

-  assert(next_pinuse(p));

-  /* If not pinuse and not mmapped, previous chunk has OK offset */

-  assert(is_mmapped(p) || pinuse(p) || next_chunk(prev_chunk(p)) == p);

-  if (is_mmapped(p))

-    do_check_mmapped_chunk(m, p);

-}

-

-/* Check properties of free chunks */

-static void do_check_free_chunk(mstate m, mchunkptr p) {

-  size_t sz = chunksize(p);

-  mchunkptr next = chunk_plus_offset(p, sz);

-  do_check_any_chunk(m, p);

-  assert(!is_inuse(p));

-  assert(!next_pinuse(p));

-  assert (!is_mmapped(p));

-  if (p != m->dv && p != m->top) {

-    if (sz >= MIN_CHUNK_SIZE) {

-      assert((sz & CHUNK_ALIGN_MASK) == 0);

-      assert(is_aligned(chunk2mem(p)));

-      assert(next->prev_foot == sz);

-      assert(pinuse(p));

-      assert (next == m->top || is_inuse(next));

-      assert(p->fd->bk == p);

-      assert(p->bk->fd == p);

-    }

-    else  /* markers are always of size SIZE_T_SIZE */

-      assert(sz == SIZE_T_SIZE);

-  }

-}

-

-/* Check properties of malloced chunks at the point they are malloced */

-static void do_check_malloced_chunk(mstate m, void* mem, size_t s) {

-  if (mem != 0) {

-    mchunkptr p = mem2chunk(mem);

-    size_t sz = p->head & ~INUSE_BITS;

-    do_check_inuse_chunk(m, p);

-    assert((sz & CHUNK_ALIGN_MASK) == 0);

-    assert(sz >= MIN_CHUNK_SIZE);

-    assert(sz >= s);

-    /* unless mmapped, size is less than MIN_CHUNK_SIZE more than request */

-    assert(is_mmapped(p) || sz < (s + MIN_CHUNK_SIZE));

-  }

-}

-

-/* Check a tree and its subtrees.  */

-static void do_check_tree(mstate m, tchunkptr t) {

-  tchunkptr head = 0;

-  tchunkptr u = t;

-  bindex_t tindex = t->index;

-  size_t tsize = chunksize(t);

-  bindex_t idx;

-  compute_tree_index(tsize, idx);

-  assert(tindex == idx);

-  assert(tsize >= MIN_LARGE_SIZE);

-  assert(tsize >= minsize_for_tree_index(idx));

-  assert((idx == NTREEBINS-1) || (tsize < minsize_for_tree_index((idx+1))));

-

-  do { /* traverse through chain of same-sized nodes */

-    do_check_any_chunk(m, ((mchunkptr)u));

-    assert(u->index == tindex);

-    assert(chunksize(u) == tsize);

-    assert(!is_inuse(u));

-    assert(!next_pinuse(u));

-    assert(u->fd->bk == u);

-    assert(u->bk->fd == u);

-    if (u->parent == 0) {

-      assert(u->child[0] == 0);

-      assert(u->child[1] == 0);

-    }

-    else {

-      assert(head == 0); /* only one node on chain has parent */

-      head = u;

-      assert(u->parent != u);

-      assert (u->parent->child[0] == u ||

-              u->parent->child[1] == u ||

-              *((tbinptr*)(u->parent)) == u);

-      if (u->child[0] != 0) {

-        assert(u->child[0]->parent == u);

-        assert(u->child[0] != u);

-        do_check_tree(m, u->child[0]);

-      }

-      if (u->child[1] != 0) {

-        assert(u->child[1]->parent == u);

-        assert(u->child[1] != u);

-        do_check_tree(m, u->child[1]);

-      }

-      if (u->child[0] != 0 && u->child[1] != 0) {

-        assert(chunksize(u->child[0]) < chunksize(u->child[1]));

-      }

-    }

-    u = u->fd;

-  } while (u != t);

-  assert(head != 0);

-}

-

-/*  Check all the chunks in a treebin.  */

-static void do_check_treebin(mstate m, bindex_t i) {

-  tbinptr* tb = treebin_at(m, i);

-  tchunkptr t = *tb;

-  int empty = (m->treemap & (1U << i)) == 0;

-  if (t == 0)

-    assert(empty);

-  if (!empty)

-    do_check_tree(m, t);

-}

-

-/*  Check all the chunks in a smallbin.  */

-static void do_check_smallbin(mstate m, bindex_t i) {

-  sbinptr b = smallbin_at(m, i);

-  mchunkptr p = b->bk;

-  unsigned int empty = (m->smallmap & (1U << i)) == 0;

-  if (p == b)

-    assert(empty);

-  if (!empty) {

-    for (; p != b; p = p->bk) {

-      size_t size = chunksize(p);

-      mchunkptr q;

-      /* each chunk claims to be free */

-      do_check_free_chunk(m, p);

-      /* chunk belongs in bin */

-      assert(small_index(size) == i);

-      assert(p->bk == b || chunksize(p->bk) == chunksize(p));

-      /* chunk is followed by an inuse chunk */

-      q = next_chunk(p);

-      if (q->head != FENCEPOST_HEAD)

-        do_check_inuse_chunk(m, q);

-    }

-  }

-}

-

-/* Find x in a bin. Used in other check functions. */

-static int bin_find(mstate m, mchunkptr x) {

-  size_t size = chunksize(x);

-  if (is_small(size)) {

-    bindex_t sidx = small_index(size);

-    sbinptr b = smallbin_at(m, sidx);

-    if (smallmap_is_marked(m, sidx)) {

-      mchunkptr p = b;

-      do {

-        if (p == x)

-          return 1;

-      } while ((p = p->fd) != b);

-    }

-  }

-  else {

-    bindex_t tidx;

-    compute_tree_index(size, tidx);

-    if (treemap_is_marked(m, tidx)) {

-      tchunkptr t = *treebin_at(m, tidx);

-      size_t sizebits = size << leftshift_for_tree_index(tidx);

-      while (t != 0 && chunksize(t) != size) {

-        t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];

-        sizebits <<= 1;

-      }

-      if (t != 0) {

-        tchunkptr u = t;

-        do {

-          if (u == (tchunkptr)x)

-            return 1;

-        } while ((u = u->fd) != t);

-      }

-    }

-  }

-  return 0;

-}

-

-/* Traverse each chunk and check it; return total */

-static size_t traverse_and_check(mstate m) {

-  size_t sum = 0;

-  if (is_initialized(m)) {

-    msegmentptr s = &m->seg;

-    sum += m->topsize + TOP_FOOT_SIZE;

-    while (s != 0) {

-      mchunkptr q = align_as_chunk(s->base);

-      mchunkptr lastq = 0;

-      assert(pinuse(q));

-      while (segment_holds(s, q) &&

-             q != m->top && q->head != FENCEPOST_HEAD) {

-        sum += chunksize(q);

-        if (is_inuse(q)) {

-          assert(!bin_find(m, q));

-          do_check_inuse_chunk(m, q);

-        }

-        else {

-          assert(q == m->dv || bin_find(m, q));

-          assert(lastq == 0 || is_inuse(lastq)); /* Not 2 consecutive free */

-          do_check_free_chunk(m, q);

-        }

-        lastq = q;

-        q = next_chunk(q);

-      }

-      s = s->next;

-    }

-  }

-  return sum;

-}

-

-/* Check all properties of malloc_state. */

-static void do_check_malloc_state(mstate m) {

-  bindex_t i;

-  size_t total;

-  /* check bins */

-  for (i = 0; i < NSMALLBINS; ++i)

-    do_check_smallbin(m, i);

-  for (i = 0; i < NTREEBINS; ++i)

-    do_check_treebin(m, i);

-

-  if (m->dvsize != 0) { /* check dv chunk */

-    do_check_any_chunk(m, m->dv);

-    assert(m->dvsize == chunksize(m->dv));

-    assert(m->dvsize >= MIN_CHUNK_SIZE);

-    assert(bin_find(m, m->dv) == 0);

-  }

-

-  if (m->top != 0) {   /* check top chunk */

-    do_check_top_chunk(m, m->top);

-    /*assert(m->topsize == chunksize(m->top)); redundant */

-    assert(m->topsize > 0);

-    assert(bin_find(m, m->top) == 0);

-  }

-

-  total = traverse_and_check(m);

-  assert(total <= m->footprint);

-  assert(m->footprint <= m->max_footprint);

-}

-#endif /* DEBUG */

-

-/* ----------------------------- statistics ------------------------------ */

-

-#if !NO_MALLINFO

-static struct mallinfo internal_mallinfo(mstate m) {

-  struct mallinfo nm = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };

-  ensure_initialization();

-  if (!PREACTION(m)) {

-    check_malloc_state(m);

-    if (is_initialized(m)) {

-      size_t nfree = SIZE_T_ONE; /* top always free */

-      size_t mfree = m->topsize + TOP_FOOT_SIZE;

-      size_t sum = mfree;

-      msegmentptr s = &m->seg;

-      while (s != 0) {

-        mchunkptr q = align_as_chunk(s->base);

-        while (segment_holds(s, q) &&

-               q != m->top && q->head != FENCEPOST_HEAD) {

-          size_t sz = chunksize(q);

-          sum += sz;

-          if (!is_inuse(q)) {

-            mfree += sz;

-            ++nfree;

-          }

-          q = next_chunk(q);

-        }

-        s = s->next;

-      }

-

-      nm.arena    = sum;

-      nm.ordblks  = nfree;

-      nm.hblkhd   = m->footprint - sum;

-      nm.usmblks  = m->max_footprint;

-      nm.uordblks = m->footprint - mfree;

-      nm.fordblks = mfree;

-      nm.keepcost = m->topsize;

-    }

-

-    POSTACTION(m);

-  }

-  return nm;

-}

-#endif /* !NO_MALLINFO */

-

-static void internal_malloc_stats(mstate m) {

-  ensure_initialization();

-  if (!PREACTION(m)) {

-    size_t maxfp = 0;

-    size_t fp = 0;

-    size_t used = 0;

-    check_malloc_state(m);

-    if (is_initialized(m)) {

-      msegmentptr s = &m->seg;

-      maxfp = m->max_footprint;

-      fp = m->footprint;

-      used = fp - (m->topsize + TOP_FOOT_SIZE);

-

-      while (s != 0) {

-        mchunkptr q = align_as_chunk(s->base);

-        while (segment_holds(s, q) &&

-               q != m->top && q->head != FENCEPOST_HEAD) {

-          if (!is_inuse(q))

-            used -= chunksize(q);

-          q = next_chunk(q);

-        }

-        s = s->next;

-      }

-    }

-

-    fprintf(stderr, "max system bytes = %10lu\n", (unsigned long)(maxfp));

-    fprintf(stderr, "system bytes     = %10lu\n", (unsigned long)(fp));

-    fprintf(stderr, "in use bytes     = %10lu\n", (unsigned long)(used));

-

-    POSTACTION(m);

-  }

-}

-

-/* ----------------------- Operations on smallbins ----------------------- */

-

-/*

-  Various forms of linking and unlinking are defined as macros.  Even

-  the ones for trees, which are very long but have very short typical

-  paths.  This is ugly but reduces reliance on inlining support of

-  compilers.

-*/

-

-/* Link a free chunk into a smallbin  */

-#define insert_small_chunk(M, P, S) {\

-  bindex_t I  = small_index(S);\

-  mchunkptr B = smallbin_at(M, I);\

-  mchunkptr F = B;\

-  assert(S >= MIN_CHUNK_SIZE);\

-  if (!smallmap_is_marked(M, I))\

-    mark_smallmap(M, I);\

-  else if (RTCHECK(ok_address(M, B->fd)))\

-    F = B->fd;\

-  else {\

-    CORRUPTION_ERROR_ACTION(M);\

-  }\

-  B->fd = P;\

-  F->bk = P;\

-  P->fd = F;\

-  P->bk = B;\

-}

-

-/* Unlink a chunk from a smallbin  */

-#define unlink_small_chunk(M, P, S) {\

-  mchunkptr F = P->fd;\

-  mchunkptr B = P->bk;\

-  bindex_t I = small_index(S);\

-  assert(P != B);\

-  assert(P != F);\

-  assert(chunksize(P) == small_index2size(I));\

-  if (F == B)\

-    clear_smallmap(M, I);\

-  else if (RTCHECK((F == smallbin_at(M,I) || ok_address(M, F)) &&\

-                   (B == smallbin_at(M,I) || ok_address(M, B)))) {\

-    F->bk = B;\

-    B->fd = F;\

-  }\

-  else {\

-    CORRUPTION_ERROR_ACTION(M);\

-  }\

-}

-

-/* Unlink the first chunk from a smallbin */

-#define unlink_first_small_chunk(M, B, P, I) {\

-  mchunkptr F = P->fd;\

-  assert(P != B);\

-  assert(P != F);\

-  assert(chunksize(P) == small_index2size(I));\

-  if (B == F)\

-    clear_smallmap(M, I);\

-  else if (RTCHECK(ok_address(M, F))) {\

-    B->fd = F;\

-    F->bk = B;\

-  }\

-  else {\

-    CORRUPTION_ERROR_ACTION(M);\

-  }\

-}

-

-

-

-/* Replace dv node, binning the old one */

-/* Used only when dvsize known to be small */

-#define replace_dv(M, P, S) {\

-  size_t DVS = M->dvsize;\

-  if (DVS != 0) {\

-    mchunkptr DV = M->dv;\

-    assert(is_small(DVS));\

-    insert_small_chunk(M, DV, DVS);\

-  }\

-  M->dvsize = S;\

-  M->dv = P;\

-}

-

-/* ------------------------- Operations on trees ------------------------- */

-

-/* Insert chunk into tree */

-#define insert_large_chunk(M, X, S) {\

-  tbinptr* H;\

-  bindex_t I;\

-  compute_tree_index(S, I);\

-  H = treebin_at(M, I);\

-  X->index = I;\

-  X->child[0] = X->child[1] = 0;\

-  if (!treemap_is_marked(M, I)) {\

-    mark_treemap(M, I);\

-    *H = X;\

-    X->parent = (tchunkptr)H;\

-    X->fd = X->bk = X;\

-  }\

-  else {\

-    tchunkptr T = *H;\

-    size_t K = S << leftshift_for_tree_index(I);\

-    for (;;) {\

-      if (chunksize(T) != S) {\

-        tchunkptr* C = &(T->child[(K >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1]);\

-        K <<= 1;\

-        if (*C != 0)\

-          T = *C;\

-        else if (RTCHECK(ok_address(M, C))) {\

-          *C = X;\

-          X->parent = T;\

-          X->fd = X->bk = X;\

-          break;\

-        }\

-        else {\

-          CORRUPTION_ERROR_ACTION(M);\

-          break;\

-        }\

-      }\

-      else {\

-        tchunkptr F = T->fd;\

-        if (RTCHECK(ok_address(M, T) && ok_address(M, F))) {\

-          T->fd = F->bk = X;\

-          X->fd = F;\

-          X->bk = T;\

-          X->parent = 0;\

-          break;\

-        }\

-        else {\

-          CORRUPTION_ERROR_ACTION(M);\

-          break;\

-        }\

-      }\

-    }\

-  }\

-}

-

-/*

-  Unlink steps:

-

-  1. If x is a chained node, unlink it from its same-sized fd/bk links

-     and choose its bk node as its replacement.

-  2. If x was the last node of its size, but not a leaf node, it must

-     be replaced with a leaf node (not merely one with an open left or

-     right), to make sure that lefts and rights of descendents

-     correspond properly to bit masks.  We use the rightmost descendent

-     of x.  We could use any other leaf, but this is easy to locate and

-     tends to counteract removal of leftmosts elsewhere, and so keeps

-     paths shorter than minimally guaranteed.  This doesn't loop much

-     because on average a node in a tree is near the bottom.

-  3. If x is the base of a chain (i.e., has parent links) relink

-     x's parent and children to x's replacement (or null if none).

-*/

-

-#define unlink_large_chunk(M, X) {\

-  tchunkptr XP = X->parent;\

-  tchunkptr R;\

-  if (X->bk != X) {\

-    tchunkptr F = X->fd;\

-    R = X->bk;\

-    if (RTCHECK(ok_address(M, F))) {\

-      F->bk = R;\

-      R->fd = F;\

-    }\

-    else {\

-      CORRUPTION_ERROR_ACTION(M);\

-    }\

-  }\

-  else {\

-    tchunkptr* RP;\

-    if (((R = *(RP = &(X->child[1]))) != 0) ||\

-        ((R = *(RP = &(X->child[0]))) != 0)) {\

-      tchunkptr* CP;\

-      while ((*(CP = &(R->child[1])) != 0) ||\

-             (*(CP = &(R->child[0])) != 0)) {\

-        R = *(RP = CP);\

-      }\

-      if (RTCHECK(ok_address(M, RP)))\

-        *RP = 0;\

-      else {\

-        CORRUPTION_ERROR_ACTION(M);\

-      }\

-    }\

-  }\

-  if (XP != 0) {\

-    tbinptr* H = treebin_at(M, X->index);\

-    if (X == *H) {\

-      if ((*H = R) == 0) \

-        clear_treemap(M, X->index);\

-    }\

-    else if (RTCHECK(ok_address(M, XP))) {\

-      if (XP->child[0] == X) \

-        XP->child[0] = R;\

-      else \

-        XP->child[1] = R;\

-    }\

-    else\

-      CORRUPTION_ERROR_ACTION(M);\

-    if (R != 0) {\

-      if (RTCHECK(ok_address(M, R))) {\

-        tchunkptr C0, C1;\

-        R->parent = XP;\

-        if ((C0 = X->child[0]) != 0) {\

-          if (RTCHECK(ok_address(M, C0))) {\

-            R->child[0] = C0;\

-            C0->parent = R;\

-          }\

-          else\

-            CORRUPTION_ERROR_ACTION(M);\

-        }\

-        if ((C1 = X->child[1]) != 0) {\

-          if (RTCHECK(ok_address(M, C1))) {\

-            R->child[1] = C1;\

-            C1->parent = R;\

-          }\

-          else\

-            CORRUPTION_ERROR_ACTION(M);\

-        }\

-      }\

-      else\

-        CORRUPTION_ERROR_ACTION(M);\

-    }\

-  }\

-}

-

-/* Relays to large vs small bin operations */

-

-#define insert_chunk(M, P, S)\

-  if (is_small(S)) insert_small_chunk(M, P, S)\

-  else { tchunkptr TP = (tchunkptr)(P); insert_large_chunk(M, TP, S); }

-

-#define unlink_chunk(M, P, S)\

-  if (is_small(S)) unlink_small_chunk(M, P, S)\

-  else { tchunkptr TP = (tchunkptr)(P); unlink_large_chunk(M, TP); }

-

-

-/* Relays to internal calls to malloc/free from realloc, memalign etc */

-

-#if ONLY_MSPACES

-#define internal_malloc(m, b) mspace_malloc(m, b)

-#define internal_free(m, mem) mspace_free(m,mem);

-#else /* ONLY_MSPACES */

-#if MSPACES

-#define internal_malloc(m, b)\

-   (m == gm)? dlmalloc(b) : mspace_malloc(m, b)

-#define internal_free(m, mem)\

-   if (m == gm) dlfree(mem); else mspace_free(m,mem);

-#else /* MSPACES */

-#define internal_malloc(m, b) dlmalloc(b)

-#define internal_free(m, mem) dlfree(mem)

-#endif /* MSPACES */

-#endif /* ONLY_MSPACES */

-

-/* -----------------------  Direct-mmapping chunks ----------------------- */

-

-/*

-  Directly mmapped chunks are set up with an offset to the start of

-  the mmapped region stored in the prev_foot field of the chunk. This

-  allows reconstruction of the required argument to MUNMAP when freed,

-  and also allows adjustment of the returned chunk to meet alignment

-  requirements (especially in memalign).

-*/

-

-/* Malloc using mmap */

-static void* mmap_alloc(mstate m, size_t nb) {

-  size_t mmsize = mmap_align(nb + SIX_SIZE_T_SIZES + CHUNK_ALIGN_MASK);

-  if (mmsize > nb) {     /* Check for wrap around 0 */

-    char* mm = (char*)(CALL_DIRECT_MMAP(mmsize));

-    if (mm != CMFAIL) {

-      size_t offset = align_offset(chunk2mem(mm));

-      size_t psize = mmsize - offset - MMAP_FOOT_PAD;

-      mchunkptr p = (mchunkptr)(mm + offset);

-      p->prev_foot = offset;

-      p->head = psize;

-      mark_inuse_foot(m, p, psize);

-      chunk_plus_offset(p, psize)->head = FENCEPOST_HEAD;

-      chunk_plus_offset(p, psize+SIZE_T_SIZE)->head = 0;

-

-      if (m->least_addr == 0 || mm < m->least_addr)

-        m->least_addr = mm;

-      if ((m->footprint += mmsize) > m->max_footprint)

-        m->max_footprint = m->footprint;

-      assert(is_aligned(chunk2mem(p)));

-      check_mmapped_chunk(m, p);

-      return chunk2mem(p);

-    }

-  }

-  return 0;

-}

-

-/* Realloc using mmap */

-static mchunkptr mmap_resize(mstate m, mchunkptr oldp, size_t nb) {

-  size_t oldsize = chunksize(oldp);

-  if (is_small(nb)) /* Can't shrink mmap regions below small size */

-    return 0;

-  /* Keep old chunk if big enough but not too big */

-  if (oldsize >= nb + SIZE_T_SIZE &&

-      (oldsize - nb) <= (mparams.granularity << 1))

-    return oldp;

-  else {

-    size_t offset = oldp->prev_foot;

-    size_t oldmmsize = oldsize + offset + MMAP_FOOT_PAD;

-    size_t newmmsize = mmap_align(nb + SIX_SIZE_T_SIZES + CHUNK_ALIGN_MASK);

-    char* cp = (char*)CALL_MREMAP((char*)oldp - offset,

-                                  oldmmsize, newmmsize, 1);

-    if (cp != CMFAIL) {

-      mchunkptr newp = (mchunkptr)(cp + offset);

-      size_t psize = newmmsize - offset - MMAP_FOOT_PAD;

-      newp->head = psize;

-      mark_inuse_foot(m, newp, psize);

-      chunk_plus_offset(newp, psize)->head = FENCEPOST_HEAD;

-      chunk_plus_offset(newp, psize+SIZE_T_SIZE)->head = 0;

-

-      if (cp < m->least_addr)

-        m->least_addr = cp;

-      if ((m->footprint += newmmsize - oldmmsize) > m->max_footprint)

-        m->max_footprint = m->footprint;

-      check_mmapped_chunk(m, newp);

-      return newp;

-    }

-  }

-  return 0;

-}

-

-/* -------------------------- mspace management -------------------------- */

-

-/* Initialize top chunk and its size */

-static void init_top(mstate m, mchunkptr p, size_t psize) {

-  /* Ensure alignment */

-  size_t offset = align_offset(chunk2mem(p));

-  p = (mchunkptr)((char*)p + offset);

-  psize -= offset;

-

-  m->top = p;

-  m->topsize = psize;

-  p->head = psize | PINUSE_BIT;

-  /* set size of fake trailing chunk holding overhead space only once */

-  chunk_plus_offset(p, psize)->head = TOP_FOOT_SIZE;

-  m->trim_check = mparams.trim_threshold; /* reset on each update */

-}

-

-/* Initialize bins for a new mstate that is otherwise zeroed out */

-static void init_bins(mstate m) {

-  /* Establish circular links for smallbins */

-  bindex_t i;

-  for (i = 0; i < NSMALLBINS; ++i) {

-    sbinptr bin = smallbin_at(m,i);

-    bin->fd = bin->bk = bin;

-  }

-}

-

-#if PROCEED_ON_ERROR

-

-/* default corruption action */

-static void reset_on_error(mstate m) {

-  int i;

-  ++malloc_corruption_error_count;

-  /* Reinitialize fields to forget about all memory */

-  m->smallbins = m->treebins = 0;

-  m->dvsize = m->topsize = 0;

-  m->seg.base = 0;

-  m->seg.size = 0;

-  m->seg.next = 0;

-  m->top = m->dv = 0;

-  for (i = 0; i < NTREEBINS; ++i)

-    *treebin_at(m, i) = 0;

-  init_bins(m);

-}

-#endif /* PROCEED_ON_ERROR */

-

-/* Allocate chunk and prepend remainder with chunk in successor base. */

-static void* prepend_alloc(mstate m, char* newbase, char* oldbase,

-                           size_t nb) {

-  mchunkptr p = align_as_chunk(newbase);

-  mchunkptr oldfirst = align_as_chunk(oldbase);

-  size_t psize = (char*)oldfirst - (char*)p;

-  mchunkptr q = chunk_plus_offset(p, nb);

-  size_t qsize = psize - nb;

-  set_size_and_pinuse_of_inuse_chunk(m, p, nb);

-

-  assert((char*)oldfirst > (char*)q);

-  assert(pinuse(oldfirst));

-  assert(qsize >= MIN_CHUNK_SIZE);

-

-  /* consolidate remainder with first chunk of old base */

-  if (oldfirst == m->top) {

-    size_t tsize = m->topsize += qsize;

-    m->top = q;

-    q->head = tsize | PINUSE_BIT;

-    check_top_chunk(m, q);

-  }

-  else if (oldfirst == m->dv) {

-    size_t dsize = m->dvsize += qsize;

-    m->dv = q;

-    set_size_and_pinuse_of_free_chunk(q, dsize);

-  }

-  else {

-    if (!is_inuse(oldfirst)) {

-      size_t nsize = chunksize(oldfirst);

-      unlink_chunk(m, oldfirst, nsize);

-      oldfirst = chunk_plus_offset(oldfirst, nsize);

-      qsize += nsize;

-    }

-    set_free_with_pinuse(q, qsize, oldfirst);

-    insert_chunk(m, q, qsize);

-    check_free_chunk(m, q);

-  }

-

-  check_malloced_chunk(m, chunk2mem(p), nb);

-  return chunk2mem(p);

-}

-

-/* Add a segment to hold a new noncontiguous region */

-static void add_segment(mstate m, char* tbase, size_t tsize, flag_t mmapped) {

-  /* Determine locations and sizes of segment, fenceposts, old top */

-  char* old_top = (char*)m->top;

-  msegmentptr oldsp = segment_holding(m, old_top);

-  char* old_end = oldsp->base + oldsp->size;

-  size_t ssize = pad_request(sizeof(struct malloc_segment));

-  char* rawsp = old_end - (ssize + FOUR_SIZE_T_SIZES + CHUNK_ALIGN_MASK);

-  size_t offset = align_offset(chunk2mem(rawsp));

-  char* asp = rawsp + offset;

-  char* csp = (asp < (old_top + MIN_CHUNK_SIZE))? old_top : asp;

-  mchunkptr sp = (mchunkptr)csp;

-  msegmentptr ss = (msegmentptr)(chunk2mem(sp));

-  mchunkptr tnext = chunk_plus_offset(sp, ssize);

-  mchunkptr p = tnext;

-  int nfences = 0;

-

-  /* reset top to new space */

-  init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);

-

-  /* Set up segment record */

-  assert(is_aligned(ss));

-  set_size_and_pinuse_of_inuse_chunk(m, sp, ssize);

-  *ss = m->seg; /* Push current record */

-  m->seg.base = tbase;

-  m->seg.size = tsize;

-  m->seg.sflags = mmapped;

-  m->seg.next = ss;

-

-  /* Insert trailing fenceposts */

-  for (;;) {

-    mchunkptr nextp = chunk_plus_offset(p, SIZE_T_SIZE);

-    p->head = FENCEPOST_HEAD;

-    ++nfences;

-    if ((char*)(&(nextp->head)) < old_end)

-      p = nextp;

-    else

-      break;

-  }

-  assert(nfences >= 2);

-

-  /* Insert the rest of old top into a bin as an ordinary free chunk */

-  if (csp != old_top) {

-    mchunkptr q = (mchunkptr)old_top;

-    size_t psize = csp - old_top;

-    mchunkptr tn = chunk_plus_offset(q, psize);

-    set_free_with_pinuse(q, psize, tn);

-    insert_chunk(m, q, psize);

-  }

-

-  check_top_chunk(m, m->top);

-}

-

-/* -------------------------- System allocation -------------------------- */

-

-/* Get memory from system using MORECORE or MMAP */

-static void* sys_alloc(mstate m, size_t nb) {

-  char* tbase = CMFAIL;

-  size_t tsize = 0;

-  flag_t mmap_flag = 0;

-

-  ensure_initialization();

-

-  /* Directly map large chunks, but only if already initialized */

-  if (use_mmap(m) && nb >= mparams.mmap_threshold && m->topsize != 0) {

-    void* mem = mmap_alloc(m, nb);

-    if (mem != 0)

-      return mem;

-  }

-

-  /*

-    Try getting memory in any of three ways (in most-preferred to

-    least-preferred order):

-    1. A call to MORECORE that can normally contiguously extend memory.

-       (disabled if not MORECORE_CONTIGUOUS or not HAVE_MORECORE or

-       or main space is mmapped or a previous contiguous call failed)

-    2. A call to MMAP new space (disabled if not HAVE_MMAP).

-       Note that under the default settings, if MORECORE is unable to

-       fulfill a request, and HAVE_MMAP is true, then mmap is

-       used as a noncontiguous system allocator. This is a useful backup

-       strategy for systems with holes in address spaces -- in this case

-       sbrk cannot contiguously expand the heap, but mmap may be able to

-       find space.

-    3. A call to MORECORE that cannot usually contiguously extend memory.

-       (disabled if not HAVE_MORECORE)

-

-   In all cases, we need to request enough bytes from system to ensure

-   we can malloc nb bytes upon success, so pad with enough space for

-   top_foot, plus alignment-pad to make sure we don't lose bytes if

-   not on boundary, and round this up to a granularity unit.

-  */

-

-  if (MORECORE_CONTIGUOUS && !use_noncontiguous(m)) {

-    char* br = CMFAIL;

-    msegmentptr ss = (m->top == 0)? 0 : segment_holding(m, (char*)m->top);

-    size_t asize = 0;

-    ACQUIRE_MALLOC_GLOBAL_LOCK();

-

-    if (ss == 0) {  /* First time through or recovery */

-      char* base = (char*)CALL_MORECORE(0);

-      if (base != CMFAIL) {

-        asize = granularity_align(nb + SYS_ALLOC_PADDING);

-        /* Adjust to end on a page boundary */

-        if (!is_page_aligned(base))

-          asize += (page_align((size_t)base) - (size_t)base);

-        /* Can't call MORECORE if size is negative when treated as signed */

-        if (asize < HALF_MAX_SIZE_T &&

-            (br = (char*)(CALL_MORECORE(asize))) == base) {

-          tbase = base;

-          tsize = asize;

-        }

-      }

-    }

-    else {

-      /* Subtract out existing available top space from MORECORE request. */

-      asize = granularity_align(nb - m->topsize + SYS_ALLOC_PADDING);

-      /* Use mem here only if it did continuously extend old space */

-      if (asize < HALF_MAX_SIZE_T &&

-          (br = (char*)(CALL_MORECORE(asize))) == ss->base+ss->size) {

-        tbase = br;

-        tsize = asize;

-      }

-    }

-

-    if (tbase == CMFAIL) {    /* Cope with partial failure */

-      if (br != CMFAIL) {    /* Try to use/extend the space we did get */

-        if (asize < HALF_MAX_SIZE_T &&

-            asize < nb + SYS_ALLOC_PADDING) {

-          size_t esize = granularity_align(nb + SYS_ALLOC_PADDING - asize);

-          if (esize < HALF_MAX_SIZE_T) {

-            char* end = (char*)CALL_MORECORE(esize);

-            if (end != CMFAIL)

-              asize += esize;

-            else {            /* Can't use; try to release */

-              (void) CALL_MORECORE(-asize);

-              br = CMFAIL;

-            }

-          }

-        }

-      }

-      if (br != CMFAIL) {    /* Use the space we did get */

-        tbase = br;

-        tsize = asize;

-      }

-      else

-        disable_contiguous(m); /* Don't try contiguous path in the future */

-    }

-

-    RELEASE_MALLOC_GLOBAL_LOCK();

-  }

-

-  if (HAVE_MMAP && tbase == CMFAIL) {  /* Try MMAP */

-    size_t rsize = granularity_align(nb + SYS_ALLOC_PADDING);

-    if (rsize > nb) { /* Fail if wraps around zero */

-      char* mp = (char*)(CALL_MMAP(rsize));

-      if (mp != CMFAIL) {

-        tbase = mp;

-        tsize = rsize;

-        mmap_flag = USE_MMAP_BIT;

-      }

-    }

-  }

-

-  if (HAVE_MORECORE && tbase == CMFAIL) { /* Try noncontiguous MORECORE */

-    size_t asize = granularity_align(nb + SYS_ALLOC_PADDING);

-    if (asize < HALF_MAX_SIZE_T) {

-      char* br = CMFAIL;

-      char* end = CMFAIL;

-      ACQUIRE_MALLOC_GLOBAL_LOCK();

-      br = (char*)(CALL_MORECORE(asize));

-      end = (char*)(CALL_MORECORE(0));

-      RELEASE_MALLOC_GLOBAL_LOCK();

-      if (br != CMFAIL && end != CMFAIL && br < end) {

-        size_t ssize = end - br;

-        if (ssize > nb + TOP_FOOT_SIZE) {

-          tbase = br;

-          tsize = ssize;

-        }

-      }

-    }

-  }

-

-  if (tbase != CMFAIL) {

-

-    if ((m->footprint += tsize) > m->max_footprint)

-      m->max_footprint = m->footprint;

-

-    if (!is_initialized(m)) { /* first-time initialization */

-      if (m->least_addr == 0 || tbase < m->least_addr)

-        m->least_addr = tbase;

-      m->seg.base = tbase;

-      m->seg.size = tsize;

-      m->seg.sflags = mmap_flag;

-      m->magic = mparams.magic;

-      m->release_checks = MAX_RELEASE_CHECK_RATE;

-      init_bins(m);

-#if !ONLY_MSPACES

-      if (is_global(m))

-        init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);

-      else

-#endif

-      {

-        /* Offset top by embedded malloc_state */

-        mchunkptr mn = next_chunk(mem2chunk(m));

-        init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) -TOP_FOOT_SIZE);

-      }

-    }

-

-    else {

-      /* Try to merge with an existing segment */

-      msegmentptr sp = &m->seg;

-      /* Only consider most recent segment if traversal suppressed */

-      while (sp != 0 && tbase != sp->base + sp->size)

-        sp = (NO_SEGMENT_TRAVERSAL) ? 0 : sp->next;

-      if (sp != 0 &&

-          !is_extern_segment(sp) &&

-          (sp->sflags & USE_MMAP_BIT) == mmap_flag &&

-          segment_holds(sp, m->top)) { /* append */

-        sp->size += tsize;

-        init_top(m, m->top, m->topsize + tsize);

-      }

-      else {

-        if (tbase < m->least_addr)

-          m->least_addr = tbase;

-        sp = &m->seg;

-        while (sp != 0 && sp->base != tbase + tsize)

-          sp = (NO_SEGMENT_TRAVERSAL) ? 0 : sp->next;

-        if (sp != 0 &&

-            !is_extern_segment(sp) &&

-            (sp->sflags & USE_MMAP_BIT) == mmap_flag) {

-          char* oldbase = sp->base;

-          sp->base = tbase;

-          sp->size += tsize;

-          return prepend_alloc(m, tbase, oldbase, nb);

-        }

-        else

-          add_segment(m, tbase, tsize, mmap_flag);

-      }

-    }

-

-    if (nb < m->topsize) { /* Allocate from new or extended top space */

-      size_t rsize = m->topsize -= nb;

-      mchunkptr p = m->top;

-      mchunkptr r = m->top = chunk_plus_offset(p, nb);

-      r->head = rsize | PINUSE_BIT;

-      set_size_and_pinuse_of_inuse_chunk(m, p, nb);

-      check_top_chunk(m, m->top);

-      check_malloced_chunk(m, chunk2mem(p), nb);

-      return chunk2mem(p);

-    }

-  }

-

-  MALLOC_FAILURE_ACTION;

-  return 0;

-}

-

-/* -----------------------  system deallocation -------------------------- */

-

-/* Unmap and unlink any mmapped segments that don't contain used chunks */

-static size_t release_unused_segments(mstate m) {

-  size_t released = 0;

-  int nsegs = 0;

-  msegmentptr pred = &m->seg;

-  msegmentptr sp = pred->next;

-  while (sp != 0) {

-    char* base = sp->base;

-    size_t size = sp->size;

-    msegmentptr next = sp->next;

-    ++nsegs;

-    if (is_mmapped_segment(sp) && !is_extern_segment(sp)) {

-      mchunkptr p = align_as_chunk(base);

-      size_t psize = chunksize(p);

-      /* Can unmap if first chunk holds entire segment and not pinned */

-      if (!is_inuse(p) && (char*)p + psize >= base + size - TOP_FOOT_SIZE) {

-        tchunkptr tp = (tchunkptr)p;

-        assert(segment_holds(sp, (char*)sp));

-        if (p == m->dv) {

-          m->dv = 0;

-          m->dvsize = 0;

-        }

-        else {

-          unlink_large_chunk(m, tp);

-        }

-        if (CALL_MUNMAP(base, size) == 0) {

-          released += size;

-          m->footprint -= size;

-          /* unlink obsoleted record */

-          sp = pred;

-          sp->next = next;

-        }

-        else { /* back out if cannot unmap */

-          insert_large_chunk(m, tp, psize);

-        }

-      }

-    }

-    if (NO_SEGMENT_TRAVERSAL) /* scan only first segment */

-      break;

-    pred = sp;

-    sp = next;

-  }

-  /* Reset check counter */

-  m->release_checks = ((nsegs > MAX_RELEASE_CHECK_RATE)?

-                       nsegs : MAX_RELEASE_CHECK_RATE);

-  return released;

-}

-

-static int sys_trim(mstate m, size_t pad) {

-  size_t released = 0;

-  ensure_initialization();

-  if (pad < MAX_REQUEST && is_initialized(m)) {

-    pad += TOP_FOOT_SIZE; /* ensure enough room for segment overhead */

-

-    if (m->topsize > pad) {

-      /* Shrink top space in granularity-size units, keeping at least one */

-      size_t unit = mparams.granularity;

-      size_t extra = ((m->topsize - pad + (unit - SIZE_T_ONE)) / unit -

-                      SIZE_T_ONE) * unit;

-      msegmentptr sp = segment_holding(m, (char*)m->top);

-

-      if (!is_extern_segment(sp)) {

-        if (is_mmapped_segment(sp)) {

-          if (HAVE_MMAP &&

-              sp->size >= extra &&

-              !has_segment_link(m, sp)) { /* can't shrink if pinned */

-            size_t newsize = sp->size - extra;

-            /* Prefer mremap, fall back to munmap */

-            if ((CALL_MREMAP(sp->base, sp->size, newsize, 0) != MFAIL) ||

-                (CALL_MUNMAP(sp->base + newsize, extra) == 0)) {

-              released = extra;

-            }

-          }

-        }

-        else if (HAVE_MORECORE) {

-          if (extra >= HALF_MAX_SIZE_T) /* Avoid wrapping negative */

-            extra = (HALF_MAX_SIZE_T) + SIZE_T_ONE - unit;

-          ACQUIRE_MALLOC_GLOBAL_LOCK();

-          {

-            /* Make sure end of memory is where we last set it. */

-            char* old_br = (char*)(CALL_MORECORE(0));

-            if (old_br == sp->base + sp->size) {

-              char* rel_br = (char*)(CALL_MORECORE(-extra));

-              char* new_br = (char*)(CALL_MORECORE(0));

-              if (rel_br != CMFAIL && new_br < old_br)

-                released = old_br - new_br;

-            }

-          }

-          RELEASE_MALLOC_GLOBAL_LOCK();

-        }

-      }

-

-      if (released != 0) {

-        sp->size -= released;

-        m->footprint -= released;

-        init_top(m, m->top, m->topsize - released);

-        check_top_chunk(m, m->top);

-      }

-    }

-

-    /* Unmap any unused mmapped segments */

-    if (HAVE_MMAP)

-      released += release_unused_segments(m);

-

-    /* On failure, disable autotrim to avoid repeated failed future calls */

-    if (released == 0 && m->topsize > m->trim_check)

-      m->trim_check = MAX_SIZE_T;

-  }

-

-  return (released != 0)? 1 : 0;

-}

-

-

-/* ---------------------------- malloc support --------------------------- */

-

-/* allocate a large request from the best fitting chunk in a treebin */

-static void* tmalloc_large(mstate m, size_t nb) {

-  tchunkptr v = 0;

-  size_t rsize = -nb; /* Unsigned negation */

-  tchunkptr t;

-  bindex_t idx;

-  compute_tree_index(nb, idx);

-  if ((t = *treebin_at(m, idx)) != 0) {

-    /* Traverse tree for this bin looking for node with size == nb */

-    size_t sizebits = nb << leftshift_for_tree_index(idx);

-    tchunkptr rst = 0;  /* The deepest untaken right subtree */

-    for (;;) {

-      tchunkptr rt;

-      size_t trem = chunksize(t) - nb;

-      if (trem < rsize) {

-        v = t;

-        if ((rsize = trem) == 0)

-          break;

-      }

-      rt = t->child[1];

-      t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];

-      if (rt != 0 && rt != t)

-        rst = rt;

-      if (t == 0) {

-        t = rst; /* set t to least subtree holding sizes > nb */

-        break;

-      }

-      sizebits <<= 1;

-    }

-  }

-  if (t == 0 && v == 0) { /* set t to root of next non-empty treebin */

-    binmap_t leftbits = left_bits(idx2bit(idx)) & m->treemap;

-    if (leftbits != 0) {

-      bindex_t i;

-      binmap_t leastbit = least_bit(leftbits);

-      compute_bit2idx(leastbit, i);

-      t = *treebin_at(m, i);

-    }

-  }

-

-  while (t != 0) { /* find smallest of tree or subtree */

-    size_t trem = chunksize(t) - nb;

-    if (trem < rsize) {

-      rsize = trem;

-      v = t;

-    }

-    t = leftmost_child(t);

-  }

-

-  /*  If dv is a better fit, return 0 so malloc will use it */

-  if (v != 0 && rsize < (size_t)(m->dvsize - nb)) {

-    if (RTCHECK(ok_address(m, v))) { /* split */

-      mchunkptr r = chunk_plus_offset(v, nb);

-      assert(chunksize(v) == rsize + nb);

-      if (RTCHECK(ok_next(v, r))) {

-        unlink_large_chunk(m, v);

-        if (rsize < MIN_CHUNK_SIZE)

-          set_inuse_and_pinuse(m, v, (rsize + nb));

-        else {

-          set_size_and_pinuse_of_inuse_chunk(m, v, nb);

-          set_size_and_pinuse_of_free_chunk(r, rsize);

-          insert_chunk(m, r, rsize);

-        }

-        return chunk2mem(v);

-      }

-    }

-    CORRUPTION_ERROR_ACTION(m);

-  }

-  return 0;

-}

-

-/* allocate a small request from the best fitting chunk in a treebin */

-static void* tmalloc_small(mstate m, size_t nb) {

-  tchunkptr t, v;

-  size_t rsize;

-  bindex_t i;

-  binmap_t leastbit = least_bit(m->treemap);

-  compute_bit2idx(leastbit, i);

-  v = t = *treebin_at(m, i);

-  rsize = chunksize(t) - nb;

-

-  while ((t = leftmost_child(t)) != 0) {

-    size_t trem = chunksize(t) - nb;

-    if (trem < rsize) {

-      rsize = trem;

-      v = t;

-    }

-  }

-

-  if (RTCHECK(ok_address(m, v))) {

-    mchunkptr r = chunk_plus_offset(v, nb);

-    assert(chunksize(v) == rsize + nb);

-    if (RTCHECK(ok_next(v, r))) {

-      unlink_large_chunk(m, v);

-      if (rsize < MIN_CHUNK_SIZE)

-        set_inuse_and_pinuse(m, v, (rsize + nb));

-      else {

-        set_size_and_pinuse_of_inuse_chunk(m, v, nb);

-        set_size_and_pinuse_of_free_chunk(r, rsize);

-        replace_dv(m, r, rsize);

-      }

-      return chunk2mem(v);

-    }

-  }

-

-  CORRUPTION_ERROR_ACTION(m);

-  return 0;

-}

-

-/* --------------------------- realloc support --------------------------- */

-

-static void* internal_realloc(mstate m, void* oldmem, size_t bytes) {

-  if (bytes >= MAX_REQUEST) {

-    MALLOC_FAILURE_ACTION;

-    return 0;

-  }

-  if (!PREACTION(m)) {

-    mchunkptr oldp = mem2chunk(oldmem);

-    size_t oldsize = chunksize(oldp);

-    mchunkptr next = chunk_plus_offset(oldp, oldsize);

-    mchunkptr newp = 0;

-    void* extra = 0;

-

-    /* Try to either shrink or extend into top. Else malloc-copy-free */

-

-    if (RTCHECK(ok_address(m, oldp) && ok_inuse(oldp) &&

-                ok_next(oldp, next) && ok_pinuse(next))) {

-      size_t nb = request2size(bytes);

-      if (is_mmapped(oldp))

-        newp = mmap_resize(m, oldp, nb);

-      else if (oldsize >= nb) { /* already big enough */

-        size_t rsize = oldsize - nb;

-        newp = oldp;

-        if (rsize >= MIN_CHUNK_SIZE) {

-          mchunkptr remainder = chunk_plus_offset(newp, nb);

-          set_inuse(m, newp, nb);

-          set_inuse_and_pinuse(m, remainder, rsize);

-          extra = chunk2mem(remainder);

-        }

-      }

-      else if (next == m->top && oldsize + m->topsize > nb) {

-        /* Expand into top */

-        size_t newsize = oldsize + m->topsize;

-        size_t newtopsize = newsize - nb;

-        mchunkptr newtop = chunk_plus_offset(oldp, nb);

-        set_inuse(m, oldp, nb);

-        newtop->head = newtopsize |PINUSE_BIT;

-        m->top = newtop;

-        m->topsize = newtopsize;

-        newp = oldp;

-      }

-    }

-    else {

-      USAGE_ERROR_ACTION(m, oldmem);

-      POSTACTION(m);

-      return 0;

-    }

-#if DEBUG

-    if (newp != 0) {

-      check_inuse_chunk(m, newp); /* Check requires lock */

-    }

-#endif

-

-    POSTACTION(m);

-

-    if (newp != 0) {

-      if (extra != 0) {

-        internal_free(m, extra);

-      }

-      return chunk2mem(newp);

-    }

-    else {

-      void* newmem = internal_malloc(m, bytes);

-      if (newmem != 0) {

-        size_t oc = oldsize - overhead_for(oldp);

-        memcpy(newmem, oldmem, (oc < bytes)? oc : bytes);

-        internal_free(m, oldmem);

-      }

-      return newmem;

-    }

-  }

-  return 0;

-}

-

-/* --------------------------- memalign support -------------------------- */

-

-static void* internal_memalign(mstate m, size_t alignment, size_t bytes) {

-  if (alignment <= MALLOC_ALIGNMENT)    /* Can just use malloc */

-    return internal_malloc(m, bytes);

-  if (alignment <  MIN_CHUNK_SIZE) /* must be at least a minimum chunk size */

-    alignment = MIN_CHUNK_SIZE;

-  if ((alignment & (alignment-SIZE_T_ONE)) != 0) {/* Ensure a power of 2 */

-    size_t a = MALLOC_ALIGNMENT << 1;

-    while (a < alignment) a <<= 1;

-    alignment = a;

-  }

-

-  if (bytes >= MAX_REQUEST - alignment) {

-    if (m != 0)  { /* Test isn't needed but avoids compiler warning */

-      MALLOC_FAILURE_ACTION;

-    }

-  }

-  else {

-    size_t nb = request2size(bytes);

-    size_t req = nb + alignment + MIN_CHUNK_SIZE - CHUNK_OVERHEAD;

-    char* mem = (char*)internal_malloc(m, req);

-    if (mem != 0) {

-      void* leader = 0;

-      void* trailer = 0;

-      mchunkptr p = mem2chunk(mem);

-

-      if (PREACTION(m)) return 0;

-      if ((((size_t)(mem)) % alignment) != 0) { /* misaligned */

-        /*

-          Find an aligned spot inside chunk.  Since we need to give

-          back leading space in a chunk of at least MIN_CHUNK_SIZE, if

-          the first calculation places us at a spot with less than

-          MIN_CHUNK_SIZE leader, we can move to the next aligned spot.

-          We've allocated enough total room so that this is always

-          possible.

-        */

-        char* br = (char*)mem2chunk((size_t)(((size_t)(mem +

-                                                       alignment -

-                                                       SIZE_T_ONE)) &

-                                             -alignment));

-        char* pos = ((size_t)(br - (char*)(p)) >= MIN_CHUNK_SIZE)?

-          br : br+alignment;

-        mchunkptr newp = (mchunkptr)pos;

-        size_t leadsize = pos - (char*)(p);

-        size_t newsize = chunksize(p) - leadsize;

-

-        if (is_mmapped(p)) { /* For mmapped chunks, just adjust offset */

-          newp->prev_foot = p->prev_foot + leadsize;

-          newp->head = newsize;

-        }

-        else { /* Otherwise, give back leader, use the rest */

-          set_inuse(m, newp, newsize);

-          set_inuse(m, p, leadsize);

-          leader = chunk2mem(p);

-        }

-        p = newp;

-      }

-

-      /* Give back spare room at the end */

-      if (!is_mmapped(p)) {

-        size_t size = chunksize(p);

-        if (size > nb + MIN_CHUNK_SIZE) {

-          size_t remainder_size = size - nb;

-          mchunkptr remainder = chunk_plus_offset(p, nb);

-          set_inuse(m, p, nb);

-          set_inuse(m, remainder, remainder_size);

-          trailer = chunk2mem(remainder);

-        }

-      }

-

-      assert (chunksize(p) >= nb);

-      assert((((size_t)(chunk2mem(p))) % alignment) == 0);

-      check_inuse_chunk(m, p);

-      POSTACTION(m);

-      if (leader != 0) {

-        internal_free(m, leader);

-      }

-      if (trailer != 0) {

-        internal_free(m, trailer);

-      }

-      return chunk2mem(p);

-    }

-  }

-  return 0;

-}

-

-/* ------------------------ comalloc/coalloc support --------------------- */

-

-static void** ialloc(mstate m,

-                     size_t n_elements,

-                     size_t* sizes,

-                     int opts,

-                     void* chunks[]) {

-  /*

-    This provides common support for independent_X routines, handling

-    all of the combinations that can result.

-

-    The opts arg has:

-    bit 0 set if all elements are same size (using sizes[0])

-    bit 1 set if elements should be zeroed

-  */

-

-  size_t    element_size;   /* chunksize of each element, if all same */

-  size_t    contents_size;  /* total size of elements */

-  size_t    array_size;     /* request size of pointer array */

-  void*     mem;            /* malloced aggregate space */

-  mchunkptr p;              /* corresponding chunk */

-  size_t    remainder_size; /* remaining bytes while splitting */

-  void**    marray;         /* either "chunks" or malloced ptr array */

-  mchunkptr array_chunk;    /* chunk for malloced ptr array */

-  flag_t    was_enabled;    /* to disable mmap */

-  size_t    size;

-  size_t    i;

-

-  ensure_initialization();

-  /* compute array length, if needed */

-  if (chunks != 0) {

-    if (n_elements == 0)

-      return chunks; /* nothing to do */

-    marray = chunks;

-    array_size = 0;

-  }

-  else {

-    /* if empty req, must still return chunk representing empty array */

-    if (n_elements == 0)

-      return (void**)internal_malloc(m, 0);

-    marray = 0;

-    array_size = request2size(n_elements * (sizeof(void*)));

-  }

-

-  /* compute total element size */

-  if (opts & 0x1) { /* all-same-size */

-    element_size = request2size(*sizes);

-    contents_size = n_elements * element_size;

-  }

-  else { /* add up all the sizes */

-    element_size = 0;

-    contents_size = 0;

-    for (i = 0; i != n_elements; ++i)

-      contents_size += request2size(sizes[i]);

-  }

-

-  size = contents_size + array_size;

-

-  /*

-     Allocate the aggregate chunk.  First disable direct-mmapping so

-     malloc won't use it, since we would not be able to later

-     free/realloc space internal to a segregated mmap region.

-  */

-  was_enabled = use_mmap(m);

-  disable_mmap(m);

-  mem = internal_malloc(m, size - CHUNK_OVERHEAD);

-  if (was_enabled)

-    enable_mmap(m);

-  if (mem == 0)

-    return 0;

-

-  if (PREACTION(m)) return 0;

-  p = mem2chunk(mem);

-  remainder_size = chunksize(p);

-

-  assert(!is_mmapped(p));

-

-  if (opts & 0x2) {       /* optionally clear the elements */

-    memset((size_t*)mem, 0, remainder_size - SIZE_T_SIZE - array_size);

-  }

-

-  /* If not provided, allocate the pointer array as final part of chunk */

-  if (marray == 0) {

-    size_t  array_chunk_size;

-    array_chunk = chunk_plus_offset(p, contents_size);

-    array_chunk_size = remainder_size - contents_size;

-    marray = (void**) (chunk2mem(array_chunk));

-    set_size_and_pinuse_of_inuse_chunk(m, array_chunk, array_chunk_size);

-    remainder_size = contents_size;

-  }

-

-  /* split out elements */

-  for (i = 0; ; ++i) {

-    marray[i] = chunk2mem(p);

-    if (i != n_elements-1) {

-      if (element_size != 0)

-        size = element_size;

-      else

-        size = request2size(sizes[i]);

-      remainder_size -= size;

-      set_size_and_pinuse_of_inuse_chunk(m, p, size);

-      p = chunk_plus_offset(p, size);

-    }

-    else { /* the final element absorbs any overallocation slop */

-      set_size_and_pinuse_of_inuse_chunk(m, p, remainder_size);

-      break;

-    }

-  }

-

-#if DEBUG

-  if (marray != chunks) {

-    /* final element must have exactly exhausted chunk */

-    if (element_size != 0) {

-      assert(remainder_size == element_size);

-    }

-    else {

-      assert(remainder_size == request2size(sizes[i]));

-    }

-    check_inuse_chunk(m, mem2chunk(marray));

-  }

-  for (i = 0; i != n_elements; ++i)

-    check_inuse_chunk(m, mem2chunk(marray[i]));

-

-#endif /* DEBUG */

-

-  POSTACTION(m);

-  return marray;

-}

-

-

-/* -------------------------- public routines ---------------------------- */

-

-#if !ONLY_MSPACES

-

-void* dlmalloc(size_t bytes) {

-  /*

-     Basic algorithm:

-     If a small request (< 256 bytes minus per-chunk overhead):

-       1. If one exists, use a remainderless chunk in associated smallbin.

-          (Remainderless means that there are too few excess bytes to

-          represent as a chunk.)

-       2. If it is big enough, use the dv chunk, which is normally the

-          chunk adjacent to the one used for the most recent small request.

-       3. If one exists, split the smallest available chunk in a bin,

-          saving remainder in dv.

-       4. If it is big enough, use the top chunk.

-       5. If available, get memory from system and use it

-     Otherwise, for a large request:

-       1. Find the smallest available binned chunk that fits, and use it

-          if it is better fitting than dv chunk, splitting if necessary.

-       2. If better fitting than any binned chunk, use the dv chunk.

-       3. If it is big enough, use the top chunk.

-       4. If request size >= mmap threshold, try to directly mmap this chunk.

-       5. If available, get memory from system and use it

-

-     The ugly goto's here ensure that postaction occurs along all paths.

-  */

-

-#if USE_LOCKS

-  ensure_initialization(); /* initialize in sys_alloc if not using locks */

-#endif

-

-  if (!PREACTION(gm)) {

-    void* mem;

-    size_t nb;

-    if (bytes <= MAX_SMALL_REQUEST) {

-      bindex_t idx;

-      binmap_t smallbits;

-      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);

-      idx = small_index(nb);

-      smallbits = gm->smallmap >> idx;

-

-      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */

-        mchunkptr b, p;

-        idx += ~smallbits & 1;       /* Uses next bin if idx empty */

-        b = smallbin_at(gm, idx);

-        p = b->fd;

-        assert(chunksize(p) == small_index2size(idx));

-        unlink_first_small_chunk(gm, b, p, idx);

-        set_inuse_and_pinuse(gm, p, small_index2size(idx));

-        mem = chunk2mem(p);

-        check_malloced_chunk(gm, mem, nb);

-        goto postaction;

-      }

-

-      else if (nb > gm->dvsize) {

-        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */

-          mchunkptr b, p, r;

-          size_t rsize;

-          bindex_t i;

-          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));

-          binmap_t leastbit = least_bit(leftbits);

-          compute_bit2idx(leastbit, i);

-          b = smallbin_at(gm, i);

-          p = b->fd;

-          assert(chunksize(p) == small_index2size(i));

-          unlink_first_small_chunk(gm, b, p, i);

-          rsize = small_index2size(i) - nb;

-          /* Fit here cannot be remainderless if 4byte sizes */

-          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)

-            set_inuse_and_pinuse(gm, p, small_index2size(i));

-          else {

-            set_size_and_pinuse_of_inuse_chunk(gm, p, nb);

-            r = chunk_plus_offset(p, nb);

-            set_size_and_pinuse_of_free_chunk(r, rsize);

-            replace_dv(gm, r, rsize);

-          }

-          mem = chunk2mem(p);

-          check_malloced_chunk(gm, mem, nb);

-          goto postaction;

-        }

-

-        else if (gm->treemap != 0 && (mem = tmalloc_small(gm, nb)) != 0) {

-          check_malloced_chunk(gm, mem, nb);

-          goto postaction;

-        }

-      }

-    }

-    else if (bytes >= MAX_REQUEST)

-      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */

-    else {

-      nb = pad_request(bytes);

-      if (gm->treemap != 0 && (mem = tmalloc_large(gm, nb)) != 0) {

-        check_malloced_chunk(gm, mem, nb);

-        goto postaction;

-      }

-    }

-

-    if (nb <= gm->dvsize) {

-      size_t rsize = gm->dvsize - nb;

-      mchunkptr p = gm->dv;

-      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */

-        mchunkptr r = gm->dv = chunk_plus_offset(p, nb);

-        gm->dvsize = rsize;

-        set_size_and_pinuse_of_free_chunk(r, rsize);

-        set_size_and_pinuse_of_inuse_chunk(gm, p, nb);

-      }

-      else { /* exhaust dv */

-        size_t dvs = gm->dvsize;

-        gm->dvsize = 0;

-        gm->dv = 0;

-        set_inuse_and_pinuse(gm, p, dvs);

-      }

-      mem = chunk2mem(p);

-      check_malloced_chunk(gm, mem, nb);

-      goto postaction;

-    }

-

-    else if (nb < gm->topsize) { /* Split top */

-      size_t rsize = gm->topsize -= nb;

-      mchunkptr p = gm->top;

-      mchunkptr r = gm->top = chunk_plus_offset(p, nb);

-      r->head = rsize | PINUSE_BIT;

-      set_size_and_pinuse_of_inuse_chunk(gm, p, nb);

-      mem = chunk2mem(p);

-      check_top_chunk(gm, gm->top);

-      check_malloced_chunk(gm, mem, nb);

-      goto postaction;

-    }

-

-    mem = sys_alloc(gm, nb);

-

-  postaction:

-    POSTACTION(gm);

-    return mem;

-  }

-

-  return 0;

-}

-

-void dlfree(void* mem) {

-  /*

-     Consolidate freed chunks with preceeding or succeeding bordering

-     free chunks, if they exist, and then place in a bin.  Intermixed

-     with special cases for top, dv, mmapped chunks, and usage errors.

-  */

-

-  if (mem != 0) {

-    mchunkptr p  = mem2chunk(mem);

-#if FOOTERS

-    mstate fm = get_mstate_for(p);

-    if (!ok_magic(fm)) {

-      USAGE_ERROR_ACTION(fm, p);

-      return;

-    }

-#else /* FOOTERS */

-#define fm gm

-#endif /* FOOTERS */

-    if (!PREACTION(fm)) {

-      check_inuse_chunk(fm, p);

-      if (RTCHECK(ok_address(fm, p) && ok_inuse(p))) {

-        size_t psize = chunksize(p);

-        mchunkptr next = chunk_plus_offset(p, psize);

-        if (!pinuse(p)) {

-          size_t prevsize = p->prev_foot;

-          if (is_mmapped(p)) {

-            psize += prevsize + MMAP_FOOT_PAD;

-            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)

-              fm->footprint -= psize;

-            goto postaction;

-          }

-          else {

-            mchunkptr prev = chunk_minus_offset(p, prevsize);

-            psize += prevsize;

-            p = prev;

-            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */

-              if (p != fm->dv) {

-                unlink_chunk(fm, p, prevsize);

-              }

-              else if ((next->head & INUSE_BITS) == INUSE_BITS) {

-                fm->dvsize = psize;

-                set_free_with_pinuse(p, psize, next);

-                goto postaction;

-              }

-            }

-            else

-              goto erroraction;

-          }

-        }

-

-        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {

-          if (!cinuse(next)) {  /* consolidate forward */

-            if (next == fm->top) {

-              size_t tsize = fm->topsize += psize;

-              fm->top = p;

-              p->head = tsize | PINUSE_BIT;

-              if (p == fm->dv) {

-                fm->dv = 0;

-                fm->dvsize = 0;

-              }

-              if (should_trim(fm, tsize))

-                sys_trim(fm, 0);

-              goto postaction;

-            }

-            else if (next == fm->dv) {

-              size_t dsize = fm->dvsize += psize;

-              fm->dv = p;

-              set_size_and_pinuse_of_free_chunk(p, dsize);

-              goto postaction;

-            }

-            else {

-              size_t nsize = chunksize(next);

-              psize += nsize;

-              unlink_chunk(fm, next, nsize);

-              set_size_and_pinuse_of_free_chunk(p, psize);

-              if (p == fm->dv) {

-                fm->dvsize = psize;

-                goto postaction;

-              }

-            }

-          }

-          else

-            set_free_with_pinuse(p, psize, next);

-

-          if (is_small(psize)) {

-            insert_small_chunk(fm, p, psize);

-            check_free_chunk(fm, p);

-          }

-          else {

-            tchunkptr tp = (tchunkptr)p;

-            insert_large_chunk(fm, tp, psize);

-            check_free_chunk(fm, p);

-            if (--fm->release_checks == 0)

-              release_unused_segments(fm);

-          }

-          goto postaction;

-        }

-      }

-    erroraction:

-      USAGE_ERROR_ACTION(fm, p);

-    postaction:

-      POSTACTION(fm);

-    }

-  }

-#if !FOOTERS

-#undef fm

-#endif /* FOOTERS */

-}

-

-void* dlcalloc(size_t n_elements, size_t elem_size) {

-  void* mem;

-  size_t req = 0;

-  if (n_elements != 0) {

-    req = n_elements * elem_size;

-    if (((n_elements | elem_size) & ~(size_t)0xffff) &&

-        (req / n_elements != elem_size))

-      req = MAX_SIZE_T; /* force downstream failure on overflow */

-  }

-  mem = dlmalloc(req);

-  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))

-    memset(mem, 0, req);

-  return mem;

-}

-

-void* dlrealloc(void* oldmem, size_t bytes) {

-  if (oldmem == 0)

-    return dlmalloc(bytes);

-#ifdef REALLOC_ZERO_BYTES_FREES

-  if (bytes == 0) {

-    dlfree(oldmem);

-    return 0;

-  }

-#endif /* REALLOC_ZERO_BYTES_FREES */

-  else {

-#if ! FOOTERS

-    mstate m = gm;

-#else /* FOOTERS */

-    mstate m = get_mstate_for(mem2chunk(oldmem));

-    if (!ok_magic(m)) {

-      USAGE_ERROR_ACTION(m, oldmem);

-      return 0;

-    }

-#endif /* FOOTERS */

-    return internal_realloc(m, oldmem, bytes);

-  }

-}

-

-void* dlmemalign(size_t alignment, size_t bytes) {

-  return internal_memalign(gm, alignment, bytes);

-}

-

-void** dlindependent_calloc(size_t n_elements, size_t elem_size,

-                                 void* chunks[]) {

-  size_t sz = elem_size; /* serves as 1-element array */

-  return ialloc(gm, n_elements, &sz, 3, chunks);

-}

-

-void** dlindependent_comalloc(size_t n_elements, size_t sizes[],

-                                   void* chunks[]) {

-  return ialloc(gm, n_elements, sizes, 0, chunks);

-}

-

-void* dlvalloc(size_t bytes) {

-  size_t pagesz;

-  ensure_initialization();

-  pagesz = mparams.page_size;

-  return dlmemalign(pagesz, bytes);

-}

-

-void* dlpvalloc(size_t bytes) {

-  size_t pagesz;

-  ensure_initialization();

-  pagesz = mparams.page_size;

-  return dlmemalign(pagesz, (bytes + pagesz - SIZE_T_ONE) & ~(pagesz - SIZE_T_ONE));

-}

-

-int dlmalloc_trim(size_t pad) {

-  int result = 0;

-  ensure_initialization();

-  if (!PREACTION(gm)) {

-    result = sys_trim(gm, pad);

-    POSTACTION(gm);

-  }

-  return result;

-}

-

-size_t dlmalloc_footprint(void) {

-  return gm->footprint;

-}

-

-size_t dlmalloc_max_footprint(void) {

-  return gm->max_footprint;

-}

-

-#if !NO_MALLINFO

-struct mallinfo dlmallinfo(void) {

-  return internal_mallinfo(gm);

-}

-#endif /* NO_MALLINFO */

-

-void dlmalloc_stats() {

-  internal_malloc_stats(gm);

-}

-

-int dlmallopt(int param_number, int value) {

-  return change_mparam(param_number, value);

-}

-

-#endif /* !ONLY_MSPACES */

-

-size_t dlmalloc_usable_size(void* mem) {

-  if (mem != 0) {

-    mchunkptr p = mem2chunk(mem);

-    if (is_inuse(p))

-      return chunksize(p) - overhead_for(p);

-  }

-  return 0;

-}

-

-/* ----------------------------- user mspaces ---------------------------- */

-

-#if MSPACES

-

-static mstate init_user_mstate(char* tbase, size_t tsize) {

-  size_t msize = pad_request(sizeof(struct malloc_state));

-  mchunkptr mn;

-  mchunkptr msp = align_as_chunk(tbase);

-  mstate m = (mstate)(chunk2mem(msp));

-  memset(m, 0, msize);

-  INITIAL_LOCK(&m->mutex);

-  msp->head = (msize|INUSE_BITS);

-  m->seg.base = m->least_addr = tbase;

-  m->seg.size = m->footprint = m->max_footprint = tsize;

-  m->magic = mparams.magic;

-  m->release_checks = MAX_RELEASE_CHECK_RATE;

-  m->mflags = mparams.default_mflags;

-  m->extp = 0;

-  m->exts = 0;

-  disable_contiguous(m);

-  init_bins(m);

-  mn = next_chunk(mem2chunk(m));

-  init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) - TOP_FOOT_SIZE);

-  check_top_chunk(m, m->top);

-  return m;

-}

-

-mspace create_mspace(size_t capacity, int locked) {

-  mstate m = 0;

-  size_t msize;

-  ensure_initialization();

-  msize = pad_request(sizeof(struct malloc_state));

-  if (capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {

-    size_t rs = ((capacity == 0)? mparams.granularity :

-                 (capacity + TOP_FOOT_SIZE + msize));

-    size_t tsize = granularity_align(rs);

-    char* tbase = (char*)(CALL_MMAP(tsize));

-    if (tbase != CMFAIL) {

-      m = init_user_mstate(tbase, tsize);

-      m->seg.sflags = USE_MMAP_BIT;

-      set_lock(m, locked);

-    }

-  }

-  return (mspace)m;

-}

-

-mspace create_mspace_with_base(void* base, size_t capacity, int locked) {

-  mstate m = 0;

-  size_t msize;

-  ensure_initialization();

-  msize = pad_request(sizeof(struct malloc_state));

-  if (capacity > msize + TOP_FOOT_SIZE &&

-      capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {

-    m = init_user_mstate((char*)base, capacity);

-    m->seg.sflags = EXTERN_BIT;

-    set_lock(m, locked);

-  }

-  return (mspace)m;

-}

-

-int mspace_track_large_chunks(mspace msp, int enable) {

-  int ret = 0;

-  mstate ms = (mstate)msp;

-  if (!PREACTION(ms)) {

-    if (!use_mmap(ms))

-      ret = 1;

-    if (!enable)

-      enable_mmap(ms);

-    else

-      disable_mmap(ms);

-    POSTACTION(ms);

-  }

-  return ret;

-}

-

-size_t destroy_mspace(mspace msp) {

-  size_t freed = 0;

-  mstate ms = (mstate)msp;

-  if (ok_magic(ms)) {

-    msegmentptr sp = &ms->seg;

-    while (sp != 0) {

-      char* base = sp->base;

-      size_t size = sp->size;

-      flag_t flag = sp->sflags;

-      sp = sp->next;

-      if ((flag & USE_MMAP_BIT) && !(flag & EXTERN_BIT) &&

-          CALL_MUNMAP(base, size) == 0)

-        freed += size;

-    }

-  }

-  else {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-  return freed;

-}

-

-/*

-  mspace versions of routines are near-clones of the global

-  versions. This is not so nice but better than the alternatives.

-*/

-

-

-void* mspace_malloc(mspace msp, size_t bytes) {

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-    return 0;

-  }

-  if (!PREACTION(ms)) {

-    void* mem;

-    size_t nb;

-    if (bytes <= MAX_SMALL_REQUEST) {

-      bindex_t idx;

-      binmap_t smallbits;

-      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);

-      idx = small_index(nb);

-      smallbits = ms->smallmap >> idx;

-

-      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */

-        mchunkptr b, p;

-        idx += ~smallbits & 1;       /* Uses next bin if idx empty */

-        b = smallbin_at(ms, idx);

-        p = b->fd;

-        assert(chunksize(p) == small_index2size(idx));

-        unlink_first_small_chunk(ms, b, p, idx);

-        set_inuse_and_pinuse(ms, p, small_index2size(idx));

-        mem = chunk2mem(p);

-        check_malloced_chunk(ms, mem, nb);

-        goto postaction;

-      }

-

-      else if (nb > ms->dvsize) {

-        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */

-          mchunkptr b, p, r;

-          size_t rsize;

-          bindex_t i;

-          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));

-          binmap_t leastbit = least_bit(leftbits);

-          compute_bit2idx(leastbit, i);

-          b = smallbin_at(ms, i);

-          p = b->fd;

-          assert(chunksize(p) == small_index2size(i));

-          unlink_first_small_chunk(ms, b, p, i);

-          rsize = small_index2size(i) - nb;

-          /* Fit here cannot be remainderless if 4byte sizes */

-          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)

-            set_inuse_and_pinuse(ms, p, small_index2size(i));

-          else {

-            set_size_and_pinuse_of_inuse_chunk(ms, p, nb);

-            r = chunk_plus_offset(p, nb);

-            set_size_and_pinuse_of_free_chunk(r, rsize);

-            replace_dv(ms, r, rsize);

-          }

-          mem = chunk2mem(p);

-          check_malloced_chunk(ms, mem, nb);

-          goto postaction;

-        }

-

-        else if (ms->treemap != 0 && (mem = tmalloc_small(ms, nb)) != 0) {

-          check_malloced_chunk(ms, mem, nb);

-          goto postaction;

-        }

-      }

-    }

-    else if (bytes >= MAX_REQUEST)

-      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */

-    else {

-      nb = pad_request(bytes);

-      if (ms->treemap != 0 && (mem = tmalloc_large(ms, nb)) != 0) {

-        check_malloced_chunk(ms, mem, nb);

-        goto postaction;

-      }

-    }

-

-    if (nb <= ms->dvsize) {

-      size_t rsize = ms->dvsize - nb;

-      mchunkptr p = ms->dv;

-      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */

-        mchunkptr r = ms->dv = chunk_plus_offset(p, nb);

-        ms->dvsize = rsize;

-        set_size_and_pinuse_of_free_chunk(r, rsize);

-        set_size_and_pinuse_of_inuse_chunk(ms, p, nb);

-      }

-      else { /* exhaust dv */

-        size_t dvs = ms->dvsize;

-        ms->dvsize = 0;

-        ms->dv = 0;

-        set_inuse_and_pinuse(ms, p, dvs);

-      }

-      mem = chunk2mem(p);

-      check_malloced_chunk(ms, mem, nb);

-      goto postaction;

-    }

-

-    else if (nb < ms->topsize) { /* Split top */

-      size_t rsize = ms->topsize -= nb;

-      mchunkptr p = ms->top;

-      mchunkptr r = ms->top = chunk_plus_offset(p, nb);

-      r->head = rsize | PINUSE_BIT;

-      set_size_and_pinuse_of_inuse_chunk(ms, p, nb);

-      mem = chunk2mem(p);

-      check_top_chunk(ms, ms->top);

-      check_malloced_chunk(ms, mem, nb);

-      goto postaction;

-    }

-

-    mem = sys_alloc(ms, nb);

-

-  postaction:

-    POSTACTION(ms);

-    return mem;

-  }

-

-  return 0;

-}

-

-void mspace_free(mspace msp, void* mem) {

-  if (mem != 0) {

-    mchunkptr p  = mem2chunk(mem);

-#if FOOTERS

-    mstate fm = get_mstate_for(p);

-    msp = msp; /* placate people compiling -Wunused */

-#else /* FOOTERS */

-    mstate fm = (mstate)msp;

-#endif /* FOOTERS */

-    if (!ok_magic(fm)) {

-      USAGE_ERROR_ACTION(fm, p);

-      return;

-    }

-    if (!PREACTION(fm)) {

-      check_inuse_chunk(fm, p);

-      if (RTCHECK(ok_address(fm, p) && ok_inuse(p))) {

-        size_t psize = chunksize(p);

-        mchunkptr next = chunk_plus_offset(p, psize);

-        if (!pinuse(p)) {

-          size_t prevsize = p->prev_foot;

-          if (is_mmapped(p)) {

-            psize += prevsize + MMAP_FOOT_PAD;

-            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)

-              fm->footprint -= psize;

-            goto postaction;

-          }

-          else {

-            mchunkptr prev = chunk_minus_offset(p, prevsize);

-            psize += prevsize;

-            p = prev;

-            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */

-              if (p != fm->dv) {

-                unlink_chunk(fm, p, prevsize);

-              }

-              else if ((next->head & INUSE_BITS) == INUSE_BITS) {

-                fm->dvsize = psize;

-                set_free_with_pinuse(p, psize, next);

-                goto postaction;

-              }

-            }

-            else

-              goto erroraction;

-          }

-        }

-

-        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {

-          if (!cinuse(next)) {  /* consolidate forward */

-            if (next == fm->top) {

-              size_t tsize = fm->topsize += psize;

-              fm->top = p;

-              p->head = tsize | PINUSE_BIT;

-              if (p == fm->dv) {

-                fm->dv = 0;

-                fm->dvsize = 0;

-              }

-              if (should_trim(fm, tsize))

-                sys_trim(fm, 0);

-              goto postaction;

-            }

-            else if (next == fm->dv) {

-              size_t dsize = fm->dvsize += psize;

-              fm->dv = p;

-              set_size_and_pinuse_of_free_chunk(p, dsize);

-              goto postaction;

-            }

-            else {

-              size_t nsize = chunksize(next);

-              psize += nsize;

-              unlink_chunk(fm, next, nsize);

-              set_size_and_pinuse_of_free_chunk(p, psize);

-              if (p == fm->dv) {

-                fm->dvsize = psize;

-                goto postaction;

-              }

-            }

-          }

-          else

-            set_free_with_pinuse(p, psize, next);

-

-          if (is_small(psize)) {

-            insert_small_chunk(fm, p, psize);

-            check_free_chunk(fm, p);

-          }

-          else {

-            tchunkptr tp = (tchunkptr)p;

-            insert_large_chunk(fm, tp, psize);

-            check_free_chunk(fm, p);

-            if (--fm->release_checks == 0)

-              release_unused_segments(fm);

-          }

-          goto postaction;

-        }

-      }

-    erroraction:

-      USAGE_ERROR_ACTION(fm, p);

-    postaction:

-      POSTACTION(fm);

-    }

-  }

-}

-

-void* mspace_calloc(mspace msp, size_t n_elements, size_t elem_size) {

-  void* mem;

-  size_t req = 0;

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-    return 0;

-  }

-  if (n_elements != 0) {

-    req = n_elements * elem_size;

-    if (((n_elements | elem_size) & ~(size_t)0xffff) &&

-        (req / n_elements != elem_size))

-      req = MAX_SIZE_T; /* force downstream failure on overflow */

-  }

-  mem = internal_malloc(ms, req);

-  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))

-    memset(mem, 0, req);

-  return mem;

-}

-

-void* mspace_realloc(mspace msp, void* oldmem, size_t bytes) {

-  if (oldmem == 0)

-    return mspace_malloc(msp, bytes);

-#ifdef REALLOC_ZERO_BYTES_FREES

-  if (bytes == 0) {

-    mspace_free(msp, oldmem);

-    return 0;

-  }

-#endif /* REALLOC_ZERO_BYTES_FREES */

-  else {

-#if FOOTERS

-    mchunkptr p  = mem2chunk(oldmem);

-    mstate ms = get_mstate_for(p);

-#else /* FOOTERS */

-    mstate ms = (mstate)msp;

-#endif /* FOOTERS */

-    if (!ok_magic(ms)) {

-      USAGE_ERROR_ACTION(ms,ms);

-      return 0;

-    }

-    return internal_realloc(ms, oldmem, bytes);

-  }

-}

-

-void* mspace_memalign(mspace msp, size_t alignment, size_t bytes) {

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-    return 0;

-  }

-  return internal_memalign(ms, alignment, bytes);

-}

-

-void** mspace_independent_calloc(mspace msp, size_t n_elements,

-                                 size_t elem_size, void* chunks[]) {

-  size_t sz = elem_size; /* serves as 1-element array */

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-    return 0;

-  }

-  return ialloc(ms, n_elements, &sz, 3, chunks);

-}

-

-void** mspace_independent_comalloc(mspace msp, size_t n_elements,

-                                   size_t sizes[], void* chunks[]) {

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-    return 0;

-  }

-  return ialloc(ms, n_elements, sizes, 0, chunks);

-}

-

-int mspace_trim(mspace msp, size_t pad) {

-  int result = 0;

-  mstate ms = (mstate)msp;

-  if (ok_magic(ms)) {

-    if (!PREACTION(ms)) {

-      result = sys_trim(ms, pad);

-      POSTACTION(ms);

-    }

-  }

-  else {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-  return result;

-}

-

-void mspace_malloc_stats(mspace msp) {

-  mstate ms = (mstate)msp;

-  if (ok_magic(ms)) {

-    internal_malloc_stats(ms);

-  }

-  else {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-}

-

-size_t mspace_footprint(mspace msp) {

-  size_t result = 0;

-  mstate ms = (mstate)msp;

-  if (ok_magic(ms)) {

-    result = ms->footprint;

-  }

-  else {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-  return result;

-}

-

-

-size_t mspace_max_footprint(mspace msp) {

-  size_t result = 0;

-  mstate ms = (mstate)msp;

-  if (ok_magic(ms)) {

-    result = ms->max_footprint;

-  }

-  else {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-  return result;

-}

-

-

-#if !NO_MALLINFO

-struct mallinfo mspace_mallinfo(mspace msp) {

-  mstate ms = (mstate)msp;

-  if (!ok_magic(ms)) {

-    USAGE_ERROR_ACTION(ms,ms);

-  }

-  return internal_mallinfo(ms);

-}

-#endif /* NO_MALLINFO */

-

-size_t mspace_usable_size(void* mem) {

-  if (mem != 0) {

-    mchunkptr p = mem2chunk(mem);

-    if (is_inuse(p))

-      return chunksize(p) - overhead_for(p);

-  }

-  return 0;

-}

-

-int mspace_mallopt(int param_number, int value) {

-  return change_mparam(param_number, value);

-}

-

-#endif /* MSPACES */

-

-

-/* -------------------- Alternative MORECORE functions ------------------- */

-

-/*

-  Guidelines for creating a custom version of MORECORE:

-

-  * For best performance, MORECORE should allocate in multiples of pagesize.

-  * MORECORE may allocate more memory than requested. (Or even less,

-      but this will usually result in a malloc failure.)

-  * MORECORE must not allocate memory when given argument zero, but

-      instead return one past the end address of memory from previous

-      nonzero call.

-  * For best performance, consecutive calls to MORECORE with positive

-      arguments should return increasing addresses, indicating that

-      space has been contiguously extended.

-  * Even though consecutive calls to MORECORE need not return contiguous

-      addresses, it must be OK for malloc'ed chunks to span multiple

-      regions in those cases where they do happen to be contiguous.

-  * MORECORE need not handle negative arguments -- it may instead

-      just return MFAIL when given negative arguments.

-      Negative arguments are always multiples of pagesize. MORECORE

-      must not misinterpret negative args as large positive unsigned

-      args. You can suppress all such calls from even occurring by defining

-      MORECORE_CANNOT_TRIM,

-

-  As an example alternative MORECORE, here is a custom allocator

-  kindly contributed for pre-OSX macOS.  It uses virtually but not

-  necessarily physically contiguous non-paged memory (locked in,

-  present and won't get swapped out).  You can use it by uncommenting

-  this section, adding some #includes, and setting up the appropriate

-  defines above:

-

-      #define MORECORE osMoreCore

-

-  There is also a shutdown routine that should somehow be called for

-  cleanup upon program exit.

-

-  #define MAX_POOL_ENTRIES 100

-  #define MINIMUM_MORECORE_SIZE  (64 * 1024U)

-  static int next_os_pool;

-  void *our_os_pools[MAX_POOL_ENTRIES];

-

-  void *osMoreCore(int size)

-  {

-    void *ptr = 0;

-    static void *sbrk_top = 0;

-

-    if (size > 0)

-    {

-      if (size < MINIMUM_MORECORE_SIZE)

-         size = MINIMUM_MORECORE_SIZE;

-      if (CurrentExecutionLevel() == kTaskLevel)

-         ptr = PoolAllocateResident(size + RM_PAGE_SIZE, 0);

-      if (ptr == 0)

-      {

-        return (void *) MFAIL;

-      }

-      // save ptrs so they can be freed during cleanup

-      our_os_pools[next_os_pool] = ptr;

-      next_os_pool++;

-      ptr = (void *) ((((size_t) ptr) + RM_PAGE_MASK) & ~RM_PAGE_MASK);

-      sbrk_top = (char *) ptr + size;

-      return ptr;

-    }

-    else if (size < 0)

-    {

-      // we don't currently support shrink behavior

-      return (void *) MFAIL;

-    }

-    else

-    {

-      return sbrk_top;

-    }

-  }

-

-  // cleanup any allocated memory pools

-  // called as last thing before shutting down driver

-

-  void osCleanupMem(void)

-  {

-    void **ptr;

-

-    for (ptr = our_os_pools; ptr < &our_os_pools[MAX_POOL_ENTRIES]; ptr++)

-      if (*ptr)

-      {

-         PoolDeallocate(*ptr);

-         *ptr = 0;

-      }

-  }

-

-*/

-

-

-/* -----------------------------------------------------------------------

-History:

-    V2.8.4 Wed May 27 09:56:23 2009  Doug Lea  (dl at gee)

-      * Use zeros instead of prev foot for is_mmapped

-      * Add mspace_track_large_chunks; thanks to Jean Brouwers

-      * Fix set_inuse in internal_realloc; thanks to Jean Brouwers

-      * Fix insufficient sys_alloc padding when using 16byte alignment

-      * Fix bad error check in mspace_footprint

-      * Adaptations for ptmalloc; thanks to Wolfram Gloger.

-      * Reentrant spin locks; thanks to Earl Chew and others

-      * Win32 improvements; thanks to Niall Douglas and Earl Chew

-      * Add NO_SEGMENT_TRAVERSAL and MAX_RELEASE_CHECK_RATE options

-      * Extension hook in malloc_state

-      * Various small adjustments to reduce warnings on some compilers

-      * Various configuration extensions/changes for more platforms. Thanks

-         to all who contributed these.

-

-    V2.8.3 Thu Sep 22 11:16:32 2005  Doug Lea  (dl at gee)

-      * Add max_footprint functions

-      * Ensure all appropriate literals are size_t

-      * Fix conditional compilation problem for some #define settings

-      * Avoid concatenating segments with the one provided

-        in create_mspace_with_base

-      * Rename some variables to avoid compiler shadowing warnings

-      * Use explicit lock initialization.

-      * Better handling of sbrk interference.

-      * Simplify and fix segment insertion, trimming and mspace_destroy

-      * Reinstate REALLOC_ZERO_BYTES_FREES option from 2.7.x

-      * Thanks especially to Dennis Flanagan for help on these.

-

-    V2.8.2 Sun Jun 12 16:01:10 2005  Doug Lea  (dl at gee)

-      * Fix memalign brace error.

-

-    V2.8.1 Wed Jun  8 16:11:46 2005  Doug Lea  (dl at gee)

-      * Fix improper #endif nesting in C++

-      * Add explicit casts needed for C++

-

-    V2.8.0 Mon May 30 14:09:02 2005  Doug Lea  (dl at gee)

-      * Use trees for large bins

-      * Support mspaces

-      * Use segments to unify sbrk-based and mmap-based system allocation,

-        removing need for emulation on most platforms without sbrk.

-      * Default safety checks

-      * Optional footer checks. Thanks to William Robertson for the idea.

-      * Internal code refactoring

-      * Incorporate suggestions and platform-specific changes.

-        Thanks to Dennis Flanagan, Colin Plumb, Niall Douglas,

-        Aaron Bachmann,  Emery Berger, and others.

-      * Speed up non-fastbin processing enough to remove fastbins.

-      * Remove useless cfree() to avoid conflicts with other apps.

-      * Remove internal memcpy, memset. Compilers handle builtins better.

-      * Remove some options that no one ever used and rename others.

-

-    V2.7.2 Sat Aug 17 09:07:30 2002  Doug Lea  (dl at gee)

-      * Fix malloc_state bitmap array misdeclaration

-

-    V2.7.1 Thu Jul 25 10:58:03 2002  Doug Lea  (dl at gee)

-      * Allow tuning of FIRST_SORTED_BIN_SIZE

-      * Use PTR_UINT as type for all ptr->int casts. Thanks to John Belmonte.

-      * Better detection and support for non-contiguousness of MORECORE.

-        Thanks to Andreas Mueller, Conal Walsh, and Wolfram Gloger

-      * Bypass most of malloc if no frees. Thanks To Emery Berger.

-      * Fix freeing of old top non-contiguous chunk im sysmalloc.

-      * Raised default trim and map thresholds to 256K.

-      * Fix mmap-related #defines. Thanks to Lubos Lunak.

-      * Fix copy macros; added LACKS_FCNTL_H. Thanks to Neal Walfield.

-      * Branch-free bin calculation

-      * Default trim and mmap thresholds now 256K.

-

-    V2.7.0 Sun Mar 11 14:14:06 2001  Doug Lea  (dl at gee)

-      * Introduce independent_comalloc and independent_calloc.

-        Thanks to Michael Pachos for motivation and help.

-      * Make optional .h file available

-      * Allow > 2GB requests on 32bit systems.

-      * new WIN32 sbrk, mmap, munmap, lock code from <Walter@GeNeSys-e.de>.

-        Thanks also to Andreas Mueller <a.mueller at paradatec.de>,

-        and Anonymous.

-      * Allow override of MALLOC_ALIGNMENT (Thanks to Ruud Waij for

-        helping test this.)

-      * memalign: check alignment arg

-      * realloc: don't try to shift chunks backwards, since this

-        leads to  more fragmentation in some programs and doesn't

-        seem to help in any others.

-      * Collect all cases in malloc requiring system memory into sysmalloc

-      * Use mmap as backup to sbrk

-      * Place all internal state in malloc_state

-      * Introduce fastbins (although similar to 2.5.1)

-      * Many minor tunings and cosmetic improvements

-      * Introduce USE_PUBLIC_MALLOC_WRAPPERS, USE_MALLOC_LOCK

-      * Introduce MALLOC_FAILURE_ACTION, MORECORE_CONTIGUOUS

-        Thanks to Tony E. Bennett <tbennett@nvidia.com> and others.

-      * Include errno.h to support default failure action.

-

-    V2.6.6 Sun Dec  5 07:42:19 1999  Doug Lea  (dl at gee)

-      * return null for negative arguments

-      * Added Several WIN32 cleanups from Martin C. Fong <mcfong at yahoo.com>

-         * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'

-          (e.g. WIN32 platforms)

-         * Cleanup header file inclusion for WIN32 platforms

-         * Cleanup code to avoid Microsoft Visual C++ compiler complaints

-         * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing

-           memory allocation routines

-         * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)

-         * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to

-           usage of 'assert' in non-WIN32 code

-         * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to

-           avoid infinite loop

-      * Always call 'fREe()' rather than 'free()'

-

-    V2.6.5 Wed Jun 17 15:57:31 1998  Doug Lea  (dl at gee)

-      * Fixed ordering problem with boundary-stamping

-

-    V2.6.3 Sun May 19 08:17:58 1996  Doug Lea  (dl at gee)

-      * Added pvalloc, as recommended by H.J. Liu

-      * Added 64bit pointer support mainly from Wolfram Gloger

-      * Added anonymously donated WIN32 sbrk emulation

-      * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen

-      * malloc_extend_top: fix mask error that caused wastage after

-        foreign sbrks

-      * Add linux mremap support code from HJ Liu

-

-    V2.6.2 Tue Dec  5 06:52:55 1995  Doug Lea  (dl at gee)

-      * Integrated most documentation with the code.

-      * Add support for mmap, with help from

-        Wolfram Gloger (Gloger@lrz.uni-muenchen.de).

-      * Use last_remainder in more cases.

-      * Pack bins using idea from  colin@nyx10.cs.du.edu

-      * Use ordered bins instead of best-fit threshhold

-      * Eliminate block-local decls to simplify tracing and debugging.

-      * Support another case of realloc via move into top

-      * Fix error occuring when initial sbrk_base not word-aligned.

-      * Rely on page size for units instead of SBRK_UNIT to

-        avoid surprises about sbrk alignment conventions.

-      * Add mallinfo, mallopt. Thanks to Raymond Nijssen

-        (raymond@es.ele.tue.nl) for the suggestion.

-      * Add `pad' argument to malloc_trim and top_pad mallopt parameter.

-      * More precautions for cases where other routines call sbrk,

-        courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).

-      * Added macros etc., allowing use in linux libc from

-        H.J. Lu (hjl@gnu.ai.mit.edu)

-      * Inverted this history list

-

-    V2.6.1 Sat Dec  2 14:10:57 1995  Doug Lea  (dl at gee)

-      * Re-tuned and fixed to behave more nicely with V2.6.0 changes.

-      * Removed all preallocation code since under current scheme

-        the work required to undo bad preallocations exceeds

-        the work saved in good cases for most test programs.

-      * No longer use return list or unconsolidated bins since

-        no scheme using them consistently outperforms those that don't

-        given above changes.

-      * Use best fit for very large chunks to prevent some worst-cases.

-      * Added some support for debugging

-

-    V2.6.0 Sat Nov  4 07:05:23 1995  Doug Lea  (dl at gee)

-      * Removed footers when chunks are in use. Thanks to

-        Paul Wilson (wilson@cs.texas.edu) for the suggestion.

-

-    V2.5.4 Wed Nov  1 07:54:51 1995  Doug Lea  (dl at gee)

-      * Added malloc_trim, with help from Wolfram Gloger

-        (wmglo@Dent.MED.Uni-Muenchen.DE).

-

-    V2.5.3 Tue Apr 26 10:16:01 1994  Doug Lea  (dl at g)

-

-    V2.5.2 Tue Apr  5 16:20:40 1994  Doug Lea  (dl at g)

-      * realloc: try to expand in both directions

-      * malloc: swap order of clean-bin strategy;

-      * realloc: only conditionally expand backwards

-      * Try not to scavenge used bins

-      * Use bin counts as a guide to preallocation

-      * Occasionally bin return list chunks in first scan

-      * Add a few optimizations from colin@nyx10.cs.du.edu

-

-    V2.5.1 Sat Aug 14 15:40:43 1993  Doug Lea  (dl at g)

-      * faster bin computation & slightly different binning

-      * merged all consolidations to one part of malloc proper

-         (eliminating old malloc_find_space & malloc_clean_bin)

-      * Scan 2 returns chunks (not just 1)

-      * Propagate failure in realloc if malloc returns 0

-      * Add stuff to allow compilation on non-ANSI compilers

-          from kpv@research.att.com

-

-    V2.5 Sat Aug  7 07:41:59 1993  Doug Lea  (dl at g.oswego.edu)

-      * removed potential for odd address access in prev_chunk

-      * removed dependency on getpagesize.h

-      * misc cosmetics and a bit more internal documentation

-      * anticosmetics: mangled names in macros to evade debugger strangeness

-      * tested on sparc, hp-700, dec-mips, rs6000

-          with gcc & native cc (hp, dec only) allowing

-          Detlefs & Zorn comparison study (in SIGPLAN Notices.)

-

-    Trial version Fri Aug 28 13:14:29 1992  Doug Lea  (dl at g.oswego.edu)

-      * Based loosely on libg++-1.2X malloc. (It retains some of the overall

-         structure of old version,  but most details differ.)

-

-*/

-

-#endif

+#ifdef NEDMALLOC_ENABLED
+/*
+  This is a version (aka dlmalloc) of malloc/free/realloc written by
+  Doug Lea and released to the public domain, as explained at
+  http://creativecommons.org/licenses/publicdomain.  Send questions,
+  comments, complaints, performance data, etc to dl@cs.oswego.edu
+
+* Version 2.8.4 Wed May 27 09:56:23 2009  Doug Lea  (dl at gee)
+
+   Note: There may be an updated version of this malloc obtainable at
+           ftp://gee.cs.oswego.edu/pub/misc/malloc.c
+         Check before installing!
+
+* Quickstart
+
+  This library is all in one file to simplify the most common usage:
+  ftp it, compile it (-O3), and link it into another program. All of
+  the compile-time options default to reasonable values for use on
+  most platforms.  You might later want to step through various
+  compile-time and dynamic tuning options.
+
+  For convenience, an include file for code using this malloc is at:
+     ftp://gee.cs.oswego.edu/pub/misc/malloc-2.8.4.h
+  You don't really need this .h file unless you call functions not
+  defined in your system include files.  The .h file contains only the
+  excerpts from this file needed for using this malloc on ANSI C/C++
+  systems, so long as you haven't changed compile-time options about
+  naming and tuning parameters.  If you do, then you can create your
+  own malloc.h that does include all settings by cutting at the point
+  indicated below. Note that you may already by default be using a C
+  library containing a malloc that is based on some version of this
+  malloc (for example in linux). You might still want to use the one
+  in this file to customize settings or to avoid overheads associated
+  with library versions.
+
+* Vital statistics:
+
+  Supported pointer/size_t representation:       4 or 8 bytes
+       size_t MUST be an unsigned type of the same width as
+       pointers. (If you are using an ancient system that declares
+       size_t as a signed type, or need it to be a different width
+       than pointers, you can use a previous release of this malloc
+       (e.g. 2.7.2) supporting these.)
+
+  Alignment:                                     8 bytes (default)
+       This suffices for nearly all current machines and C compilers.
+       However, you can define MALLOC_ALIGNMENT to be wider than this
+       if necessary (up to 128bytes), at the expense of using more space.
+
+  Minimum overhead per allocated chunk:   4 or  8 bytes (if 4byte sizes)
+                                          8 or 16 bytes (if 8byte sizes)
+       Each malloced chunk has a hidden word of overhead holding size
+       and status information, and additional cross-check word
+       if FOOTERS is defined.
+
+  Minimum allocated size: 4-byte ptrs:  16 bytes    (including overhead)
+                          8-byte ptrs:  32 bytes    (including overhead)
+
+       Even a request for zero bytes (i.e., malloc(0)) returns a
+       pointer to something of the minimum allocatable size.
+       The maximum overhead wastage (i.e., number of extra bytes
+       allocated than were requested in malloc) is less than or equal
+       to the minimum size, except for requests >= mmap_threshold that
+       are serviced via mmap(), where the worst case wastage is about
+       32 bytes plus the remainder from a system page (the minimal
+       mmap unit); typically 4096 or 8192 bytes.
+
+  Security: static-safe; optionally more or less
+       The "security" of malloc refers to the ability of malicious
+       code to accentuate the effects of errors (for example, freeing
+       space that is not currently malloc'ed or overwriting past the
+       ends of chunks) in code that calls malloc.  This malloc
+       guarantees not to modify any memory locations below the base of
+       heap, i.e., static variables, even in the presence of usage
+       errors.  The routines additionally detect most improper frees
+       and reallocs.  All this holds as long as the static bookkeeping
+       for malloc itself is not corrupted by some other means.  This
+       is only one aspect of security -- these checks do not, and
+       cannot, detect all possible programming errors.
+
+       If FOOTERS is defined nonzero, then each allocated chunk
+       carries an additional check word to verify that it was malloced
+       from its space.  These check words are the same within each
+       execution of a program using malloc, but differ across
+       executions, so externally crafted fake chunks cannot be
+       freed. This improves security by rejecting frees/reallocs that
+       could corrupt heap memory, in addition to the checks preventing
+       writes to statics that are always on.  This may further improve
+       security at the expense of time and space overhead.  (Note that
+       FOOTERS may also be worth using with MSPACES.)
+
+       By default detected errors cause the program to abort (calling
+       "abort()"). You can override this to instead proceed past
+       errors by defining PROCEED_ON_ERROR.  In this case, a bad free
+       has no effect, and a malloc that encounters a bad address
+       caused by user overwrites will ignore the bad address by
+       dropping pointers and indices to all known memory. This may
+       be appropriate for programs that should continue if at all
+       possible in the face of programming errors, although they may
+       run out of memory because dropped memory is never reclaimed.
+
+       If you don't like either of these options, you can define
+       CORRUPTION_ERROR_ACTION and USAGE_ERROR_ACTION to do anything
+       else. And if if you are sure that your program using malloc has
+       no errors or vulnerabilities, you can define INSECURE to 1,
+       which might (or might not) provide a small performance improvement.
+
+  Thread-safety: NOT thread-safe unless USE_LOCKS defined
+       When USE_LOCKS is defined, each public call to malloc, free,
+       etc is surrounded with either a pthread mutex or a win32
+       spinlock (depending on WIN32). This is not especially fast, and
+       can be a major bottleneck.  It is designed only to provide
+       minimal protection in concurrent environments, and to provide a
+       basis for extensions.  If you are using malloc in a concurrent
+       program, consider instead using nedmalloc
+       (http://www.nedprod.com/programs/portable/nedmalloc/) or
+       ptmalloc (See http://www.malloc.de), which are derived
+       from versions of this malloc.
+
+  System requirements: Any combination of MORECORE and/or MMAP/MUNMAP
+       This malloc can use unix sbrk or any emulation (invoked using
+       the CALL_MORECORE macro) and/or mmap/munmap or any emulation
+       (invoked using CALL_MMAP/CALL_MUNMAP) to get and release system
+       memory.  On most unix systems, it tends to work best if both
+       MORECORE and MMAP are enabled.  On Win32, it uses emulations
+       based on VirtualAlloc. It also uses common C library functions
+       like memset.
+
+  Compliance: I believe it is compliant with the Single Unix Specification
+       (See http://www.unix.org). Also SVID/XPG, ANSI C, and probably
+       others as well.
+
+* Overview of algorithms
+
+  This is not the fastest, most space-conserving, most portable, or
+  most tunable malloc ever written. However it is among the fastest
+  while also being among the most space-conserving, portable and
+  tunable.  Consistent balance across these factors results in a good
+  general-purpose allocator for malloc-intensive programs.
+
+  In most ways, this malloc is a best-fit allocator. Generally, it
+  chooses the best-fitting existing chunk for a request, with ties
+  broken in approximately least-recently-used order. (This strategy
+  normally maintains low fragmentation.) However, for requests less
+  than 256bytes, it deviates from best-fit when there is not an
+  exactly fitting available chunk by preferring to use space adjacent
+  to that used for the previous small request, as well as by breaking
+  ties in approximately most-recently-used order. (These enhance
+  locality of series of small allocations.)  And for very large requests
+  (>= 256Kb by default), it relies on system memory mapping
+  facilities, if supported.  (This helps avoid carrying around and
+  possibly fragmenting memory used only for large chunks.)
+
+  All operations (except malloc_stats and mallinfo) have execution
+  times that are bounded by a constant factor of the number of bits in
+  a size_t, not counting any clearing in calloc or copying in realloc,
+  or actions surrounding MORECORE and MMAP that have times
+  proportional to the number of non-contiguous regions returned by
+  system allocation routines, which is often just 1. In real-time
+  applications, you can optionally suppress segment traversals using
+  NO_SEGMENT_TRAVERSAL, which assures bounded execution even when
+  system allocators return non-contiguous spaces, at the typical
+  expense of carrying around more memory and increased fragmentation.
+
+  The implementation is not very modular and seriously overuses
+  macros. Perhaps someday all C compilers will do as good a job
+  inlining modular code as can now be done by brute-force expansion,
+  but now, enough of them seem not to.
+
+  Some compilers issue a lot of warnings about code that is
+  dead/unreachable only on some platforms, and also about intentional
+  uses of negation on unsigned types. All known cases of each can be
+  ignored.
+
+  For a longer but out of date high-level description, see
+     http://gee.cs.oswego.edu/dl/html/malloc.html
+
+* MSPACES
+  If MSPACES is defined, then in addition to malloc, free, etc.,
+  this file also defines mspace_malloc, mspace_free, etc. These
+  are versions of malloc routines that take an "mspace" argument
+  obtained using create_mspace, to control all internal bookkeeping.
+  If ONLY_MSPACES is defined, only these versions are compiled.
+  So if you would like to use this allocator for only some allocations,
+  and your system malloc for others, you can compile with
+  ONLY_MSPACES and then do something like...
+    static mspace mymspace = create_mspace(0,0); // for example
+    #define mymalloc(bytes)  mspace_malloc(mymspace, bytes)
+
+  (Note: If you only need one instance of an mspace, you can instead
+  use "USE_DL_PREFIX" to relabel the global malloc.)
+
+  You can similarly create thread-local allocators by storing
+  mspaces as thread-locals. For example:
+    static __thread mspace tlms = 0;
+    void*  tlmalloc(size_t bytes) {
+      if (tlms == 0) tlms = create_mspace(0, 0);
+      return mspace_malloc(tlms, bytes);
+    }
+    void  tlfree(void* mem) { mspace_free(tlms, mem); }
+
+  Unless FOOTERS is defined, each mspace is completely independent.
+  You cannot allocate from one and free to another (although
+  conformance is only weakly checked, so usage errors are not always
+  caught). If FOOTERS is defined, then each chunk carries around a tag
+  indicating its originating mspace, and frees are directed to their
+  originating spaces.
+
+ -------------------------  Compile-time options ---------------------------
+
+Be careful in setting #define values for numerical constants of type
+size_t. On some systems, literal values are not automatically extended
+to size_t precision unless they are explicitly casted. You can also
+use the symbolic values MAX_SIZE_T, SIZE_T_ONE, etc below.
+
+WIN32                    default: defined if _WIN32 defined
+  Defining WIN32 sets up defaults for MS environment and compilers.
+  Otherwise defaults are for unix. Beware that there seem to be some
+  cases where this malloc might not be a pure drop-in replacement for
+  Win32 malloc: Random-looking failures from Win32 GDI API's (eg;
+  SetDIBits()) may be due to bugs in some video driver implementations
+  when pixel buffers are malloc()ed, and the region spans more than
+  one VirtualAlloc()ed region. Because dlmalloc uses a small (64Kb)
+  default granularity, pixel buffers may straddle virtual allocation
+  regions more often than when using the Microsoft allocator.  You can
+  avoid this by using VirtualAlloc() and VirtualFree() for all pixel
+  buffers rather than using malloc().  If this is not possible,
+  recompile this malloc with a larger DEFAULT_GRANULARITY.
+
+MALLOC_ALIGNMENT         default: (size_t)8
+  Controls the minimum alignment for malloc'ed chunks.  It must be a
+  power of two and at least 8, even on machines for which smaller
+  alignments would suffice. It may be defined as larger than this
+  though. Note however that code and data structures are optimized for
+  the case of 8-byte alignment.
+
+MSPACES                  default: 0 (false)
+  If true, compile in support for independent allocation spaces.
+  This is only supported if HAVE_MMAP is true.
+
+ONLY_MSPACES             default: 0 (false)
+  If true, only compile in mspace versions, not regular versions.
+
+USE_LOCKS                default: 0 (false)
+  Causes each call to each public routine to be surrounded with
+  pthread or WIN32 mutex lock/unlock. (If set true, this can be
+  overridden on a per-mspace basis for mspace versions.) If set to a
+  non-zero value other than 1, locks are used, but their
+  implementation is left out, so lock functions must be supplied manually,
+  as described below.
+
+USE_SPIN_LOCKS           default: 1 iff USE_LOCKS and on x86 using gcc or MSC
+  If true, uses custom spin locks for locking. This is currently
+  supported only for x86 platforms using gcc or recent MS compilers.
+  Otherwise, posix locks or win32 critical sections are used.
+
+FOOTERS                  default: 0
+  If true, provide extra checking and dispatching by placing
+  information in the footers of allocated chunks. This adds
+  space and time overhead.
+
+INSECURE                 default: 0
+  If true, omit checks for usage errors and heap space overwrites.
+
+USE_DL_PREFIX            default: NOT defined
+  Causes compiler to prefix all public routines with the string 'dl'.
+  This can be useful when you only want to use this malloc in one part
+  of a program, using your regular system malloc elsewhere.
+
+ABORT                    default: defined as abort()
+  Defines how to abort on failed checks.  On most systems, a failed
+  check cannot die with an "assert" or even print an informative
+  message, because the underlying print routines in turn call malloc,
+  which will fail again.  Generally, the best policy is to simply call
+  abort(). It's not very useful to do more than this because many
+  errors due to overwriting will show up as address faults (null, odd
+  addresses etc) rather than malloc-triggered checks, so will also
+  abort.  Also, most compilers know that abort() does not return, so
+  can better optimize code conditionally calling it.
+
+PROCEED_ON_ERROR           default: defined as 0 (false)
+  Controls whether detected bad addresses cause them to bypassed
+  rather than aborting. If set, detected bad arguments to free and
+  realloc are ignored. And all bookkeeping information is zeroed out
+  upon a detected overwrite of freed heap space, thus losing the
+  ability to ever return it from malloc again, but enabling the
+  application to proceed. If PROCEED_ON_ERROR is defined, the
+  static variable malloc_corruption_error_count is compiled in
+  and can be examined to see if errors have occurred. This option
+  generates slower code than the default abort policy.
+
+DEBUG                    default: NOT defined
+  The DEBUG setting is mainly intended for people trying to modify
+  this code or diagnose problems when porting to new platforms.
+  However, it may also be able to better isolate user errors than just
+  using runtime checks.  The assertions in the check routines spell
+  out in more detail the assumptions and invariants underlying the
+  algorithms.  The checking is fairly extensive, and will slow down
+  execution noticeably. Calling malloc_stats or mallinfo with DEBUG
+  set will attempt to check every non-mmapped allocated and free chunk
+  in the course of computing the summaries.
+
+ABORT_ON_ASSERT_FAILURE   default: defined as 1 (true)
+  Debugging assertion failures can be nearly impossible if your
+  version of the assert macro causes malloc to be called, which will
+  lead to a cascade of further failures, blowing the runtime stack.
+  ABORT_ON_ASSERT_FAILURE cause assertions failures to call abort(),
+  which will usually make debugging easier.
+
+MALLOC_FAILURE_ACTION     default: sets errno to ENOMEM, or no-op on win32
+  The action to take before "return 0" when malloc fails to be able to
+  return memory because there is none available.
+
+HAVE_MORECORE             default: 1 (true) unless win32 or ONLY_MSPACES
+  True if this system supports sbrk or an emulation of it.
+
+MORECORE                  default: sbrk
+  The name of the sbrk-style system routine to call to obtain more
+  memory.  See below for guidance on writing custom MORECORE
+  functions. The type of the argument to sbrk/MORECORE varies across
+  systems.  It cannot be size_t, because it supports negative
+  arguments, so it is normally the signed type of the same width as
+  size_t (sometimes declared as "intptr_t").  It doesn't much matter
+  though. Internally, we only call it with arguments less than half
+  the max value of a size_t, which should work across all reasonable
+  possibilities, although sometimes generating compiler warnings.
+
+MORECORE_CONTIGUOUS       default: 1 (true) if HAVE_MORECORE
+  If true, take advantage of fact that consecutive calls to MORECORE
+  with positive arguments always return contiguous increasing
+  addresses.  This is true of unix sbrk. It does not hurt too much to
+  set it true anyway, since malloc copes with non-contiguities.
+  Setting it false when definitely non-contiguous saves time
+  and possibly wasted space it would take to discover this though.
+
+MORECORE_CANNOT_TRIM      default: NOT defined
+  True if MORECORE cannot release space back to the system when given
+  negative arguments. This is generally necessary only if you are
+  using a hand-crafted MORECORE function that cannot handle negative
+  arguments.
+
+NO_SEGMENT_TRAVERSAL       default: 0
+  If non-zero, suppresses traversals of memory segments
+  returned by either MORECORE or CALL_MMAP. This disables
+  merging of segments that are contiguous, and selectively
+  releasing them to the OS if unused, but bounds execution times.
+
+HAVE_MMAP                 default: 1 (true)
+  True if this system supports mmap or an emulation of it.  If so, and
+  HAVE_MORECORE is not true, MMAP is used for all system
+  allocation. If set and HAVE_MORECORE is true as well, MMAP is
+  primarily used to directly allocate very large blocks. It is also
+  used as a backup strategy in cases where MORECORE fails to provide
+  space from system. Note: A single call to MUNMAP is assumed to be
+  able to unmap memory that may have be allocated using multiple calls
+  to MMAP, so long as they are adjacent.
+
+HAVE_MREMAP               default: 1 on linux, else 0
+  If true realloc() uses mremap() to re-allocate large blocks and
+  extend or shrink allocation spaces.
+
+MMAP_CLEARS               default: 1 except on WINCE.
+  True if mmap clears memory so calloc doesn't need to. This is true
+  for standard unix mmap using /dev/zero and on WIN32 except for WINCE.
+
+USE_BUILTIN_FFS            default: 0 (i.e., not used)
+  Causes malloc to use the builtin ffs() function to compute indices.
+  Some compilers may recognize and intrinsify ffs to be faster than the
+  supplied C version. Also, the case of x86 using gcc is special-cased
+  to an asm instruction, so is already as fast as it can be, and so
+  this setting has no effect. Similarly for Win32 under recent MS compilers.
+  (On most x86s, the asm version is only slightly faster than the C version.)
+
+malloc_getpagesize         default: derive from system includes, or 4096.
+  The system page size. To the extent possible, this malloc manages
+  memory from the system in page-size units.  This may be (and
+  usually is) a function rather than a constant. This is ignored
+  if WIN32, where page size is determined using getSystemInfo during
+  initialization. This may be several megabytes if ENABLE_LARGE_PAGES
+  is enabled.
+
+ENABLE_LARGE_PAGES         default: NOT defined
+  Causes the system page size to be the value of GetLargePageMinimum()
+  if that function is available (Windows Server 2003/Vista or later).
+  This allows the use of large page entries in the MMU which can
+  significantly improve performance in large working set applications
+  as TLB cache load is reduced by a factor of three. Note that enabling
+  this option is equal to locking the process' memory in current
+  implementations of Windows and requires the SE_LOCK_MEMORY_PRIVILEGE
+  to be held by the process in order to succeed.
+
+USE_DEV_RANDOM             default: 0 (i.e., not used)
+  Causes malloc to use /dev/random to initialize secure magic seed for
+  stamping footers. Otherwise, the current time is used.
+
+NO_MALLINFO                default: 0
+  If defined, don't compile "mallinfo". This can be a simple way
+  of dealing with mismatches between system declarations and
+  those in this file.
+
+MALLINFO_FIELD_TYPE        default: size_t
+  The type of the fields in the mallinfo struct. This was originally
+  defined as "int" in SVID etc, but is more usefully defined as
+  size_t. The value is used only if  HAVE_USR_INCLUDE_MALLOC_H is not set
+
+REALLOC_ZERO_BYTES_FREES    default: not defined
+  This should be set if a call to realloc with zero bytes should
+  be the same as a call to free. Some people think it should. Otherwise,
+  since this malloc returns a unique pointer for malloc(0), so does
+  realloc(p, 0).
+
+LACKS_UNISTD_H, LACKS_FCNTL_H, LACKS_SYS_PARAM_H, LACKS_SYS_MMAN_H
+LACKS_STRINGS_H, LACKS_STRING_H, LACKS_SYS_TYPES_H,  LACKS_ERRNO_H
+LACKS_STDLIB_H                default: NOT defined unless on WIN32
+  Define these if your system does not have these header files.
+  You might need to manually insert some of the declarations they provide.
+
+DEFAULT_GRANULARITY        default: page size if MORECORE_CONTIGUOUS,
+                                system_info.dwAllocationGranularity in WIN32,
+                                GetLargePageMinimum() if ENABLE_LARGE_PAGES,
+                                otherwise 64K.
+      Also settable using mallopt(M_GRANULARITY, x)
+  The unit for allocating and deallocating memory from the system.  On
+  most systems with contiguous MORECORE, there is no reason to
+  make this more than a page. However, systems with MMAP tend to
+  either require or encourage larger granularities.  You can increase
+  this value to prevent system allocation functions to be called so
+  often, especially if they are slow.  The value must be at least one
+  page and must be a power of two.  Setting to 0 causes initialization
+  to either page size or win32 region size.  (Note: In previous
+  versions of malloc, the equivalent of this option was called
+  "TOP_PAD")
+
+DEFAULT_GRANULARITY_ALIGNED default: undefined (which means page size)
+  Whether to enforce alignment when allocating and deallocating memory
+  from the system i.e. the base address of all allocations will be
+  aligned to DEFAULT_GRANULARITY if it is set. Note that enabling this carries
+  some overhead as multiple calls must now be made when probing for a valid
+  aligned value, however it does greatly ease the checking for whether
+  a given memory pointer was allocated by this allocator rather than
+  some other.
+
+DEFAULT_TRIM_THRESHOLD    default: 2MB
+      Also settable using mallopt(M_TRIM_THRESHOLD, x)
+  The maximum amount of unused top-most memory to keep before
+  releasing via malloc_trim in free().  Automatic trimming is mainly
+  useful in long-lived programs using contiguous MORECORE.  Because
+  trimming via sbrk can be slow on some systems, and can sometimes be
+  wasteful (in cases where programs immediately afterward allocate
+  more large chunks) the value should be high enough so that your
+  overall system performance would improve by releasing this much
+  memory.  As a rough guide, you might set to a value close to the
+  average size of a process (program) running on your system.
+  Releasing this much memory would allow such a process to run in
+  memory.  Generally, it is worth tuning trim thresholds when a
+  program undergoes phases where several large chunks are allocated
+  and released in ways that can reuse each other's storage, perhaps
+  mixed with phases where there are no such chunks at all. The trim
+  value must be greater than page size to have any useful effect.  To
+  disable trimming completely, you can set to MAX_SIZE_T. Note that the trick
+  some people use of mallocing a huge space and then freeing it at
+  program startup, in an attempt to reserve system memory, doesn't
+  have the intended effect under automatic trimming, since that memory
+  will immediately be returned to the system.
+
+DEFAULT_MMAP_THRESHOLD       default: 256K
+      Also settable using mallopt(M_MMAP_THRESHOLD, x)
+  The request size threshold for using MMAP to directly service a
+  request. Requests of at least this size that cannot be allocated
+  using already-existing space will be serviced via mmap.  (If enough
+  normal freed space already exists it is used instead.)  Using mmap
+  segregates relatively large chunks of memory so that they can be
+  individually obtained and released from the host system. A request
+  serviced through mmap is never reused by any other request (at least
+  not directly; the system may just so happen to remap successive
+  requests to the same locations).  Segregating space in this way has
+  the benefits that: Mmapped space can always be individually released
+  back to the system, which helps keep the system level memory demands
+  of a long-lived program low.  Also, mapped memory doesn't become
+  `locked' between other chunks, as can happen with normally allocated
+  chunks, which means that even trimming via malloc_trim would not
+  release them.  However, it has the disadvantage that the space
+  cannot be reclaimed, consolidated, and then used to service later
+  requests, as happens with normal chunks.  The advantages of mmap
+  nearly always outweigh disadvantages for "large" chunks, but the
+  value of "large" may vary across systems.  The default is an
+  empirically derived value that works well in most systems. You can
+  disable mmap by setting to MAX_SIZE_T.
+
+MAX_RELEASE_CHECK_RATE   default: 4095 unless not HAVE_MMAP
+  The number of consolidated frees between checks to release
+  unused segments when freeing. When using non-contiguous segments,
+  especially with multiple mspaces, checking only for topmost space
+  doesn't always suffice to trigger trimming. To compensate for this,
+  free() will, with a period of MAX_RELEASE_CHECK_RATE (or the
+  current number of segments, if greater) try to release unused
+  segments to the OS when freeing chunks that result in
+  consolidation. The best value for this parameter is a compromise
+  between slowing down frees with relatively costly checks that
+  rarely trigger versus holding on to unused memory. To effectively
+  disable, set to MAX_SIZE_T. This may lead to a very slight speed
+  improvement at the expense of carrying around more memory.
+*/
+
+/* Version identifier to allow people to support multiple versions */
+#ifndef DLMALLOC_VERSION
+#define DLMALLOC_VERSION 20804
+#endif /* DLMALLOC_VERSION */
+
+#ifndef WIN32
+#ifdef _WIN32
+#define WIN32 1
+#endif  /* _WIN32 */
+#ifdef _WIN32_WCE
+#define LACKS_FCNTL_H
+#define WIN32 1
+#endif /* _WIN32_WCE */
+#endif  /* WIN32 */
+#ifdef WIN32
+#define WIN32_LEAN_AND_MEAN
+#include <windows.h>
+#include <tchar.h>
+#define HAVE_MMAP 1
+#define HAVE_MORECORE 0
+#define LACKS_UNISTD_H
+#define LACKS_SYS_PARAM_H
+#define LACKS_SYS_MMAN_H
+#define LACKS_STRING_H
+#define LACKS_STRINGS_H
+#define LACKS_SYS_TYPES_H
+#define LACKS_ERRNO_H
+#ifndef MALLOC_FAILURE_ACTION
+#define MALLOC_FAILURE_ACTION
+#endif /* MALLOC_FAILURE_ACTION */
+#ifdef _WIN32_WCE /* WINCE reportedly does not clear */
+#define MMAP_CLEARS 0
+#else
+#define MMAP_CLEARS 1
+#endif /* _WIN32_WCE */
+#endif  /* WIN32 */
+
+#if defined(DARWIN) || defined(_DARWIN)
+/* Mac OSX docs advise not to use sbrk; it seems better to use mmap */
+#ifndef HAVE_MORECORE
+#define HAVE_MORECORE 0
+#define HAVE_MMAP 1
+/* OSX allocators provide 16 byte alignment */
+#ifndef MALLOC_ALIGNMENT
+#define MALLOC_ALIGNMENT ((size_t)16U)
+#endif
+#endif  /* HAVE_MORECORE */
+#endif  /* DARWIN */
+
+#ifndef LACKS_SYS_TYPES_H
+#include <sys/types.h>  /* For size_t */
+#endif  /* LACKS_SYS_TYPES_H */
+
+#if (defined(__GNUC__) && ((defined(__i386__) || defined(__x86_64__)))) || (defined(_MSC_VER) && _MSC_VER>=1310)
+#define SPIN_LOCKS_AVAILABLE 1
+#else
+#define SPIN_LOCKS_AVAILABLE 0
+#endif
+
+/* The maximum possible size_t value has all bits set */
+#define MAX_SIZE_T           (~(size_t)0)
+
+#ifndef ONLY_MSPACES
+#define ONLY_MSPACES 0     /* define to a value */
+#else
+#define ONLY_MSPACES 1
+#endif  /* ONLY_MSPACES */
+#ifndef MSPACES
+#if ONLY_MSPACES
+#define MSPACES 1
+#else   /* ONLY_MSPACES */
+#define MSPACES 0
+#endif  /* ONLY_MSPACES */
+#endif  /* MSPACES */
+#ifndef MALLOC_ALIGNMENT
+#define MALLOC_ALIGNMENT ((size_t)8U)
+#endif  /* MALLOC_ALIGNMENT */
+#ifndef FOOTERS
+#define FOOTERS 0
+#endif  /* FOOTERS */
+#ifndef ABORT
+#define ABORT  abort()
+#endif  /* ABORT */
+#ifndef ABORT_ON_ASSERT_FAILURE
+#define ABORT_ON_ASSERT_FAILURE 1
+#endif  /* ABORT_ON_ASSERT_FAILURE */
+#ifndef PROCEED_ON_ERROR
+#define PROCEED_ON_ERROR 0
+#endif  /* PROCEED_ON_ERROR */
+#ifndef USE_LOCKS
+#define USE_LOCKS 0
+#endif  /* USE_LOCKS */
+#ifndef USE_SPIN_LOCKS
+#if USE_LOCKS && SPIN_LOCKS_AVAILABLE
+#define USE_SPIN_LOCKS 1
+#else
+#define USE_SPIN_LOCKS 0
+#endif /* USE_LOCKS && SPIN_LOCKS_AVAILABLE. */
+#endif /* USE_SPIN_LOCKS */
+#ifndef INSECURE
+#define INSECURE 0
+#endif  /* INSECURE */
+#ifndef HAVE_MMAP
+#define HAVE_MMAP 1
+#endif  /* HAVE_MMAP */
+#ifndef MMAP_CLEARS
+#define MMAP_CLEARS 1
+#endif  /* MMAP_CLEARS */
+#ifndef HAVE_MREMAP
+#ifdef linux
+#define HAVE_MREMAP 1
+#else   /* linux */
+#define HAVE_MREMAP 0
+#endif  /* linux */
+#endif  /* HAVE_MREMAP */
+#ifndef MALLOC_FAILURE_ACTION
+#define MALLOC_FAILURE_ACTION  errno = ENOMEM;
+#endif  /* MALLOC_FAILURE_ACTION */
+#ifndef HAVE_MORECORE
+#if ONLY_MSPACES
+#define HAVE_MORECORE 0
+#else   /* ONLY_MSPACES */
+#define HAVE_MORECORE 1
+#endif  /* ONLY_MSPACES */
+#endif  /* HAVE_MORECORE */
+#if !HAVE_MORECORE
+#define MORECORE_CONTIGUOUS 0
+#else   /* !HAVE_MORECORE */
+#define MORECORE_DEFAULT sbrk
+#ifndef MORECORE_CONTIGUOUS
+#define MORECORE_CONTIGUOUS 1
+#endif  /* MORECORE_CONTIGUOUS */
+#endif  /* HAVE_MORECORE */
+#ifndef DEFAULT_GRANULARITY
+#if (MORECORE_CONTIGUOUS || defined(WIN32))
+#define DEFAULT_GRANULARITY (0)  /* 0 means to compute in init_mparams */
+#else   /* MORECORE_CONTIGUOUS */
+#define DEFAULT_GRANULARITY ((size_t)64U * (size_t)1024U)
+#endif  /* MORECORE_CONTIGUOUS */
+#endif  /* DEFAULT_GRANULARITY */
+#ifndef DEFAULT_TRIM_THRESHOLD
+#ifndef MORECORE_CANNOT_TRIM
+#define DEFAULT_TRIM_THRESHOLD ((size_t)2U * (size_t)1024U * (size_t)1024U)
+#else   /* MORECORE_CANNOT_TRIM */
+#define DEFAULT_TRIM_THRESHOLD MAX_SIZE_T
+#endif  /* MORECORE_CANNOT_TRIM */
+#endif  /* DEFAULT_TRIM_THRESHOLD */
+#ifndef DEFAULT_MMAP_THRESHOLD
+#if HAVE_MMAP
+#define DEFAULT_MMAP_THRESHOLD ((size_t)256U * (size_t)1024U)
+#else   /* HAVE_MMAP */
+#define DEFAULT_MMAP_THRESHOLD MAX_SIZE_T
+#endif  /* HAVE_MMAP */
+#endif  /* DEFAULT_MMAP_THRESHOLD */
+#ifndef MAX_RELEASE_CHECK_RATE
+#if HAVE_MMAP
+#define MAX_RELEASE_CHECK_RATE 4095
+#else
+#define MAX_RELEASE_CHECK_RATE MAX_SIZE_T
+#endif /* HAVE_MMAP */
+#endif /* MAX_RELEASE_CHECK_RATE */
+#ifndef USE_BUILTIN_FFS
+#define USE_BUILTIN_FFS 0
+#endif  /* USE_BUILTIN_FFS */
+#ifndef USE_DEV_RANDOM
+#define USE_DEV_RANDOM 0
+#endif  /* USE_DEV_RANDOM */
+#ifndef NO_MALLINFO
+#define NO_MALLINFO 0
+#endif  /* NO_MALLINFO */
+#ifndef MALLINFO_FIELD_TYPE
+#define MALLINFO_FIELD_TYPE size_t
+#endif  /* MALLINFO_FIELD_TYPE */
+#ifndef NO_SEGMENT_TRAVERSAL
+#define NO_SEGMENT_TRAVERSAL 0
+#endif /* NO_SEGMENT_TRAVERSAL */
+
+/*
+  mallopt tuning options.  SVID/XPG defines four standard parameter
+  numbers for mallopt, normally defined in malloc.h.  None of these
+  are used in this malloc, so setting them has no effect. But this
+  malloc does support the following options.
+*/
+
+#define M_TRIM_THRESHOLD     (-1)
+#define M_GRANULARITY        (-2)
+#define M_MMAP_THRESHOLD     (-3)
+
+/* ------------------------ Mallinfo declarations ------------------------ */
+
+#if !NO_MALLINFO
+/*
+  This version of malloc supports the standard SVID/XPG mallinfo
+  routine that returns a struct containing usage properties and
+  statistics. It should work on any system that has a
+  /usr/include/malloc.h defining struct mallinfo.  The main
+  declaration needed is the mallinfo struct that is returned (by-copy)
+  by mallinfo().  The malloinfo struct contains a bunch of fields that
+  are not even meaningful in this version of malloc.  These fields are
+  are instead filled by mallinfo() with other numbers that might be of
+  interest.
+
+  HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
+  /usr/include/malloc.h file that includes a declaration of struct
+  mallinfo.  If so, it is included; else a compliant version is
+  declared below.  These must be precisely the same for mallinfo() to
+  work.  The original SVID version of this struct, defined on most
+  systems with mallinfo, declares all fields as ints. But some others
+  define as unsigned long. If your system defines the fields using a
+  type of different width than listed here, you MUST #include your
+  system version and #define HAVE_USR_INCLUDE_MALLOC_H.
+*/
+
+/* #define HAVE_USR_INCLUDE_MALLOC_H */
+
+#ifdef HAVE_USR_INCLUDE_MALLOC_H
+#include "/usr/include/malloc.h"
+#else /* HAVE_USR_INCLUDE_MALLOC_H */
+#ifndef STRUCT_MALLINFO_DECLARED
+#define STRUCT_MALLINFO_DECLARED 1
+struct mallinfo {
+  MALLINFO_FIELD_TYPE arena;    /* non-mmapped space allocated from system */
+  MALLINFO_FIELD_TYPE ordblks;  /* number of free chunks */
+  MALLINFO_FIELD_TYPE smblks;   /* always 0 */
+  MALLINFO_FIELD_TYPE hblks;    /* always 0 */
+  MALLINFO_FIELD_TYPE hblkhd;   /* space in mmapped regions */
+  MALLINFO_FIELD_TYPE usmblks;  /* maximum total allocated space */
+  MALLINFO_FIELD_TYPE fsmblks;  /* always 0 */
+  MALLINFO_FIELD_TYPE uordblks; /* total allocated space */
+  MALLINFO_FIELD_TYPE fordblks; /* total free space */
+  MALLINFO_FIELD_TYPE keepcost; /* releasable (via malloc_trim) space */
+};
+#endif /* STRUCT_MALLINFO_DECLARED */
+#endif /* HAVE_USR_INCLUDE_MALLOC_H */
+#endif /* NO_MALLINFO */
+
+/*
+  Try to persuade compilers to inline. The most critical functions for
+  inlining are defined as macros, so these aren't used for them.
+*/
+
+#ifndef FORCEINLINE
+  #if defined(__GNUC__)
+#define FORCEINLINE __inline __attribute__ ((always_inline))
+  #elif defined(_MSC_VER)
+    #define FORCEINLINE __forceinline
+  #endif
+#endif
+#ifndef NOINLINE
+  #if defined(__GNUC__)
+    #define NOINLINE __attribute__ ((noinline))
+  #elif defined(_MSC_VER)
+    #define NOINLINE __declspec(noinline)
+  #else
+    #define NOINLINE
+  #endif
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#ifndef FORCEINLINE
+ #define FORCEINLINE inline
+#endif
+#endif /* __cplusplus */
+#ifndef FORCEINLINE
+ #define FORCEINLINE
+#endif
+
+#if !ONLY_MSPACES
+
+/* ------------------- Declarations of public routines ------------------- */
+
+#ifndef USE_DL_PREFIX
+#define dlcalloc               calloc
+#define dlfree                 free
+#define dlmalloc               malloc
+#define dlmemalign             memalign
+#define dlrealloc              realloc
+#define dlvalloc               valloc
+#define dlpvalloc              pvalloc
+#define dlmallinfo             mallinfo
+#define dlmallopt              mallopt
+#define dlmalloc_trim          malloc_trim
+#define dlmalloc_stats         malloc_stats
+#define dlmalloc_usable_size   malloc_usable_size
+#define dlmalloc_footprint     malloc_footprint
+#define dlmalloc_max_footprint malloc_max_footprint
+#define dlindependent_calloc   independent_calloc
+#define dlindependent_comalloc independent_comalloc
+#endif /* USE_DL_PREFIX */
+
+
+/*
+  malloc(size_t n)
+  Returns a pointer to a newly allocated chunk of at least n bytes, or
+  null if no space is available, in which case errno is set to ENOMEM
+  on ANSI C systems.
+
+  If n is zero, malloc returns a minimum-sized chunk. (The minimum
+  size is 16 bytes on most 32bit systems, and 32 bytes on 64bit
+  systems.)  Note that size_t is an unsigned type, so calls with
+  arguments that would be negative if signed are interpreted as
+  requests for huge amounts of space, which will often fail. The
+  maximum supported value of n differs across systems, but is in all
+  cases less than the maximum representable value of a size_t.
+*/
+void* dlmalloc(size_t);
+
+/*
+  free(void* p)
+  Releases the chunk of memory pointed to by p, that had been previously
+  allocated using malloc or a related routine such as realloc.
+  It has no effect if p is null. If p was not malloced or already
+  freed, free(p) will by default cause the current program to abort.
+*/
+void  dlfree(void*);
+
+/*
+  calloc(size_t n_elements, size_t element_size);
+  Returns a pointer to n_elements * element_size bytes, with all locations
+  set to zero.
+*/
+void* dlcalloc(size_t, size_t);
+
+/*
+  realloc(void* p, size_t n)
+  Returns a pointer to a chunk of size n that contains the same data
+  as does chunk p up to the minimum of (n, p's size) bytes, or null
+  if no space is available.
+
+  The returned pointer may or may not be the same as p. The algorithm
+  prefers extending p in most cases when possible, otherwise it
+  employs the equivalent of a malloc-copy-free sequence.
+
+  If p is null, realloc is equivalent to malloc.
+
+  If space is not available, realloc returns null, errno is set (if on
+  ANSI) and p is NOT freed.
+
+  if n is for fewer bytes than already held by p, the newly unused
+  space is lopped off and freed if possible.  realloc with a size
+  argument of zero (re)allocates a minimum-sized chunk.
+
+  The old unix realloc convention of allowing the last-free'd chunk
+  to be used as an argument to realloc is not supported.
+*/
+
+void* dlrealloc(void*, size_t);
+
+/*
+  memalign(size_t alignment, size_t n);
+  Returns a pointer to a newly allocated chunk of n bytes, aligned
+  in accord with the alignment argument.
+
+  The alignment argument should be a power of two. If the argument is
+  not a power of two, the nearest greater power is used.
+  8-byte alignment is guaranteed by normal malloc calls, so don't
+  bother calling memalign with an argument of 8 or less.
+
+  Overreliance on memalign is a sure way to fragment space.
+*/
+void* dlmemalign(size_t, size_t);
+
+/*
+  valloc(size_t n);
+  Equivalent to memalign(pagesize, n), where pagesize is the page
+  size of the system. If the pagesize is unknown, 4096 is used.
+*/
+void* dlvalloc(size_t);
+
+/*
+  mallopt(int parameter_number, int parameter_value)
+  Sets tunable parameters The format is to provide a
+  (parameter-number, parameter-value) pair.  mallopt then sets the
+  corresponding parameter to the argument value if it can (i.e., so
+  long as the value is meaningful), and returns 1 if successful else
+  0.  To workaround the fact that mallopt is specified to use int,
+  not size_t parameters, the value -1 is specially treated as the
+  maximum unsigned size_t value.
+
+  SVID/XPG/ANSI defines four standard param numbers for mallopt,
+  normally defined in malloc.h.  None of these are use in this malloc,
+  so setting them has no effect. But this malloc also supports other
+  options in mallopt. See below for details.  Briefly, supported
+  parameters are as follows (listed defaults are for "typical"
+  configurations).
+
+  Symbol            param #  default    allowed param values
+  M_TRIM_THRESHOLD     -1   2*1024*1024   any   (-1 disables)
+  M_GRANULARITY        -2     page size   any power of 2 >= page size
+  M_MMAP_THRESHOLD     -3      256*1024   any   (or 0 if no MMAP support)
+*/
+int dlmallopt(int, int);
+
+/*
+  malloc_footprint();
+  Returns the number of bytes obtained from the system.  The total
+  number of bytes allocated by malloc, realloc etc., is less than this
+  value. Unlike mallinfo, this function returns only a precomputed
+  result, so can be called frequently to monitor memory consumption.
+  Even if locks are otherwise defined, this function does not use them,
+  so results might not be up to date.
+*/
+size_t dlmalloc_footprint(void);
+
+/*
+  malloc_max_footprint();
+  Returns the maximum number of bytes obtained from the system. This
+  value will be greater than current footprint if deallocated space
+  has been reclaimed by the system. The peak number of bytes allocated
+  by malloc, realloc etc., is less than this value. Unlike mallinfo,
+  this function returns only a precomputed result, so can be called
+  frequently to monitor memory consumption.  Even if locks are
+  otherwise defined, this function does not use them, so results might
+  not be up to date.
+*/
+size_t dlmalloc_max_footprint(void);
+
+#if !NO_MALLINFO
+/*
+  mallinfo()
+  Returns (by copy) a struct containing various summary statistics:
+
+  arena:     current total non-mmapped bytes allocated from system
+  ordblks:   the number of free chunks
+  smblks:    always zero.
+  hblks:     current number of mmapped regions
+  hblkhd:    total bytes held in mmapped regions
+  usmblks:   the maximum total allocated space. This will be greater
+                than current total if trimming has occurred.
+  fsmblks:   always zero
+  uordblks:  current total allocated space (normal or mmapped)
+  fordblks:  total free space
+  keepcost:  the maximum number of bytes that could ideally be released
+               back to system via malloc_trim. ("ideally" means that
+               it ignores page restrictions etc.)
+
+  Because these fields are ints, but internal bookkeeping may
+  be kept as longs, the reported values may wrap around zero and
+  thus be inaccurate.
+*/
+struct mallinfo dlmallinfo(void);
+#endif /* NO_MALLINFO */
+
+/*
+  independent_calloc(size_t n_elements, size_t element_size, void* chunks[]);
+
+  independent_calloc is similar to calloc, but instead of returning a
+  single cleared space, it returns an array of pointers to n_elements
+  independent elements that can hold contents of size elem_size, each
+  of which starts out cleared, and can be independently freed,
+  realloc'ed etc. The elements are guaranteed to be adjacently
+  allocated (this is not guaranteed to occur with multiple callocs or
+  mallocs), which may also improve cache locality in some
+  applications.
+
+  The "chunks" argument is optional (i.e., may be null, which is
+  probably the most typical usage). If it is null, the returned array
+  is itself dynamically allocated and should also be freed when it is
+  no longer needed. Otherwise, the chunks array must be of at least
+  n_elements in length. It is filled in with the pointers to the
+  chunks.
+
+  In either case, independent_calloc returns this pointer array, or
+  null if the allocation failed.  If n_elements is zero and "chunks"
+  is null, it returns a chunk representing an array with zero elements
+  (which should be freed if not wanted).
+
+  Each element must be individually freed when it is no longer
+  needed. If you'd like to instead be able to free all at once, you
+  should instead use regular calloc and assign pointers into this
+  space to represent elements.  (In this case though, you cannot
+  independently free elements.)
+
+  independent_calloc simplifies and speeds up implementations of many
+  kinds of pools.  It may also be useful when constructing large data
+  structures that initially have a fixed number of fixed-sized nodes,
+  but the number is not known at compile time, and some of the nodes
+  may later need to be freed. For example:
+
+  struct Node { int item; struct Node* next; };
+
+  struct Node* build_list() {
+    struct Node** pool;
+    int n = read_number_of_nodes_needed();
+    if (n <= 0) return 0;
+    pool = (struct Node**)(independent_calloc(n, sizeof(struct Node), 0);
+    if (pool == 0) die();
+    // organize into a linked list...
+    struct Node* first = pool[0];
+    for (i = 0; i < n-1; ++i)
+      pool[i]->next = pool[i+1];
+    free(pool);     // Can now free the array (or not, if it is needed later)
+    return first;
+  }
+*/
+void** dlindependent_calloc(size_t, size_t, void**);
+
+/*
+  independent_comalloc(size_t n_elements, size_t sizes[], void* chunks[]);
+
+  independent_comalloc allocates, all at once, a set of n_elements
+  chunks with sizes indicated in the "sizes" array.    It returns
+  an array of pointers to these elements, each of which can be
+  independently freed, realloc'ed etc. The elements are guaranteed to
+  be adjacently allocated (this is not guaranteed to occur with
+  multiple callocs or mallocs), which may also improve cache locality
+  in some applications.
+
+  The "chunks" argument is optional (i.e., may be null). If it is null
+  the returned array is itself dynamically allocated and should also
+  be freed when it is no longer needed. Otherwise, the chunks array
+  must be of at least n_elements in length. It is filled in with the
+  pointers to the chunks.
+
+  In either case, independent_comalloc returns this pointer array, or
+  null if the allocation failed.  If n_elements is zero and chunks is
+  null, it returns a chunk representing an array with zero elements
+  (which should be freed if not wanted).
+
+  Each element must be individually freed when it is no longer
+  needed. If you'd like to instead be able to free all at once, you
+  should instead use a single regular malloc, and assign pointers at
+  particular offsets in the aggregate space. (In this case though, you
+  cannot independently free elements.)
+
+  independent_comallac differs from independent_calloc in that each
+  element may have a different size, and also that it does not
+  automatically clear elements.
+
+  independent_comalloc can be used to speed up allocation in cases
+  where several structs or objects must always be allocated at the
+  same time.  For example:
+
+  struct Head { ... }
+  struct Foot { ... }
+
+  void send_message(char* msg) {
+    int msglen = strlen(msg);
+    size_t sizes[3] = { sizeof(struct Head), msglen, sizeof(struct Foot) };
+    void* chunks[3];
+    if (independent_comalloc(3, sizes, chunks) == 0)
+      die();
+    struct Head* head = (struct Head*)(chunks[0]);
+    char*        body = (char*)(chunks[1]);
+    struct Foot* foot = (struct Foot*)(chunks[2]);
+    // ...
+  }
+
+  In general though, independent_comalloc is worth using only for
+  larger values of n_elements. For small values, you probably won't
+  detect enough difference from series of malloc calls to bother.
+
+  Overuse of independent_comalloc can increase overall memory usage,
+  since it cannot reuse existing noncontiguous small chunks that
+  might be available for some of the elements.
+*/
+void** dlindependent_comalloc(size_t, size_t*, void**);
+
+
+/*
+  pvalloc(size_t n);
+  Equivalent to valloc(minimum-page-that-holds(n)), that is,
+  round up n to nearest pagesize.
+ */
+void*  dlpvalloc(size_t);
+
+/*
+  malloc_trim(size_t pad);
+
+  If possible, gives memory back to the system (via negative arguments
+  to sbrk) if there is unused memory at the `high' end of the malloc
+  pool or in unused MMAP segments. You can call this after freeing
+  large blocks of memory to potentially reduce the system-level memory
+  requirements of a program. However, it cannot guarantee to reduce
+  memory. Under some allocation patterns, some large free blocks of
+  memory will be locked between two used chunks, so they cannot be
+  given back to the system.
+
+  The `pad' argument to malloc_trim represents the amount of free
+  trailing space to leave untrimmed. If this argument is zero, only
+  the minimum amount of memory to maintain internal data structures
+  will be left. Non-zero arguments can be supplied to maintain enough
+  trailing space to service future expected allocations without having
+  to re-obtain memory from the system.
+
+  Malloc_trim returns 1 if it actually released any memory, else 0.
+*/
+int  dlmalloc_trim(size_t);
+
+/*
+  malloc_stats();
+  Prints on stderr the amount of space obtained from the system (both
+  via sbrk and mmap), the maximum amount (which may be more than
+  current if malloc_trim and/or munmap got called), and the current
+  number of bytes allocated via malloc (or realloc, etc) but not yet
+  freed. Note that this is the number of bytes allocated, not the
+  number requested. It will be larger than the number requested
+  because of alignment and bookkeeping overhead. Because it includes
+  alignment wastage as being in use, this figure may be greater than
+  zero even when no user-level chunks are allocated.
+
+  The reported current and maximum system memory can be inaccurate if
+  a program makes other calls to system memory allocation functions
+  (normally sbrk) outside of malloc.
+
+  malloc_stats prints only the most commonly interesting statistics.
+  More information can be obtained by calling mallinfo.
+*/
+void  dlmalloc_stats(void);
+
+#endif /* ONLY_MSPACES */
+
+/*
+  malloc_usable_size(void* p);
+
+  Returns the number of bytes you can actually use in
+  an allocated chunk, which may be more than you requested (although
+  often not) due to alignment and minimum size constraints.
+  You can use this many bytes without worrying about
+  overwriting other allocated objects. This is not a particularly great
+  programming practice. malloc_usable_size can be more useful in
+  debugging and assertions, for example:
+
+  p = malloc(n);
+  assert(malloc_usable_size(p) >= 256);
+*/
+size_t dlmalloc_usable_size(void*);
+
+
+#if MSPACES
+
+/*
+  mspace is an opaque type representing an independent
+  region of space that supports mspace_malloc, etc.
+*/
+typedef void* mspace;
+
+/*
+  create_mspace creates and returns a new independent space with the
+  given initial capacity, or, if 0, the default granularity size.  It
+  returns null if there is no system memory available to create the
+  space.  If argument locked is non-zero, the space uses a separate
+  lock to control access. The capacity of the space will grow
+  dynamically as needed to service mspace_malloc requests.  You can
+  control the sizes of incremental increases of this space by
+  compiling with a different DEFAULT_GRANULARITY or dynamically
+  setting with mallopt(M_GRANULARITY, value).
+*/
+mspace create_mspace(size_t capacity, int locked);
+
+/*
+  destroy_mspace destroys the given space, and attempts to return all
+  of its memory back to the system, returning the total number of
+  bytes freed. After destruction, the results of access to all memory
+  used by the space become undefined.
+*/
+size_t destroy_mspace(mspace msp);
+
+/*
+  create_mspace_with_base uses the memory supplied as the initial base
+  of a new mspace. Part (less than 128*sizeof(size_t) bytes) of this
+  space is used for bookkeeping, so the capacity must be at least this
+  large. (Otherwise 0 is returned.) When this initial space is
+  exhausted, additional memory will be obtained from the system.
+  Destroying this space will deallocate all additionally allocated
+  space (if possible) but not the initial base.
+*/
+mspace create_mspace_with_base(void* base, size_t capacity, int locked);
+
+/*
+  mspace_track_large_chunks controls whether requests for large chunks
+  are allocated in their own untracked mmapped regions, separate from
+  others in this mspace. By default large chunks are not tracked,
+  which reduces fragmentation. However, such chunks are not
+  necessarily released to the system upon destroy_mspace.  Enabling
+  tracking by setting to true may increase fragmentation, but avoids
+  leakage when relying on destroy_mspace to release all memory
+  allocated using this space.  The function returns the previous
+  setting.
+*/
+int mspace_track_large_chunks(mspace msp, int enable);
+
+
+/*
+  mspace_malloc behaves as malloc, but operates within
+  the given space.
+*/
+void* mspace_malloc(mspace msp, size_t bytes);
+
+/*
+  mspace_free behaves as free, but operates within
+  the given space.
+
+  If compiled with FOOTERS==1, mspace_free is not actually needed.
+  free may be called instead of mspace_free because freed chunks from
+  any space are handled by their originating spaces.
+*/
+void mspace_free(mspace msp, void* mem);
+
+/*
+  mspace_realloc behaves as realloc, but operates within
+  the given space.
+
+  If compiled with FOOTERS==1, mspace_realloc is not actually
+  needed.  realloc may be called instead of mspace_realloc because
+  realloced chunks from any space are handled by their originating
+  spaces.
+*/
+void* mspace_realloc(mspace msp, void* mem, size_t newsize);
+
+/*
+  mspace_calloc behaves as calloc, but operates within
+  the given space.
+*/
+void* mspace_calloc(mspace msp, size_t n_elements, size_t elem_size);
+
+/*
+  mspace_memalign behaves as memalign, but operates within
+  the given space.
+*/
+void* mspace_memalign(mspace msp, size_t alignment, size_t bytes);
+
+/*
+  mspace_independent_calloc behaves as independent_calloc, but
+  operates within the given space.
+*/
+void** mspace_independent_calloc(mspace msp, size_t n_elements,
+                                 size_t elem_size, void* chunks[]);
+
+/*
+  mspace_independent_comalloc behaves as independent_comalloc, but
+  operates within the given space.
+*/
+void** mspace_independent_comalloc(mspace msp, size_t n_elements,
+                                   size_t sizes[], void* chunks[]);
+
+/*
+  mspace_footprint() returns the number of bytes obtained from the
+  system for this space.
+*/
+size_t mspace_footprint(mspace msp);
+
+/*
+  mspace_max_footprint() returns the peak number of bytes obtained from the
+  system for this space.
+*/
+size_t mspace_max_footprint(mspace msp);
+
+
+#if !NO_MALLINFO
+/*
+  mspace_mallinfo behaves as mallinfo, but reports properties of
+  the given space.
+*/
+struct mallinfo mspace_mallinfo(mspace msp);
+#endif /* NO_MALLINFO */
+
+/*
+  malloc_usable_size(void* p) behaves the same as malloc_usable_size;
+*/
+  size_t mspace_usable_size(void* mem);
+
+/*
+  mspace_malloc_stats behaves as malloc_stats, but reports
+  properties of the given space.
+*/
+void mspace_malloc_stats(mspace msp);
+
+/*
+  mspace_trim behaves as malloc_trim, but
+  operates within the given space.
+*/
+int mspace_trim(mspace msp, size_t pad);
+
+/*
+  An alias for mallopt.
+*/
+int mspace_mallopt(int, int);
+
+#endif /* MSPACES */
+
+#ifdef __cplusplus
+}  /* end of extern "C" */
+#endif /* __cplusplus */
+
+/*
+  ========================================================================
+  To make a fully customizable malloc.h header file, cut everything
+  above this line, put into file malloc.h, edit to suit, and #include it
+  on the next line, as well as in programs that use this malloc.
+  ========================================================================
+*/
+
+/* #include "malloc.h" */
+
+/*------------------------------ internal #includes ---------------------- */
+
+#ifdef WIN32
+#pragma warning( disable : 4146 ) /* no "unsigned" warnings */
+#endif /* WIN32 */
+
+#include <stdio.h>       /* for printing in malloc_stats */
+
+#ifndef LACKS_ERRNO_H
+#include <errno.h>       /* for MALLOC_FAILURE_ACTION */
+#endif /* LACKS_ERRNO_H */
+#if FOOTERS || DEBUG
+#include <time.h>        /* for magic initialization */
+#endif /* FOOTERS */
+#ifndef LACKS_STDLIB_H
+#include <stdlib.h>      /* for abort() */
+#endif /* LACKS_STDLIB_H */
+#ifdef DEBUG
+#if ABORT_ON_ASSERT_FAILURE
+#undef assert
+#define assert(x) if(!(x)) ABORT
+#else /* ABORT_ON_ASSERT_FAILURE */
+#include <assert.h>
+#endif /* ABORT_ON_ASSERT_FAILURE */
+#else  /* DEBUG */
+#ifndef assert
+#define assert(x)
+#endif
+#define DEBUG 0
+#endif /* DEBUG */
+#ifndef LACKS_STRING_H
+#include <string.h>      /* for memset etc */
+#endif  /* LACKS_STRING_H */
+#if USE_BUILTIN_FFS
+#ifndef LACKS_STRINGS_H
+#include <strings.h>     /* for ffs */
+#endif /* LACKS_STRINGS_H */
+#endif /* USE_BUILTIN_FFS */
+#if HAVE_MMAP
+#ifndef LACKS_SYS_MMAN_H
+/* On some versions of linux, mremap decl in mman.h needs __USE_GNU set */
+#if (defined(linux) && !defined(__USE_GNU))
+#define __USE_GNU 1
+#include <sys/mman.h>    /* for mmap */
+#undef __USE_GNU
+#else
+#include <sys/mman.h>    /* for mmap */
+#endif /* linux */
+#endif /* LACKS_SYS_MMAN_H */
+#ifndef LACKS_FCNTL_H
+#include <fcntl.h>
+#endif /* LACKS_FCNTL_H */
+#endif /* HAVE_MMAP */
+#ifndef LACKS_UNISTD_H
+#include <unistd.h>     /* for sbrk, sysconf */
+#else /* LACKS_UNISTD_H */
+#if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
+extern void*     sbrk(ptrdiff_t);
+#endif /* FreeBSD etc */
+#endif /* LACKS_UNISTD_H */
+
+/* Declarations for locking */
+#if USE_LOCKS
+#ifndef WIN32
+#include <pthread.h>
+#if defined (__SVR4) && defined (__sun)  /* solaris */
+#include <thread.h>
+#endif /* solaris */
+#else
+#ifndef _M_AMD64
+/* These are already defined on AMD64 builds */
+#ifdef __cplusplus
+extern "C" {
+#endif /* __cplusplus */
+LONG __cdecl _InterlockedCompareExchange(LONG volatile *Dest, LONG Exchange, LONG Comp);
+LONG __cdecl _InterlockedExchange(LONG volatile *Target, LONG Value);
+#ifdef __cplusplus
+}
+#endif /* __cplusplus */
+#endif /* _M_AMD64 */
+#pragma intrinsic (_InterlockedCompareExchange)
+#pragma intrinsic (_InterlockedExchange)
+#define interlockedcompareexchange _InterlockedCompareExchange
+#define interlockedexchange _InterlockedExchange
+#endif /* Win32 */
+#endif /* USE_LOCKS */
+
+/* Declarations for bit scanning on win32 */
+#if defined(_MSC_VER) && _MSC_VER>=1300
+#ifndef BitScanForward	/* Try to avoid pulling in WinNT.h */
+#ifdef __cplusplus
+extern "C" {
+#endif /* __cplusplus */
+unsigned char _BitScanForward(unsigned long *index, unsigned long mask);
+unsigned char _BitScanReverse(unsigned long *index, unsigned long mask);
+#ifdef __cplusplus
+}
+#endif /* __cplusplus */
+
+#define BitScanForward _BitScanForward
+#define BitScanReverse _BitScanReverse
+#pragma intrinsic(_BitScanForward)
+#pragma intrinsic(_BitScanReverse)
+#endif /* BitScanForward */
+#endif /* defined(_MSC_VER) && _MSC_VER>=1300 */
+
+#ifndef WIN32
+#ifndef malloc_getpagesize
+#  ifdef _SC_PAGESIZE         /* some SVR4 systems omit an underscore */
+#    ifndef _SC_PAGE_SIZE
+#      define _SC_PAGE_SIZE _SC_PAGESIZE
+#    endif
+#  endif
+#  ifdef _SC_PAGE_SIZE
+#    define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
+#  else
+#    if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
+       extern size_t getpagesize();
+#      define malloc_getpagesize getpagesize()
+#    else
+#      ifdef WIN32 /* use supplied emulation of getpagesize */
+#        define malloc_getpagesize getpagesize()
+#      else
+#        ifndef LACKS_SYS_PARAM_H
+#          include <sys/param.h>
+#        endif
+#        ifdef EXEC_PAGESIZE
+#          define malloc_getpagesize EXEC_PAGESIZE
+#        else
+#          ifdef NBPG
+#            ifndef CLSIZE
+#              define malloc_getpagesize NBPG
+#            else
+#              define malloc_getpagesize (NBPG * CLSIZE)
+#            endif
+#          else
+#            ifdef NBPC
+#              define malloc_getpagesize NBPC
+#            else
+#              ifdef PAGESIZE
+#                define malloc_getpagesize PAGESIZE
+#              else /* just guess */
+#                define malloc_getpagesize ((size_t)4096U)
+#              endif
+#            endif
+#          endif
+#        endif
+#      endif
+#    endif
+#  endif
+#endif
+#endif
+
+
+
+/* ------------------- size_t and alignment properties -------------------- */
+
+/* The byte and bit size of a size_t */
+#define SIZE_T_SIZE         (sizeof(size_t))
+#define SIZE_T_BITSIZE      (sizeof(size_t) << 3)
+
+/* Some constants coerced to size_t */
+/* Annoying but necessary to avoid errors on some platforms */
+#define SIZE_T_ZERO         ((size_t)0)
+#define SIZE_T_ONE          ((size_t)1)
+#define SIZE_T_TWO          ((size_t)2)
+#define SIZE_T_FOUR         ((size_t)4)
+#define TWO_SIZE_T_SIZES    (SIZE_T_SIZE<<1)
+#define FOUR_SIZE_T_SIZES   (SIZE_T_SIZE<<2)
+#define SIX_SIZE_T_SIZES    (FOUR_SIZE_T_SIZES+TWO_SIZE_T_SIZES)
+#define HALF_MAX_SIZE_T     (MAX_SIZE_T / 2U)
+
+/* The bit mask value corresponding to MALLOC_ALIGNMENT */
+#define CHUNK_ALIGN_MASK    (MALLOC_ALIGNMENT - SIZE_T_ONE)
+
+/* True if address a has acceptable alignment */
+#define is_aligned(A)       (((size_t)((A)) & (CHUNK_ALIGN_MASK)) == 0)
+
+/* the number of bytes to offset an address to align it */
+#define align_offset(A)\
+ ((((size_t)(A) & CHUNK_ALIGN_MASK) == 0)? 0 :\
+  ((MALLOC_ALIGNMENT - ((size_t)(A) & CHUNK_ALIGN_MASK)) & CHUNK_ALIGN_MASK))
+
+/*
+  malloc_params holds global properties, including those that can be
+  dynamically set using mallopt. There is a single instance, mparams,
+  initialized in init_mparams. Note that the non-zeroness of "magic"
+  also serves as an initialization flag.
+*/
+typedef unsigned int flag_t;
+struct malloc_params {
+  volatile size_t magic;
+  size_t page_size;
+  size_t granularity;
+  size_t mmap_threshold;
+  size_t trim_threshold;
+  flag_t default_mflags;
+};
+
+static struct malloc_params mparams;
+
+/* Ensure mparams initialized */
+#define ensure_initialization() (void)(mparams.magic != 0 || init_mparams())
+
+/* -------------------------- MMAP preliminaries ------------------------- */
+
+/*
+   If HAVE_MORECORE or HAVE_MMAP are false, we just define calls and
+   checks to fail so compiler optimizer can delete code rather than
+   using so many "#if"s.
+*/
+
+
+/* MORECORE and MMAP must return MFAIL on failure */
+#define MFAIL                ((void*)(MAX_SIZE_T))
+#define CMFAIL               ((char*)(MFAIL)) /* defined for convenience */
+
+#if HAVE_MMAP
+
+#ifndef WIN32
+#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
+#define MAP_ANONYMOUS        MAP_ANON
+#endif /* MAP_ANON */
+#ifdef DEFAULT_GRANULARITY_ALIGNED
+#define MMAP_IMPL mmap_aligned
+static void* lastAlignedmmap; /* Used as a hint */
+static void* mmap_aligned(void *start, size_t length, int prot, int flags, int fd, off_t offset) {
+  void* baseaddress = 0;
+  void* ptr = 0;
+  if(!start) {
+    baseaddress = lastAlignedmmap;
+    for(;;) {
+      if(baseaddress) flags|=MAP_FIXED;
+      ptr = mmap(baseaddress, length, prot, flags, fd, offset);
+      if(!ptr)
+        baseaddress = (void*)((size_t)baseaddress + mparams.granularity);
+      else if((size_t)ptr & (mparams.granularity - SIZE_T_ONE)) {
+        munmap(ptr, length);
+        baseaddress = (void*)(((size_t)ptr + mparams.granularity) & ~(mparams.granularity - SIZE_T_ONE));
+      }
+      else break;
+    }
+  }
+  else ptr = mmap(start, length, prot, flags, fd, offset);
+  if(ptr) lastAlignedmmap = (void*)((size_t) ptr + mparams.granularity);
+  return ptr;
+}
+#else
+#define MMAP_IMPL mmap
+#endif /* DEFAULT_GRANULARITY_ALIGNED */
+#define MUNMAP_DEFAULT(a, s)  munmap((a), (s))
+#define MMAP_PROT            (PROT_READ|PROT_WRITE)
+#ifdef MAP_ANONYMOUS
+#define MMAP_FLAGS           (MAP_PRIVATE|MAP_ANONYMOUS)
+#define MMAP_DEFAULT(s)       MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, -1, 0)
+#else /* MAP_ANONYMOUS */
+/*
+   Nearly all versions of mmap support MAP_ANONYMOUS, so the following
+   is unlikely to be needed, but is supplied just in case.
+*/
+#define MMAP_FLAGS           (MAP_PRIVATE)
+static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */
+#define MMAP_DEFAULT(s) ((dev_zero_fd < 0) ? \
+           (dev_zero_fd = open("/dev/zero", O_RDWR), \
+            MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0)) : \
+            MMAP_IMPL(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0))
+#endif /* MAP_ANONYMOUS */
+
+#define DIRECT_MMAP_DEFAULT(s) MMAP_DEFAULT(s)
+
+#else /* WIN32 */
+
+/* Win32 MMAP via VirtualAlloc */
+#ifdef DEFAULT_GRANULARITY_ALIGNED
+static void* lastWin32mmap; /* Used as a hint */
+#endif /* DEFAULT_GRANULARITY_ALIGNED */
+#ifdef ENABLE_LARGE_PAGES
+static int largepagesavailable = 1;
+#endif /* ENABLE_LARGE_PAGES */
+static FORCEINLINE void* win32mmap(size_t size) {
+  void* baseaddress = 0;
+  void* ptr = 0;
+#ifdef ENABLE_LARGE_PAGES
+  /* Note that large pages are *always* allocated on a large page boundary.
+  If however granularity is small then don't waste a kernel call if size
+  isn't around the size of a large page */
+  if(largepagesavailable && size >= 1*1024*1024) {
+    ptr = VirtualAlloc(baseaddress, size, MEM_RESERVE|MEM_COMMIT|MEM_LARGE_PAGES, PAGE_READWRITE);
+    if(!ptr && ERROR_PRIVILEGE_NOT_HELD==GetLastError()) largepagesavailable=0;
+  }
+#endif
+  if(!ptr) {
+#ifdef DEFAULT_GRANULARITY_ALIGNED
+    /* We try to avoid overhead by speculatively reserving at aligned
+    addresses until we succeed */
+    baseaddress = lastWin32mmap;
+    for(;;) {
+      void* reserveaddr = VirtualAlloc(baseaddress, size, MEM_RESERVE, PAGE_READWRITE);
+      if(!reserveaddr)
+        baseaddress = (void*)((size_t)baseaddress + mparams.granularity);
+      else if((size_t)reserveaddr & (mparams.granularity - SIZE_T_ONE)) {
+        VirtualFree(reserveaddr, 0, MEM_RELEASE);
+        baseaddress = (void*)(((size_t)reserveaddr + mparams.granularity) & ~(mparams.granularity - SIZE_T_ONE));
+      }
+      else break;
+    }
+#endif
+    if(!ptr) ptr = VirtualAlloc(baseaddress, size, baseaddress ? MEM_COMMIT : MEM_RESERVE|MEM_COMMIT, PAGE_READWRITE);
+#if DEBUG
+    if(lastWin32mmap && ptr!=lastWin32mmap) printf("Non-contiguous VirtualAlloc between %p and %p\n", ptr, lastWin32mmap);
+#endif
+#ifdef DEFAULT_GRANULARITY_ALIGNED
+    if(ptr) lastWin32mmap = (void*)((size_t) ptr + mparams.granularity);
+#endif
+  }
+#if DEBUG
+#ifdef ENABLE_LARGE_PAGES
+  printf("VirtualAlloc returns %p size %u. LargePagesAvailable=%d\n", ptr, size, largepagesavailable);
+#else
+  printf("VirtualAlloc returns %p size %u\n", ptr, size);
+#endif
+#endif
+  return (ptr != 0)? ptr: MFAIL;
+}
+
+/* For direct MMAP, use MEM_TOP_DOWN to minimize interference */
+static FORCEINLINE void* win32direct_mmap(size_t size) {
+  void* ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT|MEM_TOP_DOWN,
+                           PAGE_READWRITE);
+  return (ptr != 0)? ptr: MFAIL;
+}
+
+/* This function supports releasing coalesed segments */
+static FORCEINLINE int win32munmap(void* ptr, size_t size) {
+  MEMORY_BASIC_INFORMATION minfo;
+  char* cptr = (char*)ptr;
+  while (size) {
+    if (VirtualQuery(cptr, &minfo, sizeof(minfo)) == 0)
+      return -1;
+    if (minfo.BaseAddress != cptr || minfo.AllocationBase != cptr ||
+        minfo.State != MEM_COMMIT || minfo.RegionSize > size)
+      return -1;
+    if (VirtualFree(cptr, 0, MEM_RELEASE) == 0)
+      return -1;
+    cptr += minfo.RegionSize;
+    size -= minfo.RegionSize;
+  }
+  return 0;
+}
+
+#define MMAP_DEFAULT(s)             win32mmap(s)
+#define MUNMAP_DEFAULT(a, s)        win32munmap((a), (s))
+#define DIRECT_MMAP_DEFAULT(s)      win32direct_mmap(s)
+#endif /* WIN32 */
+#endif /* HAVE_MMAP */
+
+#if HAVE_MREMAP
+#ifndef WIN32
+#define MREMAP_DEFAULT(addr, osz, nsz, mv) mremap((addr), (osz), (nsz), (mv))
+#endif /* WIN32 */
+#endif /* HAVE_MREMAP */
+
+
+/**
+ * Define CALL_MORECORE
+ */
+#if HAVE_MORECORE
+    #ifdef MORECORE
+        #define CALL_MORECORE(S)    MORECORE(S)
+    #else  /* MORECORE */
+        #define CALL_MORECORE(S)    MORECORE_DEFAULT(S)
+    #endif /* MORECORE */
+#else  /* HAVE_MORECORE */
+    #define CALL_MORECORE(S)        MFAIL
+#endif /* HAVE_MORECORE */
+
+/**
+ * Define CALL_MMAP/CALL_MUNMAP/CALL_DIRECT_MMAP
+ */
+#if HAVE_MMAP
+    #define USE_MMAP_BIT            (SIZE_T_ONE)
+
+    #ifdef MMAP
+        #define CALL_MMAP(s)        MMAP(s)
+    #else /* MMAP */
+        #define CALL_MMAP(s)        MMAP_DEFAULT(s)
+    #endif /* MMAP */
+    #ifdef MUNMAP
+        #define CALL_MUNMAP(a, s)   MUNMAP((a), (s))
+    #else /* MUNMAP */
+        #define CALL_MUNMAP(a, s)   MUNMAP_DEFAULT((a), (s))
+    #endif /* MUNMAP */
+    #ifdef DIRECT_MMAP
+        #define CALL_DIRECT_MMAP(s) DIRECT_MMAP(s)
+    #else /* DIRECT_MMAP */
+        #define CALL_DIRECT_MMAP(s) DIRECT_MMAP_DEFAULT(s)
+    #endif /* DIRECT_MMAP */
+#else  /* HAVE_MMAP */
+    #define USE_MMAP_BIT            (SIZE_T_ZERO)
+
+    #define MMAP(s)                 MFAIL
+    #define MUNMAP(a, s)            (-1)
+    #define DIRECT_MMAP(s)          MFAIL
+    #define CALL_DIRECT_MMAP(s)     DIRECT_MMAP(s)
+    #define CALL_MMAP(s)            MMAP(s)
+    #define CALL_MUNMAP(a, s)       MUNMAP((a), (s))
+#endif /* HAVE_MMAP */
+
+/**
+ * Define CALL_MREMAP
+ */
+#if HAVE_MMAP && HAVE_MREMAP
+    #ifdef MREMAP
+        #define CALL_MREMAP(addr, osz, nsz, mv) MREMAP((addr), (osz), (nsz), (mv))
+    #else /* MREMAP */
+        #define CALL_MREMAP(addr, osz, nsz, mv) MREMAP_DEFAULT((addr), (osz), (nsz), (mv))
+    #endif /* MREMAP */
+#else  /* HAVE_MMAP && HAVE_MREMAP */
+    #define CALL_MREMAP(addr, osz, nsz, mv)     MFAIL
+#endif /* HAVE_MMAP && HAVE_MREMAP */
+
+/* mstate bit set if continguous morecore disabled or failed */
+#define USE_NONCONTIGUOUS_BIT (4U)
+
+/* segment bit set in create_mspace_with_base */
+#define EXTERN_BIT            (8U)
+
+
+/* --------------------------- Lock preliminaries ------------------------ */
+
+/*
+  When locks are defined, there is one global lock, plus
+  one per-mspace lock.
+
+  The global lock_ensures that mparams.magic and other unique
+  mparams values are initialized only once. It also protects
+  sequences of calls to MORECORE.  In many cases sys_alloc requires
+  two calls, that should not be interleaved with calls by other
+  threads.  This does not protect against direct calls to MORECORE
+  by other threads not using this lock, so there is still code to
+  cope the best we can on interference.
+
+  Per-mspace locks surround calls to malloc, free, etc.  To enable use
+  in layered extensions, per-mspace locks are reentrant.
+
+  Because lock-protected regions generally have bounded times, it is
+  OK to use the supplied simple spinlocks in the custom versions for
+  x86. Spinlocks are likely to improve performance for lightly
+  contended applications, but worsen performance under heavy
+  contention.
+
+  If USE_LOCKS is > 1, the definitions of lock routines here are
+  bypassed, in which case you will need to define the type MLOCK_T,
+  and at least INITIAL_LOCK, ACQUIRE_LOCK, RELEASE_LOCK and possibly
+  TRY_LOCK (which is not used in this malloc, but commonly needed in
+  extensions.)  You must also declare a
+    static MLOCK_T malloc_global_mutex = { initialization values };.
+
+*/
+
+#if USE_LOCKS == 1
+
+#if USE_SPIN_LOCKS && SPIN_LOCKS_AVAILABLE
+#ifndef WIN32
+
+/* Custom pthread-style spin locks on x86 and x64 for gcc */
+struct pthread_mlock_t {
+  volatile unsigned int l;
+  char cachelinepadding[64];
+  unsigned int c;
+  pthread_t threadid;
+};
+#define MLOCK_T               struct pthread_mlock_t
+#define CURRENT_THREAD        pthread_self()
+#define INITIAL_LOCK(sl)      ((sl)->threadid = 0, (sl)->l = (sl)->c = 0, 0)
+#define ACQUIRE_LOCK(sl)      pthread_acquire_lock(sl)
+#define RELEASE_LOCK(sl)      pthread_release_lock(sl)
+#define TRY_LOCK(sl)          pthread_try_lock(sl)
+#define SPINS_PER_YIELD       63
+
+static MLOCK_T malloc_global_mutex = { 0, "", 0, 0};
+
+static FORCEINLINE int pthread_acquire_lock (MLOCK_T *sl) {
+  int spins = 0;
+  volatile unsigned int* lp = &sl->l;
+  for (;;) {
+    if (*lp != 0) {
+      if (sl->threadid == CURRENT_THREAD) {
+        ++sl->c;
+        return 0;
+      }
+    }
+    else {
+      /* place args to cmpxchgl in locals to evade oddities in some gccs */
+      int cmp = 0;
+      int val = 1;
+      int ret;
+      __asm__ __volatile__  ("lock; cmpxchgl %1, %2"
+                             : "=a" (ret)
+                             : "r" (val), "m" (*(lp)), "0"(cmp)
+                             : "memory", "cc");
+      if (!ret) {
+        assert(!sl->threadid);
+        sl->threadid = CURRENT_THREAD;
+        sl->c = 1;
+        return 0;
+      }
+    }
+    if ((++spins & SPINS_PER_YIELD) == 0) {
+#if defined (__SVR4) && defined (__sun) /* solaris */
+      thr_yield();
+#else
+#if defined(__linux__) || defined(__FreeBSD__) || defined(__APPLE__)
+      sched_yield();
+#else  /* no-op yield on unknown systems */
+      ;
+#endif /* __linux__ || __FreeBSD__ || __APPLE__ */
+#endif /* solaris */
+    }
+  }
+}
+
+static FORCEINLINE void pthread_release_lock (MLOCK_T *sl) {
+  volatile unsigned int* lp = &sl->l;
+  assert(*lp != 0);
+  assert(sl->threadid == CURRENT_THREAD);
+  if (--sl->c == 0) {
+    sl->threadid = 0;
+    int prev = 0;
+    int ret;
+    __asm__ __volatile__ ("lock; xchgl %0, %1"
+                          : "=r" (ret)
+                          : "m" (*(lp)), "0"(prev)
+                          : "memory");
+  }
+}
+
+static FORCEINLINE int pthread_try_lock (MLOCK_T *sl) {
+  volatile unsigned int* lp = &sl->l;
+  if (*lp != 0) {
+    if (sl->threadid == CURRENT_THREAD) {
+      ++sl->c;
+      return 1;
+    }
+  }
+  else {
+    int cmp = 0;
+    int val = 1;
+    int ret;
+    __asm__ __volatile__  ("lock; cmpxchgl %1, %2"
+                           : "=a" (ret)
+                           : "r" (val), "m" (*(lp)), "0"(cmp)
+                           : "memory", "cc");
+    if (!ret) {
+      assert(!sl->threadid);
+      sl->threadid = CURRENT_THREAD;
+      sl->c = 1;
+      return 1;
+    }
+  }
+  return 0;
+}
+
+
+#else /* WIN32 */
+/* Custom win32-style spin locks on x86 and x64 for MSC */
+struct win32_mlock_t {
+  volatile long l;
+  char cachelinepadding[64];
+  unsigned int c;
+  long threadid;
+};
+
+#define MLOCK_T               struct win32_mlock_t
+#define CURRENT_THREAD        ((long)GetCurrentThreadId())
+#define INITIAL_LOCK(sl)      ((sl)->threadid = 0, (sl)->l = (sl)->c = 0, 0)
+#define ACQUIRE_LOCK(sl)      win32_acquire_lock(sl)
+#define RELEASE_LOCK(sl)      win32_release_lock(sl)
+#define TRY_LOCK(sl)          win32_try_lock(sl)
+#define SPINS_PER_YIELD       63
+
+static MLOCK_T malloc_global_mutex = { 0, 0, 0};
+
+static FORCEINLINE int win32_acquire_lock (MLOCK_T *sl) {
+  int spins = 0;
+  for (;;) {
+    if (sl->l != 0) {
+      if (sl->threadid == CURRENT_THREAD) {
+        ++sl->c;
+        return 0;
+      }
+    }
+    else {
+      if (!interlockedexchange(&sl->l, 1)) {
+        assert(!sl->threadid);
+        sl->threadid = CURRENT_THREAD;
+        sl->c = 1;
+        return 0;
+      }
+    }
+    if ((++spins & SPINS_PER_YIELD) == 0)
+      SleepEx(0, FALSE);
+  }
+}
+
+static FORCEINLINE void win32_release_lock (MLOCK_T *sl) {
+  assert(sl->threadid == CURRENT_THREAD);
+  assert(sl->l != 0);
+  if (--sl->c == 0) {
+    sl->threadid = 0;
+    interlockedexchange (&sl->l, 0);
+  }
+}
+
+static FORCEINLINE int win32_try_lock (MLOCK_T *sl) {
+  if (sl->l != 0) {
+    if (sl->threadid == CURRENT_THREAD) {
+      ++sl->c;
+      return 1;
+    }
+  }
+  else {
+    if (!interlockedexchange(&sl->l, 1)){
+      assert(!sl->threadid);
+      sl->threadid = CURRENT_THREAD;
+      sl->c = 1;
+      return 1;
+    }
+  }
+  return 0;
+}
+
+#endif /* WIN32 */
+#else /* USE_SPIN_LOCKS */
+
+#ifndef WIN32
+/* pthreads-based locks */
+
+#define MLOCK_T               pthread_mutex_t
+#define CURRENT_THREAD        pthread_self()
+#define INITIAL_LOCK(sl)      pthread_init_lock(sl)
+#define ACQUIRE_LOCK(sl)      pthread_mutex_lock(sl)
+#define RELEASE_LOCK(sl)      pthread_mutex_unlock(sl)
+#define TRY_LOCK(sl)          (!pthread_mutex_trylock(sl))
+
+static MLOCK_T malloc_global_mutex = PTHREAD_MUTEX_INITIALIZER;
+
+/* Cope with old-style linux recursive lock initialization by adding */
+/* skipped internal declaration from pthread.h */
+#ifdef linux
+#ifndef PTHREAD_MUTEX_RECURSIVE
+extern int pthread_mutexattr_setkind_np __P ((pthread_mutexattr_t *__attr,
+					   int __kind));
+#define PTHREAD_MUTEX_RECURSIVE PTHREAD_MUTEX_RECURSIVE_NP
+#define pthread_mutexattr_settype(x,y) pthread_mutexattr_setkind_np(x,y)
+#endif
+#endif
+
+static int pthread_init_lock (MLOCK_T *sl) {
+  pthread_mutexattr_t attr;
+  if (pthread_mutexattr_init(&attr)) return 1;
+  if (pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_RECURSIVE)) return 1;
+  if (pthread_mutex_init(sl, &attr)) return 1;
+  if (pthread_mutexattr_destroy(&attr)) return 1;
+  return 0;
+}
+
+#else /* WIN32 */
+/* Win32 critical sections */
+#define MLOCK_T               CRITICAL_SECTION
+#define CURRENT_THREAD        GetCurrentThreadId()
+#define INITIAL_LOCK(s)       (!InitializeCriticalSectionAndSpinCount((s), 0x80000000|4000))
+#define ACQUIRE_LOCK(s)       (EnterCriticalSection(sl), 0)
+#define RELEASE_LOCK(s)       LeaveCriticalSection(sl)
+#define TRY_LOCK(s)           TryEnterCriticalSection(sl)
+#define NEED_GLOBAL_LOCK_INIT
+
+static MLOCK_T malloc_global_mutex;
+static volatile long malloc_global_mutex_status;
+
+/* Use spin loop to initialize global lock */
+static void init_malloc_global_mutex() {
+  for (;;) {
+    long stat = malloc_global_mutex_status;
+    if (stat > 0)
+      return;
+    /* transition to < 0 while initializing, then to > 0) */
+    if (stat == 0 &&
+        interlockedcompareexchange(&malloc_global_mutex_status, -1, 0) == 0) {
+      InitializeCriticalSection(&malloc_global_mutex);
+      interlockedexchange(&malloc_global_mutex_status,1);
+      return;
+    }
+    SleepEx(0, FALSE);
+  }
+}
+
+#endif /* WIN32 */
+#endif /* USE_SPIN_LOCKS */
+#endif /* USE_LOCKS == 1 */
+
+/* -----------------------  User-defined locks ------------------------ */
+
+#if USE_LOCKS > 1
+/* Define your own lock implementation here */
+/* #define INITIAL_LOCK(sl)  ... */
+/* #define ACQUIRE_LOCK(sl)  ... */
+/* #define RELEASE_LOCK(sl)  ... */
+/* #define TRY_LOCK(sl) ... */
+/* static MLOCK_T malloc_global_mutex = ... */
+#endif /* USE_LOCKS > 1 */
+
+/* -----------------------  Lock-based state ------------------------ */
+
+#if USE_LOCKS
+#define USE_LOCK_BIT               (2U)
+#else  /* USE_LOCKS */
+#define USE_LOCK_BIT               (0U)
+#define INITIAL_LOCK(l)
+#endif /* USE_LOCKS */
+
+#if USE_LOCKS
+#ifndef ACQUIRE_MALLOC_GLOBAL_LOCK
+#define ACQUIRE_MALLOC_GLOBAL_LOCK()  ACQUIRE_LOCK(&malloc_global_mutex);
+#endif
+#ifndef RELEASE_MALLOC_GLOBAL_LOCK
+#define RELEASE_MALLOC_GLOBAL_LOCK()  RELEASE_LOCK(&malloc_global_mutex);
+#endif
+#else  /* USE_LOCKS */
+#define ACQUIRE_MALLOC_GLOBAL_LOCK()
+#define RELEASE_MALLOC_GLOBAL_LOCK()
+#endif /* USE_LOCKS */
+
+
+/* -----------------------  Chunk representations ------------------------ */
+
+/*
+  (The following includes lightly edited explanations by Colin Plumb.)
+
+  The malloc_chunk declaration below is misleading (but accurate and
+  necessary).  It declares a "view" into memory allowing access to
+  necessary fields at known offsets from a given base.
+
+  Chunks of memory are maintained using a `boundary tag' method as
+  originally described by Knuth.  (See the paper by Paul Wilson
+  ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a survey of such
+  techniques.)  Sizes of free chunks are stored both in the front of
+  each chunk and at the end.  This makes consolidating fragmented
+  chunks into bigger chunks fast.  The head fields also hold bits
+  representing whether chunks are free or in use.
+
+  Here are some pictures to make it clearer.  They are "exploded" to
+  show that the state of a chunk can be thought of as extending from
+  the high 31 bits of the head field of its header through the
+  prev_foot and PINUSE_BIT bit of the following chunk header.
+
+  A chunk that's in use looks like:
+
+   chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+           | Size of previous chunk (if P = 0)                             |
+           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|
+         | Size of this chunk                                         1| +-+
+   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         |                                                               |
+         +-                                                             -+
+         |                                                               |
+         +-                                                             -+
+         |                                                               :
+         +-      size - sizeof(size_t) available payload bytes          -+
+         :                                                               |
+ chunk-> +-                                                             -+
+         |                                                               |
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|
+       | Size of next chunk (may or may not be in use)               | +-+
+ mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+
+    And if it's free, it looks like this:
+
+   chunk-> +-                                                             -+
+           | User payload (must be in use, or we would have merged!)       |
+           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|
+         | Size of this chunk                                         0| +-+
+   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         | Next pointer                                                  |
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         | Prev pointer                                                  |
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         |                                                               :
+         +-      size - sizeof(struct chunk) unused bytes               -+
+         :                                                               |
+ chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+         | Size of this chunk                                            |
+         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0|
+       | Size of next chunk (must be in use, or we would have merged)| +-+
+ mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+       |                                                               :
+       +- User payload                                                -+
+       :                                                               |
+       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+                                                                     |0|
+                                                                     +-+
+  Note that since we always merge adjacent free chunks, the chunks
+  adjacent to a free chunk must be in use.
+
+  Given a pointer to a chunk (which can be derived trivially from the
+  payload pointer) we can, in O(1) time, find out whether the adjacent
+  chunks are free, and if so, unlink them from the lists that they
+  are on and merge them with the current chunk.
+
+  Chunks always begin on even word boundaries, so the mem portion
+  (which is returned to the user) is also on an even word boundary, and
+  thus at least double-word aligned.
+
+  The P (PINUSE_BIT) bit, stored in the unused low-order bit of the
+  chunk size (which is always a multiple of two words), is an in-use
+  bit for the *previous* chunk.  If that bit is *clear*, then the
+  word before the current chunk size contains the previous chunk
+  size, and can be used to find the front of the previous chunk.
+  The very first chunk allocated always has this bit set, preventing
+  access to non-existent (or non-owned) memory. If pinuse is set for
+  any given chunk, then you CANNOT determine the size of the
+  previous chunk, and might even get a memory addressing fault when
+  trying to do so.
+
+  The C (CINUSE_BIT) bit, stored in the unused second-lowest bit of
+  the chunk size redundantly records whether the current chunk is
+  inuse (unless the chunk is mmapped). This redundancy enables usage
+  checks within free and realloc, and reduces indirection when freeing
+  and consolidating chunks.
+
+  Each freshly allocated chunk must have both cinuse and pinuse set.
+  That is, each allocated chunk borders either a previously allocated
+  and still in-use chunk, or the base of its memory arena. This is
+  ensured by making all allocations from the the `lowest' part of any
+  found chunk.  Further, no free chunk physically borders another one,
+  so each free chunk is known to be preceded and followed by either
+  inuse chunks or the ends of memory.
+
+  Note that the `foot' of the current chunk is actually represented
+  as the prev_foot of the NEXT chunk. This makes it easier to
+  deal with alignments etc but can be very confusing when trying
+  to extend or adapt this code.
+
+  The exceptions to all this are
+
+     1. The special chunk `top' is the top-most available chunk (i.e.,
+        the one bordering the end of available memory). It is treated
+        specially.  Top is never included in any bin, is used only if
+        no other chunk is available, and is released back to the
+        system if it is very large (see M_TRIM_THRESHOLD).  In effect,
+        the top chunk is treated as larger (and thus less well
+        fitting) than any other available chunk.  The top chunk
+        doesn't update its trailing size field since there is no next
+        contiguous chunk that would have to index off it. However,
+        space is still allocated for it (TOP_FOOT_SIZE) to enable
+        separation or merging when space is extended.
+
+     3. Chunks allocated via mmap, have both cinuse and pinuse bits
+        cleared in their head fields.  Because they are allocated
+        one-by-one, each must carry its own prev_foot field, which is
+        also used to hold the offset this chunk has within its mmapped
+        region, which is needed to preserve alignment. Each mmapped
+        chunk is trailed by the first two fields of a fake next-chunk
+        for sake of usage checks.
+
+*/
+
+struct malloc_chunk {
+  size_t               prev_foot;  /* Size of previous chunk (if free).  */
+  size_t               head;       /* Size and inuse bits. */
+  struct malloc_chunk* fd;         /* double links -- used only if free. */
+  struct malloc_chunk* bk;
+};
+
+typedef struct malloc_chunk  mchunk;
+typedef struct malloc_chunk* mchunkptr;
+typedef struct malloc_chunk* sbinptr;  /* The type of bins of chunks */
+typedef unsigned int bindex_t;         /* Described below */
+typedef unsigned int binmap_t;         /* Described below */
+
+/* ------------------- Chunks sizes and alignments ----------------------- */
+
+#define MCHUNK_SIZE         (sizeof(mchunk))
+
+#if FOOTERS
+#define CHUNK_OVERHEAD      (TWO_SIZE_T_SIZES)
+#else /* FOOTERS */
+#define CHUNK_OVERHEAD      (SIZE_T_SIZE)
+#endif /* FOOTERS */
+
+/* MMapped chunks need a second word of overhead ... */
+#define MMAP_CHUNK_OVERHEAD (TWO_SIZE_T_SIZES)
+/* ... and additional padding for fake next-chunk at foot */
+#define MMAP_FOOT_PAD       (FOUR_SIZE_T_SIZES)
+
+/* The smallest size we can malloc is an aligned minimal chunk */
+#define MIN_CHUNK_SIZE\
+  ((MCHUNK_SIZE + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)
+
+/* conversion from malloc headers to user pointers, and back */
+#define chunk2mem(p)        ((void*)((char*)(p)       + TWO_SIZE_T_SIZES))
+#define mem2chunk(mem)      ((mchunkptr)((char*)(mem) - TWO_SIZE_T_SIZES))
+/* chunk associated with aligned address A */
+#define align_as_chunk(A)   (mchunkptr)((A) + align_offset(chunk2mem(A)))
+
+/* Bounds on request (not chunk) sizes. */
+#define MAX_REQUEST         ((-MIN_CHUNK_SIZE) << 2)
+#define MIN_REQUEST         (MIN_CHUNK_SIZE - CHUNK_OVERHEAD - SIZE_T_ONE)
+
+/* pad request bytes into a usable size */
+#define pad_request(req) \
+   (((req) + CHUNK_OVERHEAD + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)
+
+/* pad request, checking for minimum (but not maximum) */
+#define request2size(req) \
+  (((req) < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(req))
+
+
+/* ------------------ Operations on head and foot fields ----------------- */
+
+/*
+  The head field of a chunk is or'ed with PINUSE_BIT when previous
+  adjacent chunk in use, and or'ed with CINUSE_BIT if this chunk is in
+  use, unless mmapped, in which case both bits are cleared.
+
+  FLAG4_BIT is not used by this malloc, but might be useful in extensions.
+*/
+
+#define PINUSE_BIT          (SIZE_T_ONE)
+#define CINUSE_BIT          (SIZE_T_TWO)
+#define FLAG4_BIT           (SIZE_T_FOUR)
+#define INUSE_BITS          (PINUSE_BIT|CINUSE_BIT)
+#define FLAG_BITS           (PINUSE_BIT|CINUSE_BIT|FLAG4_BIT)
+
+/* Head value for fenceposts */
+#define FENCEPOST_HEAD      (INUSE_BITS|SIZE_T_SIZE)
+
+/* extraction of fields from head words */
+#define cinuse(p)           ((p)->head & CINUSE_BIT)
+#define pinuse(p)           ((p)->head & PINUSE_BIT)
+#define is_inuse(p)         (((p)->head & INUSE_BITS) != PINUSE_BIT)
+#define is_mmapped(p)       (((p)->head & INUSE_BITS) == 0)
+
+#define chunksize(p)        ((p)->head & ~(FLAG_BITS))
+
+#define clear_pinuse(p)     ((p)->head &= ~PINUSE_BIT)
+
+/* Treat space at ptr +/- offset as a chunk */
+#define chunk_plus_offset(p, s)  ((mchunkptr)(((char*)(p)) + (s)))
+#define chunk_minus_offset(p, s) ((mchunkptr)(((char*)(p)) - (s)))
+
+/* Ptr to next or previous physical malloc_chunk. */
+#define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->head & ~FLAG_BITS)))
+#define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_foot) ))
+
+/* extract next chunk's pinuse bit */
+#define next_pinuse(p)  ((next_chunk(p)->head) & PINUSE_BIT)
+
+/* Get/set size at footer */
+#define get_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot)
+#define set_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot = (s))
+
+/* Set size, pinuse bit, and foot */
+#define set_size_and_pinuse_of_free_chunk(p, s)\
+  ((p)->head = (s|PINUSE_BIT), set_foot(p, s))
+
+/* Set size, pinuse bit, foot, and clear next pinuse */
+#define set_free_with_pinuse(p, s, n)\
+  (clear_pinuse(n), set_size_and_pinuse_of_free_chunk(p, s))
+
+/* Get the internal overhead associated with chunk p */
+#define overhead_for(p)\
+ (is_mmapped(p)? MMAP_CHUNK_OVERHEAD : CHUNK_OVERHEAD)
+
+/* Return true if malloced space is not necessarily cleared */
+#if MMAP_CLEARS
+#define calloc_must_clear(p) (!is_mmapped(p))
+#else /* MMAP_CLEARS */
+#define calloc_must_clear(p) (1)
+#endif /* MMAP_CLEARS */
+
+/* ---------------------- Overlaid data structures ----------------------- */
+
+/*
+  When chunks are not in use, they are treated as nodes of either
+  lists or trees.
+
+  "Small"  chunks are stored in circular doubly-linked lists, and look
+  like this:
+
+    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Size of previous chunk                            |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+    `head:' |             Size of chunk, in bytes                         |P|
+      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Forward pointer to next chunk in list             |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Back pointer to previous chunk in list            |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Unused space (may be 0 bytes long)                .
+            .                                                               .
+            .                                                               |
+nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+    `foot:' |             Size of chunk, in bytes                           |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+
+  Larger chunks are kept in a form of bitwise digital trees (aka
+  tries) keyed on chunksizes.  Because malloc_tree_chunks are only for
+  free chunks greater than 256 bytes, their size doesn't impose any
+  constraints on user chunk sizes.  Each node looks like:
+
+    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Size of previous chunk                            |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+    `head:' |             Size of chunk, in bytes                         |P|
+      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Forward pointer to next chunk of same size        |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Back pointer to previous chunk of same size       |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Pointer to left child (child[0])                  |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Pointer to right child (child[1])                 |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Pointer to parent                                 |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             bin index of this chunk                           |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+            |             Unused space                                      .
+            .                                                               |
+nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+    `foot:' |             Size of chunk, in bytes                           |
+            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+
+  Each tree holding treenodes is a tree of unique chunk sizes.  Chunks
+  of the same size are arranged in a circularly-linked list, with only
+  the oldest chunk (the next to be used, in our FIFO ordering)
+  actually in the tree.  (Tree members are distinguished by a non-null
+  parent pointer.)  If a chunk with the same size an an existing node
+  is inserted, it is linked off the existing node using pointers that
+  work in the same way as fd/bk pointers of small chunks.
+
+  Each tree contains a power of 2 sized range of chunk sizes (the
+  smallest is 0x100 <= x < 0x180), which is is divided in half at each
+  tree level, with the chunks in the smaller half of the range (0x100
+  <= x < 0x140 for the top nose) in the left subtree and the larger
+  half (0x140 <= x < 0x180) in the right subtree.  This is, of course,
+  done by inspecting individual bits.
+
+  Using these rules, each node's left subtree contains all smaller
+  sizes than its right subtree.  However, the node at the root of each
+  subtree has no particular ordering relationship to either.  (The
+  dividing line between the subtree sizes is based on trie relation.)
+  If we remove the last chunk of a given size from the interior of the
+  tree, we need to replace it with a leaf node.  The tree ordering
+  rules permit a node to be replaced by any leaf below it.
+
+  The smallest chunk in a tree (a common operation in a best-fit
+  allocator) can be found by walking a path to the leftmost leaf in
+  the tree.  Unlike a usual binary tree, where we follow left child
+  pointers until we reach a null, here we follow the right child
+  pointer any time the left one is null, until we reach a leaf with
+  both child pointers null. The smallest chunk in the tree will be
+  somewhere along that path.
+
+  The worst case number of steps to add, find, or remove a node is
+  bounded by the number of bits differentiating chunks within
+  bins. Under current bin calculations, this ranges from 6 up to 21
+  (for 32 bit sizes) or up to 53 (for 64 bit sizes). The typical case
+  is of course much better.
+*/
+
+struct malloc_tree_chunk {
+  /* The first four fields must be compatible with malloc_chunk */
+  size_t                    prev_foot;
+  size_t                    head;
+  struct malloc_tree_chunk* fd;
+  struct malloc_tree_chunk* bk;
+
+  struct malloc_tree_chunk* child[2];
+  struct malloc_tree_chunk* parent;
+  bindex_t                  index;
+};
+
+typedef struct malloc_tree_chunk  tchunk;
+typedef struct malloc_tree_chunk* tchunkptr;
+typedef struct malloc_tree_chunk* tbinptr; /* The type of bins of trees */
+
+/* A little helper macro for trees */
+#define leftmost_child(t) ((t)->child[0] != 0? (t)->child[0] : (t)->child[1])
+
+/* ----------------------------- Segments -------------------------------- */
+
+/*
+  Each malloc space may include non-contiguous segments, held in a
+  list headed by an embedded malloc_segment record representing the
+  top-most space. Segments also include flags holding properties of
+  the space. Large chunks that are directly allocated by mmap are not
+  included in this list. They are instead independently created and
+  destroyed without otherwise keeping track of them.
+
+  Segment management mainly comes into play for spaces allocated by
+  MMAP.  Any call to MMAP might or might not return memory that is
+  adjacent to an existing segment.  MORECORE normally contiguously
+  extends the current space, so this space is almost always adjacent,
+  which is simpler and faster to deal with. (This is why MORECORE is
+  used preferentially to MMAP when both are available -- see
+  sys_alloc.)  When allocating using MMAP, we don't use any of the
+  hinting mechanisms (inconsistently) supported in various
+  implementations of unix mmap, or distinguish reserving from
+  committing memory. Instead, we just ask for space, and exploit
+  contiguity when we get it.  It is probably possible to do
+  better than this on some systems, but no general scheme seems
+  to be significantly better.
+
+  Management entails a simpler variant of the consolidation scheme
+  used for chunks to reduce fragmentation -- new adjacent memory is
+  normally prepended or appended to an existing segment. However,
+  there are limitations compared to chunk consolidation that mostly
+  reflect the fact that segment processing is relatively infrequent
+  (occurring only when getting memory from system) and that we
+  don't expect to have huge numbers of segments:
+
+  * Segments are not indexed, so traversal requires linear scans.  (It
+    would be possible to index these, but is not worth the extra
+    overhead and complexity for most programs on most platforms.)
+  * New segments are only appended to old ones when holding top-most
+    memory; if they cannot be prepended to others, they are held in
+    different segments.
+
+  Except for the top-most segment of an mstate, each segment record
+  is kept at the tail of its segment. Segments are added by pushing
+  segment records onto the list headed by &mstate.seg for the
+  containing mstate.
+
+  Segment flags control allocation/merge/deallocation policies:
+  * If EXTERN_BIT set, then we did not allocate this segment,
+    and so should not try to deallocate or merge with others.
+    (This currently holds only for the initial segment passed
+    into create_mspace_with_base.)
+  * If USE_MMAP_BIT set, the segment may be merged with
+    other surrounding mmapped segments and trimmed/de-allocated
+    using munmap.
+  * If neither bit is set, then the segment was obtained using
+    MORECORE so can be merged with surrounding MORECORE'd segments
+    and deallocated/trimmed using MORECORE with negative arguments.
+*/
+
+struct malloc_segment {
+  char*        base;             /* base address */
+  size_t       size;             /* allocated size */
+  struct malloc_segment* next;   /* ptr to next segment */
+  flag_t       sflags;           /* mmap and extern flag */
+};
+
+#define is_mmapped_segment(S)  ((S)->sflags & USE_MMAP_BIT)
+#define is_extern_segment(S)   ((S)->sflags & EXTERN_BIT)
+
+typedef struct malloc_segment  msegment;
+typedef struct malloc_segment* msegmentptr;
+
+/* ---------------------------- malloc_state ----------------------------- */
+
+/*
+   A malloc_state holds all of the bookkeeping for a space.
+   The main fields are:
+
+  Top
+    The topmost chunk of the currently active segment. Its size is
+    cached in topsize.  The actual size of topmost space is
+    topsize+TOP_FOOT_SIZE, which includes space reserved for adding
+    fenceposts and segment records if necessary when getting more
+    space from the system.  The size at which to autotrim top is
+    cached from mparams in trim_check, except that it is disabled if
+    an autotrim fails.
+
+  Designated victim (dv)
+    This is the preferred chunk for servicing small requests that
+    don't have exact fits.  It is normally the chunk split off most
+    recently to service another small request.  Its size is cached in
+    dvsize. The link fields of this chunk are not maintained since it
+    is not kept in a bin.
+
+  SmallBins
+    An array of bin headers for free chunks.  These bins hold chunks
+    with sizes less than MIN_LARGE_SIZE bytes. Each bin contains
+    chunks of all the same size, spaced 8 bytes apart.  To simplify
+    use in double-linked lists, each bin header acts as a malloc_chunk
+    pointing to the real first node, if it exists (else pointing to
+    itself).  This avoids special-casing for headers.  But to avoid
+    waste, we allocate only the fd/bk pointers of bins, and then use
+    repositioning tricks to treat these as the fields of a chunk.
+
+  TreeBins
+    Treebins are pointers to the roots of trees holding a range of
+    sizes. There are 2 equally spaced treebins for each power of two
+    from TREE_SHIFT to TREE_SHIFT+16. The last bin holds anything
+    larger.
+
+  Bin maps
+    There is one bit map for small bins ("smallmap") and one for
+    treebins ("treemap).  Each bin sets its bit when non-empty, and
+    clears the bit when empty.  Bit operations are then used to avoid
+    bin-by-bin searching -- nearly all "search" is done without ever
+    looking at bins that won't be selected.  The bit maps
+    conservatively use 32 bits per map word, even if on 64bit system.
+    For a good description of some of the bit-based techniques used
+    here, see Henry S. Warren Jr's book "Hacker's Delight" (and
+    supplement at http://hackersdelight.org/). Many of these are
+    intended to reduce the branchiness of paths through malloc etc, as
+    well as to reduce the number of memory locations read or written.
+
+  Segments
+    A list of segments headed by an embedded malloc_segment record
+    representing the initial space.
+
+  Address check support
+    The least_addr field is the least address ever obtained from
+    MORECORE or MMAP. Attempted frees and reallocs of any address less
+    than this are trapped (unless INSECURE is defined).
+
+  Magic tag
+    A cross-check field that should always hold same value as mparams.magic.
+
+  Flags
+    Bits recording whether to use MMAP, locks, or contiguous MORECORE
+
+  Statistics
+    Each space keeps track of current and maximum system memory
+    obtained via MORECORE or MMAP.
+
+  Trim support
+    Fields holding the amount of unused topmost memory that should trigger
+    timming, and a counter to force periodic scanning to release unused
+    non-topmost segments.
+
+  Locking
+    If USE_LOCKS is defined, the "mutex" lock is acquired and released
+    around every public call using this mspace.
+
+  Extension support
+    A void* pointer and a size_t field that can be used to help implement
+    extensions to this malloc.
+*/
+
+/* Bin types, widths and sizes */
+#define NSMALLBINS        (32U)
+#define NTREEBINS         (32U)
+#define SMALLBIN_SHIFT    (3U)
+#define SMALLBIN_WIDTH    (SIZE_T_ONE << SMALLBIN_SHIFT)
+#define TREEBIN_SHIFT     (8U)
+#define MIN_LARGE_SIZE    (SIZE_T_ONE << TREEBIN_SHIFT)
+#define MAX_SMALL_SIZE    (MIN_LARGE_SIZE - SIZE_T_ONE)
+#define MAX_SMALL_REQUEST (MAX_SMALL_SIZE - CHUNK_ALIGN_MASK - CHUNK_OVERHEAD)
+
+struct malloc_state {
+  binmap_t   smallmap;
+  binmap_t   treemap;
+  size_t     dvsize;
+  size_t     topsize;
+  char*      least_addr;
+  mchunkptr  dv;
+  mchunkptr  top;
+  size_t     trim_check;
+  size_t     release_checks;
+  size_t     magic;
+  mchunkptr  smallbins[(NSMALLBINS+1)*2];
+  tbinptr    treebins[NTREEBINS];
+  size_t     footprint;
+  size_t     max_footprint;
+  flag_t     mflags;
+  msegment   seg;
+#if USE_LOCKS
+  MLOCK_T    mutex;     /* locate lock among fields that rarely change */
+#endif /* USE_LOCKS */
+  void*      extp;      /* Unused but available for extensions */
+  size_t     exts;
+};
+
+typedef struct malloc_state*    mstate;
+
+/* ------------- Global malloc_state and malloc_params ------------------- */
+
+#if !ONLY_MSPACES
+
+/* The global malloc_state used for all non-"mspace" calls */
+static struct malloc_state _gm_;
+#define gm                 (&_gm_)
+#define is_global(M)       ((M) == &_gm_)
+
+#endif /* !ONLY_MSPACES */
+
+#define is_initialized(M)  ((M)->top != 0)
+
+/* -------------------------- system alloc setup ------------------------- */
+
+/* Operations on mflags */
+
+#define use_lock(M)           ((M)->mflags &   USE_LOCK_BIT)
+#define enable_lock(M)        ((M)->mflags |=  USE_LOCK_BIT)
+#define disable_lock(M)       ((M)->mflags &= ~USE_LOCK_BIT)
+
+#define use_mmap(M)           ((M)->mflags &   USE_MMAP_BIT)
+#define enable_mmap(M)        ((M)->mflags |=  USE_MMAP_BIT)
+#define disable_mmap(M)       ((M)->mflags &= ~USE_MMAP_BIT)
+
+#define use_noncontiguous(M)  ((M)->mflags &   USE_NONCONTIGUOUS_BIT)
+#define disable_contiguous(M) ((M)->mflags |=  USE_NONCONTIGUOUS_BIT)
+
+#define set_lock(M,L)\
+ ((M)->mflags = (L)?\
+  ((M)->mflags | USE_LOCK_BIT) :\
+  ((M)->mflags & ~USE_LOCK_BIT))
+
+/* page-align a size */
+#define page_align(S)\
+ (((S) + (mparams.page_size - SIZE_T_ONE)) & ~(mparams.page_size - SIZE_T_ONE))
+
+/* granularity-align a size */
+#define granularity_align(S)\
+  (((S) + (mparams.granularity - SIZE_T_ONE))\
+   & ~(mparams.granularity - SIZE_T_ONE))
+
+
+/* For mmap, use granularity alignment on windows, else page-align */
+#ifdef WIN32
+#define mmap_align(S) granularity_align(S)
+#else
+#define mmap_align(S) page_align(S)
+#endif
+
+/* For sys_alloc, enough padding to ensure can malloc request on success */
+#define SYS_ALLOC_PADDING (TOP_FOOT_SIZE + MALLOC_ALIGNMENT)
+
+#define is_page_aligned(S)\
+   (((size_t)(S) & (mparams.page_size - SIZE_T_ONE)) == 0)
+#define is_granularity_aligned(S)\
+   (((size_t)(S) & (mparams.granularity - SIZE_T_ONE)) == 0)
+
+/*  True if segment S holds address A */
+#define segment_holds(S, A)\
+  ((char*)(A) >= S->base && (char*)(A) < S->base + S->size)
+
+/* Return segment holding given address */
+static msegmentptr segment_holding(mstate m, char* addr) {
+  msegmentptr sp = &m->seg;
+  for (;;) {
+    if (addr >= sp->base && addr < sp->base + sp->size)
+      return sp;
+    if ((sp = sp->next) == 0)
+      return 0;
+  }
+}
+
+/* Return true if segment contains a segment link */
+static int has_segment_link(mstate m, msegmentptr ss) {
+  msegmentptr sp = &m->seg;
+  for (;;) {
+    if ((char*)sp >= ss->base && (char*)sp < ss->base + ss->size)
+      return 1;
+    if ((sp = sp->next) == 0)
+      return 0;
+  }
+}
+
+#ifndef MORECORE_CANNOT_TRIM
+#define should_trim(M,s)  ((s) > (M)->trim_check)
+#else  /* MORECORE_CANNOT_TRIM */
+#define should_trim(M,s)  (0)
+#endif /* MORECORE_CANNOT_TRIM */
+
+/*
+  TOP_FOOT_SIZE is padding at the end of a segment, including space
+  that may be needed to place segment records and fenceposts when new
+  noncontiguous segments are added.
+*/
+#define TOP_FOOT_SIZE\
+  (align_offset(chunk2mem(0))+pad_request(sizeof(struct malloc_segment))+MIN_CHUNK_SIZE)
+
+
+/* -------------------------------  Hooks -------------------------------- */
+
+/*
+  PREACTION should be defined to return 0 on success, and nonzero on
+  failure. If you are not using locking, you can redefine these to do
+  anything you like.
+*/
+
+#if USE_LOCKS
+
+#define PREACTION(M)  ((use_lock(M))? ACQUIRE_LOCK(&(M)->mutex) : 0)
+#define POSTACTION(M) { if (use_lock(M)) RELEASE_LOCK(&(M)->mutex); }
+#else /* USE_LOCKS */
+
+#ifndef PREACTION
+#define PREACTION(M) (0)
+#endif  /* PREACTION */
+
+#ifndef POSTACTION
+#define POSTACTION(M)
+#endif  /* POSTACTION */
+
+#endif /* USE_LOCKS */
+
+/*
+  CORRUPTION_ERROR_ACTION is triggered upon detected bad addresses.
+  USAGE_ERROR_ACTION is triggered on detected bad frees and
+  reallocs. The argument p is an address that might have triggered the
+  fault. It is ignored by the two predefined actions, but might be
+  useful in custom actions that try to help diagnose errors.
+*/
+
+#if PROCEED_ON_ERROR
+
+/* A count of the number of corruption errors causing resets */
+int malloc_corruption_error_count;
+
+/* default corruption action */
+static void reset_on_error(mstate m);
+
+#define CORRUPTION_ERROR_ACTION(m)  reset_on_error(m)
+#define USAGE_ERROR_ACTION(m, p)
+
+#else /* PROCEED_ON_ERROR */
+
+#ifndef CORRUPTION_ERROR_ACTION
+#define CORRUPTION_ERROR_ACTION(m) ABORT
+#endif /* CORRUPTION_ERROR_ACTION */
+
+#ifndef USAGE_ERROR_ACTION
+#define USAGE_ERROR_ACTION(m,p) ABORT
+#endif /* USAGE_ERROR_ACTION */
+
+#endif /* PROCEED_ON_ERROR */
+
+/* -------------------------- Debugging setup ---------------------------- */
+
+#if ! DEBUG
+
+#define check_free_chunk(M,P)
+#define check_inuse_chunk(M,P)
+#define check_malloced_chunk(M,P,N)
+#define check_mmapped_chunk(M,P)
+#define check_malloc_state(M)
+#define check_top_chunk(M,P)
+
+#else /* DEBUG */
+#define check_free_chunk(M,P)       do_check_free_chunk(M,P)
+#define check_inuse_chunk(M,P)      do_check_inuse_chunk(M,P)
+#define check_top_chunk(M,P)        do_check_top_chunk(M,P)
+#define check_malloced_chunk(M,P,N) do_check_malloced_chunk(M,P,N)
+#define check_mmapped_chunk(M,P)    do_check_mmapped_chunk(M,P)
+#define check_malloc_state(M)       do_check_malloc_state(M)
+
+static void   do_check_any_chunk(mstate m, mchunkptr p);
+static void   do_check_top_chunk(mstate m, mchunkptr p);
+static void   do_check_mmapped_chunk(mstate m, mchunkptr p);
+static void   do_check_inuse_chunk(mstate m, mchunkptr p);
+static void   do_check_free_chunk(mstate m, mchunkptr p);
+static void   do_check_malloced_chunk(mstate m, void* mem, size_t s);
+static void   do_check_tree(mstate m, tchunkptr t);
+static void   do_check_treebin(mstate m, bindex_t i);
+static void   do_check_smallbin(mstate m, bindex_t i);
+static void   do_check_malloc_state(mstate m);
+static int    bin_find(mstate m, mchunkptr x);
+static size_t traverse_and_check(mstate m);
+#endif /* DEBUG */
+
+/* ---------------------------- Indexing Bins ---------------------------- */
+
+#define is_small(s)         (((s) >> SMALLBIN_SHIFT) < NSMALLBINS)
+#define small_index(s)      (bindex_t)((s)  >> SMALLBIN_SHIFT)
+#define small_index2size(i) ((i)  << SMALLBIN_SHIFT)
+#define MIN_SMALL_INDEX     (small_index(MIN_CHUNK_SIZE))
+
+/* addressing by index. See above about smallbin repositioning */
+#define smallbin_at(M, i)   ((sbinptr)((char*)&((M)->smallbins[(i)<<1])))
+#define treebin_at(M,i)     (&((M)->treebins[i]))
+
+/* assign tree index for size S to variable I. Use x86 asm if possible  */
+#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
+#define compute_tree_index(S, I)\
+{\
+  unsigned int X = S >> TREEBIN_SHIFT;\
+  if (X == 0)\
+    I = 0;\
+  else if (X > 0xFFFF)\
+    I = NTREEBINS-1;\
+  else {\
+    unsigned int K;\
+    __asm__("bsrl\t%1, %0\n\t" : "=r" (K) : "g"  (X));\
+    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\
+  }\
+}
+
+#elif defined (__INTEL_COMPILER)
+#define compute_tree_index(S, I)\
+{\
+  size_t X = S >> TREEBIN_SHIFT;\
+  if (X == 0)\
+    I = 0;\
+  else if (X > 0xFFFF)\
+    I = NTREEBINS-1;\
+  else {\
+    unsigned int K = _bit_scan_reverse (X); \
+    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\
+  }\
+}
+
+#elif defined(_MSC_VER) && _MSC_VER>=1300
+#define compute_tree_index(S, I)\
+{\
+  size_t X = S >> TREEBIN_SHIFT;\
+  if (X == 0)\
+    I = 0;\
+  else if (X > 0xFFFF)\
+    I = NTREEBINS-1;\
+  else {\
+    unsigned int K;\
+    _BitScanReverse((DWORD *) &K, (DWORD) X);\
+    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\
+  }\
+}
+
+#else /* GNUC */
+#define compute_tree_index(S, I)\
+{\
+  size_t X = S >> TREEBIN_SHIFT;\
+  if (X == 0)\
+    I = 0;\
+  else if (X > 0xFFFF)\
+    I = NTREEBINS-1;\
+  else {\
+    unsigned int Y = (unsigned int)X;\
+    unsigned int N = ((Y - 0x100) >> 16) & 8;\
+    unsigned int K = (((Y <<= N) - 0x1000) >> 16) & 4;\
+    N += K;\
+    N += K = (((Y <<= K) - 0x4000) >> 16) & 2;\
+    K = 14 - N + ((Y <<= K) >> 15);\
+    I = (K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1));\
+  }\
+}
+#endif /* GNUC */
+
+/* Bit representing maximum resolved size in a treebin at i */
+#define bit_for_tree_index(i) \
+   (i == NTREEBINS-1)? (SIZE_T_BITSIZE-1) : (((i) >> 1) + TREEBIN_SHIFT - 2)
+
+/* Shift placing maximum resolved bit in a treebin at i as sign bit */
+#define leftshift_for_tree_index(i) \
+   ((i == NTREEBINS-1)? 0 : \
+    ((SIZE_T_BITSIZE-SIZE_T_ONE) - (((i) >> 1) + TREEBIN_SHIFT - 2)))
+
+/* The size of the smallest chunk held in bin with index i */
+#define minsize_for_tree_index(i) \
+   ((SIZE_T_ONE << (((i) >> 1) + TREEBIN_SHIFT)) |  \
+   (((size_t)((i) & SIZE_T_ONE)) << (((i) >> 1) + TREEBIN_SHIFT - 1)))
+
+
+/* ------------------------ Operations on bin maps ----------------------- */
+
+/* bit corresponding to given index */
+#define idx2bit(i)              ((binmap_t)(1) << (i))
+
+/* Mark/Clear bits with given index */
+#define mark_smallmap(M,i)      ((M)->smallmap |=  idx2bit(i))
+#define clear_smallmap(M,i)     ((M)->smallmap &= ~idx2bit(i))
+#define smallmap_is_marked(M,i) ((M)->smallmap &   idx2bit(i))
+
+#define mark_treemap(M,i)       ((M)->treemap  |=  idx2bit(i))
+#define clear_treemap(M,i)      ((M)->treemap  &= ~idx2bit(i))
+#define treemap_is_marked(M,i)  ((M)->treemap  &   idx2bit(i))
+
+/* isolate the least set bit of a bitmap */
+#define least_bit(x)         ((x) & -(x))
+
+/* mask with all bits to left of least bit of x on */
+#define left_bits(x)         ((x<<1) | -(x<<1))
+
+/* mask with all bits to left of or equal to least bit of x on */
+#define same_or_left_bits(x) ((x) | -(x))
+
+/* index corresponding to given bit. Use x86 asm if possible */
+
+#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
+#define compute_bit2idx(X, I)\
+{\
+  unsigned int J;\
+  __asm__("bsfl\t%1, %0\n\t" : "=r" (J) : "g" (X));\
+  I = (bindex_t)J;\
+}
+
+#elif defined (__INTEL_COMPILER)
+#define compute_bit2idx(X, I)\
+{\
+  unsigned int J;\
+  J = _bit_scan_forward (X); \
+  I = (bindex_t)J;\
+}
+
+#elif defined(_MSC_VER) && _MSC_VER>=1300
+#define compute_bit2idx(X, I)\
+{\
+  unsigned int J;\
+  _BitScanForward((DWORD *) &J, X);\
+  I = (bindex_t)J;\
+}
+
+#elif USE_BUILTIN_FFS
+#define compute_bit2idx(X, I) I = ffs(X)-1
+
+#else
+#define compute_bit2idx(X, I)\
+{\
+  unsigned int Y = X - 1;\
+  unsigned int K = Y >> (16-4) & 16;\
+  unsigned int N = K;        Y >>= K;\
+  N += K = Y >> (8-3) &  8;  Y >>= K;\
+  N += K = Y >> (4-2) &  4;  Y >>= K;\
+  N += K = Y >> (2-1) &  2;  Y >>= K;\
+  N += K = Y >> (1-0) &  1;  Y >>= K;\
+  I = (bindex_t)(N + Y);\
+}
+#endif /* GNUC */
+
+
+/* ----------------------- Runtime Check Support ------------------------- */
+
+/*
+  For security, the main invariant is that malloc/free/etc never
+  writes to a static address other than malloc_state, unless static
+  malloc_state itself has been corrupted, which cannot occur via
+  malloc (because of these checks). In essence this means that we
+  believe all pointers, sizes, maps etc held in malloc_state, but
+  check all of those linked or offsetted from other embedded data
+  structures.  These checks are interspersed with main code in a way
+  that tends to minimize their run-time cost.
+
+  When FOOTERS is defined, in addition to range checking, we also
+  verify footer fields of inuse chunks, which can be used guarantee
+  that the mstate controlling malloc/free is intact.  This is a
+  streamlined version of the approach described by William Robertson
+  et al in "Run-time Detection of Heap-based Overflows" LISA'03
+  http://www.usenix.org/events/lisa03/tech/robertson.html The footer
+  of an inuse chunk holds the xor of its mstate and a random seed,
+  that is checked upon calls to free() and realloc().  This is
+  (probablistically) unguessable from outside the program, but can be
+  computed by any code successfully malloc'ing any chunk, so does not
+  itself provide protection against code that has already broken
+  security through some other means.  Unlike Robertson et al, we
+  always dynamically check addresses of all offset chunks (previous,
+  next, etc). This turns out to be cheaper than relying on hashes.
+*/
+
+#if !INSECURE
+/* Check if address a is at least as high as any from MORECORE or MMAP */
+#define ok_address(M, a) ((char*)(a) >= (M)->least_addr)
+/* Check if address of next chunk n is higher than base chunk p */
+#define ok_next(p, n)    ((char*)(p) < (char*)(n))
+/* Check if p has inuse status */
+#define ok_inuse(p)     is_inuse(p)
+/* Check if p has its pinuse bit on */
+#define ok_pinuse(p)     pinuse(p)
+
+#else /* !INSECURE */
+#define ok_address(M, a) (1)
+#define ok_next(b, n)    (1)
+#define ok_inuse(p)      (1)
+#define ok_pinuse(p)     (1)
+#endif /* !INSECURE */
+
+#if (FOOTERS && !INSECURE)
+/* Check if (alleged) mstate m has expected magic field */
+#define ok_magic(M)      ((M)->magic == mparams.magic)
+#else  /* (FOOTERS && !INSECURE) */
+#define ok_magic(M)      (1)
+#endif /* (FOOTERS && !INSECURE) */
+
+
+/* In gcc, use __builtin_expect to minimize impact of checks */
+#if !INSECURE
+#if defined(__GNUC__) && __GNUC__ >= 3
+#define RTCHECK(e)  __builtin_expect(e, 1)
+#else /* GNUC */
+#define RTCHECK(e)  (e)
+#endif /* GNUC */
+#else /* !INSECURE */
+#define RTCHECK(e)  (1)
+#endif /* !INSECURE */
+
+/* macros to set up inuse chunks with or without footers */
+
+#if !FOOTERS
+
+#define mark_inuse_foot(M,p,s)
+
+/* Macros for setting head/foot of non-mmapped chunks */
+
+/* Set cinuse bit and pinuse bit of next chunk */
+#define set_inuse(M,p,s)\
+  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\
+  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)
+
+/* Set cinuse and pinuse of this chunk and pinuse of next chunk */
+#define set_inuse_and_pinuse(M,p,s)\
+  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
+  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)
+
+/* Set size, cinuse and pinuse bit of this chunk */
+#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\
+  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT))
+
+#else /* FOOTERS */
+
+/* Set foot of inuse chunk to be xor of mstate and seed */
+#define mark_inuse_foot(M,p,s)\
+  (((mchunkptr)((char*)(p) + (s)))->prev_foot = ((size_t)(M) ^ mparams.magic))
+
+#define get_mstate_for(p)\
+  ((mstate)(((mchunkptr)((char*)(p) +\
+    (chunksize(p))))->prev_foot ^ mparams.magic))
+
+#define set_inuse(M,p,s)\
+  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\
+  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT), \
+  mark_inuse_foot(M,p,s))
+
+#define set_inuse_and_pinuse(M,p,s)\
+  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
+  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT),\
+ mark_inuse_foot(M,p,s))
+
+#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\
+  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
+  mark_inuse_foot(M, p, s))
+
+#endif /* !FOOTERS */
+
+/* ---------------------------- setting mparams -------------------------- */
+
+#ifdef ENABLE_LARGE_PAGES
+typedef size_t (WINAPI *GetLargePageMinimum_t)(void);
+#endif
+
+/* Initialize mparams */
+static int init_mparams(void) {
+#ifdef NEED_GLOBAL_LOCK_INIT
+  if (malloc_global_mutex_status <= 0)
+    init_malloc_global_mutex();
+#endif
+
+  ACQUIRE_MALLOC_GLOBAL_LOCK();
+  if (mparams.magic == 0) {
+    size_t magic;
+    size_t psize;
+    size_t gsize;
+
+#ifndef WIN32
+    psize = malloc_getpagesize;
+    gsize = ((DEFAULT_GRANULARITY != 0)? DEFAULT_GRANULARITY : psize);
+#else /* WIN32 */
+    {
+      SYSTEM_INFO system_info;
+      GetSystemInfo(&system_info);
+      psize = system_info.dwPageSize;
+      gsize = ((DEFAULT_GRANULARITY != 0)?
+               DEFAULT_GRANULARITY : system_info.dwAllocationGranularity);
+#ifdef ENABLE_LARGE_PAGES
+      { 
+          GetLargePageMinimum_t GetLargePageMinimum_ = (GetLargePageMinimum_t) GetProcAddress(GetModuleHandle(__T("kernel32.dll")), "GetLargePageMinimum");
+          if(GetLargePageMinimum_) {
+              size_t largepagesize = GetLargePageMinimum_();
+              if(largepagesize) {
+                  psize = largepagesize;
+                  gsize = ((DEFAULT_GRANULARITY != 0)?
+                           DEFAULT_GRANULARITY : largepagesize);
+                  if(gsize < largepagesize) gsize = largepagesize;
+              }
+          }
+      }
+#endif
+    }
+#endif /* WIN32 */
+
+    /* Sanity-check configuration:
+       size_t must be unsigned and as wide as pointer type.
+       ints must be at least 4 bytes.
+       alignment must be at least 8.
+       Alignment, min chunk size, and page size must all be powers of 2.
+    */
+    if ((sizeof(size_t) != sizeof(char*)) ||
+        (MAX_SIZE_T < MIN_CHUNK_SIZE)  ||
+        (sizeof(int) < 4)  ||
+        (MALLOC_ALIGNMENT < (size_t)8U) ||
+        ((MALLOC_ALIGNMENT & (MALLOC_ALIGNMENT-SIZE_T_ONE)) != 0) ||
+        ((MCHUNK_SIZE      & (MCHUNK_SIZE-SIZE_T_ONE))      != 0) ||
+        ((gsize            & (gsize-SIZE_T_ONE))            != 0) ||
+        ((psize            & (psize-SIZE_T_ONE))            != 0))
+      ABORT;
+
+    mparams.granularity = gsize;
+    mparams.page_size = psize;
+    mparams.mmap_threshold = DEFAULT_MMAP_THRESHOLD;
+    mparams.trim_threshold = DEFAULT_TRIM_THRESHOLD;
+#if MORECORE_CONTIGUOUS
+    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT;
+#else  /* MORECORE_CONTIGUOUS */
+    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT|USE_NONCONTIGUOUS_BIT;
+#endif /* MORECORE_CONTIGUOUS */
+
+#if !ONLY_MSPACES
+    /* Set up lock for main malloc area */
+    gm->mflags = mparams.default_mflags;
+    INITIAL_LOCK(&gm->mutex);
+#endif
+
+    {
+#if USE_DEV_RANDOM
+      int fd;
+      unsigned char buf[sizeof(size_t)];
+      /* Try to use /dev/urandom, else fall back on using time */
+      if ((fd = open("/dev/urandom", O_RDONLY)) >= 0 &&
+          read(fd, buf, sizeof(buf)) == sizeof(buf)) {
+        magic = *((size_t *) buf);
+        close(fd);
+      }
+      else
+#endif /* USE_DEV_RANDOM */
+#ifdef WIN32
+        magic = (size_t)(GetTickCount() ^ (size_t)0x55555555U);
+#else
+        magic = (size_t)(time(0) ^ (size_t)0x55555555U);
+#endif
+      magic |= (size_t)8U;    /* ensure nonzero */
+      magic &= ~(size_t)7U;   /* improve chances of fault for bad values */
+      mparams.magic = magic;
+    }
+  }
+
+  RELEASE_MALLOC_GLOBAL_LOCK();
+  return 1;
+}
+
+/* support for mallopt */
+static int change_mparam(int param_number, int value) {
+  size_t val;
+  ensure_initialization();
+  val = (value == -1)? MAX_SIZE_T : (size_t)value;
+  switch(param_number) {
+  case M_TRIM_THRESHOLD:
+    mparams.trim_threshold = val;
+    return 1;
+  case M_GRANULARITY:
+    if (val >= mparams.page_size && ((val & (val-1)) == 0)) {
+      mparams.granularity = val;
+      return 1;
+    }
+    else
+      return 0;
+  case M_MMAP_THRESHOLD:
+    mparams.mmap_threshold = val;
+    return 1;
+  default:
+    return 0;
+  }
+}
+
+#if DEBUG
+/* ------------------------- Debugging Support --------------------------- */
+
+/* Check properties of any chunk, whether free, inuse, mmapped etc  */
+static void do_check_any_chunk(mstate m, mchunkptr p) {
+  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
+  assert(ok_address(m, p));
+}
+
+/* Check properties of top chunk */
+static void do_check_top_chunk(mstate m, mchunkptr p) {
+  msegmentptr sp = segment_holding(m, (char*)p);
+  size_t  sz = p->head & ~INUSE_BITS; /* third-lowest bit can be set! */
+  assert(sp != 0);
+  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
+  assert(ok_address(m, p));
+  assert(sz == m->topsize);
+  assert(sz > 0);
+  assert(sz == ((sp->base + sp->size) - (char*)p) - TOP_FOOT_SIZE);
+  assert(pinuse(p));
+  assert(!pinuse(chunk_plus_offset(p, sz)));
+}
+
+/* Check properties of (inuse) mmapped chunks */
+static void do_check_mmapped_chunk(mstate m, mchunkptr p) {
+  size_t  sz = chunksize(p);
+  size_t len = (sz + (p->prev_foot) + MMAP_FOOT_PAD);
+  assert(is_mmapped(p));
+  assert(use_mmap(m));
+  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
+  assert(ok_address(m, p));
+  assert(!is_small(sz));
+  assert((len & (mparams.page_size-SIZE_T_ONE)) == 0);
+  assert(chunk_plus_offset(p, sz)->head == FENCEPOST_HEAD);
+  assert(chunk_plus_offset(p, sz+SIZE_T_SIZE)->head == 0);
+}
+
+/* Check properties of inuse chunks */
+static void do_check_inuse_chunk(mstate m, mchunkptr p) {
+  do_check_any_chunk(m, p);
+  assert(is_inuse(p));
+  assert(next_pinuse(p));
+  /* If not pinuse and not mmapped, previous chunk has OK offset */
+  assert(is_mmapped(p) || pinuse(p) || next_chunk(prev_chunk(p)) == p);
+  if (is_mmapped(p))
+    do_check_mmapped_chunk(m, p);
+}
+
+/* Check properties of free chunks */
+static void do_check_free_chunk(mstate m, mchunkptr p) {
+  size_t sz = chunksize(p);
+  mchunkptr next = chunk_plus_offset(p, sz);
+  do_check_any_chunk(m, p);
+  assert(!is_inuse(p));
+  assert(!next_pinuse(p));
+  assert (!is_mmapped(p));
+  if (p != m->dv && p != m->top) {
+    if (sz >= MIN_CHUNK_SIZE) {
+      assert((sz & CHUNK_ALIGN_MASK) == 0);
+      assert(is_aligned(chunk2mem(p)));
+      assert(next->prev_foot == sz);
+      assert(pinuse(p));
+      assert (next == m->top || is_inuse(next));
+      assert(p->fd->bk == p);
+      assert(p->bk->fd == p);
+    }
+    else  /* markers are always of size SIZE_T_SIZE */
+      assert(sz == SIZE_T_SIZE);
+  }
+}
+
+/* Check properties of malloced chunks at the point they are malloced */
+static void do_check_malloced_chunk(mstate m, void* mem, size_t s) {
+  if (mem != 0) {
+    mchunkptr p = mem2chunk(mem);
+    size_t sz = p->head & ~INUSE_BITS;
+    do_check_inuse_chunk(m, p);
+    assert((sz & CHUNK_ALIGN_MASK) == 0);
+    assert(sz >= MIN_CHUNK_SIZE);
+    assert(sz >= s);
+    /* unless mmapped, size is less than MIN_CHUNK_SIZE more than request */
+    assert(is_mmapped(p) || sz < (s + MIN_CHUNK_SIZE));
+  }
+}
+
+/* Check a tree and its subtrees.  */
+static void do_check_tree(mstate m, tchunkptr t) {
+  tchunkptr head = 0;
+  tchunkptr u = t;
+  bindex_t tindex = t->index;
+  size_t tsize = chunksize(t);
+  bindex_t idx;
+  compute_tree_index(tsize, idx);
+  assert(tindex == idx);
+  assert(tsize >= MIN_LARGE_SIZE);
+  assert(tsize >= minsize_for_tree_index(idx));
+  assert((idx == NTREEBINS-1) || (tsize < minsize_for_tree_index((idx+1))));
+
+  do { /* traverse through chain of same-sized nodes */
+    do_check_any_chunk(m, ((mchunkptr)u));
+    assert(u->index == tindex);
+    assert(chunksize(u) == tsize);
+    assert(!is_inuse(u));
+    assert(!next_pinuse(u));
+    assert(u->fd->bk == u);
+    assert(u->bk->fd == u);
+    if (u->parent == 0) {
+      assert(u->child[0] == 0);
+      assert(u->child[1] == 0);
+    }
+    else {
+      assert(head == 0); /* only one node on chain has parent */
+      head = u;
+      assert(u->parent != u);
+      assert (u->parent->child[0] == u ||
+              u->parent->child[1] == u ||
+              *((tbinptr*)(u->parent)) == u);
+      if (u->child[0] != 0) {
+        assert(u->child[0]->parent == u);
+        assert(u->child[0] != u);
+        do_check_tree(m, u->child[0]);
+      }
+      if (u->child[1] != 0) {
+        assert(u->child[1]->parent == u);
+        assert(u->child[1] != u);
+        do_check_tree(m, u->child[1]);
+      }
+      if (u->child[0] != 0 && u->child[1] != 0) {
+        assert(chunksize(u->child[0]) < chunksize(u->child[1]));
+      }
+    }
+    u = u->fd;
+  } while (u != t);
+  assert(head != 0);
+}
+
+/*  Check all the chunks in a treebin.  */
+static void do_check_treebin(mstate m, bindex_t i) {
+  tbinptr* tb = treebin_at(m, i);
+  tchunkptr t = *tb;
+  int empty = (m->treemap & (1U << i)) == 0;
+  if (t == 0)
+    assert(empty);
+  if (!empty)
+    do_check_tree(m, t);
+}
+
+/*  Check all the chunks in a smallbin.  */
+static void do_check_smallbin(mstate m, bindex_t i) {
+  sbinptr b = smallbin_at(m, i);
+  mchunkptr p = b->bk;
+  unsigned int empty = (m->smallmap & (1U << i)) == 0;
+  if (p == b)
+    assert(empty);
+  if (!empty) {
+    for (; p != b; p = p->bk) {
+      size_t size = chunksize(p);
+      mchunkptr q;
+      /* each chunk claims to be free */
+      do_check_free_chunk(m, p);
+      /* chunk belongs in bin */
+      assert(small_index(size) == i);
+      assert(p->bk == b || chunksize(p->bk) == chunksize(p));
+      /* chunk is followed by an inuse chunk */
+      q = next_chunk(p);
+      if (q->head != FENCEPOST_HEAD)
+        do_check_inuse_chunk(m, q);
+    }
+  }
+}
+
+/* Find x in a bin. Used in other check functions. */
+static int bin_find(mstate m, mchunkptr x) {
+  size_t size = chunksize(x);
+  if (is_small(size)) {
+    bindex_t sidx = small_index(size);
+    sbinptr b = smallbin_at(m, sidx);
+    if (smallmap_is_marked(m, sidx)) {
+      mchunkptr p = b;
+      do {
+        if (p == x)
+          return 1;
+      } while ((p = p->fd) != b);
+    }
+  }
+  else {
+    bindex_t tidx;
+    compute_tree_index(size, tidx);
+    if (treemap_is_marked(m, tidx)) {
+      tchunkptr t = *treebin_at(m, tidx);
+      size_t sizebits = size << leftshift_for_tree_index(tidx);
+      while (t != 0 && chunksize(t) != size) {
+        t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];
+        sizebits <<= 1;
+      }
+      if (t != 0) {
+        tchunkptr u = t;
+        do {
+          if (u == (tchunkptr)x)
+            return 1;
+        } while ((u = u->fd) != t);
+      }
+    }
+  }
+  return 0;
+}
+
+/* Traverse each chunk and check it; return total */
+static size_t traverse_and_check(mstate m) {
+  size_t sum = 0;
+  if (is_initialized(m)) {
+    msegmentptr s = &m->seg;
+    sum += m->topsize + TOP_FOOT_SIZE;
+    while (s != 0) {
+      mchunkptr q = align_as_chunk(s->base);
+      mchunkptr lastq = 0;
+      assert(pinuse(q));
+      while (segment_holds(s, q) &&
+             q != m->top && q->head != FENCEPOST_HEAD) {
+        sum += chunksize(q);
+        if (is_inuse(q)) {
+          assert(!bin_find(m, q));
+          do_check_inuse_chunk(m, q);
+        }
+        else {
+          assert(q == m->dv || bin_find(m, q));
+          assert(lastq == 0 || is_inuse(lastq)); /* Not 2 consecutive free */
+          do_check_free_chunk(m, q);
+        }
+        lastq = q;
+        q = next_chunk(q);
+      }
+      s = s->next;
+    }
+  }
+  return sum;
+}
+
+/* Check all properties of malloc_state. */
+static void do_check_malloc_state(mstate m) {
+  bindex_t i;
+  size_t total;
+  /* check bins */
+  for (i = 0; i < NSMALLBINS; ++i)
+    do_check_smallbin(m, i);
+  for (i = 0; i < NTREEBINS; ++i)
+    do_check_treebin(m, i);
+
+  if (m->dvsize != 0) { /* check dv chunk */
+    do_check_any_chunk(m, m->dv);
+    assert(m->dvsize == chunksize(m->dv));
+    assert(m->dvsize >= MIN_CHUNK_SIZE);
+    assert(bin_find(m, m->dv) == 0);
+  }
+
+  if (m->top != 0) {   /* check top chunk */
+    do_check_top_chunk(m, m->top);
+    /*assert(m->topsize == chunksize(m->top)); redundant */
+    assert(m->topsize > 0);
+    assert(bin_find(m, m->top) == 0);
+  }
+
+  total = traverse_and_check(m);
+  assert(total <= m->footprint);
+  assert(m->footprint <= m->max_footprint);
+}
+#endif /* DEBUG */
+
+/* ----------------------------- statistics ------------------------------ */
+
+#if !NO_MALLINFO
+static struct mallinfo internal_mallinfo(mstate m) {
+  struct mallinfo nm = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
+  ensure_initialization();
+  if (!PREACTION(m)) {
+    check_malloc_state(m);
+    if (is_initialized(m)) {
+      size_t nfree = SIZE_T_ONE; /* top always free */
+      size_t mfree = m->topsize + TOP_FOOT_SIZE;
+      size_t sum = mfree;
+      msegmentptr s = &m->seg;
+      while (s != 0) {
+        mchunkptr q = align_as_chunk(s->base);
+        while (segment_holds(s, q) &&
+               q != m->top && q->head != FENCEPOST_HEAD) {
+          size_t sz = chunksize(q);
+          sum += sz;
+          if (!is_inuse(q)) {
+            mfree += sz;
+            ++nfree;
+          }
+          q = next_chunk(q);
+        }
+        s = s->next;
+      }
+
+      nm.arena    = sum;
+      nm.ordblks  = nfree;
+      nm.hblkhd   = m->footprint - sum;
+      nm.usmblks  = m->max_footprint;
+      nm.uordblks = m->footprint - mfree;
+      nm.fordblks = mfree;
+      nm.keepcost = m->topsize;
+    }
+
+    POSTACTION(m);
+  }
+  return nm;
+}
+#endif /* !NO_MALLINFO */
+
+static void internal_malloc_stats(mstate m) {
+  ensure_initialization();
+  if (!PREACTION(m)) {
+    size_t maxfp = 0;
+    size_t fp = 0;
+    size_t used = 0;
+    check_malloc_state(m);
+    if (is_initialized(m)) {
+      msegmentptr s = &m->seg;
+      maxfp = m->max_footprint;
+      fp = m->footprint;
+      used = fp - (m->topsize + TOP_FOOT_SIZE);
+
+      while (s != 0) {
+        mchunkptr q = align_as_chunk(s->base);
+        while (segment_holds(s, q) &&
+               q != m->top && q->head != FENCEPOST_HEAD) {
+          if (!is_inuse(q))
+            used -= chunksize(q);
+          q = next_chunk(q);
+        }
+        s = s->next;
+      }
+    }
+
+    fprintf(stderr, "max system bytes = %10lu\n", (unsigned long)(maxfp));
+    fprintf(stderr, "system bytes     = %10lu\n", (unsigned long)(fp));
+    fprintf(stderr, "in use bytes     = %10lu\n", (unsigned long)(used));
+
+    POSTACTION(m);
+  }
+}
+
+/* ----------------------- Operations on smallbins ----------------------- */
+
+/*
+  Various forms of linking and unlinking are defined as macros.  Even
+  the ones for trees, which are very long but have very short typical
+  paths.  This is ugly but reduces reliance on inlining support of
+  compilers.
+*/
+
+/* Link a free chunk into a smallbin  */
+#define insert_small_chunk(M, P, S) {\
+  bindex_t I  = small_index(S);\
+  mchunkptr B = smallbin_at(M, I);\
+  mchunkptr F = B;\
+  assert(S >= MIN_CHUNK_SIZE);\
+  if (!smallmap_is_marked(M, I))\
+    mark_smallmap(M, I);\
+  else if (RTCHECK(ok_address(M, B->fd)))\
+    F = B->fd;\
+  else {\
+    CORRUPTION_ERROR_ACTION(M);\
+  }\
+  B->fd = P;\
+  F->bk = P;\
+  P->fd = F;\
+  P->bk = B;\
+}
+
+/* Unlink a chunk from a smallbin  */
+#define unlink_small_chunk(M, P, S) {\
+  mchunkptr F = P->fd;\
+  mchunkptr B = P->bk;\
+  bindex_t I = small_index(S);\
+  assert(P != B);\
+  assert(P != F);\
+  assert(chunksize(P) == small_index2size(I));\
+  if (F == B)\
+    clear_smallmap(M, I);\
+  else if (RTCHECK((F == smallbin_at(M,I) || ok_address(M, F)) &&\
+                   (B == smallbin_at(M,I) || ok_address(M, B)))) {\
+    F->bk = B;\
+    B->fd = F;\
+  }\
+  else {\
+    CORRUPTION_ERROR_ACTION(M);\
+  }\
+}
+
+/* Unlink the first chunk from a smallbin */
+#define unlink_first_small_chunk(M, B, P, I) {\
+  mchunkptr F = P->fd;\
+  assert(P != B);\
+  assert(P != F);\
+  assert(chunksize(P) == small_index2size(I));\
+  if (B == F)\
+    clear_smallmap(M, I);\
+  else if (RTCHECK(ok_address(M, F))) {\
+    B->fd = F;\
+    F->bk = B;\
+  }\
+  else {\
+    CORRUPTION_ERROR_ACTION(M);\
+  }\
+}
+
+
+
+/* Replace dv node, binning the old one */
+/* Used only when dvsize known to be small */
+#define replace_dv(M, P, S) {\
+  size_t DVS = M->dvsize;\
+  if (DVS != 0) {\
+    mchunkptr DV = M->dv;\
+    assert(is_small(DVS));\
+    insert_small_chunk(M, DV, DVS);\
+  }\
+  M->dvsize = S;\
+  M->dv = P;\
+}
+
+/* ------------------------- Operations on trees ------------------------- */
+
+/* Insert chunk into tree */
+#define insert_large_chunk(M, X, S) {\
+  tbinptr* H;\
+  bindex_t I;\
+  compute_tree_index(S, I);\
+  H = treebin_at(M, I);\
+  X->index = I;\
+  X->child[0] = X->child[1] = 0;\
+  if (!treemap_is_marked(M, I)) {\
+    mark_treemap(M, I);\
+    *H = X;\
+    X->parent = (tchunkptr)H;\
+    X->fd = X->bk = X;\
+  }\
+  else {\
+    tchunkptr T = *H;\
+    size_t K = S << leftshift_for_tree_index(I);\
+    for (;;) {\
+      if (chunksize(T) != S) {\
+        tchunkptr* C = &(T->child[(K >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1]);\
+        K <<= 1;\
+        if (*C != 0)\
+          T = *C;\
+        else if (RTCHECK(ok_address(M, C))) {\
+          *C = X;\
+          X->parent = T;\
+          X->fd = X->bk = X;\
+          break;\
+        }\
+        else {\
+          CORRUPTION_ERROR_ACTION(M);\
+          break;\
+        }\
+      }\
+      else {\
+        tchunkptr F = T->fd;\
+        if (RTCHECK(ok_address(M, T) && ok_address(M, F))) {\
+          T->fd = F->bk = X;\
+          X->fd = F;\
+          X->bk = T;\
+          X->parent = 0;\
+          break;\
+        }\
+        else {\
+          CORRUPTION_ERROR_ACTION(M);\
+          break;\
+        }\
+      }\
+    }\
+  }\
+}
+
+/*
+  Unlink steps:
+
+  1. If x is a chained node, unlink it from its same-sized fd/bk links
+     and choose its bk node as its replacement.
+  2. If x was the last node of its size, but not a leaf node, it must
+     be replaced with a leaf node (not merely one with an open left or
+     right), to make sure that lefts and rights of descendents
+     correspond properly to bit masks.  We use the rightmost descendent
+     of x.  We could use any other leaf, but this is easy to locate and
+     tends to counteract removal of leftmosts elsewhere, and so keeps
+     paths shorter than minimally guaranteed.  This doesn't loop much
+     because on average a node in a tree is near the bottom.
+  3. If x is the base of a chain (i.e., has parent links) relink
+     x's parent and children to x's replacement (or null if none).
+*/
+
+#define unlink_large_chunk(M, X) {\
+  tchunkptr XP = X->parent;\
+  tchunkptr R;\
+  if (X->bk != X) {\
+    tchunkptr F = X->fd;\
+    R = X->bk;\
+    if (RTCHECK(ok_address(M, F))) {\
+      F->bk = R;\
+      R->fd = F;\
+    }\
+    else {\
+      CORRUPTION_ERROR_ACTION(M);\
+    }\
+  }\
+  else {\
+    tchunkptr* RP;\
+    if (((R = *(RP = &(X->child[1]))) != 0) ||\
+        ((R = *(RP = &(X->child[0]))) != 0)) {\
+      tchunkptr* CP;\
+      while ((*(CP = &(R->child[1])) != 0) ||\
+             (*(CP = &(R->child[0])) != 0)) {\
+        R = *(RP = CP);\
+      }\
+      if (RTCHECK(ok_address(M, RP)))\
+        *RP = 0;\
+      else {\
+        CORRUPTION_ERROR_ACTION(M);\
+      }\
+    }\
+  }\
+  if (XP != 0) {\
+    tbinptr* H = treebin_at(M, X->index);\
+    if (X == *H) {\
+      if ((*H = R) == 0) \
+        clear_treemap(M, X->index);\
+    }\
+    else if (RTCHECK(ok_address(M, XP))) {\
+      if (XP->child[0] == X) \
+        XP->child[0] = R;\
+      else \
+        XP->child[1] = R;\
+    }\
+    else\
+      CORRUPTION_ERROR_ACTION(M);\
+    if (R != 0) {\
+      if (RTCHECK(ok_address(M, R))) {\
+        tchunkptr C0, C1;\
+        R->parent = XP;\
+        if ((C0 = X->child[0]) != 0) {\
+          if (RTCHECK(ok_address(M, C0))) {\
+            R->child[0] = C0;\
+            C0->parent = R;\
+          }\
+          else\
+            CORRUPTION_ERROR_ACTION(M);\
+        }\
+        if ((C1 = X->child[1]) != 0) {\
+          if (RTCHECK(ok_address(M, C1))) {\
+            R->child[1] = C1;\
+            C1->parent = R;\
+          }\
+          else\
+            CORRUPTION_ERROR_ACTION(M);\
+        }\
+      }\
+      else\
+        CORRUPTION_ERROR_ACTION(M);\
+    }\
+  }\
+}
+
+/* Relays to large vs small bin operations */
+
+#define insert_chunk(M, P, S)\
+  if (is_small(S)) insert_small_chunk(M, P, S)\
+  else { tchunkptr TP = (tchunkptr)(P); insert_large_chunk(M, TP, S); }
+
+#define unlink_chunk(M, P, S)\
+  if (is_small(S)) unlink_small_chunk(M, P, S)\
+  else { tchunkptr TP = (tchunkptr)(P); unlink_large_chunk(M, TP); }
+
+
+/* Relays to internal calls to malloc/free from realloc, memalign etc */
+
+#if ONLY_MSPACES
+#define internal_malloc(m, b) mspace_malloc(m, b)
+#define internal_free(m, mem) mspace_free(m,mem);
+#else /* ONLY_MSPACES */
+#if MSPACES
+#define internal_malloc(m, b)\
+   (m == gm)? dlmalloc(b) : mspace_malloc(m, b)
+#define internal_free(m, mem)\
+   if (m == gm) dlfree(mem); else mspace_free(m,mem);
+#else /* MSPACES */
+#define internal_malloc(m, b) dlmalloc(b)
+#define internal_free(m, mem) dlfree(mem)
+#endif /* MSPACES */
+#endif /* ONLY_MSPACES */
+
+/* -----------------------  Direct-mmapping chunks ----------------------- */
+
+/*
+  Directly mmapped chunks are set up with an offset to the start of
+  the mmapped region stored in the prev_foot field of the chunk. This
+  allows reconstruction of the required argument to MUNMAP when freed,
+  and also allows adjustment of the returned chunk to meet alignment
+  requirements (especially in memalign).
+*/
+
+/* Malloc using mmap */
+static void* mmap_alloc(mstate m, size_t nb) {
+  size_t mmsize = mmap_align(nb + SIX_SIZE_T_SIZES + CHUNK_ALIGN_MASK);
+  if (mmsize > nb) {     /* Check for wrap around 0 */
+    char* mm = (char*)(CALL_DIRECT_MMAP(mmsize));
+    if (mm != CMFAIL) {
+      size_t offset = align_offset(chunk2mem(mm));
+      size_t psize = mmsize - offset - MMAP_FOOT_PAD;
+      mchunkptr p = (mchunkptr)(mm + offset);
+      p->prev_foot = offset;
+      p->head = psize;
+      mark_inuse_foot(m, p, psize);
+      chunk_plus_offset(p, psize)->head = FENCEPOST_HEAD;
+      chunk_plus_offset(p, psize+SIZE_T_SIZE)->head = 0;
+
+      if (m->least_addr == 0 || mm < m->least_addr)
+        m->least_addr = mm;
+      if ((m->footprint += mmsize) > m->max_footprint)
+        m->max_footprint = m->footprint;
+      assert(is_aligned(chunk2mem(p)));
+      check_mmapped_chunk(m, p);
+      return chunk2mem(p);
+    }
+  }
+  return 0;
+}
+
+/* Realloc using mmap */
+static mchunkptr mmap_resize(mstate m, mchunkptr oldp, size_t nb) {
+  size_t oldsize = chunksize(oldp);
+  if (is_small(nb)) /* Can't shrink mmap regions below small size */
+    return 0;
+  /* Keep old chunk if big enough but not too big */
+  if (oldsize >= nb + SIZE_T_SIZE &&
+      (oldsize - nb) <= (mparams.granularity << 1))
+    return oldp;
+  else {
+    size_t offset = oldp->prev_foot;
+    size_t oldmmsize = oldsize + offset + MMAP_FOOT_PAD;
+    size_t newmmsize = mmap_align(nb + SIX_SIZE_T_SIZES + CHUNK_ALIGN_MASK);
+    char* cp = (char*)CALL_MREMAP((char*)oldp - offset,
+                                  oldmmsize, newmmsize, 1);
+    if (cp != CMFAIL) {
+      mchunkptr newp = (mchunkptr)(cp + offset);
+      size_t psize = newmmsize - offset - MMAP_FOOT_PAD;
+      newp->head = psize;
+      mark_inuse_foot(m, newp, psize);
+      chunk_plus_offset(newp, psize)->head = FENCEPOST_HEAD;
+      chunk_plus_offset(newp, psize+SIZE_T_SIZE)->head = 0;
+
+      if (cp < m->least_addr)
+        m->least_addr = cp;
+      if ((m->footprint += newmmsize - oldmmsize) > m->max_footprint)
+        m->max_footprint = m->footprint;
+      check_mmapped_chunk(m, newp);
+      return newp;
+    }
+  }
+  return 0;
+}
+
+/* -------------------------- mspace management -------------------------- */
+
+/* Initialize top chunk and its size */
+static void init_top(mstate m, mchunkptr p, size_t psize) {
+  /* Ensure alignment */
+  size_t offset = align_offset(chunk2mem(p));
+  p = (mchunkptr)((char*)p + offset);
+  psize -= offset;
+
+  m->top = p;
+  m->topsize = psize;
+  p->head = psize | PINUSE_BIT;
+  /* set size of fake trailing chunk holding overhead space only once */
+  chunk_plus_offset(p, psize)->head = TOP_FOOT_SIZE;
+  m->trim_check = mparams.trim_threshold; /* reset on each update */
+}
+
+/* Initialize bins for a new mstate that is otherwise zeroed out */
+static void init_bins(mstate m) {
+  /* Establish circular links for smallbins */
+  bindex_t i;
+  for (i = 0; i < NSMALLBINS; ++i) {
+    sbinptr bin = smallbin_at(m,i);
+    bin->fd = bin->bk = bin;
+  }
+}
+
+#if PROCEED_ON_ERROR
+
+/* default corruption action */
+static void reset_on_error(mstate m) {
+  int i;
+  ++malloc_corruption_error_count;
+  /* Reinitialize fields to forget about all memory */
+  m->smallbins = m->treebins = 0;
+  m->dvsize = m->topsize = 0;
+  m->seg.base = 0;
+  m->seg.size = 0;
+  m->seg.next = 0;
+  m->top = m->dv = 0;
+  for (i = 0; i < NTREEBINS; ++i)
+    *treebin_at(m, i) = 0;
+  init_bins(m);
+}
+#endif /* PROCEED_ON_ERROR */
+
+/* Allocate chunk and prepend remainder with chunk in successor base. */
+static void* prepend_alloc(mstate m, char* newbase, char* oldbase,
+                           size_t nb) {
+  mchunkptr p = align_as_chunk(newbase);
+  mchunkptr oldfirst = align_as_chunk(oldbase);
+  size_t psize = (char*)oldfirst - (char*)p;
+  mchunkptr q = chunk_plus_offset(p, nb);
+  size_t qsize = psize - nb;
+  set_size_and_pinuse_of_inuse_chunk(m, p, nb);
+
+  assert((char*)oldfirst > (char*)q);
+  assert(pinuse(oldfirst));
+  assert(qsize >= MIN_CHUNK_SIZE);
+
+  /* consolidate remainder with first chunk of old base */
+  if (oldfirst == m->top) {
+    size_t tsize = m->topsize += qsize;
+    m->top = q;
+    q->head = tsize | PINUSE_BIT;
+    check_top_chunk(m, q);
+  }
+  else if (oldfirst == m->dv) {
+    size_t dsize = m->dvsize += qsize;
+    m->dv = q;
+    set_size_and_pinuse_of_free_chunk(q, dsize);
+  }
+  else {
+    if (!is_inuse(oldfirst)) {
+      size_t nsize = chunksize(oldfirst);
+      unlink_chunk(m, oldfirst, nsize);
+      oldfirst = chunk_plus_offset(oldfirst, nsize);
+      qsize += nsize;
+    }
+    set_free_with_pinuse(q, qsize, oldfirst);
+    insert_chunk(m, q, qsize);
+    check_free_chunk(m, q);
+  }
+
+  check_malloced_chunk(m, chunk2mem(p), nb);
+  return chunk2mem(p);
+}
+
+/* Add a segment to hold a new noncontiguous region */
+static void add_segment(mstate m, char* tbase, size_t tsize, flag_t mmapped) {
+  /* Determine locations and sizes of segment, fenceposts, old top */
+  char* old_top = (char*)m->top;
+  msegmentptr oldsp = segment_holding(m, old_top);
+  char* old_end = oldsp->base + oldsp->size;
+  size_t ssize = pad_request(sizeof(struct malloc_segment));
+  char* rawsp = old_end - (ssize + FOUR_SIZE_T_SIZES + CHUNK_ALIGN_MASK);
+  size_t offset = align_offset(chunk2mem(rawsp));
+  char* asp = rawsp + offset;
+  char* csp = (asp < (old_top + MIN_CHUNK_SIZE))? old_top : asp;
+  mchunkptr sp = (mchunkptr)csp;
+  msegmentptr ss = (msegmentptr)(chunk2mem(sp));
+  mchunkptr tnext = chunk_plus_offset(sp, ssize);
+  mchunkptr p = tnext;
+  int nfences = 0;
+
+  /* reset top to new space */
+  init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);
+
+  /* Set up segment record */
+  assert(is_aligned(ss));
+  set_size_and_pinuse_of_inuse_chunk(m, sp, ssize);
+  *ss = m->seg; /* Push current record */
+  m->seg.base = tbase;
+  m->seg.size = tsize;
+  m->seg.sflags = mmapped;
+  m->seg.next = ss;
+
+  /* Insert trailing fenceposts */
+  for (;;) {
+    mchunkptr nextp = chunk_plus_offset(p, SIZE_T_SIZE);
+    p->head = FENCEPOST_HEAD;
+    ++nfences;
+    if ((char*)(&(nextp->head)) < old_end)
+      p = nextp;
+    else
+      break;
+  }
+  assert(nfences >= 2);
+
+  /* Insert the rest of old top into a bin as an ordinary free chunk */
+  if (csp != old_top) {
+    mchunkptr q = (mchunkptr)old_top;
+    size_t psize = csp - old_top;
+    mchunkptr tn = chunk_plus_offset(q, psize);
+    set_free_with_pinuse(q, psize, tn);
+    insert_chunk(m, q, psize);
+  }
+
+  check_top_chunk(m, m->top);
+}
+
+/* -------------------------- System allocation -------------------------- */
+
+/* Get memory from system using MORECORE or MMAP */
+static void* sys_alloc(mstate m, size_t nb) {
+  char* tbase = CMFAIL;
+  size_t tsize = 0;
+  flag_t mmap_flag = 0;
+
+  ensure_initialization();
+
+  /* Directly map large chunks, but only if already initialized */
+  if (use_mmap(m) && nb >= mparams.mmap_threshold && m->topsize != 0) {
+    void* mem = mmap_alloc(m, nb);
+    if (mem != 0)
+      return mem;
+  }
+
+  /*
+    Try getting memory in any of three ways (in most-preferred to
+    least-preferred order):
+    1. A call to MORECORE that can normally contiguously extend memory.
+       (disabled if not MORECORE_CONTIGUOUS or not HAVE_MORECORE or
+       or main space is mmapped or a previous contiguous call failed)
+    2. A call to MMAP new space (disabled if not HAVE_MMAP).
+       Note that under the default settings, if MORECORE is unable to
+       fulfill a request, and HAVE_MMAP is true, then mmap is
+       used as a noncontiguous system allocator. This is a useful backup
+       strategy for systems with holes in address spaces -- in this case
+       sbrk cannot contiguously expand the heap, but mmap may be able to
+       find space.
+    3. A call to MORECORE that cannot usually contiguously extend memory.
+       (disabled if not HAVE_MORECORE)
+
+   In all cases, we need to request enough bytes from system to ensure
+   we can malloc nb bytes upon success, so pad with enough space for
+   top_foot, plus alignment-pad to make sure we don't lose bytes if
+   not on boundary, and round this up to a granularity unit.
+  */
+
+  if (MORECORE_CONTIGUOUS && !use_noncontiguous(m)) {
+    char* br = CMFAIL;
+    msegmentptr ss = (m->top == 0)? 0 : segment_holding(m, (char*)m->top);
+    size_t asize = 0;
+    ACQUIRE_MALLOC_GLOBAL_LOCK();
+
+    if (ss == 0) {  /* First time through or recovery */
+      char* base = (char*)CALL_MORECORE(0);
+      if (base != CMFAIL) {
+        asize = granularity_align(nb + SYS_ALLOC_PADDING);
+        /* Adjust to end on a page boundary */
+        if (!is_page_aligned(base))
+          asize += (page_align((size_t)base) - (size_t)base);
+        /* Can't call MORECORE if size is negative when treated as signed */
+        if (asize < HALF_MAX_SIZE_T &&
+            (br = (char*)(CALL_MORECORE(asize))) == base) {
+          tbase = base;
+          tsize = asize;
+        }
+      }
+    }
+    else {
+      /* Subtract out existing available top space from MORECORE request. */
+      asize = granularity_align(nb - m->topsize + SYS_ALLOC_PADDING);
+      /* Use mem here only if it did continuously extend old space */
+      if (asize < HALF_MAX_SIZE_T &&
+          (br = (char*)(CALL_MORECORE(asize))) == ss->base+ss->size) {
+        tbase = br;
+        tsize = asize;
+      }
+    }
+
+    if (tbase == CMFAIL) {    /* Cope with partial failure */
+      if (br != CMFAIL) {    /* Try to use/extend the space we did get */
+        if (asize < HALF_MAX_SIZE_T &&
+            asize < nb + SYS_ALLOC_PADDING) {
+          size_t esize = granularity_align(nb + SYS_ALLOC_PADDING - asize);
+          if (esize < HALF_MAX_SIZE_T) {
+            char* end = (char*)CALL_MORECORE(esize);
+            if (end != CMFAIL)
+              asize += esize;
+            else {            /* Can't use; try to release */
+              (void) CALL_MORECORE(-asize);
+              br = CMFAIL;
+            }
+          }
+        }
+      }
+      if (br != CMFAIL) {    /* Use the space we did get */
+        tbase = br;
+        tsize = asize;
+      }
+      else
+        disable_contiguous(m); /* Don't try contiguous path in the future */
+    }
+
+    RELEASE_MALLOC_GLOBAL_LOCK();
+  }
+
+  if (HAVE_MMAP && tbase == CMFAIL) {  /* Try MMAP */
+    size_t rsize = granularity_align(nb + SYS_ALLOC_PADDING);
+    if (rsize > nb) { /* Fail if wraps around zero */
+      char* mp = (char*)(CALL_MMAP(rsize));
+      if (mp != CMFAIL) {
+        tbase = mp;
+        tsize = rsize;
+        mmap_flag = USE_MMAP_BIT;
+      }
+    }
+  }
+
+  if (HAVE_MORECORE && tbase == CMFAIL) { /* Try noncontiguous MORECORE */
+    size_t asize = granularity_align(nb + SYS_ALLOC_PADDING);
+    if (asize < HALF_MAX_SIZE_T) {
+      char* br = CMFAIL;
+      char* end = CMFAIL;
+      ACQUIRE_MALLOC_GLOBAL_LOCK();
+      br = (char*)(CALL_MORECORE(asize));
+      end = (char*)(CALL_MORECORE(0));
+      RELEASE_MALLOC_GLOBAL_LOCK();
+      if (br != CMFAIL && end != CMFAIL && br < end) {
+        size_t ssize = end - br;
+        if (ssize > nb + TOP_FOOT_SIZE) {
+          tbase = br;
+          tsize = ssize;
+        }
+      }
+    }
+  }
+
+  if (tbase != CMFAIL) {
+
+    if ((m->footprint += tsize) > m->max_footprint)
+      m->max_footprint = m->footprint;
+
+    if (!is_initialized(m)) { /* first-time initialization */
+      if (m->least_addr == 0 || tbase < m->least_addr)
+        m->least_addr = tbase;
+      m->seg.base = tbase;
+      m->seg.size = tsize;
+      m->seg.sflags = mmap_flag;
+      m->magic = mparams.magic;
+      m->release_checks = MAX_RELEASE_CHECK_RATE;
+      init_bins(m);
+#if !ONLY_MSPACES
+      if (is_global(m))
+        init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);
+      else
+#endif
+      {
+        /* Offset top by embedded malloc_state */
+        mchunkptr mn = next_chunk(mem2chunk(m));
+        init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) -TOP_FOOT_SIZE);
+      }
+    }
+
+    else {
+      /* Try to merge with an existing segment */
+      msegmentptr sp = &m->seg;
+      /* Only consider most recent segment if traversal suppressed */
+      while (sp != 0 && tbase != sp->base + sp->size)
+        sp = (NO_SEGMENT_TRAVERSAL) ? 0 : sp->next;
+      if (sp != 0 &&
+          !is_extern_segment(sp) &&
+          (sp->sflags & USE_MMAP_BIT) == mmap_flag &&
+          segment_holds(sp, m->top)) { /* append */
+        sp->size += tsize;
+        init_top(m, m->top, m->topsize + tsize);
+      }
+      else {
+        if (tbase < m->least_addr)
+          m->least_addr = tbase;
+        sp = &m->seg;
+        while (sp != 0 && sp->base != tbase + tsize)
+          sp = (NO_SEGMENT_TRAVERSAL) ? 0 : sp->next;
+        if (sp != 0 &&
+            !is_extern_segment(sp) &&
+            (sp->sflags & USE_MMAP_BIT) == mmap_flag) {
+          char* oldbase = sp->base;
+          sp->base = tbase;
+          sp->size += tsize;
+          return prepend_alloc(m, tbase, oldbase, nb);
+        }
+        else
+          add_segment(m, tbase, tsize, mmap_flag);
+      }
+    }
+
+    if (nb < m->topsize) { /* Allocate from new or extended top space */
+      size_t rsize = m->topsize -= nb;
+      mchunkptr p = m->top;
+      mchunkptr r = m->top = chunk_plus_offset(p, nb);
+      r->head = rsize | PINUSE_BIT;
+      set_size_and_pinuse_of_inuse_chunk(m, p, nb);
+      check_top_chunk(m, m->top);
+      check_malloced_chunk(m, chunk2mem(p), nb);
+      return chunk2mem(p);
+    }
+  }
+
+  MALLOC_FAILURE_ACTION;
+  return 0;
+}
+
+/* -----------------------  system deallocation -------------------------- */
+
+/* Unmap and unlink any mmapped segments that don't contain used chunks */
+static size_t release_unused_segments(mstate m) {
+  size_t released = 0;
+  int nsegs = 0;
+  msegmentptr pred = &m->seg;
+  msegmentptr sp = pred->next;
+  while (sp != 0) {
+    char* base = sp->base;
+    size_t size = sp->size;
+    msegmentptr next = sp->next;
+    ++nsegs;
+    if (is_mmapped_segment(sp) && !is_extern_segment(sp)) {
+      mchunkptr p = align_as_chunk(base);
+      size_t psize = chunksize(p);
+      /* Can unmap if first chunk holds entire segment and not pinned */
+      if (!is_inuse(p) && (char*)p + psize >= base + size - TOP_FOOT_SIZE) {
+        tchunkptr tp = (tchunkptr)p;
+        assert(segment_holds(sp, (char*)sp));
+        if (p == m->dv) {
+          m->dv = 0;
+          m->dvsize = 0;
+        }
+        else {
+          unlink_large_chunk(m, tp);
+        }
+        if (CALL_MUNMAP(base, size) == 0) {
+          released += size;
+          m->footprint -= size;
+          /* unlink obsoleted record */
+          sp = pred;
+          sp->next = next;
+        }
+        else { /* back out if cannot unmap */
+          insert_large_chunk(m, tp, psize);
+        }
+      }
+    }
+    if (NO_SEGMENT_TRAVERSAL) /* scan only first segment */
+      break;
+    pred = sp;
+    sp = next;
+  }
+  /* Reset check counter */
+  m->release_checks = ((nsegs > MAX_RELEASE_CHECK_RATE)?
+                       nsegs : MAX_RELEASE_CHECK_RATE);
+  return released;
+}
+
+static int sys_trim(mstate m, size_t pad) {
+  size_t released = 0;
+  ensure_initialization();
+  if (pad < MAX_REQUEST && is_initialized(m)) {
+    pad += TOP_FOOT_SIZE; /* ensure enough room for segment overhead */
+
+    if (m->topsize > pad) {
+      /* Shrink top space in granularity-size units, keeping at least one */
+      size_t unit = mparams.granularity;
+      size_t extra = ((m->topsize - pad + (unit - SIZE_T_ONE)) / unit -
+                      SIZE_T_ONE) * unit;
+      msegmentptr sp = segment_holding(m, (char*)m->top);
+
+      if (!is_extern_segment(sp)) {
+        if (is_mmapped_segment(sp)) {
+          if (HAVE_MMAP &&
+              sp->size >= extra &&
+              !has_segment_link(m, sp)) { /* can't shrink if pinned */
+            size_t newsize = sp->size - extra;
+            /* Prefer mremap, fall back to munmap */
+            if ((CALL_MREMAP(sp->base, sp->size, newsize, 0) != MFAIL) ||
+                (CALL_MUNMAP(sp->base + newsize, extra) == 0)) {
+              released = extra;
+            }
+          }
+        }
+        else if (HAVE_MORECORE) {
+          if (extra >= HALF_MAX_SIZE_T) /* Avoid wrapping negative */
+            extra = (HALF_MAX_SIZE_T) + SIZE_T_ONE - unit;
+          ACQUIRE_MALLOC_GLOBAL_LOCK();
+          {
+            /* Make sure end of memory is where we last set it. */
+            char* old_br = (char*)(CALL_MORECORE(0));
+            if (old_br == sp->base + sp->size) {
+              char* rel_br = (char*)(CALL_MORECORE(-extra));
+              char* new_br = (char*)(CALL_MORECORE(0));
+              if (rel_br != CMFAIL && new_br < old_br)
+                released = old_br - new_br;
+            }
+          }
+          RELEASE_MALLOC_GLOBAL_LOCK();
+        }
+      }
+
+      if (released != 0) {
+        sp->size -= released;
+        m->footprint -= released;
+        init_top(m, m->top, m->topsize - released);
+        check_top_chunk(m, m->top);
+      }
+    }
+
+    /* Unmap any unused mmapped segments */
+    if (HAVE_MMAP)
+      released += release_unused_segments(m);
+
+    /* On failure, disable autotrim to avoid repeated failed future calls */
+    if (released == 0 && m->topsize > m->trim_check)
+      m->trim_check = MAX_SIZE_T;
+  }
+
+  return (released != 0)? 1 : 0;
+}
+
+
+/* ---------------------------- malloc support --------------------------- */
+
+/* allocate a large request from the best fitting chunk in a treebin */
+static void* tmalloc_large(mstate m, size_t nb) {
+  tchunkptr v = 0;
+  size_t rsize = -nb; /* Unsigned negation */
+  tchunkptr t;
+  bindex_t idx;
+  compute_tree_index(nb, idx);
+  if ((t = *treebin_at(m, idx)) != 0) {
+    /* Traverse tree for this bin looking for node with size == nb */
+    size_t sizebits = nb << leftshift_for_tree_index(idx);
+    tchunkptr rst = 0;  /* The deepest untaken right subtree */
+    for (;;) {
+      tchunkptr rt;
+      size_t trem = chunksize(t) - nb;
+      if (trem < rsize) {
+        v = t;
+        if ((rsize = trem) == 0)
+          break;
+      }
+      rt = t->child[1];
+      t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];
+      if (rt != 0 && rt != t)
+        rst = rt;
+      if (t == 0) {
+        t = rst; /* set t to least subtree holding sizes > nb */
+        break;
+      }
+      sizebits <<= 1;
+    }
+  }
+  if (t == 0 && v == 0) { /* set t to root of next non-empty treebin */
+    binmap_t leftbits = left_bits(idx2bit(idx)) & m->treemap;
+    if (leftbits != 0) {
+      bindex_t i;
+      binmap_t leastbit = least_bit(leftbits);
+      compute_bit2idx(leastbit, i);
+      t = *treebin_at(m, i);
+    }
+  }
+
+  while (t != 0) { /* find smallest of tree or subtree */
+    size_t trem = chunksize(t) - nb;
+    if (trem < rsize) {
+      rsize = trem;
+      v = t;
+    }
+    t = leftmost_child(t);
+  }
+
+  /*  If dv is a better fit, return 0 so malloc will use it */
+  if (v != 0 && rsize < (size_t)(m->dvsize - nb)) {
+    if (RTCHECK(ok_address(m, v))) { /* split */
+      mchunkptr r = chunk_plus_offset(v, nb);
+      assert(chunksize(v) == rsize + nb);
+      if (RTCHECK(ok_next(v, r))) {
+        unlink_large_chunk(m, v);
+        if (rsize < MIN_CHUNK_SIZE)
+          set_inuse_and_pinuse(m, v, (rsize + nb));
+        else {
+          set_size_and_pinuse_of_inuse_chunk(m, v, nb);
+          set_size_and_pinuse_of_free_chunk(r, rsize);
+          insert_chunk(m, r, rsize);
+        }
+        return chunk2mem(v);
+      }
+    }
+    CORRUPTION_ERROR_ACTION(m);
+  }
+  return 0;
+}
+
+/* allocate a small request from the best fitting chunk in a treebin */
+static void* tmalloc_small(mstate m, size_t nb) {
+  tchunkptr t, v;
+  size_t rsize;
+  bindex_t i;
+  binmap_t leastbit = least_bit(m->treemap);
+  compute_bit2idx(leastbit, i);
+  v = t = *treebin_at(m, i);
+  rsize = chunksize(t) - nb;
+
+  while ((t = leftmost_child(t)) != 0) {
+    size_t trem = chunksize(t) - nb;
+    if (trem < rsize) {
+      rsize = trem;
+      v = t;
+    }
+  }
+
+  if (RTCHECK(ok_address(m, v))) {
+    mchunkptr r = chunk_plus_offset(v, nb);
+    assert(chunksize(v) == rsize + nb);
+    if (RTCHECK(ok_next(v, r))) {
+      unlink_large_chunk(m, v);
+      if (rsize < MIN_CHUNK_SIZE)
+        set_inuse_and_pinuse(m, v, (rsize + nb));
+      else {
+        set_size_and_pinuse_of_inuse_chunk(m, v, nb);
+        set_size_and_pinuse_of_free_chunk(r, rsize);
+        replace_dv(m, r, rsize);
+      }
+      return chunk2mem(v);
+    }
+  }
+
+  CORRUPTION_ERROR_ACTION(m);
+  return 0;
+}
+
+/* --------------------------- realloc support --------------------------- */
+
+static void* internal_realloc(mstate m, void* oldmem, size_t bytes) {
+  if (bytes >= MAX_REQUEST) {
+    MALLOC_FAILURE_ACTION;
+    return 0;
+  }
+  if (!PREACTION(m)) {
+    mchunkptr oldp = mem2chunk(oldmem);
+    size_t oldsize = chunksize(oldp);
+    mchunkptr next = chunk_plus_offset(oldp, oldsize);
+    mchunkptr newp = 0;
+    void* extra = 0;
+
+    /* Try to either shrink or extend into top. Else malloc-copy-free */
+
+    if (RTCHECK(ok_address(m, oldp) && ok_inuse(oldp) &&
+                ok_next(oldp, next) && ok_pinuse(next))) {
+      size_t nb = request2size(bytes);
+      if (is_mmapped(oldp))
+        newp = mmap_resize(m, oldp, nb);
+      else if (oldsize >= nb) { /* already big enough */
+        size_t rsize = oldsize - nb;
+        newp = oldp;
+        if (rsize >= MIN_CHUNK_SIZE) {
+          mchunkptr remainder = chunk_plus_offset(newp, nb);
+          set_inuse(m, newp, nb);
+          set_inuse_and_pinuse(m, remainder, rsize);
+          extra = chunk2mem(remainder);
+        }
+      }
+      else if (next == m->top && oldsize + m->topsize > nb) {
+        /* Expand into top */
+        size_t newsize = oldsize + m->topsize;
+        size_t newtopsize = newsize - nb;
+        mchunkptr newtop = chunk_plus_offset(oldp, nb);
+        set_inuse(m, oldp, nb);
+        newtop->head = newtopsize |PINUSE_BIT;
+        m->top = newtop;
+        m->topsize = newtopsize;
+        newp = oldp;
+      }
+    }
+    else {
+      USAGE_ERROR_ACTION(m, oldmem);
+      POSTACTION(m);
+      return 0;
+    }
+#if DEBUG
+    if (newp != 0) {
+      check_inuse_chunk(m, newp); /* Check requires lock */
+    }
+#endif
+
+    POSTACTION(m);
+
+    if (newp != 0) {
+      if (extra != 0) {
+        internal_free(m, extra);
+      }
+      return chunk2mem(newp);
+    }
+    else {
+      void* newmem = internal_malloc(m, bytes);
+      if (newmem != 0) {
+        size_t oc = oldsize - overhead_for(oldp);
+        memcpy(newmem, oldmem, (oc < bytes)? oc : bytes);
+        internal_free(m, oldmem);
+      }
+      return newmem;
+    }
+  }
+  return 0;
+}
+
+/* --------------------------- memalign support -------------------------- */
+
+static void* internal_memalign(mstate m, size_t alignment, size_t bytes) {
+  if (alignment <= MALLOC_ALIGNMENT)    /* Can just use malloc */
+    return internal_malloc(m, bytes);
+  if (alignment <  MIN_CHUNK_SIZE) /* must be at least a minimum chunk size */
+    alignment = MIN_CHUNK_SIZE;
+  if ((alignment & (alignment-SIZE_T_ONE)) != 0) {/* Ensure a power of 2 */
+    size_t a = MALLOC_ALIGNMENT << 1;
+    while (a < alignment) a <<= 1;
+    alignment = a;
+  }
+
+  if (bytes >= MAX_REQUEST - alignment) {
+    if (m != 0)  { /* Test isn't needed but avoids compiler warning */
+      MALLOC_FAILURE_ACTION;
+    }
+  }
+  else {
+    size_t nb = request2size(bytes);
+    size_t req = nb + alignment + MIN_CHUNK_SIZE - CHUNK_OVERHEAD;
+    char* mem = (char*)internal_malloc(m, req);
+    if (mem != 0) {
+      void* leader = 0;
+      void* trailer = 0;
+      mchunkptr p = mem2chunk(mem);
+
+      if (PREACTION(m)) return 0;
+      if ((((size_t)(mem)) % alignment) != 0) { /* misaligned */
+        /*
+          Find an aligned spot inside chunk.  Since we need to give
+          back leading space in a chunk of at least MIN_CHUNK_SIZE, if
+          the first calculation places us at a spot with less than
+          MIN_CHUNK_SIZE leader, we can move to the next aligned spot.
+          We've allocated enough total room so that this is always
+          possible.
+        */
+        char* br = (char*)mem2chunk((size_t)(((size_t)(mem +
+                                                       alignment -
+                                                       SIZE_T_ONE)) &
+                                             -alignment));
+        char* pos = ((size_t)(br - (char*)(p)) >= MIN_CHUNK_SIZE)?
+          br : br+alignment;
+        mchunkptr newp = (mchunkptr)pos;
+        size_t leadsize = pos - (char*)(p);
+        size_t newsize = chunksize(p) - leadsize;
+
+        if (is_mmapped(p)) { /* For mmapped chunks, just adjust offset */
+          newp->prev_foot = p->prev_foot + leadsize;
+          newp->head = newsize;
+        }
+        else { /* Otherwise, give back leader, use the rest */
+          set_inuse(m, newp, newsize);
+          set_inuse(m, p, leadsize);
+          leader = chunk2mem(p);
+        }
+        p = newp;
+      }
+
+      /* Give back spare room at the end */
+      if (!is_mmapped(p)) {
+        size_t size = chunksize(p);
+        if (size > nb + MIN_CHUNK_SIZE) {
+          size_t remainder_size = size - nb;
+          mchunkptr remainder = chunk_plus_offset(p, nb);
+          set_inuse(m, p, nb);
+          set_inuse(m, remainder, remainder_size);
+          trailer = chunk2mem(remainder);
+        }
+      }
+
+      assert (chunksize(p) >= nb);
+      assert((((size_t)(chunk2mem(p))) % alignment) == 0);
+      check_inuse_chunk(m, p);
+      POSTACTION(m);
+      if (leader != 0) {
+        internal_free(m, leader);
+      }
+      if (trailer != 0) {
+        internal_free(m, trailer);
+      }
+      return chunk2mem(p);
+    }
+  }
+  return 0;
+}
+
+/* ------------------------ comalloc/coalloc support --------------------- */
+
+static void** ialloc(mstate m,
+                     size_t n_elements,
+                     size_t* sizes,
+                     int opts,
+                     void* chunks[]) {
+  /*
+    This provides common support for independent_X routines, handling
+    all of the combinations that can result.
+
+    The opts arg has:
+    bit 0 set if all elements are same size (using sizes[0])
+    bit 1 set if elements should be zeroed
+  */
+
+  size_t    element_size;   /* chunksize of each element, if all same */
+  size_t    contents_size;  /* total size of elements */
+  size_t    array_size;     /* request size of pointer array */
+  void*     mem;            /* malloced aggregate space */
+  mchunkptr p;              /* corresponding chunk */
+  size_t    remainder_size; /* remaining bytes while splitting */
+  void**    marray;         /* either "chunks" or malloced ptr array */
+  mchunkptr array_chunk;    /* chunk for malloced ptr array */
+  flag_t    was_enabled;    /* to disable mmap */
+  size_t    size;
+  size_t    i;
+
+  ensure_initialization();
+  /* compute array length, if needed */
+  if (chunks != 0) {
+    if (n_elements == 0)
+      return chunks; /* nothing to do */
+    marray = chunks;
+    array_size = 0;
+  }
+  else {
+    /* if empty req, must still return chunk representing empty array */
+    if (n_elements == 0)
+      return (void**)internal_malloc(m, 0);
+    marray = 0;
+    array_size = request2size(n_elements * (sizeof(void*)));
+  }
+
+  /* compute total element size */
+  if (opts & 0x1) { /* all-same-size */
+    element_size = request2size(*sizes);
+    contents_size = n_elements * element_size;
+  }
+  else { /* add up all the sizes */
+    element_size = 0;
+    contents_size = 0;
+    for (i = 0; i != n_elements; ++i)
+      contents_size += request2size(sizes[i]);
+  }
+
+  size = contents_size + array_size;
+
+  /*
+     Allocate the aggregate chunk.  First disable direct-mmapping so
+     malloc won't use it, since we would not be able to later
+     free/realloc space internal to a segregated mmap region.
+  */
+  was_enabled = use_mmap(m);
+  disable_mmap(m);
+  mem = internal_malloc(m, size - CHUNK_OVERHEAD);
+  if (was_enabled)
+    enable_mmap(m);
+  if (mem == 0)
+    return 0;
+
+  if (PREACTION(m)) return 0;
+  p = mem2chunk(mem);
+  remainder_size = chunksize(p);
+
+  assert(!is_mmapped(p));
+
+  if (opts & 0x2) {       /* optionally clear the elements */
+    memset((size_t*)mem, 0, remainder_size - SIZE_T_SIZE - array_size);
+  }
+
+  /* If not provided, allocate the pointer array as final part of chunk */
+  if (marray == 0) {
+    size_t  array_chunk_size;
+    array_chunk = chunk_plus_offset(p, contents_size);
+    array_chunk_size = remainder_size - contents_size;
+    marray = (void**) (chunk2mem(array_chunk));
+    set_size_and_pinuse_of_inuse_chunk(m, array_chunk, array_chunk_size);
+    remainder_size = contents_size;
+  }
+
+  /* split out elements */
+  for (i = 0; ; ++i) {
+    marray[i] = chunk2mem(p);
+    if (i != n_elements-1) {
+      if (element_size != 0)
+        size = element_size;
+      else
+        size = request2size(sizes[i]);
+      remainder_size -= size;
+      set_size_and_pinuse_of_inuse_chunk(m, p, size);
+      p = chunk_plus_offset(p, size);
+    }
+    else { /* the final element absorbs any overallocation slop */
+      set_size_and_pinuse_of_inuse_chunk(m, p, remainder_size);
+      break;
+    }
+  }
+
+#if DEBUG
+  if (marray != chunks) {
+    /* final element must have exactly exhausted chunk */
+    if (element_size != 0) {
+      assert(remainder_size == element_size);
+    }
+    else {
+      assert(remainder_size == request2size(sizes[i]));
+    }
+    check_inuse_chunk(m, mem2chunk(marray));
+  }
+  for (i = 0; i != n_elements; ++i)
+    check_inuse_chunk(m, mem2chunk(marray[i]));
+
+#endif /* DEBUG */
+
+  POSTACTION(m);
+  return marray;
+}
+
+
+/* -------------------------- public routines ---------------------------- */
+
+#if !ONLY_MSPACES
+
+void* dlmalloc(size_t bytes) {
+  /*
+     Basic algorithm:
+     If a small request (< 256 bytes minus per-chunk overhead):
+       1. If one exists, use a remainderless chunk in associated smallbin.
+          (Remainderless means that there are too few excess bytes to
+          represent as a chunk.)
+       2. If it is big enough, use the dv chunk, which is normally the
+          chunk adjacent to the one used for the most recent small request.
+       3. If one exists, split the smallest available chunk in a bin,
+          saving remainder in dv.
+       4. If it is big enough, use the top chunk.
+       5. If available, get memory from system and use it
+     Otherwise, for a large request:
+       1. Find the smallest available binned chunk that fits, and use it
+          if it is better fitting than dv chunk, splitting if necessary.
+       2. If better fitting than any binned chunk, use the dv chunk.
+       3. If it is big enough, use the top chunk.
+       4. If request size >= mmap threshold, try to directly mmap this chunk.
+       5. If available, get memory from system and use it
+
+     The ugly goto's here ensure that postaction occurs along all paths.
+  */
+
+#if USE_LOCKS
+  ensure_initialization(); /* initialize in sys_alloc if not using locks */
+#endif
+
+  if (!PREACTION(gm)) {
+    void* mem;
+    size_t nb;
+    if (bytes <= MAX_SMALL_REQUEST) {
+      bindex_t idx;
+      binmap_t smallbits;
+      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);
+      idx = small_index(nb);
+      smallbits = gm->smallmap >> idx;
+
+      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */
+        mchunkptr b, p;
+        idx += ~smallbits & 1;       /* Uses next bin if idx empty */
+        b = smallbin_at(gm, idx);
+        p = b->fd;
+        assert(chunksize(p) == small_index2size(idx));
+        unlink_first_small_chunk(gm, b, p, idx);
+        set_inuse_and_pinuse(gm, p, small_index2size(idx));
+        mem = chunk2mem(p);
+        check_malloced_chunk(gm, mem, nb);
+        goto postaction;
+      }
+
+      else if (nb > gm->dvsize) {
+        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */
+          mchunkptr b, p, r;
+          size_t rsize;
+          bindex_t i;
+          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));
+          binmap_t leastbit = least_bit(leftbits);
+          compute_bit2idx(leastbit, i);
+          b = smallbin_at(gm, i);
+          p = b->fd;
+          assert(chunksize(p) == small_index2size(i));
+          unlink_first_small_chunk(gm, b, p, i);
+          rsize = small_index2size(i) - nb;
+          /* Fit here cannot be remainderless if 4byte sizes */
+          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)
+            set_inuse_and_pinuse(gm, p, small_index2size(i));
+          else {
+            set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
+            r = chunk_plus_offset(p, nb);
+            set_size_and_pinuse_of_free_chunk(r, rsize);
+            replace_dv(gm, r, rsize);
+          }
+          mem = chunk2mem(p);
+          check_malloced_chunk(gm, mem, nb);
+          goto postaction;
+        }
+
+        else if (gm->treemap != 0 && (mem = tmalloc_small(gm, nb)) != 0) {
+          check_malloced_chunk(gm, mem, nb);
+          goto postaction;
+        }
+      }
+    }
+    else if (bytes >= MAX_REQUEST)
+      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */
+    else {
+      nb = pad_request(bytes);
+      if (gm->treemap != 0 && (mem = tmalloc_large(gm, nb)) != 0) {
+        check_malloced_chunk(gm, mem, nb);
+        goto postaction;
+      }
+    }
+
+    if (nb <= gm->dvsize) {
+      size_t rsize = gm->dvsize - nb;
+      mchunkptr p = gm->dv;
+      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */
+        mchunkptr r = gm->dv = chunk_plus_offset(p, nb);
+        gm->dvsize = rsize;
+        set_size_and_pinuse_of_free_chunk(r, rsize);
+        set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
+      }
+      else { /* exhaust dv */
+        size_t dvs = gm->dvsize;
+        gm->dvsize = 0;
+        gm->dv = 0;
+        set_inuse_and_pinuse(gm, p, dvs);
+      }
+      mem = chunk2mem(p);
+      check_malloced_chunk(gm, mem, nb);
+      goto postaction;
+    }
+
+    else if (nb < gm->topsize) { /* Split top */
+      size_t rsize = gm->topsize -= nb;
+      mchunkptr p = gm->top;
+      mchunkptr r = gm->top = chunk_plus_offset(p, nb);
+      r->head = rsize | PINUSE_BIT;
+      set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
+      mem = chunk2mem(p);
+      check_top_chunk(gm, gm->top);
+      check_malloced_chunk(gm, mem, nb);
+      goto postaction;
+    }
+
+    mem = sys_alloc(gm, nb);
+
+  postaction:
+    POSTACTION(gm);
+    return mem;
+  }
+
+  return 0;
+}
+
+void dlfree(void* mem) {
+  /*
+     Consolidate freed chunks with preceeding or succeeding bordering
+     free chunks, if they exist, and then place in a bin.  Intermixed
+     with special cases for top, dv, mmapped chunks, and usage errors.
+  */
+
+  if (mem != 0) {
+    mchunkptr p  = mem2chunk(mem);
+#if FOOTERS
+    mstate fm = get_mstate_for(p);
+    if (!ok_magic(fm)) {
+      USAGE_ERROR_ACTION(fm, p);
+      return;
+    }
+#else /* FOOTERS */
+#define fm gm
+#endif /* FOOTERS */
+    if (!PREACTION(fm)) {
+      check_inuse_chunk(fm, p);
+      if (RTCHECK(ok_address(fm, p) && ok_inuse(p))) {
+        size_t psize = chunksize(p);
+        mchunkptr next = chunk_plus_offset(p, psize);
+        if (!pinuse(p)) {
+          size_t prevsize = p->prev_foot;
+          if (is_mmapped(p)) {
+            psize += prevsize + MMAP_FOOT_PAD;
+            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)
+              fm->footprint -= psize;
+            goto postaction;
+          }
+          else {
+            mchunkptr prev = chunk_minus_offset(p, prevsize);
+            psize += prevsize;
+            p = prev;
+            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */
+              if (p != fm->dv) {
+                unlink_chunk(fm, p, prevsize);
+              }
+              else if ((next->head & INUSE_BITS) == INUSE_BITS) {
+                fm->dvsize = psize;
+                set_free_with_pinuse(p, psize, next);
+                goto postaction;
+              }
+            }
+            else
+              goto erroraction;
+          }
+        }
+
+        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {
+          if (!cinuse(next)) {  /* consolidate forward */
+            if (next == fm->top) {
+              size_t tsize = fm->topsize += psize;
+              fm->top = p;
+              p->head = tsize | PINUSE_BIT;
+              if (p == fm->dv) {
+                fm->dv = 0;
+                fm->dvsize = 0;
+              }
+              if (should_trim(fm, tsize))
+                sys_trim(fm, 0);
+              goto postaction;
+            }
+            else if (next == fm->dv) {
+              size_t dsize = fm->dvsize += psize;
+              fm->dv = p;
+              set_size_and_pinuse_of_free_chunk(p, dsize);
+              goto postaction;
+            }
+            else {
+              size_t nsize = chunksize(next);
+              psize += nsize;
+              unlink_chunk(fm, next, nsize);
+              set_size_and_pinuse_of_free_chunk(p, psize);
+              if (p == fm->dv) {
+                fm->dvsize = psize;
+                goto postaction;
+              }
+            }
+          }
+          else
+            set_free_with_pinuse(p, psize, next);
+
+          if (is_small(psize)) {
+            insert_small_chunk(fm, p, psize);
+            check_free_chunk(fm, p);
+          }
+          else {
+            tchunkptr tp = (tchunkptr)p;
+            insert_large_chunk(fm, tp, psize);
+            check_free_chunk(fm, p);
+            if (--fm->release_checks == 0)
+              release_unused_segments(fm);
+          }
+          goto postaction;
+        }
+      }
+    erroraction:
+      USAGE_ERROR_ACTION(fm, p);
+    postaction:
+      POSTACTION(fm);
+    }
+  }
+#if !FOOTERS
+#undef fm
+#endif /* FOOTERS */
+}
+
+void* dlcalloc(size_t n_elements, size_t elem_size) {
+  void* mem;
+  size_t req = 0;
+  if (n_elements != 0) {
+    req = n_elements * elem_size;
+    if (((n_elements | elem_size) & ~(size_t)0xffff) &&
+        (req / n_elements != elem_size))
+      req = MAX_SIZE_T; /* force downstream failure on overflow */
+  }
+  mem = dlmalloc(req);
+  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))
+    memset(mem, 0, req);
+  return mem;
+}
+
+void* dlrealloc(void* oldmem, size_t bytes) {
+  if (oldmem == 0)
+    return dlmalloc(bytes);
+#ifdef REALLOC_ZERO_BYTES_FREES
+  if (bytes == 0) {
+    dlfree(oldmem);
+    return 0;
+  }
+#endif /* REALLOC_ZERO_BYTES_FREES */
+  else {
+#if ! FOOTERS
+    mstate m = gm;
+#else /* FOOTERS */
+    mstate m = get_mstate_for(mem2chunk(oldmem));
+    if (!ok_magic(m)) {
+      USAGE_ERROR_ACTION(m, oldmem);
+      return 0;
+    }
+#endif /* FOOTERS */
+    return internal_realloc(m, oldmem, bytes);
+  }
+}
+
+void* dlmemalign(size_t alignment, size_t bytes) {
+  return internal_memalign(gm, alignment, bytes);
+}
+
+void** dlindependent_calloc(size_t n_elements, size_t elem_size,
+                                 void* chunks[]) {
+  size_t sz = elem_size; /* serves as 1-element array */
+  return ialloc(gm, n_elements, &sz, 3, chunks);
+}
+
+void** dlindependent_comalloc(size_t n_elements, size_t sizes[],
+                                   void* chunks[]) {
+  return ialloc(gm, n_elements, sizes, 0, chunks);
+}
+
+void* dlvalloc(size_t bytes) {
+  size_t pagesz;
+  ensure_initialization();
+  pagesz = mparams.page_size;
+  return dlmemalign(pagesz, bytes);
+}
+
+void* dlpvalloc(size_t bytes) {
+  size_t pagesz;
+  ensure_initialization();
+  pagesz = mparams.page_size;
+  return dlmemalign(pagesz, (bytes + pagesz - SIZE_T_ONE) & ~(pagesz - SIZE_T_ONE));
+}
+
+int dlmalloc_trim(size_t pad) {
+  int result = 0;
+  ensure_initialization();
+  if (!PREACTION(gm)) {
+    result = sys_trim(gm, pad);
+    POSTACTION(gm);
+  }
+  return result;
+}
+
+size_t dlmalloc_footprint(void) {
+  return gm->footprint;
+}
+
+size_t dlmalloc_max_footprint(void) {
+  return gm->max_footprint;
+}
+
+#if !NO_MALLINFO
+struct mallinfo dlmallinfo(void) {
+  return internal_mallinfo(gm);
+}
+#endif /* NO_MALLINFO */
+
+void dlmalloc_stats() {
+  internal_malloc_stats(gm);
+}
+
+int dlmallopt(int param_number, int value) {
+  return change_mparam(param_number, value);
+}
+
+#endif /* !ONLY_MSPACES */
+
+size_t dlmalloc_usable_size(void* mem) {
+  if (mem != 0) {
+    mchunkptr p = mem2chunk(mem);
+    if (is_inuse(p))
+      return chunksize(p) - overhead_for(p);
+  }
+  return 0;
+}
+
+/* ----------------------------- user mspaces ---------------------------- */
+
+#if MSPACES
+
+static mstate init_user_mstate(char* tbase, size_t tsize) {
+  size_t msize = pad_request(sizeof(struct malloc_state));
+  mchunkptr mn;
+  mchunkptr msp = align_as_chunk(tbase);
+  mstate m = (mstate)(chunk2mem(msp));
+  memset(m, 0, msize);
+  INITIAL_LOCK(&m->mutex);
+  msp->head = (msize|INUSE_BITS);
+  m->seg.base = m->least_addr = tbase;
+  m->seg.size = m->footprint = m->max_footprint = tsize;
+  m->magic = mparams.magic;
+  m->release_checks = MAX_RELEASE_CHECK_RATE;
+  m->mflags = mparams.default_mflags;
+  m->extp = 0;
+  m->exts = 0;
+  disable_contiguous(m);
+  init_bins(m);
+  mn = next_chunk(mem2chunk(m));
+  init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) - TOP_FOOT_SIZE);
+  check_top_chunk(m, m->top);
+  return m;
+}
+
+mspace create_mspace(size_t capacity, int locked) {
+  mstate m = 0;
+  size_t msize;
+  ensure_initialization();
+  msize = pad_request(sizeof(struct malloc_state));
+  if (capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {
+    size_t rs = ((capacity == 0)? mparams.granularity :
+                 (capacity + TOP_FOOT_SIZE + msize));
+    size_t tsize = granularity_align(rs);
+    char* tbase = (char*)(CALL_MMAP(tsize));
+    if (tbase != CMFAIL) {
+      m = init_user_mstate(tbase, tsize);
+      m->seg.sflags = USE_MMAP_BIT;
+      set_lock(m, locked);
+    }
+  }
+  return (mspace)m;
+}
+
+mspace create_mspace_with_base(void* base, size_t capacity, int locked) {
+  mstate m = 0;
+  size_t msize;
+  ensure_initialization();
+  msize = pad_request(sizeof(struct malloc_state));
+  if (capacity > msize + TOP_FOOT_SIZE &&
+      capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {
+    m = init_user_mstate((char*)base, capacity);
+    m->seg.sflags = EXTERN_BIT;
+    set_lock(m, locked);
+  }
+  return (mspace)m;
+}
+
+int mspace_track_large_chunks(mspace msp, int enable) {
+  int ret = 0;
+  mstate ms = (mstate)msp;
+  if (!PREACTION(ms)) {
+    if (!use_mmap(ms))
+      ret = 1;
+    if (!enable)
+      enable_mmap(ms);
+    else
+      disable_mmap(ms);
+    POSTACTION(ms);
+  }
+  return ret;
+}
+
+size_t destroy_mspace(mspace msp) {
+  size_t freed = 0;
+  mstate ms = (mstate)msp;
+  if (ok_magic(ms)) {
+    msegmentptr sp = &ms->seg;
+    while (sp != 0) {
+      char* base = sp->base;
+      size_t size = sp->size;
+      flag_t flag = sp->sflags;
+      sp = sp->next;
+      if ((flag & USE_MMAP_BIT) && !(flag & EXTERN_BIT) &&
+          CALL_MUNMAP(base, size) == 0)
+        freed += size;
+    }
+  }
+  else {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+  return freed;
+}
+
+/*
+  mspace versions of routines are near-clones of the global
+  versions. This is not so nice but better than the alternatives.
+*/
+
+
+void* mspace_malloc(mspace msp, size_t bytes) {
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+    return 0;
+  }
+  if (!PREACTION(ms)) {
+    void* mem;
+    size_t nb;
+    if (bytes <= MAX_SMALL_REQUEST) {
+      bindex_t idx;
+      binmap_t smallbits;
+      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);
+      idx = small_index(nb);
+      smallbits = ms->smallmap >> idx;
+
+      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */
+        mchunkptr b, p;
+        idx += ~smallbits & 1;       /* Uses next bin if idx empty */
+        b = smallbin_at(ms, idx);
+        p = b->fd;
+        assert(chunksize(p) == small_index2size(idx));
+        unlink_first_small_chunk(ms, b, p, idx);
+        set_inuse_and_pinuse(ms, p, small_index2size(idx));
+        mem = chunk2mem(p);
+        check_malloced_chunk(ms, mem, nb);
+        goto postaction;
+      }
+
+      else if (nb > ms->dvsize) {
+        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */
+          mchunkptr b, p, r;
+          size_t rsize;
+          bindex_t i;
+          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));
+          binmap_t leastbit = least_bit(leftbits);
+          compute_bit2idx(leastbit, i);
+          b = smallbin_at(ms, i);
+          p = b->fd;
+          assert(chunksize(p) == small_index2size(i));
+          unlink_first_small_chunk(ms, b, p, i);
+          rsize = small_index2size(i) - nb;
+          /* Fit here cannot be remainderless if 4byte sizes */
+          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)
+            set_inuse_and_pinuse(ms, p, small_index2size(i));
+          else {
+            set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
+            r = chunk_plus_offset(p, nb);
+            set_size_and_pinuse_of_free_chunk(r, rsize);
+            replace_dv(ms, r, rsize);
+          }
+          mem = chunk2mem(p);
+          check_malloced_chunk(ms, mem, nb);
+          goto postaction;
+        }
+
+        else if (ms->treemap != 0 && (mem = tmalloc_small(ms, nb)) != 0) {
+          check_malloced_chunk(ms, mem, nb);
+          goto postaction;
+        }
+      }
+    }
+    else if (bytes >= MAX_REQUEST)
+      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */
+    else {
+      nb = pad_request(bytes);
+      if (ms->treemap != 0 && (mem = tmalloc_large(ms, nb)) != 0) {
+        check_malloced_chunk(ms, mem, nb);
+        goto postaction;
+      }
+    }
+
+    if (nb <= ms->dvsize) {
+      size_t rsize = ms->dvsize - nb;
+      mchunkptr p = ms->dv;
+      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */
+        mchunkptr r = ms->dv = chunk_plus_offset(p, nb);
+        ms->dvsize = rsize;
+        set_size_and_pinuse_of_free_chunk(r, rsize);
+        set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
+      }
+      else { /* exhaust dv */
+        size_t dvs = ms->dvsize;
+        ms->dvsize = 0;
+        ms->dv = 0;
+        set_inuse_and_pinuse(ms, p, dvs);
+      }
+      mem = chunk2mem(p);
+      check_malloced_chunk(ms, mem, nb);
+      goto postaction;
+    }
+
+    else if (nb < ms->topsize) { /* Split top */
+      size_t rsize = ms->topsize -= nb;
+      mchunkptr p = ms->top;
+      mchunkptr r = ms->top = chunk_plus_offset(p, nb);
+      r->head = rsize | PINUSE_BIT;
+      set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
+      mem = chunk2mem(p);
+      check_top_chunk(ms, ms->top);
+      check_malloced_chunk(ms, mem, nb);
+      goto postaction;
+    }
+
+    mem = sys_alloc(ms, nb);
+
+  postaction:
+    POSTACTION(ms);
+    return mem;
+  }
+
+  return 0;
+}
+
+void mspace_free(mspace msp, void* mem) {
+  if (mem != 0) {
+    mchunkptr p  = mem2chunk(mem);
+#if FOOTERS
+    mstate fm = get_mstate_for(p);
+    msp = msp; /* placate people compiling -Wunused */
+#else /* FOOTERS */
+    mstate fm = (mstate)msp;
+#endif /* FOOTERS */
+    if (!ok_magic(fm)) {
+      USAGE_ERROR_ACTION(fm, p);
+      return;
+    }
+    if (!PREACTION(fm)) {
+      check_inuse_chunk(fm, p);
+      if (RTCHECK(ok_address(fm, p) && ok_inuse(p))) {
+        size_t psize = chunksize(p);
+        mchunkptr next = chunk_plus_offset(p, psize);
+        if (!pinuse(p)) {
+          size_t prevsize = p->prev_foot;
+          if (is_mmapped(p)) {
+            psize += prevsize + MMAP_FOOT_PAD;
+            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)
+              fm->footprint -= psize;
+            goto postaction;
+          }
+          else {
+            mchunkptr prev = chunk_minus_offset(p, prevsize);
+            psize += prevsize;
+            p = prev;
+            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */
+              if (p != fm->dv) {
+                unlink_chunk(fm, p, prevsize);
+              }
+              else if ((next->head & INUSE_BITS) == INUSE_BITS) {
+                fm->dvsize = psize;
+                set_free_with_pinuse(p, psize, next);
+                goto postaction;
+              }
+            }
+            else
+              goto erroraction;
+          }
+        }
+
+        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {
+          if (!cinuse(next)) {  /* consolidate forward */
+            if (next == fm->top) {
+              size_t tsize = fm->topsize += psize;
+              fm->top = p;
+              p->head = tsize | PINUSE_BIT;
+              if (p == fm->dv) {
+                fm->dv = 0;
+                fm->dvsize = 0;
+              }
+              if (should_trim(fm, tsize))
+                sys_trim(fm, 0);
+              goto postaction;
+            }
+            else if (next == fm->dv) {
+              size_t dsize = fm->dvsize += psize;
+              fm->dv = p;
+              set_size_and_pinuse_of_free_chunk(p, dsize);
+              goto postaction;
+            }
+            else {
+              size_t nsize = chunksize(next);
+              psize += nsize;
+              unlink_chunk(fm, next, nsize);
+              set_size_and_pinuse_of_free_chunk(p, psize);
+              if (p == fm->dv) {
+                fm->dvsize = psize;
+                goto postaction;
+              }
+            }
+          }
+          else
+            set_free_with_pinuse(p, psize, next);
+
+          if (is_small(psize)) {
+            insert_small_chunk(fm, p, psize);
+            check_free_chunk(fm, p);
+          }
+          else {
+            tchunkptr tp = (tchunkptr)p;
+            insert_large_chunk(fm, tp, psize);
+            check_free_chunk(fm, p);
+            if (--fm->release_checks == 0)
+              release_unused_segments(fm);
+          }
+          goto postaction;
+        }
+      }
+    erroraction:
+      USAGE_ERROR_ACTION(fm, p);
+    postaction:
+      POSTACTION(fm);
+    }
+  }
+}
+
+void* mspace_calloc(mspace msp, size_t n_elements, size_t elem_size) {
+  void* mem;
+  size_t req = 0;
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+    return 0;
+  }
+  if (n_elements != 0) {
+    req = n_elements * elem_size;
+    if (((n_elements | elem_size) & ~(size_t)0xffff) &&
+        (req / n_elements != elem_size))
+      req = MAX_SIZE_T; /* force downstream failure on overflow */
+  }
+  mem = internal_malloc(ms, req);
+  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))
+    memset(mem, 0, req);
+  return mem;
+}
+
+void* mspace_realloc(mspace msp, void* oldmem, size_t bytes) {
+  if (oldmem == 0)
+    return mspace_malloc(msp, bytes);
+#ifdef REALLOC_ZERO_BYTES_FREES
+  if (bytes == 0) {
+    mspace_free(msp, oldmem);
+    return 0;
+  }
+#endif /* REALLOC_ZERO_BYTES_FREES */
+  else {
+#if FOOTERS
+    mchunkptr p  = mem2chunk(oldmem);
+    mstate ms = get_mstate_for(p);
+#else /* FOOTERS */
+    mstate ms = (mstate)msp;
+#endif /* FOOTERS */
+    if (!ok_magic(ms)) {
+      USAGE_ERROR_ACTION(ms,ms);
+      return 0;
+    }
+    return internal_realloc(ms, oldmem, bytes);
+  }
+}
+
+void* mspace_memalign(mspace msp, size_t alignment, size_t bytes) {
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+    return 0;
+  }
+  return internal_memalign(ms, alignment, bytes);
+}
+
+void** mspace_independent_calloc(mspace msp, size_t n_elements,
+                                 size_t elem_size, void* chunks[]) {
+  size_t sz = elem_size; /* serves as 1-element array */
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+    return 0;
+  }
+  return ialloc(ms, n_elements, &sz, 3, chunks);
+}
+
+void** mspace_independent_comalloc(mspace msp, size_t n_elements,
+                                   size_t sizes[], void* chunks[]) {
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+    return 0;
+  }
+  return ialloc(ms, n_elements, sizes, 0, chunks);
+}
+
+int mspace_trim(mspace msp, size_t pad) {
+  int result = 0;
+  mstate ms = (mstate)msp;
+  if (ok_magic(ms)) {
+    if (!PREACTION(ms)) {
+      result = sys_trim(ms, pad);
+      POSTACTION(ms);
+    }
+  }
+  else {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+  return result;
+}
+
+void mspace_malloc_stats(mspace msp) {
+  mstate ms = (mstate)msp;
+  if (ok_magic(ms)) {
+    internal_malloc_stats(ms);
+  }
+  else {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+}
+
+size_t mspace_footprint(mspace msp) {
+  size_t result = 0;
+  mstate ms = (mstate)msp;
+  if (ok_magic(ms)) {
+    result = ms->footprint;
+  }
+  else {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+  return result;
+}
+
+
+size_t mspace_max_footprint(mspace msp) {
+  size_t result = 0;
+  mstate ms = (mstate)msp;
+  if (ok_magic(ms)) {
+    result = ms->max_footprint;
+  }
+  else {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+  return result;
+}
+
+
+#if !NO_MALLINFO
+struct mallinfo mspace_mallinfo(mspace msp) {
+  mstate ms = (mstate)msp;
+  if (!ok_magic(ms)) {
+    USAGE_ERROR_ACTION(ms,ms);
+  }
+  return internal_mallinfo(ms);
+}
+#endif /* NO_MALLINFO */
+
+size_t mspace_usable_size(void* mem) {
+  if (mem != 0) {
+    mchunkptr p = mem2chunk(mem);
+    if (is_inuse(p))
+      return chunksize(p) - overhead_for(p);
+  }
+  return 0;
+}
+
+int mspace_mallopt(int param_number, int value) {
+  return change_mparam(param_number, value);
+}
+
+#endif /* MSPACES */
+
+
+/* -------------------- Alternative MORECORE functions ------------------- */
+
+/*
+  Guidelines for creating a custom version of MORECORE:
+
+  * For best performance, MORECORE should allocate in multiples of pagesize.
+  * MORECORE may allocate more memory than requested. (Or even less,
+      but this will usually result in a malloc failure.)
+  * MORECORE must not allocate memory when given argument zero, but
+      instead return one past the end address of memory from previous
+      nonzero call.
+  * For best performance, consecutive calls to MORECORE with positive
+      arguments should return increasing addresses, indicating that
+      space has been contiguously extended.
+  * Even though consecutive calls to MORECORE need not return contiguous
+      addresses, it must be OK for malloc'ed chunks to span multiple
+      regions in those cases where they do happen to be contiguous.
+  * MORECORE need not handle negative arguments -- it may instead
+      just return MFAIL when given negative arguments.
+      Negative arguments are always multiples of pagesize. MORECORE
+      must not misinterpret negative args as large positive unsigned
+      args. You can suppress all such calls from even occurring by defining
+      MORECORE_CANNOT_TRIM,
+
+  As an example alternative MORECORE, here is a custom allocator
+  kindly contributed for pre-OSX macOS.  It uses virtually but not
+  necessarily physically contiguous non-paged memory (locked in,
+  present and won't get swapped out).  You can use it by uncommenting
+  this section, adding some #includes, and setting up the appropriate
+  defines above:
+
+      #define MORECORE osMoreCore
+
+  There is also a shutdown routine that should somehow be called for
+  cleanup upon program exit.
+
+  #define MAX_POOL_ENTRIES 100
+  #define MINIMUM_MORECORE_SIZE  (64 * 1024U)
+  static int next_os_pool;
+  void *our_os_pools[MAX_POOL_ENTRIES];
+
+  void *osMoreCore(int size)
+  {
+    void *ptr = 0;
+    static void *sbrk_top = 0;
+
+    if (size > 0)
+    {
+      if (size < MINIMUM_MORECORE_SIZE)
+         size = MINIMUM_MORECORE_SIZE;
+      if (CurrentExecutionLevel() == kTaskLevel)
+         ptr = PoolAllocateResident(size + RM_PAGE_SIZE, 0);
+      if (ptr == 0)
+      {
+        return (void *) MFAIL;
+      }
+      // save ptrs so they can be freed during cleanup
+      our_os_pools[next_os_pool] = ptr;
+      next_os_pool++;
+      ptr = (void *) ((((size_t) ptr) + RM_PAGE_MASK) & ~RM_PAGE_MASK);
+      sbrk_top = (char *) ptr + size;
+      return ptr;
+    }
+    else if (size < 0)
+    {
+      // we don't currently support shrink behavior
+      return (void *) MFAIL;
+    }
+    else
+    {
+      return sbrk_top;
+    }
+  }
+
+  // cleanup any allocated memory pools
+  // called as last thing before shutting down driver
+
+  void osCleanupMem(void)
+  {
+    void **ptr;
+
+    for (ptr = our_os_pools; ptr < &our_os_pools[MAX_POOL_ENTRIES]; ptr++)
+      if (*ptr)
+      {
+         PoolDeallocate(*ptr);
+         *ptr = 0;
+      }
+  }
+
+*/
+
+
+/* -----------------------------------------------------------------------
+History:
+    V2.8.4 Wed May 27 09:56:23 2009  Doug Lea  (dl at gee)
+      * Use zeros instead of prev foot for is_mmapped
+      * Add mspace_track_large_chunks; thanks to Jean Brouwers
+      * Fix set_inuse in internal_realloc; thanks to Jean Brouwers
+      * Fix insufficient sys_alloc padding when using 16byte alignment
+      * Fix bad error check in mspace_footprint
+      * Adaptations for ptmalloc; thanks to Wolfram Gloger.
+      * Reentrant spin locks; thanks to Earl Chew and others
+      * Win32 improvements; thanks to Niall Douglas and Earl Chew
+      * Add NO_SEGMENT_TRAVERSAL and MAX_RELEASE_CHECK_RATE options
+      * Extension hook in malloc_state
+      * Various small adjustments to reduce warnings on some compilers
+      * Various configuration extensions/changes for more platforms. Thanks
+         to all who contributed these.
+
+    V2.8.3 Thu Sep 22 11:16:32 2005  Doug Lea  (dl at gee)
+      * Add max_footprint functions
+      * Ensure all appropriate literals are size_t
+      * Fix conditional compilation problem for some #define settings
+      * Avoid concatenating segments with the one provided
+        in create_mspace_with_base
+      * Rename some variables to avoid compiler shadowing warnings
+      * Use explicit lock initialization.
+      * Better handling of sbrk interference.
+      * Simplify and fix segment insertion, trimming and mspace_destroy
+      * Reinstate REALLOC_ZERO_BYTES_FREES option from 2.7.x
+      * Thanks especially to Dennis Flanagan for help on these.
+
+    V2.8.2 Sun Jun 12 16:01:10 2005  Doug Lea  (dl at gee)
+      * Fix memalign brace error.
+
+    V2.8.1 Wed Jun  8 16:11:46 2005  Doug Lea  (dl at gee)
+      * Fix improper #endif nesting in C++
+      * Add explicit casts needed for C++
+
+    V2.8.0 Mon May 30 14:09:02 2005  Doug Lea  (dl at gee)
+      * Use trees for large bins
+      * Support mspaces
+      * Use segments to unify sbrk-based and mmap-based system allocation,
+        removing need for emulation on most platforms without sbrk.
+      * Default safety checks
+      * Optional footer checks. Thanks to William Robertson for the idea.
+      * Internal code refactoring
+      * Incorporate suggestions and platform-specific changes.
+        Thanks to Dennis Flanagan, Colin Plumb, Niall Douglas,
+        Aaron Bachmann,  Emery Berger, and others.
+      * Speed up non-fastbin processing enough to remove fastbins.
+      * Remove useless cfree() to avoid conflicts with other apps.
+      * Remove internal memcpy, memset. Compilers handle builtins better.
+      * Remove some options that no one ever used and rename others.
+
+    V2.7.2 Sat Aug 17 09:07:30 2002  Doug Lea  (dl at gee)
+      * Fix malloc_state bitmap array misdeclaration
+
+    V2.7.1 Thu Jul 25 10:58:03 2002  Doug Lea  (dl at gee)
+      * Allow tuning of FIRST_SORTED_BIN_SIZE
+      * Use PTR_UINT as type for all ptr->int casts. Thanks to John Belmonte.
+      * Better detection and support for non-contiguousness of MORECORE.
+        Thanks to Andreas Mueller, Conal Walsh, and Wolfram Gloger
+      * Bypass most of malloc if no frees. Thanks To Emery Berger.
+      * Fix freeing of old top non-contiguous chunk im sysmalloc.
+      * Raised default trim and map thresholds to 256K.
+      * Fix mmap-related #defines. Thanks to Lubos Lunak.
+      * Fix copy macros; added LACKS_FCNTL_H. Thanks to Neal Walfield.
+      * Branch-free bin calculation
+      * Default trim and mmap thresholds now 256K.
+
+    V2.7.0 Sun Mar 11 14:14:06 2001  Doug Lea  (dl at gee)
+      * Introduce independent_comalloc and independent_calloc.
+        Thanks to Michael Pachos for motivation and help.
+      * Make optional .h file available
+      * Allow > 2GB requests on 32bit systems.
+      * new WIN32 sbrk, mmap, munmap, lock code from <Walter@GeNeSys-e.de>.
+        Thanks also to Andreas Mueller <a.mueller at paradatec.de>,
+        and Anonymous.
+      * Allow override of MALLOC_ALIGNMENT (Thanks to Ruud Waij for
+        helping test this.)
+      * memalign: check alignment arg
+      * realloc: don't try to shift chunks backwards, since this
+        leads to  more fragmentation in some programs and doesn't
+        seem to help in any others.
+      * Collect all cases in malloc requiring system memory into sysmalloc
+      * Use mmap as backup to sbrk
+      * Place all internal state in malloc_state
+      * Introduce fastbins (although similar to 2.5.1)
+      * Many minor tunings and cosmetic improvements
+      * Introduce USE_PUBLIC_MALLOC_WRAPPERS, USE_MALLOC_LOCK
+      * Introduce MALLOC_FAILURE_ACTION, MORECORE_CONTIGUOUS
+        Thanks to Tony E. Bennett <tbennett@nvidia.com> and others.
+      * Include errno.h to support default failure action.
+
+    V2.6.6 Sun Dec  5 07:42:19 1999  Doug Lea  (dl at gee)
+      * return null for negative arguments
+      * Added Several WIN32 cleanups from Martin C. Fong <mcfong at yahoo.com>
+         * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'
+          (e.g. WIN32 platforms)
+         * Cleanup header file inclusion for WIN32 platforms
+         * Cleanup code to avoid Microsoft Visual C++ compiler complaints
+         * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing
+           memory allocation routines
+         * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)
+         * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to
+           usage of 'assert' in non-WIN32 code
+         * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to
+           avoid infinite loop
+      * Always call 'fREe()' rather than 'free()'
+
+    V2.6.5 Wed Jun 17 15:57:31 1998  Doug Lea  (dl at gee)
+      * Fixed ordering problem with boundary-stamping
+
+    V2.6.3 Sun May 19 08:17:58 1996  Doug Lea  (dl at gee)
+      * Added pvalloc, as recommended by H.J. Liu
+      * Added 64bit pointer support mainly from Wolfram Gloger
+      * Added anonymously donated WIN32 sbrk emulation
+      * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
+      * malloc_extend_top: fix mask error that caused wastage after
+        foreign sbrks
+      * Add linux mremap support code from HJ Liu
+
+    V2.6.2 Tue Dec  5 06:52:55 1995  Doug Lea  (dl at gee)
+      * Integrated most documentation with the code.
+      * Add support for mmap, with help from
+        Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
+      * Use last_remainder in more cases.
+      * Pack bins using idea from  colin@nyx10.cs.du.edu
+      * Use ordered bins instead of best-fit threshhold
+      * Eliminate block-local decls to simplify tracing and debugging.
+      * Support another case of realloc via move into top
+      * Fix error occuring when initial sbrk_base not word-aligned.
+      * Rely on page size for units instead of SBRK_UNIT to
+        avoid surprises about sbrk alignment conventions.
+      * Add mallinfo, mallopt. Thanks to Raymond Nijssen
+        (raymond@es.ele.tue.nl) for the suggestion.
+      * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
+      * More precautions for cases where other routines call sbrk,
+        courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
+      * Added macros etc., allowing use in linux libc from
+        H.J. Lu (hjl@gnu.ai.mit.edu)
+      * Inverted this history list
+
+    V2.6.1 Sat Dec  2 14:10:57 1995  Doug Lea  (dl at gee)
+      * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
+      * Removed all preallocation code since under current scheme
+        the work required to undo bad preallocations exceeds
+        the work saved in good cases for most test programs.
+      * No longer use return list or unconsolidated bins since
+        no scheme using them consistently outperforms those that don't
+        given above changes.
+      * Use best fit for very large chunks to prevent some worst-cases.
+      * Added some support for debugging
+
+    V2.6.0 Sat Nov  4 07:05:23 1995  Doug Lea  (dl at gee)
+      * Removed footers when chunks are in use. Thanks to
+        Paul Wilson (wilson@cs.texas.edu) for the suggestion.
+
+    V2.5.4 Wed Nov  1 07:54:51 1995  Doug Lea  (dl at gee)
+      * Added malloc_trim, with help from Wolfram Gloger
+        (wmglo@Dent.MED.Uni-Muenchen.DE).
+
+    V2.5.3 Tue Apr 26 10:16:01 1994  Doug Lea  (dl at g)
+
+    V2.5.2 Tue Apr  5 16:20:40 1994  Doug Lea  (dl at g)
+      * realloc: try to expand in both directions
+      * malloc: swap order of clean-bin strategy;
+      * realloc: only conditionally expand backwards
+      * Try not to scavenge used bins
+      * Use bin counts as a guide to preallocation
+      * Occasionally bin return list chunks in first scan
+      * Add a few optimizations from colin@nyx10.cs.du.edu
+
+    V2.5.1 Sat Aug 14 15:40:43 1993  Doug Lea  (dl at g)
+      * faster bin computation & slightly different binning
+      * merged all consolidations to one part of malloc proper
+         (eliminating old malloc_find_space & malloc_clean_bin)
+      * Scan 2 returns chunks (not just 1)
+      * Propagate failure in realloc if malloc returns 0
+      * Add stuff to allow compilation on non-ANSI compilers
+          from kpv@research.att.com
+
+    V2.5 Sat Aug  7 07:41:59 1993  Doug Lea  (dl at g.oswego.edu)
+      * removed potential for odd address access in prev_chunk
+      * removed dependency on getpagesize.h
+      * misc cosmetics and a bit more internal documentation
+      * anticosmetics: mangled names in macros to evade debugger strangeness
+      * tested on sparc, hp-700, dec-mips, rs6000
+          with gcc & native cc (hp, dec only) allowing
+          Detlefs & Zorn comparison study (in SIGPLAN Notices.)
+
+    Trial version Fri Aug 28 13:14:29 1992  Doug Lea  (dl at g.oswego.edu)
+      * Based loosely on libg++-1.2X malloc. (It retains some of the overall
+         structure of old version,  but most details differ.)
+
+*/
+
+#endif
diff --git a/drivers/nedmalloc/nedmalloc.cpp b/drivers/nedmalloc/nedmalloc.cpp
index 8845d96549..9aac277a2a 100644
--- a/drivers/nedmalloc/nedmalloc.cpp
+++ b/drivers/nedmalloc/nedmalloc.cpp
@@ -1,1467 +1,1467 @@
-#ifdef NEDMALLOC_ENABLED

-/* Alternative malloc implementation for multiple threads without

-lock contention based on dlmalloc. (C) 2005-2009 Niall Douglas

-

-Boost Software License - Version 1.0 - August 17th, 2003

-

-Permission is hereby granted, free of charge, to any person or organization

-obtaining a copy of the software and accompanying documentation covered by

-this license (the "Software") to use, reproduce, display, distribute,

-execute, and transmit the Software, and to prepare derivative works of the

-Software, and to permit third-parties to whom the Software is furnished to

-do so, all subject to the following:

-

-The copyright notices in the Software and this entire statement, including

-the above license grant, this restriction and the following disclaimer,

-must be included in all copies of the Software, in whole or in part, and

-all derivative works of the Software, unless such copies or derivative

-works are solely in the form of machine-executable object code generated by

-a source language processor.

-

-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR

-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

-FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT

-SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE

-FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,

-ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER

-DEALINGS IN THE SOFTWARE.

-*/

-

-#ifdef _MSC_VER

-/* Enable full aliasing on MSVC */

-/*#pragma optimize("a", on)*/

-#pragma warning(push)

-#pragma warning(disable:4100)	/* unreferenced formal parameter */

-#pragma warning(disable:4127)	/* conditional expression is constant */

-#pragma warning(disable:4706)	/* assignment within conditional expression */

-#endif

-

-/*#define ENABLE_TOLERANT_NEDMALLOC 1*/

-/*#define ENABLE_FAST_HEAP_DETECTION 1*/

-/*#define NEDMALLOC_DEBUG 1*/

-

-/*#define FULLSANITYCHECKS*/

-/* If link time code generation is on, don't force or prevent inlining */

-#if defined(_MSC_VER) && defined(NEDMALLOC_DLL_EXPORTS)

-#define FORCEINLINE

-#define NOINLINE

-#endif

-

-

-#include "nedmalloc.h"

-#ifdef WIN32

- #include <malloc.h>

- #include <stddef.h>

-#endif

-#if USE_ALLOCATOR==1

- #define MSPACES 1

- #define ONLY_MSPACES 1

-#endif

-#define USE_DL_PREFIX 1

-#ifndef USE_LOCKS

- #define USE_LOCKS 1

-#endif

-#define FOOTERS 1           /* Need to enable footers so frees lock the right mspace */

-#ifndef NEDMALLOC_DEBUG

- #if defined(DEBUG) || defined(_DEBUG)

-  #define NEDMALLOC_DEBUG 1

- #else

-  #define NEDMALLOC_DEBUG 0

- #endif

-#endif

-/* We need to consistently define DEBUG=0|1, _DEBUG and NDEBUG for dlmalloc */

-#undef DEBUG

-#undef _DEBUG

-#if NEDMALLOC_DEBUG

- #define _DEBUG

- #define DEBUG 1

-#else

- #define DEBUG 0

-#endif

-#ifdef NDEBUG               /* Disable assert checking on release builds */

- #undef DEBUG

- #undef _DEBUG

-#endif

-/* The default of 64Kb means we spend too much time kernel-side */

-#ifndef DEFAULT_GRANULARITY

-#define DEFAULT_GRANULARITY (1*1024*1024)

-#if DEBUG

-#define DEFAULT_GRANULARITY_ALIGNED

-#endif

-#endif

-/*#define USE_SPIN_LOCKS 0*/

-

-

-#include "malloc.c.h"

-#ifdef NDEBUG               /* Disable assert checking on release builds */

- #undef DEBUG

-#elif !NEDMALLOC_DEBUG

- #ifdef __GNUC__

-  #warning DEBUG is defined so allocator will run with assert checking! Define NDEBUG to run at full speed.

- #elif defined(_MSC_VER)

-  #pragma message(__FILE__ ": WARNING: DEBUG is defined so allocator will run with assert checking! Define NDEBUG to run at full speed.")

- #endif

-#endif

-

-/* The maximum concurrent threads in a pool possible */

-#ifndef MAXTHREADSINPOOL

-#define MAXTHREADSINPOOL 16

-#endif

-/* The maximum number of threadcaches which can be allocated */

-#ifndef THREADCACHEMAXCACHES

-#define THREADCACHEMAXCACHES 256

-#endif

-/* The maximum size to be allocated from the thread cache */

-#ifndef THREADCACHEMAX

-#define THREADCACHEMAX 8192

-#endif

-#if 0

-/* The number of cache entries for finer grained bins. This is (topbitpos(THREADCACHEMAX)-4)*2 */

-#define THREADCACHEMAXBINS ((13-4)*2)

-#else

-/* The number of cache entries. This is (topbitpos(THREADCACHEMAX)-4) */

-#define THREADCACHEMAXBINS (13-4)

-#endif

-/* Point at which the free space in a thread cache is garbage collected */

-#ifndef THREADCACHEMAXFREESPACE

-#define THREADCACHEMAXFREESPACE (512*1024)

-#endif

-

-

-#ifdef WIN32

- #define TLSVAR			DWORD

- #define TLSALLOC(k)	(*(k)=TlsAlloc(), TLS_OUT_OF_INDEXES==*(k))

- #define TLSFREE(k)		(!TlsFree(k))

- #define TLSGET(k)		TlsGetValue(k)

- #define TLSSET(k, a)	(!TlsSetValue(k, a))

- #ifdef DEBUG

-static LPVOID ChkedTlsGetValue(DWORD idx)

-{

-	LPVOID ret=TlsGetValue(idx);

-	assert(S_OK==GetLastError());

-	return ret;

-}

-  #undef TLSGET

-  #define TLSGET(k) ChkedTlsGetValue(k)

- #endif

-#else

- #define TLSVAR			pthread_key_t

- #define TLSALLOC(k)	pthread_key_create(k, 0)

- #define TLSFREE(k)		pthread_key_delete(k)

- #define TLSGET(k)		pthread_getspecific(k)

- #define TLSSET(k, a)	pthread_setspecific(k, a)

-#endif

-

-#if defined(__cplusplus)

-#if !defined(NO_NED_NAMESPACE)

-namespace nedalloc {

-#else

-extern "C" {

-#endif

-#endif

-

-#if USE_ALLOCATOR==0

-static void *unsupported_operation(const char *opname) THROWSPEC

-{

-	fprintf(stderr, "nedmalloc: The operation %s is not supported under this build configuration\n", opname);

-	abort();

-	return 0;

-}

-static size_t mspacecounter=(size_t) 0xdeadbeef;

-#endif

-#ifndef ENABLE_FAST_HEAP_DETECTION

-static void *RESTRICT leastusedaddress;

-static size_t largestusedblock;

-#endif

-

-static FORCEINLINE void *CallMalloc(void *RESTRICT mspace, size_t size, size_t alignment) THROWSPEC

-{

-	void *RESTRICT ret=0;

-	size_t _alignment=alignment;

-#if USE_MAGIC_HEADERS

-	size_t *_ret=0;

-	size+=alignment+3*sizeof(size_t);

-	_alignment=0;

-#endif

-#if USE_ALLOCATOR==0

-	ret=_alignment ? 

-#ifdef _MSC_VER

-	/* This is the MSVCRT equivalent */

-		_aligned_malloc(size, _alignment)

-#elif defined(__linux__) || defined(__FreeBSD__) || defined(__APPLE__)

-	/* This is the glibc/ptmalloc2/dlmalloc/BSD libc equivalent.  */

-		memalign(_alignment, size)

-#else

-#error Cannot aligned allocate with the memory allocator of an unknown system!

-#endif

-		: malloc(size);

-#elif USE_ALLOCATOR==1

-	ret=_alignment ? mspace_memalign((mstate) mspace, _alignment, size) : mspace_malloc((mstate) mspace, size);

-#ifndef ENABLE_FAST_HEAP_DETECTION

-	if(ret)

-	{

-		size_t truesize=chunksize(mem2chunk(ret));

-		if(!leastusedaddress || (void *)((mstate) mspace)->least_addr<leastusedaddress) leastusedaddress=(void *)((mstate) mspace)->least_addr;

-		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);

-	}

-#endif

-#endif

-	if(!ret) return 0;

-#if USE_MAGIC_HEADERS

-	_ret=(size_t *) ret;

-	ret=(void *)(_ret+3);

-	if(alignment) ret=(void *)(((size_t) ret+alignment-1)&~(alignment-1));

-	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *)"NEDMALOC";

-	_ret[0]=(size_t) mspace;

-	_ret[1]=size-3*sizeof(size_t);

-#endif

-	return ret;

-}

-

-static FORCEINLINE void *CallCalloc(void *RESTRICT mspace, size_t size, size_t alignment) THROWSPEC

-{

-	void *RESTRICT ret=0;

-#if USE_MAGIC_HEADERS

-	size_t *_ret=0;

-	size+=alignment+3*sizeof(size_t);

-#endif

-#if USE_ALLOCATOR==0

-	ret=calloc(1, size);

-#elif USE_ALLOCATOR==1

-	ret=mspace_calloc((mstate) mspace, 1, size);

-#ifndef ENABLE_FAST_HEAP_DETECTION

-	if(ret)

-	{

-		size_t truesize=chunksize(mem2chunk(ret));

-		if(!leastusedaddress || (void *)((mstate) mspace)->least_addr<leastusedaddress) leastusedaddress=(void *)((mstate) mspace)->least_addr;

-		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);

-	}

-#endif

-#endif

-	if(!ret) return 0;

-#if USE_MAGIC_HEADERS

-	_ret=(size_t *) ret;

-	ret=(void *)(_ret+3);

-	if(alignment) ret=(void *)(((size_t) ret+alignment-1)&~(alignment-1));

-	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *) "NEDMALOC";

-	_ret[0]=(size_t) mspace;

-	_ret[1]=size-3*sizeof(size_t);

-#endif

-	return ret;

-}

-

-static FORCEINLINE void *CallRealloc(void *RESTRICT mspace, void *RESTRICT mem, int isforeign, size_t oldsize, size_t newsize) THROWSPEC

-{

-	void *RESTRICT ret=0;

-#if USE_MAGIC_HEADERS

-	mstate oldmspace=0;

-	size_t *_ret=0, *_mem=(size_t *) mem-3;

-#endif

-	if(isforeign)

-	{	/* Transfer */

-#if USE_MAGIC_HEADERS

-		assert(_mem[0]!=*(size_t *) "NEDMALOC");

-#endif

-		if((ret=CallMalloc(mspace, newsize, 0)))

-		{

-#if defined(DEBUG)

-			printf("*** nedmalloc frees system allocated block %p\n", mem);

-#endif

-			memcpy(ret, mem, oldsize<newsize ? oldsize : newsize);

-			free(mem);

-		}

-		return ret;

-	}

-#if USE_MAGIC_HEADERS

-	assert(_mem[0]==*(size_t *) "NEDMALOC");

-	newsize+=3*sizeof(size_t);

-	oldmspace=(mstate) _mem[1];

-	assert(oldsize>=_mem[2]);

-	for(; *_mem==*(size_t *) "NEDMALOC"; *_mem--=*(size_t *) "nedmaloc");

-	mem=(void *)(++_mem);

-#endif

-#if USE_ALLOCATOR==0

-	ret=realloc(mem, newsize);

-#elif USE_ALLOCATOR==1

-	ret=mspace_realloc((mstate) mspace, mem, newsize);

-#ifndef ENABLE_FAST_HEAP_DETECTION

-	if(ret)

-	{

-		size_t truesize=chunksize(mem2chunk(ret));

-		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);

-	}

-#endif

-#endif

-	if(!ret)

-	{	/* Put it back the way it was */

-#if USE_MAGIC_HEADERS

-		for(; *_mem==0; *_mem++=*(size_t *) "NEDMALOC");

-#endif

-		return 0;

-	}

-#if USE_MAGIC_HEADERS

-	_ret=(size_t *) ret;

-	ret=(void *)(_ret+3);

-	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *) "NEDMALOC";

-	_ret[0]=(size_t) mspace;

-	_ret[1]=newsize-3*sizeof(size_t);

-#endif

-	return ret;

-}

-

-static FORCEINLINE void CallFree(void *RESTRICT mspace, void *RESTRICT mem, int isforeign) THROWSPEC

-{

-#if USE_MAGIC_HEADERS

-	mstate oldmspace=0;

-	size_t *_mem=(size_t *) mem-3, oldsize=0;

-#endif

-	if(isforeign)

-	{

-#if USE_MAGIC_HEADERS

-		assert(_mem[0]!=*(size_t *) "NEDMALOC");

-#endif

-#if defined(DEBUG)

-		printf("*** nedmalloc frees system allocated block %p\n", mem);

-#endif

-		free(mem);

-		return;

-	}

-#if USE_MAGIC_HEADERS

-	assert(_mem[0]==*(size_t *) "NEDMALOC");

-	oldmspace=(mstate) _mem[1];

-	oldsize=_mem[2];

-	for(; *_mem==*(size_t *) "NEDMALOC"; *_mem--=*(size_t *) "nedmaloc");

-	mem=(void *)(++_mem);

-#endif

-#if USE_ALLOCATOR==0

-	free(mem);

-#elif USE_ALLOCATOR==1

-	mspace_free((mstate) mspace, mem);

-#endif

-}

-

-static NEDMALLOCNOALIASATTR mstate nedblkmstate(void *RESTRICT mem) THROWSPEC

-{

-	if(mem)

-	{

-#if USE_MAGIC_HEADERS

-		size_t *_mem=(size_t *) mem-3;

-		if(_mem[0]==*(size_t *) "NEDMALOC")

-		{

-			return (mstate) _mem[1];

-		}

-		else return 0;

-#else

-#if USE_ALLOCATOR==0

-		/* Fail everything */

-		return 0;

-#elif USE_ALLOCATOR==1

-#ifdef ENABLE_FAST_HEAP_DETECTION

-#ifdef WIN32

-		/*  On Windows for RELEASE both x86 and x64 the NT heap precedes each block with an eight byte header

-			which looks like:

-				normal: 4 bytes of size, 4 bytes of [char < 64, char < 64, char < 64 bit 0 always set, char random ]

-				mmaped: 4 bytes of size  4 bytes of [zero,      zero,      0xb,                        zero        ]

-

-			On Windows for DEBUG both x86 and x64 the preceding four bytes is always 0xfdfdfdfd (no man's land).

-		*/

-#pragma pack(push, 1)

-		struct _HEAP_ENTRY

-		{

-			USHORT Size;

-			USHORT PreviousSize;

-			UCHAR Cookie;			/* SegmentIndex */

-			UCHAR Flags;			/* always bit 0 (HEAP_ENTRY_BUSY). bit 1=(HEAP_ENTRY_EXTRA_PRESENT), bit 2=normal block (HEAP_ENTRY_FILL_PATTERN), bit 3=mmap block (HEAP_ENTRY_VIRTUAL_ALLOC). Bit 4 (HEAP_ENTRY_LAST_ENTRY) could be set */

-			UCHAR UnusedBytes;

-			UCHAR SmallTagIndex;	/* fastbin index. Always one of 0x02, 0x03, 0x04 < 0x80 */

-		} *RESTRICT he=((struct _HEAP_ENTRY *) mem)-1;

-#pragma pack(pop)

-		unsigned int header=((unsigned int *)mem)[-1], mask1=0x8080E100, result1, mask2=0xFFFFFF06, result2;

-		result1=header & mask1;	/* Positive testing for NT heap */

-		result2=header & mask2;	/* Positive testing for dlmalloc */

-		if(result1==0x00000100 && result2!=0x00000102)

-		{	/* This is likely a NT heap block */

-			return 0;

-		}

-#endif

-#ifdef __linux__

-		/* On Linux glibc uses ptmalloc2 (really dlmalloc) just as we do, but prev_foot contains rubbish

-		when the preceding block is allocated because ptmalloc2 finds the local mstate by rounding the ptr

-		down to the nearest megabyte. It's like dlmalloc with FOOTERS disabled. */

-		mchunkptr p=mem2chunk(mem);

-		mstate fm=get_mstate_for(p);

-		/* If it's a ptmalloc2 block, fm is likely to be some crazy value */

-		if(!is_aligned(fm)) return 0;

-		if((size_t)mem-(size_t)fm>=(size_t)1<<(SIZE_T_BITSIZE-1)) return 0;

-		if(ok_magic(fm))

-			return fm;

-		else

-			return 0;

-		if(1) { }

-#endif

-		else

-		{

-			mchunkptr p=mem2chunk(mem);

-			mstate fm=get_mstate_for(p);

-			assert(ok_magic(fm));	/* If this fails, someone tried to free a block twice */

-			if(ok_magic(fm))

-				return fm;

-		}

-#else

-//#ifdef WIN32

-//		__try

-//#endif

-		{

-			/* We try to return zero here if it isn't one of our own blocks, however

-			the current block annotation scheme used by dlmalloc makes it impossible

-			to be absolutely sure of avoiding a segfault.

-

-			mchunkptr->prev_foot = mem-(2*size_t) = mstate ^ mparams.magic for PRECEDING block;

-			mchunkptr->head      = mem-(1*size_t) = 8 multiple size of this block with bottom three bits = FLAG_BITS

-			    FLAG_BITS = bit 0 is CINUSE (currently in use unless is mmap), bit 1 is PINUSE (previous block currently

-				            in use unless mmap), bit 2 is UNUSED and currently is always zero.

-			*/

-			register void *RESTRICT leastusedaddress_=leastusedaddress;		/* Cache these to avoid register reloading */

-			register size_t largestusedblock_=largestusedblock;

-			if(!is_aligned(mem)) return 0;		/* Would fail very rarely as all allocators return aligned blocks */

-			if(mem<leastusedaddress_) return 0;	/* Simple but effective */

-			{

-				mchunkptr p=mem2chunk(mem);

-				mstate fm=0;

-				int ismmapped=is_mmapped(p);

-				if((!ismmapped && !is_inuse(p)) || (p->head & FLAG4_BIT)) return 0;

-				/* Reduced uncertainty by 0.5^2 = 25.0% */

-				/* size should never exceed largestusedblock */

-				if(chunksize(p)>largestusedblock_) return 0;

-				/* Reduced uncertainty by a minimum of 0.5^3 = 12.5%, maximum 0.5^16 = 0.0015% */

-				/* Having sanity checked prev_foot and head, check next block */

-				if(!ismmapped && (!next_pinuse(p) || (next_chunk(p)->head & FLAG4_BIT))) return 0;

-				/* Reduced uncertainty by 0.5^5 = 3.13% or 0.5^18 = 0.00038% */

-	#if 0

-				/* If previous block is free, check that its next block pointer equals us */

-				if(!ismmapped && !pinuse(p))

-					if(next_chunk(prev_chunk(p))!=p) return 0;

-				/* We could start comparing prev_foot's for similarity but it starts getting slow. */

-	#endif

-				fm = get_mstate_for(p);

-				if(!is_aligned(fm) || (void *)fm<leastusedaddress_) return 0;

-				if((size_t)mem-(size_t)fm>=(size_t)1<<(SIZE_T_BITSIZE-1)) return 0;

-				assert(ok_magic(fm));	/* If this fails, someone tried to free a block twice */

-				if(ok_magic(fm))

-					return fm;

-			}

-		}

-//#ifdef WIN32

-//		__except(1) { }

-//#endif

-#endif

-#endif

-#endif

-	}

-	return 0;

-}

-NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC

-{

-	if(mem)

-	{

-		if(isforeign) *isforeign=1;

-#if USE_MAGIC_HEADERS

-		{

-			size_t *_mem=(size_t *) mem-3;

-			if(_mem[0]==*(size_t *) "NEDMALOC")

-			{

-				mstate mspace=(mstate) _mem[1];

-				size_t size=_mem[2];

-				if(isforeign) *isforeign=0;

-				return size;

-			}

-		}

-#elif USE_ALLOCATOR==1

-		if(nedblkmstate(mem))

-		{

-			mchunkptr p=mem2chunk(mem);

-			if(isforeign) *isforeign=0;

-			return chunksize(p)-overhead_for(p);

-		}

-#ifdef DEBUG

-		else

-		{

-			int a=1; /* Set breakpoints here if needed */

-		}

-#endif

-#endif

-#if defined(ENABLE_TOLERANT_NEDMALLOC) || USE_ALLOCATOR==0

-#ifdef _MSC_VER

-		/* This is the MSVCRT equivalent */

-		return _msize(mem);

-#elif defined(__linux__)

-		/* This is the glibc/ptmalloc2/dlmalloc equivalent.  */

-		return malloc_usable_size(mem);

-#elif defined(__FreeBSD__) || defined(__APPLE__)

-		/* This is the BSD libc equivalent.  */

-		return malloc_size(mem);

-#else

-#error Cannot tolerate the memory allocator of an unknown system!

-#endif

-#endif

-	}

-	return 0;

-}

-

-NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC											{ nedpsetvalue((nedpool *) 0, v); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC						{ return nedpmalloc((nedpool *) 0, size); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC			{ return nedpcalloc((nedpool *) 0, no, size); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC			{ return nedprealloc((nedpool *) 0, mem, size); }

-NEDMALLOCNOALIASATTR void   nedfree(void *mem) THROWSPEC											{ nedpfree((nedpool *) 0, mem); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC	{ return nedpmemalign((nedpool *) 0, alignment, bytes); }

-NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC									{ return nedpmallinfo((nedpool *) 0); }

-NEDMALLOCNOALIASATTR int    nedmallopt(int parno, int value) THROWSPEC								{ return nedpmallopt((nedpool *) 0, parno, value); }

-NEDMALLOCNOALIASATTR int    nedmalloc_trim(size_t pad) THROWSPEC									{ return nedpmalloc_trim((nedpool *) 0, pad); }

-void   nedmalloc_stats() THROWSPEC																	{ nedpmalloc_stats((nedpool *) 0); }

-NEDMALLOCNOALIASATTR size_t nedmalloc_footprint() THROWSPEC											{ return nedpmalloc_footprint((nedpool *) 0); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC	{ return nedpindependent_calloc((nedpool *) 0, elemsno, elemsize, chunks); }

-NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC		{ return nedpindependent_comalloc((nedpool *) 0, elems, sizes, chunks); }

-

-struct threadcacheblk_t;

-typedef struct threadcacheblk_t threadcacheblk;

-struct threadcacheblk_t

-{	/* Keep less than 16 bytes on 32 bit systems and 32 bytes on 64 bit systems */

-#ifdef FULLSANITYCHECKS

-	unsigned int magic;

-#endif

-	unsigned int lastUsed, size;

-	threadcacheblk *next, *prev;

-};

-typedef struct threadcache_t

-{

-#ifdef FULLSANITYCHECKS

-	unsigned int magic1;

-#endif

-	int mymspace;						/* Last mspace entry this thread used */

-	long threadid;

-	unsigned int mallocs, frees, successes;

-	size_t freeInCache;					/* How much free space is stored in this cache */

-	threadcacheblk *bins[(THREADCACHEMAXBINS+1)*2];

-#ifdef FULLSANITYCHECKS

-	unsigned int magic2;

-#endif

-} threadcache;

-struct nedpool_t

-{

-	MLOCK_T mutex;

-	void *uservalue;

-	int threads;						/* Max entries in m to use */

-	threadcache *caches[THREADCACHEMAXCACHES];

-	TLSVAR mycache;						/* Thread cache for this thread. 0 for unset, negative for use mspace-1 directly, otherwise is cache-1 */

-	mstate m[MAXTHREADSINPOOL+1];		/* mspace entries for this pool */

-};

-static nedpool syspool;

-

-static FORCEINLINE NEDMALLOCNOALIASATTR unsigned int size2binidx(size_t _size) THROWSPEC

-{	/* 8=1000	16=10000	20=10100	24=11000	32=100000	48=110000	4096=1000000000000 */

-	unsigned int topbit, size=(unsigned int)(_size>>4);

-	/* 16=1		20=1	24=1	32=10	48=11	64=100	96=110	128=1000	4096=100000000 */

-

-#if defined(__GNUC__)

-        topbit = sizeof(size)*__CHAR_BIT__ - 1 - __builtin_clz(size);

-#elif defined(_MSC_VER) && _MSC_VER>=1300

-	{

-            unsigned long bsrTopBit;

-

-            _BitScanReverse(&bsrTopBit, size);

-

-            topbit = bsrTopBit;

-        }

-#else

-#if 0

-	union {

-		unsigned asInt[2];

-		double asDouble;

-	};

-	int n;

-

-	asDouble = (double)size + 0.5;

-	topbit = (asInt[!FOX_BIGENDIAN] >> 20) - 1023;

-#else

-	{

-		unsigned int x=size;

-		x = x | (x >> 1);

-		x = x | (x >> 2);

-		x = x | (x >> 4);

-		x = x | (x >> 8);

-		x = x | (x >>16);

-		x = ~x;

-		x = x - ((x >> 1) & 0x55555555);

-		x = (x & 0x33333333) + ((x >> 2) & 0x33333333);

-		x = (x + (x >> 4)) & 0x0F0F0F0F;

-		x = x + (x << 8);

-		x = x + (x << 16);

-		topbit=31 - (x >> 24);

-	}

-#endif

-#endif

-	return topbit;

-}

-

-

-#ifdef FULLSANITYCHECKS

-static void tcsanitycheck(threadcacheblk **ptr) THROWSPEC

-{

-	assert((ptr[0] && ptr[1]) || (!ptr[0] && !ptr[1]));

-	if(ptr[0] && ptr[1])

-	{

-		assert(nedblksize(ptr[0])>=sizeof(threadcacheblk));

-		assert(nedblksize(ptr[1])>=sizeof(threadcacheblk));

-		assert(*(unsigned int *) "NEDN"==ptr[0]->magic);

-		assert(*(unsigned int *) "NEDN"==ptr[1]->magic);

-		assert(!ptr[0]->prev);

-		assert(!ptr[1]->next);

-		if(ptr[0]==ptr[1])

-		{

-			assert(!ptr[0]->next);

-			assert(!ptr[1]->prev);

-		}

-	}

-}

-static void tcfullsanitycheck(threadcache *tc) THROWSPEC

-{

-	threadcacheblk **tcbptr=tc->bins;

-	int n;

-	for(n=0; n<=THREADCACHEMAXBINS; n++, tcbptr+=2)

-	{

-		threadcacheblk *b, *ob=0;

-		tcsanitycheck(tcbptr);

-		for(b=tcbptr[0]; b; ob=b, b=b->next)

-		{

-			assert(*(unsigned int *) "NEDN"==b->magic);

-			assert(!ob || ob->next==b);

-			assert(!ob || b->prev==ob);

-		}

-	}

-}

-#endif

-

-static NOINLINE void RemoveCacheEntries(nedpool *RESTRICT p, threadcache *RESTRICT tc, unsigned int age) THROWSPEC

-{

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-	if(tc->freeInCache)

-	{

-		threadcacheblk **tcbptr=tc->bins;

-		int n;

-		for(n=0; n<=THREADCACHEMAXBINS; n++, tcbptr+=2)

-		{

-			threadcacheblk **tcb=tcbptr+1;		/* come from oldest end of list */

-			/*tcsanitycheck(tcbptr);*/

-			for(; *tcb && tc->frees-(*tcb)->lastUsed>=age; )

-			{

-				threadcacheblk *f=*tcb;

-				size_t blksize=f->size; /*nedblksize(f);*/

-				assert(blksize<=nedblksize(0, f));

-				assert(blksize);

-#ifdef FULLSANITYCHECKS

-				assert(*(unsigned int *) "NEDN"==(*tcb)->magic);

-#endif

-				*tcb=(*tcb)->prev;

-				if(*tcb)

-					(*tcb)->next=0;

-				else

-					*tcbptr=0;

-				tc->freeInCache-=blksize;

-				assert((long) tc->freeInCache>=0);

-				CallFree(0, f, 0);

-				/*tcsanitycheck(tcbptr);*/

-			}

-		}

-	}

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-}

-static void DestroyCaches(nedpool *RESTRICT p) THROWSPEC

-{

-	if(p->caches)

-	{

-		threadcache *tc;

-		int n;

-		for(n=0; n<THREADCACHEMAXCACHES; n++)

-		{

-			if((tc=p->caches[n]))

-			{

-				tc->frees++;

-				RemoveCacheEntries(p, tc, 0);

-				assert(!tc->freeInCache);

-				tc->mymspace=-1;

-				tc->threadid=0;

-				CallFree(0, tc, 0);

-				p->caches[n]=0;

-			}

-		}

-	}

-}

-

-static NOINLINE threadcache *AllocCache(nedpool *RESTRICT p) THROWSPEC

-{

-	threadcache *tc=0;

-	int n, end;

-	ACQUIRE_LOCK(&p->mutex);

-	for(n=0; n<THREADCACHEMAXCACHES && p->caches[n]; n++);

-	if(THREADCACHEMAXCACHES==n)

-	{	/* List exhausted, so disable for this thread */

-		RELEASE_LOCK(&p->mutex);

-		return 0;

-	}

-	tc=p->caches[n]=(threadcache *) CallCalloc(p->m[0], sizeof(threadcache), 0);

-	if(!tc)

-	{

-		RELEASE_LOCK(&p->mutex);

-		return 0;

-	}

-#ifdef FULLSANITYCHECKS

-	tc->magic1=*(unsigned int *)"NEDMALC1";

-	tc->magic2=*(unsigned int *)"NEDMALC2";

-#endif

-	tc->threadid=(long)(size_t)CURRENT_THREAD;

-	for(end=0; p->m[end]; end++);

-	tc->mymspace=abs(tc->threadid) % end;

-	RELEASE_LOCK(&p->mutex);

-	if(TLSSET(p->mycache, (void *)(size_t)(n+1))) abort();

-	return tc;

-}

-

-static void *threadcache_malloc(nedpool *RESTRICT p, threadcache *RESTRICT tc, size_t *RESTRICT _size) THROWSPEC

-{

-	void *RESTRICT ret=0;

-	size_t size=*_size, blksize=0;

-	unsigned int bestsize;

-	unsigned int idx=size2binidx(size);

-	threadcacheblk *RESTRICT blk, **RESTRICT binsptr;

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-	/* Calculate best fit bin size */

-	bestsize=1<<(idx+4);

-#if 0

-	/* Finer grained bin fit */

-	idx<<=1;

-	if(size>bestsize)

-	{

-		idx++;

-		bestsize+=bestsize>>1;

-	}

-	if(size>bestsize)

-	{

-		idx++;

-		bestsize=1<<(4+(idx>>1));

-	}

-#else

-	if(size>bestsize)

-	{

-		idx++;

-		bestsize<<=1;

-	}

-#endif

-	assert(bestsize>=size);

-	if(size<bestsize) size=bestsize;

-	assert(size<=THREADCACHEMAX);

-	assert(idx<=THREADCACHEMAXBINS);

-	binsptr=&tc->bins[idx*2];

-	/* Try to match close, but move up a bin if necessary */

-	blk=*binsptr;

-	if(!blk || blk->size<size)

-	{	/* Bump it up a bin */

-		if(idx<THREADCACHEMAXBINS)

-		{

-			idx++;

-			binsptr+=2;

-			blk=*binsptr;

-		}

-	}

-	if(blk)

-	{

-		blksize=blk->size; /*nedblksize(blk);*/

-		assert(nedblksize(0, blk)>=blksize);

-		assert(blksize>=size);

-		if(blk->next)

-			blk->next->prev=0;

-		*binsptr=blk->next;

-		if(!*binsptr)

-			binsptr[1]=0;

-#ifdef FULLSANITYCHECKS

-		blk->magic=0;

-#endif

-		assert(binsptr[0]!=blk && binsptr[1]!=blk);

-		assert(nedblksize(0, blk)>=sizeof(threadcacheblk) && nedblksize(0, blk)<=THREADCACHEMAX+CHUNK_OVERHEAD);

-		/*printf("malloc: %p, %p, %p, %lu\n", p, tc, blk, (long) _size);*/

-		ret=(void *) blk;

-	}

-	++tc->mallocs;

-	if(ret)

-	{

-		assert(blksize>=size);

-		++tc->successes;

-		tc->freeInCache-=blksize;

-		assert((long) tc->freeInCache>=0);

-	}

-#if defined(DEBUG) && 0

-	if(!(tc->mallocs & 0xfff))

-	{

-		printf("*** threadcache=%u, mallocs=%u (%f), free=%u (%f), freeInCache=%u\n", (unsigned int) tc->threadid, tc->mallocs,

-			(float) tc->successes/tc->mallocs, tc->frees, (float) tc->successes/tc->frees, (unsigned int) tc->freeInCache);

-	}

-#endif

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-	*_size=size;

-	return ret;

-}

-static NOINLINE void ReleaseFreeInCache(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace) THROWSPEC

-{

-	unsigned int age=THREADCACHEMAXFREESPACE/8192;

-	/*ACQUIRE_LOCK(&p->m[mymspace]->mutex);*/

-	while(age && tc->freeInCache>=THREADCACHEMAXFREESPACE)

-	{

-		RemoveCacheEntries(p, tc, age);

-		/*printf("*** Removing cache entries older than %u (%u)\n", age, (unsigned int) tc->freeInCache);*/

-		age>>=1;

-	}

-	/*RELEASE_LOCK(&p->m[mymspace]->mutex);*/

-}

-static void threadcache_free(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace, void *RESTRICT mem, size_t size) THROWSPEC

-{

-	unsigned int bestsize;

-	unsigned int idx=size2binidx(size);

-	threadcacheblk **RESTRICT binsptr, *RESTRICT tck=(threadcacheblk *) mem;

-	assert(size>=sizeof(threadcacheblk) && size<=THREADCACHEMAX+CHUNK_OVERHEAD);

-#ifdef DEBUG

-	/* Make sure this is a valid memory block */

-	assert(nedblksize(0, mem));

-#endif

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-	/* Calculate best fit bin size */

-	bestsize=1<<(idx+4);

-#if 0

-	/* Finer grained bin fit */

-	idx<<=1;

-	if(size>bestsize)

-	{

-		unsigned int biggerbestsize=bestsize+bestsize<<1;

-		if(size>=biggerbestsize)

-		{

-			idx++;

-			bestsize=biggerbestsize;

-		}

-	}

-#endif

-	if(bestsize!=size)	/* dlmalloc can round up, so we round down to preserve indexing */

-		size=bestsize;

-	binsptr=&tc->bins[idx*2];

-	assert(idx<=THREADCACHEMAXBINS);

-	if(tck==*binsptr)

-	{

-		fprintf(stderr, "nedmalloc: Attempt to free already freed memory block %p - aborting!\n", tck);

-		abort();

-	}

-#ifdef FULLSANITYCHECKS

-	tck->magic=*(unsigned int *) "NEDN";

-#endif

-	tck->lastUsed=++tc->frees;

-	tck->size=(unsigned int) size;

-	tck->next=*binsptr;

-	tck->prev=0;

-	if(tck->next)

-		tck->next->prev=tck;

-	else

-		binsptr[1]=tck;

-	assert(!*binsptr || (*binsptr)->size==tck->size);

-	*binsptr=tck;

-	assert(tck==tc->bins[idx*2]);

-	assert(tc->bins[idx*2+1]==tck || binsptr[0]->next->prev==tck);

-	/*printf("free: %p, %p, %p, %lu\n", p, tc, mem, (long) size);*/

-	tc->freeInCache+=size;

-#ifdef FULLSANITYCHECKS

-	tcfullsanitycheck(tc);

-#endif

-#if 1

-	if(tc->freeInCache>=THREADCACHEMAXFREESPACE)

-		ReleaseFreeInCache(p, tc, mymspace);

-#endif

-}

-

-

-

-

-static NOINLINE int InitPool(nedpool *RESTRICT p, size_t capacity, int threads) THROWSPEC

-{	/* threads is -1 for system pool */

-	ensure_initialization();

-	ACQUIRE_MALLOC_GLOBAL_LOCK();

-	if(p->threads) goto done;

-	if(INITIAL_LOCK(&p->mutex)) goto err;

-	if(TLSALLOC(&p->mycache)) goto err;

-#if USE_ALLOCATOR==0

-	p->m[0]=(mstate) mspacecounter++;

-#elif USE_ALLOCATOR==1

-	if(!(p->m[0]=(mstate) create_mspace(capacity, 1))) goto err;

-	p->m[0]->extp=p;

-#endif

-	p->threads=(threads<1 || threads>MAXTHREADSINPOOL) ? MAXTHREADSINPOOL : threads;

-done:

-	RELEASE_MALLOC_GLOBAL_LOCK();

-	return 1;

-err:

-	if(threads<0)

-		abort();			/* If you can't allocate for system pool, we're screwed */

-	DestroyCaches(p);

-	if(p->m[0])

-	{

-#if USE_ALLOCATOR==1

-		destroy_mspace(p->m[0]);

-#endif

-		p->m[0]=0;

-	}

-	if(p->mycache)

-	{

-		if(TLSFREE(p->mycache)) abort();

-		p->mycache=0;

-	}

-	RELEASE_MALLOC_GLOBAL_LOCK();

-	return 0;

-}

-static NOINLINE mstate FindMSpace(nedpool *RESTRICT p, threadcache *RESTRICT tc, int *RESTRICT lastUsed, size_t size) THROWSPEC

-{	/* Gets called when thread's last used mspace is in use. The strategy

-	is to run through the list of all available mspaces looking for an

-	unlocked one and if we fail, we create a new one so long as we don't

-	exceed p->threads */

-	int n, end;

-	for(n=end=*lastUsed+1; p->m[n]; end=++n)

-	{

-		if(TRY_LOCK(&p->m[n]->mutex)) goto found;

-	}

-	for(n=0; n<*lastUsed && p->m[n]; n++)

-	{

-		if(TRY_LOCK(&p->m[n]->mutex)) goto found;

-	}

-	if(end<p->threads)

-	{

-		mstate temp;

-#if USE_ALLOCATOR==0

-		temp=(mstate) mspacecounter++;

-#elif USE_ALLOCATOR==1

-		if(!(temp=(mstate) create_mspace(size, 1)))

-			goto badexit;

-#endif

-		/* Now we're ready to modify the lists, we lock */

-		ACQUIRE_LOCK(&p->mutex);

-		while(p->m[end] && end<p->threads)

-			end++;

-		if(end>=p->threads)

-		{	/* Drat, must destroy it now */

-			RELEASE_LOCK(&p->mutex);

-#if USE_ALLOCATOR==1

-			destroy_mspace((mstate) temp);

-#endif

-			goto badexit;

-		}

-		/* We really want to make sure this goes into memory now but we

-		have to be careful of breaking aliasing rules, so write it twice */

-		*((volatile struct malloc_state **) &p->m[end])=p->m[end]=temp;

-		ACQUIRE_LOCK(&p->m[end]->mutex);

-		/*printf("Created mspace idx %d\n", end);*/

-		RELEASE_LOCK(&p->mutex);

-		n=end;

-		goto found;

-	}

-	/* Let it lock on the last one it used */

-badexit:

-	ACQUIRE_LOCK(&p->m[*lastUsed]->mutex);

-	return p->m[*lastUsed];

-found:

-	*lastUsed=n;

-	if(tc)

-		tc->mymspace=n;

-	else

-	{

-		if(TLSSET(p->mycache, (void *)(size_t)(-(n+1)))) abort();

-	}

-	return p->m[n];

-}

-

-typedef struct PoolList_t

-{

-	size_t size;			/* Size of list */

-	size_t length;			/* Actual entries in list */

-#ifdef DEBUG

-	nedpool *list[1];		/* Force testing of list expansion */

-#else

-	nedpool *list[16];

-#endif

-} PoolList;

-static MLOCK_T poollistlock;

-static PoolList *poollist;

-NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC

-{

-	nedpool *ret=0;

-	if(!poollist)

-	{

-		PoolList *newpoollist=0;

-		if(!(newpoollist=(PoolList *) nedpcalloc(0, 1, sizeof(PoolList)+sizeof(nedpool *)))) return 0;

-		INITIAL_LOCK(&poollistlock);

-		ACQUIRE_LOCK(&poollistlock);

-		poollist=newpoollist;

-		poollist->size=sizeof(poollist->list)/sizeof(nedpool *);

-	}

-	else

-		ACQUIRE_LOCK(&poollistlock);

-	if(poollist->length==poollist->size)

-	{

-		PoolList *newpoollist=0;

-		size_t newsize=0;

-		newsize=sizeof(PoolList)+(poollist->size+1)*sizeof(nedpool *);

-		if(!(newpoollist=(PoolList *) nedprealloc(0, poollist, newsize))) goto badexit;

-		poollist=newpoollist;

-		memset(&poollist->list[poollist->size], 0, newsize-((size_t)&poollist->list[poollist->size]-(size_t)&poollist->list[0]));

-		poollist->size=((newsize-((char *)&poollist->list[0]-(char *)poollist))/sizeof(nedpool *))-1;

-		assert(poollist->size>poollist->length);

-	}

-	if(!(ret=(nedpool *) nedpcalloc(0, 1, sizeof(nedpool)))) goto badexit;

-	if(!InitPool(ret, capacity, threads))

-	{

-		nedpfree(0, ret);

-		goto badexit;

-	}

-	poollist->list[poollist->length++]=ret;

-badexit:

-	RELEASE_LOCK(&poollistlock);

-	return ret;

-}

-void neddestroypool(nedpool *p) THROWSPEC

-{

-	unsigned int n;

-	ACQUIRE_LOCK(&p->mutex);

-	DestroyCaches(p);

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1

-		destroy_mspace(p->m[n]);

-#endif

-		p->m[n]=0;

-	}

-	RELEASE_LOCK(&p->mutex);

-	if(TLSFREE(p->mycache)) abort();

-	nedpfree(0, p);

-	ACQUIRE_LOCK(&poollistlock);

-	assert(poollist);

-	for(n=0; n<poollist->length && poollist->list[n]!=p; n++);

-	assert(n!=poollist->length);

-	memmove(&poollist->list[n], &poollist->list[n+1], (size_t)&poollist->list[poollist->length]-(size_t)&poollist->list[n]);

-	if(!--poollist->length)

-	{

-		assert(!poollist->list[0]);

-		nedpfree(0, poollist);

-		poollist=0;

-	}

-	RELEASE_LOCK(&poollistlock);

-}

-void neddestroysyspool() THROWSPEC

-{

-	nedpool *p=&syspool;

-	int n;

-	ACQUIRE_LOCK(&p->mutex);

-	DestroyCaches(p);

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1

-		destroy_mspace(p->m[n]);

-#endif

-		p->m[n]=0;

-	}

-	/* Render syspool unusable */

-	for(n=0; n<THREADCACHEMAXCACHES; n++)

-		p->caches[n]=(threadcache *)(size_t)(sizeof(size_t)>4 ? 0xdeadbeefdeadbeefULL : 0xdeadbeefUL);

-	for(n=0; n<MAXTHREADSINPOOL+1; n++)

-		p->m[n]=(mstate)(size_t)(sizeof(size_t)>4 ? 0xdeadbeefdeadbeefULL : 0xdeadbeefUL);

-	if(TLSFREE(p->mycache)) abort();

-	RELEASE_LOCK(&p->mutex);

-}

-nedpool **nedpoollist() THROWSPEC

-{

-	nedpool **ret=0;

-	if(poollist)

-	{

-		ACQUIRE_LOCK(&poollistlock);

-		if(!(ret=(nedpool **) nedmalloc((poollist->length+1)*sizeof(nedpool *)))) goto badexit;

-		memcpy(ret, poollist->list, (poollist->length+1)*sizeof(nedpool *));

-badexit:

-		RELEASE_LOCK(&poollistlock);

-	}

-	return ret;

-}

-

-void nedpsetvalue(nedpool *p, void *v) THROWSPEC

-{

-	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }

-	p->uservalue=v;

-}

-void *nedgetvalue(nedpool **p, void *mem) THROWSPEC

-{

-	nedpool *np=0;

-	mstate fm=nedblkmstate(mem);

-	if(!fm || !fm->extp) return 0;

-	np=(nedpool *) fm->extp;

-	if(p) *p=np;

-	return np->uservalue;

-}

-

-void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC

-{

-	int mycache;

-	if(!p)

-	{

-		p=&syspool;

-		if(!syspool.threads) InitPool(&syspool, 0, -1);

-	}

-	mycache=(int)(size_t) TLSGET(p->mycache);

-	if(!mycache)

-	{	/* Set to mspace 0 */

-		if(disable && TLSSET(p->mycache, (void *)(size_t)-1)) abort();

-	}

-	else if(mycache>0)

-	{	/* Set to last used mspace */

-		threadcache *tc=p->caches[mycache-1];

-#if defined(DEBUG)

-		printf("Threadcache utilisation: %lf%% in cache with %lf%% lost to other threads\n",

-			100.0*tc->successes/tc->mallocs, 100.0*((double) tc->mallocs-tc->frees)/tc->mallocs);

-#endif

-		if(disable && TLSSET(p->mycache, (void *)(size_t)(-tc->mymspace))) abort();

-		tc->frees++;

-		RemoveCacheEntries(p, tc, 0);

-		assert(!tc->freeInCache);

-		if(disable)

-		{

-			tc->mymspace=-1;

-			tc->threadid=0;

-			CallFree(0, p->caches[mycache-1], 0);

-			p->caches[mycache-1]=0;

-		}

-	}

-}

-void neddisablethreadcache(nedpool *p) THROWSPEC

-{

-	nedtrimthreadcache(p, 1);

-}

-

-#define GETMSPACE(m,p,tc,ms,s,action)                 \

-  do                                                  \

-  {                                                   \

-    mstate m = GetMSpace((p),(tc),(ms),(s));          \

-    action;                                           \

-	if(USE_ALLOCATOR==1) { RELEASE_LOCK(&m->mutex); } \

-  } while (0)

-

-static FORCEINLINE mstate GetMSpace(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace, size_t size) THROWSPEC

-{	/* Returns a locked and ready for use mspace */

-	mstate m=p->m[mymspace];

-	assert(m);

-#if USE_ALLOCATOR==1

-	if(!TRY_LOCK(&p->m[mymspace]->mutex)) m=FindMSpace(p, tc, &mymspace, size);

-	/*assert(IS_LOCKED(&p->m[mymspace]->mutex));*/

-#endif

-	return m;

-}

-static NOINLINE void GetThreadCache_cold1(nedpool *RESTRICT *RESTRICT p) THROWSPEC

-{

-	*p=&syspool;

-	if(!syspool.threads) InitPool(&syspool, 0, -1);

-}

-static NOINLINE void GetThreadCache_cold2(nedpool *RESTRICT *RESTRICT p, threadcache *RESTRICT *RESTRICT tc, int *RESTRICT mymspace, int mycache) THROWSPEC

-{

-	if(!mycache)

-	{	/* Need to allocate a new cache */

-		*tc=AllocCache(*p);

-		if(!*tc)

-		{	/* Disable */

-			if(TLSSET((*p)->mycache, (void *)(size_t)-1)) abort();

-			*mymspace=0;

-		}

-		else

-			*mymspace=(*tc)->mymspace;

-	}

-	else

-	{	/* Cache disabled, but we do have an assigned thread pool */

-		*tc=0;

-		*mymspace=-mycache-1;

-	}

-}

-static FORCEINLINE void GetThreadCache(nedpool *RESTRICT *RESTRICT p, threadcache *RESTRICT *RESTRICT tc, int *RESTRICT mymspace, size_t *RESTRICT size) THROWSPEC

-{

-	int mycache;

-	if(size && *size<sizeof(threadcacheblk)) *size=sizeof(threadcacheblk);

-	if(!*p)

-		GetThreadCache_cold1(p);

-	mycache=(int)(size_t) TLSGET((*p)->mycache);

-	if(mycache>0)

-	{	/* Already have a cache */

-		*tc=(*p)->caches[mycache-1];

-		*mymspace=(*tc)->mymspace;

-	}

-	else GetThreadCache_cold2(p, tc, mymspace, mycache);

-	assert(*mymspace>=0);

-	assert(!(*tc) || (long)(size_t)CURRENT_THREAD==(*tc)->threadid);

-#ifdef FULLSANITYCHECKS

-	if(*tc)

-	{

-		if(*(unsigned int *)"NEDMALC1"!=(*tc)->magic1 || *(unsigned int *)"NEDMALC2"!=(*tc)->magic2)

-		{

-			abort();

-		}

-	}

-#endif

-}

-

-NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC

-{

-	void *ret=0;

-	threadcache *tc;

-	int mymspace;

-	GetThreadCache(&p, &tc, &mymspace, &size);

-#if THREADCACHEMAX

-	if(tc && size<=THREADCACHEMAX)

-	{	/* Use the thread cache */

-		ret=threadcache_malloc(p, tc, &size);

-	}

-#endif

-	if(!ret)

-	{	/* Use this thread's mspace */

-        GETMSPACE(m, p, tc, mymspace, size,

-                  ret=CallMalloc(m, size, 0));

-	}

-	return ret;

-}

-NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC

-{

-	size_t rsize=size*no;

-	void *ret=0;

-	threadcache *tc;

-	int mymspace;

-	GetThreadCache(&p, &tc, &mymspace, &rsize);

-#if THREADCACHEMAX

-	if(tc && rsize<=THREADCACHEMAX)

-	{	/* Use the thread cache */

-		if((ret=threadcache_malloc(p, tc, &rsize)))

-			memset(ret, 0, rsize);

-	}

-#endif

-	if(!ret)

-	{	/* Use this thread's mspace */

-        GETMSPACE(m, p, tc, mymspace, rsize,

-                  ret=CallCalloc(m, rsize, 0));

-	}

-	return ret;

-}

-NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC

-{

-	void *ret=0;

-	threadcache *tc;

-	int mymspace, isforeign=1;

-	size_t memsize;

-	if(!mem) return nedpmalloc(p, size);

-	memsize=nedblksize(&isforeign, mem);

-	assert(memsize);

-	if(!memsize)

-	{

-		fprintf(stderr, "nedmalloc: nedprealloc() called with a block not created by nedmalloc!\n");

-		abort();

-	}

-	else if(size<=memsize && memsize-size<

-#ifdef DEBUG

-		32

-#else

-		1024

-#endif

-		)		/* If realloc size is within 1Kb smaller than existing, noop it */

-		return mem;

-	GetThreadCache(&p, &tc, &mymspace, &size);

-#if THREADCACHEMAX

-	if(tc && size && size<=THREADCACHEMAX)

-	{	/* Use the thread cache */

-		if((ret=threadcache_malloc(p, tc, &size)))

-		{

-			memcpy(ret, mem, memsize<size ? memsize : size);

-			if(memsize>=sizeof(threadcacheblk) && memsize<=(THREADCACHEMAX+CHUNK_OVERHEAD))

-				threadcache_free(p, tc, mymspace, mem, memsize);

-			else

-				CallFree(0, mem, isforeign);

-		}

-	}

-#endif

-	if(!ret)

-	{	/* Reallocs always happen in the mspace they happened in, so skip

-		locking the preferred mspace for this thread */

-		ret=CallRealloc(p->m[mymspace], mem, isforeign, memsize, size);

-	}

-	return ret;

-}

-void   nedpfree(nedpool *p, void *mem) THROWSPEC

-{	/* Frees always happen in the mspace they happened in, so skip

-	locking the preferred mspace for this thread */

-	threadcache *tc;

-	int mymspace, isforeign=1;

-	size_t memsize;

-	if(!mem)

-	{	/* If you tried this on FreeBSD you'd be sorry! */

-#ifdef DEBUG

-		fprintf(stderr, "nedmalloc: WARNING nedpfree() called with zero. This is not portable behaviour!\n");

-#endif

-		return;

-	}

-	memsize=nedblksize(&isforeign, mem);

-	assert(memsize);

-	if(!memsize)

-	{

-		fprintf(stderr, "nedmalloc: nedpfree() called with a block not created by nedmalloc!\n");

-		abort();

-	}

-	GetThreadCache(&p, &tc, &mymspace, 0);

-#if THREADCACHEMAX

-	if(mem && tc && memsize>=sizeof(threadcacheblk) && memsize<=(THREADCACHEMAX+CHUNK_OVERHEAD))

-		threadcache_free(p, tc, mymspace, mem, memsize);

-	else

-#endif

-		CallFree(0, mem, isforeign);

-}

-NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC

-{

-	void *ret;

-	threadcache *tc;

-	int mymspace;

-	GetThreadCache(&p, &tc, &mymspace, &bytes);

-	{	/* Use this thread's mspace */

-        GETMSPACE(m, p, tc, mymspace, bytes,

-                  ret=CallMalloc(m, bytes, alignment));

-	}

-	return ret;

-}

-struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC

-{

-	int n;

-	struct nedmallinfo ret={0};

-	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1 && !NO_MALLINFO

-		struct mallinfo t=mspace_mallinfo(p->m[n]);

-		ret.arena+=t.arena;

-		ret.ordblks+=t.ordblks;

-		ret.hblkhd+=t.hblkhd;

-		ret.usmblks+=t.usmblks;

-		ret.uordblks+=t.uordblks;

-		ret.fordblks+=t.fordblks;

-		ret.keepcost+=t.keepcost;

-#endif

-	}

-	return ret;

-}

-int    nedpmallopt(nedpool *p, int parno, int value) THROWSPEC

-{

-#if USE_ALLOCATOR==1

-	return mspace_mallopt(parno, value);

-#else

-	return 0;

-#endif

-}

-NEDMALLOCNOALIASATTR void*  nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC

-{

-#if USE_ALLOCATOR==1

-	if(granularity) *granularity=mparams.granularity;

-	if(magic) *magic=mparams.magic;

-	return (void *) &syspool;

-#else

-	if(granularity) *granularity=0;

-	if(magic) *magic=0;

-	return 0;

-#endif

-}

-int    nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC

-{

-	int n, ret=0;

-	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1

-		ret+=mspace_trim(p->m[n], pad);

-#endif

-	}

-	return ret;

-}

-void   nedpmalloc_stats(nedpool *p) THROWSPEC

-{

-	int n;

-	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1

-		mspace_malloc_stats(p->m[n]);

-#endif

-	}

-}

-size_t nedpmalloc_footprint(nedpool *p) THROWSPEC

-{

-	size_t ret=0;

-	int n;

-	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }

-	for(n=0; p->m[n]; n++)

-	{

-#if USE_ALLOCATOR==1

-		ret+=mspace_footprint(p->m[n]);

-#endif

-	}

-	return ret;

-}

-NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC

-{

-	void **ret;

-	threadcache *tc;

-	int mymspace;

-	GetThreadCache(&p, &tc, &mymspace, &elemsize);

-#if USE_ALLOCATOR==0

-    GETMSPACE(m, p, tc, mymspace, elemsno*elemsize,

-              ret=unsupported_operation("independent_calloc"));

-#elif USE_ALLOCATOR==1

-    GETMSPACE(m, p, tc, mymspace, elemsno*elemsize,

-              ret=mspace_independent_calloc(m, elemsno, elemsize, chunks));

-#endif

-	return ret;

-}

-NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC

-{

-	void **ret;

-	threadcache *tc;

-	int mymspace;

-    size_t i, *adjustedsizes=(size_t *) alloca(elems*sizeof(size_t));

-    if(!adjustedsizes) return 0;

-    for(i=0; i<elems; i++)

-        adjustedsizes[i]=sizes[i]<sizeof(threadcacheblk) ? sizeof(threadcacheblk) : sizes[i];

-	GetThreadCache(&p, &tc, &mymspace, 0);

-#if USE_ALLOCATOR==0

-	GETMSPACE(m, p, tc, mymspace, 0,

-              ret=unsupported_operation("independent_comalloc"));

-#elif USE_ALLOCATOR==1

-	GETMSPACE(m, p, tc, mymspace, 0,

-              ret=mspace_independent_comalloc(m, elems, adjustedsizes, chunks));

-#endif

-	return ret;

-}

-

-#if defined(__cplusplus)

-}

-#endif

-

-#ifdef _MSC_VER

-#pragma warning(pop)

-#endif

-

-#endif

+#ifdef NEDMALLOC_ENABLED
+/* Alternative malloc implementation for multiple threads without
+lock contention based on dlmalloc. (C) 2005-2009 Niall Douglas
+
+Boost Software License - Version 1.0 - August 17th, 2003
+
+Permission is hereby granted, free of charge, to any person or organization
+obtaining a copy of the software and accompanying documentation covered by
+this license (the "Software") to use, reproduce, display, distribute,
+execute, and transmit the Software, and to prepare derivative works of the
+Software, and to permit third-parties to whom the Software is furnished to
+do so, all subject to the following:
+
+The copyright notices in the Software and this entire statement, including
+the above license grant, this restriction and the following disclaimer,
+must be included in all copies of the Software, in whole or in part, and
+all derivative works of the Software, unless such copies or derivative
+works are solely in the form of machine-executable object code generated by
+a source language processor.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
+SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
+FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
+ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+DEALINGS IN THE SOFTWARE.
+*/
+
+#ifdef _MSC_VER
+/* Enable full aliasing on MSVC */
+/*#pragma optimize("a", on)*/
+#pragma warning(push)
+#pragma warning(disable:4100)	/* unreferenced formal parameter */
+#pragma warning(disable:4127)	/* conditional expression is constant */
+#pragma warning(disable:4706)	/* assignment within conditional expression */
+#endif
+
+/*#define ENABLE_TOLERANT_NEDMALLOC 1*/
+/*#define ENABLE_FAST_HEAP_DETECTION 1*/
+/*#define NEDMALLOC_DEBUG 1*/
+
+/*#define FULLSANITYCHECKS*/
+/* If link time code generation is on, don't force or prevent inlining */
+#if defined(_MSC_VER) && defined(NEDMALLOC_DLL_EXPORTS)
+#define FORCEINLINE
+#define NOINLINE
+#endif
+
+
+#include "nedmalloc.h"
+#ifdef WIN32
+ #include <malloc.h>
+ #include <stddef.h>
+#endif
+#if USE_ALLOCATOR==1
+ #define MSPACES 1
+ #define ONLY_MSPACES 1
+#endif
+#define USE_DL_PREFIX 1
+#ifndef USE_LOCKS
+ #define USE_LOCKS 1
+#endif
+#define FOOTERS 1           /* Need to enable footers so frees lock the right mspace */
+#ifndef NEDMALLOC_DEBUG
+ #if defined(DEBUG) || defined(_DEBUG)
+  #define NEDMALLOC_DEBUG 1
+ #else
+  #define NEDMALLOC_DEBUG 0
+ #endif
+#endif
+/* We need to consistently define DEBUG=0|1, _DEBUG and NDEBUG for dlmalloc */
+#undef DEBUG
+#undef _DEBUG
+#if NEDMALLOC_DEBUG
+ #define _DEBUG
+ #define DEBUG 1
+#else
+ #define DEBUG 0
+#endif
+#ifdef NDEBUG               /* Disable assert checking on release builds */
+ #undef DEBUG
+ #undef _DEBUG
+#endif
+/* The default of 64Kb means we spend too much time kernel-side */
+#ifndef DEFAULT_GRANULARITY
+#define DEFAULT_GRANULARITY (1*1024*1024)
+#if DEBUG
+#define DEFAULT_GRANULARITY_ALIGNED
+#endif
+#endif
+/*#define USE_SPIN_LOCKS 0*/
+
+
+#include "malloc.c.h"
+#ifdef NDEBUG               /* Disable assert checking on release builds */
+ #undef DEBUG
+#elif !NEDMALLOC_DEBUG
+ #ifdef __GNUC__
+  #warning DEBUG is defined so allocator will run with assert checking! Define NDEBUG to run at full speed.
+ #elif defined(_MSC_VER)
+  #pragma message(__FILE__ ": WARNING: DEBUG is defined so allocator will run with assert checking! Define NDEBUG to run at full speed.")
+ #endif
+#endif
+
+/* The maximum concurrent threads in a pool possible */
+#ifndef MAXTHREADSINPOOL
+#define MAXTHREADSINPOOL 16
+#endif
+/* The maximum number of threadcaches which can be allocated */
+#ifndef THREADCACHEMAXCACHES
+#define THREADCACHEMAXCACHES 256
+#endif
+/* The maximum size to be allocated from the thread cache */
+#ifndef THREADCACHEMAX
+#define THREADCACHEMAX 8192
+#endif
+#if 0
+/* The number of cache entries for finer grained bins. This is (topbitpos(THREADCACHEMAX)-4)*2 */
+#define THREADCACHEMAXBINS ((13-4)*2)
+#else
+/* The number of cache entries. This is (topbitpos(THREADCACHEMAX)-4) */
+#define THREADCACHEMAXBINS (13-4)
+#endif
+/* Point at which the free space in a thread cache is garbage collected */
+#ifndef THREADCACHEMAXFREESPACE
+#define THREADCACHEMAXFREESPACE (512*1024)
+#endif
+
+
+#ifdef WIN32
+ #define TLSVAR			DWORD
+ #define TLSALLOC(k)	(*(k)=TlsAlloc(), TLS_OUT_OF_INDEXES==*(k))
+ #define TLSFREE(k)		(!TlsFree(k))
+ #define TLSGET(k)		TlsGetValue(k)
+ #define TLSSET(k, a)	(!TlsSetValue(k, a))
+ #ifdef DEBUG
+static LPVOID ChkedTlsGetValue(DWORD idx)
+{
+	LPVOID ret=TlsGetValue(idx);
+	assert(S_OK==GetLastError());
+	return ret;
+}
+  #undef TLSGET
+  #define TLSGET(k) ChkedTlsGetValue(k)
+ #endif
+#else
+ #define TLSVAR			pthread_key_t
+ #define TLSALLOC(k)	pthread_key_create(k, 0)
+ #define TLSFREE(k)		pthread_key_delete(k)
+ #define TLSGET(k)		pthread_getspecific(k)
+ #define TLSSET(k, a)	pthread_setspecific(k, a)
+#endif
+
+#if defined(__cplusplus)
+#if !defined(NO_NED_NAMESPACE)
+namespace nedalloc {
+#else
+extern "C" {
+#endif
+#endif
+
+#if USE_ALLOCATOR==0
+static void *unsupported_operation(const char *opname) THROWSPEC
+{
+	fprintf(stderr, "nedmalloc: The operation %s is not supported under this build configuration\n", opname);
+	abort();
+	return 0;
+}
+static size_t mspacecounter=(size_t) 0xdeadbeef;
+#endif
+#ifndef ENABLE_FAST_HEAP_DETECTION
+static void *RESTRICT leastusedaddress;
+static size_t largestusedblock;
+#endif
+
+static FORCEINLINE void *CallMalloc(void *RESTRICT mspace, size_t size, size_t alignment) THROWSPEC
+{
+	void *RESTRICT ret=0;
+	size_t _alignment=alignment;
+#if USE_MAGIC_HEADERS
+	size_t *_ret=0;
+	size+=alignment+3*sizeof(size_t);
+	_alignment=0;
+#endif
+#if USE_ALLOCATOR==0
+	ret=_alignment ? 
+#ifdef _MSC_VER
+	/* This is the MSVCRT equivalent */
+		_aligned_malloc(size, _alignment)
+#elif defined(__linux__) || defined(__FreeBSD__) || defined(__APPLE__)
+	/* This is the glibc/ptmalloc2/dlmalloc/BSD libc equivalent.  */
+		memalign(_alignment, size)
+#else
+#error Cannot aligned allocate with the memory allocator of an unknown system!
+#endif
+		: malloc(size);
+#elif USE_ALLOCATOR==1
+	ret=_alignment ? mspace_memalign((mstate) mspace, _alignment, size) : mspace_malloc((mstate) mspace, size);
+#ifndef ENABLE_FAST_HEAP_DETECTION
+	if(ret)
+	{
+		size_t truesize=chunksize(mem2chunk(ret));
+		if(!leastusedaddress || (void *)((mstate) mspace)->least_addr<leastusedaddress) leastusedaddress=(void *)((mstate) mspace)->least_addr;
+		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);
+	}
+#endif
+#endif
+	if(!ret) return 0;
+#if USE_MAGIC_HEADERS
+	_ret=(size_t *) ret;
+	ret=(void *)(_ret+3);
+	if(alignment) ret=(void *)(((size_t) ret+alignment-1)&~(alignment-1));
+	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *)"NEDMALOC";
+	_ret[0]=(size_t) mspace;
+	_ret[1]=size-3*sizeof(size_t);
+#endif
+	return ret;
+}
+
+static FORCEINLINE void *CallCalloc(void *RESTRICT mspace, size_t size, size_t alignment) THROWSPEC
+{
+	void *RESTRICT ret=0;
+#if USE_MAGIC_HEADERS
+	size_t *_ret=0;
+	size+=alignment+3*sizeof(size_t);
+#endif
+#if USE_ALLOCATOR==0
+	ret=calloc(1, size);
+#elif USE_ALLOCATOR==1
+	ret=mspace_calloc((mstate) mspace, 1, size);
+#ifndef ENABLE_FAST_HEAP_DETECTION
+	if(ret)
+	{
+		size_t truesize=chunksize(mem2chunk(ret));
+		if(!leastusedaddress || (void *)((mstate) mspace)->least_addr<leastusedaddress) leastusedaddress=(void *)((mstate) mspace)->least_addr;
+		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);
+	}
+#endif
+#endif
+	if(!ret) return 0;
+#if USE_MAGIC_HEADERS
+	_ret=(size_t *) ret;
+	ret=(void *)(_ret+3);
+	if(alignment) ret=(void *)(((size_t) ret+alignment-1)&~(alignment-1));
+	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *) "NEDMALOC";
+	_ret[0]=(size_t) mspace;
+	_ret[1]=size-3*sizeof(size_t);
+#endif
+	return ret;
+}
+
+static FORCEINLINE void *CallRealloc(void *RESTRICT mspace, void *RESTRICT mem, int isforeign, size_t oldsize, size_t newsize) THROWSPEC
+{
+	void *RESTRICT ret=0;
+#if USE_MAGIC_HEADERS
+	mstate oldmspace=0;
+	size_t *_ret=0, *_mem=(size_t *) mem-3;
+#endif
+	if(isforeign)
+	{	/* Transfer */
+#if USE_MAGIC_HEADERS
+		assert(_mem[0]!=*(size_t *) "NEDMALOC");
+#endif
+		if((ret=CallMalloc(mspace, newsize, 0)))
+		{
+#if defined(DEBUG)
+			printf("*** nedmalloc frees system allocated block %p\n", mem);
+#endif
+			memcpy(ret, mem, oldsize<newsize ? oldsize : newsize);
+			free(mem);
+		}
+		return ret;
+	}
+#if USE_MAGIC_HEADERS
+	assert(_mem[0]==*(size_t *) "NEDMALOC");
+	newsize+=3*sizeof(size_t);
+	oldmspace=(mstate) _mem[1];
+	assert(oldsize>=_mem[2]);
+	for(; *_mem==*(size_t *) "NEDMALOC"; *_mem--=*(size_t *) "nedmaloc");
+	mem=(void *)(++_mem);
+#endif
+#if USE_ALLOCATOR==0
+	ret=realloc(mem, newsize);
+#elif USE_ALLOCATOR==1
+	ret=mspace_realloc((mstate) mspace, mem, newsize);
+#ifndef ENABLE_FAST_HEAP_DETECTION
+	if(ret)
+	{
+		size_t truesize=chunksize(mem2chunk(ret));
+		if(!largestusedblock || truesize>largestusedblock) largestusedblock=(truesize+mparams.page_size) & ~(mparams.page_size-1);
+	}
+#endif
+#endif
+	if(!ret)
+	{	/* Put it back the way it was */
+#if USE_MAGIC_HEADERS
+		for(; *_mem==0; *_mem++=*(size_t *) "NEDMALOC");
+#endif
+		return 0;
+	}
+#if USE_MAGIC_HEADERS
+	_ret=(size_t *) ret;
+	ret=(void *)(_ret+3);
+	for(; _ret<(size_t *)ret-2; _ret++) *_ret=*(size_t *) "NEDMALOC";
+	_ret[0]=(size_t) mspace;
+	_ret[1]=newsize-3*sizeof(size_t);
+#endif
+	return ret;
+}
+
+static FORCEINLINE void CallFree(void *RESTRICT mspace, void *RESTRICT mem, int isforeign) THROWSPEC
+{
+#if USE_MAGIC_HEADERS
+	mstate oldmspace=0;
+	size_t *_mem=(size_t *) mem-3, oldsize=0;
+#endif
+	if(isforeign)
+	{
+#if USE_MAGIC_HEADERS
+		assert(_mem[0]!=*(size_t *) "NEDMALOC");
+#endif
+#if defined(DEBUG)
+		printf("*** nedmalloc frees system allocated block %p\n", mem);
+#endif
+		free(mem);
+		return;
+	}
+#if USE_MAGIC_HEADERS
+	assert(_mem[0]==*(size_t *) "NEDMALOC");
+	oldmspace=(mstate) _mem[1];
+	oldsize=_mem[2];
+	for(; *_mem==*(size_t *) "NEDMALOC"; *_mem--=*(size_t *) "nedmaloc");
+	mem=(void *)(++_mem);
+#endif
+#if USE_ALLOCATOR==0
+	free(mem);
+#elif USE_ALLOCATOR==1
+	mspace_free((mstate) mspace, mem);
+#endif
+}
+
+static NEDMALLOCNOALIASATTR mstate nedblkmstate(void *RESTRICT mem) THROWSPEC
+{
+	if(mem)
+	{
+#if USE_MAGIC_HEADERS
+		size_t *_mem=(size_t *) mem-3;
+		if(_mem[0]==*(size_t *) "NEDMALOC")
+		{
+			return (mstate) _mem[1];
+		}
+		else return 0;
+#else
+#if USE_ALLOCATOR==0
+		/* Fail everything */
+		return 0;
+#elif USE_ALLOCATOR==1
+#ifdef ENABLE_FAST_HEAP_DETECTION
+#ifdef WIN32
+		/*  On Windows for RELEASE both x86 and x64 the NT heap precedes each block with an eight byte header
+			which looks like:
+				normal: 4 bytes of size, 4 bytes of [char < 64, char < 64, char < 64 bit 0 always set, char random ]
+				mmaped: 4 bytes of size  4 bytes of [zero,      zero,      0xb,                        zero        ]
+
+			On Windows for DEBUG both x86 and x64 the preceding four bytes is always 0xfdfdfdfd (no man's land).
+		*/
+#pragma pack(push, 1)
+		struct _HEAP_ENTRY
+		{
+			USHORT Size;
+			USHORT PreviousSize;
+			UCHAR Cookie;			/* SegmentIndex */
+			UCHAR Flags;			/* always bit 0 (HEAP_ENTRY_BUSY). bit 1=(HEAP_ENTRY_EXTRA_PRESENT), bit 2=normal block (HEAP_ENTRY_FILL_PATTERN), bit 3=mmap block (HEAP_ENTRY_VIRTUAL_ALLOC). Bit 4 (HEAP_ENTRY_LAST_ENTRY) could be set */
+			UCHAR UnusedBytes;
+			UCHAR SmallTagIndex;	/* fastbin index. Always one of 0x02, 0x03, 0x04 < 0x80 */
+		} *RESTRICT he=((struct _HEAP_ENTRY *) mem)-1;
+#pragma pack(pop)
+		unsigned int header=((unsigned int *)mem)[-1], mask1=0x8080E100, result1, mask2=0xFFFFFF06, result2;
+		result1=header & mask1;	/* Positive testing for NT heap */
+		result2=header & mask2;	/* Positive testing for dlmalloc */
+		if(result1==0x00000100 && result2!=0x00000102)
+		{	/* This is likely a NT heap block */
+			return 0;
+		}
+#endif
+#ifdef __linux__
+		/* On Linux glibc uses ptmalloc2 (really dlmalloc) just as we do, but prev_foot contains rubbish
+		when the preceding block is allocated because ptmalloc2 finds the local mstate by rounding the ptr
+		down to the nearest megabyte. It's like dlmalloc with FOOTERS disabled. */
+		mchunkptr p=mem2chunk(mem);
+		mstate fm=get_mstate_for(p);
+		/* If it's a ptmalloc2 block, fm is likely to be some crazy value */
+		if(!is_aligned(fm)) return 0;
+		if((size_t)mem-(size_t)fm>=(size_t)1<<(SIZE_T_BITSIZE-1)) return 0;
+		if(ok_magic(fm))
+			return fm;
+		else
+			return 0;
+		if(1) { }
+#endif
+		else
+		{
+			mchunkptr p=mem2chunk(mem);
+			mstate fm=get_mstate_for(p);
+			assert(ok_magic(fm));	/* If this fails, someone tried to free a block twice */
+			if(ok_magic(fm))
+				return fm;
+		}
+#else
+//#ifdef WIN32
+//		__try
+//#endif
+		{
+			/* We try to return zero here if it isn't one of our own blocks, however
+			the current block annotation scheme used by dlmalloc makes it impossible
+			to be absolutely sure of avoiding a segfault.
+
+			mchunkptr->prev_foot = mem-(2*size_t) = mstate ^ mparams.magic for PRECEDING block;
+			mchunkptr->head      = mem-(1*size_t) = 8 multiple size of this block with bottom three bits = FLAG_BITS
+			    FLAG_BITS = bit 0 is CINUSE (currently in use unless is mmap), bit 1 is PINUSE (previous block currently
+				            in use unless mmap), bit 2 is UNUSED and currently is always zero.
+			*/
+			register void *RESTRICT leastusedaddress_=leastusedaddress;		/* Cache these to avoid register reloading */
+			register size_t largestusedblock_=largestusedblock;
+			if(!is_aligned(mem)) return 0;		/* Would fail very rarely as all allocators return aligned blocks */
+			if(mem<leastusedaddress_) return 0;	/* Simple but effective */
+			{
+				mchunkptr p=mem2chunk(mem);
+				mstate fm=0;
+				int ismmapped=is_mmapped(p);
+				if((!ismmapped && !is_inuse(p)) || (p->head & FLAG4_BIT)) return 0;
+				/* Reduced uncertainty by 0.5^2 = 25.0% */
+				/* size should never exceed largestusedblock */
+				if(chunksize(p)>largestusedblock_) return 0;
+				/* Reduced uncertainty by a minimum of 0.5^3 = 12.5%, maximum 0.5^16 = 0.0015% */
+				/* Having sanity checked prev_foot and head, check next block */
+				if(!ismmapped && (!next_pinuse(p) || (next_chunk(p)->head & FLAG4_BIT))) return 0;
+				/* Reduced uncertainty by 0.5^5 = 3.13% or 0.5^18 = 0.00038% */
+	#if 0
+				/* If previous block is free, check that its next block pointer equals us */
+				if(!ismmapped && !pinuse(p))
+					if(next_chunk(prev_chunk(p))!=p) return 0;
+				/* We could start comparing prev_foot's for similarity but it starts getting slow. */
+	#endif
+				fm = get_mstate_for(p);
+				if(!is_aligned(fm) || (void *)fm<leastusedaddress_) return 0;
+				if((size_t)mem-(size_t)fm>=(size_t)1<<(SIZE_T_BITSIZE-1)) return 0;
+				assert(ok_magic(fm));	/* If this fails, someone tried to free a block twice */
+				if(ok_magic(fm))
+					return fm;
+			}
+		}
+//#ifdef WIN32
+//		__except(1) { }
+//#endif
+#endif
+#endif
+#endif
+	}
+	return 0;
+}
+NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC
+{
+	if(mem)
+	{
+		if(isforeign) *isforeign=1;
+#if USE_MAGIC_HEADERS
+		{
+			size_t *_mem=(size_t *) mem-3;
+			if(_mem[0]==*(size_t *) "NEDMALOC")
+			{
+				mstate mspace=(mstate) _mem[1];
+				size_t size=_mem[2];
+				if(isforeign) *isforeign=0;
+				return size;
+			}
+		}
+#elif USE_ALLOCATOR==1
+		if(nedblkmstate(mem))
+		{
+			mchunkptr p=mem2chunk(mem);
+			if(isforeign) *isforeign=0;
+			return chunksize(p)-overhead_for(p);
+		}
+#ifdef DEBUG
+		else
+		{
+			int a=1; /* Set breakpoints here if needed */
+		}
+#endif
+#endif
+#if defined(ENABLE_TOLERANT_NEDMALLOC) || USE_ALLOCATOR==0
+#ifdef _MSC_VER
+		/* This is the MSVCRT equivalent */
+		return _msize(mem);
+#elif defined(__linux__)
+		/* This is the glibc/ptmalloc2/dlmalloc equivalent.  */
+		return malloc_usable_size(mem);
+#elif defined(__FreeBSD__) || defined(__APPLE__)
+		/* This is the BSD libc equivalent.  */
+		return malloc_size(mem);
+#else
+#error Cannot tolerate the memory allocator of an unknown system!
+#endif
+#endif
+	}
+	return 0;
+}
+
+NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC											{ nedpsetvalue((nedpool *) 0, v); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC						{ return nedpmalloc((nedpool *) 0, size); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC			{ return nedpcalloc((nedpool *) 0, no, size); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC			{ return nedprealloc((nedpool *) 0, mem, size); }
+NEDMALLOCNOALIASATTR void   nedfree(void *mem) THROWSPEC											{ nedpfree((nedpool *) 0, mem); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC	{ return nedpmemalign((nedpool *) 0, alignment, bytes); }
+NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC									{ return nedpmallinfo((nedpool *) 0); }
+NEDMALLOCNOALIASATTR int    nedmallopt(int parno, int value) THROWSPEC								{ return nedpmallopt((nedpool *) 0, parno, value); }
+NEDMALLOCNOALIASATTR int    nedmalloc_trim(size_t pad) THROWSPEC									{ return nedpmalloc_trim((nedpool *) 0, pad); }
+void   nedmalloc_stats() THROWSPEC																	{ nedpmalloc_stats((nedpool *) 0); }
+NEDMALLOCNOALIASATTR size_t nedmalloc_footprint() THROWSPEC											{ return nedpmalloc_footprint((nedpool *) 0); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC	{ return nedpindependent_calloc((nedpool *) 0, elemsno, elemsize, chunks); }
+NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC		{ return nedpindependent_comalloc((nedpool *) 0, elems, sizes, chunks); }
+
+struct threadcacheblk_t;
+typedef struct threadcacheblk_t threadcacheblk;
+struct threadcacheblk_t
+{	/* Keep less than 16 bytes on 32 bit systems and 32 bytes on 64 bit systems */
+#ifdef FULLSANITYCHECKS
+	unsigned int magic;
+#endif
+	unsigned int lastUsed, size;
+	threadcacheblk *next, *prev;
+};
+typedef struct threadcache_t
+{
+#ifdef FULLSANITYCHECKS
+	unsigned int magic1;
+#endif
+	int mymspace;						/* Last mspace entry this thread used */
+	long threadid;
+	unsigned int mallocs, frees, successes;
+	size_t freeInCache;					/* How much free space is stored in this cache */
+	threadcacheblk *bins[(THREADCACHEMAXBINS+1)*2];
+#ifdef FULLSANITYCHECKS
+	unsigned int magic2;
+#endif
+} threadcache;
+struct nedpool_t
+{
+	MLOCK_T mutex;
+	void *uservalue;
+	int threads;						/* Max entries in m to use */
+	threadcache *caches[THREADCACHEMAXCACHES];
+	TLSVAR mycache;						/* Thread cache for this thread. 0 for unset, negative for use mspace-1 directly, otherwise is cache-1 */
+	mstate m[MAXTHREADSINPOOL+1];		/* mspace entries for this pool */
+};
+static nedpool syspool;
+
+static FORCEINLINE NEDMALLOCNOALIASATTR unsigned int size2binidx(size_t _size) THROWSPEC
+{	/* 8=1000	16=10000	20=10100	24=11000	32=100000	48=110000	4096=1000000000000 */
+	unsigned int topbit, size=(unsigned int)(_size>>4);
+	/* 16=1		20=1	24=1	32=10	48=11	64=100	96=110	128=1000	4096=100000000 */
+
+#if defined(__GNUC__)
+        topbit = sizeof(size)*__CHAR_BIT__ - 1 - __builtin_clz(size);
+#elif defined(_MSC_VER) && _MSC_VER>=1300
+	{
+            unsigned long bsrTopBit;
+
+            _BitScanReverse(&bsrTopBit, size);
+
+            topbit = bsrTopBit;
+        }
+#else
+#if 0
+	union {
+		unsigned asInt[2];
+		double asDouble;
+	};
+	int n;
+
+	asDouble = (double)size + 0.5;
+	topbit = (asInt[!FOX_BIGENDIAN] >> 20) - 1023;
+#else
+	{
+		unsigned int x=size;
+		x = x | (x >> 1);
+		x = x | (x >> 2);
+		x = x | (x >> 4);
+		x = x | (x >> 8);
+		x = x | (x >>16);
+		x = ~x;
+		x = x - ((x >> 1) & 0x55555555);
+		x = (x & 0x33333333) + ((x >> 2) & 0x33333333);
+		x = (x + (x >> 4)) & 0x0F0F0F0F;
+		x = x + (x << 8);
+		x = x + (x << 16);
+		topbit=31 - (x >> 24);
+	}
+#endif
+#endif
+	return topbit;
+}
+
+
+#ifdef FULLSANITYCHECKS
+static void tcsanitycheck(threadcacheblk **ptr) THROWSPEC
+{
+	assert((ptr[0] && ptr[1]) || (!ptr[0] && !ptr[1]));
+	if(ptr[0] && ptr[1])
+	{
+		assert(nedblksize(ptr[0])>=sizeof(threadcacheblk));
+		assert(nedblksize(ptr[1])>=sizeof(threadcacheblk));
+		assert(*(unsigned int *) "NEDN"==ptr[0]->magic);
+		assert(*(unsigned int *) "NEDN"==ptr[1]->magic);
+		assert(!ptr[0]->prev);
+		assert(!ptr[1]->next);
+		if(ptr[0]==ptr[1])
+		{
+			assert(!ptr[0]->next);
+			assert(!ptr[1]->prev);
+		}
+	}
+}
+static void tcfullsanitycheck(threadcache *tc) THROWSPEC
+{
+	threadcacheblk **tcbptr=tc->bins;
+	int n;
+	for(n=0; n<=THREADCACHEMAXBINS; n++, tcbptr+=2)
+	{
+		threadcacheblk *b, *ob=0;
+		tcsanitycheck(tcbptr);
+		for(b=tcbptr[0]; b; ob=b, b=b->next)
+		{
+			assert(*(unsigned int *) "NEDN"==b->magic);
+			assert(!ob || ob->next==b);
+			assert(!ob || b->prev==ob);
+		}
+	}
+}
+#endif
+
+static NOINLINE void RemoveCacheEntries(nedpool *RESTRICT p, threadcache *RESTRICT tc, unsigned int age) THROWSPEC
+{
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+	if(tc->freeInCache)
+	{
+		threadcacheblk **tcbptr=tc->bins;
+		int n;
+		for(n=0; n<=THREADCACHEMAXBINS; n++, tcbptr+=2)
+		{
+			threadcacheblk **tcb=tcbptr+1;		/* come from oldest end of list */
+			/*tcsanitycheck(tcbptr);*/
+			for(; *tcb && tc->frees-(*tcb)->lastUsed>=age; )
+			{
+				threadcacheblk *f=*tcb;
+				size_t blksize=f->size; /*nedblksize(f);*/
+				assert(blksize<=nedblksize(0, f));
+				assert(blksize);
+#ifdef FULLSANITYCHECKS
+				assert(*(unsigned int *) "NEDN"==(*tcb)->magic);
+#endif
+				*tcb=(*tcb)->prev;
+				if(*tcb)
+					(*tcb)->next=0;
+				else
+					*tcbptr=0;
+				tc->freeInCache-=blksize;
+				assert((long) tc->freeInCache>=0);
+				CallFree(0, f, 0);
+				/*tcsanitycheck(tcbptr);*/
+			}
+		}
+	}
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+}
+static void DestroyCaches(nedpool *RESTRICT p) THROWSPEC
+{
+	if(p->caches)
+	{
+		threadcache *tc;
+		int n;
+		for(n=0; n<THREADCACHEMAXCACHES; n++)
+		{
+			if((tc=p->caches[n]))
+			{
+				tc->frees++;
+				RemoveCacheEntries(p, tc, 0);
+				assert(!tc->freeInCache);
+				tc->mymspace=-1;
+				tc->threadid=0;
+				CallFree(0, tc, 0);
+				p->caches[n]=0;
+			}
+		}
+	}
+}
+
+static NOINLINE threadcache *AllocCache(nedpool *RESTRICT p) THROWSPEC
+{
+	threadcache *tc=0;
+	int n, end;
+	ACQUIRE_LOCK(&p->mutex);
+	for(n=0; n<THREADCACHEMAXCACHES && p->caches[n]; n++);
+	if(THREADCACHEMAXCACHES==n)
+	{	/* List exhausted, so disable for this thread */
+		RELEASE_LOCK(&p->mutex);
+		return 0;
+	}
+	tc=p->caches[n]=(threadcache *) CallCalloc(p->m[0], sizeof(threadcache), 0);
+	if(!tc)
+	{
+		RELEASE_LOCK(&p->mutex);
+		return 0;
+	}
+#ifdef FULLSANITYCHECKS
+	tc->magic1=*(unsigned int *)"NEDMALC1";
+	tc->magic2=*(unsigned int *)"NEDMALC2";
+#endif
+	tc->threadid=(long)(size_t)CURRENT_THREAD;
+	for(end=0; p->m[end]; end++);
+	tc->mymspace=abs(tc->threadid) % end;
+	RELEASE_LOCK(&p->mutex);
+	if(TLSSET(p->mycache, (void *)(size_t)(n+1))) abort();
+	return tc;
+}
+
+static void *threadcache_malloc(nedpool *RESTRICT p, threadcache *RESTRICT tc, size_t *RESTRICT _size) THROWSPEC
+{
+	void *RESTRICT ret=0;
+	size_t size=*_size, blksize=0;
+	unsigned int bestsize;
+	unsigned int idx=size2binidx(size);
+	threadcacheblk *RESTRICT blk, **RESTRICT binsptr;
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+	/* Calculate best fit bin size */
+	bestsize=1<<(idx+4);
+#if 0
+	/* Finer grained bin fit */
+	idx<<=1;
+	if(size>bestsize)
+	{
+		idx++;
+		bestsize+=bestsize>>1;
+	}
+	if(size>bestsize)
+	{
+		idx++;
+		bestsize=1<<(4+(idx>>1));
+	}
+#else
+	if(size>bestsize)
+	{
+		idx++;
+		bestsize<<=1;
+	}
+#endif
+	assert(bestsize>=size);
+	if(size<bestsize) size=bestsize;
+	assert(size<=THREADCACHEMAX);
+	assert(idx<=THREADCACHEMAXBINS);
+	binsptr=&tc->bins[idx*2];
+	/* Try to match close, but move up a bin if necessary */
+	blk=*binsptr;
+	if(!blk || blk->size<size)
+	{	/* Bump it up a bin */
+		if(idx<THREADCACHEMAXBINS)
+		{
+			idx++;
+			binsptr+=2;
+			blk=*binsptr;
+		}
+	}
+	if(blk)
+	{
+		blksize=blk->size; /*nedblksize(blk);*/
+		assert(nedblksize(0, blk)>=blksize);
+		assert(blksize>=size);
+		if(blk->next)
+			blk->next->prev=0;
+		*binsptr=blk->next;
+		if(!*binsptr)
+			binsptr[1]=0;
+#ifdef FULLSANITYCHECKS
+		blk->magic=0;
+#endif
+		assert(binsptr[0]!=blk && binsptr[1]!=blk);
+		assert(nedblksize(0, blk)>=sizeof(threadcacheblk) && nedblksize(0, blk)<=THREADCACHEMAX+CHUNK_OVERHEAD);
+		/*printf("malloc: %p, %p, %p, %lu\n", p, tc, blk, (long) _size);*/
+		ret=(void *) blk;
+	}
+	++tc->mallocs;
+	if(ret)
+	{
+		assert(blksize>=size);
+		++tc->successes;
+		tc->freeInCache-=blksize;
+		assert((long) tc->freeInCache>=0);
+	}
+#if defined(DEBUG) && 0
+	if(!(tc->mallocs & 0xfff))
+	{
+		printf("*** threadcache=%u, mallocs=%u (%f), free=%u (%f), freeInCache=%u\n", (unsigned int) tc->threadid, tc->mallocs,
+			(float) tc->successes/tc->mallocs, tc->frees, (float) tc->successes/tc->frees, (unsigned int) tc->freeInCache);
+	}
+#endif
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+	*_size=size;
+	return ret;
+}
+static NOINLINE void ReleaseFreeInCache(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace) THROWSPEC
+{
+	unsigned int age=THREADCACHEMAXFREESPACE/8192;
+	/*ACQUIRE_LOCK(&p->m[mymspace]->mutex);*/
+	while(age && tc->freeInCache>=THREADCACHEMAXFREESPACE)
+	{
+		RemoveCacheEntries(p, tc, age);
+		/*printf("*** Removing cache entries older than %u (%u)\n", age, (unsigned int) tc->freeInCache);*/
+		age>>=1;
+	}
+	/*RELEASE_LOCK(&p->m[mymspace]->mutex);*/
+}
+static void threadcache_free(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace, void *RESTRICT mem, size_t size) THROWSPEC
+{
+	unsigned int bestsize;
+	unsigned int idx=size2binidx(size);
+	threadcacheblk **RESTRICT binsptr, *RESTRICT tck=(threadcacheblk *) mem;
+	assert(size>=sizeof(threadcacheblk) && size<=THREADCACHEMAX+CHUNK_OVERHEAD);
+#ifdef DEBUG
+	/* Make sure this is a valid memory block */
+	assert(nedblksize(0, mem));
+#endif
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+	/* Calculate best fit bin size */
+	bestsize=1<<(idx+4);
+#if 0
+	/* Finer grained bin fit */
+	idx<<=1;
+	if(size>bestsize)
+	{
+		unsigned int biggerbestsize=bestsize+bestsize<<1;
+		if(size>=biggerbestsize)
+		{
+			idx++;
+			bestsize=biggerbestsize;
+		}
+	}
+#endif
+	if(bestsize!=size)	/* dlmalloc can round up, so we round down to preserve indexing */
+		size=bestsize;
+	binsptr=&tc->bins[idx*2];
+	assert(idx<=THREADCACHEMAXBINS);
+	if(tck==*binsptr)
+	{
+		fprintf(stderr, "nedmalloc: Attempt to free already freed memory block %p - aborting!\n", tck);
+		abort();
+	}
+#ifdef FULLSANITYCHECKS
+	tck->magic=*(unsigned int *) "NEDN";
+#endif
+	tck->lastUsed=++tc->frees;
+	tck->size=(unsigned int) size;
+	tck->next=*binsptr;
+	tck->prev=0;
+	if(tck->next)
+		tck->next->prev=tck;
+	else
+		binsptr[1]=tck;
+	assert(!*binsptr || (*binsptr)->size==tck->size);
+	*binsptr=tck;
+	assert(tck==tc->bins[idx*2]);
+	assert(tc->bins[idx*2+1]==tck || binsptr[0]->next->prev==tck);
+	/*printf("free: %p, %p, %p, %lu\n", p, tc, mem, (long) size);*/
+	tc->freeInCache+=size;
+#ifdef FULLSANITYCHECKS
+	tcfullsanitycheck(tc);
+#endif
+#if 1
+	if(tc->freeInCache>=THREADCACHEMAXFREESPACE)
+		ReleaseFreeInCache(p, tc, mymspace);
+#endif
+}
+
+
+
+
+static NOINLINE int InitPool(nedpool *RESTRICT p, size_t capacity, int threads) THROWSPEC
+{	/* threads is -1 for system pool */
+	ensure_initialization();
+	ACQUIRE_MALLOC_GLOBAL_LOCK();
+	if(p->threads) goto done;
+	if(INITIAL_LOCK(&p->mutex)) goto err;
+	if(TLSALLOC(&p->mycache)) goto err;
+#if USE_ALLOCATOR==0
+	p->m[0]=(mstate) mspacecounter++;
+#elif USE_ALLOCATOR==1
+	if(!(p->m[0]=(mstate) create_mspace(capacity, 1))) goto err;
+	p->m[0]->extp=p;
+#endif
+	p->threads=(threads<1 || threads>MAXTHREADSINPOOL) ? MAXTHREADSINPOOL : threads;
+done:
+	RELEASE_MALLOC_GLOBAL_LOCK();
+	return 1;
+err:
+	if(threads<0)
+		abort();			/* If you can't allocate for system pool, we're screwed */
+	DestroyCaches(p);
+	if(p->m[0])
+	{
+#if USE_ALLOCATOR==1
+		destroy_mspace(p->m[0]);
+#endif
+		p->m[0]=0;
+	}
+	if(p->mycache)
+	{
+		if(TLSFREE(p->mycache)) abort();
+		p->mycache=0;
+	}
+	RELEASE_MALLOC_GLOBAL_LOCK();
+	return 0;
+}
+static NOINLINE mstate FindMSpace(nedpool *RESTRICT p, threadcache *RESTRICT tc, int *RESTRICT lastUsed, size_t size) THROWSPEC
+{	/* Gets called when thread's last used mspace is in use. The strategy
+	is to run through the list of all available mspaces looking for an
+	unlocked one and if we fail, we create a new one so long as we don't
+	exceed p->threads */
+	int n, end;
+	for(n=end=*lastUsed+1; p->m[n]; end=++n)
+	{
+		if(TRY_LOCK(&p->m[n]->mutex)) goto found;
+	}
+	for(n=0; n<*lastUsed && p->m[n]; n++)
+	{
+		if(TRY_LOCK(&p->m[n]->mutex)) goto found;
+	}
+	if(end<p->threads)
+	{
+		mstate temp;
+#if USE_ALLOCATOR==0
+		temp=(mstate) mspacecounter++;
+#elif USE_ALLOCATOR==1
+		if(!(temp=(mstate) create_mspace(size, 1)))
+			goto badexit;
+#endif
+		/* Now we're ready to modify the lists, we lock */
+		ACQUIRE_LOCK(&p->mutex);
+		while(p->m[end] && end<p->threads)
+			end++;
+		if(end>=p->threads)
+		{	/* Drat, must destroy it now */
+			RELEASE_LOCK(&p->mutex);
+#if USE_ALLOCATOR==1
+			destroy_mspace((mstate) temp);
+#endif
+			goto badexit;
+		}
+		/* We really want to make sure this goes into memory now but we
+		have to be careful of breaking aliasing rules, so write it twice */
+		*((volatile struct malloc_state **) &p->m[end])=p->m[end]=temp;
+		ACQUIRE_LOCK(&p->m[end]->mutex);
+		/*printf("Created mspace idx %d\n", end);*/
+		RELEASE_LOCK(&p->mutex);
+		n=end;
+		goto found;
+	}
+	/* Let it lock on the last one it used */
+badexit:
+	ACQUIRE_LOCK(&p->m[*lastUsed]->mutex);
+	return p->m[*lastUsed];
+found:
+	*lastUsed=n;
+	if(tc)
+		tc->mymspace=n;
+	else
+	{
+		if(TLSSET(p->mycache, (void *)(size_t)(-(n+1)))) abort();
+	}
+	return p->m[n];
+}
+
+typedef struct PoolList_t
+{
+	size_t size;			/* Size of list */
+	size_t length;			/* Actual entries in list */
+#ifdef DEBUG
+	nedpool *list[1];		/* Force testing of list expansion */
+#else
+	nedpool *list[16];
+#endif
+} PoolList;
+static MLOCK_T poollistlock;
+static PoolList *poollist;
+NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC
+{
+	nedpool *ret=0;
+	if(!poollist)
+	{
+		PoolList *newpoollist=0;
+		if(!(newpoollist=(PoolList *) nedpcalloc(0, 1, sizeof(PoolList)+sizeof(nedpool *)))) return 0;
+		INITIAL_LOCK(&poollistlock);
+		ACQUIRE_LOCK(&poollistlock);
+		poollist=newpoollist;
+		poollist->size=sizeof(poollist->list)/sizeof(nedpool *);
+	}
+	else
+		ACQUIRE_LOCK(&poollistlock);
+	if(poollist->length==poollist->size)
+	{
+		PoolList *newpoollist=0;
+		size_t newsize=0;
+		newsize=sizeof(PoolList)+(poollist->size+1)*sizeof(nedpool *);
+		if(!(newpoollist=(PoolList *) nedprealloc(0, poollist, newsize))) goto badexit;
+		poollist=newpoollist;
+		memset(&poollist->list[poollist->size], 0, newsize-((size_t)&poollist->list[poollist->size]-(size_t)&poollist->list[0]));
+		poollist->size=((newsize-((char *)&poollist->list[0]-(char *)poollist))/sizeof(nedpool *))-1;
+		assert(poollist->size>poollist->length);
+	}
+	if(!(ret=(nedpool *) nedpcalloc(0, 1, sizeof(nedpool)))) goto badexit;
+	if(!InitPool(ret, capacity, threads))
+	{
+		nedpfree(0, ret);
+		goto badexit;
+	}
+	poollist->list[poollist->length++]=ret;
+badexit:
+	RELEASE_LOCK(&poollistlock);
+	return ret;
+}
+void neddestroypool(nedpool *p) THROWSPEC
+{
+	unsigned int n;
+	ACQUIRE_LOCK(&p->mutex);
+	DestroyCaches(p);
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1
+		destroy_mspace(p->m[n]);
+#endif
+		p->m[n]=0;
+	}
+	RELEASE_LOCK(&p->mutex);
+	if(TLSFREE(p->mycache)) abort();
+	nedpfree(0, p);
+	ACQUIRE_LOCK(&poollistlock);
+	assert(poollist);
+	for(n=0; n<poollist->length && poollist->list[n]!=p; n++);
+	assert(n!=poollist->length);
+	memmove(&poollist->list[n], &poollist->list[n+1], (size_t)&poollist->list[poollist->length]-(size_t)&poollist->list[n]);
+	if(!--poollist->length)
+	{
+		assert(!poollist->list[0]);
+		nedpfree(0, poollist);
+		poollist=0;
+	}
+	RELEASE_LOCK(&poollistlock);
+}
+void neddestroysyspool() THROWSPEC
+{
+	nedpool *p=&syspool;
+	int n;
+	ACQUIRE_LOCK(&p->mutex);
+	DestroyCaches(p);
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1
+		destroy_mspace(p->m[n]);
+#endif
+		p->m[n]=0;
+	}
+	/* Render syspool unusable */
+	for(n=0; n<THREADCACHEMAXCACHES; n++)
+		p->caches[n]=(threadcache *)(size_t)(sizeof(size_t)>4 ? 0xdeadbeefdeadbeefULL : 0xdeadbeefUL);
+	for(n=0; n<MAXTHREADSINPOOL+1; n++)
+		p->m[n]=(mstate)(size_t)(sizeof(size_t)>4 ? 0xdeadbeefdeadbeefULL : 0xdeadbeefUL);
+	if(TLSFREE(p->mycache)) abort();
+	RELEASE_LOCK(&p->mutex);
+}
+nedpool **nedpoollist() THROWSPEC
+{
+	nedpool **ret=0;
+	if(poollist)
+	{
+		ACQUIRE_LOCK(&poollistlock);
+		if(!(ret=(nedpool **) nedmalloc((poollist->length+1)*sizeof(nedpool *)))) goto badexit;
+		memcpy(ret, poollist->list, (poollist->length+1)*sizeof(nedpool *));
+badexit:
+		RELEASE_LOCK(&poollistlock);
+	}
+	return ret;
+}
+
+void nedpsetvalue(nedpool *p, void *v) THROWSPEC
+{
+	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }
+	p->uservalue=v;
+}
+void *nedgetvalue(nedpool **p, void *mem) THROWSPEC
+{
+	nedpool *np=0;
+	mstate fm=nedblkmstate(mem);
+	if(!fm || !fm->extp) return 0;
+	np=(nedpool *) fm->extp;
+	if(p) *p=np;
+	return np->uservalue;
+}
+
+void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC
+{
+	int mycache;
+	if(!p)
+	{
+		p=&syspool;
+		if(!syspool.threads) InitPool(&syspool, 0, -1);
+	}
+	mycache=(int)(size_t) TLSGET(p->mycache);
+	if(!mycache)
+	{	/* Set to mspace 0 */
+		if(disable && TLSSET(p->mycache, (void *)(size_t)-1)) abort();
+	}
+	else if(mycache>0)
+	{	/* Set to last used mspace */
+		threadcache *tc=p->caches[mycache-1];
+#if defined(DEBUG)
+		printf("Threadcache utilisation: %lf%% in cache with %lf%% lost to other threads\n",
+			100.0*tc->successes/tc->mallocs, 100.0*((double) tc->mallocs-tc->frees)/tc->mallocs);
+#endif
+		if(disable && TLSSET(p->mycache, (void *)(size_t)(-tc->mymspace))) abort();
+		tc->frees++;
+		RemoveCacheEntries(p, tc, 0);
+		assert(!tc->freeInCache);
+		if(disable)
+		{
+			tc->mymspace=-1;
+			tc->threadid=0;
+			CallFree(0, p->caches[mycache-1], 0);
+			p->caches[mycache-1]=0;
+		}
+	}
+}
+void neddisablethreadcache(nedpool *p) THROWSPEC
+{
+	nedtrimthreadcache(p, 1);
+}
+
+#define GETMSPACE(m,p,tc,ms,s,action)                 \
+  do                                                  \
+  {                                                   \
+    mstate m = GetMSpace((p),(tc),(ms),(s));          \
+    action;                                           \
+	if(USE_ALLOCATOR==1) { RELEASE_LOCK(&m->mutex); } \
+  } while (0)
+
+static FORCEINLINE mstate GetMSpace(nedpool *RESTRICT p, threadcache *RESTRICT tc, int mymspace, size_t size) THROWSPEC
+{	/* Returns a locked and ready for use mspace */
+	mstate m=p->m[mymspace];
+	assert(m);
+#if USE_ALLOCATOR==1
+	if(!TRY_LOCK(&p->m[mymspace]->mutex)) m=FindMSpace(p, tc, &mymspace, size);
+	/*assert(IS_LOCKED(&p->m[mymspace]->mutex));*/
+#endif
+	return m;
+}
+static NOINLINE void GetThreadCache_cold1(nedpool *RESTRICT *RESTRICT p) THROWSPEC
+{
+	*p=&syspool;
+	if(!syspool.threads) InitPool(&syspool, 0, -1);
+}
+static NOINLINE void GetThreadCache_cold2(nedpool *RESTRICT *RESTRICT p, threadcache *RESTRICT *RESTRICT tc, int *RESTRICT mymspace, int mycache) THROWSPEC
+{
+	if(!mycache)
+	{	/* Need to allocate a new cache */
+		*tc=AllocCache(*p);
+		if(!*tc)
+		{	/* Disable */
+			if(TLSSET((*p)->mycache, (void *)(size_t)-1)) abort();
+			*mymspace=0;
+		}
+		else
+			*mymspace=(*tc)->mymspace;
+	}
+	else
+	{	/* Cache disabled, but we do have an assigned thread pool */
+		*tc=0;
+		*mymspace=-mycache-1;
+	}
+}
+static FORCEINLINE void GetThreadCache(nedpool *RESTRICT *RESTRICT p, threadcache *RESTRICT *RESTRICT tc, int *RESTRICT mymspace, size_t *RESTRICT size) THROWSPEC
+{
+	int mycache;
+	if(size && *size<sizeof(threadcacheblk)) *size=sizeof(threadcacheblk);
+	if(!*p)
+		GetThreadCache_cold1(p);
+	mycache=(int)(size_t) TLSGET((*p)->mycache);
+	if(mycache>0)
+	{	/* Already have a cache */
+		*tc=(*p)->caches[mycache-1];
+		*mymspace=(*tc)->mymspace;
+	}
+	else GetThreadCache_cold2(p, tc, mymspace, mycache);
+	assert(*mymspace>=0);
+	assert(!(*tc) || (long)(size_t)CURRENT_THREAD==(*tc)->threadid);
+#ifdef FULLSANITYCHECKS
+	if(*tc)
+	{
+		if(*(unsigned int *)"NEDMALC1"!=(*tc)->magic1 || *(unsigned int *)"NEDMALC2"!=(*tc)->magic2)
+		{
+			abort();
+		}
+	}
+#endif
+}
+
+NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC
+{
+	void *ret=0;
+	threadcache *tc;
+	int mymspace;
+	GetThreadCache(&p, &tc, &mymspace, &size);
+#if THREADCACHEMAX
+	if(tc && size<=THREADCACHEMAX)
+	{	/* Use the thread cache */
+		ret=threadcache_malloc(p, tc, &size);
+	}
+#endif
+	if(!ret)
+	{	/* Use this thread's mspace */
+        GETMSPACE(m, p, tc, mymspace, size,
+                  ret=CallMalloc(m, size, 0));
+	}
+	return ret;
+}
+NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC
+{
+	size_t rsize=size*no;
+	void *ret=0;
+	threadcache *tc;
+	int mymspace;
+	GetThreadCache(&p, &tc, &mymspace, &rsize);
+#if THREADCACHEMAX
+	if(tc && rsize<=THREADCACHEMAX)
+	{	/* Use the thread cache */
+		if((ret=threadcache_malloc(p, tc, &rsize)))
+			memset(ret, 0, rsize);
+	}
+#endif
+	if(!ret)
+	{	/* Use this thread's mspace */
+        GETMSPACE(m, p, tc, mymspace, rsize,
+                  ret=CallCalloc(m, rsize, 0));
+	}
+	return ret;
+}
+NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC
+{
+	void *ret=0;
+	threadcache *tc;
+	int mymspace, isforeign=1;
+	size_t memsize;
+	if(!mem) return nedpmalloc(p, size);
+	memsize=nedblksize(&isforeign, mem);
+	assert(memsize);
+	if(!memsize)
+	{
+		fprintf(stderr, "nedmalloc: nedprealloc() called with a block not created by nedmalloc!\n");
+		abort();
+	}
+	else if(size<=memsize && memsize-size<
+#ifdef DEBUG
+		32
+#else
+		1024
+#endif
+		)		/* If realloc size is within 1Kb smaller than existing, noop it */
+		return mem;
+	GetThreadCache(&p, &tc, &mymspace, &size);
+#if THREADCACHEMAX
+	if(tc && size && size<=THREADCACHEMAX)
+	{	/* Use the thread cache */
+		if((ret=threadcache_malloc(p, tc, &size)))
+		{
+			memcpy(ret, mem, memsize<size ? memsize : size);
+			if(memsize>=sizeof(threadcacheblk) && memsize<=(THREADCACHEMAX+CHUNK_OVERHEAD))
+				threadcache_free(p, tc, mymspace, mem, memsize);
+			else
+				CallFree(0, mem, isforeign);
+		}
+	}
+#endif
+	if(!ret)
+	{	/* Reallocs always happen in the mspace they happened in, so skip
+		locking the preferred mspace for this thread */
+		ret=CallRealloc(p->m[mymspace], mem, isforeign, memsize, size);
+	}
+	return ret;
+}
+void   nedpfree(nedpool *p, void *mem) THROWSPEC
+{	/* Frees always happen in the mspace they happened in, so skip
+	locking the preferred mspace for this thread */
+	threadcache *tc;
+	int mymspace, isforeign=1;
+	size_t memsize;
+	if(!mem)
+	{	/* If you tried this on FreeBSD you'd be sorry! */
+#ifdef DEBUG
+		fprintf(stderr, "nedmalloc: WARNING nedpfree() called with zero. This is not portable behaviour!\n");
+#endif
+		return;
+	}
+	memsize=nedblksize(&isforeign, mem);
+	assert(memsize);
+	if(!memsize)
+	{
+		fprintf(stderr, "nedmalloc: nedpfree() called with a block not created by nedmalloc!\n");
+		abort();
+	}
+	GetThreadCache(&p, &tc, &mymspace, 0);
+#if THREADCACHEMAX
+	if(mem && tc && memsize>=sizeof(threadcacheblk) && memsize<=(THREADCACHEMAX+CHUNK_OVERHEAD))
+		threadcache_free(p, tc, mymspace, mem, memsize);
+	else
+#endif
+		CallFree(0, mem, isforeign);
+}
+NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC
+{
+	void *ret;
+	threadcache *tc;
+	int mymspace;
+	GetThreadCache(&p, &tc, &mymspace, &bytes);
+	{	/* Use this thread's mspace */
+        GETMSPACE(m, p, tc, mymspace, bytes,
+                  ret=CallMalloc(m, bytes, alignment));
+	}
+	return ret;
+}
+struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC
+{
+	int n;
+	struct nedmallinfo ret={0};
+	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1 && !NO_MALLINFO
+		struct mallinfo t=mspace_mallinfo(p->m[n]);
+		ret.arena+=t.arena;
+		ret.ordblks+=t.ordblks;
+		ret.hblkhd+=t.hblkhd;
+		ret.usmblks+=t.usmblks;
+		ret.uordblks+=t.uordblks;
+		ret.fordblks+=t.fordblks;
+		ret.keepcost+=t.keepcost;
+#endif
+	}
+	return ret;
+}
+int    nedpmallopt(nedpool *p, int parno, int value) THROWSPEC
+{
+#if USE_ALLOCATOR==1
+	return mspace_mallopt(parno, value);
+#else
+	return 0;
+#endif
+}
+NEDMALLOCNOALIASATTR void*  nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC
+{
+#if USE_ALLOCATOR==1
+	if(granularity) *granularity=mparams.granularity;
+	if(magic) *magic=mparams.magic;
+	return (void *) &syspool;
+#else
+	if(granularity) *granularity=0;
+	if(magic) *magic=0;
+	return 0;
+#endif
+}
+int    nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC
+{
+	int n, ret=0;
+	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1
+		ret+=mspace_trim(p->m[n], pad);
+#endif
+	}
+	return ret;
+}
+void   nedpmalloc_stats(nedpool *p) THROWSPEC
+{
+	int n;
+	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1
+		mspace_malloc_stats(p->m[n]);
+#endif
+	}
+}
+size_t nedpmalloc_footprint(nedpool *p) THROWSPEC
+{
+	size_t ret=0;
+	int n;
+	if(!p) { p=&syspool; if(!syspool.threads) InitPool(&syspool, 0, -1); }
+	for(n=0; p->m[n]; n++)
+	{
+#if USE_ALLOCATOR==1
+		ret+=mspace_footprint(p->m[n]);
+#endif
+	}
+	return ret;
+}
+NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC
+{
+	void **ret;
+	threadcache *tc;
+	int mymspace;
+	GetThreadCache(&p, &tc, &mymspace, &elemsize);
+#if USE_ALLOCATOR==0
+    GETMSPACE(m, p, tc, mymspace, elemsno*elemsize,
+              ret=unsupported_operation("independent_calloc"));
+#elif USE_ALLOCATOR==1
+    GETMSPACE(m, p, tc, mymspace, elemsno*elemsize,
+              ret=mspace_independent_calloc(m, elemsno, elemsize, chunks));
+#endif
+	return ret;
+}
+NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC
+{
+	void **ret;
+	threadcache *tc;
+	int mymspace;
+    size_t i, *adjustedsizes=(size_t *) alloca(elems*sizeof(size_t));
+    if(!adjustedsizes) return 0;
+    for(i=0; i<elems; i++)
+        adjustedsizes[i]=sizes[i]<sizeof(threadcacheblk) ? sizeof(threadcacheblk) : sizes[i];
+	GetThreadCache(&p, &tc, &mymspace, 0);
+#if USE_ALLOCATOR==0
+	GETMSPACE(m, p, tc, mymspace, 0,
+              ret=unsupported_operation("independent_comalloc"));
+#elif USE_ALLOCATOR==1
+	GETMSPACE(m, p, tc, mymspace, 0,
+              ret=mspace_independent_comalloc(m, elems, adjustedsizes, chunks));
+#endif
+	return ret;
+}
+
+#if defined(__cplusplus)
+}
+#endif
+
+#ifdef _MSC_VER
+#pragma warning(pop)
+#endif
+
+#endif
diff --git a/drivers/nedmalloc/nedmalloc.h b/drivers/nedmalloc/nedmalloc.h
index b9add1683a..7ec65849fc 100644
--- a/drivers/nedmalloc/nedmalloc.h
+++ b/drivers/nedmalloc/nedmalloc.h
@@ -1,302 +1,302 @@
-#ifdef NEDMALLOC_ENABLED

-

-/* nedalloc, an alternative malloc implementation for multiple threads without

-lock contention based on dlmalloc v2.8.3. (C) 2005-2009 Niall Douglas

-

-Boost Software License - Version 1.0 - August 17th, 2003

-

-Permission is hereby granted, free of charge, to any person or organization

-obtaining a copy of the software and accompanying documentation covered by

-this license (the "Software") to use, reproduce, display, distribute,

-execute, and transmit the Software, and to prepare derivative works of the

-Software, and to permit third-parties to whom the Software is furnished to

-do so, all subject to the following:

-

-The copyright notices in the Software and this entire statement, including

-the above license grant, this restriction and the following disclaimer,

-must be included in all copies of the Software, in whole or in part, and

-all derivative works of the Software, unless such copies or derivative

-works are solely in the form of machine-executable object code generated by

-a source language processor.

-

-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR

-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

-FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT

-SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE

-FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,

-ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER

-DEALINGS IN THE SOFTWARE.

-*/

-

-#ifndef NEDMALLOC_H

-#define NEDMALLOC_H

-

-#include "typedefs.h"

-#define MALLOC_ALIGNMENT DEFAULT_ALIGNMENT

-

-#ifdef PSP_ENABLED

-#define USE_LOCKS 0

-#define HAVE_MMAP 0

-#endif

-

-/* See malloc.c.h for what each function does.

-

-REPLACE_SYSTEM_ALLOCATOR on POSIX causes nedalloc's functions to be called

-malloc, free etc. instead of nedmalloc, nedfree etc. You may or may not want

-this. On Windows it causes nedmalloc to patch all loaded DLLs and binaries

-to replace usage of the system allocator.

-

-NO_NED_NAMESPACE prevents the functions from being defined in the nedalloc

-namespace when in C++ (uses the global namespace instead).

-

-NEDMALLOCEXTSPEC can be defined to be __declspec(dllexport) or

-__attribute__ ((visibility("default"))) or whatever you like. It defaults

-to extern unless NEDMALLOC_DLL_EXPORTS is set as it would be when building

-nedmalloc.dll.

-

-USE_LOCKS can be 2 if you want to define your own MLOCK_T, INITIAL_LOCK,

-ACQUIRE_LOCK, RELEASE_LOCK, TRY_LOCK, IS_LOCKED and NULL_LOCK_INITIALIZER.

-

-NEDMALLOC_DEBUG can be defined to cause DEBUG to be set differently for nedmalloc

-than for the rest of the build. Remember to set NDEBUG to disable all assertion

-checking too.

-

-USE_MAGIC_HEADERS causes nedalloc to allocate an extra three sizeof(size_t)

-to each block. nedpfree() and nedprealloc() can then automagically know when

-to free a system allocated block. Enabling this typically adds 20-50% to

-application memory usage.

-

-ENABLE_TOLERANT_NEDMALLOC is automatically turned on if REPLACE_SYSTEM_ALLOCATOR

-is set or the Windows DLL is being built. This causes nedmalloc to detect when a

-system allocator block is passed to it and to handle it appropriately. Note that

-without USE_MAGIC_HEADERS there is a very tiny chance that nedmalloc will segfault

-on non-Windows builds (it uses Win32 SEH to trap segfaults on Windows and there

-is no comparable system on POSIX).

-

-USE_ALLOCATOR can be one of these settings (it defaults to 1):

-  0: System allocator (nedmalloc now simply acts as a threadcache).

-     WARNING: Intended for DEBUG USE ONLY - not all functions work correctly.

-  1: dlmalloc

-

-ENABLE_LARGE_PAGES enables support for requesting memory from the system in large

-(typically >=2Mb) pages if the host OS supports this. These occupy just a single

-TLB entry and can significantly improve performance in large working set applications.

-

-ENABLE_FAST_HEAP_DETECTION enables special logic to detect blocks allocated

-by the system heap. This avoids 1.5%-2% overhead when checking for non-nedmalloc

-blocks, but it assumes that the NT and glibc heaps function in a very specific

-fashion which may not hold true across OS upgrades.

-*/

-

-#include <stddef.h>   /* for size_t */

-

-#ifndef NEDMALLOCEXTSPEC

- #ifdef NEDMALLOC_DLL_EXPORTS

-  #ifdef WIN32

-   #define NEDMALLOCEXTSPEC extern __declspec(dllexport)

-  #elif defined(__GNUC__)

-   #define NEDMALLOCEXTSPEC extern __attribute__ ((visibility("default")))

-  #endif

-  #ifndef ENABLE_TOLERANT_NEDMALLOC

-   #define ENABLE_TOLERANT_NEDMALLOC 1

-  #endif

- #else

-  #define NEDMALLOCEXTSPEC extern

- #endif

-#endif

-

-#if __STDC_VERSION__ >= 199901L		/* C99 or better */

- #define RESTRICT restrict

-#else

- #if defined(_MSC_VER) && _MSC_VER>=1400

-  #define RESTRICT __restrict

- #endif

- #ifdef __GNUC__

-  #define RESTRICT __restrict

- #endif

-#endif

-#ifndef RESTRICT

- #define RESTRICT

-#endif

-

-#if defined(_MSC_VER) && _MSC_VER>=1400

- #define NEDMALLOCPTRATTR __declspec(restrict)

- #define NEDMALLOCNOALIASATTR __declspec(noalias)

-#endif

-#ifdef __GNUC__

- #define NEDMALLOCPTRATTR __attribute__ ((malloc))

-#endif

-#ifndef NEDMALLOCPTRATTR

- #define NEDMALLOCPTRATTR

-#endif

-#ifndef NEDMALLOCNOALIASATTR

- #define NEDMALLOCNOALIASATTR

-#endif

-

-#ifndef USE_MAGIC_HEADERS

- #define USE_MAGIC_HEADERS 0

-#endif

-

-#ifndef USE_ALLOCATOR

- #define USE_ALLOCATOR 1 /* dlmalloc */

-#endif

-

-#if !USE_ALLOCATOR && !USE_MAGIC_HEADERS

-#error If you are using the system allocator then you MUST use magic headers

-#endif

-

-#ifdef REPLACE_SYSTEM_ALLOCATOR

- #if USE_ALLOCATOR==0

-  #error Cannot combine using the system allocator with replacing the system allocator

- #endif

- #ifndef ENABLE_TOLERANT_NEDMALLOC

-  #define ENABLE_TOLERANT_NEDMALLOC 1

- #endif

- #ifndef WIN32	/* We have a dedicated patcher for Windows */

-  #define nedmalloc               malloc

-  #define nedcalloc               calloc

-  #define nedrealloc              realloc

-  #define nedfree                 free

-  #define nedmemalign             memalign

-  #define nedmallinfo             mallinfo

-  #define nedmallopt              mallopt

-  #define nedmalloc_trim          malloc_trim

-  #define nedmalloc_stats         malloc_stats

-  #define nedmalloc_footprint     malloc_footprint

-  #define nedindependent_calloc   independent_calloc

-  #define nedindependent_comalloc independent_comalloc

-  #ifdef _MSC_VER

-   #define nedblksize              _msize

-  #endif

- #endif

-#endif

-

-#if defined(__cplusplus)

-extern "C" {

-#endif

-struct nedmallinfo {

-  size_t arena;    /* non-mmapped space allocated from system */

-  size_t ordblks;  /* number of free chunks */

-  size_t smblks;   /* always 0 */

-  size_t hblks;    /* always 0 */

-  size_t hblkhd;   /* space in mmapped regions */

-  size_t usmblks;  /* maximum total allocated space */

-  size_t fsmblks;  /* always 0 */

-  size_t uordblks; /* total allocated space */

-  size_t fordblks; /* total free space */

-  size_t keepcost; /* releasable (via malloc_trim) space */

-};

-#if defined(__cplusplus)

-}

-#endif

-

-#if defined(__cplusplus)

- #if !defined(NO_NED_NAMESPACE)

-namespace nedalloc {

- #else

-extern "C" {

- #endif

- #define THROWSPEC throw()

-#else

- #define THROWSPEC

-#endif

-

-/* These are the global functions */

-

-/* Gets the usable size of an allocated block. Note this will always be bigger than what was

-asked for due to rounding etc. Optionally returns 1 in isforeign if the block came from the

-system allocator - note that there is a small (>0.01%) but real chance of segfault on non-Windows

-systems when passing non-nedmalloc blocks if you don't use USE_MAGIC_HEADERS.

-*/

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC;

-

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC;

-

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void   nedfree(void *mem) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int    nedmallopt(int parno, int value) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void*  nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int    nedmalloc_trim(size_t pad) THROWSPEC;

-NEDMALLOCEXTSPEC void   nedmalloc_stats(void) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedmalloc_footprint(void) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC;

-

-/* Destroys the system memory pool used by the functions above.

-Useful for when you have nedmalloc in a DLL you're about to unload.

-If you call ANY nedmalloc functions after calling this you will

-get a fatal exception!

-*/

-NEDMALLOCEXTSPEC void neddestroysyspool() THROWSPEC;

-

-/* These are the pool functions */

-struct nedpool_t;

-typedef struct nedpool_t nedpool;

-

-/* Creates a memory pool for use with the nedp* functions below.

-Capacity is how much to allocate immediately (if you know you'll be allocating a lot

-of memory very soon) which you can leave at zero. Threads specifies how many threads

-will *normally* be accessing the pool concurrently. Setting this to zero means it

-extends on demand, but be careful of this as it can rapidly consume system resources

-where bursts of concurrent threads use a pool at once.

-*/

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC;

-

-/* Destroys a memory pool previously created by nedcreatepool().

-*/

-NEDMALLOCEXTSPEC void neddestroypool(nedpool *p) THROWSPEC;

-

-/* Returns a zero terminated snapshot of threadpools existing at the time of call. Call

-nedfree() on the returned list when you are done. Returns zero if there is only the

-system pool in existence.

-*/

-NEDMALLOCEXTSPEC nedpool **nedpoollist() THROWSPEC;

-

-/* Sets a value to be associated with a pool. You can retrieve this value by passing

-any memory block allocated from that pool.

-*/

-NEDMALLOCEXTSPEC void nedpsetvalue(nedpool *p, void *v) THROWSPEC;

-

-/* Gets a previously set value using nedpsetvalue() or zero if memory is unknown.

-Optionally can also retrieve pool. You can detect an unknown block by the return

-being zero and *p being unmodifed.

-*/

-NEDMALLOCEXTSPEC void *nedgetvalue(nedpool **p, void *mem) THROWSPEC;

-

-/* Trims the thread cache for the calling thread, returning any existing cache

-data to the central pool. Remember to ALWAYS call with zero if you used the

-system pool. Setting disable to non-zero replicates neddisablethreadcache().

-*/

-NEDMALLOCEXTSPEC void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC;

-

-/* Disables the thread cache for the calling thread, returning any existing cache

-data to the central pool. Remember to ALWAYS call with zero if you used the

-system pool.

-*/

-NEDMALLOCEXTSPEC void neddisablethreadcache(nedpool *p) THROWSPEC;

-

-

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC;

-NEDMALLOCEXTSPEC void   nedpfree(nedpool *p, void *mem) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC;

-NEDMALLOCEXTSPEC struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC;

-NEDMALLOCEXTSPEC int    nedpmallopt(nedpool *p, int parno, int value) THROWSPEC;

-NEDMALLOCEXTSPEC int    nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC;

-NEDMALLOCEXTSPEC void   nedpmalloc_stats(nedpool *p) THROWSPEC;

-NEDMALLOCEXTSPEC size_t nedpmalloc_footprint(nedpool *p) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;

-NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC;

-

-#if defined(__cplusplus)

-}

-#endif

-

-#endif

-

-#endif

+#ifdef NEDMALLOC_ENABLED
+
+/* nedalloc, an alternative malloc implementation for multiple threads without
+lock contention based on dlmalloc v2.8.3. (C) 2005-2009 Niall Douglas
+
+Boost Software License - Version 1.0 - August 17th, 2003
+
+Permission is hereby granted, free of charge, to any person or organization
+obtaining a copy of the software and accompanying documentation covered by
+this license (the "Software") to use, reproduce, display, distribute,
+execute, and transmit the Software, and to prepare derivative works of the
+Software, and to permit third-parties to whom the Software is furnished to
+do so, all subject to the following:
+
+The copyright notices in the Software and this entire statement, including
+the above license grant, this restriction and the following disclaimer,
+must be included in all copies of the Software, in whole or in part, and
+all derivative works of the Software, unless such copies or derivative
+works are solely in the form of machine-executable object code generated by
+a source language processor.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
+SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
+FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
+ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+DEALINGS IN THE SOFTWARE.
+*/
+
+#ifndef NEDMALLOC_H
+#define NEDMALLOC_H
+
+#include "typedefs.h"
+#define MALLOC_ALIGNMENT DEFAULT_ALIGNMENT
+
+#ifdef PSP_ENABLED
+#define USE_LOCKS 0
+#define HAVE_MMAP 0
+#endif
+
+/* See malloc.c.h for what each function does.
+
+REPLACE_SYSTEM_ALLOCATOR on POSIX causes nedalloc's functions to be called
+malloc, free etc. instead of nedmalloc, nedfree etc. You may or may not want
+this. On Windows it causes nedmalloc to patch all loaded DLLs and binaries
+to replace usage of the system allocator.
+
+NO_NED_NAMESPACE prevents the functions from being defined in the nedalloc
+namespace when in C++ (uses the global namespace instead).
+
+NEDMALLOCEXTSPEC can be defined to be __declspec(dllexport) or
+__attribute__ ((visibility("default"))) or whatever you like. It defaults
+to extern unless NEDMALLOC_DLL_EXPORTS is set as it would be when building
+nedmalloc.dll.
+
+USE_LOCKS can be 2 if you want to define your own MLOCK_T, INITIAL_LOCK,
+ACQUIRE_LOCK, RELEASE_LOCK, TRY_LOCK, IS_LOCKED and NULL_LOCK_INITIALIZER.
+
+NEDMALLOC_DEBUG can be defined to cause DEBUG to be set differently for nedmalloc
+than for the rest of the build. Remember to set NDEBUG to disable all assertion
+checking too.
+
+USE_MAGIC_HEADERS causes nedalloc to allocate an extra three sizeof(size_t)
+to each block. nedpfree() and nedprealloc() can then automagically know when
+to free a system allocated block. Enabling this typically adds 20-50% to
+application memory usage.
+
+ENABLE_TOLERANT_NEDMALLOC is automatically turned on if REPLACE_SYSTEM_ALLOCATOR
+is set or the Windows DLL is being built. This causes nedmalloc to detect when a
+system allocator block is passed to it and to handle it appropriately. Note that
+without USE_MAGIC_HEADERS there is a very tiny chance that nedmalloc will segfault
+on non-Windows builds (it uses Win32 SEH to trap segfaults on Windows and there
+is no comparable system on POSIX).
+
+USE_ALLOCATOR can be one of these settings (it defaults to 1):
+  0: System allocator (nedmalloc now simply acts as a threadcache).
+     WARNING: Intended for DEBUG USE ONLY - not all functions work correctly.
+  1: dlmalloc
+
+ENABLE_LARGE_PAGES enables support for requesting memory from the system in large
+(typically >=2Mb) pages if the host OS supports this. These occupy just a single
+TLB entry and can significantly improve performance in large working set applications.
+
+ENABLE_FAST_HEAP_DETECTION enables special logic to detect blocks allocated
+by the system heap. This avoids 1.5%-2% overhead when checking for non-nedmalloc
+blocks, but it assumes that the NT and glibc heaps function in a very specific
+fashion which may not hold true across OS upgrades.
+*/
+
+#include <stddef.h>   /* for size_t */
+
+#ifndef NEDMALLOCEXTSPEC
+ #ifdef NEDMALLOC_DLL_EXPORTS
+  #ifdef WIN32
+   #define NEDMALLOCEXTSPEC extern __declspec(dllexport)
+  #elif defined(__GNUC__)
+   #define NEDMALLOCEXTSPEC extern __attribute__ ((visibility("default")))
+  #endif
+  #ifndef ENABLE_TOLERANT_NEDMALLOC
+   #define ENABLE_TOLERANT_NEDMALLOC 1
+  #endif
+ #else
+  #define NEDMALLOCEXTSPEC extern
+ #endif
+#endif
+
+#if __STDC_VERSION__ >= 199901L		/* C99 or better */
+ #define RESTRICT restrict
+#else
+ #if defined(_MSC_VER) && _MSC_VER>=1400
+  #define RESTRICT __restrict
+ #endif
+ #ifdef __GNUC__
+  #define RESTRICT __restrict
+ #endif
+#endif
+#ifndef RESTRICT
+ #define RESTRICT
+#endif
+
+#if defined(_MSC_VER) && _MSC_VER>=1400
+ #define NEDMALLOCPTRATTR __declspec(restrict)
+ #define NEDMALLOCNOALIASATTR __declspec(noalias)
+#endif
+#ifdef __GNUC__
+ #define NEDMALLOCPTRATTR __attribute__ ((malloc))
+#endif
+#ifndef NEDMALLOCPTRATTR
+ #define NEDMALLOCPTRATTR
+#endif
+#ifndef NEDMALLOCNOALIASATTR
+ #define NEDMALLOCNOALIASATTR
+#endif
+
+#ifndef USE_MAGIC_HEADERS
+ #define USE_MAGIC_HEADERS 0
+#endif
+
+#ifndef USE_ALLOCATOR
+ #define USE_ALLOCATOR 1 /* dlmalloc */
+#endif
+
+#if !USE_ALLOCATOR && !USE_MAGIC_HEADERS
+#error If you are using the system allocator then you MUST use magic headers
+#endif
+
+#ifdef REPLACE_SYSTEM_ALLOCATOR
+ #if USE_ALLOCATOR==0
+  #error Cannot combine using the system allocator with replacing the system allocator
+ #endif
+ #ifndef ENABLE_TOLERANT_NEDMALLOC
+  #define ENABLE_TOLERANT_NEDMALLOC 1
+ #endif
+ #ifndef WIN32	/* We have a dedicated patcher for Windows */
+  #define nedmalloc               malloc
+  #define nedcalloc               calloc
+  #define nedrealloc              realloc
+  #define nedfree                 free
+  #define nedmemalign             memalign
+  #define nedmallinfo             mallinfo
+  #define nedmallopt              mallopt
+  #define nedmalloc_trim          malloc_trim
+  #define nedmalloc_stats         malloc_stats
+  #define nedmalloc_footprint     malloc_footprint
+  #define nedindependent_calloc   independent_calloc
+  #define nedindependent_comalloc independent_comalloc
+  #ifdef _MSC_VER
+   #define nedblksize              _msize
+  #endif
+ #endif
+#endif
+
+#if defined(__cplusplus)
+extern "C" {
+#endif
+struct nedmallinfo {
+  size_t arena;    /* non-mmapped space allocated from system */
+  size_t ordblks;  /* number of free chunks */
+  size_t smblks;   /* always 0 */
+  size_t hblks;    /* always 0 */
+  size_t hblkhd;   /* space in mmapped regions */
+  size_t usmblks;  /* maximum total allocated space */
+  size_t fsmblks;  /* always 0 */
+  size_t uordblks; /* total allocated space */
+  size_t fordblks; /* total free space */
+  size_t keepcost; /* releasable (via malloc_trim) space */
+};
+#if defined(__cplusplus)
+}
+#endif
+
+#if defined(__cplusplus)
+ #if !defined(NO_NED_NAMESPACE)
+namespace nedalloc {
+ #else
+extern "C" {
+ #endif
+ #define THROWSPEC throw()
+#else
+ #define THROWSPEC
+#endif
+
+/* These are the global functions */
+
+/* Gets the usable size of an allocated block. Note this will always be bigger than what was
+asked for due to rounding etc. Optionally returns 1 in isforeign if the block came from the
+system allocator - note that there is a small (>0.01%) but real chance of segfault on non-Windows
+systems when passing non-nedmalloc blocks if you don't use USE_MAGIC_HEADERS.
+*/
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC;
+
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC;
+
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void   nedfree(void *mem) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int    nedmallopt(int parno, int value) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void*  nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int    nedmalloc_trim(size_t pad) THROWSPEC;
+NEDMALLOCEXTSPEC void   nedmalloc_stats(void) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedmalloc_footprint(void) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC;
+
+/* Destroys the system memory pool used by the functions above.
+Useful for when you have nedmalloc in a DLL you're about to unload.
+If you call ANY nedmalloc functions after calling this you will
+get a fatal exception!
+*/
+NEDMALLOCEXTSPEC void neddestroysyspool() THROWSPEC;
+
+/* These are the pool functions */
+struct nedpool_t;
+typedef struct nedpool_t nedpool;
+
+/* Creates a memory pool for use with the nedp* functions below.
+Capacity is how much to allocate immediately (if you know you'll be allocating a lot
+of memory very soon) which you can leave at zero. Threads specifies how many threads
+will *normally* be accessing the pool concurrently. Setting this to zero means it
+extends on demand, but be careful of this as it can rapidly consume system resources
+where bursts of concurrent threads use a pool at once.
+*/
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC;
+
+/* Destroys a memory pool previously created by nedcreatepool().
+*/
+NEDMALLOCEXTSPEC void neddestroypool(nedpool *p) THROWSPEC;
+
+/* Returns a zero terminated snapshot of threadpools existing at the time of call. Call
+nedfree() on the returned list when you are done. Returns zero if there is only the
+system pool in existence.
+*/
+NEDMALLOCEXTSPEC nedpool **nedpoollist() THROWSPEC;
+
+/* Sets a value to be associated with a pool. You can retrieve this value by passing
+any memory block allocated from that pool.
+*/
+NEDMALLOCEXTSPEC void nedpsetvalue(nedpool *p, void *v) THROWSPEC;
+
+/* Gets a previously set value using nedpsetvalue() or zero if memory is unknown.
+Optionally can also retrieve pool. You can detect an unknown block by the return
+being zero and *p being unmodifed.
+*/
+NEDMALLOCEXTSPEC void *nedgetvalue(nedpool **p, void *mem) THROWSPEC;
+
+/* Trims the thread cache for the calling thread, returning any existing cache
+data to the central pool. Remember to ALWAYS call with zero if you used the
+system pool. Setting disable to non-zero replicates neddisablethreadcache().
+*/
+NEDMALLOCEXTSPEC void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC;
+
+/* Disables the thread cache for the calling thread, returning any existing cache
+data to the central pool. Remember to ALWAYS call with zero if you used the
+system pool.
+*/
+NEDMALLOCEXTSPEC void neddisablethreadcache(nedpool *p) THROWSPEC;
+
+
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC;
+NEDMALLOCEXTSPEC void   nedpfree(nedpool *p, void *mem) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC;
+NEDMALLOCEXTSPEC struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC;
+NEDMALLOCEXTSPEC int    nedpmallopt(nedpool *p, int parno, int value) THROWSPEC;
+NEDMALLOCEXTSPEC int    nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC;
+NEDMALLOCEXTSPEC void   nedpmalloc_stats(nedpool *p) THROWSPEC;
+NEDMALLOCEXTSPEC size_t nedpmalloc_footprint(nedpool *p) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
+NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC;
+
+#if defined(__cplusplus)
+}
+#endif
+
+#endif
+
+#endif
diff --git a/drivers/openssl/register_openssl.cpp b/drivers/openssl/register_openssl.cpp
index a4a60813b6..ed2150bef5 100644
--- a/drivers/openssl/register_openssl.cpp
+++ b/drivers/openssl/register_openssl.cpp
@@ -1,19 +1,19 @@
-#include "register_openssl.h"

-

-#include "stream_peer_openssl.h"

-#ifdef OPENSSL_ENABLED

-

-void register_openssl() {

-

-	ObjectTypeDB::register_type<StreamPeerOpenSSL>();

-	StreamPeerOpenSSL::initialize_ssl();

-

-}

-

-void unregister_openssl() {

-

-	StreamPeerOpenSSL::finalize_ssl();

-

-}

-#endif

-

+#include "register_openssl.h"
+
+#include "stream_peer_openssl.h"
+#ifdef OPENSSL_ENABLED
+
+void register_openssl() {
+
+	ObjectTypeDB::register_type<StreamPeerOpenSSL>();
+	StreamPeerOpenSSL::initialize_ssl();
+
+}
+
+void unregister_openssl() {
+
+	StreamPeerOpenSSL::finalize_ssl();
+
+}
+#endif
+
diff --git a/drivers/openssl/register_openssl.h b/drivers/openssl/register_openssl.h
index e1c554ca4a..e547a2b750 100644
--- a/drivers/openssl/register_openssl.h
+++ b/drivers/openssl/register_openssl.h
@@ -1,11 +1,11 @@
-#ifndef REGISTER_OPENSSL_H

-#define REGISTER_OPENSSL_H

-

-#ifdef OPENSSL_ENABLED

-

-void register_openssl();

-void unregister_openssl();

-

-#endif

-

-#endif // REGISTER_OPENSSL_H

+#ifndef REGISTER_OPENSSL_H
+#define REGISTER_OPENSSL_H
+
+#ifdef OPENSSL_ENABLED
+
+void register_openssl();
+void unregister_openssl();
+
+#endif
+
+#endif // REGISTER_OPENSSL_H
diff --git a/drivers/opus/audio_stream_opus.cpp b/drivers/opus/audio_stream_opus.cpp
index be31420453..eb9c81e152 100644
--- a/drivers/opus/audio_stream_opus.cpp
+++ b/drivers/opus/audio_stream_opus.cpp
@@ -30,6 +30,8 @@
 /*************************************************************************/
 #include "audio_stream_opus.h"
 
+const float AudioStreamPlaybackOpus::osrate=48000.0f;
+
 int AudioStreamPlaybackOpus::_op_read_func(void *_stream, unsigned char *_ptr, int _nbytes) {
 	FileAccess *fa=(FileAccess*)_stream;
 
diff --git a/drivers/opus/audio_stream_opus.h b/drivers/opus/audio_stream_opus.h
index 5bb838cd31..2f173cc270 100644
--- a/drivers/opus/audio_stream_opus.h
+++ b/drivers/opus/audio_stream_opus.h
@@ -54,7 +54,7 @@ class AudioStreamPlaybackOpus : public AudioStreamPlayback {
 	static int _op_seek_func(void *_stream, opus_int64 _offset, int _whence);
 	static int _op_close_func(void *_stream);
 	static opus_int64 _op_tell_func(void *_stream);
-	static const float osrate=48000.0f;
+	static const float osrate;
 
 	String file;
 	int64_t frames_mixed;
diff --git a/drivers/opus/opus_config.h b/drivers/opus/opus_config.h
index e75550409f..c6470e92c3 100644
--- a/drivers/opus/opus_config.h
+++ b/drivers/opus/opus_config.h
@@ -91,8 +91,13 @@
 /* This is a build of OPUS */
 #define OPUS_BUILD /**/
 
-/* Use C99 variable-size arrays */
-#define VAR_ARRAYS 1
+#ifndef WIN32
+	/* Use C99 variable-size arrays */
+	#define VAR_ARRAYS 1
+#else
+	/* Fixes VS 2013 compile error */
+	#define USE_ALLOCA 1
+#endif
 
 
 /* Define to `__inline__' or `__inline' if that's what the C compiler
diff --git a/drivers/pulseaudio/audio_driver_pulseaudio.cpp b/drivers/pulseaudio/audio_driver_pulseaudio.cpp
index dfe9ddc55f..4cda141f92 100644
--- a/drivers/pulseaudio/audio_driver_pulseaudio.cpp
+++ b/drivers/pulseaudio/audio_driver_pulseaudio.cpp
@@ -82,6 +82,17 @@ Error AudioDriverPulseAudio::init() {
 	return OK;
 }
 
+float AudioDriverPulseAudio::get_latency() {
+
+	if (latency==0)	{ //only do this once since it's approximate anyway
+		int error_code;
+		pa_usec_t palat = pa_simple_get_latency( pulse,&error_code);
+		latency=double(palat)/1000000.0;
+	}
+
+	return latency;
+}
+
 void AudioDriverPulseAudio::thread_func(void* p_udata) {
 
     AudioDriverPulseAudio* ad = (AudioDriverPulseAudio*)p_udata;
@@ -121,6 +132,7 @@ void AudioDriverPulseAudio::thread_func(void* p_udata) {
             ad->exit_thread = true;
             break;
         }
+
     }
 
     ad->thread_exited = true;
@@ -185,6 +197,7 @@ AudioDriverPulseAudio::AudioDriverPulseAudio() {
 	mutex = NULL;
     thread = NULL;
     pulse = NULL;
+    latency=0;
 }
 
 AudioDriverPulseAudio::~AudioDriverPulseAudio() {
diff --git a/drivers/pulseaudio/audio_driver_pulseaudio.h b/drivers/pulseaudio/audio_driver_pulseaudio.h
index e82e0c24be..e7c8bcce36 100644
--- a/drivers/pulseaudio/audio_driver_pulseaudio.h
+++ b/drivers/pulseaudio/audio_driver_pulseaudio.h
@@ -58,6 +58,8 @@ class AudioDriverPulseAudio : public AudioDriverSW {
 	mutable bool exit_thread;
 	bool pcm_open;
 
+	float latency;
+
 public:
 
 	const char* get_name() const {
@@ -72,6 +74,9 @@ public:
 	virtual void unlock();
 	virtual void finish();
 
+	virtual float get_latency();
+
+
     AudioDriverPulseAudio();
     ~AudioDriverPulseAudio();
 };
diff --git a/drivers/register_driver_types.cpp b/drivers/register_driver_types.cpp
index ea95975c18..2647d23011 100644
--- a/drivers/register_driver_types.cpp
+++ b/drivers/register_driver_types.cpp
@@ -46,7 +46,10 @@
 #endif
 
 #ifdef THEORA_ENABLED
-//#include "theora/video_stream_theora.h"
+#include "theora/video_stream_theora.h"
+#endif
+
+#ifdef THEORAPLAYER_ENABLED
 #include "theoraplayer/video_stream_theoraplayer.h"
 #endif
 
@@ -97,7 +100,10 @@ static ResourceFormatLoaderAudioStreamSpeex *speex_stream_loader=NULL;
 #endif
 
 #ifdef THEORA_ENABLED
-//static ResourceFormatLoaderVideoStreamTheora* theora_stream_loader = NULL;
+static ResourceFormatLoaderVideoStreamTheora* theora_stream_loader = NULL;
+#endif
+
+#ifdef THEORAPLAYER_ENABLED
 static ResourceFormatLoaderVideoStreamTheoraplayer* theoraplayer_stream_loader = NULL;
 #endif
 
@@ -217,9 +223,12 @@ void register_driver_types() {
 #endif
 
 #ifdef THEORA_ENABLED
-	//theora_stream_loader = memnew( ResourceFormatLoaderVideoStreamTheora );
-	//ResourceLoader::add_resource_format_loader(theora_stream_loader);
-	//ObjectTypeDB::register_type<VideoStreamTheora>();
+	theora_stream_loader = memnew( ResourceFormatLoaderVideoStreamTheora );
+	ResourceLoader::add_resource_format_loader(theora_stream_loader);
+	ObjectTypeDB::register_type<VideoStreamTheora>();
+#endif
+
+#ifdef THEORAPLAYER_ENABLED
 	theoraplayer_stream_loader = memnew( ResourceFormatLoaderVideoStreamTheoraplayer );
 	ResourceLoader::add_resource_format_loader(theoraplayer_stream_loader);
 	ObjectTypeDB::register_type<VideoStreamTheoraplayer>();
@@ -260,7 +269,10 @@ void unregister_driver_types() {
 #endif
 
 #ifdef THEORA_ENABLED
-	//memdelete (theora_stream_loader);
+	memdelete (theora_stream_loader);
+#endif
+
+#ifdef THEORAPLAYER_ENABLED
 	memdelete (theoraplayer_stream_loader);
 #endif
 
diff --git a/drivers/rtaudio/RtAudio.cpp b/drivers/rtaudio/RtAudio.cpp
index 8876f72e21..72ca836907 100644
--- a/drivers/rtaudio/RtAudio.cpp
+++ b/drivers/rtaudio/RtAudio.cpp
@@ -1,10234 +1,10234 @@
-#ifdef RTAUDIO_ENABLED

-/************************************************************************/

-/*! \class RtAudio

-    \brief Realtime audio i/o C++ classes.

-

-    RtAudio provides a common API (Application Programming Interface)

-    for realtime audio input/output across Linux (native ALSA, Jack,

-    and OSS), Macintosh OS X (CoreAudio and Jack), and Windows

-    (DirectSound, ASIO and WASAPI) operating systems.

-

-    RtAudio WWW site: http://www.music.mcgill.ca/~gary/rtaudio/

-

-    RtAudio: realtime audio i/o C++ classes

-    Copyright (c) 2001-2014 Gary P. Scavone

-

-    Permission is hereby granted, free of charge, to any person

-    obtaining a copy of this software and associated documentation files

-    (the "Software"), to deal in the Software without restriction,

-    including without limitation the rights to use, copy, modify, merge,

-    publish, distribute, sublicense, and/or sell copies of the Software,

-    and to permit persons to whom the Software is furnished to do so,

-    subject to the following conditions:

-

-    The above copyright notice and this permission notice shall be

-    included in all copies or substantial portions of the Software.

-

-    Any person wishing to distribute modifications to the Software is

-    asked to send the modifications to the original developer so that

-    they can be incorporated into the canonical version.  This is,

-    however, not a binding provision of this license.

-

-    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-    EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-    MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-    IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR

-    ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF

-    CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION

-    WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

-*/

-/************************************************************************/

-

-// RtAudio: Version 4.1.1

-

-#include "RtAudio.h"

-#include <iostream>

-#include <cstdlib>

-#include <cstring>

-#include <climits>

-#include <algorithm>

-

-// Static variable definitions.

-const unsigned int RtApi::MAX_SAMPLE_RATES = 14;

-const unsigned int RtApi::SAMPLE_RATES[] = {

-  4000, 5512, 8000, 9600, 11025, 16000, 22050,

-  32000, 44100, 48000, 88200, 96000, 176400, 192000

-};

-

-#if defined(__WINDOWS_DS__) || defined(__WINDOWS_ASIO__) || defined(__WINDOWS_WASAPI__)

-#ifdef WINRT_ENABLED

-  #define MUTEX_INITIALIZE(A) InitializeCriticalSectionEx(A, 0, 0)

-#else

-  #define MUTEX_INITIALIZE(A) InitializeCriticalSection(A)

-#endif

-  #define MUTEX_DESTROY(A)    DeleteCriticalSection(A)

-  #define MUTEX_LOCK(A)       EnterCriticalSection(A)

-  #define MUTEX_UNLOCK(A)     LeaveCriticalSection(A)

-

-  #include "tchar.h"

-

-  static std::string convertCharPointerToStdString(const char *text)

-  {

-    return std::string(text);

-  }

-

-  static std::string convertCharPointerToStdString(const wchar_t *text)

-  {

-    int length = WideCharToMultiByte(CP_UTF8, 0, text, -1, NULL, 0, NULL, NULL);

-    std::string s( length-1, '\0' );

-    WideCharToMultiByte(CP_UTF8, 0, text, -1, &s[0], length, NULL, NULL);

-    return s;

-  }

-

-#elif defined(__LINUX_ALSA__) || defined(__LINUX_PULSE__) || defined(__UNIX_JACK__) || defined(__LINUX_OSS__) || defined(__MACOSX_CORE__)

-  // pthread API

-  #define MUTEX_INITIALIZE(A) pthread_mutex_init(A, NULL)

-  #define MUTEX_DESTROY(A)    pthread_mutex_destroy(A)

-  #define MUTEX_LOCK(A)       pthread_mutex_lock(A)

-  #define MUTEX_UNLOCK(A)     pthread_mutex_unlock(A)

-#else

-  #define MUTEX_INITIALIZE(A) abs(*A) // dummy definitions

-  #define MUTEX_DESTROY(A)    abs(*A) // dummy definitions

-#endif

-

-// *************************************************** //

-//

-// RtAudio definitions.

-//

-// *************************************************** //

-

-std::string RtAudio :: getVersion( void ) throw()

-{

-  return RTAUDIO_VERSION;

-}

-

-void RtAudio :: getCompiledApi( std::vector<RtAudio::Api> &apis ) throw()

-{

-  apis.clear();

-

-  // The order here will control the order of RtAudio's API search in

-  // the constructor.

-#if defined(__UNIX_JACK__)

-  apis.push_back( UNIX_JACK );

-#endif

-#if defined(__LINUX_ALSA__)

-  apis.push_back( LINUX_ALSA );

-#endif

-#if defined(__LINUX_PULSE__)

-  apis.push_back( LINUX_PULSE );

-#endif

-#if defined(__LINUX_OSS__)

-  apis.push_back( LINUX_OSS );

-#endif

-#if defined(__WINDOWS_ASIO__)

-  apis.push_back( WINDOWS_ASIO );

-#endif

-#if defined(__WINDOWS_WASAPI__)

-  apis.push_back( WINDOWS_WASAPI );

-#endif

-#if defined(__WINDOWS_DS__)

-  apis.push_back( WINDOWS_DS );

-#endif

-#if defined(__MACOSX_CORE__)

-  apis.push_back( MACOSX_CORE );

-#endif

-#if defined(__RTAUDIO_DUMMY__)

-  apis.push_back( RTAUDIO_DUMMY );

-#endif

-}

-

-void RtAudio :: openRtApi( RtAudio::Api api )

-{

-  if ( rtapi_ )

-    delete rtapi_;

-  rtapi_ = 0;

-

-#if defined(__UNIX_JACK__)

-  if ( api == UNIX_JACK )

-    rtapi_ = new RtApiJack();

-#endif

-#if defined(__LINUX_ALSA__)

-  if ( api == LINUX_ALSA )

-    rtapi_ = new RtApiAlsa();

-#endif

-#if defined(__LINUX_PULSE__)

-  if ( api == LINUX_PULSE )

-    rtapi_ = new RtApiPulse();

-#endif

-#if defined(__LINUX_OSS__)

-  if ( api == LINUX_OSS )

-    rtapi_ = new RtApiOss();

-#endif

-#if defined(__WINDOWS_ASIO__)

-  if ( api == WINDOWS_ASIO )

-    rtapi_ = new RtApiAsio();

-#endif

-#if defined(__WINDOWS_WASAPI__)

-  if ( api == WINDOWS_WASAPI )

-    rtapi_ = new RtApiWasapi();

-#endif

-#if defined(__WINDOWS_DS__)

-  if ( api == WINDOWS_DS )

-    rtapi_ = new RtApiDs();

-#endif

-#if defined(__MACOSX_CORE__)

-  if ( api == MACOSX_CORE )

-    rtapi_ = new RtApiCore();

-#endif

-#if defined(__RTAUDIO_DUMMY__)

-  if ( api == RTAUDIO_DUMMY )

-    rtapi_ = new RtApiDummy();

-#endif

-}

-

-RtAudio :: RtAudio( RtAudio::Api api )

-{

-  rtapi_ = 0;

-

-  if ( api != UNSPECIFIED ) {

-    // Attempt to open the specified API.

-    openRtApi( api );

-    if ( rtapi_ ) return;

-

-    // No compiled support for specified API value.  Issue a debug

-    // warning and continue as if no API was specified.

-    std::cerr << "\nRtAudio: no compiled support for specified API argument!\n" << std::endl;

-  }

-

-  // Iterate through the compiled APIs and return as soon as we find

-  // one with at least one device or we reach the end of the list.

-  std::vector< RtAudio::Api > apis;

-  getCompiledApi( apis );

-  for ( unsigned int i=0; i<apis.size(); i++ ) {

-    openRtApi( apis[i] );

-    if ( rtapi_ && rtapi_->getDeviceCount() ) break;

-  }

-

-  if ( rtapi_ ) return;

-

-  // It should not be possible to get here because the preprocessor

-  // definition __RTAUDIO_DUMMY__ is automatically defined if no

-  // API-specific definitions are passed to the compiler. But just in

-  // case something weird happens, we'll thow an error.

-  std::string errorText = "\nRtAudio: no compiled API support found ... critical error!!\n\n";

-  throw( RtAudioError( errorText, RtAudioError::UNSPECIFIED ) );

-}

-

-RtAudio :: ~RtAudio() throw()

-{

-  if ( rtapi_ )

-    delete rtapi_;

-}

-

-void RtAudio :: openStream( RtAudio::StreamParameters *outputParameters,

-                            RtAudio::StreamParameters *inputParameters,

-                            RtAudioFormat format, unsigned int sampleRate,

-                            unsigned int *bufferFrames,

-                            RtAudioCallback callback, void *userData,

-                            RtAudio::StreamOptions *options,

-                            RtAudioErrorCallback errorCallback )

-{

-  return rtapi_->openStream( outputParameters, inputParameters, format,

-                             sampleRate, bufferFrames, callback,

-                             userData, options, errorCallback );

-}

-

-// *************************************************** //

-//

-// Public RtApi definitions (see end of file for

-// private or protected utility functions).

-//

-// *************************************************** //

-

-RtApi :: RtApi()

-{

-  stream_.state = STREAM_CLOSED;

-  stream_.mode = UNINITIALIZED;

-  stream_.apiHandle = 0;

-  stream_.userBuffer[0] = 0;

-  stream_.userBuffer[1] = 0;

-  MUTEX_INITIALIZE( &stream_.mutex );

-  showWarnings_ = true;

-  firstErrorOccurred_ = false;

-}

-

-RtApi :: ~RtApi()

-{

-  MUTEX_DESTROY( &stream_.mutex );

-}

-

-void RtApi :: openStream( RtAudio::StreamParameters *oParams,

-                          RtAudio::StreamParameters *iParams,

-                          RtAudioFormat format, unsigned int sampleRate,

-                          unsigned int *bufferFrames,

-                          RtAudioCallback callback, void *userData,

-                          RtAudio::StreamOptions *options,

-                          RtAudioErrorCallback errorCallback )

-{

-  if ( stream_.state != STREAM_CLOSED ) {

-    errorText_ = "RtApi::openStream: a stream is already open!";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-

-  // Clear stream information potentially left from a previously open stream.

-  clearStreamInfo();

-

-  if ( oParams && oParams->nChannels < 1 ) {

-    errorText_ = "RtApi::openStream: a non-NULL output StreamParameters structure cannot have an nChannels value less than one.";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-

-  if ( iParams && iParams->nChannels < 1 ) {

-    errorText_ = "RtApi::openStream: a non-NULL input StreamParameters structure cannot have an nChannels value less than one.";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-

-  if ( oParams == NULL && iParams == NULL ) {

-    errorText_ = "RtApi::openStream: input and output StreamParameters structures are both NULL!";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-

-  if ( formatBytes(format) == 0 ) {

-    errorText_ = "RtApi::openStream: 'format' parameter value is undefined.";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-

-  unsigned int nDevices = getDeviceCount();

-  unsigned int oChannels = 0;

-  if ( oParams ) {

-    oChannels = oParams->nChannels;

-    if ( oParams->deviceId >= nDevices ) {

-      errorText_ = "RtApi::openStream: output device parameter value is invalid.";

-      error( RtAudioError::INVALID_USE );

-      return;

-    }

-  }

-

-  unsigned int iChannels = 0;

-  if ( iParams ) {

-    iChannels = iParams->nChannels;

-    if ( iParams->deviceId >= nDevices ) {

-      errorText_ = "RtApi::openStream: input device parameter value is invalid.";

-      error( RtAudioError::INVALID_USE );

-      return;

-    }

-  }

-

-  bool result;

-

-  if ( oChannels > 0 ) {

-

-    result = probeDeviceOpen( oParams->deviceId, OUTPUT, oChannels, oParams->firstChannel,

-                              sampleRate, format, bufferFrames, options );

-    if ( result == false ) {

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-  }

-

-  if ( iChannels > 0 ) {

-

-    result = probeDeviceOpen( iParams->deviceId, INPUT, iChannels, iParams->firstChannel,

-                              sampleRate, format, bufferFrames, options );

-    if ( result == false ) {

-      if ( oChannels > 0 ) closeStream();

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-  }

-

-  stream_.callbackInfo.callback = (void *) callback;

-  stream_.callbackInfo.userData = userData;

-  stream_.callbackInfo.errorCallback = (void *) errorCallback;

-

-  if ( options ) options->numberOfBuffers = stream_.nBuffers;

-  stream_.state = STREAM_STOPPED;

-}

-

-unsigned int RtApi :: getDefaultInputDevice( void )

-{

-  // Should be implemented in subclasses if possible.

-  return 0;

-}

-

-unsigned int RtApi :: getDefaultOutputDevice( void )

-{

-  // Should be implemented in subclasses if possible.

-  return 0;

-}

-

-void RtApi :: closeStream( void )

-{

-  // MUST be implemented in subclasses!

-  return;

-}

-

-bool RtApi :: probeDeviceOpen( unsigned int /*device*/, StreamMode /*mode*/, unsigned int /*channels*/,

-                               unsigned int /*firstChannel*/, unsigned int /*sampleRate*/,

-                               RtAudioFormat /*format*/, unsigned int * /*bufferSize*/,

-                               RtAudio::StreamOptions * /*options*/ )

-{

-  // MUST be implemented in subclasses!

-  return FAILURE;

-}

-

-void RtApi :: tickStreamTime( void )

-{

-  // Subclasses that do not provide their own implementation of

-  // getStreamTime should call this function once per buffer I/O to

-  // provide basic stream time support.

-

-  stream_.streamTime += ( stream_.bufferSize * 1.0 / stream_.sampleRate );

-

-#if defined( HAVE_GETTIMEOFDAY )

-  gettimeofday( &stream_.lastTickTimestamp, NULL );

-#endif

-}

-

-long RtApi :: getStreamLatency( void )

-{

-  verifyStream();

-

-  long totalLatency = 0;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )

-    totalLatency = stream_.latency[0];

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX )

-    totalLatency += stream_.latency[1];

-

-  return totalLatency;

-}

-

-double RtApi :: getStreamTime( void )

-{

-  verifyStream();

-

-#if defined( HAVE_GETTIMEOFDAY )

-  // Return a very accurate estimate of the stream time by

-  // adding in the elapsed time since the last tick.

-  struct timeval then;

-  struct timeval now;

-

-  if ( stream_.state != STREAM_RUNNING || stream_.streamTime == 0.0 )

-    return stream_.streamTime;

-

-  gettimeofday( &now, NULL );

-  then = stream_.lastTickTimestamp;

-  return stream_.streamTime +

-    ((now.tv_sec + 0.000001 * now.tv_usec) -

-     (then.tv_sec + 0.000001 * then.tv_usec));     

-#else

-  return stream_.streamTime;

-#endif

-}

-

-void RtApi :: setStreamTime( double time )

-{

-  verifyStream();

-

-  if ( time >= 0.0 )

-    stream_.streamTime = time;

-}

-

-unsigned int RtApi :: getStreamSampleRate( void )

-{

- verifyStream();

-

- return stream_.sampleRate;

-}

-

-

-// *************************************************** //

-//

-// OS/API-specific methods.

-//

-// *************************************************** //

-

-#if defined(__MACOSX_CORE__)

-

-// The OS X CoreAudio API is designed to use a separate callback

-// procedure for each of its audio devices.  A single RtAudio duplex

-// stream using two different devices is supported here, though it

-// cannot be guaranteed to always behave correctly because we cannot

-// synchronize these two callbacks.

-//

-// A property listener is installed for over/underrun information.

-// However, no functionality is currently provided to allow property

-// listeners to trigger user handlers because it is unclear what could

-// be done if a critical stream parameter (buffer size, sample rate,

-// device disconnect) notification arrived.  The listeners entail

-// quite a bit of extra code and most likely, a user program wouldn't

-// be prepared for the result anyway.  However, we do provide a flag

-// to the client callback function to inform of an over/underrun.

-

-// A structure to hold various information related to the CoreAudio API

-// implementation.

-struct CoreHandle {

-  AudioDeviceID id[2];    // device ids

-#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )

-  AudioDeviceIOProcID procId[2];

-#endif

-  UInt32 iStream[2];      // device stream index (or first if using multiple)

-  UInt32 nStreams[2];     // number of streams to use

-  bool xrun[2];

-  char *deviceBuffer;

-  pthread_cond_t condition;

-  int drainCounter;       // Tracks callback counts when draining

-  bool internalDrain;     // Indicates if stop is initiated from callback or not.

-

-  CoreHandle()

-    :deviceBuffer(0), drainCounter(0), internalDrain(false) { nStreams[0] = 1; nStreams[1] = 1; id[0] = 0; id[1] = 0; xrun[0] = false; xrun[1] = false; }

-};

-

-RtApiCore:: RtApiCore()

-{

-#if defined( AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER )

-  // This is a largely undocumented but absolutely necessary

-  // requirement starting with OS-X 10.6.  If not called, queries and

-  // updates to various audio device properties are not handled

-  // correctly.

-  CFRunLoopRef theRunLoop = NULL;

-  AudioObjectPropertyAddress property = { kAudioHardwarePropertyRunLoop,

-                                          kAudioObjectPropertyScopeGlobal,

-                                          kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectSetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, sizeof(CFRunLoopRef), &theRunLoop);

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::RtApiCore: error setting run loop property!";

-    error( RtAudioError::WARNING );

-  }

-#endif

-}

-

-RtApiCore :: ~RtApiCore()

-{

-  // The subclass destructor gets called before the base class

-  // destructor, so close an existing stream before deallocating

-  // apiDeviceId memory.

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-}

-

-unsigned int RtApiCore :: getDeviceCount( void )

-{

-  // Find out how many audio devices there are, if any.

-  UInt32 dataSize;

-  AudioObjectPropertyAddress propertyAddress = { kAudioHardwarePropertyDevices, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectGetPropertyDataSize( kAudioObjectSystemObject, &propertyAddress, 0, NULL, &dataSize );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDeviceCount: OS-X error getting device info!";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  return dataSize / sizeof( AudioDeviceID );

-}

-

-unsigned int RtApiCore :: getDefaultInputDevice( void )

-{

-  unsigned int nDevices = getDeviceCount();

-  if ( nDevices <= 1 ) return 0;

-

-  AudioDeviceID id;

-  UInt32 dataSize = sizeof( AudioDeviceID );

-  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDefaultInputDevice, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, &id );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDefaultInputDevice: OS-X system error getting device.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  dataSize *= nDevices;

-  AudioDeviceID deviceList[ nDevices ];

-  property.mSelector = kAudioHardwarePropertyDevices;

-  result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, (void *) &deviceList );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDefaultInputDevice: OS-X system error getting device IDs.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  for ( unsigned int i=0; i<nDevices; i++ )

-    if ( id == deviceList[i] ) return i;

-

-  errorText_ = "RtApiCore::getDefaultInputDevice: No default device found!";

-  error( RtAudioError::WARNING );

-  return 0;

-}

-

-unsigned int RtApiCore :: getDefaultOutputDevice( void )

-{

-  unsigned int nDevices = getDeviceCount();

-  if ( nDevices <= 1 ) return 0;

-

-  AudioDeviceID id;

-  UInt32 dataSize = sizeof( AudioDeviceID );

-  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDefaultOutputDevice, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, &id );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDefaultOutputDevice: OS-X system error getting device.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  dataSize = sizeof( AudioDeviceID ) * nDevices;

-  AudioDeviceID deviceList[ nDevices ];

-  property.mSelector = kAudioHardwarePropertyDevices;

-  result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, (void *) &deviceList );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDefaultOutputDevice: OS-X system error getting device IDs.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  for ( unsigned int i=0; i<nDevices; i++ )

-    if ( id == deviceList[i] ) return i;

-

-  errorText_ = "RtApiCore::getDefaultOutputDevice: No default device found!";

-  error( RtAudioError::WARNING );

-  return 0;

-}

-

-RtAudio::DeviceInfo RtApiCore :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  // Get device ID

-  unsigned int nDevices = getDeviceCount();

-  if ( nDevices == 0 ) {

-    errorText_ = "RtApiCore::getDeviceInfo: no devices found!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  if ( device >= nDevices ) {

-    errorText_ = "RtApiCore::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  AudioDeviceID deviceList[ nDevices ];

-  UInt32 dataSize = sizeof( AudioDeviceID ) * nDevices;

-  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,

-                                          kAudioObjectPropertyScopeGlobal,

-                                          kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property,

-                                                0, NULL, &dataSize, (void *) &deviceList );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::getDeviceInfo: OS-X system error getting device IDs.";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  AudioDeviceID id = deviceList[ device ];

-

-  // Get the device name.

-  info.name.erase();

-  CFStringRef cfname;

-  dataSize = sizeof( CFStringRef );

-  property.mSelector = kAudioObjectPropertyManufacturer;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &cfname );

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceInfo: system error (" << getErrorCode( result ) << ") getting device manufacturer.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  //const char *mname = CFStringGetCStringPtr( cfname, CFStringGetSystemEncoding() );

-  int length = CFStringGetLength(cfname);

-  char *mname = (char *)malloc(length * 3 + 1);

-#if defined( UNICODE ) || defined( _UNICODE )

-  CFStringGetCString(cfname, mname, length * 3 + 1, kCFStringEncodingUTF8);

-#else

-  CFStringGetCString(cfname, mname, length * 3 + 1, CFStringGetSystemEncoding());

-#endif

-  info.name.append( (const char *)mname, strlen(mname) );

-  info.name.append( ": " );

-  CFRelease( cfname );

-  free(mname);

-

-  property.mSelector = kAudioObjectPropertyName;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &cfname );

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceInfo: system error (" << getErrorCode( result ) << ") getting device name.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  //const char *name = CFStringGetCStringPtr( cfname, CFStringGetSystemEncoding() );

-  length = CFStringGetLength(cfname);

-  char *name = (char *)malloc(length * 3 + 1);

-#if defined( UNICODE ) || defined( _UNICODE )

-  CFStringGetCString(cfname, name, length * 3 + 1, kCFStringEncodingUTF8);

-#else

-  CFStringGetCString(cfname, name, length * 3 + 1, CFStringGetSystemEncoding());

-#endif

-  info.name.append( (const char *)name, strlen(name) );

-  CFRelease( cfname );

-  free(name);

-

-  // Get the output stream "configuration".

-  AudioBufferList	*bufferList = nil;

-  property.mSelector = kAudioDevicePropertyStreamConfiguration;

-  property.mScope = kAudioDevicePropertyScopeOutput;

-  //  property.mElement = kAudioObjectPropertyElementWildcard;

-  dataSize = 0;

-  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );

-  if ( result != noErr || dataSize == 0 ) {

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting output stream configuration info for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Allocate the AudioBufferList.

-  bufferList = (AudioBufferList *) malloc( dataSize );

-  if ( bufferList == NULL ) {

-    errorText_ = "RtApiCore::getDeviceInfo: memory error allocating output AudioBufferList.";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );

-  if ( result != noErr || dataSize == 0 ) {

-    free( bufferList );

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting output stream configuration for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Get output channel information.

-  unsigned int i, nStreams = bufferList->mNumberBuffers;

-  for ( i=0; i<nStreams; i++ )

-    info.outputChannels += bufferList->mBuffers[i].mNumberChannels;

-  free( bufferList );

-

-  // Get the input stream "configuration".

-  property.mScope = kAudioDevicePropertyScopeInput;

-  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );

-  if ( result != noErr || dataSize == 0 ) {

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting input stream configuration info for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Allocate the AudioBufferList.

-  bufferList = (AudioBufferList *) malloc( dataSize );

-  if ( bufferList == NULL ) {

-    errorText_ = "RtApiCore::getDeviceInfo: memory error allocating input AudioBufferList.";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );

-  if (result != noErr || dataSize == 0) {

-    free( bufferList );

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting input stream configuration for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Get input channel information.

-  nStreams = bufferList->mNumberBuffers;

-  for ( i=0; i<nStreams; i++ )

-    info.inputChannels += bufferList->mBuffers[i].mNumberChannels;

-  free( bufferList );

-

-  // If device opens for both playback and capture, we determine the channels.

-  if ( info.outputChannels > 0 && info.inputChannels > 0 )

-    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-

-  // Probe the device sample rates.

-  bool isInput = false;

-  if ( info.outputChannels == 0 ) isInput = true;

-

-  // Determine the supported sample rates.

-  property.mSelector = kAudioDevicePropertyAvailableNominalSampleRates;

-  if ( isInput == false ) property.mScope = kAudioDevicePropertyScopeOutput;

-  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );

-  if ( result != kAudioHardwareNoError || dataSize == 0 ) {

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting sample rate info.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  UInt32 nRanges = dataSize / sizeof( AudioValueRange );

-  AudioValueRange rangeList[ nRanges ];

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &rangeList );

-  if ( result != kAudioHardwareNoError ) {

-    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting sample rates.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // The sample rate reporting mechanism is a bit of a mystery.  It

-  // seems that it can either return individual rates or a range of

-  // rates.  I assume that if the min / max range values are the same,

-  // then that represents a single supported rate and if the min / max

-  // range values are different, the device supports an arbitrary

-  // range of values (though there might be multiple ranges, so we'll

-  // use the most conservative range).

-  Float64 minimumRate = 1.0, maximumRate = 10000000000.0;

-  bool haveValueRange = false;

-  info.sampleRates.clear();

-  for ( UInt32 i=0; i<nRanges; i++ ) {

-    if ( rangeList[i].mMinimum == rangeList[i].mMaximum ) {

-      unsigned int tmpSr = (unsigned int) rangeList[i].mMinimum;

-      info.sampleRates.push_back( tmpSr );

-

-      if ( !info.preferredSampleRate || ( tmpSr <= 48000 && tmpSr > info.preferredSampleRate ) )

-        info.preferredSampleRate = tmpSr;

-

-    } else {

-      haveValueRange = true;

-      if ( rangeList[i].mMinimum > minimumRate ) minimumRate = rangeList[i].mMinimum;

-      if ( rangeList[i].mMaximum < maximumRate ) maximumRate = rangeList[i].mMaximum;

-    }

-  }

-

-  if ( haveValueRange ) {

-    for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {

-      if ( SAMPLE_RATES[k] >= (unsigned int) minimumRate && SAMPLE_RATES[k] <= (unsigned int) maximumRate ) {

-        info.sampleRates.push_back( SAMPLE_RATES[k] );

-

-        if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )

-          info.preferredSampleRate = SAMPLE_RATES[k];

-      }

-    }

-  }

-

-  // Sort and remove any redundant values

-  std::sort( info.sampleRates.begin(), info.sampleRates.end() );

-  info.sampleRates.erase( unique( info.sampleRates.begin(), info.sampleRates.end() ), info.sampleRates.end() );

-

-  if ( info.sampleRates.size() == 0 ) {

-    errorStream_ << "RtApiCore::probeDeviceInfo: No supported sample rates found for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // CoreAudio always uses 32-bit floating point data for PCM streams.

-  // Thus, any other "physical" formats supported by the device are of

-  // no interest to the client.

-  info.nativeFormats = RTAUDIO_FLOAT32;

-

-  if ( info.outputChannels > 0 )

-    if ( getDefaultOutputDevice() == device ) info.isDefaultOutput = true;

-  if ( info.inputChannels > 0 )

-    if ( getDefaultInputDevice() == device ) info.isDefaultInput = true;

-

-  info.probed = true;

-  return info;

-}

-

-static OSStatus callbackHandler( AudioDeviceID inDevice,

-                                 const AudioTimeStamp* /*inNow*/,

-                                 const AudioBufferList* inInputData,

-                                 const AudioTimeStamp* /*inInputTime*/,

-                                 AudioBufferList* outOutputData,

-                                 const AudioTimeStamp* /*inOutputTime*/,

-                                 void* infoPointer )

-{

-  CallbackInfo *info = (CallbackInfo *) infoPointer;

-

-  RtApiCore *object = (RtApiCore *) info->object;

-  if ( object->callbackEvent( inDevice, inInputData, outOutputData ) == false )

-    return kAudioHardwareUnspecifiedError;

-  else

-    return kAudioHardwareNoError;

-}

-

-static OSStatus xrunListener( AudioObjectID /*inDevice*/,

-                              UInt32 nAddresses,

-                              const AudioObjectPropertyAddress properties[],

-                              void* handlePointer )

-{

-  CoreHandle *handle = (CoreHandle *) handlePointer;

-  for ( UInt32 i=0; i<nAddresses; i++ ) {

-    if ( properties[i].mSelector == kAudioDeviceProcessorOverload ) {

-      if ( properties[i].mScope == kAudioDevicePropertyScopeInput )

-        handle->xrun[1] = true;

-      else

-        handle->xrun[0] = true;

-    }

-  }

-

-  return kAudioHardwareNoError;

-}

-

-static OSStatus rateListener( AudioObjectID inDevice,

-                              UInt32 /*nAddresses*/,

-                              const AudioObjectPropertyAddress /*properties*/[],

-                              void* ratePointer )

-{

-  Float64 *rate = (Float64 *) ratePointer;

-  UInt32 dataSize = sizeof( Float64 );

-  AudioObjectPropertyAddress property = { kAudioDevicePropertyNominalSampleRate,

-                                          kAudioObjectPropertyScopeGlobal,

-                                          kAudioObjectPropertyElementMaster };

-  AudioObjectGetPropertyData( inDevice, &property, 0, NULL, &dataSize, rate );

-  return kAudioHardwareNoError;

-}

-

-bool RtApiCore :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                   unsigned int firstChannel, unsigned int sampleRate,

-                                   RtAudioFormat format, unsigned int *bufferSize,

-                                   RtAudio::StreamOptions *options )

-{

-  // Get device ID

-  unsigned int nDevices = getDeviceCount();

-  if ( nDevices == 0 ) {

-    // This should not happen because a check is made before this function is called.

-    errorText_ = "RtApiCore::probeDeviceOpen: no devices found!";

-    return FAILURE;

-  }

-

-  if ( device >= nDevices ) {

-    // This should not happen because a check is made before this function is called.

-    errorText_ = "RtApiCore::probeDeviceOpen: device ID is invalid!";

-    return FAILURE;

-  }

-

-  AudioDeviceID deviceList[ nDevices ];

-  UInt32 dataSize = sizeof( AudioDeviceID ) * nDevices;

-  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,

-                                          kAudioObjectPropertyScopeGlobal,

-                                          kAudioObjectPropertyElementMaster };

-  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property,

-                                                0, NULL, &dataSize, (void *) &deviceList );

-  if ( result != noErr ) {

-    errorText_ = "RtApiCore::probeDeviceOpen: OS-X system error getting device IDs.";

-    return FAILURE;

-  }

-

-  AudioDeviceID id = deviceList[ device ];

-

-  // Setup for stream mode.

-  bool isInput = false;

-  if ( mode == INPUT ) {

-    isInput = true;

-    property.mScope = kAudioDevicePropertyScopeInput;

-  }

-  else

-    property.mScope = kAudioDevicePropertyScopeOutput;

-

-  // Get the stream "configuration".

-  AudioBufferList	*bufferList = nil;

-  dataSize = 0;

-  property.mSelector = kAudioDevicePropertyStreamConfiguration;

-  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );

-  if ( result != noErr || dataSize == 0 ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream configuration info for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Allocate the AudioBufferList.

-  bufferList = (AudioBufferList *) malloc( dataSize );

-  if ( bufferList == NULL ) {

-    errorText_ = "RtApiCore::probeDeviceOpen: memory error allocating AudioBufferList.";

-    return FAILURE;

-  }

-

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );

-  if (result != noErr || dataSize == 0) {

-    free( bufferList );

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream configuration for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Search for one or more streams that contain the desired number of

-  // channels. CoreAudio devices can have an arbitrary number of

-  // streams and each stream can have an arbitrary number of channels.

-  // For each stream, a single buffer of interleaved samples is

-  // provided.  RtAudio prefers the use of one stream of interleaved

-  // data or multiple consecutive single-channel streams.  However, we

-  // now support multiple consecutive multi-channel streams of

-  // interleaved data as well.

-  UInt32 iStream, offsetCounter = firstChannel;

-  UInt32 nStreams = bufferList->mNumberBuffers;

-  bool monoMode = false;

-  bool foundStream = false;

-

-  // First check that the device supports the requested number of

-  // channels.

-  UInt32 deviceChannels = 0;

-  for ( iStream=0; iStream<nStreams; iStream++ )

-    deviceChannels += bufferList->mBuffers[iStream].mNumberChannels;

-

-  if ( deviceChannels < ( channels + firstChannel ) ) {

-    free( bufferList );

-    errorStream_ << "RtApiCore::probeDeviceOpen: the device (" << device << ") does not support the requested channel count.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Look for a single stream meeting our needs.

-  UInt32 firstStream, streamCount = 1, streamChannels = 0, channelOffset = 0;

-  for ( iStream=0; iStream<nStreams; iStream++ ) {

-    streamChannels = bufferList->mBuffers[iStream].mNumberChannels;

-    if ( streamChannels >= channels + offsetCounter ) {

-      firstStream = iStream;

-      channelOffset = offsetCounter;

-      foundStream = true;

-      break;

-    }

-    if ( streamChannels > offsetCounter ) break;

-    offsetCounter -= streamChannels;

-  }

-

-  // If we didn't find a single stream above, then we should be able

-  // to meet the channel specification with multiple streams.

-  if ( foundStream == false ) {

-    monoMode = true;

-    offsetCounter = firstChannel;

-    for ( iStream=0; iStream<nStreams; iStream++ ) {

-      streamChannels = bufferList->mBuffers[iStream].mNumberChannels;

-      if ( streamChannels > offsetCounter ) break;

-      offsetCounter -= streamChannels;

-    }

-

-    firstStream = iStream;

-    channelOffset = offsetCounter;

-    Int32 channelCounter = channels + offsetCounter - streamChannels;

-

-    if ( streamChannels > 1 ) monoMode = false;

-    while ( channelCounter > 0 ) {

-      streamChannels = bufferList->mBuffers[++iStream].mNumberChannels;

-      if ( streamChannels > 1 ) monoMode = false;

-      channelCounter -= streamChannels;

-      streamCount++;

-    }

-  }

-

-  free( bufferList );

-

-  // Determine the buffer size.

-  AudioValueRange	bufferRange;

-  dataSize = sizeof( AudioValueRange );

-  property.mSelector = kAudioDevicePropertyBufferFrameSizeRange;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &bufferRange );

-

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting buffer size range for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  if ( bufferRange.mMinimum > *bufferSize ) *bufferSize = (unsigned long) bufferRange.mMinimum;

-  else if ( bufferRange.mMaximum < *bufferSize ) *bufferSize = (unsigned long) bufferRange.mMaximum;

-  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) *bufferSize = (unsigned long) bufferRange.mMinimum;

-

-  // Set the buffer size.  For multiple streams, I'm assuming we only

-  // need to make this setting for the master channel.

-  UInt32 theSize = (UInt32) *bufferSize;

-  dataSize = sizeof( UInt32 );

-  property.mSelector = kAudioDevicePropertyBufferFrameSize;

-  result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &theSize );

-

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting the buffer size for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // If attempting to setup a duplex stream, the bufferSize parameter

-  // MUST be the same in both directions!

-  *bufferSize = theSize;

-  if ( stream_.mode == OUTPUT && mode == INPUT && *bufferSize != stream_.bufferSize ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error setting buffer size for duplex stream on device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  stream_.bufferSize = *bufferSize;

-  stream_.nBuffers = 1;

-

-  // Try to set "hog" mode ... it's not clear to me this is working.

-  if ( options && options->flags & RTAUDIO_HOG_DEVICE ) {

-    pid_t hog_pid;

-    dataSize = sizeof( hog_pid );

-    property.mSelector = kAudioDevicePropertyHogMode;

-    result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &hog_pid );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting 'hog' state!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    if ( hog_pid != getpid() ) {

-      hog_pid = getpid();

-      result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &hog_pid );

-      if ( result != noErr ) {

-        errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting 'hog' state!";

-        errorText_ = errorStream_.str();

-        return FAILURE;

-      }

-    }

-  }

-

-  // Check and if necessary, change the sample rate for the device.

-  Float64 nominalRate;

-  dataSize = sizeof( Float64 );

-  property.mSelector = kAudioDevicePropertyNominalSampleRate;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &nominalRate );

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting current sample rate.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Only change the sample rate if off by more than 1 Hz.

-  if ( fabs( nominalRate - (double)sampleRate ) > 1.0 ) {

-

-    // Set a property listener for the sample rate change

-    Float64 reportedRate = 0.0;

-    AudioObjectPropertyAddress tmp = { kAudioDevicePropertyNominalSampleRate, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };

-    result = AudioObjectAddPropertyListener( id, &tmp, rateListener, (void *) &reportedRate );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate property listener for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    nominalRate = (Float64) sampleRate;

-    result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &nominalRate );

-    if ( result != noErr ) {

-      AudioObjectRemovePropertyListener( id, &tmp, rateListener, (void *) &reportedRate );

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Now wait until the reported nominal rate is what we just set.

-    UInt32 microCounter = 0;

-    while ( reportedRate != nominalRate ) {

-      microCounter += 5000;

-      if ( microCounter > 5000000 ) break;

-      usleep( 5000 );

-    }

-

-    // Remove the property listener.

-    AudioObjectRemovePropertyListener( id, &tmp, rateListener, (void *) &reportedRate );

-

-    if ( microCounter > 5000000 ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: timeout waiting for sample rate update for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-

-  // Now set the stream format for all streams.  Also, check the

-  // physical format of the device and change that if necessary.

-  AudioStreamBasicDescription	description;

-  dataSize = sizeof( AudioStreamBasicDescription );

-  property.mSelector = kAudioStreamPropertyVirtualFormat;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &description );

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream format for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Set the sample rate and data format id.  However, only make the

-  // change if the sample rate is not within 1.0 of the desired

-  // rate and the format is not linear pcm.

-  bool updateFormat = false;

-  if ( fabs( description.mSampleRate - (Float64)sampleRate ) > 1.0 ) {

-    description.mSampleRate = (Float64) sampleRate;

-    updateFormat = true;

-  }

-

-  if ( description.mFormatID != kAudioFormatLinearPCM ) {

-    description.mFormatID = kAudioFormatLinearPCM;

-    updateFormat = true;

-  }

-

-  if ( updateFormat ) {

-    result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &description );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate or data format for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-

-  // Now check the physical format.

-  property.mSelector = kAudioStreamPropertyPhysicalFormat;

-  result = AudioObjectGetPropertyData( id, &property, 0, NULL,  &dataSize, &description );

-  if ( result != noErr ) {

-    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream physical format for device (" << device << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  //std::cout << "Current physical stream format:" << std::endl;

-  //std::cout << "   mBitsPerChan = " << description.mBitsPerChannel << std::endl;

-  //std::cout << "   aligned high = " << (description.mFormatFlags & kAudioFormatFlagIsAlignedHigh) << ", isPacked = " << (description.mFormatFlags & kAudioFormatFlagIsPacked) << std::endl;

-  //std::cout << "   bytesPerFrame = " << description.mBytesPerFrame << std::endl;

-  //std::cout << "   sample rate = " << description.mSampleRate << std::endl;

-

-  if ( description.mFormatID != kAudioFormatLinearPCM || description.mBitsPerChannel < 16 ) {

-    description.mFormatID = kAudioFormatLinearPCM;

-    //description.mSampleRate = (Float64) sampleRate;

-    AudioStreamBasicDescription	testDescription = description;

-    UInt32 formatFlags;

-

-    // We'll try higher bit rates first and then work our way down.

-    std::vector< std::pair<UInt32, UInt32>  > physicalFormats;

-    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsFloat) & ~kLinearPCMFormatFlagIsSignedInteger;

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 32, formatFlags ) );

-    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked) & ~kLinearPCMFormatFlagIsFloat;

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 32, formatFlags ) );

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 24, formatFlags ) );   // 24-bit packed

-    formatFlags &= ~( kAudioFormatFlagIsPacked | kAudioFormatFlagIsAlignedHigh );

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 24.2, formatFlags ) ); // 24-bit in 4 bytes, aligned low

-    formatFlags |= kAudioFormatFlagIsAlignedHigh;

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 24.4, formatFlags ) ); // 24-bit in 4 bytes, aligned high

-    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked) & ~kLinearPCMFormatFlagIsFloat;

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 16, formatFlags ) );

-    physicalFormats.push_back( std::pair<Float32, UInt32>( 8, formatFlags ) );

-

-    bool setPhysicalFormat = false;

-    for( unsigned int i=0; i<physicalFormats.size(); i++ ) {

-      testDescription = description;

-      testDescription.mBitsPerChannel = (UInt32) physicalFormats[i].first;

-      testDescription.mFormatFlags = physicalFormats[i].second;

-      if ( (24 == (UInt32)physicalFormats[i].first) && ~( physicalFormats[i].second & kAudioFormatFlagIsPacked ) )

-        testDescription.mBytesPerFrame =  4 * testDescription.mChannelsPerFrame;

-      else

-        testDescription.mBytesPerFrame =  testDescription.mBitsPerChannel/8 * testDescription.mChannelsPerFrame;

-      testDescription.mBytesPerPacket = testDescription.mBytesPerFrame * testDescription.mFramesPerPacket;

-      result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &testDescription );

-      if ( result == noErr ) {

-        setPhysicalFormat = true;

-        //std::cout << "Updated physical stream format:" << std::endl;

-        //std::cout << "   mBitsPerChan = " << testDescription.mBitsPerChannel << std::endl;

-        //std::cout << "   aligned high = " << (testDescription.mFormatFlags & kAudioFormatFlagIsAlignedHigh) << ", isPacked = " << (testDescription.mFormatFlags & kAudioFormatFlagIsPacked) << std::endl;

-        //std::cout << "   bytesPerFrame = " << testDescription.mBytesPerFrame << std::endl;

-        //std::cout << "   sample rate = " << testDescription.mSampleRate << std::endl;

-        break;

-      }

-    }

-

-    if ( !setPhysicalFormat ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting physical data format for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  } // done setting virtual/physical formats.

-

-  // Get the stream / device latency.

-  UInt32 latency;

-  dataSize = sizeof( UInt32 );

-  property.mSelector = kAudioDevicePropertyLatency;

-  if ( AudioObjectHasProperty( id, &property ) == true ) {

-    result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &latency );

-    if ( result == kAudioHardwareNoError ) stream_.latency[ mode ] = latency;

-    else {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting device latency for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      error( RtAudioError::WARNING );

-    }

-  }

-

-  // Byte-swapping: According to AudioHardware.h, the stream data will

-  // always be presented in native-endian format, so we should never

-  // need to byte swap.

-  stream_.doByteSwap[mode] = false;

-

-  // From the CoreAudio documentation, PCM data must be supplied as

-  // 32-bit floats.

-  stream_.userFormat = format;

-  stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;

-

-  if ( streamCount == 1 )

-    stream_.nDeviceChannels[mode] = description.mChannelsPerFrame;

-  else // multiple streams

-    stream_.nDeviceChannels[mode] = channels;

-  stream_.nUserChannels[mode] = channels;

-  stream_.channelOffset[mode] = channelOffset;  // offset within a CoreAudio stream

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;

-  else stream_.userInterleaved = true;

-  stream_.deviceInterleaved[mode] = true;

-  if ( monoMode == true ) stream_.deviceInterleaved[mode] = false;

-

-  // Set flags for buffer conversion.

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( streamCount == 1 ) {

-    if ( stream_.nUserChannels[mode] > 1 &&

-         stream_.userInterleaved != stream_.deviceInterleaved[mode] )

-      stream_.doConvertBuffer[mode] = true;

-  }

-  else if ( monoMode && stream_.userInterleaved )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate our CoreHandle structure for the stream.

-  CoreHandle *handle = 0;

-  if ( stream_.apiHandle == 0 ) {

-    try {

-      handle = new CoreHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiCore::probeDeviceOpen: error allocating CoreHandle memory.";

-      goto error;

-    }

-

-    if ( pthread_cond_init( &handle->condition, NULL ) ) {

-      errorText_ = "RtApiCore::probeDeviceOpen: error initializing pthread condition variable.";

-      goto error;

-    }

-    stream_.apiHandle = (void *) handle;

-  }

-  else

-    handle = (CoreHandle *) stream_.apiHandle;

-  handle->iStream[mode] = firstStream;

-  handle->nStreams[mode] = streamCount;

-  handle->id[mode] = id;

-

-  // Allocate necessary internal buffers.

-  unsigned long bufferBytes;

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  //  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  stream_.userBuffer[mode] = (char *) malloc( bufferBytes * sizeof(char) );

-  memset( stream_.userBuffer[mode], 0, bufferBytes * sizeof(char) );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiCore::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  // If possible, we will make use of the CoreAudio stream buffers as

-  // "device buffers".  However, we can't do this if using multiple

-  // streams.

-  if ( stream_.doConvertBuffer[mode] && handle->nStreams[mode] > 1 ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( mode == INPUT ) {

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-        if ( bufferBytes <= bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiCore::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  stream_.sampleRate = sampleRate;

-  stream_.device[mode] = device;

-  stream_.state = STREAM_STOPPED;

-  stream_.callbackInfo.object = (void *) this;

-

-  // Setup the buffer conversion information structure.

-  if ( stream_.doConvertBuffer[mode] ) {

-    if ( streamCount > 1 ) setConvertInfo( mode, 0 );

-    else setConvertInfo( mode, channelOffset );

-  }

-

-  if ( mode == INPUT && stream_.mode == OUTPUT && stream_.device[0] == device )

-    // Only one callback procedure per device.

-    stream_.mode = DUPLEX;

-  else {

-#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )

-    result = AudioDeviceCreateIOProcID( id, callbackHandler, (void *) &stream_.callbackInfo, &handle->procId[mode] );

-#else

-    // deprecated in favor of AudioDeviceCreateIOProcID()

-    result = AudioDeviceAddIOProc( id, callbackHandler, (void *) &stream_.callbackInfo );

-#endif

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::probeDeviceOpen: system error setting callback for device (" << device << ").";

-      errorText_ = errorStream_.str();

-      goto error;

-    }

-    if ( stream_.mode == OUTPUT && mode == INPUT )

-      stream_.mode = DUPLEX;

-    else

-      stream_.mode = mode;

-  }

-

-  // Setup the device property listener for over/underload.

-  property.mSelector = kAudioDeviceProcessorOverload;

-  property.mScope = kAudioObjectPropertyScopeGlobal;

-  result = AudioObjectAddPropertyListener( id, &property, xrunListener, (void *) handle );

-

-  return SUCCESS;

-

- error:

-  if ( handle ) {

-    pthread_cond_destroy( &handle->condition );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.state = STREAM_CLOSED;

-  return FAILURE;

-}

-

-void RtApiCore :: closeStream( void )

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiCore::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    if (handle) {

-      AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,

-        kAudioObjectPropertyScopeGlobal,

-        kAudioObjectPropertyElementMaster };

-

-      property.mSelector = kAudioDeviceProcessorOverload;

-      property.mScope = kAudioObjectPropertyScopeGlobal;

-      if (AudioObjectRemovePropertyListener( handle->id[0], &property, xrunListener, (void *) handle ) != noErr) {

-        errorText_ = "RtApiCore::closeStream(): error removing property listener!";

-        error( RtAudioError::WARNING );

-      }

-    }

-    if ( stream_.state == STREAM_RUNNING )

-      AudioDeviceStop( handle->id[0], callbackHandler );

-#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )

-    AudioDeviceDestroyIOProcID( handle->id[0], handle->procId[0] );

-#else

-    // deprecated in favor of AudioDeviceDestroyIOProcID()

-    AudioDeviceRemoveIOProc( handle->id[0], callbackHandler );

-#endif

-  }

-

-  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {

-    if (handle) {

-      AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,

-        kAudioObjectPropertyScopeGlobal,

-        kAudioObjectPropertyElementMaster };

-

-      property.mSelector = kAudioDeviceProcessorOverload;

-      property.mScope = kAudioObjectPropertyScopeGlobal;

-      if (AudioObjectRemovePropertyListener( handle->id[1], &property, xrunListener, (void *) handle ) != noErr) {

-        errorText_ = "RtApiCore::closeStream(): error removing property listener!";

-        error( RtAudioError::WARNING );

-      }

-    }

-    if ( stream_.state == STREAM_RUNNING )

-      AudioDeviceStop( handle->id[1], callbackHandler );

-#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )

-    AudioDeviceDestroyIOProcID( handle->id[1], handle->procId[1] );

-#else

-    // deprecated in favor of AudioDeviceDestroyIOProcID()

-    AudioDeviceRemoveIOProc( handle->id[1], callbackHandler );

-#endif

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  // Destroy pthread condition variable.

-  pthread_cond_destroy( &handle->condition );

-  delete handle;

-  stream_.apiHandle = 0;

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-void RtApiCore :: startStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiCore::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  OSStatus result = noErr;

-  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    result = AudioDeviceStart( handle->id[0], callbackHandler );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::startStream: system error (" << getErrorCode( result ) << ") starting callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  if ( stream_.mode == INPUT ||

-       ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {

-

-    result = AudioDeviceStart( handle->id[1], callbackHandler );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::startStream: system error starting input callback procedure on device (" << stream_.device[1] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  handle->drainCounter = 0;

-  handle->internalDrain = false;

-  stream_.state = STREAM_RUNNING;

-

- unlock:

-  if ( result == noErr ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiCore :: stopStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiCore::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  OSStatus result = noErr;

-  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    if ( handle->drainCounter == 0 ) {

-      handle->drainCounter = 2;

-      pthread_cond_wait( &handle->condition, &stream_.mutex ); // block until signaled

-    }

-

-    result = AudioDeviceStop( handle->id[0], callbackHandler );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::stopStream: system error (" << getErrorCode( result ) << ") stopping callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {

-

-    result = AudioDeviceStop( handle->id[1], callbackHandler );

-    if ( result != noErr ) {

-      errorStream_ << "RtApiCore::stopStream: system error (" << getErrorCode( result ) << ") stopping input callback procedure on device (" << stream_.device[1] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  stream_.state = STREAM_STOPPED;

-

- unlock:

-  if ( result == noErr ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiCore :: abortStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiCore::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;

-  handle->drainCounter = 2;

-

-  stopStream();

-}

-

-// This function will be called by a spawned thread when the user

-// callback function signals that the stream should be stopped or

-// aborted.  It is better to handle it this way because the

-// callbackEvent() function probably should return before the AudioDeviceStop()

-// function is called.

-static void *coreStopStream( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiCore *object = (RtApiCore *) info->object;

-

-  object->stopStream();

-  pthread_exit( NULL );

-}

-

-bool RtApiCore :: callbackEvent( AudioDeviceID deviceId,

-                                 const AudioBufferList *inBufferList,

-                                 const AudioBufferList *outBufferList )

-{

-  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiCore::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return FAILURE;

-  }

-

-  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;

-  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;

-

-  // Check if we were draining the stream and signal is finished.

-  if ( handle->drainCounter > 3 ) {

-    ThreadHandle threadId;

-

-    stream_.state = STREAM_STOPPING;

-    if ( handle->internalDrain == true )

-      pthread_create( &threadId, NULL, coreStopStream, info );

-    else // external call to stopStream()

-      pthread_cond_signal( &handle->condition );

-    return SUCCESS;

-  }

-

-  AudioDeviceID outputDevice = handle->id[0];

-

-  // Invoke user callback to get fresh output data UNLESS we are

-  // draining stream or duplex mode AND the input/output devices are

-  // different AND this function is called for the input device.

-  if ( handle->drainCounter == 0 && ( stream_.mode != DUPLEX || deviceId == outputDevice ) ) {

-    RtAudioCallback callback = (RtAudioCallback) info->callback;

-    double streamTime = getStreamTime();

-    RtAudioStreamStatus status = 0;

-    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {

-      status |= RTAUDIO_OUTPUT_UNDERFLOW;

-      handle->xrun[0] = false;

-    }

-    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {

-      status |= RTAUDIO_INPUT_OVERFLOW;

-      handle->xrun[1] = false;

-    }

-

-    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                                  stream_.bufferSize, streamTime, status, info->userData );

-    if ( cbReturnValue == 2 ) {

-      stream_.state = STREAM_STOPPING;

-      handle->drainCounter = 2;

-      abortStream();

-      return SUCCESS;

-    }

-    else if ( cbReturnValue == 1 ) {

-      handle->drainCounter = 1;

-      handle->internalDrain = true;

-    }

-  }

-

-  if ( stream_.mode == OUTPUT || ( stream_.mode == DUPLEX && deviceId == outputDevice ) ) {

-

-    if ( handle->drainCounter > 1 ) { // write zeros to the output stream

-

-      if ( handle->nStreams[0] == 1 ) {

-        memset( outBufferList->mBuffers[handle->iStream[0]].mData,

-                0,

-                outBufferList->mBuffers[handle->iStream[0]].mDataByteSize );

-      }

-      else { // fill multiple streams with zeros

-        for ( unsigned int i=0; i<handle->nStreams[0]; i++ ) {

-          memset( outBufferList->mBuffers[handle->iStream[0]+i].mData,

-                  0,

-                  outBufferList->mBuffers[handle->iStream[0]+i].mDataByteSize );

-        }

-      }

-    }

-    else if ( handle->nStreams[0] == 1 ) {

-      if ( stream_.doConvertBuffer[0] ) { // convert directly to CoreAudio stream buffer

-        convertBuffer( (char *) outBufferList->mBuffers[handle->iStream[0]].mData,

-                       stream_.userBuffer[0], stream_.convertInfo[0] );

-      }

-      else { // copy from user buffer

-        memcpy( outBufferList->mBuffers[handle->iStream[0]].mData,

-                stream_.userBuffer[0],

-                outBufferList->mBuffers[handle->iStream[0]].mDataByteSize );

-      }

-    }

-    else { // fill multiple streams

-      Float32 *inBuffer = (Float32 *) stream_.userBuffer[0];

-      if ( stream_.doConvertBuffer[0] ) {

-        convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-        inBuffer = (Float32 *) stream_.deviceBuffer;

-      }

-

-      if ( stream_.deviceInterleaved[0] == false ) { // mono mode

-        UInt32 bufferBytes = outBufferList->mBuffers[handle->iStream[0]].mDataByteSize;

-        for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {

-          memcpy( outBufferList->mBuffers[handle->iStream[0]+i].mData,

-                  (void *)&inBuffer[i*stream_.bufferSize], bufferBytes );

-        }

-      }

-      else { // fill multiple multi-channel streams with interleaved data

-        UInt32 streamChannels, channelsLeft, inJump, outJump, inOffset;

-        Float32 *out, *in;

-

-        bool inInterleaved = ( stream_.userInterleaved ) ? true : false;

-        UInt32 inChannels = stream_.nUserChannels[0];

-        if ( stream_.doConvertBuffer[0] ) {

-          inInterleaved = true; // device buffer will always be interleaved for nStreams > 1 and not mono mode

-          inChannels = stream_.nDeviceChannels[0];

-        }

-

-        if ( inInterleaved ) inOffset = 1;

-        else inOffset = stream_.bufferSize;

-

-        channelsLeft = inChannels;

-        for ( unsigned int i=0; i<handle->nStreams[0]; i++ ) {

-          in = inBuffer;

-          out = (Float32 *) outBufferList->mBuffers[handle->iStream[0]+i].mData;

-          streamChannels = outBufferList->mBuffers[handle->iStream[0]+i].mNumberChannels;

-

-          outJump = 0;

-          // Account for possible channel offset in first stream

-          if ( i == 0 && stream_.channelOffset[0] > 0 ) {

-            streamChannels -= stream_.channelOffset[0];

-            outJump = stream_.channelOffset[0];

-            out += outJump;

-          }

-

-          // Account for possible unfilled channels at end of the last stream

-          if ( streamChannels > channelsLeft ) {

-            outJump = streamChannels - channelsLeft;

-            streamChannels = channelsLeft;

-          }

-

-          // Determine input buffer offsets and skips

-          if ( inInterleaved ) {

-            inJump = inChannels;

-            in += inChannels - channelsLeft;

-          }

-          else {

-            inJump = 1;

-            in += (inChannels - channelsLeft) * inOffset;

-          }

-

-          for ( unsigned int i=0; i<stream_.bufferSize; i++ ) {

-            for ( unsigned int j=0; j<streamChannels; j++ ) {

-              *out++ = in[j*inOffset];

-            }

-            out += outJump;

-            in += inJump;

-          }

-          channelsLeft -= streamChannels;

-        }

-      }

-    }

-  }

-

-  // Don't bother draining input

-  if ( handle->drainCounter ) {

-    handle->drainCounter++;

-    goto unlock;

-  }

-

-  AudioDeviceID inputDevice;

-  inputDevice = handle->id[1];

-  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && deviceId == inputDevice ) ) {

-

-    if ( handle->nStreams[1] == 1 ) {

-      if ( stream_.doConvertBuffer[1] ) { // convert directly from CoreAudio stream buffer

-        convertBuffer( stream_.userBuffer[1],

-                       (char *) inBufferList->mBuffers[handle->iStream[1]].mData,

-                       stream_.convertInfo[1] );

-      }

-      else { // copy to user buffer

-        memcpy( stream_.userBuffer[1],

-                inBufferList->mBuffers[handle->iStream[1]].mData,

-                inBufferList->mBuffers[handle->iStream[1]].mDataByteSize );

-      }

-    }

-    else { // read from multiple streams

-      Float32 *outBuffer = (Float32 *) stream_.userBuffer[1];

-      if ( stream_.doConvertBuffer[1] ) outBuffer = (Float32 *) stream_.deviceBuffer;

-

-      if ( stream_.deviceInterleaved[1] == false ) { // mono mode

-        UInt32 bufferBytes = inBufferList->mBuffers[handle->iStream[1]].mDataByteSize;

-        for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {

-          memcpy( (void *)&outBuffer[i*stream_.bufferSize],

-                  inBufferList->mBuffers[handle->iStream[1]+i].mData, bufferBytes );

-        }

-      }

-      else { // read from multiple multi-channel streams

-        UInt32 streamChannels, channelsLeft, inJump, outJump, outOffset;

-        Float32 *out, *in;

-

-        bool outInterleaved = ( stream_.userInterleaved ) ? true : false;

-        UInt32 outChannels = stream_.nUserChannels[1];

-        if ( stream_.doConvertBuffer[1] ) {

-          outInterleaved = true; // device buffer will always be interleaved for nStreams > 1 and not mono mode

-          outChannels = stream_.nDeviceChannels[1];

-        }

-

-        if ( outInterleaved ) outOffset = 1;

-        else outOffset = stream_.bufferSize;

-

-        channelsLeft = outChannels;

-        for ( unsigned int i=0; i<handle->nStreams[1]; i++ ) {

-          out = outBuffer;

-          in = (Float32 *) inBufferList->mBuffers[handle->iStream[1]+i].mData;

-          streamChannels = inBufferList->mBuffers[handle->iStream[1]+i].mNumberChannels;

-

-          inJump = 0;

-          // Account for possible channel offset in first stream

-          if ( i == 0 && stream_.channelOffset[1] > 0 ) {

-            streamChannels -= stream_.channelOffset[1];

-            inJump = stream_.channelOffset[1];

-            in += inJump;

-          }

-

-          // Account for possible unread channels at end of the last stream

-          if ( streamChannels > channelsLeft ) {

-            inJump = streamChannels - channelsLeft;

-            streamChannels = channelsLeft;

-          }

-

-          // Determine output buffer offsets and skips

-          if ( outInterleaved ) {

-            outJump = outChannels;

-            out += outChannels - channelsLeft;

-          }

-          else {

-            outJump = 1;

-            out += (outChannels - channelsLeft) * outOffset;

-          }

-

-          for ( unsigned int i=0; i<stream_.bufferSize; i++ ) {

-            for ( unsigned int j=0; j<streamChannels; j++ ) {

-              out[j*outOffset] = *in++;

-            }

-            out += outJump;

-            in += inJump;

-          }

-          channelsLeft -= streamChannels;

-        }

-      }

-      

-      if ( stream_.doConvertBuffer[1] ) { // convert from our internal "device" buffer

-        convertBuffer( stream_.userBuffer[1],

-                       stream_.deviceBuffer,

-                       stream_.convertInfo[1] );

-      }

-    }

-  }

-

- unlock:

-  //MUTEX_UNLOCK( &stream_.mutex );

-

-  RtApi::tickStreamTime();

-  return SUCCESS;

-}

-

-const char* RtApiCore :: getErrorCode( OSStatus code )

-{

-  switch( code ) {

-

-  case kAudioHardwareNotRunningError:

-    return "kAudioHardwareNotRunningError";

-

-  case kAudioHardwareUnspecifiedError:

-    return "kAudioHardwareUnspecifiedError";

-

-  case kAudioHardwareUnknownPropertyError:

-    return "kAudioHardwareUnknownPropertyError";

-

-  case kAudioHardwareBadPropertySizeError:

-    return "kAudioHardwareBadPropertySizeError";

-

-  case kAudioHardwareIllegalOperationError:

-    return "kAudioHardwareIllegalOperationError";

-

-  case kAudioHardwareBadObjectError:

-    return "kAudioHardwareBadObjectError";

-

-  case kAudioHardwareBadDeviceError:

-    return "kAudioHardwareBadDeviceError";

-

-  case kAudioHardwareBadStreamError:

-    return "kAudioHardwareBadStreamError";

-

-  case kAudioHardwareUnsupportedOperationError:

-    return "kAudioHardwareUnsupportedOperationError";

-

-  case kAudioDeviceUnsupportedFormatError:

-    return "kAudioDeviceUnsupportedFormatError";

-

-  case kAudioDevicePermissionsError:

-    return "kAudioDevicePermissionsError";

-

-  default:

-    return "CoreAudio unknown error";

-  }

-}

-

-  //******************** End of __MACOSX_CORE__ *********************//

-#endif

-

-#if defined(__UNIX_JACK__)

-

-// JACK is a low-latency audio server, originally written for the

-// GNU/Linux operating system and now also ported to OS-X. It can

-// connect a number of different applications to an audio device, as

-// well as allowing them to share audio between themselves.

-//

-// When using JACK with RtAudio, "devices" refer to JACK clients that

-// have ports connected to the server.  The JACK server is typically

-// started in a terminal as follows:

-//

-// .jackd -d alsa -d hw:0

-//

-// or through an interface program such as qjackctl.  Many of the

-// parameters normally set for a stream are fixed by the JACK server

-// and can be specified when the JACK server is started.  In

-// particular,

-//

-// .jackd -d alsa -d hw:0 -r 44100 -p 512 -n 4

-//

-// specifies a sample rate of 44100 Hz, a buffer size of 512 sample

-// frames, and number of buffers = 4.  Once the server is running, it

-// is not possible to override these values.  If the values are not

-// specified in the command-line, the JACK server uses default values.

-//

-// The JACK server does not have to be running when an instance of

-// RtApiJack is created, though the function getDeviceCount() will

-// report 0 devices found until JACK has been started.  When no

-// devices are available (i.e., the JACK server is not running), a

-// stream cannot be opened.

-

-#include <jack/jack.h>

-#include <unistd.h>

-#include <cstdio>

-

-// A structure to hold various information related to the Jack API

-// implementation.

-struct JackHandle {

-  jack_client_t *client;

-  jack_port_t **ports[2];

-  std::string deviceName[2];

-  bool xrun[2];

-  pthread_cond_t condition;

-  int drainCounter;       // Tracks callback counts when draining

-  bool internalDrain;     // Indicates if stop is initiated from callback or not.

-

-  JackHandle()

-    :client(0), drainCounter(0), internalDrain(false) { ports[0] = 0; ports[1] = 0; xrun[0] = false; xrun[1] = false; }

-};

-

-static void jackSilentError( const char * ) {};

-

-RtApiJack :: RtApiJack()

-{

-  // Nothing to do here.

-#if !defined(__RTAUDIO_DEBUG__)

-  // Turn off Jack's internal error reporting.

-  jack_set_error_function( &jackSilentError );

-#endif

-}

-

-RtApiJack :: ~RtApiJack()

-{

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-}

-

-unsigned int RtApiJack :: getDeviceCount( void )

-{

-  // See if we can become a jack client.

-  jack_options_t options = (jack_options_t) ( JackNoStartServer ); //JackNullOption;

-  jack_status_t *status = NULL;

-  jack_client_t *client = jack_client_open( "RtApiJackCount", options, status );

-  if ( client == 0 ) return 0;

-

-  const char **ports;

-  std::string port, previousPort;

-  unsigned int nChannels = 0, nDevices = 0;

-  ports = jack_get_ports( client, NULL, NULL, 0 );

-  if ( ports ) {

-    // Parse the port names up to the first colon (:).

-    size_t iColon = 0;

-    do {

-      port = (char *) ports[ nChannels ];

-      iColon = port.find(":");

-      if ( iColon != std::string::npos ) {

-        port = port.substr( 0, iColon + 1 );

-        if ( port != previousPort ) {

-          nDevices++;

-          previousPort = port;

-        }

-      }

-    } while ( ports[++nChannels] );

-    free( ports );

-  }

-

-  jack_client_close( client );

-  return nDevices;

-}

-

-RtAudio::DeviceInfo RtApiJack :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  jack_options_t options = (jack_options_t) ( JackNoStartServer ); //JackNullOption

-  jack_status_t *status = NULL;

-  jack_client_t *client = jack_client_open( "RtApiJackInfo", options, status );

-  if ( client == 0 ) {

-    errorText_ = "RtApiJack::getDeviceInfo: Jack server not found or connection error!";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  const char **ports;

-  std::string port, previousPort;

-  unsigned int nPorts = 0, nDevices = 0;

-  ports = jack_get_ports( client, NULL, NULL, 0 );

-  if ( ports ) {

-    // Parse the port names up to the first colon (:).

-    size_t iColon = 0;

-    do {

-      port = (char *) ports[ nPorts ];

-      iColon = port.find(":");

-      if ( iColon != std::string::npos ) {

-        port = port.substr( 0, iColon );

-        if ( port != previousPort ) {

-          if ( nDevices == device ) info.name = port;

-          nDevices++;

-          previousPort = port;

-        }

-      }

-    } while ( ports[++nPorts] );

-    free( ports );

-  }

-

-  if ( device >= nDevices ) {

-    jack_client_close( client );

-    errorText_ = "RtApiJack::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  // Get the current jack server sample rate.

-  info.sampleRates.clear();

-

-  info.preferredSampleRate = jack_get_sample_rate( client );

-  info.sampleRates.push_back( info.preferredSampleRate );

-

-  // Count the available ports containing the client name as device

-  // channels.  Jack "input ports" equal RtAudio output channels.

-  unsigned int nChannels = 0;

-  ports = jack_get_ports( client, info.name.c_str(), NULL, JackPortIsInput );

-  if ( ports ) {

-    while ( ports[ nChannels ] ) nChannels++;

-    free( ports );

-    info.outputChannels = nChannels;

-  }

-

-  // Jack "output ports" equal RtAudio input channels.

-  nChannels = 0;

-  ports = jack_get_ports( client, info.name.c_str(), NULL, JackPortIsOutput );

-  if ( ports ) {

-    while ( ports[ nChannels ] ) nChannels++;

-    free( ports );

-    info.inputChannels = nChannels;

-  }

-

-  if ( info.outputChannels == 0 && info.inputChannels == 0 ) {

-    jack_client_close(client);

-    errorText_ = "RtApiJack::getDeviceInfo: error determining Jack input/output channels!";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // If device opens for both playback and capture, we determine the channels.

-  if ( info.outputChannels > 0 && info.inputChannels > 0 )

-    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-

-  // Jack always uses 32-bit floats.

-  info.nativeFormats = RTAUDIO_FLOAT32;

-

-  // Jack doesn't provide default devices so we'll use the first available one.

-  if ( device == 0 && info.outputChannels > 0 )

-    info.isDefaultOutput = true;

-  if ( device == 0 && info.inputChannels > 0 )

-    info.isDefaultInput = true;

-

-  jack_client_close(client);

-  info.probed = true;

-  return info;

-}

-

-static int jackCallbackHandler( jack_nframes_t nframes, void *infoPointer )

-{

-  CallbackInfo *info = (CallbackInfo *) infoPointer;

-

-  RtApiJack *object = (RtApiJack *) info->object;

-  if ( object->callbackEvent( (unsigned long) nframes ) == false ) return 1;

-

-  return 0;

-}

-

-// This function will be called by a spawned thread when the Jack

-// server signals that it is shutting down.  It is necessary to handle

-// it this way because the jackShutdown() function must return before

-// the jack_deactivate() function (in closeStream()) will return.

-static void *jackCloseStream( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiJack *object = (RtApiJack *) info->object;

-

-  object->closeStream();

-

-  pthread_exit( NULL );

-}

-static void jackShutdown( void *infoPointer )

-{

-  CallbackInfo *info = (CallbackInfo *) infoPointer;

-  RtApiJack *object = (RtApiJack *) info->object;

-

-  // Check current stream state.  If stopped, then we'll assume this

-  // was called as a result of a call to RtApiJack::stopStream (the

-  // deactivation of a client handle causes this function to be called).

-  // If not, we'll assume the Jack server is shutting down or some

-  // other problem occurred and we should close the stream.

-  if ( object->isStreamRunning() == false ) return;

-

-  ThreadHandle threadId;

-  pthread_create( &threadId, NULL, jackCloseStream, info );

-  std::cerr << "\nRtApiJack: the Jack server is shutting down this client ... stream stopped and closed!!\n" << std::endl;

-}

-

-static int jackXrun( void *infoPointer )

-{

-  JackHandle *handle = (JackHandle *) infoPointer;

-

-  if ( handle->ports[0] ) handle->xrun[0] = true;

-  if ( handle->ports[1] ) handle->xrun[1] = true;

-

-  return 0;

-}

-

-bool RtApiJack :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                   unsigned int firstChannel, unsigned int sampleRate,

-                                   RtAudioFormat format, unsigned int *bufferSize,

-                                   RtAudio::StreamOptions *options )

-{

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-

-  // Look for jack server and try to become a client (only do once per stream).

-  jack_client_t *client = 0;

-  if ( mode == OUTPUT || ( mode == INPUT && stream_.mode != OUTPUT ) ) {

-    jack_options_t jackoptions = (jack_options_t) ( JackNoStartServer ); //JackNullOption;

-    jack_status_t *status = NULL;

-    if ( options && !options->streamName.empty() )

-      client = jack_client_open( options->streamName.c_str(), jackoptions, status );

-    else

-      client = jack_client_open( "RtApiJack", jackoptions, status );

-    if ( client == 0 ) {

-      errorText_ = "RtApiJack::probeDeviceOpen: Jack server not found or connection error!";

-      error( RtAudioError::WARNING );

-      return FAILURE;

-    }

-  }

-  else {

-    // The handle must have been created on an earlier pass.

-    client = handle->client;

-  }

-

-  const char **ports;

-  std::string port, previousPort, deviceName;

-  unsigned int nPorts = 0, nDevices = 0;

-  ports = jack_get_ports( client, NULL, NULL, 0 );

-  if ( ports ) {

-    // Parse the port names up to the first colon (:).

-    size_t iColon = 0;

-    do {

-      port = (char *) ports[ nPorts ];

-      iColon = port.find(":");

-      if ( iColon != std::string::npos ) {

-        port = port.substr( 0, iColon );

-        if ( port != previousPort ) {

-          if ( nDevices == device ) deviceName = port;

-          nDevices++;

-          previousPort = port;

-        }

-      }

-    } while ( ports[++nPorts] );

-    free( ports );

-  }

-

-  if ( device >= nDevices ) {

-    errorText_ = "RtApiJack::probeDeviceOpen: device ID is invalid!";

-    return FAILURE;

-  }

-

-  // Count the available ports containing the client name as device

-  // channels.  Jack "input ports" equal RtAudio output channels.

-  unsigned int nChannels = 0;

-  unsigned long flag = JackPortIsInput;

-  if ( mode == INPUT ) flag = JackPortIsOutput;

-  ports = jack_get_ports( client, deviceName.c_str(), NULL, flag );

-  if ( ports ) {

-    while ( ports[ nChannels ] ) nChannels++;

-    free( ports );

-  }

-

-  // Compare the jack ports for specified client to the requested number of channels.

-  if ( nChannels < (channels + firstChannel) ) {

-    errorStream_ << "RtApiJack::probeDeviceOpen: requested number of channels (" << channels << ") + offset (" << firstChannel << ") not found for specified device (" << device << ":" << deviceName << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Check the jack server sample rate.

-  unsigned int jackRate = jack_get_sample_rate( client );

-  if ( sampleRate != jackRate ) {

-    jack_client_close( client );

-    errorStream_ << "RtApiJack::probeDeviceOpen: the requested sample rate (" << sampleRate << ") is different than the JACK server rate (" << jackRate << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  stream_.sampleRate = jackRate;

-

-  // Get the latency of the JACK port.

-  ports = jack_get_ports( client, deviceName.c_str(), NULL, flag );

-  if ( ports[ firstChannel ] ) {

-    // Added by Ge Wang

-    jack_latency_callback_mode_t cbmode = (mode == INPUT ? JackCaptureLatency : JackPlaybackLatency);

-    // the range (usually the min and max are equal)

-    jack_latency_range_t latrange; latrange.min = latrange.max = 0;

-    // get the latency range

-    jack_port_get_latency_range( jack_port_by_name( client, ports[firstChannel] ), cbmode, &latrange );

-    // be optimistic, use the min!

-    stream_.latency[mode] = latrange.min;

-    //stream_.latency[mode] = jack_port_get_latency( jack_port_by_name( client, ports[ firstChannel ] ) );

-  }

-  free( ports );

-

-  // The jack server always uses 32-bit floating-point data.

-  stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;

-  stream_.userFormat = format;

-

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;

-  else stream_.userInterleaved = true;

-

-  // Jack always uses non-interleaved buffers.

-  stream_.deviceInterleaved[mode] = false;

-

-  // Jack always provides host byte-ordered data.

-  stream_.doByteSwap[mode] = false;

-

-  // Get the buffer size.  The buffer size and number of buffers

-  // (periods) is set when the jack server is started.

-  stream_.bufferSize = (int) jack_get_buffer_size( client );

-  *bufferSize = stream_.bufferSize;

-

-  stream_.nDeviceChannels[mode] = channels;

-  stream_.nUserChannels[mode] = channels;

-

-  // Set flags for buffer conversion.

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-       stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate our JackHandle structure for the stream.

-  if ( handle == 0 ) {

-    try {

-      handle = new JackHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiJack::probeDeviceOpen: error allocating JackHandle memory.";

-      goto error;

-    }

-

-    if ( pthread_cond_init(&handle->condition, NULL) ) {

-      errorText_ = "RtApiJack::probeDeviceOpen: error initializing pthread condition variable.";

-      goto error;

-    }

-    stream_.apiHandle = (void *) handle;

-    handle->client = client;

-  }

-  handle->deviceName[mode] = deviceName;

-

-  // Allocate necessary internal buffers.

-  unsigned long bufferBytes;

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiJack::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    if ( mode == OUTPUT )

-      bufferBytes = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-    else { // mode == INPUT

-      bufferBytes = stream_.nDeviceChannels[1] * formatBytes( stream_.deviceFormat[1] );

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes(stream_.deviceFormat[0]);

-        if ( bufferBytes < bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiJack::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  // Allocate memory for the Jack ports (channels) identifiers.

-  handle->ports[mode] = (jack_port_t **) malloc ( sizeof (jack_port_t *) * channels );

-  if ( handle->ports[mode] == NULL )  {

-    errorText_ = "RtApiJack::probeDeviceOpen: error allocating port memory.";

-    goto error;

-  }

-

-  stream_.device[mode] = device;

-  stream_.channelOffset[mode] = firstChannel;

-  stream_.state = STREAM_STOPPED;

-  stream_.callbackInfo.object = (void *) this;

-

-  if ( stream_.mode == OUTPUT && mode == INPUT )

-    // We had already set up the stream for output.

-    stream_.mode = DUPLEX;

-  else {

-    stream_.mode = mode;

-    jack_set_process_callback( handle->client, jackCallbackHandler, (void *) &stream_.callbackInfo );

-    jack_set_xrun_callback( handle->client, jackXrun, (void *) &handle );

-    jack_on_shutdown( handle->client, jackShutdown, (void *) &stream_.callbackInfo );

-  }

-

-  // Register our ports.

-  char label[64];

-  if ( mode == OUTPUT ) {

-    for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {

-      snprintf( label, 64, "outport %d", i );

-      handle->ports[0][i] = jack_port_register( handle->client, (const char *)label,

-                                                JACK_DEFAULT_AUDIO_TYPE, JackPortIsOutput, 0 );

-    }

-  }

-  else {

-    for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {

-      snprintf( label, 64, "inport %d", i );

-      handle->ports[1][i] = jack_port_register( handle->client, (const char *)label,

-                                                JACK_DEFAULT_AUDIO_TYPE, JackPortIsInput, 0 );

-    }

-  }

-

-  // Setup the buffer conversion information structure.  We don't use

-  // buffers to do channel offsets, so we override that parameter

-  // here.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, 0 );

-

-  return SUCCESS;

-

- error:

-  if ( handle ) {

-    pthread_cond_destroy( &handle->condition );

-    jack_client_close( handle->client );

-

-    if ( handle->ports[0] ) free( handle->ports[0] );

-    if ( handle->ports[1] ) free( handle->ports[1] );

-

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  return FAILURE;

-}

-

-void RtApiJack :: closeStream( void )

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiJack::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-  if ( handle ) {

-

-    if ( stream_.state == STREAM_RUNNING )

-      jack_deactivate( handle->client );

-

-    jack_client_close( handle->client );

-  }

-

-  if ( handle ) {

-    if ( handle->ports[0] ) free( handle->ports[0] );

-    if ( handle->ports[1] ) free( handle->ports[1] );

-    pthread_cond_destroy( &handle->condition );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-void RtApiJack :: startStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiJack::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-  int result = jack_activate( handle->client );

-  if ( result ) {

-    errorText_ = "RtApiJack::startStream(): unable to activate JACK client!";

-    goto unlock;

-  }

-

-  const char **ports;

-

-  // Get the list of available ports.

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    result = 1;

-    ports = jack_get_ports( handle->client, handle->deviceName[0].c_str(), NULL, JackPortIsInput);

-    if ( ports == NULL) {

-      errorText_ = "RtApiJack::startStream(): error determining available JACK input ports!";

-      goto unlock;

-    }

-

-    // Now make the port connections.  Since RtAudio wasn't designed to

-    // allow the user to select particular channels of a device, we'll

-    // just open the first "nChannels" ports with offset.

-    for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {

-      result = 1;

-      if ( ports[ stream_.channelOffset[0] + i ] )

-        result = jack_connect( handle->client, jack_port_name( handle->ports[0][i] ), ports[ stream_.channelOffset[0] + i ] );

-      if ( result ) {

-        free( ports );

-        errorText_ = "RtApiJack::startStream(): error connecting output ports!";

-        goto unlock;

-      }

-    }

-    free(ports);

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-    result = 1;

-    ports = jack_get_ports( handle->client, handle->deviceName[1].c_str(), NULL, JackPortIsOutput );

-    if ( ports == NULL) {

-      errorText_ = "RtApiJack::startStream(): error determining available JACK output ports!";

-      goto unlock;

-    }

-

-    // Now make the port connections.  See note above.

-    for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {

-      result = 1;

-      if ( ports[ stream_.channelOffset[1] + i ] )

-        result = jack_connect( handle->client, ports[ stream_.channelOffset[1] + i ], jack_port_name( handle->ports[1][i] ) );

-      if ( result ) {

-        free( ports );

-        errorText_ = "RtApiJack::startStream(): error connecting input ports!";

-        goto unlock;

-      }

-    }

-    free(ports);

-  }

-

-  handle->drainCounter = 0;

-  handle->internalDrain = false;

-  stream_.state = STREAM_RUNNING;

-

- unlock:

-  if ( result == 0 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiJack :: stopStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiJack::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    if ( handle->drainCounter == 0 ) {

-      handle->drainCounter = 2;

-      pthread_cond_wait( &handle->condition, &stream_.mutex ); // block until signaled

-    }

-  }

-

-  jack_deactivate( handle->client );

-  stream_.state = STREAM_STOPPED;

-}

-

-void RtApiJack :: abortStream( void )

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiJack::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-  handle->drainCounter = 2;

-

-  stopStream();

-}

-

-// This function will be called by a spawned thread when the user

-// callback function signals that the stream should be stopped or

-// aborted.  It is necessary to handle it this way because the

-// callbackEvent() function must return before the jack_deactivate()

-// function will return.

-static void *jackStopStream( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiJack *object = (RtApiJack *) info->object;

-

-  object->stopStream();

-  pthread_exit( NULL );

-}

-

-bool RtApiJack :: callbackEvent( unsigned long nframes )

-{

-  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiCore::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return FAILURE;

-  }

-  if ( stream_.bufferSize != nframes ) {

-    errorText_ = "RtApiCore::callbackEvent(): the JACK buffer size has changed ... cannot process!";

-    error( RtAudioError::WARNING );

-    return FAILURE;

-  }

-

-  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;

-  JackHandle *handle = (JackHandle *) stream_.apiHandle;

-

-  // Check if we were draining the stream and signal is finished.

-  if ( handle->drainCounter > 3 ) {

-    ThreadHandle threadId;

-

-    stream_.state = STREAM_STOPPING;

-    if ( handle->internalDrain == true )

-      pthread_create( &threadId, NULL, jackStopStream, info );

-    else

-      pthread_cond_signal( &handle->condition );

-    return SUCCESS;

-  }

-

-  // Invoke user callback first, to get fresh output data.

-  if ( handle->drainCounter == 0 ) {

-    RtAudioCallback callback = (RtAudioCallback) info->callback;

-    double streamTime = getStreamTime();

-    RtAudioStreamStatus status = 0;

-    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {

-      status |= RTAUDIO_OUTPUT_UNDERFLOW;

-      handle->xrun[0] = false;

-    }

-    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {

-      status |= RTAUDIO_INPUT_OVERFLOW;

-      handle->xrun[1] = false;

-    }

-    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                                  stream_.bufferSize, streamTime, status, info->userData );

-    if ( cbReturnValue == 2 ) {

-      stream_.state = STREAM_STOPPING;

-      handle->drainCounter = 2;

-      ThreadHandle id;

-      pthread_create( &id, NULL, jackStopStream, info );

-      return SUCCESS;

-    }

-    else if ( cbReturnValue == 1 ) {

-      handle->drainCounter = 1;

-      handle->internalDrain = true;

-    }

-  }

-

-  jack_default_audio_sample_t *jackbuffer;

-  unsigned long bufferBytes = nframes * sizeof( jack_default_audio_sample_t );

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    if ( handle->drainCounter > 1 ) { // write zeros to the output stream

-

-      for ( unsigned int i=0; i<stream_.nDeviceChannels[0]; i++ ) {

-        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );

-        memset( jackbuffer, 0, bufferBytes );

-      }

-

-    }

-    else if ( stream_.doConvertBuffer[0] ) {

-

-      convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-

-      for ( unsigned int i=0; i<stream_.nDeviceChannels[0]; i++ ) {

-        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );

-        memcpy( jackbuffer, &stream_.deviceBuffer[i*bufferBytes], bufferBytes );

-      }

-    }

-    else { // no buffer conversion

-      for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {

-        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );

-        memcpy( jackbuffer, &stream_.userBuffer[0][i*bufferBytes], bufferBytes );

-      }

-    }

-  }

-

-  // Don't bother draining input

-  if ( handle->drainCounter ) {

-    handle->drainCounter++;

-    goto unlock;

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    if ( stream_.doConvertBuffer[1] ) {

-      for ( unsigned int i=0; i<stream_.nDeviceChannels[1]; i++ ) {

-        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[1][i], (jack_nframes_t) nframes );

-        memcpy( &stream_.deviceBuffer[i*bufferBytes], jackbuffer, bufferBytes );

-      }

-      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );

-    }

-    else { // no buffer conversion

-      for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {

-        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[1][i], (jack_nframes_t) nframes );

-        memcpy( &stream_.userBuffer[1][i*bufferBytes], jackbuffer, bufferBytes );

-      }

-    }

-  }

-

- unlock:

-  RtApi::tickStreamTime();

-  return SUCCESS;

-}

-  //******************** End of __UNIX_JACK__ *********************//

-#endif

-

-#if defined(__WINDOWS_ASIO__) // ASIO API on Windows

-

-// The ASIO API is designed around a callback scheme, so this

-// implementation is similar to that used for OS-X CoreAudio and Linux

-// Jack.  The primary constraint with ASIO is that it only allows

-// access to a single driver at a time.  Thus, it is not possible to

-// have more than one simultaneous RtAudio stream.

-//

-// This implementation also requires a number of external ASIO files

-// and a few global variables.  The ASIO callback scheme does not

-// allow for the passing of user data, so we must create a global

-// pointer to our callbackInfo structure.

-//

-// On unix systems, we make use of a pthread condition variable.

-// Since there is no equivalent in Windows, I hacked something based

-// on information found in

-// http://www.cs.wustl.edu/~schmidt/win32-cv-1.html.

-

-#include "asiosys.h"

-#include "asio.h"

-#include "iasiothiscallresolver.h"

-#include "asiodrivers.h"

-#include <cmath>

-

-static AsioDrivers drivers;

-static ASIOCallbacks asioCallbacks;

-static ASIODriverInfo driverInfo;

-static CallbackInfo *asioCallbackInfo;

-static bool asioXRun;

-

-struct AsioHandle {

-  int drainCounter;       // Tracks callback counts when draining

-  bool internalDrain;     // Indicates if stop is initiated from callback or not.

-  ASIOBufferInfo *bufferInfos;

-  HANDLE condition;

-

-  AsioHandle()

-    :drainCounter(0), internalDrain(false), bufferInfos(0) {}

-};

-

-// Function declarations (definitions at end of section)

-static const char* getAsioErrorString( ASIOError result );

-static void sampleRateChanged( ASIOSampleRate sRate );

-static long asioMessages( long selector, long value, void* message, double* opt );

-

-RtApiAsio :: RtApiAsio()

-{

-  // ASIO cannot run on a multi-threaded appartment. You can call

-  // CoInitialize beforehand, but it must be for appartment threading

-  // (in which case, CoInitilialize will return S_FALSE here).

-  coInitialized_ = false;

-  HRESULT hr = CoInitialize( NULL ); 

-  if ( FAILED(hr) ) {

-    errorText_ = "RtApiAsio::ASIO requires a single-threaded appartment. Call CoInitializeEx(0,COINIT_APARTMENTTHREADED)";

-    error( RtAudioError::WARNING );

-  }

-  coInitialized_ = true;

-

-  drivers.removeCurrentDriver();

-  driverInfo.asioVersion = 2;

-

-  // See note in DirectSound implementation about GetDesktopWindow().

-  driverInfo.sysRef = GetForegroundWindow();

-}

-

-RtApiAsio :: ~RtApiAsio()

-{

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-  if ( coInitialized_ ) CoUninitialize();

-}

-

-unsigned int RtApiAsio :: getDeviceCount( void )

-{

-  return (unsigned int) drivers.asioGetNumDev();

-}

-

-RtAudio::DeviceInfo RtApiAsio :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  // Get device ID

-  unsigned int nDevices = getDeviceCount();

-  if ( nDevices == 0 ) {

-    errorText_ = "RtApiAsio::getDeviceInfo: no devices found!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  if ( device >= nDevices ) {

-    errorText_ = "RtApiAsio::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  // If a stream is already open, we cannot probe other devices.  Thus, use the saved results.

-  if ( stream_.state != STREAM_CLOSED ) {

-    if ( device >= devices_.size() ) {

-      errorText_ = "RtApiAsio::getDeviceInfo: device ID was not present before stream was opened.";

-      error( RtAudioError::WARNING );

-      return info;

-    }

-    return devices_[ device ];

-  }

-

-  char driverName[32];

-  ASIOError result = drivers.asioGetDriverName( (int) device, driverName, 32 );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::getDeviceInfo: unable to get driver name (" << getAsioErrorString( result ) << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  info.name = driverName;

-

-  if ( !drivers.loadDriver( driverName ) ) {

-    errorStream_ << "RtApiAsio::getDeviceInfo: unable to load driver (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  result = ASIOInit( &driverInfo );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") initializing driver (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Determine the device channel information.

-  long inputChannels, outputChannels;

-  result = ASIOGetChannels( &inputChannels, &outputChannels );

-  if ( result != ASE_OK ) {

-    drivers.removeCurrentDriver();

-    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") getting channel count (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  info.outputChannels = outputChannels;

-  info.inputChannels = inputChannels;

-  if ( info.outputChannels > 0 && info.inputChannels > 0 )

-    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-

-  // Determine the supported sample rates.

-  info.sampleRates.clear();

-  for ( unsigned int i=0; i<MAX_SAMPLE_RATES; i++ ) {

-    result = ASIOCanSampleRate( (ASIOSampleRate) SAMPLE_RATES[i] );

-    if ( result == ASE_OK ) {

-      info.sampleRates.push_back( SAMPLE_RATES[i] );

-

-      if ( !info.preferredSampleRate || ( SAMPLE_RATES[i] <= 48000 && SAMPLE_RATES[i] > info.preferredSampleRate ) )

-        info.preferredSampleRate = SAMPLE_RATES[i];

-    }

-  }

-

-  // Determine supported data types ... just check first channel and assume rest are the same.

-  ASIOChannelInfo channelInfo;

-  channelInfo.channel = 0;

-  channelInfo.isInput = true;

-  if ( info.inputChannels <= 0 ) channelInfo.isInput = false;

-  result = ASIOGetChannelInfo( &channelInfo );

-  if ( result != ASE_OK ) {

-    drivers.removeCurrentDriver();

-    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") getting driver channel info (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  info.nativeFormats = 0;

-  if ( channelInfo.type == ASIOSTInt16MSB || channelInfo.type == ASIOSTInt16LSB )

-    info.nativeFormats |= RTAUDIO_SINT16;

-  else if ( channelInfo.type == ASIOSTInt32MSB || channelInfo.type == ASIOSTInt32LSB )

-    info.nativeFormats |= RTAUDIO_SINT32;

-  else if ( channelInfo.type == ASIOSTFloat32MSB || channelInfo.type == ASIOSTFloat32LSB )

-    info.nativeFormats |= RTAUDIO_FLOAT32;

-  else if ( channelInfo.type == ASIOSTFloat64MSB || channelInfo.type == ASIOSTFloat64LSB )

-    info.nativeFormats |= RTAUDIO_FLOAT64;

-  else if ( channelInfo.type == ASIOSTInt24MSB || channelInfo.type == ASIOSTInt24LSB )

-    info.nativeFormats |= RTAUDIO_SINT24;

-

-  if ( info.outputChannels > 0 )

-    if ( getDefaultOutputDevice() == device ) info.isDefaultOutput = true;

-  if ( info.inputChannels > 0 )

-    if ( getDefaultInputDevice() == device ) info.isDefaultInput = true;

-

-  info.probed = true;

-  drivers.removeCurrentDriver();

-  return info;

-}

-

-static void bufferSwitch( long index, ASIOBool /*processNow*/ )

-{

-  RtApiAsio *object = (RtApiAsio *) asioCallbackInfo->object;

-  object->callbackEvent( index );

-}

-

-void RtApiAsio :: saveDeviceInfo( void )

-{

-  devices_.clear();

-

-  unsigned int nDevices = getDeviceCount();

-  devices_.resize( nDevices );

-  for ( unsigned int i=0; i<nDevices; i++ )

-    devices_[i] = getDeviceInfo( i );

-}

-

-bool RtApiAsio :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                   unsigned int firstChannel, unsigned int sampleRate,

-                                   RtAudioFormat format, unsigned int *bufferSize,

-                                   RtAudio::StreamOptions *options )

-{////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

-

-  bool isDuplexInput =  mode == INPUT && stream_.mode == OUTPUT;

-

-  // For ASIO, a duplex stream MUST use the same driver.

-  if ( isDuplexInput && stream_.device[0] != device ) {

-    errorText_ = "RtApiAsio::probeDeviceOpen: an ASIO duplex stream must use the same device for input and output!";

-    return FAILURE;

-  }

-

-  char driverName[32];

-  ASIOError result = drivers.asioGetDriverName( (int) device, driverName, 32 );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: unable to get driver name (" << getAsioErrorString( result ) << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Only load the driver once for duplex stream.

-  if ( !isDuplexInput ) {

-    // The getDeviceInfo() function will not work when a stream is open

-    // because ASIO does not allow multiple devices to run at the same

-    // time.  Thus, we'll probe the system before opening a stream and

-    // save the results for use by getDeviceInfo().

-    this->saveDeviceInfo();

-

-    if ( !drivers.loadDriver( driverName ) ) {

-      errorStream_ << "RtApiAsio::probeDeviceOpen: unable to load driver (" << driverName << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    result = ASIOInit( &driverInfo );

-    if ( result != ASE_OK ) {

-      errorStream_ << "RtApiAsio::probeDeviceOpen: error (" << getAsioErrorString( result ) << ") initializing driver (" << driverName << ").";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-

-  // keep them before any "goto error", they are used for error cleanup + goto device boundary checks

-  bool buffersAllocated = false;

-  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-  unsigned int nChannels;

-

-

-  // Check the device channel count.

-  long inputChannels, outputChannels;

-  result = ASIOGetChannels( &inputChannels, &outputChannels );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: error (" << getAsioErrorString( result ) << ") getting channel count (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  if ( ( mode == OUTPUT && (channels+firstChannel) > (unsigned int) outputChannels) ||

-       ( mode == INPUT && (channels+firstChannel) > (unsigned int) inputChannels) ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") does not support requested channel count (" << channels << ") + offset (" << firstChannel << ").";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-  stream_.nDeviceChannels[mode] = channels;

-  stream_.nUserChannels[mode] = channels;

-  stream_.channelOffset[mode] = firstChannel;

-

-  // Verify the sample rate is supported.

-  result = ASIOCanSampleRate( (ASIOSampleRate) sampleRate );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") does not support requested sample rate (" << sampleRate << ").";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  // Get the current sample rate

-  ASIOSampleRate currentRate;

-  result = ASIOGetSampleRate( &currentRate );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error getting sample rate.";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  // Set the sample rate only if necessary

-  if ( currentRate != sampleRate ) {

-    result = ASIOSetSampleRate( (ASIOSampleRate) sampleRate );

-    if ( result != ASE_OK ) {

-      errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error setting sample rate (" << sampleRate << ").";

-      errorText_ = errorStream_.str();

-      goto error;

-    }

-  }

-

-  // Determine the driver data type.

-  ASIOChannelInfo channelInfo;

-  channelInfo.channel = 0;

-  if ( mode == OUTPUT ) channelInfo.isInput = false;

-  else channelInfo.isInput = true;

-  result = ASIOGetChannelInfo( &channelInfo );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting data format.";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  // Assuming WINDOWS host is always little-endian.

-  stream_.doByteSwap[mode] = false;

-  stream_.userFormat = format;

-  stream_.deviceFormat[mode] = 0;

-  if ( channelInfo.type == ASIOSTInt16MSB || channelInfo.type == ASIOSTInt16LSB ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-    if ( channelInfo.type == ASIOSTInt16MSB ) stream_.doByteSwap[mode] = true;

-  }

-  else if ( channelInfo.type == ASIOSTInt32MSB || channelInfo.type == ASIOSTInt32LSB ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-    if ( channelInfo.type == ASIOSTInt32MSB ) stream_.doByteSwap[mode] = true;

-  }

-  else if ( channelInfo.type == ASIOSTFloat32MSB || channelInfo.type == ASIOSTFloat32LSB ) {

-    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;

-    if ( channelInfo.type == ASIOSTFloat32MSB ) stream_.doByteSwap[mode] = true;

-  }

-  else if ( channelInfo.type == ASIOSTFloat64MSB || channelInfo.type == ASIOSTFloat64LSB ) {

-    stream_.deviceFormat[mode] = RTAUDIO_FLOAT64;

-    if ( channelInfo.type == ASIOSTFloat64MSB ) stream_.doByteSwap[mode] = true;

-  }

-  else if ( channelInfo.type == ASIOSTInt24MSB || channelInfo.type == ASIOSTInt24LSB ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-    if ( channelInfo.type == ASIOSTInt24MSB ) stream_.doByteSwap[mode] = true;

-  }

-

-  if ( stream_.deviceFormat[mode] == 0 ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") data format not supported by RtAudio.";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  // Set the buffer size.  For a duplex stream, this will end up

-  // setting the buffer size based on the input constraints, which

-  // should be ok.

-  long minSize, maxSize, preferSize, granularity;

-  result = ASIOGetBufferSize( &minSize, &maxSize, &preferSize, &granularity );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting buffer size.";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  if ( isDuplexInput ) {

-    // When this is the duplex input (output was opened before), then we have to use the same

-    // buffersize as the output, because it might use the preferred buffer size, which most

-    // likely wasn't passed as input to this. The buffer sizes have to be identically anyway,

-    // So instead of throwing an error, make them equal. The caller uses the reference

-    // to the "bufferSize" param as usual to set up processing buffers.

-

-    *bufferSize = stream_.bufferSize;

-

-  } else {

-    if ( *bufferSize == 0 ) *bufferSize = preferSize;

-    else if ( *bufferSize < (unsigned int) minSize ) *bufferSize = (unsigned int) minSize;

-    else if ( *bufferSize > (unsigned int) maxSize ) *bufferSize = (unsigned int) maxSize;

-    else if ( granularity == -1 ) {

-      // Make sure bufferSize is a power of two.

-      int log2_of_min_size = 0;

-      int log2_of_max_size = 0;

-

-      for ( unsigned int i = 0; i < sizeof(long) * 8; i++ ) {

-        if ( minSize & ((long)1 << i) ) log2_of_min_size = i;

-        if ( maxSize & ((long)1 << i) ) log2_of_max_size = i;

-      }

-

-      long min_delta = std::abs( (long)*bufferSize - ((long)1 << log2_of_min_size) );

-      int min_delta_num = log2_of_min_size;

-

-      for (int i = log2_of_min_size + 1; i <= log2_of_max_size; i++) {

-        long current_delta = std::abs( (long)*bufferSize - ((long)1 << i) );

-        if (current_delta < min_delta) {

-          min_delta = current_delta;

-          min_delta_num = i;

-        }

-      }

-

-      *bufferSize = ( (unsigned int)1 << min_delta_num );

-      if ( *bufferSize < (unsigned int) minSize ) *bufferSize = (unsigned int) minSize;

-      else if ( *bufferSize > (unsigned int) maxSize ) *bufferSize = (unsigned int) maxSize;

-    }

-    else if ( granularity != 0 ) {

-      // Set to an even multiple of granularity, rounding up.

-      *bufferSize = (*bufferSize + granularity-1) / granularity * granularity;

-    }

-  }

-

-  /*

-  // we don't use it anymore, see above!

-  // Just left it here for the case...

-  if ( isDuplexInput && stream_.bufferSize != *bufferSize ) {

-    errorText_ = "RtApiAsio::probeDeviceOpen: input/output buffersize discrepancy!";

-    goto error;

-  }

-  */

-

-  stream_.bufferSize = *bufferSize;

-  stream_.nBuffers = 2;

-

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;

-  else stream_.userInterleaved = true;

-

-  // ASIO always uses non-interleaved buffers.

-  stream_.deviceInterleaved[mode] = false;

-

-  // Allocate, if necessary, our AsioHandle structure for the stream.

-  if ( handle == 0 ) {

-    try {

-      handle = new AsioHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiAsio::probeDeviceOpen: error allocating AsioHandle memory.";

-      goto error;

-    }

-    handle->bufferInfos = 0;

-

-    // Create a manual-reset event.

-    handle->condition = CreateEvent( NULL,   // no security

-                                     TRUE,   // manual-reset

-                                     FALSE,  // non-signaled initially

-                                     NULL ); // unnamed

-    stream_.apiHandle = (void *) handle;

-  }

-

-  // Create the ASIO internal buffers.  Since RtAudio sets up input

-  // and output separately, we'll have to dispose of previously

-  // created output buffers for a duplex stream.

-  if ( mode == INPUT && stream_.mode == OUTPUT ) {

-    ASIODisposeBuffers();

-    if ( handle->bufferInfos ) free( handle->bufferInfos );

-  }

-

-  // Allocate, initialize, and save the bufferInfos in our stream callbackInfo structure.

-  unsigned int i;

-  nChannels = stream_.nDeviceChannels[0] + stream_.nDeviceChannels[1];

-  handle->bufferInfos = (ASIOBufferInfo *) malloc( nChannels * sizeof(ASIOBufferInfo) );

-  if ( handle->bufferInfos == NULL ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: error allocating bufferInfo memory for driver (" << driverName << ").";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-

-  ASIOBufferInfo *infos;

-  infos = handle->bufferInfos;

-  for ( i=0; i<stream_.nDeviceChannels[0]; i++, infos++ ) {

-    infos->isInput = ASIOFalse;

-    infos->channelNum = i + stream_.channelOffset[0];

-    infos->buffers[0] = infos->buffers[1] = 0;

-  }

-  for ( i=0; i<stream_.nDeviceChannels[1]; i++, infos++ ) {

-    infos->isInput = ASIOTrue;

-    infos->channelNum = i + stream_.channelOffset[1];

-    infos->buffers[0] = infos->buffers[1] = 0;

-  }

-

-  // prepare for callbacks

-  stream_.sampleRate = sampleRate;

-  stream_.device[mode] = device;

-  stream_.mode = isDuplexInput ? DUPLEX : mode;

-

-  // store this class instance before registering callbacks, that are going to use it

-  asioCallbackInfo = &stream_.callbackInfo;

-  stream_.callbackInfo.object = (void *) this;

-

-  // Set up the ASIO callback structure and create the ASIO data buffers.

-  asioCallbacks.bufferSwitch = &bufferSwitch;

-  asioCallbacks.sampleRateDidChange = &sampleRateChanged;

-  asioCallbacks.asioMessage = &asioMessages;

-  asioCallbacks.bufferSwitchTimeInfo = NULL;

-  result = ASIOCreateBuffers( handle->bufferInfos, nChannels, stream_.bufferSize, &asioCallbacks );

-  if ( result != ASE_OK ) {

-    // Standard method failed. This can happen with strict/misbehaving drivers that return valid buffer size ranges

-    // but only accept the preferred buffer size as parameter for ASIOCreateBuffers. eg. Creatives ASIO driver

-    // in that case, let's be naïve and try that instead

-    *bufferSize = preferSize;

-    stream_.bufferSize = *bufferSize;

-    result = ASIOCreateBuffers( handle->bufferInfos, nChannels, stream_.bufferSize, &asioCallbacks );

-  }

-

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") creating buffers.";

-    errorText_ = errorStream_.str();

-    goto error;

-  }

-  buffersAllocated = true;  

-  stream_.state = STREAM_STOPPED;

-

-  // Set flags for buffer conversion.

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-       stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate necessary internal buffers

-  unsigned long bufferBytes;

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiAsio::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( isDuplexInput && stream_.deviceBuffer ) {

-      unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-      if ( bufferBytes <= bytesOut ) makeBuffer = false;

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiAsio::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  // Determine device latencies

-  long inputLatency, outputLatency;

-  result = ASIOGetLatencies( &inputLatency, &outputLatency );

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting latency.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING); // warn but don't fail

-  }

-  else {

-    stream_.latency[0] = outputLatency;

-    stream_.latency[1] = inputLatency;

-  }

-

-  // Setup the buffer conversion information structure.  We don't use

-  // buffers to do channel offsets, so we override that parameter

-  // here.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, 0 );

-

-  return SUCCESS;

-

- error:

-  if ( !isDuplexInput ) {

-    // the cleanup for error in the duplex input, is done by RtApi::openStream

-    // So we clean up for single channel only

-

-    if ( buffersAllocated )

-      ASIODisposeBuffers();

-

-    drivers.removeCurrentDriver();

-

-    if ( handle ) {

-      CloseHandle( handle->condition );

-      if ( handle->bufferInfos )

-        free( handle->bufferInfos );

-

-      delete handle;

-      stream_.apiHandle = 0;

-    }

-

-

-    if ( stream_.userBuffer[mode] ) {

-      free( stream_.userBuffer[mode] );

-      stream_.userBuffer[mode] = 0;

-    }

-

-    if ( stream_.deviceBuffer ) {

-      free( stream_.deviceBuffer );

-      stream_.deviceBuffer = 0;

-    }

-  }

-

-  return FAILURE;

-}////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

-

-void RtApiAsio :: closeStream()

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiAsio::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  if ( stream_.state == STREAM_RUNNING ) {

-    stream_.state = STREAM_STOPPED;

-    ASIOStop();

-  }

-  ASIODisposeBuffers();

-  drivers.removeCurrentDriver();

-

-  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-  if ( handle ) {

-    CloseHandle( handle->condition );

-    if ( handle->bufferInfos )

-      free( handle->bufferInfos );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-bool stopThreadCalled = false;

-

-void RtApiAsio :: startStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiAsio::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-  ASIOError result = ASIOStart();

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::startStream: error (" << getAsioErrorString( result ) << ") starting device.";

-    errorText_ = errorStream_.str();

-    goto unlock;

-  }

-

-  handle->drainCounter = 0;

-  handle->internalDrain = false;

-  ResetEvent( handle->condition );

-  stream_.state = STREAM_RUNNING;

-  asioXRun = false;

-

- unlock:

-  stopThreadCalled = false;

-

-  if ( result == ASE_OK ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiAsio :: stopStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiAsio::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    if ( handle->drainCounter == 0 ) {

-      handle->drainCounter = 2;

-      WaitForSingleObject( handle->condition, INFINITE );  // block until signaled

-    }

-  }

-

-  stream_.state = STREAM_STOPPED;

-

-  ASIOError result = ASIOStop();

-  if ( result != ASE_OK ) {

-    errorStream_ << "RtApiAsio::stopStream: error (" << getAsioErrorString( result ) << ") stopping device.";

-    errorText_ = errorStream_.str();

-  }

-

-  if ( result == ASE_OK ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiAsio :: abortStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiAsio::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // The following lines were commented-out because some behavior was

-  // noted where the device buffers need to be zeroed to avoid

-  // continuing sound, even when the device buffers are completely

-  // disposed.  So now, calling abort is the same as calling stop.

-  // AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-  // handle->drainCounter = 2;

-  stopStream();

-}

-

-// This function will be called by a spawned thread when the user

-// callback function signals that the stream should be stopped or

-// aborted.  It is necessary to handle it this way because the

-// callbackEvent() function must return before the ASIOStop()

-// function will return.

-static unsigned __stdcall asioStopStream( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiAsio *object = (RtApiAsio *) info->object;

-

-  object->stopStream();

-  _endthreadex( 0 );

-  return 0;

-}

-

-bool RtApiAsio :: callbackEvent( long bufferIndex )

-{

-  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiAsio::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return FAILURE;

-  }

-

-  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;

-  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;

-

-  // Check if we were draining the stream and signal if finished.

-  if ( handle->drainCounter > 3 ) {

-

-    stream_.state = STREAM_STOPPING;

-    if ( handle->internalDrain == false )

-      SetEvent( handle->condition );

-    else { // spawn a thread to stop the stream

-      unsigned threadId;

-      stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &asioStopStream,

-                                                    &stream_.callbackInfo, 0, &threadId );

-    }

-    return SUCCESS;

-  }

-

-  // Invoke user callback to get fresh output data UNLESS we are

-  // draining stream.

-  if ( handle->drainCounter == 0 ) {

-    RtAudioCallback callback = (RtAudioCallback) info->callback;

-    double streamTime = getStreamTime();

-    RtAudioStreamStatus status = 0;

-    if ( stream_.mode != INPUT && asioXRun == true ) {

-      status |= RTAUDIO_OUTPUT_UNDERFLOW;

-      asioXRun = false;

-    }

-    if ( stream_.mode != OUTPUT && asioXRun == true ) {

-      status |= RTAUDIO_INPUT_OVERFLOW;

-      asioXRun = false;

-    }

-    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                                     stream_.bufferSize, streamTime, status, info->userData );

-    if ( cbReturnValue == 2 ) {

-      stream_.state = STREAM_STOPPING;

-      handle->drainCounter = 2;

-      unsigned threadId;

-      stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &asioStopStream,

-                                                    &stream_.callbackInfo, 0, &threadId );

-      return SUCCESS;

-    }

-    else if ( cbReturnValue == 1 ) {

-      handle->drainCounter = 1;

-      handle->internalDrain = true;

-    }

-  }

-

-  unsigned int nChannels, bufferBytes, i, j;

-  nChannels = stream_.nDeviceChannels[0] + stream_.nDeviceChannels[1];

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    bufferBytes = stream_.bufferSize * formatBytes( stream_.deviceFormat[0] );

-

-    if ( handle->drainCounter > 1 ) { // write zeros to the output stream

-

-      for ( i=0, j=0; i<nChannels; i++ ) {

-        if ( handle->bufferInfos[i].isInput != ASIOTrue )

-          memset( handle->bufferInfos[i].buffers[bufferIndex], 0, bufferBytes );

-      }

-

-    }

-    else if ( stream_.doConvertBuffer[0] ) {

-

-      convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-      if ( stream_.doByteSwap[0] )

-        byteSwapBuffer( stream_.deviceBuffer,

-                        stream_.bufferSize * stream_.nDeviceChannels[0],

-                        stream_.deviceFormat[0] );

-

-      for ( i=0, j=0; i<nChannels; i++ ) {

-        if ( handle->bufferInfos[i].isInput != ASIOTrue )

-          memcpy( handle->bufferInfos[i].buffers[bufferIndex],

-                  &stream_.deviceBuffer[j++*bufferBytes], bufferBytes );

-      }

-

-    }

-    else {

-

-      if ( stream_.doByteSwap[0] )

-        byteSwapBuffer( stream_.userBuffer[0],

-                        stream_.bufferSize * stream_.nUserChannels[0],

-                        stream_.userFormat );

-

-      for ( i=0, j=0; i<nChannels; i++ ) {

-        if ( handle->bufferInfos[i].isInput != ASIOTrue )

-          memcpy( handle->bufferInfos[i].buffers[bufferIndex],

-                  &stream_.userBuffer[0][bufferBytes*j++], bufferBytes );

-      }

-

-    }

-  }

-

-  // Don't bother draining input

-  if ( handle->drainCounter ) {

-    handle->drainCounter++;

-    goto unlock;

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    bufferBytes = stream_.bufferSize * formatBytes(stream_.deviceFormat[1]);

-

-    if (stream_.doConvertBuffer[1]) {

-

-      // Always interleave ASIO input data.

-      for ( i=0, j=0; i<nChannels; i++ ) {

-        if ( handle->bufferInfos[i].isInput == ASIOTrue )

-          memcpy( &stream_.deviceBuffer[j++*bufferBytes],

-                  handle->bufferInfos[i].buffers[bufferIndex],

-                  bufferBytes );

-      }

-

-      if ( stream_.doByteSwap[1] )

-        byteSwapBuffer( stream_.deviceBuffer,

-                        stream_.bufferSize * stream_.nDeviceChannels[1],

-                        stream_.deviceFormat[1] );

-      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );

-

-    }

-    else {

-      for ( i=0, j=0; i<nChannels; i++ ) {

-        if ( handle->bufferInfos[i].isInput == ASIOTrue ) {

-          memcpy( &stream_.userBuffer[1][bufferBytes*j++],

-                  handle->bufferInfos[i].buffers[bufferIndex],

-                  bufferBytes );

-        }

-      }

-

-      if ( stream_.doByteSwap[1] )

-        byteSwapBuffer( stream_.userBuffer[1],

-                        stream_.bufferSize * stream_.nUserChannels[1],

-                        stream_.userFormat );

-    }

-  }

-

- unlock:

-  // The following call was suggested by Malte Clasen.  While the API

-  // documentation indicates it should not be required, some device

-  // drivers apparently do not function correctly without it.

-  ASIOOutputReady();

-

-  RtApi::tickStreamTime();

-  return SUCCESS;

-}

-

-static void sampleRateChanged( ASIOSampleRate sRate )

-{

-  // The ASIO documentation says that this usually only happens during

-  // external sync.  Audio processing is not stopped by the driver,

-  // actual sample rate might not have even changed, maybe only the

-  // sample rate status of an AES/EBU or S/PDIF digital input at the

-  // audio device.

-

-  RtApi *object = (RtApi *) asioCallbackInfo->object;

-  try {

-    object->stopStream();

-  }

-  catch ( RtAudioError &exception ) {

-    std::cerr << "\nRtApiAsio: sampleRateChanged() error (" << exception.getMessage() << ")!\n" << std::endl;

-    return;

-  }

-

-  std::cerr << "\nRtApiAsio: driver reports sample rate changed to " << sRate << " ... stream stopped!!!\n" << std::endl;

-}

-

-static long asioMessages( long selector, long value, void* /*message*/, double* /*opt*/ )

-{

-  long ret = 0;

-

-  switch( selector ) {

-  case kAsioSelectorSupported:

-    if ( value == kAsioResetRequest

-         || value == kAsioEngineVersion

-         || value == kAsioResyncRequest

-         || value == kAsioLatenciesChanged

-         // The following three were added for ASIO 2.0, you don't

-         // necessarily have to support them.

-         || value == kAsioSupportsTimeInfo

-         || value == kAsioSupportsTimeCode

-         || value == kAsioSupportsInputMonitor)

-      ret = 1L;

-    break;

-  case kAsioResetRequest:

-    // Defer the task and perform the reset of the driver during the

-    // next "safe" situation.  You cannot reset the driver right now,

-    // as this code is called from the driver.  Reset the driver is

-    // done by completely destruct is. I.e. ASIOStop(),

-    // ASIODisposeBuffers(), Destruction Afterwards you initialize the

-    // driver again.

-    std::cerr << "\nRtApiAsio: driver reset requested!!!" << std::endl;

-    ret = 1L;

-    break;

-  case kAsioResyncRequest:

-    // This informs the application that the driver encountered some

-    // non-fatal data loss.  It is used for synchronization purposes

-    // of different media.  Added mainly to work around the Win16Mutex

-    // problems in Windows 95/98 with the Windows Multimedia system,

-    // which could lose data because the Mutex was held too long by

-    // another thread.  However a driver can issue it in other

-    // situations, too.

-    // std::cerr << "\nRtApiAsio: driver resync requested!!!" << std::endl;

-    asioXRun = true;

-    ret = 1L;

-    break;

-  case kAsioLatenciesChanged:

-    // This will inform the host application that the drivers were

-    // latencies changed.  Beware, it this does not mean that the

-    // buffer sizes have changed!  You might need to update internal

-    // delay data.

-    std::cerr << "\nRtApiAsio: driver latency may have changed!!!" << std::endl;

-    ret = 1L;

-    break;

-  case kAsioEngineVersion:

-    // Return the supported ASIO version of the host application.  If

-    // a host application does not implement this selector, ASIO 1.0

-    // is assumed by the driver.

-    ret = 2L;

-    break;

-  case kAsioSupportsTimeInfo:

-    // Informs the driver whether the

-    // asioCallbacks.bufferSwitchTimeInfo() callback is supported.

-    // For compatibility with ASIO 1.0 drivers the host application

-    // should always support the "old" bufferSwitch method, too.

-    ret = 0;

-    break;

-  case kAsioSupportsTimeCode:

-    // Informs the driver whether application is interested in time

-    // code info.  If an application does not need to know about time

-    // code, the driver has less work to do.

-    ret = 0;

-    break;

-  }

-  return ret;

-}

-

-static const char* getAsioErrorString( ASIOError result )

-{

-  struct Messages 

-  {

-    ASIOError value;

-    const char*message;

-  };

-

-  static const Messages m[] = 

-    {

-      {   ASE_NotPresent,    "Hardware input or output is not present or available." },

-      {   ASE_HWMalfunction,  "Hardware is malfunctioning." },

-      {   ASE_InvalidParameter, "Invalid input parameter." },

-      {   ASE_InvalidMode,      "Invalid mode." },

-      {   ASE_SPNotAdvancing,     "Sample position not advancing." },

-      {   ASE_NoClock,            "Sample clock or rate cannot be determined or is not present." },

-      {   ASE_NoMemory,           "Not enough memory to complete the request." }

-    };

-

-  for ( unsigned int i = 0; i < sizeof(m)/sizeof(m[0]); ++i )

-    if ( m[i].value == result ) return m[i].message;

-

-  return "Unknown error.";

-}

-

-//******************** End of __WINDOWS_ASIO__ *********************//

-#endif

-

-

-#if defined(__WINDOWS_WASAPI__) // Windows WASAPI API

-

-// Authored by Marcus Tomlinson <themarcustomlinson@gmail.com>, April 2014

-// - Introduces support for the Windows WASAPI API

-// - Aims to deliver bit streams to and from hardware at the lowest possible latency, via the absolute minimum buffer sizes required

-// - Provides flexible stream configuration to an otherwise strict and inflexible WASAPI interface

-// - Includes automatic internal conversion of sample rate and buffer size between hardware and the user

-

-#ifndef INITGUID

-  #define INITGUID

-#endif

-#include <audioclient.h>

-#include <avrt.h>

-#include <mmdeviceapi.h>

-#include <functiondiscoverykeys_devpkey.h>

-

-//=============================================================================

-

-#define SAFE_RELEASE( objectPtr )\

-if ( objectPtr )\

-{\

-  objectPtr->Release();\

-  objectPtr = NULL;\

-}

-

-typedef HANDLE ( __stdcall *TAvSetMmThreadCharacteristicsPtr )( LPCWSTR TaskName, LPDWORD TaskIndex );

-

-//-----------------------------------------------------------------------------

-

-// WASAPI dictates stream sample rate, format, channel count, and in some cases, buffer size.

-// Therefore we must perform all necessary conversions to user buffers in order to satisfy these

-// requirements. WasapiBuffer ring buffers are used between HwIn->UserIn and UserOut->HwOut to

-// provide intermediate storage for read / write synchronization.

-class WasapiBuffer

-{

-public:

-  WasapiBuffer()

-    : buffer_( NULL ),

-      bufferSize_( 0 ),

-      inIndex_( 0 ),

-      outIndex_( 0 ) {}

-

-  ~WasapiBuffer() {

-    free( buffer_ );

-  }

-

-  // sets the length of the internal ring buffer

-  void setBufferSize( unsigned int bufferSize, unsigned int formatBytes ) {

-    free( buffer_ );

-

-    buffer_ = ( char* ) calloc( bufferSize, formatBytes );

-

-    bufferSize_ = bufferSize;

-    inIndex_ = 0;

-    outIndex_ = 0;

-  }

-

-  // attempt to push a buffer into the ring buffer at the current "in" index

-  bool pushBuffer( char* buffer, unsigned int bufferSize, RtAudioFormat format )

-  {

-    if ( !buffer ||                 // incoming buffer is NULL

-         bufferSize == 0 ||         // incoming buffer has no data

-         bufferSize > bufferSize_ ) // incoming buffer too large

-    {

-      return false;

-    }

-

-    unsigned int relOutIndex = outIndex_;

-    unsigned int inIndexEnd = inIndex_ + bufferSize;

-    if ( relOutIndex < inIndex_ && inIndexEnd >= bufferSize_ ) {

-      relOutIndex += bufferSize_;

-    }

-

-    // "in" index can end on the "out" index but cannot begin at it

-    if ( inIndex_ <= relOutIndex && inIndexEnd > relOutIndex ) {

-      return false; // not enough space between "in" index and "out" index

-    }

-

-    // copy buffer from external to internal

-    int fromZeroSize = inIndex_ + bufferSize - bufferSize_;

-    fromZeroSize = fromZeroSize < 0 ? 0 : fromZeroSize;

-    int fromInSize = bufferSize - fromZeroSize;

-

-    switch( format )

-      {

-      case RTAUDIO_SINT8:

-        memcpy( &( ( char* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( char ) );

-        memcpy( buffer_, &( ( char* ) buffer )[fromInSize], fromZeroSize * sizeof( char ) );

-        break;

-      case RTAUDIO_SINT16:

-        memcpy( &( ( short* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( short ) );

-        memcpy( buffer_, &( ( short* ) buffer )[fromInSize], fromZeroSize * sizeof( short ) );

-        break;

-      case RTAUDIO_SINT24:

-        memcpy( &( ( S24* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( S24 ) );

-        memcpy( buffer_, &( ( S24* ) buffer )[fromInSize], fromZeroSize * sizeof( S24 ) );

-        break;

-      case RTAUDIO_SINT32:

-        memcpy( &( ( int* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( int ) );

-        memcpy( buffer_, &( ( int* ) buffer )[fromInSize], fromZeroSize * sizeof( int ) );

-        break;

-      case RTAUDIO_FLOAT32:

-        memcpy( &( ( float* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( float ) );

-        memcpy( buffer_, &( ( float* ) buffer )[fromInSize], fromZeroSize * sizeof( float ) );

-        break;

-      case RTAUDIO_FLOAT64:

-        memcpy( &( ( double* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( double ) );

-        memcpy( buffer_, &( ( double* ) buffer )[fromInSize], fromZeroSize * sizeof( double ) );

-        break;

-    }

-

-    // update "in" index

-    inIndex_ += bufferSize;

-    inIndex_ %= bufferSize_;

-

-    return true;

-  }

-

-  // attempt to pull a buffer from the ring buffer from the current "out" index

-  bool pullBuffer( char* buffer, unsigned int bufferSize, RtAudioFormat format )

-  {

-    if ( !buffer ||                 // incoming buffer is NULL

-         bufferSize == 0 ||         // incoming buffer has no data

-         bufferSize > bufferSize_ ) // incoming buffer too large

-    {

-      return false;

-    }

-

-    unsigned int relInIndex = inIndex_;

-    unsigned int outIndexEnd = outIndex_ + bufferSize;

-    if ( relInIndex < outIndex_ && outIndexEnd >= bufferSize_ ) {

-      relInIndex += bufferSize_;

-    }

-

-    // "out" index can begin at and end on the "in" index

-    if ( outIndex_ < relInIndex && outIndexEnd > relInIndex ) {

-      return false; // not enough space between "out" index and "in" index

-    }

-

-    // copy buffer from internal to external

-    int fromZeroSize = outIndex_ + bufferSize - bufferSize_;

-    fromZeroSize = fromZeroSize < 0 ? 0 : fromZeroSize;

-    int fromOutSize = bufferSize - fromZeroSize;

-

-    switch( format )

-    {

-      case RTAUDIO_SINT8:

-        memcpy( buffer, &( ( char* ) buffer_ )[outIndex_], fromOutSize * sizeof( char ) );

-        memcpy( &( ( char* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( char ) );

-        break;

-      case RTAUDIO_SINT16:

-        memcpy( buffer, &( ( short* ) buffer_ )[outIndex_], fromOutSize * sizeof( short ) );

-        memcpy( &( ( short* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( short ) );

-        break;

-      case RTAUDIO_SINT24:

-        memcpy( buffer, &( ( S24* ) buffer_ )[outIndex_], fromOutSize * sizeof( S24 ) );

-        memcpy( &( ( S24* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( S24 ) );

-        break;

-      case RTAUDIO_SINT32:

-        memcpy( buffer, &( ( int* ) buffer_ )[outIndex_], fromOutSize * sizeof( int ) );

-        memcpy( &( ( int* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( int ) );

-        break;

-      case RTAUDIO_FLOAT32:

-        memcpy( buffer, &( ( float* ) buffer_ )[outIndex_], fromOutSize * sizeof( float ) );

-        memcpy( &( ( float* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( float ) );

-        break;

-      case RTAUDIO_FLOAT64:

-        memcpy( buffer, &( ( double* ) buffer_ )[outIndex_], fromOutSize * sizeof( double ) );

-        memcpy( &( ( double* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( double ) );

-        break;

-    }

-

-    // update "out" index

-    outIndex_ += bufferSize;

-    outIndex_ %= bufferSize_;

-

-    return true;

-  }

-

-private:

-  char* buffer_;

-  unsigned int bufferSize_;

-  unsigned int inIndex_;

-  unsigned int outIndex_;

-};

-

-//-----------------------------------------------------------------------------

-

-// In order to satisfy WASAPI's buffer requirements, we need a means of converting sample rate

-// between HW and the user. The convertBufferWasapi function is used to perform this conversion

-// between HwIn->UserIn and UserOut->HwOut during the stream callback loop.

-// This sample rate converter favors speed over quality, and works best with conversions between

-// one rate and its multiple.

-void convertBufferWasapi( char* outBuffer,

-                          const char* inBuffer,

-                          const unsigned int& channelCount,

-                          const unsigned int& inSampleRate,

-                          const unsigned int& outSampleRate,

-                          const unsigned int& inSampleCount,

-                          unsigned int& outSampleCount,

-                          const RtAudioFormat& format )

-{

-  // calculate the new outSampleCount and relative sampleStep

-  float sampleRatio = ( float ) outSampleRate / inSampleRate;

-  float sampleStep = 1.0f / sampleRatio;

-  float inSampleFraction = 0.0f;

-

-  outSampleCount = ( unsigned int ) roundf( inSampleCount * sampleRatio );

-

-  // frame-by-frame, copy each relative input sample into it's corresponding output sample

-  for ( unsigned int outSample = 0; outSample < outSampleCount; outSample++ )

-  {

-    unsigned int inSample = ( unsigned int ) inSampleFraction;

-

-    switch ( format )

-    {

-      case RTAUDIO_SINT8:

-        memcpy( &( ( char* ) outBuffer )[ outSample * channelCount ], &( ( char* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( char ) );

-        break;

-      case RTAUDIO_SINT16:

-        memcpy( &( ( short* ) outBuffer )[ outSample * channelCount ], &( ( short* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( short ) );

-        break;

-      case RTAUDIO_SINT24:

-        memcpy( &( ( S24* ) outBuffer )[ outSample * channelCount ], &( ( S24* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( S24 ) );

-        break;

-      case RTAUDIO_SINT32:

-        memcpy( &( ( int* ) outBuffer )[ outSample * channelCount ], &( ( int* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( int ) );

-        break;

-      case RTAUDIO_FLOAT32:

-        memcpy( &( ( float* ) outBuffer )[ outSample * channelCount ], &( ( float* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( float ) );

-        break;

-      case RTAUDIO_FLOAT64:

-        memcpy( &( ( double* ) outBuffer )[ outSample * channelCount ], &( ( double* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( double ) );

-        break;

-    }

-

-    // jump to next in sample

-    inSampleFraction += sampleStep;

-  }

-}

-

-//-----------------------------------------------------------------------------

-

-// A structure to hold various information related to the WASAPI implementation.

-struct WasapiHandle

-{

-  IAudioClient* captureAudioClient;

-  IAudioClient* renderAudioClient;

-  IAudioCaptureClient* captureClient;

-  IAudioRenderClient* renderClient;

-  HANDLE captureEvent;

-  HANDLE renderEvent;

-

-  WasapiHandle()

-  : captureAudioClient( NULL ),

-    renderAudioClient( NULL ),

-    captureClient( NULL ),

-    renderClient( NULL ),

-    captureEvent( NULL ),

-    renderEvent( NULL ) {}

-};

-

-//=============================================================================

-

-RtApiWasapi::RtApiWasapi()

-  : coInitialized_( false ), deviceEnumerator_( NULL )

-{

-  // WASAPI can run either apartment or multi-threaded

-  HRESULT hr = CoInitialize( NULL );

-  if ( !FAILED( hr ) )

-    coInitialized_ = true;

-

-  // Instantiate device enumerator

-  hr = CoCreateInstance( __uuidof( MMDeviceEnumerator ), NULL,

-                         CLSCTX_ALL, __uuidof( IMMDeviceEnumerator ),

-                         ( void** ) &deviceEnumerator_ );

-

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::RtApiWasapi: Unable to instantiate device enumerator";

-    error( RtAudioError::DRIVER_ERROR );

-  }

-}

-

-//-----------------------------------------------------------------------------

-

-RtApiWasapi::~RtApiWasapi()

-{

-  if ( stream_.state != STREAM_CLOSED )

-    closeStream();

-

-  SAFE_RELEASE( deviceEnumerator_ );

-

-  // If this object previously called CoInitialize()

-  if ( coInitialized_ )

-    CoUninitialize();

-}

-

-//=============================================================================

-

-unsigned int RtApiWasapi::getDeviceCount( void )

-{

-  unsigned int captureDeviceCount = 0;

-  unsigned int renderDeviceCount = 0;

-

-  IMMDeviceCollection* captureDevices = NULL;

-  IMMDeviceCollection* renderDevices = NULL;

-

-  // Count capture devices

-  errorText_.clear();

-  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve capture device collection.";

-    goto Exit;

-  }

-

-  hr = captureDevices->GetCount( &captureDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve capture device count.";

-    goto Exit;

-  }

-

-  // Count render devices

-  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve render device collection.";

-    goto Exit;

-  }

-

-  hr = renderDevices->GetCount( &renderDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve render device count.";

-    goto Exit;

-  }

-

-Exit:

-  // release all references

-  SAFE_RELEASE( captureDevices );

-  SAFE_RELEASE( renderDevices );

-

-  if ( errorText_.empty() )

-    return captureDeviceCount + renderDeviceCount;

-

-  error( RtAudioError::DRIVER_ERROR );

-  return 0;

-}

-

-//-----------------------------------------------------------------------------

-

-RtAudio::DeviceInfo RtApiWasapi::getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  unsigned int captureDeviceCount = 0;

-  unsigned int renderDeviceCount = 0;

-  std::string defaultDeviceName;

-  bool isCaptureDevice = false;

-

-  PROPVARIANT deviceNameProp;

-  PROPVARIANT defaultDeviceNameProp;

-

-  IMMDeviceCollection* captureDevices = NULL;

-  IMMDeviceCollection* renderDevices = NULL;

-  IMMDevice* devicePtr = NULL;

-  IMMDevice* defaultDevicePtr = NULL;

-  IAudioClient* audioClient = NULL;

-  IPropertyStore* devicePropStore = NULL;

-  IPropertyStore* defaultDevicePropStore = NULL;

-

-  WAVEFORMATEX* deviceFormat = NULL;

-  WAVEFORMATEX* closestMatchFormat = NULL;

-

-  // probed

-  info.probed = false;

-

-  // Count capture devices

-  errorText_.clear();

-  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;

-  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device collection.";

-    goto Exit;

-  }

-

-  hr = captureDevices->GetCount( &captureDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device count.";

-    goto Exit;

-  }

-

-  // Count render devices

-  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device collection.";

-    goto Exit;

-  }

-

-  hr = renderDevices->GetCount( &renderDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device count.";

-    goto Exit;

-  }

-

-  // validate device index

-  if ( device >= captureDeviceCount + renderDeviceCount ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Invalid device index.";

-    errorType = RtAudioError::INVALID_USE;

-    goto Exit;

-  }

-

-  // determine whether index falls within capture or render devices

-  if ( device >= renderDeviceCount ) {

-    hr = captureDevices->Item( device - renderDeviceCount, &devicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device handle.";

-      goto Exit;

-    }

-    isCaptureDevice = true;

-  }

-  else {

-    hr = renderDevices->Item( device, &devicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device handle.";

-      goto Exit;

-    }

-    isCaptureDevice = false;

-  }

-

-  // get default device name

-  if ( isCaptureDevice ) {

-    hr = deviceEnumerator_->GetDefaultAudioEndpoint( eCapture, eConsole, &defaultDevicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default capture device handle.";

-      goto Exit;

-    }

-  }

-  else {

-    hr = deviceEnumerator_->GetDefaultAudioEndpoint( eRender, eConsole, &defaultDevicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default render device handle.";

-      goto Exit;

-    }

-  }

-

-  hr = defaultDevicePtr->OpenPropertyStore( STGM_READ, &defaultDevicePropStore );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to open default device property store.";

-    goto Exit;

-  }

-  PropVariantInit( &defaultDeviceNameProp );

-

-  hr = defaultDevicePropStore->GetValue( PKEY_Device_FriendlyName, &defaultDeviceNameProp );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default device property: PKEY_Device_FriendlyName.";

-    goto Exit;

-  }

-

-  defaultDeviceName = convertCharPointerToStdString(defaultDeviceNameProp.pwszVal);

-

-  // name

-  hr = devicePtr->OpenPropertyStore( STGM_READ, &devicePropStore );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to open device property store.";

-    goto Exit;

-  }

-

-  PropVariantInit( &deviceNameProp );

-

-  hr = devicePropStore->GetValue( PKEY_Device_FriendlyName, &deviceNameProp );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device property: PKEY_Device_FriendlyName.";

-    goto Exit;

-  }

-

-  info.name =convertCharPointerToStdString(deviceNameProp.pwszVal);

-

-  // is default

-  if ( isCaptureDevice ) {

-    info.isDefaultInput = info.name == defaultDeviceName;

-    info.isDefaultOutput = false;

-  }

-  else {

-    info.isDefaultInput = false;

-    info.isDefaultOutput = info.name == defaultDeviceName;

-  }

-

-  // channel count

-  hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL, NULL, ( void** ) &audioClient );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device audio client.";

-    goto Exit;

-  }

-

-  hr = audioClient->GetMixFormat( &deviceFormat );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device mix format.";

-    goto Exit;

-  }

-

-  if ( isCaptureDevice ) {

-    info.inputChannels = deviceFormat->nChannels;

-    info.outputChannels = 0;

-    info.duplexChannels = 0;

-  }

-  else {

-    info.inputChannels = 0;

-    info.outputChannels = deviceFormat->nChannels;

-    info.duplexChannels = 0;

-  }

-

-  // sample rates

-  info.sampleRates.clear();

-

-  // allow support for all sample rates as we have a built-in sample rate converter

-  for ( unsigned int i = 0; i < MAX_SAMPLE_RATES; i++ ) {

-    info.sampleRates.push_back( SAMPLE_RATES[i] );

-  }

-  info.preferredSampleRate = deviceFormat->nSamplesPerSec;

-

-  // native format

-  info.nativeFormats = 0;

-

-  if ( deviceFormat->wFormatTag == WAVE_FORMAT_IEEE_FLOAT ||

-       ( deviceFormat->wFormatTag == WAVE_FORMAT_EXTENSIBLE &&

-         ( ( WAVEFORMATEXTENSIBLE* ) deviceFormat )->SubFormat == KSDATAFORMAT_SUBTYPE_IEEE_FLOAT ) )

-  {

-    if ( deviceFormat->wBitsPerSample == 32 ) {

-      info.nativeFormats |= RTAUDIO_FLOAT32;

-    }

-    else if ( deviceFormat->wBitsPerSample == 64 ) {

-      info.nativeFormats |= RTAUDIO_FLOAT64;

-    }

-  }

-  else if ( deviceFormat->wFormatTag == WAVE_FORMAT_PCM ||

-           ( deviceFormat->wFormatTag == WAVE_FORMAT_EXTENSIBLE &&

-             ( ( WAVEFORMATEXTENSIBLE* ) deviceFormat )->SubFormat == KSDATAFORMAT_SUBTYPE_PCM ) )

-  {

-    if ( deviceFormat->wBitsPerSample == 8 ) {

-      info.nativeFormats |= RTAUDIO_SINT8;

-    }

-    else if ( deviceFormat->wBitsPerSample == 16 ) {

-      info.nativeFormats |= RTAUDIO_SINT16;

-    }

-    else if ( deviceFormat->wBitsPerSample == 24 ) {

-      info.nativeFormats |= RTAUDIO_SINT24;

-    }

-    else if ( deviceFormat->wBitsPerSample == 32 ) {

-      info.nativeFormats |= RTAUDIO_SINT32;

-    }

-  }

-

-  // probed

-  info.probed = true;

-

-Exit:

-  // release all references

-  PropVariantClear( &deviceNameProp );

-  PropVariantClear( &defaultDeviceNameProp );

-

-  SAFE_RELEASE( captureDevices );

-  SAFE_RELEASE( renderDevices );

-  SAFE_RELEASE( devicePtr );

-  SAFE_RELEASE( defaultDevicePtr );

-  SAFE_RELEASE( audioClient );

-  SAFE_RELEASE( devicePropStore );

-  SAFE_RELEASE( defaultDevicePropStore );

-

-  CoTaskMemFree( deviceFormat );

-  CoTaskMemFree( closestMatchFormat );

-

-  if ( !errorText_.empty() )

-    error( errorType );

-  return info;

-}

-

-//-----------------------------------------------------------------------------

-

-unsigned int RtApiWasapi::getDefaultOutputDevice( void )

-{

-  for ( unsigned int i = 0; i < getDeviceCount(); i++ ) {

-    if ( getDeviceInfo( i ).isDefaultOutput ) {

-      return i;

-    }

-  }

-

-  return 0;

-}

-

-//-----------------------------------------------------------------------------

-

-unsigned int RtApiWasapi::getDefaultInputDevice( void )

-{

-  for ( unsigned int i = 0; i < getDeviceCount(); i++ ) {

-    if ( getDeviceInfo( i ).isDefaultInput ) {

-      return i;

-    }

-  }

-

-  return 0;

-}

-

-//-----------------------------------------------------------------------------

-

-void RtApiWasapi::closeStream( void )

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiWasapi::closeStream: No open stream to close.";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  if ( stream_.state != STREAM_STOPPED )

-    stopStream();

-

-  // clean up stream memory

-  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient )

-  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient )

-

-  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->captureClient )

-  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->renderClient )

-

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent )

-    CloseHandle( ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent );

-

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent )

-    CloseHandle( ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent );

-

-  delete ( WasapiHandle* ) stream_.apiHandle;

-  stream_.apiHandle = NULL;

-

-  for ( int i = 0; i < 2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  // update stream state

-  stream_.state = STREAM_CLOSED;

-}

-

-//-----------------------------------------------------------------------------

-

-void RtApiWasapi::startStream( void )

-{

-  verifyStream();

-

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiWasapi::startStream: The stream is already running.";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // update stream state

-  stream_.state = STREAM_RUNNING;

-

-  // create WASAPI stream thread

-  stream_.callbackInfo.thread = ( ThreadHandle ) CreateThread( NULL, 0, runWasapiThread, this, CREATE_SUSPENDED, NULL );

-

-  if ( !stream_.callbackInfo.thread ) {

-    errorText_ = "RtApiWasapi::startStream: Unable to instantiate callback thread.";

-    error( RtAudioError::THREAD_ERROR );

-  }

-  else {

-    SetThreadPriority( ( void* ) stream_.callbackInfo.thread, stream_.callbackInfo.priority );

-    ResumeThread( ( void* ) stream_.callbackInfo.thread );

-  }

-}

-

-//-----------------------------------------------------------------------------

-

-void RtApiWasapi::stopStream( void )

-{

-  verifyStream();

-

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiWasapi::stopStream: The stream is already stopped.";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // inform stream thread by setting stream state to STREAM_STOPPING

-  stream_.state = STREAM_STOPPING;

-

-  // wait until stream thread is stopped

-  while( stream_.state != STREAM_STOPPED ) {

-    Sleep( 1 );

-  }

-

-  // Wait for the last buffer to play before stopping.

-  Sleep( 1000 * stream_.bufferSize / stream_.sampleRate );

-

-  // stop capture client if applicable

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient ) {

-    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient->Stop();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::stopStream: Unable to stop capture stream.";

-      error( RtAudioError::DRIVER_ERROR );

-      return;

-    }

-  }

-

-  // stop render client if applicable

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient ) {

-    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient->Stop();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::stopStream: Unable to stop render stream.";

-      error( RtAudioError::DRIVER_ERROR );

-      return;

-    }

-  }

-

-  // close thread handle

-  if ( stream_.callbackInfo.thread && !CloseHandle( ( void* ) stream_.callbackInfo.thread ) ) {

-    errorText_ = "RtApiWasapi::stopStream: Unable to close callback thread.";

-    error( RtAudioError::THREAD_ERROR );

-    return;

-  }

-

-  stream_.callbackInfo.thread = (ThreadHandle) NULL;

-}

-

-//-----------------------------------------------------------------------------

-

-void RtApiWasapi::abortStream( void )

-{

-  verifyStream();

-

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiWasapi::abortStream: The stream is already stopped.";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // inform stream thread by setting stream state to STREAM_STOPPING

-  stream_.state = STREAM_STOPPING;

-

-  // wait until stream thread is stopped

-  while ( stream_.state != STREAM_STOPPED ) {

-    Sleep( 1 );

-  }

-

-  // stop capture client if applicable

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient ) {

-    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient->Stop();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::abortStream: Unable to stop capture stream.";

-      error( RtAudioError::DRIVER_ERROR );

-      return;

-    }

-  }

-

-  // stop render client if applicable

-  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient ) {

-    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient->Stop();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::abortStream: Unable to stop render stream.";

-      error( RtAudioError::DRIVER_ERROR );

-      return;

-    }

-  }

-

-  // close thread handle

-  if ( stream_.callbackInfo.thread && !CloseHandle( ( void* ) stream_.callbackInfo.thread ) ) {

-    errorText_ = "RtApiWasapi::abortStream: Unable to close callback thread.";

-    error( RtAudioError::THREAD_ERROR );

-    return;

-  }

-

-  stream_.callbackInfo.thread = (ThreadHandle) NULL;

-}

-

-//-----------------------------------------------------------------------------

-

-bool RtApiWasapi::probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                   unsigned int firstChannel, unsigned int sampleRate,

-                                   RtAudioFormat format, unsigned int* bufferSize,

-                                   RtAudio::StreamOptions* options )

-{

-  bool methodResult = FAILURE;

-  unsigned int captureDeviceCount = 0;

-  unsigned int renderDeviceCount = 0;

-

-  IMMDeviceCollection* captureDevices = NULL;

-  IMMDeviceCollection* renderDevices = NULL;

-  IMMDevice* devicePtr = NULL;

-  WAVEFORMATEX* deviceFormat = NULL;

-  unsigned int bufferBytes;

-  stream_.state = STREAM_STOPPED;

-

-  // create API Handle if not already created

-  if ( !stream_.apiHandle )

-    stream_.apiHandle = ( void* ) new WasapiHandle();

-

-  // Count capture devices

-  errorText_.clear();

-  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;

-  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device collection.";

-    goto Exit;

-  }

-

-  hr = captureDevices->GetCount( &captureDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device count.";

-    goto Exit;

-  }

-

-  // Count render devices

-  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device collection.";

-    goto Exit;

-  }

-

-  hr = renderDevices->GetCount( &renderDeviceCount );

-  if ( FAILED( hr ) ) {

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device count.";

-    goto Exit;

-  }

-

-  // validate device index

-  if ( device >= captureDeviceCount + renderDeviceCount ) {

-    errorType = RtAudioError::INVALID_USE;

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Invalid device index.";

-    goto Exit;

-  }

-

-  // determine whether index falls within capture or render devices

-  if ( device >= renderDeviceCount ) {

-    if ( mode != INPUT ) {

-      errorType = RtAudioError::INVALID_USE;

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Capture device selected as output device.";

-      goto Exit;

-    }

-

-    // retrieve captureAudioClient from devicePtr

-    IAudioClient*& captureAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient;

-

-    hr = captureDevices->Item( device - renderDeviceCount, &devicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device handle.";

-      goto Exit;

-    }

-

-    hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL,

-                              NULL, ( void** ) &captureAudioClient );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device audio client.";

-      goto Exit;

-    }

-

-    hr = captureAudioClient->GetMixFormat( &deviceFormat );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device mix format.";

-      goto Exit;

-    }

-

-    stream_.nDeviceChannels[mode] = deviceFormat->nChannels;

-    captureAudioClient->GetStreamLatency( ( long long* ) &stream_.latency[mode] );

-  }

-  else {

-    if ( mode != OUTPUT ) {

-      errorType = RtAudioError::INVALID_USE;

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Render device selected as input device.";

-      goto Exit;

-    }

-

-    // retrieve renderAudioClient from devicePtr

-    IAudioClient*& renderAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient;

-

-    hr = renderDevices->Item( device, &devicePtr );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device handle.";

-      goto Exit;

-    }

-

-    hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL,

-                              NULL, ( void** ) &renderAudioClient );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device audio client.";

-      goto Exit;

-    }

-

-    hr = renderAudioClient->GetMixFormat( &deviceFormat );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device mix format.";

-      goto Exit;

-    }

-

-    stream_.nDeviceChannels[mode] = deviceFormat->nChannels;

-    renderAudioClient->GetStreamLatency( ( long long* ) &stream_.latency[mode] );

-  }

-

-  // fill stream data

-  if ( ( stream_.mode == OUTPUT && mode == INPUT ) ||

-       ( stream_.mode == INPUT && mode == OUTPUT ) ) {

-    stream_.mode = DUPLEX;

-  }

-  else {

-    stream_.mode = mode;

-  }

-

-  stream_.device[mode] = device;

-  stream_.doByteSwap[mode] = false;

-  stream_.sampleRate = sampleRate;

-  stream_.bufferSize = *bufferSize;

-  stream_.nBuffers = 1;

-  stream_.nUserChannels[mode] = channels;

-  stream_.channelOffset[mode] = firstChannel;

-  stream_.userFormat = format;

-  stream_.deviceFormat[mode] = getDeviceInfo( device ).nativeFormats;

-

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED )

-    stream_.userInterleaved = false;

-  else

-    stream_.userInterleaved = true;

-  stream_.deviceInterleaved[mode] = true;

-

-  // Set flags for buffer conversion.

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] ||

-       stream_.nUserChannels != stream_.nDeviceChannels )

-    stream_.doConvertBuffer[mode] = true;

-  else if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-            stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  if ( stream_.doConvertBuffer[mode] )

-    setConvertInfo( mode, 0 );

-

-  // Allocate necessary internal buffers

-  bufferBytes = stream_.nUserChannels[mode] * stream_.bufferSize * formatBytes( stream_.userFormat );

-

-  stream_.userBuffer[mode] = ( char* ) calloc( bufferBytes, 1 );

-  if ( !stream_.userBuffer[mode] ) {

-    errorType = RtAudioError::MEMORY_ERROR;

-    errorText_ = "RtApiWasapi::probeDeviceOpen: Error allocating user buffer memory.";

-    goto Exit;

-  }

-

-  if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME )

-    stream_.callbackInfo.priority = 15;

-  else

-    stream_.callbackInfo.priority = 0;

-

-  ///! TODO: RTAUDIO_MINIMIZE_LATENCY // Provide stream buffers directly to callback

-  ///! TODO: RTAUDIO_HOG_DEVICE       // Exclusive mode

-

-  methodResult = SUCCESS;

-

-Exit:

-  //clean up

-  SAFE_RELEASE( captureDevices );

-  SAFE_RELEASE( renderDevices );

-  SAFE_RELEASE( devicePtr );

-  CoTaskMemFree( deviceFormat );

-

-  // if method failed, close the stream

-  if ( methodResult == FAILURE )

-    closeStream();

-

-  if ( !errorText_.empty() )

-    error( errorType );

-  return methodResult;

-}

-

-//=============================================================================

-

-DWORD WINAPI RtApiWasapi::runWasapiThread( void* wasapiPtr )

-{

-  if ( wasapiPtr )

-    ( ( RtApiWasapi* ) wasapiPtr )->wasapiThread();

-

-  return 0;

-}

-

-DWORD WINAPI RtApiWasapi::stopWasapiThread( void* wasapiPtr )

-{

-  if ( wasapiPtr )

-    ( ( RtApiWasapi* ) wasapiPtr )->stopStream();

-

-  return 0;

-}

-

-DWORD WINAPI RtApiWasapi::abortWasapiThread( void* wasapiPtr )

-{

-  if ( wasapiPtr )

-    ( ( RtApiWasapi* ) wasapiPtr )->abortStream();

-

-  return 0;

-}

-

-//-----------------------------------------------------------------------------

-

-void RtApiWasapi::wasapiThread()

-{

-  // as this is a new thread, we must CoInitialize it

-  CoInitialize( NULL );

-

-  HRESULT hr;

-

-  IAudioClient* captureAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient;

-  IAudioClient* renderAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient;

-  IAudioCaptureClient* captureClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureClient;

-  IAudioRenderClient* renderClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderClient;

-  HANDLE captureEvent = ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent;

-  HANDLE renderEvent = ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent;

-

-  WAVEFORMATEX* captureFormat = NULL;

-  WAVEFORMATEX* renderFormat = NULL;

-  float captureSrRatio = 0.0f;

-  float renderSrRatio = 0.0f;

-  WasapiBuffer captureBuffer;

-  WasapiBuffer renderBuffer;

-

-  // declare local stream variables

-  RtAudioCallback callback = ( RtAudioCallback ) stream_.callbackInfo.callback;

-  BYTE* streamBuffer = NULL;

-  unsigned long captureFlags = 0;

-  unsigned int bufferFrameCount = 0;

-  unsigned int numFramesPadding = 0;

-  unsigned int convBufferSize = 0;

-  bool callbackPushed = false;

-  bool callbackPulled = false;

-  bool callbackStopped = false;

-  int callbackResult = 0;

-

-  // convBuffer is used to store converted buffers between WASAPI and the user

-  char* convBuffer = NULL;

-  unsigned int convBuffSize = 0;

-  unsigned int deviceBuffSize = 0;

-

-  errorText_.clear();

-  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;

-

-  // Attempt to assign "Pro Audio" characteristic to thread

-  HMODULE AvrtDll = LoadLibrary( (LPCTSTR) "AVRT.dll" );

-  if ( AvrtDll ) {

-    DWORD taskIndex = 0;

-    TAvSetMmThreadCharacteristicsPtr AvSetMmThreadCharacteristicsPtr = ( TAvSetMmThreadCharacteristicsPtr ) GetProcAddress( AvrtDll, "AvSetMmThreadCharacteristicsW" );

-    AvSetMmThreadCharacteristicsPtr( L"Pro Audio", &taskIndex );

-    FreeLibrary( AvrtDll );

-  }

-

-  // start capture stream if applicable

-  if ( captureAudioClient ) {

-    hr = captureAudioClient->GetMixFormat( &captureFormat );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve device mix format.";

-      goto Exit;

-    }

-

-    captureSrRatio = ( ( float ) captureFormat->nSamplesPerSec / stream_.sampleRate );

-

-    // initialize capture stream according to desire buffer size

-    float desiredBufferSize = stream_.bufferSize * captureSrRatio;

-    REFERENCE_TIME desiredBufferPeriod = ( REFERENCE_TIME ) ( ( float ) desiredBufferSize * 10000000 / captureFormat->nSamplesPerSec );

-

-    if ( !captureClient ) {

-      hr = captureAudioClient->Initialize( AUDCLNT_SHAREMODE_SHARED,

-                                           AUDCLNT_STREAMFLAGS_EVENTCALLBACK,

-                                           desiredBufferPeriod,

-                                           desiredBufferPeriod,

-                                           captureFormat,

-                                           NULL );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to initialize capture audio client.";

-        goto Exit;

-      }

-

-      hr = captureAudioClient->GetService( __uuidof( IAudioCaptureClient ),

-                                           ( void** ) &captureClient );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve capture client handle.";

-        goto Exit;

-      }

-

-      // configure captureEvent to trigger on every available capture buffer

-      captureEvent = CreateEvent( NULL, FALSE, FALSE, NULL );

-      if ( !captureEvent ) {

-        errorType = RtAudioError::SYSTEM_ERROR;

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to create capture event.";

-        goto Exit;

-      }

-

-      hr = captureAudioClient->SetEventHandle( captureEvent );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to set capture event handle.";

-        goto Exit;

-      }

-

-      ( ( WasapiHandle* ) stream_.apiHandle )->captureClient = captureClient;

-      ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent = captureEvent;

-    }

-

-    unsigned int inBufferSize = 0;

-    hr = captureAudioClient->GetBufferSize( &inBufferSize );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to get capture buffer size.";

-      goto Exit;

-    }

-

-    // scale outBufferSize according to stream->user sample rate ratio

-    unsigned int outBufferSize = ( unsigned int ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT];

-    inBufferSize *= stream_.nDeviceChannels[INPUT];

-

-    // set captureBuffer size

-    captureBuffer.setBufferSize( inBufferSize + outBufferSize, formatBytes( stream_.deviceFormat[INPUT] ) );

-

-    // reset the capture stream

-    hr = captureAudioClient->Reset();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to reset capture stream.";

-      goto Exit;

-    }

-

-    // start the capture stream

-    hr = captureAudioClient->Start();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to start capture stream.";

-      goto Exit;

-    }

-  }

-

-  // start render stream if applicable

-  if ( renderAudioClient ) {

-    hr = renderAudioClient->GetMixFormat( &renderFormat );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve device mix format.";

-      goto Exit;

-    }

-

-    renderSrRatio = ( ( float ) renderFormat->nSamplesPerSec / stream_.sampleRate );

-

-    // initialize render stream according to desire buffer size

-    float desiredBufferSize = stream_.bufferSize * renderSrRatio;

-    REFERENCE_TIME desiredBufferPeriod = ( REFERENCE_TIME ) ( ( float ) desiredBufferSize * 10000000 / renderFormat->nSamplesPerSec );

-

-    if ( !renderClient ) {

-      hr = renderAudioClient->Initialize( AUDCLNT_SHAREMODE_SHARED,

-                                          AUDCLNT_STREAMFLAGS_EVENTCALLBACK,

-                                          desiredBufferPeriod,

-                                          desiredBufferPeriod,

-                                          renderFormat,

-                                          NULL );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to initialize render audio client.";

-        goto Exit;

-      }

-

-      hr = renderAudioClient->GetService( __uuidof( IAudioRenderClient ),

-                                          ( void** ) &renderClient );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render client handle.";

-        goto Exit;

-      }

-

-      // configure renderEvent to trigger on every available render buffer

-      renderEvent = CreateEvent( NULL, FALSE, FALSE, NULL );

-      if ( !renderEvent ) {

-        errorType = RtAudioError::SYSTEM_ERROR;

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to create render event.";

-        goto Exit;

-      }

-

-      hr = renderAudioClient->SetEventHandle( renderEvent );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to set render event handle.";

-        goto Exit;

-      }

-

-      ( ( WasapiHandle* ) stream_.apiHandle )->renderClient = renderClient;

-      ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent = renderEvent;

-    }

-

-    unsigned int outBufferSize = 0;

-    hr = renderAudioClient->GetBufferSize( &outBufferSize );

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to get render buffer size.";

-      goto Exit;

-    }

-

-    // scale inBufferSize according to user->stream sample rate ratio

-    unsigned int inBufferSize = ( unsigned int ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT];

-    outBufferSize *= stream_.nDeviceChannels[OUTPUT];

-

-    // set renderBuffer size

-    renderBuffer.setBufferSize( inBufferSize + outBufferSize, formatBytes( stream_.deviceFormat[OUTPUT] ) );

-

-    // reset the render stream

-    hr = renderAudioClient->Reset();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to reset render stream.";

-      goto Exit;

-    }

-

-    // start the render stream

-    hr = renderAudioClient->Start();

-    if ( FAILED( hr ) ) {

-      errorText_ = "RtApiWasapi::wasapiThread: Unable to start render stream.";

-      goto Exit;

-    }

-  }

-

-  if ( stream_.mode == INPUT ) {

-    convBuffSize = ( size_t ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] );

-    deviceBuffSize = stream_.bufferSize * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] );

-  }

-  else if ( stream_.mode == OUTPUT ) {

-    convBuffSize = ( size_t ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] );

-    deviceBuffSize = stream_.bufferSize * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] );

-  }

-  else if ( stream_.mode == DUPLEX ) {

-    convBuffSize = std::max( ( size_t ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] ),

-                             ( size_t ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] ) );

-    deviceBuffSize = std::max( stream_.bufferSize * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] ),

-                               stream_.bufferSize * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] ) );

-  }

-

-  convBuffer = ( char* ) malloc( convBuffSize );

-  stream_.deviceBuffer = ( char* ) malloc( deviceBuffSize );

-  if ( !convBuffer || !stream_.deviceBuffer ) {

-    errorType = RtAudioError::MEMORY_ERROR;

-    errorText_ = "RtApiWasapi::wasapiThread: Error allocating device buffer memory.";

-    goto Exit;

-  }

-

-  // stream process loop

-  while ( stream_.state != STREAM_STOPPING ) {

-    if ( !callbackPulled ) {

-      // Callback Input

-      // ==============

-      // 1. Pull callback buffer from inputBuffer

-      // 2. If 1. was successful: Convert callback buffer to user sample rate and channel count

-      //                          Convert callback buffer to user format

-

-      if ( captureAudioClient ) {

-        // Pull callback buffer from inputBuffer

-        callbackPulled = captureBuffer.pullBuffer( convBuffer,

-                                                   ( unsigned int ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT],

-                                                   stream_.deviceFormat[INPUT] );

-

-        if ( callbackPulled ) {

-          // Convert callback buffer to user sample rate

-          convertBufferWasapi( stream_.deviceBuffer,

-                               convBuffer,

-                               stream_.nDeviceChannels[INPUT],

-                               captureFormat->nSamplesPerSec,

-                               stream_.sampleRate,

-                               ( unsigned int ) ( stream_.bufferSize * captureSrRatio ),

-                               convBufferSize,

-                               stream_.deviceFormat[INPUT] );

-

-          if ( stream_.doConvertBuffer[INPUT] ) {

-            // Convert callback buffer to user format

-            convertBuffer( stream_.userBuffer[INPUT],

-                           stream_.deviceBuffer,

-                           stream_.convertInfo[INPUT] );

-          }

-          else {

-            // no further conversion, simple copy deviceBuffer to userBuffer

-            memcpy( stream_.userBuffer[INPUT],

-                    stream_.deviceBuffer,

-                    stream_.bufferSize * stream_.nUserChannels[INPUT] * formatBytes( stream_.userFormat ) );

-          }

-        }

-      }

-      else {

-        // if there is no capture stream, set callbackPulled flag

-        callbackPulled = true;

-      }

-

-      // Execute Callback

-      // ================

-      // 1. Execute user callback method

-      // 2. Handle return value from callback

-

-      // if callback has not requested the stream to stop

-      if ( callbackPulled && !callbackStopped ) {

-        // Execute user callback method

-        callbackResult = callback( stream_.userBuffer[OUTPUT],

-                                   stream_.userBuffer[INPUT],

-                                   stream_.bufferSize,

-                                   getStreamTime(),

-                                   captureFlags & AUDCLNT_BUFFERFLAGS_DATA_DISCONTINUITY ? RTAUDIO_INPUT_OVERFLOW : 0,

-                                   stream_.callbackInfo.userData );

-

-        // Handle return value from callback

-        if ( callbackResult == 1 ) {

-          // instantiate a thread to stop this thread

-          HANDLE threadHandle = CreateThread( NULL, 0, stopWasapiThread, this, 0, NULL );

-          if ( !threadHandle ) {

-            errorType = RtAudioError::THREAD_ERROR;

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to instantiate stream stop thread.";

-            goto Exit;

-          }

-          else if ( !CloseHandle( threadHandle ) ) {

-            errorType = RtAudioError::THREAD_ERROR;

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to close stream stop thread handle.";

-            goto Exit;

-          }

-

-          callbackStopped = true;

-        }

-        else if ( callbackResult == 2 ) {

-          // instantiate a thread to stop this thread

-          HANDLE threadHandle = CreateThread( NULL, 0, abortWasapiThread, this, 0, NULL );

-          if ( !threadHandle ) {

-            errorType = RtAudioError::THREAD_ERROR;

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to instantiate stream abort thread.";

-            goto Exit;

-          }

-          else if ( !CloseHandle( threadHandle ) ) {

-            errorType = RtAudioError::THREAD_ERROR;

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to close stream abort thread handle.";

-            goto Exit;

-          }

-

-          callbackStopped = true;

-        }

-      }

-    }

-

-    // Callback Output

-    // ===============

-    // 1. Convert callback buffer to stream format

-    // 2. Convert callback buffer to stream sample rate and channel count

-    // 3. Push callback buffer into outputBuffer

-

-    if ( renderAudioClient && callbackPulled ) {

-      if ( stream_.doConvertBuffer[OUTPUT] ) {

-        // Convert callback buffer to stream format

-        convertBuffer( stream_.deviceBuffer,

-                       stream_.userBuffer[OUTPUT],

-                       stream_.convertInfo[OUTPUT] );

-

-      }

-

-      // Convert callback buffer to stream sample rate

-      convertBufferWasapi( convBuffer,

-                           stream_.deviceBuffer,

-                           stream_.nDeviceChannels[OUTPUT],

-                           stream_.sampleRate,

-                           renderFormat->nSamplesPerSec,

-                           stream_.bufferSize,

-                           convBufferSize,

-                           stream_.deviceFormat[OUTPUT] );

-

-      // Push callback buffer into outputBuffer

-      callbackPushed = renderBuffer.pushBuffer( convBuffer,

-                                                convBufferSize * stream_.nDeviceChannels[OUTPUT],

-                                                stream_.deviceFormat[OUTPUT] );

-    }

-    else {

-      // if there is no render stream, set callbackPushed flag

-      callbackPushed = true;

-    }

-

-    // Stream Capture

-    // ==============

-    // 1. Get capture buffer from stream

-    // 2. Push capture buffer into inputBuffer

-    // 3. If 2. was successful: Release capture buffer

-

-    if ( captureAudioClient ) {

-      // if the callback input buffer was not pulled from captureBuffer, wait for next capture event

-      if ( !callbackPulled ) {

-        WaitForSingleObject( captureEvent, INFINITE );

-      }

-

-      // Get capture buffer from stream

-      hr = captureClient->GetBuffer( &streamBuffer,

-                                     &bufferFrameCount,

-                                     &captureFlags, NULL, NULL );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve capture buffer.";

-        goto Exit;

-      }

-

-      if ( bufferFrameCount != 0 ) {

-        // Push capture buffer into inputBuffer

-        if ( captureBuffer.pushBuffer( ( char* ) streamBuffer,

-                                       bufferFrameCount * stream_.nDeviceChannels[INPUT],

-                                       stream_.deviceFormat[INPUT] ) )

-        {

-          // Release capture buffer

-          hr = captureClient->ReleaseBuffer( bufferFrameCount );

-          if ( FAILED( hr ) ) {

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";

-            goto Exit;

-          }

-        }

-        else

-        {

-          // Inform WASAPI that capture was unsuccessful

-          hr = captureClient->ReleaseBuffer( 0 );

-          if ( FAILED( hr ) ) {

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";

-            goto Exit;

-          }

-        }

-      }

-      else

-      {

-        // Inform WASAPI that capture was unsuccessful

-        hr = captureClient->ReleaseBuffer( 0 );

-        if ( FAILED( hr ) ) {

-          errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";

-          goto Exit;

-        }

-      }

-    }

-

-    // Stream Render

-    // =============

-    // 1. Get render buffer from stream

-    // 2. Pull next buffer from outputBuffer

-    // 3. If 2. was successful: Fill render buffer with next buffer

-    //                          Release render buffer

-

-    if ( renderAudioClient ) {

-      // if the callback output buffer was not pushed to renderBuffer, wait for next render event

-      if ( callbackPulled && !callbackPushed ) {

-        WaitForSingleObject( renderEvent, INFINITE );

-      }

-

-      // Get render buffer from stream

-      hr = renderAudioClient->GetBufferSize( &bufferFrameCount );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer size.";

-        goto Exit;

-      }

-

-      hr = renderAudioClient->GetCurrentPadding( &numFramesPadding );

-      if ( FAILED( hr ) ) {

-        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer padding.";

-        goto Exit;

-      }

-

-      bufferFrameCount -= numFramesPadding;

-

-      if ( bufferFrameCount != 0 ) {

-        hr = renderClient->GetBuffer( bufferFrameCount, &streamBuffer );

-        if ( FAILED( hr ) ) {

-          errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer.";

-          goto Exit;

-        }

-

-        // Pull next buffer from outputBuffer

-        // Fill render buffer with next buffer

-        if ( renderBuffer.pullBuffer( ( char* ) streamBuffer,

-                                      bufferFrameCount * stream_.nDeviceChannels[OUTPUT],

-                                      stream_.deviceFormat[OUTPUT] ) )

-        {

-          // Release render buffer

-          hr = renderClient->ReleaseBuffer( bufferFrameCount, 0 );

-          if ( FAILED( hr ) ) {

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";

-            goto Exit;

-          }

-        }

-        else

-        {

-          // Inform WASAPI that render was unsuccessful

-          hr = renderClient->ReleaseBuffer( 0, 0 );

-          if ( FAILED( hr ) ) {

-            errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";

-            goto Exit;

-          }

-        }

-      }

-      else

-      {

-        // Inform WASAPI that render was unsuccessful

-        hr = renderClient->ReleaseBuffer( 0, 0 );

-        if ( FAILED( hr ) ) {

-          errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";

-          goto Exit;

-        }

-      }

-    }

-

-    // if the callback buffer was pushed renderBuffer reset callbackPulled flag

-    if ( callbackPushed ) {

-      callbackPulled = false;

-    }

-

-    // tick stream time

-    RtApi::tickStreamTime();

-  }

-

-Exit:

-  // clean up

-  CoTaskMemFree( captureFormat );

-  CoTaskMemFree( renderFormat );

-

-  free ( convBuffer );

-

-  CoUninitialize();

-

-  // update stream state

-  stream_.state = STREAM_STOPPED;

-

-  if ( errorText_.empty() )

-    return;

-  else

-    error( errorType );

-}

-

-//******************** End of __WINDOWS_WASAPI__ *********************//

-#endif

-

-

-#if defined(__WINDOWS_DS__) // Windows DirectSound API

-

-// Modified by Robin Davies, October 2005

-// - Improvements to DirectX pointer chasing. 

-// - Bug fix for non-power-of-two Asio granularity used by Edirol PCR-A30.

-// - Auto-call CoInitialize for DSOUND and ASIO platforms.

-// Various revisions for RtAudio 4.0 by Gary Scavone, April 2007

-// Changed device query structure for RtAudio 4.0.7, January 2010

-

-#include <dsound.h>

-#include <assert.h>

-#include <algorithm>

-

-#if defined(__MINGW32__)

-  // missing from latest mingw winapi

-#define WAVE_FORMAT_96M08 0x00010000 /* 96 kHz, Mono, 8-bit */

-#define WAVE_FORMAT_96S08 0x00020000 /* 96 kHz, Stereo, 8-bit */

-#define WAVE_FORMAT_96M16 0x00040000 /* 96 kHz, Mono, 16-bit */

-#define WAVE_FORMAT_96S16 0x00080000 /* 96 kHz, Stereo, 16-bit */

-#endif

-

-#define MINIMUM_DEVICE_BUFFER_SIZE 32768

-

-#ifdef _MSC_VER // if Microsoft Visual C++

-#pragma comment( lib, "winmm.lib" ) // then, auto-link winmm.lib. Otherwise, it has to be added manually.

-#endif

-

-static inline DWORD dsPointerBetween( DWORD pointer, DWORD laterPointer, DWORD earlierPointer, DWORD bufferSize )

-{

-  if ( pointer > bufferSize ) pointer -= bufferSize;

-  if ( laterPointer < earlierPointer ) laterPointer += bufferSize;

-  if ( pointer < earlierPointer ) pointer += bufferSize;

-  return pointer >= earlierPointer && pointer < laterPointer;

-}

-

-// A structure to hold various information related to the DirectSound

-// API implementation.

-struct DsHandle {

-  unsigned int drainCounter; // Tracks callback counts when draining

-  bool internalDrain;        // Indicates if stop is initiated from callback or not.

-  void *id[2];

-  void *buffer[2];

-  bool xrun[2];

-  UINT bufferPointer[2];  

-  DWORD dsBufferSize[2];

-  DWORD dsPointerLeadTime[2]; // the number of bytes ahead of the safe pointer to lead by.

-  HANDLE condition;

-

-  DsHandle()

-    :drainCounter(0), internalDrain(false) { id[0] = 0; id[1] = 0; buffer[0] = 0; buffer[1] = 0; xrun[0] = false; xrun[1] = false; bufferPointer[0] = 0; bufferPointer[1] = 0; }

-};

-

-// Declarations for utility functions, callbacks, and structures

-// specific to the DirectSound implementation.

-static BOOL CALLBACK deviceQueryCallback( LPGUID lpguid,

-                                          LPCTSTR description,

-                                          LPCTSTR module,

-                                          LPVOID lpContext );

-

-static const char* getErrorString( int code );

-

-static unsigned __stdcall callbackHandler( void *ptr );

-

-struct DsDevice {

-  LPGUID id[2];

-  bool validId[2];

-  bool found;

-  std::string name;

-

-  DsDevice()

-  : found(false) { validId[0] = false; validId[1] = false; }

-};

-

-struct DsProbeData {

-  bool isInput;

-  std::vector<struct DsDevice>* dsDevices;

-};

-

-RtApiDs :: RtApiDs()

-{

-  // Dsound will run both-threaded. If CoInitialize fails, then just

-  // accept whatever the mainline chose for a threading model.

-  coInitialized_ = false;

-  HRESULT hr = CoInitialize( NULL );

-  if ( !FAILED( hr ) ) coInitialized_ = true;

-}

-

-RtApiDs :: ~RtApiDs()

-{

-  if ( coInitialized_ ) CoUninitialize(); // balanced call.

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-}

-

-// The DirectSound default output is always the first device.

-unsigned int RtApiDs :: getDefaultOutputDevice( void )

-{

-  return 0;

-}

-

-// The DirectSound default input is always the first input device,

-// which is the first capture device enumerated.

-unsigned int RtApiDs :: getDefaultInputDevice( void )

-{

-  return 0;

-}

-

-unsigned int RtApiDs :: getDeviceCount( void )

-{

-  // Set query flag for previously found devices to false, so that we

-  // can check for any devices that have disappeared.

-  for ( unsigned int i=0; i<dsDevices.size(); i++ )

-    dsDevices[i].found = false;

-

-  // Query DirectSound devices.

-  struct DsProbeData probeInfo;

-  probeInfo.isInput = false;

-  probeInfo.dsDevices = &dsDevices;

-  HRESULT result = DirectSoundEnumerate( (LPDSENUMCALLBACK) deviceQueryCallback, &probeInfo );

-  if ( FAILED( result ) ) {

-    errorStream_ << "RtApiDs::getDeviceCount: error (" << getErrorString( result ) << ") enumerating output devices!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-  }

-

-  // Query DirectSoundCapture devices.

-  probeInfo.isInput = true;

-  result = DirectSoundCaptureEnumerate( (LPDSENUMCALLBACK) deviceQueryCallback, &probeInfo );

-  if ( FAILED( result ) ) {

-    errorStream_ << "RtApiDs::getDeviceCount: error (" << getErrorString( result ) << ") enumerating input devices!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-  }

-

-  // Clean out any devices that may have disappeared (code update submitted by Eli Zehngut).

-  for ( unsigned int i=0; i<dsDevices.size(); ) {

-    if ( dsDevices[i].found == false ) dsDevices.erase( dsDevices.begin() + i );

-    else i++;

-  }

-

-  return static_cast<unsigned int>(dsDevices.size());

-}

-

-RtAudio::DeviceInfo RtApiDs :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  if ( dsDevices.size() == 0 ) {

-    // Force a query of all devices

-    getDeviceCount();

-    if ( dsDevices.size() == 0 ) {

-      errorText_ = "RtApiDs::getDeviceInfo: no devices found!";

-      error( RtAudioError::INVALID_USE );

-      return info;

-    }

-  }

-

-  if ( device >= dsDevices.size() ) {

-    errorText_ = "RtApiDs::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  HRESULT result;

-  if ( dsDevices[ device ].validId[0] == false ) goto probeInput;

-

-  LPDIRECTSOUND output;

-  DSCAPS outCaps;

-  result = DirectSoundCreate( dsDevices[ device ].id[0], &output, NULL );

-  if ( FAILED( result ) ) {

-    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") opening output device (" << dsDevices[ device ].name << ")!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    goto probeInput;

-  }

-

-  outCaps.dwSize = sizeof( outCaps );

-  result = output->GetCaps( &outCaps );

-  if ( FAILED( result ) ) {

-    output->Release();

-    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") getting capabilities!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    goto probeInput;

-  }

-

-  // Get output channel information.

-  info.outputChannels = ( outCaps.dwFlags & DSCAPS_PRIMARYSTEREO ) ? 2 : 1;

-

-  // Get sample rate information.

-  info.sampleRates.clear();

-  for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {

-    if ( SAMPLE_RATES[k] >= (unsigned int) outCaps.dwMinSecondarySampleRate &&

-         SAMPLE_RATES[k] <= (unsigned int) outCaps.dwMaxSecondarySampleRate ) {

-      info.sampleRates.push_back( SAMPLE_RATES[k] );

-

-      if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )

-        info.preferredSampleRate = SAMPLE_RATES[k];

-    }

-  }

-

-  // Get format information.

-  if ( outCaps.dwFlags & DSCAPS_PRIMARY16BIT ) info.nativeFormats |= RTAUDIO_SINT16;

-  if ( outCaps.dwFlags & DSCAPS_PRIMARY8BIT ) info.nativeFormats |= RTAUDIO_SINT8;

-

-  output->Release();

-

-  if ( getDefaultOutputDevice() == device )

-    info.isDefaultOutput = true;

-

-  if ( dsDevices[ device ].validId[1] == false ) {

-    info.name = dsDevices[ device ].name;

-    info.probed = true;

-    return info;

-  }

-

- probeInput:

-

-  LPDIRECTSOUNDCAPTURE input;

-  result = DirectSoundCaptureCreate( dsDevices[ device ].id[1], &input, NULL );

-  if ( FAILED( result ) ) {

-    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") opening input device (" << dsDevices[ device ].name << ")!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  DSCCAPS inCaps;

-  inCaps.dwSize = sizeof( inCaps );

-  result = input->GetCaps( &inCaps );

-  if ( FAILED( result ) ) {

-    input->Release();

-    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") getting object capabilities (" << dsDevices[ device ].name << ")!";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Get input channel information.

-  info.inputChannels = inCaps.dwChannels;

-

-  // Get sample rate and format information.

-  std::vector<unsigned int> rates;

-  if ( inCaps.dwChannels >= 2 ) {

-    if ( inCaps.dwFormats & WAVE_FORMAT_1S16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_2S16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_4S16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_96S16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_1S08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_2S08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_4S08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_96S08 ) info.nativeFormats |= RTAUDIO_SINT8;

-

-    if ( info.nativeFormats & RTAUDIO_SINT16 ) {

-      if ( inCaps.dwFormats & WAVE_FORMAT_1S16 ) rates.push_back( 11025 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_2S16 ) rates.push_back( 22050 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_4S16 ) rates.push_back( 44100 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_96S16 ) rates.push_back( 96000 );

-    }

-    else if ( info.nativeFormats & RTAUDIO_SINT8 ) {

-      if ( inCaps.dwFormats & WAVE_FORMAT_1S08 ) rates.push_back( 11025 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_2S08 ) rates.push_back( 22050 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_4S08 ) rates.push_back( 44100 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_96S08 ) rates.push_back( 96000 );

-    }

-  }

-  else if ( inCaps.dwChannels == 1 ) {

-    if ( inCaps.dwFormats & WAVE_FORMAT_1M16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_2M16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_4M16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_96M16 ) info.nativeFormats |= RTAUDIO_SINT16;

-    if ( inCaps.dwFormats & WAVE_FORMAT_1M08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_2M08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_4M08 ) info.nativeFormats |= RTAUDIO_SINT8;

-    if ( inCaps.dwFormats & WAVE_FORMAT_96M08 ) info.nativeFormats |= RTAUDIO_SINT8;

-

-    if ( info.nativeFormats & RTAUDIO_SINT16 ) {

-      if ( inCaps.dwFormats & WAVE_FORMAT_1M16 ) rates.push_back( 11025 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_2M16 ) rates.push_back( 22050 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_4M16 ) rates.push_back( 44100 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_96M16 ) rates.push_back( 96000 );

-    }

-    else if ( info.nativeFormats & RTAUDIO_SINT8 ) {

-      if ( inCaps.dwFormats & WAVE_FORMAT_1M08 ) rates.push_back( 11025 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_2M08 ) rates.push_back( 22050 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_4M08 ) rates.push_back( 44100 );

-      if ( inCaps.dwFormats & WAVE_FORMAT_96M08 ) rates.push_back( 96000 );

-    }

-  }

-  else info.inputChannels = 0; // technically, this would be an error

-

-  input->Release();

-

-  if ( info.inputChannels == 0 ) return info;

-

-  // Copy the supported rates to the info structure but avoid duplication.

-  bool found;

-  for ( unsigned int i=0; i<rates.size(); i++ ) {

-    found = false;

-    for ( unsigned int j=0; j<info.sampleRates.size(); j++ ) {

-      if ( rates[i] == info.sampleRates[j] ) {

-        found = true;

-        break;

-      }

-    }

-    if ( found == false ) info.sampleRates.push_back( rates[i] );

-  }

-  std::sort( info.sampleRates.begin(), info.sampleRates.end() );

-

-  // If device opens for both playback and capture, we determine the channels.

-  if ( info.outputChannels > 0 && info.inputChannels > 0 )

-    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-

-  if ( device == 0 ) info.isDefaultInput = true;

-

-  // Copy name and return.

-  info.name = dsDevices[ device ].name;

-  info.probed = true;

-  return info;

-}

-

-bool RtApiDs :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                 unsigned int firstChannel, unsigned int sampleRate,

-                                 RtAudioFormat format, unsigned int *bufferSize,

-                                 RtAudio::StreamOptions *options )

-{

-  if ( channels + firstChannel > 2 ) {

-    errorText_ = "RtApiDs::probeDeviceOpen: DirectSound does not support more than 2 channels per device.";

-    return FAILURE;

-  }

-

-  size_t nDevices = dsDevices.size();

-  if ( nDevices == 0 ) {

-    // This should not happen because a check is made before this function is called.

-    errorText_ = "RtApiDs::probeDeviceOpen: no devices found!";

-    return FAILURE;

-  }

-

-  if ( device >= nDevices ) {

-    // This should not happen because a check is made before this function is called.

-    errorText_ = "RtApiDs::probeDeviceOpen: device ID is invalid!";

-    return FAILURE;

-  }

-

-  if ( mode == OUTPUT ) {

-    if ( dsDevices[ device ].validId[0] == false ) {

-      errorStream_ << "RtApiDs::probeDeviceOpen: device (" << device << ") does not support output!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-  else { // mode == INPUT

-    if ( dsDevices[ device ].validId[1] == false ) {

-      errorStream_ << "RtApiDs::probeDeviceOpen: device (" << device << ") does not support input!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-

-  // According to a note in PortAudio, using GetDesktopWindow()

-  // instead of GetForegroundWindow() is supposed to avoid problems

-  // that occur when the application's window is not the foreground

-  // window.  Also, if the application window closes before the

-  // DirectSound buffer, DirectSound can crash.  In the past, I had

-  // problems when using GetDesktopWindow() but it seems fine now

-  // (January 2010).  I'll leave it commented here.

-  // HWND hWnd = GetForegroundWindow();

-  HWND hWnd = GetDesktopWindow();

-

-  // Check the numberOfBuffers parameter and limit the lowest value to

-  // two.  This is a judgement call and a value of two is probably too

-  // low for capture, but it should work for playback.

-  int nBuffers = 0;

-  if ( options ) nBuffers = options->numberOfBuffers;

-  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) nBuffers = 2;

-  if ( nBuffers < 2 ) nBuffers = 3;

-

-  // Check the lower range of the user-specified buffer size and set

-  // (arbitrarily) to a lower bound of 32.

-  if ( *bufferSize < 32 ) *bufferSize = 32;

-

-  // Create the wave format structure.  The data format setting will

-  // be determined later.

-  WAVEFORMATEX waveFormat;

-  ZeroMemory( &waveFormat, sizeof(WAVEFORMATEX) );

-  waveFormat.wFormatTag = WAVE_FORMAT_PCM;

-  waveFormat.nChannels = channels + firstChannel;

-  waveFormat.nSamplesPerSec = (unsigned long) sampleRate;

-

-  // Determine the device buffer size. By default, we'll use the value

-  // defined above (32K), but we will grow it to make allowances for

-  // very large software buffer sizes.

-  DWORD dsBufferSize = MINIMUM_DEVICE_BUFFER_SIZE;

-  DWORD dsPointerLeadTime = 0;

-

-  void *ohandle = 0, *bhandle = 0;

-  HRESULT result;

-  if ( mode == OUTPUT ) {

-

-    LPDIRECTSOUND output;

-    result = DirectSoundCreate( dsDevices[ device ].id[0], &output, NULL );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") opening output device (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    DSCAPS outCaps;

-    outCaps.dwSize = sizeof( outCaps );

-    result = output->GetCaps( &outCaps );

-    if ( FAILED( result ) ) {

-      output->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting capabilities (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Check channel information.

-    if ( channels + firstChannel == 2 && !( outCaps.dwFlags & DSCAPS_PRIMARYSTEREO ) ) {

-      errorStream_ << "RtApiDs::getDeviceInfo: the output device (" << dsDevices[ device ].name << ") does not support stereo playback.";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Check format information.  Use 16-bit format unless not

-    // supported or user requests 8-bit.

-    if ( outCaps.dwFlags & DSCAPS_PRIMARY16BIT &&

-         !( format == RTAUDIO_SINT8 && outCaps.dwFlags & DSCAPS_PRIMARY8BIT ) ) {

-      waveFormat.wBitsPerSample = 16;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-    }

-    else {

-      waveFormat.wBitsPerSample = 8;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-    }

-    stream_.userFormat = format;

-

-    // Update wave format structure and buffer information.

-    waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;

-    waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;

-    dsPointerLeadTime = nBuffers * (*bufferSize) * (waveFormat.wBitsPerSample / 8) * channels;

-

-    // If the user wants an even bigger buffer, increase the device buffer size accordingly.

-    while ( dsPointerLeadTime * 2U > dsBufferSize )

-      dsBufferSize *= 2;

-

-    // Set cooperative level to DSSCL_EXCLUSIVE ... sound stops when window focus changes.

-    // result = output->SetCooperativeLevel( hWnd, DSSCL_EXCLUSIVE );

-    // Set cooperative level to DSSCL_PRIORITY ... sound remains when window focus changes.

-    result = output->SetCooperativeLevel( hWnd, DSSCL_PRIORITY );

-    if ( FAILED( result ) ) {

-      output->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") setting cooperative level (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Even though we will write to the secondary buffer, we need to

-    // access the primary buffer to set the correct output format

-    // (since the default is 8-bit, 22 kHz!).  Setup the DS primary

-    // buffer description.

-    DSBUFFERDESC bufferDescription;

-    ZeroMemory( &bufferDescription, sizeof( DSBUFFERDESC ) );

-    bufferDescription.dwSize = sizeof( DSBUFFERDESC );

-    bufferDescription.dwFlags = DSBCAPS_PRIMARYBUFFER;

-

-    // Obtain the primary buffer

-    LPDIRECTSOUNDBUFFER buffer;

-    result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );

-    if ( FAILED( result ) ) {

-      output->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") accessing primary buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Set the primary DS buffer sound format.

-    result = buffer->SetFormat( &waveFormat );

-    if ( FAILED( result ) ) {

-      output->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") setting primary buffer format (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Setup the secondary DS buffer description.

-    ZeroMemory( &bufferDescription, sizeof( DSBUFFERDESC ) );

-    bufferDescription.dwSize = sizeof( DSBUFFERDESC );

-    bufferDescription.dwFlags = ( DSBCAPS_STICKYFOCUS |

-                                  DSBCAPS_GLOBALFOCUS |

-                                  DSBCAPS_GETCURRENTPOSITION2 |

-                                  DSBCAPS_LOCHARDWARE );  // Force hardware mixing

-    bufferDescription.dwBufferBytes = dsBufferSize;

-    bufferDescription.lpwfxFormat = &waveFormat;

-

-    // Try to create the secondary DS buffer.  If that doesn't work,

-    // try to use software mixing.  Otherwise, there's a problem.

-    result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );

-    if ( FAILED( result ) ) {

-      bufferDescription.dwFlags = ( DSBCAPS_STICKYFOCUS |

-                                    DSBCAPS_GLOBALFOCUS |

-                                    DSBCAPS_GETCURRENTPOSITION2 |

-                                    DSBCAPS_LOCSOFTWARE );  // Force software mixing

-      result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );

-      if ( FAILED( result ) ) {

-        output->Release();

-        errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") creating secondary buffer (" << dsDevices[ device ].name << ")!";

-        errorText_ = errorStream_.str();

-        return FAILURE;

-      }

-    }

-

-    // Get the buffer size ... might be different from what we specified.

-    DSBCAPS dsbcaps;

-    dsbcaps.dwSize = sizeof( DSBCAPS );

-    result = buffer->GetCaps( &dsbcaps );

-    if ( FAILED( result ) ) {

-      output->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting buffer settings (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    dsBufferSize = dsbcaps.dwBufferBytes;

-

-    // Lock the DS buffer

-    LPVOID audioPtr;

-    DWORD dataLen;

-    result = buffer->Lock( 0, dsBufferSize, &audioPtr, &dataLen, NULL, NULL, 0 );

-    if ( FAILED( result ) ) {

-      output->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") locking buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Zero the DS buffer

-    ZeroMemory( audioPtr, dataLen );

-

-    // Unlock the DS buffer

-    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );

-    if ( FAILED( result ) ) {

-      output->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") unlocking buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    ohandle = (void *) output;

-    bhandle = (void *) buffer;

-  }

-

-  if ( mode == INPUT ) {

-

-    LPDIRECTSOUNDCAPTURE input;

-    result = DirectSoundCaptureCreate( dsDevices[ device ].id[1], &input, NULL );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") opening input device (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    DSCCAPS inCaps;

-    inCaps.dwSize = sizeof( inCaps );

-    result = input->GetCaps( &inCaps );

-    if ( FAILED( result ) ) {

-      input->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting input capabilities (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Check channel information.

-    if ( inCaps.dwChannels < channels + firstChannel ) {

-      errorText_ = "RtApiDs::getDeviceInfo: the input device does not support requested input channels.";

-      return FAILURE;

-    }

-

-    // Check format information.  Use 16-bit format unless user

-    // requests 8-bit.

-    DWORD deviceFormats;

-    if ( channels + firstChannel == 2 ) {

-      deviceFormats = WAVE_FORMAT_1S08 | WAVE_FORMAT_2S08 | WAVE_FORMAT_4S08 | WAVE_FORMAT_96S08;

-      if ( format == RTAUDIO_SINT8 && inCaps.dwFormats & deviceFormats ) {

-        waveFormat.wBitsPerSample = 8;

-        stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-      }

-      else { // assume 16-bit is supported

-        waveFormat.wBitsPerSample = 16;

-        stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-      }

-    }

-    else { // channel == 1

-      deviceFormats = WAVE_FORMAT_1M08 | WAVE_FORMAT_2M08 | WAVE_FORMAT_4M08 | WAVE_FORMAT_96M08;

-      if ( format == RTAUDIO_SINT8 && inCaps.dwFormats & deviceFormats ) {

-        waveFormat.wBitsPerSample = 8;

-        stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-      }

-      else { // assume 16-bit is supported

-        waveFormat.wBitsPerSample = 16;

-        stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-      }

-    }

-    stream_.userFormat = format;

-

-    // Update wave format structure and buffer information.

-    waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;

-    waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;

-    dsPointerLeadTime = nBuffers * (*bufferSize) * (waveFormat.wBitsPerSample / 8) * channels;

-

-    // If the user wants an even bigger buffer, increase the device buffer size accordingly.

-    while ( dsPointerLeadTime * 2U > dsBufferSize )

-      dsBufferSize *= 2;

-

-    // Setup the secondary DS buffer description.

-    DSCBUFFERDESC bufferDescription;

-    ZeroMemory( &bufferDescription, sizeof( DSCBUFFERDESC ) );

-    bufferDescription.dwSize = sizeof( DSCBUFFERDESC );

-    bufferDescription.dwFlags = 0;

-    bufferDescription.dwReserved = 0;

-    bufferDescription.dwBufferBytes = dsBufferSize;

-    bufferDescription.lpwfxFormat = &waveFormat;

-

-    // Create the capture buffer.

-    LPDIRECTSOUNDCAPTUREBUFFER buffer;

-    result = input->CreateCaptureBuffer( &bufferDescription, &buffer, NULL );

-    if ( FAILED( result ) ) {

-      input->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") creating input buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Get the buffer size ... might be different from what we specified.

-    DSCBCAPS dscbcaps;

-    dscbcaps.dwSize = sizeof( DSCBCAPS );

-    result = buffer->GetCaps( &dscbcaps );

-    if ( FAILED( result ) ) {

-      input->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting buffer settings (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    dsBufferSize = dscbcaps.dwBufferBytes;

-

-    // NOTE: We could have a problem here if this is a duplex stream

-    // and the play and capture hardware buffer sizes are different

-    // (I'm actually not sure if that is a problem or not).

-    // Currently, we are not verifying that.

-

-    // Lock the capture buffer

-    LPVOID audioPtr;

-    DWORD dataLen;

-    result = buffer->Lock( 0, dsBufferSize, &audioPtr, &dataLen, NULL, NULL, 0 );

-    if ( FAILED( result ) ) {

-      input->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") locking input buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    // Zero the buffer

-    ZeroMemory( audioPtr, dataLen );

-

-    // Unlock the buffer

-    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );

-    if ( FAILED( result ) ) {

-      input->Release();

-      buffer->Release();

-      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") unlocking input buffer (" << dsDevices[ device ].name << ")!";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-

-    ohandle = (void *) input;

-    bhandle = (void *) buffer;

-  }

-

-  // Set various stream parameters

-  DsHandle *handle = 0;

-  stream_.nDeviceChannels[mode] = channels + firstChannel;

-  stream_.nUserChannels[mode] = channels;

-  stream_.bufferSize = *bufferSize;

-  stream_.channelOffset[mode] = firstChannel;

-  stream_.deviceInterleaved[mode] = true;

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;

-  else stream_.userInterleaved = true;

-

-  // Set flag for buffer conversion

-  stream_.doConvertBuffer[mode] = false;

-  if (stream_.nUserChannels[mode] != stream_.nDeviceChannels[mode])

-    stream_.doConvertBuffer[mode] = true;

-  if (stream_.userFormat != stream_.deviceFormat[mode])

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-       stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate necessary internal buffers

-  long bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiDs::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( mode == INPUT ) {

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-        if ( bufferBytes <= (long) bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiDs::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  // Allocate our DsHandle structures for the stream.

-  if ( stream_.apiHandle == 0 ) {

-    try {

-      handle = new DsHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiDs::probeDeviceOpen: error allocating AsioHandle memory.";

-      goto error;

-    }

-

-    // Create a manual-reset event.

-    handle->condition = CreateEvent( NULL,   // no security

-                                     TRUE,   // manual-reset

-                                     FALSE,  // non-signaled initially

-                                     NULL ); // unnamed

-    stream_.apiHandle = (void *) handle;

-  }

-  else

-    handle = (DsHandle *) stream_.apiHandle;

-  handle->id[mode] = ohandle;

-  handle->buffer[mode] = bhandle;

-  handle->dsBufferSize[mode] = dsBufferSize;

-  handle->dsPointerLeadTime[mode] = dsPointerLeadTime;

-

-  stream_.device[mode] = device;

-  stream_.state = STREAM_STOPPED;

-  if ( stream_.mode == OUTPUT && mode == INPUT )

-    // We had already set up an output stream.

-    stream_.mode = DUPLEX;

-  else

-    stream_.mode = mode;

-  stream_.nBuffers = nBuffers;

-  stream_.sampleRate = sampleRate;

-

-  // Setup the buffer conversion information structure.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );

-

-  // Setup the callback thread.

-  if ( stream_.callbackInfo.isRunning == false ) {

-    unsigned threadId;

-    stream_.callbackInfo.isRunning = true;

-    stream_.callbackInfo.object = (void *) this;

-    stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &callbackHandler,

-                                                  &stream_.callbackInfo, 0, &threadId );

-    if ( stream_.callbackInfo.thread == 0 ) {

-      errorText_ = "RtApiDs::probeDeviceOpen: error creating callback thread!";

-      goto error;

-    }

-

-    // Boost DS thread priority

-    SetThreadPriority( (HANDLE) stream_.callbackInfo.thread, THREAD_PRIORITY_HIGHEST );

-  }

-  return SUCCESS;

-

- error:

-  if ( handle ) {

-    if ( handle->buffer[0] ) { // the object pointer can be NULL and valid

-      LPDIRECTSOUND object = (LPDIRECTSOUND) handle->id[0];

-      LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-      if ( buffer ) buffer->Release();

-      object->Release();

-    }

-    if ( handle->buffer[1] ) {

-      LPDIRECTSOUNDCAPTURE object = (LPDIRECTSOUNDCAPTURE) handle->id[1];

-      LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-      if ( buffer ) buffer->Release();

-      object->Release();

-    }

-    CloseHandle( handle->condition );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.state = STREAM_CLOSED;

-  return FAILURE;

-}

-

-void RtApiDs :: closeStream()

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiDs::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // Stop the callback thread.

-  stream_.callbackInfo.isRunning = false;

-  WaitForSingleObject( (HANDLE) stream_.callbackInfo.thread, INFINITE );

-  CloseHandle( (HANDLE) stream_.callbackInfo.thread );

-

-  DsHandle *handle = (DsHandle *) stream_.apiHandle;

-  if ( handle ) {

-    if ( handle->buffer[0] ) { // the object pointer can be NULL and valid

-      LPDIRECTSOUND object = (LPDIRECTSOUND) handle->id[0];

-      LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-      if ( buffer ) {

-        buffer->Stop();

-        buffer->Release();

-      }

-      object->Release();

-    }

-    if ( handle->buffer[1] ) {

-      LPDIRECTSOUNDCAPTURE object = (LPDIRECTSOUNDCAPTURE) handle->id[1];

-      LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-      if ( buffer ) {

-        buffer->Stop();

-        buffer->Release();

-      }

-      object->Release();

-    }

-    CloseHandle( handle->condition );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-void RtApiDs :: startStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiDs::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  DsHandle *handle = (DsHandle *) stream_.apiHandle;

-

-  // Increase scheduler frequency on lesser windows (a side-effect of

-  // increasing timer accuracy).  On greater windows (Win2K or later),

-  // this is already in effect.

-  timeBeginPeriod( 1 ); 

-

-  buffersRolling = false;

-  duplexPrerollBytes = 0;

-

-  if ( stream_.mode == DUPLEX ) {

-    // 0.5 seconds of silence in DUPLEX mode while the devices spin up and synchronize.

-    duplexPrerollBytes = (int) ( 0.5 * stream_.sampleRate * formatBytes( stream_.deviceFormat[1] ) * stream_.nDeviceChannels[1] );

-  }

-

-  HRESULT result = 0;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-    result = buffer->Play( 0, 0, DSBPLAY_LOOPING );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::startStream: error (" << getErrorString( result ) << ") starting output buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-    result = buffer->Start( DSCBSTART_LOOPING );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::startStream: error (" << getErrorString( result ) << ") starting input buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  handle->drainCounter = 0;

-  handle->internalDrain = false;

-  ResetEvent( handle->condition );

-  stream_.state = STREAM_RUNNING;

-

- unlock:

-  if ( FAILED( result ) ) error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiDs :: stopStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiDs::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  HRESULT result = 0;

-  LPVOID audioPtr;

-  DWORD dataLen;

-  DsHandle *handle = (DsHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    if ( handle->drainCounter == 0 ) {

-      handle->drainCounter = 2;

-      WaitForSingleObject( handle->condition, INFINITE );  // block until signaled

-    }

-

-    stream_.state = STREAM_STOPPED;

-

-    MUTEX_LOCK( &stream_.mutex );

-

-    // Stop the buffer and clear memory

-    LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-    result = buffer->Stop();

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") stopping output buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // Lock the buffer and clear it so that if we start to play again,

-    // we won't have old data playing.

-    result = buffer->Lock( 0, handle->dsBufferSize[0], &audioPtr, &dataLen, NULL, NULL, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") locking output buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // Zero the DS buffer

-    ZeroMemory( audioPtr, dataLen );

-

-    // Unlock the DS buffer

-    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") unlocking output buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // If we start playing again, we must begin at beginning of buffer.

-    handle->bufferPointer[0] = 0;

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-    LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-    audioPtr = NULL;

-    dataLen = 0;

-

-    stream_.state = STREAM_STOPPED;

-

-    if ( stream_.mode != DUPLEX )

-      MUTEX_LOCK( &stream_.mutex );

-

-    result = buffer->Stop();

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") stopping input buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // Lock the buffer and clear it so that if we start to play again,

-    // we won't have old data playing.

-    result = buffer->Lock( 0, handle->dsBufferSize[1], &audioPtr, &dataLen, NULL, NULL, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") locking input buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // Zero the DS buffer

-    ZeroMemory( audioPtr, dataLen );

-

-    // Unlock the DS buffer

-    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") unlocking input buffer!";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-

-    // If we start recording again, we must begin at beginning of buffer.

-    handle->bufferPointer[1] = 0;

-  }

-

- unlock:

-  timeEndPeriod( 1 ); // revert to normal scheduler frequency on lesser windows.

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( FAILED( result ) ) error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiDs :: abortStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiDs::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  DsHandle *handle = (DsHandle *) stream_.apiHandle;

-  handle->drainCounter = 2;

-

-  stopStream();

-}

-

-void RtApiDs :: callbackEvent()

-{

-  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) {

-    Sleep( 50 ); // sleep 50 milliseconds

-    return;

-  }

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiDs::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;

-  DsHandle *handle = (DsHandle *) stream_.apiHandle;

-

-  // Check if we were draining the stream and signal is finished.

-  if ( handle->drainCounter > stream_.nBuffers + 2 ) {

-

-    stream_.state = STREAM_STOPPING;

-    if ( handle->internalDrain == false )

-      SetEvent( handle->condition );

-    else

-      stopStream();

-    return;

-  }

-

-  // Invoke user callback to get fresh output data UNLESS we are

-  // draining stream.

-  if ( handle->drainCounter == 0 ) {

-    RtAudioCallback callback = (RtAudioCallback) info->callback;

-    double streamTime = getStreamTime();

-    RtAudioStreamStatus status = 0;

-    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {

-      status |= RTAUDIO_OUTPUT_UNDERFLOW;

-      handle->xrun[0] = false;

-    }

-    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {

-      status |= RTAUDIO_INPUT_OVERFLOW;

-      handle->xrun[1] = false;

-    }

-    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                                  stream_.bufferSize, streamTime, status, info->userData );

-    if ( cbReturnValue == 2 ) {

-      stream_.state = STREAM_STOPPING;

-      handle->drainCounter = 2;

-      abortStream();

-      return;

-    }

-    else if ( cbReturnValue == 1 ) {

-      handle->drainCounter = 1;

-      handle->internalDrain = true;

-    }

-  }

-

-  HRESULT result;

-  DWORD currentWritePointer, safeWritePointer;

-  DWORD currentReadPointer, safeReadPointer;

-  UINT nextWritePointer;

-

-  LPVOID buffer1 = NULL;

-  LPVOID buffer2 = NULL;

-  DWORD bufferSize1 = 0;

-  DWORD bufferSize2 = 0;

-

-  char *buffer;

-  long bufferBytes;

-

-  MUTEX_LOCK( &stream_.mutex );

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_UNLOCK( &stream_.mutex );

-    return;

-  }

-

-  if ( buffersRolling == false ) {

-    if ( stream_.mode == DUPLEX ) {

-      //assert( handle->dsBufferSize[0] == handle->dsBufferSize[1] );

-

-      // It takes a while for the devices to get rolling. As a result,

-      // there's no guarantee that the capture and write device pointers

-      // will move in lockstep.  Wait here for both devices to start

-      // rolling, and then set our buffer pointers accordingly.

-      // e.g. Crystal Drivers: the capture buffer starts up 5700 to 9600

-      // bytes later than the write buffer.

-

-      // Stub: a serious risk of having a pre-emptive scheduling round

-      // take place between the two GetCurrentPosition calls... but I'm

-      // really not sure how to solve the problem.  Temporarily boost to

-      // Realtime priority, maybe; but I'm not sure what priority the

-      // DirectSound service threads run at. We *should* be roughly

-      // within a ms or so of correct.

-

-      LPDIRECTSOUNDBUFFER dsWriteBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-      LPDIRECTSOUNDCAPTUREBUFFER dsCaptureBuffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-

-      DWORD startSafeWritePointer, startSafeReadPointer;

-

-      result = dsWriteBuffer->GetCurrentPosition( NULL, &startSafeWritePointer );

-      if ( FAILED( result ) ) {

-        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";

-        errorText_ = errorStream_.str();

-        MUTEX_UNLOCK( &stream_.mutex );

-        error( RtAudioError::SYSTEM_ERROR );

-        return;

-      }

-      result = dsCaptureBuffer->GetCurrentPosition( NULL, &startSafeReadPointer );

-      if ( FAILED( result ) ) {

-        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";

-        errorText_ = errorStream_.str();

-        MUTEX_UNLOCK( &stream_.mutex );

-        error( RtAudioError::SYSTEM_ERROR );

-        return;

-      }

-      while ( true ) {

-        result = dsWriteBuffer->GetCurrentPosition( NULL, &safeWritePointer );

-        if ( FAILED( result ) ) {

-          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";

-          errorText_ = errorStream_.str();

-          MUTEX_UNLOCK( &stream_.mutex );

-          error( RtAudioError::SYSTEM_ERROR );

-          return;

-        }

-        result = dsCaptureBuffer->GetCurrentPosition( NULL, &safeReadPointer );

-        if ( FAILED( result ) ) {

-          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";

-          errorText_ = errorStream_.str();

-          MUTEX_UNLOCK( &stream_.mutex );

-          error( RtAudioError::SYSTEM_ERROR );

-          return;

-        }

-        if ( safeWritePointer != startSafeWritePointer && safeReadPointer != startSafeReadPointer ) break;

-        Sleep( 1 );

-      }

-

-      //assert( handle->dsBufferSize[0] == handle->dsBufferSize[1] );

-

-      handle->bufferPointer[0] = safeWritePointer + handle->dsPointerLeadTime[0];

-      if ( handle->bufferPointer[0] >= handle->dsBufferSize[0] ) handle->bufferPointer[0] -= handle->dsBufferSize[0];

-      handle->bufferPointer[1] = safeReadPointer;

-    }

-    else if ( stream_.mode == OUTPUT ) {

-

-      // Set the proper nextWritePosition after initial startup.

-      LPDIRECTSOUNDBUFFER dsWriteBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-      result = dsWriteBuffer->GetCurrentPosition( &currentWritePointer, &safeWritePointer );

-      if ( FAILED( result ) ) {

-        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";

-        errorText_ = errorStream_.str();

-        MUTEX_UNLOCK( &stream_.mutex );

-        error( RtAudioError::SYSTEM_ERROR );

-        return;

-      }

-      handle->bufferPointer[0] = safeWritePointer + handle->dsPointerLeadTime[0];

-      if ( handle->bufferPointer[0] >= handle->dsBufferSize[0] ) handle->bufferPointer[0] -= handle->dsBufferSize[0];

-    }

-

-    buffersRolling = true;

-  }

-

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    

-    LPDIRECTSOUNDBUFFER dsBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];

-

-    if ( handle->drainCounter > 1 ) { // write zeros to the output stream

-      bufferBytes = stream_.bufferSize * stream_.nUserChannels[0];

-      bufferBytes *= formatBytes( stream_.userFormat );

-      memset( stream_.userBuffer[0], 0, bufferBytes );

-    }

-

-    // Setup parameters and do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[0] ) {

-      buffer = stream_.deviceBuffer;

-      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-      bufferBytes = stream_.bufferSize * stream_.nDeviceChannels[0];

-      bufferBytes *= formatBytes( stream_.deviceFormat[0] );

-    }

-    else {

-      buffer = stream_.userBuffer[0];

-      bufferBytes = stream_.bufferSize * stream_.nUserChannels[0];

-      bufferBytes *= formatBytes( stream_.userFormat );

-    }

-

-    // No byte swapping necessary in DirectSound implementation.

-

-    // Ahhh ... windoze.  16-bit data is signed but 8-bit data is

-    // unsigned.  So, we need to convert our signed 8-bit data here to

-    // unsigned.

-    if ( stream_.deviceFormat[0] == RTAUDIO_SINT8 )

-      for ( int i=0; i<bufferBytes; i++ ) buffer[i] = (unsigned char) ( buffer[i] + 128 );

-

-    DWORD dsBufferSize = handle->dsBufferSize[0];

-    nextWritePointer = handle->bufferPointer[0];

-

-    DWORD endWrite, leadPointer;

-    while ( true ) {

-      // Find out where the read and "safe write" pointers are.

-      result = dsBuffer->GetCurrentPosition( &currentWritePointer, &safeWritePointer );

-      if ( FAILED( result ) ) {

-        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";

-        errorText_ = errorStream_.str();

-        error( RtAudioError::SYSTEM_ERROR );

-        return;

-      }

-

-      // We will copy our output buffer into the region between

-      // safeWritePointer and leadPointer.  If leadPointer is not

-      // beyond the next endWrite position, wait until it is.

-      leadPointer = safeWritePointer + handle->dsPointerLeadTime[0];

-      //std::cout << "safeWritePointer = " << safeWritePointer << ", leadPointer = " << leadPointer << ", nextWritePointer = " << nextWritePointer << std::endl;

-      if ( leadPointer > dsBufferSize ) leadPointer -= dsBufferSize;

-      if ( leadPointer < nextWritePointer ) leadPointer += dsBufferSize; // unwrap offset

-      endWrite = nextWritePointer + bufferBytes;

-

-      // Check whether the entire write region is behind the play pointer.

-      if ( leadPointer >= endWrite ) break;

-

-      // If we are here, then we must wait until the leadPointer advances

-      // beyond the end of our next write region. We use the

-      // Sleep() function to suspend operation until that happens.

-      double millis = ( endWrite - leadPointer ) * 1000.0;

-      millis /= ( formatBytes( stream_.deviceFormat[0]) * stream_.nDeviceChannels[0] * stream_.sampleRate);

-      if ( millis < 1.0 ) millis = 1.0;

-      Sleep( (DWORD) millis );

-    }

-

-    if ( dsPointerBetween( nextWritePointer, safeWritePointer, currentWritePointer, dsBufferSize )

-         || dsPointerBetween( endWrite, safeWritePointer, currentWritePointer, dsBufferSize ) ) { 

-      // We've strayed into the forbidden zone ... resync the read pointer.

-      handle->xrun[0] = true;

-      nextWritePointer = safeWritePointer + handle->dsPointerLeadTime[0] - bufferBytes;

-      if ( nextWritePointer >= dsBufferSize ) nextWritePointer -= dsBufferSize;

-      handle->bufferPointer[0] = nextWritePointer;

-      endWrite = nextWritePointer + bufferBytes;

-    }

-

-    // Lock free space in the buffer

-    result = dsBuffer->Lock( nextWritePointer, bufferBytes, &buffer1,

-                             &bufferSize1, &buffer2, &bufferSize2, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") locking buffer during playback!";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-

-    // Copy our buffer into the DS buffer

-    CopyMemory( buffer1, buffer, bufferSize1 );

-    if ( buffer2 != NULL ) CopyMemory( buffer2, buffer+bufferSize1, bufferSize2 );

-

-    // Update our buffer offset and unlock sound buffer

-    dsBuffer->Unlock( buffer1, bufferSize1, buffer2, bufferSize2 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") unlocking buffer during playback!";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-    nextWritePointer = ( nextWritePointer + bufferSize1 + bufferSize2 ) % dsBufferSize;

-    handle->bufferPointer[0] = nextWritePointer;

-  }

-

-  // Don't bother draining input

-  if ( handle->drainCounter ) {

-    handle->drainCounter++;

-    goto unlock;

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    // Setup parameters.

-    if ( stream_.doConvertBuffer[1] ) {

-      buffer = stream_.deviceBuffer;

-      bufferBytes = stream_.bufferSize * stream_.nDeviceChannels[1];

-      bufferBytes *= formatBytes( stream_.deviceFormat[1] );

-    }

-    else {

-      buffer = stream_.userBuffer[1];

-      bufferBytes = stream_.bufferSize * stream_.nUserChannels[1];

-      bufferBytes *= formatBytes( stream_.userFormat );

-    }

-

-    LPDIRECTSOUNDCAPTUREBUFFER dsBuffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];

-    long nextReadPointer = handle->bufferPointer[1];

-    DWORD dsBufferSize = handle->dsBufferSize[1];

-

-    // Find out where the write and "safe read" pointers are.

-    result = dsBuffer->GetCurrentPosition( &currentReadPointer, &safeReadPointer );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-

-    if ( safeReadPointer < (DWORD)nextReadPointer ) safeReadPointer += dsBufferSize; // unwrap offset

-    DWORD endRead = nextReadPointer + bufferBytes;

-

-    // Handling depends on whether we are INPUT or DUPLEX. 

-    // If we're in INPUT mode then waiting is a good thing. If we're in DUPLEX mode,

-    // then a wait here will drag the write pointers into the forbidden zone.

-    // 

-    // In DUPLEX mode, rather than wait, we will back off the read pointer until 

-    // it's in a safe position. This causes dropouts, but it seems to be the only 

-    // practical way to sync up the read and write pointers reliably, given the 

-    // the very complex relationship between phase and increment of the read and write 

-    // pointers.

-    //

-    // In order to minimize audible dropouts in DUPLEX mode, we will

-    // provide a pre-roll period of 0.5 seconds in which we return

-    // zeros from the read buffer while the pointers sync up.

-

-    if ( stream_.mode == DUPLEX ) {

-      if ( safeReadPointer < endRead ) {

-        if ( duplexPrerollBytes <= 0 ) {

-          // Pre-roll time over. Be more agressive.

-          int adjustment = endRead-safeReadPointer;

-

-          handle->xrun[1] = true;

-          // Two cases:

-          //   - large adjustments: we've probably run out of CPU cycles, so just resync exactly,

-          //     and perform fine adjustments later.

-          //   - small adjustments: back off by twice as much.

-          if ( adjustment >= 2*bufferBytes )

-            nextReadPointer = safeReadPointer-2*bufferBytes;

-          else

-            nextReadPointer = safeReadPointer-bufferBytes-adjustment;

-

-          if ( nextReadPointer < 0 ) nextReadPointer += dsBufferSize;

-

-        }

-        else {

-          // In pre=roll time. Just do it.

-          nextReadPointer = safeReadPointer - bufferBytes;

-          while ( nextReadPointer < 0 ) nextReadPointer += dsBufferSize;

-        }

-        endRead = nextReadPointer + bufferBytes;

-      }

-    }

-    else { // mode == INPUT

-      while ( safeReadPointer < endRead && stream_.callbackInfo.isRunning ) {

-        // See comments for playback.

-        double millis = (endRead - safeReadPointer) * 1000.0;

-        millis /= ( formatBytes(stream_.deviceFormat[1]) * stream_.nDeviceChannels[1] * stream_.sampleRate);

-        if ( millis < 1.0 ) millis = 1.0;

-        Sleep( (DWORD) millis );

-

-        // Wake up and find out where we are now.

-        result = dsBuffer->GetCurrentPosition( &currentReadPointer, &safeReadPointer );

-        if ( FAILED( result ) ) {

-          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";

-          errorText_ = errorStream_.str();

-          MUTEX_UNLOCK( &stream_.mutex );

-          error( RtAudioError::SYSTEM_ERROR );

-          return;

-        }

-      

-        if ( safeReadPointer < (DWORD)nextReadPointer ) safeReadPointer += dsBufferSize; // unwrap offset

-      }

-    }

-

-    // Lock free space in the buffer

-    result = dsBuffer->Lock( nextReadPointer, bufferBytes, &buffer1,

-                             &bufferSize1, &buffer2, &bufferSize2, 0 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") locking capture buffer!";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-

-    if ( duplexPrerollBytes <= 0 ) {

-      // Copy our buffer into the DS buffer

-      CopyMemory( buffer, buffer1, bufferSize1 );

-      if ( buffer2 != NULL ) CopyMemory( buffer+bufferSize1, buffer2, bufferSize2 );

-    }

-    else {

-      memset( buffer, 0, bufferSize1 );

-      if ( buffer2 != NULL ) memset( buffer + bufferSize1, 0, bufferSize2 );

-      duplexPrerollBytes -= bufferSize1 + bufferSize2;

-    }

-

-    // Update our buffer offset and unlock sound buffer

-    nextReadPointer = ( nextReadPointer + bufferSize1 + bufferSize2 ) % dsBufferSize;

-    dsBuffer->Unlock( buffer1, bufferSize1, buffer2, bufferSize2 );

-    if ( FAILED( result ) ) {

-      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") unlocking capture buffer!";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-    handle->bufferPointer[1] = nextReadPointer;

-

-    // No byte swapping necessary in DirectSound implementation.

-

-    // If necessary, convert 8-bit data from unsigned to signed.

-    if ( stream_.deviceFormat[1] == RTAUDIO_SINT8 )

-      for ( int j=0; j<bufferBytes; j++ ) buffer[j] = (signed char) ( buffer[j] - 128 );

-

-    // Do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[1] )

-      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );

-  }

-

- unlock:

-  MUTEX_UNLOCK( &stream_.mutex );

-  RtApi::tickStreamTime();

-}

-

-// Definitions for utility functions and callbacks

-// specific to the DirectSound implementation.

-

-static unsigned __stdcall callbackHandler( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiDs *object = (RtApiDs *) info->object;

-  bool* isRunning = &info->isRunning;

-

-  while ( *isRunning == true ) {

-    object->callbackEvent();

-  }

-

-  _endthreadex( 0 );

-  return 0;

-}

-

-static BOOL CALLBACK deviceQueryCallback( LPGUID lpguid,

-                                          LPCTSTR description,

-                                          LPCTSTR /*module*/,

-                                          LPVOID lpContext )

-{

-  struct DsProbeData& probeInfo = *(struct DsProbeData*) lpContext;

-  std::vector<struct DsDevice>& dsDevices = *probeInfo.dsDevices;

-

-  HRESULT hr;

-  bool validDevice = false;

-  if ( probeInfo.isInput == true ) {

-    DSCCAPS caps;

-    LPDIRECTSOUNDCAPTURE object;

-

-    hr = DirectSoundCaptureCreate(  lpguid, &object,   NULL );

-    if ( hr != DS_OK ) return TRUE;

-

-    caps.dwSize = sizeof(caps);

-    hr = object->GetCaps( &caps );

-    if ( hr == DS_OK ) {

-      if ( caps.dwChannels > 0 && caps.dwFormats > 0 )

-        validDevice = true;

-    }

-    object->Release();

-  }

-  else {

-    DSCAPS caps;

-    LPDIRECTSOUND object;

-    hr = DirectSoundCreate(  lpguid, &object,   NULL );

-    if ( hr != DS_OK ) return TRUE;

-

-    caps.dwSize = sizeof(caps);

-    hr = object->GetCaps( &caps );

-    if ( hr == DS_OK ) {

-      if ( caps.dwFlags & DSCAPS_PRIMARYMONO || caps.dwFlags & DSCAPS_PRIMARYSTEREO )

-        validDevice = true;

-    }

-    object->Release();

-  }

-

-  // If good device, then save its name and guid.

-  std::string name = convertCharPointerToStdString( description );

-  //if ( name == "Primary Sound Driver" || name == "Primary Sound Capture Driver" )

-  if ( lpguid == NULL )

-    name = "Default Device";

-  if ( validDevice ) {

-    for ( unsigned int i=0; i<dsDevices.size(); i++ ) {

-      if ( dsDevices[i].name == name ) {

-        dsDevices[i].found = true;

-        if ( probeInfo.isInput ) {

-          dsDevices[i].id[1] = lpguid;

-          dsDevices[i].validId[1] = true;

-        }

-        else {

-          dsDevices[i].id[0] = lpguid;

-          dsDevices[i].validId[0] = true;

-        }

-        return TRUE;

-      }

-    }

-

-    DsDevice device;

-    device.name = name;

-    device.found = true;

-    if ( probeInfo.isInput ) {

-      device.id[1] = lpguid;

-      device.validId[1] = true;

-    }

-    else {

-      device.id[0] = lpguid;

-      device.validId[0] = true;

-    }

-    dsDevices.push_back( device );

-  }

-

-  return TRUE;

-}

-

-static const char* getErrorString( int code )

-{

-  switch ( code ) {

-

-  case DSERR_ALLOCATED:

-    return "Already allocated";

-

-  case DSERR_CONTROLUNAVAIL:

-    return "Control unavailable";

-

-  case DSERR_INVALIDPARAM:

-    return "Invalid parameter";

-

-  case DSERR_INVALIDCALL:

-    return "Invalid call";

-

-  case DSERR_GENERIC:

-    return "Generic error";

-

-  case DSERR_PRIOLEVELNEEDED:

-    return "Priority level needed";

-

-  case DSERR_OUTOFMEMORY:

-    return "Out of memory";

-

-  case DSERR_BADFORMAT:

-    return "The sample rate or the channel format is not supported";

-

-  case DSERR_UNSUPPORTED:

-    return "Not supported";

-

-  case DSERR_NODRIVER:

-    return "No driver";

-

-  case DSERR_ALREADYINITIALIZED:

-    return "Already initialized";

-

-  case DSERR_NOAGGREGATION:

-    return "No aggregation";

-

-  case DSERR_BUFFERLOST:

-    return "Buffer lost";

-

-  case DSERR_OTHERAPPHASPRIO:

-    return "Another application already has priority";

-

-  case DSERR_UNINITIALIZED:

-    return "Uninitialized";

-

-  default:

-    return "DirectSound unknown error";

-  }

-}

-//******************** End of __WINDOWS_DS__ *********************//

-#endif

-

-

-#if defined(__LINUX_ALSA__)

-

-#include <alsa/asoundlib.h>

-#include <unistd.h>

-

-  // A structure to hold various information related to the ALSA API

-  // implementation.

-struct AlsaHandle {

-  snd_pcm_t *handles[2];

-  bool synchronized;

-  bool xrun[2];

-  pthread_cond_t runnable_cv;

-  bool runnable;

-

-  AlsaHandle()

-    :synchronized(false), runnable(false) { xrun[0] = false; xrun[1] = false; }

-};

-

-static void *alsaCallbackHandler( void * ptr );

-

-RtApiAlsa :: RtApiAlsa()

-{

-  // Nothing to do here.

-}

-

-RtApiAlsa :: ~RtApiAlsa()

-{

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-}

-

-unsigned int RtApiAlsa :: getDeviceCount( void )

-{

-  unsigned nDevices = 0;

-  int result, subdevice, card;

-  char name[64];

-  snd_ctl_t *handle;

-

-  // Count cards and devices

-  card = -1;

-  snd_card_next( &card );

-  while ( card >= 0 ) {

-    sprintf( name, "hw:%d", card );

-    result = snd_ctl_open( &handle, name, 0 );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::getDeviceCount: control open, card = " << card << ", " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      error( RtAudioError::WARNING );

-      goto nextcard;

-    }

-    subdevice = -1;

-    while( 1 ) {

-      result = snd_ctl_pcm_next_device( handle, &subdevice );

-      if ( result < 0 ) {

-        errorStream_ << "RtApiAlsa::getDeviceCount: control next device, card = " << card << ", " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-        error( RtAudioError::WARNING );

-        break;

-      }

-      if ( subdevice < 0 )

-        break;

-      nDevices++;

-    }

-  nextcard:

-    snd_ctl_close( handle );

-    snd_card_next( &card );

-  }

-

-  result = snd_ctl_open( &handle, "default", 0 );

-  if (result == 0) {

-    nDevices++;

-    snd_ctl_close( handle );

-  }

-

-  return nDevices;

-}

-

-RtAudio::DeviceInfo RtApiAlsa :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  unsigned nDevices = 0;

-  int result, subdevice, card;

-  char name[64];

-  snd_ctl_t *chandle;

-

-  // Count cards and devices

-  card = -1;

-  subdevice = -1;

-  snd_card_next( &card );

-  while ( card >= 0 ) {

-    sprintf( name, "hw:%d", card );

-    result = snd_ctl_open( &chandle, name, SND_CTL_NONBLOCK );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::getDeviceInfo: control open, card = " << card << ", " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      error( RtAudioError::WARNING );

-      goto nextcard;

-    }

-    subdevice = -1;

-    while( 1 ) {

-      result = snd_ctl_pcm_next_device( chandle, &subdevice );

-      if ( result < 0 ) {

-        errorStream_ << "RtApiAlsa::getDeviceInfo: control next device, card = " << card << ", " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-        error( RtAudioError::WARNING );

-        break;

-      }

-      if ( subdevice < 0 ) break;

-      if ( nDevices == device ) {

-        sprintf( name, "hw:%d,%d", card, subdevice );

-        goto foundDevice;

-      }

-      nDevices++;

-    }

-  nextcard:

-    snd_ctl_close( chandle );

-    snd_card_next( &card );

-  }

-

-  result = snd_ctl_open( &chandle, "default", SND_CTL_NONBLOCK );

-  if ( result == 0 ) {

-    if ( nDevices == device ) {

-      strcpy( name, "default" );

-      goto foundDevice;

-    }

-    nDevices++;

-  }

-

-  if ( nDevices == 0 ) {

-    errorText_ = "RtApiAlsa::getDeviceInfo: no devices found!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  if ( device >= nDevices ) {

-    errorText_ = "RtApiAlsa::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

- foundDevice:

-

-  // If a stream is already open, we cannot probe the stream devices.

-  // Thus, use the saved results.

-  if ( stream_.state != STREAM_CLOSED &&

-       ( stream_.device[0] == device || stream_.device[1] == device ) ) {

-    snd_ctl_close( chandle );

-    if ( device >= devices_.size() ) {

-      errorText_ = "RtApiAlsa::getDeviceInfo: device ID was not present before stream was opened.";

-      error( RtAudioError::WARNING );

-      return info;

-    }

-    return devices_[ device ];

-  }

-

-  int openMode = SND_PCM_ASYNC;

-  snd_pcm_stream_t stream;

-  snd_pcm_info_t *pcminfo;

-  snd_pcm_info_alloca( &pcminfo );

-  snd_pcm_t *phandle;

-  snd_pcm_hw_params_t *params;

-  snd_pcm_hw_params_alloca( &params );

-

-  // First try for playback unless default device (which has subdev -1)

-  stream = SND_PCM_STREAM_PLAYBACK;

-  snd_pcm_info_set_stream( pcminfo, stream );

-  if ( subdevice != -1 ) {

-    snd_pcm_info_set_device( pcminfo, subdevice );

-    snd_pcm_info_set_subdevice( pcminfo, 0 );

-

-    result = snd_ctl_pcm_info( chandle, pcminfo );

-    if ( result < 0 ) {

-      // Device probably doesn't support playback.

-      goto captureProbe;

-    }

-  }

-

-  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK );

-  if ( result < 0 ) {

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    goto captureProbe;

-  }

-

-  // The device is open ... fill the parameter structure.

-  result = snd_pcm_hw_params_any( phandle, params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    goto captureProbe;

-  }

-

-  // Get output channel information.

-  unsigned int value;

-  result = snd_pcm_hw_params_get_channels_max( params, &value );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: error getting device (" << name << ") output channels, " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    goto captureProbe;

-  }

-  info.outputChannels = value;

-  snd_pcm_close( phandle );

-

- captureProbe:

-  stream = SND_PCM_STREAM_CAPTURE;

-  snd_pcm_info_set_stream( pcminfo, stream );

-

-  // Now try for capture unless default device (with subdev = -1)

-  if ( subdevice != -1 ) {

-    result = snd_ctl_pcm_info( chandle, pcminfo );

-    snd_ctl_close( chandle );

-    if ( result < 0 ) {

-      // Device probably doesn't support capture.

-      if ( info.outputChannels == 0 ) return info;

-      goto probeParameters;

-    }

-  }

-  else

-    snd_ctl_close( chandle );

-

-  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK);

-  if ( result < 0 ) {

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    if ( info.outputChannels == 0 ) return info;

-    goto probeParameters;

-  }

-

-  // The device is open ... fill the parameter structure.

-  result = snd_pcm_hw_params_any( phandle, params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    if ( info.outputChannels == 0 ) return info;

-    goto probeParameters;

-  }

-

-  result = snd_pcm_hw_params_get_channels_max( params, &value );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: error getting device (" << name << ") input channels, " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    if ( info.outputChannels == 0 ) return info;

-    goto probeParameters;

-  }

-  info.inputChannels = value;

-  snd_pcm_close( phandle );

-

-  // If device opens for both playback and capture, we determine the channels.

-  if ( info.outputChannels > 0 && info.inputChannels > 0 )

-    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-

-  // ALSA doesn't provide default devices so we'll use the first available one.

-  if ( device == 0 && info.outputChannels > 0 )

-    info.isDefaultOutput = true;

-  if ( device == 0 && info.inputChannels > 0 )

-    info.isDefaultInput = true;

-

- probeParameters:

-  // At this point, we just need to figure out the supported data

-  // formats and sample rates.  We'll proceed by opening the device in

-  // the direction with the maximum number of channels, or playback if

-  // they are equal.  This might limit our sample rate options, but so

-  // be it.

-

-  if ( info.outputChannels >= info.inputChannels )

-    stream = SND_PCM_STREAM_PLAYBACK;

-  else

-    stream = SND_PCM_STREAM_CAPTURE;

-  snd_pcm_info_set_stream( pcminfo, stream );

-

-  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK);

-  if ( result < 0 ) {

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // The device is open ... fill the parameter structure.

-  result = snd_pcm_hw_params_any( phandle, params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Test our discrete set of sample rate values.

-  info.sampleRates.clear();

-  for ( unsigned int i=0; i<MAX_SAMPLE_RATES; i++ ) {

-    if ( snd_pcm_hw_params_test_rate( phandle, params, SAMPLE_RATES[i], 0 ) == 0 ) {

-      info.sampleRates.push_back( SAMPLE_RATES[i] );

-

-      if ( !info.preferredSampleRate || ( SAMPLE_RATES[i] <= 48000 && SAMPLE_RATES[i] > info.preferredSampleRate ) )

-        info.preferredSampleRate = SAMPLE_RATES[i];

-    }

-  }

-  if ( info.sampleRates.size() == 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: no supported sample rates found for device (" << name << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Probe the supported data formats ... we don't care about endian-ness just yet

-  snd_pcm_format_t format;

-  info.nativeFormats = 0;

-  format = SND_PCM_FORMAT_S8;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_SINT8;

-  format = SND_PCM_FORMAT_S16;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_SINT16;

-  format = SND_PCM_FORMAT_S24;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_SINT24;

-  format = SND_PCM_FORMAT_S32;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_SINT32;

-  format = SND_PCM_FORMAT_FLOAT;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_FLOAT32;

-  format = SND_PCM_FORMAT_FLOAT64;

-  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )

-    info.nativeFormats |= RTAUDIO_FLOAT64;

-

-  // Check that we have at least one supported format

-  if ( info.nativeFormats == 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::getDeviceInfo: pcm device (" << name << ") data format not supported by RtAudio.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Get the device name

-  char *cardname;

-  result = snd_card_get_name( card, &cardname );

-  if ( result >= 0 ) {

-    sprintf( name, "hw:%s,%d", cardname, subdevice );

-    free( cardname );

-  }

-  info.name = name;

-

-  // That's all ... close the device and return

-  snd_pcm_close( phandle );

-  info.probed = true;

-  return info;

-}

-

-void RtApiAlsa :: saveDeviceInfo( void )

-{

-  devices_.clear();

-

-  unsigned int nDevices = getDeviceCount();

-  devices_.resize( nDevices );

-  for ( unsigned int i=0; i<nDevices; i++ )

-    devices_[i] = getDeviceInfo( i );

-}

-

-bool RtApiAlsa :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                   unsigned int firstChannel, unsigned int sampleRate,

-                                   RtAudioFormat format, unsigned int *bufferSize,

-                                   RtAudio::StreamOptions *options )

-

-{

-#if defined(__RTAUDIO_DEBUG__)

-  snd_output_t *out;

-  snd_output_stdio_attach(&out, stderr, 0);

-#endif

-

-  // I'm not using the "plug" interface ... too much inconsistent behavior.

-

-  unsigned nDevices = 0;

-  int result, subdevice, card;

-  char name[64];

-  snd_ctl_t *chandle;

-

-  if ( options && options->flags & RTAUDIO_ALSA_USE_DEFAULT )

-    snprintf(name, sizeof(name), "%s", "default");

-  else {

-    // Count cards and devices

-    card = -1;

-    snd_card_next( &card );

-    while ( card >= 0 ) {

-      sprintf( name, "hw:%d", card );

-      result = snd_ctl_open( &chandle, name, SND_CTL_NONBLOCK );

-      if ( result < 0 ) {

-        errorStream_ << "RtApiAlsa::probeDeviceOpen: control open, card = " << card << ", " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-        return FAILURE;

-      }

-      subdevice = -1;

-      while( 1 ) {

-        result = snd_ctl_pcm_next_device( chandle, &subdevice );

-        if ( result < 0 ) break;

-        if ( subdevice < 0 ) break;

-        if ( nDevices == device ) {

-          sprintf( name, "hw:%d,%d", card, subdevice );

-          snd_ctl_close( chandle );

-          goto foundDevice;

-        }

-        nDevices++;

-      }

-      snd_ctl_close( chandle );

-      snd_card_next( &card );

-    }

-

-    result = snd_ctl_open( &chandle, "default", SND_CTL_NONBLOCK );

-    if ( result == 0 ) {

-      if ( nDevices == device ) {

-        strcpy( name, "default" );

-        goto foundDevice;

-      }

-      nDevices++;

-    }

-

-    if ( nDevices == 0 ) {

-      // This should not happen because a check is made before this function is called.

-      errorText_ = "RtApiAlsa::probeDeviceOpen: no devices found!";

-      return FAILURE;

-    }

-

-    if ( device >= nDevices ) {

-      // This should not happen because a check is made before this function is called.

-      errorText_ = "RtApiAlsa::probeDeviceOpen: device ID is invalid!";

-      return FAILURE;

-    }

-  }

-

- foundDevice:

-

-  // The getDeviceInfo() function will not work for a device that is

-  // already open.  Thus, we'll probe the system before opening a

-  // stream and save the results for use by getDeviceInfo().

-  if ( mode == OUTPUT || ( mode == INPUT && stream_.mode != OUTPUT ) ) // only do once

-    this->saveDeviceInfo();

-

-  snd_pcm_stream_t stream;

-  if ( mode == OUTPUT )

-    stream = SND_PCM_STREAM_PLAYBACK;

-  else

-    stream = SND_PCM_STREAM_CAPTURE;

-

-  snd_pcm_t *phandle;

-  int openMode = SND_PCM_ASYNC;

-  result = snd_pcm_open( &phandle, name, stream, openMode );

-  if ( result < 0 ) {

-    if ( mode == OUTPUT )

-      errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device (" << name << ") won't open for output.";

-    else

-      errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device (" << name << ") won't open for input.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Fill the parameter structure.

-  snd_pcm_hw_params_t *hw_params;

-  snd_pcm_hw_params_alloca( &hw_params );

-  result = snd_pcm_hw_params_any( phandle, hw_params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting pcm device (" << name << ") parameters, " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-#if defined(__RTAUDIO_DEBUG__)

-  fprintf( stderr, "\nRtApiAlsa: dump hardware params just after device open:\n\n" );

-  snd_pcm_hw_params_dump( hw_params, out );

-#endif

-

-  // Set access ... check user preference.

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) {

-    stream_.userInterleaved = false;

-    result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED );

-    if ( result < 0 ) {

-      result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_INTERLEAVED );

-      stream_.deviceInterleaved[mode] =  true;

-    }

-    else

-      stream_.deviceInterleaved[mode] = false;

-  }

-  else {

-    stream_.userInterleaved = true;

-    result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_INTERLEAVED );

-    if ( result < 0 ) {

-      result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED );

-      stream_.deviceInterleaved[mode] =  false;

-    }

-    else

-      stream_.deviceInterleaved[mode] =  true;

-  }

-

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting pcm device (" << name << ") access, " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Determine how to set the device format.

-  stream_.userFormat = format;

-  snd_pcm_format_t deviceFormat = SND_PCM_FORMAT_UNKNOWN;

-

-  if ( format == RTAUDIO_SINT8 )

-    deviceFormat = SND_PCM_FORMAT_S8;

-  else if ( format == RTAUDIO_SINT16 )

-    deviceFormat = SND_PCM_FORMAT_S16;

-  else if ( format == RTAUDIO_SINT24 )

-    deviceFormat = SND_PCM_FORMAT_S24;

-  else if ( format == RTAUDIO_SINT32 )

-    deviceFormat = SND_PCM_FORMAT_S32;

-  else if ( format == RTAUDIO_FLOAT32 )

-    deviceFormat = SND_PCM_FORMAT_FLOAT;

-  else if ( format == RTAUDIO_FLOAT64 )

-    deviceFormat = SND_PCM_FORMAT_FLOAT64;

-

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat) == 0) {

-    stream_.deviceFormat[mode] = format;

-    goto setFormat;

-  }

-

-  // The user requested format is not natively supported by the device.

-  deviceFormat = SND_PCM_FORMAT_FLOAT64;

-  if ( snd_pcm_hw_params_test_format( phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_FLOAT64;

-    goto setFormat;

-  }

-

-  deviceFormat = SND_PCM_FORMAT_FLOAT;

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;

-    goto setFormat;

-  }

-

-  deviceFormat = SND_PCM_FORMAT_S32;

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-    goto setFormat;

-  }

-

-  deviceFormat = SND_PCM_FORMAT_S24;

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-    goto setFormat;

-  }

-

-  deviceFormat = SND_PCM_FORMAT_S16;

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-    goto setFormat;

-  }

-

-  deviceFormat = SND_PCM_FORMAT_S8;

-  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {

-    stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-    goto setFormat;

-  }

-

-  // If we get here, no supported format was found.

-  snd_pcm_close( phandle );

-  errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device " << device << " data format not supported by RtAudio.";

-  errorText_ = errorStream_.str();

-  return FAILURE;

-

- setFormat:

-  result = snd_pcm_hw_params_set_format( phandle, hw_params, deviceFormat );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting pcm device (" << name << ") data format, " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Determine whether byte-swaping is necessary.

-  stream_.doByteSwap[mode] = false;

-  if ( deviceFormat != SND_PCM_FORMAT_S8 ) {

-    result = snd_pcm_format_cpu_endian( deviceFormat );

-    if ( result == 0 )

-      stream_.doByteSwap[mode] = true;

-    else if (result < 0) {

-      snd_pcm_close( phandle );

-      errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting pcm device (" << name << ") endian-ness, " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      return FAILURE;

-    }

-  }

-

-  // Set the sample rate.

-  result = snd_pcm_hw_params_set_rate_near( phandle, hw_params, (unsigned int*) &sampleRate, 0 );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting sample rate on device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Determine the number of channels for this device.  We support a possible

-  // minimum device channel number > than the value requested by the user.

-  stream_.nUserChannels[mode] = channels;

-  unsigned int value;

-  result = snd_pcm_hw_params_get_channels_max( hw_params, &value );

-  unsigned int deviceChannels = value;

-  if ( result < 0 || deviceChannels < channels + firstChannel ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: requested channel parameters not supported by device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  result = snd_pcm_hw_params_get_channels_min( hw_params, &value );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting minimum channels for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  deviceChannels = value;

-  if ( deviceChannels < channels + firstChannel ) deviceChannels = channels + firstChannel;

-  stream_.nDeviceChannels[mode] = deviceChannels;

-

-  // Set the device channels.

-  result = snd_pcm_hw_params_set_channels( phandle, hw_params, deviceChannels );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting channels for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Set the buffer (or period) size.

-  int dir = 0;

-  snd_pcm_uframes_t periodSize = *bufferSize;

-  result = snd_pcm_hw_params_set_period_size_near( phandle, hw_params, &periodSize, &dir );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting period size for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  *bufferSize = periodSize;

-

-  // Set the buffer number, which in ALSA is referred to as the "period".

-  unsigned int periods = 0;

-  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) periods = 2;

-  if ( options && options->numberOfBuffers > 0 ) periods = options->numberOfBuffers;

-  if ( periods < 2 ) periods = 4; // a fairly safe default value

-  result = snd_pcm_hw_params_set_periods_near( phandle, hw_params, &periods, &dir );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting periods for device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // If attempting to setup a duplex stream, the bufferSize parameter

-  // MUST be the same in both directions!

-  if ( stream_.mode == OUTPUT && mode == INPUT && *bufferSize != stream_.bufferSize ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: system error setting buffer size for duplex stream on device (" << name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  stream_.bufferSize = *bufferSize;

-

-  // Install the hardware configuration

-  result = snd_pcm_hw_params( phandle, hw_params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error installing hardware configuration on device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-#if defined(__RTAUDIO_DEBUG__)

-  fprintf(stderr, "\nRtApiAlsa: dump hardware params after installation:\n\n");

-  snd_pcm_hw_params_dump( hw_params, out );

-#endif

-

-  // Set the software configuration to fill buffers with zeros and prevent device stopping on xruns.

-  snd_pcm_sw_params_t *sw_params = NULL;

-  snd_pcm_sw_params_alloca( &sw_params );

-  snd_pcm_sw_params_current( phandle, sw_params );

-  snd_pcm_sw_params_set_start_threshold( phandle, sw_params, *bufferSize );

-  snd_pcm_sw_params_set_stop_threshold( phandle, sw_params, ULONG_MAX );

-  snd_pcm_sw_params_set_silence_threshold( phandle, sw_params, 0 );

-

-  // The following two settings were suggested by Theo Veenker

-  //snd_pcm_sw_params_set_avail_min( phandle, sw_params, *bufferSize );

-  //snd_pcm_sw_params_set_xfer_align( phandle, sw_params, 1 );

-

-  // here are two options for a fix

-  //snd_pcm_sw_params_set_silence_size( phandle, sw_params, ULONG_MAX );

-  snd_pcm_uframes_t val;

-  snd_pcm_sw_params_get_boundary( sw_params, &val );

-  snd_pcm_sw_params_set_silence_size( phandle, sw_params, val );

-

-  result = snd_pcm_sw_params( phandle, sw_params );

-  if ( result < 0 ) {

-    snd_pcm_close( phandle );

-    errorStream_ << "RtApiAlsa::probeDeviceOpen: error installing software configuration on device (" << name << "), " << snd_strerror( result ) << ".";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-#if defined(__RTAUDIO_DEBUG__)

-  fprintf(stderr, "\nRtApiAlsa: dump software params after installation:\n\n");

-  snd_pcm_sw_params_dump( sw_params, out );

-#endif

-

-  // Set flags for buffer conversion

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-       stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate the ApiHandle if necessary and then save.

-  AlsaHandle *apiInfo = 0;

-  if ( stream_.apiHandle == 0 ) {

-    try {

-      apiInfo = (AlsaHandle *) new AlsaHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating AlsaHandle memory.";

-      goto error;

-    }

-

-    if ( pthread_cond_init( &apiInfo->runnable_cv, NULL ) ) {

-      errorText_ = "RtApiAlsa::probeDeviceOpen: error initializing pthread condition variable.";

-      goto error;

-    }

-

-    stream_.apiHandle = (void *) apiInfo;

-    apiInfo->handles[0] = 0;

-    apiInfo->handles[1] = 0;

-  }

-  else {

-    apiInfo = (AlsaHandle *) stream_.apiHandle;

-  }

-  apiInfo->handles[mode] = phandle;

-  phandle = 0;

-

-  // Allocate necessary internal buffers.

-  unsigned long bufferBytes;

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( mode == INPUT ) {

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-        if ( bufferBytes <= bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  stream_.sampleRate = sampleRate;

-  stream_.nBuffers = periods;

-  stream_.device[mode] = device;

-  stream_.state = STREAM_STOPPED;

-

-  // Setup the buffer conversion information structure.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );

-

-  // Setup thread if necessary.

-  if ( stream_.mode == OUTPUT && mode == INPUT ) {

-    // We had already set up an output stream.

-    stream_.mode = DUPLEX;

-    // Link the streams if possible.

-    apiInfo->synchronized = false;

-    if ( snd_pcm_link( apiInfo->handles[0], apiInfo->handles[1] ) == 0 )

-      apiInfo->synchronized = true;

-    else {

-      errorText_ = "RtApiAlsa::probeDeviceOpen: unable to synchronize input and output devices.";

-      error( RtAudioError::WARNING );

-    }

-  }

-  else {

-    stream_.mode = mode;

-

-    // Setup callback thread.

-    stream_.callbackInfo.object = (void *) this;

-

-    // Set the thread attributes for joinable and realtime scheduling

-    // priority (optional).  The higher priority will only take affect

-    // if the program is run as root or suid. Note, under Linux

-    // processes with CAP_SYS_NICE privilege, a user can change

-    // scheduling policy and priority (thus need not be root). See

-    // POSIX "capabilities".

-    pthread_attr_t attr;

-    pthread_attr_init( &attr );

-    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE );

-

-#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)

-    if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME ) {

-      // We previously attempted to increase the audio callback priority

-      // to SCHED_RR here via the attributes.  However, while no errors

-      // were reported in doing so, it did not work.  So, now this is

-      // done in the alsaCallbackHandler function.

-      stream_.callbackInfo.doRealtime = true;

-      int priority = options->priority;

-      int min = sched_get_priority_min( SCHED_RR );

-      int max = sched_get_priority_max( SCHED_RR );

-      if ( priority < min ) priority = min;

-      else if ( priority > max ) priority = max;

-      stream_.callbackInfo.priority = priority;

-    }

-#endif

-

-    stream_.callbackInfo.isRunning = true;

-    result = pthread_create( &stream_.callbackInfo.thread, &attr, alsaCallbackHandler, &stream_.callbackInfo );

-    pthread_attr_destroy( &attr );

-    if ( result ) {

-      stream_.callbackInfo.isRunning = false;

-      errorText_ = "RtApiAlsa::error creating callback thread!";

-      goto error;

-    }

-  }

-

-  return SUCCESS;

-

- error:

-  if ( apiInfo ) {

-    pthread_cond_destroy( &apiInfo->runnable_cv );

-    if ( apiInfo->handles[0] ) snd_pcm_close( apiInfo->handles[0] );

-    if ( apiInfo->handles[1] ) snd_pcm_close( apiInfo->handles[1] );

-    delete apiInfo;

-    stream_.apiHandle = 0;

-  }

-

-  if ( phandle) snd_pcm_close( phandle );

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.state = STREAM_CLOSED;

-  return FAILURE;

-}

-

-void RtApiAlsa :: closeStream()

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiAlsa::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;

-  stream_.callbackInfo.isRunning = false;

-  MUTEX_LOCK( &stream_.mutex );

-  if ( stream_.state == STREAM_STOPPED ) {

-    apiInfo->runnable = true;

-    pthread_cond_signal( &apiInfo->runnable_cv );

-  }

-  MUTEX_UNLOCK( &stream_.mutex );

-  pthread_join( stream_.callbackInfo.thread, NULL );

-

-  if ( stream_.state == STREAM_RUNNING ) {

-    stream_.state = STREAM_STOPPED;

-    if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )

-      snd_pcm_drop( apiInfo->handles[0] );

-    if ( stream_.mode == INPUT || stream_.mode == DUPLEX )

-      snd_pcm_drop( apiInfo->handles[1] );

-  }

-

-  if ( apiInfo ) {

-    pthread_cond_destroy( &apiInfo->runnable_cv );

-    if ( apiInfo->handles[0] ) snd_pcm_close( apiInfo->handles[0] );

-    if ( apiInfo->handles[1] ) snd_pcm_close( apiInfo->handles[1] );

-    delete apiInfo;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-void RtApiAlsa :: startStream()

-{

-  // This method calls snd_pcm_prepare if the device isn't already in that state.

-

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiAlsa::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  int result = 0;

-  snd_pcm_state_t state;

-  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;

-  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    state = snd_pcm_state( handle[0] );

-    if ( state != SND_PCM_STATE_PREPARED ) {

-      result = snd_pcm_prepare( handle[0] );

-      if ( result < 0 ) {

-        errorStream_ << "RtApiAlsa::startStream: error preparing output pcm device, " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-        goto unlock;

-      }

-    }

-  }

-

-  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {

-    result = snd_pcm_drop(handle[1]); // fix to remove stale data received since device has been open

-    state = snd_pcm_state( handle[1] );

-    if ( state != SND_PCM_STATE_PREPARED ) {

-      result = snd_pcm_prepare( handle[1] );

-      if ( result < 0 ) {

-        errorStream_ << "RtApiAlsa::startStream: error preparing input pcm device, " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-        goto unlock;

-      }

-    }

-  }

-

-  stream_.state = STREAM_RUNNING;

-

- unlock:

-  apiInfo->runnable = true;

-  pthread_cond_signal( &apiInfo->runnable_cv );

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( result >= 0 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiAlsa :: stopStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiAlsa::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_LOCK( &stream_.mutex );

-

-  int result = 0;

-  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;

-  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    if ( apiInfo->synchronized ) 

-      result = snd_pcm_drop( handle[0] );

-    else

-      result = snd_pcm_drain( handle[0] );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::stopStream: error draining output pcm device, " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {

-    result = snd_pcm_drop( handle[1] );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::stopStream: error stopping input pcm device, " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

- unlock:

-  apiInfo->runnable = false; // fixes high CPU usage when stopped

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( result >= 0 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiAlsa :: abortStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiAlsa::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_LOCK( &stream_.mutex );

-

-  int result = 0;

-  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;

-  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    result = snd_pcm_drop( handle[0] );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::abortStream: error aborting output pcm device, " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

-  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {

-    result = snd_pcm_drop( handle[1] );

-    if ( result < 0 ) {

-      errorStream_ << "RtApiAlsa::abortStream: error aborting input pcm device, " << snd_strerror( result ) << ".";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

- unlock:

-  apiInfo->runnable = false; // fixes high CPU usage when stopped

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( result >= 0 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiAlsa :: callbackEvent()

-{

-  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_LOCK( &stream_.mutex );

-    while ( !apiInfo->runnable )

-      pthread_cond_wait( &apiInfo->runnable_cv, &stream_.mutex );

-

-    if ( stream_.state != STREAM_RUNNING ) {

-      MUTEX_UNLOCK( &stream_.mutex );

-      return;

-    }

-    MUTEX_UNLOCK( &stream_.mutex );

-  }

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiAlsa::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  int doStopStream = 0;

-  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;

-  double streamTime = getStreamTime();

-  RtAudioStreamStatus status = 0;

-  if ( stream_.mode != INPUT && apiInfo->xrun[0] == true ) {

-    status |= RTAUDIO_OUTPUT_UNDERFLOW;

-    apiInfo->xrun[0] = false;

-  }

-  if ( stream_.mode != OUTPUT && apiInfo->xrun[1] == true ) {

-    status |= RTAUDIO_INPUT_OVERFLOW;

-    apiInfo->xrun[1] = false;

-  }

-  doStopStream = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                           stream_.bufferSize, streamTime, status, stream_.callbackInfo.userData );

-

-  if ( doStopStream == 2 ) {

-    abortStream();

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  // The state might change while waiting on a mutex.

-  if ( stream_.state == STREAM_STOPPED ) goto unlock;

-

-  int result;

-  char *buffer;

-  int channels;

-  snd_pcm_t **handle;

-  snd_pcm_sframes_t frames;

-  RtAudioFormat format;

-  handle = (snd_pcm_t **) apiInfo->handles;

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    // Setup parameters.

-    if ( stream_.doConvertBuffer[1] ) {

-      buffer = stream_.deviceBuffer;

-      channels = stream_.nDeviceChannels[1];

-      format = stream_.deviceFormat[1];

-    }

-    else {

-      buffer = stream_.userBuffer[1];

-      channels = stream_.nUserChannels[1];

-      format = stream_.userFormat;

-    }

-

-    // Read samples from device in interleaved/non-interleaved format.

-    if ( stream_.deviceInterleaved[1] )

-      result = snd_pcm_readi( handle[1], buffer, stream_.bufferSize );

-    else {

-      void *bufs[channels];

-      size_t offset = stream_.bufferSize * formatBytes( format );

-      for ( int i=0; i<channels; i++ )

-        bufs[i] = (void *) (buffer + (i * offset));

-      result = snd_pcm_readn( handle[1], bufs, stream_.bufferSize );

-    }

-

-    if ( result < (int) stream_.bufferSize ) {

-      // Either an error or overrun occured.

-      if ( result == -EPIPE ) {

-        snd_pcm_state_t state = snd_pcm_state( handle[1] );

-        if ( state == SND_PCM_STATE_XRUN ) {

-          apiInfo->xrun[1] = true;

-          result = snd_pcm_prepare( handle[1] );

-          if ( result < 0 ) {

-            errorStream_ << "RtApiAlsa::callbackEvent: error preparing device after overrun, " << snd_strerror( result ) << ".";

-            errorText_ = errorStream_.str();

-          }

-        }

-        else {

-          errorStream_ << "RtApiAlsa::callbackEvent: error, current state is " << snd_pcm_state_name( state ) << ", " << snd_strerror( result ) << ".";

-          errorText_ = errorStream_.str();

-        }

-      }

-      else {

-        errorStream_ << "RtApiAlsa::callbackEvent: audio read error, " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-      }

-      error( RtAudioError::WARNING );

-      goto tryOutput;

-    }

-

-    // Do byte swapping if necessary.

-    if ( stream_.doByteSwap[1] )

-      byteSwapBuffer( buffer, stream_.bufferSize * channels, format );

-

-    // Do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[1] )

-      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );

-

-    // Check stream latency

-    result = snd_pcm_delay( handle[1], &frames );

-    if ( result == 0 && frames > 0 ) stream_.latency[1] = frames;

-  }

-

- tryOutput:

-

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    // Setup parameters and do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[0] ) {

-      buffer = stream_.deviceBuffer;

-      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-      channels = stream_.nDeviceChannels[0];

-      format = stream_.deviceFormat[0];

-    }

-    else {

-      buffer = stream_.userBuffer[0];

-      channels = stream_.nUserChannels[0];

-      format = stream_.userFormat;

-    }

-

-    // Do byte swapping if necessary.

-    if ( stream_.doByteSwap[0] )

-      byteSwapBuffer(buffer, stream_.bufferSize * channels, format);

-

-    // Write samples to device in interleaved/non-interleaved format.

-    if ( stream_.deviceInterleaved[0] )

-      result = snd_pcm_writei( handle[0], buffer, stream_.bufferSize );

-    else {

-      void *bufs[channels];

-      size_t offset = stream_.bufferSize * formatBytes( format );

-      for ( int i=0; i<channels; i++ )

-        bufs[i] = (void *) (buffer + (i * offset));

-      result = snd_pcm_writen( handle[0], bufs, stream_.bufferSize );

-    }

-

-    if ( result < (int) stream_.bufferSize ) {

-      // Either an error or underrun occured.

-      if ( result == -EPIPE ) {

-        snd_pcm_state_t state = snd_pcm_state( handle[0] );

-        if ( state == SND_PCM_STATE_XRUN ) {

-          apiInfo->xrun[0] = true;

-          result = snd_pcm_prepare( handle[0] );

-          if ( result < 0 ) {

-            errorStream_ << "RtApiAlsa::callbackEvent: error preparing device after underrun, " << snd_strerror( result ) << ".";

-            errorText_ = errorStream_.str();

-          }

-          else

-            errorText_ =  "RtApiAlsa::callbackEvent: audio write error, underrun.";

-        }

-        else {

-          errorStream_ << "RtApiAlsa::callbackEvent: error, current state is " << snd_pcm_state_name( state ) << ", " << snd_strerror( result ) << ".";

-          errorText_ = errorStream_.str();

-        }

-      }

-      else {

-        errorStream_ << "RtApiAlsa::callbackEvent: audio write error, " << snd_strerror( result ) << ".";

-        errorText_ = errorStream_.str();

-      }

-      error( RtAudioError::WARNING );

-      goto unlock;

-    }

-

-    // Check stream latency

-    result = snd_pcm_delay( handle[0], &frames );

-    if ( result == 0 && frames > 0 ) stream_.latency[0] = frames;

-  }

-

- unlock:

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  RtApi::tickStreamTime();

-  if ( doStopStream == 1 ) this->stopStream();

-}

-

-static void *alsaCallbackHandler( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiAlsa *object = (RtApiAlsa *) info->object;

-  bool *isRunning = &info->isRunning;

-

-#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)

-  if ( &info->doRealtime ) {

-    pthread_t tID = pthread_self();	 // ID of this thread

-    sched_param prio = { info->priority }; // scheduling priority of thread

-    pthread_setschedparam( tID, SCHED_RR, &prio );

-  }

-#endif

-

-  while ( *isRunning == true ) {

-    pthread_testcancel();

-    object->callbackEvent();

-  }

-

-  pthread_exit( NULL );

-}

-

-//******************** End of __LINUX_ALSA__ *********************//

-#endif

-

-#if defined(__LINUX_PULSE__)

-

-// Code written by Peter Meerwald, pmeerw@pmeerw.net

-// and Tristan Matthews.

-

-#include <pulse/error.h>

-#include <pulse/simple.h>

-#include <cstdio>

-

-static const unsigned int SUPPORTED_SAMPLERATES[] = { 8000, 16000, 22050, 32000,

-                                                      44100, 48000, 96000, 0};

-

-struct rtaudio_pa_format_mapping_t {

-  RtAudioFormat rtaudio_format;

-  pa_sample_format_t pa_format;

-};

-

-static const rtaudio_pa_format_mapping_t supported_sampleformats[] = {

-  {RTAUDIO_SINT16, PA_SAMPLE_S16LE},

-  {RTAUDIO_SINT32, PA_SAMPLE_S32LE},

-  {RTAUDIO_FLOAT32, PA_SAMPLE_FLOAT32LE},

-  {0, PA_SAMPLE_INVALID}};

-

-struct PulseAudioHandle {

-  pa_simple *s_play;

-  pa_simple *s_rec;

-  pthread_t thread;

-  pthread_cond_t runnable_cv;

-  bool runnable;

-  PulseAudioHandle() : s_play(0), s_rec(0), runnable(false) { }

-};

-

-RtApiPulse::~RtApiPulse()

-{

-  if ( stream_.state != STREAM_CLOSED )

-    closeStream();

-}

-

-unsigned int RtApiPulse::getDeviceCount( void )

-{

-  return 1;

-}

-

-RtAudio::DeviceInfo RtApiPulse::getDeviceInfo( unsigned int /*device*/ )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = true;

-  info.name = "PulseAudio";

-  info.outputChannels = 2;

-  info.inputChannels = 2;

-  info.duplexChannels = 2;

-  info.isDefaultOutput = true;

-  info.isDefaultInput = true;

-

-  for ( const unsigned int *sr = SUPPORTED_SAMPLERATES; *sr; ++sr )

-    info.sampleRates.push_back( *sr );

-

-  info.preferredSampleRate = 48000;

-  info.nativeFormats = RTAUDIO_SINT16 | RTAUDIO_SINT32 | RTAUDIO_FLOAT32;

-

-  return info;

-}

-

-static void *pulseaudio_callback( void * user )

-{

-  CallbackInfo *cbi = static_cast<CallbackInfo *>( user );

-  RtApiPulse *context = static_cast<RtApiPulse *>( cbi->object );

-  volatile bool *isRunning = &cbi->isRunning;

-

-  while ( *isRunning ) {

-    pthread_testcancel();

-    context->callbackEvent();

-  }

-

-  pthread_exit( NULL );

-}

-

-void RtApiPulse::closeStream( void )

-{

-  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );

-

-  stream_.callbackInfo.isRunning = false;

-  if ( pah ) {

-    MUTEX_LOCK( &stream_.mutex );

-    if ( stream_.state == STREAM_STOPPED ) {

-      pah->runnable = true;

-      pthread_cond_signal( &pah->runnable_cv );

-    }

-    MUTEX_UNLOCK( &stream_.mutex );

-

-    pthread_join( pah->thread, 0 );

-    if ( pah->s_play ) {

-      pa_simple_flush( pah->s_play, NULL );

-      pa_simple_free( pah->s_play );

-    }

-    if ( pah->s_rec )

-      pa_simple_free( pah->s_rec );

-

-    pthread_cond_destroy( &pah->runnable_cv );

-    delete pah;

-    stream_.apiHandle = 0;

-  }

-

-  if ( stream_.userBuffer[0] ) {

-    free( stream_.userBuffer[0] );

-    stream_.userBuffer[0] = 0;

-  }

-  if ( stream_.userBuffer[1] ) {

-    free( stream_.userBuffer[1] );

-    stream_.userBuffer[1] = 0;

-  }

-

-  stream_.state = STREAM_CLOSED;

-  stream_.mode = UNINITIALIZED;

-}

-

-void RtApiPulse::callbackEvent( void )

-{

-  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );

-

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_LOCK( &stream_.mutex );

-    while ( !pah->runnable )

-      pthread_cond_wait( &pah->runnable_cv, &stream_.mutex );

-

-    if ( stream_.state != STREAM_RUNNING ) {

-      MUTEX_UNLOCK( &stream_.mutex );

-      return;

-    }

-    MUTEX_UNLOCK( &stream_.mutex );

-  }

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiPulse::callbackEvent(): the stream is closed ... "

-      "this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;

-  double streamTime = getStreamTime();

-  RtAudioStreamStatus status = 0;

-  int doStopStream = callback( stream_.userBuffer[OUTPUT], stream_.userBuffer[INPUT],

-                               stream_.bufferSize, streamTime, status,

-                               stream_.callbackInfo.userData );

-

-  if ( doStopStream == 2 ) {

-    abortStream();

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-  void *pulse_in = stream_.doConvertBuffer[INPUT] ? stream_.deviceBuffer : stream_.userBuffer[INPUT];

-  void *pulse_out = stream_.doConvertBuffer[OUTPUT] ? stream_.deviceBuffer : stream_.userBuffer[OUTPUT];

-

-  if ( stream_.state != STREAM_RUNNING )

-    goto unlock;

-

-  int pa_error;

-  size_t bytes;

-  if (stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    if ( stream_.doConvertBuffer[OUTPUT] ) {

-        convertBuffer( stream_.deviceBuffer,

-                       stream_.userBuffer[OUTPUT],

-                       stream_.convertInfo[OUTPUT] );

-        bytes = stream_.nDeviceChannels[OUTPUT] * stream_.bufferSize *

-                formatBytes( stream_.deviceFormat[OUTPUT] );

-    } else

-        bytes = stream_.nUserChannels[OUTPUT] * stream_.bufferSize *

-                formatBytes( stream_.userFormat );

-

-    if ( pa_simple_write( pah->s_play, pulse_out, bytes, &pa_error ) < 0 ) {

-      errorStream_ << "RtApiPulse::callbackEvent: audio write error, " <<

-        pa_strerror( pa_error ) << ".";

-      errorText_ = errorStream_.str();

-      error( RtAudioError::WARNING );

-    }

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX) {

-    if ( stream_.doConvertBuffer[INPUT] )

-      bytes = stream_.nDeviceChannels[INPUT] * stream_.bufferSize *

-        formatBytes( stream_.deviceFormat[INPUT] );

-    else

-      bytes = stream_.nUserChannels[INPUT] * stream_.bufferSize *

-        formatBytes( stream_.userFormat );

-            

-    if ( pa_simple_read( pah->s_rec, pulse_in, bytes, &pa_error ) < 0 ) {

-      errorStream_ << "RtApiPulse::callbackEvent: audio read error, " <<

-        pa_strerror( pa_error ) << ".";

-      errorText_ = errorStream_.str();

-      error( RtAudioError::WARNING );

-    }

-    if ( stream_.doConvertBuffer[INPUT] ) {

-      convertBuffer( stream_.userBuffer[INPUT],

-                     stream_.deviceBuffer,

-                     stream_.convertInfo[INPUT] );

-    }

-  }

-

- unlock:

-  MUTEX_UNLOCK( &stream_.mutex );

-  RtApi::tickStreamTime();

-

-  if ( doStopStream == 1 )

-    stopStream();

-}

-

-void RtApiPulse::startStream( void )

-{

-  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiPulse::startStream(): the stream is not open!";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiPulse::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  stream_.state = STREAM_RUNNING;

-

-  pah->runnable = true;

-  pthread_cond_signal( &pah->runnable_cv );

-  MUTEX_UNLOCK( &stream_.mutex );

-}

-

-void RtApiPulse::stopStream( void )

-{

-  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiPulse::stopStream(): the stream is not open!";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiPulse::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_LOCK( &stream_.mutex );

-

-  if ( pah && pah->s_play ) {

-    int pa_error;

-    if ( pa_simple_drain( pah->s_play, &pa_error ) < 0 ) {

-      errorStream_ << "RtApiPulse::stopStream: error draining output device, " <<

-        pa_strerror( pa_error ) << ".";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_UNLOCK( &stream_.mutex );

-}

-

-void RtApiPulse::abortStream( void )

-{

-  PulseAudioHandle *pah = static_cast<PulseAudioHandle*>( stream_.apiHandle );

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiPulse::abortStream(): the stream is not open!";

-    error( RtAudioError::INVALID_USE );

-    return;

-  }

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiPulse::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_LOCK( &stream_.mutex );

-

-  if ( pah && pah->s_play ) {

-    int pa_error;

-    if ( pa_simple_flush( pah->s_play, &pa_error ) < 0 ) {

-      errorStream_ << "RtApiPulse::abortStream: error flushing output device, " <<

-        pa_strerror( pa_error ) << ".";

-      errorText_ = errorStream_.str();

-      MUTEX_UNLOCK( &stream_.mutex );

-      error( RtAudioError::SYSTEM_ERROR );

-      return;

-    }

-  }

-

-  stream_.state = STREAM_STOPPED;

-  MUTEX_UNLOCK( &stream_.mutex );

-}

-

-bool RtApiPulse::probeDeviceOpen( unsigned int device, StreamMode mode,

-                                  unsigned int channels, unsigned int firstChannel,

-                                  unsigned int sampleRate, RtAudioFormat format,

-                                  unsigned int *bufferSize, RtAudio::StreamOptions *options )

-{

-  PulseAudioHandle *pah = 0;

-  unsigned long bufferBytes = 0;

-  pa_sample_spec ss;

-

-  if ( device != 0 ) return false;

-  if ( mode != INPUT && mode != OUTPUT ) return false;

-  if ( channels != 1 && channels != 2 ) {

-    errorText_ = "RtApiPulse::probeDeviceOpen: unsupported number of channels.";

-    return false;

-  }

-  ss.channels = channels;

-

-  if ( firstChannel != 0 ) return false;

-

-  bool sr_found = false;

-  for ( const unsigned int *sr = SUPPORTED_SAMPLERATES; *sr; ++sr ) {

-    if ( sampleRate == *sr ) {

-      sr_found = true;

-      stream_.sampleRate = sampleRate;

-      ss.rate = sampleRate;

-      break;

-    }

-  }

-  if ( !sr_found ) {

-    errorText_ = "RtApiPulse::probeDeviceOpen: unsupported sample rate.";

-    return false;

-  }

-

-  bool sf_found = 0;

-  for ( const rtaudio_pa_format_mapping_t *sf = supported_sampleformats;

-        sf->rtaudio_format && sf->pa_format != PA_SAMPLE_INVALID; ++sf ) {

-    if ( format == sf->rtaudio_format ) {

-      sf_found = true;

-      stream_.userFormat = sf->rtaudio_format;

-      stream_.deviceFormat[mode] = stream_.userFormat;

-      ss.format = sf->pa_format;

-      break;

-    }

-  }

-  if ( !sf_found ) { // Use internal data format conversion.

-    stream_.userFormat = format;

-    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;

-    ss.format = PA_SAMPLE_FLOAT32LE;

-  }

-

-  // Set other stream parameters.

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;

-  else stream_.userInterleaved = true;

-  stream_.deviceInterleaved[mode] = true;

-  stream_.nBuffers = 1;

-  stream_.doByteSwap[mode] = false;

-  stream_.nUserChannels[mode] = channels;

-  stream_.nDeviceChannels[mode] = channels + firstChannel;

-  stream_.channelOffset[mode] = 0;

-  std::string streamName = "RtAudio";

-

-  // Set flags for buffer conversion.

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate necessary internal buffers.

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiPulse::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-  stream_.bufferSize = *bufferSize;

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( mode == INPUT ) {

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-        if ( bufferBytes <= bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiPulse::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  stream_.device[mode] = device;

-

-  // Setup the buffer conversion information structure.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );

-

-  if ( !stream_.apiHandle ) {

-    PulseAudioHandle *pah = new PulseAudioHandle;

-    if ( !pah ) {

-      errorText_ = "RtApiPulse::probeDeviceOpen: error allocating memory for handle.";

-      goto error;

-    }

-

-    stream_.apiHandle = pah;

-    if ( pthread_cond_init( &pah->runnable_cv, NULL ) != 0 ) {

-      errorText_ = "RtApiPulse::probeDeviceOpen: error creating condition variable.";

-      goto error;

-    }

-  }

-  pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );

-

-  int error;

-  if ( options && !options->streamName.empty() ) streamName = options->streamName;

-  switch ( mode ) {

-  case INPUT:

-    pa_buffer_attr buffer_attr;

-    buffer_attr.fragsize = bufferBytes;

-    buffer_attr.maxlength = -1;

-

-    pah->s_rec = pa_simple_new( NULL, streamName.c_str(), PA_STREAM_RECORD, NULL, "Record", &ss, NULL, &buffer_attr, &error );

-    if ( !pah->s_rec ) {

-      errorText_ = "RtApiPulse::probeDeviceOpen: error connecting input to PulseAudio server.";

-      goto error;

-    }

-    break;

-  case OUTPUT:

-    pah->s_play = pa_simple_new( NULL, "RtAudio", PA_STREAM_PLAYBACK, NULL, "Playback", &ss, NULL, NULL, &error );

-    if ( !pah->s_play ) {

-      errorText_ = "RtApiPulse::probeDeviceOpen: error connecting output to PulseAudio server.";

-      goto error;

-    }

-    break;

-  default:

-    goto error;

-  }

-

-  if ( stream_.mode == UNINITIALIZED )

-    stream_.mode = mode;

-  else if ( stream_.mode == mode )

-    goto error;

-  else

-    stream_.mode = DUPLEX;

-

-  if ( !stream_.callbackInfo.isRunning ) {

-    stream_.callbackInfo.object = this;

-    stream_.callbackInfo.isRunning = true;

-    if ( pthread_create( &pah->thread, NULL, pulseaudio_callback, (void *)&stream_.callbackInfo) != 0 ) {

-      errorText_ = "RtApiPulse::probeDeviceOpen: error creating thread.";

-      goto error;

-    }

-  }

-

-  stream_.state = STREAM_STOPPED;

-  return true;

- 

- error:

-  if ( pah && stream_.callbackInfo.isRunning ) {

-    pthread_cond_destroy( &pah->runnable_cv );

-    delete pah;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  return FAILURE;

-}

-

-//******************** End of __LINUX_PULSE__ *********************//

-#endif

-

-#if defined(__LINUX_OSS__)

-

-#include <unistd.h>

-#include <sys/ioctl.h>

-#include <unistd.h>

-#include <fcntl.h>

-#include <sys/soundcard.h>

-#include <errno.h>

-#include <math.h>

-

-static void *ossCallbackHandler(void * ptr);

-

-// A structure to hold various information related to the OSS API

-// implementation.

-struct OssHandle {

-  int id[2];    // device ids

-  bool xrun[2];

-  bool triggered;

-  pthread_cond_t runnable;

-

-  OssHandle()

-    :triggered(false) { id[0] = 0; id[1] = 0; xrun[0] = false; xrun[1] = false; }

-};

-

-RtApiOss :: RtApiOss()

-{

-  // Nothing to do here.

-}

-

-RtApiOss :: ~RtApiOss()

-{

-  if ( stream_.state != STREAM_CLOSED ) closeStream();

-}

-

-unsigned int RtApiOss :: getDeviceCount( void )

-{

-  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );

-  if ( mixerfd == -1 ) {

-    errorText_ = "RtApiOss::getDeviceCount: error opening '/dev/mixer'.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  oss_sysinfo sysinfo;

-  if ( ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo ) == -1 ) {

-    close( mixerfd );

-    errorText_ = "RtApiOss::getDeviceCount: error getting sysinfo, OSS version >= 4.0 is required.";

-    error( RtAudioError::WARNING );

-    return 0;

-  }

-

-  close( mixerfd );

-  return sysinfo.numaudios;

-}

-

-RtAudio::DeviceInfo RtApiOss :: getDeviceInfo( unsigned int device )

-{

-  RtAudio::DeviceInfo info;

-  info.probed = false;

-

-  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );

-  if ( mixerfd == -1 ) {

-    errorText_ = "RtApiOss::getDeviceInfo: error opening '/dev/mixer'.";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  oss_sysinfo sysinfo;

-  int result = ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo );

-  if ( result == -1 ) {

-    close( mixerfd );

-    errorText_ = "RtApiOss::getDeviceInfo: error getting sysinfo, OSS version >= 4.0 is required.";

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  unsigned nDevices = sysinfo.numaudios;

-  if ( nDevices == 0 ) {

-    close( mixerfd );

-    errorText_ = "RtApiOss::getDeviceInfo: no devices found!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  if ( device >= nDevices ) {

-    close( mixerfd );

-    errorText_ = "RtApiOss::getDeviceInfo: device ID is invalid!";

-    error( RtAudioError::INVALID_USE );

-    return info;

-  }

-

-  oss_audioinfo ainfo;

-  ainfo.dev = device;

-  result = ioctl( mixerfd, SNDCTL_AUDIOINFO, &ainfo );

-  close( mixerfd );

-  if ( result == -1 ) {

-    errorStream_ << "RtApiOss::getDeviceInfo: error getting device (" << ainfo.name << ") info.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Probe channels

-  if ( ainfo.caps & PCM_CAP_OUTPUT ) info.outputChannels = ainfo.max_channels;

-  if ( ainfo.caps & PCM_CAP_INPUT ) info.inputChannels = ainfo.max_channels;

-  if ( ainfo.caps & PCM_CAP_DUPLEX ) {

-    if ( info.outputChannels > 0 && info.inputChannels > 0 && ainfo.caps & PCM_CAP_DUPLEX )

-      info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;

-  }

-

-  // Probe data formats ... do for input

-  unsigned long mask = ainfo.iformats;

-  if ( mask & AFMT_S16_LE || mask & AFMT_S16_BE )

-    info.nativeFormats |= RTAUDIO_SINT16;

-  if ( mask & AFMT_S8 )

-    info.nativeFormats |= RTAUDIO_SINT8;

-  if ( mask & AFMT_S32_LE || mask & AFMT_S32_BE )

-    info.nativeFormats |= RTAUDIO_SINT32;

-  if ( mask & AFMT_FLOAT )

-    info.nativeFormats |= RTAUDIO_FLOAT32;

-  if ( mask & AFMT_S24_LE || mask & AFMT_S24_BE )

-    info.nativeFormats |= RTAUDIO_SINT24;

-

-  // Check that we have at least one supported format

-  if ( info.nativeFormats == 0 ) {

-    errorStream_ << "RtApiOss::getDeviceInfo: device (" << ainfo.name << ") data format not supported by RtAudio.";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-    return info;

-  }

-

-  // Probe the supported sample rates.

-  info.sampleRates.clear();

-  if ( ainfo.nrates ) {

-    for ( unsigned int i=0; i<ainfo.nrates; i++ ) {

-      for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {

-        if ( ainfo.rates[i] == SAMPLE_RATES[k] ) {

-          info.sampleRates.push_back( SAMPLE_RATES[k] );

-

-          if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )

-            info.preferredSampleRate = SAMPLE_RATES[k];

-

-          break;

-        }

-      }

-    }

-  }

-  else {

-    // Check min and max rate values;

-    for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {

-      if ( ainfo.min_rate <= (int) SAMPLE_RATES[k] && ainfo.max_rate >= (int) SAMPLE_RATES[k] ) {

-        info.sampleRates.push_back( SAMPLE_RATES[k] );

-

-        if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )

-          info.preferredSampleRate = SAMPLE_RATES[k];

-      }

-    }

-  }

-

-  if ( info.sampleRates.size() == 0 ) {

-    errorStream_ << "RtApiOss::getDeviceInfo: no supported sample rates found for device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    error( RtAudioError::WARNING );

-  }

-  else {

-    info.probed = true;

-    info.name = ainfo.name;

-  }

-

-  return info;

-}

-

-

-bool RtApiOss :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,

-                                  unsigned int firstChannel, unsigned int sampleRate,

-                                  RtAudioFormat format, unsigned int *bufferSize,

-                                  RtAudio::StreamOptions *options )

-{

-  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );

-  if ( mixerfd == -1 ) {

-    errorText_ = "RtApiOss::probeDeviceOpen: error opening '/dev/mixer'.";

-    return FAILURE;

-  }

-

-  oss_sysinfo sysinfo;

-  int result = ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo );

-  if ( result == -1 ) {

-    close( mixerfd );

-    errorText_ = "RtApiOss::probeDeviceOpen: error getting sysinfo, OSS version >= 4.0 is required.";

-    return FAILURE;

-  }

-

-  unsigned nDevices = sysinfo.numaudios;

-  if ( nDevices == 0 ) {

-    // This should not happen because a check is made before this function is called.

-    close( mixerfd );

-    errorText_ = "RtApiOss::probeDeviceOpen: no devices found!";

-    return FAILURE;

-  }

-

-  if ( device >= nDevices ) {

-    // This should not happen because a check is made before this function is called.

-    close( mixerfd );

-    errorText_ = "RtApiOss::probeDeviceOpen: device ID is invalid!";

-    return FAILURE;

-  }

-

-  oss_audioinfo ainfo;

-  ainfo.dev = device;

-  result = ioctl( mixerfd, SNDCTL_AUDIOINFO, &ainfo );

-  close( mixerfd );

-  if ( result == -1 ) {

-    errorStream_ << "RtApiOss::getDeviceInfo: error getting device (" << ainfo.name << ") info.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Check if device supports input or output

-  if ( ( mode == OUTPUT && !( ainfo.caps & PCM_CAP_OUTPUT ) ) ||

-       ( mode == INPUT && !( ainfo.caps & PCM_CAP_INPUT ) ) ) {

-    if ( mode == OUTPUT )

-      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support output.";

-    else

-      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support input.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  int flags = 0;

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  if ( mode == OUTPUT )

-    flags |= O_WRONLY;

-  else { // mode == INPUT

-    if (stream_.mode == OUTPUT && stream_.device[0] == device) {

-      // We just set the same device for playback ... close and reopen for duplex (OSS only).

-      close( handle->id[0] );

-      handle->id[0] = 0;

-      if ( !( ainfo.caps & PCM_CAP_DUPLEX ) ) {

-        errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support duplex mode.";

-        errorText_ = errorStream_.str();

-        return FAILURE;

-      }

-      // Check that the number previously set channels is the same.

-      if ( stream_.nUserChannels[0] != channels ) {

-        errorStream_ << "RtApiOss::probeDeviceOpen: input/output channels must be equal for OSS duplex device (" << ainfo.name << ").";

-        errorText_ = errorStream_.str();

-        return FAILURE;

-      }

-      flags |= O_RDWR;

-    }

-    else

-      flags |= O_RDONLY;

-  }

-

-  // Set exclusive access if specified.

-  if ( options && options->flags & RTAUDIO_HOG_DEVICE ) flags |= O_EXCL;

-

-  // Try to open the device.

-  int fd;

-  fd = open( ainfo.devnode, flags, 0 );

-  if ( fd == -1 ) {

-    if ( errno == EBUSY )

-      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") is busy.";

-    else

-      errorStream_ << "RtApiOss::probeDeviceOpen: error opening device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // For duplex operation, specifically set this mode (this doesn't seem to work).

-  /*

-    if ( flags | O_RDWR ) {

-    result = ioctl( fd, SNDCTL_DSP_SETDUPLEX, NULL );

-    if ( result == -1) {

-    errorStream_ << "RtApiOss::probeDeviceOpen: error setting duplex mode for device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-    }

-    }

-  */

-

-  // Check the device channel support.

-  stream_.nUserChannels[mode] = channels;

-  if ( ainfo.max_channels < (int)(channels + firstChannel) ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: the device (" << ainfo.name << ") does not support requested channel parameters.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Set the number of channels.

-  int deviceChannels = channels + firstChannel;

-  result = ioctl( fd, SNDCTL_DSP_CHANNELS, &deviceChannels );

-  if ( result == -1 || deviceChannels < (int)(channels + firstChannel) ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: error setting channel parameters on device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  stream_.nDeviceChannels[mode] = deviceChannels;

-

-  // Get the data format mask

-  int mask;

-  result = ioctl( fd, SNDCTL_DSP_GETFMTS, &mask );

-  if ( result == -1 ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: error getting device (" << ainfo.name << ") data formats.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Determine how to set the device format.

-  stream_.userFormat = format;

-  int deviceFormat = -1;

-  stream_.doByteSwap[mode] = false;

-  if ( format == RTAUDIO_SINT8 ) {

-    if ( mask & AFMT_S8 ) {

-      deviceFormat = AFMT_S8;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-    }

-  }

-  else if ( format == RTAUDIO_SINT16 ) {

-    if ( mask & AFMT_S16_NE ) {

-      deviceFormat = AFMT_S16_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-    }

-    else if ( mask & AFMT_S16_OE ) {

-      deviceFormat = AFMT_S16_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-      stream_.doByteSwap[mode] = true;

-    }

-  }

-  else if ( format == RTAUDIO_SINT24 ) {

-    if ( mask & AFMT_S24_NE ) {

-      deviceFormat = AFMT_S24_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-    }

-    else if ( mask & AFMT_S24_OE ) {

-      deviceFormat = AFMT_S24_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-      stream_.doByteSwap[mode] = true;

-    }

-  }

-  else if ( format == RTAUDIO_SINT32 ) {

-    if ( mask & AFMT_S32_NE ) {

-      deviceFormat = AFMT_S32_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-    }

-    else if ( mask & AFMT_S32_OE ) {

-      deviceFormat = AFMT_S32_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-      stream_.doByteSwap[mode] = true;

-    }

-  }

-

-  if ( deviceFormat == -1 ) {

-    // The user requested format is not natively supported by the device.

-    if ( mask & AFMT_S16_NE ) {

-      deviceFormat = AFMT_S16_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-    }

-    else if ( mask & AFMT_S32_NE ) {

-      deviceFormat = AFMT_S32_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-    }

-    else if ( mask & AFMT_S24_NE ) {

-      deviceFormat = AFMT_S24_NE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-    }

-    else if ( mask & AFMT_S16_OE ) {

-      deviceFormat = AFMT_S16_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT16;

-      stream_.doByteSwap[mode] = true;

-    }

-    else if ( mask & AFMT_S32_OE ) {

-      deviceFormat = AFMT_S32_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT32;

-      stream_.doByteSwap[mode] = true;

-    }

-    else if ( mask & AFMT_S24_OE ) {

-      deviceFormat = AFMT_S24_OE;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT24;

-      stream_.doByteSwap[mode] = true;

-    }

-    else if ( mask & AFMT_S8) {

-      deviceFormat = AFMT_S8;

-      stream_.deviceFormat[mode] = RTAUDIO_SINT8;

-    }

-  }

-

-  if ( stream_.deviceFormat[mode] == 0 ) {

-    // This really shouldn't happen ...

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") data format not supported by RtAudio.";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Set the data format.

-  int temp = deviceFormat;

-  result = ioctl( fd, SNDCTL_DSP_SETFMT, &deviceFormat );

-  if ( result == -1 || deviceFormat != temp ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: error setting data format on device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Attempt to set the buffer size.  According to OSS, the minimum

-  // number of buffers is two.  The supposed minimum buffer size is 16

-  // bytes, so that will be our lower bound.  The argument to this

-  // call is in the form 0xMMMMSSSS (hex), where the buffer size (in

-  // bytes) is given as 2^SSSS and the number of buffers as 2^MMMM.

-  // We'll check the actual value used near the end of the setup

-  // procedure.

-  int ossBufferBytes = *bufferSize * formatBytes( stream_.deviceFormat[mode] ) * deviceChannels;

-  if ( ossBufferBytes < 16 ) ossBufferBytes = 16;

-  int buffers = 0;

-  if ( options ) buffers = options->numberOfBuffers;

-  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) buffers = 2;

-  if ( buffers < 2 ) buffers = 3;

-  temp = ((int) buffers << 16) + (int)( log10( (double)ossBufferBytes ) / log10( 2.0 ) );

-  result = ioctl( fd, SNDCTL_DSP_SETFRAGMENT, &temp );

-  if ( result == -1 ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: error setting buffer size on device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  stream_.nBuffers = buffers;

-

-  // Save buffer size (in sample frames).

-  *bufferSize = ossBufferBytes / ( formatBytes(stream_.deviceFormat[mode]) * deviceChannels );

-  stream_.bufferSize = *bufferSize;

-

-  // Set the sample rate.

-  int srate = sampleRate;

-  result = ioctl( fd, SNDCTL_DSP_SPEED, &srate );

-  if ( result == -1 ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: error setting sample rate (" << sampleRate << ") on device (" << ainfo.name << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-

-  // Verify the sample rate setup worked.

-  if ( abs( srate - sampleRate ) > 100 ) {

-    close( fd );

-    errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support sample rate (" << sampleRate << ").";

-    errorText_ = errorStream_.str();

-    return FAILURE;

-  }

-  stream_.sampleRate = sampleRate;

-

-  if ( mode == INPUT && stream_.mode == OUTPUT && stream_.device[0] == device) {

-    // We're doing duplex setup here.

-    stream_.deviceFormat[0] = stream_.deviceFormat[1];

-    stream_.nDeviceChannels[0] = deviceChannels;

-  }

-

-  // Set interleaving parameters.

-  stream_.userInterleaved = true;

-  stream_.deviceInterleaved[mode] =  true;

-  if ( options && options->flags & RTAUDIO_NONINTERLEAVED )

-    stream_.userInterleaved = false;

-

-  // Set flags for buffer conversion

-  stream_.doConvertBuffer[mode] = false;

-  if ( stream_.userFormat != stream_.deviceFormat[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )

-    stream_.doConvertBuffer[mode] = true;

-  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&

-       stream_.nUserChannels[mode] > 1 )

-    stream_.doConvertBuffer[mode] = true;

-

-  // Allocate the stream handles if necessary and then save.

-  if ( stream_.apiHandle == 0 ) {

-    try {

-      handle = new OssHandle;

-    }

-    catch ( std::bad_alloc& ) {

-      errorText_ = "RtApiOss::probeDeviceOpen: error allocating OssHandle memory.";

-      goto error;

-    }

-

-    if ( pthread_cond_init( &handle->runnable, NULL ) ) {

-      errorText_ = "RtApiOss::probeDeviceOpen: error initializing pthread condition variable.";

-      goto error;

-    }

-

-    stream_.apiHandle = (void *) handle;

-  }

-  else {

-    handle = (OssHandle *) stream_.apiHandle;

-  }

-  handle->id[mode] = fd;

-

-  // Allocate necessary internal buffers.

-  unsigned long bufferBytes;

-  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );

-  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );

-  if ( stream_.userBuffer[mode] == NULL ) {

-    errorText_ = "RtApiOss::probeDeviceOpen: error allocating user buffer memory.";

-    goto error;

-  }

-

-  if ( stream_.doConvertBuffer[mode] ) {

-

-    bool makeBuffer = true;

-    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );

-    if ( mode == INPUT ) {

-      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {

-        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );

-        if ( bufferBytes <= bytesOut ) makeBuffer = false;

-      }

-    }

-

-    if ( makeBuffer ) {

-      bufferBytes *= *bufferSize;

-      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );

-      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );

-      if ( stream_.deviceBuffer == NULL ) {

-        errorText_ = "RtApiOss::probeDeviceOpen: error allocating device buffer memory.";

-        goto error;

-      }

-    }

-  }

-

-  stream_.device[mode] = device;

-  stream_.state = STREAM_STOPPED;

-

-  // Setup the buffer conversion information structure.

-  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );

-

-  // Setup thread if necessary.

-  if ( stream_.mode == OUTPUT && mode == INPUT ) {

-    // We had already set up an output stream.

-    stream_.mode = DUPLEX;

-    if ( stream_.device[0] == device ) handle->id[0] = fd;

-  }

-  else {

-    stream_.mode = mode;

-

-    // Setup callback thread.

-    stream_.callbackInfo.object = (void *) this;

-

-    // Set the thread attributes for joinable and realtime scheduling

-    // priority.  The higher priority will only take affect if the

-    // program is run as root or suid.

-    pthread_attr_t attr;

-    pthread_attr_init( &attr );

-    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE );

-#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)

-    if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME ) {

-      struct sched_param param;

-      int priority = options->priority;

-      int min = sched_get_priority_min( SCHED_RR );

-      int max = sched_get_priority_max( SCHED_RR );

-      if ( priority < min ) priority = min;

-      else if ( priority > max ) priority = max;

-      param.sched_priority = priority;

-      pthread_attr_setschedparam( &attr, &param );

-      pthread_attr_setschedpolicy( &attr, SCHED_RR );

-    }

-    else

-      pthread_attr_setschedpolicy( &attr, SCHED_OTHER );

-#else

-    pthread_attr_setschedpolicy( &attr, SCHED_OTHER );

-#endif

-

-    stream_.callbackInfo.isRunning = true;

-    result = pthread_create( &stream_.callbackInfo.thread, &attr, ossCallbackHandler, &stream_.callbackInfo );

-    pthread_attr_destroy( &attr );

-    if ( result ) {

-      stream_.callbackInfo.isRunning = false;

-      errorText_ = "RtApiOss::error creating callback thread!";

-      goto error;

-    }

-  }

-

-  return SUCCESS;

-

- error:

-  if ( handle ) {

-    pthread_cond_destroy( &handle->runnable );

-    if ( handle->id[0] ) close( handle->id[0] );

-    if ( handle->id[1] ) close( handle->id[1] );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  return FAILURE;

-}

-

-void RtApiOss :: closeStream()

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiOss::closeStream(): no open stream to close!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  stream_.callbackInfo.isRunning = false;

-  MUTEX_LOCK( &stream_.mutex );

-  if ( stream_.state == STREAM_STOPPED )

-    pthread_cond_signal( &handle->runnable );

-  MUTEX_UNLOCK( &stream_.mutex );

-  pthread_join( stream_.callbackInfo.thread, NULL );

-

-  if ( stream_.state == STREAM_RUNNING ) {

-    if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )

-      ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );

-    else

-      ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );

-    stream_.state = STREAM_STOPPED;

-  }

-

-  if ( handle ) {

-    pthread_cond_destroy( &handle->runnable );

-    if ( handle->id[0] ) close( handle->id[0] );

-    if ( handle->id[1] ) close( handle->id[1] );

-    delete handle;

-    stream_.apiHandle = 0;

-  }

-

-  for ( int i=0; i<2; i++ ) {

-    if ( stream_.userBuffer[i] ) {

-      free( stream_.userBuffer[i] );

-      stream_.userBuffer[i] = 0;

-    }

-  }

-

-  if ( stream_.deviceBuffer ) {

-    free( stream_.deviceBuffer );

-    stream_.deviceBuffer = 0;

-  }

-

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-}

-

-void RtApiOss :: startStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_RUNNING ) {

-    errorText_ = "RtApiOss::startStream(): the stream is already running!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  stream_.state = STREAM_RUNNING;

-

-  // No need to do anything else here ... OSS automatically starts

-  // when fed samples.

-

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  pthread_cond_signal( &handle->runnable );

-}

-

-void RtApiOss :: stopStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiOss::stopStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  // The state might change while waiting on a mutex.

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_UNLOCK( &stream_.mutex );

-    return;

-  }

-

-  int result = 0;

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    // Flush the output with zeros a few times.

-    char *buffer;

-    int samples;

-    RtAudioFormat format;

-

-    if ( stream_.doConvertBuffer[0] ) {

-      buffer = stream_.deviceBuffer;

-      samples = stream_.bufferSize * stream_.nDeviceChannels[0];

-      format = stream_.deviceFormat[0];

-    }

-    else {

-      buffer = stream_.userBuffer[0];

-      samples = stream_.bufferSize * stream_.nUserChannels[0];

-      format = stream_.userFormat;

-    }

-

-    memset( buffer, 0, samples * formatBytes(format) );

-    for ( unsigned int i=0; i<stream_.nBuffers+1; i++ ) {

-      result = write( handle->id[0], buffer, samples * formatBytes(format) );

-      if ( result == -1 ) {

-        errorText_ = "RtApiOss::stopStream: audio write error.";

-        error( RtAudioError::WARNING );

-      }

-    }

-

-    result = ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );

-    if ( result == -1 ) {

-      errorStream_ << "RtApiOss::stopStream: system error stopping callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-    handle->triggered = false;

-  }

-

-  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && handle->id[0] != handle->id[1] ) ) {

-    result = ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );

-    if ( result == -1 ) {

-      errorStream_ << "RtApiOss::stopStream: system error stopping input callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

- unlock:

-  stream_.state = STREAM_STOPPED;

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( result != -1 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiOss :: abortStream()

-{

-  verifyStream();

-  if ( stream_.state == STREAM_STOPPED ) {

-    errorText_ = "RtApiOss::abortStream(): the stream is already stopped!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  // The state might change while waiting on a mutex.

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_UNLOCK( &stream_.mutex );

-    return;

-  }

-

-  int result = 0;

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-    result = ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );

-    if ( result == -1 ) {

-      errorStream_ << "RtApiOss::abortStream: system error stopping callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-    handle->triggered = false;

-  }

-

-  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && handle->id[0] != handle->id[1] ) ) {

-    result = ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );

-    if ( result == -1 ) {

-      errorStream_ << "RtApiOss::abortStream: system error stopping input callback procedure on device (" << stream_.device[0] << ").";

-      errorText_ = errorStream_.str();

-      goto unlock;

-    }

-  }

-

- unlock:

-  stream_.state = STREAM_STOPPED;

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  if ( result != -1 ) return;

-  error( RtAudioError::SYSTEM_ERROR );

-}

-

-void RtApiOss :: callbackEvent()

-{

-  OssHandle *handle = (OssHandle *) stream_.apiHandle;

-  if ( stream_.state == STREAM_STOPPED ) {

-    MUTEX_LOCK( &stream_.mutex );

-    pthread_cond_wait( &handle->runnable, &stream_.mutex );

-    if ( stream_.state != STREAM_RUNNING ) {

-      MUTEX_UNLOCK( &stream_.mutex );

-      return;

-    }

-    MUTEX_UNLOCK( &stream_.mutex );

-  }

-

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApiOss::callbackEvent(): the stream is closed ... this shouldn't happen!";

-    error( RtAudioError::WARNING );

-    return;

-  }

-

-  // Invoke user callback to get fresh output data.

-  int doStopStream = 0;

-  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;

-  double streamTime = getStreamTime();

-  RtAudioStreamStatus status = 0;

-  if ( stream_.mode != INPUT && handle->xrun[0] == true ) {

-    status |= RTAUDIO_OUTPUT_UNDERFLOW;

-    handle->xrun[0] = false;

-  }

-  if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {

-    status |= RTAUDIO_INPUT_OVERFLOW;

-    handle->xrun[1] = false;

-  }

-  doStopStream = callback( stream_.userBuffer[0], stream_.userBuffer[1],

-                           stream_.bufferSize, streamTime, status, stream_.callbackInfo.userData );

-  if ( doStopStream == 2 ) {

-    this->abortStream();

-    return;

-  }

-

-  MUTEX_LOCK( &stream_.mutex );

-

-  // The state might change while waiting on a mutex.

-  if ( stream_.state == STREAM_STOPPED ) goto unlock;

-

-  int result;

-  char *buffer;

-  int samples;

-  RtAudioFormat format;

-

-  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {

-

-    // Setup parameters and do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[0] ) {

-      buffer = stream_.deviceBuffer;

-      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );

-      samples = stream_.bufferSize * stream_.nDeviceChannels[0];

-      format = stream_.deviceFormat[0];

-    }

-    else {

-      buffer = stream_.userBuffer[0];

-      samples = stream_.bufferSize * stream_.nUserChannels[0];

-      format = stream_.userFormat;

-    }

-

-    // Do byte swapping if necessary.

-    if ( stream_.doByteSwap[0] )

-      byteSwapBuffer( buffer, samples, format );

-

-    if ( stream_.mode == DUPLEX && handle->triggered == false ) {

-      int trig = 0;

-      ioctl( handle->id[0], SNDCTL_DSP_SETTRIGGER, &trig );

-      result = write( handle->id[0], buffer, samples * formatBytes(format) );

-      trig = PCM_ENABLE_INPUT|PCM_ENABLE_OUTPUT;

-      ioctl( handle->id[0], SNDCTL_DSP_SETTRIGGER, &trig );

-      handle->triggered = true;

-    }

-    else

-      // Write samples to device.

-      result = write( handle->id[0], buffer, samples * formatBytes(format) );

-

-    if ( result == -1 ) {

-      // We'll assume this is an underrun, though there isn't a

-      // specific means for determining that.

-      handle->xrun[0] = true;

-      errorText_ = "RtApiOss::callbackEvent: audio write error.";

-      error( RtAudioError::WARNING );

-      // Continue on to input section.

-    }

-  }

-

-  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {

-

-    // Setup parameters.

-    if ( stream_.doConvertBuffer[1] ) {

-      buffer = stream_.deviceBuffer;

-      samples = stream_.bufferSize * stream_.nDeviceChannels[1];

-      format = stream_.deviceFormat[1];

-    }

-    else {

-      buffer = stream_.userBuffer[1];

-      samples = stream_.bufferSize * stream_.nUserChannels[1];

-      format = stream_.userFormat;

-    }

-

-    // Read samples from device.

-    result = read( handle->id[1], buffer, samples * formatBytes(format) );

-

-    if ( result == -1 ) {

-      // We'll assume this is an overrun, though there isn't a

-      // specific means for determining that.

-      handle->xrun[1] = true;

-      errorText_ = "RtApiOss::callbackEvent: audio read error.";

-      error( RtAudioError::WARNING );

-      goto unlock;

-    }

-

-    // Do byte swapping if necessary.

-    if ( stream_.doByteSwap[1] )

-      byteSwapBuffer( buffer, samples, format );

-

-    // Do buffer conversion if necessary.

-    if ( stream_.doConvertBuffer[1] )

-      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );

-  }

-

- unlock:

-  MUTEX_UNLOCK( &stream_.mutex );

-

-  RtApi::tickStreamTime();

-  if ( doStopStream == 1 ) this->stopStream();

-}

-

-static void *ossCallbackHandler( void *ptr )

-{

-  CallbackInfo *info = (CallbackInfo *) ptr;

-  RtApiOss *object = (RtApiOss *) info->object;

-  bool *isRunning = &info->isRunning;

-

-  while ( *isRunning == true ) {

-    pthread_testcancel();

-    object->callbackEvent();

-  }

-

-  pthread_exit( NULL );

-}

-

-//******************** End of __LINUX_OSS__ *********************//

-#endif

-

-

-// *************************************************** //

-//

-// Protected common (OS-independent) RtAudio methods.

-//

-// *************************************************** //

-

-// This method can be modified to control the behavior of error

-// message printing.

-void RtApi :: error( RtAudioError::Type type )

-{

-  errorStream_.str(""); // clear the ostringstream

-

-  RtAudioErrorCallback errorCallback = (RtAudioErrorCallback) stream_.callbackInfo.errorCallback;

-  if ( errorCallback ) {

-    // abortStream() can generate new error messages. Ignore them. Just keep original one.

-

-    if ( firstErrorOccurred_ )

-      return;

-

-    firstErrorOccurred_ = true;

-    const std::string errorMessage = errorText_;

-

-    if ( type != RtAudioError::WARNING && stream_.state != STREAM_STOPPED) {

-      stream_.callbackInfo.isRunning = false; // exit from the thread

-      abortStream();

-    }

-

-    errorCallback( type, errorMessage );

-    firstErrorOccurred_ = false;

-    return;

-  }

-

-  if ( type == RtAudioError::WARNING && showWarnings_ == true )

-    std::cerr << '\n' << errorText_ << "\n\n";

-  else if ( type != RtAudioError::WARNING )

-    throw( RtAudioError( errorText_, type ) );

-}

-

-void RtApi :: verifyStream()

-{

-  if ( stream_.state == STREAM_CLOSED ) {

-    errorText_ = "RtApi:: a stream is not open!";

-    error( RtAudioError::INVALID_USE );

-  }

-}

-

-void RtApi :: clearStreamInfo()

-{

-  stream_.mode = UNINITIALIZED;

-  stream_.state = STREAM_CLOSED;

-  stream_.sampleRate = 0;

-  stream_.bufferSize = 0;

-  stream_.nBuffers = 0;

-  stream_.userFormat = 0;

-  stream_.userInterleaved = true;

-  stream_.streamTime = 0.0;

-  stream_.apiHandle = 0;

-  stream_.deviceBuffer = 0;

-  stream_.callbackInfo.callback = 0;

-  stream_.callbackInfo.userData = 0;

-  stream_.callbackInfo.isRunning = false;

-  stream_.callbackInfo.errorCallback = 0;

-  for ( int i=0; i<2; i++ ) {

-    stream_.device[i] = 11111;

-    stream_.doConvertBuffer[i] = false;

-    stream_.deviceInterleaved[i] = true;

-    stream_.doByteSwap[i] = false;

-    stream_.nUserChannels[i] = 0;

-    stream_.nDeviceChannels[i] = 0;

-    stream_.channelOffset[i] = 0;

-    stream_.deviceFormat[i] = 0;

-    stream_.latency[i] = 0;

-    stream_.userBuffer[i] = 0;

-    stream_.convertInfo[i].channels = 0;

-    stream_.convertInfo[i].inJump = 0;

-    stream_.convertInfo[i].outJump = 0;

-    stream_.convertInfo[i].inFormat = 0;

-    stream_.convertInfo[i].outFormat = 0;

-    stream_.convertInfo[i].inOffset.clear();

-    stream_.convertInfo[i].outOffset.clear();

-  }

-}

-

-unsigned int RtApi :: formatBytes( RtAudioFormat format )

-{

-  if ( format == RTAUDIO_SINT16 )

-    return 2;

-  else if ( format == RTAUDIO_SINT32 || format == RTAUDIO_FLOAT32 )

-    return 4;

-  else if ( format == RTAUDIO_FLOAT64 )

-    return 8;

-  else if ( format == RTAUDIO_SINT24 )

-    return 3;

-  else if ( format == RTAUDIO_SINT8 )

-    return 1;

-

-  errorText_ = "RtApi::formatBytes: undefined format.";

-  error( RtAudioError::WARNING );

-

-  return 0;

-}

-

-void RtApi :: setConvertInfo( StreamMode mode, unsigned int firstChannel )

-{

-  if ( mode == INPUT ) { // convert device to user buffer

-    stream_.convertInfo[mode].inJump = stream_.nDeviceChannels[1];

-    stream_.convertInfo[mode].outJump = stream_.nUserChannels[1];

-    stream_.convertInfo[mode].inFormat = stream_.deviceFormat[1];

-    stream_.convertInfo[mode].outFormat = stream_.userFormat;

-  }

-  else { // convert user to device buffer

-    stream_.convertInfo[mode].inJump = stream_.nUserChannels[0];

-    stream_.convertInfo[mode].outJump = stream_.nDeviceChannels[0];

-    stream_.convertInfo[mode].inFormat = stream_.userFormat;

-    stream_.convertInfo[mode].outFormat = stream_.deviceFormat[0];

-  }

-

-  if ( stream_.convertInfo[mode].inJump < stream_.convertInfo[mode].outJump )

-    stream_.convertInfo[mode].channels = stream_.convertInfo[mode].inJump;

-  else

-    stream_.convertInfo[mode].channels = stream_.convertInfo[mode].outJump;

-

-  // Set up the interleave/deinterleave offsets.

-  if ( stream_.deviceInterleaved[mode] != stream_.userInterleaved ) {

-    if ( ( mode == OUTPUT && stream_.deviceInterleaved[mode] ) ||

-         ( mode == INPUT && stream_.userInterleaved ) ) {

-      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {

-        stream_.convertInfo[mode].inOffset.push_back( k * stream_.bufferSize );

-        stream_.convertInfo[mode].outOffset.push_back( k );

-        stream_.convertInfo[mode].inJump = 1;

-      }

-    }

-    else {

-      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {

-        stream_.convertInfo[mode].inOffset.push_back( k );

-        stream_.convertInfo[mode].outOffset.push_back( k * stream_.bufferSize );

-        stream_.convertInfo[mode].outJump = 1;

-      }

-    }

-  }

-  else { // no (de)interleaving

-    if ( stream_.userInterleaved ) {

-      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {

-        stream_.convertInfo[mode].inOffset.push_back( k );

-        stream_.convertInfo[mode].outOffset.push_back( k );

-      }

-    }

-    else {

-      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {

-        stream_.convertInfo[mode].inOffset.push_back( k * stream_.bufferSize );

-        stream_.convertInfo[mode].outOffset.push_back( k * stream_.bufferSize );

-        stream_.convertInfo[mode].inJump = 1;

-        stream_.convertInfo[mode].outJump = 1;

-      }

-    }

-  }

-

-  // Add channel offset.

-  if ( firstChannel > 0 ) {

-    if ( stream_.deviceInterleaved[mode] ) {

-      if ( mode == OUTPUT ) {

-        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )

-          stream_.convertInfo[mode].outOffset[k] += firstChannel;

-      }

-      else {

-        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )

-          stream_.convertInfo[mode].inOffset[k] += firstChannel;

-      }

-    }

-    else {

-      if ( mode == OUTPUT ) {

-        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )

-          stream_.convertInfo[mode].outOffset[k] += ( firstChannel * stream_.bufferSize );

-      }

-      else {

-        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )

-          stream_.convertInfo[mode].inOffset[k] += ( firstChannel  * stream_.bufferSize );

-      }

-    }

-  }

-}

-

-void RtApi :: convertBuffer( char *outBuffer, char *inBuffer, ConvertInfo &info )

-{

-  // This function does format conversion, input/output channel compensation, and

-  // data interleaving/deinterleaving.  24-bit integers are assumed to occupy

-  // the lower three bytes of a 32-bit integer.

-

-  // Clear our device buffer when in/out duplex device channels are different

-  if ( outBuffer == stream_.deviceBuffer && stream_.mode == DUPLEX &&

-       ( stream_.nDeviceChannels[0] < stream_.nDeviceChannels[1] ) )

-    memset( outBuffer, 0, stream_.bufferSize * info.outJump * formatBytes( info.outFormat ) );

-

-  int j;

-  if (info.outFormat == RTAUDIO_FLOAT64) {

-    Float64 scale;

-    Float64 *out = (Float64 *)outBuffer;

-

-    if (info.inFormat == RTAUDIO_SINT8) {

-      signed char *in = (signed char *)inBuffer;

-      scale = 1.0 / 127.5;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT16) {

-      Int16 *in = (Int16 *)inBuffer;

-      scale = 1.0 / 32767.5;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      Int24 *in = (Int24 *)inBuffer;

-      scale = 1.0 / 8388607.5;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float64) (in[info.inOffset[j]].asInt());

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      Int32 *in = (Int32 *)inBuffer;

-      scale = 1.0 / 2147483647.5;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      // Channel compensation and/or (de)interleaving only.

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-  else if (info.outFormat == RTAUDIO_FLOAT32) {

-    Float32 scale;

-    Float32 *out = (Float32 *)outBuffer;

-

-    if (info.inFormat == RTAUDIO_SINT8) {

-      signed char *in = (signed char *)inBuffer;

-      scale = (Float32) ( 1.0 / 127.5 );

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT16) {

-      Int16 *in = (Int16 *)inBuffer;

-      scale = (Float32) ( 1.0 / 32767.5 );

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      Int24 *in = (Int24 *)inBuffer;

-      scale = (Float32) ( 1.0 / 8388607.5 );

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float32) (in[info.inOffset[j]].asInt());

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      Int32 *in = (Int32 *)inBuffer;

-      scale = (Float32) ( 1.0 / 2147483647.5 );

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];

-          out[info.outOffset[j]] += 0.5;

-          out[info.outOffset[j]] *= scale;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      // Channel compensation and/or (de)interleaving only.

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-  else if (info.outFormat == RTAUDIO_SINT32) {

-    Int32 *out = (Int32 *)outBuffer;

-    if (info.inFormat == RTAUDIO_SINT8) {

-      signed char *in = (signed char *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]];

-          out[info.outOffset[j]] <<= 24;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT16) {

-      Int16 *in = (Int16 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]];

-          out[info.outOffset[j]] <<= 16;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      Int24 *in = (Int24 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]].asInt();

-          out[info.outOffset[j]] <<= 8;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      // Channel compensation and/or (de)interleaving only.

-      Int32 *in = (Int32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 2147483647.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 2147483647.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-  else if (info.outFormat == RTAUDIO_SINT24) {

-    Int24 *out = (Int24 *)outBuffer;

-    if (info.inFormat == RTAUDIO_SINT8) {

-      signed char *in = (signed char *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] << 16);

-          //out[info.outOffset[j]] <<= 16;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT16) {

-      Int16 *in = (Int16 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] << 8);

-          //out[info.outOffset[j]] <<= 8;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      // Channel compensation and/or (de)interleaving only.

-      Int24 *in = (Int24 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      Int32 *in = (Int32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] >> 8);

-          //out[info.outOffset[j]] >>= 8;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 8388607.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 8388607.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-  else if (info.outFormat == RTAUDIO_SINT16) {

-    Int16 *out = (Int16 *)outBuffer;

-    if (info.inFormat == RTAUDIO_SINT8) {

-      signed char *in = (signed char *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int16) in[info.inOffset[j]];

-          out[info.outOffset[j]] <<= 8;

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT16) {

-      // Channel compensation and/or (de)interleaving only.

-      Int16 *in = (Int16 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      Int24 *in = (Int24 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]].asInt() >> 8);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      Int32 *in = (Int32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int16) ((in[info.inOffset[j]] >> 16) & 0x0000ffff);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]] * 32767.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]] * 32767.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-  else if (info.outFormat == RTAUDIO_SINT8) {

-    signed char *out = (signed char *)outBuffer;

-    if (info.inFormat == RTAUDIO_SINT8) {

-      // Channel compensation and/or (de)interleaving only.

-      signed char *in = (signed char *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = in[info.inOffset[j]];

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    if (info.inFormat == RTAUDIO_SINT16) {

-      Int16 *in = (Int16 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (signed char) ((in[info.inOffset[j]] >> 8) & 0x00ff);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT24) {

-      Int24 *in = (Int24 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]].asInt() >> 16);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_SINT32) {

-      Int32 *in = (Int32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (signed char) ((in[info.inOffset[j]] >> 24) & 0x000000ff);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT32) {

-      Float32 *in = (Float32 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]] * 127.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-    else if (info.inFormat == RTAUDIO_FLOAT64) {

-      Float64 *in = (Float64 *)inBuffer;

-      for (unsigned int i=0; i<stream_.bufferSize; i++) {

-        for (j=0; j<info.channels; j++) {

-          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]] * 127.5 - 0.5);

-        }

-        in += info.inJump;

-        out += info.outJump;

-      }

-    }

-  }

-}

-

-//static inline uint16_t bswap_16(uint16_t x) { return (x>>8) | (x<<8); }

-//static inline uint32_t bswap_32(uint32_t x) { return (bswap_16(x&0xffff)<<16) | (bswap_16(x>>16)); }

-//static inline uint64_t bswap_64(uint64_t x) { return (((unsigned long long)bswap_32(x&0xffffffffull))<<32) | (bswap_32(x>>32)); }

-

-void RtApi :: byteSwapBuffer( char *buffer, unsigned int samples, RtAudioFormat format )

-{

-  register char val;

-  register char *ptr;

-

-  ptr = buffer;

-  if ( format == RTAUDIO_SINT16 ) {

-    for ( unsigned int i=0; i<samples; i++ ) {

-      // Swap 1st and 2nd bytes.

-      val = *(ptr);

-      *(ptr) = *(ptr+1);

-      *(ptr+1) = val;

-

-      // Increment 2 bytes.

-      ptr += 2;

-    }

-  }

-  else if ( format == RTAUDIO_SINT32 ||

-            format == RTAUDIO_FLOAT32 ) {

-    for ( unsigned int i=0; i<samples; i++ ) {

-      // Swap 1st and 4th bytes.

-      val = *(ptr);

-      *(ptr) = *(ptr+3);

-      *(ptr+3) = val;

-

-      // Swap 2nd and 3rd bytes.

-      ptr += 1;

-      val = *(ptr);

-      *(ptr) = *(ptr+1);

-      *(ptr+1) = val;

-

-      // Increment 3 more bytes.

-      ptr += 3;

-    }

-  }

-  else if ( format == RTAUDIO_SINT24 ) {

-    for ( unsigned int i=0; i<samples; i++ ) {

-      // Swap 1st and 3rd bytes.

-      val = *(ptr);

-      *(ptr) = *(ptr+2);

-      *(ptr+2) = val;

-

-      // Increment 2 more bytes.

-      ptr += 2;

-    }

-  }

-  else if ( format == RTAUDIO_FLOAT64 ) {

-    for ( unsigned int i=0; i<samples; i++ ) {

-      // Swap 1st and 8th bytes

-      val = *(ptr);

-      *(ptr) = *(ptr+7);

-      *(ptr+7) = val;

-

-      // Swap 2nd and 7th bytes

-      ptr += 1;

-      val = *(ptr);

-      *(ptr) = *(ptr+5);

-      *(ptr+5) = val;

-

-      // Swap 3rd and 6th bytes

-      ptr += 1;

-      val = *(ptr);

-      *(ptr) = *(ptr+3);

-      *(ptr+3) = val;

-

-      // Swap 4th and 5th bytes

-      ptr += 1;

-      val = *(ptr);

-      *(ptr) = *(ptr+1);

-      *(ptr+1) = val;

-

-      // Increment 5 more bytes.

-      ptr += 5;

-    }

-  }

-}

-

-  // Indentation settings for Vim and Emacs

-  //

-  // Local Variables:

-  // c-basic-offset: 2

-  // indent-tabs-mode: nil

-  // End:

-  //

-  // vim: et sts=2 sw=2

-

-#endif

+#ifdef RTAUDIO_ENABLED
+/************************************************************************/
+/*! \class RtAudio
+    \brief Realtime audio i/o C++ classes.
+
+    RtAudio provides a common API (Application Programming Interface)
+    for realtime audio input/output across Linux (native ALSA, Jack,
+    and OSS), Macintosh OS X (CoreAudio and Jack), and Windows
+    (DirectSound, ASIO and WASAPI) operating systems.
+
+    RtAudio WWW site: http://www.music.mcgill.ca/~gary/rtaudio/
+
+    RtAudio: realtime audio i/o C++ classes
+    Copyright (c) 2001-2014 Gary P. Scavone
+
+    Permission is hereby granted, free of charge, to any person
+    obtaining a copy of this software and associated documentation files
+    (the "Software"), to deal in the Software without restriction,
+    including without limitation the rights to use, copy, modify, merge,
+    publish, distribute, sublicense, and/or sell copies of the Software,
+    and to permit persons to whom the Software is furnished to do so,
+    subject to the following conditions:
+
+    The above copyright notice and this permission notice shall be
+    included in all copies or substantial portions of the Software.
+
+    Any person wishing to distribute modifications to the Software is
+    asked to send the modifications to the original developer so that
+    they can be incorporated into the canonical version.  This is,
+    however, not a binding provision of this license.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+    EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+    MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+    IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR
+    ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
+    CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+    WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+*/
+/************************************************************************/
+
+// RtAudio: Version 4.1.1
+
+#include "RtAudio.h"
+#include <iostream>
+#include <cstdlib>
+#include <cstring>
+#include <climits>
+#include <algorithm>
+
+// Static variable definitions.
+const unsigned int RtApi::MAX_SAMPLE_RATES = 14;
+const unsigned int RtApi::SAMPLE_RATES[] = {
+  4000, 5512, 8000, 9600, 11025, 16000, 22050,
+  32000, 44100, 48000, 88200, 96000, 176400, 192000
+};
+
+#if defined(__WINDOWS_DS__) || defined(__WINDOWS_ASIO__) || defined(__WINDOWS_WASAPI__)
+#ifdef WINRT_ENABLED
+  #define MUTEX_INITIALIZE(A) InitializeCriticalSectionEx(A, 0, 0)
+#else
+  #define MUTEX_INITIALIZE(A) InitializeCriticalSection(A)
+#endif
+  #define MUTEX_DESTROY(A)    DeleteCriticalSection(A)
+  #define MUTEX_LOCK(A)       EnterCriticalSection(A)
+  #define MUTEX_UNLOCK(A)     LeaveCriticalSection(A)
+
+  #include "tchar.h"
+
+  static std::string convertCharPointerToStdString(const char *text)
+  {
+    return std::string(text);
+  }
+
+  static std::string convertCharPointerToStdString(const wchar_t *text)
+  {
+    int length = WideCharToMultiByte(CP_UTF8, 0, text, -1, NULL, 0, NULL, NULL);
+    std::string s( length-1, '\0' );
+    WideCharToMultiByte(CP_UTF8, 0, text, -1, &s[0], length, NULL, NULL);
+    return s;
+  }
+
+#elif defined(__LINUX_ALSA__) || defined(__LINUX_PULSE__) || defined(__UNIX_JACK__) || defined(__LINUX_OSS__) || defined(__MACOSX_CORE__)
+  // pthread API
+  #define MUTEX_INITIALIZE(A) pthread_mutex_init(A, NULL)
+  #define MUTEX_DESTROY(A)    pthread_mutex_destroy(A)
+  #define MUTEX_LOCK(A)       pthread_mutex_lock(A)
+  #define MUTEX_UNLOCK(A)     pthread_mutex_unlock(A)
+#else
+  #define MUTEX_INITIALIZE(A) abs(*A) // dummy definitions
+  #define MUTEX_DESTROY(A)    abs(*A) // dummy definitions
+#endif
+
+// *************************************************** //
+//
+// RtAudio definitions.
+//
+// *************************************************** //
+
+std::string RtAudio :: getVersion( void ) throw()
+{
+  return RTAUDIO_VERSION;
+}
+
+void RtAudio :: getCompiledApi( std::vector<RtAudio::Api> &apis ) throw()
+{
+  apis.clear();
+
+  // The order here will control the order of RtAudio's API search in
+  // the constructor.
+#if defined(__UNIX_JACK__)
+  apis.push_back( UNIX_JACK );
+#endif
+#if defined(__LINUX_ALSA__)
+  apis.push_back( LINUX_ALSA );
+#endif
+#if defined(__LINUX_PULSE__)
+  apis.push_back( LINUX_PULSE );
+#endif
+#if defined(__LINUX_OSS__)
+  apis.push_back( LINUX_OSS );
+#endif
+#if defined(__WINDOWS_ASIO__)
+  apis.push_back( WINDOWS_ASIO );
+#endif
+#if defined(__WINDOWS_WASAPI__)
+  apis.push_back( WINDOWS_WASAPI );
+#endif
+#if defined(__WINDOWS_DS__)
+  apis.push_back( WINDOWS_DS );
+#endif
+#if defined(__MACOSX_CORE__)
+  apis.push_back( MACOSX_CORE );
+#endif
+#if defined(__RTAUDIO_DUMMY__)
+  apis.push_back( RTAUDIO_DUMMY );
+#endif
+}
+
+void RtAudio :: openRtApi( RtAudio::Api api )
+{
+  if ( rtapi_ )
+    delete rtapi_;
+  rtapi_ = 0;
+
+#if defined(__UNIX_JACK__)
+  if ( api == UNIX_JACK )
+    rtapi_ = new RtApiJack();
+#endif
+#if defined(__LINUX_ALSA__)
+  if ( api == LINUX_ALSA )
+    rtapi_ = new RtApiAlsa();
+#endif
+#if defined(__LINUX_PULSE__)
+  if ( api == LINUX_PULSE )
+    rtapi_ = new RtApiPulse();
+#endif
+#if defined(__LINUX_OSS__)
+  if ( api == LINUX_OSS )
+    rtapi_ = new RtApiOss();
+#endif
+#if defined(__WINDOWS_ASIO__)
+  if ( api == WINDOWS_ASIO )
+    rtapi_ = new RtApiAsio();
+#endif
+#if defined(__WINDOWS_WASAPI__)
+  if ( api == WINDOWS_WASAPI )
+    rtapi_ = new RtApiWasapi();
+#endif
+#if defined(__WINDOWS_DS__)
+  if ( api == WINDOWS_DS )
+    rtapi_ = new RtApiDs();
+#endif
+#if defined(__MACOSX_CORE__)
+  if ( api == MACOSX_CORE )
+    rtapi_ = new RtApiCore();
+#endif
+#if defined(__RTAUDIO_DUMMY__)
+  if ( api == RTAUDIO_DUMMY )
+    rtapi_ = new RtApiDummy();
+#endif
+}
+
+RtAudio :: RtAudio( RtAudio::Api api )
+{
+  rtapi_ = 0;
+
+  if ( api != UNSPECIFIED ) {
+    // Attempt to open the specified API.
+    openRtApi( api );
+    if ( rtapi_ ) return;
+
+    // No compiled support for specified API value.  Issue a debug
+    // warning and continue as if no API was specified.
+    std::cerr << "\nRtAudio: no compiled support for specified API argument!\n" << std::endl;
+  }
+
+  // Iterate through the compiled APIs and return as soon as we find
+  // one with at least one device or we reach the end of the list.
+  std::vector< RtAudio::Api > apis;
+  getCompiledApi( apis );
+  for ( unsigned int i=0; i<apis.size(); i++ ) {
+    openRtApi( apis[i] );
+    if ( rtapi_ && rtapi_->getDeviceCount() ) break;
+  }
+
+  if ( rtapi_ ) return;
+
+  // It should not be possible to get here because the preprocessor
+  // definition __RTAUDIO_DUMMY__ is automatically defined if no
+  // API-specific definitions are passed to the compiler. But just in
+  // case something weird happens, we'll thow an error.
+  std::string errorText = "\nRtAudio: no compiled API support found ... critical error!!\n\n";
+  throw( RtAudioError( errorText, RtAudioError::UNSPECIFIED ) );
+}
+
+RtAudio :: ~RtAudio() throw()
+{
+  if ( rtapi_ )
+    delete rtapi_;
+}
+
+void RtAudio :: openStream( RtAudio::StreamParameters *outputParameters,
+                            RtAudio::StreamParameters *inputParameters,
+                            RtAudioFormat format, unsigned int sampleRate,
+                            unsigned int *bufferFrames,
+                            RtAudioCallback callback, void *userData,
+                            RtAudio::StreamOptions *options,
+                            RtAudioErrorCallback errorCallback )
+{
+  return rtapi_->openStream( outputParameters, inputParameters, format,
+                             sampleRate, bufferFrames, callback,
+                             userData, options, errorCallback );
+}
+
+// *************************************************** //
+//
+// Public RtApi definitions (see end of file for
+// private or protected utility functions).
+//
+// *************************************************** //
+
+RtApi :: RtApi()
+{
+  stream_.state = STREAM_CLOSED;
+  stream_.mode = UNINITIALIZED;
+  stream_.apiHandle = 0;
+  stream_.userBuffer[0] = 0;
+  stream_.userBuffer[1] = 0;
+  MUTEX_INITIALIZE( &stream_.mutex );
+  showWarnings_ = true;
+  firstErrorOccurred_ = false;
+}
+
+RtApi :: ~RtApi()
+{
+  MUTEX_DESTROY( &stream_.mutex );
+}
+
+void RtApi :: openStream( RtAudio::StreamParameters *oParams,
+                          RtAudio::StreamParameters *iParams,
+                          RtAudioFormat format, unsigned int sampleRate,
+                          unsigned int *bufferFrames,
+                          RtAudioCallback callback, void *userData,
+                          RtAudio::StreamOptions *options,
+                          RtAudioErrorCallback errorCallback )
+{
+  if ( stream_.state != STREAM_CLOSED ) {
+    errorText_ = "RtApi::openStream: a stream is already open!";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+
+  // Clear stream information potentially left from a previously open stream.
+  clearStreamInfo();
+
+  if ( oParams && oParams->nChannels < 1 ) {
+    errorText_ = "RtApi::openStream: a non-NULL output StreamParameters structure cannot have an nChannels value less than one.";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+
+  if ( iParams && iParams->nChannels < 1 ) {
+    errorText_ = "RtApi::openStream: a non-NULL input StreamParameters structure cannot have an nChannels value less than one.";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+
+  if ( oParams == NULL && iParams == NULL ) {
+    errorText_ = "RtApi::openStream: input and output StreamParameters structures are both NULL!";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+
+  if ( formatBytes(format) == 0 ) {
+    errorText_ = "RtApi::openStream: 'format' parameter value is undefined.";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+
+  unsigned int nDevices = getDeviceCount();
+  unsigned int oChannels = 0;
+  if ( oParams ) {
+    oChannels = oParams->nChannels;
+    if ( oParams->deviceId >= nDevices ) {
+      errorText_ = "RtApi::openStream: output device parameter value is invalid.";
+      error( RtAudioError::INVALID_USE );
+      return;
+    }
+  }
+
+  unsigned int iChannels = 0;
+  if ( iParams ) {
+    iChannels = iParams->nChannels;
+    if ( iParams->deviceId >= nDevices ) {
+      errorText_ = "RtApi::openStream: input device parameter value is invalid.";
+      error( RtAudioError::INVALID_USE );
+      return;
+    }
+  }
+
+  bool result;
+
+  if ( oChannels > 0 ) {
+
+    result = probeDeviceOpen( oParams->deviceId, OUTPUT, oChannels, oParams->firstChannel,
+                              sampleRate, format, bufferFrames, options );
+    if ( result == false ) {
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+  }
+
+  if ( iChannels > 0 ) {
+
+    result = probeDeviceOpen( iParams->deviceId, INPUT, iChannels, iParams->firstChannel,
+                              sampleRate, format, bufferFrames, options );
+    if ( result == false ) {
+      if ( oChannels > 0 ) closeStream();
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+  }
+
+  stream_.callbackInfo.callback = (void *) callback;
+  stream_.callbackInfo.userData = userData;
+  stream_.callbackInfo.errorCallback = (void *) errorCallback;
+
+  if ( options ) options->numberOfBuffers = stream_.nBuffers;
+  stream_.state = STREAM_STOPPED;
+}
+
+unsigned int RtApi :: getDefaultInputDevice( void )
+{
+  // Should be implemented in subclasses if possible.
+  return 0;
+}
+
+unsigned int RtApi :: getDefaultOutputDevice( void )
+{
+  // Should be implemented in subclasses if possible.
+  return 0;
+}
+
+void RtApi :: closeStream( void )
+{
+  // MUST be implemented in subclasses!
+  return;
+}
+
+bool RtApi :: probeDeviceOpen( unsigned int /*device*/, StreamMode /*mode*/, unsigned int /*channels*/,
+                               unsigned int /*firstChannel*/, unsigned int /*sampleRate*/,
+                               RtAudioFormat /*format*/, unsigned int * /*bufferSize*/,
+                               RtAudio::StreamOptions * /*options*/ )
+{
+  // MUST be implemented in subclasses!
+  return FAILURE;
+}
+
+void RtApi :: tickStreamTime( void )
+{
+  // Subclasses that do not provide their own implementation of
+  // getStreamTime should call this function once per buffer I/O to
+  // provide basic stream time support.
+
+  stream_.streamTime += ( stream_.bufferSize * 1.0 / stream_.sampleRate );
+
+#if defined( HAVE_GETTIMEOFDAY )
+  gettimeofday( &stream_.lastTickTimestamp, NULL );
+#endif
+}
+
+long RtApi :: getStreamLatency( void )
+{
+  verifyStream();
+
+  long totalLatency = 0;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )
+    totalLatency = stream_.latency[0];
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX )
+    totalLatency += stream_.latency[1];
+
+  return totalLatency;
+}
+
+double RtApi :: getStreamTime( void )
+{
+  verifyStream();
+
+#if defined( HAVE_GETTIMEOFDAY )
+  // Return a very accurate estimate of the stream time by
+  // adding in the elapsed time since the last tick.
+  struct timeval then;
+  struct timeval now;
+
+  if ( stream_.state != STREAM_RUNNING || stream_.streamTime == 0.0 )
+    return stream_.streamTime;
+
+  gettimeofday( &now, NULL );
+  then = stream_.lastTickTimestamp;
+  return stream_.streamTime +
+    ((now.tv_sec + 0.000001 * now.tv_usec) -
+     (then.tv_sec + 0.000001 * then.tv_usec));     
+#else
+  return stream_.streamTime;
+#endif
+}
+
+void RtApi :: setStreamTime( double time )
+{
+  verifyStream();
+
+  if ( time >= 0.0 )
+    stream_.streamTime = time;
+}
+
+unsigned int RtApi :: getStreamSampleRate( void )
+{
+ verifyStream();
+
+ return stream_.sampleRate;
+}
+
+
+// *************************************************** //
+//
+// OS/API-specific methods.
+//
+// *************************************************** //
+
+#if defined(__MACOSX_CORE__)
+
+// The OS X CoreAudio API is designed to use a separate callback
+// procedure for each of its audio devices.  A single RtAudio duplex
+// stream using two different devices is supported here, though it
+// cannot be guaranteed to always behave correctly because we cannot
+// synchronize these two callbacks.
+//
+// A property listener is installed for over/underrun information.
+// However, no functionality is currently provided to allow property
+// listeners to trigger user handlers because it is unclear what could
+// be done if a critical stream parameter (buffer size, sample rate,
+// device disconnect) notification arrived.  The listeners entail
+// quite a bit of extra code and most likely, a user program wouldn't
+// be prepared for the result anyway.  However, we do provide a flag
+// to the client callback function to inform of an over/underrun.
+
+// A structure to hold various information related to the CoreAudio API
+// implementation.
+struct CoreHandle {
+  AudioDeviceID id[2];    // device ids
+#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )
+  AudioDeviceIOProcID procId[2];
+#endif
+  UInt32 iStream[2];      // device stream index (or first if using multiple)
+  UInt32 nStreams[2];     // number of streams to use
+  bool xrun[2];
+  char *deviceBuffer;
+  pthread_cond_t condition;
+  int drainCounter;       // Tracks callback counts when draining
+  bool internalDrain;     // Indicates if stop is initiated from callback or not.
+
+  CoreHandle()
+    :deviceBuffer(0), drainCounter(0), internalDrain(false) { nStreams[0] = 1; nStreams[1] = 1; id[0] = 0; id[1] = 0; xrun[0] = false; xrun[1] = false; }
+};
+
+RtApiCore:: RtApiCore()
+{
+#if defined( AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER )
+  // This is a largely undocumented but absolutely necessary
+  // requirement starting with OS-X 10.6.  If not called, queries and
+  // updates to various audio device properties are not handled
+  // correctly.
+  CFRunLoopRef theRunLoop = NULL;
+  AudioObjectPropertyAddress property = { kAudioHardwarePropertyRunLoop,
+                                          kAudioObjectPropertyScopeGlobal,
+                                          kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectSetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, sizeof(CFRunLoopRef), &theRunLoop);
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::RtApiCore: error setting run loop property!";
+    error( RtAudioError::WARNING );
+  }
+#endif
+}
+
+RtApiCore :: ~RtApiCore()
+{
+  // The subclass destructor gets called before the base class
+  // destructor, so close an existing stream before deallocating
+  // apiDeviceId memory.
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+}
+
+unsigned int RtApiCore :: getDeviceCount( void )
+{
+  // Find out how many audio devices there are, if any.
+  UInt32 dataSize;
+  AudioObjectPropertyAddress propertyAddress = { kAudioHardwarePropertyDevices, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectGetPropertyDataSize( kAudioObjectSystemObject, &propertyAddress, 0, NULL, &dataSize );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDeviceCount: OS-X error getting device info!";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  return dataSize / sizeof( AudioDeviceID );
+}
+
+unsigned int RtApiCore :: getDefaultInputDevice( void )
+{
+  unsigned int nDevices = getDeviceCount();
+  if ( nDevices <= 1 ) return 0;
+
+  AudioDeviceID id;
+  UInt32 dataSize = sizeof( AudioDeviceID );
+  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDefaultInputDevice, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, &id );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDefaultInputDevice: OS-X system error getting device.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  dataSize *= nDevices;
+  AudioDeviceID deviceList[ nDevices ];
+  property.mSelector = kAudioHardwarePropertyDevices;
+  result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, (void *) &deviceList );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDefaultInputDevice: OS-X system error getting device IDs.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  for ( unsigned int i=0; i<nDevices; i++ )
+    if ( id == deviceList[i] ) return i;
+
+  errorText_ = "RtApiCore::getDefaultInputDevice: No default device found!";
+  error( RtAudioError::WARNING );
+  return 0;
+}
+
+unsigned int RtApiCore :: getDefaultOutputDevice( void )
+{
+  unsigned int nDevices = getDeviceCount();
+  if ( nDevices <= 1 ) return 0;
+
+  AudioDeviceID id;
+  UInt32 dataSize = sizeof( AudioDeviceID );
+  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDefaultOutputDevice, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, &id );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDefaultOutputDevice: OS-X system error getting device.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  dataSize = sizeof( AudioDeviceID ) * nDevices;
+  AudioDeviceID deviceList[ nDevices ];
+  property.mSelector = kAudioHardwarePropertyDevices;
+  result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property, 0, NULL, &dataSize, (void *) &deviceList );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDefaultOutputDevice: OS-X system error getting device IDs.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  for ( unsigned int i=0; i<nDevices; i++ )
+    if ( id == deviceList[i] ) return i;
+
+  errorText_ = "RtApiCore::getDefaultOutputDevice: No default device found!";
+  error( RtAudioError::WARNING );
+  return 0;
+}
+
+RtAudio::DeviceInfo RtApiCore :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  // Get device ID
+  unsigned int nDevices = getDeviceCount();
+  if ( nDevices == 0 ) {
+    errorText_ = "RtApiCore::getDeviceInfo: no devices found!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  if ( device >= nDevices ) {
+    errorText_ = "RtApiCore::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  AudioDeviceID deviceList[ nDevices ];
+  UInt32 dataSize = sizeof( AudioDeviceID ) * nDevices;
+  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,
+                                          kAudioObjectPropertyScopeGlobal,
+                                          kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property,
+                                                0, NULL, &dataSize, (void *) &deviceList );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::getDeviceInfo: OS-X system error getting device IDs.";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  AudioDeviceID id = deviceList[ device ];
+
+  // Get the device name.
+  info.name.erase();
+  CFStringRef cfname;
+  dataSize = sizeof( CFStringRef );
+  property.mSelector = kAudioObjectPropertyManufacturer;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &cfname );
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceInfo: system error (" << getErrorCode( result ) << ") getting device manufacturer.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  //const char *mname = CFStringGetCStringPtr( cfname, CFStringGetSystemEncoding() );
+  int length = CFStringGetLength(cfname);
+  char *mname = (char *)malloc(length * 3 + 1);
+#if defined( UNICODE ) || defined( _UNICODE )
+  CFStringGetCString(cfname, mname, length * 3 + 1, kCFStringEncodingUTF8);
+#else
+  CFStringGetCString(cfname, mname, length * 3 + 1, CFStringGetSystemEncoding());
+#endif
+  info.name.append( (const char *)mname, strlen(mname) );
+  info.name.append( ": " );
+  CFRelease( cfname );
+  free(mname);
+
+  property.mSelector = kAudioObjectPropertyName;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &cfname );
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceInfo: system error (" << getErrorCode( result ) << ") getting device name.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  //const char *name = CFStringGetCStringPtr( cfname, CFStringGetSystemEncoding() );
+  length = CFStringGetLength(cfname);
+  char *name = (char *)malloc(length * 3 + 1);
+#if defined( UNICODE ) || defined( _UNICODE )
+  CFStringGetCString(cfname, name, length * 3 + 1, kCFStringEncodingUTF8);
+#else
+  CFStringGetCString(cfname, name, length * 3 + 1, CFStringGetSystemEncoding());
+#endif
+  info.name.append( (const char *)name, strlen(name) );
+  CFRelease( cfname );
+  free(name);
+
+  // Get the output stream "configuration".
+  AudioBufferList	*bufferList = nil;
+  property.mSelector = kAudioDevicePropertyStreamConfiguration;
+  property.mScope = kAudioDevicePropertyScopeOutput;
+  //  property.mElement = kAudioObjectPropertyElementWildcard;
+  dataSize = 0;
+  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );
+  if ( result != noErr || dataSize == 0 ) {
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting output stream configuration info for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Allocate the AudioBufferList.
+  bufferList = (AudioBufferList *) malloc( dataSize );
+  if ( bufferList == NULL ) {
+    errorText_ = "RtApiCore::getDeviceInfo: memory error allocating output AudioBufferList.";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );
+  if ( result != noErr || dataSize == 0 ) {
+    free( bufferList );
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting output stream configuration for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Get output channel information.
+  unsigned int i, nStreams = bufferList->mNumberBuffers;
+  for ( i=0; i<nStreams; i++ )
+    info.outputChannels += bufferList->mBuffers[i].mNumberChannels;
+  free( bufferList );
+
+  // Get the input stream "configuration".
+  property.mScope = kAudioDevicePropertyScopeInput;
+  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );
+  if ( result != noErr || dataSize == 0 ) {
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting input stream configuration info for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Allocate the AudioBufferList.
+  bufferList = (AudioBufferList *) malloc( dataSize );
+  if ( bufferList == NULL ) {
+    errorText_ = "RtApiCore::getDeviceInfo: memory error allocating input AudioBufferList.";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );
+  if (result != noErr || dataSize == 0) {
+    free( bufferList );
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting input stream configuration for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Get input channel information.
+  nStreams = bufferList->mNumberBuffers;
+  for ( i=0; i<nStreams; i++ )
+    info.inputChannels += bufferList->mBuffers[i].mNumberChannels;
+  free( bufferList );
+
+  // If device opens for both playback and capture, we determine the channels.
+  if ( info.outputChannels > 0 && info.inputChannels > 0 )
+    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+
+  // Probe the device sample rates.
+  bool isInput = false;
+  if ( info.outputChannels == 0 ) isInput = true;
+
+  // Determine the supported sample rates.
+  property.mSelector = kAudioDevicePropertyAvailableNominalSampleRates;
+  if ( isInput == false ) property.mScope = kAudioDevicePropertyScopeOutput;
+  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );
+  if ( result != kAudioHardwareNoError || dataSize == 0 ) {
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting sample rate info.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  UInt32 nRanges = dataSize / sizeof( AudioValueRange );
+  AudioValueRange rangeList[ nRanges ];
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &rangeList );
+  if ( result != kAudioHardwareNoError ) {
+    errorStream_ << "RtApiCore::getDeviceInfo: system error (" << getErrorCode( result ) << ") getting sample rates.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // The sample rate reporting mechanism is a bit of a mystery.  It
+  // seems that it can either return individual rates or a range of
+  // rates.  I assume that if the min / max range values are the same,
+  // then that represents a single supported rate and if the min / max
+  // range values are different, the device supports an arbitrary
+  // range of values (though there might be multiple ranges, so we'll
+  // use the most conservative range).
+  Float64 minimumRate = 1.0, maximumRate = 10000000000.0;
+  bool haveValueRange = false;
+  info.sampleRates.clear();
+  for ( UInt32 i=0; i<nRanges; i++ ) {
+    if ( rangeList[i].mMinimum == rangeList[i].mMaximum ) {
+      unsigned int tmpSr = (unsigned int) rangeList[i].mMinimum;
+      info.sampleRates.push_back( tmpSr );
+
+      if ( !info.preferredSampleRate || ( tmpSr <= 48000 && tmpSr > info.preferredSampleRate ) )
+        info.preferredSampleRate = tmpSr;
+
+    } else {
+      haveValueRange = true;
+      if ( rangeList[i].mMinimum > minimumRate ) minimumRate = rangeList[i].mMinimum;
+      if ( rangeList[i].mMaximum < maximumRate ) maximumRate = rangeList[i].mMaximum;
+    }
+  }
+
+  if ( haveValueRange ) {
+    for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {
+      if ( SAMPLE_RATES[k] >= (unsigned int) minimumRate && SAMPLE_RATES[k] <= (unsigned int) maximumRate ) {
+        info.sampleRates.push_back( SAMPLE_RATES[k] );
+
+        if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )
+          info.preferredSampleRate = SAMPLE_RATES[k];
+      }
+    }
+  }
+
+  // Sort and remove any redundant values
+  std::sort( info.sampleRates.begin(), info.sampleRates.end() );
+  info.sampleRates.erase( unique( info.sampleRates.begin(), info.sampleRates.end() ), info.sampleRates.end() );
+
+  if ( info.sampleRates.size() == 0 ) {
+    errorStream_ << "RtApiCore::probeDeviceInfo: No supported sample rates found for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // CoreAudio always uses 32-bit floating point data for PCM streams.
+  // Thus, any other "physical" formats supported by the device are of
+  // no interest to the client.
+  info.nativeFormats = RTAUDIO_FLOAT32;
+
+  if ( info.outputChannels > 0 )
+    if ( getDefaultOutputDevice() == device ) info.isDefaultOutput = true;
+  if ( info.inputChannels > 0 )
+    if ( getDefaultInputDevice() == device ) info.isDefaultInput = true;
+
+  info.probed = true;
+  return info;
+}
+
+static OSStatus callbackHandler( AudioDeviceID inDevice,
+                                 const AudioTimeStamp* /*inNow*/,
+                                 const AudioBufferList* inInputData,
+                                 const AudioTimeStamp* /*inInputTime*/,
+                                 AudioBufferList* outOutputData,
+                                 const AudioTimeStamp* /*inOutputTime*/,
+                                 void* infoPointer )
+{
+  CallbackInfo *info = (CallbackInfo *) infoPointer;
+
+  RtApiCore *object = (RtApiCore *) info->object;
+  if ( object->callbackEvent( inDevice, inInputData, outOutputData ) == false )
+    return kAudioHardwareUnspecifiedError;
+  else
+    return kAudioHardwareNoError;
+}
+
+static OSStatus xrunListener( AudioObjectID /*inDevice*/,
+                              UInt32 nAddresses,
+                              const AudioObjectPropertyAddress properties[],
+                              void* handlePointer )
+{
+  CoreHandle *handle = (CoreHandle *) handlePointer;
+  for ( UInt32 i=0; i<nAddresses; i++ ) {
+    if ( properties[i].mSelector == kAudioDeviceProcessorOverload ) {
+      if ( properties[i].mScope == kAudioDevicePropertyScopeInput )
+        handle->xrun[1] = true;
+      else
+        handle->xrun[0] = true;
+    }
+  }
+
+  return kAudioHardwareNoError;
+}
+
+static OSStatus rateListener( AudioObjectID inDevice,
+                              UInt32 /*nAddresses*/,
+                              const AudioObjectPropertyAddress /*properties*/[],
+                              void* ratePointer )
+{
+  Float64 *rate = (Float64 *) ratePointer;
+  UInt32 dataSize = sizeof( Float64 );
+  AudioObjectPropertyAddress property = { kAudioDevicePropertyNominalSampleRate,
+                                          kAudioObjectPropertyScopeGlobal,
+                                          kAudioObjectPropertyElementMaster };
+  AudioObjectGetPropertyData( inDevice, &property, 0, NULL, &dataSize, rate );
+  return kAudioHardwareNoError;
+}
+
+bool RtApiCore :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                   unsigned int firstChannel, unsigned int sampleRate,
+                                   RtAudioFormat format, unsigned int *bufferSize,
+                                   RtAudio::StreamOptions *options )
+{
+  // Get device ID
+  unsigned int nDevices = getDeviceCount();
+  if ( nDevices == 0 ) {
+    // This should not happen because a check is made before this function is called.
+    errorText_ = "RtApiCore::probeDeviceOpen: no devices found!";
+    return FAILURE;
+  }
+
+  if ( device >= nDevices ) {
+    // This should not happen because a check is made before this function is called.
+    errorText_ = "RtApiCore::probeDeviceOpen: device ID is invalid!";
+    return FAILURE;
+  }
+
+  AudioDeviceID deviceList[ nDevices ];
+  UInt32 dataSize = sizeof( AudioDeviceID ) * nDevices;
+  AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,
+                                          kAudioObjectPropertyScopeGlobal,
+                                          kAudioObjectPropertyElementMaster };
+  OSStatus result = AudioObjectGetPropertyData( kAudioObjectSystemObject, &property,
+                                                0, NULL, &dataSize, (void *) &deviceList );
+  if ( result != noErr ) {
+    errorText_ = "RtApiCore::probeDeviceOpen: OS-X system error getting device IDs.";
+    return FAILURE;
+  }
+
+  AudioDeviceID id = deviceList[ device ];
+
+  // Setup for stream mode.
+  bool isInput = false;
+  if ( mode == INPUT ) {
+    isInput = true;
+    property.mScope = kAudioDevicePropertyScopeInput;
+  }
+  else
+    property.mScope = kAudioDevicePropertyScopeOutput;
+
+  // Get the stream "configuration".
+  AudioBufferList	*bufferList = nil;
+  dataSize = 0;
+  property.mSelector = kAudioDevicePropertyStreamConfiguration;
+  result = AudioObjectGetPropertyDataSize( id, &property, 0, NULL, &dataSize );
+  if ( result != noErr || dataSize == 0 ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream configuration info for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Allocate the AudioBufferList.
+  bufferList = (AudioBufferList *) malloc( dataSize );
+  if ( bufferList == NULL ) {
+    errorText_ = "RtApiCore::probeDeviceOpen: memory error allocating AudioBufferList.";
+    return FAILURE;
+  }
+
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, bufferList );
+  if (result != noErr || dataSize == 0) {
+    free( bufferList );
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream configuration for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Search for one or more streams that contain the desired number of
+  // channels. CoreAudio devices can have an arbitrary number of
+  // streams and each stream can have an arbitrary number of channels.
+  // For each stream, a single buffer of interleaved samples is
+  // provided.  RtAudio prefers the use of one stream of interleaved
+  // data or multiple consecutive single-channel streams.  However, we
+  // now support multiple consecutive multi-channel streams of
+  // interleaved data as well.
+  UInt32 iStream, offsetCounter = firstChannel;
+  UInt32 nStreams = bufferList->mNumberBuffers;
+  bool monoMode = false;
+  bool foundStream = false;
+
+  // First check that the device supports the requested number of
+  // channels.
+  UInt32 deviceChannels = 0;
+  for ( iStream=0; iStream<nStreams; iStream++ )
+    deviceChannels += bufferList->mBuffers[iStream].mNumberChannels;
+
+  if ( deviceChannels < ( channels + firstChannel ) ) {
+    free( bufferList );
+    errorStream_ << "RtApiCore::probeDeviceOpen: the device (" << device << ") does not support the requested channel count.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Look for a single stream meeting our needs.
+  UInt32 firstStream, streamCount = 1, streamChannels = 0, channelOffset = 0;
+  for ( iStream=0; iStream<nStreams; iStream++ ) {
+    streamChannels = bufferList->mBuffers[iStream].mNumberChannels;
+    if ( streamChannels >= channels + offsetCounter ) {
+      firstStream = iStream;
+      channelOffset = offsetCounter;
+      foundStream = true;
+      break;
+    }
+    if ( streamChannels > offsetCounter ) break;
+    offsetCounter -= streamChannels;
+  }
+
+  // If we didn't find a single stream above, then we should be able
+  // to meet the channel specification with multiple streams.
+  if ( foundStream == false ) {
+    monoMode = true;
+    offsetCounter = firstChannel;
+    for ( iStream=0; iStream<nStreams; iStream++ ) {
+      streamChannels = bufferList->mBuffers[iStream].mNumberChannels;
+      if ( streamChannels > offsetCounter ) break;
+      offsetCounter -= streamChannels;
+    }
+
+    firstStream = iStream;
+    channelOffset = offsetCounter;
+    Int32 channelCounter = channels + offsetCounter - streamChannels;
+
+    if ( streamChannels > 1 ) monoMode = false;
+    while ( channelCounter > 0 ) {
+      streamChannels = bufferList->mBuffers[++iStream].mNumberChannels;
+      if ( streamChannels > 1 ) monoMode = false;
+      channelCounter -= streamChannels;
+      streamCount++;
+    }
+  }
+
+  free( bufferList );
+
+  // Determine the buffer size.
+  AudioValueRange	bufferRange;
+  dataSize = sizeof( AudioValueRange );
+  property.mSelector = kAudioDevicePropertyBufferFrameSizeRange;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &bufferRange );
+
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting buffer size range for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  if ( bufferRange.mMinimum > *bufferSize ) *bufferSize = (unsigned long) bufferRange.mMinimum;
+  else if ( bufferRange.mMaximum < *bufferSize ) *bufferSize = (unsigned long) bufferRange.mMaximum;
+  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) *bufferSize = (unsigned long) bufferRange.mMinimum;
+
+  // Set the buffer size.  For multiple streams, I'm assuming we only
+  // need to make this setting for the master channel.
+  UInt32 theSize = (UInt32) *bufferSize;
+  dataSize = sizeof( UInt32 );
+  property.mSelector = kAudioDevicePropertyBufferFrameSize;
+  result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &theSize );
+
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting the buffer size for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // If attempting to setup a duplex stream, the bufferSize parameter
+  // MUST be the same in both directions!
+  *bufferSize = theSize;
+  if ( stream_.mode == OUTPUT && mode == INPUT && *bufferSize != stream_.bufferSize ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error setting buffer size for duplex stream on device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  stream_.bufferSize = *bufferSize;
+  stream_.nBuffers = 1;
+
+  // Try to set "hog" mode ... it's not clear to me this is working.
+  if ( options && options->flags & RTAUDIO_HOG_DEVICE ) {
+    pid_t hog_pid;
+    dataSize = sizeof( hog_pid );
+    property.mSelector = kAudioDevicePropertyHogMode;
+    result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &hog_pid );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting 'hog' state!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    if ( hog_pid != getpid() ) {
+      hog_pid = getpid();
+      result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &hog_pid );
+      if ( result != noErr ) {
+        errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting 'hog' state!";
+        errorText_ = errorStream_.str();
+        return FAILURE;
+      }
+    }
+  }
+
+  // Check and if necessary, change the sample rate for the device.
+  Float64 nominalRate;
+  dataSize = sizeof( Float64 );
+  property.mSelector = kAudioDevicePropertyNominalSampleRate;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &nominalRate );
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting current sample rate.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Only change the sample rate if off by more than 1 Hz.
+  if ( fabs( nominalRate - (double)sampleRate ) > 1.0 ) {
+
+    // Set a property listener for the sample rate change
+    Float64 reportedRate = 0.0;
+    AudioObjectPropertyAddress tmp = { kAudioDevicePropertyNominalSampleRate, kAudioObjectPropertyScopeGlobal, kAudioObjectPropertyElementMaster };
+    result = AudioObjectAddPropertyListener( id, &tmp, rateListener, (void *) &reportedRate );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate property listener for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    nominalRate = (Float64) sampleRate;
+    result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &nominalRate );
+    if ( result != noErr ) {
+      AudioObjectRemovePropertyListener( id, &tmp, rateListener, (void *) &reportedRate );
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Now wait until the reported nominal rate is what we just set.
+    UInt32 microCounter = 0;
+    while ( reportedRate != nominalRate ) {
+      microCounter += 5000;
+      if ( microCounter > 5000000 ) break;
+      usleep( 5000 );
+    }
+
+    // Remove the property listener.
+    AudioObjectRemovePropertyListener( id, &tmp, rateListener, (void *) &reportedRate );
+
+    if ( microCounter > 5000000 ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: timeout waiting for sample rate update for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+
+  // Now set the stream format for all streams.  Also, check the
+  // physical format of the device and change that if necessary.
+  AudioStreamBasicDescription	description;
+  dataSize = sizeof( AudioStreamBasicDescription );
+  property.mSelector = kAudioStreamPropertyVirtualFormat;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &description );
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream format for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Set the sample rate and data format id.  However, only make the
+  // change if the sample rate is not within 1.0 of the desired
+  // rate and the format is not linear pcm.
+  bool updateFormat = false;
+  if ( fabs( description.mSampleRate - (Float64)sampleRate ) > 1.0 ) {
+    description.mSampleRate = (Float64) sampleRate;
+    updateFormat = true;
+  }
+
+  if ( description.mFormatID != kAudioFormatLinearPCM ) {
+    description.mFormatID = kAudioFormatLinearPCM;
+    updateFormat = true;
+  }
+
+  if ( updateFormat ) {
+    result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &description );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting sample rate or data format for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+
+  // Now check the physical format.
+  property.mSelector = kAudioStreamPropertyPhysicalFormat;
+  result = AudioObjectGetPropertyData( id, &property, 0, NULL,  &dataSize, &description );
+  if ( result != noErr ) {
+    errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting stream physical format for device (" << device << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  //std::cout << "Current physical stream format:" << std::endl;
+  //std::cout << "   mBitsPerChan = " << description.mBitsPerChannel << std::endl;
+  //std::cout << "   aligned high = " << (description.mFormatFlags & kAudioFormatFlagIsAlignedHigh) << ", isPacked = " << (description.mFormatFlags & kAudioFormatFlagIsPacked) << std::endl;
+  //std::cout << "   bytesPerFrame = " << description.mBytesPerFrame << std::endl;
+  //std::cout << "   sample rate = " << description.mSampleRate << std::endl;
+
+  if ( description.mFormatID != kAudioFormatLinearPCM || description.mBitsPerChannel < 16 ) {
+    description.mFormatID = kAudioFormatLinearPCM;
+    //description.mSampleRate = (Float64) sampleRate;
+    AudioStreamBasicDescription	testDescription = description;
+    UInt32 formatFlags;
+
+    // We'll try higher bit rates first and then work our way down.
+    std::vector< std::pair<UInt32, UInt32>  > physicalFormats;
+    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsFloat) & ~kLinearPCMFormatFlagIsSignedInteger;
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 32, formatFlags ) );
+    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked) & ~kLinearPCMFormatFlagIsFloat;
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 32, formatFlags ) );
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 24, formatFlags ) );   // 24-bit packed
+    formatFlags &= ~( kAudioFormatFlagIsPacked | kAudioFormatFlagIsAlignedHigh );
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 24.2, formatFlags ) ); // 24-bit in 4 bytes, aligned low
+    formatFlags |= kAudioFormatFlagIsAlignedHigh;
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 24.4, formatFlags ) ); // 24-bit in 4 bytes, aligned high
+    formatFlags = (description.mFormatFlags | kLinearPCMFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked) & ~kLinearPCMFormatFlagIsFloat;
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 16, formatFlags ) );
+    physicalFormats.push_back( std::pair<Float32, UInt32>( 8, formatFlags ) );
+
+    bool setPhysicalFormat = false;
+    for( unsigned int i=0; i<physicalFormats.size(); i++ ) {
+      testDescription = description;
+      testDescription.mBitsPerChannel = (UInt32) physicalFormats[i].first;
+      testDescription.mFormatFlags = physicalFormats[i].second;
+      if ( (24 == (UInt32)physicalFormats[i].first) && ~( physicalFormats[i].second & kAudioFormatFlagIsPacked ) )
+        testDescription.mBytesPerFrame =  4 * testDescription.mChannelsPerFrame;
+      else
+        testDescription.mBytesPerFrame =  testDescription.mBitsPerChannel/8 * testDescription.mChannelsPerFrame;
+      testDescription.mBytesPerPacket = testDescription.mBytesPerFrame * testDescription.mFramesPerPacket;
+      result = AudioObjectSetPropertyData( id, &property, 0, NULL, dataSize, &testDescription );
+      if ( result == noErr ) {
+        setPhysicalFormat = true;
+        //std::cout << "Updated physical stream format:" << std::endl;
+        //std::cout << "   mBitsPerChan = " << testDescription.mBitsPerChannel << std::endl;
+        //std::cout << "   aligned high = " << (testDescription.mFormatFlags & kAudioFormatFlagIsAlignedHigh) << ", isPacked = " << (testDescription.mFormatFlags & kAudioFormatFlagIsPacked) << std::endl;
+        //std::cout << "   bytesPerFrame = " << testDescription.mBytesPerFrame << std::endl;
+        //std::cout << "   sample rate = " << testDescription.mSampleRate << std::endl;
+        break;
+      }
+    }
+
+    if ( !setPhysicalFormat ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") setting physical data format for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  } // done setting virtual/physical formats.
+
+  // Get the stream / device latency.
+  UInt32 latency;
+  dataSize = sizeof( UInt32 );
+  property.mSelector = kAudioDevicePropertyLatency;
+  if ( AudioObjectHasProperty( id, &property ) == true ) {
+    result = AudioObjectGetPropertyData( id, &property, 0, NULL, &dataSize, &latency );
+    if ( result == kAudioHardwareNoError ) stream_.latency[ mode ] = latency;
+    else {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error (" << getErrorCode( result ) << ") getting device latency for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      error( RtAudioError::WARNING );
+    }
+  }
+
+  // Byte-swapping: According to AudioHardware.h, the stream data will
+  // always be presented in native-endian format, so we should never
+  // need to byte swap.
+  stream_.doByteSwap[mode] = false;
+
+  // From the CoreAudio documentation, PCM data must be supplied as
+  // 32-bit floats.
+  stream_.userFormat = format;
+  stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;
+
+  if ( streamCount == 1 )
+    stream_.nDeviceChannels[mode] = description.mChannelsPerFrame;
+  else // multiple streams
+    stream_.nDeviceChannels[mode] = channels;
+  stream_.nUserChannels[mode] = channels;
+  stream_.channelOffset[mode] = channelOffset;  // offset within a CoreAudio stream
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;
+  else stream_.userInterleaved = true;
+  stream_.deviceInterleaved[mode] = true;
+  if ( monoMode == true ) stream_.deviceInterleaved[mode] = false;
+
+  // Set flags for buffer conversion.
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( streamCount == 1 ) {
+    if ( stream_.nUserChannels[mode] > 1 &&
+         stream_.userInterleaved != stream_.deviceInterleaved[mode] )
+      stream_.doConvertBuffer[mode] = true;
+  }
+  else if ( monoMode && stream_.userInterleaved )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate our CoreHandle structure for the stream.
+  CoreHandle *handle = 0;
+  if ( stream_.apiHandle == 0 ) {
+    try {
+      handle = new CoreHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiCore::probeDeviceOpen: error allocating CoreHandle memory.";
+      goto error;
+    }
+
+    if ( pthread_cond_init( &handle->condition, NULL ) ) {
+      errorText_ = "RtApiCore::probeDeviceOpen: error initializing pthread condition variable.";
+      goto error;
+    }
+    stream_.apiHandle = (void *) handle;
+  }
+  else
+    handle = (CoreHandle *) stream_.apiHandle;
+  handle->iStream[mode] = firstStream;
+  handle->nStreams[mode] = streamCount;
+  handle->id[mode] = id;
+
+  // Allocate necessary internal buffers.
+  unsigned long bufferBytes;
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  //  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  stream_.userBuffer[mode] = (char *) malloc( bufferBytes * sizeof(char) );
+  memset( stream_.userBuffer[mode], 0, bufferBytes * sizeof(char) );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiCore::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  // If possible, we will make use of the CoreAudio stream buffers as
+  // "device buffers".  However, we can't do this if using multiple
+  // streams.
+  if ( stream_.doConvertBuffer[mode] && handle->nStreams[mode] > 1 ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( mode == INPUT ) {
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+        if ( bufferBytes <= bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiCore::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  stream_.sampleRate = sampleRate;
+  stream_.device[mode] = device;
+  stream_.state = STREAM_STOPPED;
+  stream_.callbackInfo.object = (void *) this;
+
+  // Setup the buffer conversion information structure.
+  if ( stream_.doConvertBuffer[mode] ) {
+    if ( streamCount > 1 ) setConvertInfo( mode, 0 );
+    else setConvertInfo( mode, channelOffset );
+  }
+
+  if ( mode == INPUT && stream_.mode == OUTPUT && stream_.device[0] == device )
+    // Only one callback procedure per device.
+    stream_.mode = DUPLEX;
+  else {
+#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )
+    result = AudioDeviceCreateIOProcID( id, callbackHandler, (void *) &stream_.callbackInfo, &handle->procId[mode] );
+#else
+    // deprecated in favor of AudioDeviceCreateIOProcID()
+    result = AudioDeviceAddIOProc( id, callbackHandler, (void *) &stream_.callbackInfo );
+#endif
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::probeDeviceOpen: system error setting callback for device (" << device << ").";
+      errorText_ = errorStream_.str();
+      goto error;
+    }
+    if ( stream_.mode == OUTPUT && mode == INPUT )
+      stream_.mode = DUPLEX;
+    else
+      stream_.mode = mode;
+  }
+
+  // Setup the device property listener for over/underload.
+  property.mSelector = kAudioDeviceProcessorOverload;
+  property.mScope = kAudioObjectPropertyScopeGlobal;
+  result = AudioObjectAddPropertyListener( id, &property, xrunListener, (void *) handle );
+
+  return SUCCESS;
+
+ error:
+  if ( handle ) {
+    pthread_cond_destroy( &handle->condition );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.state = STREAM_CLOSED;
+  return FAILURE;
+}
+
+void RtApiCore :: closeStream( void )
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiCore::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    if (handle) {
+      AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,
+        kAudioObjectPropertyScopeGlobal,
+        kAudioObjectPropertyElementMaster };
+
+      property.mSelector = kAudioDeviceProcessorOverload;
+      property.mScope = kAudioObjectPropertyScopeGlobal;
+      if (AudioObjectRemovePropertyListener( handle->id[0], &property, xrunListener, (void *) handle ) != noErr) {
+        errorText_ = "RtApiCore::closeStream(): error removing property listener!";
+        error( RtAudioError::WARNING );
+      }
+    }
+    if ( stream_.state == STREAM_RUNNING )
+      AudioDeviceStop( handle->id[0], callbackHandler );
+#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )
+    AudioDeviceDestroyIOProcID( handle->id[0], handle->procId[0] );
+#else
+    // deprecated in favor of AudioDeviceDestroyIOProcID()
+    AudioDeviceRemoveIOProc( handle->id[0], callbackHandler );
+#endif
+  }
+
+  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {
+    if (handle) {
+      AudioObjectPropertyAddress property = { kAudioHardwarePropertyDevices,
+        kAudioObjectPropertyScopeGlobal,
+        kAudioObjectPropertyElementMaster };
+
+      property.mSelector = kAudioDeviceProcessorOverload;
+      property.mScope = kAudioObjectPropertyScopeGlobal;
+      if (AudioObjectRemovePropertyListener( handle->id[1], &property, xrunListener, (void *) handle ) != noErr) {
+        errorText_ = "RtApiCore::closeStream(): error removing property listener!";
+        error( RtAudioError::WARNING );
+      }
+    }
+    if ( stream_.state == STREAM_RUNNING )
+      AudioDeviceStop( handle->id[1], callbackHandler );
+#if defined( MAC_OS_X_VERSION_10_5 ) && ( MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_5 )
+    AudioDeviceDestroyIOProcID( handle->id[1], handle->procId[1] );
+#else
+    // deprecated in favor of AudioDeviceDestroyIOProcID()
+    AudioDeviceRemoveIOProc( handle->id[1], callbackHandler );
+#endif
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  // Destroy pthread condition variable.
+  pthread_cond_destroy( &handle->condition );
+  delete handle;
+  stream_.apiHandle = 0;
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+void RtApiCore :: startStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiCore::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  OSStatus result = noErr;
+  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    result = AudioDeviceStart( handle->id[0], callbackHandler );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::startStream: system error (" << getErrorCode( result ) << ") starting callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  if ( stream_.mode == INPUT ||
+       ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {
+
+    result = AudioDeviceStart( handle->id[1], callbackHandler );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::startStream: system error starting input callback procedure on device (" << stream_.device[1] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  handle->drainCounter = 0;
+  handle->internalDrain = false;
+  stream_.state = STREAM_RUNNING;
+
+ unlock:
+  if ( result == noErr ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiCore :: stopStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiCore::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  OSStatus result = noErr;
+  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    if ( handle->drainCounter == 0 ) {
+      handle->drainCounter = 2;
+      pthread_cond_wait( &handle->condition, &stream_.mutex ); // block until signaled
+    }
+
+    result = AudioDeviceStop( handle->id[0], callbackHandler );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::stopStream: system error (" << getErrorCode( result ) << ") stopping callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && stream_.device[0] != stream_.device[1] ) ) {
+
+    result = AudioDeviceStop( handle->id[1], callbackHandler );
+    if ( result != noErr ) {
+      errorStream_ << "RtApiCore::stopStream: system error (" << getErrorCode( result ) << ") stopping input callback procedure on device (" << stream_.device[1] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  stream_.state = STREAM_STOPPED;
+
+ unlock:
+  if ( result == noErr ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiCore :: abortStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiCore::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;
+  handle->drainCounter = 2;
+
+  stopStream();
+}
+
+// This function will be called by a spawned thread when the user
+// callback function signals that the stream should be stopped or
+// aborted.  It is better to handle it this way because the
+// callbackEvent() function probably should return before the AudioDeviceStop()
+// function is called.
+static void *coreStopStream( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiCore *object = (RtApiCore *) info->object;
+
+  object->stopStream();
+  pthread_exit( NULL );
+}
+
+bool RtApiCore :: callbackEvent( AudioDeviceID deviceId,
+                                 const AudioBufferList *inBufferList,
+                                 const AudioBufferList *outBufferList )
+{
+  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiCore::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return FAILURE;
+  }
+
+  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;
+  CoreHandle *handle = (CoreHandle *) stream_.apiHandle;
+
+  // Check if we were draining the stream and signal is finished.
+  if ( handle->drainCounter > 3 ) {
+    ThreadHandle threadId;
+
+    stream_.state = STREAM_STOPPING;
+    if ( handle->internalDrain == true )
+      pthread_create( &threadId, NULL, coreStopStream, info );
+    else // external call to stopStream()
+      pthread_cond_signal( &handle->condition );
+    return SUCCESS;
+  }
+
+  AudioDeviceID outputDevice = handle->id[0];
+
+  // Invoke user callback to get fresh output data UNLESS we are
+  // draining stream or duplex mode AND the input/output devices are
+  // different AND this function is called for the input device.
+  if ( handle->drainCounter == 0 && ( stream_.mode != DUPLEX || deviceId == outputDevice ) ) {
+    RtAudioCallback callback = (RtAudioCallback) info->callback;
+    double streamTime = getStreamTime();
+    RtAudioStreamStatus status = 0;
+    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {
+      status |= RTAUDIO_OUTPUT_UNDERFLOW;
+      handle->xrun[0] = false;
+    }
+    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {
+      status |= RTAUDIO_INPUT_OVERFLOW;
+      handle->xrun[1] = false;
+    }
+
+    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                                  stream_.bufferSize, streamTime, status, info->userData );
+    if ( cbReturnValue == 2 ) {
+      stream_.state = STREAM_STOPPING;
+      handle->drainCounter = 2;
+      abortStream();
+      return SUCCESS;
+    }
+    else if ( cbReturnValue == 1 ) {
+      handle->drainCounter = 1;
+      handle->internalDrain = true;
+    }
+  }
+
+  if ( stream_.mode == OUTPUT || ( stream_.mode == DUPLEX && deviceId == outputDevice ) ) {
+
+    if ( handle->drainCounter > 1 ) { // write zeros to the output stream
+
+      if ( handle->nStreams[0] == 1 ) {
+        memset( outBufferList->mBuffers[handle->iStream[0]].mData,
+                0,
+                outBufferList->mBuffers[handle->iStream[0]].mDataByteSize );
+      }
+      else { // fill multiple streams with zeros
+        for ( unsigned int i=0; i<handle->nStreams[0]; i++ ) {
+          memset( outBufferList->mBuffers[handle->iStream[0]+i].mData,
+                  0,
+                  outBufferList->mBuffers[handle->iStream[0]+i].mDataByteSize );
+        }
+      }
+    }
+    else if ( handle->nStreams[0] == 1 ) {
+      if ( stream_.doConvertBuffer[0] ) { // convert directly to CoreAudio stream buffer
+        convertBuffer( (char *) outBufferList->mBuffers[handle->iStream[0]].mData,
+                       stream_.userBuffer[0], stream_.convertInfo[0] );
+      }
+      else { // copy from user buffer
+        memcpy( outBufferList->mBuffers[handle->iStream[0]].mData,
+                stream_.userBuffer[0],
+                outBufferList->mBuffers[handle->iStream[0]].mDataByteSize );
+      }
+    }
+    else { // fill multiple streams
+      Float32 *inBuffer = (Float32 *) stream_.userBuffer[0];
+      if ( stream_.doConvertBuffer[0] ) {
+        convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+        inBuffer = (Float32 *) stream_.deviceBuffer;
+      }
+
+      if ( stream_.deviceInterleaved[0] == false ) { // mono mode
+        UInt32 bufferBytes = outBufferList->mBuffers[handle->iStream[0]].mDataByteSize;
+        for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {
+          memcpy( outBufferList->mBuffers[handle->iStream[0]+i].mData,
+                  (void *)&inBuffer[i*stream_.bufferSize], bufferBytes );
+        }
+      }
+      else { // fill multiple multi-channel streams with interleaved data
+        UInt32 streamChannels, channelsLeft, inJump, outJump, inOffset;
+        Float32 *out, *in;
+
+        bool inInterleaved = ( stream_.userInterleaved ) ? true : false;
+        UInt32 inChannels = stream_.nUserChannels[0];
+        if ( stream_.doConvertBuffer[0] ) {
+          inInterleaved = true; // device buffer will always be interleaved for nStreams > 1 and not mono mode
+          inChannels = stream_.nDeviceChannels[0];
+        }
+
+        if ( inInterleaved ) inOffset = 1;
+        else inOffset = stream_.bufferSize;
+
+        channelsLeft = inChannels;
+        for ( unsigned int i=0; i<handle->nStreams[0]; i++ ) {
+          in = inBuffer;
+          out = (Float32 *) outBufferList->mBuffers[handle->iStream[0]+i].mData;
+          streamChannels = outBufferList->mBuffers[handle->iStream[0]+i].mNumberChannels;
+
+          outJump = 0;
+          // Account for possible channel offset in first stream
+          if ( i == 0 && stream_.channelOffset[0] > 0 ) {
+            streamChannels -= stream_.channelOffset[0];
+            outJump = stream_.channelOffset[0];
+            out += outJump;
+          }
+
+          // Account for possible unfilled channels at end of the last stream
+          if ( streamChannels > channelsLeft ) {
+            outJump = streamChannels - channelsLeft;
+            streamChannels = channelsLeft;
+          }
+
+          // Determine input buffer offsets and skips
+          if ( inInterleaved ) {
+            inJump = inChannels;
+            in += inChannels - channelsLeft;
+          }
+          else {
+            inJump = 1;
+            in += (inChannels - channelsLeft) * inOffset;
+          }
+
+          for ( unsigned int i=0; i<stream_.bufferSize; i++ ) {
+            for ( unsigned int j=0; j<streamChannels; j++ ) {
+              *out++ = in[j*inOffset];
+            }
+            out += outJump;
+            in += inJump;
+          }
+          channelsLeft -= streamChannels;
+        }
+      }
+    }
+  }
+
+  // Don't bother draining input
+  if ( handle->drainCounter ) {
+    handle->drainCounter++;
+    goto unlock;
+  }
+
+  AudioDeviceID inputDevice;
+  inputDevice = handle->id[1];
+  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && deviceId == inputDevice ) ) {
+
+    if ( handle->nStreams[1] == 1 ) {
+      if ( stream_.doConvertBuffer[1] ) { // convert directly from CoreAudio stream buffer
+        convertBuffer( stream_.userBuffer[1],
+                       (char *) inBufferList->mBuffers[handle->iStream[1]].mData,
+                       stream_.convertInfo[1] );
+      }
+      else { // copy to user buffer
+        memcpy( stream_.userBuffer[1],
+                inBufferList->mBuffers[handle->iStream[1]].mData,
+                inBufferList->mBuffers[handle->iStream[1]].mDataByteSize );
+      }
+    }
+    else { // read from multiple streams
+      Float32 *outBuffer = (Float32 *) stream_.userBuffer[1];
+      if ( stream_.doConvertBuffer[1] ) outBuffer = (Float32 *) stream_.deviceBuffer;
+
+      if ( stream_.deviceInterleaved[1] == false ) { // mono mode
+        UInt32 bufferBytes = inBufferList->mBuffers[handle->iStream[1]].mDataByteSize;
+        for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {
+          memcpy( (void *)&outBuffer[i*stream_.bufferSize],
+                  inBufferList->mBuffers[handle->iStream[1]+i].mData, bufferBytes );
+        }
+      }
+      else { // read from multiple multi-channel streams
+        UInt32 streamChannels, channelsLeft, inJump, outJump, outOffset;
+        Float32 *out, *in;
+
+        bool outInterleaved = ( stream_.userInterleaved ) ? true : false;
+        UInt32 outChannels = stream_.nUserChannels[1];
+        if ( stream_.doConvertBuffer[1] ) {
+          outInterleaved = true; // device buffer will always be interleaved for nStreams > 1 and not mono mode
+          outChannels = stream_.nDeviceChannels[1];
+        }
+
+        if ( outInterleaved ) outOffset = 1;
+        else outOffset = stream_.bufferSize;
+
+        channelsLeft = outChannels;
+        for ( unsigned int i=0; i<handle->nStreams[1]; i++ ) {
+          out = outBuffer;
+          in = (Float32 *) inBufferList->mBuffers[handle->iStream[1]+i].mData;
+          streamChannels = inBufferList->mBuffers[handle->iStream[1]+i].mNumberChannels;
+
+          inJump = 0;
+          // Account for possible channel offset in first stream
+          if ( i == 0 && stream_.channelOffset[1] > 0 ) {
+            streamChannels -= stream_.channelOffset[1];
+            inJump = stream_.channelOffset[1];
+            in += inJump;
+          }
+
+          // Account for possible unread channels at end of the last stream
+          if ( streamChannels > channelsLeft ) {
+            inJump = streamChannels - channelsLeft;
+            streamChannels = channelsLeft;
+          }
+
+          // Determine output buffer offsets and skips
+          if ( outInterleaved ) {
+            outJump = outChannels;
+            out += outChannels - channelsLeft;
+          }
+          else {
+            outJump = 1;
+            out += (outChannels - channelsLeft) * outOffset;
+          }
+
+          for ( unsigned int i=0; i<stream_.bufferSize; i++ ) {
+            for ( unsigned int j=0; j<streamChannels; j++ ) {
+              out[j*outOffset] = *in++;
+            }
+            out += outJump;
+            in += inJump;
+          }
+          channelsLeft -= streamChannels;
+        }
+      }
+      
+      if ( stream_.doConvertBuffer[1] ) { // convert from our internal "device" buffer
+        convertBuffer( stream_.userBuffer[1],
+                       stream_.deviceBuffer,
+                       stream_.convertInfo[1] );
+      }
+    }
+  }
+
+ unlock:
+  //MUTEX_UNLOCK( &stream_.mutex );
+
+  RtApi::tickStreamTime();
+  return SUCCESS;
+}
+
+const char* RtApiCore :: getErrorCode( OSStatus code )
+{
+  switch( code ) {
+
+  case kAudioHardwareNotRunningError:
+    return "kAudioHardwareNotRunningError";
+
+  case kAudioHardwareUnspecifiedError:
+    return "kAudioHardwareUnspecifiedError";
+
+  case kAudioHardwareUnknownPropertyError:
+    return "kAudioHardwareUnknownPropertyError";
+
+  case kAudioHardwareBadPropertySizeError:
+    return "kAudioHardwareBadPropertySizeError";
+
+  case kAudioHardwareIllegalOperationError:
+    return "kAudioHardwareIllegalOperationError";
+
+  case kAudioHardwareBadObjectError:
+    return "kAudioHardwareBadObjectError";
+
+  case kAudioHardwareBadDeviceError:
+    return "kAudioHardwareBadDeviceError";
+
+  case kAudioHardwareBadStreamError:
+    return "kAudioHardwareBadStreamError";
+
+  case kAudioHardwareUnsupportedOperationError:
+    return "kAudioHardwareUnsupportedOperationError";
+
+  case kAudioDeviceUnsupportedFormatError:
+    return "kAudioDeviceUnsupportedFormatError";
+
+  case kAudioDevicePermissionsError:
+    return "kAudioDevicePermissionsError";
+
+  default:
+    return "CoreAudio unknown error";
+  }
+}
+
+  //******************** End of __MACOSX_CORE__ *********************//
+#endif
+
+#if defined(__UNIX_JACK__)
+
+// JACK is a low-latency audio server, originally written for the
+// GNU/Linux operating system and now also ported to OS-X. It can
+// connect a number of different applications to an audio device, as
+// well as allowing them to share audio between themselves.
+//
+// When using JACK with RtAudio, "devices" refer to JACK clients that
+// have ports connected to the server.  The JACK server is typically
+// started in a terminal as follows:
+//
+// .jackd -d alsa -d hw:0
+//
+// or through an interface program such as qjackctl.  Many of the
+// parameters normally set for a stream are fixed by the JACK server
+// and can be specified when the JACK server is started.  In
+// particular,
+//
+// .jackd -d alsa -d hw:0 -r 44100 -p 512 -n 4
+//
+// specifies a sample rate of 44100 Hz, a buffer size of 512 sample
+// frames, and number of buffers = 4.  Once the server is running, it
+// is not possible to override these values.  If the values are not
+// specified in the command-line, the JACK server uses default values.
+//
+// The JACK server does not have to be running when an instance of
+// RtApiJack is created, though the function getDeviceCount() will
+// report 0 devices found until JACK has been started.  When no
+// devices are available (i.e., the JACK server is not running), a
+// stream cannot be opened.
+
+#include <jack/jack.h>
+#include <unistd.h>
+#include <cstdio>
+
+// A structure to hold various information related to the Jack API
+// implementation.
+struct JackHandle {
+  jack_client_t *client;
+  jack_port_t **ports[2];
+  std::string deviceName[2];
+  bool xrun[2];
+  pthread_cond_t condition;
+  int drainCounter;       // Tracks callback counts when draining
+  bool internalDrain;     // Indicates if stop is initiated from callback or not.
+
+  JackHandle()
+    :client(0), drainCounter(0), internalDrain(false) { ports[0] = 0; ports[1] = 0; xrun[0] = false; xrun[1] = false; }
+};
+
+static void jackSilentError( const char * ) {};
+
+RtApiJack :: RtApiJack()
+{
+  // Nothing to do here.
+#if !defined(__RTAUDIO_DEBUG__)
+  // Turn off Jack's internal error reporting.
+  jack_set_error_function( &jackSilentError );
+#endif
+}
+
+RtApiJack :: ~RtApiJack()
+{
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+}
+
+unsigned int RtApiJack :: getDeviceCount( void )
+{
+  // See if we can become a jack client.
+  jack_options_t options = (jack_options_t) ( JackNoStartServer ); //JackNullOption;
+  jack_status_t *status = NULL;
+  jack_client_t *client = jack_client_open( "RtApiJackCount", options, status );
+  if ( client == 0 ) return 0;
+
+  const char **ports;
+  std::string port, previousPort;
+  unsigned int nChannels = 0, nDevices = 0;
+  ports = jack_get_ports( client, NULL, NULL, 0 );
+  if ( ports ) {
+    // Parse the port names up to the first colon (:).
+    size_t iColon = 0;
+    do {
+      port = (char *) ports[ nChannels ];
+      iColon = port.find(":");
+      if ( iColon != std::string::npos ) {
+        port = port.substr( 0, iColon + 1 );
+        if ( port != previousPort ) {
+          nDevices++;
+          previousPort = port;
+        }
+      }
+    } while ( ports[++nChannels] );
+    free( ports );
+  }
+
+  jack_client_close( client );
+  return nDevices;
+}
+
+RtAudio::DeviceInfo RtApiJack :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  jack_options_t options = (jack_options_t) ( JackNoStartServer ); //JackNullOption
+  jack_status_t *status = NULL;
+  jack_client_t *client = jack_client_open( "RtApiJackInfo", options, status );
+  if ( client == 0 ) {
+    errorText_ = "RtApiJack::getDeviceInfo: Jack server not found or connection error!";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  const char **ports;
+  std::string port, previousPort;
+  unsigned int nPorts = 0, nDevices = 0;
+  ports = jack_get_ports( client, NULL, NULL, 0 );
+  if ( ports ) {
+    // Parse the port names up to the first colon (:).
+    size_t iColon = 0;
+    do {
+      port = (char *) ports[ nPorts ];
+      iColon = port.find(":");
+      if ( iColon != std::string::npos ) {
+        port = port.substr( 0, iColon );
+        if ( port != previousPort ) {
+          if ( nDevices == device ) info.name = port;
+          nDevices++;
+          previousPort = port;
+        }
+      }
+    } while ( ports[++nPorts] );
+    free( ports );
+  }
+
+  if ( device >= nDevices ) {
+    jack_client_close( client );
+    errorText_ = "RtApiJack::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  // Get the current jack server sample rate.
+  info.sampleRates.clear();
+
+  info.preferredSampleRate = jack_get_sample_rate( client );
+  info.sampleRates.push_back( info.preferredSampleRate );
+
+  // Count the available ports containing the client name as device
+  // channels.  Jack "input ports" equal RtAudio output channels.
+  unsigned int nChannels = 0;
+  ports = jack_get_ports( client, info.name.c_str(), NULL, JackPortIsInput );
+  if ( ports ) {
+    while ( ports[ nChannels ] ) nChannels++;
+    free( ports );
+    info.outputChannels = nChannels;
+  }
+
+  // Jack "output ports" equal RtAudio input channels.
+  nChannels = 0;
+  ports = jack_get_ports( client, info.name.c_str(), NULL, JackPortIsOutput );
+  if ( ports ) {
+    while ( ports[ nChannels ] ) nChannels++;
+    free( ports );
+    info.inputChannels = nChannels;
+  }
+
+  if ( info.outputChannels == 0 && info.inputChannels == 0 ) {
+    jack_client_close(client);
+    errorText_ = "RtApiJack::getDeviceInfo: error determining Jack input/output channels!";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // If device opens for both playback and capture, we determine the channels.
+  if ( info.outputChannels > 0 && info.inputChannels > 0 )
+    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+
+  // Jack always uses 32-bit floats.
+  info.nativeFormats = RTAUDIO_FLOAT32;
+
+  // Jack doesn't provide default devices so we'll use the first available one.
+  if ( device == 0 && info.outputChannels > 0 )
+    info.isDefaultOutput = true;
+  if ( device == 0 && info.inputChannels > 0 )
+    info.isDefaultInput = true;
+
+  jack_client_close(client);
+  info.probed = true;
+  return info;
+}
+
+static int jackCallbackHandler( jack_nframes_t nframes, void *infoPointer )
+{
+  CallbackInfo *info = (CallbackInfo *) infoPointer;
+
+  RtApiJack *object = (RtApiJack *) info->object;
+  if ( object->callbackEvent( (unsigned long) nframes ) == false ) return 1;
+
+  return 0;
+}
+
+// This function will be called by a spawned thread when the Jack
+// server signals that it is shutting down.  It is necessary to handle
+// it this way because the jackShutdown() function must return before
+// the jack_deactivate() function (in closeStream()) will return.
+static void *jackCloseStream( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiJack *object = (RtApiJack *) info->object;
+
+  object->closeStream();
+
+  pthread_exit( NULL );
+}
+static void jackShutdown( void *infoPointer )
+{
+  CallbackInfo *info = (CallbackInfo *) infoPointer;
+  RtApiJack *object = (RtApiJack *) info->object;
+
+  // Check current stream state.  If stopped, then we'll assume this
+  // was called as a result of a call to RtApiJack::stopStream (the
+  // deactivation of a client handle causes this function to be called).
+  // If not, we'll assume the Jack server is shutting down or some
+  // other problem occurred and we should close the stream.
+  if ( object->isStreamRunning() == false ) return;
+
+  ThreadHandle threadId;
+  pthread_create( &threadId, NULL, jackCloseStream, info );
+  std::cerr << "\nRtApiJack: the Jack server is shutting down this client ... stream stopped and closed!!\n" << std::endl;
+}
+
+static int jackXrun( void *infoPointer )
+{
+  JackHandle *handle = (JackHandle *) infoPointer;
+
+  if ( handle->ports[0] ) handle->xrun[0] = true;
+  if ( handle->ports[1] ) handle->xrun[1] = true;
+
+  return 0;
+}
+
+bool RtApiJack :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                   unsigned int firstChannel, unsigned int sampleRate,
+                                   RtAudioFormat format, unsigned int *bufferSize,
+                                   RtAudio::StreamOptions *options )
+{
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+
+  // Look for jack server and try to become a client (only do once per stream).
+  jack_client_t *client = 0;
+  if ( mode == OUTPUT || ( mode == INPUT && stream_.mode != OUTPUT ) ) {
+    jack_options_t jackoptions = (jack_options_t) ( JackNoStartServer ); //JackNullOption;
+    jack_status_t *status = NULL;
+    if ( options && !options->streamName.empty() )
+      client = jack_client_open( options->streamName.c_str(), jackoptions, status );
+    else
+      client = jack_client_open( "RtApiJack", jackoptions, status );
+    if ( client == 0 ) {
+      errorText_ = "RtApiJack::probeDeviceOpen: Jack server not found or connection error!";
+      error( RtAudioError::WARNING );
+      return FAILURE;
+    }
+  }
+  else {
+    // The handle must have been created on an earlier pass.
+    client = handle->client;
+  }
+
+  const char **ports;
+  std::string port, previousPort, deviceName;
+  unsigned int nPorts = 0, nDevices = 0;
+  ports = jack_get_ports( client, NULL, NULL, 0 );
+  if ( ports ) {
+    // Parse the port names up to the first colon (:).
+    size_t iColon = 0;
+    do {
+      port = (char *) ports[ nPorts ];
+      iColon = port.find(":");
+      if ( iColon != std::string::npos ) {
+        port = port.substr( 0, iColon );
+        if ( port != previousPort ) {
+          if ( nDevices == device ) deviceName = port;
+          nDevices++;
+          previousPort = port;
+        }
+      }
+    } while ( ports[++nPorts] );
+    free( ports );
+  }
+
+  if ( device >= nDevices ) {
+    errorText_ = "RtApiJack::probeDeviceOpen: device ID is invalid!";
+    return FAILURE;
+  }
+
+  // Count the available ports containing the client name as device
+  // channels.  Jack "input ports" equal RtAudio output channels.
+  unsigned int nChannels = 0;
+  unsigned long flag = JackPortIsInput;
+  if ( mode == INPUT ) flag = JackPortIsOutput;
+  ports = jack_get_ports( client, deviceName.c_str(), NULL, flag );
+  if ( ports ) {
+    while ( ports[ nChannels ] ) nChannels++;
+    free( ports );
+  }
+
+  // Compare the jack ports for specified client to the requested number of channels.
+  if ( nChannels < (channels + firstChannel) ) {
+    errorStream_ << "RtApiJack::probeDeviceOpen: requested number of channels (" << channels << ") + offset (" << firstChannel << ") not found for specified device (" << device << ":" << deviceName << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Check the jack server sample rate.
+  unsigned int jackRate = jack_get_sample_rate( client );
+  if ( sampleRate != jackRate ) {
+    jack_client_close( client );
+    errorStream_ << "RtApiJack::probeDeviceOpen: the requested sample rate (" << sampleRate << ") is different than the JACK server rate (" << jackRate << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  stream_.sampleRate = jackRate;
+
+  // Get the latency of the JACK port.
+  ports = jack_get_ports( client, deviceName.c_str(), NULL, flag );
+  if ( ports[ firstChannel ] ) {
+    // Added by Ge Wang
+    jack_latency_callback_mode_t cbmode = (mode == INPUT ? JackCaptureLatency : JackPlaybackLatency);
+    // the range (usually the min and max are equal)
+    jack_latency_range_t latrange; latrange.min = latrange.max = 0;
+    // get the latency range
+    jack_port_get_latency_range( jack_port_by_name( client, ports[firstChannel] ), cbmode, &latrange );
+    // be optimistic, use the min!
+    stream_.latency[mode] = latrange.min;
+    //stream_.latency[mode] = jack_port_get_latency( jack_port_by_name( client, ports[ firstChannel ] ) );
+  }
+  free( ports );
+
+  // The jack server always uses 32-bit floating-point data.
+  stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;
+  stream_.userFormat = format;
+
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;
+  else stream_.userInterleaved = true;
+
+  // Jack always uses non-interleaved buffers.
+  stream_.deviceInterleaved[mode] = false;
+
+  // Jack always provides host byte-ordered data.
+  stream_.doByteSwap[mode] = false;
+
+  // Get the buffer size.  The buffer size and number of buffers
+  // (periods) is set when the jack server is started.
+  stream_.bufferSize = (int) jack_get_buffer_size( client );
+  *bufferSize = stream_.bufferSize;
+
+  stream_.nDeviceChannels[mode] = channels;
+  stream_.nUserChannels[mode] = channels;
+
+  // Set flags for buffer conversion.
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+       stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate our JackHandle structure for the stream.
+  if ( handle == 0 ) {
+    try {
+      handle = new JackHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiJack::probeDeviceOpen: error allocating JackHandle memory.";
+      goto error;
+    }
+
+    if ( pthread_cond_init(&handle->condition, NULL) ) {
+      errorText_ = "RtApiJack::probeDeviceOpen: error initializing pthread condition variable.";
+      goto error;
+    }
+    stream_.apiHandle = (void *) handle;
+    handle->client = client;
+  }
+  handle->deviceName[mode] = deviceName;
+
+  // Allocate necessary internal buffers.
+  unsigned long bufferBytes;
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiJack::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    if ( mode == OUTPUT )
+      bufferBytes = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+    else { // mode == INPUT
+      bufferBytes = stream_.nDeviceChannels[1] * formatBytes( stream_.deviceFormat[1] );
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes(stream_.deviceFormat[0]);
+        if ( bufferBytes < bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiJack::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  // Allocate memory for the Jack ports (channels) identifiers.
+  handle->ports[mode] = (jack_port_t **) malloc ( sizeof (jack_port_t *) * channels );
+  if ( handle->ports[mode] == NULL )  {
+    errorText_ = "RtApiJack::probeDeviceOpen: error allocating port memory.";
+    goto error;
+  }
+
+  stream_.device[mode] = device;
+  stream_.channelOffset[mode] = firstChannel;
+  stream_.state = STREAM_STOPPED;
+  stream_.callbackInfo.object = (void *) this;
+
+  if ( stream_.mode == OUTPUT && mode == INPUT )
+    // We had already set up the stream for output.
+    stream_.mode = DUPLEX;
+  else {
+    stream_.mode = mode;
+    jack_set_process_callback( handle->client, jackCallbackHandler, (void *) &stream_.callbackInfo );
+    jack_set_xrun_callback( handle->client, jackXrun, (void *) &handle );
+    jack_on_shutdown( handle->client, jackShutdown, (void *) &stream_.callbackInfo );
+  }
+
+  // Register our ports.
+  char label[64];
+  if ( mode == OUTPUT ) {
+    for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {
+      snprintf( label, 64, "outport %d", i );
+      handle->ports[0][i] = jack_port_register( handle->client, (const char *)label,
+                                                JACK_DEFAULT_AUDIO_TYPE, JackPortIsOutput, 0 );
+    }
+  }
+  else {
+    for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {
+      snprintf( label, 64, "inport %d", i );
+      handle->ports[1][i] = jack_port_register( handle->client, (const char *)label,
+                                                JACK_DEFAULT_AUDIO_TYPE, JackPortIsInput, 0 );
+    }
+  }
+
+  // Setup the buffer conversion information structure.  We don't use
+  // buffers to do channel offsets, so we override that parameter
+  // here.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, 0 );
+
+  return SUCCESS;
+
+ error:
+  if ( handle ) {
+    pthread_cond_destroy( &handle->condition );
+    jack_client_close( handle->client );
+
+    if ( handle->ports[0] ) free( handle->ports[0] );
+    if ( handle->ports[1] ) free( handle->ports[1] );
+
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  return FAILURE;
+}
+
+void RtApiJack :: closeStream( void )
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiJack::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+  if ( handle ) {
+
+    if ( stream_.state == STREAM_RUNNING )
+      jack_deactivate( handle->client );
+
+    jack_client_close( handle->client );
+  }
+
+  if ( handle ) {
+    if ( handle->ports[0] ) free( handle->ports[0] );
+    if ( handle->ports[1] ) free( handle->ports[1] );
+    pthread_cond_destroy( &handle->condition );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+void RtApiJack :: startStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiJack::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+  int result = jack_activate( handle->client );
+  if ( result ) {
+    errorText_ = "RtApiJack::startStream(): unable to activate JACK client!";
+    goto unlock;
+  }
+
+  const char **ports;
+
+  // Get the list of available ports.
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    result = 1;
+    ports = jack_get_ports( handle->client, handle->deviceName[0].c_str(), NULL, JackPortIsInput);
+    if ( ports == NULL) {
+      errorText_ = "RtApiJack::startStream(): error determining available JACK input ports!";
+      goto unlock;
+    }
+
+    // Now make the port connections.  Since RtAudio wasn't designed to
+    // allow the user to select particular channels of a device, we'll
+    // just open the first "nChannels" ports with offset.
+    for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {
+      result = 1;
+      if ( ports[ stream_.channelOffset[0] + i ] )
+        result = jack_connect( handle->client, jack_port_name( handle->ports[0][i] ), ports[ stream_.channelOffset[0] + i ] );
+      if ( result ) {
+        free( ports );
+        errorText_ = "RtApiJack::startStream(): error connecting output ports!";
+        goto unlock;
+      }
+    }
+    free(ports);
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+    result = 1;
+    ports = jack_get_ports( handle->client, handle->deviceName[1].c_str(), NULL, JackPortIsOutput );
+    if ( ports == NULL) {
+      errorText_ = "RtApiJack::startStream(): error determining available JACK output ports!";
+      goto unlock;
+    }
+
+    // Now make the port connections.  See note above.
+    for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {
+      result = 1;
+      if ( ports[ stream_.channelOffset[1] + i ] )
+        result = jack_connect( handle->client, ports[ stream_.channelOffset[1] + i ], jack_port_name( handle->ports[1][i] ) );
+      if ( result ) {
+        free( ports );
+        errorText_ = "RtApiJack::startStream(): error connecting input ports!";
+        goto unlock;
+      }
+    }
+    free(ports);
+  }
+
+  handle->drainCounter = 0;
+  handle->internalDrain = false;
+  stream_.state = STREAM_RUNNING;
+
+ unlock:
+  if ( result == 0 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiJack :: stopStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiJack::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    if ( handle->drainCounter == 0 ) {
+      handle->drainCounter = 2;
+      pthread_cond_wait( &handle->condition, &stream_.mutex ); // block until signaled
+    }
+  }
+
+  jack_deactivate( handle->client );
+  stream_.state = STREAM_STOPPED;
+}
+
+void RtApiJack :: abortStream( void )
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiJack::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+  handle->drainCounter = 2;
+
+  stopStream();
+}
+
+// This function will be called by a spawned thread when the user
+// callback function signals that the stream should be stopped or
+// aborted.  It is necessary to handle it this way because the
+// callbackEvent() function must return before the jack_deactivate()
+// function will return.
+static void *jackStopStream( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiJack *object = (RtApiJack *) info->object;
+
+  object->stopStream();
+  pthread_exit( NULL );
+}
+
+bool RtApiJack :: callbackEvent( unsigned long nframes )
+{
+  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiCore::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return FAILURE;
+  }
+  if ( stream_.bufferSize != nframes ) {
+    errorText_ = "RtApiCore::callbackEvent(): the JACK buffer size has changed ... cannot process!";
+    error( RtAudioError::WARNING );
+    return FAILURE;
+  }
+
+  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;
+  JackHandle *handle = (JackHandle *) stream_.apiHandle;
+
+  // Check if we were draining the stream and signal is finished.
+  if ( handle->drainCounter > 3 ) {
+    ThreadHandle threadId;
+
+    stream_.state = STREAM_STOPPING;
+    if ( handle->internalDrain == true )
+      pthread_create( &threadId, NULL, jackStopStream, info );
+    else
+      pthread_cond_signal( &handle->condition );
+    return SUCCESS;
+  }
+
+  // Invoke user callback first, to get fresh output data.
+  if ( handle->drainCounter == 0 ) {
+    RtAudioCallback callback = (RtAudioCallback) info->callback;
+    double streamTime = getStreamTime();
+    RtAudioStreamStatus status = 0;
+    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {
+      status |= RTAUDIO_OUTPUT_UNDERFLOW;
+      handle->xrun[0] = false;
+    }
+    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {
+      status |= RTAUDIO_INPUT_OVERFLOW;
+      handle->xrun[1] = false;
+    }
+    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                                  stream_.bufferSize, streamTime, status, info->userData );
+    if ( cbReturnValue == 2 ) {
+      stream_.state = STREAM_STOPPING;
+      handle->drainCounter = 2;
+      ThreadHandle id;
+      pthread_create( &id, NULL, jackStopStream, info );
+      return SUCCESS;
+    }
+    else if ( cbReturnValue == 1 ) {
+      handle->drainCounter = 1;
+      handle->internalDrain = true;
+    }
+  }
+
+  jack_default_audio_sample_t *jackbuffer;
+  unsigned long bufferBytes = nframes * sizeof( jack_default_audio_sample_t );
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    if ( handle->drainCounter > 1 ) { // write zeros to the output stream
+
+      for ( unsigned int i=0; i<stream_.nDeviceChannels[0]; i++ ) {
+        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );
+        memset( jackbuffer, 0, bufferBytes );
+      }
+
+    }
+    else if ( stream_.doConvertBuffer[0] ) {
+
+      convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+
+      for ( unsigned int i=0; i<stream_.nDeviceChannels[0]; i++ ) {
+        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );
+        memcpy( jackbuffer, &stream_.deviceBuffer[i*bufferBytes], bufferBytes );
+      }
+    }
+    else { // no buffer conversion
+      for ( unsigned int i=0; i<stream_.nUserChannels[0]; i++ ) {
+        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[0][i], (jack_nframes_t) nframes );
+        memcpy( jackbuffer, &stream_.userBuffer[0][i*bufferBytes], bufferBytes );
+      }
+    }
+  }
+
+  // Don't bother draining input
+  if ( handle->drainCounter ) {
+    handle->drainCounter++;
+    goto unlock;
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    if ( stream_.doConvertBuffer[1] ) {
+      for ( unsigned int i=0; i<stream_.nDeviceChannels[1]; i++ ) {
+        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[1][i], (jack_nframes_t) nframes );
+        memcpy( &stream_.deviceBuffer[i*bufferBytes], jackbuffer, bufferBytes );
+      }
+      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );
+    }
+    else { // no buffer conversion
+      for ( unsigned int i=0; i<stream_.nUserChannels[1]; i++ ) {
+        jackbuffer = (jack_default_audio_sample_t *) jack_port_get_buffer( handle->ports[1][i], (jack_nframes_t) nframes );
+        memcpy( &stream_.userBuffer[1][i*bufferBytes], jackbuffer, bufferBytes );
+      }
+    }
+  }
+
+ unlock:
+  RtApi::tickStreamTime();
+  return SUCCESS;
+}
+  //******************** End of __UNIX_JACK__ *********************//
+#endif
+
+#if defined(__WINDOWS_ASIO__) // ASIO API on Windows
+
+// The ASIO API is designed around a callback scheme, so this
+// implementation is similar to that used for OS-X CoreAudio and Linux
+// Jack.  The primary constraint with ASIO is that it only allows
+// access to a single driver at a time.  Thus, it is not possible to
+// have more than one simultaneous RtAudio stream.
+//
+// This implementation also requires a number of external ASIO files
+// and a few global variables.  The ASIO callback scheme does not
+// allow for the passing of user data, so we must create a global
+// pointer to our callbackInfo structure.
+//
+// On unix systems, we make use of a pthread condition variable.
+// Since there is no equivalent in Windows, I hacked something based
+// on information found in
+// http://www.cs.wustl.edu/~schmidt/win32-cv-1.html.
+
+#include "asiosys.h"
+#include "asio.h"
+#include "iasiothiscallresolver.h"
+#include "asiodrivers.h"
+#include <cmath>
+
+static AsioDrivers drivers;
+static ASIOCallbacks asioCallbacks;
+static ASIODriverInfo driverInfo;
+static CallbackInfo *asioCallbackInfo;
+static bool asioXRun;
+
+struct AsioHandle {
+  int drainCounter;       // Tracks callback counts when draining
+  bool internalDrain;     // Indicates if stop is initiated from callback or not.
+  ASIOBufferInfo *bufferInfos;
+  HANDLE condition;
+
+  AsioHandle()
+    :drainCounter(0), internalDrain(false), bufferInfos(0) {}
+};
+
+// Function declarations (definitions at end of section)
+static const char* getAsioErrorString( ASIOError result );
+static void sampleRateChanged( ASIOSampleRate sRate );
+static long asioMessages( long selector, long value, void* message, double* opt );
+
+RtApiAsio :: RtApiAsio()
+{
+  // ASIO cannot run on a multi-threaded appartment. You can call
+  // CoInitialize beforehand, but it must be for appartment threading
+  // (in which case, CoInitilialize will return S_FALSE here).
+  coInitialized_ = false;
+  HRESULT hr = CoInitialize( NULL ); 
+  if ( FAILED(hr) ) {
+    errorText_ = "RtApiAsio::ASIO requires a single-threaded appartment. Call CoInitializeEx(0,COINIT_APARTMENTTHREADED)";
+    error( RtAudioError::WARNING );
+  }
+  coInitialized_ = true;
+
+  drivers.removeCurrentDriver();
+  driverInfo.asioVersion = 2;
+
+  // See note in DirectSound implementation about GetDesktopWindow().
+  driverInfo.sysRef = GetForegroundWindow();
+}
+
+RtApiAsio :: ~RtApiAsio()
+{
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+  if ( coInitialized_ ) CoUninitialize();
+}
+
+unsigned int RtApiAsio :: getDeviceCount( void )
+{
+  return (unsigned int) drivers.asioGetNumDev();
+}
+
+RtAudio::DeviceInfo RtApiAsio :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  // Get device ID
+  unsigned int nDevices = getDeviceCount();
+  if ( nDevices == 0 ) {
+    errorText_ = "RtApiAsio::getDeviceInfo: no devices found!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  if ( device >= nDevices ) {
+    errorText_ = "RtApiAsio::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  // If a stream is already open, we cannot probe other devices.  Thus, use the saved results.
+  if ( stream_.state != STREAM_CLOSED ) {
+    if ( device >= devices_.size() ) {
+      errorText_ = "RtApiAsio::getDeviceInfo: device ID was not present before stream was opened.";
+      error( RtAudioError::WARNING );
+      return info;
+    }
+    return devices_[ device ];
+  }
+
+  char driverName[32];
+  ASIOError result = drivers.asioGetDriverName( (int) device, driverName, 32 );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::getDeviceInfo: unable to get driver name (" << getAsioErrorString( result ) << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  info.name = driverName;
+
+  if ( !drivers.loadDriver( driverName ) ) {
+    errorStream_ << "RtApiAsio::getDeviceInfo: unable to load driver (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  result = ASIOInit( &driverInfo );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") initializing driver (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Determine the device channel information.
+  long inputChannels, outputChannels;
+  result = ASIOGetChannels( &inputChannels, &outputChannels );
+  if ( result != ASE_OK ) {
+    drivers.removeCurrentDriver();
+    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") getting channel count (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  info.outputChannels = outputChannels;
+  info.inputChannels = inputChannels;
+  if ( info.outputChannels > 0 && info.inputChannels > 0 )
+    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+
+  // Determine the supported sample rates.
+  info.sampleRates.clear();
+  for ( unsigned int i=0; i<MAX_SAMPLE_RATES; i++ ) {
+    result = ASIOCanSampleRate( (ASIOSampleRate) SAMPLE_RATES[i] );
+    if ( result == ASE_OK ) {
+      info.sampleRates.push_back( SAMPLE_RATES[i] );
+
+      if ( !info.preferredSampleRate || ( SAMPLE_RATES[i] <= 48000 && SAMPLE_RATES[i] > info.preferredSampleRate ) )
+        info.preferredSampleRate = SAMPLE_RATES[i];
+    }
+  }
+
+  // Determine supported data types ... just check first channel and assume rest are the same.
+  ASIOChannelInfo channelInfo;
+  channelInfo.channel = 0;
+  channelInfo.isInput = true;
+  if ( info.inputChannels <= 0 ) channelInfo.isInput = false;
+  result = ASIOGetChannelInfo( &channelInfo );
+  if ( result != ASE_OK ) {
+    drivers.removeCurrentDriver();
+    errorStream_ << "RtApiAsio::getDeviceInfo: error (" << getAsioErrorString( result ) << ") getting driver channel info (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  info.nativeFormats = 0;
+  if ( channelInfo.type == ASIOSTInt16MSB || channelInfo.type == ASIOSTInt16LSB )
+    info.nativeFormats |= RTAUDIO_SINT16;
+  else if ( channelInfo.type == ASIOSTInt32MSB || channelInfo.type == ASIOSTInt32LSB )
+    info.nativeFormats |= RTAUDIO_SINT32;
+  else if ( channelInfo.type == ASIOSTFloat32MSB || channelInfo.type == ASIOSTFloat32LSB )
+    info.nativeFormats |= RTAUDIO_FLOAT32;
+  else if ( channelInfo.type == ASIOSTFloat64MSB || channelInfo.type == ASIOSTFloat64LSB )
+    info.nativeFormats |= RTAUDIO_FLOAT64;
+  else if ( channelInfo.type == ASIOSTInt24MSB || channelInfo.type == ASIOSTInt24LSB )
+    info.nativeFormats |= RTAUDIO_SINT24;
+
+  if ( info.outputChannels > 0 )
+    if ( getDefaultOutputDevice() == device ) info.isDefaultOutput = true;
+  if ( info.inputChannels > 0 )
+    if ( getDefaultInputDevice() == device ) info.isDefaultInput = true;
+
+  info.probed = true;
+  drivers.removeCurrentDriver();
+  return info;
+}
+
+static void bufferSwitch( long index, ASIOBool /*processNow*/ )
+{
+  RtApiAsio *object = (RtApiAsio *) asioCallbackInfo->object;
+  object->callbackEvent( index );
+}
+
+void RtApiAsio :: saveDeviceInfo( void )
+{
+  devices_.clear();
+
+  unsigned int nDevices = getDeviceCount();
+  devices_.resize( nDevices );
+  for ( unsigned int i=0; i<nDevices; i++ )
+    devices_[i] = getDeviceInfo( i );
+}
+
+bool RtApiAsio :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                   unsigned int firstChannel, unsigned int sampleRate,
+                                   RtAudioFormat format, unsigned int *bufferSize,
+                                   RtAudio::StreamOptions *options )
+{////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+  bool isDuplexInput =  mode == INPUT && stream_.mode == OUTPUT;
+
+  // For ASIO, a duplex stream MUST use the same driver.
+  if ( isDuplexInput && stream_.device[0] != device ) {
+    errorText_ = "RtApiAsio::probeDeviceOpen: an ASIO duplex stream must use the same device for input and output!";
+    return FAILURE;
+  }
+
+  char driverName[32];
+  ASIOError result = drivers.asioGetDriverName( (int) device, driverName, 32 );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: unable to get driver name (" << getAsioErrorString( result ) << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Only load the driver once for duplex stream.
+  if ( !isDuplexInput ) {
+    // The getDeviceInfo() function will not work when a stream is open
+    // because ASIO does not allow multiple devices to run at the same
+    // time.  Thus, we'll probe the system before opening a stream and
+    // save the results for use by getDeviceInfo().
+    this->saveDeviceInfo();
+
+    if ( !drivers.loadDriver( driverName ) ) {
+      errorStream_ << "RtApiAsio::probeDeviceOpen: unable to load driver (" << driverName << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    result = ASIOInit( &driverInfo );
+    if ( result != ASE_OK ) {
+      errorStream_ << "RtApiAsio::probeDeviceOpen: error (" << getAsioErrorString( result ) << ") initializing driver (" << driverName << ").";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+
+  // keep them before any "goto error", they are used for error cleanup + goto device boundary checks
+  bool buffersAllocated = false;
+  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+  unsigned int nChannels;
+
+
+  // Check the device channel count.
+  long inputChannels, outputChannels;
+  result = ASIOGetChannels( &inputChannels, &outputChannels );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: error (" << getAsioErrorString( result ) << ") getting channel count (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  if ( ( mode == OUTPUT && (channels+firstChannel) > (unsigned int) outputChannels) ||
+       ( mode == INPUT && (channels+firstChannel) > (unsigned int) inputChannels) ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") does not support requested channel count (" << channels << ") + offset (" << firstChannel << ").";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+  stream_.nDeviceChannels[mode] = channels;
+  stream_.nUserChannels[mode] = channels;
+  stream_.channelOffset[mode] = firstChannel;
+
+  // Verify the sample rate is supported.
+  result = ASIOCanSampleRate( (ASIOSampleRate) sampleRate );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") does not support requested sample rate (" << sampleRate << ").";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  // Get the current sample rate
+  ASIOSampleRate currentRate;
+  result = ASIOGetSampleRate( &currentRate );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error getting sample rate.";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  // Set the sample rate only if necessary
+  if ( currentRate != sampleRate ) {
+    result = ASIOSetSampleRate( (ASIOSampleRate) sampleRate );
+    if ( result != ASE_OK ) {
+      errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error setting sample rate (" << sampleRate << ").";
+      errorText_ = errorStream_.str();
+      goto error;
+    }
+  }
+
+  // Determine the driver data type.
+  ASIOChannelInfo channelInfo;
+  channelInfo.channel = 0;
+  if ( mode == OUTPUT ) channelInfo.isInput = false;
+  else channelInfo.isInput = true;
+  result = ASIOGetChannelInfo( &channelInfo );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting data format.";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  // Assuming WINDOWS host is always little-endian.
+  stream_.doByteSwap[mode] = false;
+  stream_.userFormat = format;
+  stream_.deviceFormat[mode] = 0;
+  if ( channelInfo.type == ASIOSTInt16MSB || channelInfo.type == ASIOSTInt16LSB ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+    if ( channelInfo.type == ASIOSTInt16MSB ) stream_.doByteSwap[mode] = true;
+  }
+  else if ( channelInfo.type == ASIOSTInt32MSB || channelInfo.type == ASIOSTInt32LSB ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+    if ( channelInfo.type == ASIOSTInt32MSB ) stream_.doByteSwap[mode] = true;
+  }
+  else if ( channelInfo.type == ASIOSTFloat32MSB || channelInfo.type == ASIOSTFloat32LSB ) {
+    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;
+    if ( channelInfo.type == ASIOSTFloat32MSB ) stream_.doByteSwap[mode] = true;
+  }
+  else if ( channelInfo.type == ASIOSTFloat64MSB || channelInfo.type == ASIOSTFloat64LSB ) {
+    stream_.deviceFormat[mode] = RTAUDIO_FLOAT64;
+    if ( channelInfo.type == ASIOSTFloat64MSB ) stream_.doByteSwap[mode] = true;
+  }
+  else if ( channelInfo.type == ASIOSTInt24MSB || channelInfo.type == ASIOSTInt24LSB ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+    if ( channelInfo.type == ASIOSTInt24MSB ) stream_.doByteSwap[mode] = true;
+  }
+
+  if ( stream_.deviceFormat[mode] == 0 ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") data format not supported by RtAudio.";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  // Set the buffer size.  For a duplex stream, this will end up
+  // setting the buffer size based on the input constraints, which
+  // should be ok.
+  long minSize, maxSize, preferSize, granularity;
+  result = ASIOGetBufferSize( &minSize, &maxSize, &preferSize, &granularity );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting buffer size.";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  if ( isDuplexInput ) {
+    // When this is the duplex input (output was opened before), then we have to use the same
+    // buffersize as the output, because it might use the preferred buffer size, which most
+    // likely wasn't passed as input to this. The buffer sizes have to be identically anyway,
+    // So instead of throwing an error, make them equal. The caller uses the reference
+    // to the "bufferSize" param as usual to set up processing buffers.
+
+    *bufferSize = stream_.bufferSize;
+
+  } else {
+    if ( *bufferSize == 0 ) *bufferSize = preferSize;
+    else if ( *bufferSize < (unsigned int) minSize ) *bufferSize = (unsigned int) minSize;
+    else if ( *bufferSize > (unsigned int) maxSize ) *bufferSize = (unsigned int) maxSize;
+    else if ( granularity == -1 ) {
+      // Make sure bufferSize is a power of two.
+      int log2_of_min_size = 0;
+      int log2_of_max_size = 0;
+
+      for ( unsigned int i = 0; i < sizeof(long) * 8; i++ ) {
+        if ( minSize & ((long)1 << i) ) log2_of_min_size = i;
+        if ( maxSize & ((long)1 << i) ) log2_of_max_size = i;
+      }
+
+      long min_delta = std::abs( (long)*bufferSize - ((long)1 << log2_of_min_size) );
+      int min_delta_num = log2_of_min_size;
+
+      for (int i = log2_of_min_size + 1; i <= log2_of_max_size; i++) {
+        long current_delta = std::abs( (long)*bufferSize - ((long)1 << i) );
+        if (current_delta < min_delta) {
+          min_delta = current_delta;
+          min_delta_num = i;
+        }
+      }
+
+      *bufferSize = ( (unsigned int)1 << min_delta_num );
+      if ( *bufferSize < (unsigned int) minSize ) *bufferSize = (unsigned int) minSize;
+      else if ( *bufferSize > (unsigned int) maxSize ) *bufferSize = (unsigned int) maxSize;
+    }
+    else if ( granularity != 0 ) {
+      // Set to an even multiple of granularity, rounding up.
+      *bufferSize = (*bufferSize + granularity-1) / granularity * granularity;
+    }
+  }
+
+  /*
+  // we don't use it anymore, see above!
+  // Just left it here for the case...
+  if ( isDuplexInput && stream_.bufferSize != *bufferSize ) {
+    errorText_ = "RtApiAsio::probeDeviceOpen: input/output buffersize discrepancy!";
+    goto error;
+  }
+  */
+
+  stream_.bufferSize = *bufferSize;
+  stream_.nBuffers = 2;
+
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;
+  else stream_.userInterleaved = true;
+
+  // ASIO always uses non-interleaved buffers.
+  stream_.deviceInterleaved[mode] = false;
+
+  // Allocate, if necessary, our AsioHandle structure for the stream.
+  if ( handle == 0 ) {
+    try {
+      handle = new AsioHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiAsio::probeDeviceOpen: error allocating AsioHandle memory.";
+      goto error;
+    }
+    handle->bufferInfos = 0;
+
+    // Create a manual-reset event.
+    handle->condition = CreateEvent( NULL,   // no security
+                                     TRUE,   // manual-reset
+                                     FALSE,  // non-signaled initially
+                                     NULL ); // unnamed
+    stream_.apiHandle = (void *) handle;
+  }
+
+  // Create the ASIO internal buffers.  Since RtAudio sets up input
+  // and output separately, we'll have to dispose of previously
+  // created output buffers for a duplex stream.
+  if ( mode == INPUT && stream_.mode == OUTPUT ) {
+    ASIODisposeBuffers();
+    if ( handle->bufferInfos ) free( handle->bufferInfos );
+  }
+
+  // Allocate, initialize, and save the bufferInfos in our stream callbackInfo structure.
+  unsigned int i;
+  nChannels = stream_.nDeviceChannels[0] + stream_.nDeviceChannels[1];
+  handle->bufferInfos = (ASIOBufferInfo *) malloc( nChannels * sizeof(ASIOBufferInfo) );
+  if ( handle->bufferInfos == NULL ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: error allocating bufferInfo memory for driver (" << driverName << ").";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+
+  ASIOBufferInfo *infos;
+  infos = handle->bufferInfos;
+  for ( i=0; i<stream_.nDeviceChannels[0]; i++, infos++ ) {
+    infos->isInput = ASIOFalse;
+    infos->channelNum = i + stream_.channelOffset[0];
+    infos->buffers[0] = infos->buffers[1] = 0;
+  }
+  for ( i=0; i<stream_.nDeviceChannels[1]; i++, infos++ ) {
+    infos->isInput = ASIOTrue;
+    infos->channelNum = i + stream_.channelOffset[1];
+    infos->buffers[0] = infos->buffers[1] = 0;
+  }
+
+  // prepare for callbacks
+  stream_.sampleRate = sampleRate;
+  stream_.device[mode] = device;
+  stream_.mode = isDuplexInput ? DUPLEX : mode;
+
+  // store this class instance before registering callbacks, that are going to use it
+  asioCallbackInfo = &stream_.callbackInfo;
+  stream_.callbackInfo.object = (void *) this;
+
+  // Set up the ASIO callback structure and create the ASIO data buffers.
+  asioCallbacks.bufferSwitch = &bufferSwitch;
+  asioCallbacks.sampleRateDidChange = &sampleRateChanged;
+  asioCallbacks.asioMessage = &asioMessages;
+  asioCallbacks.bufferSwitchTimeInfo = NULL;
+  result = ASIOCreateBuffers( handle->bufferInfos, nChannels, stream_.bufferSize, &asioCallbacks );
+  if ( result != ASE_OK ) {
+    // Standard method failed. This can happen with strict/misbehaving drivers that return valid buffer size ranges
+    // but only accept the preferred buffer size as parameter for ASIOCreateBuffers. eg. Creatives ASIO driver
+    // in that case, let's be naïve and try that instead
+    *bufferSize = preferSize;
+    stream_.bufferSize = *bufferSize;
+    result = ASIOCreateBuffers( handle->bufferInfos, nChannels, stream_.bufferSize, &asioCallbacks );
+  }
+
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") creating buffers.";
+    errorText_ = errorStream_.str();
+    goto error;
+  }
+  buffersAllocated = true;  
+  stream_.state = STREAM_STOPPED;
+
+  // Set flags for buffer conversion.
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+       stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate necessary internal buffers
+  unsigned long bufferBytes;
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiAsio::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( isDuplexInput && stream_.deviceBuffer ) {
+      unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+      if ( bufferBytes <= bytesOut ) makeBuffer = false;
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiAsio::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  // Determine device latencies
+  long inputLatency, outputLatency;
+  result = ASIOGetLatencies( &inputLatency, &outputLatency );
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::probeDeviceOpen: driver (" << driverName << ") error (" << getAsioErrorString( result ) << ") getting latency.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING); // warn but don't fail
+  }
+  else {
+    stream_.latency[0] = outputLatency;
+    stream_.latency[1] = inputLatency;
+  }
+
+  // Setup the buffer conversion information structure.  We don't use
+  // buffers to do channel offsets, so we override that parameter
+  // here.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, 0 );
+
+  return SUCCESS;
+
+ error:
+  if ( !isDuplexInput ) {
+    // the cleanup for error in the duplex input, is done by RtApi::openStream
+    // So we clean up for single channel only
+
+    if ( buffersAllocated )
+      ASIODisposeBuffers();
+
+    drivers.removeCurrentDriver();
+
+    if ( handle ) {
+      CloseHandle( handle->condition );
+      if ( handle->bufferInfos )
+        free( handle->bufferInfos );
+
+      delete handle;
+      stream_.apiHandle = 0;
+    }
+
+
+    if ( stream_.userBuffer[mode] ) {
+      free( stream_.userBuffer[mode] );
+      stream_.userBuffer[mode] = 0;
+    }
+
+    if ( stream_.deviceBuffer ) {
+      free( stream_.deviceBuffer );
+      stream_.deviceBuffer = 0;
+    }
+  }
+
+  return FAILURE;
+}////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+void RtApiAsio :: closeStream()
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiAsio::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  if ( stream_.state == STREAM_RUNNING ) {
+    stream_.state = STREAM_STOPPED;
+    ASIOStop();
+  }
+  ASIODisposeBuffers();
+  drivers.removeCurrentDriver();
+
+  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+  if ( handle ) {
+    CloseHandle( handle->condition );
+    if ( handle->bufferInfos )
+      free( handle->bufferInfos );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+bool stopThreadCalled = false;
+
+void RtApiAsio :: startStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiAsio::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+  ASIOError result = ASIOStart();
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::startStream: error (" << getAsioErrorString( result ) << ") starting device.";
+    errorText_ = errorStream_.str();
+    goto unlock;
+  }
+
+  handle->drainCounter = 0;
+  handle->internalDrain = false;
+  ResetEvent( handle->condition );
+  stream_.state = STREAM_RUNNING;
+  asioXRun = false;
+
+ unlock:
+  stopThreadCalled = false;
+
+  if ( result == ASE_OK ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiAsio :: stopStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiAsio::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    if ( handle->drainCounter == 0 ) {
+      handle->drainCounter = 2;
+      WaitForSingleObject( handle->condition, INFINITE );  // block until signaled
+    }
+  }
+
+  stream_.state = STREAM_STOPPED;
+
+  ASIOError result = ASIOStop();
+  if ( result != ASE_OK ) {
+    errorStream_ << "RtApiAsio::stopStream: error (" << getAsioErrorString( result ) << ") stopping device.";
+    errorText_ = errorStream_.str();
+  }
+
+  if ( result == ASE_OK ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiAsio :: abortStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiAsio::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // The following lines were commented-out because some behavior was
+  // noted where the device buffers need to be zeroed to avoid
+  // continuing sound, even when the device buffers are completely
+  // disposed.  So now, calling abort is the same as calling stop.
+  // AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+  // handle->drainCounter = 2;
+  stopStream();
+}
+
+// This function will be called by a spawned thread when the user
+// callback function signals that the stream should be stopped or
+// aborted.  It is necessary to handle it this way because the
+// callbackEvent() function must return before the ASIOStop()
+// function will return.
+static unsigned __stdcall asioStopStream( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiAsio *object = (RtApiAsio *) info->object;
+
+  object->stopStream();
+  _endthreadex( 0 );
+  return 0;
+}
+
+bool RtApiAsio :: callbackEvent( long bufferIndex )
+{
+  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) return SUCCESS;
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiAsio::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return FAILURE;
+  }
+
+  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;
+  AsioHandle *handle = (AsioHandle *) stream_.apiHandle;
+
+  // Check if we were draining the stream and signal if finished.
+  if ( handle->drainCounter > 3 ) {
+
+    stream_.state = STREAM_STOPPING;
+    if ( handle->internalDrain == false )
+      SetEvent( handle->condition );
+    else { // spawn a thread to stop the stream
+      unsigned threadId;
+      stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &asioStopStream,
+                                                    &stream_.callbackInfo, 0, &threadId );
+    }
+    return SUCCESS;
+  }
+
+  // Invoke user callback to get fresh output data UNLESS we are
+  // draining stream.
+  if ( handle->drainCounter == 0 ) {
+    RtAudioCallback callback = (RtAudioCallback) info->callback;
+    double streamTime = getStreamTime();
+    RtAudioStreamStatus status = 0;
+    if ( stream_.mode != INPUT && asioXRun == true ) {
+      status |= RTAUDIO_OUTPUT_UNDERFLOW;
+      asioXRun = false;
+    }
+    if ( stream_.mode != OUTPUT && asioXRun == true ) {
+      status |= RTAUDIO_INPUT_OVERFLOW;
+      asioXRun = false;
+    }
+    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                                     stream_.bufferSize, streamTime, status, info->userData );
+    if ( cbReturnValue == 2 ) {
+      stream_.state = STREAM_STOPPING;
+      handle->drainCounter = 2;
+      unsigned threadId;
+      stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &asioStopStream,
+                                                    &stream_.callbackInfo, 0, &threadId );
+      return SUCCESS;
+    }
+    else if ( cbReturnValue == 1 ) {
+      handle->drainCounter = 1;
+      handle->internalDrain = true;
+    }
+  }
+
+  unsigned int nChannels, bufferBytes, i, j;
+  nChannels = stream_.nDeviceChannels[0] + stream_.nDeviceChannels[1];
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    bufferBytes = stream_.bufferSize * formatBytes( stream_.deviceFormat[0] );
+
+    if ( handle->drainCounter > 1 ) { // write zeros to the output stream
+
+      for ( i=0, j=0; i<nChannels; i++ ) {
+        if ( handle->bufferInfos[i].isInput != ASIOTrue )
+          memset( handle->bufferInfos[i].buffers[bufferIndex], 0, bufferBytes );
+      }
+
+    }
+    else if ( stream_.doConvertBuffer[0] ) {
+
+      convertBuffer( stream_.deviceBuffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+      if ( stream_.doByteSwap[0] )
+        byteSwapBuffer( stream_.deviceBuffer,
+                        stream_.bufferSize * stream_.nDeviceChannels[0],
+                        stream_.deviceFormat[0] );
+
+      for ( i=0, j=0; i<nChannels; i++ ) {
+        if ( handle->bufferInfos[i].isInput != ASIOTrue )
+          memcpy( handle->bufferInfos[i].buffers[bufferIndex],
+                  &stream_.deviceBuffer[j++*bufferBytes], bufferBytes );
+      }
+
+    }
+    else {
+
+      if ( stream_.doByteSwap[0] )
+        byteSwapBuffer( stream_.userBuffer[0],
+                        stream_.bufferSize * stream_.nUserChannels[0],
+                        stream_.userFormat );
+
+      for ( i=0, j=0; i<nChannels; i++ ) {
+        if ( handle->bufferInfos[i].isInput != ASIOTrue )
+          memcpy( handle->bufferInfos[i].buffers[bufferIndex],
+                  &stream_.userBuffer[0][bufferBytes*j++], bufferBytes );
+      }
+
+    }
+  }
+
+  // Don't bother draining input
+  if ( handle->drainCounter ) {
+    handle->drainCounter++;
+    goto unlock;
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    bufferBytes = stream_.bufferSize * formatBytes(stream_.deviceFormat[1]);
+
+    if (stream_.doConvertBuffer[1]) {
+
+      // Always interleave ASIO input data.
+      for ( i=0, j=0; i<nChannels; i++ ) {
+        if ( handle->bufferInfos[i].isInput == ASIOTrue )
+          memcpy( &stream_.deviceBuffer[j++*bufferBytes],
+                  handle->bufferInfos[i].buffers[bufferIndex],
+                  bufferBytes );
+      }
+
+      if ( stream_.doByteSwap[1] )
+        byteSwapBuffer( stream_.deviceBuffer,
+                        stream_.bufferSize * stream_.nDeviceChannels[1],
+                        stream_.deviceFormat[1] );
+      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );
+
+    }
+    else {
+      for ( i=0, j=0; i<nChannels; i++ ) {
+        if ( handle->bufferInfos[i].isInput == ASIOTrue ) {
+          memcpy( &stream_.userBuffer[1][bufferBytes*j++],
+                  handle->bufferInfos[i].buffers[bufferIndex],
+                  bufferBytes );
+        }
+      }
+
+      if ( stream_.doByteSwap[1] )
+        byteSwapBuffer( stream_.userBuffer[1],
+                        stream_.bufferSize * stream_.nUserChannels[1],
+                        stream_.userFormat );
+    }
+  }
+
+ unlock:
+  // The following call was suggested by Malte Clasen.  While the API
+  // documentation indicates it should not be required, some device
+  // drivers apparently do not function correctly without it.
+  ASIOOutputReady();
+
+  RtApi::tickStreamTime();
+  return SUCCESS;
+}
+
+static void sampleRateChanged( ASIOSampleRate sRate )
+{
+  // The ASIO documentation says that this usually only happens during
+  // external sync.  Audio processing is not stopped by the driver,
+  // actual sample rate might not have even changed, maybe only the
+  // sample rate status of an AES/EBU or S/PDIF digital input at the
+  // audio device.
+
+  RtApi *object = (RtApi *) asioCallbackInfo->object;
+  try {
+    object->stopStream();
+  }
+  catch ( RtAudioError &exception ) {
+    std::cerr << "\nRtApiAsio: sampleRateChanged() error (" << exception.getMessage() << ")!\n" << std::endl;
+    return;
+  }
+
+  std::cerr << "\nRtApiAsio: driver reports sample rate changed to " << sRate << " ... stream stopped!!!\n" << std::endl;
+}
+
+static long asioMessages( long selector, long value, void* /*message*/, double* /*opt*/ )
+{
+  long ret = 0;
+
+  switch( selector ) {
+  case kAsioSelectorSupported:
+    if ( value == kAsioResetRequest
+         || value == kAsioEngineVersion
+         || value == kAsioResyncRequest
+         || value == kAsioLatenciesChanged
+         // The following three were added for ASIO 2.0, you don't
+         // necessarily have to support them.
+         || value == kAsioSupportsTimeInfo
+         || value == kAsioSupportsTimeCode
+         || value == kAsioSupportsInputMonitor)
+      ret = 1L;
+    break;
+  case kAsioResetRequest:
+    // Defer the task and perform the reset of the driver during the
+    // next "safe" situation.  You cannot reset the driver right now,
+    // as this code is called from the driver.  Reset the driver is
+    // done by completely destruct is. I.e. ASIOStop(),
+    // ASIODisposeBuffers(), Destruction Afterwards you initialize the
+    // driver again.
+    std::cerr << "\nRtApiAsio: driver reset requested!!!" << std::endl;
+    ret = 1L;
+    break;
+  case kAsioResyncRequest:
+    // This informs the application that the driver encountered some
+    // non-fatal data loss.  It is used for synchronization purposes
+    // of different media.  Added mainly to work around the Win16Mutex
+    // problems in Windows 95/98 with the Windows Multimedia system,
+    // which could lose data because the Mutex was held too long by
+    // another thread.  However a driver can issue it in other
+    // situations, too.
+    // std::cerr << "\nRtApiAsio: driver resync requested!!!" << std::endl;
+    asioXRun = true;
+    ret = 1L;
+    break;
+  case kAsioLatenciesChanged:
+    // This will inform the host application that the drivers were
+    // latencies changed.  Beware, it this does not mean that the
+    // buffer sizes have changed!  You might need to update internal
+    // delay data.
+    std::cerr << "\nRtApiAsio: driver latency may have changed!!!" << std::endl;
+    ret = 1L;
+    break;
+  case kAsioEngineVersion:
+    // Return the supported ASIO version of the host application.  If
+    // a host application does not implement this selector, ASIO 1.0
+    // is assumed by the driver.
+    ret = 2L;
+    break;
+  case kAsioSupportsTimeInfo:
+    // Informs the driver whether the
+    // asioCallbacks.bufferSwitchTimeInfo() callback is supported.
+    // For compatibility with ASIO 1.0 drivers the host application
+    // should always support the "old" bufferSwitch method, too.
+    ret = 0;
+    break;
+  case kAsioSupportsTimeCode:
+    // Informs the driver whether application is interested in time
+    // code info.  If an application does not need to know about time
+    // code, the driver has less work to do.
+    ret = 0;
+    break;
+  }
+  return ret;
+}
+
+static const char* getAsioErrorString( ASIOError result )
+{
+  struct Messages 
+  {
+    ASIOError value;
+    const char*message;
+  };
+
+  static const Messages m[] = 
+    {
+      {   ASE_NotPresent,    "Hardware input or output is not present or available." },
+      {   ASE_HWMalfunction,  "Hardware is malfunctioning." },
+      {   ASE_InvalidParameter, "Invalid input parameter." },
+      {   ASE_InvalidMode,      "Invalid mode." },
+      {   ASE_SPNotAdvancing,     "Sample position not advancing." },
+      {   ASE_NoClock,            "Sample clock or rate cannot be determined or is not present." },
+      {   ASE_NoMemory,           "Not enough memory to complete the request." }
+    };
+
+  for ( unsigned int i = 0; i < sizeof(m)/sizeof(m[0]); ++i )
+    if ( m[i].value == result ) return m[i].message;
+
+  return "Unknown error.";
+}
+
+//******************** End of __WINDOWS_ASIO__ *********************//
+#endif
+
+
+#if defined(__WINDOWS_WASAPI__) // Windows WASAPI API
+
+// Authored by Marcus Tomlinson <themarcustomlinson@gmail.com>, April 2014
+// - Introduces support for the Windows WASAPI API
+// - Aims to deliver bit streams to and from hardware at the lowest possible latency, via the absolute minimum buffer sizes required
+// - Provides flexible stream configuration to an otherwise strict and inflexible WASAPI interface
+// - Includes automatic internal conversion of sample rate and buffer size between hardware and the user
+
+#ifndef INITGUID
+  #define INITGUID
+#endif
+#include <audioclient.h>
+#include <avrt.h>
+#include <mmdeviceapi.h>
+#include <functiondiscoverykeys_devpkey.h>
+
+//=============================================================================
+
+#define SAFE_RELEASE( objectPtr )\
+if ( objectPtr )\
+{\
+  objectPtr->Release();\
+  objectPtr = NULL;\
+}
+
+typedef HANDLE ( __stdcall *TAvSetMmThreadCharacteristicsPtr )( LPCWSTR TaskName, LPDWORD TaskIndex );
+
+//-----------------------------------------------------------------------------
+
+// WASAPI dictates stream sample rate, format, channel count, and in some cases, buffer size.
+// Therefore we must perform all necessary conversions to user buffers in order to satisfy these
+// requirements. WasapiBuffer ring buffers are used between HwIn->UserIn and UserOut->HwOut to
+// provide intermediate storage for read / write synchronization.
+class WasapiBuffer
+{
+public:
+  WasapiBuffer()
+    : buffer_( NULL ),
+      bufferSize_( 0 ),
+      inIndex_( 0 ),
+      outIndex_( 0 ) {}
+
+  ~WasapiBuffer() {
+    free( buffer_ );
+  }
+
+  // sets the length of the internal ring buffer
+  void setBufferSize( unsigned int bufferSize, unsigned int formatBytes ) {
+    free( buffer_ );
+
+    buffer_ = ( char* ) calloc( bufferSize, formatBytes );
+
+    bufferSize_ = bufferSize;
+    inIndex_ = 0;
+    outIndex_ = 0;
+  }
+
+  // attempt to push a buffer into the ring buffer at the current "in" index
+  bool pushBuffer( char* buffer, unsigned int bufferSize, RtAudioFormat format )
+  {
+    if ( !buffer ||                 // incoming buffer is NULL
+         bufferSize == 0 ||         // incoming buffer has no data
+         bufferSize > bufferSize_ ) // incoming buffer too large
+    {
+      return false;
+    }
+
+    unsigned int relOutIndex = outIndex_;
+    unsigned int inIndexEnd = inIndex_ + bufferSize;
+    if ( relOutIndex < inIndex_ && inIndexEnd >= bufferSize_ ) {
+      relOutIndex += bufferSize_;
+    }
+
+    // "in" index can end on the "out" index but cannot begin at it
+    if ( inIndex_ <= relOutIndex && inIndexEnd > relOutIndex ) {
+      return false; // not enough space between "in" index and "out" index
+    }
+
+    // copy buffer from external to internal
+    int fromZeroSize = inIndex_ + bufferSize - bufferSize_;
+    fromZeroSize = fromZeroSize < 0 ? 0 : fromZeroSize;
+    int fromInSize = bufferSize - fromZeroSize;
+
+    switch( format )
+      {
+      case RTAUDIO_SINT8:
+        memcpy( &( ( char* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( char ) );
+        memcpy( buffer_, &( ( char* ) buffer )[fromInSize], fromZeroSize * sizeof( char ) );
+        break;
+      case RTAUDIO_SINT16:
+        memcpy( &( ( short* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( short ) );
+        memcpy( buffer_, &( ( short* ) buffer )[fromInSize], fromZeroSize * sizeof( short ) );
+        break;
+      case RTAUDIO_SINT24:
+        memcpy( &( ( S24* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( S24 ) );
+        memcpy( buffer_, &( ( S24* ) buffer )[fromInSize], fromZeroSize * sizeof( S24 ) );
+        break;
+      case RTAUDIO_SINT32:
+        memcpy( &( ( int* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( int ) );
+        memcpy( buffer_, &( ( int* ) buffer )[fromInSize], fromZeroSize * sizeof( int ) );
+        break;
+      case RTAUDIO_FLOAT32:
+        memcpy( &( ( float* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( float ) );
+        memcpy( buffer_, &( ( float* ) buffer )[fromInSize], fromZeroSize * sizeof( float ) );
+        break;
+      case RTAUDIO_FLOAT64:
+        memcpy( &( ( double* ) buffer_ )[inIndex_], buffer, fromInSize * sizeof( double ) );
+        memcpy( buffer_, &( ( double* ) buffer )[fromInSize], fromZeroSize * sizeof( double ) );
+        break;
+    }
+
+    // update "in" index
+    inIndex_ += bufferSize;
+    inIndex_ %= bufferSize_;
+
+    return true;
+  }
+
+  // attempt to pull a buffer from the ring buffer from the current "out" index
+  bool pullBuffer( char* buffer, unsigned int bufferSize, RtAudioFormat format )
+  {
+    if ( !buffer ||                 // incoming buffer is NULL
+         bufferSize == 0 ||         // incoming buffer has no data
+         bufferSize > bufferSize_ ) // incoming buffer too large
+    {
+      return false;
+    }
+
+    unsigned int relInIndex = inIndex_;
+    unsigned int outIndexEnd = outIndex_ + bufferSize;
+    if ( relInIndex < outIndex_ && outIndexEnd >= bufferSize_ ) {
+      relInIndex += bufferSize_;
+    }
+
+    // "out" index can begin at and end on the "in" index
+    if ( outIndex_ < relInIndex && outIndexEnd > relInIndex ) {
+      return false; // not enough space between "out" index and "in" index
+    }
+
+    // copy buffer from internal to external
+    int fromZeroSize = outIndex_ + bufferSize - bufferSize_;
+    fromZeroSize = fromZeroSize < 0 ? 0 : fromZeroSize;
+    int fromOutSize = bufferSize - fromZeroSize;
+
+    switch( format )
+    {
+      case RTAUDIO_SINT8:
+        memcpy( buffer, &( ( char* ) buffer_ )[outIndex_], fromOutSize * sizeof( char ) );
+        memcpy( &( ( char* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( char ) );
+        break;
+      case RTAUDIO_SINT16:
+        memcpy( buffer, &( ( short* ) buffer_ )[outIndex_], fromOutSize * sizeof( short ) );
+        memcpy( &( ( short* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( short ) );
+        break;
+      case RTAUDIO_SINT24:
+        memcpy( buffer, &( ( S24* ) buffer_ )[outIndex_], fromOutSize * sizeof( S24 ) );
+        memcpy( &( ( S24* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( S24 ) );
+        break;
+      case RTAUDIO_SINT32:
+        memcpy( buffer, &( ( int* ) buffer_ )[outIndex_], fromOutSize * sizeof( int ) );
+        memcpy( &( ( int* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( int ) );
+        break;
+      case RTAUDIO_FLOAT32:
+        memcpy( buffer, &( ( float* ) buffer_ )[outIndex_], fromOutSize * sizeof( float ) );
+        memcpy( &( ( float* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( float ) );
+        break;
+      case RTAUDIO_FLOAT64:
+        memcpy( buffer, &( ( double* ) buffer_ )[outIndex_], fromOutSize * sizeof( double ) );
+        memcpy( &( ( double* ) buffer )[fromOutSize], buffer_, fromZeroSize * sizeof( double ) );
+        break;
+    }
+
+    // update "out" index
+    outIndex_ += bufferSize;
+    outIndex_ %= bufferSize_;
+
+    return true;
+  }
+
+private:
+  char* buffer_;
+  unsigned int bufferSize_;
+  unsigned int inIndex_;
+  unsigned int outIndex_;
+};
+
+//-----------------------------------------------------------------------------
+
+// In order to satisfy WASAPI's buffer requirements, we need a means of converting sample rate
+// between HW and the user. The convertBufferWasapi function is used to perform this conversion
+// between HwIn->UserIn and UserOut->HwOut during the stream callback loop.
+// This sample rate converter favors speed over quality, and works best with conversions between
+// one rate and its multiple.
+void convertBufferWasapi( char* outBuffer,
+                          const char* inBuffer,
+                          const unsigned int& channelCount,
+                          const unsigned int& inSampleRate,
+                          const unsigned int& outSampleRate,
+                          const unsigned int& inSampleCount,
+                          unsigned int& outSampleCount,
+                          const RtAudioFormat& format )
+{
+  // calculate the new outSampleCount and relative sampleStep
+  float sampleRatio = ( float ) outSampleRate / inSampleRate;
+  float sampleStep = 1.0f / sampleRatio;
+  float inSampleFraction = 0.0f;
+
+  outSampleCount = ( unsigned int ) roundf( inSampleCount * sampleRatio );
+
+  // frame-by-frame, copy each relative input sample into it's corresponding output sample
+  for ( unsigned int outSample = 0; outSample < outSampleCount; outSample++ )
+  {
+    unsigned int inSample = ( unsigned int ) inSampleFraction;
+
+    switch ( format )
+    {
+      case RTAUDIO_SINT8:
+        memcpy( &( ( char* ) outBuffer )[ outSample * channelCount ], &( ( char* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( char ) );
+        break;
+      case RTAUDIO_SINT16:
+        memcpy( &( ( short* ) outBuffer )[ outSample * channelCount ], &( ( short* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( short ) );
+        break;
+      case RTAUDIO_SINT24:
+        memcpy( &( ( S24* ) outBuffer )[ outSample * channelCount ], &( ( S24* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( S24 ) );
+        break;
+      case RTAUDIO_SINT32:
+        memcpy( &( ( int* ) outBuffer )[ outSample * channelCount ], &( ( int* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( int ) );
+        break;
+      case RTAUDIO_FLOAT32:
+        memcpy( &( ( float* ) outBuffer )[ outSample * channelCount ], &( ( float* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( float ) );
+        break;
+      case RTAUDIO_FLOAT64:
+        memcpy( &( ( double* ) outBuffer )[ outSample * channelCount ], &( ( double* ) inBuffer )[ inSample * channelCount ], channelCount * sizeof( double ) );
+        break;
+    }
+
+    // jump to next in sample
+    inSampleFraction += sampleStep;
+  }
+}
+
+//-----------------------------------------------------------------------------
+
+// A structure to hold various information related to the WASAPI implementation.
+struct WasapiHandle
+{
+  IAudioClient* captureAudioClient;
+  IAudioClient* renderAudioClient;
+  IAudioCaptureClient* captureClient;
+  IAudioRenderClient* renderClient;
+  HANDLE captureEvent;
+  HANDLE renderEvent;
+
+  WasapiHandle()
+  : captureAudioClient( NULL ),
+    renderAudioClient( NULL ),
+    captureClient( NULL ),
+    renderClient( NULL ),
+    captureEvent( NULL ),
+    renderEvent( NULL ) {}
+};
+
+//=============================================================================
+
+RtApiWasapi::RtApiWasapi()
+  : coInitialized_( false ), deviceEnumerator_( NULL )
+{
+  // WASAPI can run either apartment or multi-threaded
+  HRESULT hr = CoInitialize( NULL );
+  if ( !FAILED( hr ) )
+    coInitialized_ = true;
+
+  // Instantiate device enumerator
+  hr = CoCreateInstance( __uuidof( MMDeviceEnumerator ), NULL,
+                         CLSCTX_ALL, __uuidof( IMMDeviceEnumerator ),
+                         ( void** ) &deviceEnumerator_ );
+
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::RtApiWasapi: Unable to instantiate device enumerator";
+    error( RtAudioError::DRIVER_ERROR );
+  }
+}
+
+//-----------------------------------------------------------------------------
+
+RtApiWasapi::~RtApiWasapi()
+{
+  if ( stream_.state != STREAM_CLOSED )
+    closeStream();
+
+  SAFE_RELEASE( deviceEnumerator_ );
+
+  // If this object previously called CoInitialize()
+  if ( coInitialized_ )
+    CoUninitialize();
+}
+
+//=============================================================================
+
+unsigned int RtApiWasapi::getDeviceCount( void )
+{
+  unsigned int captureDeviceCount = 0;
+  unsigned int renderDeviceCount = 0;
+
+  IMMDeviceCollection* captureDevices = NULL;
+  IMMDeviceCollection* renderDevices = NULL;
+
+  // Count capture devices
+  errorText_.clear();
+  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve capture device collection.";
+    goto Exit;
+  }
+
+  hr = captureDevices->GetCount( &captureDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve capture device count.";
+    goto Exit;
+  }
+
+  // Count render devices
+  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve render device collection.";
+    goto Exit;
+  }
+
+  hr = renderDevices->GetCount( &renderDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceCount: Unable to retrieve render device count.";
+    goto Exit;
+  }
+
+Exit:
+  // release all references
+  SAFE_RELEASE( captureDevices );
+  SAFE_RELEASE( renderDevices );
+
+  if ( errorText_.empty() )
+    return captureDeviceCount + renderDeviceCount;
+
+  error( RtAudioError::DRIVER_ERROR );
+  return 0;
+}
+
+//-----------------------------------------------------------------------------
+
+RtAudio::DeviceInfo RtApiWasapi::getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  unsigned int captureDeviceCount = 0;
+  unsigned int renderDeviceCount = 0;
+  std::string defaultDeviceName;
+  bool isCaptureDevice = false;
+
+  PROPVARIANT deviceNameProp;
+  PROPVARIANT defaultDeviceNameProp;
+
+  IMMDeviceCollection* captureDevices = NULL;
+  IMMDeviceCollection* renderDevices = NULL;
+  IMMDevice* devicePtr = NULL;
+  IMMDevice* defaultDevicePtr = NULL;
+  IAudioClient* audioClient = NULL;
+  IPropertyStore* devicePropStore = NULL;
+  IPropertyStore* defaultDevicePropStore = NULL;
+
+  WAVEFORMATEX* deviceFormat = NULL;
+  WAVEFORMATEX* closestMatchFormat = NULL;
+
+  // probed
+  info.probed = false;
+
+  // Count capture devices
+  errorText_.clear();
+  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;
+  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device collection.";
+    goto Exit;
+  }
+
+  hr = captureDevices->GetCount( &captureDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device count.";
+    goto Exit;
+  }
+
+  // Count render devices
+  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device collection.";
+    goto Exit;
+  }
+
+  hr = renderDevices->GetCount( &renderDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device count.";
+    goto Exit;
+  }
+
+  // validate device index
+  if ( device >= captureDeviceCount + renderDeviceCount ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Invalid device index.";
+    errorType = RtAudioError::INVALID_USE;
+    goto Exit;
+  }
+
+  // determine whether index falls within capture or render devices
+  if ( device >= renderDeviceCount ) {
+    hr = captureDevices->Item( device - renderDeviceCount, &devicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve capture device handle.";
+      goto Exit;
+    }
+    isCaptureDevice = true;
+  }
+  else {
+    hr = renderDevices->Item( device, &devicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve render device handle.";
+      goto Exit;
+    }
+    isCaptureDevice = false;
+  }
+
+  // get default device name
+  if ( isCaptureDevice ) {
+    hr = deviceEnumerator_->GetDefaultAudioEndpoint( eCapture, eConsole, &defaultDevicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default capture device handle.";
+      goto Exit;
+    }
+  }
+  else {
+    hr = deviceEnumerator_->GetDefaultAudioEndpoint( eRender, eConsole, &defaultDevicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default render device handle.";
+      goto Exit;
+    }
+  }
+
+  hr = defaultDevicePtr->OpenPropertyStore( STGM_READ, &defaultDevicePropStore );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to open default device property store.";
+    goto Exit;
+  }
+  PropVariantInit( &defaultDeviceNameProp );
+
+  hr = defaultDevicePropStore->GetValue( PKEY_Device_FriendlyName, &defaultDeviceNameProp );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve default device property: PKEY_Device_FriendlyName.";
+    goto Exit;
+  }
+
+  defaultDeviceName = convertCharPointerToStdString(defaultDeviceNameProp.pwszVal);
+
+  // name
+  hr = devicePtr->OpenPropertyStore( STGM_READ, &devicePropStore );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to open device property store.";
+    goto Exit;
+  }
+
+  PropVariantInit( &deviceNameProp );
+
+  hr = devicePropStore->GetValue( PKEY_Device_FriendlyName, &deviceNameProp );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device property: PKEY_Device_FriendlyName.";
+    goto Exit;
+  }
+
+  info.name =convertCharPointerToStdString(deviceNameProp.pwszVal);
+
+  // is default
+  if ( isCaptureDevice ) {
+    info.isDefaultInput = info.name == defaultDeviceName;
+    info.isDefaultOutput = false;
+  }
+  else {
+    info.isDefaultInput = false;
+    info.isDefaultOutput = info.name == defaultDeviceName;
+  }
+
+  // channel count
+  hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL, NULL, ( void** ) &audioClient );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device audio client.";
+    goto Exit;
+  }
+
+  hr = audioClient->GetMixFormat( &deviceFormat );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::getDeviceInfo: Unable to retrieve device mix format.";
+    goto Exit;
+  }
+
+  if ( isCaptureDevice ) {
+    info.inputChannels = deviceFormat->nChannels;
+    info.outputChannels = 0;
+    info.duplexChannels = 0;
+  }
+  else {
+    info.inputChannels = 0;
+    info.outputChannels = deviceFormat->nChannels;
+    info.duplexChannels = 0;
+  }
+
+  // sample rates
+  info.sampleRates.clear();
+
+  // allow support for all sample rates as we have a built-in sample rate converter
+  for ( unsigned int i = 0; i < MAX_SAMPLE_RATES; i++ ) {
+    info.sampleRates.push_back( SAMPLE_RATES[i] );
+  }
+  info.preferredSampleRate = deviceFormat->nSamplesPerSec;
+
+  // native format
+  info.nativeFormats = 0;
+
+  if ( deviceFormat->wFormatTag == WAVE_FORMAT_IEEE_FLOAT ||
+       ( deviceFormat->wFormatTag == WAVE_FORMAT_EXTENSIBLE &&
+         ( ( WAVEFORMATEXTENSIBLE* ) deviceFormat )->SubFormat == KSDATAFORMAT_SUBTYPE_IEEE_FLOAT ) )
+  {
+    if ( deviceFormat->wBitsPerSample == 32 ) {
+      info.nativeFormats |= RTAUDIO_FLOAT32;
+    }
+    else if ( deviceFormat->wBitsPerSample == 64 ) {
+      info.nativeFormats |= RTAUDIO_FLOAT64;
+    }
+  }
+  else if ( deviceFormat->wFormatTag == WAVE_FORMAT_PCM ||
+           ( deviceFormat->wFormatTag == WAVE_FORMAT_EXTENSIBLE &&
+             ( ( WAVEFORMATEXTENSIBLE* ) deviceFormat )->SubFormat == KSDATAFORMAT_SUBTYPE_PCM ) )
+  {
+    if ( deviceFormat->wBitsPerSample == 8 ) {
+      info.nativeFormats |= RTAUDIO_SINT8;
+    }
+    else if ( deviceFormat->wBitsPerSample == 16 ) {
+      info.nativeFormats |= RTAUDIO_SINT16;
+    }
+    else if ( deviceFormat->wBitsPerSample == 24 ) {
+      info.nativeFormats |= RTAUDIO_SINT24;
+    }
+    else if ( deviceFormat->wBitsPerSample == 32 ) {
+      info.nativeFormats |= RTAUDIO_SINT32;
+    }
+  }
+
+  // probed
+  info.probed = true;
+
+Exit:
+  // release all references
+  PropVariantClear( &deviceNameProp );
+  PropVariantClear( &defaultDeviceNameProp );
+
+  SAFE_RELEASE( captureDevices );
+  SAFE_RELEASE( renderDevices );
+  SAFE_RELEASE( devicePtr );
+  SAFE_RELEASE( defaultDevicePtr );
+  SAFE_RELEASE( audioClient );
+  SAFE_RELEASE( devicePropStore );
+  SAFE_RELEASE( defaultDevicePropStore );
+
+  CoTaskMemFree( deviceFormat );
+  CoTaskMemFree( closestMatchFormat );
+
+  if ( !errorText_.empty() )
+    error( errorType );
+  return info;
+}
+
+//-----------------------------------------------------------------------------
+
+unsigned int RtApiWasapi::getDefaultOutputDevice( void )
+{
+  for ( unsigned int i = 0; i < getDeviceCount(); i++ ) {
+    if ( getDeviceInfo( i ).isDefaultOutput ) {
+      return i;
+    }
+  }
+
+  return 0;
+}
+
+//-----------------------------------------------------------------------------
+
+unsigned int RtApiWasapi::getDefaultInputDevice( void )
+{
+  for ( unsigned int i = 0; i < getDeviceCount(); i++ ) {
+    if ( getDeviceInfo( i ).isDefaultInput ) {
+      return i;
+    }
+  }
+
+  return 0;
+}
+
+//-----------------------------------------------------------------------------
+
+void RtApiWasapi::closeStream( void )
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiWasapi::closeStream: No open stream to close.";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  if ( stream_.state != STREAM_STOPPED )
+    stopStream();
+
+  // clean up stream memory
+  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient )
+  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient )
+
+  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->captureClient )
+  SAFE_RELEASE( ( ( WasapiHandle* ) stream_.apiHandle )->renderClient )
+
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent )
+    CloseHandle( ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent );
+
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent )
+    CloseHandle( ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent );
+
+  delete ( WasapiHandle* ) stream_.apiHandle;
+  stream_.apiHandle = NULL;
+
+  for ( int i = 0; i < 2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  // update stream state
+  stream_.state = STREAM_CLOSED;
+}
+
+//-----------------------------------------------------------------------------
+
+void RtApiWasapi::startStream( void )
+{
+  verifyStream();
+
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiWasapi::startStream: The stream is already running.";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // update stream state
+  stream_.state = STREAM_RUNNING;
+
+  // create WASAPI stream thread
+  stream_.callbackInfo.thread = ( ThreadHandle ) CreateThread( NULL, 0, runWasapiThread, this, CREATE_SUSPENDED, NULL );
+
+  if ( !stream_.callbackInfo.thread ) {
+    errorText_ = "RtApiWasapi::startStream: Unable to instantiate callback thread.";
+    error( RtAudioError::THREAD_ERROR );
+  }
+  else {
+    SetThreadPriority( ( void* ) stream_.callbackInfo.thread, stream_.callbackInfo.priority );
+    ResumeThread( ( void* ) stream_.callbackInfo.thread );
+  }
+}
+
+//-----------------------------------------------------------------------------
+
+void RtApiWasapi::stopStream( void )
+{
+  verifyStream();
+
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiWasapi::stopStream: The stream is already stopped.";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // inform stream thread by setting stream state to STREAM_STOPPING
+  stream_.state = STREAM_STOPPING;
+
+  // wait until stream thread is stopped
+  while( stream_.state != STREAM_STOPPED ) {
+    Sleep( 1 );
+  }
+
+  // Wait for the last buffer to play before stopping.
+  Sleep( 1000 * stream_.bufferSize / stream_.sampleRate );
+
+  // stop capture client if applicable
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient ) {
+    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient->Stop();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::stopStream: Unable to stop capture stream.";
+      error( RtAudioError::DRIVER_ERROR );
+      return;
+    }
+  }
+
+  // stop render client if applicable
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient ) {
+    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient->Stop();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::stopStream: Unable to stop render stream.";
+      error( RtAudioError::DRIVER_ERROR );
+      return;
+    }
+  }
+
+  // close thread handle
+  if ( stream_.callbackInfo.thread && !CloseHandle( ( void* ) stream_.callbackInfo.thread ) ) {
+    errorText_ = "RtApiWasapi::stopStream: Unable to close callback thread.";
+    error( RtAudioError::THREAD_ERROR );
+    return;
+  }
+
+  stream_.callbackInfo.thread = (ThreadHandle) NULL;
+}
+
+//-----------------------------------------------------------------------------
+
+void RtApiWasapi::abortStream( void )
+{
+  verifyStream();
+
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiWasapi::abortStream: The stream is already stopped.";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // inform stream thread by setting stream state to STREAM_STOPPING
+  stream_.state = STREAM_STOPPING;
+
+  // wait until stream thread is stopped
+  while ( stream_.state != STREAM_STOPPED ) {
+    Sleep( 1 );
+  }
+
+  // stop capture client if applicable
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient ) {
+    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient->Stop();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::abortStream: Unable to stop capture stream.";
+      error( RtAudioError::DRIVER_ERROR );
+      return;
+    }
+  }
+
+  // stop render client if applicable
+  if ( ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient ) {
+    HRESULT hr = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient->Stop();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::abortStream: Unable to stop render stream.";
+      error( RtAudioError::DRIVER_ERROR );
+      return;
+    }
+  }
+
+  // close thread handle
+  if ( stream_.callbackInfo.thread && !CloseHandle( ( void* ) stream_.callbackInfo.thread ) ) {
+    errorText_ = "RtApiWasapi::abortStream: Unable to close callback thread.";
+    error( RtAudioError::THREAD_ERROR );
+    return;
+  }
+
+  stream_.callbackInfo.thread = (ThreadHandle) NULL;
+}
+
+//-----------------------------------------------------------------------------
+
+bool RtApiWasapi::probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                   unsigned int firstChannel, unsigned int sampleRate,
+                                   RtAudioFormat format, unsigned int* bufferSize,
+                                   RtAudio::StreamOptions* options )
+{
+  bool methodResult = FAILURE;
+  unsigned int captureDeviceCount = 0;
+  unsigned int renderDeviceCount = 0;
+
+  IMMDeviceCollection* captureDevices = NULL;
+  IMMDeviceCollection* renderDevices = NULL;
+  IMMDevice* devicePtr = NULL;
+  WAVEFORMATEX* deviceFormat = NULL;
+  unsigned int bufferBytes;
+  stream_.state = STREAM_STOPPED;
+
+  // create API Handle if not already created
+  if ( !stream_.apiHandle )
+    stream_.apiHandle = ( void* ) new WasapiHandle();
+
+  // Count capture devices
+  errorText_.clear();
+  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;
+  HRESULT hr = deviceEnumerator_->EnumAudioEndpoints( eCapture, DEVICE_STATE_ACTIVE, &captureDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device collection.";
+    goto Exit;
+  }
+
+  hr = captureDevices->GetCount( &captureDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device count.";
+    goto Exit;
+  }
+
+  // Count render devices
+  hr = deviceEnumerator_->EnumAudioEndpoints( eRender, DEVICE_STATE_ACTIVE, &renderDevices );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device collection.";
+    goto Exit;
+  }
+
+  hr = renderDevices->GetCount( &renderDeviceCount );
+  if ( FAILED( hr ) ) {
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device count.";
+    goto Exit;
+  }
+
+  // validate device index
+  if ( device >= captureDeviceCount + renderDeviceCount ) {
+    errorType = RtAudioError::INVALID_USE;
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Invalid device index.";
+    goto Exit;
+  }
+
+  // determine whether index falls within capture or render devices
+  if ( device >= renderDeviceCount ) {
+    if ( mode != INPUT ) {
+      errorType = RtAudioError::INVALID_USE;
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Capture device selected as output device.";
+      goto Exit;
+    }
+
+    // retrieve captureAudioClient from devicePtr
+    IAudioClient*& captureAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient;
+
+    hr = captureDevices->Item( device - renderDeviceCount, &devicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve capture device handle.";
+      goto Exit;
+    }
+
+    hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL,
+                              NULL, ( void** ) &captureAudioClient );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device audio client.";
+      goto Exit;
+    }
+
+    hr = captureAudioClient->GetMixFormat( &deviceFormat );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device mix format.";
+      goto Exit;
+    }
+
+    stream_.nDeviceChannels[mode] = deviceFormat->nChannels;
+    captureAudioClient->GetStreamLatency( ( long long* ) &stream_.latency[mode] );
+  }
+  else {
+    if ( mode != OUTPUT ) {
+      errorType = RtAudioError::INVALID_USE;
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Render device selected as input device.";
+      goto Exit;
+    }
+
+    // retrieve renderAudioClient from devicePtr
+    IAudioClient*& renderAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient;
+
+    hr = renderDevices->Item( device, &devicePtr );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve render device handle.";
+      goto Exit;
+    }
+
+    hr = devicePtr->Activate( __uuidof( IAudioClient ), CLSCTX_ALL,
+                              NULL, ( void** ) &renderAudioClient );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device audio client.";
+      goto Exit;
+    }
+
+    hr = renderAudioClient->GetMixFormat( &deviceFormat );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::probeDeviceOpen: Unable to retrieve device mix format.";
+      goto Exit;
+    }
+
+    stream_.nDeviceChannels[mode] = deviceFormat->nChannels;
+    renderAudioClient->GetStreamLatency( ( long long* ) &stream_.latency[mode] );
+  }
+
+  // fill stream data
+  if ( ( stream_.mode == OUTPUT && mode == INPUT ) ||
+       ( stream_.mode == INPUT && mode == OUTPUT ) ) {
+    stream_.mode = DUPLEX;
+  }
+  else {
+    stream_.mode = mode;
+  }
+
+  stream_.device[mode] = device;
+  stream_.doByteSwap[mode] = false;
+  stream_.sampleRate = sampleRate;
+  stream_.bufferSize = *bufferSize;
+  stream_.nBuffers = 1;
+  stream_.nUserChannels[mode] = channels;
+  stream_.channelOffset[mode] = firstChannel;
+  stream_.userFormat = format;
+  stream_.deviceFormat[mode] = getDeviceInfo( device ).nativeFormats;
+
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED )
+    stream_.userInterleaved = false;
+  else
+    stream_.userInterleaved = true;
+  stream_.deviceInterleaved[mode] = true;
+
+  // Set flags for buffer conversion.
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] ||
+       stream_.nUserChannels != stream_.nDeviceChannels )
+    stream_.doConvertBuffer[mode] = true;
+  else if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+            stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  if ( stream_.doConvertBuffer[mode] )
+    setConvertInfo( mode, 0 );
+
+  // Allocate necessary internal buffers
+  bufferBytes = stream_.nUserChannels[mode] * stream_.bufferSize * formatBytes( stream_.userFormat );
+
+  stream_.userBuffer[mode] = ( char* ) calloc( bufferBytes, 1 );
+  if ( !stream_.userBuffer[mode] ) {
+    errorType = RtAudioError::MEMORY_ERROR;
+    errorText_ = "RtApiWasapi::probeDeviceOpen: Error allocating user buffer memory.";
+    goto Exit;
+  }
+
+  if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME )
+    stream_.callbackInfo.priority = 15;
+  else
+    stream_.callbackInfo.priority = 0;
+
+  ///! TODO: RTAUDIO_MINIMIZE_LATENCY // Provide stream buffers directly to callback
+  ///! TODO: RTAUDIO_HOG_DEVICE       // Exclusive mode
+
+  methodResult = SUCCESS;
+
+Exit:
+  //clean up
+  SAFE_RELEASE( captureDevices );
+  SAFE_RELEASE( renderDevices );
+  SAFE_RELEASE( devicePtr );
+  CoTaskMemFree( deviceFormat );
+
+  // if method failed, close the stream
+  if ( methodResult == FAILURE )
+    closeStream();
+
+  if ( !errorText_.empty() )
+    error( errorType );
+  return methodResult;
+}
+
+//=============================================================================
+
+DWORD WINAPI RtApiWasapi::runWasapiThread( void* wasapiPtr )
+{
+  if ( wasapiPtr )
+    ( ( RtApiWasapi* ) wasapiPtr )->wasapiThread();
+
+  return 0;
+}
+
+DWORD WINAPI RtApiWasapi::stopWasapiThread( void* wasapiPtr )
+{
+  if ( wasapiPtr )
+    ( ( RtApiWasapi* ) wasapiPtr )->stopStream();
+
+  return 0;
+}
+
+DWORD WINAPI RtApiWasapi::abortWasapiThread( void* wasapiPtr )
+{
+  if ( wasapiPtr )
+    ( ( RtApiWasapi* ) wasapiPtr )->abortStream();
+
+  return 0;
+}
+
+//-----------------------------------------------------------------------------
+
+void RtApiWasapi::wasapiThread()
+{
+  // as this is a new thread, we must CoInitialize it
+  CoInitialize( NULL );
+
+  HRESULT hr;
+
+  IAudioClient* captureAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureAudioClient;
+  IAudioClient* renderAudioClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderAudioClient;
+  IAudioCaptureClient* captureClient = ( ( WasapiHandle* ) stream_.apiHandle )->captureClient;
+  IAudioRenderClient* renderClient = ( ( WasapiHandle* ) stream_.apiHandle )->renderClient;
+  HANDLE captureEvent = ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent;
+  HANDLE renderEvent = ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent;
+
+  WAVEFORMATEX* captureFormat = NULL;
+  WAVEFORMATEX* renderFormat = NULL;
+  float captureSrRatio = 0.0f;
+  float renderSrRatio = 0.0f;
+  WasapiBuffer captureBuffer;
+  WasapiBuffer renderBuffer;
+
+  // declare local stream variables
+  RtAudioCallback callback = ( RtAudioCallback ) stream_.callbackInfo.callback;
+  BYTE* streamBuffer = NULL;
+  unsigned long captureFlags = 0;
+  unsigned int bufferFrameCount = 0;
+  unsigned int numFramesPadding = 0;
+  unsigned int convBufferSize = 0;
+  bool callbackPushed = false;
+  bool callbackPulled = false;
+  bool callbackStopped = false;
+  int callbackResult = 0;
+
+  // convBuffer is used to store converted buffers between WASAPI and the user
+  char* convBuffer = NULL;
+  unsigned int convBuffSize = 0;
+  unsigned int deviceBuffSize = 0;
+
+  errorText_.clear();
+  RtAudioError::Type errorType = RtAudioError::DRIVER_ERROR;
+
+  // Attempt to assign "Pro Audio" characteristic to thread
+  HMODULE AvrtDll = LoadLibrary( (LPCTSTR) "AVRT.dll" );
+  if ( AvrtDll ) {
+    DWORD taskIndex = 0;
+    TAvSetMmThreadCharacteristicsPtr AvSetMmThreadCharacteristicsPtr = ( TAvSetMmThreadCharacteristicsPtr ) GetProcAddress( AvrtDll, "AvSetMmThreadCharacteristicsW" );
+    AvSetMmThreadCharacteristicsPtr( L"Pro Audio", &taskIndex );
+    FreeLibrary( AvrtDll );
+  }
+
+  // start capture stream if applicable
+  if ( captureAudioClient ) {
+    hr = captureAudioClient->GetMixFormat( &captureFormat );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve device mix format.";
+      goto Exit;
+    }
+
+    captureSrRatio = ( ( float ) captureFormat->nSamplesPerSec / stream_.sampleRate );
+
+    // initialize capture stream according to desire buffer size
+    float desiredBufferSize = stream_.bufferSize * captureSrRatio;
+    REFERENCE_TIME desiredBufferPeriod = ( REFERENCE_TIME ) ( ( float ) desiredBufferSize * 10000000 / captureFormat->nSamplesPerSec );
+
+    if ( !captureClient ) {
+      hr = captureAudioClient->Initialize( AUDCLNT_SHAREMODE_SHARED,
+                                           AUDCLNT_STREAMFLAGS_EVENTCALLBACK,
+                                           desiredBufferPeriod,
+                                           desiredBufferPeriod,
+                                           captureFormat,
+                                           NULL );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to initialize capture audio client.";
+        goto Exit;
+      }
+
+      hr = captureAudioClient->GetService( __uuidof( IAudioCaptureClient ),
+                                           ( void** ) &captureClient );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve capture client handle.";
+        goto Exit;
+      }
+
+      // configure captureEvent to trigger on every available capture buffer
+      captureEvent = CreateEvent( NULL, FALSE, FALSE, NULL );
+      if ( !captureEvent ) {
+        errorType = RtAudioError::SYSTEM_ERROR;
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to create capture event.";
+        goto Exit;
+      }
+
+      hr = captureAudioClient->SetEventHandle( captureEvent );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to set capture event handle.";
+        goto Exit;
+      }
+
+      ( ( WasapiHandle* ) stream_.apiHandle )->captureClient = captureClient;
+      ( ( WasapiHandle* ) stream_.apiHandle )->captureEvent = captureEvent;
+    }
+
+    unsigned int inBufferSize = 0;
+    hr = captureAudioClient->GetBufferSize( &inBufferSize );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to get capture buffer size.";
+      goto Exit;
+    }
+
+    // scale outBufferSize according to stream->user sample rate ratio
+    unsigned int outBufferSize = ( unsigned int ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT];
+    inBufferSize *= stream_.nDeviceChannels[INPUT];
+
+    // set captureBuffer size
+    captureBuffer.setBufferSize( inBufferSize + outBufferSize, formatBytes( stream_.deviceFormat[INPUT] ) );
+
+    // reset the capture stream
+    hr = captureAudioClient->Reset();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to reset capture stream.";
+      goto Exit;
+    }
+
+    // start the capture stream
+    hr = captureAudioClient->Start();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to start capture stream.";
+      goto Exit;
+    }
+  }
+
+  // start render stream if applicable
+  if ( renderAudioClient ) {
+    hr = renderAudioClient->GetMixFormat( &renderFormat );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve device mix format.";
+      goto Exit;
+    }
+
+    renderSrRatio = ( ( float ) renderFormat->nSamplesPerSec / stream_.sampleRate );
+
+    // initialize render stream according to desire buffer size
+    float desiredBufferSize = stream_.bufferSize * renderSrRatio;
+    REFERENCE_TIME desiredBufferPeriod = ( REFERENCE_TIME ) ( ( float ) desiredBufferSize * 10000000 / renderFormat->nSamplesPerSec );
+
+    if ( !renderClient ) {
+      hr = renderAudioClient->Initialize( AUDCLNT_SHAREMODE_SHARED,
+                                          AUDCLNT_STREAMFLAGS_EVENTCALLBACK,
+                                          desiredBufferPeriod,
+                                          desiredBufferPeriod,
+                                          renderFormat,
+                                          NULL );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to initialize render audio client.";
+        goto Exit;
+      }
+
+      hr = renderAudioClient->GetService( __uuidof( IAudioRenderClient ),
+                                          ( void** ) &renderClient );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render client handle.";
+        goto Exit;
+      }
+
+      // configure renderEvent to trigger on every available render buffer
+      renderEvent = CreateEvent( NULL, FALSE, FALSE, NULL );
+      if ( !renderEvent ) {
+        errorType = RtAudioError::SYSTEM_ERROR;
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to create render event.";
+        goto Exit;
+      }
+
+      hr = renderAudioClient->SetEventHandle( renderEvent );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to set render event handle.";
+        goto Exit;
+      }
+
+      ( ( WasapiHandle* ) stream_.apiHandle )->renderClient = renderClient;
+      ( ( WasapiHandle* ) stream_.apiHandle )->renderEvent = renderEvent;
+    }
+
+    unsigned int outBufferSize = 0;
+    hr = renderAudioClient->GetBufferSize( &outBufferSize );
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to get render buffer size.";
+      goto Exit;
+    }
+
+    // scale inBufferSize according to user->stream sample rate ratio
+    unsigned int inBufferSize = ( unsigned int ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT];
+    outBufferSize *= stream_.nDeviceChannels[OUTPUT];
+
+    // set renderBuffer size
+    renderBuffer.setBufferSize( inBufferSize + outBufferSize, formatBytes( stream_.deviceFormat[OUTPUT] ) );
+
+    // reset the render stream
+    hr = renderAudioClient->Reset();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to reset render stream.";
+      goto Exit;
+    }
+
+    // start the render stream
+    hr = renderAudioClient->Start();
+    if ( FAILED( hr ) ) {
+      errorText_ = "RtApiWasapi::wasapiThread: Unable to start render stream.";
+      goto Exit;
+    }
+  }
+
+  if ( stream_.mode == INPUT ) {
+    convBuffSize = ( size_t ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] );
+    deviceBuffSize = stream_.bufferSize * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] );
+  }
+  else if ( stream_.mode == OUTPUT ) {
+    convBuffSize = ( size_t ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] );
+    deviceBuffSize = stream_.bufferSize * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] );
+  }
+  else if ( stream_.mode == DUPLEX ) {
+    convBuffSize = std::max( ( size_t ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] ),
+                             ( size_t ) ( stream_.bufferSize * renderSrRatio ) * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] ) );
+    deviceBuffSize = std::max( stream_.bufferSize * stream_.nDeviceChannels[INPUT] * formatBytes( stream_.deviceFormat[INPUT] ),
+                               stream_.bufferSize * stream_.nDeviceChannels[OUTPUT] * formatBytes( stream_.deviceFormat[OUTPUT] ) );
+  }
+
+  convBuffer = ( char* ) malloc( convBuffSize );
+  stream_.deviceBuffer = ( char* ) malloc( deviceBuffSize );
+  if ( !convBuffer || !stream_.deviceBuffer ) {
+    errorType = RtAudioError::MEMORY_ERROR;
+    errorText_ = "RtApiWasapi::wasapiThread: Error allocating device buffer memory.";
+    goto Exit;
+  }
+
+  // stream process loop
+  while ( stream_.state != STREAM_STOPPING ) {
+    if ( !callbackPulled ) {
+      // Callback Input
+      // ==============
+      // 1. Pull callback buffer from inputBuffer
+      // 2. If 1. was successful: Convert callback buffer to user sample rate and channel count
+      //                          Convert callback buffer to user format
+
+      if ( captureAudioClient ) {
+        // Pull callback buffer from inputBuffer
+        callbackPulled = captureBuffer.pullBuffer( convBuffer,
+                                                   ( unsigned int ) ( stream_.bufferSize * captureSrRatio ) * stream_.nDeviceChannels[INPUT],
+                                                   stream_.deviceFormat[INPUT] );
+
+        if ( callbackPulled ) {
+          // Convert callback buffer to user sample rate
+          convertBufferWasapi( stream_.deviceBuffer,
+                               convBuffer,
+                               stream_.nDeviceChannels[INPUT],
+                               captureFormat->nSamplesPerSec,
+                               stream_.sampleRate,
+                               ( unsigned int ) ( stream_.bufferSize * captureSrRatio ),
+                               convBufferSize,
+                               stream_.deviceFormat[INPUT] );
+
+          if ( stream_.doConvertBuffer[INPUT] ) {
+            // Convert callback buffer to user format
+            convertBuffer( stream_.userBuffer[INPUT],
+                           stream_.deviceBuffer,
+                           stream_.convertInfo[INPUT] );
+          }
+          else {
+            // no further conversion, simple copy deviceBuffer to userBuffer
+            memcpy( stream_.userBuffer[INPUT],
+                    stream_.deviceBuffer,
+                    stream_.bufferSize * stream_.nUserChannels[INPUT] * formatBytes( stream_.userFormat ) );
+          }
+        }
+      }
+      else {
+        // if there is no capture stream, set callbackPulled flag
+        callbackPulled = true;
+      }
+
+      // Execute Callback
+      // ================
+      // 1. Execute user callback method
+      // 2. Handle return value from callback
+
+      // if callback has not requested the stream to stop
+      if ( callbackPulled && !callbackStopped ) {
+        // Execute user callback method
+        callbackResult = callback( stream_.userBuffer[OUTPUT],
+                                   stream_.userBuffer[INPUT],
+                                   stream_.bufferSize,
+                                   getStreamTime(),
+                                   captureFlags & AUDCLNT_BUFFERFLAGS_DATA_DISCONTINUITY ? RTAUDIO_INPUT_OVERFLOW : 0,
+                                   stream_.callbackInfo.userData );
+
+        // Handle return value from callback
+        if ( callbackResult == 1 ) {
+          // instantiate a thread to stop this thread
+          HANDLE threadHandle = CreateThread( NULL, 0, stopWasapiThread, this, 0, NULL );
+          if ( !threadHandle ) {
+            errorType = RtAudioError::THREAD_ERROR;
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to instantiate stream stop thread.";
+            goto Exit;
+          }
+          else if ( !CloseHandle( threadHandle ) ) {
+            errorType = RtAudioError::THREAD_ERROR;
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to close stream stop thread handle.";
+            goto Exit;
+          }
+
+          callbackStopped = true;
+        }
+        else if ( callbackResult == 2 ) {
+          // instantiate a thread to stop this thread
+          HANDLE threadHandle = CreateThread( NULL, 0, abortWasapiThread, this, 0, NULL );
+          if ( !threadHandle ) {
+            errorType = RtAudioError::THREAD_ERROR;
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to instantiate stream abort thread.";
+            goto Exit;
+          }
+          else if ( !CloseHandle( threadHandle ) ) {
+            errorType = RtAudioError::THREAD_ERROR;
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to close stream abort thread handle.";
+            goto Exit;
+          }
+
+          callbackStopped = true;
+        }
+      }
+    }
+
+    // Callback Output
+    // ===============
+    // 1. Convert callback buffer to stream format
+    // 2. Convert callback buffer to stream sample rate and channel count
+    // 3. Push callback buffer into outputBuffer
+
+    if ( renderAudioClient && callbackPulled ) {
+      if ( stream_.doConvertBuffer[OUTPUT] ) {
+        // Convert callback buffer to stream format
+        convertBuffer( stream_.deviceBuffer,
+                       stream_.userBuffer[OUTPUT],
+                       stream_.convertInfo[OUTPUT] );
+
+      }
+
+      // Convert callback buffer to stream sample rate
+      convertBufferWasapi( convBuffer,
+                           stream_.deviceBuffer,
+                           stream_.nDeviceChannels[OUTPUT],
+                           stream_.sampleRate,
+                           renderFormat->nSamplesPerSec,
+                           stream_.bufferSize,
+                           convBufferSize,
+                           stream_.deviceFormat[OUTPUT] );
+
+      // Push callback buffer into outputBuffer
+      callbackPushed = renderBuffer.pushBuffer( convBuffer,
+                                                convBufferSize * stream_.nDeviceChannels[OUTPUT],
+                                                stream_.deviceFormat[OUTPUT] );
+    }
+    else {
+      // if there is no render stream, set callbackPushed flag
+      callbackPushed = true;
+    }
+
+    // Stream Capture
+    // ==============
+    // 1. Get capture buffer from stream
+    // 2. Push capture buffer into inputBuffer
+    // 3. If 2. was successful: Release capture buffer
+
+    if ( captureAudioClient ) {
+      // if the callback input buffer was not pulled from captureBuffer, wait for next capture event
+      if ( !callbackPulled ) {
+        WaitForSingleObject( captureEvent, INFINITE );
+      }
+
+      // Get capture buffer from stream
+      hr = captureClient->GetBuffer( &streamBuffer,
+                                     &bufferFrameCount,
+                                     &captureFlags, NULL, NULL );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve capture buffer.";
+        goto Exit;
+      }
+
+      if ( bufferFrameCount != 0 ) {
+        // Push capture buffer into inputBuffer
+        if ( captureBuffer.pushBuffer( ( char* ) streamBuffer,
+                                       bufferFrameCount * stream_.nDeviceChannels[INPUT],
+                                       stream_.deviceFormat[INPUT] ) )
+        {
+          // Release capture buffer
+          hr = captureClient->ReleaseBuffer( bufferFrameCount );
+          if ( FAILED( hr ) ) {
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";
+            goto Exit;
+          }
+        }
+        else
+        {
+          // Inform WASAPI that capture was unsuccessful
+          hr = captureClient->ReleaseBuffer( 0 );
+          if ( FAILED( hr ) ) {
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";
+            goto Exit;
+          }
+        }
+      }
+      else
+      {
+        // Inform WASAPI that capture was unsuccessful
+        hr = captureClient->ReleaseBuffer( 0 );
+        if ( FAILED( hr ) ) {
+          errorText_ = "RtApiWasapi::wasapiThread: Unable to release capture buffer.";
+          goto Exit;
+        }
+      }
+    }
+
+    // Stream Render
+    // =============
+    // 1. Get render buffer from stream
+    // 2. Pull next buffer from outputBuffer
+    // 3. If 2. was successful: Fill render buffer with next buffer
+    //                          Release render buffer
+
+    if ( renderAudioClient ) {
+      // if the callback output buffer was not pushed to renderBuffer, wait for next render event
+      if ( callbackPulled && !callbackPushed ) {
+        WaitForSingleObject( renderEvent, INFINITE );
+      }
+
+      // Get render buffer from stream
+      hr = renderAudioClient->GetBufferSize( &bufferFrameCount );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer size.";
+        goto Exit;
+      }
+
+      hr = renderAudioClient->GetCurrentPadding( &numFramesPadding );
+      if ( FAILED( hr ) ) {
+        errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer padding.";
+        goto Exit;
+      }
+
+      bufferFrameCount -= numFramesPadding;
+
+      if ( bufferFrameCount != 0 ) {
+        hr = renderClient->GetBuffer( bufferFrameCount, &streamBuffer );
+        if ( FAILED( hr ) ) {
+          errorText_ = "RtApiWasapi::wasapiThread: Unable to retrieve render buffer.";
+          goto Exit;
+        }
+
+        // Pull next buffer from outputBuffer
+        // Fill render buffer with next buffer
+        if ( renderBuffer.pullBuffer( ( char* ) streamBuffer,
+                                      bufferFrameCount * stream_.nDeviceChannels[OUTPUT],
+                                      stream_.deviceFormat[OUTPUT] ) )
+        {
+          // Release render buffer
+          hr = renderClient->ReleaseBuffer( bufferFrameCount, 0 );
+          if ( FAILED( hr ) ) {
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";
+            goto Exit;
+          }
+        }
+        else
+        {
+          // Inform WASAPI that render was unsuccessful
+          hr = renderClient->ReleaseBuffer( 0, 0 );
+          if ( FAILED( hr ) ) {
+            errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";
+            goto Exit;
+          }
+        }
+      }
+      else
+      {
+        // Inform WASAPI that render was unsuccessful
+        hr = renderClient->ReleaseBuffer( 0, 0 );
+        if ( FAILED( hr ) ) {
+          errorText_ = "RtApiWasapi::wasapiThread: Unable to release render buffer.";
+          goto Exit;
+        }
+      }
+    }
+
+    // if the callback buffer was pushed renderBuffer reset callbackPulled flag
+    if ( callbackPushed ) {
+      callbackPulled = false;
+    }
+
+    // tick stream time
+    RtApi::tickStreamTime();
+  }
+
+Exit:
+  // clean up
+  CoTaskMemFree( captureFormat );
+  CoTaskMemFree( renderFormat );
+
+  free ( convBuffer );
+
+  CoUninitialize();
+
+  // update stream state
+  stream_.state = STREAM_STOPPED;
+
+  if ( errorText_.empty() )
+    return;
+  else
+    error( errorType );
+}
+
+//******************** End of __WINDOWS_WASAPI__ *********************//
+#endif
+
+
+#if defined(__WINDOWS_DS__) // Windows DirectSound API
+
+// Modified by Robin Davies, October 2005
+// - Improvements to DirectX pointer chasing. 
+// - Bug fix for non-power-of-two Asio granularity used by Edirol PCR-A30.
+// - Auto-call CoInitialize for DSOUND and ASIO platforms.
+// Various revisions for RtAudio 4.0 by Gary Scavone, April 2007
+// Changed device query structure for RtAudio 4.0.7, January 2010
+
+#include <dsound.h>
+#include <assert.h>
+#include <algorithm>
+
+#if defined(__MINGW32__)
+  // missing from latest mingw winapi
+#define WAVE_FORMAT_96M08 0x00010000 /* 96 kHz, Mono, 8-bit */
+#define WAVE_FORMAT_96S08 0x00020000 /* 96 kHz, Stereo, 8-bit */
+#define WAVE_FORMAT_96M16 0x00040000 /* 96 kHz, Mono, 16-bit */
+#define WAVE_FORMAT_96S16 0x00080000 /* 96 kHz, Stereo, 16-bit */
+#endif
+
+#define MINIMUM_DEVICE_BUFFER_SIZE 32768
+
+#ifdef _MSC_VER // if Microsoft Visual C++
+#pragma comment( lib, "winmm.lib" ) // then, auto-link winmm.lib. Otherwise, it has to be added manually.
+#endif
+
+static inline DWORD dsPointerBetween( DWORD pointer, DWORD laterPointer, DWORD earlierPointer, DWORD bufferSize )
+{
+  if ( pointer > bufferSize ) pointer -= bufferSize;
+  if ( laterPointer < earlierPointer ) laterPointer += bufferSize;
+  if ( pointer < earlierPointer ) pointer += bufferSize;
+  return pointer >= earlierPointer && pointer < laterPointer;
+}
+
+// A structure to hold various information related to the DirectSound
+// API implementation.
+struct DsHandle {
+  unsigned int drainCounter; // Tracks callback counts when draining
+  bool internalDrain;        // Indicates if stop is initiated from callback or not.
+  void *id[2];
+  void *buffer[2];
+  bool xrun[2];
+  UINT bufferPointer[2];  
+  DWORD dsBufferSize[2];
+  DWORD dsPointerLeadTime[2]; // the number of bytes ahead of the safe pointer to lead by.
+  HANDLE condition;
+
+  DsHandle()
+    :drainCounter(0), internalDrain(false) { id[0] = 0; id[1] = 0; buffer[0] = 0; buffer[1] = 0; xrun[0] = false; xrun[1] = false; bufferPointer[0] = 0; bufferPointer[1] = 0; }
+};
+
+// Declarations for utility functions, callbacks, and structures
+// specific to the DirectSound implementation.
+static BOOL CALLBACK deviceQueryCallback( LPGUID lpguid,
+                                          LPCTSTR description,
+                                          LPCTSTR module,
+                                          LPVOID lpContext );
+
+static const char* getErrorString( int code );
+
+static unsigned __stdcall callbackHandler( void *ptr );
+
+struct DsDevice {
+  LPGUID id[2];
+  bool validId[2];
+  bool found;
+  std::string name;
+
+  DsDevice()
+  : found(false) { validId[0] = false; validId[1] = false; }
+};
+
+struct DsProbeData {
+  bool isInput;
+  std::vector<struct DsDevice>* dsDevices;
+};
+
+RtApiDs :: RtApiDs()
+{
+  // Dsound will run both-threaded. If CoInitialize fails, then just
+  // accept whatever the mainline chose for a threading model.
+  coInitialized_ = false;
+  HRESULT hr = CoInitialize( NULL );
+  if ( !FAILED( hr ) ) coInitialized_ = true;
+}
+
+RtApiDs :: ~RtApiDs()
+{
+  if ( coInitialized_ ) CoUninitialize(); // balanced call.
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+}
+
+// The DirectSound default output is always the first device.
+unsigned int RtApiDs :: getDefaultOutputDevice( void )
+{
+  return 0;
+}
+
+// The DirectSound default input is always the first input device,
+// which is the first capture device enumerated.
+unsigned int RtApiDs :: getDefaultInputDevice( void )
+{
+  return 0;
+}
+
+unsigned int RtApiDs :: getDeviceCount( void )
+{
+  // Set query flag for previously found devices to false, so that we
+  // can check for any devices that have disappeared.
+  for ( unsigned int i=0; i<dsDevices.size(); i++ )
+    dsDevices[i].found = false;
+
+  // Query DirectSound devices.
+  struct DsProbeData probeInfo;
+  probeInfo.isInput = false;
+  probeInfo.dsDevices = &dsDevices;
+  HRESULT result = DirectSoundEnumerate( (LPDSENUMCALLBACK) deviceQueryCallback, &probeInfo );
+  if ( FAILED( result ) ) {
+    errorStream_ << "RtApiDs::getDeviceCount: error (" << getErrorString( result ) << ") enumerating output devices!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+  }
+
+  // Query DirectSoundCapture devices.
+  probeInfo.isInput = true;
+  result = DirectSoundCaptureEnumerate( (LPDSENUMCALLBACK) deviceQueryCallback, &probeInfo );
+  if ( FAILED( result ) ) {
+    errorStream_ << "RtApiDs::getDeviceCount: error (" << getErrorString( result ) << ") enumerating input devices!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+  }
+
+  // Clean out any devices that may have disappeared (code update submitted by Eli Zehngut).
+  for ( unsigned int i=0; i<dsDevices.size(); ) {
+    if ( dsDevices[i].found == false ) dsDevices.erase( dsDevices.begin() + i );
+    else i++;
+  }
+
+  return static_cast<unsigned int>(dsDevices.size());
+}
+
+RtAudio::DeviceInfo RtApiDs :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  if ( dsDevices.size() == 0 ) {
+    // Force a query of all devices
+    getDeviceCount();
+    if ( dsDevices.size() == 0 ) {
+      errorText_ = "RtApiDs::getDeviceInfo: no devices found!";
+      error( RtAudioError::INVALID_USE );
+      return info;
+    }
+  }
+
+  if ( device >= dsDevices.size() ) {
+    errorText_ = "RtApiDs::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  HRESULT result;
+  if ( dsDevices[ device ].validId[0] == false ) goto probeInput;
+
+  LPDIRECTSOUND output;
+  DSCAPS outCaps;
+  result = DirectSoundCreate( dsDevices[ device ].id[0], &output, NULL );
+  if ( FAILED( result ) ) {
+    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") opening output device (" << dsDevices[ device ].name << ")!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    goto probeInput;
+  }
+
+  outCaps.dwSize = sizeof( outCaps );
+  result = output->GetCaps( &outCaps );
+  if ( FAILED( result ) ) {
+    output->Release();
+    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") getting capabilities!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    goto probeInput;
+  }
+
+  // Get output channel information.
+  info.outputChannels = ( outCaps.dwFlags & DSCAPS_PRIMARYSTEREO ) ? 2 : 1;
+
+  // Get sample rate information.
+  info.sampleRates.clear();
+  for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {
+    if ( SAMPLE_RATES[k] >= (unsigned int) outCaps.dwMinSecondarySampleRate &&
+         SAMPLE_RATES[k] <= (unsigned int) outCaps.dwMaxSecondarySampleRate ) {
+      info.sampleRates.push_back( SAMPLE_RATES[k] );
+
+      if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )
+        info.preferredSampleRate = SAMPLE_RATES[k];
+    }
+  }
+
+  // Get format information.
+  if ( outCaps.dwFlags & DSCAPS_PRIMARY16BIT ) info.nativeFormats |= RTAUDIO_SINT16;
+  if ( outCaps.dwFlags & DSCAPS_PRIMARY8BIT ) info.nativeFormats |= RTAUDIO_SINT8;
+
+  output->Release();
+
+  if ( getDefaultOutputDevice() == device )
+    info.isDefaultOutput = true;
+
+  if ( dsDevices[ device ].validId[1] == false ) {
+    info.name = dsDevices[ device ].name;
+    info.probed = true;
+    return info;
+  }
+
+ probeInput:
+
+  LPDIRECTSOUNDCAPTURE input;
+  result = DirectSoundCaptureCreate( dsDevices[ device ].id[1], &input, NULL );
+  if ( FAILED( result ) ) {
+    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") opening input device (" << dsDevices[ device ].name << ")!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  DSCCAPS inCaps;
+  inCaps.dwSize = sizeof( inCaps );
+  result = input->GetCaps( &inCaps );
+  if ( FAILED( result ) ) {
+    input->Release();
+    errorStream_ << "RtApiDs::getDeviceInfo: error (" << getErrorString( result ) << ") getting object capabilities (" << dsDevices[ device ].name << ")!";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Get input channel information.
+  info.inputChannels = inCaps.dwChannels;
+
+  // Get sample rate and format information.
+  std::vector<unsigned int> rates;
+  if ( inCaps.dwChannels >= 2 ) {
+    if ( inCaps.dwFormats & WAVE_FORMAT_1S16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_2S16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_4S16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_96S16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_1S08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_2S08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_4S08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_96S08 ) info.nativeFormats |= RTAUDIO_SINT8;
+
+    if ( info.nativeFormats & RTAUDIO_SINT16 ) {
+      if ( inCaps.dwFormats & WAVE_FORMAT_1S16 ) rates.push_back( 11025 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_2S16 ) rates.push_back( 22050 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_4S16 ) rates.push_back( 44100 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_96S16 ) rates.push_back( 96000 );
+    }
+    else if ( info.nativeFormats & RTAUDIO_SINT8 ) {
+      if ( inCaps.dwFormats & WAVE_FORMAT_1S08 ) rates.push_back( 11025 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_2S08 ) rates.push_back( 22050 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_4S08 ) rates.push_back( 44100 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_96S08 ) rates.push_back( 96000 );
+    }
+  }
+  else if ( inCaps.dwChannels == 1 ) {
+    if ( inCaps.dwFormats & WAVE_FORMAT_1M16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_2M16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_4M16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_96M16 ) info.nativeFormats |= RTAUDIO_SINT16;
+    if ( inCaps.dwFormats & WAVE_FORMAT_1M08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_2M08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_4M08 ) info.nativeFormats |= RTAUDIO_SINT8;
+    if ( inCaps.dwFormats & WAVE_FORMAT_96M08 ) info.nativeFormats |= RTAUDIO_SINT8;
+
+    if ( info.nativeFormats & RTAUDIO_SINT16 ) {
+      if ( inCaps.dwFormats & WAVE_FORMAT_1M16 ) rates.push_back( 11025 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_2M16 ) rates.push_back( 22050 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_4M16 ) rates.push_back( 44100 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_96M16 ) rates.push_back( 96000 );
+    }
+    else if ( info.nativeFormats & RTAUDIO_SINT8 ) {
+      if ( inCaps.dwFormats & WAVE_FORMAT_1M08 ) rates.push_back( 11025 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_2M08 ) rates.push_back( 22050 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_4M08 ) rates.push_back( 44100 );
+      if ( inCaps.dwFormats & WAVE_FORMAT_96M08 ) rates.push_back( 96000 );
+    }
+  }
+  else info.inputChannels = 0; // technically, this would be an error
+
+  input->Release();
+
+  if ( info.inputChannels == 0 ) return info;
+
+  // Copy the supported rates to the info structure but avoid duplication.
+  bool found;
+  for ( unsigned int i=0; i<rates.size(); i++ ) {
+    found = false;
+    for ( unsigned int j=0; j<info.sampleRates.size(); j++ ) {
+      if ( rates[i] == info.sampleRates[j] ) {
+        found = true;
+        break;
+      }
+    }
+    if ( found == false ) info.sampleRates.push_back( rates[i] );
+  }
+  std::sort( info.sampleRates.begin(), info.sampleRates.end() );
+
+  // If device opens for both playback and capture, we determine the channels.
+  if ( info.outputChannels > 0 && info.inputChannels > 0 )
+    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+
+  if ( device == 0 ) info.isDefaultInput = true;
+
+  // Copy name and return.
+  info.name = dsDevices[ device ].name;
+  info.probed = true;
+  return info;
+}
+
+bool RtApiDs :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                 unsigned int firstChannel, unsigned int sampleRate,
+                                 RtAudioFormat format, unsigned int *bufferSize,
+                                 RtAudio::StreamOptions *options )
+{
+  if ( channels + firstChannel > 2 ) {
+    errorText_ = "RtApiDs::probeDeviceOpen: DirectSound does not support more than 2 channels per device.";
+    return FAILURE;
+  }
+
+  size_t nDevices = dsDevices.size();
+  if ( nDevices == 0 ) {
+    // This should not happen because a check is made before this function is called.
+    errorText_ = "RtApiDs::probeDeviceOpen: no devices found!";
+    return FAILURE;
+  }
+
+  if ( device >= nDevices ) {
+    // This should not happen because a check is made before this function is called.
+    errorText_ = "RtApiDs::probeDeviceOpen: device ID is invalid!";
+    return FAILURE;
+  }
+
+  if ( mode == OUTPUT ) {
+    if ( dsDevices[ device ].validId[0] == false ) {
+      errorStream_ << "RtApiDs::probeDeviceOpen: device (" << device << ") does not support output!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+  else { // mode == INPUT
+    if ( dsDevices[ device ].validId[1] == false ) {
+      errorStream_ << "RtApiDs::probeDeviceOpen: device (" << device << ") does not support input!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+
+  // According to a note in PortAudio, using GetDesktopWindow()
+  // instead of GetForegroundWindow() is supposed to avoid problems
+  // that occur when the application's window is not the foreground
+  // window.  Also, if the application window closes before the
+  // DirectSound buffer, DirectSound can crash.  In the past, I had
+  // problems when using GetDesktopWindow() but it seems fine now
+  // (January 2010).  I'll leave it commented here.
+  // HWND hWnd = GetForegroundWindow();
+  HWND hWnd = GetDesktopWindow();
+
+  // Check the numberOfBuffers parameter and limit the lowest value to
+  // two.  This is a judgement call and a value of two is probably too
+  // low for capture, but it should work for playback.
+  int nBuffers = 0;
+  if ( options ) nBuffers = options->numberOfBuffers;
+  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) nBuffers = 2;
+  if ( nBuffers < 2 ) nBuffers = 3;
+
+  // Check the lower range of the user-specified buffer size and set
+  // (arbitrarily) to a lower bound of 32.
+  if ( *bufferSize < 32 ) *bufferSize = 32;
+
+  // Create the wave format structure.  The data format setting will
+  // be determined later.
+  WAVEFORMATEX waveFormat;
+  ZeroMemory( &waveFormat, sizeof(WAVEFORMATEX) );
+  waveFormat.wFormatTag = WAVE_FORMAT_PCM;
+  waveFormat.nChannels = channels + firstChannel;
+  waveFormat.nSamplesPerSec = (unsigned long) sampleRate;
+
+  // Determine the device buffer size. By default, we'll use the value
+  // defined above (32K), but we will grow it to make allowances for
+  // very large software buffer sizes.
+  DWORD dsBufferSize = MINIMUM_DEVICE_BUFFER_SIZE;
+  DWORD dsPointerLeadTime = 0;
+
+  void *ohandle = 0, *bhandle = 0;
+  HRESULT result;
+  if ( mode == OUTPUT ) {
+
+    LPDIRECTSOUND output;
+    result = DirectSoundCreate( dsDevices[ device ].id[0], &output, NULL );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") opening output device (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    DSCAPS outCaps;
+    outCaps.dwSize = sizeof( outCaps );
+    result = output->GetCaps( &outCaps );
+    if ( FAILED( result ) ) {
+      output->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting capabilities (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Check channel information.
+    if ( channels + firstChannel == 2 && !( outCaps.dwFlags & DSCAPS_PRIMARYSTEREO ) ) {
+      errorStream_ << "RtApiDs::getDeviceInfo: the output device (" << dsDevices[ device ].name << ") does not support stereo playback.";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Check format information.  Use 16-bit format unless not
+    // supported or user requests 8-bit.
+    if ( outCaps.dwFlags & DSCAPS_PRIMARY16BIT &&
+         !( format == RTAUDIO_SINT8 && outCaps.dwFlags & DSCAPS_PRIMARY8BIT ) ) {
+      waveFormat.wBitsPerSample = 16;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+    }
+    else {
+      waveFormat.wBitsPerSample = 8;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+    }
+    stream_.userFormat = format;
+
+    // Update wave format structure and buffer information.
+    waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;
+    waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;
+    dsPointerLeadTime = nBuffers * (*bufferSize) * (waveFormat.wBitsPerSample / 8) * channels;
+
+    // If the user wants an even bigger buffer, increase the device buffer size accordingly.
+    while ( dsPointerLeadTime * 2U > dsBufferSize )
+      dsBufferSize *= 2;
+
+    // Set cooperative level to DSSCL_EXCLUSIVE ... sound stops when window focus changes.
+    // result = output->SetCooperativeLevel( hWnd, DSSCL_EXCLUSIVE );
+    // Set cooperative level to DSSCL_PRIORITY ... sound remains when window focus changes.
+    result = output->SetCooperativeLevel( hWnd, DSSCL_PRIORITY );
+    if ( FAILED( result ) ) {
+      output->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") setting cooperative level (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Even though we will write to the secondary buffer, we need to
+    // access the primary buffer to set the correct output format
+    // (since the default is 8-bit, 22 kHz!).  Setup the DS primary
+    // buffer description.
+    DSBUFFERDESC bufferDescription;
+    ZeroMemory( &bufferDescription, sizeof( DSBUFFERDESC ) );
+    bufferDescription.dwSize = sizeof( DSBUFFERDESC );
+    bufferDescription.dwFlags = DSBCAPS_PRIMARYBUFFER;
+
+    // Obtain the primary buffer
+    LPDIRECTSOUNDBUFFER buffer;
+    result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );
+    if ( FAILED( result ) ) {
+      output->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") accessing primary buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Set the primary DS buffer sound format.
+    result = buffer->SetFormat( &waveFormat );
+    if ( FAILED( result ) ) {
+      output->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") setting primary buffer format (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Setup the secondary DS buffer description.
+    ZeroMemory( &bufferDescription, sizeof( DSBUFFERDESC ) );
+    bufferDescription.dwSize = sizeof( DSBUFFERDESC );
+    bufferDescription.dwFlags = ( DSBCAPS_STICKYFOCUS |
+                                  DSBCAPS_GLOBALFOCUS |
+                                  DSBCAPS_GETCURRENTPOSITION2 |
+                                  DSBCAPS_LOCHARDWARE );  // Force hardware mixing
+    bufferDescription.dwBufferBytes = dsBufferSize;
+    bufferDescription.lpwfxFormat = &waveFormat;
+
+    // Try to create the secondary DS buffer.  If that doesn't work,
+    // try to use software mixing.  Otherwise, there's a problem.
+    result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );
+    if ( FAILED( result ) ) {
+      bufferDescription.dwFlags = ( DSBCAPS_STICKYFOCUS |
+                                    DSBCAPS_GLOBALFOCUS |
+                                    DSBCAPS_GETCURRENTPOSITION2 |
+                                    DSBCAPS_LOCSOFTWARE );  // Force software mixing
+      result = output->CreateSoundBuffer( &bufferDescription, &buffer, NULL );
+      if ( FAILED( result ) ) {
+        output->Release();
+        errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") creating secondary buffer (" << dsDevices[ device ].name << ")!";
+        errorText_ = errorStream_.str();
+        return FAILURE;
+      }
+    }
+
+    // Get the buffer size ... might be different from what we specified.
+    DSBCAPS dsbcaps;
+    dsbcaps.dwSize = sizeof( DSBCAPS );
+    result = buffer->GetCaps( &dsbcaps );
+    if ( FAILED( result ) ) {
+      output->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting buffer settings (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    dsBufferSize = dsbcaps.dwBufferBytes;
+
+    // Lock the DS buffer
+    LPVOID audioPtr;
+    DWORD dataLen;
+    result = buffer->Lock( 0, dsBufferSize, &audioPtr, &dataLen, NULL, NULL, 0 );
+    if ( FAILED( result ) ) {
+      output->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") locking buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Zero the DS buffer
+    ZeroMemory( audioPtr, dataLen );
+
+    // Unlock the DS buffer
+    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );
+    if ( FAILED( result ) ) {
+      output->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") unlocking buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    ohandle = (void *) output;
+    bhandle = (void *) buffer;
+  }
+
+  if ( mode == INPUT ) {
+
+    LPDIRECTSOUNDCAPTURE input;
+    result = DirectSoundCaptureCreate( dsDevices[ device ].id[1], &input, NULL );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") opening input device (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    DSCCAPS inCaps;
+    inCaps.dwSize = sizeof( inCaps );
+    result = input->GetCaps( &inCaps );
+    if ( FAILED( result ) ) {
+      input->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting input capabilities (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Check channel information.
+    if ( inCaps.dwChannels < channels + firstChannel ) {
+      errorText_ = "RtApiDs::getDeviceInfo: the input device does not support requested input channels.";
+      return FAILURE;
+    }
+
+    // Check format information.  Use 16-bit format unless user
+    // requests 8-bit.
+    DWORD deviceFormats;
+    if ( channels + firstChannel == 2 ) {
+      deviceFormats = WAVE_FORMAT_1S08 | WAVE_FORMAT_2S08 | WAVE_FORMAT_4S08 | WAVE_FORMAT_96S08;
+      if ( format == RTAUDIO_SINT8 && inCaps.dwFormats & deviceFormats ) {
+        waveFormat.wBitsPerSample = 8;
+        stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+      }
+      else { // assume 16-bit is supported
+        waveFormat.wBitsPerSample = 16;
+        stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+      }
+    }
+    else { // channel == 1
+      deviceFormats = WAVE_FORMAT_1M08 | WAVE_FORMAT_2M08 | WAVE_FORMAT_4M08 | WAVE_FORMAT_96M08;
+      if ( format == RTAUDIO_SINT8 && inCaps.dwFormats & deviceFormats ) {
+        waveFormat.wBitsPerSample = 8;
+        stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+      }
+      else { // assume 16-bit is supported
+        waveFormat.wBitsPerSample = 16;
+        stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+      }
+    }
+    stream_.userFormat = format;
+
+    // Update wave format structure and buffer information.
+    waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;
+    waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;
+    dsPointerLeadTime = nBuffers * (*bufferSize) * (waveFormat.wBitsPerSample / 8) * channels;
+
+    // If the user wants an even bigger buffer, increase the device buffer size accordingly.
+    while ( dsPointerLeadTime * 2U > dsBufferSize )
+      dsBufferSize *= 2;
+
+    // Setup the secondary DS buffer description.
+    DSCBUFFERDESC bufferDescription;
+    ZeroMemory( &bufferDescription, sizeof( DSCBUFFERDESC ) );
+    bufferDescription.dwSize = sizeof( DSCBUFFERDESC );
+    bufferDescription.dwFlags = 0;
+    bufferDescription.dwReserved = 0;
+    bufferDescription.dwBufferBytes = dsBufferSize;
+    bufferDescription.lpwfxFormat = &waveFormat;
+
+    // Create the capture buffer.
+    LPDIRECTSOUNDCAPTUREBUFFER buffer;
+    result = input->CreateCaptureBuffer( &bufferDescription, &buffer, NULL );
+    if ( FAILED( result ) ) {
+      input->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") creating input buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Get the buffer size ... might be different from what we specified.
+    DSCBCAPS dscbcaps;
+    dscbcaps.dwSize = sizeof( DSCBCAPS );
+    result = buffer->GetCaps( &dscbcaps );
+    if ( FAILED( result ) ) {
+      input->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") getting buffer settings (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    dsBufferSize = dscbcaps.dwBufferBytes;
+
+    // NOTE: We could have a problem here if this is a duplex stream
+    // and the play and capture hardware buffer sizes are different
+    // (I'm actually not sure if that is a problem or not).
+    // Currently, we are not verifying that.
+
+    // Lock the capture buffer
+    LPVOID audioPtr;
+    DWORD dataLen;
+    result = buffer->Lock( 0, dsBufferSize, &audioPtr, &dataLen, NULL, NULL, 0 );
+    if ( FAILED( result ) ) {
+      input->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") locking input buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    // Zero the buffer
+    ZeroMemory( audioPtr, dataLen );
+
+    // Unlock the buffer
+    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );
+    if ( FAILED( result ) ) {
+      input->Release();
+      buffer->Release();
+      errorStream_ << "RtApiDs::probeDeviceOpen: error (" << getErrorString( result ) << ") unlocking input buffer (" << dsDevices[ device ].name << ")!";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+
+    ohandle = (void *) input;
+    bhandle = (void *) buffer;
+  }
+
+  // Set various stream parameters
+  DsHandle *handle = 0;
+  stream_.nDeviceChannels[mode] = channels + firstChannel;
+  stream_.nUserChannels[mode] = channels;
+  stream_.bufferSize = *bufferSize;
+  stream_.channelOffset[mode] = firstChannel;
+  stream_.deviceInterleaved[mode] = true;
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;
+  else stream_.userInterleaved = true;
+
+  // Set flag for buffer conversion
+  stream_.doConvertBuffer[mode] = false;
+  if (stream_.nUserChannels[mode] != stream_.nDeviceChannels[mode])
+    stream_.doConvertBuffer[mode] = true;
+  if (stream_.userFormat != stream_.deviceFormat[mode])
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+       stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate necessary internal buffers
+  long bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiDs::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( mode == INPUT ) {
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+        if ( bufferBytes <= (long) bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiDs::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  // Allocate our DsHandle structures for the stream.
+  if ( stream_.apiHandle == 0 ) {
+    try {
+      handle = new DsHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiDs::probeDeviceOpen: error allocating AsioHandle memory.";
+      goto error;
+    }
+
+    // Create a manual-reset event.
+    handle->condition = CreateEvent( NULL,   // no security
+                                     TRUE,   // manual-reset
+                                     FALSE,  // non-signaled initially
+                                     NULL ); // unnamed
+    stream_.apiHandle = (void *) handle;
+  }
+  else
+    handle = (DsHandle *) stream_.apiHandle;
+  handle->id[mode] = ohandle;
+  handle->buffer[mode] = bhandle;
+  handle->dsBufferSize[mode] = dsBufferSize;
+  handle->dsPointerLeadTime[mode] = dsPointerLeadTime;
+
+  stream_.device[mode] = device;
+  stream_.state = STREAM_STOPPED;
+  if ( stream_.mode == OUTPUT && mode == INPUT )
+    // We had already set up an output stream.
+    stream_.mode = DUPLEX;
+  else
+    stream_.mode = mode;
+  stream_.nBuffers = nBuffers;
+  stream_.sampleRate = sampleRate;
+
+  // Setup the buffer conversion information structure.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );
+
+  // Setup the callback thread.
+  if ( stream_.callbackInfo.isRunning == false ) {
+    unsigned threadId;
+    stream_.callbackInfo.isRunning = true;
+    stream_.callbackInfo.object = (void *) this;
+    stream_.callbackInfo.thread = _beginthreadex( NULL, 0, &callbackHandler,
+                                                  &stream_.callbackInfo, 0, &threadId );
+    if ( stream_.callbackInfo.thread == 0 ) {
+      errorText_ = "RtApiDs::probeDeviceOpen: error creating callback thread!";
+      goto error;
+    }
+
+    // Boost DS thread priority
+    SetThreadPriority( (HANDLE) stream_.callbackInfo.thread, THREAD_PRIORITY_HIGHEST );
+  }
+  return SUCCESS;
+
+ error:
+  if ( handle ) {
+    if ( handle->buffer[0] ) { // the object pointer can be NULL and valid
+      LPDIRECTSOUND object = (LPDIRECTSOUND) handle->id[0];
+      LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+      if ( buffer ) buffer->Release();
+      object->Release();
+    }
+    if ( handle->buffer[1] ) {
+      LPDIRECTSOUNDCAPTURE object = (LPDIRECTSOUNDCAPTURE) handle->id[1];
+      LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+      if ( buffer ) buffer->Release();
+      object->Release();
+    }
+    CloseHandle( handle->condition );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.state = STREAM_CLOSED;
+  return FAILURE;
+}
+
+void RtApiDs :: closeStream()
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiDs::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // Stop the callback thread.
+  stream_.callbackInfo.isRunning = false;
+  WaitForSingleObject( (HANDLE) stream_.callbackInfo.thread, INFINITE );
+  CloseHandle( (HANDLE) stream_.callbackInfo.thread );
+
+  DsHandle *handle = (DsHandle *) stream_.apiHandle;
+  if ( handle ) {
+    if ( handle->buffer[0] ) { // the object pointer can be NULL and valid
+      LPDIRECTSOUND object = (LPDIRECTSOUND) handle->id[0];
+      LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+      if ( buffer ) {
+        buffer->Stop();
+        buffer->Release();
+      }
+      object->Release();
+    }
+    if ( handle->buffer[1] ) {
+      LPDIRECTSOUNDCAPTURE object = (LPDIRECTSOUNDCAPTURE) handle->id[1];
+      LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+      if ( buffer ) {
+        buffer->Stop();
+        buffer->Release();
+      }
+      object->Release();
+    }
+    CloseHandle( handle->condition );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+void RtApiDs :: startStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiDs::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  DsHandle *handle = (DsHandle *) stream_.apiHandle;
+
+  // Increase scheduler frequency on lesser windows (a side-effect of
+  // increasing timer accuracy).  On greater windows (Win2K or later),
+  // this is already in effect.
+  timeBeginPeriod( 1 ); 
+
+  buffersRolling = false;
+  duplexPrerollBytes = 0;
+
+  if ( stream_.mode == DUPLEX ) {
+    // 0.5 seconds of silence in DUPLEX mode while the devices spin up and synchronize.
+    duplexPrerollBytes = (int) ( 0.5 * stream_.sampleRate * formatBytes( stream_.deviceFormat[1] ) * stream_.nDeviceChannels[1] );
+  }
+
+  HRESULT result = 0;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+    result = buffer->Play( 0, 0, DSBPLAY_LOOPING );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::startStream: error (" << getErrorString( result ) << ") starting output buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+    result = buffer->Start( DSCBSTART_LOOPING );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::startStream: error (" << getErrorString( result ) << ") starting input buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  handle->drainCounter = 0;
+  handle->internalDrain = false;
+  ResetEvent( handle->condition );
+  stream_.state = STREAM_RUNNING;
+
+ unlock:
+  if ( FAILED( result ) ) error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiDs :: stopStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiDs::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  HRESULT result = 0;
+  LPVOID audioPtr;
+  DWORD dataLen;
+  DsHandle *handle = (DsHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    if ( handle->drainCounter == 0 ) {
+      handle->drainCounter = 2;
+      WaitForSingleObject( handle->condition, INFINITE );  // block until signaled
+    }
+
+    stream_.state = STREAM_STOPPED;
+
+    MUTEX_LOCK( &stream_.mutex );
+
+    // Stop the buffer and clear memory
+    LPDIRECTSOUNDBUFFER buffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+    result = buffer->Stop();
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") stopping output buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // Lock the buffer and clear it so that if we start to play again,
+    // we won't have old data playing.
+    result = buffer->Lock( 0, handle->dsBufferSize[0], &audioPtr, &dataLen, NULL, NULL, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") locking output buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // Zero the DS buffer
+    ZeroMemory( audioPtr, dataLen );
+
+    // Unlock the DS buffer
+    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") unlocking output buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // If we start playing again, we must begin at beginning of buffer.
+    handle->bufferPointer[0] = 0;
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+    LPDIRECTSOUNDCAPTUREBUFFER buffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+    audioPtr = NULL;
+    dataLen = 0;
+
+    stream_.state = STREAM_STOPPED;
+
+    if ( stream_.mode != DUPLEX )
+      MUTEX_LOCK( &stream_.mutex );
+
+    result = buffer->Stop();
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") stopping input buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // Lock the buffer and clear it so that if we start to play again,
+    // we won't have old data playing.
+    result = buffer->Lock( 0, handle->dsBufferSize[1], &audioPtr, &dataLen, NULL, NULL, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") locking input buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // Zero the DS buffer
+    ZeroMemory( audioPtr, dataLen );
+
+    // Unlock the DS buffer
+    result = buffer->Unlock( audioPtr, dataLen, NULL, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::stopStream: error (" << getErrorString( result ) << ") unlocking input buffer!";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+
+    // If we start recording again, we must begin at beginning of buffer.
+    handle->bufferPointer[1] = 0;
+  }
+
+ unlock:
+  timeEndPeriod( 1 ); // revert to normal scheduler frequency on lesser windows.
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( FAILED( result ) ) error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiDs :: abortStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiDs::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  DsHandle *handle = (DsHandle *) stream_.apiHandle;
+  handle->drainCounter = 2;
+
+  stopStream();
+}
+
+void RtApiDs :: callbackEvent()
+{
+  if ( stream_.state == STREAM_STOPPED || stream_.state == STREAM_STOPPING ) {
+    Sleep( 50 ); // sleep 50 milliseconds
+    return;
+  }
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiDs::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  CallbackInfo *info = (CallbackInfo *) &stream_.callbackInfo;
+  DsHandle *handle = (DsHandle *) stream_.apiHandle;
+
+  // Check if we were draining the stream and signal is finished.
+  if ( handle->drainCounter > stream_.nBuffers + 2 ) {
+
+    stream_.state = STREAM_STOPPING;
+    if ( handle->internalDrain == false )
+      SetEvent( handle->condition );
+    else
+      stopStream();
+    return;
+  }
+
+  // Invoke user callback to get fresh output data UNLESS we are
+  // draining stream.
+  if ( handle->drainCounter == 0 ) {
+    RtAudioCallback callback = (RtAudioCallback) info->callback;
+    double streamTime = getStreamTime();
+    RtAudioStreamStatus status = 0;
+    if ( stream_.mode != INPUT && handle->xrun[0] == true ) {
+      status |= RTAUDIO_OUTPUT_UNDERFLOW;
+      handle->xrun[0] = false;
+    }
+    if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {
+      status |= RTAUDIO_INPUT_OVERFLOW;
+      handle->xrun[1] = false;
+    }
+    int cbReturnValue = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                                  stream_.bufferSize, streamTime, status, info->userData );
+    if ( cbReturnValue == 2 ) {
+      stream_.state = STREAM_STOPPING;
+      handle->drainCounter = 2;
+      abortStream();
+      return;
+    }
+    else if ( cbReturnValue == 1 ) {
+      handle->drainCounter = 1;
+      handle->internalDrain = true;
+    }
+  }
+
+  HRESULT result;
+  DWORD currentWritePointer, safeWritePointer;
+  DWORD currentReadPointer, safeReadPointer;
+  UINT nextWritePointer;
+
+  LPVOID buffer1 = NULL;
+  LPVOID buffer2 = NULL;
+  DWORD bufferSize1 = 0;
+  DWORD bufferSize2 = 0;
+
+  char *buffer;
+  long bufferBytes;
+
+  MUTEX_LOCK( &stream_.mutex );
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_UNLOCK( &stream_.mutex );
+    return;
+  }
+
+  if ( buffersRolling == false ) {
+    if ( stream_.mode == DUPLEX ) {
+      //assert( handle->dsBufferSize[0] == handle->dsBufferSize[1] );
+
+      // It takes a while for the devices to get rolling. As a result,
+      // there's no guarantee that the capture and write device pointers
+      // will move in lockstep.  Wait here for both devices to start
+      // rolling, and then set our buffer pointers accordingly.
+      // e.g. Crystal Drivers: the capture buffer starts up 5700 to 9600
+      // bytes later than the write buffer.
+
+      // Stub: a serious risk of having a pre-emptive scheduling round
+      // take place between the two GetCurrentPosition calls... but I'm
+      // really not sure how to solve the problem.  Temporarily boost to
+      // Realtime priority, maybe; but I'm not sure what priority the
+      // DirectSound service threads run at. We *should* be roughly
+      // within a ms or so of correct.
+
+      LPDIRECTSOUNDBUFFER dsWriteBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+      LPDIRECTSOUNDCAPTUREBUFFER dsCaptureBuffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+
+      DWORD startSafeWritePointer, startSafeReadPointer;
+
+      result = dsWriteBuffer->GetCurrentPosition( NULL, &startSafeWritePointer );
+      if ( FAILED( result ) ) {
+        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";
+        errorText_ = errorStream_.str();
+        MUTEX_UNLOCK( &stream_.mutex );
+        error( RtAudioError::SYSTEM_ERROR );
+        return;
+      }
+      result = dsCaptureBuffer->GetCurrentPosition( NULL, &startSafeReadPointer );
+      if ( FAILED( result ) ) {
+        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";
+        errorText_ = errorStream_.str();
+        MUTEX_UNLOCK( &stream_.mutex );
+        error( RtAudioError::SYSTEM_ERROR );
+        return;
+      }
+      while ( true ) {
+        result = dsWriteBuffer->GetCurrentPosition( NULL, &safeWritePointer );
+        if ( FAILED( result ) ) {
+          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";
+          errorText_ = errorStream_.str();
+          MUTEX_UNLOCK( &stream_.mutex );
+          error( RtAudioError::SYSTEM_ERROR );
+          return;
+        }
+        result = dsCaptureBuffer->GetCurrentPosition( NULL, &safeReadPointer );
+        if ( FAILED( result ) ) {
+          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";
+          errorText_ = errorStream_.str();
+          MUTEX_UNLOCK( &stream_.mutex );
+          error( RtAudioError::SYSTEM_ERROR );
+          return;
+        }
+        if ( safeWritePointer != startSafeWritePointer && safeReadPointer != startSafeReadPointer ) break;
+        Sleep( 1 );
+      }
+
+      //assert( handle->dsBufferSize[0] == handle->dsBufferSize[1] );
+
+      handle->bufferPointer[0] = safeWritePointer + handle->dsPointerLeadTime[0];
+      if ( handle->bufferPointer[0] >= handle->dsBufferSize[0] ) handle->bufferPointer[0] -= handle->dsBufferSize[0];
+      handle->bufferPointer[1] = safeReadPointer;
+    }
+    else if ( stream_.mode == OUTPUT ) {
+
+      // Set the proper nextWritePosition after initial startup.
+      LPDIRECTSOUNDBUFFER dsWriteBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+      result = dsWriteBuffer->GetCurrentPosition( &currentWritePointer, &safeWritePointer );
+      if ( FAILED( result ) ) {
+        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";
+        errorText_ = errorStream_.str();
+        MUTEX_UNLOCK( &stream_.mutex );
+        error( RtAudioError::SYSTEM_ERROR );
+        return;
+      }
+      handle->bufferPointer[0] = safeWritePointer + handle->dsPointerLeadTime[0];
+      if ( handle->bufferPointer[0] >= handle->dsBufferSize[0] ) handle->bufferPointer[0] -= handle->dsBufferSize[0];
+    }
+
+    buffersRolling = true;
+  }
+
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    
+    LPDIRECTSOUNDBUFFER dsBuffer = (LPDIRECTSOUNDBUFFER) handle->buffer[0];
+
+    if ( handle->drainCounter > 1 ) { // write zeros to the output stream
+      bufferBytes = stream_.bufferSize * stream_.nUserChannels[0];
+      bufferBytes *= formatBytes( stream_.userFormat );
+      memset( stream_.userBuffer[0], 0, bufferBytes );
+    }
+
+    // Setup parameters and do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[0] ) {
+      buffer = stream_.deviceBuffer;
+      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+      bufferBytes = stream_.bufferSize * stream_.nDeviceChannels[0];
+      bufferBytes *= formatBytes( stream_.deviceFormat[0] );
+    }
+    else {
+      buffer = stream_.userBuffer[0];
+      bufferBytes = stream_.bufferSize * stream_.nUserChannels[0];
+      bufferBytes *= formatBytes( stream_.userFormat );
+    }
+
+    // No byte swapping necessary in DirectSound implementation.
+
+    // Ahhh ... windoze.  16-bit data is signed but 8-bit data is
+    // unsigned.  So, we need to convert our signed 8-bit data here to
+    // unsigned.
+    if ( stream_.deviceFormat[0] == RTAUDIO_SINT8 )
+      for ( int i=0; i<bufferBytes; i++ ) buffer[i] = (unsigned char) ( buffer[i] + 128 );
+
+    DWORD dsBufferSize = handle->dsBufferSize[0];
+    nextWritePointer = handle->bufferPointer[0];
+
+    DWORD endWrite, leadPointer;
+    while ( true ) {
+      // Find out where the read and "safe write" pointers are.
+      result = dsBuffer->GetCurrentPosition( &currentWritePointer, &safeWritePointer );
+      if ( FAILED( result ) ) {
+        errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current write position!";
+        errorText_ = errorStream_.str();
+        error( RtAudioError::SYSTEM_ERROR );
+        return;
+      }
+
+      // We will copy our output buffer into the region between
+      // safeWritePointer and leadPointer.  If leadPointer is not
+      // beyond the next endWrite position, wait until it is.
+      leadPointer = safeWritePointer + handle->dsPointerLeadTime[0];
+      //std::cout << "safeWritePointer = " << safeWritePointer << ", leadPointer = " << leadPointer << ", nextWritePointer = " << nextWritePointer << std::endl;
+      if ( leadPointer > dsBufferSize ) leadPointer -= dsBufferSize;
+      if ( leadPointer < nextWritePointer ) leadPointer += dsBufferSize; // unwrap offset
+      endWrite = nextWritePointer + bufferBytes;
+
+      // Check whether the entire write region is behind the play pointer.
+      if ( leadPointer >= endWrite ) break;
+
+      // If we are here, then we must wait until the leadPointer advances
+      // beyond the end of our next write region. We use the
+      // Sleep() function to suspend operation until that happens.
+      double millis = ( endWrite - leadPointer ) * 1000.0;
+      millis /= ( formatBytes( stream_.deviceFormat[0]) * stream_.nDeviceChannels[0] * stream_.sampleRate);
+      if ( millis < 1.0 ) millis = 1.0;
+      Sleep( (DWORD) millis );
+    }
+
+    if ( dsPointerBetween( nextWritePointer, safeWritePointer, currentWritePointer, dsBufferSize )
+         || dsPointerBetween( endWrite, safeWritePointer, currentWritePointer, dsBufferSize ) ) { 
+      // We've strayed into the forbidden zone ... resync the read pointer.
+      handle->xrun[0] = true;
+      nextWritePointer = safeWritePointer + handle->dsPointerLeadTime[0] - bufferBytes;
+      if ( nextWritePointer >= dsBufferSize ) nextWritePointer -= dsBufferSize;
+      handle->bufferPointer[0] = nextWritePointer;
+      endWrite = nextWritePointer + bufferBytes;
+    }
+
+    // Lock free space in the buffer
+    result = dsBuffer->Lock( nextWritePointer, bufferBytes, &buffer1,
+                             &bufferSize1, &buffer2, &bufferSize2, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") locking buffer during playback!";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+
+    // Copy our buffer into the DS buffer
+    CopyMemory( buffer1, buffer, bufferSize1 );
+    if ( buffer2 != NULL ) CopyMemory( buffer2, buffer+bufferSize1, bufferSize2 );
+
+    // Update our buffer offset and unlock sound buffer
+    dsBuffer->Unlock( buffer1, bufferSize1, buffer2, bufferSize2 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") unlocking buffer during playback!";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+    nextWritePointer = ( nextWritePointer + bufferSize1 + bufferSize2 ) % dsBufferSize;
+    handle->bufferPointer[0] = nextWritePointer;
+  }
+
+  // Don't bother draining input
+  if ( handle->drainCounter ) {
+    handle->drainCounter++;
+    goto unlock;
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    // Setup parameters.
+    if ( stream_.doConvertBuffer[1] ) {
+      buffer = stream_.deviceBuffer;
+      bufferBytes = stream_.bufferSize * stream_.nDeviceChannels[1];
+      bufferBytes *= formatBytes( stream_.deviceFormat[1] );
+    }
+    else {
+      buffer = stream_.userBuffer[1];
+      bufferBytes = stream_.bufferSize * stream_.nUserChannels[1];
+      bufferBytes *= formatBytes( stream_.userFormat );
+    }
+
+    LPDIRECTSOUNDCAPTUREBUFFER dsBuffer = (LPDIRECTSOUNDCAPTUREBUFFER) handle->buffer[1];
+    long nextReadPointer = handle->bufferPointer[1];
+    DWORD dsBufferSize = handle->dsBufferSize[1];
+
+    // Find out where the write and "safe read" pointers are.
+    result = dsBuffer->GetCurrentPosition( &currentReadPointer, &safeReadPointer );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+
+    if ( safeReadPointer < (DWORD)nextReadPointer ) safeReadPointer += dsBufferSize; // unwrap offset
+    DWORD endRead = nextReadPointer + bufferBytes;
+
+    // Handling depends on whether we are INPUT or DUPLEX. 
+    // If we're in INPUT mode then waiting is a good thing. If we're in DUPLEX mode,
+    // then a wait here will drag the write pointers into the forbidden zone.
+    // 
+    // In DUPLEX mode, rather than wait, we will back off the read pointer until 
+    // it's in a safe position. This causes dropouts, but it seems to be the only 
+    // practical way to sync up the read and write pointers reliably, given the 
+    // the very complex relationship between phase and increment of the read and write 
+    // pointers.
+    //
+    // In order to minimize audible dropouts in DUPLEX mode, we will
+    // provide a pre-roll period of 0.5 seconds in which we return
+    // zeros from the read buffer while the pointers sync up.
+
+    if ( stream_.mode == DUPLEX ) {
+      if ( safeReadPointer < endRead ) {
+        if ( duplexPrerollBytes <= 0 ) {
+          // Pre-roll time over. Be more agressive.
+          int adjustment = endRead-safeReadPointer;
+
+          handle->xrun[1] = true;
+          // Two cases:
+          //   - large adjustments: we've probably run out of CPU cycles, so just resync exactly,
+          //     and perform fine adjustments later.
+          //   - small adjustments: back off by twice as much.
+          if ( adjustment >= 2*bufferBytes )
+            nextReadPointer = safeReadPointer-2*bufferBytes;
+          else
+            nextReadPointer = safeReadPointer-bufferBytes-adjustment;
+
+          if ( nextReadPointer < 0 ) nextReadPointer += dsBufferSize;
+
+        }
+        else {
+          // In pre=roll time. Just do it.
+          nextReadPointer = safeReadPointer - bufferBytes;
+          while ( nextReadPointer < 0 ) nextReadPointer += dsBufferSize;
+        }
+        endRead = nextReadPointer + bufferBytes;
+      }
+    }
+    else { // mode == INPUT
+      while ( safeReadPointer < endRead && stream_.callbackInfo.isRunning ) {
+        // See comments for playback.
+        double millis = (endRead - safeReadPointer) * 1000.0;
+        millis /= ( formatBytes(stream_.deviceFormat[1]) * stream_.nDeviceChannels[1] * stream_.sampleRate);
+        if ( millis < 1.0 ) millis = 1.0;
+        Sleep( (DWORD) millis );
+
+        // Wake up and find out where we are now.
+        result = dsBuffer->GetCurrentPosition( &currentReadPointer, &safeReadPointer );
+        if ( FAILED( result ) ) {
+          errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") getting current read position!";
+          errorText_ = errorStream_.str();
+          MUTEX_UNLOCK( &stream_.mutex );
+          error( RtAudioError::SYSTEM_ERROR );
+          return;
+        }
+      
+        if ( safeReadPointer < (DWORD)nextReadPointer ) safeReadPointer += dsBufferSize; // unwrap offset
+      }
+    }
+
+    // Lock free space in the buffer
+    result = dsBuffer->Lock( nextReadPointer, bufferBytes, &buffer1,
+                             &bufferSize1, &buffer2, &bufferSize2, 0 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") locking capture buffer!";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+
+    if ( duplexPrerollBytes <= 0 ) {
+      // Copy our buffer into the DS buffer
+      CopyMemory( buffer, buffer1, bufferSize1 );
+      if ( buffer2 != NULL ) CopyMemory( buffer+bufferSize1, buffer2, bufferSize2 );
+    }
+    else {
+      memset( buffer, 0, bufferSize1 );
+      if ( buffer2 != NULL ) memset( buffer + bufferSize1, 0, bufferSize2 );
+      duplexPrerollBytes -= bufferSize1 + bufferSize2;
+    }
+
+    // Update our buffer offset and unlock sound buffer
+    nextReadPointer = ( nextReadPointer + bufferSize1 + bufferSize2 ) % dsBufferSize;
+    dsBuffer->Unlock( buffer1, bufferSize1, buffer2, bufferSize2 );
+    if ( FAILED( result ) ) {
+      errorStream_ << "RtApiDs::callbackEvent: error (" << getErrorString( result ) << ") unlocking capture buffer!";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+    handle->bufferPointer[1] = nextReadPointer;
+
+    // No byte swapping necessary in DirectSound implementation.
+
+    // If necessary, convert 8-bit data from unsigned to signed.
+    if ( stream_.deviceFormat[1] == RTAUDIO_SINT8 )
+      for ( int j=0; j<bufferBytes; j++ ) buffer[j] = (signed char) ( buffer[j] - 128 );
+
+    // Do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[1] )
+      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );
+  }
+
+ unlock:
+  MUTEX_UNLOCK( &stream_.mutex );
+  RtApi::tickStreamTime();
+}
+
+// Definitions for utility functions and callbacks
+// specific to the DirectSound implementation.
+
+static unsigned __stdcall callbackHandler( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiDs *object = (RtApiDs *) info->object;
+  bool* isRunning = &info->isRunning;
+
+  while ( *isRunning == true ) {
+    object->callbackEvent();
+  }
+
+  _endthreadex( 0 );
+  return 0;
+}
+
+static BOOL CALLBACK deviceQueryCallback( LPGUID lpguid,
+                                          LPCTSTR description,
+                                          LPCTSTR /*module*/,
+                                          LPVOID lpContext )
+{
+  struct DsProbeData& probeInfo = *(struct DsProbeData*) lpContext;
+  std::vector<struct DsDevice>& dsDevices = *probeInfo.dsDevices;
+
+  HRESULT hr;
+  bool validDevice = false;
+  if ( probeInfo.isInput == true ) {
+    DSCCAPS caps;
+    LPDIRECTSOUNDCAPTURE object;
+
+    hr = DirectSoundCaptureCreate(  lpguid, &object,   NULL );
+    if ( hr != DS_OK ) return TRUE;
+
+    caps.dwSize = sizeof(caps);
+    hr = object->GetCaps( &caps );
+    if ( hr == DS_OK ) {
+      if ( caps.dwChannels > 0 && caps.dwFormats > 0 )
+        validDevice = true;
+    }
+    object->Release();
+  }
+  else {
+    DSCAPS caps;
+    LPDIRECTSOUND object;
+    hr = DirectSoundCreate(  lpguid, &object,   NULL );
+    if ( hr != DS_OK ) return TRUE;
+
+    caps.dwSize = sizeof(caps);
+    hr = object->GetCaps( &caps );
+    if ( hr == DS_OK ) {
+      if ( caps.dwFlags & DSCAPS_PRIMARYMONO || caps.dwFlags & DSCAPS_PRIMARYSTEREO )
+        validDevice = true;
+    }
+    object->Release();
+  }
+
+  // If good device, then save its name and guid.
+  std::string name = convertCharPointerToStdString( description );
+  //if ( name == "Primary Sound Driver" || name == "Primary Sound Capture Driver" )
+  if ( lpguid == NULL )
+    name = "Default Device";
+  if ( validDevice ) {
+    for ( unsigned int i=0; i<dsDevices.size(); i++ ) {
+      if ( dsDevices[i].name == name ) {
+        dsDevices[i].found = true;
+        if ( probeInfo.isInput ) {
+          dsDevices[i].id[1] = lpguid;
+          dsDevices[i].validId[1] = true;
+        }
+        else {
+          dsDevices[i].id[0] = lpguid;
+          dsDevices[i].validId[0] = true;
+        }
+        return TRUE;
+      }
+    }
+
+    DsDevice device;
+    device.name = name;
+    device.found = true;
+    if ( probeInfo.isInput ) {
+      device.id[1] = lpguid;
+      device.validId[1] = true;
+    }
+    else {
+      device.id[0] = lpguid;
+      device.validId[0] = true;
+    }
+    dsDevices.push_back( device );
+  }
+
+  return TRUE;
+}
+
+static const char* getErrorString( int code )
+{
+  switch ( code ) {
+
+  case DSERR_ALLOCATED:
+    return "Already allocated";
+
+  case DSERR_CONTROLUNAVAIL:
+    return "Control unavailable";
+
+  case DSERR_INVALIDPARAM:
+    return "Invalid parameter";
+
+  case DSERR_INVALIDCALL:
+    return "Invalid call";
+
+  case DSERR_GENERIC:
+    return "Generic error";
+
+  case DSERR_PRIOLEVELNEEDED:
+    return "Priority level needed";
+
+  case DSERR_OUTOFMEMORY:
+    return "Out of memory";
+
+  case DSERR_BADFORMAT:
+    return "The sample rate or the channel format is not supported";
+
+  case DSERR_UNSUPPORTED:
+    return "Not supported";
+
+  case DSERR_NODRIVER:
+    return "No driver";
+
+  case DSERR_ALREADYINITIALIZED:
+    return "Already initialized";
+
+  case DSERR_NOAGGREGATION:
+    return "No aggregation";
+
+  case DSERR_BUFFERLOST:
+    return "Buffer lost";
+
+  case DSERR_OTHERAPPHASPRIO:
+    return "Another application already has priority";
+
+  case DSERR_UNINITIALIZED:
+    return "Uninitialized";
+
+  default:
+    return "DirectSound unknown error";
+  }
+}
+//******************** End of __WINDOWS_DS__ *********************//
+#endif
+
+
+#if defined(__LINUX_ALSA__)
+
+#include <alsa/asoundlib.h>
+#include <unistd.h>
+
+  // A structure to hold various information related to the ALSA API
+  // implementation.
+struct AlsaHandle {
+  snd_pcm_t *handles[2];
+  bool synchronized;
+  bool xrun[2];
+  pthread_cond_t runnable_cv;
+  bool runnable;
+
+  AlsaHandle()
+    :synchronized(false), runnable(false) { xrun[0] = false; xrun[1] = false; }
+};
+
+static void *alsaCallbackHandler( void * ptr );
+
+RtApiAlsa :: RtApiAlsa()
+{
+  // Nothing to do here.
+}
+
+RtApiAlsa :: ~RtApiAlsa()
+{
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+}
+
+unsigned int RtApiAlsa :: getDeviceCount( void )
+{
+  unsigned nDevices = 0;
+  int result, subdevice, card;
+  char name[64];
+  snd_ctl_t *handle;
+
+  // Count cards and devices
+  card = -1;
+  snd_card_next( &card );
+  while ( card >= 0 ) {
+    sprintf( name, "hw:%d", card );
+    result = snd_ctl_open( &handle, name, 0 );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::getDeviceCount: control open, card = " << card << ", " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      error( RtAudioError::WARNING );
+      goto nextcard;
+    }
+    subdevice = -1;
+    while( 1 ) {
+      result = snd_ctl_pcm_next_device( handle, &subdevice );
+      if ( result < 0 ) {
+        errorStream_ << "RtApiAlsa::getDeviceCount: control next device, card = " << card << ", " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+        error( RtAudioError::WARNING );
+        break;
+      }
+      if ( subdevice < 0 )
+        break;
+      nDevices++;
+    }
+  nextcard:
+    snd_ctl_close( handle );
+    snd_card_next( &card );
+  }
+
+  result = snd_ctl_open( &handle, "default", 0 );
+  if (result == 0) {
+    nDevices++;
+    snd_ctl_close( handle );
+  }
+
+  return nDevices;
+}
+
+RtAudio::DeviceInfo RtApiAlsa :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  unsigned nDevices = 0;
+  int result, subdevice, card;
+  char name[64];
+  snd_ctl_t *chandle;
+
+  // Count cards and devices
+  card = -1;
+  subdevice = -1;
+  snd_card_next( &card );
+  while ( card >= 0 ) {
+    sprintf( name, "hw:%d", card );
+    result = snd_ctl_open( &chandle, name, SND_CTL_NONBLOCK );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::getDeviceInfo: control open, card = " << card << ", " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      error( RtAudioError::WARNING );
+      goto nextcard;
+    }
+    subdevice = -1;
+    while( 1 ) {
+      result = snd_ctl_pcm_next_device( chandle, &subdevice );
+      if ( result < 0 ) {
+        errorStream_ << "RtApiAlsa::getDeviceInfo: control next device, card = " << card << ", " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+        error( RtAudioError::WARNING );
+        break;
+      }
+      if ( subdevice < 0 ) break;
+      if ( nDevices == device ) {
+        sprintf( name, "hw:%d,%d", card, subdevice );
+        goto foundDevice;
+      }
+      nDevices++;
+    }
+  nextcard:
+    snd_ctl_close( chandle );
+    snd_card_next( &card );
+  }
+
+  result = snd_ctl_open( &chandle, "default", SND_CTL_NONBLOCK );
+  if ( result == 0 ) {
+    if ( nDevices == device ) {
+      strcpy( name, "default" );
+      goto foundDevice;
+    }
+    nDevices++;
+  }
+
+  if ( nDevices == 0 ) {
+    errorText_ = "RtApiAlsa::getDeviceInfo: no devices found!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  if ( device >= nDevices ) {
+    errorText_ = "RtApiAlsa::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+ foundDevice:
+
+  // If a stream is already open, we cannot probe the stream devices.
+  // Thus, use the saved results.
+  if ( stream_.state != STREAM_CLOSED &&
+       ( stream_.device[0] == device || stream_.device[1] == device ) ) {
+    snd_ctl_close( chandle );
+    if ( device >= devices_.size() ) {
+      errorText_ = "RtApiAlsa::getDeviceInfo: device ID was not present before stream was opened.";
+      error( RtAudioError::WARNING );
+      return info;
+    }
+    return devices_[ device ];
+  }
+
+  int openMode = SND_PCM_ASYNC;
+  snd_pcm_stream_t stream;
+  snd_pcm_info_t *pcminfo;
+  snd_pcm_info_alloca( &pcminfo );
+  snd_pcm_t *phandle;
+  snd_pcm_hw_params_t *params;
+  snd_pcm_hw_params_alloca( &params );
+
+  // First try for playback unless default device (which has subdev -1)
+  stream = SND_PCM_STREAM_PLAYBACK;
+  snd_pcm_info_set_stream( pcminfo, stream );
+  if ( subdevice != -1 ) {
+    snd_pcm_info_set_device( pcminfo, subdevice );
+    snd_pcm_info_set_subdevice( pcminfo, 0 );
+
+    result = snd_ctl_pcm_info( chandle, pcminfo );
+    if ( result < 0 ) {
+      // Device probably doesn't support playback.
+      goto captureProbe;
+    }
+  }
+
+  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK );
+  if ( result < 0 ) {
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    goto captureProbe;
+  }
+
+  // The device is open ... fill the parameter structure.
+  result = snd_pcm_hw_params_any( phandle, params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    goto captureProbe;
+  }
+
+  // Get output channel information.
+  unsigned int value;
+  result = snd_pcm_hw_params_get_channels_max( params, &value );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: error getting device (" << name << ") output channels, " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    goto captureProbe;
+  }
+  info.outputChannels = value;
+  snd_pcm_close( phandle );
+
+ captureProbe:
+  stream = SND_PCM_STREAM_CAPTURE;
+  snd_pcm_info_set_stream( pcminfo, stream );
+
+  // Now try for capture unless default device (with subdev = -1)
+  if ( subdevice != -1 ) {
+    result = snd_ctl_pcm_info( chandle, pcminfo );
+    snd_ctl_close( chandle );
+    if ( result < 0 ) {
+      // Device probably doesn't support capture.
+      if ( info.outputChannels == 0 ) return info;
+      goto probeParameters;
+    }
+  }
+  else
+    snd_ctl_close( chandle );
+
+  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK);
+  if ( result < 0 ) {
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    if ( info.outputChannels == 0 ) return info;
+    goto probeParameters;
+  }
+
+  // The device is open ... fill the parameter structure.
+  result = snd_pcm_hw_params_any( phandle, params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    if ( info.outputChannels == 0 ) return info;
+    goto probeParameters;
+  }
+
+  result = snd_pcm_hw_params_get_channels_max( params, &value );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: error getting device (" << name << ") input channels, " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    if ( info.outputChannels == 0 ) return info;
+    goto probeParameters;
+  }
+  info.inputChannels = value;
+  snd_pcm_close( phandle );
+
+  // If device opens for both playback and capture, we determine the channels.
+  if ( info.outputChannels > 0 && info.inputChannels > 0 )
+    info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+
+  // ALSA doesn't provide default devices so we'll use the first available one.
+  if ( device == 0 && info.outputChannels > 0 )
+    info.isDefaultOutput = true;
+  if ( device == 0 && info.inputChannels > 0 )
+    info.isDefaultInput = true;
+
+ probeParameters:
+  // At this point, we just need to figure out the supported data
+  // formats and sample rates.  We'll proceed by opening the device in
+  // the direction with the maximum number of channels, or playback if
+  // they are equal.  This might limit our sample rate options, but so
+  // be it.
+
+  if ( info.outputChannels >= info.inputChannels )
+    stream = SND_PCM_STREAM_PLAYBACK;
+  else
+    stream = SND_PCM_STREAM_CAPTURE;
+  snd_pcm_info_set_stream( pcminfo, stream );
+
+  result = snd_pcm_open( &phandle, name, stream, openMode | SND_PCM_NONBLOCK);
+  if ( result < 0 ) {
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_open error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // The device is open ... fill the parameter structure.
+  result = snd_pcm_hw_params_any( phandle, params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: snd_pcm_hw_params error for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Test our discrete set of sample rate values.
+  info.sampleRates.clear();
+  for ( unsigned int i=0; i<MAX_SAMPLE_RATES; i++ ) {
+    if ( snd_pcm_hw_params_test_rate( phandle, params, SAMPLE_RATES[i], 0 ) == 0 ) {
+      info.sampleRates.push_back( SAMPLE_RATES[i] );
+
+      if ( !info.preferredSampleRate || ( SAMPLE_RATES[i] <= 48000 && SAMPLE_RATES[i] > info.preferredSampleRate ) )
+        info.preferredSampleRate = SAMPLE_RATES[i];
+    }
+  }
+  if ( info.sampleRates.size() == 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: no supported sample rates found for device (" << name << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Probe the supported data formats ... we don't care about endian-ness just yet
+  snd_pcm_format_t format;
+  info.nativeFormats = 0;
+  format = SND_PCM_FORMAT_S8;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_SINT8;
+  format = SND_PCM_FORMAT_S16;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_SINT16;
+  format = SND_PCM_FORMAT_S24;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_SINT24;
+  format = SND_PCM_FORMAT_S32;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_SINT32;
+  format = SND_PCM_FORMAT_FLOAT;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_FLOAT32;
+  format = SND_PCM_FORMAT_FLOAT64;
+  if ( snd_pcm_hw_params_test_format( phandle, params, format ) == 0 )
+    info.nativeFormats |= RTAUDIO_FLOAT64;
+
+  // Check that we have at least one supported format
+  if ( info.nativeFormats == 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::getDeviceInfo: pcm device (" << name << ") data format not supported by RtAudio.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Get the device name
+  char *cardname;
+  result = snd_card_get_name( card, &cardname );
+  if ( result >= 0 ) {
+    sprintf( name, "hw:%s,%d", cardname, subdevice );
+    free( cardname );
+  }
+  info.name = name;
+
+  // That's all ... close the device and return
+  snd_pcm_close( phandle );
+  info.probed = true;
+  return info;
+}
+
+void RtApiAlsa :: saveDeviceInfo( void )
+{
+  devices_.clear();
+
+  unsigned int nDevices = getDeviceCount();
+  devices_.resize( nDevices );
+  for ( unsigned int i=0; i<nDevices; i++ )
+    devices_[i] = getDeviceInfo( i );
+}
+
+bool RtApiAlsa :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                   unsigned int firstChannel, unsigned int sampleRate,
+                                   RtAudioFormat format, unsigned int *bufferSize,
+                                   RtAudio::StreamOptions *options )
+
+{
+#if defined(__RTAUDIO_DEBUG__)
+  snd_output_t *out;
+  snd_output_stdio_attach(&out, stderr, 0);
+#endif
+
+  // I'm not using the "plug" interface ... too much inconsistent behavior.
+
+  unsigned nDevices = 0;
+  int result, subdevice, card;
+  char name[64];
+  snd_ctl_t *chandle;
+
+  if ( options && options->flags & RTAUDIO_ALSA_USE_DEFAULT )
+    snprintf(name, sizeof(name), "%s", "default");
+  else {
+    // Count cards and devices
+    card = -1;
+    snd_card_next( &card );
+    while ( card >= 0 ) {
+      sprintf( name, "hw:%d", card );
+      result = snd_ctl_open( &chandle, name, SND_CTL_NONBLOCK );
+      if ( result < 0 ) {
+        errorStream_ << "RtApiAlsa::probeDeviceOpen: control open, card = " << card << ", " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+        return FAILURE;
+      }
+      subdevice = -1;
+      while( 1 ) {
+        result = snd_ctl_pcm_next_device( chandle, &subdevice );
+        if ( result < 0 ) break;
+        if ( subdevice < 0 ) break;
+        if ( nDevices == device ) {
+          sprintf( name, "hw:%d,%d", card, subdevice );
+          snd_ctl_close( chandle );
+          goto foundDevice;
+        }
+        nDevices++;
+      }
+      snd_ctl_close( chandle );
+      snd_card_next( &card );
+    }
+
+    result = snd_ctl_open( &chandle, "default", SND_CTL_NONBLOCK );
+    if ( result == 0 ) {
+      if ( nDevices == device ) {
+        strcpy( name, "default" );
+        goto foundDevice;
+      }
+      nDevices++;
+    }
+
+    if ( nDevices == 0 ) {
+      // This should not happen because a check is made before this function is called.
+      errorText_ = "RtApiAlsa::probeDeviceOpen: no devices found!";
+      return FAILURE;
+    }
+
+    if ( device >= nDevices ) {
+      // This should not happen because a check is made before this function is called.
+      errorText_ = "RtApiAlsa::probeDeviceOpen: device ID is invalid!";
+      return FAILURE;
+    }
+  }
+
+ foundDevice:
+
+  // The getDeviceInfo() function will not work for a device that is
+  // already open.  Thus, we'll probe the system before opening a
+  // stream and save the results for use by getDeviceInfo().
+  if ( mode == OUTPUT || ( mode == INPUT && stream_.mode != OUTPUT ) ) // only do once
+    this->saveDeviceInfo();
+
+  snd_pcm_stream_t stream;
+  if ( mode == OUTPUT )
+    stream = SND_PCM_STREAM_PLAYBACK;
+  else
+    stream = SND_PCM_STREAM_CAPTURE;
+
+  snd_pcm_t *phandle;
+  int openMode = SND_PCM_ASYNC;
+  result = snd_pcm_open( &phandle, name, stream, openMode );
+  if ( result < 0 ) {
+    if ( mode == OUTPUT )
+      errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device (" << name << ") won't open for output.";
+    else
+      errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device (" << name << ") won't open for input.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Fill the parameter structure.
+  snd_pcm_hw_params_t *hw_params;
+  snd_pcm_hw_params_alloca( &hw_params );
+  result = snd_pcm_hw_params_any( phandle, hw_params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting pcm device (" << name << ") parameters, " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+#if defined(__RTAUDIO_DEBUG__)
+  fprintf( stderr, "\nRtApiAlsa: dump hardware params just after device open:\n\n" );
+  snd_pcm_hw_params_dump( hw_params, out );
+#endif
+
+  // Set access ... check user preference.
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) {
+    stream_.userInterleaved = false;
+    result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED );
+    if ( result < 0 ) {
+      result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_INTERLEAVED );
+      stream_.deviceInterleaved[mode] =  true;
+    }
+    else
+      stream_.deviceInterleaved[mode] = false;
+  }
+  else {
+    stream_.userInterleaved = true;
+    result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_INTERLEAVED );
+    if ( result < 0 ) {
+      result = snd_pcm_hw_params_set_access( phandle, hw_params, SND_PCM_ACCESS_RW_NONINTERLEAVED );
+      stream_.deviceInterleaved[mode] =  false;
+    }
+    else
+      stream_.deviceInterleaved[mode] =  true;
+  }
+
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting pcm device (" << name << ") access, " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Determine how to set the device format.
+  stream_.userFormat = format;
+  snd_pcm_format_t deviceFormat = SND_PCM_FORMAT_UNKNOWN;
+
+  if ( format == RTAUDIO_SINT8 )
+    deviceFormat = SND_PCM_FORMAT_S8;
+  else if ( format == RTAUDIO_SINT16 )
+    deviceFormat = SND_PCM_FORMAT_S16;
+  else if ( format == RTAUDIO_SINT24 )
+    deviceFormat = SND_PCM_FORMAT_S24;
+  else if ( format == RTAUDIO_SINT32 )
+    deviceFormat = SND_PCM_FORMAT_S32;
+  else if ( format == RTAUDIO_FLOAT32 )
+    deviceFormat = SND_PCM_FORMAT_FLOAT;
+  else if ( format == RTAUDIO_FLOAT64 )
+    deviceFormat = SND_PCM_FORMAT_FLOAT64;
+
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat) == 0) {
+    stream_.deviceFormat[mode] = format;
+    goto setFormat;
+  }
+
+  // The user requested format is not natively supported by the device.
+  deviceFormat = SND_PCM_FORMAT_FLOAT64;
+  if ( snd_pcm_hw_params_test_format( phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_FLOAT64;
+    goto setFormat;
+  }
+
+  deviceFormat = SND_PCM_FORMAT_FLOAT;
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;
+    goto setFormat;
+  }
+
+  deviceFormat = SND_PCM_FORMAT_S32;
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+    goto setFormat;
+  }
+
+  deviceFormat = SND_PCM_FORMAT_S24;
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+    goto setFormat;
+  }
+
+  deviceFormat = SND_PCM_FORMAT_S16;
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+    goto setFormat;
+  }
+
+  deviceFormat = SND_PCM_FORMAT_S8;
+  if ( snd_pcm_hw_params_test_format(phandle, hw_params, deviceFormat ) == 0 ) {
+    stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+    goto setFormat;
+  }
+
+  // If we get here, no supported format was found.
+  snd_pcm_close( phandle );
+  errorStream_ << "RtApiAlsa::probeDeviceOpen: pcm device " << device << " data format not supported by RtAudio.";
+  errorText_ = errorStream_.str();
+  return FAILURE;
+
+ setFormat:
+  result = snd_pcm_hw_params_set_format( phandle, hw_params, deviceFormat );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting pcm device (" << name << ") data format, " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Determine whether byte-swaping is necessary.
+  stream_.doByteSwap[mode] = false;
+  if ( deviceFormat != SND_PCM_FORMAT_S8 ) {
+    result = snd_pcm_format_cpu_endian( deviceFormat );
+    if ( result == 0 )
+      stream_.doByteSwap[mode] = true;
+    else if (result < 0) {
+      snd_pcm_close( phandle );
+      errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting pcm device (" << name << ") endian-ness, " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      return FAILURE;
+    }
+  }
+
+  // Set the sample rate.
+  result = snd_pcm_hw_params_set_rate_near( phandle, hw_params, (unsigned int*) &sampleRate, 0 );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting sample rate on device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Determine the number of channels for this device.  We support a possible
+  // minimum device channel number > than the value requested by the user.
+  stream_.nUserChannels[mode] = channels;
+  unsigned int value;
+  result = snd_pcm_hw_params_get_channels_max( hw_params, &value );
+  unsigned int deviceChannels = value;
+  if ( result < 0 || deviceChannels < channels + firstChannel ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: requested channel parameters not supported by device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  result = snd_pcm_hw_params_get_channels_min( hw_params, &value );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error getting minimum channels for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  deviceChannels = value;
+  if ( deviceChannels < channels + firstChannel ) deviceChannels = channels + firstChannel;
+  stream_.nDeviceChannels[mode] = deviceChannels;
+
+  // Set the device channels.
+  result = snd_pcm_hw_params_set_channels( phandle, hw_params, deviceChannels );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting channels for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Set the buffer (or period) size.
+  int dir = 0;
+  snd_pcm_uframes_t periodSize = *bufferSize;
+  result = snd_pcm_hw_params_set_period_size_near( phandle, hw_params, &periodSize, &dir );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting period size for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  *bufferSize = periodSize;
+
+  // Set the buffer number, which in ALSA is referred to as the "period".
+  unsigned int periods = 0;
+  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) periods = 2;
+  if ( options && options->numberOfBuffers > 0 ) periods = options->numberOfBuffers;
+  if ( periods < 2 ) periods = 4; // a fairly safe default value
+  result = snd_pcm_hw_params_set_periods_near( phandle, hw_params, &periods, &dir );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error setting periods for device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // If attempting to setup a duplex stream, the bufferSize parameter
+  // MUST be the same in both directions!
+  if ( stream_.mode == OUTPUT && mode == INPUT && *bufferSize != stream_.bufferSize ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: system error setting buffer size for duplex stream on device (" << name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  stream_.bufferSize = *bufferSize;
+
+  // Install the hardware configuration
+  result = snd_pcm_hw_params( phandle, hw_params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error installing hardware configuration on device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+#if defined(__RTAUDIO_DEBUG__)
+  fprintf(stderr, "\nRtApiAlsa: dump hardware params after installation:\n\n");
+  snd_pcm_hw_params_dump( hw_params, out );
+#endif
+
+  // Set the software configuration to fill buffers with zeros and prevent device stopping on xruns.
+  snd_pcm_sw_params_t *sw_params = NULL;
+  snd_pcm_sw_params_alloca( &sw_params );
+  snd_pcm_sw_params_current( phandle, sw_params );
+  snd_pcm_sw_params_set_start_threshold( phandle, sw_params, *bufferSize );
+  snd_pcm_sw_params_set_stop_threshold( phandle, sw_params, ULONG_MAX );
+  snd_pcm_sw_params_set_silence_threshold( phandle, sw_params, 0 );
+
+  // The following two settings were suggested by Theo Veenker
+  //snd_pcm_sw_params_set_avail_min( phandle, sw_params, *bufferSize );
+  //snd_pcm_sw_params_set_xfer_align( phandle, sw_params, 1 );
+
+  // here are two options for a fix
+  //snd_pcm_sw_params_set_silence_size( phandle, sw_params, ULONG_MAX );
+  snd_pcm_uframes_t val;
+  snd_pcm_sw_params_get_boundary( sw_params, &val );
+  snd_pcm_sw_params_set_silence_size( phandle, sw_params, val );
+
+  result = snd_pcm_sw_params( phandle, sw_params );
+  if ( result < 0 ) {
+    snd_pcm_close( phandle );
+    errorStream_ << "RtApiAlsa::probeDeviceOpen: error installing software configuration on device (" << name << "), " << snd_strerror( result ) << ".";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+#if defined(__RTAUDIO_DEBUG__)
+  fprintf(stderr, "\nRtApiAlsa: dump software params after installation:\n\n");
+  snd_pcm_sw_params_dump( sw_params, out );
+#endif
+
+  // Set flags for buffer conversion
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+       stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate the ApiHandle if necessary and then save.
+  AlsaHandle *apiInfo = 0;
+  if ( stream_.apiHandle == 0 ) {
+    try {
+      apiInfo = (AlsaHandle *) new AlsaHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating AlsaHandle memory.";
+      goto error;
+    }
+
+    if ( pthread_cond_init( &apiInfo->runnable_cv, NULL ) ) {
+      errorText_ = "RtApiAlsa::probeDeviceOpen: error initializing pthread condition variable.";
+      goto error;
+    }
+
+    stream_.apiHandle = (void *) apiInfo;
+    apiInfo->handles[0] = 0;
+    apiInfo->handles[1] = 0;
+  }
+  else {
+    apiInfo = (AlsaHandle *) stream_.apiHandle;
+  }
+  apiInfo->handles[mode] = phandle;
+  phandle = 0;
+
+  // Allocate necessary internal buffers.
+  unsigned long bufferBytes;
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( mode == INPUT ) {
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+        if ( bufferBytes <= bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiAlsa::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  stream_.sampleRate = sampleRate;
+  stream_.nBuffers = periods;
+  stream_.device[mode] = device;
+  stream_.state = STREAM_STOPPED;
+
+  // Setup the buffer conversion information structure.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );
+
+  // Setup thread if necessary.
+  if ( stream_.mode == OUTPUT && mode == INPUT ) {
+    // We had already set up an output stream.
+    stream_.mode = DUPLEX;
+    // Link the streams if possible.
+    apiInfo->synchronized = false;
+    if ( snd_pcm_link( apiInfo->handles[0], apiInfo->handles[1] ) == 0 )
+      apiInfo->synchronized = true;
+    else {
+      errorText_ = "RtApiAlsa::probeDeviceOpen: unable to synchronize input and output devices.";
+      error( RtAudioError::WARNING );
+    }
+  }
+  else {
+    stream_.mode = mode;
+
+    // Setup callback thread.
+    stream_.callbackInfo.object = (void *) this;
+
+    // Set the thread attributes for joinable and realtime scheduling
+    // priority (optional).  The higher priority will only take affect
+    // if the program is run as root or suid. Note, under Linux
+    // processes with CAP_SYS_NICE privilege, a user can change
+    // scheduling policy and priority (thus need not be root). See
+    // POSIX "capabilities".
+    pthread_attr_t attr;
+    pthread_attr_init( &attr );
+    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE );
+
+#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)
+    if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME ) {
+      // We previously attempted to increase the audio callback priority
+      // to SCHED_RR here via the attributes.  However, while no errors
+      // were reported in doing so, it did not work.  So, now this is
+      // done in the alsaCallbackHandler function.
+      stream_.callbackInfo.doRealtime = true;
+      int priority = options->priority;
+      int min = sched_get_priority_min( SCHED_RR );
+      int max = sched_get_priority_max( SCHED_RR );
+      if ( priority < min ) priority = min;
+      else if ( priority > max ) priority = max;
+      stream_.callbackInfo.priority = priority;
+    }
+#endif
+
+    stream_.callbackInfo.isRunning = true;
+    result = pthread_create( &stream_.callbackInfo.thread, &attr, alsaCallbackHandler, &stream_.callbackInfo );
+    pthread_attr_destroy( &attr );
+    if ( result ) {
+      stream_.callbackInfo.isRunning = false;
+      errorText_ = "RtApiAlsa::error creating callback thread!";
+      goto error;
+    }
+  }
+
+  return SUCCESS;
+
+ error:
+  if ( apiInfo ) {
+    pthread_cond_destroy( &apiInfo->runnable_cv );
+    if ( apiInfo->handles[0] ) snd_pcm_close( apiInfo->handles[0] );
+    if ( apiInfo->handles[1] ) snd_pcm_close( apiInfo->handles[1] );
+    delete apiInfo;
+    stream_.apiHandle = 0;
+  }
+
+  if ( phandle) snd_pcm_close( phandle );
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.state = STREAM_CLOSED;
+  return FAILURE;
+}
+
+void RtApiAlsa :: closeStream()
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiAlsa::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;
+  stream_.callbackInfo.isRunning = false;
+  MUTEX_LOCK( &stream_.mutex );
+  if ( stream_.state == STREAM_STOPPED ) {
+    apiInfo->runnable = true;
+    pthread_cond_signal( &apiInfo->runnable_cv );
+  }
+  MUTEX_UNLOCK( &stream_.mutex );
+  pthread_join( stream_.callbackInfo.thread, NULL );
+
+  if ( stream_.state == STREAM_RUNNING ) {
+    stream_.state = STREAM_STOPPED;
+    if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )
+      snd_pcm_drop( apiInfo->handles[0] );
+    if ( stream_.mode == INPUT || stream_.mode == DUPLEX )
+      snd_pcm_drop( apiInfo->handles[1] );
+  }
+
+  if ( apiInfo ) {
+    pthread_cond_destroy( &apiInfo->runnable_cv );
+    if ( apiInfo->handles[0] ) snd_pcm_close( apiInfo->handles[0] );
+    if ( apiInfo->handles[1] ) snd_pcm_close( apiInfo->handles[1] );
+    delete apiInfo;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+void RtApiAlsa :: startStream()
+{
+  // This method calls snd_pcm_prepare if the device isn't already in that state.
+
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiAlsa::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  int result = 0;
+  snd_pcm_state_t state;
+  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;
+  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    state = snd_pcm_state( handle[0] );
+    if ( state != SND_PCM_STATE_PREPARED ) {
+      result = snd_pcm_prepare( handle[0] );
+      if ( result < 0 ) {
+        errorStream_ << "RtApiAlsa::startStream: error preparing output pcm device, " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+        goto unlock;
+      }
+    }
+  }
+
+  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {
+    result = snd_pcm_drop(handle[1]); // fix to remove stale data received since device has been open
+    state = snd_pcm_state( handle[1] );
+    if ( state != SND_PCM_STATE_PREPARED ) {
+      result = snd_pcm_prepare( handle[1] );
+      if ( result < 0 ) {
+        errorStream_ << "RtApiAlsa::startStream: error preparing input pcm device, " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+        goto unlock;
+      }
+    }
+  }
+
+  stream_.state = STREAM_RUNNING;
+
+ unlock:
+  apiInfo->runnable = true;
+  pthread_cond_signal( &apiInfo->runnable_cv );
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( result >= 0 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiAlsa :: stopStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiAlsa::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_LOCK( &stream_.mutex );
+
+  int result = 0;
+  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;
+  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    if ( apiInfo->synchronized ) 
+      result = snd_pcm_drop( handle[0] );
+    else
+      result = snd_pcm_drain( handle[0] );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::stopStream: error draining output pcm device, " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {
+    result = snd_pcm_drop( handle[1] );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::stopStream: error stopping input pcm device, " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+ unlock:
+  apiInfo->runnable = false; // fixes high CPU usage when stopped
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( result >= 0 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiAlsa :: abortStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiAlsa::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_LOCK( &stream_.mutex );
+
+  int result = 0;
+  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;
+  snd_pcm_t **handle = (snd_pcm_t **) apiInfo->handles;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    result = snd_pcm_drop( handle[0] );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::abortStream: error aborting output pcm device, " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+  if ( ( stream_.mode == INPUT || stream_.mode == DUPLEX ) && !apiInfo->synchronized ) {
+    result = snd_pcm_drop( handle[1] );
+    if ( result < 0 ) {
+      errorStream_ << "RtApiAlsa::abortStream: error aborting input pcm device, " << snd_strerror( result ) << ".";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+ unlock:
+  apiInfo->runnable = false; // fixes high CPU usage when stopped
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( result >= 0 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiAlsa :: callbackEvent()
+{
+  AlsaHandle *apiInfo = (AlsaHandle *) stream_.apiHandle;
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_LOCK( &stream_.mutex );
+    while ( !apiInfo->runnable )
+      pthread_cond_wait( &apiInfo->runnable_cv, &stream_.mutex );
+
+    if ( stream_.state != STREAM_RUNNING ) {
+      MUTEX_UNLOCK( &stream_.mutex );
+      return;
+    }
+    MUTEX_UNLOCK( &stream_.mutex );
+  }
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiAlsa::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  int doStopStream = 0;
+  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;
+  double streamTime = getStreamTime();
+  RtAudioStreamStatus status = 0;
+  if ( stream_.mode != INPUT && apiInfo->xrun[0] == true ) {
+    status |= RTAUDIO_OUTPUT_UNDERFLOW;
+    apiInfo->xrun[0] = false;
+  }
+  if ( stream_.mode != OUTPUT && apiInfo->xrun[1] == true ) {
+    status |= RTAUDIO_INPUT_OVERFLOW;
+    apiInfo->xrun[1] = false;
+  }
+  doStopStream = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                           stream_.bufferSize, streamTime, status, stream_.callbackInfo.userData );
+
+  if ( doStopStream == 2 ) {
+    abortStream();
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  // The state might change while waiting on a mutex.
+  if ( stream_.state == STREAM_STOPPED ) goto unlock;
+
+  int result;
+  char *buffer;
+  int channels;
+  snd_pcm_t **handle;
+  snd_pcm_sframes_t frames;
+  RtAudioFormat format;
+  handle = (snd_pcm_t **) apiInfo->handles;
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    // Setup parameters.
+    if ( stream_.doConvertBuffer[1] ) {
+      buffer = stream_.deviceBuffer;
+      channels = stream_.nDeviceChannels[1];
+      format = stream_.deviceFormat[1];
+    }
+    else {
+      buffer = stream_.userBuffer[1];
+      channels = stream_.nUserChannels[1];
+      format = stream_.userFormat;
+    }
+
+    // Read samples from device in interleaved/non-interleaved format.
+    if ( stream_.deviceInterleaved[1] )
+      result = snd_pcm_readi( handle[1], buffer, stream_.bufferSize );
+    else {
+      void *bufs[channels];
+      size_t offset = stream_.bufferSize * formatBytes( format );
+      for ( int i=0; i<channels; i++ )
+        bufs[i] = (void *) (buffer + (i * offset));
+      result = snd_pcm_readn( handle[1], bufs, stream_.bufferSize );
+    }
+
+    if ( result < (int) stream_.bufferSize ) {
+      // Either an error or overrun occured.
+      if ( result == -EPIPE ) {
+        snd_pcm_state_t state = snd_pcm_state( handle[1] );
+        if ( state == SND_PCM_STATE_XRUN ) {
+          apiInfo->xrun[1] = true;
+          result = snd_pcm_prepare( handle[1] );
+          if ( result < 0 ) {
+            errorStream_ << "RtApiAlsa::callbackEvent: error preparing device after overrun, " << snd_strerror( result ) << ".";
+            errorText_ = errorStream_.str();
+          }
+        }
+        else {
+          errorStream_ << "RtApiAlsa::callbackEvent: error, current state is " << snd_pcm_state_name( state ) << ", " << snd_strerror( result ) << ".";
+          errorText_ = errorStream_.str();
+        }
+      }
+      else {
+        errorStream_ << "RtApiAlsa::callbackEvent: audio read error, " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+      }
+      error( RtAudioError::WARNING );
+      goto tryOutput;
+    }
+
+    // Do byte swapping if necessary.
+    if ( stream_.doByteSwap[1] )
+      byteSwapBuffer( buffer, stream_.bufferSize * channels, format );
+
+    // Do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[1] )
+      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );
+
+    // Check stream latency
+    result = snd_pcm_delay( handle[1], &frames );
+    if ( result == 0 && frames > 0 ) stream_.latency[1] = frames;
+  }
+
+ tryOutput:
+
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    // Setup parameters and do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[0] ) {
+      buffer = stream_.deviceBuffer;
+      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+      channels = stream_.nDeviceChannels[0];
+      format = stream_.deviceFormat[0];
+    }
+    else {
+      buffer = stream_.userBuffer[0];
+      channels = stream_.nUserChannels[0];
+      format = stream_.userFormat;
+    }
+
+    // Do byte swapping if necessary.
+    if ( stream_.doByteSwap[0] )
+      byteSwapBuffer(buffer, stream_.bufferSize * channels, format);
+
+    // Write samples to device in interleaved/non-interleaved format.
+    if ( stream_.deviceInterleaved[0] )
+      result = snd_pcm_writei( handle[0], buffer, stream_.bufferSize );
+    else {
+      void *bufs[channels];
+      size_t offset = stream_.bufferSize * formatBytes( format );
+      for ( int i=0; i<channels; i++ )
+        bufs[i] = (void *) (buffer + (i * offset));
+      result = snd_pcm_writen( handle[0], bufs, stream_.bufferSize );
+    }
+
+    if ( result < (int) stream_.bufferSize ) {
+      // Either an error or underrun occured.
+      if ( result == -EPIPE ) {
+        snd_pcm_state_t state = snd_pcm_state( handle[0] );
+        if ( state == SND_PCM_STATE_XRUN ) {
+          apiInfo->xrun[0] = true;
+          result = snd_pcm_prepare( handle[0] );
+          if ( result < 0 ) {
+            errorStream_ << "RtApiAlsa::callbackEvent: error preparing device after underrun, " << snd_strerror( result ) << ".";
+            errorText_ = errorStream_.str();
+          }
+          else
+            errorText_ =  "RtApiAlsa::callbackEvent: audio write error, underrun.";
+        }
+        else {
+          errorStream_ << "RtApiAlsa::callbackEvent: error, current state is " << snd_pcm_state_name( state ) << ", " << snd_strerror( result ) << ".";
+          errorText_ = errorStream_.str();
+        }
+      }
+      else {
+        errorStream_ << "RtApiAlsa::callbackEvent: audio write error, " << snd_strerror( result ) << ".";
+        errorText_ = errorStream_.str();
+      }
+      error( RtAudioError::WARNING );
+      goto unlock;
+    }
+
+    // Check stream latency
+    result = snd_pcm_delay( handle[0], &frames );
+    if ( result == 0 && frames > 0 ) stream_.latency[0] = frames;
+  }
+
+ unlock:
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  RtApi::tickStreamTime();
+  if ( doStopStream == 1 ) this->stopStream();
+}
+
+static void *alsaCallbackHandler( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiAlsa *object = (RtApiAlsa *) info->object;
+  bool *isRunning = &info->isRunning;
+
+#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)
+  if ( &info->doRealtime ) {
+    pthread_t tID = pthread_self();	 // ID of this thread
+    sched_param prio = { info->priority }; // scheduling priority of thread
+    pthread_setschedparam( tID, SCHED_RR, &prio );
+  }
+#endif
+
+  while ( *isRunning == true ) {
+    pthread_testcancel();
+    object->callbackEvent();
+  }
+
+  pthread_exit( NULL );
+}
+
+//******************** End of __LINUX_ALSA__ *********************//
+#endif
+
+#if defined(__LINUX_PULSE__)
+
+// Code written by Peter Meerwald, pmeerw@pmeerw.net
+// and Tristan Matthews.
+
+#include <pulse/error.h>
+#include <pulse/simple.h>
+#include <cstdio>
+
+static const unsigned int SUPPORTED_SAMPLERATES[] = { 8000, 16000, 22050, 32000,
+                                                      44100, 48000, 96000, 0};
+
+struct rtaudio_pa_format_mapping_t {
+  RtAudioFormat rtaudio_format;
+  pa_sample_format_t pa_format;
+};
+
+static const rtaudio_pa_format_mapping_t supported_sampleformats[] = {
+  {RTAUDIO_SINT16, PA_SAMPLE_S16LE},
+  {RTAUDIO_SINT32, PA_SAMPLE_S32LE},
+  {RTAUDIO_FLOAT32, PA_SAMPLE_FLOAT32LE},
+  {0, PA_SAMPLE_INVALID}};
+
+struct PulseAudioHandle {
+  pa_simple *s_play;
+  pa_simple *s_rec;
+  pthread_t thread;
+  pthread_cond_t runnable_cv;
+  bool runnable;
+  PulseAudioHandle() : s_play(0), s_rec(0), runnable(false) { }
+};
+
+RtApiPulse::~RtApiPulse()
+{
+  if ( stream_.state != STREAM_CLOSED )
+    closeStream();
+}
+
+unsigned int RtApiPulse::getDeviceCount( void )
+{
+  return 1;
+}
+
+RtAudio::DeviceInfo RtApiPulse::getDeviceInfo( unsigned int /*device*/ )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = true;
+  info.name = "PulseAudio";
+  info.outputChannels = 2;
+  info.inputChannels = 2;
+  info.duplexChannels = 2;
+  info.isDefaultOutput = true;
+  info.isDefaultInput = true;
+
+  for ( const unsigned int *sr = SUPPORTED_SAMPLERATES; *sr; ++sr )
+    info.sampleRates.push_back( *sr );
+
+  info.preferredSampleRate = 48000;
+  info.nativeFormats = RTAUDIO_SINT16 | RTAUDIO_SINT32 | RTAUDIO_FLOAT32;
+
+  return info;
+}
+
+static void *pulseaudio_callback( void * user )
+{
+  CallbackInfo *cbi = static_cast<CallbackInfo *>( user );
+  RtApiPulse *context = static_cast<RtApiPulse *>( cbi->object );
+  volatile bool *isRunning = &cbi->isRunning;
+
+  while ( *isRunning ) {
+    pthread_testcancel();
+    context->callbackEvent();
+  }
+
+  pthread_exit( NULL );
+}
+
+void RtApiPulse::closeStream( void )
+{
+  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );
+
+  stream_.callbackInfo.isRunning = false;
+  if ( pah ) {
+    MUTEX_LOCK( &stream_.mutex );
+    if ( stream_.state == STREAM_STOPPED ) {
+      pah->runnable = true;
+      pthread_cond_signal( &pah->runnable_cv );
+    }
+    MUTEX_UNLOCK( &stream_.mutex );
+
+    pthread_join( pah->thread, 0 );
+    if ( pah->s_play ) {
+      pa_simple_flush( pah->s_play, NULL );
+      pa_simple_free( pah->s_play );
+    }
+    if ( pah->s_rec )
+      pa_simple_free( pah->s_rec );
+
+    pthread_cond_destroy( &pah->runnable_cv );
+    delete pah;
+    stream_.apiHandle = 0;
+  }
+
+  if ( stream_.userBuffer[0] ) {
+    free( stream_.userBuffer[0] );
+    stream_.userBuffer[0] = 0;
+  }
+  if ( stream_.userBuffer[1] ) {
+    free( stream_.userBuffer[1] );
+    stream_.userBuffer[1] = 0;
+  }
+
+  stream_.state = STREAM_CLOSED;
+  stream_.mode = UNINITIALIZED;
+}
+
+void RtApiPulse::callbackEvent( void )
+{
+  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );
+
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_LOCK( &stream_.mutex );
+    while ( !pah->runnable )
+      pthread_cond_wait( &pah->runnable_cv, &stream_.mutex );
+
+    if ( stream_.state != STREAM_RUNNING ) {
+      MUTEX_UNLOCK( &stream_.mutex );
+      return;
+    }
+    MUTEX_UNLOCK( &stream_.mutex );
+  }
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiPulse::callbackEvent(): the stream is closed ... "
+      "this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;
+  double streamTime = getStreamTime();
+  RtAudioStreamStatus status = 0;
+  int doStopStream = callback( stream_.userBuffer[OUTPUT], stream_.userBuffer[INPUT],
+                               stream_.bufferSize, streamTime, status,
+                               stream_.callbackInfo.userData );
+
+  if ( doStopStream == 2 ) {
+    abortStream();
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+  void *pulse_in = stream_.doConvertBuffer[INPUT] ? stream_.deviceBuffer : stream_.userBuffer[INPUT];
+  void *pulse_out = stream_.doConvertBuffer[OUTPUT] ? stream_.deviceBuffer : stream_.userBuffer[OUTPUT];
+
+  if ( stream_.state != STREAM_RUNNING )
+    goto unlock;
+
+  int pa_error;
+  size_t bytes;
+  if (stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    if ( stream_.doConvertBuffer[OUTPUT] ) {
+        convertBuffer( stream_.deviceBuffer,
+                       stream_.userBuffer[OUTPUT],
+                       stream_.convertInfo[OUTPUT] );
+        bytes = stream_.nDeviceChannels[OUTPUT] * stream_.bufferSize *
+                formatBytes( stream_.deviceFormat[OUTPUT] );
+    } else
+        bytes = stream_.nUserChannels[OUTPUT] * stream_.bufferSize *
+                formatBytes( stream_.userFormat );
+
+    if ( pa_simple_write( pah->s_play, pulse_out, bytes, &pa_error ) < 0 ) {
+      errorStream_ << "RtApiPulse::callbackEvent: audio write error, " <<
+        pa_strerror( pa_error ) << ".";
+      errorText_ = errorStream_.str();
+      error( RtAudioError::WARNING );
+    }
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX) {
+    if ( stream_.doConvertBuffer[INPUT] )
+      bytes = stream_.nDeviceChannels[INPUT] * stream_.bufferSize *
+        formatBytes( stream_.deviceFormat[INPUT] );
+    else
+      bytes = stream_.nUserChannels[INPUT] * stream_.bufferSize *
+        formatBytes( stream_.userFormat );
+            
+    if ( pa_simple_read( pah->s_rec, pulse_in, bytes, &pa_error ) < 0 ) {
+      errorStream_ << "RtApiPulse::callbackEvent: audio read error, " <<
+        pa_strerror( pa_error ) << ".";
+      errorText_ = errorStream_.str();
+      error( RtAudioError::WARNING );
+    }
+    if ( stream_.doConvertBuffer[INPUT] ) {
+      convertBuffer( stream_.userBuffer[INPUT],
+                     stream_.deviceBuffer,
+                     stream_.convertInfo[INPUT] );
+    }
+  }
+
+ unlock:
+  MUTEX_UNLOCK( &stream_.mutex );
+  RtApi::tickStreamTime();
+
+  if ( doStopStream == 1 )
+    stopStream();
+}
+
+void RtApiPulse::startStream( void )
+{
+  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiPulse::startStream(): the stream is not open!";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiPulse::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  stream_.state = STREAM_RUNNING;
+
+  pah->runnable = true;
+  pthread_cond_signal( &pah->runnable_cv );
+  MUTEX_UNLOCK( &stream_.mutex );
+}
+
+void RtApiPulse::stopStream( void )
+{
+  PulseAudioHandle *pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiPulse::stopStream(): the stream is not open!";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiPulse::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_LOCK( &stream_.mutex );
+
+  if ( pah && pah->s_play ) {
+    int pa_error;
+    if ( pa_simple_drain( pah->s_play, &pa_error ) < 0 ) {
+      errorStream_ << "RtApiPulse::stopStream: error draining output device, " <<
+        pa_strerror( pa_error ) << ".";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_UNLOCK( &stream_.mutex );
+}
+
+void RtApiPulse::abortStream( void )
+{
+  PulseAudioHandle *pah = static_cast<PulseAudioHandle*>( stream_.apiHandle );
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiPulse::abortStream(): the stream is not open!";
+    error( RtAudioError::INVALID_USE );
+    return;
+  }
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiPulse::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_LOCK( &stream_.mutex );
+
+  if ( pah && pah->s_play ) {
+    int pa_error;
+    if ( pa_simple_flush( pah->s_play, &pa_error ) < 0 ) {
+      errorStream_ << "RtApiPulse::abortStream: error flushing output device, " <<
+        pa_strerror( pa_error ) << ".";
+      errorText_ = errorStream_.str();
+      MUTEX_UNLOCK( &stream_.mutex );
+      error( RtAudioError::SYSTEM_ERROR );
+      return;
+    }
+  }
+
+  stream_.state = STREAM_STOPPED;
+  MUTEX_UNLOCK( &stream_.mutex );
+}
+
+bool RtApiPulse::probeDeviceOpen( unsigned int device, StreamMode mode,
+                                  unsigned int channels, unsigned int firstChannel,
+                                  unsigned int sampleRate, RtAudioFormat format,
+                                  unsigned int *bufferSize, RtAudio::StreamOptions *options )
+{
+  PulseAudioHandle *pah = 0;
+  unsigned long bufferBytes = 0;
+  pa_sample_spec ss;
+
+  if ( device != 0 ) return false;
+  if ( mode != INPUT && mode != OUTPUT ) return false;
+  if ( channels != 1 && channels != 2 ) {
+    errorText_ = "RtApiPulse::probeDeviceOpen: unsupported number of channels.";
+    return false;
+  }
+  ss.channels = channels;
+
+  if ( firstChannel != 0 ) return false;
+
+  bool sr_found = false;
+  for ( const unsigned int *sr = SUPPORTED_SAMPLERATES; *sr; ++sr ) {
+    if ( sampleRate == *sr ) {
+      sr_found = true;
+      stream_.sampleRate = sampleRate;
+      ss.rate = sampleRate;
+      break;
+    }
+  }
+  if ( !sr_found ) {
+    errorText_ = "RtApiPulse::probeDeviceOpen: unsupported sample rate.";
+    return false;
+  }
+
+  bool sf_found = 0;
+  for ( const rtaudio_pa_format_mapping_t *sf = supported_sampleformats;
+        sf->rtaudio_format && sf->pa_format != PA_SAMPLE_INVALID; ++sf ) {
+    if ( format == sf->rtaudio_format ) {
+      sf_found = true;
+      stream_.userFormat = sf->rtaudio_format;
+      stream_.deviceFormat[mode] = stream_.userFormat;
+      ss.format = sf->pa_format;
+      break;
+    }
+  }
+  if ( !sf_found ) { // Use internal data format conversion.
+    stream_.userFormat = format;
+    stream_.deviceFormat[mode] = RTAUDIO_FLOAT32;
+    ss.format = PA_SAMPLE_FLOAT32LE;
+  }
+
+  // Set other stream parameters.
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED ) stream_.userInterleaved = false;
+  else stream_.userInterleaved = true;
+  stream_.deviceInterleaved[mode] = true;
+  stream_.nBuffers = 1;
+  stream_.doByteSwap[mode] = false;
+  stream_.nUserChannels[mode] = channels;
+  stream_.nDeviceChannels[mode] = channels + firstChannel;
+  stream_.channelOffset[mode] = 0;
+  std::string streamName = "RtAudio";
+
+  // Set flags for buffer conversion.
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate necessary internal buffers.
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiPulse::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+  stream_.bufferSize = *bufferSize;
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( mode == INPUT ) {
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+        if ( bufferBytes <= bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiPulse::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  stream_.device[mode] = device;
+
+  // Setup the buffer conversion information structure.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );
+
+  if ( !stream_.apiHandle ) {
+    PulseAudioHandle *pah = new PulseAudioHandle;
+    if ( !pah ) {
+      errorText_ = "RtApiPulse::probeDeviceOpen: error allocating memory for handle.";
+      goto error;
+    }
+
+    stream_.apiHandle = pah;
+    if ( pthread_cond_init( &pah->runnable_cv, NULL ) != 0 ) {
+      errorText_ = "RtApiPulse::probeDeviceOpen: error creating condition variable.";
+      goto error;
+    }
+  }
+  pah = static_cast<PulseAudioHandle *>( stream_.apiHandle );
+
+  int error;
+  if ( options && !options->streamName.empty() ) streamName = options->streamName;
+  switch ( mode ) {
+  case INPUT:
+    pa_buffer_attr buffer_attr;
+    buffer_attr.fragsize = bufferBytes;
+    buffer_attr.maxlength = -1;
+
+    pah->s_rec = pa_simple_new( NULL, streamName.c_str(), PA_STREAM_RECORD, NULL, "Record", &ss, NULL, &buffer_attr, &error );
+    if ( !pah->s_rec ) {
+      errorText_ = "RtApiPulse::probeDeviceOpen: error connecting input to PulseAudio server.";
+      goto error;
+    }
+    break;
+  case OUTPUT:
+    pah->s_play = pa_simple_new( NULL, "RtAudio", PA_STREAM_PLAYBACK, NULL, "Playback", &ss, NULL, NULL, &error );
+    if ( !pah->s_play ) {
+      errorText_ = "RtApiPulse::probeDeviceOpen: error connecting output to PulseAudio server.";
+      goto error;
+    }
+    break;
+  default:
+    goto error;
+  }
+
+  if ( stream_.mode == UNINITIALIZED )
+    stream_.mode = mode;
+  else if ( stream_.mode == mode )
+    goto error;
+  else
+    stream_.mode = DUPLEX;
+
+  if ( !stream_.callbackInfo.isRunning ) {
+    stream_.callbackInfo.object = this;
+    stream_.callbackInfo.isRunning = true;
+    if ( pthread_create( &pah->thread, NULL, pulseaudio_callback, (void *)&stream_.callbackInfo) != 0 ) {
+      errorText_ = "RtApiPulse::probeDeviceOpen: error creating thread.";
+      goto error;
+    }
+  }
+
+  stream_.state = STREAM_STOPPED;
+  return true;
+ 
+ error:
+  if ( pah && stream_.callbackInfo.isRunning ) {
+    pthread_cond_destroy( &pah->runnable_cv );
+    delete pah;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  return FAILURE;
+}
+
+//******************** End of __LINUX_PULSE__ *********************//
+#endif
+
+#if defined(__LINUX_OSS__)
+
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <sys/soundcard.h>
+#include <errno.h>
+#include <math.h>
+
+static void *ossCallbackHandler(void * ptr);
+
+// A structure to hold various information related to the OSS API
+// implementation.
+struct OssHandle {
+  int id[2];    // device ids
+  bool xrun[2];
+  bool triggered;
+  pthread_cond_t runnable;
+
+  OssHandle()
+    :triggered(false) { id[0] = 0; id[1] = 0; xrun[0] = false; xrun[1] = false; }
+};
+
+RtApiOss :: RtApiOss()
+{
+  // Nothing to do here.
+}
+
+RtApiOss :: ~RtApiOss()
+{
+  if ( stream_.state != STREAM_CLOSED ) closeStream();
+}
+
+unsigned int RtApiOss :: getDeviceCount( void )
+{
+  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );
+  if ( mixerfd == -1 ) {
+    errorText_ = "RtApiOss::getDeviceCount: error opening '/dev/mixer'.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  oss_sysinfo sysinfo;
+  if ( ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo ) == -1 ) {
+    close( mixerfd );
+    errorText_ = "RtApiOss::getDeviceCount: error getting sysinfo, OSS version >= 4.0 is required.";
+    error( RtAudioError::WARNING );
+    return 0;
+  }
+
+  close( mixerfd );
+  return sysinfo.numaudios;
+}
+
+RtAudio::DeviceInfo RtApiOss :: getDeviceInfo( unsigned int device )
+{
+  RtAudio::DeviceInfo info;
+  info.probed = false;
+
+  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );
+  if ( mixerfd == -1 ) {
+    errorText_ = "RtApiOss::getDeviceInfo: error opening '/dev/mixer'.";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  oss_sysinfo sysinfo;
+  int result = ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo );
+  if ( result == -1 ) {
+    close( mixerfd );
+    errorText_ = "RtApiOss::getDeviceInfo: error getting sysinfo, OSS version >= 4.0 is required.";
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  unsigned nDevices = sysinfo.numaudios;
+  if ( nDevices == 0 ) {
+    close( mixerfd );
+    errorText_ = "RtApiOss::getDeviceInfo: no devices found!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  if ( device >= nDevices ) {
+    close( mixerfd );
+    errorText_ = "RtApiOss::getDeviceInfo: device ID is invalid!";
+    error( RtAudioError::INVALID_USE );
+    return info;
+  }
+
+  oss_audioinfo ainfo;
+  ainfo.dev = device;
+  result = ioctl( mixerfd, SNDCTL_AUDIOINFO, &ainfo );
+  close( mixerfd );
+  if ( result == -1 ) {
+    errorStream_ << "RtApiOss::getDeviceInfo: error getting device (" << ainfo.name << ") info.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Probe channels
+  if ( ainfo.caps & PCM_CAP_OUTPUT ) info.outputChannels = ainfo.max_channels;
+  if ( ainfo.caps & PCM_CAP_INPUT ) info.inputChannels = ainfo.max_channels;
+  if ( ainfo.caps & PCM_CAP_DUPLEX ) {
+    if ( info.outputChannels > 0 && info.inputChannels > 0 && ainfo.caps & PCM_CAP_DUPLEX )
+      info.duplexChannels = (info.outputChannels > info.inputChannels) ? info.inputChannels : info.outputChannels;
+  }
+
+  // Probe data formats ... do for input
+  unsigned long mask = ainfo.iformats;
+  if ( mask & AFMT_S16_LE || mask & AFMT_S16_BE )
+    info.nativeFormats |= RTAUDIO_SINT16;
+  if ( mask & AFMT_S8 )
+    info.nativeFormats |= RTAUDIO_SINT8;
+  if ( mask & AFMT_S32_LE || mask & AFMT_S32_BE )
+    info.nativeFormats |= RTAUDIO_SINT32;
+  if ( mask & AFMT_FLOAT )
+    info.nativeFormats |= RTAUDIO_FLOAT32;
+  if ( mask & AFMT_S24_LE || mask & AFMT_S24_BE )
+    info.nativeFormats |= RTAUDIO_SINT24;
+
+  // Check that we have at least one supported format
+  if ( info.nativeFormats == 0 ) {
+    errorStream_ << "RtApiOss::getDeviceInfo: device (" << ainfo.name << ") data format not supported by RtAudio.";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+    return info;
+  }
+
+  // Probe the supported sample rates.
+  info.sampleRates.clear();
+  if ( ainfo.nrates ) {
+    for ( unsigned int i=0; i<ainfo.nrates; i++ ) {
+      for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {
+        if ( ainfo.rates[i] == SAMPLE_RATES[k] ) {
+          info.sampleRates.push_back( SAMPLE_RATES[k] );
+
+          if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )
+            info.preferredSampleRate = SAMPLE_RATES[k];
+
+          break;
+        }
+      }
+    }
+  }
+  else {
+    // Check min and max rate values;
+    for ( unsigned int k=0; k<MAX_SAMPLE_RATES; k++ ) {
+      if ( ainfo.min_rate <= (int) SAMPLE_RATES[k] && ainfo.max_rate >= (int) SAMPLE_RATES[k] ) {
+        info.sampleRates.push_back( SAMPLE_RATES[k] );
+
+        if ( !info.preferredSampleRate || ( SAMPLE_RATES[k] <= 48000 && SAMPLE_RATES[k] > info.preferredSampleRate ) )
+          info.preferredSampleRate = SAMPLE_RATES[k];
+      }
+    }
+  }
+
+  if ( info.sampleRates.size() == 0 ) {
+    errorStream_ << "RtApiOss::getDeviceInfo: no supported sample rates found for device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    error( RtAudioError::WARNING );
+  }
+  else {
+    info.probed = true;
+    info.name = ainfo.name;
+  }
+
+  return info;
+}
+
+
+bool RtApiOss :: probeDeviceOpen( unsigned int device, StreamMode mode, unsigned int channels,
+                                  unsigned int firstChannel, unsigned int sampleRate,
+                                  RtAudioFormat format, unsigned int *bufferSize,
+                                  RtAudio::StreamOptions *options )
+{
+  int mixerfd = open( "/dev/mixer", O_RDWR, 0 );
+  if ( mixerfd == -1 ) {
+    errorText_ = "RtApiOss::probeDeviceOpen: error opening '/dev/mixer'.";
+    return FAILURE;
+  }
+
+  oss_sysinfo sysinfo;
+  int result = ioctl( mixerfd, SNDCTL_SYSINFO, &sysinfo );
+  if ( result == -1 ) {
+    close( mixerfd );
+    errorText_ = "RtApiOss::probeDeviceOpen: error getting sysinfo, OSS version >= 4.0 is required.";
+    return FAILURE;
+  }
+
+  unsigned nDevices = sysinfo.numaudios;
+  if ( nDevices == 0 ) {
+    // This should not happen because a check is made before this function is called.
+    close( mixerfd );
+    errorText_ = "RtApiOss::probeDeviceOpen: no devices found!";
+    return FAILURE;
+  }
+
+  if ( device >= nDevices ) {
+    // This should not happen because a check is made before this function is called.
+    close( mixerfd );
+    errorText_ = "RtApiOss::probeDeviceOpen: device ID is invalid!";
+    return FAILURE;
+  }
+
+  oss_audioinfo ainfo;
+  ainfo.dev = device;
+  result = ioctl( mixerfd, SNDCTL_AUDIOINFO, &ainfo );
+  close( mixerfd );
+  if ( result == -1 ) {
+    errorStream_ << "RtApiOss::getDeviceInfo: error getting device (" << ainfo.name << ") info.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Check if device supports input or output
+  if ( ( mode == OUTPUT && !( ainfo.caps & PCM_CAP_OUTPUT ) ) ||
+       ( mode == INPUT && !( ainfo.caps & PCM_CAP_INPUT ) ) ) {
+    if ( mode == OUTPUT )
+      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support output.";
+    else
+      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support input.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  int flags = 0;
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  if ( mode == OUTPUT )
+    flags |= O_WRONLY;
+  else { // mode == INPUT
+    if (stream_.mode == OUTPUT && stream_.device[0] == device) {
+      // We just set the same device for playback ... close and reopen for duplex (OSS only).
+      close( handle->id[0] );
+      handle->id[0] = 0;
+      if ( !( ainfo.caps & PCM_CAP_DUPLEX ) ) {
+        errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support duplex mode.";
+        errorText_ = errorStream_.str();
+        return FAILURE;
+      }
+      // Check that the number previously set channels is the same.
+      if ( stream_.nUserChannels[0] != channels ) {
+        errorStream_ << "RtApiOss::probeDeviceOpen: input/output channels must be equal for OSS duplex device (" << ainfo.name << ").";
+        errorText_ = errorStream_.str();
+        return FAILURE;
+      }
+      flags |= O_RDWR;
+    }
+    else
+      flags |= O_RDONLY;
+  }
+
+  // Set exclusive access if specified.
+  if ( options && options->flags & RTAUDIO_HOG_DEVICE ) flags |= O_EXCL;
+
+  // Try to open the device.
+  int fd;
+  fd = open( ainfo.devnode, flags, 0 );
+  if ( fd == -1 ) {
+    if ( errno == EBUSY )
+      errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") is busy.";
+    else
+      errorStream_ << "RtApiOss::probeDeviceOpen: error opening device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // For duplex operation, specifically set this mode (this doesn't seem to work).
+  /*
+    if ( flags | O_RDWR ) {
+    result = ioctl( fd, SNDCTL_DSP_SETDUPLEX, NULL );
+    if ( result == -1) {
+    errorStream_ << "RtApiOss::probeDeviceOpen: error setting duplex mode for device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+    }
+    }
+  */
+
+  // Check the device channel support.
+  stream_.nUserChannels[mode] = channels;
+  if ( ainfo.max_channels < (int)(channels + firstChannel) ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: the device (" << ainfo.name << ") does not support requested channel parameters.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Set the number of channels.
+  int deviceChannels = channels + firstChannel;
+  result = ioctl( fd, SNDCTL_DSP_CHANNELS, &deviceChannels );
+  if ( result == -1 || deviceChannels < (int)(channels + firstChannel) ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: error setting channel parameters on device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  stream_.nDeviceChannels[mode] = deviceChannels;
+
+  // Get the data format mask
+  int mask;
+  result = ioctl( fd, SNDCTL_DSP_GETFMTS, &mask );
+  if ( result == -1 ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: error getting device (" << ainfo.name << ") data formats.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Determine how to set the device format.
+  stream_.userFormat = format;
+  int deviceFormat = -1;
+  stream_.doByteSwap[mode] = false;
+  if ( format == RTAUDIO_SINT8 ) {
+    if ( mask & AFMT_S8 ) {
+      deviceFormat = AFMT_S8;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+    }
+  }
+  else if ( format == RTAUDIO_SINT16 ) {
+    if ( mask & AFMT_S16_NE ) {
+      deviceFormat = AFMT_S16_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+    }
+    else if ( mask & AFMT_S16_OE ) {
+      deviceFormat = AFMT_S16_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+      stream_.doByteSwap[mode] = true;
+    }
+  }
+  else if ( format == RTAUDIO_SINT24 ) {
+    if ( mask & AFMT_S24_NE ) {
+      deviceFormat = AFMT_S24_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+    }
+    else if ( mask & AFMT_S24_OE ) {
+      deviceFormat = AFMT_S24_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+      stream_.doByteSwap[mode] = true;
+    }
+  }
+  else if ( format == RTAUDIO_SINT32 ) {
+    if ( mask & AFMT_S32_NE ) {
+      deviceFormat = AFMT_S32_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+    }
+    else if ( mask & AFMT_S32_OE ) {
+      deviceFormat = AFMT_S32_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+      stream_.doByteSwap[mode] = true;
+    }
+  }
+
+  if ( deviceFormat == -1 ) {
+    // The user requested format is not natively supported by the device.
+    if ( mask & AFMT_S16_NE ) {
+      deviceFormat = AFMT_S16_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+    }
+    else if ( mask & AFMT_S32_NE ) {
+      deviceFormat = AFMT_S32_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+    }
+    else if ( mask & AFMT_S24_NE ) {
+      deviceFormat = AFMT_S24_NE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+    }
+    else if ( mask & AFMT_S16_OE ) {
+      deviceFormat = AFMT_S16_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT16;
+      stream_.doByteSwap[mode] = true;
+    }
+    else if ( mask & AFMT_S32_OE ) {
+      deviceFormat = AFMT_S32_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT32;
+      stream_.doByteSwap[mode] = true;
+    }
+    else if ( mask & AFMT_S24_OE ) {
+      deviceFormat = AFMT_S24_OE;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT24;
+      stream_.doByteSwap[mode] = true;
+    }
+    else if ( mask & AFMT_S8) {
+      deviceFormat = AFMT_S8;
+      stream_.deviceFormat[mode] = RTAUDIO_SINT8;
+    }
+  }
+
+  if ( stream_.deviceFormat[mode] == 0 ) {
+    // This really shouldn't happen ...
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") data format not supported by RtAudio.";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Set the data format.
+  int temp = deviceFormat;
+  result = ioctl( fd, SNDCTL_DSP_SETFMT, &deviceFormat );
+  if ( result == -1 || deviceFormat != temp ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: error setting data format on device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Attempt to set the buffer size.  According to OSS, the minimum
+  // number of buffers is two.  The supposed minimum buffer size is 16
+  // bytes, so that will be our lower bound.  The argument to this
+  // call is in the form 0xMMMMSSSS (hex), where the buffer size (in
+  // bytes) is given as 2^SSSS and the number of buffers as 2^MMMM.
+  // We'll check the actual value used near the end of the setup
+  // procedure.
+  int ossBufferBytes = *bufferSize * formatBytes( stream_.deviceFormat[mode] ) * deviceChannels;
+  if ( ossBufferBytes < 16 ) ossBufferBytes = 16;
+  int buffers = 0;
+  if ( options ) buffers = options->numberOfBuffers;
+  if ( options && options->flags & RTAUDIO_MINIMIZE_LATENCY ) buffers = 2;
+  if ( buffers < 2 ) buffers = 3;
+  temp = ((int) buffers << 16) + (int)( log10( (double)ossBufferBytes ) / log10( 2.0 ) );
+  result = ioctl( fd, SNDCTL_DSP_SETFRAGMENT, &temp );
+  if ( result == -1 ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: error setting buffer size on device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  stream_.nBuffers = buffers;
+
+  // Save buffer size (in sample frames).
+  *bufferSize = ossBufferBytes / ( formatBytes(stream_.deviceFormat[mode]) * deviceChannels );
+  stream_.bufferSize = *bufferSize;
+
+  // Set the sample rate.
+  int srate = sampleRate;
+  result = ioctl( fd, SNDCTL_DSP_SPEED, &srate );
+  if ( result == -1 ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: error setting sample rate (" << sampleRate << ") on device (" << ainfo.name << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+
+  // Verify the sample rate setup worked.
+  if ( abs( srate - sampleRate ) > 100 ) {
+    close( fd );
+    errorStream_ << "RtApiOss::probeDeviceOpen: device (" << ainfo.name << ") does not support sample rate (" << sampleRate << ").";
+    errorText_ = errorStream_.str();
+    return FAILURE;
+  }
+  stream_.sampleRate = sampleRate;
+
+  if ( mode == INPUT && stream_.mode == OUTPUT && stream_.device[0] == device) {
+    // We're doing duplex setup here.
+    stream_.deviceFormat[0] = stream_.deviceFormat[1];
+    stream_.nDeviceChannels[0] = deviceChannels;
+  }
+
+  // Set interleaving parameters.
+  stream_.userInterleaved = true;
+  stream_.deviceInterleaved[mode] =  true;
+  if ( options && options->flags & RTAUDIO_NONINTERLEAVED )
+    stream_.userInterleaved = false;
+
+  // Set flags for buffer conversion
+  stream_.doConvertBuffer[mode] = false;
+  if ( stream_.userFormat != stream_.deviceFormat[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.nUserChannels[mode] < stream_.nDeviceChannels[mode] )
+    stream_.doConvertBuffer[mode] = true;
+  if ( stream_.userInterleaved != stream_.deviceInterleaved[mode] &&
+       stream_.nUserChannels[mode] > 1 )
+    stream_.doConvertBuffer[mode] = true;
+
+  // Allocate the stream handles if necessary and then save.
+  if ( stream_.apiHandle == 0 ) {
+    try {
+      handle = new OssHandle;
+    }
+    catch ( std::bad_alloc& ) {
+      errorText_ = "RtApiOss::probeDeviceOpen: error allocating OssHandle memory.";
+      goto error;
+    }
+
+    if ( pthread_cond_init( &handle->runnable, NULL ) ) {
+      errorText_ = "RtApiOss::probeDeviceOpen: error initializing pthread condition variable.";
+      goto error;
+    }
+
+    stream_.apiHandle = (void *) handle;
+  }
+  else {
+    handle = (OssHandle *) stream_.apiHandle;
+  }
+  handle->id[mode] = fd;
+
+  // Allocate necessary internal buffers.
+  unsigned long bufferBytes;
+  bufferBytes = stream_.nUserChannels[mode] * *bufferSize * formatBytes( stream_.userFormat );
+  stream_.userBuffer[mode] = (char *) calloc( bufferBytes, 1 );
+  if ( stream_.userBuffer[mode] == NULL ) {
+    errorText_ = "RtApiOss::probeDeviceOpen: error allocating user buffer memory.";
+    goto error;
+  }
+
+  if ( stream_.doConvertBuffer[mode] ) {
+
+    bool makeBuffer = true;
+    bufferBytes = stream_.nDeviceChannels[mode] * formatBytes( stream_.deviceFormat[mode] );
+    if ( mode == INPUT ) {
+      if ( stream_.mode == OUTPUT && stream_.deviceBuffer ) {
+        unsigned long bytesOut = stream_.nDeviceChannels[0] * formatBytes( stream_.deviceFormat[0] );
+        if ( bufferBytes <= bytesOut ) makeBuffer = false;
+      }
+    }
+
+    if ( makeBuffer ) {
+      bufferBytes *= *bufferSize;
+      if ( stream_.deviceBuffer ) free( stream_.deviceBuffer );
+      stream_.deviceBuffer = (char *) calloc( bufferBytes, 1 );
+      if ( stream_.deviceBuffer == NULL ) {
+        errorText_ = "RtApiOss::probeDeviceOpen: error allocating device buffer memory.";
+        goto error;
+      }
+    }
+  }
+
+  stream_.device[mode] = device;
+  stream_.state = STREAM_STOPPED;
+
+  // Setup the buffer conversion information structure.
+  if ( stream_.doConvertBuffer[mode] ) setConvertInfo( mode, firstChannel );
+
+  // Setup thread if necessary.
+  if ( stream_.mode == OUTPUT && mode == INPUT ) {
+    // We had already set up an output stream.
+    stream_.mode = DUPLEX;
+    if ( stream_.device[0] == device ) handle->id[0] = fd;
+  }
+  else {
+    stream_.mode = mode;
+
+    // Setup callback thread.
+    stream_.callbackInfo.object = (void *) this;
+
+    // Set the thread attributes for joinable and realtime scheduling
+    // priority.  The higher priority will only take affect if the
+    // program is run as root or suid.
+    pthread_attr_t attr;
+    pthread_attr_init( &attr );
+    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE );
+#ifdef SCHED_RR // Undefined with some OSes (eg: NetBSD 1.6.x with GNU Pthread)
+    if ( options && options->flags & RTAUDIO_SCHEDULE_REALTIME ) {
+      struct sched_param param;
+      int priority = options->priority;
+      int min = sched_get_priority_min( SCHED_RR );
+      int max = sched_get_priority_max( SCHED_RR );
+      if ( priority < min ) priority = min;
+      else if ( priority > max ) priority = max;
+      param.sched_priority = priority;
+      pthread_attr_setschedparam( &attr, &param );
+      pthread_attr_setschedpolicy( &attr, SCHED_RR );
+    }
+    else
+      pthread_attr_setschedpolicy( &attr, SCHED_OTHER );
+#else
+    pthread_attr_setschedpolicy( &attr, SCHED_OTHER );
+#endif
+
+    stream_.callbackInfo.isRunning = true;
+    result = pthread_create( &stream_.callbackInfo.thread, &attr, ossCallbackHandler, &stream_.callbackInfo );
+    pthread_attr_destroy( &attr );
+    if ( result ) {
+      stream_.callbackInfo.isRunning = false;
+      errorText_ = "RtApiOss::error creating callback thread!";
+      goto error;
+    }
+  }
+
+  return SUCCESS;
+
+ error:
+  if ( handle ) {
+    pthread_cond_destroy( &handle->runnable );
+    if ( handle->id[0] ) close( handle->id[0] );
+    if ( handle->id[1] ) close( handle->id[1] );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  return FAILURE;
+}
+
+void RtApiOss :: closeStream()
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiOss::closeStream(): no open stream to close!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  stream_.callbackInfo.isRunning = false;
+  MUTEX_LOCK( &stream_.mutex );
+  if ( stream_.state == STREAM_STOPPED )
+    pthread_cond_signal( &handle->runnable );
+  MUTEX_UNLOCK( &stream_.mutex );
+  pthread_join( stream_.callbackInfo.thread, NULL );
+
+  if ( stream_.state == STREAM_RUNNING ) {
+    if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX )
+      ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );
+    else
+      ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );
+    stream_.state = STREAM_STOPPED;
+  }
+
+  if ( handle ) {
+    pthread_cond_destroy( &handle->runnable );
+    if ( handle->id[0] ) close( handle->id[0] );
+    if ( handle->id[1] ) close( handle->id[1] );
+    delete handle;
+    stream_.apiHandle = 0;
+  }
+
+  for ( int i=0; i<2; i++ ) {
+    if ( stream_.userBuffer[i] ) {
+      free( stream_.userBuffer[i] );
+      stream_.userBuffer[i] = 0;
+    }
+  }
+
+  if ( stream_.deviceBuffer ) {
+    free( stream_.deviceBuffer );
+    stream_.deviceBuffer = 0;
+  }
+
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+}
+
+void RtApiOss :: startStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_RUNNING ) {
+    errorText_ = "RtApiOss::startStream(): the stream is already running!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  stream_.state = STREAM_RUNNING;
+
+  // No need to do anything else here ... OSS automatically starts
+  // when fed samples.
+
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  pthread_cond_signal( &handle->runnable );
+}
+
+void RtApiOss :: stopStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiOss::stopStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  // The state might change while waiting on a mutex.
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_UNLOCK( &stream_.mutex );
+    return;
+  }
+
+  int result = 0;
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    // Flush the output with zeros a few times.
+    char *buffer;
+    int samples;
+    RtAudioFormat format;
+
+    if ( stream_.doConvertBuffer[0] ) {
+      buffer = stream_.deviceBuffer;
+      samples = stream_.bufferSize * stream_.nDeviceChannels[0];
+      format = stream_.deviceFormat[0];
+    }
+    else {
+      buffer = stream_.userBuffer[0];
+      samples = stream_.bufferSize * stream_.nUserChannels[0];
+      format = stream_.userFormat;
+    }
+
+    memset( buffer, 0, samples * formatBytes(format) );
+    for ( unsigned int i=0; i<stream_.nBuffers+1; i++ ) {
+      result = write( handle->id[0], buffer, samples * formatBytes(format) );
+      if ( result == -1 ) {
+        errorText_ = "RtApiOss::stopStream: audio write error.";
+        error( RtAudioError::WARNING );
+      }
+    }
+
+    result = ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );
+    if ( result == -1 ) {
+      errorStream_ << "RtApiOss::stopStream: system error stopping callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+    handle->triggered = false;
+  }
+
+  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && handle->id[0] != handle->id[1] ) ) {
+    result = ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );
+    if ( result == -1 ) {
+      errorStream_ << "RtApiOss::stopStream: system error stopping input callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+ unlock:
+  stream_.state = STREAM_STOPPED;
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( result != -1 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiOss :: abortStream()
+{
+  verifyStream();
+  if ( stream_.state == STREAM_STOPPED ) {
+    errorText_ = "RtApiOss::abortStream(): the stream is already stopped!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  // The state might change while waiting on a mutex.
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_UNLOCK( &stream_.mutex );
+    return;
+  }
+
+  int result = 0;
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+    result = ioctl( handle->id[0], SNDCTL_DSP_HALT, 0 );
+    if ( result == -1 ) {
+      errorStream_ << "RtApiOss::abortStream: system error stopping callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+    handle->triggered = false;
+  }
+
+  if ( stream_.mode == INPUT || ( stream_.mode == DUPLEX && handle->id[0] != handle->id[1] ) ) {
+    result = ioctl( handle->id[1], SNDCTL_DSP_HALT, 0 );
+    if ( result == -1 ) {
+      errorStream_ << "RtApiOss::abortStream: system error stopping input callback procedure on device (" << stream_.device[0] << ").";
+      errorText_ = errorStream_.str();
+      goto unlock;
+    }
+  }
+
+ unlock:
+  stream_.state = STREAM_STOPPED;
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  if ( result != -1 ) return;
+  error( RtAudioError::SYSTEM_ERROR );
+}
+
+void RtApiOss :: callbackEvent()
+{
+  OssHandle *handle = (OssHandle *) stream_.apiHandle;
+  if ( stream_.state == STREAM_STOPPED ) {
+    MUTEX_LOCK( &stream_.mutex );
+    pthread_cond_wait( &handle->runnable, &stream_.mutex );
+    if ( stream_.state != STREAM_RUNNING ) {
+      MUTEX_UNLOCK( &stream_.mutex );
+      return;
+    }
+    MUTEX_UNLOCK( &stream_.mutex );
+  }
+
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApiOss::callbackEvent(): the stream is closed ... this shouldn't happen!";
+    error( RtAudioError::WARNING );
+    return;
+  }
+
+  // Invoke user callback to get fresh output data.
+  int doStopStream = 0;
+  RtAudioCallback callback = (RtAudioCallback) stream_.callbackInfo.callback;
+  double streamTime = getStreamTime();
+  RtAudioStreamStatus status = 0;
+  if ( stream_.mode != INPUT && handle->xrun[0] == true ) {
+    status |= RTAUDIO_OUTPUT_UNDERFLOW;
+    handle->xrun[0] = false;
+  }
+  if ( stream_.mode != OUTPUT && handle->xrun[1] == true ) {
+    status |= RTAUDIO_INPUT_OVERFLOW;
+    handle->xrun[1] = false;
+  }
+  doStopStream = callback( stream_.userBuffer[0], stream_.userBuffer[1],
+                           stream_.bufferSize, streamTime, status, stream_.callbackInfo.userData );
+  if ( doStopStream == 2 ) {
+    this->abortStream();
+    return;
+  }
+
+  MUTEX_LOCK( &stream_.mutex );
+
+  // The state might change while waiting on a mutex.
+  if ( stream_.state == STREAM_STOPPED ) goto unlock;
+
+  int result;
+  char *buffer;
+  int samples;
+  RtAudioFormat format;
+
+  if ( stream_.mode == OUTPUT || stream_.mode == DUPLEX ) {
+
+    // Setup parameters and do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[0] ) {
+      buffer = stream_.deviceBuffer;
+      convertBuffer( buffer, stream_.userBuffer[0], stream_.convertInfo[0] );
+      samples = stream_.bufferSize * stream_.nDeviceChannels[0];
+      format = stream_.deviceFormat[0];
+    }
+    else {
+      buffer = stream_.userBuffer[0];
+      samples = stream_.bufferSize * stream_.nUserChannels[0];
+      format = stream_.userFormat;
+    }
+
+    // Do byte swapping if necessary.
+    if ( stream_.doByteSwap[0] )
+      byteSwapBuffer( buffer, samples, format );
+
+    if ( stream_.mode == DUPLEX && handle->triggered == false ) {
+      int trig = 0;
+      ioctl( handle->id[0], SNDCTL_DSP_SETTRIGGER, &trig );
+      result = write( handle->id[0], buffer, samples * formatBytes(format) );
+      trig = PCM_ENABLE_INPUT|PCM_ENABLE_OUTPUT;
+      ioctl( handle->id[0], SNDCTL_DSP_SETTRIGGER, &trig );
+      handle->triggered = true;
+    }
+    else
+      // Write samples to device.
+      result = write( handle->id[0], buffer, samples * formatBytes(format) );
+
+    if ( result == -1 ) {
+      // We'll assume this is an underrun, though there isn't a
+      // specific means for determining that.
+      handle->xrun[0] = true;
+      errorText_ = "RtApiOss::callbackEvent: audio write error.";
+      error( RtAudioError::WARNING );
+      // Continue on to input section.
+    }
+  }
+
+  if ( stream_.mode == INPUT || stream_.mode == DUPLEX ) {
+
+    // Setup parameters.
+    if ( stream_.doConvertBuffer[1] ) {
+      buffer = stream_.deviceBuffer;
+      samples = stream_.bufferSize * stream_.nDeviceChannels[1];
+      format = stream_.deviceFormat[1];
+    }
+    else {
+      buffer = stream_.userBuffer[1];
+      samples = stream_.bufferSize * stream_.nUserChannels[1];
+      format = stream_.userFormat;
+    }
+
+    // Read samples from device.
+    result = read( handle->id[1], buffer, samples * formatBytes(format) );
+
+    if ( result == -1 ) {
+      // We'll assume this is an overrun, though there isn't a
+      // specific means for determining that.
+      handle->xrun[1] = true;
+      errorText_ = "RtApiOss::callbackEvent: audio read error.";
+      error( RtAudioError::WARNING );
+      goto unlock;
+    }
+
+    // Do byte swapping if necessary.
+    if ( stream_.doByteSwap[1] )
+      byteSwapBuffer( buffer, samples, format );
+
+    // Do buffer conversion if necessary.
+    if ( stream_.doConvertBuffer[1] )
+      convertBuffer( stream_.userBuffer[1], stream_.deviceBuffer, stream_.convertInfo[1] );
+  }
+
+ unlock:
+  MUTEX_UNLOCK( &stream_.mutex );
+
+  RtApi::tickStreamTime();
+  if ( doStopStream == 1 ) this->stopStream();
+}
+
+static void *ossCallbackHandler( void *ptr )
+{
+  CallbackInfo *info = (CallbackInfo *) ptr;
+  RtApiOss *object = (RtApiOss *) info->object;
+  bool *isRunning = &info->isRunning;
+
+  while ( *isRunning == true ) {
+    pthread_testcancel();
+    object->callbackEvent();
+  }
+
+  pthread_exit( NULL );
+}
+
+//******************** End of __LINUX_OSS__ *********************//
+#endif
+
+
+// *************************************************** //
+//
+// Protected common (OS-independent) RtAudio methods.
+//
+// *************************************************** //
+
+// This method can be modified to control the behavior of error
+// message printing.
+void RtApi :: error( RtAudioError::Type type )
+{
+  errorStream_.str(""); // clear the ostringstream
+
+  RtAudioErrorCallback errorCallback = (RtAudioErrorCallback) stream_.callbackInfo.errorCallback;
+  if ( errorCallback ) {
+    // abortStream() can generate new error messages. Ignore them. Just keep original one.
+
+    if ( firstErrorOccurred_ )
+      return;
+
+    firstErrorOccurred_ = true;
+    const std::string errorMessage = errorText_;
+
+    if ( type != RtAudioError::WARNING && stream_.state != STREAM_STOPPED) {
+      stream_.callbackInfo.isRunning = false; // exit from the thread
+      abortStream();
+    }
+
+    errorCallback( type, errorMessage );
+    firstErrorOccurred_ = false;
+    return;
+  }
+
+  if ( type == RtAudioError::WARNING && showWarnings_ == true )
+    std::cerr << '\n' << errorText_ << "\n\n";
+  else if ( type != RtAudioError::WARNING )
+    throw( RtAudioError( errorText_, type ) );
+}
+
+void RtApi :: verifyStream()
+{
+  if ( stream_.state == STREAM_CLOSED ) {
+    errorText_ = "RtApi:: a stream is not open!";
+    error( RtAudioError::INVALID_USE );
+  }
+}
+
+void RtApi :: clearStreamInfo()
+{
+  stream_.mode = UNINITIALIZED;
+  stream_.state = STREAM_CLOSED;
+  stream_.sampleRate = 0;
+  stream_.bufferSize = 0;
+  stream_.nBuffers = 0;
+  stream_.userFormat = 0;
+  stream_.userInterleaved = true;
+  stream_.streamTime = 0.0;
+  stream_.apiHandle = 0;
+  stream_.deviceBuffer = 0;
+  stream_.callbackInfo.callback = 0;
+  stream_.callbackInfo.userData = 0;
+  stream_.callbackInfo.isRunning = false;
+  stream_.callbackInfo.errorCallback = 0;
+  for ( int i=0; i<2; i++ ) {
+    stream_.device[i] = 11111;
+    stream_.doConvertBuffer[i] = false;
+    stream_.deviceInterleaved[i] = true;
+    stream_.doByteSwap[i] = false;
+    stream_.nUserChannels[i] = 0;
+    stream_.nDeviceChannels[i] = 0;
+    stream_.channelOffset[i] = 0;
+    stream_.deviceFormat[i] = 0;
+    stream_.latency[i] = 0;
+    stream_.userBuffer[i] = 0;
+    stream_.convertInfo[i].channels = 0;
+    stream_.convertInfo[i].inJump = 0;
+    stream_.convertInfo[i].outJump = 0;
+    stream_.convertInfo[i].inFormat = 0;
+    stream_.convertInfo[i].outFormat = 0;
+    stream_.convertInfo[i].inOffset.clear();
+    stream_.convertInfo[i].outOffset.clear();
+  }
+}
+
+unsigned int RtApi :: formatBytes( RtAudioFormat format )
+{
+  if ( format == RTAUDIO_SINT16 )
+    return 2;
+  else if ( format == RTAUDIO_SINT32 || format == RTAUDIO_FLOAT32 )
+    return 4;
+  else if ( format == RTAUDIO_FLOAT64 )
+    return 8;
+  else if ( format == RTAUDIO_SINT24 )
+    return 3;
+  else if ( format == RTAUDIO_SINT8 )
+    return 1;
+
+  errorText_ = "RtApi::formatBytes: undefined format.";
+  error( RtAudioError::WARNING );
+
+  return 0;
+}
+
+void RtApi :: setConvertInfo( StreamMode mode, unsigned int firstChannel )
+{
+  if ( mode == INPUT ) { // convert device to user buffer
+    stream_.convertInfo[mode].inJump = stream_.nDeviceChannels[1];
+    stream_.convertInfo[mode].outJump = stream_.nUserChannels[1];
+    stream_.convertInfo[mode].inFormat = stream_.deviceFormat[1];
+    stream_.convertInfo[mode].outFormat = stream_.userFormat;
+  }
+  else { // convert user to device buffer
+    stream_.convertInfo[mode].inJump = stream_.nUserChannels[0];
+    stream_.convertInfo[mode].outJump = stream_.nDeviceChannels[0];
+    stream_.convertInfo[mode].inFormat = stream_.userFormat;
+    stream_.convertInfo[mode].outFormat = stream_.deviceFormat[0];
+  }
+
+  if ( stream_.convertInfo[mode].inJump < stream_.convertInfo[mode].outJump )
+    stream_.convertInfo[mode].channels = stream_.convertInfo[mode].inJump;
+  else
+    stream_.convertInfo[mode].channels = stream_.convertInfo[mode].outJump;
+
+  // Set up the interleave/deinterleave offsets.
+  if ( stream_.deviceInterleaved[mode] != stream_.userInterleaved ) {
+    if ( ( mode == OUTPUT && stream_.deviceInterleaved[mode] ) ||
+         ( mode == INPUT && stream_.userInterleaved ) ) {
+      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {
+        stream_.convertInfo[mode].inOffset.push_back( k * stream_.bufferSize );
+        stream_.convertInfo[mode].outOffset.push_back( k );
+        stream_.convertInfo[mode].inJump = 1;
+      }
+    }
+    else {
+      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {
+        stream_.convertInfo[mode].inOffset.push_back( k );
+        stream_.convertInfo[mode].outOffset.push_back( k * stream_.bufferSize );
+        stream_.convertInfo[mode].outJump = 1;
+      }
+    }
+  }
+  else { // no (de)interleaving
+    if ( stream_.userInterleaved ) {
+      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {
+        stream_.convertInfo[mode].inOffset.push_back( k );
+        stream_.convertInfo[mode].outOffset.push_back( k );
+      }
+    }
+    else {
+      for ( int k=0; k<stream_.convertInfo[mode].channels; k++ ) {
+        stream_.convertInfo[mode].inOffset.push_back( k * stream_.bufferSize );
+        stream_.convertInfo[mode].outOffset.push_back( k * stream_.bufferSize );
+        stream_.convertInfo[mode].inJump = 1;
+        stream_.convertInfo[mode].outJump = 1;
+      }
+    }
+  }
+
+  // Add channel offset.
+  if ( firstChannel > 0 ) {
+    if ( stream_.deviceInterleaved[mode] ) {
+      if ( mode == OUTPUT ) {
+        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )
+          stream_.convertInfo[mode].outOffset[k] += firstChannel;
+      }
+      else {
+        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )
+          stream_.convertInfo[mode].inOffset[k] += firstChannel;
+      }
+    }
+    else {
+      if ( mode == OUTPUT ) {
+        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )
+          stream_.convertInfo[mode].outOffset[k] += ( firstChannel * stream_.bufferSize );
+      }
+      else {
+        for ( int k=0; k<stream_.convertInfo[mode].channels; k++ )
+          stream_.convertInfo[mode].inOffset[k] += ( firstChannel  * stream_.bufferSize );
+      }
+    }
+  }
+}
+
+void RtApi :: convertBuffer( char *outBuffer, char *inBuffer, ConvertInfo &info )
+{
+  // This function does format conversion, input/output channel compensation, and
+  // data interleaving/deinterleaving.  24-bit integers are assumed to occupy
+  // the lower three bytes of a 32-bit integer.
+
+  // Clear our device buffer when in/out duplex device channels are different
+  if ( outBuffer == stream_.deviceBuffer && stream_.mode == DUPLEX &&
+       ( stream_.nDeviceChannels[0] < stream_.nDeviceChannels[1] ) )
+    memset( outBuffer, 0, stream_.bufferSize * info.outJump * formatBytes( info.outFormat ) );
+
+  int j;
+  if (info.outFormat == RTAUDIO_FLOAT64) {
+    Float64 scale;
+    Float64 *out = (Float64 *)outBuffer;
+
+    if (info.inFormat == RTAUDIO_SINT8) {
+      signed char *in = (signed char *)inBuffer;
+      scale = 1.0 / 127.5;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT16) {
+      Int16 *in = (Int16 *)inBuffer;
+      scale = 1.0 / 32767.5;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      Int24 *in = (Int24 *)inBuffer;
+      scale = 1.0 / 8388607.5;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float64) (in[info.inOffset[j]].asInt());
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      Int32 *in = (Int32 *)inBuffer;
+      scale = 1.0 / 2147483647.5;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float64) in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      // Channel compensation and/or (de)interleaving only.
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+  else if (info.outFormat == RTAUDIO_FLOAT32) {
+    Float32 scale;
+    Float32 *out = (Float32 *)outBuffer;
+
+    if (info.inFormat == RTAUDIO_SINT8) {
+      signed char *in = (signed char *)inBuffer;
+      scale = (Float32) ( 1.0 / 127.5 );
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT16) {
+      Int16 *in = (Int16 *)inBuffer;
+      scale = (Float32) ( 1.0 / 32767.5 );
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      Int24 *in = (Int24 *)inBuffer;
+      scale = (Float32) ( 1.0 / 8388607.5 );
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float32) (in[info.inOffset[j]].asInt());
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      Int32 *in = (Int32 *)inBuffer;
+      scale = (Float32) ( 1.0 / 2147483647.5 );
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];
+          out[info.outOffset[j]] += 0.5;
+          out[info.outOffset[j]] *= scale;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      // Channel compensation and/or (de)interleaving only.
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Float32) in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+  else if (info.outFormat == RTAUDIO_SINT32) {
+    Int32 *out = (Int32 *)outBuffer;
+    if (info.inFormat == RTAUDIO_SINT8) {
+      signed char *in = (signed char *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]];
+          out[info.outOffset[j]] <<= 24;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT16) {
+      Int16 *in = (Int16 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]];
+          out[info.outOffset[j]] <<= 16;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      Int24 *in = (Int24 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) in[info.inOffset[j]].asInt();
+          out[info.outOffset[j]] <<= 8;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      // Channel compensation and/or (de)interleaving only.
+      Int32 *in = (Int32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 2147483647.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 2147483647.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+  else if (info.outFormat == RTAUDIO_SINT24) {
+    Int24 *out = (Int24 *)outBuffer;
+    if (info.inFormat == RTAUDIO_SINT8) {
+      signed char *in = (signed char *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] << 16);
+          //out[info.outOffset[j]] <<= 16;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT16) {
+      Int16 *in = (Int16 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] << 8);
+          //out[info.outOffset[j]] <<= 8;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      // Channel compensation and/or (de)interleaving only.
+      Int24 *in = (Int24 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      Int32 *in = (Int32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] >> 8);
+          //out[info.outOffset[j]] >>= 8;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 8388607.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int32) (in[info.inOffset[j]] * 8388607.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+  else if (info.outFormat == RTAUDIO_SINT16) {
+    Int16 *out = (Int16 *)outBuffer;
+    if (info.inFormat == RTAUDIO_SINT8) {
+      signed char *in = (signed char *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int16) in[info.inOffset[j]];
+          out[info.outOffset[j]] <<= 8;
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT16) {
+      // Channel compensation and/or (de)interleaving only.
+      Int16 *in = (Int16 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      Int24 *in = (Int24 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]].asInt() >> 8);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      Int32 *in = (Int32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int16) ((in[info.inOffset[j]] >> 16) & 0x0000ffff);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]] * 32767.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (Int16) (in[info.inOffset[j]] * 32767.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+  else if (info.outFormat == RTAUDIO_SINT8) {
+    signed char *out = (signed char *)outBuffer;
+    if (info.inFormat == RTAUDIO_SINT8) {
+      // Channel compensation and/or (de)interleaving only.
+      signed char *in = (signed char *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = in[info.inOffset[j]];
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    if (info.inFormat == RTAUDIO_SINT16) {
+      Int16 *in = (Int16 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (signed char) ((in[info.inOffset[j]] >> 8) & 0x00ff);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT24) {
+      Int24 *in = (Int24 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]].asInt() >> 16);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_SINT32) {
+      Int32 *in = (Int32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (signed char) ((in[info.inOffset[j]] >> 24) & 0x000000ff);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT32) {
+      Float32 *in = (Float32 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]] * 127.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+    else if (info.inFormat == RTAUDIO_FLOAT64) {
+      Float64 *in = (Float64 *)inBuffer;
+      for (unsigned int i=0; i<stream_.bufferSize; i++) {
+        for (j=0; j<info.channels; j++) {
+          out[info.outOffset[j]] = (signed char) (in[info.inOffset[j]] * 127.5 - 0.5);
+        }
+        in += info.inJump;
+        out += info.outJump;
+      }
+    }
+  }
+}
+
+//static inline uint16_t bswap_16(uint16_t x) { return (x>>8) | (x<<8); }
+//static inline uint32_t bswap_32(uint32_t x) { return (bswap_16(x&0xffff)<<16) | (bswap_16(x>>16)); }
+//static inline uint64_t bswap_64(uint64_t x) { return (((unsigned long long)bswap_32(x&0xffffffffull))<<32) | (bswap_32(x>>32)); }
+
+void RtApi :: byteSwapBuffer( char *buffer, unsigned int samples, RtAudioFormat format )
+{
+  register char val;
+  register char *ptr;
+
+  ptr = buffer;
+  if ( format == RTAUDIO_SINT16 ) {
+    for ( unsigned int i=0; i<samples; i++ ) {
+      // Swap 1st and 2nd bytes.
+      val = *(ptr);
+      *(ptr) = *(ptr+1);
+      *(ptr+1) = val;
+
+      // Increment 2 bytes.
+      ptr += 2;
+    }
+  }
+  else if ( format == RTAUDIO_SINT32 ||
+            format == RTAUDIO_FLOAT32 ) {
+    for ( unsigned int i=0; i<samples; i++ ) {
+      // Swap 1st and 4th bytes.
+      val = *(ptr);
+      *(ptr) = *(ptr+3);
+      *(ptr+3) = val;
+
+      // Swap 2nd and 3rd bytes.
+      ptr += 1;
+      val = *(ptr);
+      *(ptr) = *(ptr+1);
+      *(ptr+1) = val;
+
+      // Increment 3 more bytes.
+      ptr += 3;
+    }
+  }
+  else if ( format == RTAUDIO_SINT24 ) {
+    for ( unsigned int i=0; i<samples; i++ ) {
+      // Swap 1st and 3rd bytes.
+      val = *(ptr);
+      *(ptr) = *(ptr+2);
+      *(ptr+2) = val;
+
+      // Increment 2 more bytes.
+      ptr += 2;
+    }
+  }
+  else if ( format == RTAUDIO_FLOAT64 ) {
+    for ( unsigned int i=0; i<samples; i++ ) {
+      // Swap 1st and 8th bytes
+      val = *(ptr);
+      *(ptr) = *(ptr+7);
+      *(ptr+7) = val;
+
+      // Swap 2nd and 7th bytes
+      ptr += 1;
+      val = *(ptr);
+      *(ptr) = *(ptr+5);
+      *(ptr+5) = val;
+
+      // Swap 3rd and 6th bytes
+      ptr += 1;
+      val = *(ptr);
+      *(ptr) = *(ptr+3);
+      *(ptr+3) = val;
+
+      // Swap 4th and 5th bytes
+      ptr += 1;
+      val = *(ptr);
+      *(ptr) = *(ptr+1);
+      *(ptr+1) = val;
+
+      // Increment 5 more bytes.
+      ptr += 5;
+    }
+  }
+}
+
+  // Indentation settings for Vim and Emacs
+  //
+  // Local Variables:
+  // c-basic-offset: 2
+  // indent-tabs-mode: nil
+  // End:
+  //
+  // vim: et sts=2 sw=2
+
+#endif
diff --git a/drivers/speex/audio_stream_speex.cpp b/drivers/speex/audio_stream_speex.cpp
index 2440969345..1bb4952cc8 100644
--- a/drivers/speex/audio_stream_speex.cpp
+++ b/drivers/speex/audio_stream_speex.cpp
@@ -22,7 +22,7 @@ int AudioStreamPlaybackSpeex::mix(int16_t* p_buffer,int p_frames) {
 	//printf("update, loops %i, read ofs %i\n", (int)loops, read_ofs);
 	//printf("playing %i, paused %i\n", (int)playing, (int)paused);
 
-	if (!active || !playing || paused || !data.size())
+	if (!active || !playing || !data.size())
 		return 0;
 
 	/*
@@ -490,6 +490,8 @@ AudioStreamPlaybackSpeex::AudioStreamPlaybackSpeex() {
 	stream_channels=1;
 	stream_srate=1;
 	stream_minbuff_size=1;
+	playing=false;
+
 
 }
 
diff --git a/drivers/speex/audio_stream_speex.h b/drivers/speex/audio_stream_speex.h
index 570e846734..f0617b302f 100644
--- a/drivers/speex/audio_stream_speex.h
+++ b/drivers/speex/audio_stream_speex.h
@@ -29,7 +29,6 @@ class AudioStreamPlaybackSpeex : public AudioStreamPlayback {
 	bool loops;
 	int page_size;
 	bool playing;
-	bool paused;
 	bool packets_available;
 
 	void unload();
diff --git a/drivers/speex/config.h b/drivers/speex/config.h
index d31382702c..8c48e3b99d 100644
--- a/drivers/speex/config.h
+++ b/drivers/speex/config.h
@@ -1,52 +1,52 @@
-/*

-   Copyright (C) 2003 Commonwealth Scientific and Industrial Research

-   Organisation (CSIRO) Australia

-

-   Redistribution and use in source and binary forms, with or without

-   modification, are permitted provided that the following conditions

-   are met:

-

-   - Redistributions of source code must retain the above copyright

-   notice, this list of conditions and the following disclaimer.

-

-   - Redistributions in binary form must reproduce the above copyright

-   notice, this list of conditions and the following disclaimer in the

-   documentation and/or other materials provided with the distribution.

-

-   - Neither the name of CSIRO Australia nor the names of its

-   contributors may be used to endorse or promote products derived from

-   this software without specific prior written permission.

-

-   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS

-   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT

-   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A

-   PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE ORGANISATION OR

-   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,

-   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,

-   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR

-   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF

-   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING

-   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS

-   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

-*/

-

-#ifndef CONFIG_H

-#define CONFIG_H

-

-/* An inline macro is required for use of the inline keyword as not all C compilers support */

-/* inline.  It is officially C99 and C++ only */

-

-

-/* Use only fixed point arithmetic */

-

-//#ifdef _MSC_VER

-//#define inline _inline

-//#endif

-

-#define FIXED_POINT 1

-

-#ifdef _MSC_VER

-#define inline __inline

-#endif

-

-#endif /* ! CONFIG_H */

+/*
+   Copyright (C) 2003 Commonwealth Scientific and Industrial Research
+   Organisation (CSIRO) Australia
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+
+   - Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+
+   - Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+
+   - Neither the name of CSIRO Australia nor the names of its
+   contributors may be used to endorse or promote products derived from
+   this software without specific prior written permission.
+
+   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
+   PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE ORGANISATION OR
+   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef CONFIG_H
+#define CONFIG_H
+
+/* An inline macro is required for use of the inline keyword as not all C compilers support */
+/* inline.  It is officially C99 and C++ only */
+
+
+/* Use only fixed point arithmetic */
+
+//#ifdef _MSC_VER
+//#define inline _inline
+//#endif
+
+#define FIXED_POINT 1
+
+#ifdef _MSC_VER
+#define inline __inline
+#endif
+
+#endif /* ! CONFIG_H */
diff --git a/drivers/speex/lsp.h b/drivers/speex/lsp.h
index 648652fb9e..b55bd42f2c 100644
--- a/drivers/speex/lsp.h
+++ b/drivers/speex/lsp.h
@@ -1,64 +1,64 @@
-/*---------------------------------------------------------------------------*\

-Original Copyright

-	FILE........: AK2LSPD.H

-	TYPE........: Turbo C header file

-	COMPANY.....: Voicetronix

-	AUTHOR......: James Whitehall

-	DATE CREATED: 21/11/95

-

-Modified by Jean-Marc Valin

-

-    This file contains functions for converting Linear Prediction

-    Coefficients (LPC) to Line Spectral Pair (LSP) and back. Note that the

-    LSP coefficients are not in radians format but in the x domain of the

-    unit circle.

-

-\*---------------------------------------------------------------------------*/

-/**

-   @file lsp.h

-   @brief Line Spectral Pair (LSP) functions.

-*/

-/* Speex License:

-

-   Redistribution and use in source and binary forms, with or without

-   modification, are permitted provided that the following conditions

-   are met:

-   

-   - Redistributions of source code must retain the above copyright

-   notice, this list of conditions and the following disclaimer.

-   

-   - Redistributions in binary form must reproduce the above copyright

-   notice, this list of conditions and the following disclaimer in the

-   documentation and/or other materials provided with the distribution.

-   

-   - Neither the name of the Xiph.org Foundation nor the names of its

-   contributors may be used to endorse or promote products derived from

-   this software without specific prior written permission.

-   

-   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS

-   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT

-   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR

-   A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR

-   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,

-   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,

-   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR

-   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF

-   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING

-   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS

-   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

-*/

-

-#ifndef __AK2LSPD__

-#define __AK2LSPD__

-

-#include "arch.h"

-

-int lpc_to_lsp (spx_coef_t *a, int lpcrdr, spx_lsp_t *freq, int nb, spx_word16_t delta, char *stack);

-void lsp_to_lpc(spx_lsp_t *freq, spx_coef_t *ak, int lpcrdr, char *stack);

-

-/*Added by JMV*/

-void lsp_enforce_margin(spx_lsp_t *lsp, int len, spx_word16_t margin);

-

-void lsp_interpolate(spx_lsp_t *old_lsp, spx_lsp_t *new_lsp, spx_lsp_t *interp_lsp, int len, int subframe, int nb_subframes);

-

-#endif	/* __AK2LSPD__ */

+/*---------------------------------------------------------------------------*\
+Original Copyright
+	FILE........: AK2LSPD.H
+	TYPE........: Turbo C header file
+	COMPANY.....: Voicetronix
+	AUTHOR......: James Whitehall
+	DATE CREATED: 21/11/95
+
+Modified by Jean-Marc Valin
+
+    This file contains functions for converting Linear Prediction
+    Coefficients (LPC) to Line Spectral Pair (LSP) and back. Note that the
+    LSP coefficients are not in radians format but in the x domain of the
+    unit circle.
+
+\*---------------------------------------------------------------------------*/
+/**
+   @file lsp.h
+   @brief Line Spectral Pair (LSP) functions.
+*/
+/* Speex License:
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   
+   - Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+   
+   - Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+   
+   - Neither the name of the Xiph.org Foundation nor the names of its
+   contributors may be used to endorse or promote products derived from
+   this software without specific prior written permission.
+   
+   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+   A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
+   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef __AK2LSPD__
+#define __AK2LSPD__
+
+#include "arch.h"
+
+int lpc_to_lsp (spx_coef_t *a, int lpcrdr, spx_lsp_t *freq, int nb, spx_word16_t delta, char *stack);
+void lsp_to_lpc(spx_lsp_t *freq, spx_coef_t *ak, int lpcrdr, char *stack);
+
+/*Added by JMV*/
+void lsp_enforce_margin(spx_lsp_t *lsp, int len, spx_word16_t margin);
+
+void lsp_interpolate(spx_lsp_t *old_lsp, spx_lsp_t *new_lsp, spx_lsp_t *interp_lsp, int len, int subframe, int nb_subframes);
+
+#endif	/* __AK2LSPD__ */
diff --git a/drivers/speex/speex_bind.cpp b/drivers/speex/speex_bind.cpp
index 6e9eb638a2..d15bb3da8c 100644
--- a/drivers/speex/speex_bind.cpp
+++ b/drivers/speex/speex_bind.cpp
@@ -1,64 +1,64 @@
-

-#include "memory.h"

-#include "speex_bind.h"

-#include

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-void *speex_alloc (int size) {

-

-	uint8_t * mem = (uint8_t*)memalloc(size);

-	for(int i=0;i<size;i++)

-		mem[i]=0;

-	return mem;

-}

-

-void *speex_alloc_scratch (int size) {

-

-	return memalloc(size);

-}

-

-void *speex_realloc (void *ptr, int size) {

-

-	return memrealloc(ptr,size);

-}

-

-void speex_free (void *ptr) {

-

-	memfree(ptr);

-}

-

-void speex_free_scratch (void *ptr) {

-

-	memfree(ptr);

-}

-

-void _speex_fatal(const char *str, const char *file, int line) {

-

-	_err_print_error("SPEEX ERROR",p_file,p_line,str);

-}

-

-void speex_warning(const char *str) {

-

-	_err_print_error("SPEEX WARNING","",0,str);

-}

-

-void speex_warning_int(const char *str, int val) {

-

-	_err_print_error("SPEEX WARNING INT","Value",val,str);

-}

-

-void speex_notify(const char *str) {

-

-	print_line(str);

-}

-

-void _speex_putc(int ch, void *file) {

-

-	// will not putc, no.

-}

-

-#ifdef __cplusplus

-}

-#endif

+
+#include "memory.h"
+#include "speex_bind.h"
+#include
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void *speex_alloc (int size) {
+
+	uint8_t * mem = (uint8_t*)memalloc(size);
+	for(int i=0;i<size;i++)
+		mem[i]=0;
+	return mem;
+}
+
+void *speex_alloc_scratch (int size) {
+
+	return memalloc(size);
+}
+
+void *speex_realloc (void *ptr, int size) {
+
+	return memrealloc(ptr,size);
+}
+
+void speex_free (void *ptr) {
+
+	memfree(ptr);
+}
+
+void speex_free_scratch (void *ptr) {
+
+	memfree(ptr);
+}
+
+void _speex_fatal(const char *str, const char *file, int line) {
+
+	_err_print_error("SPEEX ERROR",p_file,p_line,str);
+}
+
+void speex_warning(const char *str) {
+
+	_err_print_error("SPEEX WARNING","",0,str);
+}
+
+void speex_warning_int(const char *str, int val) {
+
+	_err_print_error("SPEEX WARNING INT","Value",val,str);
+}
+
+void speex_notify(const char *str) {
+
+	print_line(str);
+}
+
+void _speex_putc(int ch, void *file) {
+
+	// will not putc, no.
+}
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/drivers/speex/speex_bind.h b/drivers/speex/speex_bind.h
index e842960d3c..c928430a33 100644
--- a/drivers/speex/speex_bind.h
+++ b/drivers/speex/speex_bind.h
@@ -1,48 +1,48 @@
-#ifndef SPEEX_BIND_H

-#define SPEEX_BIND_H

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

-#define OVERRIDE_SPEEX_ALLOC

-#define OVERRIDE_SPEEX_ALLOC_SCRATCH

-#define OVERRIDE_SPEEX_REALLOC

-#define OVERRIDE_SPEEX_FREE

-#define OVERRIDE_SPEEX_FREE_SCRATCH

-#define OVERRIDE_SPEEX_FATAL

-#define OVERRIDE_SPEEX_WARNING

-#define OVERRIDE_SPEEX_WARNING_INT

-#define OVERRIDE_SPEEX_NOTIFY

-#define OVERRIDE_SPEEX_PUTC

-

-void *speex_alloc (int size);

-void *speex_alloc_scratch (int size);

-void *speex_realloc (void *ptr, int size);

-void speex_free (void *ptr);

-void speex_free_scratch (void *ptr);

-void _speex_fatal(const char *str, const char *file, int line);

-void speex_warning(const char *str);

-void speex_warning_int(const char *str, int val);

-void speex_notify(const char *str);

-void _speex_putc(int ch, void *file);

-

-

-*/

-#define RELEASE

-#define SPEEX_PI 3.14159265358979323846

-

-#ifdef _MSC_VER

-#define SPEEX_INLINE __inline

-#else

-#define SPEEX_INLINE inline

-#endif

-

-

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif // SPEEX_BIND_H

+#ifndef SPEEX_BIND_H
+#define SPEEX_BIND_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+#define OVERRIDE_SPEEX_ALLOC
+#define OVERRIDE_SPEEX_ALLOC_SCRATCH
+#define OVERRIDE_SPEEX_REALLOC
+#define OVERRIDE_SPEEX_FREE
+#define OVERRIDE_SPEEX_FREE_SCRATCH
+#define OVERRIDE_SPEEX_FATAL
+#define OVERRIDE_SPEEX_WARNING
+#define OVERRIDE_SPEEX_WARNING_INT
+#define OVERRIDE_SPEEX_NOTIFY
+#define OVERRIDE_SPEEX_PUTC
+
+void *speex_alloc (int size);
+void *speex_alloc_scratch (int size);
+void *speex_realloc (void *ptr, int size);
+void speex_free (void *ptr);
+void speex_free_scratch (void *ptr);
+void _speex_fatal(const char *str, const char *file, int line);
+void speex_warning(const char *str);
+void speex_warning_int(const char *str, int val);
+void speex_notify(const char *str);
+void _speex_putc(int ch, void *file);
+
+
+*/
+#define RELEASE
+#define SPEEX_PI 3.14159265358979323846
+
+#ifdef _MSC_VER
+#define SPEEX_INLINE __inline
+#else
+#define SPEEX_INLINE inline
+#endif
+
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif // SPEEX_BIND_H
diff --git a/drivers/theora/video_stream_theora.cpp b/drivers/theora/video_stream_theora.cpp
index 214185cf88..bea49e34b7 100644
--- a/drivers/theora/video_stream_theora.cpp
+++ b/drivers/theora/video_stream_theora.cpp
@@ -1,16 +1,12 @@
 #ifdef THEORA_ENABLED
-#if 0
+
 #include "video_stream_theora.h"
 #include "os/os.h"
 #include "yuv2rgb.h"
+#include "globals.h"
 
 
-AudioStream::UpdateMode VideoStreamTheora::get_update_mode() const {
-
-	return UPDATE_IDLE;
-};
-
-int VideoStreamTheora::	buffer_data() {
+int VideoStreamPlaybackTheora::	buffer_data() {
   char *buffer=ogg_sync_buffer(&oy,4096);
   int bytes=file->get_buffer((uint8_t*)buffer, 4096);
 
@@ -18,33 +14,13 @@ int VideoStreamTheora::	buffer_data() {
   return(bytes);
 }
 
-int VideoStreamTheora::queue_page(ogg_page *page){
+int VideoStreamPlaybackTheora::queue_page(ogg_page *page){
   if(theora_p)ogg_stream_pagein(&to,page);
   if(vorbis_p)ogg_stream_pagein(&vo,page);
   return 0;
 }
 
-Image VideoStreamTheora::peek_frame() const {
-
-	if (frames_pending == 0)
-		return Image();
-	return Image(size.x, size.y, 0, format, frame_data);
-};
-
-Image VideoStreamTheora::pop_frame() {
-
-	Image ret = peek_frame();
-	frames_pending = 0;
-
-	return ret;
-};
-
-int VideoStreamTheora::get_pending_frame_count() const {
-
-	return frames_pending;
-};
-
-void VideoStreamTheora::video_write(void){
+void VideoStreamPlaybackTheora::video_write(void){
 	th_ycbcr_buffer yuv;
 	int y_offset, uv_offset;
 	th_decode_ycbcr_out(td,yuv);
@@ -78,25 +54,31 @@ void VideoStreamTheora::video_write(void){
 
 	int pitch = 4;
 	frame_data.resize(size.x * size.y * pitch);
-	DVector<uint8_t>::Write w = frame_data.write();
-	char* dst = (char*)w.ptr();
+	{
+		DVector<uint8_t>::Write w = frame_data.write();
+		char* dst = (char*)w.ptr();
 
-	uv_offset=(ti.pic_x/2)+(yuv[1].stride)*(ti.pic_y/2);
+		uv_offset=(ti.pic_x/2)+(yuv[1].stride)*(ti.pic_y/2);
 
-	if (px_fmt == TH_PF_444) {
+		if (px_fmt == TH_PF_444) {
 
-		yuv444_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[1].data, (uint8_t*)yuv[2].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
+			yuv444_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[1].data, (uint8_t*)yuv[2].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
 
-	} else if (px_fmt == TH_PF_422) {
+		} else if (px_fmt == TH_PF_422) {
 
-		yuv422_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[1].data, (uint8_t*)yuv[2].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
+			yuv422_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[1].data, (uint8_t*)yuv[2].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
 
-	} else if (px_fmt == TH_PF_420) {
+		} else if (px_fmt == TH_PF_420) {
 
-		yuv420_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[2].data, (uint8_t*)yuv[1].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
-	};
+			yuv420_2_rgb8888((uint8_t*)dst, (uint8_t*)yuv[0].data, (uint8_t*)yuv[2].data, (uint8_t*)yuv[1].data, size.x, size.y, yuv[0].stride, yuv[1].stride, size.x<<2, 0);
+		};
 
-	format = Image::FORMAT_RGBA;
+		format = Image::FORMAT_RGBA;
+	}
+
+	Image img(size.x,size.y,0,Image::FORMAT_RGBA,frame_data); //zero copy image creation
+
+	texture->set_data(img); //zero copy send to visual server
 
 	/*
 
@@ -194,7 +176,7 @@ void VideoStreamTheora::video_write(void){
 	frames_pending = 1;
 }
 
-void VideoStreamTheora::clear() {
+void VideoStreamPlaybackTheora::clear() {
 
 	if (file_name == "")
 		return;
@@ -218,7 +200,7 @@ void VideoStreamTheora::clear() {
 	}
 	ogg_sync_clear(&oy);
 
-	file_name = "";
+	//file_name = "";
 
 	theora_p = 0;
 	vorbis_p = 0;
@@ -229,7 +211,7 @@ void VideoStreamTheora::clear() {
 	playing = false;
 };
 
-void VideoStreamTheora::set_file(const String& p_file) {
+void VideoStreamPlaybackTheora::set_file(const String& p_file) {
 
 	ogg_packet op;
 	th_setup_info    *ts = NULL;
@@ -241,7 +223,7 @@ void VideoStreamTheora::set_file(const String& p_file) {
 	file = FileAccess::open(p_file, FileAccess::READ);
 	ERR_FAIL_COND(!file);
 
-	audio_frames_wrote = 0;
+
 
 	ogg_sync_init(&oy);
 
@@ -256,6 +238,8 @@ void VideoStreamTheora::set_file(const String& p_file) {
 	/* Ogg file open; parse the headers */
 	/* Only interested in Vorbis/Theora streams */
 	int stateflag = 0;
+
+    int audio_track_skip=audio_track;
 	while(!stateflag){
 		int ret=buffer_data();
 		if(ret==0)break;
@@ -282,8 +266,14 @@ void VideoStreamTheora::set_file(const String& p_file) {
 				theora_p=1;
 			}else if(!vorbis_p && vorbis_synthesis_headerin(&vi,&vc,&op)>=0){
 				/* it is vorbis */
-				copymem(&vo,&test,sizeof(test));
-				vorbis_p=1;
+                if (audio_track_skip) {
+                    vorbis_info_clear(&vi);
+                    vorbis_comment_clear(&vc);
+                    audio_track_skip--;
+                } else {
+                    copymem(&vo,&test,sizeof(test));
+                    vorbis_p=1;
+                }
 			}else{
 				/* whatever it is, we don't care about it */
 				ogg_stream_clear(&test);
@@ -386,6 +376,8 @@ void VideoStreamTheora::set_file(const String& p_file) {
 		size.x = w;
 		size.y = h;
 
+		texture->create(w,h,Image::FORMAT_RGBA,Texture::FLAG_FILTER|Texture::FLAG_VIDEO_SURFACE);
+
 	}else{
 		/* tear down the partial theora setup */
 		th_info_clear(&ti);
@@ -399,7 +391,7 @@ void VideoStreamTheora::set_file(const String& p_file) {
 		vorbis_block_init(&vd,&vb);
 		fprintf(stderr,"Ogg logical stream %lx is Vorbis %d channel %ld Hz audio.\n",
 				vo.serialno,vi.channels,vi.rate);
-		_setup(vi.channels, vi.rate);
+		//_setup(vi.channels, vi.rate);
 	}else{
 		/* tear down the partial vorbis setup */
 		vorbis_info_clear(&vi);
@@ -411,227 +403,299 @@ void VideoStreamTheora::set_file(const String& p_file) {
 	time=0;
 };
 
-float VideoStreamTheora::get_time() const {
+float VideoStreamPlaybackTheora::get_time() const {
 
 	//print_line("total: "+itos(get_total())+" todo: "+itos(get_todo()));
 	//return MAX(0,time-((get_total())/(float)vi.rate));
-	return time-((get_total())/(float)vi.rate);
+	return time-AudioServer::get_singleton()->get_output_delay()-delay_compensation;//-((get_total())/(float)vi.rate);
 };
 
-void VideoStreamTheora::update() {
+Ref<Texture> VideoStreamPlaybackTheora::get_texture() {
+
+	return texture;
+}
+
+void VideoStreamPlaybackTheora::update(float p_delta) {
 
 	if (!playing) {
 		//printf("not playing\n");
 		return;
 	};
 
-	double ctime =AudioServer::get_singleton()->get_mix_time();
+	//double ctime =AudioServer::get_singleton()->get_mix_time();
 
-	if (last_update_time) {
-		double delta = (ctime-last_update_time);
-		time+=delta;
-		//print_line("delta: "+rtos(delta));
-	}
-	last_update_time=ctime;
+	//print_line("play "+rtos(p_delta));
+	time+=p_delta;
 
+	if (videobuf_time>get_time())
+		return; //no new frames need to be produced
 
-	int audio_todo = get_todo();
-	ogg_packet op;
-	int audio_pending = 0;
+	bool frame_done=false;
 
+	while (!frame_done) {
+		//a frame needs to be produced
 
-	while (vorbis_p && audio_todo) {
-		int ret;
-		float **pcm;
-
-		/* if there's pending, decoded audio, grab it */
-		if ((ret=vorbis_synthesis_pcmout(&vd,&pcm))>0) {
-
-			audio_pending = ret;
-			int16_t* out = get_write_buffer();
-			int count = 0;
-			int to_read = MIN(ret, audio_todo);
-			for (int i=0; i<to_read; i++) {
-
-				for(int j=0;j<vi.channels;j++){
-					int val=Math::fast_ftoi(pcm[j][i]*32767.f);
-					if(val>32767)val=32767;
-					if(val<-32768)val=-32768;
-					out[count++] = val;
-				};
-			};
-			int tr = vorbis_synthesis_read(&vd, to_read);
-			audio_todo -= to_read;
-			audio_frames_wrote += to_read;
-			write(to_read);
-			audio_pending -= to_read;
-			if (audio_todo==0)
-				buffering=false;
+		ogg_packet op;
+		bool audio_pending = false;
 
 
-		} else {
+		while (vorbis_p) {
+			int ret;
+			float **pcm;
+
+			bool buffer_full=false;
+
+			/* if there's pending, decoded audio, grab it */
+			if ((ret=vorbis_synthesis_pcmout(&vd,&pcm))>0) {
+
+
+
+				const int AUXBUF_LEN=4096;
+				int to_read = ret;
+				int16_t aux_buffer[AUXBUF_LEN];
+
+				while(to_read) {
+
+					int m = MIN(AUXBUF_LEN/vi.channels,to_read);
+
+					int count = 0;
+
+					for(int j=0;j<m;j++){
+						for(int i=0;i<vi.channels;i++){
+
+							int val=Math::fast_ftoi(pcm[i][j]*32767.f);
+							if(val>32767)val=32767;
+							if(val<-32768)val=-32768;
+							aux_buffer[count++] = val;
+						}
+					}
+
+					if (mix_callback) {
+						int mixed = mix_callback(mix_udata,aux_buffer,m);
+						to_read-=mixed;
+						if (mixed!=m) { //could mix no more
+							buffer_full=true;
+							break;
+						}
+					} else {
+						to_read-=m; //just pretend we sent the audio
+					}
+
 
-			/* no pending audio; is there a pending packet to decode? */
-			if (ogg_stream_packetout(&vo,&op)>0){
-				if(vorbis_synthesis(&vb,&op)==0) { /* test for success! */
-					vorbis_synthesis_blockin(&vd,&vb);
 				}
-			} else {  /* we need more data; break out to suck in another page */
-				//printf("need moar data\n");
+
+
+				int tr = vorbis_synthesis_read(&vd, ret-to_read);
+
+				audio_pending=true;
+
+
+			} else {
+
+				/* no pending audio; is there a pending packet to decode? */
+				if (ogg_stream_packetout(&vo,&op)>0){
+					if(vorbis_synthesis(&vb,&op)==0) { /* test for success! */
+						vorbis_synthesis_blockin(&vd,&vb);
+					}
+				} else {  /* we need more data; break out to suck in another page */
+					//printf("need moar data\n");
+					break;
+				};
+			}
+
+			if (buffer_full)
 				break;
-			};
 		}
-	}
 
-	while(theora_p && !videobuf_ready){
-		/* theora is one in, one out... */
-		if(ogg_stream_packetout(&to,&op)>0){
+		while(theora_p && !frame_done){
+			/* theora is one in, one out... */
+			if(ogg_stream_packetout(&to,&op)>0){
 
 
-			if(pp_inc){
-				pp_level+=pp_inc;
-				th_decode_ctl(td,TH_DECCTL_SET_PPLEVEL,&pp_level,
-							  sizeof(pp_level));
-				pp_inc=0;
-			}
-			/*HACK: This should be set after a seek or a gap, but we might not have
-			a granulepos for the first packet (we only have them for the last
-			packet on a page), so we just set it as often as we get it.
-			To do this right, we should back-track from the last packet on the
-			page and compute the correct granulepos for the first packet after
-			a seek or a gap.*/
-			if(op.granulepos>=0){
-				th_decode_ctl(td,TH_DECCTL_SET_GRANPOS,&op.granulepos,
-							  sizeof(op.granulepos));
-			}
-			ogg_int64_t videobuf_granulepos;
-			if(th_decode_packetin(td,&op,&videobuf_granulepos)==0){
-				videobuf_time=th_granule_time(td,videobuf_granulepos);
-				//printf("frame time %f, play time %f, ready %i\n", (float)videobuf_time, get_time(), videobuf_ready);
-
-				/* is it already too old to be useful?  This is only actually
-				 useful cosmetically after a SIGSTOP.  Note that we have to
-				 decode the frame even if we don't show it (for now) due to
-				 keyframing.  Soon enough libtheora will be able to deal
-				 with non-keyframe seeks.  */
-
-				if(videobuf_time>=get_time())
-					videobuf_ready=1;
-				else{
-					/*If we are too slow, reduce the pp level.*/
-					pp_inc=pp_level>0?-1:0;
+				if(pp_inc){
+					pp_level+=pp_inc;
+					th_decode_ctl(td,TH_DECCTL_SET_PPLEVEL,&pp_level,
+								  sizeof(pp_level));
+					pp_inc=0;
+				}
+				/*HACK: This should be set after a seek or a gap, but we might not have
+				a granulepos for the first packet (we only have them for the last
+				packet on a page), so we just set it as often as we get it.
+				To do this right, we should back-track from the last packet on the
+				page and compute the correct granulepos for the first packet after
+				a seek or a gap.*/
+				if(op.granulepos>=0){
+					th_decode_ctl(td,TH_DECCTL_SET_GRANPOS,&op.granulepos,
+								  sizeof(op.granulepos));
+				}
+				ogg_int64_t videobuf_granulepos;
+				if(th_decode_packetin(td,&op,&videobuf_granulepos)==0){
+					videobuf_time=th_granule_time(td,videobuf_granulepos);
+
+					//printf("frame time %f, play time %f, ready %i\n", (float)videobuf_time, get_time(), videobuf_ready);
+
+					/* is it already too old to be useful?  This is only actually
+					 useful cosmetically after a SIGSTOP.  Note that we have to
+					 decode the frame even if we don't show it (for now) due to
+					 keyframing.  Soon enough libtheora will be able to deal
+					 with non-keyframe seeks.  */
+
+					if(videobuf_time>=get_time())
+						frame_done=true;
+					else{
+						/*If we are too slow, reduce the pp level.*/
+						pp_inc=pp_level>0?-1:0;
+					}
 				}
-			}
-
-		} else
-			break;
-	}
 
-	if (/*!videobuf_ready && */ audio_pending == 0 && file->eof_reached()) {
-		printf("video done, stopping\n");
-		stop();
-		return;
-	};
+			} else
+				break;
+		}
 
-	if (!videobuf_ready || audio_todo > 0){
-		/* no data yet for somebody.  Grab another page */
+		if (file && /*!videobuf_ready && */ file->eof_reached()) {
+			printf("video done, stopping\n");
+			stop();
+			return;
+		};
+	#if 0
+		if (!videobuf_ready || audio_todo > 0){
+			/* no data yet for somebody.  Grab another page */
 
-		buffer_data();
-		while(ogg_sync_pageout(&oy,&og)>0){
-			queue_page(&og);
+			buffer_data();
+			while(ogg_sync_pageout(&oy,&og)>0){
+				queue_page(&og);
+			}
 		}
-	}
+	#else
+		if (!frame_done){
+			//what's the point of waiting for audio to grab a page?
 
-	/* If playback has begun, top audio buffer off immediately. */
-	//if(stateflag) audio_write_nonblocking();
+			buffer_data();
+			while(ogg_sync_pageout(&oy,&og)>0){
+				queue_page(&og);
+			}
+		}
+	#endif
+		/* If playback has begun, top audio buffer off immediately. */
+		//if(stateflag) audio_write_nonblocking();
 
-	/* are we at or past time for this video frame? */
-	if(videobuf_ready && videobuf_time<=get_time()){
+		/* are we at or past time for this video frame? */
+		if(videobuf_ready && videobuf_time<=get_time()){
 
-		video_write();
-		videobuf_ready=0;
-	} else {
-		//printf("frame at %f not ready (time %f), ready %i\n", (float)videobuf_time, get_time(), videobuf_ready);
-	}
+			//video_write();
+			//videobuf_ready=0;
+		} else {
+			//printf("frame at %f not ready (time %f), ready %i\n", (float)videobuf_time, get_time(), videobuf_ready);
+		}
 
-	float tdiff=videobuf_time-get_time();
-	/*If we have lots of extra time, increase the post-processing level.*/
-	if(tdiff>ti.fps_denominator*0.25/ti.fps_numerator){
-		pp_inc=pp_level<pp_level_max?1:0;
-	}
-	else if(tdiff<ti.fps_denominator*0.05/ti.fps_numerator){
-		pp_inc=pp_level>0?-1:0;
+		float tdiff=videobuf_time-get_time();
+		/*If we have lots of extra time, increase the post-processing level.*/
+		if(tdiff>ti.fps_denominator*0.25/ti.fps_numerator){
+			pp_inc=pp_level<pp_level_max?1:0;
+		}
+		else if(tdiff<ti.fps_denominator*0.05/ti.fps_numerator){
+			pp_inc=pp_level>0?-1:0;
+		}
 	}
-};
 
-bool VideoStreamTheora::_can_mix() const {
+	video_write();
 
-	return !buffering;
 };
 
-void VideoStreamTheora::play() {
+
+void VideoStreamPlaybackTheora::play() {
 
 	if (!playing)
-		last_update_time=0;
+		time=0;
 	playing = true;
+	delay_compensation=Globals::get_singleton()->get("audio/video_delay_compensation_ms");
+	delay_compensation/=1000.0;
+
 };
 
-void VideoStreamTheora::stop() {
+void VideoStreamPlaybackTheora::stop() {
 
+	if (playing) {
+		clear();
+		set_file(file_name); //reset
+	}
 	playing = false;
-	last_update_time=0;
+	time=0;
 };
 
-bool VideoStreamTheora::is_playing() const {
+bool VideoStreamPlaybackTheora::is_playing() const {
 
 	return playing;
 };
 
-void VideoStreamTheora::set_paused(bool p_paused) {
+void VideoStreamPlaybackTheora::set_paused(bool p_paused) {
 
 	playing = !p_paused;
 };
 
-bool VideoStreamTheora::is_paused(bool p_paused) const {
+bool VideoStreamPlaybackTheora::is_paused(bool p_paused) const {
 
 	return playing;
 };
 
-void VideoStreamTheora::set_loop(bool p_enable) {
+void VideoStreamPlaybackTheora::set_loop(bool p_enable) {
 
 };
 
-bool VideoStreamTheora::has_loop() const {
+bool VideoStreamPlaybackTheora::has_loop() const {
 
 	return false;
 };
 
-float VideoStreamTheora::get_length() const {
+float VideoStreamPlaybackTheora::get_length() const {
 
 	return 0;
 };
 
-String VideoStreamTheora::get_stream_name() const {
+String VideoStreamPlaybackTheora::get_stream_name() const {
 
 	return "";
 };
 
-int VideoStreamTheora::get_loop_count() const {
+int VideoStreamPlaybackTheora::get_loop_count() const {
 
 	return 0;
 };
 
-float VideoStreamTheora::get_pos() const {
+float VideoStreamPlaybackTheora::get_pos() const {
 
 	return get_time();
 };
 
-void VideoStreamTheora::seek_pos(float p_time) {
+void VideoStreamPlaybackTheora::seek_pos(float p_time) {
 
 	// no
 };
 
-VideoStreamTheora::VideoStreamTheora() {
+void VideoStreamPlaybackTheora::set_mix_callback(AudioMixCallback p_callback,void *p_userdata) {
+
+	mix_callback=p_callback;
+	mix_udata=p_userdata;
+}
+
+int VideoStreamPlaybackTheora::get_channels() const{
+
+	return vi.channels;
+}
+
+void VideoStreamPlaybackTheora::set_audio_track(int p_idx) {
+
+    audio_track=p_idx;
+}
+
+int VideoStreamPlaybackTheora::get_mix_rate() const{
+
+	return vi.rate;
+}
+
+
+
+VideoStreamPlaybackTheora::VideoStreamPlaybackTheora() {
 
 	file = NULL;
 	theora_p = 0;
@@ -640,11 +704,16 @@ VideoStreamTheora::VideoStreamTheora() {
 	playing = false;
 	frames_pending = 0;
 	videobuf_time = 0;
-	last_update_time =0;
+
 	buffering=false;
+	texture = Ref<ImageTexture>( memnew(ImageTexture ));
+	mix_callback=NULL;
+	mix_udata=NULL;
+    audio_track=0;
+	delay_compensation=0;
 };
 
-VideoStreamTheora::~VideoStreamTheora() {
+VideoStreamPlaybackTheora::~VideoStreamPlaybackTheora() {
 
 	clear();
 
@@ -653,10 +722,16 @@ VideoStreamTheora::~VideoStreamTheora() {
 };
 
 
-RES ResourceFormatLoaderVideoStreamTheora::load(const String &p_path,const String& p_original_path) {
+RES ResourceFormatLoaderVideoStreamTheora::load(const String &p_path,const String& p_original_path, Error *r_error) {
+	if (r_error)
+		*r_error=ERR_FILE_CANT_OPEN;
 
 	VideoStreamTheora *stream = memnew(VideoStreamTheora);
 	stream->set_file(p_path);
+
+	if (r_error)
+		*r_error=OK;
+
 	return Ref<VideoStreamTheora>(stream);
 }
 
@@ -666,16 +741,16 @@ void ResourceFormatLoaderVideoStreamTheora::get_recognized_extensions(List<Strin
 	p_extensions->push_back("ogv");
 }
 bool ResourceFormatLoaderVideoStreamTheora::handles_type(const String& p_type) const {
-	return (p_type=="AudioStream" || p_type=="VideoStreamTheora");
+	return (p_type=="VideoStream" || p_type=="VideoStreamTheora");
 }
 
 String ResourceFormatLoaderVideoStreamTheora::get_resource_type(const String &p_path) const {
 
 	String exl=p_path.extension().to_lower();
 	if (exl=="ogm" || exl=="ogv")
-		return "AudioStreamTheora";
+		return "VideoStreamTheora";
 	return "";
 }
 
 #endif
-#endif
+
diff --git a/drivers/theora/video_stream_theora.h b/drivers/theora/video_stream_theora.h
index 12aac731fc..95c7fe88f6 100644
--- a/drivers/theora/video_stream_theora.h
+++ b/drivers/theora/video_stream_theora.h
@@ -10,9 +10,9 @@
 #include "io/resource_loader.h"
 #include "scene/resources/video_stream.h"
 
-class VideoStreamTheora : public VideoStream {
+class VideoStreamPlaybackTheora : public VideoStreamPlayback {
 
-	OBJ_TYPE(VideoStreamTheora, VideoStream);
+	OBJ_TYPE(VideoStreamPlaybackTheora, VideoStreamPlayback);
 
 	enum {
 		MAX_FRAMES = 4,
@@ -58,16 +58,19 @@ class VideoStreamTheora : public VideoStream {
 
 	double last_update_time;
 	double time;
+	double delay_compensation;
 
-protected:
+	Ref<ImageTexture> texture;
 
-	virtual UpdateMode get_update_mode() const;
-	virtual void update();
+	AudioMixCallback mix_callback;
+	void* mix_udata;
 
-	void clear();
+    int audio_track;
 
-	virtual bool _can_mix() const;
+protected:
 
+	void clear();
+	
 public:
 
 	virtual void play();
@@ -92,12 +95,43 @@ public:
 
 	void set_file(const String& p_file);
 
-	int get_pending_frame_count() const;
-	Image pop_frame();
-	Image peek_frame() const;
+	virtual Ref<Texture> get_texture();
+	virtual void update(float p_delta);
+
+	virtual void set_mix_callback(AudioMixCallback p_callback,void *p_userdata);
+	virtual int get_channels() const;
+	virtual int get_mix_rate() const;
+
+	virtual void set_audio_track(int p_idx);
+
+	VideoStreamPlaybackTheora();
+	~VideoStreamPlaybackTheora();
+};
+
+
+
+class VideoStreamTheora : public VideoStream {
+
+	OBJ_TYPE(VideoStreamTheora,VideoStream);
+
+	String file;
+    int audio_track;
+
+
+public:
+
+	Ref<VideoStreamPlayback> instance_playback() {
+		Ref<VideoStreamPlaybackTheora> pb = memnew( VideoStreamPlaybackTheora );
+        pb->set_audio_track(audio_track);
+		pb->set_file(file);
+		return pb;
+	}
+
+	void set_file(const String& p_file) { file=p_file; }
+    void set_audio_track(int p_track) { audio_track=p_track; }
+
+    VideoStreamTheora() { audio_track=0; }
 
-	VideoStreamTheora();
-	~VideoStreamTheora();
 };
 
 class ResourceFormatLoaderVideoStreamTheora : public ResourceFormatLoader {
diff --git a/drivers/theoraplayer/SCsub b/drivers/theoraplayer/SCsub
deleted file mode 100644
index 09fb13d8e9..0000000000
--- a/drivers/theoraplayer/SCsub
+++ /dev/null
@@ -1,106 +0,0 @@
-Import("env")
-
-import string
-
-sources = string.split("""
-src/TheoraVideoClip.cpp
-src/FFmpeg/TheoraVideoClip_FFmpeg.cpp
-src/TheoraAsync.cpp
-src/TheoraAudioInterface.cpp
-src/TheoraException.cpp
-src/TheoraWorkerThread.cpp
-src/TheoraVideoManager.cpp
-src/TheoraTimer.cpp
-src/TheoraUtil.cpp
-src/TheoraDataSource.cpp
-src/TheoraAudioPacketQueue.cpp
-src/TheoraFrameQueue.cpp
-src/Theora/TheoraVideoClip_Theora.cpp
-src/YUV/yuv_util.c
-src/YUV/libyuv/src/row_any.cc
-src/YUV/libyuv/src/compare_common.cc
-src/YUV/libyuv/src/scale_neon.cc
-src/YUV/libyuv/src/planar_functions.cc
-src/YUV/libyuv/src/compare.cc
-src/YUV/libyuv/src/scale_mips.cc
-src/YUV/libyuv/src/scale_posix.cc
-src/YUV/libyuv/src/row_posix.cc
-src/YUV/libyuv/src/row_win.cc
-src/YUV/libyuv/src/compare_neon.cc
-src/YUV/libyuv/src/convert_from_argb.cc
-src/YUV/libyuv/src/mjpeg_validate.cc
-src/YUV/libyuv/src/convert_from.cc
-src/YUV/libyuv/src/rotate_neon.cc
-src/YUV/libyuv/src/row_neon.cc
-src/YUV/libyuv/src/rotate_mips.cc
-src/YUV/libyuv/src/compare_posix.cc
-src/YUV/libyuv/src/row_mips.cc
-src/YUV/libyuv/src/scale.cc
-src/YUV/libyuv/src/scale_argb.cc
-src/YUV/libyuv/src/mjpeg_decoder.cc
-src/YUV/libyuv/src/scale_win.cc
-src/YUV/libyuv/src/scale_common.cc
-src/YUV/libyuv/src/scale_argb_neon.cc
-src/YUV/libyuv/src/row_common.cc
-src/YUV/libyuv/src/convert.cc
-src/YUV/libyuv/src/format_conversion.cc
-src/YUV/libyuv/src/rotate_argb.cc
-src/YUV/libyuv/src/rotate.cc
-src/YUV/libyuv/src/convert_argb.cc
-src/YUV/libyuv/src/cpu_id.cc
-src/YUV/libyuv/src/video_common.cc
-src/YUV/libyuv/src/convert_to_argb.cc
-src/YUV/libyuv/src/compare_win.cc
-src/YUV/libyuv/src/convert_to_i420.cc
-src/YUV/libyuv/src/convert_jpeg.cc
-src/YUV/libyuv/yuv_libyuv.c
-src/YUV/android/cpu-features.c
-src/YUV/C/yuv420_grey_c.c
-src/YUV/C/yuv420_yuv_c.c
-src/YUV/C/yuv420_rgb_c.c
-src/TheoraVideoFrame.cpp
-""")
-
-env_theora = env.Clone()
-
-if env["platform"] == "iphone":
-	sources.append("src/AVFoundation/TheoraVideoClip_AVFoundation.mm")
-	env.Append(LINKFLAGS=['-framework', 'CoreVideo', '-framework', 'CoreMedia', '-framework', 'AVFoundation'])
-	if env["target"] == "release":
-		env_theora.Append(CPPFLAGS=["-D_IOS", "-D__ARM_NEON__", "-fstrict-aliasing", "-fmessage-length=210", "-fdiagnostics-show-note-include-stack", "-fmacro-backtrace-limit=0", "-fcolor-diagnostics", "-Wno-trigraphs", "-fpascal-strings", "-fvisibility=hidden", "-fvisibility-inlines-hidden"])
-
-env_theora.Append(CPPFLAGS=["-D_LIB", "-D__THEORA"]) # removed -D_YUV_C
-env_theora.Append(CPPFLAGS=["-D_YUV_LIBYUV"])
-#env_theora.Append(CPPFLAGS=["-D_YUV_C"])
-
-if env["platform"] == "iphone":
-	env_theora.Append(CPPFLAGS=["-D__AVFOUNDATION"])
-else:
-	pass
-	#env_theora.Append(CPPFLAGS=["-D__FFMPEG"])
-
-if env["platform"] == "android":
-	env_theora.Append(CPPFLAGS=["-D_ANDROID"])
-
-if env["platform"] == "winrt":
-		env_theora.Append(CPPFLAGS=["-D_WINRT"])
-
-env_theora.Append(CPPPATH=["#drivers/theoraplayer/include/theoraplayer", "#drivers/theoraplayer/src/YUV", "#drivers/theoraplayer/src/YUV/libyuv/include", "#drivers/theoraplayer/src/Theora", "#drivers/theoraplayer/src/AVFoundation"])
-
-objs = []
-
-env_theora.add_source_files(objs, ["video_stream_theoraplayer.cpp"])
-
-if env['use_theoraplayer_binary'] == "yes":
-	if env["platform"] == "iphone":
-		env.Append(LIBPATH=['#drivers/theoraplayer/lib/ios'])
-		env.Append(LIBS=['theoraplayer', 'ogg', 'theora', 'tremor'])
-	if env["platform"] == "windows":
-		env.Append(LIBPATH=['#drivers/theoraplayer/lib/windows'])
-		env.Append(LINKFLAGS=['libtheoraplayer_static.lib', 'libogg.lib', 'libtheora.lib', 'libvorbis.lib'])
-else:
-	env_theora.add_source_files(objs, sources)
-
-env.drivers_sources += objs
-
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraAsync.h b/drivers/theoraplayer/include/theoraplayer/TheoraAsync.h
deleted file mode 100644
index 7f1b49b9af..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraAsync.h
+++ /dev/null
@@ -1,51 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraAsync_h
-#define _TheoraAsync_h
-
-#ifndef _WIN32
-#include <pthread.h>
-#endif
-
-/// @note Based on hltypes::Thread
-class TheoraMutex
-{
-public:
-	TheoraMutex();
-	~TheoraMutex();
-	void lock();
-	void unlock();
-		
-protected:
-	void* mHandle;
-		
-};
-
-/// @note Based on hltypes::Thread
-class TheoraThread
-{
-	TheoraMutex mRunningMutex;
-public:
-	TheoraThread();
-	virtual ~TheoraThread();
-	void start();
-	void stop();
-	void resume();
-	void pause();
-	bool isRunning();
-	virtual void execute() = 0;
-	void join();
-		
-protected:
-	void* mId;
-	volatile bool mRunning;
-		
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraAudioInterface.h b/drivers/theoraplayer/include/theoraplayer/TheoraAudioInterface.h
deleted file mode 100644
index aa03293806..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraAudioInterface.h
+++ /dev/null
@@ -1,51 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraAudioInterface_h
-#define _TheoraAudioInterface_h
-
-#include "TheoraExport.h"
-
-class TheoraVideoClip;
-
-
-/**
- This is the class that serves as an interface between the library's audio
- output and the audio playback library of your choice.
- The class gets mono or stereo PCM data in in floating point data
- */
-class TheoraPlayerExport TheoraAudioInterface
-{
-public:
-	//! PCM frequency, usualy 44100 Hz
-	int mFreq;
-	//! Mono or stereo
-	int mNumChannels;
-	//! Pointer to the parent TheoraVideoClip object
-	TheoraVideoClip* mClip;
-	
-	TheoraAudioInterface(TheoraVideoClip* owner, int nChannels, int freq);
-	virtual ~TheoraAudioInterface();
-    //! A function that the TheoraVideoClip object calls once more audio packets are decoded
-    /*!
-	 \param data contains one or two channels of float PCM data in the range [-1,1]
-	 \param nSamples contains the number of samples that the data parameter contains in each channel
-	 */
-	virtual void insertData(float* data, int nSamples)=0;	
-};
-
-class TheoraPlayerExport TheoraAudioInterfaceFactory
-{
-public:
-	//! VideoManager calls this when creating a new TheoraVideoClip object
-	virtual TheoraAudioInterface* createInstance(TheoraVideoClip* owner, int nChannels, int freq) = 0;
-};
-
-
-#endif
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraAudioPacketQueue.h b/drivers/theoraplayer/include/theoraplayer/TheoraAudioPacketQueue.h
deleted file mode 100644
index e0d17516e6..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraAudioPacketQueue.h
+++ /dev/null
@@ -1,48 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraAudioPacketQueue_h
-#define _TheoraAudioPacketQueue_h
-
-#include "TheoraExport.h"
-
-class TheoraAudioInterface;
-/**
- This is an internal structure which TheoraVideoClip_Theora uses to store audio packets
- */
-struct TheoraAudioPacket
-{
-	float* pcm;
-	int numSamples; //! size in number of float samples (stereo has twice the number of samples)
-	TheoraAudioPacket* next; // pointer to the next audio packet, to implement a linked list
-};
-
-/**
-    This is a Mutex object, used in thread syncronization.
- */
-class TheoraPlayerExport TheoraAudioPacketQueue
-{
-protected:
-	unsigned int mAudioFrequency, mNumAudioChannels;
-	TheoraAudioPacket* mTheoraAudioPacketQueue;
-	void _addAudioPacket(float* data, int numSamples);
-public:
-	TheoraAudioPacketQueue();
-	~TheoraAudioPacketQueue();
-	
-	float getAudioPacketQueueLength();
-	void addAudioPacket(float** buffer, int numSamples, float gain);
-	void addAudioPacket(float* buffer, int numSamples, float gain);
-	TheoraAudioPacket* popAudioPacket();
-	void destroyAudioPacket(TheoraAudioPacket* p);
-	void destroyAllAudioPackets();
-	
-	void flushAudioPackets(TheoraAudioInterface* audioInterface);
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraDataSource.h b/drivers/theoraplayer/include/theoraplayer/TheoraDataSource.h
deleted file mode 100644
index b7427e97d3..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraDataSource.h
+++ /dev/null
@@ -1,89 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraDataSource_h
-#define _TheoraDataSource_h
-
-#include <stdio.h>
-#include <string>
-#include "TheoraExport.h"
-
-/**
-	This is a simple class that provides abstracted data feeding. You can use the
-	TheoraFileDataSource for regular file playback or you can implement your own
-	internet streaming solution, or a class that uses encrypted datafiles etc.
-	The sky is the limit
-*/
-class TheoraPlayerExport TheoraDataSource
-{
-public:
-
-    virtual ~TheoraDataSource();
-	/**
-		Reads nBytes bytes from data source and returns number of read bytes.
-		if function returns less bytes then nBytes, the system assumes EOF is reached.
-	*/
-	virtual int read(void* output,int nBytes)=0;
-    //! returns a string representation of the DataSource, eg 'File: source.ogg'
-	virtual std::string repr()=0;
-	//! position the source pointer to byte_index from the start of the source
-	virtual void seek(unsigned long byte_index)=0;
-	//! return the size of the stream in bytes
-	virtual unsigned long size()=0;
-	//! return the current position of the source pointer
-	virtual unsigned long tell()=0;
-};
-
-
-/**
-	provides standard file IO
-*/
-class TheoraPlayerExport TheoraFileDataSource : public TheoraDataSource
-{
-	FILE* mFilePtr;
-	std::string mFilename;
-	unsigned long mSize;
-	
-	void openFile();
-public:
-	TheoraFileDataSource(std::string filename);
-	~TheoraFileDataSource();
-
-	int read(void* output,int nBytes);
-	void seek(unsigned long byte_index);
-	std::string repr() { return mFilename; }
-	unsigned long size();
-	unsigned long tell();
-	
-	std::string getFilename() { return mFilename; }
-};
-
-/**
-	Pre-loads the entire file and streams from memory.
-	Very useful if you're continuously displaying a video and want to avoid disk reads.
-	Not very practical for large files.
-*/
-class TheoraPlayerExport TheoraMemoryFileDataSource : public TheoraDataSource
-{
-	std::string mFilename;
-	unsigned long mSize, mReadPointer;
-	unsigned char* mData;
-public:
-	TheoraMemoryFileDataSource(unsigned char* data, long size, const std::string& filename = "memory");
-	TheoraMemoryFileDataSource(std::string filename);
-	~TheoraMemoryFileDataSource();
-
-	int read(void* output,int nBytes);
-	void seek(unsigned long byte_index);
-	std::string repr() { return "MEM:"+mFilename; }
-	unsigned long size();
-	unsigned long tell();
-	std::string getFilename() { return mFilename; }
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraException.h b/drivers/theoraplayer/include/theoraplayer/TheoraException.h
deleted file mode 100644
index f79368fa1e..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraException.h
+++ /dev/null
@@ -1,46 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef EXCEPTION_H
-#define EXCEPTION_H
-
-#include <string>
-#include "TheoraExport.h"
-
-class TheoraPlayerExport _TheoraGenericException
-{
-public:
-    std::string mErrText,mFile,mType;
-	int mLineNumber;
-
-	_TheoraGenericException(const std::string& errorText, std::string type = "",std::string file = "", int line = 0);
-    virtual ~_TheoraGenericException() {}
-    
-	virtual std::string repr();
-    
-	void writeOutput();
-	
-	virtual const std::string& getErrorText() { return mErrText; }
-    
-	const std::string getType(){ return mType; }
-};
-
-#define TheoraGenericException(msg) _TheoraGenericException(msg, "TheoraGenericException", __FILE__, __LINE__)
-
-
-#define exception_cls(name) class name : public _TheoraGenericException \
-{ \
-public: \
-	name(const std::string& errorText,std::string type = "",std::string file = "",int line = 0) : \
-	  _TheoraGenericException(errorText, type, file, line){} \
-}
-
-exception_cls(_KeyException);
-
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraExport.h b/drivers/theoraplayer/include/theoraplayer/TheoraExport.h
deleted file mode 100644
index cf16d1004c..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraExport.h
+++ /dev/null
@@ -1,38 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _theoraVideoExport_h
-#define _theoraVideoExport_h
-
-	#ifdef _LIB
-		#define TheoraPlayerExport
-		#define TheoraPlayerFnExport
-	#else
-		#ifdef _WIN32
-			#ifdef THEORAVIDEO_EXPORTS
-				#define TheoraPlayerExport __declspec(dllexport)
-				#define TheoraPlayerFnExport __declspec(dllexport)
-			#else
-				#define TheoraPlayerExport __declspec(dllimport)
-				#define TheoraPlayerFnExport __declspec(dllimport)
-			#endif
-		#else
-			#define TheoraPlayerExport __attribute__ ((visibility("default")))
-			#define TheoraPlayerFnExport __attribute__ ((visibility("default")))
-		#endif
-	#endif
-	#ifndef DEPRECATED_ATTRIBUTE
-		#ifdef _MSC_VER
-			#define DEPRECATED_ATTRIBUTE __declspec(deprecated("function is deprecated"))
-		#else
-			#define DEPRECATED_ATTRIBUTE __attribute__((deprecated))
-		#endif
-	#endif
-
-#endif
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraFrameQueue.h b/drivers/theoraplayer/include/theoraplayer/TheoraFrameQueue.h
deleted file mode 100644
index fd985bb65a..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraFrameQueue.h
+++ /dev/null
@@ -1,95 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-
-#ifndef _TheoraFrameQueue_h
-#define _TheoraFrameQueue_h
-
-#include "TheoraAsync.h"
-#include <list>
-#include "TheoraExport.h"
-
-class TheoraVideoFrame;
-class TheoraVideoClip;
-
-/**
-	This class handles the frame queue. contains frames and handles their alloctation/deallocation
-	it is designed to be thread-safe
-*/
-class TheoraPlayerExport TheoraFrameQueue
-{
-protected:
-	std::list<TheoraVideoFrame*> mQueue;
-	TheoraVideoClip* mParent;
-	TheoraMutex mMutex;
-	
-	//! implementation function that returns a TheoraVideoFrame instance
-	TheoraVideoFrame* createFrameInstance(TheoraVideoClip* clip);
-public:
-	TheoraFrameQueue(TheoraVideoClip* parent);
-	~TheoraFrameQueue();
-
-	/**
-	    \brief Returns the first available frame in the queue or NULL if no frames are available.
-
-		This function DOES NOT remove the frame from the queue, you have to do it manually
-		when you want to mark the frame as used by calling the pop() function.
-	*/
-	TheoraVideoFrame* getFirstAvailableFrame();
-    //! non-mutex version
-	TheoraVideoFrame* _getFirstAvailableFrame();
-
-	//! return the number of used (not ready) frames
-	int getUsedCount();
-
-	//! return the number of ready frames
-	int getReadyCount();
-    //! non-mutex version
-	int _getReadyCount();
-
-	/**
-	    \brief remove the first N available frame from the queue.
-
-	    Use this every time you display a frame	so you can get the next one when the time comes.
-		This function marks the frame on the front of the queue as unused and it's memory then
-		get's used again in the decoding process.
-		If you don't call this, the frame queue will fill up with precached frames up to the
-		specified amount in the TheoraVideoManager class and you won't be able to advance the video.
-	*/
-	void pop(int n = 1);
-    
-    //! This is an internal _pop function. use externally only in combination with lock() / unlock() calls
-	void _pop(int n);
-
-	//! frees all decoded frames for reuse (does not destroy memory, just marks them as free)
-	void clear();
-	//! Called by WorkerThreads when they need to unload frame data, do not call directly!
-	TheoraVideoFrame* requestEmptyFrame();
-
-	/** 
-	    \brief set's the size of the frame queue.
-		
-		Beware, currently stored ready frames will be lost upon this call
-	*/
-	void setSize(int n);
-	//! return the size of the queue
-	int getSize();
-	
-	//! return whether all frames in the queue are ready for display
-	bool isFull();
-	
-	//! lock the queue's mutex manually
-	void lock();
-	//! unlock the queue's mutex manually
-	void unlock();
-    
-    //! returns the internal frame queue. Warning: Always lock / unlock queue's mutex before accessing frames directly!
-    std::list<TheoraVideoFrame*>& _getFrameQueue();
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraPixelTransform.h b/drivers/theoraplayer/include/theoraplayer/TheoraPixelTransform.h
deleted file mode 100644
index 73d853cd03..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraPixelTransform.h
+++ /dev/null
@@ -1,18 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraPixelTransform_h
-#define _TheoraPixelTransform_h
-
-struct TheoraPixelTransform
-{
-	unsigned char *raw, *y, *u, *v, *out;
-	unsigned int w, h, rawStride, yStride, uStride, vStride;
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraPlayer.h b/drivers/theoraplayer/include/theoraplayer/TheoraPlayer.h
deleted file mode 100644
index 8c5f2c735c..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraPlayer.h
+++ /dev/null
@@ -1,17 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraPlayer_h
-#define _TheoraPlayer_h
-
-#include "TheoraVideoManager.h"
-#include "TheoraVideoClip.h"
-#include "TheoraVideoFrame.h"
-
-#endif
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraTimer.h b/drivers/theoraplayer/include/theoraplayer/TheoraTimer.h
deleted file mode 100644
index 14fdbf47fc..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraTimer.h
+++ /dev/null
@@ -1,69 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-
-#ifndef _TheoraTimer_h
-#define _TheoraTimer_h
-
-#include "TheoraExport.h"
-
-/**
-    This is a Timer object, it is used to control the playback of a TheoraVideoClip.
-
-	You can inherit this class and make a timer that eg. plays twice as fast,
-	or playbacks an audio track and uses it's time offset for syncronizing Video etc.
- */
-class TheoraPlayerExport TheoraTimer
-{
-protected:
-	//! Current time in seconds
-	float mTime,mSpeed;
-	//! Is the timer paused or not
-	bool mPaused;
-public:
-	TheoraTimer();
-	virtual ~TheoraTimer();
-
-	virtual float getTime();
-	/**
-	    \brief advance the time.
-
-		If you're using another synronization system, eg. an audio track,
-		then you can ignore this call or use it to perform other updates.
-
-		NOTE: this is called by TheoraVideoManager from the main thread
-	 */
-	virtual void update(float timeDelta);
-
-	virtual void pause();
-	virtual void play();
-	virtual bool isPaused();
-	virtual void stop();
-	/**
-	    \brief set's playback speed
-
-        1.0 is the default. The speed factor multiplies time advance, thus
-        setting the value higher will increase playback speed etc.
-    
-        NOTE: depending on Timer implementation, it may not support setting the speed
-         
-	 */
-    virtual void setSpeed(float speed);
-    //! return the update speed 1.0 is the default
-    virtual float getSpeed();
-
-	/**
-	    \brief change the current time.
-
-		if you're using another syncronization mechanism, make sure to adjust
-		the time offset there
-	 */
-	virtual void seek(float time);
-};
-#endif
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraUtil.h b/drivers/theoraplayer/include/theoraplayer/TheoraUtil.h
deleted file mode 100644
index f168971ac7..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraUtil.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraUtil_h
-#define _TheoraUtil_h
-
-#include <string>
-#include <vector>
-
-#ifndef THEORAUTIL_NOMACROS
-
-#define foreach(type,lst) for (std::vector<type>::iterator it=lst.begin();it != lst.end(); ++it)
-#define foreach_l(type,lst) for (std::list<type>::iterator it=lst.begin();it != lst.end(); ++it)
-#define foreach_r(type,lst) for (std::vector<type>::reverse_iterator it=lst.rbegin();it != lst.rend(); ++it)
-#define foreach_in_map(type,lst) for (std::map<std::string,type>::iterator it=lst.begin();it != lst.end(); ++it)
-
-#endif
-
-#define th_writelog(x) TheoraVideoManager::getSingleton().logMessage(x)
-
-
-std::string str(int i);
-std::string strf(float i);
-void _psleep(int milliseconds);
-int _nextPow2(int x);
-
-#endif
\ No newline at end of file
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraVideoClip.h b/drivers/theoraplayer/include/theoraplayer/TheoraVideoClip.h
deleted file mode 100644
index fe71cf8566..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraVideoClip.h
+++ /dev/null
@@ -1,282 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-
-#ifndef _TheoraVideoClip_h
-#define _TheoraVideoClip_h
-
-#include <string>
-#include "TheoraExport.h"
-
-// forward class declarations
-class TheoraMutex;
-class TheoraFrameQueue;
-class TheoraTimer;
-class TheoraAudioInterface;
-class TheoraWorkerThread;
-class TheoraDataSource;
-class TheoraVideoFrame;
-
-/**
-    format of the TheoraVideoFrame pixels. Affects decoding time
- */
-enum TheoraOutputMode
-{
-	// A = full alpha (255), order of letters represents the byte order for a pixel
-	// A means the image is treated as if it contains an alpha channel, while X formats
-	// just mean that RGB frame is transformed to a 4 byte format
-	TH_UNDEFINED = 0,
-	TH_RGB    =  1,
-	TH_RGBA   =  2,
-	TH_RGBX   =  3,
-	TH_ARGB   =  4,
-	TH_XRGB   =  5,
-	TH_BGR    =  6,
-	TH_BGRA   =  7,
-	TH_BGRX   =  8,
-	TH_ABGR   =  9,
-	TH_XBGR   = 10,
-	TH_GREY   = 11,
-	TH_GREY3  = 12,
-	TH_GREY3A = 13,
-	TH_GREY3X = 14,
-	TH_AGREY3 = 15,
-	TH_XGREY3 = 16,
-	TH_YUV    = 17,
-	TH_YUVA   = 18,
-	TH_YUVX   = 19,
-	TH_AYUV   = 20,
-	TH_XYUV   = 21
-};
-
-/**
-	This object contains all data related to video playback, eg. the open source file,
-	the frame queue etc.
-*/
-class TheoraPlayerExport TheoraVideoClip
-{
-	friend class TheoraWorkerThread;
-	friend class TheoraVideoFrame;
-	friend class TheoraVideoManager;
-protected:
-	TheoraFrameQueue* mFrameQueue;
-	TheoraAudioInterface* mAudioInterface;
-	TheoraDataSource* mStream;
-
-	TheoraTimer *mTimer, *mDefaultTimer;
-
-	TheoraWorkerThread* mAssignedWorkerThread;
-	
-	bool mUseAlpha;
-	
-	bool mWaitingForCache;
-	
-	// benchmark vars
-	int mNumDroppedFrames, mNumDisplayedFrames, mNumPrecachedFrames;
-
-	int mThreadAccessCount; //! counter used by TheoraVideoManager to schedule workload
-	
-	int mSeekFrame; //! stores desired seek position as a frame number. next worker thread will do the seeking and reset this var to -1
-	float mDuration, mFrameDuration, mFPS;
-	float mPriority; //! User assigned priority. Default value is 1
-    std::string mName;
-	int mWidth, mHeight, mStride;
-	int mNumFrames;
-	int audio_track;
-
-	int mSubFrameWidth, mSubFrameHeight, mSubFrameOffsetX, mSubFrameOffsetY;
-	float mAudioGain; //! multiplier for audio samples. between 0 and 1
-
-	TheoraOutputMode mOutputMode, mRequestedOutputMode;
-	bool mFirstFrameDisplayed;
-	bool mAutoRestart;
-	bool mEndOfFile, mRestarted;
-	int mIteration, mPlaybackIteration; //! used to ensure smooth playback of looping videos
-
-	TheoraMutex* mAudioMutex; //! syncs audio decoding and extraction
-	TheoraMutex* mThreadAccessMutex;
-	
-	/**
-	 * Get the priority of a video clip. based on a forumula that includes user
-	 * priority factor, whether the video is paused or not, how many precached
-	 * frames it has etc.
-	 * This function is used in TheoraVideoManager to efficiently distribute job
-	 * assignments among worker threads
-	 * @return priority number of this video clip
-	 */
-	int calculatePriority();
-	void readTheoraVorbisHeaders();
-	virtual void doSeek() = 0; //! called by WorkerThread to seek to mSeekFrame
-	virtual bool _readData() = 0;
-	bool isBusy();
-
-	/**
-	 * decodes audio from the vorbis stream and stores it in audio packets
-	 * This is an internal function of TheoraVideoClip, called regularly if playing an
-	 * audio enabled video clip.
-	 * @return last decoded timestamp (if found in decoded packet's granule position)
-	 */
-	virtual float decodeAudio() = 0;
-    
-    int _getNumReadyFrames();
-    void resetFrameQueue();
-    int discardOutdatedFrames(float absTime);
-    float getAbsPlaybackTime();
-	virtual void load(TheoraDataSource* source) = 0;
-
-	virtual void _restart() = 0; // resets the decoder and stream but leaves the frame queue intact
-public:
-	TheoraVideoClip(TheoraDataSource* data_source,
-		            TheoraOutputMode output_mode,
-					int nPrecachedFrames,
-					bool usePower2Stride);
-	virtual ~TheoraVideoClip();
-
-	std::string getName();
-	//! Returns the string name of the decoder backend (eg. Theora, AVFoundation)
-	virtual std::string getDecoderName() = 0;
-
-	//! benchmark function
-	int getNumDisplayedFrames() { return mNumDisplayedFrames; }
-	//! benchmark function
-	int getNumDroppedFrames() { return mNumDroppedFrames; }
-
-	//! return width in pixels of the video clip
-	int getWidth();
-	//! return height in pixels of the video clip
-	int getHeight();
-    
-    //! Width of the actual picture inside a video frame (depending on implementation, this may be equal to mWidth or differ within a codec block size (usually 16))
-    int getSubFrameWidth();
-    //! Height of the actual picture inside a video frame (depending on implementation, this may be equal to mHeight or differ within a codec block size (usually 16))
-	int getSubFrameHeight();
-    //! X Offset of the actual picture inside a video frame (depending on implementation, this may be 0 or within a codec block size (usually 16))
-	int getSubFrameOffsetX();
-    //! Y Offset of the actual picture inside a video frame (depending on implementation, this may be 0 or differ within a codec block size (usually 16))
-	int getSubFrameOffsetY();
-    /**
-	    \brief return stride in pixels
-
-		If you've specified usePower2Stride when creating the TheoraVideoClip object
-		then this value will be the next power of two size compared to width,
-		eg: w=376, stride=512.
-
-		Otherwise, stride will be equal to width
-	 */
-	int getStride() { return mStride; }
-
-	//! retur the timer objet associated with this object
-	TheoraTimer* getTimer();
-	//! replace the timer object with a new one
-	void setTimer(TheoraTimer* timer);
-
-	//! used by TheoraWorkerThread, do not call directly
-	virtual bool decodeNextFrame() = 0;
-
-	//! advance time. TheoraVideoManager calls this
-	void update(float timeDelta);
-	/**
-	    \brief update timer to the display time of the next frame
-
-		useful if you want to grab frames instead of regular display
-		\return time advanced. 0 if no frames are ready
-	*/
-	float updateToNextFrame();
-
-	
-	TheoraFrameQueue* getFrameQueue();
-	
-	/**
-	    \brief pop the frame from the front of the FrameQueue
-
-		see TheoraFrameQueue::pop() for more details
-	 */
-	void popFrame();
-
-	/**
-	    \brief Returns the first available frame in the queue or NULL if no frames are available.
-
-		see TheoraFrameQueue::getFirstAvailableFrame() for more details
-	*/
-	TheoraVideoFrame* getNextFrame();
-	/**
-	    check if there is enough audio data decoded to submit to the audio interface
-
-		TheoraWorkerThread calls this
-	 */
-	virtual void decodedAudioCheck() = 0;
-
-	void setAudioInterface(TheoraAudioInterface* iface);
-	TheoraAudioInterface* getAudioInterface();
-
-	/**
-	    \brief resize the frame queues
-
-		Warning: this call discards ready frames in the frame queue
-	 */
-	void setNumPrecachedFrames(int n);
-	//! returns the size of the frame queue
-	int getNumPrecachedFrames();
-	//! returns the number of ready frames in the frame queue
-	int getNumReadyFrames();
-
-	//! if you want to adjust the audio gain. range [0,1]
-	void setAudioGain(float gain);
-	float getAudioGain();
-
-	//! if you want the video to automatically and smoothly restart when the last frame is reached
-	void setAutoRestart(bool value);
-	bool getAutoRestart() { return mAutoRestart; }
-
-
-	void set_audio_track(int p_track) { audio_track=p_track; }
-
-	/**
-	    TODO: user priority. Useful only when more than one video is being decoded
-	 */
-	void setPriority(float priority);
-	float getPriority();
-
-	//! Used by TheoraVideoManager to schedule work
-	float getPriorityIndex();
-
-	//! get the current time index from the timer object
-	float getTimePosition();
-	//! get the duration of the movie in seconds
-	float getDuration();
-	//! return the clips' frame rate, warning, fps can be a non integer number!
-	float getFPS();
-	//! get the number of frames in this movie
-	int getNumFrames() { return mNumFrames; }
-
-	//! return the current output mode for this video object
-	TheoraOutputMode getOutputMode();
-	/**
-	    set a new output mode
-
-		Warning: this discards the frame queue. ready frames will be lost.
-	 */
-	void setOutputMode(TheoraOutputMode mode);
-
-    bool isDone();
-	void play();
-	void pause();
-	void restart();
-	bool isPaused();
-	void stop();
-    void setPlaybackSpeed(float speed);
-    float getPlaybackSpeed();
-	//! seek to a given time position
-	void seek(float time);
-	//! seek to a given frame number
-	void seekToFrame(int frame);
-	//! wait max_time for the clip to cache a given percentage of frames, factor in range [0,1]
-	void waitForCache(float desired_cache_factor = 0.5f, float max_wait_time = 1.0f);
-};
-
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraVideoFrame.h b/drivers/theoraplayer/include/theoraplayer/TheoraVideoFrame.h
deleted file mode 100644
index 5d27f54d1c..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraVideoFrame.h
+++ /dev/null
@@ -1,56 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraVideoFrame_h
-#define _TheoraVideoFrame_h
-
-#include "TheoraExport.h"
-#include "TheoraVideoClip.h"
-
-struct TheoraPixelTransform;
-/**
-	
-*/
-class TheoraPlayerExport TheoraVideoFrame
-{
-protected:
-	TheoraVideoClip* mParent;
-	unsigned char* mBuffer;
-	unsigned long mFrameNumber;
-public:
-	//! global time in seconds this frame should be displayed on
-	float mTimeToDisplay;
-	//! whether the frame is ready for display or not
-	bool mReady;
-	//! indicates the frame is being used by TheoraWorkerThread instance
-	bool mInUse;
-	//! used to keep track of linear time in looping videos
-	int mIteration;
-	
-	int mBpp;
-
-	TheoraVideoFrame(TheoraVideoClip* parent);
-	virtual ~TheoraVideoFrame();
-
-	//! internal function, do not use directly
-	void _setFrameNumber(unsigned long number) { mFrameNumber = number; }
-	//! returns the frame number of this frame in the theora stream
-	unsigned long getFrameNumber() { return mFrameNumber; }
-
-	void clear();
-
-	int getWidth();
-	int getStride();
-	int getHeight();
-
-	unsigned char* getBuffer();
-
-	//! Called by TheoraVideoClip to decode a source buffer onto itself
-	virtual void decode(struct TheoraPixelTransform* t);
-};
-#endif
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraVideoManager.h b/drivers/theoraplayer/include/theoraplayer/TheoraVideoManager.h
deleted file mode 100644
index d94c51b4d4..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraVideoManager.h
+++ /dev/null
@@ -1,110 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-
-#ifndef _TheoraVideoManager_h
-#define _TheoraVideoManager_h
-
-#include <vector>
-#include <list>
-#include <string>
-#include "TheoraExport.h"
-#include "TheoraVideoClip.h"
-#ifdef _WIN32
-#pragma warning( disable: 4251 ) // MSVC++
-#endif
-// forward class declarations
-class TheoraWorkerThread;
-class TheoraMutex;
-class TheoraDataSource;
-class TheoraAudioInterfaceFactory;
-/**
-	This is the main singleton class that handles all playback/sync operations
-*/
-class TheoraPlayerExport TheoraVideoManager
-{
-protected:
-	friend class TheoraWorkerThread;
-	typedef std::vector<TheoraVideoClip*> ClipList;
-	typedef std::vector<TheoraWorkerThread*> ThreadList;
-
-	//! stores pointers to worker threads which are decoding video and audio
-	ThreadList mWorkerThreads;
-	//! stores pointers to created video clips
-	ClipList mClips;
-	
-	//! stores pointer to clips that were docoded in the past in order to achieve fair scheduling
-	std::list<TheoraVideoClip*> mWorkLog;
-
-	int mDefaultNumPrecachedFrames;
-
-	TheoraMutex* mWorkMutex;
-	TheoraAudioInterfaceFactory* mAudioFactory;
-
-	void createWorkerThreads(int n);
-	void destroyWorkerThreads();
-	
-	float calcClipWorkTime(TheoraVideoClip* clip);
-
-	/**
-	 * Called by TheoraWorkerThread to request a TheoraVideoClip instance to work on decoding
-	 */
-	TheoraVideoClip* requestWork(TheoraWorkerThread* caller);
-public:
-	TheoraVideoManager(int num_worker_threads=1);
-	virtual ~TheoraVideoManager();
-
-	//! get the global reference to the manager instance
-	static TheoraVideoManager& getSingleton();
-	//! get the global pointer to the manager instance
-	static TheoraVideoManager* getSingletonPtr();
-
-	//! search registered clips by name
-	TheoraVideoClip* getVideoClipByName(std::string name);
-
-	TheoraVideoClip* createVideoClip(std::string filename,TheoraOutputMode output_mode=TH_RGB,int numPrecachedOverride=0,bool usePower2Stride=0, int p_track=0);
-	TheoraVideoClip* createVideoClip(TheoraDataSource* data_source,TheoraOutputMode output_mode=TH_RGB,int numPrecachedOverride=0,bool usePower2Stride=0, int p_audio_track=0);
-
-	void update(float timeDelta);
-
-	void destroyVideoClip(TheoraVideoClip* clip);
-
-	void setAudioInterfaceFactory(TheoraAudioInterfaceFactory* factory);
-	TheoraAudioInterfaceFactory* getAudioInterfaceFactory();
-
-	int getNumWorkerThreads();
-	void setNumWorkerThreads(int n);
-
-	void setDefaultNumPrecachedFrames(int n) { mDefaultNumPrecachedFrames=n; }
-	int getDefaultNumPrecachedFrames() { return mDefaultNumPrecachedFrames; }
-
-	//! used by libtheoraplayer functions
-	void logMessage(std::string msg);
-
-	/**
-		\brief you can set your own log function to recieve theora's log calls
-
-		This way you can integrate libtheoraplayer's log messages in your own
-		logging system, prefix them, mute them or whatever you want
-	 */
-	static void setLogFunction(void (*fn)(std::string));
-
-	//! get nicely formated version string
-	std::string getVersionString();
-	/**
-	    \brief get version numbers
-
-		if c is negative, it means it's a release candidate -c
-	 */
-	void getVersion(int* a,int* b,int* c);
-
-	//! returns the supported decoders (eg. Theora, AVFoundation...)
-	std::vector<std::string> getSupportedDecoders();
-};
-#endif
-
diff --git a/drivers/theoraplayer/include/theoraplayer/TheoraWorkerThread.h b/drivers/theoraplayer/include/theoraplayer/TheoraWorkerThread.h
deleted file mode 100644
index 2299acedbd..0000000000
--- a/drivers/theoraplayer/include/theoraplayer/TheoraWorkerThread.h
+++ /dev/null
@@ -1,32 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _TheoraWorkerThread_h
-#define _TheoraWorkerThread_h
-
-#include "TheoraAsync.h"
-
-class TheoraVideoClip;
-
-/**
-	This is the worker thread, requests work from TheoraVideoManager
-	and decodes assigned TheoraVideoClip objects
-*/
-class TheoraWorkerThread : public TheoraThread
-{
-	TheoraVideoClip* mClip;
-public:
-	TheoraWorkerThread();
-	~TheoraWorkerThread();
-
-	TheoraVideoClip* getAssignedClip() { return mClip; }
-
-    //! Main Thread Body - do not call directly!
-	void execute();
-};
-#endif
diff --git a/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.h b/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.h
deleted file mode 100644
index abd898aa01..0000000000
--- a/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.h
+++ /dev/null
@@ -1,47 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#if defined(__AVFOUNDATION) && !defined(_TheoraVideoClip_AVFoundation_h)
-#define _TheoraVideoClip_AVFoundation_h
-
-#include "TheoraAudioPacketQueue.h"
-#include "TheoraVideoClip.h"
-
-#ifndef AVFOUNDATION_CLASSES_DEFINED
-class AVAssetReader;
-class AVAssetReaderTrackOutput;
-#endif
-
-class TheoraVideoClip_AVFoundation : public TheoraVideoClip, public TheoraAudioPacketQueue
-{
-protected:
-	bool mLoaded;
-	int mFrameNumber;
-	AVAssetReader* mReader;
-	AVAssetReaderTrackOutput *mOutput, *mAudioOutput;
-	unsigned int mReadAudioSamples;
-	
-	void unload();
-	void doSeek();
-public:
-	TheoraVideoClip_AVFoundation(TheoraDataSource* data_source,
-								 TheoraOutputMode output_mode,
-								 int nPrecachedFrames,
-								 bool usePower2Stride);
-	~TheoraVideoClip_AVFoundation();
-	
-	bool _readData();
-	bool decodeNextFrame();
-	void _restart();
-	void load(TheoraDataSource* source);
-	float decodeAudio();
-	void decodedAudioCheck();
-	std::string getDecoderName() { return "AVFoundation"; }
-};
-
-#endif
diff --git a/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.mm b/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.mm
deleted file mode 100644
index 1b5cf0ab13..0000000000
--- a/drivers/theoraplayer/src/AVFoundation/TheoraVideoClip_AVFoundation.mm
+++ /dev/null
@@ -1,457 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef __AVFOUNDATION
-#define AVFOUNDATION_CLASSES_DEFINED
-#import <AVFoundation/AVFoundation.h>
-#include "TheoraAudioInterface.h"
-#include "TheoraDataSource.h"
-#include "TheoraException.h"
-#include "TheoraTimer.h"
-#include "TheoraUtil.h"
-#include "TheoraFrameQueue.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraVideoManager.h"
-#include "TheoraVideoClip_AVFoundation.h"
-#include "TheoraPixelTransform.h"
-
-#ifdef _AVFOUNDATION_BGRX
-// a fast function developed to use kernel byte swapping calls to optimize alpha decoding.
-// In AVFoundation, BGRX mode conversion is prefered to YUV conversion because apple's YUV
-// conversion on iOS seems to run faster than libtheoraplayer's implementation
-// This may change in the future with more optimizations to libtheoraplayers's YUV conversion
-// code, making this function obsolete.
-static void bgrx2rgba(unsigned char* dest, int w, int h, struct TheoraPixelTransform* t)
-{
-	unsigned register int a;
-	unsigned int *dst = (unsigned int*) dest, *dstEnd;
-	unsigned char* src = t->raw;
-	int y, x, ax;
-	
-	for (y = 0; y < h; ++y, src += t->rawStride)
-	{
-		for (x = 0, ax = w * 4, dstEnd = dst + w; dst != dstEnd; x += 4, ax += 4, ++dst)
-		{
-            // use the full alpha range here because the Y channel has already been converted
-            // to RGB and that's in [0, 255] range.
-			a = src[ax];
-            *dst = (OSReadSwapInt32(src, x) >> 8) | (a << 24);
-		}
-	}
-}
-#endif
-
-static CVPlanarPixelBufferInfo_YCbCrPlanar getYUVStruct(void* src)
-{
-	CVPlanarPixelBufferInfo_YCbCrPlanar* bigEndianYuv = (CVPlanarPixelBufferInfo_YCbCrPlanar*) src;
-	CVPlanarPixelBufferInfo_YCbCrPlanar yuv;
-	yuv.componentInfoY.offset = OSSwapInt32(bigEndianYuv->componentInfoY.offset);
-	yuv.componentInfoY.rowBytes = OSSwapInt32(bigEndianYuv->componentInfoY.rowBytes);
-	yuv.componentInfoCb.offset = OSSwapInt32(bigEndianYuv->componentInfoCb.offset);
-	yuv.componentInfoCb.rowBytes = OSSwapInt32(bigEndianYuv->componentInfoCb.rowBytes);
-	yuv.componentInfoCr.offset = OSSwapInt32(bigEndianYuv->componentInfoCr.offset);
-	yuv.componentInfoCr.rowBytes = OSSwapInt32(bigEndianYuv->componentInfoCr.rowBytes);
-	return yuv;
-}
-
-TheoraVideoClip_AVFoundation::TheoraVideoClip_AVFoundation(TheoraDataSource* data_source,
-											   TheoraOutputMode output_mode,
-											   int nPrecachedFrames,
-											   bool usePower2Stride):
-	TheoraVideoClip(data_source, output_mode, nPrecachedFrames, usePower2Stride),
-	TheoraAudioPacketQueue()
-{
-	mLoaded = 0;
-	mReader = NULL;
-	mOutput = mAudioOutput = NULL;
-	mReadAudioSamples = mAudioFrequency = mNumAudioChannels = 0;
-}
-
-TheoraVideoClip_AVFoundation::~TheoraVideoClip_AVFoundation()
-{
-	unload();
-}
-
-void TheoraVideoClip_AVFoundation::unload()
-{
-	if (mOutput != NULL || mAudioOutput != NULL || mReader != NULL)
-	{
-		NSAutoreleasePool* pool = [[NSAutoreleasePool alloc] init];
-
-		if (mOutput != NULL)
-		{
-			[mOutput release];
-			mOutput = NULL;
-		}
-		
-		if (mAudioOutput)
-		{
-			[mAudioOutput release];
-			mAudioOutput = NULL;
-		}
-		
-		if (mReader != NULL)
-		{
-			[mReader release];
-			mReader = NULL;
-		}
-		
-		[pool release];
-	}
-}
-
-bool TheoraVideoClip_AVFoundation::_readData()
-{
-	return 1;
-}
-
-bool TheoraVideoClip_AVFoundation::decodeNextFrame()
-{
-	if (mReader == NULL || mEndOfFile) return 0;
-	AVAssetReaderStatus status = [mReader status];
-	if (status == AVAssetReaderStatusFailed)
-	{
-		// This can happen on iOS when you suspend the app... Only happens on the device, iOS simulator seems to work fine.
-		th_writelog("AVAssetReader reading failed, restarting...");
-
-		mSeekFrame = mTimer->getTime() * mFPS;
-		// just in case
-		if (mSeekFrame < 0) mSeekFrame = 0;
-		if (mSeekFrame > mDuration * mFPS - 1) mSeekFrame = mDuration * mFPS - 1;
-		_restart();
-		status = [mReader status];
-		if (status == AVAssetReaderStatusFailed)
-		{
-			th_writelog("AVAssetReader restart failed!");
-			return 0;
-		}
-		th_writelog("AVAssetReader restart succeeded!");
-	}
-
-	TheoraVideoFrame* frame = mFrameQueue->requestEmptyFrame();
-	if (!frame) return 0;
-
-	CMSampleBufferRef sampleBuffer = NULL;
-	NSAutoreleasePool* pool = NULL;
-	CMTime presentationTime;
-	
-	if (mAudioInterface) decodeAudio();
-	
-	if (status == AVAssetReaderStatusReading)
-	{
-		pool = [[NSAutoreleasePool alloc] init];
-		
-		while ((sampleBuffer = [mOutput copyNextSampleBuffer]))
-		{
-			presentationTime = CMSampleBufferGetOutputPresentationTimeStamp(sampleBuffer);
-			frame->mTimeToDisplay = (float) CMTimeGetSeconds(presentationTime);
-			frame->mIteration = mIteration;
-			frame->_setFrameNumber(mFrameNumber);
-			++mFrameNumber;
-			if (frame->mTimeToDisplay < mTimer->getTime() && !mRestarted && mFrameNumber % 16 != 0)
-			{
-				// %16 operation is here to prevent a playback halt during video playback if the decoder can't keep up with demand.
-#ifdef _DEBUG
-				th_writelog(mName + ": pre-dropped frame " + str(mFrameNumber - 1));
-#endif
-				++mNumDisplayedFrames;
-				++mNumDroppedFrames;
-				CMSampleBufferInvalidate(sampleBuffer);
-				CFRelease(sampleBuffer);
-				sampleBuffer = NULL;
-				continue; // drop frame
-			}
-
-			CVImageBufferRef imageBuffer = CMSampleBufferGetImageBuffer(sampleBuffer);
-			CVPixelBufferLockBaseAddress(imageBuffer, 0);
-			void *baseAddress = CVPixelBufferGetBaseAddress(imageBuffer);
-			
-			mStride = CVPixelBufferGetBytesPerRow(imageBuffer);
-			size_t width = CVPixelBufferGetWidth(imageBuffer);
-			size_t height = CVPixelBufferGetHeight(imageBuffer);
-
-			TheoraPixelTransform t;
-			memset(&t, 0, sizeof(TheoraPixelTransform));
-#ifdef _AVFOUNDATION_BGRX
-			if (mOutputMode == TH_BGRX || mOutputMode == TH_RGBA)
-			{
-				t.raw = (unsigned char*) baseAddress;
-				t.rawStride = mStride;
-			}
-			else
-#endif
-			{
-				CVPlanarPixelBufferInfo_YCbCrPlanar yuv = getYUVStruct(baseAddress);
-				
-				t.y = (unsigned char*) baseAddress + yuv.componentInfoY.offset;  t.yStride = yuv.componentInfoY.rowBytes;
-				t.u = (unsigned char*) baseAddress + yuv.componentInfoCb.offset; t.uStride = yuv.componentInfoCb.rowBytes;
-				t.v = (unsigned char*) baseAddress + yuv.componentInfoCr.offset; t.vStride = yuv.componentInfoCr.rowBytes;
-			}
-#ifdef _AVFOUNDATION_BGRX
-			if (mOutputMode == TH_RGBA)
-			{
-				for (int i = 0; i < 1000; ++i)
-					bgrx2rgba(frame->getBuffer(), mWidth / 2, mHeight, &t);
-				frame->mReady = true;
-			}
-			else
-#endif
-			frame->decode(&t);
-
-			CVPixelBufferUnlockBaseAddress(imageBuffer, 0);
-			CMSampleBufferInvalidate(sampleBuffer);
-			CFRelease(sampleBuffer);
-
-			break; // TODO - should this really be a while loop instead of an if block?
-		}
-	}
-	if (pool) [pool release];
-
-	if (!frame->mReady) // in case the frame wasn't used
-	{
-		frame->mInUse = 0;
-	}
-
-	if (sampleBuffer == NULL && mReader.status == AVAssetReaderStatusCompleted) // other cases could be app suspended
-	{
-		if (mAutoRestart)
-        {
-            ++mIteration;
-			_restart();
-        }
-		else
-		{
-			unload();
-			mEndOfFile = true;
-		}
-		return 0;
-	}
-	
-	
-	return 1;
-}
-
-void TheoraVideoClip_AVFoundation::_restart()
-{
-	mEndOfFile = false;
-	unload();
-	load(mStream);
-	mRestarted = true;
-}
-
-void TheoraVideoClip_AVFoundation::load(TheoraDataSource* source)
-{
-	mStream = source;
-	mFrameNumber = 0;
-	mEndOfFile = false;
-	TheoraFileDataSource* fileDataSource = dynamic_cast<TheoraFileDataSource*>(source);
-	std::string filename;
-	if (fileDataSource != NULL) filename = fileDataSource->getFilename();
-	else
-	{
-		TheoraMemoryFileDataSource* memoryDataSource = dynamic_cast<TheoraMemoryFileDataSource*>(source);
-		if (memoryDataSource != NULL) filename = memoryDataSource->getFilename();
-		else throw TheoraGenericException("Unable to load MP4 file");
-	}
-	
-	NSAutoreleasePool* pool = [[NSAutoreleasePool alloc] init];
-	NSString* path = [NSString stringWithUTF8String:filename.c_str()];
-	NSError* err;
-	NSURL *url = [NSURL fileURLWithPath:path];
-	AVAsset* asset = [[AVURLAsset alloc] initWithURL:url options:nil];
-	mReader = [[AVAssetReader alloc] initWithAsset:asset error:&err];
-	NSArray* tracks = [asset tracksWithMediaType:AVMediaTypeVideo];
-	if ([tracks count] == 0)
-		throw TheoraGenericException("Unable to open video file: " + filename);
-	AVAssetTrack *videoTrack = [tracks objectAtIndex:0];
-
-	NSArray* audioTracks = [asset tracksWithMediaType:AVMediaTypeAudio];
-	if (audio_track >= audioTracks.count)
-		audio_track = 0;
-	AVAssetTrack *audioTrack = audioTracks.count > 0 ? [audioTracks objectAtIndex:audio_track] : NULL;
-	printf("*********** using audio track %i\n", audio_track);
-	
-#ifdef _AVFOUNDATION_BGRX
-	bool yuv_output = (mOutputMode != TH_BGRX && mOutputMode != TH_RGBA);
-#else
-	bool yuv_output = true;
-#endif
-	
-	NSDictionary *videoOptions = [NSDictionary dictionaryWithObjectsAndKeys:[NSNumber numberWithInt:(yuv_output) ? kCVPixelFormatType_420YpCbCr8Planar : kCVPixelFormatType_32BGRA], kCVPixelBufferPixelFormatTypeKey, nil];
-
-	mOutput = [[AVAssetReaderTrackOutput alloc] initWithTrack:videoTrack outputSettings:videoOptions];
-	[mReader addOutput:mOutput];
-	if ([mOutput respondsToSelector:@selector(setAlwaysCopiesSampleData:)]) // Not supported on iOS versions older than 5.0
-		mOutput.alwaysCopiesSampleData = NO;
-
-	mFPS = videoTrack.nominalFrameRate;
-	mWidth = mSubFrameWidth = mStride = videoTrack.naturalSize.width;
-	mHeight = mSubFrameHeight = videoTrack.naturalSize.height;
-	mFrameDuration = 1.0f / mFPS;
-	mDuration = (float) CMTimeGetSeconds(asset.duration);
-	if (mFrameQueue == NULL)
-	{
-		mFrameQueue = new TheoraFrameQueue(this);
-		mFrameQueue->setSize(mNumPrecachedFrames);
-	}
-
-	if (mSeekFrame != -1)
-	{
-		mFrameNumber = mSeekFrame;
-		[mReader setTimeRange: CMTimeRangeMake(CMTimeMakeWithSeconds(mSeekFrame / mFPS, 1), kCMTimePositiveInfinity)];
-	}
-	if (audioTrack)
-	{
-		TheoraAudioInterfaceFactory* audio_factory = TheoraVideoManager::getSingleton().getAudioInterfaceFactory();
-		if (audio_factory)
-		{
-			NSDictionary *audioOptions = [NSDictionary dictionaryWithObjectsAndKeys:
-										  [NSNumber numberWithInt:kAudioFormatLinearPCM], AVFormatIDKey,
-										  [NSNumber numberWithBool:NO], AVLinearPCMIsNonInterleaved,
-										  [NSNumber numberWithBool:NO], AVLinearPCMIsBigEndianKey,
-										  [NSNumber numberWithBool:YES], AVLinearPCMIsFloatKey,
-										  [NSNumber numberWithInt:32], AVLinearPCMBitDepthKey,
-										  nil];
-
-			mAudioOutput = [[AVAssetReaderTrackOutput alloc] initWithTrack:audioTrack outputSettings:audioOptions];
-			[mReader addOutput:mAudioOutput];
-			if ([mAudioOutput respondsToSelector:@selector(setAlwaysCopiesSampleData:)]) // Not supported on iOS versions older than 5.0
-				mAudioOutput.alwaysCopiesSampleData = NO;
-			
-			NSArray* desclst = audioTrack.formatDescriptions;
-			CMAudioFormatDescriptionRef desc = (CMAudioFormatDescriptionRef) [desclst objectAtIndex:0];
-			const AudioStreamBasicDescription* audioDesc = CMAudioFormatDescriptionGetStreamBasicDescription(desc);
-			mAudioFrequency = (unsigned int) audioDesc->mSampleRate;
-			mNumAudioChannels = audioDesc->mChannelsPerFrame;
-			
-			if (mSeekFrame != -1)
-			{
-				mReadAudioSamples = mFrameNumber * (mAudioFrequency * mNumAudioChannels) / mFPS;
-			}
-			else mReadAudioSamples = 0;
-
-			if (mAudioInterface == NULL)
-				setAudioInterface(audio_factory->createInstance(this, mNumAudioChannels, mAudioFrequency));
-		}
-	}
-	
-#ifdef _DEBUG
-	else if (!mLoaded)
-	{
-		th_writelog("-----\nwidth: " + str(mWidth) + ", height: " + str(mHeight) + ", fps: " + str((int) getFPS()));
-		th_writelog("duration: " + strf(mDuration) + " seconds\n-----");
-	}
-#endif
-	[mReader startReading];
-	[pool release];
-	mLoaded = true;
-}
- 
-void TheoraVideoClip_AVFoundation::decodedAudioCheck()
-{
-	if (!mAudioInterface || mTimer->isPaused()) return;
-	
-	mAudioMutex->lock();
-	flushAudioPackets(mAudioInterface);
-	mAudioMutex->unlock();
-}
-
-float TheoraVideoClip_AVFoundation::decodeAudio()
-{
-	if (mRestarted) return -1;
-
-	if (mReader == NULL || mEndOfFile) return 0;
-	AVAssetReaderStatus status = [mReader status];
-
-	if (mAudioOutput)
-	{
-		CMSampleBufferRef sampleBuffer = NULL;
-		NSAutoreleasePool* pool = NULL;
-		bool mutexLocked = 0;
-
-		float factor = 1.0f / (mAudioFrequency * mNumAudioChannels);
-		float videoTime = (float) mFrameNumber / mFPS;
-		float min = mFrameQueue->getSize() / mFPS + 1.0f;
-		
-		if (status == AVAssetReaderStatusReading)
-		{
-			pool = [[NSAutoreleasePool alloc] init];
-
-			// always buffer up of audio ahead of the frames
-			while (mReadAudioSamples * factor - videoTime < min)
-			{
-				if ((sampleBuffer = [mAudioOutput copyNextSampleBuffer]))
-				{
-					AudioBufferList audioBufferList;
-					
-					CMBlockBufferRef blockBuffer = NULL;
-					CMSampleBufferGetAudioBufferListWithRetainedBlockBuffer(sampleBuffer, NULL, &audioBufferList, sizeof(audioBufferList), NULL, NULL, 0, &blockBuffer);
-					
-					for (int y = 0; y < audioBufferList.mNumberBuffers; ++y)
-					{
-						AudioBuffer audioBuffer = audioBufferList.mBuffers[y];
-						float *frame = (float*) audioBuffer.mData;
-
-						if (!mutexLocked)
-						{
-							mAudioMutex->lock();
-							mutexLocked = 1;
-						}
-						addAudioPacket(frame, audioBuffer.mDataByteSize / (mNumAudioChannels * sizeof(float)), mAudioGain);
-						
-						mReadAudioSamples += audioBuffer.mDataByteSize / (sizeof(float));
-					}
-
-					CFRelease(blockBuffer);
-					CMSampleBufferInvalidate(sampleBuffer);
-					CFRelease(sampleBuffer);
-				}
-				else
-				{
-					[mAudioOutput release];
-					mAudioOutput = nil;
-					break;
-				}
-			}
-			[pool release];
-		}
-		if (mutexLocked) mAudioMutex->unlock();
-	}
-	
-	return -1;
-}
-
-void TheoraVideoClip_AVFoundation::doSeek()
-{
-#if _DEBUG
-	th_writelog(mName + " [seek]: seeking to frame " + str(mSeekFrame));
-#endif
-	int frame;
-	float time = mSeekFrame / getFPS();
-	mTimer->seek(time);
-	bool paused = mTimer->isPaused();
-	if (!paused) mTimer->pause(); // pause until seeking is done
-	
-	mEndOfFile = false;
-	mRestarted = false;
-	
-    resetFrameQueue();
-	unload();
-	load(mStream);
-
-	if (mAudioInterface)
-	{
-		mAudioMutex->lock();
-		destroyAllAudioPackets();
-		mAudioMutex->unlock();
-	}
-
-	if (!paused) mTimer->play();
-	mSeekFrame = -1;
-}
-#endif
diff --git a/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.cpp b/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.cpp
deleted file mode 100644
index fa3fd43a47..0000000000
--- a/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.cpp
+++ /dev/null
@@ -1,439 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef __FFMPEG
-#include "TheoraAudioInterface.h"
-#include "TheoraDataSource.h"
-#include "TheoraException.h"
-#include "TheoraTimer.h"
-#include "TheoraUtil.h"
-#include "TheoraFrameQueue.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraVideoManager.h"
-#include "TheoraVideoClip_FFmpeg.h"
-#include "TheoraPixelTransform.h"
-
-#define READ_BUFFER_SIZE 4096
-
-#ifdef __cplusplus
-#define __STDC_CONSTANT_MACROS
-#ifdef _STDINT_H
-#undef _STDINT_H
-#endif
-# include <stdint.h>
-#endif
-
-#define _FFMPEG_DEBUG
-
-extern "C"
-{
-#include <libavcodec/avcodec.h>
-#include <libavformat/avformat.h>
-#include "libavutil/avassert.h"
-}
-
-static bool ffmpegInitialised = 0;
-
-static int readFunction(void* data, uint8_t* buf, int buf_size)
-{
-#ifdef _FFMPEG_DEBUG
-	th_writelog("reading " + str(buf_size) + " bytes");
-#endif
-
-	TheoraDataSource* src = (TheoraDataSource*) data;
-	return src->read(buf, buf_size);
-}
-
-static int64_t seekFunction(void* data, int64_t offset, int whence)
-{
-#ifdef _FFMPEG_DEBUG
-	th_writelog("seeking: offset = " + str((long) offset) + ", whence = " + str(whence));
-#endif
-
-	TheoraDataSource* src = (TheoraDataSource*) data;
-    if (whence == AVSEEK_SIZE)
-		return src->size();
-	else if (whence == SEEK_SET)
-		src->seek((long) offset);
-	else if (whence == SEEK_END)
-		src->seek(src->size() - (long) offset);
-    return src->tell();
-}
-
-static void avlog_theoraplayer(void* p, int level, const char* fmt, va_list vargs)
-{
-	th_writelog(fmt);
-	static char logstr[2048];
-	vsprintf(logstr, fmt, vargs);
-	th_writelog("ffmpeg: " + std::string(logstr));
-}
-
-
-std::string text;
-
-static void _log(const char* s)
-{
-	text += s;
-//	th_writelog(text);
-//	text = "";
-}
-
-static void _log(const char c)
-{
-	char s[2] = {c, 0};
-	_log(s);
-}
-
-static const AVCodec *next_codec_for_id(enum AVCodecID id, const AVCodec *prev,
-                                        int encoder)
-{
-    while ((prev = av_codec_next(prev))) {
-        if (prev->id == id &&
-            (encoder ? av_codec_is_encoder(prev) : av_codec_is_decoder(prev)))
-            return prev;
-    }
-    return NULL;
-}
-
-static int compare_codec_desc(const void *a, const void *b)
-{
-    const AVCodecDescriptor **da = (const AVCodecDescriptor **) a;
-    const AVCodecDescriptor **db = (const AVCodecDescriptor **) b;
-	
-    return (*da)->type != (*db)->type ? (*da)->type - (*db)->type :
-	strcmp((*da)->name, (*db)->name);
-}
-
-static unsigned get_codecs_sorted(const AVCodecDescriptor ***rcodecs)
-{
-    const AVCodecDescriptor *desc = NULL;
-    const AVCodecDescriptor **codecs;
-    unsigned nb_codecs = 0, i = 0;
-	
-    while ((desc = avcodec_descriptor_next(desc)))
-        ++nb_codecs;
-    if (!(codecs = (const AVCodecDescriptor**) av_calloc(nb_codecs, sizeof(*codecs)))) {
-        av_log(NULL, AV_LOG_ERROR, "Out of memory\n");
-        exit(1);
-    }
-    desc = NULL;
-    while ((desc = avcodec_descriptor_next(desc)))
-        codecs[i++] = desc;
-    av_assert0(i == nb_codecs);
-    qsort(codecs, nb_codecs, sizeof(*codecs), compare_codec_desc);
-    *rcodecs = codecs;
-    return nb_codecs;
-}
-
-static char get_media_type_char(enum AVMediaType type)
-{
-    switch (type) {
-        case AVMEDIA_TYPE_VIDEO:    return 'V';
-        case AVMEDIA_TYPE_AUDIO:    return 'A';
-        case AVMEDIA_TYPE_DATA:     return 'D';
-        case AVMEDIA_TYPE_SUBTITLE: return 'S';
-        case AVMEDIA_TYPE_ATTACHMENT:return 'T';
-        default:                    return '?';
-    }
-}
-
-static void print_codecs_for_id(enum AVCodecID id, int encoder)
-{
-    const AVCodec *codec = NULL;
-	
-    _log(encoder ? "encoders" : "decoders");
-	
-    while ((codec = next_codec_for_id(id, codec, encoder)))
-        _log(codec->name);
-	
-    _log(")");
-}
-
-int show_codecs(void *optctx, const char *opt, const char *arg)
-{
-    const AVCodecDescriptor **codecs;
-    unsigned i, nb_codecs = get_codecs_sorted(&codecs);
-	
-	char tmp[1024];
-    th_writelog("Codecs:\n"
-           " D..... = Decoding supported\n"
-           " .E.... = Encoding supported\n"
-           " ..V... = Video codec\n"
-           " ..A... = Audio codec\n"
-           " ..S... = Subtitle codec\n"
-           " ...I.. = Intra frame-only codec\n"
-           " ....L. = Lossy compression\n"
-           " .....S = Lossless compression\n"
-           " -------\n");
-    for (i = 0; i < nb_codecs; ++i) {
-        const AVCodecDescriptor *desc = codecs[i];
-        const AVCodec *codec = NULL;
-		
-        _log(" ");
-        _log(avcodec_find_decoder(desc->id) ? "D" : ".");
-        _log(avcodec_find_encoder(desc->id) ? "E" : ".");
-		
-        _log(get_media_type_char(desc->type));
-        _log((desc->props & AV_CODEC_PROP_INTRA_ONLY) ? "I" : ".");
-        _log((desc->props & AV_CODEC_PROP_LOSSY)      ? "L" : ".");
-        _log((desc->props & AV_CODEC_PROP_LOSSLESS)   ? "S" : ".");
-		
-		
-        sprintf(tmp, " %-20s %s", desc->name, desc->long_name ? desc->long_name : "");
-		
-		_log(tmp);
-        /* print decoders/encoders when there's more than one or their
-         * names are different from codec name */
-        while ((codec = next_codec_for_id(desc->id, codec, 0))) {
-            if (strcmp(codec->name, desc->name)) {
-                print_codecs_for_id(desc->id, 0);
-                break;
-            }
-        }
-        codec = NULL;
-        while ((codec = next_codec_for_id(desc->id, codec, 1))) {
-            if (strcmp(codec->name, desc->name)) {
-                print_codecs_for_id(desc->id, 1);
-                break;
-            }
-        }
-		_log("\n");
-    }
-    av_free(codecs);
-	
-	av_log(0, 0, "%s", text.c_str());
-    return 0;
-}
-
-TheoraVideoClip_FFmpeg::TheoraVideoClip_FFmpeg(TheoraDataSource* data_source,
-														 TheoraOutputMode output_mode,
-														 int nPrecachedFrames,
-														 bool usePower2Stride):
-								 						 TheoraVideoClip(data_source, output_mode, nPrecachedFrames, usePower2Stride),
-														 TheoraAudioPacketQueue()
-{
-	mFormatContext = NULL;
-	mCodecContext = NULL;
-	mCodec = NULL;
-	mFrame = NULL;
-	mVideoStreamIndex = -1;
-}
-
-TheoraVideoClip_FFmpeg::~TheoraVideoClip_FFmpeg()
-{
-	unload();
-}
-
-void TheoraVideoClip_FFmpeg::load(TheoraDataSource* source)
-{
-	mVideoStreamIndex = -1;
-	mFrameNumber = 0;
-	AVDictionary* optionsDict = NULL;
-	
-	if (!ffmpegInitialised)
-	{
-#ifdef _FFMPEG_DEBUG
-		th_writelog("Initializing ffmpeg");
-#endif
-		th_writelog("avcodec version: " + str(avcodec_version()));
-		av_register_all();
-		av_log_set_level(AV_LOG_DEBUG);
-		av_log_set_callback(avlog_theoraplayer);
-		ffmpegInitialised = 1;
-		//show_codecs(0, 0, 0);
-	}
-	
-	mInputBuffer = (unsigned char*) av_malloc(READ_BUFFER_SIZE);
-	mAvioContext = avio_alloc_context(mInputBuffer, READ_BUFFER_SIZE, 0, source, &readFunction, NULL, &seekFunction);
-	
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": avio context created");
-#endif
-
-	mFormatContext = avformat_alloc_context();
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": avformat context created");
-#endif
-	mFormatContext->pb = mAvioContext;
-	
-	int err;
-	if ((err = avformat_open_input(&mFormatContext, "", NULL, NULL)) != 0)
-	{
-		th_writelog(mName + ": avformat input opening failed!");
-		th_writelog(mName + ": error_code: " + str(err));
-		return;
-	}
-	
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": avformat input opened");
-#endif
-	
-	// Retrieve stream information
-	if (avformat_find_stream_info(mFormatContext, NULL) < 0)
-		return; // Couldn't find stream information
-	
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": got stream info");
-#endif
-	
-	// Dump information about file onto standard error
-	//	av_dump_format(mFormatContext, 0, "", 0);
-	
-	// Find the first video stream
-	for (int i = 0; i < mFormatContext->nb_streams; ++i)
-	{
-		if(mFormatContext->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO)
-		{
-			mVideoStreamIndex = i;
-			break;
-		}
-	}
-	if (mVideoStreamIndex == -1)
-		return; // Didn't find a video stream
-
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": Found video stream at index " + str(mVideoStreamIndex));
-#endif
-
-	// Get a pointer to the codec context for the video stream
-	mCodecContext = mFormatContext->streams[mVideoStreamIndex]->codec;
-	
-	// Find the decoder for the video stream
-	mCodec = avcodec_find_decoder(mCodecContext->codec_id);
-	if (mCodec == NULL)
-	{
-		th_writelog("Unsupported codec!");
-		return; // Codec not found
-	}
-	// Open codec
-	if(avcodec_open2(mCodecContext, mCodec, &optionsDict) < 0)
-		return; // Could not open codec
-	
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": Codec opened");
-#endif
-
-	
-	mFrame = avcodec_alloc_frame();
-	
-#ifdef _FFMPEG_DEBUG
-	th_writelog(mName + ": Frame allocated");
-#endif
-		
-	//AVRational rational = mCodecContext->time_base;
-
-	mFPS = 25; //TODOOOOOO!!!
-	
-	mWidth = mStride = mCodecContext->width;
-	mHeight = mCodecContext->height;
-	mFrameDuration = 1.0f / mFPS;
-	mDuration = mFormatContext->duration / AV_TIME_BASE;
-	
-	if (mFrameQueue == NULL) // todo - why is this set in the backend class? it should be set in the base class, check other backends as well
-	{
-		mFrameQueue = new TheoraFrameQueue(this);
-		mFrameQueue->setSize(mNumPrecachedFrames);
-	}
-}
-
-void TheoraVideoClip_FFmpeg::unload()
-{
-	if (mInputBuffer)
-	{
-//		av_free(mInputBuffer);
-		mInputBuffer = NULL;
-	}
-	if (mAvioContext)
-	{
-		av_free(mAvioContext);
-		mAvioContext = NULL;
-	}
-	if (mFrame)
-	{
-		av_free(mFrame);
-		mFrame = NULL;
-	}
-	if (mCodecContext)
-	{
-		avcodec_close(mCodecContext);
-		mCodecContext = NULL;
-	}
-	if (mFormatContext)
-	{
-		avformat_close_input(&mFormatContext);
-		mFormatContext = NULL;
-	}
-}
-
-bool TheoraVideoClip_FFmpeg::_readData()
-{
-	return 1;
-}
-
-bool TheoraVideoClip_FFmpeg::decodeNextFrame()
-{
-	TheoraVideoFrame* frame = mFrameQueue->requestEmptyFrame();
-	if (!frame) return 0;
-
-	AVPacket packet;
-	int frameFinished;
-	
-	while (av_read_frame(mFormatContext, &packet) >= 0)
-	{
-		if (packet.stream_index == mVideoStreamIndex)
-		{
-			avcodec_decode_video2(mCodecContext, mFrame, &frameFinished, &packet);
-			
-			if (frameFinished)
-			{
-				TheoraPixelTransform t;
-				memset(&t, 0, sizeof(TheoraPixelTransform));
-
-				t.y = mFrame->data[0]; t.yStride = mFrame->linesize[0];
-				t.u = mFrame->data[1]; t.uStride = mFrame->linesize[1];
-				t.v = mFrame->data[2]; t.vStride = mFrame->linesize[2];
-				
-				frame->decode(&t);
-				frame->mTimeToDisplay = mFrameNumber / mFPS;
-				frame->mIteration = mIteration;
-				frame->_setFrameNumber(mFrameNumber++);
-
-				av_free_packet(&packet);
-				break;
-			}
-		}
-		av_free_packet(&packet);
-	}
-	return 1;
-}
- 
-void TheoraVideoClip_FFmpeg::decodedAudioCheck()
-{
-	if (!mAudioInterface || mTimer->isPaused()) return;
-	
-	mAudioMutex->lock();
-	flushAudioPackets(mAudioInterface);
-	mAudioMutex->unlock();
-}
-
-float TheoraVideoClip_FFmpeg::decodeAudio()
-{
-	return -1;
-}
-
-void TheoraVideoClip_FFmpeg::doSeek()
-{
-
-}
-
-void TheoraVideoClip_FFmpeg::_restart()
-{
-
-}
-
-#endif
diff --git a/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.h b/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.h
deleted file mode 100644
index 03f9a3d964..0000000000
--- a/drivers/theoraplayer/src/FFmpeg/TheoraVideoClip_FFmpeg.h
+++ /dev/null
@@ -1,53 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#if defined(__FFMPEG) && !defined(_TheoraVideoClip_FFmpeg_h)
-#define _TheoraVideoClip_FFmpeg_h
-
-#include "TheoraAudioPacketQueue.h"
-#include "TheoraVideoClip.h"
-
-struct AVFormatContext;
-struct AVCodecContext;
-struct AVCodec;
-struct AVFrame;
-struct AVIOContext;
-
-class TheoraVideoClip_FFmpeg : public TheoraVideoClip, public TheoraAudioPacketQueue
-{
-protected:
-	bool mLoaded;
-	
-	AVFormatContext* mFormatContext;
-	AVCodecContext* mCodecContext;
-	AVIOContext* mAvioContext;
-	AVCodec* mCodec;
-	AVFrame* mFrame;
-	unsigned char* mInputBuffer;
-	int mVideoStreamIndex;
-	int mFrameNumber;
-	
-	void unload();
-	void doSeek();
-public:
-	TheoraVideoClip_FFmpeg(TheoraDataSource* data_source,
-								 TheoraOutputMode output_mode,
-								 int nPrecachedFrames,
-								 bool usePower2Stride);
-	~TheoraVideoClip_FFmpeg();
-	
-	bool _readData();
-	bool decodeNextFrame();
-	void _restart();
-	void load(TheoraDataSource* source);
-	float decodeAudio();
-	void decodedAudioCheck();
-	std::string getDecoderName() { return "FFmpeg"; }
-};
-
-#endif
diff --git a/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.cpp b/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.cpp
deleted file mode 100644
index c4f070ec50..0000000000
--- a/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.cpp
+++ /dev/null
@@ -1,703 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef __THEORA
-#include <memory.h>
-#include <algorithm>
-#include "TheoraVideoManager.h"
-#include "TheoraFrameQueue.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraAudioInterface.h"
-#include "TheoraTimer.h"
-#include "TheoraDataSource.h"
-#include "TheoraUtil.h"
-#include "TheoraException.h"
-#include "TheoraVideoClip_Theora.h"
-#include "TheoraPixelTransform.h"
-
-TheoraVideoClip_Theora::TheoraVideoClip_Theora(TheoraDataSource* data_source,
-										TheoraOutputMode output_mode,
-										int nPrecachedFrames,
-										bool usePower2Stride):
-	TheoraVideoClip(data_source, output_mode, nPrecachedFrames, usePower2Stride),
-	TheoraAudioPacketQueue()
-{
-	mInfo.TheoraDecoder = NULL;
-	mInfo.TheoraSetup = NULL;
-	mVorbisStreams = mTheoraStreams = 0;
-	mReadAudioSamples = 0;
-	mLastDecodedFrameNumber = 0;
-}
-
-TheoraVideoClip_Theora::~TheoraVideoClip_Theora()
-{
-	if (mInfo.TheoraDecoder)
-	{
-		th_decode_free(mInfo.TheoraDecoder);
-		th_setup_free(mInfo.TheoraSetup);
-
-		if (mAudioInterface)
-		{
-			vorbis_dsp_clear(&mInfo.VorbisDSPState);
-			vorbis_block_clear(&mInfo.VorbisBlock);
-		}
-
-		ogg_stream_clear(&mInfo.TheoraStreamState);
-		th_comment_clear(&mInfo.TheoraComment);
-		th_info_clear(&mInfo.TheoraInfo);
-		
-		ogg_stream_clear(&mInfo.VorbisStreamState);
-		vorbis_comment_clear(&mInfo.VorbisComment);
-		vorbis_info_clear(&mInfo.VorbisInfo);
-		
-		ogg_sync_clear(&mInfo.OggSyncState);
-	}
-}
-
-bool TheoraVideoClip_Theora::_readData()
-{
-	int audio_eos = 0, serno;
-	float audio_time = 0;
-	float time = mTimer->getTime();
-	if (mRestarted) time = 0;
-	
-	for (;;)
-	{
-		char *buffer = ogg_sync_buffer(&mInfo.OggSyncState, 4096);
-		int bytes_read = mStream->read(buffer, 4096);
-		ogg_sync_wrote(&mInfo.OggSyncState, bytes_read);
-		
-		if (bytes_read < 4096)
-		{
-			if (bytes_read == 0)
-			{
-				if (!mAutoRestart) mEndOfFile = true;
-				return 0;
-			}
-		}
-		// when we fill the stream with enough pages, it'll start spitting out packets
-		// which contain keyframes, delta frames or audio data
-		while (ogg_sync_pageout(&mInfo.OggSyncState, &mInfo.OggPage) > 0)
-		{
-			serno = ogg_page_serialno(&mInfo.OggPage);
-			if (serno == mInfo.TheoraStreamState.serialno) ogg_stream_pagein(&mInfo.TheoraStreamState, &mInfo.OggPage);
-			if (mAudioInterface && serno == mInfo.VorbisStreamState.serialno)
-			{
-				ogg_int64_t g = ogg_page_granulepos(&mInfo.OggPage);
-				audio_time = (float) vorbis_granule_time(&mInfo.VorbisDSPState, g);
-				audio_eos = ogg_page_eos(&mInfo.OggPage);
-				ogg_stream_pagein(&mInfo.VorbisStreamState, &mInfo.OggPage);
-			}
-		}
-		if (!(mAudioInterface && !audio_eos && audio_time < time + 1.0f))
-			break;
-	}
-	return 1;
-}
-
-bool TheoraVideoClip_Theora::decodeNextFrame()
-{
-	if (mEndOfFile) return 0;
-	
-	TheoraVideoFrame* frame = mFrameQueue->requestEmptyFrame();
-	if (!frame) return 0; // max number of precached frames reached
-	bool should_restart = 0;
-	ogg_packet opTheora;
-	ogg_int64_t granulePos;
-	th_ycbcr_buffer buff;
-    int ret, nAttempts;
-	for (;;)
-	{
-        // ogg_stream_packetout can return -1 and the official docs suggest to do subsequent calls until it succeeds
-        // because the data is out of sync. still will limit the number of attempts just in case
-        for (ret = -1, nAttempts = 0; ret < 0 && nAttempts < 100; nAttempts++)
-        {
-            ret = ogg_stream_packetout(&mInfo.TheoraStreamState, &opTheora);
-        }
-		
-		if (ret > 0)
-		{
-			int status = th_decode_packetin(mInfo.TheoraDecoder, &opTheora, &granulePos);
-            if (status != 0 && status != TH_DUPFRAME) continue; // 0 means success
-
-			float time = (float) th_granule_time(mInfo.TheoraDecoder, granulePos);
-			unsigned long frame_number = (unsigned long) th_granule_frame(mInfo.TheoraDecoder, granulePos);
-			
-			if (time < mTimer->getTime() && !mRestarted && frame_number % 16 != 0)
-			{
-				// %16 operation is here to prevent a playback halt during video playback if the decoder can't keep up with demand.
-#ifdef _DEBUG
-				th_writelog(mName + ": pre-dropped frame " + str((int) frame_number));
-#endif
-				++mNumDroppedFrames;
-				continue; // drop frame
-			}
-			frame->mTimeToDisplay = time - mFrameDuration;
-			frame->mIteration     = mIteration;
-			frame->_setFrameNumber(frame_number);
-			mLastDecodedFrameNumber = frame_number;
-			th_decode_ycbcr_out(mInfo.TheoraDecoder, buff);
-			TheoraPixelTransform t;
-			memset(&t, 0, sizeof(TheoraPixelTransform));
-			
-			t.y = buff[0].data; t.yStride = buff[0].stride;
-			t.u = buff[1].data; t.uStride = buff[1].stride;
-			t.v = buff[2].data; t.vStride = buff[2].stride;
-			frame->decode(&t);
-			break;
-		}
-		else
-		{
-			if (!_readData())
-			{
-				frame->mInUse = 0;
-				should_restart = mAutoRestart;
-				break;
-			}
-		}
-	}
-	
-	if (mAudioInterface != NULL)
-	{
-		mAudioMutex->lock();
-		decodeAudio();
-		mAudioMutex->unlock();
-	}
-	if (should_restart)
-    {
-        ++mIteration;
-		_restart();
-	}
-	return 1;
-}
-
-void TheoraVideoClip_Theora::_restart()
-{
-	bool paused = mTimer->isPaused();
-	if (!paused) mTimer->pause();
-	long granule=0;
-	th_decode_ctl(mInfo.TheoraDecoder,TH_DECCTL_SET_GRANPOS,&granule,sizeof(granule));
-	th_decode_free(mInfo.TheoraDecoder);
-	mInfo.TheoraDecoder=th_decode_alloc(&mInfo.TheoraInfo,mInfo.TheoraSetup);
-	ogg_stream_reset(&mInfo.TheoraStreamState);
-	if (mAudioInterface)
-	{
-		// empty the DSP buffer
-		//float **pcm;
-		//int len = vorbis_synthesis_pcmout(&mInfo.VorbisDSPState,&pcm);
-		//if (len) vorbis_synthesis_read(&mInfo.VorbisDSPState,len);
-		ogg_packet opVorbis;
-		mReadAudioSamples = 0;
-		while (ogg_stream_packetout(&mInfo.VorbisStreamState,&opVorbis) > 0)
-		{
-			if (vorbis_synthesis(&mInfo.VorbisBlock,&opVorbis) == 0)
-				vorbis_synthesis_blockin(&mInfo.VorbisDSPState,&mInfo.VorbisBlock);
-		}
-		ogg_stream_reset(&mInfo.VorbisStreamState);
-	}
-	
-	ogg_sync_reset(&mInfo.OggSyncState);
-	mStream->seek(0);
-	ogg_int64_t granulePos = 0;
-	th_decode_ctl(mInfo.TheoraDecoder, TH_DECCTL_SET_GRANPOS, &granulePos, sizeof(granule));
-	
-	mEndOfFile = false;
-	
-	mRestarted = 1;
-	
-	if (!paused) mTimer->play();
-}
-
-void TheoraVideoClip_Theora::load(TheoraDataSource* source)
-{
-#ifdef _DEBUG
-	th_writelog("-----");
-#endif
-	mStream = source;
-	readTheoraVorbisHeaders();
-	
-	mInfo.TheoraDecoder = th_decode_alloc(&mInfo.TheoraInfo,mInfo.TheoraSetup);
-	
-	mWidth = mInfo.TheoraInfo.frame_width;
-	mHeight = mInfo.TheoraInfo.frame_height;
-    mSubFrameWidth	 = mInfo.TheoraInfo.pic_width;
-    mSubFrameHeight	 = mInfo.TheoraInfo.pic_height;
-    mSubFrameOffsetX = mInfo.TheoraInfo.pic_x;
-    mSubFrameOffsetY = mInfo.TheoraInfo.pic_y;
-    mStride = (mStride == 1) ? mStride = _nextPow2(getWidth()) : getWidth();
-	mFPS = mInfo.TheoraInfo.fps_numerator / (float) mInfo.TheoraInfo.fps_denominator;
-	
-#ifdef _DEBUG
-	th_writelog("width: " + str(mWidth) + ", height: " + str(mHeight) + ", fps: " + str((int) getFPS()));
-#endif
-	mFrameQueue = new TheoraFrameQueue(this);
-	mFrameQueue->setSize(mNumPrecachedFrames);
-	// find out the duration of the file by seeking to the end
-	// having ogg decode pages, extract the granule pos from
-	// the last theora page and seek back to beginning of the file
-	long streamSize = mStream->size(), seekPos;
-	for (int i = 1; i <= 50; ++i)
-	{
-		ogg_sync_reset(&mInfo.OggSyncState);
-		seekPos = streamSize - 4096 * i;
-		if (seekPos < 0) seekPos = 0;
-		mStream->seek(seekPos);
-		
-		char *buffer = ogg_sync_buffer(&mInfo.OggSyncState, 4096 * i);
-		int bytes_read = mStream->read(buffer, 4096 * i);
-		ogg_sync_wrote(&mInfo.OggSyncState, bytes_read);
-		ogg_sync_pageseek(&mInfo.OggSyncState, &mInfo.OggPage);
-		
-		for (;;)
-		{
-			int ret = ogg_sync_pageout(&mInfo.OggSyncState, &mInfo.OggPage);
-			if (ret == 0) break;
-			// if page is not a theora page, skip it
-			if (ogg_page_serialno(&mInfo.OggPage) != mInfo.TheoraStreamState.serialno) continue;
-			
-			ogg_int64_t granule = ogg_page_granulepos(&mInfo.OggPage);
-			if (granule >= 0)
-			{
-				mNumFrames = (int) th_granule_frame(mInfo.TheoraDecoder, granule) + 1;
-			}
-			else if (mNumFrames > 0)
-				++mNumFrames; // append delta frames at the end to get the exact numbe
-		}
-		if (mNumFrames > 0 || streamSize - 4096 * i < 0) break;
-		
-	}
-	if (mNumFrames < 0)
-		th_writelog("unable to determine file duration!");
-	else
-	{
-		mDuration = mNumFrames / mFPS;
-#ifdef _DEBUG
-		th_writelog("duration: " + strf(mDuration) + " seconds");
-#endif
-	}
-	// restore to beginning of stream.
-	ogg_sync_reset(&mInfo.OggSyncState);
-	mStream->seek(0);
-	
-	if (mVorbisStreams) // if there is no audio interface factory defined, even though the video
-		// clip might have audio, it will be ignored
-	{
-		vorbis_synthesis_init(&mInfo.VorbisDSPState, &mInfo.VorbisInfo);
-		vorbis_block_init(&mInfo.VorbisDSPState, &mInfo.VorbisBlock);
-		mNumAudioChannels = mInfo.VorbisInfo.channels;
-		mAudioFrequency = (int) mInfo.VorbisInfo.rate;
-
-		// create an audio interface instance if available
-		TheoraAudioInterfaceFactory* audio_factory = TheoraVideoManager::getSingleton().getAudioInterfaceFactory();
-		printf("**** audio factory is %p\n", audio_factory);
-		if (audio_factory) setAudioInterface(audio_factory->createInstance(this, mNumAudioChannels, mAudioFrequency));
-	}
-	
-	mFrameDuration = 1.0f / getFPS();
-#ifdef _DEBUG
-	th_writelog("-----");
-#endif
-}
-
-void TheoraVideoClip_Theora::readTheoraVorbisHeaders()
-{
-	ogg_packet tempOggPacket;
-	bool done = false;
-	bool decode_audio=TheoraVideoManager::getSingleton().getAudioInterfaceFactory() != NULL;
-	//init Vorbis/Theora Layer
-	//Ensure all structures get cleared out.
-	memset(&mInfo.OggSyncState, 0, sizeof(ogg_sync_state));
-	memset(&mInfo.OggPage, 0, sizeof(ogg_page));
-	memset(&mInfo.VorbisStreamState, 0, sizeof(ogg_stream_state));
-	memset(&mInfo.TheoraStreamState, 0, sizeof(ogg_stream_state));
-	memset(&mInfo.TheoraInfo, 0, sizeof(th_info));
-	memset(&mInfo.TheoraComment, 0, sizeof(th_comment));
-	memset(&mInfo.VorbisInfo, 0, sizeof(vorbis_info));
-	memset(&mInfo.VorbisDSPState, 0, sizeof(vorbis_dsp_state));
-	memset(&mInfo.VorbisBlock, 0, sizeof(vorbis_block));
-	memset(&mInfo.VorbisComment, 0, sizeof(vorbis_comment));
-	
-	ogg_sync_init(&mInfo.OggSyncState);
-	th_comment_init(&mInfo.TheoraComment);
-	th_info_init(&mInfo.TheoraInfo);
-	vorbis_info_init(&mInfo.VorbisInfo);
-	vorbis_comment_init(&mInfo.VorbisComment);
-	
-	while (!done)
-	{
-		char *buffer = ogg_sync_buffer(&mInfo.OggSyncState, 4096);
-		int bytes_read = mStream->read(buffer, 4096);
-		ogg_sync_wrote(&mInfo.OggSyncState, bytes_read);
-		
-		if (bytes_read == 0)
-			break;
-		
-		while (ogg_sync_pageout(&mInfo.OggSyncState, &mInfo.OggPage) > 0)
-		{
-			ogg_stream_state OggStateTest;
-			
-			//is this an initial header? If not, stop
-			if (!ogg_page_bos(&mInfo.OggPage))
-			{
-				//This is done blindly, because stream only accept themselves
-				if (mTheoraStreams) ogg_stream_pagein(&mInfo.TheoraStreamState, &mInfo.OggPage);
-				if (mVorbisStreams) ogg_stream_pagein(&mInfo.VorbisStreamState, &mInfo.OggPage);
-				
-				done=true;
-				break;
-			}
-			
-			ogg_stream_init(&OggStateTest, ogg_page_serialno(&mInfo.OggPage));
-			ogg_stream_pagein(&OggStateTest, &mInfo.OggPage);
-			ogg_stream_packetout(&OggStateTest, &tempOggPacket);
-			
-			//identify the codec
-			int ret;
-			if (!mTheoraStreams)
-			{
-				ret = th_decode_headerin(&mInfo.TheoraInfo, &mInfo.TheoraComment, &mInfo.TheoraSetup, &tempOggPacket);
-				
-				if (ret > 0)
-				{
-					//This is the Theora Header
-					memcpy(&mInfo.TheoraStreamState, &OggStateTest, sizeof(OggStateTest));
-					mTheoraStreams = 1;
-					continue;
-				}
-			}
-			if (decode_audio && !mVorbisStreams &&
-				vorbis_synthesis_headerin(&mInfo.VorbisInfo, &mInfo.VorbisComment, &tempOggPacket) >=0)
-			{
-				//This is vorbis header
-				memcpy(&mInfo.VorbisStreamState, &OggStateTest, sizeof(OggStateTest));
-				mVorbisStreams = 1;
-				continue;
-			}
-			//Hmm. I guess it's not a header we support, so erase it
-			ogg_stream_clear(&OggStateTest);
-		}
-	}
-	
-	while ((mTheoraStreams && (mTheoraStreams < 3)) ||
-		   (mVorbisStreams && (mVorbisStreams < 3)))
-	{
-		//Check 2nd'dary headers... Theora First
-		int iSuccess;
-		while (mTheoraStreams && mTheoraStreams < 3 &&
-			  (iSuccess = ogg_stream_packetout(&mInfo.TheoraStreamState, &tempOggPacket)))
-		{
-			if (iSuccess < 0)
-				throw TheoraGenericException("Error parsing Theora stream headers.");
-			if (!th_decode_headerin(&mInfo.TheoraInfo, &mInfo.TheoraComment, &mInfo.TheoraSetup, &tempOggPacket))
-				throw TheoraGenericException("invalid theora stream");
-			
-			++mTheoraStreams;
-		} //end while looking for more theora headers
-		
-		//look 2nd vorbis header packets
-		while (mVorbisStreams < 3 && (iSuccess = ogg_stream_packetout(&mInfo.VorbisStreamState, &tempOggPacket)))
-		{
-			if (iSuccess < 0)
-				throw TheoraGenericException("Error parsing vorbis stream headers");
-			
-			if (vorbis_synthesis_headerin(&mInfo.VorbisInfo, &mInfo.VorbisComment,&tempOggPacket))
-				throw TheoraGenericException("invalid stream");
-			
-			++mVorbisStreams;
-		} //end while looking for more vorbis headers
-		
-		//Not finished with Headers, get some more file data
-		if (ogg_sync_pageout(&mInfo.OggSyncState, &mInfo.OggPage) > 0)
-		{
-			if (mTheoraStreams) ogg_stream_pagein(&mInfo.TheoraStreamState, &mInfo.OggPage);
-			if (mVorbisStreams) ogg_stream_pagein(&mInfo.VorbisStreamState, &mInfo.OggPage);
-		}
-		else
-		{
-			char *buffer = ogg_sync_buffer(&mInfo.OggSyncState, 4096);
-			int bytes_read = mStream->read(buffer, 4096);
-			ogg_sync_wrote(&mInfo.OggSyncState, bytes_read);
-			
-			if (bytes_read == 0)
-				throw TheoraGenericException("End of file found prematurely");
-		}
-	} //end while looking for all headers
-	//	writelog("Vorbis Headers: " + str(mVorbisHeaders) + " Theora Headers : " + str(mTheoraHeaders));
-}
-
-void TheoraVideoClip_Theora::decodedAudioCheck()
-{
-	if (!mAudioInterface || mTimer->isPaused()) return;
-
-	mAudioMutex->lock();
-	flushAudioPackets(mAudioInterface);
-	mAudioMutex->unlock();
-}
-
-float TheoraVideoClip_Theora::decodeAudio()
-{
-	if (mRestarted) return -1;
-	
-	ogg_packet opVorbis;
-	float **pcm;
-	int len = 0;
-	float timestamp = -1;
-	bool read_past_timestamp = 0;
-	
-	float factor = 1.0f / mAudioFrequency;
-	float videoTime = (float) mLastDecodedFrameNumber / mFPS;
-	float min = mFrameQueue->getSize() / mFPS + 1.0f;
-
-	for (;;)
-	{
-		len = vorbis_synthesis_pcmout(&mInfo.VorbisDSPState, &pcm);
-		if (len == 0)
-		{
-			if (ogg_stream_packetout(&mInfo.VorbisStreamState, &opVorbis) > 0)
-			{
-				if (vorbis_synthesis(&mInfo.VorbisBlock, &opVorbis) == 0)
-				{
-					if (timestamp < 0 && opVorbis.granulepos >= 0)
-					{
-						timestamp = (float) vorbis_granule_time(&mInfo.VorbisDSPState, opVorbis.granulepos);
-					}
-					else if (timestamp >= 0) read_past_timestamp = 1;
-					vorbis_synthesis_blockin(&mInfo.VorbisDSPState, &mInfo.VorbisBlock);
-				}
-				continue;
-			}
-			else
-			{
-				float audioTime = mReadAudioSamples * factor;
-				// always buffer up of audio ahead of the frames
-				if (audioTime - videoTime < min)
-				{
-					if (!_readData()) break;
-				}
-				else
-					break;
-			}
-		}
-		addAudioPacket(pcm, len, mAudioGain);
-		mReadAudioSamples += len;
-		if (read_past_timestamp) timestamp += (float) len / mInfo.VorbisInfo.rate;
-		vorbis_synthesis_read(&mInfo.VorbisDSPState, len); // tell vorbis we read a number of samples
-	}
-	return timestamp;
-}
-
-long TheoraVideoClip_Theora::seekPage(long targetFrame, bool return_keyframe)
-{
-	int i,seek_min = 0, seek_max = (int) mStream->size();
-	long frame;
-	ogg_int64_t granule = 0;
-	
-	if (targetFrame == 0) mStream->seek(0);
-	for (i = (targetFrame == 0) ? 100 : 0; i < 100; ++i)
-	{
-		ogg_sync_reset(&mInfo.OggSyncState);
-		mStream->seek((seek_min + seek_max) / 2); // do a binary search
-		memset(&mInfo.OggPage, 0, sizeof(ogg_page));
-		ogg_sync_pageseek(&mInfo.OggSyncState, &mInfo.OggPage);
-		
-		for (;i < 1000;)
-		{
-			int ret = ogg_sync_pageout(&mInfo.OggSyncState, &mInfo.OggPage);
-			if (ret == 1)
-			{
-				int serno = ogg_page_serialno(&mInfo.OggPage);
-				if (serno == mInfo.TheoraStreamState.serialno)
-				{
-					granule = ogg_page_granulepos(&mInfo.OggPage);
-					if (granule >= 0)
-					{
-						frame = (long) th_granule_frame(mInfo.TheoraDecoder, granule);
-						if (frame < targetFrame && targetFrame - frame < 10)
-						{
-							// we're close enough, let's break this.
-							i = 1000;
-							break;
-						}
-						// we're not close enough, let's shorten the borders of the binary search
-						if (targetFrame - 1 > frame) seek_min = (seek_min + seek_max) / 2;
-						else				         seek_max = (seek_min + seek_max) / 2;
-						break;
-					}
-				}
-			}
-			else
-			{
-				char *buffer = ogg_sync_buffer(&mInfo.OggSyncState, 4096);
-				int bytes_read = mStream->read(buffer, 4096);
-				if (bytes_read == 0) break;
-				ogg_sync_wrote(&mInfo.OggSyncState, bytes_read);
-			}
-		}
-	}
-	if (return_keyframe) return (long) (granule >> mInfo.TheoraInfo.keyframe_granule_shift);
-	
-	ogg_sync_reset(&mInfo.OggSyncState);
-	memset(&mInfo.OggPage, 0, sizeof(ogg_page));
-	ogg_sync_pageseek(&mInfo.OggSyncState, &mInfo.OggPage);
-	if (targetFrame == 0) return -1;
-	mStream->seek((seek_min + seek_max) / 2); // do a binary search
-	return -1;
-}
-
-void TheoraVideoClip_Theora::doSeek()
-{
-#if _DEBUG
-	th_writelog(mName + " [seek]: seeking to frame " + str(mSeekFrame));
-#endif
-	int frame;
-	float time = mSeekFrame / getFPS();
-	mTimer->seek(time);
-	bool paused = mTimer->isPaused();
-	if (!paused) mTimer->pause(); // pause until seeking is done
-	
-	mEndOfFile = false;
-	mRestarted = false;
-	
-	resetFrameQueue();
-	// reset the video decoder.
-	ogg_stream_reset(&mInfo.TheoraStreamState);
-	th_decode_free(mInfo.TheoraDecoder);
-	mInfo.TheoraDecoder = th_decode_alloc(&mInfo.TheoraInfo, mInfo.TheoraSetup);
-	
-	if (mAudioInterface)
-	{
-		mAudioMutex->lock();
-		ogg_stream_reset(&mInfo.VorbisStreamState);
-		vorbis_synthesis_restart(&mInfo.VorbisDSPState);
-		destroyAllAudioPackets();
-	}
-	// first seek to desired frame, then figure out the location of the
-	// previous keyframe and seek to it.
-	// then by setting the correct time, the decoder will skip N frames untill
-	// we get the frame we want.
-	frame = (int) seekPage(mSeekFrame, 1); // find the keyframe nearest to the target frame
-#ifdef _DEBUG
-	//		th_writelog(mName + " [seek]: nearest keyframe for frame " + str(mSeekFrame) + " is frame: " + str(frame));
-#endif
-	seekPage(std::max(0, frame - 1), 0);
-	
-	ogg_packet opTheora;
-	ogg_int64_t granulePos;
-	bool granule_set = 0;
-	if (frame <= 1)
-	{
-		if (mInfo.TheoraInfo.version_major == 3 && mInfo.TheoraInfo.version_minor == 2 && mInfo.TheoraInfo.version_subminor == 0)
-			granulePos = 0;
-		else
-			granulePos = 1; // because of difference in granule interpretation in theora streams 3.2.0 and newer ones
-		th_decode_ctl(mInfo.TheoraDecoder, TH_DECCTL_SET_GRANPOS, &granulePos, sizeof(granulePos));
-		granule_set = 1;
-	}
-	
-	// now that we've found the keyframe that preceeds our desired frame, lets keep on decoding frames until we
-	// reach our target frame.
-	
-    int status, ret;
-	for (;mSeekFrame != 0;)
-	{
-		ret = ogg_stream_packetout(&mInfo.TheoraStreamState, &opTheora);
-		if (ret > 0)
-		{
-			if (!granule_set)
-			{
-				// theora decoder requires to set the granule pos after seek to be able to determine the current frame
-				if (opTheora.granulepos >= 0)
-				{
-					th_decode_ctl(mInfo.TheoraDecoder, TH_DECCTL_SET_GRANPOS, &opTheora.granulepos, sizeof(opTheora.granulepos));
-					granule_set = 1;
-				}
-				else continue; // ignore prev delta frames until we hit a keyframe
-			}
-			status = th_decode_packetin(mInfo.TheoraDecoder, &opTheora, &granulePos);
-            if (status != 0 && status != TH_DUPFRAME) continue;
-			frame = (int) th_granule_frame(mInfo.TheoraDecoder, granulePos);
-			if (frame >= mSeekFrame - 1) break;
-		}
-		else
-		{
-			if (!_readData())
-			{
-				th_writelog(mName + " [seek]: fineseeking failed, _readData failed!");
-				if (mAudioInterface) mAudioMutex->unlock();
-				return;
-			}
-		}
-	}
-#ifdef _DEBUG
-	//	th_writelog(mName + " [seek]: fineseeked to frame " + str(frame + 1) + ", requested: " + str(mSeekFrame));
-#endif
-	if (mAudioInterface)
-	{
-		// read audio data until we reach a timestamp. this usually takes only one iteration, but just in case let's
-		// wrap it in a loop
-		float timestamp;
-		for (;;)
-		{
-			timestamp = decodeAudio();
-			if (timestamp >= 0) break;
-			else _readData();
-		}
-		float rate = (float) mAudioFrequency * mNumAudioChannels;
-		float queued_time = getAudioPacketQueueLength();
-		// at this point there are only 2 possibilities: either we have too much packets and we have to delete
-		// the first N ones, or we don't have enough, so let's fill the gap with silence.
- 		if (time > timestamp - queued_time)
-		{
-			while (mTheoraAudioPacketQueue != NULL)
-			{
-				if (time > timestamp - queued_time + mTheoraAudioPacketQueue->numSamples / rate)
-				{
-					queued_time -= mTheoraAudioPacketQueue->numSamples / rate;
-					destroyAudioPacket(popAudioPacket());
-				}
-				else
-				{
-					int n_trim = (int) ((timestamp - queued_time + mTheoraAudioPacketQueue->numSamples / rate - time) * rate);
-					if (mTheoraAudioPacketQueue->numSamples - n_trim <= 0)
-						destroyAudioPacket(popAudioPacket()); // if there's no data to be left, just destroy it
-					else
-					{
-						for (int i = n_trim, j = 0; i < mTheoraAudioPacketQueue->numSamples; ++i, ++j)
-							mTheoraAudioPacketQueue->pcm[j] = mTheoraAudioPacketQueue->pcm[i];
-						mTheoraAudioPacketQueue->numSamples -= n_trim;
-					}
-					break;
-				}
-			}
-		}
-		else
-		{
-			// expand the first packet with silence.
-			if (mTheoraAudioPacketQueue) // just in case!
-			{
-				int i, j, nmissing = (int) ((timestamp - queued_time - time) * rate);
-				if (nmissing > 0)
-				{
-					float* samples = new float[nmissing + mTheoraAudioPacketQueue->numSamples];
-					for (i = 0; i < nmissing; ++i) samples[i] = 0;
-					for (j = 0; i < nmissing + mTheoraAudioPacketQueue->numSamples; ++i, ++j)
-						samples[i] = mTheoraAudioPacketQueue->pcm[j];
-					delete [] mTheoraAudioPacketQueue->pcm;
-					mTheoraAudioPacketQueue->pcm = samples;
-				}
-			}
-		}
-		mLastDecodedFrameNumber = mSeekFrame;
-		mReadAudioSamples = (unsigned int) (timestamp * mAudioFrequency);
-		
-		mAudioMutex->unlock();
-	}
-	if (!paused) mTimer->play();
-	mSeekFrame = -1;
-}
-#endif
diff --git a/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.h b/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.h
deleted file mode 100644
index c64c183029..0000000000
--- a/drivers/theoraplayer/src/Theora/TheoraVideoClip_Theora.h
+++ /dev/null
@@ -1,64 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#if defined(__THEORA) && !defined(_TheoraVideoClip_Theora_h)
-#define _TheoraVideoClip_Theora_h
-
-#include <ogg/ogg.h>
-#include <vorbis/vorbisfile.h>
-#include <theora/theoradec.h>
-#include "TheoraAudioPacketQueue.h"
-#include "TheoraVideoClip.h"
-
-struct TheoraInfoStruct
-{
-	// ogg/vorbis/theora variables
-	ogg_sync_state   OggSyncState;
-	ogg_page         OggPage;
-	ogg_stream_state VorbisStreamState;
-	ogg_stream_state TheoraStreamState;
-	//Theora State
-	th_info        TheoraInfo;
-	th_comment     TheoraComment;
-	th_setup_info* TheoraSetup;
-	th_dec_ctx*    TheoraDecoder;
-	//Vorbis State
-	vorbis_info      VorbisInfo;
-	vorbis_dsp_state VorbisDSPState;
-	vorbis_block     VorbisBlock;
-	vorbis_comment   VorbisComment;
-};
-
-class TheoraVideoClip_Theora : public TheoraVideoClip, public TheoraAudioPacketQueue
-{
-protected:
-	TheoraInfoStruct mInfo; // a pointer is used to avoid having to include theora & vorbis headers
-	int mTheoraStreams, mVorbisStreams;	// Keeps track of Theora and Vorbis Streams
-
-	long seekPage(long targetFrame, bool return_keyframe);
-	void doSeek();
-	void readTheoraVorbisHeaders();
-	unsigned int mReadAudioSamples;
-	unsigned long mLastDecodedFrameNumber;
-public:
-	TheoraVideoClip_Theora(TheoraDataSource* data_source,
-						   TheoraOutputMode output_mode,
-						   int nPrecachedFrames,
-						   bool usePower2Stride);
-	~TheoraVideoClip_Theora();
-
-	bool _readData();
-	bool decodeNextFrame();
-	void _restart();
-	void load(TheoraDataSource* source);
-	float decodeAudio();
-	void decodedAudioCheck();
-	std::string getDecoderName() { return "Theora"; }
-};
-
-#endif
diff --git a/drivers/theoraplayer/src/TheoraAsync.cpp b/drivers/theoraplayer/src/TheoraAsync.cpp
deleted file mode 100644
index cc3b7a4bf5..0000000000
--- a/drivers/theoraplayer/src/TheoraAsync.cpp
+++ /dev/null
@@ -1,253 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-
-#include <stdio.h>
-#include <stdlib.h>
-
-#ifdef _WIN32
-#include <windows.h>
-#else
-#include <unistd.h>
-#include <pthread.h>
-#endif
-
-#include "TheoraAsync.h"
-#include "TheoraUtil.h"
-
-#ifdef _WINRT
-#include <wrl.h>
-#endif
-
-///////////////////////////////////////////////////////////////////////////////////////////////////
-// Mutex
-///////////////////////////////////////////////////////////////////////////////////////////////////
-
-TheoraMutex::TheoraMutex()
-{
-#ifdef _WIN32
-#ifndef _WINRT // WinXP does not have CreateTheoraMutexEx()
-	mHandle = CreateMutex(0, 0, 0);
-#else
-	mHandle = CreateMutexEx(NULL, NULL, 0, SYNCHRONIZE);
-#endif
-#else
-	mHandle = (pthread_mutex_t*)malloc(sizeof(pthread_mutex_t));
-	pthread_mutex_init((pthread_mutex_t*)mHandle, 0);
-#endif
-}
-
-TheoraMutex::~TheoraMutex()
-{
-#ifdef _WIN32
-	CloseHandle(mHandle);
-#else
-	pthread_mutex_destroy((pthread_mutex_t*)mHandle);
-	free((pthread_mutex_t*)mHandle);
-	mHandle = NULL;
-#endif
-}
-
-void TheoraMutex::lock()
-{
-#ifdef _WIN32
-	WaitForSingleObjectEx(mHandle, INFINITE, FALSE);
-#else
-	pthread_mutex_lock((pthread_mutex_t*)mHandle);
-#endif
-}
-
-void TheoraMutex::unlock()
-{
-#ifdef _WIN32
-	ReleaseMutex(mHandle);
-#else
-	pthread_mutex_unlock((pthread_mutex_t*)mHandle);
-#endif
-}
-	
-///////////////////////////////////////////////////////////////////////////////////////////////////
-// Thread
-///////////////////////////////////////////////////////////////////////////////////////////////////
-
-#ifdef _WINRT
-using namespace Windows::Foundation;
-using namespace Windows::System::Threading;
-#endif
-
-#ifdef _WIN32
-unsigned long WINAPI theoraAsyncCall(void* param)
-#else
-void* theoraAsyncCall(void* param)
-#endif
-{
-	TheoraThread* t = (TheoraThread*)param;
-	t->execute();
-#ifdef _WIN32
-	return 0;
-#else
-	pthread_exit(NULL);
-	return NULL;
-#endif
-}
-
-#ifdef _WINRT
-struct TheoraAsyncActionWrapper
-{
-public:
-	IAsyncAction^ mAsyncAction;
-	TheoraAsyncActionWrapper(IAsyncAction^ asyncAction)
-	{
-		mAsyncAction = asyncAction;
-	}
-};
-#endif
-	
-TheoraThread::TheoraThread() : mRunning(false), mId(0)
-{
-#ifndef _WIN32
-	mId = (pthread_t*)malloc(sizeof(pthread_t));
-#endif
-}
-
-TheoraThread::~TheoraThread()
-{
-	if (mRunning)
-	{
-		stop();
-	}
-	if (mId != NULL)
-	{
-#ifdef _WIN32
-#ifndef _WINRT
-		CloseHandle(mId);
-#else
-		delete mId;
-#endif
-#else
-		free((pthread_t*)mId);
-#endif
-		mId = NULL;
-	}
-}
-
-void TheoraThread::start()
-{
-	mRunning = true;
-#ifdef _WIN32
-#ifndef _WINRT
-	mId = CreateThread(0, 0, &theoraAsyncCall, this, 0, 0);
-#else
-	mId = new TheoraAsyncActionWrapper(ThreadPool::RunAsync(
-		ref new WorkItemHandler([&](IAsyncAction^ work_item)
-		{
-			execute();
-		}),
-		WorkItemPriority::Normal, WorkItemOptions::TimeSliced));
-#endif
-#else
-	pthread_create((pthread_t*)mId, NULL, &theoraAsyncCall, this);
-#endif
-}
-
-bool TheoraThread::isRunning()
-{
-	bool ret;
-	mRunningMutex.lock();
-	ret = mRunning;
-	mRunningMutex.unlock();
-	
-	return ret;
-}
-
-void TheoraThread::join()
-{
-	mRunningMutex.lock();
-	mRunning = false;
-	mRunningMutex.unlock();
-#ifdef _WIN32
-#ifndef _WINRT
-	WaitForSingleObject(mId, INFINITE);
-	if (mId != NULL)
-	{
-		CloseHandle(mId);
-		mId = NULL;
-	}
-#else
-	IAsyncAction^ action = ((TheoraAsyncActionWrapper*)mId)->mAsyncAction;
-	int i = 0;
-	while (action->Status != AsyncStatus::Completed &&
-		action->Status != AsyncStatus::Canceled &&
-		action->Status != AsyncStatus::Error &&
-		i < 100)
-	{
-		_psleep(50);
-		++i;
-	}
-	if (i >= 100)
-	{
-		i = 0;
-		action->Cancel();
-		while (action->Status != AsyncStatus::Completed &&
-			action->Status != AsyncStatus::Canceled &&
-			action->Status != AsyncStatus::Error &&
-			i < 100)
-		{
-			_psleep(50);
-			++i;
-		}
-	}
-#endif
-#else
-	pthread_join(*((pthread_t*)mId), 0);
-#endif
-}
-	
-void TheoraThread::resume()
-{
-#ifdef _WIN32
-#ifndef _WINRT
-	ResumeThread(mId);
-#else
-	// not available in WinRT
-#endif
-#endif
-}
-	
-void TheoraThread::pause()
-{
-#ifdef _WIN32
-#ifndef _WINRT
-	SuspendThread(mId);
-#else
-	// not available in WinRT
-#endif
-#endif
-}
-	
-void TheoraThread::stop()
-{
-	if (mRunning)
-	{
-		mRunningMutex.lock();
-		mRunning = false;
-		mRunningMutex.unlock();
-#ifdef _WIN32
-#ifndef _WINRT
-		TerminateThread(mId, 0);
-#else
-		((TheoraAsyncActionWrapper*)mId)->mAsyncAction->Cancel();
-#endif
-#elif defined(_ANDROID)
-		pthread_kill(*((pthread_t*)mId), 0);
-#else
-		pthread_cancel(*((pthread_t*)mId));
-#endif
-	}
-}
-	
diff --git a/drivers/theoraplayer/src/TheoraAudioInterface.cpp b/drivers/theoraplayer/src/TheoraAudioInterface.cpp
deleted file mode 100644
index a265cb57b5..0000000000
--- a/drivers/theoraplayer/src/TheoraAudioInterface.cpp
+++ /dev/null
@@ -1,21 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraAudioInterface.h"
-
-TheoraAudioInterface::TheoraAudioInterface(TheoraVideoClip* owner, int nChannels, int freq)
-{
-	mFreq = freq;
-	mNumChannels = nChannels;
-	mClip = owner;
-}
-
-TheoraAudioInterface::~TheoraAudioInterface()
-{
-	
-}
diff --git a/drivers/theoraplayer/src/TheoraAudioPacketQueue.cpp b/drivers/theoraplayer/src/TheoraAudioPacketQueue.cpp
deleted file mode 100644
index be5e1018f9..0000000000
--- a/drivers/theoraplayer/src/TheoraAudioPacketQueue.cpp
+++ /dev/null
@@ -1,126 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include <stdlib.h>
-#include "TheoraAudioPacketQueue.h"
-#include "TheoraAudioInterface.h"
-
-TheoraAudioPacketQueue::TheoraAudioPacketQueue()
-{
-	mTheoraAudioPacketQueue = NULL;
-}
-
-TheoraAudioPacketQueue::~TheoraAudioPacketQueue()
-{
-	destroyAllAudioPackets();
-}
-
-float TheoraAudioPacketQueue::getAudioPacketQueueLength()
-{
-	float len = 0;
-	for (TheoraAudioPacket* p = mTheoraAudioPacketQueue; p != NULL; p = p->next)
-		len += p->numSamples;
-	
-	return len / (mAudioFrequency * mNumAudioChannels);
-}
-
-void TheoraAudioPacketQueue::_addAudioPacket(float* data, int numSamples)
-{
-	TheoraAudioPacket* packet = new TheoraAudioPacket;
-	packet->pcm = data;
-	packet->numSamples = numSamples;
-	packet->next = NULL;
-
-
-	if (mTheoraAudioPacketQueue == NULL) mTheoraAudioPacketQueue = packet;
-	else
-	{
-		TheoraAudioPacket* last = mTheoraAudioPacketQueue;
-		for (TheoraAudioPacket* p = last; p != NULL; p = p->next)
-			last = p;
-		last->next = packet;
-	}
-}
-
-void TheoraAudioPacketQueue::addAudioPacket(float** buffer, int numSamples, float gain)
-{
-	float* data = new float[numSamples * mNumAudioChannels];
-	float* dataptr = data;
-	int i;
-	unsigned int j;
-	
-	if (gain < 1.0f)
-	{
-		// apply gain, let's attenuate the samples
-		for (i = 0; i < numSamples; ++i)
-			for (j = 0; j < mNumAudioChannels; j++, ++dataptr)
-				*dataptr = buffer[i][j] * gain;
-	}
-	else
-	{
-		// do a simple copy, faster then the above method, when gain is 1.0f
-		for (i = 0; i < numSamples; ++i)
-			for (j = 0; j < mNumAudioChannels; j++, ++dataptr)
-				*dataptr = buffer[j][i];
-	}
-		
-	_addAudioPacket(data, numSamples * mNumAudioChannels);
-}
-
-void TheoraAudioPacketQueue::addAudioPacket(float* buffer, int numSamples, float gain)
-{
-	float* data = new float[numSamples * mNumAudioChannels];
-	float* dataptr = data;
-	int i, numFloats = numSamples * mNumAudioChannels;
-	
-	if (gain < 1.0f)
-	{
-		// apply gain, let's attenuate the samples
-		for (i = 0; i < numFloats; ++i, dataptr++)
-			*dataptr = buffer[i] * gain;
-	}
-	else
-	{
-		// do a simple copy, faster then the above method, when gain is 1.0f
-		for (i = 0; i < numFloats; ++i, dataptr++)
-			*dataptr = buffer[i];
-	}
-	
-	_addAudioPacket(data, numFloats);
-}
-
-TheoraAudioPacket* TheoraAudioPacketQueue::popAudioPacket()
-{
-	if (mTheoraAudioPacketQueue == NULL) return NULL;
-	TheoraAudioPacket* p = mTheoraAudioPacketQueue;
-	mTheoraAudioPacketQueue = mTheoraAudioPacketQueue->next;
-	return p;
-}
-
-void TheoraAudioPacketQueue::destroyAudioPacket(TheoraAudioPacket* p)
-{
-	if (p == NULL) return;
-	delete [] p->pcm;
-	delete p;
-}
-
-void TheoraAudioPacketQueue::destroyAllAudioPackets()
-{
-	for (TheoraAudioPacket* p = popAudioPacket(); p != NULL; p = popAudioPacket())
-		destroyAudioPacket(p);
-}
-
-void TheoraAudioPacketQueue::flushAudioPackets(TheoraAudioInterface* audioInterface)
-{
-	
-	for (TheoraAudioPacket* p = popAudioPacket(); p != NULL; p = popAudioPacket())
-	{
-		audioInterface->insertData(p->pcm, p->numSamples);
-		destroyAudioPacket(p);
-	}
-}
\ No newline at end of file
diff --git a/drivers/theoraplayer/src/TheoraDataSource.cpp b/drivers/theoraplayer/src/TheoraDataSource.cpp
deleted file mode 100644
index 6011dc6783..0000000000
--- a/drivers/theoraplayer/src/TheoraDataSource.cpp
+++ /dev/null
@@ -1,128 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include <stdio.h>
-#include <memory.h>
-#include "TheoraDataSource.h"
-#include "TheoraException.h"
-#include "TheoraVideoManager.h"
-#include "TheoraUtil.h"
-
-TheoraDataSource::~TheoraDataSource()
-{
-
-}
-
-TheoraFileDataSource::TheoraFileDataSource(std::string filename)
-{
-	mFilename = filename;
-	mFilePtr = NULL;
-}
-
-TheoraFileDataSource::~TheoraFileDataSource()
-{
-	if (mFilePtr)
-	{
-		fclose(mFilePtr);
-		mFilePtr = NULL;
-	}
-}
-
-void TheoraFileDataSource::openFile()
-{
-	if (mFilePtr == NULL)
-	{
-		mFilePtr=fopen(mFilename.c_str(), "rb");
-		if (!mFilePtr)
-        {
-            std::string msg = "Can't open video file: " + mFilename;
-            th_writelog(msg);
-            throw TheoraGenericException(msg);
-        }
-		fseek(mFilePtr, 0, SEEK_END);
-		mSize = ftell(mFilePtr);
-		fseek(mFilePtr, 0, SEEK_SET);
-	}
-}
-
-int TheoraFileDataSource::read(void* output, int nBytes)
-{
-	if (mFilePtr == NULL) openFile();
-	size_t n = fread(output, 1, nBytes, mFilePtr);
-	return (int) n;
-}
-
-void TheoraFileDataSource::seek(unsigned long byte_index)
-{
-	if (mFilePtr == NULL) openFile();
-	fseek(mFilePtr, byte_index, SEEK_SET);
-}
-
-unsigned long TheoraFileDataSource::size()
-{
-	if (mFilePtr == NULL) openFile();
-	return mSize;
-}
-
-unsigned long TheoraFileDataSource::tell()
-{
-	if (mFilePtr == NULL) return 0;
-	return ftell(mFilePtr);
-}
-
-TheoraMemoryFileDataSource::TheoraMemoryFileDataSource(std::string filename) :
-	mReadPointer(0),
-	mData(0)
-{
-	mFilename=filename;
-	FILE* f=fopen(filename.c_str(),"rb");
-	if (!f) throw TheoraGenericException("Can't open video file: "+filename);
-	fseek(f,0,SEEK_END);
-	mSize=ftell(f);
-	fseek(f,0,SEEK_SET);
-	mData=new unsigned char[mSize];
-	fread(mData,1,mSize,f);
-	fclose(f);
-}
-
-TheoraMemoryFileDataSource::TheoraMemoryFileDataSource(unsigned char* data, long size, const std::string& filename)
-{
-	mFilename = filename;
-	mData = data;
-	mSize = size;
-	mReadPointer = 0;
-}
-
-TheoraMemoryFileDataSource::~TheoraMemoryFileDataSource()
-{
-	if (mData) delete [] mData;
-}
-
-int TheoraMemoryFileDataSource::read(void* output, int nBytes)
-{
-	int n = (int) ((mReadPointer+nBytes <= mSize) ? nBytes : mSize - mReadPointer);
-	if (!n) return 0;
-	memcpy(output, mData + mReadPointer, n);
-	mReadPointer += n;
-	return n;
-}
-
-void TheoraMemoryFileDataSource::seek(unsigned long byte_index)
-{
-	mReadPointer=byte_index;
-}
-
-unsigned long TheoraMemoryFileDataSource::size()
-{
-	return mSize;
-}
-
-unsigned long TheoraMemoryFileDataSource::tell()
-{
-	return mReadPointer;
-}
diff --git a/drivers/theoraplayer/src/TheoraException.cpp b/drivers/theoraplayer/src/TheoraException.cpp
deleted file mode 100644
index 4588a81397..0000000000
--- a/drivers/theoraplayer/src/TheoraException.cpp
+++ /dev/null
@@ -1,37 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraException.h"
-#include "TheoraUtil.h"
-#include "TheoraVideoManager.h"
-#include <stdio.h>
-
-_TheoraGenericException::_TheoraGenericException(const std::string& errorText, std::string type, std::string file, int line)
-{
-    mErrText = errorText;
-	int src = (int) file.find("src");
-	if (src >= 0) file = file.substr(src + 4, 1000);
-	mLineNumber = line;
-	mFile = file;
-}
-
-
-std::string _TheoraGenericException::repr()
-{
-	std::string text = getType();
-	if (text != "") text += ": ";
-
-	if (mFile != "") text += "[" + mFile + ":" + str(mLineNumber) + "] - ";
-
-	return text + getErrorText();
-}
-
-void _TheoraGenericException::writeOutput()
-{
-	th_writelog("----------------\nException Error!\n\n" + repr() + "\n----------------");
-}
diff --git a/drivers/theoraplayer/src/TheoraFrameQueue.cpp b/drivers/theoraplayer/src/TheoraFrameQueue.cpp
deleted file mode 100644
index f402144795..0000000000
--- a/drivers/theoraplayer/src/TheoraFrameQueue.cpp
+++ /dev/null
@@ -1,174 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraFrameQueue.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraVideoManager.h"
-#include "TheoraUtil.h"
-
-
-TheoraFrameQueue::TheoraFrameQueue(TheoraVideoClip* parent)
-{
-	mParent = parent;
-}
-
-TheoraFrameQueue::~TheoraFrameQueue()
-{
-	foreach_l(TheoraVideoFrame*, mQueue)
-    {
-		delete (*it);
-    }
-	mQueue.clear();
-}
-
-TheoraVideoFrame* TheoraFrameQueue::createFrameInstance(TheoraVideoClip* clip)
-{
-	TheoraVideoFrame* frame = new TheoraVideoFrame(clip);
-	if (frame->getBuffer() == NULL) // This can happen if you run out of memory
-	{
-		delete frame;
-		return NULL;
-	}
-	return frame;
-}
-
-void TheoraFrameQueue::setSize(int n)
-{
-	mMutex.lock();
-	if (mQueue.size() > 0)
-	{
-		foreach_l (TheoraVideoFrame*, mQueue)
-        {
-			delete (*it);
-        }
-		mQueue.clear();
-	}
-	TheoraVideoFrame* frame;
-	for (int i = 0;i < n; ++i)
-	{
-		frame = createFrameInstance(mParent);
-		if (frame != NULL) mQueue.push_back(frame);
-		else
-		{
-			TheoraVideoManager::getSingleton().logMessage("TheoraFrameQueue: unable to create " + str(n) + " frames, out of memory. Created " + str((int) mQueue.size()) + " frames.");
-			break;
-		}
-	}
-	mMutex.unlock();
-}
-
-int TheoraFrameQueue::getSize()
-{
-	return (int) mQueue.size();
-}
-
-TheoraVideoFrame* TheoraFrameQueue::_getFirstAvailableFrame()
-{
-	TheoraVideoFrame* frame = mQueue.front();
-	if (frame->mReady) return frame;
-	else               return NULL;
-}
-
-TheoraVideoFrame* TheoraFrameQueue::getFirstAvailableFrame()
-{
-	mMutex.lock();
-	TheoraVideoFrame* frame = _getFirstAvailableFrame();
-	mMutex.unlock();
-	return frame;
-}
-
-void TheoraFrameQueue::clear()
-{
-	mMutex.lock();
-	foreach_l (TheoraVideoFrame*, mQueue)
-		(*it)->clear();
-	mMutex.unlock();
-}
-
-void TheoraFrameQueue::_pop(int n)
-{
-    for (int i = 0; i < n; ++i)
-    {
-        TheoraVideoFrame* first = mQueue.front();
-        first->clear();
-        mQueue.pop_front();
-        mQueue.push_back(first);
-    }
-}
-
-void TheoraFrameQueue::pop(int n)
-{
-	mMutex.lock();
-    _pop(n);
-	mMutex.unlock();
-}
-
-TheoraVideoFrame* TheoraFrameQueue::requestEmptyFrame()
-{
-	TheoraVideoFrame* frame = NULL;
-	mMutex.lock();
-	foreach_l (TheoraVideoFrame*, mQueue)
-	{
-		if (!(*it)->mInUse)
-		{
-			(*it)->mInUse = 1;
-			(*it)->mReady = 0;
-			frame = (*it);
-			break;
-		}
-	}
-	mMutex.unlock();
-	return frame;
-}
-
-int TheoraFrameQueue::getUsedCount()
-{
-	mMutex.lock();
-	int n=0;
-	foreach_l(TheoraVideoFrame*,mQueue)
-		if ((*it)->mInUse) ++n;
-	mMutex.unlock();
-	return n;
-}
-
-int TheoraFrameQueue::_getReadyCount()
-{
-	int n = 0;
-	foreach_l (TheoraVideoFrame*, mQueue)
-    if ((*it)->mReady) ++n;
-	return n;
-}
-
-
-int TheoraFrameQueue::getReadyCount()
-{
-	mMutex.lock();
-	int n = _getReadyCount();
-	mMutex.unlock();
-	return n;
-}
-
-bool TheoraFrameQueue::isFull()
-{
-	return getReadyCount() == mQueue.size();
-}
-
-void TheoraFrameQueue::lock()
-{
-	mMutex.lock();
-}
-
-void TheoraFrameQueue::unlock()
-{
-	mMutex.unlock();
-}
-
-std::list<TheoraVideoFrame*>& TheoraFrameQueue::_getFrameQueue()
-{
-    return mQueue;
-}
diff --git a/drivers/theoraplayer/src/TheoraTimer.cpp b/drivers/theoraplayer/src/TheoraTimer.cpp
deleted file mode 100644
index 644d1c2ab7..0000000000
--- a/drivers/theoraplayer/src/TheoraTimer.cpp
+++ /dev/null
@@ -1,70 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraTimer.h"
-
-TheoraTimer::TheoraTimer()
-{
-	mTime = 0;
-	mPaused = 0;
-    mSpeed = 1.0f;
-}
-
-TheoraTimer::~TheoraTimer()
-{
-
-}
-
-void TheoraTimer::update(float timeDelta)
-{
-	if (!isPaused())
-	{
-		mTime += timeDelta * mSpeed;
-	}
-}
-
-float TheoraTimer::getTime()
-{
-	return mTime;
-}
-
-void TheoraTimer::pause()
-{
-	mPaused = true;
-}
-
-void TheoraTimer::play()
-{
-	mPaused = false;
-}
-
-
-bool TheoraTimer::isPaused()
-{
-	return mPaused;
-}
-
-void TheoraTimer::stop()
-{
-
-}
-
-void TheoraTimer::seek(float time)
-{
-	mTime = time;
-}
-
-void TheoraTimer::setSpeed(float speed)
-{
-    mSpeed = speed;
-}
-
-float TheoraTimer::getSpeed()
-{
-    return mSpeed;
-}
diff --git a/drivers/theoraplayer/src/TheoraUtil.cpp b/drivers/theoraplayer/src/TheoraUtil.cpp
deleted file mode 100644
index 8f1ad0c9c1..0000000000
--- a/drivers/theoraplayer/src/TheoraUtil.cpp
+++ /dev/null
@@ -1,59 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include <stdio.h>
-#include <algorithm>
-#include <math.h>
-#include <map>
-#ifndef _WIN32
-#include <unistd.h>
-#include <pthread.h>
-#endif
-
-#include "TheoraUtil.h"
-#include "TheoraException.h"
-
-#ifdef _WIN32
-#include <windows.h>
-#pragma warning( disable: 4996 ) // MSVC++
-#endif
-
-std::string str(int i)
-{
-    char s[32];
-    sprintf(s, "%d", i);
-    return std::string(s);
-}
-
-std::string strf(float i)
-{
-    char s[32];
-    sprintf(s, "%.3f", i);
-    return std::string(s);
-}
-
-void _psleep(int miliseconds)
-{
-#ifdef _WIN32
-#ifndef _WINRT
-	Sleep(miliseconds);
-#else
-	WaitForSingleObjectEx(GetCurrentThread(), miliseconds, 0);
-#endif
-#else
-	usleep(miliseconds * 1000);
-#endif
-}
-
-
-int _nextPow2(int x)
-{
-	int y;
-	for (y = 1; y < x; y *= 2);
-	return y;
-}
diff --git a/drivers/theoraplayer/src/TheoraVideoClip.cpp b/drivers/theoraplayer/src/TheoraVideoClip.cpp
deleted file mode 100644
index 16897ee80e..0000000000
--- a/drivers/theoraplayer/src/TheoraVideoClip.cpp
+++ /dev/null
@@ -1,496 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraVideoClip.h"
-#include "TheoraVideoManager.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraFrameQueue.h"
-#include "TheoraAudioInterface.h"
-#include "TheoraTimer.h"
-#include "TheoraDataSource.h"
-#include "TheoraUtil.h"
-#include "TheoraException.h"
-
-#include "core/os/memory.h"
-
-TheoraVideoClip::TheoraVideoClip(TheoraDataSource* data_source,
-								 TheoraOutputMode output_mode,
-								 int nPrecachedFrames,
-								 bool usePower2Stride):
-	mAudioInterface(NULL),
-	mNumDroppedFrames(0),
-	mNumDisplayedFrames(0),
-	mSeekFrame(-1),
-	mDuration(-1),
-	mNumFrames(-1),
-	mFPS(1),
-	mUseAlpha(0),
-	mFrameDuration(0),
-	mName(data_source->repr()),
-	mStride(usePower2Stride),
-	mSubFrameWidth(0),
-	mSubFrameHeight(0),
-	mSubFrameOffsetX(0),
-	mSubFrameOffsetY(0),
-	mAudioGain(1),
-	mRequestedOutputMode(output_mode),
-	mAutoRestart(0),
-	mEndOfFile(0),
-	mRestarted(0),
-	mIteration(0),
-    mPlaybackIteration(0),
-	mStream(0),
-	mThreadAccessCount(0),
-	mPriority(1),
-	mFirstFrameDisplayed(0),
-	mWaitingForCache(false),
-	mOutputMode(TH_UNDEFINED)
-{
-
-	audio_track=0;
-	mAudioMutex = NULL;
-	mThreadAccessMutex = new TheoraMutex();
-	mTimer = mDefaultTimer = new TheoraTimer();
-
-	mFrameQueue = NULL;
-	mAssignedWorkerThread = NULL;
-	mNumPrecachedFrames = nPrecachedFrames;
-	setOutputMode(output_mode);
-}
-
-TheoraVideoClip::~TheoraVideoClip()
-{
-	// wait untill a worker thread is done decoding the frame
-	mThreadAccessMutex->lock();
-
-	delete mDefaultTimer;
-
-	if (mStream) memdelete(mStream);
-
-	if (mFrameQueue) delete mFrameQueue;
-
-	if (mAudioInterface)
-	{
-		mAudioMutex->lock(); // ensure a thread isn't using this mutex
-		delete mAudioInterface; // notify audio interface it's time to call it a day
-		mAudioMutex ->unlock();
-		delete mAudioMutex;
-	}
-	
-	mThreadAccessMutex->unlock();
-
-	delete mThreadAccessMutex;
-}
-
-TheoraTimer* TheoraVideoClip::getTimer()
-{
-	return mTimer;
-}
-
-void TheoraVideoClip::setTimer(TheoraTimer* timer)
-{
-	if (!timer) mTimer = mDefaultTimer;
-	else mTimer = timer;
-}
-
-void TheoraVideoClip::resetFrameQueue()
-{
-	mFrameQueue->clear();
-    mPlaybackIteration = mIteration = 0;
-}
-
-void TheoraVideoClip::restart()
-{
-	mEndOfFile = true; //temp, to prevent threads to decode while restarting
-	mThreadAccessMutex->lock();
-	_restart();
-	mTimer->seek(0);
-	mFirstFrameDisplayed = false;
-    resetFrameQueue();
-	mEndOfFile = false;
-	mRestarted = false;
-	mSeekFrame = -1;
-	mThreadAccessMutex->unlock();
-}
-
-void TheoraVideoClip::update(float timeDelta)
-{
-	if (mTimer->isPaused())
-	{
-		mTimer->update(0); // update timer in case there is some code that needs to execute each frame
-		return;
-	}
-	float time = mTimer->getTime(), speed = mTimer->getSpeed();
-    if (time + timeDelta * speed >= mDuration)
-    {
-        if (mAutoRestart && mRestarted)
-        {
-            float seekTime = time + timeDelta * speed;
-            for (;seekTime >= mDuration;)
-            {
-                seekTime -= mDuration;
-                ++mPlaybackIteration;
-            }
-
-            mTimer->seek(seekTime);
-        }
-        else
-        {
-            if (time != mDuration)
-            {
-                mTimer->update((mDuration - time) / speed);
-            }
-        }
-    }
-    else
-    {
-        mTimer->update(timeDelta);
-    }
-}
-
-float TheoraVideoClip::updateToNextFrame()
-{
-	TheoraVideoFrame* f = mFrameQueue->getFirstAvailableFrame();
-	if (!f) return 0;
-
-	float time = f->mTimeToDisplay - mTimer->getTime();
-	update(time);
-	return time;
-}
-
-TheoraFrameQueue* TheoraVideoClip::getFrameQueue()
-{
-	return mFrameQueue;
-}
-
-void TheoraVideoClip::popFrame()
-{
-	++mNumDisplayedFrames;
-	
-    // after transfering frame data to the texture, free the frame
-	// so it can be used again
-	if (!mFirstFrameDisplayed)
-	{
-		mFrameQueue->lock();
-		mFrameQueue->_pop(1);
-		mFirstFrameDisplayed = true;
-		mFrameQueue->unlock();
-	}
-	else
-	{
-		mFrameQueue->pop();
-	}
-}
-
-int TheoraVideoClip::getWidth()
-{
-	return mUseAlpha ? mWidth / 2 : mWidth;
-}
-
-int TheoraVideoClip::getHeight()
-{
-	return mHeight;
-}
-
-int TheoraVideoClip::getSubFrameWidth()
-{
-	return mUseAlpha ? mWidth / 2 : mSubFrameWidth;
-}
-
-int TheoraVideoClip::getSubFrameHeight()
-{
-	return mUseAlpha ? mHeight : mSubFrameHeight;
-}
-
-int TheoraVideoClip::getSubFrameOffsetX()
-{
-	return mUseAlpha ? 0 : mSubFrameOffsetX;
-}
-
-int TheoraVideoClip::getSubFrameOffsetY()
-{
-	return mUseAlpha ? 0 : mSubFrameOffsetY;
-}
-
-float TheoraVideoClip::getAbsPlaybackTime()
-{
-    return mTimer->getTime() + mPlaybackIteration * mDuration;
-}
-
-int TheoraVideoClip::discardOutdatedFrames(float absTime)
-{
-    int nReady = mFrameQueue->_getReadyCount();
-    // only drop frames if you have more frames to show. otherwise even the late frame will do..
-    if (nReady == 1) return 0;
-    float time = absTime;
-
-    int nPop = 0;
-    TheoraVideoFrame* frame;
-    float timeToDisplay;
-    
-    std::list<TheoraVideoFrame*>& queue = mFrameQueue->_getFrameQueue();
-    foreach_l (TheoraVideoFrame*, queue)
-    {
-        frame = *it;
-        if (!frame->mReady) break;
-        timeToDisplay = frame->mTimeToDisplay + frame->mIteration * mDuration;
-        if (time > timeToDisplay + mFrameDuration)
-        {
-            ++nPop;
-            if (nReady - nPop == 1) break; // always leave at least one in the queue
-        }
-        else break;
-    }
-    
-	if (nPop > 0)
-    {
-#define _DEBUG
-#ifdef _DEBUG
-        std::string log = getName() + ": dropped frame ";
-    
-        int i = nPop;
-        foreach_l (TheoraVideoFrame*, queue)
-        {
-            log += str((int) (*it)->getFrameNumber());
-            if (i-- > 1)
-            {
-                log += ", ";
-            }
-            else break;
-        }
-        th_writelog(log);
-#endif
-        mNumDroppedFrames += nPop;
-        mFrameQueue->_pop(nPop);
-	}
-    
-    return nPop;
-}
-
-TheoraVideoFrame* TheoraVideoClip::getNextFrame()
-{
-	TheoraVideoFrame* frame;
-    // if we are about to seek, then the current frame queue is invalidated
-	// (will be cleared when a worker thread does the actual seek)
-    if (mSeekFrame != -1) return NULL;
-
-    mFrameQueue->lock();
-	float time = getAbsPlaybackTime();
-    discardOutdatedFrames(time);
-    
-    frame = mFrameQueue->_getFirstAvailableFrame();
-    if (frame != NULL)
-    {
-        if (frame->mTimeToDisplay + frame->mIteration * mDuration > time && mFirstFrameDisplayed)
-        {
-            frame = NULL; // frame is ready but it's not yet time to display it, except when we haven't displayed any frames yet
-        }
-    }
-
-    mFrameQueue->unlock();
-	return frame;
-}
-
-std::string TheoraVideoClip::getName()
-{
-	return mName;
-}
-
-bool TheoraVideoClip::isBusy()
-{
-	return mAssignedWorkerThread || mOutputMode != mRequestedOutputMode;
-}
-
-TheoraOutputMode TheoraVideoClip::getOutputMode()
-{
-	return mOutputMode;
-}
-
-void TheoraVideoClip::setOutputMode(TheoraOutputMode mode)
-{
-	if (mode == TH_UNDEFINED) throw TheoraGenericException("Invalid output mode: TH_UNDEFINED for video: " + mName);
-	if (mOutputMode == mode) return;
-	mRequestedOutputMode = mode;
-	mUseAlpha = (mode == TH_RGBA   ||
-				 mode == TH_ARGB   ||
-				 mode == TH_BGRA   ||
-				 mode == TH_ABGR   ||
-				 mode == TH_GREY3A ||
-				 mode == TH_AGREY3 ||
-				 mode == TH_YUVA   ||
-				 mode == TH_AYUV);
-	if (mAssignedWorkerThread)
-	{
-		mThreadAccessMutex->lock();
-		// discard current frames and recreate them
-		mFrameQueue->setSize(mFrameQueue->getSize());
-		mThreadAccessMutex->unlock();
-
-	}
-	mOutputMode = mRequestedOutputMode;
-}
-
-float TheoraVideoClip::getTimePosition()
-{
-	return mTimer->getTime();
-}
-
-int TheoraVideoClip::getNumPrecachedFrames()
-{
-	return mFrameQueue->getSize();
-}
-
-void TheoraVideoClip::setNumPrecachedFrames(int n)
-{
-	if (mFrameQueue->getSize() != n)
-		mFrameQueue->setSize(n);
-}
-
-int TheoraVideoClip::_getNumReadyFrames()
-{
-	if (mSeekFrame != -1) return 0;
-	return mFrameQueue->_getReadyCount();
-}
-
-int TheoraVideoClip::getNumReadyFrames()
-{
-	if (mSeekFrame != -1) return 0; // we are about to seek, consider frame queue empty even though it will be emptied upon seek
-	return mFrameQueue->getReadyCount();
-}
-
-float TheoraVideoClip::getDuration()
-{
-	return mDuration;
-}
-
-float TheoraVideoClip::getFPS()
-{
-	return mFPS;
-}
-
-void TheoraVideoClip::play()
-{
-	mTimer->play();
-}
-
-void TheoraVideoClip::pause()
-{
-	mTimer->pause();
-}
-
-bool TheoraVideoClip::isPaused()
-{
-	return mTimer->isPaused();
-}
-
-bool TheoraVideoClip::isDone()
-{
-	return mEndOfFile && !mFrameQueue->getFirstAvailableFrame();
-}
-
-void TheoraVideoClip::stop()
-{
-	pause();
-    resetFrameQueue();
-	mFirstFrameDisplayed = false;
-	seek(0);
-}
-
-void TheoraVideoClip::setPlaybackSpeed(float speed)
-{
-	mTimer->setSpeed(speed);
-}
-
-float TheoraVideoClip::getPlaybackSpeed()
-{
-	return mTimer->getSpeed();
-}
-
-void TheoraVideoClip::seek(float time)
-{
-	seekToFrame((int) (time * getFPS()));
-}
-
-void TheoraVideoClip::seekToFrame(int frame)
-{
-	if      (frame < 0)          mSeekFrame = 0;
-	else if (frame > mNumFrames) mSeekFrame = mNumFrames;
-	else                         mSeekFrame = frame;
-
-	mFirstFrameDisplayed = false;
-	mEndOfFile = false;
-}
-
-void TheoraVideoClip::waitForCache(float desired_cache_factor, float max_wait_time)
-{
-	mWaitingForCache = true;
-	bool paused = mTimer->isPaused();
-	if (!paused) mTimer->pause();
-	int elapsed = 0;
-	int desired_num_precached_frames = (int) (desired_cache_factor * getNumPrecachedFrames());
-	while (getNumReadyFrames() < desired_num_precached_frames)
-	{
-		_psleep(10);
-		elapsed += 10;
-		if (elapsed >= max_wait_time * 1000) break;
-	}
-	if (!paused) mTimer->play();
-	mWaitingForCache = false;
-}
-
-float TheoraVideoClip::getPriority()
-{
-	return mPriority;
-}
-
-void TheoraVideoClip::setPriority(float priority)
-{
-	mPriority = priority;
-}
-
-float TheoraVideoClip::getPriorityIndex()
-{
-	float priority = (float) getNumReadyFrames();
-	if (mTimer->isPaused()) priority += getNumPrecachedFrames() / 2;
-	
-	return priority;
-}
-
-void TheoraVideoClip::setAudioInterface(TheoraAudioInterface* iface)
-{
-	mAudioInterface = iface;
-	if (iface && !mAudioMutex) mAudioMutex = new TheoraMutex;
-	if (!iface && mAudioMutex)
-	{
-		delete mAudioMutex;
-		mAudioMutex = NULL;
-	}
-}
-
-TheoraAudioInterface* TheoraVideoClip::getAudioInterface()
-{
-	return mAudioInterface;
-}
-
-void TheoraVideoClip::setAudioGain(float gain)
-{
-	if (gain > 1) mAudioGain=1;
-	if (gain < 0) mAudioGain=0;
-	else          mAudioGain=gain;
-}
-
-float TheoraVideoClip::getAudioGain()
-{
-	return mAudioGain;
-}
-
-void TheoraVideoClip::setAutoRestart(bool value)
-{
-	mAutoRestart = value;
-	if (value) mEndOfFile = false;
-}
diff --git a/drivers/theoraplayer/src/TheoraVideoFrame.cpp b/drivers/theoraplayer/src/TheoraVideoFrame.cpp
deleted file mode 100644
index b70253dabf..0000000000
--- a/drivers/theoraplayer/src/TheoraVideoFrame.cpp
+++ /dev/null
@@ -1,159 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include <memory.h>
-#include "TheoraPixelTransform.h"
-#include "TheoraVideoClip.h"
-#include "TheoraVideoFrame.h"
-#include "TheoraVideoManager.h"
-
-//#define YUV_TEST // uncomment this if you want to benchmark YUV decoding functions
-
-extern "C"
-{
-void decodeRGB  (struct TheoraPixelTransform* t);
-void decodeRGBA (struct TheoraPixelTransform* t);
-void decodeRGBX (struct TheoraPixelTransform* t);
-void decodeARGB (struct TheoraPixelTransform* t);
-void decodeXRGB (struct TheoraPixelTransform* t);
-void decodeBGR  (struct TheoraPixelTransform* t);
-void decodeBGRA (struct TheoraPixelTransform* t);
-void decodeBGRX (struct TheoraPixelTransform* t);
-void decodeABGR (struct TheoraPixelTransform* t);
-void decodeXBGR (struct TheoraPixelTransform* t);
-void decodeGrey (struct TheoraPixelTransform* t);
-void decodeGrey3(struct TheoraPixelTransform* t);
-void decodeGreyA(struct TheoraPixelTransform* t);
-void decodeGreyX(struct TheoraPixelTransform* t);
-void decodeAGrey(struct TheoraPixelTransform* t);
-void decodeXGrey(struct TheoraPixelTransform* t);
-void decodeYUV  (struct TheoraPixelTransform* t);
-void decodeYUVA (struct TheoraPixelTransform* t);
-void decodeYUVX (struct TheoraPixelTransform* t);
-void decodeAYUV (struct TheoraPixelTransform* t);
-void decodeXYUV (struct TheoraPixelTransform* t);
-}
-
-static void (*conversion_functions[])(struct TheoraPixelTransform*) = {0,
-	decodeRGB,
-	decodeRGBA,
-	decodeRGBX,
-	decodeARGB,
-	decodeXRGB,
-	decodeBGR,
-	decodeBGRA,
-	decodeBGRX,
-	decodeABGR,
-	decodeXBGR,
-	decodeGrey,
-	decodeGrey3,
-	decodeGreyA,
-	decodeGreyX,
-	decodeAGrey,
-	decodeXGrey,
-	decodeYUV,
-	decodeYUVA,
-	decodeYUVX,
-	decodeAYUV,
-	decodeXYUV
-};
-
-TheoraVideoFrame::TheoraVideoFrame(TheoraVideoClip* parent)
-{
-	mReady = mInUse = false;
-	mParent = parent;
-	mIteration = 0;
-	// number of bytes based on output mode
-	int bytemap[]={0, 3, 4, 4, 4, 4, 3, 4, 4, 4, 4, 1, 3, 4, 4, 4, 4, 3, 4, 4, 4, 4};
-	mBpp = bytemap[mParent->getOutputMode()];
-	unsigned int size = mParent->getStride() * mParent->mHeight * mBpp;
-	try
-	{
-		mBuffer = new unsigned char[size];
-	}
-	catch (std::bad_alloc)
-	{
-		mBuffer = NULL;
-		return;
-	}
-	memset(mBuffer, 255, size);
-}
-
-TheoraVideoFrame::~TheoraVideoFrame()
-{
-	if (mBuffer) delete [] mBuffer;
-}
-
-int TheoraVideoFrame::getWidth()
-{
-	return mParent->getWidth();
-}
-
-int TheoraVideoFrame::getStride()
-{
-	return mParent->mStride;
-}
-
-int TheoraVideoFrame::getHeight()
-{
-	return mParent->getHeight();
-}
-
-unsigned char* TheoraVideoFrame::getBuffer()
-{
-	return mBuffer;
-}
-
-void TheoraVideoFrame::decode(struct TheoraPixelTransform* t)
-{
-	if (t->raw != NULL)
-	{
-		int bufferStride = mParent->getWidth() * mBpp;
-		if (bufferStride == t->rawStride)
-		{
-			memcpy(mBuffer, t->raw, t->rawStride * mParent->getHeight());
-		}
-		else
-		{
-			unsigned char *buff = mBuffer, *src = t->raw;
-			int i, h = mParent->getHeight();
-			for (i = 0; i < h; ++i, buff += bufferStride, src += t->rawStride)
-			{
-				memcpy(buff, src, bufferStride);
-			}
-		}
-	}
-	else
-	{
-		t->out = mBuffer;
-		t->w = mParent->getWidth();
-		t->h = mParent->getHeight();
-        
-#ifdef YUV_TEST // when benchmarking yuv conversion functions during development, do a timed average
-        #define N 1000
-        clock_t time = clock();
-        for (int i = 0; i < N; ++i)
-        {
-            conversion_functions[mParent->getOutputMode()](t);
-        }
-        float diff = (clock() - time) * 1000.0f / CLOCKS_PER_SEC;
-        
-		char s[128];
-		sprintf(s, "%.2f", diff / N);
-        TheoraVideoManager::getSingleton().logMessage("YUV Decoding time: " + std::string(s) + " ms\n");
-#else
-		conversion_functions[mParent->getOutputMode()](t);
-#endif
-	}
-	mReady = true;
-}
-
-void TheoraVideoFrame::clear()
-{
-	mInUse = mReady = false;
-}
diff --git a/drivers/theoraplayer/src/TheoraVideoManager.cpp b/drivers/theoraplayer/src/TheoraVideoManager.cpp
deleted file mode 100644
index 53b211374a..0000000000
--- a/drivers/theoraplayer/src/TheoraVideoManager.cpp
+++ /dev/null
@@ -1,485 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "TheoraVideoManager.h"
-#include "TheoraWorkerThread.h"
-#include "TheoraVideoClip.h"
-#include "TheoraFrameQueue.h"
-#include "TheoraAudioInterface.h"
-#include "TheoraUtil.h"
-#include "TheoraDataSource.h"
-#include "TheoraException.h"
-#ifdef __THEORA
-	#include <theora/codec.h>
-	#include <vorbis/codec.h>
-	#include "TheoraVideoClip_Theora.h"
-#endif
-#ifdef __AVFOUNDATION
-	#include "TheoraVideoClip_AVFoundation.h"
-#endif
-#ifdef __FFMPEG
-	#include "TheoraVideoClip_FFmpeg.h"
-#endif
-#ifdef _ANDROID //libtheoraplayer addition for cpu feature detection
-	#include "cpu-features.h"
-#endif
-// declaring function prototype here so I don't have to put it in a header file
-// it only needs to be used by this plugin and called once
-extern "C"
-{
-	void initYUVConversionModule();
-}
-
-#include "core/os/memory.h"
-
-//#define _DECODING_BENCHMARK //uncomment to test average decoding time on a given device
-
-
-// --------------------------
-//#define _SCHEDULING_DEBUG
-#ifdef _SCHEDULING_DEBUG
-float gThreadDiagnosticTimer = 0;
-#endif
-// --------------------------
-
-#ifdef _DECODING_BENCHMARK
-void benchmark(TheoraVideoClip* clip)
-{
-	int nPrecached = 256;
-	int n = nPrecached;
-	char msg[1024];
-	clock_t t = clock();
-	while (n > 0)
-	{
-		clip->waitForCache(1.0f, 1000000);
-		n -= 32;
-		clip->getFrameQueue()->clear();
-	}
-	float diff = ((float) (clock() - t) * 1000.0f) / CLOCKS_PER_SEC;
-	sprintf(msg, "BENCHMARK: %s: Decoding %d frames took %.1fms (%.2fms average per frame)\n",clip->getName().c_str(), nPrecached, diff, diff / nPrecached);
-	TheoraVideoManager::getSingleton().logMessage(msg);
-	clip->seek(0);
-}
-#endif
-
-struct TheoraWorkCandidate
-{
-	TheoraVideoClip* clip;
-	float priority, queuedTime, workTime, entitledTime;
-};
-
-TheoraVideoManager* g_ManagerSingleton = NULL;
-
-void theora_writelog(std::string output)
-{
-	printf("%s\n", output.c_str());
-}
-
-void (*g_LogFuction)(std::string) = theora_writelog;
-
-void TheoraVideoManager::setLogFunction(void (*fn)(std::string))
-{
-	g_LogFuction = fn;
-}
-
-TheoraVideoManager* TheoraVideoManager::getSingletonPtr()
-{
-    return g_ManagerSingleton;
-}
-
-TheoraVideoManager& TheoraVideoManager::getSingleton()
-{  
-    return *g_ManagerSingleton;  
-}
-
-TheoraVideoManager::TheoraVideoManager(int num_worker_threads) : 
-	mDefaultNumPrecachedFrames(8)
-{
-	if (num_worker_threads < 1) throw TheoraGenericException("Unable to create TheoraVideoManager, at least one worker thread is reqired");
-
-	g_ManagerSingleton = this;
-
-	std::string msg = "Initializing Theora Playback Library (" + getVersionString() + ")\n";
-#ifdef __THEORA
-	msg += "  - libtheora version: " + std::string(th_version_string()) + "\n" +
-	       "  - libvorbis version: " +  std::string(vorbis_version_string()) + "\n";
-#endif
-#ifdef _ANDROID
-	uint64_t features = android_getCpuFeaturesExt();
-	char s[128];
-	sprintf(s, "  - Android: CPU Features: %u\n", (unsigned int) features);
-	msg += s;
-	if ((features & ANDROID_CPU_ARM_FEATURE_NEON) == 0)
-		msg += "  - Android: NEON features NOT SUPPORTED by CPU\n";
-	else
-		msg += "  - Android: Detected NEON CPU features\n";
-#endif
-
-#ifdef __AVFOUNDATION
-	msg += "  - using Apple AVFoundation classes.\n";
-#endif
-#ifdef __FFMPEG
-	msg += "  - using FFmpeg library.\n";
-#endif
-	
-	logMessage(msg + "------------------------------------");
-	mAudioFactory = NULL;
-	mWorkMutex = new TheoraMutex();
-
-	// for CPU based yuv2rgb decoding
-	initYUVConversionModule();
-
-	createWorkerThreads(num_worker_threads);
-}
-
-TheoraVideoManager::~TheoraVideoManager()
-{
-	destroyWorkerThreads();
-
-	mWorkMutex->lock();
-	ClipList::iterator ci;
-	for (ci = mClips.begin(); ci != mClips.end(); ++ci)
-		delete (*ci);
-	mClips.clear();
-	mWorkMutex->unlock();
-	delete mWorkMutex;
-}
-
-void TheoraVideoManager::logMessage(std::string msg)
-{
-	g_LogFuction(msg);
-}
-
-TheoraVideoClip* TheoraVideoManager::getVideoClipByName(std::string name)
-{
-	TheoraVideoClip* clip = NULL;
-	mWorkMutex->lock();
-
-	foreach(TheoraVideoClip*, mClips)
-	{
-		if ((*it)->getName() == name)
-		{
-			clip = *it;
-			break;
-		}
-	}
-	mWorkMutex->unlock();
-
-	return clip;
-}
-
-void TheoraVideoManager::setAudioInterfaceFactory(TheoraAudioInterfaceFactory* factory)
-{
-	mAudioFactory = factory;
-}
-
-TheoraAudioInterfaceFactory* TheoraVideoManager::getAudioInterfaceFactory()
-{
-	return mAudioFactory;
-}
-
-TheoraVideoClip* TheoraVideoManager::createVideoClip(std::string filename,
-													 TheoraOutputMode output_mode,
-													 int numPrecachedOverride,
-													 bool usePower2Stride,
-													 int p_track)
-{
-	TheoraDataSource* src=memnew(TheoraFileDataSource(filename));
-	return createVideoClip(src,output_mode,numPrecachedOverride,usePower2Stride, p_track);
-}
-
-TheoraVideoClip* TheoraVideoManager::createVideoClip(TheoraDataSource* data_source,
-													 TheoraOutputMode output_mode,
-													 int numPrecachedOverride,
-													 bool usePower2Stride,
-													 int p_audio_track)
-{
-	mWorkMutex->lock();
-
-	TheoraVideoClip* clip = NULL;
-	int nPrecached = numPrecachedOverride ? numPrecachedOverride : mDefaultNumPrecachedFrames;
-	logMessage("Creating video from data source: " + data_source->repr() + " [" + str(nPrecached) + " precached frames].");
-	
-#ifdef __AVFOUNDATION
-	TheoraFileDataSource* fileDataSource = dynamic_cast<TheoraFileDataSource*>(data_source);
-	std::string filename;
-	if (fileDataSource == NULL)
-	{
-		TheoraMemoryFileDataSource* memoryDataSource = dynamic_cast<TheoraMemoryFileDataSource*>(data_source);
-		if (memoryDataSource != NULL) filename = memoryDataSource->getFilename();
-		// if the user has his own data source, it's going to be a problem for AVAssetReader since it only supports reading from files...
-	}
-	else filename = fileDataSource->getFilename();
-
-	if (filename.size() > 4 && filename.substr(filename.size() - 4, filename.size()) == ".mp4")
-	{
-		clip = new TheoraVideoClip_AVFoundation(data_source, output_mode, nPrecached, usePower2Stride);
-	}
-#endif
-#if defined(__AVFOUNDATION) && defined(__THEORA)
-	else
-#endif
-#ifdef __THEORA
-		clip = new TheoraVideoClip_Theora(data_source, output_mode, nPrecached, usePower2Stride);
-#endif
-#ifdef __FFMPEG
-		clip = new TheoraVideoClip_FFmpeg(data_source, output_mode, nPrecached, usePower2Stride);
-#endif
-
-	clip->set_audio_track(p_audio_track);
-	clip->load(data_source);
-	clip->decodeNextFrame(); // ensure the first frame is always preloaded and have the main thread do it to prevent potential thread starvatio
-
-	mClips.push_back(clip);
-	mWorkMutex->unlock();
-	
-#ifdef _DECODING_BENCHMARK
-	benchmark(clip);
-#endif
-	return clip;
-}
-
-void TheoraVideoManager::destroyVideoClip(TheoraVideoClip* clip)
-{
-	if (clip)
-	{
-		th_writelog("Destroying video clip: " + clip->getName());
-		mWorkMutex->lock();
-		bool reported = 0;
-		while (clip->mAssignedWorkerThread)
-		{
-			if (!reported)
-			{
-				th_writelog(" - Waiting for WorkerThread to finish decoding in order to destroy");
-				reported = 1;
-			}
-			_psleep(1);
-		}
-		if (reported) th_writelog(" - WorkerThread done, destroying...");
-		
-		// erase the clip from the clip list
-		foreach (TheoraVideoClip*, mClips)
-		{
-			if ((*it) == clip)
-			{
-				mClips.erase(it);
-				break;
-			}
-		}
-		// remove all it's references from the work log
-		mWorkLog.remove(clip);
-
-		// delete the actual clip
-		delete clip;
-#ifdef _DEBUG
-		th_writelog("Destroyed video.");
-#endif
-		mWorkMutex->unlock();
-	}
-}
-
-TheoraVideoClip* TheoraVideoManager::requestWork(TheoraWorkerThread* caller)
-{
-	if (!mWorkMutex) return NULL;
-	mWorkMutex->lock();
-
-	TheoraVideoClip* selectedClip = NULL;
-	float maxQueuedTime = 0, totalAccessCount = 0, prioritySum = 0, diff, maxDiff = -1;
-	int nReadyFrames;
-	std::vector<TheoraWorkCandidate> candidates;
-	TheoraVideoClip* clip;
-	TheoraWorkCandidate candidate;
-
-	// first pass is for playing videos, but if no such videos are available for decoding
-	// paused videos are selected in the second pass.
-    // Note that paused videos that are waiting for cache are considered equal to playing
-    // videos in the scheduling context
-
-	for (int i = 0; i < 2 && candidates.size() == 0; ++i)
-	{
-		foreach (TheoraVideoClip*, mClips)
-		{
-			clip = *it;
-			if (clip->isBusy() || (i == 0 && clip->isPaused() && !clip->mWaitingForCache)) continue;
-			nReadyFrames = clip->getNumReadyFrames();
-			if (nReadyFrames == clip->getFrameQueue()->getSize()) continue;
-
-			candidate.clip = clip;
-			candidate.priority = clip->getPriority();
-			candidate.queuedTime = (float) nReadyFrames / (clip->getFPS() * clip->getPlaybackSpeed());
-			candidate.workTime = (float) clip->mThreadAccessCount;
-			
-			totalAccessCount += candidate.workTime;
-			if (maxQueuedTime < candidate.queuedTime) maxQueuedTime = candidate.queuedTime;
-
-			candidates.push_back(candidate);
-		}
-	}
-
-	// prevent division by zero
-	if (totalAccessCount == 0) totalAccessCount = 1;
-	if (maxQueuedTime == 0) maxQueuedTime = 1;
-
-	// normalize candidate values
-	foreach (TheoraWorkCandidate, candidates)
-	{
-		it->workTime /= totalAccessCount;
-		// adjust user priorities to favor clips that have fewer frames queued
-		it->priority *= 1.0f - (it->queuedTime / maxQueuedTime) * 0.5f;
-		prioritySum += it->priority;
-	}
-	foreach (TheoraWorkCandidate, candidates)
-	{
-		it->entitledTime = it->priority / prioritySum;
-	}
-
-	// now, based on how much access time has been given to each clip in the work log
-	// and how much time should be given to each clip based on calculated priorities,
-	// we choose a best suited clip for this worker thread to decode next
-	foreach (TheoraWorkCandidate, candidates)
-	{
-		diff = it->entitledTime - it->workTime;
-
-		if (maxDiff < diff)
-		{
-			maxDiff = diff;
-			selectedClip = it->clip;
-		}
-	}
-
-	if (selectedClip)
-	{
-		selectedClip->mAssignedWorkerThread = caller;
-		
-		int nClips = (int) mClips.size();
-		unsigned int maxWorkLogSize = (nClips - 1) * 50;
-
-		if (nClips > 1)
-		{
-			mWorkLog.push_front(selectedClip);
-			++selectedClip->mThreadAccessCount;
-		}
-		
-		TheoraVideoClip* c;
-		while (mWorkLog.size() > maxWorkLogSize)
-		{
-			c = mWorkLog.back();
-			mWorkLog.pop_back();
-			c->mThreadAccessCount--;
-		}
-#ifdef _SCHEDULING_DEBUG
-		if (mClips.size() > 1)
-		{
-			int accessCount = mWorkLog.size();
-			if (gThreadDiagnosticTimer > 2.0f)
-			{
-				gThreadDiagnosticTimer = 0;
-				std::string logstr = "-----\nTheora Playback Library debug CPU time analysis (" + str(accessCount) + "):\n";
-				int percent;
-				foreach (TheoraVideoClip*, mClips)
-				{
-					percent = ((float) (*it)->mThreadAccessCount / mWorkLog.size()) * 100.0f;
-					logstr += (*it)->getName() + " (" + str((*it)->getPriority()) + "): " + str((*it)->mThreadAccessCount) + ", " + str(percent) + "%\n";
-				}
-				logstr += "-----";
-				th_writelog(logstr);
-			}
-		}
-#endif
-	}
-
-	mWorkMutex->unlock();
-	return selectedClip;
-}
-
-void TheoraVideoManager::update(float timeDelta)
-{
-	mWorkMutex->lock();
-	foreach (TheoraVideoClip*, mClips)
-	{
-		(*it)->update(timeDelta);
-		(*it)->decodedAudioCheck();
-	}
-	mWorkMutex->unlock();
-#ifdef _SCHEDULING_DEBUG
-	gThreadDiagnosticTimer += timeDelta;
-#endif
-}
-
-int TheoraVideoManager::getNumWorkerThreads()
-{
-	return (int) mWorkerThreads.size();
-}
-
-void TheoraVideoManager::createWorkerThreads(int n)
-{
-	TheoraWorkerThread* t;
-	for (int i=0;i<n;++i)
-	{
-		t=new TheoraWorkerThread();
-		t->start();
-		mWorkerThreads.push_back(t);
-	}
-}
-
-void TheoraVideoManager::destroyWorkerThreads()
-{
-	foreach(TheoraWorkerThread*,mWorkerThreads)
-	{
-		(*it)->join();
-		delete (*it);
-	}
-	mWorkerThreads.clear();
-}
-
-void TheoraVideoManager::setNumWorkerThreads(int n)
-{
-	if (n == getNumWorkerThreads()) return;
-	if (n < 1) throw TheoraGenericException("Unable to change the number of worker threads in TheoraVideoManager, at least one worker thread is reqired");
-
-	th_writelog("changing number of worker threats to: "+str(n));
-
-	destroyWorkerThreads();
-	createWorkerThreads(n);
-}
-
-std::string TheoraVideoManager::getVersionString()
-{
-	int a, b, c;
-	getVersion(&a, &b, &c);
-	std::string out = str(a) + "." + str(b);
-	if (c != 0)
-	{
-		if (c < 0) out += " RC" + str(-c);
-		else       out += "." + str(c);
-	}
-	return out;
-}
-
-void TheoraVideoManager::getVersion(int* a, int* b, int* c) // TODO, return a struct instead of the current solution.
-{
-	*a = 1;
-	*b = 1;
-	*c = 0;
-}
-
-std::vector<std::string> TheoraVideoManager::getSupportedDecoders()
-{
-	std::vector<std::string> lst;
-#ifdef __THEORA
-	lst.push_back("Theora");
-#endif
-#ifdef __AVFOUNDATION
-	lst.push_back("AVFoundation");
-#endif
-#ifdef __FFMPEG
-	lst.push_back("FFmpeg");
-#endif
-	
-	return lst;
-}
diff --git a/drivers/theoraplayer/src/TheoraWorkerThread.cpp b/drivers/theoraplayer/src/TheoraWorkerThread.cpp
deleted file mode 100644
index cef8545b8d..0000000000
--- a/drivers/theoraplayer/src/TheoraWorkerThread.cpp
+++ /dev/null
@@ -1,49 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef _WIN32
-#pragma warning( disable: 4251 ) // MSVC++
-#endif
-#include "TheoraWorkerThread.h"
-#include "TheoraVideoManager.h"
-#include "TheoraVideoClip.h"
-#include "TheoraUtil.h"
-
-TheoraWorkerThread::TheoraWorkerThread() : TheoraThread()
-{
-	mClip = NULL;
-}
-
-TheoraWorkerThread::~TheoraWorkerThread()
-{
-
-}
-
-void TheoraWorkerThread::execute()
-{
-	while (isRunning())
-	{
-		mClip = TheoraVideoManager::getSingleton().requestWork(this);
-		if (!mClip)
-		{
-			_psleep(100);
-			continue;
-		}
-
-		mClip->mThreadAccessMutex->lock();
-		// if user requested seeking, do that then.
-		if (mClip->mSeekFrame >= 0) mClip->doSeek();
-
-		if (!mClip->decodeNextFrame())
-			_psleep(1); // this happens when the video frame queue is full.
-
-		mClip->mAssignedWorkerThread = NULL;
-		mClip->mThreadAccessMutex->unlock();
-		mClip = NULL;
-	}
-}
diff --git a/drivers/theoraplayer/src/YUV/C/yuv420_grey_c.c b/drivers/theoraplayer/src/YUV/C/yuv420_grey_c.c
deleted file mode 100644
index 8af5dd1f58..0000000000
--- a/drivers/theoraplayer/src/YUV/C/yuv420_grey_c.c
+++ /dev/null
@@ -1,56 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "yuv_util.h"
-
-static void _decodeGrey3(struct TheoraPixelTransform* t, int stride, int nBytes)
-{
-	unsigned char *ySrc = t->y, *yLineEnd, *out = t->out;
-	unsigned int y;
-	for (y = 0; y < t->h; ++y, ySrc += t->yStride - t->w, out += stride-t->w * nBytes)
-		for (yLineEnd = ySrc + t->w; ySrc != yLineEnd; ++ySrc, out += nBytes)
-			out[0] = out[1] = out[2] = *ySrc;
-}
-
-void decodeGrey(struct TheoraPixelTransform* t)
-{
-	unsigned char *ySrc = t->y, *yLineEnd, *out = t->out;
-	unsigned int y;
-	for (y = 0; y < t->h; ++y, ySrc += t->yStride - t->w)
-		for (yLineEnd = ySrc + t->w; ySrc != yLineEnd; ++ySrc, ++out)
-			*out = *ySrc;
-
-}
-
-void decodeGrey3(struct TheoraPixelTransform* t)
-{
-	_decodeGrey3(t, t->w * 3, 3);
-}
-
-void decodeGreyA(struct TheoraPixelTransform* t)
-{
-	_decodeGrey3(t, t->w * 4, 4);
-	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeGreyX(struct TheoraPixelTransform* t)
-{
-	_decodeGrey3(t, t->w * 4, 4);
-}
-
-void decodeAGrey(struct TheoraPixelTransform* t)
-{
-	_decodeGrey3(incOut(t, 1), t->w * 4, 4);
-	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXGrey(struct TheoraPixelTransform* t)
-{
-	_decodeGrey3(incOut(t, 1), t->w * 4, 4);
-}
-
diff --git a/drivers/theoraplayer/src/YUV/C/yuv420_rgb_c.c b/drivers/theoraplayer/src/YUV/C/yuv420_rgb_c.c
deleted file mode 100644
index e981e75ead..0000000000
--- a/drivers/theoraplayer/src/YUV/C/yuv420_rgb_c.c
+++ /dev/null
@@ -1,358 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef _YUV_C
-#include "yuv_util.h"
-
-int YTable [256];
-int BUTable[256];
-int GUTable[256];
-int GVTable[256];
-int RVTable[256];
-
-#define CLIP_RGB_COLOR(dst, x) \
-	tmp = (x) >> 13;\
-	if ((tmp & ~0xFF) == 0) dst = tmp;\
-	else                    dst = (-tmp) >> 31;
-
-#define _decodeRGB(t, stride, nBytes, maxWidth, i1, i2, i3, j1, j2, j3)\
-	register int tmp;\
-	int nBytes2 = nBytes * 2, cv, cu, rgbY1, rgbY2, rgbY3, rgbY4, rV, gUV, bU, width = maxWidth == 0 ? t->w : maxWidth;\
-	unsigned int y;\
-	unsigned char *ySrcEven, *ySrcOdd, *yLineEnd, *uSrc, *vSrc, *out1, *out2;\
-	\
-	for (y = 0; y < t->h; y += 2)\
-	{\
-		ySrcEven = t->y + y * t->yStride;\
-		ySrcOdd  = t->y + (y + 1) * t->yStride;\
-		uSrc = t->u + y * t->uStride / 2;\
-		vSrc = t->v + y * t->vStride / 2;\
-		out1 = t->out + y * stride;\
-		out2 = t->out + (y + 1) * stride;\
-		\
-		for (yLineEnd = ySrcEven + width; ySrcEven != yLineEnd;)\
-		{\
-			cu = *uSrc; ++uSrc;\
-			cv = *vSrc; ++vSrc;\
-			rV   = RVTable[cv];\
-			gUV  = GUTable[cu] + GVTable[cv];\
-			bU   = BUTable[cu];\
-			\
-			rgbY1 = YTable[*ySrcEven]; ++ySrcEven;\
-			rgbY2 = YTable[*ySrcOdd];  ++ySrcOdd;\
-			rgbY3 = YTable[*ySrcEven]; ++ySrcEven;\
-			rgbY4 = YTable[*ySrcOdd];  ++ySrcOdd;\
-			\
-			CLIP_RGB_COLOR(out1[i1], rgbY1 + rV );\
-			CLIP_RGB_COLOR(out1[i2], rgbY1 - gUV);\
-			CLIP_RGB_COLOR(out1[i3], rgbY1 + bU );\
-			\
-			CLIP_RGB_COLOR(out2[i1], rgbY2 + rV );\
-			CLIP_RGB_COLOR(out2[i2], rgbY2 - gUV);\
-			CLIP_RGB_COLOR(out2[i3], rgbY2 + bU );\
-			\
-			CLIP_RGB_COLOR(out1[j1], rgbY3 + rV );\
-			CLIP_RGB_COLOR(out1[j2], rgbY3 - gUV);\
-			CLIP_RGB_COLOR(out1[j3], rgbY3 + bU );\
-			\
-			CLIP_RGB_COLOR(out2[j1], rgbY4 + rV );\
-			CLIP_RGB_COLOR(out2[j2], rgbY4 - gUV);\
-			CLIP_RGB_COLOR(out2[j3], rgbY4 + bU );\
-			\
-			out1 += nBytes2;  out2 += nBytes2;\
-		}\
-	}
-
-// The 'trick' with this function is that it skips decoding YUV pixels if the alpha value is 0, thus improving the decoding speed of a frame
-#define _decodeRGBA(t, stride, nBytes, maxWidth, i1, i2, i3, j1, j2, j3, aindex1, aindex2)\
-\
-	register int tmp;\
-	int nBytes2 = nBytes * 2, cv, cu, rgbY1, rgbY2, rgbY3, rgbY4, a1, a2, a3, a4, rV, gUV, bU, width = maxWidth == 0 ? t->w : maxWidth;\
-	int alphaStride = t->w;\
-	unsigned int y;\
-	unsigned char *ySrcEven, *ySrcOdd, *yLineEnd, *uSrc, *vSrc, *out1, *out2;\
-	\
-	for (y = 0; y < t->h; y += 2)\
-	{\
-		ySrcEven = t->y + y * t->yStride;\
-		ySrcOdd  = t->y + (y + 1) * t->yStride;\
-		uSrc = t->u + y * t->uStride / 2;\
-		vSrc = t->v + y * t->vStride / 2;\
-		out1 = t->out + y * stride;\
-		out2 = t->out + (y + 1) * stride;\
-		\
-		for (yLineEnd = ySrcEven + width; ySrcEven != yLineEnd;)\
-		{\
-			cu = *uSrc; ++uSrc;\
-			cv = *vSrc; ++vSrc;\
-			rV   = RVTable[cv];\
-			gUV  = GUTable[cu] + GVTable[cv];\
-			bU   = BUTable[cu];\
-			\
-			rgbY1 = YTable[*ySrcEven]; a1 = ySrcEven[alphaStride]; ++ySrcEven;\
-			rgbY2 = YTable[*ySrcOdd];  a2 = ySrcOdd [alphaStride];  ++ySrcOdd;\
-			rgbY3 = YTable[*ySrcEven]; a3 = ySrcEven[alphaStride]; ++ySrcEven;\
-			rgbY4 = YTable[*ySrcOdd];  a4 = ySrcOdd [alphaStride];  ++ySrcOdd;\
-			\
-			if (a1 > 16)\
-			{\
-				CLIP_RGB_COLOR(out1[i1], rgbY1 + rV );\
-				CLIP_RGB_COLOR(out1[i2], rgbY1 - gUV);\
-				CLIP_RGB_COLOR(out1[i3], rgbY1 + bU );\
-				out1[aindex1] = a1 >= 235 ? 255 : (unsigned char) (((a1 - 16) * 255) / 219);\
-			}\
-			else *((unsigned int*) out1) = 0;\
-			\
-			if (a2 > 16)\
-			{\
-				CLIP_RGB_COLOR(out2[i1], rgbY2 + rV );\
-				CLIP_RGB_COLOR(out2[i2], rgbY2 - gUV);\
-				CLIP_RGB_COLOR(out2[i3], rgbY2 + bU );\
-				out2[aindex1] = a2 >= 235 ? 255 : (unsigned char) (((a2 - 16) * 255) / 219);\
-			}\
-			else *((unsigned int*) out2) = 0;\
-			\
-			if (a3 > 16)\
-			{\
-				CLIP_RGB_COLOR(out1[j1], rgbY3 + rV );\
-				CLIP_RGB_COLOR(out1[j2], rgbY3 - gUV);\
-				CLIP_RGB_COLOR(out1[j3], rgbY3 + bU );\
-				out1[aindex2] = a3 >= 235 ? 255 : (unsigned char) (((a3 - 16) * 255) / 219);\
-			}\
-			else *((unsigned int*) &out1[4]) = 0;\
-			\
-			if (a4 > 16)\
-			{\
-				CLIP_RGB_COLOR(out2[j1], rgbY4 + rV );\
-				CLIP_RGB_COLOR(out2[j2], rgbY4 - gUV);\
-				CLIP_RGB_COLOR(out2[j3], rgbY4 + bU );\
-				out2[aindex2] = a4 >= 235 ? 255 : (unsigned char) (((a4 - 16) * 255) / 219);\
-			}\
-			else *((unsigned int*) &out2[4]) = 0;\
-			\
-			out1 += nBytes2;  out2 += nBytes2;\
-		}\
-	}\
-
-void decodeRGB(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 3, 3, 0, 0, 1, 2, 3, 4, 5);
-}
-
-void decodeRGBA(struct TheoraPixelTransform* t)
-{
-	_decodeRGBA(t, t->w * 4, 4, 0, 0, 1, 2, 4, 5, 6, 3, 7);
-// This is the old 2-phase version, leaving it here in case more debugging is needed
-//	_decodeRGB(t, t->w * 4, 4, 0, 0, 1, 2, 4, 5, 6);
-//	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeRGBX(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 4, 4, 0, 0, 1, 2, 4, 5, 6);
-}
-
-void decodeARGB(struct TheoraPixelTransform* t)
-{
-	_decodeRGBA(t, t->w * 4, 4, 0, 1, 2, 3, 5, 6, 7, 0, 4);
-// This is the old 2-phase version, leaving it here in case more debugging is needed
-//	_decodeRGB(t, t->w * 4, 4, 0, 1, 2, 3, 5, 6, 7);
-//	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXRGB(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 4, 4, 0, 1, 2, 3, 5, 6, 7);
-}
-
-void decodeBGR(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 3, 3, 0, 2, 1, 0, 5, 4, 3);
-}
-
-void decodeBGRA(struct TheoraPixelTransform* t)
-{
-	_decodeRGBA(t, t->w * 4, 4, 0, 2, 1, 0, 6, 5, 4, 3, 7);
-// This is the old 2-phase version, leaving it here in case more debugging is needed
-//	_decodeRGB(t, t->w * 4, 4, 0, 2, 1, 0, 6, 5, 4);
-//	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeBGRX(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 4, 4, 0, 2, 1, 0, 6, 5, 4);
-}
-
-void decodeABGR(struct TheoraPixelTransform* t)
-{
-	_decodeRGBA(t, t->w * 4, 4, 0, 3, 2, 1, 7, 6, 5, 0, 4);
-// This is the old 2-phase version, leaving it here in case more debugging is needed
-//	_decodeRGB(t, t->w * 4, 4, 0, 3, 2, 1, 7, 6, 5);
-//	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXBGR(struct TheoraPixelTransform* t)
-{
-	_decodeRGB(t, t->w * 4, 4, 0, 3, 2, 1, 7, 6, 5);
-}
-
-void initYUVConversionModule()
-{
-	//used to bring the table into the high side (scale up) so we
-	//can maintain high precision and not use floats (FIXED POINT)
-	
-	// this is the pseudocode for yuv->rgb conversion
-	//        r = 1.164*(*ySrc - 16) + 1.596*(cv - 128);
-	//        b = 1.164*(*ySrc - 16)                   + 2.018*(cu - 128);
-	//        g = 1.164*(*ySrc - 16) - 0.813*(cv - 128) - 0.391*(cu - 128);
-	
-    double scale = 1L << 13, temp;
-	
-	int i;
-	for (i = 0; i < 256; ++i)
-	{
-		temp = i - 128;
-		
-		YTable[i]  = (int)((1.164 * scale + 0.5) * (i - 16));	//Calc Y component
-		RVTable[i] = (int)((1.596 * scale + 0.5) * temp);		//Calc R component
-		GUTable[i] = (int)((0.391 * scale + 0.5) * temp);		//Calc G u & v components
-		GVTable[i] = (int)((0.813 * scale + 0.5) * temp);
-		BUTable[i] = (int)((2.018 * scale + 0.5) * temp);		//Calc B component
-	}
-}
-
-/*
- * Below are the function versions of the above macros, use those for debugging, but leave the macros for maximum CPU execution speed
- *
- *
- *
- *
-
-void _decodeRGB(struct TheoraPixelTransform* t, int stride, int nBytes, int maxWidth, int i1, int i2, int i3, int j1, int j2, int j3)
-{
-	register int tmp;
-	int nBytes2 = nBytes * 2, cv, cu, rgbY1, rgbY2, rgbY3, rgbY4, rV, gUV, bU, width = maxWidth == 0 ? t->w : maxWidth;
-	unsigned int y;
-	unsigned char *ySrcEven, *ySrcOdd, *yLineEnd, *uSrc, *vSrc, *out1, *out2;
-	
-	for (y = 0; y < t->h; y += 2)
-	{
-		ySrcEven = t->y + y * t->yStride;
-		ySrcOdd  = t->y + (y + 1) * t->yStride;
-		uSrc = t->u + y * t->uStride / 2;
-		vSrc = t->v + y * t->vStride / 2;
-		out1 = t->out + y * stride;
-		out2 = t->out + (y + 1) * stride;
-		
-		for (yLineEnd = ySrcEven + width; ySrcEven != yLineEnd;)
-		{
-			cu = *uSrc; ++uSrc;
-			cv = *vSrc; ++vSrc;
-			rV   = RVTable[cv];
-			gUV  = GUTable[cu] + GVTable[cv];
-			bU   = BUTable[cu];
-			
-			rgbY1 = YTable[*ySrcEven]; ++ySrcEven;
-			rgbY2 = YTable[*ySrcOdd];  ++ySrcOdd;
-			rgbY3 = YTable[*ySrcEven]; ++ySrcEven;
-			rgbY4 = YTable[*ySrcOdd];  ++ySrcOdd;
-			
-			CLIP_RGB_COLOR(out1[i1], rgbY1 + rV );
-			CLIP_RGB_COLOR(out1[i2], rgbY1 - gUV);
-			CLIP_RGB_COLOR(out1[i3], rgbY1 + bU );
-			
-			CLIP_RGB_COLOR(out2[i1], rgbY2 + rV );
-			CLIP_RGB_COLOR(out2[i2], rgbY2 - gUV);
-			CLIP_RGB_COLOR(out2[i3], rgbY2 + bU );
-			
-			CLIP_RGB_COLOR(out1[j1], rgbY3 + rV );
-			CLIP_RGB_COLOR(out1[j2], rgbY3 - gUV);
-			CLIP_RGB_COLOR(out1[j3], rgbY3 + bU );
-			
-			CLIP_RGB_COLOR(out2[j1], rgbY4 + rV );
-			CLIP_RGB_COLOR(out2[j2], rgbY4 - gUV);
-			CLIP_RGB_COLOR(out2[j3], rgbY4 + bU );
-			
-			out1 += nBytes2;  out2 += nBytes2;
-		}
-	}
-}
- 
-void _decodeRGBA(struct TheoraPixelTransform* t, int stride, int nBytes, int maxWidth, int i1, int i2, int i3, int j1, int j2, int j3, int aindex1, int aindex2)
-{
-	register int tmp;
-	int nBytes2 = nBytes * 2, cv, cu, rgbY1, rgbY2, rgbY3, rgbY4, a1, a2, a3, a4, rV, gUV, bU, width = maxWidth == 0 ? t->w : maxWidth;
-	int alphaStride = t->w;
-	unsigned int y;
-	unsigned char *ySrcEven, *ySrcOdd, *yLineEnd, *uSrc, *vSrc, *out1, *out2;
-	
-	for (y = 0; y < t->h; y += 2)
-	{
-		ySrcEven = t->y + y * t->yStride;
-		ySrcOdd  = t->y + (y + 1) * t->yStride;
-		uSrc = t->u + y * t->uStride / 2;
-		vSrc = t->v + y * t->vStride / 2;
-		out1 = t->out + y * stride;
-		out2 = t->out + (y + 1) * stride;
-		
-		for (yLineEnd = ySrcEven + width; ySrcEven != yLineEnd;)
-		{
-			cu = *uSrc; ++uSrc;
-			cv = *vSrc; ++vSrc;
-			rV   = RVTable[cv];
-			gUV  = GUTable[cu] + GVTable[cv];
-			bU   = BUTable[cu];
-			
-			rgbY1 = YTable[*ySrcEven]; a1 = ySrcEven[alphaStride]; ++ySrcEven;
-			rgbY2 = YTable[*ySrcOdd];  a2 = ySrcOdd [alphaStride];  ++ySrcOdd;
-			rgbY3 = YTable[*ySrcEven]; a3 = ySrcEven[alphaStride]; ++ySrcEven;
-			rgbY4 = YTable[*ySrcOdd];  a4 = ySrcOdd [alphaStride];  ++ySrcOdd;
-			
-			if (a1 >= 32)
-			{
-				CLIP_RGB_COLOR(out1[i1], rgbY1 + rV );
-				CLIP_RGB_COLOR(out1[i2], rgbY1 - gUV);
-				CLIP_RGB_COLOR(out1[i3], rgbY1 + bU );
-				out1[aindex1] = a1 > 224 ? 255 : a1;
-			}
-			else *((unsigned int*) out1) = 0;
-			
-			if (a2 >= 32)
-			{
-				CLIP_RGB_COLOR(out2[i1], rgbY2 + rV );
-				CLIP_RGB_COLOR(out2[i2], rgbY2 - gUV);
-				CLIP_RGB_COLOR(out2[i3], rgbY2 + bU );
-				out2[aindex1] = a2 > 224 ? 255 : a2;
-			}
-			else *((unsigned int*) out2) = 0;
-
-			
-			if (a3 >= 32)
-			{
-				CLIP_RGB_COLOR(out1[j1], rgbY3 + rV );
-				CLIP_RGB_COLOR(out1[j2], rgbY3 - gUV);
-				CLIP_RGB_COLOR(out1[j3], rgbY3 + bU );
-				out1[aindex2] = a3 > 224 ? 255 : a3;
-			}
-			else *((unsigned int*) &out1[4]) = 0;
-
-			if (a4 >= 32)
-			{
-				CLIP_RGB_COLOR(out2[j1], rgbY4 + rV );
-				CLIP_RGB_COLOR(out2[j2], rgbY4 - gUV);
-				CLIP_RGB_COLOR(out2[j3], rgbY4 + bU );
-				out2[aindex2] = a4 > 224 ? 255 : a4;
-			}
-			else *((unsigned int*) &out2[4]) = 0;
-
-			out1 += nBytes2;  out2 += nBytes2;
-		}
-	}
-}
-*/
-#endif
diff --git a/drivers/theoraplayer/src/YUV/C/yuv420_yuv_c.c b/drivers/theoraplayer/src/YUV/C/yuv420_yuv_c.c
deleted file mode 100644
index fea74eca71..0000000000
--- a/drivers/theoraplayer/src/YUV/C/yuv420_yuv_c.c
+++ /dev/null
@@ -1,86 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "yuv_util.h"
-
-static void _decodeYUV(struct TheoraPixelTransform* t, int stride, int nBytes, int maxWidth)
-{
-	int cv, cu, y1, y2, y3, y4, width = maxWidth == 0 ? t->w : maxWidth;
-	unsigned char *ySrcEven, *ySrcOdd, *yLineEnd, *uSrc, *vSrc, *out1, *out2;
-	unsigned int y;
-
-	for (y=0; y < t->h; y += 2)
-	{
-		ySrcEven = t->y + y * t->yStride;
-		ySrcOdd  = t->y + (y + 1) * t->yStride;
-		uSrc = t->u + y * t->uStride / 2;
-		vSrc = t->v + y * t->vStride / 2;
-		out1 = t->out + y * stride;
-		out2 = t->out + (y + 1) * stride;
-		
-		for (yLineEnd = ySrcEven + width; ySrcEven != yLineEnd;)
-		{
-			// EVEN columns
-			cu = *uSrc; ++uSrc;
-			cv = *vSrc; ++vSrc;
-			
-			y1 = *ySrcEven; ++ySrcEven;
-			y2 = *ySrcOdd;  ++ySrcOdd;
-			y3 = *ySrcEven; ++ySrcEven;
-			y4 = *ySrcOdd;  ++ySrcOdd;
-			
-			// EVEN columns
-			out1[0] = y1;
-			out1[1] = cu;
-			out1[2] = cv;
-			
-			out2[0] = y2;
-			out2[1] = cu;
-			out2[2] = cv;
-			
-			out1 += nBytes;  out2 += nBytes;
-			// ODD columns
-			out1[0] = y3;
-			out1[1] = cu;
-			out1[2] = cv;
-			
-			out2[0] = y4;
-			out2[1] = cu;
-			out2[2] = cv;
-			out1 += nBytes;  out2 += nBytes;
-		}
-	}
-}
-
-void decodeYUV(struct TheoraPixelTransform* t)
-{
-	_decodeYUV(t, t->w * 3, 3, 0);
-}
-
-void decodeYUVA(struct TheoraPixelTransform* t)
-{
-	_decodeYUV(t, t->w * 4, 4, 0);
-	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeYUVX(struct TheoraPixelTransform* t)
-{
-	_decodeYUV(t, t->w * 4, 4, 0);
-}
-
-void decodeAYUV(struct TheoraPixelTransform* t)
-{
-	_decodeYUV(incOut(t, 1), t->w * 4, 4, 0);
-	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXYUV(struct TheoraPixelTransform* t)
-{
-	_decodeYUV(incOut(t, 1), t->w * 4, 4, 0);
-}
-
diff --git a/drivers/theoraplayer/src/YUV/android/cpu-features.c b/drivers/theoraplayer/src/YUV/android/cpu-features.c
deleted file mode 100644
index 623dc94e0e..0000000000
--- a/drivers/theoraplayer/src/YUV/android/cpu-features.c
+++ /dev/null
@@ -1,1095 +0,0 @@
-/*
- * Copyright (C) 2010 The Android Open Source Project
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- *  * Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in
- *    the documentation and/or other materials provided with the
- *    distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
- * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
- * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
- * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
- * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
- * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
- * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
- * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-
-/* ChangeLog for this library:
- *
- * NDK r8d: Add android_setCpu().
- *
- * NDK r8c: Add new ARM CPU features: VFPv2, VFP_D32, VFP_FP16,
- *          VFP_FMA, NEON_FMA, IDIV_ARM, IDIV_THUMB2 and iWMMXt.
- *
- *          Rewrite the code to parse /proc/self/auxv instead of
- *          the "Features" field in /proc/cpuinfo.
- *
- *          Dynamically allocate the buffer that hold the content
- *          of /proc/cpuinfo to deal with newer hardware.
- *
- * NDK r7c: Fix CPU count computation. The old method only reported the
- *           number of _active_ CPUs when the library was initialized,
- *           which could be less than the real total.
- *
- * NDK r5: Handle buggy kernels which report a CPU Architecture number of 7
- *         for an ARMv6 CPU (see below).
- *
- *         Handle kernels that only report 'neon', and not 'vfpv3'
- *         (VFPv3 is mandated by the ARM architecture is Neon is implemented)
- *
- *         Handle kernels that only report 'vfpv3d16', and not 'vfpv3'
- *
- *         Fix x86 compilation. Report ANDROID_CPU_FAMILY_X86 in
- *         android_getCpuFamily().
- *
- * NDK r4: Initial release
- */
-
-#if 0
-
-#ifdef _ANDROID
-#if defined(__le32__)
-
-// When users enter this, we should only provide interface and
-// libportable will give the implementations.
-
-#else // !__le32__
-
-#include <sys/system_properties.h>
-#include <pthread.h>
-#include "cpu-features.h"
-#include <stdio.h>
-#include <stdlib.h>
-#include <fcntl.h>
-#include <errno.h>
-
-static  pthread_once_t     g_once;
-static  int                g_inited;
-static  AndroidCpuFamily   g_cpuFamily;
-static  uint64_t           g_cpuFeatures;
-static  int                g_cpuCount;
-
-#ifdef __arm__
-static  uint32_t           g_cpuIdArm;
-#endif
-
-static const int  android_cpufeatures_debug = 0;
-
-#ifdef __arm__
-#  define DEFAULT_CPU_FAMILY  ANDROID_CPU_FAMILY_ARM
-#elif defined __i386__
-#  define DEFAULT_CPU_FAMILY  ANDROID_CPU_FAMILY_X86
-#else
-#  define DEFAULT_CPU_FAMILY  ANDROID_CPU_FAMILY_UNKNOWN
-#endif
-
-#define  D(...) \
-    do { \
-        if (android_cpufeatures_debug) { \
-            printf(__VA_ARGS__); fflush(stdout); \
-        } \
-    } while (0)
-
-#ifdef __i386__
-static __inline__ void x86_cpuid(int func, int values[4])
-{
-    int a, b, c, d;
-    /* We need to preserve ebx since we're compiling PIC code */
-    /* this means we can't use "=b" for the second output register */
-    __asm__ __volatile__ ( \
-      "push %%ebx\n"
-      "cpuid\n" \
-      "mov %%ebx, %1\n"
-      "pop %%ebx\n"
-      : "=a" (a), "=r" (b), "=c" (c), "=d" (d) \
-      : "a" (func) \
-    );
-    values[0] = a;
-    values[1] = b;
-    values[2] = c;
-    values[3] = d;
-}
-#endif
-
-/* Get the size of a file by reading it until the end. This is needed
- * because files under /proc do not always return a valid size when
- * using fseek(0, SEEK_END) + ftell(). Nor can they be mmap()-ed.
- */
-static int
-get_file_size(const char* pathname)
-{
-    int fd, ret, result = 0;
-    char buffer[256];
-
-    fd = open(pathname, O_RDONLY);
-    if (fd < 0) {
-        D("Can't open %s: %s\n", pathname, strerror(errno));
-        return -1;
-    }
-
-    for (;;) {
-        int ret = read(fd, buffer, sizeof buffer);
-        if (ret < 0) {
-            if (errno == EINTR)
-                continue;
-            D("Error while reading %s: %s\n", pathname, strerror(errno));
-            break;
-        }
-        if (ret == 0)
-            break;
-
-        result += ret;
-    }
-    close(fd);
-    return result;
-}
-
-/* Read the content of /proc/cpuinfo into a user-provided buffer.
- * Return the length of the data, or -1 on error. Does *not*
- * zero-terminate the content. Will not read more
- * than 'buffsize' bytes.
- */
-static int
-read_file(const char*  pathname, char*  buffer, size_t  buffsize)
-{
-    int  fd, count;
-
-    fd = open(pathname, O_RDONLY);
-    if (fd < 0) {
-        D("Could not open %s: %s\n", pathname, strerror(errno));
-        return -1;
-    }
-    count = 0;
-    while (count < (int)buffsize) {
-        int ret = read(fd, buffer + count, buffsize - count);
-        if (ret < 0) {
-            if (errno == EINTR)
-                continue;
-            D("Error while reading from %s: %s\n", pathname, strerror(errno));
-            if (count == 0)
-                count = -1;
-            break;
-        }
-        if (ret == 0)
-            break;
-        count += ret;
-    }
-    close(fd);
-    return count;
-}
-
-/* Extract the content of a the first occurence of a given field in
- * the content of /proc/cpuinfo and return it as a heap-allocated
- * string that must be freed by the caller.
- *
- * Return NULL if not found
- */
-static char*
-extract_cpuinfo_field(const char* buffer, int buflen, const char* field)
-{
-    int  fieldlen = strlen(field);
-    const char* bufend = buffer + buflen;
-    char* result = NULL;
-    int len, ignore;
-    const char *p, *q;
-
-    /* Look for first field occurence, and ensures it starts the line. */
-    p = buffer;
-    for (;;) {
-        p = memmem(p, bufend-p, field, fieldlen);
-        if (p == NULL)
-            goto EXIT;
-
-        if (p == buffer || p[-1] == '\n')
-            break;
-
-        p += fieldlen;
-    }
-
-    /* Skip to the first column followed by a space */
-    p += fieldlen;
-    p  = memchr(p, ':', bufend-p);
-    if (p == NULL || p[1] != ' ')
-        goto EXIT;
-
-    /* Find the end of the line */
-    p += 2;
-    q = memchr(p, '\n', bufend-p);
-    if (q == NULL)
-        q = bufend;
-
-    /* Copy the line into a heap-allocated buffer */
-    len = q-p;
-    result = malloc(len+1);
-    if (result == NULL)
-        goto EXIT;
-
-    memcpy(result, p, len);
-    result[len] = '\0';
-
-EXIT:
-    return result;
-}
-
-/* Checks that a space-separated list of items contains one given 'item'.
- * Returns 1 if found, 0 otherwise.
- */
-static int
-has_list_item(const char* list, const char* item)
-{
-    const char*  p = list;
-    int itemlen = strlen(item);
-
-    if (list == NULL)
-        return 0;
-
-    while (*p) {
-        const char*  q;
-
-        /* skip spaces */
-        while (*p == ' ' || *p == '\t')
-            p++;
-
-        /* find end of current list item */
-        q = p;
-        while (*q && *q != ' ' && *q != '\t')
-            q++;
-
-        if (itemlen == q-p && !memcmp(p, item, itemlen))
-            return 1;
-
-        /* skip to next item */
-        p = q;
-    }
-    return 0;
-}
-
-/* Parse a number starting from 'input', but not going further
- * than 'limit'. Return the value into '*result'.
- *
- * NOTE: Does not skip over leading spaces, or deal with sign characters.
- * NOTE: Ignores overflows.
- *
- * The function returns NULL in case of error (bad format), or the new
- * position after the decimal number in case of success (which will always
- * be <= 'limit').
- */
-static const char*
-parse_number(const char* input, const char* limit, int base, int* result)
-{
-    const char* p = input;
-    int val = 0;
-    while (p < limit) {
-        int d = (*p - '0');
-        if ((unsigned)d >= 10U) {
-            d = (*p - 'a');
-            if ((unsigned)d >= 6U)
-              d = (*p - 'A');
-            if ((unsigned)d >= 6U)
-              break;
-            d += 10;
-        }
-        if (d >= base)
-          break;
-        val = val*base + d;
-        p++;
-    }
-    if (p == input)
-        return NULL;
-
-    *result = val;
-    return p;
-}
-
-static const char*
-parse_decimal(const char* input, const char* limit, int* result)
-{
-    return parse_number(input, limit, 10, result);
-}
-
-static const char*
-parse_hexadecimal(const char* input, const char* limit, int* result)
-{
-    return parse_number(input, limit, 16, result);
-}
-
-/* This small data type is used to represent a CPU list / mask, as read
- * from sysfs on Linux. See http://www.kernel.org/doc/Documentation/cputopology.txt
- *
- * For now, we don't expect more than 32 cores on mobile devices, so keep
- * everything simple.
- */
-typedef struct {
-    uint32_t mask;
-} CpuList;
-
-static __inline__ void
-cpulist_init(CpuList* list) {
-    list->mask = 0;
-}
-
-static __inline__ void
-cpulist_and(CpuList* list1, CpuList* list2) {
-    list1->mask &= list2->mask;
-}
-
-static __inline__ void
-cpulist_set(CpuList* list, int index) {
-    if ((unsigned)index < 32) {
-        list->mask |= (uint32_t)(1U << index);
-    }
-}
-
-static __inline__ int
-cpulist_count(CpuList* list) {
-    return __builtin_popcount(list->mask);
-}
-
-/* Parse a textual list of cpus and store the result inside a CpuList object.
- * Input format is the following:
- * - comma-separated list of items (no spaces)
- * - each item is either a single decimal number (cpu index), or a range made
- *   of two numbers separated by a single dash (-). Ranges are inclusive.
- *
- * Examples:   0
- *             2,4-127,128-143
- *             0-1
- */
-static void
-cpulist_parse(CpuList* list, const char* line, int line_len)
-{
-    const char* p = line;
-    const char* end = p + line_len;
-    const char* q;
-
-    /* NOTE: the input line coming from sysfs typically contains a
-     * trailing newline, so take care of it in the code below
-     */
-    while (p < end && *p != '\n')
-    {
-        int val, start_value, end_value;
-
-        /* Find the end of current item, and put it into 'q' */
-        q = memchr(p, ',', end-p);
-        if (q == NULL) {
-            q = end;
-        }
-
-        /* Get first value */
-        p = parse_decimal(p, q, &start_value);
-        if (p == NULL)
-            goto BAD_FORMAT;
-
-        end_value = start_value;
-
-        /* If we're not at the end of the item, expect a dash and
-         * and integer; extract end value.
-         */
-        if (p < q && *p == '-') {
-            p = parse_decimal(p+1, q, &end_value);
-            if (p == NULL)
-                goto BAD_FORMAT;
-        }
-
-        /* Set bits CPU list bits */
-        for (val = start_value; val <= end_value; val++) {
-            cpulist_set(list, val);
-        }
-
-        /* Jump to next item */
-        p = q;
-        if (p < end)
-            p++;
-    }
-
-BAD_FORMAT:
-    ;
-}
-
-/* Read a CPU list from one sysfs file */
-static void
-cpulist_read_from(CpuList* list, const char* filename)
-{
-    char   file[64];
-    int    filelen;
-
-    cpulist_init(list);
-
-    filelen = read_file(filename, file, sizeof file);
-    if (filelen < 0) {
-        D("Could not read %s: %s\n", filename, strerror(errno));
-        return;
-    }
-
-    cpulist_parse(list, file, filelen);
-}
-
-// See <asm/hwcap.h> kernel header.
-#define HWCAP_VFP       (1 << 6)
-#define HWCAP_IWMMXT    (1 << 9)
-#define HWCAP_NEON      (1 << 12)
-#define HWCAP_VFPv3     (1 << 13)
-#define HWCAP_VFPv3D16  (1 << 14)
-#define HWCAP_VFPv4     (1 << 16)
-#define HWCAP_IDIVA     (1 << 17)
-#define HWCAP_IDIVT     (1 << 18)
-
-#define AT_HWCAP 16
-
-#if defined(__arm__)
-/* Compute the ELF HWCAP flags.
- */
-static uint32_t
-get_elf_hwcap(const char* cpuinfo, int cpuinfo_len)
-{
-  /* IMPORTANT:
-   *   Accessing /proc/self/auxv doesn't work anymore on all
-   *   platform versions. More specifically, when running inside
-   *   a regular application process, most of /proc/self/ will be
-   *   non-readable, including /proc/self/auxv. This doesn't
-   *   happen however if the application is debuggable, or when
-   *   running under the "shell" UID, which is why this was not
-   *   detected appropriately.
-   */
-#if 0
-    uint32_t result = 0;
-    const char filepath[] = "/proc/self/auxv";
-    int fd = open(filepath, O_RDONLY);
-    if (fd < 0) {
-        D("Could not open %s: %s\n", filepath, strerror(errno));
-        return 0;
-    }
-
-    struct { uint32_t tag; uint32_t value; } entry;
-
-    for (;;) {
-        int ret = read(fd, (char*)&entry, sizeof entry);
-        if (ret < 0) {
-            if (errno == EINTR)
-                continue;
-            D("Error while reading %s: %s\n", filepath, strerror(errno));
-            break;
-        }
-        // Detect end of list.
-        if (ret == 0 || (entry.tag == 0 && entry.value == 0))
-          break;
-        if (entry.tag == AT_HWCAP) {
-          result = entry.value;
-          break;
-        }
-    }
-    close(fd);
-    return result;
-#else
-    // Recreate ELF hwcaps by parsing /proc/cpuinfo Features tag.
-    uint32_t hwcaps = 0;
-
-    char* cpuFeatures = extract_cpuinfo_field(cpuinfo, cpuinfo_len, "Features");
-
-    if (cpuFeatures != NULL) {
-        D("Found cpuFeatures = '%s'\n", cpuFeatures);
-
-        if (has_list_item(cpuFeatures, "vfp"))
-            hwcaps |= HWCAP_VFP;
-        if (has_list_item(cpuFeatures, "vfpv3"))
-            hwcaps |= HWCAP_VFPv3;
-        if (has_list_item(cpuFeatures, "vfpv3d16"))
-            hwcaps |= HWCAP_VFPv3D16;
-        if (has_list_item(cpuFeatures, "vfpv4"))
-            hwcaps |= HWCAP_VFPv4;
-        if (has_list_item(cpuFeatures, "neon"))
-            hwcaps |= HWCAP_NEON;
-        if (has_list_item(cpuFeatures, "idiva"))
-            hwcaps |= HWCAP_IDIVA;
-        if (has_list_item(cpuFeatures, "idivt"))
-            hwcaps |= HWCAP_IDIVT;
-        if (has_list_item(cpuFeatures, "idiv"))
-            hwcaps |= HWCAP_IDIVA | HWCAP_IDIVT;
-        if (has_list_item(cpuFeatures, "iwmmxt"))
-            hwcaps |= HWCAP_IWMMXT;
-
-        free(cpuFeatures);
-    }
-    return hwcaps;
-#endif
-}
-#endif  /* __arm__ */
-
-/* Return the number of cpus present on a given device.
- *
- * To handle all weird kernel configurations, we need to compute the
- * intersection of the 'present' and 'possible' CPU lists and count
- * the result.
- */
-static int
-get_cpu_count(void)
-{
-    CpuList cpus_present[1];
-    CpuList cpus_possible[1];
-
-    cpulist_read_from(cpus_present, "/sys/devices/system/cpu/present");
-    cpulist_read_from(cpus_possible, "/sys/devices/system/cpu/possible");
-
-    /* Compute the intersection of both sets to get the actual number of
-     * CPU cores that can be used on this device by the kernel.
-     */
-    cpulist_and(cpus_present, cpus_possible);
-
-    return cpulist_count(cpus_present);
-}
-
-static void
-android_cpuInitFamily(void)
-{
-#if defined(__arm__)
-    g_cpuFamily = ANDROID_CPU_FAMILY_ARM;
-#elif defined(__i386__)
-    g_cpuFamily = ANDROID_CPU_FAMILY_X86;
-#elif defined(__mips64)
-/* Needs to be before __mips__ since the compiler defines both */
-    g_cpuFamily = ANDROID_CPU_FAMILY_MIPS64;
-#elif defined(__mips__)
-    g_cpuFamily = ANDROID_CPU_FAMILY_MIPS;
-#elif defined(__aarch64__)
-    g_cpuFamily = ANDROID_CPU_FAMILY_ARM64;
-#elif defined(__x86_64__)
-    g_cpuFamily = ANDROID_CPU_FAMILY_X86_64;
-#else
-    g_cpuFamily = ANDROID_CPU_FAMILY_UNKNOWN;
-#endif
-}
-
-static void
-android_cpuInit(void)
-{
-    char* cpuinfo = NULL;
-    int   cpuinfo_len;
-
-    android_cpuInitFamily();
-
-    g_cpuFeatures = 0;
-    g_cpuCount    = 1;
-    g_inited      = 1;
-
-    cpuinfo_len = get_file_size("/proc/cpuinfo");
-    if (cpuinfo_len < 0) {
-      D("cpuinfo_len cannot be computed!");
-      return;
-    }
-    cpuinfo = malloc(cpuinfo_len);
-    if (cpuinfo == NULL) {
-      D("cpuinfo buffer could not be allocated");
-      return;
-    }
-    cpuinfo_len = read_file("/proc/cpuinfo", cpuinfo, cpuinfo_len);
-    D("cpuinfo_len is (%d):\n%.*s\n", cpuinfo_len,
-      cpuinfo_len >= 0 ? cpuinfo_len : 0, cpuinfo);
-
-    if (cpuinfo_len < 0)  /* should not happen */ {
-        free(cpuinfo);
-        return;
-    }
-
-    /* Count the CPU cores, the value may be 0 for single-core CPUs */
-    g_cpuCount = get_cpu_count();
-    if (g_cpuCount == 0) {
-        g_cpuCount = 1;
-    }
-
-    D("found cpuCount = %d\n", g_cpuCount);
-
-#ifdef __arm__
-    {
-        char*  features = NULL;
-        char*  architecture = NULL;
-
-        /* Extract architecture from the "CPU Architecture" field.
-         * The list is well-known, unlike the the output of
-         * the 'Processor' field which can vary greatly.
-         *
-         * See the definition of the 'proc_arch' array in
-         * $KERNEL/arch/arm/kernel/setup.c and the 'c_show' function in
-         * same file.
-         */
-        char* cpuArch = extract_cpuinfo_field(cpuinfo, cpuinfo_len, "CPU architecture");
-
-        if (cpuArch != NULL) {
-            char*  end;
-            long   archNumber;
-            int    hasARMv7 = 0;
-
-            D("found cpuArch = '%s'\n", cpuArch);
-
-            /* read the initial decimal number, ignore the rest */
-            archNumber = strtol(cpuArch, &end, 10);
-
-            /* Here we assume that ARMv8 will be upwards compatible with v7
-             * in the future. Unfortunately, there is no 'Features' field to
-             * indicate that Thumb-2 is supported.
-             */
-            if (end > cpuArch && archNumber >= 7) {
-                hasARMv7 = 1;
-            }
-
-            /* Unfortunately, it seems that certain ARMv6-based CPUs
-             * report an incorrect architecture number of 7!
-             *
-             * See http://code.google.com/p/android/issues/detail?id=10812
-             *
-             * We try to correct this by looking at the 'elf_format'
-             * field reported by the 'Processor' field, which is of the
-             * form of "(v7l)" for an ARMv7-based CPU, and "(v6l)" for
-             * an ARMv6-one.
-             */
-            if (hasARMv7) {
-                char* cpuProc = extract_cpuinfo_field(cpuinfo, cpuinfo_len,
-                                                      "Processor");
-                if (cpuProc != NULL) {
-                    D("found cpuProc = '%s'\n", cpuProc);
-                    if (has_list_item(cpuProc, "(v6l)")) {
-                        D("CPU processor and architecture mismatch!!\n");
-                        hasARMv7 = 0;
-                    }
-                    free(cpuProc);
-                }
-            }
-
-            if (hasARMv7) {
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_ARMv7;
-            }
-
-            /* The LDREX / STREX instructions are available from ARMv6 */
-            if (archNumber >= 6) {
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_LDREX_STREX;
-            }
-
-            free(cpuArch);
-        }
-
-        /* Extract the list of CPU features from ELF hwcaps */
-        uint32_t hwcaps = get_elf_hwcap(cpuinfo, cpuinfo_len);
-
-        if (hwcaps != 0) {
-            int has_vfp = (hwcaps & HWCAP_VFP);
-            int has_vfpv3 = (hwcaps & HWCAP_VFPv3);
-            int has_vfpv3d16 = (hwcaps & HWCAP_VFPv3D16);
-            int has_vfpv4 = (hwcaps & HWCAP_VFPv4);
-            int has_neon = (hwcaps & HWCAP_NEON);
-            int has_idiva = (hwcaps & HWCAP_IDIVA);
-            int has_idivt = (hwcaps & HWCAP_IDIVT);
-            int has_iwmmxt = (hwcaps & HWCAP_IWMMXT);
-
-            // The kernel does a poor job at ensuring consistency when
-            // describing CPU features. So lots of guessing is needed.
-
-            // 'vfpv4' implies VFPv3|VFP_FMA|FP16
-            if (has_vfpv4)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv3    |
-                                 ANDROID_CPU_ARM_FEATURE_VFP_FP16 |
-                                 ANDROID_CPU_ARM_FEATURE_VFP_FMA;
-
-            // 'vfpv3' or 'vfpv3d16' imply VFPv3. Note that unlike GCC,
-            // a value of 'vfpv3' doesn't necessarily mean that the D32
-            // feature is present, so be conservative. All CPUs in the
-            // field that support D32 also support NEON, so this should
-            // not be a problem in practice.
-            if (has_vfpv3 || has_vfpv3d16)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv3;
-
-            // 'vfp' is super ambiguous. Depending on the kernel, it can
-            // either mean VFPv2 or VFPv3. Make it depend on ARMv7.
-            if (has_vfp) {
-              if (g_cpuFeatures & ANDROID_CPU_ARM_FEATURE_ARMv7)
-                  g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv3;
-              else
-                  g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv2;
-            }
-
-            // Neon implies VFPv3|D32, and if vfpv4 is detected, NEON_FMA
-            if (has_neon) {
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv3 |
-                                 ANDROID_CPU_ARM_FEATURE_NEON |
-                                 ANDROID_CPU_ARM_FEATURE_VFP_D32;
-              if (has_vfpv4)
-                  g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_NEON_FMA;
-            }
-
-            // VFPv3 implies VFPv2 and ARMv7
-            if (g_cpuFeatures & ANDROID_CPU_ARM_FEATURE_VFPv3)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_VFPv2 |
-                                 ANDROID_CPU_ARM_FEATURE_ARMv7;
-
-            if (has_idiva)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_IDIV_ARM;
-            if (has_idivt)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_IDIV_THUMB2;
-
-            if (has_iwmmxt)
-                g_cpuFeatures |= ANDROID_CPU_ARM_FEATURE_iWMMXt;
-        }
-
-        /* Extract the cpuid value from various fields */
-        // The CPUID value is broken up in several entries in /proc/cpuinfo.
-        // This table is used to rebuild it from the entries.
-        static const struct CpuIdEntry {
-            const char* field;
-            char        format;
-            char        bit_lshift;
-            char        bit_length;
-        } cpu_id_entries[] = {
-            { "CPU implementer", 'x', 24, 8 },
-            { "CPU variant", 'x', 20, 4 },
-            { "CPU part", 'x', 4, 12 },
-            { "CPU revision", 'd', 0, 4 },
-        };
-        size_t i;
-        D("Parsing /proc/cpuinfo to recover CPUID\n");
-        for (i = 0;
-             i < sizeof(cpu_id_entries)/sizeof(cpu_id_entries[0]);
-             ++i) {
-            const struct CpuIdEntry* entry = &cpu_id_entries[i];
-            char* value = extract_cpuinfo_field(cpuinfo,
-                                                cpuinfo_len,
-                                                entry->field);
-            if (value == NULL)
-                continue;
-
-            D("field=%s value='%s'\n", entry->field, value);
-            char* value_end = value + strlen(value);
-            int val = 0;
-            const char* start = value;
-            const char* p;
-            if (value[0] == '0' && (value[1] == 'x' || value[1] == 'X')) {
-              start += 2;
-              p = parse_hexadecimal(start, value_end, &val);
-            } else if (entry->format == 'x')
-              p = parse_hexadecimal(value, value_end, &val);
-            else
-              p = parse_decimal(value, value_end, &val);
-
-            if (p > (const char*)start) {
-              val &= ((1 << entry->bit_length)-1);
-              val <<= entry->bit_lshift;
-              g_cpuIdArm |= (uint32_t) val;
-            }
-
-            free(value);
-        }
-
-        // Handle kernel configuration bugs that prevent the correct
-        // reporting of CPU features.
-        static const struct CpuFix {
-            uint32_t  cpuid;
-            uint64_t  or_flags;
-        } cpu_fixes[] = {
-            /* The Nexus 4 (Qualcomm Krait) kernel configuration
-             * forgets to report IDIV support. */
-            { 0x510006f2, ANDROID_CPU_ARM_FEATURE_IDIV_ARM |
-                          ANDROID_CPU_ARM_FEATURE_IDIV_THUMB2 },
-            { 0x510006f3, ANDROID_CPU_ARM_FEATURE_IDIV_ARM |
-                          ANDROID_CPU_ARM_FEATURE_IDIV_THUMB2 },
-        };
-        size_t n;
-        for (n = 0; n < sizeof(cpu_fixes)/sizeof(cpu_fixes[0]); ++n) {
-            const struct CpuFix* entry = &cpu_fixes[n];
-
-            if (g_cpuIdArm == entry->cpuid)
-                g_cpuFeatures |= entry->or_flags;
-        }
-
-    }
-#endif /* __arm__ */
-
-#ifdef __i386__
-    int regs[4];
-
-/* According to http://en.wikipedia.org/wiki/CPUID */
-#define VENDOR_INTEL_b  0x756e6547
-#define VENDOR_INTEL_c  0x6c65746e
-#define VENDOR_INTEL_d  0x49656e69
-
-    x86_cpuid(0, regs);
-    int vendorIsIntel = (regs[1] == VENDOR_INTEL_b &&
-                         regs[2] == VENDOR_INTEL_c &&
-                         regs[3] == VENDOR_INTEL_d);
-
-    x86_cpuid(1, regs);
-    if ((regs[2] & (1 << 9)) != 0) {
-        g_cpuFeatures |= ANDROID_CPU_X86_FEATURE_SSSE3;
-    }
-    if ((regs[2] & (1 << 23)) != 0) {
-        g_cpuFeatures |= ANDROID_CPU_X86_FEATURE_POPCNT;
-    }
-    if (vendorIsIntel && (regs[2] & (1 << 22)) != 0) {
-        g_cpuFeatures |= ANDROID_CPU_X86_FEATURE_MOVBE;
-    }
-#endif
-
-    free(cpuinfo);
-}
-
-
-AndroidCpuFamily
-android_getCpuFamily(void)
-{
-    pthread_once(&g_once, android_cpuInit);
-    return g_cpuFamily;
-}
-
-
-uint64_t
-android_getCpuFeaturesExt(void)
-{
-    pthread_once(&g_once, android_cpuInit);
-    return g_cpuFeatures;
-}
-
-
-int
-android_getCpuCount(void)
-{
-    pthread_once(&g_once, android_cpuInit);
-    return g_cpuCount;
-}
-
-static void
-android_cpuInitDummy(void)
-{
-    g_inited = 1;
-}
-
-int
-android_setCpu(int cpu_count, uint64_t cpu_features)
-{
-    /* Fail if the library was already initialized. */
-    if (g_inited)
-        return 0;
-
-    android_cpuInitFamily();
-    g_cpuCount = (cpu_count <= 0 ? 1 : cpu_count);
-    g_cpuFeatures = cpu_features;
-    pthread_once(&g_once, android_cpuInitDummy);
-
-    return 1;
-}
-
-#ifdef __arm__
-uint32_t
-android_getCpuIdArm(void)
-{
-    pthread_once(&g_once, android_cpuInit);
-    return g_cpuIdArm;
-}
-
-int
-android_setCpuArm(int cpu_count, uint64_t cpu_features, uint32_t cpu_id)
-{
-    if (!android_setCpu(cpu_count, cpu_features))
-        return 0;
-
-    g_cpuIdArm = cpu_id;
-    return 1;
-}
-#endif  /* __arm__ */
-
-/*
- * Technical note: Making sense of ARM's FPU architecture versions.
- *
- * FPA was ARM's first attempt at an FPU architecture. There is no Android
- * device that actually uses it since this technology was already obsolete
- * when the project started. If you see references to FPA instructions
- * somewhere, you can be sure that this doesn't apply to Android at all.
- *
- * FPA was followed by "VFP", soon renamed "VFPv1" due to the emergence of
- * new versions / additions to it. ARM considers this obsolete right now,
- * and no known Android device implements it either.
- *
- * VFPv2 added a few instructions to VFPv1, and is an *optional* extension
- * supported by some ARMv5TE, ARMv6 and ARMv6T2 CPUs. Note that a device
- * supporting the 'armeabi' ABI doesn't necessarily support these.
- *
- * VFPv3-D16 adds a few instructions on top of VFPv2 and is typically used
- * on ARMv7-A CPUs which implement a FPU. Note that it is also mandated
- * by the Android 'armeabi-v7a' ABI. The -D16 suffix in its name means
- * that it provides 16 double-precision FPU registers (d0-d15) and 32
- * single-precision ones (s0-s31) which happen to be mapped to the same
- * register banks.
- *
- * VFPv3-D32 is the name of an extension to VFPv3-D16 that provides 16
- * additional double precision registers (d16-d31). Note that there are
- * still only 32 single precision registers.
- *
- * VFPv3xD is a *subset* of VFPv3-D16 that only provides single-precision
- * registers. It is only used on ARMv7-M (i.e. on micro-controllers) which
- * are not supported by Android. Note that it is not compatible with VFPv2.
- *
- * NOTE: The term 'VFPv3' usually designate either VFPv3-D16 or VFPv3-D32
- *       depending on context. For example GCC uses it for VFPv3-D32, but
- *       the Linux kernel code uses it for VFPv3-D16 (especially in
- *       /proc/cpuinfo). Always try to use the full designation when
- *       possible.
- *
- * NEON, a.k.a. "ARM Advanced SIMD" is an extension that provides
- * instructions to perform parallel computations on vectors of 8, 16,
- * 32, 64 and 128 bit quantities. NEON requires VFPv32-D32 since all
- * NEON registers are also mapped to the same register banks.
- *
- * VFPv4-D16, adds a few instructions on top of VFPv3-D16 in order to
- * perform fused multiply-accumulate on VFP registers, as well as
- * half-precision (16-bit) conversion operations.
- *
- * VFPv4-D32 is VFPv4-D16 with 32, instead of 16, FPU double precision
- * registers.
- *
- * VPFv4-NEON is VFPv4-D32 with NEON instructions. It also adds fused
- * multiply-accumulate instructions that work on the NEON registers.
- *
- * NOTE: Similarly, "VFPv4" might either reference VFPv4-D16 or VFPv4-D32
- *       depending on context.
- *
- * The following information was determined by scanning the binutils-2.22
- * sources:
- *
- * Basic VFP instruction subsets:
- *
- * #define FPU_VFP_EXT_V1xD 0x08000000     // Base VFP instruction set.
- * #define FPU_VFP_EXT_V1   0x04000000     // Double-precision insns.
- * #define FPU_VFP_EXT_V2   0x02000000     // ARM10E VFPr1.
- * #define FPU_VFP_EXT_V3xD 0x01000000     // VFPv3 single-precision.
- * #define FPU_VFP_EXT_V3   0x00800000     // VFPv3 double-precision.
- * #define FPU_NEON_EXT_V1  0x00400000     // Neon (SIMD) insns.
- * #define FPU_VFP_EXT_D32  0x00200000     // Registers D16-D31.
- * #define FPU_VFP_EXT_FP16 0x00100000     // Half-precision extensions.
- * #define FPU_NEON_EXT_FMA 0x00080000     // Neon fused multiply-add
- * #define FPU_VFP_EXT_FMA  0x00040000     // VFP fused multiply-add
- *
- * FPU types (excluding NEON)
- *
- * FPU_VFP_V1xD (EXT_V1xD)
- *    |
- *    +--------------------------+
- *    |                          |
- * FPU_VFP_V1 (+EXT_V1)       FPU_VFP_V3xD (+EXT_V2+EXT_V3xD)
- *    |                          |
- *    |                          |
- * FPU_VFP_V2 (+EXT_V2)       FPU_VFP_V4_SP_D16 (+EXT_FP16+EXT_FMA)
- *    |
- * FPU_VFP_V3D16 (+EXT_Vx3D+EXT_V3)
- *    |
- *    +--------------------------+
- *    |                          |
- * FPU_VFP_V3 (+EXT_D32)     FPU_VFP_V4D16 (+EXT_FP16+EXT_FMA)
- *    |                          |
- *    |                      FPU_VFP_V4 (+EXT_D32)
- *    |
- * FPU_VFP_HARD (+EXT_FMA+NEON_EXT_FMA)
- *
- * VFP architectures:
- *
- * ARCH_VFP_V1xD  (EXT_V1xD)
- *   |
- *   +------------------+
- *   |                  |
- *   |             ARCH_VFP_V3xD (+EXT_V2+EXT_V3xD)
- *   |                  |
- *   |             ARCH_VFP_V3xD_FP16 (+EXT_FP16)
- *   |                  |
- *   |             ARCH_VFP_V4_SP_D16 (+EXT_FMA)
- *   |
- * ARCH_VFP_V1 (+EXT_V1)
- *   |
- * ARCH_VFP_V2 (+EXT_V2)
- *   |
- * ARCH_VFP_V3D16 (+EXT_V3xD+EXT_V3)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V3D16_FP16  (+EXT_FP16)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V4_D16 (+EXT_FP16+EXT_FMA)
- *   |                   |
- *   |         ARCH_VFP_V4 (+EXT_D32)
- *   |                   |
- *   |         ARCH_NEON_VFP_V4 (+EXT_NEON+EXT_NEON_FMA)
- *   |
- * ARCH_VFP_V3 (+EXT_D32)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V3_FP16 (+EXT_FP16)
- *   |
- * ARCH_VFP_V3_PLUS_NEON_V1 (+EXT_NEON)
- *   |
- * ARCH_NEON_FP16 (+EXT_FP16)
- *
- * -fpu=<name> values and their correspondance with FPU architectures above:
- *
- *   {"vfp",               FPU_ARCH_VFP_V2},
- *   {"vfp9",              FPU_ARCH_VFP_V2},
- *   {"vfp3",              FPU_ARCH_VFP_V3}, // For backwards compatbility.
- *   {"vfp10",             FPU_ARCH_VFP_V2},
- *   {"vfp10-r0",          FPU_ARCH_VFP_V1},
- *   {"vfpxd",             FPU_ARCH_VFP_V1xD},
- *   {"vfpv2",             FPU_ARCH_VFP_V2},
- *   {"vfpv3",             FPU_ARCH_VFP_V3},
- *   {"vfpv3-fp16",        FPU_ARCH_VFP_V3_FP16},
- *   {"vfpv3-d16",         FPU_ARCH_VFP_V3D16},
- *   {"vfpv3-d16-fp16",    FPU_ARCH_VFP_V3D16_FP16},
- *   {"vfpv3xd",           FPU_ARCH_VFP_V3xD},
- *   {"vfpv3xd-fp16",      FPU_ARCH_VFP_V3xD_FP16},
- *   {"neon",              FPU_ARCH_VFP_V3_PLUS_NEON_V1},
- *   {"neon-fp16",         FPU_ARCH_NEON_FP16},
- *   {"vfpv4",             FPU_ARCH_VFP_V4},
- *   {"vfpv4-d16",         FPU_ARCH_VFP_V4D16},
- *   {"fpv4-sp-d16",       FPU_ARCH_VFP_V4_SP_D16},
- *   {"neon-vfpv4",        FPU_ARCH_NEON_VFP_V4},
- *
- *
- * Simplified diagram that only includes FPUs supported by Android:
- * Only ARCH_VFP_V3D16 is actually mandated by the armeabi-v7a ABI,
- * all others are optional and must be probed at runtime.
- *
- * ARCH_VFP_V3D16 (EXT_V1xD+EXT_V1+EXT_V2+EXT_V3xD+EXT_V3)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V3D16_FP16  (+EXT_FP16)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V4_D16 (+EXT_FP16+EXT_FMA)
- *   |                   |
- *   |         ARCH_VFP_V4 (+EXT_D32)
- *   |                   |
- *   |         ARCH_NEON_VFP_V4 (+EXT_NEON+EXT_NEON_FMA)
- *   |
- * ARCH_VFP_V3 (+EXT_D32)
- *   |
- *   +-------------------+
- *   |                   |
- *   |         ARCH_VFP_V3_FP16 (+EXT_FP16)
- *   |
- * ARCH_VFP_V3_PLUS_NEON_V1 (+EXT_NEON)
- *   |
- * ARCH_NEON_FP16 (+EXT_FP16)
- *
- */
-
-#endif // defined(__le32__)
-#endif
-
-#endif
diff --git a/drivers/theoraplayer/src/YUV/android/cpu-features.h b/drivers/theoraplayer/src/YUV/android/cpu-features.h
deleted file mode 100644
index 12d3ad5645..0000000000
--- a/drivers/theoraplayer/src/YUV/android/cpu-features.h
+++ /dev/null
@@ -1,212 +0,0 @@
-/*
- * Copyright (C) 2010 The Android Open Source Project
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- *  * Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in
- *    the documentation and/or other materials provided with the
- *    distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
- * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
- * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
- * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
- * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
- * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
- * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
- * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-#ifndef CPU_FEATURES_H
-#define CPU_FEATURES_H
-
-#include <sys/cdefs.h>
-#include <stdint.h>
-
-__BEGIN_DECLS
-
-typedef enum {
-    ANDROID_CPU_FAMILY_UNKNOWN = 0,
-    ANDROID_CPU_FAMILY_ARM,
-    ANDROID_CPU_FAMILY_X86,
-    ANDROID_CPU_FAMILY_MIPS,
-
-    ANDROID_CPU_FAMILY_MAX  /* do not remove */
-
-} AndroidCpuFamily;
-
-/* Return family of the device's CPU */
-extern AndroidCpuFamily   android_getCpuFamily(void);
-
-/* The list of feature flags for ARM CPUs that can be recognized by the
- * library. Value details are:
- *
- *   VFPv2:
- *     CPU supports the VFPv2 instruction set. Many, but not all, ARMv6 CPUs
- *     support these instructions. VFPv2 is a subset of VFPv3 so this will
- *     be set whenever VFPv3 is set too.
- *
- *   ARMv7:
- *     CPU supports the ARMv7-A basic instruction set.
- *     This feature is mandated by the 'armeabi-v7a' ABI.
- *
- *   VFPv3:
- *     CPU supports the VFPv3-D16 instruction set, providing hardware FPU
- *     support for single and double precision floating point registers.
- *     Note that only 16 FPU registers are available by default, unless
- *     the D32 bit is set too. This feature is also mandated by the
- *     'armeabi-v7a' ABI.
- *
- *   VFP_D32:
- *     CPU VFP optional extension that provides 32 FPU registers,
- *     instead of 16. Note that ARM mandates this feature is the 'NEON'
- *     feature is implemented by the CPU.
- *
- *   NEON:
- *     CPU FPU supports "ARM Advanced SIMD" instructions, also known as
- *     NEON. Note that this mandates the VFP_D32 feature as well, per the
- *     ARM Architecture specification.
- *
- *   VFP_FP16:
- *     Half-width floating precision VFP extension. If set, the CPU
- *     supports instructions to perform floating-point operations on
- *     16-bit registers. This is part of the VFPv4 specification, but
- *     not mandated by any Android ABI.
- *
- *   VFP_FMA:
- *     Fused multiply-accumulate VFP instructions extension. Also part of
- *     the VFPv4 specification, but not mandated by any Android ABI.
- *
- *   NEON_FMA:
- *     Fused multiply-accumulate NEON instructions extension. Optional
- *     extension from the VFPv4 specification, but not mandated by any
- *     Android ABI.
- *
- *   IDIV_ARM:
- *     Integer division available in ARM mode. Only available
- *     on recent CPUs (e.g. Cortex-A15).
- *
- *   IDIV_THUMB2:
- *     Integer division available in Thumb-2 mode. Only available
- *     on recent CPUs (e.g. Cortex-A15).
- *
- *   iWMMXt:
- *     Optional extension that adds MMX registers and operations to an
- *     ARM CPU. This is only available on a few XScale-based CPU designs
- *     sold by Marvell. Pretty rare in practice.
- *
- * If you want to tell the compiler to generate code that targets one of
- * the feature set above, you should probably use one of the following
- * flags (for more details, see technical note at the end of this file):
- *
- *   -mfpu=vfp
- *   -mfpu=vfpv2
- *     These are equivalent and tell GCC to use VFPv2 instructions for
- *     floating-point operations. Use this if you want your code to
- *     run on *some* ARMv6 devices, and any ARMv7-A device supported
- *     by Android.
- *
- *     Generated code requires VFPv2 feature.
- *
- *   -mfpu=vfpv3-d16
- *     Tell GCC to use VFPv3 instructions (using only 16 FPU registers).
- *     This should be generic code that runs on any CPU that supports the
- *     'armeabi-v7a' Android ABI. Note that no ARMv6 CPU supports this.
- *
- *     Generated code requires VFPv3 feature.
- *
- *   -mfpu=vfpv3
- *     Tell GCC to use VFPv3 instructions with 32 FPU registers.
- *     Generated code requires VFPv3|VFP_D32 features.
- *
- *   -mfpu=neon
- *     Tell GCC to use VFPv3 instructions with 32 FPU registers, and
- *     also support NEON intrinsics (see <arm_neon.h>).
- *     Generated code requires VFPv3|VFP_D32|NEON features.
- *
- *   -mfpu=vfpv4-d16
- *     Generated code requires VFPv3|VFP_FP16|VFP_FMA features.
- *
- *   -mfpu=vfpv4
- *     Generated code requires VFPv3|VFP_FP16|VFP_FMA|VFP_D32 features.
- *
- *   -mfpu=neon-vfpv4
- *     Generated code requires VFPv3|VFP_FP16|VFP_FMA|VFP_D32|NEON|NEON_FMA
- *     features.
- *
- *   -mcpu=cortex-a7
- *   -mcpu=cortex-a15
- *     Generated code requires VFPv3|VFP_FP16|VFP_FMA|VFP_D32|
- *                             NEON|NEON_FMA|IDIV_ARM|IDIV_THUMB2
- *     This flag implies -mfpu=neon-vfpv4.
- *
- *   -mcpu=iwmmxt
- *     Allows the use of iWMMXt instrinsics with GCC.
- */
-enum {
-    ANDROID_CPU_ARM_FEATURE_ARMv7       = (1 << 0),
-    ANDROID_CPU_ARM_FEATURE_VFPv3       = (1 << 1),
-    ANDROID_CPU_ARM_FEATURE_NEON        = (1 << 2),
-    ANDROID_CPU_ARM_FEATURE_LDREX_STREX = (1 << 3),
-    ANDROID_CPU_ARM_FEATURE_VFPv2       = (1 << 4),
-    ANDROID_CPU_ARM_FEATURE_VFP_D32     = (1 << 5),
-    ANDROID_CPU_ARM_FEATURE_VFP_FP16    = (1 << 6),
-    ANDROID_CPU_ARM_FEATURE_VFP_FMA     = (1 << 7),
-    ANDROID_CPU_ARM_FEATURE_NEON_FMA    = (1 << 8),
-    ANDROID_CPU_ARM_FEATURE_IDIV_ARM    = (1 << 9),
-    ANDROID_CPU_ARM_FEATURE_IDIV_THUMB2 = (1 << 10),
-    ANDROID_CPU_ARM_FEATURE_iWMMXt      = (1 << 11),
-};
-
-enum {
-    ANDROID_CPU_X86_FEATURE_SSSE3  = (1 << 0),
-    ANDROID_CPU_X86_FEATURE_POPCNT = (1 << 1),
-    ANDROID_CPU_X86_FEATURE_MOVBE  = (1 << 2),
-};
-
-// libtheoraplayer addition, renamed this to "Ext" as not to conflict with your own project if you've included cpu-features.c in it
-//extern uint64_t    android_getCpuFeaturesExt(void);
-#define android_getCpuFeaturesExt android_getCpuFeatures
-
-/* Return the number of CPU cores detected on this device. */
-extern int         android_getCpuCount(void);
-
-/* The following is used to force the CPU count and features
- * mask in sandboxed processes. Under 4.1 and higher, these processes
- * cannot access /proc, which is the only way to get information from
- * the kernel about the current hardware (at least on ARM).
- *
- * It _must_ be called only once, and before any android_getCpuXXX
- * function, any other case will fail.
- *
- * This function return 1 on success, and 0 on failure.
- */
-extern int android_setCpu(int      cpu_count,
-                          uint64_t cpu_features);
-
-#ifdef __arm__
-/* Retrieve the ARM 32-bit CPUID value from the kernel.
- * Note that this cannot work on sandboxed processes under 4.1 and
- * higher, unless you called android_setCpuArm() before.
- */
-extern uint32_t android_getCpuIdArm(void);
-
-/* An ARM-specific variant of android_setCpu() that also allows you
- * to set the ARM CPUID field.
- */
-extern int android_setCpuArm(int      cpu_count,
-                             uint64_t cpu_features,
-                             uint32_t cpu_id);
-#endif
-
-__END_DECLS
-
-#endif /* CPU_FEATURES_H */
diff --git a/drivers/theoraplayer/src/YUV/libyuv/LICENSE b/drivers/theoraplayer/src/YUV/libyuv/LICENSE
deleted file mode 100755
index c911747a6b..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/LICENSE
+++ /dev/null
@@ -1,29 +0,0 @@
-Copyright 2011 The LibYuv Project Authors. All rights reserved.
-
-Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions are
-met:
-
-  * Redistributions of source code must retain the above copyright
-    notice, this list of conditions and the following disclaimer.
-
-  * Redistributions in binary form must reproduce the above copyright
-    notice, this list of conditions and the following disclaimer in
-    the documentation and/or other materials provided with the
-    distribution.
-
-  * Neither the name of Google nor the names of its contributors may
-    be used to endorse or promote products derived from this software
-    without specific prior written permission.
-
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/drivers/theoraplayer/src/YUV/libyuv/LICENSE_THIRD_PARTY b/drivers/theoraplayer/src/YUV/libyuv/LICENSE_THIRD_PARTY
deleted file mode 100755
index a71591e771..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/LICENSE_THIRD_PARTY
+++ /dev/null
@@ -1,8 +0,0 @@
-This source tree contains third party source code which is governed by third
-party licenses. This file contains references to files which are under other
-licenses than the one provided in the LICENSE file in the root of the source
-tree.
-
-Files governed by third party licenses:
-source/x86inc.asm
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv.h
deleted file mode 100755
index 3bebe642cc..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_H_  // NOLINT
-#define INCLUDE_LIBYUV_H_
-
-#include "libyuv/basic_types.h"
-#include "libyuv/compare.h"
-#include "libyuv/convert.h"
-#include "libyuv/convert_argb.h"
-#include "libyuv/convert_from.h"
-#include "libyuv/convert_from_argb.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#include "libyuv/mjpeg_decoder.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/rotate.h"
-#include "libyuv/rotate_argb.h"
-#include "libyuv/row.h"
-#include "libyuv/scale.h"
-#include "libyuv/scale_argb.h"
-#include "libyuv/scale_row.h"
-#include "libyuv/version.h"
-#include "libyuv/video_common.h"
-
-#endif  // INCLUDE_LIBYUV_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/basic_types.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/basic_types.h
deleted file mode 100755
index beb750ba65..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/basic_types.h
+++ /dev/null
@@ -1,118 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_BASIC_TYPES_H_  // NOLINT
-#define INCLUDE_LIBYUV_BASIC_TYPES_H_
-
-#include <stddef.h>  // for NULL, size_t
-
-#if defined(__ANDROID__) || (defined(_MSC_VER) && (_MSC_VER < 1600))
-#include <sys/types.h>  // for uintptr_t on x86
-#else
-#include <stdint.h>  // for uintptr_t
-#endif
-
-#ifndef GG_LONGLONG
-#ifndef INT_TYPES_DEFINED
-#define INT_TYPES_DEFINED
-#ifdef COMPILER_MSVC
-typedef unsigned __int64 uint64;
-typedef __int64 int64;
-#ifndef INT64_C
-#define INT64_C(x) x ## I64
-#endif
-#ifndef UINT64_C
-#define UINT64_C(x) x ## UI64
-#endif
-#define INT64_F "I64"
-#else  // COMPILER_MSVC
-#if defined(__LP64__) && !defined(__OpenBSD__) && !defined(__APPLE__)
-typedef unsigned long uint64;  // NOLINT
-typedef long int64;  // NOLINT
-#ifndef INT64_C
-#define INT64_C(x) x ## L
-#endif
-#ifndef UINT64_C
-#define UINT64_C(x) x ## UL
-#endif
-#define INT64_F "l"
-#else  // defined(__LP64__) && !defined(__OpenBSD__) && !defined(__APPLE__)
-typedef unsigned long long uint64;  // NOLINT
-typedef long long int64;  // NOLINT
-#ifndef INT64_C
-#define INT64_C(x) x ## LL
-#endif
-#ifndef UINT64_C
-#define UINT64_C(x) x ## ULL
-#endif
-#define INT64_F "ll"
-#endif  // __LP64__
-#endif  // COMPILER_MSVC
-typedef unsigned int uint32;
-typedef int int32;
-typedef unsigned short uint16;  // NOLINT
-typedef short int16;  // NOLINT
-typedef unsigned char uint8;
-typedef signed char int8;
-#endif  // INT_TYPES_DEFINED
-#endif  // GG_LONGLONG
-
-// Detect compiler is for x86 or x64.
-#if defined(__x86_64__) || defined(_M_X64) || \
-    defined(__i386__) || defined(_M_IX86)
-#define CPU_X86 1
-#endif
-// Detect compiler is for ARM.
-#if defined(__arm__) || defined(_M_ARM)
-#define CPU_ARM 1
-#endif
-
-#ifndef ALIGNP
-#ifdef __cplusplus
-#define ALIGNP(p, t) \
-    (reinterpret_cast<uint8*>(((reinterpret_cast<uintptr_t>(p) + \
-    ((t) - 1)) & ~((t) - 1))))
-#else
-#define ALIGNP(p, t) \
-    ((uint8*)((((uintptr_t)(p) + ((t) - 1)) & ~((t) - 1))))  /* NOLINT */
-#endif
-#endif
-
-#if !defined(LIBYUV_API)
-#if defined(_WIN32) || defined(__CYGWIN__)
-#if defined(LIBYUV_BUILDING_SHARED_LIBRARY)
-#define LIBYUV_API __declspec(dllexport)
-#elif defined(LIBYUV_USING_SHARED_LIBRARY)
-#define LIBYUV_API __declspec(dllimport)
-#else
-#define LIBYUV_API
-#endif  // LIBYUV_BUILDING_SHARED_LIBRARY
-#elif defined(__GNUC__) && (__GNUC__ >= 4) && !defined(__APPLE__) && \
-    (defined(LIBYUV_BUILDING_SHARED_LIBRARY) || \
-    defined(LIBYUV_USING_SHARED_LIBRARY))
-#define LIBYUV_API __attribute__ ((visibility ("default")))
-#else
-#define LIBYUV_API
-#endif  // __GNUC__
-#endif  // LIBYUV_API
-
-#define LIBYUV_BOOL int
-#define LIBYUV_FALSE 0
-#define LIBYUV_TRUE 1
-
-// Visual C x86 or GCC little endian.
-#if defined(__x86_64__) || defined(_M_X64) || \
-  defined(__i386__) || defined(_M_IX86) || \
-  defined(__arm__) || defined(_M_ARM) || \
-  (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
-#define LIBYUV_LITTLE_ENDIAN
-#endif
-
-#endif  // INCLUDE_LIBYUV_BASIC_TYPES_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/compare.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/compare.h
deleted file mode 100755
index 5dfac7c86a..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/compare.h
+++ /dev/null
@@ -1,73 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_COMPARE_H_  // NOLINT
-#define INCLUDE_LIBYUV_COMPARE_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Compute a hash for specified memory. Seed of 5381 recommended.
-LIBYUV_API
-uint32 HashDjb2(const uint8* src, uint64 count, uint32 seed);
-
-// Sum Square Error - used to compute Mean Square Error or PSNR.
-LIBYUV_API
-uint64 ComputeSumSquareError(const uint8* src_a,
-                             const uint8* src_b, int count);
-
-LIBYUV_API
-uint64 ComputeSumSquareErrorPlane(const uint8* src_a, int stride_a,
-                                  const uint8* src_b, int stride_b,
-                                  int width, int height);
-
-static const int kMaxPsnr = 128;
-
-LIBYUV_API
-double SumSquareErrorToPsnr(uint64 sse, uint64 count);
-
-LIBYUV_API
-double CalcFramePsnr(const uint8* src_a, int stride_a,
-                     const uint8* src_b, int stride_b,
-                     int width, int height);
-
-LIBYUV_API
-double I420Psnr(const uint8* src_y_a, int stride_y_a,
-                const uint8* src_u_a, int stride_u_a,
-                const uint8* src_v_a, int stride_v_a,
-                const uint8* src_y_b, int stride_y_b,
-                const uint8* src_u_b, int stride_u_b,
-                const uint8* src_v_b, int stride_v_b,
-                int width, int height);
-
-LIBYUV_API
-double CalcFrameSsim(const uint8* src_a, int stride_a,
-                     const uint8* src_b, int stride_b,
-                     int width, int height);
-
-LIBYUV_API
-double I420Ssim(const uint8* src_y_a, int stride_y_a,
-                const uint8* src_u_a, int stride_u_a,
-                const uint8* src_v_a, int stride_v_a,
-                const uint8* src_y_b, int stride_y_b,
-                const uint8* src_u_b, int stride_u_b,
-                const uint8* src_v_b, int stride_v_b,
-                int width, int height);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_COMPARE_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert.h
deleted file mode 100755
index 1bd45c837f..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert.h
+++ /dev/null
@@ -1,254 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_CONVERT_H_  // NOLINT
-#define INCLUDE_LIBYUV_CONVERT_H_
-
-#include "libyuv/basic_types.h"
-// TODO(fbarchard): Remove the following headers includes.
-#include "libyuv/convert_from.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/rotate.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Convert I444 to I420.
-LIBYUV_API
-int I444ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert I422 to I420.
-LIBYUV_API
-int I422ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert I411 to I420.
-LIBYUV_API
-int I411ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Copy I420 to I420.
-#define I420ToI420 I420Copy
-LIBYUV_API
-int I420Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height);
-
-// Convert I400 (grey) to I420.
-LIBYUV_API
-int I400ToI420(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert NV12 to I420.
-LIBYUV_API
-int NV12ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_uv, int src_stride_uv,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert NV21 to I420.
-LIBYUV_API
-int NV21ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_vu, int src_stride_vu,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert YUY2 to I420.
-LIBYUV_API
-int YUY2ToI420(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert UYVY to I420.
-LIBYUV_API
-int UYVYToI420(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert M420 to I420.
-LIBYUV_API
-int M420ToI420(const uint8* src_m420, int src_stride_m420,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert Q420 to I420.
-LIBYUV_API
-int Q420ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// ARGB little endian (bgra in memory) to I420.
-LIBYUV_API
-int ARGBToI420(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// BGRA little endian (argb in memory) to I420.
-LIBYUV_API
-int BGRAToI420(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// ABGR little endian (rgba in memory) to I420.
-LIBYUV_API
-int ABGRToI420(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// RGBA little endian (abgr in memory) to I420.
-LIBYUV_API
-int RGBAToI420(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// RGB little endian (bgr in memory) to I420.
-LIBYUV_API
-int RGB24ToI420(const uint8* src_frame, int src_stride_frame,
-                uint8* dst_y, int dst_stride_y,
-                uint8* dst_u, int dst_stride_u,
-                uint8* dst_v, int dst_stride_v,
-                int width, int height);
-
-// RGB big endian (rgb in memory) to I420.
-LIBYUV_API
-int RAWToI420(const uint8* src_frame, int src_stride_frame,
-              uint8* dst_y, int dst_stride_y,
-              uint8* dst_u, int dst_stride_u,
-              uint8* dst_v, int dst_stride_v,
-              int width, int height);
-
-// RGB16 (RGBP fourcc) little endian to I420.
-LIBYUV_API
-int RGB565ToI420(const uint8* src_frame, int src_stride_frame,
-                 uint8* dst_y, int dst_stride_y,
-                 uint8* dst_u, int dst_stride_u,
-                 uint8* dst_v, int dst_stride_v,
-                 int width, int height);
-
-// RGB15 (RGBO fourcc) little endian to I420.
-LIBYUV_API
-int ARGB1555ToI420(const uint8* src_frame, int src_stride_frame,
-                   uint8* dst_y, int dst_stride_y,
-                   uint8* dst_u, int dst_stride_u,
-                   uint8* dst_v, int dst_stride_v,
-                   int width, int height);
-
-// RGB12 (R444 fourcc) little endian to I420.
-LIBYUV_API
-int ARGB4444ToI420(const uint8* src_frame, int src_stride_frame,
-                   uint8* dst_y, int dst_stride_y,
-                   uint8* dst_u, int dst_stride_u,
-                   uint8* dst_v, int dst_stride_v,
-                   int width, int height);
-
-#ifdef HAVE_JPEG
-// src_width/height provided by capture.
-// dst_width/height for clipping determine final size.
-LIBYUV_API
-int MJPGToI420(const uint8* sample, size_t sample_size,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int src_width, int src_height,
-               int dst_width, int dst_height);
-
-// Query size of MJPG in pixels.
-LIBYUV_API
-int MJPGSize(const uint8* sample, size_t sample_size,
-             int* width, int* height);
-#endif
-
-// Note Bayer formats (BGGR) To I420 are in format_conversion.h
-
-// Convert camera sample to I420 with cropping, rotation and vertical flip.
-// "src_size" is needed to parse MJPG.
-// "dst_stride_y" number of bytes in a row of the dst_y plane.
-//   Normally this would be the same as dst_width, with recommended alignment
-//   to 16 bytes for better efficiency.
-//   If rotation of 90 or 270 is used, stride is affected. The caller should
-//   allocate the I420 buffer according to rotation.
-// "dst_stride_u" number of bytes in a row of the dst_u plane.
-//   Normally this would be the same as (dst_width + 1) / 2, with
-//   recommended alignment to 16 bytes for better efficiency.
-//   If rotation of 90 or 270 is used, stride is affected.
-// "crop_x" and "crop_y" are starting position for cropping.
-//   To center, crop_x = (src_width - dst_width) / 2
-//              crop_y = (src_height - dst_height) / 2
-// "src_width" / "src_height" is size of src_frame in pixels.
-//   "src_height" can be negative indicating a vertically flipped image source.
-// "crop_width" / "crop_height" is the size to crop the src to.
-//    Must be less than or equal to src_width/src_height
-//    Cropping parameters are pre-rotation.
-// "rotation" can be 0, 90, 180 or 270.
-// "format" is a fourcc. ie 'I420', 'YUY2'
-// Returns 0 for successful; -1 for invalid parameter. Non-zero for failure.
-LIBYUV_API
-int ConvertToI420(const uint8* src_frame, size_t src_size,
-                  uint8* dst_y, int dst_stride_y,
-                  uint8* dst_u, int dst_stride_u,
-                  uint8* dst_v, int dst_stride_v,
-                  int crop_x, int crop_y,
-                  int src_width, int src_height,
-                  int crop_width, int crop_height,
-                  enum RotationMode rotation,
-                  uint32 format);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_CONVERT_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_argb.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_argb.h
deleted file mode 100755
index a18014ca2c..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_argb.h
+++ /dev/null
@@ -1,225 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_CONVERT_ARGB_H_  // NOLINT
-#define INCLUDE_LIBYUV_CONVERT_ARGB_H_
-
-#include "libyuv/basic_types.h"
-// TODO(fbarchard): Remove the following headers includes
-#include "libyuv/convert_from.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/rotate.h"
-
-// TODO(fbarchard): This set of functions should exactly match convert.h
-// Add missing Q420.
-// TODO(fbarchard): Add tests. Create random content of right size and convert
-// with C vs Opt and or to I420 and compare.
-// TODO(fbarchard): Some of these functions lack parameter setting.
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Alias.
-#define ARGBToARGB ARGBCopy
-
-// Copy ARGB to ARGB.
-LIBYUV_API
-int ARGBCopy(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int width, int height);
-
-// Convert I420 to ARGB.
-LIBYUV_API
-int I420ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert I422 to ARGB.
-LIBYUV_API
-int I422ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert I444 to ARGB.
-LIBYUV_API
-int I444ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert I411 to ARGB.
-LIBYUV_API
-int I411ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert I400 (grey) to ARGB.
-LIBYUV_API
-int I400ToARGB(const uint8* src_y, int src_stride_y,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Alias.
-#define YToARGB I400ToARGB_Reference
-
-// Convert I400 to ARGB. Reverse of ARGBToI400.
-LIBYUV_API
-int I400ToARGB_Reference(const uint8* src_y, int src_stride_y,
-                         uint8* dst_argb, int dst_stride_argb,
-                         int width, int height);
-
-// Convert NV12 to ARGB.
-LIBYUV_API
-int NV12ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_uv, int src_stride_uv,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert NV21 to ARGB.
-LIBYUV_API
-int NV21ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_vu, int src_stride_vu,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert M420 to ARGB.
-LIBYUV_API
-int M420ToARGB(const uint8* src_m420, int src_stride_m420,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// TODO(fbarchard): Convert Q420 to ARGB.
-// LIBYUV_API
-// int Q420ToARGB(const uint8* src_y, int src_stride_y,
-//                const uint8* src_yuy2, int src_stride_yuy2,
-//                uint8* dst_argb, int dst_stride_argb,
-//                int width, int height);
-
-// Convert YUY2 to ARGB.
-LIBYUV_API
-int YUY2ToARGB(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert UYVY to ARGB.
-LIBYUV_API
-int UYVYToARGB(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// BGRA little endian (argb in memory) to ARGB.
-LIBYUV_API
-int BGRAToARGB(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// ABGR little endian (rgba in memory) to ARGB.
-LIBYUV_API
-int ABGRToARGB(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// RGBA little endian (abgr in memory) to ARGB.
-LIBYUV_API
-int RGBAToARGB(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Deprecated function name.
-#define BG24ToARGB RGB24ToARGB
-
-// RGB little endian (bgr in memory) to ARGB.
-LIBYUV_API
-int RGB24ToARGB(const uint8* src_frame, int src_stride_frame,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height);
-
-// RGB big endian (rgb in memory) to ARGB.
-LIBYUV_API
-int RAWToARGB(const uint8* src_frame, int src_stride_frame,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height);
-
-// RGB16 (RGBP fourcc) little endian to ARGB.
-LIBYUV_API
-int RGB565ToARGB(const uint8* src_frame, int src_stride_frame,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height);
-
-// RGB15 (RGBO fourcc) little endian to ARGB.
-LIBYUV_API
-int ARGB1555ToARGB(const uint8* src_frame, int src_stride_frame,
-                   uint8* dst_argb, int dst_stride_argb,
-                   int width, int height);
-
-// RGB12 (R444 fourcc) little endian to ARGB.
-LIBYUV_API
-int ARGB4444ToARGB(const uint8* src_frame, int src_stride_frame,
-                   uint8* dst_argb, int dst_stride_argb,
-                   int width, int height);
-
-#ifdef HAVE_JPEG
-// src_width/height provided by capture
-// dst_width/height for clipping determine final size.
-LIBYUV_API
-int MJPGToARGB(const uint8* sample, size_t sample_size,
-               uint8* dst_argb, int dst_stride_argb,
-               int src_width, int src_height,
-               int dst_width, int dst_height);
-#endif
-
-// Note Bayer formats (BGGR) to ARGB are in format_conversion.h.
-
-// Convert camera sample to ARGB with cropping, rotation and vertical flip.
-// "src_size" is needed to parse MJPG.
-// "dst_stride_argb" number of bytes in a row of the dst_argb plane.
-//   Normally this would be the same as dst_width, with recommended alignment
-//   to 16 bytes for better efficiency.
-//   If rotation of 90 or 270 is used, stride is affected. The caller should
-//   allocate the I420 buffer according to rotation.
-// "dst_stride_u" number of bytes in a row of the dst_u plane.
-//   Normally this would be the same as (dst_width + 1) / 2, with
-//   recommended alignment to 16 bytes for better efficiency.
-//   If rotation of 90 or 270 is used, stride is affected.
-// "crop_x" and "crop_y" are starting position for cropping.
-//   To center, crop_x = (src_width - dst_width) / 2
-//              crop_y = (src_height - dst_height) / 2
-// "src_width" / "src_height" is size of src_frame in pixels.
-//   "src_height" can be negative indicating a vertically flipped image source.
-// "crop_width" / "crop_height" is the size to crop the src to.
-//    Must be less than or equal to src_width/src_height
-//    Cropping parameters are pre-rotation.
-// "rotation" can be 0, 90, 180 or 270.
-// "format" is a fourcc. ie 'I420', 'YUY2'
-// Returns 0 for successful; -1 for invalid parameter. Non-zero for failure.
-LIBYUV_API
-int ConvertToARGB(const uint8* src_frame, size_t src_size,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int crop_x, int crop_y,
-                  int src_width, int src_height,
-                  int crop_width, int crop_height,
-                  enum RotationMode rotation,
-                  uint32 format);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_CONVERT_ARGB_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from.h
deleted file mode 100755
index b1cf57f7dc..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from.h
+++ /dev/null
@@ -1,173 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_CONVERT_FROM_H_  // NOLINT
-#define INCLUDE_LIBYUV_CONVERT_FROM_H_
-
-#include "libyuv/basic_types.h"
-#include "libyuv/rotate.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// See Also convert.h for conversions from formats to I420.
-
-// I420Copy in convert to I420ToI420.
-
-LIBYUV_API
-int I420ToI422(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-LIBYUV_API
-int I420ToI444(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-LIBYUV_API
-int I420ToI411(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Copy to I400. Source can be I420, I422, I444, I400, NV12 or NV21.
-LIBYUV_API
-int I400Copy(const uint8* src_y, int src_stride_y,
-             uint8* dst_y, int dst_stride_y,
-             int width, int height);
-
-// TODO(fbarchard): I420ToM420
-// TODO(fbarchard): I420ToQ420
-
-LIBYUV_API
-int I420ToNV12(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height);
-
-LIBYUV_API
-int I420ToNV21(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_vu, int dst_stride_vu,
-               int width, int height);
-
-LIBYUV_API
-int I420ToYUY2(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_frame, int dst_stride_frame,
-               int width, int height);
-
-LIBYUV_API
-int I420ToUYVY(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_frame, int dst_stride_frame,
-               int width, int height);
-
-LIBYUV_API
-int I420ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-LIBYUV_API
-int I420ToBGRA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-LIBYUV_API
-int I420ToABGR(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-LIBYUV_API
-int I420ToRGBA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height);
-
-LIBYUV_API
-int I420ToRGB24(const uint8* src_y, int src_stride_y,
-                const uint8* src_u, int src_stride_u,
-                const uint8* src_v, int src_stride_v,
-                uint8* dst_frame, int dst_stride_frame,
-                int width, int height);
-
-LIBYUV_API
-int I420ToRAW(const uint8* src_y, int src_stride_y,
-              const uint8* src_u, int src_stride_u,
-              const uint8* src_v, int src_stride_v,
-              uint8* dst_frame, int dst_stride_frame,
-              int width, int height);
-
-LIBYUV_API
-int I420ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_u, int src_stride_u,
-                 const uint8* src_v, int src_stride_v,
-                 uint8* dst_frame, int dst_stride_frame,
-                 int width, int height);
-
-LIBYUV_API
-int I420ToARGB1555(const uint8* src_y, int src_stride_y,
-                   const uint8* src_u, int src_stride_u,
-                   const uint8* src_v, int src_stride_v,
-                   uint8* dst_frame, int dst_stride_frame,
-                   int width, int height);
-
-LIBYUV_API
-int I420ToARGB4444(const uint8* src_y, int src_stride_y,
-                   const uint8* src_u, int src_stride_u,
-                   const uint8* src_v, int src_stride_v,
-                   uint8* dst_frame, int dst_stride_frame,
-                   int width, int height);
-
-// Note Bayer formats (BGGR) To I420 are in format_conversion.h.
-
-// Convert I420 to specified format.
-// "dst_sample_stride" is bytes in a row for the destination. Pass 0 if the
-//    buffer has contiguous rows. Can be negative. A multiple of 16 is optimal.
-LIBYUV_API
-int ConvertFromI420(const uint8* y, int y_stride,
-                    const uint8* u, int u_stride,
-                    const uint8* v, int v_stride,
-                    uint8* dst_sample, int dst_sample_stride,
-                    int width, int height,
-                    uint32 format);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_CONVERT_FROM_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from_argb.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from_argb.h
deleted file mode 100755
index f0343a77d3..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/convert_from_argb.h
+++ /dev/null
@@ -1,168 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_CONVERT_FROM_ARGB_H_  // NOLINT
-#define INCLUDE_LIBYUV_CONVERT_FROM_ARGB_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Copy ARGB to ARGB.
-#define ARGBToARGB ARGBCopy
-LIBYUV_API
-int ARGBCopy(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int width, int height);
-
-// Convert ARGB To BGRA. (alias)
-#define ARGBToBGRA BGRAToARGB
-LIBYUV_API
-int BGRAToARGB(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert ARGB To ABGR. (alias)
-#define ARGBToABGR ABGRToARGB
-LIBYUV_API
-int ABGRToARGB(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert ARGB To RGBA.
-LIBYUV_API
-int ARGBToRGBA(const uint8* src_frame, int src_stride_frame,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert ARGB To RGB24.
-LIBYUV_API
-int ARGBToRGB24(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_rgb24, int dst_stride_rgb24,
-                int width, int height);
-
-// Convert ARGB To RAW.
-LIBYUV_API
-int ARGBToRAW(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_rgb, int dst_stride_rgb,
-              int width, int height);
-
-// Convert ARGB To RGB565.
-LIBYUV_API
-int ARGBToRGB565(const uint8* src_argb, int src_stride_argb,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height);
-
-// Convert ARGB To ARGB1555.
-LIBYUV_API
-int ARGBToARGB1555(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb1555, int dst_stride_argb1555,
-                   int width, int height);
-
-// Convert ARGB To ARGB4444.
-LIBYUV_API
-int ARGBToARGB4444(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb4444, int dst_stride_argb4444,
-                   int width, int height);
-
-// Convert ARGB To I444.
-LIBYUV_API
-int ARGBToI444(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert ARGB To I422.
-LIBYUV_API
-int ARGBToI422(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert ARGB To I420. (also in convert.h)
-LIBYUV_API
-int ARGBToI420(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert ARGB to J420. (JPeg full range I420).
-LIBYUV_API
-int ARGBToJ420(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yj, int dst_stride_yj,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert ARGB To I411.
-LIBYUV_API
-int ARGBToI411(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert ARGB to J400. (JPeg full range).
-LIBYUV_API
-int ARGBToJ400(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yj, int dst_stride_yj,
-               int width, int height);
-
-// Convert ARGB to I400.
-LIBYUV_API
-int ARGBToI400(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height);
-
-// Convert ARGB To NV12.
-LIBYUV_API
-int ARGBToNV12(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height);
-
-// Convert ARGB To NV21.
-LIBYUV_API
-int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_vu, int dst_stride_vu,
-               int width, int height);
-
-// Convert ARGB To NV21.
-LIBYUV_API
-int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_vu, int dst_stride_vu,
-               int width, int height);
-
-// Convert ARGB To YUY2.
-LIBYUV_API
-int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yuy2, int dst_stride_yuy2,
-               int width, int height);
-
-// Convert ARGB To UYVY.
-LIBYUV_API
-int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_uyvy, int dst_stride_uyvy,
-               int width, int height);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_CONVERT_FROM_ARGB_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/cpu_id.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/cpu_id.h
deleted file mode 100755
index dc858a814a..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/cpu_id.h
+++ /dev/null
@@ -1,81 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_CPU_ID_H_  // NOLINT
-#define INCLUDE_LIBYUV_CPU_ID_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// TODO(fbarchard): Consider overlapping bits for different architectures.
-// Internal flag to indicate cpuid requires initialization.
-#define kCpuInit 0x1
-
-// These flags are only valid on ARM processors.
-static const int kCpuHasARM = 0x2;
-static const int kCpuHasNEON = 0x4;
-// 0x8 reserved for future ARM flag.
-
-// These flags are only valid on x86 processors.
-static const int kCpuHasX86 = 0x10;
-static const int kCpuHasSSE2 = 0x20;
-static const int kCpuHasSSSE3 = 0x40;
-static const int kCpuHasSSE41 = 0x80;
-static const int kCpuHasSSE42 = 0x100;
-static const int kCpuHasAVX = 0x200;
-static const int kCpuHasAVX2 = 0x400;
-static const int kCpuHasERMS = 0x800;
-static const int kCpuHasFMA3 = 0x1000;
-// 0x2000, 0x4000, 0x8000 reserved for future X86 flags.
-
-// These flags are only valid on MIPS processors.
-static const int kCpuHasMIPS = 0x10000;
-static const int kCpuHasMIPS_DSP = 0x20000;
-static const int kCpuHasMIPS_DSPR2 = 0x40000;
-
-// Internal function used to auto-init.
-LIBYUV_API
-int InitCpuFlags(void);
-
-// Internal function for parsing /proc/cpuinfo.
-LIBYUV_API
-int ArmCpuCaps(const char* cpuinfo_name);
-
-// Detect CPU has SSE2 etc.
-// Test_flag parameter should be one of kCpuHas constants above.
-// returns non-zero if instruction set is detected
-static __inline int TestCpuFlag(int test_flag) {
-  LIBYUV_API extern int cpu_info_;
-  return (cpu_info_ == kCpuInit ? InitCpuFlags() : cpu_info_) & test_flag;
-}
-
-// For testing, allow CPU flags to be disabled.
-// ie MaskCpuFlags(~kCpuHasSSSE3) to disable SSSE3.
-// MaskCpuFlags(-1) to enable all cpu specific optimizations.
-// MaskCpuFlags(0) to disable all cpu specific optimizations.
-LIBYUV_API
-void MaskCpuFlags(int enable_flags);
-
-// Low level cpuid for X86. Returns zeros on other CPUs.
-// eax is the info type that you want.
-// ecx is typically the cpu number, and should normally be zero.
-LIBYUV_API
-void CpuId(uint32 eax, uint32 ecx, uint32* cpu_info);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_CPU_ID_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/format_conversion.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/format_conversion.h
deleted file mode 100755
index b18bf05343..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/format_conversion.h
+++ /dev/null
@@ -1,168 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_FORMATCONVERSION_H_  // NOLINT
-#define INCLUDE_LIBYUV_FORMATCONVERSION_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Convert Bayer RGB formats to I420.
-LIBYUV_API
-int BayerBGGRToI420(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_y, int dst_stride_y,
-                    uint8* dst_u, int dst_stride_u,
-                    uint8* dst_v, int dst_stride_v,
-                    int width, int height);
-
-LIBYUV_API
-int BayerGBRGToI420(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_y, int dst_stride_y,
-                    uint8* dst_u, int dst_stride_u,
-                    uint8* dst_v, int dst_stride_v,
-                    int width, int height);
-
-LIBYUV_API
-int BayerGRBGToI420(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_y, int dst_stride_y,
-                    uint8* dst_u, int dst_stride_u,
-                    uint8* dst_v, int dst_stride_v,
-                    int width, int height);
-
-LIBYUV_API
-int BayerRGGBToI420(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_y, int dst_stride_y,
-                    uint8* dst_u, int dst_stride_u,
-                    uint8* dst_v, int dst_stride_v,
-                    int width, int height);
-
-// Temporary API mapper.
-#define BayerRGBToI420(b, bs, f, y, ys, u, us, v, vs, w, h) \
-    BayerToI420(b, bs, y, ys, u, us, v, vs, w, h, f)
-
-LIBYUV_API
-int BayerToI420(const uint8* src_bayer, int src_stride_bayer,
-                uint8* dst_y, int dst_stride_y,
-                uint8* dst_u, int dst_stride_u,
-                uint8* dst_v, int dst_stride_v,
-                int width, int height,
-                uint32 src_fourcc_bayer);
-
-// Convert I420 to Bayer RGB formats.
-LIBYUV_API
-int I420ToBayerBGGR(const uint8* src_y, int src_stride_y,
-                    const uint8* src_u, int src_stride_u,
-                    const uint8* src_v, int src_stride_v,
-                    uint8* dst_frame, int dst_stride_frame,
-                    int width, int height);
-
-LIBYUV_API
-int I420ToBayerGBRG(const uint8* src_y, int src_stride_y,
-                    const uint8* src_u, int src_stride_u,
-                    const uint8* src_v, int src_stride_v,
-                    uint8* dst_frame, int dst_stride_frame,
-                    int width, int height);
-
-LIBYUV_API
-int I420ToBayerGRBG(const uint8* src_y, int src_stride_y,
-                    const uint8* src_u, int src_stride_u,
-                    const uint8* src_v, int src_stride_v,
-                    uint8* dst_frame, int dst_stride_frame,
-                    int width, int height);
-
-LIBYUV_API
-int I420ToBayerRGGB(const uint8* src_y, int src_stride_y,
-                    const uint8* src_u, int src_stride_u,
-                    const uint8* src_v, int src_stride_v,
-                    uint8* dst_frame, int dst_stride_frame,
-                    int width, int height);
-
-// Temporary API mapper.
-#define I420ToBayerRGB(y, ys, u, us, v, vs, b, bs, f, w, h) \
-    I420ToBayer(y, ys, u, us, v, vs, b, bs, w, h, f)
-
-LIBYUV_API
-int I420ToBayer(const uint8* src_y, int src_stride_y,
-                const uint8* src_u, int src_stride_u,
-                const uint8* src_v, int src_stride_v,
-                uint8* dst_frame, int dst_stride_frame,
-                int width, int height,
-                uint32 dst_fourcc_bayer);
-
-// Convert Bayer RGB formats to ARGB.
-LIBYUV_API
-int BayerBGGRToARGB(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height);
-
-LIBYUV_API
-int BayerGBRGToARGB(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height);
-
-LIBYUV_API
-int BayerGRBGToARGB(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height);
-
-LIBYUV_API
-int BayerRGGBToARGB(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height);
-
-// Temporary API mapper.
-#define BayerRGBToARGB(b, bs, f, a, as, w, h) BayerToARGB(b, bs, a, as, w, h, f)
-
-LIBYUV_API
-int BayerToARGB(const uint8* src_bayer, int src_stride_bayer,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height,
-                uint32 src_fourcc_bayer);
-
-// Converts ARGB to Bayer RGB formats.
-LIBYUV_API
-int ARGBToBayerBGGR(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_bayer, int dst_stride_bayer,
-                    int width, int height);
-
-LIBYUV_API
-int ARGBToBayerGBRG(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_bayer, int dst_stride_bayer,
-                    int width, int height);
-
-LIBYUV_API
-int ARGBToBayerGRBG(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_bayer, int dst_stride_bayer,
-                    int width, int height);
-
-LIBYUV_API
-int ARGBToBayerRGGB(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_bayer, int dst_stride_bayer,
-                    int width, int height);
-
-// Temporary API mapper.
-#define ARGBToBayerRGB(a, as, b, bs, f, w, h) ARGBToBayer(b, bs, a, as, w, h, f)
-
-LIBYUV_API
-int ARGBToBayer(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_bayer, int dst_stride_bayer,
-                int width, int height,
-                uint32 dst_fourcc_bayer);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_FORMATCONVERSION_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/mjpeg_decoder.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/mjpeg_decoder.h
deleted file mode 100755
index faffaea8fa..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/mjpeg_decoder.h
+++ /dev/null
@@ -1,201 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_MJPEG_DECODER_H_  // NOLINT
-#define INCLUDE_LIBYUV_MJPEG_DECODER_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-// NOTE: For a simplified public API use convert.h MJPGToI420().
-
-struct jpeg_common_struct;
-struct jpeg_decompress_struct;
-struct jpeg_source_mgr;
-
-namespace libyuv {
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-LIBYUV_BOOL ValidateJpeg(const uint8* sample, size_t sample_size);
-
-#ifdef __cplusplus
-}  // extern "C"
-#endif
-
-static const uint32 kUnknownDataSize = 0xFFFFFFFF;
-
-enum JpegSubsamplingType {
-  kJpegYuv420,
-  kJpegYuv422,
-  kJpegYuv411,
-  kJpegYuv444,
-  kJpegYuv400,
-  kJpegUnknown
-};
-
-struct SetJmpErrorMgr;
-
-// MJPEG ("Motion JPEG") is a pseudo-standard video codec where the frames are
-// simply independent JPEG images with a fixed huffman table (which is omitted).
-// It is rarely used in video transmission, but is common as a camera capture
-// format, especially in Logitech devices. This class implements a decoder for
-// MJPEG frames.
-//
-// See http://tools.ietf.org/html/rfc2435
-class LIBYUV_API MJpegDecoder {
- public:
-  typedef void (*CallbackFunction)(void* opaque,
-                                   const uint8* const* data,
-                                   const int* strides,
-                                   int rows);
-
-  static const int kColorSpaceUnknown;
-  static const int kColorSpaceGrayscale;
-  static const int kColorSpaceRgb;
-  static const int kColorSpaceYCbCr;
-  static const int kColorSpaceCMYK;
-  static const int kColorSpaceYCCK;
-
-  MJpegDecoder();
-  ~MJpegDecoder();
-
-  // Loads a new frame, reads its headers, and determines the uncompressed
-  // image format.
-  // Returns LIBYUV_TRUE if image looks valid and format is supported.
-  // If return value is LIBYUV_TRUE, then the values for all the following
-  // getters are populated.
-  // src_len is the size of the compressed mjpeg frame in bytes.
-  LIBYUV_BOOL LoadFrame(const uint8* src, size_t src_len);
-
-  // Returns width of the last loaded frame in pixels.
-  int GetWidth();
-
-  // Returns height of the last loaded frame in pixels.
-  int GetHeight();
-
-  // Returns format of the last loaded frame. The return value is one of the
-  // kColorSpace* constants.
-  int GetColorSpace();
-
-  // Number of color components in the color space.
-  int GetNumComponents();
-
-  // Sample factors of the n-th component.
-  int GetHorizSampFactor(int component);
-
-  int GetVertSampFactor(int component);
-
-  int GetHorizSubSampFactor(int component);
-
-  int GetVertSubSampFactor(int component);
-
-  // Public for testability.
-  int GetImageScanlinesPerImcuRow();
-
-  // Public for testability.
-  int GetComponentScanlinesPerImcuRow(int component);
-
-  // Width of a component in bytes.
-  int GetComponentWidth(int component);
-
-  // Height of a component.
-  int GetComponentHeight(int component);
-
-  // Width of a component in bytes with padding for DCTSIZE. Public for testing.
-  int GetComponentStride(int component);
-
-  // Size of a component in bytes.
-  int GetComponentSize(int component);
-
-  // Call this after LoadFrame() if you decide you don't want to decode it
-  // after all.
-  LIBYUV_BOOL UnloadFrame();
-
-  // Decodes the entire image into a one-buffer-per-color-component format.
-  // dst_width must match exactly. dst_height must be <= to image height; if
-  // less, the image is cropped. "planes" must have size equal to at least
-  // GetNumComponents() and they must point to non-overlapping buffers of size
-  // at least GetComponentSize(i). The pointers in planes are incremented
-  // to point to after the end of the written data.
-  // TODO(fbarchard): Add dst_x, dst_y to allow specific rect to be decoded.
-  LIBYUV_BOOL DecodeToBuffers(uint8** planes, int dst_width, int dst_height);
-
-  // Decodes the entire image and passes the data via repeated calls to a
-  // callback function. Each call will get the data for a whole number of
-  // image scanlines.
-  // TODO(fbarchard): Add dst_x, dst_y to allow specific rect to be decoded.
-  LIBYUV_BOOL DecodeToCallback(CallbackFunction fn, void* opaque,
-                        int dst_width, int dst_height);
-
-  // The helper function which recognizes the jpeg sub-sampling type.
-  static JpegSubsamplingType JpegSubsamplingTypeHelper(
-     int* subsample_x, int* subsample_y, int number_of_components);
-
- private:
-  struct Buffer {
-    const uint8* data;
-    int len;
-  };
-
-  struct BufferVector {
-    Buffer* buffers;
-    int len;
-    int pos;
-  };
-
-  // Methods that are passed to jpeglib.
-  static int fill_input_buffer(jpeg_decompress_struct* cinfo);
-  static void init_source(jpeg_decompress_struct* cinfo);
-  static void skip_input_data(jpeg_decompress_struct* cinfo,
-                              long num_bytes);  // NOLINT
-  static void term_source(jpeg_decompress_struct* cinfo);
-
-  static void ErrorHandler(jpeg_common_struct* cinfo);
-
-  void AllocOutputBuffers(int num_outbufs);
-  void DestroyOutputBuffers();
-
-  LIBYUV_BOOL StartDecode();
-  LIBYUV_BOOL FinishDecode();
-
-  void SetScanlinePointers(uint8** data);
-  LIBYUV_BOOL DecodeImcuRow();
-
-  int GetComponentScanlinePadding(int component);
-
-  // A buffer holding the input data for a frame.
-  Buffer buf_;
-  BufferVector buf_vec_;
-
-  jpeg_decompress_struct* decompress_struct_;
-  jpeg_source_mgr* source_mgr_;
-  SetJmpErrorMgr* error_mgr_;
-
-  // LIBYUV_TRUE iff at least one component has scanline padding. (i.e.,
-  // GetComponentScanlinePadding() != 0.)
-  LIBYUV_BOOL has_scanline_padding_;
-
-  // Temporaries used to point to scanline outputs.
-  int num_outbufs_;  // Outermost size of all arrays below.
-  uint8*** scanlines_;
-  int* scanlines_sizes_;
-  // Temporary buffer used for decoding when we can't decode directly to the
-  // output buffers. Large enough for just one iMCU row.
-  uint8** databuf_;
-  int* databuf_strides_;
-};
-
-}  // namespace libyuv
-
-#endif  //  __cplusplus
-#endif  // INCLUDE_LIBYUV_MJPEG_DECODER_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/planar_functions.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/planar_functions.h
deleted file mode 100755
index ac516c5ba5..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/planar_functions.h
+++ /dev/null
@@ -1,434 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_PLANAR_FUNCTIONS_H_  // NOLINT
-#define INCLUDE_LIBYUV_PLANAR_FUNCTIONS_H_
-
-#include "libyuv/basic_types.h"
-
-// TODO(fbarchard): Remove the following headers includes.
-#include "libyuv/convert.h"
-#include "libyuv/convert_argb.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Copy a plane of data.
-LIBYUV_API
-void CopyPlane(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height);
-
-// Set a plane of data to a 32 bit value.
-LIBYUV_API
-void SetPlane(uint8* dst_y, int dst_stride_y,
-              int width, int height,
-              uint32 value);
-
-// Copy I400.  Supports inverting.
-LIBYUV_API
-int I400ToI400(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height);
-
-
-// Copy I422 to I422.
-#define I422ToI422 I422Copy
-LIBYUV_API
-int I422Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height);
-
-// Copy I444 to I444.
-#define I444ToI444 I444Copy
-LIBYUV_API
-int I444Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height);
-
-// Convert YUY2 to I422.
-LIBYUV_API
-int YUY2ToI422(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert UYVY to I422.
-LIBYUV_API
-int UYVYToI422(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Convert I420 to I400. (calls CopyPlane ignoring u/v).
-LIBYUV_API
-int I420ToI400(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height);
-
-// Alias
-#define I420ToI420Mirror I420Mirror
-
-// I420 mirror.
-LIBYUV_API
-int I420Mirror(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height);
-
-// Alias
-#define I400ToI400Mirror I400Mirror
-
-// I400 mirror.  A single plane is mirrored horizontally.
-// Pass negative height to achieve 180 degree rotation.
-LIBYUV_API
-int I400Mirror(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height);
-
-// Alias
-#define ARGBToARGBMirror ARGBMirror
-
-// ARGB mirror.
-LIBYUV_API
-int ARGBMirror(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Convert NV12 to RGB565.
-LIBYUV_API
-int NV12ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_uv, int src_stride_uv,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height);
-
-// Convert NV21 to RGB565.
-LIBYUV_API
-int NV21ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_uv, int src_stride_uv,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height);
-
-// I422ToARGB is in convert_argb.h
-// Convert I422 to BGRA.
-LIBYUV_API
-int I422ToBGRA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_bgra, int dst_stride_bgra,
-               int width, int height);
-
-// Convert I422 to ABGR.
-LIBYUV_API
-int I422ToABGR(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_abgr, int dst_stride_abgr,
-               int width, int height);
-
-// Convert I422 to RGBA.
-LIBYUV_API
-int I422ToRGBA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height);
-
-// Draw a rectangle into I420.
-LIBYUV_API
-int I420Rect(uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int x, int y, int width, int height,
-             int value_y, int value_u, int value_v);
-
-// Draw a rectangle into ARGB.
-LIBYUV_API
-int ARGBRect(uint8* dst_argb, int dst_stride_argb,
-             int x, int y, int width, int height, uint32 value);
-
-// Convert ARGB to gray scale ARGB.
-LIBYUV_API
-int ARGBGrayTo(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height);
-
-// Make a rectangle of ARGB gray scale.
-LIBYUV_API
-int ARGBGray(uint8* dst_argb, int dst_stride_argb,
-             int x, int y, int width, int height);
-
-// Make a rectangle of ARGB Sepia tone.
-LIBYUV_API
-int ARGBSepia(uint8* dst_argb, int dst_stride_argb,
-              int x, int y, int width, int height);
-
-// Apply a matrix rotation to each ARGB pixel.
-// matrix_argb is 4 signed ARGB values. -128 to 127 representing -2 to 2.
-// The first 4 coefficients apply to B, G, R, A and produce B of the output.
-// The next 4 coefficients apply to B, G, R, A and produce G of the output.
-// The next 4 coefficients apply to B, G, R, A and produce R of the output.
-// The last 4 coefficients apply to B, G, R, A and produce A of the output.
-LIBYUV_API
-int ARGBColorMatrix(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_argb, int dst_stride_argb,
-                    const int8* matrix_argb,
-                    int width, int height);
-
-// Deprecated. Use ARGBColorMatrix instead.
-// Apply a matrix rotation to each ARGB pixel.
-// matrix_argb is 3 signed ARGB values. -128 to 127 representing -1 to 1.
-// The first 4 coefficients apply to B, G, R, A and produce B of the output.
-// The next 4 coefficients apply to B, G, R, A and produce G of the output.
-// The last 4 coefficients apply to B, G, R, A and produce R of the output.
-LIBYUV_API
-int RGBColorMatrix(uint8* dst_argb, int dst_stride_argb,
-                   const int8* matrix_rgb,
-                   int x, int y, int width, int height);
-
-// Apply a color table each ARGB pixel.
-// Table contains 256 ARGB values.
-LIBYUV_API
-int ARGBColorTable(uint8* dst_argb, int dst_stride_argb,
-                   const uint8* table_argb,
-                   int x, int y, int width, int height);
-
-// Apply a color table each ARGB pixel but preserve destination alpha.
-// Table contains 256 ARGB values.
-LIBYUV_API
-int RGBColorTable(uint8* dst_argb, int dst_stride_argb,
-                  const uint8* table_argb,
-                  int x, int y, int width, int height);
-
-// Apply a luma/color table each ARGB pixel but preserve destination alpha.
-// Table contains 32768 values indexed by [Y][C] where 7 it 7 bit luma from
-// RGB (YJ style) and C is an 8 bit color component (R, G or B).
-LIBYUV_API
-int ARGBLumaColorTable(const uint8* src_argb, int src_stride_argb,
-                       uint8* dst_argb, int dst_stride_argb,
-                       const uint8* luma_rgb_table,
-                       int width, int height);
-
-// Apply a 3 term polynomial to ARGB values.
-// poly points to a 4x4 matrix.  The first row is constants.  The 2nd row is
-// coefficients for b, g, r and a.  The 3rd row is coefficients for b squared,
-// g squared, r squared and a squared.  The 4rd row is coefficients for b to
-// the 3, g to the 3, r to the 3 and a to the 3.  The values are summed and
-// result clamped to 0 to 255.
-// A polynomial approximation can be dirived using software such as 'R'.
-
-LIBYUV_API
-int ARGBPolynomial(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb, int dst_stride_argb,
-                   const float* poly,
-                   int width, int height);
-
-// Quantize a rectangle of ARGB. Alpha unaffected.
-// scale is a 16 bit fractional fixed point scaler between 0 and 65535.
-// interval_size should be a value between 1 and 255.
-// interval_offset should be a value between 0 and 255.
-LIBYUV_API
-int ARGBQuantize(uint8* dst_argb, int dst_stride_argb,
-                 int scale, int interval_size, int interval_offset,
-                 int x, int y, int width, int height);
-
-// Copy ARGB to ARGB.
-LIBYUV_API
-int ARGBCopy(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int width, int height);
-
-// Copy ARGB to ARGB.
-LIBYUV_API
-int ARGBCopyAlpha(const uint8* src_argb, int src_stride_argb,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int width, int height);
-
-// Copy ARGB to ARGB.
-LIBYUV_API
-int ARGBCopyYToAlpha(const uint8* src_y, int src_stride_y,
-                     uint8* dst_argb, int dst_stride_argb,
-                     int width, int height);
-
-typedef void (*ARGBBlendRow)(const uint8* src_argb0, const uint8* src_argb1,
-                             uint8* dst_argb, int width);
-
-// Get function to Alpha Blend ARGB pixels and store to destination.
-LIBYUV_API
-ARGBBlendRow GetARGBBlend();
-
-// Alpha Blend ARGB images and store to destination.
-// Alpha of destination is set to 255.
-LIBYUV_API
-int ARGBBlend(const uint8* src_argb0, int src_stride_argb0,
-              const uint8* src_argb1, int src_stride_argb1,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height);
-
-// Multiply ARGB image by ARGB image. Shifted down by 8. Saturates to 255.
-LIBYUV_API
-int ARGBMultiply(const uint8* src_argb0, int src_stride_argb0,
-                 const uint8* src_argb1, int src_stride_argb1,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height);
-
-// Add ARGB image with ARGB image. Saturates to 255.
-LIBYUV_API
-int ARGBAdd(const uint8* src_argb0, int src_stride_argb0,
-            const uint8* src_argb1, int src_stride_argb1,
-            uint8* dst_argb, int dst_stride_argb,
-            int width, int height);
-
-// Subtract ARGB image (argb1) from ARGB image (argb0). Saturates to 0.
-LIBYUV_API
-int ARGBSubtract(const uint8* src_argb0, int src_stride_argb0,
-                 const uint8* src_argb1, int src_stride_argb1,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height);
-
-// Convert I422 to YUY2.
-LIBYUV_API
-int I422ToYUY2(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_frame, int dst_stride_frame,
-               int width, int height);
-
-// Convert I422 to UYVY.
-LIBYUV_API
-int I422ToUYVY(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_frame, int dst_stride_frame,
-               int width, int height);
-
-// Convert unattentuated ARGB to preattenuated ARGB.
-LIBYUV_API
-int ARGBAttenuate(const uint8* src_argb, int src_stride_argb,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int width, int height);
-
-// Convert preattentuated ARGB to unattenuated ARGB.
-LIBYUV_API
-int ARGBUnattenuate(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height);
-
-// Convert MJPG to ARGB.
-LIBYUV_API
-int MJPGToARGB(const uint8* sample, size_t sample_size,
-               uint8* argb, int argb_stride,
-               int w, int h, int dw, int dh);
-
-// Internal function - do not call directly.
-// Computes table of cumulative sum for image where the value is the sum
-// of all values above and to the left of the entry. Used by ARGBBlur.
-LIBYUV_API
-int ARGBComputeCumulativeSum(const uint8* src_argb, int src_stride_argb,
-                             int32* dst_cumsum, int dst_stride32_cumsum,
-                             int width, int height);
-
-// Blur ARGB image.
-// dst_cumsum table of width * (height + 1) * 16 bytes aligned to
-//   16 byte boundary.
-// dst_stride32_cumsum is number of ints in a row (width * 4).
-// radius is number of pixels around the center.  e.g. 1 = 3x3. 2=5x5.
-// Blur is optimized for radius of 5 (11x11) or less.
-LIBYUV_API
-int ARGBBlur(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int32* dst_cumsum, int dst_stride32_cumsum,
-             int width, int height, int radius);
-
-// Multiply ARGB image by ARGB value.
-LIBYUV_API
-int ARGBShade(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height, uint32 value);
-
-// Interpolate between two ARGB images using specified amount of interpolation
-// (0 to 255) and store to destination.
-// 'interpolation' is specified as 8 bit fraction where 0 means 100% src_argb0
-// and 255 means 1% src_argb0 and 99% src_argb1.
-// Internally uses ARGBScale bilinear filtering.
-// Caveat: This function will write up to 16 bytes beyond the end of dst_argb.
-LIBYUV_API
-int ARGBInterpolate(const uint8* src_argb0, int src_stride_argb0,
-                    const uint8* src_argb1, int src_stride_argb1,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height, int interpolation);
-
-#if defined(__pnacl__) || defined(__CLR_VER) || defined(COVERAGE_ENABLED) || \
-    defined(TARGET_IPHONE_SIMULATOR)
-#define LIBYUV_DISABLE_X86
-#endif
-
-// Row functions for copying a pixels from a source with a slope to a row
-// of destination. Useful for scaling, rotation, mirror, texture mapping.
-LIBYUV_API
-void ARGBAffineRow_C(const uint8* src_argb, int src_argb_stride,
-                     uint8* dst_argb, const float* uv_dudv, int width);
-// The following are available on all x86 platforms:
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__))
-LIBYUV_API
-void ARGBAffineRow_SSE2(const uint8* src_argb, int src_argb_stride,
-                        uint8* dst_argb, const float* uv_dudv, int width);
-#define HAS_ARGBAFFINEROW_SSE2
-#endif  // LIBYUV_DISABLE_X86
-
-// Shuffle ARGB channel order.  e.g. BGRA to ARGB.
-// shuffler is 16 bytes and must be aligned.
-LIBYUV_API
-int ARGBShuffle(const uint8* src_bgra, int src_stride_bgra,
-                uint8* dst_argb, int dst_stride_argb,
-                const uint8* shuffler, int width, int height);
-
-// Sobel ARGB effect with planar output.
-LIBYUV_API
-int ARGBSobelToPlane(const uint8* src_argb, int src_stride_argb,
-                     uint8* dst_y, int dst_stride_y,
-                     int width, int height);
-
-// Sobel ARGB effect.
-LIBYUV_API
-int ARGBSobel(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height);
-
-// Sobel ARGB effect w/ Sobel X, Sobel, Sobel Y in ARGB.
-LIBYUV_API
-int ARGBSobelXY(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_PLANAR_FUNCTIONS_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate.h
deleted file mode 100755
index 8af60b8955..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate.h
+++ /dev/null
@@ -1,117 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_ROTATE_H_  // NOLINT
-#define INCLUDE_LIBYUV_ROTATE_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Supported rotation.
-typedef enum RotationMode {
-  kRotate0 = 0,  // No rotation.
-  kRotate90 = 90,  // Rotate 90 degrees clockwise.
-  kRotate180 = 180,  // Rotate 180 degrees.
-  kRotate270 = 270,  // Rotate 270 degrees clockwise.
-
-  // Deprecated.
-  kRotateNone = 0,
-  kRotateClockwise = 90,
-  kRotateCounterClockwise = 270,
-} RotationModeEnum;
-
-// Rotate I420 frame.
-LIBYUV_API
-int I420Rotate(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int src_width, int src_height, enum RotationMode mode);
-
-// Rotate NV12 input and store in I420.
-LIBYUV_API
-int NV12ToI420Rotate(const uint8* src_y, int src_stride_y,
-                     const uint8* src_uv, int src_stride_uv,
-                     uint8* dst_y, int dst_stride_y,
-                     uint8* dst_u, int dst_stride_u,
-                     uint8* dst_v, int dst_stride_v,
-                     int src_width, int src_height, enum RotationMode mode);
-
-// Rotate a plane by 0, 90, 180, or 270.
-LIBYUV_API
-int RotatePlane(const uint8* src, int src_stride,
-                uint8* dst, int dst_stride,
-                int src_width, int src_height, enum RotationMode mode);
-
-// Rotate planes by 90, 180, 270. Deprecated.
-LIBYUV_API
-void RotatePlane90(const uint8* src, int src_stride,
-                   uint8* dst, int dst_stride,
-                   int width, int height);
-
-LIBYUV_API
-void RotatePlane180(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height);
-
-LIBYUV_API
-void RotatePlane270(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height);
-
-LIBYUV_API
-void RotateUV90(const uint8* src, int src_stride,
-                uint8* dst_a, int dst_stride_a,
-                uint8* dst_b, int dst_stride_b,
-                int width, int height);
-
-// Rotations for when U and V are interleaved.
-// These functions take one input pointer and
-// split the data into two buffers while
-// rotating them. Deprecated.
-LIBYUV_API
-void RotateUV180(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height);
-
-LIBYUV_API
-void RotateUV270(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height);
-
-// The 90 and 270 functions are based on transposes.
-// Doing a transpose with reversing the read/write
-// order will result in a rotation by +- 90 degrees.
-// Deprecated.
-LIBYUV_API
-void TransposePlane(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height);
-
-LIBYUV_API
-void TransposeUV(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_ROTATE_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate_argb.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate_argb.h
deleted file mode 100755
index 660ff5573e..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/rotate_argb.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_ROTATE_ARGB_H_  // NOLINT
-#define INCLUDE_LIBYUV_ROTATE_ARGB_H_
-
-#include "libyuv/basic_types.h"
-#include "libyuv/rotate.h"  // For RotationMode.
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Rotate ARGB frame
-LIBYUV_API
-int ARGBRotate(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int src_width, int src_height, enum RotationMode mode);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_ROTATE_ARGB_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/row.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/row.h
deleted file mode 100755
index 757020da86..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/row.h
+++ /dev/null
@@ -1,1694 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_ROW_H_  // NOLINT
-#define INCLUDE_LIBYUV_ROW_H_
-
-#include <stdlib.h>  // For malloc.
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#define IS_ALIGNED(p, a) (!((uintptr_t)(p) & ((a) - 1)))
-
-#ifdef __cplusplus
-#define align_buffer_64(var, size)                                             \
-  uint8* var##_mem = reinterpret_cast<uint8*>(malloc((size) + 63));            \
-  uint8* var = reinterpret_cast<uint8*>                                        \
-      ((reinterpret_cast<intptr_t>(var##_mem) + 63) & ~63)
-#else
-#define align_buffer_64(var, size)                                             \
-  uint8* var##_mem = (uint8*)(malloc((size) + 63));               /* NOLINT */ \
-  uint8* var = (uint8*)(((intptr_t)(var##_mem) + 63) & ~63)       /* NOLINT */
-#endif
-
-#define free_aligned_buffer_64(var) \
-  free(var##_mem);  \
-  var = 0
-
-#if defined(__pnacl__) || defined(__CLR_VER) || defined(COVERAGE_ENABLED) || \
-    defined(TARGET_IPHONE_SIMULATOR)
-#define LIBYUV_DISABLE_X86
-#endif
-// True if compiling for SSSE3 as a requirement.
-#if defined(__SSSE3__) || (defined(_M_IX86_FP) && (_M_IX86_FP >= 3))
-#define LIBYUV_SSSE3_ONLY
-#endif
-
-// Enable for NaCL pepper 33 for bundle and AVX2 support.
-//  #define NEW_BINUTILS
-
-// The following are available on all x86 platforms:
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__))
-// Effects:
-#define HAS_ARGBADDROW_SSE2
-#define HAS_ARGBAFFINEROW_SSE2
-#define HAS_ARGBATTENUATEROW_SSSE3
-#define HAS_ARGBBLENDROW_SSSE3
-#define HAS_ARGBCOLORMATRIXROW_SSSE3
-#define HAS_ARGBCOLORTABLEROW_X86
-#define HAS_ARGBCOPYALPHAROW_SSE2
-#define HAS_ARGBCOPYYTOALPHAROW_SSE2
-#define HAS_ARGBGRAYROW_SSSE3
-#define HAS_ARGBLUMACOLORTABLEROW_SSSE3
-#define HAS_ARGBMIRRORROW_SSSE3
-#define HAS_ARGBMULTIPLYROW_SSE2
-#define HAS_ARGBPOLYNOMIALROW_SSE2
-#define HAS_ARGBQUANTIZEROW_SSE2
-#define HAS_ARGBSEPIAROW_SSSE3
-#define HAS_ARGBSHADEROW_SSE2
-#define HAS_ARGBSUBTRACTROW_SSE2
-#define HAS_ARGBTOUVROW_SSSE3
-#define HAS_ARGBUNATTENUATEROW_SSE2
-#define HAS_COMPUTECUMULATIVESUMROW_SSE2
-#define HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
-#define HAS_INTERPOLATEROW_SSE2
-#define HAS_INTERPOLATEROW_SSSE3
-#define HAS_RGBCOLORTABLEROW_X86
-#define HAS_SOBELROW_SSE2
-#define HAS_SOBELTOPLANEROW_SSE2
-#define HAS_SOBELXROW_SSE2
-#define HAS_SOBELXYROW_SSE2
-#define HAS_SOBELYROW_SSE2
-
-// Conversions:
-#define HAS_ABGRTOUVROW_SSSE3
-#define HAS_ABGRTOYROW_SSSE3
-#define HAS_ARGB1555TOARGBROW_SSE2
-#define HAS_ARGB4444TOARGBROW_SSE2
-#define HAS_ARGBSHUFFLEROW_SSE2
-#define HAS_ARGBSHUFFLEROW_SSSE3
-#define HAS_ARGBTOARGB1555ROW_SSE2
-#define HAS_ARGBTOARGB4444ROW_SSE2
-#define HAS_ARGBTOBAYERGGROW_SSE2
-#define HAS_ARGBTOBAYERROW_SSSE3
-#define HAS_ARGBTORAWROW_SSSE3
-#define HAS_ARGBTORGB24ROW_SSSE3
-#define HAS_ARGBTORGB565ROW_SSE2
-#define HAS_ARGBTOUV422ROW_SSSE3
-#define HAS_ARGBTOUV444ROW_SSSE3
-#define HAS_ARGBTOUVJROW_SSSE3
-#define HAS_ARGBTOYJROW_SSSE3
-#define HAS_ARGBTOYROW_SSSE3
-#define HAS_BGRATOUVROW_SSSE3
-#define HAS_BGRATOYROW_SSSE3
-#define HAS_COPYROW_ERMS
-#define HAS_COPYROW_SSE2
-#define HAS_COPYROW_X86
-#define HAS_HALFROW_SSE2
-#define HAS_I400TOARGBROW_SSE2
-#define HAS_I411TOARGBROW_SSSE3
-#define HAS_I422TOARGB1555ROW_SSSE3
-#define HAS_I422TOABGRROW_SSSE3
-#define HAS_I422TOARGB1555ROW_SSSE3
-#define HAS_I422TOARGB4444ROW_SSSE3
-#define HAS_I422TOARGBROW_SSSE3
-#define HAS_I422TOBGRAROW_SSSE3
-#define HAS_I422TORAWROW_SSSE3
-#define HAS_I422TORGB24ROW_SSSE3
-#define HAS_I422TORGB565ROW_SSSE3
-#define HAS_I422TORGBAROW_SSSE3
-#define HAS_I422TOUYVYROW_SSE2
-#define HAS_I422TOYUY2ROW_SSE2
-#define HAS_I444TOARGBROW_SSSE3
-#define HAS_MERGEUVROW_SSE2
-#define HAS_MIRRORROW_SSE2
-#define HAS_MIRRORROW_SSSE3
-#define HAS_MIRRORROW_UV_SSSE3
-#define HAS_MIRRORUVROW_SSSE3
-#define HAS_NV12TOARGBROW_SSSE3
-#define HAS_NV12TORGB565ROW_SSSE3
-#define HAS_NV21TOARGBROW_SSSE3
-#define HAS_NV21TORGB565ROW_SSSE3
-#define HAS_RAWTOARGBROW_SSSE3
-#define HAS_RAWTOYROW_SSSE3
-#define HAS_RGB24TOARGBROW_SSSE3
-#define HAS_RGB24TOYROW_SSSE3
-#define HAS_RGB565TOARGBROW_SSE2
-#define HAS_RGBATOUVROW_SSSE3
-#define HAS_RGBATOYROW_SSSE3
-#define HAS_SETROW_X86
-#define HAS_SPLITUVROW_SSE2
-#define HAS_UYVYTOARGBROW_SSSE3
-#define HAS_UYVYTOUV422ROW_SSE2
-#define HAS_UYVYTOUVROW_SSE2
-#define HAS_UYVYTOYROW_SSE2
-#define HAS_YTOARGBROW_SSE2
-#define HAS_YUY2TOARGBROW_SSSE3
-#define HAS_YUY2TOUV422ROW_SSE2
-#define HAS_YUY2TOUVROW_SSE2
-#define HAS_YUY2TOYROW_SSE2
-#endif
-
-// GCC >= 4.7.0 required for AVX2.
-#if defined(__GNUC__) && (defined(__x86_64__) || defined(__i386__))
-#if (__GNUC__ > 4) || (__GNUC__ == 4 && (__GNUC_MINOR__ >= 7))
-#define GCC_HAS_AVX2 1
-#endif  // GNUC >= 4.7
-#endif  // __GNUC__
-
-// clang >= 3.4.0 required for AVX2.
-#if defined(__clang__) && (defined(__x86_64__) || defined(__i386__))
-#if (__clang_major__ > 3) || (__clang_major__ == 3 && (__clang_minor__ >= 4))
-#define CLANG_HAS_AVX2 1
-#endif  // clang >= 3.4
-#endif  // __clang__
-
-// Visual C 2012 required for AVX2.
-#if defined(_M_IX86) && defined(_MSC_VER) && _MSC_VER >= 1700
-#define VISUALC_HAS_AVX2 1
-#endif  // VisualStudio >= 2012
-
-// The following are available on all x86 platforms, but
-// require VS2012, clang 3.4 or gcc 4.7.
-// The code supports NaCL but requires a new compiler and validator.
-#if !defined(LIBYUV_DISABLE_X86) && (defined(VISUALC_HAS_AVX2) || \
-    defined(CLANG_HAS_AVX2) || defined(GCC_HAS_AVX2))
-// Effects:
-#define HAS_ARGBPOLYNOMIALROW_AVX2
-#define HAS_ARGBSHUFFLEROW_AVX2
-#define HAS_ARGBCOPYALPHAROW_AVX2
-#define HAS_ARGBCOPYYTOALPHAROW_AVX2
-#endif
-
-// The following are require VS2012.
-// TODO(fbarchard): Port to gcc.
-#if !defined(LIBYUV_DISABLE_X86) && defined(VISUALC_HAS_AVX2)
-#define HAS_ARGBTOUVROW_AVX2
-#define HAS_ARGBTOYJROW_AVX2
-#define HAS_ARGBTOYROW_AVX2
-#define HAS_HALFROW_AVX2
-#define HAS_I422TOARGBROW_AVX2
-#define HAS_INTERPOLATEROW_AVX2
-#define HAS_MERGEUVROW_AVX2
-#define HAS_MIRRORROW_AVX2
-#define HAS_SPLITUVROW_AVX2
-#define HAS_UYVYTOUV422ROW_AVX2
-#define HAS_UYVYTOUVROW_AVX2
-#define HAS_UYVYTOYROW_AVX2
-#define HAS_YUY2TOUV422ROW_AVX2
-#define HAS_YUY2TOUVROW_AVX2
-#define HAS_YUY2TOYROW_AVX2
-
-// Effects:
-#define HAS_ARGBADDROW_AVX2
-#define HAS_ARGBATTENUATEROW_AVX2
-#define HAS_ARGBMIRRORROW_AVX2
-#define HAS_ARGBMULTIPLYROW_AVX2
-#define HAS_ARGBSUBTRACTROW_AVX2
-#define HAS_ARGBUNATTENUATEROW_AVX2
-#endif  // defined(VISUALC_HAS_AVX2)
-
-// The following are Yasm x86 only:
-// TODO(fbarchard): Port AVX2 to inline.
-#if !defined(LIBYUV_DISABLE_X86) && defined(HAVE_YASM)
-    (defined(_M_IX86) || defined(_M_X64) || \
-    defined(__x86_64__) || defined(__i386__))
-#define HAS_MERGEUVROW_AVX2
-#define HAS_MERGEUVROW_MMX
-#define HAS_SPLITUVROW_AVX2
-#define HAS_SPLITUVROW_MMX
-#define HAS_UYVYTOYROW_AVX2
-#define HAS_UYVYTOYROW_MMX
-#define HAS_YUY2TOYROW_AVX2
-#define HAS_YUY2TOYROW_MMX
-#endif
-
-// The following are disabled when SSSE3 is available:
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)) && \
-    !defined(LIBYUV_SSSE3_ONLY)
-#define HAS_ARGBBLENDROW_SSE2
-#define HAS_ARGBATTENUATEROW_SSE2
-#define HAS_MIRRORROW_SSE2
-#endif
-
-// The following are available on Neon platforms:
-#if !defined(LIBYUV_DISABLE_NEON) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-#define HAS_ABGRTOUVROW_NEON
-#define HAS_ABGRTOYROW_NEON
-#define HAS_ARGB1555TOARGBROW_NEON
-#define HAS_ARGB1555TOUVROW_NEON
-#define HAS_ARGB1555TOYROW_NEON
-#define HAS_ARGB4444TOARGBROW_NEON
-#define HAS_ARGB4444TOUVROW_NEON
-#define HAS_ARGB4444TOYROW_NEON
-#define HAS_ARGBTOARGB1555ROW_NEON
-#define HAS_ARGBTOARGB4444ROW_NEON
-#define HAS_ARGBTOBAYERROW_NEON
-#define HAS_ARGBTOBAYERGGROW_NEON
-#define HAS_ARGBTORAWROW_NEON
-#define HAS_ARGBTORGB24ROW_NEON
-#define HAS_ARGBTORGB565ROW_NEON
-#define HAS_ARGBTOUV411ROW_NEON
-#define HAS_ARGBTOUV422ROW_NEON
-#define HAS_ARGBTOUV444ROW_NEON
-#define HAS_ARGBTOUVROW_NEON
-#define HAS_ARGBTOUVJROW_NEON
-#define HAS_ARGBTOYROW_NEON
-#define HAS_ARGBTOYJROW_NEON
-#define HAS_BGRATOUVROW_NEON
-#define HAS_BGRATOYROW_NEON
-#define HAS_COPYROW_NEON
-#define HAS_HALFROW_NEON
-#define HAS_I400TOARGBROW_NEON
-#define HAS_I411TOARGBROW_NEON
-#define HAS_I422TOABGRROW_NEON
-#define HAS_I422TOARGB1555ROW_NEON
-#define HAS_I422TOARGB4444ROW_NEON
-#define HAS_I422TOARGBROW_NEON
-#define HAS_I422TOBGRAROW_NEON
-#define HAS_I422TORAWROW_NEON
-#define HAS_I422TORGB24ROW_NEON
-#define HAS_I422TORGB565ROW_NEON
-#define HAS_I422TORGBAROW_NEON
-#define HAS_I422TOUYVYROW_NEON
-#define HAS_I422TOYUY2ROW_NEON
-#define HAS_I444TOARGBROW_NEON
-#define HAS_MERGEUVROW_NEON
-#define HAS_MIRRORROW_NEON
-#define HAS_MIRRORUVROW_NEON
-#define HAS_NV12TOARGBROW_NEON
-#define HAS_NV12TORGB565ROW_NEON
-#define HAS_NV21TOARGBROW_NEON
-#define HAS_NV21TORGB565ROW_NEON
-#define HAS_RAWTOARGBROW_NEON
-#define HAS_RAWTOUVROW_NEON
-#define HAS_RAWTOYROW_NEON
-#define HAS_RGB24TOARGBROW_NEON
-#define HAS_RGB24TOUVROW_NEON
-#define HAS_RGB24TOYROW_NEON
-#define HAS_RGB565TOARGBROW_NEON
-#define HAS_RGB565TOUVROW_NEON
-#define HAS_RGB565TOYROW_NEON
-#define HAS_RGBATOUVROW_NEON
-#define HAS_RGBATOYROW_NEON
-#define HAS_SETROW_NEON
-#define HAS_SPLITUVROW_NEON
-#define HAS_UYVYTOARGBROW_NEON
-#define HAS_UYVYTOUV422ROW_NEON
-#define HAS_UYVYTOUVROW_NEON
-#define HAS_UYVYTOYROW_NEON
-#define HAS_YTOARGBROW_NEON
-#define HAS_YUY2TOARGBROW_NEON
-#define HAS_YUY2TOUV422ROW_NEON
-#define HAS_YUY2TOUVROW_NEON
-#define HAS_YUY2TOYROW_NEON
-
-// Effects:
-#define HAS_ARGBADDROW_NEON
-#define HAS_ARGBATTENUATEROW_NEON
-#define HAS_ARGBBLENDROW_NEON
-#define HAS_ARGBCOLORMATRIXROW_NEON
-#define HAS_ARGBGRAYROW_NEON
-#define HAS_ARGBMIRRORROW_NEON
-#define HAS_ARGBMULTIPLYROW_NEON
-#define HAS_ARGBQUANTIZEROW_NEON
-#define HAS_ARGBSEPIAROW_NEON
-#define HAS_ARGBSHADEROW_NEON
-#define HAS_ARGBSUBTRACTROW_NEON
-#define HAS_SOBELROW_NEON
-#define HAS_SOBELTOPLANEROW_NEON
-#define HAS_SOBELXYROW_NEON
-#define HAS_SOBELXROW_NEON
-#define HAS_SOBELYROW_NEON
-#define HAS_INTERPOLATEROW_NEON
-#endif
-
-// The following are available on Mips platforms:
-#if !defined(LIBYUV_DISABLE_MIPS) && defined(__mips__)
-#define HAS_COPYROW_MIPS
-#if defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-#define HAS_I422TOABGRROW_MIPS_DSPR2
-#define HAS_I422TOARGBROW_MIPS_DSPR2
-#define HAS_I422TOBGRAROW_MIPS_DSPR2
-#define HAS_INTERPOLATEROWS_MIPS_DSPR2
-#define HAS_MIRRORROW_MIPS_DSPR2
-#define HAS_MIRRORUVROW_MIPS_DSPR2
-#define HAS_SPLITUVROW_MIPS_DSPR2
-#endif
-#endif
-
-#if defined(_MSC_VER) && !defined(__CLR_VER)
-#define SIMD_ALIGNED(var) __declspec(align(16)) var
-typedef __declspec(align(16)) int16 vec16[8];
-typedef __declspec(align(16)) int32 vec32[4];
-typedef __declspec(align(16)) int8 vec8[16];
-typedef __declspec(align(16)) uint16 uvec16[8];
-typedef __declspec(align(16)) uint32 uvec32[4];
-typedef __declspec(align(16)) uint8 uvec8[16];
-typedef __declspec(align(32)) int16 lvec16[16];
-typedef __declspec(align(32)) int32 lvec32[8];
-typedef __declspec(align(32)) int8 lvec8[32];
-typedef __declspec(align(32)) uint16 ulvec16[16];
-typedef __declspec(align(32)) uint32 ulvec32[8];
-typedef __declspec(align(32)) uint8 ulvec8[32];
-
-#elif defined(__GNUC__)
-// Caveat GCC 4.2 to 4.7 have a known issue using vectors with const.
-#define SIMD_ALIGNED(var) var __attribute__((aligned(16)))
-typedef int16 __attribute__((vector_size(16))) vec16;
-typedef int32 __attribute__((vector_size(16))) vec32;
-typedef int8 __attribute__((vector_size(16))) vec8;
-typedef uint16 __attribute__((vector_size(16))) uvec16;
-typedef uint32 __attribute__((vector_size(16))) uvec32;
-typedef uint8 __attribute__((vector_size(16))) uvec8;
-#else
-#define SIMD_ALIGNED(var) var
-typedef int16 vec16[8];
-typedef int32 vec32[4];
-typedef int8 vec8[16];
-typedef uint16 uvec16[8];
-typedef uint32 uvec32[4];
-typedef uint8 uvec8[16];
-#endif
-
-#if defined(__APPLE__) || defined(__x86_64__) || defined(__llvm__)
-#define OMITFP
-#else
-#define OMITFP __attribute__((optimize("omit-frame-pointer")))
-#endif
-
-// NaCL macros for GCC x86 and x64.
-
-// TODO(nfullagar): When pepper_33 toolchain is distributed, default to
-// NEW_BINUTILS and remove all BUNDLEALIGN occurances.
-#if defined(__native_client__)
-#define LABELALIGN ".p2align 5\n"
-#else
-#define LABELALIGN ".p2align 2\n"
-#endif
-#if defined(__native_client__) && defined(__x86_64__)
-#if defined(NEW_BINUTILS)
-#define BUNDLELOCK ".bundle_lock\n"
-#define BUNDLEUNLOCK ".bundle_unlock\n"
-#define BUNDLEALIGN "\n"
-#else
-#define BUNDLELOCK "\n"
-#define BUNDLEUNLOCK "\n"
-#define BUNDLEALIGN ".p2align 5\n"
-#endif
-#define MEMACCESS(base) "%%nacl:(%%r15,%q" #base ")"
-#define MEMACCESS2(offset, base) "%%nacl:" #offset "(%%r15,%q" #base ")"
-#define MEMLEA(offset, base) #offset "(%q" #base ")"
-#define MEMLEA3(offset, index, scale) \
-    #offset "(,%q" #index "," #scale ")"
-#define MEMLEA4(offset, base, index, scale) \
-    #offset "(%q" #base ",%q" #index "," #scale ")"
-#define MEMMOVESTRING(s, d) "%%nacl:(%q" #s "),%%nacl:(%q" #d "), %%r15"
-#define MEMSTORESTRING(reg, d) "%%" #reg ",%%nacl:(%q" #d "), %%r15"
-#define MEMOPREG(opcode, offset, base, index, scale, reg) \
-    BUNDLELOCK \
-    "lea " #offset "(%q" #base ",%q" #index "," #scale "),%%r14d\n" \
-    #opcode " (%%r15,%%r14),%%" #reg "\n" \
-    BUNDLEUNLOCK
-#define MEMOPMEM(opcode, reg, offset, base, index, scale) \
-    BUNDLELOCK \
-    "lea " #offset "(%q" #base ",%q" #index "," #scale "),%%r14d\n" \
-    #opcode " %%" #reg ",(%%r15,%%r14)\n" \
-    BUNDLEUNLOCK
-#define MEMOPARG(opcode, offset, base, index, scale, arg) \
-    BUNDLELOCK \
-    "lea " #offset "(%q" #base ",%q" #index "," #scale "),%%r14d\n" \
-    #opcode " (%%r15,%%r14),%" #arg "\n" \
-    BUNDLEUNLOCK
-#else
-#define BUNDLEALIGN "\n"
-#define MEMACCESS(base) "(%" #base ")"
-#define MEMACCESS2(offset, base) #offset "(%" #base ")"
-#define MEMLEA(offset, base) #offset "(%" #base ")"
-#define MEMLEA3(offset, index, scale) \
-    #offset "(,%" #index "," #scale ")"
-#define MEMLEA4(offset, base, index, scale) \
-    #offset "(%" #base ",%" #index "," #scale ")"
-#define MEMMOVESTRING(s, d)
-#define MEMSTORESTRING(reg, d)
-#define MEMOPREG(opcode, offset, base, index, scale, reg) \
-    #opcode " " #offset "(%" #base ",%" #index "," #scale "),%%" #reg "\n"
-#define MEMOPMEM(opcode, reg, offset, base, index, scale) \
-    #opcode " %%" #reg ","#offset "(%" #base ",%" #index "," #scale ")\n"
-#define MEMOPARG(opcode, offset, base, index, scale, arg) \
-    #opcode " " #offset "(%" #base ",%" #index "," #scale "),%" #arg "\n"
-#endif
-
-void I444ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width);
-void I422ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width);
-void I411ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width);
-void I422ToBGRARow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_bgra,
-                        int width);
-void I422ToABGRRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_abgr,
-                        int width);
-void I422ToRGBARow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_rgba,
-                        int width);
-void I422ToRGB24Row_NEON(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_rgb24,
-                         int width);
-void I422ToRAWRow_NEON(const uint8* src_y,
-                       const uint8* src_u,
-                       const uint8* src_v,
-                       uint8* dst_raw,
-                       int width);
-void I422ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_u,
-                          const uint8* src_v,
-                          uint8* dst_rgb565,
-                          int width);
-void I422ToARGB1555Row_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb1555,
-                            int width);
-void I422ToARGB4444Row_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb4444,
-                            int width);
-void NV12ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_uv,
-                        uint8* dst_argb,
-                        int width);
-void NV21ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_vu,
-                        uint8* dst_argb,
-                        int width);
-void NV12ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_uv,
-                          uint8* dst_rgb565,
-                          int width);
-void NV21ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_vu,
-                          uint8* dst_rgb565,
-                          int width);
-void YUY2ToARGBRow_NEON(const uint8* src_yuy2,
-                        uint8* dst_argb,
-                        int width);
-void UYVYToARGBRow_NEON(const uint8* src_uyvy,
-                        uint8* dst_argb,
-                        int width);
-
-void ARGBToYRow_AVX2(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYRow_Any_AVX2(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_AVX2(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_Any_AVX2(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void BGRAToYRow_SSSE3(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_SSSE3(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_SSSE3(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_SSSE3(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_SSSE3(const uint8* src_raw, uint8* dst_y, int pix);
-void ARGBToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void BGRAToYRow_Unaligned_SSSE3(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_Unaligned_SSSE3(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_Unaligned_SSSE3(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_Unaligned_SSSE3(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_Unaligned_SSSE3(const uint8* src_raw, uint8* dst_y, int pix);
-void ARGBToYRow_NEON(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_NEON(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToUV444Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void ARGBToUV422Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void ARGBToUV411Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void ARGBToUVRow_NEON(const uint8* src_argb, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void ARGBToUVJRow_NEON(const uint8* src_argb, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int pix);
-void BGRAToUVRow_NEON(const uint8* src_bgra, int src_stride_bgra,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void ABGRToUVRow_NEON(const uint8* src_abgr, int src_stride_abgr,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void RGBAToUVRow_NEON(const uint8* src_rgba, int src_stride_rgba,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void RGB24ToUVRow_NEON(const uint8* src_rgb24, int src_stride_rgb24,
-                       uint8* dst_u, uint8* dst_v, int pix);
-void RAWToUVRow_NEON(const uint8* src_raw, int src_stride_raw,
-                     uint8* dst_u, uint8* dst_v, int pix);
-void RGB565ToUVRow_NEON(const uint8* src_rgb565, int src_stride_rgb565,
-                        uint8* dst_u, uint8* dst_v, int pix);
-void ARGB1555ToUVRow_NEON(const uint8* src_argb1555, int src_stride_argb1555,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void ARGB4444ToUVRow_NEON(const uint8* src_argb4444, int src_stride_argb4444,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void BGRAToYRow_NEON(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_NEON(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_NEON(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_NEON(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_NEON(const uint8* src_raw, uint8* dst_y, int pix);
-void RGB565ToYRow_NEON(const uint8* src_rgb565, uint8* dst_y, int pix);
-void ARGB1555ToYRow_NEON(const uint8* src_argb1555, uint8* dst_y, int pix);
-void ARGB4444ToYRow_NEON(const uint8* src_argb4444, uint8* dst_y, int pix);
-void ARGBToYRow_C(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_C(const uint8* src_argb, uint8* dst_y, int pix);
-void BGRAToYRow_C(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_C(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_C(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_C(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_C(const uint8* src_raw, uint8* dst_y, int pix);
-void RGB565ToYRow_C(const uint8* src_rgb565, uint8* dst_y, int pix);
-void ARGB1555ToYRow_C(const uint8* src_argb1555, uint8* dst_y, int pix);
-void ARGB4444ToYRow_C(const uint8* src_argb4444, uint8* dst_y, int pix);
-void ARGBToYRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_Any_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
-void BGRAToYRow_Any_SSSE3(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_Any_SSSE3(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_Any_SSSE3(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_Any_SSSE3(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_Any_SSSE3(const uint8* src_raw, uint8* dst_y, int pix);
-void ARGBToYRow_Any_NEON(const uint8* src_argb, uint8* dst_y, int pix);
-void ARGBToYJRow_Any_NEON(const uint8* src_argb, uint8* dst_y, int pix);
-void BGRAToYRow_Any_NEON(const uint8* src_bgra, uint8* dst_y, int pix);
-void ABGRToYRow_Any_NEON(const uint8* src_abgr, uint8* dst_y, int pix);
-void RGBAToYRow_Any_NEON(const uint8* src_rgba, uint8* dst_y, int pix);
-void RGB24ToYRow_Any_NEON(const uint8* src_rgb24, uint8* dst_y, int pix);
-void RAWToYRow_Any_NEON(const uint8* src_raw, uint8* dst_y, int pix);
-void RGB565ToYRow_Any_NEON(const uint8* src_rgb565, uint8* dst_y, int pix);
-void ARGB1555ToYRow_Any_NEON(const uint8* src_argb1555, uint8* dst_y, int pix);
-void ARGB4444ToYRow_Any_NEON(const uint8* src_argb4444, uint8* dst_y, int pix);
-
-void ARGBToUVRow_AVX2(const uint8* src_argb, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVRow_Any_AVX2(const uint8* src_argb, int src_stride_argb,
-                          uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVRow_SSSE3(const uint8* src_argb, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVJRow_SSSE3(const uint8* src_argb, int src_stride_argb,
-                        uint8* dst_u, uint8* dst_v, int width);
-void BGRAToUVRow_SSSE3(const uint8* src_bgra, int src_stride_bgra,
-                       uint8* dst_u, uint8* dst_v, int width);
-void ABGRToUVRow_SSSE3(const uint8* src_abgr, int src_stride_abgr,
-                       uint8* dst_u, uint8* dst_v, int width);
-void RGBAToUVRow_SSSE3(const uint8* src_rgba, int src_stride_rgba,
-                       uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVRow_Unaligned_SSSE3(const uint8* src_argb, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVJRow_Unaligned_SSSE3(const uint8* src_argb, int src_stride_argb,
-                                  uint8* dst_u, uint8* dst_v, int width);
-void BGRAToUVRow_Unaligned_SSSE3(const uint8* src_bgra, int src_stride_bgra,
-                                 uint8* dst_u, uint8* dst_v, int width);
-void ABGRToUVRow_Unaligned_SSSE3(const uint8* src_abgr, int src_stride_abgr,
-                                 uint8* dst_u, uint8* dst_v, int width);
-void RGBAToUVRow_Unaligned_SSSE3(const uint8* src_rgba, int src_stride_rgba,
-                                 uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVRow_Any_SSSE3(const uint8* src_argb, int src_stride_argb,
-                           uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVJRow_Any_SSSE3(const uint8* src_argb, int src_stride_argb,
-                            uint8* dst_u, uint8* dst_v, int width);
-void BGRAToUVRow_Any_SSSE3(const uint8* src_bgra, int src_stride_bgra,
-                           uint8* dst_u, uint8* dst_v, int width);
-void ABGRToUVRow_Any_SSSE3(const uint8* src_abgr, int src_stride_abgr,
-                           uint8* dst_u, uint8* dst_v, int width);
-void RGBAToUVRow_Any_SSSE3(const uint8* src_rgba, int src_stride_rgba,
-                           uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV444Row_Any_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                             int pix);
-void ARGBToUV422Row_Any_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                             int pix);
-void ARGBToUV411Row_Any_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                             int pix);
-void ARGBToUVRow_Any_NEON(const uint8* src_argb, int src_stride_argb,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void ARGBToUVJRow_Any_NEON(const uint8* src_argb, int src_stride_argb,
-                           uint8* dst_u, uint8* dst_v, int pix);
-void BGRAToUVRow_Any_NEON(const uint8* src_bgra, int src_stride_bgra,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void ABGRToUVRow_Any_NEON(const uint8* src_abgr, int src_stride_abgr,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void RGBAToUVRow_Any_NEON(const uint8* src_rgba, int src_stride_rgba,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void RGB24ToUVRow_Any_NEON(const uint8* src_rgb24, int src_stride_rgb24,
-                           uint8* dst_u, uint8* dst_v, int pix);
-void RAWToUVRow_Any_NEON(const uint8* src_raw, int src_stride_raw,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void RGB565ToUVRow_Any_NEON(const uint8* src_rgb565, int src_stride_rgb565,
-                            uint8* dst_u, uint8* dst_v, int pix);
-void ARGB1555ToUVRow_Any_NEON(const uint8* src_argb1555,
-                              int src_stride_argb1555,
-                              uint8* dst_u, uint8* dst_v, int pix);
-void ARGB4444ToUVRow_Any_NEON(const uint8* src_argb4444,
-                              int src_stride_argb4444,
-                              uint8* dst_u, uint8* dst_v, int pix);
-void ARGBToUVRow_C(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUVJRow_C(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_u, uint8* dst_v, int width);
-void BGRAToUVRow_C(const uint8* src_bgra, int src_stride_bgra,
-                   uint8* dst_u, uint8* dst_v, int width);
-void ABGRToUVRow_C(const uint8* src_abgr, int src_stride_abgr,
-                   uint8* dst_u, uint8* dst_v, int width);
-void RGBAToUVRow_C(const uint8* src_rgba, int src_stride_rgba,
-                   uint8* dst_u, uint8* dst_v, int width);
-void RGB24ToUVRow_C(const uint8* src_rgb24, int src_stride_rgb24,
-                    uint8* dst_u, uint8* dst_v, int width);
-void RAWToUVRow_C(const uint8* src_raw, int src_stride_raw,
-                  uint8* dst_u, uint8* dst_v, int width);
-void RGB565ToUVRow_C(const uint8* src_rgb565, int src_stride_rgb565,
-                     uint8* dst_u, uint8* dst_v, int width);
-void ARGB1555ToUVRow_C(const uint8* src_argb1555, int src_stride_argb1555,
-                       uint8* dst_u, uint8* dst_v, int width);
-void ARGB4444ToUVRow_C(const uint8* src_argb4444, int src_stride_argb4444,
-                       uint8* dst_u, uint8* dst_v, int width);
-
-void ARGBToUV444Row_SSSE3(const uint8* src_argb,
-                          uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb,
-                                    uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV444Row_Any_SSSE3(const uint8* src_argb,
-                              uint8* dst_u, uint8* dst_v, int width);
-
-void ARGBToUV422Row_SSSE3(const uint8* src_argb,
-                          uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV422Row_Unaligned_SSSE3(const uint8* src_argb,
-                                    uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV422Row_Any_SSSE3(const uint8* src_argb,
-                              uint8* dst_u, uint8* dst_v, int width);
-
-void ARGBToUV444Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV422Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width);
-void ARGBToUV411Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width);
-
-void MirrorRow_AVX2(const uint8* src, uint8* dst, int width);
-void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width);
-void MirrorRow_SSE2(const uint8* src, uint8* dst, int width);
-void MirrorRow_NEON(const uint8* src, uint8* dst, int width);
-void MirrorRow_MIPS_DSPR2(const uint8* src, uint8* dst, int width);
-void MirrorRow_C(const uint8* src, uint8* dst, int width);
-
-void MirrorUVRow_SSSE3(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                       int width);
-void MirrorUVRow_NEON(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                      int width);
-void MirrorUVRow_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                            int width);
-void MirrorUVRow_C(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                   int width);
-
-void ARGBMirrorRow_AVX2(const uint8* src, uint8* dst, int width);
-void ARGBMirrorRow_SSSE3(const uint8* src, uint8* dst, int width);
-void ARGBMirrorRow_NEON(const uint8* src, uint8* dst, int width);
-void ARGBMirrorRow_C(const uint8* src, uint8* dst, int width);
-
-void SplitUVRow_C(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix);
-void SplitUVRow_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix);
-void SplitUVRow_AVX2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix);
-void SplitUVRow_NEON(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix);
-void SplitUVRow_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                           int pix);
-void SplitUVRow_Unaligned_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                               int pix);
-void SplitUVRow_Unaligned_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u,
-                                     uint8* dst_v, int pix);
-void SplitUVRow_Any_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void SplitUVRow_Any_AVX2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void SplitUVRow_Any_NEON(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                         int pix);
-void SplitUVRow_Any_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                               int pix);
-
-void MergeUVRow_C(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                  int width);
-void MergeUVRow_SSE2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width);
-void MergeUVRow_AVX2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width);
-void MergeUVRow_NEON(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width);
-void MergeUVRow_Unaligned_SSE2(const uint8* src_u, const uint8* src_v,
-                               uint8* dst_uv, int width);
-void MergeUVRow_Any_SSE2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                         int width);
-void MergeUVRow_Any_AVX2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                         int width);
-void MergeUVRow_Any_NEON(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                         int width);
-
-void CopyRow_SSE2(const uint8* src, uint8* dst, int count);
-void CopyRow_ERMS(const uint8* src, uint8* dst, int count);
-void CopyRow_X86(const uint8* src, uint8* dst, int count);
-void CopyRow_NEON(const uint8* src, uint8* dst, int count);
-void CopyRow_MIPS(const uint8* src, uint8* dst, int count);
-void CopyRow_C(const uint8* src, uint8* dst, int count);
-
-void ARGBCopyAlphaRow_C(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBCopyAlphaRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBCopyAlphaRow_AVX2(const uint8* src_argb, uint8* dst_argb, int width);
-
-void ARGBCopyYToAlphaRow_C(const uint8* src_y, uint8* dst_argb, int width);
-void ARGBCopyYToAlphaRow_SSE2(const uint8* src_y, uint8* dst_argb, int width);
-void ARGBCopyYToAlphaRow_AVX2(const uint8* src_y, uint8* dst_argb, int width);
-
-void SetRow_X86(uint8* dst, uint32 v32, int count);
-void ARGBSetRows_X86(uint8* dst, uint32 v32, int width,
-                     int dst_stride, int height);
-void SetRow_NEON(uint8* dst, uint32 v32, int count);
-void ARGBSetRows_NEON(uint8* dst, uint32 v32, int width,
-                      int dst_stride, int height);
-void SetRow_C(uint8* dst, uint32 v32, int count);
-void ARGBSetRows_C(uint8* dst, uint32 v32, int width, int dst_stride,
-                   int height);
-
-// ARGBShufflers for BGRAToARGB etc.
-void ARGBShuffleRow_C(const uint8* src_argb, uint8* dst_argb,
-                      const uint8* shuffler, int pix);
-void ARGBShuffleRow_SSE2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix);
-void ARGBShuffleRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                          const uint8* shuffler, int pix);
-void ARGBShuffleRow_AVX2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix);
-void ARGBShuffleRow_NEON(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix);
-void ARGBShuffleRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                    const uint8* shuffler, int pix);
-void ARGBShuffleRow_Any_SSE2(const uint8* src_argb, uint8* dst_argb,
-                             const uint8* shuffler, int pix);
-void ARGBShuffleRow_Any_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                              const uint8* shuffler, int pix);
-void ARGBShuffleRow_Any_AVX2(const uint8* src_argb, uint8* dst_argb,
-                             const uint8* shuffler, int pix);
-void ARGBShuffleRow_Any_NEON(const uint8* src_argb, uint8* dst_argb,
-                             const uint8* shuffler, int pix);
-
-void RGB24ToARGBRow_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix);
-void RAWToARGBRow_SSSE3(const uint8* src_raw, uint8* dst_argb, int pix);
-void RGB565ToARGBRow_SSE2(const uint8* src_rgb565, uint8* dst_argb, int pix);
-void ARGB1555ToARGBRow_SSE2(const uint8* src_argb1555, uint8* dst_argb,
-                            int pix);
-void ARGB4444ToARGBRow_SSE2(const uint8* src_argb4444, uint8* dst_argb,
-                            int pix);
-
-void RGB24ToARGBRow_NEON(const uint8* src_rgb24, uint8* dst_argb, int pix);
-void RAWToARGBRow_NEON(const uint8* src_raw, uint8* dst_argb, int pix);
-void RGB565ToARGBRow_NEON(const uint8* src_rgb565, uint8* dst_argb, int pix);
-void ARGB1555ToARGBRow_NEON(const uint8* src_argb1555, uint8* dst_argb,
-                            int pix);
-void ARGB4444ToARGBRow_NEON(const uint8* src_argb4444, uint8* dst_argb,
-                            int pix);
-void RGB24ToARGBRow_C(const uint8* src_rgb24, uint8* dst_argb, int pix);
-void RAWToARGBRow_C(const uint8* src_raw, uint8* dst_argb, int pix);
-void RGB565ToARGBRow_C(const uint8* src_rgb, uint8* dst_argb, int pix);
-void ARGB1555ToARGBRow_C(const uint8* src_argb, uint8* dst_argb, int pix);
-void ARGB4444ToARGBRow_C(const uint8* src_argb, uint8* dst_argb, int pix);
-void RGB24ToARGBRow_Any_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix);
-void RAWToARGBRow_Any_SSSE3(const uint8* src_raw, uint8* dst_argb, int pix);
-void RGB565ToARGBRow_Any_SSE2(const uint8* src_rgb565, uint8* dst_argb,
-                              int pix);
-void ARGB1555ToARGBRow_Any_SSE2(const uint8* src_argb1555, uint8* dst_argb,
-                                int pix);
-void ARGB4444ToARGBRow_Any_SSE2(const uint8* src_argb4444, uint8* dst_argb,
-                                int pix);
-void RGB24ToARGBRow_Any_NEON(const uint8* src_rgb24, uint8* dst_argb, int pix);
-void RAWToARGBRow_Any_NEON(const uint8* src_raw, uint8* dst_argb, int pix);
-void RGB565ToARGBRow_Any_NEON(const uint8* src_rgb565, uint8* dst_argb,
-                              int pix);
-void ARGB1555ToARGBRow_Any_NEON(const uint8* src_argb1555, uint8* dst_argb,
-                                int pix);
-void ARGB4444ToARGBRow_Any_NEON(const uint8* src_argb4444, uint8* dst_argb,
-                                int pix);
-
-void ARGBToRGB24Row_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRAWRow_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB565Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB1555Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB4444Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-
-void ARGBToRGB24Row_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRAWRow_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB565Row_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB1555Row_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB4444Row_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-
-void ARGBToRGBARow_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB24Row_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRAWRow_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB565Row_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB1555Row_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB4444Row_C(const uint8* src_argb, uint8* dst_rgb, int pix);
-
-void I400ToARGBRow_SSE2(const uint8* src_y, uint8* dst_argb, int pix);
-void I400ToARGBRow_Unaligned_SSE2(const uint8* src_y, uint8* dst_argb, int pix);
-void I400ToARGBRow_NEON(const uint8* src_y, uint8* dst_argb, int pix);
-void I400ToARGBRow_C(const uint8* src_y, uint8* dst_argb, int pix);
-void I400ToARGBRow_Any_SSE2(const uint8* src_y, uint8* dst_argb, int pix);
-void I400ToARGBRow_Any_NEON(const uint8* src_y, uint8* dst_argb, int pix);
-
-void I444ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_argb,
-                     int width);
-void I422ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_argb,
-                     int width);
-void I411ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_argb,
-                     int width);
-void NV12ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_uv,
-                     uint8* dst_argb,
-                     int width);
-void NV21ToRGB565Row_C(const uint8* src_y,
-                       const uint8* src_vu,
-                       uint8* dst_argb,
-                       int width);
-void NV12ToRGB565Row_C(const uint8* src_y,
-                       const uint8* src_uv,
-                       uint8* dst_argb,
-                       int width);
-void NV21ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_vu,
-                     uint8* dst_argb,
-                     int width);
-void YUY2ToARGBRow_C(const uint8* src_yuy2,
-                     uint8* dst_argb,
-                     int width);
-void UYVYToARGBRow_C(const uint8* src_uyvy,
-                     uint8* dst_argb,
-                     int width);
-void I422ToBGRARow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_bgra,
-                     int width);
-void I422ToABGRRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_abgr,
-                     int width);
-void I422ToRGBARow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_rgba,
-                     int width);
-void I422ToRGB24Row_C(const uint8* src_y,
-                      const uint8* src_u,
-                      const uint8* src_v,
-                      uint8* dst_rgb24,
-                      int width);
-void I422ToRAWRow_C(const uint8* src_y,
-                    const uint8* src_u,
-                    const uint8* src_v,
-                    uint8* dst_raw,
-                    int width);
-void I422ToARGB4444Row_C(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb4444,
-                         int width);
-void I422ToARGB1555Row_C(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb4444,
-                         int width);
-void I422ToRGB565Row_C(const uint8* src_y,
-                       const uint8* src_u,
-                       const uint8* src_v,
-                       uint8* dst_rgb565,
-                       int width);
-void YToARGBRow_C(const uint8* src_y,
-                  uint8* dst_argb,
-                  int width);
-void I422ToARGBRow_AVX2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width);
-void I444ToARGBRow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb,
-                         int width);
-void I422ToARGBRow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb,
-                         int width);
-void I411ToARGBRow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb,
-                         int width);
-void NV12ToARGBRow_SSSE3(const uint8* src_y,
-                         const uint8* src_uv,
-                         uint8* dst_argb,
-                         int width);
-void NV21ToARGBRow_SSSE3(const uint8* src_y,
-                         const uint8* src_vu,
-                         uint8* dst_argb,
-                         int width);
-void NV12ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_uv,
-                           uint8* dst_argb,
-                           int width);
-void NV21ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_vu,
-                           uint8* dst_argb,
-                           int width);
-void YUY2ToARGBRow_SSSE3(const uint8* src_yuy2,
-                         uint8* dst_argb,
-                         int width);
-void UYVYToARGBRow_SSSE3(const uint8* src_uyvy,
-                         uint8* dst_argb,
-                         int width);
-void I422ToBGRARow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_bgra,
-                         int width);
-void I422ToABGRRow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_abgr,
-                         int width);
-void I422ToRGBARow_SSSE3(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_rgba,
-                         int width);
-void I422ToARGB4444Row_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void I422ToARGB1555Row_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void I422ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_u,
-                           const uint8* src_v,
-                           uint8* dst_argb,
-                           int width);
-// RGB24/RAW are unaligned.
-void I422ToRGB24Row_SSSE3(const uint8* src_y,
-                          const uint8* src_u,
-                          const uint8* src_v,
-                          uint8* dst_rgb24,
-                          int width);
-void I422ToRAWRow_SSSE3(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_raw,
-                        int width);
-
-void I444ToARGBRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_argb,
-                                   int width);
-void I422ToARGBRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_argb,
-                                   int width);
-void I411ToARGBRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_argb,
-                                   int width);
-void NV12ToARGBRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_uv,
-                                   uint8* dst_argb,
-                                   int width);
-void NV21ToARGBRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_vu,
-                                   uint8* dst_argb,
-                                   int width);
-void YUY2ToARGBRow_Unaligned_SSSE3(const uint8* src_yuy2,
-                                   uint8* dst_argb,
-                                   int width);
-void UYVYToARGBRow_Unaligned_SSSE3(const uint8* src_uyvy,
-                                   uint8* dst_argb,
-                                   int width);
-void I422ToBGRARow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_bgra,
-                                   int width);
-void I422ToABGRRow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_abgr,
-                                   int width);
-void I422ToRGBARow_Unaligned_SSSE3(const uint8* src_y,
-                                   const uint8* src_u,
-                                   const uint8* src_v,
-                                   uint8* dst_rgba,
-                                   int width);
-void I422ToARGBRow_Any_AVX2(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I444ToARGBRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void I422ToARGBRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void I411ToARGBRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void NV12ToARGBRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_uv,
-                             uint8* dst_argb,
-                             int width);
-void NV21ToARGBRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_vu,
-                             uint8* dst_argb,
-                             int width);
-void NV12ToRGB565Row_Any_SSSE3(const uint8* src_y,
-                               const uint8* src_uv,
-                               uint8* dst_argb,
-                               int width);
-void NV21ToRGB565Row_Any_SSSE3(const uint8* src_y,
-                               const uint8* src_vu,
-                               uint8* dst_argb,
-                               int width);
-void YUY2ToARGBRow_Any_SSSE3(const uint8* src_yuy2,
-                             uint8* dst_argb,
-                             int width);
-void UYVYToARGBRow_Any_SSSE3(const uint8* src_uyvy,
-                             uint8* dst_argb,
-                             int width);
-void I422ToBGRARow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_bgra,
-                             int width);
-void I422ToABGRRow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_abgr,
-                             int width);
-void I422ToRGBARow_Any_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_rgba,
-                             int width);
-void I422ToARGB4444Row_Any_SSSE3(const uint8* src_y,
-                                 const uint8* src_u,
-                                 const uint8* src_v,
-                                 uint8* dst_rgba,
-                                 int width);
-void I422ToARGB1555Row_Any_SSSE3(const uint8* src_y,
-                                 const uint8* src_u,
-                                 const uint8* src_v,
-                                 uint8* dst_rgba,
-                                 int width);
-void I422ToRGB565Row_Any_SSSE3(const uint8* src_y,
-                               const uint8* src_u,
-                               const uint8* src_v,
-                               uint8* dst_rgba,
-                               int width);
-// RGB24/RAW are unaligned.
-void I422ToRGB24Row_Any_SSSE3(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToRAWRow_Any_SSSE3(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void YToARGBRow_SSE2(const uint8* src_y,
-                     uint8* dst_argb,
-                     int width);
-void YToARGBRow_NEON(const uint8* src_y,
-                     uint8* dst_argb,
-                     int width);
-void YToARGBRow_Any_SSE2(const uint8* src_y,
-                         uint8* dst_argb,
-                         int width);
-void YToARGBRow_Any_NEON(const uint8* src_y,
-                         uint8* dst_argb,
-                         int width);
-
-// ARGB preattenuated alpha blend.
-void ARGBBlendRow_SSSE3(const uint8* src_argb, const uint8* src_argb1,
-                        uint8* dst_argb, int width);
-void ARGBBlendRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width);
-void ARGBBlendRow_NEON(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width);
-void ARGBBlendRow_C(const uint8* src_argb, const uint8* src_argb1,
-                    uint8* dst_argb, int width);
-
-// ARGB multiply images. Same API as Blend, but these require
-// pointer and width alignment for SSE2.
-void ARGBMultiplyRow_C(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width);
-void ARGBMultiplyRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBMultiplyRow_Any_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-void ARGBMultiplyRow_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBMultiplyRow_Any_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-void ARGBMultiplyRow_NEON(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBMultiplyRow_Any_NEON(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-
-// ARGB add images.
-void ARGBAddRow_C(const uint8* src_argb, const uint8* src_argb1,
-                  uint8* dst_argb, int width);
-void ARGBAddRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                     uint8* dst_argb, int width);
-void ARGBAddRow_Any_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                         uint8* dst_argb, int width);
-void ARGBAddRow_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                     uint8* dst_argb, int width);
-void ARGBAddRow_Any_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                         uint8* dst_argb, int width);
-void ARGBAddRow_NEON(const uint8* src_argb, const uint8* src_argb1,
-                     uint8* dst_argb, int width);
-void ARGBAddRow_Any_NEON(const uint8* src_argb, const uint8* src_argb1,
-                         uint8* dst_argb, int width);
-
-// ARGB subtract images. Same API as Blend, but these require
-// pointer and width alignment for SSE2.
-void ARGBSubtractRow_C(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width);
-void ARGBSubtractRow_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBSubtractRow_Any_SSE2(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-void ARGBSubtractRow_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBSubtractRow_Any_AVX2(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-void ARGBSubtractRow_NEON(const uint8* src_argb, const uint8* src_argb1,
-                          uint8* dst_argb, int width);
-void ARGBSubtractRow_Any_NEON(const uint8* src_argb, const uint8* src_argb1,
-                              uint8* dst_argb, int width);
-
-void ARGBToRGB24Row_Any_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRAWRow_Any_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB565Row_Any_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB1555Row_Any_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB4444Row_Any_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix);
-
-void ARGBToRGB24Row_Any_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRAWRow_Any_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToRGB565Row_Any_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB1555Row_Any_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-void ARGBToARGB4444Row_Any_NEON(const uint8* src_argb, uint8* dst_rgb, int pix);
-
-void I444ToARGBRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I422ToARGBRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I411ToARGBRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I422ToBGRARow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I422ToABGRRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I422ToRGBARow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb,
-                            int width);
-void I422ToRGB24Row_Any_NEON(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* dst_argb,
-                             int width);
-void I422ToRAWRow_Any_NEON(const uint8* src_y,
-                           const uint8* src_u,
-                           const uint8* src_v,
-                           uint8* dst_argb,
-                           int width);
-void I422ToARGB4444Row_Any_NEON(const uint8* src_y,
-                                const uint8* src_u,
-                                const uint8* src_v,
-                                uint8* dst_argb,
-                                int width);
-void I422ToARGB1555Row_Any_NEON(const uint8* src_y,
-                                const uint8* src_u,
-                                const uint8* src_v,
-                                uint8* dst_argb,
-                                int width);
-void I422ToRGB565Row_Any_NEON(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void NV12ToARGBRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_uv,
-                            uint8* dst_argb,
-                            int width);
-void NV21ToARGBRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_uv,
-                            uint8* dst_argb,
-                            int width);
-void NV12ToRGB565Row_Any_NEON(const uint8* src_y,
-                              const uint8* src_uv,
-                              uint8* dst_argb,
-                              int width);
-void NV21ToRGB565Row_Any_NEON(const uint8* src_y,
-                              const uint8* src_uv,
-                              uint8* dst_argb,
-                              int width);
-void YUY2ToARGBRow_Any_NEON(const uint8* src_yuy2,
-                            uint8* dst_argb,
-                            int width);
-void UYVYToARGBRow_Any_NEON(const uint8* src_uyvy,
-                            uint8* dst_argb,
-                            int width);
-void I422ToARGBRow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToBGRARow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToABGRRow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToARGBRow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToBGRARow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-void I422ToABGRRow_MIPS_DSPR2(const uint8* src_y,
-                              const uint8* src_u,
-                              const uint8* src_v,
-                              uint8* dst_argb,
-                              int width);
-
-void YUY2ToYRow_AVX2(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_AVX2(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_AVX2(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_SSE2(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_SSE2(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_Unaligned_SSE2(const uint8* src_yuy2,
-                               uint8* dst_y, int pix);
-void YUY2ToUVRow_Unaligned_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                                uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_Unaligned_SSE2(const uint8* src_yuy2,
-                                   uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_NEON(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_NEON(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_NEON(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_C(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_C(const uint8* src_yuy2, int stride_yuy2,
-                   uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_C(const uint8* src_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_Any_AVX2(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_Any_AVX2(const uint8* src_yuy2, int stride_yuy2,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_Any_AVX2(const uint8* src_yuy2,
-                             uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_Any_SSE2(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_Any_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_Any_SSE2(const uint8* src_yuy2,
-                             uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToYRow_Any_NEON(const uint8* src_yuy2, uint8* dst_y, int pix);
-void YUY2ToUVRow_Any_NEON(const uint8* src_yuy2, int stride_yuy2,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void YUY2ToUV422Row_Any_NEON(const uint8* src_yuy2,
-                             uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_AVX2(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_AVX2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_AVX2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_SSE2(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_SSE2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_Unaligned_SSE2(const uint8* src_uyvy,
-                               uint8* dst_y, int pix);
-void UYVYToUVRow_Unaligned_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                                uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_Unaligned_SSE2(const uint8* src_uyvy,
-                                   uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_AVX2(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_AVX2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_AVX2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_NEON(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_NEON(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_NEON(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix);
-
-void UYVYToYRow_C(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_C(const uint8* src_uyvy, int stride_uyvy,
-                   uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_C(const uint8* src_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_Any_AVX2(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_Any_AVX2(const uint8* src_uyvy, int stride_uyvy,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_Any_AVX2(const uint8* src_uyvy,
-                             uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_Any_SSE2(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_Any_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_Any_SSE2(const uint8* src_uyvy,
-                             uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToYRow_Any_NEON(const uint8* src_uyvy, uint8* dst_y, int pix);
-void UYVYToUVRow_Any_NEON(const uint8* src_uyvy, int stride_uyvy,
-                          uint8* dst_u, uint8* dst_v, int pix);
-void UYVYToUV422Row_Any_NEON(const uint8* src_uyvy,
-                             uint8* dst_u, uint8* dst_v, int pix);
-
-void HalfRow_C(const uint8* src_uv, int src_uv_stride,
-               uint8* dst_uv, int pix);
-void HalfRow_SSE2(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix);
-void HalfRow_AVX2(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix);
-void HalfRow_NEON(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix);
-
-void ARGBToBayerRow_C(const uint8* src_argb, uint8* dst_bayer,
-                      uint32 selector, int pix);
-void ARGBToBayerRow_SSSE3(const uint8* src_argb, uint8* dst_bayer,
-                          uint32 selector, int pix);
-void ARGBToBayerRow_NEON(const uint8* src_argb, uint8* dst_bayer,
-                         uint32 selector, int pix);
-void ARGBToBayerRow_Any_SSSE3(const uint8* src_argb, uint8* dst_bayer,
-                              uint32 selector, int pix);
-void ARGBToBayerRow_Any_NEON(const uint8* src_argb, uint8* dst_bayer,
-                             uint32 selector, int pix);
-void ARGBToBayerGGRow_C(const uint8* src_argb, uint8* dst_bayer,
-                        uint32 /* selector */, int pix);
-void ARGBToBayerGGRow_SSE2(const uint8* src_argb, uint8* dst_bayer,
-                           uint32 /* selector */, int pix);
-void ARGBToBayerGGRow_NEON(const uint8* src_argb, uint8* dst_bayer,
-                           uint32 /* selector */, int pix);
-void ARGBToBayerGGRow_Any_SSE2(const uint8* src_argb, uint8* dst_bayer,
-                               uint32 /* selector */, int pix);
-void ARGBToBayerGGRow_Any_NEON(const uint8* src_argb, uint8* dst_bayer,
-                               uint32 /* selector */, int pix);
-
-void I422ToYUY2Row_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_yuy2, int width);
-void I422ToUYVYRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_uyvy, int width);
-void I422ToYUY2Row_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_yuy2, int width);
-void I422ToUYVYRow_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_uyvy, int width);
-void I422ToYUY2Row_Any_SSE2(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_yuy2, int width);
-void I422ToUYVYRow_Any_SSE2(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_uyvy, int width);
-void I422ToYUY2Row_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_yuy2, int width);
-void I422ToUYVYRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_uyvy, int width);
-void I422ToYUY2Row_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_yuy2, int width);
-void I422ToUYVYRow_Any_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_uyvy, int width);
-
-// Effects related row functions.
-void ARGBAttenuateRow_C(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_NEON(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBAttenuateRow_Any_SSE2(const uint8* src_argb, uint8* dst_argb,
-                               int width);
-void ARGBAttenuateRow_Any_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                int width);
-void ARGBAttenuateRow_Any_AVX2(const uint8* src_argb, uint8* dst_argb,
-                               int width);
-void ARGBAttenuateRow_Any_NEON(const uint8* src_argb, uint8* dst_argb,
-                               int width);
-
-// Inverse table for unattenuate, shared by C and SSE2.
-extern const uint32 fixed_invtbl8[256];
-void ARGBUnattenuateRow_C(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBUnattenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBUnattenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBUnattenuateRow_Any_SSE2(const uint8* src_argb, uint8* dst_argb,
-                                 int width);
-void ARGBUnattenuateRow_Any_AVX2(const uint8* src_argb, uint8* dst_argb,
-                                 int width);
-
-void ARGBGrayRow_C(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBGrayRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width);
-void ARGBGrayRow_NEON(const uint8* src_argb, uint8* dst_argb, int width);
-
-void ARGBSepiaRow_C(uint8* dst_argb, int width);
-void ARGBSepiaRow_SSSE3(uint8* dst_argb, int width);
-void ARGBSepiaRow_NEON(uint8* dst_argb, int width);
-
-void ARGBColorMatrixRow_C(const uint8* src_argb, uint8* dst_argb,
-                          const int8* matrix_argb, int width);
-void ARGBColorMatrixRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                              const int8* matrix_argb, int width);
-void ARGBColorMatrixRow_NEON(const uint8* src_argb, uint8* dst_argb,
-                             const int8* matrix_argb, int width);
-
-void ARGBColorTableRow_C(uint8* dst_argb, const uint8* table_argb, int width);
-void ARGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb, int width);
-
-void RGBColorTableRow_C(uint8* dst_argb, const uint8* table_argb, int width);
-void RGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb, int width);
-
-void ARGBQuantizeRow_C(uint8* dst_argb, int scale, int interval_size,
-                       int interval_offset, int width);
-void ARGBQuantizeRow_SSE2(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width);
-void ARGBQuantizeRow_NEON(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width);
-
-void ARGBShadeRow_C(const uint8* src_argb, uint8* dst_argb, int width,
-                    uint32 value);
-void ARGBShadeRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width,
-                       uint32 value);
-void ARGBShadeRow_NEON(const uint8* src_argb, uint8* dst_argb, int width,
-                       uint32 value);
-
-// Used for blur.
-void CumulativeSumToAverageRow_SSE2(const int32* topleft, const int32* botleft,
-                                    int width, int area, uint8* dst, int count);
-void ComputeCumulativeSumRow_SSE2(const uint8* row, int32* cumsum,
-                                  const int32* previous_cumsum, int width);
-
-void CumulativeSumToAverageRow_C(const int32* topleft, const int32* botleft,
-                                 int width, int area, uint8* dst, int count);
-void ComputeCumulativeSumRow_C(const uint8* row, int32* cumsum,
-                               const int32* previous_cumsum, int width);
-
-LIBYUV_API
-void ARGBAffineRow_C(const uint8* src_argb, int src_argb_stride,
-                     uint8* dst_argb, const float* uv_dudv, int width);
-LIBYUV_API
-void ARGBAffineRow_SSE2(const uint8* src_argb, int src_argb_stride,
-                        uint8* dst_argb, const float* uv_dudv, int width);
-
-// Used for I420Scale, ARGBScale, and ARGBInterpolate.
-void InterpolateRow_C(uint8* dst_ptr, const uint8* src_ptr,
-                      ptrdiff_t src_stride_ptr,
-                      int width, int source_y_fraction);
-void InterpolateRow_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride_ptr, int width,
-                         int source_y_fraction);
-void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                          ptrdiff_t src_stride_ptr, int width,
-                          int source_y_fraction);
-void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride_ptr, int width,
-                         int source_y_fraction);
-void InterpolateRow_NEON(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride_ptr, int width,
-                         int source_y_fraction);
-void InterpolateRows_MIPS_DSPR2(uint8* dst_ptr, const uint8* src_ptr,
-                                ptrdiff_t src_stride_ptr, int width,
-                                int source_y_fraction);
-void InterpolateRow_Unaligned_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                                   ptrdiff_t src_stride_ptr, int width,
-                                   int source_y_fraction);
-void InterpolateRow_Unaligned_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                                    ptrdiff_t src_stride_ptr, int width,
-                                    int source_y_fraction);
-void InterpolateRow_Any_NEON(uint8* dst_ptr, const uint8* src_ptr,
-                             ptrdiff_t src_stride_ptr, int width,
-                             int source_y_fraction);
-void InterpolateRow_Any_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                             ptrdiff_t src_stride_ptr, int width,
-                             int source_y_fraction);
-void InterpolateRow_Any_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                              ptrdiff_t src_stride_ptr, int width,
-                              int source_y_fraction);
-void InterpolateRow_Any_AVX2(uint8* dst_ptr, const uint8* src_ptr,
-                             ptrdiff_t src_stride_ptr, int width,
-                             int source_y_fraction);
-void InterpolateRows_Any_MIPS_DSPR2(uint8* dst_ptr, const uint8* src_ptr,
-                                    ptrdiff_t src_stride_ptr, int width,
-                                    int source_y_fraction);
-
-// Sobel images.
-void SobelXRow_C(const uint8* src_y0, const uint8* src_y1, const uint8* src_y2,
-                 uint8* dst_sobelx, int width);
-void SobelXRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobelx, int width);
-void SobelXRow_NEON(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobelx, int width);
-void SobelYRow_C(const uint8* src_y0, const uint8* src_y1,
-                 uint8* dst_sobely, int width);
-void SobelYRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width);
-void SobelYRow_NEON(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width);
-void SobelRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                uint8* dst_argb, int width);
-void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                   uint8* dst_argb, int width);
-void SobelRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                   uint8* dst_argb, int width);
-void SobelToPlaneRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                       uint8* dst_y, int width);
-void SobelToPlaneRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_y, int width);
-void SobelToPlaneRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_y, int width);
-void SobelXYRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                  uint8* dst_argb, int width);
-void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width);
-void SobelXYRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width);
-
-void ARGBPolynomialRow_C(const uint8* src_argb,
-                         uint8* dst_argb, const float* poly,
-                         int width);
-void ARGBPolynomialRow_SSE2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width);
-void ARGBPolynomialRow_AVX2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width);
-
-void ARGBLumaColorTableRow_C(const uint8* src_argb, uint8* dst_argb, int width,
-                             const uint8* luma, uint32 lumacoeff);
-void ARGBLumaColorTableRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                 int width,
-                                 const uint8* luma, uint32 lumacoeff);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_ROW_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale.h
deleted file mode 100755
index 592b8ed5fa..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale.h
+++ /dev/null
@@ -1,85 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_SCALE_H_  // NOLINT
-#define INCLUDE_LIBYUV_SCALE_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Supported filtering.
-typedef enum FilterMode {
-  kFilterNone = 0,  // Point sample; Fastest.
-  kFilterLinear = 1,  // Filter horizontally only.
-  kFilterBilinear = 2,  // Faster than box, but lower quality scaling down.
-  kFilterBox = 3  // Highest quality.
-} FilterModeEnum;
-
-// Scale a YUV plane.
-LIBYUV_API
-void ScalePlane(const uint8* src, int src_stride,
-                int src_width, int src_height,
-                uint8* dst, int dst_stride,
-                int dst_width, int dst_height,
-                enum FilterMode filtering);
-
-// Scales a YUV 4:2:0 image from the src width and height to the
-// dst width and height.
-// If filtering is kFilterNone, a simple nearest-neighbor algorithm is
-// used. This produces basic (blocky) quality at the fastest speed.
-// If filtering is kFilterBilinear, interpolation is used to produce a better
-// quality image, at the expense of speed.
-// If filtering is kFilterBox, averaging is used to produce ever better
-// quality image, at further expense of speed.
-// Returns 0 if successful.
-
-LIBYUV_API
-int I420Scale(const uint8* src_y, int src_stride_y,
-              const uint8* src_u, int src_stride_u,
-              const uint8* src_v, int src_stride_v,
-              int src_width, int src_height,
-              uint8* dst_y, int dst_stride_y,
-              uint8* dst_u, int dst_stride_u,
-              uint8* dst_v, int dst_stride_v,
-              int dst_width, int dst_height,
-              enum FilterMode filtering);
-
-#ifdef __cplusplus
-// Legacy API.  Deprecated.
-LIBYUV_API
-int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
-          int src_stride_y, int src_stride_u, int src_stride_v,
-          int src_width, int src_height,
-          uint8* dst_y, uint8* dst_u, uint8* dst_v,
-          int dst_stride_y, int dst_stride_u, int dst_stride_v,
-          int dst_width, int dst_height,
-          LIBYUV_BOOL interpolate);
-
-// Legacy API.  Deprecated.
-LIBYUV_API
-int ScaleOffset(const uint8* src_i420, int src_width, int src_height,
-                uint8* dst_i420, int dst_width, int dst_height, int dst_yoffset,
-                LIBYUV_BOOL interpolate);
-
-// For testing, allow disabling of specialized scalers.
-LIBYUV_API
-void SetUseReferenceImpl(LIBYUV_BOOL use);
-#endif  // __cplusplus
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_SCALE_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_argb.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_argb.h
deleted file mode 100755
index 0c9b362575..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_argb.h
+++ /dev/null
@@ -1,57 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_SCALE_ARGB_H_  // NOLINT
-#define INCLUDE_LIBYUV_SCALE_ARGB_H_
-
-#include "libyuv/basic_types.h"
-#include "libyuv/scale.h"  // For FilterMode
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-LIBYUV_API
-int ARGBScale(const uint8* src_argb, int src_stride_argb,
-              int src_width, int src_height,
-              uint8* dst_argb, int dst_stride_argb,
-              int dst_width, int dst_height,
-              enum FilterMode filtering);
-
-// Clipped scale takes destination rectangle coordinates for clip values.
-LIBYUV_API
-int ARGBScaleClip(const uint8* src_argb, int src_stride_argb,
-                  int src_width, int src_height,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int dst_width, int dst_height,
-                  int clip_x, int clip_y, int clip_width, int clip_height,
-                  enum FilterMode filtering);
-
-// TODO(fbarchard): Implement this.
-// Scale with YUV conversion to ARGB and clipping.
-LIBYUV_API
-int YUVToARGBScaleClip(const uint8* src_y, int src_stride_y,
-                       const uint8* src_u, int src_stride_u,
-                       const uint8* src_v, int src_stride_v,
-                       uint32 src_fourcc,
-                       int src_width, int src_height,
-                       uint8* dst_argb, int dst_stride_argb,
-                       uint32 dst_fourcc,
-                       int dst_width, int dst_height,
-                       int clip_x, int clip_y, int clip_width, int clip_height,
-                       enum FilterMode filtering);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_SCALE_ARGB_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_row.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_row.h
deleted file mode 100644
index 13eccc4d77..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/scale_row.h
+++ /dev/null
@@ -1,301 +0,0 @@
-/*
- *  Copyright 2013 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_SCALE_ROW_H_  // NOLINT
-#define INCLUDE_LIBYUV_SCALE_ROW_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if defined(__pnacl__) || defined(__CLR_VER) || defined(COVERAGE_ENABLED) || \
-    defined(TARGET_IPHONE_SIMULATOR)
-#define LIBYUV_DISABLE_X86
-#endif
-
-// The following are available on all x86 platforms:
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__))
-#define HAS_SCALEROWDOWN2_SSE2
-#define HAS_SCALEROWDOWN4_SSE2
-#define HAS_SCALEROWDOWN34_SSSE3
-#define HAS_SCALEROWDOWN38_SSSE3
-#define HAS_SCALEADDROWS_SSE2
-#define HAS_SCALEFILTERCOLS_SSSE3
-#define HAS_SCALECOLSUP2_SSE2
-#define HAS_SCALEARGBROWDOWN2_SSE2
-#define HAS_SCALEARGBROWDOWNEVEN_SSE2
-#define HAS_SCALEARGBCOLS_SSE2
-#define HAS_SCALEARGBFILTERCOLS_SSSE3
-#define HAS_SCALEARGBCOLSUP2_SSE2
-#define HAS_FIXEDDIV_X86
-#define HAS_FIXEDDIV1_X86
-#endif
-
-// The following are available on Neon platforms:
-#if !defined(LIBYUV_DISABLE_NEON) && !defined(__native_client__) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-#define HAS_SCALEROWDOWN2_NEON
-#define HAS_SCALEROWDOWN4_NEON
-#define HAS_SCALEROWDOWN34_NEON
-#define HAS_SCALEROWDOWN38_NEON
-#define HAS_SCALEARGBROWDOWNEVEN_NEON
-#define HAS_SCALEARGBROWDOWN2_NEON
-#endif
-
-// The following are available on Mips platforms:
-#if !defined(LIBYUV_DISABLE_MIPS) && !defined(__native_client__) && \
-    defined(__mips__) && defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-#define HAS_SCALEROWDOWN2_MIPS_DSPR2
-#define HAS_SCALEROWDOWN4_MIPS_DSPR2
-#define HAS_SCALEROWDOWN34_MIPS_DSPR2
-#define HAS_SCALEROWDOWN38_MIPS_DSPR2
-#endif
-
-// Scale ARGB vertically with bilinear interpolation.
-void ScalePlaneVertical(int src_height,
-                        int dst_width, int dst_height,
-                        int src_stride, int dst_stride,
-                        const uint8* src_argb, uint8* dst_argb,
-                        int x, int y, int dy,
-                        int bpp, enum FilterMode filtering);
-
-// Simplify the filtering based on scale factors.
-enum FilterMode ScaleFilterReduce(int src_width, int src_height,
-                                  int dst_width, int dst_height,
-                                  enum FilterMode filtering);
-
-// Divide num by div and return as 16.16 fixed point result.
-int FixedDiv_C(int num, int div);
-int FixedDiv_X86(int num, int div);
-// Divide num - 1 by div - 1 and return as 16.16 fixed point result.
-int FixedDiv1_C(int num, int div);
-int FixedDiv1_X86(int num, int div);
-#ifdef HAS_FIXEDDIV_X86
-#define FixedDiv FixedDiv_X86
-#define FixedDiv1 FixedDiv1_X86
-#else
-#define FixedDiv FixedDiv_C
-#define FixedDiv1 FixedDiv1_C
-#endif
-
-// Compute slope values for stepping.
-void ScaleSlope(int src_width, int src_height,
-                int dst_width, int dst_height,
-                enum FilterMode filtering,
-                int* x, int* y, int* dx, int* dy);
-
-void ScaleRowDown2_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                     uint8* dst, int dst_width);
-void ScaleRowDown2Linear_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst, int dst_width);
-void ScaleRowDown2Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width);
-void ScaleRowDown4_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                     uint8* dst, int dst_width);
-void ScaleRowDown4Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width);
-void ScaleRowDown34_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                      uint8* dst, int dst_width);
-void ScaleRowDown34_0_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* d, int dst_width);
-void ScaleRowDown34_1_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* d, int dst_width);
-void ScaleCols_C(uint8* dst_ptr, const uint8* src_ptr,
-                 int dst_width, int x, int dx);
-void ScaleColsUp2_C(uint8* dst_ptr, const uint8* src_ptr,
-                    int dst_width, int, int);
-void ScaleFilterCols_C(uint8* dst_ptr, const uint8* src_ptr,
-                       int dst_width, int x, int dx);
-void ScaleFilterCols64_C(uint8* dst_ptr, const uint8* src_ptr,
-                         int dst_width, int x, int dx);
-void ScaleRowDown38_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                      uint8* dst, int dst_width);
-void ScaleRowDown38_3_Box_C(const uint8* src_ptr,
-                            ptrdiff_t src_stride,
-                            uint8* dst_ptr, int dst_width);
-void ScaleRowDown38_2_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* dst_ptr, int dst_width);
-void ScaleAddRows_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                    uint16* dst_ptr, int src_width, int src_height);
-void ScaleARGBRowDown2_C(const uint8* src_argb,
-                         ptrdiff_t src_stride,
-                         uint8* dst_argb, int dst_width);
-void ScaleARGBRowDown2Linear_C(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               uint8* dst_argb, int dst_width);
-void ScaleARGBRowDown2Box_C(const uint8* src_argb, ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width);
-void ScaleARGBRowDownEven_C(const uint8* src_argb, ptrdiff_t src_stride,
-                            int src_stepx,
-                            uint8* dst_argb, int dst_width);
-void ScaleARGBRowDownEvenBox_C(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width);
-void ScaleARGBCols_C(uint8* dst_argb, const uint8* src_argb,
-                     int dst_width, int x, int dx);
-void ScaleARGBCols64_C(uint8* dst_argb, const uint8* src_argb,
-                       int dst_width, int x, int dx);
-void ScaleARGBColsUp2_C(uint8* dst_argb, const uint8* src_argb,
-                        int dst_width, int, int);
-void ScaleARGBFilterCols_C(uint8* dst_argb, const uint8* src_argb,
-                           int dst_width, int x, int dx);
-void ScaleARGBFilterCols64_C(uint8* dst_argb, const uint8* src_argb,
-                             int dst_width, int x, int dx);
-
-void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width);
-void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst_ptr, int dst_width);
-void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-void ScaleRowDown2_Unaligned_SSE2(const uint8* src_ptr,
-                                  ptrdiff_t src_stride,
-                                  uint8* dst_ptr, int dst_width);
-void ScaleRowDown2Linear_Unaligned_SSE2(const uint8* src_ptr,
-                                        ptrdiff_t src_stride,
-                                        uint8* dst_ptr, int dst_width);
-void ScaleRowDown2Box_Unaligned_SSE2(const uint8* src_ptr,
-                                     ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width);
-void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width);
-void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width);
-void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width);
-void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width);
-void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width);
-void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width);
-void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width);
-void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                       uint16* dst_ptr, int src_width,
-                       int src_height);
-void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                           int dst_width, int x, int dx);
-void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                       int dst_width, int x, int dx);
-void ScaleARGBRowDown2_SSE2(const uint8* src_argb,
-                            ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width);
-void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride,
-                                  uint8* dst_argb, int dst_width);
-void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               uint8* dst_argb, int dst_width);
-void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width);
-void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride,
-                                  int src_stepx,
-                                  uint8* dst_argb, int dst_width);
-void ScaleARGBCols_SSE2(uint8* dst_argb, const uint8* src_argb,
-                        int dst_width, int x, int dx);
-void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb,
-                               int dst_width, int x, int dx);
-void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb,
-                           int dst_width, int x, int dx);
-// Row functions.
-void ScaleARGBRowDownEven_NEON(const uint8* src_argb, int src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width);
-void ScaleARGBRowDownEvenBox_NEON(const uint8* src_argb, int src_stride,
-                                  int src_stepx,
-                                  uint8* dst_argb, int dst_width);
-void ScaleARGBRowDown2_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* dst, int dst_width);
-void ScaleARGBRowDown2Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width);
-
-// ScaleRowDown2Box also used by planar functions
-// NEON downscalers with interpolation.
-
-// Note - not static due to reuse in convert for 444 to 420.
-void ScaleRowDown2_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width);
-
-void ScaleRowDown2Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst, int dst_width);
-
-void ScaleRowDown4_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width);
-void ScaleRowDown4Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-
-// Down scale from 4 to 3 pixels. Use the neon multilane read/write
-//  to load up the every 4th pixel into a 4 different registers.
-// Point samples 32 pixels to 24 pixels.
-void ScaleRowDown34_NEON(const uint8* src_ptr,
-                         ptrdiff_t src_stride,
-                         uint8* dst_ptr, int dst_width);
-void ScaleRowDown34_0_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width);
-void ScaleRowDown34_1_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width);
-
-// 32 -> 12
-void ScaleRowDown38_NEON(const uint8* src_ptr,
-                         ptrdiff_t src_stride,
-                         uint8* dst_ptr, int dst_width);
-// 32x3 -> 12x1
-void ScaleRowDown38_3_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width);
-// 32x2 -> 12x1
-void ScaleRowDown38_2_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width);
-
-void ScaleRowDown2_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst, int dst_width);
-void ScaleRowDown2Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                 uint8* dst, int dst_width);
-void ScaleRowDown4_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst, int dst_width);
-void ScaleRowDown4Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                 uint8* dst, int dst_width);
-void ScaleRowDown34_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width);
-void ScaleRowDown34_0_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* d, int dst_width);
-void ScaleRowDown34_1_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* d, int dst_width);
-void ScaleRowDown38_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width);
-void ScaleRowDown38_2_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width);
-void ScaleRowDown38_3_Box_MIPS_DSPR2(const uint8* src_ptr,
-                                     ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_SCALE_ROW_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/version.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/version.h
deleted file mode 100755
index 4881861866..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/version.h
+++ /dev/null
@@ -1,16 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
-#define INCLUDE_LIBYUV_VERSION_H_
-
-#define LIBYUV_VERSION 998
-
-#endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/video_common.h b/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/video_common.h
deleted file mode 100755
index 039efb96d1..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/include/libyuv/video_common.h
+++ /dev/null
@@ -1,182 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-// Common definitions for video, including fourcc and VideoFormat.
-
-#ifndef INCLUDE_LIBYUV_VIDEO_COMMON_H_  // NOLINT
-#define INCLUDE_LIBYUV_VIDEO_COMMON_H_
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-//////////////////////////////////////////////////////////////////////////////
-// Definition of FourCC codes
-//////////////////////////////////////////////////////////////////////////////
-
-// Convert four characters to a FourCC code.
-// Needs to be a macro otherwise the OS X compiler complains when the kFormat*
-// constants are used in a switch.
-#ifdef __cplusplus
-#define FOURCC(a, b, c, d) ( \
-    (static_cast<uint32>(a)) | (static_cast<uint32>(b) << 8) | \
-    (static_cast<uint32>(c) << 16) | (static_cast<uint32>(d) << 24))
-#else
-#define FOURCC(a, b, c, d) ( \
-    ((uint32)(a)) | ((uint32)(b) << 8) | /* NOLINT */ \
-    ((uint32)(c) << 16) | ((uint32)(d) << 24))  /* NOLINT */
-#endif
-
-// Some pages discussing FourCC codes:
-//   http://www.fourcc.org/yuv.php
-//   http://v4l2spec.bytesex.org/spec/book1.htm
-//   http://developer.apple.com/quicktime/icefloe/dispatch020.html
-//   http://msdn.microsoft.com/library/windows/desktop/dd206750.aspx#nv12
-//   http://people.xiph.org/~xiphmont/containers/nut/nut4cc.txt
-
-// FourCC codes grouped according to implementation efficiency.
-// Primary formats should convert in 1 efficient step.
-// Secondary formats are converted in 2 steps.
-// Auxilliary formats call primary converters.
-enum FourCC {
-  // 9 Primary YUV formats: 5 planar, 2 biplanar, 2 packed.
-  FOURCC_I420 = FOURCC('I', '4', '2', '0'),
-  FOURCC_I422 = FOURCC('I', '4', '2', '2'),
-  FOURCC_I444 = FOURCC('I', '4', '4', '4'),
-  FOURCC_I411 = FOURCC('I', '4', '1', '1'),
-  FOURCC_I400 = FOURCC('I', '4', '0', '0'),
-  FOURCC_NV21 = FOURCC('N', 'V', '2', '1'),
-  FOURCC_NV12 = FOURCC('N', 'V', '1', '2'),
-  FOURCC_YUY2 = FOURCC('Y', 'U', 'Y', '2'),
-  FOURCC_UYVY = FOURCC('U', 'Y', 'V', 'Y'),
-
-  // 2 Secondary YUV formats: row biplanar.
-  FOURCC_M420 = FOURCC('M', '4', '2', '0'),
-  FOURCC_Q420 = FOURCC('Q', '4', '2', '0'),
-
-  // 9 Primary RGB formats: 4 32 bpp, 2 24 bpp, 3 16 bpp.
-  FOURCC_ARGB = FOURCC('A', 'R', 'G', 'B'),
-  FOURCC_BGRA = FOURCC('B', 'G', 'R', 'A'),
-  FOURCC_ABGR = FOURCC('A', 'B', 'G', 'R'),
-  FOURCC_24BG = FOURCC('2', '4', 'B', 'G'),
-  FOURCC_RAW  = FOURCC('r', 'a', 'w', ' '),
-  FOURCC_RGBA = FOURCC('R', 'G', 'B', 'A'),
-  FOURCC_RGBP = FOURCC('R', 'G', 'B', 'P'),  // rgb565 LE.
-  FOURCC_RGBO = FOURCC('R', 'G', 'B', 'O'),  // argb1555 LE.
-  FOURCC_R444 = FOURCC('R', '4', '4', '4'),  // argb4444 LE.
-
-  // 4 Secondary RGB formats: 4 Bayer Patterns.
-  FOURCC_RGGB = FOURCC('R', 'G', 'G', 'B'),
-  FOURCC_BGGR = FOURCC('B', 'G', 'G', 'R'),
-  FOURCC_GRBG = FOURCC('G', 'R', 'B', 'G'),
-  FOURCC_GBRG = FOURCC('G', 'B', 'R', 'G'),
-
-  // 1 Primary Compressed YUV format.
-  FOURCC_MJPG = FOURCC('M', 'J', 'P', 'G'),
-
-  // 5 Auxiliary YUV variations: 3 with U and V planes are swapped, 1 Alias.
-  FOURCC_YV12 = FOURCC('Y', 'V', '1', '2'),
-  FOURCC_YV16 = FOURCC('Y', 'V', '1', '6'),
-  FOURCC_YV24 = FOURCC('Y', 'V', '2', '4'),
-  FOURCC_YU12 = FOURCC('Y', 'U', '1', '2'),  // Linux version of I420.
-  FOURCC_J420 = FOURCC('J', '4', '2', '0'),
-  FOURCC_J400 = FOURCC('J', '4', '0', '0'),
-
-  // 14 Auxiliary aliases.  CanonicalFourCC() maps these to canonical fourcc.
-  FOURCC_IYUV = FOURCC('I', 'Y', 'U', 'V'),  // Alias for I420.
-  FOURCC_YU16 = FOURCC('Y', 'U', '1', '6'),  // Alias for I422.
-  FOURCC_YU24 = FOURCC('Y', 'U', '2', '4'),  // Alias for I444.
-  FOURCC_YUYV = FOURCC('Y', 'U', 'Y', 'V'),  // Alias for YUY2.
-  FOURCC_YUVS = FOURCC('y', 'u', 'v', 's'),  // Alias for YUY2 on Mac.
-  FOURCC_HDYC = FOURCC('H', 'D', 'Y', 'C'),  // Alias for UYVY.
-  FOURCC_2VUY = FOURCC('2', 'v', 'u', 'y'),  // Alias for UYVY on Mac.
-  FOURCC_JPEG = FOURCC('J', 'P', 'E', 'G'),  // Alias for MJPG.
-  FOURCC_DMB1 = FOURCC('d', 'm', 'b', '1'),  // Alias for MJPG on Mac.
-  FOURCC_BA81 = FOURCC('B', 'A', '8', '1'),  // Alias for BGGR.
-  FOURCC_RGB3 = FOURCC('R', 'G', 'B', '3'),  // Alias for RAW.
-  FOURCC_BGR3 = FOURCC('B', 'G', 'R', '3'),  // Alias for 24BG.
-  FOURCC_CM32 = FOURCC(0, 0, 0, 32),  // Alias for BGRA kCMPixelFormat_32ARGB
-  FOURCC_CM24 = FOURCC(0, 0, 0, 24),  // Alias for RAW kCMPixelFormat_24RGB
-  FOURCC_L555 = FOURCC('L', '5', '5', '5'),  // Alias for RGBO.
-  FOURCC_L565 = FOURCC('L', '5', '6', '5'),  // Alias for RGBP.
-  FOURCC_5551 = FOURCC('5', '5', '5', '1'),  // Alias for RGBO.
-
-  // 1 Auxiliary compressed YUV format set aside for capturer.
-  FOURCC_H264 = FOURCC('H', '2', '6', '4'),
-
-  // Match any fourcc.
-  FOURCC_ANY  = 0xFFFFFFFF,
-};
-
-enum FourCCBpp {
-  // Canonical fourcc codes used in our code.
-  FOURCC_BPP_I420 = 12,
-  FOURCC_BPP_I422 = 16,
-  FOURCC_BPP_I444 = 24,
-  FOURCC_BPP_I411 = 12,
-  FOURCC_BPP_I400 = 8,
-  FOURCC_BPP_NV21 = 12,
-  FOURCC_BPP_NV12 = 12,
-  FOURCC_BPP_YUY2 = 16,
-  FOURCC_BPP_UYVY = 16,
-  FOURCC_BPP_M420 = 12,
-  FOURCC_BPP_Q420 = 12,
-  FOURCC_BPP_ARGB = 32,
-  FOURCC_BPP_BGRA = 32,
-  FOURCC_BPP_ABGR = 32,
-  FOURCC_BPP_RGBA = 32,
-  FOURCC_BPP_24BG = 24,
-  FOURCC_BPP_RAW  = 24,
-  FOURCC_BPP_RGBP = 16,
-  FOURCC_BPP_RGBO = 16,
-  FOURCC_BPP_R444 = 16,
-  FOURCC_BPP_RGGB = 8,
-  FOURCC_BPP_BGGR = 8,
-  FOURCC_BPP_GRBG = 8,
-  FOURCC_BPP_GBRG = 8,
-  FOURCC_BPP_YV12 = 12,
-  FOURCC_BPP_YV16 = 16,
-  FOURCC_BPP_YV24 = 24,
-  FOURCC_BPP_YU12 = 12,
-  FOURCC_BPP_J420 = 12,
-  FOURCC_BPP_J400 = 8,
-  FOURCC_BPP_MJPG = 0,  // 0 means unknown.
-  FOURCC_BPP_H264 = 0,
-  FOURCC_BPP_IYUV = 12,
-  FOURCC_BPP_YU16 = 16,
-  FOURCC_BPP_YU24 = 24,
-  FOURCC_BPP_YUYV = 16,
-  FOURCC_BPP_YUVS = 16,
-  FOURCC_BPP_HDYC = 16,
-  FOURCC_BPP_2VUY = 16,
-  FOURCC_BPP_JPEG = 1,
-  FOURCC_BPP_DMB1 = 1,
-  FOURCC_BPP_BA81 = 8,
-  FOURCC_BPP_RGB3 = 24,
-  FOURCC_BPP_BGR3 = 24,
-  FOURCC_BPP_CM32 = 32,
-  FOURCC_BPP_CM24 = 24,
-
-  // Match any fourcc.
-  FOURCC_BPP_ANY  = 0,  // 0 means unknown.
-};
-
-// Converts fourcc aliases into canonical ones.
-LIBYUV_API uint32 CanonicalFourCC(uint32 fourcc);
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
-#endif  // INCLUDE_LIBYUV_VIDEO_COMMON_H_  NOLINT
diff --git a/drivers/theoraplayer/src/YUV/libyuv/libtheoraplayer-readme.txt b/drivers/theoraplayer/src/YUV/libyuv/libtheoraplayer-readme.txt
deleted file mode 100755
index 680e4a1c36..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/libtheoraplayer-readme.txt
+++ /dev/null
@@ -1,15 +0,0 @@
-libyuv's source code is here provided in minimalist distribution format
-with all source files not needed for compiling libtheoraplayer removed.
-
-- The project files were modified to fit libtheoraplayer's binary output
-  folder structure.
-- Some project files missing in the original source distibution were added to support
-  compiling the libtheoraplayer on those platforms.
-- Also, some code may have been changed to address certain compiler/platform
-  specific problems and is so indicated in the source code.
-
-libyuv is owned and maintained by the Google Inc. and this distribution
-is present here only for convenience and easier compilation of libtheoraplayer.
-
-If you want to use libyuv outside of libtheoraplayer, it is encouraged to use the
-original source distribution by Google Inc: https://code.google.com/p/libyuv/
\ No newline at end of file
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/compare.cc b/drivers/theoraplayer/src/YUV/libyuv/src/compare.cc
deleted file mode 100755
index 9ea81b4e21..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/compare.cc
+++ /dev/null
@@ -1,325 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/compare.h"
-
-#include <float.h>
-#include <math.h>
-#ifdef _OPENMP
-#include <omp.h>
-#endif
-
-#include "libyuv/basic_types.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// hash seed of 5381 recommended.
-// Internal C version of HashDjb2 with int sized count for efficiency.
-uint32 HashDjb2_C(const uint8* src, int count, uint32 seed);
-
-// This module is for Visual C x86
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || \
-    (defined(__x86_64__) || (defined(__i386__) && !defined(__pic__))))
-#define HAS_HASHDJB2_SSE41
-uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed);
-
-#if _MSC_VER >= 1700
-#define HAS_HASHDJB2_AVX2
-uint32 HashDjb2_AVX2(const uint8* src, int count, uint32 seed);
-#endif
-
-#endif  // HAS_HASHDJB2_SSE41
-
-// hash seed of 5381 recommended.
-LIBYUV_API
-uint32 HashDjb2(const uint8* src, uint64 count, uint32 seed) {
-  const int kBlockSize = 1 << 15;  // 32768;
-  int remainder;
-  uint32 (*HashDjb2_SSE)(const uint8* src, int count, uint32 seed) = HashDjb2_C;
-#if defined(HAS_HASHDJB2_SSE41)
-  if (TestCpuFlag(kCpuHasSSE41)) {
-    HashDjb2_SSE = HashDjb2_SSE41;
-  }
-#endif
-#if defined(HAS_HASHDJB2_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2)) {
-    HashDjb2_SSE = HashDjb2_AVX2;
-  }
-#endif
-
-  while (count >= (uint64)(kBlockSize)) {
-    seed = HashDjb2_SSE(src, kBlockSize, seed);
-    src += kBlockSize;
-    count -= kBlockSize;
-  }
-  remainder = (int)(count) & ~15;
-  if (remainder) {
-    seed = HashDjb2_SSE(src, remainder, seed);
-    src += remainder;
-    count -= remainder;
-  }
-  remainder = (int)(count) & 15;
-  if (remainder) {
-    seed = HashDjb2_C(src, remainder, seed);
-  }
-  return seed;
-}
-
-uint32 SumSquareError_C(const uint8* src_a, const uint8* src_b, int count);
-#if !defined(LIBYUV_DISABLE_NEON) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-#define HAS_SUMSQUAREERROR_NEON
-uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count);
-#endif
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__))
-#define HAS_SUMSQUAREERROR_SSE2
-uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count);
-#endif
-// Visual C 2012 required for AVX2.
-#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && _MSC_VER >= 1700
-#define HAS_SUMSQUAREERROR_AVX2
-uint32 SumSquareError_AVX2(const uint8* src_a, const uint8* src_b, int count);
-#endif
-
-// TODO(fbarchard): Refactor into row function.
-LIBYUV_API
-uint64 ComputeSumSquareError(const uint8* src_a, const uint8* src_b,
-                             int count) {
-  // SumSquareError returns values 0 to 65535 for each squared difference.
-  // Up to 65536 of those can be summed and remain within a uint32.
-  // After each block of 65536 pixels, accumulate into a uint64.
-  const int kBlockSize = 65536;
-  int remainder = count & (kBlockSize - 1) & ~31;
-  uint64 sse = 0;
-  int i;
-  uint32 (*SumSquareError)(const uint8* src_a, const uint8* src_b, int count) =
-      SumSquareError_C;
-#if defined(HAS_SUMSQUAREERROR_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    SumSquareError = SumSquareError_NEON;
-  }
-#endif
-#if defined(HAS_SUMSQUAREERROR_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) &&
-      IS_ALIGNED(src_a, 16) && IS_ALIGNED(src_b, 16)) {
-    // Note only used for multiples of 16 so count is not checked.
-    SumSquareError = SumSquareError_SSE2;
-  }
-#endif
-#if defined(HAS_SUMSQUAREERROR_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2)) {
-    // Note only used for multiples of 32 so count is not checked.
-    SumSquareError = SumSquareError_AVX2;
-  }
-#endif
-#ifdef _OPENMP
-#pragma omp parallel for reduction(+: sse)
-#endif
-  for (i = 0; i < (count - (kBlockSize - 1)); i += kBlockSize) {
-    sse += SumSquareError(src_a + i, src_b + i, kBlockSize);
-  }
-  src_a += count & ~(kBlockSize - 1);
-  src_b += count & ~(kBlockSize - 1);
-  if (remainder) {
-    sse += SumSquareError(src_a, src_b, remainder);
-    src_a += remainder;
-    src_b += remainder;
-  }
-  remainder = count & 31;
-  if (remainder) {
-    sse += SumSquareError_C(src_a, src_b, remainder);
-  }
-  return sse;
-}
-
-LIBYUV_API
-uint64 ComputeSumSquareErrorPlane(const uint8* src_a, int stride_a,
-                                  const uint8* src_b, int stride_b,
-                                  int width, int height) {
-  uint64 sse = 0;
-  int h;
-  // Coalesce rows.
-  if (stride_a == width &&
-      stride_b == width) {
-    width *= height;
-    height = 1;
-    stride_a = stride_b = 0;
-  }
-  for (h = 0; h < height; ++h) {
-    sse += ComputeSumSquareError(src_a, src_b, width);
-    src_a += stride_a;
-    src_b += stride_b;
-  }
-  return sse;
-}
-
-LIBYUV_API
-double SumSquareErrorToPsnr(uint64 sse, uint64 count) {
-  double psnr;
-  if (sse > 0) {
-    double mse = (double)(count) / (double)(sse);
-    psnr = 10.0 * log10(255.0 * 255.0 * mse);
-  } else {
-    psnr = kMaxPsnr;      // Limit to prevent divide by 0
-  }
-
-  if (psnr > kMaxPsnr)
-    psnr = kMaxPsnr;
-
-  return psnr;
-}
-
-LIBYUV_API
-double CalcFramePsnr(const uint8* src_a, int stride_a,
-                     const uint8* src_b, int stride_b,
-                     int width, int height) {
-  const uint64 samples = width * height;
-  const uint64 sse = ComputeSumSquareErrorPlane(src_a, stride_a,
-                                                src_b, stride_b,
-                                                width, height);
-  return SumSquareErrorToPsnr(sse, samples);
-}
-
-LIBYUV_API
-double I420Psnr(const uint8* src_y_a, int stride_y_a,
-                const uint8* src_u_a, int stride_u_a,
-                const uint8* src_v_a, int stride_v_a,
-                const uint8* src_y_b, int stride_y_b,
-                const uint8* src_u_b, int stride_u_b,
-                const uint8* src_v_b, int stride_v_b,
-                int width, int height) {
-  const uint64 sse_y = ComputeSumSquareErrorPlane(src_y_a, stride_y_a,
-                                                  src_y_b, stride_y_b,
-                                                  width, height);
-  const int width_uv = (width + 1) >> 1;
-  const int height_uv = (height + 1) >> 1;
-  const uint64 sse_u = ComputeSumSquareErrorPlane(src_u_a, stride_u_a,
-                                                  src_u_b, stride_u_b,
-                                                  width_uv, height_uv);
-  const uint64 sse_v = ComputeSumSquareErrorPlane(src_v_a, stride_v_a,
-                                                  src_v_b, stride_v_b,
-                                                  width_uv, height_uv);
-  const uint64 samples = width * height + 2 * (width_uv * height_uv);
-  const uint64 sse = sse_y + sse_u + sse_v;
-  return SumSquareErrorToPsnr(sse, samples);
-}
-
-static const int64 cc1 =  26634;  // (64^2*(.01*255)^2
-static const int64 cc2 = 239708;  // (64^2*(.03*255)^2
-
-static double Ssim8x8_C(const uint8* src_a, int stride_a,
-                        const uint8* src_b, int stride_b) {
-  int64 sum_a = 0;
-  int64 sum_b = 0;
-  int64 sum_sq_a = 0;
-  int64 sum_sq_b = 0;
-  int64 sum_axb = 0;
-
-  int i;
-  for (i = 0; i < 8; ++i) {
-    int j;
-    for (j = 0; j < 8; ++j) {
-      sum_a += src_a[j];
-      sum_b += src_b[j];
-      sum_sq_a += src_a[j] * src_a[j];
-      sum_sq_b += src_b[j] * src_b[j];
-      sum_axb += src_a[j] * src_b[j];
-    }
-
-    src_a += stride_a;
-    src_b += stride_b;
-  }
-
-  {
-    const int64 count = 64;
-    // scale the constants by number of pixels
-    const int64 c1 = (cc1 * count * count) >> 12;
-    const int64 c2 = (cc2 * count * count) >> 12;
-
-    const int64 sum_a_x_sum_b = sum_a * sum_b;
-
-    const int64 ssim_n = (2 * sum_a_x_sum_b + c1) *
-                         (2 * count * sum_axb - 2 * sum_a_x_sum_b + c2);
-
-    const int64 sum_a_sq = sum_a*sum_a;
-    const int64 sum_b_sq = sum_b*sum_b;
-
-    const int64 ssim_d = (sum_a_sq + sum_b_sq + c1) *
-                         (count * sum_sq_a - sum_a_sq +
-                          count * sum_sq_b - sum_b_sq + c2);
-
-    if (ssim_d == 0.0) {
-      return DBL_MAX;
-    }
-    return ssim_n * 1.0 / ssim_d;
-  }
-}
-
-// We are using a 8x8 moving window with starting location of each 8x8 window
-// on the 4x4 pixel grid. Such arrangement allows the windows to overlap
-// block boundaries to penalize blocking artifacts.
-LIBYUV_API
-double CalcFrameSsim(const uint8* src_a, int stride_a,
-                     const uint8* src_b, int stride_b,
-                     int width, int height) {
-  int samples = 0;
-  double ssim_total = 0;
-  double (*Ssim8x8)(const uint8* src_a, int stride_a,
-                    const uint8* src_b, int stride_b) = Ssim8x8_C;
-
-  // sample point start with each 4x4 location
-  int i;
-  for (i = 0; i < height - 8; i += 4) {
-    int j;
-    for (j = 0; j < width - 8; j += 4) {
-      ssim_total += Ssim8x8(src_a + j, stride_a, src_b + j, stride_b);
-      samples++;
-    }
-
-    src_a += stride_a * 4;
-    src_b += stride_b * 4;
-  }
-
-  ssim_total /= samples;
-  return ssim_total;
-}
-
-LIBYUV_API
-double I420Ssim(const uint8* src_y_a, int stride_y_a,
-                const uint8* src_u_a, int stride_u_a,
-                const uint8* src_v_a, int stride_v_a,
-                const uint8* src_y_b, int stride_y_b,
-                const uint8* src_u_b, int stride_u_b,
-                const uint8* src_v_b, int stride_v_b,
-                int width, int height) {
-  const double ssim_y = CalcFrameSsim(src_y_a, stride_y_a,
-                                      src_y_b, stride_y_b, width, height);
-  const int width_uv = (width + 1) >> 1;
-  const int height_uv = (height + 1) >> 1;
-  const double ssim_u = CalcFrameSsim(src_u_a, stride_u_a,
-                                      src_u_b, stride_u_b,
-                                      width_uv, height_uv);
-  const double ssim_v = CalcFrameSsim(src_v_a, stride_v_a,
-                                      src_v_b, stride_v_b,
-                                      width_uv, height_uv);
-  return ssim_y * 0.8 + 0.1 * (ssim_u + ssim_v);
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/compare_common.cc b/drivers/theoraplayer/src/YUV/libyuv/src/compare_common.cc
deleted file mode 100755
index c546b51829..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/compare_common.cc
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-uint32 SumSquareError_C(const uint8* src_a, const uint8* src_b, int count) {
-  uint32 sse = 0u;
-  int i;
-  for (i = 0; i < count; ++i) {
-    int diff = src_a[i] - src_b[i];
-    sse += (uint32)(diff * diff);
-  }
-  return sse;
-}
-
-// hash seed of 5381 recommended.
-// Internal C version of HashDjb2 with int sized count for efficiency.
-uint32 HashDjb2_C(const uint8* src, int count, uint32 seed) {
-  uint32 hash = seed;
-  int i;
-  for (i = 0; i < count; ++i) {
-    hash += (hash << 5) + src[i];
-  }
-  return hash;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/compare_neon.cc b/drivers/theoraplayer/src/YUV/libyuv/src/compare_neon.cc
deleted file mode 100755
index bb843a6ab8..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/compare_neon.cc
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_NEON) && defined(__ARM_NEON__)
-
-uint32 SumSquareError_NEON(const uint8* src_a, const uint8* src_b, int count) {
-  volatile uint32 sse;
-  asm volatile (
-#ifdef _ANDROID
-	".fpu neon\n"
-#endif
-    "vmov.u8    q8, #0                         \n"
-    "vmov.u8    q10, #0                        \n"
-    "vmov.u8    q9, #0                         \n"
-    "vmov.u8    q11, #0                        \n"
-
-    ".p2align  2                               \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"
-    "vld1.8     {q1}, [%1]!                    \n"
-    "subs       %2, %2, #16                    \n"
-    "vsubl.u8   q2, d0, d2                     \n"
-    "vsubl.u8   q3, d1, d3                     \n"
-    "vmlal.s16  q8, d4, d4                     \n"
-    "vmlal.s16  q9, d6, d6                     \n"
-    "vmlal.s16  q10, d5, d5                    \n"
-    "vmlal.s16  q11, d7, d7                    \n"
-    "bgt        1b                             \n"
-
-    "vadd.u32   q8, q8, q9                     \n"
-    "vadd.u32   q10, q10, q11                  \n"
-    "vadd.u32   q11, q8, q10                   \n"
-    "vpaddl.u32 q1, q11                        \n"
-    "vadd.u64   d0, d2, d3                     \n"
-    "vmov.32    %3, d0[0]                      \n"
-    : "+r"(src_a),
-      "+r"(src_b),
-      "+r"(count),
-      "=r"(sse)
-    :
-    : "memory", "cc", "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11");
-  return sse;
-}
-
-#endif  // __ARM_NEON__
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/compare_posix.cc b/drivers/theoraplayer/src/YUV/libyuv/src/compare_posix.cc
deleted file mode 100755
index ac361190e8..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/compare_posix.cc
+++ /dev/null
@@ -1,158 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_X86) && (defined(__x86_64__) || defined(__i386__))
-
-uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
-  uint32 sse;
-  asm volatile (  // NOLINT
-    "pxor      %%xmm0,%%xmm0                   \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10, 0) ",%0          \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10, 1) ",%1          \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psubusb   %%xmm2,%%xmm1                   \n"
-    "psubusb   %%xmm3,%%xmm2                   \n"
-    "por       %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "punpckhbw %%xmm5,%%xmm2                   \n"
-    "pmaddwd   %%xmm1,%%xmm1                   \n"
-    "pmaddwd   %%xmm2,%%xmm2                   \n"
-    "paddd     %%xmm1,%%xmm0                   \n"
-    "paddd     %%xmm2,%%xmm0                   \n"
-    "jg        1b                              \n"
-
-    "pshufd    $0xee,%%xmm0,%%xmm1             \n"
-    "paddd     %%xmm1,%%xmm0                   \n"
-    "pshufd    $0x1,%%xmm0,%%xmm1              \n"
-    "paddd     %%xmm1,%%xmm0                   \n"
-    "movd      %%xmm0,%3                       \n"
-
-  : "+r"(src_a),      // %0
-    "+r"(src_b),      // %1
-    "+r"(count),      // %2
-    "=g"(sse)         // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );  // NOLINT
-  return sse;
-}
-
-#endif  // defined(__x86_64__) || defined(__i386__)
-
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(__x86_64__) || (defined(__i386__) && !defined(__pic__)))
-#define HAS_HASHDJB2_SSE41
-static uvec32 kHash16x33 = { 0x92d9e201, 0, 0, 0 };  // 33 ^ 16
-static uvec32 kHashMul0 = {
-  0x0c3525e1,  // 33 ^ 15
-  0xa3476dc1,  // 33 ^ 14
-  0x3b4039a1,  // 33 ^ 13
-  0x4f5f0981,  // 33 ^ 12
-};
-static uvec32 kHashMul1 = {
-  0x30f35d61,  // 33 ^ 11
-  0x855cb541,  // 33 ^ 10
-  0x040a9121,  // 33 ^ 9
-  0x747c7101,  // 33 ^ 8
-};
-static uvec32 kHashMul2 = {
-  0xec41d4e1,  // 33 ^ 7
-  0x4cfa3cc1,  // 33 ^ 6
-  0x025528a1,  // 33 ^ 5
-  0x00121881,  // 33 ^ 4
-};
-static uvec32 kHashMul3 = {
-  0x00008c61,  // 33 ^ 3
-  0x00000441,  // 33 ^ 2
-  0x00000021,  // 33 ^ 1
-  0x00000001,  // 33 ^ 0
-};
-
-uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
-  uint32 hash;
-  asm volatile (  // NOLINT
-    "movd      %2,%%xmm0                       \n"
-    "pxor      %%xmm7,%%xmm7                   \n"
-    "movdqa    %4,%%xmm6                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10, 0) ",%0          \n"
-    "pmulld    %%xmm6,%%xmm0                   \n"
-    "movdqa    %5,%%xmm5                       \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "punpcklwd %%xmm7,%%xmm3                   \n"
-    "pmulld    %%xmm5,%%xmm3                   \n"
-    "movdqa    %6,%%xmm5                       \n"
-    "movdqa    %%xmm2,%%xmm4                   \n"
-    "punpckhwd %%xmm7,%%xmm4                   \n"
-    "pmulld    %%xmm5,%%xmm4                   \n"
-    "movdqa    %7,%%xmm5                       \n"
-    "punpckhbw %%xmm7,%%xmm1                   \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "punpcklwd %%xmm7,%%xmm2                   \n"
-    "pmulld    %%xmm5,%%xmm2                   \n"
-    "movdqa    %8,%%xmm5                       \n"
-    "punpckhwd %%xmm7,%%xmm1                   \n"
-    "pmulld    %%xmm5,%%xmm1                   \n"
-    "paddd     %%xmm4,%%xmm3                   \n"
-    "paddd     %%xmm2,%%xmm1                   \n"
-    "sub       $0x10,%1                        \n"
-    "paddd     %%xmm3,%%xmm1                   \n"
-    "pshufd    $0xe,%%xmm1,%%xmm2              \n"
-    "paddd     %%xmm2,%%xmm1                   \n"
-    "pshufd    $0x1,%%xmm1,%%xmm2              \n"
-    "paddd     %%xmm2,%%xmm1                   \n"
-    "paddd     %%xmm1,%%xmm0                   \n"
-    "jg        1b                              \n"
-    "movd      %%xmm0,%3                       \n"
-  : "+r"(src),        // %0
-    "+r"(count),      // %1
-    "+rm"(seed),      // %2
-    "=g"(hash)        // %3
-  : "m"(kHash16x33),  // %4
-    "m"(kHashMul0),   // %5
-    "m"(kHashMul1),   // %6
-    "m"(kHashMul2),   // %7
-    "m"(kHashMul3)    // %8
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );  // NOLINT
-  return hash;
-}
-#endif  // defined(__x86_64__) || (defined(__i386__) && !defined(__pic__)))
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/compare_win.cc b/drivers/theoraplayer/src/YUV/libyuv/src/compare_win.cc
deleted file mode 100755
index 99831651f5..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/compare_win.cc
+++ /dev/null
@@ -1,232 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-__declspec(naked) __declspec(align(16))
-uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
-  __asm {
-    mov        eax, [esp + 4]    // src_a
-    mov        edx, [esp + 8]    // src_b
-    mov        ecx, [esp + 12]   // count
-    pxor       xmm0, xmm0
-    pxor       xmm5, xmm5
-
-    align      4
-  wloop:
-    movdqa     xmm1, [eax]
-    lea        eax,  [eax + 16]
-    movdqa     xmm2, [edx]
-    lea        edx,  [edx + 16]
-    sub        ecx, 16
-    movdqa     xmm3, xmm1  // abs trick
-    psubusb    xmm1, xmm2
-    psubusb    xmm2, xmm3
-    por        xmm1, xmm2
-    movdqa     xmm2, xmm1
-    punpcklbw  xmm1, xmm5
-    punpckhbw  xmm2, xmm5
-    pmaddwd    xmm1, xmm1
-    pmaddwd    xmm2, xmm2
-    paddd      xmm0, xmm1
-    paddd      xmm0, xmm2
-    jg         wloop
-
-    pshufd     xmm1, xmm0, 0xee
-    paddd      xmm0, xmm1
-    pshufd     xmm1, xmm0, 0x01
-    paddd      xmm0, xmm1
-    movd       eax, xmm0
-    ret
-  }
-}
-
-// Visual C 2012 required for AVX2.
-#if _MSC_VER >= 1700
-// C4752: found Intel(R) Advanced Vector Extensions; consider using /arch:AVX.
-#pragma warning(disable: 4752)
-__declspec(naked) __declspec(align(16))
-uint32 SumSquareError_AVX2(const uint8* src_a, const uint8* src_b, int count) {
-  __asm {
-    mov        eax, [esp + 4]    // src_a
-    mov        edx, [esp + 8]    // src_b
-    mov        ecx, [esp + 12]   // count
-    vpxor      ymm0, ymm0, ymm0  // sum
-    vpxor      ymm5, ymm5, ymm5  // constant 0 for unpck
-    sub        edx, eax
-
-    align      4
-  wloop:
-    vmovdqu    ymm1, [eax]
-    vmovdqu    ymm2, [eax + edx]
-    lea        eax,  [eax + 32]
-    sub        ecx, 32
-    vpsubusb   ymm3, ymm1, ymm2  // abs difference trick
-    vpsubusb   ymm2, ymm2, ymm1
-    vpor       ymm1, ymm2, ymm3
-    vpunpcklbw ymm2, ymm1, ymm5  // u16.  mutates order.
-    vpunpckhbw ymm1, ymm1, ymm5
-    vpmaddwd   ymm2, ymm2, ymm2  // square + hadd to u32.
-    vpmaddwd   ymm1, ymm1, ymm1
-    vpaddd     ymm0, ymm0, ymm1
-    vpaddd     ymm0, ymm0, ymm2
-    jg         wloop
-
-    vpshufd    ymm1, ymm0, 0xee  // 3, 2 + 1, 0 both lanes.
-    vpaddd     ymm0, ymm0, ymm1
-    vpshufd    ymm1, ymm0, 0x01  // 1 + 0 both lanes.
-    vpaddd     ymm0, ymm0, ymm1
-    vpermq     ymm1, ymm0, 0x02  // high + low lane.
-    vpaddd     ymm0, ymm0, ymm1
-    vmovd      eax, xmm0
-    vzeroupper
-    ret
-  }
-}
-#endif  // _MSC_VER >= 1700
-
-#define HAS_HASHDJB2_SSE41
-static uvec32 kHash16x33 = { 0x92d9e201, 0, 0, 0 };  // 33 ^ 16
-static uvec32 kHashMul0 = {
-  0x0c3525e1,  // 33 ^ 15
-  0xa3476dc1,  // 33 ^ 14
-  0x3b4039a1,  // 33 ^ 13
-  0x4f5f0981,  // 33 ^ 12
-};
-static uvec32 kHashMul1 = {
-  0x30f35d61,  // 33 ^ 11
-  0x855cb541,  // 33 ^ 10
-  0x040a9121,  // 33 ^ 9
-  0x747c7101,  // 33 ^ 8
-};
-static uvec32 kHashMul2 = {
-  0xec41d4e1,  // 33 ^ 7
-  0x4cfa3cc1,  // 33 ^ 6
-  0x025528a1,  // 33 ^ 5
-  0x00121881,  // 33 ^ 4
-};
-static uvec32 kHashMul3 = {
-  0x00008c61,  // 33 ^ 3
-  0x00000441,  // 33 ^ 2
-  0x00000021,  // 33 ^ 1
-  0x00000001,  // 33 ^ 0
-};
-
-// 27: 66 0F 38 40 C6     pmulld      xmm0,xmm6
-// 44: 66 0F 38 40 DD     pmulld      xmm3,xmm5
-// 59: 66 0F 38 40 E5     pmulld      xmm4,xmm5
-// 72: 66 0F 38 40 D5     pmulld      xmm2,xmm5
-// 83: 66 0F 38 40 CD     pmulld      xmm1,xmm5
-#define pmulld(reg) _asm _emit 0x66 _asm _emit 0x0F _asm _emit 0x38 \
-    _asm _emit 0x40 _asm _emit reg
-
-__declspec(naked) __declspec(align(16))
-uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
-  __asm {
-    mov        eax, [esp + 4]    // src
-    mov        ecx, [esp + 8]    // count
-    movd       xmm0, [esp + 12]  // seed
-
-    pxor       xmm7, xmm7        // constant 0 for unpck
-    movdqa     xmm6, kHash16x33
-
-    align      4
-  wloop:
-    movdqu     xmm1, [eax]       // src[0-15]
-    lea        eax, [eax + 16]
-    pmulld(0xc6)                 // pmulld      xmm0,xmm6  hash *= 33 ^ 16
-    movdqa     xmm5, kHashMul0
-    movdqa     xmm2, xmm1
-    punpcklbw  xmm2, xmm7        // src[0-7]
-    movdqa     xmm3, xmm2
-    punpcklwd  xmm3, xmm7        // src[0-3]
-    pmulld(0xdd)                 // pmulld     xmm3, xmm5
-    movdqa     xmm5, kHashMul1
-    movdqa     xmm4, xmm2
-    punpckhwd  xmm4, xmm7        // src[4-7]
-    pmulld(0xe5)                 // pmulld     xmm4, xmm5
-    movdqa     xmm5, kHashMul2
-    punpckhbw  xmm1, xmm7        // src[8-15]
-    movdqa     xmm2, xmm1
-    punpcklwd  xmm2, xmm7        // src[8-11]
-    pmulld(0xd5)                 // pmulld     xmm2, xmm5
-    movdqa     xmm5, kHashMul3
-    punpckhwd  xmm1, xmm7        // src[12-15]
-    pmulld(0xcd)                 // pmulld     xmm1, xmm5
-    paddd      xmm3, xmm4        // add 16 results
-    paddd      xmm1, xmm2
-    sub        ecx, 16
-    paddd      xmm1, xmm3
-
-    pshufd     xmm2, xmm1, 0x0e  // upper 2 dwords
-    paddd      xmm1, xmm2
-    pshufd     xmm2, xmm1, 0x01
-    paddd      xmm1, xmm2
-    paddd      xmm0, xmm1
-    jg         wloop
-
-    movd       eax, xmm0         // return hash
-    ret
-  }
-}
-
-// Visual C 2012 required for AVX2.
-#if _MSC_VER >= 1700
-__declspec(naked) __declspec(align(16))
-uint32 HashDjb2_AVX2(const uint8* src, int count, uint32 seed) {
-  __asm {
-    mov        eax, [esp + 4]    // src
-    mov        ecx, [esp + 8]    // count
-    movd       xmm0, [esp + 12]  // seed
-    movdqa     xmm6, kHash16x33
-
-    align      4
-  wloop:
-    vpmovzxbd  xmm3, dword ptr [eax]  // src[0-3]
-    pmulld     xmm0, xmm6  // hash *= 33 ^ 16
-    vpmovzxbd  xmm4, dword ptr [eax + 4]  // src[4-7]
-    pmulld     xmm3, kHashMul0
-    vpmovzxbd  xmm2, dword ptr [eax + 8]  // src[8-11]
-    pmulld     xmm4, kHashMul1
-    vpmovzxbd  xmm1, dword ptr [eax + 12]  // src[12-15]
-    pmulld     xmm2, kHashMul2
-    lea        eax, [eax + 16]
-    pmulld     xmm1, kHashMul3
-    paddd      xmm3, xmm4        // add 16 results
-    paddd      xmm1, xmm2
-    sub        ecx, 16
-    paddd      xmm1, xmm3
-    pshufd     xmm2, xmm1, 0x0e  // upper 2 dwords
-    paddd      xmm1, xmm2
-    pshufd     xmm2, xmm1, 0x01
-    paddd      xmm1, xmm2
-    paddd      xmm0, xmm1
-    jg         wloop
-
-    movd       eax, xmm0         // return hash
-    ret
-  }
-}
-#endif  // _MSC_VER >= 1700
-
-#endif  // !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert.cc
deleted file mode 100755
index c8408dc798..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert.cc
+++ /dev/null
@@ -1,1491 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert.h"
-
-#include "libyuv/basic_types.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/rotate.h"
-#include "libyuv/scale.h"  // For ScalePlane()
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#define SUBSAMPLE(v, a, s) (v < 0) ? (-((-v + a) >> s)) : ((v + a) >> s)
-static __inline int Abs(int v) {
-  return v >= 0 ? v : -v;
-}
-
-// Any I4xx To I420 format with mirroring.
-static int I4xxToI420(const uint8* src_y, int src_stride_y,
-                      const uint8* src_u, int src_stride_u,
-                      const uint8* src_v, int src_stride_v,
-                      uint8* dst_y, int dst_stride_y,
-                      uint8* dst_u, int dst_stride_u,
-                      uint8* dst_v, int dst_stride_v,
-                      int src_y_width, int src_y_height,
-                      int src_uv_width, int src_uv_height) {
-  if (src_y_width == 0 || src_y_height == 0 ||
-      src_uv_width == 0 || src_uv_height == 0) {
-    return -1;
-  }
-  const int dst_y_width = Abs(src_y_width);
-  const int dst_y_height = Abs(src_y_height);
-  const int dst_uv_width = SUBSAMPLE(dst_y_width, 1, 1);
-  const int dst_uv_height = SUBSAMPLE(dst_y_height, 1, 1);
-  ScalePlane(src_y, src_stride_y, src_y_width, src_y_height,
-             dst_y, dst_stride_y, dst_y_width, dst_y_height,
-             kFilterBilinear);
-  ScalePlane(src_u, src_stride_u, src_uv_width, src_uv_height,
-             dst_u, dst_stride_u, dst_uv_width, dst_uv_height,
-             kFilterBilinear);
-  ScalePlane(src_v, src_stride_v, src_uv_width, src_uv_height,
-             dst_v, dst_stride_v, dst_uv_width, dst_uv_height,
-             kFilterBilinear);
-  return 0;
-}
-
-// Copy I420 with optional flipping
-// TODO(fbarchard): Use Scale plane which supports mirroring, but ensure
-// is does row coalescing.
-LIBYUV_API
-int I420Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height) {
-  if (!src_y || !src_u || !src_v ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    const int halfheight = (height + 1) >> 1;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (halfheight - 1) * src_stride_u;
-    src_v = src_v + (halfheight - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-
-  if (dst_y) {
-    CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  }
-  // Copy UV planes.
-  const int halfwidth = (width + 1) >> 1;
-  const int halfheight = (height + 1) >> 1;
-  CopyPlane(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight);
-  CopyPlane(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight);
-  return 0;
-}
-
-// 422 chroma is 1/2 width, 1x height
-// 420 chroma is 1/2 width, 1/2 height
-LIBYUV_API
-int I422ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  const int src_uv_width = SUBSAMPLE(width, 1, 1);
-  return I4xxToI420(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    src_uv_width, height);
-}
-
-// 444 chroma is 1x width, 1x height
-// 420 chroma is 1/2 width, 1/2 height
-LIBYUV_API
-int I444ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  return I4xxToI420(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    width, height);
-}
-
-// 411 chroma is 1/4 width, 1x height
-// 420 chroma is 1/2 width, 1/2 height
-LIBYUV_API
-int I411ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  const int src_uv_width = SUBSAMPLE(width, 3, 2);
-  return I4xxToI420(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    src_uv_width, height);
-}
-
-// I400 is greyscale typically used in MJPG
-LIBYUV_API
-int I400ToI420(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_y || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  SetPlane(dst_u, dst_stride_u, halfwidth, halfheight, 128);
-  SetPlane(dst_v, dst_stride_v, halfwidth, halfheight, 128);
-  return 0;
-}
-
-static void CopyPlane2(const uint8* src, int src_stride_0, int src_stride_1,
-                       uint8* dst, int dst_stride,
-                       int width, int height) {
-  void (*CopyRow)(const uint8* src, uint8* dst, int width) = CopyRow_C;
-#if defined(HAS_COPYROW_X86)
-  if (TestCpuFlag(kCpuHasX86) && IS_ALIGNED(width, 4)) {
-    CopyRow = CopyRow_X86;
-  }
-#endif
-#if defined(HAS_COPYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 32) &&
-      IS_ALIGNED(src, 16) &&
-      IS_ALIGNED(src_stride_0, 16) && IS_ALIGNED(src_stride_1, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    CopyRow = CopyRow_SSE2;
-  }
-#endif
-#if defined(HAS_COPYROW_ERMS)
-  if (TestCpuFlag(kCpuHasERMS)) {
-    CopyRow = CopyRow_ERMS;
-  }
-#endif
-#if defined(HAS_COPYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 32)) {
-    CopyRow = CopyRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_MIPS)
-  if (TestCpuFlag(kCpuHasMIPS)) {
-    CopyRow = CopyRow_MIPS;
-  }
-#endif
-
-  // Copy plane
-  for (int y = 0; y < height - 1; y += 2) {
-    CopyRow(src, dst, width);
-    CopyRow(src + src_stride_0, dst + dst_stride, width);
-    src += src_stride_0 + src_stride_1;
-    dst += dst_stride * 2;
-  }
-  if (height & 1) {
-    CopyRow(src, dst, width);
-  }
-}
-
-// Support converting from FOURCC_M420
-// Useful for bandwidth constrained transports like USB 1.0 and 2.0 and for
-// easy conversion to I420.
-// M420 format description:
-// M420 is row biplanar 420: 2 rows of Y and 1 row of UV.
-// Chroma is half width / half height. (420)
-// src_stride_m420 is row planar. Normally this will be the width in pixels.
-//   The UV plane is half width, but 2 values, so src_stride_m420 applies to
-//   this as well as the two Y planes.
-static int X420ToI420(const uint8* src_y,
-                      int src_stride_y0, int src_stride_y1,
-                      const uint8* src_uv, int src_stride_uv,
-                      uint8* dst_y, int dst_stride_y,
-                      uint8* dst_u, int dst_stride_u,
-                      uint8* dst_v, int dst_stride_v,
-                      int width, int height) {
-  if (!src_y || !src_uv ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    int halfheight = (height + 1) >> 1;
-    dst_y = dst_y + (height - 1) * dst_stride_y;
-    dst_u = dst_u + (halfheight - 1) * dst_stride_u;
-    dst_v = dst_v + (halfheight - 1) * dst_stride_v;
-    dst_stride_y = -dst_stride_y;
-    dst_stride_u = -dst_stride_u;
-    dst_stride_v = -dst_stride_v;
-  }
-  // Coalesce rows.
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  if (src_stride_y0 == width &&
-      src_stride_y1 == width &&
-      dst_stride_y == width) {
-    width *= height;
-    height = 1;
-    src_stride_y0 = src_stride_y1 = dst_stride_y = 0;
-  }
-  // Coalesce rows.
-  if (src_stride_uv == halfwidth * 2 &&
-      dst_stride_u == halfwidth &&
-      dst_stride_v == halfwidth) {
-    halfwidth *= halfheight;
-    halfheight = 1;
-    src_stride_uv = dst_stride_u = dst_stride_v = 0;
-  }
-  void (*SplitUVRow)(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix) =
-      SplitUVRow_C;
-#if defined(HAS_SPLITUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    SplitUVRow = SplitUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      SplitUVRow = SplitUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_uv, 16) && IS_ALIGNED(src_stride_uv, 16) &&
-          IS_ALIGNED(dst_u, 16) && IS_ALIGNED(dst_stride_u, 16) &&
-          IS_ALIGNED(dst_v, 16) && IS_ALIGNED(dst_stride_v, 16)) {
-        SplitUVRow = SplitUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_SPLITUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    SplitUVRow = SplitUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      SplitUVRow = SplitUVRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_SPLITUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    SplitUVRow = SplitUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      SplitUVRow = SplitUVRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_SPLITUVROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && halfwidth >= 16) {
-    SplitUVRow = SplitUVRow_Any_MIPS_DSPR2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      SplitUVRow = SplitUVRow_Unaligned_MIPS_DSPR2;
-      if (IS_ALIGNED(src_uv, 4) && IS_ALIGNED(src_stride_uv, 4) &&
-          IS_ALIGNED(dst_u, 4) && IS_ALIGNED(dst_stride_u, 4) &&
-          IS_ALIGNED(dst_v, 4) && IS_ALIGNED(dst_stride_v, 4)) {
-        SplitUVRow = SplitUVRow_MIPS_DSPR2;
-      }
-    }
-  }
-#endif
-
-  if (dst_y) {
-    if (src_stride_y0 == src_stride_y1) {
-      CopyPlane(src_y, src_stride_y0, dst_y, dst_stride_y, width, height);
-    } else {
-      CopyPlane2(src_y, src_stride_y0, src_stride_y1, dst_y, dst_stride_y,
-                 width, height);
-    }
-  }
-
-  for (int y = 0; y < halfheight; ++y) {
-    // Copy a row of UV.
-    SplitUVRow(src_uv, dst_u, dst_v, halfwidth);
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-    src_uv += src_stride_uv;
-  }
-  return 0;
-}
-
-// Convert NV12 to I420.
-LIBYUV_API
-int NV12ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_uv, int src_stride_uv,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  return X420ToI420(src_y, src_stride_y, src_stride_y,
-                    src_uv, src_stride_uv,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height);
-}
-
-// Convert NV21 to I420.  Same as NV12 but u and v pointers swapped.
-LIBYUV_API
-int NV21ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_vu, int src_stride_vu,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  return X420ToI420(src_y, src_stride_y, src_stride_y,
-                    src_vu, src_stride_vu,
-                    dst_y, dst_stride_y,
-                    dst_v, dst_stride_v,
-                    dst_u, dst_stride_u,
-                    width, height);
-}
-
-// Convert M420 to I420.
-LIBYUV_API
-int M420ToI420(const uint8* src_m420, int src_stride_m420,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  return X420ToI420(src_m420, src_stride_m420, src_stride_m420 * 2,
-                    src_m420 + src_stride_m420 * 2, src_stride_m420 * 3,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height);
-}
-
-// Convert Q420 to I420.
-// Format is rows of YY/YUYV
-LIBYUV_API
-int Q420ToI420(const uint8* src_y, int src_stride_y,
-               const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_y || !src_yuy2 ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    int halfheight = (height + 1) >> 1;
-    dst_y = dst_y + (height - 1) * dst_stride_y;
-    dst_u = dst_u + (halfheight - 1) * dst_stride_u;
-    dst_v = dst_v + (halfheight - 1) * dst_stride_v;
-    dst_stride_y = -dst_stride_y;
-    dst_stride_u = -dst_stride_u;
-    dst_stride_v = -dst_stride_v;
-  }
-  // CopyRow for rows of just Y in Q420 copied to Y plane of I420.
-  void (*CopyRow)(const uint8* src, uint8* dst, int width) = CopyRow_C;
-#if defined(HAS_COPYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 32)) {
-    CopyRow = CopyRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_X86)
-  if (IS_ALIGNED(width, 4)) {
-    CopyRow = CopyRow_X86;
-  }
-#endif
-#if defined(HAS_COPYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 32) &&
-      IS_ALIGNED(src_y, 16) && IS_ALIGNED(src_stride_y, 16) &&
-      IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-    CopyRow = CopyRow_SSE2;
-  }
-#endif
-#if defined(HAS_COPYROW_ERMS)
-  if (TestCpuFlag(kCpuHasERMS)) {
-    CopyRow = CopyRow_ERMS;
-  }
-#endif
-#if defined(HAS_COPYROW_MIPS)
-  if (TestCpuFlag(kCpuHasMIPS)) {
-    CopyRow = CopyRow_MIPS;
-  }
-#endif
-
-  void (*YUY2ToUV422Row)(const uint8* src_yuy2, uint8* dst_u, uint8* dst_v,
-      int pix) = YUY2ToUV422Row_C;
-  void (*YUY2ToYRow)(const uint8* src_yuy2, uint8* dst_y, int pix) =
-      YUY2ToYRow_C;
-#if defined(HAS_YUY2TOYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    YUY2ToUV422Row = YUY2ToUV422Row_Any_SSE2;
-    YUY2ToYRow = YUY2ToYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToUV422Row = YUY2ToUV422Row_Unaligned_SSE2;
-      YUY2ToYRow = YUY2ToYRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_yuy2, 16) && IS_ALIGNED(src_stride_yuy2, 16)) {
-        YUY2ToUV422Row = YUY2ToUV422Row_SSE2;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          YUY2ToYRow = YUY2ToYRow_SSE2;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    YUY2ToUV422Row = YUY2ToUV422Row_Any_AVX2;
-    YUY2ToYRow = YUY2ToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      YUY2ToUV422Row = YUY2ToUV422Row_AVX2;
-      YUY2ToYRow = YUY2ToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    YUY2ToYRow = YUY2ToYRow_Any_NEON;
-    if (width >= 16) {
-      YUY2ToUV422Row = YUY2ToUV422Row_Any_NEON;
-    }
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToYRow = YUY2ToYRow_NEON;
-      YUY2ToUV422Row = YUY2ToUV422Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    CopyRow(src_y, dst_y, width);
-    src_y += src_stride_y;
-    dst_y += dst_stride_y;
-
-    YUY2ToUV422Row(src_yuy2, dst_u, dst_v, width);
-    YUY2ToYRow(src_yuy2, dst_y, width);
-    src_yuy2 += src_stride_yuy2;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    CopyRow(src_y, dst_y, width);
-    YUY2ToUV422Row(src_yuy2, dst_u, dst_v, width);
-  }
-  return 0;
-}
-
-// Convert YUY2 to I420.
-LIBYUV_API
-int YUY2ToI420(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_yuy2 = src_yuy2 + (height - 1) * src_stride_yuy2;
-    src_stride_yuy2 = -src_stride_yuy2;
-  }
-  void (*YUY2ToUVRow)(const uint8* src_yuy2, int src_stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix);
-  void (*YUY2ToYRow)(const uint8* src_yuy2,
-                     uint8* dst_y, int pix);
-  YUY2ToYRow = YUY2ToYRow_C;
-  YUY2ToUVRow = YUY2ToUVRow_C;
-#if defined(HAS_YUY2TOYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    YUY2ToUVRow = YUY2ToUVRow_Any_SSE2;
-    YUY2ToYRow = YUY2ToYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToUVRow = YUY2ToUVRow_Unaligned_SSE2;
-      YUY2ToYRow = YUY2ToYRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_yuy2, 16) && IS_ALIGNED(src_stride_yuy2, 16)) {
-        YUY2ToUVRow = YUY2ToUVRow_SSE2;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          YUY2ToYRow = YUY2ToYRow_SSE2;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    YUY2ToUVRow = YUY2ToUVRow_Any_AVX2;
-    YUY2ToYRow = YUY2ToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      YUY2ToUVRow = YUY2ToUVRow_AVX2;
-      YUY2ToYRow = YUY2ToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    YUY2ToYRow = YUY2ToYRow_Any_NEON;
-    if (width >= 16) {
-      YUY2ToUVRow = YUY2ToUVRow_Any_NEON;
-    }
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToYRow = YUY2ToYRow_NEON;
-      YUY2ToUVRow = YUY2ToUVRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    YUY2ToUVRow(src_yuy2, src_stride_yuy2, dst_u, dst_v, width);
-    YUY2ToYRow(src_yuy2, dst_y, width);
-    YUY2ToYRow(src_yuy2 + src_stride_yuy2, dst_y + dst_stride_y, width);
-    src_yuy2 += src_stride_yuy2 * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    YUY2ToUVRow(src_yuy2, 0, dst_u, dst_v, width);
-    YUY2ToYRow(src_yuy2, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert UYVY to I420.
-LIBYUV_API
-int UYVYToI420(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy;
-    src_stride_uyvy = -src_stride_uyvy;
-  }
-  void (*UYVYToUVRow)(const uint8* src_uyvy, int src_stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix);
-  void (*UYVYToYRow)(const uint8* src_uyvy,
-                     uint8* dst_y, int pix);
-  UYVYToYRow = UYVYToYRow_C;
-  UYVYToUVRow = UYVYToUVRow_C;
-#if defined(HAS_UYVYTOYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    UYVYToUVRow = UYVYToUVRow_Any_SSE2;
-    UYVYToYRow = UYVYToYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      UYVYToUVRow = UYVYToUVRow_Unaligned_SSE2;
-      UYVYToYRow = UYVYToYRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_uyvy, 16) && IS_ALIGNED(src_stride_uyvy, 16)) {
-        UYVYToUVRow = UYVYToUVRow_SSE2;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          UYVYToYRow = UYVYToYRow_SSE2;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_UYVYTOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    UYVYToUVRow = UYVYToUVRow_Any_AVX2;
-    UYVYToYRow = UYVYToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      UYVYToUVRow = UYVYToUVRow_AVX2;
-      UYVYToYRow = UYVYToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_UYVYTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    UYVYToYRow = UYVYToYRow_Any_NEON;
-    if (width >= 16) {
-      UYVYToUVRow = UYVYToUVRow_Any_NEON;
-    }
-    if (IS_ALIGNED(width, 16)) {
-      UYVYToYRow = UYVYToYRow_NEON;
-      UYVYToUVRow = UYVYToUVRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    UYVYToUVRow(src_uyvy, src_stride_uyvy, dst_u, dst_v, width);
-    UYVYToYRow(src_uyvy, dst_y, width);
-    UYVYToYRow(src_uyvy + src_stride_uyvy, dst_y + dst_stride_y, width);
-    src_uyvy += src_stride_uyvy * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    UYVYToUVRow(src_uyvy, 0, dst_u, dst_v, width);
-    UYVYToYRow(src_uyvy, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert ARGB to I420.
-LIBYUV_API
-int ARGBToI420(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3) && defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVRow = ARGBToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ARGBToYRow = ARGBToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_AVX2) && defined(HAS_ARGBTOUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    ARGBToUVRow = ARGBToUVRow_Any_AVX2;
-    ARGBToYRow = ARGBToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      ARGBToUVRow = ARGBToUVRow_AVX2;
-      ARGBToYRow = ARGBToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVRow(src_argb, src_stride_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
-    src_argb += src_stride_argb * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    ARGBToUVRow(src_argb, 0, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert BGRA to I420.
-LIBYUV_API
-int BGRAToI420(const uint8* src_bgra, int src_stride_bgra,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_bgra ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_bgra = src_bgra + (height - 1) * src_stride_bgra;
-    src_stride_bgra = -src_stride_bgra;
-  }
-  void (*BGRAToUVRow)(const uint8* src_bgra0, int src_stride_bgra,
-                      uint8* dst_u, uint8* dst_v, int width) = BGRAToUVRow_C;
-  void (*BGRAToYRow)(const uint8* src_bgra, uint8* dst_y, int pix) =
-      BGRAToYRow_C;
-#if defined(HAS_BGRATOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    BGRAToUVRow = BGRAToUVRow_Any_SSSE3;
-    BGRAToYRow = BGRAToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      BGRAToUVRow = BGRAToUVRow_Unaligned_SSSE3;
-      BGRAToYRow = BGRAToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_bgra, 16) && IS_ALIGNED(src_stride_bgra, 16)) {
-        BGRAToUVRow = BGRAToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          BGRAToYRow = BGRAToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_BGRATOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    BGRAToYRow = BGRAToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      BGRAToYRow = BGRAToYRow_NEON;
-    }
-    if (width >= 16) {
-      BGRAToUVRow = BGRAToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        BGRAToUVRow = BGRAToUVRow_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    BGRAToUVRow(src_bgra, src_stride_bgra, dst_u, dst_v, width);
-    BGRAToYRow(src_bgra, dst_y, width);
-    BGRAToYRow(src_bgra + src_stride_bgra, dst_y + dst_stride_y, width);
-    src_bgra += src_stride_bgra * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    BGRAToUVRow(src_bgra, 0, dst_u, dst_v, width);
-    BGRAToYRow(src_bgra, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert ABGR to I420.
-LIBYUV_API
-int ABGRToI420(const uint8* src_abgr, int src_stride_abgr,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_abgr ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_abgr = src_abgr + (height - 1) * src_stride_abgr;
-    src_stride_abgr = -src_stride_abgr;
-  }
-  void (*ABGRToUVRow)(const uint8* src_abgr0, int src_stride_abgr,
-                      uint8* dst_u, uint8* dst_v, int width) = ABGRToUVRow_C;
-  void (*ABGRToYRow)(const uint8* src_abgr, uint8* dst_y, int pix) =
-      ABGRToYRow_C;
-#if defined(HAS_ABGRTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ABGRToUVRow = ABGRToUVRow_Any_SSSE3;
-    ABGRToYRow = ABGRToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ABGRToUVRow = ABGRToUVRow_Unaligned_SSSE3;
-      ABGRToYRow = ABGRToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_abgr, 16) && IS_ALIGNED(src_stride_abgr, 16)) {
-        ABGRToUVRow = ABGRToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ABGRToYRow = ABGRToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_ABGRTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ABGRToYRow = ABGRToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ABGRToYRow = ABGRToYRow_NEON;
-    }
-    if (width >= 16) {
-      ABGRToUVRow = ABGRToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ABGRToUVRow = ABGRToUVRow_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ABGRToUVRow(src_abgr, src_stride_abgr, dst_u, dst_v, width);
-    ABGRToYRow(src_abgr, dst_y, width);
-    ABGRToYRow(src_abgr + src_stride_abgr, dst_y + dst_stride_y, width);
-    src_abgr += src_stride_abgr * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    ABGRToUVRow(src_abgr, 0, dst_u, dst_v, width);
-    ABGRToYRow(src_abgr, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert RGBA to I420.
-LIBYUV_API
-int RGBAToI420(const uint8* src_rgba, int src_stride_rgba,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_rgba ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_rgba = src_rgba + (height - 1) * src_stride_rgba;
-    src_stride_rgba = -src_stride_rgba;
-  }
-  void (*RGBAToUVRow)(const uint8* src_rgba0, int src_stride_rgba,
-                      uint8* dst_u, uint8* dst_v, int width) = RGBAToUVRow_C;
-  void (*RGBAToYRow)(const uint8* src_rgba, uint8* dst_y, int pix) =
-      RGBAToYRow_C;
-#if defined(HAS_RGBATOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    RGBAToUVRow = RGBAToUVRow_Any_SSSE3;
-    RGBAToYRow = RGBAToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      RGBAToUVRow = RGBAToUVRow_Unaligned_SSSE3;
-      RGBAToYRow = RGBAToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_rgba, 16) && IS_ALIGNED(src_stride_rgba, 16)) {
-        RGBAToUVRow = RGBAToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          RGBAToYRow = RGBAToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_RGBATOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RGBAToYRow = RGBAToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RGBAToYRow = RGBAToYRow_NEON;
-    }
-    if (width >= 16) {
-      RGBAToUVRow = RGBAToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        RGBAToUVRow = RGBAToUVRow_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    RGBAToUVRow(src_rgba, src_stride_rgba, dst_u, dst_v, width);
-    RGBAToYRow(src_rgba, dst_y, width);
-    RGBAToYRow(src_rgba + src_stride_rgba, dst_y + dst_stride_y, width);
-    src_rgba += src_stride_rgba * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    RGBAToUVRow(src_rgba, 0, dst_u, dst_v, width);
-    RGBAToYRow(src_rgba, dst_y, width);
-  }
-  return 0;
-}
-
-// Convert RGB24 to I420.
-LIBYUV_API
-int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
-                uint8* dst_y, int dst_stride_y,
-                uint8* dst_u, int dst_stride_u,
-                uint8* dst_v, int dst_stride_v,
-                int width, int height) {
-  if (!src_rgb24 || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_rgb24 = src_rgb24 + (height - 1) * src_stride_rgb24;
-    src_stride_rgb24 = -src_stride_rgb24;
-  }
-
-#if defined(HAS_RGB24TOYROW_NEON)
-  void (*RGB24ToUVRow)(const uint8* src_rgb24, int src_stride_rgb24,
-      uint8* dst_u, uint8* dst_v, int width) = RGB24ToUVRow_C;
-  void (*RGB24ToYRow)(const uint8* src_rgb24, uint8* dst_y, int pix) =
-      RGB24ToYRow_C;
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RGB24ToYRow = RGB24ToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RGB24ToYRow = RGB24ToYRow_NEON;
-    }
-    if (width >= 16) {
-      RGB24ToUVRow = RGB24ToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        RGB24ToUVRow = RGB24ToUVRow_NEON;
-      }
-    }
-  }
-#else  // HAS_RGB24TOYROW_NEON
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  void (*RGB24ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      RGB24ToARGBRow_C;
-#if defined(HAS_RGB24TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    RGB24ToARGBRow = RGB24ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      RGB24ToARGBRow = RGB24ToARGBRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif  // HAS_ARGBTOUVROW_SSSE3
-#endif  // HAS_RGB24TOYROW_NEON
-
-  for (int y = 0; y < height - 1; y += 2) {
-#if defined(HAS_RGB24TOYROW_NEON)
-    RGB24ToUVRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width);
-    RGB24ToYRow(src_rgb24, dst_y, width);
-    RGB24ToYRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width);
-#else
-    RGB24ToARGBRow(src_rgb24, row, width);
-    RGB24ToARGBRow(src_rgb24 + src_stride_rgb24, row + kRowSize, width);
-    ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-    ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-#endif
-    src_rgb24 += src_stride_rgb24 * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-#if defined(HAS_RGB24TOYROW_NEON)
-    RGB24ToUVRow(src_rgb24, 0, dst_u, dst_v, width);
-    RGB24ToYRow(src_rgb24, dst_y, width);
-#else
-    RGB24ToARGBRow(src_rgb24, row, width);
-    ARGBToUVRow(row, 0, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-#endif
-  }
-#if !defined(HAS_RGB24TOYROW_NEON)
-  free_aligned_buffer_64(row);
-#endif
-  return 0;
-}
-
-// Convert RAW to I420.
-LIBYUV_API
-int RAWToI420(const uint8* src_raw, int src_stride_raw,
-              uint8* dst_y, int dst_stride_y,
-              uint8* dst_u, int dst_stride_u,
-              uint8* dst_v, int dst_stride_v,
-              int width, int height) {
-  if (!src_raw || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_raw = src_raw + (height - 1) * src_stride_raw;
-    src_stride_raw = -src_stride_raw;
-  }
-
-#if defined(HAS_RAWTOYROW_NEON)
-  void (*RAWToUVRow)(const uint8* src_raw, int src_stride_raw,
-      uint8* dst_u, uint8* dst_v, int width) = RAWToUVRow_C;
-  void (*RAWToYRow)(const uint8* src_raw, uint8* dst_y, int pix) =
-      RAWToYRow_C;
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RAWToYRow = RAWToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RAWToYRow = RAWToYRow_NEON;
-    }
-    if (width >= 16) {
-      RAWToUVRow = RAWToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        RAWToUVRow = RAWToUVRow_NEON;
-      }
-    }
-  }
-#else  // HAS_RAWTOYROW_NEON
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  void (*RAWToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      RAWToARGBRow_C;
-#if defined(HAS_RAWTOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    RAWToARGBRow = RAWToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      RAWToARGBRow = RAWToARGBRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif  // HAS_ARGBTOUVROW_SSSE3
-#endif  // HAS_RAWTOYROW_NEON
-
-  for (int y = 0; y < height - 1; y += 2) {
-#if defined(HAS_RAWTOYROW_NEON)
-    RAWToUVRow(src_raw, src_stride_raw, dst_u, dst_v, width);
-    RAWToYRow(src_raw, dst_y, width);
-    RAWToYRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width);
-#else
-    RAWToARGBRow(src_raw, row, width);
-    RAWToARGBRow(src_raw + src_stride_raw, row + kRowSize, width);
-    ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-    ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-#endif
-    src_raw += src_stride_raw * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-#if defined(HAS_RAWTOYROW_NEON)
-    RAWToUVRow(src_raw, 0, dst_u, dst_v, width);
-    RAWToYRow(src_raw, dst_y, width);
-#else
-    RAWToARGBRow(src_raw, row, width);
-    ARGBToUVRow(row, 0, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-#endif
-  }
-#if !defined(HAS_RAWTOYROW_NEON)
-  free_aligned_buffer_64(row);
-#endif
-  return 0;
-}
-
-// Convert RGB565 to I420.
-LIBYUV_API
-int RGB565ToI420(const uint8* src_rgb565, int src_stride_rgb565,
-                uint8* dst_y, int dst_stride_y,
-                uint8* dst_u, int dst_stride_u,
-                uint8* dst_v, int dst_stride_v,
-                int width, int height) {
-  if (!src_rgb565 || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_rgb565 = src_rgb565 + (height - 1) * src_stride_rgb565;
-    src_stride_rgb565 = -src_stride_rgb565;
-  }
-
-#if defined(HAS_RGB565TOYROW_NEON)
-  void (*RGB565ToUVRow)(const uint8* src_rgb565, int src_stride_rgb565,
-      uint8* dst_u, uint8* dst_v, int width) = RGB565ToUVRow_C;
-  void (*RGB565ToYRow)(const uint8* src_rgb565, uint8* dst_y, int pix) =
-      RGB565ToYRow_C;
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RGB565ToYRow = RGB565ToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RGB565ToYRow = RGB565ToYRow_NEON;
-    }
-    if (width >= 16) {
-      RGB565ToUVRow = RGB565ToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        RGB565ToUVRow = RGB565ToUVRow_NEON;
-      }
-    }
-  }
-#else  // HAS_RGB565TOYROW_NEON
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  void (*RGB565ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      RGB565ToARGBRow_C;
-#if defined(HAS_RGB565TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8) {
-    RGB565ToARGBRow = RGB565ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      RGB565ToARGBRow = RGB565ToARGBRow_SSE2;
-    }
-  }
-#endif
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif  // HAS_ARGBTOUVROW_SSSE3
-#endif  // HAS_RGB565TOYROW_NEON
-
-  for (int y = 0; y < height - 1; y += 2) {
-#if defined(HAS_RGB565TOYROW_NEON)
-    RGB565ToUVRow(src_rgb565, src_stride_rgb565, dst_u, dst_v, width);
-    RGB565ToYRow(src_rgb565, dst_y, width);
-    RGB565ToYRow(src_rgb565 + src_stride_rgb565, dst_y + dst_stride_y, width);
-#else
-    RGB565ToARGBRow(src_rgb565, row, width);
-    RGB565ToARGBRow(src_rgb565 + src_stride_rgb565, row + kRowSize, width);
-    ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-    ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-#endif
-    src_rgb565 += src_stride_rgb565 * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-#if defined(HAS_RGB565TOYROW_NEON)
-    RGB565ToUVRow(src_rgb565, 0, dst_u, dst_v, width);
-    RGB565ToYRow(src_rgb565, dst_y, width);
-#else
-    RGB565ToARGBRow(src_rgb565, row, width);
-    ARGBToUVRow(row, 0, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-#endif
-  }
-#if !defined(HAS_RGB565TOYROW_NEON)
-  free_aligned_buffer_64(row);
-#endif
-  return 0;
-}
-
-// Convert ARGB1555 to I420.
-LIBYUV_API
-int ARGB1555ToI420(const uint8* src_argb1555, int src_stride_argb1555,
-                   uint8* dst_y, int dst_stride_y,
-                   uint8* dst_u, int dst_stride_u,
-                   uint8* dst_v, int dst_stride_v,
-                   int width, int height) {
-  if (!src_argb1555 || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb1555 = src_argb1555 + (height - 1) * src_stride_argb1555;
-    src_stride_argb1555 = -src_stride_argb1555;
-  }
-
-#if defined(HAS_ARGB1555TOYROW_NEON)
-  void (*ARGB1555ToUVRow)(const uint8* src_argb1555, int src_stride_argb1555,
-      uint8* dst_u, uint8* dst_v, int width) = ARGB1555ToUVRow_C;
-  void (*ARGB1555ToYRow)(const uint8* src_argb1555, uint8* dst_y, int pix) =
-      ARGB1555ToYRow_C;
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGB1555ToYRow = ARGB1555ToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB1555ToYRow = ARGB1555ToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGB1555ToUVRow = ARGB1555ToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGB1555ToUVRow = ARGB1555ToUVRow_NEON;
-      }
-    }
-  }
-#else  // HAS_ARGB1555TOYROW_NEON
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  void (*ARGB1555ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      ARGB1555ToARGBRow_C;
-#if defined(HAS_ARGB1555TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8) {
-    ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB1555ToARGBRow = ARGB1555ToARGBRow_SSE2;
-    }
-  }
-#endif
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif  // HAS_ARGBTOUVROW_SSSE3
-#endif  // HAS_ARGB1555TOYROW_NEON
-
-  for (int y = 0; y < height - 1; y += 2) {
-#if defined(HAS_ARGB1555TOYROW_NEON)
-    ARGB1555ToUVRow(src_argb1555, src_stride_argb1555, dst_u, dst_v, width);
-    ARGB1555ToYRow(src_argb1555, dst_y, width);
-    ARGB1555ToYRow(src_argb1555 + src_stride_argb1555, dst_y + dst_stride_y,
-                   width);
-#else
-    ARGB1555ToARGBRow(src_argb1555, row, width);
-    ARGB1555ToARGBRow(src_argb1555 + src_stride_argb1555, row + kRowSize,
-                      width);
-    ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-    ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-#endif
-    src_argb1555 += src_stride_argb1555 * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-#if defined(HAS_ARGB1555TOYROW_NEON)
-    ARGB1555ToUVRow(src_argb1555, 0, dst_u, dst_v, width);
-    ARGB1555ToYRow(src_argb1555, dst_y, width);
-#else
-    ARGB1555ToARGBRow(src_argb1555, row, width);
-    ARGBToUVRow(row, 0, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-#endif
-  }
-#if !defined(HAS_ARGB1555TOYROW_NEON)
-  free_aligned_buffer_64(row);
-#endif
-  return 0;
-}
-
-// Convert ARGB4444 to I420.
-LIBYUV_API
-int ARGB4444ToI420(const uint8* src_argb4444, int src_stride_argb4444,
-                   uint8* dst_y, int dst_stride_y,
-                   uint8* dst_u, int dst_stride_u,
-                   uint8* dst_v, int dst_stride_v,
-                   int width, int height) {
-  if (!src_argb4444 || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb4444 = src_argb4444 + (height - 1) * src_stride_argb4444;
-    src_stride_argb4444 = -src_stride_argb4444;
-  }
-
-#if defined(HAS_ARGB4444TOYROW_NEON)
-  void (*ARGB4444ToUVRow)(const uint8* src_argb4444, int src_stride_argb4444,
-      uint8* dst_u, uint8* dst_v, int width) = ARGB4444ToUVRow_C;
-  void (*ARGB4444ToYRow)(const uint8* src_argb4444, uint8* dst_y, int pix) =
-      ARGB4444ToYRow_C;
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGB4444ToYRow = ARGB4444ToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB4444ToYRow = ARGB4444ToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGB4444ToUVRow = ARGB4444ToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGB4444ToUVRow = ARGB4444ToUVRow_NEON;
-      }
-    }
-  }
-#else  // HAS_ARGB4444TOYROW_NEON
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  void (*ARGB4444ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      ARGB4444ToARGBRow_C;
-#if defined(HAS_ARGB4444TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8) {
-    ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB4444ToARGBRow = ARGB4444ToARGBRow_SSE2;
-    }
-  }
-#endif
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif  // HAS_ARGBTOUVROW_SSSE3
-#endif  // HAS_ARGB4444TOYROW_NEON
-
-  for (int y = 0; y < height - 1; y += 2) {
-#if defined(HAS_ARGB4444TOYROW_NEON)
-    ARGB4444ToUVRow(src_argb4444, src_stride_argb4444, dst_u, dst_v, width);
-    ARGB4444ToYRow(src_argb4444, dst_y, width);
-    ARGB4444ToYRow(src_argb4444 + src_stride_argb4444, dst_y + dst_stride_y,
-                   width);
-#else
-    ARGB4444ToARGBRow(src_argb4444, row, width);
-    ARGB4444ToARGBRow(src_argb4444 + src_stride_argb4444, row + kRowSize,
-                      width);
-    ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-    ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-#endif
-    src_argb4444 += src_stride_argb4444 * 2;
-    dst_y += dst_stride_y * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-#if defined(HAS_ARGB4444TOYROW_NEON)
-    ARGB4444ToUVRow(src_argb4444, 0, dst_u, dst_v, width);
-    ARGB4444ToYRow(src_argb4444, dst_y, width);
-#else
-    ARGB4444ToARGBRow(src_argb4444, row, width);
-    ARGBToUVRow(row, 0, dst_u, dst_v, width);
-    ARGBToYRow(row, dst_y, width);
-#endif
-  }
-#if !defined(HAS_ARGB4444TOYROW_NEON)
-  free_aligned_buffer_64(row);
-#endif
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_argb.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_argb.cc
deleted file mode 100755
index a8aab91478..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_argb.cc
+++ /dev/null
@@ -1,901 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert_argb.h"
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#ifdef HAVE_JPEG
-#include "libyuv/mjpeg_decoder.h"
-#endif
-#include "libyuv/rotate_argb.h"
-#include "libyuv/row.h"
-#include "libyuv/video_common.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Copy ARGB with optional flipping
-LIBYUV_API
-int ARGBCopy(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int width, int height) {
-  if (!src_argb || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-
-  CopyPlane(src_argb, src_stride_argb, dst_argb, dst_stride_argb,
-            width * 4, height);
-  return 0;
-}
-
-// Convert I444 to ARGB.
-LIBYUV_API
-int I444ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v ||
-      !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u == width &&
-      src_stride_v == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0;
-  }
-  void (*I444ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I444ToARGBRow_C;
-#if defined(HAS_I444TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I444ToARGBRow = I444ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I444ToARGBRow = I444ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I444ToARGBRow = I444ToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I444TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I444ToARGBRow = I444ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I444ToARGBRow = I444ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I444ToARGBRow(src_y, src_u, src_v, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert I422 to ARGB.
-LIBYUV_API
-int I422ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v ||
-      !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0;
-  }
-  void (*I422ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToARGBRow_C;
-#if defined(HAS_I422TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I422ToARGBRow = I422ToARGBRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 16) {
-    I422ToARGBRow = I422ToARGBRow_Any_AVX2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToARGBRow = I422ToARGBRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2) &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride_argb, 4)) {
-    I422ToARGBRow = I422ToARGBRow_MIPS_DSPR2;
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToARGBRow(src_y, src_u, src_v, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert I411 to ARGB.
-LIBYUV_API
-int I411ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v ||
-      !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 4 == width &&
-      src_stride_v * 4 == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0;
-  }
-  void (*I411ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I411ToARGBRow_C;
-#if defined(HAS_I411TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I411ToARGBRow = I411ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I411ToARGBRow = I411ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I411ToARGBRow = I411ToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I411TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I411ToARGBRow = I411ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I411ToARGBRow = I411ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I411ToARGBRow(src_y, src_u, src_v, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert I400 to ARGB.
-LIBYUV_API
-int I400ToARGB_Reference(const uint8* src_y, int src_stride_y,
-                         uint8* dst_argb, int dst_stride_argb,
-                         int width, int height) {
-  if (!src_y || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = dst_stride_argb = 0;
-  }
-  void (*YToARGBRow)(const uint8* y_buf,
-                     uint8* rgb_buf,
-                     int width) = YToARGBRow_C;
-#if defined(HAS_YTOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    YToARGBRow = YToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      YToARGBRow = YToARGBRow_SSE2;
-    }
-  }
-#elif defined(HAS_YTOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    YToARGBRow = YToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      YToARGBRow = YToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    YToARGBRow(src_y, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-  }
-  return 0;
-}
-
-// Convert I400 to ARGB.
-LIBYUV_API
-int I400ToARGB(const uint8* src_y, int src_stride_y,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = dst_stride_argb = 0;
-  }
-  void (*I400ToARGBRow)(const uint8* src_y, uint8* dst_argb, int pix) =
-      I400ToARGBRow_C;
-#if defined(HAS_I400TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8) {
-    I400ToARGBRow = I400ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      I400ToARGBRow = I400ToARGBRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I400ToARGBRow = I400ToARGBRow_SSE2;
-      }
-    }
-  }
-#elif defined(HAS_I400TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I400ToARGBRow = I400ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I400ToARGBRow = I400ToARGBRow_NEON;
-    }
-  }
-#endif
-  for (int y = 0; y < height; ++y) {
-    I400ToARGBRow(src_y, dst_argb, width);
-    src_y += src_stride_y;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Shuffle table for converting BGRA to ARGB.
-static uvec8 kShuffleMaskBGRAToARGB = {
-  3u, 2u, 1u, 0u, 7u, 6u, 5u, 4u, 11u, 10u, 9u, 8u, 15u, 14u, 13u, 12u
-};
-
-// Shuffle table for converting ABGR to ARGB.
-static uvec8 kShuffleMaskABGRToARGB = {
-  2u, 1u, 0u, 3u, 6u, 5u, 4u, 7u, 10u, 9u, 8u, 11u, 14u, 13u, 12u, 15u
-};
-
-// Shuffle table for converting RGBA to ARGB.
-static uvec8 kShuffleMaskRGBAToARGB = {
-  1u, 2u, 3u, 0u, 5u, 6u, 7u, 4u, 9u, 10u, 11u, 8u, 13u, 14u, 15u, 12u
-};
-
-// Convert BGRA to ARGB.
-LIBYUV_API
-int BGRAToARGB(const uint8* src_bgra, int src_stride_bgra,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  return ARGBShuffle(src_bgra, src_stride_bgra,
-                     dst_argb, dst_stride_argb,
-                     (const uint8*)(&kShuffleMaskBGRAToARGB),
-                     width, height);
-}
-
-// Convert ABGR to ARGB.
-LIBYUV_API
-int ABGRToARGB(const uint8* src_abgr, int src_stride_abgr,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  return ARGBShuffle(src_abgr, src_stride_abgr,
-                     dst_argb, dst_stride_argb,
-                     (const uint8*)(&kShuffleMaskABGRToARGB),
-                     width, height);
-}
-
-// Convert RGBA to ARGB.
-LIBYUV_API
-int RGBAToARGB(const uint8* src_rgba, int src_stride_rgba,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  return ARGBShuffle(src_rgba, src_stride_rgba,
-                     dst_argb, dst_stride_argb,
-                     (const uint8*)(&kShuffleMaskRGBAToARGB),
-                     width, height);
-}
-
-// Convert RGB24 to ARGB.
-LIBYUV_API
-int RGB24ToARGB(const uint8* src_rgb24, int src_stride_rgb24,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height) {
-  if (!src_rgb24 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_rgb24 = src_rgb24 + (height - 1) * src_stride_rgb24;
-    src_stride_rgb24 = -src_stride_rgb24;
-  }
-  // Coalesce rows.
-  if (src_stride_rgb24 == width * 3 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_rgb24 = dst_stride_argb = 0;
-  }
-  void (*RGB24ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      RGB24ToARGBRow_C;
-#if defined(HAS_RGB24TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    RGB24ToARGBRow = RGB24ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      RGB24ToARGBRow = RGB24ToARGBRow_SSSE3;
-    }
-  }
-#elif defined(HAS_RGB24TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RGB24ToARGBRow = RGB24ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RGB24ToARGBRow = RGB24ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    RGB24ToARGBRow(src_rgb24, dst_argb, width);
-    src_rgb24 += src_stride_rgb24;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert RAW to ARGB.
-LIBYUV_API
-int RAWToARGB(const uint8* src_raw, int src_stride_raw,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height) {
-  if (!src_raw || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_raw = src_raw + (height - 1) * src_stride_raw;
-    src_stride_raw = -src_stride_raw;
-  }
-  // Coalesce rows.
-  if (src_stride_raw == width * 3 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_raw = dst_stride_argb = 0;
-  }
-  void (*RAWToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
-      RAWToARGBRow_C;
-#if defined(HAS_RAWTOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    RAWToARGBRow = RAWToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      RAWToARGBRow = RAWToARGBRow_SSSE3;
-    }
-  }
-#elif defined(HAS_RAWTOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RAWToARGBRow = RAWToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RAWToARGBRow = RAWToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    RAWToARGBRow(src_raw, dst_argb, width);
-    src_raw += src_stride_raw;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert RGB565 to ARGB.
-LIBYUV_API
-int RGB565ToARGB(const uint8* src_rgb565, int src_stride_rgb565,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height) {
-  if (!src_rgb565 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_rgb565 = src_rgb565 + (height - 1) * src_stride_rgb565;
-    src_stride_rgb565 = -src_stride_rgb565;
-  }
-  // Coalesce rows.
-  if (src_stride_rgb565 == width * 2 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_rgb565 = dst_stride_argb = 0;
-  }
-  void (*RGB565ToARGBRow)(const uint8* src_rgb565, uint8* dst_argb, int pix) =
-      RGB565ToARGBRow_C;
-#if defined(HAS_RGB565TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    RGB565ToARGBRow = RGB565ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      RGB565ToARGBRow = RGB565ToARGBRow_SSE2;
-    }
-  }
-#elif defined(HAS_RGB565TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    RGB565ToARGBRow = RGB565ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      RGB565ToARGBRow = RGB565ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    RGB565ToARGBRow(src_rgb565, dst_argb, width);
-    src_rgb565 += src_stride_rgb565;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert ARGB1555 to ARGB.
-LIBYUV_API
-int ARGB1555ToARGB(const uint8* src_argb1555, int src_stride_argb1555,
-                   uint8* dst_argb, int dst_stride_argb,
-                   int width, int height) {
-  if (!src_argb1555 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb1555 = src_argb1555 + (height - 1) * src_stride_argb1555;
-    src_stride_argb1555 = -src_stride_argb1555;
-  }
-  // Coalesce rows.
-  if (src_stride_argb1555 == width * 2 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb1555 = dst_stride_argb = 0;
-  }
-  void (*ARGB1555ToARGBRow)(const uint8* src_argb1555, uint8* dst_argb,
-                            int pix) = ARGB1555ToARGBRow_C;
-#if defined(HAS_ARGB1555TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB1555ToARGBRow = ARGB1555ToARGBRow_SSE2;
-    }
-  }
-#elif defined(HAS_ARGB1555TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB1555ToARGBRow = ARGB1555ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGB1555ToARGBRow(src_argb1555, dst_argb, width);
-    src_argb1555 += src_stride_argb1555;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert ARGB4444 to ARGB.
-LIBYUV_API
-int ARGB4444ToARGB(const uint8* src_argb4444, int src_stride_argb4444,
-                   uint8* dst_argb, int dst_stride_argb,
-                   int width, int height) {
-  if (!src_argb4444 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb4444 = src_argb4444 + (height - 1) * src_stride_argb4444;
-    src_stride_argb4444 = -src_stride_argb4444;
-  }
-  // Coalesce rows.
-  if (src_stride_argb4444 == width * 2 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb4444 = dst_stride_argb = 0;
-  }
-  void (*ARGB4444ToARGBRow)(const uint8* src_argb4444, uint8* dst_argb,
-                            int pix) = ARGB4444ToARGBRow_C;
-#if defined(HAS_ARGB4444TOARGBROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB4444ToARGBRow = ARGB4444ToARGBRow_SSE2;
-    }
-  }
-#elif defined(HAS_ARGB4444TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGB4444ToARGBRow = ARGB4444ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGB4444ToARGBRow(src_argb4444, dst_argb, width);
-    src_argb4444 += src_stride_argb4444;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert NV12 to ARGB.
-LIBYUV_API
-int NV12ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_uv, int src_stride_uv,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_uv || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  void (*NV12ToARGBRow)(const uint8* y_buf,
-                        const uint8* uv_buf,
-                        uint8* rgb_buf,
-                        int width) = NV12ToARGBRow_C;
-#if defined(HAS_NV12TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    NV12ToARGBRow = NV12ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToARGBRow = NV12ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        NV12ToARGBRow = NV12ToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_NV12TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    NV12ToARGBRow = NV12ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToARGBRow = NV12ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    NV12ToARGBRow(src_y, src_uv, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_uv += src_stride_uv;
-    }
-  }
-  return 0;
-}
-
-// Convert NV21 to ARGB.
-LIBYUV_API
-int NV21ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_uv, int src_stride_uv,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_uv || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  void (*NV21ToARGBRow)(const uint8* y_buf,
-                        const uint8* uv_buf,
-                        uint8* rgb_buf,
-                        int width) = NV21ToARGBRow_C;
-#if defined(HAS_NV21TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    NV21ToARGBRow = NV21ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      NV21ToARGBRow = NV21ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        NV21ToARGBRow = NV21ToARGBRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_NV21TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    NV21ToARGBRow = NV21ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      NV21ToARGBRow = NV21ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    NV21ToARGBRow(src_y, src_uv, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_uv += src_stride_uv;
-    }
-  }
-  return 0;
-}
-
-// Convert M420 to ARGB.
-LIBYUV_API
-int M420ToARGB(const uint8* src_m420, int src_stride_m420,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_m420 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  void (*NV12ToARGBRow)(const uint8* y_buf,
-                        const uint8* uv_buf,
-                        uint8* rgb_buf,
-                        int width) = NV12ToARGBRow_C;
-#if defined(HAS_NV12TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    NV12ToARGBRow = NV12ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToARGBRow = NV12ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        NV12ToARGBRow = NV12ToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_NV12TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    NV12ToARGBRow = NV12ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToARGBRow = NV12ToARGBRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    NV12ToARGBRow(src_m420, src_m420 + src_stride_m420 * 2, dst_argb, width);
-    NV12ToARGBRow(src_m420 + src_stride_m420, src_m420 + src_stride_m420 * 2,
-                  dst_argb + dst_stride_argb, width);
-    dst_argb += dst_stride_argb * 2;
-    src_m420 += src_stride_m420 * 3;
-  }
-  if (height & 1) {
-    NV12ToARGBRow(src_m420, src_m420 + src_stride_m420 * 2, dst_argb, width);
-  }
-  return 0;
-}
-
-// Convert YUY2 to ARGB.
-LIBYUV_API
-int YUY2ToARGB(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_yuy2 || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_yuy2 = src_yuy2 + (height - 1) * src_stride_yuy2;
-    src_stride_yuy2 = -src_stride_yuy2;
-  }
-  // Coalesce rows.
-  if (src_stride_yuy2 == width * 2 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_yuy2 = dst_stride_argb = 0;
-  }
-  void (*YUY2ToARGBRow)(const uint8* src_yuy2, uint8* dst_argb, int pix) =
-      YUY2ToARGBRow_C;
-#if defined(HAS_YUY2TOARGBROW_SSSE3)
-  // Posix is 16, Windows is 8.
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    YUY2ToARGBRow = YUY2ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToARGBRow = YUY2ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_yuy2, 16) && IS_ALIGNED(src_stride_yuy2, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        YUY2ToARGBRow = YUY2ToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_YUY2TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    YUY2ToARGBRow = YUY2ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      YUY2ToARGBRow = YUY2ToARGBRow_NEON;
-    }
-  }
-#endif
-  for (int y = 0; y < height; ++y) {
-    YUY2ToARGBRow(src_yuy2, dst_argb, width);
-    src_yuy2 += src_stride_yuy2;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert UYVY to ARGB.
-LIBYUV_API
-int UYVYToARGB(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_uyvy || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy;
-    src_stride_uyvy = -src_stride_uyvy;
-  }
-  // Coalesce rows.
-  if (src_stride_uyvy == width * 2 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_uyvy = dst_stride_argb = 0;
-  }
-  void (*UYVYToARGBRow)(const uint8* src_uyvy, uint8* dst_argb, int pix) =
-      UYVYToARGBRow_C;
-#if defined(HAS_UYVYTOARGBROW_SSSE3)
-  // Posix is 16, Windows is 8.
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    UYVYToARGBRow = UYVYToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      UYVYToARGBRow = UYVYToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_uyvy, 16) && IS_ALIGNED(src_stride_uyvy, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        UYVYToARGBRow = UYVYToARGBRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_UYVYTOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    UYVYToARGBRow = UYVYToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      UYVYToARGBRow = UYVYToARGBRow_NEON;
-    }
-  }
-#endif
-  for (int y = 0; y < height; ++y) {
-    UYVYToARGBRow(src_uyvy, dst_argb, width);
-    src_uyvy += src_stride_uyvy;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_from.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_from.cc
deleted file mode 100755
index 1e10832856..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_from.cc
+++ /dev/null
@@ -1,1196 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert_from.h"
-
-#include "libyuv/basic_types.h"
-#include "libyuv/convert.h"  // For I420Copy
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/rotate.h"
-#include "libyuv/scale.h"  // For ScalePlane()
-#include "libyuv/video_common.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#define SUBSAMPLE(v, a, s) (v < 0) ? (-((-v + a) >> s)) : ((v + a) >> s)
-static __inline int Abs(int v) {
-  return v >= 0 ? v : -v;
-}
-
-// I420 To any I4xx YUV format with mirroring.
-static int I420ToI4xx(const uint8* src_y, int src_stride_y,
-                      const uint8* src_u, int src_stride_u,
-                      const uint8* src_v, int src_stride_v,
-                      uint8* dst_y, int dst_stride_y,
-                      uint8* dst_u, int dst_stride_u,
-                      uint8* dst_v, int dst_stride_v,
-                      int src_y_width, int src_y_height,
-                      int dst_uv_width, int dst_uv_height) {
-  if (src_y_width == 0 || src_y_height == 0 ||
-      dst_uv_width <= 0 || dst_uv_height <= 0) {
-    return -1;
-  }
-  const int dst_y_width = Abs(src_y_width);
-  const int dst_y_height = Abs(src_y_height);
-  const int src_uv_width = SUBSAMPLE(src_y_width, 1, 1);
-  const int src_uv_height = SUBSAMPLE(src_y_height, 1, 1);
-  ScalePlane(src_y, src_stride_y, src_y_width, src_y_height,
-             dst_y, dst_stride_y, dst_y_width, dst_y_height,
-             kFilterBilinear);
-  ScalePlane(src_u, src_stride_u, src_uv_width, src_uv_height,
-             dst_u, dst_stride_u, dst_uv_width, dst_uv_height,
-             kFilterBilinear);
-  ScalePlane(src_v, src_stride_v, src_uv_width, src_uv_height,
-             dst_v, dst_stride_v, dst_uv_width, dst_uv_height,
-             kFilterBilinear);
-  return 0;
-}
-
-// 420 chroma is 1/2 width, 1/2 height
-// 422 chroma is 1/2 width, 1x height
-LIBYUV_API
-int I420ToI422(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  const int dst_uv_width = (Abs(width) + 1) >> 1;
-  const int dst_uv_height = Abs(height);
-  return I420ToI4xx(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    dst_uv_width, dst_uv_height);
-}
-
-// 420 chroma is 1/2 width, 1/2 height
-// 444 chroma is 1x width, 1x height
-LIBYUV_API
-int I420ToI444(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  const int dst_uv_width = Abs(width);
-  const int dst_uv_height = Abs(height);
-  return I420ToI4xx(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    dst_uv_width, dst_uv_height);
-}
-
-// 420 chroma is 1/2 width, 1/2 height
-// 411 chroma is 1/4 width, 1x height
-LIBYUV_API
-int I420ToI411(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  const int dst_uv_width = (Abs(width) + 3) >> 2;
-  const int dst_uv_height = Abs(height);
-  return I420ToI4xx(src_y, src_stride_y,
-                    src_u, src_stride_u,
-                    src_v, src_stride_v,
-                    dst_y, dst_stride_y,
-                    dst_u, dst_stride_u,
-                    dst_v, dst_stride_v,
-                    width, height,
-                    dst_uv_width, dst_uv_height);
-}
-
-// Copy to I400. Source can be I420,422,444,400,NV12,NV21
-LIBYUV_API
-int I400Copy(const uint8* src_y, int src_stride_y,
-             uint8* dst_y, int dst_stride_y,
-             int width, int height) {
-  if (!src_y || !dst_y ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  return 0;
-}
-
-LIBYUV_API
-int I422ToYUY2(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_yuy2, int dst_stride_yuy2,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_yuy2 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
-    dst_stride_yuy2 = -dst_stride_yuy2;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_yuy2 == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_yuy2 = 0;
-  }
-  void (*I422ToYUY2Row)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_yuy2, int width) =
-      I422ToYUY2Row_C;
-#if defined(HAS_I422TOYUY2ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOYUY2ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToYUY2Row(src_y, src_u, src_v, dst_yuy2, width);
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-    dst_yuy2 += dst_stride_yuy2;
-  }
-  return 0;
-}
-
-LIBYUV_API
-int I420ToYUY2(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_yuy2, int dst_stride_yuy2,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_yuy2 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
-    dst_stride_yuy2 = -dst_stride_yuy2;
-  }
-  void (*I422ToYUY2Row)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_yuy2, int width) =
-      I422ToYUY2Row_C;
-#if defined(HAS_I422TOYUY2ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOYUY2ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    I422ToYUY2Row(src_y, src_u, src_v, dst_yuy2, width);
-    I422ToYUY2Row(src_y + src_stride_y, src_u, src_v,
-                  dst_yuy2 + dst_stride_yuy2, width);
-    src_y += src_stride_y * 2;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-    dst_yuy2 += dst_stride_yuy2 * 2;
-  }
-  if (height & 1) {
-    I422ToYUY2Row(src_y, src_u, src_v, dst_yuy2, width);
-  }
-  return 0;
-}
-
-LIBYUV_API
-int I422ToUYVY(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_uyvy, int dst_stride_uyvy,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_uyvy ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
-    dst_stride_uyvy = -dst_stride_uyvy;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_uyvy == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_uyvy = 0;
-  }
-  void (*I422ToUYVYRow)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_uyvy, int width) =
-      I422ToUYVYRow_C;
-#if defined(HAS_I422TOUYVYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOUYVYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToUYVYRow(src_y, src_u, src_v, dst_uyvy, width);
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-    dst_uyvy += dst_stride_uyvy;
-  }
-  return 0;
-}
-
-LIBYUV_API
-int I420ToUYVY(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_uyvy, int dst_stride_uyvy,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_uyvy ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
-    dst_stride_uyvy = -dst_stride_uyvy;
-  }
-  void (*I422ToUYVYRow)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_uyvy, int width) =
-      I422ToUYVYRow_C;
-#if defined(HAS_I422TOUYVYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOUYVYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    I422ToUYVYRow(src_y, src_u, src_v, dst_uyvy, width);
-    I422ToUYVYRow(src_y + src_stride_y, src_u, src_v,
-                  dst_uyvy + dst_stride_uyvy, width);
-    src_y += src_stride_y * 2;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-    dst_uyvy += dst_stride_uyvy * 2;
-  }
-  if (height & 1) {
-    I422ToUYVYRow(src_y, src_u, src_v, dst_uyvy, width);
-  }
-  return 0;
-}
-
-LIBYUV_API
-int I420ToNV12(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_y || !dst_uv ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    int halfheight = (height + 1) >> 1;
-    dst_y = dst_y + (height - 1) * dst_stride_y;
-    dst_uv = dst_uv + (halfheight - 1) * dst_stride_uv;
-    dst_stride_y = -dst_stride_y;
-    dst_stride_uv = -dst_stride_uv;
-  }
-  // Coalesce rows.
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  if (src_stride_y == width &&
-      dst_stride_y == width) {
-    width *= height;
-    height = 1;
-    src_stride_y = dst_stride_y = 0;
-  }
-  // Coalesce rows.
-  if (src_stride_u == halfwidth &&
-      src_stride_v == halfwidth &&
-      dst_stride_uv == halfwidth * 2) {
-    halfwidth *= halfheight;
-    halfheight = 1;
-    src_stride_u = src_stride_v = dst_stride_uv = 0;
-  }
-  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                      int width) = MergeUVRow_C;
-#if defined(HAS_MERGEUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_u, 16) && IS_ALIGNED(src_stride_u, 16) &&
-          IS_ALIGNED(src_v, 16) && IS_ALIGNED(src_stride_v, 16) &&
-          IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
-        MergeUVRow_ = MergeUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    MergeUVRow_ = MergeUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      MergeUVRow_ = MergeUVRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_NEON;
-    }
-  }
-#endif
-
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  for (int y = 0; y < halfheight; ++y) {
-    // Merge a row of U and V into a row of UV.
-    MergeUVRow_(src_u, src_v, dst_uv, halfwidth);
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-    dst_uv += dst_stride_uv;
-  }
-  return 0;
-}
-
-LIBYUV_API
-int I420ToNV21(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_vu, int dst_stride_vu,
-               int width, int height) {
-  return I420ToNV12(src_y, src_stride_y,
-                    src_v, src_stride_v,
-                    src_u, src_stride_u,
-                    dst_y, src_stride_y,
-                    dst_vu, dst_stride_vu,
-                    width, height);
-}
-
-// Convert I420 to ARGB.
-LIBYUV_API
-int I420ToARGB(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  void (*I422ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToARGBRow_C;
-#if defined(HAS_I422TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I422ToARGBRow = I422ToARGBRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 16) {
-    I422ToARGBRow = I422ToARGBRow_Any_AVX2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToARGBRow = I422ToARGBRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2) &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride_argb, 4)) {
-    I422ToARGBRow = I422ToARGBRow_MIPS_DSPR2;
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToARGBRow(src_y, src_u, src_v, dst_argb, width);
-    dst_argb += dst_stride_argb;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to BGRA.
-LIBYUV_API
-int I420ToBGRA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_bgra, int dst_stride_bgra,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_bgra ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_bgra = dst_bgra + (height - 1) * dst_stride_bgra;
-    dst_stride_bgra = -dst_stride_bgra;
-  }
-  void (*I422ToBGRARow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToBGRARow_C;
-#if defined(HAS_I422TOBGRAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToBGRARow = I422ToBGRARow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToBGRARow = I422ToBGRARow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_bgra, 16) && IS_ALIGNED(dst_stride_bgra, 16)) {
-        I422ToBGRARow = I422ToBGRARow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I422TOBGRAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToBGRARow = I422ToBGRARow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToBGRARow = I422ToBGRARow_NEON;
-    }
-  }
-#elif defined(HAS_I422TOBGRAROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2) &&
-      IS_ALIGNED(dst_bgra, 4) && IS_ALIGNED(dst_stride_bgra, 4)) {
-    I422ToBGRARow = I422ToBGRARow_MIPS_DSPR2;
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToBGRARow(src_y, src_u, src_v, dst_bgra, width);
-    dst_bgra += dst_stride_bgra;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to ABGR.
-LIBYUV_API
-int I420ToABGR(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_abgr, int dst_stride_abgr,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_abgr ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_abgr = dst_abgr + (height - 1) * dst_stride_abgr;
-    dst_stride_abgr = -dst_stride_abgr;
-  }
-  void (*I422ToABGRRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToABGRRow_C;
-#if defined(HAS_I422TOABGRROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToABGRRow = I422ToABGRRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToABGRRow = I422ToABGRRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_abgr, 16) && IS_ALIGNED(dst_stride_abgr, 16)) {
-        I422ToABGRRow = I422ToABGRRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I422TOABGRROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToABGRRow = I422ToABGRRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToABGRRow = I422ToABGRRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToABGRRow(src_y, src_u, src_v, dst_abgr, width);
-    dst_abgr += dst_stride_abgr;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to RGBA.
-LIBYUV_API
-int I420ToRGBA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_rgba ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba;
-    dst_stride_rgba = -dst_stride_rgba;
-  }
-  void (*I422ToRGBARow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToRGBARow_C;
-#if defined(HAS_I422TORGBAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToRGBARow = I422ToRGBARow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGBARow = I422ToRGBARow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_rgba, 16) && IS_ALIGNED(dst_stride_rgba, 16)) {
-        I422ToRGBARow = I422ToRGBARow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I422TORGBAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToRGBARow = I422ToRGBARow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGBARow = I422ToRGBARow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToRGBARow(src_y, src_u, src_v, dst_rgba, width);
-    dst_rgba += dst_stride_rgba;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to RGB24.
-LIBYUV_API
-int I420ToRGB24(const uint8* src_y, int src_stride_y,
-                const uint8* src_u, int src_stride_u,
-                const uint8* src_v, int src_stride_v,
-                uint8* dst_rgb24, int dst_stride_rgb24,
-                int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_rgb24 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24;
-    dst_stride_rgb24 = -dst_stride_rgb24;
-  }
-  void (*I422ToRGB24Row)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToRGB24Row_C;
-#if defined(HAS_I422TORGB24ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToRGB24Row = I422ToRGB24Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGB24Row = I422ToRGB24Row_SSSE3;
-    }
-  }
-#elif defined(HAS_I422TORGB24ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToRGB24Row = I422ToRGB24Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGB24Row = I422ToRGB24Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, width);
-    dst_rgb24 += dst_stride_rgb24;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to RAW.
-LIBYUV_API
-int I420ToRAW(const uint8* src_y, int src_stride_y,
-                const uint8* src_u, int src_stride_u,
-                const uint8* src_v, int src_stride_v,
-                uint8* dst_raw, int dst_stride_raw,
-                int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_raw ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_raw = dst_raw + (height - 1) * dst_stride_raw;
-    dst_stride_raw = -dst_stride_raw;
-  }
-  void (*I422ToRAWRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToRAWRow_C;
-#if defined(HAS_I422TORAWROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToRAWRow = I422ToRAWRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRAWRow = I422ToRAWRow_SSSE3;
-    }
-  }
-#elif defined(HAS_I422TORAWROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToRAWRow = I422ToRAWRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRAWRow = I422ToRAWRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToRAWRow(src_y, src_u, src_v, dst_raw, width);
-    dst_raw += dst_stride_raw;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to ARGB1555.
-LIBYUV_API
-int I420ToARGB1555(const uint8* src_y, int src_stride_y,
-                   const uint8* src_u, int src_stride_u,
-                   const uint8* src_v, int src_stride_v,
-                   uint8* dst_argb1555, int dst_stride_argb1555,
-                   int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_argb1555 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb1555 = dst_argb1555 + (height - 1) * dst_stride_argb1555;
-    dst_stride_argb1555 = -dst_stride_argb1555;
-  }
-  void (*I422ToARGB1555Row)(const uint8* y_buf,
-                          const uint8* u_buf,
-                          const uint8* v_buf,
-                          uint8* rgb_buf,
-                          int width) = I422ToARGB1555Row_C;
-#if defined(HAS_I422TOARGB1555ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToARGB1555Row = I422ToARGB1555Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGB1555Row = I422ToARGB1555Row_SSSE3;
-    }
-  }
-#elif defined(HAS_I422TOARGB1555ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToARGB1555Row = I422ToARGB1555Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGB1555Row = I422ToARGB1555Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToARGB1555Row(src_y, src_u, src_v, dst_argb1555, width);
-    dst_argb1555 += dst_stride_argb1555;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-
-// Convert I420 to ARGB4444.
-LIBYUV_API
-int I420ToARGB4444(const uint8* src_y, int src_stride_y,
-                   const uint8* src_u, int src_stride_u,
-                   const uint8* src_v, int src_stride_v,
-                   uint8* dst_argb4444, int dst_stride_argb4444,
-                   int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_argb4444 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb4444 = dst_argb4444 + (height - 1) * dst_stride_argb4444;
-    dst_stride_argb4444 = -dst_stride_argb4444;
-  }
-  void (*I422ToARGB4444Row)(const uint8* y_buf,
-                          const uint8* u_buf,
-                          const uint8* v_buf,
-                          uint8* rgb_buf,
-                          int width) = I422ToARGB4444Row_C;
-#if defined(HAS_I422TOARGB4444ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToARGB4444Row = I422ToARGB4444Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGB4444Row = I422ToARGB4444Row_SSSE3;
-    }
-  }
-#elif defined(HAS_I422TOARGB4444ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToARGB4444Row = I422ToARGB4444Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGB4444Row = I422ToARGB4444Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToARGB4444Row(src_y, src_u, src_v, dst_argb4444, width);
-    dst_argb4444 += dst_stride_argb4444;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to RGB565.
-LIBYUV_API
-int I420ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_u, int src_stride_u,
-                 const uint8* src_v, int src_stride_v,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height) {
-  if (!src_y || !src_u || !src_v || !dst_rgb565 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565;
-    dst_stride_rgb565 = -dst_stride_rgb565;
-  }
-  void (*I422ToRGB565Row)(const uint8* y_buf,
-                          const uint8* u_buf,
-                          const uint8* v_buf,
-                          uint8* rgb_buf,
-                          int width) = I422ToRGB565Row_C;
-#if defined(HAS_I422TORGB565ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToRGB565Row = I422ToRGB565Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGB565Row = I422ToRGB565Row_SSSE3;
-    }
-  }
-#elif defined(HAS_I422TORGB565ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToRGB565Row = I422ToRGB565Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGB565Row = I422ToRGB565Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, width);
-    dst_rgb565 += dst_stride_rgb565;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_u += src_stride_u;
-      src_v += src_stride_v;
-    }
-  }
-  return 0;
-}
-
-// Convert I420 to specified format
-LIBYUV_API
-int ConvertFromI420(const uint8* y, int y_stride,
-                    const uint8* u, int u_stride,
-                    const uint8* v, int v_stride,
-                    uint8* dst_sample, int dst_sample_stride,
-                    int width, int height,
-                    uint32 fourcc) {
-  uint32 format = CanonicalFourCC(fourcc);
-  if (!y || !u|| !v || !dst_sample ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  int r = 0;
-  switch (format) {
-    // Single plane formats
-    case FOURCC_YUY2:
-      r = I420ToYUY2(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 2,
-                     width, height);
-      break;
-    case FOURCC_UYVY:
-      r = I420ToUYVY(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 2,
-                     width, height);
-      break;
-    case FOURCC_RGBP:
-      r = I420ToRGB565(y, y_stride,
-                       u, u_stride,
-                       v, v_stride,
-                       dst_sample,
-                       dst_sample_stride ? dst_sample_stride : width * 2,
-                       width, height);
-      break;
-    case FOURCC_RGBO:
-      r = I420ToARGB1555(y, y_stride,
-                         u, u_stride,
-                         v, v_stride,
-                         dst_sample,
-                         dst_sample_stride ? dst_sample_stride : width * 2,
-                         width, height);
-      break;
-    case FOURCC_R444:
-      r = I420ToARGB4444(y, y_stride,
-                         u, u_stride,
-                         v, v_stride,
-                         dst_sample,
-                         dst_sample_stride ? dst_sample_stride : width * 2,
-                         width, height);
-      break;
-    case FOURCC_24BG:
-      r = I420ToRGB24(y, y_stride,
-                      u, u_stride,
-                      v, v_stride,
-                      dst_sample,
-                      dst_sample_stride ? dst_sample_stride : width * 3,
-                      width, height);
-      break;
-    case FOURCC_RAW:
-      r = I420ToRAW(y, y_stride,
-                    u, u_stride,
-                    v, v_stride,
-                    dst_sample,
-                    dst_sample_stride ? dst_sample_stride : width * 3,
-                    width, height);
-      break;
-    case FOURCC_ARGB:
-      r = I420ToARGB(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 4,
-                     width, height);
-      break;
-    case FOURCC_BGRA:
-      r = I420ToBGRA(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 4,
-                     width, height);
-      break;
-    case FOURCC_ABGR:
-      r = I420ToABGR(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 4,
-                     width, height);
-      break;
-    case FOURCC_RGBA:
-      r = I420ToRGBA(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width * 4,
-                     width, height);
-      break;
-    case FOURCC_BGGR:
-      r = I420ToBayerBGGR(y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          dst_sample,
-                          dst_sample_stride ? dst_sample_stride : width,
-                          width, height);
-      break;
-    case FOURCC_GBRG:
-      r = I420ToBayerGBRG(y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          dst_sample,
-                          dst_sample_stride ? dst_sample_stride : width,
-                          width, height);
-      break;
-    case FOURCC_GRBG:
-      r = I420ToBayerGRBG(y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          dst_sample,
-                          dst_sample_stride ? dst_sample_stride : width,
-                          width, height);
-      break;
-    case FOURCC_RGGB:
-      r = I420ToBayerRGGB(y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          dst_sample,
-                          dst_sample_stride ? dst_sample_stride : width,
-                          width, height);
-      break;
-    case FOURCC_I400:
-      r = I400Copy(y, y_stride,
-                   dst_sample,
-                   dst_sample_stride ? dst_sample_stride : width,
-                   width, height);
-      break;
-    case FOURCC_NV12: {
-      uint8* dst_uv = dst_sample + width * height;
-      r = I420ToNV12(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width,
-                     dst_uv,
-                     dst_sample_stride ? dst_sample_stride : width,
-                     width, height);
-      break;
-    }
-    case FOURCC_NV21: {
-      uint8* dst_vu = dst_sample + width * height;
-      r = I420ToNV21(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample,
-                     dst_sample_stride ? dst_sample_stride : width,
-                     dst_vu,
-                     dst_sample_stride ? dst_sample_stride : width,
-                     width, height);
-      break;
-    }
-    // TODO(fbarchard): Add M420 and Q420.
-    // Triplanar formats
-    // TODO(fbarchard): halfstride instead of halfwidth
-    case FOURCC_I420:
-    case FOURCC_YU12:
-    case FOURCC_YV12: {
-      int halfwidth = (width + 1) / 2;
-      int halfheight = (height + 1) / 2;
-      uint8* dst_u;
-      uint8* dst_v;
-      if (format == FOURCC_YV12) {
-        dst_v = dst_sample + width * height;
-        dst_u = dst_v + halfwidth * halfheight;
-      } else {
-        dst_u = dst_sample + width * height;
-        dst_v = dst_u + halfwidth * halfheight;
-      }
-      r = I420Copy(y, y_stride,
-                   u, u_stride,
-                   v, v_stride,
-                   dst_sample, width,
-                   dst_u, halfwidth,
-                   dst_v, halfwidth,
-                   width, height);
-      break;
-    }
-    case FOURCC_I422:
-    case FOURCC_YV16: {
-      int halfwidth = (width + 1) / 2;
-      uint8* dst_u;
-      uint8* dst_v;
-      if (format == FOURCC_YV16) {
-        dst_v = dst_sample + width * height;
-        dst_u = dst_v + halfwidth * height;
-      } else {
-        dst_u = dst_sample + width * height;
-        dst_v = dst_u + halfwidth * height;
-      }
-      r = I420ToI422(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample, width,
-                     dst_u, halfwidth,
-                     dst_v, halfwidth,
-                     width, height);
-      break;
-    }
-    case FOURCC_I444:
-    case FOURCC_YV24: {
-      uint8* dst_u;
-      uint8* dst_v;
-      if (format == FOURCC_YV24) {
-        dst_v = dst_sample + width * height;
-        dst_u = dst_v + width * height;
-      } else {
-        dst_u = dst_sample + width * height;
-        dst_v = dst_u + width * height;
-      }
-      r = I420ToI444(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample, width,
-                     dst_u, width,
-                     dst_v, width,
-                     width, height);
-      break;
-    }
-    case FOURCC_I411: {
-      int quarterwidth = (width + 3) / 4;
-      uint8* dst_u = dst_sample + width * height;
-      uint8* dst_v = dst_u + quarterwidth * height;
-      r = I420ToI411(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     dst_sample, width,
-                     dst_u, quarterwidth,
-                     dst_v, quarterwidth,
-                     width, height);
-      break;
-    }
-
-    // Formats not supported - MJPG, biplanar, some rgb formats.
-    default:
-      return -1;  // unknown fourcc - return failure code.
-  }
-  return r;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_from_argb.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_from_argb.cc
deleted file mode 100755
index 41421fb30b..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_from_argb.cc
+++ /dev/null
@@ -1,1096 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert_from_argb.h"
-
-#include "libyuv/basic_types.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// ARGB little endian (bgra in memory) to I444
-LIBYUV_API
-int ARGBToI444(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_y == width &&
-      dst_stride_u == width &&
-      dst_stride_v == width) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_y = dst_stride_u = dst_stride_v = 0;
-  }
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-  void (*ARGBToUV444Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV444Row_C;
-#if defined(HAS_ARGBTOUV444ROW_SSSE3)
-    if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-      ARGBToUV444Row = ARGBToUV444Row_Any_SSSE3;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV444Row = ARGBToUV444Row_Unaligned_SSSE3;
-        if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-          ARGBToUV444Row = ARGBToUV444Row_SSSE3;
-        }
-      }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    ARGBToUV444Row = ARGBToUV444Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-      ARGBToUV444Row = ARGBToUV444Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV444Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// ARGB little endian (bgra in memory) to I422
-LIBYUV_API
-int ARGBToI422(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_y == width &&
-      dst_stride_u * 2 == width &&
-      dst_stride_v * 2 == width) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_y = dst_stride_u = dst_stride_v = 0;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// ARGB little endian (bgra in memory) to I411
-LIBYUV_API
-int ARGBToI411(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_y == width &&
-      dst_stride_u * 4 == width &&
-      dst_stride_v * 4 == width) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_y = dst_stride_u = dst_stride_v = 0;
-  }
-  void (*ARGBToUV411Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV411Row_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    ARGBToYRow = ARGBToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      ARGBToYRow = ARGBToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 32) {
-      ARGBToUV411Row = ARGBToUV411Row_Any_NEON;
-      if (IS_ALIGNED(width, 32)) {
-        ARGBToUV411Row = ARGBToUV411Row_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV411Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-LIBYUV_API
-int ARGBToNV12(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_y || !dst_uv ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3) && defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVRow = ARGBToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ARGBToYRow = ARGBToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
-  }
-#endif
-  int halfwidth = (width + 1) >> 1;
-  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                      int width) = MergeUVRow_C;
-#if defined(HAS_MERGEUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
-        MergeUVRow_ = MergeUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    MergeUVRow_ = MergeUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      MergeUVRow_ = MergeUVRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_NEON;
-    }
-  }
-#endif
-
-  // Allocate a rows of uv.
-  align_buffer_64(row_u, ((halfwidth + 15) & ~15) * 2);
-  uint8* row_v = row_u + ((halfwidth + 15) & ~15);
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
-    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
-    src_argb += src_stride_argb * 2;
-    dst_y += dst_stride_y * 2;
-    dst_uv += dst_stride_uv;
-  }
-  if (height & 1) {
-    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
-    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-  }
-  free_aligned_buffer_64(row_u);
-  return 0;
-}
-
-// Same as NV12 but U and V swapped.
-LIBYUV_API
-int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_y || !dst_uv ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3) && defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVRow = ARGBToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ARGBToYRow = ARGBToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
-  }
-#endif
-  int halfwidth = (width + 1) >> 1;
-  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                      int width) = MergeUVRow_C;
-#if defined(HAS_MERGEUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
-        MergeUVRow_ = MergeUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    MergeUVRow_ = MergeUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      MergeUVRow_ = MergeUVRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_NEON;
-    }
-  }
-#endif
-
-  // Allocate a rows of uv.
-  align_buffer_64(row_u, ((halfwidth + 15) & ~15) * 2);
-  uint8* row_v = row_u + ((halfwidth + 15) & ~15);
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
-    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
-    src_argb += src_stride_argb * 2;
-    dst_y += dst_stride_y * 2;
-    dst_uv += dst_stride_uv;
-  }
-  if (height & 1) {
-    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
-    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-  }
-  free_aligned_buffer_64(row_u);
-  return 0;
-}
-
-// Convert ARGB to YUY2.
-LIBYUV_API
-int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yuy2, int dst_stride_yuy2,
-               int width, int height) {
-  if (!src_argb || !dst_yuy2 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
-    dst_stride_yuy2 = -dst_stride_yuy2;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_yuy2 == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_yuy2 = 0;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  void (*I422ToYUY2Row)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_yuy2, int width) =
-      I422ToYUY2Row_C;
-#if defined(HAS_I422TOYUY2ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOYUY2ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_NEON;
-    }
-  }
-#endif
-
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, row_u, row_v, width);
-    ARGBToYRow(src_argb, row_y, width);
-    I422ToYUY2Row(row_y, row_u, row_v, dst_yuy2, width);
-    src_argb += src_stride_argb;
-    dst_yuy2 += dst_stride_yuy2;
-  }
-
-  free_aligned_buffer_64(row_y);
-  return 0;
-}
-
-// Convert ARGB to UYVY.
-LIBYUV_API
-int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_uyvy, int dst_stride_uyvy,
-               int width, int height) {
-  if (!src_argb || !dst_uyvy ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
-    dst_stride_uyvy = -dst_stride_uyvy;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_uyvy == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_uyvy = 0;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  void (*I422ToUYVYRow)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_uyvy, int width) =
-      I422ToUYVYRow_C;
-#if defined(HAS_I422TOUYVYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOUYVYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_NEON;
-    }
-  }
-#endif
-
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, row_u, row_v, width);
-    ARGBToYRow(src_argb, row_y, width);
-    I422ToUYVYRow(row_y, row_u, row_v, dst_uyvy, width);
-    src_argb += src_stride_argb;
-    dst_uyvy += dst_stride_uyvy;
-  }
-
-  free_aligned_buffer_64(row_y);
-  return 0;
-}
-
-// Convert ARGB to I400.
-LIBYUV_API
-int ARGBToI400(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  if (!src_argb || !dst_y || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_y == width) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_y = 0;
-  }
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    ARGBToYRow = ARGBToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      ARGBToYRow = ARGBToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-  }
-  return 0;
-}
-
-// Shuffle table for converting ARGB to RGBA.
-static uvec8 kShuffleMaskARGBToRGBA = {
-  3u, 0u, 1u, 2u, 7u, 4u, 5u, 6u, 11u, 8u, 9u, 10u, 15u, 12u, 13u, 14u
-};
-
-// Convert ARGB to RGBA.
-LIBYUV_API
-int ARGBToRGBA(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height) {
-  return ARGBShuffle(src_argb, src_stride_argb,
-                     dst_rgba, dst_stride_rgba,
-                     (const uint8*)(&kShuffleMaskARGBToRGBA),
-                     width, height);
-}
-
-// Convert ARGB To RGB24.
-LIBYUV_API
-int ARGBToRGB24(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_rgb24, int dst_stride_rgb24,
-                int width, int height) {
-  if (!src_argb || !dst_rgb24 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_rgb24 == width * 3) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_rgb24 = 0;
-  }
-  void (*ARGBToRGB24Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRGB24Row_C;
-#if defined(HAS_ARGBTORGB24ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToRGB24Row = ARGBToRGB24Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToRGB24Row = ARGBToRGB24Row_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTORGB24ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRGB24Row = ARGBToRGB24Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRGB24Row = ARGBToRGB24Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRGB24Row(src_argb, dst_rgb24, width);
-    src_argb += src_stride_argb;
-    dst_rgb24 += dst_stride_rgb24;
-  }
-  return 0;
-}
-
-// Convert ARGB To RAW.
-LIBYUV_API
-int ARGBToRAW(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_raw, int dst_stride_raw,
-              int width, int height) {
-  if (!src_argb || !dst_raw || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_raw == width * 3) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_raw = 0;
-  }
-  void (*ARGBToRAWRow)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRAWRow_C;
-#if defined(HAS_ARGBTORAWROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToRAWRow = ARGBToRAWRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToRAWRow = ARGBToRAWRow_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTORAWROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRAWRow = ARGBToRAWRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRAWRow = ARGBToRAWRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRAWRow(src_argb, dst_raw, width);
-    src_argb += src_stride_argb;
-    dst_raw += dst_stride_raw;
-  }
-  return 0;
-}
-
-// Convert ARGB To RGB565.
-LIBYUV_API
-int ARGBToRGB565(const uint8* src_argb, int src_stride_argb,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height) {
-  if (!src_argb || !dst_rgb565 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_rgb565 == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_rgb565 = 0;
-  }
-  void (*ARGBToRGB565Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRGB565Row_C;
-#if defined(HAS_ARGBTORGB565ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToRGB565Row = ARGBToRGB565Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToRGB565Row = ARGBToRGB565Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTORGB565ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRGB565Row = ARGBToRGB565Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRGB565Row = ARGBToRGB565Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRGB565Row(src_argb, dst_rgb565, width);
-    src_argb += src_stride_argb;
-    dst_rgb565 += dst_stride_rgb565;
-  }
-  return 0;
-}
-
-// Convert ARGB To ARGB1555.
-LIBYUV_API
-int ARGBToARGB1555(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb1555, int dst_stride_argb1555,
-                   int width, int height) {
-  if (!src_argb || !dst_argb1555 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb1555 == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb1555 = 0;
-  }
-  void (*ARGBToARGB1555Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToARGB1555Row_C;
-#if defined(HAS_ARGBTOARGB1555ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToARGB1555Row = ARGBToARGB1555Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTOARGB1555ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToARGB1555Row = ARGBToARGB1555Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToARGB1555Row(src_argb, dst_argb1555, width);
-    src_argb += src_stride_argb;
-    dst_argb1555 += dst_stride_argb1555;
-  }
-  return 0;
-}
-
-// Convert ARGB To ARGB4444.
-LIBYUV_API
-int ARGBToARGB4444(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb4444, int dst_stride_argb4444,
-                   int width, int height) {
-  if (!src_argb || !dst_argb4444 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb4444 == width * 2) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb4444 = 0;
-  }
-  void (*ARGBToARGB4444Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToARGB4444Row_C;
-#if defined(HAS_ARGBTOARGB4444ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToARGB4444Row = ARGBToARGB4444Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTOARGB4444ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToARGB4444Row = ARGBToARGB4444Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToARGB4444Row(src_argb, dst_argb4444, width);
-    src_argb += src_stride_argb;
-    dst_argb4444 += dst_stride_argb4444;
-  }
-  return 0;
-}
-
-// Convert ARGB to J420. (JPeg full range I420).
-LIBYUV_API
-int ARGBToJ420(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yj, int dst_stride_yj,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_yj || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVJRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVJRow_C;
-  void (*ARGBToYJRow)(const uint8* src_argb, uint8* dst_yj, int pix) =
-      ARGBToYJRow_C;
-#if defined(HAS_ARGBTOYJROW_SSSE3) && defined(HAS_ARGBTOUVJROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVJRow = ARGBToUVJRow_Any_SSSE3;
-    ARGBToYJRow = ARGBToYJRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVJRow = ARGBToUVJRow_Unaligned_SSSE3;
-      ARGBToYJRow = ARGBToYJRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVJRow = ARGBToUVJRow_SSSE3;
-        if (IS_ALIGNED(dst_yj, 16) && IS_ALIGNED(dst_stride_yj, 16)) {
-          ARGBToYJRow = ARGBToYJRow_SSSE3;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYJROW_AVX2) && defined(HAS_ARGBTOUVJROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    ARGBToYJRow = ARGBToYJRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      ARGBToYJRow = ARGBToYJRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYJROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYJRow = ARGBToYJRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYJRow = ARGBToYJRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVJRow = ARGBToUVJRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVJRow = ARGBToUVJRow_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVJRow(src_argb, src_stride_argb, dst_u, dst_v, width);
-    ARGBToYJRow(src_argb, dst_yj, width);
-    ARGBToYJRow(src_argb + src_stride_argb, dst_yj + dst_stride_yj, width);
-    src_argb += src_stride_argb * 2;
-    dst_yj += dst_stride_yj * 2;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  if (height & 1) {
-    ARGBToUVJRow(src_argb, 0, dst_u, dst_v, width);
-    ARGBToYJRow(src_argb, dst_yj, width);
-  }
-  return 0;
-}
-
-// Convert ARGB to J400.
-LIBYUV_API
-int ARGBToJ400(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yj, int dst_stride_yj,
-               int width, int height) {
-  if (!src_argb || !dst_yj || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_yj == width) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_yj = 0;
-  }
-  void (*ARGBToYJRow)(const uint8* src_argb, uint8* dst_yj, int pix) =
-      ARGBToYJRow_C;
-#if defined(HAS_ARGBTOYJROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYJRow = ARGBToYJRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYJRow = ARGBToYJRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_yj, 16) && IS_ALIGNED(dst_stride_yj, 16)) {
-        ARGBToYJRow = ARGBToYJRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYJROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    ARGBToYJRow = ARGBToYJRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      ARGBToYJRow = ARGBToYJRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOYJROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYJRow = ARGBToYJRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYJRow = ARGBToYJRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToYJRow(src_argb, dst_yj, width);
-    src_argb += src_stride_argb;
-    dst_yj += dst_stride_yj;
-  }
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_jpeg.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_jpeg.cc
deleted file mode 100755
index bcb980f7f1..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_jpeg.cc
+++ /dev/null
@@ -1,392 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert.h"
-
-#ifdef HAVE_JPEG
-#include "libyuv/mjpeg_decoder.h"
-#endif
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#ifdef HAVE_JPEG
-struct I420Buffers {
-  uint8* y;
-  int y_stride;
-  uint8* u;
-  int u_stride;
-  uint8* v;
-  int v_stride;
-  int w;
-  int h;
-};
-
-static void JpegCopyI420(void* opaque,
-                         const uint8* const* data,
-                         const int* strides,
-                         int rows) {
-  I420Buffers* dest = (I420Buffers*)(opaque);
-  I420Copy(data[0], strides[0],
-           data[1], strides[1],
-           data[2], strides[2],
-           dest->y, dest->y_stride,
-           dest->u, dest->u_stride,
-           dest->v, dest->v_stride,
-           dest->w, rows);
-  dest->y += rows * dest->y_stride;
-  dest->u += ((rows + 1) >> 1) * dest->u_stride;
-  dest->v += ((rows + 1) >> 1) * dest->v_stride;
-  dest->h -= rows;
-}
-
-static void JpegI422ToI420(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  I420Buffers* dest = (I420Buffers*)(opaque);
-  I422ToI420(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->y, dest->y_stride,
-             dest->u, dest->u_stride,
-             dest->v, dest->v_stride,
-             dest->w, rows);
-  dest->y += rows * dest->y_stride;
-  dest->u += ((rows + 1) >> 1) * dest->u_stride;
-  dest->v += ((rows + 1) >> 1) * dest->v_stride;
-  dest->h -= rows;
-}
-
-static void JpegI444ToI420(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  I420Buffers* dest = (I420Buffers*)(opaque);
-  I444ToI420(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->y, dest->y_stride,
-             dest->u, dest->u_stride,
-             dest->v, dest->v_stride,
-             dest->w, rows);
-  dest->y += rows * dest->y_stride;
-  dest->u += ((rows + 1) >> 1) * dest->u_stride;
-  dest->v += ((rows + 1) >> 1) * dest->v_stride;
-  dest->h -= rows;
-}
-
-static void JpegI411ToI420(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  I420Buffers* dest = (I420Buffers*)(opaque);
-  I411ToI420(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->y, dest->y_stride,
-             dest->u, dest->u_stride,
-             dest->v, dest->v_stride,
-             dest->w, rows);
-  dest->y += rows * dest->y_stride;
-  dest->u += ((rows + 1) >> 1) * dest->u_stride;
-  dest->v += ((rows + 1) >> 1) * dest->v_stride;
-  dest->h -= rows;
-}
-
-static void JpegI400ToI420(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  I420Buffers* dest = (I420Buffers*)(opaque);
-  I400ToI420(data[0], strides[0],
-             dest->y, dest->y_stride,
-             dest->u, dest->u_stride,
-             dest->v, dest->v_stride,
-             dest->w, rows);
-  dest->y += rows * dest->y_stride;
-  dest->u += ((rows + 1) >> 1) * dest->u_stride;
-  dest->v += ((rows + 1) >> 1) * dest->v_stride;
-  dest->h -= rows;
-}
-
-// Query size of MJPG in pixels.
-LIBYUV_API
-int MJPGSize(const uint8* sample, size_t sample_size,
-             int* width, int* height) {
-  MJpegDecoder mjpeg_decoder;
-  LIBYUV_BOOL ret = mjpeg_decoder.LoadFrame(sample, sample_size);
-  if (ret) {
-    *width = mjpeg_decoder.GetWidth();
-    *height = mjpeg_decoder.GetHeight();
-  }
-  mjpeg_decoder.UnloadFrame();
-  return ret ? 0 : -1;  // -1 for runtime failure.
-}
-
-// MJPG (Motion JPeg) to I420
-// TODO(fbarchard): review w and h requirement. dw and dh may be enough.
-LIBYUV_API
-int MJPGToI420(const uint8* sample,
-               size_t sample_size,
-               uint8* y, int y_stride,
-               uint8* u, int u_stride,
-               uint8* v, int v_stride,
-               int w, int h,
-               int dw, int dh) {
-  if (sample_size == kUnknownDataSize) {
-    // ERROR: MJPEG frame size unknown
-    return -1;
-  }
-
-  // TODO(fbarchard): Port MJpeg to C.
-  MJpegDecoder mjpeg_decoder;
-  LIBYUV_BOOL ret = mjpeg_decoder.LoadFrame(sample, sample_size);
-  if (ret && (mjpeg_decoder.GetWidth() != w ||
-              mjpeg_decoder.GetHeight() != h)) {
-    // ERROR: MJPEG frame has unexpected dimensions
-    mjpeg_decoder.UnloadFrame();
-    return 1;  // runtime failure
-  }
-  if (ret) {
-    I420Buffers bufs = { y, y_stride, u, u_stride, v, v_stride, dw, dh };
-    // YUV420
-    if (mjpeg_decoder.GetColorSpace() ==
-            MJpegDecoder::kColorSpaceYCbCr &&
-        mjpeg_decoder.GetNumComponents() == 3 &&
-        mjpeg_decoder.GetVertSampFactor(0) == 2 &&
-        mjpeg_decoder.GetHorizSampFactor(0) == 2 &&
-        mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-        mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-        mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-        mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegCopyI420, &bufs, dw, dh);
-    // YUV422
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 2 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI422ToI420, &bufs, dw, dh);
-    // YUV444
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI444ToI420, &bufs, dw, dh);
-    // YUV411
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 4 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI411ToI420, &bufs, dw, dh);
-    // YUV400
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceGrayscale &&
-               mjpeg_decoder.GetNumComponents() == 1 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI400ToI420, &bufs, dw, dh);
-    } else {
-      // TODO(fbarchard): Implement conversion for any other colorspace/sample
-      // factors that occur in practice. 411 is supported by libjpeg
-      // ERROR: Unable to convert MJPEG frame because format is not supported
-      mjpeg_decoder.UnloadFrame();
-      return 1;
-    }
-  }
-  return ret ? 0 : 1;
-}
-
-#ifdef HAVE_JPEG
-struct ARGBBuffers {
-  uint8* argb;
-  int argb_stride;
-  int w;
-  int h;
-};
-
-static void JpegI420ToARGB(void* opaque,
-                         const uint8* const* data,
-                         const int* strides,
-                         int rows) {
-  ARGBBuffers* dest = (ARGBBuffers*)(opaque);
-  I420ToARGB(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->argb, dest->argb_stride,
-             dest->w, rows);
-  dest->argb += rows * dest->argb_stride;
-  dest->h -= rows;
-}
-
-static void JpegI422ToARGB(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  ARGBBuffers* dest = (ARGBBuffers*)(opaque);
-  I422ToARGB(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->argb, dest->argb_stride,
-             dest->w, rows);
-  dest->argb += rows * dest->argb_stride;
-  dest->h -= rows;
-}
-
-static void JpegI444ToARGB(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  ARGBBuffers* dest = (ARGBBuffers*)(opaque);
-  I444ToARGB(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->argb, dest->argb_stride,
-             dest->w, rows);
-  dest->argb += rows * dest->argb_stride;
-  dest->h -= rows;
-}
-
-static void JpegI411ToARGB(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  ARGBBuffers* dest = (ARGBBuffers*)(opaque);
-  I411ToARGB(data[0], strides[0],
-             data[1], strides[1],
-             data[2], strides[2],
-             dest->argb, dest->argb_stride,
-             dest->w, rows);
-  dest->argb += rows * dest->argb_stride;
-  dest->h -= rows;
-}
-
-static void JpegI400ToARGB(void* opaque,
-                           const uint8* const* data,
-                           const int* strides,
-                           int rows) {
-  ARGBBuffers* dest = (ARGBBuffers*)(opaque);
-  I400ToARGB(data[0], strides[0],
-             dest->argb, dest->argb_stride,
-             dest->w, rows);
-  dest->argb += rows * dest->argb_stride;
-  dest->h -= rows;
-}
-
-// MJPG (Motion JPeg) to ARGB
-// TODO(fbarchard): review w and h requirement. dw and dh may be enough.
-LIBYUV_API
-int MJPGToARGB(const uint8* sample,
-               size_t sample_size,
-               uint8* argb, int argb_stride,
-               int w, int h,
-               int dw, int dh) {
-  if (sample_size == kUnknownDataSize) {
-    // ERROR: MJPEG frame size unknown
-    return -1;
-  }
-
-  // TODO(fbarchard): Port MJpeg to C.
-  MJpegDecoder mjpeg_decoder;
-  LIBYUV_BOOL ret = mjpeg_decoder.LoadFrame(sample, sample_size);
-  if (ret && (mjpeg_decoder.GetWidth() != w ||
-              mjpeg_decoder.GetHeight() != h)) {
-    // ERROR: MJPEG frame has unexpected dimensions
-    mjpeg_decoder.UnloadFrame();
-    return 1;  // runtime failure
-  }
-  if (ret) {
-    ARGBBuffers bufs = { argb, argb_stride, dw, dh };
-    // YUV420
-    if (mjpeg_decoder.GetColorSpace() ==
-            MJpegDecoder::kColorSpaceYCbCr &&
-        mjpeg_decoder.GetNumComponents() == 3 &&
-        mjpeg_decoder.GetVertSampFactor(0) == 2 &&
-        mjpeg_decoder.GetHorizSampFactor(0) == 2 &&
-        mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-        mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-        mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-        mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI420ToARGB, &bufs, dw, dh);
-    // YUV422
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 2 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI422ToARGB, &bufs, dw, dh);
-    // YUV444
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI444ToARGB, &bufs, dw, dh);
-    // YUV411
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceYCbCr &&
-               mjpeg_decoder.GetNumComponents() == 3 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 4 &&
-               mjpeg_decoder.GetVertSampFactor(1) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(1) == 1 &&
-               mjpeg_decoder.GetVertSampFactor(2) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(2) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI411ToARGB, &bufs, dw, dh);
-    // YUV400
-    } else if (mjpeg_decoder.GetColorSpace() ==
-                   MJpegDecoder::kColorSpaceGrayscale &&
-               mjpeg_decoder.GetNumComponents() == 1 &&
-               mjpeg_decoder.GetVertSampFactor(0) == 1 &&
-               mjpeg_decoder.GetHorizSampFactor(0) == 1) {
-      ret = mjpeg_decoder.DecodeToCallback(&JpegI400ToARGB, &bufs, dw, dh);
-    } else {
-      // TODO(fbarchard): Implement conversion for any other colorspace/sample
-      // factors that occur in practice. 411 is supported by libjpeg
-      // ERROR: Unable to convert MJPEG frame because format is not supported
-      mjpeg_decoder.UnloadFrame();
-      return 1;
-    }
-  }
-  return ret ? 0 : 1;
-}
-#endif
-
-#endif
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_argb.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_argb.cc
deleted file mode 100755
index 1b228a7b4d..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_argb.cc
+++ /dev/null
@@ -1,327 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert_argb.h"
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#ifdef HAVE_JPEG
-#include "libyuv/mjpeg_decoder.h"
-#endif
-#include "libyuv/rotate_argb.h"
-#include "libyuv/row.h"
-#include "libyuv/video_common.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Convert camera sample to I420 with cropping, rotation and vertical flip.
-// src_width is used for source stride computation
-// src_height is used to compute location of planes, and indicate inversion
-// sample_size is measured in bytes and is the size of the frame.
-//   With MJPEG it is the compressed size of the frame.
-LIBYUV_API
-int ConvertToARGB(const uint8* sample, size_t sample_size,
-                  uint8* crop_argb, int argb_stride,
-                  int crop_x, int crop_y,
-                  int src_width, int src_height,
-                  int crop_width, int crop_height,
-                  enum RotationMode rotation,
-                  uint32 fourcc) {
-  uint32 format = CanonicalFourCC(fourcc);
-  int aligned_src_width = (src_width + 1) & ~1;
-  const uint8* src;
-  const uint8* src_uv;
-  int abs_src_height = (src_height < 0) ? -src_height : src_height;
-  int inv_crop_height = (crop_height < 0) ? -crop_height : crop_height;
-  int r = 0;
-
-  // One pass rotation is available for some formats. For the rest, convert
-  // to I420 (with optional vertical flipping) into a temporary I420 buffer,
-  // and then rotate the I420 to the final destination buffer.
-  // For in-place conversion, if destination crop_argb is same as source sample,
-  // also enable temporary buffer.
-  LIBYUV_BOOL need_buf = (rotation && format != FOURCC_ARGB) ||
-      crop_argb == sample;
-  uint8* tmp_argb = crop_argb;
-  int tmp_argb_stride = argb_stride;
-  uint8* rotate_buffer = NULL;
-  int abs_crop_height = (crop_height < 0) ? -crop_height : crop_height;
-
-  if (crop_argb == NULL || sample == NULL ||
-      src_width <= 0 || crop_width <= 0 ||
-      src_height == 0 || crop_height == 0) {
-    return -1;
-  }
-  if (src_height < 0) {
-    inv_crop_height = -inv_crop_height;
-  }
-
-  if (need_buf) {
-    int argb_size = crop_width * abs_crop_height * 4;
-    rotate_buffer = (uint8*)malloc(argb_size);
-    if (!rotate_buffer) {
-      return 1;  // Out of memory runtime error.
-    }
-    crop_argb = rotate_buffer;
-    argb_stride = crop_width;
-  }
-
-  switch (format) {
-    // Single plane formats
-    case FOURCC_YUY2:
-      src = sample + (aligned_src_width * crop_y + crop_x) * 2;
-      r = YUY2ToARGB(src, aligned_src_width * 2,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_UYVY:
-      src = sample + (aligned_src_width * crop_y + crop_x) * 2;
-      r = UYVYToARGB(src, aligned_src_width * 2,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_24BG:
-      src = sample + (src_width * crop_y + crop_x) * 3;
-      r = RGB24ToARGB(src, src_width * 3,
-                      crop_argb, argb_stride,
-                      crop_width, inv_crop_height);
-      break;
-    case FOURCC_RAW:
-      src = sample + (src_width * crop_y + crop_x) * 3;
-      r = RAWToARGB(src, src_width * 3,
-                    crop_argb, argb_stride,
-                    crop_width, inv_crop_height);
-      break;
-    case FOURCC_ARGB:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = ARGBToARGB(src, src_width * 4,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_BGRA:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = BGRAToARGB(src, src_width * 4,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_ABGR:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = ABGRToARGB(src, src_width * 4,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBA:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = RGBAToARGB(src, src_width * 4,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBP:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = RGB565ToARGB(src, src_width * 2,
-                       crop_argb, argb_stride,
-                       crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBO:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = ARGB1555ToARGB(src, src_width * 2,
-                         crop_argb, argb_stride,
-                         crop_width, inv_crop_height);
-      break;
-    case FOURCC_R444:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = ARGB4444ToARGB(src, src_width * 2,
-                         crop_argb, argb_stride,
-                         crop_width, inv_crop_height);
-      break;
-    // TODO(fbarchard): Support cropping Bayer by odd numbers
-    // by adjusting fourcc.
-    case FOURCC_BGGR:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerBGGRToARGB(src, src_width,
-                          crop_argb, argb_stride,
-                          crop_width, inv_crop_height);
-      break;
-
-    case FOURCC_GBRG:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerGBRGToARGB(src, src_width,
-                          crop_argb, argb_stride,
-                          crop_width, inv_crop_height);
-      break;
-
-    case FOURCC_GRBG:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerGRBGToARGB(src, src_width,
-                          crop_argb, argb_stride,
-                          crop_width, inv_crop_height);
-      break;
-
-    case FOURCC_RGGB:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerRGGBToARGB(src, src_width,
-                          crop_argb, argb_stride,
-                          crop_width, inv_crop_height);
-      break;
-
-    case FOURCC_I400:
-      src = sample + src_width * crop_y + crop_x;
-      r = I400ToARGB(src, src_width,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-
-    // Biplanar formats
-    case FOURCC_NV12:
-      src = sample + (src_width * crop_y + crop_x);
-      src_uv = sample + aligned_src_width * (src_height + crop_y / 2) + crop_x;
-      r = NV12ToARGB(src, src_width,
-                     src_uv, aligned_src_width,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_NV21:
-      src = sample + (src_width * crop_y + crop_x);
-      src_uv = sample + aligned_src_width * (src_height + crop_y / 2) + crop_x;
-      // Call NV12 but with u and v parameters swapped.
-      r = NV21ToARGB(src, src_width,
-                     src_uv, aligned_src_width,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_M420:
-      src = sample + (src_width * crop_y) * 12 / 8 + crop_x;
-      r = M420ToARGB(src, src_width,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-//    case FOURCC_Q420:
-//      src = sample + (src_width + aligned_src_width * 2) * crop_y + crop_x;
-//      src_uv = sample + (src_width + aligned_src_width * 2) * crop_y +
-//               src_width + crop_x * 2;
-//      r = Q420ToARGB(src, src_width * 3,
-//                    src_uv, src_width * 3,
-//                    crop_argb, argb_stride,
-//                    crop_width, inv_crop_height);
-//      break;
-    // Triplanar formats
-    case FOURCC_I420:
-    case FOURCC_YU12:
-    case FOURCC_YV12: {
-      const uint8* src_y = sample + (src_width * crop_y + crop_x);
-      const uint8* src_u;
-      const uint8* src_v;
-      int halfwidth = (src_width + 1) / 2;
-      int halfheight = (abs_src_height + 1) / 2;
-      if (format == FOURCC_YV12) {
-        src_v = sample + src_width * abs_src_height +
-            (halfwidth * crop_y + crop_x) / 2;
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * (halfheight + crop_y / 2) + crop_x / 2;
-      } else {
-        src_u = sample + src_width * abs_src_height +
-            (halfwidth * crop_y + crop_x) / 2;
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * (halfheight + crop_y / 2) + crop_x / 2;
-      }
-      r = I420ToARGB(src_y, src_width,
-                     src_u, halfwidth,
-                     src_v, halfwidth,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-    case FOURCC_I422:
-    case FOURCC_YV16: {
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u;
-      const uint8* src_v;
-      int halfwidth = (src_width + 1) / 2;
-      if (format == FOURCC_YV16) {
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * crop_y + crop_x / 2;
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * (abs_src_height + crop_y) + crop_x / 2;
-      } else {
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * crop_y + crop_x / 2;
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * (abs_src_height + crop_y) + crop_x / 2;
-      }
-      r = I422ToARGB(src_y, src_width,
-                     src_u, halfwidth,
-                     src_v, halfwidth,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-    case FOURCC_I444:
-    case FOURCC_YV24: {
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u;
-      const uint8* src_v;
-      if (format == FOURCC_YV24) {
-        src_v = sample + src_width * (abs_src_height + crop_y) + crop_x;
-        src_u = sample + src_width * (abs_src_height * 2 + crop_y) + crop_x;
-      } else {
-        src_u = sample + src_width * (abs_src_height + crop_y) + crop_x;
-        src_v = sample + src_width * (abs_src_height * 2 + crop_y) + crop_x;
-      }
-      r = I444ToARGB(src_y, src_width,
-                     src_u, src_width,
-                     src_v, src_width,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-    case FOURCC_I411: {
-      int quarterwidth = (src_width + 3) / 4;
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u = sample + src_width * abs_src_height +
-          quarterwidth * crop_y + crop_x / 4;
-      const uint8* src_v = sample + src_width * abs_src_height +
-          quarterwidth * (abs_src_height + crop_y) + crop_x / 4;
-      r = I411ToARGB(src_y, src_width,
-                     src_u, quarterwidth,
-                     src_v, quarterwidth,
-                     crop_argb, argb_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-#ifdef HAVE_JPEG
-    case FOURCC_MJPG:
-      r = MJPGToARGB(sample, sample_size,
-                     crop_argb, argb_stride,
-                     src_width, abs_src_height, crop_width, inv_crop_height);
-      break;
-#endif
-    default:
-      r = -1;  // unknown fourcc - return failure code.
-  }
-
-  if (need_buf) {
-    if (!r) {
-      r = ARGBRotate(crop_argb, argb_stride,
-                     tmp_argb, tmp_argb_stride,
-                     crop_width, abs_crop_height, rotation);
-    }
-    free(rotate_buffer);
-  }
-
-  return r;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_i420.cc b/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_i420.cc
deleted file mode 100755
index 7b194fff72..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/convert_to_i420.cc
+++ /dev/null
@@ -1,383 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include <stdlib.h>
-
-#include "libyuv/convert.h"
-
-#include "libyuv/format_conversion.h"
-#include "libyuv/video_common.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Convert camera sample to I420 with cropping, rotation and vertical flip.
-// src_width is used for source stride computation
-// src_height is used to compute location of planes, and indicate inversion
-// sample_size is measured in bytes and is the size of the frame.
-//   With MJPEG it is the compressed size of the frame.
-LIBYUV_API
-int ConvertToI420(const uint8* sample,
-                  size_t sample_size,
-                  uint8* y, int y_stride,
-                  uint8* u, int u_stride,
-                  uint8* v, int v_stride,
-                  int crop_x, int crop_y,
-                  int src_width, int src_height,
-                  int crop_width, int crop_height,
-                  enum RotationMode rotation,
-                  uint32 fourcc) {
-  uint32 format = CanonicalFourCC(fourcc);
-  int aligned_src_width = (src_width + 1) & ~1;
-  const uint8* src;
-  const uint8* src_uv;
-  int abs_src_height = (src_height < 0) ? -src_height : src_height;
-  int inv_crop_height = (crop_height < 0) ? -crop_height : crop_height;
-  int r = 0;
-  LIBYUV_BOOL need_buf = (rotation && format != FOURCC_I420 &&
-      format != FOURCC_NV12 && format != FOURCC_NV21 &&
-      format != FOURCC_YU12 && format != FOURCC_YV12) || y == sample;
-  uint8* tmp_y = y;
-  uint8* tmp_u = u;
-  uint8* tmp_v = v;
-  int tmp_y_stride = y_stride;
-  int tmp_u_stride = u_stride;
-  int tmp_v_stride = v_stride;
-  uint8* rotate_buffer = NULL;
-  int abs_crop_height = (crop_height < 0) ? -crop_height : crop_height;
-
-  if (!y || !u || !v || !sample ||
-      src_width <= 0 || crop_width <= 0  ||
-      src_height == 0 || crop_height == 0) {
-    return -1;
-  }
-  if (src_height < 0) {
-    inv_crop_height = -inv_crop_height;
-  }
-
-  // One pass rotation is available for some formats. For the rest, convert
-  // to I420 (with optional vertical flipping) into a temporary I420 buffer,
-  // and then rotate the I420 to the final destination buffer.
-  // For in-place conversion, if destination y is same as source sample,
-  // also enable temporary buffer.
-  if (need_buf) {
-    int y_size = crop_width * abs_crop_height;
-    int uv_size = ((crop_width + 1) / 2) * ((abs_crop_height + 1) / 2);
-    rotate_buffer = (uint8*)malloc(y_size + uv_size * 2);
-    if (!rotate_buffer) {
-      return 1;  // Out of memory runtime error.
-    }
-    y = rotate_buffer;
-    u = y + y_size;
-    v = u + uv_size;
-    y_stride = crop_width;
-    u_stride = v_stride = ((crop_width + 1) / 2);
-  }
-
-  switch (format) {
-    // Single plane formats
-    case FOURCC_YUY2:
-      src = sample + (aligned_src_width * crop_y + crop_x) * 2;
-      r = YUY2ToI420(src, aligned_src_width * 2,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_UYVY:
-      src = sample + (aligned_src_width * crop_y + crop_x) * 2;
-      r = UYVYToI420(src, aligned_src_width * 2,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBP:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = RGB565ToI420(src, src_width * 2,
-                       y, y_stride,
-                       u, u_stride,
-                       v, v_stride,
-                       crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBO:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = ARGB1555ToI420(src, src_width * 2,
-                         y, y_stride,
-                         u, u_stride,
-                         v, v_stride,
-                         crop_width, inv_crop_height);
-      break;
-    case FOURCC_R444:
-      src = sample + (src_width * crop_y + crop_x) * 2;
-      r = ARGB4444ToI420(src, src_width * 2,
-                         y, y_stride,
-                         u, u_stride,
-                         v, v_stride,
-                         crop_width, inv_crop_height);
-      break;
-    case FOURCC_24BG:
-      src = sample + (src_width * crop_y + crop_x) * 3;
-      r = RGB24ToI420(src, src_width * 3,
-                      y, y_stride,
-                      u, u_stride,
-                      v, v_stride,
-                      crop_width, inv_crop_height);
-      break;
-    case FOURCC_RAW:
-      src = sample + (src_width * crop_y + crop_x) * 3;
-      r = RAWToI420(src, src_width * 3,
-                    y, y_stride,
-                    u, u_stride,
-                    v, v_stride,
-                    crop_width, inv_crop_height);
-      break;
-    case FOURCC_ARGB:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = ARGBToI420(src, src_width * 4,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_BGRA:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = BGRAToI420(src, src_width * 4,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_ABGR:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = ABGRToI420(src, src_width * 4,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGBA:
-      src = sample + (src_width * crop_y + crop_x) * 4;
-      r = RGBAToI420(src, src_width * 4,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    // TODO(fbarchard): Support cropping Bayer by odd numbers
-    // by adjusting fourcc.
-    case FOURCC_BGGR:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerBGGRToI420(src, src_width,
-                          y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          crop_width, inv_crop_height);
-      break;
-    case FOURCC_GBRG:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerGBRGToI420(src, src_width,
-                          y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          crop_width, inv_crop_height);
-      break;
-    case FOURCC_GRBG:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerGRBGToI420(src, src_width,
-                          y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          crop_width, inv_crop_height);
-      break;
-    case FOURCC_RGGB:
-      src = sample + (src_width * crop_y + crop_x);
-      r = BayerRGGBToI420(src, src_width,
-                          y, y_stride,
-                          u, u_stride,
-                          v, v_stride,
-                          crop_width, inv_crop_height);
-      break;
-    case FOURCC_I400:
-      src = sample + src_width * crop_y + crop_x;
-      r = I400ToI420(src, src_width,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    // Biplanar formats
-    case FOURCC_NV12:
-      src = sample + (src_width * crop_y + crop_x);
-      src_uv = sample + aligned_src_width * (src_height + crop_y / 2) + crop_x;
-      r = NV12ToI420Rotate(src, src_width,
-                           src_uv, aligned_src_width,
-                           y, y_stride,
-                           u, u_stride,
-                           v, v_stride,
-                           crop_width, inv_crop_height, rotation);
-      break;
-    case FOURCC_NV21:
-      src = sample + (src_width * crop_y + crop_x);
-      src_uv = sample + aligned_src_width * (src_height + crop_y / 2) + crop_x;
-      // Call NV12 but with u and v parameters swapped.
-      r = NV12ToI420Rotate(src, src_width,
-                           src_uv, aligned_src_width,
-                           y, y_stride,
-                           v, v_stride,
-                           u, u_stride,
-                           crop_width, inv_crop_height, rotation);
-      break;
-    case FOURCC_M420:
-      src = sample + (src_width * crop_y) * 12 / 8 + crop_x;
-      r = M420ToI420(src, src_width,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    case FOURCC_Q420:
-      src = sample + (src_width + aligned_src_width * 2) * crop_y + crop_x;
-      src_uv = sample + (src_width + aligned_src_width * 2) * crop_y +
-               src_width + crop_x * 2;
-      r = Q420ToI420(src, src_width * 3,
-                    src_uv, src_width * 3,
-                    y, y_stride,
-                    u, u_stride,
-                    v, v_stride,
-                    crop_width, inv_crop_height);
-      break;
-    // Triplanar formats
-    case FOURCC_I420:
-    case FOURCC_YU12:
-    case FOURCC_YV12: {
-      const uint8* src_y = sample + (src_width * crop_y + crop_x);
-      const uint8* src_u;
-      const uint8* src_v;
-      int halfwidth = (src_width + 1) / 2;
-      int halfheight = (abs_src_height + 1) / 2;
-      if (format == FOURCC_YV12) {
-        src_v = sample + src_width * abs_src_height +
-            (halfwidth * crop_y + crop_x) / 2;
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * (halfheight + crop_y / 2) + crop_x / 2;
-      } else {
-        src_u = sample + src_width * abs_src_height +
-            (halfwidth * crop_y + crop_x) / 2;
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * (halfheight + crop_y / 2) + crop_x / 2;
-      }
-      r = I420Rotate(src_y, src_width,
-                     src_u, halfwidth,
-                     src_v, halfwidth,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height, rotation);
-      break;
-    }
-    case FOURCC_I422:
-    case FOURCC_YV16: {
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u;
-      const uint8* src_v;
-      int halfwidth = (src_width + 1) / 2;
-      if (format == FOURCC_YV16) {
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * crop_y + crop_x / 2;
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * (abs_src_height + crop_y) + crop_x / 2;
-      } else {
-        src_u = sample + src_width * abs_src_height +
-            halfwidth * crop_y + crop_x / 2;
-        src_v = sample + src_width * abs_src_height +
-            halfwidth * (abs_src_height + crop_y) + crop_x / 2;
-      }
-      r = I422ToI420(src_y, src_width,
-                     src_u, halfwidth,
-                     src_v, halfwidth,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-    case FOURCC_I444:
-    case FOURCC_YV24: {
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u;
-      const uint8* src_v;
-      if (format == FOURCC_YV24) {
-        src_v = sample + src_width * (abs_src_height + crop_y) + crop_x;
-        src_u = sample + src_width * (abs_src_height * 2 + crop_y) + crop_x;
-      } else {
-        src_u = sample + src_width * (abs_src_height + crop_y) + crop_x;
-        src_v = sample + src_width * (abs_src_height * 2 + crop_y) + crop_x;
-      }
-      r = I444ToI420(src_y, src_width,
-                     src_u, src_width,
-                     src_v, src_width,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-    case FOURCC_I411: {
-      int quarterwidth = (src_width + 3) / 4;
-      const uint8* src_y = sample + src_width * crop_y + crop_x;
-      const uint8* src_u = sample + src_width * abs_src_height +
-          quarterwidth * crop_y + crop_x / 4;
-      const uint8* src_v = sample + src_width * abs_src_height +
-          quarterwidth * (abs_src_height + crop_y) + crop_x / 4;
-      r = I411ToI420(src_y, src_width,
-                     src_u, quarterwidth,
-                     src_v, quarterwidth,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     crop_width, inv_crop_height);
-      break;
-    }
-#ifdef HAVE_JPEG
-    case FOURCC_MJPG:
-      r = MJPGToI420(sample, sample_size,
-                     y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     src_width, abs_src_height, crop_width, inv_crop_height);
-      break;
-#endif
-    default:
-      r = -1;  // unknown fourcc - return failure code.
-  }
-
-  if (need_buf) {
-    if (!r) {
-      r = I420Rotate(y, y_stride,
-                     u, u_stride,
-                     v, v_stride,
-                     tmp_y, tmp_y_stride,
-                     tmp_u, tmp_u_stride,
-                     tmp_v, tmp_v_stride,
-                     crop_width, abs_crop_height, rotation);
-    }
-    free(rotate_buffer);
-  }
-
-  return r;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/cpu_id.cc b/drivers/theoraplayer/src/YUV/libyuv/src/cpu_id.cc
deleted file mode 100755
index aac7390b32..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/cpu_id.cc
+++ /dev/null
@@ -1,300 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/cpu_id.h"
-
-#ifdef _ANDROID //libtheoraplayer addition for cpu feature detection
-#include "cpu-features.h"
-#endif
-
-#ifdef _MSC_VER
-#include <intrin.h>  // For __cpuidex()
-#endif
-#if !defined(__pnacl__) && !defined(__CLR_VER) && \
-    !defined(__native_client__) && defined(_M_X64) && \
-    defined(_MSC_VER) && (_MSC_FULL_VER >= 160040219)
-#include <immintrin.h>  // For _xgetbv()
-#endif
-
-#if !defined(__native_client__)
-#include <stdlib.h>  // For getenv()
-#endif
-
-// For ArmCpuCaps() but unittested on all platforms
-#include <stdio.h>
-#include <string.h>
-
-#include "libyuv/basic_types.h"  // For CPU_X86
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// For functions that use the stack and have runtime checks for overflow,
-// use SAFEBUFFERS to avoid additional check.
-#if defined(_MSC_VER) && (_MSC_FULL_VER >= 160040219)
-#define SAFEBUFFERS __declspec(safebuffers)
-#else
-#define SAFEBUFFERS
-#endif
-
-// Low level cpuid for X86. Returns zeros on other CPUs.
-#if !defined(__pnacl__) && !defined(__CLR_VER) && \
-    (defined(_M_IX86) || defined(_M_X64) || \
-    defined(__i386__) || defined(__x86_64__))
-LIBYUV_API
-void CpuId(uint32 info_eax, uint32 info_ecx, uint32* cpu_info) {
-#if defined(_MSC_VER)
-#if (_MSC_FULL_VER >= 160040219)
-  __cpuidex((int*)(cpu_info), info_eax, info_ecx);
-#elif defined(_M_IX86)
-  __asm {
-    mov        eax, info_eax
-    mov        ecx, info_ecx
-    mov        edi, cpu_info
-    cpuid
-    mov        [edi], eax
-    mov        [edi + 4], ebx
-    mov        [edi + 8], ecx
-    mov        [edi + 12], edx
-  }
-#else
-  if (info_ecx == 0) {
-    __cpuid((int*)(cpu_info), info_eax);
-  } else {
-    cpu_info[3] = cpu_info[2] = cpu_info[1] = cpu_info[0] = 0;
-  }
-#endif
-#else  // defined(_MSC_VER)
-  uint32 info_ebx, info_edx;
-  asm volatile (  // NOLINT
-#if defined( __i386__) && defined(__PIC__)
-    // Preserve ebx for fpic 32 bit.
-    "mov %%ebx, %%edi                          \n"
-    "cpuid                                     \n"
-    "xchg %%edi, %%ebx                         \n"
-    : "=D" (info_ebx),
-#else
-    "cpuid                                     \n"
-    : "=b" (info_ebx),
-#endif  //  defined( __i386__) && defined(__PIC__)
-      "+a" (info_eax), "+c" (info_ecx), "=d" (info_edx));
-  cpu_info[0] = info_eax;
-  cpu_info[1] = info_ebx;
-  cpu_info[2] = info_ecx;
-  cpu_info[3] = info_edx;
-#endif  // defined(_MSC_VER)
-}
-
-#if !defined(__native_client__)
-#define HAS_XGETBV
-// X86 CPUs have xgetbv to detect OS saves high parts of ymm registers.
-int TestOsSaveYmm() {
-  uint32 xcr0 = 0u;
-#if defined(_MSC_VER) && (_MSC_FULL_VER >= 160040219)
-  xcr0 = (uint32)(_xgetbv(0));  // VS2010 SP1 required.
-#elif defined(_M_IX86)
-  __asm {
-    xor        ecx, ecx    // xcr 0
-    _asm _emit 0x0f _asm _emit 0x01 _asm _emit 0xd0  // For VS2010 and earlier.
-    mov        xcr0, eax
-  }
-#elif defined(__i386__) || defined(__x86_64__)
-  asm(".byte 0x0f, 0x01, 0xd0" : "=a" (xcr0) : "c" (0) : "%edx");
-#endif  // defined(_MSC_VER)
-  return((xcr0 & 6) == 6);  // Is ymm saved?
-}
-#endif  // !defined(__native_client__)
-#else
-LIBYUV_API
-void CpuId(uint32 eax, uint32 ecx, uint32* cpu_info) {
-  cpu_info[0] = cpu_info[1] = cpu_info[2] = cpu_info[3] = 0;
-}
-#endif
-
-// based on libvpx arm_cpudetect.c
-// For Arm, but public to allow testing on any CPU
-LIBYUV_API SAFEBUFFERS
-int ArmCpuCaps(const char* cpuinfo_name) {
-  char cpuinfo_line[512];
-  FILE* f = fopen(cpuinfo_name, "r");
-  if (!f) {
-    // Assume Neon if /proc/cpuinfo is unavailable.
-    // This will occur for Chrome sandbox for Pepper or Render process.
-    return kCpuHasNEON;
-  }
-  while (fgets(cpuinfo_line, sizeof(cpuinfo_line) - 1, f)) {
-    if (memcmp(cpuinfo_line, "Features", 8) == 0) {
-      char* p = strstr(cpuinfo_line, " neon");
-      if (p && (p[5] == ' ' || p[5] == '\n')) {
-        fclose(f);
-        return kCpuHasNEON;
-      }
-    }
-  }
-  fclose(f);
-  return 0;
-}
-
-#if defined(__mips__) && defined(__linux__)
-static int MipsCpuCaps(const char* search_string) {
-  char cpuinfo_line[512];
-  const char* file_name = "/proc/cpuinfo";
-  FILE* f = fopen(file_name, "r");
-  if (!f) {
-    // Assume DSP if /proc/cpuinfo is unavailable.
-    // This will occur for Chrome sandbox for Pepper or Render process.
-    return kCpuHasMIPS_DSP;
-  }
-  while (fgets(cpuinfo_line, sizeof(cpuinfo_line) - 1, f) != NULL) {
-    if (strstr(cpuinfo_line, search_string) != NULL) {
-      fclose(f);
-      return kCpuHasMIPS_DSP;
-    }
-  }
-  fclose(f);
-  return 0;
-}
-#endif
-
-// CPU detect function for SIMD instruction sets.
-LIBYUV_API
-int cpu_info_ = kCpuInit;  // cpu_info is not initialized yet.
-
-// Test environment variable for disabling CPU features. Any non-zero value
-// to disable. Zero ignored to make it easy to set the variable on/off.
-#if !defined(__native_client__) && !defined(_M_ARM)
-
-static LIBYUV_BOOL TestEnv(const char* name) {
-#if !defined(_WINRT) && !defined(ORBIS_ENABLED)
-  const char* var = getenv(name);
-  if (var) {
-    if (var[0] != '0') {
-      return LIBYUV_TRUE;
-    }
-  }
-#endif
-  return LIBYUV_FALSE;
-}
-#else  // nacl does not support getenv().
-static LIBYUV_BOOL TestEnv(const char*) {
-  return LIBYUV_FALSE;
-}
-#endif
-
-LIBYUV_API SAFEBUFFERS
-int InitCpuFlags(void) {
-#if !defined(__pnacl__) && !defined(__CLR_VER) && defined(CPU_X86)
-
-  uint32 cpu_info1[4] = { 0, 0, 0, 0 };
-  uint32 cpu_info7[4] = { 0, 0, 0, 0 };
-  CpuId(1, 0, cpu_info1);
-  CpuId(7, 0, cpu_info7);
-  cpu_info_ = ((cpu_info1[3] & 0x04000000) ? kCpuHasSSE2 : 0) |
-              ((cpu_info1[2] & 0x00000200) ? kCpuHasSSSE3 : 0) |
-              ((cpu_info1[2] & 0x00080000) ? kCpuHasSSE41 : 0) |
-              ((cpu_info1[2] & 0x00100000) ? kCpuHasSSE42 : 0) |
-              ((cpu_info7[1] & 0x00000200) ? kCpuHasERMS : 0) |
-              ((cpu_info1[2] & 0x00001000) ? kCpuHasFMA3 : 0) |
-              kCpuHasX86;
-#ifdef HAS_XGETBV
-  if ((cpu_info1[2] & 0x18000000) == 0x18000000 &&  // AVX and OSSave
-      TestOsSaveYmm()) {  // Saves YMM.
-    cpu_info_ |= ((cpu_info7[1] & 0x00000020) ? kCpuHasAVX2 : 0) |
-                 kCpuHasAVX;
-  }
-#endif
-  // Environment variable overrides for testing.
-  if (TestEnv("LIBYUV_DISABLE_X86")) {
-    cpu_info_ &= ~kCpuHasX86;
-  }
-  if (TestEnv("LIBYUV_DISABLE_SSE2")) {
-    cpu_info_ &= ~kCpuHasSSE2;
-  }
-  if (TestEnv("LIBYUV_DISABLE_SSSE3")) {
-    cpu_info_ &= ~kCpuHasSSSE3;
-  }
-  if (TestEnv("LIBYUV_DISABLE_SSE41")) {
-    cpu_info_ &= ~kCpuHasSSE41;
-  }
-  if (TestEnv("LIBYUV_DISABLE_SSE42")) {
-    cpu_info_ &= ~kCpuHasSSE42;
-  }
-  if (TestEnv("LIBYUV_DISABLE_AVX")) {
-    cpu_info_ &= ~kCpuHasAVX;
-  }
-  if (TestEnv("LIBYUV_DISABLE_AVX2")) {
-    cpu_info_ &= ~kCpuHasAVX2;
-  }
-  if (TestEnv("LIBYUV_DISABLE_ERMS")) {
-    cpu_info_ &= ~kCpuHasERMS;
-  }
-  if (TestEnv("LIBYUV_DISABLE_FMA3")) {
-    cpu_info_ &= ~kCpuHasFMA3;
-  }
-#elif defined(__mips__) && defined(__linux__)
-  // Linux mips parse text file for dsp detect.
-  cpu_info_ = MipsCpuCaps("dsp");  // set kCpuHasMIPS_DSP.
-#if defined(__mips_dspr2)
-  cpu_info_ |= kCpuHasMIPS_DSPR2;
-#endif
-  cpu_info_ |= kCpuHasMIPS;
-
-  if (getenv("LIBYUV_DISABLE_MIPS")) {
-    cpu_info_ &= ~kCpuHasMIPS;
-  }
-  if (getenv("LIBYUV_DISABLE_MIPS_DSP")) {
-    cpu_info_ &= ~kCpuHasMIPS_DSP;
-  }
-  if (getenv("LIBYUV_DISABLE_MIPS_DSPR2")) {
-    cpu_info_ &= ~kCpuHasMIPS_DSPR2;
-  }
-#elif defined(__arm__)
-// gcc -mfpu=neon defines __ARM_NEON__
-// __ARM_NEON__ generates code that requires Neon.  NaCL also requires Neon.
-// For Linux, /proc/cpuinfo can be tested but without that assume Neon.
-#if defined(__ARM_NEON__) || defined(__native_client__) || !defined(__linux__)
-#ifdef _ANDROID
-  cpu_info_ = ArmCpuCaps("/proc/cpuinfo"); // libtheoraplayer #ifdef addition, just in case, android gave us troubles
-#else
-  cpu_info_ = kCpuHasNEON;
-#endif
-#else
-  // Linux arm parse text file for neon detect.
-  cpu_info_ = ArmCpuCaps("/proc/cpuinfo");
-#endif
-  cpu_info_ |= kCpuHasARM;
-  if (TestEnv("LIBYUV_DISABLE_NEON")) {
-    cpu_info_ &= ~kCpuHasNEON;
-  }
-#ifdef _ANDROID
-  // libtheoraplayer addition to disable NEON support on android devices that don't support it, once again, just in case	
-  if ((android_getCpuFeaturesExt() & ANDROID_CPU_ARM_FEATURE_NEON) == 0)
-  {
- 	cpu_info_ = kCpuHasARM;
-  }
-#endif
-#endif  // __arm__
-  if (TestEnv("LIBYUV_DISABLE_ASM")) {
-    cpu_info_ = 0;
-  }
-  return cpu_info_;
-}
-
-LIBYUV_API
-void MaskCpuFlags(int enable_flags) {
-  cpu_info_ = InitCpuFlags() & enable_flags;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/format_conversion.cc b/drivers/theoraplayer/src/YUV/libyuv/src/format_conversion.cc
deleted file mode 100755
index a3daf96a98..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/format_conversion.cc
+++ /dev/null
@@ -1,552 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/format_conversion.h"
-
-#include "libyuv/basic_types.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/video_common.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// generate a selector mask useful for pshufb
-static uint32 GenerateSelector(int select0, int select1) {
-  return (uint32)(select0) |
-         (uint32)((select1 + 4) << 8) |
-         (uint32)((select0 + 8) << 16) |
-         (uint32)((select1 + 12) << 24);
-}
-
-static int MakeSelectors(const int blue_index,
-                         const int green_index,
-                         const int red_index,
-                         uint32 dst_fourcc_bayer,
-                         uint32* index_map) {
-  // Now build a lookup table containing the indices for the four pixels in each
-  // 2x2 Bayer grid.
-  switch (dst_fourcc_bayer) {
-    case FOURCC_BGGR:
-      index_map[0] = GenerateSelector(blue_index, green_index);
-      index_map[1] = GenerateSelector(green_index, red_index);
-      break;
-    case FOURCC_GBRG:
-      index_map[0] = GenerateSelector(green_index, blue_index);
-      index_map[1] = GenerateSelector(red_index, green_index);
-      break;
-    case FOURCC_RGGB:
-      index_map[0] = GenerateSelector(red_index, green_index);
-      index_map[1] = GenerateSelector(green_index, blue_index);
-      break;
-    case FOURCC_GRBG:
-      index_map[0] = GenerateSelector(green_index, red_index);
-      index_map[1] = GenerateSelector(blue_index, green_index);
-      break;
-    default:
-      return -1;  // Bad FourCC
-  }
-  return 0;
-}
-
-// Converts 32 bit ARGB to Bayer RGB formats.
-LIBYUV_API
-int ARGBToBayer(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_bayer, int dst_stride_bayer,
-                int width, int height,
-                uint32 dst_fourcc_bayer) {
-  int y;
-  const int blue_index = 0;  // Offsets for ARGB format
-  const int green_index = 1;
-  const int red_index = 2;
-  uint32 index_map[2];
-  void (*ARGBToBayerRow)(const uint8* src_argb, uint8* dst_bayer,
-                         uint32 selector, int pix) = ARGBToBayerRow_C;
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-#if defined(HAS_ARGBTOBAYERROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToBayerRow = ARGBToBayerRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerRow_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTOBAYERROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToBayerRow = ARGBToBayerRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerRow_NEON;
-    }
-  }
-#endif
-  if (MakeSelectors(blue_index, green_index, red_index,
-                    dst_fourcc_bayer, index_map)) {
-    return -1;  // Bad FourCC
-  }
-
-  for (y = 0; y < height; ++y) {
-    ARGBToBayerRow(src_argb, dst_bayer, index_map[y & 1], width);
-    src_argb += src_stride_argb;
-    dst_bayer += dst_stride_bayer;
-  }
-  return 0;
-}
-
-#define AVG(a, b) (((a) + (b)) >> 1)
-
-static void BayerRowBG(const uint8* src_bayer0, int src_stride_bayer,
-                       uint8* dst_argb, int pix) {
-  const uint8* src_bayer1 = src_bayer0 + src_stride_bayer;
-  uint8 g = src_bayer0[1];
-  uint8 r = src_bayer1[1];
-  int x;
-  for (x = 0; x < pix - 2; x += 2) {
-    dst_argb[0] = src_bayer0[0];
-    dst_argb[1] = AVG(g, src_bayer0[1]);
-    dst_argb[2] = AVG(r, src_bayer1[1]);
-    dst_argb[3] = 255U;
-    dst_argb[4] = AVG(src_bayer0[0], src_bayer0[2]);
-    dst_argb[5] = src_bayer0[1];
-    dst_argb[6] = src_bayer1[1];
-    dst_argb[7] = 255U;
-    g = src_bayer0[1];
-    r = src_bayer1[1];
-    src_bayer0 += 2;
-    src_bayer1 += 2;
-    dst_argb += 8;
-  }
-  dst_argb[0] = src_bayer0[0];
-  dst_argb[1] = AVG(g, src_bayer0[1]);
-  dst_argb[2] = AVG(r, src_bayer1[1]);
-  dst_argb[3] = 255U;
-  if (!(pix & 1)) {
-    dst_argb[4] = src_bayer0[0];
-    dst_argb[5] = src_bayer0[1];
-    dst_argb[6] = src_bayer1[1];
-    dst_argb[7] = 255U;
-  }
-}
-
-static void BayerRowRG(const uint8* src_bayer0, int src_stride_bayer,
-                       uint8* dst_argb, int pix) {
-  const uint8* src_bayer1 = src_bayer0 + src_stride_bayer;
-  uint8 g = src_bayer0[1];
-  uint8 b = src_bayer1[1];
-  int x;
-  for (x = 0; x < pix - 2; x += 2) {
-    dst_argb[0] = AVG(b, src_bayer1[1]);
-    dst_argb[1] = AVG(g, src_bayer0[1]);
-    dst_argb[2] = src_bayer0[0];
-    dst_argb[3] = 255U;
-    dst_argb[4] = src_bayer1[1];
-    dst_argb[5] = src_bayer0[1];
-    dst_argb[6] = AVG(src_bayer0[0], src_bayer0[2]);
-    dst_argb[7] = 255U;
-    g = src_bayer0[1];
-    b = src_bayer1[1];
-    src_bayer0 += 2;
-    src_bayer1 += 2;
-    dst_argb += 8;
-  }
-  dst_argb[0] = AVG(b, src_bayer1[1]);
-  dst_argb[1] = AVG(g, src_bayer0[1]);
-  dst_argb[2] = src_bayer0[0];
-  dst_argb[3] = 255U;
-  if (!(pix & 1)) {
-    dst_argb[4] = src_bayer1[1];
-    dst_argb[5] = src_bayer0[1];
-    dst_argb[6] = src_bayer0[0];
-    dst_argb[7] = 255U;
-  }
-}
-
-static void BayerRowGB(const uint8* src_bayer0, int src_stride_bayer,
-                       uint8* dst_argb, int pix) {
-  const uint8* src_bayer1 = src_bayer0 + src_stride_bayer;
-  uint8 b = src_bayer0[1];
-  int x;
-  for (x = 0; x < pix - 2; x += 2) {
-    dst_argb[0] = AVG(b, src_bayer0[1]);
-    dst_argb[1] = src_bayer0[0];
-    dst_argb[2] = src_bayer1[0];
-    dst_argb[3] = 255U;
-    dst_argb[4] = src_bayer0[1];
-    dst_argb[5] = AVG(src_bayer0[0], src_bayer0[2]);
-    dst_argb[6] = AVG(src_bayer1[0], src_bayer1[2]);
-    dst_argb[7] = 255U;
-    b = src_bayer0[1];
-    src_bayer0 += 2;
-    src_bayer1 += 2;
-    dst_argb += 8;
-  }
-  dst_argb[0] = AVG(b, src_bayer0[1]);
-  dst_argb[1] = src_bayer0[0];
-  dst_argb[2] = src_bayer1[0];
-  dst_argb[3] = 255U;
-  if (!(pix & 1)) {
-    dst_argb[4] = src_bayer0[1];
-    dst_argb[5] = src_bayer0[0];
-    dst_argb[6] = src_bayer1[0];
-    dst_argb[7] = 255U;
-  }
-}
-
-static void BayerRowGR(const uint8* src_bayer0, int src_stride_bayer,
-                       uint8* dst_argb, int pix) {
-  const uint8* src_bayer1 = src_bayer0 + src_stride_bayer;
-  uint8 r = src_bayer0[1];
-  int x;
-  for (x = 0; x < pix - 2; x += 2) {
-    dst_argb[0] = src_bayer1[0];
-    dst_argb[1] = src_bayer0[0];
-    dst_argb[2] = AVG(r, src_bayer0[1]);
-    dst_argb[3] = 255U;
-    dst_argb[4] = AVG(src_bayer1[0], src_bayer1[2]);
-    dst_argb[5] = AVG(src_bayer0[0], src_bayer0[2]);
-    dst_argb[6] = src_bayer0[1];
-    dst_argb[7] = 255U;
-    r = src_bayer0[1];
-    src_bayer0 += 2;
-    src_bayer1 += 2;
-    dst_argb += 8;
-  }
-  dst_argb[0] = src_bayer1[0];
-  dst_argb[1] = src_bayer0[0];
-  dst_argb[2] = AVG(r, src_bayer0[1]);
-  dst_argb[3] = 255U;
-  if (!(pix & 1)) {
-    dst_argb[4] = src_bayer1[0];
-    dst_argb[5] = src_bayer0[0];
-    dst_argb[6] = src_bayer0[1];
-    dst_argb[7] = 255U;
-  }
-}
-
-// Converts any Bayer RGB format to ARGB.
-LIBYUV_API
-int BayerToARGB(const uint8* src_bayer, int src_stride_bayer,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height,
-                uint32 src_fourcc_bayer) {
-  int y;
-  void (*BayerRow0)(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int pix);
-  void (*BayerRow1)(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int pix);
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  switch (src_fourcc_bayer) {
-    case FOURCC_BGGR:
-      BayerRow0 = BayerRowBG;
-      BayerRow1 = BayerRowGR;
-      break;
-    case FOURCC_GBRG:
-      BayerRow0 = BayerRowGB;
-      BayerRow1 = BayerRowRG;
-      break;
-    case FOURCC_GRBG:
-      BayerRow0 = BayerRowGR;
-      BayerRow1 = BayerRowBG;
-      break;
-    case FOURCC_RGGB:
-      BayerRow0 = BayerRowRG;
-      BayerRow1 = BayerRowGB;
-      break;
-    default:
-      return -1;    // Bad FourCC
-  }
-
-  for (y = 0; y < height - 1; y += 2) {
-    BayerRow0(src_bayer, src_stride_bayer, dst_argb, width);
-    BayerRow1(src_bayer + src_stride_bayer, -src_stride_bayer,
-              dst_argb + dst_stride_argb, width);
-    src_bayer += src_stride_bayer * 2;
-    dst_argb += dst_stride_argb * 2;
-  }
-  if (height & 1) {
-    BayerRow0(src_bayer, src_stride_bayer, dst_argb, width);
-  }
-  return 0;
-}
-
-// Converts any Bayer RGB format to ARGB.
-LIBYUV_API
-int BayerToI420(const uint8* src_bayer, int src_stride_bayer,
-                uint8* dst_y, int dst_stride_y,
-                uint8* dst_u, int dst_stride_u,
-                uint8* dst_v, int dst_stride_v,
-                int width, int height,
-                uint32 src_fourcc_bayer) {
-  void (*BayerRow0)(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int pix);
-  void (*BayerRow1)(const uint8* src_bayer, int src_stride_bayer,
-                    uint8* dst_argb, int pix);
-
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-  // Negative height means invert the image.
-  if (height < 0) {
-    int halfheight;
-    height = -height;
-    halfheight = (height + 1) >> 1;
-    dst_y = dst_y + (height - 1) * dst_stride_y;
-    dst_u = dst_u + (halfheight - 1) * dst_stride_u;
-    dst_v = dst_v + (halfheight - 1) * dst_stride_v;
-    dst_stride_y = -dst_stride_y;
-    dst_stride_u = -dst_stride_u;
-    dst_stride_v = -dst_stride_v;
-  }
-#if defined(HAS_ARGBTOYROW_SSSE3) && defined(HAS_ARGBTOUVROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      ARGBToUVRow = ARGBToUVRow_SSSE3;
-      if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
-  }
-#endif
-
-  switch (src_fourcc_bayer) {
-    case FOURCC_BGGR:
-      BayerRow0 = BayerRowBG;
-      BayerRow1 = BayerRowGR;
-      break;
-    case FOURCC_GBRG:
-      BayerRow0 = BayerRowGB;
-      BayerRow1 = BayerRowRG;
-      break;
-    case FOURCC_GRBG:
-      BayerRow0 = BayerRowGR;
-      BayerRow1 = BayerRowBG;
-      break;
-    case FOURCC_RGGB:
-      BayerRow0 = BayerRowRG;
-      BayerRow1 = BayerRowGB;
-      break;
-    default:
-      return -1;  // Bad FourCC
-  }
-
-  {
-    // Allocate 2 rows of ARGB.
-    const int kRowSize = (width * 4 + 15) & ~15;
-    align_buffer_64(row, kRowSize * 2);
-    int y;
-    for (y = 0; y < height - 1; y += 2) {
-      BayerRow0(src_bayer, src_stride_bayer, row, width);
-      BayerRow1(src_bayer + src_stride_bayer, -src_stride_bayer,
-                row + kRowSize, width);
-      ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
-      ARGBToYRow(row, dst_y, width);
-      ARGBToYRow(row + kRowSize, dst_y + dst_stride_y, width);
-      src_bayer += src_stride_bayer * 2;
-      dst_y += dst_stride_y * 2;
-      dst_u += dst_stride_u;
-      dst_v += dst_stride_v;
-    }
-    if (height & 1) {
-      BayerRow0(src_bayer, src_stride_bayer, row, width);
-      ARGBToUVRow(row, 0, dst_u, dst_v, width);
-      ARGBToYRow(row, dst_y, width);
-    }
-    free_aligned_buffer_64(row);
-  }
-  return 0;
-}
-
-// Convert I420 to Bayer.
-LIBYUV_API
-int I420ToBayer(const uint8* src_y, int src_stride_y,
-                const uint8* src_u, int src_stride_u,
-                const uint8* src_v, int src_stride_v,
-                uint8* dst_bayer, int dst_stride_bayer,
-                int width, int height,
-                uint32 dst_fourcc_bayer) {
-  void (*I422ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToARGBRow_C;
-  void (*ARGBToBayerRow)(const uint8* src_argb, uint8* dst_bayer,
-                         uint32 selector, int pix) = ARGBToBayerRow_C;
-  const int blue_index = 0;  // Offsets for ARGB format
-  const int green_index = 1;
-  const int red_index = 2;
-  uint32 index_map[2];
-  // Negative height means invert the image.
-  if (height < 0) {
-    int halfheight;
-    height = -height;
-    halfheight = (height + 1) >> 1;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (halfheight - 1) * src_stride_u;
-    src_v = src_v + (halfheight - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-#if defined(HAS_I422TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_SSSE3;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 16) {
-    I422ToARGBRow = I422ToARGBRow_Any_AVX2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToARGBRow = I422ToARGBRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2)) {
-    I422ToARGBRow = I422ToARGBRow_MIPS_DSPR2;
-  }
-#endif
-
-#if defined(HAS_ARGBTOBAYERROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    ARGBToBayerRow = ARGBToBayerRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerRow_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTOBAYERROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToBayerRow = ARGBToBayerRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerRow_NEON;
-    }
-  }
-#endif
-
-  if (MakeSelectors(blue_index, green_index, red_index,
-                    dst_fourcc_bayer, index_map)) {
-    return -1;  // Bad FourCC
-  }
-  {
-    // Allocate a row of ARGB.
-    align_buffer_64(row, width * 4);
-    int y;
-    for (y = 0; y < height; ++y) {
-      I422ToARGBRow(src_y, src_u, src_v, row, width);
-      ARGBToBayerRow(row, dst_bayer, index_map[y & 1], width);
-      dst_bayer += dst_stride_bayer;
-      src_y += src_stride_y;
-      if (y & 1) {
-        src_u += src_stride_u;
-        src_v += src_stride_v;
-      }
-    }
-    free_aligned_buffer_64(row);
-  }
-  return 0;
-}
-
-#define MAKEBAYERFOURCC(BAYER)                                                 \
-LIBYUV_API                                                                     \
-int Bayer##BAYER##ToI420(const uint8* src_bayer, int src_stride_bayer,         \
-                         uint8* dst_y, int dst_stride_y,                       \
-                         uint8* dst_u, int dst_stride_u,                       \
-                         uint8* dst_v, int dst_stride_v,                       \
-                         int width, int height) {                              \
-  return BayerToI420(src_bayer, src_stride_bayer,                              \
-                     dst_y, dst_stride_y,                                      \
-                     dst_u, dst_stride_u,                                      \
-                     dst_v, dst_stride_v,                                      \
-                     width, height,                                            \
-                     FOURCC_##BAYER);                                          \
-}                                                                              \
-                                                                               \
-LIBYUV_API                                                                     \
-int I420ToBayer##BAYER(const uint8* src_y, int src_stride_y,                   \
-                       const uint8* src_u, int src_stride_u,                   \
-                       const uint8* src_v, int src_stride_v,                   \
-                       uint8* dst_bayer, int dst_stride_bayer,                 \
-                       int width, int height) {                                \
-  return I420ToBayer(src_y, src_stride_y,                                      \
-                     src_u, src_stride_u,                                      \
-                     src_v, src_stride_v,                                      \
-                     dst_bayer, dst_stride_bayer,                              \
-                     width, height,                                            \
-                     FOURCC_##BAYER);                                          \
-}                                                                              \
-                                                                               \
-LIBYUV_API                                                                     \
-int ARGBToBayer##BAYER(const uint8* src_argb, int src_stride_argb,             \
-                       uint8* dst_bayer, int dst_stride_bayer,                 \
-                       int width, int height) {                                \
-  return ARGBToBayer(src_argb, src_stride_argb,                                \
-                     dst_bayer, dst_stride_bayer,                              \
-                     width, height,                                            \
-                     FOURCC_##BAYER);                                          \
-}                                                                              \
-                                                                               \
-LIBYUV_API                                                                     \
-int Bayer##BAYER##ToARGB(const uint8* src_bayer, int src_stride_bayer,         \
-                         uint8* dst_argb, int dst_stride_argb,                 \
-                         int width, int height) {                              \
-  return BayerToARGB(src_bayer, src_stride_bayer,                              \
-                     dst_argb, dst_stride_argb,                                \
-                     width, height,                                            \
-                     FOURCC_##BAYER);                                          \
-}
-
-MAKEBAYERFOURCC(BGGR)
-MAKEBAYERFOURCC(GBRG)
-MAKEBAYERFOURCC(GRBG)
-MAKEBAYERFOURCC(RGGB)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_decoder.cc b/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_decoder.cc
deleted file mode 100755
index 193b829ba9..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_decoder.cc
+++ /dev/null
@@ -1,558 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/mjpeg_decoder.h"
-
-#ifdef HAVE_JPEG
-#include <assert.h>
-
-#if !defined(__pnacl__) && !defined(__CLR_VER) && !defined(COVERAGE_ENABLED) &&\
-    !defined(TARGET_IPHONE_SIMULATOR)
-// Must be included before jpeglib.
-#include <setjmp.h>
-#define HAVE_SETJMP
-#endif
-struct FILE;  // For jpeglib.h.
-
-// C++ build requires extern C for jpeg internals.
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#include <jpeglib.h>
-
-#ifdef __cplusplus
-}  // extern "C"
-#endif
-
-#include "libyuv/planar_functions.h"  // For CopyPlane().
-
-namespace libyuv {
-
-#ifdef HAVE_SETJMP
-struct SetJmpErrorMgr {
-  jpeg_error_mgr base;  // Must be at the top
-  jmp_buf setjmp_buffer;
-};
-#endif
-
-const int MJpegDecoder::kColorSpaceUnknown = JCS_UNKNOWN;
-const int MJpegDecoder::kColorSpaceGrayscale = JCS_GRAYSCALE;
-const int MJpegDecoder::kColorSpaceRgb = JCS_RGB;
-const int MJpegDecoder::kColorSpaceYCbCr = JCS_YCbCr;
-const int MJpegDecoder::kColorSpaceCMYK = JCS_CMYK;
-const int MJpegDecoder::kColorSpaceYCCK = JCS_YCCK;
-
-MJpegDecoder::MJpegDecoder()
-    : has_scanline_padding_(LIBYUV_FALSE),
-      num_outbufs_(0),
-      scanlines_(NULL),
-      scanlines_sizes_(NULL),
-      databuf_(NULL),
-      databuf_strides_(NULL) {
-  decompress_struct_ = new jpeg_decompress_struct;
-  source_mgr_ = new jpeg_source_mgr;
-#ifdef HAVE_SETJMP
-  error_mgr_ = new SetJmpErrorMgr;
-  decompress_struct_->err = jpeg_std_error(&error_mgr_->base);
-  // Override standard exit()-based error handler.
-  error_mgr_->base.error_exit = &ErrorHandler;
-#endif
-  decompress_struct_->client_data = NULL;
-  source_mgr_->init_source = &init_source;
-  source_mgr_->fill_input_buffer = &fill_input_buffer;
-  source_mgr_->skip_input_data = &skip_input_data;
-  source_mgr_->resync_to_restart = &jpeg_resync_to_restart;
-  source_mgr_->term_source = &term_source;
-  jpeg_create_decompress(decompress_struct_);
-  decompress_struct_->src = source_mgr_;
-  buf_vec_.buffers = &buf_;
-  buf_vec_.len = 1;
-}
-
-MJpegDecoder::~MJpegDecoder() {
-  jpeg_destroy_decompress(decompress_struct_);
-  delete decompress_struct_;
-  delete source_mgr_;
-#ifdef HAVE_SETJMP
-  delete error_mgr_;
-#endif
-  DestroyOutputBuffers();
-}
-
-LIBYUV_BOOL MJpegDecoder::LoadFrame(const uint8* src, size_t src_len) {
-  if (!ValidateJpeg(src, src_len)) {
-    return LIBYUV_FALSE;
-  }
-
-  buf_.data = src;
-  buf_.len = (int)(src_len);
-  buf_vec_.pos = 0;
-  decompress_struct_->client_data = &buf_vec_;
-#ifdef HAVE_SETJMP
-  if (setjmp(error_mgr_->setjmp_buffer)) {
-    // We called jpeg_read_header, it experienced an error, and we called
-    // longjmp() and rewound the stack to here. Return error.
-    return LIBYUV_FALSE;
-  }
-#endif
-  if (jpeg_read_header(decompress_struct_, TRUE) != JPEG_HEADER_OK) {
-    // ERROR: Bad MJPEG header
-    return LIBYUV_FALSE;
-  }
-  AllocOutputBuffers(GetNumComponents());
-  for (int i = 0; i < num_outbufs_; ++i) {
-    int scanlines_size = GetComponentScanlinesPerImcuRow(i);
-    if (scanlines_sizes_[i] != scanlines_size) {
-      if (scanlines_[i]) {
-        delete scanlines_[i];
-      }
-      scanlines_[i] = new uint8* [scanlines_size];
-      scanlines_sizes_[i] = scanlines_size;
-    }
-
-    // We allocate padding for the final scanline to pad it up to DCTSIZE bytes
-    // to avoid memory errors, since jpeglib only reads full MCUs blocks. For
-    // the preceding scanlines, the padding is not needed/wanted because the
-    // following addresses will already be valid (they are the initial bytes of
-    // the next scanline) and will be overwritten when jpeglib writes out that
-    // next scanline.
-    int databuf_stride = GetComponentStride(i);
-    int databuf_size = scanlines_size * databuf_stride;
-    if (databuf_strides_[i] != databuf_stride) {
-      if (databuf_[i]) {
-        delete databuf_[i];
-      }
-      databuf_[i] = new uint8[databuf_size];
-      databuf_strides_[i] = databuf_stride;
-    }
-
-    if (GetComponentStride(i) != GetComponentWidth(i)) {
-      has_scanline_padding_ = LIBYUV_TRUE;
-    }
-  }
-  return LIBYUV_TRUE;
-}
-
-static int DivideAndRoundUp(int numerator, int denominator) {
-  return (numerator + denominator - 1) / denominator;
-}
-
-static int DivideAndRoundDown(int numerator, int denominator) {
-  return numerator / denominator;
-}
-
-// Returns width of the last loaded frame.
-int MJpegDecoder::GetWidth() {
-  return decompress_struct_->image_width;
-}
-
-// Returns height of the last loaded frame.
-int MJpegDecoder::GetHeight() {
-  return decompress_struct_->image_height;
-}
-
-// Returns format of the last loaded frame. The return value is one of the
-// kColorSpace* constants.
-int MJpegDecoder::GetColorSpace() {
-  return decompress_struct_->jpeg_color_space;
-}
-
-// Number of color components in the color space.
-int MJpegDecoder::GetNumComponents() {
-  return decompress_struct_->num_components;
-}
-
-// Sample factors of the n-th component.
-int MJpegDecoder::GetHorizSampFactor(int component) {
-  return decompress_struct_->comp_info[component].h_samp_factor;
-}
-
-int MJpegDecoder::GetVertSampFactor(int component) {
-  return decompress_struct_->comp_info[component].v_samp_factor;
-}
-
-int MJpegDecoder::GetHorizSubSampFactor(int component) {
-  return decompress_struct_->max_h_samp_factor /
-      GetHorizSampFactor(component);
-}
-
-int MJpegDecoder::GetVertSubSampFactor(int component) {
-  return decompress_struct_->max_v_samp_factor /
-      GetVertSampFactor(component);
-}
-
-int MJpegDecoder::GetImageScanlinesPerImcuRow() {
-  return decompress_struct_->max_v_samp_factor * DCTSIZE;
-}
-
-int MJpegDecoder::GetComponentScanlinesPerImcuRow(int component) {
-  int vs = GetVertSubSampFactor(component);
-  return DivideAndRoundUp(GetImageScanlinesPerImcuRow(), vs);
-}
-
-int MJpegDecoder::GetComponentWidth(int component) {
-  int hs = GetHorizSubSampFactor(component);
-  return DivideAndRoundUp(GetWidth(), hs);
-}
-
-int MJpegDecoder::GetComponentHeight(int component) {
-  int vs = GetVertSubSampFactor(component);
-  return DivideAndRoundUp(GetHeight(), vs);
-}
-
-// Get width in bytes padded out to a multiple of DCTSIZE
-int MJpegDecoder::GetComponentStride(int component) {
-  return (GetComponentWidth(component) + DCTSIZE - 1) & ~(DCTSIZE - 1);
-}
-
-int MJpegDecoder::GetComponentSize(int component) {
-  return GetComponentWidth(component) * GetComponentHeight(component);
-}
-
-LIBYUV_BOOL MJpegDecoder::UnloadFrame() {
-#ifdef HAVE_SETJMP
-  if (setjmp(error_mgr_->setjmp_buffer)) {
-    // We called jpeg_abort_decompress, it experienced an error, and we called
-    // longjmp() and rewound the stack to here. Return error.
-    return LIBYUV_FALSE;
-  }
-#endif
-  jpeg_abort_decompress(decompress_struct_);
-  return LIBYUV_TRUE;
-}
-
-// TODO(fbarchard): Allow rectangle to be specified: x, y, width, height.
-LIBYUV_BOOL MJpegDecoder::DecodeToBuffers(
-    uint8** planes, int dst_width, int dst_height) {
-  if (dst_width != GetWidth() ||
-      dst_height > GetHeight()) {
-    // ERROR: Bad dimensions
-    return LIBYUV_FALSE;
-  }
-#ifdef HAVE_SETJMP
-  if (setjmp(error_mgr_->setjmp_buffer)) {
-    // We called into jpeglib, it experienced an error sometime during this
-    // function call, and we called longjmp() and rewound the stack to here.
-    // Return error.
-    return LIBYUV_FALSE;
-  }
-#endif
-  if (!StartDecode()) {
-    return LIBYUV_FALSE;
-  }
-  SetScanlinePointers(databuf_);
-  int lines_left = dst_height;
-  // Compute amount of lines to skip to implement vertical crop.
-  // TODO(fbarchard): Ensure skip is a multiple of maximum component
-  // subsample. ie 2
-  int skip = (GetHeight() - dst_height) / 2;
-  if (skip > 0) {
-    // There is no API to skip lines in the output data, so we read them
-    // into the temp buffer.
-    while (skip >= GetImageScanlinesPerImcuRow()) {
-      if (!DecodeImcuRow()) {
-        FinishDecode();
-        return LIBYUV_FALSE;
-      }
-      skip -= GetImageScanlinesPerImcuRow();
-    }
-    if (skip > 0) {
-      // Have a partial iMCU row left over to skip. Must read it and then
-      // copy the parts we want into the destination.
-      if (!DecodeImcuRow()) {
-        FinishDecode();
-        return LIBYUV_FALSE;
-      }
-      for (int i = 0; i < num_outbufs_; ++i) {
-        // TODO(fbarchard): Compute skip to avoid this
-        assert(skip % GetVertSubSampFactor(i) == 0);
-        int rows_to_skip =
-            DivideAndRoundDown(skip, GetVertSubSampFactor(i));
-        int scanlines_to_copy = GetComponentScanlinesPerImcuRow(i) -
-                                rows_to_skip;
-        int data_to_skip = rows_to_skip * GetComponentStride(i);
-        CopyPlane(databuf_[i] + data_to_skip, GetComponentStride(i),
-                  planes[i], GetComponentWidth(i),
-                  GetComponentWidth(i), scanlines_to_copy);
-        planes[i] += scanlines_to_copy * GetComponentWidth(i);
-      }
-      lines_left -= (GetImageScanlinesPerImcuRow() - skip);
-    }
-  }
-
-  // Read full MCUs but cropped horizontally
-  for (; lines_left > GetImageScanlinesPerImcuRow();
-         lines_left -= GetImageScanlinesPerImcuRow()) {
-    if (!DecodeImcuRow()) {
-      FinishDecode();
-      return LIBYUV_FALSE;
-    }
-    for (int i = 0; i < num_outbufs_; ++i) {
-      int scanlines_to_copy = GetComponentScanlinesPerImcuRow(i);
-      CopyPlane(databuf_[i], GetComponentStride(i),
-                planes[i], GetComponentWidth(i),
-                GetComponentWidth(i), scanlines_to_copy);
-      planes[i] += scanlines_to_copy * GetComponentWidth(i);
-    }
-  }
-
-  if (lines_left > 0) {
-    // Have a partial iMCU row left over to decode.
-    if (!DecodeImcuRow()) {
-      FinishDecode();
-      return LIBYUV_FALSE;
-    }
-    for (int i = 0; i < num_outbufs_; ++i) {
-      int scanlines_to_copy =
-          DivideAndRoundUp(lines_left, GetVertSubSampFactor(i));
-      CopyPlane(databuf_[i], GetComponentStride(i),
-                planes[i], GetComponentWidth(i),
-                GetComponentWidth(i), scanlines_to_copy);
-      planes[i] += scanlines_to_copy * GetComponentWidth(i);
-    }
-  }
-  return FinishDecode();
-}
-
-LIBYUV_BOOL MJpegDecoder::DecodeToCallback(CallbackFunction fn, void* opaque,
-    int dst_width, int dst_height) {
-  if (dst_width != GetWidth() ||
-      dst_height > GetHeight()) {
-    // ERROR: Bad dimensions
-    return LIBYUV_FALSE;
-  }
-#ifdef HAVE_SETJMP
-  if (setjmp(error_mgr_->setjmp_buffer)) {
-    // We called into jpeglib, it experienced an error sometime during this
-    // function call, and we called longjmp() and rewound the stack to here.
-    // Return error.
-    return LIBYUV_FALSE;
-  }
-#endif
-  if (!StartDecode()) {
-    return LIBYUV_FALSE;
-  }
-  SetScanlinePointers(databuf_);
-  int lines_left = dst_height;
-  // TODO(fbarchard): Compute amount of lines to skip to implement vertical crop
-  int skip = (GetHeight() - dst_height) / 2;
-  if (skip > 0) {
-    while (skip >= GetImageScanlinesPerImcuRow()) {
-      if (!DecodeImcuRow()) {
-        FinishDecode();
-        return LIBYUV_FALSE;
-      }
-      skip -= GetImageScanlinesPerImcuRow();
-    }
-    if (skip > 0) {
-      // Have a partial iMCU row left over to skip.
-      if (!DecodeImcuRow()) {
-        FinishDecode();
-        return LIBYUV_FALSE;
-      }
-      for (int i = 0; i < num_outbufs_; ++i) {
-        // TODO(fbarchard): Compute skip to avoid this
-        assert(skip % GetVertSubSampFactor(i) == 0);
-        int rows_to_skip = DivideAndRoundDown(skip, GetVertSubSampFactor(i));
-        int data_to_skip = rows_to_skip * GetComponentStride(i);
-        // Change our own data buffer pointers so we can pass them to the
-        // callback.
-        databuf_[i] += data_to_skip;
-      }
-      int scanlines_to_copy = GetImageScanlinesPerImcuRow() - skip;
-      (*fn)(opaque, databuf_, databuf_strides_, scanlines_to_copy);
-      // Now change them back.
-      for (int i = 0; i < num_outbufs_; ++i) {
-        int rows_to_skip = DivideAndRoundDown(skip, GetVertSubSampFactor(i));
-        int data_to_skip = rows_to_skip * GetComponentStride(i);
-        databuf_[i] -= data_to_skip;
-      }
-      lines_left -= scanlines_to_copy;
-    }
-  }
-  // Read full MCUs until we get to the crop point.
-  for (; lines_left >= GetImageScanlinesPerImcuRow();
-         lines_left -= GetImageScanlinesPerImcuRow()) {
-    if (!DecodeImcuRow()) {
-      FinishDecode();
-      return LIBYUV_FALSE;
-    }
-    (*fn)(opaque, databuf_, databuf_strides_, GetImageScanlinesPerImcuRow());
-  }
-  if (lines_left > 0) {
-    // Have a partial iMCU row left over to decode.
-    if (!DecodeImcuRow()) {
-      FinishDecode();
-      return LIBYUV_FALSE;
-    }
-    (*fn)(opaque, databuf_, databuf_strides_, lines_left);
-  }
-  return FinishDecode();
-}
-
-void MJpegDecoder::init_source(j_decompress_ptr cinfo) {
-  fill_input_buffer(cinfo);
-}
-
-boolean MJpegDecoder::fill_input_buffer(j_decompress_ptr cinfo) {
-  BufferVector* buf_vec = (BufferVector*)(cinfo->client_data);
-  if (buf_vec->pos >= buf_vec->len) {
-    assert(0 && "No more data");
-    // ERROR: No more data
-    return FALSE;
-  }
-  cinfo->src->next_input_byte = buf_vec->buffers[buf_vec->pos].data;
-  cinfo->src->bytes_in_buffer = buf_vec->buffers[buf_vec->pos].len;
-  ++buf_vec->pos;
-  return TRUE;
-}
-
-void MJpegDecoder::skip_input_data(j_decompress_ptr cinfo,
-                                   long num_bytes) {  // NOLINT
-  cinfo->src->next_input_byte += num_bytes;
-}
-
-void MJpegDecoder::term_source(j_decompress_ptr cinfo) {
-  // Nothing to do.
-}
-
-#ifdef HAVE_SETJMP
-void MJpegDecoder::ErrorHandler(j_common_ptr cinfo) {
-  // This is called when a jpeglib command experiences an error. Unfortunately
-  // jpeglib's error handling model is not very flexible, because it expects the
-  // error handler to not return--i.e., it wants the program to terminate. To
-  // recover from errors we use setjmp() as shown in their example. setjmp() is
-  // C's implementation for the "call with current continuation" functionality
-  // seen in some functional programming languages.
-  // A formatted message can be output, but is unsafe for release.
-#ifdef DEBUG
-  char buf[JMSG_LENGTH_MAX];
-  (*cinfo->err->format_message)(cinfo, buf);
-  // ERROR: Error in jpeglib: buf
-#endif
-
-  SetJmpErrorMgr* mgr = (SetJmpErrorMgr*)(cinfo->err);
-  // This rewinds the call stack to the point of the corresponding setjmp()
-  // and causes it to return (for a second time) with value 1.
-  longjmp(mgr->setjmp_buffer, 1);
-}
-#endif
-
-void MJpegDecoder::AllocOutputBuffers(int num_outbufs) {
-  if (num_outbufs != num_outbufs_) {
-    // We could perhaps optimize this case to resize the output buffers without
-    // necessarily having to delete and recreate each one, but it's not worth
-    // it.
-    DestroyOutputBuffers();
-
-    scanlines_ = new uint8** [num_outbufs];
-    scanlines_sizes_ = new int[num_outbufs];
-    databuf_ = new uint8* [num_outbufs];
-    databuf_strides_ = new int[num_outbufs];
-
-    for (int i = 0; i < num_outbufs; ++i) {
-      scanlines_[i] = NULL;
-      scanlines_sizes_[i] = 0;
-      databuf_[i] = NULL;
-      databuf_strides_[i] = 0;
-    }
-
-    num_outbufs_ = num_outbufs;
-  }
-}
-
-void MJpegDecoder::DestroyOutputBuffers() {
-  for (int i = 0; i < num_outbufs_; ++i) {
-    delete [] scanlines_[i];
-    delete [] databuf_[i];
-  }
-  delete [] scanlines_;
-  delete [] databuf_;
-  delete [] scanlines_sizes_;
-  delete [] databuf_strides_;
-  scanlines_ = NULL;
-  databuf_ = NULL;
-  scanlines_sizes_ = NULL;
-  databuf_strides_ = NULL;
-  num_outbufs_ = 0;
-}
-
-// JDCT_IFAST and do_block_smoothing improve performance substantially.
-LIBYUV_BOOL MJpegDecoder::StartDecode() {
-  decompress_struct_->raw_data_out = TRUE;
-  decompress_struct_->dct_method = JDCT_IFAST;  // JDCT_ISLOW is default
-  decompress_struct_->dither_mode = JDITHER_NONE;
-  // Not applicable to 'raw':
-  decompress_struct_->do_fancy_upsampling = LIBYUV_FALSE;
-  // Only for buffered mode:
-  decompress_struct_->enable_2pass_quant = LIBYUV_FALSE;
-  // Blocky but fast:
-  decompress_struct_->do_block_smoothing = LIBYUV_FALSE;
-
-  if (!jpeg_start_decompress(decompress_struct_)) {
-    // ERROR: Couldn't start JPEG decompressor";
-    return LIBYUV_FALSE;
-  }
-  return LIBYUV_TRUE;
-}
-
-LIBYUV_BOOL MJpegDecoder::FinishDecode() {
-  // jpeglib considers it an error if we finish without decoding the whole
-  // image, so we call "abort" rather than "finish".
-  jpeg_abort_decompress(decompress_struct_);
-  return LIBYUV_TRUE;
-}
-
-void MJpegDecoder::SetScanlinePointers(uint8** data) {
-  for (int i = 0; i < num_outbufs_; ++i) {
-    uint8* data_i = data[i];
-    for (int j = 0; j < scanlines_sizes_[i]; ++j) {
-      scanlines_[i][j] = data_i;
-      data_i += GetComponentStride(i);
-    }
-  }
-}
-
-inline LIBYUV_BOOL MJpegDecoder::DecodeImcuRow() {
-  return (unsigned int)(GetImageScanlinesPerImcuRow()) ==
-      jpeg_read_raw_data(decompress_struct_,
-                         scanlines_,
-                         GetImageScanlinesPerImcuRow());
-}
-
-// The helper function which recognizes the jpeg sub-sampling type.
-JpegSubsamplingType MJpegDecoder::JpegSubsamplingTypeHelper(
-    int* subsample_x, int* subsample_y, int number_of_components) {
-  if (number_of_components == 3) {  // Color images.
-    if (subsample_x[0] == 1 && subsample_y[0] == 1 &&
-        subsample_x[1] == 2 && subsample_y[1] == 2 &&
-        subsample_x[2] == 2 && subsample_y[2] == 2) {
-      return kJpegYuv420;
-    } else if (subsample_x[0] == 1 && subsample_y[0] == 1 &&
-        subsample_x[1] == 2 && subsample_y[1] == 1 &&
-        subsample_x[2] == 2 && subsample_y[2] == 1) {
-      return kJpegYuv422;
-    } else if (subsample_x[0] == 1 && subsample_y[0] == 1 &&
-        subsample_x[1] == 1 && subsample_y[1] == 1 &&
-        subsample_x[2] == 1 && subsample_y[2] == 1) {
-      return kJpegYuv444;
-    }
-  } else if (number_of_components == 1) {  // Grey-scale images.
-    if (subsample_x[0] == 1 && subsample_y[0] == 1) {
-      return kJpegYuv400;
-    }
-  }
-  return kJpegUnknown;
-}
-
-}  // namespace libyuv
-#endif  // HAVE_JPEG
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_validate.cc b/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_validate.cc
deleted file mode 100755
index 23d22d099b..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/mjpeg_validate.cc
+++ /dev/null
@@ -1,47 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/mjpeg_decoder.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Helper function to validate the jpeg appears intact.
-// TODO(fbarchard): Optimize case where SOI is found but EOI is not.
-LIBYUV_BOOL ValidateJpeg(const uint8* sample, size_t sample_size) {
-  size_t i;
-  if (sample_size < 64) {
-    // ERROR: Invalid jpeg size: sample_size
-    return LIBYUV_FALSE;
-  }
-  if (sample[0] != 0xff || sample[1] != 0xd8) {  // Start Of Image
-    // ERROR: Invalid jpeg initial start code
-    return LIBYUV_FALSE;
-  }
-  for (i = sample_size - 2; i > 1;) {
-    if (sample[i] != 0xd9) {
-      if (sample[i] == 0xff && sample[i + 1] == 0xd9) {  // End Of Image
-        return LIBYUV_TRUE;  // Success: Valid jpeg.
-      }
-      --i;
-    }
-    --i;
-  }
-  // ERROR: Invalid jpeg end code not found. Size sample_size
-  return LIBYUV_FALSE;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/planar_functions.cc b/drivers/theoraplayer/src/YUV/libyuv/src/planar_functions.cc
deleted file mode 100755
index f0a8989051..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/planar_functions.cc
+++ /dev/null
@@ -1,2238 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/planar_functions.h"
-
-#include <string.h>  // for memset()
-
-#include "libyuv/cpu_id.h"
-#ifdef HAVE_JPEG
-#include "libyuv/mjpeg_decoder.h"
-#endif
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Copy a plane of data
-LIBYUV_API
-void CopyPlane(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  int y;
-  void (*CopyRow)(const uint8* src, uint8* dst, int width) = CopyRow_C;
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      dst_stride_y == width) {
-    width *= height;
-    height = 1;
-    src_stride_y = dst_stride_y = 0;
-  }
-#if defined(HAS_COPYROW_X86)
-  if (TestCpuFlag(kCpuHasX86) && IS_ALIGNED(width, 4)) {
-    CopyRow = CopyRow_X86;
-  }
-#endif
-#if defined(HAS_COPYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 32) &&
-      IS_ALIGNED(src_y, 16) && IS_ALIGNED(src_stride_y, 16) &&
-      IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-    CopyRow = CopyRow_SSE2;
-  }
-#endif
-#if defined(HAS_COPYROW_ERMS)
-  if (TestCpuFlag(kCpuHasERMS)) {
-    CopyRow = CopyRow_ERMS;
-  }
-#endif
-#if defined(HAS_COPYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 32)) {
-    CopyRow = CopyRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_MIPS)
-  if (TestCpuFlag(kCpuHasMIPS)) {
-    CopyRow = CopyRow_MIPS;
-  }
-#endif
-
-  // Copy plane
-  for (y = 0; y < height; ++y) {
-    CopyRow(src_y, dst_y, width);
-    src_y += src_stride_y;
-    dst_y += dst_stride_y;
-  }
-}
-
-// Copy I422.
-LIBYUV_API
-int I422Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height) {
-  int halfwidth = (width + 1) >> 1;
-  if (!src_y || !src_u || !src_v ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (height - 1) * src_stride_u;
-    src_v = src_v + (height - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  CopyPlane(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, height);
-  CopyPlane(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, height);
-  return 0;
-}
-
-// Copy I444.
-LIBYUV_API
-int I444Copy(const uint8* src_y, int src_stride_y,
-             const uint8* src_u, int src_stride_u,
-             const uint8* src_v, int src_stride_v,
-             uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int width, int height) {
-  if (!src_y || !src_u || !src_v ||
-      !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (height - 1) * src_stride_u;
-    src_v = src_v + (height - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  CopyPlane(src_u, src_stride_u, dst_u, dst_stride_u, width, height);
-  CopyPlane(src_v, src_stride_v, dst_v, dst_stride_v, width, height);
-  return 0;
-}
-
-// Copy I400.
-LIBYUV_API
-int I400ToI400(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  if (!src_y || !dst_y || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  return 0;
-}
-
-// Convert I420 to I400.
-LIBYUV_API
-int I420ToI400(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  if (!src_y || !dst_y || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  return 0;
-}
-
-// Mirror a plane of data.
-void MirrorPlane(const uint8* src_y, int src_stride_y,
-                 uint8* dst_y, int dst_stride_y,
-                 int width, int height) {
-  int y;
-  void (*MirrorRow)(const uint8* src, uint8* dst, int width) = MirrorRow_C;
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-#if defined(HAS_MIRRORROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 16)) {
-    MirrorRow = MirrorRow_NEON;
-  }
-#endif
-#if defined(HAS_MIRRORROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 16)) {
-    MirrorRow = MirrorRow_SSE2;
-  }
-#endif
-#if defined(HAS_MIRRORROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(src_y, 16) && IS_ALIGNED(src_stride_y, 16) &&
-      IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-    MirrorRow = MirrorRow_SSSE3;
-  }
-#endif
-#if defined(HAS_MIRRORROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 32)) {
-    MirrorRow = MirrorRow_AVX2;
-  }
-#endif
-
-  // Mirror plane
-  for (y = 0; y < height; ++y) {
-    MirrorRow(src_y, dst_y, width);
-    src_y += src_stride_y;
-    dst_y += dst_stride_y;
-  }
-}
-
-// Convert YUY2 to I422.
-LIBYUV_API
-int YUY2ToI422(const uint8* src_yuy2, int src_stride_yuy2,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  int y;
-  void (*YUY2ToUV422Row)(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix) =
-      YUY2ToUV422Row_C;
-  void (*YUY2ToYRow)(const uint8* src_yuy2, uint8* dst_y, int pix) =
-      YUY2ToYRow_C;
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_yuy2 = src_yuy2 + (height - 1) * src_stride_yuy2;
-    src_stride_yuy2 = -src_stride_yuy2;
-  }
-  // Coalesce rows.
-  if (src_stride_yuy2 == width * 2 &&
-      dst_stride_y == width &&
-      dst_stride_u * 2 == width &&
-      dst_stride_v * 2 == width) {
-    width *= height;
-    height = 1;
-    src_stride_yuy2 = dst_stride_y = dst_stride_u = dst_stride_v = 0;
-  }
-#if defined(HAS_YUY2TOYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    YUY2ToUV422Row = YUY2ToUV422Row_Any_SSE2;
-    YUY2ToYRow = YUY2ToYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToUV422Row = YUY2ToUV422Row_Unaligned_SSE2;
-      YUY2ToYRow = YUY2ToYRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_yuy2, 16) && IS_ALIGNED(src_stride_yuy2, 16)) {
-        YUY2ToUV422Row = YUY2ToUV422Row_SSE2;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          YUY2ToYRow = YUY2ToYRow_SSE2;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    YUY2ToUV422Row = YUY2ToUV422Row_Any_AVX2;
-    YUY2ToYRow = YUY2ToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      YUY2ToUV422Row = YUY2ToUV422Row_AVX2;
-      YUY2ToYRow = YUY2ToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_YUY2TOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    YUY2ToYRow = YUY2ToYRow_Any_NEON;
-    if (width >= 16) {
-      YUY2ToUV422Row = YUY2ToUV422Row_Any_NEON;
-    }
-    if (IS_ALIGNED(width, 16)) {
-      YUY2ToYRow = YUY2ToYRow_NEON;
-      YUY2ToUV422Row = YUY2ToUV422Row_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    YUY2ToUV422Row(src_yuy2, dst_u, dst_v, width);
-    YUY2ToYRow(src_yuy2, dst_y, width);
-    src_yuy2 += src_stride_yuy2;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// Convert UYVY to I422.
-LIBYUV_API
-int UYVYToI422(const uint8* src_uyvy, int src_stride_uyvy,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  int y;
-  void (*UYVYToUV422Row)(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix) =
-      UYVYToUV422Row_C;
-  void (*UYVYToYRow)(const uint8* src_uyvy,
-                     uint8* dst_y, int pix) = UYVYToYRow_C;
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy;
-    src_stride_uyvy = -src_stride_uyvy;
-  }
-  // Coalesce rows.
-  if (src_stride_uyvy == width * 2 &&
-      dst_stride_y == width &&
-      dst_stride_u * 2 == width &&
-      dst_stride_v * 2 == width) {
-    width *= height;
-    height = 1;
-    src_stride_uyvy = dst_stride_y = dst_stride_u = dst_stride_v = 0;
-  }
-#if defined(HAS_UYVYTOYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16) {
-    UYVYToUV422Row = UYVYToUV422Row_Any_SSE2;
-    UYVYToYRow = UYVYToYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      UYVYToUV422Row = UYVYToUV422Row_Unaligned_SSE2;
-      UYVYToYRow = UYVYToYRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_uyvy, 16) && IS_ALIGNED(src_stride_uyvy, 16)) {
-        UYVYToUV422Row = UYVYToUV422Row_SSE2;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          UYVYToYRow = UYVYToYRow_SSE2;
-        }
-      }
-    }
-  }
-#endif
-#if defined(HAS_UYVYTOYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 32) {
-    UYVYToUV422Row = UYVYToUV422Row_Any_AVX2;
-    UYVYToYRow = UYVYToYRow_Any_AVX2;
-    if (IS_ALIGNED(width, 32)) {
-      UYVYToUV422Row = UYVYToUV422Row_AVX2;
-      UYVYToYRow = UYVYToYRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_UYVYTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    UYVYToYRow = UYVYToYRow_Any_NEON;
-    if (width >= 16) {
-      UYVYToUV422Row = UYVYToUV422Row_Any_NEON;
-    }
-    if (IS_ALIGNED(width, 16)) {
-      UYVYToYRow = UYVYToYRow_NEON;
-      UYVYToUV422Row = UYVYToUV422Row_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    UYVYToUV422Row(src_uyvy, dst_u, dst_v, width);
-    UYVYToYRow(src_uyvy, dst_y, width);
-    src_uyvy += src_stride_uyvy;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// Mirror I400 with optional flipping
-LIBYUV_API
-int I400Mirror(const uint8* src_y, int src_stride_y,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  if (!src_y || !dst_y ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-
-  MirrorPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  return 0;
-}
-
-// Mirror I420 with optional flipping
-LIBYUV_API
-int I420Mirror(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  if (!src_y || !src_u || !src_v || !dst_y || !dst_u || !dst_v ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    halfheight = (height + 1) >> 1;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (halfheight - 1) * src_stride_u;
-    src_v = src_v + (halfheight - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-
-  if (dst_y) {
-    MirrorPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
-  }
-  MirrorPlane(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight);
-  MirrorPlane(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight);
-  return 0;
-}
-
-// ARGB mirror.
-LIBYUV_API
-int ARGBMirror(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  int y;
-  void (*ARGBMirrorRow)(const uint8* src, uint8* dst, int width) =
-      ARGBMirrorRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-
-#if defined(HAS_ARGBMIRRORROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBMirrorRow = ARGBMirrorRow_SSSE3;
-  }
-#endif
-#if defined(HAS_ARGBMIRRORROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 8)) {
-    ARGBMirrorRow = ARGBMirrorRow_AVX2;
-  }
-#endif
-#if defined(HAS_ARGBMIRRORROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 4)) {
-    ARGBMirrorRow = ARGBMirrorRow_NEON;
-  }
-#endif
-
-  // Mirror plane
-  for (y = 0; y < height; ++y) {
-    ARGBMirrorRow(src_argb, dst_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Get a blender that optimized for the CPU, alignment and pixel count.
-// As there are 6 blenders to choose from, the caller should try to use
-// the same blend function for all pixels if possible.
-LIBYUV_API
-ARGBBlendRow GetARGBBlend() {
-  void (*ARGBBlendRow)(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width) = ARGBBlendRow_C;
-#if defined(HAS_ARGBBLENDROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3)) {
-    ARGBBlendRow = ARGBBlendRow_SSSE3;
-    return ARGBBlendRow;
-  }
-#endif
-#if defined(HAS_ARGBBLENDROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    ARGBBlendRow = ARGBBlendRow_SSE2;
-  }
-#endif
-#if defined(HAS_ARGBBLENDROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    ARGBBlendRow = ARGBBlendRow_NEON;
-  }
-#endif
-  return ARGBBlendRow;
-}
-
-// Alpha Blend 2 ARGB images and store to destination.
-LIBYUV_API
-int ARGBBlend(const uint8* src_argb0, int src_stride_argb0,
-              const uint8* src_argb1, int src_stride_argb1,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height) {
-  int y;
-  void (*ARGBBlendRow)(const uint8* src_argb, const uint8* src_argb1,
-                       uint8* dst_argb, int width) = GetARGBBlend();
-  if (!src_argb0 || !src_argb1 || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb0 == width * 4 &&
-      src_stride_argb1 == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb0 = src_stride_argb1 = dst_stride_argb = 0;
-  }
-
-  for (y = 0; y < height; ++y) {
-    ARGBBlendRow(src_argb0, src_argb1, dst_argb, width);
-    src_argb0 += src_stride_argb0;
-    src_argb1 += src_stride_argb1;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Multiply 2 ARGB images and store to destination.
-LIBYUV_API
-int ARGBMultiply(const uint8* src_argb0, int src_stride_argb0,
-                 const uint8* src_argb1, int src_stride_argb1,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height) {
-  int y;
-  void (*ARGBMultiplyRow)(const uint8* src0, const uint8* src1, uint8* dst,
-                          int width) = ARGBMultiplyRow_C;
-  if (!src_argb0 || !src_argb1 || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb0 == width * 4 &&
-      src_stride_argb1 == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb0 = src_stride_argb1 = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBMULTIPLYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    ARGBMultiplyRow = ARGBMultiplyRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBMultiplyRow = ARGBMultiplyRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBMULTIPLYROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    ARGBMultiplyRow = ARGBMultiplyRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBMultiplyRow = ARGBMultiplyRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBMULTIPLYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBMultiplyRow = ARGBMultiplyRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBMultiplyRow = ARGBMultiplyRow_NEON;
-    }
-  }
-#endif
-
-  // Multiply plane
-  for (y = 0; y < height; ++y) {
-    ARGBMultiplyRow(src_argb0, src_argb1, dst_argb, width);
-    src_argb0 += src_stride_argb0;
-    src_argb1 += src_stride_argb1;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Add 2 ARGB images and store to destination.
-LIBYUV_API
-int ARGBAdd(const uint8* src_argb0, int src_stride_argb0,
-            const uint8* src_argb1, int src_stride_argb1,
-            uint8* dst_argb, int dst_stride_argb,
-            int width, int height) {
-  int y;
-  void (*ARGBAddRow)(const uint8* src0, const uint8* src1, uint8* dst,
-                     int width) = ARGBAddRow_C;
-  if (!src_argb0 || !src_argb1 || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb0 == width * 4 &&
-      src_stride_argb1 == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb0 = src_stride_argb1 = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBADDROW_SSE2) && defined(_MSC_VER)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    ARGBAddRow = ARGBAddRow_SSE2;
-  }
-#endif
-#if defined(HAS_ARGBADDROW_SSE2) && !defined(_MSC_VER)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    ARGBAddRow = ARGBAddRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBAddRow = ARGBAddRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBADDROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    ARGBAddRow = ARGBAddRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBAddRow = ARGBAddRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBADDROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBAddRow = ARGBAddRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBAddRow = ARGBAddRow_NEON;
-    }
-  }
-#endif
-
-  // Add plane
-  for (y = 0; y < height; ++y) {
-    ARGBAddRow(src_argb0, src_argb1, dst_argb, width);
-    src_argb0 += src_stride_argb0;
-    src_argb1 += src_stride_argb1;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Subtract 2 ARGB images and store to destination.
-LIBYUV_API
-int ARGBSubtract(const uint8* src_argb0, int src_stride_argb0,
-                 const uint8* src_argb1, int src_stride_argb1,
-                 uint8* dst_argb, int dst_stride_argb,
-                 int width, int height) {
-  int y;
-  void (*ARGBSubtractRow)(const uint8* src0, const uint8* src1, uint8* dst,
-                          int width) = ARGBSubtractRow_C;
-  if (!src_argb0 || !src_argb1 || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb0 == width * 4 &&
-      src_stride_argb1 == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb0 = src_stride_argb1 = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBSUBTRACTROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    ARGBSubtractRow = ARGBSubtractRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBSubtractRow = ARGBSubtractRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBSUBTRACTROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    ARGBSubtractRow = ARGBSubtractRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBSubtractRow = ARGBSubtractRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBSUBTRACTROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBSubtractRow = ARGBSubtractRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBSubtractRow = ARGBSubtractRow_NEON;
-    }
-  }
-#endif
-
-  // Subtract plane
-  for (y = 0; y < height; ++y) {
-    ARGBSubtractRow(src_argb0, src_argb1, dst_argb, width);
-    src_argb0 += src_stride_argb0;
-    src_argb1 += src_stride_argb1;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert I422 to BGRA.
-LIBYUV_API
-int I422ToBGRA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_bgra, int dst_stride_bgra,
-               int width, int height) {
-  int y;
-  void (*I422ToBGRARow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToBGRARow_C;
-  if (!src_y || !src_u || !src_v ||
-      !dst_bgra ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_bgra = dst_bgra + (height - 1) * dst_stride_bgra;
-    dst_stride_bgra = -dst_stride_bgra;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_bgra == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_bgra = 0;
-  }
-#if defined(HAS_I422TOBGRAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    I422ToBGRARow = I422ToBGRARow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToBGRARow = I422ToBGRARow_NEON;
-    }
-  }
-#elif defined(HAS_I422TOBGRAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToBGRARow = I422ToBGRARow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToBGRARow = I422ToBGRARow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_bgra, 16) && IS_ALIGNED(dst_stride_bgra, 16)) {
-        I422ToBGRARow = I422ToBGRARow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_I422TOBGRAROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2) &&
-      IS_ALIGNED(dst_bgra, 4) && IS_ALIGNED(dst_stride_bgra, 4)) {
-    I422ToBGRARow = I422ToBGRARow_MIPS_DSPR2;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    I422ToBGRARow(src_y, src_u, src_v, dst_bgra, width);
-    dst_bgra += dst_stride_bgra;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert I422 to ABGR.
-LIBYUV_API
-int I422ToABGR(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_abgr, int dst_stride_abgr,
-               int width, int height) {
-  int y;
-  void (*I422ToABGRRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToABGRRow_C;
-  if (!src_y || !src_u || !src_v ||
-      !dst_abgr ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_abgr = dst_abgr + (height - 1) * dst_stride_abgr;
-    dst_stride_abgr = -dst_stride_abgr;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_abgr == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_abgr = 0;
-  }
-#if defined(HAS_I422TOABGRROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    I422ToABGRRow = I422ToABGRRow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToABGRRow = I422ToABGRRow_NEON;
-    }
-  }
-#elif defined(HAS_I422TOABGRROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToABGRRow = I422ToABGRRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToABGRRow = I422ToABGRRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_abgr, 16) && IS_ALIGNED(dst_stride_abgr, 16)) {
-        I422ToABGRRow = I422ToABGRRow_SSSE3;
-      }
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    I422ToABGRRow(src_y, src_u, src_v, dst_abgr, width);
-    dst_abgr += dst_stride_abgr;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert I422 to RGBA.
-LIBYUV_API
-int I422ToRGBA(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height) {
-  int y;
-  void (*I422ToRGBARow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToRGBARow_C;
-  if (!src_y || !src_u || !src_v ||
-      !dst_rgba ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba;
-    dst_stride_rgba = -dst_stride_rgba;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      src_stride_u * 2 == width &&
-      src_stride_v * 2 == width &&
-      dst_stride_rgba == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = src_stride_u = src_stride_v = dst_stride_rgba = 0;
-  }
-#if defined(HAS_I422TORGBAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    I422ToRGBARow = I422ToRGBARow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToRGBARow = I422ToRGBARow_NEON;
-    }
-  }
-#elif defined(HAS_I422TORGBAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    I422ToRGBARow = I422ToRGBARow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      I422ToRGBARow = I422ToRGBARow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_rgba, 16) && IS_ALIGNED(dst_stride_rgba, 16)) {
-        I422ToRGBARow = I422ToRGBARow_SSSE3;
-      }
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    I422ToRGBARow(src_y, src_u, src_v, dst_rgba, width);
-    dst_rgba += dst_stride_rgba;
-    src_y += src_stride_y;
-    src_u += src_stride_u;
-    src_v += src_stride_v;
-  }
-  return 0;
-}
-
-// Convert NV12 to RGB565.
-LIBYUV_API
-int NV12ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_uv, int src_stride_uv,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height) {
-  int y;
-  void (*NV12ToRGB565Row)(const uint8* y_buf,
-                          const uint8* uv_buf,
-                          uint8* rgb_buf,
-                          int width) = NV12ToRGB565Row_C;
-  if (!src_y || !src_uv || !dst_rgb565 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565;
-    dst_stride_rgb565 = -dst_stride_rgb565;
-  }
-#if defined(HAS_NV12TORGB565ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    NV12ToRGB565Row = NV12ToRGB565Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToRGB565Row = NV12ToRGB565Row_SSSE3;
-    }
-  }
-#elif defined(HAS_NV12TORGB565ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    NV12ToRGB565Row = NV12ToRGB565Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      NV12ToRGB565Row = NV12ToRGB565Row_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    NV12ToRGB565Row(src_y, src_uv, dst_rgb565, width);
-    dst_rgb565 += dst_stride_rgb565;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_uv += src_stride_uv;
-    }
-  }
-  return 0;
-}
-
-// Convert NV21 to RGB565.
-LIBYUV_API
-int NV21ToRGB565(const uint8* src_y, int src_stride_y,
-                 const uint8* src_vu, int src_stride_vu,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height) {
-  int y;
-  void (*NV21ToRGB565Row)(const uint8* y_buf,
-                          const uint8* src_vu,
-                          uint8* rgb_buf,
-                          int width) = NV21ToRGB565Row_C;
-  if (!src_y || !src_vu || !dst_rgb565 ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565;
-    dst_stride_rgb565 = -dst_stride_rgb565;
-  }
-#if defined(HAS_NV21TORGB565ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    NV21ToRGB565Row = NV21ToRGB565Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      NV21ToRGB565Row = NV21ToRGB565Row_SSSE3;
-    }
-  }
-#elif defined(HAS_NV21TORGB565ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    NV21ToRGB565Row = NV21ToRGB565Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      NV21ToRGB565Row = NV21ToRGB565Row_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    NV21ToRGB565Row(src_y, src_vu, dst_rgb565, width);
-    dst_rgb565 += dst_stride_rgb565;
-    src_y += src_stride_y;
-    if (y & 1) {
-      src_vu += src_stride_vu;
-    }
-  }
-  return 0;
-}
-
-LIBYUV_API
-void SetPlane(uint8* dst_y, int dst_stride_y,
-              int width, int height,
-              uint32 value) {
-  int y;
-  uint32 v32 = value | (value << 8) | (value << 16) | (value << 24);
-  void (*SetRow)(uint8* dst, uint32 value, int pix) = SetRow_C;
-  // Coalesce rows.
-  if (dst_stride_y == width) {
-    width *= height;
-    height = 1;
-    dst_stride_y = 0;
-  }
-#if defined(HAS_SETROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) &&
-      IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-    SetRow = SetRow_NEON;
-  }
-#endif
-#if defined(HAS_SETROW_X86)
-  if (TestCpuFlag(kCpuHasX86) && IS_ALIGNED(width, 4)) {
-    SetRow = SetRow_X86;
-  }
-#endif
-
-  // Set plane
-  for (y = 0; y < height; ++y) {
-    SetRow(dst_y, v32, width);
-    dst_y += dst_stride_y;
-  }
-}
-
-// Draw a rectangle into I420
-LIBYUV_API
-int I420Rect(uint8* dst_y, int dst_stride_y,
-             uint8* dst_u, int dst_stride_u,
-             uint8* dst_v, int dst_stride_v,
-             int x, int y,
-             int width, int height,
-             int value_y, int value_u, int value_v) {
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  uint8* start_y = dst_y + y * dst_stride_y + x;
-  uint8* start_u = dst_u + (y / 2) * dst_stride_u + (x / 2);
-  uint8* start_v = dst_v + (y / 2) * dst_stride_v + (x / 2);
-  if (!dst_y || !dst_u || !dst_v ||
-      width <= 0 || height <= 0 ||
-      x < 0 || y < 0 ||
-      value_y < 0 || value_y > 255 ||
-      value_u < 0 || value_u > 255 ||
-      value_v < 0 || value_v > 255) {
-    return -1;
-  }
-
-  SetPlane(start_y, dst_stride_y, width, height, value_y);
-  SetPlane(start_u, dst_stride_u, halfwidth, halfheight, value_u);
-  SetPlane(start_v, dst_stride_v, halfwidth, halfheight, value_v);
-  return 0;
-}
-
-// Draw a rectangle into ARGB
-LIBYUV_API
-int ARGBRect(uint8* dst_argb, int dst_stride_argb,
-             int dst_x, int dst_y,
-             int width, int height,
-             uint32 value) {
-  if (!dst_argb ||
-      width <= 0 || height <= 0 ||
-      dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-  dst_argb += dst_y * dst_stride_argb + dst_x * 4;
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_SETROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBSetRows_NEON(dst_argb, value, width, dst_stride_argb, height);
-    return 0;
-  }
-#endif
-#if defined(HAS_SETROW_X86)
-  if (TestCpuFlag(kCpuHasX86)) {
-    ARGBSetRows_X86(dst_argb, value, width, dst_stride_argb, height);
-    return 0;
-  }
-#endif
-  ARGBSetRows_C(dst_argb, value, width, dst_stride_argb, height);
-  return 0;
-}
-
-// Convert unattentuated ARGB to preattenuated ARGB.
-// An unattenutated ARGB alpha blend uses the formula
-// p = a * f + (1 - a) * b
-// where
-//   p is output pixel
-//   f is foreground pixel
-//   b is background pixel
-//   a is alpha value from foreground pixel
-// An preattenutated ARGB alpha blend uses the formula
-// p = f + (1 - a) * b
-// where
-//   f is foreground pixel premultiplied by alpha
-
-LIBYUV_API
-int ARGBAttenuate(const uint8* src_argb, int src_stride_argb,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int width, int height) {
-  int y;
-  void (*ARGBAttenuateRow)(const uint8* src_argb, uint8* dst_argb,
-                           int width) = ARGBAttenuateRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBATTENUATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBAttenuateRow = ARGBAttenuateRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBAttenuateRow = ARGBAttenuateRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBATTENUATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 4) {
-    ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBAttenuateRow = ARGBAttenuateRow_SSSE3;
-    }
-  }
-#endif
-#if defined(HAS_ARGBATTENUATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBAttenuateRow = ARGBAttenuateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBATTENUATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBAttenuateRow = ARGBAttenuateRow_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBAttenuateRow(src_argb, dst_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert preattentuated ARGB to unattenuated ARGB.
-LIBYUV_API
-int ARGBUnattenuate(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height) {
-  int y;
-  void (*ARGBUnattenuateRow)(const uint8* src_argb, uint8* dst_argb,
-                             int width) = ARGBUnattenuateRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBUNATTENUATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    ARGBUnattenuateRow = ARGBUnattenuateRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBUnattenuateRow = ARGBUnattenuateRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBUNATTENUATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    ARGBUnattenuateRow = ARGBUnattenuateRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBUnattenuateRow = ARGBUnattenuateRow_AVX2;
-    }
-  }
-#endif
-// TODO(fbarchard): Neon version.
-
-  for (y = 0; y < height; ++y) {
-    ARGBUnattenuateRow(src_argb, dst_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Convert ARGB to Grayed ARGB.
-LIBYUV_API
-int ARGBGrayTo(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height) {
-  int y;
-  void (*ARGBGrayRow)(const uint8* src_argb, uint8* dst_argb,
-                      int width) = ARGBGrayRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBGRAYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 8) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBGrayRow = ARGBGrayRow_SSSE3;
-  }
-#elif defined(HAS_ARGBGRAYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBGrayRow = ARGBGrayRow_NEON;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBGrayRow(src_argb, dst_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Make a rectangle of ARGB gray scale.
-LIBYUV_API
-int ARGBGray(uint8* dst_argb, int dst_stride_argb,
-             int dst_x, int dst_y,
-             int width, int height) {
-  int y;
-  void (*ARGBGrayRow)(const uint8* src_argb, uint8* dst_argb,
-                      int width) = ARGBGrayRow_C;
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || width <= 0 || height <= 0 || dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBGRAYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 8) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBGrayRow = ARGBGrayRow_SSSE3;
-  }
-#elif defined(HAS_ARGBGRAYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBGrayRow = ARGBGrayRow_NEON;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    ARGBGrayRow(dst, dst, width);
-    dst += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Make a rectangle of ARGB Sepia tone.
-LIBYUV_API
-int ARGBSepia(uint8* dst_argb, int dst_stride_argb,
-              int dst_x, int dst_y, int width, int height) {
-  int y;
-  void (*ARGBSepiaRow)(uint8* dst_argb, int width) = ARGBSepiaRow_C;
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || width <= 0 || height <= 0 || dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBSEPIAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 8) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBSepiaRow = ARGBSepiaRow_SSSE3;
-  }
-#elif defined(HAS_ARGBSEPIAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBSepiaRow = ARGBSepiaRow_NEON;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    ARGBSepiaRow(dst, width);
-    dst += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Apply a 4x4 matrix to each ARGB pixel.
-// Note: Normally for shading, but can be used to swizzle or invert.
-LIBYUV_API
-int ARGBColorMatrix(const uint8* src_argb, int src_stride_argb,
-                    uint8* dst_argb, int dst_stride_argb,
-                    const int8* matrix_argb,
-                    int width, int height) {
-  int y;
-  void (*ARGBColorMatrixRow)(const uint8* src_argb, uint8* dst_argb,
-      const int8* matrix_argb, int width) = ARGBColorMatrixRow_C;
-  if (!src_argb || !dst_argb || !matrix_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBCOLORMATRIXROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 8) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBColorMatrixRow = ARGBColorMatrixRow_SSSE3;
-  }
-#elif defined(HAS_ARGBCOLORMATRIXROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBColorMatrixRow = ARGBColorMatrixRow_NEON;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    ARGBColorMatrixRow(src_argb, dst_argb, matrix_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Apply a 4x3 matrix to each ARGB pixel.
-// Deprecated.
-LIBYUV_API
-int RGBColorMatrix(uint8* dst_argb, int dst_stride_argb,
-                   const int8* matrix_rgb,
-                   int dst_x, int dst_y, int width, int height) {
-  SIMD_ALIGNED(int8 matrix_argb[16]);
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || !matrix_rgb || width <= 0 || height <= 0 ||
-      dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-
-  // Convert 4x3 7 bit matrix to 4x4 6 bit matrix.
-  matrix_argb[0] = matrix_rgb[0] / 2;
-  matrix_argb[1] = matrix_rgb[1] / 2;
-  matrix_argb[2] = matrix_rgb[2] / 2;
-  matrix_argb[3] = matrix_rgb[3] / 2;
-  matrix_argb[4] = matrix_rgb[4] / 2;
-  matrix_argb[5] = matrix_rgb[5] / 2;
-  matrix_argb[6] = matrix_rgb[6] / 2;
-  matrix_argb[7] = matrix_rgb[7] / 2;
-  matrix_argb[8] = matrix_rgb[8] / 2;
-  matrix_argb[9] = matrix_rgb[9] / 2;
-  matrix_argb[10] = matrix_rgb[10] / 2;
-  matrix_argb[11] = matrix_rgb[11] / 2;
-  matrix_argb[14] = matrix_argb[13] = matrix_argb[12] = 0;
-  matrix_argb[15] = 64;  // 1.0
-
-  return ARGBColorMatrix((const uint8*)(dst), dst_stride_argb,
-                         dst, dst_stride_argb,
-                         &matrix_argb[0], width, height);
-}
-
-// Apply a color table each ARGB pixel.
-// Table contains 256 ARGB values.
-LIBYUV_API
-int ARGBColorTable(uint8* dst_argb, int dst_stride_argb,
-                   const uint8* table_argb,
-                   int dst_x, int dst_y, int width, int height) {
-  int y;
-  void (*ARGBColorTableRow)(uint8* dst_argb, const uint8* table_argb,
-                            int width) = ARGBColorTableRow_C;
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || !table_argb || width <= 0 || height <= 0 ||
-      dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBCOLORTABLEROW_X86)
-  if (TestCpuFlag(kCpuHasX86)) {
-    ARGBColorTableRow = ARGBColorTableRow_X86;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    ARGBColorTableRow(dst, table_argb, width);
-    dst += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Apply a color table each ARGB pixel but preserve destination alpha.
-// Table contains 256 ARGB values.
-LIBYUV_API
-int RGBColorTable(uint8* dst_argb, int dst_stride_argb,
-                  const uint8* table_argb,
-                  int dst_x, int dst_y, int width, int height) {
-  int y;
-  void (*RGBColorTableRow)(uint8* dst_argb, const uint8* table_argb,
-                           int width) = RGBColorTableRow_C;
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || !table_argb || width <= 0 || height <= 0 ||
-      dst_x < 0 || dst_y < 0) {
-    return -1;
-  }
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_RGBCOLORTABLEROW_X86)
-  if (TestCpuFlag(kCpuHasX86)) {
-    RGBColorTableRow = RGBColorTableRow_X86;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    RGBColorTableRow(dst, table_argb, width);
-    dst += dst_stride_argb;
-  }
-  return 0;
-}
-
-// ARGBQuantize is used to posterize art.
-// e.g. rgb / qvalue * qvalue + qvalue / 2
-// But the low levels implement efficiently with 3 parameters, and could be
-// used for other high level operations.
-// dst_argb[0] = (b * scale >> 16) * interval_size + interval_offset;
-// where scale is 1 / interval_size as a fixed point value.
-// The divide is replaces with a multiply by reciprocal fixed point multiply.
-// Caveat - although SSE2 saturates, the C function does not and should be used
-// with care if doing anything but quantization.
-LIBYUV_API
-int ARGBQuantize(uint8* dst_argb, int dst_stride_argb,
-                 int scale, int interval_size, int interval_offset,
-                 int dst_x, int dst_y, int width, int height) {
-  int y;
-  void (*ARGBQuantizeRow)(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width) = ARGBQuantizeRow_C;
-  uint8* dst = dst_argb + dst_y * dst_stride_argb + dst_x * 4;
-  if (!dst_argb || width <= 0 || height <= 0 || dst_x < 0 || dst_y < 0 ||
-      interval_size < 1 || interval_size > 255) {
-    return -1;
-  }
-  // Coalesce rows.
-  if (dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBQUANTIZEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBQuantizeRow = ARGBQuantizeRow_SSE2;
-  }
-#elif defined(HAS_ARGBQUANTIZEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBQuantizeRow = ARGBQuantizeRow_NEON;
-  }
-#endif
-  for (y = 0; y < height; ++y) {
-    ARGBQuantizeRow(dst, scale, interval_size, interval_offset, width);
-    dst += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Computes table of cumulative sum for image where the value is the sum
-// of all values above and to the left of the entry. Used by ARGBBlur.
-LIBYUV_API
-int ARGBComputeCumulativeSum(const uint8* src_argb, int src_stride_argb,
-                             int32* dst_cumsum, int dst_stride32_cumsum,
-                             int width, int height) {
-  int y;
-  void (*ComputeCumulativeSumRow)(const uint8* row, int32* cumsum,
-      const int32* previous_cumsum, int width) = ComputeCumulativeSumRow_C;
-  int32* previous_cumsum = dst_cumsum;
-  if (!dst_cumsum || !src_argb || width <= 0 || height <= 0) {
-    return -1;
-  }
-#if defined(HAS_CUMULATIVESUMTOAVERAGEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    ComputeCumulativeSumRow = ComputeCumulativeSumRow_SSE2;
-  }
-#endif
-  memset(dst_cumsum, 0, width * sizeof(dst_cumsum[0]) * 4);  // 4 int per pixel.
-  for (y = 0; y < height; ++y) {
-    ComputeCumulativeSumRow(src_argb, dst_cumsum, previous_cumsum, width);
-    previous_cumsum = dst_cumsum;
-    dst_cumsum += dst_stride32_cumsum;
-    src_argb += src_stride_argb;
-  }
-  return 0;
-}
-
-// Blur ARGB image.
-// Caller should allocate CumulativeSum table of width * height * 16 bytes
-// aligned to 16 byte boundary. height can be radius * 2 + 2 to save memory
-// as the buffer is treated as circular.
-LIBYUV_API
-int ARGBBlur(const uint8* src_argb, int src_stride_argb,
-             uint8* dst_argb, int dst_stride_argb,
-             int32* dst_cumsum, int dst_stride32_cumsum,
-             int width, int height, int radius) {
-  int y;
-  void (*ComputeCumulativeSumRow)(const uint8 *row, int32 *cumsum,
-      const int32* previous_cumsum, int width) = ComputeCumulativeSumRow_C;
-  void (*CumulativeSumToAverageRow)(const int32* topleft, const int32* botleft,
-      int width, int area, uint8* dst, int count) = CumulativeSumToAverageRow_C;
-  int32* cumsum_bot_row;
-  int32* max_cumsum_bot_row;
-  int32* cumsum_top_row;
-
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  if (radius > height) {
-    radius = height;
-  }
-  if (radius > (width / 2 - 1)) {
-    radius = width / 2 - 1;
-  }
-  if (radius <= 0) {
-    return -1;
-  }
-#if defined(HAS_CUMULATIVESUMTOAVERAGEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    ComputeCumulativeSumRow = ComputeCumulativeSumRow_SSE2;
-    CumulativeSumToAverageRow = CumulativeSumToAverageRow_SSE2;
-  }
-#endif
-  // Compute enough CumulativeSum for first row to be blurred. After this
-  // one row of CumulativeSum is updated at a time.
-  ARGBComputeCumulativeSum(src_argb, src_stride_argb,
-                           dst_cumsum, dst_stride32_cumsum,
-                           width, radius);
-
-  src_argb = src_argb + radius * src_stride_argb;
-  cumsum_bot_row = &dst_cumsum[(radius - 1) * dst_stride32_cumsum];
-
-  max_cumsum_bot_row = &dst_cumsum[(radius * 2 + 2) * dst_stride32_cumsum];
-  cumsum_top_row = &dst_cumsum[0];
-
-  for (y = 0; y < height; ++y) {
-    int top_y = ((y - radius - 1) >= 0) ? (y - radius - 1) : 0;
-    int bot_y = ((y + radius) < height) ? (y + radius) : (height - 1);
-    int area = radius * (bot_y - top_y);
-    int boxwidth = radius * 4;
-    int x;
-    int n;
-
-    // Increment cumsum_top_row pointer with circular buffer wrap around.
-    if (top_y) {
-      cumsum_top_row += dst_stride32_cumsum;
-      if (cumsum_top_row >= max_cumsum_bot_row) {
-        cumsum_top_row = dst_cumsum;
-      }
-    }
-    // Increment cumsum_bot_row pointer with circular buffer wrap around and
-    // then fill in a row of CumulativeSum.
-    if ((y + radius) < height) {
-      const int32* prev_cumsum_bot_row = cumsum_bot_row;
-      cumsum_bot_row += dst_stride32_cumsum;
-      if (cumsum_bot_row >= max_cumsum_bot_row) {
-        cumsum_bot_row = dst_cumsum;
-      }
-      ComputeCumulativeSumRow(src_argb, cumsum_bot_row, prev_cumsum_bot_row,
-                              width);
-      src_argb += src_stride_argb;
-    }
-
-    // Left clipped.
-    for (x = 0; x < radius + 1; ++x) {
-      CumulativeSumToAverageRow(cumsum_top_row, cumsum_bot_row,
-                                boxwidth, area, &dst_argb[x * 4], 1);
-      area += (bot_y - top_y);
-      boxwidth += 4;
-    }
-
-    // Middle unclipped.
-    n = (width - 1) - radius - x + 1;
-    CumulativeSumToAverageRow(cumsum_top_row, cumsum_bot_row,
-                              boxwidth, area, &dst_argb[x * 4], n);
-
-    // Right clipped.
-    for (x += n; x <= width - 1; ++x) {
-      area -= (bot_y - top_y);
-      boxwidth -= 4;
-      CumulativeSumToAverageRow(cumsum_top_row + (x - radius - 1) * 4,
-                                cumsum_bot_row + (x - radius - 1) * 4,
-                                boxwidth, area, &dst_argb[x * 4], 1);
-    }
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Multiply ARGB image by a specified ARGB value.
-LIBYUV_API
-int ARGBShade(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height, uint32 value) {
-  int y;
-  void (*ARGBShadeRow)(const uint8* src_argb, uint8* dst_argb,
-                       int width, uint32 value) = ARGBShadeRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0 || value == 0u) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBSHADEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    ARGBShadeRow = ARGBShadeRow_SSE2;
-  }
-#elif defined(HAS_ARGBSHADEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBShadeRow = ARGBShadeRow_NEON;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBShadeRow(src_argb, dst_argb, width, value);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Interpolate 2 ARGB images by specified amount (0 to 255).
-LIBYUV_API
-int ARGBInterpolate(const uint8* src_argb0, int src_stride_argb0,
-                    const uint8* src_argb1, int src_stride_argb1,
-                    uint8* dst_argb, int dst_stride_argb,
-                    int width, int height, int interpolation) {
-  int y;
-  void (*InterpolateRow)(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride, int dst_width,
-                         int source_y_fraction) = InterpolateRow_C;
-  if (!src_argb0 || !src_argb1 || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_argb = dst_argb + (height - 1) * dst_stride_argb;
-    dst_stride_argb = -dst_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb0 == width * 4 &&
-      src_stride_argb1 == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb0 = src_stride_argb1 = dst_stride_argb = 0;
-  }
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_argb0, 16) && IS_ALIGNED(src_stride_argb0, 16) &&
-          IS_ALIGNED(src_argb1, 16) && IS_ALIGNED(src_stride_argb1, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb0, 16) && IS_ALIGNED(src_stride_argb0, 16) &&
-          IS_ALIGNED(src_argb1, 16) && IS_ALIGNED(src_stride_argb1, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 8) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(width, 8)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 4) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(width, 4)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROWS_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && width >= 1 &&
-      IS_ALIGNED(src_argb0, 4) && IS_ALIGNED(src_stride_argb0, 4) &&
-      IS_ALIGNED(src_argb1, 4) && IS_ALIGNED(src_stride_argb1, 4) &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride_argb, 4)) {
-    ScaleARGBFilterRows = InterpolateRow_MIPS_DSPR2;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    InterpolateRow(dst_argb, src_argb0, src_argb1 - src_argb0,
-                   width * 4, interpolation);
-    src_argb0 += src_stride_argb0;
-    src_argb1 += src_stride_argb1;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Shuffle ARGB channel order.  e.g. BGRA to ARGB.
-LIBYUV_API
-int ARGBShuffle(const uint8* src_bgra, int src_stride_bgra,
-                uint8* dst_argb, int dst_stride_argb,
-                const uint8* shuffler, int width, int height) {
-  int y;
-  void (*ARGBShuffleRow)(const uint8* src_bgra, uint8* dst_argb,
-                         const uint8* shuffler, int pix) = ARGBShuffleRow_C;
-  if (!src_bgra || !dst_argb ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_bgra = src_bgra + (height - 1) * src_stride_bgra;
-    src_stride_bgra = -src_stride_bgra;
-  }
-  // Coalesce rows.
-  if (src_stride_bgra == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_bgra = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBSHUFFLEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4) {
-    ARGBShuffleRow = ARGBShuffleRow_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBShuffleRow = ARGBShuffleRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBSHUFFLEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8) {
-    ARGBShuffleRow = ARGBShuffleRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBShuffleRow = ARGBShuffleRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_bgra, 16) && IS_ALIGNED(src_stride_bgra, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        ARGBShuffleRow = ARGBShuffleRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_ARGBSHUFFLEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && width >= 16) {
-    ARGBShuffleRow = ARGBShuffleRow_Any_AVX2;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBShuffleRow = ARGBShuffleRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBSHUFFLEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 4) {
-    ARGBShuffleRow = ARGBShuffleRow_Any_NEON;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBShuffleRow = ARGBShuffleRow_NEON;
-    }
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBShuffleRow(src_bgra, dst_argb, shuffler, width);
-    src_bgra += src_stride_bgra;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Sobel ARGB effect.
-static int ARGBSobelize(const uint8* src_argb, int src_stride_argb,
-                        uint8* dst_argb, int dst_stride_argb,
-                        int width, int height,
-                        void (*SobelRow)(const uint8* src_sobelx,
-                                         const uint8* src_sobely,
-                                         uint8* dst, int width)) {
-  int y;
-  void (*ARGBToBayerRow)(const uint8* src_argb, uint8* dst_bayer,
-                         uint32 selector, int pix) = ARGBToBayerGGRow_C;
-  void (*SobelYRow)(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width) = SobelYRow_C;
-  void (*SobelXRow)(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobely, int width) =
-      SobelXRow_C;
-  const int kEdge = 16;  // Extra pixels at start of row for extrude/align.
-  if (!src_argb  || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb  = src_argb  + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // ARGBToBayer used to select G channel from ARGB.
-#if defined(HAS_ARGBTOBAYERGGROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToBayerRow = ARGBToBayerGGRow_Any_SSE2;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerGGRow_SSE2;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOBAYERROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 8 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToBayerRow = ARGBToBayerRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerRow_SSSE3;
-    }
-  }
-#endif
-#if defined(HAS_ARGBTOBAYERGGROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToBayerRow = ARGBToBayerGGRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToBayerRow = ARGBToBayerGGRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_SOBELYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    SobelYRow = SobelYRow_SSE2;
-  }
-#endif
-#if defined(HAS_SOBELYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    SobelYRow = SobelYRow_NEON;
-  }
-#endif
-#if defined(HAS_SOBELXROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2)) {
-    SobelXRow = SobelXRow_SSE2;
-  }
-#endif
-#if defined(HAS_SOBELXROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    SobelXRow = SobelXRow_NEON;
-  }
-#endif
-  {
-    // 3 rows with edges before/after.
-    const int kRowSize = (width + kEdge + 15) & ~15;
-    align_buffer_64(rows, kRowSize * 2 + (kEdge + kRowSize * 3 + kEdge));
-    uint8* row_sobelx = rows;
-    uint8* row_sobely = rows + kRowSize;
-    uint8* row_y = rows + kRowSize * 2;
-
-    // Convert first row.
-    uint8* row_y0 = row_y + kEdge;
-    uint8* row_y1 = row_y0 + kRowSize;
-    uint8* row_y2 = row_y1 + kRowSize;
-    ARGBToBayerRow(src_argb, row_y0, 0x0d090501, width);
-    row_y0[-1] = row_y0[0];
-    memset(row_y0 + width, row_y0[width - 1], 16);  // Extrude 16 for valgrind.
-    ARGBToBayerRow(src_argb, row_y1, 0x0d090501, width);
-    row_y1[-1] = row_y1[0];
-    memset(row_y1 + width, row_y1[width - 1], 16);
-    memset(row_y2 + width, 0, 16);
-
-    for (y = 0; y < height; ++y) {
-      // Convert next row of ARGB to Y.
-      if (y < (height - 1)) {
-        src_argb += src_stride_argb;
-      }
-      ARGBToBayerRow(src_argb, row_y2, 0x0d090501, width);
-      row_y2[-1] = row_y2[0];
-      row_y2[width] = row_y2[width - 1];
-
-      SobelXRow(row_y0 - 1, row_y1 - 1, row_y2 - 1, row_sobelx, width);
-      SobelYRow(row_y0 - 1, row_y2 - 1, row_sobely, width);
-      SobelRow(row_sobelx, row_sobely, dst_argb, width);
-
-      // Cycle thru circular queue of 3 row_y buffers.
-      {
-        uint8* row_yt = row_y0;
-        row_y0 = row_y1;
-        row_y1 = row_y2;
-        row_y2 = row_yt;
-      }
-
-      dst_argb += dst_stride_argb;
-    }
-    free_aligned_buffer_64(rows);
-  }
-  return 0;
-}
-
-// Sobel ARGB effect.
-LIBYUV_API
-int ARGBSobel(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_argb, int dst_stride_argb,
-              int width, int height) {
-  void (*SobelRow)(const uint8* src_sobelx, const uint8* src_sobely,
-                   uint8* dst_argb, int width) = SobelRow_C;
-#if defined(HAS_SOBELROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    SobelRow = SobelRow_SSE2;
-  }
-#endif
-#if defined(HAS_SOBELROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    SobelRow = SobelRow_NEON;
-  }
-#endif
-  return ARGBSobelize(src_argb, src_stride_argb, dst_argb, dst_stride_argb,
-                      width, height, SobelRow);
-}
-
-// Sobel ARGB effect with planar output.
-LIBYUV_API
-int ARGBSobelToPlane(const uint8* src_argb, int src_stride_argb,
-                     uint8* dst_y, int dst_stride_y,
-                     int width, int height) {
-  void (*SobelToPlaneRow)(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_, int width) = SobelToPlaneRow_C;
-#if defined(HAS_SOBELTOPLANEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-    SobelToPlaneRow = SobelToPlaneRow_SSE2;
-  }
-#endif
-#if defined(HAS_SOBELTOPLANEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 16)) {
-    SobelToPlaneRow = SobelToPlaneRow_NEON;
-  }
-#endif
-  return ARGBSobelize(src_argb, src_stride_argb, dst_y, dst_stride_y,
-                      width, height, SobelToPlaneRow);
-}
-
-// SobelXY ARGB effect.
-// Similar to Sobel, but also stores Sobel X in R and Sobel Y in B.  G = Sobel.
-LIBYUV_API
-int ARGBSobelXY(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_argb, int dst_stride_argb,
-                int width, int height) {
-  void (*SobelXYRow)(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width) = SobelXYRow_C;
-#if defined(HAS_SOBELXYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-    SobelXYRow = SobelXYRow_SSE2;
-  }
-#endif
-#if defined(HAS_SOBELXYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    SobelXYRow = SobelXYRow_NEON;
-  }
-#endif
-  return ARGBSobelize(src_argb, src_stride_argb, dst_argb, dst_stride_argb,
-                      width, height, SobelXYRow);
-}
-
-// Apply a 4x4 polynomial to each ARGB pixel.
-LIBYUV_API
-int ARGBPolynomial(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb, int dst_stride_argb,
-                   const float* poly,
-                   int width, int height) {
-  int y;
-  void (*ARGBPolynomialRow)(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width) = ARGBPolynomialRow_C;
-  if (!src_argb || !dst_argb || !poly || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb  = src_argb  + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBPOLYNOMIALROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 2)) {
-    ARGBPolynomialRow = ARGBPolynomialRow_SSE2;
-  }
-#endif
-#if defined(HAS_ARGBPOLYNOMIALROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && TestCpuFlag(kCpuHasFMA3) &&
-      IS_ALIGNED(width, 2)) {
-    ARGBPolynomialRow = ARGBPolynomialRow_AVX2;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBPolynomialRow(src_argb, dst_argb, poly, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Apply a lumacolortable to each ARGB pixel.
-LIBYUV_API
-int ARGBLumaColorTable(const uint8* src_argb, int src_stride_argb,
-                       uint8* dst_argb, int dst_stride_argb,
-                       const uint8* luma,
-                       int width, int height) {
-  int y;
-  void (*ARGBLumaColorTableRow)(const uint8* src_argb, uint8* dst_argb,
-      int width, const uint8* luma, const uint32 lumacoeff) =
-      ARGBLumaColorTableRow_C;
-  if (!src_argb || !dst_argb || !luma || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb  = src_argb  + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBLUMACOLORTABLEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 4)) {
-    ARGBLumaColorTableRow = ARGBLumaColorTableRow_SSSE3;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBLumaColorTableRow(src_argb, dst_argb, width, luma, 0x00264b0f);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Copy Alpha from one ARGB image to another.
-LIBYUV_API
-int ARGBCopyAlpha(const uint8* src_argb, int src_stride_argb,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int width, int height) {
-  int y;
-  void (*ARGBCopyAlphaRow)(const uint8* src_argb, uint8* dst_argb, int width) =
-      ARGBCopyAlphaRow_C;
-  if (!src_argb || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  // Coalesce rows.
-  if (src_stride_argb == width * 4 &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_argb = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBCOPYALPHAROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16) &&
-      IS_ALIGNED(width, 8)) {
-    ARGBCopyAlphaRow = ARGBCopyAlphaRow_SSE2;
-  }
-#endif
-#if defined(HAS_ARGBCOPYALPHAROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 16)) {
-    ARGBCopyAlphaRow = ARGBCopyAlphaRow_AVX2;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBCopyAlphaRow(src_argb, dst_argb, width);
-    src_argb += src_stride_argb;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-// Copy a planar Y channel to the alpha channel of a destination ARGB image.
-LIBYUV_API
-int ARGBCopyYToAlpha(const uint8* src_y, int src_stride_y,
-                     uint8* dst_argb, int dst_stride_argb,
-                     int width, int height) {
-  int y;
-  void (*ARGBCopyYToAlphaRow)(const uint8* src_y, uint8* dst_argb, int width) =
-      ARGBCopyYToAlphaRow_C;
-  if (!src_y || !dst_argb || width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_stride_y = -src_stride_y;
-  }
-  // Coalesce rows.
-  if (src_stride_y == width &&
-      dst_stride_argb == width * 4) {
-    width *= height;
-    height = 1;
-    src_stride_y = dst_stride_argb = 0;
-  }
-#if defined(HAS_ARGBCOPYYTOALPHAROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) &&
-      IS_ALIGNED(src_y, 16) && IS_ALIGNED(src_stride_y, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16) &&
-      IS_ALIGNED(width, 8)) {
-    ARGBCopyYToAlphaRow = ARGBCopyYToAlphaRow_SSE2;
-  }
-#endif
-#if defined(HAS_ARGBCOPYYTOALPHAROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 16)) {
-    ARGBCopyYToAlphaRow = ARGBCopyYToAlphaRow_AVX2;
-  }
-#endif
-
-  for (y = 0; y < height; ++y) {
-    ARGBCopyYToAlphaRow(src_y, dst_argb, width);
-    src_y += src_stride_y;
-    dst_argb += dst_stride_argb;
-  }
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/rotate.cc b/drivers/theoraplayer/src/YUV/libyuv/src/rotate.cc
deleted file mode 100755
index b052ac1dc4..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/rotate.cc
+++ /dev/null
@@ -1,1301 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/rotate.h"
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/convert.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || defined(__x86_64__) || defined(__i386__))
-#if defined(__APPLE__) && defined(__i386__)
-#define DECLARE_FUNCTION(name)                                                 \
-    ".text                                     \n"                             \
-    ".private_extern _" #name "                \n"                             \
-    ".align 4,0x90                             \n"                             \
-"_" #name ":                                   \n"
-#elif defined(__MINGW32__) || defined(__CYGWIN__) && defined(__i386__)
-#define DECLARE_FUNCTION(name)                                                 \
-    ".text                                     \n"                             \
-    ".align 4,0x90                             \n"                             \
-"_" #name ":                                   \n"
-#else
-#define DECLARE_FUNCTION(name)                                                 \
-    ".text                                     \n"                             \
-    ".align 4,0x90                             \n"                             \
-#name ":                                       \n"
-#endif
-#endif
-
-#if !defined(LIBYUV_DISABLE_NEON) && !defined(__native_client__) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-#define HAS_MIRRORROW_NEON
-void MirrorRow_NEON(const uint8* src, uint8* dst, int width);
-#define HAS_MIRRORROW_UV_NEON
-void MirrorUVRow_NEON(const uint8* src, uint8* dst_a, uint8* dst_b, int width);
-#define HAS_TRANSPOSE_WX8_NEON
-void TransposeWx8_NEON(const uint8* src, int src_stride,
-                       uint8* dst, int dst_stride, int width);
-#define HAS_TRANSPOSE_UVWX8_NEON
-void TransposeUVWx8_NEON(const uint8* src, int src_stride,
-                         uint8* dst_a, int dst_stride_a,
-                         uint8* dst_b, int dst_stride_b,
-                         int width);
-#endif  // defined(__ARM_NEON__)
-
-#if !defined(LIBYUV_DISABLE_MIPS) && !defined(__native_client__) && \
-    defined(__mips__) && \
-    defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-#define HAS_TRANSPOSE_WX8_MIPS_DSPR2
-void TransposeWx8_MIPS_DSPR2(const uint8* src, int src_stride,
-                             uint8* dst, int dst_stride, int width);
-
-void TransposeWx8_FAST_MIPS_DSPR2(const uint8* src, int src_stride,
-                                  uint8* dst, int dst_stride, int width);
-#define HAS_TRANSPOSE_UVWx8_MIPS_DSPR2
-void TransposeUVWx8_MIPS_DSPR2(const uint8* src, int src_stride,
-                               uint8* dst_a, int dst_stride_a,
-                               uint8* dst_b, int dst_stride_b,
-                               int width);
-#endif  // defined(__mips__)
-
-#if !defined(LIBYUV_DISABLE_X86) && \
-    defined(_M_IX86) && defined(_MSC_VER)
-#define HAS_TRANSPOSE_WX8_SSSE3
-__declspec(naked) __declspec(align(16))
-static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
-                               uint8* dst, int dst_stride, int width) {
-  __asm {
-    push      edi
-    push      esi
-    push      ebp
-    mov       eax, [esp + 12 + 4]   // src
-    mov       edi, [esp + 12 + 8]   // src_stride
-    mov       edx, [esp + 12 + 12]  // dst
-    mov       esi, [esp + 12 + 16]  // dst_stride
-    mov       ecx, [esp + 12 + 20]  // width
-
-    // Read in the data from the source pointer.
-    // First round of bit swap.
-    align      4
- convertloop:
-    movq      xmm0, qword ptr [eax]
-    lea       ebp, [eax + 8]
-    movq      xmm1, qword ptr [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    punpcklbw xmm0, xmm1
-    movq      xmm2, qword ptr [eax]
-    movdqa    xmm1, xmm0
-    palignr   xmm1, xmm1, 8
-    movq      xmm3, qword ptr [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    punpcklbw xmm2, xmm3
-    movdqa    xmm3, xmm2
-    movq      xmm4, qword ptr [eax]
-    palignr   xmm3, xmm3, 8
-    movq      xmm5, qword ptr [eax + edi]
-    punpcklbw xmm4, xmm5
-    lea       eax, [eax + 2 * edi]
-    movdqa    xmm5, xmm4
-    movq      xmm6, qword ptr [eax]
-    palignr   xmm5, xmm5, 8
-    movq      xmm7, qword ptr [eax + edi]
-    punpcklbw xmm6, xmm7
-    mov       eax, ebp
-    movdqa    xmm7, xmm6
-    palignr   xmm7, xmm7, 8
-    // Second round of bit swap.
-    punpcklwd xmm0, xmm2
-    punpcklwd xmm1, xmm3
-    movdqa    xmm2, xmm0
-    movdqa    xmm3, xmm1
-    palignr   xmm2, xmm2, 8
-    palignr   xmm3, xmm3, 8
-    punpcklwd xmm4, xmm6
-    punpcklwd xmm5, xmm7
-    movdqa    xmm6, xmm4
-    movdqa    xmm7, xmm5
-    palignr   xmm6, xmm6, 8
-    palignr   xmm7, xmm7, 8
-    // Third round of bit swap.
-    // Write to the destination pointer.
-    punpckldq xmm0, xmm4
-    movq      qword ptr [edx], xmm0
-    movdqa    xmm4, xmm0
-    palignr   xmm4, xmm4, 8
-    movq      qword ptr [edx + esi], xmm4
-    lea       edx, [edx + 2 * esi]
-    punpckldq xmm2, xmm6
-    movdqa    xmm6, xmm2
-    palignr   xmm6, xmm6, 8
-    movq      qword ptr [edx], xmm2
-    punpckldq xmm1, xmm5
-    movq      qword ptr [edx + esi], xmm6
-    lea       edx, [edx + 2 * esi]
-    movdqa    xmm5, xmm1
-    movq      qword ptr [edx], xmm1
-    palignr   xmm5, xmm5, 8
-    punpckldq xmm3, xmm7
-    movq      qword ptr [edx + esi], xmm5
-    lea       edx, [edx + 2 * esi]
-    movq      qword ptr [edx], xmm3
-    movdqa    xmm7, xmm3
-    palignr   xmm7, xmm7, 8
-    sub       ecx, 8
-    movq      qword ptr [edx + esi], xmm7
-    lea       edx, [edx + 2 * esi]
-    jg        convertloop
-
-    pop       ebp
-    pop       esi
-    pop       edi
-    ret
-  }
-}
-
-#define HAS_TRANSPOSE_UVWX8_SSE2
-__declspec(naked) __declspec(align(16))
-static void TransposeUVWx8_SSE2(const uint8* src, int src_stride,
-                                uint8* dst_a, int dst_stride_a,
-                                uint8* dst_b, int dst_stride_b,
-                                int w) {
-  __asm {
-    push      ebx
-    push      esi
-    push      edi
-    push      ebp
-    mov       eax, [esp + 16 + 4]   // src
-    mov       edi, [esp + 16 + 8]   // src_stride
-    mov       edx, [esp + 16 + 12]  // dst_a
-    mov       esi, [esp + 16 + 16]  // dst_stride_a
-    mov       ebx, [esp + 16 + 20]  // dst_b
-    mov       ebp, [esp + 16 + 24]  // dst_stride_b
-    mov       ecx, esp
-    sub       esp, 4 + 16
-    and       esp, ~15
-    mov       [esp + 16], ecx
-    mov       ecx, [ecx + 16 + 28]  // w
-
-    align      4
- convertloop:
-    // Read in the data from the source pointer.
-    // First round of bit swap.
-    movdqa    xmm0, [eax]
-    movdqa    xmm1, [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    movdqa    xmm7, xmm0  // use xmm7 as temp register.
-    punpcklbw xmm0, xmm1
-    punpckhbw xmm7, xmm1
-    movdqa    xmm1, xmm7
-    movdqa    xmm2, [eax]
-    movdqa    xmm3, [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    movdqa    xmm7, xmm2
-    punpcklbw xmm2, xmm3
-    punpckhbw xmm7, xmm3
-    movdqa    xmm3, xmm7
-    movdqa    xmm4, [eax]
-    movdqa    xmm5, [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    movdqa    xmm7, xmm4
-    punpcklbw xmm4, xmm5
-    punpckhbw xmm7, xmm5
-    movdqa    xmm5, xmm7
-    movdqa    xmm6, [eax]
-    movdqa    xmm7, [eax + edi]
-    lea       eax, [eax + 2 * edi]
-    movdqa    [esp], xmm5  // backup xmm5
-    neg       edi
-    movdqa    xmm5, xmm6   // use xmm5 as temp register.
-    punpcklbw xmm6, xmm7
-    punpckhbw xmm5, xmm7
-    movdqa    xmm7, xmm5
-    lea       eax, [eax + 8 * edi + 16]
-    neg       edi
-    // Second round of bit swap.
-    movdqa    xmm5, xmm0
-    punpcklwd xmm0, xmm2
-    punpckhwd xmm5, xmm2
-    movdqa    xmm2, xmm5
-    movdqa    xmm5, xmm1
-    punpcklwd xmm1, xmm3
-    punpckhwd xmm5, xmm3
-    movdqa    xmm3, xmm5
-    movdqa    xmm5, xmm4
-    punpcklwd xmm4, xmm6
-    punpckhwd xmm5, xmm6
-    movdqa    xmm6, xmm5
-    movdqa    xmm5, [esp]  // restore xmm5
-    movdqa    [esp], xmm6  // backup xmm6
-    movdqa    xmm6, xmm5    // use xmm6 as temp register.
-    punpcklwd xmm5, xmm7
-    punpckhwd xmm6, xmm7
-    movdqa    xmm7, xmm6
-    // Third round of bit swap.
-    // Write to the destination pointer.
-    movdqa    xmm6, xmm0
-    punpckldq xmm0, xmm4
-    punpckhdq xmm6, xmm4
-    movdqa    xmm4, xmm6
-    movdqa    xmm6, [esp]  // restore xmm6
-    movlpd    qword ptr [edx], xmm0
-    movhpd    qword ptr [ebx], xmm0
-    movlpd    qword ptr [edx + esi], xmm4
-    lea       edx, [edx + 2 * esi]
-    movhpd    qword ptr [ebx + ebp], xmm4
-    lea       ebx, [ebx + 2 * ebp]
-    movdqa    xmm0, xmm2   // use xmm0 as the temp register.
-    punpckldq xmm2, xmm6
-    movlpd    qword ptr [edx], xmm2
-    movhpd    qword ptr [ebx], xmm2
-    punpckhdq xmm0, xmm6
-    movlpd    qword ptr [edx + esi], xmm0
-    lea       edx, [edx + 2 * esi]
-    movhpd    qword ptr [ebx + ebp], xmm0
-    lea       ebx, [ebx + 2 * ebp]
-    movdqa    xmm0, xmm1   // use xmm0 as the temp register.
-    punpckldq xmm1, xmm5
-    movlpd    qword ptr [edx], xmm1
-    movhpd    qword ptr [ebx], xmm1
-    punpckhdq xmm0, xmm5
-    movlpd    qword ptr [edx + esi], xmm0
-    lea       edx, [edx + 2 * esi]
-    movhpd    qword ptr [ebx + ebp], xmm0
-    lea       ebx, [ebx + 2 * ebp]
-    movdqa    xmm0, xmm3   // use xmm0 as the temp register.
-    punpckldq xmm3, xmm7
-    movlpd    qword ptr [edx], xmm3
-    movhpd    qword ptr [ebx], xmm3
-    punpckhdq xmm0, xmm7
-    sub       ecx, 8
-    movlpd    qword ptr [edx + esi], xmm0
-    lea       edx, [edx + 2 * esi]
-    movhpd    qword ptr [ebx + ebp], xmm0
-    lea       ebx, [ebx + 2 * ebp]
-    jg        convertloop
-
-    mov       esp, [esp + 16]
-    pop       ebp
-    pop       edi
-    pop       esi
-    pop       ebx
-    ret
-  }
-}
-#elif !defined(LIBYUV_DISABLE_X86) && \
-    (defined(__i386__) || (defined(__x86_64__) && !defined(__native_client__)))
-#define HAS_TRANSPOSE_WX8_SSSE3
-static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
-                               uint8* dst, int dst_stride, int width) {
-  asm volatile (
-    // Read in the data from the source pointer.
-    // First round of bit swap.
-    ".p2align  2                                 \n"
-  "1:                                            \n"
-    "movq       (%0),%%xmm0                      \n"
-    "movq       (%0,%3),%%xmm1                   \n"
-    "lea        (%0,%3,2),%0                     \n"
-    "punpcklbw  %%xmm1,%%xmm0                    \n"
-    "movq       (%0),%%xmm2                      \n"
-    "movdqa     %%xmm0,%%xmm1                    \n"
-    "palignr    $0x8,%%xmm1,%%xmm1               \n"
-    "movq       (%0,%3),%%xmm3                   \n"
-    "lea        (%0,%3,2),%0                     \n"
-    "punpcklbw  %%xmm3,%%xmm2                    \n"
-    "movdqa     %%xmm2,%%xmm3                    \n"
-    "movq       (%0),%%xmm4                      \n"
-    "palignr    $0x8,%%xmm3,%%xmm3               \n"
-    "movq       (%0,%3),%%xmm5                   \n"
-    "lea        (%0,%3,2),%0                     \n"
-    "punpcklbw  %%xmm5,%%xmm4                    \n"
-    "movdqa     %%xmm4,%%xmm5                    \n"
-    "movq       (%0),%%xmm6                      \n"
-    "palignr    $0x8,%%xmm5,%%xmm5               \n"
-    "movq       (%0,%3),%%xmm7                   \n"
-    "lea        (%0,%3,2),%0                     \n"
-    "punpcklbw  %%xmm7,%%xmm6                    \n"
-    "neg        %3                               \n"
-    "movdqa     %%xmm6,%%xmm7                    \n"
-    "lea        0x8(%0,%3,8),%0                  \n"
-    "palignr    $0x8,%%xmm7,%%xmm7               \n"
-    "neg        %3                               \n"
-     // Second round of bit swap.
-    "punpcklwd  %%xmm2,%%xmm0                    \n"
-    "punpcklwd  %%xmm3,%%xmm1                    \n"
-    "movdqa     %%xmm0,%%xmm2                    \n"
-    "movdqa     %%xmm1,%%xmm3                    \n"
-    "palignr    $0x8,%%xmm2,%%xmm2               \n"
-    "palignr    $0x8,%%xmm3,%%xmm3               \n"
-    "punpcklwd  %%xmm6,%%xmm4                    \n"
-    "punpcklwd  %%xmm7,%%xmm5                    \n"
-    "movdqa     %%xmm4,%%xmm6                    \n"
-    "movdqa     %%xmm5,%%xmm7                    \n"
-    "palignr    $0x8,%%xmm6,%%xmm6               \n"
-    "palignr    $0x8,%%xmm7,%%xmm7               \n"
-    // Third round of bit swap.
-    // Write to the destination pointer.
-    "punpckldq  %%xmm4,%%xmm0                    \n"
-    "movq       %%xmm0,(%1)                      \n"
-    "movdqa     %%xmm0,%%xmm4                    \n"
-    "palignr    $0x8,%%xmm4,%%xmm4               \n"
-    "movq       %%xmm4,(%1,%4)                   \n"
-    "lea        (%1,%4,2),%1                     \n"
-    "punpckldq  %%xmm6,%%xmm2                    \n"
-    "movdqa     %%xmm2,%%xmm6                    \n"
-    "movq       %%xmm2,(%1)                      \n"
-    "palignr    $0x8,%%xmm6,%%xmm6               \n"
-    "punpckldq  %%xmm5,%%xmm1                    \n"
-    "movq       %%xmm6,(%1,%4)                   \n"
-    "lea        (%1,%4,2),%1                     \n"
-    "movdqa     %%xmm1,%%xmm5                    \n"
-    "movq       %%xmm1,(%1)                      \n"
-    "palignr    $0x8,%%xmm5,%%xmm5               \n"
-    "movq       %%xmm5,(%1,%4)                   \n"
-    "lea        (%1,%4,2),%1                     \n"
-    "punpckldq  %%xmm7,%%xmm3                    \n"
-    "movq       %%xmm3,(%1)                      \n"
-    "movdqa     %%xmm3,%%xmm7                    \n"
-    "palignr    $0x8,%%xmm7,%%xmm7               \n"
-    "sub        $0x8,%2                          \n"
-    "movq       %%xmm7,(%1,%4)                   \n"
-    "lea        (%1,%4,2),%1                     \n"
-    "jg         1b                               \n"
-    : "+r"(src),    // %0
-      "+r"(dst),    // %1
-      "+r"(width)   // %2
-    : "r"((intptr_t)(src_stride)),  // %3
-      "r"((intptr_t)(dst_stride))   // %4
-    : "memory", "cc"
-  #if defined(__SSE2__)
-      , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-  #endif
-  );
-}
-
-#if !defined(LIBYUV_DISABLE_X86) && defined(__i386__)
-#define HAS_TRANSPOSE_UVWX8_SSE2
-extern "C" void TransposeUVWx8_SSE2(const uint8* src, int src_stride,
-                                    uint8* dst_a, int dst_stride_a,
-                                    uint8* dst_b, int dst_stride_b,
-                                    int w);
-  asm (
-    DECLARE_FUNCTION(TransposeUVWx8_SSE2)
-    "push   %ebx                               \n"
-    "push   %esi                               \n"
-    "push   %edi                               \n"
-    "push   %ebp                               \n"
-    "mov    0x14(%esp),%eax                    \n"
-    "mov    0x18(%esp),%edi                    \n"
-    "mov    0x1c(%esp),%edx                    \n"
-    "mov    0x20(%esp),%esi                    \n"
-    "mov    0x24(%esp),%ebx                    \n"
-    "mov    0x28(%esp),%ebp                    \n"
-    "mov    %esp,%ecx                          \n"
-    "sub    $0x14,%esp                         \n"
-    "and    $0xfffffff0,%esp                   \n"
-    "mov    %ecx,0x10(%esp)                    \n"
-    "mov    0x2c(%ecx),%ecx                    \n"
-
-"1:                                            \n"
-    "movdqa (%eax),%xmm0                       \n"
-    "movdqa (%eax,%edi,1),%xmm1                \n"
-    "lea    (%eax,%edi,2),%eax                 \n"
-    "movdqa %xmm0,%xmm7                        \n"
-    "punpcklbw %xmm1,%xmm0                     \n"
-    "punpckhbw %xmm1,%xmm7                     \n"
-    "movdqa %xmm7,%xmm1                        \n"
-    "movdqa (%eax),%xmm2                       \n"
-    "movdqa (%eax,%edi,1),%xmm3                \n"
-    "lea    (%eax,%edi,2),%eax                 \n"
-    "movdqa %xmm2,%xmm7                        \n"
-    "punpcklbw %xmm3,%xmm2                     \n"
-    "punpckhbw %xmm3,%xmm7                     \n"
-    "movdqa %xmm7,%xmm3                        \n"
-    "movdqa (%eax),%xmm4                       \n"
-    "movdqa (%eax,%edi,1),%xmm5                \n"
-    "lea    (%eax,%edi,2),%eax                 \n"
-    "movdqa %xmm4,%xmm7                        \n"
-    "punpcklbw %xmm5,%xmm4                     \n"
-    "punpckhbw %xmm5,%xmm7                     \n"
-    "movdqa %xmm7,%xmm5                        \n"
-    "movdqa (%eax),%xmm6                       \n"
-    "movdqa (%eax,%edi,1),%xmm7                \n"
-    "lea    (%eax,%edi,2),%eax                 \n"
-    "movdqa %xmm5,(%esp)                       \n"
-    "neg    %edi                               \n"
-    "movdqa %xmm6,%xmm5                        \n"
-    "punpcklbw %xmm7,%xmm6                     \n"
-    "punpckhbw %xmm7,%xmm5                     \n"
-    "movdqa %xmm5,%xmm7                        \n"
-    "lea    0x10(%eax,%edi,8),%eax             \n"
-    "neg    %edi                               \n"
-    "movdqa %xmm0,%xmm5                        \n"
-    "punpcklwd %xmm2,%xmm0                     \n"
-    "punpckhwd %xmm2,%xmm5                     \n"
-    "movdqa %xmm5,%xmm2                        \n"
-    "movdqa %xmm1,%xmm5                        \n"
-    "punpcklwd %xmm3,%xmm1                     \n"
-    "punpckhwd %xmm3,%xmm5                     \n"
-    "movdqa %xmm5,%xmm3                        \n"
-    "movdqa %xmm4,%xmm5                        \n"
-    "punpcklwd %xmm6,%xmm4                     \n"
-    "punpckhwd %xmm6,%xmm5                     \n"
-    "movdqa %xmm5,%xmm6                        \n"
-    "movdqa (%esp),%xmm5                       \n"
-    "movdqa %xmm6,(%esp)                       \n"
-    "movdqa %xmm5,%xmm6                        \n"
-    "punpcklwd %xmm7,%xmm5                     \n"
-    "punpckhwd %xmm7,%xmm6                     \n"
-    "movdqa %xmm6,%xmm7                        \n"
-    "movdqa %xmm0,%xmm6                        \n"
-    "punpckldq %xmm4,%xmm0                     \n"
-    "punpckhdq %xmm4,%xmm6                     \n"
-    "movdqa %xmm6,%xmm4                        \n"
-    "movdqa (%esp),%xmm6                       \n"
-    "movlpd %xmm0,(%edx)                       \n"
-    "movhpd %xmm0,(%ebx)                       \n"
-    "movlpd %xmm4,(%edx,%esi,1)                \n"
-    "lea    (%edx,%esi,2),%edx                 \n"
-    "movhpd %xmm4,(%ebx,%ebp,1)                \n"
-    "lea    (%ebx,%ebp,2),%ebx                 \n"
-    "movdqa %xmm2,%xmm0                        \n"
-    "punpckldq %xmm6,%xmm2                     \n"
-    "movlpd %xmm2,(%edx)                       \n"
-    "movhpd %xmm2,(%ebx)                       \n"
-    "punpckhdq %xmm6,%xmm0                     \n"
-    "movlpd %xmm0,(%edx,%esi,1)                \n"
-    "lea    (%edx,%esi,2),%edx                 \n"
-    "movhpd %xmm0,(%ebx,%ebp,1)                \n"
-    "lea    (%ebx,%ebp,2),%ebx                 \n"
-    "movdqa %xmm1,%xmm0                        \n"
-    "punpckldq %xmm5,%xmm1                     \n"
-    "movlpd %xmm1,(%edx)                       \n"
-    "movhpd %xmm1,(%ebx)                       \n"
-    "punpckhdq %xmm5,%xmm0                     \n"
-    "movlpd %xmm0,(%edx,%esi,1)                \n"
-    "lea    (%edx,%esi,2),%edx                 \n"
-    "movhpd %xmm0,(%ebx,%ebp,1)                \n"
-    "lea    (%ebx,%ebp,2),%ebx                 \n"
-    "movdqa %xmm3,%xmm0                        \n"
-    "punpckldq %xmm7,%xmm3                     \n"
-    "movlpd %xmm3,(%edx)                       \n"
-    "movhpd %xmm3,(%ebx)                       \n"
-    "punpckhdq %xmm7,%xmm0                     \n"
-    "sub    $0x8,%ecx                          \n"
-    "movlpd %xmm0,(%edx,%esi,1)                \n"
-    "lea    (%edx,%esi,2),%edx                 \n"
-    "movhpd %xmm0,(%ebx,%ebp,1)                \n"
-    "lea    (%ebx,%ebp,2),%ebx                 \n"
-    "jg     1b                                 \n"
-    "mov    0x10(%esp),%esp                    \n"
-    "pop    %ebp                               \n"
-    "pop    %edi                               \n"
-    "pop    %esi                               \n"
-    "pop    %ebx                               \n"
-#if defined(__native_client__)
-    "pop    %ecx                               \n"
-    "and    $0xffffffe0,%ecx                   \n"
-    "jmp    *%ecx                              \n"
-#else
-    "ret                                       \n"
-#endif
-);
-#elif !defined(LIBYUV_DISABLE_X86) && !defined(__native_client__) && \
-    defined(__x86_64__)
-// 64 bit version has enough registers to do 16x8 to 8x16 at a time.
-#define HAS_TRANSPOSE_WX8_FAST_SSSE3
-static void TransposeWx8_FAST_SSSE3(const uint8* src, int src_stride,
-                                    uint8* dst, int dst_stride, int width) {
-  asm volatile (
-  // Read in the data from the source pointer.
-  // First round of bit swap.
-  ".p2align  2                                 \n"
-"1:                                            \n"
-  "movdqa     (%0),%%xmm0                      \n"
-  "movdqa     (%0,%3),%%xmm1                   \n"
-  "lea        (%0,%3,2),%0                     \n"
-  "movdqa     %%xmm0,%%xmm8                    \n"
-  "punpcklbw  %%xmm1,%%xmm0                    \n"
-  "punpckhbw  %%xmm1,%%xmm8                    \n"
-  "movdqa     (%0),%%xmm2                      \n"
-  "movdqa     %%xmm0,%%xmm1                    \n"
-  "movdqa     %%xmm8,%%xmm9                    \n"
-  "palignr    $0x8,%%xmm1,%%xmm1               \n"
-  "palignr    $0x8,%%xmm9,%%xmm9               \n"
-  "movdqa     (%0,%3),%%xmm3                   \n"
-  "lea        (%0,%3,2),%0                     \n"
-  "movdqa     %%xmm2,%%xmm10                   \n"
-  "punpcklbw  %%xmm3,%%xmm2                    \n"
-  "punpckhbw  %%xmm3,%%xmm10                   \n"
-  "movdqa     %%xmm2,%%xmm3                    \n"
-  "movdqa     %%xmm10,%%xmm11                  \n"
-  "movdqa     (%0),%%xmm4                      \n"
-  "palignr    $0x8,%%xmm3,%%xmm3               \n"
-  "palignr    $0x8,%%xmm11,%%xmm11             \n"
-  "movdqa     (%0,%3),%%xmm5                   \n"
-  "lea        (%0,%3,2),%0                     \n"
-  "movdqa     %%xmm4,%%xmm12                   \n"
-  "punpcklbw  %%xmm5,%%xmm4                    \n"
-  "punpckhbw  %%xmm5,%%xmm12                   \n"
-  "movdqa     %%xmm4,%%xmm5                    \n"
-  "movdqa     %%xmm12,%%xmm13                  \n"
-  "movdqa     (%0),%%xmm6                      \n"
-  "palignr    $0x8,%%xmm5,%%xmm5               \n"
-  "palignr    $0x8,%%xmm13,%%xmm13             \n"
-  "movdqa     (%0,%3),%%xmm7                   \n"
-  "lea        (%0,%3,2),%0                     \n"
-  "movdqa     %%xmm6,%%xmm14                   \n"
-  "punpcklbw  %%xmm7,%%xmm6                    \n"
-  "punpckhbw  %%xmm7,%%xmm14                   \n"
-  "neg        %3                               \n"
-  "movdqa     %%xmm6,%%xmm7                    \n"
-  "movdqa     %%xmm14,%%xmm15                  \n"
-  "lea        0x10(%0,%3,8),%0                 \n"
-  "palignr    $0x8,%%xmm7,%%xmm7               \n"
-  "palignr    $0x8,%%xmm15,%%xmm15             \n"
-  "neg        %3                               \n"
-   // Second round of bit swap.
-  "punpcklwd  %%xmm2,%%xmm0                    \n"
-  "punpcklwd  %%xmm3,%%xmm1                    \n"
-  "movdqa     %%xmm0,%%xmm2                    \n"
-  "movdqa     %%xmm1,%%xmm3                    \n"
-  "palignr    $0x8,%%xmm2,%%xmm2               \n"
-  "palignr    $0x8,%%xmm3,%%xmm3               \n"
-  "punpcklwd  %%xmm6,%%xmm4                    \n"
-  "punpcklwd  %%xmm7,%%xmm5                    \n"
-  "movdqa     %%xmm4,%%xmm6                    \n"
-  "movdqa     %%xmm5,%%xmm7                    \n"
-  "palignr    $0x8,%%xmm6,%%xmm6               \n"
-  "palignr    $0x8,%%xmm7,%%xmm7               \n"
-  "punpcklwd  %%xmm10,%%xmm8                   \n"
-  "punpcklwd  %%xmm11,%%xmm9                   \n"
-  "movdqa     %%xmm8,%%xmm10                   \n"
-  "movdqa     %%xmm9,%%xmm11                   \n"
-  "palignr    $0x8,%%xmm10,%%xmm10             \n"
-  "palignr    $0x8,%%xmm11,%%xmm11             \n"
-  "punpcklwd  %%xmm14,%%xmm12                  \n"
-  "punpcklwd  %%xmm15,%%xmm13                  \n"
-  "movdqa     %%xmm12,%%xmm14                  \n"
-  "movdqa     %%xmm13,%%xmm15                  \n"
-  "palignr    $0x8,%%xmm14,%%xmm14             \n"
-  "palignr    $0x8,%%xmm15,%%xmm15             \n"
-  // Third round of bit swap.
-  // Write to the destination pointer.
-  "punpckldq  %%xmm4,%%xmm0                    \n"
-  "movq       %%xmm0,(%1)                      \n"
-  "movdqa     %%xmm0,%%xmm4                    \n"
-  "palignr    $0x8,%%xmm4,%%xmm4               \n"
-  "movq       %%xmm4,(%1,%4)                   \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "punpckldq  %%xmm6,%%xmm2                    \n"
-  "movdqa     %%xmm2,%%xmm6                    \n"
-  "movq       %%xmm2,(%1)                      \n"
-  "palignr    $0x8,%%xmm6,%%xmm6               \n"
-  "punpckldq  %%xmm5,%%xmm1                    \n"
-  "movq       %%xmm6,(%1,%4)                   \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "movdqa     %%xmm1,%%xmm5                    \n"
-  "movq       %%xmm1,(%1)                      \n"
-  "palignr    $0x8,%%xmm5,%%xmm5               \n"
-  "movq       %%xmm5,(%1,%4)                   \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "punpckldq  %%xmm7,%%xmm3                    \n"
-  "movq       %%xmm3,(%1)                      \n"
-  "movdqa     %%xmm3,%%xmm7                    \n"
-  "palignr    $0x8,%%xmm7,%%xmm7               \n"
-  "movq       %%xmm7,(%1,%4)                   \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "punpckldq  %%xmm12,%%xmm8                   \n"
-  "movq       %%xmm8,(%1)                      \n"
-  "movdqa     %%xmm8,%%xmm12                   \n"
-  "palignr    $0x8,%%xmm12,%%xmm12             \n"
-  "movq       %%xmm12,(%1,%4)                  \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "punpckldq  %%xmm14,%%xmm10                  \n"
-  "movdqa     %%xmm10,%%xmm14                  \n"
-  "movq       %%xmm10,(%1)                     \n"
-  "palignr    $0x8,%%xmm14,%%xmm14             \n"
-  "punpckldq  %%xmm13,%%xmm9                   \n"
-  "movq       %%xmm14,(%1,%4)                  \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "movdqa     %%xmm9,%%xmm13                   \n"
-  "movq       %%xmm9,(%1)                      \n"
-  "palignr    $0x8,%%xmm13,%%xmm13             \n"
-  "movq       %%xmm13,(%1,%4)                  \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "punpckldq  %%xmm15,%%xmm11                  \n"
-  "movq       %%xmm11,(%1)                     \n"
-  "movdqa     %%xmm11,%%xmm15                  \n"
-  "palignr    $0x8,%%xmm15,%%xmm15             \n"
-  "sub        $0x10,%2                         \n"
-  "movq       %%xmm15,(%1,%4)                  \n"
-  "lea        (%1,%4,2),%1                     \n"
-  "jg         1b                               \n"
-  : "+r"(src),    // %0
-    "+r"(dst),    // %1
-    "+r"(width)   // %2
-  : "r"((intptr_t)(src_stride)),  // %3
-    "r"((intptr_t)(dst_stride))   // %4
-  : "memory", "cc",
-    "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
-    "xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13",  "xmm14",  "xmm15"
-);
-}
-
-#define HAS_TRANSPOSE_UVWX8_SSE2
-static void TransposeUVWx8_SSE2(const uint8* src, int src_stride,
-                                uint8* dst_a, int dst_stride_a,
-                                uint8* dst_b, int dst_stride_b,
-                                int w) {
-  asm volatile (
-  // Read in the data from the source pointer.
-  // First round of bit swap.
-  ".p2align  2                                 \n"
-"1:                                            \n"
-  "movdqa     (%0),%%xmm0                      \n"
-  "movdqa     (%0,%4),%%xmm1                   \n"
-  "lea        (%0,%4,2),%0                     \n"
-  "movdqa     %%xmm0,%%xmm8                    \n"
-  "punpcklbw  %%xmm1,%%xmm0                    \n"
-  "punpckhbw  %%xmm1,%%xmm8                    \n"
-  "movdqa     %%xmm8,%%xmm1                    \n"
-  "movdqa     (%0),%%xmm2                      \n"
-  "movdqa     (%0,%4),%%xmm3                   \n"
-  "lea        (%0,%4,2),%0                     \n"
-  "movdqa     %%xmm2,%%xmm8                    \n"
-  "punpcklbw  %%xmm3,%%xmm2                    \n"
-  "punpckhbw  %%xmm3,%%xmm8                    \n"
-  "movdqa     %%xmm8,%%xmm3                    \n"
-  "movdqa     (%0),%%xmm4                      \n"
-  "movdqa     (%0,%4),%%xmm5                   \n"
-  "lea        (%0,%4,2),%0                     \n"
-  "movdqa     %%xmm4,%%xmm8                    \n"
-  "punpcklbw  %%xmm5,%%xmm4                    \n"
-  "punpckhbw  %%xmm5,%%xmm8                    \n"
-  "movdqa     %%xmm8,%%xmm5                    \n"
-  "movdqa     (%0),%%xmm6                      \n"
-  "movdqa     (%0,%4),%%xmm7                   \n"
-  "lea        (%0,%4,2),%0                     \n"
-  "movdqa     %%xmm6,%%xmm8                    \n"
-  "punpcklbw  %%xmm7,%%xmm6                    \n"
-  "neg        %4                               \n"
-  "lea        0x10(%0,%4,8),%0                 \n"
-  "punpckhbw  %%xmm7,%%xmm8                    \n"
-  "movdqa     %%xmm8,%%xmm7                    \n"
-  "neg        %4                               \n"
-   // Second round of bit swap.
-  "movdqa     %%xmm0,%%xmm8                    \n"
-  "movdqa     %%xmm1,%%xmm9                    \n"
-  "punpckhwd  %%xmm2,%%xmm8                    \n"
-  "punpckhwd  %%xmm3,%%xmm9                    \n"
-  "punpcklwd  %%xmm2,%%xmm0                    \n"
-  "punpcklwd  %%xmm3,%%xmm1                    \n"
-  "movdqa     %%xmm8,%%xmm2                    \n"
-  "movdqa     %%xmm9,%%xmm3                    \n"
-  "movdqa     %%xmm4,%%xmm8                    \n"
-  "movdqa     %%xmm5,%%xmm9                    \n"
-  "punpckhwd  %%xmm6,%%xmm8                    \n"
-  "punpckhwd  %%xmm7,%%xmm9                    \n"
-  "punpcklwd  %%xmm6,%%xmm4                    \n"
-  "punpcklwd  %%xmm7,%%xmm5                    \n"
-  "movdqa     %%xmm8,%%xmm6                    \n"
-  "movdqa     %%xmm9,%%xmm7                    \n"
-  // Third round of bit swap.
-  // Write to the destination pointer.
-  "movdqa     %%xmm0,%%xmm8                    \n"
-  "punpckldq  %%xmm4,%%xmm0                    \n"
-  "movlpd     %%xmm0,(%1)                      \n"  // Write back U channel
-  "movhpd     %%xmm0,(%2)                      \n"  // Write back V channel
-  "punpckhdq  %%xmm4,%%xmm8                    \n"
-  "movlpd     %%xmm8,(%1,%5)                   \n"
-  "lea        (%1,%5,2),%1                     \n"
-  "movhpd     %%xmm8,(%2,%6)                   \n"
-  "lea        (%2,%6,2),%2                     \n"
-  "movdqa     %%xmm2,%%xmm8                    \n"
-  "punpckldq  %%xmm6,%%xmm2                    \n"
-  "movlpd     %%xmm2,(%1)                      \n"
-  "movhpd     %%xmm2,(%2)                      \n"
-  "punpckhdq  %%xmm6,%%xmm8                    \n"
-  "movlpd     %%xmm8,(%1,%5)                   \n"
-  "lea        (%1,%5,2),%1                     \n"
-  "movhpd     %%xmm8,(%2,%6)                   \n"
-  "lea        (%2,%6,2),%2                     \n"
-  "movdqa     %%xmm1,%%xmm8                    \n"
-  "punpckldq  %%xmm5,%%xmm1                    \n"
-  "movlpd     %%xmm1,(%1)                      \n"
-  "movhpd     %%xmm1,(%2)                      \n"
-  "punpckhdq  %%xmm5,%%xmm8                    \n"
-  "movlpd     %%xmm8,(%1,%5)                   \n"
-  "lea        (%1,%5,2),%1                     \n"
-  "movhpd     %%xmm8,(%2,%6)                   \n"
-  "lea        (%2,%6,2),%2                     \n"
-  "movdqa     %%xmm3,%%xmm8                    \n"
-  "punpckldq  %%xmm7,%%xmm3                    \n"
-  "movlpd     %%xmm3,(%1)                      \n"
-  "movhpd     %%xmm3,(%2)                      \n"
-  "punpckhdq  %%xmm7,%%xmm8                    \n"
-  "sub        $0x8,%3                          \n"
-  "movlpd     %%xmm8,(%1,%5)                   \n"
-  "lea        (%1,%5,2),%1                     \n"
-  "movhpd     %%xmm8,(%2,%6)                   \n"
-  "lea        (%2,%6,2),%2                     \n"
-  "jg         1b                               \n"
-  : "+r"(src),    // %0
-    "+r"(dst_a),  // %1
-    "+r"(dst_b),  // %2
-    "+r"(w)   // %3
-  : "r"((intptr_t)(src_stride)),    // %4
-    "r"((intptr_t)(dst_stride_a)),  // %5
-    "r"((intptr_t)(dst_stride_b))   // %6
-  : "memory", "cc",
-    "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
-    "xmm8", "xmm9"
-);
-}
-#endif
-#endif
-
-static void TransposeWx8_C(const uint8* src, int src_stride,
-                           uint8* dst, int dst_stride,
-                           int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    dst[0] = src[0 * src_stride];
-    dst[1] = src[1 * src_stride];
-    dst[2] = src[2 * src_stride];
-    dst[3] = src[3 * src_stride];
-    dst[4] = src[4 * src_stride];
-    dst[5] = src[5 * src_stride];
-    dst[6] = src[6 * src_stride];
-    dst[7] = src[7 * src_stride];
-    ++src;
-    dst += dst_stride;
-  }
-}
-
-static void TransposeWxH_C(const uint8* src, int src_stride,
-                           uint8* dst, int dst_stride,
-                           int width, int height) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int j;
-    for (j = 0; j < height; ++j) {
-      dst[i * dst_stride + j] = src[j * src_stride + i];
-    }
-  }
-}
-
-LIBYUV_API
-void TransposePlane(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height) {
-  int i = height;
-  void (*TransposeWx8)(const uint8* src, int src_stride,
-                       uint8* dst, int dst_stride,
-                       int width) = TransposeWx8_C;
-#if defined(HAS_TRANSPOSE_WX8_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    TransposeWx8 = TransposeWx8_NEON;
-  }
-#endif
-#if defined(HAS_TRANSPOSE_WX8_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 8)) {
-    TransposeWx8 = TransposeWx8_SSSE3;
-  }
-#endif
-#if defined(HAS_TRANSPOSE_WX8_FAST_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) &&
-      IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16)) {
-    TransposeWx8 = TransposeWx8_FAST_SSSE3;
-  }
-#endif
-#if defined(HAS_TRANSPOSE_WX8_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2)) {
-    if (IS_ALIGNED(width, 4) &&
-        IS_ALIGNED(src, 4) && IS_ALIGNED(src_stride, 4)) {
-      TransposeWx8 = TransposeWx8_FAST_MIPS_DSPR2;
-    } else {
-      TransposeWx8 = TransposeWx8_MIPS_DSPR2;
-    }
-  }
-#endif
-
-  // Work across the source in 8x8 tiles
-  while (i >= 8) {
-    TransposeWx8(src, src_stride, dst, dst_stride, width);
-    src += 8 * src_stride;    // Go down 8 rows.
-    dst += 8;                 // Move over 8 columns.
-    i -= 8;
-  }
-
-  TransposeWxH_C(src, src_stride, dst, dst_stride, width, i);
-}
-
-LIBYUV_API
-void RotatePlane90(const uint8* src, int src_stride,
-                   uint8* dst, int dst_stride,
-                   int width, int height) {
-  // Rotate by 90 is a transpose with the source read
-  // from bottom to top. So set the source pointer to the end
-  // of the buffer and flip the sign of the source stride.
-  src += src_stride * (height - 1);
-  src_stride = -src_stride;
-  TransposePlane(src, src_stride, dst, dst_stride, width, height);
-}
-
-LIBYUV_API
-void RotatePlane270(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height) {
-  // Rotate by 270 is a transpose with the destination written
-  // from bottom to top. So set the destination pointer to the end
-  // of the buffer and flip the sign of the destination stride.
-  dst += dst_stride * (width - 1);
-  dst_stride = -dst_stride;
-  TransposePlane(src, src_stride, dst, dst_stride, width, height);
-}
-
-LIBYUV_API
-void RotatePlane180(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height) {
-  // Swap first and last row and mirror the content. Uses a temporary row.
-  align_buffer_64(row, width);
-  const uint8* src_bot = src + src_stride * (height - 1);
-  uint8* dst_bot = dst + dst_stride * (height - 1);
-  int half_height = (height + 1) >> 1;
-  int y;
-  void (*MirrorRow)(const uint8* src, uint8* dst, int width) = MirrorRow_C;
-  void (*CopyRow)(const uint8* src, uint8* dst, int width) = CopyRow_C;
-#if defined(HAS_MIRRORROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 16)) {
-    MirrorRow = MirrorRow_NEON;
-  }
-#endif
-#if defined(HAS_MIRRORROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    MirrorRow = MirrorRow_SSE2;
-  }
-#endif
-#if defined(HAS_MIRRORROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    MirrorRow = MirrorRow_SSSE3;
-  }
-#endif
-#if defined(HAS_MIRRORROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 32)) {
-    MirrorRow = MirrorRow_AVX2;
-  }
-#endif
-#if defined(HAS_MIRRORROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) &&
-      IS_ALIGNED(src, 4) && IS_ALIGNED(src_stride, 4) &&
-      IS_ALIGNED(dst, 4) && IS_ALIGNED(dst_stride, 4)) {
-    MirrorRow = MirrorRow_MIPS_DSPR2;
-  }
-#endif
-#if defined(HAS_COPYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 32)) {
-    CopyRow = CopyRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_X86)
-  if (TestCpuFlag(kCpuHasX86) && IS_ALIGNED(width, 4)) {
-    CopyRow = CopyRow_X86;
-  }
-#endif
-#if defined(HAS_COPYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width, 32) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    CopyRow = CopyRow_SSE2;
-  }
-#endif
-#if defined(HAS_COPYROW_ERMS)
-  if (TestCpuFlag(kCpuHasERMS)) {
-    CopyRow = CopyRow_ERMS;
-  }
-#endif
-#if defined(HAS_COPYROW_MIPS)
-  if (TestCpuFlag(kCpuHasMIPS)) {
-    CopyRow = CopyRow_MIPS;
-  }
-#endif
-
-  // Odd height will harmlessly mirror the middle row twice.
-  for (y = 0; y < half_height; ++y) {
-    MirrorRow(src, row, width);  // Mirror first row into a buffer
-    src += src_stride;
-    MirrorRow(src_bot, dst, width);  // Mirror last row into first row
-    dst += dst_stride;
-    CopyRow(row, dst_bot, width);  // Copy first mirrored row into last
-    src_bot -= src_stride;
-    dst_bot -= dst_stride;
-  }
-  free_aligned_buffer_64(row);
-}
-
-static void TransposeUVWx8_C(const uint8* src, int src_stride,
-                             uint8* dst_a, int dst_stride_a,
-                             uint8* dst_b, int dst_stride_b,
-                             int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    dst_a[0] = src[0 * src_stride + 0];
-    dst_b[0] = src[0 * src_stride + 1];
-    dst_a[1] = src[1 * src_stride + 0];
-    dst_b[1] = src[1 * src_stride + 1];
-    dst_a[2] = src[2 * src_stride + 0];
-    dst_b[2] = src[2 * src_stride + 1];
-    dst_a[3] = src[3 * src_stride + 0];
-    dst_b[3] = src[3 * src_stride + 1];
-    dst_a[4] = src[4 * src_stride + 0];
-    dst_b[4] = src[4 * src_stride + 1];
-    dst_a[5] = src[5 * src_stride + 0];
-    dst_b[5] = src[5 * src_stride + 1];
-    dst_a[6] = src[6 * src_stride + 0];
-    dst_b[6] = src[6 * src_stride + 1];
-    dst_a[7] = src[7 * src_stride + 0];
-    dst_b[7] = src[7 * src_stride + 1];
-    src += 2;
-    dst_a += dst_stride_a;
-    dst_b += dst_stride_b;
-  }
-}
-
-static void TransposeUVWxH_C(const uint8* src, int src_stride,
-                             uint8* dst_a, int dst_stride_a,
-                             uint8* dst_b, int dst_stride_b,
-                             int width, int height) {
-  int i;
-  for (i = 0; i < width * 2; i += 2) {
-    int j;
-    for (j = 0; j < height; ++j) {
-      dst_a[j + ((i >> 1) * dst_stride_a)] = src[i + (j * src_stride)];
-      dst_b[j + ((i >> 1) * dst_stride_b)] = src[i + (j * src_stride) + 1];
-    }
-  }
-}
-
-LIBYUV_API
-void TransposeUV(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height) {
-  int i = height;
-  void (*TransposeUVWx8)(const uint8* src, int src_stride,
-                         uint8* dst_a, int dst_stride_a,
-                         uint8* dst_b, int dst_stride_b,
-                         int width) = TransposeUVWx8_C;
-#if defined(HAS_TRANSPOSE_UVWX8_NEON)
-  if (TestCpuFlag(kCpuHasNEON)) {
-    TransposeUVWx8 = TransposeUVWx8_NEON;
-  }
-#elif defined(HAS_TRANSPOSE_UVWX8_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) &&
-      IS_ALIGNED(width, 8) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16)) {
-    TransposeUVWx8 = TransposeUVWx8_SSE2;
-  }
-#elif defined(HAS_TRANSPOSE_UVWx8_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(width, 2) &&
-      IS_ALIGNED(src, 4) && IS_ALIGNED(src_stride, 4)) {
-    TransposeUVWx8 = TransposeUVWx8_MIPS_DSPR2;
-  }
-#endif
-
-  // Work through the source in 8x8 tiles.
-  while (i >= 8) {
-    TransposeUVWx8(src, src_stride,
-                   dst_a, dst_stride_a,
-                   dst_b, dst_stride_b,
-                   width);
-    src += 8 * src_stride;    // Go down 8 rows.
-    dst_a += 8;               // Move over 8 columns.
-    dst_b += 8;               // Move over 8 columns.
-    i -= 8;
-  }
-
-  TransposeUVWxH_C(src, src_stride,
-                   dst_a, dst_stride_a,
-                   dst_b, dst_stride_b,
-                   width, i);
-}
-
-LIBYUV_API
-void RotateUV90(const uint8* src, int src_stride,
-                uint8* dst_a, int dst_stride_a,
-                uint8* dst_b, int dst_stride_b,
-                int width, int height) {
-  src += src_stride * (height - 1);
-  src_stride = -src_stride;
-
-  TransposeUV(src, src_stride,
-              dst_a, dst_stride_a,
-              dst_b, dst_stride_b,
-              width, height);
-}
-
-LIBYUV_API
-void RotateUV270(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height) {
-  dst_a += dst_stride_a * (width - 1);
-  dst_b += dst_stride_b * (width - 1);
-  dst_stride_a = -dst_stride_a;
-  dst_stride_b = -dst_stride_b;
-
-  TransposeUV(src, src_stride,
-              dst_a, dst_stride_a,
-              dst_b, dst_stride_b,
-              width, height);
-}
-
-// Rotate 180 is a horizontal and vertical flip.
-LIBYUV_API
-void RotateUV180(const uint8* src, int src_stride,
-                 uint8* dst_a, int dst_stride_a,
-                 uint8* dst_b, int dst_stride_b,
-                 int width, int height) {
-  int i;
-  void (*MirrorRowUV)(const uint8* src, uint8* dst_u, uint8* dst_v, int width) =
-      MirrorUVRow_C;
-#if defined(HAS_MIRRORUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    MirrorRowUV = MirrorUVRow_NEON;
-  }
-#elif defined(HAS_MIRRORROW_UV_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 16) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16)) {
-    MirrorRowUV = MirrorUVRow_SSSE3;
-  }
-#elif defined(HAS_MIRRORUVROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) &&
-      IS_ALIGNED(src, 4) && IS_ALIGNED(src_stride, 4)) {
-    MirrorRowUV = MirrorUVRow_MIPS_DSPR2;
-  }
-#endif
-
-  dst_a += dst_stride_a * (height - 1);
-  dst_b += dst_stride_b * (height - 1);
-
-  for (i = 0; i < height; ++i) {
-    MirrorRowUV(src, dst_a, dst_b, width);
-    src += src_stride;
-    dst_a -= dst_stride_a;
-    dst_b -= dst_stride_b;
-  }
-}
-
-LIBYUV_API
-int RotatePlane(const uint8* src, int src_stride,
-                uint8* dst, int dst_stride,
-                int width, int height,
-                enum RotationMode mode) {
-  if (!src || width <= 0 || height == 0 || !dst) {
-    return -1;
-  }
-
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src = src + (height - 1) * src_stride;
-    src_stride = -src_stride;
-  }
-
-  switch (mode) {
-    case kRotate0:
-      // copy frame
-      CopyPlane(src, src_stride,
-                dst, dst_stride,
-                width, height);
-      return 0;
-    case kRotate90:
-      RotatePlane90(src, src_stride,
-                    dst, dst_stride,
-                    width, height);
-      return 0;
-    case kRotate270:
-      RotatePlane270(src, src_stride,
-                     dst, dst_stride,
-                     width, height);
-      return 0;
-    case kRotate180:
-      RotatePlane180(src, src_stride,
-                     dst, dst_stride,
-                     width, height);
-      return 0;
-    default:
-      break;
-  }
-  return -1;
-}
-
-LIBYUV_API
-int I420Rotate(const uint8* src_y, int src_stride_y,
-               const uint8* src_u, int src_stride_u,
-               const uint8* src_v, int src_stride_v,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height,
-               enum RotationMode mode) {
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  if (!src_y || !src_u || !src_v || width <= 0 || height == 0 ||
-      !dst_y || !dst_u || !dst_v) {
-    return -1;
-  }
-
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    halfheight = (height + 1) >> 1;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_u = src_u + (halfheight - 1) * src_stride_u;
-    src_v = src_v + (halfheight - 1) * src_stride_v;
-    src_stride_y = -src_stride_y;
-    src_stride_u = -src_stride_u;
-    src_stride_v = -src_stride_v;
-  }
-
-  switch (mode) {
-    case kRotate0:
-      // copy frame
-      return I420Copy(src_y, src_stride_y,
-                      src_u, src_stride_u,
-                      src_v, src_stride_v,
-                      dst_y, dst_stride_y,
-                      dst_u, dst_stride_u,
-                      dst_v, dst_stride_v,
-                      width, height);
-    case kRotate90:
-      RotatePlane90(src_y, src_stride_y,
-                    dst_y, dst_stride_y,
-                    width, height);
-      RotatePlane90(src_u, src_stride_u,
-                    dst_u, dst_stride_u,
-                    halfwidth, halfheight);
-      RotatePlane90(src_v, src_stride_v,
-                    dst_v, dst_stride_v,
-                    halfwidth, halfheight);
-      return 0;
-    case kRotate270:
-      RotatePlane270(src_y, src_stride_y,
-                     dst_y, dst_stride_y,
-                     width, height);
-      RotatePlane270(src_u, src_stride_u,
-                     dst_u, dst_stride_u,
-                     halfwidth, halfheight);
-      RotatePlane270(src_v, src_stride_v,
-                     dst_v, dst_stride_v,
-                     halfwidth, halfheight);
-      return 0;
-    case kRotate180:
-      RotatePlane180(src_y, src_stride_y,
-                     dst_y, dst_stride_y,
-                     width, height);
-      RotatePlane180(src_u, src_stride_u,
-                     dst_u, dst_stride_u,
-                     halfwidth, halfheight);
-      RotatePlane180(src_v, src_stride_v,
-                     dst_v, dst_stride_v,
-                     halfwidth, halfheight);
-      return 0;
-    default:
-      break;
-  }
-  return -1;
-}
-
-LIBYUV_API
-int NV12ToI420Rotate(const uint8* src_y, int src_stride_y,
-                     const uint8* src_uv, int src_stride_uv,
-                     uint8* dst_y, int dst_stride_y,
-                     uint8* dst_u, int dst_stride_u,
-                     uint8* dst_v, int dst_stride_v,
-                     int width, int height,
-                     enum RotationMode mode) {
-  int halfwidth = (width + 1) >> 1;
-  int halfheight = (height + 1) >> 1;
-  if (!src_y || !src_uv || width <= 0 || height == 0 ||
-      !dst_y || !dst_u || !dst_v) {
-    return -1;
-  }
-
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    halfheight = (height + 1) >> 1;
-    src_y = src_y + (height - 1) * src_stride_y;
-    src_uv = src_uv + (halfheight - 1) * src_stride_uv;
-    src_stride_y = -src_stride_y;
-    src_stride_uv = -src_stride_uv;
-  }
-
-  switch (mode) {
-    case kRotate0:
-      // copy frame
-      return NV12ToI420(src_y, src_stride_y,
-                        src_uv, src_stride_uv,
-                        dst_y, dst_stride_y,
-                        dst_u, dst_stride_u,
-                        dst_v, dst_stride_v,
-                        width, height);
-    case kRotate90:
-      RotatePlane90(src_y, src_stride_y,
-                    dst_y, dst_stride_y,
-                    width, height);
-      RotateUV90(src_uv, src_stride_uv,
-                 dst_u, dst_stride_u,
-                 dst_v, dst_stride_v,
-                 halfwidth, halfheight);
-      return 0;
-    case kRotate270:
-      RotatePlane270(src_y, src_stride_y,
-                     dst_y, dst_stride_y,
-                     width, height);
-      RotateUV270(src_uv, src_stride_uv,
-                  dst_u, dst_stride_u,
-                  dst_v, dst_stride_v,
-                  halfwidth, halfheight);
-      return 0;
-    case kRotate180:
-      RotatePlane180(src_y, src_stride_y,
-                     dst_y, dst_stride_y,
-                     width, height);
-      RotateUV180(src_uv, src_stride_uv,
-                  dst_u, dst_stride_u,
-                  dst_v, dst_stride_v,
-                  halfwidth, halfheight);
-      return 0;
-    default:
-      break;
-  }
-  return -1;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_argb.cc b/drivers/theoraplayer/src/YUV/libyuv/src/rotate_argb.cc
deleted file mode 100755
index ab0f9ce070..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_argb.cc
+++ /dev/null
@@ -1,209 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/rotate.h"
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/convert.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// ARGBScale has a function to copy pixels to a row, striding each source
-// pixel by a constant.
-#if !defined(LIBYUV_DISABLE_X86) && \
-    (defined(_M_IX86) || \
-    (defined(__x86_64__) && !defined(__native_client__)) || defined(__i386__))
-#define HAS_SCALEARGBROWDOWNEVEN_SSE2
-void ScaleARGBRowDownEven_SSE2(const uint8* src_ptr, int src_stride,
-                               int src_stepx,
-                               uint8* dst_ptr, int dst_width);
-#endif
-#if !defined(LIBYUV_DISABLE_NEON) && !defined(__native_client__) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-#define HAS_SCALEARGBROWDOWNEVEN_NEON
-void ScaleARGBRowDownEven_NEON(const uint8* src_ptr, int src_stride,
-                               int src_stepx,
-                               uint8* dst_ptr, int dst_width);
-#endif
-
-void ScaleARGBRowDownEven_C(const uint8* src_ptr, int,
-                            int src_stepx,
-                            uint8* dst_ptr, int dst_width);
-
-static void ARGBTranspose(const uint8* src, int src_stride,
-                          uint8* dst, int dst_stride,
-                          int width, int height) {
-  int i;
-  int src_pixel_step = src_stride >> 2;
-  void (*ScaleARGBRowDownEven)(const uint8* src_ptr, int src_stride,
-      int src_step, uint8* dst_ptr, int dst_width) = ScaleARGBRowDownEven_C;
-#if defined(HAS_SCALEARGBROWDOWNEVEN_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(height, 4) &&  // Width of dest.
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    ScaleARGBRowDownEven = ScaleARGBRowDownEven_SSE2;
-  }
-#elif defined(HAS_SCALEARGBROWDOWNEVEN_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(height, 4) &&  // Width of dest.
-      IS_ALIGNED(src, 4)) {
-    ScaleARGBRowDownEven = ScaleARGBRowDownEven_NEON;
-  }
-#endif
-
-  for (i = 0; i < width; ++i) {  // column of source to row of dest.
-    ScaleARGBRowDownEven(src, 0, src_pixel_step, dst, height);
-    dst += dst_stride;
-    src += 4;
-  }
-}
-
-void ARGBRotate90(const uint8* src, int src_stride,
-                  uint8* dst, int dst_stride,
-                  int width, int height) {
-  // Rotate by 90 is a ARGBTranspose with the source read
-  // from bottom to top. So set the source pointer to the end
-  // of the buffer and flip the sign of the source stride.
-  src += src_stride * (height - 1);
-  src_stride = -src_stride;
-  ARGBTranspose(src, src_stride, dst, dst_stride, width, height);
-}
-
-void ARGBRotate270(const uint8* src, int src_stride,
-                    uint8* dst, int dst_stride,
-                    int width, int height) {
-  // Rotate by 270 is a ARGBTranspose with the destination written
-  // from bottom to top. So set the destination pointer to the end
-  // of the buffer and flip the sign of the destination stride.
-  dst += dst_stride * (width - 1);
-  dst_stride = -dst_stride;
-  ARGBTranspose(src, src_stride, dst, dst_stride, width, height);
-}
-
-void ARGBRotate180(const uint8* src, int src_stride,
-                   uint8* dst, int dst_stride,
-                   int width, int height) {
-  // Swap first and last row and mirror the content. Uses a temporary row.
-  align_buffer_64(row, width * 4);
-  const uint8* src_bot = src + src_stride * (height - 1);
-  uint8* dst_bot = dst + dst_stride * (height - 1);
-  int half_height = (height + 1) >> 1;
-  int y;
-  void (*ARGBMirrorRow)(const uint8* src, uint8* dst, int width) =
-      ARGBMirrorRow_C;
-  void (*CopyRow)(const uint8* src, uint8* dst, int width) = CopyRow_C;
-#if defined(HAS_ARGBMIRRORROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    ARGBMirrorRow = ARGBMirrorRow_SSSE3;
-  }
-#endif
-#if defined(HAS_ARGBMIRRORROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && IS_ALIGNED(width, 8)) {
-    ARGBMirrorRow = ARGBMirrorRow_AVX2;
-  }
-#endif
-#if defined(HAS_ARGBMIRRORROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 4)) {
-    ARGBMirrorRow = ARGBMirrorRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width * 4, 32)) {
-    CopyRow = CopyRow_NEON;
-  }
-#endif
-#if defined(HAS_COPYROW_X86)
-  if (TestCpuFlag(kCpuHasX86)) {
-    CopyRow = CopyRow_X86;
-  }
-#endif
-#if defined(HAS_COPYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(width * 4, 32) &&
-      IS_ALIGNED(src, 16) && IS_ALIGNED(src_stride, 16) &&
-      IS_ALIGNED(dst, 16) && IS_ALIGNED(dst_stride, 16)) {
-    CopyRow = CopyRow_SSE2;
-  }
-#endif
-#if defined(HAS_COPYROW_ERMS)
-  if (TestCpuFlag(kCpuHasERMS)) {
-    CopyRow = CopyRow_ERMS;
-  }
-#endif
-#if defined(HAS_COPYROW_MIPS)
-  if (TestCpuFlag(kCpuHasMIPS)) {
-    CopyRow = CopyRow_MIPS;
-  }
-#endif
-
-  // Odd height will harmlessly mirror the middle row twice.
-  for (y = 0; y < half_height; ++y) {
-    ARGBMirrorRow(src, row, width);  // Mirror first row into a buffer
-    ARGBMirrorRow(src_bot, dst, width);  // Mirror last row into first row
-    CopyRow(row, dst_bot, width * 4);  // Copy first mirrored row into last
-    src += src_stride;
-    dst += dst_stride;
-    src_bot -= src_stride;
-    dst_bot -= dst_stride;
-  }
-  free_aligned_buffer_64(row);
-}
-
-LIBYUV_API
-int ARGBRotate(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_argb, int dst_stride_argb,
-               int width, int height,
-               enum RotationMode mode) {
-  if (!src_argb || width <= 0 || height == 0 || !dst_argb) {
-    return -1;
-  }
-
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-
-  switch (mode) {
-    case kRotate0:
-      // copy frame
-      return ARGBCopy(src_argb, src_stride_argb,
-                      dst_argb, dst_stride_argb,
-                      width, height);
-    case kRotate90:
-      ARGBRotate90(src_argb, src_stride_argb,
-                   dst_argb, dst_stride_argb,
-                   width, height);
-      return 0;
-    case kRotate270:
-      ARGBRotate270(src_argb, src_stride_argb,
-                    dst_argb, dst_stride_argb,
-                    width, height);
-      return 0;
-    case kRotate180:
-      ARGBRotate180(src_argb, src_stride_argb,
-                    dst_argb, dst_stride_argb,
-                    width, height);
-      return 0;
-    default:
-      break;
-  }
-  return -1;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_mips.cc b/drivers/theoraplayer/src/YUV/libyuv/src/rotate_mips.cc
deleted file mode 100755
index 04d5a663f7..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_mips.cc
+++ /dev/null
@@ -1,486 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_MIPS) && \
-    defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-
-void TransposeWx8_MIPS_DSPR2(const uint8* src, int src_stride,
-                             uint8* dst, int dst_stride,
-                             int width) {
-   __asm__ __volatile__ (
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-      "sll              $t2, %[src_stride], 0x1          \n" // src_stride x 2
-      "sll              $t4, %[src_stride], 0x2          \n" // src_stride x 4
-      "sll              $t9, %[src_stride], 0x3          \n" // src_stride x 8
-      "addu             $t3, $t2, %[src_stride]          \n"
-      "addu             $t5, $t4, %[src_stride]          \n"
-      "addu             $t6, $t2, $t4                    \n"
-      "andi             $t0, %[dst], 0x3                 \n"
-      "andi             $t1, %[dst_stride], 0x3          \n"
-      "or               $t0, $t0, $t1                    \n"
-      "bnez             $t0, 11f                         \n"
-      " subu            $t7, $t9, %[src_stride]          \n"
-//dst + dst_stride word aligned
-    "1:                                                  \n"
-      "lbu              $t0, 0(%[src])                   \n"
-      "lbux             $t1, %[src_stride](%[src])       \n"
-      "lbux             $t8, $t2(%[src])                 \n"
-      "lbux             $t9, $t3(%[src])                 \n"
-      "sll              $t1, $t1, 16                     \n"
-      "sll              $t9, $t9, 16                     \n"
-      "or               $t0, $t0, $t1                    \n"
-      "or               $t8, $t8, $t9                    \n"
-      "precr.qb.ph      $s0, $t8, $t0                    \n"
-      "lbux             $t0, $t4(%[src])                 \n"
-      "lbux             $t1, $t5(%[src])                 \n"
-      "lbux             $t8, $t6(%[src])                 \n"
-      "lbux             $t9, $t7(%[src])                 \n"
-      "sll              $t1, $t1, 16                     \n"
-      "sll              $t9, $t9, 16                     \n"
-      "or               $t0, $t0, $t1                    \n"
-      "or               $t8, $t8, $t9                    \n"
-      "precr.qb.ph      $s1, $t8, $t0                    \n"
-      "sw               $s0, 0(%[dst])                   \n"
-      "addiu            %[width], -1                     \n"
-      "addiu            %[src], 1                        \n"
-      "sw               $s1, 4(%[dst])                   \n"
-      "bnez             %[width], 1b                     \n"
-      " addu            %[dst], %[dst], %[dst_stride]    \n"
-      "b                2f                               \n"
-//dst + dst_stride unaligned
-   "11:                                                  \n"
-      "lbu              $t0, 0(%[src])                   \n"
-      "lbux             $t1, %[src_stride](%[src])       \n"
-      "lbux             $t8, $t2(%[src])                 \n"
-      "lbux             $t9, $t3(%[src])                 \n"
-      "sll              $t1, $t1, 16                     \n"
-      "sll              $t9, $t9, 16                     \n"
-      "or               $t0, $t0, $t1                    \n"
-      "or               $t8, $t8, $t9                    \n"
-      "precr.qb.ph      $s0, $t8, $t0                    \n"
-      "lbux             $t0, $t4(%[src])                 \n"
-      "lbux             $t1, $t5(%[src])                 \n"
-      "lbux             $t8, $t6(%[src])                 \n"
-      "lbux             $t9, $t7(%[src])                 \n"
-      "sll              $t1, $t1, 16                     \n"
-      "sll              $t9, $t9, 16                     \n"
-      "or               $t0, $t0, $t1                    \n"
-      "or               $t8, $t8, $t9                    \n"
-      "precr.qb.ph      $s1, $t8, $t0                    \n"
-      "swr              $s0, 0(%[dst])                   \n"
-      "swl              $s0, 3(%[dst])                   \n"
-      "addiu            %[width], -1                     \n"
-      "addiu            %[src], 1                        \n"
-      "swr              $s1, 4(%[dst])                   \n"
-      "swl              $s1, 7(%[dst])                   \n"
-      "bnez             %[width], 11b                    \n"
-       "addu             %[dst], %[dst], %[dst_stride]   \n"
-    "2:                                                  \n"
-      ".set pop                                          \n"
-      :[src] "+r" (src),
-       [dst] "+r" (dst),
-       [width] "+r" (width)
-      :[src_stride] "r" (src_stride),
-       [dst_stride] "r" (dst_stride)
-      : "t0", "t1",  "t2", "t3", "t4", "t5",
-        "t6", "t7", "t8", "t9",
-        "s0", "s1"
-  );
-}
-
-void TransposeWx8_FAST_MIPS_DSPR2(const uint8* src, int src_stride,
-                                  uint8* dst, int dst_stride,
-                                  int width) {
-  __asm__ __volatile__ (
-      ".set noat                                         \n"
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-      "beqz             %[width], 2f                     \n"
-      " sll             $t2, %[src_stride], 0x1          \n"  // src_stride x 2
-      "sll              $t4, %[src_stride], 0x2          \n"  // src_stride x 4
-      "sll              $t9, %[src_stride], 0x3          \n"  // src_stride x 8
-      "addu             $t3, $t2, %[src_stride]          \n"
-      "addu             $t5, $t4, %[src_stride]          \n"
-      "addu             $t6, $t2, $t4                    \n"
-
-      "srl              $AT, %[width], 0x2               \n"
-      "andi             $t0, %[dst], 0x3                 \n"
-      "andi             $t1, %[dst_stride], 0x3          \n"
-      "or               $t0, $t0, $t1                    \n"
-      "bnez             $t0, 11f                         \n"
-      " subu            $t7, $t9, %[src_stride]          \n"
-//dst + dst_stride word aligned
-      "1:                                                \n"
-      "lw               $t0, 0(%[src])                   \n"
-      "lwx              $t1, %[src_stride](%[src])       \n"
-      "lwx              $t8, $t2(%[src])                 \n"
-      "lwx              $t9, $t3(%[src])                 \n"
-
-// t0 = | 30 | 20 | 10 | 00 |
-// t1 = | 31 | 21 | 11 | 01 |
-// t8 = | 32 | 22 | 12 | 02 |
-// t9 = | 33 | 23 | 13 | 03 |
-
-      "precr.qb.ph     $s0, $t1, $t0                     \n"
-      "precr.qb.ph     $s1, $t9, $t8                     \n"
-      "precrq.qb.ph    $s2, $t1, $t0                     \n"
-      "precrq.qb.ph    $s3, $t9, $t8                     \n"
-
-  // s0 = | 21 | 01 | 20 | 00 |
-  // s1 = | 23 | 03 | 22 | 02 |
-  // s2 = | 31 | 11 | 30 | 10 |
-  // s3 = | 33 | 13 | 32 | 12 |
-
-      "precr.qb.ph     $s4, $s1, $s0                     \n"
-      "precrq.qb.ph    $s5, $s1, $s0                     \n"
-      "precr.qb.ph     $s6, $s3, $s2                     \n"
-      "precrq.qb.ph    $s7, $s3, $s2                     \n"
-
-  // s4 = | 03 | 02 | 01 | 00 |
-  // s5 = | 23 | 22 | 21 | 20 |
-  // s6 = | 13 | 12 | 11 | 10 |
-  // s7 = | 33 | 32 | 31 | 30 |
-
-      "lwx              $t0, $t4(%[src])                 \n"
-      "lwx              $t1, $t5(%[src])                 \n"
-      "lwx              $t8, $t6(%[src])                 \n"
-      "lwx              $t9, $t7(%[src])                 \n"
-
-// t0 = | 34 | 24 | 14 | 04 |
-// t1 = | 35 | 25 | 15 | 05 |
-// t8 = | 36 | 26 | 16 | 06 |
-// t9 = | 37 | 27 | 17 | 07 |
-
-      "precr.qb.ph     $s0, $t1, $t0                     \n"
-      "precr.qb.ph     $s1, $t9, $t8                     \n"
-      "precrq.qb.ph    $s2, $t1, $t0                     \n"
-      "precrq.qb.ph    $s3, $t9, $t8                     \n"
-
-  // s0 = | 25 | 05 | 24 | 04 |
-  // s1 = | 27 | 07 | 26 | 06 |
-  // s2 = | 35 | 15 | 34 | 14 |
-  // s3 = | 37 | 17 | 36 | 16 |
-
-      "precr.qb.ph     $t0, $s1, $s0                     \n"
-      "precrq.qb.ph    $t1, $s1, $s0                     \n"
-      "precr.qb.ph     $t8, $s3, $s2                     \n"
-      "precrq.qb.ph    $t9, $s3, $s2                     \n"
-
-  // t0 = | 07 | 06 | 05 | 04 |
-  // t1 = | 27 | 26 | 25 | 24 |
-  // t8 = | 17 | 16 | 15 | 14 |
-  // t9 = | 37 | 36 | 35 | 34 |
-
-      "addu            $s0, %[dst], %[dst_stride]        \n"
-      "addu            $s1, $s0, %[dst_stride]           \n"
-      "addu            $s2, $s1, %[dst_stride]           \n"
-
-      "sw              $s4, 0(%[dst])                    \n"
-      "sw              $t0, 4(%[dst])                    \n"
-      "sw              $s6, 0($s0)                       \n"
-      "sw              $t8, 4($s0)                       \n"
-      "sw              $s5, 0($s1)                       \n"
-      "sw              $t1, 4($s1)                       \n"
-      "sw              $s7, 0($s2)                       \n"
-      "sw              $t9, 4($s2)                       \n"
-
-      "addiu            $AT, -1                          \n"
-      "addiu            %[src], 4                        \n"
-
-      "bnez             $AT, 1b                          \n"
-      " addu            %[dst], $s2, %[dst_stride]       \n"
-      "b                2f                               \n"
-//dst + dst_stride unaligned
-      "11:                                               \n"
-      "lw               $t0, 0(%[src])                   \n"
-      "lwx              $t1, %[src_stride](%[src])       \n"
-      "lwx              $t8, $t2(%[src])                 \n"
-      "lwx              $t9, $t3(%[src])                 \n"
-
-// t0 = | 30 | 20 | 10 | 00 |
-// t1 = | 31 | 21 | 11 | 01 |
-// t8 = | 32 | 22 | 12 | 02 |
-// t9 = | 33 | 23 | 13 | 03 |
-
-      "precr.qb.ph     $s0, $t1, $t0                     \n"
-      "precr.qb.ph     $s1, $t9, $t8                     \n"
-      "precrq.qb.ph    $s2, $t1, $t0                     \n"
-      "precrq.qb.ph    $s3, $t9, $t8                     \n"
-
-  // s0 = | 21 | 01 | 20 | 00 |
-  // s1 = | 23 | 03 | 22 | 02 |
-  // s2 = | 31 | 11 | 30 | 10 |
-  // s3 = | 33 | 13 | 32 | 12 |
-
-      "precr.qb.ph     $s4, $s1, $s0                     \n"
-      "precrq.qb.ph    $s5, $s1, $s0                     \n"
-      "precr.qb.ph     $s6, $s3, $s2                     \n"
-      "precrq.qb.ph    $s7, $s3, $s2                     \n"
-
-  // s4 = | 03 | 02 | 01 | 00 |
-  // s5 = | 23 | 22 | 21 | 20 |
-  // s6 = | 13 | 12 | 11 | 10 |
-  // s7 = | 33 | 32 | 31 | 30 |
-
-      "lwx              $t0, $t4(%[src])                 \n"
-      "lwx              $t1, $t5(%[src])                 \n"
-      "lwx              $t8, $t6(%[src])                 \n"
-      "lwx              $t9, $t7(%[src])                 \n"
-
-// t0 = | 34 | 24 | 14 | 04 |
-// t1 = | 35 | 25 | 15 | 05 |
-// t8 = | 36 | 26 | 16 | 06 |
-// t9 = | 37 | 27 | 17 | 07 |
-
-      "precr.qb.ph     $s0, $t1, $t0                     \n"
-      "precr.qb.ph     $s1, $t9, $t8                     \n"
-      "precrq.qb.ph    $s2, $t1, $t0                     \n"
-      "precrq.qb.ph    $s3, $t9, $t8                     \n"
-
-  // s0 = | 25 | 05 | 24 | 04 |
-  // s1 = | 27 | 07 | 26 | 06 |
-  // s2 = | 35 | 15 | 34 | 14 |
-  // s3 = | 37 | 17 | 36 | 16 |
-
-      "precr.qb.ph     $t0, $s1, $s0                     \n"
-      "precrq.qb.ph    $t1, $s1, $s0                     \n"
-      "precr.qb.ph     $t8, $s3, $s2                     \n"
-      "precrq.qb.ph    $t9, $s3, $s2                     \n"
-
-  // t0 = | 07 | 06 | 05 | 04 |
-  // t1 = | 27 | 26 | 25 | 24 |
-  // t8 = | 17 | 16 | 15 | 14 |
-  // t9 = | 37 | 36 | 35 | 34 |
-
-      "addu            $s0, %[dst], %[dst_stride]        \n"
-      "addu            $s1, $s0, %[dst_stride]           \n"
-      "addu            $s2, $s1, %[dst_stride]           \n"
-
-      "swr              $s4, 0(%[dst])                   \n"
-      "swl              $s4, 3(%[dst])                   \n"
-      "swr              $t0, 4(%[dst])                   \n"
-      "swl              $t0, 7(%[dst])                   \n"
-      "swr              $s6, 0($s0)                      \n"
-      "swl              $s6, 3($s0)                      \n"
-      "swr              $t8, 4($s0)                      \n"
-      "swl              $t8, 7($s0)                      \n"
-      "swr              $s5, 0($s1)                      \n"
-      "swl              $s5, 3($s1)                      \n"
-      "swr              $t1, 4($s1)                      \n"
-      "swl              $t1, 7($s1)                      \n"
-      "swr              $s7, 0($s2)                      \n"
-      "swl              $s7, 3($s2)                      \n"
-      "swr              $t9, 4($s2)                      \n"
-      "swl              $t9, 7($s2)                      \n"
-
-      "addiu            $AT, -1                          \n"
-      "addiu            %[src], 4                        \n"
-
-      "bnez             $AT, 11b                         \n"
-      " addu            %[dst], $s2, %[dst_stride]       \n"
-      "2:                                                \n"
-      ".set pop                                          \n"
-      ".set at                                           \n"
-      :[src] "+r" (src),
-       [dst] "+r" (dst),
-       [width] "+r" (width)
-      :[src_stride] "r" (src_stride),
-       [dst_stride] "r" (dst_stride)
-      : "t0", "t1",  "t2", "t3",  "t4", "t5",
-        "t6", "t7", "t8", "t9",
-        "s0", "s1", "s2", "s3", "s4",
-        "s5", "s6", "s7"
-  );
-}
-
-void TransposeUVWx8_MIPS_DSPR2(const uint8* src, int src_stride,
-                               uint8* dst_a, int dst_stride_a,
-                               uint8* dst_b, int dst_stride_b,
-                               int width) {
-  __asm__ __volatile__ (
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-      "beqz            %[width], 2f                      \n"
-      " sll            $t2, %[src_stride], 0x1           \n" // src_stride x 2
-      "sll             $t4, %[src_stride], 0x2           \n" // src_stride x 4
-      "sll             $t9, %[src_stride], 0x3           \n" // src_stride x 8
-      "addu            $t3, $t2, %[src_stride]           \n"
-      "addu            $t5, $t4, %[src_stride]           \n"
-      "addu            $t6, $t2, $t4                     \n"
-      "subu            $t7, $t9, %[src_stride]           \n"
-      "srl             $t1, %[width], 1                  \n"
-
-// check word aligment for dst_a, dst_b, dst_stride_a and dst_stride_b
-      "andi            $t0, %[dst_a], 0x3                \n"
-      "andi            $t8, %[dst_b], 0x3                \n"
-      "or              $t0, $t0, $t8                     \n"
-      "andi            $t8, %[dst_stride_a], 0x3         \n"
-      "andi            $s5, %[dst_stride_b], 0x3         \n"
-      "or              $t8, $t8, $s5                     \n"
-      "or              $t0, $t0, $t8                     \n"
-      "bnez            $t0, 11f                          \n"
-      " nop                                              \n"
-// dst + dst_stride word aligned (both, a & b dst addresses)
-    "1:                                                  \n"
-      "lw              $t0, 0(%[src])                    \n" // |B0|A0|b0|a0|
-      "lwx             $t8, %[src_stride](%[src])        \n" // |B1|A1|b1|a1|
-      "addu            $s5, %[dst_a], %[dst_stride_a]    \n"
-      "lwx             $t9, $t2(%[src])                  \n" // |B2|A2|b2|a2|
-      "lwx             $s0, $t3(%[src])                  \n" // |B3|A3|b3|a3|
-      "addu            $s6, %[dst_b], %[dst_stride_b]    \n"
-
-      "precrq.ph.w     $s1, $t8, $t0                     \n" // |B1|A1|B0|A0|
-      "precrq.ph.w     $s2, $s0, $t9                     \n" // |B3|A3|B2|A2|
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |A3|A2|A1|A0|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |B3|B2|B1|B0|
-
-      "sll             $t0, $t0, 16                      \n"
-      "packrl.ph       $s1, $t8, $t0                     \n" // |b1|a1|b0|a0|
-      "sll             $t9, $t9, 16                      \n"
-      "packrl.ph       $s2, $s0, $t9                     \n" // |b3|a3|b2|a2|
-
-      "sw              $s3, 0($s5)                       \n"
-      "sw              $s4, 0($s6)                       \n"
-
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |a3|a2|a1|a0|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |b3|b2|b1|b0|
-
-      "lwx             $t0, $t4(%[src])                  \n" // |B4|A4|b4|a4|
-      "lwx             $t8, $t5(%[src])                  \n" // |B5|A5|b5|a5|
-      "lwx             $t9, $t6(%[src])                  \n" // |B6|A6|b6|a6|
-      "lwx             $s0, $t7(%[src])                  \n" // |B7|A7|b7|a7|
-      "sw              $s3, 0(%[dst_a])                  \n"
-      "sw              $s4, 0(%[dst_b])                  \n"
-
-      "precrq.ph.w     $s1, $t8, $t0                     \n" // |B5|A5|B4|A4|
-      "precrq.ph.w     $s2, $s0, $t9                     \n" // |B6|A6|B7|A7|
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |A7|A6|A5|A4|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |B7|B6|B5|B4|
-
-      "sll             $t0, $t0, 16                      \n"
-      "packrl.ph       $s1, $t8, $t0                     \n" // |b5|a5|b4|a4|
-      "sll             $t9, $t9, 16                      \n"
-      "packrl.ph       $s2, $s0, $t9                     \n" // |b7|a7|b6|a6|
-      "sw              $s3, 4($s5)                       \n"
-      "sw              $s4, 4($s6)                       \n"
-
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |a7|a6|a5|a4|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |b7|b6|b5|b4|
-
-      "addiu           %[src], 4                         \n"
-      "addiu           $t1, -1                           \n"
-      "sll             $t0, %[dst_stride_a], 1           \n"
-      "sll             $t8, %[dst_stride_b], 1           \n"
-      "sw              $s3, 4(%[dst_a])                  \n"
-      "sw              $s4, 4(%[dst_b])                  \n"
-      "addu            %[dst_a], %[dst_a], $t0           \n"
-      "bnez            $t1, 1b                           \n"
-      " addu           %[dst_b], %[dst_b], $t8           \n"
-      "b               2f                                \n"
-      " nop                                              \n"
-
-// dst_a or dst_b or dst_stride_a or dst_stride_b not word aligned
-   "11:                                                  \n"
-      "lw              $t0, 0(%[src])                    \n" // |B0|A0|b0|a0|
-      "lwx             $t8, %[src_stride](%[src])        \n" // |B1|A1|b1|a1|
-      "addu            $s5, %[dst_a], %[dst_stride_a]    \n"
-      "lwx             $t9, $t2(%[src])                  \n" // |B2|A2|b2|a2|
-      "lwx             $s0, $t3(%[src])                  \n" // |B3|A3|b3|a3|
-      "addu            $s6, %[dst_b], %[dst_stride_b]    \n"
-
-      "precrq.ph.w     $s1, $t8, $t0                     \n" // |B1|A1|B0|A0|
-      "precrq.ph.w     $s2, $s0, $t9                     \n" // |B3|A3|B2|A2|
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |A3|A2|A1|A0|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |B3|B2|B1|B0|
-
-      "sll             $t0, $t0, 16                      \n"
-      "packrl.ph       $s1, $t8, $t0                     \n" // |b1|a1|b0|a0|
-      "sll             $t9, $t9, 16                      \n"
-      "packrl.ph       $s2, $s0, $t9                     \n" // |b3|a3|b2|a2|
-
-      "swr             $s3, 0($s5)                       \n"
-      "swl             $s3, 3($s5)                       \n"
-      "swr             $s4, 0($s6)                       \n"
-      "swl             $s4, 3($s6)                       \n"
-
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |a3|a2|a1|a0|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |b3|b2|b1|b0|
-
-      "lwx             $t0, $t4(%[src])                  \n" // |B4|A4|b4|a4|
-      "lwx             $t8, $t5(%[src])                  \n" // |B5|A5|b5|a5|
-      "lwx             $t9, $t6(%[src])                  \n" // |B6|A6|b6|a6|
-      "lwx             $s0, $t7(%[src])                  \n" // |B7|A7|b7|a7|
-      "swr             $s3, 0(%[dst_a])                  \n"
-      "swl             $s3, 3(%[dst_a])                  \n"
-      "swr             $s4, 0(%[dst_b])                  \n"
-      "swl             $s4, 3(%[dst_b])                  \n"
-
-      "precrq.ph.w     $s1, $t8, $t0                     \n" // |B5|A5|B4|A4|
-      "precrq.ph.w     $s2, $s0, $t9                     \n" // |B6|A6|B7|A7|
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |A7|A6|A5|A4|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |B7|B6|B5|B4|
-
-      "sll             $t0, $t0, 16                      \n"
-      "packrl.ph       $s1, $t8, $t0                     \n" // |b5|a5|b4|a4|
-      "sll             $t9, $t9, 16                      \n"
-      "packrl.ph       $s2, $s0, $t9                     \n" // |b7|a7|b6|a6|
-
-      "swr             $s3, 4($s5)                       \n"
-      "swl             $s3, 7($s5)                       \n"
-      "swr             $s4, 4($s6)                       \n"
-      "swl             $s4, 7($s6)                       \n"
-
-      "precr.qb.ph     $s3, $s2, $s1                     \n" // |a7|a6|a5|a4|
-      "precrq.qb.ph    $s4, $s2, $s1                     \n" // |b7|b6|b5|b4|
-
-      "addiu           %[src], 4                         \n"
-      "addiu           $t1, -1                           \n"
-      "sll             $t0, %[dst_stride_a], 1           \n"
-      "sll             $t8, %[dst_stride_b], 1           \n"
-      "swr             $s3, 4(%[dst_a])                  \n"
-      "swl             $s3, 7(%[dst_a])                  \n"
-      "swr             $s4, 4(%[dst_b])                  \n"
-      "swl             $s4, 7(%[dst_b])                  \n"
-      "addu            %[dst_a], %[dst_a], $t0           \n"
-      "bnez            $t1, 11b                          \n"
-      " addu           %[dst_b], %[dst_b], $t8           \n"
-
-      "2:                                                \n"
-      ".set pop                                          \n"
-      : [src] "+r" (src),
-        [dst_a] "+r" (dst_a),
-        [dst_b] "+r" (dst_b),
-        [width] "+r" (width),
-        [src_stride] "+r" (src_stride)
-      : [dst_stride_a] "r" (dst_stride_a),
-        [dst_stride_b] "r" (dst_stride_b)
-      : "t0", "t1",  "t2", "t3",  "t4", "t5",
-        "t6", "t7", "t8", "t9",
-        "s0", "s1", "s2", "s3",
-        "s4", "s5", "s6"
-  );
-}
-
-#endif  // defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_neon.cc b/drivers/theoraplayer/src/YUV/libyuv/src/rotate_neon.cc
deleted file mode 100755
index 274c4109cd..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/rotate_neon.cc
+++ /dev/null
@@ -1,412 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#if !defined(LIBYUV_DISABLE_NEON) && defined(__ARM_NEON__)
-static uvec8 kVTbl4x4Transpose =
-  { 0,  4,  8, 12,  1,  5,  9, 13,  2,  6, 10, 14,  3,  7, 11, 15 };
-
-void TransposeWx8_NEON(const uint8* src, int src_stride,
-                       uint8* dst, int dst_stride,
-                       int width) {
-  const uint8* src_temp = NULL;
-  asm volatile (
-    // loops are on blocks of 8. loop will stop when
-    // counter gets to or below 0. starting the counter
-    // at w-8 allow for this
-#ifdef _ANDROID
-				".fpu neon\n"
-#endif
-    "sub         %5, #8                        \n"
-
-    // handle 8x8 blocks. this should be the majority of the plane
-    ".p2align  2                               \n"
-    "1:                                        \n"
-      "mov         %0, %1                      \n"
-
-      "vld1.8      {d0}, [%0], %2              \n"
-      "vld1.8      {d1}, [%0], %2              \n"
-      "vld1.8      {d2}, [%0], %2              \n"
-      "vld1.8      {d3}, [%0], %2              \n"
-      "vld1.8      {d4}, [%0], %2              \n"
-      "vld1.8      {d5}, [%0], %2              \n"
-      "vld1.8      {d6}, [%0], %2              \n"
-      "vld1.8      {d7}, [%0]                  \n"
-
-      "vtrn.8      d1, d0                      \n"
-      "vtrn.8      d3, d2                      \n"
-      "vtrn.8      d5, d4                      \n"
-      "vtrn.8      d7, d6                      \n"
-
-      "vtrn.16     d1, d3                      \n"
-      "vtrn.16     d0, d2                      \n"
-      "vtrn.16     d5, d7                      \n"
-      "vtrn.16     d4, d6                      \n"
-
-      "vtrn.32     d1, d5                      \n"
-      "vtrn.32     d0, d4                      \n"
-      "vtrn.32     d3, d7                      \n"
-      "vtrn.32     d2, d6                      \n"
-
-      "vrev16.8    q0, q0                      \n"
-      "vrev16.8    q1, q1                      \n"
-      "vrev16.8    q2, q2                      \n"
-      "vrev16.8    q3, q3                      \n"
-
-      "mov         %0, %3                      \n"
-
-      "vst1.8      {d1}, [%0], %4              \n"
-      "vst1.8      {d0}, [%0], %4              \n"
-      "vst1.8      {d3}, [%0], %4              \n"
-      "vst1.8      {d2}, [%0], %4              \n"
-      "vst1.8      {d5}, [%0], %4              \n"
-      "vst1.8      {d4}, [%0], %4              \n"
-      "vst1.8      {d7}, [%0], %4              \n"
-      "vst1.8      {d6}, [%0]                  \n"
-
-      "add         %1, #8                      \n"  // src += 8
-      "add         %3, %3, %4, lsl #3          \n"  // dst += 8 * dst_stride
-      "subs        %5,  #8                     \n"  // w   -= 8
-      "bge         1b                          \n"
-
-    // add 8 back to counter. if the result is 0 there are
-    // no residuals.
-    "adds        %5, #8                        \n"
-    "beq         4f                            \n"
-
-    // some residual, so between 1 and 7 lines left to transpose
-    "cmp         %5, #2                        \n"
-    "blt         3f                            \n"
-
-    "cmp         %5, #4                        \n"
-    "blt         2f                            \n"
-
-    // 4x8 block
-    "mov         %0, %1                        \n"
-    "vld1.32     {d0[0]}, [%0], %2             \n"
-    "vld1.32     {d0[1]}, [%0], %2             \n"
-    "vld1.32     {d1[0]}, [%0], %2             \n"
-    "vld1.32     {d1[1]}, [%0], %2             \n"
-    "vld1.32     {d2[0]}, [%0], %2             \n"
-    "vld1.32     {d2[1]}, [%0], %2             \n"
-    "vld1.32     {d3[0]}, [%0], %2             \n"
-    "vld1.32     {d3[1]}, [%0]                 \n"
-
-    "mov         %0, %3                        \n"
-
-    "vld1.8      {q3}, [%6]                    \n"
-
-    "vtbl.8      d4, {d0, d1}, d6              \n"
-    "vtbl.8      d5, {d0, d1}, d7              \n"
-    "vtbl.8      d0, {d2, d3}, d6              \n"
-    "vtbl.8      d1, {d2, d3}, d7              \n"
-
-    // TODO(frkoenig): Rework shuffle above to
-    // write out with 4 instead of 8 writes.
-    "vst1.32     {d4[0]}, [%0], %4             \n"
-    "vst1.32     {d4[1]}, [%0], %4             \n"
-    "vst1.32     {d5[0]}, [%0], %4             \n"
-    "vst1.32     {d5[1]}, [%0]                 \n"
-
-    "add         %0, %3, #4                    \n"
-    "vst1.32     {d0[0]}, [%0], %4             \n"
-    "vst1.32     {d0[1]}, [%0], %4             \n"
-    "vst1.32     {d1[0]}, [%0], %4             \n"
-    "vst1.32     {d1[1]}, [%0]                 \n"
-
-    "add         %1, #4                        \n"  // src += 4
-    "add         %3, %3, %4, lsl #2            \n"  // dst += 4 * dst_stride
-    "subs        %5,  #4                       \n"  // w   -= 4
-    "beq         4f                            \n"
-
-    // some residual, check to see if it includes a 2x8 block,
-    // or less
-    "cmp         %5, #2                        \n"
-    "blt         3f                            \n"
-
-    // 2x8 block
-    "2:                                        \n"
-    "mov         %0, %1                        \n"
-    "vld1.16     {d0[0]}, [%0], %2             \n"
-    "vld1.16     {d1[0]}, [%0], %2             \n"
-    "vld1.16     {d0[1]}, [%0], %2             \n"
-    "vld1.16     {d1[1]}, [%0], %2             \n"
-    "vld1.16     {d0[2]}, [%0], %2             \n"
-    "vld1.16     {d1[2]}, [%0], %2             \n"
-    "vld1.16     {d0[3]}, [%0], %2             \n"
-    "vld1.16     {d1[3]}, [%0]                 \n"
-
-    "vtrn.8      d0, d1                        \n"
-
-    "mov         %0, %3                        \n"
-
-    "vst1.64     {d0}, [%0], %4                \n"
-    "vst1.64     {d1}, [%0]                    \n"
-
-    "add         %1, #2                        \n"  // src += 2
-    "add         %3, %3, %4, lsl #1            \n"  // dst += 2 * dst_stride
-    "subs        %5,  #2                       \n"  // w   -= 2
-    "beq         4f                            \n"
-
-    // 1x8 block
-    "3:                                        \n"
-    "vld1.8      {d0[0]}, [%1], %2             \n"
-    "vld1.8      {d0[1]}, [%1], %2             \n"
-    "vld1.8      {d0[2]}, [%1], %2             \n"
-    "vld1.8      {d0[3]}, [%1], %2             \n"
-    "vld1.8      {d0[4]}, [%1], %2             \n"
-    "vld1.8      {d0[5]}, [%1], %2             \n"
-    "vld1.8      {d0[6]}, [%1], %2             \n"
-    "vld1.8      {d0[7]}, [%1]                 \n"
-
-    "vst1.64     {d0}, [%3]                    \n"
-
-    "4:                                        \n"
-
-    : "+r"(src_temp),          // %0
-      "+r"(src),               // %1
-      "+r"(src_stride),        // %2
-      "+r"(dst),               // %3
-      "+r"(dst_stride),        // %4
-      "+r"(width)              // %5
-    : "r"(&kVTbl4x4Transpose)  // %6
-    : "memory", "cc", "q0", "q1", "q2", "q3"
-  );
-}
-
-static uvec8 kVTbl4x4TransposeDi =
-  { 0,  8,  1,  9,  2, 10,  3, 11,  4, 12,  5, 13,  6, 14,  7, 15 };
-
-void TransposeUVWx8_NEON(const uint8* src, int src_stride,
-                         uint8* dst_a, int dst_stride_a,
-                         uint8* dst_b, int dst_stride_b,
-                         int width) {
-  const uint8* src_temp = NULL;
-  asm volatile (
-    // loops are on blocks of 8. loop will stop when
-    // counter gets to or below 0. starting the counter
-    // at w-8 allow for this
-    "sub         %7, #8                        \n"
-
-    // handle 8x8 blocks. this should be the majority of the plane
-    ".p2align  2                               \n"
-    "1:                                        \n"
-      "mov         %0, %1                      \n"
-
-      "vld2.8      {d0,  d1},  [%0], %2        \n"
-      "vld2.8      {d2,  d3},  [%0], %2        \n"
-      "vld2.8      {d4,  d5},  [%0], %2        \n"
-      "vld2.8      {d6,  d7},  [%0], %2        \n"
-      "vld2.8      {d16, d17}, [%0], %2        \n"
-      "vld2.8      {d18, d19}, [%0], %2        \n"
-      "vld2.8      {d20, d21}, [%0], %2        \n"
-      "vld2.8      {d22, d23}, [%0]            \n"
-
-      "vtrn.8      q1, q0                      \n"
-      "vtrn.8      q3, q2                      \n"
-      "vtrn.8      q9, q8                      \n"
-      "vtrn.8      q11, q10                    \n"
-
-      "vtrn.16     q1, q3                      \n"
-      "vtrn.16     q0, q2                      \n"
-      "vtrn.16     q9, q11                     \n"
-      "vtrn.16     q8, q10                     \n"
-
-      "vtrn.32     q1, q9                      \n"
-      "vtrn.32     q0, q8                      \n"
-      "vtrn.32     q3, q11                     \n"
-      "vtrn.32     q2, q10                     \n"
-
-      "vrev16.8    q0, q0                      \n"
-      "vrev16.8    q1, q1                      \n"
-      "vrev16.8    q2, q2                      \n"
-      "vrev16.8    q3, q3                      \n"
-      "vrev16.8    q8, q8                      \n"
-      "vrev16.8    q9, q9                      \n"
-      "vrev16.8    q10, q10                    \n"
-      "vrev16.8    q11, q11                    \n"
-
-      "mov         %0, %3                      \n"
-
-      "vst1.8      {d2},  [%0], %4             \n"
-      "vst1.8      {d0},  [%0], %4             \n"
-      "vst1.8      {d6},  [%0], %4             \n"
-      "vst1.8      {d4},  [%0], %4             \n"
-      "vst1.8      {d18}, [%0], %4             \n"
-      "vst1.8      {d16}, [%0], %4             \n"
-      "vst1.8      {d22}, [%0], %4             \n"
-      "vst1.8      {d20}, [%0]                 \n"
-
-      "mov         %0, %5                      \n"
-
-      "vst1.8      {d3},  [%0], %6             \n"
-      "vst1.8      {d1},  [%0], %6             \n"
-      "vst1.8      {d7},  [%0], %6             \n"
-      "vst1.8      {d5},  [%0], %6             \n"
-      "vst1.8      {d19}, [%0], %6             \n"
-      "vst1.8      {d17}, [%0], %6             \n"
-      "vst1.8      {d23}, [%0], %6             \n"
-      "vst1.8      {d21}, [%0]                 \n"
-
-      "add         %1, #8*2                    \n"  // src   += 8*2
-      "add         %3, %3, %4, lsl #3          \n"  // dst_a += 8 * dst_stride_a
-      "add         %5, %5, %6, lsl #3          \n"  // dst_b += 8 * dst_stride_b
-      "subs        %7,  #8                     \n"  // w     -= 8
-      "bge         1b                          \n"
-
-    // add 8 back to counter. if the result is 0 there are
-    // no residuals.
-    "adds        %7, #8                        \n"
-    "beq         4f                            \n"
-
-    // some residual, so between 1 and 7 lines left to transpose
-    "cmp         %7, #2                        \n"
-    "blt         3f                            \n"
-
-    "cmp         %7, #4                        \n"
-    "blt         2f                            \n"
-
-    //TODO(frkoenig): Clean this up
-    // 4x8 block
-    "mov         %0, %1                        \n"
-    "vld1.64     {d0}, [%0], %2                \n"
-    "vld1.64     {d1}, [%0], %2                \n"
-    "vld1.64     {d2}, [%0], %2                \n"
-    "vld1.64     {d3}, [%0], %2                \n"
-    "vld1.64     {d4}, [%0], %2                \n"
-    "vld1.64     {d5}, [%0], %2                \n"
-    "vld1.64     {d6}, [%0], %2                \n"
-    "vld1.64     {d7}, [%0]                    \n"
-
-    "vld1.8      {q15}, [%8]                   \n"
-
-    "vtrn.8      q0, q1                        \n"
-    "vtrn.8      q2, q3                        \n"
-
-    "vtbl.8      d16, {d0, d1}, d30            \n"
-    "vtbl.8      d17, {d0, d1}, d31            \n"
-    "vtbl.8      d18, {d2, d3}, d30            \n"
-    "vtbl.8      d19, {d2, d3}, d31            \n"
-    "vtbl.8      d20, {d4, d5}, d30            \n"
-    "vtbl.8      d21, {d4, d5}, d31            \n"
-    "vtbl.8      d22, {d6, d7}, d30            \n"
-    "vtbl.8      d23, {d6, d7}, d31            \n"
-
-    "mov         %0, %3                        \n"
-
-    "vst1.32     {d16[0]},  [%0], %4           \n"
-    "vst1.32     {d16[1]},  [%0], %4           \n"
-    "vst1.32     {d17[0]},  [%0], %4           \n"
-    "vst1.32     {d17[1]},  [%0], %4           \n"
-
-    "add         %0, %3, #4                    \n"
-    "vst1.32     {d20[0]}, [%0], %4            \n"
-    "vst1.32     {d20[1]}, [%0], %4            \n"
-    "vst1.32     {d21[0]}, [%0], %4            \n"
-    "vst1.32     {d21[1]}, [%0]                \n"
-
-    "mov         %0, %5                        \n"
-
-    "vst1.32     {d18[0]}, [%0], %6            \n"
-    "vst1.32     {d18[1]}, [%0], %6            \n"
-    "vst1.32     {d19[0]}, [%0], %6            \n"
-    "vst1.32     {d19[1]}, [%0], %6            \n"
-
-    "add         %0, %5, #4                    \n"
-    "vst1.32     {d22[0]},  [%0], %6           \n"
-    "vst1.32     {d22[1]},  [%0], %6           \n"
-    "vst1.32     {d23[0]},  [%0], %6           \n"
-    "vst1.32     {d23[1]},  [%0]               \n"
-
-    "add         %1, #4*2                      \n"  // src   += 4 * 2
-    "add         %3, %3, %4, lsl #2            \n"  // dst_a += 4 * dst_stride_a
-    "add         %5, %5, %6, lsl #2            \n"  // dst_b += 4 * dst_stride_b
-    "subs        %7,  #4                       \n"  // w     -= 4
-    "beq         4f                            \n"
-
-    // some residual, check to see if it includes a 2x8 block,
-    // or less
-    "cmp         %7, #2                        \n"
-    "blt         3f                            \n"
-
-    // 2x8 block
-    "2:                                        \n"
-    "mov         %0, %1                        \n"
-    "vld2.16     {d0[0], d2[0]}, [%0], %2      \n"
-    "vld2.16     {d1[0], d3[0]}, [%0], %2      \n"
-    "vld2.16     {d0[1], d2[1]}, [%0], %2      \n"
-    "vld2.16     {d1[1], d3[1]}, [%0], %2      \n"
-    "vld2.16     {d0[2], d2[2]}, [%0], %2      \n"
-    "vld2.16     {d1[2], d3[2]}, [%0], %2      \n"
-    "vld2.16     {d0[3], d2[3]}, [%0], %2      \n"
-    "vld2.16     {d1[3], d3[3]}, [%0]          \n"
-
-    "vtrn.8      d0, d1                        \n"
-    "vtrn.8      d2, d3                        \n"
-
-    "mov         %0, %3                        \n"
-
-    "vst1.64     {d0}, [%0], %4                \n"
-    "vst1.64     {d2}, [%0]                    \n"
-
-    "mov         %0, %5                        \n"
-
-    "vst1.64     {d1}, [%0], %6                \n"
-    "vst1.64     {d3}, [%0]                    \n"
-
-    "add         %1, #2*2                      \n"  // src   += 2 * 2
-    "add         %3, %3, %4, lsl #1            \n"  // dst_a += 2 * dst_stride_a
-    "add         %5, %5, %6, lsl #1            \n"  // dst_b += 2 * dst_stride_b
-    "subs        %7,  #2                       \n"  // w     -= 2
-    "beq         4f                            \n"
-
-    // 1x8 block
-    "3:                                        \n"
-    "vld2.8      {d0[0], d1[0]}, [%1], %2      \n"
-    "vld2.8      {d0[1], d1[1]}, [%1], %2      \n"
-    "vld2.8      {d0[2], d1[2]}, [%1], %2      \n"
-    "vld2.8      {d0[3], d1[3]}, [%1], %2      \n"
-    "vld2.8      {d0[4], d1[4]}, [%1], %2      \n"
-    "vld2.8      {d0[5], d1[5]}, [%1], %2      \n"
-    "vld2.8      {d0[6], d1[6]}, [%1], %2      \n"
-    "vld2.8      {d0[7], d1[7]}, [%1]          \n"
-
-    "vst1.64     {d0}, [%3]                    \n"
-    "vst1.64     {d1}, [%5]                    \n"
-
-    "4:                                        \n"
-
-    : "+r"(src_temp),            // %0
-      "+r"(src),                 // %1
-      "+r"(src_stride),          // %2
-      "+r"(dst_a),               // %3
-      "+r"(dst_stride_a),        // %4
-      "+r"(dst_b),               // %5
-      "+r"(dst_stride_b),        // %6
-      "+r"(width)                // %7
-    : "r"(&kVTbl4x4TransposeDi)  // %8
-    : "memory", "cc",
-      "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11"
-  );
-}
-#endif
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_any.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_any.cc
deleted file mode 100755
index 90c6a3ff5f..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_any.cc
+++ /dev/null
@@ -1,542 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// TODO(fbarchard): Consider 'any' functions handling any quantity of pixels.
-// TODO(fbarchard): Consider 'any' functions handling odd alignment.
-// YUV to RGB does multiple of 8 with SIMD and remainder with C.
-#define YANY(NAMEANY, I420TORGB_SIMD, I420TORGB_C, UV_SHIFT, BPP, MASK)        \
-    void NAMEANY(const uint8* y_buf,                                           \
-                 const uint8* u_buf,                                           \
-                 const uint8* v_buf,                                           \
-                 uint8* rgb_buf,                                               \
-                 int width) {                                                  \
-      int n = width & ~MASK;                                                   \
-      I420TORGB_SIMD(y_buf, u_buf, v_buf, rgb_buf, n);                         \
-      I420TORGB_C(y_buf + n,                                                   \
-                  u_buf + (n >> UV_SHIFT),                                     \
-                  v_buf + (n >> UV_SHIFT),                                     \
-                  rgb_buf + n * BPP, width & MASK);                            \
-    }
-
-#ifdef HAS_I422TOARGBROW_SSSE3
-YANY(I444ToARGBRow_Any_SSSE3, I444ToARGBRow_Unaligned_SSSE3, I444ToARGBRow_C,
-     0, 4, 7)
-YANY(I422ToARGBRow_Any_SSSE3, I422ToARGBRow_Unaligned_SSSE3, I422ToARGBRow_C,
-     1, 4, 7)
-YANY(I411ToARGBRow_Any_SSSE3, I411ToARGBRow_Unaligned_SSSE3, I411ToARGBRow_C,
-     2, 4, 7)
-YANY(I422ToBGRARow_Any_SSSE3, I422ToBGRARow_Unaligned_SSSE3, I422ToBGRARow_C,
-     1, 4, 7)
-YANY(I422ToABGRRow_Any_SSSE3, I422ToABGRRow_Unaligned_SSSE3, I422ToABGRRow_C,
-     1, 4, 7)
-YANY(I422ToRGBARow_Any_SSSE3, I422ToRGBARow_Unaligned_SSSE3, I422ToRGBARow_C,
-     1, 4, 7)
-// I422ToRGB565Row_SSSE3 is unaligned.
-YANY(I422ToARGB4444Row_Any_SSSE3, I422ToARGB4444Row_SSSE3, I422ToARGB4444Row_C,
-     1, 2, 7)
-YANY(I422ToARGB1555Row_Any_SSSE3, I422ToARGB1555Row_SSSE3, I422ToARGB1555Row_C,
-     1, 2, 7)
-YANY(I422ToRGB565Row_Any_SSSE3, I422ToRGB565Row_SSSE3, I422ToRGB565Row_C,
-     1, 2, 7)
-// I422ToRGB24Row_SSSE3 is unaligned.
-YANY(I422ToRGB24Row_Any_SSSE3, I422ToRGB24Row_SSSE3, I422ToRGB24Row_C, 1, 3, 7)
-YANY(I422ToRAWRow_Any_SSSE3, I422ToRAWRow_SSSE3, I422ToRAWRow_C, 1, 3, 7)
-YANY(I422ToYUY2Row_Any_SSE2, I422ToYUY2Row_SSE2, I422ToYUY2Row_C, 1, 2, 15)
-YANY(I422ToUYVYRow_Any_SSE2, I422ToUYVYRow_SSE2, I422ToUYVYRow_C, 1, 2, 15)
-#endif  // HAS_I422TOARGBROW_SSSE3
-#ifdef HAS_I422TOARGBROW_AVX2
-YANY(I422ToARGBRow_Any_AVX2, I422ToARGBRow_AVX2, I422ToARGBRow_C, 1, 4, 15)
-#endif  // HAS_I422TOARGBROW_AVX2
-#ifdef HAS_I422TOARGBROW_NEON
-YANY(I444ToARGBRow_Any_NEON, I444ToARGBRow_NEON, I444ToARGBRow_C, 0, 4, 7)
-YANY(I422ToARGBRow_Any_NEON, I422ToARGBRow_NEON, I422ToARGBRow_C, 1, 4, 7)
-YANY(I411ToARGBRow_Any_NEON, I411ToARGBRow_NEON, I411ToARGBRow_C, 2, 4, 7)
-YANY(I422ToBGRARow_Any_NEON, I422ToBGRARow_NEON, I422ToBGRARow_C, 1, 4, 7)
-YANY(I422ToABGRRow_Any_NEON, I422ToABGRRow_NEON, I422ToABGRRow_C, 1, 4, 7)
-YANY(I422ToRGBARow_Any_NEON, I422ToRGBARow_NEON, I422ToRGBARow_C, 1, 4, 7)
-YANY(I422ToRGB24Row_Any_NEON, I422ToRGB24Row_NEON, I422ToRGB24Row_C, 1, 3, 7)
-YANY(I422ToRAWRow_Any_NEON, I422ToRAWRow_NEON, I422ToRAWRow_C, 1, 3, 7)
-YANY(I422ToARGB4444Row_Any_NEON, I422ToARGB4444Row_NEON, I422ToARGB4444Row_C,
-     1, 2, 7)
-YANY(I422ToARGB1555Row_Any_NEON, I422ToARGB1555Row_NEON, I422ToARGB1555Row_C,
-     1, 2, 7)
-YANY(I422ToRGB565Row_Any_NEON, I422ToRGB565Row_NEON, I422ToRGB565Row_C, 1, 2, 7)
-YANY(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, I422ToYUY2Row_C, 1, 2, 15)
-YANY(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, I422ToUYVYRow_C, 1, 2, 15)
-#endif  // HAS_I422TOARGBROW_NEON
-#undef YANY
-
-// Wrappers to handle odd width
-#define NV2NY(NAMEANY, NV12TORGB_SIMD, NV12TORGB_C, UV_SHIFT, BPP)             \
-    void NAMEANY(const uint8* y_buf,                                           \
-                 const uint8* uv_buf,                                          \
-                 uint8* rgb_buf,                                               \
-                 int width) {                                                  \
-      int n = width & ~7;                                                      \
-      NV12TORGB_SIMD(y_buf, uv_buf, rgb_buf, n);                               \
-      NV12TORGB_C(y_buf + n,                                                   \
-                  uv_buf + (n >> UV_SHIFT),                                    \
-                  rgb_buf + n * BPP, width & 7);                               \
-    }
-
-#ifdef HAS_NV12TOARGBROW_SSSE3
-NV2NY(NV12ToARGBRow_Any_SSSE3, NV12ToARGBRow_Unaligned_SSSE3, NV12ToARGBRow_C,
-      0, 4)
-NV2NY(NV21ToARGBRow_Any_SSSE3, NV21ToARGBRow_Unaligned_SSSE3, NV21ToARGBRow_C,
-      0, 4)
-#endif  // HAS_NV12TOARGBROW_SSSE3
-#ifdef HAS_NV12TOARGBROW_NEON
-NV2NY(NV12ToARGBRow_Any_NEON, NV12ToARGBRow_NEON, NV12ToARGBRow_C, 0, 4)
-NV2NY(NV21ToARGBRow_Any_NEON, NV21ToARGBRow_NEON, NV21ToARGBRow_C, 0, 4)
-#endif  // HAS_NV12TOARGBROW_NEON
-#ifdef HAS_NV12TORGB565ROW_SSSE3
-NV2NY(NV12ToRGB565Row_Any_SSSE3, NV12ToRGB565Row_SSSE3, NV12ToRGB565Row_C,
-      0, 2)
-NV2NY(NV21ToRGB565Row_Any_SSSE3, NV21ToRGB565Row_SSSE3, NV21ToRGB565Row_C,
-      0, 2)
-#endif  // HAS_NV12TORGB565ROW_SSSE3
-#ifdef HAS_NV12TORGB565ROW_NEON
-NV2NY(NV12ToRGB565Row_Any_NEON, NV12ToRGB565Row_NEON, NV12ToRGB565Row_C, 0, 2)
-NV2NY(NV21ToRGB565Row_Any_NEON, NV21ToRGB565Row_NEON, NV21ToRGB565Row_C, 0, 2)
-#endif  // HAS_NV12TORGB565ROW_NEON
-#undef NVANY
-
-#define RGBANY(NAMEANY, ARGBTORGB_SIMD, ARGBTORGB_C, MASK, SBPP, BPP)          \
-    void NAMEANY(const uint8* src,                                             \
-                 uint8* dst,                                                   \
-                 int width) {                                                  \
-      int n = width & ~MASK;                                                   \
-      ARGBTORGB_SIMD(src, dst, n);                                             \
-      ARGBTORGB_C(src + n * SBPP, dst + n * BPP, width & MASK);                \
-    }
-
-#if defined(HAS_ARGBTORGB24ROW_SSSE3)
-RGBANY(ARGBToRGB24Row_Any_SSSE3, ARGBToRGB24Row_SSSE3, ARGBToRGB24Row_C,
-       15, 4, 3)
-RGBANY(ARGBToRAWRow_Any_SSSE3, ARGBToRAWRow_SSSE3, ARGBToRAWRow_C,
-       15, 4, 3)
-RGBANY(ARGBToRGB565Row_Any_SSE2, ARGBToRGB565Row_SSE2, ARGBToRGB565Row_C,
-       3, 4, 2)
-RGBANY(ARGBToARGB1555Row_Any_SSE2, ARGBToARGB1555Row_SSE2, ARGBToARGB1555Row_C,
-       3, 4, 2)
-RGBANY(ARGBToARGB4444Row_Any_SSE2, ARGBToARGB4444Row_SSE2, ARGBToARGB4444Row_C,
-       3, 4, 2)
-#endif
-#if defined(HAS_I400TOARGBROW_SSE2)
-RGBANY(I400ToARGBRow_Any_SSE2, I400ToARGBRow_Unaligned_SSE2, I400ToARGBRow_C,
-       7, 1, 4)
-#endif
-#if defined(HAS_YTOARGBROW_SSE2)
-RGBANY(YToARGBRow_Any_SSE2, YToARGBRow_SSE2, YToARGBRow_C,
-       7, 1, 4)
-RGBANY(YUY2ToARGBRow_Any_SSSE3, YUY2ToARGBRow_Unaligned_SSSE3, YUY2ToARGBRow_C,
-       15, 2, 4)
-RGBANY(UYVYToARGBRow_Any_SSSE3, UYVYToARGBRow_Unaligned_SSSE3, UYVYToARGBRow_C,
-       15, 2, 4)
-// These require alignment on ARGB, so C is used for remainder.
-RGBANY(RGB24ToARGBRow_Any_SSSE3, RGB24ToARGBRow_SSSE3, RGB24ToARGBRow_C,
-       15, 3, 4)
-RGBANY(RAWToARGBRow_Any_SSSE3, RAWToARGBRow_SSSE3, RAWToARGBRow_C,
-       15, 3, 4)
-RGBANY(RGB565ToARGBRow_Any_SSE2, RGB565ToARGBRow_SSE2, RGB565ToARGBRow_C,
-       7, 2, 4)
-RGBANY(ARGB1555ToARGBRow_Any_SSE2, ARGB1555ToARGBRow_SSE2, ARGB1555ToARGBRow_C,
-       7, 2, 4)
-RGBANY(ARGB4444ToARGBRow_Any_SSE2, ARGB4444ToARGBRow_SSE2, ARGB4444ToARGBRow_C,
-       7, 2, 4)
-#endif
-#if defined(HAS_ARGBTORGB24ROW_NEON)
-RGBANY(ARGBToRGB24Row_Any_NEON, ARGBToRGB24Row_NEON, ARGBToRGB24Row_C, 7, 4, 3)
-RGBANY(ARGBToRAWRow_Any_NEON, ARGBToRAWRow_NEON, ARGBToRAWRow_C, 7, 4, 3)
-RGBANY(ARGBToRGB565Row_Any_NEON, ARGBToRGB565Row_NEON, ARGBToRGB565Row_C,
-       7, 4, 2)
-RGBANY(ARGBToARGB1555Row_Any_NEON, ARGBToARGB1555Row_NEON, ARGBToARGB1555Row_C,
-       7, 4, 2)
-RGBANY(ARGBToARGB4444Row_Any_NEON, ARGBToARGB4444Row_NEON, ARGBToARGB4444Row_C,
-       7, 4, 2)
-RGBANY(I400ToARGBRow_Any_NEON, I400ToARGBRow_NEON, I400ToARGBRow_C,
-       7, 1, 4)
-RGBANY(YToARGBRow_Any_NEON, YToARGBRow_NEON, YToARGBRow_C,
-       7, 1, 4)
-RGBANY(YUY2ToARGBRow_Any_NEON, YUY2ToARGBRow_NEON, YUY2ToARGBRow_C,
-       7, 2, 4)
-RGBANY(UYVYToARGBRow_Any_NEON, UYVYToARGBRow_NEON, UYVYToARGBRow_C,
-       7, 2, 4)
-#endif
-#undef RGBANY
-
-// ARGB to Bayer does multiple of 4 pixels, SSSE3 aligned src, unaligned dst.
-#define BAYERANY(NAMEANY, ARGBTORGB_SIMD, ARGBTORGB_C, MASK, SBPP, BPP)        \
-    void NAMEANY(const uint8* src,                                             \
-                 uint8* dst, uint32 selector,                                  \
-                 int width) {                                                  \
-      int n = width & ~MASK;                                                   \
-      ARGBTORGB_SIMD(src, dst, selector, n);                                   \
-      ARGBTORGB_C(src + n * SBPP, dst + n * BPP, selector, width & MASK);      \
-    }
-
-#if defined(HAS_ARGBTOBAYERROW_SSSE3)
-BAYERANY(ARGBToBayerRow_Any_SSSE3, ARGBToBayerRow_SSSE3, ARGBToBayerRow_C,
-         7, 4, 1)
-#endif
-#if defined(HAS_ARGBTOBAYERROW_NEON)
-BAYERANY(ARGBToBayerRow_Any_NEON, ARGBToBayerRow_NEON, ARGBToBayerRow_C,
-         7, 4, 1)
-#endif
-#if defined(HAS_ARGBTOBAYERGGROW_SSE2)
-BAYERANY(ARGBToBayerGGRow_Any_SSE2, ARGBToBayerGGRow_SSE2, ARGBToBayerGGRow_C,
-         7, 4, 1)
-#endif
-#if defined(HAS_ARGBTOBAYERGGROW_NEON)
-BAYERANY(ARGBToBayerGGRow_Any_NEON, ARGBToBayerGGRow_NEON, ARGBToBayerGGRow_C,
-         7, 4, 1)
-#endif
-
-#undef BAYERANY
-
-// RGB/YUV to Y does multiple of 16 with SIMD and last 16 with SIMD.
-#define YANY(NAMEANY, ARGBTOY_SIMD, SBPP, BPP, NUM)                            \
-    void NAMEANY(const uint8* src_argb, uint8* dst_y, int width) {             \
-      ARGBTOY_SIMD(src_argb, dst_y, width - NUM);                              \
-      ARGBTOY_SIMD(src_argb + (width - NUM) * SBPP,                            \
-                   dst_y + (width - NUM) * BPP, NUM);                          \
-    }
-
-#ifdef HAS_ARGBTOYROW_AVX2
-YANY(ARGBToYRow_Any_AVX2, ARGBToYRow_AVX2, 4, 1, 32)
-YANY(ARGBToYJRow_Any_AVX2, ARGBToYJRow_AVX2, 4, 1, 32)
-YANY(YUY2ToYRow_Any_AVX2, YUY2ToYRow_AVX2, 2, 1, 32)
-YANY(UYVYToYRow_Any_AVX2, UYVYToYRow_AVX2, 2, 1, 32)
-#endif
-#ifdef HAS_ARGBTOYROW_SSSE3
-YANY(ARGBToYRow_Any_SSSE3, ARGBToYRow_Unaligned_SSSE3, 4, 1, 16)
-#endif
-#ifdef HAS_BGRATOYROW_SSSE3
-YANY(BGRAToYRow_Any_SSSE3, BGRAToYRow_Unaligned_SSSE3, 4, 1, 16)
-YANY(ABGRToYRow_Any_SSSE3, ABGRToYRow_Unaligned_SSSE3, 4, 1, 16)
-YANY(RGBAToYRow_Any_SSSE3, RGBAToYRow_Unaligned_SSSE3, 4, 1, 16)
-YANY(YUY2ToYRow_Any_SSE2, YUY2ToYRow_Unaligned_SSE2, 2, 1, 16)
-YANY(UYVYToYRow_Any_SSE2, UYVYToYRow_Unaligned_SSE2, 2, 1, 16)
-#endif
-#ifdef HAS_ARGBTOYJROW_SSSE3
-YANY(ARGBToYJRow_Any_SSSE3, ARGBToYJRow_Unaligned_SSSE3, 4, 1, 16)
-#endif
-#ifdef HAS_ARGBTOYROW_NEON
-YANY(ARGBToYRow_Any_NEON, ARGBToYRow_NEON, 4, 1, 8)
-YANY(ARGBToYJRow_Any_NEON, ARGBToYJRow_NEON, 4, 1, 8)
-YANY(BGRAToYRow_Any_NEON, BGRAToYRow_NEON, 4, 1, 8)
-YANY(ABGRToYRow_Any_NEON, ABGRToYRow_NEON, 4, 1, 8)
-YANY(RGBAToYRow_Any_NEON, RGBAToYRow_NEON, 4, 1, 8)
-YANY(RGB24ToYRow_Any_NEON, RGB24ToYRow_NEON, 3, 1, 8)
-YANY(RAWToYRow_Any_NEON, RAWToYRow_NEON, 3, 1, 8)
-YANY(RGB565ToYRow_Any_NEON, RGB565ToYRow_NEON, 2, 1, 8)
-YANY(ARGB1555ToYRow_Any_NEON, ARGB1555ToYRow_NEON, 2, 1, 8)
-YANY(ARGB4444ToYRow_Any_NEON, ARGB4444ToYRow_NEON, 2, 1, 8)
-YANY(YUY2ToYRow_Any_NEON, YUY2ToYRow_NEON, 2, 1, 16)
-YANY(UYVYToYRow_Any_NEON, UYVYToYRow_NEON, 2, 1, 16)
-YANY(RGB24ToARGBRow_Any_NEON, RGB24ToARGBRow_NEON, 3, 4, 8)
-YANY(RAWToARGBRow_Any_NEON, RAWToARGBRow_NEON, 3, 4, 8)
-YANY(RGB565ToARGBRow_Any_NEON, RGB565ToARGBRow_NEON, 2, 4, 8)
-YANY(ARGB1555ToARGBRow_Any_NEON, ARGB1555ToARGBRow_NEON, 2, 4, 8)
-YANY(ARGB4444ToARGBRow_Any_NEON, ARGB4444ToARGBRow_NEON, 2, 4, 8)
-#endif
-#undef YANY
-
-#define YANY(NAMEANY, ARGBTOY_SIMD, ARGBTOY_C, SBPP, BPP, MASK)                \
-    void NAMEANY(const uint8* src_argb, uint8* dst_y, int width) {             \
-      int n = width & ~MASK;                                                   \
-      ARGBTOY_SIMD(src_argb, dst_y, n);                                        \
-      ARGBTOY_C(src_argb + n * SBPP,                                           \
-                dst_y  + n * BPP, width & MASK);                               \
-    }
-
-// Attenuate is destructive so last16 method can not be used due to overlap.
-#ifdef HAS_ARGBATTENUATEROW_SSSE3
-YANY(ARGBAttenuateRow_Any_SSSE3, ARGBAttenuateRow_SSSE3, ARGBAttenuateRow_C,
-     4, 4, 3)
-#endif
-#ifdef HAS_ARGBATTENUATEROW_SSE2
-YANY(ARGBAttenuateRow_Any_SSE2, ARGBAttenuateRow_SSE2, ARGBAttenuateRow_C,
-     4, 4, 3)
-#endif
-#ifdef HAS_ARGBUNATTENUATEROW_SSE2
-YANY(ARGBUnattenuateRow_Any_SSE2, ARGBUnattenuateRow_SSE2, ARGBUnattenuateRow_C,
-     4, 4, 3)
-#endif
-#ifdef HAS_ARGBATTENUATEROW_AVX2
-YANY(ARGBAttenuateRow_Any_AVX2, ARGBAttenuateRow_AVX2, ARGBAttenuateRow_C,
-     4, 4, 7)
-#endif
-#ifdef HAS_ARGBUNATTENUATEROW_AVX2
-YANY(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, ARGBUnattenuateRow_C,
-     4, 4, 7)
-#endif
-#ifdef HAS_ARGBATTENUATEROW_NEON
-YANY(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, ARGBAttenuateRow_C,
-     4, 4, 7)
-#endif
-#undef YANY
-
-// RGB/YUV to UV does multiple of 16 with SIMD and remainder with C.
-#define UVANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, MASK)                     \
-    void NAMEANY(const uint8* src_argb, int src_stride_argb,                   \
-                 uint8* dst_u, uint8* dst_v, int width) {                      \
-      int n = width & ~MASK;                                                   \
-      ANYTOUV_SIMD(src_argb, src_stride_argb, dst_u, dst_v, n);                \
-      ANYTOUV_C(src_argb  + n * BPP, src_stride_argb,                          \
-                dst_u + (n >> 1),                                              \
-                dst_v + (n >> 1),                                              \
-                width & MASK);                                                 \
-    }
-
-#ifdef HAS_ARGBTOUVROW_AVX2
-UVANY(ARGBToUVRow_Any_AVX2, ARGBToUVRow_AVX2, ARGBToUVRow_C, 4, 31)
-UVANY(YUY2ToUVRow_Any_AVX2, YUY2ToUVRow_AVX2, YUY2ToUVRow_C, 2, 31)
-UVANY(UYVYToUVRow_Any_AVX2, UYVYToUVRow_AVX2, UYVYToUVRow_C, 2, 31)
-#endif
-#ifdef HAS_ARGBTOUVROW_SSSE3
-UVANY(ARGBToUVRow_Any_SSSE3, ARGBToUVRow_Unaligned_SSSE3, ARGBToUVRow_C, 4, 15)
-UVANY(ARGBToUVJRow_Any_SSSE3, ARGBToUVJRow_Unaligned_SSSE3, ARGBToUVJRow_C,
-      4, 15)
-UVANY(BGRAToUVRow_Any_SSSE3, BGRAToUVRow_Unaligned_SSSE3, BGRAToUVRow_C, 4, 15)
-UVANY(ABGRToUVRow_Any_SSSE3, ABGRToUVRow_Unaligned_SSSE3, ABGRToUVRow_C, 4, 15)
-UVANY(RGBAToUVRow_Any_SSSE3, RGBAToUVRow_Unaligned_SSSE3, RGBAToUVRow_C, 4, 15)
-UVANY(YUY2ToUVRow_Any_SSE2, YUY2ToUVRow_Unaligned_SSE2, YUY2ToUVRow_C, 2, 15)
-UVANY(UYVYToUVRow_Any_SSE2, UYVYToUVRow_Unaligned_SSE2, UYVYToUVRow_C, 2, 15)
-#endif
-#ifdef HAS_ARGBTOUVROW_NEON
-UVANY(ARGBToUVRow_Any_NEON, ARGBToUVRow_NEON, ARGBToUVRow_C, 4, 15)
-UVANY(ARGBToUVJRow_Any_NEON, ARGBToUVJRow_NEON, ARGBToUVJRow_C, 4, 15)
-UVANY(BGRAToUVRow_Any_NEON, BGRAToUVRow_NEON, BGRAToUVRow_C, 4, 15)
-UVANY(ABGRToUVRow_Any_NEON, ABGRToUVRow_NEON, ABGRToUVRow_C, 4, 15)
-UVANY(RGBAToUVRow_Any_NEON, RGBAToUVRow_NEON, RGBAToUVRow_C, 4, 15)
-UVANY(RGB24ToUVRow_Any_NEON, RGB24ToUVRow_NEON, RGB24ToUVRow_C, 3, 15)
-UVANY(RAWToUVRow_Any_NEON, RAWToUVRow_NEON, RAWToUVRow_C, 3, 15)
-UVANY(RGB565ToUVRow_Any_NEON, RGB565ToUVRow_NEON, RGB565ToUVRow_C, 2, 15)
-UVANY(ARGB1555ToUVRow_Any_NEON, ARGB1555ToUVRow_NEON, ARGB1555ToUVRow_C, 2, 15)
-UVANY(ARGB4444ToUVRow_Any_NEON, ARGB4444ToUVRow_NEON, ARGB4444ToUVRow_C, 2, 15)
-UVANY(YUY2ToUVRow_Any_NEON, YUY2ToUVRow_NEON, YUY2ToUVRow_C, 2, 15)
-UVANY(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, UYVYToUVRow_C, 2, 15)
-#endif
-#undef UVANY
-
-#define UV422ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, MASK, SHIFT)           \
-    void NAMEANY(const uint8* src_uv,                                          \
-                 uint8* dst_u, uint8* dst_v, int width) {                      \
-      int n = width & ~MASK;                                                   \
-      ANYTOUV_SIMD(src_uv, dst_u, dst_v, n);                                   \
-      ANYTOUV_C(src_uv  + n * BPP,                                             \
-                dst_u + (n >> SHIFT),                                          \
-                dst_v + (n >> SHIFT),                                          \
-                width & MASK);                                                 \
-    }
-
-#ifdef HAS_ARGBTOUV444ROW_SSSE3
-UV422ANY(ARGBToUV444Row_Any_SSSE3, ARGBToUV444Row_Unaligned_SSSE3,
-         ARGBToUV444Row_C, 4, 15, 0)
-#endif
-#ifdef HAS_YUY2TOUV422ROW_AVX2
-UV422ANY(YUY2ToUV422Row_Any_AVX2, YUY2ToUV422Row_AVX2,
-         YUY2ToUV422Row_C, 2, 31, 1)
-UV422ANY(UYVYToUV422Row_Any_AVX2, UYVYToUV422Row_AVX2,
-         UYVYToUV422Row_C, 2, 31, 1)
-#endif
-#ifdef HAS_ARGBTOUVROW_SSSE3
-UV422ANY(ARGBToUV422Row_Any_SSSE3, ARGBToUV422Row_Unaligned_SSSE3,
-         ARGBToUV422Row_C, 4, 15, 1)
-UV422ANY(YUY2ToUV422Row_Any_SSE2, YUY2ToUV422Row_Unaligned_SSE2,
-         YUY2ToUV422Row_C, 2, 15, 1)
-UV422ANY(UYVYToUV422Row_Any_SSE2, UYVYToUV422Row_Unaligned_SSE2,
-         UYVYToUV422Row_C, 2, 15, 1)
-#endif
-#ifdef HAS_YUY2TOUV422ROW_NEON
-UV422ANY(ARGBToUV444Row_Any_NEON, ARGBToUV444Row_NEON,
-         ARGBToUV444Row_C, 4, 7, 0)
-UV422ANY(ARGBToUV422Row_Any_NEON, ARGBToUV422Row_NEON,
-         ARGBToUV422Row_C, 4, 15, 1)
-UV422ANY(ARGBToUV411Row_Any_NEON, ARGBToUV411Row_NEON,
-         ARGBToUV411Row_C, 4, 31, 2)
-UV422ANY(YUY2ToUV422Row_Any_NEON, YUY2ToUV422Row_NEON,
-         YUY2ToUV422Row_C, 2, 15, 1)
-UV422ANY(UYVYToUV422Row_Any_NEON, UYVYToUV422Row_NEON,
-         UYVYToUV422Row_C, 2, 15, 1)
-#endif
-#undef UV422ANY
-
-#define SPLITUVROWANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, MASK)                  \
-    void NAMEANY(const uint8* src_uv,                                          \
-                 uint8* dst_u, uint8* dst_v, int width) {                      \
-      int n = width & ~MASK;                                                   \
-      ANYTOUV_SIMD(src_uv, dst_u, dst_v, n);                                   \
-      ANYTOUV_C(src_uv + n * 2,                                                \
-                dst_u + n,                                                     \
-                dst_v + n,                                                     \
-                width & MASK);                                                 \
-    }
-
-#ifdef HAS_SPLITUVROW_SSE2
-SPLITUVROWANY(SplitUVRow_Any_SSE2, SplitUVRow_Unaligned_SSE2, SplitUVRow_C, 15)
-#endif
-#ifdef HAS_SPLITUVROW_AVX2
-SPLITUVROWANY(SplitUVRow_Any_AVX2, SplitUVRow_AVX2, SplitUVRow_C, 31)
-#endif
-#ifdef HAS_SPLITUVROW_NEON
-SPLITUVROWANY(SplitUVRow_Any_NEON, SplitUVRow_NEON, SplitUVRow_C, 15)
-#endif
-#ifdef HAS_SPLITUVROW_MIPS_DSPR2
-SPLITUVROWANY(SplitUVRow_Any_MIPS_DSPR2, SplitUVRow_Unaligned_MIPS_DSPR2,
-              SplitUVRow_C, 15)
-#endif
-#undef SPLITUVROWANY
-
-#define MERGEUVROW_ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, MASK)                 \
-    void NAMEANY(const uint8* src_u, const uint8* src_v,                       \
-                 uint8* dst_uv, int width) {                                   \
-      int n = width & ~MASK;                                                   \
-      ANYTOUV_SIMD(src_u, src_v, dst_uv, n);                                   \
-      ANYTOUV_C(src_u + n,                                                     \
-                src_v + n,                                                     \
-                dst_uv + n * 2,                                                \
-                width & MASK);                                                 \
-    }
-
-#ifdef HAS_MERGEUVROW_SSE2
-MERGEUVROW_ANY(MergeUVRow_Any_SSE2, MergeUVRow_Unaligned_SSE2, MergeUVRow_C, 15)
-#endif
-#ifdef HAS_MERGEUVROW_AVX2
-MERGEUVROW_ANY(MergeUVRow_Any_AVX2, MergeUVRow_AVX2, MergeUVRow_C, 31)
-#endif
-#ifdef HAS_MERGEUVROW_NEON
-MERGEUVROW_ANY(MergeUVRow_Any_NEON, MergeUVRow_NEON, MergeUVRow_C, 15)
-#endif
-#undef MERGEUVROW_ANY
-
-#define MATHROW_ANY(NAMEANY, ARGBMATH_SIMD, ARGBMATH_C, MASK)                  \
-    void NAMEANY(const uint8* src_argb0, const uint8* src_argb1,               \
-                 uint8* dst_argb, int width) {                                 \
-      int n = width & ~MASK;                                                   \
-      ARGBMATH_SIMD(src_argb0, src_argb1, dst_argb, n);                        \
-      ARGBMATH_C(src_argb0 + n * 4,                                            \
-                 src_argb1 + n * 4,                                            \
-                 dst_argb + n * 4,                                             \
-                 width & MASK);                                                \
-    }
-
-#ifdef HAS_ARGBMULTIPLYROW_SSE2
-MATHROW_ANY(ARGBMultiplyRow_Any_SSE2, ARGBMultiplyRow_SSE2, ARGBMultiplyRow_C,
-            3)
-#endif
-#ifdef HAS_ARGBADDROW_SSE2
-MATHROW_ANY(ARGBAddRow_Any_SSE2, ARGBAddRow_SSE2, ARGBAddRow_C, 3)
-#endif
-#ifdef HAS_ARGBSUBTRACTROW_SSE2
-MATHROW_ANY(ARGBSubtractRow_Any_SSE2, ARGBSubtractRow_SSE2, ARGBSubtractRow_C,
-            3)
-#endif
-#ifdef HAS_ARGBMULTIPLYROW_AVX2
-MATHROW_ANY(ARGBMultiplyRow_Any_AVX2, ARGBMultiplyRow_AVX2, ARGBMultiplyRow_C,
-            7)
-#endif
-#ifdef HAS_ARGBADDROW_AVX2
-MATHROW_ANY(ARGBAddRow_Any_AVX2, ARGBAddRow_AVX2, ARGBAddRow_C, 7)
-#endif
-#ifdef HAS_ARGBSUBTRACTROW_AVX2
-MATHROW_ANY(ARGBSubtractRow_Any_AVX2, ARGBSubtractRow_AVX2, ARGBSubtractRow_C,
-            7)
-#endif
-#ifdef HAS_ARGBMULTIPLYROW_NEON
-MATHROW_ANY(ARGBMultiplyRow_Any_NEON, ARGBMultiplyRow_NEON, ARGBMultiplyRow_C,
-            7)
-#endif
-#ifdef HAS_ARGBADDROW_NEON
-MATHROW_ANY(ARGBAddRow_Any_NEON, ARGBAddRow_NEON, ARGBAddRow_C, 7)
-#endif
-#ifdef HAS_ARGBSUBTRACTROW_NEON
-MATHROW_ANY(ARGBSubtractRow_Any_NEON, ARGBSubtractRow_NEON, ARGBSubtractRow_C,
-            7)
-#endif
-#undef MATHROW_ANY
-
-// Shuffle may want to work in place, so last16 method can not be used.
-#define YANY(NAMEANY, ARGBTOY_SIMD, ARGBTOY_C, SBPP, BPP, MASK)                \
-    void NAMEANY(const uint8* src_argb, uint8* dst_argb,                       \
-                 const uint8* shuffler, int width) {                           \
-      int n = width & ~MASK;                                                   \
-      ARGBTOY_SIMD(src_argb, dst_argb, shuffler, n);                           \
-      ARGBTOY_C(src_argb + n * SBPP,                                           \
-                dst_argb  + n * BPP, shuffler, width & MASK);                  \
-    }
-
-#ifdef HAS_ARGBSHUFFLEROW_SSE2
-YANY(ARGBShuffleRow_Any_SSE2, ARGBShuffleRow_SSE2,
-     ARGBShuffleRow_C, 4, 4, 3)
-#endif
-#ifdef HAS_ARGBSHUFFLEROW_SSSE3
-YANY(ARGBShuffleRow_Any_SSSE3, ARGBShuffleRow_Unaligned_SSSE3,
-     ARGBShuffleRow_C, 4, 4, 7)
-#endif
-#ifdef HAS_ARGBSHUFFLEROW_AVX2
-YANY(ARGBShuffleRow_Any_AVX2, ARGBShuffleRow_AVX2,
-     ARGBShuffleRow_C, 4, 4, 15)
-#endif
-#ifdef HAS_ARGBSHUFFLEROW_NEON
-YANY(ARGBShuffleRow_Any_NEON, ARGBShuffleRow_NEON,
-     ARGBShuffleRow_C, 4, 4, 3)
-#endif
-#undef YANY
-
-// Interpolate may want to work in place, so last16 method can not be used.
-#define NANY(NAMEANY, TERP_SIMD, TERP_C, SBPP, BPP, MASK)                      \
-    void NAMEANY(uint8* dst_ptr, const uint8* src_ptr,                         \
-                 ptrdiff_t src_stride_ptr, int width,                          \
-                 int source_y_fraction) {                                      \
-      int n = width & ~MASK;                                                   \
-      TERP_SIMD(dst_ptr, src_ptr, src_stride_ptr,                              \
-                n, source_y_fraction);                                         \
-      TERP_C(dst_ptr + n * BPP,                                                \
-             src_ptr + n * SBPP, src_stride_ptr,                               \
-             width & MASK, source_y_fraction);                                 \
-    }
-
-#ifdef HAS_INTERPOLATEROW_AVX2
-NANY(InterpolateRow_Any_AVX2, InterpolateRow_AVX2,
-     InterpolateRow_C, 1, 1, 32)
-#endif
-#ifdef HAS_INTERPOLATEROW_SSSE3
-NANY(InterpolateRow_Any_SSSE3, InterpolateRow_Unaligned_SSSE3,
-     InterpolateRow_C, 1, 1, 15)
-#endif
-#ifdef HAS_INTERPOLATEROW_SSE2
-NANY(InterpolateRow_Any_SSE2, InterpolateRow_Unaligned_SSE2,
-     InterpolateRow_C, 1, 1, 15)
-#endif
-#ifdef HAS_INTERPOLATEROW_NEON
-NANY(InterpolateRow_Any_NEON, InterpolateRow_NEON,
-     InterpolateRow_C, 1, 1, 15)
-#endif
-#ifdef HAS_INTERPOLATEROW_MIPS_DSPR2
-NANY(InterpolateRow_Any_MIPS_DSPR2, InterpolateRow_MIPS_DSPR2,
-     InterpolateRow_C, 1, 1, 3)
-#endif
-#undef NANY
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_common.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_common.cc
deleted file mode 100755
index 135bdc9084..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_common.cc
+++ /dev/null
@@ -1,2247 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#include <string.h>  // For memcpy and memset.
-
-#include "libyuv/basic_types.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// llvm x86 is poor at ternary operator, so use branchless min/max.
-
-#define USE_BRANCHLESS 1
-#if USE_BRANCHLESS
-static __inline int32 clamp0(int32 v) {
-  return ((-(v) >> 31) & (v));
-}
-
-static __inline int32 clamp255(int32 v) {
-  return (((255 - (v)) >> 31) | (v)) & 255;
-}
-
-static __inline uint32 Clamp(int32 val) {
-  int v = clamp0(val);
-  return (uint32)(clamp255(v));
-}
-
-static __inline uint32 Abs(int32 v) {
-  int m = v >> 31;
-  return (v + m) ^ m;
-}
-#else  // USE_BRANCHLESS
-static __inline int32 clamp0(int32 v) {
-  return (v < 0) ? 0 : v;
-}
-
-static __inline int32 clamp255(int32 v) {
-  return (v > 255) ? 255 : v;
-}
-
-static __inline uint32 Clamp(int32 val) {
-  int v = clamp0(val);
-  return (uint32)(clamp255(v));
-}
-
-static __inline uint32 Abs(int32 v) {
-  return (v < 0) ? -v : v;
-}
-#endif  // USE_BRANCHLESS
-
-#ifdef LIBYUV_LITTLE_ENDIAN
-#define WRITEWORD(p, v) *(uint32*)(p) = v
-#else
-static inline void WRITEWORD(uint8* p, uint32 v) {
-  p[0] = (uint8)(v & 255);
-  p[1] = (uint8)((v >> 8) & 255);
-  p[2] = (uint8)((v >> 16) & 255);
-  p[3] = (uint8)((v >> 24) & 255);
-}
-#endif
-
-void RGB24ToARGBRow_C(const uint8* src_rgb24, uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_rgb24[0];
-    uint8 g = src_rgb24[1];
-    uint8 r = src_rgb24[2];
-    dst_argb[0] = b;
-    dst_argb[1] = g;
-    dst_argb[2] = r;
-    dst_argb[3] = 255u;
-    dst_argb += 4;
-    src_rgb24 += 3;
-  }
-}
-
-void RAWToARGBRow_C(const uint8* src_raw, uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 r = src_raw[0];
-    uint8 g = src_raw[1];
-    uint8 b = src_raw[2];
-    dst_argb[0] = b;
-    dst_argb[1] = g;
-    dst_argb[2] = r;
-    dst_argb[3] = 255u;
-    dst_argb += 4;
-    src_raw += 3;
-  }
-}
-
-void RGB565ToARGBRow_C(const uint8* src_rgb565, uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_rgb565[0] & 0x1f;
-    uint8 g = (src_rgb565[0] >> 5) | ((src_rgb565[1] & 0x07) << 3);
-    uint8 r = src_rgb565[1] >> 3;
-    dst_argb[0] = (b << 3) | (b >> 2);
-    dst_argb[1] = (g << 2) | (g >> 4);
-    dst_argb[2] = (r << 3) | (r >> 2);
-    dst_argb[3] = 255u;
-    dst_argb += 4;
-    src_rgb565 += 2;
-  }
-}
-
-void ARGB1555ToARGBRow_C(const uint8* src_argb1555, uint8* dst_argb,
-                         int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb1555[0] & 0x1f;
-    uint8 g = (src_argb1555[0] >> 5) | ((src_argb1555[1] & 0x03) << 3);
-    uint8 r = (src_argb1555[1] & 0x7c) >> 2;
-    uint8 a = src_argb1555[1] >> 7;
-    dst_argb[0] = (b << 3) | (b >> 2);
-    dst_argb[1] = (g << 3) | (g >> 2);
-    dst_argb[2] = (r << 3) | (r >> 2);
-    dst_argb[3] = -a;
-    dst_argb += 4;
-    src_argb1555 += 2;
-  }
-}
-
-void ARGB4444ToARGBRow_C(const uint8* src_argb4444, uint8* dst_argb,
-                         int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb4444[0] & 0x0f;
-    uint8 g = src_argb4444[0] >> 4;
-    uint8 r = src_argb4444[1] & 0x0f;
-    uint8 a = src_argb4444[1] >> 4;
-    dst_argb[0] = (b << 4) | b;
-    dst_argb[1] = (g << 4) | g;
-    dst_argb[2] = (r << 4) | r;
-    dst_argb[3] = (a << 4) | a;
-    dst_argb += 4;
-    src_argb4444 += 2;
-  }
-}
-
-void ARGBToRGB24Row_C(const uint8* src_argb, uint8* dst_rgb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb[0];
-    uint8 g = src_argb[1];
-    uint8 r = src_argb[2];
-    dst_rgb[0] = b;
-    dst_rgb[1] = g;
-    dst_rgb[2] = r;
-    dst_rgb += 3;
-    src_argb += 4;
-  }
-}
-
-void ARGBToRAWRow_C(const uint8* src_argb, uint8* dst_rgb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb[0];
-    uint8 g = src_argb[1];
-    uint8 r = src_argb[2];
-    dst_rgb[0] = r;
-    dst_rgb[1] = g;
-    dst_rgb[2] = b;
-    dst_rgb += 3;
-    src_argb += 4;
-  }
-}
-
-void ARGBToRGB565Row_C(const uint8* src_argb, uint8* dst_rgb, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_argb[0] >> 3;
-    uint8 g0 = src_argb[1] >> 2;
-    uint8 r0 = src_argb[2] >> 3;
-    uint8 b1 = src_argb[4] >> 3;
-    uint8 g1 = src_argb[5] >> 2;
-    uint8 r1 = src_argb[6] >> 3;
-    WRITEWORD(dst_rgb, b0 | (g0 << 5) | (r0 << 11) |
-              (b1 << 16) | (g1 << 21) | (r1 << 27));
-    dst_rgb += 4;
-    src_argb += 8;
-  }
-  if (width & 1) {
-    uint8 b0 = src_argb[0] >> 3;
-    uint8 g0 = src_argb[1] >> 2;
-    uint8 r0 = src_argb[2] >> 3;
-    *(uint16*)(dst_rgb) = b0 | (g0 << 5) | (r0 << 11);
-  }
-}
-
-void ARGBToARGB1555Row_C(const uint8* src_argb, uint8* dst_rgb, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_argb[0] >> 3;
-    uint8 g0 = src_argb[1] >> 3;
-    uint8 r0 = src_argb[2] >> 3;
-    uint8 a0 = src_argb[3] >> 7;
-    uint8 b1 = src_argb[4] >> 3;
-    uint8 g1 = src_argb[5] >> 3;
-    uint8 r1 = src_argb[6] >> 3;
-    uint8 a1 = src_argb[7] >> 7;
-    *(uint32*)(dst_rgb) =
-        b0 | (g0 << 5) | (r0 << 10) | (a0 << 15) |
-        (b1 << 16) | (g1 << 21) | (r1 << 26) | (a1 << 31);
-    dst_rgb += 4;
-    src_argb += 8;
-  }
-  if (width & 1) {
-    uint8 b0 = src_argb[0] >> 3;
-    uint8 g0 = src_argb[1] >> 3;
-    uint8 r0 = src_argb[2] >> 3;
-    uint8 a0 = src_argb[3] >> 7;
-    *(uint16*)(dst_rgb) =
-        b0 | (g0 << 5) | (r0 << 10) | (a0 << 15);
-  }
-}
-
-void ARGBToARGB4444Row_C(const uint8* src_argb, uint8* dst_rgb, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_argb[0] >> 4;
-    uint8 g0 = src_argb[1] >> 4;
-    uint8 r0 = src_argb[2] >> 4;
-    uint8 a0 = src_argb[3] >> 4;
-    uint8 b1 = src_argb[4] >> 4;
-    uint8 g1 = src_argb[5] >> 4;
-    uint8 r1 = src_argb[6] >> 4;
-    uint8 a1 = src_argb[7] >> 4;
-    *(uint32*)(dst_rgb) =
-        b0 | (g0 << 4) | (r0 << 8) | (a0 << 12) |
-        (b1 << 16) | (g1 << 20) | (r1 << 24) | (a1 << 28);
-    dst_rgb += 4;
-    src_argb += 8;
-  }
-  if (width & 1) {
-    uint8 b0 = src_argb[0] >> 4;
-    uint8 g0 = src_argb[1] >> 4;
-    uint8 r0 = src_argb[2] >> 4;
-    uint8 a0 = src_argb[3] >> 4;
-    *(uint16*)(dst_rgb) =
-        b0 | (g0 << 4) | (r0 << 8) | (a0 << 12);
-  }
-}
-
-static __inline int RGBToY(uint8 r, uint8 g, uint8 b) {
-  return (66 * r + 129 * g +  25 * b + 0x1080) >> 8;
-}
-
-static __inline int RGBToU(uint8 r, uint8 g, uint8 b) {
-  return (112 * b - 74 * g - 38 * r + 0x8080) >> 8;
-}
-static __inline int RGBToV(uint8 r, uint8 g, uint8 b) {
-  return (112 * r - 94 * g - 18 * b + 0x8080) >> 8;
-}
-
-#define MAKEROWY(NAME, R, G, B, BPP) \
-void NAME ## ToYRow_C(const uint8* src_argb0, uint8* dst_y, int width) {       \
-  int x;                                                                       \
-  for (x = 0; x < width; ++x) {                                                \
-    dst_y[0] = RGBToY(src_argb0[R], src_argb0[G], src_argb0[B]);               \
-    src_argb0 += BPP;                                                          \
-    dst_y += 1;                                                                \
-  }                                                                            \
-}                                                                              \
-void NAME ## ToUVRow_C(const uint8* src_rgb0, int src_stride_rgb,              \
-                       uint8* dst_u, uint8* dst_v, int width) {                \
-  const uint8* src_rgb1 = src_rgb0 + src_stride_rgb;                           \
-  int x;                                                                       \
-  for (x = 0; x < width - 1; x += 2) {                                         \
-    uint8 ab = (src_rgb0[B] + src_rgb0[B + BPP] +                              \
-               src_rgb1[B] + src_rgb1[B + BPP]) >> 2;                          \
-    uint8 ag = (src_rgb0[G] + src_rgb0[G + BPP] +                              \
-               src_rgb1[G] + src_rgb1[G + BPP]) >> 2;                          \
-    uint8 ar = (src_rgb0[R] + src_rgb0[R + BPP] +                              \
-               src_rgb1[R] + src_rgb1[R + BPP]) >> 2;                          \
-    dst_u[0] = RGBToU(ar, ag, ab);                                             \
-    dst_v[0] = RGBToV(ar, ag, ab);                                             \
-    src_rgb0 += BPP * 2;                                                       \
-    src_rgb1 += BPP * 2;                                                       \
-    dst_u += 1;                                                                \
-    dst_v += 1;                                                                \
-  }                                                                            \
-  if (width & 1) {                                                             \
-    uint8 ab = (src_rgb0[B] + src_rgb1[B]) >> 1;                               \
-    uint8 ag = (src_rgb0[G] + src_rgb1[G]) >> 1;                               \
-    uint8 ar = (src_rgb0[R] + src_rgb1[R]) >> 1;                               \
-    dst_u[0] = RGBToU(ar, ag, ab);                                             \
-    dst_v[0] = RGBToV(ar, ag, ab);                                             \
-  }                                                                            \
-}
-
-MAKEROWY(ARGB, 2, 1, 0, 4)
-MAKEROWY(BGRA, 1, 2, 3, 4)
-MAKEROWY(ABGR, 0, 1, 2, 4)
-MAKEROWY(RGBA, 3, 2, 1, 4)
-MAKEROWY(RGB24, 2, 1, 0, 3)
-MAKEROWY(RAW, 0, 1, 2, 3)
-#undef MAKEROWY
-
-// JPeg uses a variation on BT.601-1 full range
-// y =  0.29900 * r + 0.58700 * g + 0.11400 * b
-// u = -0.16874 * r - 0.33126 * g + 0.50000 * b  + center
-// v =  0.50000 * r - 0.41869 * g - 0.08131 * b  + center
-// BT.601 Mpeg range uses:
-// b 0.1016 * 255 = 25.908 = 25
-// g 0.5078 * 255 = 129.489 = 129
-// r 0.2578 * 255 = 65.739 = 66
-// JPeg 8 bit Y (not used):
-// b 0.11400 * 256 = 29.184 = 29
-// g 0.58700 * 256 = 150.272 = 150
-// r 0.29900 * 256 = 76.544 = 77
-// JPeg 7 bit Y:
-// b 0.11400 * 128 = 14.592 = 15
-// g 0.58700 * 128 = 75.136 = 75
-// r 0.29900 * 128 = 38.272 = 38
-// JPeg 8 bit U:
-// b  0.50000 * 255 = 127.5 = 127
-// g -0.33126 * 255 = -84.4713 = -84
-// r -0.16874 * 255 = -43.0287 = -43
-// JPeg 8 bit V:
-// b -0.08131 * 255 = -20.73405 = -20
-// g -0.41869 * 255 = -106.76595 = -107
-// r  0.50000 * 255 = 127.5 = 127
-
-static __inline int RGBToYJ(uint8 r, uint8 g, uint8 b) {
-  return (38 * r + 75 * g +  15 * b + 64) >> 7;
-}
-
-static __inline int RGBToUJ(uint8 r, uint8 g, uint8 b) {
-  return (127 * b - 84 * g - 43 * r + 0x8080) >> 8;
-}
-static __inline int RGBToVJ(uint8 r, uint8 g, uint8 b) {
-  return (127 * r - 107 * g - 20 * b + 0x8080) >> 8;
-}
-
-#define AVGB(a, b) (((a) + (b) + 1) >> 1)
-
-#define MAKEROWYJ(NAME, R, G, B, BPP) \
-void NAME ## ToYJRow_C(const uint8* src_argb0, uint8* dst_y, int width) {      \
-  int x;                                                                       \
-  for (x = 0; x < width; ++x) {                                                \
-    dst_y[0] = RGBToYJ(src_argb0[R], src_argb0[G], src_argb0[B]);              \
-    src_argb0 += BPP;                                                          \
-    dst_y += 1;                                                                \
-  }                                                                            \
-}                                                                              \
-void NAME ## ToUVJRow_C(const uint8* src_rgb0, int src_stride_rgb,             \
-                        uint8* dst_u, uint8* dst_v, int width) {               \
-  const uint8* src_rgb1 = src_rgb0 + src_stride_rgb;                           \
-  int x;                                                                       \
-  for (x = 0; x < width - 1; x += 2) {                                         \
-    uint8 ab = AVGB(AVGB(src_rgb0[B], src_rgb1[B]),                            \
-                    AVGB(src_rgb0[B + BPP], src_rgb1[B + BPP]));               \
-    uint8 ag = AVGB(AVGB(src_rgb0[G], src_rgb1[G]),                            \
-                    AVGB(src_rgb0[G + BPP], src_rgb1[G + BPP]));               \
-    uint8 ar = AVGB(AVGB(src_rgb0[R], src_rgb1[R]),                            \
-                    AVGB(src_rgb0[R + BPP], src_rgb1[R + BPP]));               \
-    dst_u[0] = RGBToUJ(ar, ag, ab);                                            \
-    dst_v[0] = RGBToVJ(ar, ag, ab);                                            \
-    src_rgb0 += BPP * 2;                                                       \
-    src_rgb1 += BPP * 2;                                                       \
-    dst_u += 1;                                                                \
-    dst_v += 1;                                                                \
-  }                                                                            \
-  if (width & 1) {                                                             \
-    uint8 ab = AVGB(src_rgb0[B], src_rgb1[B]);                                 \
-    uint8 ag = AVGB(src_rgb0[G], src_rgb1[G]);                                 \
-    uint8 ar = AVGB(src_rgb0[R], src_rgb1[R]);                                 \
-    dst_u[0] = RGBToUJ(ar, ag, ab);                                            \
-    dst_v[0] = RGBToVJ(ar, ag, ab);                                            \
-  }                                                                            \
-}
-
-MAKEROWYJ(ARGB, 2, 1, 0, 4)
-#undef MAKEROWYJ
-
-void RGB565ToYRow_C(const uint8* src_rgb565, uint8* dst_y, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_rgb565[0] & 0x1f;
-    uint8 g = (src_rgb565[0] >> 5) | ((src_rgb565[1] & 0x07) << 3);
-    uint8 r = src_rgb565[1] >> 3;
-    b = (b << 3) | (b >> 2);
-    g = (g << 2) | (g >> 4);
-    r = (r << 3) | (r >> 2);
-    dst_y[0] = RGBToY(r, g, b);
-    src_rgb565 += 2;
-    dst_y += 1;
-  }
-}
-
-void ARGB1555ToYRow_C(const uint8* src_argb1555, uint8* dst_y, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb1555[0] & 0x1f;
-    uint8 g = (src_argb1555[0] >> 5) | ((src_argb1555[1] & 0x03) << 3);
-    uint8 r = (src_argb1555[1] & 0x7c) >> 2;
-    b = (b << 3) | (b >> 2);
-    g = (g << 3) | (g >> 2);
-    r = (r << 3) | (r >> 2);
-    dst_y[0] = RGBToY(r, g, b);
-    src_argb1555 += 2;
-    dst_y += 1;
-  }
-}
-
-void ARGB4444ToYRow_C(const uint8* src_argb4444, uint8* dst_y, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 b = src_argb4444[0] & 0x0f;
-    uint8 g = src_argb4444[0] >> 4;
-    uint8 r = src_argb4444[1] & 0x0f;
-    b = (b << 4) | b;
-    g = (g << 4) | g;
-    r = (r << 4) | r;
-    dst_y[0] = RGBToY(r, g, b);
-    src_argb4444 += 2;
-    dst_y += 1;
-  }
-}
-
-void RGB565ToUVRow_C(const uint8* src_rgb565, int src_stride_rgb565,
-                     uint8* dst_u, uint8* dst_v, int width) {
-  const uint8* next_rgb565 = src_rgb565 + src_stride_rgb565;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_rgb565[0] & 0x1f;
-    uint8 g0 = (src_rgb565[0] >> 5) | ((src_rgb565[1] & 0x07) << 3);
-    uint8 r0 = src_rgb565[1] >> 3;
-    uint8 b1 = src_rgb565[2] & 0x1f;
-    uint8 g1 = (src_rgb565[2] >> 5) | ((src_rgb565[3] & 0x07) << 3);
-    uint8 r1 = src_rgb565[3] >> 3;
-    uint8 b2 = next_rgb565[0] & 0x1f;
-    uint8 g2 = (next_rgb565[0] >> 5) | ((next_rgb565[1] & 0x07) << 3);
-    uint8 r2 = next_rgb565[1] >> 3;
-    uint8 b3 = next_rgb565[2] & 0x1f;
-    uint8 g3 = (next_rgb565[2] >> 5) | ((next_rgb565[3] & 0x07) << 3);
-    uint8 r3 = next_rgb565[3] >> 3;
-    uint8 b = (b0 + b1 + b2 + b3);  // 565 * 4 = 787.
-    uint8 g = (g0 + g1 + g2 + g3);
-    uint8 r = (r0 + r1 + r2 + r3);
-    b = (b << 1) | (b >> 6);  // 787 -> 888.
-    r = (r << 1) | (r >> 6);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-    src_rgb565 += 4;
-    next_rgb565 += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-  if (width & 1) {
-    uint8 b0 = src_rgb565[0] & 0x1f;
-    uint8 g0 = (src_rgb565[0] >> 5) | ((src_rgb565[1] & 0x07) << 3);
-    uint8 r0 = src_rgb565[1] >> 3;
-    uint8 b2 = next_rgb565[0] & 0x1f;
-    uint8 g2 = (next_rgb565[0] >> 5) | ((next_rgb565[1] & 0x07) << 3);
-    uint8 r2 = next_rgb565[1] >> 3;
-    uint8 b = (b0 + b2);  // 565 * 2 = 676.
-    uint8 g = (g0 + g2);
-    uint8 r = (r0 + r2);
-    b = (b << 2) | (b >> 4);  // 676 -> 888
-    g = (g << 1) | (g >> 6);
-    r = (r << 2) | (r >> 4);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-  }
-}
-
-void ARGB1555ToUVRow_C(const uint8* src_argb1555, int src_stride_argb1555,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  const uint8* next_argb1555 = src_argb1555 + src_stride_argb1555;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_argb1555[0] & 0x1f;
-    uint8 g0 = (src_argb1555[0] >> 5) | ((src_argb1555[1] & 0x03) << 3);
-    uint8 r0 = (src_argb1555[1] & 0x7c) >> 2;
-    uint8 b1 = src_argb1555[2] & 0x1f;
-    uint8 g1 = (src_argb1555[2] >> 5) | ((src_argb1555[3] & 0x03) << 3);
-    uint8 r1 = (src_argb1555[3] & 0x7c) >> 2;
-    uint8 b2 = next_argb1555[0] & 0x1f;
-    uint8 g2 = (next_argb1555[0] >> 5) | ((next_argb1555[1] & 0x03) << 3);
-    uint8 r2 = (next_argb1555[1] & 0x7c) >> 2;
-    uint8 b3 = next_argb1555[2] & 0x1f;
-    uint8 g3 = (next_argb1555[2] >> 5) | ((next_argb1555[3] & 0x03) << 3);
-    uint8 r3 = (next_argb1555[3] & 0x7c) >> 2;
-    uint8 b = (b0 + b1 + b2 + b3);  // 555 * 4 = 777.
-    uint8 g = (g0 + g1 + g2 + g3);
-    uint8 r = (r0 + r1 + r2 + r3);
-    b = (b << 1) | (b >> 6);  // 777 -> 888.
-    g = (g << 1) | (g >> 6);
-    r = (r << 1) | (r >> 6);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-    src_argb1555 += 4;
-    next_argb1555 += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-  if (width & 1) {
-    uint8 b0 = src_argb1555[0] & 0x1f;
-    uint8 g0 = (src_argb1555[0] >> 5) | ((src_argb1555[1] & 0x03) << 3);
-    uint8 r0 = (src_argb1555[1] & 0x7c) >> 2;
-    uint8 b2 = next_argb1555[0] & 0x1f;
-    uint8 g2 = (next_argb1555[0] >> 5) | ((next_argb1555[1] & 0x03) << 3);
-    uint8 r2 = next_argb1555[1] >> 3;
-    uint8 b = (b0 + b2);  // 555 * 2 = 666.
-    uint8 g = (g0 + g2);
-    uint8 r = (r0 + r2);
-    b = (b << 2) | (b >> 4);  // 666 -> 888.
-    g = (g << 2) | (g >> 4);
-    r = (r << 2) | (r >> 4);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-  }
-}
-
-void ARGB4444ToUVRow_C(const uint8* src_argb4444, int src_stride_argb4444,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  const uint8* next_argb4444 = src_argb4444 + src_stride_argb4444;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 b0 = src_argb4444[0] & 0x0f;
-    uint8 g0 = src_argb4444[0] >> 4;
-    uint8 r0 = src_argb4444[1] & 0x0f;
-    uint8 b1 = src_argb4444[2] & 0x0f;
-    uint8 g1 = src_argb4444[2] >> 4;
-    uint8 r1 = src_argb4444[3] & 0x0f;
-    uint8 b2 = next_argb4444[0] & 0x0f;
-    uint8 g2 = next_argb4444[0] >> 4;
-    uint8 r2 = next_argb4444[1] & 0x0f;
-    uint8 b3 = next_argb4444[2] & 0x0f;
-    uint8 g3 = next_argb4444[2] >> 4;
-    uint8 r3 = next_argb4444[3] & 0x0f;
-    uint8 b = (b0 + b1 + b2 + b3);  // 444 * 4 = 666.
-    uint8 g = (g0 + g1 + g2 + g3);
-    uint8 r = (r0 + r1 + r2 + r3);
-    b = (b << 2) | (b >> 4);  // 666 -> 888.
-    g = (g << 2) | (g >> 4);
-    r = (r << 2) | (r >> 4);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-    src_argb4444 += 4;
-    next_argb4444 += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-  if (width & 1) {
-    uint8 b0 = src_argb4444[0] & 0x0f;
-    uint8 g0 = src_argb4444[0] >> 4;
-    uint8 r0 = src_argb4444[1] & 0x0f;
-    uint8 b2 = next_argb4444[0] & 0x0f;
-    uint8 g2 = next_argb4444[0] >> 4;
-    uint8 r2 = next_argb4444[1] & 0x0f;
-    uint8 b = (b0 + b2);  // 444 * 2 = 555.
-    uint8 g = (g0 + g2);
-    uint8 r = (r0 + r2);
-    b = (b << 3) | (b >> 2);  // 555 -> 888.
-    g = (g << 3) | (g >> 2);
-    r = (r << 3) | (r >> 2);
-    dst_u[0] = RGBToU(r, g, b);
-    dst_v[0] = RGBToV(r, g, b);
-  }
-}
-
-void ARGBToUV444Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 ab = src_argb[0];
-    uint8 ag = src_argb[1];
-    uint8 ar = src_argb[2];
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-    src_argb += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-}
-
-void ARGBToUV422Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 ab = (src_argb[0] + src_argb[4]) >> 1;
-    uint8 ag = (src_argb[1] + src_argb[5]) >> 1;
-    uint8 ar = (src_argb[2] + src_argb[6]) >> 1;
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-    src_argb += 8;
-    dst_u += 1;
-    dst_v += 1;
-  }
-  if (width & 1) {
-    uint8 ab = src_argb[0];
-    uint8 ag = src_argb[1];
-    uint8 ar = src_argb[2];
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-  }
-}
-
-void ARGBToUV411Row_C(const uint8* src_argb,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  int x;
-  for (x = 0; x < width - 3; x += 4) {
-    uint8 ab = (src_argb[0] + src_argb[4] + src_argb[8] + src_argb[12]) >> 2;
-    uint8 ag = (src_argb[1] + src_argb[5] + src_argb[9] + src_argb[13]) >> 2;
-    uint8 ar = (src_argb[2] + src_argb[6] + src_argb[10] + src_argb[14]) >> 2;
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-    src_argb += 16;
-    dst_u += 1;
-    dst_v += 1;
-  }
-  if ((width & 3) == 3) {
-    uint8 ab = (src_argb[0] + src_argb[4] + src_argb[8]) / 3;
-    uint8 ag = (src_argb[1] + src_argb[5] + src_argb[9]) / 3;
-    uint8 ar = (src_argb[2] + src_argb[6] + src_argb[10]) / 3;
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-  } else if ((width & 3) == 2) {
-    uint8 ab = (src_argb[0] + src_argb[4]) >> 1;
-    uint8 ag = (src_argb[1] + src_argb[5]) >> 1;
-    uint8 ar = (src_argb[2] + src_argb[6]) >> 1;
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-  } else if ((width & 3) == 1) {
-    uint8 ab = src_argb[0];
-    uint8 ag = src_argb[1];
-    uint8 ar = src_argb[2];
-    dst_u[0] = RGBToU(ar, ag, ab);
-    dst_v[0] = RGBToV(ar, ag, ab);
-  }
-}
-
-void ARGBGrayRow_C(const uint8* src_argb, uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 y = RGBToYJ(src_argb[2], src_argb[1], src_argb[0]);
-    dst_argb[2] = dst_argb[1] = dst_argb[0] = y;
-    dst_argb[3] = src_argb[3];
-    dst_argb += 4;
-    src_argb += 4;
-  }
-}
-
-// Convert a row of image to Sepia tone.
-void ARGBSepiaRow_C(uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    int b = dst_argb[0];
-    int g = dst_argb[1];
-    int r = dst_argb[2];
-    int sb = (b * 17 + g * 68 + r * 35) >> 7;
-    int sg = (b * 22 + g * 88 + r * 45) >> 7;
-    int sr = (b * 24 + g * 98 + r * 50) >> 7;
-    // b does not over flow. a is preserved from original.
-    dst_argb[0] = sb;
-    dst_argb[1] = clamp255(sg);
-    dst_argb[2] = clamp255(sr);
-    dst_argb += 4;
-  }
-}
-
-// Apply color matrix to a row of image. Matrix is signed.
-// TODO(fbarchard): Consider adding rounding (+32).
-void ARGBColorMatrixRow_C(const uint8* src_argb, uint8* dst_argb,
-                          const int8* matrix_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    int b = src_argb[0];
-    int g = src_argb[1];
-    int r = src_argb[2];
-    int a = src_argb[3];
-    int sb = (b * matrix_argb[0] + g * matrix_argb[1] +
-              r * matrix_argb[2] + a * matrix_argb[3]) >> 6;
-    int sg = (b * matrix_argb[4] + g * matrix_argb[5] +
-              r * matrix_argb[6] + a * matrix_argb[7]) >> 6;
-    int sr = (b * matrix_argb[8] + g * matrix_argb[9] +
-              r * matrix_argb[10] + a * matrix_argb[11]) >> 6;
-    int sa = (b * matrix_argb[12] + g * matrix_argb[13] +
-              r * matrix_argb[14] + a * matrix_argb[15]) >> 6;
-    dst_argb[0] = Clamp(sb);
-    dst_argb[1] = Clamp(sg);
-    dst_argb[2] = Clamp(sr);
-    dst_argb[3] = Clamp(sa);
-    src_argb += 4;
-    dst_argb += 4;
-  }
-}
-
-// Apply color table to a row of image.
-void ARGBColorTableRow_C(uint8* dst_argb, const uint8* table_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    int b = dst_argb[0];
-    int g = dst_argb[1];
-    int r = dst_argb[2];
-    int a = dst_argb[3];
-    dst_argb[0] = table_argb[b * 4 + 0];
-    dst_argb[1] = table_argb[g * 4 + 1];
-    dst_argb[2] = table_argb[r * 4 + 2];
-    dst_argb[3] = table_argb[a * 4 + 3];
-    dst_argb += 4;
-  }
-}
-
-// Apply color table to a row of image.
-void RGBColorTableRow_C(uint8* dst_argb, const uint8* table_argb, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    int b = dst_argb[0];
-    int g = dst_argb[1];
-    int r = dst_argb[2];
-    dst_argb[0] = table_argb[b * 4 + 0];
-    dst_argb[1] = table_argb[g * 4 + 1];
-    dst_argb[2] = table_argb[r * 4 + 2];
-    dst_argb += 4;
-  }
-}
-
-void ARGBQuantizeRow_C(uint8* dst_argb, int scale, int interval_size,
-                       int interval_offset, int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    int b = dst_argb[0];
-    int g = dst_argb[1];
-    int r = dst_argb[2];
-    dst_argb[0] = (b * scale >> 16) * interval_size + interval_offset;
-    dst_argb[1] = (g * scale >> 16) * interval_size + interval_offset;
-    dst_argb[2] = (r * scale >> 16) * interval_size + interval_offset;
-    dst_argb += 4;
-  }
-}
-
-#define REPEAT8(v) (v) | ((v) << 8)
-#define SHADE(f, v) v * f >> 24
-
-void ARGBShadeRow_C(const uint8* src_argb, uint8* dst_argb, int width,
-                    uint32 value) {
-  const uint32 b_scale = REPEAT8(value & 0xff);
-  const uint32 g_scale = REPEAT8((value >> 8) & 0xff);
-  const uint32 r_scale = REPEAT8((value >> 16) & 0xff);
-  const uint32 a_scale = REPEAT8(value >> 24);
-
-  int i;
-  for (i = 0; i < width; ++i) {
-    const uint32 b = REPEAT8(src_argb[0]);
-    const uint32 g = REPEAT8(src_argb[1]);
-    const uint32 r = REPEAT8(src_argb[2]);
-    const uint32 a = REPEAT8(src_argb[3]);
-    dst_argb[0] = SHADE(b, b_scale);
-    dst_argb[1] = SHADE(g, g_scale);
-    dst_argb[2] = SHADE(r, r_scale);
-    dst_argb[3] = SHADE(a, a_scale);
-    src_argb += 4;
-    dst_argb += 4;
-  }
-}
-#undef REPEAT8
-#undef SHADE
-
-#define REPEAT8(v) (v) | ((v) << 8)
-#define SHADE(f, v) v * f >> 16
-
-void ARGBMultiplyRow_C(const uint8* src_argb0, const uint8* src_argb1,
-                       uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    const uint32 b = REPEAT8(src_argb0[0]);
-    const uint32 g = REPEAT8(src_argb0[1]);
-    const uint32 r = REPEAT8(src_argb0[2]);
-    const uint32 a = REPEAT8(src_argb0[3]);
-    const uint32 b_scale = src_argb1[0];
-    const uint32 g_scale = src_argb1[1];
-    const uint32 r_scale = src_argb1[2];
-    const uint32 a_scale = src_argb1[3];
-    dst_argb[0] = SHADE(b, b_scale);
-    dst_argb[1] = SHADE(g, g_scale);
-    dst_argb[2] = SHADE(r, r_scale);
-    dst_argb[3] = SHADE(a, a_scale);
-    src_argb0 += 4;
-    src_argb1 += 4;
-    dst_argb += 4;
-  }
-}
-#undef REPEAT8
-#undef SHADE
-
-#define SHADE(f, v) clamp255(v + f)
-
-void ARGBAddRow_C(const uint8* src_argb0, const uint8* src_argb1,
-                  uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    const int b = src_argb0[0];
-    const int g = src_argb0[1];
-    const int r = src_argb0[2];
-    const int a = src_argb0[3];
-    const int b_add = src_argb1[0];
-    const int g_add = src_argb1[1];
-    const int r_add = src_argb1[2];
-    const int a_add = src_argb1[3];
-    dst_argb[0] = SHADE(b, b_add);
-    dst_argb[1] = SHADE(g, g_add);
-    dst_argb[2] = SHADE(r, r_add);
-    dst_argb[3] = SHADE(a, a_add);
-    src_argb0 += 4;
-    src_argb1 += 4;
-    dst_argb += 4;
-  }
-}
-#undef SHADE
-
-#define SHADE(f, v) clamp0(f - v)
-
-void ARGBSubtractRow_C(const uint8* src_argb0, const uint8* src_argb1,
-                       uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    const int b = src_argb0[0];
-    const int g = src_argb0[1];
-    const int r = src_argb0[2];
-    const int a = src_argb0[3];
-    const int b_sub = src_argb1[0];
-    const int g_sub = src_argb1[1];
-    const int r_sub = src_argb1[2];
-    const int a_sub = src_argb1[3];
-    dst_argb[0] = SHADE(b, b_sub);
-    dst_argb[1] = SHADE(g, g_sub);
-    dst_argb[2] = SHADE(r, r_sub);
-    dst_argb[3] = SHADE(a, a_sub);
-    src_argb0 += 4;
-    src_argb1 += 4;
-    dst_argb += 4;
-  }
-}
-#undef SHADE
-
-// Sobel functions which mimics SSSE3.
-void SobelXRow_C(const uint8* src_y0, const uint8* src_y1, const uint8* src_y2,
-                 uint8* dst_sobelx, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int a = src_y0[i];
-    int b = src_y1[i];
-    int c = src_y2[i];
-    int a_sub = src_y0[i + 2];
-    int b_sub = src_y1[i + 2];
-    int c_sub = src_y2[i + 2];
-    int a_diff = a - a_sub;
-    int b_diff = b - b_sub;
-    int c_diff = c - c_sub;
-    int sobel = Abs(a_diff + b_diff * 2 + c_diff);
-    dst_sobelx[i] = (uint8)(clamp255(sobel));
-  }
-}
-
-void SobelYRow_C(const uint8* src_y0, const uint8* src_y1,
-                 uint8* dst_sobely, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int a = src_y0[i + 0];
-    int b = src_y0[i + 1];
-    int c = src_y0[i + 2];
-    int a_sub = src_y1[i + 0];
-    int b_sub = src_y1[i + 1];
-    int c_sub = src_y1[i + 2];
-    int a_diff = a - a_sub;
-    int b_diff = b - b_sub;
-    int c_diff = c - c_sub;
-    int sobel = Abs(a_diff + b_diff * 2 + c_diff);
-    dst_sobely[i] = (uint8)(clamp255(sobel));
-  }
-}
-
-void SobelRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int r = src_sobelx[i];
-    int b = src_sobely[i];
-    int s = clamp255(r + b);
-    dst_argb[0] = (uint8)(s);
-    dst_argb[1] = (uint8)(s);
-    dst_argb[2] = (uint8)(s);
-    dst_argb[3] = (uint8)(255u);
-    dst_argb += 4;
-  }
-}
-
-void SobelToPlaneRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                       uint8* dst_y, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int r = src_sobelx[i];
-    int b = src_sobely[i];
-    int s = clamp255(r + b);
-    dst_y[i] = (uint8)(s);
-  }
-}
-
-void SobelXYRow_C(const uint8* src_sobelx, const uint8* src_sobely,
-                  uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    int r = src_sobelx[i];
-    int b = src_sobely[i];
-    int g = clamp255(r + b);
-    dst_argb[0] = (uint8)(b);
-    dst_argb[1] = (uint8)(g);
-    dst_argb[2] = (uint8)(r);
-    dst_argb[3] = (uint8)(255u);
-    dst_argb += 4;
-  }
-}
-
-void I400ToARGBRow_C(const uint8* src_y, uint8* dst_argb, int width) {
-  // Copy a Y to RGB.
-  int x;
-  for (x = 0; x < width; ++x) {
-    uint8 y = src_y[0];
-    dst_argb[2] = dst_argb[1] = dst_argb[0] = y;
-    dst_argb[3] = 255u;
-    dst_argb += 4;
-    ++src_y;
-  }
-}
-
-// C reference code that mimics the YUV assembly.
-
-#define YG 74 /* (int8)(1.164 * 64 + 0.5) */
-
-#define UB 127 /* min(63,(int8)(2.018 * 64)) */
-#define UG -25 /* (int8)(-0.391 * 64 - 0.5) */
-#define UR 0
-
-#define VB 0
-#define VG -52 /* (int8)(-0.813 * 64 - 0.5) */
-#define VR 102 /* (int8)(1.596 * 64 + 0.5) */
-
-// Bias
-#define BB UB * 128 + VB * 128
-#define BG UG * 128 + VG * 128
-#define BR UR * 128 + VR * 128
-
-static __inline void YuvPixel(uint8 y, uint8 u, uint8 v,
-                              uint8* b, uint8* g, uint8* r) {
-  int32 y1 = ((int32)(y) - 16) * YG;
-  *b = Clamp((int32)((u * UB + v * VB) - (BB) + y1) >> 6);
-  *g = Clamp((int32)((u * UG + v * VG) - (BG) + y1) >> 6);
-  *r = Clamp((int32)((u * UR + v * VR) - (BR) + y1) >> 6);
-}
-
-#if !defined(LIBYUV_DISABLE_NEON) && \
-    (defined(__ARM_NEON__) || defined(LIBYUV_NEON))
-// C mimic assembly.
-// TODO(fbarchard): Remove subsampling from Neon.
-void I444ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint8 u = (src_u[0] + src_u[1] + 1) >> 1;
-    uint8 v = (src_v[0] + src_v[1] + 1) >> 1;
-    YuvPixel(src_y[0], u, v, rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], u, v, rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    src_u += 2;
-    src_v += 2;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-  }
-}
-#else
-void I444ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    src_y += 1;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 4;  // Advance 1 pixel.
-  }
-}
-#endif
-// Also used for 420
-void I422ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void I422ToRGB24Row_C(const uint8* src_y,
-                      const uint8* src_u,
-                      const uint8* src_v,
-                      uint8* rgb_buf,
-                      int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 3, rgb_buf + 4, rgb_buf + 5);
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 6;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-  }
-}
-
-void I422ToRAWRow_C(const uint8* src_y,
-                    const uint8* src_u,
-                    const uint8* src_v,
-                    uint8* rgb_buf,
-                    int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 2, rgb_buf + 1, rgb_buf + 0);
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 5, rgb_buf + 4, rgb_buf + 3);
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 6;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 2, rgb_buf + 1, rgb_buf + 0);
-  }
-}
-
-void I422ToARGB4444Row_C(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb4444,
-                         int width) {
-  uint8 b0;
-  uint8 g0;
-  uint8 r0;
-  uint8 b1;
-  uint8 g1;
-  uint8 r1;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    YuvPixel(src_y[1], src_u[0], src_v[0], &b1, &g1, &r1);
-    b0 = b0 >> 4;
-    g0 = g0 >> 4;
-    r0 = r0 >> 4;
-    b1 = b1 >> 4;
-    g1 = g1 >> 4;
-    r1 = r1 >> 4;
-    *(uint32*)(dst_argb4444) = b0 | (g0 << 4) | (r0 << 8) |
-        (b1 << 16) | (g1 << 20) | (r1 << 24) | 0xf000f000;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    dst_argb4444 += 4;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    b0 = b0 >> 4;
-    g0 = g0 >> 4;
-    r0 = r0 >> 4;
-    *(uint16*)(dst_argb4444) = b0 | (g0 << 4) | (r0 << 8) |
-        0xf000;
-  }
-}
-
-void I422ToARGB1555Row_C(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_argb1555,
-                         int width) {
-  uint8 b0;
-  uint8 g0;
-  uint8 r0;
-  uint8 b1;
-  uint8 g1;
-  uint8 r1;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    YuvPixel(src_y[1], src_u[0], src_v[0], &b1, &g1, &r1);
-    b0 = b0 >> 3;
-    g0 = g0 >> 3;
-    r0 = r0 >> 3;
-    b1 = b1 >> 3;
-    g1 = g1 >> 3;
-    r1 = r1 >> 3;
-    *(uint32*)(dst_argb1555) = b0 | (g0 << 5) | (r0 << 10) |
-        (b1 << 16) | (g1 << 21) | (r1 << 26) | 0x80008000;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    dst_argb1555 += 4;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    b0 = b0 >> 3;
-    g0 = g0 >> 3;
-    r0 = r0 >> 3;
-    *(uint16*)(dst_argb1555) = b0 | (g0 << 5) | (r0 << 10) |
-        0x8000;
-  }
-}
-
-void I422ToRGB565Row_C(const uint8* src_y,
-                       const uint8* src_u,
-                       const uint8* src_v,
-                       uint8* dst_rgb565,
-                       int width) {
-  uint8 b0;
-  uint8 g0;
-  uint8 r0;
-  uint8 b1;
-  uint8 g1;
-  uint8 r1;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    YuvPixel(src_y[1], src_u[0], src_v[0], &b1, &g1, &r1);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    b1 = b1 >> 3;
-    g1 = g1 >> 2;
-    r1 = r1 >> 3;
-    *(uint32*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11) |
-        (b1 << 16) | (g1 << 21) | (r1 << 27);
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    dst_rgb565 += 4;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0], &b0, &g0, &r0);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    *(uint16*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11);
-  }
-}
-
-void I411ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 3; x += 4) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    YuvPixel(src_y[2], src_u[0], src_v[0],
-             rgb_buf + 8, rgb_buf + 9, rgb_buf + 10);
-    rgb_buf[11] = 255;
-    YuvPixel(src_y[3], src_u[0], src_v[0],
-             rgb_buf + 12, rgb_buf + 13, rgb_buf + 14);
-    rgb_buf[15] = 255;
-    src_y += 4;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 16;  // Advance 4 pixels.
-  }
-  if (width & 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void NV12ToARGBRow_C(const uint8* src_y,
-                     const uint8* usrc_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], usrc_v[0], usrc_v[1],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], usrc_v[0], usrc_v[1],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    usrc_v += 2;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], usrc_v[0], usrc_v[1],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void NV21ToARGBRow_C(const uint8* src_y,
-                     const uint8* src_vu,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_vu[1], src_vu[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-
-    YuvPixel(src_y[1], src_vu[1], src_vu[0],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-
-    src_y += 2;
-    src_vu += 2;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_vu[1], src_vu[0],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void NV12ToRGB565Row_C(const uint8* src_y,
-                       const uint8* usrc_v,
-                       uint8* dst_rgb565,
-                       int width) {
-  uint8 b0;
-  uint8 g0;
-  uint8 r0;
-  uint8 b1;
-  uint8 g1;
-  uint8 r1;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], usrc_v[0], usrc_v[1], &b0, &g0, &r0);
-    YuvPixel(src_y[1], usrc_v[0], usrc_v[1], &b1, &g1, &r1);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    b1 = b1 >> 3;
-    g1 = g1 >> 2;
-    r1 = r1 >> 3;
-    *(uint32*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11) |
-        (b1 << 16) | (g1 << 21) | (r1 << 27);
-    src_y += 2;
-    usrc_v += 2;
-    dst_rgb565 += 4;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], usrc_v[0], usrc_v[1], &b0, &g0, &r0);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    *(uint16*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11);
-  }
-}
-
-void NV21ToRGB565Row_C(const uint8* src_y,
-                       const uint8* vsrc_u,
-                       uint8* dst_rgb565,
-                       int width) {
-  uint8 b0;
-  uint8 g0;
-  uint8 r0;
-  uint8 b1;
-  uint8 g1;
-  uint8 r1;
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], vsrc_u[1], vsrc_u[0], &b0, &g0, &r0);
-    YuvPixel(src_y[1], vsrc_u[1], vsrc_u[0], &b1, &g1, &r1);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    b1 = b1 >> 3;
-    g1 = g1 >> 2;
-    r1 = r1 >> 3;
-    *(uint32*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11) |
-        (b1 << 16) | (g1 << 21) | (r1 << 27);
-    src_y += 2;
-    vsrc_u += 2;
-    dst_rgb565 += 4;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], vsrc_u[1], vsrc_u[0], &b0, &g0, &r0);
-    b0 = b0 >> 3;
-    g0 = g0 >> 2;
-    r0 = r0 >> 3;
-    *(uint16*)(dst_rgb565) = b0 | (g0 << 5) | (r0 << 11);
-  }
-}
-
-void YUY2ToARGBRow_C(const uint8* src_yuy2,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_yuy2[0], src_yuy2[1], src_yuy2[3],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_yuy2[2], src_yuy2[1], src_yuy2[3],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_yuy2 += 4;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_yuy2[0], src_yuy2[1], src_yuy2[3],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void UYVYToARGBRow_C(const uint8* src_uyvy,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_uyvy[1], src_uyvy[0], src_uyvy[2],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_uyvy[3], src_uyvy[0], src_uyvy[2],
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_uyvy += 4;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_uyvy[1], src_uyvy[0], src_uyvy[2],
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void I422ToBGRARow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 3, rgb_buf + 2, rgb_buf + 1);
-    rgb_buf[0] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 7, rgb_buf + 6, rgb_buf + 5);
-    rgb_buf[4] = 255;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 3, rgb_buf + 2, rgb_buf + 1);
-    rgb_buf[0] = 255;
-  }
-}
-
-void I422ToABGRRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 2, rgb_buf + 1, rgb_buf + 0);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 6, rgb_buf + 5, rgb_buf + 4);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 2, rgb_buf + 1, rgb_buf + 0);
-    rgb_buf[3] = 255;
-  }
-}
-
-void I422ToRGBARow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* rgb_buf,
-                     int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 1, rgb_buf + 2, rgb_buf + 3);
-    rgb_buf[0] = 255;
-    YuvPixel(src_y[1], src_u[0], src_v[0],
-             rgb_buf + 5, rgb_buf + 6, rgb_buf + 7);
-    rgb_buf[4] = 255;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], src_u[0], src_v[0],
-             rgb_buf + 1, rgb_buf + 2, rgb_buf + 3);
-    rgb_buf[0] = 255;
-  }
-}
-
-void YToARGBRow_C(const uint8* src_y, uint8* rgb_buf, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    YuvPixel(src_y[0], 128, 128,
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-    YuvPixel(src_y[1], 128, 128,
-             rgb_buf + 4, rgb_buf + 5, rgb_buf + 6);
-    rgb_buf[7] = 255;
-    src_y += 2;
-    rgb_buf += 8;  // Advance 2 pixels.
-  }
-  if (width & 1) {
-    YuvPixel(src_y[0], 128, 128,
-             rgb_buf + 0, rgb_buf + 1, rgb_buf + 2);
-    rgb_buf[3] = 255;
-  }
-}
-
-void MirrorRow_C(const uint8* src, uint8* dst, int width) {
-  int x;
-  src += width - 1;
-  for (x = 0; x < width - 1; x += 2) {
-    dst[x] = src[0];
-    dst[x + 1] = src[-1];
-    src -= 2;
-  }
-  if (width & 1) {
-    dst[width - 1] = src[0];
-  }
-}
-
-void MirrorUVRow_C(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int width) {
-  int x;
-  src_uv += (width - 1) << 1;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_u[x] = src_uv[0];
-    dst_u[x + 1] = src_uv[-2];
-    dst_v[x] = src_uv[1];
-    dst_v[x + 1] = src_uv[-2 + 1];
-    src_uv -= 4;
-  }
-  if (width & 1) {
-    dst_u[width - 1] = src_uv[0];
-    dst_v[width - 1] = src_uv[1];
-  }
-}
-
-void ARGBMirrorRow_C(const uint8* src, uint8* dst, int width) {
-  int x;
-  const uint32* src32 = (const uint32*)(src);
-  uint32* dst32 = (uint32*)(dst);
-  src32 += width - 1;
-  for (x = 0; x < width - 1; x += 2) {
-    dst32[x] = src32[0];
-    dst32[x + 1] = src32[-1];
-    src32 -= 2;
-  }
-  if (width & 1) {
-    dst32[width - 1] = src32[0];
-  }
-}
-
-void SplitUVRow_C(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_u[x] = src_uv[0];
-    dst_u[x + 1] = src_uv[2];
-    dst_v[x] = src_uv[1];
-    dst_v[x + 1] = src_uv[3];
-    src_uv += 4;
-  }
-  if (width & 1) {
-    dst_u[width - 1] = src_uv[0];
-    dst_v[width - 1] = src_uv[1];
-  }
-}
-
-void MergeUVRow_C(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                  int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_uv[0] = src_u[x];
-    dst_uv[1] = src_v[x];
-    dst_uv[2] = src_u[x + 1];
-    dst_uv[3] = src_v[x + 1];
-    dst_uv += 4;
-  }
-  if (width & 1) {
-    dst_uv[0] = src_u[width - 1];
-    dst_uv[1] = src_v[width - 1];
-  }
-}
-
-void CopyRow_C(const uint8* src, uint8* dst, int count) {
-  memcpy(dst, src, count);
-}
-
-void SetRow_C(uint8* dst, uint32 v8, int count) {
-#ifdef _MSC_VER
-  // VC will generate rep stosb.
-  int x;
-  for (x = 0; x < count; ++x) {
-    dst[x] = v8;
-  }
-#else
-  memset(dst, v8, count);
-#endif
-}
-
-void ARGBSetRows_C(uint8* dst, uint32 v32, int width,
-                 int dst_stride, int height) {
-  int y;
-  for (y = 0; y < height; ++y) {
-    uint32* d = (uint32*)(dst);
-    int x;
-    for (x = 0; x < width; ++x) {
-      d[x] = v32;
-    }
-    dst += dst_stride;
-  }
-}
-
-// Filter 2 rows of YUY2 UV's (422) into U and V (420).
-void YUY2ToUVRow_C(const uint8* src_yuy2, int src_stride_yuy2,
-                   uint8* dst_u, uint8* dst_v, int width) {
-  // Output a row of UV values, filtering 2 rows of YUY2.
-  int x;
-  for (x = 0; x < width; x += 2) {
-    dst_u[0] = (src_yuy2[1] + src_yuy2[src_stride_yuy2 + 1] + 1) >> 1;
-    dst_v[0] = (src_yuy2[3] + src_yuy2[src_stride_yuy2 + 3] + 1) >> 1;
-    src_yuy2 += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-}
-
-// Copy row of YUY2 UV's (422) into U and V (422).
-void YUY2ToUV422Row_C(const uint8* src_yuy2,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  // Output a row of UV values.
-  int x;
-  for (x = 0; x < width; x += 2) {
-    dst_u[0] = src_yuy2[1];
-    dst_v[0] = src_yuy2[3];
-    src_yuy2 += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-}
-
-// Copy row of YUY2 Y's (422) into Y (420/422).
-void YUY2ToYRow_C(const uint8* src_yuy2, uint8* dst_y, int width) {
-  // Output a row of Y values.
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_y[x] = src_yuy2[0];
-    dst_y[x + 1] = src_yuy2[2];
-    src_yuy2 += 4;
-  }
-  if (width & 1) {
-    dst_y[width - 1] = src_yuy2[0];
-  }
-}
-
-// Filter 2 rows of UYVY UV's (422) into U and V (420).
-void UYVYToUVRow_C(const uint8* src_uyvy, int src_stride_uyvy,
-                   uint8* dst_u, uint8* dst_v, int width) {
-  // Output a row of UV values.
-  int x;
-  for (x = 0; x < width; x += 2) {
-    dst_u[0] = (src_uyvy[0] + src_uyvy[src_stride_uyvy + 0] + 1) >> 1;
-    dst_v[0] = (src_uyvy[2] + src_uyvy[src_stride_uyvy + 2] + 1) >> 1;
-    src_uyvy += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-}
-
-// Copy row of UYVY UV's (422) into U and V (422).
-void UYVYToUV422Row_C(const uint8* src_uyvy,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  // Output a row of UV values.
-  int x;
-  for (x = 0; x < width; x += 2) {
-    dst_u[0] = src_uyvy[0];
-    dst_v[0] = src_uyvy[2];
-    src_uyvy += 4;
-    dst_u += 1;
-    dst_v += 1;
-  }
-}
-
-// Copy row of UYVY Y's (422) into Y (420/422).
-void UYVYToYRow_C(const uint8* src_uyvy, uint8* dst_y, int width) {
-  // Output a row of Y values.
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_y[x] = src_uyvy[1];
-    dst_y[x + 1] = src_uyvy[3];
-    src_uyvy += 4;
-  }
-  if (width & 1) {
-    dst_y[width - 1] = src_uyvy[1];
-  }
-}
-
-#define BLEND(f, b, a) (((256 - a) * b) >> 8) + f
-
-// Blend src_argb0 over src_argb1 and store to dst_argb.
-// dst_argb may be src_argb0 or src_argb1.
-// This code mimics the SSSE3 version for better testability.
-void ARGBBlendRow_C(const uint8* src_argb0, const uint8* src_argb1,
-                    uint8* dst_argb, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    uint32 fb = src_argb0[0];
-    uint32 fg = src_argb0[1];
-    uint32 fr = src_argb0[2];
-    uint32 a = src_argb0[3];
-    uint32 bb = src_argb1[0];
-    uint32 bg = src_argb1[1];
-    uint32 br = src_argb1[2];
-    dst_argb[0] = BLEND(fb, bb, a);
-    dst_argb[1] = BLEND(fg, bg, a);
-    dst_argb[2] = BLEND(fr, br, a);
-    dst_argb[3] = 255u;
-
-    fb = src_argb0[4 + 0];
-    fg = src_argb0[4 + 1];
-    fr = src_argb0[4 + 2];
-    a = src_argb0[4 + 3];
-    bb = src_argb1[4 + 0];
-    bg = src_argb1[4 + 1];
-    br = src_argb1[4 + 2];
-    dst_argb[4 + 0] = BLEND(fb, bb, a);
-    dst_argb[4 + 1] = BLEND(fg, bg, a);
-    dst_argb[4 + 2] = BLEND(fr, br, a);
-    dst_argb[4 + 3] = 255u;
-    src_argb0 += 8;
-    src_argb1 += 8;
-    dst_argb += 8;
-  }
-
-  if (width & 1) {
-    uint32 fb = src_argb0[0];
-    uint32 fg = src_argb0[1];
-    uint32 fr = src_argb0[2];
-    uint32 a = src_argb0[3];
-    uint32 bb = src_argb1[0];
-    uint32 bg = src_argb1[1];
-    uint32 br = src_argb1[2];
-    dst_argb[0] = BLEND(fb, bb, a);
-    dst_argb[1] = BLEND(fg, bg, a);
-    dst_argb[2] = BLEND(fr, br, a);
-    dst_argb[3] = 255u;
-  }
-}
-#undef BLEND
-#define ATTENUATE(f, a) (a | (a << 8)) * (f | (f << 8)) >> 24
-
-// Multiply source RGB by alpha and store to destination.
-// This code mimics the SSSE3 version for better testability.
-void ARGBAttenuateRow_C(const uint8* src_argb, uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width - 1; i += 2) {
-    uint32 b = src_argb[0];
-    uint32 g = src_argb[1];
-    uint32 r = src_argb[2];
-    uint32 a = src_argb[3];
-    dst_argb[0] = ATTENUATE(b, a);
-    dst_argb[1] = ATTENUATE(g, a);
-    dst_argb[2] = ATTENUATE(r, a);
-    dst_argb[3] = a;
-    b = src_argb[4];
-    g = src_argb[5];
-    r = src_argb[6];
-    a = src_argb[7];
-    dst_argb[4] = ATTENUATE(b, a);
-    dst_argb[5] = ATTENUATE(g, a);
-    dst_argb[6] = ATTENUATE(r, a);
-    dst_argb[7] = a;
-    src_argb += 8;
-    dst_argb += 8;
-  }
-
-  if (width & 1) {
-    const uint32 b = src_argb[0];
-    const uint32 g = src_argb[1];
-    const uint32 r = src_argb[2];
-    const uint32 a = src_argb[3];
-    dst_argb[0] = ATTENUATE(b, a);
-    dst_argb[1] = ATTENUATE(g, a);
-    dst_argb[2] = ATTENUATE(r, a);
-    dst_argb[3] = a;
-  }
-}
-#undef ATTENUATE
-
-// Divide source RGB by alpha and store to destination.
-// b = (b * 255 + (a / 2)) / a;
-// g = (g * 255 + (a / 2)) / a;
-// r = (r * 255 + (a / 2)) / a;
-// Reciprocal method is off by 1 on some values. ie 125
-// 8.8 fixed point inverse table with 1.0 in upper short and 1 / a in lower.
-#define T(a) 0x01000000 + (0x10000 / a)
-const uint32 fixed_invtbl8[256] = {
-  0x01000000, 0x0100ffff, T(0x02), T(0x03), T(0x04), T(0x05), T(0x06), T(0x07),
-  T(0x08), T(0x09), T(0x0a), T(0x0b), T(0x0c), T(0x0d), T(0x0e), T(0x0f),
-  T(0x10), T(0x11), T(0x12), T(0x13), T(0x14), T(0x15), T(0x16), T(0x17),
-  T(0x18), T(0x19), T(0x1a), T(0x1b), T(0x1c), T(0x1d), T(0x1e), T(0x1f),
-  T(0x20), T(0x21), T(0x22), T(0x23), T(0x24), T(0x25), T(0x26), T(0x27),
-  T(0x28), T(0x29), T(0x2a), T(0x2b), T(0x2c), T(0x2d), T(0x2e), T(0x2f),
-  T(0x30), T(0x31), T(0x32), T(0x33), T(0x34), T(0x35), T(0x36), T(0x37),
-  T(0x38), T(0x39), T(0x3a), T(0x3b), T(0x3c), T(0x3d), T(0x3e), T(0x3f),
-  T(0x40), T(0x41), T(0x42), T(0x43), T(0x44), T(0x45), T(0x46), T(0x47),
-  T(0x48), T(0x49), T(0x4a), T(0x4b), T(0x4c), T(0x4d), T(0x4e), T(0x4f),
-  T(0x50), T(0x51), T(0x52), T(0x53), T(0x54), T(0x55), T(0x56), T(0x57),
-  T(0x58), T(0x59), T(0x5a), T(0x5b), T(0x5c), T(0x5d), T(0x5e), T(0x5f),
-  T(0x60), T(0x61), T(0x62), T(0x63), T(0x64), T(0x65), T(0x66), T(0x67),
-  T(0x68), T(0x69), T(0x6a), T(0x6b), T(0x6c), T(0x6d), T(0x6e), T(0x6f),
-  T(0x70), T(0x71), T(0x72), T(0x73), T(0x74), T(0x75), T(0x76), T(0x77),
-  T(0x78), T(0x79), T(0x7a), T(0x7b), T(0x7c), T(0x7d), T(0x7e), T(0x7f),
-  T(0x80), T(0x81), T(0x82), T(0x83), T(0x84), T(0x85), T(0x86), T(0x87),
-  T(0x88), T(0x89), T(0x8a), T(0x8b), T(0x8c), T(0x8d), T(0x8e), T(0x8f),
-  T(0x90), T(0x91), T(0x92), T(0x93), T(0x94), T(0x95), T(0x96), T(0x97),
-  T(0x98), T(0x99), T(0x9a), T(0x9b), T(0x9c), T(0x9d), T(0x9e), T(0x9f),
-  T(0xa0), T(0xa1), T(0xa2), T(0xa3), T(0xa4), T(0xa5), T(0xa6), T(0xa7),
-  T(0xa8), T(0xa9), T(0xaa), T(0xab), T(0xac), T(0xad), T(0xae), T(0xaf),
-  T(0xb0), T(0xb1), T(0xb2), T(0xb3), T(0xb4), T(0xb5), T(0xb6), T(0xb7),
-  T(0xb8), T(0xb9), T(0xba), T(0xbb), T(0xbc), T(0xbd), T(0xbe), T(0xbf),
-  T(0xc0), T(0xc1), T(0xc2), T(0xc3), T(0xc4), T(0xc5), T(0xc6), T(0xc7),
-  T(0xc8), T(0xc9), T(0xca), T(0xcb), T(0xcc), T(0xcd), T(0xce), T(0xcf),
-  T(0xd0), T(0xd1), T(0xd2), T(0xd3), T(0xd4), T(0xd5), T(0xd6), T(0xd7),
-  T(0xd8), T(0xd9), T(0xda), T(0xdb), T(0xdc), T(0xdd), T(0xde), T(0xdf),
-  T(0xe0), T(0xe1), T(0xe2), T(0xe3), T(0xe4), T(0xe5), T(0xe6), T(0xe7),
-  T(0xe8), T(0xe9), T(0xea), T(0xeb), T(0xec), T(0xed), T(0xee), T(0xef),
-  T(0xf0), T(0xf1), T(0xf2), T(0xf3), T(0xf4), T(0xf5), T(0xf6), T(0xf7),
-  T(0xf8), T(0xf9), T(0xfa), T(0xfb), T(0xfc), T(0xfd), T(0xfe), 0x01000100 };
-#undef T
-
-void ARGBUnattenuateRow_C(const uint8* src_argb, uint8* dst_argb, int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    uint32 b = src_argb[0];
-    uint32 g = src_argb[1];
-    uint32 r = src_argb[2];
-    const uint32 a = src_argb[3];
-    const uint32 ia = fixed_invtbl8[a] & 0xffff;  // 8.8 fixed point
-    b = (b * ia) >> 8;
-    g = (g * ia) >> 8;
-    r = (r * ia) >> 8;
-    // Clamping should not be necessary but is free in assembly.
-    dst_argb[0] = clamp255(b);
-    dst_argb[1] = clamp255(g);
-    dst_argb[2] = clamp255(r);
-    dst_argb[3] = a;
-    src_argb += 4;
-    dst_argb += 4;
-  }
-}
-
-void ComputeCumulativeSumRow_C(const uint8* row, int32* cumsum,
-                               const int32* previous_cumsum, int width) {
-  int32 row_sum[4] = {0, 0, 0, 0};
-  int x;
-  for (x = 0; x < width; ++x) {
-    row_sum[0] += row[x * 4 + 0];
-    row_sum[1] += row[x * 4 + 1];
-    row_sum[2] += row[x * 4 + 2];
-    row_sum[3] += row[x * 4 + 3];
-    cumsum[x * 4 + 0] = row_sum[0]  + previous_cumsum[x * 4 + 0];
-    cumsum[x * 4 + 1] = row_sum[1]  + previous_cumsum[x * 4 + 1];
-    cumsum[x * 4 + 2] = row_sum[2]  + previous_cumsum[x * 4 + 2];
-    cumsum[x * 4 + 3] = row_sum[3]  + previous_cumsum[x * 4 + 3];
-  }
-}
-
-void CumulativeSumToAverageRow_C(const int32* tl, const int32* bl,
-                                int w, int area, uint8* dst, int count) {
-  float ooa = 1.0f / area;
-  int i;
-  for (i = 0; i < count; ++i) {
-    dst[0] = (uint8)((bl[w + 0] + tl[0] - bl[0] - tl[w + 0]) * ooa);
-    dst[1] = (uint8)((bl[w + 1] + tl[1] - bl[1] - tl[w + 1]) * ooa);
-    dst[2] = (uint8)((bl[w + 2] + tl[2] - bl[2] - tl[w + 2]) * ooa);
-    dst[3] = (uint8)((bl[w + 3] + tl[3] - bl[3] - tl[w + 3]) * ooa);
-    dst += 4;
-    tl += 4;
-    bl += 4;
-  }
-}
-
-// Copy pixels from rotated source to destination row with a slope.
-LIBYUV_API
-void ARGBAffineRow_C(const uint8* src_argb, int src_argb_stride,
-                     uint8* dst_argb, const float* uv_dudv, int width) {
-  int i;
-  // Render a row of pixels from source into a buffer.
-  float uv[2];
-  uv[0] = uv_dudv[0];
-  uv[1] = uv_dudv[1];
-  for (i = 0; i < width; ++i) {
-    int x = (int)(uv[0]);
-    int y = (int)(uv[1]);
-    *(uint32*)(dst_argb) =
-        *(const uint32*)(src_argb + y * src_argb_stride +
-                                         x * 4);
-    dst_argb += 4;
-    uv[0] += uv_dudv[2];
-    uv[1] += uv_dudv[3];
-  }
-}
-
-// Blend 2 rows into 1 for conversions such as I422ToI420.
-void HalfRow_C(const uint8* src_uv, int src_uv_stride,
-               uint8* dst_uv, int pix) {
-  int x;
-  for (x = 0; x < pix; ++x) {
-    dst_uv[x] = (src_uv[x] + src_uv[src_uv_stride + x] + 1) >> 1;
-  }
-}
-
-// C version 2x2 -> 2x1.
-void InterpolateRow_C(uint8* dst_ptr, const uint8* src_ptr,
-                      ptrdiff_t src_stride,
-                      int width, int source_y_fraction) {
-  int y1_fraction = source_y_fraction;
-  int y0_fraction = 256 - y1_fraction;
-  const uint8* src_ptr1 = src_ptr + src_stride;
-  int x;
-  if (source_y_fraction == 0) {
-    memcpy(dst_ptr, src_ptr, width);
-    return;
-  }
-  if (source_y_fraction == 128) {
-    HalfRow_C(src_ptr, (int)(src_stride), dst_ptr, width);
-    return;
-  }
-  for (x = 0; x < width - 1; x += 2) {
-    dst_ptr[0] = (src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction) >> 8;
-    dst_ptr[1] = (src_ptr[1] * y0_fraction + src_ptr1[1] * y1_fraction) >> 8;
-    src_ptr += 2;
-    src_ptr1 += 2;
-    dst_ptr += 2;
-  }
-  if (width & 1) {
-    dst_ptr[0] = (src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction) >> 8;
-  }
-}
-
-// Select 2 channels from ARGB on alternating pixels.  e.g.  BGBGBGBG
-void ARGBToBayerRow_C(const uint8* src_argb,
-                      uint8* dst_bayer, uint32 selector, int pix) {
-  int index0 = selector & 0xff;
-  int index1 = (selector >> 8) & 0xff;
-  // Copy a row of Bayer.
-  int x;
-  for (x = 0; x < pix - 1; x += 2) {
-    dst_bayer[0] = src_argb[index0];
-    dst_bayer[1] = src_argb[index1];
-    src_argb += 8;
-    dst_bayer += 2;
-  }
-  if (pix & 1) {
-    dst_bayer[0] = src_argb[index0];
-  }
-}
-
-// Select G channel from ARGB.  e.g.  GGGGGGGG
-void ARGBToBayerGGRow_C(const uint8* src_argb,
-                        uint8* dst_bayer, uint32 selector, int pix) {
-  // Copy a row of G.
-  int x;
-  for (x = 0; x < pix - 1; x += 2) {
-    dst_bayer[0] = src_argb[1];
-    dst_bayer[1] = src_argb[5];
-    src_argb += 8;
-    dst_bayer += 2;
-  }
-  if (pix & 1) {
-    dst_bayer[0] = src_argb[1];
-  }
-}
-
-// Use first 4 shuffler values to reorder ARGB channels.
-void ARGBShuffleRow_C(const uint8* src_argb, uint8* dst_argb,
-                      const uint8* shuffler, int pix) {
-  int index0 = shuffler[0];
-  int index1 = shuffler[1];
-  int index2 = shuffler[2];
-  int index3 = shuffler[3];
-  // Shuffle a row of ARGB.
-  int x;
-  for (x = 0; x < pix; ++x) {
-    // To support in-place conversion.
-    uint8 b = src_argb[index0];
-    uint8 g = src_argb[index1];
-    uint8 r = src_argb[index2];
-    uint8 a = src_argb[index3];
-    dst_argb[0] = b;
-    dst_argb[1] = g;
-    dst_argb[2] = r;
-    dst_argb[3] = a;
-    src_argb += 4;
-    dst_argb += 4;
-  }
-}
-
-void I422ToYUY2Row_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_frame, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_frame[0] = src_y[0];
-    dst_frame[1] = src_u[0];
-    dst_frame[2] = src_y[1];
-    dst_frame[3] = src_v[0];
-    dst_frame += 4;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-  }
-  if (width & 1) {
-    dst_frame[0] = src_y[0];
-    dst_frame[1] = src_u[0];
-    dst_frame[2] = src_y[0];  // duplicate last y
-    dst_frame[3] = src_v[0];
-  }
-}
-
-void I422ToUYVYRow_C(const uint8* src_y,
-                     const uint8* src_u,
-                     const uint8* src_v,
-                     uint8* dst_frame, int width) {
-  int x;
-  for (x = 0; x < width - 1; x += 2) {
-    dst_frame[0] = src_u[0];
-    dst_frame[1] = src_y[0];
-    dst_frame[2] = src_v[0];
-    dst_frame[3] = src_y[1];
-    dst_frame += 4;
-    src_y += 2;
-    src_u += 1;
-    src_v += 1;
-  }
-  if (width & 1) {
-    dst_frame[0] = src_u[0];
-    dst_frame[1] = src_y[0];
-    dst_frame[2] = src_v[0];
-    dst_frame[3] = src_y[0];  // duplicate last y
-  }
-}
-
-#if !defined(LIBYUV_DISABLE_X86) && defined(HAS_I422TOARGBROW_SSSE3)
-// row_win.cc has asm version, but GCC uses 2 step wrapper.
-#if !defined(_MSC_VER) && (defined(__x86_64__) || defined(__i386__))
-void I422ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_u,
-                           const uint8* src_v,
-                           uint8* rgb_buf,
-                           int width) {
-  // Allocate a row of ARGB.
-  align_buffer_64(row, width * 4);
-  I422ToARGBRow_SSSE3(src_y, src_u, src_v, row, width);
-  ARGBToRGB565Row_SSE2(row, rgb_buf, width);
-  free_aligned_buffer_64(row);
-}
-#endif  // !defined(_MSC_VER) && (defined(__x86_64__) || defined(__i386__))
-
-#if defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)
-void I422ToARGB1555Row_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* rgb_buf,
-                             int width) {
-  // Allocate a row of ARGB.
-  align_buffer_64(row, width * 4);
-  I422ToARGBRow_SSSE3(src_y, src_u, src_v, row, width);
-  ARGBToARGB1555Row_SSE2(row, rgb_buf, width);
-  free_aligned_buffer_64(row);
-}
-
-void I422ToARGB4444Row_SSSE3(const uint8* src_y,
-                             const uint8* src_u,
-                             const uint8* src_v,
-                             uint8* rgb_buf,
-                             int width) {
-  // Allocate a row of ARGB.
-  align_buffer_64(row, width * 4);
-  I422ToARGBRow_SSSE3(src_y, src_u, src_v, row, width);
-  ARGBToARGB4444Row_SSE2(row, rgb_buf, width);
-  free_aligned_buffer_64(row);
-}
-
-void NV12ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_uv,
-                           uint8* dst_rgb565,
-                           int width) {
-  // Allocate a row of ARGB.
-  align_buffer_64(row, width * 4);
-  NV12ToARGBRow_SSSE3(src_y, src_uv, row, width);
-  ARGBToRGB565Row_SSE2(row, dst_rgb565, width);
-  free_aligned_buffer_64(row);
-}
-
-void NV21ToRGB565Row_SSSE3(const uint8* src_y,
-                           const uint8* src_vu,
-                           uint8* dst_rgb565,
-                           int width) {
-  // Allocate a row of ARGB.
-  align_buffer_64(row, width * 4);
-  NV21ToARGBRow_SSSE3(src_y, src_vu, row, width);
-  ARGBToRGB565Row_SSE2(row, dst_rgb565, width);
-  free_aligned_buffer_64(row);
-}
-
-void YUY2ToARGBRow_SSSE3(const uint8* src_yuy2,
-                         uint8* dst_argb,
-                         int width) {
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-  YUY2ToUV422Row_SSE2(src_yuy2, row_u, row_v, width);
-  YUY2ToYRow_SSE2(src_yuy2, row_y, width);
-  I422ToARGBRow_SSSE3(row_y, row_u, row_v, dst_argb, width);
-  free_aligned_buffer_64(row_y);
-}
-
-void YUY2ToARGBRow_Unaligned_SSSE3(const uint8* src_yuy2,
-                                   uint8* dst_argb,
-                                   int width) {
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-  YUY2ToUV422Row_Unaligned_SSE2(src_yuy2, row_u, row_v, width);
-  YUY2ToYRow_Unaligned_SSE2(src_yuy2, row_y, width);
-  I422ToARGBRow_Unaligned_SSSE3(row_y, row_u, row_v, dst_argb, width);
-  free_aligned_buffer_64(row_y);
-}
-
-void UYVYToARGBRow_SSSE3(const uint8* src_uyvy,
-                         uint8* dst_argb,
-                         int width) {
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-  UYVYToUV422Row_SSE2(src_uyvy, row_u, row_v, width);
-  UYVYToYRow_SSE2(src_uyvy, row_y, width);
-  I422ToARGBRow_SSSE3(row_y, row_u, row_v, dst_argb, width);
-  free_aligned_buffer_64(row_y);
-}
-
-void UYVYToARGBRow_Unaligned_SSSE3(const uint8* src_uyvy,
-                                   uint8* dst_argb,
-                                   int width) {
-  // Allocate a rows of yuv.
-  align_buffer_64(row_y, ((width + 63) & ~63) * 2);
-  uint8* row_u = row_y + ((width + 63) & ~63);
-  uint8* row_v = row_u + ((width + 63) & ~63) / 2;
-  UYVYToUV422Row_Unaligned_SSE2(src_uyvy, row_u, row_v, width);
-  UYVYToYRow_Unaligned_SSE2(src_uyvy, row_y, width);
-  I422ToARGBRow_Unaligned_SSSE3(row_y, row_u, row_v, dst_argb, width);
-  free_aligned_buffer_64(row_y);
-}
-
-#endif  // defined(_M_IX86) || defined(__x86_64__) || defined(__i386__)
-#endif  // !defined(LIBYUV_DISABLE_X86)
-
-void ARGBPolynomialRow_C(const uint8* src_argb,
-                         uint8* dst_argb, const float* poly,
-                         int width) {
-  int i;
-  for (i = 0; i < width; ++i) {
-    float b = (float)(src_argb[0]);
-    float g = (float)(src_argb[1]);
-    float r = (float)(src_argb[2]);
-    float a = (float)(src_argb[3]);
-    float b2 = b * b;
-    float g2 = g * g;
-    float r2 = r * r;
-    float a2 = a * a;
-    float db = poly[0] + poly[4] * b;
-    float dg = poly[1] + poly[5] * g;
-    float dr = poly[2] + poly[6] * r;
-    float da = poly[3] + poly[7] * a;
-    float b3 = b2 * b;
-    float g3 = g2 * g;
-    float r3 = r2 * r;
-    float a3 = a2 * a;
-    db += poly[8] * b2;
-    dg += poly[9] * g2;
-    dr += poly[10] * r2;
-    da += poly[11] * a2;
-    db += poly[12] * b3;
-    dg += poly[13] * g3;
-    dr += poly[14] * r3;
-    da += poly[15] * a3;
-
-    dst_argb[0] = Clamp((int32)(db));
-    dst_argb[1] = Clamp((int32)(dg));
-    dst_argb[2] = Clamp((int32)(dr));
-    dst_argb[3] = Clamp((int32)(da));
-    src_argb += 4;
-    dst_argb += 4;
-  }
-}
-
-void ARGBLumaColorTableRow_C(const uint8* src_argb, uint8* dst_argb, int width,
-                             const uint8* luma, uint32 lumacoeff) {
-  uint32 bc = lumacoeff & 0xff;
-  uint32 gc = (lumacoeff >> 8) & 0xff;
-  uint32 rc = (lumacoeff >> 16) & 0xff;
-
-  int i;
-  for (i = 0; i < width - 1; i += 2) {
-    // Luminance in rows, color values in columns.
-    const uint8* luma0 = ((src_argb[0] * bc + src_argb[1] * gc +
-                           src_argb[2] * rc) & 0x7F00u) + luma;
-    const uint8* luma1;
-    dst_argb[0] = luma0[src_argb[0]];
-    dst_argb[1] = luma0[src_argb[1]];
-    dst_argb[2] = luma0[src_argb[2]];
-    dst_argb[3] = src_argb[3];
-    luma1 = ((src_argb[4] * bc + src_argb[5] * gc +
-              src_argb[6] * rc) & 0x7F00u) + luma;
-    dst_argb[4] = luma1[src_argb[4]];
-    dst_argb[5] = luma1[src_argb[5]];
-    dst_argb[6] = luma1[src_argb[6]];
-    dst_argb[7] = src_argb[7];
-    src_argb += 8;
-    dst_argb += 8;
-  }
-  if (width & 1) {
-    // Luminance in rows, color values in columns.
-    const uint8* luma0 = ((src_argb[0] * bc + src_argb[1] * gc +
-                           src_argb[2] * rc) & 0x7F00u) + luma;
-    dst_argb[0] = luma0[src_argb[0]];
-    dst_argb[1] = luma0[src_argb[1]];
-    dst_argb[2] = luma0[src_argb[2]];
-    dst_argb[3] = src_argb[3];
-  }
-}
-
-void ARGBCopyAlphaRow_C(const uint8* src, uint8* dst, int width) {
-  int i;
-  for (i = 0; i < width - 1; i += 2) {
-    dst[3] = src[3];
-    dst[7] = src[7];
-    dst += 8;
-    src += 8;
-  }
-  if (width & 1) {
-    dst[3] = src[3];
-  }
-}
-
-void ARGBCopyYToAlphaRow_C(const uint8* src, uint8* dst, int width) {
-  int i;
-  for (i = 0; i < width - 1; i += 2) {
-    dst[3] = src[0];
-    dst[7] = src[1];
-    dst += 8;
-    src += 2;
-  }
-  if (width & 1) {
-    dst[3] = src[0];
-  }
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_mips.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_mips.cc
deleted file mode 100755
index 4435c55c5c..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_mips.cc
+++ /dev/null
@@ -1,991 +0,0 @@
-/*
- *  Copyright (c) 2012 The LibYuv project authors. All Rights Reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// The following are available on Mips platforms:
-#if !defined(LIBYUV_DISABLE_MIPS) && defined(__mips__)
-
-#ifdef HAS_COPYROW_MIPS
-void CopyRow_MIPS(const uint8* src, uint8* dst, int count) {
-  __asm__ __volatile__ (
-    ".set      noreorder                         \n"
-    ".set      noat                              \n"
-    "slti      $at, %[count], 8                  \n"
-    "bne       $at ,$zero, $last8                \n"
-    "xor       $t8, %[src], %[dst]               \n"
-    "andi      $t8, $t8, 0x3                     \n"
-
-    "bne       $t8, $zero, unaligned             \n"
-    "negu      $a3, %[dst]                       \n"
-    // make dst/src aligned
-    "andi      $a3, $a3, 0x3                     \n"
-    "beq       $a3, $zero, $chk16w               \n"
-    // word-aligned now count is the remining bytes count
-    "subu     %[count], %[count], $a3            \n"
-
-    "lwr       $t8, 0(%[src])                    \n"
-    "addu      %[src], %[src], $a3               \n"
-    "swr       $t8, 0(%[dst])                    \n"
-    "addu      %[dst], %[dst], $a3               \n"
-
-    // Now the dst/src are mutually word-aligned with word-aligned addresses
-    "$chk16w:                                    \n"
-    "andi      $t8, %[count], 0x3f               \n"  // whole 64-B chunks?
-    // t8 is the byte count after 64-byte chunks
-    "beq       %[count], $t8, chk8w              \n"
-    // There will be at most 1 32-byte chunk after it
-    "subu      $a3, %[count], $t8                \n"  // the reminder
-    // Here a3 counts bytes in 16w chunks
-    "addu      $a3, %[dst], $a3                  \n"
-    // Now a3 is the final dst after 64-byte chunks
-    "addu      $t0, %[dst], %[count]             \n"
-    // t0 is the "past the end" address
-
-    // When in the loop we exercise "pref 30,x(a1)", the a1+x should not be past
-    // the "t0-32" address
-    // This means: for x=128 the last "safe" a1 address is "t0-160"
-    // Alternatively, for x=64 the last "safe" a1 address is "t0-96"
-    // we will use "pref 30,128(a1)", so "t0-160" is the limit
-    "subu      $t9, $t0, 160                     \n"
-    // t9 is the "last safe pref 30,128(a1)" address
-    "pref      0, 0(%[src])                      \n"  // first line of src
-    "pref      0, 32(%[src])                     \n"  // second line of src
-    "pref      0, 64(%[src])                     \n"
-    "pref      30, 32(%[dst])                    \n"
-    // In case the a1 > t9 don't use "pref 30" at all
-    "sgtu      $v1, %[dst], $t9                  \n"
-    "bgtz      $v1, $loop16w                     \n"
-    "nop                                         \n"
-    // otherwise, start with using pref30
-    "pref      30, 64(%[dst])                    \n"
-    "$loop16w:                                    \n"
-    "pref      0, 96(%[src])                     \n"
-    "lw        $t0, 0(%[src])                    \n"
-    "bgtz      $v1, $skip_pref30_96              \n"  // skip
-    "lw        $t1, 4(%[src])                    \n"
-    "pref      30, 96(%[dst])                    \n"  // continue
-    "$skip_pref30_96:                            \n"
-    "lw        $t2, 8(%[src])                    \n"
-    "lw        $t3, 12(%[src])                   \n"
-    "lw        $t4, 16(%[src])                   \n"
-    "lw        $t5, 20(%[src])                   \n"
-    "lw        $t6, 24(%[src])                   \n"
-    "lw        $t7, 28(%[src])                   \n"
-    "pref      0, 128(%[src])                    \n"
-    //  bring the next lines of src, addr 128
-    "sw        $t0, 0(%[dst])                    \n"
-    "sw        $t1, 4(%[dst])                    \n"
-    "sw        $t2, 8(%[dst])                    \n"
-    "sw        $t3, 12(%[dst])                   \n"
-    "sw        $t4, 16(%[dst])                   \n"
-    "sw        $t5, 20(%[dst])                   \n"
-    "sw        $t6, 24(%[dst])                   \n"
-    "sw        $t7, 28(%[dst])                   \n"
-    "lw        $t0, 32(%[src])                   \n"
-    "bgtz      $v1, $skip_pref30_128             \n"  // skip pref 30,128(a1)
-    "lw        $t1, 36(%[src])                   \n"
-    "pref      30, 128(%[dst])                   \n"  // set dest, addr 128
-    "$skip_pref30_128:                           \n"
-    "lw        $t2, 40(%[src])                   \n"
-    "lw        $t3, 44(%[src])                   \n"
-    "lw        $t4, 48(%[src])                   \n"
-    "lw        $t5, 52(%[src])                   \n"
-    "lw        $t6, 56(%[src])                   \n"
-    "lw        $t7, 60(%[src])                   \n"
-    "pref      0, 160(%[src])                    \n"
-    // bring the next lines of src, addr 160
-    "sw        $t0, 32(%[dst])                   \n"
-    "sw        $t1, 36(%[dst])                   \n"
-    "sw        $t2, 40(%[dst])                   \n"
-    "sw        $t3, 44(%[dst])                   \n"
-    "sw        $t4, 48(%[dst])                   \n"
-    "sw        $t5, 52(%[dst])                   \n"
-    "sw        $t6, 56(%[dst])                   \n"
-    "sw        $t7, 60(%[dst])                   \n"
-
-    "addiu     %[dst], %[dst], 64                \n"  // adding 64 to dest
-    "sgtu      $v1, %[dst], $t9                  \n"
-    "bne       %[dst], $a3, $loop16w             \n"
-    " addiu    %[src], %[src], 64                \n"  // adding 64 to src
-    "move      %[count], $t8                     \n"
-
-    // Here we have src and dest word-aligned but less than 64-bytes to go
-
-    "chk8w:                                      \n"
-    "pref      0, 0x0(%[src])                    \n"
-    "andi      $t8, %[count], 0x1f               \n"  // 32-byte chunk?
-    // the t8 is the reminder count past 32-bytes
-    "beq       %[count], $t8, chk1w              \n"
-    // count=t8,no 32-byte chunk
-    " nop                                        \n"
-
-    "lw        $t0, 0(%[src])                    \n"
-    "lw        $t1, 4(%[src])                    \n"
-    "lw        $t2, 8(%[src])                    \n"
-    "lw        $t3, 12(%[src])                   \n"
-    "lw        $t4, 16(%[src])                   \n"
-    "lw        $t5, 20(%[src])                   \n"
-    "lw        $t6, 24(%[src])                   \n"
-    "lw        $t7, 28(%[src])                   \n"
-    "addiu     %[src], %[src], 32                \n"
-
-    "sw        $t0, 0(%[dst])                    \n"
-    "sw        $t1, 4(%[dst])                    \n"
-    "sw        $t2, 8(%[dst])                    \n"
-    "sw        $t3, 12(%[dst])                   \n"
-    "sw        $t4, 16(%[dst])                   \n"
-    "sw        $t5, 20(%[dst])                   \n"
-    "sw        $t6, 24(%[dst])                   \n"
-    "sw        $t7, 28(%[dst])                   \n"
-    "addiu     %[dst], %[dst], 32                \n"
-
-    "chk1w:                                      \n"
-    "andi      %[count], $t8, 0x3                \n"
-    // now count is the reminder past 1w chunks
-    "beq       %[count], $t8, $last8             \n"
-    " subu     $a3, $t8, %[count]                \n"
-    // a3 is count of bytes in 1w chunks
-    "addu      $a3, %[dst], $a3                  \n"
-    // now a3 is the dst address past the 1w chunks
-    // copying in words (4-byte chunks)
-    "$wordCopy_loop:                             \n"
-    "lw        $t3, 0(%[src])                    \n"
-    // the first t3 may be equal t0 ... optimize?
-    "addiu     %[src], %[src],4                  \n"
-    "addiu     %[dst], %[dst],4                  \n"
-    "bne       %[dst], $a3,$wordCopy_loop        \n"
-    " sw       $t3, -4(%[dst])                   \n"
-
-    // For the last (<8) bytes
-    "$last8:                                     \n"
-    "blez      %[count], leave                   \n"
-    " addu     $a3, %[dst], %[count]             \n"  // a3 -last dst address
-    "$last8loop:                                 \n"
-    "lb        $v1, 0(%[src])                    \n"
-    "addiu     %[src], %[src], 1                 \n"
-    "addiu     %[dst], %[dst], 1                 \n"
-    "bne       %[dst], $a3, $last8loop           \n"
-    " sb       $v1, -1(%[dst])                   \n"
-
-    "leave:                                      \n"
-    "  j       $ra                               \n"
-    "  nop                                       \n"
-
-    //
-    // UNALIGNED case
-    //
-
-    "unaligned:                                  \n"
-    // got here with a3="negu a1"
-    "andi      $a3, $a3, 0x3                     \n"  // a1 is word aligned?
-    "beqz      $a3, $ua_chk16w                   \n"
-    " subu     %[count], %[count], $a3           \n"
-    // bytes left after initial a3 bytes
-    "lwr       $v1, 0(%[src])                    \n"
-    "lwl       $v1, 3(%[src])                    \n"
-    "addu      %[src], %[src], $a3               \n"  // a3 may be 1, 2 or 3
-    "swr       $v1, 0(%[dst])                    \n"
-    "addu      %[dst], %[dst], $a3               \n"
-    // below the dst will be word aligned (NOTE1)
-    "$ua_chk16w:                                 \n"
-    "andi      $t8, %[count], 0x3f               \n"  // whole 64-B chunks?
-    // t8 is the byte count after 64-byte chunks
-    "beq       %[count], $t8, ua_chk8w           \n"
-    // if a2==t8, no 64-byte chunks
-    // There will be at most 1 32-byte chunk after it
-    "subu      $a3, %[count], $t8                \n"  // the reminder
-    // Here a3 counts bytes in 16w chunks
-    "addu      $a3, %[dst], $a3                  \n"
-    // Now a3 is the final dst after 64-byte chunks
-    "addu      $t0, %[dst], %[count]             \n"  // t0 "past the end"
-    "subu      $t9, $t0, 160                     \n"
-    // t9 is the "last safe pref 30,128(a1)" address
-    "pref      0, 0(%[src])                      \n"  // first line of src
-    "pref      0, 32(%[src])                     \n"  // second line  addr 32
-    "pref      0, 64(%[src])                     \n"
-    "pref      30, 32(%[dst])                    \n"
-    // safe, as we have at least 64 bytes ahead
-    // In case the a1 > t9 don't use "pref 30" at all
-    "sgtu      $v1, %[dst], $t9                  \n"
-    "bgtz      $v1, $ua_loop16w                  \n"
-    // skip "pref 30,64(a1)" for too short arrays
-    " nop                                        \n"
-    // otherwise, start with using pref30
-    "pref      30, 64(%[dst])                    \n"
-    "$ua_loop16w:                                \n"
-    "pref      0, 96(%[src])                     \n"
-    "lwr       $t0, 0(%[src])                    \n"
-    "lwl       $t0, 3(%[src])                    \n"
-    "lwr       $t1, 4(%[src])                    \n"
-    "bgtz      $v1, $ua_skip_pref30_96           \n"
-    " lwl      $t1, 7(%[src])                    \n"
-    "pref      30, 96(%[dst])                    \n"
-    // continue setting up the dest, addr 96
-    "$ua_skip_pref30_96:                         \n"
-    "lwr       $t2, 8(%[src])                    \n"
-    "lwl       $t2, 11(%[src])                   \n"
-    "lwr       $t3, 12(%[src])                   \n"
-    "lwl       $t3, 15(%[src])                   \n"
-    "lwr       $t4, 16(%[src])                   \n"
-    "lwl       $t4, 19(%[src])                   \n"
-    "lwr       $t5, 20(%[src])                   \n"
-    "lwl       $t5, 23(%[src])                   \n"
-    "lwr       $t6, 24(%[src])                   \n"
-    "lwl       $t6, 27(%[src])                   \n"
-    "lwr       $t7, 28(%[src])                   \n"
-    "lwl       $t7, 31(%[src])                   \n"
-    "pref      0, 128(%[src])                    \n"
-    // bring the next lines of src, addr 128
-    "sw        $t0, 0(%[dst])                    \n"
-    "sw        $t1, 4(%[dst])                    \n"
-    "sw        $t2, 8(%[dst])                    \n"
-    "sw        $t3, 12(%[dst])                   \n"
-    "sw        $t4, 16(%[dst])                   \n"
-    "sw        $t5, 20(%[dst])                   \n"
-    "sw        $t6, 24(%[dst])                   \n"
-    "sw        $t7, 28(%[dst])                   \n"
-    "lwr       $t0, 32(%[src])                   \n"
-    "lwl       $t0, 35(%[src])                   \n"
-    "lwr       $t1, 36(%[src])                   \n"
-    "bgtz      $v1, ua_skip_pref30_128           \n"
-    " lwl      $t1, 39(%[src])                   \n"
-    "pref      30, 128(%[dst])                   \n"
-    // continue setting up the dest, addr 128
-    "ua_skip_pref30_128:                         \n"
-
-    "lwr       $t2, 40(%[src])                   \n"
-    "lwl       $t2, 43(%[src])                   \n"
-    "lwr       $t3, 44(%[src])                   \n"
-    "lwl       $t3, 47(%[src])                   \n"
-    "lwr       $t4, 48(%[src])                   \n"
-    "lwl       $t4, 51(%[src])                   \n"
-    "lwr       $t5, 52(%[src])                   \n"
-    "lwl       $t5, 55(%[src])                   \n"
-    "lwr       $t6, 56(%[src])                   \n"
-    "lwl       $t6, 59(%[src])                   \n"
-    "lwr       $t7, 60(%[src])                   \n"
-    "lwl       $t7, 63(%[src])                   \n"
-    "pref      0, 160(%[src])                    \n"
-    // bring the next lines of src, addr 160
-    "sw        $t0, 32(%[dst])                   \n"
-    "sw        $t1, 36(%[dst])                   \n"
-    "sw        $t2, 40(%[dst])                   \n"
-    "sw        $t3, 44(%[dst])                   \n"
-    "sw        $t4, 48(%[dst])                   \n"
-    "sw        $t5, 52(%[dst])                   \n"
-    "sw        $t6, 56(%[dst])                   \n"
-    "sw        $t7, 60(%[dst])                   \n"
-
-    "addiu     %[dst],%[dst],64                  \n"  // adding 64 to dest
-    "sgtu      $v1,%[dst],$t9                    \n"
-    "bne       %[dst],$a3,$ua_loop16w            \n"
-    " addiu    %[src],%[src],64                  \n"  // adding 64 to src
-    "move      %[count],$t8                      \n"
-
-    // Here we have src and dest word-aligned but less than 64-bytes to go
-
-    "ua_chk8w:                                   \n"
-    "pref      0, 0x0(%[src])                    \n"
-    "andi      $t8, %[count], 0x1f               \n"  // 32-byte chunk?
-    // the t8 is the reminder count
-    "beq       %[count], $t8, $ua_chk1w          \n"
-    // when count==t8, no 32-byte chunk
-
-    "lwr       $t0, 0(%[src])                    \n"
-    "lwl       $t0, 3(%[src])                    \n"
-    "lwr       $t1, 4(%[src])                    \n"
-    "lwl       $t1, 7(%[src])                    \n"
-    "lwr       $t2, 8(%[src])                    \n"
-    "lwl       $t2, 11(%[src])                   \n"
-    "lwr       $t3, 12(%[src])                   \n"
-    "lwl       $t3, 15(%[src])                   \n"
-    "lwr       $t4, 16(%[src])                   \n"
-    "lwl       $t4, 19(%[src])                   \n"
-    "lwr       $t5, 20(%[src])                   \n"
-    "lwl       $t5, 23(%[src])                   \n"
-    "lwr       $t6, 24(%[src])                   \n"
-    "lwl       $t6, 27(%[src])                   \n"
-    "lwr       $t7, 28(%[src])                   \n"
-    "lwl       $t7, 31(%[src])                   \n"
-    "addiu     %[src], %[src], 32                \n"
-
-    "sw        $t0, 0(%[dst])                    \n"
-    "sw        $t1, 4(%[dst])                    \n"
-    "sw        $t2, 8(%[dst])                    \n"
-    "sw        $t3, 12(%[dst])                   \n"
-    "sw        $t4, 16(%[dst])                   \n"
-    "sw        $t5, 20(%[dst])                   \n"
-    "sw        $t6, 24(%[dst])                   \n"
-    "sw        $t7, 28(%[dst])                   \n"
-    "addiu     %[dst], %[dst], 32                \n"
-
-    "$ua_chk1w:                                  \n"
-    "andi      %[count], $t8, 0x3                \n"
-    // now count is the reminder past 1w chunks
-    "beq       %[count], $t8, ua_smallCopy       \n"
-    "subu      $a3, $t8, %[count]                \n"
-    // a3 is count of bytes in 1w chunks
-    "addu      $a3, %[dst], $a3                  \n"
-    // now a3 is the dst address past the 1w chunks
-
-    // copying in words (4-byte chunks)
-    "$ua_wordCopy_loop:                          \n"
-    "lwr       $v1, 0(%[src])                    \n"
-    "lwl       $v1, 3(%[src])                    \n"
-    "addiu     %[src], %[src], 4                 \n"
-    "addiu     %[dst], %[dst], 4                 \n"
-    // note: dst=a1 is word aligned here, see NOTE1
-    "bne       %[dst], $a3, $ua_wordCopy_loop    \n"
-    " sw       $v1,-4(%[dst])                    \n"
-
-    // Now less than 4 bytes (value in count) left to copy
-    "ua_smallCopy:                               \n"
-    "beqz      %[count], leave                   \n"
-    " addu     $a3, %[dst], %[count]             \n" // a3 = last dst address
-    "$ua_smallCopy_loop:                         \n"
-    "lb        $v1, 0(%[src])                    \n"
-    "addiu     %[src], %[src], 1                 \n"
-    "addiu     %[dst], %[dst], 1                 \n"
-    "bne       %[dst],$a3,$ua_smallCopy_loop     \n"
-    " sb       $v1, -1(%[dst])                   \n"
-
-    "j         $ra                               \n"
-    " nop                                        \n"
-    ".set      at                                \n"
-    ".set      reorder                           \n"
-       : [dst] "+r" (dst), [src] "+r" (src)
-       : [count] "r" (count)
-       : "t0", "t1", "t2", "t3", "t4", "t5", "t6", "t7",
-       "t8", "t9", "a3", "v1", "at"
-  );
-}
-#endif  // HAS_COPYROW_MIPS
-
-// MIPS DSPR2 functions
-#if !defined(LIBYUV_DISABLE_MIPS) && defined(__mips_dsp) && \
-    (__mips_dsp_rev >= 2)
-void SplitUVRow_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                           int width) {
-  __asm__ __volatile__ (
-    ".set push                                     \n"
-    ".set noreorder                                \n"
-    "srl             $t4, %[width], 4              \n"  // multiplies of 16
-    "blez            $t4, 2f                       \n"
-    " andi           %[width], %[width], 0xf       \n"  // residual
-
-    ".p2align        2                             \n"
-  "1:                                              \n"
-    "addiu           $t4, $t4, -1                  \n"
-    "lw              $t0, 0(%[src_uv])             \n"  // V1 | U1 | V0 | U0
-    "lw              $t1, 4(%[src_uv])             \n"  // V3 | U3 | V2 | U2
-    "lw              $t2, 8(%[src_uv])             \n"  // V5 | U5 | V4 | U4
-    "lw              $t3, 12(%[src_uv])            \n"  // V7 | U7 | V6 | U6
-    "lw              $t5, 16(%[src_uv])            \n"  // V9 | U9 | V8 | U8
-    "lw              $t6, 20(%[src_uv])            \n"  // V11 | U11 | V10 | U10
-    "lw              $t7, 24(%[src_uv])            \n"  // V13 | U13 | V12 | U12
-    "lw              $t8, 28(%[src_uv])            \n"  // V15 | U15 | V14 | U14
-    "addiu           %[src_uv], %[src_uv], 32      \n"
-    "precrq.qb.ph    $t9, $t1, $t0                 \n"  // V3 | V2 | V1 | V0
-    "precr.qb.ph     $t0, $t1, $t0                 \n"  // U3 | U2 | U1 | U0
-    "precrq.qb.ph    $t1, $t3, $t2                 \n"  // V7 | V6 | V5 | V4
-    "precr.qb.ph     $t2, $t3, $t2                 \n"  // U7 | U6 | U5 | U4
-    "precrq.qb.ph    $t3, $t6, $t5                 \n"  // V11 | V10 | V9 | V8
-    "precr.qb.ph     $t5, $t6, $t5                 \n"  // U11 | U10 | U9 | U8
-    "precrq.qb.ph    $t6, $t8, $t7                 \n"  // V15 | V14 | V13 | V12
-    "precr.qb.ph     $t7, $t8, $t7                 \n"  // U15 | U14 | U13 | U12
-    "sw              $t9, 0(%[dst_v])              \n"
-    "sw              $t0, 0(%[dst_u])              \n"
-    "sw              $t1, 4(%[dst_v])              \n"
-    "sw              $t2, 4(%[dst_u])              \n"
-    "sw              $t3, 8(%[dst_v])              \n"
-    "sw              $t5, 8(%[dst_u])              \n"
-    "sw              $t6, 12(%[dst_v])             \n"
-    "sw              $t7, 12(%[dst_u])             \n"
-    "addiu           %[dst_v], %[dst_v], 16        \n"
-    "bgtz            $t4, 1b                       \n"
-    " addiu          %[dst_u], %[dst_u], 16        \n"
-
-    "beqz            %[width], 3f                  \n"
-    " nop                                          \n"
-
-  "2:                                              \n"
-    "lbu             $t0, 0(%[src_uv])             \n"
-    "lbu             $t1, 1(%[src_uv])             \n"
-    "addiu           %[src_uv], %[src_uv], 2       \n"
-    "addiu           %[width], %[width], -1        \n"
-    "sb              $t0, 0(%[dst_u])              \n"
-    "sb              $t1, 0(%[dst_v])              \n"
-    "addiu           %[dst_u], %[dst_u], 1         \n"
-    "bgtz            %[width], 2b                  \n"
-    " addiu          %[dst_v], %[dst_v], 1         \n"
-
-  "3:                                              \n"
-    ".set pop                                      \n"
-     : [src_uv] "+r" (src_uv),
-       [width] "+r" (width),
-       [dst_u] "+r" (dst_u),
-       [dst_v] "+r" (dst_v)
-     :
-     : "t0", "t1", "t2", "t3",
-     "t4", "t5", "t6", "t7", "t8", "t9"
-  );
-}
-
-void SplitUVRow_Unaligned_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u,
-                                     uint8* dst_v, int width) {
-  __asm__ __volatile__ (
-    ".set push                                     \n"
-    ".set noreorder                                \n"
-    "srl             $t4, %[width], 4              \n"  // multiplies of 16
-    "blez            $t4, 2f                       \n"
-    " andi           %[width], %[width], 0xf       \n"  // residual
-
-    ".p2align        2                             \n"
-  "1:                                              \n"
-    "addiu           $t4, $t4, -1                  \n"
-    "lwr             $t0, 0(%[src_uv])             \n"
-    "lwl             $t0, 3(%[src_uv])             \n"  // V1 | U1 | V0 | U0
-    "lwr             $t1, 4(%[src_uv])             \n"
-    "lwl             $t1, 7(%[src_uv])             \n"  // V3 | U3 | V2 | U2
-    "lwr             $t2, 8(%[src_uv])             \n"
-    "lwl             $t2, 11(%[src_uv])            \n"  // V5 | U5 | V4 | U4
-    "lwr             $t3, 12(%[src_uv])            \n"
-    "lwl             $t3, 15(%[src_uv])            \n"  // V7 | U7 | V6 | U6
-    "lwr             $t5, 16(%[src_uv])            \n"
-    "lwl             $t5, 19(%[src_uv])            \n"  // V9 | U9 | V8 | U8
-    "lwr             $t6, 20(%[src_uv])            \n"
-    "lwl             $t6, 23(%[src_uv])            \n"  // V11 | U11 | V10 | U10
-    "lwr             $t7, 24(%[src_uv])            \n"
-    "lwl             $t7, 27(%[src_uv])            \n"  // V13 | U13 | V12 | U12
-    "lwr             $t8, 28(%[src_uv])            \n"
-    "lwl             $t8, 31(%[src_uv])            \n"  // V15 | U15 | V14 | U14
-    "precrq.qb.ph    $t9, $t1, $t0                 \n"  // V3 | V2 | V1 | V0
-    "precr.qb.ph     $t0, $t1, $t0                 \n"  // U3 | U2 | U1 | U0
-    "precrq.qb.ph    $t1, $t3, $t2                 \n"  // V7 | V6 | V5 | V4
-    "precr.qb.ph     $t2, $t3, $t2                 \n"  // U7 | U6 | U5 | U4
-    "precrq.qb.ph    $t3, $t6, $t5                 \n"  // V11 | V10 | V9 | V8
-    "precr.qb.ph     $t5, $t6, $t5                 \n"  // U11 | U10 | U9 | U8
-    "precrq.qb.ph    $t6, $t8, $t7                 \n"  // V15 | V14 | V13 | V12
-    "precr.qb.ph     $t7, $t8, $t7                 \n"  // U15 | U14 | U13 | U12
-    "addiu           %[src_uv], %[src_uv], 32      \n"
-    "swr             $t9, 0(%[dst_v])              \n"
-    "swl             $t9, 3(%[dst_v])              \n"
-    "swr             $t0, 0(%[dst_u])              \n"
-    "swl             $t0, 3(%[dst_u])              \n"
-    "swr             $t1, 4(%[dst_v])              \n"
-    "swl             $t1, 7(%[dst_v])              \n"
-    "swr             $t2, 4(%[dst_u])              \n"
-    "swl             $t2, 7(%[dst_u])              \n"
-    "swr             $t3, 8(%[dst_v])              \n"
-    "swl             $t3, 11(%[dst_v])             \n"
-    "swr             $t5, 8(%[dst_u])              \n"
-    "swl             $t5, 11(%[dst_u])             \n"
-    "swr             $t6, 12(%[dst_v])             \n"
-    "swl             $t6, 15(%[dst_v])             \n"
-    "swr             $t7, 12(%[dst_u])             \n"
-    "swl             $t7, 15(%[dst_u])             \n"
-    "addiu           %[dst_u], %[dst_u], 16        \n"
-    "bgtz            $t4, 1b                       \n"
-    " addiu          %[dst_v], %[dst_v], 16        \n"
-
-    "beqz            %[width], 3f                  \n"
-    " nop                                          \n"
-
-  "2:                                              \n"
-    "lbu             $t0, 0(%[src_uv])             \n"
-    "lbu             $t1, 1(%[src_uv])             \n"
-    "addiu           %[src_uv], %[src_uv], 2       \n"
-    "addiu           %[width], %[width], -1        \n"
-    "sb              $t0, 0(%[dst_u])              \n"
-    "sb              $t1, 0(%[dst_v])              \n"
-    "addiu           %[dst_u], %[dst_u], 1         \n"
-    "bgtz            %[width], 2b                  \n"
-    " addiu          %[dst_v], %[dst_v], 1         \n"
-
-  "3:                                              \n"
-    ".set pop                                      \n"
-     : [src_uv] "+r" (src_uv),
-       [width] "+r" (width),
-       [dst_u] "+r" (dst_u),
-       [dst_v] "+r" (dst_v)
-     :
-     : "t0", "t1", "t2", "t3",
-     "t4", "t5", "t6", "t7", "t8", "t9"
-  );
-}
-
-void MirrorRow_MIPS_DSPR2(const uint8* src, uint8* dst, int width) {
-  __asm__ __volatile__ (
-    ".set push                             \n"
-    ".set noreorder                        \n"
-
-    "srl       $t4, %[width], 4            \n"  // multiplies of 16
-    "andi      $t5, %[width], 0xf          \n"
-    "blez      $t4, 2f                     \n"
-    " addu     %[src], %[src], %[width]    \n"  // src += width
-
-    ".p2align  2                           \n"
-   "1:                                     \n"
-    "lw        $t0, -16(%[src])            \n"  // |3|2|1|0|
-    "lw        $t1, -12(%[src])            \n"  // |7|6|5|4|
-    "lw        $t2, -8(%[src])             \n"  // |11|10|9|8|
-    "lw        $t3, -4(%[src])             \n"  // |15|14|13|12|
-    "wsbh      $t0, $t0                    \n"  // |2|3|0|1|
-    "wsbh      $t1, $t1                    \n"  // |6|7|4|5|
-    "wsbh      $t2, $t2                    \n"  // |10|11|8|9|
-    "wsbh      $t3, $t3                    \n"  // |14|15|12|13|
-    "rotr      $t0, $t0, 16                \n"  // |0|1|2|3|
-    "rotr      $t1, $t1, 16                \n"  // |4|5|6|7|
-    "rotr      $t2, $t2, 16                \n"  // |8|9|10|11|
-    "rotr      $t3, $t3, 16                \n"  // |12|13|14|15|
-    "addiu     %[src], %[src], -16         \n"
-    "addiu     $t4, $t4, -1                \n"
-    "sw        $t3, 0(%[dst])              \n"  // |15|14|13|12|
-    "sw        $t2, 4(%[dst])              \n"  // |11|10|9|8|
-    "sw        $t1, 8(%[dst])              \n"  // |7|6|5|4|
-    "sw        $t0, 12(%[dst])             \n"  // |3|2|1|0|
-    "bgtz      $t4, 1b                     \n"
-    " addiu    %[dst], %[dst], 16          \n"
-    "beqz      $t5, 3f                     \n"
-    " nop                                  \n"
-
-   "2:                                     \n"
-    "lbu       $t0, -1(%[src])             \n"
-    "addiu     $t5, $t5, -1                \n"
-    "addiu     %[src], %[src], -1          \n"
-    "sb        $t0, 0(%[dst])              \n"
-    "bgez      $t5, 2b                     \n"
-    " addiu    %[dst], %[dst], 1           \n"
-
-   "3:                                     \n"
-    ".set pop                              \n"
-      : [src] "+r" (src), [dst] "+r" (dst)
-      : [width] "r" (width)
-      : "t0", "t1", "t2", "t3", "t4", "t5"
-  );
-}
-
-void MirrorUVRow_MIPS_DSPR2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                            int width) {
-  int x = 0;
-  int y = 0;
-  __asm__ __volatile__ (
-    ".set push                                    \n"
-    ".set noreorder                               \n"
-
-    "addu            $t4, %[width], %[width]      \n"
-    "srl             %[x], %[width], 4            \n"
-    "andi            %[y], %[width], 0xf          \n"
-    "blez            %[x], 2f                     \n"
-    " addu           %[src_uv], %[src_uv], $t4    \n"
-
-    ".p2align        2                            \n"
-   "1:                                            \n"
-    "lw              $t0, -32(%[src_uv])          \n"  // |3|2|1|0|
-    "lw              $t1, -28(%[src_uv])          \n"  // |7|6|5|4|
-    "lw              $t2, -24(%[src_uv])          \n"  // |11|10|9|8|
-    "lw              $t3, -20(%[src_uv])          \n"  // |15|14|13|12|
-    "lw              $t4, -16(%[src_uv])          \n"  // |19|18|17|16|
-    "lw              $t6, -12(%[src_uv])          \n"  // |23|22|21|20|
-    "lw              $t7, -8(%[src_uv])           \n"  // |27|26|25|24|
-    "lw              $t8, -4(%[src_uv])           \n"  // |31|30|29|28|
-
-    "rotr            $t0, $t0, 16                 \n"  // |1|0|3|2|
-    "rotr            $t1, $t1, 16                 \n"  // |5|4|7|6|
-    "rotr            $t2, $t2, 16                 \n"  // |9|8|11|10|
-    "rotr            $t3, $t3, 16                 \n"  // |13|12|15|14|
-    "rotr            $t4, $t4, 16                 \n"  // |17|16|19|18|
-    "rotr            $t6, $t6, 16                 \n"  // |21|20|23|22|
-    "rotr            $t7, $t7, 16                 \n"  // |25|24|27|26|
-    "rotr            $t8, $t8, 16                 \n"  // |29|28|31|30|
-    "precr.qb.ph     $t9, $t0, $t1                \n"  // |0|2|4|6|
-    "precrq.qb.ph    $t5, $t0, $t1                \n"  // |1|3|5|7|
-    "precr.qb.ph     $t0, $t2, $t3                \n"  // |8|10|12|14|
-    "precrq.qb.ph    $t1, $t2, $t3                \n"  // |9|11|13|15|
-    "precr.qb.ph     $t2, $t4, $t6                \n"  // |16|18|20|22|
-    "precrq.qb.ph    $t3, $t4, $t6                \n"  // |17|19|21|23|
-    "precr.qb.ph     $t4, $t7, $t8                \n"  // |24|26|28|30|
-    "precrq.qb.ph    $t6, $t7, $t8                \n"  // |25|27|29|31|
-    "addiu           %[src_uv], %[src_uv], -32    \n"
-    "addiu           %[x], %[x], -1               \n"
-    "swr             $t4, 0(%[dst_u])             \n"
-    "swl             $t4, 3(%[dst_u])             \n"  // |30|28|26|24|
-    "swr             $t6, 0(%[dst_v])             \n"
-    "swl             $t6, 3(%[dst_v])             \n"  // |31|29|27|25|
-    "swr             $t2, 4(%[dst_u])             \n"
-    "swl             $t2, 7(%[dst_u])             \n"  // |22|20|18|16|
-    "swr             $t3, 4(%[dst_v])             \n"
-    "swl             $t3, 7(%[dst_v])             \n"  // |23|21|19|17|
-    "swr             $t0, 8(%[dst_u])             \n"
-    "swl             $t0, 11(%[dst_u])            \n"  // |14|12|10|8|
-    "swr             $t1, 8(%[dst_v])             \n"
-    "swl             $t1, 11(%[dst_v])            \n"  // |15|13|11|9|
-    "swr             $t9, 12(%[dst_u])            \n"
-    "swl             $t9, 15(%[dst_u])            \n"  // |6|4|2|0|
-    "swr             $t5, 12(%[dst_v])            \n"
-    "swl             $t5, 15(%[dst_v])            \n"  // |7|5|3|1|
-    "addiu           %[dst_v], %[dst_v], 16       \n"
-    "bgtz            %[x], 1b                     \n"
-    " addiu          %[dst_u], %[dst_u], 16       \n"
-    "beqz            %[y], 3f                     \n"
-    " nop                                         \n"
-    "b               2f                           \n"
-    " nop                                         \n"
-
-   "2:                                            \n"
-    "lbu             $t0, -2(%[src_uv])           \n"
-    "lbu             $t1, -1(%[src_uv])           \n"
-    "addiu           %[src_uv], %[src_uv], -2     \n"
-    "addiu           %[y], %[y], -1               \n"
-    "sb              $t0, 0(%[dst_u])             \n"
-    "sb              $t1, 0(%[dst_v])             \n"
-    "addiu           %[dst_u], %[dst_u], 1        \n"
-    "bgtz            %[y], 2b                     \n"
-    " addiu          %[dst_v], %[dst_v], 1        \n"
-
-   "3:                                            \n"
-    ".set pop                                     \n"
-      : [src_uv] "+r" (src_uv),
-        [dst_u] "+r" (dst_u),
-        [dst_v] "+r" (dst_v),
-        [x] "=&r" (x),
-        [y] "+r" (y)
-      : [width] "r" (width)
-      : "t0", "t1", "t2", "t3", "t4",
-      "t5", "t7", "t8", "t9"
-  );
-}
-
-// Convert (4 Y and 2 VU) I422 and arrange RGB values into
-// t5 = | 0 | B0 | 0 | b0 |
-// t4 = | 0 | B1 | 0 | b1 |
-// t9 = | 0 | G0 | 0 | g0 |
-// t8 = | 0 | G1 | 0 | g1 |
-// t2 = | 0 | R0 | 0 | r0 |
-// t1 = | 0 | R1 | 0 | r1 |
-#define I422ToTransientMipsRGB                                                 \
-      "lw                $t0, 0(%[y_buf])       \n"                            \
-      "lhu               $t1, 0(%[u_buf])       \n"                            \
-      "lhu               $t2, 0(%[v_buf])       \n"                            \
-      "preceu.ph.qbr     $t1, $t1               \n"                            \
-      "preceu.ph.qbr     $t2, $t2               \n"                            \
-      "preceu.ph.qbra    $t3, $t0               \n"                            \
-      "preceu.ph.qbla    $t0, $t0               \n"                            \
-      "subu.ph           $t1, $t1, $s5          \n"                            \
-      "subu.ph           $t2, $t2, $s5          \n"                            \
-      "subu.ph           $t3, $t3, $s4          \n"                            \
-      "subu.ph           $t0, $t0, $s4          \n"                            \
-      "mul.ph            $t3, $t3, $s0          \n"                            \
-      "mul.ph            $t0, $t0, $s0          \n"                            \
-      "shll.ph           $t4, $t1, 0x7          \n"                            \
-      "subu.ph           $t4, $t4, $t1          \n"                            \
-      "mul.ph            $t6, $t1, $s1          \n"                            \
-      "mul.ph            $t1, $t2, $s2          \n"                            \
-      "addq_s.ph         $t5, $t4, $t3          \n"                            \
-      "addq_s.ph         $t4, $t4, $t0          \n"                            \
-      "shra.ph           $t5, $t5, 6            \n"                            \
-      "shra.ph           $t4, $t4, 6            \n"                            \
-      "addiu             %[u_buf], 2            \n"                            \
-      "addiu             %[v_buf], 2            \n"                            \
-      "addu.ph           $t6, $t6, $t1          \n"                            \
-      "mul.ph            $t1, $t2, $s3          \n"                            \
-      "addu.ph           $t9, $t6, $t3          \n"                            \
-      "addu.ph           $t8, $t6, $t0          \n"                            \
-      "shra.ph           $t9, $t9, 6            \n"                            \
-      "shra.ph           $t8, $t8, 6            \n"                            \
-      "addu.ph           $t2, $t1, $t3          \n"                            \
-      "addu.ph           $t1, $t1, $t0          \n"                            \
-      "shra.ph           $t2, $t2, 6            \n"                            \
-      "shra.ph           $t1, $t1, 6            \n"                            \
-      "subu.ph           $t5, $t5, $s5          \n"                            \
-      "subu.ph           $t4, $t4, $s5          \n"                            \
-      "subu.ph           $t9, $t9, $s5          \n"                            \
-      "subu.ph           $t8, $t8, $s5          \n"                            \
-      "subu.ph           $t2, $t2, $s5          \n"                            \
-      "subu.ph           $t1, $t1, $s5          \n"                            \
-      "shll_s.ph         $t5, $t5, 8            \n"                            \
-      "shll_s.ph         $t4, $t4, 8            \n"                            \
-      "shll_s.ph         $t9, $t9, 8            \n"                            \
-      "shll_s.ph         $t8, $t8, 8            \n"                            \
-      "shll_s.ph         $t2, $t2, 8            \n"                            \
-      "shll_s.ph         $t1, $t1, 8            \n"                            \
-      "shra.ph           $t5, $t5, 8            \n"                            \
-      "shra.ph           $t4, $t4, 8            \n"                            \
-      "shra.ph           $t9, $t9, 8            \n"                            \
-      "shra.ph           $t8, $t8, 8            \n"                            \
-      "shra.ph           $t2, $t2, 8            \n"                            \
-      "shra.ph           $t1, $t1, 8            \n"                            \
-      "addu.ph           $t5, $t5, $s5          \n"                            \
-      "addu.ph           $t4, $t4, $s5          \n"                            \
-      "addu.ph           $t9, $t9, $s5          \n"                            \
-      "addu.ph           $t8, $t8, $s5          \n"                            \
-      "addu.ph           $t2, $t2, $s5          \n"                            \
-      "addu.ph           $t1, $t1, $s5          \n"
-
-void I422ToARGBRow_MIPS_DSPR2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* rgb_buf,
-                              int width) {
-  __asm__ __volatile__ (
-    ".set push                                \n"
-    ".set noreorder                           \n"
-    "beqz              %[width], 2f           \n"
-    " repl.ph          $s0, 74                \n"  // |YG|YG| = |74|74|
-    "repl.ph           $s1, -25               \n"  // |UG|UG| = |-25|-25|
-    "repl.ph           $s2, -52               \n"  // |VG|VG| = |-52|-52|
-    "repl.ph           $s3, 102               \n"  // |VR|VR| = |102|102|
-    "repl.ph           $s4, 16                \n"  // |0|16|0|16|
-    "repl.ph           $s5, 128               \n"  // |128|128| // clipping
-    "lui               $s6, 0xff00            \n"
-    "ori               $s6, 0xff00            \n"  // |ff|00|ff|00|ff|
-
-    ".p2align          2                      \n"
-   "1:                                        \n"
-      I422ToTransientMipsRGB
-// Arranging into argb format
-    "precr.qb.ph       $t4, $t8, $t4          \n"  // |G1|g1|B1|b1|
-    "precr.qb.ph       $t5, $t9, $t5          \n"  // |G0|g0|B0|b0|
-    "addiu             %[width], -4           \n"
-    "precrq.qb.ph      $t8, $t4, $t5          \n"  // |G1|B1|G0|B0|
-    "precr.qb.ph       $t9, $t4, $t5          \n"  // |g1|b1|g0|b0|
-    "precr.qb.ph       $t2, $t1, $t2          \n"  // |R1|r1|R0|r0|
-
-    "addiu             %[y_buf], 4            \n"
-    "preceu.ph.qbla    $t1, $t2               \n"  // |0 |R1|0 |R0|
-    "preceu.ph.qbra    $t2, $t2               \n"  // |0 |r1|0 |r0|
-    "or                $t1, $t1, $s6          \n"  // |ff|R1|ff|R0|
-    "or                $t2, $t2, $s6          \n"  // |ff|r1|ff|r0|
-    "precrq.ph.w       $t0, $t2, $t9          \n"  // |ff|r1|g1|b1|
-    "precrq.ph.w       $t3, $t1, $t8          \n"  // |ff|R1|G1|B1|
-    "sll               $t9, $t9, 16           \n"
-    "sll               $t8, $t8, 16           \n"
-    "packrl.ph         $t2, $t2, $t9          \n"  // |ff|r0|g0|b0|
-    "packrl.ph         $t1, $t1, $t8          \n"  // |ff|R0|G0|B0|
-// Store results.
-    "sw                $t2, 0(%[rgb_buf])     \n"
-    "sw                $t0, 4(%[rgb_buf])     \n"
-    "sw                $t1, 8(%[rgb_buf])     \n"
-    "sw                $t3, 12(%[rgb_buf])    \n"
-    "bnez              %[width], 1b           \n"
-    " addiu            %[rgb_buf], 16         \n"
-   "2:                                        \n"
-    ".set pop                                 \n"
-      :[y_buf] "+r" (y_buf),
-       [u_buf] "+r" (u_buf),
-       [v_buf] "+r" (v_buf),
-       [width] "+r" (width),
-       [rgb_buf] "+r" (rgb_buf)
-      :
-      : "t0", "t1",  "t2", "t3",  "t4", "t5",
-      "t6", "t7", "t8", "t9",
-      "s0", "s1", "s2", "s3",
-      "s4", "s5", "s6"
-  );
-}
-
-void I422ToABGRRow_MIPS_DSPR2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* rgb_buf,
-                              int width) {
-  __asm__ __volatile__ (
-    ".set push                                \n"
-    ".set noreorder                           \n"
-    "beqz              %[width], 2f           \n"
-    " repl.ph          $s0, 74                \n"  // |YG|YG| = |74|74|
-    "repl.ph           $s1, -25               \n"  // |UG|UG| = |-25|-25|
-    "repl.ph           $s2, -52               \n"  // |VG|VG| = |-52|-52|
-    "repl.ph           $s3, 102               \n"  // |VR|VR| = |102|102|
-    "repl.ph           $s4, 16                \n"  // |0|16|0|16|
-    "repl.ph           $s5, 128               \n"  // |128|128|
-    "lui               $s6, 0xff00            \n"
-    "ori               $s6, 0xff00            \n"  // |ff|00|ff|00|
-
-    ".p2align          2                       \n"
-   "1:                                         \n"
-      I422ToTransientMipsRGB
-// Arranging into abgr format
-    "precr.qb.ph      $t0, $t8, $t1           \n"  // |G1|g1|R1|r1|
-    "precr.qb.ph      $t3, $t9, $t2           \n"  // |G0|g0|R0|r0|
-    "precrq.qb.ph     $t8, $t0, $t3           \n"  // |G1|R1|G0|R0|
-    "precr.qb.ph      $t9, $t0, $t3           \n"  // |g1|r1|g0|r0|
-
-    "precr.qb.ph       $t2, $t4, $t5          \n"  // |B1|b1|B0|b0|
-    "addiu             %[width], -4           \n"
-    "addiu             %[y_buf], 4            \n"
-    "preceu.ph.qbla    $t1, $t2               \n"  // |0 |B1|0 |B0|
-    "preceu.ph.qbra    $t2, $t2               \n"  // |0 |b1|0 |b0|
-    "or                $t1, $t1, $s6          \n"  // |ff|B1|ff|B0|
-    "or                $t2, $t2, $s6          \n"  // |ff|b1|ff|b0|
-    "precrq.ph.w       $t0, $t2, $t9          \n"  // |ff|b1|g1|r1|
-    "precrq.ph.w       $t3, $t1, $t8          \n"  // |ff|B1|G1|R1|
-    "sll               $t9, $t9, 16           \n"
-    "sll               $t8, $t8, 16           \n"
-    "packrl.ph         $t2, $t2, $t9          \n"  // |ff|b0|g0|r0|
-    "packrl.ph         $t1, $t1, $t8          \n"  // |ff|B0|G0|R0|
-// Store results.
-    "sw                $t2, 0(%[rgb_buf])     \n"
-    "sw                $t0, 4(%[rgb_buf])     \n"
-    "sw                $t1, 8(%[rgb_buf])     \n"
-    "sw                $t3, 12(%[rgb_buf])    \n"
-    "bnez              %[width], 1b           \n"
-    " addiu            %[rgb_buf], 16         \n"
-   "2:                                        \n"
-    ".set pop                                 \n"
-      :[y_buf] "+r" (y_buf),
-       [u_buf] "+r" (u_buf),
-       [v_buf] "+r" (v_buf),
-       [width] "+r" (width),
-       [rgb_buf] "+r" (rgb_buf)
-      :
-      : "t0", "t1",  "t2", "t3",  "t4", "t5",
-      "t6", "t7", "t8", "t9",
-      "s0", "s1", "s2", "s3",
-      "s4", "s5", "s6"
-  );
-}
-
-void I422ToBGRARow_MIPS_DSPR2(const uint8* y_buf,
-                              const uint8* u_buf,
-                              const uint8* v_buf,
-                              uint8* rgb_buf,
-                              int width) {
-  __asm__ __volatile__ (
-    ".set push                                \n"
-    ".set noreorder                           \n"
-    "beqz              %[width], 2f           \n"
-    " repl.ph          $s0, 74                \n"  // |YG|YG| = |74 |74 |
-    "repl.ph           $s1, -25               \n"  // |UG|UG| = |-25|-25|
-    "repl.ph           $s2, -52               \n"  // |VG|VG| = |-52|-52|
-    "repl.ph           $s3, 102               \n"  // |VR|VR| = |102|102|
-    "repl.ph           $s4, 16                \n"  // |0|16|0|16|
-    "repl.ph           $s5, 128               \n"  // |128|128|
-    "lui               $s6, 0xff              \n"
-    "ori               $s6, 0xff              \n"  // |00|ff|00|ff|
-
-    ".p2align          2                      \n"
-   "1:                                        \n"
-      I422ToTransientMipsRGB
-      // Arranging into bgra format
-    "precr.qb.ph       $t4, $t4, $t8          \n"  // |B1|b1|G1|g1|
-    "precr.qb.ph       $t5, $t5, $t9          \n"  // |B0|b0|G0|g0|
-    "precrq.qb.ph      $t8, $t4, $t5          \n"  // |B1|G1|B0|G0|
-    "precr.qb.ph       $t9, $t4, $t5          \n"  // |b1|g1|b0|g0|
-
-    "precr.qb.ph       $t2, $t1, $t2          \n"  // |R1|r1|R0|r0|
-    "addiu             %[width], -4           \n"
-    "addiu             %[y_buf], 4            \n"
-    "preceu.ph.qbla    $t1, $t2               \n"  // |0 |R1|0 |R0|
-    "preceu.ph.qbra    $t2, $t2               \n"  // |0 |r1|0 |r0|
-    "sll               $t1, $t1, 8            \n"  // |R1|0 |R0|0 |
-    "sll               $t2, $t2, 8            \n"  // |r1|0 |r0|0 |
-    "or                $t1, $t1, $s6          \n"  // |R1|ff|R0|ff|
-    "or                $t2, $t2, $s6          \n"  // |r1|ff|r0|ff|
-    "precrq.ph.w       $t0, $t9, $t2          \n"  // |b1|g1|r1|ff|
-    "precrq.ph.w       $t3, $t8, $t1          \n"  // |B1|G1|R1|ff|
-    "sll               $t1, $t1, 16           \n"
-    "sll               $t2, $t2, 16           \n"
-    "packrl.ph         $t2, $t9, $t2          \n"  // |b0|g0|r0|ff|
-    "packrl.ph         $t1, $t8, $t1          \n"  // |B0|G0|R0|ff|
-// Store results.
-    "sw                $t2, 0(%[rgb_buf])     \n"
-    "sw                $t0, 4(%[rgb_buf])     \n"
-    "sw                $t1, 8(%[rgb_buf])     \n"
-    "sw                $t3, 12(%[rgb_buf])    \n"
-    "bnez              %[width], 1b           \n"
-    " addiu            %[rgb_buf], 16         \n"
-   "2:                                        \n"
-    ".set pop                                 \n"
-      :[y_buf] "+r" (y_buf),
-       [u_buf] "+r" (u_buf),
-       [v_buf] "+r" (v_buf),
-       [width] "+r" (width),
-       [rgb_buf] "+r" (rgb_buf)
-      :
-      : "t0", "t1",  "t2", "t3",  "t4", "t5",
-      "t6", "t7", "t8", "t9",
-      "s0", "s1", "s2", "s3",
-      "s4", "s5", "s6"
-  );
-}
-
-// Bilinear filter 8x2 -> 8x1
-void InterpolateRows_MIPS_DSPR2(uint8* dst_ptr, const uint8* src_ptr,
-                                ptrdiff_t src_stride, int dst_width,
-                                int source_y_fraction) {
-    int y0_fraction = 256 - source_y_fraction;
-    const uint8* src_ptr1 = src_ptr + src_stride;
-
-  __asm__ __volatile__ (
-     ".set push                                           \n"
-     ".set noreorder                                      \n"
-
-     "replv.ph          $t0, %[y0_fraction]               \n"
-     "replv.ph          $t1, %[source_y_fraction]         \n"
-
-    ".p2align           2                                 \n"
-   "1:                                                    \n"
-     "lw                $t2, 0(%[src_ptr])                \n"
-     "lw                $t3, 0(%[src_ptr1])               \n"
-     "lw                $t4, 4(%[src_ptr])                \n"
-     "lw                $t5, 4(%[src_ptr1])               \n"
-     "muleu_s.ph.qbl    $t6, $t2, $t0                     \n"
-     "muleu_s.ph.qbr    $t7, $t2, $t0                     \n"
-     "muleu_s.ph.qbl    $t8, $t3, $t1                     \n"
-     "muleu_s.ph.qbr    $t9, $t3, $t1                     \n"
-     "muleu_s.ph.qbl    $t2, $t4, $t0                     \n"
-     "muleu_s.ph.qbr    $t3, $t4, $t0                     \n"
-     "muleu_s.ph.qbl    $t4, $t5, $t1                     \n"
-     "muleu_s.ph.qbr    $t5, $t5, $t1                     \n"
-     "addq.ph           $t6, $t6, $t8                     \n"
-     "addq.ph           $t7, $t7, $t9                     \n"
-     "addq.ph           $t2, $t2, $t4                     \n"
-     "addq.ph           $t3, $t3, $t5                     \n"
-     "shra.ph           $t6, $t6, 8                       \n"
-     "shra.ph           $t7, $t7, 8                       \n"
-     "shra.ph           $t2, $t2, 8                       \n"
-     "shra.ph           $t3, $t3, 8                       \n"
-     "precr.qb.ph       $t6, $t6, $t7                     \n"
-     "precr.qb.ph       $t2, $t2, $t3                     \n"
-     "addiu             %[src_ptr], %[src_ptr], 8         \n"
-     "addiu             %[src_ptr1], %[src_ptr1], 8       \n"
-     "addiu             %[dst_width], %[dst_width], -8    \n"
-     "sw                $t6, 0(%[dst_ptr])                \n"
-     "sw                $t2, 4(%[dst_ptr])                \n"
-     "bgtz              %[dst_width], 1b                  \n"
-     " addiu            %[dst_ptr], %[dst_ptr], 8         \n"
-
-     ".set pop                                            \n"
-  : [dst_ptr] "+r" (dst_ptr),
-    [src_ptr1] "+r" (src_ptr1),
-    [src_ptr] "+r" (src_ptr),
-    [dst_width] "+r" (dst_width)
-  : [source_y_fraction] "r" (source_y_fraction),
-    [y0_fraction] "r" (y0_fraction),
-    [src_stride] "r" (src_stride)
-  : "t0", "t1", "t2", "t3", "t4", "t5",
-    "t6", "t7", "t8", "t9"
-  );
-}
-#endif  // __mips_dsp_rev >= 2
-
-#endif  // defined(__mips__)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_neon.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_neon.cc
deleted file mode 100755
index 68e380051b..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_neon.cc
+++ /dev/null
@@ -1,2847 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC Neon
-#if !defined(LIBYUV_DISABLE_NEON) && defined(__ARM_NEON__)
-
-// Read 8 Y, 4 U and 4 V from 422
-#define READYUV422                                                             \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vld1.32    {d2[0]}, [%1]!                 \n"                             \
-    "vld1.32    {d2[1]}, [%2]!                 \n"
-
-// Read 8 Y, 2 U and 2 V from 422
-#define READYUV411                                                             \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vld1.16    {d2[0]}, [%1]!                 \n"                             \
-    "vld1.16    {d2[1]}, [%2]!                 \n"                             \
-    "vmov.u8    d3, d2                         \n"                             \
-    "vzip.u8    d2, d3                         \n"
-
-// Read 8 Y, 8 U and 8 V from 444
-#define READYUV444                                                             \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vld1.8     {d2}, [%1]!                    \n"                             \
-    "vld1.8     {d3}, [%2]!                    \n"                             \
-    "vpaddl.u8  q1, q1                         \n"                             \
-    "vrshrn.u16 d2, q1, #1                     \n"
-
-// Read 8 Y, and set 4 U and 4 V to 128
-#define READYUV400                                                             \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vmov.u8    d2, #128                       \n"
-
-// Read 8 Y and 4 UV from NV12
-#define READNV12                                                               \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vld1.8     {d2}, [%1]!                    \n"                             \
-    "vmov.u8    d3, d2                         \n"/* split odd/even uv apart */\
-    "vuzp.u8    d2, d3                         \n"                             \
-    "vtrn.u32   d2, d3                         \n"
-
-// Read 8 Y and 4 VU from NV21
-#define READNV21                                                               \
-    "vld1.8     {d0}, [%0]!                    \n"                             \
-    "vld1.8     {d2}, [%1]!                    \n"                             \
-    "vmov.u8    d3, d2                         \n"/* split odd/even uv apart */\
-    "vuzp.u8    d3, d2                         \n"                             \
-    "vtrn.u32   d2, d3                         \n"
-
-// Read 8 YUY2
-#define READYUY2                                                               \
-    "vld2.8     {d0, d2}, [%0]!                \n"                             \
-    "vmov.u8    d3, d2                         \n"                             \
-    "vuzp.u8    d2, d3                         \n"                             \
-    "vtrn.u32   d2, d3                         \n"
-
-// Read 8 UYVY
-#define READUYVY                                                               \
-    "vld2.8     {d2, d3}, [%0]!                \n"                             \
-    "vmov.u8    d0, d3                         \n"                             \
-    "vmov.u8    d3, d2                         \n"                             \
-    "vuzp.u8    d2, d3                         \n"                             \
-    "vtrn.u32   d2, d3                         \n"
-
-#define YUV422TORGB                                                            \
-    "veor.u8    d2, d26                        \n"/*subtract 128 from u and v*/\
-    "vmull.s8   q8, d2, d24                    \n"/*  u/v B/R component      */\
-    "vmull.s8   q9, d2, d25                    \n"/*  u/v G component        */\
-    "vmov.u8    d1, #0                         \n"/*  split odd/even y apart */\
-    "vtrn.u8    d0, d1                         \n"                             \
-    "vsub.s16   q0, q0, q15                    \n"/*  offset y               */\
-    "vmul.s16   q0, q0, q14                    \n"                             \
-    "vadd.s16   d18, d19                       \n"                             \
-    "vqadd.s16  d20, d0, d16                   \n" /* B */                     \
-    "vqadd.s16  d21, d1, d16                   \n"                             \
-    "vqadd.s16  d22, d0, d17                   \n" /* R */                     \
-    "vqadd.s16  d23, d1, d17                   \n"                             \
-    "vqadd.s16  d16, d0, d18                   \n" /* G */                     \
-    "vqadd.s16  d17, d1, d18                   \n"                             \
-    "vqshrun.s16 d0, q10, #6                   \n" /* B */                     \
-    "vqshrun.s16 d1, q11, #6                   \n" /* G */                     \
-    "vqshrun.s16 d2, q8, #6                    \n" /* R */                     \
-    "vmovl.u8   q10, d0                        \n"/*  set up for reinterleave*/\
-    "vmovl.u8   q11, d1                        \n"                             \
-    "vmovl.u8   q8, d2                         \n"                             \
-    "vtrn.u8    d20, d21                       \n"                             \
-    "vtrn.u8    d22, d23                       \n"                             \
-    "vtrn.u8    d16, d17                       \n"                             \
-    "vmov.u8    d21, d16                       \n"
-
-static vec8 kUVToRB  = { 127, 127, 127, 127, 102, 102, 102, 102,
-                         0, 0, 0, 0, 0, 0, 0, 0 };
-static vec8 kUVToG = { -25, -25, -25, -25, -52, -52, -52, -52,
-                       0, 0, 0, 0, 0, 0, 0, 0 };
-
-void I444ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-#ifdef _ANDROID
-				".fpu neon\n"
-#endif
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV444
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_argb),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_argb),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I411ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV411
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_argb),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToBGRARow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_bgra,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vswp.u8    d20, d22                       \n"
-    "vmov.u8    d19, #255                      \n"
-    "vst4.8     {d19, d20, d21, d22}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_bgra),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToABGRRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_abgr,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vswp.u8    d20, d22                       \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_abgr),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToRGBARow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_rgba,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d19, #255                      \n"
-    "vst4.8     {d19, d20, d21, d22}, [%3]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_u),     // %1
-      "+r"(src_v),     // %2
-      "+r"(dst_rgba),  // %3
-      "+r"(width)      // %4
-    : "r"(&kUVToRB),   // %5
-      "r"(&kUVToG)     // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToRGB24Row_NEON(const uint8* src_y,
-                         const uint8* src_u,
-                         const uint8* src_v,
-                         uint8* dst_rgb24,
-                         int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vst3.8     {d20, d21, d22}, [%3]!         \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),      // %0
-      "+r"(src_u),      // %1
-      "+r"(src_v),      // %2
-      "+r"(dst_rgb24),  // %3
-      "+r"(width)       // %4
-    : "r"(&kUVToRB),    // %5
-      "r"(&kUVToG)      // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I422ToRAWRow_NEON(const uint8* src_y,
-                       const uint8* src_u,
-                       const uint8* src_v,
-                       uint8* dst_raw,
-                       int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vswp.u8    d20, d22                       \n"
-    "vst3.8     {d20, d21, d22}, [%3]!         \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),    // %0
-      "+r"(src_u),    // %1
-      "+r"(src_v),    // %2
-      "+r"(dst_raw),  // %3
-      "+r"(width)     // %4
-    : "r"(&kUVToRB),  // %5
-      "r"(&kUVToG)    // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-#define ARGBTORGB565                                                           \
-    "vshr.u8    d20, d20, #3                   \n"  /* B                    */ \
-    "vshr.u8    d21, d21, #2                   \n"  /* G                    */ \
-    "vshr.u8    d22, d22, #3                   \n"  /* R                    */ \
-    "vmovl.u8   q8, d20                        \n"  /* B                    */ \
-    "vmovl.u8   q9, d21                        \n"  /* G                    */ \
-    "vmovl.u8   q10, d22                       \n"  /* R                    */ \
-    "vshl.u16   q9, q9, #5                     \n"  /* G                    */ \
-    "vshl.u16   q10, q10, #11                  \n"  /* R                    */ \
-    "vorr       q0, q8, q9                     \n"  /* BG                   */ \
-    "vorr       q0, q0, q10                    \n"  /* BGR                  */
-
-void I422ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_u,
-                          const uint8* src_v,
-                          uint8* dst_rgb565,
-                          int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    ARGBTORGB565
-    "vst1.8     {q0}, [%3]!                    \n"  // store 8 pixels RGB565.
-    "bgt        1b                             \n"
-    : "+r"(src_y),    // %0
-      "+r"(src_u),    // %1
-      "+r"(src_v),    // %2
-      "+r"(dst_rgb565),  // %3
-      "+r"(width)     // %4
-    : "r"(&kUVToRB),  // %5
-      "r"(&kUVToG)    // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-#define ARGBTOARGB1555                                                         \
-    "vshr.u8    q10, q10, #3                   \n"  /* B                    */ \
-    "vshr.u8    d22, d22, #3                   \n"  /* R                    */ \
-    "vshr.u8    d23, d23, #7                   \n"  /* A                    */ \
-    "vmovl.u8   q8, d20                        \n"  /* B                    */ \
-    "vmovl.u8   q9, d21                        \n"  /* G                    */ \
-    "vmovl.u8   q10, d22                       \n"  /* R                    */ \
-    "vmovl.u8   q11, d23                       \n"  /* A                    */ \
-    "vshl.u16   q9, q9, #5                     \n"  /* G                    */ \
-    "vshl.u16   q10, q10, #10                  \n"  /* R                    */ \
-    "vshl.u16   q11, q11, #15                  \n"  /* A                    */ \
-    "vorr       q0, q8, q9                     \n"  /* BG                   */ \
-    "vorr       q1, q10, q11                   \n"  /* RA                   */ \
-    "vorr       q0, q0, q1                     \n"  /* BGRA                 */
-
-void I422ToARGB1555Row_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb1555,
-                            int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    ARGBTOARGB1555
-    "vst1.8     {q0}, [%3]!                    \n"  // store 8 pixels ARGB1555.
-    "bgt        1b                             \n"
-    : "+r"(src_y),    // %0
-      "+r"(src_u),    // %1
-      "+r"(src_v),    // %2
-      "+r"(dst_argb1555),  // %3
-      "+r"(width)     // %4
-    : "r"(&kUVToRB),  // %5
-      "r"(&kUVToG)    // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-#define ARGBTOARGB4444                                                         \
-    "vshr.u8    d20, d20, #4                   \n"  /* B                    */ \
-    "vbic.32    d21, d21, d4                   \n"  /* G                    */ \
-    "vshr.u8    d22, d22, #4                   \n"  /* R                    */ \
-    "vbic.32    d23, d23, d4                   \n"  /* A                    */ \
-    "vorr       d0, d20, d21                   \n"  /* BG                   */ \
-    "vorr       d1, d22, d23                   \n"  /* RA                   */ \
-    "vzip.u8    d0, d1                         \n"  /* BGRA                 */
-
-void I422ToARGB4444Row_NEON(const uint8* src_y,
-                            const uint8* src_u,
-                            const uint8* src_v,
-                            uint8* dst_argb4444,
-                            int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%5]                    \n"
-    "vld1.8     {d25}, [%6]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    "vmov.u8    d4, #0x0f                      \n"  // bits to clear with vbic.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV422
-    YUV422TORGB
-    "subs       %4, %4, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    ARGBTOARGB4444
-    "vst1.8     {q0}, [%3]!                    \n"  // store 8 pixels ARGB4444.
-    "bgt        1b                             \n"
-    : "+r"(src_y),    // %0
-      "+r"(src_u),    // %1
-      "+r"(src_v),    // %2
-      "+r"(dst_argb4444),  // %3
-      "+r"(width)     // %4
-    : "r"(&kUVToRB),  // %5
-      "r"(&kUVToG)    // %6
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void YToARGBRow_NEON(const uint8* src_y,
-                     uint8* dst_argb,
-                     int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%3]                    \n"
-    "vld1.8     {d25}, [%4]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUV400
-    YUV422TORGB
-    "subs       %2, %2, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%1]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(dst_argb),  // %1
-      "+r"(width)      // %2
-    : "r"(&kUVToRB),   // %3
-      "r"(&kUVToG)     // %4
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void I400ToARGBRow_NEON(const uint8* src_y,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-    "vmov.u8    d23, #255                      \n"
-  "1:                                          \n"
-    "vld1.8     {d20}, [%0]!                   \n"
-    "vmov       d21, d20                       \n"
-    "vmov       d22, d20                       \n"
-    "subs       %2, %2, #8                     \n"
-    "vst4.8     {d20, d21, d22, d23}, [%1]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(dst_argb),  // %1
-      "+r"(width)      // %2
-    :
-    : "cc", "memory", "d20", "d21", "d22", "d23"
-  );
-}
-
-void NV12ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_uv,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%4]                    \n"
-    "vld1.8     {d25}, [%5]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READNV12
-    YUV422TORGB
-    "subs       %3, %3, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%2]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_uv),    // %1
-      "+r"(dst_argb),  // %2
-      "+r"(width)      // %3
-    : "r"(&kUVToRB),   // %4
-      "r"(&kUVToG)     // %5
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void NV21ToARGBRow_NEON(const uint8* src_y,
-                        const uint8* src_uv,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%4]                    \n"
-    "vld1.8     {d25}, [%5]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READNV21
-    YUV422TORGB
-    "subs       %3, %3, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%2]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_uv),    // %1
-      "+r"(dst_argb),  // %2
-      "+r"(width)      // %3
-    : "r"(&kUVToRB),   // %4
-      "r"(&kUVToG)     // %5
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void NV12ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_uv,
-                          uint8* dst_rgb565,
-                          int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%4]                    \n"
-    "vld1.8     {d25}, [%5]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READNV12
-    YUV422TORGB
-    "subs       %3, %3, #8                     \n"
-    ARGBTORGB565
-    "vst1.8     {q0}, [%2]!                    \n"  // store 8 pixels RGB565.
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_uv),    // %1
-      "+r"(dst_rgb565),  // %2
-      "+r"(width)      // %3
-    : "r"(&kUVToRB),   // %4
-      "r"(&kUVToG)     // %5
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void NV21ToRGB565Row_NEON(const uint8* src_y,
-                          const uint8* src_uv,
-                          uint8* dst_rgb565,
-                          int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%4]                    \n"
-    "vld1.8     {d25}, [%5]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READNV21
-    YUV422TORGB
-    "subs       %3, %3, #8                     \n"
-    ARGBTORGB565
-    "vst1.8     {q0}, [%2]!                    \n"  // store 8 pixels RGB565.
-    "bgt        1b                             \n"
-    : "+r"(src_y),     // %0
-      "+r"(src_uv),    // %1
-      "+r"(dst_rgb565),  // %2
-      "+r"(width)      // %3
-    : "r"(&kUVToRB),   // %4
-      "r"(&kUVToG)     // %5
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void YUY2ToARGBRow_NEON(const uint8* src_yuy2,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%3]                    \n"
-    "vld1.8     {d25}, [%4]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READYUY2
-    YUV422TORGB
-    "subs       %2, %2, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%1]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_yuy2),  // %0
-      "+r"(dst_argb),  // %1
-      "+r"(width)      // %2
-    : "r"(&kUVToRB),   // %3
-      "r"(&kUVToG)     // %4
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void UYVYToARGBRow_NEON(const uint8* src_uyvy,
-                        uint8* dst_argb,
-                        int width) {
-  asm volatile (
-    "vld1.8     {d24}, [%3]                    \n"
-    "vld1.8     {d25}, [%4]                    \n"
-    "vmov.u8    d26, #128                      \n"
-    "vmov.u16   q14, #74                       \n"
-    "vmov.u16   q15, #16                       \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    READUYVY
-    YUV422TORGB
-    "subs       %2, %2, #8                     \n"
-    "vmov.u8    d23, #255                      \n"
-    "vst4.8     {d20, d21, d22, d23}, [%1]!    \n"
-    "bgt        1b                             \n"
-    : "+r"(src_uyvy),  // %0
-      "+r"(dst_argb),  // %1
-      "+r"(width)      // %2
-    : "r"(&kUVToRB),   // %3
-      "r"(&kUVToG)     // %4
-    : "cc", "memory", "q0", "q1", "q2", "q3",
-      "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// Reads 16 pairs of UV and write even values to dst_u and odd to dst_v.
-void SplitUVRow_NEON(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                     int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {q0, q1}, [%0]!                \n"  // load 16 pairs of UV
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop
-    "vst1.8     {q0}, [%1]!                    \n"  // store U
-    "vst1.8     {q1}, [%2]!                    \n"  // store V
-    "bgt        1b                             \n"
-    : "+r"(src_uv),  // %0
-      "+r"(dst_u),   // %1
-      "+r"(dst_v),   // %2
-      "+r"(width)    // %3  // Output registers
-    :                       // Input registers
-    : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// Reads 16 U's and V's and writes out 16 pairs of UV.
-void MergeUVRow_NEON(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load U
-    "vld1.8     {q1}, [%1]!                    \n"  // load V
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop
-    "vst2.u8    {q0, q1}, [%2]!                \n"  // store 16 pairs of UV
-    "bgt        1b                             \n"
-    :
-      "+r"(src_u),   // %0
-      "+r"(src_v),   // %1
-      "+r"(dst_uv),  // %2
-      "+r"(width)    // %3  // Output registers
-    :                       // Input registers
-    : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// Copy multiple of 32.  vld4.8  allow unaligned and is fastest on a15.
-void CopyRow_NEON(const uint8* src, uint8* dst, int count) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 32
-    "subs       %2, %2, #32                    \n"  // 32 processed per loop
-    "vst1.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 32
-    "bgt        1b                             \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(count)  // %2  // Output registers
-  :                     // Input registers
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// SetRow8 writes 'count' bytes using a 32 bit value repeated.
-void SetRow_NEON(uint8* dst, uint32 v32, int count) {
-  asm volatile (
-    "vdup.u32  q0, %2                          \n"  // duplicate 4 ints
-    "1:                                        \n"
-    "subs      %1, %1, #16                     \n"  // 16 bytes per loop
-    "vst1.8    {q0}, [%0]!                     \n"  // store
-    "bgt       1b                              \n"
-  : "+r"(dst),   // %0
-    "+r"(count)  // %1
-  : "r"(v32)     // %2
-  : "cc", "memory", "q0"
-  );
-}
-
-// TODO(fbarchard): Make fully assembler
-// SetRow32 writes 'count' words using a 32 bit value repeated.
-void ARGBSetRows_NEON(uint8* dst, uint32 v32, int width,
-                      int dst_stride, int height) {
-  for (int y = 0; y < height; ++y) {
-    SetRow_NEON(dst, v32, width << 2);
-    dst += dst_stride;
-  }
-}
-
-void MirrorRow_NEON(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    // Start at end of source row.
-    "mov        r3, #-16                       \n"
-    "add        %0, %0, %2                     \n"
-    "sub        %0, #16                        \n"
-
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0], r3                 \n"  // src -= 16
-    "subs       %2, #16                        \n"  // 16 pixels per loop.
-    "vrev64.8   q0, q0                         \n"
-    "vst1.8     {d1}, [%1]!                    \n"  // dst += 16
-    "vst1.8     {d0}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "cc", "memory", "r3", "q0"
-  );
-}
-
-void MirrorUVRow_NEON(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                      int width) {
-  asm volatile (
-    // Start at end of source row.
-    "mov        r12, #-16                      \n"
-    "add        %0, %0, %3, lsl #1             \n"
-    "sub        %0, #16                        \n"
-
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {d0, d1}, [%0], r12            \n"  // src -= 16
-    "subs       %3, #8                         \n"  // 8 pixels per loop.
-    "vrev64.8   q0, q0                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // dst += 8
-    "vst1.8     {d1}, [%2]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_uv),  // %0
-    "+r"(dst_u),   // %1
-    "+r"(dst_v),   // %2
-    "+r"(width)    // %3
-  :
-  : "cc", "memory", "r12", "q0"
-  );
-}
-
-void ARGBMirrorRow_NEON(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    // Start at end of source row.
-    "mov        r3, #-16                       \n"
-    "add        %0, %0, %2, lsl #2             \n"
-    "sub        %0, #16                        \n"
-
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0], r3                 \n"  // src -= 16
-    "subs       %2, #4                         \n"  // 4 pixels per loop.
-    "vrev64.32  q0, q0                         \n"
-    "vst1.8     {d1}, [%1]!                    \n"  // dst += 16
-    "vst1.8     {d0}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "cc", "memory", "r3", "q0"
-  );
-}
-
-void RGB24ToARGBRow_NEON(const uint8* src_rgb24, uint8* dst_argb, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #255                       \n"  // Alpha
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d1, d2, d3}, [%0]!            \n"  // load 8 pixels of RGB24.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vst4.8     {d1, d2, d3, d4}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb24),  // %0
-    "+r"(dst_argb),   // %1
-    "+r"(pix)         // %2
-  :
-  : "cc", "memory", "d1", "d2", "d3", "d4"  // Clobber List
-  );
-}
-
-void RAWToARGBRow_NEON(const uint8* src_raw, uint8* dst_argb, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #255                       \n"  // Alpha
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d1, d2, d3}, [%0]!            \n"  // load 8 pixels of RAW.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vswp.u8    d1, d3                         \n"  // swap R, B
-    "vst4.8     {d1, d2, d3, d4}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_raw),   // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d1", "d2", "d3", "d4"  // Clobber List
-  );
-}
-
-#define RGB565TOARGB                                                           \
-    "vshrn.u16  d6, q0, #5                     \n"  /* G xxGGGGGG           */ \
-    "vuzp.u8    d0, d1                         \n"  /* d0 xxxBBBBB RRRRRxxx */ \
-    "vshl.u8    d6, d6, #2                     \n"  /* G GGGGGG00 upper 6   */ \
-    "vshr.u8    d1, d1, #3                     \n"  /* R 000RRRRR lower 5   */ \
-    "vshl.u8    q0, q0, #3                     \n"  /* B,R BBBBB000 upper 5 */ \
-    "vshr.u8    q2, q0, #5                     \n"  /* B,R 00000BBB lower 3 */ \
-    "vorr.u8    d0, d0, d4                     \n"  /* B                    */ \
-    "vshr.u8    d4, d6, #6                     \n"  /* G 000000GG lower 2   */ \
-    "vorr.u8    d2, d1, d5                     \n"  /* R                    */ \
-    "vorr.u8    d1, d4, d6                     \n"  /* G                    */
-
-void RGB565ToARGBRow_NEON(const uint8* src_rgb565, uint8* dst_argb, int pix) {
-  asm volatile (
-    "vmov.u8    d3, #255                       \n"  // Alpha
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 RGB565 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    RGB565TOARGB
-    "vst4.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb565),  // %0
-    "+r"(dst_argb),    // %1
-    "+r"(pix)          // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3"  // Clobber List
-  );
-}
-
-#define ARGB1555TOARGB                                                         \
-    "vshrn.u16  d7, q0, #8                     \n"  /* A Arrrrrxx           */ \
-    "vshr.u8    d6, d7, #2                     \n"  /* R xxxRRRRR           */ \
-    "vshrn.u16  d5, q0, #5                     \n"  /* G xxxGGGGG           */ \
-    "vmovn.u16  d4, q0                         \n"  /* B xxxBBBBB           */ \
-    "vshr.u8    d7, d7, #7                     \n"  /* A 0000000A           */ \
-    "vneg.s8    d7, d7                         \n"  /* A AAAAAAAA upper 8   */ \
-    "vshl.u8    d6, d6, #3                     \n"  /* R RRRRR000 upper 5   */ \
-    "vshr.u8    q1, q3, #5                     \n"  /* R,A 00000RRR lower 3 */ \
-    "vshl.u8    q0, q2, #3                     \n"  /* B,G BBBBB000 upper 5 */ \
-    "vshr.u8    q2, q0, #5                     \n"  /* B,G 00000BBB lower 3 */ \
-    "vorr.u8    q1, q1, q3                     \n"  /* R,A                  */ \
-    "vorr.u8    q0, q0, q2                     \n"  /* B,G                  */ \
-
-// RGB555TOARGB is same as ARGB1555TOARGB but ignores alpha.
-#define RGB555TOARGB                                                           \
-    "vshrn.u16  d6, q0, #5                     \n"  /* G xxxGGGGG           */ \
-    "vuzp.u8    d0, d1                         \n"  /* d0 xxxBBBBB xRRRRRxx */ \
-    "vshl.u8    d6, d6, #3                     \n"  /* G GGGGG000 upper 5   */ \
-    "vshr.u8    d1, d1, #2                     \n"  /* R 00xRRRRR lower 5   */ \
-    "vshl.u8    q0, q0, #3                     \n"  /* B,R BBBBB000 upper 5 */ \
-    "vshr.u8    q2, q0, #5                     \n"  /* B,R 00000BBB lower 3 */ \
-    "vorr.u8    d0, d0, d4                     \n"  /* B                    */ \
-    "vshr.u8    d4, d6, #5                     \n"  /* G 00000GGG lower 3   */ \
-    "vorr.u8    d2, d1, d5                     \n"  /* R                    */ \
-    "vorr.u8    d1, d4, d6                     \n"  /* G                    */
-
-void ARGB1555ToARGBRow_NEON(const uint8* src_argb1555, uint8* dst_argb,
-                            int pix) {
-  asm volatile (
-    "vmov.u8    d3, #255                       \n"  // Alpha
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB1555 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGB1555TOARGB
-    "vst4.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_argb1555),  // %0
-    "+r"(dst_argb),    // %1
-    "+r"(pix)          // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3"  // Clobber List
-  );
-}
-
-#define ARGB4444TOARGB                                                         \
-    "vuzp.u8    d0, d1                         \n"  /* d0 BG, d1 RA         */ \
-    "vshl.u8    q2, q0, #4                     \n"  /* B,R BBBB0000         */ \
-    "vshr.u8    q1, q0, #4                     \n"  /* G,A 0000GGGG         */ \
-    "vshr.u8    q0, q2, #4                     \n"  /* B,R 0000BBBB         */ \
-    "vorr.u8    q0, q0, q2                     \n"  /* B,R BBBBBBBB         */ \
-    "vshl.u8    q2, q1, #4                     \n"  /* G,A GGGG0000         */ \
-    "vorr.u8    q1, q1, q2                     \n"  /* G,A GGGGGGGG         */ \
-    "vswp.u8    d1, d2                         \n"  /* B,R,G,A -> B,G,R,A   */
-
-void ARGB4444ToARGBRow_NEON(const uint8* src_argb4444, uint8* dst_argb,
-                            int pix) {
-  asm volatile (
-    "vmov.u8    d3, #255                       \n"  // Alpha
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB4444 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGB4444TOARGB
-    "vst4.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_argb4444),  // %0
-    "+r"(dst_argb),    // %1
-    "+r"(pix)          // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2"  // Clobber List
-  );
-}
-
-void ARGBToRGB24Row_NEON(const uint8* src_argb, uint8* dst_rgb24, int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d1, d2, d3, d4}, [%0]!        \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vst3.8     {d1, d2, d3}, [%1]!            \n"  // store 8 pixels of RGB24.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_rgb24),  // %1
-    "+r"(pix)         // %2
-  :
-  : "cc", "memory", "d1", "d2", "d3", "d4"  // Clobber List
-  );
-}
-
-void ARGBToRAWRow_NEON(const uint8* src_argb, uint8* dst_raw, int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d1, d2, d3, d4}, [%0]!        \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vswp.u8    d1, d3                         \n"  // swap R, B
-    "vst3.8     {d1, d2, d3}, [%1]!            \n"  // store 8 pixels of RAW.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_raw),   // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d1", "d2", "d3", "d4"  // Clobber List
-  );
-}
-
-void YUY2ToYRow_NEON(const uint8* src_yuy2, uint8* dst_y, int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {q0, q1}, [%0]!                \n"  // load 16 pixels of YUY2.
-    "subs       %2, %2, #16                    \n"  // 16 processed per loop.
-    "vst1.8     {q0}, [%1]!                    \n"  // store 16 pixels of Y.
-    "bgt        1b                             \n"
-  : "+r"(src_yuy2),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-void UYVYToYRow_NEON(const uint8* src_uyvy, uint8* dst_y, int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {q0, q1}, [%0]!                \n"  // load 16 pixels of UYVY.
-    "subs       %2, %2, #16                    \n"  // 16 processed per loop.
-    "vst1.8     {q1}, [%1]!                    \n"  // store 16 pixels of Y.
-    "bgt        1b                             \n"
-  : "+r"(src_uyvy),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-void YUY2ToUV422Row_NEON(const uint8* src_yuy2, uint8* dst_u, uint8* dst_v,
-                         int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 16 pixels of YUY2.
-    "subs       %3, %3, #16                    \n"  // 16 pixels = 8 UVs.
-    "vst1.8     {d1}, [%1]!                    \n"  // store 8 U.
-    "vst1.8     {d3}, [%2]!                    \n"  // store 8 V.
-    "bgt        1b                             \n"
-  : "+r"(src_yuy2),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(pix)        // %3
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3"  // Clobber List
-  );
-}
-
-void UYVYToUV422Row_NEON(const uint8* src_uyvy, uint8* dst_u, uint8* dst_v,
-                         int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 16 pixels of UYVY.
-    "subs       %3, %3, #16                    \n"  // 16 pixels = 8 UVs.
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 U.
-    "vst1.8     {d2}, [%2]!                    \n"  // store 8 V.
-    "bgt        1b                             \n"
-  : "+r"(src_uyvy),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(pix)        // %3
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3"  // Clobber List
-  );
-}
-
-void YUY2ToUVRow_NEON(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // stride + src_yuy2
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 16 pixels of YUY2.
-    "subs       %4, %4, #16                    \n"  // 16 pixels = 8 UVs.
-    "vld4.8     {d4, d5, d6, d7}, [%1]!        \n"  // load next row YUY2.
-    "vrhadd.u8  d1, d1, d5                     \n"  // average rows of U
-    "vrhadd.u8  d3, d3, d7                     \n"  // average rows of V
-    "vst1.8     {d1}, [%2]!                    \n"  // store 8 U.
-    "vst1.8     {d3}, [%3]!                    \n"  // store 8 V.
-    "bgt        1b                             \n"
-  : "+r"(src_yuy2),     // %0
-    "+r"(stride_yuy2),  // %1
-    "+r"(dst_u),        // %2
-    "+r"(dst_v),        // %3
-    "+r"(pix)           // %4
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7"  // Clobber List
-  );
-}
-
-void UYVYToUVRow_NEON(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // stride + src_uyvy
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 16 pixels of UYVY.
-    "subs       %4, %4, #16                    \n"  // 16 pixels = 8 UVs.
-    "vld4.8     {d4, d5, d6, d7}, [%1]!        \n"  // load next row UYVY.
-    "vrhadd.u8  d0, d0, d4                     \n"  // average rows of U
-    "vrhadd.u8  d2, d2, d6                     \n"  // average rows of V
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 U.
-    "vst1.8     {d2}, [%3]!                    \n"  // store 8 V.
-    "bgt        1b                             \n"
-  : "+r"(src_uyvy),     // %0
-    "+r"(stride_uyvy),  // %1
-    "+r"(dst_u),        // %2
-    "+r"(dst_v),        // %3
-    "+r"(pix)           // %4
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7"  // Clobber List
-  );
-}
-
-void HalfRow_NEON(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix) {
-  asm volatile (
-    // change the stride to row 2 pointer
-    "add        %1, %0                         \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load row 1 16 pixels.
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop
-    "vld1.8     {q1}, [%1]!                    \n"  // load row 2 16 pixels.
-    "vrhadd.u8  q0, q1                         \n"  // average row 1 and 2
-    "vst1.8     {q0}, [%2]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_uv),         // %0
-    "+r"(src_uv_stride),  // %1
-    "+r"(dst_uv),         // %2
-    "+r"(pix)             // %3
-  :
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// Select 2 channels from ARGB on alternating pixels.  e.g.  BGBGBGBG
-void ARGBToBayerRow_NEON(const uint8* src_argb, uint8* dst_bayer,
-                         uint32 selector, int pix) {
-  asm volatile (
-    "vmov.u32   d6[0], %3                      \n"  // selector
-  "1:                                          \n"
-    "vld1.8     {q0, q1}, [%0]!                \n"  // load row 8 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop
-    "vtbl.8     d4, {d0, d1}, d6               \n"  // look up 4 pixels
-    "vtbl.8     d5, {d2, d3}, d6               \n"  // look up 4 pixels
-    "vtrn.u32   d4, d5                         \n"  // combine 8 pixels
-    "vst1.8     {d4}, [%1]!                    \n"  // store 8.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_bayer),  // %1
-    "+r"(pix)         // %2
-  : "r"(selector)     // %3
-  : "cc", "memory", "q0", "q1", "q2", "q3"  // Clobber List
-  );
-}
-
-// Select G channels from ARGB.  e.g.  GGGGGGGG
-void ARGBToBayerGGRow_NEON(const uint8* src_argb, uint8* dst_bayer,
-                           uint32 /*selector*/, int pix) {
-  asm volatile (
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load row 8 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop
-    "vst1.8     {d1}, [%1]!                    \n"  // store 8 G's.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_bayer),  // %1
-    "+r"(pix)         // %2
-  :
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// For BGRAToARGB, ABGRToARGB, RGBAToARGB, and ARGBToRGBA.
-void ARGBShuffleRow_NEON(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix) {
-  asm volatile (
-    "vld1.8     {q2}, [%3]                     \n"  // shuffler
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 4 pixels.
-    "subs       %2, %2, #4                     \n"  // 4 processed per loop
-    "vtbl.8     d2, {d0, d1}, d4               \n"  // look up 2 first pixels
-    "vtbl.8     d3, {d0, d1}, d5               \n"  // look up 2 next pixels
-    "vst1.8     {q1}, [%1]!                    \n"  // store 4.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "r"(shuffler)    // %3
-  : "cc", "memory", "q0", "q1", "q2"  // Clobber List
-  );
-}
-
-void I422ToYUY2Row_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_yuy2, int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {d0, d2}, [%0]!                \n"  // load 16 Ys
-    "vld1.8     {d1}, [%1]!                    \n"  // load 8 Us
-    "vld1.8     {d3}, [%2]!                    \n"  // load 8 Vs
-    "subs       %4, %4, #16                    \n"  // 16 pixels
-    "vst4.8     {d0, d1, d2, d3}, [%3]!        \n"  // Store 8 YUY2/16 pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_y),     // %0
-    "+r"(src_u),     // %1
-    "+r"(src_v),     // %2
-    "+r"(dst_yuy2),  // %3
-    "+r"(width)      // %4
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3"
-  );
-}
-
-void I422ToUYVYRow_NEON(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_uyvy, int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld2.8     {d1, d3}, [%0]!                \n"  // load 16 Ys
-    "vld1.8     {d0}, [%1]!                    \n"  // load 8 Us
-    "vld1.8     {d2}, [%2]!                    \n"  // load 8 Vs
-    "subs       %4, %4, #16                    \n"  // 16 pixels
-    "vst4.8     {d0, d1, d2, d3}, [%3]!        \n"  // Store 8 UYVY/16 pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_y),     // %0
-    "+r"(src_u),     // %1
-    "+r"(src_v),     // %2
-    "+r"(dst_uyvy),  // %3
-    "+r"(width)      // %4
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3"
-  );
-}
-
-void ARGBToRGB565Row_NEON(const uint8* src_argb, uint8* dst_rgb565, int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d20, d21, d22, d23}, [%0]!    \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGBTORGB565
-    "vst1.8     {q0}, [%1]!                    \n"  // store 8 pixels RGB565.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_rgb565),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q8", "q9", "q10", "q11"
-  );
-}
-
-void ARGBToARGB1555Row_NEON(const uint8* src_argb, uint8* dst_argb1555,
-                            int pix) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d20, d21, d22, d23}, [%0]!    \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGBTOARGB1555
-    "vst1.8     {q0}, [%1]!                    \n"  // store 8 pixels ARGB1555.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb1555),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q8", "q9", "q10", "q11"
-  );
-}
-
-void ARGBToARGB4444Row_NEON(const uint8* src_argb, uint8* dst_argb4444,
-                            int pix) {
-  asm volatile (
-    "vmov.u8    d4, #0x0f                      \n"  // bits to clear with vbic.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d20, d21, d22, d23}, [%0]!    \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGBTOARGB4444
-    "vst1.8     {q0}, [%1]!                    \n"  // store 8 pixels ARGB4444.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),      // %0
-    "+r"(dst_argb4444),  // %1
-    "+r"(pix)            // %2
-  :
-  : "cc", "memory", "q0", "q8", "q9", "q10", "q11"
-  );
-}
-
-void ARGBToYRow_NEON(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d24, #13                       \n"  // B * 0.1016 coefficient
-    "vmov.u8    d25, #65                       \n"  // G * 0.5078 coefficient
-    "vmov.u8    d26, #33                       \n"  // R * 0.2578 coefficient
-    "vmov.u8    d27, #16                       \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d27                        \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q12", "q13"
-  );
-}
-
-void ARGBToYJRow_NEON(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d24, #15                       \n"  // B * 0.11400 coefficient
-    "vmov.u8    d25, #75                       \n"  // G * 0.58700 coefficient
-    "vmov.u8    d26, #38                       \n"  // R * 0.29900 coefficient
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 15 bit to 8 bit Y
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q12", "q13"
-  );
-}
-
-// 8x1 pixels.
-void ARGBToUV444Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) {
-  asm volatile (
-    "vmov.u8    d24, #112                      \n"  // UB / VR 0.875 coefficient
-    "vmov.u8    d25, #74                       \n"  // UG -0.5781 coefficient
-    "vmov.u8    d26, #38                       \n"  // UR -0.2969 coefficient
-    "vmov.u8    d27, #18                       \n"  // VB -0.1406 coefficient
-    "vmov.u8    d28, #94                       \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlsl.u8   q2, d1, d25                    \n"  // G
-    "vmlsl.u8   q2, d2, d26                    \n"  // R
-    "vadd.u16   q2, q2, q15                    \n"  // +128 -> unsigned
-
-    "vmull.u8   q3, d2, d24                    \n"  // R
-    "vmlsl.u8   q3, d1, d28                    \n"  // G
-    "vmlsl.u8   q3, d0, d27                    \n"  // B
-    "vadd.u16   q3, q3, q15                    \n"  // +128 -> unsigned
-
-    "vqshrn.u16  d0, q2, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q3, #8                    \n"  // 16 bit to 8 bit V
-
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%2]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(pix)        // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 16x1 pixels -> 8x1.  pix is number of argb pixels. e.g. 16.
-void ARGBToUV422Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) {
-  asm volatile (
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop.
-    "vmul.s16   q8, q0, q10                    \n"  // B
-    "vmls.s16   q8, q1, q11                    \n"  // G
-    "vmls.s16   q8, q2, q12                    \n"  // R
-    "vadd.u16   q8, q8, q15                    \n"  // +128 -> unsigned
-
-    "vmul.s16   q9, q2, q10                    \n"  // R
-    "vmls.s16   q9, q1, q14                    \n"  // G
-    "vmls.s16   q9, q0, q13                    \n"  // B
-    "vadd.u16   q9, q9, q15                    \n"  // +128 -> unsigned
-
-    "vqshrn.u16  d0, q8, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q9, #8                    \n"  // 16 bit to 8 bit V
-
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%2]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(pix)        // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 32x1 pixels -> 8x1.  pix is number of argb pixels. e.g. 32.
-void ARGBToUV411Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) {
-  asm volatile (
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%0]!     \n"  // load 8 more ARGB pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%0]!     \n"  // load last 8 ARGB pixels.
-    "vpaddl.u8  q4, q4                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q5, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q6, q6                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vpadd.u16  d0, d0, d1                     \n"  // B 16 shorts -> 8 shorts.
-    "vpadd.u16  d1, d8, d9                     \n"  // B
-    "vpadd.u16  d2, d2, d3                     \n"  // G 16 shorts -> 8 shorts.
-    "vpadd.u16  d3, d10, d11                   \n"  // G
-    "vpadd.u16  d4, d4, d5                     \n"  // R 16 shorts -> 8 shorts.
-    "vpadd.u16  d5, d12, d13                   \n"  // R
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %3, %3, #32                    \n"  // 32 processed per loop.
-    "vmul.s16   q8, q0, q10                    \n"  // B
-    "vmls.s16   q8, q1, q11                    \n"  // G
-    "vmls.s16   q8, q2, q12                    \n"  // R
-    "vadd.u16   q8, q8, q15                    \n"  // +128 -> unsigned
-    "vmul.s16   q9, q2, q10                    \n"  // R
-    "vmls.s16   q9, q1, q14                    \n"  // G
-    "vmls.s16   q9, q0, q13                    \n"  // B
-    "vadd.u16   q9, q9, q15                    \n"  // +128 -> unsigned
-    "vqshrn.u16  d0, q8, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q9, #8                    \n"  // 16 bit to 8 bit V
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%2]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_u),     // %1
-    "+r"(dst_v),     // %2
-    "+r"(pix)        // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 16x2 pixels -> 8x1.  pix is number of argb pixels. e.g. 16.
-#define RGBTOUV(QB, QG, QR) \
-    "vmul.s16   q8, " #QB ", q10               \n"  /* B                    */ \
-    "vmls.s16   q8, " #QG ", q11               \n"  /* G                    */ \
-    "vmls.s16   q8, " #QR ", q12               \n"  /* R                    */ \
-    "vadd.u16   q8, q8, q15                    \n"  /* +128 -> unsigned     */ \
-    "vmul.s16   q9, " #QR ", q10               \n"  /* R                    */ \
-    "vmls.s16   q9, " #QG ", q14               \n"  /* G                    */ \
-    "vmls.s16   q9, " #QB ", q13               \n"  /* B                    */ \
-    "vadd.u16   q9, q9, q15                    \n"  /* +128 -> unsigned     */ \
-    "vqshrn.u16  d0, q8, #8                    \n"  /* 16 bit to 8 bit U    */ \
-    "vqshrn.u16  d1, q9, #8                    \n"  /* 16 bit to 8 bit V    */
-
-// TODO(fbarchard): Consider vhadd vertical, then vpaddl horizontal, avoid shr.
-void ARGBToUVRow_NEON(const uint8* src_argb, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_argb
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%1]!     \n"  // load 8 more ARGB pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%1]!     \n"  // load last 8 ARGB pixels.
-    "vpadal.u8  q0, q4                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q6                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q0, q1, q2)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(src_stride_argb),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// TODO(fbarchard): Subsample match C code.
-void ARGBToUVJRow_NEON(const uint8* src_argb, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_argb
-    "vmov.s16   q10, #127 / 2                  \n"  // UB / VR 0.500 coefficient
-    "vmov.s16   q11, #84 / 2                   \n"  // UG -0.33126 coefficient
-    "vmov.s16   q12, #43 / 2                   \n"  // UR -0.16874 coefficient
-    "vmov.s16   q13, #20 / 2                   \n"  // VB -0.08131 coefficient
-    "vmov.s16   q14, #107 / 2                  \n"  // VG -0.41869 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%1]!     \n"  // load 8 more ARGB pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%1]!     \n"  // load last 8 ARGB pixels.
-    "vpadal.u8  q0, q4                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q6                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q0, q1, q2)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(src_stride_argb),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void BGRAToUVRow_NEON(const uint8* src_bgra, int src_stride_bgra,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_bgra
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 BGRA pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 BGRA pixels.
-    "vpaddl.u8  q3, q3                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%1]!     \n"  // load 8 more BGRA pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%1]!     \n"  // load last 8 BGRA pixels.
-    "vpadal.u8  q3, q7                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q6                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q1, q1, #1                     \n"  // 2x average
-    "vrshr.u16  q2, q2, #1                     \n"
-    "vrshr.u16  q3, q3, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q3, q2, q1)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_bgra),  // %0
-    "+r"(src_stride_bgra),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void ABGRToUVRow_NEON(const uint8* src_abgr, int src_stride_abgr,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_abgr
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ABGR pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ABGR pixels.
-    "vpaddl.u8  q2, q2                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q0, q0                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%1]!     \n"  // load 8 more ABGR pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%1]!     \n"  // load last 8 ABGR pixels.
-    "vpadal.u8  q2, q6                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q0, q4                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q2, q1, q0)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_abgr),  // %0
-    "+r"(src_stride_abgr),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void RGBAToUVRow_NEON(const uint8* src_rgba, int src_stride_rgba,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_rgba
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 RGBA pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 RGBA pixels.
-    "vpaddl.u8  q0, q1                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q2                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q3                         \n"  // R 16 bytes -> 8 shorts.
-    "vld4.8     {d8, d10, d12, d14}, [%1]!     \n"  // load 8 more RGBA pixels.
-    "vld4.8     {d9, d11, d13, d15}, [%1]!     \n"  // load last 8 RGBA pixels.
-    "vpadal.u8  q0, q5                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q6                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q7                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q0, q1, q2)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_rgba),  // %0
-    "+r"(src_stride_rgba),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void RGB24ToUVRow_NEON(const uint8* src_rgb24, int src_stride_rgb24,
-                       uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_rgb24
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d0, d2, d4}, [%0]!            \n"  // load 8 RGB24 pixels.
-    "vld3.8     {d1, d3, d5}, [%0]!            \n"  // load next 8 RGB24 pixels.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vld3.8     {d8, d10, d12}, [%1]!          \n"  // load 8 more RGB24 pixels.
-    "vld3.8     {d9, d11, d13}, [%1]!          \n"  // load last 8 RGB24 pixels.
-    "vpadal.u8  q0, q4                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q6                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q0, q1, q2)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb24),  // %0
-    "+r"(src_stride_rgb24),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void RAWToUVRow_NEON(const uint8* src_raw, int src_stride_raw,
-                     uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_raw
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d0, d2, d4}, [%0]!            \n"  // load 8 RAW pixels.
-    "vld3.8     {d1, d3, d5}, [%0]!            \n"  // load next 8 RAW pixels.
-    "vpaddl.u8  q2, q2                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q0, q0                         \n"  // R 16 bytes -> 8 shorts.
-    "vld3.8     {d8, d10, d12}, [%1]!          \n"  // load 8 more RAW pixels.
-    "vld3.8     {d9, d11, d13}, [%1]!          \n"  // load last 8 RAW pixels.
-    "vpadal.u8  q2, q6                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q5                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q0, q4                         \n"  // R 16 bytes -> 8 shorts.
-
-    "vrshr.u16  q0, q0, #1                     \n"  // 2x average
-    "vrshr.u16  q1, q1, #1                     \n"
-    "vrshr.u16  q2, q2, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 32 processed per loop.
-    RGBTOUV(q2, q1, q0)
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_raw),  // %0
-    "+r"(src_stride_raw),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 16x2 pixels -> 8x1.  pix is number of argb pixels. e.g. 16.
-void RGB565ToUVRow_NEON(const uint8* src_rgb565, int src_stride_rgb565,
-                        uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_argb
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 RGB565 pixels.
-    RGB565TOARGB
-    "vpaddl.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%0]!                    \n"  // next 8 RGB565 pixels.
-    RGB565TOARGB
-    "vpaddl.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vld1.8     {q0}, [%1]!                    \n"  // load 8 RGB565 pixels.
-    RGB565TOARGB
-    "vpadal.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%1]!                    \n"  // next 8 RGB565 pixels.
-    RGB565TOARGB
-    "vpadal.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vrshr.u16  q4, q4, #1                     \n"  // 2x average
-    "vrshr.u16  q5, q5, #1                     \n"
-    "vrshr.u16  q6, q6, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 16 processed per loop.
-    "vmul.s16   q8, q4, q10                    \n"  // B
-    "vmls.s16   q8, q5, q11                    \n"  // G
-    "vmls.s16   q8, q6, q12                    \n"  // R
-    "vadd.u16   q8, q8, q15                    \n"  // +128 -> unsigned
-    "vmul.s16   q9, q6, q10                    \n"  // R
-    "vmls.s16   q9, q5, q14                    \n"  // G
-    "vmls.s16   q9, q4, q13                    \n"  // B
-    "vadd.u16   q9, q9, q15                    \n"  // +128 -> unsigned
-    "vqshrn.u16  d0, q8, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q9, #8                    \n"  // 16 bit to 8 bit V
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb565),  // %0
-    "+r"(src_stride_rgb565),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 16x2 pixels -> 8x1.  pix is number of argb pixels. e.g. 16.
-void ARGB1555ToUVRow_NEON(const uint8* src_argb1555, int src_stride_argb1555,
-                        uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_argb
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB1555 pixels.
-    RGB555TOARGB
-    "vpaddl.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%0]!                    \n"  // next 8 ARGB1555 pixels.
-    RGB555TOARGB
-    "vpaddl.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vld1.8     {q0}, [%1]!                    \n"  // load 8 ARGB1555 pixels.
-    RGB555TOARGB
-    "vpadal.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%1]!                    \n"  // next 8 ARGB1555 pixels.
-    RGB555TOARGB
-    "vpadal.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vrshr.u16  q4, q4, #1                     \n"  // 2x average
-    "vrshr.u16  q5, q5, #1                     \n"
-    "vrshr.u16  q6, q6, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 16 processed per loop.
-    "vmul.s16   q8, q4, q10                    \n"  // B
-    "vmls.s16   q8, q5, q11                    \n"  // G
-    "vmls.s16   q8, q6, q12                    \n"  // R
-    "vadd.u16   q8, q8, q15                    \n"  // +128 -> unsigned
-    "vmul.s16   q9, q6, q10                    \n"  // R
-    "vmls.s16   q9, q5, q14                    \n"  // G
-    "vmls.s16   q9, q4, q13                    \n"  // B
-    "vadd.u16   q9, q9, q15                    \n"  // +128 -> unsigned
-    "vqshrn.u16  d0, q8, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q9, #8                    \n"  // 16 bit to 8 bit V
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb1555),  // %0
-    "+r"(src_stride_argb1555),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// 16x2 pixels -> 8x1.  pix is number of argb pixels. e.g. 16.
-void ARGB4444ToUVRow_NEON(const uint8* src_argb4444, int src_stride_argb4444,
-                          uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "add        %1, %0, %1                     \n"  // src_stride + src_argb
-    "vmov.s16   q10, #112 / 2                  \n"  // UB / VR 0.875 coefficient
-    "vmov.s16   q11, #74 / 2                   \n"  // UG -0.5781 coefficient
-    "vmov.s16   q12, #38 / 2                   \n"  // UR -0.2969 coefficient
-    "vmov.s16   q13, #18 / 2                   \n"  // VB -0.1406 coefficient
-    "vmov.s16   q14, #94 / 2                   \n"  // VG -0.7344 coefficient
-    "vmov.u16   q15, #0x8080                   \n"  // 128.5
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB4444 pixels.
-    ARGB4444TOARGB
-    "vpaddl.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%0]!                    \n"  // next 8 ARGB4444 pixels.
-    ARGB4444TOARGB
-    "vpaddl.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpaddl.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpaddl.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vld1.8     {q0}, [%1]!                    \n"  // load 8 ARGB4444 pixels.
-    ARGB4444TOARGB
-    "vpadal.u8  d8, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d10, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d12, d2                        \n"  // R 8 bytes -> 4 shorts.
-    "vld1.8     {q0}, [%1]!                    \n"  // next 8 ARGB4444 pixels.
-    ARGB4444TOARGB
-    "vpadal.u8  d9, d0                         \n"  // B 8 bytes -> 4 shorts.
-    "vpadal.u8  d11, d1                        \n"  // G 8 bytes -> 4 shorts.
-    "vpadal.u8  d13, d2                        \n"  // R 8 bytes -> 4 shorts.
-
-    "vrshr.u16  q4, q4, #1                     \n"  // 2x average
-    "vrshr.u16  q5, q5, #1                     \n"
-    "vrshr.u16  q6, q6, #1                     \n"
-
-    "subs       %4, %4, #16                    \n"  // 16 processed per loop.
-    "vmul.s16   q8, q4, q10                    \n"  // B
-    "vmls.s16   q8, q5, q11                    \n"  // G
-    "vmls.s16   q8, q6, q12                    \n"  // R
-    "vadd.u16   q8, q8, q15                    \n"  // +128 -> unsigned
-    "vmul.s16   q9, q6, q10                    \n"  // R
-    "vmls.s16   q9, q5, q14                    \n"  // G
-    "vmls.s16   q9, q4, q13                    \n"  // B
-    "vadd.u16   q9, q9, q15                    \n"  // +128 -> unsigned
-    "vqshrn.u16  d0, q8, #8                    \n"  // 16 bit to 8 bit U
-    "vqshrn.u16  d1, q9, #8                    \n"  // 16 bit to 8 bit V
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 pixels U.
-    "vst1.8     {d1}, [%3]!                    \n"  // store 8 pixels V.
-    "bgt        1b                             \n"
-  : "+r"(src_argb4444),  // %0
-    "+r"(src_stride_argb4444),  // %1
-    "+r"(dst_u),     // %2
-    "+r"(dst_v),     // %3
-    "+r"(pix)        // %4
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7",
-    "q8", "q9", "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-void RGB565ToYRow_NEON(const uint8* src_rgb565, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d24, #13                       \n"  // B * 0.1016 coefficient
-    "vmov.u8    d25, #65                       \n"  // G * 0.5078 coefficient
-    "vmov.u8    d26, #33                       \n"  // R * 0.2578 coefficient
-    "vmov.u8    d27, #16                       \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 RGB565 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    RGB565TOARGB
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d27                        \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb565),  // %0
-    "+r"(dst_y),       // %1
-    "+r"(pix)          // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q12", "q13"
-  );
-}
-
-void ARGB1555ToYRow_NEON(const uint8* src_argb1555, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d24, #13                       \n"  // B * 0.1016 coefficient
-    "vmov.u8    d25, #65                       \n"  // G * 0.5078 coefficient
-    "vmov.u8    d26, #33                       \n"  // R * 0.2578 coefficient
-    "vmov.u8    d27, #16                       \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB1555 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGB1555TOARGB
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d27                        \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_argb1555),  // %0
-    "+r"(dst_y),         // %1
-    "+r"(pix)            // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q12", "q13"
-  );
-}
-
-void ARGB4444ToYRow_NEON(const uint8* src_argb4444, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d24, #13                       \n"  // B * 0.1016 coefficient
-    "vmov.u8    d25, #65                       \n"  // G * 0.5078 coefficient
-    "vmov.u8    d26, #33                       \n"  // R * 0.2578 coefficient
-    "vmov.u8    d27, #16                       \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 8 ARGB4444 pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    ARGB4444TOARGB
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d27                        \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_argb4444),  // %0
-    "+r"(dst_y),         // %1
-    "+r"(pix)            // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q12", "q13"
-  );
-}
-
-void BGRAToYRow_NEON(const uint8* src_bgra, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #33                        \n"  // R * 0.2578 coefficient
-    "vmov.u8    d5, #65                        \n"  // G * 0.5078 coefficient
-    "vmov.u8    d6, #13                        \n"  // B * 0.1016 coefficient
-    "vmov.u8    d7, #16                        \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 pixels of BGRA.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q8, d1, d4                     \n"  // R
-    "vmlal.u8   q8, d2, d5                     \n"  // G
-    "vmlal.u8   q8, d3, d6                     \n"  // B
-    "vqrshrun.s16 d0, q8, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d7                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_bgra),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", "q8"
-  );
-}
-
-void ABGRToYRow_NEON(const uint8* src_abgr, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #33                        \n"  // R * 0.2578 coefficient
-    "vmov.u8    d5, #65                        \n"  // G * 0.5078 coefficient
-    "vmov.u8    d6, #13                        \n"  // B * 0.1016 coefficient
-    "vmov.u8    d7, #16                        \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 pixels of ABGR.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q8, d0, d4                     \n"  // R
-    "vmlal.u8   q8, d1, d5                     \n"  // G
-    "vmlal.u8   q8, d2, d6                     \n"  // B
-    "vqrshrun.s16 d0, q8, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d7                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_abgr),  // %0
-    "+r"(dst_y),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", "q8"
-  );
-}
-
-void RGBAToYRow_NEON(const uint8* src_rgba, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #13                        \n"  // B * 0.1016 coefficient
-    "vmov.u8    d5, #65                        \n"  // G * 0.5078 coefficient
-    "vmov.u8    d6, #33                        \n"  // R * 0.2578 coefficient
-    "vmov.u8    d7, #16                        \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 pixels of RGBA.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q8, d1, d4                     \n"  // B
-    "vmlal.u8   q8, d2, d5                     \n"  // G
-    "vmlal.u8   q8, d3, d6                     \n"  // R
-    "vqrshrun.s16 d0, q8, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d7                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_rgba),  // %0
-    "+r"(dst_y),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", "q8"
-  );
-}
-
-void RGB24ToYRow_NEON(const uint8* src_rgb24, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #13                        \n"  // B * 0.1016 coefficient
-    "vmov.u8    d5, #65                        \n"  // G * 0.5078 coefficient
-    "vmov.u8    d6, #33                        \n"  // R * 0.2578 coefficient
-    "vmov.u8    d7, #16                        \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d0, d1, d2}, [%0]!            \n"  // load 8 pixels of RGB24.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q8, d0, d4                     \n"  // B
-    "vmlal.u8   q8, d1, d5                     \n"  // G
-    "vmlal.u8   q8, d2, d6                     \n"  // R
-    "vqrshrun.s16 d0, q8, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d7                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_rgb24),  // %0
-    "+r"(dst_y),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", "q8"
-  );
-}
-
-void RAWToYRow_NEON(const uint8* src_raw, uint8* dst_y, int pix) {
-  asm volatile (
-    "vmov.u8    d4, #33                        \n"  // R * 0.2578 coefficient
-    "vmov.u8    d5, #65                        \n"  // G * 0.5078 coefficient
-    "vmov.u8    d6, #13                        \n"  // B * 0.1016 coefficient
-    "vmov.u8    d7, #16                        \n"  // Add 16 constant
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld3.8     {d0, d1, d2}, [%0]!            \n"  // load 8 pixels of RAW.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q8, d0, d4                     \n"  // B
-    "vmlal.u8   q8, d1, d5                     \n"  // G
-    "vmlal.u8   q8, d2, d6                     \n"  // R
-    "vqrshrun.s16 d0, q8, #7                   \n"  // 16 bit to 8 bit Y
-    "vqadd.u8   d0, d7                         \n"
-    "vst1.8     {d0}, [%1]!                    \n"  // store 8 pixels Y.
-    "bgt        1b                             \n"
-  : "+r"(src_raw),  // %0
-    "+r"(dst_y),  // %1
-    "+r"(pix)        // %2
-  :
-  : "cc", "memory", "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", "q8"
-  );
-}
-
-// Bilinear filter 16x2 -> 16x1
-void InterpolateRow_NEON(uint8* dst_ptr,
-                         const uint8* src_ptr, ptrdiff_t src_stride,
-                         int dst_width, int source_y_fraction) {
-  asm volatile (
-    "cmp        %4, #0                         \n"
-    "beq        100f                           \n"
-    "add        %2, %1                         \n"
-    "cmp        %4, #64                        \n"
-    "beq        75f                            \n"
-    "cmp        %4, #128                       \n"
-    "beq        50f                            \n"
-    "cmp        %4, #192                       \n"
-    "beq        25f                            \n"
-
-    "vdup.8     d5, %4                         \n"
-    "rsb        %4, #256                       \n"
-    "vdup.8     d4, %4                         \n"
-    // General purpose row blend.
-  "1:                                          \n"
-    "vld1.8     {q0}, [%1]!                    \n"
-    "vld1.8     {q1}, [%2]!                    \n"
-    "subs       %3, %3, #16                    \n"
-    "vmull.u8   q13, d0, d4                    \n"
-    "vmull.u8   q14, d1, d4                    \n"
-    "vmlal.u8   q13, d2, d5                    \n"
-    "vmlal.u8   q14, d3, d5                    \n"
-    "vrshrn.u16 d0, q13, #8                    \n"
-    "vrshrn.u16 d1, q14, #8                    \n"
-    "vst1.8     {q0}, [%0]!                    \n"
-    "bgt        1b                             \n"
-    "b          99f                            \n"
-
-    // Blend 25 / 75.
-  "25:                                         \n"
-    "vld1.8     {q0}, [%1]!                    \n"
-    "vld1.8     {q1}, [%2]!                    \n"
-    "subs       %3, %3, #16                    \n"
-    "vrhadd.u8  q0, q1                         \n"
-    "vrhadd.u8  q0, q1                         \n"
-    "vst1.8     {q0}, [%0]!                    \n"
-    "bgt        25b                            \n"
-    "b          99f                            \n"
-
-    // Blend 50 / 50.
-  "50:                                         \n"
-    "vld1.8     {q0}, [%1]!                    \n"
-    "vld1.8     {q1}, [%2]!                    \n"
-    "subs       %3, %3, #16                    \n"
-    "vrhadd.u8  q0, q1                         \n"
-    "vst1.8     {q0}, [%0]!                    \n"
-    "bgt        50b                            \n"
-    "b          99f                            \n"
-
-    // Blend 75 / 25.
-  "75:                                         \n"
-    "vld1.8     {q1}, [%1]!                    \n"
-    "vld1.8     {q0}, [%2]!                    \n"
-    "subs       %3, %3, #16                    \n"
-    "vrhadd.u8  q0, q1                         \n"
-    "vrhadd.u8  q0, q1                         \n"
-    "vst1.8     {q0}, [%0]!                    \n"
-    "bgt        75b                            \n"
-    "b          99f                            \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-  "100:                                        \n"
-    "vld1.8     {q0}, [%1]!                    \n"
-    "subs       %3, %3, #16                    \n"
-    "vst1.8     {q0}, [%0]!                    \n"
-    "bgt        100b                           \n"
-
-  "99:                                         \n"
-  : "+r"(dst_ptr),          // %0
-    "+r"(src_ptr),          // %1
-    "+r"(src_stride),       // %2
-    "+r"(dst_width),        // %3
-    "+r"(source_y_fraction) // %4
-  :
-  : "cc", "memory", "q0", "q1", "d4", "d5", "q13", "q14"
-  );
-}
-
-// dr * (256 - sa) / 256 + sr = dr - dr * sa / 256 + sr
-void ARGBBlendRow_NEON(const uint8* src_argb0, const uint8* src_argb1,
-                       uint8* dst_argb, int width) {
-  asm volatile (
-    "subs       %3, #8                         \n"
-    "blt        89f                            \n"
-    // Blend 8 pixels.
-  "8:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 pixels of ARGB0.
-    "vld4.8     {d4, d5, d6, d7}, [%1]!        \n"  // load 8 pixels of ARGB1.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q10, d4, d3                    \n"  // db * a
-    "vmull.u8   q11, d5, d3                    \n"  // dg * a
-    "vmull.u8   q12, d6, d3                    \n"  // dr * a
-    "vqrshrn.u16 d20, q10, #8                  \n"  // db >>= 8
-    "vqrshrn.u16 d21, q11, #8                  \n"  // dg >>= 8
-    "vqrshrn.u16 d22, q12, #8                  \n"  // dr >>= 8
-    "vqsub.u8   q2, q2, q10                    \n"  // dbg - dbg * a / 256
-    "vqsub.u8   d6, d6, d22                    \n"  // dr - dr * a / 256
-    "vqadd.u8   q0, q0, q2                     \n"  // + sbg
-    "vqadd.u8   d2, d2, d6                     \n"  // + sr
-    "vmov.u8    d3, #255                       \n"  // a = 255
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 pixels of ARGB.
-    "bge        8b                             \n"
-
-  "89:                                         \n"
-    "adds       %3, #8-1                       \n"
-    "blt        99f                            \n"
-
-    // Blend 1 pixels.
-  "1:                                          \n"
-    "vld4.8     {d0[0],d1[0],d2[0],d3[0]}, [%0]! \n"  // load 1 pixel ARGB0.
-    "vld4.8     {d4[0],d5[0],d6[0],d7[0]}, [%1]! \n"  // load 1 pixel ARGB1.
-    "subs       %3, %3, #1                     \n"  // 1 processed per loop.
-    "vmull.u8   q10, d4, d3                    \n"  // db * a
-    "vmull.u8   q11, d5, d3                    \n"  // dg * a
-    "vmull.u8   q12, d6, d3                    \n"  // dr * a
-    "vqrshrn.u16 d20, q10, #8                  \n"  // db >>= 8
-    "vqrshrn.u16 d21, q11, #8                  \n"  // dg >>= 8
-    "vqrshrn.u16 d22, q12, #8                  \n"  // dr >>= 8
-    "vqsub.u8   q2, q2, q10                    \n"  // dbg - dbg * a / 256
-    "vqsub.u8   d6, d6, d22                    \n"  // dr - dr * a / 256
-    "vqadd.u8   q0, q0, q2                     \n"  // + sbg
-    "vqadd.u8   d2, d2, d6                     \n"  // + sr
-    "vmov.u8    d3, #255                       \n"  // a = 255
-    "vst4.8     {d0[0],d1[0],d2[0],d3[0]}, [%2]! \n"  // store 1 pixel.
-    "bge        1b                             \n"
-
-  "99:                                         \n"
-
-  : "+r"(src_argb0),    // %0
-    "+r"(src_argb1),    // %1
-    "+r"(dst_argb),     // %2
-    "+r"(width)         // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q10", "q11", "q12"
-  );
-}
-
-// Attenuate 8 pixels at a time.
-void ARGBAttenuateRow_NEON(const uint8* src_argb, uint8* dst_argb, int width) {
-  asm volatile (
-    // Attenuate 8 pixels.
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q10, d0, d3                    \n"  // b * a
-    "vmull.u8   q11, d1, d3                    \n"  // g * a
-    "vmull.u8   q12, d2, d3                    \n"  // r * a
-    "vqrshrn.u16 d0, q10, #8                   \n"  // b >>= 8
-    "vqrshrn.u16 d1, q11, #8                   \n"  // g >>= 8
-    "vqrshrn.u16 d2, q12, #8                   \n"  // r >>= 8
-    "vst4.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_argb),   // %1
-    "+r"(width)       // %2
-  :
-  : "cc", "memory", "q0", "q1", "q10", "q11", "q12"
-  );
-}
-
-// Quantize 8 ARGB pixels (32 bytes).
-// dst = (dst * scale >> 16) * interval_size + interval_offset;
-void ARGBQuantizeRow_NEON(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width) {
-  asm volatile (
-    "vdup.u16   q8, %2                         \n"
-    "vshr.u16   q8, q8, #1                     \n"  // scale >>= 1
-    "vdup.u16   q9, %3                         \n"  // interval multiply.
-    "vdup.u16   q10, %4                        \n"  // interval add
-
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]         \n"  // load 8 pixels of ARGB.
-    "subs       %1, %1, #8                     \n"  // 8 processed per loop.
-    "vmovl.u8   q0, d0                         \n"  // b (0 .. 255)
-    "vmovl.u8   q1, d2                         \n"
-    "vmovl.u8   q2, d4                         \n"
-    "vqdmulh.s16 q0, q0, q8                    \n"  // b * scale
-    "vqdmulh.s16 q1, q1, q8                    \n"  // g
-    "vqdmulh.s16 q2, q2, q8                    \n"  // r
-    "vmul.u16   q0, q0, q9                     \n"  // b * interval_size
-    "vmul.u16   q1, q1, q9                     \n"  // g
-    "vmul.u16   q2, q2, q9                     \n"  // r
-    "vadd.u16   q0, q0, q10                    \n"  // b + interval_offset
-    "vadd.u16   q1, q1, q10                    \n"  // g
-    "vadd.u16   q2, q2, q10                    \n"  // r
-    "vqmovn.u16 d0, q0                         \n"
-    "vqmovn.u16 d2, q1                         \n"
-    "vqmovn.u16 d4, q2                         \n"
-    "vst4.8     {d0, d2, d4, d6}, [%0]!        \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(dst_argb),       // %0
-    "+r"(width)           // %1
-  : "r"(scale),           // %2
-    "r"(interval_size),   // %3
-    "r"(interval_offset)  // %4
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q8", "q9", "q10"
-  );
-}
-
-// Shade 8 pixels at a time by specified value.
-// NOTE vqrdmulh.s16 q10, q10, d0[0] must use a scaler register from 0 to 8.
-// Rounding in vqrdmulh does +1 to high if high bit of low s16 is set.
-void ARGBShadeRow_NEON(const uint8* src_argb, uint8* dst_argb, int width,
-                       uint32 value) {
-  asm volatile (
-    "vdup.u32   q0, %3                         \n"  // duplicate scale value.
-    "vzip.u8    d0, d1                         \n"  // d0 aarrggbb.
-    "vshr.u16   q0, q0, #1                     \n"  // scale / 2.
-
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d20, d22, d24, d26}, [%0]!    \n"  // load 8 pixels of ARGB.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmovl.u8   q10, d20                       \n"  // b (0 .. 255)
-    "vmovl.u8   q11, d22                       \n"
-    "vmovl.u8   q12, d24                       \n"
-    "vmovl.u8   q13, d26                       \n"
-    "vqrdmulh.s16 q10, q10, d0[0]              \n"  // b * scale * 2
-    "vqrdmulh.s16 q11, q11, d0[1]              \n"  // g
-    "vqrdmulh.s16 q12, q12, d0[2]              \n"  // r
-    "vqrdmulh.s16 q13, q13, d0[3]              \n"  // a
-    "vqmovn.u16 d20, q10                       \n"
-    "vqmovn.u16 d22, q11                       \n"
-    "vqmovn.u16 d24, q12                       \n"
-    "vqmovn.u16 d26, q13                       \n"
-    "vst4.8     {d20, d22, d24, d26}, [%1]!    \n"  // store 8 pixels of ARGB.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),       // %0
-    "+r"(dst_argb),       // %1
-    "+r"(width)           // %2
-  : "r"(value)            // %3
-  : "cc", "memory", "q0", "q10", "q11", "q12", "q13"
-  );
-}
-
-// Convert 8 ARGB pixels (64 bytes) to 8 Gray ARGB pixels
-// Similar to ARGBToYJ but stores ARGB.
-// C code is (15 * b + 75 * g + 38 * r + 64) >> 7;
-void ARGBGrayRow_NEON(const uint8* src_argb, uint8* dst_argb, int width) {
-  asm volatile (
-    "vmov.u8    d24, #15                       \n"  // B * 0.11400 coefficient
-    "vmov.u8    d25, #75                       \n"  // G * 0.58700 coefficient
-    "vmov.u8    d26, #38                       \n"  // R * 0.29900 coefficient
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q2, d0, d24                    \n"  // B
-    "vmlal.u8   q2, d1, d25                    \n"  // G
-    "vmlal.u8   q2, d2, d26                    \n"  // R
-    "vqrshrun.s16 d0, q2, #7                   \n"  // 15 bit to 8 bit B
-    "vmov       d1, d0                         \n"  // G
-    "vmov       d2, d0                         \n"  // R
-    "vst4.8     {d0, d1, d2, d3}, [%1]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(width)      // %2
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q12", "q13"
-  );
-}
-
-// Convert 8 ARGB pixels (32 bytes) to 8 Sepia ARGB pixels.
-//    b = (r * 35 + g * 68 + b * 17) >> 7
-//    g = (r * 45 + g * 88 + b * 22) >> 7
-//    r = (r * 50 + g * 98 + b * 24) >> 7
-void ARGBSepiaRow_NEON(uint8* dst_argb, int width) {
-  asm volatile (
-    "vmov.u8    d20, #17                       \n"  // BB coefficient
-    "vmov.u8    d21, #68                       \n"  // BG coefficient
-    "vmov.u8    d22, #35                       \n"  // BR coefficient
-    "vmov.u8    d24, #22                       \n"  // GB coefficient
-    "vmov.u8    d25, #88                       \n"  // GG coefficient
-    "vmov.u8    d26, #45                       \n"  // GR coefficient
-    "vmov.u8    d28, #24                       \n"  // BB coefficient
-    "vmov.u8    d29, #98                       \n"  // BG coefficient
-    "vmov.u8    d30, #50                       \n"  // BR coefficient
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]         \n"  // load 8 ARGB pixels.
-    "subs       %1, %1, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q2, d0, d20                    \n"  // B to Sepia B
-    "vmlal.u8   q2, d1, d21                    \n"  // G
-    "vmlal.u8   q2, d2, d22                    \n"  // R
-    "vmull.u8   q3, d0, d24                    \n"  // B to Sepia G
-    "vmlal.u8   q3, d1, d25                    \n"  // G
-    "vmlal.u8   q3, d2, d26                    \n"  // R
-    "vmull.u8   q8, d0, d28                    \n"  // B to Sepia R
-    "vmlal.u8   q8, d1, d29                    \n"  // G
-    "vmlal.u8   q8, d2, d30                    \n"  // R
-    "vqshrn.u16 d0, q2, #7                     \n"  // 16 bit to 8 bit B
-    "vqshrn.u16 d1, q3, #7                     \n"  // 16 bit to 8 bit G
-    "vqshrn.u16 d2, q8, #7                     \n"  // 16 bit to 8 bit R
-    "vst4.8     {d0, d1, d2, d3}, [%0]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-  : "+r"(dst_argb),  // %0
-    "+r"(width)      // %1
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3",
-    "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// Tranform 8 ARGB pixels (32 bytes) with color matrix.
-// TODO(fbarchard): Was same as Sepia except matrix is provided.  This function
-// needs to saturate.  Consider doing a non-saturating version.
-void ARGBColorMatrixRow_NEON(const uint8* src_argb, uint8* dst_argb,
-                             const int8* matrix_argb, int width) {
-  asm volatile (
-    "vld1.8     {q2}, [%3]                     \n"  // load 3 ARGB vectors.
-    "vmovl.s8   q0, d4                         \n"  // B,G coefficients s16.
-    "vmovl.s8   q1, d5                         \n"  // R,A coefficients s16.
-
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d16, d18, d20, d22}, [%0]!    \n"  // load 8 ARGB pixels.
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop.
-    "vmovl.u8   q8, d16                        \n"  // b (0 .. 255) 16 bit
-    "vmovl.u8   q9, d18                        \n"  // g
-    "vmovl.u8   q10, d20                       \n"  // r
-    "vmovl.u8   q15, d22                       \n"  // a
-    "vmul.s16   q12, q8, d0[0]                 \n"  // B = B * Matrix B
-    "vmul.s16   q13, q8, d1[0]                 \n"  // G = B * Matrix G
-    "vmul.s16   q14, q8, d2[0]                 \n"  // R = B * Matrix R
-    "vmul.s16   q15, q8, d3[0]                 \n"  // A = B * Matrix A
-    "vmul.s16   q4, q9, d0[1]                  \n"  // B += G * Matrix B
-    "vmul.s16   q5, q9, d1[1]                  \n"  // G += G * Matrix G
-    "vmul.s16   q6, q9, d2[1]                  \n"  // R += G * Matrix R
-    "vmul.s16   q7, q9, d3[1]                  \n"  // A += G * Matrix A
-    "vqadd.s16  q12, q12, q4                   \n"  // Accumulate B
-    "vqadd.s16  q13, q13, q5                   \n"  // Accumulate G
-    "vqadd.s16  q14, q14, q6                   \n"  // Accumulate R
-    "vqadd.s16  q15, q15, q7                   \n"  // Accumulate A
-    "vmul.s16   q4, q10, d0[2]                 \n"  // B += R * Matrix B
-    "vmul.s16   q5, q10, d1[2]                 \n"  // G += R * Matrix G
-    "vmul.s16   q6, q10, d2[2]                 \n"  // R += R * Matrix R
-    "vmul.s16   q7, q10, d3[2]                 \n"  // A += R * Matrix A
-    "vqadd.s16  q12, q12, q4                   \n"  // Accumulate B
-    "vqadd.s16  q13, q13, q5                   \n"  // Accumulate G
-    "vqadd.s16  q14, q14, q6                   \n"  // Accumulate R
-    "vqadd.s16  q15, q15, q7                   \n"  // Accumulate A
-    "vmul.s16   q4, q15, d0[3]                 \n"  // B += A * Matrix B
-    "vmul.s16   q5, q15, d1[3]                 \n"  // G += A * Matrix G
-    "vmul.s16   q6, q15, d2[3]                 \n"  // R += A * Matrix R
-    "vmul.s16   q7, q15, d3[3]                 \n"  // A += A * Matrix A
-    "vqadd.s16  q12, q12, q4                   \n"  // Accumulate B
-    "vqadd.s16  q13, q13, q5                   \n"  // Accumulate G
-    "vqadd.s16  q14, q14, q6                   \n"  // Accumulate R
-    "vqadd.s16  q15, q15, q7                   \n"  // Accumulate A
-    "vqshrun.s16 d16, q12, #6                  \n"  // 16 bit to 8 bit B
-    "vqshrun.s16 d18, q13, #6                  \n"  // 16 bit to 8 bit G
-    "vqshrun.s16 d20, q14, #6                  \n"  // 16 bit to 8 bit R
-    "vqshrun.s16 d22, q15, #6                  \n"  // 16 bit to 8 bit A
-    "vst4.8     {d16, d18, d20, d22}, [%1]!    \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_argb),   // %1
-    "+r"(width)       // %2
-  : "r"(matrix_argb)  // %3
-  : "cc", "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7", "q8", "q9",
-    "q10", "q11", "q12", "q13", "q14", "q15"
-  );
-}
-
-// TODO(fbarchard): fix vqshrun in ARGBMultiplyRow_NEON and reenable.
-#ifdef HAS_ARGBMULTIPLYROW_NEON
-// Multiply 2 rows of ARGB pixels together, 8 pixels at a time.
-void ARGBMultiplyRow_NEON(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  asm volatile (
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%1]!        \n"  // load 8 more ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vmull.u8   q0, d0, d1                     \n"  // multiply B
-    "vmull.u8   q1, d2, d3                     \n"  // multiply G
-    "vmull.u8   q2, d4, d5                     \n"  // multiply R
-    "vmull.u8   q3, d6, d7                     \n"  // multiply A
-    "vrshrn.u16 d0, q0, #8                     \n"  // 16 bit to 8 bit B
-    "vrshrn.u16 d1, q1, #8                     \n"  // 16 bit to 8 bit G
-    "vrshrn.u16 d2, q2, #8                     \n"  // 16 bit to 8 bit R
-    "vrshrn.u16 d3, q3, #8                     \n"  // 16 bit to 8 bit A
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3"
-  );
-}
-#endif  // HAS_ARGBMULTIPLYROW_NEON
-
-// Add 2 rows of ARGB pixels together, 8 pixels at a time.
-void ARGBAddRow_NEON(const uint8* src_argb0, const uint8* src_argb1,
-                     uint8* dst_argb, int width) {
-  asm volatile (
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d4, d5, d6, d7}, [%1]!        \n"  // load 8 more ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vqadd.u8   q0, q0, q2                     \n"  // add B, G
-    "vqadd.u8   q1, q1, q3                     \n"  // add R, A
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3"
-  );
-}
-
-// Subtract 2 rows of ARGB pixels, 8 pixels at a time.
-void ARGBSubtractRow_NEON(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  asm volatile (
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d4, d5, d6, d7}, [%1]!        \n"  // load 8 more ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vqsub.u8   q0, q0, q2                     \n"  // subtract B, G
-    "vqsub.u8   q1, q1, q3                     \n"  // subtract R, A
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "cc", "memory", "q0", "q1", "q2", "q3"
-  );
-}
-
-// Adds Sobel X and Sobel Y and stores Sobel into ARGB.
-// A = 255
-// R = Sobel
-// G = Sobel
-// B = Sobel
-void SobelRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width) {
-  asm volatile (
-    "vmov.u8    d3, #255                       \n"  // alpha
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0}, [%0]!                    \n"  // load 8 sobelx.
-    "vld1.8     {d1}, [%1]!                    \n"  // load 8 sobely.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vqadd.u8   d0, d0, d1                     \n"  // add
-    "vmov.u8    d1, d0                         \n"
-    "vmov.u8    d2, d0                         \n"
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(width)        // %3
-  :
-  : "cc", "memory", "q0", "q1"
-  );
-}
-
-// Adds Sobel X and Sobel Y and stores Sobel into plane.
-void SobelToPlaneRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_y, int width) {
-  asm volatile (
-    // 16 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"  // load 16 sobelx.
-    "vld1.8     {q1}, [%1]!                    \n"  // load 16 sobely.
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop.
-    "vqadd.u8   q0, q0, q1                     \n"  // add
-    "vst1.8     {q0}, [%2]!                    \n"  // store 16 pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_y),       // %2
-    "+r"(width)        // %3
-  :
-  : "cc", "memory", "q0", "q1"
-  );
-}
-
-// Mixes Sobel X, Sobel Y and Sobel into ARGB.
-// A = 255
-// R = Sobel X
-// G = Sobel
-// B = Sobel Y
-void SobelXYRow_NEON(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width) {
-  asm volatile (
-    "vmov.u8    d3, #255                       \n"  // alpha
-    // 8 pixel loop.
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d2}, [%0]!                    \n"  // load 8 sobelx.
-    "vld1.8     {d0}, [%1]!                    \n"  // load 8 sobely.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vqadd.u8   d1, d0, d2                     \n"  // add
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"  // store 8 ARGB pixels.
-    "bgt        1b                             \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(width)        // %3
-  :
-  : "cc", "memory", "q0", "q1"
-  );
-}
-
-// SobelX as a matrix is
-// -1  0  1
-// -2  0  2
-// -1  0  1
-void SobelXRow_NEON(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobelx, int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0}, [%0],%5                  \n"  // top
-    "vld1.8     {d1}, [%0],%6                  \n"
-    "vsubl.u8   q0, d0, d1                     \n"
-    "vld1.8     {d2}, [%1],%5                  \n"  // center * 2
-    "vld1.8     {d3}, [%1],%6                  \n"
-    "vsubl.u8   q1, d2, d3                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vld1.8     {d2}, [%2],%5                  \n"  // bottom
-    "vld1.8     {d3}, [%2],%6                  \n"
-    "subs       %4, %4, #8                     \n"  // 8 pixels
-    "vsubl.u8   q1, d2, d3                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vabs.s16   q0, q0                         \n"
-    "vqmovn.u16 d0, q0                         \n"
-    "vst1.8     {d0}, [%3]!                    \n"  // store 8 sobelx
-    "bgt        1b                             \n"
-  : "+r"(src_y0),      // %0
-    "+r"(src_y1),      // %1
-    "+r"(src_y2),      // %2
-    "+r"(dst_sobelx),  // %3
-    "+r"(width)        // %4
-  : "r"(2),            // %5
-    "r"(6)             // %6
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-
-// SobelY as a matrix is
-// -1 -2 -1
-//  0  0  0
-//  1  2  1
-void SobelYRow_NEON(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0}, [%0],%4                  \n"  // left
-    "vld1.8     {d1}, [%1],%4                  \n"
-    "vsubl.u8   q0, d0, d1                     \n"
-    "vld1.8     {d2}, [%0],%4                  \n"  // center * 2
-    "vld1.8     {d3}, [%1],%4                  \n"
-    "vsubl.u8   q1, d2, d3                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vld1.8     {d2}, [%0],%5                  \n"  // right
-    "vld1.8     {d3}, [%1],%5                  \n"
-    "subs       %3, %3, #8                     \n"  // 8 pixels
-    "vsubl.u8   q1, d2, d3                     \n"
-    "vadd.s16   q0, q0, q1                     \n"
-    "vabs.s16   q0, q0                         \n"
-    "vqmovn.u16 d0, q0                         \n"
-    "vst1.8     {d0}, [%2]!                    \n"  // store 8 sobely
-    "bgt        1b                             \n"
-  : "+r"(src_y0),      // %0
-    "+r"(src_y1),      // %1
-    "+r"(dst_sobely),  // %2
-    "+r"(width)        // %3
-  : "r"(1),            // %4
-    "r"(6)             // %5
-  : "cc", "memory", "q0", "q1"  // Clobber List
-  );
-}
-#endif  // __ARM_NEON__
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_posix.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_posix.cc
deleted file mode 100755
index 106fda5689..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_posix.cc
+++ /dev/null
@@ -1,6443 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC x86 and x64.
-#if !defined(LIBYUV_DISABLE_X86) && (defined(__x86_64__) || defined(__i386__))
-
-#if defined(HAS_ARGBTOYROW_SSSE3) || defined(HAS_ARGBGRAYROW_SSSE3)
-
-// Constants for ARGB
-static vec8 kARGBToY = {
-  13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0
-};
-
-// JPeg full range.
-static vec8 kARGBToYJ = {
-  15, 75, 38, 0, 15, 75, 38, 0, 15, 75, 38, 0, 15, 75, 38, 0
-};
-#endif  // defined(HAS_ARGBTOYROW_SSSE3) || defined(HAS_ARGBGRAYROW_SSSE3)
-
-#if defined(HAS_ARGBTOYROW_SSSE3) || defined(HAS_I422TOARGBROW_SSSE3)
-
-static vec8 kARGBToU = {
-  112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0
-};
-
-static vec8 kARGBToUJ = {
-  127, -84, -43, 0, 127, -84, -43, 0, 127, -84, -43, 0, 127, -84, -43, 0
-};
-
-static vec8 kARGBToV = {
-  -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0,
-};
-
-static vec8 kARGBToVJ = {
-  -20, -107, 127, 0, -20, -107, 127, 0, -20, -107, 127, 0, -20, -107, 127, 0
-};
-
-// Constants for BGRA
-static vec8 kBGRAToY = {
-  0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13
-};
-
-static vec8 kBGRAToU = {
-  0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112
-};
-
-static vec8 kBGRAToV = {
-  0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18
-};
-
-// Constants for ABGR
-static vec8 kABGRToY = {
-  33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0
-};
-
-static vec8 kABGRToU = {
-  -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0
-};
-
-static vec8 kABGRToV = {
-  112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0
-};
-
-// Constants for RGBA.
-static vec8 kRGBAToY = {
-  0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33
-};
-
-static vec8 kRGBAToU = {
-  0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38
-};
-
-static vec8 kRGBAToV = {
-  0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112
-};
-
-static uvec8 kAddY16 = {
-  16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u
-};
-
-static vec16 kAddYJ64 = {
-  64, 64, 64, 64, 64, 64, 64, 64
-};
-
-static uvec8 kAddUV128 = {
-  128u, 128u, 128u, 128u, 128u, 128u, 128u, 128u,
-  128u, 128u, 128u, 128u, 128u, 128u, 128u, 128u
-};
-
-static uvec16 kAddUVJ128 = {
-  0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u
-};
-#endif  // defined(HAS_ARGBTOYROW_SSSE3) || defined(HAS_I422TOARGBROW_SSSE3)
-
-#ifdef HAS_RGB24TOARGBROW_SSSE3
-
-// Shuffle table for converting RGB24 to ARGB.
-static uvec8 kShuffleMaskRGB24ToARGB = {
-  0u, 1u, 2u, 12u, 3u, 4u, 5u, 13u, 6u, 7u, 8u, 14u, 9u, 10u, 11u, 15u
-};
-
-// Shuffle table for converting RAW to ARGB.
-static uvec8 kShuffleMaskRAWToARGB = {
-  2u, 1u, 0u, 12u, 5u, 4u, 3u, 13u, 8u, 7u, 6u, 14u, 11u, 10u, 9u, 15u
-};
-
-// Shuffle table for converting ARGB to RGB24.
-static uvec8 kShuffleMaskARGBToRGB24 = {
-  0u, 1u, 2u, 4u, 5u, 6u, 8u, 9u, 10u, 12u, 13u, 14u, 128u, 128u, 128u, 128u
-};
-
-// Shuffle table for converting ARGB to RAW.
-static uvec8 kShuffleMaskARGBToRAW = {
-  2u, 1u, 0u, 6u, 5u, 4u, 10u, 9u, 8u, 14u, 13u, 12u, 128u, 128u, 128u, 128u
-};
-
-// Shuffle table for converting ARGBToRGB24 for I422ToRGB24.  First 8 + next 4
-static uvec8 kShuffleMaskARGBToRGB24_0 = {
-  0u, 1u, 2u, 4u, 5u, 6u, 8u, 9u, 128u, 128u, 128u, 128u, 10u, 12u, 13u, 14u
-};
-
-// Shuffle table for converting ARGB to RAW.
-static uvec8 kShuffleMaskARGBToRAW_0 = {
-  2u, 1u, 0u, 6u, 5u, 4u, 10u, 9u, 128u, 128u, 128u, 128u, 8u, 14u, 13u, 12u
-};
-#endif  // HAS_RGB24TOARGBROW_SSSE3
-
-#if defined(TESTING) && defined(__x86_64__)
-void TestRow_SSE2(const uint8* src_y, uint8* dst_argb, int pix) {
-  asm volatile (
-    ".p2align  5                               \n"
-    "mov       %%eax,%%eax                     \n"
-    "mov       %%ebx,%%ebx                     \n"
-    "mov       %%ecx,%%ecx                     \n"
-    "mov       %%edx,%%edx                     \n"
-    "mov       %%esi,%%esi                     \n"
-    "mov       %%edi,%%edi                     \n"
-    "mov       %%ebp,%%ebp                     \n"
-    "mov       %%esp,%%esp                     \n"
-    ".p2align  5                               \n"
-    "mov       %%r8d,%%r8d                     \n"
-    "mov       %%r9d,%%r9d                     \n"
-    "mov       %%r10d,%%r10d                   \n"
-    "mov       %%r11d,%%r11d                   \n"
-    "mov       %%r12d,%%r12d                   \n"
-    "mov       %%r13d,%%r13d                   \n"
-    "mov       %%r14d,%%r14d                   \n"
-    "mov       %%r15d,%%r15d                   \n"
-    ".p2align  5                               \n"
-    "lea       (%%rax),%%eax                   \n"
-    "lea       (%%rbx),%%ebx                   \n"
-    "lea       (%%rcx),%%ecx                   \n"
-    "lea       (%%rdx),%%edx                   \n"
-    "lea       (%%rsi),%%esi                   \n"
-    "lea       (%%rdi),%%edi                   \n"
-    "lea       (%%rbp),%%ebp                   \n"
-    "lea       (%%rsp),%%esp                   \n"
-    ".p2align  5                               \n"
-    "lea       (%%r8),%%r8d                    \n"
-    "lea       (%%r9),%%r9d                    \n"
-    "lea       (%%r10),%%r10d                  \n"
-    "lea       (%%r11),%%r11d                  \n"
-    "lea       (%%r12),%%r12d                  \n"
-    "lea       (%%r13),%%r13d                  \n"
-    "lea       (%%r14),%%r14d                  \n"
-    "lea       (%%r15),%%r15d                  \n"
-
-    ".p2align  5                               \n"
-    "lea       0x10(%%rax),%%eax               \n"
-    "lea       0x10(%%rbx),%%ebx               \n"
-    "lea       0x10(%%rcx),%%ecx               \n"
-    "lea       0x10(%%rdx),%%edx               \n"
-    "lea       0x10(%%rsi),%%esi               \n"
-    "lea       0x10(%%rdi),%%edi               \n"
-    "lea       0x10(%%rbp),%%ebp               \n"
-    "lea       0x10(%%rsp),%%esp               \n"
-    ".p2align  5                               \n"
-    "lea       0x10(%%r8),%%r8d                \n"
-    "lea       0x10(%%r9),%%r9d                \n"
-    "lea       0x10(%%r10),%%r10d              \n"
-    "lea       0x10(%%r11),%%r11d              \n"
-    "lea       0x10(%%r12),%%r12d              \n"
-    "lea       0x10(%%r13),%%r13d              \n"
-    "lea       0x10(%%r14),%%r14d              \n"
-    "lea       0x10(%%r15),%%r15d              \n"
-
-    ".p2align  5                               \n"
-    "add       0x10,%%eax                      \n"
-    "add       0x10,%%ebx                      \n"
-    "add       0x10,%%ecx                      \n"
-    "add       0x10,%%edx                      \n"
-    "add       0x10,%%esi                      \n"
-    "add       0x10,%%edi                      \n"
-    "add       0x10,%%ebp                      \n"
-    "add       0x10,%%esp                      \n"
-    ".p2align  5                               \n"
-    "add       0x10,%%r8d                      \n"
-    "add       0x10,%%r9d                      \n"
-    "add       0x10,%%r10d                     \n"
-    "add       0x10,%%r11d                     \n"
-    "add       0x10,%%r12d                     \n"
-    "add       0x10,%%r13d                     \n"
-    "add       0x10,%%r14d                     \n"
-    "add       0x10,%%r15d                     \n"
-
-    ".p2align  2                               \n"
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_y),     // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // TESTING
-
-#ifdef HAS_I400TOARGBROW_SSE2
-void I400ToARGBRow_SSE2(const uint8* src_y, uint8* dst_argb, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pslld     $0x18,%%xmm5                    \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm0,%%xmm0                   \n"
-    "punpckhwd %%xmm1,%%xmm1                   \n"
-    "por       %%xmm5,%%xmm0                   \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_y),     // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void I400ToARGBRow_Unaligned_SSE2(const uint8* src_y, uint8* dst_argb,
-                                  int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pslld     $0x18,%%xmm5                    \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm0,%%xmm0                   \n"
-    "punpckhwd %%xmm1,%%xmm1                   \n"
-    "por       %%xmm5,%%xmm0                   \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_y),     // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_I400TOARGBROW_SSE2
-
-#ifdef HAS_RGB24TOARGBROW_SSSE3
-void RGB24ToARGBRow_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"  // generate mask 0xff000000
-    "pslld     $0x18,%%xmm5                    \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x30,0) ",%0           \n"
-    "movdqa    %%xmm3,%%xmm2                   \n"
-    "palignr   $0x8,%%xmm1,%%xmm2              \n"
-    "pshufb    %%xmm4,%%xmm2                   \n"
-    "por       %%xmm5,%%xmm2                   \n"
-    "palignr   $0xc,%%xmm0,%%xmm1              \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "movdqa    %%xmm2," MEMACCESS2(0x20,1) "   \n"
-    "por       %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm4,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "palignr   $0x4,%%xmm3,%%xmm3              \n"
-    "pshufb    %%xmm4,%%xmm3                   \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "por       %%xmm5,%%xmm3                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x30,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_rgb24),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "m"(kShuffleMaskRGB24ToARGB)  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void RAWToARGBRow_SSSE3(const uint8* src_raw, uint8* dst_argb, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"  // generate mask 0xff000000
-    "pslld     $0x18,%%xmm5                    \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x30,0) ",%0           \n"
-    "movdqa    %%xmm3,%%xmm2                   \n"
-    "palignr   $0x8,%%xmm1,%%xmm2              \n"
-    "pshufb    %%xmm4,%%xmm2                   \n"
-    "por       %%xmm5,%%xmm2                   \n"
-    "palignr   $0xc,%%xmm0,%%xmm1              \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "movdqa    %%xmm2," MEMACCESS2(0x20,1) "   \n"
-    "por       %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm4,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "palignr   $0x4,%%xmm3,%%xmm3              \n"
-    "pshufb    %%xmm4,%%xmm3                   \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "por       %%xmm5,%%xmm3                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x30,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_raw),   // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "m"(kShuffleMaskRAWToARGB)  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void RGB565ToARGBRow_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "mov       $0x1080108,%%eax                \n"
-    "movd      %%eax,%%xmm5                    \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "mov       $0x20802080,%%eax               \n"
-    "movd      %%eax,%%xmm6                    \n"
-    "pshufd    $0x0,%%xmm6,%%xmm6              \n"
-    "pcmpeqb   %%xmm3,%%xmm3                   \n"
-    "psllw     $0xb,%%xmm3                     \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "psllw     $0xa,%%xmm4                     \n"
-    "psrlw     $0x5,%%xmm4                     \n"
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "psllw     $0x8,%%xmm7                     \n"
-    "sub       %0,%1                           \n"
-    "sub       %0,%1                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "pand      %%xmm3,%%xmm1                   \n"
-    "psllw     $0xb,%%xmm2                     \n"
-    "pmulhuw   %%xmm5,%%xmm1                   \n"
-    "pmulhuw   %%xmm5,%%xmm2                   \n"
-    "psllw     $0x8,%%xmm1                     \n"
-    "por       %%xmm2,%%xmm1                   \n"
-    "pand      %%xmm4,%%xmm0                   \n"
-    "pmulhuw   %%xmm6,%%xmm0                   \n"
-    "por       %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm0,%%xmm1                   \n"
-    "punpckhbw %%xmm0,%%xmm2                   \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm1,0x00,1,0,2)           //  movdqa  %%xmm1,(%1,%0,2)
-    MEMOPMEM(movdqa,xmm2,0x10,1,0,2)           //  movdqa  %%xmm2,0x10(%1,%0,2)
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc", "eax"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGB1555ToARGBRow_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "mov       $0x1080108,%%eax                \n"
-    "movd      %%eax,%%xmm5                    \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "mov       $0x42004200,%%eax               \n"
-    "movd      %%eax,%%xmm6                    \n"
-    "pshufd    $0x0,%%xmm6,%%xmm6              \n"
-    "pcmpeqb   %%xmm3,%%xmm3                   \n"
-    "psllw     $0xb,%%xmm3                     \n"
-    "movdqa    %%xmm3,%%xmm4                   \n"
-    "psrlw     $0x6,%%xmm4                     \n"
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "psllw     $0x8,%%xmm7                     \n"
-    "sub       %0,%1                           \n"
-    "sub       %0,%1                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psllw     $0x1,%%xmm1                     \n"
-    "psllw     $0xb,%%xmm2                     \n"
-    "pand      %%xmm3,%%xmm1                   \n"
-    "pmulhuw   %%xmm5,%%xmm2                   \n"
-    "pmulhuw   %%xmm5,%%xmm1                   \n"
-    "psllw     $0x8,%%xmm1                     \n"
-    "por       %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "pand      %%xmm4,%%xmm0                   \n"
-    "psraw     $0x8,%%xmm2                     \n"
-    "pmulhuw   %%xmm6,%%xmm0                   \n"
-    "pand      %%xmm7,%%xmm2                   \n"
-    "por       %%xmm2,%%xmm0                   \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm0,%%xmm1                   \n"
-    "punpckhbw %%xmm0,%%xmm2                   \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm1,0x00,1,0,2)           //  movdqa  %%xmm1,(%1,%0,2)
-    MEMOPMEM(movdqa,xmm2,0x10,1,0,2)           //  movdqa  %%xmm2,0x10(%1,%0,2)
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc", "eax"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGB4444ToARGBRow_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "mov       $0xf0f0f0f,%%eax                \n"
-    "movd      %%eax,%%xmm4                    \n"
-    "pshufd    $0x0,%%xmm4,%%xmm4              \n"
-    "movdqa    %%xmm4,%%xmm5                   \n"
-    "pslld     $0x4,%%xmm5                     \n"
-    "sub       %0,%1                           \n"
-    "sub       %0,%1                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "pand      %%xmm4,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "psllw     $0x4,%%xmm1                     \n"
-    "psrlw     $0x4,%%xmm3                     \n"
-    "por       %%xmm1,%%xmm0                   \n"
-    "por       %%xmm3,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm2,%%xmm0                   \n"
-    "punpckhbw %%xmm2,%%xmm1                   \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,2)           //  movdqa  %%xmm0,(%1,%0,2)
-    MEMOPMEM(movdqa,xmm1,0x10,1,0,2)           //  movdqa  %%xmm1,0x10(%1,%0,2)
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc", "eax"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ARGBToRGB24Row_SSSE3(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "movdqa    %3,%%xmm6                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "pshufb    %%xmm6,%%xmm0                   \n"
-    "pshufb    %%xmm6,%%xmm1                   \n"
-    "pshufb    %%xmm6,%%xmm2                   \n"
-    "pshufb    %%xmm6,%%xmm3                   \n"
-    "movdqa    %%xmm1,%%xmm4                   \n"
-    "psrldq    $0x4,%%xmm1                     \n"
-    "pslldq    $0xc,%%xmm4                     \n"
-    "movdqa    %%xmm2,%%xmm5                   \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pslldq    $0x8,%%xmm5                     \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "psrldq    $0x8,%%xmm2                     \n"
-    "pslldq    $0x4,%%xmm3                     \n"
-    "por       %%xmm3,%%xmm2                   \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "movdqu    %%xmm2," MEMACCESS2(0x20,1) "   \n"
-    "lea       " MEMLEA(0x30,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  : "m"(kShuffleMaskARGBToRGB24)  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-void ARGBToRAWRow_SSSE3(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "movdqa    %3,%%xmm6                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "pshufb    %%xmm6,%%xmm0                   \n"
-    "pshufb    %%xmm6,%%xmm1                   \n"
-    "pshufb    %%xmm6,%%xmm2                   \n"
-    "pshufb    %%xmm6,%%xmm3                   \n"
-    "movdqa    %%xmm1,%%xmm4                   \n"
-    "psrldq    $0x4,%%xmm1                     \n"
-    "pslldq    $0xc,%%xmm4                     \n"
-    "movdqa    %%xmm2,%%xmm5                   \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pslldq    $0x8,%%xmm5                     \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "psrldq    $0x8,%%xmm2                     \n"
-    "pslldq    $0x4,%%xmm3                     \n"
-    "por       %%xmm3,%%xmm2                   \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "movdqu    %%xmm2," MEMACCESS2(0x20,1) "   \n"
-    "lea       " MEMLEA(0x30,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  : "m"(kShuffleMaskARGBToRAW)  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-void ARGBToRGB565Row_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm3,%%xmm3                   \n"
-    "psrld     $0x1b,%%xmm3                    \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "psrld     $0x1a,%%xmm4                    \n"
-    "pslld     $0x5,%%xmm4                     \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pslld     $0xb,%%xmm5                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "pslld     $0x8,%%xmm0                     \n"
-    "psrld     $0x3,%%xmm1                     \n"
-    "psrld     $0x5,%%xmm2                     \n"
-    "psrad     $0x10,%%xmm0                    \n"
-    "pand      %%xmm3,%%xmm1                   \n"
-    "pand      %%xmm4,%%xmm2                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "por       %%xmm2,%%xmm1                   \n"
-    "por       %%xmm1,%%xmm0                   \n"
-    "packssdw  %%xmm0,%%xmm0                   \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x4,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ARGBToARGB1555Row_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "psrld     $0x1b,%%xmm4                    \n"
-    "movdqa    %%xmm4,%%xmm5                   \n"
-    "pslld     $0x5,%%xmm5                     \n"
-    "movdqa    %%xmm4,%%xmm6                   \n"
-    "pslld     $0xa,%%xmm6                     \n"
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "pslld     $0xf,%%xmm7                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm3                   \n"
-    "psrad     $0x10,%%xmm0                    \n"
-    "psrld     $0x3,%%xmm1                     \n"
-    "psrld     $0x6,%%xmm2                     \n"
-    "psrld     $0x9,%%xmm3                     \n"
-    "pand      %%xmm7,%%xmm0                   \n"
-    "pand      %%xmm4,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "pand      %%xmm6,%%xmm3                   \n"
-    "por       %%xmm1,%%xmm0                   \n"
-    "por       %%xmm3,%%xmm2                   \n"
-    "por       %%xmm2,%%xmm0                   \n"
-    "packssdw  %%xmm0,%%xmm0                   \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMACCESS2(0x8,1) ",%1        \n"
-    "sub       $0x4,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGBToARGB4444Row_SSE2(const uint8* src, uint8* dst, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "psllw     $0xc,%%xmm4                     \n"
-    "movdqa    %%xmm4,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm3                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm3,%%xmm0                   \n"
-    "pand      %%xmm4,%%xmm1                   \n"
-    "psrlq     $0x4,%%xmm0                     \n"
-    "psrlq     $0x8,%%xmm1                     \n"
-    "por       %%xmm1,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x4,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(pix)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4"
-#endif
-  );
-}
-#endif  // HAS_RGB24TOARGBROW_SSSE3
-
-#ifdef HAS_ARGBTOYROW_SSSE3
-void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kARGBToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ARGBToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kARGBToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBTOYROW_SSSE3
-
-#ifdef HAS_ARGBTOYJROW_SSSE3
-void ARGBToYJRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %3,%%xmm4                       \n"
-    "movdqa    %4,%%xmm5                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "paddw     %%xmm5,%%xmm0                   \n"
-    "paddw     %%xmm5,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kARGBToYJ),  // %3
-    "m"(kAddYJ64)    // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ARGBToYJRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %3,%%xmm4                       \n"
-    "movdqa    %4,%%xmm5                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "paddw     %%xmm5,%%xmm0                   \n"
-    "paddw     %%xmm5,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kARGBToYJ),  // %3
-    "m"(kAddYJ64)    // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBTOYJROW_SSSE3
-
-#ifdef HAS_ARGBTOUVROW_SSSE3
-// TODO(fbarchard): pass xmm constants to single block of assembly.
-// fpic on GCC 4.2 for OSX runs out of GPR registers. "m" effectively takes
-// 3 registers - ebx, ebp and eax. "m" can be passed with 3 normal registers,
-// or 4 if stack frame is disabled. Doing 2 assembly blocks is a work around
-// and considered unsafe.
-void ARGBToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),  // %0
-    "m"(kARGBToV),  // %1
-    "m"(kAddUV128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(pavgb,0x00,0,4,1,xmm0)            //  pavgb   (%0,%4,1),%%xmm0
-    MEMOPREG(pavgb,0x10,0,4,1,xmm1)            //  pavgb   0x10(%0,%4,1),%%xmm1
-    MEMOPREG(pavgb,0x20,0,4,1,xmm2)            //  pavgb   0x20(%0,%4,1),%%xmm2
-    MEMOPREG(pavgb,0x30,0,4,1,xmm6)            //  pavgb   0x30(%0,%4,1),%%xmm6
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps    %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_argb)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-// TODO(fbarchard): Share code with ARGBToUVRow_SSSE3.
-void ARGBToUVJRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                        uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToUJ),  // %0
-    "m"(kARGBToVJ),  // %1
-    "m"(kAddUVJ128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(pavgb,0x00,0,4,1,xmm0)            //  pavgb   (%0,%4,1),%%xmm0
-    MEMOPREG(pavgb,0x10,0,4,1,xmm1)            //  pavgb   0x10(%0,%4,1),%%xmm1
-    MEMOPREG(pavgb,0x20,0,4,1,xmm2)            //  pavgb   0x20(%0,%4,1),%%xmm2
-    MEMOPREG(pavgb,0x30,0,4,1,xmm6)            //  pavgb   0x30(%0,%4,1),%%xmm6
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "paddw     %%xmm5,%%xmm0                   \n"
-    "paddw     %%xmm5,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_argb)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGBToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),         // %0
-    "m"(kARGBToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm7)           //  movdqu  (%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    MEMOPREG(movdqu,0x10,0,4,1,xmm7)           //  movdqu  0x10(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqu,0x20,0,4,1,xmm7)           //  movdqu  0x20(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    MEMOPREG(movdqu,0x30,0,4,1,xmm7)           //  movdqu  0x30(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_argb)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGBToUVJRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                  uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToUJ),         // %0
-    "m"(kARGBToVJ),         // %1
-    "m"(kAddUVJ128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm7)           //  movdqu  (%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    MEMOPREG(movdqu,0x10,0,4,1,xmm7)           //  movdqu  0x10(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqu,0x20,0,4,1,xmm7)           //  movdqu  0x20(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    MEMOPREG(movdqu,0x30,0,4,1,xmm7)           //  movdqu  0x30(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "paddw     %%xmm5,%%xmm0                   \n"
-    "paddw     %%xmm5,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_argb))
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGBToUV444Row_SSSE3(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                          int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),  // %0
-    "m"(kARGBToV),  // %1
-    "m"(kAddUV128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm6                   \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm2                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm2                     \n"
-    "packsswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "pmaddubsw %%xmm3,%%xmm0                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm2                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm2                     \n"
-    "packsswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,2,1)           //  movdqa  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),        // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6"
-#endif
-  );
-}
-
-void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_u,
-                                    uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),  // %0
-    "m"(kARGBToV),  // %1
-    "m"(kAddUV128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm6                   \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm2                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm2                     \n"
-    "packsswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "pmaddubsw %%xmm3,%%xmm0                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm2                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm2                     \n"
-    "packsswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,2,1)           //  movdqu  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),        // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6"
-#endif
-  );
-}
-
-void ARGBToUV422Row_SSSE3(const uint8* src_argb0,
-                          uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),  // %0
-    "m"(kARGBToV),  // %1
-    "m"(kAddUV128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ARGBToUV422Row_Unaligned_SSSE3(const uint8* src_argb0,
-                                    uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kARGBToU),  // %0
-    "m"(kARGBToV),  // %1
-    "m"(kAddUV128)  // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void BGRAToYRow_SSSE3(const uint8* src_bgra, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_bgra),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kBGRAToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void BGRAToYRow_Unaligned_SSSE3(const uint8* src_bgra, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_bgra),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kBGRAToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void BGRAToUVRow_SSSE3(const uint8* src_bgra0, int src_stride_bgra,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kBGRAToU),         // %0
-    "m"(kBGRAToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(pavgb,0x00,0,4,1,xmm0)            //  pavgb   (%0,%4,1),%%xmm0
-    MEMOPREG(pavgb,0x10,0,4,1,xmm1)            //  pavgb   0x10(%0,%4,1),%%xmm1
-    MEMOPREG(pavgb,0x20,0,4,1,xmm2)            //  pavgb   0x20(%0,%4,1),%%xmm2
-    MEMOPREG(pavgb,0x30,0,4,1,xmm6)            //  pavgb   0x30(%0,%4,1),%%xmm6
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_bgra0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_bgra)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void BGRAToUVRow_Unaligned_SSSE3(const uint8* src_bgra0, int src_stride_bgra,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kBGRAToU),         // %0
-    "m"(kBGRAToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm7)           //  movdqu  (%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    MEMOPREG(movdqu,0x10,0,4,1,xmm7)           //  movdqu  0x10(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqu,0x20,0,4,1,xmm7)           //  movdqu  0x20(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    MEMOPREG(movdqu,0x30,0,4,1,xmm7)           //  movdqu  0x30(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_bgra0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_bgra)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ABGRToYRow_SSSE3(const uint8* src_abgr, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_abgr),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kABGRToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ABGRToYRow_Unaligned_SSSE3(const uint8* src_abgr, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_abgr),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kABGRToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void RGBAToYRow_SSSE3(const uint8* src_rgba, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_rgba),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kRGBAToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void RGBAToYRow_Unaligned_SSSE3(const uint8* src_rgba, uint8* dst_y, int pix) {
-  asm volatile (
-    "movdqa    %4,%%xmm5                       \n"
-    "movdqa    %3,%%xmm4                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm4,%%xmm3                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "phaddw    %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm2                     \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_rgba),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  : "m"(kRGBAToY),   // %3
-    "m"(kAddY16)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ABGRToUVRow_SSSE3(const uint8* src_abgr0, int src_stride_abgr,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kABGRToU),         // %0
-    "m"(kABGRToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(pavgb,0x00,0,4,1,xmm0)            //  pavgb   (%0,%4,1),%%xmm0
-    MEMOPREG(pavgb,0x10,0,4,1,xmm1)            //  pavgb   0x10(%0,%4,1),%%xmm1
-    MEMOPREG(pavgb,0x20,0,4,1,xmm2)            //  pavgb   0x20(%0,%4,1),%%xmm2
-    MEMOPREG(pavgb,0x30,0,4,1,xmm6)            //  pavgb   0x30(%0,%4,1),%%xmm6
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_abgr0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_abgr)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ABGRToUVRow_Unaligned_SSSE3(const uint8* src_abgr0, int src_stride_abgr,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kABGRToU),         // %0
-    "m"(kABGRToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm7)           //  movdqu  (%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    MEMOPREG(movdqu,0x10,0,4,1,xmm7)           //  movdqu  0x10(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqu,0x20,0,4,1,xmm7)           //  movdqu  0x20(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    MEMOPREG(movdqu,0x30,0,4,1,xmm7)           //  movdqu  0x30(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_abgr0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_abgr)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void RGBAToUVRow_SSSE3(const uint8* src_rgba0, int src_stride_rgba,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kRGBAToU),         // %0
-    "m"(kRGBAToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(pavgb,0x00,0,4,1,xmm0)            //  pavgb   (%0,%4,1),%%xmm0
-    MEMOPREG(pavgb,0x10,0,4,1,xmm1)            //  pavgb   0x10(%0,%4,1),%%xmm1
-    MEMOPREG(pavgb,0x20,0,4,1,xmm2)            //  pavgb   0x20(%0,%4,1),%%xmm2
-    MEMOPREG(pavgb,0x30,0,4,1,xmm6)            //  pavgb   0x30(%0,%4,1),%%xmm6
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_rgba0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_rgba))
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void RGBAToUVRow_Unaligned_SSSE3(const uint8* src_rgba0, int src_stride_rgba,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kRGBAToU),         // %0
-    "m"(kRGBAToV),         // %1
-    "m"(kAddUV128)         // %2
-  );
-  asm volatile (
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqu    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqu    " MEMACCESS2(0x30,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm7)           //  movdqu  (%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    MEMOPREG(movdqu,0x10,0,4,1,xmm7)           //  movdqu  0x10(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqu,0x20,0,4,1,xmm7)           //  movdqu  0x20(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    MEMOPREG(movdqu,0x30,0,4,1,xmm7)           //  movdqu  0x30(%0,%4,1),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm7                   \n"
-    "shufps    $0x88,%%xmm6,%%xmm2             \n"
-    "shufps    $0xdd,%%xmm6,%%xmm7             \n"
-    "pavgb     %%xmm7,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm2                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "phaddw    %%xmm2,%%xmm0                   \n"
-    "phaddw    %%xmm6,%%xmm1                   \n"
-    "psraw     $0x8,%%xmm0                     \n"
-    "psraw     $0x8,%%xmm1                     \n"
-    "packsswb  %%xmm1,%%xmm0                   \n"
-    "paddb     %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movlps    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhps,xmm0,0x00,1,2,1)           //  movhps  %%xmm0,(%1,%2,1)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_rgba0),       // %0
-    "+r"(dst_u),           // %1
-    "+r"(dst_v),           // %2
-    "+rm"(width)           // %3
-  : "r"((intptr_t)(src_stride_rgba)) // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBTOUVROW_SSSE3
-
-#ifdef HAS_I422TOARGBROW_SSSE3
-#define UB 127 /* min(63,(int8)(2.018 * 64)) */
-#define UG -25 /* (int8)(-0.391 * 64 - 0.5) */
-#define UR 0
-
-#define VB 0
-#define VG -52 /* (int8)(-0.813 * 64 - 0.5) */
-#define VR 102 /* (int8)(1.596 * 64 + 0.5) */
-
-// Bias
-#define BB UB * 128 + VB * 128
-#define BG UG * 128 + VG * 128
-#define BR UR * 128 + VR * 128
-
-#define YG 74 /* (int8)(1.164 * 64 + 0.5) */
-
-struct {
-  vec8 kUVToB;  // 0
-  vec8 kUVToG;  // 16
-  vec8 kUVToR;  // 32
-  vec16 kUVBiasB;  // 48
-  vec16 kUVBiasG;  // 64
-  vec16 kUVBiasR;  // 80
-  vec16 kYSub16;  // 96
-  vec16 kYToRgb;  // 112
-  vec8 kVUToB;  // 128
-  vec8 kVUToG;  // 144
-  vec8 kVUToR;  // 160
-} static SIMD_ALIGNED(kYuvConstants) = {
-  { UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB },
-  { UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG },
-  { UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR },
-  { BB, BB, BB, BB, BB, BB, BB, BB },
-  { BG, BG, BG, BG, BG, BG, BG, BG },
-  { BR, BR, BR, BR, BR, BR, BR, BR },
-  { 16, 16, 16, 16, 16, 16, 16, 16 },
-  { YG, YG, YG, YG, YG, YG, YG, YG },
-  { VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB },
-  { VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG },
-  { VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR }
-};
-
-
-// Read 8 UV from 411
-#define READYUV444                                                             \
-    "movq       " MEMACCESS([u_buf]) ",%%xmm0                   \n"            \
-    BUNDLEALIGN                                                                \
-    MEMOPREG(movq, 0x00, [u_buf], [v_buf], 1, xmm1)                            \
-    "lea        " MEMLEA(0x8, [u_buf]) ",%[u_buf]               \n"            \
-    "punpcklbw  %%xmm1,%%xmm0                                   \n"
-
-// Read 4 UV from 422, upsample to 8 UV
-#define READYUV422                                                             \
-    "movd       " MEMACCESS([u_buf]) ",%%xmm0                   \n"            \
-    BUNDLEALIGN                                                                \
-    MEMOPREG(movd, 0x00, [u_buf], [v_buf], 1, xmm1)                            \
-    "lea        " MEMLEA(0x4, [u_buf]) ",%[u_buf]               \n"            \
-    "punpcklbw  %%xmm1,%%xmm0                                   \n"            \
-    "punpcklwd  %%xmm0,%%xmm0                                   \n"
-
-// Read 2 UV from 411, upsample to 8 UV
-#define READYUV411                                                             \
-    "movd       " MEMACCESS([u_buf]) ",%%xmm0                   \n"            \
-    BUNDLEALIGN                                                                \
-    MEMOPREG(movd, 0x00, [u_buf], [v_buf], 1, xmm1)                            \
-    "lea        " MEMLEA(0x2, [u_buf]) ",%[u_buf]               \n"            \
-    "punpcklbw  %%xmm1,%%xmm0                                   \n"            \
-    "punpcklwd  %%xmm0,%%xmm0                                   \n"            \
-    "punpckldq  %%xmm0,%%xmm0                                   \n"
-
-// Read 4 UV from NV12, upsample to 8 UV
-#define READNV12                                                               \
-    "movq       " MEMACCESS([uv_buf]) ",%%xmm0                  \n"            \
-    "lea        " MEMLEA(0x8, [uv_buf]) ",%[uv_buf]             \n"            \
-    "punpcklwd  %%xmm0,%%xmm0                                   \n"
-
-// Convert 8 pixels: 8 UV and 8 Y
-#define YUVTORGB                                                               \
-    "movdqa     %%xmm0,%%xmm1                                   \n"            \
-    "movdqa     %%xmm0,%%xmm2                                   \n"            \
-    "pmaddubsw  " MEMACCESS([kYuvConstants]) ",%%xmm0           \n"            \
-    "pmaddubsw  " MEMACCESS2(16, [kYuvConstants]) ",%%xmm1      \n"            \
-    "pmaddubsw  " MEMACCESS2(32, [kYuvConstants]) ",%%xmm2      \n"            \
-    "psubw      " MEMACCESS2(48, [kYuvConstants]) ",%%xmm0      \n"            \
-    "psubw      " MEMACCESS2(64, [kYuvConstants]) ",%%xmm1      \n"            \
-    "psubw      " MEMACCESS2(80, [kYuvConstants]) ",%%xmm2      \n"            \
-    "movq       " MEMACCESS([y_buf]) ",%%xmm3                   \n"            \
-    "lea        " MEMLEA(0x8, [y_buf]) ",%[y_buf]               \n"            \
-    "punpcklbw  %%xmm4,%%xmm3                                   \n"            \
-    "psubsw     " MEMACCESS2(96, [kYuvConstants]) ",%%xmm3      \n"            \
-    "pmullw     " MEMACCESS2(112, [kYuvConstants]) ",%%xmm3     \n"            \
-    "paddsw     %%xmm3,%%xmm0                                   \n"            \
-    "paddsw     %%xmm3,%%xmm1                                   \n"            \
-    "paddsw     %%xmm3,%%xmm2                                   \n"            \
-    "psraw      $0x6,%%xmm0                                     \n"            \
-    "psraw      $0x6,%%xmm1                                     \n"            \
-    "psraw      $0x6,%%xmm2                                     \n"            \
-    "packuswb   %%xmm0,%%xmm0                                   \n"            \
-    "packuswb   %%xmm1,%%xmm1                                   \n"            \
-    "packuswb   %%xmm2,%%xmm2                                   \n"
-
-// Convert 8 pixels: 8 VU and 8 Y
-#define YVUTORGB                                                               \
-    "movdqa     %%xmm0,%%xmm1                                   \n"            \
-    "movdqa     %%xmm0,%%xmm2                                   \n"            \
-    "pmaddubsw  " MEMACCESS2(128, [kYuvConstants]) ",%%xmm0     \n"            \
-    "pmaddubsw  " MEMACCESS2(144, [kYuvConstants]) ",%%xmm1     \n"            \
-    "pmaddubsw  " MEMACCESS2(160, [kYuvConstants]) ",%%xmm2     \n"            \
-    "psubw      " MEMACCESS2(48, [kYuvConstants]) ",%%xmm0      \n"            \
-    "psubw      " MEMACCESS2(64, [kYuvConstants]) ",%%xmm1      \n"            \
-    "psubw      " MEMACCESS2(80, [kYuvConstants]) ",%%xmm2      \n"            \
-    "movq       " MEMACCESS([y_buf]) ",%%xmm3                   \n"            \
-    "lea        " MEMLEA(0x8, [y_buf]) ",%[y_buf]               \n"            \
-    "punpcklbw  %%xmm4,%%xmm3                                   \n"            \
-    "psubsw     " MEMACCESS2(96, [kYuvConstants]) ",%%xmm3      \n"            \
-    "pmullw     " MEMACCESS2(112, [kYuvConstants]) ",%%xmm3     \n"            \
-    "paddsw     %%xmm3,%%xmm0                                   \n"            \
-    "paddsw     %%xmm3,%%xmm1                                   \n"            \
-    "paddsw     %%xmm3,%%xmm2                                   \n"            \
-    "psraw      $0x6,%%xmm0                                     \n"            \
-    "psraw      $0x6,%%xmm1                                     \n"            \
-    "psraw      $0x6,%%xmm2                                     \n"            \
-    "packuswb   %%xmm0,%%xmm0                                   \n"            \
-    "packuswb   %%xmm1,%%xmm1                                   \n"            \
-    "packuswb   %%xmm2,%%xmm2                                   \n"
-
-void OMITFP I444ToARGBRow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_argb,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV444
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS([dst_argb]) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "   \n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb]  \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToRGB24Row_SSSE3(const uint8* y_buf,
-                                 const uint8* u_buf,
-                                 const uint8* v_buf,
-                                 uint8* dst_rgb24,
-                                 int width) {
-// fpic 32 bit gcc 4.2 on OSX runs out of GPR regs.
-#if defined(__i386__)
-  asm volatile (
-    "movdqa    %[kShuffleMaskARGBToRGB24_0],%%xmm5 \n"
-    "movdqa    %[kShuffleMaskARGBToRGB24],%%xmm6   \n"
-  :: [kShuffleMaskARGBToRGB24_0]"m"(kShuffleMaskARGBToRGB24_0),
-    [kShuffleMaskARGBToRGB24]"m"(kShuffleMaskARGBToRGB24));
-#endif
-
-  asm volatile (
-#if !defined(__i386__)
-    "movdqa    %[kShuffleMaskARGBToRGB24_0],%%xmm5 \n"
-    "movdqa    %[kShuffleMaskARGBToRGB24],%%xmm6   \n"
-#endif
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm2,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm6,%%xmm1                   \n"
-    "palignr   $0xc,%%xmm0,%%xmm1              \n"
-    "movq      %%xmm0," MEMACCESS([dst_rgb24]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x8,[dst_rgb24]) "\n"
-    "lea       " MEMLEA(0x18,[dst_rgb24]) ",%[dst_rgb24] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_rgb24]"+r"(dst_rgb24),  // %[dst_rgb24]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB)
-#if !defined(__i386__)
-    , [kShuffleMaskARGBToRGB24_0]"m"(kShuffleMaskARGBToRGB24_0),
-    [kShuffleMaskARGBToRGB24]"m"(kShuffleMaskARGBToRGB24)
-#endif
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-void OMITFP I422ToRAWRow_SSSE3(const uint8* y_buf,
-                               const uint8* u_buf,
-                               const uint8* v_buf,
-                               uint8* dst_raw,
-                               int width) {
-// fpic 32 bit gcc 4.2 on OSX runs out of GPR regs.
-#if defined(__i386__)
-  asm volatile (
-    "movdqa    %[kShuffleMaskARGBToRAW_0],%%xmm5 \n"
-    "movdqa    %[kShuffleMaskARGBToRAW],%%xmm6   \n"
-  :: [kShuffleMaskARGBToRAW_0]"m"(kShuffleMaskARGBToRAW_0),
-    [kShuffleMaskARGBToRAW]"m"(kShuffleMaskARGBToRAW));
-#endif
-
-  asm volatile (
-#if !defined(__i386__)
-    "movdqa    %[kShuffleMaskARGBToRAW_0],%%xmm5 \n"
-    "movdqa    %[kShuffleMaskARGBToRAW],%%xmm6   \n"
-#endif
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm2,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm6,%%xmm1                   \n"
-    "palignr   $0xc,%%xmm0,%%xmm1              \n"
-    "movq      %%xmm0," MEMACCESS([dst_raw]) " \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x8,[dst_raw]) "\n"
-    "lea       " MEMLEA(0x18,[dst_raw]) ",%[dst_raw] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_raw]"+r"(dst_raw),  // %[dst_raw]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB)
-#if !defined(__i386__)
-    , [kShuffleMaskARGBToRAW_0]"m"(kShuffleMaskARGBToRAW_0),
-    [kShuffleMaskARGBToRAW]"m"(kShuffleMaskARGBToRAW)
-#endif
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-void OMITFP I422ToARGBRow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_argb,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I411ToARGBRow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_argb,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV411
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP NV12ToARGBRow_SSSE3(const uint8* y_buf,
-                                const uint8* uv_buf,
-                                uint8* dst_argb,
-                                int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READNV12
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [uv_buf]"+r"(uv_buf),    // %[uv_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-  // Does not use r14.
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP NV21ToARGBRow_SSSE3(const uint8* y_buf,
-                                const uint8* uv_buf,
-                                uint8* dst_argb,
-                                int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READNV12
-    YVUTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [uv_buf]"+r"(uv_buf),    // %[uv_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-  // Does not use r14.
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I444ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_argb,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV444
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_argb,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I411ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_argb,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV411
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP NV12ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* uv_buf,
-                                          uint8* dst_argb,
-                                          int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READNV12
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [uv_buf]"+r"(uv_buf),    // %[uv_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-  // Does not use r14.
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP NV21ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* uv_buf,
-                                          uint8* dst_argb,
-                                          int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READNV12
-    YVUTORGB
-    "punpcklbw %%xmm1,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm0                   \n"
-    "punpckhwd %%xmm2,%%xmm1                   \n"
-    "movdqu    %%xmm0," MEMACCESS([dst_argb]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_argb]) "\n"
-    "lea       " MEMLEA(0x20,[dst_argb]) ",%[dst_argb] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [uv_buf]"+r"(uv_buf),    // %[uv_buf]
-    [dst_argb]"+r"(dst_argb),  // %[dst_argb]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-  // Does not use r14.
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToBGRARow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_bgra,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "punpcklbw %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm2,%%xmm5                   \n"
-    "movdqa    %%xmm5,%%xmm0                   \n"
-    "punpcklwd %%xmm1,%%xmm5                   \n"
-    "punpckhwd %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm5," MEMACCESS([dst_bgra]) "\n"
-    "movdqa    %%xmm0," MEMACCESS2(0x10,[dst_bgra]) "\n"
-    "lea       " MEMLEA(0x20,[dst_bgra]) ",%[dst_bgra] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_bgra]"+r"(dst_bgra),  // %[dst_bgra]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToABGRRow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_abgr,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "punpcklwd %%xmm0,%%xmm2                   \n"
-    "punpckhwd %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2," MEMACCESS([dst_abgr]) "\n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,[dst_abgr]) "\n"
-    "lea       " MEMLEA(0x20,[dst_abgr]) ",%[dst_abgr] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_abgr]"+r"(dst_abgr),  // %[dst_abgr]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToRGBARow_SSSE3(const uint8* y_buf,
-                                const uint8* u_buf,
-                                const uint8* v_buf,
-                                uint8* dst_rgba,
-                                int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "punpcklbw %%xmm2,%%xmm1                   \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "movdqa    %%xmm5,%%xmm0                   \n"
-    "punpcklwd %%xmm1,%%xmm5                   \n"
-    "punpckhwd %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm5," MEMACCESS([dst_rgba]) "\n"
-    "movdqa    %%xmm0," MEMACCESS2(0x10,[dst_rgba]) "\n"
-    "lea       " MEMLEA(0x20,[dst_rgba]) ",%[dst_rgba] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_rgba]"+r"(dst_rgba),  // %[dst_rgba]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToBGRARow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_bgra,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "punpcklbw %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm2,%%xmm5                   \n"
-    "movdqa    %%xmm5,%%xmm0                   \n"
-    "punpcklwd %%xmm1,%%xmm5                   \n"
-    "punpckhwd %%xmm1,%%xmm0                   \n"
-    "movdqu    %%xmm5," MEMACCESS([dst_bgra]) "\n"
-    "movdqu    %%xmm0," MEMACCESS2(0x10,[dst_bgra]) "\n"
-    "lea       " MEMLEA(0x20,[dst_bgra]) ",%[dst_bgra] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_bgra]"+r"(dst_bgra),  // %[dst_bgra]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToABGRRow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_abgr,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "punpcklbw %%xmm1,%%xmm2                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "punpcklwd %%xmm0,%%xmm2                   \n"
-    "punpckhwd %%xmm0,%%xmm1                   \n"
-    "movdqu    %%xmm2," MEMACCESS([dst_abgr]) "\n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,[dst_abgr]) "\n"
-    "lea       " MEMLEA(0x20,[dst_abgr]) ",%[dst_abgr] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_abgr]"+r"(dst_abgr),  // %[dst_abgr]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void OMITFP I422ToRGBARow_Unaligned_SSSE3(const uint8* y_buf,
-                                          const uint8* u_buf,
-                                          const uint8* v_buf,
-                                          uint8* dst_rgba,
-                                          int width) {
-  asm volatile (
-    "sub       %[u_buf],%[v_buf]               \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-    LABELALIGN
-  "1:                                          \n"
-    READYUV422
-    YUVTORGB
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "punpcklbw %%xmm2,%%xmm1                   \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "movdqa    %%xmm5,%%xmm0                   \n"
-    "punpcklwd %%xmm1,%%xmm5                   \n"
-    "punpckhwd %%xmm1,%%xmm0                   \n"
-    "movdqu    %%xmm5," MEMACCESS([dst_rgba]) "\n"
-    "movdqu    %%xmm0," MEMACCESS2(0x10,[dst_rgba]) "\n"
-    "lea       " MEMLEA(0x20,[dst_rgba]) ",%[dst_rgba] \n"
-    "sub       $0x8,%[width]                   \n"
-    "jg        1b                              \n"
-  : [y_buf]"+r"(y_buf),    // %[y_buf]
-    [u_buf]"+r"(u_buf),    // %[u_buf]
-    [v_buf]"+r"(v_buf),    // %[v_buf]
-    [dst_rgba]"+r"(dst_rgba),  // %[dst_rgba]
-    [width]"+rm"(width)    // %[width]
-  : [kYuvConstants]"r"(&kYuvConstants.kUVToB) // %[kYuvConstants]
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-#endif  // HAS_I422TOARGBROW_SSSE3
-
-#ifdef HAS_YTOARGBROW_SSE2
-void YToARGBRow_SSE2(const uint8* y_buf,
-                     uint8* dst_argb,
-                     int width) {
-  asm volatile (
-    "pxor      %%xmm5,%%xmm5                   \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "pslld     $0x18,%%xmm4                    \n"
-    "mov       $0x00100010,%%eax               \n"
-    "movd      %%eax,%%xmm3                    \n"
-    "pshufd    $0x0,%%xmm3,%%xmm3              \n"
-    "mov       $0x004a004a,%%eax               \n"
-    "movd      %%eax,%%xmm2                    \n"
-    "pshufd    $0x0,%%xmm2,%%xmm2              \n"
-    LABELALIGN
-  "1:                                          \n"
-    // Step 1: Scale Y contribution to 8 G values. G = (y - 16) * 1.164
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "psubusw   %%xmm3,%%xmm0                   \n"
-    "pmullw    %%xmm2,%%xmm0                   \n"
-    "psrlw     $6, %%xmm0                      \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-
-    // Step 2: Weave into ARGB
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm0,%%xmm0                   \n"
-    "punpckhwd %%xmm1,%%xmm1                   \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "por       %%xmm4,%%xmm1                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(y_buf),     // %0
-    "+r"(dst_argb),  // %1
-    "+rm"(width)     // %2
-  :
-  : "memory", "cc", "eax"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4"
-#endif
-  );
-}
-#endif  // HAS_YTOARGBROW_SSE2
-
-#ifdef HAS_MIRRORROW_SSSE3
-// Shuffle table for reversing the bytes.
-static uvec8 kShuffleMirror = {
-  15u, 14u, 13u, 12u, 11u, 10u, 9u, 8u, 7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u
-};
-
-void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width) {
-  intptr_t temp_width = (intptr_t)(width);
-  asm volatile (
-    "movdqa    %3,%%xmm5                       \n"
-    "lea       " MEMLEA(-0x10,0) ",%0          \n"
-    LABELALIGN
-  "1:                                          \n"
-    MEMOPREG(movdqa,0x00,0,2,1,xmm0)           //  movdqa  (%0,%2),%%xmm0
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(temp_width)  // %2
-  : "m"(kShuffleMirror) // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_MIRRORROW_SSSE3
-
-#ifdef HAS_MIRRORROW_SSE2
-void MirrorRow_SSE2(const uint8* src, uint8* dst, int width) {
-  intptr_t temp_width = (intptr_t)(width);
-  asm volatile (
-    "lea       " MEMLEA(-0x10,0) ",%0          \n"
-    LABELALIGN
-  "1:                                          \n"
-    MEMOPREG(movdqu,0x00,0,2,1,xmm0)           //  movdqu  (%0,%2),%%xmm0
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "psllw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm1,%%xmm0                   \n"
-    "pshuflw   $0x1b,%%xmm0,%%xmm0             \n"
-    "pshufhw   $0x1b,%%xmm0,%%xmm0             \n"
-    "pshufd    $0x4e,%%xmm0,%%xmm0             \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1)",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(temp_width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_MIRRORROW_SSE2
-
-#ifdef HAS_MIRRORROW_UV_SSSE3
-// Shuffle table for reversing the bytes of UV channels.
-static uvec8 kShuffleMirrorUV = {
-  14u, 12u, 10u, 8u, 6u, 4u, 2u, 0u, 15u, 13u, 11u, 9u, 7u, 5u, 3u, 1u
-};
-void MirrorUVRow_SSSE3(const uint8* src, uint8* dst_u, uint8* dst_v,
-                       int width) {
-  intptr_t temp_width = (intptr_t)(width);
-  asm volatile (
-    "movdqa    %4,%%xmm1                       \n"
-    "lea       " MEMLEA4(-0x10,0,3,2) ",%0       \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(-0x10,0) ",%0            \n"
-    "pshufb    %%xmm1,%%xmm0                   \n"
-    "sub       $8,%3                           \n"
-    "movlpd    %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movhpd,xmm0,0x00,1,2,1)           //  movhpd    %%xmm0,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src),      // %0
-    "+r"(dst_u),    // %1
-    "+r"(dst_v),    // %2
-    "+r"(temp_width)  // %3
-  : "m"(kShuffleMirrorUV)  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_MIRRORROW_UV_SSSE3
-
-#ifdef HAS_ARGBMIRRORROW_SSSE3
-// Shuffle table for reversing the bytes.
-static uvec8 kARGBShuffleMirror = {
-  12u, 13u, 14u, 15u, 8u, 9u, 10u, 11u, 4u, 5u, 6u, 7u, 0u, 1u, 2u, 3u
-};
-
-void ARGBMirrorRow_SSSE3(const uint8* src, uint8* dst, int width) {
-  intptr_t temp_width = (intptr_t)(width);
-  asm volatile (
-    "lea       " MEMLEA4(-0x10,0,2,4) ",%0     \n"
-    "movdqa    %3,%%xmm5                       \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "lea       " MEMLEA(-0x10,0) ",%0          \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src),  // %0
-    "+r"(dst),  // %1
-    "+r"(temp_width)  // %2
-  : "m"(kARGBShuffleMirror)  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBMIRRORROW_SSSE3
-
-#ifdef HAS_SPLITUVROW_SSE2
-void SplitUVRow_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb    %%xmm5,%%xmm5                    \n"
-    "psrlw      $0x8,%%xmm5                      \n"
-    "sub        %1,%2                            \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movdqa     " MEMACCESS(0) ",%%xmm0          \n"
-    "movdqa     " MEMACCESS2(0x10,0) ",%%xmm1    \n"
-    "lea        " MEMLEA(0x20,0) ",%0            \n"
-    "movdqa     %%xmm0,%%xmm2                    \n"
-    "movdqa     %%xmm1,%%xmm3                    \n"
-    "pand       %%xmm5,%%xmm0                    \n"
-    "pand       %%xmm5,%%xmm1                    \n"
-    "packuswb   %%xmm1,%%xmm0                    \n"
-    "psrlw      $0x8,%%xmm2                      \n"
-    "psrlw      $0x8,%%xmm3                      \n"
-    "packuswb   %%xmm3,%%xmm2                    \n"
-    "movdqa     %%xmm0," MEMACCESS(1) "          \n"
-    MEMOPMEM(movdqa,xmm2,0x00,1,2,1)             // movdqa     %%xmm2,(%1,%2)
-    "lea        " MEMLEA(0x10,1) ",%1            \n"
-    "sub        $0x10,%3                         \n"
-    "jg         1b                               \n"
-  : "+r"(src_uv),     // %0
-    "+r"(dst_u),      // %1
-    "+r"(dst_v),      // %2
-    "+r"(pix)         // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void SplitUVRow_Unaligned_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                               int pix) {
-  asm volatile (
-    "pcmpeqb    %%xmm5,%%xmm5                    \n"
-    "psrlw      $0x8,%%xmm5                      \n"
-    "sub        %1,%2                            \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movdqu     " MEMACCESS(0) ",%%xmm0          \n"
-    "movdqu     " MEMACCESS2(0x10,0) ",%%xmm1    \n"
-    "lea        " MEMLEA(0x20,0) ",%0            \n"
-    "movdqa     %%xmm0,%%xmm2                    \n"
-    "movdqa     %%xmm1,%%xmm3                    \n"
-    "pand       %%xmm5,%%xmm0                    \n"
-    "pand       %%xmm5,%%xmm1                    \n"
-    "packuswb   %%xmm1,%%xmm0                    \n"
-    "psrlw      $0x8,%%xmm2                      \n"
-    "psrlw      $0x8,%%xmm3                      \n"
-    "packuswb   %%xmm3,%%xmm2                    \n"
-    "movdqu     %%xmm0," MEMACCESS(1) "          \n"
-    MEMOPMEM(movdqu,xmm2,0x00,1,2,1)             //  movdqu     %%xmm2,(%1,%2)
-    "lea        " MEMLEA(0x10,1) ",%1            \n"
-    "sub        $0x10,%3                         \n"
-    "jg         1b                               \n"
-  : "+r"(src_uv),     // %0
-    "+r"(dst_u),      // %1
-    "+r"(dst_v),      // %2
-    "+r"(pix)         // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_SPLITUVROW_SSE2
-
-#ifdef HAS_MERGEUVROW_SSE2
-void MergeUVRow_SSE2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width) {
-  asm volatile (
-    "sub       %0,%1                             \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0           \n"
-    MEMOPREG(movdqa,0x00,0,1,1,xmm1)             //  movdqa    (%0,%1,1),%%xmm1
-    "lea       " MEMLEA(0x10,0) ",%0             \n"
-    "movdqa    %%xmm0,%%xmm2                     \n"
-    "punpcklbw %%xmm1,%%xmm0                     \n"
-    "punpckhbw %%xmm1,%%xmm2                     \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "           \n"
-    "movdqa    %%xmm2," MEMACCESS2(0x10,2) "     \n"
-    "lea       " MEMLEA(0x20,2) ",%2             \n"
-    "sub       $0x10,%3                          \n"
-    "jg        1b                                \n"
-  : "+r"(src_u),     // %0
-    "+r"(src_v),     // %1
-    "+r"(dst_uv),    // %2
-    "+r"(width)      // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2"
-#endif
-  );
-}
-
-void MergeUVRow_Unaligned_SSE2(const uint8* src_u, const uint8* src_v,
-                               uint8* dst_uv, int width) {
-  asm volatile (
-    "sub       %0,%1                             \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0           \n"
-    MEMOPREG(movdqu,0x00,0,1,1,xmm1)             //  movdqu    (%0,%1,1),%%xmm1
-    "lea       " MEMLEA(0x10,0) ",%0             \n"
-    "movdqa    %%xmm0,%%xmm2                     \n"
-    "punpcklbw %%xmm1,%%xmm0                     \n"
-    "punpckhbw %%xmm1,%%xmm2                     \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "           \n"
-    "movdqu    %%xmm2," MEMACCESS2(0x10,2) "     \n"
-    "lea       " MEMLEA(0x20,2) ",%2             \n"
-    "sub       $0x10,%3                          \n"
-    "jg        1b                                \n"
-  : "+r"(src_u),     // %0
-    "+r"(src_v),     // %1
-    "+r"(dst_uv),    // %2
-    "+r"(width)      // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2"
-#endif
-  );
-}
-#endif  // HAS_MERGEUVROW_SSE2
-
-#ifdef HAS_COPYROW_SSE2
-void CopyRow_SSE2(const uint8* src, uint8* dst, int count) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x20,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(count)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_COPYROW_SSE2
-
-#ifdef HAS_COPYROW_X86
-void CopyRow_X86(const uint8* src, uint8* dst, int width) {
-  size_t width_tmp = (size_t)(width);
-  asm volatile (
-    "shr       $0x2,%2                         \n"
-    "rep movsl " MEMMOVESTRING(0,1) "          \n"
-  : "+S"(src),  // %0
-    "+D"(dst),  // %1
-    "+c"(width_tmp) // %2
-  :
-  : "memory", "cc"
-  );
-}
-#endif  // HAS_COPYROW_X86
-
-#ifdef HAS_COPYROW_ERMS
-// Unaligned Multiple of 1.
-void CopyRow_ERMS(const uint8* src, uint8* dst, int width) {
-  size_t width_tmp = (size_t)(width);
-  asm volatile (
-    "rep movsb " MEMMOVESTRING(0,1) "          \n"
-  : "+S"(src),  // %0
-    "+D"(dst),  // %1
-    "+c"(width_tmp) // %2
-  :
-  : "memory", "cc"
-  );
-}
-#endif  // HAS_COPYROW_ERMS
-
-#ifdef HAS_ARGBCOPYALPHAROW_SSE2
-// width in pixels
-void ARGBCopyAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm0,%%xmm0                   \n"
-    "pslld     $0x18,%%xmm0                    \n"
-    "pcmpeqb   %%xmm1,%%xmm1                   \n"
-    "psrld     $0x8,%%xmm1                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm2         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm4         \n"
-    "movdqa    " MEMACCESS2(0x10,1) ",%%xmm5   \n"
-    "pand      %%xmm0,%%xmm2                   \n"
-    "pand      %%xmm0,%%xmm3                   \n"
-    "pand      %%xmm1,%%xmm4                   \n"
-    "pand      %%xmm1,%%xmm5                   \n"
-    "por       %%xmm4,%%xmm2                   \n"
-    "por       %%xmm5,%%xmm3                   \n"
-    "movdqa    %%xmm2," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBCOPYALPHAROW_SSE2
-
-#ifdef HAS_ARGBCOPYALPHAROW_AVX2
-// width in pixels
-void ARGBCopyAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    "vpcmpeqb  %%ymm0,%%ymm0,%%ymm0            \n"
-    "vpsrld    $0x8,%%ymm0,%%ymm0              \n"
-    LABELALIGN
-  "1:                                          \n"
-    "vmovdqu   " MEMACCESS(0) ",%%ymm1         \n"
-    "vmovdqu   " MEMACCESS2(0x20,0) ",%%ymm2   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "vpblendvb %%ymm0," MEMACCESS(1) ",%%ymm1,%%ymm1        \n"
-    "vpblendvb %%ymm0," MEMACCESS2(0x20,1) ",%%ymm2,%%ymm2  \n"
-    "vmovdqu   %%ymm1," MEMACCESS(1) "         \n"
-    "vmovdqu   %%ymm2," MEMACCESS2(0x20,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-    "vzeroupper                                \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2"
-#endif
-  );
-}
-#endif  // HAS_ARGBCOPYALPHAROW_AVX2
-
-#ifdef HAS_ARGBCOPYYTOALPHAROW_SSE2
-// width in pixels
-void ARGBCopyYToAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm0,%%xmm0                   \n"
-    "pslld     $0x18,%%xmm0                    \n"
-    "pcmpeqb   %%xmm1,%%xmm1                   \n"
-    "psrld     $0x8,%%xmm1                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "punpcklbw %%xmm2,%%xmm2                   \n"
-    "punpckhwd %%xmm2,%%xmm3                   \n"
-    "punpcklwd %%xmm2,%%xmm2                   \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm4         \n"
-    "movdqa    " MEMACCESS2(0x10,1) ",%%xmm5   \n"
-    "pand      %%xmm0,%%xmm2                   \n"
-    "pand      %%xmm0,%%xmm3                   \n"
-    "pand      %%xmm1,%%xmm4                   \n"
-    "pand      %%xmm1,%%xmm5                   \n"
-    "por       %%xmm4,%%xmm2                   \n"
-    "por       %%xmm5,%%xmm3                   \n"
-    "movdqa    %%xmm2," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBCOPYYTOALPHAROW_SSE2
-
-#ifdef HAS_ARGBCOPYYTOALPHAROW_AVX2
-// width in pixels
-void ARGBCopyYToAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
-  asm volatile (
-    "vpcmpeqb  %%ymm0,%%ymm0,%%ymm0            \n"
-    "vpsrld    $0x8,%%ymm0,%%ymm0              \n"
-    LABELALIGN
-  "1:                                          \n"
-    "vpmovzxbd " MEMACCESS(0) ",%%ymm1         \n"
-    "vpmovzxbd " MEMACCESS2(0x8,0) ",%%ymm2    \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "vpslld    $0x18,%%ymm1,%%ymm1             \n"
-    "vpslld    $0x18,%%ymm2,%%ymm2             \n"
-    "vpblendvb %%ymm0," MEMACCESS(1) ",%%ymm1,%%ymm1        \n"
-    "vpblendvb %%ymm0," MEMACCESS2(0x20,1) ",%%ymm2,%%ymm2  \n"
-    "vmovdqu   %%ymm1," MEMACCESS(1) "         \n"
-    "vmovdqu   %%ymm2," MEMACCESS2(0x20,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-    "vzeroupper                                \n"
-  : "+r"(src),   // %0
-    "+r"(dst),   // %1
-    "+r"(width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2"
-#endif
-  );
-}
-#endif  // HAS_ARGBCOPYYTOALPHAROW_AVX2
-
-#ifdef HAS_SETROW_X86
-void SetRow_X86(uint8* dst, uint32 v32, int width) {
-  size_t width_tmp = (size_t)(width);
-  asm volatile (
-    "shr       $0x2,%1                         \n"
-    "rep stosl " MEMSTORESTRING(eax,0) "       \n"
-    : "+D"(dst),       // %0
-      "+c"(width_tmp)  // %1
-    : "a"(v32)         // %2
-    : "memory", "cc");
-}
-
-void ARGBSetRows_X86(uint8* dst, uint32 v32, int width,
-                   int dst_stride, int height) {
-  for (int y = 0; y < height; ++y) {
-    size_t width_tmp = (size_t)(width);
-    uint32* d = (uint32*)(dst);
-    asm volatile (
-      "rep stosl " MEMSTORESTRING(eax,0) "     \n"
-      : "+D"(d),         // %0
-        "+c"(width_tmp)  // %1
-      : "a"(v32)         // %2
-      : "memory", "cc");
-    dst += dst_stride;
-  }
-}
-#endif  // HAS_SETROW_X86
-
-#ifdef HAS_YUY2TOYROW_SSE2
-void YUY2ToYRow_SSE2(const uint8* src_yuy2, uint8* dst_y, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void YUY2ToUVRow_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x00,0,4,1,xmm2)           //  movdqa  (%0,%4,1),%%xmm2
-    MEMOPREG(movdqa,0x10,0,4,1,xmm3)           //  movdqa  0x10(%0,%4,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  : "r"((intptr_t)(stride_yuy2))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void YUY2ToUV422Row_SSE2(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void YUY2ToYRow_Unaligned_SSE2(const uint8* src_yuy2,
-                               uint8* dst_y, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void YUY2ToUVRow_Unaligned_SSE2(const uint8* src_yuy2,
-                                int stride_yuy2,
-                                uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm2)           //  movdqu  (%0,%4,1),%%xmm2
-    MEMOPREG(movdqu,0x10,0,4,1,xmm3)           //  movdqu  0x10(%0,%4,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  : "r"((intptr_t)(stride_yuy2))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void YUY2ToUV422Row_Unaligned_SSE2(const uint8* src_yuy2,
-                                   uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_yuy2),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void UYVYToYRow_SSE2(const uint8* src_uyvy, uint8* dst_y, int pix) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void UYVYToUVRow_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x00,0,4,1,xmm2)           //  movdqa  (%0,%4,1),%%xmm2
-    MEMOPREG(movdqa,0x10,0,4,1,xmm3)           //  movdqa  0x10(%0,%4,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  : "r"((intptr_t)(stride_uyvy))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void UYVYToUV422Row_SSE2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void UYVYToYRow_Unaligned_SSE2(const uint8* src_uyvy,
-                               uint8* dst_y, int pix) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),  // %0
-    "+r"(dst_y),     // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void UYVYToUVRow_Unaligned_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                                uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x00,0,4,1,xmm2)           //  movdqu  (%0,%4,1),%%xmm2
-    MEMOPREG(movdqu,0x10,0,4,1,xmm3)           //  movdqu  0x10(%0,%4,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  : "r"((intptr_t)(stride_uyvy))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void UYVYToUV422Row_Unaligned_SSE2(const uint8* src_uyvy,
-                                   uint8* dst_u, uint8* dst_v, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-    "sub       %1,%2                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm1,0x00,1,2,1)             //  movq    %%xmm1,(%1,%2)
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x10,%3                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_uyvy),    // %0
-    "+r"(dst_u),       // %1
-    "+r"(dst_v),       // %2
-    "+r"(pix)          // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_YUY2TOYROW_SSE2
-
-#ifdef HAS_ARGBBLENDROW_SSE2
-// Blend 8 pixels at a time.
-void ARGBBlendRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                       uint8* dst_argb, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "psrlw     $0xf,%%xmm7                     \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "psrlw     $0x8,%%xmm6                     \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psllw     $0x8,%%xmm5                     \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "pslld     $0x18,%%xmm4                    \n"
-    "sub       $0x1,%3                         \n"
-    "je        91f                             \n"
-    "jl        99f                             \n"
-
-    // 1 pixel loop until destination pointer is aligned.
-  "10:                                         \n"
-    "test      $0xf,%2                         \n"
-    "je        19f                             \n"
-    "movd      " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm2         \n"
-    "psrlw     $0x8,%%xmm3                     \n"
-    "pshufhw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pshuflw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x4,1) ",%1            \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x1,%3                         \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x4,2) ",%2            \n"
-    "jge       10b                             \n"
-
-  "19:                                         \n"
-    "add       $1-4,%3                         \n"
-    "jl        49f                             \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "41:                                         \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm2         \n"
-    "psrlw     $0x8,%%xmm3                     \n"
-    "pshufhw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pshuflw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jge       41b                             \n"
-
-  "49:                                         \n"
-    "add       $0x3,%3                         \n"
-    "jl        99f                             \n"
-
-    // 1 pixel loop.
-  "91:                                         \n"
-    "movd      " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm2         \n"
-    "psrlw     $0x8,%%xmm3                     \n"
-    "pshufhw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pshuflw   $0xf5,%%xmm3,%%xmm3             \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x4,1) ",%1            \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x1,%3                         \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x4,2) ",%2            \n"
-    "jge       91b                             \n"
-  "99:                                         \n"
-  : "+r"(src_argb0),    // %0
-    "+r"(src_argb1),    // %1
-    "+r"(dst_argb),     // %2
-    "+r"(width)         // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBBLENDROW_SSE2
-
-#ifdef HAS_ARGBBLENDROW_SSSE3
-// Shuffle table for isolating alpha.
-static uvec8 kShuffleAlpha = {
-  3u, 0x80, 3u, 0x80, 7u, 0x80, 7u, 0x80,
-  11u, 0x80, 11u, 0x80, 15u, 0x80, 15u, 0x80
-};
-
-// Blend 8 pixels at a time
-// Shuffle table for reversing the bytes.
-
-// Same as SSE2, but replaces
-//    psrlw      xmm3, 8          // alpha
-//    pshufhw    xmm3, xmm3,0F5h  // 8 alpha words
-//    pshuflw    xmm3, xmm3,0F5h
-// with..
-//    pshufb     xmm3, kShuffleAlpha // alpha
-
-void ARGBBlendRow_SSSE3(const uint8* src_argb0, const uint8* src_argb1,
-                        uint8* dst_argb, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "psrlw     $0xf,%%xmm7                     \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "psrlw     $0x8,%%xmm6                     \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psllw     $0x8,%%xmm5                     \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "pslld     $0x18,%%xmm4                    \n"
-    "sub       $0x1,%3                         \n"
-    "je        91f                             \n"
-    "jl        99f                             \n"
-
-    // 1 pixel loop until destination pointer is aligned.
-  "10:                                         \n"
-    "test      $0xf,%2                         \n"
-    "je        19f                             \n"
-    "movd      " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm2         \n"
-    "pshufb    %4,%%xmm3                       \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x4,1) ",%1            \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x1,%3                         \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x4,2) ",%2            \n"
-    "jge       10b                             \n"
-
-  "19:                                         \n"
-    "add       $1-4,%3                         \n"
-    "jl        49f                             \n"
-    "test      $0xf,%0                         \n"
-    "jne       41f                             \n"
-    "test      $0xf,%1                         \n"
-    "jne       41f                             \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "40:                                         \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm2         \n"
-    "pshufb    %4,%%xmm3                       \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jge       40b                             \n"
-    "jmp       49f                             \n"
-
-    // 4 pixel unaligned loop.
-    LABELALIGN
-  "41:                                         \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm2         \n"
-    "pshufb    %4,%%xmm3                       \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jge       41b                             \n"
-
-  "49:                                         \n"
-    "add       $0x3,%3                         \n"
-    "jl        99f                             \n"
-
-    // 1 pixel loop.
-  "91:                                         \n"
-    "movd      " MEMACCESS(0) ",%%xmm3         \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "movdqa    %%xmm3,%%xmm0                   \n"
-    "pxor      %%xmm4,%%xmm3                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm2         \n"
-    "pshufb    %4,%%xmm3                       \n"
-    "pand      %%xmm6,%%xmm2                   \n"
-    "paddw     %%xmm7,%%xmm3                   \n"
-    "pmullw    %%xmm3,%%xmm2                   \n"
-    "movd      " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x4,1) ",%1            \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "por       %%xmm4,%%xmm0                   \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm2                     \n"
-    "paddusb   %%xmm2,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x1,%3                         \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x4,2) ",%2            \n"
-    "jge       91b                             \n"
-  "99:                                         \n"
-  : "+r"(src_argb0),    // %0
-    "+r"(src_argb1),    // %1
-    "+r"(dst_argb),     // %2
-    "+r"(width)         // %3
-  : "m"(kShuffleAlpha)  // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBBLENDROW_SSSE3
-
-#ifdef HAS_ARGBATTENUATEROW_SSE2
-// Attenuate 4 pixels at a time.
-// aligned to 16 bytes
-void ARGBAttenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "pslld     $0x18,%%xmm4                    \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrld     $0x8,%%xmm5                     \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "pshufhw   $0xff,%%xmm0,%%xmm2             \n"
-    "pshuflw   $0xff,%%xmm2,%%xmm2             \n"
-    "pmulhuw   %%xmm2,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm1         \n"
-    "punpckhbw %%xmm1,%%xmm1                   \n"
-    "pshufhw   $0xff,%%xmm1,%%xmm2             \n"
-    "pshuflw   $0xff,%%xmm2,%%xmm2             \n"
-    "pmulhuw   %%xmm2,%%xmm1                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "pand      %%xmm4,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "por       %%xmm2,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+r"(width)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBATTENUATEROW_SSE2
-
-#ifdef HAS_ARGBATTENUATEROW_SSSE3
-// Shuffle table duplicating alpha
-static uvec8 kShuffleAlpha0 = {
-  3u, 3u, 3u, 3u, 3u, 3u, 128u, 128u, 7u, 7u, 7u, 7u, 7u, 7u, 128u, 128u,
-};
-static uvec8 kShuffleAlpha1 = {
-  11u, 11u, 11u, 11u, 11u, 11u, 128u, 128u,
-  15u, 15u, 15u, 15u, 15u, 15u, 128u, 128u,
-};
-// Attenuate 4 pixels at a time.
-// aligned to 16 bytes
-void ARGBAttenuateRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width) {
-  asm volatile (
-    "pcmpeqb   %%xmm3,%%xmm3                   \n"
-    "pslld     $0x18,%%xmm3                    \n"
-    "movdqa    %3,%%xmm4                       \n"
-    "movdqa    %4,%%xmm5                       \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm1         \n"
-    "punpcklbw %%xmm1,%%xmm1                   \n"
-    "pmulhuw   %%xmm1,%%xmm0                   \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm1         \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm2         \n"
-    "punpckhbw %%xmm2,%%xmm2                   \n"
-    "pmulhuw   %%xmm2,%%xmm1                   \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "pand      %%xmm3,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "por       %%xmm2,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+r"(width)        // %2
-  : "m"(kShuffleAlpha0),  // %3
-    "m"(kShuffleAlpha1)  // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBATTENUATEROW_SSSE3
-
-#ifdef HAS_ARGBUNATTENUATEROW_SSE2
-// Unattenuate 4 pixels at a time.
-// aligned to 16 bytes
-void ARGBUnattenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb,
-                             int width) {
-  uintptr_t alpha = 0;
-  asm volatile (
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movzb     " MEMACCESS2(0x03,0) ",%3       \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    MEMOPREG(movd,0x00,4,3,4,xmm2)             //  movd      0x0(%4,%3,4),%%xmm2
-    "movzb     " MEMACCESS2(0x07,0) ",%3       \n"
-    MEMOPREG(movd,0x00,4,3,4,xmm3)             //  movd      0x0(%4,%3,4),%%xmm3
-    "pshuflw   $0x40,%%xmm2,%%xmm2             \n"
-    "pshuflw   $0x40,%%xmm3,%%xmm3             \n"
-    "movlhps   %%xmm3,%%xmm2                   \n"
-    "pmulhuw   %%xmm2,%%xmm0                   \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm1         \n"
-    "movzb     " MEMACCESS2(0x0b,0) ",%3       \n"
-    "punpckhbw %%xmm1,%%xmm1                   \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,4,3,4,xmm2)             //  movd      0x0(%4,%3,4),%%xmm2
-    "movzb     " MEMACCESS2(0x0f,0) ",%3       \n"
-    MEMOPREG(movd,0x00,4,3,4,xmm3)             //  movd      0x0(%4,%3,4),%%xmm3
-    "pshuflw   $0x40,%%xmm2,%%xmm2             \n"
-    "pshuflw   $0x40,%%xmm3,%%xmm3             \n"
-    "movlhps   %%xmm3,%%xmm2                   \n"
-    "pmulhuw   %%xmm2,%%xmm1                   \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+r"(width),       // %2
-    "+r"(alpha)        // %3
-  : "r"(fixed_invtbl8)  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBUNATTENUATEROW_SSE2
-
-#ifdef HAS_ARGBGRAYROW_SSSE3
-// Convert 8 ARGB pixels (64 bytes) to 8 Gray ARGB pixels
-void ARGBGrayRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width) {
-  asm volatile (
-    "movdqa    %3,%%xmm4                       \n"
-    "movdqa    %4,%%xmm5                       \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "pmaddubsw %%xmm4,%%xmm0                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "phaddw    %%xmm1,%%xmm0                   \n"
-    "paddw     %%xmm5,%%xmm0                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm2         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm3   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrld     $0x18,%%xmm2                    \n"
-    "psrld     $0x18,%%xmm3                    \n"
-    "packuswb  %%xmm3,%%xmm2                   \n"
-    "packuswb  %%xmm2,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm3                   \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "punpcklbw %%xmm2,%%xmm3                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm3,%%xmm0                   \n"
-    "punpckhwd %%xmm3,%%xmm1                   \n"
-    "sub       $0x8,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_argb),   // %1
-    "+r"(width)       // %2
-  : "m"(kARGBToYJ),   // %3
-    "m"(kAddYJ64)     // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBGRAYROW_SSSE3
-
-#ifdef HAS_ARGBSEPIAROW_SSSE3
-//    b = (r * 35 + g * 68 + b * 17) >> 7
-//    g = (r * 45 + g * 88 + b * 22) >> 7
-//    r = (r * 50 + g * 98 + b * 24) >> 7
-// Constant for ARGB color to sepia tone
-static vec8 kARGBToSepiaB = {
-  17, 68, 35, 0, 17, 68, 35, 0, 17, 68, 35, 0, 17, 68, 35, 0
-};
-
-static vec8 kARGBToSepiaG = {
-  22, 88, 45, 0, 22, 88, 45, 0, 22, 88, 45, 0, 22, 88, 45, 0
-};
-
-static vec8 kARGBToSepiaR = {
-  24, 98, 50, 0, 24, 98, 50, 0, 24, 98, 50, 0, 24, 98, 50, 0
-};
-
-// Convert 8 ARGB pixels (32 bytes) to 8 Sepia ARGB pixels.
-void ARGBSepiaRow_SSSE3(uint8* dst_argb, int width) {
-  asm volatile (
-    "movdqa    %2,%%xmm2                       \n"
-    "movdqa    %3,%%xmm3                       \n"
-    "movdqa    %4,%%xmm4                       \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm6   \n"
-    "pmaddubsw %%xmm2,%%xmm0                   \n"
-    "pmaddubsw %%xmm2,%%xmm6                   \n"
-    "phaddw    %%xmm6,%%xmm0                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm5         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "pmaddubsw %%xmm3,%%xmm5                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "phaddw    %%xmm1,%%xmm5                   \n"
-    "psrlw     $0x7,%%xmm5                     \n"
-    "packuswb  %%xmm5,%%xmm5                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm5         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "pmaddubsw %%xmm4,%%xmm5                   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "phaddw    %%xmm1,%%xmm5                   \n"
-    "psrlw     $0x7,%%xmm5                     \n"
-    "packuswb  %%xmm5,%%xmm5                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm6         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "psrld     $0x18,%%xmm6                    \n"
-    "psrld     $0x18,%%xmm1                    \n"
-    "packuswb  %%xmm1,%%xmm6                   \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "punpcklbw %%xmm6,%%xmm5                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklwd %%xmm5,%%xmm0                   \n"
-    "punpckhwd %%xmm5,%%xmm1                   \n"
-    "sub       $0x8,%1                         \n"
-    "movdqa    %%xmm0," MEMACCESS(0) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,0) "   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "jg        1b                              \n"
-  : "+r"(dst_argb),      // %0
-    "+r"(width)          // %1
-  : "m"(kARGBToSepiaB),  // %2
-    "m"(kARGBToSepiaG),  // %3
-    "m"(kARGBToSepiaR)   // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-#endif  // HAS_ARGBSEPIAROW_SSSE3
-
-#ifdef HAS_ARGBCOLORMATRIXROW_SSSE3
-// Tranform 8 ARGB pixels (32 bytes) with color matrix.
-// Same as Sepia except matrix is provided.
-void ARGBColorMatrixRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                              const int8* matrix_argb, int width) {
-  asm volatile (
-    "movdqu    " MEMACCESS(3) ",%%xmm5         \n"
-    "pshufd    $0x00,%%xmm5,%%xmm2             \n"
-    "pshufd    $0x55,%%xmm5,%%xmm3             \n"
-    "pshufd    $0xaa,%%xmm5,%%xmm4             \n"
-    "pshufd    $0xff,%%xmm5,%%xmm5             \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm7   \n"
-    "pmaddubsw %%xmm2,%%xmm0                   \n"
-    "pmaddubsw %%xmm2,%%xmm7                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm6         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "pmaddubsw %%xmm3,%%xmm6                   \n"
-    "pmaddubsw %%xmm3,%%xmm1                   \n"
-    "phaddsw   %%xmm7,%%xmm0                   \n"
-    "phaddsw   %%xmm1,%%xmm6                   \n"
-    "psraw     $0x6,%%xmm0                     \n"
-    "psraw     $0x6,%%xmm6                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "punpcklbw %%xmm6,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm1         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm7   \n"
-    "pmaddubsw %%xmm4,%%xmm1                   \n"
-    "pmaddubsw %%xmm4,%%xmm7                   \n"
-    "phaddsw   %%xmm7,%%xmm1                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm6         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm7   \n"
-    "pmaddubsw %%xmm5,%%xmm6                   \n"
-    "pmaddubsw %%xmm5,%%xmm7                   \n"
-    "phaddsw   %%xmm7,%%xmm6                   \n"
-    "psraw     $0x6,%%xmm1                     \n"
-    "psraw     $0x6,%%xmm6                     \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "punpcklbw %%xmm6,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm6                   \n"
-    "punpcklwd %%xmm1,%%xmm0                   \n"
-    "punpckhwd %%xmm1,%%xmm6                   \n"
-    "sub       $0x8,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm6," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),      // %0
-    "+r"(dst_argb),      // %1
-    "+r"(width)          // %2
-  : "r"(matrix_argb)     // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBCOLORMATRIXROW_SSSE3
-
-#ifdef HAS_ARGBQUANTIZEROW_SSE2
-// Quantize 4 ARGB pixels (16 bytes).
-// aligned to 16 bytes
-void ARGBQuantizeRow_SSE2(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width) {
-  asm volatile (
-    "movd      %2,%%xmm2                       \n"
-    "movd      %3,%%xmm3                       \n"
-    "movd      %4,%%xmm4                       \n"
-    "pshuflw   $0x40,%%xmm2,%%xmm2             \n"
-    "pshufd    $0x44,%%xmm2,%%xmm2             \n"
-    "pshuflw   $0x40,%%xmm3,%%xmm3             \n"
-    "pshufd    $0x44,%%xmm3,%%xmm3             \n"
-    "pshuflw   $0x40,%%xmm4,%%xmm4             \n"
-    "pshufd    $0x44,%%xmm4,%%xmm4             \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "pslld     $0x18,%%xmm6                    \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "pmulhuw   %%xmm2,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm1         \n"
-    "punpckhbw %%xmm5,%%xmm1                   \n"
-    "pmulhuw   %%xmm2,%%xmm1                   \n"
-    "pmullw    %%xmm3,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm7         \n"
-    "pmullw    %%xmm3,%%xmm1                   \n"
-    "pand      %%xmm6,%%xmm7                   \n"
-    "paddw     %%xmm4,%%xmm0                   \n"
-    "paddw     %%xmm4,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "por       %%xmm7,%%xmm0                   \n"
-    "sub       $0x4,%1                         \n"
-    "movdqa    %%xmm0," MEMACCESS(0) "         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "jg        1b                              \n"
-  : "+r"(dst_argb),       // %0
-    "+r"(width)           // %1
-  : "r"(scale),           // %2
-    "r"(interval_size),   // %3
-    "r"(interval_offset)  // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBQUANTIZEROW_SSE2
-
-#ifdef HAS_ARGBSHADEROW_SSE2
-// Shade 4 pixels at a time by specified value.
-// Aligned to 16 bytes.
-void ARGBShadeRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width,
-                       uint32 value) {
-  asm volatile (
-    "movd      %3,%%xmm2                       \n"
-    "punpcklbw %%xmm2,%%xmm2                   \n"
-    "punpcklqdq %%xmm2,%%xmm2                  \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "punpckhbw %%xmm1,%%xmm1                   \n"
-    "pmulhuw   %%xmm2,%%xmm0                   \n"
-    "pmulhuw   %%xmm2,%%xmm1                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(width)      // %2
-  : "r"(value)       // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2"
-#endif
-  );
-}
-#endif  // HAS_ARGBSHADEROW_SSE2
-
-#ifdef HAS_ARGBMULTIPLYROW_SSE2
-// Multiply 2 rows of ARGB pixels together, 4 pixels at a time.
-void ARGBMultiplyRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  asm volatile (
-    "pxor      %%xmm5,%%xmm5                   \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "movdqu    %%xmm0,%%xmm1                   \n"
-    "movdqu    %%xmm2,%%xmm3                   \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "punpckhbw %%xmm1,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "punpckhbw %%xmm5,%%xmm3                   \n"
-    "pmulhuw   %%xmm2,%%xmm0                   \n"
-    "pmulhuw   %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBMULTIPLYROW_SSE2
-
-#ifdef HAS_ARGBADDROW_SSE2
-// Add 2 rows of ARGB pixels together, 4 pixels at a time.
-void ARGBAddRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                     uint8* dst_argb, int width) {
-  asm volatile (
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_ARGBADDROW_SSE2
-
-#ifdef HAS_ARGBSUBTRACTROW_SSE2
-// Subtract 2 rows of ARGB pixels, 4 pixels at a time.
-void ARGBSubtractRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  asm volatile (
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "psubusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb0),  // %0
-    "+r"(src_argb1),  // %1
-    "+r"(dst_argb),   // %2
-    "+r"(width)       // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_ARGBSUBTRACTROW_SSE2
-
-#ifdef HAS_SOBELXROW_SSE2
-// SobelX as a matrix is
-// -1  0  1
-// -2  0  2
-// -1  0  1
-void SobelXRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobelx, int width) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    "sub       %0,%2                           \n"
-    "sub       %0,%3                           \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "movq      " MEMACCESS2(0x2,0) ",%%xmm1    \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "psubw     %%xmm1,%%xmm0                   \n"
-    BUNDLEALIGN
-    MEMOPREG(movq,0x00,0,1,1,xmm1)             //  movq      (%0,%1,1),%%xmm1
-    MEMOPREG(movq,0x02,0,1,1,xmm2)             //  movq      0x2(%0,%1,1),%%xmm2
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "psubw     %%xmm2,%%xmm1                   \n"
-    BUNDLEALIGN
-    MEMOPREG(movq,0x00,0,2,1,xmm2)             //  movq      (%0,%2,1),%%xmm2
-    MEMOPREG(movq,0x02,0,2,1,xmm3)             //  movq      0x2(%0,%2,1),%%xmm3
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "punpcklbw %%xmm5,%%xmm3                   \n"
-    "psubw     %%xmm3,%%xmm2                   \n"
-    "paddw     %%xmm2,%%xmm0                   \n"
-    "paddw     %%xmm1,%%xmm0                   \n"
-    "paddw     %%xmm1,%%xmm0                   \n"
-    "pxor      %%xmm1,%%xmm1                   \n"
-    "psubw     %%xmm0,%%xmm1                   \n"
-    "pmaxsw    %%xmm1,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "sub       $0x8,%4                         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm0,0x00,0,3,1)             //  movq      %%xmm0,(%0,%3,1)
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "jg        1b                              \n"
-  : "+r"(src_y0),      // %0
-    "+r"(src_y1),      // %1
-    "+r"(src_y2),      // %2
-    "+r"(dst_sobelx),  // %3
-    "+r"(width)        // %4
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_SOBELXROW_SSE2
-
-#ifdef HAS_SOBELYROW_SSE2
-// SobelY as a matrix is
-// -1 -2 -1
-//  0  0  0
-//  1  2  1
-void SobelYRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    "sub       %0,%2                           \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movq,0x00,0,1,1,xmm1)             //  movq      (%0,%1,1),%%xmm1
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "psubw     %%xmm1,%%xmm0                   \n"
-    BUNDLEALIGN
-    "movq      " MEMACCESS2(0x1,0) ",%%xmm1    \n"
-    MEMOPREG(movq,0x01,0,1,1,xmm2)             //  movq      0x1(%0,%1,1),%%xmm2
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "psubw     %%xmm2,%%xmm1                   \n"
-    BUNDLEALIGN
-    "movq      " MEMACCESS2(0x2,0) ",%%xmm2    \n"
-    MEMOPREG(movq,0x02,0,1,1,xmm3)             //  movq      0x2(%0,%1,1),%%xmm3
-    "punpcklbw %%xmm5,%%xmm2                   \n"
-    "punpcklbw %%xmm5,%%xmm3                   \n"
-    "psubw     %%xmm3,%%xmm2                   \n"
-    "paddw     %%xmm2,%%xmm0                   \n"
-    "paddw     %%xmm1,%%xmm0                   \n"
-    "paddw     %%xmm1,%%xmm0                   \n"
-    "pxor      %%xmm1,%%xmm1                   \n"
-    "psubw     %%xmm0,%%xmm1                   \n"
-    "pmaxsw    %%xmm1,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "sub       $0x8,%3                         \n"
-    BUNDLEALIGN
-    MEMOPMEM(movq,xmm0,0x00,0,2,1)             //  movq      %%xmm0,(%0,%2,1)
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "jg        1b                              \n"
-  : "+r"(src_y0),      // %0
-    "+r"(src_y1),      // %1
-    "+r"(dst_sobely),  // %2
-    "+r"(width)        // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_SOBELYROW_SSE2
-
-#ifdef HAS_SOBELROW_SSE2
-// Adds Sobel X and Sobel Y and stores Sobel into ARGB.
-// A = 255
-// R = Sobel
-// G = Sobel
-// B = Sobel
-void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                   uint8* dst_argb, int width) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pslld     $0x18,%%xmm5                    \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,0,1,1,xmm1)           //  movdqa    (%0,%1,1),%%xmm1
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "punpcklbw %%xmm0,%%xmm2                   \n"
-    "punpckhbw %%xmm0,%%xmm0                   \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "punpcklwd %%xmm2,%%xmm1                   \n"
-    "punpckhwd %%xmm2,%%xmm2                   \n"
-    "por       %%xmm5,%%xmm1                   \n"
-    "por       %%xmm5,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm3                   \n"
-    "punpcklwd %%xmm0,%%xmm3                   \n"
-    "punpckhwd %%xmm0,%%xmm0                   \n"
-    "por       %%xmm5,%%xmm3                   \n"
-    "por       %%xmm5,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movdqa    %%xmm1," MEMACCESS(2) "         \n"
-    "movdqa    %%xmm2," MEMACCESS2(0x10,2) "   \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x20,2) "   \n"
-    "movdqa    %%xmm0," MEMACCESS2(0x30,2) "   \n"
-    "lea       " MEMLEA(0x40,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(width)        // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_SOBELROW_SSE2
-
-#ifdef HAS_SOBELTOPLANEROW_SSE2
-// Adds Sobel X and Sobel Y and stores Sobel into a plane.
-void SobelToPlaneRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_y, int width) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "pslld     $0x18,%%xmm5                    \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,0,1,1,xmm1)           //  movdqa    (%0,%1,1),%%xmm1
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%3                        \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_y),       // %2
-    "+r"(width)        // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-#endif  // HAS_SOBELTOPLANEROW_SSE2
-
-#ifdef HAS_SOBELXYROW_SSE2
-// Mixes Sobel X, Sobel Y and Sobel into ARGB.
-// A = 255
-// R = Sobel X
-// G = Sobel
-// B = Sobel Y
-void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-
-    // 8 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,0,1,1,xmm1)           //  movdqa    (%0,%1,1),%%xmm1
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "paddusb   %%xmm1,%%xmm2                   \n"
-    "movdqa    %%xmm0,%%xmm3                   \n"
-    "punpcklbw %%xmm5,%%xmm3                   \n"
-    "punpckhbw %%xmm5,%%xmm0                   \n"
-    "movdqa    %%xmm1,%%xmm4                   \n"
-    "punpcklbw %%xmm2,%%xmm4                   \n"
-    "punpckhbw %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm4,%%xmm6                   \n"
-    "punpcklwd %%xmm3,%%xmm6                   \n"
-    "punpckhwd %%xmm3,%%xmm4                   \n"
-    "movdqa    %%xmm1,%%xmm7                   \n"
-    "punpcklwd %%xmm0,%%xmm7                   \n"
-    "punpckhwd %%xmm0,%%xmm1                   \n"
-    "sub       $0x10,%3                        \n"
-    "movdqa    %%xmm6," MEMACCESS(2) "         \n"
-    "movdqa    %%xmm4," MEMACCESS2(0x10,2) "   \n"
-    "movdqa    %%xmm7," MEMACCESS2(0x20,2) "   \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x30,2) "   \n"
-    "lea       " MEMLEA(0x40,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_sobelx),  // %0
-    "+r"(src_sobely),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(width)        // %3
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_SOBELXYROW_SSE2
-
-#ifdef HAS_COMPUTECUMULATIVESUMROW_SSE2
-// Creates a table of cumulative sums where each value is a sum of all values
-// above and to the left of the value, inclusive of the value.
-void ComputeCumulativeSumRow_SSE2(const uint8* row, int32* cumsum,
-                                  const int32* previous_cumsum, int width) {
-  asm volatile (
-    "pxor      %%xmm0,%%xmm0                   \n"
-    "pxor      %%xmm1,%%xmm1                   \n"
-    "sub       $0x4,%3                         \n"
-    "jl        49f                             \n"
-    "test      $0xf,%1                         \n"
-    "jne       49f                             \n"
-
-  // 4 pixel loop                              \n"
-    LABELALIGN
-  "40:                                         \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm2,%%xmm4                   \n"
-    "punpcklbw %%xmm1,%%xmm2                   \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "punpcklwd %%xmm1,%%xmm2                   \n"
-    "punpckhwd %%xmm1,%%xmm3                   \n"
-    "punpckhbw %%xmm1,%%xmm4                   \n"
-    "movdqa    %%xmm4,%%xmm5                   \n"
-    "punpcklwd %%xmm1,%%xmm4                   \n"
-    "punpckhwd %%xmm1,%%xmm5                   \n"
-    "paddd     %%xmm2,%%xmm0                   \n"
-    "movdqa    " MEMACCESS(2) ",%%xmm2         \n"
-    "paddd     %%xmm0,%%xmm2                   \n"
-    "paddd     %%xmm3,%%xmm0                   \n"
-    "movdqa    " MEMACCESS2(0x10,2) ",%%xmm3   \n"
-    "paddd     %%xmm0,%%xmm3                   \n"
-    "paddd     %%xmm4,%%xmm0                   \n"
-    "movdqa    " MEMACCESS2(0x20,2) ",%%xmm4   \n"
-    "paddd     %%xmm0,%%xmm4                   \n"
-    "paddd     %%xmm5,%%xmm0                   \n"
-    "movdqa    " MEMACCESS2(0x30,2) ",%%xmm5   \n"
-    "lea       " MEMLEA(0x40,2) ",%2           \n"
-    "paddd     %%xmm0,%%xmm5                   \n"
-    "movdqa    %%xmm2," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm3," MEMACCESS2(0x10,1) "   \n"
-    "movdqa    %%xmm4," MEMACCESS2(0x20,1) "   \n"
-    "movdqa    %%xmm5," MEMACCESS2(0x30,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "sub       $0x4,%3                         \n"
-    "jge       40b                             \n"
-
-  "49:                                         \n"
-    "add       $0x3,%3                         \n"
-    "jl        19f                             \n"
-
-  // 1 pixel loop                              \n"
-    LABELALIGN
-  "10:                                         \n"
-    "movd      " MEMACCESS(0) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "punpcklbw %%xmm1,%%xmm2                   \n"
-    "punpcklwd %%xmm1,%%xmm2                   \n"
-    "paddd     %%xmm2,%%xmm0                   \n"
-    "movdqu    " MEMACCESS(2) ",%%xmm2         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "paddd     %%xmm0,%%xmm2                   \n"
-    "movdqu    %%xmm2," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x1,%3                         \n"
-    "jge       10b                             \n"
-
-  "19:                                         \n"
-  : "+r"(row),  // %0
-    "+r"(cumsum),  // %1
-    "+r"(previous_cumsum),  // %2
-    "+r"(width)  // %3
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_COMPUTECUMULATIVESUMROW_SSE2
-
-#ifdef HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
-void CumulativeSumToAverageRow_SSE2(const int32* topleft, const int32* botleft,
-                                    int width, int area, uint8* dst,
-                                    int count) {
-  asm volatile (
-    "movd      %5,%%xmm5                       \n"
-    "cvtdq2ps  %%xmm5,%%xmm5                   \n"
-    "rcpss     %%xmm5,%%xmm4                   \n"
-    "pshufd    $0x0,%%xmm4,%%xmm4              \n"
-    "sub       $0x4,%3                         \n"
-    "jl        49f                             \n"
-    "cmpl      $0x80,%5                        \n"
-    "ja        40f                             \n"
-
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "psrld     $0x10,%%xmm6                    \n"
-    "cvtdq2ps  %%xmm6,%%xmm6                   \n"
-    "addps     %%xmm6,%%xmm5                   \n"
-    "mulps     %%xmm4,%%xmm5                   \n"
-    "cvtps2dq  %%xmm5,%%xmm5                   \n"
-    "packssdw  %%xmm5,%%xmm5                   \n"
-
-  // 4 pixel small loop                        \n"
-    LABELALIGN
-  "4:                                         \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    BUNDLEALIGN
-    MEMOPREG(psubd,0x00,0,4,4,xmm0)            // psubd    0x00(%0,%4,4),%%xmm0
-    MEMOPREG(psubd,0x10,0,4,4,xmm1)            // psubd    0x10(%0,%4,4),%%xmm1
-    MEMOPREG(psubd,0x20,0,4,4,xmm2)            // psubd    0x20(%0,%4,4),%%xmm2
-    MEMOPREG(psubd,0x30,0,4,4,xmm3)            // psubd    0x30(%0,%4,4),%%xmm3
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "psubd     " MEMACCESS(1) ",%%xmm0         \n"
-    "psubd     " MEMACCESS2(0x10,1) ",%%xmm1   \n"
-    "psubd     " MEMACCESS2(0x20,1) ",%%xmm2   \n"
-    "psubd     " MEMACCESS2(0x30,1) ",%%xmm3   \n"
-    BUNDLEALIGN
-    MEMOPREG(paddd,0x00,1,4,4,xmm0)            // paddd    0x00(%1,%4,4),%%xmm0
-    MEMOPREG(paddd,0x10,1,4,4,xmm1)            // paddd    0x10(%1,%4,4),%%xmm1
-    MEMOPREG(paddd,0x20,1,4,4,xmm2)            // paddd    0x20(%1,%4,4),%%xmm2
-    MEMOPREG(paddd,0x30,1,4,4,xmm3)            // paddd    0x30(%1,%4,4),%%xmm3
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "packssdw  %%xmm1,%%xmm0                   \n"
-    "packssdw  %%xmm3,%%xmm2                   \n"
-    "pmulhuw   %%xmm5,%%xmm0                   \n"
-    "pmulhuw   %%xmm5,%%xmm2                   \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "sub       $0x4,%3                         \n"
-    "jge       4b                              \n"
-    "jmp       49f                             \n"
-
-  // 4 pixel loop                              \n"
-    LABELALIGN
-  "40:                                         \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "movdqa    " MEMACCESS2(0x20,0) ",%%xmm2   \n"
-    "movdqa    " MEMACCESS2(0x30,0) ",%%xmm3   \n"
-    BUNDLEALIGN
-    MEMOPREG(psubd,0x00,0,4,4,xmm0)            // psubd    0x00(%0,%4,4),%%xmm0
-    MEMOPREG(psubd,0x10,0,4,4,xmm1)            // psubd    0x10(%0,%4,4),%%xmm1
-    MEMOPREG(psubd,0x20,0,4,4,xmm2)            // psubd    0x20(%0,%4,4),%%xmm2
-    MEMOPREG(psubd,0x30,0,4,4,xmm3)            // psubd    0x30(%0,%4,4),%%xmm3
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "psubd     " MEMACCESS(1) ",%%xmm0         \n"
-    "psubd     " MEMACCESS2(0x10,1) ",%%xmm1   \n"
-    "psubd     " MEMACCESS2(0x20,1) ",%%xmm2   \n"
-    "psubd     " MEMACCESS2(0x30,1) ",%%xmm3   \n"
-    BUNDLEALIGN
-    MEMOPREG(paddd,0x00,1,4,4,xmm0)            // paddd    0x00(%1,%4,4),%%xmm0
-    MEMOPREG(paddd,0x10,1,4,4,xmm1)            // paddd    0x10(%1,%4,4),%%xmm1
-    MEMOPREG(paddd,0x20,1,4,4,xmm2)            // paddd    0x20(%1,%4,4),%%xmm2
-    MEMOPREG(paddd,0x30,1,4,4,xmm3)            // paddd    0x30(%1,%4,4),%%xmm3
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "cvtdq2ps  %%xmm0,%%xmm0                   \n"
-    "cvtdq2ps  %%xmm1,%%xmm1                   \n"
-    "mulps     %%xmm4,%%xmm0                   \n"
-    "mulps     %%xmm4,%%xmm1                   \n"
-    "cvtdq2ps  %%xmm2,%%xmm2                   \n"
-    "cvtdq2ps  %%xmm3,%%xmm3                   \n"
-    "mulps     %%xmm4,%%xmm2                   \n"
-    "mulps     %%xmm4,%%xmm3                   \n"
-    "cvtps2dq  %%xmm0,%%xmm0                   \n"
-    "cvtps2dq  %%xmm1,%%xmm1                   \n"
-    "cvtps2dq  %%xmm2,%%xmm2                   \n"
-    "cvtps2dq  %%xmm3,%%xmm3                   \n"
-    "packssdw  %%xmm1,%%xmm0                   \n"
-    "packssdw  %%xmm3,%%xmm2                   \n"
-    "packuswb  %%xmm2,%%xmm0                   \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "sub       $0x4,%3                         \n"
-    "jge       40b                             \n"
-
-  "49:                                         \n"
-    "add       $0x3,%3                         \n"
-    "jl        19f                             \n"
-
-  // 1 pixel loop                              \n"
-    LABELALIGN
-  "10:                                         \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(psubd,0x00,0,4,4,xmm0)            // psubd    0x00(%0,%4,4),%%xmm0
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "psubd     " MEMACCESS(1) ",%%xmm0         \n"
-    BUNDLEALIGN
-    MEMOPREG(paddd,0x00,1,4,4,xmm0)            // paddd    0x00(%1,%4,4),%%xmm0
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "cvtdq2ps  %%xmm0,%%xmm0                   \n"
-    "mulps     %%xmm4,%%xmm0                   \n"
-    "cvtps2dq  %%xmm0,%%xmm0                   \n"
-    "packssdw  %%xmm0,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x4,2) ",%2            \n"
-    "sub       $0x1,%3                         \n"
-    "jge       10b                             \n"
-  "19:                                         \n"
-  : "+r"(topleft),  // %0
-    "+r"(botleft),  // %1
-    "+r"(dst),      // %2
-    "+rm"(count)    // %3
-  : "r"((intptr_t)(width)),  // %4
-    "rm"(area)     // %5
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-#endif  // HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
-
-#ifdef HAS_ARGBAFFINEROW_SSE2
-// Copy ARGB pixels from source image with slope to a row of destination.
-LIBYUV_API
-void ARGBAffineRow_SSE2(const uint8* src_argb, int src_argb_stride,
-                        uint8* dst_argb, const float* src_dudv, int width) {
-  intptr_t src_argb_stride_temp = src_argb_stride;
-  intptr_t temp = 0;
-  asm volatile (
-    "movq      " MEMACCESS(3) ",%%xmm2         \n"
-    "movq      " MEMACCESS2(0x08,3) ",%%xmm7   \n"
-    "shl       $0x10,%1                        \n"
-    "add       $0x4,%1                         \n"
-    "movd      %1,%%xmm5                       \n"
-    "sub       $0x4,%4                         \n"
-    "jl        49f                             \n"
-
-    "pshufd    $0x44,%%xmm7,%%xmm7             \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "movdqa    %%xmm2,%%xmm0                   \n"
-    "addps     %%xmm7,%%xmm0                   \n"
-    "movlhps   %%xmm0,%%xmm2                   \n"
-    "movdqa    %%xmm7,%%xmm4                   \n"
-    "addps     %%xmm4,%%xmm4                   \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "addps     %%xmm4,%%xmm3                   \n"
-    "addps     %%xmm4,%%xmm4                   \n"
-
-  // 4 pixel loop                              \n"
-    LABELALIGN
-  "40:                                         \n"
-    "cvttps2dq %%xmm2,%%xmm0                   \n"  // x, y float to int first 2
-    "cvttps2dq %%xmm3,%%xmm1                   \n"  // x, y float to int next 2
-    "packssdw  %%xmm1,%%xmm0                   \n"  // x, y as 8 shorts
-    "pmaddwd   %%xmm5,%%xmm0                   \n"  // off = x * 4 + y * stride
-    "movd      %%xmm0,%k1                      \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-    "movd      %%xmm0,%k5                      \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,0,1,1,xmm1)             //  movd      (%0,%1,1),%%xmm1
-    MEMOPREG(movd,0x00,0,5,1,xmm6)             //  movd      (%0,%5,1),%%xmm6
-    "punpckldq %%xmm6,%%xmm1                   \n"
-    "addps     %%xmm4,%%xmm2                   \n"
-    "movq      %%xmm1," MEMACCESS(2) "         \n"
-    "movd      %%xmm0,%k1                      \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-    "movd      %%xmm0,%k5                      \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,0,1,1,xmm0)             //  movd      (%0,%1,1),%%xmm0
-    MEMOPREG(movd,0x00,0,5,1,xmm6)             //  movd      (%0,%5,1),%%xmm6
-    "punpckldq %%xmm6,%%xmm0                   \n"
-    "addps     %%xmm4,%%xmm3                   \n"
-    "sub       $0x4,%4                         \n"
-    "movq      %%xmm0," MEMACCESS2(0x08,2) "   \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jge       40b                             \n"
-
-  "49:                                         \n"
-    "add       $0x3,%4                         \n"
-    "jl        19f                             \n"
-
-  // 1 pixel loop                              \n"
-    LABELALIGN
-  "10:                                         \n"
-    "cvttps2dq %%xmm2,%%xmm0                   \n"
-    "packssdw  %%xmm0,%%xmm0                   \n"
-    "pmaddwd   %%xmm5,%%xmm0                   \n"
-    "addps     %%xmm7,%%xmm2                   \n"
-    "movd      %%xmm0,%k1                      \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,0,1,1,xmm0)             //  movd      (%0,%1,1),%%xmm0
-    "sub       $0x1,%4                         \n"
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x04,2) ",%2           \n"
-    "jge       10b                             \n"
-  "19:                                         \n"
-  : "+r"(src_argb),  // %0
-    "+r"(src_argb_stride_temp),  // %1
-    "+r"(dst_argb),  // %2
-    "+r"(src_dudv),  // %3
-    "+rm"(width),    // %4
-    "+r"(temp)   // %5
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBAFFINEROW_SSE2
-
-#ifdef HAS_INTERPOLATEROW_SSSE3
-// Bilinear filter 16x2 -> 16x1
-void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                          ptrdiff_t src_stride, int dst_width,
-                          int source_y_fraction) {
-  asm volatile (
-    "sub       %1,%0                           \n"
-    "shr       %3                              \n"
-    "cmp       $0x0,%3                         \n"
-    "je        100f                            \n"
-    "cmp       $0x20,%3                        \n"
-    "je        75f                             \n"
-    "cmp       $0x40,%3                        \n"
-    "je        50f                             \n"
-    "cmp       $0x60,%3                        \n"
-    "je        25f                             \n"
-
-    "movd      %3,%%xmm0                       \n"
-    "neg       %3                              \n"
-    "add       $0x80,%3                        \n"
-    "movd      %3,%%xmm5                       \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "punpcklwd %%xmm5,%%xmm5                   \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-
-    // General purpose row blend.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm2)
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm2,%%xmm0                   \n"
-    "punpckhbw %%xmm2,%%xmm1                   \n"
-    "pmaddubsw %%xmm5,%%xmm0                   \n"
-    "pmaddubsw %%xmm5,%%xmm1                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-    "jmp       99f                             \n"
-
-    // Blend 25 / 75.
-    LABELALIGN
-  "25:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm1)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        25b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 50 / 50.
-    LABELALIGN
-  "50:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm1)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        50b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 75 / 25.
-    LABELALIGN
-  "75:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm1         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm0)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        75b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-    LABELALIGN
-  "100:                                        \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    "sub       $0x10,%2                        \n"
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        100b                            \n"
-
-  "99:                                         \n"
-  : "+r"(dst_ptr),    // %0
-    "+r"(src_ptr),    // %1
-    "+r"(dst_width),  // %2
-    "+r"(source_y_fraction)  // %3
-  : "r"((intptr_t)(src_stride))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_INTERPOLATEROW_SSSE3
-
-#ifdef HAS_INTERPOLATEROW_SSE2
-// Bilinear filter 16x2 -> 16x1
-void InterpolateRow_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride, int dst_width,
-                         int source_y_fraction) {
-  asm volatile (
-    "sub       %1,%0                           \n"
-    "shr       %3                              \n"
-    "cmp       $0x0,%3                         \n"
-    "je        100f                            \n"
-    "cmp       $0x20,%3                        \n"
-    "je        75f                             \n"
-    "cmp       $0x40,%3                        \n"
-    "je        50f                             \n"
-    "cmp       $0x60,%3                        \n"
-    "je        25f                             \n"
-
-    "movd      %3,%%xmm0                       \n"
-    "neg       %3                              \n"
-    "add       $0x80,%3                        \n"
-    "movd      %3,%%xmm5                       \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "punpcklwd %%xmm5,%%xmm5                   \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-
-    // General purpose row blend.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm2)           //  movdqa    (%1,%4,1),%%xmm2
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "punpcklbw %%xmm4,%%xmm2                   \n"
-    "punpckhbw %%xmm4,%%xmm3                   \n"
-    "punpcklbw %%xmm4,%%xmm0                   \n"
-    "punpckhbw %%xmm4,%%xmm1                   \n"
-    "psubw     %%xmm0,%%xmm2                   \n"
-    "psubw     %%xmm1,%%xmm3                   \n"
-    "paddw     %%xmm2,%%xmm2                   \n"
-    "paddw     %%xmm3,%%xmm3                   \n"
-    "pmulhw    %%xmm5,%%xmm2                   \n"
-    "pmulhw    %%xmm5,%%xmm3                   \n"
-    "paddw     %%xmm2,%%xmm0                   \n"
-    "paddw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)           //  movdqa    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-    "jmp       99f                             \n"
-
-    // Blend 25 / 75.
-    LABELALIGN
-  "25:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm1)           //  movdqa    (%1,%4,1),%%xmm1
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)           //  movdqa    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        25b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 50 / 50.
-    LABELALIGN
-  "50:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm1)           //  movdqa    (%1,%4,1),%%xmm1
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)           //  movdqa    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        50b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 75 / 25.
-    LABELALIGN
-  "75:                                         \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm1         \n"
-    MEMOPREG(movdqa,0x00,1,4,1,xmm0)           //  movdqa    (%1,%4,1),%%xmm0
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)           //  movdqa    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        75b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-    LABELALIGN
-  "100:                                        \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    "sub       $0x10,%2                        \n"
-    MEMOPMEM(movdqa,xmm0,0x00,1,0,1)           //  movdqa    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        100b                            \n"
-
-  "99:                                         \n"
-  : "+r"(dst_ptr),    // %0
-    "+r"(src_ptr),    // %1
-    "+r"(dst_width),  // %2
-    "+r"(source_y_fraction)  // %3
-  : "r"((intptr_t)(src_stride))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_INTERPOLATEROW_SSE2
-
-#ifdef HAS_INTERPOLATEROW_SSSE3
-// Bilinear filter 16x2 -> 16x1
-void InterpolateRow_Unaligned_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                                    ptrdiff_t src_stride, int dst_width,
-                                    int source_y_fraction) {
-  asm volatile (
-    "sub       %1,%0                           \n"
-    "shr       %3                              \n"
-    "cmp       $0x0,%3                         \n"
-    "je        100f                            \n"
-    "cmp       $0x20,%3                        \n"
-    "je        75f                             \n"
-    "cmp       $0x40,%3                        \n"
-    "je        50f                             \n"
-    "cmp       $0x60,%3                        \n"
-    "je        25f                             \n"
-
-    "movd      %3,%%xmm0                       \n"
-    "neg       %3                              \n"
-    "add       $0x80,%3                        \n"
-    "movd      %3,%%xmm5                       \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "punpcklwd %%xmm5,%%xmm5                   \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-
-    // General purpose row blend.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm2)
-    "movdqu    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm2,%%xmm0                   \n"
-    "punpckhbw %%xmm2,%%xmm1                   \n"
-    "pmaddubsw %%xmm5,%%xmm0                   \n"
-    "pmaddubsw %%xmm5,%%xmm1                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "psrlw     $0x7,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-    "jmp       99f                             \n"
-
-    // Blend 25 / 75.
-    LABELALIGN
-  "25:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm1)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        25b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 50 / 50.
-    LABELALIGN
-  "50:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm1)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        50b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 75 / 25.
-    LABELALIGN
-  "75:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm0)
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        75b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-    LABELALIGN
-  "100:                                        \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    "sub       $0x10,%2                        \n"
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        100b                            \n"
-
-  "99:                                         \n"
-  : "+r"(dst_ptr),    // %0
-    "+r"(src_ptr),    // %1
-    "+r"(dst_width),  // %2
-    "+r"(source_y_fraction)  // %3
-  : "r"((intptr_t)(src_stride))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm5"
-#endif
-  );
-}
-#endif   // HAS_INTERPOLATEROW_SSSE3
-
-#ifdef HAS_INTERPOLATEROW_SSE2
-// Bilinear filter 16x2 -> 16x1
-void InterpolateRow_Unaligned_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                                   ptrdiff_t src_stride, int dst_width,
-                                   int source_y_fraction) {
-  asm volatile (
-    "sub       %1,%0                           \n"
-    "shr       %3                              \n"
-    "cmp       $0x0,%3                         \n"
-    "je        100f                            \n"
-    "cmp       $0x20,%3                        \n"
-    "je        75f                             \n"
-    "cmp       $0x40,%3                        \n"
-    "je        50f                             \n"
-    "cmp       $0x60,%3                        \n"
-    "je        25f                             \n"
-
-    "movd      %3,%%xmm0                       \n"
-    "neg       %3                              \n"
-    "add       $0x80,%3                        \n"
-    "movd      %3,%%xmm5                       \n"
-    "punpcklbw %%xmm0,%%xmm5                   \n"
-    "punpcklwd %%xmm5,%%xmm5                   \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    "pxor      %%xmm4,%%xmm4                   \n"
-
-    // General purpose row blend.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm2)           //  movdqu    (%1,%4,1),%%xmm2
-    "movdqu    %%xmm0,%%xmm1                   \n"
-    "movdqu    %%xmm2,%%xmm3                   \n"
-    "punpcklbw %%xmm4,%%xmm2                   \n"
-    "punpckhbw %%xmm4,%%xmm3                   \n"
-    "punpcklbw %%xmm4,%%xmm0                   \n"
-    "punpckhbw %%xmm4,%%xmm1                   \n"
-    "psubw     %%xmm0,%%xmm2                   \n"
-    "psubw     %%xmm1,%%xmm3                   \n"
-    "paddw     %%xmm2,%%xmm2                   \n"
-    "paddw     %%xmm3,%%xmm3                   \n"
-    "pmulhw    %%xmm5,%%xmm2                   \n"
-    "pmulhw    %%xmm5,%%xmm3                   \n"
-    "paddw     %%xmm2,%%xmm0                   \n"
-    "paddw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)           //  movdqu    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-    "jmp       99f                             \n"
-
-    // Blend 25 / 75.
-    LABELALIGN
-  "25:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm1)           //  movdqu    (%1,%4,1),%%xmm1
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)           //  movdqu    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        25b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 50 / 50.
-    LABELALIGN
-  "50:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm1)           //  movdqu    (%1,%4,1),%%xmm1
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)           //  movdqu    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        50b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 75 / 25.
-    LABELALIGN
-  "75:                                         \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm1         \n"
-    MEMOPREG(movdqu,0x00,1,4,1,xmm0)           //  movdqu    (%1,%4,1),%%xmm0
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "pavgb     %%xmm1,%%xmm0                   \n"
-    "sub       $0x10,%2                        \n"
-    BUNDLEALIGN
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)           //  movdqu    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        75b                             \n"
-    "jmp       99f                             \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-    LABELALIGN
-  "100:                                        \n"
-    "movdqu    " MEMACCESS(1) ",%%xmm0         \n"
-    "sub       $0x10,%2                        \n"
-    MEMOPMEM(movdqu,xmm0,0x00,1,0,1)           //  movdqu    %%xmm0,(%1,%0,1)
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        100b                            \n"
-
-  "99:                                         \n"
-  : "+r"(dst_ptr),    // %0
-    "+r"(src_ptr),    // %1
-    "+r"(dst_width),  // %2
-    "+r"(source_y_fraction)  // %3
-  : "r"((intptr_t)(src_stride))  // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_INTERPOLATEROW_SSE2
-
-#ifdef HAS_HALFROW_SSE2
-void HalfRow_SSE2(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix) {
-  asm volatile (
-    "sub       %0,%1                           \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(pavgb,0x00,0,3,1,xmm0)            //  pavgb     (%0,%3),%%xmm0
-    "sub       $0x10,%2                        \n"
-    MEMOPMEM(movdqa,xmm0,0x00,0,1,1)           //  movdqa    %%xmm0,(%0,%1)
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "jg        1b                              \n"
-  : "+r"(src_uv),  // %0
-    "+r"(dst_uv),  // %1
-    "+r"(pix)      // %2
-  : "r"((intptr_t)(src_uv_stride))  // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-      , "xmm0"
-#endif
-  );
-}
-#endif  // HAS_HALFROW_SSE2
-
-#ifdef HAS_ARGBTOBAYERROW_SSSE3
-void ARGBToBayerRow_SSSE3(const uint8* src_argb, uint8* dst_bayer,
-                          uint32 selector, int pix) {
-  asm volatile (
-    // NaCL caveat - assumes movd is from GPR
-    "movd      %3,%%xmm5                       \n"
-    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "punpckldq %%xmm1,%%xmm0                   \n"
-    "sub       $0x8,%2                         \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_bayer), // %1
-    "+r"(pix)        // %2
-  : "g"(selector)    // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBTOBAYERROW_SSSE3
-
-#ifdef HAS_ARGBTOBAYERGGROW_SSE2
-void ARGBToBayerGGRow_SSE2(const uint8* src_argb, uint8* dst_bayer,
-                           uint32 selector, int pix) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrld     $0x18,%%xmm5                    \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrld     $0x8,%%xmm0                     \n"
-    "psrld     $0x8,%%xmm1                     \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packssdw  %%xmm1,%%xmm0                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x8,%2                         \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_bayer), // %1
-    "+r"(pix)        // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBTOBAYERGGROW_SSE2
-
-#ifdef HAS_ARGBSHUFFLEROW_SSSE3
-// For BGRAToARGB, ABGRToARGB, RGBAToARGB, and ARGBToRGBA.
-void ARGBShuffleRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                          const uint8* shuffler, int pix) {
-  asm volatile (
-    "movdqa    " MEMACCESS(3) ",%%xmm5         \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "sub       $0x8,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "r"(shuffler)    // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void ARGBShuffleRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                    const uint8* shuffler, int pix) {
-  asm volatile (
-    "movdqa    " MEMACCESS(3) ",%%xmm5         \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pshufb    %%xmm5,%%xmm0                   \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "sub       $0x8,%2                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "r"(shuffler)    // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBSHUFFLEROW_SSSE3
-
-#ifdef HAS_ARGBSHUFFLEROW_AVX2
-// For BGRAToARGB, ABGRToARGB, RGBAToARGB, and ARGBToRGBA.
-void ARGBShuffleRow_AVX2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix) {
-  asm volatile (
-    "vbroadcastf128 " MEMACCESS(3) ",%%ymm5    \n"
-    LABELALIGN
-  "1:                                          \n"
-    "vmovdqu   " MEMACCESS(0) ",%%ymm0         \n"
-    "vmovdqu   " MEMACCESS2(0x20,0) ",%%ymm1   \n"
-    "lea       " MEMLEA(0x40,0) ",%0           \n"
-    "vpshufb   %%ymm5,%%ymm0,%%ymm0            \n"
-    "vpshufb   %%ymm5,%%ymm1,%%ymm1            \n"
-    "sub       $0x10,%2                        \n"
-    "vmovdqu   %%ymm0," MEMACCESS(1) "         \n"
-    "vmovdqu   %%ymm1," MEMACCESS2(0x20,1) "   \n"
-    "lea       " MEMLEA(0x40,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(pix)        // %2
-  : "r"(shuffler)    // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBSHUFFLEROW_AVX2
-
-#ifdef HAS_ARGBSHUFFLEROW_SSE2
-// For BGRAToARGB, ABGRToARGB, RGBAToARGB, and ARGBToRGBA.
-void ARGBShuffleRow_SSE2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix) {
-  uintptr_t pixel_temp = 0u;
-  asm volatile (
-    "pxor      %%xmm5,%%xmm5                   \n"
-    "mov       " MEMACCESS(4) ",%k2            \n"
-    "cmp       $0x3000102,%k2                  \n"
-    "je        3012f                           \n"
-    "cmp       $0x10203,%k2                    \n"
-    "je        123f                            \n"
-    "cmp       $0x30201,%k2                    \n"
-    "je        321f                            \n"
-    "cmp       $0x2010003,%k2                  \n"
-    "je        2103f                           \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movzb     " MEMACCESS(4) ",%2             \n"
-    MEMOPARG(movzb,0x00,0,2,1,2) "             \n"  //  movzb     (%0,%2,1),%2
-    "mov       %b2," MEMACCESS(1) "            \n"
-    "movzb     " MEMACCESS2(0x1,4) ",%2        \n"
-    MEMOPARG(movzb,0x00,0,2,1,2) "             \n"  //  movzb     (%0,%2,1),%2
-    "mov       %b2," MEMACCESS2(0x1,1) "       \n"
-    BUNDLEALIGN
-    "movzb     " MEMACCESS2(0x2,4) ",%2        \n"
-    MEMOPARG(movzb,0x00,0,2,1,2) "             \n"  //  movzb     (%0,%2,1),%2
-    "mov       %b2," MEMACCESS2(0x2,1) "       \n"
-    "movzb     " MEMACCESS2(0x3,4) ",%2        \n"
-    MEMOPARG(movzb,0x00,0,2,1,2) "             \n"  //  movzb     (%0,%2,1),%2
-    "mov       %b2," MEMACCESS2(0x3,1) "       \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    "lea       " MEMLEA(0x4,1) ",%1            \n"
-    "sub       $0x1,%3                         \n"
-    "jg        1b                              \n"
-    "jmp       99f                             \n"
-
-    LABELALIGN
-  "123:                                        \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpckhbw %%xmm5,%%xmm1                   \n"
-    "pshufhw   $0x1b,%%xmm0,%%xmm0             \n"
-    "pshuflw   $0x1b,%%xmm0,%%xmm0             \n"
-    "pshufhw   $0x1b,%%xmm1,%%xmm1             \n"
-    "pshuflw   $0x1b,%%xmm1,%%xmm1             \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        123b                            \n"
-    "jmp       99f                             \n"
-
-    LABELALIGN
-  "321:                                        \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpckhbw %%xmm5,%%xmm1                   \n"
-    "pshufhw   $0x39,%%xmm0,%%xmm0             \n"
-    "pshuflw   $0x39,%%xmm0,%%xmm0             \n"
-    "pshufhw   $0x39,%%xmm1,%%xmm1             \n"
-    "pshuflw   $0x39,%%xmm1,%%xmm1             \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        321b                            \n"
-    "jmp       99f                             \n"
-
-    LABELALIGN
-  "2103:                                       \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpckhbw %%xmm5,%%xmm1                   \n"
-    "pshufhw   $0x93,%%xmm0,%%xmm0             \n"
-    "pshuflw   $0x93,%%xmm0,%%xmm0             \n"
-    "pshufhw   $0x93,%%xmm1,%%xmm1             \n"
-    "pshuflw   $0x93,%%xmm1,%%xmm1             \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        2103b                           \n"
-    "jmp       99f                             \n"
-
-    LABELALIGN
-  "3012:                                       \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpckhbw %%xmm5,%%xmm1                   \n"
-    "pshufhw   $0xc6,%%xmm0,%%xmm0             \n"
-    "pshuflw   $0xc6,%%xmm0,%%xmm0             \n"
-    "pshufhw   $0xc6,%%xmm1,%%xmm1             \n"
-    "pshuflw   $0xc6,%%xmm1,%%xmm1             \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        3012b                           \n"
-
-  "99:                                         \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+d"(pixel_temp),  // %2
-    "+r"(pix)         // %3
-  : "r"(shuffler)      // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBSHUFFLEROW_SSE2
-
-#ifdef HAS_I422TOYUY2ROW_SSE2
-void I422ToYUY2Row_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_frame, int width) {
- asm volatile (
-    "sub       %1,%2                             \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movq      " MEMACCESS(1) ",%%xmm2           \n"
-    MEMOPREG(movq,0x00,1,2,1,xmm3)               //  movq    (%1,%2,1),%%xmm3
-    "lea       " MEMLEA(0x8,1) ",%1              \n"
-    "punpcklbw %%xmm3,%%xmm2                     \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0           \n"
-    "lea       " MEMLEA(0x10,0) ",%0             \n"
-    "movdqa    %%xmm0,%%xmm1                     \n"
-    "punpcklbw %%xmm2,%%xmm0                     \n"
-    "punpckhbw %%xmm2,%%xmm1                     \n"
-    "movdqu    %%xmm0," MEMACCESS(3) "           \n"
-    "movdqu    %%xmm1," MEMACCESS2(0x10,3) "     \n"
-    "lea       " MEMLEA(0x20,3) ",%3             \n"
-    "sub       $0x10,%4                          \n"
-    "jg         1b                               \n"
-    : "+r"(src_y),  // %0
-      "+r"(src_u),  // %1
-      "+r"(src_v),  // %2
-      "+r"(dst_frame),  // %3
-      "+rm"(width)  // %4
-    :
-    : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3"
-#endif
-  );
-}
-#endif  // HAS_I422TOYUY2ROW_SSE2
-
-#ifdef HAS_I422TOUYVYROW_SSE2
-void I422ToUYVYRow_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_frame, int width) {
- asm volatile (
-    "sub        %1,%2                            \n"
-    LABELALIGN
-  "1:                                            \n"
-    "movq      " MEMACCESS(1) ",%%xmm2           \n"
-    MEMOPREG(movq,0x00,1,2,1,xmm3)               //  movq    (%1,%2,1),%%xmm3
-    "lea       " MEMLEA(0x8,1) ",%1              \n"
-    "punpcklbw %%xmm3,%%xmm2                     \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0           \n"
-    "movdqa    %%xmm2,%%xmm1                     \n"
-    "lea       " MEMLEA(0x10,0) ",%0             \n"
-    "punpcklbw %%xmm0,%%xmm1                     \n"
-    "punpckhbw %%xmm0,%%xmm2                     \n"
-    "movdqu    %%xmm1," MEMACCESS(3) "           \n"
-    "movdqu    %%xmm2," MEMACCESS2(0x10,3) "     \n"
-    "lea       " MEMLEA(0x20,3) ",%3             \n"
-    "sub       $0x10,%4                          \n"
-    "jg         1b                               \n"
-    : "+r"(src_y),  // %0
-      "+r"(src_u),  // %1
-      "+r"(src_v),  // %2
-      "+r"(dst_frame),  // %3
-      "+rm"(width)  // %4
-    :
-    : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3"
-#endif
-  );
-}
-#endif  // HAS_I422TOUYVYROW_SSE2
-
-#ifdef HAS_ARGBPOLYNOMIALROW_SSE2
-void ARGBPolynomialRow_SSE2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width) {
-  asm volatile (
-    "pxor      %%xmm3,%%xmm3                   \n"
-
-    // 2 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "punpcklbw %%xmm3,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm4                   \n"
-    "punpcklwd %%xmm3,%%xmm0                   \n"
-    "punpckhwd %%xmm3,%%xmm4                   \n"
-    "cvtdq2ps  %%xmm0,%%xmm0                   \n"
-    "cvtdq2ps  %%xmm4,%%xmm4                   \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "movdqa    %%xmm4,%%xmm5                   \n"
-    "mulps     " MEMACCESS2(0x10,3) ",%%xmm0   \n"
-    "mulps     " MEMACCESS2(0x10,3) ",%%xmm4   \n"
-    "addps     " MEMACCESS(3) ",%%xmm0         \n"
-    "addps     " MEMACCESS(3) ",%%xmm4         \n"
-    "movdqa    %%xmm1,%%xmm2                   \n"
-    "movdqa    %%xmm5,%%xmm6                   \n"
-    "mulps     %%xmm1,%%xmm2                   \n"
-    "mulps     %%xmm5,%%xmm6                   \n"
-    "mulps     %%xmm2,%%xmm1                   \n"
-    "mulps     %%xmm6,%%xmm5                   \n"
-    "mulps     " MEMACCESS2(0x20,3) ",%%xmm2   \n"
-    "mulps     " MEMACCESS2(0x20,3) ",%%xmm6   \n"
-    "mulps     " MEMACCESS2(0x30,3) ",%%xmm1   \n"
-    "mulps     " MEMACCESS2(0x30,3) ",%%xmm5   \n"
-    "addps     %%xmm2,%%xmm0                   \n"
-    "addps     %%xmm6,%%xmm4                   \n"
-    "addps     %%xmm1,%%xmm0                   \n"
-    "addps     %%xmm5,%%xmm4                   \n"
-    "cvttps2dq %%xmm0,%%xmm0                   \n"
-    "cvttps2dq %%xmm4,%%xmm4                   \n"
-    "packuswb  %%xmm4,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "sub       $0x2,%2                         \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(width)      // %2
-  : "r"(poly)        // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-#endif  // HAS_ARGBPOLYNOMIALROW_SSE2
-
-#ifdef HAS_ARGBPOLYNOMIALROW_AVX2
-void ARGBPolynomialRow_AVX2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width) {
-  asm volatile (
-    "vbroadcastf128 " MEMACCESS(3) ",%%ymm4     \n"
-    "vbroadcastf128 " MEMACCESS2(0x10,3) ",%%ymm5 \n"
-    "vbroadcastf128 " MEMACCESS2(0x20,3) ",%%ymm6 \n"
-    "vbroadcastf128 " MEMACCESS2(0x30,3) ",%%ymm7 \n"
-
-    // 2 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "vpmovzxbd   " MEMACCESS(0) ",%%ymm0       \n"  // 2 ARGB pixels
-    "lea         " MEMLEA(0x8,0) ",%0          \n"
-    "vcvtdq2ps   %%ymm0,%%ymm0                 \n"  // X 8 floats
-    "vmulps      %%ymm0,%%ymm0,%%ymm2          \n"  // X * X
-    "vmulps      %%ymm7,%%ymm0,%%ymm3          \n"  // C3 * X
-    "vfmadd132ps %%ymm5,%%ymm4,%%ymm0          \n"  // result = C0 + C1 * X
-    "vfmadd231ps %%ymm6,%%ymm2,%%ymm0          \n"  // result += C2 * X * X
-    "vfmadd231ps %%ymm3,%%ymm2,%%ymm0          \n"  // result += C3 * X * X * X
-    "vcvttps2dq  %%ymm0,%%ymm0                 \n"
-    "vpackusdw   %%ymm0,%%ymm0,%%ymm0          \n"
-    "vpermq      $0xd8,%%ymm0,%%ymm0           \n"
-    "vpackuswb   %%xmm0,%%xmm0,%%xmm0          \n"
-    "sub         $0x2,%2                       \n"
-    "vmovq       %%xmm0," MEMACCESS(1) "       \n"
-    "lea         " MEMLEA(0x8,1) ",%1          \n"
-    "jg          1b                            \n"
-    "vzeroupper                                \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(width)      // %2
-  : "r"(poly)        // %3
-  : "memory", "cc"
-#if defined(__SSE2__)
-// TODO(fbarchard): declare ymm usage when applicable.
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-#endif  // HAS_ARGBPOLYNOMIALROW_AVX2
-
-#ifdef HAS_ARGBCOLORTABLEROW_X86
-// Tranform ARGB pixels with color table.
-void ARGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb,
-                           int width) {
-  uintptr_t pixel_temp = 0u;
-  asm volatile (
-    // 1 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movzb     " MEMACCESS(0) ",%1             \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    MEMOPARG(movzb,0x00,3,1,4,1) "             \n"  // movzb (%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x4,0) "      \n"
-    "movzb     " MEMACCESS2(-0x3,0) ",%1       \n"
-    MEMOPARG(movzb,0x01,3,1,4,1) "             \n"  // movzb 0x1(%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x3,0) "      \n"
-    "movzb     " MEMACCESS2(-0x2,0) ",%1       \n"
-    MEMOPARG(movzb,0x02,3,1,4,1) "             \n"  // movzb 0x2(%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x2,0) "      \n"
-    "movzb     " MEMACCESS2(-0x1,0) ",%1       \n"
-    MEMOPARG(movzb,0x03,3,1,4,1) "             \n"  // movzb 0x3(%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x1,0) "      \n"
-    "dec       %2                              \n"
-    "jg        1b                              \n"
-  : "+r"(dst_argb),   // %0
-    "+d"(pixel_temp), // %1
-    "+r"(width)       // %2
-  : "r"(table_argb)   // %3
-  : "memory", "cc");
-}
-#endif  // HAS_ARGBCOLORTABLEROW_X86
-
-#ifdef HAS_RGBCOLORTABLEROW_X86
-// Tranform RGB pixels with color table.
-void RGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb, int width) {
-  uintptr_t pixel_temp = 0u;
-  asm volatile (
-    // 1 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movzb     " MEMACCESS(0) ",%1             \n"
-    "lea       " MEMLEA(0x4,0) ",%0            \n"
-    MEMOPARG(movzb,0x00,3,1,4,1) "             \n"  // movzb (%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x4,0) "      \n"
-    "movzb     " MEMACCESS2(-0x3,0) ",%1       \n"
-    MEMOPARG(movzb,0x01,3,1,4,1) "             \n"  // movzb 0x1(%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x3,0) "      \n"
-    "movzb     " MEMACCESS2(-0x2,0) ",%1       \n"
-    MEMOPARG(movzb,0x02,3,1,4,1) "             \n"  // movzb 0x2(%3,%1,4),%1
-    "mov       %b1," MEMACCESS2(-0x2,0) "      \n"
-    "dec       %2                              \n"
-    "jg        1b                              \n"
-  : "+r"(dst_argb),   // %0
-    "+d"(pixel_temp), // %1
-    "+r"(width)       // %2
-  : "r"(table_argb)   // %3
-  : "memory", "cc");
-}
-#endif  // HAS_RGBCOLORTABLEROW_X86
-
-#ifdef HAS_ARGBLUMACOLORTABLEROW_SSSE3
-// Tranform RGB pixels with luma table.
-void ARGBLumaColorTableRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                 int width,
-                                 const uint8* luma, uint32 lumacoeff) {
-  uintptr_t pixel_temp = 0u;
-  uintptr_t table_temp = 0u;
-  asm volatile (
-    "movd      %6,%%xmm3                       \n"
-    "pshufd    $0x0,%%xmm3,%%xmm3              \n"
-    "pcmpeqb   %%xmm4,%%xmm4                   \n"
-    "psllw     $0x8,%%xmm4                     \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-
-    // 4 pixel loop.
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(2) ",%%xmm0         \n"
-    "pmaddubsw %%xmm3,%%xmm0                   \n"
-    "phaddw    %%xmm0,%%xmm0                   \n"
-    "pand      %%xmm4,%%xmm0                   \n"
-    "punpcklwd %%xmm5,%%xmm0                   \n"
-    "movd      %%xmm0,%k1                      \n"  // 32 bit offset
-    "add       %5,%1                           \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-
-    "movzb     " MEMACCESS(2) ",%0             \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS(3) "            \n"
-    "movzb     " MEMACCESS2(0x1,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x1,3) "       \n"
-    "movzb     " MEMACCESS2(0x2,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x2,3) "       \n"
-    "movzb     " MEMACCESS2(0x3,2) ",%0        \n"
-    "mov       %b0," MEMACCESS2(0x3,3) "       \n"
-
-    "movd      %%xmm0,%k1                      \n"  // 32 bit offset
-    "add       %5,%1                           \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-
-    "movzb     " MEMACCESS2(0x4,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x4,3) "       \n"
-    BUNDLEALIGN
-    "movzb     " MEMACCESS2(0x5,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x5,3) "       \n"
-    "movzb     " MEMACCESS2(0x6,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x6,3) "       \n"
-    "movzb     " MEMACCESS2(0x7,2) ",%0        \n"
-    "mov       %b0," MEMACCESS2(0x7,3) "       \n"
-
-    "movd      %%xmm0,%k1                      \n"  // 32 bit offset
-    "add       %5,%1                           \n"
-    "pshufd    $0x39,%%xmm0,%%xmm0             \n"
-
-    "movzb     " MEMACCESS2(0x8,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x8,3) "       \n"
-    "movzb     " MEMACCESS2(0x9,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0x9,3) "       \n"
-    "movzb     " MEMACCESS2(0xa,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0xa,3) "       \n"
-    "movzb     " MEMACCESS2(0xb,2) ",%0        \n"
-    "mov       %b0," MEMACCESS2(0xb,3) "       \n"
-
-    "movd      %%xmm0,%k1                      \n"  // 32 bit offset
-    "add       %5,%1                           \n"
-
-    "movzb     " MEMACCESS2(0xc,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0xc,3) "       \n"
-    "movzb     " MEMACCESS2(0xd,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0xd,3) "       \n"
-    "movzb     " MEMACCESS2(0xe,2) ",%0        \n"
-    MEMOPARG(movzb,0x00,1,0,1,0) "             \n"  // movzb     (%1,%0,1),%0
-    "mov       %b0," MEMACCESS2(0xe,3) "       \n"
-    "movzb     " MEMACCESS2(0xf,2) ",%0        \n"
-    "mov       %b0," MEMACCESS2(0xf,3) "       \n"
-    "sub       $0x4,%4                         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "lea       " MEMLEA(0x10,3) ",%3           \n"
-    "jg        1b                              \n"
-  : "+d"(pixel_temp),  // %0
-    "+a"(table_temp),  // %1
-    "+r"(src_argb),    // %2
-    "+r"(dst_argb),    // %3
-    "+rm"(width)       // %4
-  : "r"(luma),         // %5
-    "rm"(lumacoeff)    // %6
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-#endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
-
-#endif  // defined(__x86_64__) || defined(__i386__)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_win.cc b/drivers/theoraplayer/src/YUV/libyuv/src/row_win.cc
deleted file mode 100755
index f13e4d7ae5..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_win.cc
+++ /dev/null
@@ -1,7284 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for Visual C x86.
-#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-#ifdef HAS_ARGBTOYROW_SSSE3
-
-// Constants for ARGB.
-static const vec8 kARGBToY = {
-  13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0
-};
-
-// JPeg full range.
-static const vec8 kARGBToYJ = {
-  15, 75, 38, 0, 15, 75, 38, 0, 15, 75, 38, 0, 15, 75, 38, 0
-};
-
-static const vec8 kARGBToU = {
-  112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0
-};
-
-static const vec8 kARGBToUJ = {
-  127, -84, -43, 0, 127, -84, -43, 0, 127, -84, -43, 0, 127, -84, -43, 0
-};
-
-static const vec8 kARGBToV = {
-  -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0,
-};
-
-static const vec8 kARGBToVJ = {
-  -20, -107, 127, 0, -20, -107, 127, 0, -20, -107, 127, 0, -20, -107, 127, 0
-};
-
-// vpermd for vphaddw + vpackuswb vpermd.
-static const lvec32 kPermdARGBToY_AVX = {
-  0, 4, 1, 5, 2, 6, 3, 7
-};
-
-// vpshufb for vphaddw + vpackuswb packed to shorts.
-static const lvec8 kShufARGBToUV_AVX = {
-  0, 1, 8, 9, 2, 3, 10, 11, 4, 5, 12, 13, 6, 7, 14, 15,
-  0, 1, 8, 9, 2, 3, 10, 11, 4, 5, 12, 13, 6, 7, 14, 15,
-};
-
-// Constants for BGRA.
-static const vec8 kBGRAToY = {
-  0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13
-};
-
-static const vec8 kBGRAToU = {
-  0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112
-};
-
-static const vec8 kBGRAToV = {
-  0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18
-};
-
-// Constants for ABGR.
-static const vec8 kABGRToY = {
-  33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0, 33, 65, 13, 0
-};
-
-static const vec8 kABGRToU = {
-  -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0, -38, -74, 112, 0
-};
-
-static const vec8 kABGRToV = {
-  112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0, 112, -94, -18, 0
-};
-
-// Constants for RGBA.
-static const vec8 kRGBAToY = {
-  0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33, 0, 13, 65, 33
-};
-
-static const vec8 kRGBAToU = {
-  0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38, 0, 112, -74, -38
-};
-
-static const vec8 kRGBAToV = {
-  0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112, 0, -18, -94, 112
-};
-
-static const uvec8 kAddY16 = {
-  16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u, 16u
-};
-
-static const vec16 kAddYJ64 = {
-  64, 64, 64, 64, 64, 64, 64, 64
-};
-
-static const uvec8 kAddUV128 = {
-  128u, 128u, 128u, 128u, 128u, 128u, 128u, 128u,
-  128u, 128u, 128u, 128u, 128u, 128u, 128u, 128u
-};
-
-static const uvec16 kAddUVJ128 = {
-  0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u, 0x8080u
-};
-
-// Shuffle table for converting RGB24 to ARGB.
-static const uvec8 kShuffleMaskRGB24ToARGB = {
-  0u, 1u, 2u, 12u, 3u, 4u, 5u, 13u, 6u, 7u, 8u, 14u, 9u, 10u, 11u, 15u
-};
-
-// Shuffle table for converting RAW to ARGB.
-static const uvec8 kShuffleMaskRAWToARGB = {
-  2u, 1u, 0u, 12u, 5u, 4u, 3u, 13u, 8u, 7u, 6u, 14u, 11u, 10u, 9u, 15u
-};
-
-// Shuffle table for converting ARGB to RGB24.
-static const uvec8 kShuffleMaskARGBToRGB24 = {
-  0u, 1u, 2u, 4u, 5u, 6u, 8u, 9u, 10u, 12u, 13u, 14u, 128u, 128u, 128u, 128u
-};
-
-// Shuffle table for converting ARGB to RAW.
-static const uvec8 kShuffleMaskARGBToRAW = {
-  2u, 1u, 0u, 6u, 5u, 4u, 10u, 9u, 8u, 14u, 13u, 12u, 128u, 128u, 128u, 128u
-};
-
-// Shuffle table for converting ARGBToRGB24 for I422ToRGB24.  First 8 + next 4
-static const uvec8 kShuffleMaskARGBToRGB24_0 = {
-  0u, 1u, 2u, 4u, 5u, 6u, 8u, 9u, 128u, 128u, 128u, 128u, 10u, 12u, 13u, 14u
-};
-
-// Shuffle table for converting ARGB to RAW.
-static const uvec8 kShuffleMaskARGBToRAW_0 = {
-  2u, 1u, 0u, 6u, 5u, 4u, 10u, 9u, 128u, 128u, 128u, 128u, 8u, 14u, 13u, 12u
-};
-
-// Duplicates gray value 3 times and fills in alpha opaque.
-__declspec(naked) __declspec(align(16))
-void I400ToARGBRow_SSE2(const uint8* src_y, uint8* dst_argb, int pix) {
-  __asm {
-    mov        eax, [esp + 4]        // src_y
-    mov        edx, [esp + 8]        // dst_argb
-    mov        ecx, [esp + 12]       // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0xff000000
-    pslld      xmm5, 24
-
-    align      4
-  convertloop:
-    movq       xmm0, qword ptr [eax]
-    lea        eax,  [eax + 8]
-    punpcklbw  xmm0, xmm0
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm0
-    punpckhwd  xmm1, xmm1
-    por        xmm0, xmm5
-    por        xmm1, xmm5
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I400ToARGBRow_Unaligned_SSE2(const uint8* src_y, uint8* dst_argb,
-                                  int pix) {
-  __asm {
-    mov        eax, [esp + 4]        // src_y
-    mov        edx, [esp + 8]        // dst_argb
-    mov        ecx, [esp + 12]       // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0xff000000
-    pslld      xmm5, 24
-
-    align      4
-  convertloop:
-    movq       xmm0, qword ptr [eax]
-    lea        eax,  [eax + 8]
-    punpcklbw  xmm0, xmm0
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm0
-    punpckhwd  xmm1, xmm1
-    por        xmm0, xmm5
-    por        xmm1, xmm5
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RGB24ToARGBRow_SSSE3(const uint8* src_rgb24, uint8* dst_argb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_rgb24
-    mov       edx, [esp + 8]   // dst_argb
-    mov       ecx, [esp + 12]  // pix
-    pcmpeqb   xmm5, xmm5       // generate mask 0xff000000
-    pslld     xmm5, 24
-    movdqa    xmm4, kShuffleMaskRGB24ToARGB
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]
-    movdqu    xmm1, [eax + 16]
-    movdqu    xmm3, [eax + 32]
-    lea       eax, [eax + 48]
-    movdqa    xmm2, xmm3
-    palignr   xmm2, xmm1, 8    // xmm2 = { xmm3[0:3] xmm1[8:15]}
-    pshufb    xmm2, xmm4
-    por       xmm2, xmm5
-    palignr   xmm1, xmm0, 12   // xmm1 = { xmm3[0:7] xmm0[12:15]}
-    pshufb    xmm0, xmm4
-    movdqa    [edx + 32], xmm2
-    por       xmm0, xmm5
-    pshufb    xmm1, xmm4
-    movdqa    [edx], xmm0
-    por       xmm1, xmm5
-    palignr   xmm3, xmm3, 4    // xmm3 = { xmm3[4:15]}
-    pshufb    xmm3, xmm4
-    movdqa    [edx + 16], xmm1
-    por       xmm3, xmm5
-    sub       ecx, 16
-    movdqa    [edx + 48], xmm3
-    lea       edx, [edx + 64]
-    jg        convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RAWToARGBRow_SSSE3(const uint8* src_raw, uint8* dst_argb,
-                        int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_raw
-    mov       edx, [esp + 8]   // dst_argb
-    mov       ecx, [esp + 12]  // pix
-    pcmpeqb   xmm5, xmm5       // generate mask 0xff000000
-    pslld     xmm5, 24
-    movdqa    xmm4, kShuffleMaskRAWToARGB
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]
-    movdqu    xmm1, [eax + 16]
-    movdqu    xmm3, [eax + 32]
-    lea       eax, [eax + 48]
-    movdqa    xmm2, xmm3
-    palignr   xmm2, xmm1, 8    // xmm2 = { xmm3[0:3] xmm1[8:15]}
-    pshufb    xmm2, xmm4
-    por       xmm2, xmm5
-    palignr   xmm1, xmm0, 12   // xmm1 = { xmm3[0:7] xmm0[12:15]}
-    pshufb    xmm0, xmm4
-    movdqa    [edx + 32], xmm2
-    por       xmm0, xmm5
-    pshufb    xmm1, xmm4
-    movdqa    [edx], xmm0
-    por       xmm1, xmm5
-    palignr   xmm3, xmm3, 4    // xmm3 = { xmm3[4:15]}
-    pshufb    xmm3, xmm4
-    movdqa    [edx + 16], xmm1
-    por       xmm3, xmm5
-    sub       ecx, 16
-    movdqa    [edx + 48], xmm3
-    lea       edx, [edx + 64]
-    jg        convertloop
-    ret
-  }
-}
-
-// pmul method to replicate bits.
-// Math to replicate bits:
-// (v << 8) | (v << 3)
-// v * 256 + v * 8
-// v * (256 + 8)
-// G shift of 5 is incorporated, so shift is 5 + 8 and 5 + 3
-// 20 instructions.
-__declspec(naked) __declspec(align(16))
-void RGB565ToARGBRow_SSE2(const uint8* src_rgb565, uint8* dst_argb,
-                          int pix) {
-  __asm {
-    mov       eax, 0x01080108  // generate multiplier to repeat 5 bits
-    movd      xmm5, eax
-    pshufd    xmm5, xmm5, 0
-    mov       eax, 0x20802080  // multiplier shift by 5 and then repeat 6 bits
-    movd      xmm6, eax
-    pshufd    xmm6, xmm6, 0
-    pcmpeqb   xmm3, xmm3       // generate mask 0xf800f800 for Red
-    psllw     xmm3, 11
-    pcmpeqb   xmm4, xmm4       // generate mask 0x07e007e0 for Green
-    psllw     xmm4, 10
-    psrlw     xmm4, 5
-    pcmpeqb   xmm7, xmm7       // generate mask 0xff00ff00 for Alpha
-    psllw     xmm7, 8
-
-    mov       eax, [esp + 4]   // src_rgb565
-    mov       edx, [esp + 8]   // dst_argb
-    mov       ecx, [esp + 12]  // pix
-    sub       edx, eax
-    sub       edx, eax
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]   // fetch 8 pixels of bgr565
-    movdqa    xmm1, xmm0
-    movdqa    xmm2, xmm0
-    pand      xmm1, xmm3    // R in upper 5 bits
-    psllw     xmm2, 11      // B in upper 5 bits
-    pmulhuw   xmm1, xmm5    // * (256 + 8)
-    pmulhuw   xmm2, xmm5    // * (256 + 8)
-    psllw     xmm1, 8
-    por       xmm1, xmm2    // RB
-    pand      xmm0, xmm4    // G in middle 6 bits
-    pmulhuw   xmm0, xmm6    // << 5 * (256 + 4)
-    por       xmm0, xmm7    // AG
-    movdqa    xmm2, xmm1
-    punpcklbw xmm1, xmm0
-    punpckhbw xmm2, xmm0
-    movdqa    [eax * 2 + edx], xmm1  // store 4 pixels of ARGB
-    movdqa    [eax * 2 + edx + 16], xmm2  // store next 4 pixels of ARGB
-    lea       eax, [eax + 16]
-    sub       ecx, 8
-    jg        convertloop
-    ret
-  }
-}
-
-// 24 instructions
-__declspec(naked) __declspec(align(16))
-void ARGB1555ToARGBRow_SSE2(const uint8* src_argb1555, uint8* dst_argb,
-                            int pix) {
-  __asm {
-    mov       eax, 0x01080108  // generate multiplier to repeat 5 bits
-    movd      xmm5, eax
-    pshufd    xmm5, xmm5, 0
-    mov       eax, 0x42004200  // multiplier shift by 6 and then repeat 5 bits
-    movd      xmm6, eax
-    pshufd    xmm6, xmm6, 0
-    pcmpeqb   xmm3, xmm3       // generate mask 0xf800f800 for Red
-    psllw     xmm3, 11
-    movdqa    xmm4, xmm3       // generate mask 0x03e003e0 for Green
-    psrlw     xmm4, 6
-    pcmpeqb   xmm7, xmm7       // generate mask 0xff00ff00 for Alpha
-    psllw     xmm7, 8
-
-    mov       eax, [esp + 4]   // src_argb1555
-    mov       edx, [esp + 8]   // dst_argb
-    mov       ecx, [esp + 12]  // pix
-    sub       edx, eax
-    sub       edx, eax
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]   // fetch 8 pixels of 1555
-    movdqa    xmm1, xmm0
-    movdqa    xmm2, xmm0
-    psllw     xmm1, 1       // R in upper 5 bits
-    psllw     xmm2, 11      // B in upper 5 bits
-    pand      xmm1, xmm3
-    pmulhuw   xmm2, xmm5    // * (256 + 8)
-    pmulhuw   xmm1, xmm5    // * (256 + 8)
-    psllw     xmm1, 8
-    por       xmm1, xmm2    // RB
-    movdqa    xmm2, xmm0
-    pand      xmm0, xmm4    // G in middle 5 bits
-    psraw     xmm2, 8       // A
-    pmulhuw   xmm0, xmm6    // << 6 * (256 + 8)
-    pand      xmm2, xmm7
-    por       xmm0, xmm2    // AG
-    movdqa    xmm2, xmm1
-    punpcklbw xmm1, xmm0
-    punpckhbw xmm2, xmm0
-    movdqa    [eax * 2 + edx], xmm1  // store 4 pixels of ARGB
-    movdqa    [eax * 2 + edx + 16], xmm2  // store next 4 pixels of ARGB
-    lea       eax, [eax + 16]
-    sub       ecx, 8
-    jg        convertloop
-    ret
-  }
-}
-
-// 18 instructions.
-__declspec(naked) __declspec(align(16))
-void ARGB4444ToARGBRow_SSE2(const uint8* src_argb4444, uint8* dst_argb,
-                            int pix) {
-  __asm {
-    mov       eax, 0x0f0f0f0f  // generate mask 0x0f0f0f0f
-    movd      xmm4, eax
-    pshufd    xmm4, xmm4, 0
-    movdqa    xmm5, xmm4       // 0xf0f0f0f0 for high nibbles
-    pslld     xmm5, 4
-    mov       eax, [esp + 4]   // src_argb4444
-    mov       edx, [esp + 8]   // dst_argb
-    mov       ecx, [esp + 12]  // pix
-    sub       edx, eax
-    sub       edx, eax
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]   // fetch 8 pixels of bgra4444
-    movdqa    xmm2, xmm0
-    pand      xmm0, xmm4    // mask low nibbles
-    pand      xmm2, xmm5    // mask high nibbles
-    movdqa    xmm1, xmm0
-    movdqa    xmm3, xmm2
-    psllw     xmm1, 4
-    psrlw     xmm3, 4
-    por       xmm0, xmm1
-    por       xmm2, xmm3
-    movdqa    xmm1, xmm0
-    punpcklbw xmm0, xmm2
-    punpckhbw xmm1, xmm2
-    movdqa    [eax * 2 + edx], xmm0  // store 4 pixels of ARGB
-    movdqa    [eax * 2 + edx + 16], xmm1  // store next 4 pixels of ARGB
-    lea       eax, [eax + 16]
-    sub       ecx, 8
-    jg        convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToRGB24Row_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_argb
-    mov       edx, [esp + 8]   // dst_rgb
-    mov       ecx, [esp + 12]  // pix
-    movdqa    xmm6, kShuffleMaskARGBToRGB24
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]   // fetch 16 pixels of argb
-    movdqu    xmm1, [eax + 16]
-    movdqu    xmm2, [eax + 32]
-    movdqu    xmm3, [eax + 48]
-    lea       eax, [eax + 64]
-    pshufb    xmm0, xmm6    // pack 16 bytes of ARGB to 12 bytes of RGB
-    pshufb    xmm1, xmm6
-    pshufb    xmm2, xmm6
-    pshufb    xmm3, xmm6
-    movdqa    xmm4, xmm1   // 4 bytes from 1 for 0
-    psrldq    xmm1, 4      // 8 bytes from 1
-    pslldq    xmm4, 12     // 4 bytes from 1 for 0
-    movdqa    xmm5, xmm2   // 8 bytes from 2 for 1
-    por       xmm0, xmm4   // 4 bytes from 1 for 0
-    pslldq    xmm5, 8      // 8 bytes from 2 for 1
-    movdqu    [edx], xmm0  // store 0
-    por       xmm1, xmm5   // 8 bytes from 2 for 1
-    psrldq    xmm2, 8      // 4 bytes from 2
-    pslldq    xmm3, 4      // 12 bytes from 3 for 2
-    por       xmm2, xmm3   // 12 bytes from 3 for 2
-    movdqu    [edx + 16], xmm1   // store 1
-    movdqu    [edx + 32], xmm2   // store 2
-    lea       edx, [edx + 48]
-    sub       ecx, 16
-    jg        convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToRAWRow_SSSE3(const uint8* src_argb, uint8* dst_rgb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_argb
-    mov       edx, [esp + 8]   // dst_rgb
-    mov       ecx, [esp + 12]  // pix
-    movdqa    xmm6, kShuffleMaskARGBToRAW
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax]   // fetch 16 pixels of argb
-    movdqu    xmm1, [eax + 16]
-    movdqu    xmm2, [eax + 32]
-    movdqu    xmm3, [eax + 48]
-    lea       eax, [eax + 64]
-    pshufb    xmm0, xmm6    // pack 16 bytes of ARGB to 12 bytes of RGB
-    pshufb    xmm1, xmm6
-    pshufb    xmm2, xmm6
-    pshufb    xmm3, xmm6
-    movdqa    xmm4, xmm1   // 4 bytes from 1 for 0
-    psrldq    xmm1, 4      // 8 bytes from 1
-    pslldq    xmm4, 12     // 4 bytes from 1 for 0
-    movdqa    xmm5, xmm2   // 8 bytes from 2 for 1
-    por       xmm0, xmm4   // 4 bytes from 1 for 0
-    pslldq    xmm5, 8      // 8 bytes from 2 for 1
-    movdqu    [edx], xmm0  // store 0
-    por       xmm1, xmm5   // 8 bytes from 2 for 1
-    psrldq    xmm2, 8      // 4 bytes from 2
-    pslldq    xmm3, 4      // 12 bytes from 3 for 2
-    por       xmm2, xmm3   // 12 bytes from 3 for 2
-    movdqu    [edx + 16], xmm1   // store 1
-    movdqu    [edx + 32], xmm2   // store 2
-    lea       edx, [edx + 48]
-    sub       ecx, 16
-    jg        convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToRGB565Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_argb
-    mov       edx, [esp + 8]   // dst_rgb
-    mov       ecx, [esp + 12]  // pix
-    pcmpeqb   xmm3, xmm3       // generate mask 0x0000001f
-    psrld     xmm3, 27
-    pcmpeqb   xmm4, xmm4       // generate mask 0x000007e0
-    psrld     xmm4, 26
-    pslld     xmm4, 5
-    pcmpeqb   xmm5, xmm5       // generate mask 0xfffff800
-    pslld     xmm5, 11
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax]   // fetch 4 pixels of argb
-    movdqa    xmm1, xmm0    // B
-    movdqa    xmm2, xmm0    // G
-    pslld     xmm0, 8       // R
-    psrld     xmm1, 3       // B
-    psrld     xmm2, 5       // G
-    psrad     xmm0, 16      // R
-    pand      xmm1, xmm3    // B
-    pand      xmm2, xmm4    // G
-    pand      xmm0, xmm5    // R
-    por       xmm1, xmm2    // BG
-    por       xmm0, xmm1    // BGR
-    packssdw  xmm0, xmm0
-    lea       eax, [eax + 16]
-    movq      qword ptr [edx], xmm0  // store 4 pixels of RGB565
-    lea       edx, [edx + 8]
-    sub       ecx, 4
-    jg        convertloop
-    ret
-  }
-}
-
-// TODO(fbarchard): Improve sign extension/packing.
-__declspec(naked) __declspec(align(16))
-void ARGBToARGB1555Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_argb
-    mov       edx, [esp + 8]   // dst_rgb
-    mov       ecx, [esp + 12]  // pix
-    pcmpeqb   xmm4, xmm4       // generate mask 0x0000001f
-    psrld     xmm4, 27
-    movdqa    xmm5, xmm4       // generate mask 0x000003e0
-    pslld     xmm5, 5
-    movdqa    xmm6, xmm4       // generate mask 0x00007c00
-    pslld     xmm6, 10
-    pcmpeqb   xmm7, xmm7       // generate mask 0xffff8000
-    pslld     xmm7, 15
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax]   // fetch 4 pixels of argb
-    movdqa    xmm1, xmm0    // B
-    movdqa    xmm2, xmm0    // G
-    movdqa    xmm3, xmm0    // R
-    psrad     xmm0, 16      // A
-    psrld     xmm1, 3       // B
-    psrld     xmm2, 6       // G
-    psrld     xmm3, 9       // R
-    pand      xmm0, xmm7    // A
-    pand      xmm1, xmm4    // B
-    pand      xmm2, xmm5    // G
-    pand      xmm3, xmm6    // R
-    por       xmm0, xmm1    // BA
-    por       xmm2, xmm3    // GR
-    por       xmm0, xmm2    // BGRA
-    packssdw  xmm0, xmm0
-    lea       eax, [eax + 16]
-    movq      qword ptr [edx], xmm0  // store 4 pixels of ARGB1555
-    lea       edx, [edx + 8]
-    sub       ecx, 4
-    jg        convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToARGB4444Row_SSE2(const uint8* src_argb, uint8* dst_rgb, int pix) {
-  __asm {
-    mov       eax, [esp + 4]   // src_argb
-    mov       edx, [esp + 8]   // dst_rgb
-    mov       ecx, [esp + 12]  // pix
-    pcmpeqb   xmm4, xmm4       // generate mask 0xf000f000
-    psllw     xmm4, 12
-    movdqa    xmm3, xmm4       // generate mask 0x00f000f0
-    psrlw     xmm3, 8
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax]   // fetch 4 pixels of argb
-    movdqa    xmm1, xmm0
-    pand      xmm0, xmm3    // low nibble
-    pand      xmm1, xmm4    // high nibble
-    psrl      xmm0, 4
-    psrl      xmm1, 8
-    por       xmm0, xmm1
-    packuswb  xmm0, xmm0
-    lea       eax, [eax + 16]
-    movq      qword ptr [edx], xmm0  // store 4 pixels of ARGB4444
-    lea       edx, [edx + 8]
-    sub       ecx, 4
-    jg        convertloop
-    ret
-  }
-}
-
-// Convert 16 ARGB pixels (64 bytes) to 16 Y values.
-__declspec(naked) __declspec(align(16))
-void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kARGBToY
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-// Convert 16 ARGB pixels (64 bytes) to 16 Y values.
-__declspec(naked) __declspec(align(16))
-void ARGBToYJRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm4, kARGBToYJ
-    movdqa     xmm5, kAddYJ64
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    paddw      xmm0, xmm5  // Add .5 for rounding.
-    paddw      xmm2, xmm5
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-#ifdef HAS_ARGBTOYROW_AVX2
-// Convert 32 ARGB pixels (128 bytes) to 32 Y values.
-__declspec(naked) __declspec(align(32))
-void ARGBToYRow_AVX2(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    vbroadcastf128 ymm4, kARGBToY
-    vbroadcastf128 ymm5, kAddY16
-    vmovdqa    ymm6, kPermdARGBToY_AVX
-
-    align      4
- convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    vmovdqu    ymm2, [eax + 64]
-    vmovdqu    ymm3, [eax + 96]
-    vpmaddubsw ymm0, ymm0, ymm4
-    vpmaddubsw ymm1, ymm1, ymm4
-    vpmaddubsw ymm2, ymm2, ymm4
-    vpmaddubsw ymm3, ymm3, ymm4
-    lea        eax, [eax + 128]
-    vphaddw    ymm0, ymm0, ymm1  // mutates.
-    vphaddw    ymm2, ymm2, ymm3
-    vpsrlw     ymm0, ymm0, 7
-    vpsrlw     ymm2, ymm2, 7
-    vpackuswb  ymm0, ymm0, ymm2  // mutates.
-    vpermd     ymm0, ymm6, ymm0  // For vphaddw + vpackuswb mutation.
-    vpaddb     ymm0, ymm0, ymm5
-    sub        ecx, 32
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    jg         convertloop
-    vzeroupper
-    ret
-  }
-}
-#endif  //  HAS_ARGBTOYROW_AVX2
-
-#ifdef HAS_ARGBTOYROW_AVX2
-// Convert 32 ARGB pixels (128 bytes) to 32 Y values.
-__declspec(naked) __declspec(align(32))
-void ARGBToYJRow_AVX2(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    vbroadcastf128 ymm4, kARGBToYJ
-    vbroadcastf128 ymm5, kAddYJ64
-    vmovdqa    ymm6, kPermdARGBToY_AVX
-
-    align      4
- convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    vmovdqu    ymm2, [eax + 64]
-    vmovdqu    ymm3, [eax + 96]
-    vpmaddubsw ymm0, ymm0, ymm4
-    vpmaddubsw ymm1, ymm1, ymm4
-    vpmaddubsw ymm2, ymm2, ymm4
-    vpmaddubsw ymm3, ymm3, ymm4
-    lea        eax, [eax + 128]
-    vphaddw    ymm0, ymm0, ymm1  // mutates.
-    vphaddw    ymm2, ymm2, ymm3
-    vpaddw     ymm0, ymm0, ymm5  // Add .5 for rounding.
-    vpaddw     ymm2, ymm2, ymm5
-    vpsrlw     ymm0, ymm0, 7
-    vpsrlw     ymm2, ymm2, 7
-    vpackuswb  ymm0, ymm0, ymm2  // mutates.
-    vpermd     ymm0, ymm6, ymm0  // For vphaddw + vpackuswb mutation.
-    sub        ecx, 32
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    jg         convertloop
-
-    vzeroupper
-    ret
-  }
-}
-#endif  //  HAS_ARGBTOYJROW_AVX2
-
-__declspec(naked) __declspec(align(16))
-void ARGBToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kARGBToY
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToYJRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm4, kARGBToYJ
-    movdqa     xmm5, kAddYJ64
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    paddw      xmm0, xmm5
-    paddw      xmm2, xmm5
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void BGRAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kBGRAToY
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void BGRAToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kBGRAToY
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ABGRToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kABGRToY
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ABGRToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kABGRToY
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RGBAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kRGBAToY
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RGBAToYRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_y */
-    mov        ecx, [esp + 12]  /* pix */
-    movdqa     xmm5, kAddY16
-    movdqa     xmm4, kRGBAToY
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm2, xmm4
-    pmaddubsw  xmm3, xmm4
-    lea        eax, [eax + 64]
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psrlw      xmm0, 7
-    psrlw      xmm2, 7
-    packuswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pavgb      xmm0, [eax + esi]
-    pavgb      xmm1, [eax + esi + 16]
-    pavgb      xmm2, [eax + esi + 32]
-    pavgb      xmm3, [eax + esi + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUVJRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                        uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kARGBToUJ
-    movdqa     xmm6, kARGBToVJ
-    movdqa     xmm5, kAddUVJ128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pavgb      xmm0, [eax + esi]
-    pavgb      xmm1, [eax + esi + 16]
-    pavgb      xmm2, [eax + esi + 32]
-    pavgb      xmm3, [eax + esi + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    paddw      xmm0, xmm5            // +.5 rounding -> unsigned
-    paddw      xmm1, xmm5
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-#ifdef HAS_ARGBTOUVROW_AVX2
-__declspec(naked) __declspec(align(32))
-void ARGBToUVRow_AVX2(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    vbroadcastf128 ymm5, kAddUV128
-    vbroadcastf128 ymm6, kARGBToV
-    vbroadcastf128 ymm7, kARGBToU
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 32x2 argb pixels to 16x1 */
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    vmovdqu    ymm2, [eax + 64]
-    vmovdqu    ymm3, [eax + 96]
-    vpavgb     ymm0, ymm0, [eax + esi]
-    vpavgb     ymm1, ymm1, [eax + esi + 32]
-    vpavgb     ymm2, ymm2, [eax + esi + 64]
-    vpavgb     ymm3, ymm3, [eax + esi + 96]
-    lea        eax,  [eax + 128]
-    vshufps    ymm4, ymm0, ymm1, 0x88
-    vshufps    ymm0, ymm0, ymm1, 0xdd
-    vpavgb     ymm0, ymm0, ymm4  // mutated by vshufps
-    vshufps    ymm4, ymm2, ymm3, 0x88
-    vshufps    ymm2, ymm2, ymm3, 0xdd
-    vpavgb     ymm2, ymm2, ymm4  // mutated by vshufps
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 32 different pixels, its 16 pixels of U and 16 of V
-    vpmaddubsw ymm1, ymm0, ymm7  // U
-    vpmaddubsw ymm3, ymm2, ymm7
-    vpmaddubsw ymm0, ymm0, ymm6  // V
-    vpmaddubsw ymm2, ymm2, ymm6
-    vphaddw    ymm1, ymm1, ymm3  // mutates
-    vphaddw    ymm0, ymm0, ymm2
-    vpsraw     ymm1, ymm1, 8
-    vpsraw     ymm0, ymm0, 8
-    vpacksswb  ymm0, ymm1, ymm0  // mutates
-    vpermq     ymm0, ymm0, 0xd8  // For vpacksswb
-    vpshufb    ymm0, ymm0, kShufARGBToUV_AVX  // For vshufps + vphaddw
-    vpaddb     ymm0, ymm0, ymm5  // -> unsigned
-
-    // step 3 - store 16 U and 16 V values
-    sub         ecx, 32
-    vextractf128 [edx], ymm0, 0 // U
-    vextractf128 [edx + edi], ymm0, 1 // V
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBTOUVROW_AVX2
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    movdqu     xmm4, [eax + esi]
-    pavgb      xmm0, xmm4
-    movdqu     xmm4, [eax + esi + 16]
-    pavgb      xmm1, xmm4
-    movdqu     xmm4, [eax + esi + 32]
-    pavgb      xmm2, xmm4
-    movdqu     xmm4, [eax + esi + 48]
-    pavgb      xmm3, xmm4
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUVJRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kARGBToUJ
-    movdqa     xmm6, kARGBToVJ
-    movdqa     xmm5, kAddUVJ128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    movdqu     xmm4, [eax + esi]
-    pavgb      xmm0, xmm4
-    movdqu     xmm4, [eax + esi + 16]
-    pavgb      xmm1, xmm4
-    movdqu     xmm4, [eax + esi + 32]
-    pavgb      xmm2, xmm4
-    movdqu     xmm4, [eax + esi + 48]
-    pavgb      xmm3, xmm4
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    paddw      xmm0, xmm5            // +.5 rounding -> unsigned
-    paddw      xmm1, xmm5
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUV444Row_SSSE3(const uint8* src_argb0,
-                          uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]   // src_argb
-    mov        edx, [esp + 4 + 8]   // dst_u
-    mov        edi, [esp + 4 + 12]  // dst_v
-    mov        ecx, [esp + 4 + 16]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* convert to U and V */
-    movdqa     xmm0, [eax]          // U
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm7
-    pmaddubsw  xmm1, xmm7
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm3, xmm7
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psraw      xmm0, 8
-    psraw      xmm2, 8
-    packsswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx,  16
-    movdqa     [edx], xmm0
-
-    movdqa     xmm0, [eax]          // V
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm6
-    pmaddubsw  xmm1, xmm6
-    pmaddubsw  xmm2, xmm6
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psraw      xmm0, 8
-    psraw      xmm2, 8
-    packsswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    lea        eax,  [eax + 64]
-    movdqa     [edx + edi], xmm0
-    lea        edx,  [edx + 16]
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb0,
-                                    uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]   // src_argb
-    mov        edx, [esp + 4 + 8]   // dst_u
-    mov        edi, [esp + 4 + 12]  // dst_v
-    mov        ecx, [esp + 4 + 16]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* convert to U and V */
-    movdqu     xmm0, [eax]          // U
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm7
-    pmaddubsw  xmm1, xmm7
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm3, xmm7
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psraw      xmm0, 8
-    psraw      xmm2, 8
-    packsswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    sub        ecx,  16
-    movdqu     [edx], xmm0
-
-    movdqu     xmm0, [eax]          // V
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    pmaddubsw  xmm0, xmm6
-    pmaddubsw  xmm1, xmm6
-    pmaddubsw  xmm2, xmm6
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm1
-    phaddw     xmm2, xmm3
-    psraw      xmm0, 8
-    psraw      xmm2, 8
-    packsswb   xmm0, xmm2
-    paddb      xmm0, xmm5
-    lea        eax,  [eax + 64]
-    movdqu     [edx + edi], xmm0
-    lea        edx,  [edx + 16]
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUV422Row_SSSE3(const uint8* src_argb0,
-                          uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]   // src_argb
-    mov        edx, [esp + 4 + 8]   // dst_u
-    mov        edi, [esp + 4 + 12]  // dst_v
-    mov        ecx, [esp + 4 + 16]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBToUV422Row_Unaligned_SSSE3(const uint8* src_argb0,
-                                    uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]   // src_argb
-    mov        edx, [esp + 4 + 8]   // dst_u
-    mov        edi, [esp + 4 + 12]  // dst_v
-    mov        ecx, [esp + 4 + 16]  // pix
-    movdqa     xmm7, kARGBToU
-    movdqa     xmm6, kARGBToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void BGRAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kBGRAToU
-    movdqa     xmm6, kBGRAToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pavgb      xmm0, [eax + esi]
-    pavgb      xmm1, [eax + esi + 16]
-    pavgb      xmm2, [eax + esi + 32]
-    pavgb      xmm3, [eax + esi + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void BGRAToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kBGRAToU
-    movdqa     xmm6, kBGRAToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    movdqu     xmm4, [eax + esi]
-    pavgb      xmm0, xmm4
-    movdqu     xmm4, [eax + esi + 16]
-    pavgb      xmm1, xmm4
-    movdqu     xmm4, [eax + esi + 32]
-    pavgb      xmm2, xmm4
-    movdqu     xmm4, [eax + esi + 48]
-    pavgb      xmm3, xmm4
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ABGRToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kABGRToU
-    movdqa     xmm6, kABGRToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pavgb      xmm0, [eax + esi]
-    pavgb      xmm1, [eax + esi + 16]
-    pavgb      xmm2, [eax + esi + 32]
-    pavgb      xmm3, [eax + esi + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ABGRToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kABGRToU
-    movdqa     xmm6, kABGRToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    movdqu     xmm4, [eax + esi]
-    pavgb      xmm0, xmm4
-    movdqu     xmm4, [eax + esi + 16]
-    pavgb      xmm1, xmm4
-    movdqu     xmm4, [eax + esi + 32]
-    pavgb      xmm2, xmm4
-    movdqu     xmm4, [eax + esi + 48]
-    pavgb      xmm3, xmm4
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RGBAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                       uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kRGBAToU
-    movdqa     xmm6, kRGBAToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-    pavgb      xmm0, [eax + esi]
-    pavgb      xmm1, [eax + esi + 16]
-    pavgb      xmm2, [eax + esi + 32]
-    pavgb      xmm3, [eax + esi + 48]
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void RGBAToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
-                                 uint8* dst_u, uint8* dst_v, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb
-    mov        esi, [esp + 8 + 8]   // src_stride_argb
-    mov        edx, [esp + 8 + 12]  // dst_u
-    mov        edi, [esp + 8 + 16]  // dst_v
-    mov        ecx, [esp + 8 + 20]  // pix
-    movdqa     xmm7, kRGBAToU
-    movdqa     xmm6, kRGBAToV
-    movdqa     xmm5, kAddUV128
-    sub        edi, edx             // stride from u to v
-
-    align      4
- convertloop:
-    /* step 1 - subsample 16x2 argb pixels to 8x1 */
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + 32]
-    movdqu     xmm3, [eax + 48]
-    movdqu     xmm4, [eax + esi]
-    pavgb      xmm0, xmm4
-    movdqu     xmm4, [eax + esi + 16]
-    pavgb      xmm1, xmm4
-    movdqu     xmm4, [eax + esi + 32]
-    pavgb      xmm2, xmm4
-    movdqu     xmm4, [eax + esi + 48]
-    pavgb      xmm3, xmm4
-    lea        eax,  [eax + 64]
-    movdqa     xmm4, xmm0
-    shufps     xmm0, xmm1, 0x88
-    shufps     xmm4, xmm1, 0xdd
-    pavgb      xmm0, xmm4
-    movdqa     xmm4, xmm2
-    shufps     xmm2, xmm3, 0x88
-    shufps     xmm4, xmm3, 0xdd
-    pavgb      xmm2, xmm4
-
-    // step 2 - convert to U and V
-    // from here down is very similar to Y code except
-    // instead of 16 different pixels, its 8 pixels of U and 8 of V
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    pmaddubsw  xmm0, xmm7  // U
-    pmaddubsw  xmm2, xmm7
-    pmaddubsw  xmm1, xmm6  // V
-    pmaddubsw  xmm3, xmm6
-    phaddw     xmm0, xmm2
-    phaddw     xmm1, xmm3
-    psraw      xmm0, 8
-    psraw      xmm1, 8
-    packsswb   xmm0, xmm1
-    paddb      xmm0, xmm5            // -> unsigned
-
-    // step 3 - store 8 U and 8 V values
-    sub        ecx, 16
-    movlps     qword ptr [edx], xmm0 // U
-    movhps     qword ptr [edx + edi], xmm0 // V
-    lea        edx, [edx + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBTOYROW_SSSE3
-
-#define YG 74 /* (int8)(1.164 * 64 + 0.5) */
-
-#define UB 127 /* min(63,(int8)(2.018 * 64)) */
-#define UG -25 /* (int8)(-0.391 * 64 - 0.5) */
-#define UR 0
-
-#define VB 0
-#define VG -52 /* (int8)(-0.813 * 64 - 0.5) */
-#define VR 102 /* (int8)(1.596 * 64 + 0.5) */
-
-// Bias
-#define BB UB * 128 + VB * 128
-#define BG UG * 128 + VG * 128
-#define BR UR * 128 + VR * 128
-
-#ifdef HAS_I422TOARGBROW_AVX2
-
-static const lvec8 kUVToB_AVX = {
-  UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB,
-  UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB
-};
-static const lvec8 kUVToR_AVX = {
-  UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR,
-  UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR
-};
-static const lvec8 kUVToG_AVX = {
-  UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG,
-  UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG
-};
-static const lvec16 kYToRgb_AVX = {
-  YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG, YG
-};
-static const lvec16 kYSub16_AVX = {
-  16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16
-};
-static const lvec16 kUVBiasB_AVX = {
-  BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB, BB
-};
-static const lvec16 kUVBiasG_AVX = {
-  BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG, BG
-};
-static const lvec16 kUVBiasR_AVX = {
-  BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR, BR
-};
-
-// 16 pixels
-// 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ARGB (64 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToARGBRow_AVX2(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    vpcmpeqb   ymm5, ymm5, ymm5     // generate 0xffffffffffffffff for alpha
-    vpxor      ymm4, ymm4, ymm4
-
-    align      4
- convertloop:
-    vmovq      xmm0, qword ptr [esi]          //  U
-    vmovq      xmm1, qword ptr [esi + edi]    //  V
-    lea        esi,  [esi + 8]
-    vpunpcklbw ymm0, ymm0, ymm1               // UV
-    vpermq     ymm0, ymm0, 0xd8
-    vpunpcklwd ymm0, ymm0, ymm0              // UVUV
-    vpmaddubsw ymm2, ymm0, kUVToB_AVX        // scale B UV
-    vpmaddubsw ymm1, ymm0, kUVToG_AVX        // scale G UV
-    vpmaddubsw ymm0, ymm0, kUVToR_AVX        // scale R UV
-    vpsubw     ymm2, ymm2, kUVBiasB_AVX      // unbias back to signed
-    vpsubw     ymm1, ymm1, kUVBiasG_AVX
-    vpsubw     ymm0, ymm0, kUVBiasR_AVX
-
-    // Step 2: Find Y contribution to 16 R,G,B values
-    vmovdqu    xmm3, [eax]                  // NOLINT
-    lea        eax, [eax + 16]
-    vpermq     ymm3, ymm3, 0xd8
-    vpunpcklbw ymm3, ymm3, ymm4
-    vpsubsw    ymm3, ymm3, kYSub16_AVX
-    vpmullw    ymm3, ymm3, kYToRgb_AVX
-    vpaddsw    ymm2, ymm2, ymm3           // B += Y
-    vpaddsw    ymm1, ymm1, ymm3           // G += Y
-    vpaddsw    ymm0, ymm0, ymm3           // R += Y
-    vpsraw     ymm2, ymm2, 6
-    vpsraw     ymm1, ymm1, 6
-    vpsraw     ymm0, ymm0, 6
-    vpackuswb  ymm2, ymm2, ymm2           // B
-    vpackuswb  ymm1, ymm1, ymm1           // G
-    vpackuswb  ymm0, ymm0, ymm0           // R
-
-    // Step 3: Weave into ARGB
-    vpunpcklbw ymm2, ymm2, ymm1           // BG
-    vpermq     ymm2, ymm2, 0xd8
-    vpunpcklbw ymm0, ymm0, ymm5           // RA
-    vpermq     ymm0, ymm0, 0xd8
-    vpunpcklwd ymm1, ymm2, ymm0           // BGRA first 8 pixels
-    vpunpckhwd ymm2, ymm2, ymm0           // BGRA next 8 pixels
-    vmovdqu    [edx], ymm1
-    vmovdqu    [edx + 32], ymm2
-    lea        edx,  [edx + 64]
-    sub        ecx, 16
-    jg         convertloop
-    vzeroupper
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_I422TOARGBROW_AVX2
-
-#ifdef HAS_I422TOARGBROW_SSSE3
-
-static const vec8 kUVToB = {
-  UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB
-};
-
-static const vec8 kUVToR = {
-  UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR
-};
-
-static const vec8 kUVToG = {
-  UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG
-};
-
-static const vec8 kVUToB = {
-  VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB, VB, UB,
-};
-
-static const vec8 kVUToR = {
-  VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR, VR, UR,
-};
-
-static const vec8 kVUToG = {
-  VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG, VG, UG,
-};
-
-static const vec16 kYToRgb = { YG, YG, YG, YG, YG, YG, YG, YG };
-static const vec16 kYSub16 = { 16, 16, 16, 16, 16, 16, 16, 16 };
-static const vec16 kUVBiasB = { BB, BB, BB, BB, BB, BB, BB, BB };
-static const vec16 kUVBiasG = { BG, BG, BG, BG, BG, BG, BG, BG };
-static const vec16 kUVBiasR = { BR, BR, BR, BR, BR, BR, BR, BR };
-
-// TODO(fbarchard): Read that does half size on Y and treats 420 as 444.
-
-// Read 8 UV from 444.
-#define READYUV444 __asm {                                                     \
-    __asm movq       xmm0, qword ptr [esi] /* U */                /* NOLINT */ \
-    __asm movq       xmm1, qword ptr [esi + edi] /* V */          /* NOLINT */ \
-    __asm lea        esi,  [esi + 8]                                           \
-    __asm punpcklbw  xmm0, xmm1           /* UV */                             \
-  }
-
-// Read 4 UV from 422, upsample to 8 UV.
-#define READYUV422 __asm {                                                     \
-    __asm movd       xmm0, [esi]          /* U */                              \
-    __asm movd       xmm1, [esi + edi]    /* V */                              \
-    __asm lea        esi,  [esi + 4]                                           \
-    __asm punpcklbw  xmm0, xmm1           /* UV */                             \
-    __asm punpcklwd  xmm0, xmm0           /* UVUV (upsample) */                \
-  }
-
-// Read 2 UV from 411, upsample to 8 UV.
-#define READYUV411 __asm {                                                     \
-    __asm movzx      ebx, word ptr [esi]        /* U */           /* NOLINT */ \
-    __asm movd       xmm0, ebx                                                 \
-    __asm movzx      ebx, word ptr [esi + edi]  /* V */           /* NOLINT */ \
-    __asm movd       xmm1, ebx                                                 \
-    __asm lea        esi,  [esi + 2]                                           \
-    __asm punpcklbw  xmm0, xmm1           /* UV */                             \
-    __asm punpcklwd  xmm0, xmm0           /* UVUV (upsample) */                \
-    __asm punpckldq  xmm0, xmm0           /* UVUV (upsample) */                \
-  }
-
-// Read 4 UV from NV12, upsample to 8 UV.
-#define READNV12 __asm {                                                       \
-    __asm movq       xmm0, qword ptr [esi] /* UV */               /* NOLINT */ \
-    __asm lea        esi,  [esi + 8]                                           \
-    __asm punpcklwd  xmm0, xmm0           /* UVUV (upsample) */                \
-  }
-
-// Convert 8 pixels: 8 UV and 8 Y.
-#define YUVTORGB __asm {                                                       \
-    /* Step 1: Find 4 UV contributions to 8 R,G,B values */                    \
-    __asm movdqa     xmm1, xmm0                                                \
-    __asm movdqa     xmm2, xmm0                                                \
-    __asm pmaddubsw  xmm0, kUVToB        /* scale B UV */                      \
-    __asm pmaddubsw  xmm1, kUVToG        /* scale G UV */                      \
-    __asm pmaddubsw  xmm2, kUVToR        /* scale R UV */                      \
-    __asm psubw      xmm0, kUVBiasB      /* unbias back to signed */           \
-    __asm psubw      xmm1, kUVBiasG                                            \
-    __asm psubw      xmm2, kUVBiasR                                            \
-    /* Step 2: Find Y contribution to 8 R,G,B values */                        \
-    __asm movq       xmm3, qword ptr [eax]                        /* NOLINT */ \
-    __asm lea        eax, [eax + 8]                                            \
-    __asm punpcklbw  xmm3, xmm4                                                \
-    __asm psubsw     xmm3, kYSub16                                             \
-    __asm pmullw     xmm3, kYToRgb                                             \
-    __asm paddsw     xmm0, xmm3           /* B += Y */                         \
-    __asm paddsw     xmm1, xmm3           /* G += Y */                         \
-    __asm paddsw     xmm2, xmm3           /* R += Y */                         \
-    __asm psraw      xmm0, 6                                                   \
-    __asm psraw      xmm1, 6                                                   \
-    __asm psraw      xmm2, 6                                                   \
-    __asm packuswb   xmm0, xmm0           /* B */                              \
-    __asm packuswb   xmm1, xmm1           /* G */                              \
-    __asm packuswb   xmm2, xmm2           /* R */                              \
-  }
-
-// Convert 8 pixels: 8 VU and 8 Y.
-#define YVUTORGB __asm {                                                       \
-    /* Step 1: Find 4 UV contributions to 8 R,G,B values */                    \
-    __asm movdqa     xmm1, xmm0                                                \
-    __asm movdqa     xmm2, xmm0                                                \
-    __asm pmaddubsw  xmm0, kVUToB        /* scale B UV */                      \
-    __asm pmaddubsw  xmm1, kVUToG        /* scale G UV */                      \
-    __asm pmaddubsw  xmm2, kVUToR        /* scale R UV */                      \
-    __asm psubw      xmm0, kUVBiasB      /* unbias back to signed */           \
-    __asm psubw      xmm1, kUVBiasG                                            \
-    __asm psubw      xmm2, kUVBiasR                                            \
-    /* Step 2: Find Y contribution to 8 R,G,B values */                        \
-    __asm movq       xmm3, qword ptr [eax]                        /* NOLINT */ \
-    __asm lea        eax, [eax + 8]                                            \
-    __asm punpcklbw  xmm3, xmm4                                                \
-    __asm psubsw     xmm3, kYSub16                                             \
-    __asm pmullw     xmm3, kYToRgb                                             \
-    __asm paddsw     xmm0, xmm3           /* B += Y */                         \
-    __asm paddsw     xmm1, xmm3           /* G += Y */                         \
-    __asm paddsw     xmm2, xmm3           /* R += Y */                         \
-    __asm psraw      xmm0, 6                                                   \
-    __asm psraw      xmm1, 6                                                   \
-    __asm psraw      xmm2, 6                                                   \
-    __asm packuswb   xmm0, xmm0           /* B */                              \
-    __asm packuswb   xmm1, xmm1           /* G */                              \
-    __asm packuswb   xmm2, xmm2           /* R */                              \
-  }
-
-// 8 pixels, dest aligned 16.
-// 8 UV values, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I444ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV444
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToRGB24Row_SSSE3(const uint8* y_buf,
-                          const uint8* u_buf,
-                          const uint8* v_buf,
-                          uint8* dst_rgb24,
-                          int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgb24
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-    movdqa     xmm5, kShuffleMaskARGBToRGB24_0
-    movdqa     xmm6, kShuffleMaskARGBToRGB24
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into RRGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm2           // RR
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRR first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRR next 4 pixels
-    pshufb     xmm0, xmm5           // Pack into first 8 and last 4 bytes.
-    pshufb     xmm1, xmm6           // Pack into first 12 bytes.
-    palignr    xmm1, xmm0, 12       // last 4 bytes of xmm0 + 12 from xmm1
-    movq       qword ptr [edx], xmm0  // First 8 bytes
-    movdqu     [edx + 8], xmm1      // Last 16 bytes. = 24 bytes, 8 RGB pixels.
-    lea        edx,  [edx + 24]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToRAWRow_SSSE3(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* dst_raw,
-                        int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // raw
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-    movdqa     xmm5, kShuffleMaskARGBToRAW_0
-    movdqa     xmm6, kShuffleMaskARGBToRAW
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into RRGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm2           // RR
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRR first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRR next 4 pixels
-    pshufb     xmm0, xmm5           // Pack into first 8 and last 4 bytes.
-    pshufb     xmm1, xmm6           // Pack into first 12 bytes.
-    palignr    xmm1, xmm0, 12       // last 4 bytes of xmm0 + 12 from xmm1
-    movq       qword ptr [edx], xmm0  // First 8 bytes
-    movdqu     [edx + 8], xmm1      // Last 16 bytes. = 24 bytes, 8 RGB pixels.
-    lea        edx,  [edx + 24]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest unaligned.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToRGB565Row_SSSE3(const uint8* y_buf,
-                           const uint8* u_buf,
-                           const uint8* v_buf,
-                           uint8* rgb565_buf,
-                           int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgb565
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-    pcmpeqb    xmm5, xmm5       // generate mask 0x0000001f
-    psrld      xmm5, 27
-    pcmpeqb    xmm6, xmm6       // generate mask 0x000007e0
-    psrld      xmm6, 26
-    pslld      xmm6, 5
-    pcmpeqb    xmm7, xmm7       // generate mask 0xfffff800
-    pslld      xmm7, 11
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into RRGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm2           // RR
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRR first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRR next 4 pixels
-
-    // Step 3b: RRGB -> RGB565
-    movdqa     xmm3, xmm0    // B  first 4 pixels of argb
-    movdqa     xmm2, xmm0    // G
-    pslld      xmm0, 8       // R
-    psrld      xmm3, 3       // B
-    psrld      xmm2, 5       // G
-    psrad      xmm0, 16      // R
-    pand       xmm3, xmm5    // B
-    pand       xmm2, xmm6    // G
-    pand       xmm0, xmm7    // R
-    por        xmm3, xmm2    // BG
-    por        xmm0, xmm3    // BGR
-    movdqa     xmm3, xmm1    // B  next 4 pixels of argb
-    movdqa     xmm2, xmm1    // G
-    pslld      xmm1, 8       // R
-    psrld      xmm3, 3       // B
-    psrld      xmm2, 5       // G
-    psrad      xmm1, 16      // R
-    pand       xmm3, xmm5    // B
-    pand       xmm2, xmm6    // G
-    pand       xmm1, xmm7    // R
-    por        xmm3, xmm2    // BG
-    por        xmm1, xmm3    // BGR
-    packssdw   xmm0, xmm1
-    sub        ecx, 8
-    movdqu     [edx], xmm0   // store 8 pixels of RGB565
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 2 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-// Similar to I420 but duplicate UV once more.
-__declspec(naked) __declspec(align(16))
-void I411ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       ebx
-    push       esi
-    push       edi
-    mov        eax, [esp + 12 + 4]   // Y
-    mov        esi, [esp + 12 + 8]   // U
-    mov        edi, [esp + 12 + 12]  // V
-    mov        edx, [esp + 12 + 16]  // argb
-    mov        ecx, [esp + 12 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV411  // modifies EBX
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    pop        ebx
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void NV12ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* uv_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READNV12
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void NV21ToARGBRow_SSSE3(const uint8* y_buf,
-                         const uint8* uv_buf,
-                         uint8* dst_argb,
-                         int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // VU
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READNV12
-    YVUTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, unaligned.
-// 8 UV values, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I444ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_argb,
-                                   int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV444
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, unaligned.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void I422ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_argb,
-                                   int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // argb
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, unaligned.
-// 2 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-// Similar to I420 but duplicate UV once more.
-__declspec(naked) __declspec(align(16))
-void I411ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_argb,
-                                   int width) {
-  __asm {
-    push       ebx
-    push       esi
-    push       edi
-    mov        eax, [esp + 12 + 4]   // Y
-    mov        esi, [esp + 12 + 8]   // U
-    mov        edi, [esp + 12 + 12]  // V
-    mov        edx, [esp + 12 + 16]  // argb
-    mov        ecx, [esp + 12 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV411  // modifies EBX
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    pop        ebx
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void NV12ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* uv_buf,
-                                   uint8* dst_argb,
-                                   int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // UV
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READNV12
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-
-// 8 pixels, dest aligned 16.
-// 4 UV values upsampled to 8 UV, mixed with 8 Y producing 8 ARGB (32 bytes).
-__declspec(naked) __declspec(align(16))
-void NV21ToARGBRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* uv_buf,
-                                   uint8* dst_argb,
-                                   int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // Y
-    mov        esi, [esp + 4 + 8]   // VU
-    mov        edx, [esp + 4 + 12]  // argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READNV12
-    YVUTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm0, xmm1           // BG
-    punpcklbw  xmm2, xmm5           // RA
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm2           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm2           // BGRA next 4 pixels
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToBGRARow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_bgra,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // bgra
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into BGRA
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    punpcklbw  xmm1, xmm0           // GB
-    punpcklbw  xmm5, xmm2           // AR
-    movdqa     xmm0, xmm5
-    punpcklwd  xmm5, xmm1           // BGRA first 4 pixels
-    punpckhwd  xmm0, xmm1           // BGRA next 4 pixels
-    movdqa     [edx], xmm5
-    movdqa     [edx + 16], xmm0
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToBGRARow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_bgra,
-                                   int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // bgra
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into BGRA
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    punpcklbw  xmm1, xmm0           // GB
-    punpcklbw  xmm5, xmm2           // AR
-    movdqa     xmm0, xmm5
-    punpcklwd  xmm5, xmm1           // BGRA first 4 pixels
-    punpckhwd  xmm0, xmm1           // BGRA next 4 pixels
-    movdqu     [edx], xmm5
-    movdqu     [edx + 16], xmm0
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToABGRRow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_abgr,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // abgr
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm2, xmm1           // RG
-    punpcklbw  xmm0, xmm5           // BA
-    movdqa     xmm1, xmm2
-    punpcklwd  xmm2, xmm0           // RGBA first 4 pixels
-    punpckhwd  xmm1, xmm0           // RGBA next 4 pixels
-    movdqa     [edx], xmm2
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToABGRRow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_abgr,
-                                   int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // abgr
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into ARGB
-    punpcklbw  xmm2, xmm1           // RG
-    punpcklbw  xmm0, xmm5           // BA
-    movdqa     xmm1, xmm2
-    punpcklwd  xmm2, xmm0           // RGBA first 4 pixels
-    punpckhwd  xmm1, xmm0           // RGBA next 4 pixels
-    movdqu     [edx], xmm2
-    movdqu     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToRGBARow_SSSE3(const uint8* y_buf,
-                         const uint8* u_buf,
-                         const uint8* v_buf,
-                         uint8* dst_rgba,
-                         int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgba
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into RGBA
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    punpcklbw  xmm1, xmm2           // GR
-    punpcklbw  xmm5, xmm0           // AB
-    movdqa     xmm0, xmm5
-    punpcklwd  xmm5, xmm1           // RGBA first 4 pixels
-    punpckhwd  xmm0, xmm1           // RGBA next 4 pixels
-    movdqa     [edx], xmm5
-    movdqa     [edx + 16], xmm0
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToRGBARow_Unaligned_SSSE3(const uint8* y_buf,
-                                   const uint8* u_buf,
-                                   const uint8* v_buf,
-                                   uint8* dst_rgba,
-                                   int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // Y
-    mov        esi, [esp + 8 + 8]   // U
-    mov        edi, [esp + 8 + 12]  // V
-    mov        edx, [esp + 8 + 16]  // rgba
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        edi, esi
-    pxor       xmm4, xmm4
-
-    align      4
- convertloop:
-    READYUV422
-    YUVTORGB
-
-    // Step 3: Weave into RGBA
-    pcmpeqb    xmm5, xmm5           // generate 0xffffffff for alpha
-    punpcklbw  xmm1, xmm2           // GR
-    punpcklbw  xmm5, xmm0           // AB
-    movdqa     xmm0, xmm5
-    punpcklwd  xmm5, xmm1           // RGBA first 4 pixels
-    punpckhwd  xmm0, xmm1           // RGBA next 4 pixels
-    movdqu     [edx], xmm5
-    movdqu     [edx + 16], xmm0
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-#endif  // HAS_I422TOARGBROW_SSSE3
-
-#ifdef HAS_YTOARGBROW_SSE2
-__declspec(naked) __declspec(align(16))
-void YToARGBRow_SSE2(const uint8* y_buf,
-                     uint8* rgb_buf,
-                     int width) {
-  __asm {
-    pxor       xmm5, xmm5
-    pcmpeqb    xmm4, xmm4           // generate mask 0xff000000
-    pslld      xmm4, 24
-    mov        eax, 0x00100010
-    movd       xmm3, eax
-    pshufd     xmm3, xmm3, 0
-    mov        eax, 0x004a004a       // 74
-    movd       xmm2, eax
-    pshufd     xmm2, xmm2,0
-    mov        eax, [esp + 4]       // Y
-    mov        edx, [esp + 8]       // rgb
-    mov        ecx, [esp + 12]      // width
-
-    align      4
- convertloop:
-    // Step 1: Scale Y contribution to 8 G values. G = (y - 16) * 1.164
-    movq       xmm0, qword ptr [eax]
-    lea        eax, [eax + 8]
-    punpcklbw  xmm0, xmm5           // 0.Y
-    psubusw    xmm0, xmm3
-    pmullw     xmm0, xmm2
-    psrlw      xmm0, 6
-    packuswb   xmm0, xmm0           // G
-
-    // Step 2: Weave into ARGB
-    punpcklbw  xmm0, xmm0           // GG
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm0           // BGRA first 4 pixels
-    punpckhwd  xmm1, xmm1           // BGRA next 4 pixels
-    por        xmm0, xmm4
-    por        xmm1, xmm4
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx,  [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_YTOARGBROW_SSE2
-
-#ifdef HAS_MIRRORROW_SSSE3
-// Shuffle table for reversing the bytes.
-static const uvec8 kShuffleMirror = {
-  15u, 14u, 13u, 12u, 11u, 10u, 9u, 8u, 7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u
-};
-
-__declspec(naked) __declspec(align(16))
-void MirrorRow_SSSE3(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov       eax, [esp + 4]   // src
-    mov       edx, [esp + 8]   // dst
-    mov       ecx, [esp + 12]  // width
-    movdqa    xmm5, kShuffleMirror
-    lea       eax, [eax - 16]
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax + ecx]
-    pshufb    xmm0, xmm5
-    sub       ecx, 16
-    movdqa    [edx], xmm0
-    lea       edx, [edx + 16]
-    jg        convertloop
-    ret
-  }
-}
-#endif  // HAS_MIRRORROW_SSSE3
-
-#ifdef HAS_MIRRORROW_AVX2
-// Shuffle table for reversing the bytes.
-static const ulvec8 kShuffleMirror_AVX2 = {
-  15u, 14u, 13u, 12u, 11u, 10u, 9u, 8u, 7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u,
-  15u, 14u, 13u, 12u, 11u, 10u, 9u, 8u, 7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u
-};
-
-__declspec(naked) __declspec(align(16))
-void MirrorRow_AVX2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov       eax, [esp + 4]   // src
-    mov       edx, [esp + 8]   // dst
-    mov       ecx, [esp + 12]  // width
-    vmovdqa   ymm5, kShuffleMirror_AVX2
-    lea       eax, [eax - 32]
-
-    align      4
- convertloop:
-    vmovdqu   ymm0, [eax + ecx]
-    vpshufb   ymm0, ymm0, ymm5
-    vpermq    ymm0, ymm0, 0x4e  // swap high and low halfs
-    sub       ecx, 32
-    vmovdqu   [edx], ymm0
-    lea       edx, [edx + 32]
-    jg        convertloop
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_MIRRORROW_AVX2
-
-#ifdef HAS_MIRRORROW_SSE2
-// SSE2 version has movdqu so it can be used on unaligned buffers when SSSE3
-// version can not.
-__declspec(naked) __declspec(align(16))
-void MirrorRow_SSE2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov       eax, [esp + 4]   // src
-    mov       edx, [esp + 8]   // dst
-    mov       ecx, [esp + 12]  // width
-    lea       eax, [eax - 16]
-
-    align      4
- convertloop:
-    movdqu    xmm0, [eax + ecx]
-    movdqa    xmm1, xmm0        // swap bytes
-    psllw     xmm0, 8
-    psrlw     xmm1, 8
-    por       xmm0, xmm1
-    pshuflw   xmm0, xmm0, 0x1b  // swap words
-    pshufhw   xmm0, xmm0, 0x1b
-    pshufd    xmm0, xmm0, 0x4e  // swap qwords
-    sub       ecx, 16
-    movdqu    [edx], xmm0
-    lea       edx, [edx + 16]
-    jg        convertloop
-    ret
-  }
-}
-#endif  // HAS_MIRRORROW_SSE2
-
-#ifdef HAS_MIRRORROW_UV_SSSE3
-// Shuffle table for reversing the bytes of UV channels.
-static const uvec8 kShuffleMirrorUV = {
-  14u, 12u, 10u, 8u, 6u, 4u, 2u, 0u, 15u, 13u, 11u, 9u, 7u, 5u, 3u, 1u
-};
-
-__declspec(naked) __declspec(align(16))
-void MirrorUVRow_SSSE3(const uint8* src, uint8* dst_u, uint8* dst_v,
-                       int width) {
-  __asm {
-    push      edi
-    mov       eax, [esp + 4 + 4]   // src
-    mov       edx, [esp + 4 + 8]   // dst_u
-    mov       edi, [esp + 4 + 12]  // dst_v
-    mov       ecx, [esp + 4 + 16]  // width
-    movdqa    xmm1, kShuffleMirrorUV
-    lea       eax, [eax + ecx * 2 - 16]
-    sub       edi, edx
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax]
-    lea       eax, [eax - 16]
-    pshufb    xmm0, xmm1
-    sub       ecx, 8
-    movlpd    qword ptr [edx], xmm0
-    movhpd    qword ptr [edx + edi], xmm0
-    lea       edx, [edx + 8]
-    jg        convertloop
-
-    pop       edi
-    ret
-  }
-}
-#endif  // HAS_MIRRORROW_UV_SSSE3
-
-#ifdef HAS_ARGBMIRRORROW_SSSE3
-// Shuffle table for reversing the bytes.
-static const uvec8 kARGBShuffleMirror = {
-  12u, 13u, 14u, 15u, 8u, 9u, 10u, 11u, 4u, 5u, 6u, 7u, 0u, 1u, 2u, 3u
-};
-
-__declspec(naked) __declspec(align(16))
-void ARGBMirrorRow_SSSE3(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov       eax, [esp + 4]   // src
-    mov       edx, [esp + 8]   // dst
-    mov       ecx, [esp + 12]  // width
-    lea       eax, [eax - 16 + ecx * 4]  // last 4 pixels.
-    movdqa    xmm5, kARGBShuffleMirror
-
-    align      4
- convertloop:
-    movdqa    xmm0, [eax]
-    lea       eax, [eax - 16]
-    pshufb    xmm0, xmm5
-    sub       ecx, 4
-    movdqa    [edx], xmm0
-    lea       edx, [edx + 16]
-    jg        convertloop
-    ret
-  }
-}
-#endif  // HAS_ARGBMIRRORROW_SSSE3
-
-#ifdef HAS_ARGBMIRRORROW_AVX2
-// Shuffle table for reversing the bytes.
-static const ulvec32 kARGBShuffleMirror_AVX2 = {
-  7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u
-};
-
-__declspec(naked) __declspec(align(16))
-void ARGBMirrorRow_AVX2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov       eax, [esp + 4]   // src
-    mov       edx, [esp + 8]   // dst
-    mov       ecx, [esp + 12]  // width
-    lea       eax, [eax - 32]
-    vmovdqa   ymm5, kARGBShuffleMirror_AVX2
-
-    align      4
- convertloop:
-    vpermd    ymm0, ymm5, [eax + ecx * 4]  // permute dword order
-    sub       ecx, 8
-    vmovdqu   [edx], ymm0
-    lea       edx, [edx + 32]
-    jg        convertloop
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBMIRRORROW_AVX2
-
-#ifdef HAS_SPLITUVROW_SSE2
-__declspec(naked) __declspec(align(16))
-void SplitUVRow_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_uv
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    movdqa     xmm2, xmm0
-    movdqa     xmm3, xmm1
-    pand       xmm0, xmm5   // even bytes
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    psrlw      xmm2, 8      // odd bytes
-    psrlw      xmm3, 8
-    packuswb   xmm2, xmm3
-    movdqa     [edx], xmm0
-    movdqa     [edx + edi], xmm2
-    lea        edx, [edx + 16]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void SplitUVRow_Unaligned_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v,
-                               int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_uv
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    movdqa     xmm2, xmm0
-    movdqa     xmm3, xmm1
-    pand       xmm0, xmm5   // even bytes
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    psrlw      xmm2, 8      // odd bytes
-    psrlw      xmm3, 8
-    packuswb   xmm2, xmm3
-    movdqu     [edx], xmm0
-    movdqu     [edx + edi], xmm2
-    lea        edx, [edx + 16]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-#endif  // HAS_SPLITUVROW_SSE2
-
-#ifdef HAS_SPLITUVROW_AVX2
-__declspec(naked) __declspec(align(16))
-void SplitUVRow_AVX2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_uv
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5      // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax,  [eax + 64]
-    vpsrlw     ymm2, ymm0, 8      // odd bytes
-    vpsrlw     ymm3, ymm1, 8
-    vpand      ymm0, ymm0, ymm5   // even bytes
-    vpand      ymm1, ymm1, ymm5
-    vpackuswb  ymm0, ymm0, ymm1
-    vpackuswb  ymm2, ymm2, ymm3
-    vpermq     ymm0, ymm0, 0xd8
-    vpermq     ymm2, ymm2, 0xd8
-    vmovdqu    [edx], ymm0
-    vmovdqu    [edx + edi], ymm2
-    lea        edx, [edx + 32]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_SPLITUVROW_AVX2
-
-#ifdef HAS_MERGEUVROW_SSE2
-__declspec(naked) __declspec(align(16))
-void MergeUVRow_SSE2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_u
-    mov        edx, [esp + 4 + 8]    // src_v
-    mov        edi, [esp + 4 + 12]   // dst_uv
-    mov        ecx, [esp + 4 + 16]   // width
-    sub        edx, eax
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]      // read 16 U's
-    movdqa     xmm1, [eax + edx]  // and 16 V's
-    lea        eax,  [eax + 16]
-    movdqa     xmm2, xmm0
-    punpcklbw  xmm0, xmm1       // first 8 UV pairs
-    punpckhbw  xmm2, xmm1       // next 8 UV pairs
-    movdqa     [edi], xmm0
-    movdqa     [edi + 16], xmm2
-    lea        edi, [edi + 32]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void MergeUVRow_Unaligned_SSE2(const uint8* src_u, const uint8* src_v,
-                               uint8* dst_uv, int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_u
-    mov        edx, [esp + 4 + 8]    // src_v
-    mov        edi, [esp + 4 + 12]   // dst_uv
-    mov        ecx, [esp + 4 + 16]   // width
-    sub        edx, eax
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]      // read 16 U's
-    movdqu     xmm1, [eax + edx]  // and 16 V's
-    lea        eax,  [eax + 16]
-    movdqa     xmm2, xmm0
-    punpcklbw  xmm0, xmm1       // first 8 UV pairs
-    punpckhbw  xmm2, xmm1       // next 8 UV pairs
-    movdqu     [edi], xmm0
-    movdqu     [edi + 16], xmm2
-    lea        edi, [edi + 32]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-#endif  //  HAS_MERGEUVROW_SSE2
-
-#ifdef HAS_MERGEUVROW_AVX2
-__declspec(naked) __declspec(align(16))
-void MergeUVRow_AVX2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                     int width) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_u
-    mov        edx, [esp + 4 + 8]    // src_v
-    mov        edi, [esp + 4 + 12]   // dst_uv
-    mov        ecx, [esp + 4 + 16]   // width
-    sub        edx, eax
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]           // read 32 U's
-    vmovdqu    ymm1, [eax + edx]     // and 32 V's
-    lea        eax,  [eax + 32]
-    vpunpcklbw ymm2, ymm0, ymm1      // low 16 UV pairs. mutated qqword 0,2
-    vpunpckhbw ymm0, ymm0, ymm1      // high 16 UV pairs. mutated qqword 1,3
-    vperm2i128 ymm1, ymm2, ymm0, 0x20  // low 128 of ymm2 and low 128 of ymm0
-    vperm2i128 ymm2, ymm2, ymm0, 0x31  // high 128 of ymm2 and high 128 of ymm0
-    vmovdqu    [edi], ymm1
-    vmovdqu    [edi + 32], ymm2
-    lea        edi, [edi + 64]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    vzeroupper
-    ret
-  }
-}
-#endif  //  HAS_MERGEUVROW_AVX2
-
-#ifdef HAS_COPYROW_SSE2
-// CopyRow copys 'count' bytes using a 16 byte load/store, 32 bytes at time.
-__declspec(naked) __declspec(align(16))
-void CopyRow_SSE2(const uint8* src, uint8* dst, int count) {
-  __asm {
-    mov        eax, [esp + 4]   // src
-    mov        edx, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax, [eax + 32]
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    sub        ecx, 32
-    jg         convertloop
-    ret
-  }
-}
-#endif  // HAS_COPYROW_SSE2
-
-// Unaligned Multiple of 1.
-__declspec(naked) __declspec(align(16))
-void CopyRow_ERMS(const uint8* src, uint8* dst, int count) {
-  __asm {
-    mov        eax, esi
-    mov        edx, edi
-    mov        esi, [esp + 4]   // src
-    mov        edi, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    rep movsb
-    mov        edi, edx
-    mov        esi, eax
-    ret
-  }
-}
-
-#ifdef HAS_COPYROW_X86
-__declspec(naked) __declspec(align(16))
-void CopyRow_X86(const uint8* src, uint8* dst, int count) {
-  __asm {
-    mov        eax, esi
-    mov        edx, edi
-    mov        esi, [esp + 4]   // src
-    mov        edi, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    shr        ecx, 2
-    rep movsd
-    mov        edi, edx
-    mov        esi, eax
-    ret
-  }
-}
-#endif  // HAS_COPYROW_X86
-
-#ifdef HAS_ARGBCOPYALPHAROW_SSE2
-// width in pixels
-__declspec(naked) __declspec(align(16))
-void ARGBCopyAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src
-    mov        edx, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    pcmpeqb    xmm0, xmm0       // generate mask 0xff000000
-    pslld      xmm0, 24
-    pcmpeqb    xmm1, xmm1       // generate mask 0x00ffffff
-    psrld      xmm1, 8
-
-    align      4
-  convertloop:
-    movdqa     xmm2, [eax]
-    movdqa     xmm3, [eax + 16]
-    lea        eax, [eax + 32]
-    movdqa     xmm4, [edx]
-    movdqa     xmm5, [edx + 16]
-    pand       xmm2, xmm0
-    pand       xmm3, xmm0
-    pand       xmm4, xmm1
-    pand       xmm5, xmm1
-    por        xmm2, xmm4
-    por        xmm3, xmm5
-    movdqa     [edx], xmm2
-    movdqa     [edx + 16], xmm3
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_ARGBCOPYALPHAROW_SSE2
-
-#ifdef HAS_ARGBCOPYALPHAROW_AVX2
-// width in pixels
-__declspec(naked) __declspec(align(16))
-void ARGBCopyAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src
-    mov        edx, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    vpcmpeqb   ymm0, ymm0, ymm0
-    vpsrld     ymm0, ymm0, 8    // generate mask 0x00ffffff
-
-    align      4
-  convertloop:
-    vmovdqu    ymm1, [eax]
-    vmovdqu    ymm2, [eax + 32]
-    lea        eax, [eax + 64]
-    vpblendvb  ymm1, ymm1, [edx], ymm0
-    vpblendvb  ymm2, ymm2, [edx + 32], ymm0
-    vmovdqu    [edx], ymm1
-    vmovdqu    [edx + 32], ymm2
-    lea        edx, [edx + 64]
-    sub        ecx, 16
-    jg         convertloop
-
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBCOPYALPHAROW_AVX2
-
-#ifdef HAS_ARGBCOPYYTOALPHAROW_SSE2
-// width in pixels
-__declspec(naked) __declspec(align(16))
-void ARGBCopyYToAlphaRow_SSE2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src
-    mov        edx, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    pcmpeqb    xmm0, xmm0       // generate mask 0xff000000
-    pslld      xmm0, 24
-    pcmpeqb    xmm1, xmm1       // generate mask 0x00ffffff
-    psrld      xmm1, 8
-
-    align      4
-  convertloop:
-    movq       xmm2, qword ptr [eax]  // 8 Y's
-    lea        eax, [eax + 8]
-    punpcklbw  xmm2, xmm2
-    punpckhwd  xmm3, xmm2
-    punpcklwd  xmm2, xmm2
-    movdqa     xmm4, [edx]
-    movdqa     xmm5, [edx + 16]
-    pand       xmm2, xmm0
-    pand       xmm3, xmm0
-    pand       xmm4, xmm1
-    pand       xmm5, xmm1
-    por        xmm2, xmm4
-    por        xmm3, xmm5
-    movdqa     [edx], xmm2
-    movdqa     [edx + 16], xmm3
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_ARGBCOPYYTOALPHAROW_SSE2
-
-#ifdef HAS_ARGBCOPYYTOALPHAROW_AVX2
-// width in pixels
-__declspec(naked) __declspec(align(16))
-void ARGBCopyYToAlphaRow_AVX2(const uint8* src, uint8* dst, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src
-    mov        edx, [esp + 8]   // dst
-    mov        ecx, [esp + 12]  // count
-    vpcmpeqb   ymm0, ymm0, ymm0
-    vpsrld     ymm0, ymm0, 8    // generate mask 0x00ffffff
-
-    align      4
-  convertloop:
-    vpmovzxbd  ymm1, qword ptr [eax]
-    vpmovzxbd  ymm2, qword ptr [eax + 8]
-    lea        eax, [eax + 16]
-    vpslld     ymm1, ymm1, 24
-    vpslld     ymm2, ymm2, 24
-    vpblendvb  ymm1, ymm1, [edx], ymm0
-    vpblendvb  ymm2, ymm2, [edx + 32], ymm0
-    vmovdqu    [edx], ymm1
-    vmovdqu    [edx + 32], ymm2
-    lea        edx, [edx + 64]
-    sub        ecx, 16
-    jg         convertloop
-
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBCOPYYTOALPHAROW_AVX2
-
-#ifdef HAS_SETROW_X86
-// SetRow8 writes 'count' bytes using a 32 bit value repeated.
-__declspec(naked) __declspec(align(16))
-void SetRow_X86(uint8* dst, uint32 v32, int count) {
-  __asm {
-    mov        edx, edi
-    mov        edi, [esp + 4]   // dst
-    mov        eax, [esp + 8]   // v32
-    mov        ecx, [esp + 12]  // count
-    shr        ecx, 2
-    rep stosd
-    mov        edi, edx
-    ret
-  }
-}
-
-// SetRow32 writes 'count' words using a 32 bit value repeated.
-__declspec(naked) __declspec(align(16))
-void ARGBSetRows_X86(uint8* dst, uint32 v32, int width,
-                   int dst_stride, int height) {
-  __asm {
-    push       esi
-    push       edi
-    push       ebp
-    mov        edi, [esp + 12 + 4]   // dst
-    mov        eax, [esp + 12 + 8]   // v32
-    mov        ebp, [esp + 12 + 12]  // width
-    mov        edx, [esp + 12 + 16]  // dst_stride
-    mov        esi, [esp + 12 + 20]  // height
-    lea        ecx, [ebp * 4]
-    sub        edx, ecx             // stride - width * 4
-
-    align      4
-  convertloop:
-    mov        ecx, ebp
-    rep stosd
-    add        edi, edx
-    sub        esi, 1
-    jg         convertloop
-
-    pop        ebp
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SETROW_X86
-
-#ifdef HAS_YUY2TOYROW_AVX2
-__declspec(naked) __declspec(align(16))
-void YUY2ToYRow_AVX2(const uint8* src_yuy2,
-                     uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_yuy2
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5  // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax,  [eax + 64]
-    vpand      ymm0, ymm0, ymm5   // even bytes are Y
-    vpand      ymm1, ymm1, ymm5
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    sub        ecx, 32
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    jg         convertloop
-    vzeroupper
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUVRow_AVX2(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5      // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    vpavgb     ymm0, ymm0, [eax + esi]
-    vpavgb     ymm1, ymm1, [eax + esi + 32]
-    lea        eax,  [eax + 64]
-    vpsrlw     ymm0, ymm0, 8      // YUYV -> UVUV
-    vpsrlw     ymm1, ymm1, 8
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    vpand      ymm1, ymm0, ymm5  // U
-    vpsrlw     ymm0, ymm0, 8     // V
-    vpackuswb  ymm1, ymm1, ymm1  // mutates.
-    vpackuswb  ymm0, ymm0, ymm0  // mutates.
-    vpermq     ymm1, ymm1, 0xd8
-    vpermq     ymm0, ymm0, 0xd8
-    vextractf128 [edx], ymm1, 0  // U
-    vextractf128 [edx + edi], ymm0, 0 // V
-    lea        edx, [edx + 16]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUV422Row_AVX2(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5      // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax,  [eax + 64]
-    vpsrlw     ymm0, ymm0, 8      // YUYV -> UVUV
-    vpsrlw     ymm1, ymm1, 8
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    vpand      ymm1, ymm0, ymm5  // U
-    vpsrlw     ymm0, ymm0, 8     // V
-    vpackuswb  ymm1, ymm1, ymm1  // mutates.
-    vpackuswb  ymm0, ymm0, ymm0  // mutates.
-    vpermq     ymm1, ymm1, 0xd8
-    vpermq     ymm0, ymm0, 0xd8
-    vextractf128 [edx], ymm1, 0  // U
-    vextractf128 [edx + edi], ymm0, 0 // V
-    lea        edx, [edx + 16]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    vzeroupper
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToYRow_AVX2(const uint8* src_uyvy,
-                     uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_uyvy
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax,  [eax + 64]
-    vpsrlw     ymm0, ymm0, 8      // odd bytes are Y
-    vpsrlw     ymm1, ymm1, 8
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    sub        ecx, 32
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    jg         convertloop
-    ret
-    vzeroupper
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUVRow_AVX2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5      // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    vpavgb     ymm0, ymm0, [eax + esi]
-    vpavgb     ymm1, ymm1, [eax + esi + 32]
-    lea        eax,  [eax + 64]
-    vpand      ymm0, ymm0, ymm5   // UYVY -> UVUV
-    vpand      ymm1, ymm1, ymm5
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    vpand      ymm1, ymm0, ymm5  // U
-    vpsrlw     ymm0, ymm0, 8     // V
-    vpackuswb  ymm1, ymm1, ymm1  // mutates.
-    vpackuswb  ymm0, ymm0, ymm0  // mutates.
-    vpermq     ymm1, ymm1, 0xd8
-    vpermq     ymm0, ymm0, 0xd8
-    vextractf128 [edx], ymm1, 0  // U
-    vextractf128 [edx + edi], ymm0, 0 // V
-    lea        edx, [edx + 16]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUV422Row_AVX2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    vpcmpeqb   ymm5, ymm5, ymm5      // generate mask 0x00ff00ff
-    vpsrlw     ymm5, ymm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax,  [eax + 64]
-    vpand      ymm0, ymm0, ymm5   // UYVY -> UVUV
-    vpand      ymm1, ymm1, ymm5
-    vpackuswb  ymm0, ymm0, ymm1   // mutates.
-    vpermq     ymm0, ymm0, 0xd8
-    vpand      ymm1, ymm0, ymm5  // U
-    vpsrlw     ymm0, ymm0, 8     // V
-    vpackuswb  ymm1, ymm1, ymm1  // mutates.
-    vpackuswb  ymm0, ymm0, ymm0  // mutates.
-    vpermq     ymm1, ymm1, 0xd8
-    vpermq     ymm0, ymm0, 0xd8
-    vextractf128 [edx], ymm1, 0  // U
-    vextractf128 [edx + edi], ymm0, 0 // V
-    lea        edx, [edx + 16]
-    sub        ecx, 32
-    jg         convertloop
-
-    pop        edi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_YUY2TOYROW_AVX2
-
-#ifdef HAS_YUY2TOYROW_SSE2
-__declspec(naked) __declspec(align(16))
-void YUY2ToYRow_SSE2(const uint8* src_yuy2,
-                     uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_yuy2
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-    pcmpeqb    xmm5, xmm5        // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    pand       xmm0, xmm5   // even bytes are Y
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUVRow_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + esi]
-    movdqa     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2
-    pavgb      xmm1, xmm3
-    psrlw      xmm0, 8      // YUYV -> UVUV
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUV422Row_SSE2(const uint8* src_yuy2,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8      // YUYV -> UVUV
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToYRow_Unaligned_SSE2(const uint8* src_yuy2,
-                               uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_yuy2
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-    pcmpeqb    xmm5, xmm5        // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    pand       xmm0, xmm5   // even bytes are Y
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUVRow_Unaligned_SSE2(const uint8* src_yuy2, int stride_yuy2,
-                                uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + esi]
-    movdqu     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2
-    pavgb      xmm1, xmm3
-    psrlw      xmm0, 8      // YUYV -> UVUV
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void YUY2ToUV422Row_Unaligned_SSE2(const uint8* src_yuy2,
-                                   uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8      // YUYV -> UVUV
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToYRow_SSE2(const uint8* src_uyvy,
-                     uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_uyvy
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8    // odd bytes are Y
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUVRow_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                      uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + esi]
-    movdqa     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2
-    pavgb      xmm1, xmm3
-    pand       xmm0, xmm5   // UYVY -> UVUV
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUV422Row_SSE2(const uint8* src_uyvy,
-                         uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    pand       xmm0, xmm5   // UYVY -> UVUV
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToYRow_Unaligned_SSE2(const uint8* src_uyvy,
-                               uint8* dst_y, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_uyvy
-    mov        edx, [esp + 8]    // dst_y
-    mov        ecx, [esp + 12]   // pix
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8    // odd bytes are Y
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUVRow_Unaligned_SSE2(const uint8* src_uyvy, int stride_uyvy,
-                                uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_yuy2
-    mov        esi, [esp + 8 + 8]    // stride_yuy2
-    mov        edx, [esp + 8 + 12]   // dst_u
-    mov        edi, [esp + 8 + 16]   // dst_v
-    mov        ecx, [esp + 8 + 20]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + esi]
-    movdqu     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2
-    pavgb      xmm1, xmm3
-    pand       xmm0, xmm5   // UYVY -> UVUV
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void UYVYToUV422Row_Unaligned_SSE2(const uint8* src_uyvy,
-                                   uint8* dst_u, uint8* dst_v, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_yuy2
-    mov        edx, [esp + 4 + 8]    // dst_u
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-    sub        edi, edx
-
-    align      4
-  convertloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    pand       xmm0, xmm5   // UYVY -> UVUV
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    movdqa     xmm1, xmm0
-    pand       xmm0, xmm5  // U
-    packuswb   xmm0, xmm0
-    psrlw      xmm1, 8     // V
-    packuswb   xmm1, xmm1
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + edi], xmm1
-    lea        edx, [edx + 8]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    ret
-  }
-}
-#endif  // HAS_YUY2TOYROW_SSE2
-
-#ifdef HAS_ARGBBLENDROW_SSE2
-// Blend 8 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBBlendRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                       uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm7, xmm7       // generate constant 1
-    psrlw      xmm7, 15
-    pcmpeqb    xmm6, xmm6       // generate mask 0x00ff00ff
-    psrlw      xmm6, 8
-    pcmpeqb    xmm5, xmm5       // generate mask 0xff00ff00
-    psllw      xmm5, 8
-    pcmpeqb    xmm4, xmm4       // generate mask 0xff000000
-    pslld      xmm4, 24
-
-    sub        ecx, 1
-    je         convertloop1     // only 1 pixel?
-    jl         convertloop1b
-
-    // 1 pixel loop until destination pointer is aligned.
-  alignloop1:
-    test       edx, 15          // aligned?
-    je         alignloop1b
-    movd       xmm3, [eax]
-    lea        eax, [eax + 4]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movd       xmm2, [esi]      // _r_b
-    psrlw      xmm3, 8          // alpha
-    pshufhw    xmm3, xmm3, 0F5h // 8 alpha words
-    pshuflw    xmm3, xmm3, 0F5h
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movd       xmm1, [esi]      // _a_g
-    lea        esi, [esi + 4]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        alignloop1
-
-  alignloop1b:
-    add        ecx, 1 - 4
-    jl         convertloop4b
-
-    // 4 pixel loop.
-  convertloop4:
-    movdqu     xmm3, [eax]      // src argb
-    lea        eax, [eax + 16]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movdqu     xmm2, [esi]      // _r_b
-    psrlw      xmm3, 8          // alpha
-    pshufhw    xmm3, xmm3, 0F5h // 8 alpha words
-    pshuflw    xmm3, xmm3, 0F5h
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movdqu     xmm1, [esi]      // _a_g
-    lea        esi, [esi + 16]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jge        convertloop4
-
-  convertloop4b:
-    add        ecx, 4 - 1
-    jl         convertloop1b
-
-    // 1 pixel loop.
-  convertloop1:
-    movd       xmm3, [eax]      // src argb
-    lea        eax, [eax + 4]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movd       xmm2, [esi]      // _r_b
-    psrlw      xmm3, 8          // alpha
-    pshufhw    xmm3, xmm3, 0F5h // 8 alpha words
-    pshuflw    xmm3, xmm3, 0F5h
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movd       xmm1, [esi]      // _a_g
-    lea        esi, [esi + 4]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        convertloop1
-
-  convertloop1b:
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBBLENDROW_SSE2
-
-#ifdef HAS_ARGBBLENDROW_SSSE3
-// Shuffle table for isolating alpha.
-static const uvec8 kShuffleAlpha = {
-  3u, 0x80, 3u, 0x80, 7u, 0x80, 7u, 0x80,
-  11u, 0x80, 11u, 0x80, 15u, 0x80, 15u, 0x80
-};
-// Same as SSE2, but replaces:
-//    psrlw      xmm3, 8          // alpha
-//    pshufhw    xmm3, xmm3, 0F5h // 8 alpha words
-//    pshuflw    xmm3, xmm3, 0F5h
-// with..
-//    pshufb     xmm3, kShuffleAlpha // alpha
-// Blend 8 pixels at a time.
-
-__declspec(naked) __declspec(align(16))
-void ARGBBlendRow_SSSE3(const uint8* src_argb0, const uint8* src_argb1,
-                        uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pcmpeqb    xmm7, xmm7       // generate constant 0x0001
-    psrlw      xmm7, 15
-    pcmpeqb    xmm6, xmm6       // generate mask 0x00ff00ff
-    psrlw      xmm6, 8
-    pcmpeqb    xmm5, xmm5       // generate mask 0xff00ff00
-    psllw      xmm5, 8
-    pcmpeqb    xmm4, xmm4       // generate mask 0xff000000
-    pslld      xmm4, 24
-
-    sub        ecx, 1
-    je         convertloop1     // only 1 pixel?
-    jl         convertloop1b
-
-    // 1 pixel loop until destination pointer is aligned.
-  alignloop1:
-    test       edx, 15          // aligned?
-    je         alignloop1b
-    movd       xmm3, [eax]
-    lea        eax, [eax + 4]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movd       xmm2, [esi]      // _r_b
-    pshufb     xmm3, kShuffleAlpha // alpha
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movd       xmm1, [esi]      // _a_g
-    lea        esi, [esi + 4]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        alignloop1
-
-  alignloop1b:
-    add        ecx, 1 - 4
-    jl         convertloop4b
-
-    test       eax, 15          // unaligned?
-    jne        convertuloop4
-    test       esi, 15          // unaligned?
-    jne        convertuloop4
-
-    // 4 pixel loop.
-  convertloop4:
-    movdqa     xmm3, [eax]      // src argb
-    lea        eax, [eax + 16]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movdqa     xmm2, [esi]      // _r_b
-    pshufb     xmm3, kShuffleAlpha // alpha
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movdqa     xmm1, [esi]      // _a_g
-    lea        esi, [esi + 16]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jge        convertloop4
-    jmp        convertloop4b
-
-    // 4 pixel unaligned loop.
-  convertuloop4:
-    movdqu     xmm3, [eax]      // src argb
-    lea        eax, [eax + 16]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movdqu     xmm2, [esi]      // _r_b
-    pshufb     xmm3, kShuffleAlpha // alpha
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movdqu     xmm1, [esi]      // _a_g
-    lea        esi, [esi + 16]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jge        convertuloop4
-
-  convertloop4b:
-    add        ecx, 4 - 1
-    jl         convertloop1b
-
-    // 1 pixel loop.
-  convertloop1:
-    movd       xmm3, [eax]      // src argb
-    lea        eax, [eax + 4]
-    movdqa     xmm0, xmm3       // src argb
-    pxor       xmm3, xmm4       // ~alpha
-    movd       xmm2, [esi]      // _r_b
-    pshufb     xmm3, kShuffleAlpha // alpha
-    pand       xmm2, xmm6       // _r_b
-    paddw      xmm3, xmm7       // 256 - alpha
-    pmullw     xmm2, xmm3       // _r_b * alpha
-    movd       xmm1, [esi]      // _a_g
-    lea        esi, [esi + 4]
-    psrlw      xmm1, 8          // _a_g
-    por        xmm0, xmm4       // set alpha to 255
-    pmullw     xmm1, xmm3       // _a_g * alpha
-    psrlw      xmm2, 8          // _r_b convert to 8 bits again
-    paddusb    xmm0, xmm2       // + src argb
-    pand       xmm1, xmm5       // a_g_ convert to 8 bits again
-    paddusb    xmm0, xmm1       // + src argb
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        convertloop1
-
-  convertloop1b:
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBBLENDROW_SSSE3
-
-#ifdef HAS_ARGBATTENUATEROW_SSE2
-// Attenuate 4 pixels at a time.
-// Aligned to 16 bytes.
-__declspec(naked) __declspec(align(16))
-void ARGBAttenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src_argb0
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    pcmpeqb    xmm4, xmm4       // generate mask 0xff000000
-    pslld      xmm4, 24
-    pcmpeqb    xmm5, xmm5       // generate mask 0x00ffffff
-    psrld      xmm5, 8
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]      // read 4 pixels
-    punpcklbw  xmm0, xmm0       // first 2
-    pshufhw    xmm2, xmm0, 0FFh // 8 alpha words
-    pshuflw    xmm2, xmm2, 0FFh
-    pmulhuw    xmm0, xmm2       // rgb * a
-    movdqa     xmm1, [eax]      // read 4 pixels
-    punpckhbw  xmm1, xmm1       // next 2 pixels
-    pshufhw    xmm2, xmm1, 0FFh // 8 alpha words
-    pshuflw    xmm2, xmm2, 0FFh
-    pmulhuw    xmm1, xmm2       // rgb * a
-    movdqa     xmm2, [eax]      // alphas
-    lea        eax, [eax + 16]
-    psrlw      xmm0, 8
-    pand       xmm2, xmm4
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    pand       xmm0, xmm5       // keep original alphas
-    por        xmm0, xmm2
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_ARGBATTENUATEROW_SSE2
-
-#ifdef HAS_ARGBATTENUATEROW_SSSE3
-// Shuffle table duplicating alpha.
-static const uvec8 kShuffleAlpha0 = {
-  3u, 3u, 3u, 3u, 3u, 3u, 128u, 128u, 7u, 7u, 7u, 7u, 7u, 7u, 128u, 128u,
-};
-static const uvec8 kShuffleAlpha1 = {
-  11u, 11u, 11u, 11u, 11u, 11u, 128u, 128u,
-  15u, 15u, 15u, 15u, 15u, 15u, 128u, 128u,
-};
-__declspec(naked) __declspec(align(16))
-void ARGBAttenuateRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src_argb0
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    pcmpeqb    xmm3, xmm3       // generate mask 0xff000000
-    pslld      xmm3, 24
-    movdqa     xmm4, kShuffleAlpha0
-    movdqa     xmm5, kShuffleAlpha1
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]      // read 4 pixels
-    pshufb     xmm0, xmm4       // isolate first 2 alphas
-    movdqu     xmm1, [eax]      // read 4 pixels
-    punpcklbw  xmm1, xmm1       // first 2 pixel rgbs
-    pmulhuw    xmm0, xmm1       // rgb * a
-    movdqu     xmm1, [eax]      // read 4 pixels
-    pshufb     xmm1, xmm5       // isolate next 2 alphas
-    movdqu     xmm2, [eax]      // read 4 pixels
-    punpckhbw  xmm2, xmm2       // next 2 pixel rgbs
-    pmulhuw    xmm1, xmm2       // rgb * a
-    movdqu     xmm2, [eax]      // mask original alpha
-    lea        eax, [eax + 16]
-    pand       xmm2, xmm3
-    psrlw      xmm0, 8
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    por        xmm0, xmm2       // copy original alpha
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_ARGBATTENUATEROW_SSSE3
-
-#ifdef HAS_ARGBATTENUATEROW_AVX2
-// Shuffle table duplicating alpha.
-static const ulvec8 kShuffleAlpha_AVX2 = {
-  6u, 7u, 6u, 7u, 6u, 7u, 128u, 128u,
-  14u, 15u, 14u, 15u, 14u, 15u, 128u, 128u,
-  6u, 7u, 6u, 7u, 6u, 7u, 128u, 128u,
-  14u, 15u, 14u, 15u, 14u, 15u, 128u, 128u,
-};
-__declspec(naked) __declspec(align(16))
-void ARGBAttenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src_argb0
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    sub        edx, eax
-    vmovdqa    ymm4, kShuffleAlpha_AVX2
-    vpcmpeqb   ymm5, ymm5, ymm5 // generate mask 0xff000000
-    vpslld     ymm5, ymm5, 24
-
-    align      4
- convertloop:
-    vmovdqu    ymm6, [eax]       // read 8 pixels.
-    vpunpcklbw ymm0, ymm6, ymm6  // low 4 pixels. mutated.
-    vpunpckhbw ymm1, ymm6, ymm6  // high 4 pixels. mutated.
-    vpshufb    ymm2, ymm0, ymm4  // low 4 alphas
-    vpshufb    ymm3, ymm1, ymm4  // high 4 alphas
-    vpmulhuw   ymm0, ymm0, ymm2  // rgb * a
-    vpmulhuw   ymm1, ymm1, ymm3  // rgb * a
-    vpand      ymm6, ymm6, ymm5  // isolate alpha
-    vpsrlw     ymm0, ymm0, 8
-    vpsrlw     ymm1, ymm1, 8
-    vpackuswb  ymm0, ymm0, ymm1  // unmutated.
-    vpor       ymm0, ymm0, ymm6  // copy original alpha
-    sub        ecx, 8
-    vmovdqu    [eax + edx], ymm0
-    lea        eax, [eax + 32]
-    jg         convertloop
-
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBATTENUATEROW_AVX2
-
-#ifdef HAS_ARGBUNATTENUATEROW_SSE2
-// Unattenuate 4 pixels at a time.
-// Aligned to 16 bytes.
-__declspec(naked) __declspec(align(16))
-void ARGBUnattenuateRow_SSE2(const uint8* src_argb, uint8* dst_argb,
-                             int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_argb0
-    mov        edx, [esp + 8 + 8]   // dst_argb
-    mov        ecx, [esp + 8 + 12]  // width
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]      // read 4 pixels
-    movzx      esi, byte ptr [eax + 3]  // first alpha
-    movzx      edi, byte ptr [eax + 7]  // second alpha
-    punpcklbw  xmm0, xmm0       // first 2
-    movd       xmm2, dword ptr fixed_invtbl8[esi * 4]
-    movd       xmm3, dword ptr fixed_invtbl8[edi * 4]
-    pshuflw    xmm2, xmm2, 040h // first 4 inv_alpha words.  1, a, a, a
-    pshuflw    xmm3, xmm3, 040h // next 4 inv_alpha words
-    movlhps    xmm2, xmm3
-    pmulhuw    xmm0, xmm2       // rgb * a
-
-    movdqu     xmm1, [eax]      // read 4 pixels
-    movzx      esi, byte ptr [eax + 11]  // third alpha
-    movzx      edi, byte ptr [eax + 15]  // forth alpha
-    punpckhbw  xmm1, xmm1       // next 2
-    movd       xmm2, dword ptr fixed_invtbl8[esi * 4]
-    movd       xmm3, dword ptr fixed_invtbl8[edi * 4]
-    pshuflw    xmm2, xmm2, 040h // first 4 inv_alpha words
-    pshuflw    xmm3, xmm3, 040h // next 4 inv_alpha words
-    movlhps    xmm2, xmm3
-    pmulhuw    xmm1, xmm2       // rgb * a
-    lea        eax, [eax + 16]
-
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBUNATTENUATEROW_SSE2
-
-#ifdef HAS_ARGBUNATTENUATEROW_AVX2
-// Shuffle table duplicating alpha.
-static const ulvec8 kUnattenShuffleAlpha_AVX2 = {
-  0u, 1u, 0u, 1u, 0u, 1u, 6u, 7u, 8u, 9u, 8u, 9u, 8u, 9u, 14u, 15,
-  0u, 1u, 0u, 1u, 0u, 1u, 6u, 7u, 8u, 9u, 8u, 9u, 8u, 9u, 14u, 15,
-};
-// TODO(fbarchard): Enable USE_GATHER for future hardware if faster.
-// USE_GATHER is not on by default, due to being a slow instruction.
-#ifdef USE_GATHER
-__declspec(naked) __declspec(align(16))
-void ARGBUnattenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb,
-                             int width) {
-  __asm {
-    mov        eax, [esp + 4]   // src_argb0
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    sub        edx, eax
-    vmovdqa    ymm4, kUnattenShuffleAlpha_AVX2
-
-    align      4
- convertloop:
-    vmovdqu    ymm6, [eax]       // read 8 pixels.
-    vpcmpeqb   ymm5, ymm5, ymm5  // generate mask 0xffffffff for gather.
-    vpsrld     ymm2, ymm6, 24    // alpha in low 8 bits.
-    vpunpcklbw ymm0, ymm6, ymm6  // low 4 pixels. mutated.
-    vpunpckhbw ymm1, ymm6, ymm6  // high 4 pixels. mutated.
-    vpgatherdd ymm3, [ymm2 * 4 + fixed_invtbl8], ymm5  // ymm5 cleared.  1, a
-    vpunpcklwd ymm2, ymm3, ymm3  // low 4 inverted alphas. mutated. 1, 1, a, a
-    vpunpckhwd ymm3, ymm3, ymm3  // high 4 inverted alphas. mutated.
-    vpshufb    ymm2, ymm2, ymm4  // replicate low 4 alphas. 1, a, a, a
-    vpshufb    ymm3, ymm3, ymm4  // replicate high 4 alphas
-    vpmulhuw   ymm0, ymm0, ymm2  // rgb * ia
-    vpmulhuw   ymm1, ymm1, ymm3  // rgb * ia
-    vpackuswb  ymm0, ymm0, ymm1  // unmutated.
-    sub        ecx, 8
-    vmovdqu    [eax + edx], ymm0
-    lea        eax, [eax + 32]
-    jg         convertloop
-
-    vzeroupper
-    ret
-  }
-}
-#else  // USE_GATHER
-__declspec(naked) __declspec(align(16))
-void ARGBUnattenuateRow_AVX2(const uint8* src_argb, uint8* dst_argb,
-                             int width) {
-  __asm {
-
-    mov        eax, [esp + 4]   // src_argb0
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    sub        edx, eax
-    vmovdqa    ymm5, kUnattenShuffleAlpha_AVX2
-
-    push       esi
-    push       edi
-
-    align      4
- convertloop:
-    // replace VPGATHER
-    movzx      esi, byte ptr [eax + 3]                 // alpha0
-    movzx      edi, byte ptr [eax + 7]                 // alpha1
-    vmovd      xmm0, dword ptr fixed_invtbl8[esi * 4]  // [1,a0]
-    vmovd      xmm1, dword ptr fixed_invtbl8[edi * 4]  // [1,a1]
-    movzx      esi, byte ptr [eax + 11]                // alpha2
-    movzx      edi, byte ptr [eax + 15]                // alpha3
-    vpunpckldq xmm6, xmm0, xmm1                        // [1,a1,1,a0]
-    vmovd      xmm2, dword ptr fixed_invtbl8[esi * 4]  // [1,a2]
-    vmovd      xmm3, dword ptr fixed_invtbl8[edi * 4]  // [1,a3]
-    movzx      esi, byte ptr [eax + 19]                // alpha4
-    movzx      edi, byte ptr [eax + 23]                // alpha5
-    vpunpckldq xmm7, xmm2, xmm3                        // [1,a3,1,a2]
-    vmovd      xmm0, dword ptr fixed_invtbl8[esi * 4]  // [1,a4]
-    vmovd      xmm1, dword ptr fixed_invtbl8[edi * 4]  // [1,a5]
-    movzx      esi, byte ptr [eax + 27]                // alpha6
-    movzx      edi, byte ptr [eax + 31]                // alpha7
-    vpunpckldq xmm0, xmm0, xmm1                        // [1,a5,1,a4]
-    vmovd      xmm2, dword ptr fixed_invtbl8[esi * 4]  // [1,a6]
-    vmovd      xmm3, dword ptr fixed_invtbl8[edi * 4]  // [1,a7]
-    vpunpckldq xmm2, xmm2, xmm3                        // [1,a7,1,a6]
-    vpunpcklqdq xmm3, xmm6, xmm7                       // [1,a3,1,a2,1,a1,1,a0]
-    vpunpcklqdq xmm0, xmm0, xmm2                       // [1,a7,1,a6,1,a5,1,a4]
-    vinserti128 ymm3, ymm3, xmm0, 1 // [1,a7,1,a6,1,a5,1,a4,1,a3,1,a2,1,a1,1,a0]
-    // end of VPGATHER
-
-    vmovdqu    ymm6, [eax]       // read 8 pixels.
-    vpunpcklbw ymm0, ymm6, ymm6  // low 4 pixels. mutated.
-    vpunpckhbw ymm1, ymm6, ymm6  // high 4 pixels. mutated.
-    vpunpcklwd ymm2, ymm3, ymm3  // low 4 inverted alphas. mutated. 1, 1, a, a
-    vpunpckhwd ymm3, ymm3, ymm3  // high 4 inverted alphas. mutated.
-    vpshufb    ymm2, ymm2, ymm5  // replicate low 4 alphas. 1, a, a, a
-    vpshufb    ymm3, ymm3, ymm5  // replicate high 4 alphas
-    vpmulhuw   ymm0, ymm0, ymm2  // rgb * ia
-    vpmulhuw   ymm1, ymm1, ymm3  // rgb * ia
-    vpackuswb  ymm0, ymm0, ymm1  // unmutated.
-    sub        ecx, 8
-    vmovdqu    [eax + edx], ymm0
-    lea        eax, [eax + 32]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // USE_GATHER
-#endif  // HAS_ARGBATTENUATEROW_AVX2
-
-#ifdef HAS_ARGBGRAYROW_SSSE3
-// Convert 8 ARGB pixels (64 bytes) to 8 Gray ARGB pixels.
-__declspec(naked) __declspec(align(16))
-void ARGBGrayRow_SSSE3(const uint8* src_argb, uint8* dst_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_argb */
-    mov        ecx, [esp + 12]  /* width */
-    movdqa     xmm4, kARGBToYJ
-    movdqa     xmm5, kAddYJ64
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]  // G
-    movdqa     xmm1, [eax + 16]
-    pmaddubsw  xmm0, xmm4
-    pmaddubsw  xmm1, xmm4
-    phaddw     xmm0, xmm1
-    paddw      xmm0, xmm5  // Add .5 for rounding.
-    psrlw      xmm0, 7
-    packuswb   xmm0, xmm0   // 8 G bytes
-    movdqa     xmm2, [eax]  // A
-    movdqa     xmm3, [eax + 16]
-    lea        eax, [eax + 32]
-    psrld      xmm2, 24
-    psrld      xmm3, 24
-    packuswb   xmm2, xmm3
-    packuswb   xmm2, xmm2   // 8 A bytes
-    movdqa     xmm3, xmm0   // Weave into GG, GA, then GGGA
-    punpcklbw  xmm0, xmm0   // 8 GG words
-    punpcklbw  xmm3, xmm2   // 8 GA words
-    movdqa     xmm1, xmm0
-    punpcklwd  xmm0, xmm3   // GGGA first 4
-    punpckhwd  xmm1, xmm3   // GGGA next 4
-    sub        ecx, 8
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    jg         convertloop
-    ret
-  }
-}
-#endif  // HAS_ARGBGRAYROW_SSSE3
-
-#ifdef HAS_ARGBSEPIAROW_SSSE3
-//    b = (r * 35 + g * 68 + b * 17) >> 7
-//    g = (r * 45 + g * 88 + b * 22) >> 7
-//    r = (r * 50 + g * 98 + b * 24) >> 7
-// Constant for ARGB color to sepia tone.
-static const vec8 kARGBToSepiaB = {
-  17, 68, 35, 0, 17, 68, 35, 0, 17, 68, 35, 0, 17, 68, 35, 0
-};
-
-static const vec8 kARGBToSepiaG = {
-  22, 88, 45, 0, 22, 88, 45, 0, 22, 88, 45, 0, 22, 88, 45, 0
-};
-
-static const vec8 kARGBToSepiaR = {
-  24, 98, 50, 0, 24, 98, 50, 0, 24, 98, 50, 0, 24, 98, 50, 0
-};
-
-// Convert 8 ARGB pixels (32 bytes) to 8 Sepia ARGB pixels.
-__declspec(naked) __declspec(align(16))
-void ARGBSepiaRow_SSSE3(uint8* dst_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   /* dst_argb */
-    mov        ecx, [esp + 8]   /* width */
-    movdqa     xmm2, kARGBToSepiaB
-    movdqa     xmm3, kARGBToSepiaG
-    movdqa     xmm4, kARGBToSepiaR
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]  // B
-    movdqa     xmm6, [eax + 16]
-    pmaddubsw  xmm0, xmm2
-    pmaddubsw  xmm6, xmm2
-    phaddw     xmm0, xmm6
-    psrlw      xmm0, 7
-    packuswb   xmm0, xmm0   // 8 B values
-    movdqa     xmm5, [eax]  // G
-    movdqa     xmm1, [eax + 16]
-    pmaddubsw  xmm5, xmm3
-    pmaddubsw  xmm1, xmm3
-    phaddw     xmm5, xmm1
-    psrlw      xmm5, 7
-    packuswb   xmm5, xmm5   // 8 G values
-    punpcklbw  xmm0, xmm5   // 8 BG values
-    movdqa     xmm5, [eax]  // R
-    movdqa     xmm1, [eax + 16]
-    pmaddubsw  xmm5, xmm4
-    pmaddubsw  xmm1, xmm4
-    phaddw     xmm5, xmm1
-    psrlw      xmm5, 7
-    packuswb   xmm5, xmm5   // 8 R values
-    movdqa     xmm6, [eax]  // A
-    movdqa     xmm1, [eax + 16]
-    psrld      xmm6, 24
-    psrld      xmm1, 24
-    packuswb   xmm6, xmm1
-    packuswb   xmm6, xmm6   // 8 A values
-    punpcklbw  xmm5, xmm6   // 8 RA values
-    movdqa     xmm1, xmm0   // Weave BG, RA together
-    punpcklwd  xmm0, xmm5   // BGRA first 4
-    punpckhwd  xmm1, xmm5   // BGRA next 4
-    sub        ecx, 8
-    movdqa     [eax], xmm0
-    movdqa     [eax + 16], xmm1
-    lea        eax, [eax + 32]
-    jg         convertloop
-    ret
-  }
-}
-#endif  // HAS_ARGBSEPIAROW_SSSE3
-
-#ifdef HAS_ARGBCOLORMATRIXROW_SSSE3
-// Tranform 8 ARGB pixels (32 bytes) with color matrix.
-// Same as Sepia except matrix is provided.
-// TODO(fbarchard): packuswbs only use half of the reg. To make RGBA, combine R
-// and B into a high and low, then G/A, unpackl/hbw and then unpckl/hwd.
-__declspec(naked) __declspec(align(16))
-void ARGBColorMatrixRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                              const int8* matrix_argb, int width) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_argb */
-    mov        ecx, [esp + 12]  /* matrix_argb */
-    movdqu     xmm5, [ecx]
-    pshufd     xmm2, xmm5, 0x00
-    pshufd     xmm3, xmm5, 0x55
-    pshufd     xmm4, xmm5, 0xaa
-    pshufd     xmm5, xmm5, 0xff
-    mov        ecx, [esp + 16]  /* width */
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]  // B
-    movdqa     xmm7, [eax + 16]
-    pmaddubsw  xmm0, xmm2
-    pmaddubsw  xmm7, xmm2
-    movdqa     xmm6, [eax]  // G
-    movdqa     xmm1, [eax + 16]
-    pmaddubsw  xmm6, xmm3
-    pmaddubsw  xmm1, xmm3
-    phaddsw    xmm0, xmm7   // B
-    phaddsw    xmm6, xmm1   // G
-    psraw      xmm0, 6      // B
-    psraw      xmm6, 6      // G
-    packuswb   xmm0, xmm0   // 8 B values
-    packuswb   xmm6, xmm6   // 8 G values
-    punpcklbw  xmm0, xmm6   // 8 BG values
-    movdqa     xmm1, [eax]  // R
-    movdqa     xmm7, [eax + 16]
-    pmaddubsw  xmm1, xmm4
-    pmaddubsw  xmm7, xmm4
-    phaddsw    xmm1, xmm7   // R
-    movdqa     xmm6, [eax]  // A
-    movdqa     xmm7, [eax + 16]
-    pmaddubsw  xmm6, xmm5
-    pmaddubsw  xmm7, xmm5
-    phaddsw    xmm6, xmm7   // A
-    psraw      xmm1, 6      // R
-    psraw      xmm6, 6      // A
-    packuswb   xmm1, xmm1   // 8 R values
-    packuswb   xmm6, xmm6   // 8 A values
-    punpcklbw  xmm1, xmm6   // 8 RA values
-    movdqa     xmm6, xmm0   // Weave BG, RA together
-    punpcklwd  xmm0, xmm1   // BGRA first 4
-    punpckhwd  xmm6, xmm1   // BGRA next 4
-    sub        ecx, 8
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm6
-    lea        eax, [eax + 32]
-    lea        edx, [edx + 32]
-    jg         convertloop
-    ret
-  }
-}
-#endif  // HAS_ARGBCOLORMATRIXROW_SSSE3
-
-#ifdef HAS_ARGBQUANTIZEROW_SSE2
-// Quantize 4 ARGB pixels (16 bytes).
-// Aligned to 16 bytes.
-__declspec(naked) __declspec(align(16))
-void ARGBQuantizeRow_SSE2(uint8* dst_argb, int scale, int interval_size,
-                          int interval_offset, int width) {
-  __asm {
-    mov        eax, [esp + 4]    /* dst_argb */
-    movd       xmm2, [esp + 8]   /* scale */
-    movd       xmm3, [esp + 12]  /* interval_size */
-    movd       xmm4, [esp + 16]  /* interval_offset */
-    mov        ecx, [esp + 20]   /* width */
-    pshuflw    xmm2, xmm2, 040h
-    pshufd     xmm2, xmm2, 044h
-    pshuflw    xmm3, xmm3, 040h
-    pshufd     xmm3, xmm3, 044h
-    pshuflw    xmm4, xmm4, 040h
-    pshufd     xmm4, xmm4, 044h
-    pxor       xmm5, xmm5  // constant 0
-    pcmpeqb    xmm6, xmm6  // generate mask 0xff000000
-    pslld      xmm6, 24
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]  // read 4 pixels
-    punpcklbw  xmm0, xmm5   // first 2 pixels
-    pmulhuw    xmm0, xmm2   // pixel * scale >> 16
-    movdqa     xmm1, [eax]  // read 4 pixels
-    punpckhbw  xmm1, xmm5   // next 2 pixels
-    pmulhuw    xmm1, xmm2
-    pmullw     xmm0, xmm3   // * interval_size
-    movdqa     xmm7, [eax]  // read 4 pixels
-    pmullw     xmm1, xmm3
-    pand       xmm7, xmm6   // mask alpha
-    paddw      xmm0, xmm4   // + interval_size / 2
-    paddw      xmm1, xmm4
-    packuswb   xmm0, xmm1
-    por        xmm0, xmm7
-    sub        ecx, 4
-    movdqa     [eax], xmm0
-    lea        eax, [eax + 16]
-    jg         convertloop
-    ret
-  }
-}
-#endif  // HAS_ARGBQUANTIZEROW_SSE2
-
-#ifdef HAS_ARGBSHADEROW_SSE2
-// Shade 4 pixels at a time by specified value.
-// Aligned to 16 bytes.
-__declspec(naked) __declspec(align(16))
-void ARGBShadeRow_SSE2(const uint8* src_argb, uint8* dst_argb, int width,
-                       uint32 value) {
-  __asm {
-    mov        eax, [esp + 4]   // src_argb
-    mov        edx, [esp + 8]   // dst_argb
-    mov        ecx, [esp + 12]  // width
-    movd       xmm2, [esp + 16]  // value
-    punpcklbw  xmm2, xmm2
-    punpcklqdq xmm2, xmm2
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]      // read 4 pixels
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm0       // first 2
-    punpckhbw  xmm1, xmm1       // next 2
-    pmulhuw    xmm0, xmm2       // argb * value
-    pmulhuw    xmm1, xmm2       // argb * value
-    psrlw      xmm0, 8
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    ret
-  }
-}
-#endif  // HAS_ARGBSHADEROW_SSE2
-
-#ifdef HAS_ARGBMULTIPLYROW_SSE2
-// Multiply 2 rows of ARGB pixels together, 4 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBMultiplyRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    pxor       xmm5, xmm5  // constant 0
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]        // read 4 pixels from src_argb0
-    movdqu     xmm2, [esi]        // read 4 pixels from src_argb1
-    movdqu     xmm1, xmm0
-    movdqu     xmm3, xmm2
-    punpcklbw  xmm0, xmm0         // first 2
-    punpckhbw  xmm1, xmm1         // next 2
-    punpcklbw  xmm2, xmm5         // first 2
-    punpckhbw  xmm3, xmm5         // next 2
-    pmulhuw    xmm0, xmm2         // src_argb0 * src_argb1 first 2
-    pmulhuw    xmm1, xmm3         // src_argb0 * src_argb1 next 2
-    lea        eax, [eax + 16]
-    lea        esi, [esi + 16]
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBMULTIPLYROW_SSE2
-
-#ifdef HAS_ARGBADDROW_SSE2
-// Add 2 rows of ARGB pixels together, 4 pixels at a time.
-// TODO(fbarchard): Port this to posix, neon and other math functions.
-__declspec(naked) __declspec(align(16))
-void ARGBAddRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                     uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-
-    sub        ecx, 4
-    jl         convertloop49
-
-    align      4
- convertloop4:
-    movdqu     xmm0, [eax]        // read 4 pixels from src_argb0
-    lea        eax, [eax + 16]
-    movdqu     xmm1, [esi]        // read 4 pixels from src_argb1
-    lea        esi, [esi + 16]
-    paddusb    xmm0, xmm1         // src_argb0 + src_argb1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jge        convertloop4
-
- convertloop49:
-    add        ecx, 4 - 1
-    jl         convertloop19
-
- convertloop1:
-    movd       xmm0, [eax]        // read 1 pixels from src_argb0
-    lea        eax, [eax + 4]
-    movd       xmm1, [esi]        // read 1 pixels from src_argb1
-    lea        esi, [esi + 4]
-    paddusb    xmm0, xmm1         // src_argb0 + src_argb1
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        convertloop1
-
- convertloop19:
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBADDROW_SSE2
-
-#ifdef HAS_ARGBSUBTRACTROW_SSE2
-// Subtract 2 rows of ARGB pixels together, 4 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBSubtractRow_SSE2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-
-    align      4
- convertloop:
-    movdqu     xmm0, [eax]        // read 4 pixels from src_argb0
-    lea        eax, [eax + 16]
-    movdqu     xmm1, [esi]        // read 4 pixels from src_argb1
-    lea        esi, [esi + 16]
-    psubusb    xmm0, xmm1         // src_argb0 - src_argb1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBSUBTRACTROW_SSE2
-
-#ifdef HAS_ARGBMULTIPLYROW_AVX2
-// Multiply 2 rows of ARGB pixels together, 8 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBMultiplyRow_AVX2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    vpxor      ymm5, ymm5, ymm5     // constant 0
-
-    align      4
- convertloop:
-    vmovdqu    ymm1, [eax]        // read 8 pixels from src_argb0
-    lea        eax, [eax + 32]
-    vmovdqu    ymm3, [esi]        // read 8 pixels from src_argb1
-    lea        esi, [esi + 32]
-    vpunpcklbw ymm0, ymm1, ymm1   // low 4
-    vpunpckhbw ymm1, ymm1, ymm1   // high 4
-    vpunpcklbw ymm2, ymm3, ymm5   // low 4
-    vpunpckhbw ymm3, ymm3, ymm5   // high 4
-    vpmulhuw   ymm0, ymm0, ymm2   // src_argb0 * src_argb1 low 4
-    vpmulhuw   ymm1, ymm1, ymm3   // src_argb0 * src_argb1 high 4
-    vpackuswb  ymm0, ymm0, ymm1
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBMULTIPLYROW_AVX2
-
-#ifdef HAS_ARGBADDROW_AVX2
-// Add 2 rows of ARGB pixels together, 8 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBAddRow_AVX2(const uint8* src_argb0, const uint8* src_argb1,
-                     uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-
-    align      4
- convertloop:
-    vmovdqu    ymm0, [eax]              // read 8 pixels from src_argb0
-    lea        eax, [eax + 32]
-    vpaddusb   ymm0, ymm0, [esi]        // add 8 pixels from src_argb1
-    lea        esi, [esi + 32]
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBADDROW_AVX2
-
-#ifdef HAS_ARGBSUBTRACTROW_AVX2
-// Subtract 2 rows of ARGB pixels together, 8 pixels at a time.
-__declspec(naked) __declspec(align(16))
-void ARGBSubtractRow_AVX2(const uint8* src_argb0, const uint8* src_argb1,
-                          uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_argb0
-    mov        esi, [esp + 4 + 8]   // src_argb1
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-
-    align      4
- convertloop:
-    vmovdqu    ymm0, [eax]              // read 8 pixels from src_argb0
-    lea        eax, [eax + 32]
-    vpsubusb   ymm0, ymm0, [esi]        // src_argb0 - src_argb1
-    lea        esi, [esi + 32]
-    vmovdqu    [edx], ymm0
-    lea        edx, [edx + 32]
-    sub        ecx, 8
-    jg         convertloop
-
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBSUBTRACTROW_AVX2
-
-#ifdef HAS_SOBELXROW_SSE2
-// SobelX as a matrix is
-// -1  0  1
-// -2  0  2
-// -1  0  1
-__declspec(naked) __declspec(align(16))
-void SobelXRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    const uint8* src_y2, uint8* dst_sobelx, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   // src_y0
-    mov        esi, [esp + 8 + 8]   // src_y1
-    mov        edi, [esp + 8 + 12]  // src_y2
-    mov        edx, [esp + 8 + 16]  // dst_sobelx
-    mov        ecx, [esp + 8 + 20]  // width
-    sub        esi, eax
-    sub        edi, eax
-    sub        edx, eax
-    pxor       xmm5, xmm5  // constant 0
-
-    align      4
- convertloop:
-    movq       xmm0, qword ptr [eax]            // read 8 pixels from src_y0[0]
-    movq       xmm1, qword ptr [eax + 2]        // read 8 pixels from src_y0[2]
-    punpcklbw  xmm0, xmm5
-    punpcklbw  xmm1, xmm5
-    psubw      xmm0, xmm1
-    movq       xmm1, qword ptr [eax + esi]      // read 8 pixels from src_y1[0]
-    movq       xmm2, qword ptr [eax + esi + 2]  // read 8 pixels from src_y1[2]
-    punpcklbw  xmm1, xmm5
-    punpcklbw  xmm2, xmm5
-    psubw      xmm1, xmm2
-    movq       xmm2, qword ptr [eax + edi]      // read 8 pixels from src_y2[0]
-    movq       xmm3, qword ptr [eax + edi + 2]  // read 8 pixels from src_y2[2]
-    punpcklbw  xmm2, xmm5
-    punpcklbw  xmm3, xmm5
-    psubw      xmm2, xmm3
-    paddw      xmm0, xmm2
-    paddw      xmm0, xmm1
-    paddw      xmm0, xmm1
-    pxor       xmm1, xmm1   // abs = max(xmm0, -xmm0).  SSSE3 could use pabsw
-    psubw      xmm1, xmm0
-    pmaxsw     xmm0, xmm1
-    packuswb   xmm0, xmm0
-    sub        ecx, 8
-    movq       qword ptr [eax + edx], xmm0
-    lea        eax, [eax + 8]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SOBELXROW_SSE2
-
-#ifdef HAS_SOBELYROW_SSE2
-// SobelY as a matrix is
-// -1 -2 -1
-//  0  0  0
-//  1  2  1
-__declspec(naked) __declspec(align(16))
-void SobelYRow_SSE2(const uint8* src_y0, const uint8* src_y1,
-                    uint8* dst_sobely, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_y0
-    mov        esi, [esp + 4 + 8]   // src_y1
-    mov        edx, [esp + 4 + 12]  // dst_sobely
-    mov        ecx, [esp + 4 + 16]  // width
-    sub        esi, eax
-    sub        edx, eax
-    pxor       xmm5, xmm5  // constant 0
-
-    align      4
- convertloop:
-    movq       xmm0, qword ptr [eax]            // read 8 pixels from src_y0[0]
-    movq       xmm1, qword ptr [eax + esi]      // read 8 pixels from src_y1[0]
-    punpcklbw  xmm0, xmm5
-    punpcklbw  xmm1, xmm5
-    psubw      xmm0, xmm1
-    movq       xmm1, qword ptr [eax + 1]        // read 8 pixels from src_y0[1]
-    movq       xmm2, qword ptr [eax + esi + 1]  // read 8 pixels from src_y1[1]
-    punpcklbw  xmm1, xmm5
-    punpcklbw  xmm2, xmm5
-    psubw      xmm1, xmm2
-    movq       xmm2, qword ptr [eax + 2]        // read 8 pixels from src_y0[2]
-    movq       xmm3, qword ptr [eax + esi + 2]  // read 8 pixels from src_y1[2]
-    punpcklbw  xmm2, xmm5
-    punpcklbw  xmm3, xmm5
-    psubw      xmm2, xmm3
-    paddw      xmm0, xmm2
-    paddw      xmm0, xmm1
-    paddw      xmm0, xmm1
-    pxor       xmm1, xmm1   // abs = max(xmm0, -xmm0).  SSSE3 could use pabsw
-    psubw      xmm1, xmm0
-    pmaxsw     xmm0, xmm1
-    packuswb   xmm0, xmm0
-    sub        ecx, 8
-    movq       qword ptr [eax + edx], xmm0
-    lea        eax, [eax + 8]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SOBELYROW_SSE2
-
-#ifdef HAS_SOBELROW_SSE2
-// Adds Sobel X and Sobel Y and stores Sobel into ARGB.
-// A = 255
-// R = Sobel
-// G = Sobel
-// B = Sobel
-__declspec(naked) __declspec(align(16))
-void SobelRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                   uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_sobelx
-    mov        esi, [esp + 4 + 8]   // src_sobely
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    sub        esi, eax
-    pcmpeqb    xmm5, xmm5           // alpha 255
-    pslld      xmm5, 24             // 0xff000000
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]            // read 16 pixels src_sobelx
-    movdqa     xmm1, [eax + esi]      // read 16 pixels src_sobely
-    lea        eax, [eax + 16]
-    paddusb    xmm0, xmm1             // sobel = sobelx + sobely
-    movdqa     xmm2, xmm0             // GG
-    punpcklbw  xmm2, xmm0             // First 8
-    punpckhbw  xmm0, xmm0             // Next 8
-    movdqa     xmm1, xmm2             // GGGG
-    punpcklwd  xmm1, xmm2             // First 4
-    punpckhwd  xmm2, xmm2             // Next 4
-    por        xmm1, xmm5             // GGGA
-    por        xmm2, xmm5
-    movdqa     xmm3, xmm0             // GGGG
-    punpcklwd  xmm3, xmm0             // Next 4
-    punpckhwd  xmm0, xmm0             // Last 4
-    por        xmm3, xmm5             // GGGA
-    por        xmm0, xmm5
-    sub        ecx, 16
-    movdqa     [edx], xmm1
-    movdqa     [edx + 16], xmm2
-    movdqa     [edx + 32], xmm3
-    movdqa     [edx + 48], xmm0
-    lea        edx, [edx + 64]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SOBELROW_SSE2
-
-#ifdef HAS_SOBELTOPLANEROW_SSE2
-// Adds Sobel X and Sobel Y and stores Sobel into a plane.
-__declspec(naked) __declspec(align(16))
-void SobelToPlaneRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                          uint8* dst_y, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_sobelx
-    mov        esi, [esp + 4 + 8]   // src_sobely
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    sub        esi, eax
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]            // read 16 pixels src_sobelx
-    movdqa     xmm1, [eax + esi]      // read 16 pixels src_sobely
-    lea        eax, [eax + 16]
-    paddusb    xmm0, xmm1             // sobel = sobelx + sobely
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SOBELTOPLANEROW_SSE2
-
-#ifdef HAS_SOBELXYROW_SSE2
-// Mixes Sobel X, Sobel Y and Sobel into ARGB.
-// A = 255
-// R = Sobel X
-// G = Sobel
-// B = Sobel Y
-__declspec(naked) __declspec(align(16))
-void SobelXYRow_SSE2(const uint8* src_sobelx, const uint8* src_sobely,
-                     uint8* dst_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   // src_sobelx
-    mov        esi, [esp + 4 + 8]   // src_sobely
-    mov        edx, [esp + 4 + 12]  // dst_argb
-    mov        ecx, [esp + 4 + 16]  // width
-    sub        esi, eax
-    pcmpeqb    xmm5, xmm5           // alpha 255
-
-    align      4
- convertloop:
-    movdqa     xmm0, [eax]            // read 16 pixels src_sobelx
-    movdqa     xmm1, [eax + esi]      // read 16 pixels src_sobely
-    lea        eax, [eax + 16]
-    movdqa     xmm2, xmm0
-    paddusb    xmm2, xmm1             // sobel = sobelx + sobely
-    movdqa     xmm3, xmm0             // XA
-    punpcklbw  xmm3, xmm5
-    punpckhbw  xmm0, xmm5
-    movdqa     xmm4, xmm1             // YS
-    punpcklbw  xmm4, xmm2
-    punpckhbw  xmm1, xmm2
-    movdqa     xmm6, xmm4             // YSXA
-    punpcklwd  xmm6, xmm3             // First 4
-    punpckhwd  xmm4, xmm3             // Next 4
-    movdqa     xmm7, xmm1             // YSXA
-    punpcklwd  xmm7, xmm0             // Next 4
-    punpckhwd  xmm1, xmm0             // Last 4
-    sub        ecx, 16
-    movdqa     [edx], xmm6
-    movdqa     [edx + 16], xmm4
-    movdqa     [edx + 32], xmm7
-    movdqa     [edx + 48], xmm1
-    lea        edx, [edx + 64]
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_SOBELXYROW_SSE2
-
-#ifdef HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
-// Consider float CumulativeSum.
-// Consider calling CumulativeSum one row at time as needed.
-// Consider circular CumulativeSum buffer of radius * 2 + 1 height.
-// Convert cumulative sum for an area to an average for 1 pixel.
-// topleft is pointer to top left of CumulativeSum buffer for area.
-// botleft is pointer to bottom left of CumulativeSum buffer.
-// width is offset from left to right of area in CumulativeSum buffer measured
-//   in number of ints.
-// area is the number of pixels in the area being averaged.
-// dst points to pixel to store result to.
-// count is number of averaged pixels to produce.
-// Does 4 pixels at a time, requires CumulativeSum pointers to be 16 byte
-// aligned.
-void CumulativeSumToAverageRow_SSE2(const int32* topleft, const int32* botleft,
-                                    int width, int area, uint8* dst,
-                                    int count) {
-  __asm {
-    mov        eax, topleft  // eax topleft
-    mov        esi, botleft  // esi botleft
-    mov        edx, width
-    movd       xmm5, area
-    mov        edi, dst
-    mov        ecx, count
-    cvtdq2ps   xmm5, xmm5
-    rcpss      xmm4, xmm5  // 1.0f / area
-    pshufd     xmm4, xmm4, 0
-    sub        ecx, 4
-    jl         l4b
-
-    cmp        area, 128  // 128 pixels will not overflow 15 bits.
-    ja         l4
-
-    pshufd     xmm5, xmm5, 0        // area
-    pcmpeqb    xmm6, xmm6           // constant of 65536.0 - 1 = 65535.0
-    psrld      xmm6, 16
-    cvtdq2ps   xmm6, xmm6
-    addps      xmm5, xmm6           // (65536.0 + area - 1)
-    mulps      xmm5, xmm4           // (65536.0 + area - 1) * 1 / area
-    cvtps2dq   xmm5, xmm5           // 0.16 fixed point
-    packssdw   xmm5, xmm5           // 16 bit shorts
-
-    // 4 pixel loop small blocks.
-    align      4
-  s4:
-    // top left
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-
-    // - top right
-    psubd      xmm0, [eax + edx * 4]
-    psubd      xmm1, [eax + edx * 4 + 16]
-    psubd      xmm2, [eax + edx * 4 + 32]
-    psubd      xmm3, [eax + edx * 4 + 48]
-    lea        eax, [eax + 64]
-
-    // - bottom left
-    psubd      xmm0, [esi]
-    psubd      xmm1, [esi + 16]
-    psubd      xmm2, [esi + 32]
-    psubd      xmm3, [esi + 48]
-
-    // + bottom right
-    paddd      xmm0, [esi + edx * 4]
-    paddd      xmm1, [esi + edx * 4 + 16]
-    paddd      xmm2, [esi + edx * 4 + 32]
-    paddd      xmm3, [esi + edx * 4 + 48]
-    lea        esi, [esi + 64]
-
-    packssdw   xmm0, xmm1  // pack 4 pixels into 2 registers
-    packssdw   xmm2, xmm3
-
-    pmulhuw    xmm0, xmm5
-    pmulhuw    xmm2, xmm5
-
-    packuswb   xmm0, xmm2
-    movdqu     [edi], xmm0
-    lea        edi, [edi + 16]
-    sub        ecx, 4
-    jge        s4
-
-    jmp        l4b
-
-    // 4 pixel loop
-    align      4
-  l4:
-    // top left
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + 32]
-    movdqa     xmm3, [eax + 48]
-
-    // - top right
-    psubd      xmm0, [eax + edx * 4]
-    psubd      xmm1, [eax + edx * 4 + 16]
-    psubd      xmm2, [eax + edx * 4 + 32]
-    psubd      xmm3, [eax + edx * 4 + 48]
-    lea        eax, [eax + 64]
-
-    // - bottom left
-    psubd      xmm0, [esi]
-    psubd      xmm1, [esi + 16]
-    psubd      xmm2, [esi + 32]
-    psubd      xmm3, [esi + 48]
-
-    // + bottom right
-    paddd      xmm0, [esi + edx * 4]
-    paddd      xmm1, [esi + edx * 4 + 16]
-    paddd      xmm2, [esi + edx * 4 + 32]
-    paddd      xmm3, [esi + edx * 4 + 48]
-    lea        esi, [esi + 64]
-
-    cvtdq2ps   xmm0, xmm0   // Average = Sum * 1 / Area
-    cvtdq2ps   xmm1, xmm1
-    mulps      xmm0, xmm4
-    mulps      xmm1, xmm4
-    cvtdq2ps   xmm2, xmm2
-    cvtdq2ps   xmm3, xmm3
-    mulps      xmm2, xmm4
-    mulps      xmm3, xmm4
-    cvtps2dq   xmm0, xmm0
-    cvtps2dq   xmm1, xmm1
-    cvtps2dq   xmm2, xmm2
-    cvtps2dq   xmm3, xmm3
-    packssdw   xmm0, xmm1
-    packssdw   xmm2, xmm3
-    packuswb   xmm0, xmm2
-    movdqu     [edi], xmm0
-    lea        edi, [edi + 16]
-    sub        ecx, 4
-    jge        l4
-
-  l4b:
-    add        ecx, 4 - 1
-    jl         l1b
-
-    // 1 pixel loop
-    align      4
-  l1:
-    movdqa     xmm0, [eax]
-    psubd      xmm0, [eax + edx * 4]
-    lea        eax, [eax + 16]
-    psubd      xmm0, [esi]
-    paddd      xmm0, [esi + edx * 4]
-    lea        esi, [esi + 16]
-    cvtdq2ps   xmm0, xmm0
-    mulps      xmm0, xmm4
-    cvtps2dq   xmm0, xmm0
-    packssdw   xmm0, xmm0
-    packuswb   xmm0, xmm0
-    movd       dword ptr [edi], xmm0
-    lea        edi, [edi + 4]
-    sub        ecx, 1
-    jge        l1
-  l1b:
-  }
-}
-#endif  // HAS_CUMULATIVESUMTOAVERAGEROW_SSE2
-
-#ifdef HAS_COMPUTECUMULATIVESUMROW_SSE2
-// Creates a table of cumulative sums where each value is a sum of all values
-// above and to the left of the value.
-void ComputeCumulativeSumRow_SSE2(const uint8* row, int32* cumsum,
-                                  const int32* previous_cumsum, int width) {
-  __asm {
-    mov        eax, row
-    mov        edx, cumsum
-    mov        esi, previous_cumsum
-    mov        ecx, width
-    pxor       xmm0, xmm0
-    pxor       xmm1, xmm1
-
-    sub        ecx, 4
-    jl         l4b
-    test       edx, 15
-    jne        l4b
-
-    // 4 pixel loop
-    align      4
-  l4:
-    movdqu     xmm2, [eax]  // 4 argb pixels 16 bytes.
-    lea        eax, [eax + 16]
-    movdqa     xmm4, xmm2
-
-    punpcklbw  xmm2, xmm1
-    movdqa     xmm3, xmm2
-    punpcklwd  xmm2, xmm1
-    punpckhwd  xmm3, xmm1
-
-    punpckhbw  xmm4, xmm1
-    movdqa     xmm5, xmm4
-    punpcklwd  xmm4, xmm1
-    punpckhwd  xmm5, xmm1
-
-    paddd      xmm0, xmm2
-    movdqa     xmm2, [esi]  // previous row above.
-    paddd      xmm2, xmm0
-
-    paddd      xmm0, xmm3
-    movdqa     xmm3, [esi + 16]
-    paddd      xmm3, xmm0
-
-    paddd      xmm0, xmm4
-    movdqa     xmm4, [esi + 32]
-    paddd      xmm4, xmm0
-
-    paddd      xmm0, xmm5
-    movdqa     xmm5, [esi + 48]
-    lea        esi, [esi + 64]
-    paddd      xmm5, xmm0
-
-    movdqa     [edx], xmm2
-    movdqa     [edx + 16], xmm3
-    movdqa     [edx + 32], xmm4
-    movdqa     [edx + 48], xmm5
-
-    lea        edx, [edx + 64]
-    sub        ecx, 4
-    jge        l4
-
-  l4b:
-    add        ecx, 4 - 1
-    jl         l1b
-
-    // 1 pixel loop
-    align      4
-  l1:
-    movd       xmm2, dword ptr [eax]  // 1 argb pixel 4 bytes.
-    lea        eax, [eax + 4]
-    punpcklbw  xmm2, xmm1
-    punpcklwd  xmm2, xmm1
-    paddd      xmm0, xmm2
-    movdqu     xmm2, [esi]
-    lea        esi, [esi + 16]
-    paddd      xmm2, xmm0
-    movdqu     [edx], xmm2
-    lea        edx, [edx + 16]
-    sub        ecx, 1
-    jge        l1
-
- l1b:
-  }
-}
-#endif  // HAS_COMPUTECUMULATIVESUMROW_SSE2
-
-#ifdef HAS_ARGBAFFINEROW_SSE2
-// Copy ARGB pixels from source image with slope to a row of destination.
-__declspec(naked) __declspec(align(16))
-LIBYUV_API
-void ARGBAffineRow_SSE2(const uint8* src_argb, int src_argb_stride,
-                        uint8* dst_argb, const float* uv_dudv, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 12]  // src_argb
-    mov        esi, [esp + 16]  // stride
-    mov        edx, [esp + 20]  // dst_argb
-    mov        ecx, [esp + 24]  // pointer to uv_dudv
-    movq       xmm2, qword ptr [ecx]  // uv
-    movq       xmm7, qword ptr [ecx + 8]  // dudv
-    mov        ecx, [esp + 28]  // width
-    shl        esi, 16          // 4, stride
-    add        esi, 4
-    movd       xmm5, esi
-    sub        ecx, 4
-    jl         l4b
-
-    // setup for 4 pixel loop
-    pshufd     xmm7, xmm7, 0x44  // dup dudv
-    pshufd     xmm5, xmm5, 0  // dup 4, stride
-    movdqa     xmm0, xmm2    // x0, y0, x1, y1
-    addps      xmm0, xmm7
-    movlhps    xmm2, xmm0
-    movdqa     xmm4, xmm7
-    addps      xmm4, xmm4    // dudv *= 2
-    movdqa     xmm3, xmm2    // x2, y2, x3, y3
-    addps      xmm3, xmm4
-    addps      xmm4, xmm4    // dudv *= 4
-
-    // 4 pixel loop
-    align      4
-  l4:
-    cvttps2dq  xmm0, xmm2    // x, y float to int first 2
-    cvttps2dq  xmm1, xmm3    // x, y float to int next 2
-    packssdw   xmm0, xmm1    // x, y as 8 shorts
-    pmaddwd    xmm0, xmm5    // offsets = x * 4 + y * stride.
-    movd       esi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // shift right
-    movd       edi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // shift right
-    movd       xmm1, [eax + esi]  // read pixel 0
-    movd       xmm6, [eax + edi]  // read pixel 1
-    punpckldq  xmm1, xmm6     // combine pixel 0 and 1
-    addps      xmm2, xmm4    // x, y += dx, dy first 2
-    movq       qword ptr [edx], xmm1
-    movd       esi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // shift right
-    movd       edi, xmm0
-    movd       xmm6, [eax + esi]  // read pixel 2
-    movd       xmm0, [eax + edi]  // read pixel 3
-    punpckldq  xmm6, xmm0     // combine pixel 2 and 3
-    addps      xmm3, xmm4    // x, y += dx, dy next 2
-    sub        ecx, 4
-    movq       qword ptr 8[edx], xmm6
-    lea        edx, [edx + 16]
-    jge        l4
-
-  l4b:
-    add        ecx, 4 - 1
-    jl         l1b
-
-    // 1 pixel loop
-    align      4
-  l1:
-    cvttps2dq  xmm0, xmm2    // x, y float to int
-    packssdw   xmm0, xmm0    // x, y as shorts
-    pmaddwd    xmm0, xmm5    // offset = x * 4 + y * stride
-    addps      xmm2, xmm7    // x, y += dx, dy
-    movd       esi, xmm0
-    movd       xmm0, [eax + esi]  // copy a pixel
-    sub        ecx, 1
-    movd       [edx], xmm0
-    lea        edx, [edx + 4]
-    jge        l1
-  l1b:
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBAFFINEROW_SSE2
-
-#ifdef HAS_INTERPOLATEROW_AVX2
-// Bilinear filter 16x2 -> 16x1
-__declspec(naked) __declspec(align(16))
-void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
-                          ptrdiff_t src_stride, int dst_width,
-                          int source_y_fraction) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]   // dst_ptr
-    mov        esi, [esp + 8 + 8]   // src_ptr
-    mov        edx, [esp + 8 + 12]  // src_stride
-    mov        ecx, [esp + 8 + 16]  // dst_width
-    mov        eax, [esp + 8 + 20]  // source_y_fraction (0..255)
-    shr        eax, 1
-    // Dispatch to specialized filters if applicable.
-    cmp        eax, 0
-    je         xloop100  // 0 / 128.  Blend 100 / 0.
-    sub        edi, esi
-    cmp        eax, 32
-    je         xloop75   // 32 / 128 is 0.25.  Blend 75 / 25.
-    cmp        eax, 64
-    je         xloop50   // 64 / 128 is 0.50.  Blend 50 / 50.
-    cmp        eax, 96
-    je         xloop25   // 96 / 128 is 0.75.  Blend 25 / 75.
-
-    vmovd      xmm0, eax  // high fraction 0..127
-    neg        eax
-    add        eax, 128
-    vmovd      xmm5, eax  // low fraction 128..1
-    vpunpcklbw xmm5, xmm5, xmm0
-    vpunpcklwd xmm5, xmm5, xmm5
-    vpxor      ymm0, ymm0, ymm0
-    vpermd     ymm5, ymm0, ymm5
-
-    align      4
-  xloop:
-    vmovdqu    ymm0, [esi]
-    vmovdqu    ymm2, [esi + edx]
-    vpunpckhbw ymm1, ymm0, ymm2  // mutates
-    vpunpcklbw ymm0, ymm0, ymm2  // mutates
-    vpmaddubsw ymm0, ymm0, ymm5
-    vpmaddubsw ymm1, ymm1, ymm5
-    vpsrlw     ymm0, ymm0, 7
-    vpsrlw     ymm1, ymm1, 7
-    vpackuswb  ymm0, ymm0, ymm1  // unmutates
-    sub        ecx, 32
-    vmovdqu    [esi + edi], ymm0
-    lea        esi, [esi + 32]
-    jg         xloop
-    jmp        xloop99
-
-    // Blend 25 / 75.
-    align      4
-  xloop25:
-    vmovdqu    ymm0, [esi]
-    vpavgb     ymm0, ymm0, [esi + edx]
-    vpavgb     ymm0, ymm0, [esi + edx]
-    sub        ecx, 32
-    vmovdqu    [esi + edi], ymm0
-    lea        esi, [esi + 32]
-    jg         xloop25
-    jmp        xloop99
-
-    // Blend 50 / 50.
-    align      4
-  xloop50:
-    vmovdqu    ymm0, [esi]
-    vpavgb     ymm0, ymm0, [esi + edx]
-    sub        ecx, 32
-    vmovdqu    [esi + edi], ymm0
-    lea        esi, [esi + 32]
-    jg         xloop50
-    jmp        xloop99
-
-    // Blend 75 / 25.
-    align      4
-  xloop75:
-    vmovdqu    ymm0, [esi + edx]
-    vpavgb     ymm0, ymm0, [esi]
-    vpavgb     ymm0, ymm0, [esi]
-    sub        ecx, 32
-    vmovdqu     [esi + edi], ymm0
-    lea        esi, [esi + 32]
-    jg         xloop75
-    jmp        xloop99
-
-    // Blend 100 / 0 - Copy row unchanged.
-    align      4
-  xloop100:
-    rep movsb
-
-  xloop99:
-    pop        edi
-    pop        esi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_INTERPOLATEROW_AVX2
-
-#ifdef HAS_INTERPOLATEROW_SSSE3
-// Bilinear filter 16x2 -> 16x1
-__declspec(naked) __declspec(align(16))
-void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                          ptrdiff_t src_stride, int dst_width,
-                          int source_y_fraction) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]   // dst_ptr
-    mov        esi, [esp + 8 + 8]   // src_ptr
-    mov        edx, [esp + 8 + 12]  // src_stride
-    mov        ecx, [esp + 8 + 16]  // dst_width
-    mov        eax, [esp + 8 + 20]  // source_y_fraction (0..255)
-    sub        edi, esi
-    shr        eax, 1
-    // Dispatch to specialized filters if applicable.
-    cmp        eax, 0
-    je         xloop100  // 0 / 128.  Blend 100 / 0.
-    cmp        eax, 32
-    je         xloop75   // 32 / 128 is 0.25.  Blend 75 / 25.
-    cmp        eax, 64
-    je         xloop50   // 64 / 128 is 0.50.  Blend 50 / 50.
-    cmp        eax, 96
-    je         xloop25   // 96 / 128 is 0.75.  Blend 25 / 75.
-
-    movd       xmm0, eax  // high fraction 0..127
-    neg        eax
-    add        eax, 128
-    movd       xmm5, eax  // low fraction 128..1
-    punpcklbw  xmm5, xmm0
-    punpcklwd  xmm5, xmm5
-    pshufd     xmm5, xmm5, 0
-
-    align      4
-  xloop:
-    movdqa     xmm0, [esi]
-    movdqa     xmm2, [esi + edx]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm2
-    punpckhbw  xmm1, xmm2
-    pmaddubsw  xmm0, xmm5
-    pmaddubsw  xmm1, xmm5
-    psrlw      xmm0, 7
-    psrlw      xmm1, 7
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop
-    jmp        xloop99
-
-    // Blend 25 / 75.
-    align      4
-  xloop25:
-    movdqa     xmm0, [esi]
-    movdqa     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop25
-    jmp        xloop99
-
-    // Blend 50 / 50.
-    align      4
-  xloop50:
-    movdqa     xmm0, [esi]
-    movdqa     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop50
-    jmp        xloop99
-
-    // Blend 75 / 25.
-    align      4
-  xloop75:
-    movdqa     xmm1, [esi]
-    movdqa     xmm0, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop75
-    jmp        xloop99
-
-    // Blend 100 / 0 - Copy row unchanged.
-    align      4
-  xloop100:
-    movdqa     xmm0, [esi]
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop100
-
-  xloop99:
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_INTERPOLATEROW_SSSE3
-
-#ifdef HAS_INTERPOLATEROW_SSE2
-// Bilinear filter 16x2 -> 16x1
-__declspec(naked) __declspec(align(16))
-void InterpolateRow_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                         ptrdiff_t src_stride, int dst_width,
-                         int source_y_fraction) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]   // dst_ptr
-    mov        esi, [esp + 8 + 8]   // src_ptr
-    mov        edx, [esp + 8 + 12]  // src_stride
-    mov        ecx, [esp + 8 + 16]  // dst_width
-    mov        eax, [esp + 8 + 20]  // source_y_fraction (0..255)
-    sub        edi, esi
-    // Dispatch to specialized filters if applicable.
-    cmp        eax, 0
-    je         xloop100  // 0 / 256.  Blend 100 / 0.
-    cmp        eax, 64
-    je         xloop75   // 64 / 256 is 0.25.  Blend 75 / 25.
-    cmp        eax, 128
-    je         xloop50   // 128 / 256 is 0.50.  Blend 50 / 50.
-    cmp        eax, 192
-    je         xloop25   // 192 / 256 is 0.75.  Blend 25 / 75.
-
-    movd       xmm5, eax            // xmm5 = y fraction
-    punpcklbw  xmm5, xmm5
-    psrlw      xmm5, 1
-    punpcklwd  xmm5, xmm5
-    punpckldq  xmm5, xmm5
-    punpcklqdq xmm5, xmm5
-    pxor       xmm4, xmm4
-
-    align      4
-  xloop:
-    movdqa     xmm0, [esi]  // row0
-    movdqa     xmm2, [esi + edx]  // row1
-    movdqa     xmm1, xmm0
-    movdqa     xmm3, xmm2
-    punpcklbw  xmm2, xmm4
-    punpckhbw  xmm3, xmm4
-    punpcklbw  xmm0, xmm4
-    punpckhbw  xmm1, xmm4
-    psubw      xmm2, xmm0  // row1 - row0
-    psubw      xmm3, xmm1
-    paddw      xmm2, xmm2  // 9 bits * 15 bits = 8.16
-    paddw      xmm3, xmm3
-    pmulhw     xmm2, xmm5  // scale diff
-    pmulhw     xmm3, xmm5
-    paddw      xmm0, xmm2  // sum rows
-    paddw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop
-    jmp        xloop99
-
-    // Blend 25 / 75.
-    align      4
-  xloop25:
-    movdqa     xmm0, [esi]
-    movdqa     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop25
-    jmp        xloop99
-
-    // Blend 50 / 50.
-    align      4
-  xloop50:
-    movdqa     xmm0, [esi]
-    movdqa     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop50
-    jmp        xloop99
-
-    // Blend 75 / 25.
-    align      4
-  xloop75:
-    movdqa     xmm1, [esi]
-    movdqa     xmm0, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop75
-    jmp        xloop99
-
-    // Blend 100 / 0 - Copy row unchanged.
-    align      4
-  xloop100:
-    movdqa     xmm0, [esi]
-    sub        ecx, 16
-    movdqa     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop100
-
-  xloop99:
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_INTERPOLATEROW_SSE2
-
-// Bilinear filter 16x2 -> 16x1
-__declspec(naked) __declspec(align(16))
-void InterpolateRow_Unaligned_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                                    ptrdiff_t src_stride, int dst_width,
-                                    int source_y_fraction) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]   // dst_ptr
-    mov        esi, [esp + 8 + 8]   // src_ptr
-    mov        edx, [esp + 8 + 12]  // src_stride
-    mov        ecx, [esp + 8 + 16]  // dst_width
-    mov        eax, [esp + 8 + 20]  // source_y_fraction (0..255)
-    sub        edi, esi
-    shr        eax, 1
-    // Dispatch to specialized filters if applicable.
-    cmp        eax, 0
-    je         xloop100  // 0 / 128.  Blend 100 / 0.
-    cmp        eax, 32
-    je         xloop75   // 32 / 128 is 0.25.  Blend 75 / 25.
-    cmp        eax, 64
-    je         xloop50   // 64 / 128 is 0.50.  Blend 50 / 50.
-    cmp        eax, 96
-    je         xloop25   // 96 / 128 is 0.75.  Blend 25 / 75.
-
-    movd       xmm0, eax  // high fraction 0..127
-    neg        eax
-    add        eax, 128
-    movd       xmm5, eax  // low fraction 128..1
-    punpcklbw  xmm5, xmm0
-    punpcklwd  xmm5, xmm5
-    pshufd     xmm5, xmm5, 0
-
-    align      4
-  xloop:
-    movdqu     xmm0, [esi]
-    movdqu     xmm2, [esi + edx]
-    movdqu     xmm1, xmm0
-    punpcklbw  xmm0, xmm2
-    punpckhbw  xmm1, xmm2
-    pmaddubsw  xmm0, xmm5
-    pmaddubsw  xmm1, xmm5
-    psrlw      xmm0, 7
-    psrlw      xmm1, 7
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop
-    jmp        xloop99
-
-    // Blend 25 / 75.
-    align      4
-  xloop25:
-    movdqu     xmm0, [esi]
-    movdqu     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop25
-    jmp        xloop99
-
-    // Blend 50 / 50.
-    align      4
-  xloop50:
-    movdqu     xmm0, [esi]
-    movdqu     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop50
-    jmp        xloop99
-
-    // Blend 75 / 25.
-    align      4
-  xloop75:
-    movdqu     xmm1, [esi]
-    movdqu     xmm0, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop75
-    jmp        xloop99
-
-    // Blend 100 / 0 - Copy row unchanged.
-    align      4
-  xloop100:
-    movdqu     xmm0, [esi]
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop100
-
-  xloop99:
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-#ifdef HAS_INTERPOLATEROW_SSE2
-// Bilinear filter 16x2 -> 16x1
-__declspec(naked) __declspec(align(16))
-void InterpolateRow_Unaligned_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                                   ptrdiff_t src_stride, int dst_width,
-                                   int source_y_fraction) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]   // dst_ptr
-    mov        esi, [esp + 8 + 8]   // src_ptr
-    mov        edx, [esp + 8 + 12]  // src_stride
-    mov        ecx, [esp + 8 + 16]  // dst_width
-    mov        eax, [esp + 8 + 20]  // source_y_fraction (0..255)
-    sub        edi, esi
-    // Dispatch to specialized filters if applicable.
-    cmp        eax, 0
-    je         xloop100  // 0 / 256.  Blend 100 / 0.
-    cmp        eax, 64
-    je         xloop75   // 64 / 256 is 0.25.  Blend 75 / 25.
-    cmp        eax, 128
-    je         xloop50   // 128 / 256 is 0.50.  Blend 50 / 50.
-    cmp        eax, 192
-    je         xloop25   // 192 / 256 is 0.75.  Blend 25 / 75.
-
-    movd       xmm5, eax            // xmm5 = y fraction
-    punpcklbw  xmm5, xmm5
-    psrlw      xmm5, 1
-    punpcklwd  xmm5, xmm5
-    punpckldq  xmm5, xmm5
-    punpcklqdq xmm5, xmm5
-    pxor       xmm4, xmm4
-
-    align      4
-  xloop:
-    movdqu     xmm0, [esi]  // row0
-    movdqu     xmm2, [esi + edx]  // row1
-    movdqu     xmm1, xmm0
-    movdqu     xmm3, xmm2
-    punpcklbw  xmm2, xmm4
-    punpckhbw  xmm3, xmm4
-    punpcklbw  xmm0, xmm4
-    punpckhbw  xmm1, xmm4
-    psubw      xmm2, xmm0  // row1 - row0
-    psubw      xmm3, xmm1
-    paddw      xmm2, xmm2  // 9 bits * 15 bits = 8.16
-    paddw      xmm3, xmm3
-    pmulhw     xmm2, xmm5  // scale diff
-    pmulhw     xmm3, xmm5
-    paddw      xmm0, xmm2  // sum rows
-    paddw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop
-    jmp        xloop99
-
-    // Blend 25 / 75.
-    align      4
-  xloop25:
-    movdqu     xmm0, [esi]
-    movdqu     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop25
-    jmp        xloop99
-
-    // Blend 50 / 50.
-    align      4
-  xloop50:
-    movdqu     xmm0, [esi]
-    movdqu     xmm1, [esi + edx]
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop50
-    jmp        xloop99
-
-    // Blend 75 / 25.
-    align      4
-  xloop75:
-    movdqu     xmm1, [esi]
-    movdqu     xmm0, [esi + edx]
-    pavgb      xmm0, xmm1
-    pavgb      xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop75
-    jmp        xloop99
-
-    // Blend 100 / 0 - Copy row unchanged.
-    align      4
-  xloop100:
-    movdqu     xmm0, [esi]
-    sub        ecx, 16
-    movdqu     [esi + edi], xmm0
-    lea        esi, [esi + 16]
-    jg         xloop100
-
-  xloop99:
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_INTERPOLATEROW_SSE2
-
-__declspec(naked) __declspec(align(16))
-void HalfRow_SSE2(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_uv
-    mov        edx, [esp + 4 + 8]    // src_uv_stride
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    sub        edi, eax
-
-    align      4
-  convertloop:
-    movdqa     xmm0, [eax]
-    pavgb      xmm0, [eax + edx]
-    sub        ecx, 16
-    movdqa     [eax + edi], xmm0
-    lea        eax,  [eax + 16]
-    jg         convertloop
-    pop        edi
-    ret
-  }
-}
-
-#ifdef HAS_HALFROW_AVX2
-__declspec(naked) __declspec(align(16))
-void HalfRow_AVX2(const uint8* src_uv, int src_uv_stride,
-                  uint8* dst_uv, int pix) {
-  __asm {
-    push       edi
-    mov        eax, [esp + 4 + 4]    // src_uv
-    mov        edx, [esp + 4 + 8]    // src_uv_stride
-    mov        edi, [esp + 4 + 12]   // dst_v
-    mov        ecx, [esp + 4 + 16]   // pix
-    sub        edi, eax
-
-    align      4
-  convertloop:
-    vmovdqu    ymm0, [eax]
-    vpavgb     ymm0, ymm0, [eax + edx]
-    sub        ecx, 32
-    vmovdqu    [eax + edi], ymm0
-    lea        eax,  [eax + 32]
-    jg         convertloop
-
-    pop        edi
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_HALFROW_AVX2
-
-__declspec(naked) __declspec(align(16))
-void ARGBToBayerRow_SSSE3(const uint8* src_argb, uint8* dst_bayer,
-                          uint32 selector, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_argb
-    mov        edx, [esp + 8]    // dst_bayer
-    movd       xmm5, [esp + 12]  // selector
-    mov        ecx, [esp + 16]   // pix
-    pshufd     xmm5, xmm5, 0
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax, [eax + 32]
-    pshufb     xmm0, xmm5
-    pshufb     xmm1, xmm5
-    punpckldq  xmm0, xmm1
-    sub        ecx, 8
-    movq       qword ptr [edx], xmm0
-    lea        edx, [edx + 8]
-    jg         wloop
-    ret
-  }
-}
-
-// Specialized ARGB to Bayer that just isolates G channel.
-__declspec(naked) __declspec(align(16))
-void ARGBToBayerGGRow_SSE2(const uint8* src_argb, uint8* dst_bayer,
-                           uint32 selector, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_argb
-    mov        edx, [esp + 8]    // dst_bayer
-                                 // selector
-    mov        ecx, [esp + 16]   // pix
-    pcmpeqb    xmm5, xmm5        // generate mask 0x000000ff
-    psrld      xmm5, 24
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax, [eax + 32]
-    psrld      xmm0, 8  // Move green to bottom.
-    psrld      xmm1, 8
-    pand       xmm0, xmm5
-    pand       xmm1, xmm5
-    packssdw   xmm0, xmm1
-    packuswb   xmm0, xmm1
-    sub        ecx, 8
-    movq       qword ptr [edx], xmm0
-    lea        edx, [edx + 8]
-    jg         wloop
-    ret
-  }
-}
-
-// For BGRAToARGB, ABGRToARGB, RGBAToARGB, and ARGBToRGBA.
-__declspec(naked) __declspec(align(16))
-void ARGBShuffleRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                          const uint8* shuffler, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_argb
-    mov        edx, [esp + 8]    // dst_argb
-    mov        ecx, [esp + 12]   // shuffler
-    movdqa     xmm5, [ecx]
-    mov        ecx, [esp + 16]   // pix
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax, [eax + 32]
-    pshufb     xmm0, xmm5
-    pshufb     xmm1, xmm5
-    sub        ecx, 8
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    jg         wloop
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void ARGBShuffleRow_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                    const uint8* shuffler, int pix) {
-  __asm {
-    mov        eax, [esp + 4]    // src_argb
-    mov        edx, [esp + 8]    // dst_argb
-    mov        ecx, [esp + 12]   // shuffler
-    movdqa     xmm5, [ecx]
-    mov        ecx, [esp + 16]   // pix
-
-    align      4
-  wloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax, [eax + 32]
-    pshufb     xmm0, xmm5
-    pshufb     xmm1, xmm5
-    sub        ecx, 8
-    movdqu     [edx], xmm0
-    movdqu     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    jg         wloop
-    ret
-  }
-}
-
-#ifdef HAS_ARGBSHUFFLEROW_AVX2
-__declspec(naked) __declspec(align(16))
-void ARGBShuffleRow_AVX2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix) {
-  __asm {
-    mov        eax, [esp + 4]     // src_argb
-    mov        edx, [esp + 8]     // dst_argb
-    mov        ecx, [esp + 12]    // shuffler
-    vbroadcastf128 ymm5, [ecx]    // same shuffle in high as low.
-    mov        ecx, [esp + 16]    // pix
-
-    align      4
-  wloop:
-    vmovdqu    ymm0, [eax]
-    vmovdqu    ymm1, [eax + 32]
-    lea        eax, [eax + 64]
-    vpshufb    ymm0, ymm0, ymm5
-    vpshufb    ymm1, ymm1, ymm5
-    sub        ecx, 16
-    vmovdqu    [edx], ymm0
-    vmovdqu    [edx + 32], ymm1
-    lea        edx, [edx + 64]
-    jg         wloop
-
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBSHUFFLEROW_AVX2
-
-__declspec(naked) __declspec(align(16))
-void ARGBShuffleRow_SSE2(const uint8* src_argb, uint8* dst_argb,
-                         const uint8* shuffler, int pix) {
-  __asm {
-    push       ebx
-    push       esi
-    mov        eax, [esp + 8 + 4]    // src_argb
-    mov        edx, [esp + 8 + 8]    // dst_argb
-    mov        esi, [esp + 8 + 12]   // shuffler
-    mov        ecx, [esp + 8 + 16]   // pix
-    pxor       xmm5, xmm5
-
-    mov        ebx, [esi]   // shuffler
-    cmp        ebx, 0x03000102
-    je         shuf_3012
-    cmp        ebx, 0x00010203
-    je         shuf_0123
-    cmp        ebx, 0x00030201
-    je         shuf_0321
-    cmp        ebx, 0x02010003
-    je         shuf_2103
-
-  // TODO(fbarchard): Use one source pointer and 3 offsets.
-  shuf_any1:
-    movzx      ebx, byte ptr [esi]
-    movzx      ebx, byte ptr [eax + ebx]
-    mov        [edx], bl
-    movzx      ebx, byte ptr [esi + 1]
-    movzx      ebx, byte ptr [eax + ebx]
-    mov        [edx + 1], bl
-    movzx      ebx, byte ptr [esi + 2]
-    movzx      ebx, byte ptr [eax + ebx]
-    mov        [edx + 2], bl
-    movzx      ebx, byte ptr [esi + 3]
-    movzx      ebx, byte ptr [eax + ebx]
-    mov        [edx + 3], bl
-    lea        eax, [eax + 4]
-    lea        edx, [edx + 4]
-    sub        ecx, 1
-    jg         shuf_any1
-    jmp        shuf99
-
-    align      4
-  shuf_0123:
-    movdqu     xmm0, [eax]
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm5
-    punpckhbw  xmm1, xmm5
-    pshufhw    xmm0, xmm0, 01Bh   // 1B = 00011011 = 0x0123 = BGRAToARGB
-    pshuflw    xmm0, xmm0, 01Bh
-    pshufhw    xmm1, xmm1, 01Bh
-    pshuflw    xmm1, xmm1, 01Bh
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         shuf_0123
-    jmp        shuf99
-
-    align      4
-  shuf_0321:
-    movdqu     xmm0, [eax]
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm5
-    punpckhbw  xmm1, xmm5
-    pshufhw    xmm0, xmm0, 039h   // 39 = 00111001 = 0x0321 = RGBAToARGB
-    pshuflw    xmm0, xmm0, 039h
-    pshufhw    xmm1, xmm1, 039h
-    pshuflw    xmm1, xmm1, 039h
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         shuf_0321
-    jmp        shuf99
-
-    align      4
-  shuf_2103:
-    movdqu     xmm0, [eax]
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm5
-    punpckhbw  xmm1, xmm5
-    pshufhw    xmm0, xmm0, 093h   // 93 = 10010011 = 0x2103 = ARGBToRGBA
-    pshuflw    xmm0, xmm0, 093h
-    pshufhw    xmm1, xmm1, 093h
-    pshuflw    xmm1, xmm1, 093h
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         shuf_2103
-    jmp        shuf99
-
-    align      4
-  shuf_3012:
-    movdqu     xmm0, [eax]
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm5
-    punpckhbw  xmm1, xmm5
-    pshufhw    xmm0, xmm0, 0C6h   // C6 = 11000110 = 0x3012 = ABGRToARGB
-    pshuflw    xmm0, xmm0, 0C6h
-    pshufhw    xmm1, xmm1, 0C6h
-    pshuflw    xmm1, xmm1, 0C6h
-    packuswb   xmm0, xmm1
-    sub        ecx, 4
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         shuf_3012
-
-  shuf99:
-    pop        esi
-    pop        ebx
-    ret
-  }
-}
-
-// YUY2 - Macro-pixel = 2 image pixels
-// Y0U0Y1V0....Y2U2Y3V2...Y4U4Y5V4....
-
-// UYVY - Macro-pixel = 2 image pixels
-// U0Y0V0Y1
-
-__declspec(naked) __declspec(align(16))
-void I422ToYUY2Row_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_frame, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_y
-    mov        esi, [esp + 8 + 8]    // src_u
-    mov        edx, [esp + 8 + 12]   // src_v
-    mov        edi, [esp + 8 + 16]   // dst_frame
-    mov        ecx, [esp + 8 + 20]   // width
-    sub        edx, esi
-
-    align      4
-  convertloop:
-    movq       xmm2, qword ptr [esi] // U
-    movq       xmm3, qword ptr [esi + edx] // V
-    lea        esi, [esi + 8]
-    punpcklbw  xmm2, xmm3 // UV
-    movdqu     xmm0, [eax] // Y
-    lea        eax, [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm2 // YUYV
-    punpckhbw  xmm1, xmm2
-    movdqu     [edi], xmm0
-    movdqu     [edi + 16], xmm1
-    lea        edi, [edi + 32]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-__declspec(naked) __declspec(align(16))
-void I422ToUYVYRow_SSE2(const uint8* src_y,
-                        const uint8* src_u,
-                        const uint8* src_v,
-                        uint8* dst_frame, int width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_y
-    mov        esi, [esp + 8 + 8]    // src_u
-    mov        edx, [esp + 8 + 12]   // src_v
-    mov        edi, [esp + 8 + 16]   // dst_frame
-    mov        ecx, [esp + 8 + 20]   // width
-    sub        edx, esi
-
-    align      4
-  convertloop:
-    movq       xmm2, qword ptr [esi] // U
-    movq       xmm3, qword ptr [esi + edx] // V
-    lea        esi, [esi + 8]
-    punpcklbw  xmm2, xmm3 // UV
-    movdqu     xmm0, [eax] // Y
-    movdqa     xmm1, xmm2
-    lea        eax, [eax + 16]
-    punpcklbw  xmm1, xmm0 // UYVY
-    punpckhbw  xmm2, xmm0
-    movdqu     [edi], xmm1
-    movdqu     [edi + 16], xmm2
-    lea        edi, [edi + 32]
-    sub        ecx, 16
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-#ifdef HAS_ARGBPOLYNOMIALROW_SSE2
-__declspec(naked) __declspec(align(16))
-void ARGBPolynomialRow_SSE2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   /* src_argb */
-    mov        edx, [esp + 4 + 8]   /* dst_argb */
-    mov        esi, [esp + 4 + 12]  /* poly */
-    mov        ecx, [esp + 4 + 16]  /* width */
-    pxor       xmm3, xmm3  // 0 constant for zero extending bytes to ints.
-
-    // 2 pixel loop.
-    align      4
- convertloop:
-//    pmovzxbd  xmm0, dword ptr [eax]  // BGRA pixel
-//    pmovzxbd  xmm4, dword ptr [eax + 4]  // BGRA pixel
-    movq       xmm0, qword ptr [eax]  // BGRABGRA
-    lea        eax, [eax + 8]
-    punpcklbw  xmm0, xmm3
-    movdqa     xmm4, xmm0
-    punpcklwd  xmm0, xmm3  // pixel 0
-    punpckhwd  xmm4, xmm3  // pixel 1
-    cvtdq2ps   xmm0, xmm0  // 4 floats
-    cvtdq2ps   xmm4, xmm4
-    movdqa     xmm1, xmm0  // X
-    movdqa     xmm5, xmm4
-    mulps      xmm0, [esi + 16]  // C1 * X
-    mulps      xmm4, [esi + 16]
-    addps      xmm0, [esi]  // result = C0 + C1 * X
-    addps      xmm4, [esi]
-    movdqa     xmm2, xmm1
-    movdqa     xmm6, xmm5
-    mulps      xmm2, xmm1  // X * X
-    mulps      xmm6, xmm5
-    mulps      xmm1, xmm2  // X * X * X
-    mulps      xmm5, xmm6
-    mulps      xmm2, [esi + 32]  // C2 * X * X
-    mulps      xmm6, [esi + 32]
-    mulps      xmm1, [esi + 48]  // C3 * X * X * X
-    mulps      xmm5, [esi + 48]
-    addps      xmm0, xmm2  // result += C2 * X * X
-    addps      xmm4, xmm6
-    addps      xmm0, xmm1  // result += C3 * X * X * X
-    addps      xmm4, xmm5
-    cvttps2dq  xmm0, xmm0
-    cvttps2dq  xmm4, xmm4
-    packuswb   xmm0, xmm4
-    packuswb   xmm0, xmm0
-    sub        ecx, 2
-    movq       qword ptr [edx], xmm0
-    lea        edx, [edx + 8]
-    jg         convertloop
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBPOLYNOMIALROW_SSE2
-
-#ifdef HAS_ARGBPOLYNOMIALROW_AVX2
-__declspec(naked) __declspec(align(16))
-void ARGBPolynomialRow_AVX2(const uint8* src_argb,
-                            uint8* dst_argb, const float* poly,
-                            int width) {
-  __asm {
-    mov        eax, [esp + 4]   /* src_argb */
-    mov        edx, [esp + 8]   /* dst_argb */
-    mov        ecx, [esp + 12]   /* poly */
-    vbroadcastf128 ymm4, [ecx]       // C0
-    vbroadcastf128 ymm5, [ecx + 16]  // C1
-    vbroadcastf128 ymm6, [ecx + 32]  // C2
-    vbroadcastf128 ymm7, [ecx + 48]  // C3
-    mov        ecx, [esp + 16]  /* width */
-
-    // 2 pixel loop.
-    align      4
- convertloop:
-    vpmovzxbd   ymm0, qword ptr [eax]  // 2 BGRA pixels
-    lea         eax, [eax + 8]
-    vcvtdq2ps   ymm0, ymm0        // X 8 floats
-    vmulps      ymm2, ymm0, ymm0  // X * X
-    vmulps      ymm3, ymm0, ymm7  // C3 * X
-    vfmadd132ps ymm0, ymm4, ymm5  // result = C0 + C1 * X
-    vfmadd231ps ymm0, ymm2, ymm6  // result += C2 * X * X
-    vfmadd231ps ymm0, ymm2, ymm3  // result += C3 * X * X * X
-    vcvttps2dq  ymm0, ymm0
-    vpackusdw   ymm0, ymm0, ymm0  // b0g0r0a0_00000000_b0g0r0a0_00000000
-    vpermq      ymm0, ymm0, 0xd8  // b0g0r0a0_b0g0r0a0_00000000_00000000
-    vpackuswb   xmm0, xmm0, xmm0  // bgrabgra_00000000_00000000_00000000
-    sub         ecx, 2
-    vmovq       qword ptr [edx], xmm0
-    lea         edx, [edx + 8]
-    jg          convertloop
-    vzeroupper
-    ret
-  }
-}
-#endif  // HAS_ARGBPOLYNOMIALROW_AVX2
-
-#ifdef HAS_ARGBCOLORTABLEROW_X86
-// Tranform ARGB pixels with color table.
-__declspec(naked) __declspec(align(16))
-void ARGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb,
-                           int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   /* dst_argb */
-    mov        esi, [esp + 4 + 8]   /* table_argb */
-    mov        ecx, [esp + 4 + 12]  /* width */
-
-    // 1 pixel loop.
-    align      4
-  convertloop:
-    movzx      edx, byte ptr [eax]
-    lea        eax, [eax + 4]
-    movzx      edx, byte ptr [esi + edx * 4]
-    mov        byte ptr [eax - 4], dl
-    movzx      edx, byte ptr [eax - 4 + 1]
-    movzx      edx, byte ptr [esi + edx * 4 + 1]
-    mov        byte ptr [eax - 4 + 1], dl
-    movzx      edx, byte ptr [eax - 4 + 2]
-    movzx      edx, byte ptr [esi + edx * 4 + 2]
-    mov        byte ptr [eax - 4 + 2], dl
-    movzx      edx, byte ptr [eax - 4 + 3]
-    movzx      edx, byte ptr [esi + edx * 4 + 3]
-    mov        byte ptr [eax - 4 + 3], dl
-    dec        ecx
-    jg         convertloop
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBCOLORTABLEROW_X86
-
-#ifdef HAS_RGBCOLORTABLEROW_X86
-// Tranform RGB pixels with color table.
-__declspec(naked) __declspec(align(16))
-void RGBColorTableRow_X86(uint8* dst_argb, const uint8* table_argb, int width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]   /* dst_argb */
-    mov        esi, [esp + 4 + 8]   /* table_argb */
-    mov        ecx, [esp + 4 + 12]  /* width */
-
-    // 1 pixel loop.
-    align      4
-  convertloop:
-    movzx      edx, byte ptr [eax]
-    lea        eax, [eax + 4]
-    movzx      edx, byte ptr [esi + edx * 4]
-    mov        byte ptr [eax - 4], dl
-    movzx      edx, byte ptr [eax - 4 + 1]
-    movzx      edx, byte ptr [esi + edx * 4 + 1]
-    mov        byte ptr [eax - 4 + 1], dl
-    movzx      edx, byte ptr [eax - 4 + 2]
-    movzx      edx, byte ptr [esi + edx * 4 + 2]
-    mov        byte ptr [eax - 4 + 2], dl
-    dec        ecx
-    jg         convertloop
-
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_RGBCOLORTABLEROW_X86
-
-#ifdef HAS_ARGBLUMACOLORTABLEROW_SSSE3
-// Tranform RGB pixels with luma table.
-__declspec(naked) __declspec(align(16))
-void ARGBLumaColorTableRow_SSSE3(const uint8* src_argb, uint8* dst_argb,
-                                 int width,
-                                 const uint8* luma, uint32 lumacoeff) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]   /* src_argb */
-    mov        edi, [esp + 8 + 8]   /* dst_argb */
-    mov        ecx, [esp + 8 + 12]  /* width */
-    movd       xmm2, dword ptr [esp + 8 + 16]  // luma table
-    movd       xmm3, dword ptr [esp + 8 + 20]  // lumacoeff
-    pshufd     xmm2, xmm2, 0
-    pshufd     xmm3, xmm3, 0
-    pcmpeqb    xmm4, xmm4        // generate mask 0xff00ff00
-    psllw      xmm4, 8
-    pxor       xmm5, xmm5
-
-    // 4 pixel loop.
-    align      4
-  convertloop:
-    movdqu     xmm0, qword ptr [eax]      // generate luma ptr
-    pmaddubsw  xmm0, xmm3
-    phaddw     xmm0, xmm0
-    pand       xmm0, xmm4  // mask out low bits
-    punpcklwd  xmm0, xmm5
-    paddd      xmm0, xmm2  // add table base
-    movd       esi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // 00111001 to rotate right 32
-
-    movzx      edx, byte ptr [eax]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi], dl
-    movzx      edx, byte ptr [eax + 1]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 1], dl
-    movzx      edx, byte ptr [eax + 2]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 2], dl
-    movzx      edx, byte ptr [eax + 3]  // copy alpha.
-    mov        byte ptr [edi + 3], dl
-
-    movd       esi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // 00111001 to rotate right 32
-
-    movzx      edx, byte ptr [eax + 4]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 4], dl
-    movzx      edx, byte ptr [eax + 5]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 5], dl
-    movzx      edx, byte ptr [eax + 6]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 6], dl
-    movzx      edx, byte ptr [eax + 7]  // copy alpha.
-    mov        byte ptr [edi + 7], dl
-
-    movd       esi, xmm0
-    pshufd     xmm0, xmm0, 0x39  // 00111001 to rotate right 32
-
-    movzx      edx, byte ptr [eax + 8]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 8], dl
-    movzx      edx, byte ptr [eax + 9]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 9], dl
-    movzx      edx, byte ptr [eax + 10]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 10], dl
-    movzx      edx, byte ptr [eax + 11]  // copy alpha.
-    mov        byte ptr [edi + 11], dl
-
-    movd       esi, xmm0
-
-    movzx      edx, byte ptr [eax + 12]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 12], dl
-    movzx      edx, byte ptr [eax + 13]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 13], dl
-    movzx      edx, byte ptr [eax + 14]
-    movzx      edx, byte ptr [esi + edx]
-    mov        byte ptr [edi + 14], dl
-    movzx      edx, byte ptr [eax + 15]  // copy alpha.
-    mov        byte ptr [edi + 15], dl
-
-    sub        ecx, 4
-    lea        eax, [eax + 16]
-    lea        edi, [edi + 16]
-    jg         convertloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-#endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
-
-#endif  // !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/row_x86.asm b/drivers/theoraplayer/src/YUV/libyuv/src/row_x86.asm
deleted file mode 100755
index 0cb326f8e5..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/row_x86.asm
+++ /dev/null
@@ -1,146 +0,0 @@
-;
-; Copyright 2012 The LibYuv Project Authors. All rights reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-%ifdef __YASM_VERSION_ID__
-%if __YASM_VERSION_ID__ < 01020000h
-%error AVX2 is supported only by yasm 1.2.0 or later.
-%endif
-%endif
-%include "x86inc.asm"
-
-SECTION .text
-
-; cglobal numeric constants are parameters, gpr regs, mm regs
-
-; void YUY2ToYRow_SSE2(const uint8* src_yuy2, uint8* dst_y, int pix)
-
-%macro YUY2TOYROW 2-3
-cglobal %1ToYRow%3, 3, 3, 3, src_yuy2, dst_y, pix
-%ifidn %1,YUY2
-    pcmpeqb    m2, m2, m2        ; generate mask 0x00ff00ff
-    psrlw      m2, m2, 8
-%endif
-
-    ALIGN      4
-.convertloop:
-    mov%2      m0, [src_yuy2q]
-    mov%2      m1, [src_yuy2q + mmsize]
-    lea        src_yuy2q, [src_yuy2q + mmsize * 2]
-%ifidn %1,YUY2
-    pand       m0, m0, m2   ; YUY2 even bytes are Y
-    pand       m1, m1, m2
-%else
-    psrlw      m0, m0, 8    ; UYVY odd bytes are Y
-    psrlw      m1, m1, 8
-%endif
-    packuswb   m0, m0, m1
-%if cpuflag(AVX2)
-    vpermq     m0, m0, 0xd8
-%endif
-    sub        pixd, mmsize
-    mov%2      [dst_yq], m0
-    lea        dst_yq, [dst_yq + mmsize]
-    jg         .convertloop
-    REP_RET
-%endmacro
-
-; TODO(fbarchard): Remove MMX.  Add SSSE3 pshufb version.
-INIT_MMX MMX
-YUY2TOYROW YUY2,a,
-YUY2TOYROW YUY2,u,_Unaligned
-YUY2TOYROW UYVY,a,
-YUY2TOYROW UYVY,u,_Unaligned
-INIT_XMM SSE2
-YUY2TOYROW YUY2,a,
-YUY2TOYROW YUY2,u,_Unaligned
-YUY2TOYROW UYVY,a,
-YUY2TOYROW UYVY,u,_Unaligned
-INIT_YMM AVX2
-YUY2TOYROW YUY2,a,
-YUY2TOYROW UYVY,a,
-
-; void SplitUVRow_SSE2(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int pix)
-
-%macro SplitUVRow 1-2
-cglobal SplitUVRow%2, 4, 4, 5, src_uv, dst_u, dst_v, pix
-    pcmpeqb    m4, m4, m4        ; generate mask 0x00ff00ff
-    psrlw      m4, m4, 8
-    sub        dst_vq, dst_uq
-
-    ALIGN      4
-.convertloop:
-    mov%1      m0, [src_uvq]
-    mov%1      m1, [src_uvq + mmsize]
-    lea        src_uvq, [src_uvq + mmsize * 2]
-    psrlw      m2, m0, 8         ; odd bytes
-    psrlw      m3, m1, 8
-    pand       m0, m0, m4        ; even bytes
-    pand       m1, m1, m4
-    packuswb   m0, m0, m1
-    packuswb   m2, m2, m3
-%if cpuflag(AVX2)
-    vpermq     m0, m0, 0xd8
-    vpermq     m2, m2, 0xd8
-%endif
-    mov%1      [dst_uq], m0
-    mov%1      [dst_uq + dst_vq], m2
-    lea        dst_uq, [dst_uq + mmsize]
-    sub        pixd, mmsize
-    jg         .convertloop
-    REP_RET
-%endmacro
-
-INIT_MMX MMX
-SplitUVRow a,
-SplitUVRow u,_Unaligned
-INIT_XMM SSE2
-SplitUVRow a,
-SplitUVRow u,_Unaligned
-INIT_YMM AVX2
-SplitUVRow a,
-
-; void MergeUVRow_SSE2(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-;                      int width);
-
-%macro MergeUVRow_ 1-2
-cglobal MergeUVRow_%2, 4, 4, 3, src_u, src_v, dst_uv, pix
-    sub        src_vq, src_uq
-
-    ALIGN      4
-.convertloop:
-    mov%1      m0, [src_uq]
-    mov%1      m1, [src_vq]
-    lea        src_uq, [src_uq + mmsize]
-    punpcklbw  m2, m0, m1       // first 8 UV pairs
-    punpckhbw  m0, m0, m1       // next 8 UV pairs
-%if cpuflag(AVX2)
-    vperm2i128 m1, m2, m0, 0x20  // low 128 of ymm2 and low 128 of ymm0
-    vperm2i128 m2, m2, m0, 0x31  // high 128 of ymm2 and high 128 of ymm0
-    mov%1      [dst_uvq], m1
-    mov%1      [dst_uvq + mmsize], m2
-%else
-    mov%1      [dst_uvq], m2
-    mov%1      [dst_uvq + mmsize], m0
-%endif
-    lea        dst_uvq, [dst_uvq + mmsize * 2]
-    sub        pixd, mmsize
-    jg         .convertloop
-    REP_RET
-%endmacro
-
-INIT_MMX MMX
-MergeUVRow_ a,
-MergeUVRow_ u,_Unaligned
-INIT_XMM SSE2
-MergeUVRow_ a,
-MergeUVRow_ u,_Unaligned
-INIT_YMM AVX2
-MergeUVRow_ a,
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale.cc
deleted file mode 100755
index b3893cc00c..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale.cc
+++ /dev/null
@@ -1,926 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/scale.h"
-
-#include <assert.h>
-#include <string.h>
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/planar_functions.h"  // For CopyPlane
-#include "libyuv/row.h"
-#include "libyuv/scale_row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// Remove this macro if OVERREAD is safe.
-#define AVOID_OVERREAD 1
-
-static __inline int Abs(int v) {
-  return v >= 0 ? v : -v;
-}
-
-#define SUBSAMPLE(v, a, s) (v < 0) ? (-((-v + a) >> s)) : ((v + a) >> s)
-
-// Scale plane, 1/2
-// This is an optimized version for scaling down a plane to 1/2 of
-// its original size.
-
-static void ScalePlaneDown2(int src_width, int src_height,
-                            int dst_width, int dst_height,
-                            int src_stride, int dst_stride,
-                            const uint8* src_ptr, uint8* dst_ptr,
-                            enum FilterMode filtering) {
-  int y;
-  void (*ScaleRowDown2)(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) =
-    filtering == kFilterNone ? ScaleRowDown2_C :
-        (filtering == kFilterLinear ? ScaleRowDown2Linear_C :
-        ScaleRowDown2Box_C);
-  int row_stride = src_stride << 1;
-  if (!filtering) {
-    src_ptr += src_stride;  // Point to odd rows.
-    src_stride = 0;
-  }
-
-#if defined(HAS_SCALEROWDOWN2_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(dst_width, 16)) {
-    ScaleRowDown2 = filtering ? ScaleRowDown2Box_NEON : ScaleRowDown2_NEON;
-  }
-#elif defined(HAS_SCALEROWDOWN2_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 16)) {
-    ScaleRowDown2 = filtering == kFilterNone ? ScaleRowDown2_Unaligned_SSE2 :
-        (filtering == kFilterLinear ? ScaleRowDown2Linear_Unaligned_SSE2 :
-        ScaleRowDown2Box_Unaligned_SSE2);
-    if (IS_ALIGNED(src_ptr, 16) &&
-        IS_ALIGNED(src_stride, 16) && IS_ALIGNED(row_stride, 16) &&
-        IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleRowDown2 = filtering == kFilterNone ? ScaleRowDown2_SSE2 :
-          (filtering == kFilterLinear ? ScaleRowDown2Linear_SSE2 :
-          ScaleRowDown2Box_SSE2);
-    }
-  }
-#elif defined(HAS_SCALEROWDOWN2_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(src_ptr, 4) &&
-      IS_ALIGNED(src_stride, 4) && IS_ALIGNED(row_stride, 4) &&
-      IS_ALIGNED(dst_ptr, 4) && IS_ALIGNED(dst_stride, 4)) {
-    ScaleRowDown2 = filtering ?
-        ScaleRowDown2Box_MIPS_DSPR2 : ScaleRowDown2_MIPS_DSPR2;
-  }
-#endif
-
-  if (filtering == kFilterLinear) {
-    src_stride = 0;
-  }
-  // TODO(fbarchard): Loop through source height to allow odd height.
-  for (y = 0; y < dst_height; ++y) {
-    ScaleRowDown2(src_ptr, src_stride, dst_ptr, dst_width);
-    src_ptr += row_stride;
-    dst_ptr += dst_stride;
-  }
-}
-
-// Scale plane, 1/4
-// This is an optimized version for scaling down a plane to 1/4 of
-// its original size.
-
-static void ScalePlaneDown4(int src_width, int src_height,
-                            int dst_width, int dst_height,
-                            int src_stride, int dst_stride,
-                            const uint8* src_ptr, uint8* dst_ptr,
-                            enum FilterMode filtering) {
-  int y;
-  void (*ScaleRowDown4)(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) =
-      filtering ? ScaleRowDown4Box_C : ScaleRowDown4_C;
-  int row_stride = src_stride << 2;
-  if (!filtering) {
-    src_ptr += src_stride * 2;  // Point to row 2.
-    src_stride = 0;
-  }
-#if defined(HAS_SCALEROWDOWN4_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(dst_width, 8)) {
-    ScaleRowDown4 = filtering ? ScaleRowDown4Box_NEON : ScaleRowDown4_NEON;
-  }
-#elif defined(HAS_SCALEROWDOWN4_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) &&
-      IS_ALIGNED(dst_width, 8) && IS_ALIGNED(row_stride, 16) &&
-      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-    ScaleRowDown4 = filtering ? ScaleRowDown4Box_SSE2 : ScaleRowDown4_SSE2;
-  }
-#elif defined(HAS_SCALEROWDOWN4_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(row_stride, 4) &&
-      IS_ALIGNED(src_ptr, 4) && IS_ALIGNED(src_stride, 4) &&
-      IS_ALIGNED(dst_ptr, 4) && IS_ALIGNED(dst_stride, 4)) {
-    ScaleRowDown4 = filtering ?
-        ScaleRowDown4Box_MIPS_DSPR2 : ScaleRowDown4_MIPS_DSPR2;
-  }
-#endif
-
-  if (filtering == kFilterLinear) {
-    src_stride = 0;
-  }
-  for (y = 0; y < dst_height; ++y) {
-    ScaleRowDown4(src_ptr, src_stride, dst_ptr, dst_width);
-    src_ptr += row_stride;
-    dst_ptr += dst_stride;
-  }
-}
-
-// Scale plane down, 3/4
-
-static void ScalePlaneDown34(int src_width, int src_height,
-                             int dst_width, int dst_height,
-                             int src_stride, int dst_stride,
-                             const uint8* src_ptr, uint8* dst_ptr,
-                             enum FilterMode filtering) {
-  int y;
-  void (*ScaleRowDown34_0)(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-  void (*ScaleRowDown34_1)(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-  const int filter_stride = (filtering == kFilterLinear) ? 0 : src_stride;
-  assert(dst_width % 3 == 0);
-  if (!filtering) {
-    ScaleRowDown34_0 = ScaleRowDown34_C;
-    ScaleRowDown34_1 = ScaleRowDown34_C;
-  } else {
-    ScaleRowDown34_0 = ScaleRowDown34_0_Box_C;
-    ScaleRowDown34_1 = ScaleRowDown34_1_Box_C;
-  }
-#if defined(HAS_SCALEROWDOWN34_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && (dst_width % 24 == 0)) {
-    if (!filtering) {
-      ScaleRowDown34_0 = ScaleRowDown34_NEON;
-      ScaleRowDown34_1 = ScaleRowDown34_NEON;
-    } else {
-      ScaleRowDown34_0 = ScaleRowDown34_0_Box_NEON;
-      ScaleRowDown34_1 = ScaleRowDown34_1_Box_NEON;
-    }
-  }
-#endif
-#if defined(HAS_SCALEROWDOWN34_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && (dst_width % 24 == 0) &&
-      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-    if (!filtering) {
-      ScaleRowDown34_0 = ScaleRowDown34_SSSE3;
-      ScaleRowDown34_1 = ScaleRowDown34_SSSE3;
-    } else {
-      ScaleRowDown34_0 = ScaleRowDown34_0_Box_SSSE3;
-      ScaleRowDown34_1 = ScaleRowDown34_1_Box_SSSE3;
-    }
-  }
-#endif
-#if defined(HAS_SCALEROWDOWN34_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && (dst_width % 24 == 0) &&
-      IS_ALIGNED(src_ptr, 4) && IS_ALIGNED(src_stride, 4) &&
-      IS_ALIGNED(dst_ptr, 4) && IS_ALIGNED(dst_stride, 4)) {
-    if (!filtering) {
-      ScaleRowDown34_0 = ScaleRowDown34_MIPS_DSPR2;
-      ScaleRowDown34_1 = ScaleRowDown34_MIPS_DSPR2;
-    } else {
-      ScaleRowDown34_0 = ScaleRowDown34_0_Box_MIPS_DSPR2;
-      ScaleRowDown34_1 = ScaleRowDown34_1_Box_MIPS_DSPR2;
-    }
-  }
-#endif
-
-  for (y = 0; y < dst_height - 2; y += 3) {
-    ScaleRowDown34_0(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride;
-    dst_ptr += dst_stride;
-    ScaleRowDown34_1(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride;
-    dst_ptr += dst_stride;
-    ScaleRowDown34_0(src_ptr + src_stride, -filter_stride,
-                     dst_ptr, dst_width);
-    src_ptr += src_stride * 2;
-    dst_ptr += dst_stride;
-  }
-
-  // Remainder 1 or 2 rows with last row vertically unfiltered
-  if ((dst_height % 3) == 2) {
-    ScaleRowDown34_0(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride;
-    dst_ptr += dst_stride;
-    ScaleRowDown34_1(src_ptr, 0, dst_ptr, dst_width);
-  } else if ((dst_height % 3) == 1) {
-    ScaleRowDown34_0(src_ptr, 0, dst_ptr, dst_width);
-  }
-}
-
-
-// Scale plane, 3/8
-// This is an optimized version for scaling down a plane to 3/8
-// of its original size.
-//
-// Uses box filter arranges like this
-// aaabbbcc -> abc
-// aaabbbcc    def
-// aaabbbcc    ghi
-// dddeeeff
-// dddeeeff
-// dddeeeff
-// ggghhhii
-// ggghhhii
-// Boxes are 3x3, 2x3, 3x2 and 2x2
-
-static void ScalePlaneDown38(int src_width, int src_height,
-                             int dst_width, int dst_height,
-                             int src_stride, int dst_stride,
-                             const uint8* src_ptr, uint8* dst_ptr,
-                             enum FilterMode filtering) {
-  int y;
-  void (*ScaleRowDown38_3)(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-  void (*ScaleRowDown38_2)(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width);
-  const int filter_stride = (filtering == kFilterLinear) ? 0 : src_stride;
-  assert(dst_width % 3 == 0);
-  if (!filtering) {
-    ScaleRowDown38_3 = ScaleRowDown38_C;
-    ScaleRowDown38_2 = ScaleRowDown38_C;
-  } else {
-    ScaleRowDown38_3 = ScaleRowDown38_3_Box_C;
-    ScaleRowDown38_2 = ScaleRowDown38_2_Box_C;
-  }
-#if defined(HAS_SCALEROWDOWN38_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && (dst_width % 12 == 0)) {
-    if (!filtering) {
-      ScaleRowDown38_3 = ScaleRowDown38_NEON;
-      ScaleRowDown38_2 = ScaleRowDown38_NEON;
-    } else {
-      ScaleRowDown38_3 = ScaleRowDown38_3_Box_NEON;
-      ScaleRowDown38_2 = ScaleRowDown38_2_Box_NEON;
-    }
-  }
-#elif defined(HAS_SCALEROWDOWN38_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && (dst_width % 24 == 0) &&
-      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-    if (!filtering) {
-      ScaleRowDown38_3 = ScaleRowDown38_SSSE3;
-      ScaleRowDown38_2 = ScaleRowDown38_SSSE3;
-    } else {
-      ScaleRowDown38_3 = ScaleRowDown38_3_Box_SSSE3;
-      ScaleRowDown38_2 = ScaleRowDown38_2_Box_SSSE3;
-    }
-  }
-#elif defined(HAS_SCALEROWDOWN38_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && (dst_width % 12 == 0) &&
-      IS_ALIGNED(src_ptr, 4) && IS_ALIGNED(src_stride, 4) &&
-      IS_ALIGNED(dst_ptr, 4) && IS_ALIGNED(dst_stride, 4)) {
-    if (!filtering) {
-      ScaleRowDown38_3 = ScaleRowDown38_MIPS_DSPR2;
-      ScaleRowDown38_2 = ScaleRowDown38_MIPS_DSPR2;
-    } else {
-      ScaleRowDown38_3 = ScaleRowDown38_3_Box_MIPS_DSPR2;
-      ScaleRowDown38_2 = ScaleRowDown38_2_Box_MIPS_DSPR2;
-    }
-  }
-#endif
-
-  for (y = 0; y < dst_height - 2; y += 3) {
-    ScaleRowDown38_3(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride * 3;
-    dst_ptr += dst_stride;
-    ScaleRowDown38_3(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride * 3;
-    dst_ptr += dst_stride;
-    ScaleRowDown38_2(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride * 2;
-    dst_ptr += dst_stride;
-  }
-
-  // Remainder 1 or 2 rows with last row vertically unfiltered
-  if ((dst_height % 3) == 2) {
-    ScaleRowDown38_3(src_ptr, filter_stride, dst_ptr, dst_width);
-    src_ptr += src_stride * 3;
-    dst_ptr += dst_stride;
-    ScaleRowDown38_3(src_ptr, 0, dst_ptr, dst_width);
-  } else if ((dst_height % 3) == 1) {
-    ScaleRowDown38_3(src_ptr, 0, dst_ptr, dst_width);
-  }
-}
-
-static __inline uint32 SumBox(int iboxwidth, int iboxheight,
-                              ptrdiff_t src_stride, const uint8* src_ptr) {
-  uint32 sum = 0u;
-  int y;
-  assert(iboxwidth > 0);
-  assert(iboxheight > 0);
-  for (y = 0; y < iboxheight; ++y) {
-    int x;
-    for (x = 0; x < iboxwidth; ++x) {
-      sum += src_ptr[x];
-    }
-    src_ptr += src_stride;
-  }
-  return sum;
-}
-
-static void ScalePlaneBoxRow_C(int dst_width, int boxheight,
-                               int x, int dx, ptrdiff_t src_stride,
-                               const uint8* src_ptr, uint8* dst_ptr) {
-  int i;
-  int boxwidth;
-  for (i = 0; i < dst_width; ++i) {
-    int ix = x >> 16;
-    x += dx;
-    boxwidth = (x >> 16) - ix;
-    *dst_ptr++ = SumBox(boxwidth, boxheight, src_stride, src_ptr + ix) /
-        (boxwidth * boxheight);
-  }
-}
-
-static __inline uint32 SumPixels(int iboxwidth, const uint16* src_ptr) {
-  uint32 sum = 0u;
-  int x;
-  assert(iboxwidth > 0);
-  for (x = 0; x < iboxwidth; ++x) {
-    sum += src_ptr[x];
-  }
-  return sum;
-}
-
-static void ScaleAddCols2_C(int dst_width, int boxheight, int x, int dx,
-                            const uint16* src_ptr, uint8* dst_ptr) {
-  int i;
-  int scaletbl[2];
-  int minboxwidth = (dx >> 16);
-  int* scaleptr = scaletbl - minboxwidth;
-  int boxwidth;
-  scaletbl[0] = 65536 / (minboxwidth * boxheight);
-  scaletbl[1] = 65536 / ((minboxwidth + 1) * boxheight);
-  for (i = 0; i < dst_width; ++i) {
-    int ix = x >> 16;
-    x += dx;
-    boxwidth = (x >> 16) - ix;
-    *dst_ptr++ = SumPixels(boxwidth, src_ptr + ix) * scaleptr[boxwidth] >> 16;
-  }
-}
-
-static void ScaleAddCols1_C(int dst_width, int boxheight, int x, int dx,
-                            const uint16* src_ptr, uint8* dst_ptr) {
-  int boxwidth = (dx >> 16);
-  int scaleval = 65536 / (boxwidth * boxheight);
-  int i;
-  for (i = 0; i < dst_width; ++i) {
-    *dst_ptr++ = SumPixels(boxwidth, src_ptr + x) * scaleval >> 16;
-    x += boxwidth;
-  }
-}
-
-// Scale plane down to any dimensions, with interpolation.
-// (boxfilter).
-//
-// Same method as SimpleScale, which is fixed point, outputting
-// one pixel of destination using fixed point (16.16) to step
-// through source, sampling a box of pixel with simple
-// averaging.
-static void ScalePlaneBox(int src_width, int src_height,
-                          int dst_width, int dst_height,
-                          int src_stride, int dst_stride,
-                          const uint8* src_ptr, uint8* dst_ptr) {
-  int j;
-  // Initial source x/y coordinate and step values as 16.16 fixed point.
-  int x = 0;
-  int y = 0;
-  int dx = 0;
-  int dy = 0;
-  const int max_y = (src_height << 16);
-  ScaleSlope(src_width, src_height, dst_width, dst_height, kFilterBox,
-             &x, &y, &dx, &dy);
-  src_width = Abs(src_width);
-  // TODO(fbarchard): Remove this and make AddRows handle boxheight 1.
-  if (!IS_ALIGNED(src_width, 16) || dst_height * 2 > src_height) {
-    uint8* dst = dst_ptr;
-    int j;
-    for (j = 0; j < dst_height; ++j) {
-      int boxheight;
-      int iy = y >> 16;
-      const uint8* src = src_ptr + iy * src_stride;
-      y += dy;
-      if (y > max_y) {
-        y = max_y;
-      }
-      boxheight = (y >> 16) - iy;
-      ScalePlaneBoxRow_C(dst_width, boxheight,
-                         x, dx, src_stride,
-                         src, dst);
-      dst += dst_stride;
-    }
-    return;
-  }
-  {
-    // Allocate a row buffer of uint16.
-    align_buffer_64(row16, src_width * 2);
-    void (*ScaleAddCols)(int dst_width, int boxheight, int x, int dx,
-        const uint16* src_ptr, uint8* dst_ptr) =
-        (dx & 0xffff) ? ScaleAddCols2_C: ScaleAddCols1_C;
-    void (*ScaleAddRows)(const uint8* src_ptr, ptrdiff_t src_stride,
-        uint16* dst_ptr, int src_width, int src_height) = ScaleAddRows_C;
-
-#if defined(HAS_SCALEADDROWS_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) &&
-#ifdef AVOID_OVERREAD
-        IS_ALIGNED(src_width, 16) &&
-#endif
-        IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-      ScaleAddRows = ScaleAddRows_SSE2;
-    }
-#endif
-
-    for (j = 0; j < dst_height; ++j) {
-      int boxheight;
-      int iy = y >> 16;
-      const uint8* src = src_ptr + iy * src_stride;
-      y += dy;
-      if (y > (src_height << 16)) {
-        y = (src_height << 16);
-      }
-      boxheight = (y >> 16) - iy;
-      ScaleAddRows(src, src_stride, (uint16*)(row16),
-                 src_width, boxheight);
-      ScaleAddCols(dst_width, boxheight, x, dx, (uint16*)(row16),
-                 dst_ptr);
-      dst_ptr += dst_stride;
-    }
-    free_aligned_buffer_64(row16);
-  }
-}
-
-// Scale plane down with bilinear interpolation.
-void ScalePlaneBilinearDown(int src_width, int src_height,
-                            int dst_width, int dst_height,
-                            int src_stride, int dst_stride,
-                            const uint8* src_ptr, uint8* dst_ptr,
-                            enum FilterMode filtering) {
-  // Initial source x/y coordinate and step values as 16.16 fixed point.
-  int x = 0;
-  int y = 0;
-  int dx = 0;
-  int dy = 0;
-  // TODO(fbarchard): Consider not allocating row buffer for kFilterLinear.
-  // Allocate a row buffer.
-  align_buffer_64(row, src_width);
-
-  const int max_y = (src_height - 1) << 16;
-  int j;
-  void (*ScaleFilterCols)(uint8* dst_ptr, const uint8* src_ptr,
-      int dst_width, int x, int dx) =
-      (src_width >= 32768) ? ScaleFilterCols64_C : ScaleFilterCols_C;
-  void (*InterpolateRow)(uint8* dst_ptr, const uint8* src_ptr,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-  ScaleSlope(src_width, src_height, dst_width, dst_height, filtering,
-             &x, &y, &dx, &dy);
-  src_width = Abs(src_width);
-
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(src_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(src_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && src_width >= 32) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(src_width, 32)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(src_width, 16)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && src_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_MIPS_DSPR2;
-    if (IS_ALIGNED(src_width, 4)) {
-      InterpolateRow = InterpolateRow_MIPS_DSPR2;
-    }
-  }
-#endif
-
-
-#if defined(HAS_SCALEFILTERCOLS_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && src_width < 32768) {
-    ScaleFilterCols = ScaleFilterCols_SSSE3;
-  }
-#endif
-  if (y > max_y) {
-    y = max_y;
-  }
-
-  for (j = 0; j < dst_height; ++j) {
-    int yi = y >> 16;
-    const uint8* src = src_ptr + yi * src_stride;
-    if (filtering == kFilterLinear) {
-      ScaleFilterCols(dst_ptr, src, dst_width, x, dx);
-    } else {
-      int yf = (y >> 8) & 255;
-      InterpolateRow(row, src, src_stride, src_width, yf);
-      ScaleFilterCols(dst_ptr, row, dst_width, x, dx);
-    }
-    dst_ptr += dst_stride;
-    y += dy;
-    if (y > max_y) {
-      y = max_y;
-    }
-  }
-  free_aligned_buffer_64(row);
-}
-
-// Scale up down with bilinear interpolation.
-void ScalePlaneBilinearUp(int src_width, int src_height,
-                          int dst_width, int dst_height,
-                          int src_stride, int dst_stride,
-                          const uint8* src_ptr, uint8* dst_ptr,
-                          enum FilterMode filtering) {
-  int j;
-  // Initial source x/y coordinate and step values as 16.16 fixed point.
-  int x = 0;
-  int y = 0;
-  int dx = 0;
-  int dy = 0;
-  const int max_y = (src_height - 1) << 16;
-  void (*InterpolateRow)(uint8* dst_ptr, const uint8* src_ptr,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-  void (*ScaleFilterCols)(uint8* dst_ptr, const uint8* src_ptr,
-       int dst_width, int x, int dx) =
-       filtering ? ScaleFilterCols_C : ScaleCols_C;
-  ScaleSlope(src_width, src_height, dst_width, dst_height, filtering,
-             &x, &y, &dx, &dy);
-  src_width = Abs(src_width);
-
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && dst_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(dst_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && dst_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(dst_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && dst_width >= 32) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(dst_width, 32)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && dst_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(dst_width, 16)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_MIPS_DSPR2;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_MIPS_DSPR2;
-    }
-  }
-#endif
-
-  if (filtering && src_width >= 32768) {
-    ScaleFilterCols = ScaleFilterCols64_C;
-  }
-#if defined(HAS_SCALEFILTERCOLS_SSSE3)
-  if (filtering && TestCpuFlag(kCpuHasSSSE3) && src_width < 32768) {
-    ScaleFilterCols = ScaleFilterCols_SSSE3;
-  }
-#endif
-  if (!filtering && src_width * 2 == dst_width && x < 0x8000) {
-    ScaleFilterCols = ScaleColsUp2_C;
-#if defined(HAS_SCALECOLS_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 8) &&
-        IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
-        IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleFilterCols = ScaleColsUp2_SSE2;
-    }
-#endif
-  }
-
-  if (y > max_y) {
-    y = max_y;
-  }
-  {
-    int yi = y >> 16;
-    const uint8* src = src_ptr + yi * src_stride;
-
-    // Allocate 2 row buffers.
-    const int kRowSize = (dst_width + 15) & ~15;
-    align_buffer_64(row, kRowSize * 2);
-
-    uint8* rowptr = row;
-    int rowstride = kRowSize;
-    int lasty = yi;
-
-    ScaleFilterCols(rowptr, src, dst_width, x, dx);
-    if (src_height > 1) {
-      src += src_stride;
-    }
-    ScaleFilterCols(rowptr + rowstride, src, dst_width, x, dx);
-    src += src_stride;
-
-    for (j = 0; j < dst_height; ++j) {
-      yi = y >> 16;
-      if (yi != lasty) {
-        if (y > max_y) {
-          y = max_y;
-          yi = y >> 16;
-          src = src_ptr + yi * src_stride;
-        }
-        if (yi != lasty) {
-          ScaleFilterCols(rowptr, src, dst_width, x, dx);
-          rowptr += rowstride;
-          rowstride = -rowstride;
-          lasty = yi;
-          src += src_stride;
-        }
-      }
-      if (filtering == kFilterLinear) {
-        InterpolateRow(dst_ptr, rowptr, 0, dst_width, 0);
-      } else {
-        int yf = (y >> 8) & 255;
-        InterpolateRow(dst_ptr, rowptr, rowstride, dst_width, yf);
-      }
-      dst_ptr += dst_stride;
-      y += dy;
-    }
-    free_aligned_buffer_64(row);
-  }
-}
-
-// Scale Plane to/from any dimensions, without interpolation.
-// Fixed point math is used for performance: The upper 16 bits
-// of x and dx is the integer part of the source position and
-// the lower 16 bits are the fixed decimal part.
-
-static void ScalePlaneSimple(int src_width, int src_height,
-                             int dst_width, int dst_height,
-                             int src_stride, int dst_stride,
-                             const uint8* src_ptr, uint8* dst_ptr) {
-  int i;
-  void (*ScaleCols)(uint8* dst_ptr, const uint8* src_ptr,
-      int dst_width, int x, int dx) = ScaleCols_C;
-  // Initial source x/y coordinate and step values as 16.16 fixed point.
-  int x = 0;
-  int y = 0;
-  int dx = 0;
-  int dy = 0;
-  ScaleSlope(src_width, src_height, dst_width, dst_height, kFilterNone,
-             &x, &y, &dx, &dy);
-  src_width = Abs(src_width);
-
-  if (src_width * 2 == dst_width && x < 0x8000) {
-    ScaleCols = ScaleColsUp2_C;
-#if defined(HAS_SCALECOLS_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 8) &&
-        IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
-        IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleCols = ScaleColsUp2_SSE2;
-    }
-#endif
-  }
-
-  for (i = 0; i < dst_height; ++i) {
-    ScaleCols(dst_ptr, src_ptr + (y >> 16) * src_stride,
-              dst_width, x, dx);
-    dst_ptr += dst_stride;
-    y += dy;
-  }
-}
-
-// Scale a plane.
-// This function dispatches to a specialized scaler based on scale factor.
-
-LIBYUV_API
-void ScalePlane(const uint8* src, int src_stride,
-                int src_width, int src_height,
-                uint8* dst, int dst_stride,
-                int dst_width, int dst_height,
-                enum FilterMode filtering) {
-  // Simplify filtering when possible.
-  filtering = ScaleFilterReduce(src_width, src_height,
-                                dst_width, dst_height,
-                                filtering);
-
-  // Negative height means invert the image.
-  if (src_height < 0) {
-    src_height = -src_height;
-    src = src + (src_height - 1) * src_stride;
-    src_stride = -src_stride;
-  }
-
-  // Use specialized scales to improve performance for common resolutions.
-  // For example, all the 1/2 scalings will use ScalePlaneDown2()
-  if (dst_width == src_width && dst_height == src_height) {
-    // Straight copy.
-    CopyPlane(src, src_stride, dst, dst_stride, dst_width, dst_height);
-    return;
-  }
-  if (dst_width == src_width) {
-    int dy = FixedDiv(src_height, dst_height);
-    // Arbitrary scale vertically, but unscaled vertically.
-    ScalePlaneVertical(src_height,
-                       dst_width, dst_height,
-                       src_stride, dst_stride, src, dst,
-                       0, 0, dy, 1, filtering);
-    return;
-  }
-  if (dst_width <= Abs(src_width) && dst_height <= src_height) {
-    // Scale down.
-    if (4 * dst_width == 3 * src_width &&
-        4 * dst_height == 3 * src_height) {
-      // optimized, 3/4
-      ScalePlaneDown34(src_width, src_height, dst_width, dst_height,
-                       src_stride, dst_stride, src, dst, filtering);
-      return;
-    }
-    if (2 * dst_width == src_width && 2 * dst_height == src_height) {
-      // optimized, 1/2
-      ScalePlaneDown2(src_width, src_height, dst_width, dst_height,
-                      src_stride, dst_stride, src, dst, filtering);
-      return;
-    }
-    // 3/8 rounded up for odd sized chroma height.
-    if (8 * dst_width == 3 * src_width &&
-        dst_height == ((src_height * 3 + 7) / 8)) {
-      // optimized, 3/8
-      ScalePlaneDown38(src_width, src_height, dst_width, dst_height,
-                       src_stride, dst_stride, src, dst, filtering);
-      return;
-    }
-    if (4 * dst_width == src_width && 4 * dst_height == src_height &&
-               filtering != kFilterBilinear) {
-      // optimized, 1/4
-      ScalePlaneDown4(src_width, src_height, dst_width, dst_height,
-                      src_stride, dst_stride, src, dst, filtering);
-      return;
-    }
-  }
-  if (filtering == kFilterBox && dst_height * 2 < src_height) {
-    ScalePlaneBox(src_width, src_height, dst_width, dst_height,
-                  src_stride, dst_stride, src, dst);
-    return;
-  }
-  if (filtering && dst_height > src_height) {
-    ScalePlaneBilinearUp(src_width, src_height, dst_width, dst_height,
-                         src_stride, dst_stride, src, dst, filtering);
-    return;
-  }
-  if (filtering) {
-    ScalePlaneBilinearDown(src_width, src_height, dst_width, dst_height,
-                           src_stride, dst_stride, src, dst, filtering);
-    return;
-  }
-  ScalePlaneSimple(src_width, src_height, dst_width, dst_height,
-                   src_stride, dst_stride, src, dst);
-}
-
-// Scale an I420 image.
-// This function in turn calls a scaling function for each plane.
-
-LIBYUV_API
-int I420Scale(const uint8* src_y, int src_stride_y,
-              const uint8* src_u, int src_stride_u,
-              const uint8* src_v, int src_stride_v,
-              int src_width, int src_height,
-              uint8* dst_y, int dst_stride_y,
-              uint8* dst_u, int dst_stride_u,
-              uint8* dst_v, int dst_stride_v,
-              int dst_width, int dst_height,
-              enum FilterMode filtering) {
-  int src_halfwidth = SUBSAMPLE(src_width, 1, 1);
-  int src_halfheight = SUBSAMPLE(src_height, 1, 1);
-  int dst_halfwidth = SUBSAMPLE(dst_width, 1, 1);
-  int dst_halfheight = SUBSAMPLE(dst_height, 1, 1);
-  if (!src_y || !src_u || !src_v || src_width == 0 || src_height == 0 ||
-      !dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
-    return -1;
-  }
-
-  ScalePlane(src_y, src_stride_y, src_width, src_height,
-             dst_y, dst_stride_y, dst_width, dst_height,
-             filtering);
-  ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
-             dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
-             filtering);
-  ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
-             dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
-             filtering);
-  return 0;
-}
-
-// Deprecated api
-LIBYUV_API
-int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
-          int src_stride_y, int src_stride_u, int src_stride_v,
-          int src_width, int src_height,
-          uint8* dst_y, uint8* dst_u, uint8* dst_v,
-          int dst_stride_y, int dst_stride_u, int dst_stride_v,
-          int dst_width, int dst_height,
-          LIBYUV_BOOL interpolate) {
-  return I420Scale(src_y, src_stride_y,
-                   src_u, src_stride_u,
-                   src_v, src_stride_v,
-                   src_width, src_height,
-                   dst_y, dst_stride_y,
-                   dst_u, dst_stride_u,
-                   dst_v, dst_stride_v,
-                   dst_width, dst_height,
-                   interpolate ? kFilterBox : kFilterNone);
-}
-
-// Deprecated api
-LIBYUV_API
-int ScaleOffset(const uint8* src, int src_width, int src_height,
-                uint8* dst, int dst_width, int dst_height, int dst_yoffset,
-                LIBYUV_BOOL interpolate) {
-  // Chroma requires offset to multiple of 2.
-  int dst_yoffset_even = dst_yoffset & ~1;
-  int src_halfwidth = SUBSAMPLE(src_width, 1, 1);
-  int src_halfheight = SUBSAMPLE(src_height, 1, 1);
-  int dst_halfwidth = SUBSAMPLE(dst_width, 1, 1);
-  int dst_halfheight = SUBSAMPLE(dst_height, 1, 1);
-  int aheight = dst_height - dst_yoffset_even * 2;  // actual output height
-  const uint8* src_y = src;
-  const uint8* src_u = src + src_width * src_height;
-  const uint8* src_v = src + src_width * src_height +
-                             src_halfwidth * src_halfheight;
-  uint8* dst_y = dst + dst_yoffset_even * dst_width;
-  uint8* dst_u = dst + dst_width * dst_height +
-                 (dst_yoffset_even >> 1) * dst_halfwidth;
-  uint8* dst_v = dst + dst_width * dst_height + dst_halfwidth * dst_halfheight +
-                 (dst_yoffset_even >> 1) * dst_halfwidth;
-  if (!src || src_width <= 0 || src_height <= 0 ||
-      !dst || dst_width <= 0 || dst_height <= 0 || dst_yoffset_even < 0 ||
-      dst_yoffset_even >= dst_height) {
-    return -1;
-  }
-  return I420Scale(src_y, src_width,
-                   src_u, src_halfwidth,
-                   src_v, src_halfwidth,
-                   src_width, src_height,
-                   dst_y, dst_width,
-                   dst_u, dst_halfwidth,
-                   dst_v, dst_halfwidth,
-                   dst_width, aheight,
-                   interpolate ? kFilterBox : kFilterNone);
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb.cc
deleted file mode 100755
index e339cd7c79..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb.cc
+++ /dev/null
@@ -1,809 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/scale.h"
-
-#include <assert.h>
-#include <string.h>
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/planar_functions.h"  // For CopyARGB
-#include "libyuv/row.h"
-#include "libyuv/scale_row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-static __inline int Abs(int v) {
-  return v >= 0 ? v : -v;
-}
-
-// ScaleARGB ARGB, 1/2
-// This is an optimized version for scaling down a ARGB to 1/2 of
-// its original size.
-static void ScaleARGBDown2(int src_width, int src_height,
-                           int dst_width, int dst_height,
-                           int src_stride, int dst_stride,
-                           const uint8* src_argb, uint8* dst_argb,
-                           int x, int dx, int y, int dy,
-                           enum FilterMode filtering) {
-  int j;
-  int row_stride = src_stride * (dy >> 16);
-  void (*ScaleARGBRowDown2)(const uint8* src_argb, ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width) =
-    filtering == kFilterNone ? ScaleARGBRowDown2_C :
-        (filtering == kFilterLinear ? ScaleARGBRowDown2Linear_C :
-        ScaleARGBRowDown2Box_C);
-  assert(dx == 65536 * 2);  // Test scale factor of 2.
-  assert((dy & 0x1ffff) == 0);  // Test vertical scale is multiple of 2.
-  // Advance to odd row, even column.
-  if (filtering == kFilterBilinear) {
-    src_argb += (y >> 16) * src_stride + (x >> 16) * 4;
-  } else {
-    src_argb += (y >> 16) * src_stride + ((x >> 16) - 1) * 4;
-  }
-
-#if defined(HAS_SCALEARGBROWDOWN2_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 4) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(row_stride, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-    ScaleARGBRowDown2 = filtering == kFilterNone ? ScaleARGBRowDown2_SSE2 :
-        (filtering == kFilterLinear ? ScaleARGBRowDown2Linear_SSE2 :
-        ScaleARGBRowDown2Box_SSE2);
-  }
-#elif defined(HAS_SCALEARGBROWDOWN2_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(dst_width, 8) &&
-      IS_ALIGNED(src_argb, 4) && IS_ALIGNED(row_stride, 4)) {
-    ScaleARGBRowDown2 = filtering ? ScaleARGBRowDown2Box_NEON :
-        ScaleARGBRowDown2_NEON;
-  }
-#endif
-
-  if (filtering == kFilterLinear) {
-    src_stride = 0;
-  }
-  for (j = 0; j < dst_height; ++j) {
-    ScaleARGBRowDown2(src_argb, src_stride, dst_argb, dst_width);
-    src_argb += row_stride;
-    dst_argb += dst_stride;
-  }
-}
-
-// ScaleARGB ARGB, 1/4
-// This is an optimized version for scaling down a ARGB to 1/4 of
-// its original size.
-static void ScaleARGBDown4Box(int src_width, int src_height,
-                              int dst_width, int dst_height,
-                              int src_stride, int dst_stride,
-                              const uint8* src_argb, uint8* dst_argb,
-                              int x, int dx, int y, int dy) {
-  int j;
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (dst_width * 2 * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-  int row_stride = src_stride * (dy >> 16);
-  void (*ScaleARGBRowDown2)(const uint8* src_argb, ptrdiff_t src_stride,
-    uint8* dst_argb, int dst_width) = ScaleARGBRowDown2Box_C;
-  // Advance to odd row, even column.
-  src_argb += (y >> 16) * src_stride + (x >> 16) * 4;
-  assert(dx == 65536 * 4);  // Test scale factor of 4.
-  assert((dy & 0x3ffff) == 0);  // Test vertical scale is multiple of 4.
-#if defined(HAS_SCALEARGBROWDOWN2_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 4) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(row_stride, 16) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-    ScaleARGBRowDown2 = ScaleARGBRowDown2Box_SSE2;
-  }
-#elif defined(HAS_SCALEARGBROWDOWN2_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(dst_width, 8) &&
-      IS_ALIGNED(src_argb, 4) && IS_ALIGNED(row_stride, 4)) {
-    ScaleARGBRowDown2 = ScaleARGBRowDown2Box_NEON;
-  }
-#endif
-  for (j = 0; j < dst_height; ++j) {
-    ScaleARGBRowDown2(src_argb, src_stride, row, dst_width * 2);
-    ScaleARGBRowDown2(src_argb + src_stride * 2, src_stride,
-                      row + kRowSize, dst_width * 2);
-    ScaleARGBRowDown2(row, kRowSize, dst_argb, dst_width);
-    src_argb += row_stride;
-    dst_argb += dst_stride;
-  }
-  free_aligned_buffer_64(row);
-}
-
-// ScaleARGB ARGB Even
-// This is an optimized version for scaling down a ARGB to even
-// multiple of its original size.
-static void ScaleARGBDownEven(int src_width, int src_height,
-                              int dst_width, int dst_height,
-                              int src_stride, int dst_stride,
-                              const uint8* src_argb, uint8* dst_argb,
-                              int x, int dx, int y, int dy,
-                              enum FilterMode filtering) {
-  int j;
-  int col_step = dx >> 16;
-  int row_stride = (dy >> 16) * src_stride;
-  void (*ScaleARGBRowDownEven)(const uint8* src_argb, ptrdiff_t src_stride,
-                               int src_step, uint8* dst_argb, int dst_width) =
-      filtering ? ScaleARGBRowDownEvenBox_C : ScaleARGBRowDownEven_C;
-  assert(IS_ALIGNED(src_width, 2));
-  assert(IS_ALIGNED(src_height, 2));
-  src_argb += (y >> 16) * src_stride + (x >> 16) * 4;
-#if defined(HAS_SCALEARGBROWDOWNEVEN_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 4) &&
-      IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-    ScaleARGBRowDownEven = filtering ? ScaleARGBRowDownEvenBox_SSE2 :
-        ScaleARGBRowDownEven_SSE2;
-  }
-#elif defined(HAS_SCALEARGBROWDOWNEVEN_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(dst_width, 4) &&
-      IS_ALIGNED(src_argb, 4)) {
-    ScaleARGBRowDownEven = filtering ? ScaleARGBRowDownEvenBox_NEON :
-        ScaleARGBRowDownEven_NEON;
-  }
-#endif
-
-  if (filtering == kFilterLinear) {
-    src_stride = 0;
-  }
-  for (j = 0; j < dst_height; ++j) {
-    ScaleARGBRowDownEven(src_argb, src_stride, col_step, dst_argb, dst_width);
-    src_argb += row_stride;
-    dst_argb += dst_stride;
-  }
-}
-
-// Scale ARGB down with bilinear interpolation.
-static void ScaleARGBBilinearDown(int src_width, int src_height,
-                                  int dst_width, int dst_height,
-                                  int src_stride, int dst_stride,
-                                  const uint8* src_argb, uint8* dst_argb,
-                                  int x, int dx, int y, int dy,
-                                  enum FilterMode filtering) {
-  int j;
-  void (*InterpolateRow)(uint8* dst_argb, const uint8* src_argb,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-  void (*ScaleARGBFilterCols)(uint8* dst_argb, const uint8* src_argb,
-      int dst_width, int x, int dx) =
-      (src_width >= 32768) ? ScaleARGBFilterCols64_C : ScaleARGBFilterCols_C;
-  int64 xlast = x + (int64)(dst_width - 1) * dx;
-  int64 xl = (dx >= 0) ? x : xlast;
-  int64 xr = (dx >= 0) ? xlast : x;
-  int clip_src_width;
-  xl = (xl >> 16) & ~3;  // Left edge aligned.
-  xr = (xr >> 16) + 1;  // Right most pixel used.  Bilinear uses 2 pixels.
-  xr = (xr + 1 + 3) & ~3;  // 1 beyond 4 pixel aligned right most pixel.
-  if (xr > src_width) {
-    xr = src_width;
-  }
-  clip_src_width = (int)(xr - xl) * 4;  // Width aligned to 4.
-  src_argb += xl * 4;
-  x -= (int)(xl << 16);
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && clip_src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(clip_src_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && clip_src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(clip_src_width, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && clip_src_width >= 32) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(clip_src_width, 32)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && clip_src_width >= 16) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(clip_src_width, 16)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROWS_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && clip_src_width >= 4 &&
-      IS_ALIGNED(src_argb, 4) && IS_ALIGNED(src_stride, 4)) {
-    InterpolateRow = InterpolateRow_Any_MIPS_DSPR2;
-    if (IS_ALIGNED(clip_src_width, 4)) {
-      InterpolateRow = InterpolateRow_MIPS_DSPR2;
-    }
-  }
-#endif
-#if defined(HAS_SCALEARGBFILTERCOLS_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && src_width < 32768) {
-    ScaleARGBFilterCols = ScaleARGBFilterCols_SSSE3;
-  }
-#endif
-  // TODO(fbarchard): Consider not allocating row buffer for kFilterLinear.
-  // Allocate a row of ARGB.
-  {
-    align_buffer_64(row, clip_src_width * 4);
-
-    const int max_y = (src_height - 1) << 16;
-    if (y > max_y) {
-      y = max_y;
-    }
-    for (j = 0; j < dst_height; ++j) {
-      int yi = y >> 16;
-      const uint8* src = src_argb + yi * src_stride;
-      if (filtering == kFilterLinear) {
-        ScaleARGBFilterCols(dst_argb, src, dst_width, x, dx);
-      } else {
-        int yf = (y >> 8) & 255;
-        InterpolateRow(row, src, src_stride, clip_src_width, yf);
-        ScaleARGBFilterCols(dst_argb, row, dst_width, x, dx);
-      }
-      dst_argb += dst_stride;
-      y += dy;
-      if (y > max_y) {
-        y = max_y;
-      }
-    }
-    free_aligned_buffer_64(row);
-  }
-}
-
-// Scale ARGB up with bilinear interpolation.
-static void ScaleARGBBilinearUp(int src_width, int src_height,
-                                int dst_width, int dst_height,
-                                int src_stride, int dst_stride,
-                                const uint8* src_argb, uint8* dst_argb,
-                                int x, int dx, int y, int dy,
-                                enum FilterMode filtering) {
-  int j;
-  void (*InterpolateRow)(uint8* dst_argb, const uint8* src_argb,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-  void (*ScaleARGBFilterCols)(uint8* dst_argb, const uint8* src_argb,
-      int dst_width, int x, int dx) =
-      filtering ? ScaleARGBFilterCols_C : ScaleARGBCols_C;
-  const int max_y = (src_height - 1) << 16;
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && dst_width >= 8) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(dst_width, 8)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROWS_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && dst_width >= 1 &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride, 4)) {
-    InterpolateRow = InterpolateRow_MIPS_DSPR2;
-  }
-#endif
-  if (src_width >= 32768) {
-    ScaleARGBFilterCols = filtering ?
-        ScaleARGBFilterCols64_C : ScaleARGBCols64_C;
-  }
-#if defined(HAS_SCALEARGBFILTERCOLS_SSSE3)
-  if (filtering && TestCpuFlag(kCpuHasSSSE3) && src_width < 32768) {
-    ScaleARGBFilterCols = ScaleARGBFilterCols_SSSE3;
-  }
-#endif
-#if defined(HAS_SCALEARGBCOLS_SSE2)
-  if (!filtering && TestCpuFlag(kCpuHasSSE2) && src_width < 32768) {
-    ScaleARGBFilterCols = ScaleARGBCols_SSE2;
-  }
-#endif
-  if (!filtering && src_width * 2 == dst_width && x < 0x8000) {
-    ScaleARGBFilterCols = ScaleARGBColsUp2_C;
-#if defined(HAS_SCALEARGBCOLSUP2_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 8) &&
-        IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16) &&
-        IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleARGBFilterCols = ScaleARGBColsUp2_SSE2;
-    }
-#endif
-  }
-
-  if (y > max_y) {
-    y = max_y;
-  }
-
-  {
-    int yi = y >> 16;
-    const uint8* src = src_argb + yi * src_stride;
-
-    // Allocate 2 rows of ARGB.
-    const int kRowSize = (dst_width * 4 + 15) & ~15;
-    align_buffer_64(row, kRowSize * 2);
-
-    uint8* rowptr = row;
-    int rowstride = kRowSize;
-    int lasty = yi;
-
-    ScaleARGBFilterCols(rowptr, src, dst_width, x, dx);
-    if (src_height > 1) {
-      src += src_stride;
-    }
-    ScaleARGBFilterCols(rowptr + rowstride, src, dst_width, x, dx);
-    src += src_stride;
-
-    for (j = 0; j < dst_height; ++j) {
-      yi = y >> 16;
-      if (yi != lasty) {
-        if (y > max_y) {
-          y = max_y;
-          yi = y >> 16;
-          src = src_argb + yi * src_stride;
-        }
-        if (yi != lasty) {
-          ScaleARGBFilterCols(rowptr, src, dst_width, x, dx);
-          rowptr += rowstride;
-          rowstride = -rowstride;
-          lasty = yi;
-          src += src_stride;
-        }
-      }
-      if (filtering == kFilterLinear) {
-        InterpolateRow(dst_argb, rowptr, 0, dst_width * 4, 0);
-      } else {
-        int yf = (y >> 8) & 255;
-        InterpolateRow(dst_argb, rowptr, rowstride, dst_width * 4, yf);
-      }
-      dst_argb += dst_stride;
-      y += dy;
-    }
-    free_aligned_buffer_64(row);
-  }
-}
-
-#ifdef YUVSCALEUP
-// Scale YUV to ARGB up with bilinear interpolation.
-static void ScaleYUVToARGBBilinearUp(int src_width, int src_height,
-                                     int dst_width, int dst_height,
-                                     int src_stride_y,
-                                     int src_stride_u,
-                                     int src_stride_v,
-                                     int dst_stride_argb,
-                                     const uint8* src_y,
-                                     const uint8* src_u,
-                                     const uint8* src_v,
-                                     uint8* dst_argb,
-                                     int x, int dx, int y, int dy,
-                                     enum FilterMode filtering) {
-  int j;
-  void (*I422ToARGBRow)(const uint8* y_buf,
-                        const uint8* u_buf,
-                        const uint8* v_buf,
-                        uint8* rgb_buf,
-                        int width) = I422ToARGBRow_C;
-#if defined(HAS_I422TOARGBROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && src_width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_SSSE3;
-    if (IS_ALIGNED(src_width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        I422ToARGBRow = I422ToARGBRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && src_width >= 16) {
-    I422ToARGBRow = I422ToARGBRow_Any_AVX2;
-    if (IS_ALIGNED(src_width, 16)) {
-      I422ToARGBRow = I422ToARGBRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && src_width >= 8) {
-    I422ToARGBRow = I422ToARGBRow_Any_NEON;
-    if (IS_ALIGNED(src_width, 8)) {
-      I422ToARGBRow = I422ToARGBRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_I422TOARGBROW_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && IS_ALIGNED(src_width, 4) &&
-      IS_ALIGNED(src_y, 4) && IS_ALIGNED(src_stride_y, 4) &&
-      IS_ALIGNED(src_u, 2) && IS_ALIGNED(src_stride_u, 2) &&
-      IS_ALIGNED(src_v, 2) && IS_ALIGNED(src_stride_v, 2) &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride_argb, 4)) {
-    I422ToARGBRow = I422ToARGBRow_MIPS_DSPR2;
-  }
-#endif
-
-  void (*InterpolateRow)(uint8* dst_argb, const uint8* src_argb,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && dst_width >= 8) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(dst_width, 8)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && dst_width >= 4) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(dst_width, 4)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROWS_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && dst_width >= 1 &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride_argb, 4)) {
-    InterpolateRow = InterpolateRow_MIPS_DSPR2;
-  }
-#endif
-
-  void (*ScaleARGBFilterCols)(uint8* dst_argb, const uint8* src_argb,
-      int dst_width, int x, int dx) =
-      filtering ? ScaleARGBFilterCols_C : ScaleARGBCols_C;
-  if (src_width >= 32768) {
-    ScaleARGBFilterCols = filtering ?
-        ScaleARGBFilterCols64_C : ScaleARGBCols64_C;
-  }
-#if defined(HAS_SCALEARGBFILTERCOLS_SSSE3)
-  if (filtering && TestCpuFlag(kCpuHasSSSE3) && src_width < 32768) {
-    ScaleARGBFilterCols = ScaleARGBFilterCols_SSSE3;
-  }
-#endif
-#if defined(HAS_SCALEARGBCOLS_SSE2)
-  if (!filtering && TestCpuFlag(kCpuHasSSE2) && src_width < 32768) {
-    ScaleARGBFilterCols = ScaleARGBCols_SSE2;
-  }
-#endif
-  if (!filtering && src_width * 2 == dst_width && x < 0x8000) {
-    ScaleARGBFilterCols = ScaleARGBColsUp2_C;
-#if defined(HAS_SCALEARGBCOLSUP2_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 8) &&
-        IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16) &&
-        IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleARGBFilterCols = ScaleARGBColsUp2_SSE2;
-    }
-#endif
-  }
-
-  const int max_y = (src_height - 1) << 16;
-  if (y > max_y) {
-    y = max_y;
-  }
-  const int kYShift = 1;  // Shift Y by 1 to convert Y plane to UV coordinate.
-  int yi = y >> 16;
-  int uv_yi = yi >> kYShift;
-  const uint8* src_row_y = src_y + yi * src_stride_y;
-  const uint8* src_row_u = src_u + uv_yi * src_stride_u;
-  const uint8* src_row_v = src_v + uv_yi * src_stride_v;
-
-  // Allocate 2 rows of ARGB.
-  const int kRowSize = (dst_width * 4 + 15) & ~15;
-  align_buffer_64(row, kRowSize * 2);
-
-  // Allocate 1 row of ARGB for source conversion.
-  align_buffer_64(argb_row, src_width * 4);
-
-  uint8* rowptr = row;
-  int rowstride = kRowSize;
-  int lasty = yi;
-
-  // TODO(fbarchard): Convert first 2 rows of YUV to ARGB.
-  ScaleARGBFilterCols(rowptr, src_row_y, dst_width, x, dx);
-  if (src_height > 1) {
-    src_row_y += src_stride_y;
-    if (yi & 1) {
-      src_row_u += src_stride_u;
-      src_row_v += src_stride_v;
-    }
-  }
-  ScaleARGBFilterCols(rowptr + rowstride, src_row_y, dst_width, x, dx);
-  if (src_height > 2) {
-    src_row_y += src_stride_y;
-    if (!(yi & 1)) {
-      src_row_u += src_stride_u;
-      src_row_v += src_stride_v;
-    }
-  }
-
-  for (j = 0; j < dst_height; ++j) {
-    yi = y >> 16;
-    if (yi != lasty) {
-      if (y > max_y) {
-        y = max_y;
-        yi = y >> 16;
-        uv_yi = yi >> kYShift;
-        src_row_y = src_y + yi * src_stride_y;
-        src_row_u = src_u + uv_yi * src_stride_u;
-        src_row_v = src_v + uv_yi * src_stride_v;
-      }
-      if (yi != lasty) {
-        // TODO(fbarchard): Convert the clipped region of row.
-        I422ToARGBRow(src_row_y, src_row_u, src_row_v, argb_row, src_width);
-        ScaleARGBFilterCols(rowptr, argb_row, dst_width, x, dx);
-        rowptr += rowstride;
-        rowstride = -rowstride;
-        lasty = yi;
-        src_row_y += src_stride_y;
-        if (yi & 1) {
-          src_row_u += src_stride_u;
-          src_row_v += src_stride_v;
-        }
-      }
-    }
-    if (filtering == kFilterLinear) {
-      InterpolateRow(dst_argb, rowptr, 0, dst_width * 4, 0);
-    } else {
-      int yf = (y >> 8) & 255;
-      InterpolateRow(dst_argb, rowptr, rowstride, dst_width * 4, yf);
-    }
-    dst_argb += dst_stride_argb;
-    y += dy;
-  }
-  free_aligned_buffer_64(row);
-  free_aligned_buffer_64(row_argb);
-}
-#endif
-
-// Scale ARGB to/from any dimensions, without interpolation.
-// Fixed point math is used for performance: The upper 16 bits
-// of x and dx is the integer part of the source position and
-// the lower 16 bits are the fixed decimal part.
-
-static void ScaleARGBSimple(int src_width, int src_height,
-                            int dst_width, int dst_height,
-                            int src_stride, int dst_stride,
-                            const uint8* src_argb, uint8* dst_argb,
-                            int x, int dx, int y, int dy) {
-  int j;
-  void (*ScaleARGBCols)(uint8* dst_argb, const uint8* src_argb,
-      int dst_width, int x, int dx) =
-      (src_width >= 32768) ? ScaleARGBCols64_C : ScaleARGBCols_C;
-#if defined(HAS_SCALEARGBCOLS_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && src_width < 32768) {
-    ScaleARGBCols = ScaleARGBCols_SSE2;
-  }
-#endif
-  if (src_width * 2 == dst_width && x < 0x8000) {
-    ScaleARGBCols = ScaleARGBColsUp2_C;
-#if defined(HAS_SCALEARGBCOLSUP2_SSE2)
-    if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 8) &&
-        IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16) &&
-        IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-      ScaleARGBCols = ScaleARGBColsUp2_SSE2;
-    }
-#endif
-  }
-
-  for (j = 0; j < dst_height; ++j) {
-    ScaleARGBCols(dst_argb, src_argb + (y >> 16) * src_stride,
-                  dst_width, x, dx);
-    dst_argb += dst_stride;
-    y += dy;
-  }
-}
-
-// ScaleARGB a ARGB.
-// This function in turn calls a scaling function
-// suitable for handling the desired resolutions.
-static void ScaleARGB(const uint8* src, int src_stride,
-                      int src_width, int src_height,
-                      uint8* dst, int dst_stride,
-                      int dst_width, int dst_height,
-                      int clip_x, int clip_y, int clip_width, int clip_height,
-                      enum FilterMode filtering) {
-  // Initial source x/y coordinate and step values as 16.16 fixed point.
-  int x = 0;
-  int y = 0;
-  int dx = 0;
-  int dy = 0;
-  // ARGB does not support box filter yet, but allow the user to pass it.
-  // Simplify filtering when possible.
-  filtering = ScaleFilterReduce(src_width, src_height,
-                                dst_width, dst_height,
-                                filtering);
-
-  // Negative src_height means invert the image.
-  if (src_height < 0) {
-    src_height = -src_height;
-    src = src + (src_height - 1) * src_stride;
-    src_stride = -src_stride;
-  }
-  ScaleSlope(src_width, src_height, dst_width, dst_height, filtering,
-             &x, &y, &dx, &dy);
-  src_width = Abs(src_width);
-  if (clip_x) {
-    int64 clipf = (int64)(clip_x) * dx;
-    x += (clipf & 0xffff);
-    src += (clipf >> 16) * 4;
-    dst += clip_x * 4;
-  }
-  if (clip_y) {
-    int64 clipf = (int64)(clip_y) * dy;
-    y += (clipf & 0xffff);
-    src += (clipf >> 16) * src_stride;
-    dst += clip_y * dst_stride;
-  }
-
-  // Special case for integer step values.
-  if (((dx | dy) & 0xffff) == 0) {
-    if (!dx || !dy) {  // 1 pixel wide and/or tall.
-      filtering = kFilterNone;
-    } else {
-      // Optimized even scale down. ie 2, 4, 6, 8, 10x.
-      if (!(dx & 0x10000) && !(dy & 0x10000)) {
-        if (dx == 0x20000) {
-          // Optimized 1/2 downsample.
-          ScaleARGBDown2(src_width, src_height,
-                         clip_width, clip_height,
-                         src_stride, dst_stride, src, dst,
-                         x, dx, y, dy, filtering);
-          return;
-        }
-        if (dx == 0x40000 && filtering == kFilterBox) {
-          // Optimized 1/4 box downsample.
-          ScaleARGBDown4Box(src_width, src_height,
-                            clip_width, clip_height,
-                            src_stride, dst_stride, src, dst,
-                            x, dx, y, dy);
-          return;
-        }
-        ScaleARGBDownEven(src_width, src_height,
-                          clip_width, clip_height,
-                          src_stride, dst_stride, src, dst,
-                          x, dx, y, dy, filtering);
-        return;
-      }
-      // Optimized odd scale down. ie 3, 5, 7, 9x.
-      if ((dx & 0x10000) && (dy & 0x10000)) {
-        filtering = kFilterNone;
-        if (dx == 0x10000 && dy == 0x10000) {
-          // Straight copy.
-          ARGBCopy(src + (y >> 16) * src_stride + (x >> 16) * 4, src_stride,
-                   dst, dst_stride, clip_width, clip_height);
-          return;
-        }
-      }
-    }
-  }
-  if (dx == 0x10000 && (x & 0xffff) == 0) {
-    // Arbitrary scale vertically, but unscaled vertically.
-    ScalePlaneVertical(src_height,
-                       clip_width, clip_height,
-                       src_stride, dst_stride, src, dst,
-                       x, y, dy, 4, filtering);
-    return;
-  }
-  if (filtering && dy < 65536) {
-    ScaleARGBBilinearUp(src_width, src_height,
-                        clip_width, clip_height,
-                        src_stride, dst_stride, src, dst,
-                        x, dx, y, dy, filtering);
-    return;
-  }
-  if (filtering) {
-    ScaleARGBBilinearDown(src_width, src_height,
-                          clip_width, clip_height,
-                          src_stride, dst_stride, src, dst,
-                          x, dx, y, dy, filtering);
-    return;
-  }
-  ScaleARGBSimple(src_width, src_height, clip_width, clip_height,
-                  src_stride, dst_stride, src, dst,
-                  x, dx, y, dy);
-}
-
-LIBYUV_API
-int ARGBScaleClip(const uint8* src_argb, int src_stride_argb,
-                  int src_width, int src_height,
-                  uint8* dst_argb, int dst_stride_argb,
-                  int dst_width, int dst_height,
-                  int clip_x, int clip_y, int clip_width, int clip_height,
-                  enum FilterMode filtering) {
-  if (!src_argb || src_width == 0 || src_height == 0 ||
-      !dst_argb || dst_width <= 0 || dst_height <= 0 ||
-      clip_x < 0 || clip_y < 0 ||
-      (clip_x + clip_width) > dst_width ||
-      (clip_y + clip_height) > dst_height) {
-    return -1;
-  }
-  ScaleARGB(src_argb, src_stride_argb, src_width, src_height,
-            dst_argb, dst_stride_argb, dst_width, dst_height,
-            clip_x, clip_y, clip_width, clip_height, filtering);
-  return 0;
-}
-
-// Scale an ARGB image.
-LIBYUV_API
-int ARGBScale(const uint8* src_argb, int src_stride_argb,
-              int src_width, int src_height,
-              uint8* dst_argb, int dst_stride_argb,
-              int dst_width, int dst_height,
-              enum FilterMode filtering) {
-  if (!src_argb || src_width == 0 || src_height == 0 ||
-      !dst_argb || dst_width <= 0 || dst_height <= 0) {
-    return -1;
-  }
-  ScaleARGB(src_argb, src_stride_argb, src_width, src_height,
-            dst_argb, dst_stride_argb, dst_width, dst_height,
-            0, 0, dst_width, dst_height, filtering);
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb_neon.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb_neon.cc
deleted file mode 100755
index c0b5433239..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_argb_neon.cc
+++ /dev/null
@@ -1,145 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC Neon
-#if !defined(LIBYUV_DISABLE_NEON) && defined(__ARM_NEON__)
-
-void ScaleARGBRowDown2_NEON(const uint8* src_ptr, ptrdiff_t /* src_stride */,
-                            uint8* dst, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID

-	".fpu neon\n"

-#endif
-  "1:                                          \n"
-    // load even pixels into q0, odd into q1
-    "vld2.32    {q0, q1}, [%0]!                \n"
-    "vld2.32    {q2, q3}, [%0]!                \n"
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop
-    "vst1.8     {q1}, [%1]!                    \n"  // store odd pixels
-    "vst1.8     {q3}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst),              // %1
-    "+r"(dst_width)         // %2
-  :
-  : "memory", "cc", "q0", "q1", "q2", "q3"  // Clobber List
-  );
-}
-
-void ScaleARGBRowDown2Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width) {
-  asm volatile (
-    // change the stride to row 2 pointer
-    "add        %1, %1, %0                     \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vpaddl.u8  q3, q3                         \n"  // A 16 bytes -> 8 shorts.
-    "vld4.8     {d16, d18, d20, d22}, [%1]!    \n"  // load 8 more ARGB pixels.
-    "vld4.8     {d17, d19, d21, d23}, [%1]!    \n"  // load last 8 ARGB pixels.
-    "vpadal.u8  q0, q8                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q9                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q10                        \n"  // R 16 bytes -> 8 shorts.
-    "vpadal.u8  q3, q11                        \n"  // A 16 bytes -> 8 shorts.
-    "vrshrn.u16 d0, q0, #2                     \n"  // downshift, round and pack
-    "vrshrn.u16 d1, q1, #2                     \n"
-    "vrshrn.u16 d2, q2, #2                     \n"
-    "vrshrn.u16 d3, q3, #2                     \n"
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(src_stride),       // %1
-    "+r"(dst),              // %2
-    "+r"(dst_width)         // %3
-  :
-  : "memory", "cc", "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11"
-  );
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: src_argb 4 byte aligned.
-void ScaleARGBRowDownEven_NEON(const uint8* src_argb, ptrdiff_t, int src_stepx,
-                               uint8* dst_argb, int dst_width) {
-  asm volatile (
-    "mov        r12, %3, lsl #2                \n"
-    ".p2align  2                               \n"
-  "1:                                          \n"
-    "vld1.32    {d0[0]}, [%0], r12             \n"
-    "vld1.32    {d0[1]}, [%0], r12             \n"
-    "vld1.32    {d1[0]}, [%0], r12             \n"
-    "vld1.32    {d1[1]}, [%0], r12             \n"
-    "subs       %2, %2, #4                     \n"  // 4 pixels per loop.
-    "vst1.8     {q0}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+r"(dst_width)    // %2
-  : "r"(src_stepx)     // %3
-  : "memory", "cc", "r12", "q0"
-  );
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: src_argb 4 byte aligned.
-void ScaleARGBRowDownEvenBox_NEON(const uint8* src_argb, ptrdiff_t src_stride,
-                                  int src_stepx,
-                                  uint8* dst_argb, int dst_width) {
-  asm volatile (
-    "mov       r12, %4, lsl #2                 \n"
-    "add       %1, %1, %0                      \n"
-    ".p2align  2                               \n"
-  "1:                                          \n"
-    "vld1.8    {d0}, [%0], r12                 \n"  // Read 4 2x2 blocks -> 2x1
-    "vld1.8    {d1}, [%1], r12                 \n"
-    "vld1.8    {d2}, [%0], r12                 \n"
-    "vld1.8    {d3}, [%1], r12                 \n"
-    "vld1.8    {d4}, [%0], r12                 \n"
-    "vld1.8    {d5}, [%1], r12                 \n"
-    "vld1.8    {d6}, [%0], r12                 \n"
-    "vld1.8    {d7}, [%1], r12                 \n"
-    "vaddl.u8  q0, d0, d1                      \n"
-    "vaddl.u8  q1, d2, d3                      \n"
-    "vaddl.u8  q2, d4, d5                      \n"
-    "vaddl.u8  q3, d6, d7                      \n"
-    "vswp.8    d1, d2                          \n"  // ab_cd -> ac_bd
-    "vswp.8    d5, d6                          \n"  // ef_gh -> eg_fh
-    "vadd.u16  q0, q0, q1                      \n"  // (a+b)_(c+d)
-    "vadd.u16  q2, q2, q3                      \n"  // (e+f)_(g+h)
-    "vrshrn.u16 d0, q0, #2                     \n"  // first 2 pixels.
-    "vrshrn.u16 d1, q2, #2                     \n"  // next 2 pixels.
-    "subs       %3, %3, #4                     \n"  // 4 pixels per loop.
-    "vst1.8     {q0}, [%2]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_argb),    // %0
-    "+r"(src_stride),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(dst_width)    // %3
-  : "r"(src_stepx)     // %4
-  : "memory", "cc", "r12", "q0", "q1", "q2", "q3"
-  );
-}
-#endif  // __ARM_NEON__
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_common.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_common.cc
deleted file mode 100644
index 6ed8bfaf97..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_common.cc
+++ /dev/null
@@ -1,772 +0,0 @@
-/*
- *  Copyright 2013 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/scale.h"
-
-#include <assert.h>
-#include <string.h>
-
-#include "libyuv/cpu_id.h"
-#include "libyuv/planar_functions.h"  // For CopyARGB
-#include "libyuv/row.h"
-#include "libyuv/scale_row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-static __inline int Abs(int v) {
-  return v >= 0 ? v : -v;
-}
-
-// CPU agnostic row functions
-void ScaleRowDown2_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                     uint8* dst, int dst_width) {
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = src_ptr[1];
-    dst[1] = src_ptr[3];
-    dst += 2;
-    src_ptr += 4;
-  }
-  if (dst_width & 1) {
-    dst[0] = src_ptr[1];
-  }
-}
-
-void ScaleRowDown2Linear_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst, int dst_width) {
-  const uint8* s = src_ptr;
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = (s[0] + s[1] + 1) >> 1;
-    dst[1] = (s[2] + s[3] + 1) >> 1;
-    dst += 2;
-    s += 4;
-  }
-  if (dst_width & 1) {
-    dst[0] = (s[0] + s[1] + 1) >> 1;
-  }
-}
-
-void ScaleRowDown2Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width) {
-  const uint8* s = src_ptr;
-  const uint8* t = src_ptr + src_stride;
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = (s[0] + s[1] + t[0] + t[1] + 2) >> 2;
-    dst[1] = (s[2] + s[3] + t[2] + t[3] + 2) >> 2;
-    dst += 2;
-    s += 4;
-    t += 4;
-  }
-  if (dst_width & 1) {
-    dst[0] = (s[0] + s[1] + t[0] + t[1] + 2) >> 2;
-  }
-}
-
-void ScaleRowDown4_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                     uint8* dst, int dst_width) {
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = src_ptr[2];
-    dst[1] = src_ptr[6];
-    dst += 2;
-    src_ptr += 8;
-  }
-  if (dst_width & 1) {
-    dst[0] = src_ptr[2];
-  }
-}
-
-void ScaleRowDown4Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width) {
-  intptr_t stride = src_stride;
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] + src_ptr[3] +
-             src_ptr[stride + 0] + src_ptr[stride + 1] +
-             src_ptr[stride + 2] + src_ptr[stride + 3] +
-             src_ptr[stride * 2 + 0] + src_ptr[stride * 2 + 1] +
-             src_ptr[stride * 2 + 2] + src_ptr[stride * 2 + 3] +
-             src_ptr[stride * 3 + 0] + src_ptr[stride * 3 + 1] +
-             src_ptr[stride * 3 + 2] + src_ptr[stride * 3 + 3] +
-             8) >> 4;
-    dst[1] = (src_ptr[4] + src_ptr[5] + src_ptr[6] + src_ptr[7] +
-             src_ptr[stride + 4] + src_ptr[stride + 5] +
-             src_ptr[stride + 6] + src_ptr[stride + 7] +
-             src_ptr[stride * 2 + 4] + src_ptr[stride * 2 + 5] +
-             src_ptr[stride * 2 + 6] + src_ptr[stride * 2 + 7] +
-             src_ptr[stride * 3 + 4] + src_ptr[stride * 3 + 5] +
-             src_ptr[stride * 3 + 6] + src_ptr[stride * 3 + 7] +
-             8) >> 4;
-    dst += 2;
-    src_ptr += 8;
-  }
-  if (dst_width & 1) {
-    dst[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] + src_ptr[3] +
-             src_ptr[stride + 0] + src_ptr[stride + 1] +
-             src_ptr[stride + 2] + src_ptr[stride + 3] +
-             src_ptr[stride * 2 + 0] + src_ptr[stride * 2 + 1] +
-             src_ptr[stride * 2 + 2] + src_ptr[stride * 2 + 3] +
-             src_ptr[stride * 3 + 0] + src_ptr[stride * 3 + 1] +
-             src_ptr[stride * 3 + 2] + src_ptr[stride * 3 + 3] +
-             8) >> 4;
-  }
-}
-
-void ScaleRowDown34_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                      uint8* dst, int dst_width) {
-  int x;
-  assert((dst_width % 3 == 0) && (dst_width > 0));
-  for (x = 0; x < dst_width; x += 3) {
-    dst[0] = src_ptr[0];
-    dst[1] = src_ptr[1];
-    dst[2] = src_ptr[3];
-    dst += 3;
-    src_ptr += 4;
-  }
-}
-
-// Filter rows 0 and 1 together, 3 : 1
-void ScaleRowDown34_0_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* d, int dst_width) {
-  const uint8* s = src_ptr;
-  const uint8* t = src_ptr + src_stride;
-  int x;
-  assert((dst_width % 3 == 0) && (dst_width > 0));
-  for (x = 0; x < dst_width; x += 3) {
-    uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
-    uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
-    uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
-    uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
-    uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
-    uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
-    d[0] = (a0 * 3 + b0 + 2) >> 2;
-    d[1] = (a1 * 3 + b1 + 2) >> 2;
-    d[2] = (a2 * 3 + b2 + 2) >> 2;
-    d += 3;
-    s += 4;
-    t += 4;
-  }
-}
-
-// Filter rows 1 and 2 together, 1 : 1
-void ScaleRowDown34_1_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* d, int dst_width) {
-  const uint8* s = src_ptr;
-  const uint8* t = src_ptr + src_stride;
-  int x;
-  assert((dst_width % 3 == 0) && (dst_width > 0));
-  for (x = 0; x < dst_width; x += 3) {
-    uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
-    uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
-    uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
-    uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
-    uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
-    uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
-    d[0] = (a0 + b0 + 1) >> 1;
-    d[1] = (a1 + b1 + 1) >> 1;
-    d[2] = (a2 + b2 + 1) >> 1;
-    d += 3;
-    s += 4;
-    t += 4;
-  }
-}
-
-// Scales a single row of pixels using point sampling.
-void ScaleCols_C(uint8* dst_ptr, const uint8* src_ptr,
-                 int dst_width, int x, int dx) {
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    dst_ptr[0] = src_ptr[x >> 16];
-    x += dx;
-    dst_ptr[1] = src_ptr[x >> 16];
-    x += dx;
-    dst_ptr += 2;
-  }
-  if (dst_width & 1) {
-    dst_ptr[0] = src_ptr[x >> 16];
-  }
-}
-
-// Scales a single row of pixels up by 2x using point sampling.
-void ScaleColsUp2_C(uint8* dst_ptr, const uint8* src_ptr,
-                    int dst_width, int x, int dx) {
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    dst_ptr[1] = dst_ptr[0] = src_ptr[0];
-    src_ptr += 1;
-    dst_ptr += 2;
-  }
-  if (dst_width & 1) {
-    dst_ptr[0] = src_ptr[0];
-  }
-}
-
-// (1-f)a + fb can be replaced with a + f(b-a)
-#define BLENDER(a, b, f) (uint8)((int)(a) + \
-    ((int)(f) * ((int)(b) - (int)(a)) >> 16))
-
-void ScaleFilterCols_C(uint8* dst_ptr, const uint8* src_ptr,
-                       int dst_width, int x, int dx) {
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    int xi = x >> 16;
-    int a = src_ptr[xi];
-    int b = src_ptr[xi + 1];
-    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
-    x += dx;
-    xi = x >> 16;
-    a = src_ptr[xi];
-    b = src_ptr[xi + 1];
-    dst_ptr[1] = BLENDER(a, b, x & 0xffff);
-    x += dx;
-    dst_ptr += 2;
-  }
-  if (dst_width & 1) {
-    int xi = x >> 16;
-    int a = src_ptr[xi];
-    int b = src_ptr[xi + 1];
-    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
-  }
-}
-
-void ScaleFilterCols64_C(uint8* dst_ptr, const uint8* src_ptr,
-                         int dst_width, int x32, int dx) {
-  int64 x = (int64)(x32);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    int64 xi = x >> 16;
-    int a = src_ptr[xi];
-    int b = src_ptr[xi + 1];
-    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
-    x += dx;
-    xi = x >> 16;
-    a = src_ptr[xi];
-    b = src_ptr[xi + 1];
-    dst_ptr[1] = BLENDER(a, b, x & 0xffff);
-    x += dx;
-    dst_ptr += 2;
-  }
-  if (dst_width & 1) {
-    int64 xi = x >> 16;
-    int a = src_ptr[xi];
-    int b = src_ptr[xi + 1];
-    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
-  }
-}
-#undef BLENDER
-
-void ScaleRowDown38_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                      uint8* dst, int dst_width) {
-  int x;
-  assert(dst_width % 3 == 0);
-  for (x = 0; x < dst_width; x += 3) {
-    dst[0] = src_ptr[0];
-    dst[1] = src_ptr[3];
-    dst[2] = src_ptr[6];
-    dst += 3;
-    src_ptr += 8;
-  }
-}
-
-// 8x3 -> 3x1
-void ScaleRowDown38_3_Box_C(const uint8* src_ptr,
-                            ptrdiff_t src_stride,
-                            uint8* dst_ptr, int dst_width) {
-  intptr_t stride = src_stride;
-  int i;
-  assert((dst_width % 3 == 0) && (dst_width > 0));
-  for (i = 0; i < dst_width; i += 3) {
-    dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
-        src_ptr[stride + 0] + src_ptr[stride + 1] +
-        src_ptr[stride + 2] + src_ptr[stride * 2 + 0] +
-        src_ptr[stride * 2 + 1] + src_ptr[stride * 2 + 2]) *
-        (65536 / 9) >> 16;
-    dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
-        src_ptr[stride + 3] + src_ptr[stride + 4] +
-        src_ptr[stride + 5] + src_ptr[stride * 2 + 3] +
-        src_ptr[stride * 2 + 4] + src_ptr[stride * 2 + 5]) *
-        (65536 / 9) >> 16;
-    dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
-        src_ptr[stride + 6] + src_ptr[stride + 7] +
-        src_ptr[stride * 2 + 6] + src_ptr[stride * 2 + 7]) *
-        (65536 / 6) >> 16;
-    src_ptr += 8;
-    dst_ptr += 3;
-  }
-}
-
-// 8x2 -> 3x1
-void ScaleRowDown38_2_Box_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* dst_ptr, int dst_width) {
-  intptr_t stride = src_stride;
-  int i;
-  assert((dst_width % 3 == 0) && (dst_width > 0));
-  for (i = 0; i < dst_width; i += 3) {
-    dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
-        src_ptr[stride + 0] + src_ptr[stride + 1] +
-        src_ptr[stride + 2]) * (65536 / 6) >> 16;
-    dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
-        src_ptr[stride + 3] + src_ptr[stride + 4] +
-        src_ptr[stride + 5]) * (65536 / 6) >> 16;
-    dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
-        src_ptr[stride + 6] + src_ptr[stride + 7]) *
-        (65536 / 4) >> 16;
-    src_ptr += 8;
-    dst_ptr += 3;
-  }
-}
-
-void ScaleAddRows_C(const uint8* src_ptr, ptrdiff_t src_stride,
-                    uint16* dst_ptr, int src_width, int src_height) {
-  int x;
-  assert(src_width > 0);
-  assert(src_height > 0);
-  for (x = 0; x < src_width; ++x) {
-    const uint8* s = src_ptr + x;
-    unsigned int sum = 0u;
-    int y;
-    for (y = 0; y < src_height; ++y) {
-      sum += s[0];
-      s += src_stride;
-    }
-    // TODO(fbarchard): Consider limitting height to 256 to avoid overflow.
-    dst_ptr[x] = sum < 65535u ? sum : 65535u;
-  }
-}
-
-void ScaleARGBRowDown2_C(const uint8* src_argb,
-                         ptrdiff_t src_stride,
-                         uint8* dst_argb, int dst_width) {
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = src[1];
-    dst[1] = src[3];
-    src += 4;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    dst[0] = src[1];
-  }
-}
-
-void ScaleARGBRowDown2Linear_C(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               uint8* dst_argb, int dst_width) {
-  int x;
-  for (x = 0; x < dst_width; ++x) {
-    dst_argb[0] = (src_argb[0] + src_argb[4] + 1) >> 1;
-    dst_argb[1] = (src_argb[1] + src_argb[5] + 1) >> 1;
-    dst_argb[2] = (src_argb[2] + src_argb[6] + 1) >> 1;
-    dst_argb[3] = (src_argb[3] + src_argb[7] + 1) >> 1;
-    src_argb += 8;
-    dst_argb += 4;
-  }
-}
-
-void ScaleARGBRowDown2Box_C(const uint8* src_argb, ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width) {
-  int x;
-  for (x = 0; x < dst_width; ++x) {
-    dst_argb[0] = (src_argb[0] + src_argb[4] +
-                  src_argb[src_stride] + src_argb[src_stride + 4] + 2) >> 2;
-    dst_argb[1] = (src_argb[1] + src_argb[5] +
-                  src_argb[src_stride + 1] + src_argb[src_stride + 5] + 2) >> 2;
-    dst_argb[2] = (src_argb[2] + src_argb[6] +
-                  src_argb[src_stride + 2] + src_argb[src_stride + 6] + 2) >> 2;
-    dst_argb[3] = (src_argb[3] + src_argb[7] +
-                  src_argb[src_stride + 3] + src_argb[src_stride + 7] + 2) >> 2;
-    src_argb += 8;
-    dst_argb += 4;
-  }
-}
-
-void ScaleARGBRowDownEven_C(const uint8* src_argb, ptrdiff_t src_stride,
-                            int src_stepx,
-                            uint8* dst_argb, int dst_width) {
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-
-  int x;
-  for (x = 0; x < dst_width - 1; x += 2) {
-    dst[0] = src[0];
-    dst[1] = src[src_stepx];
-    src += src_stepx * 2;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    dst[0] = src[0];
-  }
-}
-
-void ScaleARGBRowDownEvenBox_C(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width) {
-  int x;
-  for (x = 0; x < dst_width; ++x) {
-    dst_argb[0] = (src_argb[0] + src_argb[4] +
-                  src_argb[src_stride] + src_argb[src_stride + 4] + 2) >> 2;
-    dst_argb[1] = (src_argb[1] + src_argb[5] +
-                  src_argb[src_stride + 1] + src_argb[src_stride + 5] + 2) >> 2;
-    dst_argb[2] = (src_argb[2] + src_argb[6] +
-                  src_argb[src_stride + 2] + src_argb[src_stride + 6] + 2) >> 2;
-    dst_argb[3] = (src_argb[3] + src_argb[7] +
-                  src_argb[src_stride + 3] + src_argb[src_stride + 7] + 2) >> 2;
-    src_argb += src_stepx * 4;
-    dst_argb += 4;
-  }
-}
-
-// Scales a single row of pixels using point sampling.
-void ScaleARGBCols_C(uint8* dst_argb, const uint8* src_argb,
-                     int dst_width, int x, int dx) {
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    dst[0] = src[x >> 16];
-    x += dx;
-    dst[1] = src[x >> 16];
-    x += dx;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    dst[0] = src[x >> 16];
-  }
-}
-
-void ScaleARGBCols64_C(uint8* dst_argb, const uint8* src_argb,
-                       int dst_width, int x32, int dx) {
-  int64 x = (int64)(x32);
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    dst[0] = src[x >> 16];
-    x += dx;
-    dst[1] = src[x >> 16];
-    x += dx;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    dst[0] = src[x >> 16];
-  }
-}
-
-// Scales a single row of pixels up by 2x using point sampling.
-void ScaleARGBColsUp2_C(uint8* dst_argb, const uint8* src_argb,
-                        int dst_width, int x, int dx) {
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    dst[1] = dst[0] = src[0];
-    src += 1;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    dst[0] = src[0];
-  }
-}
-
-// Mimics SSSE3 blender
-#define BLENDER1(a, b, f) ((a) * (0x7f ^ f) + (b) * f) >> 7
-#define BLENDERC(a, b, f, s) (uint32)( \
-    BLENDER1(((a) >> s) & 255, ((b) >> s) & 255, f) << s)
-#define BLENDER(a, b, f) \
-    BLENDERC(a, b, f, 24) | BLENDERC(a, b, f, 16) | \
-    BLENDERC(a, b, f, 8) | BLENDERC(a, b, f, 0)
-
-void ScaleARGBFilterCols_C(uint8* dst_argb, const uint8* src_argb,
-                           int dst_width, int x, int dx) {
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    int xi = x >> 16;
-    int xf = (x >> 9) & 0x7f;
-    uint32 a = src[xi];
-    uint32 b = src[xi + 1];
-    dst[0] = BLENDER(a, b, xf);
-    x += dx;
-    xi = x >> 16;
-    xf = (x >> 9) & 0x7f;
-    a = src[xi];
-    b = src[xi + 1];
-    dst[1] = BLENDER(a, b, xf);
-    x += dx;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    int xi = x >> 16;
-    int xf = (x >> 9) & 0x7f;
-    uint32 a = src[xi];
-    uint32 b = src[xi + 1];
-    dst[0] = BLENDER(a, b, xf);
-  }
-}
-
-void ScaleARGBFilterCols64_C(uint8* dst_argb, const uint8* src_argb,
-                             int dst_width, int x32, int dx) {
-  int64 x = (int64)(x32);
-  const uint32* src = (const uint32*)(src_argb);
-  uint32* dst = (uint32*)(dst_argb);
-  int j;
-  for (j = 0; j < dst_width - 1; j += 2) {
-    int64 xi = x >> 16;
-    int xf = (x >> 9) & 0x7f;
-    uint32 a = src[xi];
-    uint32 b = src[xi + 1];
-    dst[0] = BLENDER(a, b, xf);
-    x += dx;
-    xi = x >> 16;
-    xf = (x >> 9) & 0x7f;
-    a = src[xi];
-    b = src[xi + 1];
-    dst[1] = BLENDER(a, b, xf);
-    x += dx;
-    dst += 2;
-  }
-  if (dst_width & 1) {
-    int64 xi = x >> 16;
-    int xf = (x >> 9) & 0x7f;
-    uint32 a = src[xi];
-    uint32 b = src[xi + 1];
-    dst[0] = BLENDER(a, b, xf);
-  }
-}
-#undef BLENDER1
-#undef BLENDERC
-#undef BLENDER
-
-// Scale plane vertically with bilinear interpolation.
-void ScalePlaneVertical(int src_height,
-                        int dst_width, int dst_height,
-                        int src_stride, int dst_stride,
-                        const uint8* src_argb, uint8* dst_argb,
-                        int x, int y, int dy,
-                        int bpp, enum FilterMode filtering) {
-  // TODO(fbarchard): Allow higher bpp.
-  int dst_width_bytes = dst_width * bpp;
-  void (*InterpolateRow)(uint8* dst_argb, const uint8* src_argb,
-      ptrdiff_t src_stride, int dst_width, int source_y_fraction) =
-      InterpolateRow_C;
-  const int max_y = (src_height > 1) ? ((src_height - 1) << 16) - 1 : 0;
-  int j;
-  assert(bpp >= 1 && bpp <= 4);
-  assert(src_height != 0);
-  assert(dst_width > 0);
-  assert(dst_height > 0);
-  src_argb += (x >> 16) * bpp;
-#if defined(HAS_INTERPOLATEROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && dst_width_bytes >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSE2;
-    if (IS_ALIGNED(dst_width_bytes, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSE2;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && dst_width_bytes >= 16) {
-    InterpolateRow = InterpolateRow_Any_SSSE3;
-    if (IS_ALIGNED(dst_width_bytes, 16)) {
-      InterpolateRow = InterpolateRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride, 16) &&
-          IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride, 16)) {
-        InterpolateRow = InterpolateRow_SSSE3;
-      }
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && dst_width_bytes >= 32) {
-    InterpolateRow = InterpolateRow_Any_AVX2;
-    if (IS_ALIGNED(dst_width_bytes, 32)) {
-      InterpolateRow = InterpolateRow_AVX2;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && dst_width_bytes >= 16) {
-    InterpolateRow = InterpolateRow_Any_NEON;
-    if (IS_ALIGNED(dst_width_bytes, 16)) {
-      InterpolateRow = InterpolateRow_NEON;
-    }
-  }
-#endif
-#if defined(HAS_INTERPOLATEROWS_MIPS_DSPR2)
-  if (TestCpuFlag(kCpuHasMIPS_DSPR2) && dst_width_bytes >= 4 &&
-      IS_ALIGNED(src_argb, 4) && IS_ALIGNED(src_stride, 4) &&
-      IS_ALIGNED(dst_argb, 4) && IS_ALIGNED(dst_stride, 4)) {
-    InterpolateRow = InterpolateRow_Any_MIPS_DSPR2;
-    if (IS_ALIGNED(dst_width_bytes, 4)) {
-      InterpolateRow = InterpolateRow_MIPS_DSPR2;
-    }
-  }
-#endif
-  for (j = 0; j < dst_height; ++j) {
-    int yi;
-    int yf;
-    if (y > max_y) {
-      y = max_y;
-    }
-    yi = y >> 16;
-    yf = filtering ? ((y >> 8) & 255) : 0;
-    InterpolateRow(dst_argb, src_argb + yi * src_stride,
-                   src_stride, dst_width_bytes, yf);
-    dst_argb += dst_stride;
-    y += dy;
-  }
-}
-
-// Simplify the filtering based on scale factors.
-enum FilterMode ScaleFilterReduce(int src_width, int src_height,
-                                  int dst_width, int dst_height,
-                                  enum FilterMode filtering) {
-  if (src_width < 0) {
-    src_width = -src_width;
-  }
-  if (src_height < 0) {
-    src_height = -src_height;
-  }
-  if (filtering == kFilterBox) {
-    // If scaling both axis to 0.5 or larger, switch from Box to Bilinear.
-    if (dst_width * 2 >= src_width && dst_height * 2 >= src_height) {
-      filtering = kFilterBilinear;
-    }
-    // If scaling to larger, switch from Box to Bilinear.
-    if (dst_width >= src_width || dst_height >= src_height) {
-      filtering = kFilterBilinear;
-    }
-  }
-  if (filtering == kFilterBilinear) {
-    if (src_height == 1) {
-      filtering = kFilterLinear;
-    }
-    // TODO(fbarchard): Detect any odd scale factor and reduce to Linear.
-    if (dst_height == src_height || dst_height * 3 == src_height) {
-      filtering = kFilterLinear;
-    }
-    // TODO(fbarchard): Remove 1 pixel wide filter restriction, which is to
-    // avoid reading 2 pixels horizontally that causes memory exception.
-    if (src_width == 1) {
-      filtering = kFilterNone;
-    }
-  }
-  if (filtering == kFilterLinear) {
-    if (src_width == 1) {
-      filtering = kFilterNone;
-    }
-    // TODO(fbarchard): Detect any odd scale factor and reduce to None.
-    if (dst_width == src_width || dst_width * 3 == src_width) {
-      filtering = kFilterNone;
-    }
-  }
-  return filtering;
-}
-
-// Divide num by div and return as 16.16 fixed point result.
-int FixedDiv_C(int num, int div) {
-  return (int)(((int64)(num) << 16) / div);
-}
-
-// Divide num by div and return as 16.16 fixed point result.
-int FixedDiv1_C(int num, int div) {
-  return (int)((((int64)(num) << 16) - 0x00010001) /
-                          (div - 1));
-}
-
-#define CENTERSTART(dx, s) (dx < 0) ? -((-dx >> 1) + s) : ((dx >> 1) + s)
-
-// Compute slope values for stepping.
-void ScaleSlope(int src_width, int src_height,
-                int dst_width, int dst_height,
-                enum FilterMode filtering,
-                int* x, int* y, int* dx, int* dy) {
-  assert(x != NULL);
-  assert(y != NULL);
-  assert(dx != NULL);
-  assert(dy != NULL);
-  assert(src_width != 0);
-  assert(src_height != 0);
-  assert(dst_width > 0);
-  assert(dst_height > 0);
-  // Check for 1 pixel and avoid FixedDiv overflow.
-  if (dst_width == 1 && src_width >= 32768) {
-    dst_width = src_width;
-  }
-  if (dst_height == 1 && src_height >= 32768) {
-    dst_height = src_height;
-  }
-  if (filtering == kFilterBox) {
-    // Scale step for point sampling duplicates all pixels equally.
-    *dx = FixedDiv(Abs(src_width), dst_width);
-    *dy = FixedDiv(src_height, dst_height);
-    *x = 0;
-    *y = 0;
-  } else if (filtering == kFilterBilinear) {
-    // Scale step for bilinear sampling renders last pixel once for upsample.
-    if (dst_width <= Abs(src_width)) {
-      *dx = FixedDiv(Abs(src_width), dst_width);
-      *x = CENTERSTART(*dx, -32768);  // Subtract 0.5 (32768) to center filter.
-    } else if (dst_width > 1) {
-      *dx = FixedDiv1(Abs(src_width), dst_width);
-      *x = 0;
-    }
-    if (dst_height <= src_height) {
-      *dy = FixedDiv(src_height,  dst_height);
-      *y = CENTERSTART(*dy, -32768);  // Subtract 0.5 (32768) to center filter.
-    } else if (dst_height > 1) {
-      *dy = FixedDiv1(src_height, dst_height);
-      *y = 0;
-    }
-  } else if (filtering == kFilterLinear) {
-    // Scale step for bilinear sampling renders last pixel once for upsample.
-    if (dst_width <= Abs(src_width)) {
-      *dx = FixedDiv(Abs(src_width), dst_width);
-      *x = CENTERSTART(*dx, -32768);  // Subtract 0.5 (32768) to center filter.
-    } else if (dst_width > 1) {
-      *dx = FixedDiv1(Abs(src_width), dst_width);
-      *x = 0;
-    }
-    *dy = FixedDiv(src_height, dst_height);
-    *y = *dy >> 1;
-  } else {
-    // Scale step for point sampling duplicates all pixels equally.
-    *dx = FixedDiv(Abs(src_width), dst_width);
-    *dy = FixedDiv(src_height, dst_height);
-    *x = CENTERSTART(*dx, 0);
-    *y = CENTERSTART(*dy, 0);
-  }
-  // Negative src_width means horizontally mirror.
-  if (src_width < 0) {
-    *x += (dst_width - 1) * *dx;
-    *dx = -*dx;
-    // src_width = -src_width;   // Caller must do this.
-  }
-}
-#undef CENTERSTART
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_mips.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_mips.cc
deleted file mode 100755
index 4572f4504e..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_mips.cc
+++ /dev/null
@@ -1,653 +0,0 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/basic_types.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC MIPS DSPR2
-#if !defined(LIBYUV_DISABLE_MIPS) && \
-    defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-
-void ScaleRowDown2_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst, int dst_width) {
-  __asm__ __volatile__(
-    ".set push                                     \n"
-    ".set noreorder                                \n"
-
-    "srl            $t9, %[dst_width], 4           \n"  // iterations -> by 16
-    "beqz           $t9, 2f                        \n"
-    " nop                                          \n"
-
-    ".p2align       2                              \n"
-  "1:                                              \n"
-    "lw             $t0, 0(%[src_ptr])             \n"  // |3|2|1|0|
-    "lw             $t1, 4(%[src_ptr])             \n"  // |7|6|5|4|
-    "lw             $t2, 8(%[src_ptr])             \n"  // |11|10|9|8|
-    "lw             $t3, 12(%[src_ptr])            \n"  // |15|14|13|12|
-    "lw             $t4, 16(%[src_ptr])            \n"  // |19|18|17|16|
-    "lw             $t5, 20(%[src_ptr])            \n"  // |23|22|21|20|
-    "lw             $t6, 24(%[src_ptr])            \n"  // |27|26|25|24|
-    "lw             $t7, 28(%[src_ptr])            \n"  // |31|30|29|28|
-    // TODO(fbarchard): Use odd pixels instead of even.
-    "precr.qb.ph    $t8, $t1, $t0                  \n"  // |6|4|2|0|
-    "precr.qb.ph    $t0, $t3, $t2                  \n"  // |14|12|10|8|
-    "precr.qb.ph    $t1, $t5, $t4                  \n"  // |22|20|18|16|
-    "precr.qb.ph    $t2, $t7, $t6                  \n"  // |30|28|26|24|
-    "addiu          %[src_ptr], %[src_ptr], 32     \n"
-    "addiu          $t9, $t9, -1                   \n"
-    "sw             $t8, 0(%[dst])                 \n"
-    "sw             $t0, 4(%[dst])                 \n"
-    "sw             $t1, 8(%[dst])                 \n"
-    "sw             $t2, 12(%[dst])                \n"
-    "bgtz           $t9, 1b                        \n"
-    " addiu         %[dst], %[dst], 16             \n"
-
-  "2:                                              \n"
-    "andi           $t9, %[dst_width], 0xf         \n"  // residue
-    "beqz           $t9, 3f                        \n"
-    " nop                                          \n"
-
-  "21:                                             \n"
-    "lbu            $t0, 0(%[src_ptr])             \n"
-    "addiu          %[src_ptr], %[src_ptr], 2      \n"
-    "addiu          $t9, $t9, -1                   \n"
-    "sb             $t0, 0(%[dst])                 \n"
-    "bgtz           $t9, 21b                       \n"
-    " addiu         %[dst], %[dst], 1              \n"
-
-  "3:                                              \n"
-    ".set pop                                      \n"
-  : [src_ptr] "+r" (src_ptr),
-    [dst] "+r" (dst)
-  : [dst_width] "r" (dst_width)
-  : "t0", "t1", "t2", "t3", "t4", "t5",
-    "t6", "t7", "t8", "t9"
-  );
-}
-
-void ScaleRowDown2Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                 uint8* dst, int dst_width) {
-  const uint8* t = src_ptr + src_stride;
-
-  __asm__ __volatile__ (
-    ".set push                                    \n"
-    ".set noreorder                               \n"
-
-    "srl            $t9, %[dst_width], 3          \n"  // iterations -> step 8
-    "bltz           $t9, 2f                       \n"
-    " nop                                         \n"
-
-    ".p2align       2                             \n"
-  "1:                                             \n"
-    "lw             $t0, 0(%[src_ptr])            \n"  // |3|2|1|0|
-    "lw             $t1, 4(%[src_ptr])            \n"  // |7|6|5|4|
-    "lw             $t2, 8(%[src_ptr])            \n"  // |11|10|9|8|
-    "lw             $t3, 12(%[src_ptr])           \n"  // |15|14|13|12|
-    "lw             $t4, 0(%[t])                  \n"  // |19|18|17|16|
-    "lw             $t5, 4(%[t])                  \n"  // |23|22|21|20|
-    "lw             $t6, 8(%[t])                  \n"  // |27|26|25|24|
-    "lw             $t7, 12(%[t])                 \n"  // |31|30|29|28|
-    "addiu          $t9, $t9, -1                  \n"
-    "srl            $t8, $t0, 16                  \n"  // |X|X|3|2|
-    "ins            $t0, $t4, 16, 16              \n"  // |17|16|1|0|
-    "ins            $t4, $t8, 0, 16               \n"  // |19|18|3|2|
-    "raddu.w.qb     $t0, $t0                      \n"  // |17+16+1+0|
-    "raddu.w.qb     $t4, $t4                      \n"  // |19+18+3+2|
-    "shra_r.w       $t0, $t0, 2                   \n"  // |t0+2|>>2
-    "shra_r.w       $t4, $t4, 2                   \n"  // |t4+2|>>2
-    "srl            $t8, $t1, 16                  \n"  // |X|X|7|6|
-    "ins            $t1, $t5, 16, 16              \n"  // |21|20|5|4|
-    "ins            $t5, $t8, 0, 16               \n"  // |22|23|7|6|
-    "raddu.w.qb     $t1, $t1                      \n"  // |21+20+5+4|
-    "raddu.w.qb     $t5, $t5                      \n"  // |23+22+7+6|
-    "shra_r.w       $t1, $t1, 2                   \n"  // |t1+2|>>2
-    "shra_r.w       $t5, $t5, 2                   \n"  // |t5+2|>>2
-    "srl            $t8, $t2, 16                  \n"  // |X|X|11|10|
-    "ins            $t2, $t6, 16, 16              \n"  // |25|24|9|8|
-    "ins            $t6, $t8, 0, 16               \n"  // |27|26|11|10|
-    "raddu.w.qb     $t2, $t2                      \n"  // |25+24+9+8|
-    "raddu.w.qb     $t6, $t6                      \n"  // |27+26+11+10|
-    "shra_r.w       $t2, $t2, 2                   \n"  // |t2+2|>>2
-    "shra_r.w       $t6, $t6, 2                   \n"  // |t5+2|>>2
-    "srl            $t8, $t3, 16                  \n"  // |X|X|15|14|
-    "ins            $t3, $t7, 16, 16              \n"  // |29|28|13|12|
-    "ins            $t7, $t8, 0, 16               \n"  // |31|30|15|14|
-    "raddu.w.qb     $t3, $t3                      \n"  // |29+28+13+12|
-    "raddu.w.qb     $t7, $t7                      \n"  // |31+30+15+14|
-    "shra_r.w       $t3, $t3, 2                   \n"  // |t3+2|>>2
-    "shra_r.w       $t7, $t7, 2                   \n"  // |t7+2|>>2
-    "addiu          %[src_ptr], %[src_ptr], 16    \n"
-    "addiu          %[t], %[t], 16                \n"
-    "sb             $t0, 0(%[dst])                \n"
-    "sb             $t4, 1(%[dst])                \n"
-    "sb             $t1, 2(%[dst])                \n"
-    "sb             $t5, 3(%[dst])                \n"
-    "sb             $t2, 4(%[dst])                \n"
-    "sb             $t6, 5(%[dst])                \n"
-    "sb             $t3, 6(%[dst])                \n"
-    "sb             $t7, 7(%[dst])                \n"
-    "bgtz           $t9, 1b                       \n"
-    " addiu         %[dst], %[dst], 8             \n"
-
-  "2:                                             \n"
-    "andi           $t9, %[dst_width], 0x7        \n"  // x = residue
-    "beqz           $t9, 3f                       \n"
-    " nop                                         \n"
-
-    "21:                                          \n"
-    "lwr            $t1, 0(%[src_ptr])            \n"
-    "lwl            $t1, 3(%[src_ptr])            \n"
-    "lwr            $t2, 0(%[t])                  \n"
-    "lwl            $t2, 3(%[t])                  \n"
-    "srl            $t8, $t1, 16                  \n"
-    "ins            $t1, $t2, 16, 16              \n"
-    "ins            $t2, $t8, 0, 16               \n"
-    "raddu.w.qb     $t1, $t1                      \n"
-    "raddu.w.qb     $t2, $t2                      \n"
-    "shra_r.w       $t1, $t1, 2                   \n"
-    "shra_r.w       $t2, $t2, 2                   \n"
-    "sb             $t1, 0(%[dst])                \n"
-    "sb             $t2, 1(%[dst])                \n"
-    "addiu          %[src_ptr], %[src_ptr], 4     \n"
-    "addiu          $t9, $t9, -2                  \n"
-    "addiu          %[t], %[t], 4                 \n"
-    "bgtz           $t9, 21b                      \n"
-    " addiu         %[dst], %[dst], 2             \n"
-
-  "3:                                             \n"
-    ".set pop                                     \n"
-
-  : [src_ptr] "+r" (src_ptr),
-    [dst] "+r" (dst), [t] "+r" (t)
-  : [dst_width] "r" (dst_width)
-  : "t0", "t1", "t2", "t3", "t4", "t5",
-    "t6", "t7", "t8", "t9"
-  );
-}
-
-void ScaleRowDown4_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst, int dst_width) {
-  __asm__ __volatile__ (
-      ".set push                                    \n"
-      ".set noreorder                               \n"
-
-      "srl            $t9, %[dst_width], 3          \n"
-      "beqz           $t9, 2f                       \n"
-      " nop                                         \n"
-
-      ".p2align       2                             \n"
-     "1:                                            \n"
-      "lw             $t1, 0(%[src_ptr])            \n"  // |3|2|1|0|
-      "lw             $t2, 4(%[src_ptr])            \n"  // |7|6|5|4|
-      "lw             $t3, 8(%[src_ptr])            \n"  // |11|10|9|8|
-      "lw             $t4, 12(%[src_ptr])           \n"  // |15|14|13|12|
-      "lw             $t5, 16(%[src_ptr])           \n"  // |19|18|17|16|
-      "lw             $t6, 20(%[src_ptr])           \n"  // |23|22|21|20|
-      "lw             $t7, 24(%[src_ptr])           \n"  // |27|26|25|24|
-      "lw             $t8, 28(%[src_ptr])           \n"  // |31|30|29|28|
-      "precr.qb.ph    $t1, $t2, $t1                 \n"  // |6|4|2|0|
-      "precr.qb.ph    $t2, $t4, $t3                 \n"  // |14|12|10|8|
-      "precr.qb.ph    $t5, $t6, $t5                 \n"  // |22|20|18|16|
-      "precr.qb.ph    $t6, $t8, $t7                 \n"  // |30|28|26|24|
-      "precr.qb.ph    $t1, $t2, $t1                 \n"  // |12|8|4|0|
-      "precr.qb.ph    $t5, $t6, $t5                 \n"  // |28|24|20|16|
-      "addiu          %[src_ptr], %[src_ptr], 32    \n"
-      "addiu          $t9, $t9, -1                  \n"
-      "sw             $t1, 0(%[dst])                \n"
-      "sw             $t5, 4(%[dst])                \n"
-      "bgtz           $t9, 1b                       \n"
-      " addiu         %[dst], %[dst], 8             \n"
-
-    "2:                                             \n"
-      "andi           $t9, %[dst_width], 7          \n"  // residue
-      "beqz           $t9, 3f                       \n"
-      " nop                                         \n"
-
-    "21:                                            \n"
-      "lbu            $t1, 0(%[src_ptr])            \n"
-      "addiu          %[src_ptr], %[src_ptr], 4     \n"
-      "addiu          $t9, $t9, -1                  \n"
-      "sb             $t1, 0(%[dst])                \n"
-      "bgtz           $t9, 21b                      \n"
-      " addiu         %[dst], %[dst], 1             \n"
-
-    "3:                                             \n"
-      ".set pop                                     \n"
-      : [src_ptr] "+r" (src_ptr),
-        [dst] "+r" (dst)
-      : [dst_width] "r" (dst_width)
-      : "t1", "t2", "t3", "t4", "t5",
-        "t6", "t7", "t8", "t9"
-  );
-}
-
-void ScaleRowDown4Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                 uint8* dst, int dst_width) {
-  intptr_t stride = src_stride;
-  const uint8* s1 = src_ptr + stride;
-  const uint8* s2 = s1 + stride;
-  const uint8* s3 = s2 + stride;
-
-  __asm__ __volatile__ (
-      ".set push                                  \n"
-      ".set noreorder                             \n"
-
-      "srl           $t9, %[dst_width], 1         \n"
-      "andi          $t8, %[dst_width], 1         \n"
-
-      ".p2align      2                            \n"
-     "1:                                          \n"
-      "lw            $t0, 0(%[src_ptr])           \n"  // |3|2|1|0|
-      "lw            $t1, 0(%[s1])                \n"  // |7|6|5|4|
-      "lw            $t2, 0(%[s2])                \n"  // |11|10|9|8|
-      "lw            $t3, 0(%[s3])                \n"  // |15|14|13|12|
-      "lw            $t4, 4(%[src_ptr])           \n"  // |19|18|17|16|
-      "lw            $t5, 4(%[s1])                \n"  // |23|22|21|20|
-      "lw            $t6, 4(%[s2])                \n"  // |27|26|25|24|
-      "lw            $t7, 4(%[s3])                \n"  // |31|30|29|28|
-      "raddu.w.qb    $t0, $t0                     \n"  // |3 + 2 + 1 + 0|
-      "raddu.w.qb    $t1, $t1                     \n"  // |7 + 6 + 5 + 4|
-      "raddu.w.qb    $t2, $t2                     \n"  // |11 + 10 + 9 + 8|
-      "raddu.w.qb    $t3, $t3                     \n"  // |15 + 14 + 13 + 12|
-      "raddu.w.qb    $t4, $t4                     \n"  // |19 + 18 + 17 + 16|
-      "raddu.w.qb    $t5, $t5                     \n"  // |23 + 22 + 21 + 20|
-      "raddu.w.qb    $t6, $t6                     \n"  // |27 + 26 + 25 + 24|
-      "raddu.w.qb    $t7, $t7                     \n"  // |31 + 30 + 29 + 28|
-      "add           $t0, $t0, $t1                \n"
-      "add           $t1, $t2, $t3                \n"
-      "add           $t0, $t0, $t1                \n"
-      "add           $t4, $t4, $t5                \n"
-      "add           $t6, $t6, $t7                \n"
-      "add           $t4, $t4, $t6                \n"
-      "shra_r.w      $t0, $t0, 4                  \n"
-      "shra_r.w      $t4, $t4, 4                  \n"
-      "sb            $t0, 0(%[dst])               \n"
-      "sb            $t4, 1(%[dst])               \n"
-      "addiu         %[src_ptr], %[src_ptr], 8    \n"
-      "addiu         %[s1], %[s1], 8              \n"
-      "addiu         %[s2], %[s2], 8              \n"
-      "addiu         %[s3], %[s3], 8              \n"
-      "addiu         $t9, $t9, -1                 \n"
-      "bgtz          $t9, 1b                      \n"
-      " addiu        %[dst], %[dst], 2            \n"
-      "beqz          $t8, 2f                      \n"
-      " nop                                       \n"
-
-      "lw            $t0, 0(%[src_ptr])           \n"  // |3|2|1|0|
-      "lw            $t1, 0(%[s1])                \n"  // |7|6|5|4|
-      "lw            $t2, 0(%[s2])                \n"  // |11|10|9|8|
-      "lw            $t3, 0(%[s3])                \n"  // |15|14|13|12|
-      "raddu.w.qb    $t0, $t0                     \n"  // |3 + 2 + 1 + 0|
-      "raddu.w.qb    $t1, $t1                     \n"  // |7 + 6 + 5 + 4|
-      "raddu.w.qb    $t2, $t2                     \n"  // |11 + 10 + 9 + 8|
-      "raddu.w.qb    $t3, $t3                     \n"  // |15 + 14 + 13 + 12|
-      "add           $t0, $t0, $t1                \n"
-      "add           $t1, $t2, $t3                \n"
-      "add           $t0, $t0, $t1                \n"
-      "shra_r.w      $t0, $t0, 4                  \n"
-      "sb            $t0, 0(%[dst])               \n"
-
-      "2:                                         \n"
-      ".set pop                                   \n"
-
-      : [src_ptr] "+r" (src_ptr),
-        [dst] "+r" (dst),
-        [s1] "+r" (s1),
-        [s2] "+r" (s2),
-        [s3] "+r" (s3)
-      : [dst_width] "r" (dst_width)
-      : "t0", "t1", "t2", "t3", "t4", "t5",
-        "t6","t7", "t8", "t9"
-  );
-}
-
-void ScaleRowDown34_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width) {
-  __asm__ __volatile__ (
-      ".set push                                          \n"
-      ".set noreorder                                     \n"
-      ".p2align        2                                  \n"
-    "1:                                                   \n"
-      "lw              $t1, 0(%[src_ptr])                 \n"  // |3|2|1|0|
-      "lw              $t2, 4(%[src_ptr])                 \n"  // |7|6|5|4|
-      "lw              $t3, 8(%[src_ptr])                 \n"  // |11|10|9|8|
-      "lw              $t4, 12(%[src_ptr])                \n"  // |15|14|13|12|
-      "lw              $t5, 16(%[src_ptr])                \n"  // |19|18|17|16|
-      "lw              $t6, 20(%[src_ptr])                \n"  // |23|22|21|20|
-      "lw              $t7, 24(%[src_ptr])                \n"  // |27|26|25|24|
-      "lw              $t8, 28(%[src_ptr])                \n"  // |31|30|29|28|
-      "precrq.qb.ph    $t0, $t2, $t4                      \n"  // |7|5|15|13|
-      "precrq.qb.ph    $t9, $t6, $t8                      \n"  // |23|21|31|30|
-      "addiu           %[dst_width], %[dst_width], -24    \n"
-      "ins             $t1, $t1, 8, 16                    \n"  // |3|1|0|X|
-      "ins             $t4, $t0, 8, 16                    \n"  // |X|15|13|12|
-      "ins             $t5, $t5, 8, 16                    \n"  // |19|17|16|X|
-      "ins             $t8, $t9, 8, 16                    \n"  // |X|31|29|28|
-      "addiu           %[src_ptr], %[src_ptr], 32         \n"
-      "packrl.ph       $t0, $t3, $t0                      \n"  // |9|8|7|5|
-      "packrl.ph       $t9, $t7, $t9                      \n"  // |25|24|23|21|
-      "prepend         $t1, $t2, 8                        \n"  // |4|3|1|0|
-      "prepend         $t3, $t4, 24                       \n"  // |15|13|12|11|
-      "prepend         $t5, $t6, 8                        \n"  // |20|19|17|16|
-      "prepend         $t7, $t8, 24                       \n"  // |31|29|28|27|
-      "sw              $t1, 0(%[dst])                     \n"
-      "sw              $t0, 4(%[dst])                     \n"
-      "sw              $t3, 8(%[dst])                     \n"
-      "sw              $t5, 12(%[dst])                    \n"
-      "sw              $t9, 16(%[dst])                    \n"
-      "sw              $t7, 20(%[dst])                    \n"
-      "bnez            %[dst_width], 1b                   \n"
-      " addiu          %[dst], %[dst], 24                 \n"
-      ".set pop                                           \n"
-      : [src_ptr] "+r" (src_ptr),
-        [dst] "+r" (dst),
-        [dst_width] "+r" (dst_width)
-      :
-      : "t0", "t1", "t2", "t3", "t4", "t5",
-        "t6","t7", "t8", "t9"
-  );
-}
-
-void ScaleRowDown34_0_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* d, int dst_width) {
-  __asm__ __volatile__ (
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-      "repl.ph           $t3, 3                          \n"  // 0x00030003
-
-     ".p2align           2                               \n"
-    "1:                                                  \n"
-      "lw                $t0, 0(%[src_ptr])              \n"  // |S3|S2|S1|S0|
-      "lwx               $t1, %[src_stride](%[src_ptr])  \n"  // |T3|T2|T1|T0|
-      "rotr              $t2, $t0, 8                     \n"  // |S0|S3|S2|S1|
-      "rotr              $t6, $t1, 8                     \n"  // |T0|T3|T2|T1|
-      "muleu_s.ph.qbl    $t4, $t2, $t3                   \n"  // |S0*3|S3*3|
-      "muleu_s.ph.qbl    $t5, $t6, $t3                   \n"  // |T0*3|T3*3|
-      "andi              $t0, $t2, 0xFFFF                \n"  // |0|0|S2|S1|
-      "andi              $t1, $t6, 0xFFFF                \n"  // |0|0|T2|T1|
-      "raddu.w.qb        $t0, $t0                        \n"
-      "raddu.w.qb        $t1, $t1                        \n"
-      "shra_r.w          $t0, $t0, 1                     \n"
-      "shra_r.w          $t1, $t1, 1                     \n"
-      "preceu.ph.qbr     $t2, $t2                        \n"  // |0|S2|0|S1|
-      "preceu.ph.qbr     $t6, $t6                        \n"  // |0|T2|0|T1|
-      "rotr              $t2, $t2, 16                    \n"  // |0|S1|0|S2|
-      "rotr              $t6, $t6, 16                    \n"  // |0|T1|0|T2|
-      "addu.ph           $t2, $t2, $t4                   \n"
-      "addu.ph           $t6, $t6, $t5                   \n"
-      "sll               $t5, $t0, 1                     \n"
-      "add               $t0, $t5, $t0                   \n"
-      "shra_r.ph         $t2, $t2, 2                     \n"
-      "shra_r.ph         $t6, $t6, 2                     \n"
-      "shll.ph           $t4, $t2, 1                     \n"
-      "addq.ph           $t4, $t4, $t2                   \n"
-      "addu              $t0, $t0, $t1                   \n"
-      "addiu             %[src_ptr], %[src_ptr], 4       \n"
-      "shra_r.w          $t0, $t0, 2                     \n"
-      "addu.ph           $t6, $t6, $t4                   \n"
-      "shra_r.ph         $t6, $t6, 2                     \n"
-      "srl               $t1, $t6, 16                    \n"
-      "addiu             %[dst_width], %[dst_width], -3  \n"
-      "sb                $t1, 0(%[d])                    \n"
-      "sb                $t0, 1(%[d])                    \n"
-      "sb                $t6, 2(%[d])                    \n"
-      "bgtz              %[dst_width], 1b                \n"
-      " addiu            %[d], %[d], 3                   \n"
-    "3:                                                  \n"
-      ".set pop                                          \n"
-      : [src_ptr] "+r" (src_ptr),
-        [src_stride] "+r" (src_stride),
-        [d] "+r" (d),
-        [dst_width] "+r" (dst_width)
-      :
-      : "t0", "t1", "t2", "t3",
-        "t4", "t5", "t6"
-  );
-}
-
-void ScaleRowDown34_1_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* d, int dst_width) {
-  __asm__ __volatile__ (
-      ".set push                                           \n"
-      ".set noreorder                                      \n"
-      "repl.ph           $t2, 3                            \n"  // 0x00030003
-
-      ".p2align          2                                 \n"
-    "1:                                                    \n"
-      "lw                $t0, 0(%[src_ptr])                \n"  // |S3|S2|S1|S0|
-      "lwx               $t1, %[src_stride](%[src_ptr])    \n"  // |T3|T2|T1|T0|
-      "rotr              $t4, $t0, 8                       \n"  // |S0|S3|S2|S1|
-      "rotr              $t6, $t1, 8                       \n"  // |T0|T3|T2|T1|
-      "muleu_s.ph.qbl    $t3, $t4, $t2                     \n"  // |S0*3|S3*3|
-      "muleu_s.ph.qbl    $t5, $t6, $t2                     \n"  // |T0*3|T3*3|
-      "andi              $t0, $t4, 0xFFFF                  \n"  // |0|0|S2|S1|
-      "andi              $t1, $t6, 0xFFFF                  \n"  // |0|0|T2|T1|
-      "raddu.w.qb        $t0, $t0                          \n"
-      "raddu.w.qb        $t1, $t1                          \n"
-      "shra_r.w          $t0, $t0, 1                       \n"
-      "shra_r.w          $t1, $t1, 1                       \n"
-      "preceu.ph.qbr     $t4, $t4                          \n"  // |0|S2|0|S1|
-      "preceu.ph.qbr     $t6, $t6                          \n"  // |0|T2|0|T1|
-      "rotr              $t4, $t4, 16                      \n"  // |0|S1|0|S2|
-      "rotr              $t6, $t6, 16                      \n"  // |0|T1|0|T2|
-      "addu.ph           $t4, $t4, $t3                     \n"
-      "addu.ph           $t6, $t6, $t5                     \n"
-      "shra_r.ph         $t6, $t6, 2                       \n"
-      "shra_r.ph         $t4, $t4, 2                       \n"
-      "addu.ph           $t6, $t6, $t4                     \n"
-      "addiu             %[src_ptr], %[src_ptr], 4         \n"
-      "shra_r.ph         $t6, $t6, 1                       \n"
-      "addu              $t0, $t0, $t1                     \n"
-      "addiu             %[dst_width], %[dst_width], -3    \n"
-      "shra_r.w          $t0, $t0, 1                       \n"
-      "srl               $t1, $t6, 16                      \n"
-      "sb                $t1, 0(%[d])                      \n"
-      "sb                $t0, 1(%[d])                      \n"
-      "sb                $t6, 2(%[d])                      \n"
-      "bgtz              %[dst_width], 1b                  \n"
-      " addiu            %[d], %[d], 3                     \n"
-    "3:                                                    \n"
-      ".set pop                                            \n"
-      : [src_ptr] "+r" (src_ptr),
-        [src_stride] "+r" (src_stride),
-        [d] "+r" (d),
-        [dst_width] "+r" (dst_width)
-      :
-      : "t0", "t1", "t2", "t3",
-        "t4", "t5", "t6"
-  );
-}
-
-void ScaleRowDown38_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width) {
-  __asm__ __volatile__ (
-      ".set push                                     \n"
-      ".set noreorder                                \n"
-
-      ".p2align   2                                  \n"
-    "1:                                              \n"
-      "lw         $t0, 0(%[src_ptr])                 \n"  // |3|2|1|0|
-      "lw         $t1, 4(%[src_ptr])                 \n"  // |7|6|5|4|
-      "lw         $t2, 8(%[src_ptr])                 \n"  // |11|10|9|8|
-      "lw         $t3, 12(%[src_ptr])                \n"  // |15|14|13|12|
-      "lw         $t4, 16(%[src_ptr])                \n"  // |19|18|17|16|
-      "lw         $t5, 20(%[src_ptr])                \n"  // |23|22|21|20|
-      "lw         $t6, 24(%[src_ptr])                \n"  // |27|26|25|24|
-      "lw         $t7, 28(%[src_ptr])                \n"  // |31|30|29|28|
-      "wsbh       $t0, $t0                           \n"  // |2|3|0|1|
-      "wsbh       $t6, $t6                           \n"  // |26|27|24|25|
-      "srl        $t0, $t0, 8                        \n"  // |X|2|3|0|
-      "srl        $t3, $t3, 16                       \n"  // |X|X|15|14|
-      "srl        $t5, $t5, 16                       \n"  // |X|X|23|22|
-      "srl        $t7, $t7, 16                       \n"  // |X|X|31|30|
-      "ins        $t1, $t2, 24, 8                    \n"  // |8|6|5|4|
-      "ins        $t6, $t5, 0, 8                     \n"  // |26|27|24|22|
-      "ins        $t1, $t0, 0, 16                    \n"  // |8|6|3|0|
-      "ins        $t6, $t7, 24, 8                    \n"  // |30|27|24|22|
-      "prepend    $t2, $t3, 24                       \n"  // |X|15|14|11|
-      "ins        $t4, $t4, 16, 8                    \n"  // |19|16|17|X|
-      "ins        $t4, $t2, 0, 16                    \n"  // |19|16|14|11|
-      "addiu      %[src_ptr], %[src_ptr], 32         \n"
-      "addiu      %[dst_width], %[dst_width], -12    \n"
-      "addiu      $t8,%[dst_width], -12              \n"
-      "sw         $t1, 0(%[dst])                     \n"
-      "sw         $t4, 4(%[dst])                     \n"
-      "sw         $t6, 8(%[dst])                     \n"
-      "bgez       $t8, 1b                            \n"
-      " addiu     %[dst], %[dst], 12                 \n"
-      ".set pop                                      \n"
-      : [src_ptr] "+r" (src_ptr),
-        [dst] "+r" (dst),
-        [dst_width] "+r" (dst_width)
-      :
-      : "t0", "t1", "t2", "t3", "t4",
-        "t5", "t6", "t7", "t8"
-  );
-}
-
-void ScaleRowDown38_2_Box_MIPS_DSPR2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width) {
-  intptr_t stride = src_stride;
-  const uint8* t = src_ptr + stride;
-  const int c = 0x2AAA;
-
-  __asm__ __volatile__ (
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-
-      ".p2align        2                                 \n"
-    "1:                                                  \n"
-      "lw              $t0, 0(%[src_ptr])                \n"  // |S3|S2|S1|S0|
-      "lw              $t1, 4(%[src_ptr])                \n"  // |S7|S6|S5|S4|
-      "lw              $t2, 0(%[t])                      \n"  // |T3|T2|T1|T0|
-      "lw              $t3, 4(%[t])                      \n"  // |T7|T6|T5|T4|
-      "rotr            $t1, $t1, 16                      \n"  // |S5|S4|S7|S6|
-      "packrl.ph       $t4, $t1, $t3                     \n"  // |S7|S6|T7|T6|
-      "packrl.ph       $t5, $t3, $t1                     \n"  // |T5|T4|S5|S4|
-      "raddu.w.qb      $t4, $t4                          \n"  // S7+S6+T7+T6
-      "raddu.w.qb      $t5, $t5                          \n"  // T5+T4+S5+S4
-      "precrq.qb.ph    $t6, $t0, $t2                     \n"  // |S3|S1|T3|T1|
-      "precrq.qb.ph    $t6, $t6, $t6                     \n"  // |S3|T3|S3|T3|
-      "srl             $t4, $t4, 2                       \n"  // t4 / 4
-      "srl             $t6, $t6, 16                      \n"  // |0|0|S3|T3|
-      "raddu.w.qb      $t6, $t6                          \n"  // 0+0+S3+T3
-      "addu            $t6, $t5, $t6                     \n"
-      "mul             $t6, $t6, %[c]                    \n"  // t6 * 0x2AAA
-      "sll             $t0, $t0, 8                       \n"  // |S2|S1|S0|0|
-      "sll             $t2, $t2, 8                       \n"  // |T2|T1|T0|0|
-      "raddu.w.qb      $t0, $t0                          \n"  // S2+S1+S0+0
-      "raddu.w.qb      $t2, $t2                          \n"  // T2+T1+T0+0
-      "addu            $t0, $t0, $t2                     \n"
-      "mul             $t0, $t0, %[c]                    \n"  // t0 * 0x2AAA
-      "addiu           %[src_ptr], %[src_ptr], 8         \n"
-      "addiu           %[t], %[t], 8                     \n"
-      "addiu           %[dst_width], %[dst_width], -3    \n"
-      "addiu           %[dst_ptr], %[dst_ptr], 3         \n"
-      "srl             $t6, $t6, 16                      \n"
-      "srl             $t0, $t0, 16                      \n"
-      "sb              $t4, -1(%[dst_ptr])               \n"
-      "sb              $t6, -2(%[dst_ptr])               \n"
-      "bgtz            %[dst_width], 1b                  \n"
-      " sb             $t0, -3(%[dst_ptr])               \n"
-      ".set pop                                          \n"
-      : [src_ptr] "+r" (src_ptr),
-        [dst_ptr] "+r" (dst_ptr),
-        [t] "+r" (t),
-        [dst_width] "+r" (dst_width)
-      : [c] "r" (c)
-      : "t0", "t1", "t2", "t3", "t4", "t5", "t6"
-  );
-}
-
-void ScaleRowDown38_3_Box_MIPS_DSPR2(const uint8* src_ptr,
-                                     ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width) {
-  intptr_t stride = src_stride;
-  const uint8* s1 = src_ptr + stride;
-  stride += stride;
-  const uint8* s2 = src_ptr + stride;
-  const int c1 = 0x1C71;
-  const int c2 = 0x2AAA;
-
-  __asm__ __volatile__ (
-      ".set push                                         \n"
-      ".set noreorder                                    \n"
-
-      ".p2align        2                                 \n"
-    "1:                                                  \n"
-      "lw              $t0, 0(%[src_ptr])                \n"  // |S3|S2|S1|S0|
-      "lw              $t1, 4(%[src_ptr])                \n"  // |S7|S6|S5|S4|
-      "lw              $t2, 0(%[s1])                     \n"  // |T3|T2|T1|T0|
-      "lw              $t3, 4(%[s1])                     \n"  // |T7|T6|T5|T4|
-      "lw              $t4, 0(%[s2])                     \n"  // |R3|R2|R1|R0|
-      "lw              $t5, 4(%[s2])                     \n"  // |R7|R6|R5|R4|
-      "rotr            $t1, $t1, 16                      \n"  // |S5|S4|S7|S6|
-      "packrl.ph       $t6, $t1, $t3                     \n"  // |S7|S6|T7|T6|
-      "raddu.w.qb      $t6, $t6                          \n"  // S7+S6+T7+T6
-      "packrl.ph       $t7, $t3, $t1                     \n"  // |T5|T4|S5|S4|
-      "raddu.w.qb      $t7, $t7                          \n"  // T5+T4+S5+S4
-      "sll             $t8, $t5, 16                      \n"  // |R5|R4|0|0|
-      "raddu.w.qb      $t8, $t8                          \n"  // R5+R4
-      "addu            $t7, $t7, $t8                     \n"
-      "srl             $t8, $t5, 16                      \n"  // |0|0|R7|R6|
-      "raddu.w.qb      $t8, $t8                          \n"  // R7 + R6
-      "addu            $t6, $t6, $t8                     \n"
-      "mul             $t6, $t6, %[c2]                   \n"  // t6 * 0x2AAA
-      "precrq.qb.ph    $t8, $t0, $t2                     \n"  // |S3|S1|T3|T1|
-      "precrq.qb.ph    $t8, $t8, $t4                     \n"  // |S3|T3|R3|R1|
-      "srl             $t8, $t8, 8                       \n"  // |0|S3|T3|R3|
-      "raddu.w.qb      $t8, $t8                          \n"  // S3 + T3 + R3
-      "addu            $t7, $t7, $t8                     \n"
-      "mul             $t7, $t7, %[c1]                   \n"  // t7 * 0x1C71
-      "sll             $t0, $t0, 8                       \n"  // |S2|S1|S0|0|
-      "sll             $t2, $t2, 8                       \n"  // |T2|T1|T0|0|
-      "sll             $t4, $t4, 8                       \n"  // |R2|R1|R0|0|
-      "raddu.w.qb      $t0, $t0                          \n"
-      "raddu.w.qb      $t2, $t2                          \n"
-      "raddu.w.qb      $t4, $t4                          \n"
-      "addu            $t0, $t0, $t2                     \n"
-      "addu            $t0, $t0, $t4                     \n"
-      "mul             $t0, $t0, %[c1]                   \n"  // t0 * 0x1C71
-      "addiu           %[src_ptr], %[src_ptr], 8         \n"
-      "addiu           %[s1], %[s1], 8                   \n"
-      "addiu           %[s2], %[s2], 8                   \n"
-      "addiu           %[dst_width], %[dst_width], -3    \n"
-      "addiu           %[dst_ptr], %[dst_ptr], 3         \n"
-      "srl             $t6, $t6, 16                      \n"
-      "srl             $t7, $t7, 16                      \n"
-      "srl             $t0, $t0, 16                      \n"
-      "sb              $t6, -1(%[dst_ptr])               \n"
-      "sb              $t7, -2(%[dst_ptr])               \n"
-      "bgtz            %[dst_width], 1b                  \n"
-      " sb             $t0, -3(%[dst_ptr])               \n"
-      ".set pop                                          \n"
-      : [src_ptr] "+r" (src_ptr),
-        [dst_ptr] "+r" (dst_ptr),
-        [s1] "+r" (s1),
-        [s2] "+r" (s2),
-        [dst_width] "+r" (dst_width)
-      : [c1] "r" (c1), [c2] "r" (c2)
-      : "t0", "t1", "t2", "t3", "t4",
-        "t5", "t6", "t7", "t8"
-  );
-}
-
-#endif  // defined(__mips_dsp) && (__mips_dsp_rev >= 2)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_neon.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_neon.cc
deleted file mode 100755
index a9df93c055..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_neon.cc
+++ /dev/null
@@ -1,699 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC Neon.
-#if !defined(LIBYUV_DISABLE_NEON) && defined(__ARM_NEON__)
-
-// NEON downscalers with interpolation.
-// Provided by Fritz Koenig
-
-// Read 32x1 throw away even pixels, and write 16x1.
-void ScaleRowDown2_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID
-	".fpu neon\n"
-#endif
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    // load even pixels into q0, odd into q1
-    "vld2.8     {q0, q1}, [%0]!                \n"
-    "subs       %2, %2, #16                    \n"  // 16 processed per loop
-    "vst1.8     {q1}, [%1]!                    \n"  // store odd pixels
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst),              // %1
-    "+r"(dst_width)         // %2
-  :
-  : "q0", "q1"              // Clobber List
-  );
-}
-
-// Read 32x2 average down and write 16x1.
-void ScaleRowDown2Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID
-	".fpu neon\n"
-#endif
-    // change the stride to row 2 pointer
-    "add        %1, %0                         \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0, q1}, [%0]!                \n"  // load row 1 and post inc
-    "vld1.8     {q2, q3}, [%1]!                \n"  // load row 2 and post inc
-    "subs       %3, %3, #16                    \n"  // 16 processed per loop
-    "vpaddl.u8  q0, q0                         \n"  // row 1 add adjacent
-    "vpaddl.u8  q1, q1                         \n"
-    "vpadal.u8  q0, q2                         \n"  // row 2 add adjacent + row1
-    "vpadal.u8  q1, q3                         \n"
-    "vrshrn.u16 d0, q0, #2                     \n"  // downshift, round and pack
-    "vrshrn.u16 d1, q1, #2                     \n"
-    "vst1.8     {q0}, [%2]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(src_stride),       // %1
-    "+r"(dst),              // %2
-    "+r"(dst_width)         // %3
-  :
-  : "q0", "q1", "q2", "q3"     // Clobber List
-  );
-}
-
-void ScaleRowDown4_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID
-	".fpu neon\n"
-#endif
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!        \n" // src line 0
-    "subs       %2, %2, #8                     \n" // 8 processed per loop
-    "vst1.8     {d2}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width)         // %2
-  :
-  : "q0", "q1", "memory", "cc"
-  );
-}
-
-void ScaleRowDown4Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID
-	".fpu neon\n"
-#endif
-    "add        r4, %0, %3                     \n"
-    "add        r5, r4, %3                     \n"
-    "add        %3, r5, %3                     \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {q0}, [%0]!                    \n"   // load up 16x4
-    "vld1.8     {q1}, [r4]!                    \n"
-    "vld1.8     {q2}, [r5]!                    \n"
-    "vld1.8     {q3}, [%3]!                    \n"
-    "subs       %2, %2, #4                     \n"
-    "vpaddl.u8  q0, q0                         \n"
-    "vpadal.u8  q0, q1                         \n"
-    "vpadal.u8  q0, q2                         \n"
-    "vpadal.u8  q0, q3                         \n"
-    "vpaddl.u16 q0, q0                         \n"
-    "vrshrn.u32 d0, q0, #4                     \n"   // divide by 16 w/rounding
-    "vmovn.u16  d0, q0                         \n"
-    "vst1.32    {d0[0]}, [%1]!                 \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width)         // %2
-  : "r"(src_stride)         // %3
-  : "r4", "r5", "q0", "q1", "q2", "q3", "memory", "cc"
-  );
-}
-
-// Down scale from 4 to 3 pixels. Use the neon multilane read/write
-// to load up the every 4th pixel into a 4 different registers.
-// Point samples 32 pixels to 24 pixels.
-void ScaleRowDown34_NEON(const uint8* src_ptr,
-                         ptrdiff_t src_stride,
-                         uint8* dst_ptr, int dst_width) {
-  asm volatile (
-#ifdef _ANDROID
-				".fpu neon\n"
-#endif
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d1, d2, d3}, [%0]!      \n" // src line 0
-    "subs       %2, %2, #24                  \n"
-    "vmov       d2, d3                       \n" // order d0, d1, d2
-    "vst3.8     {d0, d1, d2}, [%1]!          \n"
-    "bgt        1b                           \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width)         // %2
-  :
-  : "d0", "d1", "d2", "d3", "memory", "cc"
-  );
-}
-
-void ScaleRowDown34_0_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "vmov.u8    d24, #3                        \n"
-    "add        %3, %0                         \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8       {d0, d1, d2, d3}, [%0]!      \n" // src line 0
-    "vld4.8       {d4, d5, d6, d7}, [%3]!      \n" // src line 1
-    "subs         %2, %2, #24                  \n"
-
-    // filter src line 0 with src line 1
-    // expand chars to shorts to allow for room
-    // when adding lines together
-    "vmovl.u8     q8, d4                       \n"
-    "vmovl.u8     q9, d5                       \n"
-    "vmovl.u8     q10, d6                      \n"
-    "vmovl.u8     q11, d7                      \n"
-
-    // 3 * line_0 + line_1
-    "vmlal.u8     q8, d0, d24                  \n"
-    "vmlal.u8     q9, d1, d24                  \n"
-    "vmlal.u8     q10, d2, d24                 \n"
-    "vmlal.u8     q11, d3, d24                 \n"
-
-    // (3 * line_0 + line_1) >> 2
-    "vqrshrn.u16  d0, q8, #2                   \n"
-    "vqrshrn.u16  d1, q9, #2                   \n"
-    "vqrshrn.u16  d2, q10, #2                  \n"
-    "vqrshrn.u16  d3, q11, #2                  \n"
-
-    // a0 = (src[0] * 3 + s[1] * 1) >> 2
-    "vmovl.u8     q8, d1                       \n"
-    "vmlal.u8     q8, d0, d24                  \n"
-    "vqrshrn.u16  d0, q8, #2                   \n"
-
-    // a1 = (src[1] * 1 + s[2] * 1) >> 1
-    "vrhadd.u8    d1, d1, d2                   \n"
-
-    // a2 = (src[2] * 1 + s[3] * 3) >> 2
-    "vmovl.u8     q8, d2                       \n"
-    "vmlal.u8     q8, d3, d24                  \n"
-    "vqrshrn.u16  d2, q8, #2                   \n"
-
-    "vst3.8       {d0, d1, d2}, [%1]!          \n"
-
-    "bgt          1b                           \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width),        // %2
-    "+r"(src_stride)        // %3
-  :
-  : "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11", "d24", "memory", "cc"
-  );
-}
-
-void ScaleRowDown34_1_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "vmov.u8    d24, #3                        \n"
-    "add        %3, %0                         \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8       {d0, d1, d2, d3}, [%0]!      \n" // src line 0
-    "vld4.8       {d4, d5, d6, d7}, [%3]!      \n" // src line 1
-    "subs         %2, %2, #24                  \n"
-    // average src line 0 with src line 1
-    "vrhadd.u8    q0, q0, q2                   \n"
-    "vrhadd.u8    q1, q1, q3                   \n"
-
-    // a0 = (src[0] * 3 + s[1] * 1) >> 2
-    "vmovl.u8     q3, d1                       \n"
-    "vmlal.u8     q3, d0, d24                  \n"
-    "vqrshrn.u16  d0, q3, #2                   \n"
-
-    // a1 = (src[1] * 1 + s[2] * 1) >> 1
-    "vrhadd.u8    d1, d1, d2                   \n"
-
-    // a2 = (src[2] * 1 + s[3] * 3) >> 2
-    "vmovl.u8     q3, d2                       \n"
-    "vmlal.u8     q3, d3, d24                  \n"
-    "vqrshrn.u16  d2, q3, #2                   \n"
-
-    "vst3.8       {d0, d1, d2}, [%1]!          \n"
-    "bgt          1b                           \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width),        // %2
-    "+r"(src_stride)        // %3
-  :
-  : "r4", "q0", "q1", "q2", "q3", "d24", "memory", "cc"
-  );
-}
-
-#define HAS_SCALEROWDOWN38_NEON
-static uvec8 kShuf38 =
-  { 0, 3, 6, 8, 11, 14, 16, 19, 22, 24, 27, 30, 0, 0, 0, 0 };
-static uvec8 kShuf38_2 =
-  { 0, 8, 16, 2, 10, 17, 4, 12, 18, 6, 14, 19, 0, 0, 0, 0 };
-static vec16 kMult38_Div6 =
-  { 65536 / 12, 65536 / 12, 65536 / 12, 65536 / 12,
-    65536 / 12, 65536 / 12, 65536 / 12, 65536 / 12 };
-static vec16 kMult38_Div9 =
-  { 65536 / 18, 65536 / 18, 65536 / 18, 65536 / 18,
-    65536 / 18, 65536 / 18, 65536 / 18, 65536 / 18 };
-
-// 32 -> 12
-void ScaleRowDown38_NEON(const uint8* src_ptr,
-                         ptrdiff_t src_stride,
-                         uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "vld1.8     {q3}, [%3]                     \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0, d1, d2, d3}, [%0]!        \n"
-    "subs       %2, %2, #12                    \n"
-    "vtbl.u8    d4, {d0, d1, d2, d3}, d6       \n"
-    "vtbl.u8    d5, {d0, d1, d2, d3}, d7       \n"
-    "vst1.8     {d4}, [%1]!                    \n"
-    "vst1.32    {d5[0]}, [%1]!                 \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width)         // %2
-  : "r"(&kShuf38)           // %3
-  : "d0", "d1", "d2", "d3", "d4", "d5", "memory", "cc"
-  );
-}
-
-// 32x3 -> 12x1
-void OMITFP ScaleRowDown38_3_Box_NEON(const uint8* src_ptr,
-                                      ptrdiff_t src_stride,
-                                      uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "vld1.16    {q13}, [%4]                    \n"
-    "vld1.8     {q14}, [%5]                    \n"
-    "vld1.8     {q15}, [%6]                    \n"
-    "add        r4, %0, %3, lsl #1             \n"
-    "add        %3, %0                         \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-
-    // d0 = 00 40 01 41 02 42 03 43
-    // d1 = 10 50 11 51 12 52 13 53
-    // d2 = 20 60 21 61 22 62 23 63
-    // d3 = 30 70 31 71 32 72 33 73
-    "vld4.8       {d0, d1, d2, d3}, [%0]!      \n"
-    "vld4.8       {d4, d5, d6, d7}, [%3]!      \n"
-    "vld4.8       {d16, d17, d18, d19}, [r4]!  \n"
-    "subs         %2, %2, #12                  \n"
-
-    // Shuffle the input data around to get align the data
-    //  so adjacent data can be added. 0,1 - 2,3 - 4,5 - 6,7
-    // d0 = 00 10 01 11 02 12 03 13
-    // d1 = 40 50 41 51 42 52 43 53
-    "vtrn.u8      d0, d1                       \n"
-    "vtrn.u8      d4, d5                       \n"
-    "vtrn.u8      d16, d17                     \n"
-
-    // d2 = 20 30 21 31 22 32 23 33
-    // d3 = 60 70 61 71 62 72 63 73
-    "vtrn.u8      d2, d3                       \n"
-    "vtrn.u8      d6, d7                       \n"
-    "vtrn.u8      d18, d19                     \n"
-
-    // d0 = 00+10 01+11 02+12 03+13
-    // d2 = 40+50 41+51 42+52 43+53
-    "vpaddl.u8    q0, q0                       \n"
-    "vpaddl.u8    q2, q2                       \n"
-    "vpaddl.u8    q8, q8                       \n"
-
-    // d3 = 60+70 61+71 62+72 63+73
-    "vpaddl.u8    d3, d3                       \n"
-    "vpaddl.u8    d7, d7                       \n"
-    "vpaddl.u8    d19, d19                     \n"
-
-    // combine source lines
-    "vadd.u16     q0, q2                       \n"
-    "vadd.u16     q0, q8                       \n"
-    "vadd.u16     d4, d3, d7                   \n"
-    "vadd.u16     d4, d19                      \n"
-
-    // dst_ptr[3] = (s[6 + st * 0] + s[7 + st * 0]
-    //             + s[6 + st * 1] + s[7 + st * 1]
-    //             + s[6 + st * 2] + s[7 + st * 2]) / 6
-    "vqrdmulh.s16 q2, q2, q13                  \n"
-    "vmovn.u16    d4, q2                       \n"
-
-    // Shuffle 2,3 reg around so that 2 can be added to the
-    //  0,1 reg and 3 can be added to the 4,5 reg. This
-    //  requires expanding from u8 to u16 as the 0,1 and 4,5
-    //  registers are already expanded. Then do transposes
-    //  to get aligned.
-    // q2 = xx 20 xx 30 xx 21 xx 31 xx 22 xx 32 xx 23 xx 33
-    "vmovl.u8     q1, d2                       \n"
-    "vmovl.u8     q3, d6                       \n"
-    "vmovl.u8     q9, d18                      \n"
-
-    // combine source lines
-    "vadd.u16     q1, q3                       \n"
-    "vadd.u16     q1, q9                       \n"
-
-    // d4 = xx 20 xx 30 xx 22 xx 32
-    // d5 = xx 21 xx 31 xx 23 xx 33
-    "vtrn.u32     d2, d3                       \n"
-
-    // d4 = xx 20 xx 21 xx 22 xx 23
-    // d5 = xx 30 xx 31 xx 32 xx 33
-    "vtrn.u16     d2, d3                       \n"
-
-    // 0+1+2, 3+4+5
-    "vadd.u16     q0, q1                       \n"
-
-    // Need to divide, but can't downshift as the the value
-    //  isn't a power of 2. So multiply by 65536 / n
-    //  and take the upper 16 bits.
-    "vqrdmulh.s16 q0, q0, q15                  \n"
-
-    // Align for table lookup, vtbl requires registers to
-    //  be adjacent
-    "vmov.u8      d2, d4                       \n"
-
-    "vtbl.u8      d3, {d0, d1, d2}, d28        \n"
-    "vtbl.u8      d4, {d0, d1, d2}, d29        \n"
-
-    "vst1.8       {d3}, [%1]!                  \n"
-    "vst1.32      {d4[0]}, [%1]!               \n"
-    "bgt          1b                           \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst_ptr),          // %1
-    "+r"(dst_width),        // %2
-    "+r"(src_stride)        // %3
-  : "r"(&kMult38_Div6),     // %4
-    "r"(&kShuf38_2),        // %5
-    "r"(&kMult38_Div9)      // %6
-  : "r4", "q0", "q1", "q2", "q3", "q8", "q9",
-    "q13", "q14", "q15", "memory", "cc"
-  );
-}
-
-// 32x2 -> 12x1
-void ScaleRowDown38_2_Box_NEON(const uint8* src_ptr,
-                               ptrdiff_t src_stride,
-                               uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "vld1.16    {q13}, [%4]                    \n"
-    "vld1.8     {q14}, [%5]                    \n"
-    "add        %3, %0                         \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-
-    // d0 = 00 40 01 41 02 42 03 43
-    // d1 = 10 50 11 51 12 52 13 53
-    // d2 = 20 60 21 61 22 62 23 63
-    // d3 = 30 70 31 71 32 72 33 73
-    "vld4.8       {d0, d1, d2, d3}, [%0]!      \n"
-    "vld4.8       {d4, d5, d6, d7}, [%3]!      \n"
-    "subs         %2, %2, #12                  \n"
-
-    // Shuffle the input data around to get align the data
-    //  so adjacent data can be added. 0,1 - 2,3 - 4,5 - 6,7
-    // d0 = 00 10 01 11 02 12 03 13
-    // d1 = 40 50 41 51 42 52 43 53
-    "vtrn.u8      d0, d1                       \n"
-    "vtrn.u8      d4, d5                       \n"
-
-    // d2 = 20 30 21 31 22 32 23 33
-    // d3 = 60 70 61 71 62 72 63 73
-    "vtrn.u8      d2, d3                       \n"
-    "vtrn.u8      d6, d7                       \n"
-
-    // d0 = 00+10 01+11 02+12 03+13
-    // d2 = 40+50 41+51 42+52 43+53
-    "vpaddl.u8    q0, q0                       \n"
-    "vpaddl.u8    q2, q2                       \n"
-
-    // d3 = 60+70 61+71 62+72 63+73
-    "vpaddl.u8    d3, d3                       \n"
-    "vpaddl.u8    d7, d7                       \n"
-
-    // combine source lines
-    "vadd.u16     q0, q2                       \n"
-    "vadd.u16     d4, d3, d7                   \n"
-
-    // dst_ptr[3] = (s[6] + s[7] + s[6+st] + s[7+st]) / 4
-    "vqrshrn.u16  d4, q2, #2                   \n"
-
-    // Shuffle 2,3 reg around so that 2 can be added to the
-    //  0,1 reg and 3 can be added to the 4,5 reg. This
-    //  requires expanding from u8 to u16 as the 0,1 and 4,5
-    //  registers are already expanded. Then do transposes
-    //  to get aligned.
-    // q2 = xx 20 xx 30 xx 21 xx 31 xx 22 xx 32 xx 23 xx 33
-    "vmovl.u8     q1, d2                       \n"
-    "vmovl.u8     q3, d6                       \n"
-
-    // combine source lines
-    "vadd.u16     q1, q3                       \n"
-
-    // d4 = xx 20 xx 30 xx 22 xx 32
-    // d5 = xx 21 xx 31 xx 23 xx 33
-    "vtrn.u32     d2, d3                       \n"
-
-    // d4 = xx 20 xx 21 xx 22 xx 23
-    // d5 = xx 30 xx 31 xx 32 xx 33
-    "vtrn.u16     d2, d3                       \n"
-
-    // 0+1+2, 3+4+5
-    "vadd.u16     q0, q1                       \n"
-
-    // Need to divide, but can't downshift as the the value
-    //  isn't a power of 2. So multiply by 65536 / n
-    //  and take the upper 16 bits.
-    "vqrdmulh.s16 q0, q0, q13                  \n"
-
-    // Align for table lookup, vtbl requires registers to
-    //  be adjacent
-    "vmov.u8      d2, d4                       \n"
-
-    "vtbl.u8      d3, {d0, d1, d2}, d28        \n"
-    "vtbl.u8      d4, {d0, d1, d2}, d29        \n"
-
-    "vst1.8       {d3}, [%1]!                  \n"
-    "vst1.32      {d4[0]}, [%1]!               \n"
-    "bgt          1b                           \n"
-  : "+r"(src_ptr),       // %0
-    "+r"(dst_ptr),       // %1
-    "+r"(dst_width),     // %2
-    "+r"(src_stride)     // %3
-  : "r"(&kMult38_Div6),  // %4
-    "r"(&kShuf38_2)      // %5
-  : "q0", "q1", "q2", "q3", "q13", "q14", "memory", "cc"
-  );
-}
-
-// 16x2 -> 16x1
-void ScaleFilterRows_NEON(uint8* dst_ptr,
-                          const uint8* src_ptr, ptrdiff_t src_stride,
-                          int dst_width, int source_y_fraction) {
-  asm volatile (
-    "cmp          %4, #0                       \n"
-    "beq          100f                         \n"
-    "add          %2, %1                       \n"
-    "cmp          %4, #64                      \n"
-    "beq          75f                          \n"
-    "cmp          %4, #128                     \n"
-    "beq          50f                          \n"
-    "cmp          %4, #192                     \n"
-    "beq          25f                          \n"
-
-    "vdup.8       d5, %4                       \n"
-    "rsb          %4, #256                     \n"
-    "vdup.8       d4, %4                       \n"
-    // General purpose row blend.
-  "1:                                          \n"
-    "vld1.8       {q0}, [%1]!                  \n"
-    "vld1.8       {q1}, [%2]!                  \n"
-    "subs         %3, %3, #16                  \n"
-    "vmull.u8     q13, d0, d4                  \n"
-    "vmull.u8     q14, d1, d4                  \n"
-    "vmlal.u8     q13, d2, d5                  \n"
-    "vmlal.u8     q14, d3, d5                  \n"
-    "vrshrn.u16   d0, q13, #8                  \n"
-    "vrshrn.u16   d1, q14, #8                  \n"
-    "vst1.8       {q0}, [%0]!                  \n"
-    "bgt          1b                           \n"
-    "b            99f                          \n"
-
-    // Blend 25 / 75.
-  "25:                                         \n"
-    "vld1.8       {q0}, [%1]!                  \n"
-    "vld1.8       {q1}, [%2]!                  \n"
-    "subs         %3, %3, #16                  \n"
-    "vrhadd.u8    q0, q1                       \n"
-    "vrhadd.u8    q0, q1                       \n"
-    "vst1.8       {q0}, [%0]!                  \n"
-    "bgt          25b                          \n"
-    "b            99f                          \n"
-
-    // Blend 50 / 50.
-  "50:                                         \n"
-    "vld1.8       {q0}, [%1]!                  \n"
-    "vld1.8       {q1}, [%2]!                  \n"
-    "subs         %3, %3, #16                  \n"
-    "vrhadd.u8    q0, q1                       \n"
-    "vst1.8       {q0}, [%0]!                  \n"
-    "bgt          50b                          \n"
-    "b            99f                          \n"
-
-    // Blend 75 / 25.
-  "75:                                         \n"
-    "vld1.8       {q1}, [%1]!                  \n"
-    "vld1.8       {q0}, [%2]!                  \n"
-    "subs         %3, %3, #16                  \n"
-    "vrhadd.u8    q0, q1                       \n"
-    "vrhadd.u8    q0, q1                       \n"
-    "vst1.8       {q0}, [%0]!                  \n"
-    "bgt          75b                          \n"
-    "b            99f                          \n"
-
-    // Blend 100 / 0 - Copy row unchanged.
-  "100:                                        \n"
-    "vld1.8       {q0}, [%1]!                  \n"
-    "subs         %3, %3, #16                  \n"
-    "vst1.8       {q0}, [%0]!                  \n"
-    "bgt          100b                         \n"
-
-  "99:                                         \n"
-    "vst1.8       {d1[7]}, [%0]                \n"
-  : "+r"(dst_ptr),          // %0
-    "+r"(src_ptr),          // %1
-    "+r"(src_stride),       // %2
-    "+r"(dst_width),        // %3
-    "+r"(source_y_fraction) // %4
-  :
-  : "q0", "q1", "d4", "d5", "q13", "q14", "memory", "cc"
-  );
-}
-
-void ScaleARGBRowDown2_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                            uint8* dst, int dst_width) {
-  asm volatile (
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    // load even pixels into q0, odd into q1
-    "vld2.32    {q0, q1}, [%0]!                \n"
-    "vld2.32    {q2, q3}, [%0]!                \n"
-    "subs       %2, %2, #8                     \n"  // 8 processed per loop
-    "vst1.8     {q1}, [%1]!                    \n"  // store odd pixels
-    "vst1.8     {q3}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(dst),              // %1
-    "+r"(dst_width)         // %2
-  :
-  : "memory", "cc", "q0", "q1", "q2", "q3"  // Clobber List
-  );
-}
-
-void ScaleARGBRowDown2Box_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
-                               uint8* dst, int dst_width) {
-  asm volatile (
-    // change the stride to row 2 pointer
-    "add        %1, %1, %0                     \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld4.8     {d0, d2, d4, d6}, [%0]!        \n"  // load 8 ARGB pixels.
-    "vld4.8     {d1, d3, d5, d7}, [%0]!        \n"  // load next 8 ARGB pixels.
-    "subs       %3, %3, #8                     \n"  // 8 processed per loop.
-    "vpaddl.u8  q0, q0                         \n"  // B 16 bytes -> 8 shorts.
-    "vpaddl.u8  q1, q1                         \n"  // G 16 bytes -> 8 shorts.
-    "vpaddl.u8  q2, q2                         \n"  // R 16 bytes -> 8 shorts.
-    "vpaddl.u8  q3, q3                         \n"  // A 16 bytes -> 8 shorts.
-    "vld4.8     {d16, d18, d20, d22}, [%1]!    \n"  // load 8 more ARGB pixels.
-    "vld4.8     {d17, d19, d21, d23}, [%1]!    \n"  // load last 8 ARGB pixels.
-    "vpadal.u8  q0, q8                         \n"  // B 16 bytes -> 8 shorts.
-    "vpadal.u8  q1, q9                         \n"  // G 16 bytes -> 8 shorts.
-    "vpadal.u8  q2, q10                        \n"  // R 16 bytes -> 8 shorts.
-    "vpadal.u8  q3, q11                        \n"  // A 16 bytes -> 8 shorts.
-    "vrshrn.u16 d0, q0, #2                     \n"  // downshift, round and pack
-    "vrshrn.u16 d1, q1, #2                     \n"
-    "vrshrn.u16 d2, q2, #2                     \n"
-    "vrshrn.u16 d3, q3, #2                     \n"
-    "vst4.8     {d0, d1, d2, d3}, [%2]!        \n"
-    "bgt        1b                             \n"
-  : "+r"(src_ptr),          // %0
-    "+r"(src_stride),       // %1
-    "+r"(dst),              // %2
-    "+r"(dst_width)         // %3
-  :
-  : "memory", "cc", "q0", "q1", "q2", "q3", "q8", "q9", "q10", "q11"
-  );
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: src_argb 4 byte aligned.
-void ScaleARGBRowDownEven_NEON(const uint8* src_argb,  ptrdiff_t src_stride,
-                               int src_stepx, uint8* dst_argb, int dst_width) {
-  asm volatile (
-    "mov        r12, %3, lsl #2                \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.32    {d0[0]}, [%0], r12             \n"
-    "vld1.32    {d0[1]}, [%0], r12             \n"
-    "vld1.32    {d1[0]}, [%0], r12             \n"
-    "vld1.32    {d1[1]}, [%0], r12             \n"
-    "subs       %2, %2, #4                     \n"  // 4 pixels per loop.
-    "vst1.8     {q0}, [%1]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_argb),    // %0
-    "+r"(dst_argb),    // %1
-    "+r"(dst_width)    // %2
-  : "r"(src_stepx)     // %3
-  : "memory", "cc", "r12", "q0"
-  );
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: src_argb 4 byte aligned.
-void ScaleARGBRowDownEvenBox_NEON(const uint8* src_argb, ptrdiff_t src_stride,
-                                  int src_stepx,
-                                  uint8* dst_argb, int dst_width) {
-  asm volatile (
-    "mov        r12, %4, lsl #2                \n"
-    "add        %1, %1, %0                     \n"
-    ".p2align   2                              \n"
-  "1:                                          \n"
-    "vld1.8     {d0}, [%0], r12                \n"  // Read 4 2x2 blocks -> 2x1
-    "vld1.8     {d1}, [%1], r12                \n"
-    "vld1.8     {d2}, [%0], r12                \n"
-    "vld1.8     {d3}, [%1], r12                \n"
-    "vld1.8     {d4}, [%0], r12                \n"
-    "vld1.8     {d5}, [%1], r12                \n"
-    "vld1.8     {d6}, [%0], r12                \n"
-    "vld1.8     {d7}, [%1], r12                \n"
-    "vaddl.u8   q0, d0, d1                     \n"
-    "vaddl.u8   q1, d2, d3                     \n"
-    "vaddl.u8   q2, d4, d5                     \n"
-    "vaddl.u8   q3, d6, d7                     \n"
-    "vswp.8     d1, d2                         \n"  // ab_cd -> ac_bd
-    "vswp.8     d5, d6                         \n"  // ef_gh -> eg_fh
-    "vadd.u16   q0, q0, q1                     \n"  // (a+b)_(c+d)
-    "vadd.u16   q2, q2, q3                     \n"  // (e+f)_(g+h)
-    "vrshrn.u16 d0, q0, #2                     \n"  // first 2 pixels.
-    "vrshrn.u16 d1, q2, #2                     \n"  // next 2 pixels.
-    "subs       %3, %3, #4                     \n"  // 4 pixels per loop.
-    "vst1.8     {q0}, [%2]!                    \n"
-    "bgt        1b                             \n"
-  : "+r"(src_argb),    // %0
-    "+r"(src_stride),  // %1
-    "+r"(dst_argb),    // %2
-    "+r"(dst_width)    // %3
-  : "r"(src_stepx)     // %4
-  : "memory", "cc", "r12", "q0", "q1", "q2", "q3"
-  );
-}
-
-#endif  // __ARM_NEON__
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_posix.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_posix.cc
deleted file mode 100644
index 352e667822..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_posix.cc
+++ /dev/null
@@ -1,1315 +0,0 @@
-/*
- *  Copyright 2013 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for GCC x86 and x64.
-#if !defined(LIBYUV_DISABLE_X86) && (defined(__x86_64__) || defined(__i386__))
-
-// Offsets for source bytes 0 to 9
-static uvec8 kShuf0 =
-  { 0, 1, 3, 4, 5, 7, 8, 9, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 11 to 20 with 8 subtracted = 3 to 12.
-static uvec8 kShuf1 =
-  { 3, 4, 5, 7, 8, 9, 11, 12, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
-static uvec8 kShuf2 =
-  { 5, 7, 8, 9, 11, 12, 13, 15, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 0 to 10
-static uvec8 kShuf01 =
-  { 0, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10 };
-
-// Offsets for source bytes 10 to 21 with 8 subtracted = 3 to 13.
-static uvec8 kShuf11 =
-  { 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13 };
-
-// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
-static uvec8 kShuf21 =
-  { 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 13, 14, 14, 15 };
-
-// Coefficients for source bytes 0 to 10
-static uvec8 kMadd01 =
-  { 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2 };
-
-// Coefficients for source bytes 10 to 21
-static uvec8 kMadd11 =
-  { 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1 };
-
-// Coefficients for source bytes 21 to 31
-static uvec8 kMadd21 =
-  { 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3 };
-
-// Coefficients for source bytes 21 to 31
-static vec16 kRound34 =
-  { 2, 2, 2, 2, 2, 2, 2, 2 };
-
-static uvec8 kShuf38a =
-  { 0, 3, 6, 8, 11, 14, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-static uvec8 kShuf38b =
-  { 128, 128, 128, 128, 128, 128, 0, 3, 6, 8, 11, 14, 128, 128, 128, 128 };
-
-// Arrange words 0,3,6 into 0,1,2
-static uvec8 kShufAc =
-  { 0, 1, 6, 7, 12, 13, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Arrange words 0,3,6 into 3,4,5
-static uvec8 kShufAc3 =
-  { 128, 128, 128, 128, 128, 128, 0, 1, 6, 7, 12, 13, 128, 128, 128, 128 };
-
-// Scaling values for boxes of 3x3 and 2x3
-static uvec16 kScaleAc33 =
-  { 65536 / 9, 65536 / 9, 65536 / 6, 65536 / 9, 65536 / 9, 65536 / 6, 0, 0 };
-
-// Arrange first value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb0 =
-  { 0, 128, 3, 128, 6, 128, 8, 128, 11, 128, 14, 128, 128, 128, 128, 128 };
-
-// Arrange second value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb1 =
-  { 1, 128, 4, 128, 7, 128, 9, 128, 12, 128, 15, 128, 128, 128, 128, 128 };
-
-// Arrange third value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb2 =
-  { 2, 128, 5, 128, 128, 128, 10, 128, 13, 128, 128, 128, 128, 128, 128, 128 };
-
-// Scaling values for boxes of 3x2 and 2x2
-static uvec16 kScaleAb2 =
-  { 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
-
-// GCC versions of row functions are verbatim conversions from Visual C.
-// Generated using gcc disassembly on Visual C object file:
-// objdump -D yuvscaler.obj >yuvscaler.txt
-
-void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10, 0) ",%%xmm1  \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "pand      %%xmm5,%%xmm3                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "pavgw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    MEMOPREG(movdqa,0x00,0,3,1,xmm2)           //  movdqa  (%0,%3,1),%%xmm2
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x10,0,3,1,xmm3)           //  movdqa  0x10(%0,%3,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "pand      %%xmm5,%%xmm3                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "pavgw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  : "r"((intptr_t)(src_stride))   // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown2_Unaligned_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                                  uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void ScaleRowDown2Linear_Unaligned_SSE2(const uint8* src_ptr,
-                                        ptrdiff_t src_stride,
-                                        uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "pand      %%xmm5,%%xmm3                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "pavgw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown2Box_Unaligned_SSE2(const uint8* src_ptr,
-                                     ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrlw     $0x8,%%xmm5                     \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqu    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqu    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    MEMOPREG(movdqu,0x00,0,3,1,xmm2)           //  movdqu  (%0,%3,1),%%xmm2
-    BUNDLEALIGN
-    MEMOPREG(movdqu,0x10,0,3,1,xmm3)           //  movdqu  0x10(%0,%3,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "pand      %%xmm5,%%xmm2                   \n"
-    "pand      %%xmm5,%%xmm3                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "pavgw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqu    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "sub       $0x10,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  : "r"((intptr_t)(src_stride))   // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "pcmpeqb   %%xmm5,%%xmm5                   \n"
-    "psrld     $0x18,%%xmm5                    \n"
-    "pslld     $0x10,%%xmm5                    \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pand      %%xmm5,%%xmm0                   \n"
-    "pand      %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width) {
-  intptr_t stridex3 = 0;
-  asm volatile (
-    "pcmpeqb   %%xmm7,%%xmm7                   \n"
-    "psrlw     $0x8,%%xmm7                     \n"
-    "lea       " MEMLEA4(0x00,4,4,2) ",%3      \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    MEMOPREG(movdqa,0x00,0,4,1,xmm2)           //  movdqa  (%0,%4,1),%%xmm2
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x10,0,4,1,xmm3)           //  movdqa  0x10(%0,%4,1),%%xmm3
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    MEMOPREG(movdqa,0x00,0,4,2,xmm2)           //  movdqa  (%0,%4,2),%%xmm2
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x10,0,4,2,xmm3)           //  movdqa  0x10(%0,%4,2),%%xmm3
-    MEMOPREG(movdqa,0x00,0,3,1,xmm4)           //  movdqa  (%0,%3,1),%%xmm4
-    MEMOPREG(movdqa,0x10,0,3,1,xmm5)           //  movdqa  0x10(%0,%3,1),%%xmm5
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm4,%%xmm2                   \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm5,%%xmm3                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "movdqa    %%xmm1,%%xmm3                   \n"
-    "psrlw     $0x8,%%xmm1                     \n"
-    "pand      %%xmm7,%%xmm2                   \n"
-    "pand      %%xmm7,%%xmm3                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "pavgw     %%xmm3,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm0                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "psrlw     $0x8,%%xmm0                     \n"
-    "pand      %%xmm7,%%xmm2                   \n"
-    "pavgw     %%xmm2,%%xmm0                   \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x8,1) ",%1            \n"
-    "sub       $0x8,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),     // %0
-    "+r"(dst_ptr),     // %1
-    "+r"(dst_width),   // %2
-    "+r"(stridex3)     // %3
-  : "r"((intptr_t)(src_stride))    // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm7"
-#endif
-  );
-}
-
-void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %0,%%xmm3                       \n"
-    "movdqa    %1,%%xmm4                       \n"
-    "movdqa    %2,%%xmm5                       \n"
-  :
-  : "m"(kShuf0),  // %0
-    "m"(kShuf1),  // %1
-    "m"(kShuf2)   // %2
-  );
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm2   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "palignr   $0x8,%%xmm0,%%xmm1              \n"
-    "pshufb    %%xmm3,%%xmm0                   \n"
-    "pshufb    %%xmm4,%%xmm1                   \n"
-    "pshufb    %%xmm5,%%xmm2                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "movq      %%xmm1," MEMACCESS2(0x8,1) "    \n"
-    "movq      %%xmm2," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x18,1) ",%1           \n"
-    "sub       $0x18,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),   // %0
-    "+r"(dst_ptr),   // %1
-    "+r"(dst_width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %0,%%xmm2                       \n"  // kShuf01
-    "movdqa    %1,%%xmm3                       \n"  // kShuf11
-    "movdqa    %2,%%xmm4                       \n"  // kShuf21
-  :
-  : "m"(kShuf01),  // %0
-    "m"(kShuf11),  // %1
-    "m"(kShuf21)   // %2
-  );
-  asm volatile (
-    "movdqa    %0,%%xmm5                       \n"  // kMadd01
-    "movdqa    %1,%%xmm0                       \n"  // kMadd11
-    "movdqa    %2,%%xmm1                       \n"  // kRound34
-  :
-  : "m"(kMadd01),  // %0
-    "m"(kMadd11),  // %1
-    "m"(kRound34)  // %2
-  );
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm6         \n"
-    MEMOPREG(movdqa,0x00,0,3,1,xmm7)           //  movdqa  (%0,%3),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm5,%%xmm6                   \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS(1) "         \n"
-    "movdqu    " MEMACCESS2(0x8,0) ",%%xmm6    \n"
-    MEMOPREG(movdqu,0x8,0,3,1,xmm7)            //  movdqu  0x8(%0,%3),%%xmm7
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm3,%%xmm6                   \n"
-    "pmaddubsw %%xmm0,%%xmm6                   \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS2(0x8,1) "    \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm6   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x10,0,3,1,xmm7)           //  movdqa  0x10(%0,%3),%%xmm7
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm4,%%xmm6                   \n"
-    "pmaddubsw %4,%%xmm6                       \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x18,1) ",%1           \n"
-    "sub       $0x18,%2                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),   // %0
-    "+r"(dst_ptr),   // %1
-    "+r"(dst_width)  // %2
-  : "r"((intptr_t)(src_stride)),  // %3
-    "m"(kMadd21)     // %4
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %0,%%xmm2                       \n"  // kShuf01
-    "movdqa    %1,%%xmm3                       \n"  // kShuf11
-    "movdqa    %2,%%xmm4                       \n"  // kShuf21
-  :
-  : "m"(kShuf01),  // %0
-    "m"(kShuf11),  // %1
-    "m"(kShuf21)   // %2
-  );
-  asm volatile (
-    "movdqa    %0,%%xmm5                       \n"  // kMadd01
-    "movdqa    %1,%%xmm0                       \n"  // kMadd11
-    "movdqa    %2,%%xmm1                       \n"  // kRound34
-  :
-  : "m"(kMadd01),  // %0
-    "m"(kMadd11),  // %1
-    "m"(kRound34)  // %2
-  );
-
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm6         \n"
-    MEMOPREG(movdqa,0x00,0,3,1,xmm7)           //  movdqa  (%0,%3,1),%%xmm7
-    "pavgb     %%xmm6,%%xmm7                   \n"
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm2,%%xmm6                   \n"
-    "pmaddubsw %%xmm5,%%xmm6                   \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS(1) "         \n"
-    "movdqu    " MEMACCESS2(0x8,0) ",%%xmm6    \n"
-    MEMOPREG(movdqu,0x8,0,3,1,xmm7)            //  movdqu  0x8(%0,%3,1),%%xmm7
-    "pavgb     %%xmm6,%%xmm7                   \n"
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm3,%%xmm6                   \n"
-    "pmaddubsw %%xmm0,%%xmm6                   \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS2(0x8,1) "    \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm6   \n"
-    MEMOPREG(movdqa,0x10,0,3,1,xmm7)           //  movdqa  0x10(%0,%3,1),%%xmm7
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm6,%%xmm7                   \n"
-    "pavgb     %%xmm7,%%xmm6                   \n"
-    "pshufb    %%xmm4,%%xmm6                   \n"
-    "pmaddubsw %4,%%xmm6                       \n"
-    "paddsw    %%xmm1,%%xmm6                   \n"
-    "psrlw     $0x2,%%xmm6                     \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "movq      %%xmm6," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x18,1) ",%1           \n"
-    "sub       $0x18,%2                        \n"
-    "jg        1b                              \n"
-    : "+r"(src_ptr),   // %0
-      "+r"(dst_ptr),   // %1
-      "+r"(dst_width)  // %2
-    : "r"((intptr_t)(src_stride)),  // %3
-      "m"(kMadd21)     // %4
-    : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %3,%%xmm4                       \n"
-    "movdqa    %4,%%xmm5                       \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "paddusb   %%xmm1,%%xmm0                   \n"
-    "movq      %%xmm0," MEMACCESS(1) "         \n"
-    "movhlps   %%xmm0,%%xmm1                   \n"
-    "movd      %%xmm1," MEMACCESS2(0x8,1) "    \n"
-    "lea       " MEMLEA(0xc,1) ",%1            \n"
-    "sub       $0xc,%2                         \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),   // %0
-    "+r"(dst_ptr),   // %1
-    "+r"(dst_width)  // %2
-  : "m"(kShuf38a),   // %3
-    "m"(kShuf38b)    // %4
-  : "memory", "cc"
-#if defined(__SSE2__)
-      , "xmm0", "xmm1", "xmm4", "xmm5"
-#endif
-  );
-}
-
-void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %0,%%xmm2                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm4                       \n"
-    "movdqa    %3,%%xmm5                       \n"
-  :
-  : "m"(kShufAb0),   // %0
-    "m"(kShufAb1),   // %1
-    "m"(kShufAb2),   // %2
-    "m"(kScaleAb2)   // %3
-  );
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(pavgb,0x00,0,3,1,xmm0)            //  pavgb   (%0,%3,1),%%xmm0
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "pshufb    %%xmm2,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm6                   \n"
-    "pshufb    %%xmm3,%%xmm6                   \n"
-    "paddusw   %%xmm6,%%xmm1                   \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "paddusw   %%xmm0,%%xmm1                   \n"
-    "pmulhuw   %%xmm5,%%xmm1                   \n"
-    "packuswb  %%xmm1,%%xmm1                   \n"
-    "sub       $0x6,%2                         \n"
-    "movd      %%xmm1," MEMACCESS(1) "         \n"
-    "psrlq     $0x10,%%xmm1                    \n"
-    "movd      %%xmm1," MEMACCESS2(0x2,1) "    \n"
-    "lea       " MEMLEA(0x6,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),     // %0
-    "+r"(dst_ptr),     // %1
-    "+r"(dst_width)    // %2
-  : "r"((intptr_t)(src_stride))  // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  asm volatile (
-    "movdqa    %0,%%xmm2                       \n"
-    "movdqa    %1,%%xmm3                       \n"
-    "movdqa    %2,%%xmm4                       \n"
-    "pxor      %%xmm5,%%xmm5                   \n"
-  :
-  : "m"(kShufAc),    // %0
-    "m"(kShufAc3),   // %1
-    "m"(kScaleAc33)  // %2
-  );
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movdqa,0x00,0,3,1,xmm6)           //  movdqa  (%0,%3,1),%%xmm6
-    "movhlps   %%xmm0,%%xmm1                   \n"
-    "movhlps   %%xmm6,%%xmm7                   \n"
-    "punpcklbw %%xmm5,%%xmm0                   \n"
-    "punpcklbw %%xmm5,%%xmm1                   \n"
-    "punpcklbw %%xmm5,%%xmm6                   \n"
-    "punpcklbw %%xmm5,%%xmm7                   \n"
-    "paddusw   %%xmm6,%%xmm0                   \n"
-    "paddusw   %%xmm7,%%xmm1                   \n"
-    MEMOPREG(movdqa,0x00,0,3,2,xmm6)           //  movdqa  (%0,%3,2),%%xmm6
-    "lea       " MEMLEA(0x10,0) ",%0           \n"
-    "movhlps   %%xmm6,%%xmm7                   \n"
-    "punpcklbw %%xmm5,%%xmm6                   \n"
-    "punpcklbw %%xmm5,%%xmm7                   \n"
-    "paddusw   %%xmm6,%%xmm0                   \n"
-    "paddusw   %%xmm7,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm6                   \n"
-    "psrldq    $0x2,%%xmm0                     \n"
-    "paddusw   %%xmm0,%%xmm6                   \n"
-    "psrldq    $0x2,%%xmm0                     \n"
-    "paddusw   %%xmm0,%%xmm6                   \n"
-    "pshufb    %%xmm2,%%xmm6                   \n"
-    "movdqa    %%xmm1,%%xmm7                   \n"
-    "psrldq    $0x2,%%xmm1                     \n"
-    "paddusw   %%xmm1,%%xmm7                   \n"
-    "psrldq    $0x2,%%xmm1                     \n"
-    "paddusw   %%xmm1,%%xmm7                   \n"
-    "pshufb    %%xmm3,%%xmm7                   \n"
-    "paddusw   %%xmm7,%%xmm6                   \n"
-    "pmulhuw   %%xmm4,%%xmm6                   \n"
-    "packuswb  %%xmm6,%%xmm6                   \n"
-    "sub       $0x6,%2                         \n"
-    "movd      %%xmm6," MEMACCESS(1) "         \n"
-    "psrlq     $0x10,%%xmm6                    \n"
-    "movd      %%xmm6," MEMACCESS2(0x2,1) "    \n"
-    "lea       " MEMLEA(0x6,1) ",%1            \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),    // %0
-    "+r"(dst_ptr),    // %1
-    "+r"(dst_width)   // %2
-  : "r"((intptr_t)(src_stride))   // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
-#endif
-  );
-}
-
-void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                       uint16* dst_ptr, int src_width, int src_height) {
-  int tmp_height = 0;
-  intptr_t tmp_src = 0;
-  asm volatile (
-    "pxor      %%xmm4,%%xmm4                   \n"
-    "sub       $0x1,%5                         \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "mov       %0,%3                           \n"
-    "add       %6,%0                           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm4,%%xmm0                   \n"
-    "punpckhbw %%xmm4,%%xmm1                   \n"
-    "mov       %5,%2                           \n"
-    "test      %2,%2                           \n"
-    "je        3f                              \n"
-
-    LABELALIGN
-  "2:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm2         \n"
-    "add       %6,%0                           \n"
-    "movdqa    %%xmm2,%%xmm3                   \n"
-    "punpcklbw %%xmm4,%%xmm2                   \n"
-    "punpckhbw %%xmm4,%%xmm3                   \n"
-    "paddusw   %%xmm2,%%xmm0                   \n"
-    "paddusw   %%xmm3,%%xmm1                   \n"
-    "sub       $0x1,%2                         \n"
-    "jg        2b                              \n"
-
-    LABELALIGN
-  "3:                                          \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,1) "   \n"
-    "lea       " MEMLEA(0x10,3) ",%0           \n"
-    "lea       " MEMLEA(0x20,1) ",%1           \n"
-    "sub       $0x10,%4                        \n"
-    "jg        1b                              \n"
-  : "+r"(src_ptr),     // %0
-    "+r"(dst_ptr),     // %1
-    "+r"(tmp_height),  // %2
-    "+r"(tmp_src),     // %3
-    "+r"(src_width),   // %4
-    "+rm"(src_height)  // %5
-  : "rm"((intptr_t)(src_stride))  // %6
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4"
-#endif
-  );
-}
-
-// Bilinear column filtering. SSSE3 version.
-void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                           int dst_width, int x, int dx) {
-  intptr_t x0 = 0, x1 = 0, temp_pixel = 0;
-  asm volatile (
-    "movd      %6,%%xmm2                       \n"
-    "movd      %7,%%xmm3                       \n"
-    "movl      $0x04040000,%k2                 \n"
-    "movd      %k2,%%xmm5                      \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "psrlw     $0x9,%%xmm6                     \n"
-    "pextrw    $0x1,%%xmm2,%k3                 \n"
-    "subl      $0x2,%5                         \n"
-    "jl        29f                             \n"
-    "movdqa    %%xmm2,%%xmm0                   \n"
-    "paddd     %%xmm3,%%xmm0                   \n"
-    "punpckldq %%xmm0,%%xmm2                   \n"
-    "punpckldq %%xmm3,%%xmm3                   \n"
-    "paddd     %%xmm3,%%xmm3                   \n"
-    "pextrw    $0x3,%%xmm2,%k4                 \n"
-
-    LABELALIGN
-  "2:                                          \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "paddd     %%xmm3,%%xmm2                   \n"
-    MEMOPARG(movzwl,0x00,1,3,1,k2)             //  movzwl  (%1,%3,1),%k2
-    "movd      %k2,%%xmm0                      \n"
-    "psrlw     $0x9,%%xmm1                     \n"
-    BUNDLEALIGN
-    MEMOPARG(movzwl,0x00,1,4,1,k2)             //  movzwl  (%1,%4,1),%k2
-    "movd      %k2,%%xmm4                      \n"
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "punpcklwd %%xmm4,%%xmm0                   \n"
-    "pxor      %%xmm6,%%xmm1                   \n"
-    "pmaddubsw %%xmm1,%%xmm0                   \n"
-    "pextrw    $0x1,%%xmm2,%k3                 \n"
-    "pextrw    $0x3,%%xmm2,%k4                 \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movd      %%xmm0,%k2                      \n"
-    "mov       %w2," MEMACCESS(0) "            \n"
-    "lea       " MEMLEA(0x2,0) ",%0            \n"
-    "sub       $0x2,%5                         \n"
-    "jge       2b                              \n"
-
-    LABELALIGN
-  "29:                                         \n"
-    "addl      $0x1,%5                         \n"
-    "jl        99f                             \n"
-    MEMOPARG(movzwl,0x00,1,3,1,k2)             //  movzwl  (%1,%3,1),%k2
-    "movd      %k2,%%xmm0                      \n"
-    "psrlw     $0x9,%%xmm2                     \n"
-    "pshufb    %%xmm5,%%xmm2                   \n"
-    "pxor      %%xmm6,%%xmm2                   \n"
-    "pmaddubsw %%xmm2,%%xmm0                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movd      %%xmm0,%k2                      \n"
-    "mov       %b2," MEMACCESS(0) "            \n"
-  "99:                                         \n"
-  : "+r"(dst_ptr),     // %0
-    "+r"(src_ptr),     // %1
-    "+a"(temp_pixel),  // %2
-    "+r"(x0),          // %3
-    "+r"(x1),          // %4
-    "+rm"(dst_width)   // %5
-  : "rm"(x),           // %6
-    "rm"(dx)           // %7
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-// Reads 4 pixels, duplicates them and writes 8 pixels.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                       int dst_width, int x, int dx) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpcklbw %%xmm0,%%xmm0                   \n"
-    "punpckhbw %%xmm1,%%xmm1                   \n"
-    "sub       $0x20,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(0) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,0) "   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "jg        1b                              \n"
-
-  : "+r"(dst_ptr),     // %0
-    "+r"(src_ptr),     // %1
-    "+r"(dst_width)    // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void ScaleARGBRowDown2_SSE2(const uint8* src_argb,
-                            ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "shufps    $0xdd,%%xmm1,%%xmm0             \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(dst_width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride,
-                                  uint8* dst_argb, int dst_width) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm2             \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),  // %0
-    "+r"(dst_argb),  // %1
-    "+r"(dst_width)  // %2
-  :
-  : "memory", "cc"
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               uint8* dst_argb, int dst_width) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(0) ",%%xmm0         \n"
-    "movdqa    " MEMACCESS2(0x10,0) ",%%xmm1   \n"
-    BUNDLEALIGN
-    MEMOPREG(movdqa,0x00,0,3,1,xmm2)           //  movdqa   (%0,%3,1),%%xmm2
-    MEMOPREG(movdqa,0x10,0,3,1,xmm3)           //  movdqa   0x10(%0,%3,1),%%xmm3
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm2             \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "sub       $0x4,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(1) "         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),   // %0
-    "+r"(dst_argb),   // %1
-    "+r"(dst_width)   // %2
-  : "r"((intptr_t)(src_stride))   // %3
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3"
-#endif
-  );
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: dst_argb 16 byte aligned.
-void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width) {
-  intptr_t src_stepx_x4 = (intptr_t)(src_stepx);
-  intptr_t src_stepx_x12 = 0;
-  asm volatile (
-    "lea       " MEMLEA3(0x00,1,4) ",%1        \n"
-    "lea       " MEMLEA4(0x00,1,1,2) ",%4      \n"
-    LABELALIGN
-  "1:                                          \n"
-    "movd      " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movd,0x00,0,1,1,xmm1)             //  movd      (%0,%1,1),%%xmm1
-    "punpckldq %%xmm1,%%xmm0                   \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,0,1,2,xmm2)             //  movd      (%0,%1,2),%%xmm2
-    MEMOPREG(movd,0x00,0,4,1,xmm3)             //  movd      (%0,%4,1),%%xmm3
-    "lea       " MEMLEA4(0x00,0,1,4) ",%0      \n"
-    "punpckldq %%xmm3,%%xmm2                   \n"
-    "punpcklqdq %%xmm2,%%xmm0                  \n"
-    "sub       $0x4,%3                         \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),      // %0
-    "+r"(src_stepx_x4),  // %1
-    "+r"(dst_argb),      // %2
-    "+r"(dst_width),     // %3
-    "+r"(src_stepx_x12)  // %4
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3"
-#endif
-  );
-}
-
-// Blends four 2x2 to 4x1.
-// Alignment requirement: dst_argb 16 byte aligned.
-void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride, int src_stepx,
-                                  uint8* dst_argb, int dst_width) {
-  intptr_t src_stepx_x4 = (intptr_t)(src_stepx);
-  intptr_t src_stepx_x12 = 0;
-  intptr_t row1 = (intptr_t)(src_stride);
-  asm volatile (
-    "lea       " MEMLEA3(0x00,1,4) ",%1        \n"
-    "lea       " MEMLEA4(0x00,1,1,2) ",%4      \n"
-    "lea       " MEMLEA4(0x00,0,5,1) ",%5      \n"
-
-    LABELALIGN
-  "1:                                          \n"
-    "movq      " MEMACCESS(0) ",%%xmm0         \n"
-    MEMOPREG(movhps,0x00,0,1,1,xmm0)           //  movhps    (%0,%1,1),%%xmm0
-    MEMOPREG(movq,0x00,0,1,2,xmm1)             //  movq      (%0,%1,2),%%xmm1
-    BUNDLEALIGN
-    MEMOPREG(movhps,0x00,0,4,1,xmm1)           //  movhps    (%0,%4,1),%%xmm1
-    "lea       " MEMLEA4(0x00,0,1,4) ",%0      \n"
-    "movq      " MEMACCESS(5) ",%%xmm2         \n"
-    BUNDLEALIGN
-    MEMOPREG(movhps,0x00,5,1,1,xmm2)           //  movhps    (%5,%1,1),%%xmm2
-    MEMOPREG(movq,0x00,5,1,2,xmm3)             //  movq      (%5,%1,2),%%xmm3
-    MEMOPREG(movhps,0x00,5,4,1,xmm3)           //  movhps    (%5,%4,1),%%xmm3
-    "lea       " MEMLEA4(0x00,5,1,4) ",%5      \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "pavgb     %%xmm3,%%xmm1                   \n"
-    "movdqa    %%xmm0,%%xmm2                   \n"
-    "shufps    $0x88,%%xmm1,%%xmm0             \n"
-    "shufps    $0xdd,%%xmm1,%%xmm2             \n"
-    "pavgb     %%xmm2,%%xmm0                   \n"
-    "sub       $0x4,%3                         \n"
-    "movdqa    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jg        1b                              \n"
-  : "+r"(src_argb),       // %0
-    "+r"(src_stepx_x4),   // %1
-    "+r"(dst_argb),       // %2
-    "+rm"(dst_width),     // %3
-    "+r"(src_stepx_x12),  // %4
-    "+r"(row1)            // %5
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3"
-#endif
-  );
-}
-
-void ScaleARGBCols_SSE2(uint8* dst_argb, const uint8* src_argb,
-                        int dst_width, int x, int dx) {
-  intptr_t x0 = 0, x1 = 0;
-  asm volatile (
-    "movd      %5,%%xmm2                       \n"
-    "movd      %6,%%xmm3                       \n"
-    "pshufd    $0x0,%%xmm2,%%xmm2              \n"
-    "pshufd    $0x11,%%xmm3,%%xmm0             \n"
-    "paddd     %%xmm0,%%xmm2                   \n"
-    "paddd     %%xmm3,%%xmm3                   \n"
-    "pshufd    $0x5,%%xmm3,%%xmm0              \n"
-    "paddd     %%xmm0,%%xmm2                   \n"
-    "paddd     %%xmm3,%%xmm3                   \n"
-    "pshufd    $0x0,%%xmm3,%%xmm3              \n"
-    "pextrw    $0x1,%%xmm2,%k0                 \n"
-    "pextrw    $0x3,%%xmm2,%k1                 \n"
-    "cmp       $0x0,%4                         \n"
-    "jl        99f                             \n"
-    "sub       $0x4,%4                         \n"
-    "jl        49f                             \n"
-
-    LABELALIGN
-  "40:                                         \n"
-    MEMOPREG(movd,0x00,3,0,4,xmm0)             //  movd      (%3,%0,4),%%xmm0
-    MEMOPREG(movd,0x00,3,1,4,xmm1)             //  movd      (%3,%1,4),%%xmm1
-    "pextrw    $0x5,%%xmm2,%k0                 \n"
-    "pextrw    $0x7,%%xmm2,%k1                 \n"
-    "paddd     %%xmm3,%%xmm2                   \n"
-    "punpckldq %%xmm1,%%xmm0                   \n"
-    MEMOPREG(movd,0x00,3,0,4,xmm1)             //  movd      (%3,%0,4),%%xmm1
-    MEMOPREG(movd,0x00,3,1,4,xmm4)             //  movd      (%3,%1,4),%%xmm4
-    "pextrw    $0x1,%%xmm2,%k0                 \n"
-    "pextrw    $0x3,%%xmm2,%k1                 \n"
-    "punpckldq %%xmm4,%%xmm1                   \n"
-    "punpcklqdq %%xmm1,%%xmm0                  \n"
-    "sub       $0x4,%4                         \n"
-    "movdqu    %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x10,2) ",%2           \n"
-    "jge       40b                             \n"
-
-  "49:                                         \n"
-    "test      $0x2,%4                         \n"
-    "je        29f                             \n"
-    BUNDLEALIGN
-    MEMOPREG(movd,0x00,3,0,4,xmm0)             //  movd      (%3,%0,4),%%xmm0
-    MEMOPREG(movd,0x00,3,1,4,xmm1)             //  movd      (%3,%1,4),%%xmm1
-    "pextrw    $0x5,%%xmm2,%k0                 \n"
-    "punpckldq %%xmm1,%%xmm0                   \n"
-    "movq      %%xmm0," MEMACCESS(2) "         \n"
-    "lea       " MEMLEA(0x8,2) ",%2            \n"
-  "29:                                         \n"
-    "test      $0x1,%4                         \n"
-    "je        99f                             \n"
-    MEMOPREG(movd,0x00,3,0,4,xmm0)             //  movd      (%3,%0,4),%%xmm0
-    "movd      %%xmm0," MEMACCESS(2) "         \n"
-  "99:                                         \n"
-  : "+a"(x0),          // %0
-    "+d"(x1),          // %1
-    "+r"(dst_argb),    // %2
-    "+r"(src_argb),    // %3
-    "+r"(dst_width)    // %4
-  : "rm"(x),           // %5
-    "rm"(dx)           // %6
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4"
-#endif
-  );
-}
-
-// Reads 4 pixels, duplicates them and writes 8 pixels.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb,
-                           int dst_width, int x, int dx) {
-  asm volatile (
-    LABELALIGN
-  "1:                                          \n"
-    "movdqa    " MEMACCESS(1) ",%%xmm0         \n"
-    "lea       " MEMLEA(0x10,1) ",%1           \n"
-    "movdqa    %%xmm0,%%xmm1                   \n"
-    "punpckldq %%xmm0,%%xmm0                   \n"
-    "punpckhdq %%xmm1,%%xmm1                   \n"
-    "sub       $0x8,%2                         \n"
-    "movdqa    %%xmm0," MEMACCESS(0) "         \n"
-    "movdqa    %%xmm1," MEMACCESS2(0x10,0) "   \n"
-    "lea       " MEMLEA(0x20,0) ",%0           \n"
-    "jg        1b                              \n"
-
-  : "+r"(dst_argb),    // %0
-    "+r"(src_argb),    // %1
-    "+r"(dst_width)    // %2
-  :
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1"
-#endif
-  );
-}
-
-// Shuffle table for arranging 2 pixels into pairs for pmaddubsw
-static uvec8 kShuffleColARGB = {
-  0u, 4u, 1u, 5u, 2u, 6u, 3u, 7u,  // bbggrraa 1st pixel
-  8u, 12u, 9u, 13u, 10u, 14u, 11u, 15u  // bbggrraa 2nd pixel
-};
-
-// Shuffle table for duplicating 2 fractions into 8 bytes each
-static uvec8 kShuffleFractions = {
-  0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 4u, 4u, 4u, 4u, 4u, 4u, 4u, 4u,
-};
-
-// Bilinear row filtering combines 4x2 -> 4x1. SSSE3 version
-void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb,
-                               int dst_width, int x, int dx) {
-  intptr_t x0 = 0, x1 = 0;
-  asm volatile (
-    "movdqa    %0,%%xmm4                       \n"
-    "movdqa    %1,%%xmm5                       \n"
-  :
-  : "m"(kShuffleColARGB),  // %0
-    "m"(kShuffleFractions)  // %1
-  );
-
-  asm volatile (
-    "movd      %5,%%xmm2                       \n"
-    "movd      %6,%%xmm3                       \n"
-    "pcmpeqb   %%xmm6,%%xmm6                   \n"
-    "psrlw     $0x9,%%xmm6                     \n"
-    "pextrw    $0x1,%%xmm2,%k3                 \n"
-    "sub       $0x2,%2                         \n"
-    "jl        29f                             \n"
-    "movdqa    %%xmm2,%%xmm0                   \n"
-    "paddd     %%xmm3,%%xmm0                   \n"
-    "punpckldq %%xmm0,%%xmm2                   \n"
-    "punpckldq %%xmm3,%%xmm3                   \n"
-    "paddd     %%xmm3,%%xmm3                   \n"
-    "pextrw    $0x3,%%xmm2,%k4                 \n"
-
-    LABELALIGN
-  "2:                                          \n"
-    "movdqa    %%xmm2,%%xmm1                   \n"
-    "paddd     %%xmm3,%%xmm2                   \n"
-    MEMOPREG(movq,0x00,1,3,4,xmm0)             //  movq      (%1,%3,4),%%xmm0
-    "psrlw     $0x9,%%xmm1                     \n"
-    BUNDLEALIGN
-    MEMOPREG(movhps,0x00,1,4,4,xmm0)           //  movhps    (%1,%4,4),%%xmm0
-    "pshufb    %%xmm5,%%xmm1                   \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "pxor      %%xmm6,%%xmm1                   \n"
-    "pmaddubsw %%xmm1,%%xmm0                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "pextrw    $0x1,%%xmm2,%k3                 \n"
-    "pextrw    $0x3,%%xmm2,%k4                 \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movq      %%xmm0," MEMACCESS(0) "         \n"
-    "lea       " MEMLEA(0x8,0) ",%0            \n"
-    "sub       $0x2,%2                         \n"
-    "jge       2b                              \n"
-
-    LABELALIGN
-  "29:                                         \n"
-    "add       $0x1,%2                         \n"
-    "jl        99f                             \n"
-    "psrlw     $0x9,%%xmm2                     \n"
-    BUNDLEALIGN
-    MEMOPREG(movq,0x00,1,3,4,xmm0)             //  movq      (%1,%3,4),%%xmm0
-    "pshufb    %%xmm5,%%xmm2                   \n"
-    "pshufb    %%xmm4,%%xmm0                   \n"
-    "pxor      %%xmm6,%%xmm2                   \n"
-    "pmaddubsw %%xmm2,%%xmm0                   \n"
-    "psrlw     $0x7,%%xmm0                     \n"
-    "packuswb  %%xmm0,%%xmm0                   \n"
-    "movd      %%xmm0," MEMACCESS(0) "         \n"
-
-    LABELALIGN
-  "99:                                         \n"
-  : "+r"(dst_argb),    // %0
-    "+r"(src_argb),    // %1
-    "+rm"(dst_width),  // %2
-    "+r"(x0),          // %3
-    "+r"(x1)           // %4
-  : "rm"(x),           // %5
-    "rm"(dx)           // %6
-  : "memory", "cc"
-#if defined(__native_client__) && defined(__x86_64__)
-    , "r14"
-#endif
-#if defined(__SSE2__)
-    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
-#endif
-  );
-}
-
-// Divide num by div and return as 16.16 fixed point result.
-int FixedDiv_X86(int num, int div) {
-  asm volatile (
-    "cdq                                       \n"
-    "shld      $0x10,%%eax,%%edx               \n"
-    "shl       $0x10,%%eax                     \n"
-    "idiv      %1                              \n"
-    "mov       %0, %%eax                       \n"
-    : "+a"(num)  // %0
-    : "c"(div)   // %1
-    : "memory", "cc", "edx"
-  );
-  return num;
-}
-
-// Divide num - 1 by div - 1 and return as 16.16 fixed point result.
-int FixedDiv1_X86(int num, int div) {
-  asm volatile (
-    "cdq                                       \n"
-    "shld      $0x10,%%eax,%%edx               \n"
-    "shl       $0x10,%%eax                     \n"
-    "sub       $0x10001,%%eax                  \n"
-    "sbb       $0x0,%%edx                      \n"
-    "sub       $0x1,%1                         \n"
-    "idiv      %1                              \n"
-    "mov       %0, %%eax                       \n"
-    : "+a"(num)  // %0
-    : "c"(div)   // %1
-    : "memory", "cc", "edx"
-  );
-  return num;
-}
-
-#endif  // defined(__x86_64__) || defined(__i386__)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/scale_win.cc b/drivers/theoraplayer/src/YUV/libyuv/src/scale_win.cc
deleted file mode 100644
index 840b9738da..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/scale_win.cc
+++ /dev/null
@@ -1,1320 +0,0 @@
-/*
- *  Copyright 2013 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// This module is for Visual C x86.
-#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-// Offsets for source bytes 0 to 9
-static uvec8 kShuf0 =
-  { 0, 1, 3, 4, 5, 7, 8, 9, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 11 to 20 with 8 subtracted = 3 to 12.
-static uvec8 kShuf1 =
-  { 3, 4, 5, 7, 8, 9, 11, 12, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
-static uvec8 kShuf2 =
-  { 5, 7, 8, 9, 11, 12, 13, 15, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Offsets for source bytes 0 to 10
-static uvec8 kShuf01 =
-  { 0, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10 };
-
-// Offsets for source bytes 10 to 21 with 8 subtracted = 3 to 13.
-static uvec8 kShuf11 =
-  { 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13 };
-
-// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
-static uvec8 kShuf21 =
-  { 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 13, 14, 14, 15 };
-
-// Coefficients for source bytes 0 to 10
-static uvec8 kMadd01 =
-  { 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2 };
-
-// Coefficients for source bytes 10 to 21
-static uvec8 kMadd11 =
-  { 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1 };
-
-// Coefficients for source bytes 21 to 31
-static uvec8 kMadd21 =
-  { 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3 };
-
-// Coefficients for source bytes 21 to 31
-static vec16 kRound34 =
-  { 2, 2, 2, 2, 2, 2, 2, 2 };
-
-static uvec8 kShuf38a =
-  { 0, 3, 6, 8, 11, 14, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-static uvec8 kShuf38b =
-  { 128, 128, 128, 128, 128, 128, 0, 3, 6, 8, 11, 14, 128, 128, 128, 128 };
-
-// Arrange words 0,3,6 into 0,1,2
-static uvec8 kShufAc =
-  { 0, 1, 6, 7, 12, 13, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
-
-// Arrange words 0,3,6 into 3,4,5
-static uvec8 kShufAc3 =
-  { 128, 128, 128, 128, 128, 128, 0, 1, 6, 7, 12, 13, 128, 128, 128, 128 };
-
-// Scaling values for boxes of 3x3 and 2x3
-static uvec16 kScaleAc33 =
-  { 65536 / 9, 65536 / 9, 65536 / 6, 65536 / 9, 65536 / 9, 65536 / 6, 0, 0 };
-
-// Arrange first value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb0 =
-  { 0, 128, 3, 128, 6, 128, 8, 128, 11, 128, 14, 128, 128, 128, 128, 128 };
-
-// Arrange second value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb1 =
-  { 1, 128, 4, 128, 7, 128, 9, 128, 12, 128, 15, 128, 128, 128, 128, 128 };
-
-// Arrange third value for pixels 0,1,2,3,4,5
-static uvec8 kShufAb2 =
-  { 2, 128, 5, 128, 128, 128, 10, 128, 13, 128, 128, 128, 128, 128, 128, 128 };
-
-// Scaling values for boxes of 3x2 and 2x2
-static uvec16 kScaleAb2 =
-  { 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
-
-// Reads 32 pixels, throws half away and writes 16 pixels.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8               // isolate odd pixels.
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x1 rectangle to 16x1.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                              uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-
-    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
-    psrlw      xmm0, 8
-    movdqa     xmm3, xmm1
-    psrlw      xmm1, 8
-    pand       xmm2, xmm5
-    pand       xmm3, xmm5
-    pavgw      xmm0, xmm2
-    pavgw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x2 rectangle to 16x1.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + esi]
-    movdqa     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2            // average rows
-    pavgb      xmm1, xmm3
-
-    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
-    psrlw      xmm0, 8
-    movdqa     xmm3, xmm1
-    psrlw      xmm1, 8
-    pand       xmm2, xmm5
-    pand       xmm3, xmm5
-    pavgw      xmm0, xmm2
-    pavgw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-
-    sub        ecx, 16
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Reads 32 pixels, throws half away and writes 16 pixels.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2_Unaligned_SSE2(const uint8* src_ptr,
-                                  ptrdiff_t src_stride,
-                                  uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-
-    align      4
-  wloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    psrlw      xmm0, 8               // isolate odd pixels.
-    psrlw      xmm1, 8
-    packuswb   xmm0, xmm1
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x1 rectangle to 16x1.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2Linear_Unaligned_SSE2(const uint8* src_ptr,
-                                        ptrdiff_t src_stride,
-                                        uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  wloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-
-    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
-    psrlw      xmm0, 8
-    movdqa     xmm3, xmm1
-    psrlw      xmm1, 8
-    pand       xmm2, xmm5
-    pand       xmm3, xmm5
-    pavgw      xmm0, xmm2
-    pavgw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x2 rectangle to 16x1.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown2Box_Unaligned_SSE2(const uint8* src_ptr,
-                                     ptrdiff_t src_stride,
-                                     uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
-    psrlw      xmm5, 8
-
-    align      4
-  wloop:
-    movdqu     xmm0, [eax]
-    movdqu     xmm1, [eax + 16]
-    movdqu     xmm2, [eax + esi]
-    movdqu     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2            // average rows
-    pavgb      xmm1, xmm3
-
-    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
-    psrlw      xmm0, 8
-    movdqa     xmm3, xmm1
-    psrlw      xmm1, 8
-    pand       xmm2, xmm5
-    pand       xmm3, xmm5
-    pavgw      xmm0, xmm2
-    pavgw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-
-    sub        ecx, 16
-    movdqu     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Point samples 32 pixels to 8 pixels.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                        uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff0000
-    psrld      xmm5, 24
-    pslld      xmm5, 16
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    pand       xmm0, xmm5
-    pand       xmm1, xmm5
-    packuswb   xmm0, xmm1
-    psrlw      xmm0, 8
-    packuswb   xmm0, xmm0
-    sub        ecx, 8
-    movq       qword ptr [edx], xmm0
-    lea        edx, [edx + 8]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x4 rectangle to 8x1.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                           uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_ptr
-    mov        esi, [esp + 8 + 8]    // src_stride
-    mov        edx, [esp + 8 + 12]   // dst_ptr
-    mov        ecx, [esp + 8 + 16]   // dst_width
-    lea        edi, [esi + esi * 2]  // src_stride * 3
-    pcmpeqb    xmm7, xmm7            // generate mask 0x00ff00ff
-    psrlw      xmm7, 8
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + esi]
-    movdqa     xmm3, [eax + esi + 16]
-    pavgb      xmm0, xmm2            // average rows
-    pavgb      xmm1, xmm3
-    movdqa     xmm2, [eax + esi * 2]
-    movdqa     xmm3, [eax + esi * 2 + 16]
-    movdqa     xmm4, [eax + edi]
-    movdqa     xmm5, [eax + edi + 16]
-    lea        eax, [eax + 32]
-    pavgb      xmm2, xmm4
-    pavgb      xmm3, xmm5
-    pavgb      xmm0, xmm2
-    pavgb      xmm1, xmm3
-
-    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
-    psrlw      xmm0, 8
-    movdqa     xmm3, xmm1
-    psrlw      xmm1, 8
-    pand       xmm2, xmm7
-    pand       xmm3, xmm7
-    pavgw      xmm0, xmm2
-    pavgw      xmm1, xmm3
-    packuswb   xmm0, xmm1
-
-    movdqa     xmm2, xmm0            // average columns (16 to 8 pixels)
-    psrlw      xmm0, 8
-    pand       xmm2, xmm7
-    pavgw      xmm0, xmm2
-    packuswb   xmm0, xmm0
-
-    sub        ecx, 8
-    movq       qword ptr [edx], xmm0
-    lea        edx, [edx + 8]
-    jg         wloop
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// Point samples 32 pixels to 24 pixels.
-// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
-// Then shuffled to do the scaling.
-
-// Note that movdqa+palign may be better than movdqu.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-    movdqa     xmm3, kShuf0
-    movdqa     xmm4, kShuf1
-    movdqa     xmm5, kShuf2
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    movdqa     xmm2, xmm1
-    palignr    xmm1, xmm0, 8
-    pshufb     xmm0, xmm3
-    pshufb     xmm1, xmm4
-    pshufb     xmm2, xmm5
-    movq       qword ptr [edx], xmm0
-    movq       qword ptr [edx + 8], xmm1
-    movq       qword ptr [edx + 16], xmm2
-    lea        edx, [edx + 24]
-    sub        ecx, 24
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 32x2 rectangle to 24x1
-// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
-// Then shuffled to do the scaling.
-
-// Register usage:
-// xmm0 src_row 0
-// xmm1 src_row 1
-// xmm2 shuf 0
-// xmm3 shuf 1
-// xmm4 shuf 2
-// xmm5 madd 0
-// xmm6 madd 1
-// xmm7 kRound34
-
-// Note that movdqa+palign may be better than movdqu.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    movdqa     xmm2, kShuf01
-    movdqa     xmm3, kShuf11
-    movdqa     xmm4, kShuf21
-    movdqa     xmm5, kMadd01
-    movdqa     xmm6, kMadd11
-    movdqa     xmm7, kRound34
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]           // pixels 0..7
-    movdqa     xmm1, [eax + esi]
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm2
-    pmaddubsw  xmm0, xmm5
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    movq       qword ptr [edx], xmm0
-    movdqu     xmm0, [eax + 8]       // pixels 8..15
-    movdqu     xmm1, [eax + esi + 8]
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm3
-    pmaddubsw  xmm0, xmm6
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    movq       qword ptr [edx + 8], xmm0
-    movdqa     xmm0, [eax + 16]      // pixels 16..23
-    movdqa     xmm1, [eax + esi + 16]
-    lea        eax, [eax + 32]
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm4
-    movdqa     xmm1, kMadd21
-    pmaddubsw  xmm0, xmm1
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    sub        ecx, 24
-    movq       qword ptr [edx + 16], xmm0
-    lea        edx, [edx + 24]
-    jg         wloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Note that movdqa+palign may be better than movdqu.
-// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    movdqa     xmm2, kShuf01
-    movdqa     xmm3, kShuf11
-    movdqa     xmm4, kShuf21
-    movdqa     xmm5, kMadd01
-    movdqa     xmm6, kMadd11
-    movdqa     xmm7, kRound34
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]           // pixels 0..7
-    movdqa     xmm1, [eax + esi]
-    pavgb      xmm1, xmm0
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm2
-    pmaddubsw  xmm0, xmm5
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    movq       qword ptr [edx], xmm0
-    movdqu     xmm0, [eax + 8]       // pixels 8..15
-    movdqu     xmm1, [eax + esi + 8]
-    pavgb      xmm1, xmm0
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm3
-    pmaddubsw  xmm0, xmm6
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    movq       qword ptr [edx + 8], xmm0
-    movdqa     xmm0, [eax + 16]      // pixels 16..23
-    movdqa     xmm1, [eax + esi + 16]
-    lea        eax, [eax + 32]
-    pavgb      xmm1, xmm0
-    pavgb      xmm0, xmm1
-    pshufb     xmm0, xmm4
-    movdqa     xmm1, kMadd21
-    pmaddubsw  xmm0, xmm1
-    paddsw     xmm0, xmm7
-    psrlw      xmm0, 2
-    packuswb   xmm0, xmm0
-    sub        ecx, 24
-    movq       qword ptr [edx + 16], xmm0
-    lea        edx, [edx+24]
-    jg         wloop
-
-    pop        esi
-    ret
-  }
-}
-
-// 3/8 point sampler
-
-// Scale 32 pixels to 12
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
-                          uint8* dst_ptr, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_ptr
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_ptr
-    mov        ecx, [esp + 16]       // dst_width
-    movdqa     xmm4, kShuf38a
-    movdqa     xmm5, kShuf38b
-
-    align      4
-  xloop:
-    movdqa     xmm0, [eax]           // 16 pixels -> 0,1,2,3,4,5
-    movdqa     xmm1, [eax + 16]      // 16 pixels -> 6,7,8,9,10,11
-    lea        eax, [eax + 32]
-    pshufb     xmm0, xmm4
-    pshufb     xmm1, xmm5
-    paddusb    xmm0, xmm1
-
-    sub        ecx, 12
-    movq       qword ptr [edx], xmm0  // write 12 pixels
-    movhlps    xmm1, xmm0
-    movd       [edx + 8], xmm1
-    lea        edx, [edx + 12]
-    jg         xloop
-
-    ret
-  }
-}
-
-// Scale 16x3 pixels to 6x1 with interpolation
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    movdqa     xmm2, kShufAc
-    movdqa     xmm3, kShufAc3
-    movdqa     xmm4, kScaleAc33
-    pxor       xmm5, xmm5
-
-    align      4
-  xloop:
-    movdqa     xmm0, [eax]           // sum up 3 rows into xmm0/1
-    movdqa     xmm6, [eax + esi]
-    movhlps    xmm1, xmm0
-    movhlps    xmm7, xmm6
-    punpcklbw  xmm0, xmm5
-    punpcklbw  xmm1, xmm5
-    punpcklbw  xmm6, xmm5
-    punpcklbw  xmm7, xmm5
-    paddusw    xmm0, xmm6
-    paddusw    xmm1, xmm7
-    movdqa     xmm6, [eax + esi * 2]
-    lea        eax, [eax + 16]
-    movhlps    xmm7, xmm6
-    punpcklbw  xmm6, xmm5
-    punpcklbw  xmm7, xmm5
-    paddusw    xmm0, xmm6
-    paddusw    xmm1, xmm7
-
-    movdqa     xmm6, xmm0            // 8 pixels -> 0,1,2 of xmm6
-    psrldq     xmm0, 2
-    paddusw    xmm6, xmm0
-    psrldq     xmm0, 2
-    paddusw    xmm6, xmm0
-    pshufb     xmm6, xmm2
-
-    movdqa     xmm7, xmm1            // 8 pixels -> 3,4,5 of xmm6
-    psrldq     xmm1, 2
-    paddusw    xmm7, xmm1
-    psrldq     xmm1, 2
-    paddusw    xmm7, xmm1
-    pshufb     xmm7, xmm3
-    paddusw    xmm6, xmm7
-
-    pmulhuw    xmm6, xmm4            // divide by 9,9,6, 9,9,6
-    packuswb   xmm6, xmm6
-
-    sub        ecx, 6
-    movd       [edx], xmm6           // write 6 pixels
-    psrlq      xmm6, 16
-    movd       [edx + 2], xmm6
-    lea        edx, [edx + 6]
-    jg         xloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Scale 16x2 pixels to 6x1 with interpolation
-__declspec(naked) __declspec(align(16))
-void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr,
-                                ptrdiff_t src_stride,
-                                uint8* dst_ptr, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_ptr
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_ptr
-    mov        ecx, [esp + 4 + 16]   // dst_width
-    movdqa     xmm2, kShufAb0
-    movdqa     xmm3, kShufAb1
-    movdqa     xmm4, kShufAb2
-    movdqa     xmm5, kScaleAb2
-
-    align      4
-  xloop:
-    movdqa     xmm0, [eax]           // average 2 rows into xmm0
-    pavgb      xmm0, [eax + esi]
-    lea        eax, [eax + 16]
-
-    movdqa     xmm1, xmm0            // 16 pixels -> 0,1,2,3,4,5 of xmm1
-    pshufb     xmm1, xmm2
-    movdqa     xmm6, xmm0
-    pshufb     xmm6, xmm3
-    paddusw    xmm1, xmm6
-    pshufb     xmm0, xmm4
-    paddusw    xmm1, xmm0
-
-    pmulhuw    xmm1, xmm5            // divide by 3,3,2, 3,3,2
-    packuswb   xmm1, xmm1
-
-    sub        ecx, 6
-    movd       [edx], xmm1           // write 6 pixels
-    psrlq      xmm1, 16
-    movd       [edx + 2], xmm1
-    lea        edx, [edx + 6]
-    jg         xloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Reads 16xN bytes and produces 16 shorts at a time.
-// TODO(fbarchard): Make this handle 4xN bytes for any width ARGB.
-__declspec(naked) __declspec(align(16))
-void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
-                       uint16* dst_ptr, int src_width,
-                       int src_height) {
-  __asm {
-    push       esi
-    push       edi
-    push       ebx
-    push       ebp
-    mov        esi, [esp + 16 + 4]   // src_ptr
-    mov        edx, [esp + 16 + 8]   // src_stride
-    mov        edi, [esp + 16 + 12]  // dst_ptr
-    mov        ecx, [esp + 16 + 16]  // dst_width
-    mov        ebx, [esp + 16 + 20]  // height
-    pxor       xmm4, xmm4
-    dec        ebx
-
-    align      4
-  xloop:
-    // first row
-    movdqa     xmm0, [esi]
-    lea        eax, [esi + edx]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm4
-    punpckhbw  xmm1, xmm4
-    lea        esi, [esi + 16]
-    mov        ebp, ebx
-    test       ebp, ebp
-    je         ydone
-
-    // sum remaining rows
-    align      4
-  yloop:
-    movdqa     xmm2, [eax]       // read 16 pixels
-    lea        eax, [eax + edx]  // advance to next row
-    movdqa     xmm3, xmm2
-    punpcklbw  xmm2, xmm4
-    punpckhbw  xmm3, xmm4
-    paddusw    xmm0, xmm2        // sum 16 words
-    paddusw    xmm1, xmm3
-    sub        ebp, 1
-    jg         yloop
-
-    align      4
-  ydone:
-    movdqa     [edi], xmm0
-    movdqa     [edi + 16], xmm1
-    lea        edi, [edi + 32]
-
-    sub        ecx, 16
-    jg         xloop
-
-    pop        ebp
-    pop        ebx
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// Bilinear column filtering. SSSE3 version.
-// TODO(fbarchard): Port to Neon
-// TODO(fbarchard): Switch the following:
-//    xor        ebx, ebx
-//    mov        bx, word ptr [esi + eax]  // 2 source x0 pixels
-// To
-//    movzx      ebx, word ptr [esi + eax]  // 2 source x0 pixels
-// when drmemory bug fixed.
-// https://code.google.com/p/drmemory/issues/detail?id=1396
-
-__declspec(naked) __declspec(align(16))
-void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
-                           int dst_width, int x, int dx) {
-  __asm {
-    push       ebx
-    push       esi
-    push       edi
-    mov        edi, [esp + 12 + 4]    // dst_ptr
-    mov        esi, [esp + 12 + 8]    // src_ptr
-    mov        ecx, [esp + 12 + 12]   // dst_width
-    movd       xmm2, [esp + 12 + 16]  // x
-    movd       xmm3, [esp + 12 + 20]  // dx
-    mov        eax, 0x04040000      // shuffle to line up fractions with pixel.
-    movd       xmm5, eax
-    pcmpeqb    xmm6, xmm6           // generate 0x007f for inverting fraction.
-    psrlw      xmm6, 9
-    pextrw     eax, xmm2, 1         // get x0 integer. preroll
-    sub        ecx, 2
-    jl         xloop29
-
-    movdqa     xmm0, xmm2           // x1 = x0 + dx
-    paddd      xmm0, xmm3
-    punpckldq  xmm2, xmm0           // x0 x1
-    punpckldq  xmm3, xmm3           // dx dx
-    paddd      xmm3, xmm3           // dx * 2, dx * 2
-    pextrw     edx, xmm2, 3         // get x1 integer. preroll
-
-    // 2 Pixel loop.
-    align      4
-  xloop2:
-    movdqa     xmm1, xmm2           // x0, x1 fractions.
-    paddd      xmm2, xmm3           // x += dx
-    movzx      ebx, word ptr [esi + eax]  // 2 source x0 pixels
-    movd       xmm0, ebx
-    psrlw      xmm1, 9              // 7 bit fractions.
-    movzx      ebx, word ptr [esi + edx]  // 2 source x1 pixels
-    movd       xmm4, ebx
-    pshufb     xmm1, xmm5           // 0011
-    punpcklwd  xmm0, xmm4
-    pxor       xmm1, xmm6           // 0..7f and 7f..0
-    pmaddubsw  xmm0, xmm1           // 16 bit, 2 pixels.
-    pextrw     eax, xmm2, 1         // get x0 integer. next iteration.
-    pextrw     edx, xmm2, 3         // get x1 integer. next iteration.
-    psrlw      xmm0, 7              // 8.7 fixed point to low 8 bits.
-    packuswb   xmm0, xmm0           // 8 bits, 2 pixels.
-    movd       ebx, xmm0
-    mov        [edi], bx
-    lea        edi, [edi + 2]
-    sub        ecx, 2               // 2 pixels
-    jge        xloop2
-
-    align      4
- xloop29:
-
-    add        ecx, 2 - 1
-    jl         xloop99
-
-    // 1 pixel remainder
-    movzx      ebx, word ptr [esi + eax]  // 2 source x0 pixels
-    movd       xmm0, ebx
-    psrlw      xmm2, 9              // 7 bit fractions.
-    pshufb     xmm2, xmm5           // 0011
-    pxor       xmm2, xmm6           // 0..7f and 7f..0
-    pmaddubsw  xmm0, xmm2           // 16 bit
-    psrlw      xmm0, 7              // 8.7 fixed point to low 8 bits.
-    packuswb   xmm0, xmm0           // 8 bits
-    movd       ebx, xmm0
-    mov        [edi], bl
-
-    align      4
- xloop99:
-
-    pop        edi
-    pop        esi
-    pop        ebx
-    ret
-  }
-}
-
-// Reads 16 pixels, duplicates them and writes 32 pixels.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr,
-                       int dst_width, int x, int dx) {
-  __asm {
-    mov        edx, [esp + 4]    // dst_ptr
-    mov        eax, [esp + 8]    // src_ptr
-    mov        ecx, [esp + 12]   // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    lea        eax,  [eax + 16]
-    movdqa     xmm1, xmm0
-    punpcklbw  xmm0, xmm0
-    punpckhbw  xmm1, xmm1
-    sub        ecx, 32
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Reads 8 pixels, throws half away and writes 4 even pixels (0, 2, 4, 6)
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBRowDown2_SSE2(const uint8* src_argb,
-                            ptrdiff_t src_stride,
-                            uint8* dst_argb, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_argb
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_argb
-    mov        ecx, [esp + 16]       // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    shufps     xmm0, xmm1, 0xdd
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 8x1 rectangle to 4x1.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride,
-                                  uint8* dst_argb, int dst_width) {
-  __asm {
-    mov        eax, [esp + 4]        // src_argb
-                                     // src_stride ignored
-    mov        edx, [esp + 12]       // dst_argb
-    mov        ecx, [esp + 16]       // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    lea        eax,  [eax + 32]
-    movdqa     xmm2, xmm0
-    shufps     xmm0, xmm1, 0x88      // even pixels
-    shufps     xmm2, xmm1, 0xdd      // odd pixels
-    pavgb      xmm0, xmm2
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Blends 8x2 rectangle to 4x1.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb,
-                               ptrdiff_t src_stride,
-                               uint8* dst_argb, int dst_width) {
-  __asm {
-    push       esi
-    mov        eax, [esp + 4 + 4]    // src_argb
-    mov        esi, [esp + 4 + 8]    // src_stride
-    mov        edx, [esp + 4 + 12]   // dst_argb
-    mov        ecx, [esp + 4 + 16]   // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    movdqa     xmm1, [eax + 16]
-    movdqa     xmm2, [eax + esi]
-    movdqa     xmm3, [eax + esi + 16]
-    lea        eax,  [eax + 32]
-    pavgb      xmm0, xmm2            // average rows
-    pavgb      xmm1, xmm3
-    movdqa     xmm2, xmm0            // average columns (8 to 4 pixels)
-    shufps     xmm0, xmm1, 0x88      // even pixels
-    shufps     xmm2, xmm1, 0xdd      // odd pixels
-    pavgb      xmm0, xmm2
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    pop        esi
-    ret
-  }
-}
-
-// Reads 4 pixels at a time.
-// Alignment requirement: dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride,
-                               int src_stepx,
-                               uint8* dst_argb, int dst_width) {
-  __asm {
-    push       ebx
-    push       edi
-    mov        eax, [esp + 8 + 4]    // src_argb
-                                     // src_stride ignored
-    mov        ebx, [esp + 8 + 12]   // src_stepx
-    mov        edx, [esp + 8 + 16]   // dst_argb
-    mov        ecx, [esp + 8 + 20]   // dst_width
-    lea        ebx, [ebx * 4]
-    lea        edi, [ebx + ebx * 2]
-
-    align      4
-  wloop:
-    movd       xmm0, [eax]
-    movd       xmm1, [eax + ebx]
-    punpckldq  xmm0, xmm1
-    movd       xmm2, [eax + ebx * 2]
-    movd       xmm3, [eax + edi]
-    lea        eax,  [eax + ebx * 4]
-    punpckldq  xmm2, xmm3
-    punpcklqdq xmm0, xmm2
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    pop        edi
-    pop        ebx
-    ret
-  }
-}
-
-// Blends four 2x2 to 4x1.
-// Alignment requirement: dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb,
-                                  ptrdiff_t src_stride,
-                                  int src_stepx,
-                                  uint8* dst_argb, int dst_width) {
-  __asm {
-    push       ebx
-    push       esi
-    push       edi
-    mov        eax, [esp + 12 + 4]    // src_argb
-    mov        esi, [esp + 12 + 8]    // src_stride
-    mov        ebx, [esp + 12 + 12]   // src_stepx
-    mov        edx, [esp + 12 + 16]   // dst_argb
-    mov        ecx, [esp + 12 + 20]   // dst_width
-    lea        esi, [eax + esi]       // row1 pointer
-    lea        ebx, [ebx * 4]
-    lea        edi, [ebx + ebx * 2]
-
-    align      4
-  wloop:
-    movq       xmm0, qword ptr [eax]  // row0 4 pairs
-    movhps     xmm0, qword ptr [eax + ebx]
-    movq       xmm1, qword ptr [eax + ebx * 2]
-    movhps     xmm1, qword ptr [eax + edi]
-    lea        eax,  [eax + ebx * 4]
-    movq       xmm2, qword ptr [esi]  // row1 4 pairs
-    movhps     xmm2, qword ptr [esi + ebx]
-    movq       xmm3, qword ptr [esi + ebx * 2]
-    movhps     xmm3, qword ptr [esi + edi]
-    lea        esi,  [esi + ebx * 4]
-    pavgb      xmm0, xmm2            // average rows
-    pavgb      xmm1, xmm3
-    movdqa     xmm2, xmm0            // average columns (8 to 4 pixels)
-    shufps     xmm0, xmm1, 0x88      // even pixels
-    shufps     xmm2, xmm1, 0xdd      // odd pixels
-    pavgb      xmm0, xmm2
-    sub        ecx, 4
-    movdqa     [edx], xmm0
-    lea        edx, [edx + 16]
-    jg         wloop
-
-    pop        edi
-    pop        esi
-    pop        ebx
-    ret
-  }
-}
-
-// Column scaling unfiltered. SSE2 version.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBCols_SSE2(uint8* dst_argb, const uint8* src_argb,
-                        int dst_width, int x, int dx) {
-  __asm {
-    push       edi
-    push       esi
-    mov        edi, [esp + 8 + 4]    // dst_argb
-    mov        esi, [esp + 8 + 8]    // src_argb
-    mov        ecx, [esp + 8 + 12]   // dst_width
-    movd       xmm2, [esp + 8 + 16]  // x
-    movd       xmm3, [esp + 8 + 20]  // dx
-
-    pshufd     xmm2, xmm2, 0         // x0 x0 x0 x0
-    pshufd     xmm0, xmm3, 0x11      // dx  0 dx  0
-    paddd      xmm2, xmm0
-    paddd      xmm3, xmm3            // 0, 0, 0,  dx * 2
-    pshufd     xmm0, xmm3, 0x05      // dx * 2, dx * 2, 0, 0
-    paddd      xmm2, xmm0            // x3 x2 x1 x0
-    paddd      xmm3, xmm3            // 0, 0, 0,  dx * 4
-    pshufd     xmm3, xmm3, 0         // dx * 4, dx * 4, dx * 4, dx * 4
-
-    pextrw     eax, xmm2, 1          // get x0 integer.
-    pextrw     edx, xmm2, 3          // get x1 integer.
-
-    cmp        ecx, 0
-    jle        xloop99
-    sub        ecx, 4
-    jl         xloop49
-
-    // 4 Pixel loop.
-    align      4
- xloop4:
-    movd       xmm0, [esi + eax * 4]  // 1 source x0 pixels
-    movd       xmm1, [esi + edx * 4]  // 1 source x1 pixels
-    pextrw     eax, xmm2, 5           // get x2 integer.
-    pextrw     edx, xmm2, 7           // get x3 integer.
-    paddd      xmm2, xmm3             // x += dx
-    punpckldq  xmm0, xmm1             // x0 x1
-
-    movd       xmm1, [esi + eax * 4]  // 1 source x2 pixels
-    movd       xmm4, [esi + edx * 4]  // 1 source x3 pixels
-    pextrw     eax, xmm2, 1           // get x0 integer. next iteration.
-    pextrw     edx, xmm2, 3           // get x1 integer. next iteration.
-    punpckldq  xmm1, xmm4             // x2 x3
-    punpcklqdq xmm0, xmm1             // x0 x1 x2 x3
-    sub        ecx, 4                 // 4 pixels
-    movdqu     [edi], xmm0
-    lea        edi, [edi + 16]
-    jge        xloop4
-
-    align      4
- xloop49:
-    test       ecx, 2
-    je         xloop29
-
-    // 2 Pixels.
-    movd       xmm0, [esi + eax * 4]  // 1 source x0 pixels
-    movd       xmm1, [esi + edx * 4]  // 1 source x1 pixels
-    pextrw     eax, xmm2, 5           // get x2 integer.
-    punpckldq  xmm0, xmm1             // x0 x1
-
-    movq       qword ptr [edi], xmm0
-    lea        edi, [edi + 8]
-
- xloop29:
-    test       ecx, 1
-    je         xloop99
-
-    // 1 Pixels.
-    movd       xmm0, [esi + eax * 4]  // 1 source x2 pixels
-    movd       dword ptr [edi], xmm0
-    align      4
- xloop99:
-
-    pop        esi
-    pop        edi
-    ret
-  }
-}
-
-// Bilinear row filtering combines 2x1 -> 1x1. SSSE3 version.
-// TODO(fbarchard): Port to Neon
-
-// Shuffle table for arranging 2 pixels into pairs for pmaddubsw
-static uvec8 kShuffleColARGB = {
-  0u, 4u, 1u, 5u, 2u, 6u, 3u, 7u,  // bbggrraa 1st pixel
-  8u, 12u, 9u, 13u, 10u, 14u, 11u, 15u  // bbggrraa 2nd pixel
-};
-
-// Shuffle table for duplicating 2 fractions into 8 bytes each
-static uvec8 kShuffleFractions = {
-  0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 4u, 4u, 4u, 4u, 4u, 4u, 4u, 4u,
-};
-
-__declspec(naked) __declspec(align(16))
-void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb,
-                               int dst_width, int x, int dx) {
-  __asm {
-    push       esi
-    push       edi
-    mov        edi, [esp + 8 + 4]    // dst_argb
-    mov        esi, [esp + 8 + 8]    // src_argb
-    mov        ecx, [esp + 8 + 12]   // dst_width
-    movd       xmm2, [esp + 8 + 16]  // x
-    movd       xmm3, [esp + 8 + 20]  // dx
-    movdqa     xmm4, kShuffleColARGB
-    movdqa     xmm5, kShuffleFractions
-    pcmpeqb    xmm6, xmm6           // generate 0x007f for inverting fraction.
-    psrlw      xmm6, 9
-    pextrw     eax, xmm2, 1         // get x0 integer. preroll
-    sub        ecx, 2
-    jl         xloop29
-
-    movdqa     xmm0, xmm2           // x1 = x0 + dx
-    paddd      xmm0, xmm3
-    punpckldq  xmm2, xmm0           // x0 x1
-    punpckldq  xmm3, xmm3           // dx dx
-    paddd      xmm3, xmm3           // dx * 2, dx * 2
-    pextrw     edx, xmm2, 3         // get x1 integer. preroll
-
-    // 2 Pixel loop.
-    align      4
-  xloop2:
-    movdqa     xmm1, xmm2           // x0, x1 fractions.
-    paddd      xmm2, xmm3           // x += dx
-    movq       xmm0, qword ptr [esi + eax * 4]  // 2 source x0 pixels
-    psrlw      xmm1, 9              // 7 bit fractions.
-    movhps     xmm0, qword ptr [esi + edx * 4]  // 2 source x1 pixels
-    pshufb     xmm1, xmm5           // 0000000011111111
-    pshufb     xmm0, xmm4           // arrange pixels into pairs
-    pxor       xmm1, xmm6           // 0..7f and 7f..0
-    pmaddubsw  xmm0, xmm1           // argb_argb 16 bit, 2 pixels.
-    pextrw     eax, xmm2, 1         // get x0 integer. next iteration.
-    pextrw     edx, xmm2, 3         // get x1 integer. next iteration.
-    psrlw      xmm0, 7              // argb 8.7 fixed point to low 8 bits.
-    packuswb   xmm0, xmm0           // argb_argb 8 bits, 2 pixels.
-    movq       qword ptr [edi], xmm0
-    lea        edi, [edi + 8]
-    sub        ecx, 2               // 2 pixels
-    jge        xloop2
-
-    align      4
- xloop29:
-
-    add        ecx, 2 - 1
-    jl         xloop99
-
-    // 1 pixel remainder
-    psrlw      xmm2, 9              // 7 bit fractions.
-    movq       xmm0, qword ptr [esi + eax * 4]  // 2 source x0 pixels
-    pshufb     xmm2, xmm5           // 00000000
-    pshufb     xmm0, xmm4           // arrange pixels into pairs
-    pxor       xmm2, xmm6           // 0..7f and 7f..0
-    pmaddubsw  xmm0, xmm2           // argb 16 bit, 1 pixel.
-    psrlw      xmm0, 7
-    packuswb   xmm0, xmm0           // argb 8 bits, 1 pixel.
-    movd       [edi], xmm0
-
-    align      4
- xloop99:
-
-    pop        edi
-    pop        esi
-    ret
-  }
-}
-
-// Reads 4 pixels, duplicates them and writes 8 pixels.
-// Alignment requirement: src_argb 16 byte aligned, dst_argb 16 byte aligned.
-__declspec(naked) __declspec(align(16))
-void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb,
-                           int dst_width, int x, int dx) {
-  __asm {
-    mov        edx, [esp + 4]    // dst_argb
-    mov        eax, [esp + 8]    // src_argb
-    mov        ecx, [esp + 12]   // dst_width
-
-    align      4
-  wloop:
-    movdqa     xmm0, [eax]
-    lea        eax,  [eax + 16]
-    movdqa     xmm1, xmm0
-    punpckldq  xmm0, xmm0
-    punpckhdq  xmm1, xmm1
-    sub        ecx, 8
-    movdqa     [edx], xmm0
-    movdqa     [edx + 16], xmm1
-    lea        edx, [edx + 32]
-    jg         wloop
-
-    ret
-  }
-}
-
-// Divide num by div and return as 16.16 fixed point result.
-__declspec(naked) __declspec(align(16))
-int FixedDiv_X86(int num, int div) {
-  __asm {
-    mov        eax, [esp + 4]    // num
-    cdq                          // extend num to 64 bits
-    shld       edx, eax, 16      // 32.16
-    shl        eax, 16
-    idiv       dword ptr [esp + 8]
-    ret
-  }
-}
-
-// Divide num by div and return as 16.16 fixed point result.
-__declspec(naked) __declspec(align(16))
-int FixedDiv1_X86(int num, int div) {
-  __asm {
-    mov        eax, [esp + 4]    // num
-    mov        ecx, [esp + 8]    // denom
-    cdq                          // extend num to 64 bits
-    shld       edx, eax, 16      // 32.16
-    shl        eax, 16
-    sub        eax, 0x00010001
-    sbb        edx, 0
-    sub        ecx, 1
-    idiv       ecx
-    ret
-  }
-}
-
-#endif  // !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/video_common.cc b/drivers/theoraplayer/src/YUV/libyuv/src/video_common.cc
deleted file mode 100755
index efbedf46e2..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/video_common.cc
+++ /dev/null
@@ -1,64 +0,0 @@
-/*
- *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS. All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#include "libyuv/video_common.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-#define ARRAY_SIZE(x) (int)(sizeof(x) / sizeof(x[0]))
-
-struct FourCCAliasEntry {
-  uint32 alias;
-  uint32 canonical;
-};
-
-static const struct FourCCAliasEntry kFourCCAliases[] = {
-  {FOURCC_IYUV, FOURCC_I420},
-  {FOURCC_YU16, FOURCC_I422},
-  {FOURCC_YU24, FOURCC_I444},
-  {FOURCC_YUYV, FOURCC_YUY2},
-  {FOURCC_YUVS, FOURCC_YUY2},  // kCMPixelFormat_422YpCbCr8_yuvs
-  {FOURCC_HDYC, FOURCC_UYVY},
-  {FOURCC_2VUY, FOURCC_UYVY},  // kCMPixelFormat_422YpCbCr8
-  {FOURCC_JPEG, FOURCC_MJPG},  // Note: JPEG has DHT while MJPG does not.
-  {FOURCC_DMB1, FOURCC_MJPG},
-  {FOURCC_BA81, FOURCC_BGGR},
-  {FOURCC_RGB3, FOURCC_RAW },
-  {FOURCC_BGR3, FOURCC_24BG},
-  {FOURCC_CM32, FOURCC_BGRA},  // kCMPixelFormat_32ARGB
-  {FOURCC_CM24, FOURCC_RAW },  // kCMPixelFormat_24RGB
-  {FOURCC_L555, FOURCC_RGBO},  // kCMPixelFormat_16LE555
-  {FOURCC_L565, FOURCC_RGBP},  // kCMPixelFormat_16LE565
-  {FOURCC_5551, FOURCC_RGBO},  // kCMPixelFormat_16LE5551
-};
-// TODO(fbarchard): Consider mapping kCMPixelFormat_32BGRA to FOURCC_ARGB.
-//  {FOURCC_BGRA, FOURCC_ARGB},  // kCMPixelFormat_32BGRA
-
-LIBYUV_API
-uint32 CanonicalFourCC(uint32 fourcc) {
-  int i;
-  for (i = 0; i < ARRAY_SIZE(kFourCCAliases); ++i) {
-    if (kFourCCAliases[i].alias == fourcc) {
-      return kFourCCAliases[i].canonical;
-    }
-  }
-  // Not an alias, so return it as-is.
-  return fourcc;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
-
diff --git a/drivers/theoraplayer/src/YUV/libyuv/src/x86inc.asm b/drivers/theoraplayer/src/YUV/libyuv/src/x86inc.asm
deleted file mode 100755
index cb5c32df3a..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/src/x86inc.asm
+++ /dev/null
@@ -1,1136 +0,0 @@
-;*****************************************************************************
-;* x86inc.asm: x264asm abstraction layer
-;*****************************************************************************
-;* Copyright (C) 2005-2012 x264 project
-;*
-;* Authors: Loren Merritt <lorenm@u.washington.edu>
-;*          Anton Mitrofanov <BugMaster@narod.ru>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
-;*          Henrik Gramner <hengar-6@student.ltu.se>
-;*
-;* Permission to use, copy, modify, and/or distribute this software for any
-;* purpose with or without fee is hereby granted, provided that the above
-;* copyright notice and this permission notice appear in all copies.
-;*
-;* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
-;* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
-;* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
-;* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
-;* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
-;* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
-;* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
-;*****************************************************************************
-
-; This is a header file for the x264ASM assembly language, which uses
-; NASM/YASM syntax combined with a large number of macros to provide easy
-; abstraction between different calling conventions (x86_32, win64, linux64).
-; It also has various other useful features to simplify writing the kind of
-; DSP functions that are most often used in x264.
-
-; Unlike the rest of x264, this file is available under an ISC license, as it
-; has significant usefulness outside of x264 and we want it to be available
-; to the largest audience possible.  Of course, if you modify it for your own
-; purposes to add a new feature, we strongly encourage contributing a patch
-; as this feature might be useful for others as well.  Send patches or ideas
-; to x264-devel@videolan.org .
-
-; Local changes for libyuv:
-; remove %define program_name and references in labels
-; rename cpus to uppercase
-
-%define WIN64  0
-%define UNIX64 0
-%if ARCH_X86_64
-    %ifidn __OUTPUT_FORMAT__,win32
-        %define WIN64  1
-    %elifidn __OUTPUT_FORMAT__,win64
-        %define WIN64  1
-    %else
-        %define UNIX64 1
-    %endif
-%endif
-
-%ifdef PREFIX
-    %define mangle(x) _ %+ x
-%else
-    %define mangle(x) x
-%endif
-
-; Name of the .rodata section.
-; Kludge: Something on OS X fails to align .rodata even given an align attribute,
-; so use a different read-only section.
-%macro SECTION_RODATA 0-1 16
-    %ifidn __OUTPUT_FORMAT__,macho64
-        SECTION .text align=%1
-    %elifidn __OUTPUT_FORMAT__,macho
-        SECTION .text align=%1
-        fakegot:
-    %elifidn __OUTPUT_FORMAT__,aout
-        section .text
-    %else
-        SECTION .rodata align=%1
-    %endif
-%endmacro
-
-; aout does not support align=
-%macro SECTION_TEXT 0-1 16
-    %ifidn __OUTPUT_FORMAT__,aout
-        SECTION .text
-    %else
-        SECTION .text align=%1
-    %endif
-%endmacro
-
-%if WIN64
-    %define PIC
-%elif ARCH_X86_64 == 0
-; x86_32 doesn't require PIC.
-; Some distros prefer shared objects to be PIC, but nothing breaks if
-; the code contains a few textrels, so we'll skip that complexity.
-    %undef PIC
-%endif
-%ifdef PIC
-    default rel
-%endif
-
-; Always use long nops (reduces 0x90 spam in disassembly on x86_32)
-CPU amdnop
-
-; Macros to eliminate most code duplication between x86_32 and x86_64:
-; Currently this works only for leaf functions which load all their arguments
-; into registers at the start, and make no other use of the stack. Luckily that
-; covers most of x264's asm.
-
-; PROLOGUE:
-; %1 = number of arguments. loads them from stack if needed.
-; %2 = number of registers used. pushes callee-saved regs if needed.
-; %3 = number of xmm registers used. pushes callee-saved xmm regs if needed.
-; %4 = list of names to define to registers
-; PROLOGUE can also be invoked by adding the same options to cglobal
-
-; e.g.
-; cglobal foo, 2,3,0, dst, src, tmp
-; declares a function (foo), taking two args (dst and src) and one local variable (tmp)
-
-; TODO Some functions can use some args directly from the stack. If they're the
-; last args then you can just not declare them, but if they're in the middle
-; we need more flexible macro.
-
-; RET:
-; Pops anything that was pushed by PROLOGUE, and returns.
-
-; REP_RET:
-; Same, but if it doesn't pop anything it becomes a 2-byte ret, for athlons
-; which are slow when a normal ret follows a branch.
-
-; registers:
-; rN and rNq are the native-size register holding function argument N
-; rNd, rNw, rNb are dword, word, and byte size
-; rNh is the high 8 bits of the word size
-; rNm is the original location of arg N (a register or on the stack), dword
-; rNmp is native size
-
-%macro DECLARE_REG 2-3
-    %define r%1q %2
-    %define r%1d %2d
-    %define r%1w %2w
-    %define r%1b %2b
-    %define r%1h %2h
-    %if %0 == 2
-        %define r%1m  %2d
-        %define r%1mp %2
-    %elif ARCH_X86_64 ; memory
-        %define r%1m [rsp + stack_offset + %3]
-        %define r%1mp qword r %+ %1m
-    %else
-        %define r%1m [esp + stack_offset + %3]
-        %define r%1mp dword r %+ %1m
-    %endif
-    %define r%1  %2
-%endmacro
-
-%macro DECLARE_REG_SIZE 3
-    %define r%1q r%1
-    %define e%1q r%1
-    %define r%1d e%1
-    %define e%1d e%1
-    %define r%1w %1
-    %define e%1w %1
-    %define r%1h %3
-    %define e%1h %3
-    %define r%1b %2
-    %define e%1b %2
-%if ARCH_X86_64 == 0
-    %define r%1  e%1
-%endif
-%endmacro
-
-DECLARE_REG_SIZE ax, al, ah
-DECLARE_REG_SIZE bx, bl, bh
-DECLARE_REG_SIZE cx, cl, ch
-DECLARE_REG_SIZE dx, dl, dh
-DECLARE_REG_SIZE si, sil, null
-DECLARE_REG_SIZE di, dil, null
-DECLARE_REG_SIZE bp, bpl, null
-
-; t# defines for when per-arch register allocation is more complex than just function arguments
-
-%macro DECLARE_REG_TMP 1-*
-    %assign %%i 0
-    %rep %0
-        CAT_XDEFINE t, %%i, r%1
-        %assign %%i %%i+1
-        %rotate 1
-    %endrep
-%endmacro
-
-%macro DECLARE_REG_TMP_SIZE 0-*
-    %rep %0
-        %define t%1q t%1 %+ q
-        %define t%1d t%1 %+ d
-        %define t%1w t%1 %+ w
-        %define t%1h t%1 %+ h
-        %define t%1b t%1 %+ b
-        %rotate 1
-    %endrep
-%endmacro
-
-DECLARE_REG_TMP_SIZE 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14
-
-%if ARCH_X86_64
-    %define gprsize 8
-%else
-    %define gprsize 4
-%endif
-
-%macro PUSH 1
-    push %1
-    %assign stack_offset stack_offset+gprsize
-%endmacro
-
-%macro POP 1
-    pop %1
-    %assign stack_offset stack_offset-gprsize
-%endmacro
-
-%macro PUSH_IF_USED 1-*
-    %rep %0
-        %if %1 < regs_used
-            PUSH r%1
-        %endif
-        %rotate 1
-    %endrep
-%endmacro
-
-%macro POP_IF_USED 1-*
-    %rep %0
-        %if %1 < regs_used
-            pop r%1
-        %endif
-        %rotate 1
-    %endrep
-%endmacro
-
-%macro LOAD_IF_USED 1-*
-    %rep %0
-        %if %1 < num_args
-            mov r%1, r %+ %1 %+ mp
-        %endif
-        %rotate 1
-    %endrep
-%endmacro
-
-%macro SUB 2
-    sub %1, %2
-    %ifidn %1, rsp
-        %assign stack_offset stack_offset+(%2)
-    %endif
-%endmacro
-
-%macro ADD 2
-    add %1, %2
-    %ifidn %1, rsp
-        %assign stack_offset stack_offset-(%2)
-    %endif
-%endmacro
-
-%macro movifnidn 2
-    %ifnidn %1, %2
-        mov %1, %2
-    %endif
-%endmacro
-
-%macro movsxdifnidn 2
-    %ifnidn %1, %2
-        movsxd %1, %2
-    %endif
-%endmacro
-
-%macro ASSERT 1
-    %if (%1) == 0
-        %error assert failed
-    %endif
-%endmacro
-
-%macro DEFINE_ARGS 0-*
-    %ifdef n_arg_names
-        %assign %%i 0
-        %rep n_arg_names
-            CAT_UNDEF arg_name %+ %%i, q
-            CAT_UNDEF arg_name %+ %%i, d
-            CAT_UNDEF arg_name %+ %%i, w
-            CAT_UNDEF arg_name %+ %%i, h
-            CAT_UNDEF arg_name %+ %%i, b
-            CAT_UNDEF arg_name %+ %%i, m
-            CAT_UNDEF arg_name %+ %%i, mp
-            CAT_UNDEF arg_name, %%i
-            %assign %%i %%i+1
-        %endrep
-    %endif
-
-    %xdefine %%stack_offset stack_offset
-    %undef stack_offset ; so that the current value of stack_offset doesn't get baked in by xdefine
-    %assign %%i 0
-    %rep %0
-        %xdefine %1q r %+ %%i %+ q
-        %xdefine %1d r %+ %%i %+ d
-        %xdefine %1w r %+ %%i %+ w
-        %xdefine %1h r %+ %%i %+ h
-        %xdefine %1b r %+ %%i %+ b
-        %xdefine %1m r %+ %%i %+ m
-        %xdefine %1mp r %+ %%i %+ mp
-        CAT_XDEFINE arg_name, %%i, %1
-        %assign %%i %%i+1
-        %rotate 1
-    %endrep
-    %xdefine stack_offset %%stack_offset
-    %assign n_arg_names %0
-%endmacro
-
-%if WIN64 ; Windows x64 ;=================================================
-
-DECLARE_REG 0,  rcx
-DECLARE_REG 1,  rdx
-DECLARE_REG 2,  R8
-DECLARE_REG 3,  R9
-DECLARE_REG 4,  R10, 40
-DECLARE_REG 5,  R11, 48
-DECLARE_REG 6,  rax, 56
-DECLARE_REG 7,  rdi, 64
-DECLARE_REG 8,  rsi, 72
-DECLARE_REG 9,  rbx, 80
-DECLARE_REG 10, rbp, 88
-DECLARE_REG 11, R12, 96
-DECLARE_REG 12, R13, 104
-DECLARE_REG 13, R14, 112
-DECLARE_REG 14, R15, 120
-
-%macro PROLOGUE 2-4+ 0 ; #args, #regs, #xmm_regs, arg_names...
-    %assign num_args %1
-    %assign regs_used %2
-    ASSERT regs_used >= num_args
-    ASSERT regs_used <= 15
-    PUSH_IF_USED 7, 8, 9, 10, 11, 12, 13, 14
-    %if mmsize == 8
-        %assign xmm_regs_used 0
-    %else
-        WIN64_SPILL_XMM %3
-    %endif
-    LOAD_IF_USED 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14
-    DEFINE_ARGS %4
-%endmacro
-
-%macro WIN64_SPILL_XMM 1
-    %assign xmm_regs_used %1
-    ASSERT xmm_regs_used <= 16
-    %if xmm_regs_used > 6
-        SUB rsp, (xmm_regs_used-6)*16+16
-        %assign %%i xmm_regs_used
-        %rep (xmm_regs_used-6)
-            %assign %%i %%i-1
-            movdqa [rsp + (%%i-6)*16+(~stack_offset&8)], xmm %+ %%i
-        %endrep
-    %endif
-%endmacro
-
-%macro WIN64_RESTORE_XMM_INTERNAL 1
-    %if xmm_regs_used > 6
-        %assign %%i xmm_regs_used
-        %rep (xmm_regs_used-6)
-            %assign %%i %%i-1
-            movdqa xmm %+ %%i, [%1 + (%%i-6)*16+(~stack_offset&8)]
-        %endrep
-        add %1, (xmm_regs_used-6)*16+16
-    %endif
-%endmacro
-
-%macro WIN64_RESTORE_XMM 1
-    WIN64_RESTORE_XMM_INTERNAL %1
-    %assign stack_offset stack_offset-(xmm_regs_used-6)*16+16
-    %assign xmm_regs_used 0
-%endmacro
-
-%define has_epilogue regs_used > 7 || xmm_regs_used > 6 || mmsize == 32
-
-%macro RET 0
-    WIN64_RESTORE_XMM_INTERNAL rsp
-    POP_IF_USED 14, 13, 12, 11, 10, 9, 8, 7
-%if mmsize == 32
-    vzeroupper
-%endif
-    ret
-%endmacro
-
-%elif ARCH_X86_64 ; *nix x64 ;=============================================
-
-DECLARE_REG 0,  rdi
-DECLARE_REG 1,  rsi
-DECLARE_REG 2,  rdx
-DECLARE_REG 3,  rcx
-DECLARE_REG 4,  R8
-DECLARE_REG 5,  R9
-DECLARE_REG 6,  rax, 8
-DECLARE_REG 7,  R10, 16
-DECLARE_REG 8,  R11, 24
-DECLARE_REG 9,  rbx, 32
-DECLARE_REG 10, rbp, 40
-DECLARE_REG 11, R12, 48
-DECLARE_REG 12, R13, 56
-DECLARE_REG 13, R14, 64
-DECLARE_REG 14, R15, 72
-
-%macro PROLOGUE 2-4+ ; #args, #regs, #xmm_regs, arg_names...
-    %assign num_args %1
-    %assign regs_used %2
-    ASSERT regs_used >= num_args
-    ASSERT regs_used <= 15
-    PUSH_IF_USED 9, 10, 11, 12, 13, 14
-    LOAD_IF_USED 6, 7, 8, 9, 10, 11, 12, 13, 14
-    DEFINE_ARGS %4
-%endmacro
-
-%define has_epilogue regs_used > 9 || mmsize == 32
-
-%macro RET 0
-    POP_IF_USED 14, 13, 12, 11, 10, 9
-%if mmsize == 32
-    vzeroupper
-%endif
-    ret
-%endmacro
-
-%else ; X86_32 ;==============================================================
-
-DECLARE_REG 0, eax, 4
-DECLARE_REG 1, ecx, 8
-DECLARE_REG 2, edx, 12
-DECLARE_REG 3, ebx, 16
-DECLARE_REG 4, esi, 20
-DECLARE_REG 5, edi, 24
-DECLARE_REG 6, ebp, 28
-%define rsp esp
-
-%macro DECLARE_ARG 1-*
-    %rep %0
-        %define r%1m [esp + stack_offset + 4*%1 + 4]
-        %define r%1mp dword r%1m
-        %rotate 1
-    %endrep
-%endmacro
-
-DECLARE_ARG 7, 8, 9, 10, 11, 12, 13, 14
-
-%macro PROLOGUE 2-4+ ; #args, #regs, #xmm_regs, arg_names...
-    %assign num_args %1
-    %assign regs_used %2
-    %if regs_used > 7
-        %assign regs_used 7
-    %endif
-    ASSERT regs_used >= num_args
-    PUSH_IF_USED 3, 4, 5, 6
-    LOAD_IF_USED 0, 1, 2, 3, 4, 5, 6
-    DEFINE_ARGS %4
-%endmacro
-
-%define has_epilogue regs_used > 3 || mmsize == 32
-
-%macro RET 0
-    POP_IF_USED 6, 5, 4, 3
-%if mmsize == 32
-    vzeroupper
-%endif
-    ret
-%endmacro
-
-%endif ;======================================================================
-
-%if WIN64 == 0
-%macro WIN64_SPILL_XMM 1
-%endmacro
-%macro WIN64_RESTORE_XMM 1
-%endmacro
-%endif
-
-%macro REP_RET 0
-    %if has_epilogue
-        RET
-    %else
-        rep ret
-    %endif
-%endmacro
-
-%macro TAIL_CALL 2 ; callee, is_nonadjacent
-    %if has_epilogue
-        call %1
-        RET
-    %elif %2
-        jmp %1
-    %endif
-%endmacro
-
-;=============================================================================
-; arch-independent part
-;=============================================================================
-
-%assign function_align 16
-
-; Begin a function.
-; Applies any symbol mangling needed for C linkage, and sets up a define such that
-; subsequent uses of the function name automatically refer to the mangled version.
-; Appends cpuflags to the function name if cpuflags has been specified.
-%macro cglobal 1-2+ ; name, [PROLOGUE args]
-%if %0 == 1
-    cglobal_internal %1 %+ SUFFIX
-%else
-    cglobal_internal %1 %+ SUFFIX, %2
-%endif
-%endmacro
-%macro cglobal_internal 1-2+
-    %ifndef cglobaled_%1
-        %xdefine %1 mangle(%1)
-        %xdefine %1.skip_prologue %1 %+ .skip_prologue
-        CAT_XDEFINE cglobaled_, %1, 1
-    %endif
-    %xdefine current_function %1
-    %ifidn __OUTPUT_FORMAT__,elf
-        global %1:function hidden
-    %else
-        global %1
-    %endif
-    align function_align
-    %1:
-    RESET_MM_PERMUTATION ; not really needed, but makes disassembly somewhat nicer
-    %assign stack_offset 0
-    %if %0 > 1
-        PROLOGUE %2
-    %endif
-%endmacro
-
-%macro cextern 1
-    %xdefine %1 mangle(%1)
-    CAT_XDEFINE cglobaled_, %1, 1
-    extern %1
-%endmacro
-
-; like cextern, but without the prefix
-%macro cextern_naked 1
-    %xdefine %1 mangle(%1)
-    CAT_XDEFINE cglobaled_, %1, 1
-    extern %1
-%endmacro
-
-%macro const 2+
-    %xdefine %1 mangle(%1)
-    global %1
-    %1: %2
-%endmacro
-
-; This is needed for ELF, otherwise the GNU linker assumes the stack is
-; executable by default.
-%ifidn __OUTPUT_FORMAT__,elf
-SECTION .note.GNU-stack noalloc noexec nowrite progbits
-%endif
-%ifidn __OUTPUT_FORMAT__,elf32
-section .note.GNU-stack noalloc noexec nowrite progbits
-%endif
-%ifidn __OUTPUT_FORMAT__,elf64
-section .note.GNU-stack noalloc noexec nowrite progbits
-%endif
-
-; cpuflags
-
-%assign cpuflags_MMX      (1<<0)
-%assign cpuflags_MMX2     (1<<1) | cpuflags_MMX
-%assign cpuflags_3dnow    (1<<2) | cpuflags_MMX
-%assign cpuflags_3dnow2   (1<<3) | cpuflags_3dnow
-%assign cpuflags_SSE      (1<<4) | cpuflags_MMX2
-%assign cpuflags_SSE2     (1<<5) | cpuflags_SSE
-%assign cpuflags_SSE2slow (1<<6) | cpuflags_SSE2
-%assign cpuflags_SSE3     (1<<7) | cpuflags_SSE2
-%assign cpuflags_SSSE3    (1<<8) | cpuflags_SSE3
-%assign cpuflags_SSE4     (1<<9) | cpuflags_SSSE3
-%assign cpuflags_SSE42    (1<<10)| cpuflags_SSE4
-%assign cpuflags_AVX      (1<<11)| cpuflags_SSE42
-%assign cpuflags_xop      (1<<12)| cpuflags_AVX
-%assign cpuflags_fma4     (1<<13)| cpuflags_AVX
-%assign cpuflags_AVX2     (1<<14)| cpuflags_AVX
-%assign cpuflags_fma3     (1<<15)| cpuflags_AVX
-
-%assign cpuflags_cache32  (1<<16)
-%assign cpuflags_cache64  (1<<17)
-%assign cpuflags_slowctz  (1<<18)
-%assign cpuflags_lzcnt    (1<<19)
-%assign cpuflags_misalign (1<<20)
-%assign cpuflags_aligned  (1<<21) ; not a cpu feature, but a function variant
-%assign cpuflags_atom     (1<<22)
-%assign cpuflags_bmi1     (1<<23)
-%assign cpuflags_bmi2     (1<<24)|cpuflags_bmi1
-%assign cpuflags_tbm      (1<<25)|cpuflags_bmi1
-
-%define    cpuflag(x) ((cpuflags & (cpuflags_ %+ x)) == (cpuflags_ %+ x))
-%define notcpuflag(x) ((cpuflags & (cpuflags_ %+ x)) != (cpuflags_ %+ x))
-
-; Takes up to 2 cpuflags from the above list.
-; All subsequent functions (up to the next INIT_CPUFLAGS) is built for the specified cpu.
-; You shouldn't need to invoke this macro directly, it's a subroutine for INIT_MMX &co.
-%macro INIT_CPUFLAGS 0-2
-    %if %0 >= 1
-        %xdefine cpuname %1
-        %assign cpuflags cpuflags_%1
-        %if %0 >= 2
-            %xdefine cpuname %1_%2
-            %assign cpuflags cpuflags | cpuflags_%2
-        %endif
-        %xdefine SUFFIX _ %+ cpuname
-        %if cpuflag(AVX)
-            %assign AVX_enabled 1
-        %endif
-        %if mmsize == 16 && notcpuflag(SSE2)
-            %define mova movaps
-            %define movu movups
-            %define movnta movntps
-        %endif
-        %if cpuflag(aligned)
-            %define movu mova
-        %elifidn %1, SSE3
-            %define movu lddqu
-        %endif
-    %else
-        %xdefine SUFFIX
-        %undef cpuname
-        %undef cpuflags
-    %endif
-%endmacro
-
-; merge MMX and SSE*
-
-%macro CAT_XDEFINE 3
-    %xdefine %1%2 %3
-%endmacro
-
-%macro CAT_UNDEF 2
-    %undef %1%2
-%endmacro
-
-%macro INIT_MMX 0-1+
-    %assign AVX_enabled 0
-    %define RESET_MM_PERMUTATION INIT_MMX %1
-    %define mmsize 8
-    %define num_mmregs 8
-    %define mova movq
-    %define movu movq
-    %define movh movd
-    %define movnta movntq
-    %assign %%i 0
-    %rep 8
-    CAT_XDEFINE m, %%i, mm %+ %%i
-    CAT_XDEFINE nmm, %%i, %%i
-    %assign %%i %%i+1
-    %endrep
-    %rep 8
-    CAT_UNDEF m, %%i
-    CAT_UNDEF nmm, %%i
-    %assign %%i %%i+1
-    %endrep
-    INIT_CPUFLAGS %1
-%endmacro
-
-%macro INIT_XMM 0-1+
-    %assign AVX_enabled 0
-    %define RESET_MM_PERMUTATION INIT_XMM %1
-    %define mmsize 16
-    %define num_mmregs 8
-    %if ARCH_X86_64
-    %define num_mmregs 16
-    %endif
-    %define mova movdqa
-    %define movu movdqu
-    %define movh movq
-    %define movnta movntdq
-    %assign %%i 0
-    %rep num_mmregs
-    CAT_XDEFINE m, %%i, xmm %+ %%i
-    CAT_XDEFINE nxmm, %%i, %%i
-    %assign %%i %%i+1
-    %endrep
-    INIT_CPUFLAGS %1
-%endmacro
-
-%macro INIT_YMM 0-1+
-    %assign AVX_enabled 1
-    %define RESET_MM_PERMUTATION INIT_YMM %1
-    %define mmsize 32
-    %define num_mmregs 8
-    %if ARCH_X86_64
-    %define num_mmregs 16
-    %endif
-    %define mova vmovaps
-    %define movu vmovups
-    %undef movh
-    %define movnta vmovntps
-    %assign %%i 0
-    %rep num_mmregs
-    CAT_XDEFINE m, %%i, ymm %+ %%i
-    CAT_XDEFINE nymm, %%i, %%i
-    %assign %%i %%i+1
-    %endrep
-    INIT_CPUFLAGS %1
-%endmacro
-
-INIT_XMM
-
-; I often want to use macros that permute their arguments. e.g. there's no
-; efficient way to implement butterfly or transpose or dct without swapping some
-; arguments.
-;
-; I would like to not have to manually keep track of the permutations:
-; If I insert a permutation in the middle of a function, it should automatically
-; change everything that follows. For more complex macros I may also have multiple
-; implementations, e.g. the SSE2 and SSSE3 versions may have different permutations.
-;
-; Hence these macros. Insert a PERMUTE or some SWAPs at the end of a macro that
-; permutes its arguments. It's equivalent to exchanging the contents of the
-; registers, except that this way you exchange the register names instead, so it
-; doesn't cost any cycles.
-
-%macro PERMUTE 2-* ; takes a list of pairs to swap
-%rep %0/2
-    %xdefine tmp%2 m%2
-    %xdefine ntmp%2 nm%2
-    %rotate 2
-%endrep
-%rep %0/2
-    %xdefine m%1 tmp%2
-    %xdefine nm%1 ntmp%2
-    %undef tmp%2
-    %undef ntmp%2
-    %rotate 2
-%endrep
-%endmacro
-
-%macro SWAP 2-* ; swaps a single chain (sometimes more concise than pairs)
-%rep %0-1
-%ifdef m%1
-    %xdefine tmp m%1
-    %xdefine m%1 m%2
-    %xdefine m%2 tmp
-    CAT_XDEFINE n, m%1, %1
-    CAT_XDEFINE n, m%2, %2
-%else
-    ; If we were called as "SWAP m0,m1" rather than "SWAP 0,1" infer the original numbers here.
-    ; Be careful using this mode in nested macros though, as in some cases there may be
-    ; other copies of m# that have already been dereferenced and don't get updated correctly.
-    %xdefine %%n1 n %+ %1
-    %xdefine %%n2 n %+ %2
-    %xdefine tmp m %+ %%n1
-    CAT_XDEFINE m, %%n1, m %+ %%n2
-    CAT_XDEFINE m, %%n2, tmp
-    CAT_XDEFINE n, m %+ %%n1, %%n1
-    CAT_XDEFINE n, m %+ %%n2, %%n2
-%endif
-    %undef tmp
-    %rotate 1
-%endrep
-%endmacro
-
-; If SAVE_MM_PERMUTATION is placed at the end of a function, then any later
-; calls to that function will automatically load the permutation, so values can
-; be returned in mmregs.
-%macro SAVE_MM_PERMUTATION 0-1
-    %if %0
-        %xdefine %%f %1_m
-    %else
-        %xdefine %%f current_function %+ _m
-    %endif
-    %assign %%i 0
-    %rep num_mmregs
-        CAT_XDEFINE %%f, %%i, m %+ %%i
-    %assign %%i %%i+1
-    %endrep
-%endmacro
-
-%macro LOAD_MM_PERMUTATION 1 ; name to load from
-    %ifdef %1_m0
-        %assign %%i 0
-        %rep num_mmregs
-            CAT_XDEFINE m, %%i, %1_m %+ %%i
-            CAT_XDEFINE n, m %+ %%i, %%i
-        %assign %%i %%i+1
-        %endrep
-    %endif
-%endmacro
-
-; Append cpuflags to the callee's name iff the appended name is known and the plain name isn't
-%macro call 1
-    call_internal %1, %1 %+ SUFFIX
-%endmacro
-%macro call_internal 2
-    %xdefine %%i %1
-    %ifndef cglobaled_%1
-        %ifdef cglobaled_%2
-            %xdefine %%i %2
-        %endif
-    %endif
-    call %%i
-    LOAD_MM_PERMUTATION %%i
-%endmacro
-
-; Substitutions that reduce instruction size but are functionally equivalent
-%macro add 2
-    %ifnum %2
-        %if %2==128
-            sub %1, -128
-        %else
-            add %1, %2
-        %endif
-    %else
-        add %1, %2
-    %endif
-%endmacro
-
-%macro sub 2
-    %ifnum %2
-        %if %2==128
-            add %1, -128
-        %else
-            sub %1, %2
-        %endif
-    %else
-        sub %1, %2
-    %endif
-%endmacro
-
-;=============================================================================
-; AVX abstraction layer
-;=============================================================================
-
-%assign i 0
-%rep 16
-    %if i < 8
-        CAT_XDEFINE sizeofmm, i, 8
-    %endif
-    CAT_XDEFINE sizeofxmm, i, 16
-    CAT_XDEFINE sizeofymm, i, 32
-%assign i i+1
-%endrep
-%undef i
-
-%macro CHECK_AVX_INSTR_EMU 3-*
-    %xdefine %%opcode %1
-    %xdefine %%dst %2
-    %rep %0-2
-        %ifidn %%dst, %3
-            %error non-AVX emulation of ``%%opcode'' is not supported
-        %endif
-        %rotate 1
-    %endrep
-%endmacro
-
-;%1 == instruction
-;%2 == 1 if float, 0 if int
-;%3 == 1 if 4-operand (xmm, xmm, xmm, imm), 0 if 2- or 3-operand (xmm, xmm, xmm)
-;%4 == number of operands given
-;%5+: operands
-%macro RUN_AVX_INSTR 6-7+
-    %ifid %6
-        %define %%sizeofreg sizeof%6
-    %elifid %5
-        %define %%sizeofreg sizeof%5
-    %else
-        %define %%sizeofreg mmsize
-    %endif
-    %if %%sizeofreg==32
-        %if %4>=3
-            v%1 %5, %6, %7
-        %else
-            v%1 %5, %6
-        %endif
-    %else
-        %if %%sizeofreg==8
-            %define %%regmov movq
-        %elif %2
-            %define %%regmov movaps
-        %else
-            %define %%regmov movdqa
-        %endif
-
-        %if %4>=3+%3
-            %ifnidn %5, %6
-                %if AVX_enabled && %%sizeofreg==16
-                    v%1 %5, %6, %7
-                %else
-                    CHECK_AVX_INSTR_EMU {%1 %5, %6, %7}, %5, %7
-                    %%regmov %5, %6
-                    %1 %5, %7
-                %endif
-            %else
-                %1 %5, %7
-            %endif
-        %elif %4>=3
-            %1 %5, %6, %7
-        %else
-            %1 %5, %6
-        %endif
-    %endif
-%endmacro
-
-; 3arg AVX ops with a memory arg can only have it in src2,
-; whereas SSE emulation of 3arg prefers to have it in src1 (i.e. the mov).
-; So, if the op is symmetric and the wrong one is memory, swap them.
-%macro RUN_AVX_INSTR1 8
-    %assign %%swap 0
-    %if AVX_enabled
-        %ifnid %6
-            %assign %%swap 1
-        %endif
-    %elifnidn %5, %6
-        %ifnid %7
-            %assign %%swap 1
-        %endif
-    %endif
-    %if %%swap && %3 == 0 && %8 == 1
-        RUN_AVX_INSTR %1, %2, %3, %4, %5, %7, %6
-    %else
-        RUN_AVX_INSTR %1, %2, %3, %4, %5, %6, %7
-    %endif
-%endmacro
-
-;%1 == instruction
-;%2 == 1 if float, 0 if int
-;%3 == 1 if 4-operand (xmm, xmm, xmm, imm), 0 if 2- or 3-operand (xmm, xmm, xmm)
-;%4 == 1 if symmetric (i.e. doesn't matter which src arg is which), 0 if not
-%macro AVX_INSTR 4
-    %macro %1 2-9 fnord, fnord, fnord, %1, %2, %3, %4
-        %ifidn %3, fnord
-            RUN_AVX_INSTR %6, %7, %8, 2, %1, %2
-        %elifidn %4, fnord
-            RUN_AVX_INSTR1 %6, %7, %8, 3, %1, %2, %3, %9
-        %elifidn %5, fnord
-            RUN_AVX_INSTR %6, %7, %8, 4, %1, %2, %3, %4
-        %else
-            RUN_AVX_INSTR %6, %7, %8, 5, %1, %2, %3, %4, %5
-        %endif
-    %endmacro
-%endmacro
-
-AVX_INSTR addpd, 1, 0, 1
-AVX_INSTR addps, 1, 0, 1
-AVX_INSTR addsd, 1, 0, 1
-AVX_INSTR addss, 1, 0, 1
-AVX_INSTR addsubpd, 1, 0, 0
-AVX_INSTR addsubps, 1, 0, 0
-AVX_INSTR andpd, 1, 0, 1
-AVX_INSTR andps, 1, 0, 1
-AVX_INSTR andnpd, 1, 0, 0
-AVX_INSTR andnps, 1, 0, 0
-AVX_INSTR blendpd, 1, 0, 0
-AVX_INSTR blendps, 1, 0, 0
-AVX_INSTR blendvpd, 1, 0, 0
-AVX_INSTR blendvps, 1, 0, 0
-AVX_INSTR cmppd, 1, 0, 0
-AVX_INSTR cmpps, 1, 0, 0
-AVX_INSTR cmpsd, 1, 0, 0
-AVX_INSTR cmpss, 1, 0, 0
-AVX_INSTR cvtdq2ps, 1, 0, 0
-AVX_INSTR cvtps2dq, 1, 0, 0
-AVX_INSTR divpd, 1, 0, 0
-AVX_INSTR divps, 1, 0, 0
-AVX_INSTR divsd, 1, 0, 0
-AVX_INSTR divss, 1, 0, 0
-AVX_INSTR dppd, 1, 1, 0
-AVX_INSTR dpps, 1, 1, 0
-AVX_INSTR haddpd, 1, 0, 0
-AVX_INSTR haddps, 1, 0, 0
-AVX_INSTR hsubpd, 1, 0, 0
-AVX_INSTR hsubps, 1, 0, 0
-AVX_INSTR maxpd, 1, 0, 1
-AVX_INSTR maxps, 1, 0, 1
-AVX_INSTR maxsd, 1, 0, 1
-AVX_INSTR maxss, 1, 0, 1
-AVX_INSTR minpd, 1, 0, 1
-AVX_INSTR minps, 1, 0, 1
-AVX_INSTR minsd, 1, 0, 1
-AVX_INSTR minss, 1, 0, 1
-AVX_INSTR movhlps, 1, 0, 0
-AVX_INSTR movlhps, 1, 0, 0
-AVX_INSTR movsd, 1, 0, 0
-AVX_INSTR movss, 1, 0, 0
-AVX_INSTR mpsadbw, 0, 1, 0
-AVX_INSTR mulpd, 1, 0, 1
-AVX_INSTR mulps, 1, 0, 1
-AVX_INSTR mulsd, 1, 0, 1
-AVX_INSTR mulss, 1, 0, 1
-AVX_INSTR orpd, 1, 0, 1
-AVX_INSTR orps, 1, 0, 1
-AVX_INSTR pabsb, 0, 0, 0
-AVX_INSTR pabsw, 0, 0, 0
-AVX_INSTR pabsd, 0, 0, 0
-AVX_INSTR packsswb, 0, 0, 0
-AVX_INSTR packssdw, 0, 0, 0
-AVX_INSTR packuswb, 0, 0, 0
-AVX_INSTR packusdw, 0, 0, 0
-AVX_INSTR paddb, 0, 0, 1
-AVX_INSTR paddw, 0, 0, 1
-AVX_INSTR paddd, 0, 0, 1
-AVX_INSTR paddq, 0, 0, 1
-AVX_INSTR paddsb, 0, 0, 1
-AVX_INSTR paddsw, 0, 0, 1
-AVX_INSTR paddusb, 0, 0, 1
-AVX_INSTR paddusw, 0, 0, 1
-AVX_INSTR palignr, 0, 1, 0
-AVX_INSTR pand, 0, 0, 1
-AVX_INSTR pandn, 0, 0, 0
-AVX_INSTR pavgb, 0, 0, 1
-AVX_INSTR pavgw, 0, 0, 1
-AVX_INSTR pblendvb, 0, 0, 0
-AVX_INSTR pblendw, 0, 1, 0
-AVX_INSTR pcmpestri, 0, 0, 0
-AVX_INSTR pcmpestrm, 0, 0, 0
-AVX_INSTR pcmpistri, 0, 0, 0
-AVX_INSTR pcmpistrm, 0, 0, 0
-AVX_INSTR pcmpeqb, 0, 0, 1
-AVX_INSTR pcmpeqw, 0, 0, 1
-AVX_INSTR pcmpeqd, 0, 0, 1
-AVX_INSTR pcmpeqq, 0, 0, 1
-AVX_INSTR pcmpgtb, 0, 0, 0
-AVX_INSTR pcmpgtw, 0, 0, 0
-AVX_INSTR pcmpgtd, 0, 0, 0
-AVX_INSTR pcmpgtq, 0, 0, 0
-AVX_INSTR phaddw, 0, 0, 0
-AVX_INSTR phaddd, 0, 0, 0
-AVX_INSTR phaddsw, 0, 0, 0
-AVX_INSTR phsubw, 0, 0, 0
-AVX_INSTR phsubd, 0, 0, 0
-AVX_INSTR phsubsw, 0, 0, 0
-AVX_INSTR pmaddwd, 0, 0, 1
-AVX_INSTR pmaddubsw, 0, 0, 0
-AVX_INSTR pmaxsb, 0, 0, 1
-AVX_INSTR pmaxsw, 0, 0, 1
-AVX_INSTR pmaxsd, 0, 0, 1
-AVX_INSTR pmaxub, 0, 0, 1
-AVX_INSTR pmaxuw, 0, 0, 1
-AVX_INSTR pmaxud, 0, 0, 1
-AVX_INSTR pminsb, 0, 0, 1
-AVX_INSTR pminsw, 0, 0, 1
-AVX_INSTR pminsd, 0, 0, 1
-AVX_INSTR pminub, 0, 0, 1
-AVX_INSTR pminuw, 0, 0, 1
-AVX_INSTR pminud, 0, 0, 1
-AVX_INSTR pmovmskb, 0, 0, 0
-AVX_INSTR pmulhuw, 0, 0, 1
-AVX_INSTR pmulhrsw, 0, 0, 1
-AVX_INSTR pmulhw, 0, 0, 1
-AVX_INSTR pmullw, 0, 0, 1
-AVX_INSTR pmulld, 0, 0, 1
-AVX_INSTR pmuludq, 0, 0, 1
-AVX_INSTR pmuldq, 0, 0, 1
-AVX_INSTR por, 0, 0, 1
-AVX_INSTR psadbw, 0, 0, 1
-AVX_INSTR pshufb, 0, 0, 0
-AVX_INSTR pshufd, 0, 1, 0
-AVX_INSTR pshufhw, 0, 1, 0
-AVX_INSTR pshuflw, 0, 1, 0
-AVX_INSTR psignb, 0, 0, 0
-AVX_INSTR psignw, 0, 0, 0
-AVX_INSTR psignd, 0, 0, 0
-AVX_INSTR psllw, 0, 0, 0
-AVX_INSTR pslld, 0, 0, 0
-AVX_INSTR psllq, 0, 0, 0
-AVX_INSTR pslldq, 0, 0, 0
-AVX_INSTR psraw, 0, 0, 0
-AVX_INSTR psrad, 0, 0, 0
-AVX_INSTR psrlw, 0, 0, 0
-AVX_INSTR psrld, 0, 0, 0
-AVX_INSTR psrlq, 0, 0, 0
-AVX_INSTR psrldq, 0, 0, 0
-AVX_INSTR psubb, 0, 0, 0
-AVX_INSTR psubw, 0, 0, 0
-AVX_INSTR psubd, 0, 0, 0
-AVX_INSTR psubq, 0, 0, 0
-AVX_INSTR psubsb, 0, 0, 0
-AVX_INSTR psubsw, 0, 0, 0
-AVX_INSTR psubusb, 0, 0, 0
-AVX_INSTR psubusw, 0, 0, 0
-AVX_INSTR ptest, 0, 0, 0
-AVX_INSTR punpckhbw, 0, 0, 0
-AVX_INSTR punpckhwd, 0, 0, 0
-AVX_INSTR punpckhdq, 0, 0, 0
-AVX_INSTR punpckhqdq, 0, 0, 0
-AVX_INSTR punpcklbw, 0, 0, 0
-AVX_INSTR punpcklwd, 0, 0, 0
-AVX_INSTR punpckldq, 0, 0, 0
-AVX_INSTR punpcklqdq, 0, 0, 0
-AVX_INSTR pxor, 0, 0, 1
-AVX_INSTR shufps, 1, 1, 0
-AVX_INSTR subpd, 1, 0, 0
-AVX_INSTR subps, 1, 0, 0
-AVX_INSTR subsd, 1, 0, 0
-AVX_INSTR subss, 1, 0, 0
-AVX_INSTR unpckhpd, 1, 0, 0
-AVX_INSTR unpckhps, 1, 0, 0
-AVX_INSTR unpcklpd, 1, 0, 0
-AVX_INSTR unpcklps, 1, 0, 0
-AVX_INSTR xorpd, 1, 0, 1
-AVX_INSTR xorps, 1, 0, 1
-
-; 3DNow instructions, for sharing code between AVX, SSE and 3DN
-AVX_INSTR pfadd, 1, 0, 1
-AVX_INSTR pfsub, 1, 0, 0
-AVX_INSTR pfmul, 1, 0, 1
-
-; base-4 constants for shuffles
-%assign i 0
-%rep 256
-    %assign j ((i>>6)&3)*1000 + ((i>>4)&3)*100 + ((i>>2)&3)*10 + (i&3)
-    %if j < 10
-        CAT_XDEFINE q000, j, i
-    %elif j < 100
-        CAT_XDEFINE q00, j, i
-    %elif j < 1000
-        CAT_XDEFINE q0, j, i
-    %else
-        CAT_XDEFINE q, j, i
-    %endif
-%assign i i+1
-%endrep
-%undef i
-%undef j
-
-%macro FMA_INSTR 3
-    %macro %1 4-7 %1, %2, %3
-        %if cpuflag(xop)
-            v%5 %1, %2, %3, %4
-        %else
-            %6 %1, %2, %3
-            %7 %1, %4
-        %endif
-    %endmacro
-%endmacro
-
-FMA_INSTR  pmacsdd,  pmulld, paddd
-FMA_INSTR  pmacsww,  pmullw, paddw
-FMA_INSTR pmadcswd, pmaddwd, paddd
-
-; tzcnt is equivalent to "rep bsf" and is backwards-compatible with bsf.
-; This lets us use tzcnt without bumping the yasm version requirement yet.
-%define tzcnt rep bsf
diff --git a/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.c b/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.c
deleted file mode 100755
index 3712a3b6d6..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.c
+++ /dev/null
@@ -1,72 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifdef _YUV_LIBYUV
-#include <libyuv.h>
-#include "yuv_util.h"
-#include "yuv_libyuv.h"
-
-void decodeRGB(struct TheoraPixelTransform* t)
-{
-	I420ToRAW(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 3, t->w, t->h);
-}
-
-void decodeRGBA(struct TheoraPixelTransform* t)
-{
-	I420ToABGR(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeRGBX(struct TheoraPixelTransform* t)
-{
-    I420ToABGR(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-}
-
-void decodeARGB(struct TheoraPixelTransform* t)
-{
-    I420ToBGRA(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXRGB(struct TheoraPixelTransform* t)
-{
-    I420ToBGRA(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-}
-
-void decodeBGR(struct TheoraPixelTransform* t)
-{
-	I420ToRGB24(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 3, t->w, t->h);
-}
-
-void decodeBGRA(struct TheoraPixelTransform* t)
-{
-	I420ToARGB(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-	_decodeAlpha(incOut(t, 3), t->w * 4);
-}
-
-void decodeBGRX(struct TheoraPixelTransform* t)
-{
-	I420ToARGB(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-}
-
-void decodeABGR(struct TheoraPixelTransform* t)
-{
-    I420ToRGBA(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-	_decodeAlpha(t, t->w * 4);
-}
-
-void decodeXBGR(struct TheoraPixelTransform* t)
-{
-	I420ToRGBA(t->y, t->yStride, t->u, t->uStride, t->v, t->vStride, t->out, t->w * 4, t->w, t->h);
-}
-
-void initYUVConversionModule()
-{
-
-}
-#endif
diff --git a/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.h b/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.h
deleted file mode 100755
index f621af0c5f..0000000000
--- a/drivers/theoraplayer/src/YUV/libyuv/yuv_libyuv.h
+++ /dev/null
@@ -1,14 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _YUV_LIBYUV_h
-#define _YUV_LIBYUV_h
-
-#include "TheoraPixelTransform.h"
-
-#endif
diff --git a/drivers/theoraplayer/src/YUV/yuv_util.c b/drivers/theoraplayer/src/YUV/yuv_util.c
deleted file mode 100644
index f5bf3e5f9e..0000000000
--- a/drivers/theoraplayer/src/YUV/yuv_util.c
+++ /dev/null
@@ -1,39 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#include "yuv_util.h"
-
-struct TheoraPixelTransform* incOut(struct TheoraPixelTransform* t, int n)
-{
-	// used for XRGB, XBGR and similar
-	t->out += n;
-	return t;
-}
-
-void _decodeAlpha(struct TheoraPixelTransform* t, int stride)
-{
-	int width = t->w;
-	unsigned char *ySrc, *yLineEnd, *out;
-	int luma;
-	unsigned int y;
-	for (y = 0; y < t->h; y++)
-	{
-		ySrc = t->y + y * t->yStride + width;
-		out = t->out + y * stride;
-		
-		for (yLineEnd = ySrc + width; ySrc != yLineEnd; ++ySrc, out += 4)
-		{
-			luma = (*ySrc);
-            // because in YCbCr specification, luma values are in the range of [16, 235]
-            // account for 'footroom' and 'headroom' ranges while using luma values as alpha channel
-            if (luma <= 16)       *out = 0;
-            else if (luma >= 235) *out = 255;
-            else                  *out = (unsigned char) (((luma - 16) * 255) / 219);
-		}
-	}
-}
diff --git a/drivers/theoraplayer/src/YUV/yuv_util.h b/drivers/theoraplayer/src/YUV/yuv_util.h
deleted file mode 100644
index 1f9d76634a..0000000000
--- a/drivers/theoraplayer/src/YUV/yuv_util.h
+++ /dev/null
@@ -1,17 +0,0 @@
-/************************************************************************************
-This source file is part of the Theora Video Playback Library
-For latest info, see http://libtheoraplayer.googlecode.com
-*************************************************************************************
-Copyright (c) 2008-2014 Kresimir Spes (kspes@cateia.com)
-This program is free software; you can redistribute it and/or modify it under
-the terms of the BSD license: http://opensource.org/licenses/BSD-3-Clause
-*************************************************************************************/
-#ifndef _YUV_UTIL_h
-#define _YUV_UTIL_h
-
-#include "TheoraPixelTransform.h"
-
-struct TheoraPixelTransform* incOut(struct TheoraPixelTransform* t, int n);
-void _decodeAlpha(struct TheoraPixelTransform* t, int stride);
-
-#endif
diff --git a/drivers/theoraplayer/theoraplayer.xcodeproj/project.pbxproj b/drivers/theoraplayer/theoraplayer.xcodeproj/project.pbxproj
deleted file mode 100644
index 23f875fe0c..0000000000
--- a/drivers/theoraplayer/theoraplayer.xcodeproj/project.pbxproj
+++ /dev/null
@@ -1,2606 +0,0 @@
-// !$*UTF8*$!
-{
-	archiveVersion = 1;
-	classes = {
-	};
-	objectVersion = 46;
-	objects = {
-
-/* Begin PBXBuildFile section */
-		D139462D17C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139462E17C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139462F17C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463017C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463117C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463217C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463317C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463417C0ED450091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D139463617C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463717C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463817C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463917C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463A17C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463B17C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463C17C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463D17C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D139463E17C0ED450091F4A4 /* yuv_libyuv.h in Headers */ = {isa = PBXBuildFile; fileRef = D139462C17C0ED450091F4A4 /* yuv_libyuv.h */; };
-		D13946C617C110670091F4A4 /* yuv_libyuv.c in Sources */ = {isa = PBXBuildFile; fileRef = D139462B17C0ED450091F4A4 /* yuv_libyuv.c */; };
-		D13946CC17C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946CD17C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946CE17C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946CF17C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D017C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D117C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D217C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D317C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D417C119B40091F4A4 /* yuv_util.c in Sources */ = {isa = PBXBuildFile; fileRef = D13946CA17C119B30091F4A4 /* yuv_util.c */; };
-		D13946D517C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946D617C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946D717C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946D817C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946D917C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946DA17C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946DB17C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946DC17C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D13946DD17C119B40091F4A4 /* yuv_util.h in Headers */ = {isa = PBXBuildFile; fileRef = D13946CB17C119B30091F4A4 /* yuv_util.h */; };
-		D159BCB017C227F30030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB117C227F40030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB217C227F40030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB317C227F40030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB417C227F50030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB517C227F50030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB617C227F60030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB717C227F60030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB817C227F70030FAB6 /* convert_from.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05617C157CD00CA0FD2 /* convert_from.cc */; };
-		D159BCB917C228310030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBA17C228320030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBB17C228320030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBC17C228330030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBD17C228330030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBE17C228340030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCBF17C228340030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCC017C228340030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCC117C228350030FAB6 /* rotate_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */; };
-		D159BCC217C2286D0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC317C2286D0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC417C2286D0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC517C2286E0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC617C2286E0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC717C2286F0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC817C2286F0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCC917C2286F0030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D159BCCA17C228700030FAB6 /* scale.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06F17C157CD00CA0FD2 /* scale.cc */; };
-		D15D361017C386A600F40439 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D15D361117C386A600F40439 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D15D361217C386A700F40439 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D15D361317C386B100F40439 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D15D361517C386B300F40439 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D15D361617C386B400F40439 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D16775AB155C501D0050EC64 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D16775AC155C501D0050EC64 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D16775AD155C501D0050EC64 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D16775AE155C501D0050EC64 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D16775AF155C501D0050EC64 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D16775B0155C501D0050EC64 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D16775B1155C501D0050EC64 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D16775B2155C501D0050EC64 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D16775B3155C501D0050EC64 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D16775B4155C501D0050EC64 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D16775B5155C501D0050EC64 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D16775B6155C501D0050EC64 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D16775B7155C501D0050EC64 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D16775B8155C501D0050EC64 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D16775B9155C501D0050EC64 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D16775BA155C501D0050EC64 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D16775BB155C501D0050EC64 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D16775BC155C501D0050EC64 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D16775BD155C501D0050EC64 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D16775BE155C501D0050EC64 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D16775BF155C501D0050EC64 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D16775C0155C501D0050EC64 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D16775CE155C50280050EC64 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775CF155C50280050EC64 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D16775D0155C50280050EC64 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775D1155C50280050EC64 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D16775D2155C50280050EC64 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775D3155C50280050EC64 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D16775D4155C50280050EC64 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775D5155C50280050EC64 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D16775D6155C50280050EC64 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775D7155C50280050EC64 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D16775D8155C50280050EC64 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775D9155C50280050EC64 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D16775DA155C50280050EC64 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775DB155C50280050EC64 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D16775DC155C50280050EC64 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775DD155C50280050EC64 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D16775DE155C50280050EC64 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775DF155C50280050EC64 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D16775E0155C50280050EC64 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775E1155C50280050EC64 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D16775E2155C50280050EC64 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775E3155C50280050EC64 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D16775E4155C50280050EC64 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775E5155C50280050EC64 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D16775E6155C50280050EC64 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D16775E7155C50280050EC64 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D198F952177A31FC002942E3 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D198F953177A31FC002942E3 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D198F954177A31FC002942E3 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D198F955177A31FC002942E3 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D198F956177A31FC002942E3 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D198F957177A31FC002942E3 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D198F958177A31FC002942E3 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D198F959177A31FC002942E3 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D198F95A177A31FC002942E3 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D198F95B177A31FC002942E3 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D198F95C177A31FC002942E3 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D198F95D177A31FC002942E3 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D198F95F177A31FC002942E3 /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D198F960177A31FC002942E3 /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D198F961177A31FC002942E3 /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D198F962177A31FC002942E3 /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D198F965177A31FC002942E3 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D198F966177A31FC002942E3 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D198F967177A31FC002942E3 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D198F968177A31FC002942E3 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D198F969177A31FC002942E3 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D198F96A177A31FC002942E3 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D198F96B177A31FC002942E3 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D198F96C177A31FC002942E3 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D198F96D177A31FC002942E3 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D198F96E177A31FC002942E3 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D198F96F177A31FC002942E3 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D198F970177A31FC002942E3 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D198F971177A31FC002942E3 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D198F972177A31FC002942E3 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D198F974177A31FC002942E3 /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D198F97E177A31FE002942E3 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D198F97F177A31FE002942E3 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D198F980177A31FE002942E3 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D198F981177A31FE002942E3 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D198F982177A31FE002942E3 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D198F983177A31FE002942E3 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D198F984177A31FE002942E3 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D198F985177A31FE002942E3 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D198F986177A31FE002942E3 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D198F987177A31FE002942E3 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D198F988177A31FE002942E3 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D198F989177A31FE002942E3 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D198F98B177A31FE002942E3 /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D198F98C177A31FE002942E3 /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D198F98D177A31FE002942E3 /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D198F98E177A31FE002942E3 /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D198F991177A31FE002942E3 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D198F992177A31FE002942E3 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D198F993177A31FE002942E3 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D198F994177A31FE002942E3 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D198F995177A31FE002942E3 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D198F996177A31FE002942E3 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D198F997177A31FE002942E3 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D198F998177A31FE002942E3 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D198F999177A31FE002942E3 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D198F99A177A31FE002942E3 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D198F99B177A31FE002942E3 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D198F99C177A31FE002942E3 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D198F99D177A31FE002942E3 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D198F99E177A31FE002942E3 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D198F9A0177A31FE002942E3 /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D198F9AA177A3200002942E3 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D198F9AB177A3200002942E3 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D198F9AC177A3200002942E3 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D198F9AD177A3200002942E3 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D198F9AE177A3200002942E3 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D198F9AF177A3200002942E3 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D198F9B0177A3200002942E3 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D198F9B1177A3200002942E3 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D198F9B2177A3200002942E3 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D198F9B3177A3200002942E3 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D198F9B4177A3200002942E3 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D198F9B5177A3200002942E3 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D198F9B6177A3200002942E3 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D198F9B8177A3200002942E3 /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D198F9B9177A3200002942E3 /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D198F9BA177A3200002942E3 /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D198F9BB177A3200002942E3 /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D198F9BE177A3200002942E3 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D198F9BF177A3200002942E3 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D198F9C0177A3200002942E3 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D198F9C1177A3200002942E3 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D198F9C2177A3200002942E3 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D198F9C3177A3200002942E3 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D198F9C4177A3200002942E3 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D198F9C5177A3200002942E3 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D198F9C6177A3200002942E3 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D198F9C7177A3200002942E3 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D198F9C8177A3200002942E3 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D198F9C9177A3200002942E3 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D198F9CA177A3200002942E3 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D198F9CB177A3200002942E3 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D198F9CD177A3200002942E3 /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D1BCE05A18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE05B18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE05C18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE05D18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE05E18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE05F18F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE06018F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE06118F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE06218F3F7FE00C83470 /* scale_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05818F3F7FE00C83470 /* scale_common.cc */; };
-		D1BCE06318F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06418F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06518F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06618F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06718F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06818F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06918F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06A18F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1BCE06B18F3F7FE00C83470 /* scale_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1BCE05918F3F7FE00C83470 /* scale_posix.cc */; };
-		D1C3D07217C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07317C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07417C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07517C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07617C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07717C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07817C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07917C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D07A17C157CD00CA0FD2 /* compare_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */; };
-		D1C3D08117C157CD00CA0FD2 /* compare_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05017C157CD00CA0FD2 /* compare_neon.cc */; };
-		D1C3D08217C157CD00CA0FD2 /* compare_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05017C157CD00CA0FD2 /* compare_neon.cc */; };
-		D1C3D08317C157CD00CA0FD2 /* compare_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05017C157CD00CA0FD2 /* compare_neon.cc */; };
-		D1C3D08417C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D08517C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D08617C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D08717C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D08817C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D08917C157CD00CA0FD2 /* compare_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */; };
-		D1C3D09617C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09717C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09817C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09917C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09A17C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09B17C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09C17C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09D17C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09E17C157CD00CA0FD2 /* compare.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05317C157CD00CA0FD2 /* compare.cc */; };
-		D1C3D09F17C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A017C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A117C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A217C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A317C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A417C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A517C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A617C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0A717C157CD00CA0FD2 /* convert_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */; };
-		D1C3D0C317C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C417C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C517C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C617C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C717C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C817C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0C917C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0CA17C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0CB17C157CD00CA0FD2 /* convert_to_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */; };
-		D1C3D0CC17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0CD17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0CE17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0CF17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D017C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D117C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D217C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D317C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D417C157CD00CA0FD2 /* convert_to_i420.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */; };
-		D1C3D0D517C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0D617C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0D717C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0D817C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0D917C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0DA17C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0DB17C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0DC17C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0DD17C157CD00CA0FD2 /* convert.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05A17C157CD00CA0FD2 /* convert.cc */; };
-		D1C3D0DE17C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0DF17C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E017C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E117C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E217C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E317C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E417C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E517C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E617C157CD00CA0FD2 /* cpu_id.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */; };
-		D1C3D0E717C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0E817C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0E917C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0EA17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0EB17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0EC17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0ED17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0EE17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D0EF17C157CD00CA0FD2 /* format_conversion.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */; };
-		D1C3D10217C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10317C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10417C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10517C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10617C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10717C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10817C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10917C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D10A17C157CD00CA0FD2 /* planar_functions.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */; };
-		D1C3D12317C157CD00CA0FD2 /* rotate_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06217C157CD00CA0FD2 /* rotate_neon.cc */; };
-		D1C3D12417C157CD00CA0FD2 /* rotate_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06217C157CD00CA0FD2 /* rotate_neon.cc */; };
-		D1C3D12517C157CD00CA0FD2 /* rotate_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06217C157CD00CA0FD2 /* rotate_neon.cc */; };
-		D1C3D12617C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12717C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12817C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12917C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12A17C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12B17C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12C17C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12D17C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12E17C157CD00CA0FD2 /* rotate.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06317C157CD00CA0FD2 /* rotate.cc */; };
-		D1C3D12F17C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13017C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13117C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13217C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13317C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13417C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13517C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13617C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13717C157CD00CA0FD2 /* row_any.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06417C157CD00CA0FD2 /* row_any.cc */; };
-		D1C3D13817C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13917C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13A17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13B17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13C17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13D17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13E17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D13F17C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D14017C157CD00CA0FD2 /* row_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06517C157CD00CA0FD2 /* row_common.cc */; };
-		D1C3D15017C157CD00CA0FD2 /* row_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06717C157CD00CA0FD2 /* row_neon.cc */; };
-		D1C3D15117C157CD00CA0FD2 /* row_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06717C157CD00CA0FD2 /* row_neon.cc */; };
-		D1C3D15217C157CD00CA0FD2 /* row_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06717C157CD00CA0FD2 /* row_neon.cc */; };
-		D1C3D15317C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D15417C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D15517C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D15617C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D15717C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D15817C157CD00CA0FD2 /* row_posix.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06817C157CD00CA0FD2 /* row_posix.cc */; };
-		D1C3D17417C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06B17C157CD00CA0FD2 /* scale_argb_neon.cc */; };
-		D1C3D17517C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06B17C157CD00CA0FD2 /* scale_argb_neon.cc */; };
-		D1C3D17617C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06B17C157CD00CA0FD2 /* scale_argb_neon.cc */; };
-		D1C3D17717C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17817C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17917C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17A17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17B17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17C17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17D17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17E17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D17F17C157CD00CA0FD2 /* scale_argb.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */; };
-		D1C3D18F17C157CD00CA0FD2 /* scale_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06E17C157CD00CA0FD2 /* scale_neon.cc */; };
-		D1C3D19017C157CD00CA0FD2 /* scale_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06E17C157CD00CA0FD2 /* scale_neon.cc */; };
-		D1C3D19117C157CD00CA0FD2 /* scale_neon.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D06E17C157CD00CA0FD2 /* scale_neon.cc */; };
-		D1C3D19B17C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D19C17C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D19D17C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D19E17C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D19F17C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D1A017C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D1A117C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D1A217C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1C3D1A317C157CD00CA0FD2 /* video_common.cc in Sources */ = {isa = PBXBuildFile; fileRef = D1C3D07017C157CD00CA0FD2 /* video_common.cc */; };
-		D1CD00001696FC0B00609AB0 /* Theora.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFFB1696FC0100609AB0 /* Theora.framework */; };
-		D1CD00011696FC0B00609AB0 /* Vorbis.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF91696FBF700609AB0 /* Vorbis.framework */; };
-		D1CD00021696FC0B00609AB0 /* Ogg.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF71696FBF400609AB0 /* Ogg.framework */; };
-		D1CD00041696FF9400609AB0 /* CoreMedia.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CD00031696FF9400609AB0 /* CoreMedia.framework */; };
-		D1CD00051696FF9600609AB0 /* CoreMedia.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CD00031696FF9400609AB0 /* CoreMedia.framework */; };
-		D1CDFF241696C77A00609AB0 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D1CDFF251696C77A00609AB0 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D1CDFF261696C77A00609AB0 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D1CDFF271696C77A00609AB0 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D1CDFF281696C77A00609AB0 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D1CDFF291696C77A00609AB0 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D1CDFF2A1696C77A00609AB0 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D1CDFF2B1696C77A00609AB0 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D1CDFF2C1696C77A00609AB0 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D1CDFF2D1696C77A00609AB0 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D1CDFF2E1696C77A00609AB0 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D1CDFF341696C77A00609AB0 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF351696C77A00609AB0 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF361696C77A00609AB0 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF371696C77A00609AB0 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF381696C77A00609AB0 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF391696C77A00609AB0 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3A1696C77A00609AB0 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3B1696C77A00609AB0 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3C1696C77A00609AB0 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3D1696C77A00609AB0 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3E1696C77A00609AB0 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF3F1696C77A00609AB0 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF401696C77A00609AB0 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF4C1696C79700609AB0 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D1CDFF4D1696C79700609AB0 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D1CDFF4E1696C79700609AB0 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D1CDFF4F1696C79700609AB0 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D1CDFF501696C79700609AB0 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D1CDFF511696C79700609AB0 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D1CDFF521696C79700609AB0 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D1CDFF531696C79700609AB0 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D1CDFF541696C79700609AB0 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D1CDFF551696C79700609AB0 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D1CDFF561696C79700609AB0 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D1CDFF5C1696C79700609AB0 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF5D1696C79700609AB0 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF5E1696C79700609AB0 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF5F1696C79700609AB0 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF601696C79700609AB0 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF611696C79700609AB0 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF621696C79700609AB0 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF631696C79700609AB0 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF641696C79700609AB0 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF651696C79700609AB0 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF661696C79700609AB0 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF671696C79700609AB0 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF681696C79700609AB0 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1CDFF961696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D1CDFF971696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D1CDFF981696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D1CDFF991696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D1CDFF9A1696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D1CDFF9B1696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D1CDFF9E1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D1CDFF9F1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF9D1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h */; };
-		D1CDFFA21696E1CA00609AB0 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D1CDFFA31696E1CA00609AB0 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D1CDFFA41696E1CA00609AB0 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D1CDFFA51696E1CA00609AB0 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D1CDFFA61696E1CA00609AB0 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D1CDFFA71696E1CA00609AB0 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D1CDFFA81696E1CA00609AB0 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D1CDFFA91696E1CA00609AB0 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D1CDFFAA1696E1CA00609AB0 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D1CDFFAB1696E1CA00609AB0 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D1CDFFAC1696E1CA00609AB0 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D1CDFFB01696E1CA00609AB0 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D1CDFFB11696E1CA00609AB0 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D1CDFFB21696E1CA00609AB0 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D1CDFFB31696E1CA00609AB0 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D1CDFFB41696E1CA00609AB0 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D1CDFFB51696E1CA00609AB0 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D1CDFFB61696E1CA00609AB0 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D1CDFFB71696E1CA00609AB0 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D1CDFFB81696E1CA00609AB0 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D1CDFFB91696E1CA00609AB0 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D1CDFFBA1696E1CA00609AB0 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D1CDFFBB1696E1CA00609AB0 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D1CDFFBC1696E1CA00609AB0 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D1CDFFBD1696E1CA00609AB0 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D1CDFFC71696E1D700609AB0 /* TheoraAsync.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759E155C501D0050EC64 /* TheoraAsync.cpp */; };
-		D1CDFFC81696E1D700609AB0 /* TheoraAudioInterface.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */; };
-		D1CDFFC91696E1D700609AB0 /* TheoraDataSource.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */; };
-		D1CDFFCA1696E1D700609AB0 /* TheoraException.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A1155C501D0050EC64 /* TheoraException.cpp */; };
-		D1CDFFCB1696E1D700609AB0 /* TheoraFrameQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */; };
-		D1CDFFCC1696E1D700609AB0 /* TheoraTimer.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A3155C501D0050EC64 /* TheoraTimer.cpp */; };
-		D1CDFFCD1696E1D700609AB0 /* TheoraUtil.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A4155C501D0050EC64 /* TheoraUtil.cpp */; };
-		D1CDFFCE1696E1D700609AB0 /* TheoraVideoClip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */; };
-		D1CDFFCF1696E1D700609AB0 /* TheoraVideoFrame.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */; };
-		D1CDFFD01696E1D700609AB0 /* TheoraVideoManager.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */; };
-		D1CDFFD11696E1D700609AB0 /* TheoraWorkerThread.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */; };
-		D1CDFFD21696E1D700609AB0 /* TheoraVideoClip_Theora.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */; };
-		D1CDFFD51696E1D700609AB0 /* TheoraAsync.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C1155C50280050EC64 /* TheoraAsync.h */; };
-		D1CDFFD61696E1D700609AB0 /* TheoraAudioInterface.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C2155C50280050EC64 /* TheoraAudioInterface.h */; };
-		D1CDFFD71696E1D700609AB0 /* TheoraDataSource.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C3155C50280050EC64 /* TheoraDataSource.h */; };
-		D1CDFFD81696E1D700609AB0 /* TheoraException.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C4155C50280050EC64 /* TheoraException.h */; };
-		D1CDFFD91696E1D700609AB0 /* TheoraExport.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C5155C50280050EC64 /* TheoraExport.h */; };
-		D1CDFFDA1696E1D700609AB0 /* TheoraFrameQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C6155C50280050EC64 /* TheoraFrameQueue.h */; };
-		D1CDFFDB1696E1D700609AB0 /* TheoraPlayer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C7155C50280050EC64 /* TheoraPlayer.h */; };
-		D1CDFFDC1696E1D700609AB0 /* TheoraTimer.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C8155C50280050EC64 /* TheoraTimer.h */; };
-		D1CDFFDD1696E1D700609AB0 /* TheoraUtil.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775C9155C50280050EC64 /* TheoraUtil.h */; };
-		D1CDFFDE1696E1D700609AB0 /* TheoraVideoClip.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CA155C50280050EC64 /* TheoraVideoClip.h */; };
-		D1CDFFDF1696E1D700609AB0 /* TheoraVideoFrame.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CB155C50280050EC64 /* TheoraVideoFrame.h */; };
-		D1CDFFE01696E1D700609AB0 /* TheoraVideoManager.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CC155C50280050EC64 /* TheoraVideoManager.h */; };
-		D1CDFFE11696E1D700609AB0 /* TheoraWorkerThread.h in Headers */ = {isa = PBXBuildFile; fileRef = D16775CD155C50280050EC64 /* TheoraWorkerThread.h */; };
-		D1CDFFE21696E1D700609AB0 /* TheoraVideoClip_Theora.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */; };
-		D1CDFFEA1696E24B00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D1CDFFEB1696E24C00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D1CDFFEC1696E24F00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */ = {isa = PBXBuildFile; fileRef = D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */; };
-		D1CDFFEE1696FB7200609AB0 /* AVFoundation.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFED1696FB7200609AB0 /* AVFoundation.framework */; };
-		D1CDFFF11696FB8900609AB0 /* CoreVideo.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF01696FB8900609AB0 /* CoreVideo.framework */; };
-		D1CDFFF31696FBA800609AB0 /* Foundation.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF21696FBA800609AB0 /* Foundation.framework */; };
-		D1CDFFF41696FBB200609AB0 /* CoreVideo.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF01696FB8900609AB0 /* CoreVideo.framework */; };
-		D1CDFFF51696FBB200609AB0 /* AVFoundation.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFED1696FB7200609AB0 /* AVFoundation.framework */; };
-		D1CDFFF61696FBB200609AB0 /* Foundation.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF21696FBA800609AB0 /* Foundation.framework */; };
-		D1CDFFFD1696FC0800609AB0 /* Theora.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFFB1696FC0100609AB0 /* Theora.framework */; };
-		D1CDFFFE1696FC0800609AB0 /* Vorbis.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF91696FBF700609AB0 /* Vorbis.framework */; };
-		D1CDFFFF1696FC0800609AB0 /* Ogg.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = D1CDFFF71696FBF400609AB0 /* Ogg.framework */; };
-		D1D465D616C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D1D465D516C2D063007A45AA /* TheoraAudioPacketQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1D465D716C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D1D465D516C2D063007A45AA /* TheoraAudioPacketQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1D465D816C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */ = {isa = PBXBuildFile; fileRef = D1D465D516C2D063007A45AA /* TheoraAudioPacketQueue.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1D465DA16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1D465DB16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1D465DC16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1D465DD16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1D465DE16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1D465DF16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */; };
-		D1E2719916B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E2719A16B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E2719B16B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E2719C16B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E2719D16B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E2719E16B46F640046C00C /* yuv420_grey_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718A16B46F640046C00C /* yuv420_grey_c.c */; };
-		D1E271A516B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271A616B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271A716B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271A816B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271A916B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271AA16B46F640046C00C /* yuv420_yuv_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */; };
-		D1E271AC16B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271AD16B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271AE16B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271AF16B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271B016B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271B116B470210046C00C /* yuv420_rgb_c.c in Sources */ = {isa = PBXBuildFile; fileRef = D1E271AB16B470210046C00C /* yuv420_rgb_c.c */; };
-		D1E271B316B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1E271B416B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1E271B516B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; settings = {ATTRIBUTES = (Public, ); }; };
-		D1E271B616B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D1E271B716B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D1E271B816B471E80046C00C /* TheoraPixelTransform.h in Headers */ = {isa = PBXBuildFile; fileRef = D1E271B216B471E80046C00C /* TheoraPixelTransform.h */; };
-		D1F09EB1169AFEFB00DEEC63 /* TheoraVideoClip_AVFoundation.h in Headers */ = {isa = PBXBuildFile; fileRef = D1CDFF9D1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h */; };
-/* End PBXBuildFile section */
-
-/* Begin PBXFileReference section */
-		D12CA55517734B4200412E5B /* TheoraVideoClip_FFmpeg.cpp */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraVideoClip_FFmpeg.cpp; path = src/FFmpeg/TheoraVideoClip_FFmpeg.cpp; sourceTree = "<group>"; };
-		D12CA55617734B4200412E5B /* TheoraVideoClip_FFmpeg.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = TheoraVideoClip_FFmpeg.h; path = src/FFmpeg/TheoraVideoClip_FFmpeg.h; sourceTree = "<group>"; };
-		D1358BC218D7777200A36FDC /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
-		D1358BC318D7777800A36FDC /* iOS.xcconfig */ = {isa = PBXFileReference; lastKnownFileType = text.xcconfig; name = iOS.xcconfig; path = xcconfig/iOS.xcconfig; sourceTree = "<group>"; };
-		D1358BC418D7777800A36FDC /* Mac.xcconfig */ = {isa = PBXFileReference; lastKnownFileType = text.xcconfig; name = Mac.xcconfig; path = xcconfig/Mac.xcconfig; sourceTree = "<group>"; };
-		D139462B17C0ED450091F4A4 /* yuv_libyuv.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = yuv_libyuv.c; path = src/YUV/libyuv/yuv_libyuv.c; sourceTree = "<group>"; };
-		D139462C17C0ED450091F4A4 /* yuv_libyuv.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = yuv_libyuv.h; path = src/YUV/libyuv/yuv_libyuv.h; sourceTree = "<group>"; };
-		D13946CA17C119B30091F4A4 /* yuv_util.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = yuv_util.c; path = src/YUV/yuv_util.c; sourceTree = "<group>"; };
-		D13946CB17C119B30091F4A4 /* yuv_util.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = yuv_util.h; path = src/YUV/yuv_util.h; sourceTree = "<group>"; };
-		D1473F2A150CA69B00B20490 /* theoraplayer.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = theoraplayer.framework; sourceTree = BUILT_PRODUCTS_DIR; };
-		D159BCAB17C227940030FAB6 /* compare_win.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; name = compare_win.cc; path = src/YUV/libyuv/src/compare_win.cc; sourceTree = "<group>"; };
-		D159BCAC17C227940030FAB6 /* row_win.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; name = row_win.cc; path = src/YUV/libyuv/src/row_win.cc; sourceTree = "<group>"; };
-		D159BCAD17C227940030FAB6 /* row_x86.asm */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.asm.asm; name = row_x86.asm; path = src/YUV/libyuv/src/row_x86.asm; sourceTree = "<group>"; };
-		D159BCAE17C227940030FAB6 /* x86inc.asm */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.asm.asm; name = x86inc.asm; path = src/YUV/libyuv/src/x86inc.asm; sourceTree = "<group>"; };
-		D167759E155C501D0050EC64 /* TheoraAsync.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraAsync.cpp; path = src/TheoraAsync.cpp; sourceTree = "<group>"; };
-		D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraAudioInterface.cpp; path = src/TheoraAudioInterface.cpp; sourceTree = "<group>"; };
-		D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraDataSource.cpp; path = src/TheoraDataSource.cpp; sourceTree = "<group>"; };
-		D16775A1155C501D0050EC64 /* TheoraException.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraException.cpp; path = src/TheoraException.cpp; sourceTree = "<group>"; };
-		D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraFrameQueue.cpp; path = src/TheoraFrameQueue.cpp; sourceTree = "<group>"; };
-		D16775A3155C501D0050EC64 /* TheoraTimer.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraTimer.cpp; path = src/TheoraTimer.cpp; sourceTree = "<group>"; };
-		D16775A4155C501D0050EC64 /* TheoraUtil.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraUtil.cpp; path = src/TheoraUtil.cpp; sourceTree = "<group>"; };
-		D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraVideoClip.cpp; path = src/TheoraVideoClip.cpp; sourceTree = "<group>"; };
-		D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraVideoFrame.cpp; path = src/TheoraVideoFrame.cpp; sourceTree = "<group>"; };
-		D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraVideoManager.cpp; path = src/TheoraVideoManager.cpp; sourceTree = "<group>"; };
-		D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraWorkerThread.cpp; path = src/TheoraWorkerThread.cpp; sourceTree = "<group>"; };
-		D16775C1155C50280050EC64 /* TheoraAsync.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraAsync.h; path = include/theoraplayer/TheoraAsync.h; sourceTree = "<group>"; };
-		D16775C2155C50280050EC64 /* TheoraAudioInterface.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraAudioInterface.h; path = include/theoraplayer/TheoraAudioInterface.h; sourceTree = "<group>"; };
-		D16775C3155C50280050EC64 /* TheoraDataSource.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraDataSource.h; path = include/theoraplayer/TheoraDataSource.h; sourceTree = "<group>"; };
-		D16775C4155C50280050EC64 /* TheoraException.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraException.h; path = include/theoraplayer/TheoraException.h; sourceTree = "<group>"; };
-		D16775C5155C50280050EC64 /* TheoraExport.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraExport.h; path = include/theoraplayer/TheoraExport.h; sourceTree = "<group>"; };
-		D16775C6155C50280050EC64 /* TheoraFrameQueue.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraFrameQueue.h; path = include/theoraplayer/TheoraFrameQueue.h; sourceTree = "<group>"; };
-		D16775C7155C50280050EC64 /* TheoraPlayer.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraPlayer.h; path = include/theoraplayer/TheoraPlayer.h; sourceTree = "<group>"; };
-		D16775C8155C50280050EC64 /* TheoraTimer.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraTimer.h; path = include/theoraplayer/TheoraTimer.h; sourceTree = "<group>"; };
-		D16775C9155C50280050EC64 /* TheoraUtil.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraUtil.h; path = include/theoraplayer/TheoraUtil.h; sourceTree = "<group>"; };
-		D16775CA155C50280050EC64 /* TheoraVideoClip.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraVideoClip.h; path = include/theoraplayer/TheoraVideoClip.h; sourceTree = "<group>"; };
-		D16775CB155C50280050EC64 /* TheoraVideoFrame.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraVideoFrame.h; path = include/theoraplayer/TheoraVideoFrame.h; sourceTree = "<group>"; };
-		D16775CC155C50280050EC64 /* TheoraVideoManager.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraVideoManager.h; path = include/theoraplayer/TheoraVideoManager.h; sourceTree = "<group>"; };
-		D16775CD155C50280050EC64 /* TheoraWorkerThread.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraWorkerThread.h; path = include/theoraplayer/TheoraWorkerThread.h; sourceTree = "<group>"; };
-		D198F97B177A31FC002942E3 /* libtheoraplayer.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D198F9A7177A31FE002942E3 /* libtheoraplayer_avfoundation.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer_avfoundation.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D198F9D4177A3200002942E3 /* libtheoraplayer_theora_avfoundation.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer_theora_avfoundation.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1BB6FAE150E9E7100EF9400 /* libtheoraplayer.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1BCE05718F3F7D800C83470 /* scale_row.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = scale_row.h; path = src/YUV/libyuv/include/libyuv/scale_row.h; sourceTree = "<group>"; };
-		D1BCE05818F3F7FE00C83470 /* scale_common.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_common.cc; path = src/YUV/libyuv/src/scale_common.cc; sourceTree = "<group>"; };
-		D1BCE05918F3F7FE00C83470 /* scale_posix.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_posix.cc; path = src/YUV/libyuv/src/scale_posix.cc; sourceTree = "<group>"; };
-		D1BCE06C18F3F80800C83470 /* scale_win.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; name = scale_win.cc; path = src/YUV/libyuv/src/scale_win.cc; sourceTree = "<group>"; };
-		D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = compare_common.cc; path = src/YUV/libyuv/src/compare_common.cc; sourceTree = "<group>"; };
-		D1C3D05017C157CD00CA0FD2 /* compare_neon.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = compare_neon.cc; path = src/YUV/libyuv/src/compare_neon.cc; sourceTree = "<group>"; };
-		D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = compare_posix.cc; path = src/YUV/libyuv/src/compare_posix.cc; sourceTree = "<group>"; };
-		D1C3D05317C157CD00CA0FD2 /* compare.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = compare.cc; path = src/YUV/libyuv/src/compare.cc; sourceTree = "<group>"; };
-		D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_argb.cc; path = src/YUV/libyuv/src/convert_argb.cc; sourceTree = "<group>"; };
-		D1C3D05517C157CD00CA0FD2 /* convert_from_argb.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_from_argb.cc; path = src/YUV/libyuv/src/convert_from_argb.cc; sourceTree = "<group>"; };
-		D1C3D05617C157CD00CA0FD2 /* convert_from.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_from.cc; path = src/YUV/libyuv/src/convert_from.cc; sourceTree = "<group>"; };
-		D1C3D05717C157CD00CA0FD2 /* convert_jpeg.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_jpeg.cc; path = src/YUV/libyuv/src/convert_jpeg.cc; sourceTree = "<group>"; };
-		D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_to_argb.cc; path = src/YUV/libyuv/src/convert_to_argb.cc; sourceTree = "<group>"; };
-		D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert_to_i420.cc; path = src/YUV/libyuv/src/convert_to_i420.cc; sourceTree = "<group>"; };
-		D1C3D05A17C157CD00CA0FD2 /* convert.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = convert.cc; path = src/YUV/libyuv/src/convert.cc; sourceTree = "<group>"; };
-		D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = cpu_id.cc; path = src/YUV/libyuv/src/cpu_id.cc; sourceTree = "<group>"; };
-		D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = format_conversion.cc; path = src/YUV/libyuv/src/format_conversion.cc; sourceTree = "<group>"; };
-		D1C3D05D17C157CD00CA0FD2 /* mjpeg_decoder.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = mjpeg_decoder.cc; path = src/YUV/libyuv/src/mjpeg_decoder.cc; sourceTree = "<group>"; };
-		D1C3D05E17C157CD00CA0FD2 /* mjpeg_validate.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = mjpeg_validate.cc; path = src/YUV/libyuv/src/mjpeg_validate.cc; sourceTree = "<group>"; };
-		D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = planar_functions.cc; path = src/YUV/libyuv/src/planar_functions.cc; sourceTree = "<group>"; };
-		D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = rotate_argb.cc; path = src/YUV/libyuv/src/rotate_argb.cc; sourceTree = "<group>"; };
-		D1C3D06117C157CD00CA0FD2 /* rotate_mips.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = rotate_mips.cc; path = src/YUV/libyuv/src/rotate_mips.cc; sourceTree = "<group>"; };
-		D1C3D06217C157CD00CA0FD2 /* rotate_neon.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = rotate_neon.cc; path = src/YUV/libyuv/src/rotate_neon.cc; sourceTree = "<group>"; };
-		D1C3D06317C157CD00CA0FD2 /* rotate.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = rotate.cc; path = src/YUV/libyuv/src/rotate.cc; sourceTree = "<group>"; };
-		D1C3D06417C157CD00CA0FD2 /* row_any.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = row_any.cc; path = src/YUV/libyuv/src/row_any.cc; sourceTree = "<group>"; };
-		D1C3D06517C157CD00CA0FD2 /* row_common.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = row_common.cc; path = src/YUV/libyuv/src/row_common.cc; sourceTree = "<group>"; };
-		D1C3D06617C157CD00CA0FD2 /* row_mips.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = row_mips.cc; path = src/YUV/libyuv/src/row_mips.cc; sourceTree = "<group>"; };
-		D1C3D06717C157CD00CA0FD2 /* row_neon.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = row_neon.cc; path = src/YUV/libyuv/src/row_neon.cc; sourceTree = "<group>"; };
-		D1C3D06817C157CD00CA0FD2 /* row_posix.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = row_posix.cc; path = src/YUV/libyuv/src/row_posix.cc; sourceTree = "<group>"; };
-		D1C3D06B17C157CD00CA0FD2 /* scale_argb_neon.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_argb_neon.cc; path = src/YUV/libyuv/src/scale_argb_neon.cc; sourceTree = "<group>"; };
-		D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_argb.cc; path = src/YUV/libyuv/src/scale_argb.cc; sourceTree = "<group>"; };
-		D1C3D06D17C157CD00CA0FD2 /* scale_mips.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_mips.cc; path = src/YUV/libyuv/src/scale_mips.cc; sourceTree = "<group>"; };
-		D1C3D06E17C157CD00CA0FD2 /* scale_neon.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale_neon.cc; path = src/YUV/libyuv/src/scale_neon.cc; sourceTree = "<group>"; };
-		D1C3D06F17C157CD00CA0FD2 /* scale.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = scale.cc; path = src/YUV/libyuv/src/scale.cc; sourceTree = "<group>"; };
-		D1C3D07017C157CD00CA0FD2 /* video_common.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = video_common.cc; path = src/YUV/libyuv/src/video_common.cc; sourceTree = "<group>"; };
-		D1C3D1CE17C15BB400CA0FD2 /* libyuv.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = libyuv.h; path = src/YUV/libyuv/include/libyuv.h; sourceTree = "<group>"; };
-		D1C3D1CF17C15BC100CA0FD2 /* basic_types.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = basic_types.h; path = src/YUV/libyuv/include/libyuv/basic_types.h; sourceTree = "<group>"; };
-		D1C3D1D017C15BC100CA0FD2 /* compare.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = compare.h; path = src/YUV/libyuv/include/libyuv/compare.h; sourceTree = "<group>"; };
-		D1C3D1D117C15BC100CA0FD2 /* convert_argb.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = convert_argb.h; path = src/YUV/libyuv/include/libyuv/convert_argb.h; sourceTree = "<group>"; };
-		D1C3D1D217C15BC100CA0FD2 /* convert_from_argb.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = convert_from_argb.h; path = src/YUV/libyuv/include/libyuv/convert_from_argb.h; sourceTree = "<group>"; };
-		D1C3D1D317C15BC100CA0FD2 /* convert_from.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = convert_from.h; path = src/YUV/libyuv/include/libyuv/convert_from.h; sourceTree = "<group>"; };
-		D1C3D1D417C15BC100CA0FD2 /* convert.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = convert.h; path = src/YUV/libyuv/include/libyuv/convert.h; sourceTree = "<group>"; };
-		D1C3D1D517C15BC100CA0FD2 /* cpu_id.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = cpu_id.h; path = src/YUV/libyuv/include/libyuv/cpu_id.h; sourceTree = "<group>"; };
-		D1C3D1D617C15BC100CA0FD2 /* format_conversion.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = format_conversion.h; path = src/YUV/libyuv/include/libyuv/format_conversion.h; sourceTree = "<group>"; };
-		D1C3D1D717C15BC100CA0FD2 /* mjpeg_decoder.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = mjpeg_decoder.h; path = src/YUV/libyuv/include/libyuv/mjpeg_decoder.h; sourceTree = "<group>"; };
-		D1C3D1D817C15BC100CA0FD2 /* planar_functions.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = planar_functions.h; path = src/YUV/libyuv/include/libyuv/planar_functions.h; sourceTree = "<group>"; };
-		D1C3D1D917C15BC100CA0FD2 /* rotate_argb.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = rotate_argb.h; path = src/YUV/libyuv/include/libyuv/rotate_argb.h; sourceTree = "<group>"; };
-		D1C3D1DA17C15BC100CA0FD2 /* rotate.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = rotate.h; path = src/YUV/libyuv/include/libyuv/rotate.h; sourceTree = "<group>"; };
-		D1C3D1DB17C15BC100CA0FD2 /* row.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = row.h; path = src/YUV/libyuv/include/libyuv/row.h; sourceTree = "<group>"; };
-		D1C3D1DC17C15BC100CA0FD2 /* scale_argb.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = scale_argb.h; path = src/YUV/libyuv/include/libyuv/scale_argb.h; sourceTree = "<group>"; };
-		D1C3D1DD17C15BC100CA0FD2 /* scale.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = scale.h; path = src/YUV/libyuv/include/libyuv/scale.h; sourceTree = "<group>"; };
-		D1C3D1DE17C15BC100CA0FD2 /* version.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = version.h; path = src/YUV/libyuv/include/libyuv/version.h; sourceTree = "<group>"; };
-		D1C3D1DF17C15BC100CA0FD2 /* video_common.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = video_common.h; path = src/YUV/libyuv/include/libyuv/video_common.h; sourceTree = "<group>"; };
-		D1CD00031696FF9400609AB0 /* CoreMedia.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = CoreMedia.framework; path = System/Library/Frameworks/CoreMedia.framework; sourceTree = SDKROOT; };
-		D1CDFF481696C77A00609AB0 /* theoraplayer.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = theoraplayer.framework; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1CDFF701696C79700609AB0 /* theoraplayer.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = theoraplayer.framework; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraVideoClip_Theora.cpp; path = src/Theora/TheoraVideoClip_Theora.cpp; sourceTree = "<group>"; };
-		D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraVideoClip_Theora.h; path = src/Theora/TheoraVideoClip_Theora.h; sourceTree = "<group>"; };
-		D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.objcpp; name = TheoraVideoClip_AVFoundation.mm; path = src/AVFoundation/TheoraVideoClip_AVFoundation.mm; sourceTree = "<group>"; };
-		D1CDFF9D1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraVideoClip_AVFoundation.h; path = src/AVFoundation/TheoraVideoClip_AVFoundation.h; sourceTree = "<group>"; };
-		D1CDFFC41696E1CA00609AB0 /* libtheoraplayer.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1CDFFE91696E1D700609AB0 /* libtheoraplayer.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libtheoraplayer.a; sourceTree = BUILT_PRODUCTS_DIR; };
-		D1CDFFED1696FB7200609AB0 /* AVFoundation.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = AVFoundation.framework; path = System/Library/Frameworks/AVFoundation.framework; sourceTree = SDKROOT; };
-		D1CDFFF01696FB8900609AB0 /* CoreVideo.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = CoreVideo.framework; path = System/Library/Frameworks/CoreVideo.framework; sourceTree = SDKROOT; };
-		D1CDFFF21696FBA800609AB0 /* Foundation.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Foundation.framework; path = System/Library/Frameworks/Foundation.framework; sourceTree = SDKROOT; };
-		D1CDFFF71696FBF400609AB0 /* Ogg.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Ogg.framework; path = ../__build__/products/Debug/Ogg.framework; sourceTree = "<group>"; };
-		D1CDFFF91696FBF700609AB0 /* Vorbis.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Vorbis.framework; path = ../__build__/products/Debug/Vorbis.framework; sourceTree = "<group>"; };
-		D1CDFFFB1696FC0100609AB0 /* Theora.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Theora.framework; path = ../__build__/products/Debug/Theora.framework; sourceTree = "<group>"; };
-		D1D465D516C2D063007A45AA /* TheoraAudioPacketQueue.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraAudioPacketQueue.h; path = include/theoraplayer/TheoraAudioPacketQueue.h; sourceTree = "<group>"; };
-		D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = TheoraAudioPacketQueue.cpp; path = src/TheoraAudioPacketQueue.cpp; sourceTree = "<group>"; };
-		D1E2718A16B46F640046C00C /* yuv420_grey_c.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = yuv420_grey_c.c; path = src/YUV/C/yuv420_grey_c.c; sourceTree = "<group>"; };
-		D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = yuv420_yuv_c.c; path = src/YUV/C/yuv420_yuv_c.c; sourceTree = "<group>"; };
-		D1E271AB16B470210046C00C /* yuv420_rgb_c.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = yuv420_rgb_c.c; path = src/YUV/C/yuv420_rgb_c.c; sourceTree = "<group>"; };
-		D1E271B216B471E80046C00C /* TheoraPixelTransform.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TheoraPixelTransform.h; path = include/theoraplayer/TheoraPixelTransform.h; sourceTree = "<group>"; };
-		D1F4DA1D18FECACE007C1968 /* cpu-features.c */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.c; name = "cpu-features.c"; path = "src/YUV/android/cpu-features.c"; sourceTree = "<group>"; };
-		D1F4DA1E18FECACE007C1968 /* cpu-features.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = "cpu-features.h"; path = "src/YUV/android/cpu-features.h"; sourceTree = "<group>"; };
-/* End PBXFileReference section */
-
-/* Begin PBXFrameworksBuildPhase section */
-		D1473F26150CA69B00B20490 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CD00001696FC0B00609AB0 /* Theora.framework in Frameworks */,
-				D1CD00011696FC0B00609AB0 /* Vorbis.framework in Frameworks */,
-				D1CD00021696FC0B00609AB0 /* Ogg.framework in Frameworks */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F963177A31FC002942E3 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F98F177A31FE002942E3 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F9BC177A3200002942E3 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1BB6FAB150E9E7100EF9400 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF2F1696C77A00609AB0 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFFF41696FBB200609AB0 /* CoreVideo.framework in Frameworks */,
-				D1CDFFF51696FBB200609AB0 /* AVFoundation.framework in Frameworks */,
-				D1CDFFF61696FBB200609AB0 /* Foundation.framework in Frameworks */,
-				D1CD00051696FF9600609AB0 /* CoreMedia.framework in Frameworks */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF571696C79700609AB0 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CD00041696FF9400609AB0 /* CoreMedia.framework in Frameworks */,
-				D1CDFFF31696FBA800609AB0 /* Foundation.framework in Frameworks */,
-				D1CDFFF11696FB8900609AB0 /* CoreVideo.framework in Frameworks */,
-				D1CDFFEE1696FB7200609AB0 /* AVFoundation.framework in Frameworks */,
-				D1CDFFFD1696FC0800609AB0 /* Theora.framework in Frameworks */,
-				D1CDFFFE1696FC0800609AB0 /* Vorbis.framework in Frameworks */,
-				D1CDFFFF1696FC0800609AB0 /* Ogg.framework in Frameworks */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFAE1696E1CA00609AB0 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFD31696E1D700609AB0 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXFrameworksBuildPhase section */
-
-/* Begin PBXGroup section */
-		D12CA55417734B2400412E5B /* FFmpeg */ = {
-			isa = PBXGroup;
-			children = (
-				D12CA55517734B4200412E5B /* TheoraVideoClip_FFmpeg.cpp */,
-				D12CA55617734B4200412E5B /* TheoraVideoClip_FFmpeg.h */,
-			);
-			name = FFmpeg;
-			sourceTree = "<group>";
-		};
-		D1358BC118D7776700A36FDC /* config */ = {
-			isa = PBXGroup;
-			children = (
-				D1358BC318D7777800A36FDC /* iOS.xcconfig */,
-				D1358BC418D7777800A36FDC /* Mac.xcconfig */,
-				D1358BC218D7777200A36FDC /* Info.plist */,
-			);
-			name = config;
-			sourceTree = "<group>";
-		};
-		D139462A17C0ED2F0091F4A4 /* libyuv */ = {
-			isa = PBXGroup;
-			children = (
-				D1C3D04E17C157AC00CA0FD2 /* include */,
-				D1C3D04D17C157A800CA0FD2 /* src */,
-				D139462B17C0ED450091F4A4 /* yuv_libyuv.c */,
-				D139462C17C0ED450091F4A4 /* yuv_libyuv.h */,
-			);
-			name = libyuv;
-			sourceTree = "<group>";
-		};
-		D1473F1E150CA69B00B20490 = {
-			isa = PBXGroup;
-			children = (
-				D1358BC118D7776700A36FDC /* config */,
-				D147401F150CAE9600B20490 /* include */,
-				D1473F42150CA6C000B20490 /* src */,
-				D1473F2C150CA69B00B20490 /* Frameworks */,
-				D1473F2B150CA69B00B20490 /* Products */,
-			);
-			sourceTree = "<group>";
-		};
-		D1473F2B150CA69B00B20490 /* Products */ = {
-			isa = PBXGroup;
-			children = (
-				D1473F2A150CA69B00B20490 /* theoraplayer.framework */,
-				D1BB6FAE150E9E7100EF9400 /* libtheoraplayer.a */,
-				D1CDFF481696C77A00609AB0 /* theoraplayer.framework */,
-				D1CDFF701696C79700609AB0 /* theoraplayer.framework */,
-				D1CDFFC41696E1CA00609AB0 /* libtheoraplayer.a */,
-				D1CDFFE91696E1D700609AB0 /* libtheoraplayer.a */,
-				D198F97B177A31FC002942E3 /* libtheoraplayer.a */,
-				D198F9A7177A31FE002942E3 /* libtheoraplayer_avfoundation.a */,
-				D198F9D4177A3200002942E3 /* libtheoraplayer_theora_avfoundation.a */,
-			);
-			name = Products;
-			sourceTree = "<group>";
-		};
-		D1473F2C150CA69B00B20490 /* Frameworks */ = {
-			isa = PBXGroup;
-			children = (
-				D1473F82150CA7F300B20490 /* mac */,
-			);
-			name = Frameworks;
-			sourceTree = "<group>";
-		};
-		D1473F42150CA6C000B20490 /* src */ = {
-			isa = PBXGroup;
-			children = (
-				D1E2718516B46F370046C00C /* YUV */,
-				D1CDFF921696CEFA00609AB0 /* Theora */,
-				D1CDFF931696CF0000609AB0 /* AVFoundation */,
-				D12CA55417734B2400412E5B /* FFmpeg */,
-				D16775A5155C501D0050EC64 /* TheoraVideoClip.cpp */,
-				D167759E155C501D0050EC64 /* TheoraAsync.cpp */,
-				D1D465D916C2D070007A45AA /* TheoraAudioPacketQueue.cpp */,
-				D167759F155C501D0050EC64 /* TheoraAudioInterface.cpp */,
-				D16775A0155C501D0050EC64 /* TheoraDataSource.cpp */,
-				D16775A1155C501D0050EC64 /* TheoraException.cpp */,
-				D16775A2155C501D0050EC64 /* TheoraFrameQueue.cpp */,
-				D16775A3155C501D0050EC64 /* TheoraTimer.cpp */,
-				D16775A4155C501D0050EC64 /* TheoraUtil.cpp */,
-				D16775A6155C501D0050EC64 /* TheoraVideoFrame.cpp */,
-				D16775A7155C501D0050EC64 /* TheoraVideoManager.cpp */,
-				D16775A8155C501D0050EC64 /* TheoraWorkerThread.cpp */,
-			);
-			name = src;
-			sourceTree = "<group>";
-		};
-		D1473F82150CA7F300B20490 /* mac */ = {
-			isa = PBXGroup;
-			children = (
-				D1CD00031696FF9400609AB0 /* CoreMedia.framework */,
-				D1CDFFFB1696FC0100609AB0 /* Theora.framework */,
-				D1CDFFF91696FBF700609AB0 /* Vorbis.framework */,
-				D1CDFFF71696FBF400609AB0 /* Ogg.framework */,
-				D1CDFFF01696FB8900609AB0 /* CoreVideo.framework */,
-				D1CDFFED1696FB7200609AB0 /* AVFoundation.framework */,
-				D1CDFFF21696FBA800609AB0 /* Foundation.framework */,
-			);
-			name = mac;
-			sourceTree = "<group>";
-		};
-		D147401F150CAE9600B20490 /* include */ = {
-			isa = PBXGroup;
-			children = (
-				D16775C1155C50280050EC64 /* TheoraAsync.h */,
-				D16775C2155C50280050EC64 /* TheoraAudioInterface.h */,
-				D16775C3155C50280050EC64 /* TheoraDataSource.h */,
-				D16775C4155C50280050EC64 /* TheoraException.h */,
-				D16775C5155C50280050EC64 /* TheoraExport.h */,
-				D1E271B216B471E80046C00C /* TheoraPixelTransform.h */,
-				D16775CB155C50280050EC64 /* TheoraVideoFrame.h */,
-				D16775C6155C50280050EC64 /* TheoraFrameQueue.h */,
-				D16775C7155C50280050EC64 /* TheoraPlayer.h */,
-				D16775C8155C50280050EC64 /* TheoraTimer.h */,
-				D16775C9155C50280050EC64 /* TheoraUtil.h */,
-				D16775CA155C50280050EC64 /* TheoraVideoClip.h */,
-				D16775CC155C50280050EC64 /* TheoraVideoManager.h */,
-				D1D465D516C2D063007A45AA /* TheoraAudioPacketQueue.h */,
-				D16775CD155C50280050EC64 /* TheoraWorkerThread.h */,
-			);
-			name = include;
-			sourceTree = "<group>";
-		};
-		D1C3D04D17C157A800CA0FD2 /* src */ = {
-			isa = PBXGroup;
-			children = (
-				D1C3D04F17C157CD00CA0FD2 /* compare_common.cc */,
-				D1C3D05017C157CD00CA0FD2 /* compare_neon.cc */,
-				D1C3D05117C157CD00CA0FD2 /* compare_posix.cc */,
-				D159BCAB17C227940030FAB6 /* compare_win.cc */,
-				D1C3D05317C157CD00CA0FD2 /* compare.cc */,
-				D1C3D05417C157CD00CA0FD2 /* convert_argb.cc */,
-				D1C3D05517C157CD00CA0FD2 /* convert_from_argb.cc */,
-				D1C3D05617C157CD00CA0FD2 /* convert_from.cc */,
-				D1C3D05717C157CD00CA0FD2 /* convert_jpeg.cc */,
-				D1C3D05817C157CD00CA0FD2 /* convert_to_argb.cc */,
-				D1C3D05917C157CD00CA0FD2 /* convert_to_i420.cc */,
-				D1C3D05A17C157CD00CA0FD2 /* convert.cc */,
-				D1C3D05B17C157CD00CA0FD2 /* cpu_id.cc */,
-				D1C3D05C17C157CD00CA0FD2 /* format_conversion.cc */,
-				D1C3D05D17C157CD00CA0FD2 /* mjpeg_decoder.cc */,
-				D1C3D05E17C157CD00CA0FD2 /* mjpeg_validate.cc */,
-				D1C3D05F17C157CD00CA0FD2 /* planar_functions.cc */,
-				D1C3D06017C157CD00CA0FD2 /* rotate_argb.cc */,
-				D1C3D06117C157CD00CA0FD2 /* rotate_mips.cc */,
-				D1C3D06217C157CD00CA0FD2 /* rotate_neon.cc */,
-				D1C3D06317C157CD00CA0FD2 /* rotate.cc */,
-				D1C3D06417C157CD00CA0FD2 /* row_any.cc */,
-				D1C3D06517C157CD00CA0FD2 /* row_common.cc */,
-				D1C3D06617C157CD00CA0FD2 /* row_mips.cc */,
-				D1C3D06717C157CD00CA0FD2 /* row_neon.cc */,
-				D1C3D06817C157CD00CA0FD2 /* row_posix.cc */,
-				D159BCAC17C227940030FAB6 /* row_win.cc */,
-				D1C3D06B17C157CD00CA0FD2 /* scale_argb_neon.cc */,
-				D1C3D06C17C157CD00CA0FD2 /* scale_argb.cc */,
-				D1C3D06D17C157CD00CA0FD2 /* scale_mips.cc */,
-				D1C3D06E17C157CD00CA0FD2 /* scale_neon.cc */,
-				D1BCE06C18F3F80800C83470 /* scale_win.cc */,
-				D1C3D06F17C157CD00CA0FD2 /* scale.cc */,
-				D1BCE05818F3F7FE00C83470 /* scale_common.cc */,
-				D1BCE05918F3F7FE00C83470 /* scale_posix.cc */,
-				D1C3D07017C157CD00CA0FD2 /* video_common.cc */,
-				D159BCAD17C227940030FAB6 /* row_x86.asm */,
-				D159BCAE17C227940030FAB6 /* x86inc.asm */,
-			);
-			name = src;
-			sourceTree = "<group>";
-		};
-		D1C3D04E17C157AC00CA0FD2 /* include */ = {
-			isa = PBXGroup;
-			children = (
-				D1C3D1CD17C15BA900CA0FD2 /* libyuv */,
-				D1C3D1CE17C15BB400CA0FD2 /* libyuv.h */,
-			);
-			name = include;
-			sourceTree = "<group>";
-		};
-		D1C3D1CD17C15BA900CA0FD2 /* libyuv */ = {
-			isa = PBXGroup;
-			children = (
-				D1C3D1CF17C15BC100CA0FD2 /* basic_types.h */,
-				D1C3D1D017C15BC100CA0FD2 /* compare.h */,
-				D1C3D1D117C15BC100CA0FD2 /* convert_argb.h */,
-				D1C3D1D217C15BC100CA0FD2 /* convert_from_argb.h */,
-				D1C3D1D317C15BC100CA0FD2 /* convert_from.h */,
-				D1C3D1D417C15BC100CA0FD2 /* convert.h */,
-				D1C3D1D517C15BC100CA0FD2 /* cpu_id.h */,
-				D1C3D1D617C15BC100CA0FD2 /* format_conversion.h */,
-				D1C3D1D717C15BC100CA0FD2 /* mjpeg_decoder.h */,
-				D1C3D1D817C15BC100CA0FD2 /* planar_functions.h */,
-				D1C3D1D917C15BC100CA0FD2 /* rotate_argb.h */,
-				D1C3D1DA17C15BC100CA0FD2 /* rotate.h */,
-				D1C3D1DB17C15BC100CA0FD2 /* row.h */,
-				D1C3D1DC17C15BC100CA0FD2 /* scale_argb.h */,
-				D1BCE05718F3F7D800C83470 /* scale_row.h */,
-				D1C3D1DD17C15BC100CA0FD2 /* scale.h */,
-				D1C3D1DE17C15BC100CA0FD2 /* version.h */,
-				D1C3D1DF17C15BC100CA0FD2 /* video_common.h */,
-			);
-			name = libyuv;
-			sourceTree = "<group>";
-		};
-		D1CDFF921696CEFA00609AB0 /* Theora */ = {
-			isa = PBXGroup;
-			children = (
-				D1CDFF951696D0F000609AB0 /* TheoraVideoClip_Theora.h */,
-				D1CDFF941696D0F000609AB0 /* TheoraVideoClip_Theora.cpp */,
-			);
-			name = Theora;
-			sourceTree = "<group>";
-		};
-		D1CDFF931696CF0000609AB0 /* AVFoundation */ = {
-			isa = PBXGroup;
-			children = (
-				D1CDFF9D1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h */,
-				D1CDFF9C1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm */,
-			);
-			name = AVFoundation;
-			sourceTree = "<group>";
-		};
-		D1E2718516B46F370046C00C /* YUV */ = {
-			isa = PBXGroup;
-			children = (
-				D1F4DA1C18FECABC007C1968 /* android */,
-				D139462A17C0ED2F0091F4A4 /* libyuv */,
-				D1E2718716B46F4F0046C00C /* C */,
-				D13946CA17C119B30091F4A4 /* yuv_util.c */,
-				D13946CB17C119B30091F4A4 /* yuv_util.h */,
-			);
-			name = YUV;
-			sourceTree = "<group>";
-		};
-		D1E2718716B46F4F0046C00C /* C */ = {
-			isa = PBXGroup;
-			children = (
-				D1E271AB16B470210046C00C /* yuv420_rgb_c.c */,
-				D1E2718C16B46F640046C00C /* yuv420_yuv_c.c */,
-				D1E2718A16B46F640046C00C /* yuv420_grey_c.c */,
-			);
-			name = C;
-			sourceTree = "<group>";
-		};
-		D1F4DA1C18FECABC007C1968 /* android */ = {
-			isa = PBXGroup;
-			children = (
-				D1F4DA1D18FECACE007C1968 /* cpu-features.c */,
-				D1F4DA1E18FECACE007C1968 /* cpu-features.h */,
-			);
-			name = android;
-			sourceTree = "<group>";
-		};
-/* End PBXGroup section */
-
-/* Begin PBXHeadersBuildPhase section */
-		D1473F27150CA69B00B20490 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D16775CE155C50280050EC64 /* TheoraAsync.h in Headers */,
-				D16775D0155C50280050EC64 /* TheoraAudioInterface.h in Headers */,
-				D16775D2155C50280050EC64 /* TheoraDataSource.h in Headers */,
-				D16775D4155C50280050EC64 /* TheoraException.h in Headers */,
-				D16775D6155C50280050EC64 /* TheoraExport.h in Headers */,
-				D16775D8155C50280050EC64 /* TheoraFrameQueue.h in Headers */,
-				D16775DA155C50280050EC64 /* TheoraPlayer.h in Headers */,
-				D16775DC155C50280050EC64 /* TheoraTimer.h in Headers */,
-				D16775DE155C50280050EC64 /* TheoraUtil.h in Headers */,
-				D16775E0155C50280050EC64 /* TheoraVideoClip.h in Headers */,
-				D16775E2155C50280050EC64 /* TheoraVideoFrame.h in Headers */,
-				D16775E4155C50280050EC64 /* TheoraVideoManager.h in Headers */,
-				D16775E6155C50280050EC64 /* TheoraWorkerThread.h in Headers */,
-				D1D465D616C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */,
-				D1E271B316B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D1CDFF991696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */,
-				D139463617C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946D517C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F964177A31FC002942E3 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F965177A31FC002942E3 /* TheoraAsync.h in Headers */,
-				D198F966177A31FC002942E3 /* TheoraAudioInterface.h in Headers */,
-				D198F967177A31FC002942E3 /* TheoraDataSource.h in Headers */,
-				D198F968177A31FC002942E3 /* TheoraException.h in Headers */,
-				D198F969177A31FC002942E3 /* TheoraExport.h in Headers */,
-				D198F96A177A31FC002942E3 /* TheoraFrameQueue.h in Headers */,
-				D198F96B177A31FC002942E3 /* TheoraPlayer.h in Headers */,
-				D198F96C177A31FC002942E3 /* TheoraTimer.h in Headers */,
-				D198F96D177A31FC002942E3 /* TheoraUtil.h in Headers */,
-				D198F96E177A31FC002942E3 /* TheoraVideoClip.h in Headers */,
-				D198F96F177A31FC002942E3 /* TheoraVideoFrame.h in Headers */,
-				D198F970177A31FC002942E3 /* TheoraVideoManager.h in Headers */,
-				D198F971177A31FC002942E3 /* TheoraWorkerThread.h in Headers */,
-				D198F972177A31FC002942E3 /* TheoraVideoClip_Theora.h in Headers */,
-				D198F974177A31FC002942E3 /* TheoraPixelTransform.h in Headers */,
-				D139463917C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946D817C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F990177A31FE002942E3 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F991177A31FE002942E3 /* TheoraAsync.h in Headers */,
-				D198F992177A31FE002942E3 /* TheoraAudioInterface.h in Headers */,
-				D198F993177A31FE002942E3 /* TheoraDataSource.h in Headers */,
-				D198F994177A31FE002942E3 /* TheoraException.h in Headers */,
-				D198F995177A31FE002942E3 /* TheoraExport.h in Headers */,
-				D198F996177A31FE002942E3 /* TheoraFrameQueue.h in Headers */,
-				D198F997177A31FE002942E3 /* TheoraPlayer.h in Headers */,
-				D198F998177A31FE002942E3 /* TheoraTimer.h in Headers */,
-				D198F999177A31FE002942E3 /* TheoraUtil.h in Headers */,
-				D198F99A177A31FE002942E3 /* TheoraVideoClip.h in Headers */,
-				D198F99B177A31FE002942E3 /* TheoraVideoFrame.h in Headers */,
-				D198F99C177A31FE002942E3 /* TheoraVideoManager.h in Headers */,
-				D198F99D177A31FE002942E3 /* TheoraWorkerThread.h in Headers */,
-				D198F99E177A31FE002942E3 /* TheoraVideoClip_Theora.h in Headers */,
-				D198F9A0177A31FE002942E3 /* TheoraPixelTransform.h in Headers */,
-				D139463A17C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946D917C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F9BD177A3200002942E3 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F9BE177A3200002942E3 /* TheoraAsync.h in Headers */,
-				D198F9BF177A3200002942E3 /* TheoraAudioInterface.h in Headers */,
-				D198F9C0177A3200002942E3 /* TheoraDataSource.h in Headers */,
-				D198F9C1177A3200002942E3 /* TheoraException.h in Headers */,
-				D198F9C2177A3200002942E3 /* TheoraExport.h in Headers */,
-				D198F9C3177A3200002942E3 /* TheoraFrameQueue.h in Headers */,
-				D198F9C4177A3200002942E3 /* TheoraPlayer.h in Headers */,
-				D198F9C5177A3200002942E3 /* TheoraTimer.h in Headers */,
-				D198F9C6177A3200002942E3 /* TheoraUtil.h in Headers */,
-				D198F9C7177A3200002942E3 /* TheoraVideoClip.h in Headers */,
-				D198F9C8177A3200002942E3 /* TheoraVideoFrame.h in Headers */,
-				D198F9C9177A3200002942E3 /* TheoraVideoManager.h in Headers */,
-				D198F9CA177A3200002942E3 /* TheoraWorkerThread.h in Headers */,
-				D198F9CB177A3200002942E3 /* TheoraVideoClip_Theora.h in Headers */,
-				D198F9CD177A3200002942E3 /* TheoraPixelTransform.h in Headers */,
-				D139463B17C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946DA17C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1BB6FAC150E9E7100EF9400 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D16775CF155C50280050EC64 /* TheoraAsync.h in Headers */,
-				D16775D1155C50280050EC64 /* TheoraAudioInterface.h in Headers */,
-				D16775D3155C50280050EC64 /* TheoraDataSource.h in Headers */,
-				D16775D5155C50280050EC64 /* TheoraException.h in Headers */,
-				D16775D7155C50280050EC64 /* TheoraExport.h in Headers */,
-				D16775D9155C50280050EC64 /* TheoraFrameQueue.h in Headers */,
-				D16775DB155C50280050EC64 /* TheoraPlayer.h in Headers */,
-				D16775DD155C50280050EC64 /* TheoraTimer.h in Headers */,
-				D16775DF155C50280050EC64 /* TheoraUtil.h in Headers */,
-				D16775E1155C50280050EC64 /* TheoraVideoClip.h in Headers */,
-				D16775E3155C50280050EC64 /* TheoraVideoFrame.h in Headers */,
-				D16775E5155C50280050EC64 /* TheoraVideoManager.h in Headers */,
-				D16775E7155C50280050EC64 /* TheoraWorkerThread.h in Headers */,
-				D1CDFF9B1696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */,
-				D1E271B616B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D139463C17C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946DB17C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF331696C77A00609AB0 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFF341696C77A00609AB0 /* TheoraAsync.h in Headers */,
-				D1CDFF351696C77A00609AB0 /* TheoraAudioInterface.h in Headers */,
-				D1CDFF361696C77A00609AB0 /* TheoraDataSource.h in Headers */,
-				D1CDFF371696C77A00609AB0 /* TheoraException.h in Headers */,
-				D1CDFF381696C77A00609AB0 /* TheoraExport.h in Headers */,
-				D1CDFF391696C77A00609AB0 /* TheoraFrameQueue.h in Headers */,
-				D1CDFF3A1696C77A00609AB0 /* TheoraPlayer.h in Headers */,
-				D1CDFF3B1696C77A00609AB0 /* TheoraTimer.h in Headers */,
-				D1CDFF3C1696C77A00609AB0 /* TheoraUtil.h in Headers */,
-				D1CDFF3D1696C77A00609AB0 /* TheoraVideoClip.h in Headers */,
-				D1CDFF3E1696C77A00609AB0 /* TheoraVideoFrame.h in Headers */,
-				D1CDFF3F1696C77A00609AB0 /* TheoraVideoManager.h in Headers */,
-				D1CDFF401696C77A00609AB0 /* TheoraWorkerThread.h in Headers */,
-				D1E271B416B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D1D465D716C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */,
-				D1CDFF9F1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.h in Headers */,
-				D139463717C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946D617C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF5B1696C79700609AB0 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFF5C1696C79700609AB0 /* TheoraAsync.h in Headers */,
-				D1CDFF5D1696C79700609AB0 /* TheoraAudioInterface.h in Headers */,
-				D1CDFF5E1696C79700609AB0 /* TheoraDataSource.h in Headers */,
-				D1CDFF5F1696C79700609AB0 /* TheoraException.h in Headers */,
-				D1CDFF601696C79700609AB0 /* TheoraExport.h in Headers */,
-				D1CDFF611696C79700609AB0 /* TheoraFrameQueue.h in Headers */,
-				D1CDFF621696C79700609AB0 /* TheoraPlayer.h in Headers */,
-				D1CDFF631696C79700609AB0 /* TheoraTimer.h in Headers */,
-				D1CDFF641696C79700609AB0 /* TheoraUtil.h in Headers */,
-				D1CDFF651696C79700609AB0 /* TheoraVideoClip.h in Headers */,
-				D1CDFF661696C79700609AB0 /* TheoraVideoFrame.h in Headers */,
-				D1CDFF671696C79700609AB0 /* TheoraVideoManager.h in Headers */,
-				D1CDFF681696C79700609AB0 /* TheoraWorkerThread.h in Headers */,
-				D1E271B516B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D1D465D816C2D063007A45AA /* TheoraAudioPacketQueue.h in Headers */,
-				D1CDFF9A1696D0F000609AB0 /* TheoraVideoClip_Theora.h in Headers */,
-				D1F09EB1169AFEFB00DEEC63 /* TheoraVideoClip_AVFoundation.h in Headers */,
-				D139463817C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946D717C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFAF1696E1CA00609AB0 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFFB01696E1CA00609AB0 /* TheoraAsync.h in Headers */,
-				D1CDFFB11696E1CA00609AB0 /* TheoraAudioInterface.h in Headers */,
-				D1CDFFB21696E1CA00609AB0 /* TheoraDataSource.h in Headers */,
-				D1CDFFB31696E1CA00609AB0 /* TheoraException.h in Headers */,
-				D1CDFFB41696E1CA00609AB0 /* TheoraExport.h in Headers */,
-				D1CDFFB51696E1CA00609AB0 /* TheoraFrameQueue.h in Headers */,
-				D1CDFFB61696E1CA00609AB0 /* TheoraPlayer.h in Headers */,
-				D1CDFFB71696E1CA00609AB0 /* TheoraTimer.h in Headers */,
-				D1CDFFB81696E1CA00609AB0 /* TheoraUtil.h in Headers */,
-				D1CDFFB91696E1CA00609AB0 /* TheoraVideoClip.h in Headers */,
-				D1CDFFBA1696E1CA00609AB0 /* TheoraVideoFrame.h in Headers */,
-				D1CDFFBB1696E1CA00609AB0 /* TheoraVideoManager.h in Headers */,
-				D1CDFFBC1696E1CA00609AB0 /* TheoraWorkerThread.h in Headers */,
-				D1CDFFBD1696E1CA00609AB0 /* TheoraVideoClip_Theora.h in Headers */,
-				D1E271B716B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D139463D17C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946DC17C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFD41696E1D700609AB0 /* Headers */ = {
-			isa = PBXHeadersBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFFD51696E1D700609AB0 /* TheoraAsync.h in Headers */,
-				D1CDFFD61696E1D700609AB0 /* TheoraAudioInterface.h in Headers */,
-				D1CDFFD71696E1D700609AB0 /* TheoraDataSource.h in Headers */,
-				D1CDFFD81696E1D700609AB0 /* TheoraException.h in Headers */,
-				D1CDFFD91696E1D700609AB0 /* TheoraExport.h in Headers */,
-				D1CDFFDA1696E1D700609AB0 /* TheoraFrameQueue.h in Headers */,
-				D1CDFFDB1696E1D700609AB0 /* TheoraPlayer.h in Headers */,
-				D1CDFFDC1696E1D700609AB0 /* TheoraTimer.h in Headers */,
-				D1CDFFDD1696E1D700609AB0 /* TheoraUtil.h in Headers */,
-				D1CDFFDE1696E1D700609AB0 /* TheoraVideoClip.h in Headers */,
-				D1CDFFDF1696E1D700609AB0 /* TheoraVideoFrame.h in Headers */,
-				D1CDFFE01696E1D700609AB0 /* TheoraVideoManager.h in Headers */,
-				D1CDFFE11696E1D700609AB0 /* TheoraWorkerThread.h in Headers */,
-				D1CDFFE21696E1D700609AB0 /* TheoraVideoClip_Theora.h in Headers */,
-				D1E271B816B471E80046C00C /* TheoraPixelTransform.h in Headers */,
-				D139463E17C0ED450091F4A4 /* yuv_libyuv.h in Headers */,
-				D13946DD17C119B40091F4A4 /* yuv_util.h in Headers */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXHeadersBuildPhase section */
-
-/* Begin PBXNativeTarget section */
-		D1473F29150CA69B00B20490 /* theoraplayer (Theora) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1473F3F150CA69B00B20490 /* Build configuration list for PBXNativeTarget "theoraplayer (Theora)" */;
-			buildPhases = (
-				D1473F25150CA69B00B20490 /* Sources */,
-				D1473F26150CA69B00B20490 /* Frameworks */,
-				D1473F27150CA69B00B20490 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (Theora)";
-			productName = theoraplayer;
-			productReference = D1473F2A150CA69B00B20490 /* theoraplayer.framework */;
-			productType = "com.apple.product-type.framework";
-		};
-		D198F950177A31FC002942E3 /* theoraplayer (Mac Theora) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D198F975177A31FC002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac Theora)" */;
-			buildPhases = (
-				D198F951177A31FC002942E3 /* Sources */,
-				D198F963177A31FC002942E3 /* Frameworks */,
-				D198F964177A31FC002942E3 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (Mac Theora)";
-			productName = libtheoraplayer;
-			productReference = D198F97B177A31FC002942E3 /* libtheoraplayer.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-		D198F97C177A31FE002942E3 /* theoraplayer (Mac AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D198F9A1177A31FE002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac AVFoundation)" */;
-			buildPhases = (
-				D198F97D177A31FE002942E3 /* Sources */,
-				D198F98F177A31FE002942E3 /* Frameworks */,
-				D198F990177A31FE002942E3 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (Mac AVFoundation)";
-			productName = libtheoraplayer;
-			productReference = D198F9A7177A31FE002942E3 /* libtheoraplayer_avfoundation.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-		D198F9A8177A3200002942E3 /* theoraplayer (Mac Theora AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D198F9CE177A3200002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac Theora AVFoundation)" */;
-			buildPhases = (
-				D198F9A9177A3200002942E3 /* Sources */,
-				D198F9BC177A3200002942E3 /* Frameworks */,
-				D198F9BD177A3200002942E3 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (Mac Theora AVFoundation)";
-			productName = libtheoraplayer;
-			productReference = D198F9D4177A3200002942E3 /* libtheoraplayer_theora_avfoundation.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-		D1BB6FAD150E9E7100EF9400 /* theoraplayer (iOS Theora) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1BB6FBC150E9E7100EF9400 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS Theora)" */;
-			buildPhases = (
-				D1BB6FAA150E9E7100EF9400 /* Sources */,
-				D1BB6FAB150E9E7100EF9400 /* Frameworks */,
-				D1BB6FAC150E9E7100EF9400 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (iOS Theora)";
-			productName = libtheoraplayer;
-			productReference = D1BB6FAE150E9E7100EF9400 /* libtheoraplayer.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-		D1CDFF221696C77A00609AB0 /* theoraplayer (AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1CDFF421696C77A00609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (AVFoundation)" */;
-			buildPhases = (
-				D1CDFF231696C77A00609AB0 /* Sources */,
-				D1CDFF2F1696C77A00609AB0 /* Frameworks */,
-				D1CDFF331696C77A00609AB0 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (AVFoundation)";
-			productName = theoraplayer;
-			productReference = D1CDFF481696C77A00609AB0 /* theoraplayer.framework */;
-			productType = "com.apple.product-type.framework";
-		};
-		D1CDFF4A1696C79700609AB0 /* theoraplayer (Theora AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1CDFF6A1696C79700609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (Theora AVFoundation)" */;
-			buildPhases = (
-				D1CDFF4B1696C79700609AB0 /* Sources */,
-				D1CDFF571696C79700609AB0 /* Frameworks */,
-				D1CDFF5B1696C79700609AB0 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (Theora AVFoundation)";
-			productName = theoraplayer;
-			productReference = D1CDFF701696C79700609AB0 /* theoraplayer.framework */;
-			productType = "com.apple.product-type.framework";
-		};
-		D1CDFFA01696E1CA00609AB0 /* theoraplayer (iOS AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1CDFFBE1696E1CA00609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS AVFoundation)" */;
-			buildPhases = (
-				D1CDFFA11696E1CA00609AB0 /* Sources */,
-				D1CDFFAE1696E1CA00609AB0 /* Frameworks */,
-				D1CDFFAF1696E1CA00609AB0 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (iOS AVFoundation)";
-			productName = libtheoraplayer;
-			productReference = D1CDFFC41696E1CA00609AB0 /* libtheoraplayer.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-		D1CDFFC51696E1D700609AB0 /* theoraplayer (iOS Theora AVFoundation) */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = D1CDFFE31696E1D700609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS Theora AVFoundation)" */;
-			buildPhases = (
-				D1CDFFC61696E1D700609AB0 /* Sources */,
-				D1CDFFD31696E1D700609AB0 /* Frameworks */,
-				D1CDFFD41696E1D700609AB0 /* Headers */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = "theoraplayer (iOS Theora AVFoundation)";
-			productName = libtheoraplayer;
-			productReference = D1CDFFE91696E1D700609AB0 /* libtheoraplayer.a */;
-			productType = "com.apple.product-type.library.static";
-		};
-/* End PBXNativeTarget section */
-
-/* Begin PBXProject section */
-		D1473F20150CA69B00B20490 /* Project object */ = {
-			isa = PBXProject;
-			attributes = {
-				LastUpgradeCheck = 0510;
-			};
-			buildConfigurationList = D1473F23150CA69B00B20490 /* Build configuration list for PBXProject "theoraplayer" */;
-			compatibilityVersion = "Xcode 3.2";
-			developmentRegion = English;
-			hasScannedForEncodings = 0;
-			knownRegions = (
-				en,
-			);
-			mainGroup = D1473F1E150CA69B00B20490;
-			productRefGroup = D1473F2B150CA69B00B20490 /* Products */;
-			projectDirPath = "";
-			projectRoot = "";
-			targets = (
-				D1473F29150CA69B00B20490 /* theoraplayer (Theora) */,
-				D1CDFF221696C77A00609AB0 /* theoraplayer (AVFoundation) */,
-				D1CDFF4A1696C79700609AB0 /* theoraplayer (Theora AVFoundation) */,
-				D198F950177A31FC002942E3 /* theoraplayer (Mac Theora) */,
-				D198F97C177A31FE002942E3 /* theoraplayer (Mac AVFoundation) */,
-				D198F9A8177A3200002942E3 /* theoraplayer (Mac Theora AVFoundation) */,
-				D1BB6FAD150E9E7100EF9400 /* theoraplayer (iOS Theora) */,
-				D1CDFFA01696E1CA00609AB0 /* theoraplayer (iOS AVFoundation) */,
-				D1CDFFC51696E1D700609AB0 /* theoraplayer (iOS Theora AVFoundation) */,
-			);
-		};
-/* End PBXProject section */
-
-/* Begin PBXSourcesBuildPhase section */
-		D1473F25150CA69B00B20490 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D16775AB155C501D0050EC64 /* TheoraAsync.cpp in Sources */,
-				D16775AD155C501D0050EC64 /* TheoraAudioInterface.cpp in Sources */,
-				D16775AF155C501D0050EC64 /* TheoraDataSource.cpp in Sources */,
-				D16775B1155C501D0050EC64 /* TheoraException.cpp in Sources */,
-				D16775B3155C501D0050EC64 /* TheoraFrameQueue.cpp in Sources */,
-				D16775B5155C501D0050EC64 /* TheoraTimer.cpp in Sources */,
-				D16775B7155C501D0050EC64 /* TheoraUtil.cpp in Sources */,
-				D16775B9155C501D0050EC64 /* TheoraVideoClip.cpp in Sources */,
-				D16775BB155C501D0050EC64 /* TheoraVideoFrame.cpp in Sources */,
-				D16775BD155C501D0050EC64 /* TheoraVideoManager.cpp in Sources */,
-				D16775BF155C501D0050EC64 /* TheoraWorkerThread.cpp in Sources */,
-				D1CDFF961696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D1E2719916B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271A516B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271AC16B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D1BCE05A18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D1D465DA16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139462D17C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946CC17C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07217C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08417C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1BCE06318F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1C3D09617C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D09F17C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C317C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0CC17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0D517C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0DE17C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0E717C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10217C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12617C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D12F17C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13817C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15317C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17717C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19B17C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB017C227F30030FAB6 /* convert_from.cc in Sources */,
-				D159BCB917C228310030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC217C2286D0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F951177A31FC002942E3 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F952177A31FC002942E3 /* TheoraAsync.cpp in Sources */,
-				D198F953177A31FC002942E3 /* TheoraAudioInterface.cpp in Sources */,
-				D198F954177A31FC002942E3 /* TheoraDataSource.cpp in Sources */,
-				D198F955177A31FC002942E3 /* TheoraException.cpp in Sources */,
-				D198F956177A31FC002942E3 /* TheoraFrameQueue.cpp in Sources */,
-				D198F957177A31FC002942E3 /* TheoraTimer.cpp in Sources */,
-				D198F958177A31FC002942E3 /* TheoraUtil.cpp in Sources */,
-				D198F959177A31FC002942E3 /* TheoraVideoClip.cpp in Sources */,
-				D198F95A177A31FC002942E3 /* TheoraVideoFrame.cpp in Sources */,
-				D198F95B177A31FC002942E3 /* TheoraVideoManager.cpp in Sources */,
-				D198F95C177A31FC002942E3 /* TheoraWorkerThread.cpp in Sources */,
-				D198F95D177A31FC002942E3 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D198F95F177A31FC002942E3 /* yuv420_grey_c.c in Sources */,
-				D198F960177A31FC002942E3 /* yuv420_yuv_c.c in Sources */,
-				D198F961177A31FC002942E3 /* yuv420_rgb_c.c in Sources */,
-				D1BCE05D18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D198F962177A31FC002942E3 /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139463017C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946CF17C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07517C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08717C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1BCE06618F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1C3D09917C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A217C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C617C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0CF17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0D817C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E117C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0EA17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10517C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12917C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13217C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13B17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15617C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17A17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19E17C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB317C227F40030FAB6 /* convert_from.cc in Sources */,
-				D159BCBC17C228330030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC517C2286E0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F97D177A31FE002942E3 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F97E177A31FE002942E3 /* TheoraAsync.cpp in Sources */,
-				D198F97F177A31FE002942E3 /* TheoraAudioInterface.cpp in Sources */,
-				D198F980177A31FE002942E3 /* TheoraDataSource.cpp in Sources */,
-				D198F981177A31FE002942E3 /* TheoraException.cpp in Sources */,
-				D198F982177A31FE002942E3 /* TheoraFrameQueue.cpp in Sources */,
-				D198F983177A31FE002942E3 /* TheoraTimer.cpp in Sources */,
-				D198F984177A31FE002942E3 /* TheoraUtil.cpp in Sources */,
-				D198F985177A31FE002942E3 /* TheoraVideoClip.cpp in Sources */,
-				D198F986177A31FE002942E3 /* TheoraVideoFrame.cpp in Sources */,
-				D198F987177A31FE002942E3 /* TheoraVideoManager.cpp in Sources */,
-				D198F988177A31FE002942E3 /* TheoraWorkerThread.cpp in Sources */,
-				D198F989177A31FE002942E3 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D198F98B177A31FE002942E3 /* yuv420_grey_c.c in Sources */,
-				D198F98C177A31FE002942E3 /* yuv420_yuv_c.c in Sources */,
-				D198F98D177A31FE002942E3 /* yuv420_rgb_c.c in Sources */,
-				D1BCE05E18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D198F98E177A31FE002942E3 /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139463117C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946D017C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07617C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08817C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1BCE06718F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1C3D09A17C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A317C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C717C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0D017C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0D917C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E217C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0EB17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10617C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12A17C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13317C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13C17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15717C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17B17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19F17C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB417C227F50030FAB6 /* convert_from.cc in Sources */,
-				D159BCBD17C228330030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC617C2286E0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D198F9A9177A3200002942E3 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D198F9AA177A3200002942E3 /* TheoraAsync.cpp in Sources */,
-				D198F9AB177A3200002942E3 /* TheoraAudioInterface.cpp in Sources */,
-				D198F9AC177A3200002942E3 /* TheoraDataSource.cpp in Sources */,
-				D198F9AD177A3200002942E3 /* TheoraException.cpp in Sources */,
-				D198F9AE177A3200002942E3 /* TheoraFrameQueue.cpp in Sources */,
-				D198F9AF177A3200002942E3 /* TheoraTimer.cpp in Sources */,
-				D198F9B0177A3200002942E3 /* TheoraUtil.cpp in Sources */,
-				D198F9B1177A3200002942E3 /* TheoraVideoClip.cpp in Sources */,
-				D198F9B2177A3200002942E3 /* TheoraVideoFrame.cpp in Sources */,
-				D198F9B3177A3200002942E3 /* TheoraVideoManager.cpp in Sources */,
-				D198F9B4177A3200002942E3 /* TheoraWorkerThread.cpp in Sources */,
-				D198F9B5177A3200002942E3 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D1BCE06818F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D198F9B6177A3200002942E3 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D198F9B8177A3200002942E3 /* yuv420_grey_c.c in Sources */,
-				D198F9B9177A3200002942E3 /* yuv420_yuv_c.c in Sources */,
-				D198F9BA177A3200002942E3 /* yuv420_rgb_c.c in Sources */,
-				D198F9BB177A3200002942E3 /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139463217C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946D117C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07717C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08917C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1C3D09B17C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A417C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C817C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0D117C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0DA17C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E317C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0EC17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10717C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12B17C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13417C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13D17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15817C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17C17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D1A017C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB517C227F50030FAB6 /* convert_from.cc in Sources */,
-				D159BCBE17C228340030FAB6 /* rotate_argb.cc in Sources */,
-				D1BCE05F18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D159BCC717C2286F0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1BB6FAA150E9E7100EF9400 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D16775AC155C501D0050EC64 /* TheoraAsync.cpp in Sources */,
-				D1BCE06018F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D16775AE155C501D0050EC64 /* TheoraAudioInterface.cpp in Sources */,
-				D16775B0155C501D0050EC64 /* TheoraDataSource.cpp in Sources */,
-				D16775B2155C501D0050EC64 /* TheoraException.cpp in Sources */,
-				D16775B4155C501D0050EC64 /* TheoraFrameQueue.cpp in Sources */,
-				D16775B6155C501D0050EC64 /* TheoraTimer.cpp in Sources */,
-				D16775B8155C501D0050EC64 /* TheoraUtil.cpp in Sources */,
-				D16775BA155C501D0050EC64 /* TheoraVideoClip.cpp in Sources */,
-				D16775BC155C501D0050EC64 /* TheoraVideoFrame.cpp in Sources */,
-				D16775BE155C501D0050EC64 /* TheoraVideoManager.cpp in Sources */,
-				D16775C0155C501D0050EC64 /* TheoraWorkerThread.cpp in Sources */,
-				D1BCE06918F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1CDFF981696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D1E2719C16B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271A816B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271AF16B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D139463317C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D1D465DD16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D13946D217C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07817C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08117C157CD00CA0FD2 /* compare_neon.cc in Sources */,
-				D1C3D09C17C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A517C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C917C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0D217C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0DB17C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E417C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0ED17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10817C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12317C157CD00CA0FD2 /* rotate_neon.cc in Sources */,
-				D1C3D12C17C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13517C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13E17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15017C157CD00CA0FD2 /* row_neon.cc in Sources */,
-				D1C3D17417C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */,
-				D1C3D17D17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D18F17C157CD00CA0FD2 /* scale_neon.cc in Sources */,
-				D1C3D1A117C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB617C227F60030FAB6 /* convert_from.cc in Sources */,
-				D159BCBF17C228340030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC817C2286F0030FAB6 /* scale.cc in Sources */,
-				D15D361017C386A600F40439 /* row_posix.cc in Sources */,
-				D15D361317C386B100F40439 /* compare_posix.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF231696C77A00609AB0 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFF241696C77A00609AB0 /* TheoraAsync.cpp in Sources */,
-				D1CDFF251696C77A00609AB0 /* TheoraAudioInterface.cpp in Sources */,
-				D1CDFF261696C77A00609AB0 /* TheoraDataSource.cpp in Sources */,
-				D1CDFF271696C77A00609AB0 /* TheoraException.cpp in Sources */,
-				D1CDFF281696C77A00609AB0 /* TheoraFrameQueue.cpp in Sources */,
-				D1CDFF291696C77A00609AB0 /* TheoraTimer.cpp in Sources */,
-				D1CDFF2A1696C77A00609AB0 /* TheoraUtil.cpp in Sources */,
-				D1CDFF2B1696C77A00609AB0 /* TheoraVideoClip.cpp in Sources */,
-				D1CDFF2C1696C77A00609AB0 /* TheoraVideoFrame.cpp in Sources */,
-				D1CDFF2D1696C77A00609AB0 /* TheoraVideoManager.cpp in Sources */,
-				D1CDFF2E1696C77A00609AB0 /* TheoraWorkerThread.cpp in Sources */,
-				D1CDFF9E1696D0FA00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D1E2719A16B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271A616B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271AD16B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D1BCE05B18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D1D465DB16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139462E17C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946CD17C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07317C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08517C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1BCE06418F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1C3D09717C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A017C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C417C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0CD17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0D617C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0DF17C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0E817C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10317C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12717C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13017C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13917C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15417C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17817C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19C17C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB117C227F40030FAB6 /* convert_from.cc in Sources */,
-				D159BCBA17C228320030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC317C2286D0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFF4B1696C79700609AB0 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFF4C1696C79700609AB0 /* TheoraAsync.cpp in Sources */,
-				D1CDFF4D1696C79700609AB0 /* TheoraAudioInterface.cpp in Sources */,
-				D1CDFF4E1696C79700609AB0 /* TheoraDataSource.cpp in Sources */,
-				D1CDFF4F1696C79700609AB0 /* TheoraException.cpp in Sources */,
-				D1CDFF501696C79700609AB0 /* TheoraFrameQueue.cpp in Sources */,
-				D1CDFF511696C79700609AB0 /* TheoraTimer.cpp in Sources */,
-				D1CDFF521696C79700609AB0 /* TheoraUtil.cpp in Sources */,
-				D1CDFF531696C79700609AB0 /* TheoraVideoClip.cpp in Sources */,
-				D1CDFF541696C79700609AB0 /* TheoraVideoFrame.cpp in Sources */,
-				D1CDFF551696C79700609AB0 /* TheoraVideoManager.cpp in Sources */,
-				D1CDFF561696C79700609AB0 /* TheoraWorkerThread.cpp in Sources */,
-				D1CDFF971696D0F000609AB0 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D1BCE06518F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1CDFFEC1696E24F00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D1E2719B16B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271A716B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271AE16B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D1D465DC16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D139462F17C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D13946CE17C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07417C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08617C157CD00CA0FD2 /* compare_posix.cc in Sources */,
-				D1C3D09817C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A117C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0C517C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0CE17C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0D717C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E017C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0E917C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10417C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12817C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13117C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13A17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15517C157CD00CA0FD2 /* row_posix.cc in Sources */,
-				D1C3D17917C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19D17C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB217C227F40030FAB6 /* convert_from.cc in Sources */,
-				D159BCBB17C228320030FAB6 /* rotate_argb.cc in Sources */,
-				D1BCE05C18F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D159BCC417C2286D0030FAB6 /* scale.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFA11696E1CA00609AB0 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFFA21696E1CA00609AB0 /* TheoraAsync.cpp in Sources */,
-				D1BCE06118F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D1CDFFA31696E1CA00609AB0 /* TheoraAudioInterface.cpp in Sources */,
-				D1CDFFA41696E1CA00609AB0 /* TheoraDataSource.cpp in Sources */,
-				D1CDFFA51696E1CA00609AB0 /* TheoraException.cpp in Sources */,
-				D1CDFFA61696E1CA00609AB0 /* TheoraFrameQueue.cpp in Sources */,
-				D1CDFFA71696E1CA00609AB0 /* TheoraTimer.cpp in Sources */,
-				D1CDFFA81696E1CA00609AB0 /* TheoraUtil.cpp in Sources */,
-				D1CDFFA91696E1CA00609AB0 /* TheoraVideoClip.cpp in Sources */,
-				D1CDFFAA1696E1CA00609AB0 /* TheoraVideoFrame.cpp in Sources */,
-				D1CDFFAB1696E1CA00609AB0 /* TheoraVideoManager.cpp in Sources */,
-				D1CDFFAC1696E1CA00609AB0 /* TheoraWorkerThread.cpp in Sources */,
-				D1BCE06A18F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D1CDFFEA1696E24B00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D1E2719D16B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271A916B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271B016B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D139463417C0ED450091F4A4 /* yuv_libyuv.c in Sources */,
-				D1D465DE16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D13946D317C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07917C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08217C157CD00CA0FD2 /* compare_neon.cc in Sources */,
-				D1C3D09D17C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A617C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0CA17C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0D317C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0DC17C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E517C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0EE17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10917C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12417C157CD00CA0FD2 /* rotate_neon.cc in Sources */,
-				D1C3D12D17C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13617C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D13F17C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15117C157CD00CA0FD2 /* row_neon.cc in Sources */,
-				D1C3D17517C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */,
-				D1C3D17E17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19017C157CD00CA0FD2 /* scale_neon.cc in Sources */,
-				D1C3D1A217C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB717C227F60030FAB6 /* convert_from.cc in Sources */,
-				D159BCC017C228340030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCC917C2286F0030FAB6 /* scale.cc in Sources */,
-				D15D361117C386A600F40439 /* row_posix.cc in Sources */,
-				D15D361617C386B400F40439 /* compare_posix.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		D1CDFFC61696E1D700609AB0 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				D1CDFFC71696E1D700609AB0 /* TheoraAsync.cpp in Sources */,
-				D1CDFFC81696E1D700609AB0 /* TheoraAudioInterface.cpp in Sources */,
-				D1CDFFC91696E1D700609AB0 /* TheoraDataSource.cpp in Sources */,
-				D1CDFFCA1696E1D700609AB0 /* TheoraException.cpp in Sources */,
-				D1CDFFCB1696E1D700609AB0 /* TheoraFrameQueue.cpp in Sources */,
-				D1CDFFCC1696E1D700609AB0 /* TheoraTimer.cpp in Sources */,
-				D1CDFFCD1696E1D700609AB0 /* TheoraUtil.cpp in Sources */,
-				D1CDFFCE1696E1D700609AB0 /* TheoraVideoClip.cpp in Sources */,
-				D1CDFFCF1696E1D700609AB0 /* TheoraVideoFrame.cpp in Sources */,
-				D1CDFFD01696E1D700609AB0 /* TheoraVideoManager.cpp in Sources */,
-				D1CDFFD11696E1D700609AB0 /* TheoraWorkerThread.cpp in Sources */,
-				D1CDFFD21696E1D700609AB0 /* TheoraVideoClip_Theora.cpp in Sources */,
-				D1CDFFEB1696E24C00609AB0 /* TheoraVideoClip_AVFoundation.mm in Sources */,
-				D1E2719E16B46F640046C00C /* yuv420_grey_c.c in Sources */,
-				D1E271AA16B46F640046C00C /* yuv420_yuv_c.c in Sources */,
-				D1E271B116B470210046C00C /* yuv420_rgb_c.c in Sources */,
-				D13946C617C110670091F4A4 /* yuv_libyuv.c in Sources */,
-				D1D465DF16C2D070007A45AA /* TheoraAudioPacketQueue.cpp in Sources */,
-				D13946D417C119B40091F4A4 /* yuv_util.c in Sources */,
-				D1C3D07A17C157CD00CA0FD2 /* compare_common.cc in Sources */,
-				D1C3D08317C157CD00CA0FD2 /* compare_neon.cc in Sources */,
-				D1C3D09E17C157CD00CA0FD2 /* compare.cc in Sources */,
-				D1C3D0A717C157CD00CA0FD2 /* convert_argb.cc in Sources */,
-				D1C3D0CB17C157CD00CA0FD2 /* convert_to_argb.cc in Sources */,
-				D1C3D0D417C157CD00CA0FD2 /* convert_to_i420.cc in Sources */,
-				D1C3D0DD17C157CD00CA0FD2 /* convert.cc in Sources */,
-				D1C3D0E617C157CD00CA0FD2 /* cpu_id.cc in Sources */,
-				D1C3D0EF17C157CD00CA0FD2 /* format_conversion.cc in Sources */,
-				D1C3D10A17C157CD00CA0FD2 /* planar_functions.cc in Sources */,
-				D1C3D12517C157CD00CA0FD2 /* rotate_neon.cc in Sources */,
-				D1C3D12E17C157CD00CA0FD2 /* rotate.cc in Sources */,
-				D1C3D13717C157CD00CA0FD2 /* row_any.cc in Sources */,
-				D1C3D14017C157CD00CA0FD2 /* row_common.cc in Sources */,
-				D1C3D15217C157CD00CA0FD2 /* row_neon.cc in Sources */,
-				D1C3D17617C157CD00CA0FD2 /* scale_argb_neon.cc in Sources */,
-				D1C3D17F17C157CD00CA0FD2 /* scale_argb.cc in Sources */,
-				D1C3D19117C157CD00CA0FD2 /* scale_neon.cc in Sources */,
-				D1C3D1A317C157CD00CA0FD2 /* video_common.cc in Sources */,
-				D159BCB817C227F70030FAB6 /* convert_from.cc in Sources */,
-				D1BCE06218F3F7FE00C83470 /* scale_common.cc in Sources */,
-				D159BCC117C228350030FAB6 /* rotate_argb.cc in Sources */,
-				D159BCCA17C228700030FAB6 /* scale.cc in Sources */,
-				D1BCE06B18F3F7FE00C83470 /* scale_posix.cc in Sources */,
-				D15D361217C386A700F40439 /* row_posix.cc in Sources */,
-				D15D361517C386B300F40439 /* compare_posix.cc in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXSourcesBuildPhase section */
-
-/* Begin XCBuildConfiguration section */
-		D1473F43150CA6CE00B20490 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				ALWAYS_SEARCH_USER_PATHS = NO;
-				CLANG_WARN_BOOL_CONVERSION = YES;
-				CLANG_WARN_CONSTANT_CONVERSION = YES;
-				CLANG_WARN_EMPTY_BODY = YES;
-				CLANG_WARN_ENUM_CONVERSION = YES;
-				CLANG_WARN_INT_CONVERSION = YES;
-				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				COPY_PHASE_STRIP = NO;
-				GCC_DYNAMIC_NO_PIC = NO;
-				GCC_ENABLE_OBJC_EXCEPTIONS = YES;
-				GCC_GENERATE_DEBUGGING_SYMBOLS = YES;
-				GCC_INLINES_ARE_PRIVATE_EXTERN = YES;
-				GCC_OPTIMIZATION_LEVEL = 0;
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					"DEBUG=1",
-					_DEBUG,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphoneos*]" = (
-					"$(LIBYUV_PREPROCESSOR_IOS)",
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphonesimulator*]" = "$(inherited)";
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=macosx*]" = (
-					"$(LIBYUV_PREPROCESSOR_MAC)",
-					"$(inherited)",
-				);
-				GCC_SYMBOLS_PRIVATE_EXTERN = YES;
-				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
-				GCC_WARN_ABOUT_RETURN_TYPE = YES;
-				GCC_WARN_UNDECLARED_SELECTOR = YES;
-				GCC_WARN_UNINITIALIZED_AUTOS = YES;
-				GCC_WARN_UNUSED_FUNCTION = YES;
-				GCC_WARN_UNUSED_VARIABLE = YES;
-				HEADER_SEARCH_PATHS = (
-					"$(SRCROOT)/../ogg/include",
-					"$(SRCROOT)/../vorbis/include",
-					"$(SRCROOT)/../xal/lib/ogg/include",
-					"$(SRCROOT)/../xal/lib/vorbis/include",
-					"$(SRCROOT)/../theora/include",
-					"$(SRCROOT)/src/YUV/libyuv/include",
-				);
-				LIBYUV_PREPROCESSOR_IOS = "LIBYUV_NEON __ARM_NEON__";
-				LIBYUV_PREPROCESSOR_MAC = __SSSE3__;
-				ONLY_ACTIVE_ARCH = YES;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D1473F44150CA6CE00B20490 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Debug;
-		};
-		D1473F45150CA6D600B20490 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				ALWAYS_SEARCH_USER_PATHS = NO;
-				CLANG_WARN_BOOL_CONVERSION = YES;
-				CLANG_WARN_CONSTANT_CONVERSION = YES;
-				CLANG_WARN_EMPTY_BODY = YES;
-				CLANG_WARN_ENUM_CONVERSION = YES;
-				CLANG_WARN_INT_CONVERSION = YES;
-				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				COPY_PHASE_STRIP = YES;
-				GCC_ENABLE_OBJC_EXCEPTIONS = YES;
-				GCC_GENERATE_DEBUGGING_SYMBOLS = YES;
-				GCC_INLINES_ARE_PRIVATE_EXTERN = YES;
-				GCC_OPTIMIZATION_LEVEL = 3;
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphoneos*]" = (
-					"$(LIBYUV_PREPROCESSOR_IOS)",
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphonesimulator*]" = "$(inherited)";
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=macosx*]" = (
-					"$(LIBYUV_PREPROCESSOR_MAC)",
-					"$(inherited)",
-				);
-				GCC_SYMBOLS_PRIVATE_EXTERN = YES;
-				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
-				GCC_WARN_ABOUT_RETURN_TYPE = YES;
-				GCC_WARN_UNDECLARED_SELECTOR = YES;
-				GCC_WARN_UNINITIALIZED_AUTOS = YES;
-				GCC_WARN_UNUSED_FUNCTION = YES;
-				GCC_WARN_UNUSED_VARIABLE = YES;
-				HEADER_SEARCH_PATHS = (
-					"$(SRCROOT)/../ogg/include",
-					"$(SRCROOT)/../vorbis/include",
-					"$(SRCROOT)/../xal/lib/ogg/include",
-					"$(SRCROOT)/../xal/lib/vorbis/include",
-					"$(SRCROOT)/../theora/include",
-					"$(SRCROOT)/src/YUV/libyuv/include",
-				);
-				LIBYUV_PREPROCESSOR_IOS = "LIBYUV_NEON __ARM_NEON__";
-				LIBYUV_PREPROCESSOR_MAC = __SSSE3__;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D1473F46150CA6D600B20490 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = "App Store";
-		};
-		D1473F47150CA6E200B20490 /* Release */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				ALWAYS_SEARCH_USER_PATHS = NO;
-				CLANG_WARN_BOOL_CONVERSION = YES;
-				CLANG_WARN_CONSTANT_CONVERSION = YES;
-				CLANG_WARN_EMPTY_BODY = YES;
-				CLANG_WARN_ENUM_CONVERSION = YES;
-				CLANG_WARN_INT_CONVERSION = YES;
-				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				COPY_PHASE_STRIP = YES;
-				GCC_ENABLE_OBJC_EXCEPTIONS = YES;
-				GCC_GENERATE_DEBUGGING_SYMBOLS = YES;
-				GCC_INLINES_ARE_PRIVATE_EXTERN = YES;
-				GCC_OPTIMIZATION_LEVEL = 3;
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphoneos*]" = (
-					"$(LIBYUV_PREPROCESSOR_IOS)",
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=iphonesimulator*]" = "$(inherited)";
-				"GCC_PREPROCESSOR_DEFINITIONS[sdk=macosx*]" = (
-					"$(LIBYUV_PREPROCESSOR_MAC)",
-					"$(inherited)",
-				);
-				GCC_SYMBOLS_PRIVATE_EXTERN = YES;
-				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
-				GCC_WARN_ABOUT_RETURN_TYPE = YES;
-				GCC_WARN_UNDECLARED_SELECTOR = YES;
-				GCC_WARN_UNINITIALIZED_AUTOS = YES;
-				GCC_WARN_UNUSED_FUNCTION = YES;
-				GCC_WARN_UNUSED_VARIABLE = YES;
-				HEADER_SEARCH_PATHS = (
-					"$(SRCROOT)/../ogg/include",
-					"$(SRCROOT)/../vorbis/include",
-					"$(SRCROOT)/../xal/lib/ogg/include",
-					"$(SRCROOT)/../xal/lib/vorbis/include",
-					"$(SRCROOT)/../theora/include",
-					"$(SRCROOT)/src/YUV/libyuv/include",
-				);
-				LIBYUV_PREPROCESSOR_IOS = "LIBYUV_NEON __ARM_NEON__";
-				LIBYUV_PREPROCESSOR_MAC = __SSSE3__;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D1473F48150CA6E200B20490 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Release;
-		};
-		D198F977177A31FC002942E3 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D198F979177A31FC002942E3 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D198F97A177A31FC002942E3 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D198F9A3177A31FE002942E3 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D198F9A5177A31FE002942E3 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D198F9A6177A31FE002942E3 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D198F9D0177A3200002942E3 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					__AVFOUNDATION,
-					YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_theora_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D198F9D2177A3200002942E3 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					__AVFOUNDATION,
-					YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_theora_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D198F9D3177A3200002942E3 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_MAC,
-					__THEORA,
-					__AVFOUNDATION,
-					YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer_theora_avfoundation;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D1BB6FB8150E9E7100EF9400 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D1BB6FBA150E9E7100EF9400 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D1BB6FBB150E9E7100EF9400 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D1CDFF441696C77A00609AB0 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Debug;
-		};
-		D1CDFF461696C77A00609AB0 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Release;
-		};
-		D1CDFF471696C77A00609AB0 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = "App Store";
-		};
-		D1CDFF6C1696C79700609AB0 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Debug;
-		};
-		D1CDFF6E1696C79700609AB0 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = Release;
-		};
-		D1CDFF6F1696C79700609AB0 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC418D7777800A36FDC /* Mac.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				INFOPLIST_FILE = Info.plist;
-				LD_DYLIB_INSTALL_NAME = "@executable_path/../Frameworks/$(EXECUTABLE_PATH)";
-				PRODUCT_NAME = theoraplayer;
-				WRAPPER_EXTENSION = framework;
-			};
-			name = "App Store";
-		};
-		D1CDFFC01696E1CA00609AB0 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D1CDFFC21696E1CA00609AB0 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D1CDFFC31696E1CA00609AB0 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-		D1CDFFE51696E1D700609AB0 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Debug;
-		};
-		D1CDFFE71696E1D700609AB0 /* Release */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = Release;
-		};
-		D1CDFFE81696E1D700609AB0 /* App Store */ = {
-			isa = XCBuildConfiguration;
-			baseConfigurationReference = D1358BC318D7777800A36FDC /* iOS.xcconfig */;
-			buildSettings = {
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					_IOS,
-					__THEORA,
-					__AVFOUNDATION,
-					_YUV_LIBYUV,
-					"$(inherited)",
-				);
-				"GCC_PREPROCESSOR_DEFINITIONS[arch=arm64]" = LIBYUV_DISABLE_NEON;
-				GCC_WARN_UNINITIALIZED_AUTOS = NO;
-				GCC_WARN_UNUSED_VALUE = NO;
-				GCC_WARN_UNUSED_VARIABLE = NO;
-				PRODUCT_NAME = theoraplayer;
-				SKIP_INSTALL = YES;
-			};
-			name = "App Store";
-		};
-/* End XCBuildConfiguration section */
-
-/* Begin XCConfigurationList section */
-		D1473F23150CA69B00B20490 /* Build configuration list for PBXProject "theoraplayer" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1473F43150CA6CE00B20490 /* Debug */,
-				D1473F47150CA6E200B20490 /* Release */,
-				D1473F45150CA6D600B20490 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1473F3F150CA69B00B20490 /* Build configuration list for PBXNativeTarget "theoraplayer (Theora)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1473F44150CA6CE00B20490 /* Debug */,
-				D1473F48150CA6E200B20490 /* Release */,
-				D1473F46150CA6D600B20490 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D198F975177A31FC002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac Theora)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D198F977177A31FC002942E3 /* Debug */,
-				D198F979177A31FC002942E3 /* Release */,
-				D198F97A177A31FC002942E3 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D198F9A1177A31FE002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D198F9A3177A31FE002942E3 /* Debug */,
-				D198F9A5177A31FE002942E3 /* Release */,
-				D198F9A6177A31FE002942E3 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D198F9CE177A3200002942E3 /* Build configuration list for PBXNativeTarget "theoraplayer (Mac Theora AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D198F9D0177A3200002942E3 /* Debug */,
-				D198F9D2177A3200002942E3 /* Release */,
-				D198F9D3177A3200002942E3 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1BB6FBC150E9E7100EF9400 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS Theora)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1BB6FB8150E9E7100EF9400 /* Debug */,
-				D1BB6FBA150E9E7100EF9400 /* Release */,
-				D1BB6FBB150E9E7100EF9400 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1CDFF421696C77A00609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1CDFF441696C77A00609AB0 /* Debug */,
-				D1CDFF461696C77A00609AB0 /* Release */,
-				D1CDFF471696C77A00609AB0 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1CDFF6A1696C79700609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (Theora AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1CDFF6C1696C79700609AB0 /* Debug */,
-				D1CDFF6E1696C79700609AB0 /* Release */,
-				D1CDFF6F1696C79700609AB0 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1CDFFBE1696E1CA00609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1CDFFC01696E1CA00609AB0 /* Debug */,
-				D1CDFFC21696E1CA00609AB0 /* Release */,
-				D1CDFFC31696E1CA00609AB0 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-		D1CDFFE31696E1D700609AB0 /* Build configuration list for PBXNativeTarget "theoraplayer (iOS Theora AVFoundation)" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				D1CDFFE51696E1D700609AB0 /* Debug */,
-				D1CDFFE71696E1D700609AB0 /* Release */,
-				D1CDFFE81696E1D700609AB0 /* App Store */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Debug;
-		};
-/* End XCConfigurationList section */
-	};
-	rootObject = D1473F20150CA69B00B20490 /* Project object */;
-}
diff --git a/drivers/theoraplayer/video_stream_theoraplayer.cpp b/drivers/theoraplayer/video_stream_theoraplayer.cpp
deleted file mode 100644
index 876cac3425..0000000000
--- a/drivers/theoraplayer/video_stream_theoraplayer.cpp
+++ /dev/null
@@ -1,556 +0,0 @@
-/*************************************************************************/
-/*  video_stream.cpp                                                     */
-/*************************************************************************/
-/*                       This file is part of:                           */
-/*                           GODOT ENGINE                                */
-/*                    http://www.godotengine.org                         */
-/*************************************************************************/
-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
-/*                                                                       */
-/* Permission is hereby granted, free of charge, to any person obtaining */
-/* a copy of this software and associated documentation files (the       */
-/* "Software"), to deal in the Software without restriction, including   */
-/* without limitation the rights to use, copy, modify, merge, publish,   */
-/* distribute, sublicense, and/or sell copies of the Software, and to    */
-/* permit persons to whom the Software is furnished to do so, subject to */
-/* the following conditions:                                             */
-/*                                                                       */
-/* The above copyright notice and this permission notice shall be        */
-/* included in all copies or substantial portions of the Software.       */
-/*                                                                       */
-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
-/*************************************************************************/
-#include "video_stream_theoraplayer.h"
-
-#include "core/os/file_access.h"
-
-#include "include/theoraplayer/TheoraPlayer.h"
-#include "include/theoraplayer/TheoraTimer.h"
-#include "include/theoraplayer/TheoraAudioInterface.h"
-#include "include/theoraplayer/TheoraDataSource.h"
-#include "include/theoraplayer/TheoraException.h"
-
-#include "core/ring_buffer.h"
-#include "core/os/thread_safe.h"
-
-#include "core/globals.h"
-
-static TheoraVideoManager* mgr = NULL;
-
-class TPDataFA : public TheoraDataSource {
-
-	FileAccess* fa;
-	String data_name;
-
-public:
-
-	int read(void* output,int nBytes) {
-
-		if (!fa)
-			return -1;
-
-		return fa->get_buffer((uint8_t*)output, nBytes);
-	};
-
-	//! returns a string representation of the DataSource, eg 'File: source.ogg'
-	virtual std::string repr() {
-		return data_name.utf8().get_data();
-	};
-
-	//! position the source pointer to byte_index from the start of the source
-	virtual void seek(unsigned long byte_index) {
-
-		if (!fa)
-			return;
-
-		fa->seek(byte_index);
-	};
-
-
-	//! return the size of the stream in bytes
-	virtual unsigned long size() {
-
-		if (!fa)
-			return 0;
-
-		return fa->get_len();
-	};
-
-	//! return the current position of the source pointer
-	virtual unsigned long tell() {
-
-		if (!fa)
-			return 0;
-
-		return fa->get_pos();
-	};
-
-	TPDataFA(const String& p_path) {
-
-		fa = FileAccess::open(p_path, FileAccess::READ);
-		data_name = "File: " + p_path;
-	};
-
-	TPDataFA(FileAccess* p_fa, const String& p_path) {
-
-		fa = p_fa;
-		data_name = "File: " + p_path;
-	};
-
-	~TPDataFA() {
-
-		if (fa)
-			memdelete(fa);
-	};
-};
-
-class AudioStreamInput : public AudioStreamResampled {
-
-	_THREAD_SAFE_CLASS_;
-
-	int channels;
-	int freq;
-
-	RID stream_rid;
-	mutable RingBuffer<float> rb;
-	int rb_power;
-	int total_wrote;
-	bool playing;
-	bool paused;
-
-public:
-
-	virtual void play() {
-
-		_THREAD_SAFE_METHOD_
-		_setup(channels, freq, 256);
-		stream_rid=AudioServer::get_singleton()->audio_stream_create(get_audio_stream());
-		AudioServer::get_singleton()->stream_set_active(stream_rid,true);
-		AudioServer::get_singleton()->stream_set_volume_scale(stream_rid,1);
-		playing = true;
-		paused = false;
-	};
-	virtual void stop() {
-
-		_THREAD_SAFE_METHOD_
-
-		AudioServer::get_singleton()->stream_set_active(stream_rid,false);
-		//_clear_stream();
-		playing=false;
-		_clear();
-	};
-
-	virtual bool is_playing() const { return true; };
-
-	virtual void set_paused(bool p_paused) { paused = p_paused; };
-	virtual bool is_paused(bool p_paused) const { return paused; };
-
-	virtual void set_loop(bool p_enable) {};
-	virtual bool has_loop() const { return false; };
-
-	virtual float get_length() const { return 0; };
-
-	virtual String get_stream_name() const { return "Theora Audio Stream"; };
-
-	virtual int get_loop_count() const { return 1; };
-
-	virtual float get_pos() const { return 0; };
-	virtual void seek_pos(float p_time) {};
-
-	virtual UpdateMode get_update_mode() const { return UPDATE_THREAD; };
-
-	virtual bool _can_mix() const { return true; };
-
-	void input(float* p_data, int p_samples) {
-
-
-		_THREAD_SAFE_METHOD_;
-		//printf("input %i samples from %p\n", p_samples, p_data);
-		if (rb.space_left() < p_samples) {
-			rb_power += 1;
-			rb.resize(rb_power);
-		}
-		rb.write(p_data, p_samples);
-
-		update(); //update too here for less latency
-	};
-
-	void update() {
-
-		_THREAD_SAFE_METHOD_;
-		int todo = get_todo();
-		int16_t* buffer = get_write_buffer();
-		int frames = rb.data_left()/channels;
-		const int to_write = MIN(todo, frames);
-
-		for (int i=0; i<to_write*channels; i++) {
-
-			int v = rb.read() * 32767;
-			int16_t sample = CLAMP(v,-32768,32767);
-			buffer[i] = sample;
-		};
-		write(to_write);
-		total_wrote += to_write;
-	};
-
-	int get_pending() const {
-		return rb.data_left();
-	};
-
-	int get_total_wrote() {
-
-		return total_wrote - (get_total() - get_todo());
-	};
-
-	AudioStreamInput(int p_channels, int p_freq) {
-
-		playing = false;
-		paused = true;
-		channels = p_channels;
-		freq = p_freq;
-		total_wrote = 0;
-		rb_power = 22;
-		rb.resize(rb_power);
-	};
-
-	~AudioStreamInput() {
-
-		stop();
-	};
-};
-
-class TPAudioGodot : public TheoraAudioInterface, TheoraTimer {
-
-	Ref<AudioStreamInput> stream;
-	int sample_count;
-	int channels;
-	int freq;
-
-public:
-
-	void insertData(float* data, int nSamples) {
-
-		stream->input(data, nSamples);
-	};
-
-	TPAudioGodot(TheoraVideoClip* owner, int nChannels, int p_freq)
-		: TheoraAudioInterface(owner, nChannels, p_freq), TheoraTimer() {
-
-		printf("***************** audio interface constructor freq %i\n", p_freq);
-		channels = nChannels;
-		freq = p_freq;
-		stream = Ref<AudioStreamInput>(memnew(AudioStreamInput(nChannels, p_freq)));
-		stream->play();
-		sample_count = 0;
-		owner->setTimer(this);
-	};
-
-	void stop() {
-
-		stream->stop();
-	};
-
-	void update(float time_increase)
-	{
-		float prev_time = mTime;
-		//mTime = (float)(stream->get_total_wrote()) / freq;
-		//mTime = MAX(0,mTime-AudioServer::get_singleton()->get_output_delay());
-		//mTime = (float)sample_count / channels / freq;
-		mTime += time_increase;
-		if (mTime - prev_time > .02) printf("time increase %f secs\n", mTime - prev_time);
-		//float duration=mClip->getDuration();
-		//if (mTime > duration) mTime=duration;
-		//printf("time at timer is %f, %f, samples %i\n", mTime, time_increase, sample_count);
-	}
-};
-
-class TPAudioGodotFactory : public TheoraAudioInterfaceFactory {
-
-public:
-	TheoraAudioInterface* createInstance(TheoraVideoClip* owner, int nChannels, int freq) {
-
-		printf("************** creating audio output\n");
-		TheoraAudioInterface* ta = new TPAudioGodot(owner, nChannels, freq);
-		return ta;
-	};
-};
-
-static TPAudioGodotFactory* audio_factory = NULL;
-
-void VideoStreamTheoraplayer::stop() {
-
-	playing = false;
-	if (clip) {
-		clip->stop();
-		clip->seek(0);
-	};
-	started = true;
-};
-
-void VideoStreamTheoraplayer::play() {
-	if (clip)
-		playing = true;
-};
-
-bool VideoStreamTheoraplayer::is_playing() const {
-
-	return playing;
-};
-
-void VideoStreamTheoraplayer::set_paused(bool p_paused) {
-
-	paused = p_paused;
-	if (paused) {
-		clip->pause();
-	} else {
-		if (clip && playing && !started)
-			clip->play();
-	}
-};
-
-bool VideoStreamTheoraplayer::is_paused(bool p_paused) const {
-
-	return !playing;
-};
-
-void VideoStreamTheoraplayer::set_loop(bool p_enable) {
-
-	loop = p_enable;
-};
-
-bool VideoStreamTheoraplayer::has_loop() const {
-
-	return loop;
-};
-
-float VideoStreamTheoraplayer::get_length() const {
-
-	if (!clip)
-		return 0;
-
-	return clip->getDuration();
-};
-
-
-float VideoStreamTheoraplayer::get_pos() const {
-
-	if (!clip)
-		return 0;
-
-	return clip->getTimer()->getTime();
-};
-
-void VideoStreamTheoraplayer::seek_pos(float p_time) {
-
-	if (!clip)
-		return;
-
-	clip->seek(p_time);
-};
-
-int VideoStreamTheoraplayer::get_pending_frame_count() const {
-
-	if (!clip)
-		return 0;
-
-	TheoraVideoFrame* f = clip->getNextFrame();
-	return f ? 1 : 0;
-};
-
-
-void VideoStreamTheoraplayer::pop_frame(Ref<ImageTexture> p_tex) {
-
-	if (!clip)
-		return;
-
-	TheoraVideoFrame* f = clip->getNextFrame();
-	if (!f) {
-		return;
-	};
-
-#ifdef GLES2_ENABLED
-//	RasterizerGLES2* r = RasterizerGLES2::get_singleton();
-//	r->_texture_set_data(p_tex, f->mBpp == 3 ? Image::Format_RGB : Image::Format_RGBA, f->mBpp, w, h, f->getBuffer());
-
-#endif
-
-	float w=clip->getWidth(),h=clip->getHeight();
-	int imgsize = w * h * f->mBpp;
-
-	int size = f->getStride() * f->getHeight() * f->mBpp;
-	data.resize(imgsize);
-	{
-		DVector<uint8_t>::Write wr = data.write();
-		uint8_t* ptr = wr.ptr();
-		copymem(ptr, f->getBuffer(), imgsize);
-	}
-    /*
-    for (int i=0; i<h; i++) {
-        int dstofs = i * w * f->mBpp;
-        int srcofs = i * f->getStride() * f->mBpp;
-        copymem(ptr + dstofs, f->getBuffer() + dstofs, w * f->mBpp);
-    };
-     */
-	Image frame = Image();
-	frame.create(w, h, 0, f->mBpp == 3 ? Image::FORMAT_RGB : Image::FORMAT_RGBA, data);
-
-	clip->popFrame();
-
-	if (p_tex->get_width() == 0) {
-		p_tex->create(frame.get_width(),frame.get_height(),frame.get_format(),Texture::FLAG_VIDEO_SURFACE|Texture::FLAG_FILTER);
-		p_tex->set_data(frame);
-	} else {
-
-		p_tex->set_data(frame);
-	};
-};
-
-/*
-Image VideoStreamTheoraplayer::pop_frame() {
-
-	Image ret = frame;
-	frame = Image();
-	return ret;
-};
-*/
-
-Image VideoStreamTheoraplayer::peek_frame() const {
-
-	return Image();
-};
-
-void VideoStreamTheoraplayer::update(float p_time) {
-
-	if (!mgr)
-		return;
-
-	if (!clip)
-		return;
-
-	if (!playing || paused)
-		return;
-
-	//printf("video update!\n");
-	if (started) {
-		if (clip->getNumReadyFrames() < 2) {
-			printf("frames not ready, returning!\n");
-			return;
-		};
-		started = false;
-		//printf("playing clip!\n");
-		clip->play();
-	} else if (clip->isDone()) {
-		playing = false;
-	};
-
-	mgr->update(p_time);
-};
-
-
-void VideoStreamTheoraplayer::set_audio_track(int p_idx) {
-	audio_track=p_idx;
-	if (clip)
-		clip->set_audio_track(audio_track);
-}
-
-void VideoStreamTheoraplayer::set_file(const String& p_file) {
-
-	FileAccess* f = FileAccess::open(p_file, FileAccess::READ);
-	if (!f || !f->is_open())
-		return;
-
-	if (!audio_factory) {
-		audio_factory = memnew(TPAudioGodotFactory);
-	};
-
-	if (mgr == NULL) {
-		mgr = memnew(TheoraVideoManager);
-		mgr->setAudioInterfaceFactory(audio_factory);
-	};
-
-	int track = GLOBAL_DEF("theora/audio_track", 0); // hack
-
-	if (p_file.find(".mp4") != -1) {
-		
-		std::string file = p_file.replace("res://", "").utf8().get_data();
-		clip = mgr->createVideoClip(file, TH_RGBX, 2, false, track);
-		//clip->set_audio_track(audio_track);
-		memdelete(f);
-
-	} else {
-
-		TheoraDataSource* ds = memnew(TPDataFA(f, p_file));
-
-		try {
-			clip = mgr->createVideoClip(ds);
-			clip->set_audio_track(audio_track);
-		} catch (_TheoraGenericException e) {
-			printf("exception ocurred! %s\n", e.repr().c_str());
-			clip = NULL;
-		};
-	};
-
-	clip->pause();
-	started = true;
-};
-
-VideoStreamTheoraplayer::~VideoStreamTheoraplayer() {
-
-	stop();
-	//if (mgr) { // this should be a singleton or static or something
-	//	memdelete(mgr);
-	//};
-	//mgr = NULL;
-	if (clip) {
-		mgr->destroyVideoClip(clip);
-		clip = NULL;
-	};
-};
-
-VideoStreamTheoraplayer::VideoStreamTheoraplayer() {
-
-	//mgr = NULL;
-	clip = NULL;
-	started = false;
-	playing = false;
-	paused = false;
-	loop = false;
-	audio_track=0;
-};
-
-
-RES ResourceFormatLoaderVideoStreamTheoraplayer::load(const String &p_path, const String& p_original_path, Error *r_error) {
-	if (r_error)
-		*r_error=OK;
-
-	VideoStreamTheoraplayer *stream = memnew(VideoStreamTheoraplayer);
-	stream->set_file(p_path);
-	return Ref<VideoStreamTheoraplayer>(stream);
-}
-
-void ResourceFormatLoaderVideoStreamTheoraplayer::get_recognized_extensions(List<String> *p_extensions) const {
-
-	p_extensions->push_back("ogm");
-	p_extensions->push_back("ogv");
-	p_extensions->push_back("mp4");
-}
-bool ResourceFormatLoaderVideoStreamTheoraplayer::handles_type(const String& p_type) const {
-	return p_type=="VideoStream" || p_type == "VideoStreamTheoraplayer";
-}
-
-String ResourceFormatLoaderVideoStreamTheoraplayer::get_resource_type(const String &p_path) const {
-
-	String exl=p_path.extension().to_lower();
-	if (exl=="ogm" || exl=="ogv" || exl=="mp4")
-		return "VideoStream";
-	return "";
-}
-
-
-
diff --git a/drivers/theoraplayer/video_stream_theoraplayer.h b/drivers/theoraplayer/video_stream_theoraplayer.h
deleted file mode 100644
index 69cae7c4a2..0000000000
--- a/drivers/theoraplayer/video_stream_theoraplayer.h
+++ /dev/null
@@ -1,66 +0,0 @@
-#ifndef VIDEO_STREAM_THEORAPLAYER_H
-#define VIDEO_STREAM_THEORAPLAYER_H
-
-#include "scene/resources/video_stream.h"
-#include "io/resource_loader.h"
-#include "scene/resources/texture.h"
-
-class TheoraVideoManager;
-class TheoraVideoClip;
-
-class VideoStreamTheoraplayer : public VideoStream {
-
-	OBJ_TYPE(VideoStreamTheoraplayer,VideoStream);
-
-	mutable DVector<uint8_t> data;
-	TheoraVideoClip* clip;
-	bool started;
-	bool playing;
-	bool loop;
-	bool paused;
-
-	int audio_track;
-
-public:
-
-	virtual void stop();
-	virtual void play();
-
-	virtual bool is_playing() const;
-
-	virtual void set_paused(bool p_paused);
-	virtual bool is_paused(bool p_paused) const;
-
-	virtual void set_loop(bool p_enable);
-	virtual bool has_loop() const;
-
-	virtual float get_pos() const;
-	virtual void seek_pos(float p_time);
-
-	virtual float get_length() const;
-
-	virtual int get_pending_frame_count() const;
-	virtual void pop_frame(Ref<ImageTexture> p_tex);
-	virtual Image peek_frame() const;
-
-	void update(float p_time);
-
-	void set_file(const String& p_file);
-	void set_audio_track(int p_idx);
-
-	~VideoStreamTheoraplayer();
-	VideoStreamTheoraplayer();
-};
-
-class ResourceFormatLoaderVideoStreamTheoraplayer : public ResourceFormatLoader {
-public:
-	virtual RES load(const String &p_path,const String& p_original_path="",Error *r_error=NULL);
-	virtual void get_recognized_extensions(List<String> *p_extensions) const;
-	virtual bool handles_type(const String& p_type) const;
-	virtual String get_resource_type(const String &p_path) const;
-
-};
-
-
-#endif
-
diff --git a/main/main.cpp b/main/main.cpp
index 84f7c3a88e..e1bc3b9fd9 100644
--- a/main/main.cpp
+++ b/main/main.cpp
@@ -701,6 +701,7 @@ Error Main::setup(const char *execpath,int argc, char *argv[],bool p_second_phas
 	GLOBAL_DEF("display/resizable",video_mode.resizable);
 	GLOBAL_DEF("display/test_width",0);
 	GLOBAL_DEF("display/test_height",0);
+	OS::get_singleton()->_pixel_snap=GLOBAL_DEF("display/use_2d_pixel_snap",false);
 	if (rtm==-1) {
 		rtm=GLOBAL_DEF("render/thread_model",OS::RENDER_THREAD_SAFE);
 		if (rtm>=1) //hack for now
diff --git a/modules/gdscript/gd_parser.cpp b/modules/gdscript/gd_parser.cpp
index 9c39051b7f..313fb57d0e 100644
--- a/modules/gdscript/gd_parser.cpp
+++ b/modules/gdscript/gd_parser.cpp
@@ -2421,6 +2421,16 @@ void GDParser::_parse_class(ClassNode *p_class) {
 								}; //fallthrough to use the same
 								case Variant::REAL: {
 
+									if (tokenizer->get_token()==GDTokenizer::TK_IDENTIFIER && tokenizer->get_token_identifier()=="EASE") {
+										current_export.hint=PROPERTY_HINT_EXP_EASING;
+										tokenizer->advance();
+										if (tokenizer->get_token()!=GDTokenizer::TK_PARENTHESIS_CLOSE) {
+											_set_error("Expected ')' in hint.");
+											return;
+										}
+										break;
+									}
+
 									float sign=1.0;
 
 									if (tokenizer->get_token()==GDTokenizer::TK_OP_SUB) {
@@ -2571,6 +2581,17 @@ void GDParser::_parse_class(ClassNode *p_class) {
 										}
 										break;
 									}
+
+									if (tokenizer->get_token()==GDTokenizer::TK_IDENTIFIER && tokenizer->get_token_identifier()=="MULTILINE") {
+
+										current_export.hint=PROPERTY_HINT_MULTILINE_TEXT;
+										tokenizer->advance();
+										if (tokenizer->get_token()!=GDTokenizer::TK_PARENTHESIS_CLOSE) {
+											_set_error("Expected ')' in hint.");
+											return;
+										}
+										break;
+									}
 								} break;
 								case Variant::COLOR: {
 
@@ -2607,23 +2628,16 @@ void GDParser::_parse_class(ClassNode *p_class) {
 					} else if (tokenizer->get_token()==GDTokenizer::TK_IDENTIFIER) {
 
 						String identifier = tokenizer->get_token_identifier();
-						if (identifier == "flag") {
-							current_export.type=Variant::INT;
-							current_export.hint=PROPERTY_HINT_ALL_FLAGS;
-						}else if (identifier == "multiline"){
-							current_export.type=Variant::STRING;
-							current_export.hint=PROPERTY_HINT_MULTILINE_TEXT;
-						} else {
-							if (!ObjectTypeDB::is_type(identifier,"Resource")) {
-	
-								current_export=PropertyInfo();
-								_set_error("Export hint not a type or resource.");
-							}
-	
-							current_export.type=Variant::OBJECT;
-							current_export.hint=PROPERTY_HINT_RESOURCE_TYPE;
-							current_export.hint_string=identifier;
+						if (!ObjectTypeDB::is_type(identifier,"Resource")) {
+
+							current_export=PropertyInfo();
+							_set_error("Export hint not a type or resource.");
 						}
+
+						current_export.type=Variant::OBJECT;
+						current_export.hint=PROPERTY_HINT_RESOURCE_TYPE;
+						current_export.hint_string=identifier;
+
 						tokenizer->advance();
 					}
 
diff --git a/platform/android/detect.py b/platform/android/detect.py
index c0c091b20a..c36e35484e 100644
--- a/platform/android/detect.py
+++ b/platform/android/detect.py
@@ -38,8 +38,6 @@ def get_flags():
 		('nedmalloc', 'no'),
 		('builtin_zlib', 'no'),
                 ('openssl','builtin'), #use builtin openssl
-		('theora','no'), #use builtin openssl
-
         ]
 
 
diff --git a/platform/flash/os_flash.h b/platform/flash/os_flash.h
index c8013df2f2..46fce117fc 100644
--- a/platform/flash/os_flash.h
+++ b/platform/flash/os_flash.h
@@ -1,88 +1,88 @@
-#ifndef OS_FLASH_H

-#define OS_FLASH_H

-

-#include "os/input.h"

-#include "drivers/unix/os_unix.h"

-#include "os/input.h"

-#include "servers/visual_server.h"

-#include "servers/visual/rasterizer.h"

-#include "servers/physics/physics_server_sw.h"

-#include "servers/spatial_sound/spatial_sound_server_sw.h"

-#include "servers/spatial_sound_2d/spatial_sound_2d_server_sw.h"

-#include "servers/audio/audio_server_sw.h"

-#include "servers/physics_2d/physics_2d_server_sw.h"

-#include "main/input_default.h"

-

-class OSFlash : public OS_Unix {

-

-	VideoMode default_videomode;

-	MainLoop * main_loop;

-	InputDefault *input;

-	Rasterizer *rasterizer;

-	VisualServer *visual_server;

-	AudioDriverSW* audio_driver;

-	AudioServerSW *audio_server;

-	SampleManagerMallocSW *sample_manager;

-	SpatialSoundServerSW *spatial_sound_server;

-	SpatialSound2DServerSW *spatial_sound_2d_server;

-	PhysicsServer *physics_server;

-	Physics2DServer *physics_2d_server;

-

-public:

-

-	virtual int get_video_driver_count() const;

-	virtual const char * get_video_driver_name(int p_driver) const;

-

-	virtual VideoMode get_default_video_mode() const;

-

-	virtual int get_audio_driver_count() const;

-	virtual const char * get_audio_driver_name(int p_driver) const;

-

-	virtual void initialize_core();

-	virtual void initialize(const VideoMode& p_desired,int p_video_driver,int p_audio_driver);

-

-	virtual void set_main_loop( MainLoop * p_main_loop );

-	virtual void delete_main_loop();

-

-	virtual void finalize();

-

-	typedef int64_t ProcessID;

-

-	static OS* get_singleton();

-

-	virtual void set_mouse_show(bool p_show);

-	virtual void set_mouse_grab(bool p_grab);

-	virtual bool is_mouse_grab_enabled() const;

-	virtual Point2 get_mouse_pos() const;

-	virtual int get_mouse_button_state() const;

-	virtual void set_window_title(const String& p_title);

-

-	//virtual void set_clipboard(const String& p_text);

-	//virtual String get_clipboard() const;

-

-

-	virtual bool has_virtual_keyboard() const;

-	virtual void show_virtual_keyboard(const String& p_existing_text,const Rect2& p_screen_rect);

-	virtual void hide_virtual_keyboard();

-

-	virtual void set_video_mode(const VideoMode& p_video_mode,int p_screen=0);

-	virtual VideoMode get_video_mode(int p_screen=0) const;

-	virtual void get_fullscreen_mode_list(List<VideoMode> *p_list,int p_screen=0) const;

-

-	virtual String get_name();

-	virtual MainLoop *get_main_loop() const;

-

-	virtual bool can_draw() const;

-

-	virtual void set_cursor_shape(CursorShape p_shape);

-

-	virtual bool has_touchscreen_ui_hint() const;

-

-	virtual void yield();

-

-	virtual Error shell_open(String p_uri);

-

-	bool iterate();

-};

-

-#endif

+#ifndef OS_FLASH_H
+#define OS_FLASH_H
+
+#include "os/input.h"
+#include "drivers/unix/os_unix.h"
+#include "os/input.h"
+#include "servers/visual_server.h"
+#include "servers/visual/rasterizer.h"
+#include "servers/physics/physics_server_sw.h"
+#include "servers/spatial_sound/spatial_sound_server_sw.h"
+#include "servers/spatial_sound_2d/spatial_sound_2d_server_sw.h"
+#include "servers/audio/audio_server_sw.h"
+#include "servers/physics_2d/physics_2d_server_sw.h"
+#include "main/input_default.h"
+
+class OSFlash : public OS_Unix {
+
+	VideoMode default_videomode;
+	MainLoop * main_loop;
+	InputDefault *input;
+	Rasterizer *rasterizer;
+	VisualServer *visual_server;
+	AudioDriverSW* audio_driver;
+	AudioServerSW *audio_server;
+	SampleManagerMallocSW *sample_manager;
+	SpatialSoundServerSW *spatial_sound_server;
+	SpatialSound2DServerSW *spatial_sound_2d_server;
+	PhysicsServer *physics_server;
+	Physics2DServer *physics_2d_server;
+
+public:
+
+	virtual int get_video_driver_count() const;
+	virtual const char * get_video_driver_name(int p_driver) const;
+
+	virtual VideoMode get_default_video_mode() const;
+
+	virtual int get_audio_driver_count() const;
+	virtual const char * get_audio_driver_name(int p_driver) const;
+
+	virtual void initialize_core();
+	virtual void initialize(const VideoMode& p_desired,int p_video_driver,int p_audio_driver);
+
+	virtual void set_main_loop( MainLoop * p_main_loop );
+	virtual void delete_main_loop();
+
+	virtual void finalize();
+
+	typedef int64_t ProcessID;
+
+	static OS* get_singleton();
+
+	virtual void set_mouse_show(bool p_show);
+	virtual void set_mouse_grab(bool p_grab);
+	virtual bool is_mouse_grab_enabled() const;
+	virtual Point2 get_mouse_pos() const;
+	virtual int get_mouse_button_state() const;
+	virtual void set_window_title(const String& p_title);
+
+	//virtual void set_clipboard(const String& p_text);
+	//virtual String get_clipboard() const;
+
+
+	virtual bool has_virtual_keyboard() const;
+	virtual void show_virtual_keyboard(const String& p_existing_text,const Rect2& p_screen_rect);
+	virtual void hide_virtual_keyboard();
+
+	virtual void set_video_mode(const VideoMode& p_video_mode,int p_screen=0);
+	virtual VideoMode get_video_mode(int p_screen=0) const;
+	virtual void get_fullscreen_mode_list(List<VideoMode> *p_list,int p_screen=0) const;
+
+	virtual String get_name();
+	virtual MainLoop *get_main_loop() const;
+
+	virtual bool can_draw() const;
+
+	virtual void set_cursor_shape(CursorShape p_shape);
+
+	virtual bool has_touchscreen_ui_hint() const;
+
+	virtual void yield();
+
+	virtual Error shell_open(String p_uri);
+
+	bool iterate();
+};
+
+#endif
diff --git a/platform/iphone/detect.py b/platform/iphone/detect.py
index 0661b45c1b..137fcc56f5 100644
--- a/platform/iphone/detect.py
+++ b/platform/iphone/detect.py
@@ -38,7 +38,6 @@ def get_flags():
 	return [
 		('tools', 'no'),
 		('webp', 'yes'),
-		("theora","no"),       		
 		('openssl','builtin'), #use builtin openssl
 	]
 
diff --git a/platform/osx/detect.py b/platform/osx/detect.py
index 4a20ca80c1..22cee0527e 100644
--- a/platform/osx/detect.py
+++ b/platform/osx/detect.py
@@ -31,7 +31,6 @@ def get_flags():
 	('opengl', 'no'),
 	('legacygl', 'yes'),
 	('builtin_zlib', 'no'),
-        ("theora","no"),       	
 	('freetype','builtin'), #use builtin freetype
 	]
 
diff --git a/platform/osx/os_osx.h b/platform/osx/os_osx.h
index 49fa6c0862..e8277688ac 100644
--- a/platform/osx/os_osx.h
+++ b/platform/osx/os_osx.h
@@ -157,6 +157,8 @@ public:
 	Error shell_open(String p_uri);
 	void push_input(const InputEvent& p_event);
 
+	String get_locale() const;
+
 	virtual void set_video_mode(const VideoMode& p_video_mode,int p_screen=0);
 	virtual VideoMode get_video_mode(int p_screen=0) const;
 	virtual void get_fullscreen_mode_list(List<VideoMode> *p_list,int p_screen=0) const;
diff --git a/platform/osx/os_osx.mm b/platform/osx/os_osx.mm
index b0ce37cecf..4990d04ab6 100644
--- a/platform/osx/os_osx.mm
+++ b/platform/osx/os_osx.mm
@@ -1254,6 +1254,11 @@ Error OS_OSX::shell_open(String p_uri) {
 	return OK;
 }
 
+String OS_OSX::get_locale() const {
+  NSString* preferredLang = [[NSLocale preferredLanguages] objectAtIndex:0];
+	return [preferredLang UTF8String];
+}
+
 void OS_OSX::swap_buffers() {
 
 	[context flushBuffer];
diff --git a/platform/windows/tcp_server_winsock.cpp b/platform/windows/tcp_server_winsock.cpp
index bf7e85aebb..cc689d9dcf 100644
--- a/platform/windows/tcp_server_winsock.cpp
+++ b/platform/windows/tcp_server_winsock.cpp
@@ -79,6 +79,13 @@ Error TCPServerWinsock::listen(uint16_t p_port,const List<String> *p_accepted_ho
 	my_addr.sin_addr.s_addr = INADDR_ANY; // automatically fill with my IP TODO: use p_accepted_hosts
 	memset(my_addr.sin_zero, '\0', sizeof my_addr.sin_zero);
 
+	int reuse=1;
+	if(setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, (char *)&reuse, sizeof(reuse)) < 0) {
+
+		printf("REUSEADDR failed!");
+	}
+
+
 	if (bind(sockfd, (struct sockaddr *)&my_addr, sizeof my_addr) != SOCKET_ERROR) {
 
 		if (::listen(sockfd, SOMAXCONN) == SOCKET_ERROR) {
diff --git a/platform/winrt/app.cpp b/platform/winrt/app.cpp
index 662229b04e..263cd684c4 100644
--- a/platform/winrt/app.cpp
+++ b/platform/winrt/app.cpp
@@ -1,385 +1,385 @@
-//

-// This file demonstrates how to initialize EGL in a Windows Store app, using ICoreWindow.

-//

-

-#include "app.h"

-

-#include "main/main.h"

-#include "core/os/dir_access.h"

-#include "core/os/file_access.h"

-

-using namespace Windows::ApplicationModel::Core;

-using namespace Windows::ApplicationModel::Activation;

-using namespace Windows::UI::Core;

-using namespace Windows::UI::Input;

-using namespace Windows::Foundation;

-using namespace Windows::Graphics::Display;

-using namespace Microsoft::WRL;

-using namespace Platform;

-

-using namespace $ext_safeprojectname$;

-

-// Helper to convert a length in device-independent pixels (DIPs) to a length in physical pixels.

-inline float ConvertDipsToPixels(float dips, float dpi)

-{

-    static const float dipsPerInch = 96.0f;

-    return floor(dips * dpi / dipsPerInch + 0.5f); // Round to nearest integer.

-}

-

-// Implementation of the IFrameworkViewSource interface, necessary to run our app.

-ref class HelloTriangleApplicationSource sealed : Windows::ApplicationModel::Core::IFrameworkViewSource

-{

-public:

-    virtual Windows::ApplicationModel::Core::IFrameworkView^ CreateView()

-    {

-        return ref new App();

-    }

-};

-

-// The main function creates an IFrameworkViewSource for our app, and runs the app.

-[Platform::MTAThread]

-int main(Platform::Array<Platform::String^>^)

-{

-    auto helloTriangleApplicationSource = ref new HelloTriangleApplicationSource();

-    CoreApplication::Run(helloTriangleApplicationSource);

-    return 0;

-}

-

-App::App() :

-    mWindowClosed(false),

-    mWindowVisible(true),

-    mWindowWidth(0),

-    mWindowHeight(0),

-    mEglDisplay(EGL_NO_DISPLAY),

-    mEglContext(EGL_NO_CONTEXT),

-    mEglSurface(EGL_NO_SURFACE)

-{

-}

-

-// The first method called when the IFrameworkView is being created.

-void App::Initialize(CoreApplicationView^ applicationView)

-{

-    // Register event handlers for app lifecycle. This example includes Activated, so that we

-    // can make the CoreWindow active and start rendering on the window.

-    applicationView->Activated += 

-        ref new TypedEventHandler<CoreApplicationView^, IActivatedEventArgs^>(this, &App::OnActivated);

-

-    // Logic for other event handlers could go here.

-    // Information about the Suspending and Resuming event handlers can be found here:

-    // http://msdn.microsoft.com/en-us/library/windows/apps/xaml/hh994930.aspx

-

-	os = new OSWinrt;

-}

-

-// Called when the CoreWindow object is created (or re-created).

-void App::SetWindow(CoreWindow^ p_window)

-{

-	window = p_window;

-    window->VisibilityChanged +=

-        ref new TypedEventHandler<CoreWindow^, VisibilityChangedEventArgs^>(this, &App::OnVisibilityChanged);

-

-    window->Closed += 

-        ref new TypedEventHandler<CoreWindow^, CoreWindowEventArgs^>(this, &App::OnWindowClosed);

-

-    window->SizeChanged += 

-        ref new TypedEventHandler<CoreWindow^, WindowSizeChangedEventArgs^>(this, &App::OnWindowSizeChanged);

-

-#if !(WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)

-    // Disable all pointer visual feedback for better performance when touching.

-    // This is not supported on Windows Phone applications.

-    auto pointerVisualizationSettings = PointerVisualizationSettings::GetForCurrentView();

-    pointerVisualizationSettings->IsContactFeedbackEnabled = false;

-    pointerVisualizationSettings->IsBarrelButtonFeedbackEnabled = false;

-#endif

-

-

-	window->PointerPressed +=

-		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerPressed);

-

-	window->PointerMoved +=

-		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerMoved);

-

-	window->PointerReleased +=

-		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerReleased);

-

-	//window->PointerWheelChanged +=

-	//	ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerWheelChanged);

-

-

-

-	char* args[] = {"-path", "game", NULL};

-	Main::setup("winrt", 2, args, false);

-

-	// The CoreWindow has been created, so EGL can be initialized.

-	ContextEGL* context = memnew(ContextEGL(window));

-	os->set_gl_context(context);

-	UpdateWindowSize(Size(window->Bounds.Width, window->Bounds.Height));

-

-	Main::setup2();

-}

-

-static int _get_button(Windows::UI::Input::PointerPoint ^pt) {

-

-	using namespace Windows::UI::Input;

-

-#if WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP

-	return BUTTON_LEFT;

-#else

-	switch (pt->Properties->PointerUpdateKind)

-	{

-		case PointerUpdateKind::LeftButtonPressed:

-		case PointerUpdateKind::LeftButtonReleased:

-			return BUTTON_LEFT;

-

-		case PointerUpdateKind::RightButtonPressed:

-		case PointerUpdateKind::RightButtonReleased:

-			return BUTTON_RIGHT;

-

-		case PointerUpdateKind::MiddleButtonPressed:

-		case PointerUpdateKind::MiddleButtonReleased:

-			return BUTTON_MIDDLE;

-

-		case PointerUpdateKind::XButton1Pressed:

-		case PointerUpdateKind::XButton1Released:

-			return BUTTON_WHEEL_UP;

-

-		case PointerUpdateKind::XButton2Pressed:

-		case PointerUpdateKind::XButton2Released:

-			return BUTTON_WHEEL_DOWN;

-

-		default:

-			break;

-	}

-#endif

-

-	return 0;

-};

-

-static bool _is_touch(Windows::UI::Input::PointerPoint ^pointerPoint) {

-#if WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP

-	return true;

-#else

-	using namespace Windows::Devices::Input;

-	switch (pointerPoint->PointerDevice->PointerDeviceType) {

-		case PointerDeviceType::Touch:

-		case PointerDeviceType::Pen:

-			return true;

-		default:

-			return false;

-	}

-#endif

-}

-

-

-static Windows::Foundation::Point _get_pixel_position(CoreWindow^ window, Windows::Foundation::Point rawPosition, OS* os) {

-

-	Windows::Foundation::Point outputPosition;

-

-	// Compute coordinates normalized from 0..1.

-	// If the coordinates need to be sized to the SDL window,

-	// we'll do that after.

-	#if 1 || WINAPI_FAMILY != WINAPI_FAMILY_PHONE_APP

-	outputPosition.X = rawPosition.X / window->Bounds.Width;

-	outputPosition.Y = rawPosition.Y / window->Bounds.Height;

-	#else

-	switch (DisplayProperties::CurrentOrientation)

-	{

-		case DisplayOrientations::Portrait:

-			outputPosition.X = rawPosition.X / window->Bounds.Width;

-			outputPosition.Y = rawPosition.Y / window->Bounds.Height;

-			break;

-		case DisplayOrientations::PortraitFlipped:

-			outputPosition.X = 1.0f - (rawPosition.X / window->Bounds.Width);

-			outputPosition.Y = 1.0f - (rawPosition.Y / window->Bounds.Height);

-			break;

-		case DisplayOrientations::Landscape:

-			outputPosition.X = rawPosition.Y / window->Bounds.Height;

-			outputPosition.Y = 1.0f - (rawPosition.X / window->Bounds.Width);

-			break;

-		case DisplayOrientations::LandscapeFlipped:

-			outputPosition.X = 1.0f - (rawPosition.Y / window->Bounds.Height);

-			outputPosition.Y = rawPosition.X / window->Bounds.Width;

-			break;

-		default:

-			break;

-	}

-	#endif

-

-	OS::VideoMode vm = os->get_video_mode();

-	outputPosition.X *= vm.width;

-	outputPosition.Y *= vm.height;

-

-	return outputPosition;

-};

-

-static int _get_finger(uint32_t p_touch_id) {

-

-	return p_touch_id % 31; // for now

-};

-

-void App::pointer_event(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args, bool p_pressed) {

-

-	Windows::UI::Input::PointerPoint ^point = args->CurrentPoint;

-	Windows::Foundation::Point pos = _get_pixel_position(window, point->Position, os);

-	int but = _get_button(point);

-	if (_is_touch(point)) {

-

-		InputEvent event;

-		event.type = InputEvent::SCREEN_TOUCH;

-		event.device = 0;

-		event.screen_touch.pressed = p_pressed;

-		event.screen_touch.x = pos.X;

-		event.screen_touch.y = pos.Y;

-		event.screen_touch.index = _get_finger(point->PointerId);

-

-		last_touch_x[event.screen_touch.index] = pos.X;

-		last_touch_y[event.screen_touch.index] = pos.Y;

-

-		os->input_event(event);

-		if (event.screen_touch.index != 0)

-			return;

-

-	}; // fallthrought of sorts

-

-	InputEvent event;

-	event.type = InputEvent::MOUSE_BUTTON;

-	event.device = 0;

-	event.mouse_button.pressed = p_pressed;

-	event.mouse_button.button_index = but;

-	event.mouse_button.x = pos.X;

-	event.mouse_button.y = pos.Y;

-	event.mouse_button.global_x = pos.X;

-	event.mouse_button.global_y = pos.Y;

-

-	last_touch_x[31] = pos.X;

-	last_touch_y[31] = pos.Y;

-

-	os->input_event(event);

-};

-

-

-void App::OnPointerPressed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {

-

-	pointer_event(sender, args, true);

-};

-

-

-void App::OnPointerReleased(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {

-

-	pointer_event(sender, args, false);

-};

-

-void App::OnPointerMoved(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {

-

-	Windows::UI::Input::PointerPoint ^point = args->CurrentPoint;

-	Windows::Foundation::Point pos = _get_pixel_position(window, point->Position, os);

-

-	if (_is_touch(point)) {

-

-		InputEvent event;

-		event.type = InputEvent::SCREEN_DRAG;

-		event.device = 0;

-		event.screen_drag.x = pos.X;

-		event.screen_drag.y = pos.Y;

-		event.screen_drag.index = _get_finger(point->PointerId);

-		event.screen_drag.relative_x = event.screen_drag.x - last_touch_x[event.screen_drag.index];

-		event.screen_drag.relative_y = event.screen_drag.y - last_touch_y[event.screen_drag.index];

-

-		os->input_event(event);

-		if (event.screen_drag.index != 0)

-			return;

-

-	}; // fallthrought of sorts

-

-	InputEvent event;

-	event.type = InputEvent::MOUSE_MOTION;

-	event.device = 0;

-	event.mouse_motion.x = pos.X;

-	event.mouse_motion.y = pos.Y;

-	event.mouse_motion.global_x = pos.X;

-	event.mouse_motion.global_y = pos.Y;

-	event.mouse_motion.relative_x = pos.X - last_touch_x[31];

-	event.mouse_motion.relative_y = pos.Y - last_touch_y[31];

-

-	os->input_event(event);

-

-};

-

-

-// Initializes scene resources

-void App::Load(Platform::String^ entryPoint)

-{

-	//char* args[] = {"-test", "render", NULL};

-	//Main::setup("winrt", 2, args);

-}

-

-// This method is called after the window becomes active.

-void App::Run()

-{

-	if (Main::start())

-		os->run();

-}

-

-// Terminate events do not cause Uninitialize to be called. It will be called if your IFrameworkView

-// class is torn down while the app is in the foreground.

-void App::Uninitialize()

-{

-	Main::cleanup();

-	delete os;

-}

-

-// Application lifecycle event handler.

-void App::OnActivated(CoreApplicationView^ applicationView, IActivatedEventArgs^ args)

-{

-    // Run() won't start until the CoreWindow is activated.

-    CoreWindow::GetForCurrentThread()->Activate();

-}

-

-// Window event handlers.

-void App::OnVisibilityChanged(CoreWindow^ sender, VisibilityChangedEventArgs^ args)

-{

-    mWindowVisible = args->Visible;

-}

-

-void App::OnWindowClosed(CoreWindow^ sender, CoreWindowEventArgs^ args)

-{

-    mWindowClosed = true;

-}

-

-void App::OnWindowSizeChanged(CoreWindow^ sender, WindowSizeChangedEventArgs^ args)

-{

-#if (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP)

-    // On Windows 8.1, apps are resized when they are snapped alongside other apps, or when the device is rotated.

-    // The default framebuffer will be automatically resized when either of these occur.

-    // In particular, on a 90 degree rotation, the default framebuffer's width and height will switch.

-    UpdateWindowSize(args->Size);

-#else if (WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)

-    // On Windows Phone 8.1, the window size changes when the device is rotated.

-    // The default framebuffer will not be automatically resized when this occurs.

-    // It is therefore up to the app to handle rotation-specific logic in its rendering code.

-	//os->screen_size_changed();

-	UpdateWindowSize(args->Size);

-#endif

-}

-

-void App::UpdateWindowSize(Size size)

-{

-	float dpi;

-#if (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP)

-	DisplayInformation^ currentDisplayInformation = DisplayInformation::GetForCurrentView();

-	dpi = currentDisplayInformation->LogicalDpi;

-#else if (WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)

-	dpi = DisplayProperties::LogicalDpi;

-#endif

-	Size pixelSize(ConvertDipsToPixels(size.Width, dpi), ConvertDipsToPixels(size.Height, dpi));

-

-    mWindowWidth = static_cast<GLsizei>(pixelSize.Width);

-    mWindowHeight = static_cast<GLsizei>(pixelSize.Height);

-

-	OS::VideoMode vm;

-	vm.width = mWindowWidth;

-	vm.height = mWindowHeight;

-	vm.fullscreen = true;

-	vm.resizable = false;

-	os->set_video_mode(vm);

-}

+//
+// This file demonstrates how to initialize EGL in a Windows Store app, using ICoreWindow.
+//
+
+#include "app.h"
+
+#include "main/main.h"
+#include "core/os/dir_access.h"
+#include "core/os/file_access.h"
+
+using namespace Windows::ApplicationModel::Core;
+using namespace Windows::ApplicationModel::Activation;
+using namespace Windows::UI::Core;
+using namespace Windows::UI::Input;
+using namespace Windows::Foundation;
+using namespace Windows::Graphics::Display;
+using namespace Microsoft::WRL;
+using namespace Platform;
+
+using namespace $ext_safeprojectname$;
+
+// Helper to convert a length in device-independent pixels (DIPs) to a length in physical pixels.
+inline float ConvertDipsToPixels(float dips, float dpi)
+{
+    static const float dipsPerInch = 96.0f;
+    return floor(dips * dpi / dipsPerInch + 0.5f); // Round to nearest integer.
+}
+
+// Implementation of the IFrameworkViewSource interface, necessary to run our app.
+ref class HelloTriangleApplicationSource sealed : Windows::ApplicationModel::Core::IFrameworkViewSource
+{
+public:
+    virtual Windows::ApplicationModel::Core::IFrameworkView^ CreateView()
+    {
+        return ref new App();
+    }
+};
+
+// The main function creates an IFrameworkViewSource for our app, and runs the app.
+[Platform::MTAThread]
+int main(Platform::Array<Platform::String^>^)
+{
+    auto helloTriangleApplicationSource = ref new HelloTriangleApplicationSource();
+    CoreApplication::Run(helloTriangleApplicationSource);
+    return 0;
+}
+
+App::App() :
+    mWindowClosed(false),
+    mWindowVisible(true),
+    mWindowWidth(0),
+    mWindowHeight(0),
+    mEglDisplay(EGL_NO_DISPLAY),
+    mEglContext(EGL_NO_CONTEXT),
+    mEglSurface(EGL_NO_SURFACE)
+{
+}
+
+// The first method called when the IFrameworkView is being created.
+void App::Initialize(CoreApplicationView^ applicationView)
+{
+    // Register event handlers for app lifecycle. This example includes Activated, so that we
+    // can make the CoreWindow active and start rendering on the window.
+    applicationView->Activated += 
+        ref new TypedEventHandler<CoreApplicationView^, IActivatedEventArgs^>(this, &App::OnActivated);
+
+    // Logic for other event handlers could go here.
+    // Information about the Suspending and Resuming event handlers can be found here:
+    // http://msdn.microsoft.com/en-us/library/windows/apps/xaml/hh994930.aspx
+
+	os = new OSWinrt;
+}
+
+// Called when the CoreWindow object is created (or re-created).
+void App::SetWindow(CoreWindow^ p_window)
+{
+	window = p_window;
+    window->VisibilityChanged +=
+        ref new TypedEventHandler<CoreWindow^, VisibilityChangedEventArgs^>(this, &App::OnVisibilityChanged);
+
+    window->Closed += 
+        ref new TypedEventHandler<CoreWindow^, CoreWindowEventArgs^>(this, &App::OnWindowClosed);
+
+    window->SizeChanged += 
+        ref new TypedEventHandler<CoreWindow^, WindowSizeChangedEventArgs^>(this, &App::OnWindowSizeChanged);
+
+#if !(WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)
+    // Disable all pointer visual feedback for better performance when touching.
+    // This is not supported on Windows Phone applications.
+    auto pointerVisualizationSettings = PointerVisualizationSettings::GetForCurrentView();
+    pointerVisualizationSettings->IsContactFeedbackEnabled = false;
+    pointerVisualizationSettings->IsBarrelButtonFeedbackEnabled = false;
+#endif
+
+
+	window->PointerPressed +=
+		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerPressed);
+
+	window->PointerMoved +=
+		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerMoved);
+
+	window->PointerReleased +=
+		ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerReleased);
+
+	//window->PointerWheelChanged +=
+	//	ref new TypedEventHandler<CoreWindow^, PointerEventArgs^>(this, &App::OnPointerWheelChanged);
+
+
+
+	char* args[] = {"-path", "game", NULL};
+	Main::setup("winrt", 2, args, false);
+
+	// The CoreWindow has been created, so EGL can be initialized.
+	ContextEGL* context = memnew(ContextEGL(window));
+	os->set_gl_context(context);
+	UpdateWindowSize(Size(window->Bounds.Width, window->Bounds.Height));
+
+	Main::setup2();
+}
+
+static int _get_button(Windows::UI::Input::PointerPoint ^pt) {
+
+	using namespace Windows::UI::Input;
+
+#if WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP
+	return BUTTON_LEFT;
+#else
+	switch (pt->Properties->PointerUpdateKind)
+	{
+		case PointerUpdateKind::LeftButtonPressed:
+		case PointerUpdateKind::LeftButtonReleased:
+			return BUTTON_LEFT;
+
+		case PointerUpdateKind::RightButtonPressed:
+		case PointerUpdateKind::RightButtonReleased:
+			return BUTTON_RIGHT;
+
+		case PointerUpdateKind::MiddleButtonPressed:
+		case PointerUpdateKind::MiddleButtonReleased:
+			return BUTTON_MIDDLE;
+
+		case PointerUpdateKind::XButton1Pressed:
+		case PointerUpdateKind::XButton1Released:
+			return BUTTON_WHEEL_UP;
+
+		case PointerUpdateKind::XButton2Pressed:
+		case PointerUpdateKind::XButton2Released:
+			return BUTTON_WHEEL_DOWN;
+
+		default:
+			break;
+	}
+#endif
+
+	return 0;
+};
+
+static bool _is_touch(Windows::UI::Input::PointerPoint ^pointerPoint) {
+#if WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP
+	return true;
+#else
+	using namespace Windows::Devices::Input;
+	switch (pointerPoint->PointerDevice->PointerDeviceType) {
+		case PointerDeviceType::Touch:
+		case PointerDeviceType::Pen:
+			return true;
+		default:
+			return false;
+	}
+#endif
+}
+
+
+static Windows::Foundation::Point _get_pixel_position(CoreWindow^ window, Windows::Foundation::Point rawPosition, OS* os) {
+
+	Windows::Foundation::Point outputPosition;
+
+	// Compute coordinates normalized from 0..1.
+	// If the coordinates need to be sized to the SDL window,
+	// we'll do that after.
+	#if 1 || WINAPI_FAMILY != WINAPI_FAMILY_PHONE_APP
+	outputPosition.X = rawPosition.X / window->Bounds.Width;
+	outputPosition.Y = rawPosition.Y / window->Bounds.Height;
+	#else
+	switch (DisplayProperties::CurrentOrientation)
+	{
+		case DisplayOrientations::Portrait:
+			outputPosition.X = rawPosition.X / window->Bounds.Width;
+			outputPosition.Y = rawPosition.Y / window->Bounds.Height;
+			break;
+		case DisplayOrientations::PortraitFlipped:
+			outputPosition.X = 1.0f - (rawPosition.X / window->Bounds.Width);
+			outputPosition.Y = 1.0f - (rawPosition.Y / window->Bounds.Height);
+			break;
+		case DisplayOrientations::Landscape:
+			outputPosition.X = rawPosition.Y / window->Bounds.Height;
+			outputPosition.Y = 1.0f - (rawPosition.X / window->Bounds.Width);
+			break;
+		case DisplayOrientations::LandscapeFlipped:
+			outputPosition.X = 1.0f - (rawPosition.Y / window->Bounds.Height);
+			outputPosition.Y = rawPosition.X / window->Bounds.Width;
+			break;
+		default:
+			break;
+	}
+	#endif
+
+	OS::VideoMode vm = os->get_video_mode();
+	outputPosition.X *= vm.width;
+	outputPosition.Y *= vm.height;
+
+	return outputPosition;
+};
+
+static int _get_finger(uint32_t p_touch_id) {
+
+	return p_touch_id % 31; // for now
+};
+
+void App::pointer_event(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args, bool p_pressed) {
+
+	Windows::UI::Input::PointerPoint ^point = args->CurrentPoint;
+	Windows::Foundation::Point pos = _get_pixel_position(window, point->Position, os);
+	int but = _get_button(point);
+	if (_is_touch(point)) {
+
+		InputEvent event;
+		event.type = InputEvent::SCREEN_TOUCH;
+		event.device = 0;
+		event.screen_touch.pressed = p_pressed;
+		event.screen_touch.x = pos.X;
+		event.screen_touch.y = pos.Y;
+		event.screen_touch.index = _get_finger(point->PointerId);
+
+		last_touch_x[event.screen_touch.index] = pos.X;
+		last_touch_y[event.screen_touch.index] = pos.Y;
+
+		os->input_event(event);
+		if (event.screen_touch.index != 0)
+			return;
+
+	}; // fallthrought of sorts
+
+	InputEvent event;
+	event.type = InputEvent::MOUSE_BUTTON;
+	event.device = 0;
+	event.mouse_button.pressed = p_pressed;
+	event.mouse_button.button_index = but;
+	event.mouse_button.x = pos.X;
+	event.mouse_button.y = pos.Y;
+	event.mouse_button.global_x = pos.X;
+	event.mouse_button.global_y = pos.Y;
+
+	last_touch_x[31] = pos.X;
+	last_touch_y[31] = pos.Y;
+
+	os->input_event(event);
+};
+
+
+void App::OnPointerPressed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {
+
+	pointer_event(sender, args, true);
+};
+
+
+void App::OnPointerReleased(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {
+
+	pointer_event(sender, args, false);
+};
+
+void App::OnPointerMoved(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args) {
+
+	Windows::UI::Input::PointerPoint ^point = args->CurrentPoint;
+	Windows::Foundation::Point pos = _get_pixel_position(window, point->Position, os);
+
+	if (_is_touch(point)) {
+
+		InputEvent event;
+		event.type = InputEvent::SCREEN_DRAG;
+		event.device = 0;
+		event.screen_drag.x = pos.X;
+		event.screen_drag.y = pos.Y;
+		event.screen_drag.index = _get_finger(point->PointerId);
+		event.screen_drag.relative_x = event.screen_drag.x - last_touch_x[event.screen_drag.index];
+		event.screen_drag.relative_y = event.screen_drag.y - last_touch_y[event.screen_drag.index];
+
+		os->input_event(event);
+		if (event.screen_drag.index != 0)
+			return;
+
+	}; // fallthrought of sorts
+
+	InputEvent event;
+	event.type = InputEvent::MOUSE_MOTION;
+	event.device = 0;
+	event.mouse_motion.x = pos.X;
+	event.mouse_motion.y = pos.Y;
+	event.mouse_motion.global_x = pos.X;
+	event.mouse_motion.global_y = pos.Y;
+	event.mouse_motion.relative_x = pos.X - last_touch_x[31];
+	event.mouse_motion.relative_y = pos.Y - last_touch_y[31];
+
+	os->input_event(event);
+
+};
+
+
+// Initializes scene resources
+void App::Load(Platform::String^ entryPoint)
+{
+	//char* args[] = {"-test", "render", NULL};
+	//Main::setup("winrt", 2, args);
+}
+
+// This method is called after the window becomes active.
+void App::Run()
+{
+	if (Main::start())
+		os->run();
+}
+
+// Terminate events do not cause Uninitialize to be called. It will be called if your IFrameworkView
+// class is torn down while the app is in the foreground.
+void App::Uninitialize()
+{
+	Main::cleanup();
+	delete os;
+}
+
+// Application lifecycle event handler.
+void App::OnActivated(CoreApplicationView^ applicationView, IActivatedEventArgs^ args)
+{
+    // Run() won't start until the CoreWindow is activated.
+    CoreWindow::GetForCurrentThread()->Activate();
+}
+
+// Window event handlers.
+void App::OnVisibilityChanged(CoreWindow^ sender, VisibilityChangedEventArgs^ args)
+{
+    mWindowVisible = args->Visible;
+}
+
+void App::OnWindowClosed(CoreWindow^ sender, CoreWindowEventArgs^ args)
+{
+    mWindowClosed = true;
+}
+
+void App::OnWindowSizeChanged(CoreWindow^ sender, WindowSizeChangedEventArgs^ args)
+{
+#if (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP)
+    // On Windows 8.1, apps are resized when they are snapped alongside other apps, or when the device is rotated.
+    // The default framebuffer will be automatically resized when either of these occur.
+    // In particular, on a 90 degree rotation, the default framebuffer's width and height will switch.
+    UpdateWindowSize(args->Size);
+#else if (WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)
+    // On Windows Phone 8.1, the window size changes when the device is rotated.
+    // The default framebuffer will not be automatically resized when this occurs.
+    // It is therefore up to the app to handle rotation-specific logic in its rendering code.
+	//os->screen_size_changed();
+	UpdateWindowSize(args->Size);
+#endif
+}
+
+void App::UpdateWindowSize(Size size)
+{
+	float dpi;
+#if (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP)
+	DisplayInformation^ currentDisplayInformation = DisplayInformation::GetForCurrentView();
+	dpi = currentDisplayInformation->LogicalDpi;
+#else if (WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP)
+	dpi = DisplayProperties::LogicalDpi;
+#endif
+	Size pixelSize(ConvertDipsToPixels(size.Width, dpi), ConvertDipsToPixels(size.Height, dpi));
+
+    mWindowWidth = static_cast<GLsizei>(pixelSize.Width);
+    mWindowHeight = static_cast<GLsizei>(pixelSize.Height);
+
+	OS::VideoMode vm;
+	vm.width = mWindowWidth;
+	vm.height = mWindowHeight;
+	vm.fullscreen = true;
+	vm.resizable = false;
+	os->set_video_mode(vm);
+}
diff --git a/platform/winrt/app.h b/platform/winrt/app.h
index 7926465ff8..9ce2fe560f 100644
--- a/platform/winrt/app.h
+++ b/platform/winrt/app.h
@@ -1,61 +1,61 @@
-#pragma once

-

-#include <string>

-

-#include <wrl.h>

-

-#include "os_winrt.h"

-#include "GLES2/gl2.h"

-

-namespace $ext_safeprojectname$

-{

-    ref class App sealed : public Windows::ApplicationModel::Core::IFrameworkView

-    {

-    public:

-        App();

-

-        // IFrameworkView Methods.

-        virtual void Initialize(Windows::ApplicationModel::Core::CoreApplicationView^ applicationView);

-        virtual void SetWindow(Windows::UI::Core::CoreWindow^ window);

-        virtual void Load(Platform::String^ entryPoint);

-        virtual void Run();

-        virtual void Uninitialize();

-

-    private:

-		void RecreateRenderer();

-

-        // Application lifecycle event handlers.

-        void OnActivated(Windows::ApplicationModel::Core::CoreApplicationView^ applicationView, Windows::ApplicationModel::Activation::IActivatedEventArgs^ args);

-

-        // Window event handlers.

-        void OnWindowSizeChanged(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::WindowSizeChangedEventArgs^ args);

-        void OnVisibilityChanged(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::VisibilityChangedEventArgs^ args);

-        void OnWindowClosed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::CoreWindowEventArgs^ args);

-

-		void pointer_event(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args, bool p_pressed);

-		void OnPointerPressed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);

-		void OnPointerReleased(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);

-		void OnPointerMoved(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);

-

-

-        void UpdateWindowSize(Windows::Foundation::Size size);

-        void InitializeEGL(Windows::UI::Core::CoreWindow^ window);

-        void CleanupEGL();

-

-        bool mWindowClosed;

-        bool mWindowVisible;

-        GLsizei mWindowWidth;

-        GLsizei mWindowHeight;

-        

-        EGLDisplay mEglDisplay;

-        EGLContext mEglContext;

-        EGLSurface mEglSurface;

-

-		CoreWindow^ window;

-		OSWinrt* os;

-

-		int last_touch_x[32]; // 20 fingers, index 31 reserved for the mouse

-		int last_touch_y[32];

-	};

-

-}

+#pragma once
+
+#include <string>
+
+#include <wrl.h>
+
+#include "os_winrt.h"
+#include "GLES2/gl2.h"
+
+namespace $ext_safeprojectname$
+{
+    ref class App sealed : public Windows::ApplicationModel::Core::IFrameworkView
+    {
+    public:
+        App();
+
+        // IFrameworkView Methods.
+        virtual void Initialize(Windows::ApplicationModel::Core::CoreApplicationView^ applicationView);
+        virtual void SetWindow(Windows::UI::Core::CoreWindow^ window);
+        virtual void Load(Platform::String^ entryPoint);
+        virtual void Run();
+        virtual void Uninitialize();
+
+    private:
+		void RecreateRenderer();
+
+        // Application lifecycle event handlers.
+        void OnActivated(Windows::ApplicationModel::Core::CoreApplicationView^ applicationView, Windows::ApplicationModel::Activation::IActivatedEventArgs^ args);
+
+        // Window event handlers.
+        void OnWindowSizeChanged(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::WindowSizeChangedEventArgs^ args);
+        void OnVisibilityChanged(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::VisibilityChangedEventArgs^ args);
+        void OnWindowClosed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::CoreWindowEventArgs^ args);
+
+		void pointer_event(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args, bool p_pressed);
+		void OnPointerPressed(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);
+		void OnPointerReleased(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);
+		void OnPointerMoved(Windows::UI::Core::CoreWindow^ sender, Windows::UI::Core::PointerEventArgs^ args);
+
+
+        void UpdateWindowSize(Windows::Foundation::Size size);
+        void InitializeEGL(Windows::UI::Core::CoreWindow^ window);
+        void CleanupEGL();
+
+        bool mWindowClosed;
+        bool mWindowVisible;
+        GLsizei mWindowWidth;
+        GLsizei mWindowHeight;
+        
+        EGLDisplay mEglDisplay;
+        EGLContext mEglContext;
+        EGLSurface mEglSurface;
+
+		CoreWindow^ window;
+		OSWinrt* os;
+
+		int last_touch_x[32]; // 20 fingers, index 31 reserved for the mouse
+		int last_touch_y[32];
+	};
+
+}
diff --git a/platform/winrt/detect.py b/platform/winrt/detect.py
index d97d974a19..7439f0cd55 100644
--- a/platform/winrt/detect.py
+++ b/platform/winrt/detect.py
@@ -1,156 +1,156 @@
-

-

-import os

-

-import sys	

-import string

-

-

-def is_active():

-	return True

-        

-def get_name():

-		return "WinRT"

-

-def can_build():

-	if (os.name=="nt"):

-		#building natively on windows!

-		if (os.getenv("VSINSTALLDIR")):

-			return True 

-	return False

-		

-def get_opts():

-	return []

-  

-def get_flags():

-

-	return []

-

-

-def configure(env):

-

-	env.Append(CPPPATH=['#platform/winrt', '#platform/winrt/include'])

-	arch = ""

-

-	if os.getenv('PLATFORM') == "ARM":

-

-		# compiler commandline

-		# debug:   /Yu"pch.h" /MP /GS     /analyze- /W3 /wd"4453" /wd"28204"     /Zc:wchar_t /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.WindowsPhone\" /I"Generated Files\" /I"ARM\Debug\"   /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.Shared\" /ZW:nostdlib /Zi /Gm- /Od /sdl /Fd"ARM\Debug\vc120.pdb"   /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE" /D "_DEBUG" /errorReport:prompt /WX- /Zc:forScope /RTC1 /ZW /Gd /Oy- /MDd    /Fa"ARM\Debug\"   /EHsc /nologo /Fo"ARM\Debug\"   /Fp"ARM\Debug\App2.WindowsPhone.pch"

-		# release: /Yu"pch.h" /MP /GS /GL /analyze- /W3 /wd"4453" /wd"28204" /Gy /Zc:wchar_t /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.WindowsPhone\" /I"Generated Files\" /I"ARM\Release\" /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.Shared\" /ZW:nostdlib /Zi /Gm- /O2 /sdl /Fd"ARM\Release\vc120.pdb" /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE"             /errorReport:prompt /WX- /Zc:forScope       /ZW /Gd /Oy- /Oi /MD /Fa"ARM\Release\" /EHsc /nologo /Fo"ARM\Release\" /Fp"ARM\Release\App2.WindowsPhone.pch"

-

-		# linker commandline

-		# debug:   /OUT:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.exe"   /MANIFEST:NO       /NXCOMPAT /PDB:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.pdb"   /DYNAMICBASE "WindowsPhoneCore.lib" "RuntimeObject.lib" "PhoneAppModelHost.lib" /DEBUG /MACHINE:ARM /NODEFAULTLIB:"kernel32.lib" /NODEFAULTLIB:"ole32.lib" /WINMD /APPCONTAINER /INCREMENTAL /PGD:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.pgd"   /WINMDFILE:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.winmd"   /SUBSYSTEM:WINDOWS /MANIFESTUAC:NO /ManifestFile:"ARM\Debug\App2.WindowsPhone.exe.intermediate.manifest"            /ERRORREPORT:PROMPT /NOLOGO /TLBID:1

-		# release: /OUT:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.exe" /MANIFEST:NO /LTCG /NXCOMPAT /PDB:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.pdb" /DYNAMICBASE "WindowsPhoneCore.lib" "RuntimeObject.lib" "PhoneAppModelHost.lib" /DEBUG /MACHINE:ARM /NODEFAULTLIB:"kernel32.lib" /NODEFAULTLIB:"ole32.lib" /WINMD /APPCONTAINER /OPT:REF     /PGD:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.pgd" /WINMDFILE:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.winmd" /SUBSYSTEM:WINDOWS /MANIFESTUAC:NO /ManifestFile:"ARM\Release\App2.WindowsPhone.exe.intermediate.manifest" /OPT:ICF /ERRORREPORT:PROMPT /NOLOGO /TLBID:1

-

-		arch = "arm"

-

-		env.Append(LINKFLAGS=['/INCREMENTAL:NO', '/MANIFEST:NO', '/NXCOMPAT', '/DYNAMICBASE', "WindowsPhoneCore.lib", "RuntimeObject.lib", "PhoneAppModelHost.lib", "/DEBUG", "/MACHINE:ARM", '/NODEFAULTLIB:"kernel32.lib"', '/NODEFAULTLIB:"ole32.lib"', '/WINMD', '/APPCONTAINER', '/MANIFESTUAC:NO', '/ERRORREPORT:PROMPT', '/NOLOGO', '/TLBID:1'])

-		env.Append(LIBPATH=['#platform/winrt/ARM/lib'])

-

-		env.Append(CCFLAGS=string.split('/MP /GS /wd"4453" /wd"28204" /analyze- /Zc:wchar_t /Zi /Gm- /Od /fp:precise /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /DWINDOWSPHONE_ENABLED /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE" /errorReport:prompt /WX- /Zc:forScope /Gd /Oy- /Oi /MD /RTC1 /Gd /EHsc /nologo'))

-		env.Append(CXXFLAGS=string.split('/ZW'))

-

-		if (env["target"]=="release"):

-

-			env.Append(CCFLAGS=['/O2'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:WINDOWS'])

-

-		elif (env["target"]=="test"):

-

-			env.Append(CCFLAGS=['/O2','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])

-

-		elif (env["target"]=="debug"):

-

-			env.Append(CCFLAGS=['/Zi','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])

-			env.Append(LINKFLAGS=['/DEBUG', '/D_DEBUG'])

-

-		elif (env["target"]=="profile"):

-

-			env.Append(CCFLAGS=['-g','-pg'])

-			env.Append(LINKFLAGS=['-pg'])

-

-

-		env['ENV'] = os.environ;

-		# fix environment for windows phone 8.1

-		env['ENV']['WINDOWSPHONEKITDIR'] = env['ENV']['WINDOWSPHONEKITDIR'].replace("8.0", "8.1") # wtf

-		env['ENV']['INCLUDE'] = env['ENV']['INCLUDE'].replace("8.0", "8.1")

-		env['ENV']['LIB'] = env['ENV']['LIB'].replace("8.0", "8.1")

-		env['ENV']['PATH'] = env['ENV']['PATH'].replace("8.0", "8.1")

-		env['ENV']['LIBPATH'] = env['ENV']['LIBPATH'].replace("8.0\\Windows Metadata", "8.1\\References\\CommonConfiguration\\Neutral")

-

-	else:

-

-		arch = "x64"

-		env.Append(LINKFLAGS=['/MANIFEST:NO', '/NXCOMPAT', '/DYNAMICBASE', "kernel32.lib", '/MACHINE:X64', '/WINMD', '/APPCONTAINER', '/MANIFESTUAC:NO', '/ERRORREPORT:PROMPT', '/NOLOGO', '/TLBID:1'])

-

-		env.Append(LIBPATH=['#platform/winrt/x64/lib'])

-

-

-		if (env["target"]=="release"):

-

-			env.Append(CCFLAGS=['/O2'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:WINDOWS'])

-			env.Append(LINKFLAGS=['/ENTRY:mainCRTStartup'])

-

-		elif (env["target"]=="test"):

-

-			env.Append(CCFLAGS=['/O2','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])

-

-		elif (env["target"]=="debug"):

-

-			env.Append(CCFLAGS=['/Zi','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])

-			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])

-			env.Append(LINKFLAGS=['/DEBUG', '/D_DEBUG'])

-

-		elif (env["target"]=="profile"):

-

-			env.Append(CCFLAGS=['-g','-pg'])

-			env.Append(LINKFLAGS=['-pg'])

-

-

-		env.Append(CCFLAGS=string.split('/MP /GS /wd"4453" /wd"28204" /Zc:wchar_t /Gm- /Od /fp:precise /D "_UNICODE" /D "UNICODE" /D "WINAPI_FAMILY=WINAPI_FAMILY_APP" /errorReport:prompt /WX- /Zc:forScope /RTC1 /Gd /MDd /EHsc /nologo'))

-		env.Append(CXXFLAGS=string.split('/ZW'))

-		env.Append(CCFLAGS=['/AI', os.environ['VCINSTALLDIR']+'\\vcpackages', '/AI', os.environ['WINDOWSSDKDIR']+'\\References\\CommonConfiguration\\Neutral'])

-		env.Append(CCFLAGS=['/DWINAPI_FAMILY=WINAPI_FAMILY_APP', '/D_WIN32_WINNT=0x0603', '/DNTDDI_VERSION=0x06030000'])

-

-		env['ENV'] = os.environ;

-

-

-	env["PROGSUFFIX"]="."+arch+env["PROGSUFFIX"]

-	env["OBJSUFFIX"]="."+arch+env["OBJSUFFIX"]

-	env["LIBSUFFIX"]="."+arch+env["LIBSUFFIX"]

-

-

-	#env.Append(CCFLAGS=['/Gd','/GR','/nologo', '/EHsc'])

-	#env.Append(CXXFLAGS=['/TP', '/ZW'])

-	#env.Append(CPPFLAGS=['/DMSVC', '/GR', ])

-	##env.Append(CCFLAGS=['/I'+os.getenv("WindowsSdkDir")+"/Include"])

-	env.Append(CCFLAGS=['/DWINRT_ENABLED'])

-	env.Append(CCFLAGS=['/DWINDOWS_ENABLED'])

-	env.Append(CCFLAGS=['/DRTAUDIO_ENABLED'])

-	#env.Append(CCFLAGS=['/DWIN32'])

-	env.Append(CCFLAGS=['/DTYPED_METHOD_BIND'])

-

-	env.Append(CCFLAGS=['/DGLES2_ENABLED'])

-	#env.Append(CCFLAGS=['/DGLES1_ENABLED'])

-

-	LIBS=[

-		#'winmm',

-		'libEGL',

-		'libGLESv2',

-		'libANGLE',

-		#'kernel32','ole32','user32', 'advapi32'

-		]

-	env.Append(LINKFLAGS=[p+".lib" for p in LIBS])

-

-	import methods

-	env.Append( BUILDERS = { 'GLSL120' : env.Builder(action = methods.build_legacygl_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )

-	env.Append( BUILDERS = { 'GLSL' : env.Builder(action = methods.build_glsl_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )

-	env.Append( BUILDERS = { 'HLSL9' : env.Builder(action = methods.build_hlsl_dx9_headers, suffix = 'hlsl.h',src_suffix = '.hlsl') } )

-	env.Append( BUILDERS = { 'GLSL120GLES' : env.Builder(action = methods.build_gles2_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )

-

-

-#/c/Program Files (x86)/Windows Phone Kits/8.1/lib/ARM/WindowsPhoneCore.lib

+
+
+import os
+
+import sys	
+import string
+
+
+def is_active():
+	return True
+        
+def get_name():
+		return "WinRT"
+
+def can_build():
+	if (os.name=="nt"):
+		#building natively on windows!
+		if (os.getenv("VSINSTALLDIR")):
+			return True 
+	return False
+		
+def get_opts():
+	return []
+  
+def get_flags():
+
+	return []
+
+
+def configure(env):
+
+	env.Append(CPPPATH=['#platform/winrt', '#platform/winrt/include'])
+	arch = ""
+
+	if os.getenv('PLATFORM') == "ARM":
+
+		# compiler commandline
+		# debug:   /Yu"pch.h" /MP /GS     /analyze- /W3 /wd"4453" /wd"28204"     /Zc:wchar_t /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.WindowsPhone\" /I"Generated Files\" /I"ARM\Debug\"   /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.Shared\" /ZW:nostdlib /Zi /Gm- /Od /sdl /Fd"ARM\Debug\vc120.pdb"   /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE" /D "_DEBUG" /errorReport:prompt /WX- /Zc:forScope /RTC1 /ZW /Gd /Oy- /MDd    /Fa"ARM\Debug\"   /EHsc /nologo /Fo"ARM\Debug\"   /Fp"ARM\Debug\App2.WindowsPhone.pch"
+		# release: /Yu"pch.h" /MP /GS /GL /analyze- /W3 /wd"4453" /wd"28204" /Gy /Zc:wchar_t /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.WindowsPhone\" /I"Generated Files\" /I"ARM\Release\" /I"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\App2\App2.Shared\" /ZW:nostdlib /Zi /Gm- /O2 /sdl /Fd"ARM\Release\vc120.pdb" /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE"             /errorReport:prompt /WX- /Zc:forScope       /ZW /Gd /Oy- /Oi /MD /Fa"ARM\Release\" /EHsc /nologo /Fo"ARM\Release\" /Fp"ARM\Release\App2.WindowsPhone.pch"
+
+		# linker commandline
+		# debug:   /OUT:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.exe"   /MANIFEST:NO       /NXCOMPAT /PDB:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.pdb"   /DYNAMICBASE "WindowsPhoneCore.lib" "RuntimeObject.lib" "PhoneAppModelHost.lib" /DEBUG /MACHINE:ARM /NODEFAULTLIB:"kernel32.lib" /NODEFAULTLIB:"ole32.lib" /WINMD /APPCONTAINER /INCREMENTAL /PGD:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.WindowsPhone.pgd"   /WINMDFILE:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Debug\App2.WindowsPhone\App2.winmd"   /SUBSYSTEM:WINDOWS /MANIFESTUAC:NO /ManifestFile:"ARM\Debug\App2.WindowsPhone.exe.intermediate.manifest"            /ERRORREPORT:PROMPT /NOLOGO /TLBID:1
+		# release: /OUT:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.exe" /MANIFEST:NO /LTCG /NXCOMPAT /PDB:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.pdb" /DYNAMICBASE "WindowsPhoneCore.lib" "RuntimeObject.lib" "PhoneAppModelHost.lib" /DEBUG /MACHINE:ARM /NODEFAULTLIB:"kernel32.lib" /NODEFAULTLIB:"ole32.lib" /WINMD /APPCONTAINER /OPT:REF     /PGD:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.WindowsPhone.pgd" /WINMDFILE:"C:\Users\ariel\Documents\Visual Studio 2013\Projects\App2\ARM\Release\App2.WindowsPhone\App2.winmd" /SUBSYSTEM:WINDOWS /MANIFESTUAC:NO /ManifestFile:"ARM\Release\App2.WindowsPhone.exe.intermediate.manifest" /OPT:ICF /ERRORREPORT:PROMPT /NOLOGO /TLBID:1
+
+		arch = "arm"
+
+		env.Append(LINKFLAGS=['/INCREMENTAL:NO', '/MANIFEST:NO', '/NXCOMPAT', '/DYNAMICBASE', "WindowsPhoneCore.lib", "RuntimeObject.lib", "PhoneAppModelHost.lib", "/DEBUG", "/MACHINE:ARM", '/NODEFAULTLIB:"kernel32.lib"', '/NODEFAULTLIB:"ole32.lib"', '/WINMD', '/APPCONTAINER', '/MANIFESTUAC:NO', '/ERRORREPORT:PROMPT', '/NOLOGO', '/TLBID:1'])
+		env.Append(LIBPATH=['#platform/winrt/ARM/lib'])
+
+		env.Append(CCFLAGS=string.split('/MP /GS /wd"4453" /wd"28204" /analyze- /Zc:wchar_t /Zi /Gm- /Od /fp:precise /fp:precise /D "PSAPI_VERSION=2" /D "WINAPI_FAMILY=WINAPI_FAMILY_PHONE_APP" /DWINDOWSPHONE_ENABLED /D "_UITHREADCTXT_SUPPORT=0" /D "_UNICODE" /D "UNICODE" /errorReport:prompt /WX- /Zc:forScope /Gd /Oy- /Oi /MD /RTC1 /Gd /EHsc /nologo'))
+		env.Append(CXXFLAGS=string.split('/ZW'))
+
+		if (env["target"]=="release"):
+
+			env.Append(CCFLAGS=['/O2'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:WINDOWS'])
+
+		elif (env["target"]=="test"):
+
+			env.Append(CCFLAGS=['/O2','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])
+
+		elif (env["target"]=="debug"):
+
+			env.Append(CCFLAGS=['/Zi','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])
+			env.Append(LINKFLAGS=['/DEBUG', '/D_DEBUG'])
+
+		elif (env["target"]=="profile"):
+
+			env.Append(CCFLAGS=['-g','-pg'])
+			env.Append(LINKFLAGS=['-pg'])
+
+
+		env['ENV'] = os.environ;
+		# fix environment for windows phone 8.1
+		env['ENV']['WINDOWSPHONEKITDIR'] = env['ENV']['WINDOWSPHONEKITDIR'].replace("8.0", "8.1") # wtf
+		env['ENV']['INCLUDE'] = env['ENV']['INCLUDE'].replace("8.0", "8.1")
+		env['ENV']['LIB'] = env['ENV']['LIB'].replace("8.0", "8.1")
+		env['ENV']['PATH'] = env['ENV']['PATH'].replace("8.0", "8.1")
+		env['ENV']['LIBPATH'] = env['ENV']['LIBPATH'].replace("8.0\\Windows Metadata", "8.1\\References\\CommonConfiguration\\Neutral")
+
+	else:
+
+		arch = "x64"
+		env.Append(LINKFLAGS=['/MANIFEST:NO', '/NXCOMPAT', '/DYNAMICBASE', "kernel32.lib", '/MACHINE:X64', '/WINMD', '/APPCONTAINER', '/MANIFESTUAC:NO', '/ERRORREPORT:PROMPT', '/NOLOGO', '/TLBID:1'])
+
+		env.Append(LIBPATH=['#platform/winrt/x64/lib'])
+
+
+		if (env["target"]=="release"):
+
+			env.Append(CCFLAGS=['/O2'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:WINDOWS'])
+			env.Append(LINKFLAGS=['/ENTRY:mainCRTStartup'])
+
+		elif (env["target"]=="test"):
+
+			env.Append(CCFLAGS=['/O2','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])
+
+		elif (env["target"]=="debug"):
+
+			env.Append(CCFLAGS=['/Zi','/DDEBUG_ENABLED','/DD3D_DEBUG_INFO'])
+			env.Append(LINKFLAGS=['/SUBSYSTEM:CONSOLE'])
+			env.Append(LINKFLAGS=['/DEBUG', '/D_DEBUG'])
+
+		elif (env["target"]=="profile"):
+
+			env.Append(CCFLAGS=['-g','-pg'])
+			env.Append(LINKFLAGS=['-pg'])
+
+
+		env.Append(CCFLAGS=string.split('/MP /GS /wd"4453" /wd"28204" /Zc:wchar_t /Gm- /Od /fp:precise /D "_UNICODE" /D "UNICODE" /D "WINAPI_FAMILY=WINAPI_FAMILY_APP" /errorReport:prompt /WX- /Zc:forScope /RTC1 /Gd /MDd /EHsc /nologo'))
+		env.Append(CXXFLAGS=string.split('/ZW'))
+		env.Append(CCFLAGS=['/AI', os.environ['VCINSTALLDIR']+'\\vcpackages', '/AI', os.environ['WINDOWSSDKDIR']+'\\References\\CommonConfiguration\\Neutral'])
+		env.Append(CCFLAGS=['/DWINAPI_FAMILY=WINAPI_FAMILY_APP', '/D_WIN32_WINNT=0x0603', '/DNTDDI_VERSION=0x06030000'])
+
+		env['ENV'] = os.environ;
+
+
+	env["PROGSUFFIX"]="."+arch+env["PROGSUFFIX"]
+	env["OBJSUFFIX"]="."+arch+env["OBJSUFFIX"]
+	env["LIBSUFFIX"]="."+arch+env["LIBSUFFIX"]
+
+
+	#env.Append(CCFLAGS=['/Gd','/GR','/nologo', '/EHsc'])
+	#env.Append(CXXFLAGS=['/TP', '/ZW'])
+	#env.Append(CPPFLAGS=['/DMSVC', '/GR', ])
+	##env.Append(CCFLAGS=['/I'+os.getenv("WindowsSdkDir")+"/Include"])
+	env.Append(CCFLAGS=['/DWINRT_ENABLED'])
+	env.Append(CCFLAGS=['/DWINDOWS_ENABLED'])
+	env.Append(CCFLAGS=['/DRTAUDIO_ENABLED'])
+	#env.Append(CCFLAGS=['/DWIN32'])
+	env.Append(CCFLAGS=['/DTYPED_METHOD_BIND'])
+
+	env.Append(CCFLAGS=['/DGLES2_ENABLED'])
+	#env.Append(CCFLAGS=['/DGLES1_ENABLED'])
+
+	LIBS=[
+		#'winmm',
+		'libEGL',
+		'libGLESv2',
+		'libANGLE',
+		#'kernel32','ole32','user32', 'advapi32'
+		]
+	env.Append(LINKFLAGS=[p+".lib" for p in LIBS])
+
+	import methods
+	env.Append( BUILDERS = { 'GLSL120' : env.Builder(action = methods.build_legacygl_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )
+	env.Append( BUILDERS = { 'GLSL' : env.Builder(action = methods.build_glsl_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )
+	env.Append( BUILDERS = { 'HLSL9' : env.Builder(action = methods.build_hlsl_dx9_headers, suffix = 'hlsl.h',src_suffix = '.hlsl') } )
+	env.Append( BUILDERS = { 'GLSL120GLES' : env.Builder(action = methods.build_gles2_headers, suffix = 'glsl.h',src_suffix = '.glsl') } )
+
+
+#/c/Program Files (x86)/Windows Phone Kits/8.1/lib/ARM/WindowsPhoneCore.lib
diff --git a/platform/winrt/include/EGL/egl.h b/platform/winrt/include/EGL/egl.h
index fb6f9b71e0..12590a0e20 100644
--- a/platform/winrt/include/EGL/egl.h
+++ b/platform/winrt/include/EGL/egl.h
@@ -1,298 +1,298 @@
-#ifndef __egl_h_

-#define __egl_h_ 1

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

-** Copyright (c) 2013-2014 The Khronos Group Inc.

-**

-** Permission is hereby granted, free of charge, to any person obtaining a

-** copy of this software and/or associated documentation files (the

-** "Materials"), to deal in the Materials without restriction, including

-** without limitation the rights to use, copy, modify, merge, publish,

-** distribute, sublicense, and/or sell copies of the Materials, and to

-** permit persons to whom the Materials are furnished to do so, subject to

-** the following conditions:

-**

-** The above copyright notice and this permission notice shall be included

-** in all copies or substantial portions of the Materials.

-**

-** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY

-** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

-** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

-** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

-*/

-/*

-** This header is generated from the Khronos OpenGL / OpenGL ES XML

-** API Registry. The current version of the Registry, generator scripts

-** used to make the header, and the header can be found at

-**   http://www.opengl.org/registry/

-**

-** Khronos $Revision: 27018 $ on $Date: 2014-06-10 08:06:12 -0700 (Tue, 10 Jun 2014) $

-*/

-

-#include <EGL/eglplatform.h>

-

-/* Generated on date 20140610 */

-

-/* Generated C header for:

- * API: egl

- * Versions considered: .*

- * Versions emitted: .*

- * Default extensions included: None

- * Additional extensions included: _nomatch_^

- * Extensions removed: _nomatch_^

- */

-

-#ifndef EGL_VERSION_1_0

-#define EGL_VERSION_1_0 1

-typedef unsigned int EGLBoolean;

-typedef void *EGLDisplay;

-#include <KHR/khrplatform.h>

-#include <EGL/eglplatform.h>

-typedef void *EGLConfig;

-typedef void *EGLSurface;

-typedef void *EGLContext;

-typedef void (*__eglMustCastToProperFunctionPointerType)(void);

-#define EGL_ALPHA_SIZE                    0x3021

-#define EGL_BAD_ACCESS                    0x3002

-#define EGL_BAD_ALLOC                     0x3003

-#define EGL_BAD_ATTRIBUTE                 0x3004

-#define EGL_BAD_CONFIG                    0x3005

-#define EGL_BAD_CONTEXT                   0x3006

-#define EGL_BAD_CURRENT_SURFACE           0x3007

-#define EGL_BAD_DISPLAY                   0x3008

-#define EGL_BAD_MATCH                     0x3009

-#define EGL_BAD_NATIVE_PIXMAP             0x300A

-#define EGL_BAD_NATIVE_WINDOW             0x300B

-#define EGL_BAD_PARAMETER                 0x300C

-#define EGL_BAD_SURFACE                   0x300D

-#define EGL_BLUE_SIZE                     0x3022

-#define EGL_BUFFER_SIZE                   0x3020

-#define EGL_CONFIG_CAVEAT                 0x3027

-#define EGL_CONFIG_ID                     0x3028

-#define EGL_CORE_NATIVE_ENGINE            0x305B

-#define EGL_DEPTH_SIZE                    0x3025

-#define EGL_DONT_CARE                     ((EGLint)-1)

-#define EGL_DRAW                          0x3059

-#define EGL_EXTENSIONS                    0x3055

-#define EGL_FALSE                         0

-#define EGL_GREEN_SIZE                    0x3023

-#define EGL_HEIGHT                        0x3056

-#define EGL_LARGEST_PBUFFER               0x3058

-#define EGL_LEVEL                         0x3029

-#define EGL_MAX_PBUFFER_HEIGHT            0x302A

-#define EGL_MAX_PBUFFER_PIXELS            0x302B

-#define EGL_MAX_PBUFFER_WIDTH             0x302C

-#define EGL_NATIVE_RENDERABLE             0x302D

-#define EGL_NATIVE_VISUAL_ID              0x302E

-#define EGL_NATIVE_VISUAL_TYPE            0x302F

-#define EGL_NONE                          0x3038

-#define EGL_NON_CONFORMANT_CONFIG         0x3051

-#define EGL_NOT_INITIALIZED               0x3001

-#define EGL_NO_CONTEXT                    ((EGLContext)0)

-#define EGL_NO_DISPLAY                    ((EGLDisplay)0)

-#define EGL_NO_SURFACE                    ((EGLSurface)0)

-#define EGL_PBUFFER_BIT                   0x0001

-#define EGL_PIXMAP_BIT                    0x0002

-#define EGL_READ                          0x305A

-#define EGL_RED_SIZE                      0x3024

-#define EGL_SAMPLES                       0x3031

-#define EGL_SAMPLE_BUFFERS                0x3032

-#define EGL_SLOW_CONFIG                   0x3050

-#define EGL_STENCIL_SIZE                  0x3026

-#define EGL_SUCCESS                       0x3000

-#define EGL_SURFACE_TYPE                  0x3033

-#define EGL_TRANSPARENT_BLUE_VALUE        0x3035

-#define EGL_TRANSPARENT_GREEN_VALUE       0x3036

-#define EGL_TRANSPARENT_RED_VALUE         0x3037

-#define EGL_TRANSPARENT_RGB               0x3052

-#define EGL_TRANSPARENT_TYPE              0x3034

-#define EGL_TRUE                          1

-#define EGL_VENDOR                        0x3053

-#define EGL_VERSION                       0x3054

-#define EGL_WIDTH                         0x3057

-#define EGL_WINDOW_BIT                    0x0004

-EGLAPI EGLBoolean EGLAPIENTRY eglChooseConfig (EGLDisplay dpy, const EGLint *attrib_list, EGLConfig *configs, EGLint config_size, EGLint *num_config);

-EGLAPI EGLBoolean EGLAPIENTRY eglCopyBuffers (EGLDisplay dpy, EGLSurface surface, EGLNativePixmapType target);

-EGLAPI EGLContext EGLAPIENTRY eglCreateContext (EGLDisplay dpy, EGLConfig config, EGLContext share_context, const EGLint *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferSurface (EGLDisplay dpy, EGLConfig config, const EGLint *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurface (EGLDisplay dpy, EGLConfig config, EGLNativePixmapType pixmap, const EGLint *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreateWindowSurface (EGLDisplay dpy, EGLConfig config, EGLNativeWindowType win, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroyContext (EGLDisplay dpy, EGLContext ctx);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroySurface (EGLDisplay dpy, EGLSurface surface);

-EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigAttrib (EGLDisplay dpy, EGLConfig config, EGLint attribute, EGLint *value);

-EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigs (EGLDisplay dpy, EGLConfig *configs, EGLint config_size, EGLint *num_config);

-EGLAPI EGLDisplay EGLAPIENTRY eglGetCurrentDisplay (void);

-EGLAPI EGLSurface EGLAPIENTRY eglGetCurrentSurface (EGLint readdraw);

-EGLAPI EGLDisplay EGLAPIENTRY eglGetDisplay (EGLNativeDisplayType display_id);

-EGLAPI EGLint EGLAPIENTRY eglGetError (void);

-EGLAPI __eglMustCastToProperFunctionPointerType EGLAPIENTRY eglGetProcAddress (const char *procname);

-EGLAPI EGLBoolean EGLAPIENTRY eglInitialize (EGLDisplay dpy, EGLint *major, EGLint *minor);

-EGLAPI EGLBoolean EGLAPIENTRY eglMakeCurrent (EGLDisplay dpy, EGLSurface draw, EGLSurface read, EGLContext ctx);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryContext (EGLDisplay dpy, EGLContext ctx, EGLint attribute, EGLint *value);

-EGLAPI const char *EGLAPIENTRY eglQueryString (EGLDisplay dpy, EGLint name);

-EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLint *value);

-EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffers (EGLDisplay dpy, EGLSurface surface);

-EGLAPI EGLBoolean EGLAPIENTRY eglTerminate (EGLDisplay dpy);

-EGLAPI EGLBoolean EGLAPIENTRY eglWaitGL (void);

-EGLAPI EGLBoolean EGLAPIENTRY eglWaitNative (EGLint engine);

-#endif /* EGL_VERSION_1_0 */

-

-#ifndef EGL_VERSION_1_1

-#define EGL_VERSION_1_1 1

-#define EGL_BACK_BUFFER                   0x3084

-#define EGL_BIND_TO_TEXTURE_RGB           0x3039

-#define EGL_BIND_TO_TEXTURE_RGBA          0x303A

-#define EGL_CONTEXT_LOST                  0x300E

-#define EGL_MIN_SWAP_INTERVAL             0x303B

-#define EGL_MAX_SWAP_INTERVAL             0x303C

-#define EGL_MIPMAP_TEXTURE                0x3082

-#define EGL_MIPMAP_LEVEL                  0x3083

-#define EGL_NO_TEXTURE                    0x305C

-#define EGL_TEXTURE_2D                    0x305F

-#define EGL_TEXTURE_FORMAT                0x3080

-#define EGL_TEXTURE_RGB                   0x305D

-#define EGL_TEXTURE_RGBA                  0x305E

-#define EGL_TEXTURE_TARGET                0x3081

-EGLAPI EGLBoolean EGLAPIENTRY eglBindTexImage (EGLDisplay dpy, EGLSurface surface, EGLint buffer);

-EGLAPI EGLBoolean EGLAPIENTRY eglReleaseTexImage (EGLDisplay dpy, EGLSurface surface, EGLint buffer);

-EGLAPI EGLBoolean EGLAPIENTRY eglSurfaceAttrib (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLint value);

-EGLAPI EGLBoolean EGLAPIENTRY eglSwapInterval (EGLDisplay dpy, EGLint interval);

-#endif /* EGL_VERSION_1_1 */

-

-#ifndef EGL_VERSION_1_2

-#define EGL_VERSION_1_2 1

-typedef unsigned int EGLenum;

-typedef void *EGLClientBuffer;

-#define EGL_ALPHA_FORMAT                  0x3088

-#define EGL_ALPHA_FORMAT_NONPRE           0x308B

-#define EGL_ALPHA_FORMAT_PRE              0x308C

-#define EGL_ALPHA_MASK_SIZE               0x303E

-#define EGL_BUFFER_PRESERVED              0x3094

-#define EGL_BUFFER_DESTROYED              0x3095

-#define EGL_CLIENT_APIS                   0x308D

-#define EGL_COLORSPACE                    0x3087

-#define EGL_COLORSPACE_sRGB               0x3089

-#define EGL_COLORSPACE_LINEAR             0x308A

-#define EGL_COLOR_BUFFER_TYPE             0x303F

-#define EGL_CONTEXT_CLIENT_TYPE           0x3097

-#define EGL_DISPLAY_SCALING               10000

-#define EGL_HORIZONTAL_RESOLUTION         0x3090

-#define EGL_LUMINANCE_BUFFER              0x308F

-#define EGL_LUMINANCE_SIZE                0x303D

-#define EGL_OPENGL_ES_BIT                 0x0001

-#define EGL_OPENVG_BIT                    0x0002

-#define EGL_OPENGL_ES_API                 0x30A0

-#define EGL_OPENVG_API                    0x30A1

-#define EGL_OPENVG_IMAGE                  0x3096

-#define EGL_PIXEL_ASPECT_RATIO            0x3092

-#define EGL_RENDERABLE_TYPE               0x3040

-#define EGL_RENDER_BUFFER                 0x3086

-#define EGL_RGB_BUFFER                    0x308E

-#define EGL_SINGLE_BUFFER                 0x3085

-#define EGL_SWAP_BEHAVIOR                 0x3093

-#define EGL_UNKNOWN                       ((EGLint)-1)

-#define EGL_VERTICAL_RESOLUTION           0x3091

-EGLAPI EGLBoolean EGLAPIENTRY eglBindAPI (EGLenum api);

-EGLAPI EGLenum EGLAPIENTRY eglQueryAPI (void);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferFromClientBuffer (EGLDisplay dpy, EGLenum buftype, EGLClientBuffer buffer, EGLConfig config, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglReleaseThread (void);

-EGLAPI EGLBoolean EGLAPIENTRY eglWaitClient (void);

-#endif /* EGL_VERSION_1_2 */

-

-#ifndef EGL_VERSION_1_3

-#define EGL_VERSION_1_3 1

-#define EGL_CONFORMANT                    0x3042

-#define EGL_CONTEXT_CLIENT_VERSION        0x3098

-#define EGL_MATCH_NATIVE_PIXMAP           0x3041

-#define EGL_OPENGL_ES2_BIT                0x0004

-#define EGL_VG_ALPHA_FORMAT               0x3088

-#define EGL_VG_ALPHA_FORMAT_NONPRE        0x308B

-#define EGL_VG_ALPHA_FORMAT_PRE           0x308C

-#define EGL_VG_ALPHA_FORMAT_PRE_BIT       0x0040

-#define EGL_VG_COLORSPACE                 0x3087

-#define EGL_VG_COLORSPACE_sRGB            0x3089

-#define EGL_VG_COLORSPACE_LINEAR          0x308A

-#define EGL_VG_COLORSPACE_LINEAR_BIT      0x0020

-#endif /* EGL_VERSION_1_3 */

-

-#ifndef EGL_VERSION_1_4

-#define EGL_VERSION_1_4 1

-#define EGL_DEFAULT_DISPLAY               ((EGLNativeDisplayType)0)

-#define EGL_MULTISAMPLE_RESOLVE_BOX_BIT   0x0200

-#define EGL_MULTISAMPLE_RESOLVE           0x3099

-#define EGL_MULTISAMPLE_RESOLVE_DEFAULT   0x309A

-#define EGL_MULTISAMPLE_RESOLVE_BOX       0x309B

-#define EGL_OPENGL_API                    0x30A2

-#define EGL_OPENGL_BIT                    0x0008

-#define EGL_SWAP_BEHAVIOR_PRESERVED_BIT   0x0400

-EGLAPI EGLContext EGLAPIENTRY eglGetCurrentContext (void);

-#endif /* EGL_VERSION_1_4 */

-

-#ifndef EGL_VERSION_1_5

-#define EGL_VERSION_1_5 1

-typedef void *EGLSync;

-typedef intptr_t EGLAttrib;

-typedef khronos_utime_nanoseconds_t EGLTime;

-#define EGL_CONTEXT_MAJOR_VERSION         0x3098

-#define EGL_CONTEXT_MINOR_VERSION         0x30FB

-#define EGL_CONTEXT_OPENGL_PROFILE_MASK   0x30FD

-#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY 0x31BD

-#define EGL_NO_RESET_NOTIFICATION         0x31BE

-#define EGL_LOSE_CONTEXT_ON_RESET         0x31BF

-#define EGL_CONTEXT_OPENGL_CORE_PROFILE_BIT 0x00000001

-#define EGL_CONTEXT_OPENGL_COMPATIBILITY_PROFILE_BIT 0x00000002

-#define EGL_CONTEXT_OPENGL_DEBUG          0x31B0

-#define EGL_CONTEXT_OPENGL_FORWARD_COMPATIBLE 0x31B1

-#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS  0x31B2

-#define EGL_OPENGL_ES3_BIT                0x00000040

-#define EGL_CL_EVENT_HANDLE               0x309C

-#define EGL_SYNC_CL_EVENT                 0x30FE

-#define EGL_SYNC_CL_EVENT_COMPLETE        0x30FF

-#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE  0x30F0

-#define EGL_SYNC_TYPE                     0x30F7

-#define EGL_SYNC_STATUS                   0x30F1

-#define EGL_SYNC_CONDITION                0x30F8

-#define EGL_SIGNALED                      0x30F2

-#define EGL_UNSIGNALED                    0x30F3

-#define EGL_SYNC_FLUSH_COMMANDS_BIT       0x0001

-#define EGL_FOREVER                       0xFFFFFFFFFFFFFFFFull

-#define EGL_TIMEOUT_EXPIRED               0x30F5

-#define EGL_CONDITION_SATISFIED           0x30F6

-#define EGL_NO_SYNC                       ((EGLSync)0)

-#define EGL_SYNC_FENCE                    0x30F9

-#define EGL_GL_COLORSPACE                 0x309D

-#define EGL_GL_COLORSPACE_SRGB            0x3089

-#define EGL_GL_COLORSPACE_LINEAR          0x308A

-#define EGL_GL_RENDERBUFFER               0x30B9

-#define EGL_GL_TEXTURE_2D                 0x30B1

-#define EGL_GL_TEXTURE_LEVEL              0x30BC

-#define EGL_GL_TEXTURE_3D                 0x30B2

-#define EGL_GL_TEXTURE_ZOFFSET            0x30BD

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_X 0x30B3

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_X 0x30B4

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Y 0x30B5

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Y 0x30B6

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Z 0x30B7

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Z 0x30B8

-EGLAPI EGLSync EGLAPIENTRY eglCreateSync (EGLDisplay dpy, EGLenum type, const EGLAttrib *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroySync (EGLDisplay dpy, EGLSync sync);

-EGLAPI EGLint EGLAPIENTRY eglClientWaitSync (EGLDisplay dpy, EGLSync sync, EGLint flags, EGLTime timeout);

-EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttrib (EGLDisplay dpy, EGLSync sync, EGLint attribute, EGLAttrib *value);

-EGLAPI EGLDisplay EGLAPIENTRY eglGetPlatformDisplay (EGLenum platform, void *native_display, const EGLAttrib *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformWindowSurface (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLAttrib *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformPixmapSurface (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLAttrib *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglWaitSync (EGLDisplay dpy, EGLSync sync, EGLint flags);

-#endif /* EGL_VERSION_1_5 */

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif

+#ifndef __egl_h_
+#define __egl_h_ 1
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+** Copyright (c) 2013-2014 The Khronos Group Inc.
+**
+** Permission is hereby granted, free of charge, to any person obtaining a
+** copy of this software and/or associated documentation files (the
+** "Materials"), to deal in the Materials without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Materials, and to
+** permit persons to whom the Materials are furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be included
+** in all copies or substantial portions of the Materials.
+**
+** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
+*/
+/*
+** This header is generated from the Khronos OpenGL / OpenGL ES XML
+** API Registry. The current version of the Registry, generator scripts
+** used to make the header, and the header can be found at
+**   http://www.opengl.org/registry/
+**
+** Khronos $Revision: 27018 $ on $Date: 2014-06-10 08:06:12 -0700 (Tue, 10 Jun 2014) $
+*/
+
+#include <EGL/eglplatform.h>
+
+/* Generated on date 20140610 */
+
+/* Generated C header for:
+ * API: egl
+ * Versions considered: .*
+ * Versions emitted: .*
+ * Default extensions included: None
+ * Additional extensions included: _nomatch_^
+ * Extensions removed: _nomatch_^
+ */
+
+#ifndef EGL_VERSION_1_0
+#define EGL_VERSION_1_0 1
+typedef unsigned int EGLBoolean;
+typedef void *EGLDisplay;
+#include <KHR/khrplatform.h>
+#include <EGL/eglplatform.h>
+typedef void *EGLConfig;
+typedef void *EGLSurface;
+typedef void *EGLContext;
+typedef void (*__eglMustCastToProperFunctionPointerType)(void);
+#define EGL_ALPHA_SIZE                    0x3021
+#define EGL_BAD_ACCESS                    0x3002
+#define EGL_BAD_ALLOC                     0x3003
+#define EGL_BAD_ATTRIBUTE                 0x3004
+#define EGL_BAD_CONFIG                    0x3005
+#define EGL_BAD_CONTEXT                   0x3006
+#define EGL_BAD_CURRENT_SURFACE           0x3007
+#define EGL_BAD_DISPLAY                   0x3008
+#define EGL_BAD_MATCH                     0x3009
+#define EGL_BAD_NATIVE_PIXMAP             0x300A
+#define EGL_BAD_NATIVE_WINDOW             0x300B
+#define EGL_BAD_PARAMETER                 0x300C
+#define EGL_BAD_SURFACE                   0x300D
+#define EGL_BLUE_SIZE                     0x3022
+#define EGL_BUFFER_SIZE                   0x3020
+#define EGL_CONFIG_CAVEAT                 0x3027
+#define EGL_CONFIG_ID                     0x3028
+#define EGL_CORE_NATIVE_ENGINE            0x305B
+#define EGL_DEPTH_SIZE                    0x3025
+#define EGL_DONT_CARE                     ((EGLint)-1)
+#define EGL_DRAW                          0x3059
+#define EGL_EXTENSIONS                    0x3055
+#define EGL_FALSE                         0
+#define EGL_GREEN_SIZE                    0x3023
+#define EGL_HEIGHT                        0x3056
+#define EGL_LARGEST_PBUFFER               0x3058
+#define EGL_LEVEL                         0x3029
+#define EGL_MAX_PBUFFER_HEIGHT            0x302A
+#define EGL_MAX_PBUFFER_PIXELS            0x302B
+#define EGL_MAX_PBUFFER_WIDTH             0x302C
+#define EGL_NATIVE_RENDERABLE             0x302D
+#define EGL_NATIVE_VISUAL_ID              0x302E
+#define EGL_NATIVE_VISUAL_TYPE            0x302F
+#define EGL_NONE                          0x3038
+#define EGL_NON_CONFORMANT_CONFIG         0x3051
+#define EGL_NOT_INITIALIZED               0x3001
+#define EGL_NO_CONTEXT                    ((EGLContext)0)
+#define EGL_NO_DISPLAY                    ((EGLDisplay)0)
+#define EGL_NO_SURFACE                    ((EGLSurface)0)
+#define EGL_PBUFFER_BIT                   0x0001
+#define EGL_PIXMAP_BIT                    0x0002
+#define EGL_READ                          0x305A
+#define EGL_RED_SIZE                      0x3024
+#define EGL_SAMPLES                       0x3031
+#define EGL_SAMPLE_BUFFERS                0x3032
+#define EGL_SLOW_CONFIG                   0x3050
+#define EGL_STENCIL_SIZE                  0x3026
+#define EGL_SUCCESS                       0x3000
+#define EGL_SURFACE_TYPE                  0x3033
+#define EGL_TRANSPARENT_BLUE_VALUE        0x3035
+#define EGL_TRANSPARENT_GREEN_VALUE       0x3036
+#define EGL_TRANSPARENT_RED_VALUE         0x3037
+#define EGL_TRANSPARENT_RGB               0x3052
+#define EGL_TRANSPARENT_TYPE              0x3034
+#define EGL_TRUE                          1
+#define EGL_VENDOR                        0x3053
+#define EGL_VERSION                       0x3054
+#define EGL_WIDTH                         0x3057
+#define EGL_WINDOW_BIT                    0x0004
+EGLAPI EGLBoolean EGLAPIENTRY eglChooseConfig (EGLDisplay dpy, const EGLint *attrib_list, EGLConfig *configs, EGLint config_size, EGLint *num_config);
+EGLAPI EGLBoolean EGLAPIENTRY eglCopyBuffers (EGLDisplay dpy, EGLSurface surface, EGLNativePixmapType target);
+EGLAPI EGLContext EGLAPIENTRY eglCreateContext (EGLDisplay dpy, EGLConfig config, EGLContext share_context, const EGLint *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferSurface (EGLDisplay dpy, EGLConfig config, const EGLint *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurface (EGLDisplay dpy, EGLConfig config, EGLNativePixmapType pixmap, const EGLint *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreateWindowSurface (EGLDisplay dpy, EGLConfig config, EGLNativeWindowType win, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroyContext (EGLDisplay dpy, EGLContext ctx);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroySurface (EGLDisplay dpy, EGLSurface surface);
+EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigAttrib (EGLDisplay dpy, EGLConfig config, EGLint attribute, EGLint *value);
+EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigs (EGLDisplay dpy, EGLConfig *configs, EGLint config_size, EGLint *num_config);
+EGLAPI EGLDisplay EGLAPIENTRY eglGetCurrentDisplay (void);
+EGLAPI EGLSurface EGLAPIENTRY eglGetCurrentSurface (EGLint readdraw);
+EGLAPI EGLDisplay EGLAPIENTRY eglGetDisplay (EGLNativeDisplayType display_id);
+EGLAPI EGLint EGLAPIENTRY eglGetError (void);
+EGLAPI __eglMustCastToProperFunctionPointerType EGLAPIENTRY eglGetProcAddress (const char *procname);
+EGLAPI EGLBoolean EGLAPIENTRY eglInitialize (EGLDisplay dpy, EGLint *major, EGLint *minor);
+EGLAPI EGLBoolean EGLAPIENTRY eglMakeCurrent (EGLDisplay dpy, EGLSurface draw, EGLSurface read, EGLContext ctx);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryContext (EGLDisplay dpy, EGLContext ctx, EGLint attribute, EGLint *value);
+EGLAPI const char *EGLAPIENTRY eglQueryString (EGLDisplay dpy, EGLint name);
+EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLint *value);
+EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffers (EGLDisplay dpy, EGLSurface surface);
+EGLAPI EGLBoolean EGLAPIENTRY eglTerminate (EGLDisplay dpy);
+EGLAPI EGLBoolean EGLAPIENTRY eglWaitGL (void);
+EGLAPI EGLBoolean EGLAPIENTRY eglWaitNative (EGLint engine);
+#endif /* EGL_VERSION_1_0 */
+
+#ifndef EGL_VERSION_1_1
+#define EGL_VERSION_1_1 1
+#define EGL_BACK_BUFFER                   0x3084
+#define EGL_BIND_TO_TEXTURE_RGB           0x3039
+#define EGL_BIND_TO_TEXTURE_RGBA          0x303A
+#define EGL_CONTEXT_LOST                  0x300E
+#define EGL_MIN_SWAP_INTERVAL             0x303B
+#define EGL_MAX_SWAP_INTERVAL             0x303C
+#define EGL_MIPMAP_TEXTURE                0x3082
+#define EGL_MIPMAP_LEVEL                  0x3083
+#define EGL_NO_TEXTURE                    0x305C
+#define EGL_TEXTURE_2D                    0x305F
+#define EGL_TEXTURE_FORMAT                0x3080
+#define EGL_TEXTURE_RGB                   0x305D
+#define EGL_TEXTURE_RGBA                  0x305E
+#define EGL_TEXTURE_TARGET                0x3081
+EGLAPI EGLBoolean EGLAPIENTRY eglBindTexImage (EGLDisplay dpy, EGLSurface surface, EGLint buffer);
+EGLAPI EGLBoolean EGLAPIENTRY eglReleaseTexImage (EGLDisplay dpy, EGLSurface surface, EGLint buffer);
+EGLAPI EGLBoolean EGLAPIENTRY eglSurfaceAttrib (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLint value);
+EGLAPI EGLBoolean EGLAPIENTRY eglSwapInterval (EGLDisplay dpy, EGLint interval);
+#endif /* EGL_VERSION_1_1 */
+
+#ifndef EGL_VERSION_1_2
+#define EGL_VERSION_1_2 1
+typedef unsigned int EGLenum;
+typedef void *EGLClientBuffer;
+#define EGL_ALPHA_FORMAT                  0x3088
+#define EGL_ALPHA_FORMAT_NONPRE           0x308B
+#define EGL_ALPHA_FORMAT_PRE              0x308C
+#define EGL_ALPHA_MASK_SIZE               0x303E
+#define EGL_BUFFER_PRESERVED              0x3094
+#define EGL_BUFFER_DESTROYED              0x3095
+#define EGL_CLIENT_APIS                   0x308D
+#define EGL_COLORSPACE                    0x3087
+#define EGL_COLORSPACE_sRGB               0x3089
+#define EGL_COLORSPACE_LINEAR             0x308A
+#define EGL_COLOR_BUFFER_TYPE             0x303F
+#define EGL_CONTEXT_CLIENT_TYPE           0x3097
+#define EGL_DISPLAY_SCALING               10000
+#define EGL_HORIZONTAL_RESOLUTION         0x3090
+#define EGL_LUMINANCE_BUFFER              0x308F
+#define EGL_LUMINANCE_SIZE                0x303D
+#define EGL_OPENGL_ES_BIT                 0x0001
+#define EGL_OPENVG_BIT                    0x0002
+#define EGL_OPENGL_ES_API                 0x30A0
+#define EGL_OPENVG_API                    0x30A1
+#define EGL_OPENVG_IMAGE                  0x3096
+#define EGL_PIXEL_ASPECT_RATIO            0x3092
+#define EGL_RENDERABLE_TYPE               0x3040
+#define EGL_RENDER_BUFFER                 0x3086
+#define EGL_RGB_BUFFER                    0x308E
+#define EGL_SINGLE_BUFFER                 0x3085
+#define EGL_SWAP_BEHAVIOR                 0x3093
+#define EGL_UNKNOWN                       ((EGLint)-1)
+#define EGL_VERTICAL_RESOLUTION           0x3091
+EGLAPI EGLBoolean EGLAPIENTRY eglBindAPI (EGLenum api);
+EGLAPI EGLenum EGLAPIENTRY eglQueryAPI (void);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferFromClientBuffer (EGLDisplay dpy, EGLenum buftype, EGLClientBuffer buffer, EGLConfig config, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglReleaseThread (void);
+EGLAPI EGLBoolean EGLAPIENTRY eglWaitClient (void);
+#endif /* EGL_VERSION_1_2 */
+
+#ifndef EGL_VERSION_1_3
+#define EGL_VERSION_1_3 1
+#define EGL_CONFORMANT                    0x3042
+#define EGL_CONTEXT_CLIENT_VERSION        0x3098
+#define EGL_MATCH_NATIVE_PIXMAP           0x3041
+#define EGL_OPENGL_ES2_BIT                0x0004
+#define EGL_VG_ALPHA_FORMAT               0x3088
+#define EGL_VG_ALPHA_FORMAT_NONPRE        0x308B
+#define EGL_VG_ALPHA_FORMAT_PRE           0x308C
+#define EGL_VG_ALPHA_FORMAT_PRE_BIT       0x0040
+#define EGL_VG_COLORSPACE                 0x3087
+#define EGL_VG_COLORSPACE_sRGB            0x3089
+#define EGL_VG_COLORSPACE_LINEAR          0x308A
+#define EGL_VG_COLORSPACE_LINEAR_BIT      0x0020
+#endif /* EGL_VERSION_1_3 */
+
+#ifndef EGL_VERSION_1_4
+#define EGL_VERSION_1_4 1
+#define EGL_DEFAULT_DISPLAY               ((EGLNativeDisplayType)0)
+#define EGL_MULTISAMPLE_RESOLVE_BOX_BIT   0x0200
+#define EGL_MULTISAMPLE_RESOLVE           0x3099
+#define EGL_MULTISAMPLE_RESOLVE_DEFAULT   0x309A
+#define EGL_MULTISAMPLE_RESOLVE_BOX       0x309B
+#define EGL_OPENGL_API                    0x30A2
+#define EGL_OPENGL_BIT                    0x0008
+#define EGL_SWAP_BEHAVIOR_PRESERVED_BIT   0x0400
+EGLAPI EGLContext EGLAPIENTRY eglGetCurrentContext (void);
+#endif /* EGL_VERSION_1_4 */
+
+#ifndef EGL_VERSION_1_5
+#define EGL_VERSION_1_5 1
+typedef void *EGLSync;
+typedef intptr_t EGLAttrib;
+typedef khronos_utime_nanoseconds_t EGLTime;
+#define EGL_CONTEXT_MAJOR_VERSION         0x3098
+#define EGL_CONTEXT_MINOR_VERSION         0x30FB
+#define EGL_CONTEXT_OPENGL_PROFILE_MASK   0x30FD
+#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY 0x31BD
+#define EGL_NO_RESET_NOTIFICATION         0x31BE
+#define EGL_LOSE_CONTEXT_ON_RESET         0x31BF
+#define EGL_CONTEXT_OPENGL_CORE_PROFILE_BIT 0x00000001
+#define EGL_CONTEXT_OPENGL_COMPATIBILITY_PROFILE_BIT 0x00000002
+#define EGL_CONTEXT_OPENGL_DEBUG          0x31B0
+#define EGL_CONTEXT_OPENGL_FORWARD_COMPATIBLE 0x31B1
+#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS  0x31B2
+#define EGL_OPENGL_ES3_BIT                0x00000040
+#define EGL_CL_EVENT_HANDLE               0x309C
+#define EGL_SYNC_CL_EVENT                 0x30FE
+#define EGL_SYNC_CL_EVENT_COMPLETE        0x30FF
+#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE  0x30F0
+#define EGL_SYNC_TYPE                     0x30F7
+#define EGL_SYNC_STATUS                   0x30F1
+#define EGL_SYNC_CONDITION                0x30F8
+#define EGL_SIGNALED                      0x30F2
+#define EGL_UNSIGNALED                    0x30F3
+#define EGL_SYNC_FLUSH_COMMANDS_BIT       0x0001
+#define EGL_FOREVER                       0xFFFFFFFFFFFFFFFFull
+#define EGL_TIMEOUT_EXPIRED               0x30F5
+#define EGL_CONDITION_SATISFIED           0x30F6
+#define EGL_NO_SYNC                       ((EGLSync)0)
+#define EGL_SYNC_FENCE                    0x30F9
+#define EGL_GL_COLORSPACE                 0x309D
+#define EGL_GL_COLORSPACE_SRGB            0x3089
+#define EGL_GL_COLORSPACE_LINEAR          0x308A
+#define EGL_GL_RENDERBUFFER               0x30B9
+#define EGL_GL_TEXTURE_2D                 0x30B1
+#define EGL_GL_TEXTURE_LEVEL              0x30BC
+#define EGL_GL_TEXTURE_3D                 0x30B2
+#define EGL_GL_TEXTURE_ZOFFSET            0x30BD
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_X 0x30B3
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_X 0x30B4
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Y 0x30B5
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Y 0x30B6
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Z 0x30B7
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Z 0x30B8
+EGLAPI EGLSync EGLAPIENTRY eglCreateSync (EGLDisplay dpy, EGLenum type, const EGLAttrib *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroySync (EGLDisplay dpy, EGLSync sync);
+EGLAPI EGLint EGLAPIENTRY eglClientWaitSync (EGLDisplay dpy, EGLSync sync, EGLint flags, EGLTime timeout);
+EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttrib (EGLDisplay dpy, EGLSync sync, EGLint attribute, EGLAttrib *value);
+EGLAPI EGLDisplay EGLAPIENTRY eglGetPlatformDisplay (EGLenum platform, void *native_display, const EGLAttrib *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformWindowSurface (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLAttrib *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformPixmapSurface (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLAttrib *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglWaitSync (EGLDisplay dpy, EGLSync sync, EGLint flags);
+#endif /* EGL_VERSION_1_5 */
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/winrt/include/EGL/eglext.h b/platform/winrt/include/EGL/eglext.h
index 459ecf4a3d..05b2555f31 100644
--- a/platform/winrt/include/EGL/eglext.h
+++ b/platform/winrt/include/EGL/eglext.h
@@ -1,766 +1,766 @@
-#ifndef __eglext_h_

-#define __eglext_h_ 1

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

-** Copyright (c) 2013-2014 The Khronos Group Inc.

-**

-** Permission is hereby granted, free of charge, to any person obtaining a

-** copy of this software and/or associated documentation files (the

-** "Materials"), to deal in the Materials without restriction, including

-** without limitation the rights to use, copy, modify, merge, publish,

-** distribute, sublicense, and/or sell copies of the Materials, and to

-** permit persons to whom the Materials are furnished to do so, subject to

-** the following conditions:

-**

-** The above copyright notice and this permission notice shall be included

-** in all copies or substantial portions of the Materials.

-**

-** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY

-** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

-** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

-** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

-*/

-/*

-** This header is generated from the Khronos OpenGL / OpenGL ES XML

-** API Registry. The current version of the Registry, generator scripts

-** used to make the header, and the header can be found at

-**   http://www.opengl.org/registry/

-**

-** Khronos $Revision: 27018 $ on $Date: 2014-06-10 08:06:12 -0700 (Tue, 10 Jun 2014) $

-*/

-

-#include <EGL/eglplatform.h>

-

-#define EGL_EGLEXT_VERSION 20140610

-

-/* Generated C header for:

- * API: egl

- * Versions considered: .*

- * Versions emitted: _nomatch_^

- * Default extensions included: egl

- * Additional extensions included: _nomatch_^

- * Extensions removed: _nomatch_^

- */

-

-#ifndef EGL_KHR_cl_event

-#define EGL_KHR_cl_event 1

-#define EGL_CL_EVENT_HANDLE_KHR           0x309C

-#define EGL_SYNC_CL_EVENT_KHR             0x30FE

-#define EGL_SYNC_CL_EVENT_COMPLETE_KHR    0x30FF

-#endif /* EGL_KHR_cl_event */

-

-#ifndef EGL_KHR_cl_event2

-#define EGL_KHR_cl_event2 1

-typedef void *EGLSyncKHR;

-typedef intptr_t EGLAttribKHR;

-typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNC64KHRPROC) (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSync64KHR (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list);

-#endif

-#endif /* EGL_KHR_cl_event2 */

-

-#ifndef EGL_KHR_client_get_all_proc_addresses

-#define EGL_KHR_client_get_all_proc_addresses 1

-#endif /* EGL_KHR_client_get_all_proc_addresses */

-

-#ifndef EGL_KHR_config_attribs

-#define EGL_KHR_config_attribs 1

-#define EGL_CONFORMANT_KHR                0x3042

-#define EGL_VG_COLORSPACE_LINEAR_BIT_KHR  0x0020

-#define EGL_VG_ALPHA_FORMAT_PRE_BIT_KHR   0x0040

-#endif /* EGL_KHR_config_attribs */

-

-#ifndef EGL_KHR_create_context

-#define EGL_KHR_create_context 1

-#define EGL_CONTEXT_MAJOR_VERSION_KHR     0x3098

-#define EGL_CONTEXT_MINOR_VERSION_KHR     0x30FB

-#define EGL_CONTEXT_FLAGS_KHR             0x30FC

-#define EGL_CONTEXT_OPENGL_PROFILE_MASK_KHR 0x30FD

-#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_KHR 0x31BD

-#define EGL_NO_RESET_NOTIFICATION_KHR     0x31BE

-#define EGL_LOSE_CONTEXT_ON_RESET_KHR     0x31BF

-#define EGL_CONTEXT_OPENGL_DEBUG_BIT_KHR  0x00000001

-#define EGL_CONTEXT_OPENGL_FORWARD_COMPATIBLE_BIT_KHR 0x00000002

-#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_BIT_KHR 0x00000004

-#define EGL_CONTEXT_OPENGL_CORE_PROFILE_BIT_KHR 0x00000001

-#define EGL_CONTEXT_OPENGL_COMPATIBILITY_PROFILE_BIT_KHR 0x00000002

-#define EGL_OPENGL_ES3_BIT_KHR            0x00000040

-#endif /* EGL_KHR_create_context */

-

-#ifndef EGL_KHR_fence_sync

-#define EGL_KHR_fence_sync 1

-#ifdef KHRONOS_SUPPORT_INT64

-#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_KHR 0x30F0

-#define EGL_SYNC_CONDITION_KHR            0x30F8

-#define EGL_SYNC_FENCE_KHR                0x30F9

-#endif /* KHRONOS_SUPPORT_INT64 */

-#endif /* EGL_KHR_fence_sync */

-

-#ifndef EGL_KHR_get_all_proc_addresses

-#define EGL_KHR_get_all_proc_addresses 1

-#endif /* EGL_KHR_get_all_proc_addresses */

-

-#ifndef EGL_KHR_gl_colorspace

-#define EGL_KHR_gl_colorspace 1

-#define EGL_GL_COLORSPACE_KHR             0x309D

-#define EGL_GL_COLORSPACE_SRGB_KHR        0x3089

-#define EGL_GL_COLORSPACE_LINEAR_KHR      0x308A

-#endif /* EGL_KHR_gl_colorspace */

-

-#ifndef EGL_KHR_gl_renderbuffer_image

-#define EGL_KHR_gl_renderbuffer_image 1

-#define EGL_GL_RENDERBUFFER_KHR           0x30B9

-#endif /* EGL_KHR_gl_renderbuffer_image */

-

-#ifndef EGL_KHR_gl_texture_2D_image

-#define EGL_KHR_gl_texture_2D_image 1

-#define EGL_GL_TEXTURE_2D_KHR             0x30B1

-#define EGL_GL_TEXTURE_LEVEL_KHR          0x30BC

-#endif /* EGL_KHR_gl_texture_2D_image */

-

-#ifndef EGL_KHR_gl_texture_3D_image

-#define EGL_KHR_gl_texture_3D_image 1

-#define EGL_GL_TEXTURE_3D_KHR             0x30B2

-#define EGL_GL_TEXTURE_ZOFFSET_KHR        0x30BD

-#endif /* EGL_KHR_gl_texture_3D_image */

-

-#ifndef EGL_KHR_gl_texture_cubemap_image

-#define EGL_KHR_gl_texture_cubemap_image 1

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_X_KHR 0x30B3

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_X_KHR 0x30B4

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Y_KHR 0x30B5

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_KHR 0x30B6

-#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Z_KHR 0x30B7

-#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_KHR 0x30B8

-#endif /* EGL_KHR_gl_texture_cubemap_image */

-

-#ifndef EGL_KHR_image

-#define EGL_KHR_image 1

-typedef void *EGLImageKHR;

-#define EGL_NATIVE_PIXMAP_KHR             0x30B0

-#define EGL_NO_IMAGE_KHR                  ((EGLImageKHR)0)

-typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEIMAGEKHRPROC) (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYIMAGEKHRPROC) (EGLDisplay dpy, EGLImageKHR image);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLImageKHR EGLAPIENTRY eglCreateImageKHR (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroyImageKHR (EGLDisplay dpy, EGLImageKHR image);

-#endif

-#endif /* EGL_KHR_image */

-

-#ifndef EGL_KHR_image_base

-#define EGL_KHR_image_base 1

-#define EGL_IMAGE_PRESERVED_KHR           0x30D2

-#endif /* EGL_KHR_image_base */

-

-#ifndef EGL_KHR_image_pixmap

-#define EGL_KHR_image_pixmap 1

-#endif /* EGL_KHR_image_pixmap */

-

-#ifndef EGL_KHR_lock_surface

-#define EGL_KHR_lock_surface 1

-#define EGL_READ_SURFACE_BIT_KHR          0x0001

-#define EGL_WRITE_SURFACE_BIT_KHR         0x0002

-#define EGL_LOCK_SURFACE_BIT_KHR          0x0080

-#define EGL_OPTIMAL_FORMAT_BIT_KHR        0x0100

-#define EGL_MATCH_FORMAT_KHR              0x3043

-#define EGL_FORMAT_RGB_565_EXACT_KHR      0x30C0

-#define EGL_FORMAT_RGB_565_KHR            0x30C1

-#define EGL_FORMAT_RGBA_8888_EXACT_KHR    0x30C2

-#define EGL_FORMAT_RGBA_8888_KHR          0x30C3

-#define EGL_MAP_PRESERVE_PIXELS_KHR       0x30C4

-#define EGL_LOCK_USAGE_HINT_KHR           0x30C5

-#define EGL_BITMAP_POINTER_KHR            0x30C6

-#define EGL_BITMAP_PITCH_KHR              0x30C7

-#define EGL_BITMAP_ORIGIN_KHR             0x30C8

-#define EGL_BITMAP_PIXEL_RED_OFFSET_KHR   0x30C9

-#define EGL_BITMAP_PIXEL_GREEN_OFFSET_KHR 0x30CA

-#define EGL_BITMAP_PIXEL_BLUE_OFFSET_KHR  0x30CB

-#define EGL_BITMAP_PIXEL_ALPHA_OFFSET_KHR 0x30CC

-#define EGL_BITMAP_PIXEL_LUMINANCE_OFFSET_KHR 0x30CD

-#define EGL_LOWER_LEFT_KHR                0x30CE

-#define EGL_UPPER_LEFT_KHR                0x30CF

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLUNLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglLockSurfaceKHR (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglUnlockSurfaceKHR (EGLDisplay dpy, EGLSurface surface);

-#endif

-#endif /* EGL_KHR_lock_surface */

-

-#ifndef EGL_KHR_lock_surface2

-#define EGL_KHR_lock_surface2 1

-#define EGL_BITMAP_PIXEL_SIZE_KHR         0x3110

-#endif /* EGL_KHR_lock_surface2 */

-

-#ifndef EGL_KHR_lock_surface3

-#define EGL_KHR_lock_surface3 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACE64KHRPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface64KHR (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value);

-#endif

-#endif /* EGL_KHR_lock_surface3 */

-

-#ifndef EGL_KHR_platform_android

-#define EGL_KHR_platform_android 1

-#define EGL_PLATFORM_ANDROID_KHR          0x3141

-#endif /* EGL_KHR_platform_android */

-

-#ifndef EGL_KHR_platform_gbm

-#define EGL_KHR_platform_gbm 1

-#define EGL_PLATFORM_GBM_KHR              0x31D7

-#endif /* EGL_KHR_platform_gbm */

-

-#ifndef EGL_KHR_platform_wayland

-#define EGL_KHR_platform_wayland 1

-#define EGL_PLATFORM_WAYLAND_KHR          0x31D8

-#endif /* EGL_KHR_platform_wayland */

-

-#ifndef EGL_KHR_platform_x11

-#define EGL_KHR_platform_x11 1

-#define EGL_PLATFORM_X11_KHR              0x31D5

-#define EGL_PLATFORM_X11_SCREEN_KHR       0x31D6

-#endif /* EGL_KHR_platform_x11 */

-

-#ifndef EGL_KHR_reusable_sync

-#define EGL_KHR_reusable_sync 1

-typedef khronos_utime_nanoseconds_t EGLTimeKHR;

-#ifdef KHRONOS_SUPPORT_INT64

-#define EGL_SYNC_STATUS_KHR               0x30F1

-#define EGL_SIGNALED_KHR                  0x30F2

-#define EGL_UNSIGNALED_KHR                0x30F3

-#define EGL_TIMEOUT_EXPIRED_KHR           0x30F5

-#define EGL_CONDITION_SATISFIED_KHR       0x30F6

-#define EGL_SYNC_TYPE_KHR                 0x30F7

-#define EGL_SYNC_REUSABLE_KHR             0x30FA

-#define EGL_SYNC_FLUSH_COMMANDS_BIT_KHR   0x0001

-#define EGL_FOREVER_KHR                   0xFFFFFFFFFFFFFFFFull

-#define EGL_NO_SYNC_KHR                   ((EGLSyncKHR)0)

-typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNCKHRPROC) (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync);

-typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSyncKHR (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncKHR (EGLDisplay dpy, EGLSyncKHR sync);

-EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout);

-EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode);

-EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value);

-#endif

-#endif /* KHRONOS_SUPPORT_INT64 */

-#endif /* EGL_KHR_reusable_sync */

-

-#ifndef EGL_KHR_stream

-#define EGL_KHR_stream 1

-typedef void *EGLStreamKHR;

-typedef khronos_uint64_t EGLuint64KHR;

-#ifdef KHRONOS_SUPPORT_INT64

-#define EGL_NO_STREAM_KHR                 ((EGLStreamKHR)0)

-#define EGL_CONSUMER_LATENCY_USEC_KHR     0x3210

-#define EGL_PRODUCER_FRAME_KHR            0x3212

-#define EGL_CONSUMER_FRAME_KHR            0x3213

-#define EGL_STREAM_STATE_KHR              0x3214

-#define EGL_STREAM_STATE_CREATED_KHR      0x3215

-#define EGL_STREAM_STATE_CONNECTING_KHR   0x3216

-#define EGL_STREAM_STATE_EMPTY_KHR        0x3217

-#define EGL_STREAM_STATE_NEW_FRAME_AVAILABLE_KHR 0x3218

-#define EGL_STREAM_STATE_OLD_FRAME_AVAILABLE_KHR 0x3219

-#define EGL_STREAM_STATE_DISCONNECTED_KHR 0x321A

-#define EGL_BAD_STREAM_KHR                0x321B

-#define EGL_BAD_STATE_KHR                 0x321C

-typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMKHRPROC) (EGLDisplay dpy, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMATTRIBKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMU64KHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamKHR (EGLDisplay dpy, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroyStreamKHR (EGLDisplay dpy, EGLStreamKHR stream);

-EGLAPI EGLBoolean EGLAPIENTRY eglStreamAttribKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamu64KHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value);

-#endif

-#endif /* KHRONOS_SUPPORT_INT64 */

-#endif /* EGL_KHR_stream */

-

-#ifndef EGL_KHR_stream_consumer_gltexture

-#define EGL_KHR_stream_consumer_gltexture 1

-#ifdef EGL_KHR_stream

-#define EGL_CONSUMER_ACQUIRE_TIMEOUT_USEC_KHR 0x321E

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERGLTEXTUREEXTERNALKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERACQUIREKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERRELEASEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerGLTextureExternalKHR (EGLDisplay dpy, EGLStreamKHR stream);

-EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerAcquireKHR (EGLDisplay dpy, EGLStreamKHR stream);

-EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerReleaseKHR (EGLDisplay dpy, EGLStreamKHR stream);

-#endif

-#endif /* EGL_KHR_stream */

-#endif /* EGL_KHR_stream_consumer_gltexture */

-

-#ifndef EGL_KHR_stream_cross_process_fd

-#define EGL_KHR_stream_cross_process_fd 1

-typedef int EGLNativeFileDescriptorKHR;

-#ifdef EGL_KHR_stream

-#define EGL_NO_FILE_DESCRIPTOR_KHR        ((EGLNativeFileDescriptorKHR)(-1))

-typedef EGLNativeFileDescriptorKHR (EGLAPIENTRYP PFNEGLGETSTREAMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);

-typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMFROMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLNativeFileDescriptorKHR EGLAPIENTRY eglGetStreamFileDescriptorKHR (EGLDisplay dpy, EGLStreamKHR stream);

-EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamFromFileDescriptorKHR (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor);

-#endif

-#endif /* EGL_KHR_stream */

-#endif /* EGL_KHR_stream_cross_process_fd */

-

-#ifndef EGL_KHR_stream_fifo

-#define EGL_KHR_stream_fifo 1

-#ifdef EGL_KHR_stream

-#define EGL_STREAM_FIFO_LENGTH_KHR        0x31FC

-#define EGL_STREAM_TIME_NOW_KHR           0x31FD

-#define EGL_STREAM_TIME_CONSUMER_KHR      0x31FE

-#define EGL_STREAM_TIME_PRODUCER_KHR      0x31FF

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMTIMEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamTimeKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value);

-#endif

-#endif /* EGL_KHR_stream */

-#endif /* EGL_KHR_stream_fifo */

-

-#ifndef EGL_KHR_stream_producer_aldatalocator

-#define EGL_KHR_stream_producer_aldatalocator 1

-#ifdef EGL_KHR_stream

-#endif /* EGL_KHR_stream */

-#endif /* EGL_KHR_stream_producer_aldatalocator */

-

-#ifndef EGL_KHR_stream_producer_eglsurface

-#define EGL_KHR_stream_producer_eglsurface 1

-#ifdef EGL_KHR_stream

-#define EGL_STREAM_BIT_KHR                0x0800

-typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATESTREAMPRODUCERSURFACEKHRPROC) (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSurface EGLAPIENTRY eglCreateStreamProducerSurfaceKHR (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list);

-#endif

-#endif /* EGL_KHR_stream */

-#endif /* EGL_KHR_stream_producer_eglsurface */

-

-#ifndef EGL_KHR_surfaceless_context

-#define EGL_KHR_surfaceless_context 1

-#endif /* EGL_KHR_surfaceless_context */

-

-#ifndef EGL_KHR_vg_parent_image

-#define EGL_KHR_vg_parent_image 1

-#define EGL_VG_PARENT_IMAGE_KHR           0x30BA

-#endif /* EGL_KHR_vg_parent_image */

-

-#ifndef EGL_KHR_wait_sync

-#define EGL_KHR_wait_sync 1

-typedef EGLint (EGLAPIENTRYP PFNEGLWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLint EGLAPIENTRY eglWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags);

-#endif

-#endif /* EGL_KHR_wait_sync */

-

-#ifndef EGL_ANDROID_blob_cache

-#define EGL_ANDROID_blob_cache 1

-typedef khronos_ssize_t EGLsizeiANDROID;

-typedef void (*EGLSetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, const void *value, EGLsizeiANDROID valueSize);

-typedef EGLsizeiANDROID (*EGLGetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, void *value, EGLsizeiANDROID valueSize);

-typedef void (EGLAPIENTRYP PFNEGLSETBLOBCACHEFUNCSANDROIDPROC) (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI void EGLAPIENTRY eglSetBlobCacheFuncsANDROID (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get);

-#endif

-#endif /* EGL_ANDROID_blob_cache */

-

-#ifndef EGL_ANDROID_framebuffer_target

-#define EGL_ANDROID_framebuffer_target 1

-#define EGL_FRAMEBUFFER_TARGET_ANDROID    0x3147

-#endif /* EGL_ANDROID_framebuffer_target */

-

-#ifndef EGL_ANDROID_image_native_buffer

-#define EGL_ANDROID_image_native_buffer 1

-#define EGL_NATIVE_BUFFER_ANDROID         0x3140

-#endif /* EGL_ANDROID_image_native_buffer */

-

-#ifndef EGL_ANDROID_native_fence_sync

-#define EGL_ANDROID_native_fence_sync 1

-#define EGL_SYNC_NATIVE_FENCE_ANDROID     0x3144

-#define EGL_SYNC_NATIVE_FENCE_FD_ANDROID  0x3145

-#define EGL_SYNC_NATIVE_FENCE_SIGNALED_ANDROID 0x3146

-#define EGL_NO_NATIVE_FENCE_FD_ANDROID    -1

-typedef EGLint (EGLAPIENTRYP PFNEGLDUPNATIVEFENCEFDANDROIDPROC) (EGLDisplay dpy, EGLSyncKHR sync);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLint EGLAPIENTRY eglDupNativeFenceFDANDROID (EGLDisplay dpy, EGLSyncKHR sync);

-#endif

-#endif /* EGL_ANDROID_native_fence_sync */

-

-#ifndef EGL_ANDROID_recordable

-#define EGL_ANDROID_recordable 1

-#define EGL_RECORDABLE_ANDROID            0x3142

-#endif /* EGL_ANDROID_recordable */

-

-#ifndef EGL_ANGLE_d3d_share_handle_client_buffer

-#define EGL_ANGLE_d3d_share_handle_client_buffer 1

-#define EGL_D3D_TEXTURE_2D_SHARE_HANDLE_ANGLE 0x3200

-#endif /* EGL_ANGLE_d3d_share_handle_client_buffer */

-

-#ifndef EGL_ANGLE_window_fixed_size

-#define EGL_ANGLE_window_fixed_size 1

-#define EGL_FIXED_SIZE_ANGLE        0x3201

-#endif /* EGL_ANGLE_window_fixed_size */

-

-#ifndef EGL_ANGLE_query_surface_pointer

-#define EGL_ANGLE_query_surface_pointer 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACEPOINTERANGLEPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurfacePointerANGLE (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value);

-#endif

-#endif /* EGL_ANGLE_query_surface_pointer */

-

-#ifndef EGL_ANGLE_software_display

-#define EGL_ANGLE_software_display 1

-#define EGL_SOFTWARE_DISPLAY_ANGLE ((EGLNativeDisplayType)-1)

-#endif /* EGL_ANGLE_software_display */

-

-#ifndef EGL_ANGLE_direct3d_display

-#define EGL_ANGLE_direct3d_display 1

-#define EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE ((EGLNativeDisplayType)-2)

-#define EGL_D3D11_ONLY_DISPLAY_ANGLE ((EGLNativeDisplayType)-3)

-#endif /* EGL_ANGLE_direct3d_display */

-

-#ifndef EGL_ANGLE_surface_d3d_texture_2d_share_handle

-#define EGL_ANGLE_surface_d3d_texture_2d_share_handle 1

-#endif /* EGL_ANGLE_surface_d3d_texture_2d_share_handle */

-

-#ifndef EGL_ANGLE_surface_d3d_render_to_back_buffer

-#define EGL_ANGLE_surface_d3d_render_to_back_buffer 1

-#define EGL_ANGLE_DISPLAY_ALLOW_RENDER_TO_BACK_BUFFER 0x320B

-#define EGL_ANGLE_SURFACE_RENDER_TO_BACK_BUFFER 0x320C

-#endif /* EGL_ANGLE_surface_d3d_render_to_back_buffer */

-

-#ifndef EGL_ANGLE_platform_angle

-#define EGL_ANGLE_platform_angle 1

-#define EGL_PLATFORM_ANGLE_ANGLE          0x3201

-#define EGL_PLATFORM_ANGLE_TYPE_ANGLE     0x3202

-#define EGL_PLATFORM_ANGLE_MAX_VERSION_MAJOR_ANGLE 0x3203

-#define EGL_PLATFORM_ANGLE_MAX_VERSION_MINOR_ANGLE 0x3204

-#define EGL_PLATFORM_ANGLE_TYPE_DEFAULT_ANGLE 0x3205

-#endif /* EGL_ANGLE_platform_angle */

-

-#ifndef EGL_ANGLE_platform_angle_d3d

-#define EGL_ANGLE_platform_angle_d3d 1

-#define EGL_PLATFORM_ANGLE_TYPE_D3D9_ANGLE 0x3206

-#define EGL_PLATFORM_ANGLE_TYPE_D3D11_ANGLE 0x3207

-#define EGL_PLATFORM_ANGLE_USE_WARP_ANGLE 0x3208

-#endif /* EGL_ANGLE_platform_angle_d3d */

-

-#ifndef EGL_ANGLE_platform_angle_opengl

-#define EGL_ANGLE_platform_angle_opengl 1

-#define EGL_PLATFORM_ANGLE_TYPE_OPENGL_ANGLE 0x3209

-#define EGL_PLATFORM_ANGLE_TYPE_OPENGLES_ANGLE 0x320A

-#endif /* EGL_ANGLE_platform_angle_opengl */

-

-#ifndef EGL_ARM_pixmap_multisample_discard

-#define EGL_ARM_pixmap_multisample_discard 1

-#define EGL_DISCARD_SAMPLES_ARM           0x3286

-#endif /* EGL_ARM_pixmap_multisample_discard */

-

-#ifndef EGL_EXT_buffer_age

-#define EGL_EXT_buffer_age 1

-#define EGL_BUFFER_AGE_EXT                0x313D

-#endif /* EGL_EXT_buffer_age */

-

-#ifndef EGL_EXT_client_extensions

-#define EGL_EXT_client_extensions 1

-#endif /* EGL_EXT_client_extensions */

-

-#ifndef EGL_EXT_create_context_robustness

-#define EGL_EXT_create_context_robustness 1

-#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_EXT 0x30BF

-#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_EXT 0x3138

-#define EGL_NO_RESET_NOTIFICATION_EXT     0x31BE

-#define EGL_LOSE_CONTEXT_ON_RESET_EXT     0x31BF

-#endif /* EGL_EXT_create_context_robustness */

-

-#ifndef EGL_EXT_device_base

-#define EGL_EXT_device_base 1

-typedef void *EGLDeviceEXT;

-#define EGL_NO_DEVICE_EXT                 ((EGLDeviceEXT)(0))

-#define EGL_BAD_DEVICE_EXT                0x322B

-#define EGL_DEVICE_EXT                    0x322C

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDEVICEATTRIBEXTPROC) (EGLDeviceEXT device, EGLint attribute, EGLAttrib *value);

-typedef const char *(EGLAPIENTRYP PFNEGLQUERYDEVICESTRINGEXTPROC) (EGLDeviceEXT device, EGLint name);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDEVICESEXTPROC) (EGLint max_devices, EGLDeviceEXT *devices, EGLint *num_devices);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDISPLAYATTRIBEXTPROC) (EGLDisplay dpy, EGLint attribute, EGLAttrib *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryDeviceAttribEXT (EGLDeviceEXT device, EGLint attribute, EGLAttrib *value);

-EGLAPI const char *EGLAPIENTRY eglQueryDeviceStringEXT (EGLDeviceEXT device, EGLint name);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryDevicesEXT (EGLint max_devices, EGLDeviceEXT *devices, EGLint *num_devices);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryDisplayAttribEXT (EGLDisplay dpy, EGLint attribute, EGLAttrib *value);

-#endif

-#endif /* EGL_EXT_device_base */

-

-#ifndef EGL_EXT_image_dma_buf_import

-#define EGL_EXT_image_dma_buf_import 1

-#define EGL_LINUX_DMA_BUF_EXT             0x3270

-#define EGL_LINUX_DRM_FOURCC_EXT          0x3271

-#define EGL_DMA_BUF_PLANE0_FD_EXT         0x3272

-#define EGL_DMA_BUF_PLANE0_OFFSET_EXT     0x3273

-#define EGL_DMA_BUF_PLANE0_PITCH_EXT      0x3274

-#define EGL_DMA_BUF_PLANE1_FD_EXT         0x3275

-#define EGL_DMA_BUF_PLANE1_OFFSET_EXT     0x3276

-#define EGL_DMA_BUF_PLANE1_PITCH_EXT      0x3277

-#define EGL_DMA_BUF_PLANE2_FD_EXT         0x3278

-#define EGL_DMA_BUF_PLANE2_OFFSET_EXT     0x3279

-#define EGL_DMA_BUF_PLANE2_PITCH_EXT      0x327A

-#define EGL_YUV_COLOR_SPACE_HINT_EXT      0x327B

-#define EGL_SAMPLE_RANGE_HINT_EXT         0x327C

-#define EGL_YUV_CHROMA_HORIZONTAL_SITING_HINT_EXT 0x327D

-#define EGL_YUV_CHROMA_VERTICAL_SITING_HINT_EXT 0x327E

-#define EGL_ITU_REC601_EXT                0x327F

-#define EGL_ITU_REC709_EXT                0x3280

-#define EGL_ITU_REC2020_EXT               0x3281

-#define EGL_YUV_FULL_RANGE_EXT            0x3282

-#define EGL_YUV_NARROW_RANGE_EXT          0x3283

-#define EGL_YUV_CHROMA_SITING_0_EXT       0x3284

-#define EGL_YUV_CHROMA_SITING_0_5_EXT     0x3285

-#endif /* EGL_EXT_image_dma_buf_import */

-

-#ifndef EGL_EXT_multiview_window

-#define EGL_EXT_multiview_window 1

-#define EGL_MULTIVIEW_VIEW_COUNT_EXT      0x3134

-#endif /* EGL_EXT_multiview_window */

-

-#ifndef EGL_EXT_platform_base

-#define EGL_EXT_platform_base 1

-typedef EGLDisplay (EGLAPIENTRYP PFNEGLGETPLATFORMDISPLAYEXTPROC) (EGLenum platform, void *native_display, const EGLint *attrib_list);

-typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMWINDOWSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list);

-typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMPIXMAPSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLDisplay EGLAPIENTRY eglGetPlatformDisplayEXT (EGLenum platform, void *native_display, const EGLint *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformWindowSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list);

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformPixmapSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list);

-#endif

-#endif /* EGL_EXT_platform_base */

-

-#ifndef EGL_EXT_platform_device

-#define EGL_EXT_platform_device 1

-#define EGL_PLATFORM_DEVICE_EXT           0x313F

-#endif /* EGL_EXT_platform_device */

-

-#ifndef EGL_EXT_platform_wayland

-#define EGL_EXT_platform_wayland 1

-#define EGL_PLATFORM_WAYLAND_EXT          0x31D8

-#endif /* EGL_EXT_platform_wayland */

-

-#ifndef EGL_EXT_platform_x11

-#define EGL_EXT_platform_x11 1

-#define EGL_PLATFORM_X11_EXT              0x31D5

-#define EGL_PLATFORM_X11_SCREEN_EXT       0x31D6

-#endif /* EGL_EXT_platform_x11 */

-

-#ifndef EGL_EXT_protected_surface

-#define EGL_EXT_protected_surface 1

-#define EGL_PROTECTED_CONTENT_EXT         0x32C0

-#endif /* EGL_EXT_protected_surface */

-

-#ifndef EGL_EXT_swap_buffers_with_damage

-#define EGL_EXT_swap_buffers_with_damage 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSWITHDAMAGEEXTPROC) (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersWithDamageEXT (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects);

-#endif

-#endif /* EGL_EXT_swap_buffers_with_damage */

-

-#ifndef EGL_HI_clientpixmap

-#define EGL_HI_clientpixmap 1

-struct EGLClientPixmapHI {

-    void  *pData;

-    EGLint iWidth;

-    EGLint iHeight;

-    EGLint iStride;

-};

-#define EGL_CLIENT_PIXMAP_POINTER_HI      0x8F74

-typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPIXMAPSURFACEHIPROC) (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurfaceHI (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap);

-#endif

-#endif /* EGL_HI_clientpixmap */

-

-#ifndef EGL_HI_colorformats

-#define EGL_HI_colorformats 1

-#define EGL_COLOR_FORMAT_HI               0x8F70

-#define EGL_COLOR_RGB_HI                  0x8F71

-#define EGL_COLOR_RGBA_HI                 0x8F72

-#define EGL_COLOR_ARGB_HI                 0x8F73

-#endif /* EGL_HI_colorformats */

-

-#ifndef EGL_IMG_context_priority

-#define EGL_IMG_context_priority 1

-#define EGL_CONTEXT_PRIORITY_LEVEL_IMG    0x3100

-#define EGL_CONTEXT_PRIORITY_HIGH_IMG     0x3101

-#define EGL_CONTEXT_PRIORITY_MEDIUM_IMG   0x3102

-#define EGL_CONTEXT_PRIORITY_LOW_IMG      0x3103

-#endif /* EGL_IMG_context_priority */

-

-#ifndef EGL_MESA_drm_image

-#define EGL_MESA_drm_image 1

-#define EGL_DRM_BUFFER_FORMAT_MESA        0x31D0

-#define EGL_DRM_BUFFER_USE_MESA           0x31D1

-#define EGL_DRM_BUFFER_FORMAT_ARGB32_MESA 0x31D2

-#define EGL_DRM_BUFFER_MESA               0x31D3

-#define EGL_DRM_BUFFER_STRIDE_MESA        0x31D4

-#define EGL_DRM_BUFFER_USE_SCANOUT_MESA   0x00000001

-#define EGL_DRM_BUFFER_USE_SHARE_MESA     0x00000002

-typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEDRMIMAGEMESAPROC) (EGLDisplay dpy, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLEXPORTDRMIMAGEMESAPROC) (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLImageKHR EGLAPIENTRY eglCreateDRMImageMESA (EGLDisplay dpy, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglExportDRMImageMESA (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride);

-#endif

-#endif /* EGL_MESA_drm_image */

-

-#ifndef EGL_MESA_platform_gbm

-#define EGL_MESA_platform_gbm 1

-#define EGL_PLATFORM_GBM_MESA             0x31D7

-#endif /* EGL_MESA_platform_gbm */

-

-#ifndef EGL_NOK_swap_region

-#define EGL_NOK_swap_region 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSREGIONNOKPROC) (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersRegionNOK (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);

-#endif

-#endif /* EGL_NOK_swap_region */

-

-#ifndef EGL_NOK_swap_region2

-#define EGL_NOK_swap_region2 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSREGION2NOKPROC) (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersRegion2NOK (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);

-#endif

-#endif /* EGL_NOK_swap_region2 */

-

-#ifndef EGL_NOK_texture_from_pixmap

-#define EGL_NOK_texture_from_pixmap 1

-#define EGL_Y_INVERTED_NOK                0x307F

-#endif /* EGL_NOK_texture_from_pixmap */

-

-#ifndef EGL_NV_3dvision_surface

-#define EGL_NV_3dvision_surface 1

-#define EGL_AUTO_STEREO_NV                0x3136

-#endif /* EGL_NV_3dvision_surface */

-

-#ifndef EGL_NV_coverage_sample

-#define EGL_NV_coverage_sample 1

-#define EGL_COVERAGE_BUFFERS_NV           0x30E0

-#define EGL_COVERAGE_SAMPLES_NV           0x30E1

-#endif /* EGL_NV_coverage_sample */

-

-#ifndef EGL_NV_coverage_sample_resolve

-#define EGL_NV_coverage_sample_resolve 1

-#define EGL_COVERAGE_SAMPLE_RESOLVE_NV    0x3131

-#define EGL_COVERAGE_SAMPLE_RESOLVE_DEFAULT_NV 0x3132

-#define EGL_COVERAGE_SAMPLE_RESOLVE_NONE_NV 0x3133

-#endif /* EGL_NV_coverage_sample_resolve */

-

-#ifndef EGL_NV_depth_nonlinear

-#define EGL_NV_depth_nonlinear 1

-#define EGL_DEPTH_ENCODING_NV             0x30E2

-#define EGL_DEPTH_ENCODING_NONE_NV        0

-#define EGL_DEPTH_ENCODING_NONLINEAR_NV   0x30E3

-#endif /* EGL_NV_depth_nonlinear */

-

-#ifndef EGL_NV_native_query

-#define EGL_NV_native_query 1

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEDISPLAYNVPROC) (EGLDisplay dpy, EGLNativeDisplayType *display_id);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEWINDOWNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEPIXMAPNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeDisplayNV (EGLDisplay dpy, EGLNativeDisplayType *display_id);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeWindowNV (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window);

-EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativePixmapNV (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap);

-#endif

-#endif /* EGL_NV_native_query */

-

-#ifndef EGL_NV_post_convert_rounding

-#define EGL_NV_post_convert_rounding 1

-#endif /* EGL_NV_post_convert_rounding */

-

-#ifndef EGL_NV_post_sub_buffer

-#define EGL_NV_post_sub_buffer 1

-#define EGL_POST_SUB_BUFFER_SUPPORTED_NV  0x30BE

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLPOSTSUBBUFFERNVPROC) (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLBoolean EGLAPIENTRY eglPostSubBufferNV (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height);

-#endif

-#endif /* EGL_NV_post_sub_buffer */

-

-#ifndef EGL_NV_stream_sync

-#define EGL_NV_stream_sync 1

-#define EGL_SYNC_NEW_FRAME_NV             0x321F

-typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESTREAMSYNCNVPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateStreamSyncNV (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list);

-#endif

-#endif /* EGL_NV_stream_sync */

-

-#ifndef EGL_NV_sync

-#define EGL_NV_sync 1

-typedef void *EGLSyncNV;

-typedef khronos_utime_nanoseconds_t EGLTimeNV;

-#ifdef KHRONOS_SUPPORT_INT64

-#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_NV 0x30E6

-#define EGL_SYNC_STATUS_NV                0x30E7

-#define EGL_SIGNALED_NV                   0x30E8

-#define EGL_UNSIGNALED_NV                 0x30E9

-#define EGL_SYNC_FLUSH_COMMANDS_BIT_NV    0x0001

-#define EGL_FOREVER_NV                    0xFFFFFFFFFFFFFFFFull

-#define EGL_ALREADY_SIGNALED_NV           0x30EA

-#define EGL_TIMEOUT_EXPIRED_NV            0x30EB

-#define EGL_CONDITION_SATISFIED_NV        0x30EC

-#define EGL_SYNC_TYPE_NV                  0x30ED

-#define EGL_SYNC_CONDITION_NV             0x30EE

-#define EGL_SYNC_FENCE_NV                 0x30EF

-#define EGL_NO_SYNC_NV                    ((EGLSyncNV)0)

-typedef EGLSyncNV (EGLAPIENTRYP PFNEGLCREATEFENCESYNCNVPROC) (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCNVPROC) (EGLSyncNV sync);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLFENCENVPROC) (EGLSyncNV sync);

-typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCNVPROC) (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCNVPROC) (EGLSyncNV sync, EGLenum mode);

-typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBNVPROC) (EGLSyncNV sync, EGLint attribute, EGLint *value);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLSyncNV EGLAPIENTRY eglCreateFenceSyncNV (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list);

-EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncNV (EGLSyncNV sync);

-EGLAPI EGLBoolean EGLAPIENTRY eglFenceNV (EGLSyncNV sync);

-EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncNV (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout);

-EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncNV (EGLSyncNV sync, EGLenum mode);

-EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribNV (EGLSyncNV sync, EGLint attribute, EGLint *value);

-#endif

-#endif /* KHRONOS_SUPPORT_INT64 */

-#endif /* EGL_NV_sync */

-

-#ifndef EGL_NV_system_time

-#define EGL_NV_system_time 1

-typedef khronos_utime_nanoseconds_t EGLuint64NV;

-#ifdef KHRONOS_SUPPORT_INT64

-typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMEFREQUENCYNVPROC) (void);

-typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMENVPROC) (void);

-#ifdef EGL_EGLEXT_PROTOTYPES

-EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeFrequencyNV (void);

-EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeNV (void);

-#endif

-#endif /* KHRONOS_SUPPORT_INT64 */

-#endif /* EGL_NV_system_time */

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif

+#ifndef __eglext_h_
+#define __eglext_h_ 1
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+** Copyright (c) 2013-2014 The Khronos Group Inc.
+**
+** Permission is hereby granted, free of charge, to any person obtaining a
+** copy of this software and/or associated documentation files (the
+** "Materials"), to deal in the Materials without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Materials, and to
+** permit persons to whom the Materials are furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be included
+** in all copies or substantial portions of the Materials.
+**
+** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
+*/
+/*
+** This header is generated from the Khronos OpenGL / OpenGL ES XML
+** API Registry. The current version of the Registry, generator scripts
+** used to make the header, and the header can be found at
+**   http://www.opengl.org/registry/
+**
+** Khronos $Revision: 27018 $ on $Date: 2014-06-10 08:06:12 -0700 (Tue, 10 Jun 2014) $
+*/
+
+#include <EGL/eglplatform.h>
+
+#define EGL_EGLEXT_VERSION 20140610
+
+/* Generated C header for:
+ * API: egl
+ * Versions considered: .*
+ * Versions emitted: _nomatch_^
+ * Default extensions included: egl
+ * Additional extensions included: _nomatch_^
+ * Extensions removed: _nomatch_^
+ */
+
+#ifndef EGL_KHR_cl_event
+#define EGL_KHR_cl_event 1
+#define EGL_CL_EVENT_HANDLE_KHR           0x309C
+#define EGL_SYNC_CL_EVENT_KHR             0x30FE
+#define EGL_SYNC_CL_EVENT_COMPLETE_KHR    0x30FF
+#endif /* EGL_KHR_cl_event */
+
+#ifndef EGL_KHR_cl_event2
+#define EGL_KHR_cl_event2 1
+typedef void *EGLSyncKHR;
+typedef intptr_t EGLAttribKHR;
+typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNC64KHRPROC) (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSync64KHR (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list);
+#endif
+#endif /* EGL_KHR_cl_event2 */
+
+#ifndef EGL_KHR_client_get_all_proc_addresses
+#define EGL_KHR_client_get_all_proc_addresses 1
+#endif /* EGL_KHR_client_get_all_proc_addresses */
+
+#ifndef EGL_KHR_config_attribs
+#define EGL_KHR_config_attribs 1
+#define EGL_CONFORMANT_KHR                0x3042
+#define EGL_VG_COLORSPACE_LINEAR_BIT_KHR  0x0020
+#define EGL_VG_ALPHA_FORMAT_PRE_BIT_KHR   0x0040
+#endif /* EGL_KHR_config_attribs */
+
+#ifndef EGL_KHR_create_context
+#define EGL_KHR_create_context 1
+#define EGL_CONTEXT_MAJOR_VERSION_KHR     0x3098
+#define EGL_CONTEXT_MINOR_VERSION_KHR     0x30FB
+#define EGL_CONTEXT_FLAGS_KHR             0x30FC
+#define EGL_CONTEXT_OPENGL_PROFILE_MASK_KHR 0x30FD
+#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_KHR 0x31BD
+#define EGL_NO_RESET_NOTIFICATION_KHR     0x31BE
+#define EGL_LOSE_CONTEXT_ON_RESET_KHR     0x31BF
+#define EGL_CONTEXT_OPENGL_DEBUG_BIT_KHR  0x00000001
+#define EGL_CONTEXT_OPENGL_FORWARD_COMPATIBLE_BIT_KHR 0x00000002
+#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_BIT_KHR 0x00000004
+#define EGL_CONTEXT_OPENGL_CORE_PROFILE_BIT_KHR 0x00000001
+#define EGL_CONTEXT_OPENGL_COMPATIBILITY_PROFILE_BIT_KHR 0x00000002
+#define EGL_OPENGL_ES3_BIT_KHR            0x00000040
+#endif /* EGL_KHR_create_context */
+
+#ifndef EGL_KHR_fence_sync
+#define EGL_KHR_fence_sync 1
+#ifdef KHRONOS_SUPPORT_INT64
+#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_KHR 0x30F0
+#define EGL_SYNC_CONDITION_KHR            0x30F8
+#define EGL_SYNC_FENCE_KHR                0x30F9
+#endif /* KHRONOS_SUPPORT_INT64 */
+#endif /* EGL_KHR_fence_sync */
+
+#ifndef EGL_KHR_get_all_proc_addresses
+#define EGL_KHR_get_all_proc_addresses 1
+#endif /* EGL_KHR_get_all_proc_addresses */
+
+#ifndef EGL_KHR_gl_colorspace
+#define EGL_KHR_gl_colorspace 1
+#define EGL_GL_COLORSPACE_KHR             0x309D
+#define EGL_GL_COLORSPACE_SRGB_KHR        0x3089
+#define EGL_GL_COLORSPACE_LINEAR_KHR      0x308A
+#endif /* EGL_KHR_gl_colorspace */
+
+#ifndef EGL_KHR_gl_renderbuffer_image
+#define EGL_KHR_gl_renderbuffer_image 1
+#define EGL_GL_RENDERBUFFER_KHR           0x30B9
+#endif /* EGL_KHR_gl_renderbuffer_image */
+
+#ifndef EGL_KHR_gl_texture_2D_image
+#define EGL_KHR_gl_texture_2D_image 1
+#define EGL_GL_TEXTURE_2D_KHR             0x30B1
+#define EGL_GL_TEXTURE_LEVEL_KHR          0x30BC
+#endif /* EGL_KHR_gl_texture_2D_image */
+
+#ifndef EGL_KHR_gl_texture_3D_image
+#define EGL_KHR_gl_texture_3D_image 1
+#define EGL_GL_TEXTURE_3D_KHR             0x30B2
+#define EGL_GL_TEXTURE_ZOFFSET_KHR        0x30BD
+#endif /* EGL_KHR_gl_texture_3D_image */
+
+#ifndef EGL_KHR_gl_texture_cubemap_image
+#define EGL_KHR_gl_texture_cubemap_image 1
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_X_KHR 0x30B3
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_X_KHR 0x30B4
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Y_KHR 0x30B5
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_KHR 0x30B6
+#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Z_KHR 0x30B7
+#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_KHR 0x30B8
+#endif /* EGL_KHR_gl_texture_cubemap_image */
+
+#ifndef EGL_KHR_image
+#define EGL_KHR_image 1
+typedef void *EGLImageKHR;
+#define EGL_NATIVE_PIXMAP_KHR             0x30B0
+#define EGL_NO_IMAGE_KHR                  ((EGLImageKHR)0)
+typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEIMAGEKHRPROC) (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYIMAGEKHRPROC) (EGLDisplay dpy, EGLImageKHR image);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLImageKHR EGLAPIENTRY eglCreateImageKHR (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroyImageKHR (EGLDisplay dpy, EGLImageKHR image);
+#endif
+#endif /* EGL_KHR_image */
+
+#ifndef EGL_KHR_image_base
+#define EGL_KHR_image_base 1
+#define EGL_IMAGE_PRESERVED_KHR           0x30D2
+#endif /* EGL_KHR_image_base */
+
+#ifndef EGL_KHR_image_pixmap
+#define EGL_KHR_image_pixmap 1
+#endif /* EGL_KHR_image_pixmap */
+
+#ifndef EGL_KHR_lock_surface
+#define EGL_KHR_lock_surface 1
+#define EGL_READ_SURFACE_BIT_KHR          0x0001
+#define EGL_WRITE_SURFACE_BIT_KHR         0x0002
+#define EGL_LOCK_SURFACE_BIT_KHR          0x0080
+#define EGL_OPTIMAL_FORMAT_BIT_KHR        0x0100
+#define EGL_MATCH_FORMAT_KHR              0x3043
+#define EGL_FORMAT_RGB_565_EXACT_KHR      0x30C0
+#define EGL_FORMAT_RGB_565_KHR            0x30C1
+#define EGL_FORMAT_RGBA_8888_EXACT_KHR    0x30C2
+#define EGL_FORMAT_RGBA_8888_KHR          0x30C3
+#define EGL_MAP_PRESERVE_PIXELS_KHR       0x30C4
+#define EGL_LOCK_USAGE_HINT_KHR           0x30C5
+#define EGL_BITMAP_POINTER_KHR            0x30C6
+#define EGL_BITMAP_PITCH_KHR              0x30C7
+#define EGL_BITMAP_ORIGIN_KHR             0x30C8
+#define EGL_BITMAP_PIXEL_RED_OFFSET_KHR   0x30C9
+#define EGL_BITMAP_PIXEL_GREEN_OFFSET_KHR 0x30CA
+#define EGL_BITMAP_PIXEL_BLUE_OFFSET_KHR  0x30CB
+#define EGL_BITMAP_PIXEL_ALPHA_OFFSET_KHR 0x30CC
+#define EGL_BITMAP_PIXEL_LUMINANCE_OFFSET_KHR 0x30CD
+#define EGL_LOWER_LEFT_KHR                0x30CE
+#define EGL_UPPER_LEFT_KHR                0x30CF
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLUNLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglLockSurfaceKHR (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglUnlockSurfaceKHR (EGLDisplay dpy, EGLSurface surface);
+#endif
+#endif /* EGL_KHR_lock_surface */
+
+#ifndef EGL_KHR_lock_surface2
+#define EGL_KHR_lock_surface2 1
+#define EGL_BITMAP_PIXEL_SIZE_KHR         0x3110
+#endif /* EGL_KHR_lock_surface2 */
+
+#ifndef EGL_KHR_lock_surface3
+#define EGL_KHR_lock_surface3 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACE64KHRPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface64KHR (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value);
+#endif
+#endif /* EGL_KHR_lock_surface3 */
+
+#ifndef EGL_KHR_platform_android
+#define EGL_KHR_platform_android 1
+#define EGL_PLATFORM_ANDROID_KHR          0x3141
+#endif /* EGL_KHR_platform_android */
+
+#ifndef EGL_KHR_platform_gbm
+#define EGL_KHR_platform_gbm 1
+#define EGL_PLATFORM_GBM_KHR              0x31D7
+#endif /* EGL_KHR_platform_gbm */
+
+#ifndef EGL_KHR_platform_wayland
+#define EGL_KHR_platform_wayland 1
+#define EGL_PLATFORM_WAYLAND_KHR          0x31D8
+#endif /* EGL_KHR_platform_wayland */
+
+#ifndef EGL_KHR_platform_x11
+#define EGL_KHR_platform_x11 1
+#define EGL_PLATFORM_X11_KHR              0x31D5
+#define EGL_PLATFORM_X11_SCREEN_KHR       0x31D6
+#endif /* EGL_KHR_platform_x11 */
+
+#ifndef EGL_KHR_reusable_sync
+#define EGL_KHR_reusable_sync 1
+typedef khronos_utime_nanoseconds_t EGLTimeKHR;
+#ifdef KHRONOS_SUPPORT_INT64
+#define EGL_SYNC_STATUS_KHR               0x30F1
+#define EGL_SIGNALED_KHR                  0x30F2
+#define EGL_UNSIGNALED_KHR                0x30F3
+#define EGL_TIMEOUT_EXPIRED_KHR           0x30F5
+#define EGL_CONDITION_SATISFIED_KHR       0x30F6
+#define EGL_SYNC_TYPE_KHR                 0x30F7
+#define EGL_SYNC_REUSABLE_KHR             0x30FA
+#define EGL_SYNC_FLUSH_COMMANDS_BIT_KHR   0x0001
+#define EGL_FOREVER_KHR                   0xFFFFFFFFFFFFFFFFull
+#define EGL_NO_SYNC_KHR                   ((EGLSyncKHR)0)
+typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNCKHRPROC) (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync);
+typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSyncKHR (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncKHR (EGLDisplay dpy, EGLSyncKHR sync);
+EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout);
+EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode);
+EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value);
+#endif
+#endif /* KHRONOS_SUPPORT_INT64 */
+#endif /* EGL_KHR_reusable_sync */
+
+#ifndef EGL_KHR_stream
+#define EGL_KHR_stream 1
+typedef void *EGLStreamKHR;
+typedef khronos_uint64_t EGLuint64KHR;
+#ifdef KHRONOS_SUPPORT_INT64
+#define EGL_NO_STREAM_KHR                 ((EGLStreamKHR)0)
+#define EGL_CONSUMER_LATENCY_USEC_KHR     0x3210
+#define EGL_PRODUCER_FRAME_KHR            0x3212
+#define EGL_CONSUMER_FRAME_KHR            0x3213
+#define EGL_STREAM_STATE_KHR              0x3214
+#define EGL_STREAM_STATE_CREATED_KHR      0x3215
+#define EGL_STREAM_STATE_CONNECTING_KHR   0x3216
+#define EGL_STREAM_STATE_EMPTY_KHR        0x3217
+#define EGL_STREAM_STATE_NEW_FRAME_AVAILABLE_KHR 0x3218
+#define EGL_STREAM_STATE_OLD_FRAME_AVAILABLE_KHR 0x3219
+#define EGL_STREAM_STATE_DISCONNECTED_KHR 0x321A
+#define EGL_BAD_STREAM_KHR                0x321B
+#define EGL_BAD_STATE_KHR                 0x321C
+typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMKHRPROC) (EGLDisplay dpy, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMATTRIBKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMU64KHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamKHR (EGLDisplay dpy, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroyStreamKHR (EGLDisplay dpy, EGLStreamKHR stream);
+EGLAPI EGLBoolean EGLAPIENTRY eglStreamAttribKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamu64KHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value);
+#endif
+#endif /* KHRONOS_SUPPORT_INT64 */
+#endif /* EGL_KHR_stream */
+
+#ifndef EGL_KHR_stream_consumer_gltexture
+#define EGL_KHR_stream_consumer_gltexture 1
+#ifdef EGL_KHR_stream
+#define EGL_CONSUMER_ACQUIRE_TIMEOUT_USEC_KHR 0x321E
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERGLTEXTUREEXTERNALKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERACQUIREKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERRELEASEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerGLTextureExternalKHR (EGLDisplay dpy, EGLStreamKHR stream);
+EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerAcquireKHR (EGLDisplay dpy, EGLStreamKHR stream);
+EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerReleaseKHR (EGLDisplay dpy, EGLStreamKHR stream);
+#endif
+#endif /* EGL_KHR_stream */
+#endif /* EGL_KHR_stream_consumer_gltexture */
+
+#ifndef EGL_KHR_stream_cross_process_fd
+#define EGL_KHR_stream_cross_process_fd 1
+typedef int EGLNativeFileDescriptorKHR;
+#ifdef EGL_KHR_stream
+#define EGL_NO_FILE_DESCRIPTOR_KHR        ((EGLNativeFileDescriptorKHR)(-1))
+typedef EGLNativeFileDescriptorKHR (EGLAPIENTRYP PFNEGLGETSTREAMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream);
+typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMFROMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLNativeFileDescriptorKHR EGLAPIENTRY eglGetStreamFileDescriptorKHR (EGLDisplay dpy, EGLStreamKHR stream);
+EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamFromFileDescriptorKHR (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor);
+#endif
+#endif /* EGL_KHR_stream */
+#endif /* EGL_KHR_stream_cross_process_fd */
+
+#ifndef EGL_KHR_stream_fifo
+#define EGL_KHR_stream_fifo 1
+#ifdef EGL_KHR_stream
+#define EGL_STREAM_FIFO_LENGTH_KHR        0x31FC
+#define EGL_STREAM_TIME_NOW_KHR           0x31FD
+#define EGL_STREAM_TIME_CONSUMER_KHR      0x31FE
+#define EGL_STREAM_TIME_PRODUCER_KHR      0x31FF
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMTIMEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamTimeKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value);
+#endif
+#endif /* EGL_KHR_stream */
+#endif /* EGL_KHR_stream_fifo */
+
+#ifndef EGL_KHR_stream_producer_aldatalocator
+#define EGL_KHR_stream_producer_aldatalocator 1
+#ifdef EGL_KHR_stream
+#endif /* EGL_KHR_stream */
+#endif /* EGL_KHR_stream_producer_aldatalocator */
+
+#ifndef EGL_KHR_stream_producer_eglsurface
+#define EGL_KHR_stream_producer_eglsurface 1
+#ifdef EGL_KHR_stream
+#define EGL_STREAM_BIT_KHR                0x0800
+typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATESTREAMPRODUCERSURFACEKHRPROC) (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSurface EGLAPIENTRY eglCreateStreamProducerSurfaceKHR (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list);
+#endif
+#endif /* EGL_KHR_stream */
+#endif /* EGL_KHR_stream_producer_eglsurface */
+
+#ifndef EGL_KHR_surfaceless_context
+#define EGL_KHR_surfaceless_context 1
+#endif /* EGL_KHR_surfaceless_context */
+
+#ifndef EGL_KHR_vg_parent_image
+#define EGL_KHR_vg_parent_image 1
+#define EGL_VG_PARENT_IMAGE_KHR           0x30BA
+#endif /* EGL_KHR_vg_parent_image */
+
+#ifndef EGL_KHR_wait_sync
+#define EGL_KHR_wait_sync 1
+typedef EGLint (EGLAPIENTRYP PFNEGLWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLint EGLAPIENTRY eglWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags);
+#endif
+#endif /* EGL_KHR_wait_sync */
+
+#ifndef EGL_ANDROID_blob_cache
+#define EGL_ANDROID_blob_cache 1
+typedef khronos_ssize_t EGLsizeiANDROID;
+typedef void (*EGLSetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, const void *value, EGLsizeiANDROID valueSize);
+typedef EGLsizeiANDROID (*EGLGetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, void *value, EGLsizeiANDROID valueSize);
+typedef void (EGLAPIENTRYP PFNEGLSETBLOBCACHEFUNCSANDROIDPROC) (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI void EGLAPIENTRY eglSetBlobCacheFuncsANDROID (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get);
+#endif
+#endif /* EGL_ANDROID_blob_cache */
+
+#ifndef EGL_ANDROID_framebuffer_target
+#define EGL_ANDROID_framebuffer_target 1
+#define EGL_FRAMEBUFFER_TARGET_ANDROID    0x3147
+#endif /* EGL_ANDROID_framebuffer_target */
+
+#ifndef EGL_ANDROID_image_native_buffer
+#define EGL_ANDROID_image_native_buffer 1
+#define EGL_NATIVE_BUFFER_ANDROID         0x3140
+#endif /* EGL_ANDROID_image_native_buffer */
+
+#ifndef EGL_ANDROID_native_fence_sync
+#define EGL_ANDROID_native_fence_sync 1
+#define EGL_SYNC_NATIVE_FENCE_ANDROID     0x3144
+#define EGL_SYNC_NATIVE_FENCE_FD_ANDROID  0x3145
+#define EGL_SYNC_NATIVE_FENCE_SIGNALED_ANDROID 0x3146
+#define EGL_NO_NATIVE_FENCE_FD_ANDROID    -1
+typedef EGLint (EGLAPIENTRYP PFNEGLDUPNATIVEFENCEFDANDROIDPROC) (EGLDisplay dpy, EGLSyncKHR sync);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLint EGLAPIENTRY eglDupNativeFenceFDANDROID (EGLDisplay dpy, EGLSyncKHR sync);
+#endif
+#endif /* EGL_ANDROID_native_fence_sync */
+
+#ifndef EGL_ANDROID_recordable
+#define EGL_ANDROID_recordable 1
+#define EGL_RECORDABLE_ANDROID            0x3142
+#endif /* EGL_ANDROID_recordable */
+
+#ifndef EGL_ANGLE_d3d_share_handle_client_buffer
+#define EGL_ANGLE_d3d_share_handle_client_buffer 1
+#define EGL_D3D_TEXTURE_2D_SHARE_HANDLE_ANGLE 0x3200
+#endif /* EGL_ANGLE_d3d_share_handle_client_buffer */
+
+#ifndef EGL_ANGLE_window_fixed_size
+#define EGL_ANGLE_window_fixed_size 1
+#define EGL_FIXED_SIZE_ANGLE        0x3201
+#endif /* EGL_ANGLE_window_fixed_size */
+
+#ifndef EGL_ANGLE_query_surface_pointer
+#define EGL_ANGLE_query_surface_pointer 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACEPOINTERANGLEPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurfacePointerANGLE (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value);
+#endif
+#endif /* EGL_ANGLE_query_surface_pointer */
+
+#ifndef EGL_ANGLE_software_display
+#define EGL_ANGLE_software_display 1
+#define EGL_SOFTWARE_DISPLAY_ANGLE ((EGLNativeDisplayType)-1)
+#endif /* EGL_ANGLE_software_display */
+
+#ifndef EGL_ANGLE_direct3d_display
+#define EGL_ANGLE_direct3d_display 1
+#define EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE ((EGLNativeDisplayType)-2)
+#define EGL_D3D11_ONLY_DISPLAY_ANGLE ((EGLNativeDisplayType)-3)
+#endif /* EGL_ANGLE_direct3d_display */
+
+#ifndef EGL_ANGLE_surface_d3d_texture_2d_share_handle
+#define EGL_ANGLE_surface_d3d_texture_2d_share_handle 1
+#endif /* EGL_ANGLE_surface_d3d_texture_2d_share_handle */
+
+#ifndef EGL_ANGLE_surface_d3d_render_to_back_buffer
+#define EGL_ANGLE_surface_d3d_render_to_back_buffer 1
+#define EGL_ANGLE_DISPLAY_ALLOW_RENDER_TO_BACK_BUFFER 0x320B
+#define EGL_ANGLE_SURFACE_RENDER_TO_BACK_BUFFER 0x320C
+#endif /* EGL_ANGLE_surface_d3d_render_to_back_buffer */
+
+#ifndef EGL_ANGLE_platform_angle
+#define EGL_ANGLE_platform_angle 1
+#define EGL_PLATFORM_ANGLE_ANGLE          0x3201
+#define EGL_PLATFORM_ANGLE_TYPE_ANGLE     0x3202
+#define EGL_PLATFORM_ANGLE_MAX_VERSION_MAJOR_ANGLE 0x3203
+#define EGL_PLATFORM_ANGLE_MAX_VERSION_MINOR_ANGLE 0x3204
+#define EGL_PLATFORM_ANGLE_TYPE_DEFAULT_ANGLE 0x3205
+#endif /* EGL_ANGLE_platform_angle */
+
+#ifndef EGL_ANGLE_platform_angle_d3d
+#define EGL_ANGLE_platform_angle_d3d 1
+#define EGL_PLATFORM_ANGLE_TYPE_D3D9_ANGLE 0x3206
+#define EGL_PLATFORM_ANGLE_TYPE_D3D11_ANGLE 0x3207
+#define EGL_PLATFORM_ANGLE_USE_WARP_ANGLE 0x3208
+#endif /* EGL_ANGLE_platform_angle_d3d */
+
+#ifndef EGL_ANGLE_platform_angle_opengl
+#define EGL_ANGLE_platform_angle_opengl 1
+#define EGL_PLATFORM_ANGLE_TYPE_OPENGL_ANGLE 0x3209
+#define EGL_PLATFORM_ANGLE_TYPE_OPENGLES_ANGLE 0x320A
+#endif /* EGL_ANGLE_platform_angle_opengl */
+
+#ifndef EGL_ARM_pixmap_multisample_discard
+#define EGL_ARM_pixmap_multisample_discard 1
+#define EGL_DISCARD_SAMPLES_ARM           0x3286
+#endif /* EGL_ARM_pixmap_multisample_discard */
+
+#ifndef EGL_EXT_buffer_age
+#define EGL_EXT_buffer_age 1
+#define EGL_BUFFER_AGE_EXT                0x313D
+#endif /* EGL_EXT_buffer_age */
+
+#ifndef EGL_EXT_client_extensions
+#define EGL_EXT_client_extensions 1
+#endif /* EGL_EXT_client_extensions */
+
+#ifndef EGL_EXT_create_context_robustness
+#define EGL_EXT_create_context_robustness 1
+#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_EXT 0x30BF
+#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_EXT 0x3138
+#define EGL_NO_RESET_NOTIFICATION_EXT     0x31BE
+#define EGL_LOSE_CONTEXT_ON_RESET_EXT     0x31BF
+#endif /* EGL_EXT_create_context_robustness */
+
+#ifndef EGL_EXT_device_base
+#define EGL_EXT_device_base 1
+typedef void *EGLDeviceEXT;
+#define EGL_NO_DEVICE_EXT                 ((EGLDeviceEXT)(0))
+#define EGL_BAD_DEVICE_EXT                0x322B
+#define EGL_DEVICE_EXT                    0x322C
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDEVICEATTRIBEXTPROC) (EGLDeviceEXT device, EGLint attribute, EGLAttrib *value);
+typedef const char *(EGLAPIENTRYP PFNEGLQUERYDEVICESTRINGEXTPROC) (EGLDeviceEXT device, EGLint name);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDEVICESEXTPROC) (EGLint max_devices, EGLDeviceEXT *devices, EGLint *num_devices);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYDISPLAYATTRIBEXTPROC) (EGLDisplay dpy, EGLint attribute, EGLAttrib *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryDeviceAttribEXT (EGLDeviceEXT device, EGLint attribute, EGLAttrib *value);
+EGLAPI const char *EGLAPIENTRY eglQueryDeviceStringEXT (EGLDeviceEXT device, EGLint name);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryDevicesEXT (EGLint max_devices, EGLDeviceEXT *devices, EGLint *num_devices);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryDisplayAttribEXT (EGLDisplay dpy, EGLint attribute, EGLAttrib *value);
+#endif
+#endif /* EGL_EXT_device_base */
+
+#ifndef EGL_EXT_image_dma_buf_import
+#define EGL_EXT_image_dma_buf_import 1
+#define EGL_LINUX_DMA_BUF_EXT             0x3270
+#define EGL_LINUX_DRM_FOURCC_EXT          0x3271
+#define EGL_DMA_BUF_PLANE0_FD_EXT         0x3272
+#define EGL_DMA_BUF_PLANE0_OFFSET_EXT     0x3273
+#define EGL_DMA_BUF_PLANE0_PITCH_EXT      0x3274
+#define EGL_DMA_BUF_PLANE1_FD_EXT         0x3275
+#define EGL_DMA_BUF_PLANE1_OFFSET_EXT     0x3276
+#define EGL_DMA_BUF_PLANE1_PITCH_EXT      0x3277
+#define EGL_DMA_BUF_PLANE2_FD_EXT         0x3278
+#define EGL_DMA_BUF_PLANE2_OFFSET_EXT     0x3279
+#define EGL_DMA_BUF_PLANE2_PITCH_EXT      0x327A
+#define EGL_YUV_COLOR_SPACE_HINT_EXT      0x327B
+#define EGL_SAMPLE_RANGE_HINT_EXT         0x327C
+#define EGL_YUV_CHROMA_HORIZONTAL_SITING_HINT_EXT 0x327D
+#define EGL_YUV_CHROMA_VERTICAL_SITING_HINT_EXT 0x327E
+#define EGL_ITU_REC601_EXT                0x327F
+#define EGL_ITU_REC709_EXT                0x3280
+#define EGL_ITU_REC2020_EXT               0x3281
+#define EGL_YUV_FULL_RANGE_EXT            0x3282
+#define EGL_YUV_NARROW_RANGE_EXT          0x3283
+#define EGL_YUV_CHROMA_SITING_0_EXT       0x3284
+#define EGL_YUV_CHROMA_SITING_0_5_EXT     0x3285
+#endif /* EGL_EXT_image_dma_buf_import */
+
+#ifndef EGL_EXT_multiview_window
+#define EGL_EXT_multiview_window 1
+#define EGL_MULTIVIEW_VIEW_COUNT_EXT      0x3134
+#endif /* EGL_EXT_multiview_window */
+
+#ifndef EGL_EXT_platform_base
+#define EGL_EXT_platform_base 1
+typedef EGLDisplay (EGLAPIENTRYP PFNEGLGETPLATFORMDISPLAYEXTPROC) (EGLenum platform, void *native_display, const EGLint *attrib_list);
+typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMWINDOWSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list);
+typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMPIXMAPSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLDisplay EGLAPIENTRY eglGetPlatformDisplayEXT (EGLenum platform, void *native_display, const EGLint *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformWindowSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list);
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformPixmapSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list);
+#endif
+#endif /* EGL_EXT_platform_base */
+
+#ifndef EGL_EXT_platform_device
+#define EGL_EXT_platform_device 1
+#define EGL_PLATFORM_DEVICE_EXT           0x313F
+#endif /* EGL_EXT_platform_device */
+
+#ifndef EGL_EXT_platform_wayland
+#define EGL_EXT_platform_wayland 1
+#define EGL_PLATFORM_WAYLAND_EXT          0x31D8
+#endif /* EGL_EXT_platform_wayland */
+
+#ifndef EGL_EXT_platform_x11
+#define EGL_EXT_platform_x11 1
+#define EGL_PLATFORM_X11_EXT              0x31D5
+#define EGL_PLATFORM_X11_SCREEN_EXT       0x31D6
+#endif /* EGL_EXT_platform_x11 */
+
+#ifndef EGL_EXT_protected_surface
+#define EGL_EXT_protected_surface 1
+#define EGL_PROTECTED_CONTENT_EXT         0x32C0
+#endif /* EGL_EXT_protected_surface */
+
+#ifndef EGL_EXT_swap_buffers_with_damage
+#define EGL_EXT_swap_buffers_with_damage 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSWITHDAMAGEEXTPROC) (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersWithDamageEXT (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects);
+#endif
+#endif /* EGL_EXT_swap_buffers_with_damage */
+
+#ifndef EGL_HI_clientpixmap
+#define EGL_HI_clientpixmap 1
+struct EGLClientPixmapHI {
+    void  *pData;
+    EGLint iWidth;
+    EGLint iHeight;
+    EGLint iStride;
+};
+#define EGL_CLIENT_PIXMAP_POINTER_HI      0x8F74
+typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPIXMAPSURFACEHIPROC) (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurfaceHI (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap);
+#endif
+#endif /* EGL_HI_clientpixmap */
+
+#ifndef EGL_HI_colorformats
+#define EGL_HI_colorformats 1
+#define EGL_COLOR_FORMAT_HI               0x8F70
+#define EGL_COLOR_RGB_HI                  0x8F71
+#define EGL_COLOR_RGBA_HI                 0x8F72
+#define EGL_COLOR_ARGB_HI                 0x8F73
+#endif /* EGL_HI_colorformats */
+
+#ifndef EGL_IMG_context_priority
+#define EGL_IMG_context_priority 1
+#define EGL_CONTEXT_PRIORITY_LEVEL_IMG    0x3100
+#define EGL_CONTEXT_PRIORITY_HIGH_IMG     0x3101
+#define EGL_CONTEXT_PRIORITY_MEDIUM_IMG   0x3102
+#define EGL_CONTEXT_PRIORITY_LOW_IMG      0x3103
+#endif /* EGL_IMG_context_priority */
+
+#ifndef EGL_MESA_drm_image
+#define EGL_MESA_drm_image 1
+#define EGL_DRM_BUFFER_FORMAT_MESA        0x31D0
+#define EGL_DRM_BUFFER_USE_MESA           0x31D1
+#define EGL_DRM_BUFFER_FORMAT_ARGB32_MESA 0x31D2
+#define EGL_DRM_BUFFER_MESA               0x31D3
+#define EGL_DRM_BUFFER_STRIDE_MESA        0x31D4
+#define EGL_DRM_BUFFER_USE_SCANOUT_MESA   0x00000001
+#define EGL_DRM_BUFFER_USE_SHARE_MESA     0x00000002
+typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEDRMIMAGEMESAPROC) (EGLDisplay dpy, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLEXPORTDRMIMAGEMESAPROC) (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLImageKHR EGLAPIENTRY eglCreateDRMImageMESA (EGLDisplay dpy, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglExportDRMImageMESA (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride);
+#endif
+#endif /* EGL_MESA_drm_image */
+
+#ifndef EGL_MESA_platform_gbm
+#define EGL_MESA_platform_gbm 1
+#define EGL_PLATFORM_GBM_MESA             0x31D7
+#endif /* EGL_MESA_platform_gbm */
+
+#ifndef EGL_NOK_swap_region
+#define EGL_NOK_swap_region 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSREGIONNOKPROC) (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersRegionNOK (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);
+#endif
+#endif /* EGL_NOK_swap_region */
+
+#ifndef EGL_NOK_swap_region2
+#define EGL_NOK_swap_region2 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSREGION2NOKPROC) (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersRegion2NOK (EGLDisplay dpy, EGLSurface surface, EGLint numRects, const EGLint *rects);
+#endif
+#endif /* EGL_NOK_swap_region2 */
+
+#ifndef EGL_NOK_texture_from_pixmap
+#define EGL_NOK_texture_from_pixmap 1
+#define EGL_Y_INVERTED_NOK                0x307F
+#endif /* EGL_NOK_texture_from_pixmap */
+
+#ifndef EGL_NV_3dvision_surface
+#define EGL_NV_3dvision_surface 1
+#define EGL_AUTO_STEREO_NV                0x3136
+#endif /* EGL_NV_3dvision_surface */
+
+#ifndef EGL_NV_coverage_sample
+#define EGL_NV_coverage_sample 1
+#define EGL_COVERAGE_BUFFERS_NV           0x30E0
+#define EGL_COVERAGE_SAMPLES_NV           0x30E1
+#endif /* EGL_NV_coverage_sample */
+
+#ifndef EGL_NV_coverage_sample_resolve
+#define EGL_NV_coverage_sample_resolve 1
+#define EGL_COVERAGE_SAMPLE_RESOLVE_NV    0x3131
+#define EGL_COVERAGE_SAMPLE_RESOLVE_DEFAULT_NV 0x3132
+#define EGL_COVERAGE_SAMPLE_RESOLVE_NONE_NV 0x3133
+#endif /* EGL_NV_coverage_sample_resolve */
+
+#ifndef EGL_NV_depth_nonlinear
+#define EGL_NV_depth_nonlinear 1
+#define EGL_DEPTH_ENCODING_NV             0x30E2
+#define EGL_DEPTH_ENCODING_NONE_NV        0
+#define EGL_DEPTH_ENCODING_NONLINEAR_NV   0x30E3
+#endif /* EGL_NV_depth_nonlinear */
+
+#ifndef EGL_NV_native_query
+#define EGL_NV_native_query 1
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEDISPLAYNVPROC) (EGLDisplay dpy, EGLNativeDisplayType *display_id);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEWINDOWNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEPIXMAPNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeDisplayNV (EGLDisplay dpy, EGLNativeDisplayType *display_id);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeWindowNV (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window);
+EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativePixmapNV (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap);
+#endif
+#endif /* EGL_NV_native_query */
+
+#ifndef EGL_NV_post_convert_rounding
+#define EGL_NV_post_convert_rounding 1
+#endif /* EGL_NV_post_convert_rounding */
+
+#ifndef EGL_NV_post_sub_buffer
+#define EGL_NV_post_sub_buffer 1
+#define EGL_POST_SUB_BUFFER_SUPPORTED_NV  0x30BE
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLPOSTSUBBUFFERNVPROC) (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLBoolean EGLAPIENTRY eglPostSubBufferNV (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height);
+#endif
+#endif /* EGL_NV_post_sub_buffer */
+
+#ifndef EGL_NV_stream_sync
+#define EGL_NV_stream_sync 1
+#define EGL_SYNC_NEW_FRAME_NV             0x321F
+typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESTREAMSYNCNVPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateStreamSyncNV (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list);
+#endif
+#endif /* EGL_NV_stream_sync */
+
+#ifndef EGL_NV_sync
+#define EGL_NV_sync 1
+typedef void *EGLSyncNV;
+typedef khronos_utime_nanoseconds_t EGLTimeNV;
+#ifdef KHRONOS_SUPPORT_INT64
+#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_NV 0x30E6
+#define EGL_SYNC_STATUS_NV                0x30E7
+#define EGL_SIGNALED_NV                   0x30E8
+#define EGL_UNSIGNALED_NV                 0x30E9
+#define EGL_SYNC_FLUSH_COMMANDS_BIT_NV    0x0001
+#define EGL_FOREVER_NV                    0xFFFFFFFFFFFFFFFFull
+#define EGL_ALREADY_SIGNALED_NV           0x30EA
+#define EGL_TIMEOUT_EXPIRED_NV            0x30EB
+#define EGL_CONDITION_SATISFIED_NV        0x30EC
+#define EGL_SYNC_TYPE_NV                  0x30ED
+#define EGL_SYNC_CONDITION_NV             0x30EE
+#define EGL_SYNC_FENCE_NV                 0x30EF
+#define EGL_NO_SYNC_NV                    ((EGLSyncNV)0)
+typedef EGLSyncNV (EGLAPIENTRYP PFNEGLCREATEFENCESYNCNVPROC) (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCNVPROC) (EGLSyncNV sync);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLFENCENVPROC) (EGLSyncNV sync);
+typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCNVPROC) (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCNVPROC) (EGLSyncNV sync, EGLenum mode);
+typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBNVPROC) (EGLSyncNV sync, EGLint attribute, EGLint *value);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLSyncNV EGLAPIENTRY eglCreateFenceSyncNV (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list);
+EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncNV (EGLSyncNV sync);
+EGLAPI EGLBoolean EGLAPIENTRY eglFenceNV (EGLSyncNV sync);
+EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncNV (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout);
+EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncNV (EGLSyncNV sync, EGLenum mode);
+EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribNV (EGLSyncNV sync, EGLint attribute, EGLint *value);
+#endif
+#endif /* KHRONOS_SUPPORT_INT64 */
+#endif /* EGL_NV_sync */
+
+#ifndef EGL_NV_system_time
+#define EGL_NV_system_time 1
+typedef khronos_utime_nanoseconds_t EGLuint64NV;
+#ifdef KHRONOS_SUPPORT_INT64
+typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMEFREQUENCYNVPROC) (void);
+typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMENVPROC) (void);
+#ifdef EGL_EGLEXT_PROTOTYPES
+EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeFrequencyNV (void);
+EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeNV (void);
+#endif
+#endif /* KHRONOS_SUPPORT_INT64 */
+#endif /* EGL_NV_system_time */
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/winrt/include/EGL/eglplatform.h b/platform/winrt/include/EGL/eglplatform.h
index b0e88f94d4..71dadc79d5 100644
--- a/platform/winrt/include/EGL/eglplatform.h
+++ b/platform/winrt/include/EGL/eglplatform.h
@@ -1,131 +1,131 @@
-#ifndef __eglplatform_h_

-#define __eglplatform_h_

-

-/*

-** Copyright (c) 2007-2013 The Khronos Group Inc.

-**

-** Permission is hereby granted, free of charge, to any person obtaining a

-** copy of this software and/or associated documentation files (the

-** "Materials"), to deal in the Materials without restriction, including

-** without limitation the rights to use, copy, modify, merge, publish,

-** distribute, sublicense, and/or sell copies of the Materials, and to

-** permit persons to whom the Materials are furnished to do so, subject to

-** the following conditions:

-**

-** The above copyright notice and this permission notice shall be included

-** in all copies or substantial portions of the Materials.

-**

-** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY

-** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

-** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

-** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

-*/

-

-/* Platform-specific types and definitions for egl.h

- * $Revision: 23432 $ on $Date: 2013-10-09 00:57:24 -0700 (Wed, 09 Oct 2013) $

- *

- * Adopters may modify khrplatform.h and this file to suit their platform.

- * You are encouraged to submit all modifications to the Khronos group so that

- * they can be included in future versions of this file.  Please submit changes

- * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)

- * by filing a bug against product "EGL" component "Registry".

- */

-

-#include <KHR/khrplatform.h>

-

-/* Macros used in EGL function prototype declarations.

- *

- * EGL functions should be prototyped as:

- *

- * EGLAPI return-type EGLAPIENTRY eglFunction(arguments);

- * typedef return-type (EXPAPIENTRYP PFNEGLFUNCTIONPROC) (arguments);

- *

- * KHRONOS_APICALL and KHRONOS_APIENTRY are defined in KHR/khrplatform.h

- */

-

-#ifndef EGLAPI

-#define EGLAPI KHRONOS_APICALL

-#endif

-

-#ifndef EGLAPIENTRY

-#define EGLAPIENTRY  KHRONOS_APIENTRY

-#endif

-#define EGLAPIENTRYP EGLAPIENTRY*

-

-/* The types NativeDisplayType, NativeWindowType, and NativePixmapType

- * are aliases of window-system-dependent types, such as X Display * or

- * Windows Device Context. They must be defined in platform-specific

- * code below. The EGL-prefixed versions of Native*Type are the same

- * types, renamed in EGL 1.3 so all types in the API start with "EGL".

- *

- * Khronos STRONGLY RECOMMENDS that you use the default definitions

- * provided below, since these changes affect both binary and source

- * portability of applications using EGL running on different EGL

- * implementations.

- */

-

-#if defined(_WIN32) || defined(__VC32__) && !defined(__CYGWIN__) && !defined(__SCITECH_SNAP__) /* Win32 and WinCE */

-#ifndef WIN32_LEAN_AND_MEAN

-#define WIN32_LEAN_AND_MEAN 1

-#endif

-#include <windows.h>

-

-typedef HDC     EGLNativeDisplayType;

-typedef HBITMAP EGLNativePixmapType;

-

-#if defined(WINAPI_FAMILY) && (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP || WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP) /* Windows Store */

-#include <inspectable.h>

-typedef IInspectable* EGLNativeWindowType;

-#else

-typedef HWND    EGLNativeWindowType;

-#endif

-

-#elif defined(__WINSCW__) || defined(__SYMBIAN32__)  /* Symbian */

-

-typedef int   EGLNativeDisplayType;

-typedef void *EGLNativeWindowType;

-typedef void *EGLNativePixmapType;

-

-#elif defined(__ANDROID__) || defined(ANDROID)

-

-#include <android/native_window.h>

-

-struct egl_native_pixmap_t;

-

-typedef struct ANativeWindow*           EGLNativeWindowType;

-typedef struct egl_native_pixmap_t*     EGLNativePixmapType;

-typedef void*                           EGLNativeDisplayType;

-

-#elif defined(__unix__)

-

-/* X11 (tentative)  */

-#include <X11/Xlib.h>

-#include <X11/Xutil.h>

-

-typedef Display *EGLNativeDisplayType;

-typedef Pixmap   EGLNativePixmapType;

-typedef Window   EGLNativeWindowType;

-

-#else

-#error "Platform not recognized"

-#endif

-

-/* EGL 1.2 types, renamed for consistency in EGL 1.3 */

-typedef EGLNativeDisplayType NativeDisplayType;

-typedef EGLNativePixmapType  NativePixmapType;

-typedef EGLNativeWindowType  NativeWindowType;

-

-

-/* Define EGLint. This must be a signed integral type large enough to contain

- * all legal attribute names and values passed into and out of EGL, whether

- * their type is boolean, bitmask, enumerant (symbolic constant), integer,

- * handle, or other.  While in general a 32-bit integer will suffice, if

- * handles are 64 bit types, then EGLint should be defined as a signed 64-bit

- * integer type.

- */

-typedef khronos_int32_t EGLint;

-

-#endif /* __eglplatform_h */

+#ifndef __eglplatform_h_
+#define __eglplatform_h_
+
+/*
+** Copyright (c) 2007-2013 The Khronos Group Inc.
+**
+** Permission is hereby granted, free of charge, to any person obtaining a
+** copy of this software and/or associated documentation files (the
+** "Materials"), to deal in the Materials without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Materials, and to
+** permit persons to whom the Materials are furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be included
+** in all copies or substantial portions of the Materials.
+**
+** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
+*/
+
+/* Platform-specific types and definitions for egl.h
+ * $Revision: 23432 $ on $Date: 2013-10-09 00:57:24 -0700 (Wed, 09 Oct 2013) $
+ *
+ * Adopters may modify khrplatform.h and this file to suit their platform.
+ * You are encouraged to submit all modifications to the Khronos group so that
+ * they can be included in future versions of this file.  Please submit changes
+ * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)
+ * by filing a bug against product "EGL" component "Registry".
+ */
+
+#include <KHR/khrplatform.h>
+
+/* Macros used in EGL function prototype declarations.
+ *
+ * EGL functions should be prototyped as:
+ *
+ * EGLAPI return-type EGLAPIENTRY eglFunction(arguments);
+ * typedef return-type (EXPAPIENTRYP PFNEGLFUNCTIONPROC) (arguments);
+ *
+ * KHRONOS_APICALL and KHRONOS_APIENTRY are defined in KHR/khrplatform.h
+ */
+
+#ifndef EGLAPI
+#define EGLAPI KHRONOS_APICALL
+#endif
+
+#ifndef EGLAPIENTRY
+#define EGLAPIENTRY  KHRONOS_APIENTRY
+#endif
+#define EGLAPIENTRYP EGLAPIENTRY*
+
+/* The types NativeDisplayType, NativeWindowType, and NativePixmapType
+ * are aliases of window-system-dependent types, such as X Display * or
+ * Windows Device Context. They must be defined in platform-specific
+ * code below. The EGL-prefixed versions of Native*Type are the same
+ * types, renamed in EGL 1.3 so all types in the API start with "EGL".
+ *
+ * Khronos STRONGLY RECOMMENDS that you use the default definitions
+ * provided below, since these changes affect both binary and source
+ * portability of applications using EGL running on different EGL
+ * implementations.
+ */
+
+#if defined(_WIN32) || defined(__VC32__) && !defined(__CYGWIN__) && !defined(__SCITECH_SNAP__) /* Win32 and WinCE */
+#ifndef WIN32_LEAN_AND_MEAN
+#define WIN32_LEAN_AND_MEAN 1
+#endif
+#include <windows.h>
+
+typedef HDC     EGLNativeDisplayType;
+typedef HBITMAP EGLNativePixmapType;
+
+#if defined(WINAPI_FAMILY) && (WINAPI_FAMILY == WINAPI_FAMILY_PC_APP || WINAPI_FAMILY == WINAPI_FAMILY_PHONE_APP) /* Windows Store */
+#include <inspectable.h>
+typedef IInspectable* EGLNativeWindowType;
+#else
+typedef HWND    EGLNativeWindowType;
+#endif
+
+#elif defined(__WINSCW__) || defined(__SYMBIAN32__)  /* Symbian */
+
+typedef int   EGLNativeDisplayType;
+typedef void *EGLNativeWindowType;
+typedef void *EGLNativePixmapType;
+
+#elif defined(__ANDROID__) || defined(ANDROID)
+
+#include <android/native_window.h>
+
+struct egl_native_pixmap_t;
+
+typedef struct ANativeWindow*           EGLNativeWindowType;
+typedef struct egl_native_pixmap_t*     EGLNativePixmapType;
+typedef void*                           EGLNativeDisplayType;
+
+#elif defined(__unix__)
+
+/* X11 (tentative)  */
+#include <X11/Xlib.h>
+#include <X11/Xutil.h>
+
+typedef Display *EGLNativeDisplayType;
+typedef Pixmap   EGLNativePixmapType;
+typedef Window   EGLNativeWindowType;
+
+#else
+#error "Platform not recognized"
+#endif
+
+/* EGL 1.2 types, renamed for consistency in EGL 1.3 */
+typedef EGLNativeDisplayType NativeDisplayType;
+typedef EGLNativePixmapType  NativePixmapType;
+typedef EGLNativeWindowType  NativeWindowType;
+
+
+/* Define EGLint. This must be a signed integral type large enough to contain
+ * all legal attribute names and values passed into and out of EGL, whether
+ * their type is boolean, bitmask, enumerant (symbolic constant), integer,
+ * handle, or other.  While in general a 32-bit integer will suffice, if
+ * handles are 64 bit types, then EGLint should be defined as a signed 64-bit
+ * integer type.
+ */
+typedef khronos_int32_t EGLint;
+
+#endif /* __eglplatform_h */
diff --git a/platform/winrt/include/GLES2/gl2.h b/platform/winrt/include/GLES2/gl2.h
index 5b3fa03c89..c2d8357268 100644
--- a/platform/winrt/include/GLES2/gl2.h
+++ b/platform/winrt/include/GLES2/gl2.h
@@ -1,620 +1,620 @@
-#ifndef __gl2_h_

-#define __gl2_h_

-

-/* $Revision: 20555 $ on $Date:: 2013-02-12 14:32:47 -0800 #$ */

-

-#include <GLES2/gl2platform.h>

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

- * This document is licensed under the SGI Free Software B License Version

- * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .

- */

-

-/*-------------------------------------------------------------------------

- * Data type definitions

- *-----------------------------------------------------------------------*/

-

-typedef void             GLvoid;

-typedef char             GLchar;

-typedef unsigned int     GLenum;

-typedef unsigned char    GLboolean;

-typedef unsigned int     GLbitfield;

-typedef khronos_int8_t   GLbyte;

-typedef short            GLshort;

-typedef int              GLint;

-typedef int              GLsizei;

-typedef khronos_uint8_t  GLubyte;

-typedef unsigned short   GLushort;

-typedef unsigned int     GLuint;

-typedef khronos_float_t  GLfloat;

-typedef khronos_float_t  GLclampf;

-typedef khronos_int32_t  GLfixed;

-

-/* GL types for handling large vertex buffer objects */

-typedef khronos_intptr_t GLintptr;

-typedef khronos_ssize_t  GLsizeiptr;

-

-/* OpenGL ES core versions */

-#define GL_ES_VERSION_2_0                 1

-

-/* ClearBufferMask */

-#define GL_DEPTH_BUFFER_BIT               0x00000100

-#define GL_STENCIL_BUFFER_BIT             0x00000400

-#define GL_COLOR_BUFFER_BIT               0x00004000

-

-/* Boolean */

-#define GL_FALSE                          0

-#define GL_TRUE                           1

-

-/* BeginMode */

-#define GL_POINTS                         0x0000

-#define GL_LINES                          0x0001

-#define GL_LINE_LOOP                      0x0002

-#define GL_LINE_STRIP                     0x0003

-#define GL_TRIANGLES                      0x0004

-#define GL_TRIANGLE_STRIP                 0x0005

-#define GL_TRIANGLE_FAN                   0x0006

-

-/* AlphaFunction (not supported in ES20) */

-/*      GL_NEVER */

-/*      GL_LESS */

-/*      GL_EQUAL */

-/*      GL_LEQUAL */

-/*      GL_GREATER */

-/*      GL_NOTEQUAL */

-/*      GL_GEQUAL */

-/*      GL_ALWAYS */

-

-/* BlendingFactorDest */

-#define GL_ZERO                           0

-#define GL_ONE                            1

-#define GL_SRC_COLOR                      0x0300

-#define GL_ONE_MINUS_SRC_COLOR            0x0301

-#define GL_SRC_ALPHA                      0x0302

-#define GL_ONE_MINUS_SRC_ALPHA            0x0303

-#define GL_DST_ALPHA                      0x0304

-#define GL_ONE_MINUS_DST_ALPHA            0x0305

-

-/* BlendingFactorSrc */

-/*      GL_ZERO */

-/*      GL_ONE */

-#define GL_DST_COLOR                      0x0306

-#define GL_ONE_MINUS_DST_COLOR            0x0307

-#define GL_SRC_ALPHA_SATURATE             0x0308

-/*      GL_SRC_ALPHA */

-/*      GL_ONE_MINUS_SRC_ALPHA */

-/*      GL_DST_ALPHA */

-/*      GL_ONE_MINUS_DST_ALPHA */

-

-/* BlendEquationSeparate */

-#define GL_FUNC_ADD                       0x8006

-#define GL_BLEND_EQUATION                 0x8009

-#define GL_BLEND_EQUATION_RGB             0x8009    /* same as BLEND_EQUATION */

-#define GL_BLEND_EQUATION_ALPHA           0x883D

-

-/* BlendSubtract */

-#define GL_FUNC_SUBTRACT                  0x800A

-#define GL_FUNC_REVERSE_SUBTRACT          0x800B

-

-/* Separate Blend Functions */

-#define GL_BLEND_DST_RGB                  0x80C8

-#define GL_BLEND_SRC_RGB                  0x80C9

-#define GL_BLEND_DST_ALPHA                0x80CA

-#define GL_BLEND_SRC_ALPHA                0x80CB

-#define GL_CONSTANT_COLOR                 0x8001

-#define GL_ONE_MINUS_CONSTANT_COLOR       0x8002

-#define GL_CONSTANT_ALPHA                 0x8003

-#define GL_ONE_MINUS_CONSTANT_ALPHA       0x8004

-#define GL_BLEND_COLOR                    0x8005

-

-/* Buffer Objects */

-#define GL_ARRAY_BUFFER                   0x8892

-#define GL_ELEMENT_ARRAY_BUFFER           0x8893

-#define GL_ARRAY_BUFFER_BINDING           0x8894

-#define GL_ELEMENT_ARRAY_BUFFER_BINDING   0x8895

-

-#define GL_STREAM_DRAW                    0x88E0

-#define GL_STATIC_DRAW                    0x88E4

-#define GL_DYNAMIC_DRAW                   0x88E8

-

-#define GL_BUFFER_SIZE                    0x8764

-#define GL_BUFFER_USAGE                   0x8765

-

-#define GL_CURRENT_VERTEX_ATTRIB          0x8626

-

-/* CullFaceMode */

-#define GL_FRONT                          0x0404

-#define GL_BACK                           0x0405

-#define GL_FRONT_AND_BACK                 0x0408

-

-/* DepthFunction */

-/*      GL_NEVER */

-/*      GL_LESS */

-/*      GL_EQUAL */

-/*      GL_LEQUAL */

-/*      GL_GREATER */

-/*      GL_NOTEQUAL */

-/*      GL_GEQUAL */

-/*      GL_ALWAYS */

-

-/* EnableCap */

-#define GL_TEXTURE_2D                     0x0DE1

-#define GL_CULL_FACE                      0x0B44

-#define GL_BLEND                          0x0BE2

-#define GL_DITHER                         0x0BD0

-#define GL_STENCIL_TEST                   0x0B90

-#define GL_DEPTH_TEST                     0x0B71

-#define GL_SCISSOR_TEST                   0x0C11

-#define GL_POLYGON_OFFSET_FILL            0x8037

-#define GL_SAMPLE_ALPHA_TO_COVERAGE       0x809E

-#define GL_SAMPLE_COVERAGE                0x80A0

-

-/* ErrorCode */

-#define GL_NO_ERROR                       0

-#define GL_INVALID_ENUM                   0x0500

-#define GL_INVALID_VALUE                  0x0501

-#define GL_INVALID_OPERATION              0x0502

-#define GL_OUT_OF_MEMORY                  0x0505

-

-/* FrontFaceDirection */

-#define GL_CW                             0x0900

-#define GL_CCW                            0x0901

-

-/* GetPName */

-#define GL_LINE_WIDTH                     0x0B21

-#define GL_ALIASED_POINT_SIZE_RANGE       0x846D

-#define GL_ALIASED_LINE_WIDTH_RANGE       0x846E

-#define GL_CULL_FACE_MODE                 0x0B45

-#define GL_FRONT_FACE                     0x0B46

-#define GL_DEPTH_RANGE                    0x0B70

-#define GL_DEPTH_WRITEMASK                0x0B72

-#define GL_DEPTH_CLEAR_VALUE              0x0B73

-#define GL_DEPTH_FUNC                     0x0B74

-#define GL_STENCIL_CLEAR_VALUE            0x0B91

-#define GL_STENCIL_FUNC                   0x0B92

-#define GL_STENCIL_FAIL                   0x0B94

-#define GL_STENCIL_PASS_DEPTH_FAIL        0x0B95

-#define GL_STENCIL_PASS_DEPTH_PASS        0x0B96

-#define GL_STENCIL_REF                    0x0B97

-#define GL_STENCIL_VALUE_MASK             0x0B93

-#define GL_STENCIL_WRITEMASK              0x0B98

-#define GL_STENCIL_BACK_FUNC              0x8800

-#define GL_STENCIL_BACK_FAIL              0x8801

-#define GL_STENCIL_BACK_PASS_DEPTH_FAIL   0x8802

-#define GL_STENCIL_BACK_PASS_DEPTH_PASS   0x8803

-#define GL_STENCIL_BACK_REF               0x8CA3

-#define GL_STENCIL_BACK_VALUE_MASK        0x8CA4

-#define GL_STENCIL_BACK_WRITEMASK         0x8CA5

-#define GL_VIEWPORT                       0x0BA2

-#define GL_SCISSOR_BOX                    0x0C10

-/*      GL_SCISSOR_TEST */

-#define GL_COLOR_CLEAR_VALUE              0x0C22

-#define GL_COLOR_WRITEMASK                0x0C23

-#define GL_UNPACK_ALIGNMENT               0x0CF5

-#define GL_PACK_ALIGNMENT                 0x0D05

-#define GL_MAX_TEXTURE_SIZE               0x0D33

-#define GL_MAX_VIEWPORT_DIMS              0x0D3A

-#define GL_SUBPIXEL_BITS                  0x0D50

-#define GL_RED_BITS                       0x0D52

-#define GL_GREEN_BITS                     0x0D53

-#define GL_BLUE_BITS                      0x0D54

-#define GL_ALPHA_BITS                     0x0D55

-#define GL_DEPTH_BITS                     0x0D56

-#define GL_STENCIL_BITS                   0x0D57

-#define GL_POLYGON_OFFSET_UNITS           0x2A00

-/*      GL_POLYGON_OFFSET_FILL */

-#define GL_POLYGON_OFFSET_FACTOR          0x8038

-#define GL_TEXTURE_BINDING_2D             0x8069

-#define GL_SAMPLE_BUFFERS                 0x80A8

-#define GL_SAMPLES                        0x80A9

-#define GL_SAMPLE_COVERAGE_VALUE          0x80AA

-#define GL_SAMPLE_COVERAGE_INVERT         0x80AB

-

-/* GetTextureParameter */

-/*      GL_TEXTURE_MAG_FILTER */

-/*      GL_TEXTURE_MIN_FILTER */

-/*      GL_TEXTURE_WRAP_S */

-/*      GL_TEXTURE_WRAP_T */

-

-#define GL_NUM_COMPRESSED_TEXTURE_FORMATS 0x86A2

-#define GL_COMPRESSED_TEXTURE_FORMATS     0x86A3

-

-/* HintMode */

-#define GL_DONT_CARE                      0x1100

-#define GL_FASTEST                        0x1101

-#define GL_NICEST                         0x1102

-

-/* HintTarget */

-#define GL_GENERATE_MIPMAP_HINT            0x8192

-

-/* DataType */

-#define GL_BYTE                           0x1400

-#define GL_UNSIGNED_BYTE                  0x1401

-#define GL_SHORT                          0x1402

-#define GL_UNSIGNED_SHORT                 0x1403

-#define GL_INT                            0x1404

-#define GL_UNSIGNED_INT                   0x1405

-#define GL_FLOAT                          0x1406

-#define GL_FIXED                          0x140C

-

-/* PixelFormat */

-#define GL_DEPTH_COMPONENT                0x1902

-#define GL_ALPHA                          0x1906

-#define GL_RGB                            0x1907

-#define GL_RGBA                           0x1908

-#define GL_LUMINANCE                      0x1909

-#define GL_LUMINANCE_ALPHA                0x190A

-

-/* PixelType */

-/*      GL_UNSIGNED_BYTE */

-#define GL_UNSIGNED_SHORT_4_4_4_4         0x8033

-#define GL_UNSIGNED_SHORT_5_5_5_1         0x8034

-#define GL_UNSIGNED_SHORT_5_6_5           0x8363

-

-/* Shaders */

-#define GL_FRAGMENT_SHADER                  0x8B30

-#define GL_VERTEX_SHADER                    0x8B31

-#define GL_MAX_VERTEX_ATTRIBS               0x8869

-#define GL_MAX_VERTEX_UNIFORM_VECTORS       0x8DFB

-#define GL_MAX_VARYING_VECTORS              0x8DFC

-#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS 0x8B4D

-#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS   0x8B4C

-#define GL_MAX_TEXTURE_IMAGE_UNITS          0x8872

-#define GL_MAX_FRAGMENT_UNIFORM_VECTORS     0x8DFD

-#define GL_SHADER_TYPE                      0x8B4F

-#define GL_DELETE_STATUS                    0x8B80

-#define GL_LINK_STATUS                      0x8B82

-#define GL_VALIDATE_STATUS                  0x8B83

-#define GL_ATTACHED_SHADERS                 0x8B85

-#define GL_ACTIVE_UNIFORMS                  0x8B86

-#define GL_ACTIVE_UNIFORM_MAX_LENGTH        0x8B87

-#define GL_ACTIVE_ATTRIBUTES                0x8B89

-#define GL_ACTIVE_ATTRIBUTE_MAX_LENGTH      0x8B8A

-#define GL_SHADING_LANGUAGE_VERSION         0x8B8C

-#define GL_CURRENT_PROGRAM                  0x8B8D

-

-/* StencilFunction */

-#define GL_NEVER                          0x0200

-#define GL_LESS                           0x0201

-#define GL_EQUAL                          0x0202

-#define GL_LEQUAL                         0x0203

-#define GL_GREATER                        0x0204

-#define GL_NOTEQUAL                       0x0205

-#define GL_GEQUAL                         0x0206

-#define GL_ALWAYS                         0x0207

-

-/* StencilOp */

-/*      GL_ZERO */

-#define GL_KEEP                           0x1E00

-#define GL_REPLACE                        0x1E01

-#define GL_INCR                           0x1E02

-#define GL_DECR                           0x1E03

-#define GL_INVERT                         0x150A

-#define GL_INCR_WRAP                      0x8507

-#define GL_DECR_WRAP                      0x8508

-

-/* StringName */

-#define GL_VENDOR                         0x1F00

-#define GL_RENDERER                       0x1F01

-#define GL_VERSION                        0x1F02

-#define GL_EXTENSIONS                     0x1F03

-

-/* TextureMagFilter */

-#define GL_NEAREST                        0x2600

-#define GL_LINEAR                         0x2601

-

-/* TextureMinFilter */

-/*      GL_NEAREST */

-/*      GL_LINEAR */

-#define GL_NEAREST_MIPMAP_NEAREST         0x2700

-#define GL_LINEAR_MIPMAP_NEAREST          0x2701

-#define GL_NEAREST_MIPMAP_LINEAR          0x2702

-#define GL_LINEAR_MIPMAP_LINEAR           0x2703

-

-/* TextureParameterName */

-#define GL_TEXTURE_MAG_FILTER             0x2800

-#define GL_TEXTURE_MIN_FILTER             0x2801

-#define GL_TEXTURE_WRAP_S                 0x2802

-#define GL_TEXTURE_WRAP_T                 0x2803

-

-/* TextureTarget */

-/*      GL_TEXTURE_2D */

-#define GL_TEXTURE                        0x1702

-

-#define GL_TEXTURE_CUBE_MAP               0x8513

-#define GL_TEXTURE_BINDING_CUBE_MAP       0x8514

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_X    0x8515

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X    0x8516

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y    0x8517

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y    0x8518

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z    0x8519

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z    0x851A

-#define GL_MAX_CUBE_MAP_TEXTURE_SIZE      0x851C

-

-/* TextureUnit */

-#define GL_TEXTURE0                       0x84C0

-#define GL_TEXTURE1                       0x84C1

-#define GL_TEXTURE2                       0x84C2

-#define GL_TEXTURE3                       0x84C3

-#define GL_TEXTURE4                       0x84C4

-#define GL_TEXTURE5                       0x84C5

-#define GL_TEXTURE6                       0x84C6

-#define GL_TEXTURE7                       0x84C7

-#define GL_TEXTURE8                       0x84C8

-#define GL_TEXTURE9                       0x84C9

-#define GL_TEXTURE10                      0x84CA

-#define GL_TEXTURE11                      0x84CB

-#define GL_TEXTURE12                      0x84CC

-#define GL_TEXTURE13                      0x84CD

-#define GL_TEXTURE14                      0x84CE

-#define GL_TEXTURE15                      0x84CF

-#define GL_TEXTURE16                      0x84D0

-#define GL_TEXTURE17                      0x84D1

-#define GL_TEXTURE18                      0x84D2

-#define GL_TEXTURE19                      0x84D3

-#define GL_TEXTURE20                      0x84D4

-#define GL_TEXTURE21                      0x84D5

-#define GL_TEXTURE22                      0x84D6

-#define GL_TEXTURE23                      0x84D7

-#define GL_TEXTURE24                      0x84D8

-#define GL_TEXTURE25                      0x84D9

-#define GL_TEXTURE26                      0x84DA

-#define GL_TEXTURE27                      0x84DB

-#define GL_TEXTURE28                      0x84DC

-#define GL_TEXTURE29                      0x84DD

-#define GL_TEXTURE30                      0x84DE

-#define GL_TEXTURE31                      0x84DF

-#define GL_ACTIVE_TEXTURE                 0x84E0

-

-/* TextureWrapMode */

-#define GL_REPEAT                         0x2901

-#define GL_CLAMP_TO_EDGE                  0x812F

-#define GL_MIRRORED_REPEAT                0x8370

-

-/* Uniform Types */

-#define GL_FLOAT_VEC2                     0x8B50

-#define GL_FLOAT_VEC3                     0x8B51

-#define GL_FLOAT_VEC4                     0x8B52

-#define GL_INT_VEC2                       0x8B53

-#define GL_INT_VEC3                       0x8B54

-#define GL_INT_VEC4                       0x8B55

-#define GL_BOOL                           0x8B56

-#define GL_BOOL_VEC2                      0x8B57

-#define GL_BOOL_VEC3                      0x8B58

-#define GL_BOOL_VEC4                      0x8B59

-#define GL_FLOAT_MAT2                     0x8B5A

-#define GL_FLOAT_MAT3                     0x8B5B

-#define GL_FLOAT_MAT4                     0x8B5C

-#define GL_SAMPLER_2D                     0x8B5E

-#define GL_SAMPLER_CUBE                   0x8B60

-

-/* Vertex Arrays */

-#define GL_VERTEX_ATTRIB_ARRAY_ENABLED        0x8622

-#define GL_VERTEX_ATTRIB_ARRAY_SIZE           0x8623

-#define GL_VERTEX_ATTRIB_ARRAY_STRIDE         0x8624

-#define GL_VERTEX_ATTRIB_ARRAY_TYPE           0x8625

-#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED     0x886A

-#define GL_VERTEX_ATTRIB_ARRAY_POINTER        0x8645

-#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING 0x889F

-

-/* Read Format */

-#define GL_IMPLEMENTATION_COLOR_READ_TYPE   0x8B9A

-#define GL_IMPLEMENTATION_COLOR_READ_FORMAT 0x8B9B

-

-/* Shader Source */

-#define GL_COMPILE_STATUS                 0x8B81

-#define GL_INFO_LOG_LENGTH                0x8B84

-#define GL_SHADER_SOURCE_LENGTH           0x8B88

-#define GL_SHADER_COMPILER                0x8DFA

-

-/* Shader Binary */

-#define GL_SHADER_BINARY_FORMATS          0x8DF8

-#define GL_NUM_SHADER_BINARY_FORMATS      0x8DF9

-

-/* Shader Precision-Specified Types */

-#define GL_LOW_FLOAT                      0x8DF0

-#define GL_MEDIUM_FLOAT                   0x8DF1

-#define GL_HIGH_FLOAT                     0x8DF2

-#define GL_LOW_INT                        0x8DF3

-#define GL_MEDIUM_INT                     0x8DF4

-#define GL_HIGH_INT                       0x8DF5

-

-/* Framebuffer Object. */

-#define GL_FRAMEBUFFER                    0x8D40

-#define GL_RENDERBUFFER                   0x8D41

-

-#define GL_RGBA4                          0x8056

-#define GL_RGB5_A1                        0x8057

-#define GL_RGB565                         0x8D62

-#define GL_DEPTH_COMPONENT16              0x81A5

-#define GL_STENCIL_INDEX8                 0x8D48

-

-#define GL_RENDERBUFFER_WIDTH             0x8D42

-#define GL_RENDERBUFFER_HEIGHT            0x8D43

-#define GL_RENDERBUFFER_INTERNAL_FORMAT   0x8D44

-#define GL_RENDERBUFFER_RED_SIZE          0x8D50

-#define GL_RENDERBUFFER_GREEN_SIZE        0x8D51

-#define GL_RENDERBUFFER_BLUE_SIZE         0x8D52

-#define GL_RENDERBUFFER_ALPHA_SIZE        0x8D53

-#define GL_RENDERBUFFER_DEPTH_SIZE        0x8D54

-#define GL_RENDERBUFFER_STENCIL_SIZE      0x8D55

-

-#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE           0x8CD0

-#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME           0x8CD1

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL         0x8CD2

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE 0x8CD3

-

-#define GL_COLOR_ATTACHMENT0              0x8CE0

-#define GL_DEPTH_ATTACHMENT               0x8D00

-#define GL_STENCIL_ATTACHMENT             0x8D20

-

-#define GL_NONE                           0

-

-#define GL_FRAMEBUFFER_COMPLETE                      0x8CD5

-#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT         0x8CD6

-#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT 0x8CD7

-#define GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS         0x8CD9

-#define GL_FRAMEBUFFER_UNSUPPORTED                   0x8CDD

-

-#define GL_FRAMEBUFFER_BINDING            0x8CA6

-#define GL_RENDERBUFFER_BINDING           0x8CA7

-#define GL_MAX_RENDERBUFFER_SIZE          0x84E8

-

-#define GL_INVALID_FRAMEBUFFER_OPERATION  0x0506

-

-/*-------------------------------------------------------------------------

- * GL core functions.

- *-----------------------------------------------------------------------*/

-

-GL_APICALL void         GL_APIENTRY glActiveTexture (GLenum texture);

-GL_APICALL void         GL_APIENTRY glAttachShader (GLuint program, GLuint shader);

-GL_APICALL void         GL_APIENTRY glBindAttribLocation (GLuint program, GLuint index, const GLchar* name);

-GL_APICALL void         GL_APIENTRY glBindBuffer (GLenum target, GLuint buffer);

-GL_APICALL void         GL_APIENTRY glBindFramebuffer (GLenum target, GLuint framebuffer);

-GL_APICALL void         GL_APIENTRY glBindRenderbuffer (GLenum target, GLuint renderbuffer);

-GL_APICALL void         GL_APIENTRY glBindTexture (GLenum target, GLuint texture);

-GL_APICALL void         GL_APIENTRY glBlendColor (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha);

-GL_APICALL void         GL_APIENTRY glBlendEquation ( GLenum mode );

-GL_APICALL void         GL_APIENTRY glBlendEquationSeparate (GLenum modeRGB, GLenum modeAlpha);

-GL_APICALL void         GL_APIENTRY glBlendFunc (GLenum sfactor, GLenum dfactor);

-GL_APICALL void         GL_APIENTRY glBlendFuncSeparate (GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);

-GL_APICALL void         GL_APIENTRY glBufferData (GLenum target, GLsizeiptr size, const GLvoid* data, GLenum usage);

-GL_APICALL void         GL_APIENTRY glBufferSubData (GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid* data);

-GL_APICALL GLenum       GL_APIENTRY glCheckFramebufferStatus (GLenum target);

-GL_APICALL void         GL_APIENTRY glClear (GLbitfield mask);

-GL_APICALL void         GL_APIENTRY glClearColor (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha);

-GL_APICALL void         GL_APIENTRY glClearDepthf (GLclampf depth);

-GL_APICALL void         GL_APIENTRY glClearStencil (GLint s);

-GL_APICALL void         GL_APIENTRY glColorMask (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha);

-GL_APICALL void         GL_APIENTRY glCompileShader (GLuint shader);

-GL_APICALL void         GL_APIENTRY glCompressedTexImage2D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void         GL_APIENTRY glCompressedTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void         GL_APIENTRY glCopyTexImage2D (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border);

-GL_APICALL void         GL_APIENTRY glCopyTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL GLuint       GL_APIENTRY glCreateProgram (void);

-GL_APICALL GLuint       GL_APIENTRY glCreateShader (GLenum type);

-GL_APICALL void         GL_APIENTRY glCullFace (GLenum mode);

-GL_APICALL void         GL_APIENTRY glDeleteBuffers (GLsizei n, const GLuint* buffers);

-GL_APICALL void         GL_APIENTRY glDeleteFramebuffers (GLsizei n, const GLuint* framebuffers);

-GL_APICALL void         GL_APIENTRY glDeleteProgram (GLuint program);

-GL_APICALL void         GL_APIENTRY glDeleteRenderbuffers (GLsizei n, const GLuint* renderbuffers);

-GL_APICALL void         GL_APIENTRY glDeleteShader (GLuint shader);

-GL_APICALL void         GL_APIENTRY glDeleteTextures (GLsizei n, const GLuint* textures);

-GL_APICALL void         GL_APIENTRY glDepthFunc (GLenum func);

-GL_APICALL void         GL_APIENTRY glDepthMask (GLboolean flag);

-GL_APICALL void         GL_APIENTRY glDepthRangef (GLclampf zNear, GLclampf zFar);

-GL_APICALL void         GL_APIENTRY glDetachShader (GLuint program, GLuint shader);

-GL_APICALL void         GL_APIENTRY glDisable (GLenum cap);

-GL_APICALL void         GL_APIENTRY glDisableVertexAttribArray (GLuint index);

-GL_APICALL void         GL_APIENTRY glDrawArrays (GLenum mode, GLint first, GLsizei count);

-GL_APICALL void         GL_APIENTRY glDrawElements (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices);

-GL_APICALL void         GL_APIENTRY glEnable (GLenum cap);

-GL_APICALL void         GL_APIENTRY glEnableVertexAttribArray (GLuint index);

-GL_APICALL void         GL_APIENTRY glFinish (void);

-GL_APICALL void         GL_APIENTRY glFlush (void);

-GL_APICALL void         GL_APIENTRY glFramebufferRenderbuffer (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer);

-GL_APICALL void         GL_APIENTRY glFramebufferTexture2D (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level);

-GL_APICALL void         GL_APIENTRY glFrontFace (GLenum mode);

-GL_APICALL void         GL_APIENTRY glGenBuffers (GLsizei n, GLuint* buffers);

-GL_APICALL void         GL_APIENTRY glGenerateMipmap (GLenum target);

-GL_APICALL void         GL_APIENTRY glGenFramebuffers (GLsizei n, GLuint* framebuffers);

-GL_APICALL void         GL_APIENTRY glGenRenderbuffers (GLsizei n, GLuint* renderbuffers);

-GL_APICALL void         GL_APIENTRY glGenTextures (GLsizei n, GLuint* textures);

-GL_APICALL void         GL_APIENTRY glGetActiveAttrib (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);

-GL_APICALL void         GL_APIENTRY glGetActiveUniform (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);

-GL_APICALL void         GL_APIENTRY glGetAttachedShaders (GLuint program, GLsizei maxcount, GLsizei* count, GLuint* shaders);

-GL_APICALL GLint        GL_APIENTRY glGetAttribLocation (GLuint program, const GLchar* name);

-GL_APICALL void         GL_APIENTRY glGetBooleanv (GLenum pname, GLboolean* params);

-GL_APICALL void         GL_APIENTRY glGetBufferParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL GLenum       GL_APIENTRY glGetError (void);

-GL_APICALL void         GL_APIENTRY glGetFloatv (GLenum pname, GLfloat* params);

-GL_APICALL void         GL_APIENTRY glGetFramebufferAttachmentParameteriv (GLenum target, GLenum attachment, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetIntegerv (GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetProgramiv (GLuint program, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetProgramInfoLog (GLuint program, GLsizei bufsize, GLsizei* length, GLchar* infolog);

-GL_APICALL void         GL_APIENTRY glGetRenderbufferParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetShaderiv (GLuint shader, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetShaderInfoLog (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* infolog);

-GL_APICALL void         GL_APIENTRY glGetShaderPrecisionFormat (GLenum shadertype, GLenum precisiontype, GLint* range, GLint* precision);

-GL_APICALL void         GL_APIENTRY glGetShaderSource (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* source);

-GL_APICALL const GLubyte* GL_APIENTRY glGetString (GLenum name);

-GL_APICALL void         GL_APIENTRY glGetTexParameterfv (GLenum target, GLenum pname, GLfloat* params);

-GL_APICALL void         GL_APIENTRY glGetTexParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetUniformfv (GLuint program, GLint location, GLfloat* params);

-GL_APICALL void         GL_APIENTRY glGetUniformiv (GLuint program, GLint location, GLint* params);

-GL_APICALL GLint        GL_APIENTRY glGetUniformLocation (GLuint program, const GLchar* name);

-GL_APICALL void         GL_APIENTRY glGetVertexAttribfv (GLuint index, GLenum pname, GLfloat* params);

-GL_APICALL void         GL_APIENTRY glGetVertexAttribiv (GLuint index, GLenum pname, GLint* params);

-GL_APICALL void         GL_APIENTRY glGetVertexAttribPointerv (GLuint index, GLenum pname, GLvoid** pointer);

-GL_APICALL void         GL_APIENTRY glHint (GLenum target, GLenum mode);

-GL_APICALL GLboolean    GL_APIENTRY glIsBuffer (GLuint buffer);

-GL_APICALL GLboolean    GL_APIENTRY glIsEnabled (GLenum cap);

-GL_APICALL GLboolean    GL_APIENTRY glIsFramebuffer (GLuint framebuffer);

-GL_APICALL GLboolean    GL_APIENTRY glIsProgram (GLuint program);

-GL_APICALL GLboolean    GL_APIENTRY glIsRenderbuffer (GLuint renderbuffer);

-GL_APICALL GLboolean    GL_APIENTRY glIsShader (GLuint shader);

-GL_APICALL GLboolean    GL_APIENTRY glIsTexture (GLuint texture);

-GL_APICALL void         GL_APIENTRY glLineWidth (GLfloat width);

-GL_APICALL void         GL_APIENTRY glLinkProgram (GLuint program);

-GL_APICALL void         GL_APIENTRY glPixelStorei (GLenum pname, GLint param);

-GL_APICALL void         GL_APIENTRY glPolygonOffset (GLfloat factor, GLfloat units);

-GL_APICALL void         GL_APIENTRY glReadPixels (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid* pixels);

-GL_APICALL void         GL_APIENTRY glReleaseShaderCompiler (void);

-GL_APICALL void         GL_APIENTRY glRenderbufferStorage (GLenum target, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void         GL_APIENTRY glSampleCoverage (GLclampf value, GLboolean invert);

-GL_APICALL void         GL_APIENTRY glScissor (GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL void         GL_APIENTRY glShaderBinary (GLsizei n, const GLuint* shaders, GLenum binaryformat, const GLvoid* binary, GLsizei length);

-GL_APICALL void         GL_APIENTRY glShaderSource (GLuint shader, GLsizei count, const GLchar* const* string, const GLint* length);

-GL_APICALL void         GL_APIENTRY glStencilFunc (GLenum func, GLint ref, GLuint mask);

-GL_APICALL void         GL_APIENTRY glStencilFuncSeparate (GLenum face, GLenum func, GLint ref, GLuint mask);

-GL_APICALL void         GL_APIENTRY glStencilMask (GLuint mask);

-GL_APICALL void         GL_APIENTRY glStencilMaskSeparate (GLenum face, GLuint mask);

-GL_APICALL void         GL_APIENTRY glStencilOp (GLenum fail, GLenum zfail, GLenum zpass);

-GL_APICALL void         GL_APIENTRY glStencilOpSeparate (GLenum face, GLenum fail, GLenum zfail, GLenum zpass);

-GL_APICALL void         GL_APIENTRY glTexImage2D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void         GL_APIENTRY glTexParameterf (GLenum target, GLenum pname, GLfloat param);

-GL_APICALL void         GL_APIENTRY glTexParameterfv (GLenum target, GLenum pname, const GLfloat* params);

-GL_APICALL void         GL_APIENTRY glTexParameteri (GLenum target, GLenum pname, GLint param);

-GL_APICALL void         GL_APIENTRY glTexParameteriv (GLenum target, GLenum pname, const GLint* params);

-GL_APICALL void         GL_APIENTRY glTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void         GL_APIENTRY glUniform1f (GLint location, GLfloat x);

-GL_APICALL void         GL_APIENTRY glUniform1fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void         GL_APIENTRY glUniform1i (GLint location, GLint x);

-GL_APICALL void         GL_APIENTRY glUniform1iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void         GL_APIENTRY glUniform2f (GLint location, GLfloat x, GLfloat y);

-GL_APICALL void         GL_APIENTRY glUniform2fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void         GL_APIENTRY glUniform2i (GLint location, GLint x, GLint y);

-GL_APICALL void         GL_APIENTRY glUniform2iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void         GL_APIENTRY glUniform3f (GLint location, GLfloat x, GLfloat y, GLfloat z);

-GL_APICALL void         GL_APIENTRY glUniform3fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void         GL_APIENTRY glUniform3i (GLint location, GLint x, GLint y, GLint z);

-GL_APICALL void         GL_APIENTRY glUniform3iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void         GL_APIENTRY glUniform4f (GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-GL_APICALL void         GL_APIENTRY glUniform4fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void         GL_APIENTRY glUniform4i (GLint location, GLint x, GLint y, GLint z, GLint w);

-GL_APICALL void         GL_APIENTRY glUniform4iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void         GL_APIENTRY glUniformMatrix2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void         GL_APIENTRY glUniformMatrix3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void         GL_APIENTRY glUniformMatrix4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void         GL_APIENTRY glUseProgram (GLuint program);

-GL_APICALL void         GL_APIENTRY glValidateProgram (GLuint program);

-GL_APICALL void         GL_APIENTRY glVertexAttrib1f (GLuint indx, GLfloat x);

-GL_APICALL void         GL_APIENTRY glVertexAttrib1fv (GLuint indx, const GLfloat* values);

-GL_APICALL void         GL_APIENTRY glVertexAttrib2f (GLuint indx, GLfloat x, GLfloat y);

-GL_APICALL void         GL_APIENTRY glVertexAttrib2fv (GLuint indx, const GLfloat* values);

-GL_APICALL void         GL_APIENTRY glVertexAttrib3f (GLuint indx, GLfloat x, GLfloat y, GLfloat z);

-GL_APICALL void         GL_APIENTRY glVertexAttrib3fv (GLuint indx, const GLfloat* values);

-GL_APICALL void         GL_APIENTRY glVertexAttrib4f (GLuint indx, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-GL_APICALL void         GL_APIENTRY glVertexAttrib4fv (GLuint indx, const GLfloat* values);

-GL_APICALL void         GL_APIENTRY glVertexAttribPointer (GLuint indx, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid* ptr);

-GL_APICALL void         GL_APIENTRY glViewport (GLint x, GLint y, GLsizei width, GLsizei height);

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif /* __gl2_h_ */

+#ifndef __gl2_h_
+#define __gl2_h_
+
+/* $Revision: 20555 $ on $Date:: 2013-02-12 14:32:47 -0800 #$ */
+
+#include <GLES2/gl2platform.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * This document is licensed under the SGI Free Software B License Version
+ * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .
+ */
+
+/*-------------------------------------------------------------------------
+ * Data type definitions
+ *-----------------------------------------------------------------------*/
+
+typedef void             GLvoid;
+typedef char             GLchar;
+typedef unsigned int     GLenum;
+typedef unsigned char    GLboolean;
+typedef unsigned int     GLbitfield;
+typedef khronos_int8_t   GLbyte;
+typedef short            GLshort;
+typedef int              GLint;
+typedef int              GLsizei;
+typedef khronos_uint8_t  GLubyte;
+typedef unsigned short   GLushort;
+typedef unsigned int     GLuint;
+typedef khronos_float_t  GLfloat;
+typedef khronos_float_t  GLclampf;
+typedef khronos_int32_t  GLfixed;
+
+/* GL types for handling large vertex buffer objects */
+typedef khronos_intptr_t GLintptr;
+typedef khronos_ssize_t  GLsizeiptr;
+
+/* OpenGL ES core versions */
+#define GL_ES_VERSION_2_0                 1
+
+/* ClearBufferMask */
+#define GL_DEPTH_BUFFER_BIT               0x00000100
+#define GL_STENCIL_BUFFER_BIT             0x00000400
+#define GL_COLOR_BUFFER_BIT               0x00004000
+
+/* Boolean */
+#define GL_FALSE                          0
+#define GL_TRUE                           1
+
+/* BeginMode */
+#define GL_POINTS                         0x0000
+#define GL_LINES                          0x0001
+#define GL_LINE_LOOP                      0x0002
+#define GL_LINE_STRIP                     0x0003
+#define GL_TRIANGLES                      0x0004
+#define GL_TRIANGLE_STRIP                 0x0005
+#define GL_TRIANGLE_FAN                   0x0006
+
+/* AlphaFunction (not supported in ES20) */
+/*      GL_NEVER */
+/*      GL_LESS */
+/*      GL_EQUAL */
+/*      GL_LEQUAL */
+/*      GL_GREATER */
+/*      GL_NOTEQUAL */
+/*      GL_GEQUAL */
+/*      GL_ALWAYS */
+
+/* BlendingFactorDest */
+#define GL_ZERO                           0
+#define GL_ONE                            1
+#define GL_SRC_COLOR                      0x0300
+#define GL_ONE_MINUS_SRC_COLOR            0x0301
+#define GL_SRC_ALPHA                      0x0302
+#define GL_ONE_MINUS_SRC_ALPHA            0x0303
+#define GL_DST_ALPHA                      0x0304
+#define GL_ONE_MINUS_DST_ALPHA            0x0305
+
+/* BlendingFactorSrc */
+/*      GL_ZERO */
+/*      GL_ONE */
+#define GL_DST_COLOR                      0x0306
+#define GL_ONE_MINUS_DST_COLOR            0x0307
+#define GL_SRC_ALPHA_SATURATE             0x0308
+/*      GL_SRC_ALPHA */
+/*      GL_ONE_MINUS_SRC_ALPHA */
+/*      GL_DST_ALPHA */
+/*      GL_ONE_MINUS_DST_ALPHA */
+
+/* BlendEquationSeparate */
+#define GL_FUNC_ADD                       0x8006
+#define GL_BLEND_EQUATION                 0x8009
+#define GL_BLEND_EQUATION_RGB             0x8009    /* same as BLEND_EQUATION */
+#define GL_BLEND_EQUATION_ALPHA           0x883D
+
+/* BlendSubtract */
+#define GL_FUNC_SUBTRACT                  0x800A
+#define GL_FUNC_REVERSE_SUBTRACT          0x800B
+
+/* Separate Blend Functions */
+#define GL_BLEND_DST_RGB                  0x80C8
+#define GL_BLEND_SRC_RGB                  0x80C9
+#define GL_BLEND_DST_ALPHA                0x80CA
+#define GL_BLEND_SRC_ALPHA                0x80CB
+#define GL_CONSTANT_COLOR                 0x8001
+#define GL_ONE_MINUS_CONSTANT_COLOR       0x8002
+#define GL_CONSTANT_ALPHA                 0x8003
+#define GL_ONE_MINUS_CONSTANT_ALPHA       0x8004
+#define GL_BLEND_COLOR                    0x8005
+
+/* Buffer Objects */
+#define GL_ARRAY_BUFFER                   0x8892
+#define GL_ELEMENT_ARRAY_BUFFER           0x8893
+#define GL_ARRAY_BUFFER_BINDING           0x8894
+#define GL_ELEMENT_ARRAY_BUFFER_BINDING   0x8895
+
+#define GL_STREAM_DRAW                    0x88E0
+#define GL_STATIC_DRAW                    0x88E4
+#define GL_DYNAMIC_DRAW                   0x88E8
+
+#define GL_BUFFER_SIZE                    0x8764
+#define GL_BUFFER_USAGE                   0x8765
+
+#define GL_CURRENT_VERTEX_ATTRIB          0x8626
+
+/* CullFaceMode */
+#define GL_FRONT                          0x0404
+#define GL_BACK                           0x0405
+#define GL_FRONT_AND_BACK                 0x0408
+
+/* DepthFunction */
+/*      GL_NEVER */
+/*      GL_LESS */
+/*      GL_EQUAL */
+/*      GL_LEQUAL */
+/*      GL_GREATER */
+/*      GL_NOTEQUAL */
+/*      GL_GEQUAL */
+/*      GL_ALWAYS */
+
+/* EnableCap */
+#define GL_TEXTURE_2D                     0x0DE1
+#define GL_CULL_FACE                      0x0B44
+#define GL_BLEND                          0x0BE2
+#define GL_DITHER                         0x0BD0
+#define GL_STENCIL_TEST                   0x0B90
+#define GL_DEPTH_TEST                     0x0B71
+#define GL_SCISSOR_TEST                   0x0C11
+#define GL_POLYGON_OFFSET_FILL            0x8037
+#define GL_SAMPLE_ALPHA_TO_COVERAGE       0x809E
+#define GL_SAMPLE_COVERAGE                0x80A0
+
+/* ErrorCode */
+#define GL_NO_ERROR                       0
+#define GL_INVALID_ENUM                   0x0500
+#define GL_INVALID_VALUE                  0x0501
+#define GL_INVALID_OPERATION              0x0502
+#define GL_OUT_OF_MEMORY                  0x0505
+
+/* FrontFaceDirection */
+#define GL_CW                             0x0900
+#define GL_CCW                            0x0901
+
+/* GetPName */
+#define GL_LINE_WIDTH                     0x0B21
+#define GL_ALIASED_POINT_SIZE_RANGE       0x846D
+#define GL_ALIASED_LINE_WIDTH_RANGE       0x846E
+#define GL_CULL_FACE_MODE                 0x0B45
+#define GL_FRONT_FACE                     0x0B46
+#define GL_DEPTH_RANGE                    0x0B70
+#define GL_DEPTH_WRITEMASK                0x0B72
+#define GL_DEPTH_CLEAR_VALUE              0x0B73
+#define GL_DEPTH_FUNC                     0x0B74
+#define GL_STENCIL_CLEAR_VALUE            0x0B91
+#define GL_STENCIL_FUNC                   0x0B92
+#define GL_STENCIL_FAIL                   0x0B94
+#define GL_STENCIL_PASS_DEPTH_FAIL        0x0B95
+#define GL_STENCIL_PASS_DEPTH_PASS        0x0B96
+#define GL_STENCIL_REF                    0x0B97
+#define GL_STENCIL_VALUE_MASK             0x0B93
+#define GL_STENCIL_WRITEMASK              0x0B98
+#define GL_STENCIL_BACK_FUNC              0x8800
+#define GL_STENCIL_BACK_FAIL              0x8801
+#define GL_STENCIL_BACK_PASS_DEPTH_FAIL   0x8802
+#define GL_STENCIL_BACK_PASS_DEPTH_PASS   0x8803
+#define GL_STENCIL_BACK_REF               0x8CA3
+#define GL_STENCIL_BACK_VALUE_MASK        0x8CA4
+#define GL_STENCIL_BACK_WRITEMASK         0x8CA5
+#define GL_VIEWPORT                       0x0BA2
+#define GL_SCISSOR_BOX                    0x0C10
+/*      GL_SCISSOR_TEST */
+#define GL_COLOR_CLEAR_VALUE              0x0C22
+#define GL_COLOR_WRITEMASK                0x0C23
+#define GL_UNPACK_ALIGNMENT               0x0CF5
+#define GL_PACK_ALIGNMENT                 0x0D05
+#define GL_MAX_TEXTURE_SIZE               0x0D33
+#define GL_MAX_VIEWPORT_DIMS              0x0D3A
+#define GL_SUBPIXEL_BITS                  0x0D50
+#define GL_RED_BITS                       0x0D52
+#define GL_GREEN_BITS                     0x0D53
+#define GL_BLUE_BITS                      0x0D54
+#define GL_ALPHA_BITS                     0x0D55
+#define GL_DEPTH_BITS                     0x0D56
+#define GL_STENCIL_BITS                   0x0D57
+#define GL_POLYGON_OFFSET_UNITS           0x2A00
+/*      GL_POLYGON_OFFSET_FILL */
+#define GL_POLYGON_OFFSET_FACTOR          0x8038
+#define GL_TEXTURE_BINDING_2D             0x8069
+#define GL_SAMPLE_BUFFERS                 0x80A8
+#define GL_SAMPLES                        0x80A9
+#define GL_SAMPLE_COVERAGE_VALUE          0x80AA
+#define GL_SAMPLE_COVERAGE_INVERT         0x80AB
+
+/* GetTextureParameter */
+/*      GL_TEXTURE_MAG_FILTER */
+/*      GL_TEXTURE_MIN_FILTER */
+/*      GL_TEXTURE_WRAP_S */
+/*      GL_TEXTURE_WRAP_T */
+
+#define GL_NUM_COMPRESSED_TEXTURE_FORMATS 0x86A2
+#define GL_COMPRESSED_TEXTURE_FORMATS     0x86A3
+
+/* HintMode */
+#define GL_DONT_CARE                      0x1100
+#define GL_FASTEST                        0x1101
+#define GL_NICEST                         0x1102
+
+/* HintTarget */
+#define GL_GENERATE_MIPMAP_HINT            0x8192
+
+/* DataType */
+#define GL_BYTE                           0x1400
+#define GL_UNSIGNED_BYTE                  0x1401
+#define GL_SHORT                          0x1402
+#define GL_UNSIGNED_SHORT                 0x1403
+#define GL_INT                            0x1404
+#define GL_UNSIGNED_INT                   0x1405
+#define GL_FLOAT                          0x1406
+#define GL_FIXED                          0x140C
+
+/* PixelFormat */
+#define GL_DEPTH_COMPONENT                0x1902
+#define GL_ALPHA                          0x1906
+#define GL_RGB                            0x1907
+#define GL_RGBA                           0x1908
+#define GL_LUMINANCE                      0x1909
+#define GL_LUMINANCE_ALPHA                0x190A
+
+/* PixelType */
+/*      GL_UNSIGNED_BYTE */
+#define GL_UNSIGNED_SHORT_4_4_4_4         0x8033
+#define GL_UNSIGNED_SHORT_5_5_5_1         0x8034
+#define GL_UNSIGNED_SHORT_5_6_5           0x8363
+
+/* Shaders */
+#define GL_FRAGMENT_SHADER                  0x8B30
+#define GL_VERTEX_SHADER                    0x8B31
+#define GL_MAX_VERTEX_ATTRIBS               0x8869
+#define GL_MAX_VERTEX_UNIFORM_VECTORS       0x8DFB
+#define GL_MAX_VARYING_VECTORS              0x8DFC
+#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS 0x8B4D
+#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS   0x8B4C
+#define GL_MAX_TEXTURE_IMAGE_UNITS          0x8872
+#define GL_MAX_FRAGMENT_UNIFORM_VECTORS     0x8DFD
+#define GL_SHADER_TYPE                      0x8B4F
+#define GL_DELETE_STATUS                    0x8B80
+#define GL_LINK_STATUS                      0x8B82
+#define GL_VALIDATE_STATUS                  0x8B83
+#define GL_ATTACHED_SHADERS                 0x8B85
+#define GL_ACTIVE_UNIFORMS                  0x8B86
+#define GL_ACTIVE_UNIFORM_MAX_LENGTH        0x8B87
+#define GL_ACTIVE_ATTRIBUTES                0x8B89
+#define GL_ACTIVE_ATTRIBUTE_MAX_LENGTH      0x8B8A
+#define GL_SHADING_LANGUAGE_VERSION         0x8B8C
+#define GL_CURRENT_PROGRAM                  0x8B8D
+
+/* StencilFunction */
+#define GL_NEVER                          0x0200
+#define GL_LESS                           0x0201
+#define GL_EQUAL                          0x0202
+#define GL_LEQUAL                         0x0203
+#define GL_GREATER                        0x0204
+#define GL_NOTEQUAL                       0x0205
+#define GL_GEQUAL                         0x0206
+#define GL_ALWAYS                         0x0207
+
+/* StencilOp */
+/*      GL_ZERO */
+#define GL_KEEP                           0x1E00
+#define GL_REPLACE                        0x1E01
+#define GL_INCR                           0x1E02
+#define GL_DECR                           0x1E03
+#define GL_INVERT                         0x150A
+#define GL_INCR_WRAP                      0x8507
+#define GL_DECR_WRAP                      0x8508
+
+/* StringName */
+#define GL_VENDOR                         0x1F00
+#define GL_RENDERER                       0x1F01
+#define GL_VERSION                        0x1F02
+#define GL_EXTENSIONS                     0x1F03
+
+/* TextureMagFilter */
+#define GL_NEAREST                        0x2600
+#define GL_LINEAR                         0x2601
+
+/* TextureMinFilter */
+/*      GL_NEAREST */
+/*      GL_LINEAR */
+#define GL_NEAREST_MIPMAP_NEAREST         0x2700
+#define GL_LINEAR_MIPMAP_NEAREST          0x2701
+#define GL_NEAREST_MIPMAP_LINEAR          0x2702
+#define GL_LINEAR_MIPMAP_LINEAR           0x2703
+
+/* TextureParameterName */
+#define GL_TEXTURE_MAG_FILTER             0x2800
+#define GL_TEXTURE_MIN_FILTER             0x2801
+#define GL_TEXTURE_WRAP_S                 0x2802
+#define GL_TEXTURE_WRAP_T                 0x2803
+
+/* TextureTarget */
+/*      GL_TEXTURE_2D */
+#define GL_TEXTURE                        0x1702
+
+#define GL_TEXTURE_CUBE_MAP               0x8513
+#define GL_TEXTURE_BINDING_CUBE_MAP       0x8514
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_X    0x8515
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X    0x8516
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y    0x8517
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y    0x8518
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z    0x8519
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z    0x851A
+#define GL_MAX_CUBE_MAP_TEXTURE_SIZE      0x851C
+
+/* TextureUnit */
+#define GL_TEXTURE0                       0x84C0
+#define GL_TEXTURE1                       0x84C1
+#define GL_TEXTURE2                       0x84C2
+#define GL_TEXTURE3                       0x84C3
+#define GL_TEXTURE4                       0x84C4
+#define GL_TEXTURE5                       0x84C5
+#define GL_TEXTURE6                       0x84C6
+#define GL_TEXTURE7                       0x84C7
+#define GL_TEXTURE8                       0x84C8
+#define GL_TEXTURE9                       0x84C9
+#define GL_TEXTURE10                      0x84CA
+#define GL_TEXTURE11                      0x84CB
+#define GL_TEXTURE12                      0x84CC
+#define GL_TEXTURE13                      0x84CD
+#define GL_TEXTURE14                      0x84CE
+#define GL_TEXTURE15                      0x84CF
+#define GL_TEXTURE16                      0x84D0
+#define GL_TEXTURE17                      0x84D1
+#define GL_TEXTURE18                      0x84D2
+#define GL_TEXTURE19                      0x84D3
+#define GL_TEXTURE20                      0x84D4
+#define GL_TEXTURE21                      0x84D5
+#define GL_TEXTURE22                      0x84D6
+#define GL_TEXTURE23                      0x84D7
+#define GL_TEXTURE24                      0x84D8
+#define GL_TEXTURE25                      0x84D9
+#define GL_TEXTURE26                      0x84DA
+#define GL_TEXTURE27                      0x84DB
+#define GL_TEXTURE28                      0x84DC
+#define GL_TEXTURE29                      0x84DD
+#define GL_TEXTURE30                      0x84DE
+#define GL_TEXTURE31                      0x84DF
+#define GL_ACTIVE_TEXTURE                 0x84E0
+
+/* TextureWrapMode */
+#define GL_REPEAT                         0x2901
+#define GL_CLAMP_TO_EDGE                  0x812F
+#define GL_MIRRORED_REPEAT                0x8370
+
+/* Uniform Types */
+#define GL_FLOAT_VEC2                     0x8B50
+#define GL_FLOAT_VEC3                     0x8B51
+#define GL_FLOAT_VEC4                     0x8B52
+#define GL_INT_VEC2                       0x8B53
+#define GL_INT_VEC3                       0x8B54
+#define GL_INT_VEC4                       0x8B55
+#define GL_BOOL                           0x8B56
+#define GL_BOOL_VEC2                      0x8B57
+#define GL_BOOL_VEC3                      0x8B58
+#define GL_BOOL_VEC4                      0x8B59
+#define GL_FLOAT_MAT2                     0x8B5A
+#define GL_FLOAT_MAT3                     0x8B5B
+#define GL_FLOAT_MAT4                     0x8B5C
+#define GL_SAMPLER_2D                     0x8B5E
+#define GL_SAMPLER_CUBE                   0x8B60
+
+/* Vertex Arrays */
+#define GL_VERTEX_ATTRIB_ARRAY_ENABLED        0x8622
+#define GL_VERTEX_ATTRIB_ARRAY_SIZE           0x8623
+#define GL_VERTEX_ATTRIB_ARRAY_STRIDE         0x8624
+#define GL_VERTEX_ATTRIB_ARRAY_TYPE           0x8625
+#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED     0x886A
+#define GL_VERTEX_ATTRIB_ARRAY_POINTER        0x8645
+#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING 0x889F
+
+/* Read Format */
+#define GL_IMPLEMENTATION_COLOR_READ_TYPE   0x8B9A
+#define GL_IMPLEMENTATION_COLOR_READ_FORMAT 0x8B9B
+
+/* Shader Source */
+#define GL_COMPILE_STATUS                 0x8B81
+#define GL_INFO_LOG_LENGTH                0x8B84
+#define GL_SHADER_SOURCE_LENGTH           0x8B88
+#define GL_SHADER_COMPILER                0x8DFA
+
+/* Shader Binary */
+#define GL_SHADER_BINARY_FORMATS          0x8DF8
+#define GL_NUM_SHADER_BINARY_FORMATS      0x8DF9
+
+/* Shader Precision-Specified Types */
+#define GL_LOW_FLOAT                      0x8DF0
+#define GL_MEDIUM_FLOAT                   0x8DF1
+#define GL_HIGH_FLOAT                     0x8DF2
+#define GL_LOW_INT                        0x8DF3
+#define GL_MEDIUM_INT                     0x8DF4
+#define GL_HIGH_INT                       0x8DF5
+
+/* Framebuffer Object. */
+#define GL_FRAMEBUFFER                    0x8D40
+#define GL_RENDERBUFFER                   0x8D41
+
+#define GL_RGBA4                          0x8056
+#define GL_RGB5_A1                        0x8057
+#define GL_RGB565                         0x8D62
+#define GL_DEPTH_COMPONENT16              0x81A5
+#define GL_STENCIL_INDEX8                 0x8D48
+
+#define GL_RENDERBUFFER_WIDTH             0x8D42
+#define GL_RENDERBUFFER_HEIGHT            0x8D43
+#define GL_RENDERBUFFER_INTERNAL_FORMAT   0x8D44
+#define GL_RENDERBUFFER_RED_SIZE          0x8D50
+#define GL_RENDERBUFFER_GREEN_SIZE        0x8D51
+#define GL_RENDERBUFFER_BLUE_SIZE         0x8D52
+#define GL_RENDERBUFFER_ALPHA_SIZE        0x8D53
+#define GL_RENDERBUFFER_DEPTH_SIZE        0x8D54
+#define GL_RENDERBUFFER_STENCIL_SIZE      0x8D55
+
+#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE           0x8CD0
+#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME           0x8CD1
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL         0x8CD2
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE 0x8CD3
+
+#define GL_COLOR_ATTACHMENT0              0x8CE0
+#define GL_DEPTH_ATTACHMENT               0x8D00
+#define GL_STENCIL_ATTACHMENT             0x8D20
+
+#define GL_NONE                           0
+
+#define GL_FRAMEBUFFER_COMPLETE                      0x8CD5
+#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT         0x8CD6
+#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT 0x8CD7
+#define GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS         0x8CD9
+#define GL_FRAMEBUFFER_UNSUPPORTED                   0x8CDD
+
+#define GL_FRAMEBUFFER_BINDING            0x8CA6
+#define GL_RENDERBUFFER_BINDING           0x8CA7
+#define GL_MAX_RENDERBUFFER_SIZE          0x84E8
+
+#define GL_INVALID_FRAMEBUFFER_OPERATION  0x0506
+
+/*-------------------------------------------------------------------------
+ * GL core functions.
+ *-----------------------------------------------------------------------*/
+
+GL_APICALL void         GL_APIENTRY glActiveTexture (GLenum texture);
+GL_APICALL void         GL_APIENTRY glAttachShader (GLuint program, GLuint shader);
+GL_APICALL void         GL_APIENTRY glBindAttribLocation (GLuint program, GLuint index, const GLchar* name);
+GL_APICALL void         GL_APIENTRY glBindBuffer (GLenum target, GLuint buffer);
+GL_APICALL void         GL_APIENTRY glBindFramebuffer (GLenum target, GLuint framebuffer);
+GL_APICALL void         GL_APIENTRY glBindRenderbuffer (GLenum target, GLuint renderbuffer);
+GL_APICALL void         GL_APIENTRY glBindTexture (GLenum target, GLuint texture);
+GL_APICALL void         GL_APIENTRY glBlendColor (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha);
+GL_APICALL void         GL_APIENTRY glBlendEquation ( GLenum mode );
+GL_APICALL void         GL_APIENTRY glBlendEquationSeparate (GLenum modeRGB, GLenum modeAlpha);
+GL_APICALL void         GL_APIENTRY glBlendFunc (GLenum sfactor, GLenum dfactor);
+GL_APICALL void         GL_APIENTRY glBlendFuncSeparate (GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);
+GL_APICALL void         GL_APIENTRY glBufferData (GLenum target, GLsizeiptr size, const GLvoid* data, GLenum usage);
+GL_APICALL void         GL_APIENTRY glBufferSubData (GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid* data);
+GL_APICALL GLenum       GL_APIENTRY glCheckFramebufferStatus (GLenum target);
+GL_APICALL void         GL_APIENTRY glClear (GLbitfield mask);
+GL_APICALL void         GL_APIENTRY glClearColor (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha);
+GL_APICALL void         GL_APIENTRY glClearDepthf (GLclampf depth);
+GL_APICALL void         GL_APIENTRY glClearStencil (GLint s);
+GL_APICALL void         GL_APIENTRY glColorMask (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha);
+GL_APICALL void         GL_APIENTRY glCompileShader (GLuint shader);
+GL_APICALL void         GL_APIENTRY glCompressedTexImage2D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void         GL_APIENTRY glCompressedTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void         GL_APIENTRY glCopyTexImage2D (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border);
+GL_APICALL void         GL_APIENTRY glCopyTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL GLuint       GL_APIENTRY glCreateProgram (void);
+GL_APICALL GLuint       GL_APIENTRY glCreateShader (GLenum type);
+GL_APICALL void         GL_APIENTRY glCullFace (GLenum mode);
+GL_APICALL void         GL_APIENTRY glDeleteBuffers (GLsizei n, const GLuint* buffers);
+GL_APICALL void         GL_APIENTRY glDeleteFramebuffers (GLsizei n, const GLuint* framebuffers);
+GL_APICALL void         GL_APIENTRY glDeleteProgram (GLuint program);
+GL_APICALL void         GL_APIENTRY glDeleteRenderbuffers (GLsizei n, const GLuint* renderbuffers);
+GL_APICALL void         GL_APIENTRY glDeleteShader (GLuint shader);
+GL_APICALL void         GL_APIENTRY glDeleteTextures (GLsizei n, const GLuint* textures);
+GL_APICALL void         GL_APIENTRY glDepthFunc (GLenum func);
+GL_APICALL void         GL_APIENTRY glDepthMask (GLboolean flag);
+GL_APICALL void         GL_APIENTRY glDepthRangef (GLclampf zNear, GLclampf zFar);
+GL_APICALL void         GL_APIENTRY glDetachShader (GLuint program, GLuint shader);
+GL_APICALL void         GL_APIENTRY glDisable (GLenum cap);
+GL_APICALL void         GL_APIENTRY glDisableVertexAttribArray (GLuint index);
+GL_APICALL void         GL_APIENTRY glDrawArrays (GLenum mode, GLint first, GLsizei count);
+GL_APICALL void         GL_APIENTRY glDrawElements (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices);
+GL_APICALL void         GL_APIENTRY glEnable (GLenum cap);
+GL_APICALL void         GL_APIENTRY glEnableVertexAttribArray (GLuint index);
+GL_APICALL void         GL_APIENTRY glFinish (void);
+GL_APICALL void         GL_APIENTRY glFlush (void);
+GL_APICALL void         GL_APIENTRY glFramebufferRenderbuffer (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer);
+GL_APICALL void         GL_APIENTRY glFramebufferTexture2D (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level);
+GL_APICALL void         GL_APIENTRY glFrontFace (GLenum mode);
+GL_APICALL void         GL_APIENTRY glGenBuffers (GLsizei n, GLuint* buffers);
+GL_APICALL void         GL_APIENTRY glGenerateMipmap (GLenum target);
+GL_APICALL void         GL_APIENTRY glGenFramebuffers (GLsizei n, GLuint* framebuffers);
+GL_APICALL void         GL_APIENTRY glGenRenderbuffers (GLsizei n, GLuint* renderbuffers);
+GL_APICALL void         GL_APIENTRY glGenTextures (GLsizei n, GLuint* textures);
+GL_APICALL void         GL_APIENTRY glGetActiveAttrib (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);
+GL_APICALL void         GL_APIENTRY glGetActiveUniform (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);
+GL_APICALL void         GL_APIENTRY glGetAttachedShaders (GLuint program, GLsizei maxcount, GLsizei* count, GLuint* shaders);
+GL_APICALL GLint        GL_APIENTRY glGetAttribLocation (GLuint program, const GLchar* name);
+GL_APICALL void         GL_APIENTRY glGetBooleanv (GLenum pname, GLboolean* params);
+GL_APICALL void         GL_APIENTRY glGetBufferParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL GLenum       GL_APIENTRY glGetError (void);
+GL_APICALL void         GL_APIENTRY glGetFloatv (GLenum pname, GLfloat* params);
+GL_APICALL void         GL_APIENTRY glGetFramebufferAttachmentParameteriv (GLenum target, GLenum attachment, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetIntegerv (GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetProgramiv (GLuint program, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetProgramInfoLog (GLuint program, GLsizei bufsize, GLsizei* length, GLchar* infolog);
+GL_APICALL void         GL_APIENTRY glGetRenderbufferParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetShaderiv (GLuint shader, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetShaderInfoLog (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* infolog);
+GL_APICALL void         GL_APIENTRY glGetShaderPrecisionFormat (GLenum shadertype, GLenum precisiontype, GLint* range, GLint* precision);
+GL_APICALL void         GL_APIENTRY glGetShaderSource (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* source);
+GL_APICALL const GLubyte* GL_APIENTRY glGetString (GLenum name);
+GL_APICALL void         GL_APIENTRY glGetTexParameterfv (GLenum target, GLenum pname, GLfloat* params);
+GL_APICALL void         GL_APIENTRY glGetTexParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetUniformfv (GLuint program, GLint location, GLfloat* params);
+GL_APICALL void         GL_APIENTRY glGetUniformiv (GLuint program, GLint location, GLint* params);
+GL_APICALL GLint        GL_APIENTRY glGetUniformLocation (GLuint program, const GLchar* name);
+GL_APICALL void         GL_APIENTRY glGetVertexAttribfv (GLuint index, GLenum pname, GLfloat* params);
+GL_APICALL void         GL_APIENTRY glGetVertexAttribiv (GLuint index, GLenum pname, GLint* params);
+GL_APICALL void         GL_APIENTRY glGetVertexAttribPointerv (GLuint index, GLenum pname, GLvoid** pointer);
+GL_APICALL void         GL_APIENTRY glHint (GLenum target, GLenum mode);
+GL_APICALL GLboolean    GL_APIENTRY glIsBuffer (GLuint buffer);
+GL_APICALL GLboolean    GL_APIENTRY glIsEnabled (GLenum cap);
+GL_APICALL GLboolean    GL_APIENTRY glIsFramebuffer (GLuint framebuffer);
+GL_APICALL GLboolean    GL_APIENTRY glIsProgram (GLuint program);
+GL_APICALL GLboolean    GL_APIENTRY glIsRenderbuffer (GLuint renderbuffer);
+GL_APICALL GLboolean    GL_APIENTRY glIsShader (GLuint shader);
+GL_APICALL GLboolean    GL_APIENTRY glIsTexture (GLuint texture);
+GL_APICALL void         GL_APIENTRY glLineWidth (GLfloat width);
+GL_APICALL void         GL_APIENTRY glLinkProgram (GLuint program);
+GL_APICALL void         GL_APIENTRY glPixelStorei (GLenum pname, GLint param);
+GL_APICALL void         GL_APIENTRY glPolygonOffset (GLfloat factor, GLfloat units);
+GL_APICALL void         GL_APIENTRY glReadPixels (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid* pixels);
+GL_APICALL void         GL_APIENTRY glReleaseShaderCompiler (void);
+GL_APICALL void         GL_APIENTRY glRenderbufferStorage (GLenum target, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void         GL_APIENTRY glSampleCoverage (GLclampf value, GLboolean invert);
+GL_APICALL void         GL_APIENTRY glScissor (GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL void         GL_APIENTRY glShaderBinary (GLsizei n, const GLuint* shaders, GLenum binaryformat, const GLvoid* binary, GLsizei length);
+GL_APICALL void         GL_APIENTRY glShaderSource (GLuint shader, GLsizei count, const GLchar* const* string, const GLint* length);
+GL_APICALL void         GL_APIENTRY glStencilFunc (GLenum func, GLint ref, GLuint mask);
+GL_APICALL void         GL_APIENTRY glStencilFuncSeparate (GLenum face, GLenum func, GLint ref, GLuint mask);
+GL_APICALL void         GL_APIENTRY glStencilMask (GLuint mask);
+GL_APICALL void         GL_APIENTRY glStencilMaskSeparate (GLenum face, GLuint mask);
+GL_APICALL void         GL_APIENTRY glStencilOp (GLenum fail, GLenum zfail, GLenum zpass);
+GL_APICALL void         GL_APIENTRY glStencilOpSeparate (GLenum face, GLenum fail, GLenum zfail, GLenum zpass);
+GL_APICALL void         GL_APIENTRY glTexImage2D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void         GL_APIENTRY glTexParameterf (GLenum target, GLenum pname, GLfloat param);
+GL_APICALL void         GL_APIENTRY glTexParameterfv (GLenum target, GLenum pname, const GLfloat* params);
+GL_APICALL void         GL_APIENTRY glTexParameteri (GLenum target, GLenum pname, GLint param);
+GL_APICALL void         GL_APIENTRY glTexParameteriv (GLenum target, GLenum pname, const GLint* params);
+GL_APICALL void         GL_APIENTRY glTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void         GL_APIENTRY glUniform1f (GLint location, GLfloat x);
+GL_APICALL void         GL_APIENTRY glUniform1fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void         GL_APIENTRY glUniform1i (GLint location, GLint x);
+GL_APICALL void         GL_APIENTRY glUniform1iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void         GL_APIENTRY glUniform2f (GLint location, GLfloat x, GLfloat y);
+GL_APICALL void         GL_APIENTRY glUniform2fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void         GL_APIENTRY glUniform2i (GLint location, GLint x, GLint y);
+GL_APICALL void         GL_APIENTRY glUniform2iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void         GL_APIENTRY glUniform3f (GLint location, GLfloat x, GLfloat y, GLfloat z);
+GL_APICALL void         GL_APIENTRY glUniform3fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void         GL_APIENTRY glUniform3i (GLint location, GLint x, GLint y, GLint z);
+GL_APICALL void         GL_APIENTRY glUniform3iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void         GL_APIENTRY glUniform4f (GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+GL_APICALL void         GL_APIENTRY glUniform4fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void         GL_APIENTRY glUniform4i (GLint location, GLint x, GLint y, GLint z, GLint w);
+GL_APICALL void         GL_APIENTRY glUniform4iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void         GL_APIENTRY glUniformMatrix2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void         GL_APIENTRY glUniformMatrix3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void         GL_APIENTRY glUniformMatrix4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void         GL_APIENTRY glUseProgram (GLuint program);
+GL_APICALL void         GL_APIENTRY glValidateProgram (GLuint program);
+GL_APICALL void         GL_APIENTRY glVertexAttrib1f (GLuint indx, GLfloat x);
+GL_APICALL void         GL_APIENTRY glVertexAttrib1fv (GLuint indx, const GLfloat* values);
+GL_APICALL void         GL_APIENTRY glVertexAttrib2f (GLuint indx, GLfloat x, GLfloat y);
+GL_APICALL void         GL_APIENTRY glVertexAttrib2fv (GLuint indx, const GLfloat* values);
+GL_APICALL void         GL_APIENTRY glVertexAttrib3f (GLuint indx, GLfloat x, GLfloat y, GLfloat z);
+GL_APICALL void         GL_APIENTRY glVertexAttrib3fv (GLuint indx, const GLfloat* values);
+GL_APICALL void         GL_APIENTRY glVertexAttrib4f (GLuint indx, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+GL_APICALL void         GL_APIENTRY glVertexAttrib4fv (GLuint indx, const GLfloat* values);
+GL_APICALL void         GL_APIENTRY glVertexAttribPointer (GLuint indx, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid* ptr);
+GL_APICALL void         GL_APIENTRY glViewport (GLint x, GLint y, GLsizei width, GLsizei height);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __gl2_h_ */
diff --git a/platform/winrt/include/GLES2/gl2ext.h b/platform/winrt/include/GLES2/gl2ext.h
index cc6997d4b2..d77fdbaebc 100644
--- a/platform/winrt/include/GLES2/gl2ext.h
+++ b/platform/winrt/include/GLES2/gl2ext.h
@@ -1,2013 +1,2013 @@
-#ifndef __gl2ext_h_

-#define __gl2ext_h_

-

-/* $Revision: 20795 $ on $Date:: 2013-03-07 01:01:58 -0800 #$ */

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

- * This document is licensed under the SGI Free Software B License Version

- * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .

- */

-

-#ifndef GL_APIENTRYP

-#   define GL_APIENTRYP GL_APIENTRY*

-#endif

-

-/*------------------------------------------------------------------------*

- * OES extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_OES_compressed_ETC1_RGB8_texture */

-#ifndef GL_OES_compressed_ETC1_RGB8_texture

-#define GL_ETC1_RGB8_OES                                        0x8D64

-#endif

-

-/* GL_OES_compressed_paletted_texture */

-#ifndef GL_OES_compressed_paletted_texture

-#define GL_PALETTE4_RGB8_OES                                    0x8B90

-#define GL_PALETTE4_RGBA8_OES                                   0x8B91

-#define GL_PALETTE4_R5_G6_B5_OES                                0x8B92

-#define GL_PALETTE4_RGBA4_OES                                   0x8B93

-#define GL_PALETTE4_RGB5_A1_OES                                 0x8B94

-#define GL_PALETTE8_RGB8_OES                                    0x8B95

-#define GL_PALETTE8_RGBA8_OES                                   0x8B96

-#define GL_PALETTE8_R5_G6_B5_OES                                0x8B97

-#define GL_PALETTE8_RGBA4_OES                                   0x8B98

-#define GL_PALETTE8_RGB5_A1_OES                                 0x8B99

-#endif

-

-/* GL_OES_depth24 */

-#ifndef GL_OES_depth24

-#define GL_DEPTH_COMPONENT24_OES                                0x81A6

-#endif

-

-/* GL_OES_depth32 */

-#ifndef GL_OES_depth32

-#define GL_DEPTH_COMPONENT32_OES                                0x81A7

-#endif

-

-/* GL_OES_depth_texture */

-/* No new tokens introduced by this extension. */

-

-/* GL_OES_EGL_image */

-#ifndef GL_OES_EGL_image

-typedef void* GLeglImageOES;

-#endif

-

-/* GL_OES_EGL_image_external */

-#ifndef GL_OES_EGL_image_external

-/* GLeglImageOES defined in GL_OES_EGL_image already. */

-#define GL_TEXTURE_EXTERNAL_OES                                 0x8D65

-#define GL_SAMPLER_EXTERNAL_OES                                 0x8D66

-#define GL_TEXTURE_BINDING_EXTERNAL_OES                         0x8D67

-#define GL_REQUIRED_TEXTURE_IMAGE_UNITS_OES                     0x8D68

-#endif

-

-/* GL_OES_element_index_uint */

-#ifndef GL_OES_element_index_uint

-#define GL_UNSIGNED_INT                                         0x1405

-#endif

-

-/* GL_OES_get_program_binary */

-#ifndef GL_OES_get_program_binary

-#define GL_PROGRAM_BINARY_LENGTH_OES                            0x8741

-#define GL_NUM_PROGRAM_BINARY_FORMATS_OES                       0x87FE

-#define GL_PROGRAM_BINARY_FORMATS_OES                           0x87FF

-#endif

-

-/* GL_OES_mapbuffer */

-#ifndef GL_OES_mapbuffer

-#define GL_WRITE_ONLY_OES                                       0x88B9

-#define GL_BUFFER_ACCESS_OES                                    0x88BB

-#define GL_BUFFER_MAPPED_OES                                    0x88BC

-#define GL_BUFFER_MAP_POINTER_OES                               0x88BD

-#endif

-

-/* GL_OES_packed_depth_stencil */

-#ifndef GL_OES_packed_depth_stencil

-#define GL_DEPTH_STENCIL_OES                                    0x84F9

-#define GL_UNSIGNED_INT_24_8_OES                                0x84FA

-#define GL_DEPTH24_STENCIL8_OES                                 0x88F0

-#endif

-

-/* GL_OES_required_internalformat */

-#ifndef GL_OES_required_internalformat 

-#define GL_ALPHA8_OES                                           0x803C

-#define GL_DEPTH_COMPONENT16_OES                                0x81A5

-/* reuse GL_DEPTH_COMPONENT24_OES */                            

-/* reuse GL_DEPTH24_STENCIL8_OES */                             

-/* reuse GL_DEPTH_COMPONENT32_OES */                            

-#define GL_LUMINANCE4_ALPHA4_OES                                0x8043

-#define GL_LUMINANCE8_ALPHA8_OES                                0x8045

-#define GL_LUMINANCE8_OES                                       0x8040

-#define GL_RGBA4_OES                                            0x8056

-#define GL_RGB5_A1_OES                                          0x8057

-#define GL_RGB565_OES                                           0x8D62

-/* reuse GL_RGB8_OES */                              

-/* reuse GL_RGBA8_OES */  

-/* reuse GL_RGB10_EXT */

-/* reuse GL_RGB10_A2_EXT */

-#endif 

-

-/* GL_OES_rgb8_rgba8 */

-#ifndef GL_OES_rgb8_rgba8

-#define GL_RGB8_OES                                             0x8051

-#define GL_RGBA8_OES                                            0x8058

-#endif

-

-/* GL_OES_standard_derivatives */

-#ifndef GL_OES_standard_derivatives

-#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT_OES                  0x8B8B

-#endif

-

-/* GL_OES_stencil1 */

-#ifndef GL_OES_stencil1

-#define GL_STENCIL_INDEX1_OES                                   0x8D46

-#endif

-

-/* GL_OES_stencil4 */

-#ifndef GL_OES_stencil4

-#define GL_STENCIL_INDEX4_OES                                   0x8D47

-#endif

-

-#ifndef GL_OES_surfaceless_context

-#define GL_FRAMEBUFFER_UNDEFINED_OES                            0x8219

-#endif

-

-/* GL_OES_texture_3D */

-#ifndef GL_OES_texture_3D

-#define GL_TEXTURE_WRAP_R_OES                                   0x8072

-#define GL_TEXTURE_3D_OES                                       0x806F

-#define GL_TEXTURE_BINDING_3D_OES                               0x806A

-#define GL_MAX_3D_TEXTURE_SIZE_OES                              0x8073

-#define GL_SAMPLER_3D_OES                                       0x8B5F

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_OES        0x8CD4

-#endif

-

-/* GL_OES_texture_float */

-/* No new tokens introduced by this extension. */

-

-/* GL_OES_texture_float_linear */

-/* No new tokens introduced by this extension. */

-

-/* GL_OES_texture_half_float */

-#ifndef GL_OES_texture_half_float

-#define GL_HALF_FLOAT_OES                                       0x8D61

-#endif

-

-/* GL_OES_texture_half_float_linear */

-/* No new tokens introduced by this extension. */

-

-/* GL_OES_texture_npot */

-/* No new tokens introduced by this extension. */

-

-/* GL_OES_vertex_array_object */

-#ifndef GL_OES_vertex_array_object

-#define GL_VERTEX_ARRAY_BINDING_OES                             0x85B5

-#endif

-

-/* GL_OES_vertex_half_float */

-/* GL_HALF_FLOAT_OES defined in GL_OES_texture_half_float already. */

-

-/* GL_OES_vertex_type_10_10_10_2 */

-#ifndef GL_OES_vertex_type_10_10_10_2

-#define GL_UNSIGNED_INT_10_10_10_2_OES                          0x8DF6

-#define GL_INT_10_10_10_2_OES                                   0x8DF7

-#endif

-

-/*------------------------------------------------------------------------*

- * KHR extension tokens

- *------------------------------------------------------------------------*/

-

-#ifndef GL_KHR_debug

-typedef void (GL_APIENTRYP GLDEBUGPROC)(GLenum source,GLenum type,GLuint id,GLenum severity,GLsizei length,const GLchar *message,GLvoid *userParam);

-#define GL_DEBUG_OUTPUT_SYNCHRONOUS                             0x8242

-#define GL_DEBUG_NEXT_LOGGED_MESSAGE_LENGTH                     0x8243

-#define GL_DEBUG_CALLBACK_FUNCTION                              0x8244

-#define GL_DEBUG_CALLBACK_USER_PARAM                            0x8245

-#define GL_DEBUG_SOURCE_API                                     0x8246

-#define GL_DEBUG_SOURCE_WINDOW_SYSTEM                           0x8247

-#define GL_DEBUG_SOURCE_SHADER_COMPILER                         0x8248

-#define GL_DEBUG_SOURCE_THIRD_PARTY                             0x8249

-#define GL_DEBUG_SOURCE_APPLICATION                             0x824A

-#define GL_DEBUG_SOURCE_OTHER                                   0x824B

-#define GL_DEBUG_TYPE_ERROR                                     0x824C

-#define GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR                       0x824D

-#define GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR                        0x824E

-#define GL_DEBUG_TYPE_PORTABILITY                               0x824F

-#define GL_DEBUG_TYPE_PERFORMANCE                               0x8250

-#define GL_DEBUG_TYPE_OTHER                                     0x8251

-#define GL_DEBUG_TYPE_MARKER                                    0x8268

-#define GL_DEBUG_TYPE_PUSH_GROUP                                0x8269

-#define GL_DEBUG_TYPE_POP_GROUP                                 0x826A

-#define GL_DEBUG_SEVERITY_NOTIFICATION                          0x826B

-#define GL_MAX_DEBUG_GROUP_STACK_DEPTH                          0x826C

-#define GL_DEBUG_GROUP_STACK_DEPTH                              0x826D

-#define GL_BUFFER                                               0x82E0

-#define GL_SHADER                                               0x82E1

-#define GL_PROGRAM                                              0x82E2

-#define GL_QUERY                                                0x82E3

-/* PROGRAM_PIPELINE only in GL */                               

-#define GL_SAMPLER                                              0x82E6

-/* DISPLAY_LIST only in GL */                                   

-#define GL_MAX_LABEL_LENGTH                                     0x82E8

-#define GL_MAX_DEBUG_MESSAGE_LENGTH                             0x9143

-#define GL_MAX_DEBUG_LOGGED_MESSAGES                            0x9144

-#define GL_DEBUG_LOGGED_MESSAGES                                0x9145

-#define GL_DEBUG_SEVERITY_HIGH                                  0x9146

-#define GL_DEBUG_SEVERITY_MEDIUM                                0x9147

-#define GL_DEBUG_SEVERITY_LOW                                   0x9148

-#define GL_DEBUG_OUTPUT                                         0x92E0

-#define GL_CONTEXT_FLAG_DEBUG_BIT                               0x00000002

-#define GL_STACK_OVERFLOW                                       0x0503

-#define GL_STACK_UNDERFLOW                                      0x0504

-#endif

-

-#ifndef GL_KHR_texture_compression_astc_ldr

-#define GL_COMPRESSED_RGBA_ASTC_4x4_KHR                         0x93B0

-#define GL_COMPRESSED_RGBA_ASTC_5x4_KHR                         0x93B1

-#define GL_COMPRESSED_RGBA_ASTC_5x5_KHR                         0x93B2

-#define GL_COMPRESSED_RGBA_ASTC_6x5_KHR                         0x93B3

-#define GL_COMPRESSED_RGBA_ASTC_6x6_KHR                         0x93B4

-#define GL_COMPRESSED_RGBA_ASTC_8x5_KHR                         0x93B5

-#define GL_COMPRESSED_RGBA_ASTC_8x6_KHR                         0x93B6

-#define GL_COMPRESSED_RGBA_ASTC_8x8_KHR                         0x93B7

-#define GL_COMPRESSED_RGBA_ASTC_10x5_KHR                        0x93B8

-#define GL_COMPRESSED_RGBA_ASTC_10x6_KHR                        0x93B9

-#define GL_COMPRESSED_RGBA_ASTC_10x8_KHR                        0x93BA

-#define GL_COMPRESSED_RGBA_ASTC_10x10_KHR                       0x93BB

-#define GL_COMPRESSED_RGBA_ASTC_12x10_KHR                       0x93BC

-#define GL_COMPRESSED_RGBA_ASTC_12x12_KHR                       0x93BD

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR                 0x93D0

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR                 0x93D1

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR                 0x93D2

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR                 0x93D3

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR                 0x93D4

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR                 0x93D5

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR                 0x93D6

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR                 0x93D7

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x5_KHR                0x93D8

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x6_KHR                0x93D9

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR                0x93DA

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR               0x93DB

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x10_KHR               0x93DC

-#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR               0x93DD

-#endif

-

-/*------------------------------------------------------------------------*

- * AMD extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_AMD_compressed_3DC_texture */

-#ifndef GL_AMD_compressed_3DC_texture

-#define GL_3DC_X_AMD                                            0x87F9

-#define GL_3DC_XY_AMD                                           0x87FA

-#endif

-

-/* GL_AMD_compressed_ATC_texture */

-#ifndef GL_AMD_compressed_ATC_texture

-#define GL_ATC_RGB_AMD                                          0x8C92

-#define GL_ATC_RGBA_EXPLICIT_ALPHA_AMD                          0x8C93

-#define GL_ATC_RGBA_INTERPOLATED_ALPHA_AMD                      0x87EE

-#endif

-

-/* GL_AMD_performance_monitor */

-#ifndef GL_AMD_performance_monitor

-#define GL_COUNTER_TYPE_AMD                                     0x8BC0

-#define GL_COUNTER_RANGE_AMD                                    0x8BC1

-#define GL_UNSIGNED_INT64_AMD                                   0x8BC2

-#define GL_PERCENTAGE_AMD                                       0x8BC3

-#define GL_PERFMON_RESULT_AVAILABLE_AMD                         0x8BC4

-#define GL_PERFMON_RESULT_SIZE_AMD                              0x8BC5

-#define GL_PERFMON_RESULT_AMD                                   0x8BC6

-#endif

-

-/* GL_AMD_program_binary_Z400 */

-#ifndef GL_AMD_program_binary_Z400

-#define GL_Z400_BINARY_AMD                                      0x8740

-#endif

-

-/*------------------------------------------------------------------------*

- * ANGLE extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_ANGLE_depth_texture */

-#ifndef GL_ANGLE_depth_texture

-#define GL_DEPTH_COMPONENT                                      0x1902

-#define GL_DEPTH_STENCIL_OES                                    0x84F9

-#define GL_UNSIGNED_SHORT                                       0x1403

-#define GL_UNSIGNED_INT                                         0x1405

-#define GL_UNSIGNED_INT_24_8_OES                                0x84FA

-#define GL_DEPTH_COMPONENT16                                    0x81A5

-#define GL_DEPTH_COMPONENT32_OES                                0x81A7

-#define GL_DEPTH24_STENCIL8_OES                                 0x88F0

-#endif

-

-/* GL_ANGLE_framebuffer_blit */

-#ifndef GL_ANGLE_framebuffer_blit

-#define GL_READ_FRAMEBUFFER_ANGLE                               0x8CA8

-#define GL_DRAW_FRAMEBUFFER_ANGLE                               0x8CA9

-#define GL_DRAW_FRAMEBUFFER_BINDING_ANGLE                       0x8CA6

-#define GL_READ_FRAMEBUFFER_BINDING_ANGLE                       0x8CAA

-#endif

-

-/* GL_ANGLE_framebuffer_multisample */

-#ifndef GL_ANGLE_framebuffer_multisample

-#define GL_RENDERBUFFER_SAMPLES_ANGLE                           0x8CAB

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_ANGLE             0x8D56

-#define GL_MAX_SAMPLES_ANGLE                                    0x8D57

-#endif

-

-/* GL_ANGLE_instanced_arrays */

-#ifndef GL_ANGLE_instanced_arrays 

-#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR_ANGLE                    0x88FE

-#endif

-

-/* GL_ANGLE_pack_reverse_row_order */

-#ifndef GL_ANGLE_pack_reverse_row_order 

-#define GL_PACK_REVERSE_ROW_ORDER_ANGLE                         0x93A4

-#endif

-

-/* GL_ANGLE_program_binary */

-#ifndef GL_ANGLE_program_binary

-#define GL_PROGRAM_BINARY_ANGLE                                 0x93A6

-#endif

-

-/* GL_ANGLE_texture_compression_dxt3 */

-#ifndef GL_ANGLE_texture_compression_dxt3 

-#define GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE                      0x83F2

-#endif

-

-/* GL_ANGLE_texture_compression_dxt5 */

-#ifndef GL_ANGLE_texture_compression_dxt5 

-#define GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE                      0x83F3

-#endif

-

-/* GL_ANGLE_texture_usage */

-#ifndef GL_ANGLE_texture_usage 

-#define GL_TEXTURE_USAGE_ANGLE                                  0x93A2

-#define GL_FRAMEBUFFER_ATTACHMENT_ANGLE                         0x93A3

-#endif

-

-/* GL_ANGLE_translated_shader_source */

-#ifndef GL_ANGLE_translated_shader_source 

-#define GL_TRANSLATED_SHADER_SOURCE_LENGTH_ANGLE                0x93A0

-#endif

-

-/*------------------------------------------------------------------------*

- * APPLE extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_APPLE_copy_texture_levels */

-/* No new tokens introduced by this extension. */

-    

-/* GL_APPLE_framebuffer_multisample */

-#ifndef GL_APPLE_framebuffer_multisample

-#define GL_RENDERBUFFER_SAMPLES_APPLE                           0x8CAB

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_APPLE             0x8D56

-#define GL_MAX_SAMPLES_APPLE                                    0x8D57

-#define GL_READ_FRAMEBUFFER_APPLE                               0x8CA8

-#define GL_DRAW_FRAMEBUFFER_APPLE                               0x8CA9

-#define GL_DRAW_FRAMEBUFFER_BINDING_APPLE                       0x8CA6

-#define GL_READ_FRAMEBUFFER_BINDING_APPLE                       0x8CAA

-#endif

-

-/* GL_APPLE_rgb_422 */

-#ifndef GL_APPLE_rgb_422

-#define GL_RGB_422_APPLE                                        0x8A1F

-#define GL_UNSIGNED_SHORT_8_8_APPLE                             0x85BA

-#define GL_UNSIGNED_SHORT_8_8_REV_APPLE                         0x85BB

-#endif

-

-/* GL_APPLE_sync */

-#ifndef GL_APPLE_sync

-

-#ifndef __gl3_h_

-/* These types are defined with reference to <inttypes.h>

- * in the Apple extension spec, but here we use the Khronos

- * portable types in khrplatform.h, and assume those types 

- * are always defined.

- * If any other extensions using these types are defined, 

- * the typedefs must move out of this block and be shared.

- */

-typedef khronos_int64_t GLint64;

-typedef khronos_uint64_t GLuint64;

-typedef struct __GLsync *GLsync;

-#endif

-

-#define GL_SYNC_OBJECT_APPLE                                    0x8A53

-#define GL_MAX_SERVER_WAIT_TIMEOUT_APPLE                        0x9111

-#define GL_OBJECT_TYPE_APPLE                                    0x9112

-#define GL_SYNC_CONDITION_APPLE                                 0x9113

-#define GL_SYNC_STATUS_APPLE                                    0x9114

-#define GL_SYNC_FLAGS_APPLE                                     0x9115

-#define GL_SYNC_FENCE_APPLE                                     0x9116

-#define GL_SYNC_GPU_COMMANDS_COMPLETE_APPLE                     0x9117

-#define GL_UNSIGNALED_APPLE                                     0x9118

-#define GL_SIGNALED_APPLE                                       0x9119

-#define GL_ALREADY_SIGNALED_APPLE                               0x911A

-#define GL_TIMEOUT_EXPIRED_APPLE                                0x911B

-#define GL_CONDITION_SATISFIED_APPLE                            0x911C

-#define GL_WAIT_FAILED_APPLE                                    0x911D

-#define GL_SYNC_FLUSH_COMMANDS_BIT_APPLE                        0x00000001

-#define GL_TIMEOUT_IGNORED_APPLE                                0xFFFFFFFFFFFFFFFFull

-#endif

-

-/* GL_APPLE_texture_format_BGRA8888 */

-#ifndef GL_APPLE_texture_format_BGRA8888

-#define GL_BGRA_EXT                                             0x80E1

-#endif

-

-/* GL_APPLE_texture_max_level */

-#ifndef GL_APPLE_texture_max_level

-#define GL_TEXTURE_MAX_LEVEL_APPLE                              0x813D

-#endif

-

-/*------------------------------------------------------------------------*

- * ARM extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_ARM_mali_program_binary */

-#ifndef GL_ARM_mali_program_binary

-#define GL_MALI_PROGRAM_BINARY_ARM                              0x8F61

-#endif

-

-/* GL_ARM_mali_shader_binary */

-#ifndef GL_ARM_mali_shader_binary

-#define GL_MALI_SHADER_BINARY_ARM                               0x8F60

-#endif

-

-/* GL_ARM_rgba8 */

-/* No new tokens introduced by this extension. */

-

-/*------------------------------------------------------------------------*

- * EXT extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_EXT_blend_minmax */

-#ifndef GL_EXT_blend_minmax

-#define GL_MIN_EXT                                              0x8007

-#define GL_MAX_EXT                                              0x8008

-#endif

-

-/* GL_EXT_color_buffer_half_float */

-#ifndef GL_EXT_color_buffer_half_float

-#define GL_RGBA16F_EXT                                          0x881A

-#define GL_RGB16F_EXT                                           0x881B

-#define GL_RG16F_EXT                                            0x822F

-#define GL_R16F_EXT                                             0x822D

-#define GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE_EXT            0x8211

-#define GL_UNSIGNED_NORMALIZED_EXT                              0x8C17

-#endif

-

-/* GL_EXT_debug_label */

-#ifndef GL_EXT_debug_label

-#define GL_PROGRAM_PIPELINE_OBJECT_EXT                          0x8A4F

-#define GL_PROGRAM_OBJECT_EXT                                   0x8B40

-#define GL_SHADER_OBJECT_EXT                                    0x8B48

-#define GL_BUFFER_OBJECT_EXT                                    0x9151

-#define GL_QUERY_OBJECT_EXT                                     0x9153

-#define GL_VERTEX_ARRAY_OBJECT_EXT                              0x9154

-#endif

-

-/* GL_EXT_debug_marker */

-/* No new tokens introduced by this extension. */

-

-/* GL_EXT_discard_framebuffer */

-#ifndef GL_EXT_discard_framebuffer

-#define GL_COLOR_EXT                                            0x1800

-#define GL_DEPTH_EXT                                            0x1801

-#define GL_STENCIL_EXT                                          0x1802

-#endif

-

-/* GL_EXT_map_buffer_range */

-#ifndef GL_EXT_map_buffer_range

-#define GL_MAP_READ_BIT_EXT                                     0x0001

-#define GL_MAP_WRITE_BIT_EXT                                    0x0002

-#define GL_MAP_INVALIDATE_RANGE_BIT_EXT                         0x0004

-#define GL_MAP_INVALIDATE_BUFFER_BIT_EXT                        0x0008

-#define GL_MAP_FLUSH_EXPLICIT_BIT_EXT                           0x0010

-#define GL_MAP_UNSYNCHRONIZED_BIT_EXT                           0x0020

-#endif

-

-/* GL_EXT_multisampled_render_to_texture */

-#ifndef GL_EXT_multisampled_render_to_texture

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_SAMPLES_EXT           0x8D6C

-/* reuse values from GL_EXT_framebuffer_multisample (desktop extension) */ 

-#define GL_RENDERBUFFER_SAMPLES_EXT                             0x8CAB

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT               0x8D56

-#define GL_MAX_SAMPLES_EXT                                      0x8D57

-#endif

-

-/* GL_EXT_multiview_draw_buffers */

-#ifndef GL_EXT_multiview_draw_buffers

-#define GL_COLOR_ATTACHMENT_EXT                                 0x90F0

-#define GL_MULTIVIEW_EXT                                        0x90F1

-#define GL_DRAW_BUFFER_EXT                                      0x0C01

-#define GL_READ_BUFFER_EXT                                      0x0C02

-#define GL_MAX_MULTIVIEW_BUFFERS_EXT                            0x90F2

-#endif

-

-/* GL_EXT_multi_draw_arrays */

-/* No new tokens introduced by this extension. */

-

-/* GL_EXT_occlusion_query_boolean */

-#ifndef GL_EXT_occlusion_query_boolean

-#define GL_ANY_SAMPLES_PASSED_EXT                               0x8C2F

-#define GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT                  0x8D6A

-#define GL_CURRENT_QUERY_EXT                                    0x8865

-#define GL_QUERY_RESULT_EXT                                     0x8866

-#define GL_QUERY_RESULT_AVAILABLE_EXT                           0x8867

-#endif

-

-/* GL_EXT_read_format_bgra */

-#ifndef GL_EXT_read_format_bgra

-#define GL_BGRA_EXT                                             0x80E1

-#define GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT                       0x8365

-#define GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT                       0x8366

-#endif

-

-/* GL_EXT_robustness */

-#ifndef GL_EXT_robustness

-/* reuse GL_NO_ERROR */

-#define GL_GUILTY_CONTEXT_RESET_EXT                             0x8253

-#define GL_INNOCENT_CONTEXT_RESET_EXT                           0x8254

-#define GL_UNKNOWN_CONTEXT_RESET_EXT                            0x8255

-#define GL_CONTEXT_ROBUST_ACCESS_EXT                            0x90F3

-#define GL_RESET_NOTIFICATION_STRATEGY_EXT                      0x8256

-#define GL_LOSE_CONTEXT_ON_RESET_EXT                            0x8252

-#define GL_NO_RESET_NOTIFICATION_EXT                            0x8261

-#endif

-

-/* GL_EXT_separate_shader_objects */

-#ifndef GL_EXT_separate_shader_objects

-#define GL_VERTEX_SHADER_BIT_EXT                                0x00000001

-#define GL_FRAGMENT_SHADER_BIT_EXT                              0x00000002

-#define GL_ALL_SHADER_BITS_EXT                                  0xFFFFFFFF

-#define GL_PROGRAM_SEPARABLE_EXT                                0x8258

-#define GL_ACTIVE_PROGRAM_EXT                                   0x8259

-#define GL_PROGRAM_PIPELINE_BINDING_EXT                         0x825A

-#endif

-

-/* GL_EXT_shader_framebuffer_fetch */

-#ifndef GL_EXT_shader_framebuffer_fetch

-#define GL_FRAGMENT_SHADER_DISCARDS_SAMPLES_EXT                 0x8A52

-#endif

-

-/* GL_EXT_shader_texture_lod */

-/* No new tokens introduced by this extension. */

-

-/* GL_EXT_shadow_samplers */

-#ifndef GL_EXT_shadow_samplers

-#define GL_TEXTURE_COMPARE_MODE_EXT                             0x884C

-#define GL_TEXTURE_COMPARE_FUNC_EXT                             0x884D

-#define GL_COMPARE_REF_TO_TEXTURE_EXT                           0x884E

-#define GL_SAMPLER_2D_SHADOW_EXT                                0x8B62

-#endif

-

-/* GL_EXT_sRGB */

-#ifndef GL_EXT_sRGB

-#define GL_SRGB_EXT                                             0x8C40

-#define GL_SRGB_ALPHA_EXT                                       0x8C42

-#define GL_SRGB8_ALPHA8_EXT                                     0x8C43

-#define GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING_EXT            0x8210

-#endif

-

-/* GL_EXT_texture_compression_dxt1 */

-#ifndef GL_EXT_texture_compression_dxt1

-#define GL_COMPRESSED_RGB_S3TC_DXT1_EXT                         0x83F0

-#define GL_COMPRESSED_RGBA_S3TC_DXT1_EXT                        0x83F1

-#endif

-

-/* GL_EXT_texture_filter_anisotropic */

-#ifndef GL_EXT_texture_filter_anisotropic

-#define GL_TEXTURE_MAX_ANISOTROPY_EXT                           0x84FE

-#define GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT                       0x84FF

-#endif

-

-/* GL_EXT_texture_format_BGRA8888 */

-#ifndef GL_EXT_texture_format_BGRA8888

-#define GL_BGRA_EXT                                             0x80E1

-#endif

-

-/* GL_EXT_texture_rg */

-#ifndef GL_EXT_texture_rg

-#define GL_RED_EXT                                              0x1903

-#define GL_RG_EXT                                               0x8227

-#define GL_R8_EXT                                               0x8229

-#define GL_RG8_EXT                                              0x822B

-#endif

-

-/* GL_EXT_texture_storage */

-#ifndef GL_EXT_texture_storage

-#define GL_TEXTURE_IMMUTABLE_FORMAT_EXT                         0x912F

-#define GL_ALPHA8_EXT                                           0x803C  

-#define GL_LUMINANCE8_EXT                                       0x8040

-#define GL_LUMINANCE8_ALPHA8_EXT                                0x8045

-#define GL_RGBA32F_EXT                                          0x8814  

-#define GL_RGB32F_EXT                                           0x8815

-#define GL_ALPHA32F_EXT                                         0x8816

-#define GL_LUMINANCE32F_EXT                                     0x8818

-#define GL_LUMINANCE_ALPHA32F_EXT                               0x8819

-/* reuse GL_RGBA16F_EXT */

-/* reuse GL_RGB16F_EXT */

-#define GL_ALPHA16F_EXT                                         0x881C

-#define GL_LUMINANCE16F_EXT                                     0x881E

-#define GL_LUMINANCE_ALPHA16F_EXT                               0x881F

-#define GL_RGB10_A2_EXT                                         0x8059  

-#define GL_RGB10_EXT                                            0x8052

-#define GL_BGRA8_EXT                                            0x93A1

-#define GL_R8_EXT                                               0x8229

-#define GL_RG8_EXT                                              0x822B

-#define GL_R32F_EXT                                             0x822E  

-#define GL_RG32F_EXT                                            0x8230

-#define GL_R16F_EXT                                             0x822D

-#define GL_RG16F_EXT                                            0x822F

-#endif

-

-/* GL_EXT_texture_type_2_10_10_10_REV */

-#ifndef GL_EXT_texture_type_2_10_10_10_REV

-#define GL_UNSIGNED_INT_2_10_10_10_REV_EXT                      0x8368

-#endif

-

-/* GL_EXT_unpack_subimage */

-#ifndef GL_EXT_unpack_subimage

-#define GL_UNPACK_ROW_LENGTH_EXT                                0x0CF2

-#define GL_UNPACK_SKIP_ROWS_EXT                                 0x0CF3

-#define GL_UNPACK_SKIP_PIXELS_EXT                               0x0CF4

-#endif

-

-/*------------------------------------------------------------------------*

- * DMP extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_DMP_shader_binary */

-#ifndef GL_DMP_shader_binary

-#define GL_SHADER_BINARY_DMP                                    0x9250

-#endif

-

-/*------------------------------------------------------------------------*

- * FJ extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_FJ_shader_binary_GCCSO */

-#ifndef GL_FJ_shader_binary_GCCSO

-#define GL_GCCSO_SHADER_BINARY_F                                0x9260

-#endif

-

-/*------------------------------------------------------------------------*

- * IMG extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_IMG_program_binary */

-#ifndef GL_IMG_program_binary

-#define GL_SGX_PROGRAM_BINARY_IMG                               0x9130

-#endif

-

-/* GL_IMG_read_format */

-#ifndef GL_IMG_read_format

-#define GL_BGRA_IMG                                             0x80E1

-#define GL_UNSIGNED_SHORT_4_4_4_4_REV_IMG                       0x8365

-#endif

-

-/* GL_IMG_shader_binary */

-#ifndef GL_IMG_shader_binary

-#define GL_SGX_BINARY_IMG                                       0x8C0A

-#endif

-

-/* GL_IMG_texture_compression_pvrtc */

-#ifndef GL_IMG_texture_compression_pvrtc

-#define GL_COMPRESSED_RGB_PVRTC_4BPPV1_IMG                      0x8C00

-#define GL_COMPRESSED_RGB_PVRTC_2BPPV1_IMG                      0x8C01

-#define GL_COMPRESSED_RGBA_PVRTC_4BPPV1_IMG                     0x8C02

-#define GL_COMPRESSED_RGBA_PVRTC_2BPPV1_IMG                     0x8C03

-#endif

-

-/* GL_IMG_texture_compression_pvrtc2 */

-#ifndef GL_IMG_texture_compression_pvrtc2

-#define GL_COMPRESSED_RGBA_PVRTC_2BPPV2_IMG                     0x9137

-#define GL_COMPRESSED_RGBA_PVRTC_4BPPV2_IMG                     0x9138

-#endif

-

-/* GL_IMG_multisampled_render_to_texture */

-#ifndef GL_IMG_multisampled_render_to_texture

-#define GL_RENDERBUFFER_SAMPLES_IMG                             0x9133

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_IMG               0x9134

-#define GL_MAX_SAMPLES_IMG                                      0x9135

-#define GL_TEXTURE_SAMPLES_IMG                                  0x9136

-#endif

-

-/*------------------------------------------------------------------------*

- * NV extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_NV_coverage_sample */

-#ifndef GL_NV_coverage_sample

-#define GL_COVERAGE_COMPONENT_NV                                0x8ED0

-#define GL_COVERAGE_COMPONENT4_NV                               0x8ED1

-#define GL_COVERAGE_ATTACHMENT_NV                               0x8ED2

-#define GL_COVERAGE_BUFFERS_NV                                  0x8ED3

-#define GL_COVERAGE_SAMPLES_NV                                  0x8ED4

-#define GL_COVERAGE_ALL_FRAGMENTS_NV                            0x8ED5

-#define GL_COVERAGE_EDGE_FRAGMENTS_NV                           0x8ED6

-#define GL_COVERAGE_AUTOMATIC_NV                                0x8ED7

-#define GL_COVERAGE_BUFFER_BIT_NV                               0x8000

-#endif

-

-/* GL_NV_depth_nonlinear */

-#ifndef GL_NV_depth_nonlinear

-#define GL_DEPTH_COMPONENT16_NONLINEAR_NV                       0x8E2C

-#endif

-

-/* GL_NV_draw_buffers */

-#ifndef GL_NV_draw_buffers

-#define GL_MAX_DRAW_BUFFERS_NV                                  0x8824

-#define GL_DRAW_BUFFER0_NV                                      0x8825

-#define GL_DRAW_BUFFER1_NV                                      0x8826

-#define GL_DRAW_BUFFER2_NV                                      0x8827

-#define GL_DRAW_BUFFER3_NV                                      0x8828

-#define GL_DRAW_BUFFER4_NV                                      0x8829

-#define GL_DRAW_BUFFER5_NV                                      0x882A

-#define GL_DRAW_BUFFER6_NV                                      0x882B

-#define GL_DRAW_BUFFER7_NV                                      0x882C

-#define GL_DRAW_BUFFER8_NV                                      0x882D

-#define GL_DRAW_BUFFER9_NV                                      0x882E

-#define GL_DRAW_BUFFER10_NV                                     0x882F

-#define GL_DRAW_BUFFER11_NV                                     0x8830

-#define GL_DRAW_BUFFER12_NV                                     0x8831

-#define GL_DRAW_BUFFER13_NV                                     0x8832

-#define GL_DRAW_BUFFER14_NV                                     0x8833

-#define GL_DRAW_BUFFER15_NV                                     0x8834

-#define GL_COLOR_ATTACHMENT0_NV                                 0x8CE0

-#define GL_COLOR_ATTACHMENT1_NV                                 0x8CE1

-#define GL_COLOR_ATTACHMENT2_NV                                 0x8CE2

-#define GL_COLOR_ATTACHMENT3_NV                                 0x8CE3

-#define GL_COLOR_ATTACHMENT4_NV                                 0x8CE4

-#define GL_COLOR_ATTACHMENT5_NV                                 0x8CE5

-#define GL_COLOR_ATTACHMENT6_NV                                 0x8CE6

-#define GL_COLOR_ATTACHMENT7_NV                                 0x8CE7

-#define GL_COLOR_ATTACHMENT8_NV                                 0x8CE8

-#define GL_COLOR_ATTACHMENT9_NV                                 0x8CE9

-#define GL_COLOR_ATTACHMENT10_NV                                0x8CEA

-#define GL_COLOR_ATTACHMENT11_NV                                0x8CEB

-#define GL_COLOR_ATTACHMENT12_NV                                0x8CEC

-#define GL_COLOR_ATTACHMENT13_NV                                0x8CED

-#define GL_COLOR_ATTACHMENT14_NV                                0x8CEE

-#define GL_COLOR_ATTACHMENT15_NV                                0x8CEF

-#endif

-

-/* GL_EXT_draw_buffers */

-#ifndef GL_EXT_draw_buffers

-#define GL_MAX_DRAW_BUFFERS_EXT                                  0x8824

-#define GL_DRAW_BUFFER0_EXT                                      0x8825

-#define GL_DRAW_BUFFER1_EXT                                      0x8826

-#define GL_DRAW_BUFFER2_EXT                                      0x8827

-#define GL_DRAW_BUFFER3_EXT                                      0x8828

-#define GL_DRAW_BUFFER4_EXT                                      0x8829

-#define GL_DRAW_BUFFER5_EXT                                      0x882A

-#define GL_DRAW_BUFFER6_EXT                                      0x882B

-#define GL_DRAW_BUFFER7_EXT                                      0x882C

-#define GL_DRAW_BUFFER8_EXT                                      0x882D

-#define GL_DRAW_BUFFER9_EXT                                      0x882E

-#define GL_DRAW_BUFFER10_EXT                                     0x882F

-#define GL_DRAW_BUFFER11_EXT                                     0x8830

-#define GL_DRAW_BUFFER12_EXT                                     0x8831

-#define GL_DRAW_BUFFER13_EXT                                     0x8832

-#define GL_DRAW_BUFFER14_EXT                                     0x8833

-#define GL_DRAW_BUFFER15_EXT                                     0x8834

-#define GL_COLOR_ATTACHMENT0_EXT                                 0x8CE0

-#define GL_COLOR_ATTACHMENT1_EXT                                 0x8CE1

-#define GL_COLOR_ATTACHMENT2_EXT                                 0x8CE2

-#define GL_COLOR_ATTACHMENT3_EXT                                 0x8CE3

-#define GL_COLOR_ATTACHMENT4_EXT                                 0x8CE4

-#define GL_COLOR_ATTACHMENT5_EXT                                 0x8CE5

-#define GL_COLOR_ATTACHMENT6_EXT                                 0x8CE6

-#define GL_COLOR_ATTACHMENT7_EXT                                 0x8CE7

-#define GL_COLOR_ATTACHMENT8_EXT                                 0x8CE8

-#define GL_COLOR_ATTACHMENT9_EXT                                 0x8CE9

-#define GL_COLOR_ATTACHMENT10_EXT                                0x8CEA

-#define GL_COLOR_ATTACHMENT11_EXT                                0x8CEB

-#define GL_COLOR_ATTACHMENT12_EXT                                0x8CEC

-#define GL_COLOR_ATTACHMENT13_EXT                                0x8CED

-#define GL_COLOR_ATTACHMENT14_EXT                                0x8CEE

-#define GL_COLOR_ATTACHMENT15_EXT                                0x8CEF

-#define GL_MAX_COLOR_ATTACHMENTS_EXT                             0x8CDF

-#endif

-

-/* GL_NV_draw_instanced */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_fbo_color_attachments */

-#ifndef GL_NV_fbo_color_attachments

-#define GL_MAX_COLOR_ATTACHMENTS_NV                             0x8CDF

-/* GL_COLOR_ATTACHMENT{0-15}_NV defined in GL_NV_draw_buffers already. */

-#endif

-

-/* GL_NV_fence */

-#ifndef GL_NV_fence

-#define GL_ALL_COMPLETED_NV                                     0x84F2

-#define GL_FENCE_STATUS_NV                                      0x84F3

-#define GL_FENCE_CONDITION_NV                                   0x84F4

-#endif

-

-/* GL_NV_framebuffer_blit */

-#ifndef GL_NV_framebuffer_blit

-#define GL_READ_FRAMEBUFFER_NV                                  0x8CA8

-#define GL_DRAW_FRAMEBUFFER_NV                                  0x8CA9

-#define GL_DRAW_FRAMEBUFFER_BINDING_NV                          0x8CA6 

-#define GL_READ_FRAMEBUFFER_BINDING_NV                          0x8CAA

-#endif

-

-/* GL_NV_framebuffer_multisample */

-#ifndef GL_NV_framebuffer_multisample

-#define GL_RENDERBUFFER_SAMPLES_NV                              0x8CAB

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_NV                0x8D56

-#define GL_MAX_SAMPLES_NV                                       0x8D57

-#endif

-

-/* GL_NV_generate_mipmap_sRGB */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_instanced_arrays */

-#ifndef GL_NV_instanced_arrays

-#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR_NV                       0x88FE

-#endif

-

-/* GL_NV_read_buffer */

-#ifndef GL_NV_read_buffer

-#define GL_READ_BUFFER_NV                                       0x0C02

-#endif

-

-/* GL_NV_read_buffer_front */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_read_depth */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_read_depth_stencil */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_read_stencil */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_shadow_samplers_array */

-#ifndef GL_NV_shadow_samplers_array

-#define GL_SAMPLER_2D_ARRAY_SHADOW_NV                           0x8DC4

-#endif                                             

-                                                   

-/* GL_NV_shadow_samplers_cube */

-#ifndef GL_NV_shadow_samplers_cube                 

-#define GL_SAMPLER_CUBE_SHADOW_NV                               0x8DC5

-#endif

-

-/* GL_NV_sRGB_formats */

-#ifndef GL_NV_sRGB_formats

-#define GL_SLUMINANCE_NV                                        0x8C46

-#define GL_SLUMINANCE_ALPHA_NV                                  0x8C44

-#define GL_SRGB8_NV                                             0x8C41

-#define GL_SLUMINANCE8_NV                                       0x8C47

-#define GL_SLUMINANCE8_ALPHA8_NV                                0x8C45

-#define GL_COMPRESSED_SRGB_S3TC_DXT1_NV                         0x8C4C

-#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_NV                   0x8C4D

-#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_NV                   0x8C4E

-#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_NV                   0x8C4F

-#define GL_ETC1_SRGB8_NV                                        0x88EE

-#endif

-

-/* GL_NV_texture_border_clamp */

-#ifndef GL_NV_texture_border_clamp

-#define GL_TEXTURE_BORDER_COLOR_NV                              0x1004

-#define GL_CLAMP_TO_BORDER_NV                                   0x812D

-#endif

-

-/* GL_NV_texture_compression_s3tc_update */

-/* No new tokens introduced by this extension. */

-

-/* GL_NV_texture_npot_2D_mipmap */

-/* No new tokens introduced by this extension. */

-

-/*------------------------------------------------------------------------*

- * QCOM extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_QCOM_alpha_test */

-#ifndef GL_QCOM_alpha_test

-#define GL_ALPHA_TEST_QCOM                                      0x0BC0

-#define GL_ALPHA_TEST_FUNC_QCOM                                 0x0BC1

-#define GL_ALPHA_TEST_REF_QCOM                                  0x0BC2

-#endif

-

-/* GL_QCOM_binning_control */

-#ifndef GL_QCOM_binning_control

-#define GL_BINNING_CONTROL_HINT_QCOM                            0x8FB0

-#define GL_CPU_OPTIMIZED_QCOM                                   0x8FB1

-#define GL_GPU_OPTIMIZED_QCOM                                   0x8FB2

-#define GL_RENDER_DIRECT_TO_FRAMEBUFFER_QCOM                    0x8FB3

-#endif

-

-/* GL_QCOM_driver_control */

-/* No new tokens introduced by this extension. */

-

-/* GL_QCOM_extended_get */

-#ifndef GL_QCOM_extended_get

-#define GL_TEXTURE_WIDTH_QCOM                                   0x8BD2

-#define GL_TEXTURE_HEIGHT_QCOM                                  0x8BD3

-#define GL_TEXTURE_DEPTH_QCOM                                   0x8BD4

-#define GL_TEXTURE_INTERNAL_FORMAT_QCOM                         0x8BD5

-#define GL_TEXTURE_FORMAT_QCOM                                  0x8BD6

-#define GL_TEXTURE_TYPE_QCOM                                    0x8BD7

-#define GL_TEXTURE_IMAGE_VALID_QCOM                             0x8BD8

-#define GL_TEXTURE_NUM_LEVELS_QCOM                              0x8BD9

-#define GL_TEXTURE_TARGET_QCOM                                  0x8BDA

-#define GL_TEXTURE_OBJECT_VALID_QCOM                            0x8BDB

-#define GL_STATE_RESTORE                                        0x8BDC

-#endif

-

-/* GL_QCOM_extended_get2 */

-/* No new tokens introduced by this extension. */

-

-/* GL_QCOM_perfmon_global_mode */

-#ifndef GL_QCOM_perfmon_global_mode

-#define GL_PERFMON_GLOBAL_MODE_QCOM                             0x8FA0

-#endif

-

-/* GL_QCOM_writeonly_rendering */

-#ifndef GL_QCOM_writeonly_rendering

-#define GL_WRITEONLY_RENDERING_QCOM                             0x8823

-#endif

-

-/* GL_QCOM_tiled_rendering */

-#ifndef GL_QCOM_tiled_rendering

-#define GL_COLOR_BUFFER_BIT0_QCOM                               0x00000001

-#define GL_COLOR_BUFFER_BIT1_QCOM                               0x00000002

-#define GL_COLOR_BUFFER_BIT2_QCOM                               0x00000004

-#define GL_COLOR_BUFFER_BIT3_QCOM                               0x00000008

-#define GL_COLOR_BUFFER_BIT4_QCOM                               0x00000010

-#define GL_COLOR_BUFFER_BIT5_QCOM                               0x00000020

-#define GL_COLOR_BUFFER_BIT6_QCOM                               0x00000040

-#define GL_COLOR_BUFFER_BIT7_QCOM                               0x00000080

-#define GL_DEPTH_BUFFER_BIT0_QCOM                               0x00000100

-#define GL_DEPTH_BUFFER_BIT1_QCOM                               0x00000200

-#define GL_DEPTH_BUFFER_BIT2_QCOM                               0x00000400

-#define GL_DEPTH_BUFFER_BIT3_QCOM                               0x00000800

-#define GL_DEPTH_BUFFER_BIT4_QCOM                               0x00001000

-#define GL_DEPTH_BUFFER_BIT5_QCOM                               0x00002000

-#define GL_DEPTH_BUFFER_BIT6_QCOM                               0x00004000

-#define GL_DEPTH_BUFFER_BIT7_QCOM                               0x00008000

-#define GL_STENCIL_BUFFER_BIT0_QCOM                             0x00010000

-#define GL_STENCIL_BUFFER_BIT1_QCOM                             0x00020000

-#define GL_STENCIL_BUFFER_BIT2_QCOM                             0x00040000

-#define GL_STENCIL_BUFFER_BIT3_QCOM                             0x00080000

-#define GL_STENCIL_BUFFER_BIT4_QCOM                             0x00100000

-#define GL_STENCIL_BUFFER_BIT5_QCOM                             0x00200000

-#define GL_STENCIL_BUFFER_BIT6_QCOM                             0x00400000

-#define GL_STENCIL_BUFFER_BIT7_QCOM                             0x00800000

-#define GL_MULTISAMPLE_BUFFER_BIT0_QCOM                         0x01000000

-#define GL_MULTISAMPLE_BUFFER_BIT1_QCOM                         0x02000000

-#define GL_MULTISAMPLE_BUFFER_BIT2_QCOM                         0x04000000

-#define GL_MULTISAMPLE_BUFFER_BIT3_QCOM                         0x08000000

-#define GL_MULTISAMPLE_BUFFER_BIT4_QCOM                         0x10000000

-#define GL_MULTISAMPLE_BUFFER_BIT5_QCOM                         0x20000000

-#define GL_MULTISAMPLE_BUFFER_BIT6_QCOM                         0x40000000

-#define GL_MULTISAMPLE_BUFFER_BIT7_QCOM                         0x80000000

-#endif

-

-/*------------------------------------------------------------------------*

- * VIV extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_VIV_shader_binary */

-#ifndef GL_VIV_shader_binary

-#define GL_SHADER_BINARY_VIV                                    0x8FC4

-#endif

-

-/*------------------------------------------------------------------------*

- * End of extension tokens, start of corresponding extension functions

- *------------------------------------------------------------------------*/

-

-/*------------------------------------------------------------------------*

- * OES extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_OES_compressed_ETC1_RGB8_texture */

-#ifndef GL_OES_compressed_ETC1_RGB8_texture

-#define GL_OES_compressed_ETC1_RGB8_texture 1

-#endif

-

-/* GL_OES_compressed_paletted_texture */

-#ifndef GL_OES_compressed_paletted_texture

-#define GL_OES_compressed_paletted_texture 1

-#endif

-

-/* GL_OES_depth24 */

-#ifndef GL_OES_depth24

-#define GL_OES_depth24 1

-#endif

-

-/* GL_OES_depth32 */

-#ifndef GL_OES_depth32

-#define GL_OES_depth32 1

-#endif

-

-/* GL_OES_depth_texture */

-#ifndef GL_OES_depth_texture

-#define GL_OES_depth_texture 1

-#endif

-

-/* GL_OES_EGL_image */

-#ifndef GL_OES_EGL_image

-#define GL_OES_EGL_image 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glEGLImageTargetTexture2DOES (GLenum target, GLeglImageOES image);

-GL_APICALL void GL_APIENTRY glEGLImageTargetRenderbufferStorageOES (GLenum target, GLeglImageOES image);

-#endif

-typedef void (GL_APIENTRYP PFNGLEGLIMAGETARGETTEXTURE2DOESPROC) (GLenum target, GLeglImageOES image);

-typedef void (GL_APIENTRYP PFNGLEGLIMAGETARGETRENDERBUFFERSTORAGEOESPROC) (GLenum target, GLeglImageOES image);

-#endif

-

-/* GL_OES_EGL_image_external */

-#ifndef GL_OES_EGL_image_external

-#define GL_OES_EGL_image_external 1

-/* glEGLImageTargetTexture2DOES defined in GL_OES_EGL_image already. */

-#endif

-

-/* GL_OES_element_index_uint */

-#ifndef GL_OES_element_index_uint

-#define GL_OES_element_index_uint 1

-#endif

-

-/* GL_OES_fbo_render_mipmap */

-#ifndef GL_OES_fbo_render_mipmap

-#define GL_OES_fbo_render_mipmap 1

-#endif

-

-/* GL_OES_fragment_precision_high */

-#ifndef GL_OES_fragment_precision_high

-#define GL_OES_fragment_precision_high 1

-#endif

-

-/* GL_OES_get_program_binary */

-#ifndef GL_OES_get_program_binary

-#define GL_OES_get_program_binary 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glGetProgramBinaryOES (GLuint program, GLsizei bufSize, GLsizei *length, GLenum *binaryFormat, GLvoid *binary);

-GL_APICALL void GL_APIENTRY glProgramBinaryOES (GLuint program, GLenum binaryFormat, const GLvoid *binary, GLint length);

-#endif

-typedef void (GL_APIENTRYP PFNGLGETPROGRAMBINARYOESPROC) (GLuint program, GLsizei bufSize, GLsizei *length, GLenum *binaryFormat, GLvoid *binary);

-typedef void (GL_APIENTRYP PFNGLPROGRAMBINARYOESPROC) (GLuint program, GLenum binaryFormat, const GLvoid *binary, GLint length);

-#endif

-

-/* GL_OES_mapbuffer */

-#ifndef GL_OES_mapbuffer

-#define GL_OES_mapbuffer 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void* GL_APIENTRY glMapBufferOES (GLenum target, GLenum access);

-GL_APICALL GLboolean GL_APIENTRY glUnmapBufferOES (GLenum target);

-GL_APICALL void GL_APIENTRY glGetBufferPointervOES (GLenum target, GLenum pname, GLvoid** params);

-#endif

-typedef void* (GL_APIENTRYP PFNGLMAPBUFFEROESPROC) (GLenum target, GLenum access);

-typedef GLboolean (GL_APIENTRYP PFNGLUNMAPBUFFEROESPROC) (GLenum target);

-typedef void (GL_APIENTRYP PFNGLGETBUFFERPOINTERVOESPROC) (GLenum target, GLenum pname, GLvoid** params);

-#endif

-

-/* GL_OES_packed_depth_stencil */

-#ifndef GL_OES_packed_depth_stencil

-#define GL_OES_packed_depth_stencil 1

-#endif

-

-/* GL_OES_required_internalformat */

-#ifndef GL_OES_required_internalformat

-#define GL_OES_required_internalformat 1

-#endif

-

-/* GL_OES_rgb8_rgba8 */

-#ifndef GL_OES_rgb8_rgba8

-#define GL_OES_rgb8_rgba8 1

-#endif

-

-/* GL_OES_standard_derivatives */

-#ifndef GL_OES_standard_derivatives

-#define GL_OES_standard_derivatives 1

-#endif

-

-/* GL_OES_stencil1 */

-#ifndef GL_OES_stencil1

-#define GL_OES_stencil1 1

-#endif

-

-/* GL_OES_stencil4 */

-#ifndef GL_OES_stencil4

-#define GL_OES_stencil4 1

-#endif

-

-#ifndef GL_OES_surfaceless_context

-#define GL_OES_surfaceless_context 1

-#endif

-

-/* GL_OES_texture_3D */

-#ifndef GL_OES_texture_3D

-#define GL_OES_texture_3D 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glTexImage3DOES (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void GL_APIENTRY glTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void GL_APIENTRY glCopyTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL void GL_APIENTRY glCompressedTexImage3DOES (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void GL_APIENTRY glCompressedTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void GL_APIENTRY glFramebufferTexture3DOES (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset);

-#endif

-typedef void (GL_APIENTRYP PFNGLTEXIMAGE3DOESPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);

-typedef void (GL_APIENTRYP PFNGLTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);

-typedef void (GL_APIENTRYP PFNGLCOPYTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLCOMPRESSEDTEXIMAGE3DOESPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);

-typedef void (GL_APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);

-typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE3DOES) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset);

-#endif

-

-/* GL_OES_texture_float */

-#ifndef GL_OES_texture_float

-#define GL_OES_texture_float 1

-#endif

-

-/* GL_OES_texture_float_linear */

-#ifndef GL_OES_texture_float_linear

-#define GL_OES_texture_float_linear 1

-#endif

-

-/* GL_OES_texture_half_float */

-#ifndef GL_OES_texture_half_float

-#define GL_OES_texture_half_float 1

-#endif

-

-/* GL_OES_texture_half_float_linear */

-#ifndef GL_OES_texture_half_float_linear

-#define GL_OES_texture_half_float_linear 1

-#endif

-

-/* GL_OES_texture_npot */

-#ifndef GL_OES_texture_npot

-#define GL_OES_texture_npot 1

-#endif

-

-/* GL_OES_vertex_array_object */

-#ifndef GL_OES_vertex_array_object

-#define GL_OES_vertex_array_object 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glBindVertexArrayOES (GLuint array);

-GL_APICALL void GL_APIENTRY glDeleteVertexArraysOES (GLsizei n, const GLuint *arrays);

-GL_APICALL void GL_APIENTRY glGenVertexArraysOES (GLsizei n, GLuint *arrays);

-GL_APICALL GLboolean GL_APIENTRY glIsVertexArrayOES (GLuint array);

-#endif

-typedef void (GL_APIENTRYP PFNGLBINDVERTEXARRAYOESPROC) (GLuint array);

-typedef void (GL_APIENTRYP PFNGLDELETEVERTEXARRAYSOESPROC) (GLsizei n, const GLuint *arrays);

-typedef void (GL_APIENTRYP PFNGLGENVERTEXARRAYSOESPROC) (GLsizei n, GLuint *arrays);

-typedef GLboolean (GL_APIENTRYP PFNGLISVERTEXARRAYOESPROC) (GLuint array);

-#endif

-

-/* GL_OES_vertex_half_float */

-#ifndef GL_OES_vertex_half_float

-#define GL_OES_vertex_half_float 1

-#endif

-

-/* GL_OES_vertex_type_10_10_10_2 */

-#ifndef GL_OES_vertex_type_10_10_10_2

-#define GL_OES_vertex_type_10_10_10_2 1

-#endif

-

-/*------------------------------------------------------------------------*

- * KHR extension functions

- *------------------------------------------------------------------------*/

-

-#ifndef GL_KHR_debug

-#define GL_KHR_debug 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDebugMessageControl (GLenum source, GLenum type, GLenum severity, GLsizei count, const GLuint *ids, GLboolean enabled);

-GL_APICALL void GL_APIENTRY glDebugMessageInsert (GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar *buf);

-GL_APICALL void GL_APIENTRY glDebugMessageCallback (GLDEBUGPROC callback, const void *userParam);

-GL_APICALL GLuint GL_APIENTRY glGetDebugMessageLog (GLuint count, GLsizei bufsize, GLenum *sources, GLenum *types, GLuint *ids, GLenum *severities, GLsizei *lengths, GLchar *messageLog);

-GL_APICALL void GL_APIENTRY glPushDebugGroup (GLenum source, GLuint id, GLsizei length, const GLchar *message);

-GL_APICALL void GL_APIENTRY glPopDebugGroup (void);

-GL_APICALL void GL_APIENTRY glObjectLabel (GLenum identifier, GLuint name, GLsizei length, const GLchar *label);

-GL_APICALL void GL_APIENTRY glGetObjectLabel (GLenum identifier, GLuint name, GLsizei bufSize, GLsizei *length, GLchar *label);

-GL_APICALL void GL_APIENTRY glObjectPtrLabel (const void *ptr, GLsizei length, const GLchar *label);

-GL_APICALL void GL_APIENTRY glGetObjectPtrLabel (const void *ptr, GLsizei bufSize, GLsizei *length, GLchar *label);

-GL_APICALL void GL_APIENTRY glGetPointerv (GLenum pname, void **params);

-#endif 

-typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGECONTROLPROC) (GLenum source, GLenum type, GLenum severity, GLsizei count, const GLuint *ids, GLboolean enabled);

-typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGEINSERTPROC) (GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar *buf);

-typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGECALLBACKPROC) (GLDEBUGPROC callback, const void *userParam);

-typedef GLuint (GL_APIENTRYP PFNGLGETDEBUGMESSAGELOGPROC) (GLuint count, GLsizei bufsize, GLenum *sources, GLenum *types, GLuint *ids, GLenum *severities, GLsizei *lengths, GLchar *messageLog);

-typedef void (GL_APIENTRYP PFNGLPUSHDEBUGGROUPPROC) (GLenum source, GLuint id, GLsizei length, const GLchar *message);

-typedef void (GL_APIENTRYP PFNGLPOPDEBUGGROUPPROC) (void);

-typedef void (GL_APIENTRYP PFNGLOBJECTLABELPROC) (GLenum identifier, GLuint name, GLsizei length, const GLchar *label);

-typedef void (GL_APIENTRYP PFNGLGETOBJECTLABELPROC) (GLenum identifier, GLuint name, GLsizei bufSize, GLsizei *length, GLchar *label);

-typedef void (GL_APIENTRYP PFNGLOBJECTPTRLABELPROC) (const void *ptr, GLsizei length, const GLchar *label);

-typedef void (GL_APIENTRYP PFNGLGETOBJECTPTRLABELPROC) (const void *ptr, GLsizei bufSize, GLsizei *length, GLchar *label);

-typedef void (GL_APIENTRYP PFNGLGETPOINTERVPROC) (GLenum pname, void **params);

-#endif

-

-#ifndef GL_KHR_texture_compression_astc_ldr

-#define GL_KHR_texture_compression_astc_ldr 1

-#endif

-

-

-/*------------------------------------------------------------------------*

- * AMD extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_AMD_compressed_3DC_texture */

-#ifndef GL_AMD_compressed_3DC_texture

-#define GL_AMD_compressed_3DC_texture 1

-#endif

-

-/* GL_AMD_compressed_ATC_texture */

-#ifndef GL_AMD_compressed_ATC_texture

-#define GL_AMD_compressed_ATC_texture 1

-#endif

-

-/* AMD_performance_monitor */

-#ifndef GL_AMD_performance_monitor

-#define GL_AMD_performance_monitor 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glGetPerfMonitorGroupsAMD (GLint *numGroups, GLsizei groupsSize, GLuint *groups);

-GL_APICALL void GL_APIENTRY glGetPerfMonitorCountersAMD (GLuint group, GLint *numCounters, GLint *maxActiveCounters, GLsizei counterSize, GLuint *counters);

-GL_APICALL void GL_APIENTRY glGetPerfMonitorGroupStringAMD (GLuint group, GLsizei bufSize, GLsizei *length, GLchar *groupString);

-GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterStringAMD (GLuint group, GLuint counter, GLsizei bufSize, GLsizei *length, GLchar *counterString);

-GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterInfoAMD (GLuint group, GLuint counter, GLenum pname, GLvoid *data);

-GL_APICALL void GL_APIENTRY glGenPerfMonitorsAMD (GLsizei n, GLuint *monitors);

-GL_APICALL void GL_APIENTRY glDeletePerfMonitorsAMD (GLsizei n, GLuint *monitors);

-GL_APICALL void GL_APIENTRY glSelectPerfMonitorCountersAMD (GLuint monitor, GLboolean enable, GLuint group, GLint numCounters, GLuint *countersList);

-GL_APICALL void GL_APIENTRY glBeginPerfMonitorAMD (GLuint monitor);

-GL_APICALL void GL_APIENTRY glEndPerfMonitorAMD (GLuint monitor);

-GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterDataAMD (GLuint monitor, GLenum pname, GLsizei dataSize, GLuint *data, GLint *bytesWritten);

-#endif

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORGROUPSAMDPROC) (GLint *numGroups, GLsizei groupsSize, GLuint *groups);

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERSAMDPROC) (GLuint group, GLint *numCounters, GLint *maxActiveCounters, GLsizei counterSize, GLuint *counters);

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORGROUPSTRINGAMDPROC) (GLuint group, GLsizei bufSize, GLsizei *length, GLchar *groupString);

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERSTRINGAMDPROC) (GLuint group, GLuint counter, GLsizei bufSize, GLsizei *length, GLchar *counterString);

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERINFOAMDPROC) (GLuint group, GLuint counter, GLenum pname, GLvoid *data);

-typedef void (GL_APIENTRYP PFNGLGENPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors);

-typedef void (GL_APIENTRYP PFNGLDELETEPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors);

-typedef void (GL_APIENTRYP PFNGLSELECTPERFMONITORCOUNTERSAMDPROC) (GLuint monitor, GLboolean enable, GLuint group, GLint numCounters, GLuint *countersList);

-typedef void (GL_APIENTRYP PFNGLBEGINPERFMONITORAMDPROC) (GLuint monitor);

-typedef void (GL_APIENTRYP PFNGLENDPERFMONITORAMDPROC) (GLuint monitor);

-typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERDATAAMDPROC) (GLuint monitor, GLenum pname, GLsizei dataSize, GLuint *data, GLint *bytesWritten);

-#endif

-

-/* GL_AMD_program_binary_Z400 */

-#ifndef GL_AMD_program_binary_Z400

-#define GL_AMD_program_binary_Z400 1

-#endif

-

-/*------------------------------------------------------------------------*

- * ANGLE extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_ANGLE_depth_texture */

-#ifndef GL_ANGLE_depth_texture

-#define GL_ANGLE_depth_texture 1

-#endif

-

-/* GL_ANGLE_framebuffer_blit */

-#ifndef GL_ANGLE_framebuffer_blit

-#define GL_ANGLE_framebuffer_blit 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glBlitFramebufferANGLE (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);

-#endif

-typedef void (GL_APIENTRYP PFNGLBLITFRAMEBUFFERANGLEPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);

-#endif

-

-/* GL_ANGLE_framebuffer_multisample */

-#ifndef GL_ANGLE_framebuffer_multisample

-#define GL_ANGLE_framebuffer_multisample 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleANGLE (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-#endif

-typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEANGLEPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-#endif

-

-#ifndef GL_ANGLE_instanced_arrays 

-#define GL_ANGLE_instanced_arrays 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDrawArraysInstancedANGLE (GLenum mode, GLint first, GLsizei count, GLsizei primcount);

-GL_APICALL void GL_APIENTRY glDrawElementsInstancedANGLE (GLenum mode, GLsizei count, GLenum type, const void *indices, GLsizei primcount);

-GL_APICALL void GL_APIENTRY glVertexAttribDivisorANGLE (GLuint index, GLuint divisor);

-#endif

-typedef void (GL_APIENTRYP PFNGLDRAWARRAYSINSTANCEDANGLEPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount);

-typedef void (GL_APIENTRYP PFNGLDRAWELEMENTSINSTANCEDANGLEPROC) (GLenum mode, GLsizei count, GLenum type, const void *indices, GLsizei primcount);

-typedef void (GL_APIENTRYP PFNGLVERTEXATTRIBDIVISORANGLEPROC) (GLuint index, GLuint divisor);

-#endif

-

-/* GL_ANGLE_pack_reverse_row_order */

-#ifndef GL_ANGLE_pack_reverse_row_order 

-#define GL_ANGLE_pack_reverse_row_order 1

-#endif

-

-/* GL_ANGLE_program_binary */

-#ifndef GL_ANGLE_program_binary

-#define GL_ANGLE_program_binary 1

-#endif

-

-/* GL_ANGLE_texture_compression_dxt3 */

-#ifndef GL_ANGLE_texture_compression_dxt3 

-#define GL_ANGLE_texture_compression_dxt3 1

-#endif

-

-/* GL_ANGLE_texture_compression_dxt5 */

-#ifndef GL_ANGLE_texture_compression_dxt5 

-#define GL_ANGLE_texture_compression_dxt5 1

-#endif

-

-/* GL_ANGLE_texture_usage */

-#ifndef GL_ANGLE_texture_usage 

-#define GL_ANGLE_texture_usage 1

-#endif

-

-#ifndef GL_ANGLE_translated_shader_source 

-#define GL_ANGLE_translated_shader_source 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glGetTranslatedShaderSourceANGLE (GLuint shader, GLsizei bufsize, GLsizei *length, GLchar *source);

-#endif

-typedef void (GL_APIENTRYP PFNGLGETTRANSLATEDSHADERSOURCEANGLEPROC) (GLuint shader, GLsizei bufsize, GLsizei *length, GLchar *source);

-#endif

-

-/*------------------------------------------------------------------------*

- * APPLE extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_APPLE_copy_texture_levels */

-#ifndef GL_APPLE_copy_texture_levels

-#define GL_APPLE_copy_texture_levels 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glCopyTextureLevelsAPPLE (GLuint destinationTexture, GLuint sourceTexture, GLint sourceBaseLevel, GLsizei sourceLevelCount);

-#endif

-typedef void (GL_APIENTRYP PFNGLCOPYTEXTURELEVELSAPPLEPROC) (GLuint destinationTexture, GLuint sourceTexture, GLint sourceBaseLevel, GLsizei sourceLevelCount);

-#endif

-

-/* GL_APPLE_framebuffer_multisample */

-#ifndef GL_APPLE_framebuffer_multisample

-#define GL_APPLE_framebuffer_multisample 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleAPPLE (GLenum, GLsizei, GLenum, GLsizei, GLsizei);

-GL_APICALL void GL_APIENTRY glResolveMultisampleFramebufferAPPLE (void);

-#endif /* GL_GLEXT_PROTOTYPES */

-typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEAPPLEPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLRESOLVEMULTISAMPLEFRAMEBUFFERAPPLEPROC) (void);

-#endif

-

-/* GL_APPLE_rgb_422 */

-#ifndef GL_APPLE_rgb_422

-#define GL_APPLE_rgb_422 1

-#endif

-

-/* GL_APPLE_sync */

-#ifndef GL_APPLE_sync

-#define GL_APPLE_sync 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL GLsync GL_APIENTRY glFenceSyncAPPLE (GLenum condition, GLbitfield flags);

-GL_APICALL GLboolean GL_APIENTRY glIsSyncAPPLE (GLsync sync);

-GL_APICALL void GL_APIENTRY glDeleteSyncAPPLE (GLsync sync);

-GL_APICALL GLenum GL_APIENTRY glClientWaitSyncAPPLE (GLsync sync, GLbitfield flags, GLuint64 timeout);

-GL_APICALL void GL_APIENTRY glWaitSyncAPPLE (GLsync sync, GLbitfield flags, GLuint64 timeout);

-GL_APICALL void GL_APIENTRY glGetInteger64vAPPLE (GLenum pname, GLint64 *params);

-GL_APICALL void GL_APIENTRY glGetSyncivAPPLE (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values);

-#endif

-typedef GLsync (GL_APIENTRYP PFNGLFENCESYNCAPPLEPROC) (GLenum condition, GLbitfield flags);

-typedef GLboolean (GL_APIENTRYP PFNGLISSYNCAPPLEPROC) (GLsync sync);

-typedef void (GL_APIENTRYP PFNGLDELETESYNCAPPLEPROC) (GLsync sync);

-typedef GLenum (GL_APIENTRYP PFNGLCLIENTWAITSYNCAPPLEPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);

-typedef void (GL_APIENTRYP PFNGLWAITSYNCAPPLEPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);

-typedef void (GL_APIENTRYP PFNGLGETINTEGER64VAPPLEPROC) (GLenum pname, GLint64 *params);

-typedef void (GL_APIENTRYP PFNGLGETSYNCIVAPPLEPROC) (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values);

-#endif

-

-/* GL_APPLE_texture_format_BGRA8888 */

-#ifndef GL_APPLE_texture_format_BGRA8888

-#define GL_APPLE_texture_format_BGRA8888 1

-#endif

-

-/* GL_APPLE_texture_max_level */

-#ifndef GL_APPLE_texture_max_level

-#define GL_APPLE_texture_max_level 1

-#endif

-

-/*------------------------------------------------------------------------*

- * ARM extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_ARM_mali_program_binary */

-#ifndef GL_ARM_mali_program_binary

-#define GL_ARM_mali_program_binary 1

-#endif

-

-/* GL_ARM_mali_shader_binary */

-#ifndef GL_ARM_mali_shader_binary

-#define GL_ARM_mali_shader_binary 1

-#endif

-

-/* GL_ARM_rgba8 */

-#ifndef GL_ARM_rgba8

-#define GL_ARM_rgba8 1

-#endif

-

-/*------------------------------------------------------------------------*

- * EXT extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_EXT_blend_minmax */

-#ifndef GL_EXT_blend_minmax

-#define GL_EXT_blend_minmax 1

-#endif

-

-/* GL_EXT_color_buffer_half_float */

-#ifndef GL_EXT_color_buffer_half_float

-#define GL_EXT_color_buffer_half_float 1

-#endif

-

-/* GL_EXT_debug_label */

-#ifndef GL_EXT_debug_label

-#define GL_EXT_debug_label 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glLabelObjectEXT (GLenum type, GLuint object, GLsizei length, const GLchar *label);

-GL_APICALL void GL_APIENTRY glGetObjectLabelEXT (GLenum type, GLuint object, GLsizei bufSize, GLsizei *length, GLchar *label);

-#endif

-typedef void (GL_APIENTRYP PFNGLLABELOBJECTEXTPROC) (GLenum type, GLuint object, GLsizei length, const GLchar *label);

-typedef void (GL_APIENTRYP PFNGLGETOBJECTLABELEXTPROC) (GLenum type, GLuint object, GLsizei bufSize, GLsizei *length, GLchar *label);

-#endif

-

-/* GL_EXT_debug_marker */

-#ifndef GL_EXT_debug_marker

-#define GL_EXT_debug_marker 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glInsertEventMarkerEXT (GLsizei length, const GLchar *marker);

-GL_APICALL void GL_APIENTRY glPushGroupMarkerEXT (GLsizei length, const GLchar *marker);

-GL_APICALL void GL_APIENTRY glPopGroupMarkerEXT (void);

-#endif

-typedef void (GL_APIENTRYP PFNGLINSERTEVENTMARKEREXTPROC) (GLsizei length, const GLchar *marker);

-typedef void (GL_APIENTRYP PFNGLPUSHGROUPMARKEREXTPROC) (GLsizei length, const GLchar *marker);

-typedef void (GL_APIENTRYP PFNGLPOPGROUPMARKEREXTPROC) (void);

-#endif

-

-/* GL_EXT_discard_framebuffer */

-#ifndef GL_EXT_discard_framebuffer

-#define GL_EXT_discard_framebuffer 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDiscardFramebufferEXT (GLenum target, GLsizei numAttachments, const GLenum *attachments);

-#endif

-typedef void (GL_APIENTRYP PFNGLDISCARDFRAMEBUFFEREXTPROC) (GLenum target, GLsizei numAttachments, const GLenum *attachments);

-#endif

-

-/* GL_EXT_map_buffer_range */

-#ifndef GL_EXT_map_buffer_range

-#define GL_EXT_map_buffer_range 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void* GL_APIENTRY glMapBufferRangeEXT (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);

-GL_APICALL void GL_APIENTRY glFlushMappedBufferRangeEXT (GLenum target, GLintptr offset, GLsizeiptr length);

-#endif

-typedef void* (GL_APIENTRYP PFNGLMAPBUFFERRANGEEXTPROC) (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);

-typedef void (GL_APIENTRYP PFNGLFLUSHMAPPEDBUFFERRANGEEXTPROC) (GLenum target, GLintptr offset, GLsizeiptr length);

-#endif

-

-/* GL_EXT_multisampled_render_to_texture */

-#ifndef GL_EXT_multisampled_render_to_texture

-#define GL_EXT_multisampled_render_to_texture 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleEXT (GLenum, GLsizei, GLenum, GLsizei, GLsizei);

-GL_APICALL void GL_APIENTRY glFramebufferTexture2DMultisampleEXT (GLenum, GLenum, GLenum, GLuint, GLint, GLsizei);

-#endif

-typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEEXTPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DMULTISAMPLEEXTPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLsizei samples);

-#endif

-

-/* GL_EXT_multiview_draw_buffers */

-#ifndef GL_EXT_multiview_draw_buffers

-#define GL_EXT_multiview_draw_buffers 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glReadBufferIndexedEXT (GLenum src, GLint index);

-GL_APICALL void GL_APIENTRY glDrawBuffersIndexedEXT (GLint n, const GLenum *location, const GLint *indices);

-GL_APICALL void GL_APIENTRY glGetIntegeri_vEXT (GLenum target, GLuint index, GLint *data);

-#endif

-typedef void (GL_APIENTRYP PFNGLREADBUFFERINDEXEDEXTPROC) (GLenum src, GLint index);

-typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSINDEXEDEXTPROC) (GLint n, const GLenum *location, const GLint *indices);

-typedef void (GL_APIENTRYP PFNGLGETINTEGERI_VEXTPROC) (GLenum target, GLuint index, GLint *data);

-#endif

-

-#ifndef GL_EXT_multi_draw_arrays

-#define GL_EXT_multi_draw_arrays 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glMultiDrawArraysEXT (GLenum, const GLint *, const GLsizei *, GLsizei);

-GL_APICALL void GL_APIENTRY glMultiDrawElementsEXT (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei);

-#endif /* GL_GLEXT_PROTOTYPES */

-typedef void (GL_APIENTRYP PFNGLMULTIDRAWARRAYSEXTPROC) (GLenum mode, GLint *first, GLsizei *count, GLsizei primcount);

-typedef void (GL_APIENTRYP PFNGLMULTIDRAWELEMENTSEXTPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount);

-#endif

-

-/* GL_EXT_occlusion_query_boolean */

-#ifndef GL_EXT_occlusion_query_boolean

-#define GL_EXT_occlusion_query_boolean 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glGenQueriesEXT (GLsizei n, GLuint *ids);

-GL_APICALL void GL_APIENTRY glDeleteQueriesEXT (GLsizei n, const GLuint *ids);

-GL_APICALL GLboolean GL_APIENTRY glIsQueryEXT (GLuint id);

-GL_APICALL void GL_APIENTRY glBeginQueryEXT (GLenum target, GLuint id);

-GL_APICALL void GL_APIENTRY glEndQueryEXT (GLenum target);

-GL_APICALL void GL_APIENTRY glGetQueryivEXT (GLenum target, GLenum pname, GLint *params);

-GL_APICALL void GL_APIENTRY glGetQueryObjectuivEXT (GLuint id, GLenum pname, GLuint *params);

-#endif

-typedef void (GL_APIENTRYP PFNGLGENQUERIESEXTPROC) (GLsizei n, GLuint *ids);

-typedef void (GL_APIENTRYP PFNGLDELETEQUERIESEXTPROC) (GLsizei n, const GLuint *ids);

-typedef GLboolean (GL_APIENTRYP PFNGLISQUERYEXTPROC) (GLuint id);

-typedef void (GL_APIENTRYP PFNGLBEGINQUERYEXTPROC) (GLenum target, GLuint id);

-typedef void (GL_APIENTRYP PFNGLENDQUERYEXTPROC) (GLenum target);

-typedef void (GL_APIENTRYP PFNGLGETQUERYIVEXTPROC) (GLenum target, GLenum pname, GLint *params);

-typedef void (GL_APIENTRYP PFNGLGETQUERYOBJECTUIVEXTPROC) (GLuint id, GLenum pname, GLuint *params);

-#endif

-

-/* GL_EXT_read_format_bgra */

-#ifndef GL_EXT_read_format_bgra

-#define GL_EXT_read_format_bgra 1

-#endif

-

-/* GL_EXT_robustness */

-#ifndef GL_EXT_robustness

-#define GL_EXT_robustness 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL GLenum GL_APIENTRY glGetGraphicsResetStatusEXT (void);

-GL_APICALL void GL_APIENTRY glReadnPixelsEXT (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void *data);

-GL_APICALL void GL_APIENTRY glGetnUniformfvEXT (GLuint program, GLint location, GLsizei bufSize, float *params);

-GL_APICALL void GL_APIENTRY glGetnUniformivEXT (GLuint program, GLint location, GLsizei bufSize, GLint *params);

-#endif

-typedef GLenum (GL_APIENTRYP PFNGLGETGRAPHICSRESETSTATUSEXTPROC) (void);

-typedef void (GL_APIENTRYP PFNGLREADNPIXELSEXTPROC) (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void *data);

-typedef void (GL_APIENTRYP PFNGLGETNUNIFORMFVEXTPROC) (GLuint program, GLint location, GLsizei bufSize, float *params);

-typedef void (GL_APIENTRYP PFNGLGETNUNIFORMIVEXTPROC) (GLuint program, GLint location, GLsizei bufSize, GLint *params);

-#endif

-

-/* GL_EXT_separate_shader_objects */

-#ifndef GL_EXT_separate_shader_objects

-#define GL_EXT_separate_shader_objects 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glUseProgramStagesEXT (GLuint pipeline, GLbitfield stages, GLuint program);

-GL_APICALL void GL_APIENTRY glActiveShaderProgramEXT (GLuint pipeline, GLuint program);

-GL_APICALL GLuint GL_APIENTRY glCreateShaderProgramvEXT (GLenum type, GLsizei count, const GLchar **strings);

-GL_APICALL void GL_APIENTRY glBindProgramPipelineEXT (GLuint pipeline);

-GL_APICALL void GL_APIENTRY glDeleteProgramPipelinesEXT (GLsizei n, const GLuint *pipelines);

-GL_APICALL void GL_APIENTRY glGenProgramPipelinesEXT (GLsizei n, GLuint *pipelines);

-GL_APICALL GLboolean GL_APIENTRY glIsProgramPipelineEXT (GLuint pipeline);

-GL_APICALL void GL_APIENTRY glProgramParameteriEXT (GLuint program, GLenum pname, GLint value);

-GL_APICALL void GL_APIENTRY glGetProgramPipelineivEXT (GLuint pipeline, GLenum pname, GLint *params);

-GL_APICALL void GL_APIENTRY glProgramUniform1iEXT (GLuint program, GLint location, GLint x);

-GL_APICALL void GL_APIENTRY glProgramUniform2iEXT (GLuint program, GLint location, GLint x, GLint y);

-GL_APICALL void GL_APIENTRY glProgramUniform3iEXT (GLuint program, GLint location, GLint x, GLint y, GLint z);

-GL_APICALL void GL_APIENTRY glProgramUniform4iEXT (GLuint program, GLint location, GLint x, GLint y, GLint z, GLint w);

-GL_APICALL void GL_APIENTRY glProgramUniform1fEXT (GLuint program, GLint location, GLfloat x);

-GL_APICALL void GL_APIENTRY glProgramUniform2fEXT (GLuint program, GLint location, GLfloat x, GLfloat y);

-GL_APICALL void GL_APIENTRY glProgramUniform3fEXT (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z);

-GL_APICALL void GL_APIENTRY glProgramUniform4fEXT (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-GL_APICALL void GL_APIENTRY glProgramUniform1ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);

-GL_APICALL void GL_APIENTRY glProgramUniform2ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);

-GL_APICALL void GL_APIENTRY glProgramUniform3ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);

-GL_APICALL void GL_APIENTRY glProgramUniform4ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);

-GL_APICALL void GL_APIENTRY glProgramUniform1fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniform2fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniform3fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniform4fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniformMatrix2fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniformMatrix3fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glProgramUniformMatrix4fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-GL_APICALL void GL_APIENTRY glValidateProgramPipelineEXT (GLuint pipeline);

-GL_APICALL void GL_APIENTRY glGetProgramPipelineInfoLogEXT (GLuint pipeline, GLsizei bufSize, GLsizei *length, GLchar *infoLog);

-#endif

-typedef void (GL_APIENTRYP PFNGLUSEPROGRAMSTAGESEXTPROC) (GLuint pipeline, GLbitfield stages, GLuint program);

-typedef void (GL_APIENTRYP PFNGLACTIVESHADERPROGRAMEXTPROC) (GLuint pipeline, GLuint program);

-typedef GLuint (GL_APIENTRYP PFNGLCREATESHADERPROGRAMVEXTPROC) (GLenum type, GLsizei count, const GLchar **strings);

-typedef void (GL_APIENTRYP PFNGLBINDPROGRAMPIPELINEEXTPROC) (GLuint pipeline);

-typedef void (GL_APIENTRYP PFNGLDELETEPROGRAMPIPELINESEXTPROC) (GLsizei n, const GLuint *pipelines);

-typedef void (GL_APIENTRYP PFNGLGENPROGRAMPIPELINESEXTPROC) (GLsizei n, GLuint *pipelines);

-typedef GLboolean (GL_APIENTRYP PFNGLISPROGRAMPIPELINEEXTPROC) (GLuint pipeline);

-typedef void (GL_APIENTRYP PFNGLPROGRAMPARAMETERIEXTPROC) (GLuint program, GLenum pname, GLint value);

-typedef void (GL_APIENTRYP PFNGLGETPROGRAMPIPELINEIVEXTPROC) (GLuint pipeline, GLenum pname, GLint *params);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1IEXTPROC) (GLuint program, GLint location, GLint x);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2IEXTPROC) (GLuint program, GLint location, GLint x, GLint y);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3IEXTPROC) (GLuint program, GLint location, GLint x, GLint y, GLint z);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4IEXTPROC) (GLuint program, GLint location, GLint x, GLint y, GLint z, GLint w);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1FEXTPROC) (GLuint program, GLint location, GLfloat x);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX2FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX3FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX4FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);

-typedef void (GL_APIENTRYP PFNGLVALIDATEPROGRAMPIPELINEEXTPROC) (GLuint pipeline);

-typedef void (GL_APIENTRYP PFNGLGETPROGRAMPIPELINEINFOLOGEXTPROC) (GLuint pipeline, GLsizei bufSize, GLsizei *length, GLchar *infoLog);

-#endif

-

-/* GL_EXT_shader_framebuffer_fetch */

-#ifndef GL_EXT_shader_framebuffer_fetch

-#define GL_EXT_shader_framebuffer_fetch 1

-#endif

-

-/* GL_EXT_shader_texture_lod */

-#ifndef GL_EXT_shader_texture_lod

-#define GL_EXT_shader_texture_lod 1

-#endif

-

-/* GL_EXT_shadow_samplers */

-#ifndef GL_EXT_shadow_samplers

-#define GL_EXT_shadow_samplers 1

-#endif

-

-/* GL_EXT_sRGB */

-#ifndef GL_EXT_sRGB

-#define GL_EXT_sRGB 1

-#endif

-

-/* GL_EXT_texture_compression_dxt1 */

-#ifndef GL_EXT_texture_compression_dxt1

-#define GL_EXT_texture_compression_dxt1 1

-#endif

-

-/* GL_EXT_texture_filter_anisotropic */

-#ifndef GL_EXT_texture_filter_anisotropic

-#define GL_EXT_texture_filter_anisotropic 1

-#endif

-

-/* GL_EXT_texture_format_BGRA8888 */

-#ifndef GL_EXT_texture_format_BGRA8888

-#define GL_EXT_texture_format_BGRA8888 1

-#endif

-

-/* GL_EXT_texture_rg */

-#ifndef GL_EXT_texture_rg

-#define GL_EXT_texture_rg 1

-#endif

-

-/* GL_EXT_texture_storage */

-#ifndef GL_EXT_texture_storage

-#define GL_EXT_texture_storage 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glTexStorage1DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);

-GL_APICALL void GL_APIENTRY glTexStorage2DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void GL_APIENTRY glTexStorage3DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);

-GL_APICALL void GL_APIENTRY glTextureStorage1DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);

-GL_APICALL void GL_APIENTRY glTextureStorage2DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void GL_APIENTRY glTextureStorage3DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);

-#endif

-typedef void (GL_APIENTRYP PFNGLTEXSTORAGE1DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);

-typedef void (GL_APIENTRYP PFNGLTEXSTORAGE2DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLTEXSTORAGE3DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);

-typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE1DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);

-typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE2DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE3DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);

-#endif

-

-/* GL_EXT_texture_type_2_10_10_10_REV */

-#ifndef GL_EXT_texture_type_2_10_10_10_REV

-#define GL_EXT_texture_type_2_10_10_10_REV 1

-#endif

-

-/* GL_EXT_unpack_subimage */

-#ifndef GL_EXT_unpack_subimage

-#define GL_EXT_unpack_subimage 1

-#endif

-

-/*------------------------------------------------------------------------*

- * DMP extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_DMP_shader_binary */

-#ifndef GL_DMP_shader_binary

-#define GL_DMP_shader_binary 1

-#endif

-

-/*------------------------------------------------------------------------*

- * FJ extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_FJ_shader_binary_GCCSO */

-#ifndef GL_FJ_shader_binary_GCCSO

-#define GL_FJ_shader_binary_GCCSO 1

-#endif

-

-/*------------------------------------------------------------------------*

- * IMG extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_IMG_program_binary */

-#ifndef GL_IMG_program_binary

-#define GL_IMG_program_binary 1

-#endif

-

-/* GL_IMG_read_format */

-#ifndef GL_IMG_read_format

-#define GL_IMG_read_format 1

-#endif

-

-/* GL_IMG_shader_binary */

-#ifndef GL_IMG_shader_binary

-#define GL_IMG_shader_binary 1

-#endif

-

-/* GL_IMG_texture_compression_pvrtc */

-#ifndef GL_IMG_texture_compression_pvrtc

-#define GL_IMG_texture_compression_pvrtc 1

-#endif

-

-/* GL_IMG_texture_compression_pvrtc2 */

-#ifndef GL_IMG_texture_compression_pvrtc2

-#define GL_IMG_texture_compression_pvrtc2 1

-#endif

-

-/* GL_IMG_multisampled_render_to_texture */

-#ifndef GL_IMG_multisampled_render_to_texture

-#define GL_IMG_multisampled_render_to_texture 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleIMG (GLenum, GLsizei, GLenum, GLsizei, GLsizei);

-GL_APICALL void GL_APIENTRY glFramebufferTexture2DMultisampleIMG (GLenum, GLenum, GLenum, GLuint, GLint, GLsizei);

-#endif

-typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEIMGPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DMULTISAMPLEIMGPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLsizei samples);

-#endif

-

-/*------------------------------------------------------------------------*

- * NV extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_NV_coverage_sample */

-#ifndef GL_NV_coverage_sample

-#define GL_NV_coverage_sample 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glCoverageMaskNV (GLboolean mask);

-GL_APICALL void GL_APIENTRY glCoverageOperationNV (GLenum operation);

-#endif

-typedef void (GL_APIENTRYP PFNGLCOVERAGEMASKNVPROC) (GLboolean mask);

-typedef void (GL_APIENTRYP PFNGLCOVERAGEOPERATIONNVPROC) (GLenum operation);

-#endif

-

-/* GL_NV_depth_nonlinear */

-#ifndef GL_NV_depth_nonlinear

-#define GL_NV_depth_nonlinear 1

-#endif

-

-/* GL_NV_draw_buffers */

-#ifndef GL_NV_draw_buffers

-#define GL_NV_draw_buffers 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDrawBuffersNV (GLsizei n, const GLenum *bufs);

-#endif

-typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSNVPROC) (GLsizei n, const GLenum *bufs);

-#endif

-

-/* GL_EXT_draw_buffers */

-#ifndef GL_EXT_draw_buffers

-#define GL_EXT_draw_buffers 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDrawBuffersEXT (GLsizei n, const GLenum *bufs);

-#endif

-typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSEXTPROC) (GLsizei n, const GLenum *bufs);

-#endif

-

-/* GL_NV_draw_instanced */

-#ifndef GL_NV_draw_instanced

-#define GL_NV_draw_instanced 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDrawArraysInstancedNV (GLenum mode, GLint first, GLsizei count, GLsizei primcount);

-GL_APICALL void GL_APIENTRY glDrawElementsInstancedNV (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount);

-#endif

-typedef void (GL_APIENTRYP PFNDRAWARRAYSINSTANCEDNVPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount);

-typedef void (GL_APIENTRYP PFNDRAWELEMENTSINSTANCEDNVPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount);

-#endif

-

-/* GL_NV_fbo_color_attachments */

-#ifndef GL_NV_fbo_color_attachments

-#define GL_NV_fbo_color_attachments 1

-#endif

-

-/* GL_NV_fence */

-#ifndef GL_NV_fence

-#define GL_NV_fence 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glDeleteFencesNV (GLsizei, const GLuint *);

-GL_APICALL void GL_APIENTRY glGenFencesNV (GLsizei, GLuint *);

-GL_APICALL GLboolean GL_APIENTRY glIsFenceNV (GLuint);

-GL_APICALL GLboolean GL_APIENTRY glTestFenceNV (GLuint);

-GL_APICALL void GL_APIENTRY glGetFenceivNV (GLuint, GLenum, GLint *);

-GL_APICALL void GL_APIENTRY glFinishFenceNV (GLuint);

-GL_APICALL void GL_APIENTRY glSetFenceNV (GLuint, GLenum);

-#endif

-typedef void (GL_APIENTRYP PFNGLDELETEFENCESNVPROC) (GLsizei n, const GLuint *fences);

-typedef void (GL_APIENTRYP PFNGLGENFENCESNVPROC) (GLsizei n, GLuint *fences);

-typedef GLboolean (GL_APIENTRYP PFNGLISFENCENVPROC) (GLuint fence);

-typedef GLboolean (GL_APIENTRYP PFNGLTESTFENCENVPROC) (GLuint fence);

-typedef void (GL_APIENTRYP PFNGLGETFENCEIVNVPROC) (GLuint fence, GLenum pname, GLint *params);

-typedef void (GL_APIENTRYP PFNGLFINISHFENCENVPROC) (GLuint fence);

-typedef void (GL_APIENTRYP PFNGLSETFENCENVPROC) (GLuint fence, GLenum condition);

-#endif

-

-/* GL_NV_framebuffer_blit */

-#ifndef GL_NV_framebuffer_blit

-#define GL_NV_framebuffer_blit 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glBlitFramebufferNV (int srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);

-#endif

-typedef void (GL_APIENTRYP PFNBLITFRAMEBUFFERNVPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);

-#endif

-

-/* GL_NV_framebuffer_multisample */

-#ifndef GL_NV_framebuffer_multisample

-#define GL_NV_framebuffer_multisample 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleNV ( GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-#endif

-typedef void (GL_APIENTRYP PFNRENDERBUFFERSTORAGEMULTISAMPLENVPROC) ( GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-#endif

-

-/* GL_NV_generate_mipmap_sRGB */

-#ifndef GL_NV_generate_mipmap_sRGB

-#define GL_NV_generate_mipmap_sRGB 1

-#endif

-

-/* GL_NV_instanced_arrays */

-#ifndef GL_NV_instanced_arrays

-#define GL_NV_instanced_arrays 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glVertexAttribDivisorNV (GLuint index, GLuint divisor);

-#endif

-typedef void (GL_APIENTRYP PFNVERTEXATTRIBDIVISORNVPROC) (GLuint index, GLuint divisor);

-#endif

-

-/* GL_NV_read_buffer */

-#ifndef GL_NV_read_buffer

-#define GL_NV_read_buffer 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glReadBufferNV (GLenum mode);

-#endif

-typedef void (GL_APIENTRYP PFNGLREADBUFFERNVPROC) (GLenum mode);

-#endif

-

-/* GL_NV_read_buffer_front */

-#ifndef GL_NV_read_buffer_front

-#define GL_NV_read_buffer_front 1

-#endif

-

-/* GL_NV_read_depth */

-#ifndef GL_NV_read_depth

-#define GL_NV_read_depth 1

-#endif

-

-/* GL_NV_read_depth_stencil */

-#ifndef GL_NV_read_depth_stencil

-#define GL_NV_read_depth_stencil 1

-#endif

-

-/* GL_NV_read_stencil */

-#ifndef GL_NV_read_stencil

-#define GL_NV_read_stencil 1

-#endif

-

-/* GL_NV_shadow_samplers_array */

-#ifndef GL_NV_shadow_samplers_array

-#define GL_NV_shadow_samplers_array 1

-#endif

-

-/* GL_NV_shadow_samplers_cube */

-#ifndef GL_NV_shadow_samplers_cube

-#define GL_NV_shadow_samplers_cube 1

-#endif

-

-/* GL_NV_sRGB_formats */

-#ifndef GL_NV_sRGB_formats

-#define GL_NV_sRGB_formats 1

-#endif

-

-/* GL_NV_texture_border_clamp */

-#ifndef GL_NV_texture_border_clamp

-#define GL_NV_texture_border_clamp 1

-#endif

-

-/* GL_NV_texture_compression_s3tc_update */

-#ifndef GL_NV_texture_compression_s3tc_update

-#define GL_NV_texture_compression_s3tc_update 1

-#endif

-

-/* GL_NV_texture_npot_2D_mipmap */

-#ifndef GL_NV_texture_npot_2D_mipmap

-#define GL_NV_texture_npot_2D_mipmap 1

-#endif

-

-/*------------------------------------------------------------------------*

- * QCOM extension functions

- *------------------------------------------------------------------------*/

-

-/* GL_QCOM_alpha_test */

-#ifndef GL_QCOM_alpha_test

-#define GL_QCOM_alpha_test 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glAlphaFuncQCOM (GLenum func, GLclampf ref);

-#endif

-typedef void (GL_APIENTRYP PFNGLALPHAFUNCQCOMPROC) (GLenum func, GLclampf ref);

-#endif

-

-/* GL_QCOM_binning_control */

-#ifndef GL_QCOM_binning_control

-#define GL_QCOM_binning_control 1

-#endif

-

-/* GL_QCOM_driver_control */

-#ifndef GL_QCOM_driver_control

-#define GL_QCOM_driver_control 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glGetDriverControlsQCOM (GLint *num, GLsizei size, GLuint *driverControls);

-GL_APICALL void GL_APIENTRY glGetDriverControlStringQCOM (GLuint driverControl, GLsizei bufSize, GLsizei *length, GLchar *driverControlString);

-GL_APICALL void GL_APIENTRY glEnableDriverControlQCOM (GLuint driverControl);

-GL_APICALL void GL_APIENTRY glDisableDriverControlQCOM (GLuint driverControl);

-#endif

-typedef void (GL_APIENTRYP PFNGLGETDRIVERCONTROLSQCOMPROC) (GLint *num, GLsizei size, GLuint *driverControls);

-typedef void (GL_APIENTRYP PFNGLGETDRIVERCONTROLSTRINGQCOMPROC) (GLuint driverControl, GLsizei bufSize, GLsizei *length, GLchar *driverControlString);

-typedef void (GL_APIENTRYP PFNGLENABLEDRIVERCONTROLQCOMPROC) (GLuint driverControl);

-typedef void (GL_APIENTRYP PFNGLDISABLEDRIVERCONTROLQCOMPROC) (GLuint driverControl);

-#endif

-

-/* GL_QCOM_extended_get */

-#ifndef GL_QCOM_extended_get

-#define GL_QCOM_extended_get 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glExtGetTexturesQCOM (GLuint *textures, GLint maxTextures, GLint *numTextures);

-GL_APICALL void GL_APIENTRY glExtGetBuffersQCOM (GLuint *buffers, GLint maxBuffers, GLint *numBuffers);

-GL_APICALL void GL_APIENTRY glExtGetRenderbuffersQCOM (GLuint *renderbuffers, GLint maxRenderbuffers, GLint *numRenderbuffers);

-GL_APICALL void GL_APIENTRY glExtGetFramebuffersQCOM (GLuint *framebuffers, GLint maxFramebuffers, GLint *numFramebuffers);

-GL_APICALL void GL_APIENTRY glExtGetTexLevelParameterivQCOM (GLuint texture, GLenum face, GLint level, GLenum pname, GLint *params);

-GL_APICALL void GL_APIENTRY glExtTexObjectStateOverrideiQCOM (GLenum target, GLenum pname, GLint param);

-GL_APICALL void GL_APIENTRY glExtGetTexSubImageQCOM (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, GLvoid *texels);

-GL_APICALL void GL_APIENTRY glExtGetBufferPointervQCOM (GLenum target, GLvoid **params);

-#endif

-typedef void (GL_APIENTRYP PFNGLEXTGETTEXTURESQCOMPROC) (GLuint *textures, GLint maxTextures, GLint *numTextures);

-typedef void (GL_APIENTRYP PFNGLEXTGETBUFFERSQCOMPROC) (GLuint *buffers, GLint maxBuffers, GLint *numBuffers);

-typedef void (GL_APIENTRYP PFNGLEXTGETRENDERBUFFERSQCOMPROC) (GLuint *renderbuffers, GLint maxRenderbuffers, GLint *numRenderbuffers);

-typedef void (GL_APIENTRYP PFNGLEXTGETFRAMEBUFFERSQCOMPROC) (GLuint *framebuffers, GLint maxFramebuffers, GLint *numFramebuffers);

-typedef void (GL_APIENTRYP PFNGLEXTGETTEXLEVELPARAMETERIVQCOMPROC) (GLuint texture, GLenum face, GLint level, GLenum pname, GLint *params);

-typedef void (GL_APIENTRYP PFNGLEXTTEXOBJECTSTATEOVERRIDEIQCOMPROC) (GLenum target, GLenum pname, GLint param);

-typedef void (GL_APIENTRYP PFNGLEXTGETTEXSUBIMAGEQCOMPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, GLvoid *texels);

-typedef void (GL_APIENTRYP PFNGLEXTGETBUFFERPOINTERVQCOMPROC) (GLenum target, GLvoid **params);

-#endif

-

-/* GL_QCOM_extended_get2 */

-#ifndef GL_QCOM_extended_get2

-#define GL_QCOM_extended_get2 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glExtGetShadersQCOM (GLuint *shaders, GLint maxShaders, GLint *numShaders);

-GL_APICALL void GL_APIENTRY glExtGetProgramsQCOM (GLuint *programs, GLint maxPrograms, GLint *numPrograms);

-GL_APICALL GLboolean GL_APIENTRY glExtIsProgramBinaryQCOM (GLuint program);

-GL_APICALL void GL_APIENTRY glExtGetProgramBinarySourceQCOM (GLuint program, GLenum shadertype, GLchar *source, GLint *length);

-#endif

-typedef void (GL_APIENTRYP PFNGLEXTGETSHADERSQCOMPROC) (GLuint *shaders, GLint maxShaders, GLint *numShaders);

-typedef void (GL_APIENTRYP PFNGLEXTGETPROGRAMSQCOMPROC) (GLuint *programs, GLint maxPrograms, GLint *numPrograms);

-typedef GLboolean (GL_APIENTRYP PFNGLEXTISPROGRAMBINARYQCOMPROC) (GLuint program);

-typedef void (GL_APIENTRYP PFNGLEXTGETPROGRAMBINARYSOURCEQCOMPROC) (GLuint program, GLenum shadertype, GLchar *source, GLint *length);

-#endif

-

-/* GL_QCOM_perfmon_global_mode */

-#ifndef GL_QCOM_perfmon_global_mode

-#define GL_QCOM_perfmon_global_mode 1

-#endif

-

-/* GL_QCOM_writeonly_rendering */

-#ifndef GL_QCOM_writeonly_rendering

-#define GL_QCOM_writeonly_rendering 1

-#endif

-

-/* GL_QCOM_tiled_rendering */

-#ifndef GL_QCOM_tiled_rendering

-#define GL_QCOM_tiled_rendering 1

-#ifdef GL_GLEXT_PROTOTYPES

-GL_APICALL void GL_APIENTRY glStartTilingQCOM (GLuint x, GLuint y, GLuint width, GLuint height, GLbitfield preserveMask);

-GL_APICALL void GL_APIENTRY glEndTilingQCOM (GLbitfield preserveMask);

-#endif

-typedef void (GL_APIENTRYP PFNGLSTARTTILINGQCOMPROC) (GLuint x, GLuint y, GLuint width, GLuint height, GLbitfield preserveMask);

-typedef void (GL_APIENTRYP PFNGLENDTILINGQCOMPROC) (GLbitfield preserveMask);

-#endif

-

-/*------------------------------------------------------------------------*

- * VIV extension tokens

- *------------------------------------------------------------------------*/

-

-/* GL_VIV_shader_binary */

-#ifndef GL_VIV_shader_binary

-#define GL_VIV_shader_binary 1

-#endif

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif /* __gl2ext_h_ */

+#ifndef __gl2ext_h_
+#define __gl2ext_h_
+
+/* $Revision: 20795 $ on $Date:: 2013-03-07 01:01:58 -0800 #$ */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * This document is licensed under the SGI Free Software B License Version
+ * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .
+ */
+
+#ifndef GL_APIENTRYP
+#   define GL_APIENTRYP GL_APIENTRY*
+#endif
+
+/*------------------------------------------------------------------------*
+ * OES extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_OES_compressed_ETC1_RGB8_texture */
+#ifndef GL_OES_compressed_ETC1_RGB8_texture
+#define GL_ETC1_RGB8_OES                                        0x8D64
+#endif
+
+/* GL_OES_compressed_paletted_texture */
+#ifndef GL_OES_compressed_paletted_texture
+#define GL_PALETTE4_RGB8_OES                                    0x8B90
+#define GL_PALETTE4_RGBA8_OES                                   0x8B91
+#define GL_PALETTE4_R5_G6_B5_OES                                0x8B92
+#define GL_PALETTE4_RGBA4_OES                                   0x8B93
+#define GL_PALETTE4_RGB5_A1_OES                                 0x8B94
+#define GL_PALETTE8_RGB8_OES                                    0x8B95
+#define GL_PALETTE8_RGBA8_OES                                   0x8B96
+#define GL_PALETTE8_R5_G6_B5_OES                                0x8B97
+#define GL_PALETTE8_RGBA4_OES                                   0x8B98
+#define GL_PALETTE8_RGB5_A1_OES                                 0x8B99
+#endif
+
+/* GL_OES_depth24 */
+#ifndef GL_OES_depth24
+#define GL_DEPTH_COMPONENT24_OES                                0x81A6
+#endif
+
+/* GL_OES_depth32 */
+#ifndef GL_OES_depth32
+#define GL_DEPTH_COMPONENT32_OES                                0x81A7
+#endif
+
+/* GL_OES_depth_texture */
+/* No new tokens introduced by this extension. */
+
+/* GL_OES_EGL_image */
+#ifndef GL_OES_EGL_image
+typedef void* GLeglImageOES;
+#endif
+
+/* GL_OES_EGL_image_external */
+#ifndef GL_OES_EGL_image_external
+/* GLeglImageOES defined in GL_OES_EGL_image already. */
+#define GL_TEXTURE_EXTERNAL_OES                                 0x8D65
+#define GL_SAMPLER_EXTERNAL_OES                                 0x8D66
+#define GL_TEXTURE_BINDING_EXTERNAL_OES                         0x8D67
+#define GL_REQUIRED_TEXTURE_IMAGE_UNITS_OES                     0x8D68
+#endif
+
+/* GL_OES_element_index_uint */
+#ifndef GL_OES_element_index_uint
+#define GL_UNSIGNED_INT                                         0x1405
+#endif
+
+/* GL_OES_get_program_binary */
+#ifndef GL_OES_get_program_binary
+#define GL_PROGRAM_BINARY_LENGTH_OES                            0x8741
+#define GL_NUM_PROGRAM_BINARY_FORMATS_OES                       0x87FE
+#define GL_PROGRAM_BINARY_FORMATS_OES                           0x87FF
+#endif
+
+/* GL_OES_mapbuffer */
+#ifndef GL_OES_mapbuffer
+#define GL_WRITE_ONLY_OES                                       0x88B9
+#define GL_BUFFER_ACCESS_OES                                    0x88BB
+#define GL_BUFFER_MAPPED_OES                                    0x88BC
+#define GL_BUFFER_MAP_POINTER_OES                               0x88BD
+#endif
+
+/* GL_OES_packed_depth_stencil */
+#ifndef GL_OES_packed_depth_stencil
+#define GL_DEPTH_STENCIL_OES                                    0x84F9
+#define GL_UNSIGNED_INT_24_8_OES                                0x84FA
+#define GL_DEPTH24_STENCIL8_OES                                 0x88F0
+#endif
+
+/* GL_OES_required_internalformat */
+#ifndef GL_OES_required_internalformat 
+#define GL_ALPHA8_OES                                           0x803C
+#define GL_DEPTH_COMPONENT16_OES                                0x81A5
+/* reuse GL_DEPTH_COMPONENT24_OES */                            
+/* reuse GL_DEPTH24_STENCIL8_OES */                             
+/* reuse GL_DEPTH_COMPONENT32_OES */                            
+#define GL_LUMINANCE4_ALPHA4_OES                                0x8043
+#define GL_LUMINANCE8_ALPHA8_OES                                0x8045
+#define GL_LUMINANCE8_OES                                       0x8040
+#define GL_RGBA4_OES                                            0x8056
+#define GL_RGB5_A1_OES                                          0x8057
+#define GL_RGB565_OES                                           0x8D62
+/* reuse GL_RGB8_OES */                              
+/* reuse GL_RGBA8_OES */  
+/* reuse GL_RGB10_EXT */
+/* reuse GL_RGB10_A2_EXT */
+#endif 
+
+/* GL_OES_rgb8_rgba8 */
+#ifndef GL_OES_rgb8_rgba8
+#define GL_RGB8_OES                                             0x8051
+#define GL_RGBA8_OES                                            0x8058
+#endif
+
+/* GL_OES_standard_derivatives */
+#ifndef GL_OES_standard_derivatives
+#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT_OES                  0x8B8B
+#endif
+
+/* GL_OES_stencil1 */
+#ifndef GL_OES_stencil1
+#define GL_STENCIL_INDEX1_OES                                   0x8D46
+#endif
+
+/* GL_OES_stencil4 */
+#ifndef GL_OES_stencil4
+#define GL_STENCIL_INDEX4_OES                                   0x8D47
+#endif
+
+#ifndef GL_OES_surfaceless_context
+#define GL_FRAMEBUFFER_UNDEFINED_OES                            0x8219
+#endif
+
+/* GL_OES_texture_3D */
+#ifndef GL_OES_texture_3D
+#define GL_TEXTURE_WRAP_R_OES                                   0x8072
+#define GL_TEXTURE_3D_OES                                       0x806F
+#define GL_TEXTURE_BINDING_3D_OES                               0x806A
+#define GL_MAX_3D_TEXTURE_SIZE_OES                              0x8073
+#define GL_SAMPLER_3D_OES                                       0x8B5F
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_OES        0x8CD4
+#endif
+
+/* GL_OES_texture_float */
+/* No new tokens introduced by this extension. */
+
+/* GL_OES_texture_float_linear */
+/* No new tokens introduced by this extension. */
+
+/* GL_OES_texture_half_float */
+#ifndef GL_OES_texture_half_float
+#define GL_HALF_FLOAT_OES                                       0x8D61
+#endif
+
+/* GL_OES_texture_half_float_linear */
+/* No new tokens introduced by this extension. */
+
+/* GL_OES_texture_npot */
+/* No new tokens introduced by this extension. */
+
+/* GL_OES_vertex_array_object */
+#ifndef GL_OES_vertex_array_object
+#define GL_VERTEX_ARRAY_BINDING_OES                             0x85B5
+#endif
+
+/* GL_OES_vertex_half_float */
+/* GL_HALF_FLOAT_OES defined in GL_OES_texture_half_float already. */
+
+/* GL_OES_vertex_type_10_10_10_2 */
+#ifndef GL_OES_vertex_type_10_10_10_2
+#define GL_UNSIGNED_INT_10_10_10_2_OES                          0x8DF6
+#define GL_INT_10_10_10_2_OES                                   0x8DF7
+#endif
+
+/*------------------------------------------------------------------------*
+ * KHR extension tokens
+ *------------------------------------------------------------------------*/
+
+#ifndef GL_KHR_debug
+typedef void (GL_APIENTRYP GLDEBUGPROC)(GLenum source,GLenum type,GLuint id,GLenum severity,GLsizei length,const GLchar *message,GLvoid *userParam);
+#define GL_DEBUG_OUTPUT_SYNCHRONOUS                             0x8242
+#define GL_DEBUG_NEXT_LOGGED_MESSAGE_LENGTH                     0x8243
+#define GL_DEBUG_CALLBACK_FUNCTION                              0x8244
+#define GL_DEBUG_CALLBACK_USER_PARAM                            0x8245
+#define GL_DEBUG_SOURCE_API                                     0x8246
+#define GL_DEBUG_SOURCE_WINDOW_SYSTEM                           0x8247
+#define GL_DEBUG_SOURCE_SHADER_COMPILER                         0x8248
+#define GL_DEBUG_SOURCE_THIRD_PARTY                             0x8249
+#define GL_DEBUG_SOURCE_APPLICATION                             0x824A
+#define GL_DEBUG_SOURCE_OTHER                                   0x824B
+#define GL_DEBUG_TYPE_ERROR                                     0x824C
+#define GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR                       0x824D
+#define GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR                        0x824E
+#define GL_DEBUG_TYPE_PORTABILITY                               0x824F
+#define GL_DEBUG_TYPE_PERFORMANCE                               0x8250
+#define GL_DEBUG_TYPE_OTHER                                     0x8251
+#define GL_DEBUG_TYPE_MARKER                                    0x8268
+#define GL_DEBUG_TYPE_PUSH_GROUP                                0x8269
+#define GL_DEBUG_TYPE_POP_GROUP                                 0x826A
+#define GL_DEBUG_SEVERITY_NOTIFICATION                          0x826B
+#define GL_MAX_DEBUG_GROUP_STACK_DEPTH                          0x826C
+#define GL_DEBUG_GROUP_STACK_DEPTH                              0x826D
+#define GL_BUFFER                                               0x82E0
+#define GL_SHADER                                               0x82E1
+#define GL_PROGRAM                                              0x82E2
+#define GL_QUERY                                                0x82E3
+/* PROGRAM_PIPELINE only in GL */                               
+#define GL_SAMPLER                                              0x82E6
+/* DISPLAY_LIST only in GL */                                   
+#define GL_MAX_LABEL_LENGTH                                     0x82E8
+#define GL_MAX_DEBUG_MESSAGE_LENGTH                             0x9143
+#define GL_MAX_DEBUG_LOGGED_MESSAGES                            0x9144
+#define GL_DEBUG_LOGGED_MESSAGES                                0x9145
+#define GL_DEBUG_SEVERITY_HIGH                                  0x9146
+#define GL_DEBUG_SEVERITY_MEDIUM                                0x9147
+#define GL_DEBUG_SEVERITY_LOW                                   0x9148
+#define GL_DEBUG_OUTPUT                                         0x92E0
+#define GL_CONTEXT_FLAG_DEBUG_BIT                               0x00000002
+#define GL_STACK_OVERFLOW                                       0x0503
+#define GL_STACK_UNDERFLOW                                      0x0504
+#endif
+
+#ifndef GL_KHR_texture_compression_astc_ldr
+#define GL_COMPRESSED_RGBA_ASTC_4x4_KHR                         0x93B0
+#define GL_COMPRESSED_RGBA_ASTC_5x4_KHR                         0x93B1
+#define GL_COMPRESSED_RGBA_ASTC_5x5_KHR                         0x93B2
+#define GL_COMPRESSED_RGBA_ASTC_6x5_KHR                         0x93B3
+#define GL_COMPRESSED_RGBA_ASTC_6x6_KHR                         0x93B4
+#define GL_COMPRESSED_RGBA_ASTC_8x5_KHR                         0x93B5
+#define GL_COMPRESSED_RGBA_ASTC_8x6_KHR                         0x93B6
+#define GL_COMPRESSED_RGBA_ASTC_8x8_KHR                         0x93B7
+#define GL_COMPRESSED_RGBA_ASTC_10x5_KHR                        0x93B8
+#define GL_COMPRESSED_RGBA_ASTC_10x6_KHR                        0x93B9
+#define GL_COMPRESSED_RGBA_ASTC_10x8_KHR                        0x93BA
+#define GL_COMPRESSED_RGBA_ASTC_10x10_KHR                       0x93BB
+#define GL_COMPRESSED_RGBA_ASTC_12x10_KHR                       0x93BC
+#define GL_COMPRESSED_RGBA_ASTC_12x12_KHR                       0x93BD
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR                 0x93D0
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR                 0x93D1
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR                 0x93D2
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR                 0x93D3
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR                 0x93D4
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR                 0x93D5
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR                 0x93D6
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR                 0x93D7
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x5_KHR                0x93D8
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x6_KHR                0x93D9
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR                0x93DA
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR               0x93DB
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x10_KHR               0x93DC
+#define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR               0x93DD
+#endif
+
+/*------------------------------------------------------------------------*
+ * AMD extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_AMD_compressed_3DC_texture */
+#ifndef GL_AMD_compressed_3DC_texture
+#define GL_3DC_X_AMD                                            0x87F9
+#define GL_3DC_XY_AMD                                           0x87FA
+#endif
+
+/* GL_AMD_compressed_ATC_texture */
+#ifndef GL_AMD_compressed_ATC_texture
+#define GL_ATC_RGB_AMD                                          0x8C92
+#define GL_ATC_RGBA_EXPLICIT_ALPHA_AMD                          0x8C93
+#define GL_ATC_RGBA_INTERPOLATED_ALPHA_AMD                      0x87EE
+#endif
+
+/* GL_AMD_performance_monitor */
+#ifndef GL_AMD_performance_monitor
+#define GL_COUNTER_TYPE_AMD                                     0x8BC0
+#define GL_COUNTER_RANGE_AMD                                    0x8BC1
+#define GL_UNSIGNED_INT64_AMD                                   0x8BC2
+#define GL_PERCENTAGE_AMD                                       0x8BC3
+#define GL_PERFMON_RESULT_AVAILABLE_AMD                         0x8BC4
+#define GL_PERFMON_RESULT_SIZE_AMD                              0x8BC5
+#define GL_PERFMON_RESULT_AMD                                   0x8BC6
+#endif
+
+/* GL_AMD_program_binary_Z400 */
+#ifndef GL_AMD_program_binary_Z400
+#define GL_Z400_BINARY_AMD                                      0x8740
+#endif
+
+/*------------------------------------------------------------------------*
+ * ANGLE extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_ANGLE_depth_texture */
+#ifndef GL_ANGLE_depth_texture
+#define GL_DEPTH_COMPONENT                                      0x1902
+#define GL_DEPTH_STENCIL_OES                                    0x84F9
+#define GL_UNSIGNED_SHORT                                       0x1403
+#define GL_UNSIGNED_INT                                         0x1405
+#define GL_UNSIGNED_INT_24_8_OES                                0x84FA
+#define GL_DEPTH_COMPONENT16                                    0x81A5
+#define GL_DEPTH_COMPONENT32_OES                                0x81A7
+#define GL_DEPTH24_STENCIL8_OES                                 0x88F0
+#endif
+
+/* GL_ANGLE_framebuffer_blit */
+#ifndef GL_ANGLE_framebuffer_blit
+#define GL_READ_FRAMEBUFFER_ANGLE                               0x8CA8
+#define GL_DRAW_FRAMEBUFFER_ANGLE                               0x8CA9
+#define GL_DRAW_FRAMEBUFFER_BINDING_ANGLE                       0x8CA6
+#define GL_READ_FRAMEBUFFER_BINDING_ANGLE                       0x8CAA
+#endif
+
+/* GL_ANGLE_framebuffer_multisample */
+#ifndef GL_ANGLE_framebuffer_multisample
+#define GL_RENDERBUFFER_SAMPLES_ANGLE                           0x8CAB
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_ANGLE             0x8D56
+#define GL_MAX_SAMPLES_ANGLE                                    0x8D57
+#endif
+
+/* GL_ANGLE_instanced_arrays */
+#ifndef GL_ANGLE_instanced_arrays 
+#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR_ANGLE                    0x88FE
+#endif
+
+/* GL_ANGLE_pack_reverse_row_order */
+#ifndef GL_ANGLE_pack_reverse_row_order 
+#define GL_PACK_REVERSE_ROW_ORDER_ANGLE                         0x93A4
+#endif
+
+/* GL_ANGLE_program_binary */
+#ifndef GL_ANGLE_program_binary
+#define GL_PROGRAM_BINARY_ANGLE                                 0x93A6
+#endif
+
+/* GL_ANGLE_texture_compression_dxt3 */
+#ifndef GL_ANGLE_texture_compression_dxt3 
+#define GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE                      0x83F2
+#endif
+
+/* GL_ANGLE_texture_compression_dxt5 */
+#ifndef GL_ANGLE_texture_compression_dxt5 
+#define GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE                      0x83F3
+#endif
+
+/* GL_ANGLE_texture_usage */
+#ifndef GL_ANGLE_texture_usage 
+#define GL_TEXTURE_USAGE_ANGLE                                  0x93A2
+#define GL_FRAMEBUFFER_ATTACHMENT_ANGLE                         0x93A3
+#endif
+
+/* GL_ANGLE_translated_shader_source */
+#ifndef GL_ANGLE_translated_shader_source 
+#define GL_TRANSLATED_SHADER_SOURCE_LENGTH_ANGLE                0x93A0
+#endif
+
+/*------------------------------------------------------------------------*
+ * APPLE extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_APPLE_copy_texture_levels */
+/* No new tokens introduced by this extension. */
+    
+/* GL_APPLE_framebuffer_multisample */
+#ifndef GL_APPLE_framebuffer_multisample
+#define GL_RENDERBUFFER_SAMPLES_APPLE                           0x8CAB
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_APPLE             0x8D56
+#define GL_MAX_SAMPLES_APPLE                                    0x8D57
+#define GL_READ_FRAMEBUFFER_APPLE                               0x8CA8
+#define GL_DRAW_FRAMEBUFFER_APPLE                               0x8CA9
+#define GL_DRAW_FRAMEBUFFER_BINDING_APPLE                       0x8CA6
+#define GL_READ_FRAMEBUFFER_BINDING_APPLE                       0x8CAA
+#endif
+
+/* GL_APPLE_rgb_422 */
+#ifndef GL_APPLE_rgb_422
+#define GL_RGB_422_APPLE                                        0x8A1F
+#define GL_UNSIGNED_SHORT_8_8_APPLE                             0x85BA
+#define GL_UNSIGNED_SHORT_8_8_REV_APPLE                         0x85BB
+#endif
+
+/* GL_APPLE_sync */
+#ifndef GL_APPLE_sync
+
+#ifndef __gl3_h_
+/* These types are defined with reference to <inttypes.h>
+ * in the Apple extension spec, but here we use the Khronos
+ * portable types in khrplatform.h, and assume those types 
+ * are always defined.
+ * If any other extensions using these types are defined, 
+ * the typedefs must move out of this block and be shared.
+ */
+typedef khronos_int64_t GLint64;
+typedef khronos_uint64_t GLuint64;
+typedef struct __GLsync *GLsync;
+#endif
+
+#define GL_SYNC_OBJECT_APPLE                                    0x8A53
+#define GL_MAX_SERVER_WAIT_TIMEOUT_APPLE                        0x9111
+#define GL_OBJECT_TYPE_APPLE                                    0x9112
+#define GL_SYNC_CONDITION_APPLE                                 0x9113
+#define GL_SYNC_STATUS_APPLE                                    0x9114
+#define GL_SYNC_FLAGS_APPLE                                     0x9115
+#define GL_SYNC_FENCE_APPLE                                     0x9116
+#define GL_SYNC_GPU_COMMANDS_COMPLETE_APPLE                     0x9117
+#define GL_UNSIGNALED_APPLE                                     0x9118
+#define GL_SIGNALED_APPLE                                       0x9119
+#define GL_ALREADY_SIGNALED_APPLE                               0x911A
+#define GL_TIMEOUT_EXPIRED_APPLE                                0x911B
+#define GL_CONDITION_SATISFIED_APPLE                            0x911C
+#define GL_WAIT_FAILED_APPLE                                    0x911D
+#define GL_SYNC_FLUSH_COMMANDS_BIT_APPLE                        0x00000001
+#define GL_TIMEOUT_IGNORED_APPLE                                0xFFFFFFFFFFFFFFFFull
+#endif
+
+/* GL_APPLE_texture_format_BGRA8888 */
+#ifndef GL_APPLE_texture_format_BGRA8888
+#define GL_BGRA_EXT                                             0x80E1
+#endif
+
+/* GL_APPLE_texture_max_level */
+#ifndef GL_APPLE_texture_max_level
+#define GL_TEXTURE_MAX_LEVEL_APPLE                              0x813D
+#endif
+
+/*------------------------------------------------------------------------*
+ * ARM extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_ARM_mali_program_binary */
+#ifndef GL_ARM_mali_program_binary
+#define GL_MALI_PROGRAM_BINARY_ARM                              0x8F61
+#endif
+
+/* GL_ARM_mali_shader_binary */
+#ifndef GL_ARM_mali_shader_binary
+#define GL_MALI_SHADER_BINARY_ARM                               0x8F60
+#endif
+
+/* GL_ARM_rgba8 */
+/* No new tokens introduced by this extension. */
+
+/*------------------------------------------------------------------------*
+ * EXT extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_EXT_blend_minmax */
+#ifndef GL_EXT_blend_minmax
+#define GL_MIN_EXT                                              0x8007
+#define GL_MAX_EXT                                              0x8008
+#endif
+
+/* GL_EXT_color_buffer_half_float */
+#ifndef GL_EXT_color_buffer_half_float
+#define GL_RGBA16F_EXT                                          0x881A
+#define GL_RGB16F_EXT                                           0x881B
+#define GL_RG16F_EXT                                            0x822F
+#define GL_R16F_EXT                                             0x822D
+#define GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE_EXT            0x8211
+#define GL_UNSIGNED_NORMALIZED_EXT                              0x8C17
+#endif
+
+/* GL_EXT_debug_label */
+#ifndef GL_EXT_debug_label
+#define GL_PROGRAM_PIPELINE_OBJECT_EXT                          0x8A4F
+#define GL_PROGRAM_OBJECT_EXT                                   0x8B40
+#define GL_SHADER_OBJECT_EXT                                    0x8B48
+#define GL_BUFFER_OBJECT_EXT                                    0x9151
+#define GL_QUERY_OBJECT_EXT                                     0x9153
+#define GL_VERTEX_ARRAY_OBJECT_EXT                              0x9154
+#endif
+
+/* GL_EXT_debug_marker */
+/* No new tokens introduced by this extension. */
+
+/* GL_EXT_discard_framebuffer */
+#ifndef GL_EXT_discard_framebuffer
+#define GL_COLOR_EXT                                            0x1800
+#define GL_DEPTH_EXT                                            0x1801
+#define GL_STENCIL_EXT                                          0x1802
+#endif
+
+/* GL_EXT_map_buffer_range */
+#ifndef GL_EXT_map_buffer_range
+#define GL_MAP_READ_BIT_EXT                                     0x0001
+#define GL_MAP_WRITE_BIT_EXT                                    0x0002
+#define GL_MAP_INVALIDATE_RANGE_BIT_EXT                         0x0004
+#define GL_MAP_INVALIDATE_BUFFER_BIT_EXT                        0x0008
+#define GL_MAP_FLUSH_EXPLICIT_BIT_EXT                           0x0010
+#define GL_MAP_UNSYNCHRONIZED_BIT_EXT                           0x0020
+#endif
+
+/* GL_EXT_multisampled_render_to_texture */
+#ifndef GL_EXT_multisampled_render_to_texture
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_SAMPLES_EXT           0x8D6C
+/* reuse values from GL_EXT_framebuffer_multisample (desktop extension) */ 
+#define GL_RENDERBUFFER_SAMPLES_EXT                             0x8CAB
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT               0x8D56
+#define GL_MAX_SAMPLES_EXT                                      0x8D57
+#endif
+
+/* GL_EXT_multiview_draw_buffers */
+#ifndef GL_EXT_multiview_draw_buffers
+#define GL_COLOR_ATTACHMENT_EXT                                 0x90F0
+#define GL_MULTIVIEW_EXT                                        0x90F1
+#define GL_DRAW_BUFFER_EXT                                      0x0C01
+#define GL_READ_BUFFER_EXT                                      0x0C02
+#define GL_MAX_MULTIVIEW_BUFFERS_EXT                            0x90F2
+#endif
+
+/* GL_EXT_multi_draw_arrays */
+/* No new tokens introduced by this extension. */
+
+/* GL_EXT_occlusion_query_boolean */
+#ifndef GL_EXT_occlusion_query_boolean
+#define GL_ANY_SAMPLES_PASSED_EXT                               0x8C2F
+#define GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT                  0x8D6A
+#define GL_CURRENT_QUERY_EXT                                    0x8865
+#define GL_QUERY_RESULT_EXT                                     0x8866
+#define GL_QUERY_RESULT_AVAILABLE_EXT                           0x8867
+#endif
+
+/* GL_EXT_read_format_bgra */
+#ifndef GL_EXT_read_format_bgra
+#define GL_BGRA_EXT                                             0x80E1
+#define GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT                       0x8365
+#define GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT                       0x8366
+#endif
+
+/* GL_EXT_robustness */
+#ifndef GL_EXT_robustness
+/* reuse GL_NO_ERROR */
+#define GL_GUILTY_CONTEXT_RESET_EXT                             0x8253
+#define GL_INNOCENT_CONTEXT_RESET_EXT                           0x8254
+#define GL_UNKNOWN_CONTEXT_RESET_EXT                            0x8255
+#define GL_CONTEXT_ROBUST_ACCESS_EXT                            0x90F3
+#define GL_RESET_NOTIFICATION_STRATEGY_EXT                      0x8256
+#define GL_LOSE_CONTEXT_ON_RESET_EXT                            0x8252
+#define GL_NO_RESET_NOTIFICATION_EXT                            0x8261
+#endif
+
+/* GL_EXT_separate_shader_objects */
+#ifndef GL_EXT_separate_shader_objects
+#define GL_VERTEX_SHADER_BIT_EXT                                0x00000001
+#define GL_FRAGMENT_SHADER_BIT_EXT                              0x00000002
+#define GL_ALL_SHADER_BITS_EXT                                  0xFFFFFFFF
+#define GL_PROGRAM_SEPARABLE_EXT                                0x8258
+#define GL_ACTIVE_PROGRAM_EXT                                   0x8259
+#define GL_PROGRAM_PIPELINE_BINDING_EXT                         0x825A
+#endif
+
+/* GL_EXT_shader_framebuffer_fetch */
+#ifndef GL_EXT_shader_framebuffer_fetch
+#define GL_FRAGMENT_SHADER_DISCARDS_SAMPLES_EXT                 0x8A52
+#endif
+
+/* GL_EXT_shader_texture_lod */
+/* No new tokens introduced by this extension. */
+
+/* GL_EXT_shadow_samplers */
+#ifndef GL_EXT_shadow_samplers
+#define GL_TEXTURE_COMPARE_MODE_EXT                             0x884C
+#define GL_TEXTURE_COMPARE_FUNC_EXT                             0x884D
+#define GL_COMPARE_REF_TO_TEXTURE_EXT                           0x884E
+#define GL_SAMPLER_2D_SHADOW_EXT                                0x8B62
+#endif
+
+/* GL_EXT_sRGB */
+#ifndef GL_EXT_sRGB
+#define GL_SRGB_EXT                                             0x8C40
+#define GL_SRGB_ALPHA_EXT                                       0x8C42
+#define GL_SRGB8_ALPHA8_EXT                                     0x8C43
+#define GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING_EXT            0x8210
+#endif
+
+/* GL_EXT_texture_compression_dxt1 */
+#ifndef GL_EXT_texture_compression_dxt1
+#define GL_COMPRESSED_RGB_S3TC_DXT1_EXT                         0x83F0
+#define GL_COMPRESSED_RGBA_S3TC_DXT1_EXT                        0x83F1
+#endif
+
+/* GL_EXT_texture_filter_anisotropic */
+#ifndef GL_EXT_texture_filter_anisotropic
+#define GL_TEXTURE_MAX_ANISOTROPY_EXT                           0x84FE
+#define GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT                       0x84FF
+#endif
+
+/* GL_EXT_texture_format_BGRA8888 */
+#ifndef GL_EXT_texture_format_BGRA8888
+#define GL_BGRA_EXT                                             0x80E1
+#endif
+
+/* GL_EXT_texture_rg */
+#ifndef GL_EXT_texture_rg
+#define GL_RED_EXT                                              0x1903
+#define GL_RG_EXT                                               0x8227
+#define GL_R8_EXT                                               0x8229
+#define GL_RG8_EXT                                              0x822B
+#endif
+
+/* GL_EXT_texture_storage */
+#ifndef GL_EXT_texture_storage
+#define GL_TEXTURE_IMMUTABLE_FORMAT_EXT                         0x912F
+#define GL_ALPHA8_EXT                                           0x803C  
+#define GL_LUMINANCE8_EXT                                       0x8040
+#define GL_LUMINANCE8_ALPHA8_EXT                                0x8045
+#define GL_RGBA32F_EXT                                          0x8814  
+#define GL_RGB32F_EXT                                           0x8815
+#define GL_ALPHA32F_EXT                                         0x8816
+#define GL_LUMINANCE32F_EXT                                     0x8818
+#define GL_LUMINANCE_ALPHA32F_EXT                               0x8819
+/* reuse GL_RGBA16F_EXT */
+/* reuse GL_RGB16F_EXT */
+#define GL_ALPHA16F_EXT                                         0x881C
+#define GL_LUMINANCE16F_EXT                                     0x881E
+#define GL_LUMINANCE_ALPHA16F_EXT                               0x881F
+#define GL_RGB10_A2_EXT                                         0x8059  
+#define GL_RGB10_EXT                                            0x8052
+#define GL_BGRA8_EXT                                            0x93A1
+#define GL_R8_EXT                                               0x8229
+#define GL_RG8_EXT                                              0x822B
+#define GL_R32F_EXT                                             0x822E  
+#define GL_RG32F_EXT                                            0x8230
+#define GL_R16F_EXT                                             0x822D
+#define GL_RG16F_EXT                                            0x822F
+#endif
+
+/* GL_EXT_texture_type_2_10_10_10_REV */
+#ifndef GL_EXT_texture_type_2_10_10_10_REV
+#define GL_UNSIGNED_INT_2_10_10_10_REV_EXT                      0x8368
+#endif
+
+/* GL_EXT_unpack_subimage */
+#ifndef GL_EXT_unpack_subimage
+#define GL_UNPACK_ROW_LENGTH_EXT                                0x0CF2
+#define GL_UNPACK_SKIP_ROWS_EXT                                 0x0CF3
+#define GL_UNPACK_SKIP_PIXELS_EXT                               0x0CF4
+#endif
+
+/*------------------------------------------------------------------------*
+ * DMP extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_DMP_shader_binary */
+#ifndef GL_DMP_shader_binary
+#define GL_SHADER_BINARY_DMP                                    0x9250
+#endif
+
+/*------------------------------------------------------------------------*
+ * FJ extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_FJ_shader_binary_GCCSO */
+#ifndef GL_FJ_shader_binary_GCCSO
+#define GL_GCCSO_SHADER_BINARY_F                                0x9260
+#endif
+
+/*------------------------------------------------------------------------*
+ * IMG extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_IMG_program_binary */
+#ifndef GL_IMG_program_binary
+#define GL_SGX_PROGRAM_BINARY_IMG                               0x9130
+#endif
+
+/* GL_IMG_read_format */
+#ifndef GL_IMG_read_format
+#define GL_BGRA_IMG                                             0x80E1
+#define GL_UNSIGNED_SHORT_4_4_4_4_REV_IMG                       0x8365
+#endif
+
+/* GL_IMG_shader_binary */
+#ifndef GL_IMG_shader_binary
+#define GL_SGX_BINARY_IMG                                       0x8C0A
+#endif
+
+/* GL_IMG_texture_compression_pvrtc */
+#ifndef GL_IMG_texture_compression_pvrtc
+#define GL_COMPRESSED_RGB_PVRTC_4BPPV1_IMG                      0x8C00
+#define GL_COMPRESSED_RGB_PVRTC_2BPPV1_IMG                      0x8C01
+#define GL_COMPRESSED_RGBA_PVRTC_4BPPV1_IMG                     0x8C02
+#define GL_COMPRESSED_RGBA_PVRTC_2BPPV1_IMG                     0x8C03
+#endif
+
+/* GL_IMG_texture_compression_pvrtc2 */
+#ifndef GL_IMG_texture_compression_pvrtc2
+#define GL_COMPRESSED_RGBA_PVRTC_2BPPV2_IMG                     0x9137
+#define GL_COMPRESSED_RGBA_PVRTC_4BPPV2_IMG                     0x9138
+#endif
+
+/* GL_IMG_multisampled_render_to_texture */
+#ifndef GL_IMG_multisampled_render_to_texture
+#define GL_RENDERBUFFER_SAMPLES_IMG                             0x9133
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_IMG               0x9134
+#define GL_MAX_SAMPLES_IMG                                      0x9135
+#define GL_TEXTURE_SAMPLES_IMG                                  0x9136
+#endif
+
+/*------------------------------------------------------------------------*
+ * NV extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_NV_coverage_sample */
+#ifndef GL_NV_coverage_sample
+#define GL_COVERAGE_COMPONENT_NV                                0x8ED0
+#define GL_COVERAGE_COMPONENT4_NV                               0x8ED1
+#define GL_COVERAGE_ATTACHMENT_NV                               0x8ED2
+#define GL_COVERAGE_BUFFERS_NV                                  0x8ED3
+#define GL_COVERAGE_SAMPLES_NV                                  0x8ED4
+#define GL_COVERAGE_ALL_FRAGMENTS_NV                            0x8ED5
+#define GL_COVERAGE_EDGE_FRAGMENTS_NV                           0x8ED6
+#define GL_COVERAGE_AUTOMATIC_NV                                0x8ED7
+#define GL_COVERAGE_BUFFER_BIT_NV                               0x8000
+#endif
+
+/* GL_NV_depth_nonlinear */
+#ifndef GL_NV_depth_nonlinear
+#define GL_DEPTH_COMPONENT16_NONLINEAR_NV                       0x8E2C
+#endif
+
+/* GL_NV_draw_buffers */
+#ifndef GL_NV_draw_buffers
+#define GL_MAX_DRAW_BUFFERS_NV                                  0x8824
+#define GL_DRAW_BUFFER0_NV                                      0x8825
+#define GL_DRAW_BUFFER1_NV                                      0x8826
+#define GL_DRAW_BUFFER2_NV                                      0x8827
+#define GL_DRAW_BUFFER3_NV                                      0x8828
+#define GL_DRAW_BUFFER4_NV                                      0x8829
+#define GL_DRAW_BUFFER5_NV                                      0x882A
+#define GL_DRAW_BUFFER6_NV                                      0x882B
+#define GL_DRAW_BUFFER7_NV                                      0x882C
+#define GL_DRAW_BUFFER8_NV                                      0x882D
+#define GL_DRAW_BUFFER9_NV                                      0x882E
+#define GL_DRAW_BUFFER10_NV                                     0x882F
+#define GL_DRAW_BUFFER11_NV                                     0x8830
+#define GL_DRAW_BUFFER12_NV                                     0x8831
+#define GL_DRAW_BUFFER13_NV                                     0x8832
+#define GL_DRAW_BUFFER14_NV                                     0x8833
+#define GL_DRAW_BUFFER15_NV                                     0x8834
+#define GL_COLOR_ATTACHMENT0_NV                                 0x8CE0
+#define GL_COLOR_ATTACHMENT1_NV                                 0x8CE1
+#define GL_COLOR_ATTACHMENT2_NV                                 0x8CE2
+#define GL_COLOR_ATTACHMENT3_NV                                 0x8CE3
+#define GL_COLOR_ATTACHMENT4_NV                                 0x8CE4
+#define GL_COLOR_ATTACHMENT5_NV                                 0x8CE5
+#define GL_COLOR_ATTACHMENT6_NV                                 0x8CE6
+#define GL_COLOR_ATTACHMENT7_NV                                 0x8CE7
+#define GL_COLOR_ATTACHMENT8_NV                                 0x8CE8
+#define GL_COLOR_ATTACHMENT9_NV                                 0x8CE9
+#define GL_COLOR_ATTACHMENT10_NV                                0x8CEA
+#define GL_COLOR_ATTACHMENT11_NV                                0x8CEB
+#define GL_COLOR_ATTACHMENT12_NV                                0x8CEC
+#define GL_COLOR_ATTACHMENT13_NV                                0x8CED
+#define GL_COLOR_ATTACHMENT14_NV                                0x8CEE
+#define GL_COLOR_ATTACHMENT15_NV                                0x8CEF
+#endif
+
+/* GL_EXT_draw_buffers */
+#ifndef GL_EXT_draw_buffers
+#define GL_MAX_DRAW_BUFFERS_EXT                                  0x8824
+#define GL_DRAW_BUFFER0_EXT                                      0x8825
+#define GL_DRAW_BUFFER1_EXT                                      0x8826
+#define GL_DRAW_BUFFER2_EXT                                      0x8827
+#define GL_DRAW_BUFFER3_EXT                                      0x8828
+#define GL_DRAW_BUFFER4_EXT                                      0x8829
+#define GL_DRAW_BUFFER5_EXT                                      0x882A
+#define GL_DRAW_BUFFER6_EXT                                      0x882B
+#define GL_DRAW_BUFFER7_EXT                                      0x882C
+#define GL_DRAW_BUFFER8_EXT                                      0x882D
+#define GL_DRAW_BUFFER9_EXT                                      0x882E
+#define GL_DRAW_BUFFER10_EXT                                     0x882F
+#define GL_DRAW_BUFFER11_EXT                                     0x8830
+#define GL_DRAW_BUFFER12_EXT                                     0x8831
+#define GL_DRAW_BUFFER13_EXT                                     0x8832
+#define GL_DRAW_BUFFER14_EXT                                     0x8833
+#define GL_DRAW_BUFFER15_EXT                                     0x8834
+#define GL_COLOR_ATTACHMENT0_EXT                                 0x8CE0
+#define GL_COLOR_ATTACHMENT1_EXT                                 0x8CE1
+#define GL_COLOR_ATTACHMENT2_EXT                                 0x8CE2
+#define GL_COLOR_ATTACHMENT3_EXT                                 0x8CE3
+#define GL_COLOR_ATTACHMENT4_EXT                                 0x8CE4
+#define GL_COLOR_ATTACHMENT5_EXT                                 0x8CE5
+#define GL_COLOR_ATTACHMENT6_EXT                                 0x8CE6
+#define GL_COLOR_ATTACHMENT7_EXT                                 0x8CE7
+#define GL_COLOR_ATTACHMENT8_EXT                                 0x8CE8
+#define GL_COLOR_ATTACHMENT9_EXT                                 0x8CE9
+#define GL_COLOR_ATTACHMENT10_EXT                                0x8CEA
+#define GL_COLOR_ATTACHMENT11_EXT                                0x8CEB
+#define GL_COLOR_ATTACHMENT12_EXT                                0x8CEC
+#define GL_COLOR_ATTACHMENT13_EXT                                0x8CED
+#define GL_COLOR_ATTACHMENT14_EXT                                0x8CEE
+#define GL_COLOR_ATTACHMENT15_EXT                                0x8CEF
+#define GL_MAX_COLOR_ATTACHMENTS_EXT                             0x8CDF
+#endif
+
+/* GL_NV_draw_instanced */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_fbo_color_attachments */
+#ifndef GL_NV_fbo_color_attachments
+#define GL_MAX_COLOR_ATTACHMENTS_NV                             0x8CDF
+/* GL_COLOR_ATTACHMENT{0-15}_NV defined in GL_NV_draw_buffers already. */
+#endif
+
+/* GL_NV_fence */
+#ifndef GL_NV_fence
+#define GL_ALL_COMPLETED_NV                                     0x84F2
+#define GL_FENCE_STATUS_NV                                      0x84F3
+#define GL_FENCE_CONDITION_NV                                   0x84F4
+#endif
+
+/* GL_NV_framebuffer_blit */
+#ifndef GL_NV_framebuffer_blit
+#define GL_READ_FRAMEBUFFER_NV                                  0x8CA8
+#define GL_DRAW_FRAMEBUFFER_NV                                  0x8CA9
+#define GL_DRAW_FRAMEBUFFER_BINDING_NV                          0x8CA6 
+#define GL_READ_FRAMEBUFFER_BINDING_NV                          0x8CAA
+#endif
+
+/* GL_NV_framebuffer_multisample */
+#ifndef GL_NV_framebuffer_multisample
+#define GL_RENDERBUFFER_SAMPLES_NV                              0x8CAB
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_NV                0x8D56
+#define GL_MAX_SAMPLES_NV                                       0x8D57
+#endif
+
+/* GL_NV_generate_mipmap_sRGB */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_instanced_arrays */
+#ifndef GL_NV_instanced_arrays
+#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR_NV                       0x88FE
+#endif
+
+/* GL_NV_read_buffer */
+#ifndef GL_NV_read_buffer
+#define GL_READ_BUFFER_NV                                       0x0C02
+#endif
+
+/* GL_NV_read_buffer_front */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_read_depth */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_read_depth_stencil */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_read_stencil */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_shadow_samplers_array */
+#ifndef GL_NV_shadow_samplers_array
+#define GL_SAMPLER_2D_ARRAY_SHADOW_NV                           0x8DC4
+#endif                                             
+                                                   
+/* GL_NV_shadow_samplers_cube */
+#ifndef GL_NV_shadow_samplers_cube                 
+#define GL_SAMPLER_CUBE_SHADOW_NV                               0x8DC5
+#endif
+
+/* GL_NV_sRGB_formats */
+#ifndef GL_NV_sRGB_formats
+#define GL_SLUMINANCE_NV                                        0x8C46
+#define GL_SLUMINANCE_ALPHA_NV                                  0x8C44
+#define GL_SRGB8_NV                                             0x8C41
+#define GL_SLUMINANCE8_NV                                       0x8C47
+#define GL_SLUMINANCE8_ALPHA8_NV                                0x8C45
+#define GL_COMPRESSED_SRGB_S3TC_DXT1_NV                         0x8C4C
+#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_NV                   0x8C4D
+#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_NV                   0x8C4E
+#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_NV                   0x8C4F
+#define GL_ETC1_SRGB8_NV                                        0x88EE
+#endif
+
+/* GL_NV_texture_border_clamp */
+#ifndef GL_NV_texture_border_clamp
+#define GL_TEXTURE_BORDER_COLOR_NV                              0x1004
+#define GL_CLAMP_TO_BORDER_NV                                   0x812D
+#endif
+
+/* GL_NV_texture_compression_s3tc_update */
+/* No new tokens introduced by this extension. */
+
+/* GL_NV_texture_npot_2D_mipmap */
+/* No new tokens introduced by this extension. */
+
+/*------------------------------------------------------------------------*
+ * QCOM extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_QCOM_alpha_test */
+#ifndef GL_QCOM_alpha_test
+#define GL_ALPHA_TEST_QCOM                                      0x0BC0
+#define GL_ALPHA_TEST_FUNC_QCOM                                 0x0BC1
+#define GL_ALPHA_TEST_REF_QCOM                                  0x0BC2
+#endif
+
+/* GL_QCOM_binning_control */
+#ifndef GL_QCOM_binning_control
+#define GL_BINNING_CONTROL_HINT_QCOM                            0x8FB0
+#define GL_CPU_OPTIMIZED_QCOM                                   0x8FB1
+#define GL_GPU_OPTIMIZED_QCOM                                   0x8FB2
+#define GL_RENDER_DIRECT_TO_FRAMEBUFFER_QCOM                    0x8FB3
+#endif
+
+/* GL_QCOM_driver_control */
+/* No new tokens introduced by this extension. */
+
+/* GL_QCOM_extended_get */
+#ifndef GL_QCOM_extended_get
+#define GL_TEXTURE_WIDTH_QCOM                                   0x8BD2
+#define GL_TEXTURE_HEIGHT_QCOM                                  0x8BD3
+#define GL_TEXTURE_DEPTH_QCOM                                   0x8BD4
+#define GL_TEXTURE_INTERNAL_FORMAT_QCOM                         0x8BD5
+#define GL_TEXTURE_FORMAT_QCOM                                  0x8BD6
+#define GL_TEXTURE_TYPE_QCOM                                    0x8BD7
+#define GL_TEXTURE_IMAGE_VALID_QCOM                             0x8BD8
+#define GL_TEXTURE_NUM_LEVELS_QCOM                              0x8BD9
+#define GL_TEXTURE_TARGET_QCOM                                  0x8BDA
+#define GL_TEXTURE_OBJECT_VALID_QCOM                            0x8BDB
+#define GL_STATE_RESTORE                                        0x8BDC
+#endif
+
+/* GL_QCOM_extended_get2 */
+/* No new tokens introduced by this extension. */
+
+/* GL_QCOM_perfmon_global_mode */
+#ifndef GL_QCOM_perfmon_global_mode
+#define GL_PERFMON_GLOBAL_MODE_QCOM                             0x8FA0
+#endif
+
+/* GL_QCOM_writeonly_rendering */
+#ifndef GL_QCOM_writeonly_rendering
+#define GL_WRITEONLY_RENDERING_QCOM                             0x8823
+#endif
+
+/* GL_QCOM_tiled_rendering */
+#ifndef GL_QCOM_tiled_rendering
+#define GL_COLOR_BUFFER_BIT0_QCOM                               0x00000001
+#define GL_COLOR_BUFFER_BIT1_QCOM                               0x00000002
+#define GL_COLOR_BUFFER_BIT2_QCOM                               0x00000004
+#define GL_COLOR_BUFFER_BIT3_QCOM                               0x00000008
+#define GL_COLOR_BUFFER_BIT4_QCOM                               0x00000010
+#define GL_COLOR_BUFFER_BIT5_QCOM                               0x00000020
+#define GL_COLOR_BUFFER_BIT6_QCOM                               0x00000040
+#define GL_COLOR_BUFFER_BIT7_QCOM                               0x00000080
+#define GL_DEPTH_BUFFER_BIT0_QCOM                               0x00000100
+#define GL_DEPTH_BUFFER_BIT1_QCOM                               0x00000200
+#define GL_DEPTH_BUFFER_BIT2_QCOM                               0x00000400
+#define GL_DEPTH_BUFFER_BIT3_QCOM                               0x00000800
+#define GL_DEPTH_BUFFER_BIT4_QCOM                               0x00001000
+#define GL_DEPTH_BUFFER_BIT5_QCOM                               0x00002000
+#define GL_DEPTH_BUFFER_BIT6_QCOM                               0x00004000
+#define GL_DEPTH_BUFFER_BIT7_QCOM                               0x00008000
+#define GL_STENCIL_BUFFER_BIT0_QCOM                             0x00010000
+#define GL_STENCIL_BUFFER_BIT1_QCOM                             0x00020000
+#define GL_STENCIL_BUFFER_BIT2_QCOM                             0x00040000
+#define GL_STENCIL_BUFFER_BIT3_QCOM                             0x00080000
+#define GL_STENCIL_BUFFER_BIT4_QCOM                             0x00100000
+#define GL_STENCIL_BUFFER_BIT5_QCOM                             0x00200000
+#define GL_STENCIL_BUFFER_BIT6_QCOM                             0x00400000
+#define GL_STENCIL_BUFFER_BIT7_QCOM                             0x00800000
+#define GL_MULTISAMPLE_BUFFER_BIT0_QCOM                         0x01000000
+#define GL_MULTISAMPLE_BUFFER_BIT1_QCOM                         0x02000000
+#define GL_MULTISAMPLE_BUFFER_BIT2_QCOM                         0x04000000
+#define GL_MULTISAMPLE_BUFFER_BIT3_QCOM                         0x08000000
+#define GL_MULTISAMPLE_BUFFER_BIT4_QCOM                         0x10000000
+#define GL_MULTISAMPLE_BUFFER_BIT5_QCOM                         0x20000000
+#define GL_MULTISAMPLE_BUFFER_BIT6_QCOM                         0x40000000
+#define GL_MULTISAMPLE_BUFFER_BIT7_QCOM                         0x80000000
+#endif
+
+/*------------------------------------------------------------------------*
+ * VIV extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_VIV_shader_binary */
+#ifndef GL_VIV_shader_binary
+#define GL_SHADER_BINARY_VIV                                    0x8FC4
+#endif
+
+/*------------------------------------------------------------------------*
+ * End of extension tokens, start of corresponding extension functions
+ *------------------------------------------------------------------------*/
+
+/*------------------------------------------------------------------------*
+ * OES extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_OES_compressed_ETC1_RGB8_texture */
+#ifndef GL_OES_compressed_ETC1_RGB8_texture
+#define GL_OES_compressed_ETC1_RGB8_texture 1
+#endif
+
+/* GL_OES_compressed_paletted_texture */
+#ifndef GL_OES_compressed_paletted_texture
+#define GL_OES_compressed_paletted_texture 1
+#endif
+
+/* GL_OES_depth24 */
+#ifndef GL_OES_depth24
+#define GL_OES_depth24 1
+#endif
+
+/* GL_OES_depth32 */
+#ifndef GL_OES_depth32
+#define GL_OES_depth32 1
+#endif
+
+/* GL_OES_depth_texture */
+#ifndef GL_OES_depth_texture
+#define GL_OES_depth_texture 1
+#endif
+
+/* GL_OES_EGL_image */
+#ifndef GL_OES_EGL_image
+#define GL_OES_EGL_image 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glEGLImageTargetTexture2DOES (GLenum target, GLeglImageOES image);
+GL_APICALL void GL_APIENTRY glEGLImageTargetRenderbufferStorageOES (GLenum target, GLeglImageOES image);
+#endif
+typedef void (GL_APIENTRYP PFNGLEGLIMAGETARGETTEXTURE2DOESPROC) (GLenum target, GLeglImageOES image);
+typedef void (GL_APIENTRYP PFNGLEGLIMAGETARGETRENDERBUFFERSTORAGEOESPROC) (GLenum target, GLeglImageOES image);
+#endif
+
+/* GL_OES_EGL_image_external */
+#ifndef GL_OES_EGL_image_external
+#define GL_OES_EGL_image_external 1
+/* glEGLImageTargetTexture2DOES defined in GL_OES_EGL_image already. */
+#endif
+
+/* GL_OES_element_index_uint */
+#ifndef GL_OES_element_index_uint
+#define GL_OES_element_index_uint 1
+#endif
+
+/* GL_OES_fbo_render_mipmap */
+#ifndef GL_OES_fbo_render_mipmap
+#define GL_OES_fbo_render_mipmap 1
+#endif
+
+/* GL_OES_fragment_precision_high */
+#ifndef GL_OES_fragment_precision_high
+#define GL_OES_fragment_precision_high 1
+#endif
+
+/* GL_OES_get_program_binary */
+#ifndef GL_OES_get_program_binary
+#define GL_OES_get_program_binary 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glGetProgramBinaryOES (GLuint program, GLsizei bufSize, GLsizei *length, GLenum *binaryFormat, GLvoid *binary);
+GL_APICALL void GL_APIENTRY glProgramBinaryOES (GLuint program, GLenum binaryFormat, const GLvoid *binary, GLint length);
+#endif
+typedef void (GL_APIENTRYP PFNGLGETPROGRAMBINARYOESPROC) (GLuint program, GLsizei bufSize, GLsizei *length, GLenum *binaryFormat, GLvoid *binary);
+typedef void (GL_APIENTRYP PFNGLPROGRAMBINARYOESPROC) (GLuint program, GLenum binaryFormat, const GLvoid *binary, GLint length);
+#endif
+
+/* GL_OES_mapbuffer */
+#ifndef GL_OES_mapbuffer
+#define GL_OES_mapbuffer 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void* GL_APIENTRY glMapBufferOES (GLenum target, GLenum access);
+GL_APICALL GLboolean GL_APIENTRY glUnmapBufferOES (GLenum target);
+GL_APICALL void GL_APIENTRY glGetBufferPointervOES (GLenum target, GLenum pname, GLvoid** params);
+#endif
+typedef void* (GL_APIENTRYP PFNGLMAPBUFFEROESPROC) (GLenum target, GLenum access);
+typedef GLboolean (GL_APIENTRYP PFNGLUNMAPBUFFEROESPROC) (GLenum target);
+typedef void (GL_APIENTRYP PFNGLGETBUFFERPOINTERVOESPROC) (GLenum target, GLenum pname, GLvoid** params);
+#endif
+
+/* GL_OES_packed_depth_stencil */
+#ifndef GL_OES_packed_depth_stencil
+#define GL_OES_packed_depth_stencil 1
+#endif
+
+/* GL_OES_required_internalformat */
+#ifndef GL_OES_required_internalformat
+#define GL_OES_required_internalformat 1
+#endif
+
+/* GL_OES_rgb8_rgba8 */
+#ifndef GL_OES_rgb8_rgba8
+#define GL_OES_rgb8_rgba8 1
+#endif
+
+/* GL_OES_standard_derivatives */
+#ifndef GL_OES_standard_derivatives
+#define GL_OES_standard_derivatives 1
+#endif
+
+/* GL_OES_stencil1 */
+#ifndef GL_OES_stencil1
+#define GL_OES_stencil1 1
+#endif
+
+/* GL_OES_stencil4 */
+#ifndef GL_OES_stencil4
+#define GL_OES_stencil4 1
+#endif
+
+#ifndef GL_OES_surfaceless_context
+#define GL_OES_surfaceless_context 1
+#endif
+
+/* GL_OES_texture_3D */
+#ifndef GL_OES_texture_3D
+#define GL_OES_texture_3D 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glTexImage3DOES (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void GL_APIENTRY glTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void GL_APIENTRY glCopyTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL void GL_APIENTRY glCompressedTexImage3DOES (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void GL_APIENTRY glCompressedTexSubImage3DOES (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void GL_APIENTRY glFramebufferTexture3DOES (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset);
+#endif
+typedef void (GL_APIENTRYP PFNGLTEXIMAGE3DOESPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);
+typedef void (GL_APIENTRYP PFNGLTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);
+typedef void (GL_APIENTRYP PFNGLCOPYTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLCOMPRESSEDTEXIMAGE3DOESPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);
+typedef void (GL_APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE3DOESPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);
+typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE3DOES) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset);
+#endif
+
+/* GL_OES_texture_float */
+#ifndef GL_OES_texture_float
+#define GL_OES_texture_float 1
+#endif
+
+/* GL_OES_texture_float_linear */
+#ifndef GL_OES_texture_float_linear
+#define GL_OES_texture_float_linear 1
+#endif
+
+/* GL_OES_texture_half_float */
+#ifndef GL_OES_texture_half_float
+#define GL_OES_texture_half_float 1
+#endif
+
+/* GL_OES_texture_half_float_linear */
+#ifndef GL_OES_texture_half_float_linear
+#define GL_OES_texture_half_float_linear 1
+#endif
+
+/* GL_OES_texture_npot */
+#ifndef GL_OES_texture_npot
+#define GL_OES_texture_npot 1
+#endif
+
+/* GL_OES_vertex_array_object */
+#ifndef GL_OES_vertex_array_object
+#define GL_OES_vertex_array_object 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glBindVertexArrayOES (GLuint array);
+GL_APICALL void GL_APIENTRY glDeleteVertexArraysOES (GLsizei n, const GLuint *arrays);
+GL_APICALL void GL_APIENTRY glGenVertexArraysOES (GLsizei n, GLuint *arrays);
+GL_APICALL GLboolean GL_APIENTRY glIsVertexArrayOES (GLuint array);
+#endif
+typedef void (GL_APIENTRYP PFNGLBINDVERTEXARRAYOESPROC) (GLuint array);
+typedef void (GL_APIENTRYP PFNGLDELETEVERTEXARRAYSOESPROC) (GLsizei n, const GLuint *arrays);
+typedef void (GL_APIENTRYP PFNGLGENVERTEXARRAYSOESPROC) (GLsizei n, GLuint *arrays);
+typedef GLboolean (GL_APIENTRYP PFNGLISVERTEXARRAYOESPROC) (GLuint array);
+#endif
+
+/* GL_OES_vertex_half_float */
+#ifndef GL_OES_vertex_half_float
+#define GL_OES_vertex_half_float 1
+#endif
+
+/* GL_OES_vertex_type_10_10_10_2 */
+#ifndef GL_OES_vertex_type_10_10_10_2
+#define GL_OES_vertex_type_10_10_10_2 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * KHR extension functions
+ *------------------------------------------------------------------------*/
+
+#ifndef GL_KHR_debug
+#define GL_KHR_debug 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDebugMessageControl (GLenum source, GLenum type, GLenum severity, GLsizei count, const GLuint *ids, GLboolean enabled);
+GL_APICALL void GL_APIENTRY glDebugMessageInsert (GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar *buf);
+GL_APICALL void GL_APIENTRY glDebugMessageCallback (GLDEBUGPROC callback, const void *userParam);
+GL_APICALL GLuint GL_APIENTRY glGetDebugMessageLog (GLuint count, GLsizei bufsize, GLenum *sources, GLenum *types, GLuint *ids, GLenum *severities, GLsizei *lengths, GLchar *messageLog);
+GL_APICALL void GL_APIENTRY glPushDebugGroup (GLenum source, GLuint id, GLsizei length, const GLchar *message);
+GL_APICALL void GL_APIENTRY glPopDebugGroup (void);
+GL_APICALL void GL_APIENTRY glObjectLabel (GLenum identifier, GLuint name, GLsizei length, const GLchar *label);
+GL_APICALL void GL_APIENTRY glGetObjectLabel (GLenum identifier, GLuint name, GLsizei bufSize, GLsizei *length, GLchar *label);
+GL_APICALL void GL_APIENTRY glObjectPtrLabel (const void *ptr, GLsizei length, const GLchar *label);
+GL_APICALL void GL_APIENTRY glGetObjectPtrLabel (const void *ptr, GLsizei bufSize, GLsizei *length, GLchar *label);
+GL_APICALL void GL_APIENTRY glGetPointerv (GLenum pname, void **params);
+#endif 
+typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGECONTROLPROC) (GLenum source, GLenum type, GLenum severity, GLsizei count, const GLuint *ids, GLboolean enabled);
+typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGEINSERTPROC) (GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar *buf);
+typedef void (GL_APIENTRYP PFNGLDEBUGMESSAGECALLBACKPROC) (GLDEBUGPROC callback, const void *userParam);
+typedef GLuint (GL_APIENTRYP PFNGLGETDEBUGMESSAGELOGPROC) (GLuint count, GLsizei bufsize, GLenum *sources, GLenum *types, GLuint *ids, GLenum *severities, GLsizei *lengths, GLchar *messageLog);
+typedef void (GL_APIENTRYP PFNGLPUSHDEBUGGROUPPROC) (GLenum source, GLuint id, GLsizei length, const GLchar *message);
+typedef void (GL_APIENTRYP PFNGLPOPDEBUGGROUPPROC) (void);
+typedef void (GL_APIENTRYP PFNGLOBJECTLABELPROC) (GLenum identifier, GLuint name, GLsizei length, const GLchar *label);
+typedef void (GL_APIENTRYP PFNGLGETOBJECTLABELPROC) (GLenum identifier, GLuint name, GLsizei bufSize, GLsizei *length, GLchar *label);
+typedef void (GL_APIENTRYP PFNGLOBJECTPTRLABELPROC) (const void *ptr, GLsizei length, const GLchar *label);
+typedef void (GL_APIENTRYP PFNGLGETOBJECTPTRLABELPROC) (const void *ptr, GLsizei bufSize, GLsizei *length, GLchar *label);
+typedef void (GL_APIENTRYP PFNGLGETPOINTERVPROC) (GLenum pname, void **params);
+#endif
+
+#ifndef GL_KHR_texture_compression_astc_ldr
+#define GL_KHR_texture_compression_astc_ldr 1
+#endif
+
+
+/*------------------------------------------------------------------------*
+ * AMD extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_AMD_compressed_3DC_texture */
+#ifndef GL_AMD_compressed_3DC_texture
+#define GL_AMD_compressed_3DC_texture 1
+#endif
+
+/* GL_AMD_compressed_ATC_texture */
+#ifndef GL_AMD_compressed_ATC_texture
+#define GL_AMD_compressed_ATC_texture 1
+#endif
+
+/* AMD_performance_monitor */
+#ifndef GL_AMD_performance_monitor
+#define GL_AMD_performance_monitor 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glGetPerfMonitorGroupsAMD (GLint *numGroups, GLsizei groupsSize, GLuint *groups);
+GL_APICALL void GL_APIENTRY glGetPerfMonitorCountersAMD (GLuint group, GLint *numCounters, GLint *maxActiveCounters, GLsizei counterSize, GLuint *counters);
+GL_APICALL void GL_APIENTRY glGetPerfMonitorGroupStringAMD (GLuint group, GLsizei bufSize, GLsizei *length, GLchar *groupString);
+GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterStringAMD (GLuint group, GLuint counter, GLsizei bufSize, GLsizei *length, GLchar *counterString);
+GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterInfoAMD (GLuint group, GLuint counter, GLenum pname, GLvoid *data);
+GL_APICALL void GL_APIENTRY glGenPerfMonitorsAMD (GLsizei n, GLuint *monitors);
+GL_APICALL void GL_APIENTRY glDeletePerfMonitorsAMD (GLsizei n, GLuint *monitors);
+GL_APICALL void GL_APIENTRY glSelectPerfMonitorCountersAMD (GLuint monitor, GLboolean enable, GLuint group, GLint numCounters, GLuint *countersList);
+GL_APICALL void GL_APIENTRY glBeginPerfMonitorAMD (GLuint monitor);
+GL_APICALL void GL_APIENTRY glEndPerfMonitorAMD (GLuint monitor);
+GL_APICALL void GL_APIENTRY glGetPerfMonitorCounterDataAMD (GLuint monitor, GLenum pname, GLsizei dataSize, GLuint *data, GLint *bytesWritten);
+#endif
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORGROUPSAMDPROC) (GLint *numGroups, GLsizei groupsSize, GLuint *groups);
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERSAMDPROC) (GLuint group, GLint *numCounters, GLint *maxActiveCounters, GLsizei counterSize, GLuint *counters);
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORGROUPSTRINGAMDPROC) (GLuint group, GLsizei bufSize, GLsizei *length, GLchar *groupString);
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERSTRINGAMDPROC) (GLuint group, GLuint counter, GLsizei bufSize, GLsizei *length, GLchar *counterString);
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERINFOAMDPROC) (GLuint group, GLuint counter, GLenum pname, GLvoid *data);
+typedef void (GL_APIENTRYP PFNGLGENPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors);
+typedef void (GL_APIENTRYP PFNGLDELETEPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors);
+typedef void (GL_APIENTRYP PFNGLSELECTPERFMONITORCOUNTERSAMDPROC) (GLuint monitor, GLboolean enable, GLuint group, GLint numCounters, GLuint *countersList);
+typedef void (GL_APIENTRYP PFNGLBEGINPERFMONITORAMDPROC) (GLuint monitor);
+typedef void (GL_APIENTRYP PFNGLENDPERFMONITORAMDPROC) (GLuint monitor);
+typedef void (GL_APIENTRYP PFNGLGETPERFMONITORCOUNTERDATAAMDPROC) (GLuint monitor, GLenum pname, GLsizei dataSize, GLuint *data, GLint *bytesWritten);
+#endif
+
+/* GL_AMD_program_binary_Z400 */
+#ifndef GL_AMD_program_binary_Z400
+#define GL_AMD_program_binary_Z400 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * ANGLE extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_ANGLE_depth_texture */
+#ifndef GL_ANGLE_depth_texture
+#define GL_ANGLE_depth_texture 1
+#endif
+
+/* GL_ANGLE_framebuffer_blit */
+#ifndef GL_ANGLE_framebuffer_blit
+#define GL_ANGLE_framebuffer_blit 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glBlitFramebufferANGLE (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+#endif
+typedef void (GL_APIENTRYP PFNGLBLITFRAMEBUFFERANGLEPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+#endif
+
+/* GL_ANGLE_framebuffer_multisample */
+#ifndef GL_ANGLE_framebuffer_multisample
+#define GL_ANGLE_framebuffer_multisample 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleANGLE (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+#endif
+typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEANGLEPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+#endif
+
+#ifndef GL_ANGLE_instanced_arrays 
+#define GL_ANGLE_instanced_arrays 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDrawArraysInstancedANGLE (GLenum mode, GLint first, GLsizei count, GLsizei primcount);
+GL_APICALL void GL_APIENTRY glDrawElementsInstancedANGLE (GLenum mode, GLsizei count, GLenum type, const void *indices, GLsizei primcount);
+GL_APICALL void GL_APIENTRY glVertexAttribDivisorANGLE (GLuint index, GLuint divisor);
+#endif
+typedef void (GL_APIENTRYP PFNGLDRAWARRAYSINSTANCEDANGLEPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount);
+typedef void (GL_APIENTRYP PFNGLDRAWELEMENTSINSTANCEDANGLEPROC) (GLenum mode, GLsizei count, GLenum type, const void *indices, GLsizei primcount);
+typedef void (GL_APIENTRYP PFNGLVERTEXATTRIBDIVISORANGLEPROC) (GLuint index, GLuint divisor);
+#endif
+
+/* GL_ANGLE_pack_reverse_row_order */
+#ifndef GL_ANGLE_pack_reverse_row_order 
+#define GL_ANGLE_pack_reverse_row_order 1
+#endif
+
+/* GL_ANGLE_program_binary */
+#ifndef GL_ANGLE_program_binary
+#define GL_ANGLE_program_binary 1
+#endif
+
+/* GL_ANGLE_texture_compression_dxt3 */
+#ifndef GL_ANGLE_texture_compression_dxt3 
+#define GL_ANGLE_texture_compression_dxt3 1
+#endif
+
+/* GL_ANGLE_texture_compression_dxt5 */
+#ifndef GL_ANGLE_texture_compression_dxt5 
+#define GL_ANGLE_texture_compression_dxt5 1
+#endif
+
+/* GL_ANGLE_texture_usage */
+#ifndef GL_ANGLE_texture_usage 
+#define GL_ANGLE_texture_usage 1
+#endif
+
+#ifndef GL_ANGLE_translated_shader_source 
+#define GL_ANGLE_translated_shader_source 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glGetTranslatedShaderSourceANGLE (GLuint shader, GLsizei bufsize, GLsizei *length, GLchar *source);
+#endif
+typedef void (GL_APIENTRYP PFNGLGETTRANSLATEDSHADERSOURCEANGLEPROC) (GLuint shader, GLsizei bufsize, GLsizei *length, GLchar *source);
+#endif
+
+/*------------------------------------------------------------------------*
+ * APPLE extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_APPLE_copy_texture_levels */
+#ifndef GL_APPLE_copy_texture_levels
+#define GL_APPLE_copy_texture_levels 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glCopyTextureLevelsAPPLE (GLuint destinationTexture, GLuint sourceTexture, GLint sourceBaseLevel, GLsizei sourceLevelCount);
+#endif
+typedef void (GL_APIENTRYP PFNGLCOPYTEXTURELEVELSAPPLEPROC) (GLuint destinationTexture, GLuint sourceTexture, GLint sourceBaseLevel, GLsizei sourceLevelCount);
+#endif
+
+/* GL_APPLE_framebuffer_multisample */
+#ifndef GL_APPLE_framebuffer_multisample
+#define GL_APPLE_framebuffer_multisample 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleAPPLE (GLenum, GLsizei, GLenum, GLsizei, GLsizei);
+GL_APICALL void GL_APIENTRY glResolveMultisampleFramebufferAPPLE (void);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEAPPLEPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLRESOLVEMULTISAMPLEFRAMEBUFFERAPPLEPROC) (void);
+#endif
+
+/* GL_APPLE_rgb_422 */
+#ifndef GL_APPLE_rgb_422
+#define GL_APPLE_rgb_422 1
+#endif
+
+/* GL_APPLE_sync */
+#ifndef GL_APPLE_sync
+#define GL_APPLE_sync 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL GLsync GL_APIENTRY glFenceSyncAPPLE (GLenum condition, GLbitfield flags);
+GL_APICALL GLboolean GL_APIENTRY glIsSyncAPPLE (GLsync sync);
+GL_APICALL void GL_APIENTRY glDeleteSyncAPPLE (GLsync sync);
+GL_APICALL GLenum GL_APIENTRY glClientWaitSyncAPPLE (GLsync sync, GLbitfield flags, GLuint64 timeout);
+GL_APICALL void GL_APIENTRY glWaitSyncAPPLE (GLsync sync, GLbitfield flags, GLuint64 timeout);
+GL_APICALL void GL_APIENTRY glGetInteger64vAPPLE (GLenum pname, GLint64 *params);
+GL_APICALL void GL_APIENTRY glGetSyncivAPPLE (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values);
+#endif
+typedef GLsync (GL_APIENTRYP PFNGLFENCESYNCAPPLEPROC) (GLenum condition, GLbitfield flags);
+typedef GLboolean (GL_APIENTRYP PFNGLISSYNCAPPLEPROC) (GLsync sync);
+typedef void (GL_APIENTRYP PFNGLDELETESYNCAPPLEPROC) (GLsync sync);
+typedef GLenum (GL_APIENTRYP PFNGLCLIENTWAITSYNCAPPLEPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);
+typedef void (GL_APIENTRYP PFNGLWAITSYNCAPPLEPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);
+typedef void (GL_APIENTRYP PFNGLGETINTEGER64VAPPLEPROC) (GLenum pname, GLint64 *params);
+typedef void (GL_APIENTRYP PFNGLGETSYNCIVAPPLEPROC) (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values);
+#endif
+
+/* GL_APPLE_texture_format_BGRA8888 */
+#ifndef GL_APPLE_texture_format_BGRA8888
+#define GL_APPLE_texture_format_BGRA8888 1
+#endif
+
+/* GL_APPLE_texture_max_level */
+#ifndef GL_APPLE_texture_max_level
+#define GL_APPLE_texture_max_level 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * ARM extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_ARM_mali_program_binary */
+#ifndef GL_ARM_mali_program_binary
+#define GL_ARM_mali_program_binary 1
+#endif
+
+/* GL_ARM_mali_shader_binary */
+#ifndef GL_ARM_mali_shader_binary
+#define GL_ARM_mali_shader_binary 1
+#endif
+
+/* GL_ARM_rgba8 */
+#ifndef GL_ARM_rgba8
+#define GL_ARM_rgba8 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * EXT extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_EXT_blend_minmax */
+#ifndef GL_EXT_blend_minmax
+#define GL_EXT_blend_minmax 1
+#endif
+
+/* GL_EXT_color_buffer_half_float */
+#ifndef GL_EXT_color_buffer_half_float
+#define GL_EXT_color_buffer_half_float 1
+#endif
+
+/* GL_EXT_debug_label */
+#ifndef GL_EXT_debug_label
+#define GL_EXT_debug_label 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glLabelObjectEXT (GLenum type, GLuint object, GLsizei length, const GLchar *label);
+GL_APICALL void GL_APIENTRY glGetObjectLabelEXT (GLenum type, GLuint object, GLsizei bufSize, GLsizei *length, GLchar *label);
+#endif
+typedef void (GL_APIENTRYP PFNGLLABELOBJECTEXTPROC) (GLenum type, GLuint object, GLsizei length, const GLchar *label);
+typedef void (GL_APIENTRYP PFNGLGETOBJECTLABELEXTPROC) (GLenum type, GLuint object, GLsizei bufSize, GLsizei *length, GLchar *label);
+#endif
+
+/* GL_EXT_debug_marker */
+#ifndef GL_EXT_debug_marker
+#define GL_EXT_debug_marker 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glInsertEventMarkerEXT (GLsizei length, const GLchar *marker);
+GL_APICALL void GL_APIENTRY glPushGroupMarkerEXT (GLsizei length, const GLchar *marker);
+GL_APICALL void GL_APIENTRY glPopGroupMarkerEXT (void);
+#endif
+typedef void (GL_APIENTRYP PFNGLINSERTEVENTMARKEREXTPROC) (GLsizei length, const GLchar *marker);
+typedef void (GL_APIENTRYP PFNGLPUSHGROUPMARKEREXTPROC) (GLsizei length, const GLchar *marker);
+typedef void (GL_APIENTRYP PFNGLPOPGROUPMARKEREXTPROC) (void);
+#endif
+
+/* GL_EXT_discard_framebuffer */
+#ifndef GL_EXT_discard_framebuffer
+#define GL_EXT_discard_framebuffer 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDiscardFramebufferEXT (GLenum target, GLsizei numAttachments, const GLenum *attachments);
+#endif
+typedef void (GL_APIENTRYP PFNGLDISCARDFRAMEBUFFEREXTPROC) (GLenum target, GLsizei numAttachments, const GLenum *attachments);
+#endif
+
+/* GL_EXT_map_buffer_range */
+#ifndef GL_EXT_map_buffer_range
+#define GL_EXT_map_buffer_range 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void* GL_APIENTRY glMapBufferRangeEXT (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);
+GL_APICALL void GL_APIENTRY glFlushMappedBufferRangeEXT (GLenum target, GLintptr offset, GLsizeiptr length);
+#endif
+typedef void* (GL_APIENTRYP PFNGLMAPBUFFERRANGEEXTPROC) (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);
+typedef void (GL_APIENTRYP PFNGLFLUSHMAPPEDBUFFERRANGEEXTPROC) (GLenum target, GLintptr offset, GLsizeiptr length);
+#endif
+
+/* GL_EXT_multisampled_render_to_texture */
+#ifndef GL_EXT_multisampled_render_to_texture
+#define GL_EXT_multisampled_render_to_texture 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleEXT (GLenum, GLsizei, GLenum, GLsizei, GLsizei);
+GL_APICALL void GL_APIENTRY glFramebufferTexture2DMultisampleEXT (GLenum, GLenum, GLenum, GLuint, GLint, GLsizei);
+#endif
+typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEEXTPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DMULTISAMPLEEXTPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLsizei samples);
+#endif
+
+/* GL_EXT_multiview_draw_buffers */
+#ifndef GL_EXT_multiview_draw_buffers
+#define GL_EXT_multiview_draw_buffers 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glReadBufferIndexedEXT (GLenum src, GLint index);
+GL_APICALL void GL_APIENTRY glDrawBuffersIndexedEXT (GLint n, const GLenum *location, const GLint *indices);
+GL_APICALL void GL_APIENTRY glGetIntegeri_vEXT (GLenum target, GLuint index, GLint *data);
+#endif
+typedef void (GL_APIENTRYP PFNGLREADBUFFERINDEXEDEXTPROC) (GLenum src, GLint index);
+typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSINDEXEDEXTPROC) (GLint n, const GLenum *location, const GLint *indices);
+typedef void (GL_APIENTRYP PFNGLGETINTEGERI_VEXTPROC) (GLenum target, GLuint index, GLint *data);
+#endif
+
+#ifndef GL_EXT_multi_draw_arrays
+#define GL_EXT_multi_draw_arrays 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glMultiDrawArraysEXT (GLenum, const GLint *, const GLsizei *, GLsizei);
+GL_APICALL void GL_APIENTRY glMultiDrawElementsEXT (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (GL_APIENTRYP PFNGLMULTIDRAWARRAYSEXTPROC) (GLenum mode, GLint *first, GLsizei *count, GLsizei primcount);
+typedef void (GL_APIENTRYP PFNGLMULTIDRAWELEMENTSEXTPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount);
+#endif
+
+/* GL_EXT_occlusion_query_boolean */
+#ifndef GL_EXT_occlusion_query_boolean
+#define GL_EXT_occlusion_query_boolean 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glGenQueriesEXT (GLsizei n, GLuint *ids);
+GL_APICALL void GL_APIENTRY glDeleteQueriesEXT (GLsizei n, const GLuint *ids);
+GL_APICALL GLboolean GL_APIENTRY glIsQueryEXT (GLuint id);
+GL_APICALL void GL_APIENTRY glBeginQueryEXT (GLenum target, GLuint id);
+GL_APICALL void GL_APIENTRY glEndQueryEXT (GLenum target);
+GL_APICALL void GL_APIENTRY glGetQueryivEXT (GLenum target, GLenum pname, GLint *params);
+GL_APICALL void GL_APIENTRY glGetQueryObjectuivEXT (GLuint id, GLenum pname, GLuint *params);
+#endif
+typedef void (GL_APIENTRYP PFNGLGENQUERIESEXTPROC) (GLsizei n, GLuint *ids);
+typedef void (GL_APIENTRYP PFNGLDELETEQUERIESEXTPROC) (GLsizei n, const GLuint *ids);
+typedef GLboolean (GL_APIENTRYP PFNGLISQUERYEXTPROC) (GLuint id);
+typedef void (GL_APIENTRYP PFNGLBEGINQUERYEXTPROC) (GLenum target, GLuint id);
+typedef void (GL_APIENTRYP PFNGLENDQUERYEXTPROC) (GLenum target);
+typedef void (GL_APIENTRYP PFNGLGETQUERYIVEXTPROC) (GLenum target, GLenum pname, GLint *params);
+typedef void (GL_APIENTRYP PFNGLGETQUERYOBJECTUIVEXTPROC) (GLuint id, GLenum pname, GLuint *params);
+#endif
+
+/* GL_EXT_read_format_bgra */
+#ifndef GL_EXT_read_format_bgra
+#define GL_EXT_read_format_bgra 1
+#endif
+
+/* GL_EXT_robustness */
+#ifndef GL_EXT_robustness
+#define GL_EXT_robustness 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL GLenum GL_APIENTRY glGetGraphicsResetStatusEXT (void);
+GL_APICALL void GL_APIENTRY glReadnPixelsEXT (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void *data);
+GL_APICALL void GL_APIENTRY glGetnUniformfvEXT (GLuint program, GLint location, GLsizei bufSize, float *params);
+GL_APICALL void GL_APIENTRY glGetnUniformivEXT (GLuint program, GLint location, GLsizei bufSize, GLint *params);
+#endif
+typedef GLenum (GL_APIENTRYP PFNGLGETGRAPHICSRESETSTATUSEXTPROC) (void);
+typedef void (GL_APIENTRYP PFNGLREADNPIXELSEXTPROC) (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLsizei bufSize, void *data);
+typedef void (GL_APIENTRYP PFNGLGETNUNIFORMFVEXTPROC) (GLuint program, GLint location, GLsizei bufSize, float *params);
+typedef void (GL_APIENTRYP PFNGLGETNUNIFORMIVEXTPROC) (GLuint program, GLint location, GLsizei bufSize, GLint *params);
+#endif
+
+/* GL_EXT_separate_shader_objects */
+#ifndef GL_EXT_separate_shader_objects
+#define GL_EXT_separate_shader_objects 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glUseProgramStagesEXT (GLuint pipeline, GLbitfield stages, GLuint program);
+GL_APICALL void GL_APIENTRY glActiveShaderProgramEXT (GLuint pipeline, GLuint program);
+GL_APICALL GLuint GL_APIENTRY glCreateShaderProgramvEXT (GLenum type, GLsizei count, const GLchar **strings);
+GL_APICALL void GL_APIENTRY glBindProgramPipelineEXT (GLuint pipeline);
+GL_APICALL void GL_APIENTRY glDeleteProgramPipelinesEXT (GLsizei n, const GLuint *pipelines);
+GL_APICALL void GL_APIENTRY glGenProgramPipelinesEXT (GLsizei n, GLuint *pipelines);
+GL_APICALL GLboolean GL_APIENTRY glIsProgramPipelineEXT (GLuint pipeline);
+GL_APICALL void GL_APIENTRY glProgramParameteriEXT (GLuint program, GLenum pname, GLint value);
+GL_APICALL void GL_APIENTRY glGetProgramPipelineivEXT (GLuint pipeline, GLenum pname, GLint *params);
+GL_APICALL void GL_APIENTRY glProgramUniform1iEXT (GLuint program, GLint location, GLint x);
+GL_APICALL void GL_APIENTRY glProgramUniform2iEXT (GLuint program, GLint location, GLint x, GLint y);
+GL_APICALL void GL_APIENTRY glProgramUniform3iEXT (GLuint program, GLint location, GLint x, GLint y, GLint z);
+GL_APICALL void GL_APIENTRY glProgramUniform4iEXT (GLuint program, GLint location, GLint x, GLint y, GLint z, GLint w);
+GL_APICALL void GL_APIENTRY glProgramUniform1fEXT (GLuint program, GLint location, GLfloat x);
+GL_APICALL void GL_APIENTRY glProgramUniform2fEXT (GLuint program, GLint location, GLfloat x, GLfloat y);
+GL_APICALL void GL_APIENTRY glProgramUniform3fEXT (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z);
+GL_APICALL void GL_APIENTRY glProgramUniform4fEXT (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+GL_APICALL void GL_APIENTRY glProgramUniform1ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);
+GL_APICALL void GL_APIENTRY glProgramUniform2ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);
+GL_APICALL void GL_APIENTRY glProgramUniform3ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);
+GL_APICALL void GL_APIENTRY glProgramUniform4ivEXT (GLuint program, GLint location, GLsizei count, const GLint *value);
+GL_APICALL void GL_APIENTRY glProgramUniform1fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniform2fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniform3fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniform4fvEXT (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniformMatrix2fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniformMatrix3fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glProgramUniformMatrix4fvEXT (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+GL_APICALL void GL_APIENTRY glValidateProgramPipelineEXT (GLuint pipeline);
+GL_APICALL void GL_APIENTRY glGetProgramPipelineInfoLogEXT (GLuint pipeline, GLsizei bufSize, GLsizei *length, GLchar *infoLog);
+#endif
+typedef void (GL_APIENTRYP PFNGLUSEPROGRAMSTAGESEXTPROC) (GLuint pipeline, GLbitfield stages, GLuint program);
+typedef void (GL_APIENTRYP PFNGLACTIVESHADERPROGRAMEXTPROC) (GLuint pipeline, GLuint program);
+typedef GLuint (GL_APIENTRYP PFNGLCREATESHADERPROGRAMVEXTPROC) (GLenum type, GLsizei count, const GLchar **strings);
+typedef void (GL_APIENTRYP PFNGLBINDPROGRAMPIPELINEEXTPROC) (GLuint pipeline);
+typedef void (GL_APIENTRYP PFNGLDELETEPROGRAMPIPELINESEXTPROC) (GLsizei n, const GLuint *pipelines);
+typedef void (GL_APIENTRYP PFNGLGENPROGRAMPIPELINESEXTPROC) (GLsizei n, GLuint *pipelines);
+typedef GLboolean (GL_APIENTRYP PFNGLISPROGRAMPIPELINEEXTPROC) (GLuint pipeline);
+typedef void (GL_APIENTRYP PFNGLPROGRAMPARAMETERIEXTPROC) (GLuint program, GLenum pname, GLint value);
+typedef void (GL_APIENTRYP PFNGLGETPROGRAMPIPELINEIVEXTPROC) (GLuint pipeline, GLenum pname, GLint *params);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1IEXTPROC) (GLuint program, GLint location, GLint x);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2IEXTPROC) (GLuint program, GLint location, GLint x, GLint y);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3IEXTPROC) (GLuint program, GLint location, GLint x, GLint y, GLint z);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4IEXTPROC) (GLuint program, GLint location, GLint x, GLint y, GLint z, GLint w);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1FEXTPROC) (GLuint program, GLint location, GLfloat x);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4FEXTPROC) (GLuint program, GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM1FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM2FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM3FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORM4FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX2FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX3FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLPROGRAMUNIFORMMATRIX4FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value);
+typedef void (GL_APIENTRYP PFNGLVALIDATEPROGRAMPIPELINEEXTPROC) (GLuint pipeline);
+typedef void (GL_APIENTRYP PFNGLGETPROGRAMPIPELINEINFOLOGEXTPROC) (GLuint pipeline, GLsizei bufSize, GLsizei *length, GLchar *infoLog);
+#endif
+
+/* GL_EXT_shader_framebuffer_fetch */
+#ifndef GL_EXT_shader_framebuffer_fetch
+#define GL_EXT_shader_framebuffer_fetch 1
+#endif
+
+/* GL_EXT_shader_texture_lod */
+#ifndef GL_EXT_shader_texture_lod
+#define GL_EXT_shader_texture_lod 1
+#endif
+
+/* GL_EXT_shadow_samplers */
+#ifndef GL_EXT_shadow_samplers
+#define GL_EXT_shadow_samplers 1
+#endif
+
+/* GL_EXT_sRGB */
+#ifndef GL_EXT_sRGB
+#define GL_EXT_sRGB 1
+#endif
+
+/* GL_EXT_texture_compression_dxt1 */
+#ifndef GL_EXT_texture_compression_dxt1
+#define GL_EXT_texture_compression_dxt1 1
+#endif
+
+/* GL_EXT_texture_filter_anisotropic */
+#ifndef GL_EXT_texture_filter_anisotropic
+#define GL_EXT_texture_filter_anisotropic 1
+#endif
+
+/* GL_EXT_texture_format_BGRA8888 */
+#ifndef GL_EXT_texture_format_BGRA8888
+#define GL_EXT_texture_format_BGRA8888 1
+#endif
+
+/* GL_EXT_texture_rg */
+#ifndef GL_EXT_texture_rg
+#define GL_EXT_texture_rg 1
+#endif
+
+/* GL_EXT_texture_storage */
+#ifndef GL_EXT_texture_storage
+#define GL_EXT_texture_storage 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glTexStorage1DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);
+GL_APICALL void GL_APIENTRY glTexStorage2DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void GL_APIENTRY glTexStorage3DEXT (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);
+GL_APICALL void GL_APIENTRY glTextureStorage1DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);
+GL_APICALL void GL_APIENTRY glTextureStorage2DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void GL_APIENTRY glTextureStorage3DEXT (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);
+#endif
+typedef void (GL_APIENTRYP PFNGLTEXSTORAGE1DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);
+typedef void (GL_APIENTRYP PFNGLTEXSTORAGE2DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLTEXSTORAGE3DEXTPROC) (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);
+typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE1DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);
+typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE2DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLTEXTURESTORAGE3DEXTPROC) (GLuint texture, GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);
+#endif
+
+/* GL_EXT_texture_type_2_10_10_10_REV */
+#ifndef GL_EXT_texture_type_2_10_10_10_REV
+#define GL_EXT_texture_type_2_10_10_10_REV 1
+#endif
+
+/* GL_EXT_unpack_subimage */
+#ifndef GL_EXT_unpack_subimage
+#define GL_EXT_unpack_subimage 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * DMP extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_DMP_shader_binary */
+#ifndef GL_DMP_shader_binary
+#define GL_DMP_shader_binary 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * FJ extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_FJ_shader_binary_GCCSO */
+#ifndef GL_FJ_shader_binary_GCCSO
+#define GL_FJ_shader_binary_GCCSO 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * IMG extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_IMG_program_binary */
+#ifndef GL_IMG_program_binary
+#define GL_IMG_program_binary 1
+#endif
+
+/* GL_IMG_read_format */
+#ifndef GL_IMG_read_format
+#define GL_IMG_read_format 1
+#endif
+
+/* GL_IMG_shader_binary */
+#ifndef GL_IMG_shader_binary
+#define GL_IMG_shader_binary 1
+#endif
+
+/* GL_IMG_texture_compression_pvrtc */
+#ifndef GL_IMG_texture_compression_pvrtc
+#define GL_IMG_texture_compression_pvrtc 1
+#endif
+
+/* GL_IMG_texture_compression_pvrtc2 */
+#ifndef GL_IMG_texture_compression_pvrtc2
+#define GL_IMG_texture_compression_pvrtc2 1
+#endif
+
+/* GL_IMG_multisampled_render_to_texture */
+#ifndef GL_IMG_multisampled_render_to_texture
+#define GL_IMG_multisampled_render_to_texture 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleIMG (GLenum, GLsizei, GLenum, GLsizei, GLsizei);
+GL_APICALL void GL_APIENTRY glFramebufferTexture2DMultisampleIMG (GLenum, GLenum, GLenum, GLuint, GLint, GLsizei);
+#endif
+typedef void (GL_APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEIMGPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+typedef void (GL_APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DMULTISAMPLEIMGPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLsizei samples);
+#endif
+
+/*------------------------------------------------------------------------*
+ * NV extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_NV_coverage_sample */
+#ifndef GL_NV_coverage_sample
+#define GL_NV_coverage_sample 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glCoverageMaskNV (GLboolean mask);
+GL_APICALL void GL_APIENTRY glCoverageOperationNV (GLenum operation);
+#endif
+typedef void (GL_APIENTRYP PFNGLCOVERAGEMASKNVPROC) (GLboolean mask);
+typedef void (GL_APIENTRYP PFNGLCOVERAGEOPERATIONNVPROC) (GLenum operation);
+#endif
+
+/* GL_NV_depth_nonlinear */
+#ifndef GL_NV_depth_nonlinear
+#define GL_NV_depth_nonlinear 1
+#endif
+
+/* GL_NV_draw_buffers */
+#ifndef GL_NV_draw_buffers
+#define GL_NV_draw_buffers 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDrawBuffersNV (GLsizei n, const GLenum *bufs);
+#endif
+typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSNVPROC) (GLsizei n, const GLenum *bufs);
+#endif
+
+/* GL_EXT_draw_buffers */
+#ifndef GL_EXT_draw_buffers
+#define GL_EXT_draw_buffers 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDrawBuffersEXT (GLsizei n, const GLenum *bufs);
+#endif
+typedef void (GL_APIENTRYP PFNGLDRAWBUFFERSEXTPROC) (GLsizei n, const GLenum *bufs);
+#endif
+
+/* GL_NV_draw_instanced */
+#ifndef GL_NV_draw_instanced
+#define GL_NV_draw_instanced 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDrawArraysInstancedNV (GLenum mode, GLint first, GLsizei count, GLsizei primcount);
+GL_APICALL void GL_APIENTRY glDrawElementsInstancedNV (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount);
+#endif
+typedef void (GL_APIENTRYP PFNDRAWARRAYSINSTANCEDNVPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount);
+typedef void (GL_APIENTRYP PFNDRAWELEMENTSINSTANCEDNVPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount);
+#endif
+
+/* GL_NV_fbo_color_attachments */
+#ifndef GL_NV_fbo_color_attachments
+#define GL_NV_fbo_color_attachments 1
+#endif
+
+/* GL_NV_fence */
+#ifndef GL_NV_fence
+#define GL_NV_fence 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glDeleteFencesNV (GLsizei, const GLuint *);
+GL_APICALL void GL_APIENTRY glGenFencesNV (GLsizei, GLuint *);
+GL_APICALL GLboolean GL_APIENTRY glIsFenceNV (GLuint);
+GL_APICALL GLboolean GL_APIENTRY glTestFenceNV (GLuint);
+GL_APICALL void GL_APIENTRY glGetFenceivNV (GLuint, GLenum, GLint *);
+GL_APICALL void GL_APIENTRY glFinishFenceNV (GLuint);
+GL_APICALL void GL_APIENTRY glSetFenceNV (GLuint, GLenum);
+#endif
+typedef void (GL_APIENTRYP PFNGLDELETEFENCESNVPROC) (GLsizei n, const GLuint *fences);
+typedef void (GL_APIENTRYP PFNGLGENFENCESNVPROC) (GLsizei n, GLuint *fences);
+typedef GLboolean (GL_APIENTRYP PFNGLISFENCENVPROC) (GLuint fence);
+typedef GLboolean (GL_APIENTRYP PFNGLTESTFENCENVPROC) (GLuint fence);
+typedef void (GL_APIENTRYP PFNGLGETFENCEIVNVPROC) (GLuint fence, GLenum pname, GLint *params);
+typedef void (GL_APIENTRYP PFNGLFINISHFENCENVPROC) (GLuint fence);
+typedef void (GL_APIENTRYP PFNGLSETFENCENVPROC) (GLuint fence, GLenum condition);
+#endif
+
+/* GL_NV_framebuffer_blit */
+#ifndef GL_NV_framebuffer_blit
+#define GL_NV_framebuffer_blit 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glBlitFramebufferNV (int srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+#endif
+typedef void (GL_APIENTRYP PFNBLITFRAMEBUFFERNVPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+#endif
+
+/* GL_NV_framebuffer_multisample */
+#ifndef GL_NV_framebuffer_multisample
+#define GL_NV_framebuffer_multisample 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glRenderbufferStorageMultisampleNV ( GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+#endif
+typedef void (GL_APIENTRYP PFNRENDERBUFFERSTORAGEMULTISAMPLENVPROC) ( GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+#endif
+
+/* GL_NV_generate_mipmap_sRGB */
+#ifndef GL_NV_generate_mipmap_sRGB
+#define GL_NV_generate_mipmap_sRGB 1
+#endif
+
+/* GL_NV_instanced_arrays */
+#ifndef GL_NV_instanced_arrays
+#define GL_NV_instanced_arrays 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glVertexAttribDivisorNV (GLuint index, GLuint divisor);
+#endif
+typedef void (GL_APIENTRYP PFNVERTEXATTRIBDIVISORNVPROC) (GLuint index, GLuint divisor);
+#endif
+
+/* GL_NV_read_buffer */
+#ifndef GL_NV_read_buffer
+#define GL_NV_read_buffer 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glReadBufferNV (GLenum mode);
+#endif
+typedef void (GL_APIENTRYP PFNGLREADBUFFERNVPROC) (GLenum mode);
+#endif
+
+/* GL_NV_read_buffer_front */
+#ifndef GL_NV_read_buffer_front
+#define GL_NV_read_buffer_front 1
+#endif
+
+/* GL_NV_read_depth */
+#ifndef GL_NV_read_depth
+#define GL_NV_read_depth 1
+#endif
+
+/* GL_NV_read_depth_stencil */
+#ifndef GL_NV_read_depth_stencil
+#define GL_NV_read_depth_stencil 1
+#endif
+
+/* GL_NV_read_stencil */
+#ifndef GL_NV_read_stencil
+#define GL_NV_read_stencil 1
+#endif
+
+/* GL_NV_shadow_samplers_array */
+#ifndef GL_NV_shadow_samplers_array
+#define GL_NV_shadow_samplers_array 1
+#endif
+
+/* GL_NV_shadow_samplers_cube */
+#ifndef GL_NV_shadow_samplers_cube
+#define GL_NV_shadow_samplers_cube 1
+#endif
+
+/* GL_NV_sRGB_formats */
+#ifndef GL_NV_sRGB_formats
+#define GL_NV_sRGB_formats 1
+#endif
+
+/* GL_NV_texture_border_clamp */
+#ifndef GL_NV_texture_border_clamp
+#define GL_NV_texture_border_clamp 1
+#endif
+
+/* GL_NV_texture_compression_s3tc_update */
+#ifndef GL_NV_texture_compression_s3tc_update
+#define GL_NV_texture_compression_s3tc_update 1
+#endif
+
+/* GL_NV_texture_npot_2D_mipmap */
+#ifndef GL_NV_texture_npot_2D_mipmap
+#define GL_NV_texture_npot_2D_mipmap 1
+#endif
+
+/*------------------------------------------------------------------------*
+ * QCOM extension functions
+ *------------------------------------------------------------------------*/
+
+/* GL_QCOM_alpha_test */
+#ifndef GL_QCOM_alpha_test
+#define GL_QCOM_alpha_test 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glAlphaFuncQCOM (GLenum func, GLclampf ref);
+#endif
+typedef void (GL_APIENTRYP PFNGLALPHAFUNCQCOMPROC) (GLenum func, GLclampf ref);
+#endif
+
+/* GL_QCOM_binning_control */
+#ifndef GL_QCOM_binning_control
+#define GL_QCOM_binning_control 1
+#endif
+
+/* GL_QCOM_driver_control */
+#ifndef GL_QCOM_driver_control
+#define GL_QCOM_driver_control 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glGetDriverControlsQCOM (GLint *num, GLsizei size, GLuint *driverControls);
+GL_APICALL void GL_APIENTRY glGetDriverControlStringQCOM (GLuint driverControl, GLsizei bufSize, GLsizei *length, GLchar *driverControlString);
+GL_APICALL void GL_APIENTRY glEnableDriverControlQCOM (GLuint driverControl);
+GL_APICALL void GL_APIENTRY glDisableDriverControlQCOM (GLuint driverControl);
+#endif
+typedef void (GL_APIENTRYP PFNGLGETDRIVERCONTROLSQCOMPROC) (GLint *num, GLsizei size, GLuint *driverControls);
+typedef void (GL_APIENTRYP PFNGLGETDRIVERCONTROLSTRINGQCOMPROC) (GLuint driverControl, GLsizei bufSize, GLsizei *length, GLchar *driverControlString);
+typedef void (GL_APIENTRYP PFNGLENABLEDRIVERCONTROLQCOMPROC) (GLuint driverControl);
+typedef void (GL_APIENTRYP PFNGLDISABLEDRIVERCONTROLQCOMPROC) (GLuint driverControl);
+#endif
+
+/* GL_QCOM_extended_get */
+#ifndef GL_QCOM_extended_get
+#define GL_QCOM_extended_get 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glExtGetTexturesQCOM (GLuint *textures, GLint maxTextures, GLint *numTextures);
+GL_APICALL void GL_APIENTRY glExtGetBuffersQCOM (GLuint *buffers, GLint maxBuffers, GLint *numBuffers);
+GL_APICALL void GL_APIENTRY glExtGetRenderbuffersQCOM (GLuint *renderbuffers, GLint maxRenderbuffers, GLint *numRenderbuffers);
+GL_APICALL void GL_APIENTRY glExtGetFramebuffersQCOM (GLuint *framebuffers, GLint maxFramebuffers, GLint *numFramebuffers);
+GL_APICALL void GL_APIENTRY glExtGetTexLevelParameterivQCOM (GLuint texture, GLenum face, GLint level, GLenum pname, GLint *params);
+GL_APICALL void GL_APIENTRY glExtTexObjectStateOverrideiQCOM (GLenum target, GLenum pname, GLint param);
+GL_APICALL void GL_APIENTRY glExtGetTexSubImageQCOM (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, GLvoid *texels);
+GL_APICALL void GL_APIENTRY glExtGetBufferPointervQCOM (GLenum target, GLvoid **params);
+#endif
+typedef void (GL_APIENTRYP PFNGLEXTGETTEXTURESQCOMPROC) (GLuint *textures, GLint maxTextures, GLint *numTextures);
+typedef void (GL_APIENTRYP PFNGLEXTGETBUFFERSQCOMPROC) (GLuint *buffers, GLint maxBuffers, GLint *numBuffers);
+typedef void (GL_APIENTRYP PFNGLEXTGETRENDERBUFFERSQCOMPROC) (GLuint *renderbuffers, GLint maxRenderbuffers, GLint *numRenderbuffers);
+typedef void (GL_APIENTRYP PFNGLEXTGETFRAMEBUFFERSQCOMPROC) (GLuint *framebuffers, GLint maxFramebuffers, GLint *numFramebuffers);
+typedef void (GL_APIENTRYP PFNGLEXTGETTEXLEVELPARAMETERIVQCOMPROC) (GLuint texture, GLenum face, GLint level, GLenum pname, GLint *params);
+typedef void (GL_APIENTRYP PFNGLEXTTEXOBJECTSTATEOVERRIDEIQCOMPROC) (GLenum target, GLenum pname, GLint param);
+typedef void (GL_APIENTRYP PFNGLEXTGETTEXSUBIMAGEQCOMPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, GLvoid *texels);
+typedef void (GL_APIENTRYP PFNGLEXTGETBUFFERPOINTERVQCOMPROC) (GLenum target, GLvoid **params);
+#endif
+
+/* GL_QCOM_extended_get2 */
+#ifndef GL_QCOM_extended_get2
+#define GL_QCOM_extended_get2 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glExtGetShadersQCOM (GLuint *shaders, GLint maxShaders, GLint *numShaders);
+GL_APICALL void GL_APIENTRY glExtGetProgramsQCOM (GLuint *programs, GLint maxPrograms, GLint *numPrograms);
+GL_APICALL GLboolean GL_APIENTRY glExtIsProgramBinaryQCOM (GLuint program);
+GL_APICALL void GL_APIENTRY glExtGetProgramBinarySourceQCOM (GLuint program, GLenum shadertype, GLchar *source, GLint *length);
+#endif
+typedef void (GL_APIENTRYP PFNGLEXTGETSHADERSQCOMPROC) (GLuint *shaders, GLint maxShaders, GLint *numShaders);
+typedef void (GL_APIENTRYP PFNGLEXTGETPROGRAMSQCOMPROC) (GLuint *programs, GLint maxPrograms, GLint *numPrograms);
+typedef GLboolean (GL_APIENTRYP PFNGLEXTISPROGRAMBINARYQCOMPROC) (GLuint program);
+typedef void (GL_APIENTRYP PFNGLEXTGETPROGRAMBINARYSOURCEQCOMPROC) (GLuint program, GLenum shadertype, GLchar *source, GLint *length);
+#endif
+
+/* GL_QCOM_perfmon_global_mode */
+#ifndef GL_QCOM_perfmon_global_mode
+#define GL_QCOM_perfmon_global_mode 1
+#endif
+
+/* GL_QCOM_writeonly_rendering */
+#ifndef GL_QCOM_writeonly_rendering
+#define GL_QCOM_writeonly_rendering 1
+#endif
+
+/* GL_QCOM_tiled_rendering */
+#ifndef GL_QCOM_tiled_rendering
+#define GL_QCOM_tiled_rendering 1
+#ifdef GL_GLEXT_PROTOTYPES
+GL_APICALL void GL_APIENTRY glStartTilingQCOM (GLuint x, GLuint y, GLuint width, GLuint height, GLbitfield preserveMask);
+GL_APICALL void GL_APIENTRY glEndTilingQCOM (GLbitfield preserveMask);
+#endif
+typedef void (GL_APIENTRYP PFNGLSTARTTILINGQCOMPROC) (GLuint x, GLuint y, GLuint width, GLuint height, GLbitfield preserveMask);
+typedef void (GL_APIENTRYP PFNGLENDTILINGQCOMPROC) (GLbitfield preserveMask);
+#endif
+
+/*------------------------------------------------------------------------*
+ * VIV extension tokens
+ *------------------------------------------------------------------------*/
+
+/* GL_VIV_shader_binary */
+#ifndef GL_VIV_shader_binary
+#define GL_VIV_shader_binary 1
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __gl2ext_h_ */
diff --git a/platform/winrt/include/GLES2/gl2platform.h b/platform/winrt/include/GLES2/gl2platform.h
index 38cb3b7b6f..c9fa3c4d64 100644
--- a/platform/winrt/include/GLES2/gl2platform.h
+++ b/platform/winrt/include/GLES2/gl2platform.h
@@ -1,30 +1,30 @@
-#ifndef __gl2platform_h_

-#define __gl2platform_h_

-

-/* $Revision: 10602 $ on $Date:: 2010-03-04 22:35:34 -0800 #$ */

-

-/*

- * This document is licensed under the SGI Free Software B License Version

- * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .

- */

-

-/* Platform-specific types and definitions for OpenGL ES 2.X  gl2.h

- *

- * Adopters may modify khrplatform.h and this file to suit their platform.

- * You are encouraged to submit all modifications to the Khronos group so that

- * they can be included in future versions of this file.  Please submit changes

- * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)

- * by filing a bug against product "OpenGL-ES" component "Registry".

- */

-

-#include <KHR/khrplatform.h>

-

-#ifndef GL_APICALL

-#define GL_APICALL  KHRONOS_APICALL

-#endif

-

-#ifndef GL_APIENTRY

-#define GL_APIENTRY KHRONOS_APIENTRY

-#endif

-

-#endif /* __gl2platform_h_ */

+#ifndef __gl2platform_h_
+#define __gl2platform_h_
+
+/* $Revision: 10602 $ on $Date:: 2010-03-04 22:35:34 -0800 #$ */
+
+/*
+ * This document is licensed under the SGI Free Software B License Version
+ * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .
+ */
+
+/* Platform-specific types and definitions for OpenGL ES 2.X  gl2.h
+ *
+ * Adopters may modify khrplatform.h and this file to suit their platform.
+ * You are encouraged to submit all modifications to the Khronos group so that
+ * they can be included in future versions of this file.  Please submit changes
+ * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)
+ * by filing a bug against product "OpenGL-ES" component "Registry".
+ */
+
+#include <KHR/khrplatform.h>
+
+#ifndef GL_APICALL
+#define GL_APICALL  KHRONOS_APICALL
+#endif
+
+#ifndef GL_APIENTRY
+#define GL_APIENTRY KHRONOS_APIENTRY
+#endif
+
+#endif /* __gl2platform_h_ */
diff --git a/platform/winrt/include/GLES3/gl3.h b/platform/winrt/include/GLES3/gl3.h
index 024edc4306..9c79862c0d 100644
--- a/platform/winrt/include/GLES3/gl3.h
+++ b/platform/winrt/include/GLES3/gl3.h
@@ -1,1061 +1,1061 @@
-#ifndef __gl3_h_

-#define __gl3_h_

-

-/* 

- * gl3.h last updated on $Date: 2013-02-12 14:37:24 -0800 (Tue, 12 Feb 2013) $

- */

-

-#include <GLES3/gl3platform.h>

-

-#ifdef __cplusplus

-extern "C" {

-#endif

-

-/*

-** Copyright (c) 2007-2013 The Khronos Group Inc.

-**

-** Permission is hereby granted, free of charge, to any person obtaining a

-** copy of this software and/or associated documentation files (the

-** "Materials"), to deal in the Materials without restriction, including

-** without limitation the rights to use, copy, modify, merge, publish,

-** distribute, sublicense, and/or sell copies of the Materials, and to

-** permit persons to whom the Materials are furnished to do so, subject to

-** the following conditions:

-**

-** The above copyright notice and this permission notice shall be included

-** in all copies or substantial portions of the Materials.

-**

-** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY

-** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

-** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

-** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

-*/

-

-/*-------------------------------------------------------------------------

- * Data type definitions

- *-----------------------------------------------------------------------*/

-

-/* OpenGL ES 2.0 */

-

-typedef void             GLvoid;

-typedef char             GLchar;

-typedef unsigned int     GLenum;

-typedef unsigned char    GLboolean;

-typedef unsigned int     GLbitfield;

-typedef khronos_int8_t   GLbyte;

-typedef short            GLshort;

-typedef int              GLint;

-typedef int              GLsizei;

-typedef khronos_uint8_t  GLubyte;

-typedef unsigned short   GLushort;

-typedef unsigned int     GLuint;

-typedef khronos_float_t  GLfloat;

-typedef khronos_float_t  GLclampf;

-typedef khronos_int32_t  GLfixed;

-typedef khronos_intptr_t GLintptr;

-typedef khronos_ssize_t  GLsizeiptr;

-

-/* OpenGL ES 3.0 */

-

-typedef unsigned short   GLhalf;

-typedef khronos_int64_t  GLint64;

-typedef khronos_uint64_t GLuint64;

-typedef struct __GLsync *GLsync;

-

-/*-------------------------------------------------------------------------

- * Token definitions

- *-----------------------------------------------------------------------*/

-

-/* OpenGL ES core versions */

-#define GL_ES_VERSION_3_0                                1

-#define GL_ES_VERSION_2_0                                1

-

-/* OpenGL ES 2.0 */

-

-/* ClearBufferMask */

-#define GL_DEPTH_BUFFER_BIT                              0x00000100

-#define GL_STENCIL_BUFFER_BIT                            0x00000400

-#define GL_COLOR_BUFFER_BIT                              0x00004000

-

-/* Boolean */

-#define GL_FALSE                                         0

-#define GL_TRUE                                          1

-

-/* BeginMode */

-#define GL_POINTS                                        0x0000

-#define GL_LINES                                         0x0001

-#define GL_LINE_LOOP                                     0x0002

-#define GL_LINE_STRIP                                    0x0003

-#define GL_TRIANGLES                                     0x0004

-#define GL_TRIANGLE_STRIP                                0x0005

-#define GL_TRIANGLE_FAN                                  0x0006

-

-/* BlendingFactorDest */

-#define GL_ZERO                                          0

-#define GL_ONE                                           1

-#define GL_SRC_COLOR                                     0x0300

-#define GL_ONE_MINUS_SRC_COLOR                           0x0301

-#define GL_SRC_ALPHA                                     0x0302

-#define GL_ONE_MINUS_SRC_ALPHA                           0x0303

-#define GL_DST_ALPHA                                     0x0304

-#define GL_ONE_MINUS_DST_ALPHA                           0x0305

-

-/* BlendingFactorSrc */

-/*      GL_ZERO */

-/*      GL_ONE */

-#define GL_DST_COLOR                                     0x0306

-#define GL_ONE_MINUS_DST_COLOR                           0x0307

-#define GL_SRC_ALPHA_SATURATE                            0x0308

-/*      GL_SRC_ALPHA */

-/*      GL_ONE_MINUS_SRC_ALPHA */

-/*      GL_DST_ALPHA */

-/*      GL_ONE_MINUS_DST_ALPHA */

-

-/* BlendEquationSeparate */

-#define GL_FUNC_ADD                                      0x8006

-#define GL_BLEND_EQUATION                                0x8009

-#define GL_BLEND_EQUATION_RGB                            0x8009    /* same as BLEND_EQUATION */

-#define GL_BLEND_EQUATION_ALPHA                          0x883D

-

-/* BlendSubtract */

-#define GL_FUNC_SUBTRACT                                 0x800A

-#define GL_FUNC_REVERSE_SUBTRACT                         0x800B

-

-/* Separate Blend Functions */

-#define GL_BLEND_DST_RGB                                 0x80C8

-#define GL_BLEND_SRC_RGB                                 0x80C9

-#define GL_BLEND_DST_ALPHA                               0x80CA

-#define GL_BLEND_SRC_ALPHA                               0x80CB

-#define GL_CONSTANT_COLOR                                0x8001

-#define GL_ONE_MINUS_CONSTANT_COLOR                      0x8002

-#define GL_CONSTANT_ALPHA                                0x8003

-#define GL_ONE_MINUS_CONSTANT_ALPHA                      0x8004

-#define GL_BLEND_COLOR                                   0x8005

-

-/* Buffer Objects */

-#define GL_ARRAY_BUFFER                                  0x8892

-#define GL_ELEMENT_ARRAY_BUFFER                          0x8893

-#define GL_ARRAY_BUFFER_BINDING                          0x8894

-#define GL_ELEMENT_ARRAY_BUFFER_BINDING                  0x8895

-

-#define GL_STREAM_DRAW                                   0x88E0

-#define GL_STATIC_DRAW                                   0x88E4

-#define GL_DYNAMIC_DRAW                                  0x88E8

-

-#define GL_BUFFER_SIZE                                   0x8764

-#define GL_BUFFER_USAGE                                  0x8765

-

-#define GL_CURRENT_VERTEX_ATTRIB                         0x8626

-

-/* CullFaceMode */

-#define GL_FRONT                                         0x0404

-#define GL_BACK                                          0x0405

-#define GL_FRONT_AND_BACK                                0x0408

-

-/* DepthFunction */

-/*      GL_NEVER */

-/*      GL_LESS */

-/*      GL_EQUAL */

-/*      GL_LEQUAL */

-/*      GL_GREATER */

-/*      GL_NOTEQUAL */

-/*      GL_GEQUAL */

-/*      GL_ALWAYS */

-

-/* EnableCap */

-#define GL_TEXTURE_2D                                    0x0DE1

-#define GL_CULL_FACE                                     0x0B44

-#define GL_BLEND                                         0x0BE2

-#define GL_DITHER                                        0x0BD0

-#define GL_STENCIL_TEST                                  0x0B90

-#define GL_DEPTH_TEST                                    0x0B71

-#define GL_SCISSOR_TEST                                  0x0C11

-#define GL_POLYGON_OFFSET_FILL                           0x8037

-#define GL_SAMPLE_ALPHA_TO_COVERAGE                      0x809E

-#define GL_SAMPLE_COVERAGE                               0x80A0

-

-/* ErrorCode */

-#define GL_NO_ERROR                                      0

-#define GL_INVALID_ENUM                                  0x0500

-#define GL_INVALID_VALUE                                 0x0501

-#define GL_INVALID_OPERATION                             0x0502

-#define GL_OUT_OF_MEMORY                                 0x0505

-

-/* FrontFaceDirection */

-#define GL_CW                                            0x0900

-#define GL_CCW                                           0x0901

-

-/* GetPName */

-#define GL_LINE_WIDTH                                    0x0B21

-#define GL_ALIASED_POINT_SIZE_RANGE                      0x846D

-#define GL_ALIASED_LINE_WIDTH_RANGE                      0x846E

-#define GL_CULL_FACE_MODE                                0x0B45

-#define GL_FRONT_FACE                                    0x0B46

-#define GL_DEPTH_RANGE                                   0x0B70

-#define GL_DEPTH_WRITEMASK                               0x0B72

-#define GL_DEPTH_CLEAR_VALUE                             0x0B73

-#define GL_DEPTH_FUNC                                    0x0B74

-#define GL_STENCIL_CLEAR_VALUE                           0x0B91

-#define GL_STENCIL_FUNC                                  0x0B92

-#define GL_STENCIL_FAIL                                  0x0B94

-#define GL_STENCIL_PASS_DEPTH_FAIL                       0x0B95

-#define GL_STENCIL_PASS_DEPTH_PASS                       0x0B96

-#define GL_STENCIL_REF                                   0x0B97

-#define GL_STENCIL_VALUE_MASK                            0x0B93

-#define GL_STENCIL_WRITEMASK                             0x0B98

-#define GL_STENCIL_BACK_FUNC                             0x8800

-#define GL_STENCIL_BACK_FAIL                             0x8801

-#define GL_STENCIL_BACK_PASS_DEPTH_FAIL                  0x8802

-#define GL_STENCIL_BACK_PASS_DEPTH_PASS                  0x8803

-#define GL_STENCIL_BACK_REF                              0x8CA3

-#define GL_STENCIL_BACK_VALUE_MASK                       0x8CA4

-#define GL_STENCIL_BACK_WRITEMASK                        0x8CA5

-#define GL_VIEWPORT                                      0x0BA2

-#define GL_SCISSOR_BOX                                   0x0C10

-/*      GL_SCISSOR_TEST */

-#define GL_COLOR_CLEAR_VALUE                             0x0C22

-#define GL_COLOR_WRITEMASK                               0x0C23

-#define GL_UNPACK_ALIGNMENT                              0x0CF5

-#define GL_PACK_ALIGNMENT                                0x0D05

-#define GL_MAX_TEXTURE_SIZE                              0x0D33

-#define GL_MAX_VIEWPORT_DIMS                             0x0D3A

-#define GL_SUBPIXEL_BITS                                 0x0D50

-#define GL_RED_BITS                                      0x0D52

-#define GL_GREEN_BITS                                    0x0D53

-#define GL_BLUE_BITS                                     0x0D54

-#define GL_ALPHA_BITS                                    0x0D55

-#define GL_DEPTH_BITS                                    0x0D56

-#define GL_STENCIL_BITS                                  0x0D57

-#define GL_POLYGON_OFFSET_UNITS                          0x2A00

-/*      GL_POLYGON_OFFSET_FILL */

-#define GL_POLYGON_OFFSET_FACTOR                         0x8038

-#define GL_TEXTURE_BINDING_2D                            0x8069

-#define GL_SAMPLE_BUFFERS                                0x80A8

-#define GL_SAMPLES                                       0x80A9

-#define GL_SAMPLE_COVERAGE_VALUE                         0x80AA

-#define GL_SAMPLE_COVERAGE_INVERT                        0x80AB

-

-/* GetTextureParameter */

-/*      GL_TEXTURE_MAG_FILTER */

-/*      GL_TEXTURE_MIN_FILTER */

-/*      GL_TEXTURE_WRAP_S */

-/*      GL_TEXTURE_WRAP_T */

-

-#define GL_NUM_COMPRESSED_TEXTURE_FORMATS                0x86A2

-#define GL_COMPRESSED_TEXTURE_FORMATS                    0x86A3

-

-/* HintMode */

-#define GL_DONT_CARE                                     0x1100

-#define GL_FASTEST                                       0x1101

-#define GL_NICEST                                        0x1102

-

-/* HintTarget */

-#define GL_GENERATE_MIPMAP_HINT                          0x8192

-

-/* DataType */

-#define GL_BYTE                                          0x1400

-#define GL_UNSIGNED_BYTE                                 0x1401

-#define GL_SHORT                                         0x1402

-#define GL_UNSIGNED_SHORT                                0x1403

-#define GL_INT                                           0x1404

-#define GL_UNSIGNED_INT                                  0x1405

-#define GL_FLOAT                                         0x1406

-#define GL_FIXED                                         0x140C

-

-/* PixelFormat */

-#define GL_DEPTH_COMPONENT                               0x1902

-#define GL_ALPHA                                         0x1906

-#define GL_RGB                                           0x1907

-#define GL_RGBA                                          0x1908

-#define GL_LUMINANCE                                     0x1909

-#define GL_LUMINANCE_ALPHA                               0x190A

-

-/* PixelType */

-/*      GL_UNSIGNED_BYTE */

-#define GL_UNSIGNED_SHORT_4_4_4_4                        0x8033

-#define GL_UNSIGNED_SHORT_5_5_5_1                        0x8034

-#define GL_UNSIGNED_SHORT_5_6_5                          0x8363

-

-/* Shaders */

-#define GL_FRAGMENT_SHADER                               0x8B30

-#define GL_VERTEX_SHADER                                 0x8B31

-#define GL_MAX_VERTEX_ATTRIBS                            0x8869

-#define GL_MAX_VERTEX_UNIFORM_VECTORS                    0x8DFB

-#define GL_MAX_VARYING_VECTORS                           0x8DFC

-#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS              0x8B4D

-#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS                0x8B4C

-#define GL_MAX_TEXTURE_IMAGE_UNITS                       0x8872

-#define GL_MAX_FRAGMENT_UNIFORM_VECTORS                  0x8DFD

-#define GL_SHADER_TYPE                                   0x8B4F

-#define GL_DELETE_STATUS                                 0x8B80

-#define GL_LINK_STATUS                                   0x8B82

-#define GL_VALIDATE_STATUS                               0x8B83

-#define GL_ATTACHED_SHADERS                              0x8B85

-#define GL_ACTIVE_UNIFORMS                               0x8B86

-#define GL_ACTIVE_UNIFORM_MAX_LENGTH                     0x8B87

-#define GL_ACTIVE_ATTRIBUTES                             0x8B89

-#define GL_ACTIVE_ATTRIBUTE_MAX_LENGTH                   0x8B8A

-#define GL_SHADING_LANGUAGE_VERSION                      0x8B8C

-#define GL_CURRENT_PROGRAM                               0x8B8D

-

-/* StencilFunction */

-#define GL_NEVER                                         0x0200

-#define GL_LESS                                          0x0201

-#define GL_EQUAL                                         0x0202

-#define GL_LEQUAL                                        0x0203

-#define GL_GREATER                                       0x0204

-#define GL_NOTEQUAL                                      0x0205

-#define GL_GEQUAL                                        0x0206

-#define GL_ALWAYS                                        0x0207

-

-/* StencilOp */

-/*      GL_ZERO */

-#define GL_KEEP                                          0x1E00

-#define GL_REPLACE                                       0x1E01

-#define GL_INCR                                          0x1E02

-#define GL_DECR                                          0x1E03

-#define GL_INVERT                                        0x150A

-#define GL_INCR_WRAP                                     0x8507

-#define GL_DECR_WRAP                                     0x8508

-

-/* StringName */

-#define GL_VENDOR                                        0x1F00

-#define GL_RENDERER                                      0x1F01

-#define GL_VERSION                                       0x1F02

-#define GL_EXTENSIONS                                    0x1F03

-

-/* TextureMagFilter */

-#define GL_NEAREST                                       0x2600

-#define GL_LINEAR                                        0x2601

-

-/* TextureMinFilter */

-/*      GL_NEAREST */

-/*      GL_LINEAR */

-#define GL_NEAREST_MIPMAP_NEAREST                        0x2700

-#define GL_LINEAR_MIPMAP_NEAREST                         0x2701

-#define GL_NEAREST_MIPMAP_LINEAR                         0x2702

-#define GL_LINEAR_MIPMAP_LINEAR                          0x2703

-

-/* TextureParameterName */

-#define GL_TEXTURE_MAG_FILTER                            0x2800

-#define GL_TEXTURE_MIN_FILTER                            0x2801

-#define GL_TEXTURE_WRAP_S                                0x2802

-#define GL_TEXTURE_WRAP_T                                0x2803

-

-/* TextureTarget */

-/*      GL_TEXTURE_2D */

-#define GL_TEXTURE                                       0x1702

-

-#define GL_TEXTURE_CUBE_MAP                              0x8513

-#define GL_TEXTURE_BINDING_CUBE_MAP                      0x8514

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_X                   0x8515

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X                   0x8516

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y                   0x8517

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y                   0x8518

-#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z                   0x8519

-#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z                   0x851A

-#define GL_MAX_CUBE_MAP_TEXTURE_SIZE                     0x851C

-

-/* TextureUnit */

-#define GL_TEXTURE0                                      0x84C0

-#define GL_TEXTURE1                                      0x84C1

-#define GL_TEXTURE2                                      0x84C2

-#define GL_TEXTURE3                                      0x84C3

-#define GL_TEXTURE4                                      0x84C4

-#define GL_TEXTURE5                                      0x84C5

-#define GL_TEXTURE6                                      0x84C6

-#define GL_TEXTURE7                                      0x84C7

-#define GL_TEXTURE8                                      0x84C8

-#define GL_TEXTURE9                                      0x84C9

-#define GL_TEXTURE10                                     0x84CA

-#define GL_TEXTURE11                                     0x84CB

-#define GL_TEXTURE12                                     0x84CC

-#define GL_TEXTURE13                                     0x84CD

-#define GL_TEXTURE14                                     0x84CE

-#define GL_TEXTURE15                                     0x84CF

-#define GL_TEXTURE16                                     0x84D0

-#define GL_TEXTURE17                                     0x84D1

-#define GL_TEXTURE18                                     0x84D2

-#define GL_TEXTURE19                                     0x84D3

-#define GL_TEXTURE20                                     0x84D4

-#define GL_TEXTURE21                                     0x84D5

-#define GL_TEXTURE22                                     0x84D6

-#define GL_TEXTURE23                                     0x84D7

-#define GL_TEXTURE24                                     0x84D8

-#define GL_TEXTURE25                                     0x84D9

-#define GL_TEXTURE26                                     0x84DA

-#define GL_TEXTURE27                                     0x84DB

-#define GL_TEXTURE28                                     0x84DC

-#define GL_TEXTURE29                                     0x84DD

-#define GL_TEXTURE30                                     0x84DE

-#define GL_TEXTURE31                                     0x84DF

-#define GL_ACTIVE_TEXTURE                                0x84E0

-

-/* TextureWrapMode */

-#define GL_REPEAT                                        0x2901

-#define GL_CLAMP_TO_EDGE                                 0x812F

-#define GL_MIRRORED_REPEAT                               0x8370

-

-/* Uniform Types */

-#define GL_FLOAT_VEC2                                    0x8B50

-#define GL_FLOAT_VEC3                                    0x8B51

-#define GL_FLOAT_VEC4                                    0x8B52

-#define GL_INT_VEC2                                      0x8B53

-#define GL_INT_VEC3                                      0x8B54

-#define GL_INT_VEC4                                      0x8B55

-#define GL_BOOL                                          0x8B56

-#define GL_BOOL_VEC2                                     0x8B57

-#define GL_BOOL_VEC3                                     0x8B58

-#define GL_BOOL_VEC4                                     0x8B59

-#define GL_FLOAT_MAT2                                    0x8B5A

-#define GL_FLOAT_MAT3                                    0x8B5B

-#define GL_FLOAT_MAT4                                    0x8B5C

-#define GL_SAMPLER_2D                                    0x8B5E

-#define GL_SAMPLER_CUBE                                  0x8B60

-

-/* Vertex Arrays */

-#define GL_VERTEX_ATTRIB_ARRAY_ENABLED                   0x8622

-#define GL_VERTEX_ATTRIB_ARRAY_SIZE                      0x8623

-#define GL_VERTEX_ATTRIB_ARRAY_STRIDE                    0x8624

-#define GL_VERTEX_ATTRIB_ARRAY_TYPE                      0x8625

-#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED                0x886A

-#define GL_VERTEX_ATTRIB_ARRAY_POINTER                   0x8645

-#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING            0x889F

-

-/* Read Format */

-#define GL_IMPLEMENTATION_COLOR_READ_TYPE                0x8B9A

-#define GL_IMPLEMENTATION_COLOR_READ_FORMAT              0x8B9B

-

-/* Shader Source */

-#define GL_COMPILE_STATUS                                0x8B81

-#define GL_INFO_LOG_LENGTH                               0x8B84

-#define GL_SHADER_SOURCE_LENGTH                          0x8B88

-#define GL_SHADER_COMPILER                               0x8DFA

-

-/* Shader Binary */

-#define GL_SHADER_BINARY_FORMATS                         0x8DF8

-#define GL_NUM_SHADER_BINARY_FORMATS                     0x8DF9

-

-/* Shader Precision-Specified Types */

-#define GL_LOW_FLOAT                                     0x8DF0

-#define GL_MEDIUM_FLOAT                                  0x8DF1

-#define GL_HIGH_FLOAT                                    0x8DF2

-#define GL_LOW_INT                                       0x8DF3

-#define GL_MEDIUM_INT                                    0x8DF4

-#define GL_HIGH_INT                                      0x8DF5

-

-/* Framebuffer Object. */

-#define GL_FRAMEBUFFER                                   0x8D40

-#define GL_RENDERBUFFER                                  0x8D41

-

-#define GL_RGBA4                                         0x8056

-#define GL_RGB5_A1                                       0x8057

-#define GL_RGB565                                        0x8D62

-#define GL_DEPTH_COMPONENT16                             0x81A5

-#define GL_STENCIL_INDEX8                                0x8D48

-

-#define GL_RENDERBUFFER_WIDTH                            0x8D42

-#define GL_RENDERBUFFER_HEIGHT                           0x8D43

-#define GL_RENDERBUFFER_INTERNAL_FORMAT                  0x8D44

-#define GL_RENDERBUFFER_RED_SIZE                         0x8D50

-#define GL_RENDERBUFFER_GREEN_SIZE                       0x8D51

-#define GL_RENDERBUFFER_BLUE_SIZE                        0x8D52

-#define GL_RENDERBUFFER_ALPHA_SIZE                       0x8D53

-#define GL_RENDERBUFFER_DEPTH_SIZE                       0x8D54

-#define GL_RENDERBUFFER_STENCIL_SIZE                     0x8D55

-

-#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE            0x8CD0

-#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME            0x8CD1

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL          0x8CD2

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE  0x8CD3

-

-#define GL_COLOR_ATTACHMENT0                             0x8CE0

-#define GL_DEPTH_ATTACHMENT                              0x8D00

-#define GL_STENCIL_ATTACHMENT                            0x8D20

-

-#define GL_NONE                                          0

-

-#define GL_FRAMEBUFFER_COMPLETE                          0x8CD5

-#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT             0x8CD6

-#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT     0x8CD7

-#define GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS             0x8CD9

-#define GL_FRAMEBUFFER_UNSUPPORTED                       0x8CDD

-

-#define GL_FRAMEBUFFER_BINDING                           0x8CA6

-#define GL_RENDERBUFFER_BINDING                          0x8CA7

-#define GL_MAX_RENDERBUFFER_SIZE                         0x84E8

-

-#define GL_INVALID_FRAMEBUFFER_OPERATION                 0x0506

-

-/* OpenGL ES 3.0 */

-

-#define GL_READ_BUFFER                                   0x0C02

-#define GL_UNPACK_ROW_LENGTH                             0x0CF2

-#define GL_UNPACK_SKIP_ROWS                              0x0CF3

-#define GL_UNPACK_SKIP_PIXELS                            0x0CF4

-#define GL_PACK_ROW_LENGTH                               0x0D02

-#define GL_PACK_SKIP_ROWS                                0x0D03

-#define GL_PACK_SKIP_PIXELS                              0x0D04

-#define GL_COLOR                                         0x1800

-#define GL_DEPTH                                         0x1801

-#define GL_STENCIL                                       0x1802

-#define GL_RED                                           0x1903

-#define GL_RGB8                                          0x8051

-#define GL_RGBA8                                         0x8058

-#define GL_RGB10_A2                                      0x8059

-#define GL_TEXTURE_BINDING_3D                            0x806A

-#define GL_UNPACK_SKIP_IMAGES                            0x806D

-#define GL_UNPACK_IMAGE_HEIGHT                           0x806E

-#define GL_TEXTURE_3D                                    0x806F

-#define GL_TEXTURE_WRAP_R                                0x8072

-#define GL_MAX_3D_TEXTURE_SIZE                           0x8073

-#define GL_UNSIGNED_INT_2_10_10_10_REV                   0x8368

-#define GL_MAX_ELEMENTS_VERTICES                         0x80E8

-#define GL_MAX_ELEMENTS_INDICES                          0x80E9

-#define GL_TEXTURE_MIN_LOD                               0x813A

-#define GL_TEXTURE_MAX_LOD                               0x813B

-#define GL_TEXTURE_BASE_LEVEL                            0x813C

-#define GL_TEXTURE_MAX_LEVEL                             0x813D

-#define GL_MIN                                           0x8007

-#define GL_MAX                                           0x8008

-#define GL_DEPTH_COMPONENT24                             0x81A6

-#define GL_MAX_TEXTURE_LOD_BIAS                          0x84FD

-#define GL_TEXTURE_COMPARE_MODE                          0x884C

-#define GL_TEXTURE_COMPARE_FUNC                          0x884D

-#define GL_CURRENT_QUERY                                 0x8865

-#define GL_QUERY_RESULT                                  0x8866

-#define GL_QUERY_RESULT_AVAILABLE                        0x8867

-#define GL_BUFFER_MAPPED                                 0x88BC

-#define GL_BUFFER_MAP_POINTER                            0x88BD

-#define GL_STREAM_READ                                   0x88E1

-#define GL_STREAM_COPY                                   0x88E2

-#define GL_STATIC_READ                                   0x88E5

-#define GL_STATIC_COPY                                   0x88E6

-#define GL_DYNAMIC_READ                                  0x88E9

-#define GL_DYNAMIC_COPY                                  0x88EA

-#define GL_MAX_DRAW_BUFFERS                              0x8824

-#define GL_DRAW_BUFFER0                                  0x8825

-#define GL_DRAW_BUFFER1                                  0x8826

-#define GL_DRAW_BUFFER2                                  0x8827

-#define GL_DRAW_BUFFER3                                  0x8828

-#define GL_DRAW_BUFFER4                                  0x8829

-#define GL_DRAW_BUFFER5                                  0x882A

-#define GL_DRAW_BUFFER6                                  0x882B

-#define GL_DRAW_BUFFER7                                  0x882C

-#define GL_DRAW_BUFFER8                                  0x882D

-#define GL_DRAW_BUFFER9                                  0x882E

-#define GL_DRAW_BUFFER10                                 0x882F

-#define GL_DRAW_BUFFER11                                 0x8830

-#define GL_DRAW_BUFFER12                                 0x8831

-#define GL_DRAW_BUFFER13                                 0x8832

-#define GL_DRAW_BUFFER14                                 0x8833

-#define GL_DRAW_BUFFER15                                 0x8834

-#define GL_MAX_FRAGMENT_UNIFORM_COMPONENTS               0x8B49

-#define GL_MAX_VERTEX_UNIFORM_COMPONENTS                 0x8B4A

-#define GL_SAMPLER_3D                                    0x8B5F

-#define GL_SAMPLER_2D_SHADOW                             0x8B62

-#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT               0x8B8B

-#define GL_PIXEL_PACK_BUFFER                             0x88EB

-#define GL_PIXEL_UNPACK_BUFFER                           0x88EC

-#define GL_PIXEL_PACK_BUFFER_BINDING                     0x88ED

-#define GL_PIXEL_UNPACK_BUFFER_BINDING                   0x88EF

-#define GL_FLOAT_MAT2x3                                  0x8B65

-#define GL_FLOAT_MAT2x4                                  0x8B66

-#define GL_FLOAT_MAT3x2                                  0x8B67

-#define GL_FLOAT_MAT3x4                                  0x8B68

-#define GL_FLOAT_MAT4x2                                  0x8B69

-#define GL_FLOAT_MAT4x3                                  0x8B6A

-#define GL_SRGB                                          0x8C40

-#define GL_SRGB8                                         0x8C41

-#define GL_SRGB8_ALPHA8                                  0x8C43

-#define GL_COMPARE_REF_TO_TEXTURE                        0x884E

-#define GL_MAJOR_VERSION                                 0x821B

-#define GL_MINOR_VERSION                                 0x821C

-#define GL_NUM_EXTENSIONS                                0x821D

-#define GL_RGBA32F                                       0x8814

-#define GL_RGB32F                                        0x8815

-#define GL_RGBA16F                                       0x881A

-#define GL_RGB16F                                        0x881B

-#define GL_VERTEX_ATTRIB_ARRAY_INTEGER                   0x88FD

-#define GL_MAX_ARRAY_TEXTURE_LAYERS                      0x88FF

-#define GL_MIN_PROGRAM_TEXEL_OFFSET                      0x8904

-#define GL_MAX_PROGRAM_TEXEL_OFFSET                      0x8905

-#define GL_MAX_VARYING_COMPONENTS                        0x8B4B

-#define GL_TEXTURE_2D_ARRAY                              0x8C1A

-#define GL_TEXTURE_BINDING_2D_ARRAY                      0x8C1D

-#define GL_R11F_G11F_B10F                                0x8C3A

-#define GL_UNSIGNED_INT_10F_11F_11F_REV                  0x8C3B

-#define GL_RGB9_E5                                       0x8C3D

-#define GL_UNSIGNED_INT_5_9_9_9_REV                      0x8C3E

-#define GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH         0x8C76

-#define GL_TRANSFORM_FEEDBACK_BUFFER_MODE                0x8C7F

-#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS    0x8C80

-#define GL_TRANSFORM_FEEDBACK_VARYINGS                   0x8C83

-#define GL_TRANSFORM_FEEDBACK_BUFFER_START               0x8C84

-#define GL_TRANSFORM_FEEDBACK_BUFFER_SIZE                0x8C85

-#define GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN         0x8C88

-#define GL_RASTERIZER_DISCARD                            0x8C89

-#define GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS 0x8C8A

-#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS       0x8C8B

-#define GL_INTERLEAVED_ATTRIBS                           0x8C8C

-#define GL_SEPARATE_ATTRIBS                              0x8C8D

-#define GL_TRANSFORM_FEEDBACK_BUFFER                     0x8C8E

-#define GL_TRANSFORM_FEEDBACK_BUFFER_BINDING             0x8C8F

-#define GL_RGBA32UI                                      0x8D70

-#define GL_RGB32UI                                       0x8D71

-#define GL_RGBA16UI                                      0x8D76

-#define GL_RGB16UI                                       0x8D77

-#define GL_RGBA8UI                                       0x8D7C

-#define GL_RGB8UI                                        0x8D7D

-#define GL_RGBA32I                                       0x8D82

-#define GL_RGB32I                                        0x8D83

-#define GL_RGBA16I                                       0x8D88

-#define GL_RGB16I                                        0x8D89

-#define GL_RGBA8I                                        0x8D8E

-#define GL_RGB8I                                         0x8D8F

-#define GL_RED_INTEGER                                   0x8D94

-#define GL_RGB_INTEGER                                   0x8D98

-#define GL_RGBA_INTEGER                                  0x8D99

-#define GL_SAMPLER_2D_ARRAY                              0x8DC1

-#define GL_SAMPLER_2D_ARRAY_SHADOW                       0x8DC4

-#define GL_SAMPLER_CUBE_SHADOW                           0x8DC5

-#define GL_UNSIGNED_INT_VEC2                             0x8DC6

-#define GL_UNSIGNED_INT_VEC3                             0x8DC7

-#define GL_UNSIGNED_INT_VEC4                             0x8DC8

-#define GL_INT_SAMPLER_2D                                0x8DCA

-#define GL_INT_SAMPLER_3D                                0x8DCB

-#define GL_INT_SAMPLER_CUBE                              0x8DCC

-#define GL_INT_SAMPLER_2D_ARRAY                          0x8DCF

-#define GL_UNSIGNED_INT_SAMPLER_2D                       0x8DD2

-#define GL_UNSIGNED_INT_SAMPLER_3D                       0x8DD3

-#define GL_UNSIGNED_INT_SAMPLER_CUBE                     0x8DD4

-#define GL_UNSIGNED_INT_SAMPLER_2D_ARRAY                 0x8DD7

-#define GL_BUFFER_ACCESS_FLAGS                           0x911F

-#define GL_BUFFER_MAP_LENGTH                             0x9120

-#define GL_BUFFER_MAP_OFFSET                             0x9121

-#define GL_DEPTH_COMPONENT32F                            0x8CAC

-#define GL_DEPTH32F_STENCIL8                             0x8CAD

-#define GL_FLOAT_32_UNSIGNED_INT_24_8_REV                0x8DAD

-#define GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING         0x8210

-#define GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE         0x8211

-#define GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE               0x8212

-#define GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE             0x8213

-#define GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE              0x8214

-#define GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE             0x8215

-#define GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE             0x8216

-#define GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE           0x8217

-#define GL_FRAMEBUFFER_DEFAULT                           0x8218

-#define GL_FRAMEBUFFER_UNDEFINED                         0x8219

-#define GL_DEPTH_STENCIL_ATTACHMENT                      0x821A

-#define GL_DEPTH_STENCIL                                 0x84F9

-#define GL_UNSIGNED_INT_24_8                             0x84FA

-#define GL_DEPTH24_STENCIL8                              0x88F0

-#define GL_UNSIGNED_NORMALIZED                           0x8C17

-#define GL_DRAW_FRAMEBUFFER_BINDING                      GL_FRAMEBUFFER_BINDING

-#define GL_READ_FRAMEBUFFER                              0x8CA8

-#define GL_DRAW_FRAMEBUFFER                              0x8CA9

-#define GL_READ_FRAMEBUFFER_BINDING                      0x8CAA

-#define GL_RENDERBUFFER_SAMPLES                          0x8CAB

-#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER          0x8CD4

-#define GL_MAX_COLOR_ATTACHMENTS                         0x8CDF

-#define GL_COLOR_ATTACHMENT1                             0x8CE1

-#define GL_COLOR_ATTACHMENT2                             0x8CE2

-#define GL_COLOR_ATTACHMENT3                             0x8CE3

-#define GL_COLOR_ATTACHMENT4                             0x8CE4

-#define GL_COLOR_ATTACHMENT5                             0x8CE5

-#define GL_COLOR_ATTACHMENT6                             0x8CE6

-#define GL_COLOR_ATTACHMENT7                             0x8CE7

-#define GL_COLOR_ATTACHMENT8                             0x8CE8

-#define GL_COLOR_ATTACHMENT9                             0x8CE9

-#define GL_COLOR_ATTACHMENT10                            0x8CEA

-#define GL_COLOR_ATTACHMENT11                            0x8CEB

-#define GL_COLOR_ATTACHMENT12                            0x8CEC

-#define GL_COLOR_ATTACHMENT13                            0x8CED

-#define GL_COLOR_ATTACHMENT14                            0x8CEE

-#define GL_COLOR_ATTACHMENT15                            0x8CEF

-#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE            0x8D56

-#define GL_MAX_SAMPLES                                   0x8D57

-#define GL_HALF_FLOAT                                    0x140B

-#define GL_MAP_READ_BIT                                  0x0001

-#define GL_MAP_WRITE_BIT                                 0x0002

-#define GL_MAP_INVALIDATE_RANGE_BIT                      0x0004

-#define GL_MAP_INVALIDATE_BUFFER_BIT                     0x0008

-#define GL_MAP_FLUSH_EXPLICIT_BIT                        0x0010

-#define GL_MAP_UNSYNCHRONIZED_BIT                        0x0020

-#define GL_RG                                            0x8227

-#define GL_RG_INTEGER                                    0x8228

-#define GL_R8                                            0x8229

-#define GL_RG8                                           0x822B

-#define GL_R16F                                          0x822D

-#define GL_R32F                                          0x822E

-#define GL_RG16F                                         0x822F

-#define GL_RG32F                                         0x8230

-#define GL_R8I                                           0x8231

-#define GL_R8UI                                          0x8232

-#define GL_R16I                                          0x8233

-#define GL_R16UI                                         0x8234

-#define GL_R32I                                          0x8235

-#define GL_R32UI                                         0x8236

-#define GL_RG8I                                          0x8237

-#define GL_RG8UI                                         0x8238

-#define GL_RG16I                                         0x8239

-#define GL_RG16UI                                        0x823A

-#define GL_RG32I                                         0x823B

-#define GL_RG32UI                                        0x823C

-#define GL_VERTEX_ARRAY_BINDING                          0x85B5

-#define GL_R8_SNORM                                      0x8F94

-#define GL_RG8_SNORM                                     0x8F95

-#define GL_RGB8_SNORM                                    0x8F96

-#define GL_RGBA8_SNORM                                   0x8F97

-#define GL_SIGNED_NORMALIZED                             0x8F9C

-#define GL_PRIMITIVE_RESTART_FIXED_INDEX                 0x8D69

-#define GL_COPY_READ_BUFFER                              0x8F36

-#define GL_COPY_WRITE_BUFFER                             0x8F37

-#define GL_COPY_READ_BUFFER_BINDING                      GL_COPY_READ_BUFFER

-#define GL_COPY_WRITE_BUFFER_BINDING                     GL_COPY_WRITE_BUFFER

-#define GL_UNIFORM_BUFFER                                0x8A11

-#define GL_UNIFORM_BUFFER_BINDING                        0x8A28

-#define GL_UNIFORM_BUFFER_START                          0x8A29

-#define GL_UNIFORM_BUFFER_SIZE                           0x8A2A

-#define GL_MAX_VERTEX_UNIFORM_BLOCKS                     0x8A2B

-#define GL_MAX_FRAGMENT_UNIFORM_BLOCKS                   0x8A2D

-#define GL_MAX_COMBINED_UNIFORM_BLOCKS                   0x8A2E

-#define GL_MAX_UNIFORM_BUFFER_BINDINGS                   0x8A2F

-#define GL_MAX_UNIFORM_BLOCK_SIZE                        0x8A30

-#define GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS        0x8A31

-#define GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS      0x8A33

-#define GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT               0x8A34

-#define GL_ACTIVE_UNIFORM_BLOCK_MAX_NAME_LENGTH          0x8A35

-#define GL_ACTIVE_UNIFORM_BLOCKS                         0x8A36

-#define GL_UNIFORM_TYPE                                  0x8A37

-#define GL_UNIFORM_SIZE                                  0x8A38

-#define GL_UNIFORM_NAME_LENGTH                           0x8A39

-#define GL_UNIFORM_BLOCK_INDEX                           0x8A3A

-#define GL_UNIFORM_OFFSET                                0x8A3B

-#define GL_UNIFORM_ARRAY_STRIDE                          0x8A3C

-#define GL_UNIFORM_MATRIX_STRIDE                         0x8A3D

-#define GL_UNIFORM_IS_ROW_MAJOR                          0x8A3E

-#define GL_UNIFORM_BLOCK_BINDING                         0x8A3F

-#define GL_UNIFORM_BLOCK_DATA_SIZE                       0x8A40

-#define GL_UNIFORM_BLOCK_NAME_LENGTH                     0x8A41

-#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORMS                 0x8A42

-#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORM_INDICES          0x8A43

-#define GL_UNIFORM_BLOCK_REFERENCED_BY_VERTEX_SHADER     0x8A44

-#define GL_UNIFORM_BLOCK_REFERENCED_BY_FRAGMENT_SHADER   0x8A46

-#define GL_INVALID_INDEX                                 0xFFFFFFFFu

-#define GL_MAX_VERTEX_OUTPUT_COMPONENTS                  0x9122

-#define GL_MAX_FRAGMENT_INPUT_COMPONENTS                 0x9125

-#define GL_MAX_SERVER_WAIT_TIMEOUT                       0x9111

-#define GL_OBJECT_TYPE                                   0x9112

-#define GL_SYNC_CONDITION                                0x9113

-#define GL_SYNC_STATUS                                   0x9114

-#define GL_SYNC_FLAGS                                    0x9115

-#define GL_SYNC_FENCE                                    0x9116

-#define GL_SYNC_GPU_COMMANDS_COMPLETE                    0x9117

-#define GL_UNSIGNALED                                    0x9118

-#define GL_SIGNALED                                      0x9119

-#define GL_ALREADY_SIGNALED                              0x911A

-#define GL_TIMEOUT_EXPIRED                               0x911B

-#define GL_CONDITION_SATISFIED                           0x911C

-#define GL_WAIT_FAILED                                   0x911D

-#define GL_SYNC_FLUSH_COMMANDS_BIT                       0x00000001

-#define GL_TIMEOUT_IGNORED                               0xFFFFFFFFFFFFFFFFull

-#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR                   0x88FE

-#define GL_ANY_SAMPLES_PASSED                            0x8C2F

-#define GL_ANY_SAMPLES_PASSED_CONSERVATIVE               0x8D6A

-#define GL_SAMPLER_BINDING                               0x8919

-#define GL_RGB10_A2UI                                    0x906F

-#define GL_TEXTURE_SWIZZLE_R                             0x8E42

-#define GL_TEXTURE_SWIZZLE_G                             0x8E43

-#define GL_TEXTURE_SWIZZLE_B                             0x8E44

-#define GL_TEXTURE_SWIZZLE_A                             0x8E45

-#define GL_GREEN                                         0x1904

-#define GL_BLUE                                          0x1905

-#define GL_INT_2_10_10_10_REV                            0x8D9F

-#define GL_TRANSFORM_FEEDBACK                            0x8E22

-#define GL_TRANSFORM_FEEDBACK_PAUSED                     0x8E23

-#define GL_TRANSFORM_FEEDBACK_ACTIVE                     0x8E24

-#define GL_TRANSFORM_FEEDBACK_BINDING                    0x8E25

-#define GL_PROGRAM_BINARY_RETRIEVABLE_HINT               0x8257

-#define GL_PROGRAM_BINARY_LENGTH                         0x8741

-#define GL_NUM_PROGRAM_BINARY_FORMATS                    0x87FE

-#define GL_PROGRAM_BINARY_FORMATS                        0x87FF

-#define GL_COMPRESSED_R11_EAC                            0x9270

-#define GL_COMPRESSED_SIGNED_R11_EAC                     0x9271

-#define GL_COMPRESSED_RG11_EAC                           0x9272

-#define GL_COMPRESSED_SIGNED_RG11_EAC                    0x9273

-#define GL_COMPRESSED_RGB8_ETC2                          0x9274

-#define GL_COMPRESSED_SRGB8_ETC2                         0x9275

-#define GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2      0x9276

-#define GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2     0x9277

-#define GL_COMPRESSED_RGBA8_ETC2_EAC                     0x9278

-#define GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC              0x9279

-#define GL_TEXTURE_IMMUTABLE_FORMAT                      0x912F

-#define GL_MAX_ELEMENT_INDEX                             0x8D6B

-#define GL_NUM_SAMPLE_COUNTS                             0x9380

-#define GL_TEXTURE_IMMUTABLE_LEVELS                      0x82DF

-

-/*-------------------------------------------------------------------------

- * Entrypoint definitions

- *-----------------------------------------------------------------------*/

-

-/* OpenGL ES 2.0 */

-

-GL_APICALL void           GL_APIENTRY glActiveTexture (GLenum texture);

-GL_APICALL void           GL_APIENTRY glAttachShader (GLuint program, GLuint shader);

-GL_APICALL void           GL_APIENTRY glBindAttribLocation (GLuint program, GLuint index, const GLchar* name);

-GL_APICALL void           GL_APIENTRY glBindBuffer (GLenum target, GLuint buffer);

-GL_APICALL void           GL_APIENTRY glBindFramebuffer (GLenum target, GLuint framebuffer);

-GL_APICALL void           GL_APIENTRY glBindRenderbuffer (GLenum target, GLuint renderbuffer);

-GL_APICALL void           GL_APIENTRY glBindTexture (GLenum target, GLuint texture);

-GL_APICALL void           GL_APIENTRY glBlendColor (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha);

-GL_APICALL void           GL_APIENTRY glBlendEquation (GLenum mode);

-GL_APICALL void           GL_APIENTRY glBlendEquationSeparate (GLenum modeRGB, GLenum modeAlpha);

-GL_APICALL void           GL_APIENTRY glBlendFunc (GLenum sfactor, GLenum dfactor);

-GL_APICALL void           GL_APIENTRY glBlendFuncSeparate (GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);

-GL_APICALL void           GL_APIENTRY glBufferData (GLenum target, GLsizeiptr size, const GLvoid* data, GLenum usage);

-GL_APICALL void           GL_APIENTRY glBufferSubData (GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid* data);

-GL_APICALL GLenum         GL_APIENTRY glCheckFramebufferStatus (GLenum target);

-GL_APICALL void           GL_APIENTRY glClear (GLbitfield mask);

-GL_APICALL void           GL_APIENTRY glClearColor (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha);

-GL_APICALL void           GL_APIENTRY glClearDepthf (GLfloat depth);

-GL_APICALL void           GL_APIENTRY glClearStencil (GLint s);

-GL_APICALL void           GL_APIENTRY glColorMask (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha);

-GL_APICALL void           GL_APIENTRY glCompileShader (GLuint shader);

-GL_APICALL void           GL_APIENTRY glCompressedTexImage2D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void           GL_APIENTRY glCompressedTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void           GL_APIENTRY glCopyTexImage2D (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border);

-GL_APICALL void           GL_APIENTRY glCopyTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL GLuint         GL_APIENTRY glCreateProgram (void);

-GL_APICALL GLuint         GL_APIENTRY glCreateShader (GLenum type);

-GL_APICALL void           GL_APIENTRY glCullFace (GLenum mode);

-GL_APICALL void           GL_APIENTRY glDeleteBuffers (GLsizei n, const GLuint* buffers);

-GL_APICALL void           GL_APIENTRY glDeleteFramebuffers (GLsizei n, const GLuint* framebuffers);

-GL_APICALL void           GL_APIENTRY glDeleteProgram (GLuint program);

-GL_APICALL void           GL_APIENTRY glDeleteRenderbuffers (GLsizei n, const GLuint* renderbuffers);

-GL_APICALL void           GL_APIENTRY glDeleteShader (GLuint shader);

-GL_APICALL void           GL_APIENTRY glDeleteTextures (GLsizei n, const GLuint* textures);

-GL_APICALL void           GL_APIENTRY glDepthFunc (GLenum func);

-GL_APICALL void           GL_APIENTRY glDepthMask (GLboolean flag);

-GL_APICALL void           GL_APIENTRY glDepthRangef (GLfloat n, GLfloat f);

-GL_APICALL void           GL_APIENTRY glDetachShader (GLuint program, GLuint shader);

-GL_APICALL void           GL_APIENTRY glDisable (GLenum cap);

-GL_APICALL void           GL_APIENTRY glDisableVertexAttribArray (GLuint index);

-GL_APICALL void           GL_APIENTRY glDrawArrays (GLenum mode, GLint first, GLsizei count);

-GL_APICALL void           GL_APIENTRY glDrawElements (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices);

-GL_APICALL void           GL_APIENTRY glEnable (GLenum cap);

-GL_APICALL void           GL_APIENTRY glEnableVertexAttribArray (GLuint index);

-GL_APICALL void           GL_APIENTRY glFinish (void);

-GL_APICALL void           GL_APIENTRY glFlush (void);

-GL_APICALL void           GL_APIENTRY glFramebufferRenderbuffer (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer);

-GL_APICALL void           GL_APIENTRY glFramebufferTexture2D (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level);

-GL_APICALL void           GL_APIENTRY glFrontFace (GLenum mode);

-GL_APICALL void           GL_APIENTRY glGenBuffers (GLsizei n, GLuint* buffers);

-GL_APICALL void           GL_APIENTRY glGenerateMipmap (GLenum target);

-GL_APICALL void           GL_APIENTRY glGenFramebuffers (GLsizei n, GLuint* framebuffers);

-GL_APICALL void           GL_APIENTRY glGenRenderbuffers (GLsizei n, GLuint* renderbuffers);

-GL_APICALL void           GL_APIENTRY glGenTextures (GLsizei n, GLuint* textures);

-GL_APICALL void           GL_APIENTRY glGetActiveAttrib (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);

-GL_APICALL void           GL_APIENTRY glGetActiveUniform (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);

-GL_APICALL void           GL_APIENTRY glGetAttachedShaders (GLuint program, GLsizei maxcount, GLsizei* count, GLuint* shaders);

-GL_APICALL GLint          GL_APIENTRY glGetAttribLocation (GLuint program, const GLchar* name);

-GL_APICALL void           GL_APIENTRY glGetBooleanv (GLenum pname, GLboolean* params);

-GL_APICALL void           GL_APIENTRY glGetBufferParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL GLenum         GL_APIENTRY glGetError (void);

-GL_APICALL void           GL_APIENTRY glGetFloatv (GLenum pname, GLfloat* params);

-GL_APICALL void           GL_APIENTRY glGetFramebufferAttachmentParameteriv (GLenum target, GLenum attachment, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetIntegerv (GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetProgramiv (GLuint program, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetProgramInfoLog (GLuint program, GLsizei bufsize, GLsizei* length, GLchar* infolog);

-GL_APICALL void           GL_APIENTRY glGetRenderbufferParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetShaderiv (GLuint shader, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetShaderInfoLog (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* infolog);

-GL_APICALL void           GL_APIENTRY glGetShaderPrecisionFormat (GLenum shadertype, GLenum precisiontype, GLint* range, GLint* precision);

-GL_APICALL void           GL_APIENTRY glGetShaderSource (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* source);

-GL_APICALL const GLubyte* GL_APIENTRY glGetString (GLenum name);

-GL_APICALL void           GL_APIENTRY glGetTexParameterfv (GLenum target, GLenum pname, GLfloat* params);

-GL_APICALL void           GL_APIENTRY glGetTexParameteriv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetUniformfv (GLuint program, GLint location, GLfloat* params);

-GL_APICALL void           GL_APIENTRY glGetUniformiv (GLuint program, GLint location, GLint* params);

-GL_APICALL GLint          GL_APIENTRY glGetUniformLocation (GLuint program, const GLchar* name);

-GL_APICALL void           GL_APIENTRY glGetVertexAttribfv (GLuint index, GLenum pname, GLfloat* params);

-GL_APICALL void           GL_APIENTRY glGetVertexAttribiv (GLuint index, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetVertexAttribPointerv (GLuint index, GLenum pname, GLvoid** pointer);

-GL_APICALL void           GL_APIENTRY glHint (GLenum target, GLenum mode);

-GL_APICALL GLboolean      GL_APIENTRY glIsBuffer (GLuint buffer);

-GL_APICALL GLboolean      GL_APIENTRY glIsEnabled (GLenum cap);

-GL_APICALL GLboolean      GL_APIENTRY glIsFramebuffer (GLuint framebuffer);

-GL_APICALL GLboolean      GL_APIENTRY glIsProgram (GLuint program);

-GL_APICALL GLboolean      GL_APIENTRY glIsRenderbuffer (GLuint renderbuffer);

-GL_APICALL GLboolean      GL_APIENTRY glIsShader (GLuint shader);

-GL_APICALL GLboolean      GL_APIENTRY glIsTexture (GLuint texture);

-GL_APICALL void           GL_APIENTRY glLineWidth (GLfloat width);

-GL_APICALL void           GL_APIENTRY glLinkProgram (GLuint program);

-GL_APICALL void           GL_APIENTRY glPixelStorei (GLenum pname, GLint param);

-GL_APICALL void           GL_APIENTRY glPolygonOffset (GLfloat factor, GLfloat units);

-GL_APICALL void           GL_APIENTRY glReadPixels (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid* pixels);

-GL_APICALL void           GL_APIENTRY glReleaseShaderCompiler (void);

-GL_APICALL void           GL_APIENTRY glRenderbufferStorage (GLenum target, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glSampleCoverage (GLfloat value, GLboolean invert);

-GL_APICALL void           GL_APIENTRY glScissor (GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glShaderBinary (GLsizei n, const GLuint* shaders, GLenum binaryformat, const GLvoid* binary, GLsizei length);

-GL_APICALL void           GL_APIENTRY glShaderSource (GLuint shader, GLsizei count, const GLchar* const* string, const GLint* length);

-GL_APICALL void           GL_APIENTRY glStencilFunc (GLenum func, GLint ref, GLuint mask);

-GL_APICALL void           GL_APIENTRY glStencilFuncSeparate (GLenum face, GLenum func, GLint ref, GLuint mask);

-GL_APICALL void           GL_APIENTRY glStencilMask (GLuint mask);

-GL_APICALL void           GL_APIENTRY glStencilMaskSeparate (GLenum face, GLuint mask);

-GL_APICALL void           GL_APIENTRY glStencilOp (GLenum fail, GLenum zfail, GLenum zpass);

-GL_APICALL void           GL_APIENTRY glStencilOpSeparate (GLenum face, GLenum fail, GLenum zfail, GLenum zpass);

-GL_APICALL void           GL_APIENTRY glTexImage2D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void           GL_APIENTRY glTexParameterf (GLenum target, GLenum pname, GLfloat param);

-GL_APICALL void           GL_APIENTRY glTexParameterfv (GLenum target, GLenum pname, const GLfloat* params);

-GL_APICALL void           GL_APIENTRY glTexParameteri (GLenum target, GLenum pname, GLint param);

-GL_APICALL void           GL_APIENTRY glTexParameteriv (GLenum target, GLenum pname, const GLint* params);

-GL_APICALL void           GL_APIENTRY glTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void           GL_APIENTRY glUniform1f (GLint location, GLfloat x);

-GL_APICALL void           GL_APIENTRY glUniform1fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void           GL_APIENTRY glUniform1i (GLint location, GLint x);

-GL_APICALL void           GL_APIENTRY glUniform1iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void           GL_APIENTRY glUniform2f (GLint location, GLfloat x, GLfloat y);

-GL_APICALL void           GL_APIENTRY glUniform2fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void           GL_APIENTRY glUniform2i (GLint location, GLint x, GLint y);

-GL_APICALL void           GL_APIENTRY glUniform2iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void           GL_APIENTRY glUniform3f (GLint location, GLfloat x, GLfloat y, GLfloat z);

-GL_APICALL void           GL_APIENTRY glUniform3fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void           GL_APIENTRY glUniform3i (GLint location, GLint x, GLint y, GLint z);

-GL_APICALL void           GL_APIENTRY glUniform3iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void           GL_APIENTRY glUniform4f (GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-GL_APICALL void           GL_APIENTRY glUniform4fv (GLint location, GLsizei count, const GLfloat* v);

-GL_APICALL void           GL_APIENTRY glUniform4i (GLint location, GLint x, GLint y, GLint z, GLint w);

-GL_APICALL void           GL_APIENTRY glUniform4iv (GLint location, GLsizei count, const GLint* v);

-GL_APICALL void           GL_APIENTRY glUniformMatrix2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUseProgram (GLuint program);

-GL_APICALL void           GL_APIENTRY glValidateProgram (GLuint program);

-GL_APICALL void           GL_APIENTRY glVertexAttrib1f (GLuint indx, GLfloat x);

-GL_APICALL void           GL_APIENTRY glVertexAttrib1fv (GLuint indx, const GLfloat* values);

-GL_APICALL void           GL_APIENTRY glVertexAttrib2f (GLuint indx, GLfloat x, GLfloat y);

-GL_APICALL void           GL_APIENTRY glVertexAttrib2fv (GLuint indx, const GLfloat* values);

-GL_APICALL void           GL_APIENTRY glVertexAttrib3f (GLuint indx, GLfloat x, GLfloat y, GLfloat z);

-GL_APICALL void           GL_APIENTRY glVertexAttrib3fv (GLuint indx, const GLfloat* values);

-GL_APICALL void           GL_APIENTRY glVertexAttrib4f (GLuint indx, GLfloat x, GLfloat y, GLfloat z, GLfloat w);

-GL_APICALL void           GL_APIENTRY glVertexAttrib4fv (GLuint indx, const GLfloat* values);

-GL_APICALL void           GL_APIENTRY glVertexAttribPointer (GLuint indx, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid* ptr);

-GL_APICALL void           GL_APIENTRY glViewport (GLint x, GLint y, GLsizei width, GLsizei height);

-

-/* OpenGL ES 3.0 */

-

-GL_APICALL void           GL_APIENTRY glReadBuffer (GLenum mode);

-GL_APICALL void           GL_APIENTRY glDrawRangeElements (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid* indices);

-GL_APICALL void           GL_APIENTRY glTexImage3D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void           GL_APIENTRY glTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);

-GL_APICALL void           GL_APIENTRY glCopyTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glCompressedTexImage3D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void           GL_APIENTRY glCompressedTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);

-GL_APICALL void           GL_APIENTRY glGenQueries (GLsizei n, GLuint* ids);

-GL_APICALL void           GL_APIENTRY glDeleteQueries (GLsizei n, const GLuint* ids);

-GL_APICALL GLboolean      GL_APIENTRY glIsQuery (GLuint id);

-GL_APICALL void           GL_APIENTRY glBeginQuery (GLenum target, GLuint id);

-GL_APICALL void           GL_APIENTRY glEndQuery (GLenum target);

-GL_APICALL void           GL_APIENTRY glGetQueryiv (GLenum target, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetQueryObjectuiv (GLuint id, GLenum pname, GLuint* params);

-GL_APICALL GLboolean      GL_APIENTRY glUnmapBuffer (GLenum target);

-GL_APICALL void           GL_APIENTRY glGetBufferPointerv (GLenum target, GLenum pname, GLvoid** params);

-GL_APICALL void           GL_APIENTRY glDrawBuffers (GLsizei n, const GLenum* bufs);

-GL_APICALL void           GL_APIENTRY glUniformMatrix2x3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix3x2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix2x4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix4x2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix3x4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glUniformMatrix4x3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glBlitFramebuffer (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);

-GL_APICALL void           GL_APIENTRY glRenderbufferStorageMultisample (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glFramebufferTextureLayer (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer);

-GL_APICALL GLvoid*        GL_APIENTRY glMapBufferRange (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);

-GL_APICALL void           GL_APIENTRY glFlushMappedBufferRange (GLenum target, GLintptr offset, GLsizeiptr length);

-GL_APICALL void           GL_APIENTRY glBindVertexArray (GLuint array);

-GL_APICALL void           GL_APIENTRY glDeleteVertexArrays (GLsizei n, const GLuint* arrays);

-GL_APICALL void           GL_APIENTRY glGenVertexArrays (GLsizei n, GLuint* arrays);

-GL_APICALL GLboolean      GL_APIENTRY glIsVertexArray (GLuint array);

-GL_APICALL void           GL_APIENTRY glGetIntegeri_v (GLenum target, GLuint index, GLint* data);

-GL_APICALL void           GL_APIENTRY glBeginTransformFeedback (GLenum primitiveMode);

-GL_APICALL void           GL_APIENTRY glEndTransformFeedback (void);

-GL_APICALL void           GL_APIENTRY glBindBufferRange (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size);

-GL_APICALL void           GL_APIENTRY glBindBufferBase (GLenum target, GLuint index, GLuint buffer);

-GL_APICALL void           GL_APIENTRY glTransformFeedbackVaryings (GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode);

-GL_APICALL void           GL_APIENTRY glGetTransformFeedbackVarying (GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size, GLenum* type, GLchar* name);

-GL_APICALL void           GL_APIENTRY glVertexAttribIPointer (GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid* pointer);

-GL_APICALL void           GL_APIENTRY glGetVertexAttribIiv (GLuint index, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetVertexAttribIuiv (GLuint index, GLenum pname, GLuint* params);

-GL_APICALL void           GL_APIENTRY glVertexAttribI4i (GLuint index, GLint x, GLint y, GLint z, GLint w);

-GL_APICALL void           GL_APIENTRY glVertexAttribI4ui (GLuint index, GLuint x, GLuint y, GLuint z, GLuint w);

-GL_APICALL void           GL_APIENTRY glVertexAttribI4iv (GLuint index, const GLint* v);

-GL_APICALL void           GL_APIENTRY glVertexAttribI4uiv (GLuint index, const GLuint* v);

-GL_APICALL void           GL_APIENTRY glGetUniformuiv (GLuint program, GLint location, GLuint* params);

-GL_APICALL GLint          GL_APIENTRY glGetFragDataLocation (GLuint program, const GLchar *name);

-GL_APICALL void           GL_APIENTRY glUniform1ui (GLint location, GLuint v0);

-GL_APICALL void           GL_APIENTRY glUniform2ui (GLint location, GLuint v0, GLuint v1);

-GL_APICALL void           GL_APIENTRY glUniform3ui (GLint location, GLuint v0, GLuint v1, GLuint v2);

-GL_APICALL void           GL_APIENTRY glUniform4ui (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3);

-GL_APICALL void           GL_APIENTRY glUniform1uiv (GLint location, GLsizei count, const GLuint* value);

-GL_APICALL void           GL_APIENTRY glUniform2uiv (GLint location, GLsizei count, const GLuint* value);

-GL_APICALL void           GL_APIENTRY glUniform3uiv (GLint location, GLsizei count, const GLuint* value);

-GL_APICALL void           GL_APIENTRY glUniform4uiv (GLint location, GLsizei count, const GLuint* value);

-GL_APICALL void           GL_APIENTRY glClearBufferiv (GLenum buffer, GLint drawbuffer, const GLint* value);

-GL_APICALL void           GL_APIENTRY glClearBufferuiv (GLenum buffer, GLint drawbuffer, const GLuint* value);

-GL_APICALL void           GL_APIENTRY glClearBufferfv (GLenum buffer, GLint drawbuffer, const GLfloat* value);

-GL_APICALL void           GL_APIENTRY glClearBufferfi (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);

-GL_APICALL const GLubyte* GL_APIENTRY glGetStringi (GLenum name, GLuint index);

-GL_APICALL void           GL_APIENTRY glCopyBufferSubData (GLenum readTarget, GLenum writeTarget, GLintptr readOffset, GLintptr writeOffset, GLsizeiptr size);

-GL_APICALL void           GL_APIENTRY glGetUniformIndices (GLuint program, GLsizei uniformCount, const GLchar* const* uniformNames, GLuint* uniformIndices);

-GL_APICALL void           GL_APIENTRY glGetActiveUniformsiv (GLuint program, GLsizei uniformCount, const GLuint* uniformIndices, GLenum pname, GLint* params);

-GL_APICALL GLuint         GL_APIENTRY glGetUniformBlockIndex (GLuint program, const GLchar* uniformBlockName);

-GL_APICALL void           GL_APIENTRY glGetActiveUniformBlockiv (GLuint program, GLuint uniformBlockIndex, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetActiveUniformBlockName (GLuint program, GLuint uniformBlockIndex, GLsizei bufSize, GLsizei* length, GLchar* uniformBlockName);

-GL_APICALL void           GL_APIENTRY glUniformBlockBinding (GLuint program, GLuint uniformBlockIndex, GLuint uniformBlockBinding);

-GL_APICALL void           GL_APIENTRY glDrawArraysInstanced (GLenum mode, GLint first, GLsizei count, GLsizei instanceCount);

-GL_APICALL void           GL_APIENTRY glDrawElementsInstanced (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices, GLsizei instanceCount);

-GL_APICALL GLsync         GL_APIENTRY glFenceSync (GLenum condition, GLbitfield flags);

-GL_APICALL GLboolean      GL_APIENTRY glIsSync (GLsync sync);

-GL_APICALL void           GL_APIENTRY glDeleteSync (GLsync sync);

-GL_APICALL GLenum         GL_APIENTRY glClientWaitSync (GLsync sync, GLbitfield flags, GLuint64 timeout);

-GL_APICALL void           GL_APIENTRY glWaitSync (GLsync sync, GLbitfield flags, GLuint64 timeout);

-GL_APICALL void           GL_APIENTRY glGetInteger64v (GLenum pname, GLint64* params);

-GL_APICALL void           GL_APIENTRY glGetSynciv (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei* length, GLint* values);

-GL_APICALL void           GL_APIENTRY glGetInteger64i_v (GLenum target, GLuint index, GLint64* data);

-GL_APICALL void           GL_APIENTRY glGetBufferParameteri64v (GLenum target, GLenum pname, GLint64* params);

-GL_APICALL void           GL_APIENTRY glGenSamplers (GLsizei count, GLuint* samplers);

-GL_APICALL void           GL_APIENTRY glDeleteSamplers (GLsizei count, const GLuint* samplers);

-GL_APICALL GLboolean      GL_APIENTRY glIsSampler (GLuint sampler);

-GL_APICALL void           GL_APIENTRY glBindSampler (GLuint unit, GLuint sampler);

-GL_APICALL void           GL_APIENTRY glSamplerParameteri (GLuint sampler, GLenum pname, GLint param);

-GL_APICALL void           GL_APIENTRY glSamplerParameteriv (GLuint sampler, GLenum pname, const GLint* param);

-GL_APICALL void           GL_APIENTRY glSamplerParameterf (GLuint sampler, GLenum pname, GLfloat param);

-GL_APICALL void           GL_APIENTRY glSamplerParameterfv (GLuint sampler, GLenum pname, const GLfloat* param);

-GL_APICALL void           GL_APIENTRY glGetSamplerParameteriv (GLuint sampler, GLenum pname, GLint* params);

-GL_APICALL void           GL_APIENTRY glGetSamplerParameterfv (GLuint sampler, GLenum pname, GLfloat* params);

-GL_APICALL void           GL_APIENTRY glVertexAttribDivisor (GLuint index, GLuint divisor);

-GL_APICALL void           GL_APIENTRY glBindTransformFeedback (GLenum target, GLuint id);

-GL_APICALL void           GL_APIENTRY glDeleteTransformFeedbacks (GLsizei n, const GLuint* ids);

-GL_APICALL void           GL_APIENTRY glGenTransformFeedbacks (GLsizei n, GLuint* ids);

-GL_APICALL GLboolean      GL_APIENTRY glIsTransformFeedback (GLuint id);

-GL_APICALL void           GL_APIENTRY glPauseTransformFeedback (void);

-GL_APICALL void           GL_APIENTRY glResumeTransformFeedback (void);

-GL_APICALL void           GL_APIENTRY glGetProgramBinary (GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, GLvoid* binary);

-GL_APICALL void           GL_APIENTRY glProgramBinary (GLuint program, GLenum binaryFormat, const GLvoid* binary, GLsizei length);

-GL_APICALL void           GL_APIENTRY glProgramParameteri (GLuint program, GLenum pname, GLint value);

-GL_APICALL void           GL_APIENTRY glInvalidateFramebuffer (GLenum target, GLsizei numAttachments, const GLenum* attachments);

-GL_APICALL void           GL_APIENTRY glInvalidateSubFramebuffer (GLenum target, GLsizei numAttachments, const GLenum* attachments, GLint x, GLint y, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glTexStorage2D (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);

-GL_APICALL void           GL_APIENTRY glTexStorage3D (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);

-GL_APICALL void           GL_APIENTRY glGetInternalformativ (GLenum target, GLenum internalformat, GLenum pname, GLsizei bufSize, GLint* params);

-

-#ifdef __cplusplus

-}

-#endif

-

-#endif

+#ifndef __gl3_h_
+#define __gl3_h_
+
+/* 
+ * gl3.h last updated on $Date: 2013-02-12 14:37:24 -0800 (Tue, 12 Feb 2013) $
+ */
+
+#include <GLES3/gl3platform.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+** Copyright (c) 2007-2013 The Khronos Group Inc.
+**
+** Permission is hereby granted, free of charge, to any person obtaining a
+** copy of this software and/or associated documentation files (the
+** "Materials"), to deal in the Materials without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Materials, and to
+** permit persons to whom the Materials are furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be included
+** in all copies or substantial portions of the Materials.
+**
+** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
+*/
+
+/*-------------------------------------------------------------------------
+ * Data type definitions
+ *-----------------------------------------------------------------------*/
+
+/* OpenGL ES 2.0 */
+
+typedef void             GLvoid;
+typedef char             GLchar;
+typedef unsigned int     GLenum;
+typedef unsigned char    GLboolean;
+typedef unsigned int     GLbitfield;
+typedef khronos_int8_t   GLbyte;
+typedef short            GLshort;
+typedef int              GLint;
+typedef int              GLsizei;
+typedef khronos_uint8_t  GLubyte;
+typedef unsigned short   GLushort;
+typedef unsigned int     GLuint;
+typedef khronos_float_t  GLfloat;
+typedef khronos_float_t  GLclampf;
+typedef khronos_int32_t  GLfixed;
+typedef khronos_intptr_t GLintptr;
+typedef khronos_ssize_t  GLsizeiptr;
+
+/* OpenGL ES 3.0 */
+
+typedef unsigned short   GLhalf;
+typedef khronos_int64_t  GLint64;
+typedef khronos_uint64_t GLuint64;
+typedef struct __GLsync *GLsync;
+
+/*-------------------------------------------------------------------------
+ * Token definitions
+ *-----------------------------------------------------------------------*/
+
+/* OpenGL ES core versions */
+#define GL_ES_VERSION_3_0                                1
+#define GL_ES_VERSION_2_0                                1
+
+/* OpenGL ES 2.0 */
+
+/* ClearBufferMask */
+#define GL_DEPTH_BUFFER_BIT                              0x00000100
+#define GL_STENCIL_BUFFER_BIT                            0x00000400
+#define GL_COLOR_BUFFER_BIT                              0x00004000
+
+/* Boolean */
+#define GL_FALSE                                         0
+#define GL_TRUE                                          1
+
+/* BeginMode */
+#define GL_POINTS                                        0x0000
+#define GL_LINES                                         0x0001
+#define GL_LINE_LOOP                                     0x0002
+#define GL_LINE_STRIP                                    0x0003
+#define GL_TRIANGLES                                     0x0004
+#define GL_TRIANGLE_STRIP                                0x0005
+#define GL_TRIANGLE_FAN                                  0x0006
+
+/* BlendingFactorDest */
+#define GL_ZERO                                          0
+#define GL_ONE                                           1
+#define GL_SRC_COLOR                                     0x0300
+#define GL_ONE_MINUS_SRC_COLOR                           0x0301
+#define GL_SRC_ALPHA                                     0x0302
+#define GL_ONE_MINUS_SRC_ALPHA                           0x0303
+#define GL_DST_ALPHA                                     0x0304
+#define GL_ONE_MINUS_DST_ALPHA                           0x0305
+
+/* BlendingFactorSrc */
+/*      GL_ZERO */
+/*      GL_ONE */
+#define GL_DST_COLOR                                     0x0306
+#define GL_ONE_MINUS_DST_COLOR                           0x0307
+#define GL_SRC_ALPHA_SATURATE                            0x0308
+/*      GL_SRC_ALPHA */
+/*      GL_ONE_MINUS_SRC_ALPHA */
+/*      GL_DST_ALPHA */
+/*      GL_ONE_MINUS_DST_ALPHA */
+
+/* BlendEquationSeparate */
+#define GL_FUNC_ADD                                      0x8006
+#define GL_BLEND_EQUATION                                0x8009
+#define GL_BLEND_EQUATION_RGB                            0x8009    /* same as BLEND_EQUATION */
+#define GL_BLEND_EQUATION_ALPHA                          0x883D
+
+/* BlendSubtract */
+#define GL_FUNC_SUBTRACT                                 0x800A
+#define GL_FUNC_REVERSE_SUBTRACT                         0x800B
+
+/* Separate Blend Functions */
+#define GL_BLEND_DST_RGB                                 0x80C8
+#define GL_BLEND_SRC_RGB                                 0x80C9
+#define GL_BLEND_DST_ALPHA                               0x80CA
+#define GL_BLEND_SRC_ALPHA                               0x80CB
+#define GL_CONSTANT_COLOR                                0x8001
+#define GL_ONE_MINUS_CONSTANT_COLOR                      0x8002
+#define GL_CONSTANT_ALPHA                                0x8003
+#define GL_ONE_MINUS_CONSTANT_ALPHA                      0x8004
+#define GL_BLEND_COLOR                                   0x8005
+
+/* Buffer Objects */
+#define GL_ARRAY_BUFFER                                  0x8892
+#define GL_ELEMENT_ARRAY_BUFFER                          0x8893
+#define GL_ARRAY_BUFFER_BINDING                          0x8894
+#define GL_ELEMENT_ARRAY_BUFFER_BINDING                  0x8895
+
+#define GL_STREAM_DRAW                                   0x88E0
+#define GL_STATIC_DRAW                                   0x88E4
+#define GL_DYNAMIC_DRAW                                  0x88E8
+
+#define GL_BUFFER_SIZE                                   0x8764
+#define GL_BUFFER_USAGE                                  0x8765
+
+#define GL_CURRENT_VERTEX_ATTRIB                         0x8626
+
+/* CullFaceMode */
+#define GL_FRONT                                         0x0404
+#define GL_BACK                                          0x0405
+#define GL_FRONT_AND_BACK                                0x0408
+
+/* DepthFunction */
+/*      GL_NEVER */
+/*      GL_LESS */
+/*      GL_EQUAL */
+/*      GL_LEQUAL */
+/*      GL_GREATER */
+/*      GL_NOTEQUAL */
+/*      GL_GEQUAL */
+/*      GL_ALWAYS */
+
+/* EnableCap */
+#define GL_TEXTURE_2D                                    0x0DE1
+#define GL_CULL_FACE                                     0x0B44
+#define GL_BLEND                                         0x0BE2
+#define GL_DITHER                                        0x0BD0
+#define GL_STENCIL_TEST                                  0x0B90
+#define GL_DEPTH_TEST                                    0x0B71
+#define GL_SCISSOR_TEST                                  0x0C11
+#define GL_POLYGON_OFFSET_FILL                           0x8037
+#define GL_SAMPLE_ALPHA_TO_COVERAGE                      0x809E
+#define GL_SAMPLE_COVERAGE                               0x80A0
+
+/* ErrorCode */
+#define GL_NO_ERROR                                      0
+#define GL_INVALID_ENUM                                  0x0500
+#define GL_INVALID_VALUE                                 0x0501
+#define GL_INVALID_OPERATION                             0x0502
+#define GL_OUT_OF_MEMORY                                 0x0505
+
+/* FrontFaceDirection */
+#define GL_CW                                            0x0900
+#define GL_CCW                                           0x0901
+
+/* GetPName */
+#define GL_LINE_WIDTH                                    0x0B21
+#define GL_ALIASED_POINT_SIZE_RANGE                      0x846D
+#define GL_ALIASED_LINE_WIDTH_RANGE                      0x846E
+#define GL_CULL_FACE_MODE                                0x0B45
+#define GL_FRONT_FACE                                    0x0B46
+#define GL_DEPTH_RANGE                                   0x0B70
+#define GL_DEPTH_WRITEMASK                               0x0B72
+#define GL_DEPTH_CLEAR_VALUE                             0x0B73
+#define GL_DEPTH_FUNC                                    0x0B74
+#define GL_STENCIL_CLEAR_VALUE                           0x0B91
+#define GL_STENCIL_FUNC                                  0x0B92
+#define GL_STENCIL_FAIL                                  0x0B94
+#define GL_STENCIL_PASS_DEPTH_FAIL                       0x0B95
+#define GL_STENCIL_PASS_DEPTH_PASS                       0x0B96
+#define GL_STENCIL_REF                                   0x0B97
+#define GL_STENCIL_VALUE_MASK                            0x0B93
+#define GL_STENCIL_WRITEMASK                             0x0B98
+#define GL_STENCIL_BACK_FUNC                             0x8800
+#define GL_STENCIL_BACK_FAIL                             0x8801
+#define GL_STENCIL_BACK_PASS_DEPTH_FAIL                  0x8802
+#define GL_STENCIL_BACK_PASS_DEPTH_PASS                  0x8803
+#define GL_STENCIL_BACK_REF                              0x8CA3
+#define GL_STENCIL_BACK_VALUE_MASK                       0x8CA4
+#define GL_STENCIL_BACK_WRITEMASK                        0x8CA5
+#define GL_VIEWPORT                                      0x0BA2
+#define GL_SCISSOR_BOX                                   0x0C10
+/*      GL_SCISSOR_TEST */
+#define GL_COLOR_CLEAR_VALUE                             0x0C22
+#define GL_COLOR_WRITEMASK                               0x0C23
+#define GL_UNPACK_ALIGNMENT                              0x0CF5
+#define GL_PACK_ALIGNMENT                                0x0D05
+#define GL_MAX_TEXTURE_SIZE                              0x0D33
+#define GL_MAX_VIEWPORT_DIMS                             0x0D3A
+#define GL_SUBPIXEL_BITS                                 0x0D50
+#define GL_RED_BITS                                      0x0D52
+#define GL_GREEN_BITS                                    0x0D53
+#define GL_BLUE_BITS                                     0x0D54
+#define GL_ALPHA_BITS                                    0x0D55
+#define GL_DEPTH_BITS                                    0x0D56
+#define GL_STENCIL_BITS                                  0x0D57
+#define GL_POLYGON_OFFSET_UNITS                          0x2A00
+/*      GL_POLYGON_OFFSET_FILL */
+#define GL_POLYGON_OFFSET_FACTOR                         0x8038
+#define GL_TEXTURE_BINDING_2D                            0x8069
+#define GL_SAMPLE_BUFFERS                                0x80A8
+#define GL_SAMPLES                                       0x80A9
+#define GL_SAMPLE_COVERAGE_VALUE                         0x80AA
+#define GL_SAMPLE_COVERAGE_INVERT                        0x80AB
+
+/* GetTextureParameter */
+/*      GL_TEXTURE_MAG_FILTER */
+/*      GL_TEXTURE_MIN_FILTER */
+/*      GL_TEXTURE_WRAP_S */
+/*      GL_TEXTURE_WRAP_T */
+
+#define GL_NUM_COMPRESSED_TEXTURE_FORMATS                0x86A2
+#define GL_COMPRESSED_TEXTURE_FORMATS                    0x86A3
+
+/* HintMode */
+#define GL_DONT_CARE                                     0x1100
+#define GL_FASTEST                                       0x1101
+#define GL_NICEST                                        0x1102
+
+/* HintTarget */
+#define GL_GENERATE_MIPMAP_HINT                          0x8192
+
+/* DataType */
+#define GL_BYTE                                          0x1400
+#define GL_UNSIGNED_BYTE                                 0x1401
+#define GL_SHORT                                         0x1402
+#define GL_UNSIGNED_SHORT                                0x1403
+#define GL_INT                                           0x1404
+#define GL_UNSIGNED_INT                                  0x1405
+#define GL_FLOAT                                         0x1406
+#define GL_FIXED                                         0x140C
+
+/* PixelFormat */
+#define GL_DEPTH_COMPONENT                               0x1902
+#define GL_ALPHA                                         0x1906
+#define GL_RGB                                           0x1907
+#define GL_RGBA                                          0x1908
+#define GL_LUMINANCE                                     0x1909
+#define GL_LUMINANCE_ALPHA                               0x190A
+
+/* PixelType */
+/*      GL_UNSIGNED_BYTE */
+#define GL_UNSIGNED_SHORT_4_4_4_4                        0x8033
+#define GL_UNSIGNED_SHORT_5_5_5_1                        0x8034
+#define GL_UNSIGNED_SHORT_5_6_5                          0x8363
+
+/* Shaders */
+#define GL_FRAGMENT_SHADER                               0x8B30
+#define GL_VERTEX_SHADER                                 0x8B31
+#define GL_MAX_VERTEX_ATTRIBS                            0x8869
+#define GL_MAX_VERTEX_UNIFORM_VECTORS                    0x8DFB
+#define GL_MAX_VARYING_VECTORS                           0x8DFC
+#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS              0x8B4D
+#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS                0x8B4C
+#define GL_MAX_TEXTURE_IMAGE_UNITS                       0x8872
+#define GL_MAX_FRAGMENT_UNIFORM_VECTORS                  0x8DFD
+#define GL_SHADER_TYPE                                   0x8B4F
+#define GL_DELETE_STATUS                                 0x8B80
+#define GL_LINK_STATUS                                   0x8B82
+#define GL_VALIDATE_STATUS                               0x8B83
+#define GL_ATTACHED_SHADERS                              0x8B85
+#define GL_ACTIVE_UNIFORMS                               0x8B86
+#define GL_ACTIVE_UNIFORM_MAX_LENGTH                     0x8B87
+#define GL_ACTIVE_ATTRIBUTES                             0x8B89
+#define GL_ACTIVE_ATTRIBUTE_MAX_LENGTH                   0x8B8A
+#define GL_SHADING_LANGUAGE_VERSION                      0x8B8C
+#define GL_CURRENT_PROGRAM                               0x8B8D
+
+/* StencilFunction */
+#define GL_NEVER                                         0x0200
+#define GL_LESS                                          0x0201
+#define GL_EQUAL                                         0x0202
+#define GL_LEQUAL                                        0x0203
+#define GL_GREATER                                       0x0204
+#define GL_NOTEQUAL                                      0x0205
+#define GL_GEQUAL                                        0x0206
+#define GL_ALWAYS                                        0x0207
+
+/* StencilOp */
+/*      GL_ZERO */
+#define GL_KEEP                                          0x1E00
+#define GL_REPLACE                                       0x1E01
+#define GL_INCR                                          0x1E02
+#define GL_DECR                                          0x1E03
+#define GL_INVERT                                        0x150A
+#define GL_INCR_WRAP                                     0x8507
+#define GL_DECR_WRAP                                     0x8508
+
+/* StringName */
+#define GL_VENDOR                                        0x1F00
+#define GL_RENDERER                                      0x1F01
+#define GL_VERSION                                       0x1F02
+#define GL_EXTENSIONS                                    0x1F03
+
+/* TextureMagFilter */
+#define GL_NEAREST                                       0x2600
+#define GL_LINEAR                                        0x2601
+
+/* TextureMinFilter */
+/*      GL_NEAREST */
+/*      GL_LINEAR */
+#define GL_NEAREST_MIPMAP_NEAREST                        0x2700
+#define GL_LINEAR_MIPMAP_NEAREST                         0x2701
+#define GL_NEAREST_MIPMAP_LINEAR                         0x2702
+#define GL_LINEAR_MIPMAP_LINEAR                          0x2703
+
+/* TextureParameterName */
+#define GL_TEXTURE_MAG_FILTER                            0x2800
+#define GL_TEXTURE_MIN_FILTER                            0x2801
+#define GL_TEXTURE_WRAP_S                                0x2802
+#define GL_TEXTURE_WRAP_T                                0x2803
+
+/* TextureTarget */
+/*      GL_TEXTURE_2D */
+#define GL_TEXTURE                                       0x1702
+
+#define GL_TEXTURE_CUBE_MAP                              0x8513
+#define GL_TEXTURE_BINDING_CUBE_MAP                      0x8514
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_X                   0x8515
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X                   0x8516
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y                   0x8517
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y                   0x8518
+#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z                   0x8519
+#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z                   0x851A
+#define GL_MAX_CUBE_MAP_TEXTURE_SIZE                     0x851C
+
+/* TextureUnit */
+#define GL_TEXTURE0                                      0x84C0
+#define GL_TEXTURE1                                      0x84C1
+#define GL_TEXTURE2                                      0x84C2
+#define GL_TEXTURE3                                      0x84C3
+#define GL_TEXTURE4                                      0x84C4
+#define GL_TEXTURE5                                      0x84C5
+#define GL_TEXTURE6                                      0x84C6
+#define GL_TEXTURE7                                      0x84C7
+#define GL_TEXTURE8                                      0x84C8
+#define GL_TEXTURE9                                      0x84C9
+#define GL_TEXTURE10                                     0x84CA
+#define GL_TEXTURE11                                     0x84CB
+#define GL_TEXTURE12                                     0x84CC
+#define GL_TEXTURE13                                     0x84CD
+#define GL_TEXTURE14                                     0x84CE
+#define GL_TEXTURE15                                     0x84CF
+#define GL_TEXTURE16                                     0x84D0
+#define GL_TEXTURE17                                     0x84D1
+#define GL_TEXTURE18                                     0x84D2
+#define GL_TEXTURE19                                     0x84D3
+#define GL_TEXTURE20                                     0x84D4
+#define GL_TEXTURE21                                     0x84D5
+#define GL_TEXTURE22                                     0x84D6
+#define GL_TEXTURE23                                     0x84D7
+#define GL_TEXTURE24                                     0x84D8
+#define GL_TEXTURE25                                     0x84D9
+#define GL_TEXTURE26                                     0x84DA
+#define GL_TEXTURE27                                     0x84DB
+#define GL_TEXTURE28                                     0x84DC
+#define GL_TEXTURE29                                     0x84DD
+#define GL_TEXTURE30                                     0x84DE
+#define GL_TEXTURE31                                     0x84DF
+#define GL_ACTIVE_TEXTURE                                0x84E0
+
+/* TextureWrapMode */
+#define GL_REPEAT                                        0x2901
+#define GL_CLAMP_TO_EDGE                                 0x812F
+#define GL_MIRRORED_REPEAT                               0x8370
+
+/* Uniform Types */
+#define GL_FLOAT_VEC2                                    0x8B50
+#define GL_FLOAT_VEC3                                    0x8B51
+#define GL_FLOAT_VEC4                                    0x8B52
+#define GL_INT_VEC2                                      0x8B53
+#define GL_INT_VEC3                                      0x8B54
+#define GL_INT_VEC4                                      0x8B55
+#define GL_BOOL                                          0x8B56
+#define GL_BOOL_VEC2                                     0x8B57
+#define GL_BOOL_VEC3                                     0x8B58
+#define GL_BOOL_VEC4                                     0x8B59
+#define GL_FLOAT_MAT2                                    0x8B5A
+#define GL_FLOAT_MAT3                                    0x8B5B
+#define GL_FLOAT_MAT4                                    0x8B5C
+#define GL_SAMPLER_2D                                    0x8B5E
+#define GL_SAMPLER_CUBE                                  0x8B60
+
+/* Vertex Arrays */
+#define GL_VERTEX_ATTRIB_ARRAY_ENABLED                   0x8622
+#define GL_VERTEX_ATTRIB_ARRAY_SIZE                      0x8623
+#define GL_VERTEX_ATTRIB_ARRAY_STRIDE                    0x8624
+#define GL_VERTEX_ATTRIB_ARRAY_TYPE                      0x8625
+#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED                0x886A
+#define GL_VERTEX_ATTRIB_ARRAY_POINTER                   0x8645
+#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING            0x889F
+
+/* Read Format */
+#define GL_IMPLEMENTATION_COLOR_READ_TYPE                0x8B9A
+#define GL_IMPLEMENTATION_COLOR_READ_FORMAT              0x8B9B
+
+/* Shader Source */
+#define GL_COMPILE_STATUS                                0x8B81
+#define GL_INFO_LOG_LENGTH                               0x8B84
+#define GL_SHADER_SOURCE_LENGTH                          0x8B88
+#define GL_SHADER_COMPILER                               0x8DFA
+
+/* Shader Binary */
+#define GL_SHADER_BINARY_FORMATS                         0x8DF8
+#define GL_NUM_SHADER_BINARY_FORMATS                     0x8DF9
+
+/* Shader Precision-Specified Types */
+#define GL_LOW_FLOAT                                     0x8DF0
+#define GL_MEDIUM_FLOAT                                  0x8DF1
+#define GL_HIGH_FLOAT                                    0x8DF2
+#define GL_LOW_INT                                       0x8DF3
+#define GL_MEDIUM_INT                                    0x8DF4
+#define GL_HIGH_INT                                      0x8DF5
+
+/* Framebuffer Object. */
+#define GL_FRAMEBUFFER                                   0x8D40
+#define GL_RENDERBUFFER                                  0x8D41
+
+#define GL_RGBA4                                         0x8056
+#define GL_RGB5_A1                                       0x8057
+#define GL_RGB565                                        0x8D62
+#define GL_DEPTH_COMPONENT16                             0x81A5
+#define GL_STENCIL_INDEX8                                0x8D48
+
+#define GL_RENDERBUFFER_WIDTH                            0x8D42
+#define GL_RENDERBUFFER_HEIGHT                           0x8D43
+#define GL_RENDERBUFFER_INTERNAL_FORMAT                  0x8D44
+#define GL_RENDERBUFFER_RED_SIZE                         0x8D50
+#define GL_RENDERBUFFER_GREEN_SIZE                       0x8D51
+#define GL_RENDERBUFFER_BLUE_SIZE                        0x8D52
+#define GL_RENDERBUFFER_ALPHA_SIZE                       0x8D53
+#define GL_RENDERBUFFER_DEPTH_SIZE                       0x8D54
+#define GL_RENDERBUFFER_STENCIL_SIZE                     0x8D55
+
+#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE            0x8CD0
+#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME            0x8CD1
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL          0x8CD2
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE  0x8CD3
+
+#define GL_COLOR_ATTACHMENT0                             0x8CE0
+#define GL_DEPTH_ATTACHMENT                              0x8D00
+#define GL_STENCIL_ATTACHMENT                            0x8D20
+
+#define GL_NONE                                          0
+
+#define GL_FRAMEBUFFER_COMPLETE                          0x8CD5
+#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT             0x8CD6
+#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT     0x8CD7
+#define GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS             0x8CD9
+#define GL_FRAMEBUFFER_UNSUPPORTED                       0x8CDD
+
+#define GL_FRAMEBUFFER_BINDING                           0x8CA6
+#define GL_RENDERBUFFER_BINDING                          0x8CA7
+#define GL_MAX_RENDERBUFFER_SIZE                         0x84E8
+
+#define GL_INVALID_FRAMEBUFFER_OPERATION                 0x0506
+
+/* OpenGL ES 3.0 */
+
+#define GL_READ_BUFFER                                   0x0C02
+#define GL_UNPACK_ROW_LENGTH                             0x0CF2
+#define GL_UNPACK_SKIP_ROWS                              0x0CF3
+#define GL_UNPACK_SKIP_PIXELS                            0x0CF4
+#define GL_PACK_ROW_LENGTH                               0x0D02
+#define GL_PACK_SKIP_ROWS                                0x0D03
+#define GL_PACK_SKIP_PIXELS                              0x0D04
+#define GL_COLOR                                         0x1800
+#define GL_DEPTH                                         0x1801
+#define GL_STENCIL                                       0x1802
+#define GL_RED                                           0x1903
+#define GL_RGB8                                          0x8051
+#define GL_RGBA8                                         0x8058
+#define GL_RGB10_A2                                      0x8059
+#define GL_TEXTURE_BINDING_3D                            0x806A
+#define GL_UNPACK_SKIP_IMAGES                            0x806D
+#define GL_UNPACK_IMAGE_HEIGHT                           0x806E
+#define GL_TEXTURE_3D                                    0x806F
+#define GL_TEXTURE_WRAP_R                                0x8072
+#define GL_MAX_3D_TEXTURE_SIZE                           0x8073
+#define GL_UNSIGNED_INT_2_10_10_10_REV                   0x8368
+#define GL_MAX_ELEMENTS_VERTICES                         0x80E8
+#define GL_MAX_ELEMENTS_INDICES                          0x80E9
+#define GL_TEXTURE_MIN_LOD                               0x813A
+#define GL_TEXTURE_MAX_LOD                               0x813B
+#define GL_TEXTURE_BASE_LEVEL                            0x813C
+#define GL_TEXTURE_MAX_LEVEL                             0x813D
+#define GL_MIN                                           0x8007
+#define GL_MAX                                           0x8008
+#define GL_DEPTH_COMPONENT24                             0x81A6
+#define GL_MAX_TEXTURE_LOD_BIAS                          0x84FD
+#define GL_TEXTURE_COMPARE_MODE                          0x884C
+#define GL_TEXTURE_COMPARE_FUNC                          0x884D
+#define GL_CURRENT_QUERY                                 0x8865
+#define GL_QUERY_RESULT                                  0x8866
+#define GL_QUERY_RESULT_AVAILABLE                        0x8867
+#define GL_BUFFER_MAPPED                                 0x88BC
+#define GL_BUFFER_MAP_POINTER                            0x88BD
+#define GL_STREAM_READ                                   0x88E1
+#define GL_STREAM_COPY                                   0x88E2
+#define GL_STATIC_READ                                   0x88E5
+#define GL_STATIC_COPY                                   0x88E6
+#define GL_DYNAMIC_READ                                  0x88E9
+#define GL_DYNAMIC_COPY                                  0x88EA
+#define GL_MAX_DRAW_BUFFERS                              0x8824
+#define GL_DRAW_BUFFER0                                  0x8825
+#define GL_DRAW_BUFFER1                                  0x8826
+#define GL_DRAW_BUFFER2                                  0x8827
+#define GL_DRAW_BUFFER3                                  0x8828
+#define GL_DRAW_BUFFER4                                  0x8829
+#define GL_DRAW_BUFFER5                                  0x882A
+#define GL_DRAW_BUFFER6                                  0x882B
+#define GL_DRAW_BUFFER7                                  0x882C
+#define GL_DRAW_BUFFER8                                  0x882D
+#define GL_DRAW_BUFFER9                                  0x882E
+#define GL_DRAW_BUFFER10                                 0x882F
+#define GL_DRAW_BUFFER11                                 0x8830
+#define GL_DRAW_BUFFER12                                 0x8831
+#define GL_DRAW_BUFFER13                                 0x8832
+#define GL_DRAW_BUFFER14                                 0x8833
+#define GL_DRAW_BUFFER15                                 0x8834
+#define GL_MAX_FRAGMENT_UNIFORM_COMPONENTS               0x8B49
+#define GL_MAX_VERTEX_UNIFORM_COMPONENTS                 0x8B4A
+#define GL_SAMPLER_3D                                    0x8B5F
+#define GL_SAMPLER_2D_SHADOW                             0x8B62
+#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT               0x8B8B
+#define GL_PIXEL_PACK_BUFFER                             0x88EB
+#define GL_PIXEL_UNPACK_BUFFER                           0x88EC
+#define GL_PIXEL_PACK_BUFFER_BINDING                     0x88ED
+#define GL_PIXEL_UNPACK_BUFFER_BINDING                   0x88EF
+#define GL_FLOAT_MAT2x3                                  0x8B65
+#define GL_FLOAT_MAT2x4                                  0x8B66
+#define GL_FLOAT_MAT3x2                                  0x8B67
+#define GL_FLOAT_MAT3x4                                  0x8B68
+#define GL_FLOAT_MAT4x2                                  0x8B69
+#define GL_FLOAT_MAT4x3                                  0x8B6A
+#define GL_SRGB                                          0x8C40
+#define GL_SRGB8                                         0x8C41
+#define GL_SRGB8_ALPHA8                                  0x8C43
+#define GL_COMPARE_REF_TO_TEXTURE                        0x884E
+#define GL_MAJOR_VERSION                                 0x821B
+#define GL_MINOR_VERSION                                 0x821C
+#define GL_NUM_EXTENSIONS                                0x821D
+#define GL_RGBA32F                                       0x8814
+#define GL_RGB32F                                        0x8815
+#define GL_RGBA16F                                       0x881A
+#define GL_RGB16F                                        0x881B
+#define GL_VERTEX_ATTRIB_ARRAY_INTEGER                   0x88FD
+#define GL_MAX_ARRAY_TEXTURE_LAYERS                      0x88FF
+#define GL_MIN_PROGRAM_TEXEL_OFFSET                      0x8904
+#define GL_MAX_PROGRAM_TEXEL_OFFSET                      0x8905
+#define GL_MAX_VARYING_COMPONENTS                        0x8B4B
+#define GL_TEXTURE_2D_ARRAY                              0x8C1A
+#define GL_TEXTURE_BINDING_2D_ARRAY                      0x8C1D
+#define GL_R11F_G11F_B10F                                0x8C3A
+#define GL_UNSIGNED_INT_10F_11F_11F_REV                  0x8C3B
+#define GL_RGB9_E5                                       0x8C3D
+#define GL_UNSIGNED_INT_5_9_9_9_REV                      0x8C3E
+#define GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH         0x8C76
+#define GL_TRANSFORM_FEEDBACK_BUFFER_MODE                0x8C7F
+#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS    0x8C80
+#define GL_TRANSFORM_FEEDBACK_VARYINGS                   0x8C83
+#define GL_TRANSFORM_FEEDBACK_BUFFER_START               0x8C84
+#define GL_TRANSFORM_FEEDBACK_BUFFER_SIZE                0x8C85
+#define GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN         0x8C88
+#define GL_RASTERIZER_DISCARD                            0x8C89
+#define GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS 0x8C8A
+#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS       0x8C8B
+#define GL_INTERLEAVED_ATTRIBS                           0x8C8C
+#define GL_SEPARATE_ATTRIBS                              0x8C8D
+#define GL_TRANSFORM_FEEDBACK_BUFFER                     0x8C8E
+#define GL_TRANSFORM_FEEDBACK_BUFFER_BINDING             0x8C8F
+#define GL_RGBA32UI                                      0x8D70
+#define GL_RGB32UI                                       0x8D71
+#define GL_RGBA16UI                                      0x8D76
+#define GL_RGB16UI                                       0x8D77
+#define GL_RGBA8UI                                       0x8D7C
+#define GL_RGB8UI                                        0x8D7D
+#define GL_RGBA32I                                       0x8D82
+#define GL_RGB32I                                        0x8D83
+#define GL_RGBA16I                                       0x8D88
+#define GL_RGB16I                                        0x8D89
+#define GL_RGBA8I                                        0x8D8E
+#define GL_RGB8I                                         0x8D8F
+#define GL_RED_INTEGER                                   0x8D94
+#define GL_RGB_INTEGER                                   0x8D98
+#define GL_RGBA_INTEGER                                  0x8D99
+#define GL_SAMPLER_2D_ARRAY                              0x8DC1
+#define GL_SAMPLER_2D_ARRAY_SHADOW                       0x8DC4
+#define GL_SAMPLER_CUBE_SHADOW                           0x8DC5
+#define GL_UNSIGNED_INT_VEC2                             0x8DC6
+#define GL_UNSIGNED_INT_VEC3                             0x8DC7
+#define GL_UNSIGNED_INT_VEC4                             0x8DC8
+#define GL_INT_SAMPLER_2D                                0x8DCA
+#define GL_INT_SAMPLER_3D                                0x8DCB
+#define GL_INT_SAMPLER_CUBE                              0x8DCC
+#define GL_INT_SAMPLER_2D_ARRAY                          0x8DCF
+#define GL_UNSIGNED_INT_SAMPLER_2D                       0x8DD2
+#define GL_UNSIGNED_INT_SAMPLER_3D                       0x8DD3
+#define GL_UNSIGNED_INT_SAMPLER_CUBE                     0x8DD4
+#define GL_UNSIGNED_INT_SAMPLER_2D_ARRAY                 0x8DD7
+#define GL_BUFFER_ACCESS_FLAGS                           0x911F
+#define GL_BUFFER_MAP_LENGTH                             0x9120
+#define GL_BUFFER_MAP_OFFSET                             0x9121
+#define GL_DEPTH_COMPONENT32F                            0x8CAC
+#define GL_DEPTH32F_STENCIL8                             0x8CAD
+#define GL_FLOAT_32_UNSIGNED_INT_24_8_REV                0x8DAD
+#define GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING         0x8210
+#define GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE         0x8211
+#define GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE               0x8212
+#define GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE             0x8213
+#define GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE              0x8214
+#define GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE             0x8215
+#define GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE             0x8216
+#define GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE           0x8217
+#define GL_FRAMEBUFFER_DEFAULT                           0x8218
+#define GL_FRAMEBUFFER_UNDEFINED                         0x8219
+#define GL_DEPTH_STENCIL_ATTACHMENT                      0x821A
+#define GL_DEPTH_STENCIL                                 0x84F9
+#define GL_UNSIGNED_INT_24_8                             0x84FA
+#define GL_DEPTH24_STENCIL8                              0x88F0
+#define GL_UNSIGNED_NORMALIZED                           0x8C17
+#define GL_DRAW_FRAMEBUFFER_BINDING                      GL_FRAMEBUFFER_BINDING
+#define GL_READ_FRAMEBUFFER                              0x8CA8
+#define GL_DRAW_FRAMEBUFFER                              0x8CA9
+#define GL_READ_FRAMEBUFFER_BINDING                      0x8CAA
+#define GL_RENDERBUFFER_SAMPLES                          0x8CAB
+#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER          0x8CD4
+#define GL_MAX_COLOR_ATTACHMENTS                         0x8CDF
+#define GL_COLOR_ATTACHMENT1                             0x8CE1
+#define GL_COLOR_ATTACHMENT2                             0x8CE2
+#define GL_COLOR_ATTACHMENT3                             0x8CE3
+#define GL_COLOR_ATTACHMENT4                             0x8CE4
+#define GL_COLOR_ATTACHMENT5                             0x8CE5
+#define GL_COLOR_ATTACHMENT6                             0x8CE6
+#define GL_COLOR_ATTACHMENT7                             0x8CE7
+#define GL_COLOR_ATTACHMENT8                             0x8CE8
+#define GL_COLOR_ATTACHMENT9                             0x8CE9
+#define GL_COLOR_ATTACHMENT10                            0x8CEA
+#define GL_COLOR_ATTACHMENT11                            0x8CEB
+#define GL_COLOR_ATTACHMENT12                            0x8CEC
+#define GL_COLOR_ATTACHMENT13                            0x8CED
+#define GL_COLOR_ATTACHMENT14                            0x8CEE
+#define GL_COLOR_ATTACHMENT15                            0x8CEF
+#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE            0x8D56
+#define GL_MAX_SAMPLES                                   0x8D57
+#define GL_HALF_FLOAT                                    0x140B
+#define GL_MAP_READ_BIT                                  0x0001
+#define GL_MAP_WRITE_BIT                                 0x0002
+#define GL_MAP_INVALIDATE_RANGE_BIT                      0x0004
+#define GL_MAP_INVALIDATE_BUFFER_BIT                     0x0008
+#define GL_MAP_FLUSH_EXPLICIT_BIT                        0x0010
+#define GL_MAP_UNSYNCHRONIZED_BIT                        0x0020
+#define GL_RG                                            0x8227
+#define GL_RG_INTEGER                                    0x8228
+#define GL_R8                                            0x8229
+#define GL_RG8                                           0x822B
+#define GL_R16F                                          0x822D
+#define GL_R32F                                          0x822E
+#define GL_RG16F                                         0x822F
+#define GL_RG32F                                         0x8230
+#define GL_R8I                                           0x8231
+#define GL_R8UI                                          0x8232
+#define GL_R16I                                          0x8233
+#define GL_R16UI                                         0x8234
+#define GL_R32I                                          0x8235
+#define GL_R32UI                                         0x8236
+#define GL_RG8I                                          0x8237
+#define GL_RG8UI                                         0x8238
+#define GL_RG16I                                         0x8239
+#define GL_RG16UI                                        0x823A
+#define GL_RG32I                                         0x823B
+#define GL_RG32UI                                        0x823C
+#define GL_VERTEX_ARRAY_BINDING                          0x85B5
+#define GL_R8_SNORM                                      0x8F94
+#define GL_RG8_SNORM                                     0x8F95
+#define GL_RGB8_SNORM                                    0x8F96
+#define GL_RGBA8_SNORM                                   0x8F97
+#define GL_SIGNED_NORMALIZED                             0x8F9C
+#define GL_PRIMITIVE_RESTART_FIXED_INDEX                 0x8D69
+#define GL_COPY_READ_BUFFER                              0x8F36
+#define GL_COPY_WRITE_BUFFER                             0x8F37
+#define GL_COPY_READ_BUFFER_BINDING                      GL_COPY_READ_BUFFER
+#define GL_COPY_WRITE_BUFFER_BINDING                     GL_COPY_WRITE_BUFFER
+#define GL_UNIFORM_BUFFER                                0x8A11
+#define GL_UNIFORM_BUFFER_BINDING                        0x8A28
+#define GL_UNIFORM_BUFFER_START                          0x8A29
+#define GL_UNIFORM_BUFFER_SIZE                           0x8A2A
+#define GL_MAX_VERTEX_UNIFORM_BLOCKS                     0x8A2B
+#define GL_MAX_FRAGMENT_UNIFORM_BLOCKS                   0x8A2D
+#define GL_MAX_COMBINED_UNIFORM_BLOCKS                   0x8A2E
+#define GL_MAX_UNIFORM_BUFFER_BINDINGS                   0x8A2F
+#define GL_MAX_UNIFORM_BLOCK_SIZE                        0x8A30
+#define GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS        0x8A31
+#define GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS      0x8A33
+#define GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT               0x8A34
+#define GL_ACTIVE_UNIFORM_BLOCK_MAX_NAME_LENGTH          0x8A35
+#define GL_ACTIVE_UNIFORM_BLOCKS                         0x8A36
+#define GL_UNIFORM_TYPE                                  0x8A37
+#define GL_UNIFORM_SIZE                                  0x8A38
+#define GL_UNIFORM_NAME_LENGTH                           0x8A39
+#define GL_UNIFORM_BLOCK_INDEX                           0x8A3A
+#define GL_UNIFORM_OFFSET                                0x8A3B
+#define GL_UNIFORM_ARRAY_STRIDE                          0x8A3C
+#define GL_UNIFORM_MATRIX_STRIDE                         0x8A3D
+#define GL_UNIFORM_IS_ROW_MAJOR                          0x8A3E
+#define GL_UNIFORM_BLOCK_BINDING                         0x8A3F
+#define GL_UNIFORM_BLOCK_DATA_SIZE                       0x8A40
+#define GL_UNIFORM_BLOCK_NAME_LENGTH                     0x8A41
+#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORMS                 0x8A42
+#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORM_INDICES          0x8A43
+#define GL_UNIFORM_BLOCK_REFERENCED_BY_VERTEX_SHADER     0x8A44
+#define GL_UNIFORM_BLOCK_REFERENCED_BY_FRAGMENT_SHADER   0x8A46
+#define GL_INVALID_INDEX                                 0xFFFFFFFFu
+#define GL_MAX_VERTEX_OUTPUT_COMPONENTS                  0x9122
+#define GL_MAX_FRAGMENT_INPUT_COMPONENTS                 0x9125
+#define GL_MAX_SERVER_WAIT_TIMEOUT                       0x9111
+#define GL_OBJECT_TYPE                                   0x9112
+#define GL_SYNC_CONDITION                                0x9113
+#define GL_SYNC_STATUS                                   0x9114
+#define GL_SYNC_FLAGS                                    0x9115
+#define GL_SYNC_FENCE                                    0x9116
+#define GL_SYNC_GPU_COMMANDS_COMPLETE                    0x9117
+#define GL_UNSIGNALED                                    0x9118
+#define GL_SIGNALED                                      0x9119
+#define GL_ALREADY_SIGNALED                              0x911A
+#define GL_TIMEOUT_EXPIRED                               0x911B
+#define GL_CONDITION_SATISFIED                           0x911C
+#define GL_WAIT_FAILED                                   0x911D
+#define GL_SYNC_FLUSH_COMMANDS_BIT                       0x00000001
+#define GL_TIMEOUT_IGNORED                               0xFFFFFFFFFFFFFFFFull
+#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR                   0x88FE
+#define GL_ANY_SAMPLES_PASSED                            0x8C2F
+#define GL_ANY_SAMPLES_PASSED_CONSERVATIVE               0x8D6A
+#define GL_SAMPLER_BINDING                               0x8919
+#define GL_RGB10_A2UI                                    0x906F
+#define GL_TEXTURE_SWIZZLE_R                             0x8E42
+#define GL_TEXTURE_SWIZZLE_G                             0x8E43
+#define GL_TEXTURE_SWIZZLE_B                             0x8E44
+#define GL_TEXTURE_SWIZZLE_A                             0x8E45
+#define GL_GREEN                                         0x1904
+#define GL_BLUE                                          0x1905
+#define GL_INT_2_10_10_10_REV                            0x8D9F
+#define GL_TRANSFORM_FEEDBACK                            0x8E22
+#define GL_TRANSFORM_FEEDBACK_PAUSED                     0x8E23
+#define GL_TRANSFORM_FEEDBACK_ACTIVE                     0x8E24
+#define GL_TRANSFORM_FEEDBACK_BINDING                    0x8E25
+#define GL_PROGRAM_BINARY_RETRIEVABLE_HINT               0x8257
+#define GL_PROGRAM_BINARY_LENGTH                         0x8741
+#define GL_NUM_PROGRAM_BINARY_FORMATS                    0x87FE
+#define GL_PROGRAM_BINARY_FORMATS                        0x87FF
+#define GL_COMPRESSED_R11_EAC                            0x9270
+#define GL_COMPRESSED_SIGNED_R11_EAC                     0x9271
+#define GL_COMPRESSED_RG11_EAC                           0x9272
+#define GL_COMPRESSED_SIGNED_RG11_EAC                    0x9273
+#define GL_COMPRESSED_RGB8_ETC2                          0x9274
+#define GL_COMPRESSED_SRGB8_ETC2                         0x9275
+#define GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2      0x9276
+#define GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2     0x9277
+#define GL_COMPRESSED_RGBA8_ETC2_EAC                     0x9278
+#define GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC              0x9279
+#define GL_TEXTURE_IMMUTABLE_FORMAT                      0x912F
+#define GL_MAX_ELEMENT_INDEX                             0x8D6B
+#define GL_NUM_SAMPLE_COUNTS                             0x9380
+#define GL_TEXTURE_IMMUTABLE_LEVELS                      0x82DF
+
+/*-------------------------------------------------------------------------
+ * Entrypoint definitions
+ *-----------------------------------------------------------------------*/
+
+/* OpenGL ES 2.0 */
+
+GL_APICALL void           GL_APIENTRY glActiveTexture (GLenum texture);
+GL_APICALL void           GL_APIENTRY glAttachShader (GLuint program, GLuint shader);
+GL_APICALL void           GL_APIENTRY glBindAttribLocation (GLuint program, GLuint index, const GLchar* name);
+GL_APICALL void           GL_APIENTRY glBindBuffer (GLenum target, GLuint buffer);
+GL_APICALL void           GL_APIENTRY glBindFramebuffer (GLenum target, GLuint framebuffer);
+GL_APICALL void           GL_APIENTRY glBindRenderbuffer (GLenum target, GLuint renderbuffer);
+GL_APICALL void           GL_APIENTRY glBindTexture (GLenum target, GLuint texture);
+GL_APICALL void           GL_APIENTRY glBlendColor (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha);
+GL_APICALL void           GL_APIENTRY glBlendEquation (GLenum mode);
+GL_APICALL void           GL_APIENTRY glBlendEquationSeparate (GLenum modeRGB, GLenum modeAlpha);
+GL_APICALL void           GL_APIENTRY glBlendFunc (GLenum sfactor, GLenum dfactor);
+GL_APICALL void           GL_APIENTRY glBlendFuncSeparate (GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);
+GL_APICALL void           GL_APIENTRY glBufferData (GLenum target, GLsizeiptr size, const GLvoid* data, GLenum usage);
+GL_APICALL void           GL_APIENTRY glBufferSubData (GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid* data);
+GL_APICALL GLenum         GL_APIENTRY glCheckFramebufferStatus (GLenum target);
+GL_APICALL void           GL_APIENTRY glClear (GLbitfield mask);
+GL_APICALL void           GL_APIENTRY glClearColor (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha);
+GL_APICALL void           GL_APIENTRY glClearDepthf (GLfloat depth);
+GL_APICALL void           GL_APIENTRY glClearStencil (GLint s);
+GL_APICALL void           GL_APIENTRY glColorMask (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha);
+GL_APICALL void           GL_APIENTRY glCompileShader (GLuint shader);
+GL_APICALL void           GL_APIENTRY glCompressedTexImage2D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void           GL_APIENTRY glCompressedTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void           GL_APIENTRY glCopyTexImage2D (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border);
+GL_APICALL void           GL_APIENTRY glCopyTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL GLuint         GL_APIENTRY glCreateProgram (void);
+GL_APICALL GLuint         GL_APIENTRY glCreateShader (GLenum type);
+GL_APICALL void           GL_APIENTRY glCullFace (GLenum mode);
+GL_APICALL void           GL_APIENTRY glDeleteBuffers (GLsizei n, const GLuint* buffers);
+GL_APICALL void           GL_APIENTRY glDeleteFramebuffers (GLsizei n, const GLuint* framebuffers);
+GL_APICALL void           GL_APIENTRY glDeleteProgram (GLuint program);
+GL_APICALL void           GL_APIENTRY glDeleteRenderbuffers (GLsizei n, const GLuint* renderbuffers);
+GL_APICALL void           GL_APIENTRY glDeleteShader (GLuint shader);
+GL_APICALL void           GL_APIENTRY glDeleteTextures (GLsizei n, const GLuint* textures);
+GL_APICALL void           GL_APIENTRY glDepthFunc (GLenum func);
+GL_APICALL void           GL_APIENTRY glDepthMask (GLboolean flag);
+GL_APICALL void           GL_APIENTRY glDepthRangef (GLfloat n, GLfloat f);
+GL_APICALL void           GL_APIENTRY glDetachShader (GLuint program, GLuint shader);
+GL_APICALL void           GL_APIENTRY glDisable (GLenum cap);
+GL_APICALL void           GL_APIENTRY glDisableVertexAttribArray (GLuint index);
+GL_APICALL void           GL_APIENTRY glDrawArrays (GLenum mode, GLint first, GLsizei count);
+GL_APICALL void           GL_APIENTRY glDrawElements (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices);
+GL_APICALL void           GL_APIENTRY glEnable (GLenum cap);
+GL_APICALL void           GL_APIENTRY glEnableVertexAttribArray (GLuint index);
+GL_APICALL void           GL_APIENTRY glFinish (void);
+GL_APICALL void           GL_APIENTRY glFlush (void);
+GL_APICALL void           GL_APIENTRY glFramebufferRenderbuffer (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer);
+GL_APICALL void           GL_APIENTRY glFramebufferTexture2D (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level);
+GL_APICALL void           GL_APIENTRY glFrontFace (GLenum mode);
+GL_APICALL void           GL_APIENTRY glGenBuffers (GLsizei n, GLuint* buffers);
+GL_APICALL void           GL_APIENTRY glGenerateMipmap (GLenum target);
+GL_APICALL void           GL_APIENTRY glGenFramebuffers (GLsizei n, GLuint* framebuffers);
+GL_APICALL void           GL_APIENTRY glGenRenderbuffers (GLsizei n, GLuint* renderbuffers);
+GL_APICALL void           GL_APIENTRY glGenTextures (GLsizei n, GLuint* textures);
+GL_APICALL void           GL_APIENTRY glGetActiveAttrib (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);
+GL_APICALL void           GL_APIENTRY glGetActiveUniform (GLuint program, GLuint index, GLsizei bufsize, GLsizei* length, GLint* size, GLenum* type, GLchar* name);
+GL_APICALL void           GL_APIENTRY glGetAttachedShaders (GLuint program, GLsizei maxcount, GLsizei* count, GLuint* shaders);
+GL_APICALL GLint          GL_APIENTRY glGetAttribLocation (GLuint program, const GLchar* name);
+GL_APICALL void           GL_APIENTRY glGetBooleanv (GLenum pname, GLboolean* params);
+GL_APICALL void           GL_APIENTRY glGetBufferParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL GLenum         GL_APIENTRY glGetError (void);
+GL_APICALL void           GL_APIENTRY glGetFloatv (GLenum pname, GLfloat* params);
+GL_APICALL void           GL_APIENTRY glGetFramebufferAttachmentParameteriv (GLenum target, GLenum attachment, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetIntegerv (GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetProgramiv (GLuint program, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetProgramInfoLog (GLuint program, GLsizei bufsize, GLsizei* length, GLchar* infolog);
+GL_APICALL void           GL_APIENTRY glGetRenderbufferParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetShaderiv (GLuint shader, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetShaderInfoLog (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* infolog);
+GL_APICALL void           GL_APIENTRY glGetShaderPrecisionFormat (GLenum shadertype, GLenum precisiontype, GLint* range, GLint* precision);
+GL_APICALL void           GL_APIENTRY glGetShaderSource (GLuint shader, GLsizei bufsize, GLsizei* length, GLchar* source);
+GL_APICALL const GLubyte* GL_APIENTRY glGetString (GLenum name);
+GL_APICALL void           GL_APIENTRY glGetTexParameterfv (GLenum target, GLenum pname, GLfloat* params);
+GL_APICALL void           GL_APIENTRY glGetTexParameteriv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetUniformfv (GLuint program, GLint location, GLfloat* params);
+GL_APICALL void           GL_APIENTRY glGetUniformiv (GLuint program, GLint location, GLint* params);
+GL_APICALL GLint          GL_APIENTRY glGetUniformLocation (GLuint program, const GLchar* name);
+GL_APICALL void           GL_APIENTRY glGetVertexAttribfv (GLuint index, GLenum pname, GLfloat* params);
+GL_APICALL void           GL_APIENTRY glGetVertexAttribiv (GLuint index, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetVertexAttribPointerv (GLuint index, GLenum pname, GLvoid** pointer);
+GL_APICALL void           GL_APIENTRY glHint (GLenum target, GLenum mode);
+GL_APICALL GLboolean      GL_APIENTRY glIsBuffer (GLuint buffer);
+GL_APICALL GLboolean      GL_APIENTRY glIsEnabled (GLenum cap);
+GL_APICALL GLboolean      GL_APIENTRY glIsFramebuffer (GLuint framebuffer);
+GL_APICALL GLboolean      GL_APIENTRY glIsProgram (GLuint program);
+GL_APICALL GLboolean      GL_APIENTRY glIsRenderbuffer (GLuint renderbuffer);
+GL_APICALL GLboolean      GL_APIENTRY glIsShader (GLuint shader);
+GL_APICALL GLboolean      GL_APIENTRY glIsTexture (GLuint texture);
+GL_APICALL void           GL_APIENTRY glLineWidth (GLfloat width);
+GL_APICALL void           GL_APIENTRY glLinkProgram (GLuint program);
+GL_APICALL void           GL_APIENTRY glPixelStorei (GLenum pname, GLint param);
+GL_APICALL void           GL_APIENTRY glPolygonOffset (GLfloat factor, GLfloat units);
+GL_APICALL void           GL_APIENTRY glReadPixels (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid* pixels);
+GL_APICALL void           GL_APIENTRY glReleaseShaderCompiler (void);
+GL_APICALL void           GL_APIENTRY glRenderbufferStorage (GLenum target, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glSampleCoverage (GLfloat value, GLboolean invert);
+GL_APICALL void           GL_APIENTRY glScissor (GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glShaderBinary (GLsizei n, const GLuint* shaders, GLenum binaryformat, const GLvoid* binary, GLsizei length);
+GL_APICALL void           GL_APIENTRY glShaderSource (GLuint shader, GLsizei count, const GLchar* const* string, const GLint* length);
+GL_APICALL void           GL_APIENTRY glStencilFunc (GLenum func, GLint ref, GLuint mask);
+GL_APICALL void           GL_APIENTRY glStencilFuncSeparate (GLenum face, GLenum func, GLint ref, GLuint mask);
+GL_APICALL void           GL_APIENTRY glStencilMask (GLuint mask);
+GL_APICALL void           GL_APIENTRY glStencilMaskSeparate (GLenum face, GLuint mask);
+GL_APICALL void           GL_APIENTRY glStencilOp (GLenum fail, GLenum zfail, GLenum zpass);
+GL_APICALL void           GL_APIENTRY glStencilOpSeparate (GLenum face, GLenum fail, GLenum zfail, GLenum zpass);
+GL_APICALL void           GL_APIENTRY glTexImage2D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void           GL_APIENTRY glTexParameterf (GLenum target, GLenum pname, GLfloat param);
+GL_APICALL void           GL_APIENTRY glTexParameterfv (GLenum target, GLenum pname, const GLfloat* params);
+GL_APICALL void           GL_APIENTRY glTexParameteri (GLenum target, GLenum pname, GLint param);
+GL_APICALL void           GL_APIENTRY glTexParameteriv (GLenum target, GLenum pname, const GLint* params);
+GL_APICALL void           GL_APIENTRY glTexSubImage2D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void           GL_APIENTRY glUniform1f (GLint location, GLfloat x);
+GL_APICALL void           GL_APIENTRY glUniform1fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void           GL_APIENTRY glUniform1i (GLint location, GLint x);
+GL_APICALL void           GL_APIENTRY glUniform1iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void           GL_APIENTRY glUniform2f (GLint location, GLfloat x, GLfloat y);
+GL_APICALL void           GL_APIENTRY glUniform2fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void           GL_APIENTRY glUniform2i (GLint location, GLint x, GLint y);
+GL_APICALL void           GL_APIENTRY glUniform2iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void           GL_APIENTRY glUniform3f (GLint location, GLfloat x, GLfloat y, GLfloat z);
+GL_APICALL void           GL_APIENTRY glUniform3fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void           GL_APIENTRY glUniform3i (GLint location, GLint x, GLint y, GLint z);
+GL_APICALL void           GL_APIENTRY glUniform3iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void           GL_APIENTRY glUniform4f (GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+GL_APICALL void           GL_APIENTRY glUniform4fv (GLint location, GLsizei count, const GLfloat* v);
+GL_APICALL void           GL_APIENTRY glUniform4i (GLint location, GLint x, GLint y, GLint z, GLint w);
+GL_APICALL void           GL_APIENTRY glUniform4iv (GLint location, GLsizei count, const GLint* v);
+GL_APICALL void           GL_APIENTRY glUniformMatrix2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUseProgram (GLuint program);
+GL_APICALL void           GL_APIENTRY glValidateProgram (GLuint program);
+GL_APICALL void           GL_APIENTRY glVertexAttrib1f (GLuint indx, GLfloat x);
+GL_APICALL void           GL_APIENTRY glVertexAttrib1fv (GLuint indx, const GLfloat* values);
+GL_APICALL void           GL_APIENTRY glVertexAttrib2f (GLuint indx, GLfloat x, GLfloat y);
+GL_APICALL void           GL_APIENTRY glVertexAttrib2fv (GLuint indx, const GLfloat* values);
+GL_APICALL void           GL_APIENTRY glVertexAttrib3f (GLuint indx, GLfloat x, GLfloat y, GLfloat z);
+GL_APICALL void           GL_APIENTRY glVertexAttrib3fv (GLuint indx, const GLfloat* values);
+GL_APICALL void           GL_APIENTRY glVertexAttrib4f (GLuint indx, GLfloat x, GLfloat y, GLfloat z, GLfloat w);
+GL_APICALL void           GL_APIENTRY glVertexAttrib4fv (GLuint indx, const GLfloat* values);
+GL_APICALL void           GL_APIENTRY glVertexAttribPointer (GLuint indx, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid* ptr);
+GL_APICALL void           GL_APIENTRY glViewport (GLint x, GLint y, GLsizei width, GLsizei height);
+
+/* OpenGL ES 3.0 */
+
+GL_APICALL void           GL_APIENTRY glReadBuffer (GLenum mode);
+GL_APICALL void           GL_APIENTRY glDrawRangeElements (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid* indices);
+GL_APICALL void           GL_APIENTRY glTexImage3D (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void           GL_APIENTRY glTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid* pixels);
+GL_APICALL void           GL_APIENTRY glCopyTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glCompressedTexImage3D (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void           GL_APIENTRY glCompressedTexSubImage3D (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid* data);
+GL_APICALL void           GL_APIENTRY glGenQueries (GLsizei n, GLuint* ids);
+GL_APICALL void           GL_APIENTRY glDeleteQueries (GLsizei n, const GLuint* ids);
+GL_APICALL GLboolean      GL_APIENTRY glIsQuery (GLuint id);
+GL_APICALL void           GL_APIENTRY glBeginQuery (GLenum target, GLuint id);
+GL_APICALL void           GL_APIENTRY glEndQuery (GLenum target);
+GL_APICALL void           GL_APIENTRY glGetQueryiv (GLenum target, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetQueryObjectuiv (GLuint id, GLenum pname, GLuint* params);
+GL_APICALL GLboolean      GL_APIENTRY glUnmapBuffer (GLenum target);
+GL_APICALL void           GL_APIENTRY glGetBufferPointerv (GLenum target, GLenum pname, GLvoid** params);
+GL_APICALL void           GL_APIENTRY glDrawBuffers (GLsizei n, const GLenum* bufs);
+GL_APICALL void           GL_APIENTRY glUniformMatrix2x3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix3x2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix2x4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix4x2fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix3x4fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glUniformMatrix4x3fv (GLint location, GLsizei count, GLboolean transpose, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glBlitFramebuffer (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+GL_APICALL void           GL_APIENTRY glRenderbufferStorageMultisample (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glFramebufferTextureLayer (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer);
+GL_APICALL GLvoid*        GL_APIENTRY glMapBufferRange (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access);
+GL_APICALL void           GL_APIENTRY glFlushMappedBufferRange (GLenum target, GLintptr offset, GLsizeiptr length);
+GL_APICALL void           GL_APIENTRY glBindVertexArray (GLuint array);
+GL_APICALL void           GL_APIENTRY glDeleteVertexArrays (GLsizei n, const GLuint* arrays);
+GL_APICALL void           GL_APIENTRY glGenVertexArrays (GLsizei n, GLuint* arrays);
+GL_APICALL GLboolean      GL_APIENTRY glIsVertexArray (GLuint array);
+GL_APICALL void           GL_APIENTRY glGetIntegeri_v (GLenum target, GLuint index, GLint* data);
+GL_APICALL void           GL_APIENTRY glBeginTransformFeedback (GLenum primitiveMode);
+GL_APICALL void           GL_APIENTRY glEndTransformFeedback (void);
+GL_APICALL void           GL_APIENTRY glBindBufferRange (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size);
+GL_APICALL void           GL_APIENTRY glBindBufferBase (GLenum target, GLuint index, GLuint buffer);
+GL_APICALL void           GL_APIENTRY glTransformFeedbackVaryings (GLuint program, GLsizei count, const GLchar* const* varyings, GLenum bufferMode);
+GL_APICALL void           GL_APIENTRY glGetTransformFeedbackVarying (GLuint program, GLuint index, GLsizei bufSize, GLsizei* length, GLsizei* size, GLenum* type, GLchar* name);
+GL_APICALL void           GL_APIENTRY glVertexAttribIPointer (GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid* pointer);
+GL_APICALL void           GL_APIENTRY glGetVertexAttribIiv (GLuint index, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetVertexAttribIuiv (GLuint index, GLenum pname, GLuint* params);
+GL_APICALL void           GL_APIENTRY glVertexAttribI4i (GLuint index, GLint x, GLint y, GLint z, GLint w);
+GL_APICALL void           GL_APIENTRY glVertexAttribI4ui (GLuint index, GLuint x, GLuint y, GLuint z, GLuint w);
+GL_APICALL void           GL_APIENTRY glVertexAttribI4iv (GLuint index, const GLint* v);
+GL_APICALL void           GL_APIENTRY glVertexAttribI4uiv (GLuint index, const GLuint* v);
+GL_APICALL void           GL_APIENTRY glGetUniformuiv (GLuint program, GLint location, GLuint* params);
+GL_APICALL GLint          GL_APIENTRY glGetFragDataLocation (GLuint program, const GLchar *name);
+GL_APICALL void           GL_APIENTRY glUniform1ui (GLint location, GLuint v0);
+GL_APICALL void           GL_APIENTRY glUniform2ui (GLint location, GLuint v0, GLuint v1);
+GL_APICALL void           GL_APIENTRY glUniform3ui (GLint location, GLuint v0, GLuint v1, GLuint v2);
+GL_APICALL void           GL_APIENTRY glUniform4ui (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3);
+GL_APICALL void           GL_APIENTRY glUniform1uiv (GLint location, GLsizei count, const GLuint* value);
+GL_APICALL void           GL_APIENTRY glUniform2uiv (GLint location, GLsizei count, const GLuint* value);
+GL_APICALL void           GL_APIENTRY glUniform3uiv (GLint location, GLsizei count, const GLuint* value);
+GL_APICALL void           GL_APIENTRY glUniform4uiv (GLint location, GLsizei count, const GLuint* value);
+GL_APICALL void           GL_APIENTRY glClearBufferiv (GLenum buffer, GLint drawbuffer, const GLint* value);
+GL_APICALL void           GL_APIENTRY glClearBufferuiv (GLenum buffer, GLint drawbuffer, const GLuint* value);
+GL_APICALL void           GL_APIENTRY glClearBufferfv (GLenum buffer, GLint drawbuffer, const GLfloat* value);
+GL_APICALL void           GL_APIENTRY glClearBufferfi (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);
+GL_APICALL const GLubyte* GL_APIENTRY glGetStringi (GLenum name, GLuint index);
+GL_APICALL void           GL_APIENTRY glCopyBufferSubData (GLenum readTarget, GLenum writeTarget, GLintptr readOffset, GLintptr writeOffset, GLsizeiptr size);
+GL_APICALL void           GL_APIENTRY glGetUniformIndices (GLuint program, GLsizei uniformCount, const GLchar* const* uniformNames, GLuint* uniformIndices);
+GL_APICALL void           GL_APIENTRY glGetActiveUniformsiv (GLuint program, GLsizei uniformCount, const GLuint* uniformIndices, GLenum pname, GLint* params);
+GL_APICALL GLuint         GL_APIENTRY glGetUniformBlockIndex (GLuint program, const GLchar* uniformBlockName);
+GL_APICALL void           GL_APIENTRY glGetActiveUniformBlockiv (GLuint program, GLuint uniformBlockIndex, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetActiveUniformBlockName (GLuint program, GLuint uniformBlockIndex, GLsizei bufSize, GLsizei* length, GLchar* uniformBlockName);
+GL_APICALL void           GL_APIENTRY glUniformBlockBinding (GLuint program, GLuint uniformBlockIndex, GLuint uniformBlockBinding);
+GL_APICALL void           GL_APIENTRY glDrawArraysInstanced (GLenum mode, GLint first, GLsizei count, GLsizei instanceCount);
+GL_APICALL void           GL_APIENTRY glDrawElementsInstanced (GLenum mode, GLsizei count, GLenum type, const GLvoid* indices, GLsizei instanceCount);
+GL_APICALL GLsync         GL_APIENTRY glFenceSync (GLenum condition, GLbitfield flags);
+GL_APICALL GLboolean      GL_APIENTRY glIsSync (GLsync sync);
+GL_APICALL void           GL_APIENTRY glDeleteSync (GLsync sync);
+GL_APICALL GLenum         GL_APIENTRY glClientWaitSync (GLsync sync, GLbitfield flags, GLuint64 timeout);
+GL_APICALL void           GL_APIENTRY glWaitSync (GLsync sync, GLbitfield flags, GLuint64 timeout);
+GL_APICALL void           GL_APIENTRY glGetInteger64v (GLenum pname, GLint64* params);
+GL_APICALL void           GL_APIENTRY glGetSynciv (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei* length, GLint* values);
+GL_APICALL void           GL_APIENTRY glGetInteger64i_v (GLenum target, GLuint index, GLint64* data);
+GL_APICALL void           GL_APIENTRY glGetBufferParameteri64v (GLenum target, GLenum pname, GLint64* params);
+GL_APICALL void           GL_APIENTRY glGenSamplers (GLsizei count, GLuint* samplers);
+GL_APICALL void           GL_APIENTRY glDeleteSamplers (GLsizei count, const GLuint* samplers);
+GL_APICALL GLboolean      GL_APIENTRY glIsSampler (GLuint sampler);
+GL_APICALL void           GL_APIENTRY glBindSampler (GLuint unit, GLuint sampler);
+GL_APICALL void           GL_APIENTRY glSamplerParameteri (GLuint sampler, GLenum pname, GLint param);
+GL_APICALL void           GL_APIENTRY glSamplerParameteriv (GLuint sampler, GLenum pname, const GLint* param);
+GL_APICALL void           GL_APIENTRY glSamplerParameterf (GLuint sampler, GLenum pname, GLfloat param);
+GL_APICALL void           GL_APIENTRY glSamplerParameterfv (GLuint sampler, GLenum pname, const GLfloat* param);
+GL_APICALL void           GL_APIENTRY glGetSamplerParameteriv (GLuint sampler, GLenum pname, GLint* params);
+GL_APICALL void           GL_APIENTRY glGetSamplerParameterfv (GLuint sampler, GLenum pname, GLfloat* params);
+GL_APICALL void           GL_APIENTRY glVertexAttribDivisor (GLuint index, GLuint divisor);
+GL_APICALL void           GL_APIENTRY glBindTransformFeedback (GLenum target, GLuint id);
+GL_APICALL void           GL_APIENTRY glDeleteTransformFeedbacks (GLsizei n, const GLuint* ids);
+GL_APICALL void           GL_APIENTRY glGenTransformFeedbacks (GLsizei n, GLuint* ids);
+GL_APICALL GLboolean      GL_APIENTRY glIsTransformFeedback (GLuint id);
+GL_APICALL void           GL_APIENTRY glPauseTransformFeedback (void);
+GL_APICALL void           GL_APIENTRY glResumeTransformFeedback (void);
+GL_APICALL void           GL_APIENTRY glGetProgramBinary (GLuint program, GLsizei bufSize, GLsizei* length, GLenum* binaryFormat, GLvoid* binary);
+GL_APICALL void           GL_APIENTRY glProgramBinary (GLuint program, GLenum binaryFormat, const GLvoid* binary, GLsizei length);
+GL_APICALL void           GL_APIENTRY glProgramParameteri (GLuint program, GLenum pname, GLint value);
+GL_APICALL void           GL_APIENTRY glInvalidateFramebuffer (GLenum target, GLsizei numAttachments, const GLenum* attachments);
+GL_APICALL void           GL_APIENTRY glInvalidateSubFramebuffer (GLenum target, GLsizei numAttachments, const GLenum* attachments, GLint x, GLint y, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glTexStorage2D (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
+GL_APICALL void           GL_APIENTRY glTexStorage3D (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth);
+GL_APICALL void           GL_APIENTRY glGetInternalformativ (GLenum target, GLenum internalformat, GLenum pname, GLsizei bufSize, GLint* params);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/winrt/include/GLES3/gl3ext.h b/platform/winrt/include/GLES3/gl3ext.h
index 199436c082..4d4ea96c4d 100644
--- a/platform/winrt/include/GLES3/gl3ext.h
+++ b/platform/winrt/include/GLES3/gl3ext.h
@@ -1,24 +1,24 @@
-#ifndef __gl3ext_h_

-#define __gl3ext_h_

-

-/* $Revision: 17809 $ on $Date:: 2012-05-14 08:03:36 -0700 #$ */

-

-/*

- * This document is licensed under the SGI Free Software B License Version

- * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .

- */

-

-/* OpenGL ES 3 Extensions

- *

- * After an OES extension's interactions with OpenGl ES 3.0 have been documented,

- * its tokens and function definitions should be added to this file in a manner

- * that does not conflict with gl2ext.h or gl3.h.

- *

- * Tokens and function definitions for extensions that have become standard

- * features in OpenGL ES 3.0 will not be added to this file.

- *

- * Applications using OpenGL-ES-2-only extensions should include gl2ext.h

- */

-

-#endif /* __gl3ext_h_ */

-

+#ifndef __gl3ext_h_
+#define __gl3ext_h_
+
+/* $Revision: 17809 $ on $Date:: 2012-05-14 08:03:36 -0700 #$ */
+
+/*
+ * This document is licensed under the SGI Free Software B License Version
+ * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .
+ */
+
+/* OpenGL ES 3 Extensions
+ *
+ * After an OES extension's interactions with OpenGl ES 3.0 have been documented,
+ * its tokens and function definitions should be added to this file in a manner
+ * that does not conflict with gl2ext.h or gl3.h.
+ *
+ * Tokens and function definitions for extensions that have become standard
+ * features in OpenGL ES 3.0 will not be added to this file.
+ *
+ * Applications using OpenGL-ES-2-only extensions should include gl2ext.h
+ */
+
+#endif /* __gl3ext_h_ */
+
diff --git a/platform/winrt/include/GLES3/gl3platform.h b/platform/winrt/include/GLES3/gl3platform.h
index 679e0dc3ca..1bd1a850fa 100644
--- a/platform/winrt/include/GLES3/gl3platform.h
+++ b/platform/winrt/include/GLES3/gl3platform.h
@@ -1,30 +1,30 @@
-#ifndef __gl3platform_h_

-#define __gl3platform_h_

-

-/* $Revision: 18437 $ on $Date:: 2012-07-08 23:31:39 -0700 #$ */

-

-/*

- * This document is licensed under the SGI Free Software B License Version

- * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .

- */

-

-/* Platform-specific types and definitions for OpenGL ES 3.X  gl3.h

- *

- * Adopters may modify khrplatform.h and this file to suit their platform.

- * You are encouraged to submit all modifications to the Khronos group so that

- * they can be included in future versions of this file.  Please submit changes

- * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)

- * by filing a bug against product "OpenGL-ES" component "Registry".

- */

-

-#include <KHR/khrplatform.h>

-

-#ifndef GL_APICALL

-#define GL_APICALL  KHRONOS_APICALL

-#endif

-

-#ifndef GL_APIENTRY

-#define GL_APIENTRY KHRONOS_APIENTRY

-#endif

-

-#endif /* __gl3platform_h_ */

+#ifndef __gl3platform_h_
+#define __gl3platform_h_
+
+/* $Revision: 18437 $ on $Date:: 2012-07-08 23:31:39 -0700 #$ */
+
+/*
+ * This document is licensed under the SGI Free Software B License Version
+ * 2.0. For details, see http://oss.sgi.com/projects/FreeB/ .
+ */
+
+/* Platform-specific types and definitions for OpenGL ES 3.X  gl3.h
+ *
+ * Adopters may modify khrplatform.h and this file to suit their platform.
+ * You are encouraged to submit all modifications to the Khronos group so that
+ * they can be included in future versions of this file.  Please submit changes
+ * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla)
+ * by filing a bug against product "OpenGL-ES" component "Registry".
+ */
+
+#include <KHR/khrplatform.h>
+
+#ifndef GL_APICALL
+#define GL_APICALL  KHRONOS_APICALL
+#endif
+
+#ifndef GL_APIENTRY
+#define GL_APIENTRY KHRONOS_APIENTRY
+#endif
+
+#endif /* __gl3platform_h_ */
diff --git a/platform/winrt/include/GLSLANG/ShaderLang.h b/platform/winrt/include/GLSLANG/ShaderLang.h
index d925b88f24..647fed6a02 100644
--- a/platform/winrt/include/GLSLANG/ShaderLang.h
+++ b/platform/winrt/include/GLSLANG/ShaderLang.h
@@ -1,411 +1,411 @@
-//

-// Copyright (c) 2002-2013 The ANGLE Project Authors. All rights reserved.

-// Use of this source code is governed by a BSD-style license that can be

-// found in the LICENSE file.

-//

-#ifndef _COMPILER_INTERFACE_INCLUDED_

-#define _COMPILER_INTERFACE_INCLUDED_

-

-#if defined(COMPONENT_BUILD) && !defined(ANGLE_TRANSLATOR_STATIC)

-#if defined(_WIN32) || defined(_WIN64)

-

-#if defined(ANGLE_TRANSLATOR_IMPLEMENTATION)

-#define COMPILER_EXPORT __declspec(dllexport)

-#else

-#define COMPILER_EXPORT __declspec(dllimport)

-#endif  // defined(ANGLE_TRANSLATOR_IMPLEMENTATION)

-

-#else  // defined(_WIN32) || defined(_WIN64)

-#define COMPILER_EXPORT __attribute__((visibility("default")))

-#endif

-

-#else  // defined(COMPONENT_BUILD) && !defined(ANGLE_TRANSLATOR_STATIC)

-#define COMPILER_EXPORT

-#endif

-

-#include <stddef.h>

-

-#include "KHR/khrplatform.h"

-

-#include <map>

-#include <string>

-#include <vector>

-

-//

-// This is the platform independent interface between an OGL driver

-// and the shading language compiler.

-//

-

-namespace sh

-{

-// GLenum alias

-typedef unsigned int GLenum;

-}

-

-// Must be included after GLenum proxy typedef

-// Note: make sure to increment ANGLE_SH_VERSION when changing ShaderVars.h

-#include "ShaderVars.h"

-

-// Version number for shader translation API.

-// It is incremented every time the API changes.

-#define ANGLE_SH_VERSION 132

-

-typedef enum {

-  SH_GLES2_SPEC = 0x8B40,

-  SH_WEBGL_SPEC = 0x8B41,

-

-  SH_GLES3_SPEC = 0x8B86,

-  SH_WEBGL2_SPEC = 0x8B87,

-

-  // The CSS Shaders spec is a subset of the WebGL spec.

-  //

-  // In both CSS vertex and fragment shaders, ANGLE:

-  // (1) Reserves the "css_" prefix.

-  // (2) Renames the main function to css_main.

-  // (3) Disables the gl_MaxDrawBuffers built-in.

-  //

-  // In CSS fragment shaders, ANGLE:

-  // (1) Disables the gl_FragColor built-in.

-  // (2) Disables the gl_FragData built-in.

-  // (3) Enables the css_MixColor built-in.

-  // (4) Enables the css_ColorMatrix built-in.

-  //

-  // After passing a CSS shader through ANGLE, the browser is expected to append

-  // a new main function to it.

-  // This new main function will call the css_main function.

-  // It may also perform additional operations like varying assignment, texture

-  // access, and gl_FragColor assignment in order to implement the CSS Shaders

-  // blend modes.

-  //

-  SH_CSS_SHADERS_SPEC = 0x8B42

-} ShShaderSpec;

-

-typedef enum {

-  SH_ESSL_OUTPUT   = 0x8B45,

-  SH_GLSL_OUTPUT   = 0x8B46,

-  SH_HLSL_OUTPUT   = 0x8B47,

-  SH_HLSL9_OUTPUT  = 0x8B47,

-  SH_HLSL11_OUTPUT = 0x8B48

-} ShShaderOutput;

-

-// Compile options.

-typedef enum {

-  SH_VALIDATE                = 0,

-  SH_VALIDATE_LOOP_INDEXING  = 0x0001,

-  SH_INTERMEDIATE_TREE       = 0x0002,

-  SH_OBJECT_CODE             = 0x0004,

-  SH_VARIABLES               = 0x0008,

-  SH_LINE_DIRECTIVES         = 0x0010,

-  SH_SOURCE_PATH             = 0x0020,

-  SH_UNROLL_FOR_LOOP_WITH_INTEGER_INDEX = 0x0040,

-  // If a sampler array index happens to be a loop index,

-  //   1) if its type is integer, unroll the loop.

-  //   2) if its type is float, fail the shader compile.

-  // This is to work around a mac driver bug.

-  SH_UNROLL_FOR_LOOP_WITH_SAMPLER_ARRAY_INDEX = 0x0080,

-

-  // This is needed only as a workaround for certain OpenGL driver bugs.

-  SH_EMULATE_BUILT_IN_FUNCTIONS = 0x0100,

-

-  // This is an experimental flag to enforce restrictions that aim to prevent 

-  // timing attacks.

-  // It generates compilation errors for shaders that could expose sensitive

-  // texture information via the timing channel.

-  // To use this flag, you must compile the shader under the WebGL spec

-  // (using the SH_WEBGL_SPEC flag).

-  SH_TIMING_RESTRICTIONS = 0x0200,

-

-  // This flag prints the dependency graph that is used to enforce timing

-  // restrictions on fragment shaders.

-  // This flag only has an effect if all of the following are true:

-  // - The shader spec is SH_WEBGL_SPEC.

-  // - The compile options contain the SH_TIMING_RESTRICTIONS flag.

-  // - The shader type is GL_FRAGMENT_SHADER.

-  SH_DEPENDENCY_GRAPH = 0x0400,

-

-  // Enforce the GLSL 1.017 Appendix A section 7 packing restrictions.

-  // This flag only enforces (and can only enforce) the packing

-  // restrictions for uniform variables in both vertex and fragment

-  // shaders. ShCheckVariablesWithinPackingLimits() lets embedders

-  // enforce the packing restrictions for varying variables during

-  // program link time.

-  SH_ENFORCE_PACKING_RESTRICTIONS = 0x0800,

-

-  // This flag ensures all indirect (expression-based) array indexing

-  // is clamped to the bounds of the array. This ensures, for example,

-  // that you cannot read off the end of a uniform, whether an array

-  // vec234, or mat234 type. The ShArrayIndexClampingStrategy enum,

-  // specified in the ShBuiltInResources when constructing the

-  // compiler, selects the strategy for the clamping implementation.

-  SH_CLAMP_INDIRECT_ARRAY_BOUNDS = 0x1000,

-

-  // This flag limits the complexity of an expression.

-  SH_LIMIT_EXPRESSION_COMPLEXITY = 0x2000,

-

-  // This flag limits the depth of the call stack.

-  SH_LIMIT_CALL_STACK_DEPTH = 0x4000,

-

-  // This flag initializes gl_Position to vec4(0,0,0,0) at the

-  // beginning of the vertex shader's main(), and has no effect in the

-  // fragment shader. It is intended as a workaround for drivers which

-  // incorrectly fail to link programs if gl_Position is not written.

-  SH_INIT_GL_POSITION = 0x8000,

-

-  // This flag replaces

-  //   "a && b" with "a ? b : false",

-  //   "a || b" with "a ? true : b".

-  // This is to work around a MacOSX driver bug that |b| is executed

-  // independent of |a|'s value.

-  SH_UNFOLD_SHORT_CIRCUIT = 0x10000,

-

-  // This flag initializes varyings without static use in vertex shader

-  // at the beginning of main(), and has no effects in the fragment shader.

-  // It is intended as a workaround for drivers which incorrectly optimize

-  // out such varyings and cause a link failure.

-  SH_INIT_VARYINGS_WITHOUT_STATIC_USE = 0x20000,

-

-  // This flag scalarizes vec/ivec/bvec/mat constructor args.

-  // It is intended as a workaround for Linux/Mac driver bugs.

-  SH_SCALARIZE_VEC_AND_MAT_CONSTRUCTOR_ARGS = 0x40000,

-

-  // This flag overwrites a struct name with a unique prefix.

-  // It is intended as a workaround for drivers that do not handle

-  // struct scopes correctly, including all Mac drivers and Linux AMD.

-  SH_REGENERATE_STRUCT_NAMES = 0x80000,

-} ShCompileOptions;

-

-// Defines alternate strategies for implementing array index clamping.

-typedef enum {

-  // Use the clamp intrinsic for array index clamping.

-  SH_CLAMP_WITH_CLAMP_INTRINSIC = 1,

-

-  // Use a user-defined function for array index clamping.

-  SH_CLAMP_WITH_USER_DEFINED_INT_CLAMP_FUNCTION

-} ShArrayIndexClampingStrategy;

-

-//

-// Driver must call this first, once, before doing any other

-// compiler operations.

-// If the function succeeds, the return value is true, else false.

-//

-COMPILER_EXPORT bool ShInitialize();

-//

-// Driver should call this at shutdown.

-// If the function succeeds, the return value is true, else false.

-//

-COMPILER_EXPORT bool ShFinalize();

-

-// The 64 bits hash function. The first parameter is the input string; the

-// second parameter is the string length.

-typedef khronos_uint64_t (*ShHashFunction64)(const char*, size_t);

-

-//

-// Implementation dependent built-in resources (constants and extensions).

-// The names for these resources has been obtained by stripping gl_/GL_.

-//

-typedef struct

-{

-    // Constants.

-    int MaxVertexAttribs;

-    int MaxVertexUniformVectors;

-    int MaxVaryingVectors;

-    int MaxVertexTextureImageUnits;

-    int MaxCombinedTextureImageUnits;

-    int MaxTextureImageUnits;

-    int MaxFragmentUniformVectors;

-    int MaxDrawBuffers;

-

-    // Extensions.

-    // Set to 1 to enable the extension, else 0.

-    int OES_standard_derivatives;

-    int OES_EGL_image_external;

-    int ARB_texture_rectangle;

-    int EXT_draw_buffers;

-    int EXT_frag_depth;

-    int EXT_shader_texture_lod;

-

-    // Set to 1 to enable replacing GL_EXT_draw_buffers #extension directives

-    // with GL_NV_draw_buffers in ESSL output. This flag can be used to emulate

-    // EXT_draw_buffers by using it in combination with GLES3.0 glDrawBuffers

-    // function. This applies to Tegra K1 devices.

-    int NV_draw_buffers;

-

-    // Set to 1 if highp precision is supported in the fragment language.

-    // Default is 0.

-    int FragmentPrecisionHigh;

-

-    // GLSL ES 3.0 constants.

-    int MaxVertexOutputVectors;

-    int MaxFragmentInputVectors;

-    int MinProgramTexelOffset;

-    int MaxProgramTexelOffset;

-

-    // Name Hashing.

-    // Set a 64 bit hash function to enable user-defined name hashing.

-    // Default is NULL.

-    ShHashFunction64 HashFunction;

-

-    // Selects a strategy to use when implementing array index clamping.

-    // Default is SH_CLAMP_WITH_CLAMP_INTRINSIC.

-    ShArrayIndexClampingStrategy ArrayIndexClampingStrategy;

-

-    // The maximum complexity an expression can be.

-    int MaxExpressionComplexity;

-

-    // The maximum depth a call stack can be.

-    int MaxCallStackDepth;

-} ShBuiltInResources;

-

-//

-// Initialize built-in resources with minimum expected values.

-// Parameters:

-// resources: The object to initialize. Will be comparable with memcmp.

-//

-COMPILER_EXPORT void ShInitBuiltInResources(ShBuiltInResources *resources);

-

-//

-// ShHandle held by but opaque to the driver.  It is allocated,

-// managed, and de-allocated by the compiler. Its contents

-// are defined by and used by the compiler.

-//

-// If handle creation fails, 0 will be returned.

-//

-typedef void *ShHandle;

-

-//

-// Returns the a concatenated list of the items in ShBuiltInResources as a

-// null-terminated string.

-// This function must be updated whenever ShBuiltInResources is changed.

-// Parameters:

-// handle: Specifies the handle of the compiler to be used.

-COMPILER_EXPORT const std::string &ShGetBuiltInResourcesString(const ShHandle handle);

-

-//

-// Driver calls these to create and destroy compiler objects.

-//

-// Returns the handle of constructed compiler, null if the requested compiler is

-// not supported.

-// Parameters:

-// type: Specifies the type of shader - GL_FRAGMENT_SHADER or GL_VERTEX_SHADER.

-// spec: Specifies the language spec the compiler must conform to -

-//       SH_GLES2_SPEC or SH_WEBGL_SPEC.

-// output: Specifies the output code type - SH_ESSL_OUTPUT, SH_GLSL_OUTPUT,

-//         SH_HLSL9_OUTPUT or SH_HLSL11_OUTPUT.

-// resources: Specifies the built-in resources.

-COMPILER_EXPORT ShHandle ShConstructCompiler(

-    sh::GLenum type,

-    ShShaderSpec spec,

-    ShShaderOutput output,

-    const ShBuiltInResources *resources);

-COMPILER_EXPORT void ShDestruct(ShHandle handle);

-

-//

-// Compiles the given shader source.

-// If the function succeeds, the return value is true, else false.

-// Parameters:

-// handle: Specifies the handle of compiler to be used.

-// shaderStrings: Specifies an array of pointers to null-terminated strings

-//                containing the shader source code.

-// numStrings: Specifies the number of elements in shaderStrings array.

-// compileOptions: A mask containing the following parameters:

-// SH_VALIDATE: Validates shader to ensure that it conforms to the spec

-//              specified during compiler construction.

-// SH_VALIDATE_LOOP_INDEXING: Validates loop and indexing in the shader to

-//                            ensure that they do not exceed the minimum

-//                            functionality mandated in GLSL 1.0 spec,

-//                            Appendix A, Section 4 and 5.

-//                            There is no need to specify this parameter when

-//                            compiling for WebGL - it is implied.

-// SH_INTERMEDIATE_TREE: Writes intermediate tree to info log.

-//                       Can be queried by calling ShGetInfoLog().

-// SH_OBJECT_CODE: Translates intermediate tree to glsl or hlsl shader.

-//                 Can be queried by calling ShGetObjectCode().

-// SH_VARIABLES: Extracts attributes, uniforms, and varyings.

-//               Can be queried by calling ShGetVariableInfo().

-//

-COMPILER_EXPORT bool ShCompile(

-    const ShHandle handle,

-    const char * const shaderStrings[],

-    size_t numStrings,

-    int compileOptions);

-

-// Return the version of the shader language.

-COMPILER_EXPORT int ShGetShaderVersion(const ShHandle handle);

-

-// Return the currently set language output type.

-COMPILER_EXPORT ShShaderOutput ShGetShaderOutputType(

-    const ShHandle handle);

-

-// Returns null-terminated information log for a compiled shader.

-// Parameters:

-// handle: Specifies the compiler

-COMPILER_EXPORT const std::string &ShGetInfoLog(const ShHandle handle);

-

-// Returns null-terminated object code for a compiled shader.

-// Parameters:

-// handle: Specifies the compiler

-COMPILER_EXPORT const std::string &ShGetObjectCode(const ShHandle handle);

-

-// Returns a (original_name, hash) map containing all the user defined

-// names in the shader, including variable names, function names, struct

-// names, and struct field names.

-// Parameters:

-// handle: Specifies the compiler

-COMPILER_EXPORT const std::map<std::string, std::string> *ShGetNameHashingMap(

-    const ShHandle handle);

-

-// Shader variable inspection.

-// Returns a pointer to a list of variables of the designated type.

-// (See ShaderVars.h for type definitions, included above)

-// Returns NULL on failure.

-// Parameters:

-// handle: Specifies the compiler

-COMPILER_EXPORT const std::vector<sh::Uniform> *ShGetUniforms(const ShHandle handle);

-COMPILER_EXPORT const std::vector<sh::Varying> *ShGetVaryings(const ShHandle handle);

-COMPILER_EXPORT const std::vector<sh::Attribute> *ShGetAttributes(const ShHandle handle);

-COMPILER_EXPORT const std::vector<sh::Attribute> *ShGetOutputVariables(const ShHandle handle);

-COMPILER_EXPORT const std::vector<sh::InterfaceBlock> *ShGetInterfaceBlocks(const ShHandle handle);

-

-typedef struct

-{

-    sh::GLenum type;

-    int size;

-} ShVariableInfo;

-

-// Returns true if the passed in variables pack in maxVectors following

-// the packing rules from the GLSL 1.017 spec, Appendix A, section 7.

-// Returns false otherwise. Also look at the SH_ENFORCE_PACKING_RESTRICTIONS

-// flag above.

-// Parameters:

-// maxVectors: the available rows of registers.

-// varInfoArray: an array of variable info (types and sizes).

-// varInfoArraySize: the size of the variable array.

-COMPILER_EXPORT bool ShCheckVariablesWithinPackingLimits(

-    int maxVectors,

-    ShVariableInfo *varInfoArray,

-    size_t varInfoArraySize);

-

-// Gives the compiler-assigned register for an interface block.

-// The method writes the value to the output variable "indexOut".

-// Returns true if it found a valid interface block, false otherwise.

-// Parameters:

-// handle: Specifies the compiler

-// interfaceBlockName: Specifies the interface block

-// indexOut: output variable that stores the assigned register

-COMPILER_EXPORT bool ShGetInterfaceBlockRegister(const ShHandle handle,

-                                                 const std::string &interfaceBlockName,

-                                                 unsigned int *indexOut);

-

-// Gives the compiler-assigned register for uniforms in the default

-// interface block.

-// The method writes the value to the output variable "indexOut".

-// Returns true if it found a valid default uniform, false otherwise.

-// Parameters:

-// handle: Specifies the compiler

-// interfaceBlockName: Specifies the uniform

-// indexOut: output variable that stores the assigned register

-COMPILER_EXPORT bool ShGetUniformRegister(const ShHandle handle,

-                                          const std::string &uniformName,

-                                          unsigned int *indexOut);

-

-#endif // _COMPILER_INTERFACE_INCLUDED_

+//
+// Copyright (c) 2002-2013 The ANGLE Project Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+#ifndef _COMPILER_INTERFACE_INCLUDED_
+#define _COMPILER_INTERFACE_INCLUDED_
+
+#if defined(COMPONENT_BUILD) && !defined(ANGLE_TRANSLATOR_STATIC)
+#if defined(_WIN32) || defined(_WIN64)
+
+#if defined(ANGLE_TRANSLATOR_IMPLEMENTATION)
+#define COMPILER_EXPORT __declspec(dllexport)
+#else
+#define COMPILER_EXPORT __declspec(dllimport)
+#endif  // defined(ANGLE_TRANSLATOR_IMPLEMENTATION)
+
+#else  // defined(_WIN32) || defined(_WIN64)
+#define COMPILER_EXPORT __attribute__((visibility("default")))
+#endif
+
+#else  // defined(COMPONENT_BUILD) && !defined(ANGLE_TRANSLATOR_STATIC)
+#define COMPILER_EXPORT
+#endif
+
+#include <stddef.h>
+
+#include "KHR/khrplatform.h"
+
+#include <map>
+#include <string>
+#include <vector>
+
+//
+// This is the platform independent interface between an OGL driver
+// and the shading language compiler.
+//
+
+namespace sh
+{
+// GLenum alias
+typedef unsigned int GLenum;
+}
+
+// Must be included after GLenum proxy typedef
+// Note: make sure to increment ANGLE_SH_VERSION when changing ShaderVars.h
+#include "ShaderVars.h"
+
+// Version number for shader translation API.
+// It is incremented every time the API changes.
+#define ANGLE_SH_VERSION 132
+
+typedef enum {
+  SH_GLES2_SPEC = 0x8B40,
+  SH_WEBGL_SPEC = 0x8B41,
+
+  SH_GLES3_SPEC = 0x8B86,
+  SH_WEBGL2_SPEC = 0x8B87,
+
+  // The CSS Shaders spec is a subset of the WebGL spec.
+  //
+  // In both CSS vertex and fragment shaders, ANGLE:
+  // (1) Reserves the "css_" prefix.
+  // (2) Renames the main function to css_main.
+  // (3) Disables the gl_MaxDrawBuffers built-in.
+  //
+  // In CSS fragment shaders, ANGLE:
+  // (1) Disables the gl_FragColor built-in.
+  // (2) Disables the gl_FragData built-in.
+  // (3) Enables the css_MixColor built-in.
+  // (4) Enables the css_ColorMatrix built-in.
+  //
+  // After passing a CSS shader through ANGLE, the browser is expected to append
+  // a new main function to it.
+  // This new main function will call the css_main function.
+  // It may also perform additional operations like varying assignment, texture
+  // access, and gl_FragColor assignment in order to implement the CSS Shaders
+  // blend modes.
+  //
+  SH_CSS_SHADERS_SPEC = 0x8B42
+} ShShaderSpec;
+
+typedef enum {
+  SH_ESSL_OUTPUT   = 0x8B45,
+  SH_GLSL_OUTPUT   = 0x8B46,
+  SH_HLSL_OUTPUT   = 0x8B47,
+  SH_HLSL9_OUTPUT  = 0x8B47,
+  SH_HLSL11_OUTPUT = 0x8B48
+} ShShaderOutput;
+
+// Compile options.
+typedef enum {
+  SH_VALIDATE                = 0,
+  SH_VALIDATE_LOOP_INDEXING  = 0x0001,
+  SH_INTERMEDIATE_TREE       = 0x0002,
+  SH_OBJECT_CODE             = 0x0004,
+  SH_VARIABLES               = 0x0008,
+  SH_LINE_DIRECTIVES         = 0x0010,
+  SH_SOURCE_PATH             = 0x0020,
+  SH_UNROLL_FOR_LOOP_WITH_INTEGER_INDEX = 0x0040,
+  // If a sampler array index happens to be a loop index,
+  //   1) if its type is integer, unroll the loop.
+  //   2) if its type is float, fail the shader compile.
+  // This is to work around a mac driver bug.
+  SH_UNROLL_FOR_LOOP_WITH_SAMPLER_ARRAY_INDEX = 0x0080,
+
+  // This is needed only as a workaround for certain OpenGL driver bugs.
+  SH_EMULATE_BUILT_IN_FUNCTIONS = 0x0100,
+
+  // This is an experimental flag to enforce restrictions that aim to prevent 
+  // timing attacks.
+  // It generates compilation errors for shaders that could expose sensitive
+  // texture information via the timing channel.
+  // To use this flag, you must compile the shader under the WebGL spec
+  // (using the SH_WEBGL_SPEC flag).
+  SH_TIMING_RESTRICTIONS = 0x0200,
+
+  // This flag prints the dependency graph that is used to enforce timing
+  // restrictions on fragment shaders.
+  // This flag only has an effect if all of the following are true:
+  // - The shader spec is SH_WEBGL_SPEC.
+  // - The compile options contain the SH_TIMING_RESTRICTIONS flag.
+  // - The shader type is GL_FRAGMENT_SHADER.
+  SH_DEPENDENCY_GRAPH = 0x0400,
+
+  // Enforce the GLSL 1.017 Appendix A section 7 packing restrictions.
+  // This flag only enforces (and can only enforce) the packing
+  // restrictions for uniform variables in both vertex and fragment
+  // shaders. ShCheckVariablesWithinPackingLimits() lets embedders
+  // enforce the packing restrictions for varying variables during
+  // program link time.
+  SH_ENFORCE_PACKING_RESTRICTIONS = 0x0800,
+
+  // This flag ensures all indirect (expression-based) array indexing
+  // is clamped to the bounds of the array. This ensures, for example,
+  // that you cannot read off the end of a uniform, whether an array
+  // vec234, or mat234 type. The ShArrayIndexClampingStrategy enum,
+  // specified in the ShBuiltInResources when constructing the
+  // compiler, selects the strategy for the clamping implementation.
+  SH_CLAMP_INDIRECT_ARRAY_BOUNDS = 0x1000,
+
+  // This flag limits the complexity of an expression.
+  SH_LIMIT_EXPRESSION_COMPLEXITY = 0x2000,
+
+  // This flag limits the depth of the call stack.
+  SH_LIMIT_CALL_STACK_DEPTH = 0x4000,
+
+  // This flag initializes gl_Position to vec4(0,0,0,0) at the
+  // beginning of the vertex shader's main(), and has no effect in the
+  // fragment shader. It is intended as a workaround for drivers which
+  // incorrectly fail to link programs if gl_Position is not written.
+  SH_INIT_GL_POSITION = 0x8000,
+
+  // This flag replaces
+  //   "a && b" with "a ? b : false",
+  //   "a || b" with "a ? true : b".
+  // This is to work around a MacOSX driver bug that |b| is executed
+  // independent of |a|'s value.
+  SH_UNFOLD_SHORT_CIRCUIT = 0x10000,
+
+  // This flag initializes varyings without static use in vertex shader
+  // at the beginning of main(), and has no effects in the fragment shader.
+  // It is intended as a workaround for drivers which incorrectly optimize
+  // out such varyings and cause a link failure.
+  SH_INIT_VARYINGS_WITHOUT_STATIC_USE = 0x20000,
+
+  // This flag scalarizes vec/ivec/bvec/mat constructor args.
+  // It is intended as a workaround for Linux/Mac driver bugs.
+  SH_SCALARIZE_VEC_AND_MAT_CONSTRUCTOR_ARGS = 0x40000,
+
+  // This flag overwrites a struct name with a unique prefix.
+  // It is intended as a workaround for drivers that do not handle
+  // struct scopes correctly, including all Mac drivers and Linux AMD.
+  SH_REGENERATE_STRUCT_NAMES = 0x80000,
+} ShCompileOptions;
+
+// Defines alternate strategies for implementing array index clamping.
+typedef enum {
+  // Use the clamp intrinsic for array index clamping.
+  SH_CLAMP_WITH_CLAMP_INTRINSIC = 1,
+
+  // Use a user-defined function for array index clamping.
+  SH_CLAMP_WITH_USER_DEFINED_INT_CLAMP_FUNCTION
+} ShArrayIndexClampingStrategy;
+
+//
+// Driver must call this first, once, before doing any other
+// compiler operations.
+// If the function succeeds, the return value is true, else false.
+//
+COMPILER_EXPORT bool ShInitialize();
+//
+// Driver should call this at shutdown.
+// If the function succeeds, the return value is true, else false.
+//
+COMPILER_EXPORT bool ShFinalize();
+
+// The 64 bits hash function. The first parameter is the input string; the
+// second parameter is the string length.
+typedef khronos_uint64_t (*ShHashFunction64)(const char*, size_t);
+
+//
+// Implementation dependent built-in resources (constants and extensions).
+// The names for these resources has been obtained by stripping gl_/GL_.
+//
+typedef struct
+{
+    // Constants.
+    int MaxVertexAttribs;
+    int MaxVertexUniformVectors;
+    int MaxVaryingVectors;
+    int MaxVertexTextureImageUnits;
+    int MaxCombinedTextureImageUnits;
+    int MaxTextureImageUnits;
+    int MaxFragmentUniformVectors;
+    int MaxDrawBuffers;
+
+    // Extensions.
+    // Set to 1 to enable the extension, else 0.
+    int OES_standard_derivatives;
+    int OES_EGL_image_external;
+    int ARB_texture_rectangle;
+    int EXT_draw_buffers;
+    int EXT_frag_depth;
+    int EXT_shader_texture_lod;
+
+    // Set to 1 to enable replacing GL_EXT_draw_buffers #extension directives
+    // with GL_NV_draw_buffers in ESSL output. This flag can be used to emulate
+    // EXT_draw_buffers by using it in combination with GLES3.0 glDrawBuffers
+    // function. This applies to Tegra K1 devices.
+    int NV_draw_buffers;
+
+    // Set to 1 if highp precision is supported in the fragment language.
+    // Default is 0.
+    int FragmentPrecisionHigh;
+
+    // GLSL ES 3.0 constants.
+    int MaxVertexOutputVectors;
+    int MaxFragmentInputVectors;
+    int MinProgramTexelOffset;
+    int MaxProgramTexelOffset;
+
+    // Name Hashing.
+    // Set a 64 bit hash function to enable user-defined name hashing.
+    // Default is NULL.
+    ShHashFunction64 HashFunction;
+
+    // Selects a strategy to use when implementing array index clamping.
+    // Default is SH_CLAMP_WITH_CLAMP_INTRINSIC.
+    ShArrayIndexClampingStrategy ArrayIndexClampingStrategy;
+
+    // The maximum complexity an expression can be.
+    int MaxExpressionComplexity;
+
+    // The maximum depth a call stack can be.
+    int MaxCallStackDepth;
+} ShBuiltInResources;
+
+//
+// Initialize built-in resources with minimum expected values.
+// Parameters:
+// resources: The object to initialize. Will be comparable with memcmp.
+//
+COMPILER_EXPORT void ShInitBuiltInResources(ShBuiltInResources *resources);
+
+//
+// ShHandle held by but opaque to the driver.  It is allocated,
+// managed, and de-allocated by the compiler. Its contents
+// are defined by and used by the compiler.
+//
+// If handle creation fails, 0 will be returned.
+//
+typedef void *ShHandle;
+
+//
+// Returns the a concatenated list of the items in ShBuiltInResources as a
+// null-terminated string.
+// This function must be updated whenever ShBuiltInResources is changed.
+// Parameters:
+// handle: Specifies the handle of the compiler to be used.
+COMPILER_EXPORT const std::string &ShGetBuiltInResourcesString(const ShHandle handle);
+
+//
+// Driver calls these to create and destroy compiler objects.
+//
+// Returns the handle of constructed compiler, null if the requested compiler is
+// not supported.
+// Parameters:
+// type: Specifies the type of shader - GL_FRAGMENT_SHADER or GL_VERTEX_SHADER.
+// spec: Specifies the language spec the compiler must conform to -
+//       SH_GLES2_SPEC or SH_WEBGL_SPEC.
+// output: Specifies the output code type - SH_ESSL_OUTPUT, SH_GLSL_OUTPUT,
+//         SH_HLSL9_OUTPUT or SH_HLSL11_OUTPUT.
+// resources: Specifies the built-in resources.
+COMPILER_EXPORT ShHandle ShConstructCompiler(
+    sh::GLenum type,
+    ShShaderSpec spec,
+    ShShaderOutput output,
+    const ShBuiltInResources *resources);
+COMPILER_EXPORT void ShDestruct(ShHandle handle);
+
+//
+// Compiles the given shader source.
+// If the function succeeds, the return value is true, else false.
+// Parameters:
+// handle: Specifies the handle of compiler to be used.
+// shaderStrings: Specifies an array of pointers to null-terminated strings
+//                containing the shader source code.
+// numStrings: Specifies the number of elements in shaderStrings array.
+// compileOptions: A mask containing the following parameters:
+// SH_VALIDATE: Validates shader to ensure that it conforms to the spec
+//              specified during compiler construction.
+// SH_VALIDATE_LOOP_INDEXING: Validates loop and indexing in the shader to
+//                            ensure that they do not exceed the minimum
+//                            functionality mandated in GLSL 1.0 spec,
+//                            Appendix A, Section 4 and 5.
+//                            There is no need to specify this parameter when
+//                            compiling for WebGL - it is implied.
+// SH_INTERMEDIATE_TREE: Writes intermediate tree to info log.
+//                       Can be queried by calling ShGetInfoLog().
+// SH_OBJECT_CODE: Translates intermediate tree to glsl or hlsl shader.
+//                 Can be queried by calling ShGetObjectCode().
+// SH_VARIABLES: Extracts attributes, uniforms, and varyings.
+//               Can be queried by calling ShGetVariableInfo().
+//
+COMPILER_EXPORT bool ShCompile(
+    const ShHandle handle,
+    const char * const shaderStrings[],
+    size_t numStrings,
+    int compileOptions);
+
+// Return the version of the shader language.
+COMPILER_EXPORT int ShGetShaderVersion(const ShHandle handle);
+
+// Return the currently set language output type.
+COMPILER_EXPORT ShShaderOutput ShGetShaderOutputType(
+    const ShHandle handle);
+
+// Returns null-terminated information log for a compiled shader.
+// Parameters:
+// handle: Specifies the compiler
+COMPILER_EXPORT const std::string &ShGetInfoLog(const ShHandle handle);
+
+// Returns null-terminated object code for a compiled shader.
+// Parameters:
+// handle: Specifies the compiler
+COMPILER_EXPORT const std::string &ShGetObjectCode(const ShHandle handle);
+
+// Returns a (original_name, hash) map containing all the user defined
+// names in the shader, including variable names, function names, struct
+// names, and struct field names.
+// Parameters:
+// handle: Specifies the compiler
+COMPILER_EXPORT const std::map<std::string, std::string> *ShGetNameHashingMap(
+    const ShHandle handle);
+
+// Shader variable inspection.
+// Returns a pointer to a list of variables of the designated type.
+// (See ShaderVars.h for type definitions, included above)
+// Returns NULL on failure.
+// Parameters:
+// handle: Specifies the compiler
+COMPILER_EXPORT const std::vector<sh::Uniform> *ShGetUniforms(const ShHandle handle);
+COMPILER_EXPORT const std::vector<sh::Varying> *ShGetVaryings(const ShHandle handle);
+COMPILER_EXPORT const std::vector<sh::Attribute> *ShGetAttributes(const ShHandle handle);
+COMPILER_EXPORT const std::vector<sh::Attribute> *ShGetOutputVariables(const ShHandle handle);
+COMPILER_EXPORT const std::vector<sh::InterfaceBlock> *ShGetInterfaceBlocks(const ShHandle handle);
+
+typedef struct
+{
+    sh::GLenum type;
+    int size;
+} ShVariableInfo;
+
+// Returns true if the passed in variables pack in maxVectors following
+// the packing rules from the GLSL 1.017 spec, Appendix A, section 7.
+// Returns false otherwise. Also look at the SH_ENFORCE_PACKING_RESTRICTIONS
+// flag above.
+// Parameters:
+// maxVectors: the available rows of registers.
+// varInfoArray: an array of variable info (types and sizes).
+// varInfoArraySize: the size of the variable array.
+COMPILER_EXPORT bool ShCheckVariablesWithinPackingLimits(
+    int maxVectors,
+    ShVariableInfo *varInfoArray,
+    size_t varInfoArraySize);
+
+// Gives the compiler-assigned register for an interface block.
+// The method writes the value to the output variable "indexOut".
+// Returns true if it found a valid interface block, false otherwise.
+// Parameters:
+// handle: Specifies the compiler
+// interfaceBlockName: Specifies the interface block
+// indexOut: output variable that stores the assigned register
+COMPILER_EXPORT bool ShGetInterfaceBlockRegister(const ShHandle handle,
+                                                 const std::string &interfaceBlockName,
+                                                 unsigned int *indexOut);
+
+// Gives the compiler-assigned register for uniforms in the default
+// interface block.
+// The method writes the value to the output variable "indexOut".
+// Returns true if it found a valid default uniform, false otherwise.
+// Parameters:
+// handle: Specifies the compiler
+// interfaceBlockName: Specifies the uniform
+// indexOut: output variable that stores the assigned register
+COMPILER_EXPORT bool ShGetUniformRegister(const ShHandle handle,
+                                          const std::string &uniformName,
+                                          unsigned int *indexOut);
+
+#endif // _COMPILER_INTERFACE_INCLUDED_
diff --git a/platform/winrt/include/GLSLANG/ShaderVars.h b/platform/winrt/include/GLSLANG/ShaderVars.h
index dc452b0cf4..da21c3e76e 100644
--- a/platform/winrt/include/GLSLANG/ShaderVars.h
+++ b/platform/winrt/include/GLSLANG/ShaderVars.h
@@ -1,185 +1,185 @@
-//

-// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved.

-// Use of this source code is governed by a BSD-style license that can be

-// found in the LICENSE file.

-//

-// ShaderVars.h:

-//  Types to represent GL variables (varyings, uniforms, etc)

-//

-

-#ifndef _COMPILER_INTERFACE_VARIABLES_

-#define _COMPILER_INTERFACE_VARIABLES_

-

-#include <string>

-#include <vector>

-#include <algorithm>

-

-// Assume ShaderLang.h is included before ShaderVars.h, for sh::GLenum

-// Note: make sure to increment ANGLE_SH_VERSION when changing ShaderVars.h

-

-namespace sh

-{

-

-// Varying interpolation qualifier, see section 4.3.9 of the ESSL 3.00.4 spec

-enum InterpolationType

-{

-    INTERPOLATION_SMOOTH,

-    INTERPOLATION_CENTROID,

-    INTERPOLATION_FLAT

-};

-

-// Uniform block layout qualifier, see section 4.3.8.3 of the ESSL 3.00.4 spec

-enum BlockLayoutType

-{

-    BLOCKLAYOUT_STANDARD,

-    BLOCKLAYOUT_PACKED,

-    BLOCKLAYOUT_SHARED

-};

-

-// Base class for all variables defined in shaders, including Varyings, Uniforms, etc

-// Note: we must override the copy constructor and assignment operator so we can

-// work around excessive GCC binary bloating:

-// See https://code.google.com/p/angleproject/issues/detail?id=697

-struct COMPILER_EXPORT ShaderVariable

-{

-    ShaderVariable();

-    ShaderVariable(GLenum typeIn, unsigned int arraySizeIn);

-    ~ShaderVariable();

-    ShaderVariable(const ShaderVariable &other);

-    ShaderVariable &operator=(const ShaderVariable &other);

-

-    bool isArray() const { return arraySize > 0; }

-    unsigned int elementCount() const { return std::max(1u, arraySize); }

-    bool isStruct() const { return !fields.empty(); }

-

-    // All of the shader's variables are described using nested data

-    // structures. This is needed in order to disambiguate similar looking

-    // types, such as two structs containing the same fields, but in

-    // different orders. "findInfoByMappedName" provides an easy query for

-    // users to dive into the data structure and fetch the unique variable

-    // instance corresponding to a dereferencing chain of the top-level

-    // variable.

-    // Given a mapped name like 'a[0].b.c[0]', return the ShaderVariable

-    // that defines 'c' in |leafVar|, and the original name 'A[0].B.C[0]'

-    // in |originalName|, based on the assumption that |this| defines 'a'.

-    // If no match is found, return false.

-    bool findInfoByMappedName(const std::string &mappedFullName,

-                              const ShaderVariable **leafVar,

-                              std::string* originalFullName) const;

-

-    GLenum type;

-    GLenum precision;

-    std::string name;

-    std::string mappedName;

-    unsigned int arraySize;

-    bool staticUse;

-    std::vector<ShaderVariable> fields;

-    std::string structName;

-

-  protected:

-    bool isSameVariableAtLinkTime(const ShaderVariable &other,

-                                  bool matchPrecision) const;

-

-    bool operator==(const ShaderVariable &other) const;

-    bool operator!=(const ShaderVariable &other) const

-    {

-        return !operator==(other);

-    }

-};

-

-struct COMPILER_EXPORT Uniform : public ShaderVariable

-{

-    Uniform();

-    ~Uniform();

-    Uniform(const Uniform &other);

-    Uniform &operator=(const Uniform &other);

-    bool operator==(const Uniform &other) const;

-    bool operator!=(const Uniform &other) const

-    {

-        return !operator==(other);

-    }

-

-    // Decide whether two uniforms are the same at shader link time,

-    // assuming one from vertex shader and the other from fragment shader.

-    // See GLSL ES Spec 3.00.3, sec 4.3.5.

-    bool isSameUniformAtLinkTime(const Uniform &other) const;

-};

-

-struct COMPILER_EXPORT Attribute : public ShaderVariable

-{

-    Attribute();

-    ~Attribute();

-    Attribute(const Attribute &other);

-    Attribute &operator=(const Attribute &other);

-    bool operator==(const Attribute &other) const;

-    bool operator!=(const Attribute &other) const

-    {

-        return !operator==(other);

-    }

-

-    int location;

-};

-

-struct COMPILER_EXPORT InterfaceBlockField : public ShaderVariable

-{

-    InterfaceBlockField();

-    ~InterfaceBlockField();

-    InterfaceBlockField(const InterfaceBlockField &other);

-    InterfaceBlockField &operator=(const InterfaceBlockField &other);

-    bool operator==(const InterfaceBlockField &other) const;

-    bool operator!=(const InterfaceBlockField &other) const

-    {

-        return !operator==(other);

-    }

-

-    // Decide whether two InterfaceBlock fields are the same at shader

-    // link time, assuming one from vertex shader and the other from

-    // fragment shader.

-    // See GLSL ES Spec 3.00.3, sec 4.3.7.

-    bool isSameInterfaceBlockFieldAtLinkTime(

-        const InterfaceBlockField &other) const;

-

-    bool isRowMajorLayout;

-};

-

-struct COMPILER_EXPORT Varying : public ShaderVariable

-{

-    Varying();

-    ~Varying();

-    Varying(const Varying &otherg);

-    Varying &operator=(const Varying &other);

-    bool operator==(const Varying &other) const;

-    bool operator!=(const Varying &other) const

-    {

-        return !operator==(other);

-    }

-

-    // Decide whether two varyings are the same at shader link time,

-    // assuming one from vertex shader and the other from fragment shader.

-    // See GLSL ES Spec 3.00.3, sec 4.3.9.

-    bool isSameVaryingAtLinkTime(const Varying &other) const;

-

-    InterpolationType interpolation;

-    bool isInvariant;

-};

-

-struct COMPILER_EXPORT InterfaceBlock

-{

-    InterfaceBlock();

-    ~InterfaceBlock();

-    InterfaceBlock(const InterfaceBlock &other);

-    InterfaceBlock &operator=(const InterfaceBlock &other);

-

-    std::string name;

-    std::string mappedName;

-    std::string instanceName;

-    unsigned int arraySize;

-    BlockLayoutType layout;

-    bool isRowMajorLayout;

-    bool staticUse;

-    std::vector<InterfaceBlockField> fields;

-};

-

-}

-

-#endif // _COMPILER_INTERFACE_VARIABLES_

+//
+// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+// ShaderVars.h:
+//  Types to represent GL variables (varyings, uniforms, etc)
+//
+
+#ifndef _COMPILER_INTERFACE_VARIABLES_
+#define _COMPILER_INTERFACE_VARIABLES_
+
+#include <string>
+#include <vector>
+#include <algorithm>
+
+// Assume ShaderLang.h is included before ShaderVars.h, for sh::GLenum
+// Note: make sure to increment ANGLE_SH_VERSION when changing ShaderVars.h
+
+namespace sh
+{
+
+// Varying interpolation qualifier, see section 4.3.9 of the ESSL 3.00.4 spec
+enum InterpolationType
+{
+    INTERPOLATION_SMOOTH,
+    INTERPOLATION_CENTROID,
+    INTERPOLATION_FLAT
+};
+
+// Uniform block layout qualifier, see section 4.3.8.3 of the ESSL 3.00.4 spec
+enum BlockLayoutType
+{
+    BLOCKLAYOUT_STANDARD,
+    BLOCKLAYOUT_PACKED,
+    BLOCKLAYOUT_SHARED
+};
+
+// Base class for all variables defined in shaders, including Varyings, Uniforms, etc
+// Note: we must override the copy constructor and assignment operator so we can
+// work around excessive GCC binary bloating:
+// See https://code.google.com/p/angleproject/issues/detail?id=697
+struct COMPILER_EXPORT ShaderVariable
+{
+    ShaderVariable();
+    ShaderVariable(GLenum typeIn, unsigned int arraySizeIn);
+    ~ShaderVariable();
+    ShaderVariable(const ShaderVariable &other);
+    ShaderVariable &operator=(const ShaderVariable &other);
+
+    bool isArray() const { return arraySize > 0; }
+    unsigned int elementCount() const { return std::max(1u, arraySize); }
+    bool isStruct() const { return !fields.empty(); }
+
+    // All of the shader's variables are described using nested data
+    // structures. This is needed in order to disambiguate similar looking
+    // types, such as two structs containing the same fields, but in
+    // different orders. "findInfoByMappedName" provides an easy query for
+    // users to dive into the data structure and fetch the unique variable
+    // instance corresponding to a dereferencing chain of the top-level
+    // variable.
+    // Given a mapped name like 'a[0].b.c[0]', return the ShaderVariable
+    // that defines 'c' in |leafVar|, and the original name 'A[0].B.C[0]'
+    // in |originalName|, based on the assumption that |this| defines 'a'.
+    // If no match is found, return false.
+    bool findInfoByMappedName(const std::string &mappedFullName,
+                              const ShaderVariable **leafVar,
+                              std::string* originalFullName) const;
+
+    GLenum type;
+    GLenum precision;
+    std::string name;
+    std::string mappedName;
+    unsigned int arraySize;
+    bool staticUse;
+    std::vector<ShaderVariable> fields;
+    std::string structName;
+
+  protected:
+    bool isSameVariableAtLinkTime(const ShaderVariable &other,
+                                  bool matchPrecision) const;
+
+    bool operator==(const ShaderVariable &other) const;
+    bool operator!=(const ShaderVariable &other) const
+    {
+        return !operator==(other);
+    }
+};
+
+struct COMPILER_EXPORT Uniform : public ShaderVariable
+{
+    Uniform();
+    ~Uniform();
+    Uniform(const Uniform &other);
+    Uniform &operator=(const Uniform &other);
+    bool operator==(const Uniform &other) const;
+    bool operator!=(const Uniform &other) const
+    {
+        return !operator==(other);
+    }
+
+    // Decide whether two uniforms are the same at shader link time,
+    // assuming one from vertex shader and the other from fragment shader.
+    // See GLSL ES Spec 3.00.3, sec 4.3.5.
+    bool isSameUniformAtLinkTime(const Uniform &other) const;
+};
+
+struct COMPILER_EXPORT Attribute : public ShaderVariable
+{
+    Attribute();
+    ~Attribute();
+    Attribute(const Attribute &other);
+    Attribute &operator=(const Attribute &other);
+    bool operator==(const Attribute &other) const;
+    bool operator!=(const Attribute &other) const
+    {
+        return !operator==(other);
+    }
+
+    int location;
+};
+
+struct COMPILER_EXPORT InterfaceBlockField : public ShaderVariable
+{
+    InterfaceBlockField();
+    ~InterfaceBlockField();
+    InterfaceBlockField(const InterfaceBlockField &other);
+    InterfaceBlockField &operator=(const InterfaceBlockField &other);
+    bool operator==(const InterfaceBlockField &other) const;
+    bool operator!=(const InterfaceBlockField &other) const
+    {
+        return !operator==(other);
+    }
+
+    // Decide whether two InterfaceBlock fields are the same at shader
+    // link time, assuming one from vertex shader and the other from
+    // fragment shader.
+    // See GLSL ES Spec 3.00.3, sec 4.3.7.
+    bool isSameInterfaceBlockFieldAtLinkTime(
+        const InterfaceBlockField &other) const;
+
+    bool isRowMajorLayout;
+};
+
+struct COMPILER_EXPORT Varying : public ShaderVariable
+{
+    Varying();
+    ~Varying();
+    Varying(const Varying &otherg);
+    Varying &operator=(const Varying &other);
+    bool operator==(const Varying &other) const;
+    bool operator!=(const Varying &other) const
+    {
+        return !operator==(other);
+    }
+
+    // Decide whether two varyings are the same at shader link time,
+    // assuming one from vertex shader and the other from fragment shader.
+    // See GLSL ES Spec 3.00.3, sec 4.3.9.
+    bool isSameVaryingAtLinkTime(const Varying &other) const;
+
+    InterpolationType interpolation;
+    bool isInvariant;
+};
+
+struct COMPILER_EXPORT InterfaceBlock
+{
+    InterfaceBlock();
+    ~InterfaceBlock();
+    InterfaceBlock(const InterfaceBlock &other);
+    InterfaceBlock &operator=(const InterfaceBlock &other);
+
+    std::string name;
+    std::string mappedName;
+    std::string instanceName;
+    unsigned int arraySize;
+    BlockLayoutType layout;
+    bool isRowMajorLayout;
+    bool staticUse;
+    std::vector<InterfaceBlockField> fields;
+};
+
+}
+
+#endif // _COMPILER_INTERFACE_VARIABLES_
diff --git a/platform/winrt/include/KHR/khrplatform.h b/platform/winrt/include/KHR/khrplatform.h
index 0921b895a5..c9e6f17d34 100644
--- a/platform/winrt/include/KHR/khrplatform.h
+++ b/platform/winrt/include/KHR/khrplatform.h
@@ -1,282 +1,282 @@
-#ifndef __khrplatform_h_

-#define __khrplatform_h_

-

-/*

-** Copyright (c) 2008-2009 The Khronos Group Inc.

-**

-** Permission is hereby granted, free of charge, to any person obtaining a

-** copy of this software and/or associated documentation files (the

-** "Materials"), to deal in the Materials without restriction, including

-** without limitation the rights to use, copy, modify, merge, publish,

-** distribute, sublicense, and/or sell copies of the Materials, and to

-** permit persons to whom the Materials are furnished to do so, subject to

-** the following conditions:

-**

-** The above copyright notice and this permission notice shall be included

-** in all copies or substantial portions of the Materials.

-**

-** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

-** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

-** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.

-** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY

-** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

-** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

-** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

-*/

-

-/* Khronos platform-specific types and definitions.

- *

- * $Revision: 23298 $ on $Date: 2013-09-30 17:07:13 -0700 (Mon, 30 Sep 2013) $

- *

- * Adopters may modify this file to suit their platform. Adopters are

- * encouraged to submit platform specific modifications to the Khronos

- * group so that they can be included in future versions of this file.

- * Please submit changes by sending them to the public Khronos Bugzilla

- * (http://khronos.org/bugzilla) by filing a bug against product

- * "Khronos (general)" component "Registry".

- *

- * A predefined template which fills in some of the bug fields can be

- * reached using http://tinyurl.com/khrplatform-h-bugreport, but you

- * must create a Bugzilla login first.

- *

- *

- * See the Implementer's Guidelines for information about where this file

- * should be located on your system and for more details of its use:

- *    http://www.khronos.org/registry/implementers_guide.pdf

- *

- * This file should be included as

- *        #include <KHR/khrplatform.h>

- * by Khronos client API header files that use its types and defines.

- *

- * The types in khrplatform.h should only be used to define API-specific types.

- *

- * Types defined in khrplatform.h:

- *    khronos_int8_t              signed   8  bit

- *    khronos_uint8_t             unsigned 8  bit

- *    khronos_int16_t             signed   16 bit

- *    khronos_uint16_t            unsigned 16 bit

- *    khronos_int32_t             signed   32 bit

- *    khronos_uint32_t            unsigned 32 bit

- *    khronos_int64_t             signed   64 bit

- *    khronos_uint64_t            unsigned 64 bit

- *    khronos_intptr_t            signed   same number of bits as a pointer

- *    khronos_uintptr_t           unsigned same number of bits as a pointer

- *    khronos_ssize_t             signed   size

- *    khronos_usize_t             unsigned size

- *    khronos_float_t             signed   32 bit floating point

- *    khronos_time_ns_t           unsigned 64 bit time in nanoseconds

- *    khronos_utime_nanoseconds_t unsigned time interval or absolute time in

- *                                         nanoseconds

- *    khronos_stime_nanoseconds_t signed time interval in nanoseconds

- *    khronos_boolean_enum_t      enumerated boolean type. This should

- *      only be used as a base type when a client API's boolean type is

- *      an enum. Client APIs which use an integer or other type for

- *      booleans cannot use this as the base type for their boolean.

- *

- * Tokens defined in khrplatform.h:

- *

- *    KHRONOS_FALSE, KHRONOS_TRUE Enumerated boolean false/true values.

- *

- *    KHRONOS_SUPPORT_INT64 is 1 if 64 bit integers are supported; otherwise 0.

- *    KHRONOS_SUPPORT_FLOAT is 1 if floats are supported; otherwise 0.

- *

- * Calling convention macros defined in this file:

- *    KHRONOS_APICALL

- *    KHRONOS_APIENTRY

- *    KHRONOS_APIATTRIBUTES

- *

- * These may be used in function prototypes as:

- *

- *      KHRONOS_APICALL void KHRONOS_APIENTRY funcname(

- *                                  int arg1,

- *                                  int arg2) KHRONOS_APIATTRIBUTES;

- */

-

-/*-------------------------------------------------------------------------

- * Definition of KHRONOS_APICALL

- *-------------------------------------------------------------------------

- * This precedes the return type of the function in the function prototype.

- */

-#if defined(_WIN32) && !defined(__SCITECH_SNAP__)

-#   define KHRONOS_APICALL __declspec(dllimport)

-#elif defined (__SYMBIAN32__)

-#   define KHRONOS_APICALL IMPORT_C

-#else

-#   define KHRONOS_APICALL

-#endif

-

-/*-------------------------------------------------------------------------

- * Definition of KHRONOS_APIENTRY

- *-------------------------------------------------------------------------

- * This follows the return type of the function  and precedes the function

- * name in the function prototype.

- */

-#if defined(_WIN32) && !defined(_WIN32_WCE) && !defined(__SCITECH_SNAP__)

-    /* Win32 but not WinCE */

-#   define KHRONOS_APIENTRY __stdcall

-#else

-#   define KHRONOS_APIENTRY

-#endif

-

-/*-------------------------------------------------------------------------

- * Definition of KHRONOS_APIATTRIBUTES

- *-------------------------------------------------------------------------

- * This follows the closing parenthesis of the function prototype arguments.

- */

-#if defined (__ARMCC_2__)

-#define KHRONOS_APIATTRIBUTES __softfp

-#else

-#define KHRONOS_APIATTRIBUTES

-#endif

-

-/*-------------------------------------------------------------------------

- * basic type definitions

- *-----------------------------------------------------------------------*/

-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || defined(__GNUC__) || defined(__SCO__) || defined(__USLC__)

-

-

-/*

- * Using <stdint.h>

- */

-#include <stdint.h>

-typedef int32_t                 khronos_int32_t;

-typedef uint32_t                khronos_uint32_t;

-typedef int64_t                 khronos_int64_t;

-typedef uint64_t                khronos_uint64_t;

-#define KHRONOS_SUPPORT_INT64   1

-#define KHRONOS_SUPPORT_FLOAT   1

-

-#elif defined(__VMS ) || defined(__sgi)

-

-/*

- * Using <inttypes.h>

- */

-#include <inttypes.h>

-typedef int32_t                 khronos_int32_t;

-typedef uint32_t                khronos_uint32_t;

-typedef int64_t                 khronos_int64_t;

-typedef uint64_t                khronos_uint64_t;

-#define KHRONOS_SUPPORT_INT64   1

-#define KHRONOS_SUPPORT_FLOAT   1

-

-#elif defined(_WIN32) && !defined(__SCITECH_SNAP__)

-

-/*

- * Win32

- */

-typedef __int32                 khronos_int32_t;

-typedef unsigned __int32        khronos_uint32_t;

-typedef __int64                 khronos_int64_t;

-typedef unsigned __int64        khronos_uint64_t;

-#define KHRONOS_SUPPORT_INT64   1

-#define KHRONOS_SUPPORT_FLOAT   1

-

-#elif defined(__sun__) || defined(__digital__)

-

-/*

- * Sun or Digital

- */

-typedef int                     khronos_int32_t;

-typedef unsigned int            khronos_uint32_t;

-#if defined(__arch64__) || defined(_LP64)

-typedef long int                khronos_int64_t;

-typedef unsigned long int       khronos_uint64_t;

-#else

-typedef long long int           khronos_int64_t;

-typedef unsigned long long int  khronos_uint64_t;

-#endif /* __arch64__ */

-#define KHRONOS_SUPPORT_INT64   1

-#define KHRONOS_SUPPORT_FLOAT   1

-

-#elif 0

-

-/*

- * Hypothetical platform with no float or int64 support

- */

-typedef int                     khronos_int32_t;

-typedef unsigned int            khronos_uint32_t;

-#define KHRONOS_SUPPORT_INT64   0

-#define KHRONOS_SUPPORT_FLOAT   0

-

-#else

-

-/*

- * Generic fallback

- */

-#include <stdint.h>

-typedef int32_t                 khronos_int32_t;

-typedef uint32_t                khronos_uint32_t;

-typedef int64_t                 khronos_int64_t;

-typedef uint64_t                khronos_uint64_t;

-#define KHRONOS_SUPPORT_INT64   1

-#define KHRONOS_SUPPORT_FLOAT   1

-

-#endif

-

-

-/*

- * Types that are (so far) the same on all platforms

- */

-typedef signed   char          khronos_int8_t;

-typedef unsigned char          khronos_uint8_t;

-typedef signed   short int     khronos_int16_t;

-typedef unsigned short int     khronos_uint16_t;

-

-/*

- * Types that differ between LLP64 and LP64 architectures - in LLP64, 

- * pointers are 64 bits, but 'long' is still 32 bits. Win64 appears

- * to be the only LLP64 architecture in current use.

- */

-#ifdef _WIN64

-typedef signed   long long int khronos_intptr_t;

-typedef unsigned long long int khronos_uintptr_t;

-typedef signed   long long int khronos_ssize_t;

-typedef unsigned long long int khronos_usize_t;

-#else

-typedef signed   long  int     khronos_intptr_t;

-typedef unsigned long  int     khronos_uintptr_t;

-typedef signed   long  int     khronos_ssize_t;

-typedef unsigned long  int     khronos_usize_t;

-#endif

-

-#if KHRONOS_SUPPORT_FLOAT

-/*

- * Float type

- */

-typedef          float         khronos_float_t;

-#endif

-

-#if KHRONOS_SUPPORT_INT64

-/* Time types

- *

- * These types can be used to represent a time interval in nanoseconds or

- * an absolute Unadjusted System Time.  Unadjusted System Time is the number

- * of nanoseconds since some arbitrary system event (e.g. since the last

- * time the system booted).  The Unadjusted System Time is an unsigned

- * 64 bit value that wraps back to 0 every 584 years.  Time intervals

- * may be either signed or unsigned.

- */

-typedef khronos_uint64_t       khronos_utime_nanoseconds_t;

-typedef khronos_int64_t        khronos_stime_nanoseconds_t;

-#endif

-

-/*

- * Dummy value used to pad enum types to 32 bits.

- */

-#ifndef KHRONOS_MAX_ENUM

-#define KHRONOS_MAX_ENUM 0x7FFFFFFF

-#endif

-

-/*

- * Enumerated boolean type

- *

- * Values other than zero should be considered to be true.  Therefore

- * comparisons should not be made against KHRONOS_TRUE.

- */

-typedef enum {

-    KHRONOS_FALSE = 0,

-    KHRONOS_TRUE  = 1,

-    KHRONOS_BOOLEAN_ENUM_FORCE_SIZE = KHRONOS_MAX_ENUM

-} khronos_boolean_enum_t;

-

-#endif /* __khrplatform_h_ */

+#ifndef __khrplatform_h_
+#define __khrplatform_h_
+
+/*
+** Copyright (c) 2008-2009 The Khronos Group Inc.
+**
+** Permission is hereby granted, free of charge, to any person obtaining a
+** copy of this software and/or associated documentation files (the
+** "Materials"), to deal in the Materials without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Materials, and to
+** permit persons to whom the Materials are furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be included
+** in all copies or substantial portions of the Materials.
+**
+** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
+*/
+
+/* Khronos platform-specific types and definitions.
+ *
+ * $Revision: 23298 $ on $Date: 2013-09-30 17:07:13 -0700 (Mon, 30 Sep 2013) $
+ *
+ * Adopters may modify this file to suit their platform. Adopters are
+ * encouraged to submit platform specific modifications to the Khronos
+ * group so that they can be included in future versions of this file.
+ * Please submit changes by sending them to the public Khronos Bugzilla
+ * (http://khronos.org/bugzilla) by filing a bug against product
+ * "Khronos (general)" component "Registry".
+ *
+ * A predefined template which fills in some of the bug fields can be
+ * reached using http://tinyurl.com/khrplatform-h-bugreport, but you
+ * must create a Bugzilla login first.
+ *
+ *
+ * See the Implementer's Guidelines for information about where this file
+ * should be located on your system and for more details of its use:
+ *    http://www.khronos.org/registry/implementers_guide.pdf
+ *
+ * This file should be included as
+ *        #include <KHR/khrplatform.h>
+ * by Khronos client API header files that use its types and defines.
+ *
+ * The types in khrplatform.h should only be used to define API-specific types.
+ *
+ * Types defined in khrplatform.h:
+ *    khronos_int8_t              signed   8  bit
+ *    khronos_uint8_t             unsigned 8  bit
+ *    khronos_int16_t             signed   16 bit
+ *    khronos_uint16_t            unsigned 16 bit
+ *    khronos_int32_t             signed   32 bit
+ *    khronos_uint32_t            unsigned 32 bit
+ *    khronos_int64_t             signed   64 bit
+ *    khronos_uint64_t            unsigned 64 bit
+ *    khronos_intptr_t            signed   same number of bits as a pointer
+ *    khronos_uintptr_t           unsigned same number of bits as a pointer
+ *    khronos_ssize_t             signed   size
+ *    khronos_usize_t             unsigned size
+ *    khronos_float_t             signed   32 bit floating point
+ *    khronos_time_ns_t           unsigned 64 bit time in nanoseconds
+ *    khronos_utime_nanoseconds_t unsigned time interval or absolute time in
+ *                                         nanoseconds
+ *    khronos_stime_nanoseconds_t signed time interval in nanoseconds
+ *    khronos_boolean_enum_t      enumerated boolean type. This should
+ *      only be used as a base type when a client API's boolean type is
+ *      an enum. Client APIs which use an integer or other type for
+ *      booleans cannot use this as the base type for their boolean.
+ *
+ * Tokens defined in khrplatform.h:
+ *
+ *    KHRONOS_FALSE, KHRONOS_TRUE Enumerated boolean false/true values.
+ *
+ *    KHRONOS_SUPPORT_INT64 is 1 if 64 bit integers are supported; otherwise 0.
+ *    KHRONOS_SUPPORT_FLOAT is 1 if floats are supported; otherwise 0.
+ *
+ * Calling convention macros defined in this file:
+ *    KHRONOS_APICALL
+ *    KHRONOS_APIENTRY
+ *    KHRONOS_APIATTRIBUTES
+ *
+ * These may be used in function prototypes as:
+ *
+ *      KHRONOS_APICALL void KHRONOS_APIENTRY funcname(
+ *                                  int arg1,
+ *                                  int arg2) KHRONOS_APIATTRIBUTES;
+ */
+
+/*-------------------------------------------------------------------------
+ * Definition of KHRONOS_APICALL
+ *-------------------------------------------------------------------------
+ * This precedes the return type of the function in the function prototype.
+ */
+#if defined(_WIN32) && !defined(__SCITECH_SNAP__)
+#   define KHRONOS_APICALL __declspec(dllimport)
+#elif defined (__SYMBIAN32__)
+#   define KHRONOS_APICALL IMPORT_C
+#else
+#   define KHRONOS_APICALL
+#endif
+
+/*-------------------------------------------------------------------------
+ * Definition of KHRONOS_APIENTRY
+ *-------------------------------------------------------------------------
+ * This follows the return type of the function  and precedes the function
+ * name in the function prototype.
+ */
+#if defined(_WIN32) && !defined(_WIN32_WCE) && !defined(__SCITECH_SNAP__)
+    /* Win32 but not WinCE */
+#   define KHRONOS_APIENTRY __stdcall
+#else
+#   define KHRONOS_APIENTRY
+#endif
+
+/*-------------------------------------------------------------------------
+ * Definition of KHRONOS_APIATTRIBUTES
+ *-------------------------------------------------------------------------
+ * This follows the closing parenthesis of the function prototype arguments.
+ */
+#if defined (__ARMCC_2__)
+#define KHRONOS_APIATTRIBUTES __softfp
+#else
+#define KHRONOS_APIATTRIBUTES
+#endif
+
+/*-------------------------------------------------------------------------
+ * basic type definitions
+ *-----------------------------------------------------------------------*/
+#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || defined(__GNUC__) || defined(__SCO__) || defined(__USLC__)
+
+
+/*
+ * Using <stdint.h>
+ */
+#include <stdint.h>
+typedef int32_t                 khronos_int32_t;
+typedef uint32_t                khronos_uint32_t;
+typedef int64_t                 khronos_int64_t;
+typedef uint64_t                khronos_uint64_t;
+#define KHRONOS_SUPPORT_INT64   1
+#define KHRONOS_SUPPORT_FLOAT   1
+
+#elif defined(__VMS ) || defined(__sgi)
+
+/*
+ * Using <inttypes.h>
+ */
+#include <inttypes.h>
+typedef int32_t                 khronos_int32_t;
+typedef uint32_t                khronos_uint32_t;
+typedef int64_t                 khronos_int64_t;
+typedef uint64_t                khronos_uint64_t;
+#define KHRONOS_SUPPORT_INT64   1
+#define KHRONOS_SUPPORT_FLOAT   1
+
+#elif defined(_WIN32) && !defined(__SCITECH_SNAP__)
+
+/*
+ * Win32
+ */
+typedef __int32                 khronos_int32_t;
+typedef unsigned __int32        khronos_uint32_t;
+typedef __int64                 khronos_int64_t;
+typedef unsigned __int64        khronos_uint64_t;
+#define KHRONOS_SUPPORT_INT64   1
+#define KHRONOS_SUPPORT_FLOAT   1
+
+#elif defined(__sun__) || defined(__digital__)
+
+/*
+ * Sun or Digital
+ */
+typedef int                     khronos_int32_t;
+typedef unsigned int            khronos_uint32_t;
+#if defined(__arch64__) || defined(_LP64)
+typedef long int                khronos_int64_t;
+typedef unsigned long int       khronos_uint64_t;
+#else
+typedef long long int           khronos_int64_t;
+typedef unsigned long long int  khronos_uint64_t;
+#endif /* __arch64__ */
+#define KHRONOS_SUPPORT_INT64   1
+#define KHRONOS_SUPPORT_FLOAT   1
+
+#elif 0
+
+/*
+ * Hypothetical platform with no float or int64 support
+ */
+typedef int                     khronos_int32_t;
+typedef unsigned int            khronos_uint32_t;
+#define KHRONOS_SUPPORT_INT64   0
+#define KHRONOS_SUPPORT_FLOAT   0
+
+#else
+
+/*
+ * Generic fallback
+ */
+#include <stdint.h>
+typedef int32_t                 khronos_int32_t;
+typedef uint32_t                khronos_uint32_t;
+typedef int64_t                 khronos_int64_t;
+typedef uint64_t                khronos_uint64_t;
+#define KHRONOS_SUPPORT_INT64   1
+#define KHRONOS_SUPPORT_FLOAT   1
+
+#endif
+
+
+/*
+ * Types that are (so far) the same on all platforms
+ */
+typedef signed   char          khronos_int8_t;
+typedef unsigned char          khronos_uint8_t;
+typedef signed   short int     khronos_int16_t;
+typedef unsigned short int     khronos_uint16_t;
+
+/*
+ * Types that differ between LLP64 and LP64 architectures - in LLP64, 
+ * pointers are 64 bits, but 'long' is still 32 bits. Win64 appears
+ * to be the only LLP64 architecture in current use.
+ */
+#ifdef _WIN64
+typedef signed   long long int khronos_intptr_t;
+typedef unsigned long long int khronos_uintptr_t;
+typedef signed   long long int khronos_ssize_t;
+typedef unsigned long long int khronos_usize_t;
+#else
+typedef signed   long  int     khronos_intptr_t;
+typedef unsigned long  int     khronos_uintptr_t;
+typedef signed   long  int     khronos_ssize_t;
+typedef unsigned long  int     khronos_usize_t;
+#endif
+
+#if KHRONOS_SUPPORT_FLOAT
+/*
+ * Float type
+ */
+typedef          float         khronos_float_t;
+#endif
+
+#if KHRONOS_SUPPORT_INT64
+/* Time types
+ *
+ * These types can be used to represent a time interval in nanoseconds or
+ * an absolute Unadjusted System Time.  Unadjusted System Time is the number
+ * of nanoseconds since some arbitrary system event (e.g. since the last
+ * time the system booted).  The Unadjusted System Time is an unsigned
+ * 64 bit value that wraps back to 0 every 584 years.  Time intervals
+ * may be either signed or unsigned.
+ */
+typedef khronos_uint64_t       khronos_utime_nanoseconds_t;
+typedef khronos_int64_t        khronos_stime_nanoseconds_t;
+#endif
+
+/*
+ * Dummy value used to pad enum types to 32 bits.
+ */
+#ifndef KHRONOS_MAX_ENUM
+#define KHRONOS_MAX_ENUM 0x7FFFFFFF
+#endif
+
+/*
+ * Enumerated boolean type
+ *
+ * Values other than zero should be considered to be true.  Therefore
+ * comparisons should not be made against KHRONOS_TRUE.
+ */
+typedef enum {
+    KHRONOS_FALSE = 0,
+    KHRONOS_TRUE  = 1,
+    KHRONOS_BOOLEAN_ENUM_FORCE_SIZE = KHRONOS_MAX_ENUM
+} khronos_boolean_enum_t;
+
+#endif /* __khrplatform_h_ */
diff --git a/platform/winrt/include/angle_gl.h b/platform/winrt/include/angle_gl.h
index fadae33e26..d093f75ee2 100644
--- a/platform/winrt/include/angle_gl.h
+++ b/platform/winrt/include/angle_gl.h
@@ -1,23 +1,23 @@
-//

-// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved.

-// Use of this source code is governed by a BSD-style license that can be

-// found in the LICENSE file.

-//

-// angle_gl.h:

-//   Includes all necessary GL headers and definitions for ANGLE.

-//

-

-#ifndef ANGLE_GL_H_

-#define ANGLE_GL_H_

-

-#include "GLES2/gl2.h"

-#include "GLES2/gl2ext.h"

-#include "GLES3/gl3.h"

-#include "GLES3/gl3ext.h"

-

-// The following enum is used in ANGLE, but is from desktop GL

-#ifndef GL_SAMPLER_2D_RECT_ARB

-#define GL_SAMPLER_2D_RECT_ARB 0x8B63

-#endif

-

-#endif // ANGLE_GL_H_

+//
+// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+// angle_gl.h:
+//   Includes all necessary GL headers and definitions for ANGLE.
+//
+
+#ifndef ANGLE_GL_H_
+#define ANGLE_GL_H_
+
+#include "GLES2/gl2.h"
+#include "GLES2/gl2ext.h"
+#include "GLES3/gl3.h"
+#include "GLES3/gl3ext.h"
+
+// The following enum is used in ANGLE, but is from desktop GL
+#ifndef GL_SAMPLER_2D_RECT_ARB
+#define GL_SAMPLER_2D_RECT_ARB 0x8B63
+#endif
+
+#endif // ANGLE_GL_H_
diff --git a/platform/winrt/include/angle_windowsstore.h b/platform/winrt/include/angle_windowsstore.h
index fe587bf269..53ec93e037 100644
--- a/platform/winrt/include/angle_windowsstore.h
+++ b/platform/winrt/include/angle_windowsstore.h
@@ -1,37 +1,37 @@
-//

-// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved.

-// Use of this source code is governed by a BSD-style license that can be

-// found in the LICENSE file.

-//

-// angle_windowsstore.h:

-

-#ifndef ANGLE_WINDOWSSTORE_H_

-#define ANGLE_WINDOWSSTORE_H_

-

-// The following properties can be set on the CoreApplication to support additional

-// ANGLE configuration options.

-//

-// The Visual Studio sample templates provided with this version of ANGLE have examples

-// of how to set these property values.

-

-//

-// Property: EGLNativeWindowTypeProperty

-// Type: IInspectable

-// Description: Set this property to specify the window type to use for creating a surface.

-//              If this property is missing, surface creation will fail.

-//

-const wchar_t EGLNativeWindowTypeProperty[] = L"EGLNativeWindowTypeProperty";

-

-//

-// Property: EGLRenderSurfaceSizeProperty

-// Type: Size

-// Description: Set this property to specify a preferred size in pixels of the render surface.

-//              The render surface size width and height must be greater than 0.

-//              If this property is set, then the render surface size is fixed.

-//              If this property is missing, a default behavior will be provided.

-//              The default behavior uses the window size if a CoreWindow is specified or

-//              the size of the SwapChainPanel control if one is specified.

-//

-const wchar_t EGLRenderSurfaceSizeProperty[] = L"EGLRenderSurfaceSizeProperty";

-

-#endif // ANGLE_WINDOWSSTORE_H_

+//
+// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+// angle_windowsstore.h:
+
+#ifndef ANGLE_WINDOWSSTORE_H_
+#define ANGLE_WINDOWSSTORE_H_
+
+// The following properties can be set on the CoreApplication to support additional
+// ANGLE configuration options.
+//
+// The Visual Studio sample templates provided with this version of ANGLE have examples
+// of how to set these property values.
+
+//
+// Property: EGLNativeWindowTypeProperty
+// Type: IInspectable
+// Description: Set this property to specify the window type to use for creating a surface.
+//              If this property is missing, surface creation will fail.
+//
+const wchar_t EGLNativeWindowTypeProperty[] = L"EGLNativeWindowTypeProperty";
+
+//
+// Property: EGLRenderSurfaceSizeProperty
+// Type: Size
+// Description: Set this property to specify a preferred size in pixels of the render surface.
+//              The render surface size width and height must be greater than 0.
+//              If this property is set, then the render surface size is fixed.
+//              If this property is missing, a default behavior will be provided.
+//              The default behavior uses the window size if a CoreWindow is specified or
+//              the size of the SwapChainPanel control if one is specified.
+//
+const wchar_t EGLRenderSurfaceSizeProperty[] = L"EGLRenderSurfaceSizeProperty";
+
+#endif // ANGLE_WINDOWSSTORE_H_
diff --git a/platform/x11/detect.py b/platform/x11/detect.py
index 33e8fd03e2..9a52a7c92b 100644
--- a/platform/x11/detect.py
+++ b/platform/x11/detect.py
@@ -63,7 +63,7 @@ def get_flags():
 	return [
 	('builtin_zlib', 'no'),
 	("openssl", "yes"),
-	("theora","no"),
+	#("theora","no"),
         ]
 			
 
diff --git a/scene/2d/animated_sprite.cpp b/scene/2d/animated_sprite.cpp
index 0b00ac9560..342b86b4c1 100644
--- a/scene/2d/animated_sprite.cpp
+++ b/scene/2d/animated_sprite.cpp
@@ -28,6 +28,8 @@
 /*************************************************************************/
 #include "animated_sprite.h"
 #include "scene/scene_string_names.h"
+#include "os/os.h"
+
 void AnimatedSprite::edit_set_pivot(const Point2& p_pivot) {
 
 	set_offset(p_pivot);
@@ -153,6 +155,9 @@ void AnimatedSprite::_notification(int p_what) {
 			if (centered)
 				ofs-=s/2;
 
+			if (OS::get_singleton()->get_use_pixel_snap()) {
+				ofs=ofs.floor();
+			}
 			Rect2 dst_rect(ofs,s);
 
 			if (hflip)
diff --git a/scene/2d/joints_2d.cpp b/scene/2d/joints_2d.cpp
index c1e91c2ecd..1df936535f 100644
--- a/scene/2d/joints_2d.cpp
+++ b/scene/2d/joints_2d.cpp
@@ -126,7 +126,7 @@ void Joint2D::_bind_methods() {
 
 	ADD_PROPERTY( PropertyInfo( Variant::NODE_PATH, "node_a"), _SCS("set_node_a"),_SCS("get_node_a") );
 	ADD_PROPERTY( PropertyInfo( Variant::NODE_PATH, "node_b"), _SCS("set_node_b"),_SCS("get_node_b") );
-	ADD_PROPERTY( PropertyInfo( Variant::REAL, "bias/bias",PROPERTY_HINT_RANGE,"0,0.9,0.01"), _SCS("set_bias"),_SCS("get_bias") );
+	ADD_PROPERTY( PropertyInfo( Variant::REAL, "bias/bias",PROPERTY_HINT_RANGE,"0,0.9,0.001"), _SCS("set_bias"),_SCS("get_bias") );
 
 }
 
@@ -175,15 +175,37 @@ RID PinJoint2D::_configure_joint() {
 		//add a collision exception between both
 		Physics2DServer::get_singleton()->body_add_collision_exception(body_a->get_rid(),body_b->get_rid());
 	}
+	RID pj = Physics2DServer::get_singleton()->pin_joint_create(get_global_transform().get_origin(),body_a->get_rid(),body_b?body_b->get_rid():RID());
+	Physics2DServer::get_singleton()->pin_joint_set_param(pj, Physics2DServer::PIN_JOINT_SOFTNESS, softness);
+	return pj;
 
-	return Physics2DServer::get_singleton()->pin_joint_create(get_global_transform().get_origin(),body_a->get_rid(),body_b?body_b->get_rid():RID());
+}
+
+void PinJoint2D::set_softness(real_t p_softness) {
+
+	softness=p_softness;
+	update();
+	if (get_joint().is_valid())
+		Physics2DServer::get_singleton()->pin_joint_set_param(get_joint(), Physics2DServer::PIN_JOINT_SOFTNESS, p_softness);
 
 }
 
+real_t PinJoint2D::get_softness() const {
 
-PinJoint2D::PinJoint2D() {
+	return softness;
+}
+
+void PinJoint2D::_bind_methods() {
+
+	ObjectTypeDB::bind_method(_MD("set_softness","softness"), &PinJoint2D::set_softness);
+	ObjectTypeDB::bind_method(_MD("get_softness"), &PinJoint2D::get_softness);
 
+	ADD_PROPERTY( PropertyInfo( Variant::REAL, "softness", PROPERTY_HINT_EXP_RANGE,"0.00,16,0.01"), _SCS("set_softness"), _SCS("get_softness"));
+}
+
+PinJoint2D::PinJoint2D() {
 
+	softness = 0;
 }
 
 
diff --git a/scene/2d/joints_2d.h b/scene/2d/joints_2d.h
index ac72c6ce59..908e3a158e 100644
--- a/scene/2d/joints_2d.h
+++ b/scene/2d/joints_2d.h
@@ -72,13 +72,17 @@ class PinJoint2D : public Joint2D {
 
 	OBJ_TYPE(PinJoint2D,Joint2D);
 
+	real_t softness;
+
 protected:
 
 	void _notification(int p_what);
 	virtual RID _configure_joint();
+	static void _bind_methods();
 public:
 
-
+	void set_softness(real_t p_stiffness);
+	real_t get_softness() const;
 
 	PinJoint2D();
 };
diff --git a/scene/2d/navigation2d.cpp b/scene/2d/navigation2d.cpp
index 5db0e0a9fc..b7d51730a0 100644
--- a/scene/2d/navigation2d.cpp
+++ b/scene/2d/navigation2d.cpp
@@ -8,8 +8,6 @@ void Navigation2D::_navpoly_link(int p_id) {
 	NavMesh &nm=navpoly_map[p_id];
 	ERR_FAIL_COND(nm.linked);
 
-	print_line("LINK");
-
 	DVector<Vector2> vertices=nm.navpoly->get_vertices();
 	int len = vertices.size();
 	if (len==0)
@@ -48,7 +46,6 @@ void Navigation2D::_navpoly_link(int p_id) {
 			e.point=_get_point(ep);
 			p.edges[j]=e;
 
-
 			int idxn = indices[(j+1)%plen];
 			if (idxn<0 || idxn>=len) {
 				valid=false;
@@ -121,7 +118,7 @@ void Navigation2D::_navpoly_unlink(int p_id) {
 	NavMesh &nm=navpoly_map[p_id];
 	ERR_FAIL_COND(!nm.linked);
 
-	print_line("UNLINK");
+	//print_line("UNLINK");
 
 	for (List<Polygon>::Element *E=nm.polygons.front();E;E=E->next()) {
 
@@ -358,7 +355,6 @@ Vector<Vector2> Navigation2D::get_simple_path(const Vector2& p_start, const Vect
 
 	if (!begin_poly || !end_poly) {
 
-		//print_line("No Path Path");
 		return Vector<Vector2>(); //no path
 	}
 
diff --git a/scene/2d/ray_cast_2d.cpp b/scene/2d/ray_cast_2d.cpp
index 05594fd79c..acc4c620e6 100644
--- a/scene/2d/ray_cast_2d.cpp
+++ b/scene/2d/ray_cast_2d.cpp
@@ -53,6 +53,16 @@ uint32_t RayCast2D::get_layer_mask() const {
 	return layer_mask;
 }
 
+void RayCast2D::set_type_mask(uint32_t p_mask) {
+
+	type_mask=p_mask;
+}
+
+uint32_t RayCast2D::get_type_mask() const {
+
+	return type_mask;
+}
+
 bool RayCast2D::is_colliding() const{
 
 	return collided;
@@ -162,7 +172,7 @@ void RayCast2D::_notification(int p_what) {
 
 			Physics2DDirectSpaceState::RayResult rr;
 
-			if (dss->intersect_ray(gt.get_origin(),gt.xform(to),rr,exclude,layer_mask)) {
+			if (dss->intersect_ray(gt.get_origin(),gt.xform(to),rr,exclude,layer_mask,type_mask)) {
 
 				collided=true;
 				against=rr.collider_id;
@@ -241,9 +251,13 @@ void RayCast2D::_bind_methods() {
 	ObjectTypeDB::bind_method(_MD("set_layer_mask","mask"),&RayCast2D::set_layer_mask);
 	ObjectTypeDB::bind_method(_MD("get_layer_mask"),&RayCast2D::get_layer_mask);
 
+	ObjectTypeDB::bind_method(_MD("set_type_mask","mask"),&RayCast2D::set_type_mask);
+	ObjectTypeDB::bind_method(_MD("get_type_mask"),&RayCast2D::get_type_mask);
+
 	ADD_PROPERTY(PropertyInfo(Variant::BOOL,"enabled"),_SCS("set_enabled"),_SCS("is_enabled"));
 	ADD_PROPERTY(PropertyInfo(Variant::VECTOR2,"cast_to"),_SCS("set_cast_to"),_SCS("get_cast_to"));
 	ADD_PROPERTY(PropertyInfo(Variant::INT,"layer_mask",PROPERTY_HINT_ALL_FLAGS),_SCS("set_layer_mask"),_SCS("get_layer_mask"));
+	ADD_PROPERTY(PropertyInfo(Variant::INT,"type_mask",PROPERTY_HINT_FLAGS,"Static,Kinematic,Rigid,Character,Area"),_SCS("set_type_mask"),_SCS("get_type_mask"));
 }
 
 RayCast2D::RayCast2D() {
@@ -253,5 +267,6 @@ RayCast2D::RayCast2D() {
 	collided=false;
 	against_shape=0;
 	layer_mask=1;
+	type_mask=Physics2DDirectSpaceState::TYPE_MASK_COLLISION;
 	cast_to=Vector2(0,50);
 }
diff --git a/scene/2d/ray_cast_2d.h b/scene/2d/ray_cast_2d.h
index c7616be523..8c3ce8b3b3 100644
--- a/scene/2d/ray_cast_2d.h
+++ b/scene/2d/ray_cast_2d.h
@@ -44,6 +44,7 @@ class RayCast2D : public Node2D {
 	Vector2 collision_normal;
 	Set<RID> exclude;
 	uint32_t layer_mask;
+	uint32_t type_mask;
 
 
 	Vector2 cast_to;
@@ -62,6 +63,9 @@ public:
 	void set_layer_mask(uint32_t p_mask);
 	uint32_t get_layer_mask() const;
 
+	void set_type_mask(uint32_t p_mask);
+	uint32_t get_type_mask() const;
+
 	bool is_colliding() const;
 	Object *get_collider() const;
 	int get_collider_shape() const;
diff --git a/scene/2d/sprite.cpp b/scene/2d/sprite.cpp
index 87697fc073..89d9966958 100644
--- a/scene/2d/sprite.cpp
+++ b/scene/2d/sprite.cpp
@@ -30,6 +30,7 @@
 #include "core/core_string_names.h"
 #include "scene/scene_string_names.h"
 #include "scene/main/viewport.h"
+#include "os/os.h"
 
 void Sprite::edit_set_pivot(const Point2& p_pivot) {
 
@@ -85,6 +86,9 @@ void Sprite::_notification(int p_what) {
 			Point2 ofs=offset;
 			if (centered)
 				ofs-=s/2;
+			if (OS::get_singleton()->get_use_pixel_snap()) {
+				ofs=ofs.floor();
+			}
 
 			Rect2 dst_rect(ofs,s);
 
@@ -422,6 +426,9 @@ void ViewportSprite::_notification(int p_what) {
 			if (centered)
 				ofs-=s/2;
 
+			if (OS::get_singleton()->get_use_pixel_snap()) {
+				ofs=ofs.floor();
+			}
 			Rect2 dst_rect(ofs,s);
 			texture->draw_rect_region(ci,dst_rect,src_rect,modulate);
 
diff --git a/scene/3d/navigation.cpp b/scene/3d/navigation.cpp
index bfa8add09c..ce28350be0 100644
--- a/scene/3d/navigation.cpp
+++ b/scene/3d/navigation.cpp
@@ -295,12 +295,14 @@ Vector<Vector3> Navigation::get_simple_path(const Vector3& p_start, const Vector
 		}
 	}
 
+
 	if (!begin_poly || !end_poly) {
 
 		//print_line("No Path Path");
 		return Vector<Vector3>(); //no path
 	}
 
+
 	if (begin_poly==end_poly) {
 
 		Vector<Vector3> path;
diff --git a/scene/audio/stream_player.cpp b/scene/audio/stream_player.cpp
index c4e1ccc9b6..0ee9d76611 100644
--- a/scene/audio/stream_player.cpp
+++ b/scene/audio/stream_player.cpp
@@ -67,7 +67,7 @@ bool StreamPlayer::sp_mix(int32_t *p_buffer,int p_frames) {
 
 void StreamPlayer::sp_update() {
 
-	_THREAD_SAFE_METHOD_
+	//_THREAD_SAFE_METHOD_
 	if (!paused && resampler.is_ready() && playback.is_valid()) {
 
 		if (!playback->is_playing()) {
@@ -144,7 +144,7 @@ void StreamPlayer::play(float p_from_offset) {
 	if (playback->is_playing())
 		stop();
 
-	_THREAD_SAFE_METHOD_
+	//_THREAD_SAFE_METHOD_
 	playback->play(p_from_offset);
 	//feed the ringbuffer as long as no update callback is going on
 	sp_update();
@@ -162,7 +162,7 @@ void StreamPlayer::stop() {
 	if (playback.is_null())
 		return;
 
-	_THREAD_SAFE_METHOD_
+	//_THREAD_SAFE_METHOD_
 	AudioServer::get_singleton()->stream_set_active(stream_rid,false);
 	playback->stop();
 	//set_idle_process(false);
diff --git a/scene/audio/stream_player.h b/scene/audio/stream_player.h
index b5aa943067..be090f50e1 100644
--- a/scene/audio/stream_player.h
+++ b/scene/audio/stream_player.h
@@ -37,7 +37,7 @@ class StreamPlayer : public Node {
 
 	OBJ_TYPE(StreamPlayer,Node);
 
-	_THREAD_SAFE_CLASS_
+	//_THREAD_SAFE_CLASS_
 
 	struct InternalStream : public AudioServer::AudioStream {
 		StreamPlayer *player;
diff --git a/scene/gui/video_player.cpp b/scene/gui/video_player.cpp
index c0b971cb33..f50552b32c 100644
--- a/scene/gui/video_player.cpp
+++ b/scene/gui/video_player.cpp
@@ -28,6 +28,88 @@
 /*************************************************************************/
 #include "video_player.h"
 
+
+
+int VideoPlayer::InternalStream::get_channel_count() const {
+
+	return player->sp_get_channel_count();
+}
+void VideoPlayer::InternalStream::set_mix_rate(int p_rate){
+
+	return player->sp_set_mix_rate(p_rate);
+}
+bool VideoPlayer::InternalStream::mix(int32_t *p_buffer,int p_frames){
+
+	return player->sp_mix(p_buffer,p_frames);
+}
+void VideoPlayer::InternalStream::update(){
+
+	player->sp_update();
+}
+
+
+int VideoPlayer::sp_get_channel_count() const {
+
+	return playback->get_channels();
+}
+
+void VideoPlayer::sp_set_mix_rate(int p_rate){
+
+	server_mix_rate=p_rate;
+}
+
+bool VideoPlayer::sp_mix(int32_t *p_buffer,int p_frames) {
+
+	if (resampler.is_ready()) {
+		return resampler.mix(p_buffer,p_frames);
+	}
+
+	return false;
+}
+
+void VideoPlayer::sp_update() {
+#if 0
+	_THREAD_SAFE_METHOD_
+	//update is unused
+	if (!paused && playback.is_valid()) {
+
+		if (!playback->is_playing()) {
+			//stream depleted data, but there's still audio in the ringbuffer
+			//check that all this audio has been flushed before stopping the stream
+			int to_mix = resampler.get_total() - resampler.get_todo();
+			if (to_mix==0) {
+				stop();
+				return;
+			}
+
+			return;
+		}
+
+		int todo =resampler.get_todo();
+		int wrote = playback->mix(resampler.get_write_buffer(),todo);
+		resampler.write(wrote);
+	}
+#endif
+}
+
+int VideoPlayer::_audio_mix_callback(void* p_udata,const int16_t *p_data,int p_frames) {
+
+	VideoPlayer *vp=(VideoPlayer*)p_udata;
+
+	int todo=MIN(vp->resampler.get_todo(),p_frames);
+
+	int16_t *wb = vp->resampler.get_write_buffer();
+	int c = vp->resampler.get_channel_count();
+
+	for(int i=0;i<todo*c;i++) {
+		wb[i]=p_data[i];
+	}
+	vp->resampler.write(todo);
+	return todo;
+}
+
+
+
 void VideoPlayer::_notification(int p_notification) {
 
 	switch (p_notification) {
@@ -45,16 +127,25 @@ void VideoPlayer::_notification(int p_notification) {
 				return;
 			if (paused)
 				return;
-			if (!stream->is_playing())
+			if (!playback->is_playing())
 				return;
 
-			stream->update(get_tree()->get_idle_process_time());
-			int prev_width = texture->get_width();
+			double audio_time = AudioServer::get_singleton()->get_mix_time();
+
+			double delta = last_audio_time==0?0:audio_time-last_audio_time;
+			last_audio_time=audio_time;
+			if (delta==0)
+				return;
+
+
+			playback->update(delta);
+
+			/*int prev_width = texture->get_width();
 			stream->pop_frame(texture);
 			if (prev_width == 0) {
 				update();
 				minimum_size_changed();
-			};
+			};*/
 
 		} break;
 
@@ -75,6 +166,9 @@ void VideoPlayer::_notification(int p_notification) {
 
 };
 
+
+
+
 Size2 VideoPlayer::get_minimum_size() const {
 
 	if (!expand && !texture.is_null())
@@ -100,15 +194,33 @@ void VideoPlayer::set_stream(const Ref<VideoStream> &p_stream) {
 
 	stop();
 
-	texture = Ref<ImageTexture>(memnew(ImageTexture));
-
 	stream=p_stream;
-	if (!stream.is_null()) {
-
-		stream->set_loop(loops);
-		stream->set_paused(paused);
+    if (stream.is_valid()) {
+        stream->set_audio_track(audio_track);
+        playback=stream->instance_playback();
+    } else {
+        playback=Ref<VideoStreamPlayback>();
+    }
+
+	if (!playback.is_null()) {
+		playback->set_loop(loops);
+		playback->set_paused(paused);
+		texture=playback->get_texture();
+
+		AudioServer::get_singleton()->lock();
+		resampler.setup(playback->get_channels(),playback->get_mix_rate(),server_mix_rate,buffering_ms,0);
+		AudioServer::get_singleton()->unlock();
+		playback->set_mix_callback(_audio_mix_callback,this);
+
+	} else {
+		texture.unref();
+		AudioServer::get_singleton()->lock();
+		resampler.clear();
+		AudioServer::get_singleton()->unlock();
 	}
 
+	update();
+
 };
 
 Ref<VideoStream> VideoPlayer::get_stream() const {
@@ -119,36 +231,41 @@ Ref<VideoStream> VideoPlayer::get_stream() const {
 void VideoPlayer::play() {
 
 	ERR_FAIL_COND(!is_inside_tree());
-	if (stream.is_null())
+	if (playback.is_null())
 		return;
-	stream->play();
+	playback->stop();
+	playback->play();
 	set_process(true);
+	AudioServer::get_singleton()->stream_set_active(stream_rid,true);
+	AudioServer::get_singleton()->stream_set_volume_scale(stream_rid,volume);
+	last_audio_time=0;
 };
 
 void VideoPlayer::stop() {
 
 	if (!is_inside_tree())
 		return;
-	if (stream.is_null())
+	if (playback.is_null())
 		return;
 
-	stream->stop();
+	playback->stop();
 	set_process(false);
+	last_audio_time=0;
 };
 
 bool VideoPlayer::is_playing() const {
 
-	if (stream.is_null())
+	if (playback.is_null())
 		return false;
 
-	return stream->is_playing();
+	return playback->is_playing();
 };
 
 void VideoPlayer::set_paused(bool p_paused) {
 
 	paused=p_paused;
-	if (stream.is_valid()) {
-		stream->set_paused(p_paused);
+	if (playback.is_valid()) {
+		playback->set_paused(p_paused);
 		set_process(!p_paused);
 	};
 };
@@ -156,7 +273,27 @@ void VideoPlayer::set_paused(bool p_paused) {
 bool VideoPlayer::is_paused() const {
 
 	return paused;
-};
+}
+
+void VideoPlayer::set_buffering_msec(int p_msec) {
+
+	buffering_ms=p_msec;
+}
+
+int VideoPlayer::get_buffering_msec() const{
+
+	return buffering_ms;
+}
+
+void VideoPlayer::set_audio_track(int p_track) {
+    audio_track=p_track;
+}
+
+int VideoPlayer::get_audio_track() const {
+
+    return audio_track;
+}
+
 
 void VideoPlayer::set_volume(float p_vol) {
 
@@ -194,9 +331,9 @@ String VideoPlayer::get_stream_name() const {
 
 float VideoPlayer::get_stream_pos() const {
 
-	if (stream.is_null())
+	if (playback.is_null())
 		return 0;
-	return stream->get_pos();
+	return playback->get_pos();
 };
 
 
@@ -229,6 +366,9 @@ void VideoPlayer::_bind_methods() {
 	ObjectTypeDB::bind_method(_MD("set_volume_db","db"),&VideoPlayer::set_volume_db);
 	ObjectTypeDB::bind_method(_MD("get_volume_db"),&VideoPlayer::get_volume_db);
 
+    ObjectTypeDB::bind_method(_MD("set_audio_track","track"),&VideoPlayer::set_audio_track);
+    ObjectTypeDB::bind_method(_MD("get_audio_track"),&VideoPlayer::get_audio_track);
+
 	ObjectTypeDB::bind_method(_MD("get_stream_name"),&VideoPlayer::get_stream_name);
 
 	ObjectTypeDB::bind_method(_MD("get_stream_pos"),&VideoPlayer::get_stream_pos);
@@ -239,13 +379,16 @@ void VideoPlayer::_bind_methods() {
 	ObjectTypeDB::bind_method(_MD("set_expand","enable"), &VideoPlayer::set_expand );
 	ObjectTypeDB::bind_method(_MD("has_expand"), &VideoPlayer::has_expand );
 
+	ObjectTypeDB::bind_method(_MD("set_buffering_msec","msec"),&VideoPlayer::set_buffering_msec);
+	ObjectTypeDB::bind_method(_MD("get_buffering_msec"),&VideoPlayer::get_buffering_msec);
 
 	ADD_PROPERTY( PropertyInfo(Variant::OBJECT, "stream/stream", PROPERTY_HINT_RESOURCE_TYPE,"VideoStream"), _SCS("set_stream"), _SCS("get_stream") );
 //	ADD_PROPERTY( PropertyInfo(Variant::BOOL, "stream/loop"), _SCS("set_loop"), _SCS("has_loop") );
 	ADD_PROPERTY( PropertyInfo(Variant::REAL, "stream/volume_db", PROPERTY_HINT_RANGE,"-80,24,0.01"), _SCS("set_volume_db"), _SCS("get_volume_db") );
 	ADD_PROPERTY( PropertyInfo(Variant::BOOL, "stream/autoplay"), _SCS("set_autoplay"), _SCS("has_autoplay") );
 	ADD_PROPERTY( PropertyInfo(Variant::BOOL, "stream/paused"), _SCS("set_paused"), _SCS("is_paused") );
-	ADD_PROPERTY( PropertyInfo( Variant::BOOL, "expand" ), _SCS("set_expand"),_SCS("has_expand") );
+    ADD_PROPERTY( PropertyInfo(Variant::INT, "stream/audio_track",PROPERTY_HINT_RANGE,"0,128,1"), _SCS("set_audio_track"), _SCS("get_audio_track") );
+    ADD_PROPERTY( PropertyInfo( Variant::BOOL, "expand" ), _SCS("set_expand"),_SCS("has_expand") );
 }
 
 
@@ -257,6 +400,16 @@ VideoPlayer::VideoPlayer() {
 	autoplay = false;
 	expand = true;
 	loops = false;
+
+    audio_track=0;
+
+	buffering_ms=500;
+	server_mix_rate=44100;
+
+	internal_stream.player=this;
+	stream_rid=AudioServer::get_singleton()->audio_stream_create(&internal_stream);
+	last_audio_time=0;
+
 };
 
 VideoPlayer::~VideoPlayer() {
diff --git a/scene/gui/video_player.h b/scene/gui/video_player.h
index 2b850ca509..c485e3d6b6 100644
--- a/scene/gui/video_player.h
+++ b/scene/gui/video_player.h
@@ -31,22 +31,50 @@
 
 #include "scene/resources/video_stream.h"
 #include "scene/gui/control.h"
+#include "servers/audio/audio_rb_resampler.h"
 
 class VideoPlayer : public Control {
 
 	OBJ_TYPE(VideoPlayer,Control);
 
+	struct InternalStream : public AudioServer::AudioStream {
+		VideoPlayer *player;
+		virtual int get_channel_count() const;
+		virtual void set_mix_rate(int p_rate); //notify the stream of the mix rate
+		virtual bool mix(int32_t *p_buffer,int p_frames);
+		virtual void update();
+	};
+
+
+	InternalStream internal_stream;
+	Ref<VideoStreamPlayback> playback;
 	Ref<VideoStream> stream;
+
+	int sp_get_channel_count() const;
+	void sp_set_mix_rate(int p_rate); //notify the stream of the mix rate
+	bool sp_mix(int32_t *p_buffer,int p_frames);
+	void sp_update();
+
+
 	RID stream_rid;
 
 	Ref<ImageTexture> texture;
 	Image last_frame;
 
+	AudioRBResampler resampler;
+
 	bool paused;
 	bool autoplay;
 	float volume;
+	double last_audio_time;
 	bool expand;
 	bool loops;
+	int buffering_ms;
+    int server_mix_rate;
+    int audio_track;
+
+	static int _audio_mix_callback(void* p_udata,const int16_t *p_data,int p_frames);
+
 
 protected:
 
@@ -82,6 +110,12 @@ public:
 	void set_autoplay(bool p_vol);
 	bool has_autoplay() const;
 
+    void set_audio_track(int p_track);
+    int get_audio_track() const;
+
+	void set_buffering_msec(int p_msec);
+	int get_buffering_msec() const;
+
 	VideoPlayer();
 	~VideoPlayer();
 };
diff --git a/scene/io/resource_format_wav.cpp b/scene/io/resource_format_wav.cpp
index 090348c933..00b800b28b 100644
--- a/scene/io/resource_format_wav.cpp
+++ b/scene/io/resource_format_wav.cpp
@@ -252,6 +252,7 @@ RES ResourceFormatLoaderWAV::load(const String &p_path, const String& p_original
 	if (r_error)
 		*r_error=OK;
 
+
 	return sample;
 
 }
diff --git a/scene/main/instance_placeholder.cpp b/scene/main/instance_placeholder.cpp
new file mode 100644
index 0000000000..370eb1e74a
--- /dev/null
+++ b/scene/main/instance_placeholder.cpp
@@ -0,0 +1,74 @@
+#include "instance_placeholder.h"
+
+#include "scene/resources/packed_scene.h"
+#include "io/resource_loader.h"
+
+bool InstancePlaceholder::_set(const StringName& p_name, const Variant& p_value) {
+
+	PropSet ps;
+	ps.name=p_name;
+	ps.value=p_value;
+	stored_values.push_back(ps);
+	return true;
+}
+
+bool InstancePlaceholder::_get(const StringName& p_name,Variant &r_ret) const{
+
+	return false;
+}
+void InstancePlaceholder::_get_property_list( List<PropertyInfo> *p_list) const{
+
+
+}
+
+
+void InstancePlaceholder::set_path(const String& p_name) {
+
+	path=p_name;
+}
+
+String InstancePlaceholder::get_path() const {
+
+	return path;
+}
+void InstancePlaceholder::replace_by_instance(const Ref<PackedScene> &p_custom_scene){
+
+	ERR_FAIL_COND(!is_inside_tree());
+
+	Node *base = get_parent();
+	if (!base)
+		return;
+
+	Ref<PackedScene> ps;
+	if (p_custom_scene.is_valid())
+		ps = p_custom_scene;
+	else
+		ps = ResourceLoader::load(path,"PackedScene");
+
+	if (!ps.is_valid())
+		return;
+	Node *scene = ps->instance();
+	scene->set_name(get_name());
+	int pos = get_position_in_parent();
+
+	for(List<PropSet>::Element *E=stored_values.front();E;E=E->next()) {
+		scene->set(E->get().name,E->get().value);
+	}
+
+	queue_delete();
+
+	base->remove_child(this);
+	base->add_child(scene);
+	base->move_child(scene,pos);
+
+}
+
+void InstancePlaceholder::_bind_methods() {
+
+	ObjectTypeDB::bind_method(_MD("replace_by_instance","custom_scene:PackedScene"),&InstancePlaceholder::replace_by_instance,DEFVAL(Variant()));
+}
+
+InstancePlaceholder::InstancePlaceholder() {
+
+
+}
diff --git a/scene/main/instance_placeholder.h b/scene/main/instance_placeholder.h
new file mode 100644
index 0000000000..e9e76e7a2d
--- /dev/null
+++ b/scene/main/instance_placeholder.h
@@ -0,0 +1,37 @@
+#ifndef INSTANCE_PLACEHOLDER_H
+#define INSTANCE_PLACEHOLDER_H
+
+#include "scene/main/node.h"
+
+class PackedScene;
+
+class InstancePlaceholder : public Node {
+
+	OBJ_TYPE(InstancePlaceholder,Node);
+
+	String path;
+	struct PropSet {
+		StringName name;
+		Variant value;
+	};
+
+	List<PropSet> stored_values;
+
+protected:
+	bool _set(const StringName& p_name, const Variant& p_value);
+	bool _get(const StringName& p_name,Variant &r_ret) const;
+	void _get_property_list( List<PropertyInfo> *p_list) const;
+
+	static void _bind_methods();
+
+public:
+
+	void set_path(const String& p_name);
+	String get_path() const;
+
+	void replace_by_instance(const Ref<PackedScene>& p_custom_scene=Ref<PackedScene>());
+
+	InstancePlaceholder();
+};
+
+#endif // INSTANCE_PLACEHOLDER_H
diff --git a/scene/main/node.cpp b/scene/main/node.cpp
index 8336ce35f6..dd4587c19f 100644
--- a/scene/main/node.cpp
+++ b/scene/main/node.cpp
@@ -841,6 +841,20 @@ Node *Node::get_child(int p_index) const {
 	return data.children[p_index];
 }
 
+
+Node *Node::_get_child_by_name(const StringName& p_name) const {
+
+	int cc=data.children.size();
+	Node* const* cd=data.children.ptr();
+
+	for(int i=0;i<cc;i++){
+		if (cd[i]->data.name==p_name)
+			return cd[i];
+	}
+
+	return NULL;
+}
+
 Node *Node::_get_node(const NodePath& p_path) const {
 
 	ERR_FAIL_COND_V( !data.inside_tree && p_path.is_absolute(), NULL );
@@ -906,8 +920,10 @@ Node *Node::_get_node(const NodePath& p_path) const {
 Node *Node::get_node(const NodePath& p_path) const {
 
 	Node *node = _get_node(p_path);
-	ERR_EXPLAIN("Node not found: "+p_path);
-	ERR_FAIL_COND_V(!node,NULL);
+	if (!node) {
+		ERR_EXPLAIN("Node not found: "+p_path);
+		ERR_FAIL_COND_V(!node,NULL);
+	}
 	return node;
 }
 
@@ -1036,6 +1052,7 @@ void Node::get_owned_by(Node *p_by,List<Node*> *p_owned) {
 
 void Node::_set_owner_nocheck(Node* p_owner) {
 
+	ERR_FAIL_COND(data.owner);
 	data.owner=p_owner;
 	data.owner->data.owned.push_back( this );
 	data.OW = data.owner->data.owned.back();
@@ -1332,7 +1349,29 @@ String Node::get_filename() const {
 	return data.filename;
 }
 
+void Node::set_editable_instance(Node* p_node,bool p_editable) {
+
+	ERR_FAIL_NULL(p_node);
+	ERR_FAIL_COND(!is_a_parent_of(p_node));
+	NodePath p = get_path_to(p_node);
+	if (!p_editable)
+		data.editable_instances.erase(p);
+	else
+		data.editable_instances[p]=true;
+
+}
+
+bool Node::is_editable_instance(Node *p_node) const {
+
+	if (!p_node)
+		return false; //easier, null is never editable :)
+	ERR_FAIL_COND_V(!is_a_parent_of(p_node),false);
+	NodePath p = get_path_to(p_node);
+	return data.editable_instances.has(p);
+}
+
 
+#if 0
 
 void Node::generate_instance_state() {
 
@@ -1383,13 +1422,36 @@ Dictionary Node::get_instance_state() const {
 	return data.instance_state;
 }
 
-Vector<StringName> Node::get_instance_groups() const {
+#endif
+
+void Node::set_scene_instance_state(const Ref<SceneState>& p_state) {
+
+	data.instance_state=p_state;
+}
+
+Ref<SceneState> Node::get_scene_instance_state() const{
+
+	return data.instance_state;
+}
+
+void Node::set_scene_inherited_state(const Ref<SceneState>& p_state) {
+
+	data.inherited_state=p_state;
+}
+
+Ref<SceneState> Node::get_scene_inherited_state() const{
 
-	return data.instance_groups;
+	return data.inherited_state;
 }
-Vector<Node::Connection> Node::get_instance_connections() const{
 
-	return data.instance_connections;
+void Node::set_scene_instance_load_placeholder(bool p_enable) {
+
+	data.use_placeholder=p_enable;
+}
+
+bool Node::get_scene_instance_load_placeholder() const{
+
+	return data.use_placeholder;
 }
 
 int Node::get_position_in_parent() const {
@@ -2049,6 +2111,7 @@ Node::Node() {
 	data.parent_owned=false;
 	data.in_constructor=true;
 	data.viewport=NULL;
+	data.use_placeholder=false;
 }
 
 Node::~Node() {
@@ -2065,3 +2128,4 @@ Node::~Node() {
 }
 
 
+////////////////////////////////
diff --git a/scene/main/node.h b/scene/main/node.h
index a6d5bfbd9f..87fa4dd6ca 100644
--- a/scene/main/node.h
+++ b/scene/main/node.h
@@ -38,6 +38,7 @@
 
 
 class Viewport;
+class SceneState;
 class Node : public Object {
 
 	OBJ_TYPE( Node, Object );
@@ -69,9 +70,10 @@ private:
 	struct Data {
 	
 		String filename;
-		Dictionary instance_state;
-		Vector<StringName> instance_groups;
-		Vector<Connection> instance_connections;
+		Ref<SceneState> instance_state;
+		Ref<SceneState> inherited_state;
+
+		HashMap<NodePath,int> editable_instances;
 
 		Node *parent;
 		Node *owner;
@@ -96,6 +98,7 @@ private:
 		PauseMode pause_mode;
 		Node *pause_owner;
 		// variables used to properly sort the node when processing, ignored otherwise
+		//should move all the stuff below to bits
 		bool fixed_process;
 		bool idle_process;
 
@@ -105,6 +108,9 @@ private:
 
 		bool parent_owned;
 		bool in_constructor;
+		bool use_placeholder;
+
+
 	} data;
 	
 
@@ -112,6 +118,7 @@ private:
 	
 	virtual bool _use_builtin_script() const { return true; }
 	Node *_get_node(const NodePath& p_path) const;
+	Node *_get_child_by_name(const StringName& p_name) const;
 
 
 
@@ -151,7 +158,7 @@ protected:
 	
 	static void _bind_methods();
 
-friend class PackedScene;
+friend class SceneState;
 
 	void _add_child_nocheck(Node* p_child,const StringName& p_name);
 	void _set_owner_nocheck(Node* p_owner);
@@ -208,7 +215,7 @@ public:
 	
 	struct GroupInfo {
 	
-		String name;
+		StringName name;
 		bool persistent;
 	};
 	
@@ -229,7 +236,11 @@ public:
 	
 	void set_filename(const String& p_filename);
 	String get_filename() const;
-	
+
+	void set_editable_instance(Node* p_node,bool p_editable);
+	bool is_editable_instance(Node* p_node) const;
+
+
 	/* NOTIFICATIONS */
 	
 	void propagate_notification(int p_notification);
@@ -261,10 +272,14 @@ public:
 	//Node *clone_tree() const;
 
 	// used by editors, to save what has changed only
-	void generate_instance_state();
-	Dictionary get_instance_state() const;
-	Vector<StringName> get_instance_groups() const;
-	Vector<Connection> get_instance_connections() const;
+	void set_scene_instance_state(const Ref<SceneState>& p_state);
+	Ref<SceneState> get_scene_instance_state() const;
+
+	void set_scene_inherited_state(const Ref<SceneState>& p_state);
+	Ref<SceneState> get_scene_inherited_state() const;
+
+	void set_scene_instance_load_placeholder(bool p_enable);
+	bool get_scene_instance_load_placeholder() const;
 
 	static Vector<Variant> make_binds(VARIANT_ARG_LIST);
 
@@ -307,6 +322,4 @@ public:
 typedef Set<Node*,Node::Comparator> NodeSet;
 
 
-
-
 #endif
diff --git a/scene/register_scene_types.cpp b/scene/register_scene_types.cpp
index 851de4a89f..fe6b192d78 100644
--- a/scene/register_scene_types.cpp
+++ b/scene/register_scene_types.cpp
@@ -36,6 +36,7 @@
 #include "resources/default_theme/default_theme.h"
 #include "object_type_db.h"
 #include "scene/main/canvas_layer.h"
+#include "scene/main/instance_placeholder.h"
 #include "scene/main/viewport.h"
 #include "scene/gui/control.h"
 #include "scene/gui/texture_progress.h"
@@ -216,7 +217,7 @@
 #include "scene/3d/collision_polygon.h"
 #endif
 
-
+#include "scene/resources/scene_format_text.h"
 
 static ResourceFormatLoaderImage *resource_loader_image=NULL;
 static ResourceFormatLoaderWAV *resource_loader_wav=NULL;
@@ -229,6 +230,8 @@ static ResourceFormatLoaderBitMap *resource_loader_bitmap=NULL;
 static ResourceFormatLoaderTheme *resource_loader_theme=NULL;
 static ResourceFormatLoaderShader *resource_loader_shader=NULL;
 
+static ResourceFormatSaverText *resource_saver_text=NULL;
+
 //static SceneStringNames *string_names;
 
 void register_scene_types() {
@@ -267,6 +270,7 @@ void register_scene_types() {
 	ObjectTypeDB::register_type<Object>();
 
 	ObjectTypeDB::register_type<Node>();
+	ObjectTypeDB::register_virtual_type<InstancePlaceholder>();
 
 	ObjectTypeDB::register_type<Viewport>();
 	ObjectTypeDB::register_virtual_type<RenderTargetTexture>();
@@ -612,6 +616,9 @@ void register_scene_types() {
 	OS::get_singleton()->yield(); //may take time to init
 
 
+	resource_saver_text = memnew( ResourceFormatSaverText );
+	ResourceSaver::add_resource_format_saver(resource_saver_text);
+
 }
 
 void unregister_scene_types() {
@@ -629,5 +636,9 @@ void unregister_scene_types() {
 
 	memdelete( resource_loader_theme );
 	memdelete( resource_loader_shader );
+
+	if (resource_saver_text) {
+		memdelete(resource_saver_text);
+	}
 	SceneStringNames::free();
 }
diff --git a/scene/resources/packed_scene.cpp b/scene/resources/packed_scene.cpp
index b6082c3a76..a9dc0e8bfb 100644
--- a/scene/resources/packed_scene.cpp
+++ b/scene/resources/packed_scene.cpp
@@ -32,13 +32,30 @@
 #include "scene/3d/spatial.h"
 #include "scene/gui/control.h"
 #include "scene/2d/node_2d.h"
+#include "scene/main/instance_placeholder.h"
 
-bool PackedScene::can_instance() const {
+#define PACK_VERSION 2
+
+bool SceneState::can_instance() const {
 
 	return nodes.size()>0;
 }
 
-Node *PackedScene::instance(bool p_gen_edit_state) const {
+
+Node *SceneState::instance(bool p_gen_edit_state) const {
+
+	// nodes where instancing failed (because something is missing)
+	List<Node*> stray_instances;
+
+#define NODE_FROM_ID(p_name,p_id)\
+	Node *p_name;\
+	if (p_id&FLAG_ID_IS_PATH) {\
+		NodePath np=node_paths[p_id&FLAG_MASK];\
+		p_name=ret_nodes[0]->_get_node(np);\
+	} else {\
+		ERR_FAIL_INDEX_V(p_id&FLAG_MASK,nc,NULL);\
+		p_name=ret_nodes[p_id&FLAG_MASK];\
+	}
 
 	int nc = nodes.size();
 	ERR_FAIL_COND_V(nc==0,NULL);
@@ -59,29 +76,80 @@ Node *PackedScene::instance(bool p_gen_edit_state) const {
 
 	Node **ret_nodes=(Node**)alloca( sizeof(Node*)*nc );
 
+	bool gen_node_path_cache=p_gen_edit_state && node_path_cache.empty();
 
 	for(int i=0;i<nc;i++) {
 
 		const NodeData &n=nd[i];
 
-		if (!ObjectTypeDB::is_type_enabled(snames[n.type])) {
-			ret_nodes[i]=NULL;
-			continue;
+		Node *parent=NULL;
+
+		if (i>0) {
+
+			NODE_FROM_ID(nparent,n.parent);
+#ifdef DEBUG_ENABLED
+			if (!nparent && n.parent&FLAG_ID_IS_PATH) {
+
+				WARN_PRINT(String("Parent path '"+String(node_paths[n.parent&FLAG_MASK])+"' for node '"+String(snames[n.name])+"' has vanished when instancing: '"+get_path()+"'.").ascii().get_data());
+
+			}
+#endif
+			parent=nparent;
 		}
 
 		Node *node=NULL;
 
-		if (n.instance>=0) {
-			//instance existing
-			Ref<PackedScene> sdata = props[ n.instance ];
+		if (i==0 && base_scene_idx>=0) {
+			//scene inheritance on root node
+			print_line("scene inherit");
+			Ref<PackedScene> sdata = props[ base_scene_idx ];
 			ERR_FAIL_COND_V( !sdata.is_valid(), NULL);
-			node = sdata->instance();
+			node = sdata->instance(p_gen_edit_state);
 			ERR_FAIL_COND_V(!node,NULL);
-			if (p_gen_edit_state)
-				node->generate_instance_state();
+			if (p_gen_edit_state) {
+				node->set_scene_inherited_state(sdata->get_state());
+			}
 
-		} else {
-			//create anew
+		} else if (n.instance>=0) {
+			//instance a scene into this node
+			print_line("instance");
+			if (n.instance&FLAG_INSTANCE_IS_PLACEHOLDER) {
+
+				String path = props[n.instance&FLAG_MASK];
+				if (disable_placeholders) {
+
+					Ref<PackedScene> sdata = ResourceLoader::load(path,"PackedScene");
+					ERR_FAIL_COND_V( !sdata.is_valid(), NULL);
+					node = sdata->instance(p_gen_edit_state);
+					ERR_FAIL_COND_V(!node,NULL);
+				} else {
+					InstancePlaceholder *ip = memnew( InstancePlaceholder );
+					ip->set_path(path);
+					node=ip;
+				}
+				node->set_scene_instance_load_placeholder(true);
+			} else {
+				Ref<PackedScene> sdata = props[ n.instance&FLAG_MASK ];
+				ERR_FAIL_COND_V( !sdata.is_valid(), NULL);
+				node = sdata->instance(p_gen_edit_state);
+				ERR_FAIL_COND_V(!node,NULL);
+
+			}
+
+		} else if (n.type==TYPE_INSTANCED) {
+			//print_line("instanced");
+			//get the node from somewhere, it likely already exists from another instance
+			if (parent) {
+				node=parent->_get_child_by_name(snames[n.name]);
+#ifdef DEBUG_ENABLED
+				if (!node) {
+					WARN_PRINT(String("Node '"+String(ret_nodes[0]->get_path_to(parent))+"/"+String(snames[n.name])+"' was modified from inside a instance, but it has vanished.").ascii().get_data());
+				}
+#endif
+			}
+		} else if (ObjectTypeDB::is_type_enabled(snames[n.type])) {
+			print_line("created");
+			//node belongs to this scene and must be created
 			Object * obj = ObjectTypeDB::instance(snames[ n.type ]);
 			if (!obj || !obj->cast_to<Node>()) {
 				if (obj) {
@@ -109,49 +177,68 @@ Node *PackedScene::instance(bool p_gen_edit_state) const {
 		}
 
 
-		//properties
-		int nprop_count=n.properties.size();
-		if (nprop_count) {
+		if (node) {
+			// may not have found the node (part of instanced scene and removed)
+			// if found all is good, otherwise ignore
+
+			//properties
+			int nprop_count=n.properties.size();
+			if (nprop_count) {
 
-			const NodeData::Property* nprops=&n.properties[0];
+				const NodeData::Property* nprops=&n.properties[0];
 
-			for(int j=0;j<nprop_count;j++) {
+				for(int j=0;j<nprop_count;j++) {
 
-				bool valid;
-				ERR_FAIL_INDEX_V( nprops[j].name, sname_count, NULL );
-				ERR_FAIL_INDEX_V( nprops[j].value, prop_count, NULL );
+					bool valid;
+					ERR_FAIL_INDEX_V( nprops[j].name, sname_count, NULL );
+					ERR_FAIL_INDEX_V( nprops[j].value, prop_count, NULL );
 
-				node->set(snames[ nprops[j].name ],props[ nprops[j].value ],&valid);
+					node->set(snames[ nprops[j].name ],props[ nprops[j].value ],&valid);
+				}
 			}
-		}
 
-		//name
+			//name
 
-		//groups
-		for(int j=0;j<n.groups.size();j++) {
+			//groups
+			for(int j=0;j<n.groups.size();j++) {
 
-			ERR_FAIL_INDEX_V( n.groups[j], sname_count, NULL );
-			node->add_to_group( snames[ n.groups[j] ], true );
-		}
+				ERR_FAIL_INDEX_V( n.groups[j], sname_count, NULL );
+				node->add_to_group( snames[ n.groups[j] ], true );
+			}
 
+			if (n.instance>=0 || n.type!=TYPE_INSTANCED) {
+				//if node was not part of instance, must set it's name, parenthood and ownership
+				if (i>0) {
+					if (parent) {
+						parent->_add_child_nocheck(node,snames[n.name]);
+					} else {
+						//it may be possible that an instanced scene has changed
+						//and the node has nowhere to go anymore
+						stray_instances.push_back(node); //can't be added, go to stray list
+					}
+				} else {
+					node->_set_name_nocheck( snames[ n.name ] );
+				}
 
-		ret_nodes[i]=node;
+			}
 
-		if (i>0) {			
-			ERR_FAIL_INDEX_V(n.parent,i,NULL);
-			ERR_FAIL_COND_V(!ret_nodes[n.parent],NULL);
-			ret_nodes[n.parent]->_add_child_nocheck(node,snames[n.name]);
-		} else {
-			node->_set_name_nocheck( snames[ n.name ] );
-		}
+			if (n.owner>=0) {
 
+				NODE_FROM_ID(owner,n.owner);
+				if (owner)
+					node->_set_owner_nocheck(owner);
+			}
 
-		if (n.owner>=0) {
 
-			ERR_FAIL_INDEX_V(n.owner,i,NULL);
-			node->_set_owner_nocheck(ret_nodes[n.owner]);
 		}
 
+
+		ret_nodes[i]=node;
+
+		if (node && gen_node_path_cache && ret_nodes[0]) {
+			NodePath n = ret_nodes[0]->get_path_to(node);
+			node_path_cache[n]=i;
+		}
 	}
 
 
@@ -163,8 +250,14 @@ Node *PackedScene::instance(bool p_gen_edit_state) const {
 	for(int i=0;i<cc;i++) {
 
 		const ConnectionData &c=cdata[i];
-		ERR_FAIL_INDEX_V( c.from, nc, NULL );
-		ERR_FAIL_INDEX_V( c.to, nc, NULL );
+		//ERR_FAIL_INDEX_V( c.from, nc, NULL );
+		//ERR_FAIL_INDEX_V( c.to, nc, NULL );
+
+		NODE_FROM_ID(cfrom,c.from);
+		NODE_FROM_ID(cto,c.to);
+
+		if (!cfrom || !cto)
+			continue;
 
 		Vector<Variant> binds;
 		if (c.binds.size()) {
@@ -173,17 +266,25 @@ Node *PackedScene::instance(bool p_gen_edit_state) const {
 				binds[j]=props[ c.binds[j] ];
 		}
 
-		if (!ret_nodes[c.from] || !ret_nodes[c.to])
-			continue;
-		ret_nodes[c.from]->connect( snames[ c.signal], ret_nodes[ c.to ], snames[ c.method], binds,CONNECT_PERSIST|c.flags );
+
+		cfrom->connect( snames[ c.signal], cto, snames[ c.method], binds,CONNECT_PERSIST|c.flags );
 	}
 
-	Node *s = ret_nodes[0];
+	//Node *s = ret_nodes[0];
 
-	if (get_path()!="" && get_path().find("::")==-1)
-		s->set_filename(get_path());
+	//remove nodes that could not be added, likely as a result that
+	while(stray_instances.size()) {
+		memdelete(stray_instances.front()->get());
+		stray_instances.pop_front();;
+	}
+
+	for(int i=0;i<editable_instances.size();i++) {
+		Node *ei = ret_nodes[0]->_get_node(editable_instances[i]);
+		if (ei) {
+			ret_nodes[0]->set_editable_instance(ei,true);
+		}
+	}
 
-	s->notification(Node::NOTIFICATION_INSTANCED);
 	return ret_nodes[0];
 
 }
@@ -209,44 +310,157 @@ static int _vm_get_variant(const Variant& p_variant, HashMap<Variant,int,Variant
 	return idx;
 }
 
-Error PackedScene::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map) {
+Error SceneState::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map,Map<Node*,int> &nodepath_map) {
+
 
-	if (p_node!=p_owner && (p_node->get_owner()!=p_owner))
-		return OK; //nothing to do with this node, may either belong to another scene or be onowned
+	// this function handles all the work related to properly packing scenes, be it
+	// instanced or inherited.
+	// given the complexity of this process, an attempt will be made to properly
+	// document it. if you fail to understand something, please ask!
+
+	//discard nodes that do not belong to be processed
+	if (p_node!=p_owner && p_node->get_owner()!=p_owner && !p_owner->is_editable_instance(p_node->get_owner()))
+		return OK;
+
+	// save the child instanced scenes that are chosen as editable, so they can be restored
+	// upon load back
+	if (p_node!=p_owner && p_node->get_filename()!=String() && p_owner->is_editable_instance(p_node))
+		editable_instances.push_back(p_owner->get_path_to(p_node));
 
 	NodeData nd;
 
 	nd.name=_nm_get_string(p_node->get_name(),name_map);
-	nd.type=_nm_get_string(p_node->get_type(),name_map);
-	nd.parent=p_parent_idx;	
+	nd.instance=-1; //not instanced by default
+
+	// if this node is part of an instanced scene or sub-instanced scene
+	// we need to get the corresponding instance states.
+	// with the instance states, we can query for identical properties/groups
+	// and only save what has changed
+
+	List<PackState> pack_state_stack;
+
+	bool instanced_by_owner=true;
+
+	{
+		Node *n=p_node;
+
+		while(n) {
+
+			if (n==p_owner) {
+
+				Ref<SceneState> state = n->get_scene_inherited_state();
+				if (state.is_valid()) {
+					int node = state->find_node_by_path(n->get_path_to(p_node));
+					if (node>=0) {
+						//this one has state for this node, save
+						PackState ps;
+						ps.node=node;
+						ps.state=state;
+						pack_state_stack.push_front(ps);
+						instanced_by_owner=false;
+					}
+				}
+
+				if (p_node->get_filename()!=String() && p_node->get_owner()==p_owner && instanced_by_owner) {
+
+					if (p_node->get_scene_instance_load_placeholder()) {
+						//it's a placeholder, use the placeholder path
+						nd.instance=_vm_get_variant(p_node->get_filename(),variant_map);
+						nd.instance|=FLAG_INSTANCE_IS_PLACEHOLDER;
+					} else {
+						//must instance ourselves
+						Ref<PackedScene> instance = ResourceLoader::load(p_node->get_filename());
+						if (!instance.is_valid()) {
+							return ERR_CANT_OPEN;
+						}
 
+						nd.instance=_vm_get_variant(instance,variant_map);
+					}
+				}
+				n=NULL;
+			} else {
+				if (n->get_filename()!=String()) {
+					//is an instance
+					Ref<SceneState> state = n->get_scene_instance_state();
+					if (state.is_valid()) {
+						int node = state->find_node_by_path(n->get_path_to(p_node));
+						if (node>=0) {
+							//this one has state for this node, save
+							PackState ps;
+							ps.node=node;
+							ps.state=state;
+							pack_state_stack.push_back(ps);
+						}
+					}
+
+				}
+				n=n->get_owner();
+			}
+		}
+	}
+
+#if 0
+
+	Ref<SceneState> base_scene = p_node->get_scene_inherited_state(); //for inheritance
+	Ref<SceneState> instance_state;
+	int instance_state_node=-1;
+
+	if (base_scene.is_valid() && (p_node==p_owner || p_node->get_owner()==p_owner)) {
+		//scene inheritance in use, see if this node is actually inherited
+		NodePath path = p_owner->get_path_to(p_node);
+		instance_state_node = base_scene->find_node_by_path(path);
+		if (instance_state_node>=0) {
+			instance_state=base_scene;
+		}
+	}
 
-	Dictionary instance_state;
-	Set<StringName> instance_groups;
+	// check that this is a directly instanced scene from the scene being packed, if so
+	// this information must be saved. Of course, if using scene instancing and this node
+	// does belong to base scene, ignore.
 
+	if (instance_state.is_null() && p_node!=p_owner && p_node->get_owner()==p_owner && p_node->get_filename()!="") {
 
-	if (p_node!=p_owner && p_node->get_filename()!="") {
-		//instanced
+		//instanced, only direct sub-scnes are supported of course
 		Ref<PackedScene> instance = ResourceLoader::load(p_node->get_filename());
 		if (!instance.is_valid()) {
 			return ERR_CANT_OPEN;
 		}
 
 		nd.instance=_vm_get_variant(instance,variant_map);
-		instance_state = p_node->get_instance_state();
-		Vector<StringName> ig = p_node->get_instance_groups();
-		for(int i=0;i<ig.size();i++)
-			instance_groups.insert(ig[i]);
+
 	} else {
 
 		nd.instance=-1;
 	}
 
+	// finally, if this does not belong to scene inheritance, check
+	// if it belongs to scene instancing
+
+	if (instance_state.is_null() && p_node!=p_owner) {
+		//if not affected by scene inheritance, this may be
+		if (p_node->get_owner()==p_owner && p_node->get_filename()!=String()) {
+			instance_state=p_node->get_scene_instance_state();
+			if (instance_state.is_valid()) {
+				instance_state_node=instance_state->find_node_by_path(p_node->get_path_to(p_node));
+			}
 
-	//instance state makes sure that only changes to instance are saved
+		} else if (p_node->get_owner()!=p_owner && p_owner->is_editable_instance(p_node->get_owner())) {
+			instance_state=p_node->get_owner()->get_scene_instance_state();
+			if (instance_state.is_valid()) {
+				instance_state_node=instance_state->find_node_by_path(p_node->get_owner()->get_path_to(p_node));
+			}
+		}
+	}
+#endif
+	int subscene_prop_search_from=0;
+
+	// all setup, we then proceed to check all properties for the node
+	// and save the ones that are worth saving
 
 	List<PropertyInfo> plist;
 	p_node->get_property_list(&plist);
+
+
 	for (List<PropertyInfo>::Element *E=plist.front();E;E=E->next()) {
 
 
@@ -257,34 +471,63 @@ Error PackedScene::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<
 		String name = E->get().name;
 		Variant value = p_node->get( E->get().name );
 
-		if (nd.instance<0 && ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONZERO) && value.is_zero()) || ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONONE) && value.is_one())) {
-			continue;
-		}
 
-		print_line("PASSED!");
-		print_line("at: "+String(p_node->get_name())+"::"+name+": -  nz: "+itos(E->get().usage&PROPERTY_USAGE_STORE_IF_NONZERO)+" no: "+itos(E->get().usage&PROPERTY_USAGE_STORE_IF_NONONE));
-		print_line("value: "+String(value)+" is zero: "+itos(value.is_zero())+" is one" +itos(value.is_one()));
+		bool isdefault = ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONZERO) && value.is_zero()) || ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONONE) && value.is_one());
+
+//		if (nd.instance<0 && ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONZERO) && value.is_zero()) || ((E->get().usage & PROPERTY_USAGE_STORE_IF_NONONE) && value.is_one())) {
+//			continue;
+//		}
 
-		if (nd.instance>=0) {
-			//only save changed properties in instance
-			/*
-			  // this was commented because it would not save properties created from within script
-			  // done with _get_property_list, that are not in the original node.
-			  // if some property is not saved, check again
 
-			  if (!instance_state.has(name)) {
-				print_line("skip not in instance");
+
+		//print_line("PASSED!");
+		//print_line("at: "+String(p_node->get_name())+"::"+name+": -  nz: "+itos(E->get().usage&PROPERTY_USAGE_STORE_IF_NONZERO)+" no: "+itos(E->get().usage&PROPERTY_USAGE_STORE_IF_NONONE));
+		//print_line("value: "+String(value)+" is zero: "+itos(value.is_zero())+" is one" +itos(value.is_one()));
+
+		if (pack_state_stack.size()) {
+			// we are on part of an instanced subscene
+			// or part of instanced scene.
+			// only save what has been changed
+			// only save changed properties in instance
+
+			if (E->get().usage & PROPERTY_USAGE_NO_INSTANCE_STATE || E->get().name=="__meta__") {
+				//property has requested that no instance state is saved, sorry
+				//also, meta won't be overriden or saved
 				continue;
-			}*/
+			}
+
+			bool exists=false;
+			Variant original;
+
+			for (List<PackState>::Element *F=pack_state_stack.back();F;F=F->prev()) {
+				//check all levels of pack to see if the property exists somewhere
+				const PackState &ps=F->get();
 
-			if (E->get().usage & PROPERTY_USAGE_NO_INSTANCE_STATE) {
+				original = ps.state->get_property_value(ps.node,E->get().name,exists);
+				if (exists) {
+					break;
+				}
+			}
+
+
+			if (exists && bool(Variant::evaluate(Variant::OP_EQUAL,value,original))) {
+				//exists and did not change
 				continue;
 			}
 
-			if (instance_state.has(name) && instance_state[name]==value) {
+			if (!exists && isdefault) {
+				//does not exist in original node, but it's the default value
+				//so safe to skip too.
 				continue;
 			}
 
+
+		} else  {
+
+			if (isdefault) {
+				//it's the default value, no point in saving it
+				continue;
+			}
 		}
 
 		NodeData::Property prop;
@@ -295,6 +538,9 @@ Error PackedScene::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<
 	}
 
 
+	// save the groups this node is into
+	// discard groups that come from the original scene
+
 	List<Node::GroupInfo> groups;
 	p_node->get_groups(&groups);
 	for(List<Node::GroupInfo>::Element *E=groups.front();E;E=E->next()) {
@@ -302,27 +548,123 @@ Error PackedScene::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<
 
 		if (!gi.persistent)
 			continue;
-		if (nd.instance>=0 && instance_groups.has(gi.name))
-			continue; //group was instanced, don't add here
+//		if (instance_state_node>=0 && instance_state->is_node_in_group(instance_state_node,gi.name))
+//			continue; //group was instanced, don't add here
+
+		bool skip=false;
+		for (List<PackState>::Element *F=pack_state_stack.front();F;F=F->next()) {
+			//check all levels of pack to see if the group was added somewhere
+			const PackState &ps=F->get();
+			if (ps.state->is_node_in_group(ps.node,gi.name)) {
+				skip=true;
+				break;
+			}
+		}
+
+		if (skip)
+			continue;
 
 		nd.groups.push_back(_nm_get_string(gi.name,name_map));
 	}
 
-	if (node_map.has(p_node->get_owner()))
-		nd.owner=node_map[p_node->get_owner()];
-	else
+
+	// save the right owner
+	// for the saved scene root this is -1
+	// for nodes of the saved scene this is 0
+	// for nodes of instanced scenes this is >0
+
+	if (p_node==p_owner) {
+		//saved scene root
+		nd.owner=-1;
+	} else if (p_node->get_owner()==p_owner) {
+		//part of saved scene
+		nd.owner=0;
+	} else {
+
 		nd.owner=-1;
+#if 0
+		// this is pointless, if this was instanced by something else,
+		// the owner will already be set.
+
+		if (node_map.has(p_node->get_owner())) {
+			//maybe an existing saved node
+			nd.owner=node_map[p_node->get_owner()];
+		} else {
+			//not saved, use nodepath map
+			int sidx;
+			if (nodepath_map.has(p_node->get_owner())) {
+				sidx=nodepath_map[p_node->get_owner()];
+			} else {
+				sidx=nodepath_map.size();
+				nodepath_map[p_node->get_owner()]=sidx;
+			}
+
+			nd.owner=FLAG_ID_IS_PATH|sidx;
+
+		}
+#endif
+
+
+	}
+
+	// Save the right type. If this node was created by an instance
+	// then flag that the node should not be created but reused	
+	if (pack_state_stack.empty()) {
+		//this node is not part of an instancing process, so save the type
+		nd.type=_nm_get_string(p_node->get_type(),name_map);
+	} else {
+		// this node is part of an instanced process, so do not save the type.
+		// instead, save that it was instanced
+		nd.type=TYPE_INSTANCED;
+	}
+
+
+	// determine whether to save this node or not
+	// if this node is part of an instanced sub-scene, we can skip storing it if basically
+	// no properties changed and no groups were added to it.
+	// below condition is true for all nodes of the scene being saved, and ones in subscenes
+	// that hold changes
+
+	bool save_node = nd.properties.size() || nd.groups.size(); // some local properties or groups exist
+	save_node = save_node || p_node==p_owner; // owner is always saved
+	save_node = save_node || (p_node->get_owner()==p_owner && instanced_by_owner); //part of scene and not instanced
+
 
 	int idx = nodes.size();
-	node_map[p_node]=idx;
-	nodes.push_back(nd);
+	int parent_node=NO_PARENT_SAVED;
+
+	if (save_node) {
 
+		//don't save the node if nothing and subscene
+
+		node_map[p_node]=idx;
+
+		//ok validate parent node
+		if (p_parent_idx==NO_PARENT_SAVED) {
+
+			int sidx;
+			if (nodepath_map.has(p_node->get_parent())) {
+				sidx=nodepath_map[p_node->get_parent()];
+			} else {
+				sidx=nodepath_map.size();
+				nodepath_map[p_node->get_parent()]=sidx;
+			}
+
+			nd.parent=FLAG_ID_IS_PATH|sidx;
+		} else {
+			nd.parent=p_parent_idx;
+		}
+
+		parent_node=idx;
+		nodes.push_back(nd);
+
+	}
 
 
 	for(int i=0;i<p_node->get_child_count();i++) {
 
 		Node *c=p_node->get_child(i);
-		Error err = _parse_node(p_owner,c,idx,name_map,variant_map,node_map);
+		Error err = _parse_node(p_owner,c,parent_node,name_map,variant_map,node_map,nodepath_map);
 		if (err)
 			return err;
 	}
@@ -331,53 +673,135 @@ Error PackedScene::_parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<
 
 }
 
-Error PackedScene::_parse_connections(Node *p_owner,Node *p_node, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map) {
-
-	if (p_node!=p_owner && (p_node->get_owner()!=p_owner))
-		return OK; //nothing to do with this node, may either belong to another scene or be onowned
-
-	List<MethodInfo> signals;
-	p_node->get_signal_list(&signals);
+Error SceneState::_parse_connections(Node *p_owner,Node *p_node, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map,Map<Node*,int> &nodepath_map) {
 
-	ERR_FAIL_COND_V( !node_map.has(p_node), ERR_BUG);
-	NodeData &nd = nodes[node_map[p_node]];
-	Set<Connection> instance_connections;
+	if (p_node!=p_owner && p_node->get_owner() && p_node->get_owner()!=p_owner && !p_owner->is_editable_instance(p_node->get_owner()))
+		return OK;
 
-	if (nd.instance>=0) {
 
-		Vector<Connection> iconns = p_node->get_instance_connections();
-		for(int i=0;i<iconns.size();i++) {
+	List<MethodInfo> _signals;
+	p_node->get_signal_list(&_signals);
 
-			instance_connections.insert(iconns[i]);
-		}
-	}
+	//ERR_FAIL_COND_V( !node_map.has(p_node), ERR_BUG);
+	//NodeData &nd = nodes[node_map[p_node]];
 
 
-	for(List<MethodInfo>::Element *E=signals.front();E;E=E->next()) {
+	for(List<MethodInfo>::Element *E=_signals.front();E;E=E->next()) {
 
 		List<Node::Connection> conns;
 		p_node->get_signal_connection_list(E->get().name,&conns);
 		for(List<Node::Connection>::Element *F=conns.front();F;F=F->next()) {
 
 			const Node::Connection &c = F->get();
-			if (!(c.flags&CONNECT_PERSIST))
+
+			if (!(c.flags&CONNECT_PERSIST)) //only persistent connections get saved
 				continue;
 
-			if (nd.instance>=0 && instance_connections.has(c))
-				continue; //came from instance, don't save!
+			// only connections that originate or end into main saved scene are saved
+			// everything else is discarded
 
 			Node *n=c.target->cast_to<Node>();
-			if (!n)
+			if (!n) {
 				continue;
+			}
+
+			//source node is outside saved scene?
+			bool src_is_out = p_node!=p_owner && (p_node->get_filename()!=String() || p_node->get_owner()!=p_owner);
+			//target node is outside saved scene?
+			bool dst_is_out = n!=p_owner && (n->get_filename()!=String() || n->get_owner()!=p_owner);
 
-			if (!node_map.has(n)) {
-				WARN_PRINT("Connection to node outside scene??")
+			//if both are out, ignore connection
+			if (src_is_out && dst_is_out) {
 				continue;
 			}
 
+
+			{
+				Node *nl=p_node;
+
+				bool exists=false;
+
+				while(nl) {
+
+					if (nl==p_owner) {
+
+						Ref<SceneState> state = nl->get_scene_inherited_state();
+						if (state.is_valid()) {
+							int from_node = state->find_node_by_path(nl->get_path_to(p_node));
+							int to_node = state->find_node_by_path(nl->get_path_to(n));
+
+							if (from_node>=0 && to_node>=0) {
+								//this one has state for this node, save
+								if (state->is_connection(from_node,c.signal,to_node,c.method)) {
+									exists=true;
+									break;
+								}
+							}
+						}
+
+						nl=NULL;
+					} else {
+						if (nl->get_filename()!=String()) {
+							//is an instance
+							Ref<SceneState> state = nl->get_scene_instance_state();
+							if (state.is_valid()) {
+								int from_node = state->find_node_by_path(nl->get_path_to(p_node));
+								int to_node = state->find_node_by_path(nl->get_path_to(n));
+
+								if (from_node>=0 && to_node>=0) {
+									//this one has state for this node, save
+									if (state->is_connection(from_node,c.signal,to_node,c.method)) {
+										exists=true;
+										break;
+									}
+								}
+							}
+
+						}
+						nl=nl->get_owner();
+					}
+				}
+
+				if (exists) {
+					continue;
+				}
+
+			}
+
+
+			int src_id;
+
+			if (node_map.has(p_node)) {
+				src_id=node_map[p_node];
+			} else {
+				if (nodepath_map.has(p_node)) {
+					src_id=FLAG_ID_IS_PATH|nodepath_map[p_node];
+				} else {
+					int sidx=nodepath_map.size();
+					nodepath_map[p_node]=sidx;
+					src_id=FLAG_ID_IS_PATH|sidx;
+				}
+			}
+
+
+
+			int target_id;
+
+			if (node_map.has(n)) {
+				target_id=node_map[n];
+			} else {
+				if (nodepath_map.has(n)) {
+					target_id=FLAG_ID_IS_PATH|nodepath_map[n];
+				} else {
+					int sidx=nodepath_map.size();
+					nodepath_map[n]=sidx;
+					target_id=FLAG_ID_IS_PATH|sidx;
+				}
+			}
+
 			ConnectionData cd;
-			cd.from=node_map[p_node];
-			cd.to=node_map[n];
+			cd.from=src_id;
+			cd.to=target_id;
 			cd.method=_nm_get_string(c.method,name_map);
 			cd.signal=_nm_get_string(c.signal,name_map);
 			cd.flags=c.flags;
@@ -392,7 +816,7 @@ Error PackedScene::_parse_connections(Node *p_owner,Node *p_node, Map<StringName
 	for(int i=0;i<p_node->get_child_count();i++) {
 
 		Node *c=p_node->get_child(i);
-		Error err = _parse_connections(p_owner,c,name_map,variant_map,node_map);
+		Error err = _parse_connections(p_owner,c,name_map,variant_map,node_map,nodepath_map);
 		if (err)
 			return err;
 	}
@@ -401,7 +825,7 @@ Error PackedScene::_parse_connections(Node *p_owner,Node *p_node, Map<StringName
 }
 
 
-Error PackedScene::pack(Node *p_scene) {
+Error SceneState::pack(Node *p_scene) {
 	ERR_FAIL_NULL_V( p_scene, ERR_INVALID_PARAMETER );
 
 
@@ -412,14 +836,27 @@ Error PackedScene::pack(Node *p_scene) {
 	Map<StringName,int> name_map;
 	HashMap<Variant,int,VariantHasher> variant_map;
 	Map<Node*,int> node_map;
+	Map<Node*,int> nodepath_map;
+
+	//if using scene inheritance, pack the scene it inherits from
+	if (scene->get_scene_inherited_state().is_valid()) {
+		String path = scene->get_scene_inherited_state()->get_path();
+		Ref<PackedScene> instance = ResourceLoader::load(path);
+		if (instance.is_valid()) {
+
+			base_scene_idx=_vm_get_variant(instance,variant_map);
+		}
+	}
+	//instanced, only direct sub-scnes are supported of course
+
 
-	Error err = _parse_node(scene,scene,-1,name_map,variant_map,node_map);
+	Error err = _parse_node(scene,scene,-1,name_map,variant_map,node_map,nodepath_map);
 	if (err) {
 		clear();
 		ERR_FAIL_V(err);
 	}
 
-	err = _parse_connections(scene,scene,name_map,variant_map,node_map);
+	err = _parse_connections(scene,scene,name_map,variant_map,node_map,nodepath_map);
 	if (err) {
 		clear();
 		ERR_FAIL_V(err);
@@ -440,19 +877,177 @@ Error PackedScene::pack(Node *p_scene) {
 		variants[idx]=*K;
 	}
 
+	node_paths.resize(nodepath_map.size());
+	for(Map<Node*,int>::Element *E=nodepath_map.front();E;E=E->next()) {
+
+		node_paths[E->get()]=scene->get_path_to(E->key());
+	}
+
+
 	return OK;
 }
 
-void PackedScene::clear() {
+void SceneState::set_path(const String &p_path) {
+
+	path=p_path;
+}
+
+String SceneState::get_path() const{
+
+	return path;
+}
+
+void SceneState::clear() {
 
 	names.clear();
 	variants.clear();
 	nodes.clear();
 	connections.clear();
+	node_path_cache.clear();
+	node_paths.clear();
+	editable_instances.clear();
+	base_scene_idx=-1;
 
 }
 
-void PackedScene::_set_bundled_scene(const Dictionary& d) {
+Ref<SceneState> SceneState::_get_base_scene_state() const {
+
+	if (base_scene_idx>=0) {
+
+		Ref<PackedScene> ps = variants[base_scene_idx];
+		if (ps.is_valid()) {
+			return ps->get_state();
+		}
+	}
+
+	return Ref<SceneState>();
+}
+
+int SceneState::find_node_by_path(const NodePath& p_node) const {
+
+	if (!node_path_cache.has(p_node)) {		
+		if (_get_base_scene_state().is_valid()) {
+			int idx = _get_base_scene_state()->find_node_by_path(p_node);
+			if (idx>=0) {
+				if (!base_scene_node_remap.has(idx)) {
+					int ridx = nodes.size() + base_scene_node_remap.size();
+					base_scene_node_remap[ridx]=idx;
+				}
+
+				return base_scene_node_remap[idx];
+			}
+		}
+		return -1;
+	}
+
+	int nid = node_path_cache[p_node];
+
+	if (_get_base_scene_state().is_valid() && !base_scene_node_remap.has(nid)) {
+		//for nodes that _do_ exist in current scene, still try to look for
+		//the node in the instanced scene, as a property may be missing
+		//from the local one
+		int idx = _get_base_scene_state()->find_node_by_path(p_node);
+		base_scene_node_remap[nid]=idx;
+
+	}
+
+	return nid;
+}
+Variant SceneState::get_property_value(int p_node, const StringName& p_property, bool &found) const {
+
+	found=false;
+
+	ERR_FAIL_COND_V(p_node<0,Variant());
+
+	if (p_node<nodes.size()) {
+		//find in built-in nodes
+		int pc = nodes[p_node].properties.size();
+		const StringName* namep = names.ptr();
+
+		const NodeData::Property *p=nodes[p_node].properties.ptr();
+		for(int i=0;i<pc;i++) {
+			if (p_property==namep[p[i].name]) {
+				found=true;
+				return variants[p[i].value];
+			}
+		}
+	}
+
+	//property not found, try on instance
+
+	if (base_scene_node_remap.has(p_node)) {
+		return _get_base_scene_state()->get_property_value(base_scene_node_remap[p_node],p_property,found);
+	}
+
+	return Variant();
+}
+
+bool SceneState::is_node_in_group(int p_node,const StringName& p_group) const {
+
+	ERR_FAIL_COND_V(p_node<0,false);
+
+	if (p_node<nodes.size()) {
+		const StringName* namep = names.ptr();
+		for(int i=0;i<nodes[p_node].groups.size();i++) {
+			if (namep[nodes[p_node].groups[i]]==p_group)
+				return true;
+		}
+	}
+
+	if (base_scene_node_remap.has(p_node)) {
+		return _get_base_scene_state()->is_node_in_group(base_scene_node_remap[p_node],p_group);
+	}
+
+	return false;
+}
+
+bool SceneState::disable_placeholders=false;
+
+void SceneState::set_disable_placeholders(bool p_disable) {
+
+	disable_placeholders=p_disable;
+}
+
+bool SceneState::is_connection(int p_node,const StringName& p_signal,int p_to_node,const StringName& p_to_method) const {
+
+	ERR_FAIL_COND_V(p_node<0,false);
+	ERR_FAIL_COND_V(p_to_node<0,false);
+
+	if (p_node<nodes.size() && p_to_node<nodes.size()) {
+
+		int signal_idx=-1;
+		int method_idx=-1;
+		for(int i=0;i<names.size();i++) {
+			if (names[i]==p_signal) {
+				signal_idx=i;
+			} else if (names[i]==p_to_method) {
+				method_idx=i;
+			}
+		}
+
+		if (signal_idx>=0 && method_idx>=0) {
+			//signal and method strings are stored..
+
+			for(int i=0;i<connections.size();i++) {
+
+				if (connections[i].from==p_node && connections[i].to==p_to_node && connections[i].signal==signal_idx && connections[i].method==method_idx) {
+
+					return true;
+				}
+			}
+		}
+	}
+
+	if (base_scene_node_remap.has(p_node) && base_scene_node_remap.has(p_to_node)) {
+		return _get_base_scene_state()->is_connection(base_scene_node_remap[p_node],p_signal,base_scene_node_remap[p_to_node],p_to_method);
+	}
+
+	return false;
+
+}
+
+
+void SceneState::set_bundled_scene(const Dictionary& d) {
 
 
 	ERR_FAIL_COND( !d.has("names"));
@@ -463,6 +1058,15 @@ void PackedScene::_set_bundled_scene(const Dictionary& d) {
 	ERR_FAIL_COND( !d.has("conns"));
 //	ERR_FAIL_COND( !d.has("path"));
 
+	int version=1;
+	if (d.has("version"))
+		version=d["version"];
+
+	if (version>PACK_VERSION) {
+		ERR_EXPLAIN("Save format version too new!");
+		ERR_FAIL();
+	}
+
 	DVector<String> snames = d["names"];
 	if (snames.size()) {
 
@@ -540,11 +1144,34 @@ void PackedScene::_set_bundled_scene(const Dictionary& d) {
 
 	}
 
+	Array np;
+	if (d.has("node_paths")) {
+		np=d["node_paths"];
+	}
+	node_paths.resize(np.size());
+	for(int i=0;i<np.size();i++) {
+		node_paths[i]=np[i];
+	}
+
+	Array ei;
+	if (d.has("editable_instances")) {
+		ei=d["editable_instances"];
+	}
+
+	if (d.has("base_scene")) {
+		base_scene_idx=d["base_scene"];
+	}
+
+	editable_instances.resize(ei.size());
+	for(int i=0;i<editable_instances.size();i++) {
+		editable_instances[i]=ei[i];
+	}
+
 //	path=d["path"];
 
 }
 
-Dictionary PackedScene::_get_bundled_scene() const {
+Dictionary SceneState::get_bundled_scene() const {
 
 	DVector<String> rnames;
 	rnames.resize(names.size());
@@ -605,7 +1232,25 @@ Dictionary PackedScene::_get_bundled_scene() const {
 	}
 
 	d["conns"]=rconns;
-	d["version"]=1;
+
+	Array rnode_paths;
+	rnode_paths.resize(node_paths.size());
+	for(int i=0;i<node_paths.size();i++) {
+		rnode_paths[i]=node_paths[i];
+	}	
+	d["node_paths"]=rnode_paths;
+
+	Array reditable_instances;
+	reditable_instances.resize(editable_instances.size());
+	for(int i=0;i<editable_instances.size();i++) {
+		reditable_instances[i]=editable_instances[i];
+	}
+	d["editable_instances"]=reditable_instances;
+	if (base_scene_idx>=0) {
+		d["base_scene"]=base_scene_idx;
+	}
+
+	d["version"]=PACK_VERSION;
 
 //	d["path"]=path;
 
@@ -614,10 +1259,264 @@ Dictionary PackedScene::_get_bundled_scene() const {
 
 }
 
+int SceneState::get_node_count() const {
+
+	return nodes.size();
+}
+
+StringName SceneState::get_node_type(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),StringName());
+	if (nodes[p_idx].type==TYPE_INSTANCED)
+		return StringName();
+	return names[nodes[p_idx].type];
+}
+
+StringName SceneState::get_node_name(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),StringName());
+	return names[nodes[p_idx].name];
+}
+
+Ref<PackedScene> SceneState::get_node_instance(int p_idx) const {
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),Ref<PackedScene>());
+	if (nodes[p_idx].instance>=0) {
+		return variants[nodes[p_idx].instance];
+	} else if (nodes[p_idx].parent<=0 || nodes[p_idx].parent==NO_PARENT_SAVED) {
+
+		if (base_scene_idx>=0) {
+			return variants[base_scene_idx];
+		}
+	}
+
+	return Ref<PackedScene>();
+
+
+}
+Vector<StringName> SceneState::get_node_groups(int p_idx) const{
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),Vector<StringName>());
+	Vector<StringName> groups;
+	for(int i=0;i<nodes[p_idx].groups.size();i++) {
+		groups.push_back(names[nodes[p_idx].groups[i]]);
+	}
+	return groups;
+}
+
+
+NodePath SceneState::get_node_path(int p_idx,bool p_for_parent) const {
+
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),NodePath());
+
+	if (nodes[p_idx].parent<0 || nodes[p_idx].parent==NO_PARENT_SAVED) {
+		if (p_for_parent) {
+			return NodePath();
+		} else {
+			return NodePath(".");
+		}
+	}
+
+	Vector<StringName> sub_path;
+	NodePath base_path;
+	int nidx=p_idx;
+	while(true) {
+		if (nodes[nidx].parent==NO_PARENT_SAVED || nodes[nidx].parent<0) {
+
+			sub_path.insert(0,".");
+			break;
+		}
+
+		if (!p_for_parent || p_idx!=nidx) {
+			sub_path.insert(0,names[nodes[nidx].name]);
+		}
+
+		if (nodes[nidx].parent&FLAG_ID_IS_PATH) {
+			base_path=node_paths[nodes[nidx].parent&FLAG_MASK];
+			break;
+		} else {
+			nidx=nodes[nidx].parent&FLAG_MASK;
+		}
+	}
+
+	for(int i=0;i<base_path.get_name_count();i++) {
+		StringName sn = base_path.get_name(i);
+		sub_path.insert(0,base_path.get_name(i));
+	}
+
+	if (sub_path.empty()) {
+		return NodePath(".");
+	}
+
+	return NodePath(sub_path,false);
+
+}
+
+int SceneState::get_node_property_count(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),-1);
+	return nodes[p_idx].properties.size();
+
+}
+StringName SceneState::get_node_property_name(int p_idx,int p_prop) const{
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),StringName());
+	ERR_FAIL_INDEX_V(p_prop,nodes[p_idx].properties.size(),StringName());
+	return names[nodes[p_idx].properties[p_prop].name];
+
+}
+Variant SceneState::get_node_property_value(int p_idx,int p_prop) const{
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),Variant());
+	ERR_FAIL_INDEX_V(p_prop,nodes[p_idx].properties.size(),Variant());
+
+	return variants[nodes[p_idx].properties[p_prop].value];
+}
+
+
+NodePath SceneState::get_node_owner_path(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,nodes.size(),NodePath());
+	if (nodes[p_idx].owner<0 || nodes[p_idx].owner==NO_PARENT_SAVED)
+		return NodePath(); //root likely
+	if (nodes[p_idx].owner&FLAG_ID_IS_PATH) {
+		return node_paths[nodes[p_idx].owner&FLAG_MASK];
+	} else {
+		return get_node_path(nodes[p_idx].owner&FLAG_MASK);
+	}
+}
+
+int SceneState::get_connection_count() const {
+
+	return connections.size();
+}
+NodePath SceneState::get_connection_source(int p_idx) const{
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),NodePath());
+	if (connections[p_idx].from&FLAG_ID_IS_PATH) {
+		return node_paths[connections[p_idx].from&FLAG_MASK];
+	} else {
+		return get_node_path(connections[p_idx].from&FLAG_MASK);
+	}
+
+}
+
+StringName SceneState::get_connection_signal(int p_idx) const{
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),StringName());
+	return names[connections[p_idx].signal];
+
+}
+NodePath SceneState::get_connection_target(int p_idx) const{
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),NodePath());
+	if (connections[p_idx].to&FLAG_ID_IS_PATH) {
+		return node_paths[connections[p_idx].to&FLAG_MASK];
+	} else {
+		return get_node_path(connections[p_idx].to&FLAG_MASK);
+	}
+
+}
+StringName SceneState::get_connection_method(int p_idx) const{
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),StringName());
+	return names[connections[p_idx].method];
+
+}
+int SceneState::get_connection_flags(int p_idx) const{
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),-1);
+	return connections[p_idx].flags;
+}
+
+Array SceneState::get_connection_binds(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,connections.size(),-1);
+	Array binds;
+	for(int i=0;i<connections[p_idx].binds.size();i++) {
+		binds.push_back(variants[connections[p_idx].binds[i]]);
+	}
+	return binds;
+}
+
+Vector<NodePath> SceneState::get_editable_instances() const {
+	return editable_instances;
+}
+
+
+SceneState::SceneState() {
+
+	base_scene_idx=-1;
+}
+
+
+////////////////
+
+
+
+void PackedScene::_set_bundled_scene(const Dictionary& d) {
+
+	state->set_bundled_scene(d);
+}
+
+Dictionary PackedScene::_get_bundled_scene() const {
+
+	return state->get_bundled_scene();
+}
+
+
+Error PackedScene::pack(Node *p_scene) {
+
+	return state->pack(p_scene);
+}
+
+void PackedScene::clear() {
+
+	state->clear();
+}
+
+bool PackedScene::can_instance() const {
+
+	return state->can_instance();
+}
+
+Node *PackedScene::instance(bool p_gen_edit_state) const {
+
+#ifndef TOOLS_ENABLED
+	if (p_gen_edit_state) {
+		ERR_EXPLAIN("Edit state is only for editors, does not work without tools compiled");
+		ERR_FAIL_COND_V(p_gen_edit_state,NULL);
+	}
+#endif
+
+	Node *s = state->instance(p_gen_edit_state);
+	if (!s)
+		return NULL;
+
+	if (p_gen_edit_state) {
+		s->set_scene_instance_state(state);
+	}
+
+	if (get_path()!="" && get_path().find("::")==-1)
+		s->set_filename(get_path());
+
+
+	s->notification(Node::NOTIFICATION_INSTANCED);
+
+	return s;
+}
+
+Ref<SceneState> PackedScene::get_state() {
+
+	return state;
+}
+
+void PackedScene::set_path(const String& p_path,bool p_take_over) {
+
+	state->set_path(p_path);
+	Resource::set_path(p_path,p_take_over);
+}
+
 void PackedScene::_bind_methods() {
 
 	ObjectTypeDB::bind_method(_MD("pack","path:Node"),&PackedScene::pack);
-	ObjectTypeDB::bind_method(_MD("instance:Node"),&PackedScene::instance,DEFVAL(false));
+	ObjectTypeDB::bind_method(_MD("instance:Node","gen_edit_state"),&PackedScene::instance,DEFVAL(false));
 	ObjectTypeDB::bind_method(_MD("can_instance"),&PackedScene::can_instance);
 	ObjectTypeDB::bind_method(_MD("_set_bundled_scene"),&PackedScene::_set_bundled_scene);
 	ObjectTypeDB::bind_method(_MD("_get_bundled_scene"),&PackedScene::_get_bundled_scene);
@@ -628,5 +1527,6 @@ void PackedScene::_bind_methods() {
 
 PackedScene::PackedScene() {
 
+	state = Ref<SceneState>( memnew( SceneState ));
 
 }
diff --git a/scene/resources/packed_scene.h b/scene/resources/packed_scene.h
index 6c7fa545d4..3956d2abe4 100644
--- a/scene/resources/packed_scene.h
+++ b/scene/resources/packed_scene.h
@@ -32,17 +32,27 @@
 #include "resource.h"
 #include "scene/main/node.h"
 
-class PackedScene : public Resource {
 
-	OBJ_TYPE( PackedScene, Resource );
-	RES_BASE_EXTENSION("scn");
+class SceneState : public Reference {
+
+	OBJ_TYPE( SceneState, Reference );
 	Vector<StringName> names;
 	Vector<Variant> variants;
+	Vector<NodePath> node_paths;
+	Vector<NodePath> editable_instances;
+	mutable HashMap<NodePath,int> node_path_cache;
+	mutable Map<int,int> base_scene_node_remap;
 
-	//missing - instances
-	//missing groups
-	//missing - owner
-	//missing - override names and values
+	int base_scene_idx;
+
+	enum {
+		FLAG_ID_IS_PATH=(1<<30),
+		FLAG_INSTANCE_IS_PLACEHOLDER=(1<<30),
+		FLAG_MASK=(1<<24)-1,
+		NO_PARENT_SAVED=0x7FFFFFFF,
+		TYPE_INSTANCED=0x7FFFFFFF,
+
+	};
 
 	struct NodeData {
 
@@ -59,9 +69,15 @@ class PackedScene : public Resource {
 		};
 
 		Vector<Property> properties;
-		Vector<int> groups;
+		Vector<int> groups;		
+
 	};
 
+	struct PackState {
+		Ref<SceneState> state;
+		int node;
+		PackState() { node=-1; }
+	};
 
 	Vector<NodeData> nodes;
 
@@ -77,16 +93,78 @@ class PackedScene : public Resource {
 
 	Vector<ConnectionData> connections;
 
-	Error _parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map);
-	Error _parse_connections(Node *p_owner,Node *p_node, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map);
 
+	Error _parse_node(Node *p_owner,Node *p_node,int p_parent_idx, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map,Map<Node*,int> &nodepath_map);
+	Error _parse_connections(Node *p_owner,Node *p_node, Map<StringName,int> &name_map,HashMap<Variant,int,VariantHasher> &variant_map,Map<Node*,int> &node_map,Map<Node*,int> &nodepath_map);
+
+	String path;
+
+	_FORCE_INLINE_ Ref<SceneState> _get_base_scene_state() const;
+
+	static bool disable_placeholders;
+public:
+
+	static void set_disable_placeholders(bool p_disable);
+
+	int find_node_by_path(const NodePath& p_node) const;
+	Variant get_property_value(int p_node,const StringName& p_property,bool &found) const;
+	bool is_node_in_group(int p_node,const StringName& p_group) const;
+	bool is_connection(int p_node,const StringName& p_signal,int p_to_node,const StringName& p_to_method) const;
+
+
+	void set_bundled_scene(const Dictionary& p_dictionary);
+	Dictionary get_bundled_scene() const;
+
+	Error pack(Node *p_scene);
+
+	void set_path(const String &p_path);
+	String get_path() const;
+
+	void clear();
+
+	bool can_instance() const;
+	Node *instance(bool p_gen_edit_state=false) const;
+
+
+	//build-unbuild API
+
+	int get_node_count() const;
+	StringName get_node_type(int p_idx) const;
+	StringName get_node_name(int p_idx) const;
+	NodePath get_node_path(int p_idx,bool p_for_parent=false) const;
+	NodePath get_node_owner_path(int p_idx) const;
+	Ref<PackedScene> get_node_instance(int p_idx) const;
+	Vector<StringName> get_node_groups(int p_idx) const;
+
+	int get_node_property_count(int p_idx) const;
+	StringName get_node_property_name(int p_idx,int p_prop) const;
+	Variant get_node_property_value(int p_idx,int p_prop) const;
+
+	int get_connection_count() const;
+	NodePath get_connection_source(int p_idx) const;
+	StringName get_connection_signal(int p_idx) const;
+	NodePath get_connection_target(int p_idx) const;
+	StringName get_connection_method(int p_idx) const;
+	int get_connection_flags(int p_idx) const;
+	Array get_connection_binds(int p_idx) const;
+
+	Vector<NodePath> get_editable_instances() const;
+
+	SceneState();
+};
+
+class PackedScene : public Resource {
+
+	OBJ_TYPE(PackedScene, Resource );
+	RES_BASE_EXTENSION("scn");
+
+	Ref<SceneState> state;
 
 	void _set_bundled_scene(const Dictionary& p_scene);
 	Dictionary _get_bundled_scene() const;
 
 protected:
 
-
 	static void _bind_methods();
 public:
 
@@ -98,7 +176,12 @@ public:
 	bool can_instance() const;
 	Node *instance(bool p_gen_edit_state=false) const;
 
+	virtual void set_path(const String& p_path,bool p_take_over=false);
+
+	Ref<SceneState> get_state();
+
 	PackedScene();
+
 };
 
 #endif // SCENE_PRELOADER_H
diff --git a/scene/resources/scene_format_text.cpp b/scene/resources/scene_format_text.cpp
new file mode 100644
index 0000000000..8403c06ad1
--- /dev/null
+++ b/scene/resources/scene_format_text.cpp
@@ -0,0 +1,792 @@
+#include "scene_format_text.h"
+
+#include "globals.h"
+#include "version.h"
+#include "os/dir_access.h"
+
+#define FORMAT_VERSION 1
+
+void ResourceFormatSaverTextInstance::write_property(const String& p_name,const Variant& p_property,bool *r_ok) {
+
+	if (r_ok)
+		*r_ok=false;
+
+	if (p_name!=String()) {
+		f->store_string(p_name+" = ");
+	}
+
+	switch( p_property.get_type() ) {
+
+		case Variant::NIL: {
+			f->store_string("null");
+		} break;
+		case Variant::BOOL: {
+
+			f->store_string(p_property.operator bool() ? "true":"false" );
+		} break;
+		case Variant::INT: {
+
+			f->store_string( itos(p_property.operator int()) );
+		} break;
+		case Variant::REAL: {
+
+			f->store_string( rtoss(p_property.operator real_t()) );
+		} break;
+		case Variant::STRING: {
+
+			String str=p_property;
+
+			str="\""+str.c_escape()+"\"";
+			f->store_string( str );
+		} break;
+		case Variant::VECTOR2: {
+
+			Vector2 v = p_property;
+			f->store_string("Vector2( "+rtoss(v.x) +", "+rtoss(v.y)+" )" );
+		} break;
+		case Variant::RECT2: {
+
+			Rect2 aabb = p_property;
+			f->store_string("Rect2( "+rtoss(aabb.pos.x) +", "+rtoss(aabb.pos.y) +", "+rtoss(aabb.size.x) +", "+rtoss(aabb.size.y)+" )" );
+
+		} break;
+		case Variant::VECTOR3: {
+
+			Vector3 v = p_property;
+			f->store_string("Vector3( "+rtoss(v.x) +", "+rtoss(v.y)+", "+rtoss(v.z)+" )");
+		} break;
+		case Variant::PLANE: {
+
+			Plane p = p_property;
+			f->store_string("Plane( "+rtoss(p.normal.x) +", "+rtoss(p.normal.y)+", "+rtoss(p.normal.z)+", "+rtoss(p.d)+" )" );
+
+		} break;
+		case Variant::_AABB: {
+
+			AABB aabb = p_property;
+			f->store_string("AABB( "+rtoss(aabb.pos.x) +", "+rtoss(aabb.pos.y) +", "+rtoss(aabb.pos.z) +", "+rtoss(aabb.size.x) +", "+rtoss(aabb.size.y) +", "+rtoss(aabb.size.z)+" )"  );
+
+		} break;
+		case Variant::QUAT: {
+
+			Quat quat = p_property;
+			f->store_string("Quat( "+rtoss(quat.x)+", "+rtoss(quat.y)+", "+rtoss(quat.z)+", "+rtoss(quat.w)+" )");
+
+		} break;
+		case Variant::MATRIX32: {
+
+			String s="Matrix32( ";
+			Matrix32 m3 = p_property;
+			for (int i=0;i<3;i++) {
+				for (int j=0;j<2;j++) {
+
+					if (i!=0 || j!=0)
+						s+=", ";
+					s+=rtoss( m3.elements[i][j] );
+				}
+			}
+
+			f->store_string(s+" )");
+
+		} break;
+		case Variant::MATRIX3: {
+
+			String s="Matrix3( ";
+			Matrix3 m3 = p_property;
+			for (int i=0;i<3;i++) {
+				for (int j=0;j<3;j++) {
+
+					if (i!=0 || j!=0)
+						s+=", ";
+					s+=rtoss( m3.elements[i][j] );
+				}
+			}
+
+			f->store_string(s+" )");
+
+		} break;
+		case Variant::TRANSFORM: {
+
+			String s="Transform( ";
+			Transform t = p_property;
+			Matrix3 &m3 = t.basis;
+			for (int i=0;i<3;i++) {
+				for (int j=0;j<3;j++) {
+
+					if (i!=0 || j!=0)
+						s+=", ";
+					s+=rtoss( m3.elements[i][j] );
+				}
+			}
+
+			s=s+", "+rtoss(t.origin.x) +", "+rtoss(t.origin.y)+", "+rtoss(t.origin.z);
+
+			f->store_string(s+" )");
+		} break;
+
+			// misc types
+		case Variant::COLOR: {
+
+			Color c = p_property;
+			f->store_string("Color( "+rtoss(c.r) +", "+rtoss(c.g)+", "+rtoss(c.b)+", "+rtoss(c.a)+" )");
+
+		} break;
+		case Variant::IMAGE: {
+
+
+			Image img=p_property;
+
+			if (img.empty()) {
+				f->store_string("RawImage()");
+				break;
+			}
+
+			String imgstr="RawImage( ";
+			imgstr+=itos(img.get_width());
+			imgstr+=", "+itos(img.get_height());
+			imgstr+=", "+itos(img.get_mipmaps());
+			imgstr+=", ";
+
+			switch(img.get_format()) {
+
+				case Image::FORMAT_GRAYSCALE: imgstr+="GRAYSCALE"; break;
+				case Image::FORMAT_INTENSITY: imgstr+="INTENSITY"; break;
+				case Image::FORMAT_GRAYSCALE_ALPHA: imgstr+="GRAYSCALE_ALPHA"; break;
+				case Image::FORMAT_RGB: imgstr+="RGB"; break;
+				case Image::FORMAT_RGBA: imgstr+="RGBA"; break;
+				case Image::FORMAT_INDEXED : imgstr+="INDEXED"; break;
+				case Image::FORMAT_INDEXED_ALPHA: imgstr+="INDEXED_ALPHA"; break;
+				case Image::FORMAT_BC1: imgstr+="BC1"; break;
+				case Image::FORMAT_BC2: imgstr+="BC2"; break;
+				case Image::FORMAT_BC3: imgstr+="BC3"; break;
+				case Image::FORMAT_BC4: imgstr+="BC4"; break;
+				case Image::FORMAT_BC5: imgstr+="BC5"; break;
+				case Image::FORMAT_PVRTC2: imgstr+="PVRTC2"; break;
+				case Image::FORMAT_PVRTC2_ALPHA: imgstr+="PVRTC2_ALPHA"; break;
+				case Image::FORMAT_PVRTC4: imgstr+="PVRTC4"; break;
+				case Image::FORMAT_PVRTC4_ALPHA: imgstr+="PVRTC4_ALPHA"; break;
+				case Image::FORMAT_ETC: imgstr+="ETC"; break;
+				case Image::FORMAT_ATC: imgstr+="ATC"; break;
+				case Image::FORMAT_ATC_ALPHA_EXPLICIT: imgstr+="ATC_ALPHA_EXPLICIT"; break;
+				case Image::FORMAT_ATC_ALPHA_INTERPOLATED: imgstr+="ATC_ALPHA_INTERPOLATED"; break;
+				case Image::FORMAT_CUSTOM: imgstr+="CUSTOM"; break;
+				default: {}
+			}
+
+
+			String s;
+
+			DVector<uint8_t> data = img.get_data();
+			int len = data.size();
+			DVector<uint8_t>::Read r = data.read();
+			const uint8_t *ptr=r.ptr();;
+			for (int i=0;i<len;i++) {
+
+				uint8_t byte = ptr[i];
+				const char  hex[16]={'0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F'};
+				char str[3]={ hex[byte>>4], hex[byte&0xF], 0};
+				s+=str;
+			}
+
+			imgstr+=", ";
+			f->store_string(imgstr);
+			f->store_string(s);
+			f->store_string(" )");
+		} break;
+		case Variant::NODE_PATH: {
+
+			String str=p_property;
+
+			str="NodePath(\""+str.c_escape()+"\")";
+			f->store_string(str);
+
+		} break;
+
+		case Variant::OBJECT: {
+
+			RES res = p_property;
+			if (res.is_null()) {
+				f->store_string("null");
+				if (r_ok)
+					*r_ok=true;
+
+				break; // don't save it
+			}
+
+			if (external_resources.has(res)) {
+
+				f->store_string("ExtResource( "+itos(external_resources[res]+1)+" )");
+			} else {
+
+				if (internal_resources.has(res)) {
+					f->store_string("SubResource( "+itos(internal_resources[res])+" )");
+				} else 	if (res->get_path().length() && res->get_path().find("::")==-1) {
+
+					//external resource
+					String path=relative_paths?local_path.path_to_file(res->get_path()):res->get_path();
+					f->store_string("Resource( \""+path+"\" )");
+				} else {
+					f->store_string("null");
+					ERR_EXPLAIN("Resource was not pre cached for the resource section, bug?");
+					ERR_BREAK(true);
+					//internal resource
+				}
+			}
+
+		} break;
+		case Variant::INPUT_EVENT: {
+
+			f->store_string("InputEvent()"); //will be added later
+		} break;
+		case Variant::DICTIONARY: {
+
+			Dictionary dict = p_property;
+
+			List<Variant> keys;
+			dict.get_key_list(&keys);
+			keys.sort();
+
+			f->store_string("{ ");
+			for(List<Variant>::Element *E=keys.front();E;E=E->next()) {
+
+				//if (!_check_type(dict[E->get()]))
+				//	continue;
+				bool ok;
+				write_property("",E->get(),&ok);
+				ERR_CONTINUE(!ok);
+
+				f->store_string(":");
+				write_property("",dict[E->get()],&ok);
+				if (!ok)
+					write_property("",Variant()); //at least make the file consistent..
+				if (E->next())
+					f->store_string(", ");
+			}
+
+
+			f->store_string(" }");
+
+
+		} break;
+		case Variant::ARRAY: {
+
+			f->store_string("[ ");
+			Array array = p_property;
+			int len=array.size();
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				write_property("",array[i]);
+
+
+			}
+			f->store_string(" ]");
+
+		} break;
+
+		case Variant::RAW_ARRAY: {
+
+			f->store_string("RawArray( ");
+			String s;
+			DVector<uint8_t> data = p_property;
+			int len = data.size();
+			DVector<uint8_t>::Read r = data.read();
+			const uint8_t *ptr=r.ptr();;
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				uint8_t byte = ptr[i];
+				const char  hex[16]={'0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F'};
+				char str[3]={ hex[byte>>4], hex[byte&0xF], 0};
+				f->store_string(str);
+
+			}
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::INT_ARRAY: {
+
+			f->store_string("IntArray( ");
+			DVector<int> data = p_property;
+			int len = data.size();
+			DVector<int>::Read r = data.read();
+			const int *ptr=r.ptr();;
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+
+				f->store_string(itos(ptr[i]));
+			}
+
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::REAL_ARRAY: {
+
+			f->store_string("FloatArray( ");
+			DVector<real_t> data = p_property;
+			int len = data.size();
+			DVector<real_t>::Read r = data.read();
+			const real_t *ptr=r.ptr();;
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				f->store_string(rtoss(ptr[i]));
+			}
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::STRING_ARRAY: {
+
+			f->store_string("StringArray( ");
+			DVector<String> data = p_property;
+			int len = data.size();
+			DVector<String>::Read r = data.read();
+			const String *ptr=r.ptr();;
+			String s;
+			//write_string("\n");
+
+
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				String str=ptr[i];
+				f->store_string(""+str.c_escape()+"\"");
+			}
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::VECTOR2_ARRAY: {
+
+			f->store_string("Vector2Array( ");
+			DVector<Vector2> data = p_property;
+			int len = data.size();
+			DVector<Vector2>::Read r = data.read();
+			const Vector2 *ptr=r.ptr();;
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				f->store_string(rtoss(ptr[i].x)+", "+rtoss(ptr[i].y) );
+			}
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::VECTOR3_ARRAY: {
+
+			f->store_string("Vector3Array( ");
+			DVector<Vector3> data = p_property;
+			int len = data.size();
+			DVector<Vector3>::Read r = data.read();
+			const Vector3 *ptr=r.ptr();;
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+				f->store_string(rtoss(ptr[i].x)+", "+rtoss(ptr[i].y)+", "+rtoss(ptr[i].z) );
+			}
+
+			f->store_string(" )");
+
+		} break;
+		case Variant::COLOR_ARRAY: {
+
+			f->store_string("ColorArray( ");
+
+			DVector<Color> data = p_property;
+			int len = data.size();
+			DVector<Color>::Read r = data.read();
+			const Color *ptr=r.ptr();;
+
+			for (int i=0;i<len;i++) {
+
+				if (i>0)
+					f->store_string(", ");
+
+				f->store_string(rtoss(ptr[i].r)+", "+rtoss(ptr[i].g)+", "+rtoss(ptr[i].b)+", "+rtoss(ptr[i].a) );
+
+			}
+			f->store_string(" )");
+
+		} break;
+		default: {}
+
+	}
+
+	if (r_ok)
+		*r_ok=true;
+
+}
+
+
+void ResourceFormatSaverTextInstance::_find_resources(const Variant& p_variant,bool p_main) {
+
+
+	switch(p_variant.get_type()) {
+		case Variant::OBJECT: {
+
+
+			RES res = p_variant.operator RefPtr();
+
+			if (res.is_null() || external_resources.has(res))
+				return;
+
+			if (!p_main && (!bundle_resources ) && res->get_path().length() && res->get_path().find("::") == -1 ) {
+				int index = external_resources.size();
+				external_resources[res]=index;
+				return;
+			}
+
+			if (resource_set.has(res))
+				return;
+
+			List<PropertyInfo> property_list;
+
+			res->get_property_list( &property_list );
+			property_list.sort();
+
+			List<PropertyInfo>::Element *I=property_list.front();
+
+			while(I) {
+
+				PropertyInfo pi=I->get();
+
+				if (pi.usage&PROPERTY_USAGE_STORAGE || (bundle_resources && pi.usage&PROPERTY_USAGE_BUNDLE)) {
+
+					Variant v=res->get(I->get().name);
+					_find_resources(v);
+				}
+
+				I=I->next();
+			}
+
+			resource_set.insert( res ); //saved after, so the childs it needs are available when loaded
+			saved_resources.push_back(res);
+
+		} break;
+		case Variant::ARRAY: {
+
+			Array varray=p_variant;
+			int len=varray.size();
+			for(int i=0;i<len;i++) {
+
+				Variant v=varray.get(i);
+				_find_resources(v);
+			}
+
+		} break;
+		case Variant::DICTIONARY: {
+
+			Dictionary d=p_variant;
+			List<Variant> keys;
+			d.get_key_list(&keys);
+			for(List<Variant>::Element *E=keys.front();E;E=E->next()) {
+
+				Variant v = d[E->get()];
+				_find_resources(v);
+			}
+		} break;
+		default: {}
+	}
+
+}
+
+
+
+Error ResourceFormatSaverTextInstance::save(const String &p_path,const RES& p_resource,uint32_t p_flags) {
+
+	if (p_path.ends_with(".tscn")) {
+		packed_scene=p_resource;
+	}
+
+	Error err;
+	f = FileAccess::open(p_path, FileAccess::WRITE,&err);
+	ERR_FAIL_COND_V( err, ERR_CANT_OPEN );
+	FileAccessRef _fref(f);
+
+	local_path = Globals::get_singleton()->localize_path(p_path);
+
+	relative_paths=p_flags&ResourceSaver::FLAG_RELATIVE_PATHS;
+	skip_editor=p_flags&ResourceSaver::FLAG_OMIT_EDITOR_PROPERTIES;
+	bundle_resources=p_flags&ResourceSaver::FLAG_BUNDLE_RESOURCES;
+	takeover_paths=p_flags&ResourceSaver::FLAG_REPLACE_SUBRESOURCE_PATHS;
+	if (!p_path.begins_with("res://")) {
+		takeover_paths=false;
+	}
+
+	// save resources
+	_find_resources(p_resource,true);
+
+	if (packed_scene.is_valid()) {
+		//add instances to external resources if saving a packed scene
+		for(int i=0;i<packed_scene->get_state()->get_node_count();i++) {
+			Ref<PackedScene> instance=packed_scene->get_state()->get_node_instance(i);
+			if (instance.is_valid() && !external_resources.has(instance)) {
+				int index = external_resources.size();
+				external_resources[instance]=index;
+			}
+		}
+	}
+
+
+	ERR_FAIL_COND_V(err!=OK,err);
+
+	{
+		String title=packed_scene.is_valid()?"[gd_scene ":"[gd_resource ";
+		if (packed_scene.is_null())
+			title+="type=\""+p_resource->get_type()+"\" ";
+		int load_steps=saved_resources.size()+external_resources.size();
+		//if (packed_scene.is_valid()) {
+		//	load_steps+=packed_scene->get_node_count();
+		//}
+		//no, better to not use load steps from nodes, no point to that
+
+		if (load_steps>1) {
+			title+="load_steps="+itos(load_steps)+" ";
+		}
+		title+="format="+itos(FORMAT_VERSION)+"";
+		//title+="engine_version=\""+itos(VERSION_MAJOR)+"."+itos(VERSION_MINOR)+"\"";
+
+		f->store_string(title);
+		f->store_line("]\n"); //one empty line
+	}
+
+
+	for(Map<RES,int>::Element *E=external_resources.front();E;E=E->next()) {
+
+		String p = E->key()->get_path();
+
+		f->store_string("[ext_resource path=\""+p+"\" type=\""+E->key()->get_save_type()+"\" id="+itos(E->get()+1)+"]\n"); //bundled
+	}
+
+	if (external_resources.size())
+		f->store_line(String()); //separate
+
+	Set<int> used_indices;
+
+	for(List<RES>::Element *E=saved_resources.front();E;E=E->next()) {
+
+		RES res = E->get();
+		if (E->next() && (res->get_path()=="" || res->get_path().find("::") != -1 )) {
+
+			if (res->get_subindex()!=0) {
+				if (used_indices.has(res->get_subindex())) {
+					res->set_subindex(0); //repeated
+				} else {
+					used_indices.insert(res->get_subindex());
+				}
+			}
+		}
+	}
+
+	for(List<RES>::Element *E=saved_resources.front();E;E=E->next()) {
+
+		RES res = E->get();
+		ERR_CONTINUE(!resource_set.has(res));
+		bool main = (E->next()==NULL);
+
+		if (main && packed_scene.is_valid())
+			break; //save as a scene
+
+		if (main) {
+			f->store_line("[resource]\n");
+		} else {
+			String line="[sub_resource ";
+			if (res->get_subindex()==0) {
+				int new_subindex=1;
+				if (used_indices.size()) {
+					new_subindex=used_indices.back()->get()+1;
+				}
+
+				res->set_subindex(new_subindex);
+				used_indices.insert(new_subindex);
+			}
+
+			int idx = res->get_subindex();
+			line+="type=\""+res->get_type()+"\" id="+itos(idx);
+			f->store_line(line+"]\n");
+			if (takeover_paths) {
+				res->set_path(p_path+"::"+itos(idx),true);
+			}
+
+			internal_resources[res]=idx;
+
+		}
+
+
+		List<PropertyInfo> property_list;
+		res->get_property_list(&property_list);
+//		property_list.sort();
+		for(List<PropertyInfo>::Element *PE = property_list.front();PE;PE=PE->next()) {
+
+
+			if (skip_editor && PE->get().name.begins_with("__editor"))
+				continue;
+
+			if (PE->get().usage&PROPERTY_USAGE_STORAGE || (bundle_resources && PE->get().usage&PROPERTY_USAGE_BUNDLE)) {
+
+				String name = PE->get().name;
+				Variant value = res->get(name);
+
+
+				if ((PE->get().usage&PROPERTY_USAGE_STORE_IF_NONZERO && value.is_zero())||(PE->get().usage&PROPERTY_USAGE_STORE_IF_NONONE && value.is_one()) )
+					continue;
+
+				if (PE->get().type==Variant::OBJECT && value.is_zero())
+					continue;
+
+				write_property(name,value);
+				f->store_string("\n");
+			}
+
+
+		}
+
+		f->store_string("\n");
+
+	}
+
+	if (packed_scene.is_valid()) {
+		//if this is a scene, save nodes and connections!
+		Ref<SceneState> state = packed_scene->get_state();
+		for(int i=0;i<state->get_node_count();i++) {
+
+			StringName type = state->get_node_type(i);
+			StringName name = state->get_node_name(i);
+			NodePath path = state->get_node_path(i,true);
+			NodePath owner = state->get_node_owner_path(i);
+			Ref<PackedScene> instance = state->get_node_instance(i);
+			Vector<StringName> groups = state->get_node_groups(i);
+
+			String header="[node";
+			header+=" name=\""+String(name)+"\"";
+			if (type!=StringName()) {
+				header+=" type=\""+String(type)+"\"";
+			}
+			if (path!=NodePath()) {
+				header+=" parent=\""+String(path.simplified())+"\"";
+			}
+			if (owner!=NodePath() && owner!=NodePath(".")) {
+				header+=" owner=\""+String(owner.simplified())+"\"";
+			}
+
+			if (groups.size()) {
+				String sgroups=" groups=[ ";
+				for(int j=0;j<groups.size();j++) {
+					if (j>0)
+						sgroups+=", ";
+					sgroups+="\""+groups[i].operator String().c_escape()+"\"";
+				}
+				sgroups+=" ]";
+				header+=sgroups;
+			}
+
+			f->store_string(header);
+
+			if (instance.is_valid()) {
+				f->store_string(" instance=");
+				write_property("",instance);
+			}
+
+			f->store_line("]\n");
+
+			for(int j=0;j<state->get_node_property_count(i);j++) {
+
+				write_property(state->get_node_property_name(i,j),state->get_node_property_value(i,j));
+				f->store_line(String());
+
+			}
+
+			if (state->get_node_property_count(i)) {
+				//add space
+				f->store_line(String());
+			}
+
+		}
+
+		for(int i=0;i<state->get_connection_count();i++) {
+
+			String connstr="[connection";
+			connstr+=" signal=\""+String(state->get_connection_signal(i))+"\"";
+			connstr+=" from=\""+String(state->get_connection_source(i).simplified())+"\"";
+			connstr+=" to=\""+String(state->get_connection_target(i).simplified())+"\"";
+			connstr+=" method=\""+String(state->get_connection_method(i))+"\"";
+			int flags = state->get_connection_flags(i);
+			if (flags!=Object::CONNECT_PERSIST) {
+				connstr+=" flags="+itos(flags);
+			}
+
+			Array binds=state->get_connection_binds(i);
+			f->store_string(connstr);
+			if (binds.size()) {
+				f->store_string(" binds=");
+				write_property("",binds);
+			}
+
+			f->store_line("]\n");
+		}
+
+		f->store_line(String());
+
+		Vector<NodePath> editable_instances = state->get_editable_instances();
+		for(int i=0;i<editable_instances.size();i++) {
+			f->store_line("[editable path=\""+editable_instances[i].operator String()+"\"]");
+		}
+	}
+
+	if (f->get_error()!=OK && f->get_error()!=ERR_FILE_EOF) {
+		f->close();
+		return ERR_CANT_CREATE;
+	}
+
+	f->close();
+	//memdelete(f);
+
+	return OK;
+}
+
+
+
+Error ResourceFormatSaverText::save(const String &p_path,const RES& p_resource,uint32_t p_flags) {
+
+	if (p_path.ends_with(".sct") && p_resource->get_type()!="PackedScene") {
+		return ERR_FILE_UNRECOGNIZED;
+	}
+
+	ResourceFormatSaverTextInstance saver;
+	return saver.save(p_path,p_resource,p_flags);
+
+}
+
+bool ResourceFormatSaverText::recognize(const RES& p_resource) const {
+
+
+	return true; // all recognized!
+}
+void ResourceFormatSaverText::get_recognized_extensions(const RES& p_resource,List<String> *p_extensions) const {
+
+	p_extensions->push_back("tres"); //text resource
+	if (p_resource->get_type()=="PackedScene")
+		p_extensions->push_back("tscn"); //text scene
+
+}
+
+ResourceFormatSaverText* ResourceFormatSaverText::singleton=NULL;
+ResourceFormatSaverText::ResourceFormatSaverText() {
+	singleton=this;
+}
diff --git a/scene/resources/scene_format_text.h b/scene/resources/scene_format_text.h
new file mode 100644
index 0000000000..576a78d183
--- /dev/null
+++ b/scene/resources/scene_format_text.h
@@ -0,0 +1,46 @@
+#ifndef SCENE_FORMAT_TEXT_H
+#define SCENE_FORMAT_TEXT_H
+
+#include "io/resource_loader.h"
+#include "io/resource_saver.h"
+#include "os/file_access.h"
+#include "scene/resources/packed_scene.h"
+
+class ResourceFormatSaverTextInstance  {
+
+	String local_path;
+
+	Ref<PackedScene> packed_scene;
+
+	bool takeover_paths;
+	bool relative_paths;
+	bool bundle_resources;
+	bool skip_editor;
+	FileAccess *f;
+	Set<RES> resource_set;
+	List<RES> saved_resources;
+	Map<RES,int> external_resources;
+	Map<RES,int> internal_resources;
+
+	void _find_resources(const Variant& p_variant,bool p_main=false);
+	void write_property(const String& p_name,const Variant& p_property,bool *r_ok=NULL);
+
+public:
+
+	Error save(const String &p_path,const RES& p_resource,uint32_t p_flags=0);
+
+
+};
+
+class ResourceFormatSaverText : public ResourceFormatSaver {
+public:
+	static ResourceFormatSaverText* singleton;
+	virtual Error save(const String &p_path,const RES& p_resource,uint32_t p_flags=0);
+	virtual bool recognize(const RES& p_resource) const;
+	virtual void get_recognized_extensions(const RES& p_resource,List<String> *p_extensions) const;
+
+	ResourceFormatSaverText();
+};
+
+
+#endif // SCENE_FORMAT_TEXT_H
diff --git a/scene/resources/video_stream.cpp b/scene/resources/video_stream.cpp
index b27413bb68..c957fd4c67 100644
--- a/scene/resources/video_stream.cpp
+++ b/scene/resources/video_stream.cpp
@@ -28,16 +28,12 @@
 /*************************************************************************/
 #include "video_stream.h"
 
-void VideoStream::_bind_methods() {
+void VideoStreamPlayback::_bind_methods() {
 
-	ObjectTypeDB::bind_method(_MD("get_pending_frame_count"),&VideoStream::get_pending_frame_count);
-	ObjectTypeDB::bind_method(_MD("pop_frame"),&VideoStream::pop_frame);
-	ObjectTypeDB::bind_method(_MD("peek_frame"),&VideoStream::peek_frame);
-	ObjectTypeDB::bind_method(_MD("set_audio_track","idx"),&VideoStream::set_audio_track);
 };
 
 
-VideoStream::VideoStream() {
+VideoStreamPlayback::VideoStreamPlayback() {
 
 };
 
diff --git a/scene/resources/video_stream.h b/scene/resources/video_stream.h
index 2ad7457ec4..a23ef0c64f 100644
--- a/scene/resources/video_stream.h
+++ b/scene/resources/video_stream.h
@@ -33,15 +33,17 @@
 #include "scene/resources/texture.h"
 
 
-class VideoStream : public Resource {
+class VideoStreamPlayback : public Resource {
 
-	OBJ_TYPE(VideoStream,Resource);
+	OBJ_TYPE(VideoStreamPlayback,Resource);
 
 protected:
 	static void _bind_methods();
 
 public:
 
+	typedef int (*AudioMixCallback)(void* p_udata,const int16_t *p_data,int p_frames);
+
 	virtual void stop()=0;
 	virtual void play()=0;
 
@@ -58,16 +60,34 @@ public:
 	virtual float get_pos() const=0;
 	virtual void seek_pos(float p_time)=0;
 
-	virtual int get_pending_frame_count() const=0;
-	virtual void pop_frame(Ref<ImageTexture> p_tex)=0;
-	virtual Image peek_frame() const=0;
-
 	virtual void set_audio_track(int p_idx) =0;
 
-	virtual void update(float p_time)=0;
+	//virtual int mix(int16_t* p_bufer,int p_frames)=0;
+
+	virtual Ref<Texture> get_texture()=0;
+	virtual void update(float p_delta)=0;
 
-	VideoStream();
+	virtual void set_mix_callback(AudioMixCallback p_callback,void *p_userdata)=0;
+	virtual int get_channels() const=0;
+	virtual int get_mix_rate() const=0;
+
+	VideoStreamPlayback();
 };
 
+
+class VideoStream : public Resource {
+
+	OBJ_TYPE( VideoStream, Resource );
+	OBJ_SAVE_TYPE( VideoStream ); //children are all saved as AudioStream, so they can be exchanged
+
+public:
+
+    virtual void set_audio_track(int p_track)=0;
+	virtual Ref<VideoStreamPlayback> instance_playback()=0;
+
+	VideoStream() {}
+};
+
+
 #endif
 
diff --git a/servers/audio/audio_mixer_sw.cpp b/servers/audio/audio_mixer_sw.cpp
index 788eec0d4a..791f31719e 100644
--- a/servers/audio/audio_mixer_sw.cpp
+++ b/servers/audio/audio_mixer_sw.cpp
@@ -765,6 +765,10 @@ AudioMixer::ChannelID AudioMixerSW::channel_alloc(RID p_sample) {
 	c.mix.increment=1;
 	//zero everything when this errors
 	for(int i=0;i<4;i++) {
+		c.mix.vol[i]=0;
+		c.mix.reverb_vol[i]=0;
+		c.mix.chorus_vol[i]=0;
+
 		c.mix.old_vol[i]=0;
 		c.mix.old_reverb_vol[i]=0;
 		c.mix.old_chorus_vol[i]=0;
diff --git a/servers/audio/audio_server_sw.cpp b/servers/audio/audio_server_sw.cpp
index 141c940615..417a582da6 100644
--- a/servers/audio/audio_server_sw.cpp
+++ b/servers/audio/audio_server_sw.cpp
@@ -920,7 +920,7 @@ float AudioServerSW::get_event_voice_global_volume_scale() const {
 
 double AudioServerSW::get_output_delay() const {
 
-	return _output_delay;
+	return _output_delay+AudioDriverSW::get_singleton()->get_latency();
 }
 
 double AudioServerSW::get_mix_time() const {
diff --git a/servers/audio/audio_server_sw.h b/servers/audio/audio_server_sw.h
index 250855a43e..c566e129c1 100644
--- a/servers/audio/audio_server_sw.h
+++ b/servers/audio/audio_server_sw.h
@@ -256,6 +256,8 @@ public:
 	virtual void unlock()=0;
 	virtual void finish()=0;
 
+	virtual float get_latency() { return 0; }
+
 
 
 
diff --git a/servers/audio_server.cpp b/servers/audio_server.cpp
index 7c4f8b185a..c155f5204a 100644
--- a/servers/audio_server.cpp
+++ b/servers/audio_server.cpp
@@ -164,6 +164,7 @@ void AudioServer::_bind_methods() {
 	BIND_CONSTANT( REVERB_HALL );
 
 	GLOBAL_DEF("audio/stream_buffering_ms",500);
+	GLOBAL_DEF("audio/video_delay_compensation_ms",300);
 
 }
 
diff --git a/servers/physics/space_sw.cpp b/servers/physics/space_sw.cpp
index a4fe1dd3fc..ba1c737530 100644
--- a/servers/physics/space_sw.cpp
+++ b/servers/physics/space_sw.cpp
@@ -1,746 +1,746 @@
-/*************************************************************************/

-/*  space_sw.cpp                                                         */

-/*************************************************************************/

-/*                       This file is part of:                           */

-/*                           GODOT ENGINE                                */

-/*                    http://www.godotengine.org                         */

-/*************************************************************************/

-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */

-/*                                                                       */

-/* Permission is hereby granted, free of charge, to any person obtaining */

-/* a copy of this software and associated documentation files (the       */

-/* "Software"), to deal in the Software without restriction, including   */

-/* without limitation the rights to use, copy, modify, merge, publish,   */

-/* distribute, sublicense, and/or sell copies of the Software, and to    */

-/* permit persons to whom the Software is furnished to do so, subject to */

-/* the following conditions:                                             */

-/*                                                                       */

-/* The above copyright notice and this permission notice shall be        */

-/* included in all copies or substantial portions of the Software.       */

-/*                                                                       */

-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */

-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */

-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/

-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */

-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */

-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */

-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */

-/*************************************************************************/

-#include "globals.h"

-#include "space_sw.h"

-#include "collision_solver_sw.h"

-#include "physics_server_sw.h"

-

-

-_FORCE_INLINE_ static bool _match_object_type_query(CollisionObjectSW *p_object, uint32_t p_layer_mask, uint32_t p_type_mask) {

-

-	if ((p_object->get_layer_mask()&p_layer_mask)==0)

-		return false;

-

-	if (p_object->get_type()==CollisionObjectSW::TYPE_AREA && !(p_type_mask&PhysicsDirectSpaceState::TYPE_MASK_AREA))

-		return false;

-

-	BodySW *body = static_cast<BodySW*>(p_object);

-

-	return (1<<body->get_mode())&p_type_mask;

-

-}

-

-

-bool PhysicsDirectSpaceStateSW::intersect_ray(const Vector3& p_from, const Vector3& p_to,RayResult &r_result,const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {

-

-

-	ERR_FAIL_COND_V(space->locked,false);

-

-	Vector3 begin,end;

-	Vector3 normal;

-	begin=p_from;

-	end=p_to;

-	normal=(end-begin).normalized();

-

-

-	int amount = space->broadphase->cull_segment(begin,end,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);

-

-

-	//todo, create another array tha references results, compute AABBs and check closest point to ray origin, sort, and stop evaluating results when beyond first collision

-

-	bool collided=false;

-	Vector3 res_point,res_normal;

-	int res_shape;

-	const CollisionObjectSW *res_obj;

-	real_t min_d=1e10;

-

-

-

-	for(int i=0;i<amount;i++) {

-

-		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))

-			continue;

-

-		if (!(static_cast<CollisionObjectSW*>(space->intersection_query_results[i])->is_ray_pickable()))

-			continue;

-

-		if (p_exclude.has( space->intersection_query_results[i]->get_self()))

-			continue;

-

-		const CollisionObjectSW *col_obj=space->intersection_query_results[i];

-

-		int shape_idx=space->intersection_query_subindex_results[i];

-		Transform inv_xform = col_obj->get_shape_inv_transform(shape_idx) * col_obj->get_inv_transform();

-

-		Vector3 local_from = inv_xform.xform(begin);

-		Vector3 local_to = inv_xform.xform(end);

-

-		const ShapeSW *shape = col_obj->get_shape(shape_idx);

-

-		Vector3 shape_point,shape_normal;

-

-

-		if (shape->intersect_segment(local_from,local_to,shape_point,shape_normal)) {

-

-

-

-			Transform xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx);

-			shape_point=xform.xform(shape_point);

-

-			real_t ld = normal.dot(shape_point);

-

-

-			if (ld<min_d) {

-

-				min_d=ld;

-				res_point=shape_point;

-				res_normal=inv_xform.basis.xform_inv(shape_normal).normalized();

-				res_shape=shape_idx;

-				res_obj=col_obj;

-				collided=true;

-			}

-		}

-

-	}

-

-	if (!collided)

-		return false;

-

-

-	r_result.collider_id=res_obj->get_instance_id();

-	if (r_result.collider_id!=0)

-		r_result.collider=ObjectDB::get_instance(r_result.collider_id);

-	else

-		r_result.collider=NULL;

-	r_result.normal=res_normal;

-	r_result.position=res_point;

-	r_result.rid=res_obj->get_self();

-	r_result.shape=res_shape;

-

-	return true;

-

-}

-

-

-int PhysicsDirectSpaceStateSW::intersect_shape(const RID& p_shape, const Transform& p_xform,float p_margin,ShapeResult *r_results,int p_result_max,const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {

-

-	if (p_result_max<=0)

-		return 0;

-

-	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);

-	ERR_FAIL_COND_V(!shape,0);

-

-	AABB aabb = p_xform.xform(shape->get_aabb());

-

-	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);

-

-	bool collided=false;

-	int cc=0;

-

-	//Transform ai = p_xform.affine_inverse();

-

-	for(int i=0;i<amount;i++) {

-

-		if (cc>=p_result_max)

-			break;

-

-		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))

-			continue;

-

-		//area cant be picked by ray (default)

-

-		if (p_exclude.has( space->intersection_query_results[i]->get_self()))

-			continue;

-

-

-		const CollisionObjectSW *col_obj=space->intersection_query_results[i];

-		int shape_idx=space->intersection_query_subindex_results[i];

-

-		if (!CollisionSolverSW::solve_static(shape,p_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), NULL,NULL,NULL,p_margin,0))

-			continue;

-

-		r_results[cc].collider_id=col_obj->get_instance_id();

-		if (r_results[cc].collider_id!=0)

-			r_results[cc].collider=ObjectDB::get_instance(r_results[cc].collider_id);

-		else

-			r_results[cc].collider=NULL;

-		r_results[cc].rid=col_obj->get_self();

-		r_results[cc].shape=shape_idx;

-

-		cc++;

-

-	}

-

-	return cc;

-

-}

-

-

-bool PhysicsDirectSpaceStateSW::cast_motion(const RID& p_shape, const Transform& p_xform,const Vector3& p_motion,float p_margin,float &p_closest_safe,float &p_closest_unsafe, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask,ShapeRestInfo *r_info) {

-

-

-

-	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);

-	ERR_FAIL_COND_V(!shape,false);

-

-	AABB aabb = p_xform.xform(shape->get_aabb());

-	aabb=aabb.merge(AABB(aabb.pos+p_motion,aabb.size)); //motion

-	aabb=aabb.grow(p_margin);

-

-	//if (p_motion!=Vector3())

-	//	print_line(p_motion);

-

-	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);

-

-	float best_safe=1;

-	float best_unsafe=1;

-

-	Transform xform_inv = p_xform.affine_inverse();

-	MotionShapeSW mshape;

-	mshape.shape=shape;

-	mshape.motion=xform_inv.basis.xform(p_motion);

-

-	bool best_first=true;

-

-	Vector3 closest_A,closest_B;

-

-	for(int i=0;i<amount;i++) {

-

-

-		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))

-			continue;

-

-		if (p_exclude.has( space->intersection_query_results[i]->get_self()))

-			continue; //ignore excluded

-

-

-		const CollisionObjectSW *col_obj=space->intersection_query_results[i];

-		int shape_idx=space->intersection_query_subindex_results[i];

-

-		Vector3 point_A,point_B;

-		Vector3 sep_axis=p_motion.normalized();

-

-		Transform col_obj_xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx);

-		//test initial overlap, does it collide if going all the way?

-		if (CollisionSolverSW::solve_distance(&mshape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,point_A,point_B,aabb,&sep_axis)) {

-			//print_line("failed motion cast (no collision)");

-			continue;

-		}

-

-

-		//test initial overlap

-#if 0

-		if (CollisionSolverSW::solve_static(shape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,NULL,NULL,&sep_axis)) {

-			print_line("failed initial cast (collision at begining)");

-			return false;

-		}

-#else

-		sep_axis=p_motion.normalized();

-

-		if (!CollisionSolverSW::solve_distance(shape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,point_A,point_B,aabb,&sep_axis)) {

-			//print_line("failed motion cast (no collision)");

-			return false;

-		}

-#endif

-

-

-		//just do kinematic solving

-		float low=0;

-		float hi=1;

-		Vector3 mnormal=p_motion.normalized();

-

-		for(int i=0;i<8;i++) { //steps should be customizable..

-

-			Transform xfa = p_xform;

-			float ofs = (low+hi)*0.5;

-

-			Vector3 sep=mnormal; //important optimization for this to work fast enough

-

-			mshape.motion=xform_inv.basis.xform(p_motion*ofs);

-

-			Vector3 lA,lB;

-

-			bool collided = !CollisionSolverSW::solve_distance(&mshape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,lA,lB,aabb,&sep);

-

-			if (collided) {

-

-				//print_line(itos(i)+": "+rtos(ofs));

-				hi=ofs;

-			} else {

-

-				point_A=lA;

-				point_B=lB;

-				low=ofs;

-			}

-		}

-

-		if (low<best_safe) {

-			best_first=true; //force reset

-			best_safe=low;

-			best_unsafe=hi;

-		}

-

-		if (r_info && (best_first || (point_A.distance_squared_to(point_B) < closest_A.distance_squared_to(closest_B) && low<=best_safe))) {

-			closest_A=point_A;

-			closest_B=point_B;

-			r_info->collider_id=col_obj->get_instance_id();

-			r_info->rid=col_obj->get_self();

-			r_info->shape=shape_idx;

-			r_info->point=closest_B;

-			r_info->normal=(closest_A-closest_B).normalized();

-			best_first=false;

-			if (col_obj->get_type()==CollisionObjectSW::TYPE_BODY) {

-				const BodySW *body=static_cast<const BodySW*>(col_obj);

-				r_info->linear_velocity= body->get_linear_velocity() + (body->get_angular_velocity()).cross(body->get_transform().origin - closest_B);

-			}

-

-		}

-

-

-	}

-

-	p_closest_safe=best_safe;

-	p_closest_unsafe=best_unsafe;	

-

-	return true;

-}

-

-bool PhysicsDirectSpaceStateSW::collide_shape(RID p_shape, const Transform& p_shape_xform,float p_margin,Vector3 *r_results,int p_result_max,int &r_result_count, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask){

-

-	if (p_result_max<=0)

-		return 0;

-

-	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);

-	ERR_FAIL_COND_V(!shape,0);

-

-	AABB aabb = p_shape_xform.xform(shape->get_aabb());

-	aabb=aabb.grow(p_margin);

-

-	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);

-

-	bool collided=false;

-	int cc=0;

-	r_result_count=0;

-

-	PhysicsServerSW::CollCbkData cbk;

-	cbk.max=p_result_max;

-	cbk.amount=0;

-	cbk.ptr=r_results;

-	CollisionSolverSW::CallbackResult cbkres=NULL;

-

-	PhysicsServerSW::CollCbkData *cbkptr=NULL;

-	if (p_result_max>0) {

-		cbkptr=&cbk;

-		cbkres=PhysicsServerSW::_shape_col_cbk;

-	}

-

-

-	for(int i=0;i<amount;i++) {

-

-		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))

-			continue;

-

-		const CollisionObjectSW *col_obj=space->intersection_query_results[i];

-		int shape_idx=space->intersection_query_subindex_results[i];

-

-		if (p_exclude.has( col_obj->get_self() )) {

-			continue;

-		}

-

-		//print_line("AGAINST: "+itos(col_obj->get_self().get_id())+":"+itos(shape_idx));

-		//print_line("THE ABBB: "+(col_obj->get_transform() * col_obj->get_shape_transform(shape_idx)).xform(col_obj->get_shape(shape_idx)->get_aabb()));

-

-		if (CollisionSolverSW::solve_static(shape,p_shape_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx),cbkres,cbkptr,NULL,p_margin)) {

-			collided=true;

-		}

-

-	}

-

-	r_result_count=cbk.amount;

-

-	return collided;

-

-}

-

-

-struct _RestCallbackData {

-

-	const CollisionObjectSW *object;

-	const CollisionObjectSW *best_object;

-	int shape;

-	int best_shape;

-	Vector3 best_contact;

-	Vector3 best_normal;

-	float best_len;

-};

-

-static void _rest_cbk_result(const Vector3& p_point_A,const Vector3& p_point_B,void *p_userdata) {

-

-

-	_RestCallbackData *rd=(_RestCallbackData*)p_userdata;

-

-	Vector3 contact_rel = p_point_B - p_point_A;

-	float len = contact_rel.length();

-	if (len <= rd->best_len)

-		return;

-

-	rd->best_len=len;

-	rd->best_contact=p_point_B;

-	rd->best_normal=contact_rel/len;

-	rd->best_object=rd->object;

-	rd->best_shape=rd->shape;

-

-}

-bool PhysicsDirectSpaceStateSW::rest_info(RID p_shape, const Transform& p_shape_xform,float p_margin,ShapeRestInfo *r_info, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {

-

-

-	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);

-	ERR_FAIL_COND_V(!shape,0);

-

-	AABB aabb = p_shape_xform.xform(shape->get_aabb());

-	aabb=aabb.grow(p_margin);

-

-	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);

-

-	_RestCallbackData rcd;

-	rcd.best_len=0;

-	rcd.best_object=NULL;

-	rcd.best_shape=0;

-

-	for(int i=0;i<amount;i++) {

-

-

-		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))

-			continue;

-

-		const CollisionObjectSW *col_obj=space->intersection_query_results[i];

-		int shape_idx=space->intersection_query_subindex_results[i];

-

-		if (p_exclude.has( col_obj->get_self() ))

-			continue;

-

-		rcd.object=col_obj;

-		rcd.shape=shape_idx;

-		bool sc = CollisionSolverSW::solve_static(shape,p_shape_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx),_rest_cbk_result,&rcd,NULL,p_margin);

-		if (!sc)

-			continue;

-

-

-	}

-

-	if (rcd.best_len==0)

-		return false;

-

-	r_info->collider_id=rcd.best_object->get_instance_id();

-	r_info->shape=rcd.best_shape;

-	r_info->normal=rcd.best_normal;

-	r_info->point=rcd.best_contact;

-	r_info->rid=rcd.best_object->get_self();

-	if (rcd.best_object->get_type()==CollisionObjectSW::TYPE_BODY) {

-

-		const BodySW *body = static_cast<const BodySW*>(rcd.best_object);

-		Vector3 rel_vec = r_info->point-body->get_transform().get_origin();

-		r_info->linear_velocity = body->get_linear_velocity() +

-				(body->get_angular_velocity()).cross(body->get_transform().origin-rcd.best_contact);// * mPos);

-

-

-	} else {

-		r_info->linear_velocity=Vector3();

-	}

-

-	return true;

-}

-

-

-PhysicsDirectSpaceStateSW::PhysicsDirectSpaceStateSW() {

-

-

-	space=NULL;

-}

-

-

-////////////////////////////////////////////////////////////////////////////////////////////////////////////

-

-

-

-

-

-

-

-

-

-

-void* SpaceSW::_broadphase_pair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_self) {

-

-	CollisionObjectSW::Type type_A=A->get_type();

-	CollisionObjectSW::Type type_B=B->get_type();

-	if (type_A>type_B) {

-

-		SWAP(A,B);

-		SWAP(p_subindex_A,p_subindex_B);

-		SWAP(type_A,type_B);

-	}

-

-	SpaceSW *self = (SpaceSW*)p_self;

-

-	self->collision_pairs++;

-

-	if (type_A==CollisionObjectSW::TYPE_AREA) {

-

-		AreaSW *area=static_cast<AreaSW*>(A);

-		if (type_B==CollisionObjectSW::TYPE_AREA) {

-

-			AreaSW *area_b=static_cast<AreaSW*>(B);

-			Area2PairSW *area2_pair = memnew(Area2PairSW(area_b,p_subindex_B,area,p_subindex_A) );

-			return area2_pair;

-		} else {

-

-			BodySW *body=static_cast<BodySW*>(B);

-			AreaPairSW *area_pair = memnew(AreaPairSW(body,p_subindex_B,area,p_subindex_A) );

-			return area_pair;

-		}

-	} else {

-

-

-		BodyPairSW *b = memnew( BodyPairSW((BodySW*)A,p_subindex_A,(BodySW*)B,p_subindex_B) );

-		return b;

-

-	}

-

-	return NULL;

-}

-

-void SpaceSW::_broadphase_unpair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_data,void *p_self) {

-

-

-

-	SpaceSW *self = (SpaceSW*)p_self;

-	self->collision_pairs--;

-	ConstraintSW *c = (ConstraintSW*)p_data;

-	memdelete(c);

-}

-

-

-const SelfList<BodySW>::List& SpaceSW::get_active_body_list() const {

-

-	return active_list;

-}

-void SpaceSW::body_add_to_active_list(SelfList<BodySW>* p_body) {

-

-	active_list.add(p_body);

-}

-void SpaceSW::body_remove_from_active_list(SelfList<BodySW>* p_body) {

-

-	active_list.remove(p_body);

-

-}

-

-void SpaceSW::body_add_to_inertia_update_list(SelfList<BodySW>* p_body) {

-

-

-	inertia_update_list.add(p_body);

-}

-

-void SpaceSW::body_remove_from_inertia_update_list(SelfList<BodySW>* p_body) {

-

-	inertia_update_list.remove(p_body);

-}

-

-BroadPhaseSW *SpaceSW::get_broadphase() {

-

-	return broadphase;

-}

-

-void SpaceSW::add_object(CollisionObjectSW *p_object) {

-

-	ERR_FAIL_COND( objects.has(p_object) );

-	objects.insert(p_object);

-}

-

-void SpaceSW::remove_object(CollisionObjectSW *p_object) {

-

-	ERR_FAIL_COND( !objects.has(p_object) );

-	objects.erase(p_object);

-}

-

-const Set<CollisionObjectSW*> &SpaceSW::get_objects() const {

-

-	return objects;

-}

-

-void SpaceSW::body_add_to_state_query_list(SelfList<BodySW>* p_body) {

-

-	state_query_list.add(p_body);

-}

-void SpaceSW::body_remove_from_state_query_list(SelfList<BodySW>* p_body) {

-

-	state_query_list.remove(p_body);

-}

-

-void SpaceSW::area_add_to_monitor_query_list(SelfList<AreaSW>* p_area) {

-

-	monitor_query_list.add(p_area);

-}

-void SpaceSW::area_remove_from_monitor_query_list(SelfList<AreaSW>* p_area) {

-

-	monitor_query_list.remove(p_area);

-}

-

-void SpaceSW::area_add_to_moved_list(SelfList<AreaSW>* p_area) {

-

-	area_moved_list.add(p_area);

-}

-

-void SpaceSW::area_remove_from_moved_list(SelfList<AreaSW>* p_area) {

-

-	area_moved_list.remove(p_area);

-}

-

-const SelfList<AreaSW>::List& SpaceSW::get_moved_area_list() const {

-

-	return area_moved_list;

-}

-

-

-

-

-void SpaceSW::call_queries() {

-

-	while(state_query_list.first()) {

-

-		BodySW * b = state_query_list.first()->self();

-		b->call_queries();

-		state_query_list.remove(state_query_list.first());

-	}

-

-	while(monitor_query_list.first()) {

-

-		AreaSW * a = monitor_query_list.first()->self();

-		a->call_queries();

-		monitor_query_list.remove(monitor_query_list.first());

-	}

-

-}

-

-void SpaceSW::setup() {

-

-	contact_debug_count=0;

-	while(inertia_update_list.first()) {

-		inertia_update_list.first()->self()->update_inertias();

-		inertia_update_list.remove(inertia_update_list.first());

-	}

-

-

-}

-

-void SpaceSW::update() {

-

-

-	broadphase->update();

-

-}

-

-

-void SpaceSW::set_param(PhysicsServer::SpaceParameter p_param, real_t p_value) {

-

-	switch(p_param) {

-

-		case PhysicsServer::SPACE_PARAM_CONTACT_RECYCLE_RADIUS: contact_recycle_radius=p_value; break;

-		case PhysicsServer::SPACE_PARAM_CONTACT_MAX_SEPARATION: contact_max_separation=p_value; break;

-		case PhysicsServer::SPACE_PARAM_BODY_MAX_ALLOWED_PENETRATION: contact_max_allowed_penetration=p_value; break;

-		case PhysicsServer::SPACE_PARAM_BODY_LINEAR_VELOCITY_SLEEP_TRESHOLD: body_linear_velocity_sleep_threshold=p_value; break;

-		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_SLEEP_TRESHOLD: body_angular_velocity_sleep_threshold=p_value; break;

-		case PhysicsServer::SPACE_PARAM_BODY_TIME_TO_SLEEP: body_time_to_sleep=p_value; break;

-		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_DAMP_RATIO: body_angular_velocity_damp_ratio=p_value; break;

-		case PhysicsServer::SPACE_PARAM_CONSTRAINT_DEFAULT_BIAS: constraint_bias=p_value; break;

-	}

-}

-

-real_t SpaceSW::get_param(PhysicsServer::SpaceParameter p_param) const {

-

-	switch(p_param) {

-

-		case PhysicsServer::SPACE_PARAM_CONTACT_RECYCLE_RADIUS: return contact_recycle_radius;

-		case PhysicsServer::SPACE_PARAM_CONTACT_MAX_SEPARATION: return contact_max_separation;

-		case PhysicsServer::SPACE_PARAM_BODY_MAX_ALLOWED_PENETRATION: return contact_max_allowed_penetration;

-		case PhysicsServer::SPACE_PARAM_BODY_LINEAR_VELOCITY_SLEEP_TRESHOLD: return body_linear_velocity_sleep_threshold;

-		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_SLEEP_TRESHOLD: return body_angular_velocity_sleep_threshold;

-		case PhysicsServer::SPACE_PARAM_BODY_TIME_TO_SLEEP: return body_time_to_sleep;

-		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_DAMP_RATIO: return body_angular_velocity_damp_ratio;

-		case PhysicsServer::SPACE_PARAM_CONSTRAINT_DEFAULT_BIAS: return constraint_bias;

-	}

-	return 0;

-}

-

-void SpaceSW::lock() {

-

-	locked=true;

-}

-

-void SpaceSW::unlock() {

-

-	locked=false;

-}

-

-bool SpaceSW::is_locked() const {

-

-	return locked;

-}

-

-PhysicsDirectSpaceStateSW *SpaceSW::get_direct_state() {

-

-	return direct_access;

-}

-

-SpaceSW::SpaceSW() {

-

-	collision_pairs=0;

-	active_objects=0;

-	island_count=0;

-	contact_debug_count=0;

-

-	locked=false;

-	contact_recycle_radius=0.01;

-	contact_max_separation=0.05;

-	contact_max_allowed_penetration= 0.01;

-

-	constraint_bias = 0.01;

-	body_linear_velocity_sleep_threshold=GLOBAL_DEF("physics/sleep_threshold_linear",0.1);

-	body_angular_velocity_sleep_threshold=GLOBAL_DEF("physics/sleep_threshold_angular", (8.0 / 180.0 * Math_PI) );

-	body_time_to_sleep=0.5;

-	body_angular_velocity_damp_ratio=10;

-

-

-	broadphase = BroadPhaseSW::create_func();

-	broadphase->set_pair_callback(_broadphase_pair,this);

-	broadphase->set_unpair_callback(_broadphase_unpair,this);

-	area=NULL;

-

-	direct_access = memnew( PhysicsDirectSpaceStateSW );

-	direct_access->space=this;

-}

-

-SpaceSW::~SpaceSW() {

-

-	memdelete(broadphase);

-	memdelete( direct_access );

-}

-

-

-

+/*************************************************************************/
+/*  space_sw.cpp                                                         */
+/*************************************************************************/
+/*                       This file is part of:                           */
+/*                           GODOT ENGINE                                */
+/*                    http://www.godotengine.org                         */
+/*************************************************************************/
+/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
+/*                                                                       */
+/* Permission is hereby granted, free of charge, to any person obtaining */
+/* a copy of this software and associated documentation files (the       */
+/* "Software"), to deal in the Software without restriction, including   */
+/* without limitation the rights to use, copy, modify, merge, publish,   */
+/* distribute, sublicense, and/or sell copies of the Software, and to    */
+/* permit persons to whom the Software is furnished to do so, subject to */
+/* the following conditions:                                             */
+/*                                                                       */
+/* The above copyright notice and this permission notice shall be        */
+/* included in all copies or substantial portions of the Software.       */
+/*                                                                       */
+/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
+/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
+/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
+/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
+/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
+/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
+/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
+/*************************************************************************/
+#include "globals.h"
+#include "space_sw.h"
+#include "collision_solver_sw.h"
+#include "physics_server_sw.h"
+
+
+_FORCE_INLINE_ static bool _match_object_type_query(CollisionObjectSW *p_object, uint32_t p_layer_mask, uint32_t p_type_mask) {
+
+	if (p_object->get_type()==CollisionObjectSW::TYPE_AREA)
+		return p_type_mask&PhysicsDirectSpaceState::TYPE_MASK_AREA;
+
+	if ((p_object->get_layer_mask()&p_layer_mask)==0)
+		return false;
+
+	BodySW *body = static_cast<BodySW*>(p_object);
+
+	return (1<<body->get_mode())&p_type_mask;
+
+}
+
+
+bool PhysicsDirectSpaceStateSW::intersect_ray(const Vector3& p_from, const Vector3& p_to,RayResult &r_result,const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {
+
+
+	ERR_FAIL_COND_V(space->locked,false);
+
+	Vector3 begin,end;
+	Vector3 normal;
+	begin=p_from;
+	end=p_to;
+	normal=(end-begin).normalized();
+
+
+	int amount = space->broadphase->cull_segment(begin,end,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);
+
+
+	//todo, create another array tha references results, compute AABBs and check closest point to ray origin, sort, and stop evaluating results when beyond first collision
+
+	bool collided=false;
+	Vector3 res_point,res_normal;
+	int res_shape;
+	const CollisionObjectSW *res_obj;
+	real_t min_d=1e10;
+
+
+
+	for(int i=0;i<amount;i++) {
+
+		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))
+			continue;
+
+		if (!(static_cast<CollisionObjectSW*>(space->intersection_query_results[i])->is_ray_pickable()))
+			continue;
+
+		if (p_exclude.has( space->intersection_query_results[i]->get_self()))
+			continue;
+
+		const CollisionObjectSW *col_obj=space->intersection_query_results[i];
+
+		int shape_idx=space->intersection_query_subindex_results[i];
+		Transform inv_xform = col_obj->get_shape_inv_transform(shape_idx) * col_obj->get_inv_transform();
+
+		Vector3 local_from = inv_xform.xform(begin);
+		Vector3 local_to = inv_xform.xform(end);
+
+		const ShapeSW *shape = col_obj->get_shape(shape_idx);
+
+		Vector3 shape_point,shape_normal;
+
+
+		if (shape->intersect_segment(local_from,local_to,shape_point,shape_normal)) {
+
+
+
+			Transform xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx);
+			shape_point=xform.xform(shape_point);
+
+			real_t ld = normal.dot(shape_point);
+
+
+			if (ld<min_d) {
+
+				min_d=ld;
+				res_point=shape_point;
+				res_normal=inv_xform.basis.xform_inv(shape_normal).normalized();
+				res_shape=shape_idx;
+				res_obj=col_obj;
+				collided=true;
+			}
+		}
+
+	}
+
+	if (!collided)
+		return false;
+
+
+	r_result.collider_id=res_obj->get_instance_id();
+	if (r_result.collider_id!=0)
+		r_result.collider=ObjectDB::get_instance(r_result.collider_id);
+	else
+		r_result.collider=NULL;
+	r_result.normal=res_normal;
+	r_result.position=res_point;
+	r_result.rid=res_obj->get_self();
+	r_result.shape=res_shape;
+
+	return true;
+
+}
+
+
+int PhysicsDirectSpaceStateSW::intersect_shape(const RID& p_shape, const Transform& p_xform,float p_margin,ShapeResult *r_results,int p_result_max,const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {
+
+	if (p_result_max<=0)
+		return 0;
+
+	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);
+	ERR_FAIL_COND_V(!shape,0);
+
+	AABB aabb = p_xform.xform(shape->get_aabb());
+
+	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);
+
+	bool collided=false;
+	int cc=0;
+
+	//Transform ai = p_xform.affine_inverse();
+
+	for(int i=0;i<amount;i++) {
+
+		if (cc>=p_result_max)
+			break;
+
+		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))
+			continue;
+
+		//area cant be picked by ray (default)
+
+		if (p_exclude.has( space->intersection_query_results[i]->get_self()))
+			continue;
+
+
+		const CollisionObjectSW *col_obj=space->intersection_query_results[i];
+		int shape_idx=space->intersection_query_subindex_results[i];
+
+		if (!CollisionSolverSW::solve_static(shape,p_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), NULL,NULL,NULL,p_margin,0))
+			continue;
+
+		r_results[cc].collider_id=col_obj->get_instance_id();
+		if (r_results[cc].collider_id!=0)
+			r_results[cc].collider=ObjectDB::get_instance(r_results[cc].collider_id);
+		else
+			r_results[cc].collider=NULL;
+		r_results[cc].rid=col_obj->get_self();
+		r_results[cc].shape=shape_idx;
+
+		cc++;
+
+	}
+
+	return cc;
+
+}
+
+
+bool PhysicsDirectSpaceStateSW::cast_motion(const RID& p_shape, const Transform& p_xform,const Vector3& p_motion,float p_margin,float &p_closest_safe,float &p_closest_unsafe, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask,ShapeRestInfo *r_info) {
+
+
+
+	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);
+	ERR_FAIL_COND_V(!shape,false);
+
+	AABB aabb = p_xform.xform(shape->get_aabb());
+	aabb=aabb.merge(AABB(aabb.pos+p_motion,aabb.size)); //motion
+	aabb=aabb.grow(p_margin);
+
+	//if (p_motion!=Vector3())
+	//	print_line(p_motion);
+
+	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);
+
+	float best_safe=1;
+	float best_unsafe=1;
+
+	Transform xform_inv = p_xform.affine_inverse();
+	MotionShapeSW mshape;
+	mshape.shape=shape;
+	mshape.motion=xform_inv.basis.xform(p_motion);
+
+	bool best_first=true;
+
+	Vector3 closest_A,closest_B;
+
+	for(int i=0;i<amount;i++) {
+
+
+		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))
+			continue;
+
+		if (p_exclude.has( space->intersection_query_results[i]->get_self()))
+			continue; //ignore excluded
+
+
+		const CollisionObjectSW *col_obj=space->intersection_query_results[i];
+		int shape_idx=space->intersection_query_subindex_results[i];
+
+		Vector3 point_A,point_B;
+		Vector3 sep_axis=p_motion.normalized();
+
+		Transform col_obj_xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx);
+		//test initial overlap, does it collide if going all the way?
+		if (CollisionSolverSW::solve_distance(&mshape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,point_A,point_B,aabb,&sep_axis)) {
+			//print_line("failed motion cast (no collision)");
+			continue;
+		}
+
+
+		//test initial overlap
+#if 0
+		if (CollisionSolverSW::solve_static(shape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,NULL,NULL,&sep_axis)) {
+			print_line("failed initial cast (collision at begining)");
+			return false;
+		}
+#else
+		sep_axis=p_motion.normalized();
+
+		if (!CollisionSolverSW::solve_distance(shape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,point_A,point_B,aabb,&sep_axis)) {
+			//print_line("failed motion cast (no collision)");
+			return false;
+		}
+#endif
+
+
+		//just do kinematic solving
+		float low=0;
+		float hi=1;
+		Vector3 mnormal=p_motion.normalized();
+
+		for(int i=0;i<8;i++) { //steps should be customizable..
+
+			Transform xfa = p_xform;
+			float ofs = (low+hi)*0.5;
+
+			Vector3 sep=mnormal; //important optimization for this to work fast enough
+
+			mshape.motion=xform_inv.basis.xform(p_motion*ofs);
+
+			Vector3 lA,lB;
+
+			bool collided = !CollisionSolverSW::solve_distance(&mshape,p_xform,col_obj->get_shape(shape_idx),col_obj_xform,lA,lB,aabb,&sep);
+
+			if (collided) {
+
+				//print_line(itos(i)+": "+rtos(ofs));
+				hi=ofs;
+			} else {
+
+				point_A=lA;
+				point_B=lB;
+				low=ofs;
+			}
+		}
+
+		if (low<best_safe) {
+			best_first=true; //force reset
+			best_safe=low;
+			best_unsafe=hi;
+		}
+
+		if (r_info && (best_first || (point_A.distance_squared_to(point_B) < closest_A.distance_squared_to(closest_B) && low<=best_safe))) {
+			closest_A=point_A;
+			closest_B=point_B;
+			r_info->collider_id=col_obj->get_instance_id();
+			r_info->rid=col_obj->get_self();
+			r_info->shape=shape_idx;
+			r_info->point=closest_B;
+			r_info->normal=(closest_A-closest_B).normalized();
+			best_first=false;
+			if (col_obj->get_type()==CollisionObjectSW::TYPE_BODY) {
+				const BodySW *body=static_cast<const BodySW*>(col_obj);
+				r_info->linear_velocity= body->get_linear_velocity() + (body->get_angular_velocity()).cross(body->get_transform().origin - closest_B);
+			}
+
+		}
+
+
+	}
+
+	p_closest_safe=best_safe;
+	p_closest_unsafe=best_unsafe;	
+
+	return true;
+}
+
+bool PhysicsDirectSpaceStateSW::collide_shape(RID p_shape, const Transform& p_shape_xform,float p_margin,Vector3 *r_results,int p_result_max,int &r_result_count, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask){
+
+	if (p_result_max<=0)
+		return 0;
+
+	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);
+	ERR_FAIL_COND_V(!shape,0);
+
+	AABB aabb = p_shape_xform.xform(shape->get_aabb());
+	aabb=aabb.grow(p_margin);
+
+	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);
+
+	bool collided=false;
+	int cc=0;
+	r_result_count=0;
+
+	PhysicsServerSW::CollCbkData cbk;
+	cbk.max=p_result_max;
+	cbk.amount=0;
+	cbk.ptr=r_results;
+	CollisionSolverSW::CallbackResult cbkres=NULL;
+
+	PhysicsServerSW::CollCbkData *cbkptr=NULL;
+	if (p_result_max>0) {
+		cbkptr=&cbk;
+		cbkres=PhysicsServerSW::_shape_col_cbk;
+	}
+
+
+	for(int i=0;i<amount;i++) {
+
+		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))
+			continue;
+
+		const CollisionObjectSW *col_obj=space->intersection_query_results[i];
+		int shape_idx=space->intersection_query_subindex_results[i];
+
+		if (p_exclude.has( col_obj->get_self() )) {
+			continue;
+		}
+
+		//print_line("AGAINST: "+itos(col_obj->get_self().get_id())+":"+itos(shape_idx));
+		//print_line("THE ABBB: "+(col_obj->get_transform() * col_obj->get_shape_transform(shape_idx)).xform(col_obj->get_shape(shape_idx)->get_aabb()));
+
+		if (CollisionSolverSW::solve_static(shape,p_shape_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx),cbkres,cbkptr,NULL,p_margin)) {
+			collided=true;
+		}
+
+	}
+
+	r_result_count=cbk.amount;
+
+	return collided;
+
+}
+
+
+struct _RestCallbackData {
+
+	const CollisionObjectSW *object;
+	const CollisionObjectSW *best_object;
+	int shape;
+	int best_shape;
+	Vector3 best_contact;
+	Vector3 best_normal;
+	float best_len;
+};
+
+static void _rest_cbk_result(const Vector3& p_point_A,const Vector3& p_point_B,void *p_userdata) {
+
+
+	_RestCallbackData *rd=(_RestCallbackData*)p_userdata;
+
+	Vector3 contact_rel = p_point_B - p_point_A;
+	float len = contact_rel.length();
+	if (len <= rd->best_len)
+		return;
+
+	rd->best_len=len;
+	rd->best_contact=p_point_B;
+	rd->best_normal=contact_rel/len;
+	rd->best_object=rd->object;
+	rd->best_shape=rd->shape;
+
+}
+bool PhysicsDirectSpaceStateSW::rest_info(RID p_shape, const Transform& p_shape_xform,float p_margin,ShapeRestInfo *r_info, const Set<RID>& p_exclude,uint32_t p_layer_mask,uint32_t p_object_type_mask) {
+
+
+	ShapeSW *shape = static_cast<PhysicsServerSW*>(PhysicsServer::get_singleton())->shape_owner.get(p_shape);
+	ERR_FAIL_COND_V(!shape,0);
+
+	AABB aabb = p_shape_xform.xform(shape->get_aabb());
+	aabb=aabb.grow(p_margin);
+
+	int amount = space->broadphase->cull_aabb(aabb,space->intersection_query_results,SpaceSW::INTERSECTION_QUERY_MAX,space->intersection_query_subindex_results);
+
+	_RestCallbackData rcd;
+	rcd.best_len=0;
+	rcd.best_object=NULL;
+	rcd.best_shape=0;
+
+	for(int i=0;i<amount;i++) {
+
+
+		if (!_match_object_type_query(space->intersection_query_results[i],p_layer_mask,p_object_type_mask))
+			continue;
+
+		const CollisionObjectSW *col_obj=space->intersection_query_results[i];
+		int shape_idx=space->intersection_query_subindex_results[i];
+
+		if (p_exclude.has( col_obj->get_self() ))
+			continue;
+
+		rcd.object=col_obj;
+		rcd.shape=shape_idx;
+		bool sc = CollisionSolverSW::solve_static(shape,p_shape_xform,col_obj->get_shape(shape_idx),col_obj->get_transform() * col_obj->get_shape_transform(shape_idx),_rest_cbk_result,&rcd,NULL,p_margin);
+		if (!sc)
+			continue;
+
+
+	}
+
+	if (rcd.best_len==0)
+		return false;
+
+	r_info->collider_id=rcd.best_object->get_instance_id();
+	r_info->shape=rcd.best_shape;
+	r_info->normal=rcd.best_normal;
+	r_info->point=rcd.best_contact;
+	r_info->rid=rcd.best_object->get_self();
+	if (rcd.best_object->get_type()==CollisionObjectSW::TYPE_BODY) {
+
+		const BodySW *body = static_cast<const BodySW*>(rcd.best_object);
+		Vector3 rel_vec = r_info->point-body->get_transform().get_origin();
+		r_info->linear_velocity = body->get_linear_velocity() +
+				(body->get_angular_velocity()).cross(body->get_transform().origin-rcd.best_contact);// * mPos);
+
+
+	} else {
+		r_info->linear_velocity=Vector3();
+	}
+
+	return true;
+}
+
+
+PhysicsDirectSpaceStateSW::PhysicsDirectSpaceStateSW() {
+
+
+	space=NULL;
+}
+
+
+////////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+
+
+
+
+
+
+
+
+
+void* SpaceSW::_broadphase_pair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_self) {
+
+	CollisionObjectSW::Type type_A=A->get_type();
+	CollisionObjectSW::Type type_B=B->get_type();
+	if (type_A>type_B) {
+
+		SWAP(A,B);
+		SWAP(p_subindex_A,p_subindex_B);
+		SWAP(type_A,type_B);
+	}
+
+	SpaceSW *self = (SpaceSW*)p_self;
+
+	self->collision_pairs++;
+
+	if (type_A==CollisionObjectSW::TYPE_AREA) {
+
+		AreaSW *area=static_cast<AreaSW*>(A);
+		if (type_B==CollisionObjectSW::TYPE_AREA) {
+
+			AreaSW *area_b=static_cast<AreaSW*>(B);
+			Area2PairSW *area2_pair = memnew(Area2PairSW(area_b,p_subindex_B,area,p_subindex_A) );
+			return area2_pair;
+		} else {
+
+			BodySW *body=static_cast<BodySW*>(B);
+			AreaPairSW *area_pair = memnew(AreaPairSW(body,p_subindex_B,area,p_subindex_A) );
+			return area_pair;
+		}
+	} else {
+
+
+		BodyPairSW *b = memnew( BodyPairSW((BodySW*)A,p_subindex_A,(BodySW*)B,p_subindex_B) );
+		return b;
+
+	}
+
+	return NULL;
+}
+
+void SpaceSW::_broadphase_unpair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_data,void *p_self) {
+
+
+
+	SpaceSW *self = (SpaceSW*)p_self;
+	self->collision_pairs--;
+	ConstraintSW *c = (ConstraintSW*)p_data;
+	memdelete(c);
+}
+
+
+const SelfList<BodySW>::List& SpaceSW::get_active_body_list() const {
+
+	return active_list;
+}
+void SpaceSW::body_add_to_active_list(SelfList<BodySW>* p_body) {
+
+	active_list.add(p_body);
+}
+void SpaceSW::body_remove_from_active_list(SelfList<BodySW>* p_body) {
+
+	active_list.remove(p_body);
+
+}
+
+void SpaceSW::body_add_to_inertia_update_list(SelfList<BodySW>* p_body) {
+
+
+	inertia_update_list.add(p_body);
+}
+
+void SpaceSW::body_remove_from_inertia_update_list(SelfList<BodySW>* p_body) {
+
+	inertia_update_list.remove(p_body);
+}
+
+BroadPhaseSW *SpaceSW::get_broadphase() {
+
+	return broadphase;
+}
+
+void SpaceSW::add_object(CollisionObjectSW *p_object) {
+
+	ERR_FAIL_COND( objects.has(p_object) );
+	objects.insert(p_object);
+}
+
+void SpaceSW::remove_object(CollisionObjectSW *p_object) {
+
+	ERR_FAIL_COND( !objects.has(p_object) );
+	objects.erase(p_object);
+}
+
+const Set<CollisionObjectSW*> &SpaceSW::get_objects() const {
+
+	return objects;
+}
+
+void SpaceSW::body_add_to_state_query_list(SelfList<BodySW>* p_body) {
+
+	state_query_list.add(p_body);
+}
+void SpaceSW::body_remove_from_state_query_list(SelfList<BodySW>* p_body) {
+
+	state_query_list.remove(p_body);
+}
+
+void SpaceSW::area_add_to_monitor_query_list(SelfList<AreaSW>* p_area) {
+
+	monitor_query_list.add(p_area);
+}
+void SpaceSW::area_remove_from_monitor_query_list(SelfList<AreaSW>* p_area) {
+
+	monitor_query_list.remove(p_area);
+}
+
+void SpaceSW::area_add_to_moved_list(SelfList<AreaSW>* p_area) {
+
+	area_moved_list.add(p_area);
+}
+
+void SpaceSW::area_remove_from_moved_list(SelfList<AreaSW>* p_area) {
+
+	area_moved_list.remove(p_area);
+}
+
+const SelfList<AreaSW>::List& SpaceSW::get_moved_area_list() const {
+
+	return area_moved_list;
+}
+
+
+
+
+void SpaceSW::call_queries() {
+
+	while(state_query_list.first()) {
+
+		BodySW * b = state_query_list.first()->self();
+		b->call_queries();
+		state_query_list.remove(state_query_list.first());
+	}
+
+	while(monitor_query_list.first()) {
+
+		AreaSW * a = monitor_query_list.first()->self();
+		a->call_queries();
+		monitor_query_list.remove(monitor_query_list.first());
+	}
+
+}
+
+void SpaceSW::setup() {
+
+	contact_debug_count=0;
+	while(inertia_update_list.first()) {
+		inertia_update_list.first()->self()->update_inertias();
+		inertia_update_list.remove(inertia_update_list.first());
+	}
+
+
+}
+
+void SpaceSW::update() {
+
+
+	broadphase->update();
+
+}
+
+
+void SpaceSW::set_param(PhysicsServer::SpaceParameter p_param, real_t p_value) {
+
+	switch(p_param) {
+
+		case PhysicsServer::SPACE_PARAM_CONTACT_RECYCLE_RADIUS: contact_recycle_radius=p_value; break;
+		case PhysicsServer::SPACE_PARAM_CONTACT_MAX_SEPARATION: contact_max_separation=p_value; break;
+		case PhysicsServer::SPACE_PARAM_BODY_MAX_ALLOWED_PENETRATION: contact_max_allowed_penetration=p_value; break;
+		case PhysicsServer::SPACE_PARAM_BODY_LINEAR_VELOCITY_SLEEP_TRESHOLD: body_linear_velocity_sleep_threshold=p_value; break;
+		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_SLEEP_TRESHOLD: body_angular_velocity_sleep_threshold=p_value; break;
+		case PhysicsServer::SPACE_PARAM_BODY_TIME_TO_SLEEP: body_time_to_sleep=p_value; break;
+		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_DAMP_RATIO: body_angular_velocity_damp_ratio=p_value; break;
+		case PhysicsServer::SPACE_PARAM_CONSTRAINT_DEFAULT_BIAS: constraint_bias=p_value; break;
+	}
+}
+
+real_t SpaceSW::get_param(PhysicsServer::SpaceParameter p_param) const {
+
+	switch(p_param) {
+
+		case PhysicsServer::SPACE_PARAM_CONTACT_RECYCLE_RADIUS: return contact_recycle_radius;
+		case PhysicsServer::SPACE_PARAM_CONTACT_MAX_SEPARATION: return contact_max_separation;
+		case PhysicsServer::SPACE_PARAM_BODY_MAX_ALLOWED_PENETRATION: return contact_max_allowed_penetration;
+		case PhysicsServer::SPACE_PARAM_BODY_LINEAR_VELOCITY_SLEEP_TRESHOLD: return body_linear_velocity_sleep_threshold;
+		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_SLEEP_TRESHOLD: return body_angular_velocity_sleep_threshold;
+		case PhysicsServer::SPACE_PARAM_BODY_TIME_TO_SLEEP: return body_time_to_sleep;
+		case PhysicsServer::SPACE_PARAM_BODY_ANGULAR_VELOCITY_DAMP_RATIO: return body_angular_velocity_damp_ratio;
+		case PhysicsServer::SPACE_PARAM_CONSTRAINT_DEFAULT_BIAS: return constraint_bias;
+	}
+	return 0;
+}
+
+void SpaceSW::lock() {
+
+	locked=true;
+}
+
+void SpaceSW::unlock() {
+
+	locked=false;
+}
+
+bool SpaceSW::is_locked() const {
+
+	return locked;
+}
+
+PhysicsDirectSpaceStateSW *SpaceSW::get_direct_state() {
+
+	return direct_access;
+}
+
+SpaceSW::SpaceSW() {
+
+	collision_pairs=0;
+	active_objects=0;
+	island_count=0;
+	contact_debug_count=0;
+
+	locked=false;
+	contact_recycle_radius=0.01;
+	contact_max_separation=0.05;
+	contact_max_allowed_penetration= 0.01;
+
+	constraint_bias = 0.01;
+	body_linear_velocity_sleep_threshold=GLOBAL_DEF("physics/sleep_threshold_linear",0.1);
+	body_angular_velocity_sleep_threshold=GLOBAL_DEF("physics/sleep_threshold_angular", (8.0 / 180.0 * Math_PI) );
+	body_time_to_sleep=0.5;
+	body_angular_velocity_damp_ratio=10;
+
+
+	broadphase = BroadPhaseSW::create_func();
+	broadphase->set_pair_callback(_broadphase_pair,this);
+	broadphase->set_unpair_callback(_broadphase_unpair,this);
+	area=NULL;
+
+	direct_access = memnew( PhysicsDirectSpaceStateSW );
+	direct_access->space=this;
+}
+
+SpaceSW::~SpaceSW() {
+
+	memdelete(broadphase);
+	memdelete( direct_access );
+}
+
+
+
diff --git a/servers/physics/space_sw.h b/servers/physics/space_sw.h
index e88f61d881..ac788ba93f 100644
--- a/servers/physics/space_sw.h
+++ b/servers/physics/space_sw.h
@@ -1,187 +1,187 @@
-/*************************************************************************/

-/*  space_sw.h                                                           */

-/*************************************************************************/

-/*                       This file is part of:                           */

-/*                           GODOT ENGINE                                */

-/*                    http://www.godotengine.org                         */

-/*************************************************************************/

-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */

-/*                                                                       */

-/* Permission is hereby granted, free of charge, to any person obtaining */

-/* a copy of this software and associated documentation files (the       */

-/* "Software"), to deal in the Software without restriction, including   */

-/* without limitation the rights to use, copy, modify, merge, publish,   */

-/* distribute, sublicense, and/or sell copies of the Software, and to    */

-/* permit persons to whom the Software is furnished to do so, subject to */

-/* the following conditions:                                             */

-/*                                                                       */

-/* The above copyright notice and this permission notice shall be        */

-/* included in all copies or substantial portions of the Software.       */

-/*                                                                       */

-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */

-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */

-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/

-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */

-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */

-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */

-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */

-/*************************************************************************/

-#ifndef SPACE_SW_H

-#define SPACE_SW_H

-

-#include "typedefs.h"

-#include "hash_map.h"

-#include "body_sw.h"

-#include "area_sw.h"

-#include "body_pair_sw.h"

-#include "area_pair_sw.h"

-#include "broad_phase_sw.h"

-#include "collision_object_sw.h"

-

-

-class PhysicsDirectSpaceStateSW : public PhysicsDirectSpaceState {

-

-	OBJ_TYPE( PhysicsDirectSpaceStateSW, PhysicsDirectSpaceState );

-public:

-

-	SpaceSW *space;

-

-	virtual bool intersect_ray(const Vector3& p_from, const Vector3& p_to,RayResult &r_result,const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);

-	virtual int intersect_shape(const RID& p_shape, const Transform& p_xform,float p_margin,ShapeResult *r_results,int p_result_max,const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);

-	virtual bool cast_motion(const RID& p_shape, const Transform& p_xform,const Vector3& p_motion,float p_margin,float &p_closest_safe,float &p_closest_unsafe, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION,ShapeRestInfo *r_info=NULL);

-	virtual bool collide_shape(RID p_shape, const Transform& p_shape_xform,float p_margin,Vector3 *r_results,int p_result_max,int &r_result_count, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);

-	virtual bool rest_info(RID p_shape, const Transform& p_shape_xform,float p_margin,ShapeRestInfo *r_info, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);

-

-	PhysicsDirectSpaceStateSW();

-};

-

-

-

-class SpaceSW {

-

-

-	PhysicsDirectSpaceStateSW *direct_access;

-	RID self;

-

-	BroadPhaseSW *broadphase;

-	SelfList<BodySW>::List active_list;

-	SelfList<BodySW>::List inertia_update_list;

-	SelfList<BodySW>::List state_query_list;

-	SelfList<AreaSW>::List monitor_query_list;

-	SelfList<AreaSW>::List area_moved_list;

-

-	static void* _broadphase_pair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_self);

-	static void _broadphase_unpair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_data,void *p_self);

-

-	Set<CollisionObjectSW*> objects;

-

-	AreaSW *area;

-

-	real_t contact_recycle_radius;

-	real_t contact_max_separation;

-	real_t contact_max_allowed_penetration;

-	real_t constraint_bias;

-

-	enum {

-

-		INTERSECTION_QUERY_MAX=2048

-	};

-

-	CollisionObjectSW *intersection_query_results[INTERSECTION_QUERY_MAX];

-	int intersection_query_subindex_results[INTERSECTION_QUERY_MAX];

-

-	float body_linear_velocity_sleep_threshold;

-	float body_angular_velocity_sleep_threshold;

-	float body_time_to_sleep;

-	float body_angular_velocity_damp_ratio;

-

-	bool locked;

-

-	int island_count;

-	int active_objects;

-	int collision_pairs;

-

-	RID static_global_body;

-

-	Vector<Vector3> contact_debug;

-	int contact_debug_count;

-

-friend class PhysicsDirectSpaceStateSW;

-

-public:

-

-	_FORCE_INLINE_ void set_self(const RID& p_self) { self=p_self; }

-	_FORCE_INLINE_ RID get_self() const { return self; }

-

-	void set_default_area(AreaSW *p_area) { area=p_area; }

-	AreaSW *get_default_area() const { return area; }

-

-	const SelfList<BodySW>::List& get_active_body_list() const;

-	void body_add_to_active_list(SelfList<BodySW>* p_body);

-	void body_remove_from_active_list(SelfList<BodySW>* p_body);

-	void body_add_to_inertia_update_list(SelfList<BodySW>* p_body);

-	void body_remove_from_inertia_update_list(SelfList<BodySW>* p_body);

-

-	void body_add_to_state_query_list(SelfList<BodySW>* p_body);

-	void body_remove_from_state_query_list(SelfList<BodySW>* p_body);

-

-	void area_add_to_monitor_query_list(SelfList<AreaSW>* p_area);

-	void area_remove_from_monitor_query_list(SelfList<AreaSW>* p_area);

-	void area_add_to_moved_list(SelfList<AreaSW>* p_area);

-	void area_remove_from_moved_list(SelfList<AreaSW>* p_area);

-	const SelfList<AreaSW>::List& get_moved_area_list() const;

-

-	BroadPhaseSW *get_broadphase();

-

-	void add_object(CollisionObjectSW *p_object);

-	void remove_object(CollisionObjectSW *p_object);

-	const Set<CollisionObjectSW*> &get_objects() const;

-

-	_FORCE_INLINE_ real_t get_contact_recycle_radius() const { return contact_recycle_radius; }

-	_FORCE_INLINE_ real_t get_contact_max_separation() const { return contact_max_separation; }

-	_FORCE_INLINE_ real_t get_contact_max_allowed_penetration() const { return contact_max_allowed_penetration; }

-	_FORCE_INLINE_ real_t get_constraint_bias() const { return constraint_bias; }

-	_FORCE_INLINE_ real_t get_body_linear_velocity_sleep_treshold() const { return body_linear_velocity_sleep_threshold; }

-	_FORCE_INLINE_ real_t get_body_angular_velocity_sleep_treshold() const { return body_angular_velocity_sleep_threshold; }

-	_FORCE_INLINE_ real_t get_body_time_to_sleep() const { return body_time_to_sleep; }

-	_FORCE_INLINE_ real_t get_body_angular_velocity_damp_ratio() const { return body_angular_velocity_damp_ratio; }

-

-

-	void update();

-	void setup();

-	void call_queries();

-

-

-	bool is_locked() const;

-	void lock();

-	void unlock();

-

-	void set_param(PhysicsServer::SpaceParameter p_param, real_t p_value);

-	real_t get_param(PhysicsServer::SpaceParameter p_param) const;

-

-	void set_island_count(int p_island_count) { island_count=p_island_count; }

-	int get_island_count() const { return island_count; }

-

-	void set_active_objects(int p_active_objects) { active_objects=p_active_objects; }

-	int get_active_objects() const { return active_objects; }

-

-	int get_collision_pairs() const { return collision_pairs; }

-

-	PhysicsDirectSpaceStateSW *get_direct_state();

-

-	void set_debug_contacts(int p_amount) { contact_debug.resize(p_amount); }

-	_FORCE_INLINE_ bool is_debugging_contacts() const { return !contact_debug.empty(); }

-	_FORCE_INLINE_ void add_debug_contact(const Vector3& p_contact) { if (contact_debug_count<contact_debug.size()) contact_debug[contact_debug_count++]=p_contact; }

-	_FORCE_INLINE_ Vector<Vector3> get_debug_contacts() { return contact_debug; }

-	_FORCE_INLINE_ int get_debug_contact_count() { return contact_debug_count; }

-

-	void set_static_global_body(RID p_body) { static_global_body=p_body; }

-	RID get_static_global_body() { return static_global_body; }

-

-

-	SpaceSW();

-	~SpaceSW();

-};

-

-

-#endif // SPACE__SW_H

+/*************************************************************************/
+/*  space_sw.h                                                           */
+/*************************************************************************/
+/*                       This file is part of:                           */
+/*                           GODOT ENGINE                                */
+/*                    http://www.godotengine.org                         */
+/*************************************************************************/
+/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
+/*                                                                       */
+/* Permission is hereby granted, free of charge, to any person obtaining */
+/* a copy of this software and associated documentation files (the       */
+/* "Software"), to deal in the Software without restriction, including   */
+/* without limitation the rights to use, copy, modify, merge, publish,   */
+/* distribute, sublicense, and/or sell copies of the Software, and to    */
+/* permit persons to whom the Software is furnished to do so, subject to */
+/* the following conditions:                                             */
+/*                                                                       */
+/* The above copyright notice and this permission notice shall be        */
+/* included in all copies or substantial portions of the Software.       */
+/*                                                                       */
+/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
+/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
+/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
+/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
+/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
+/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
+/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
+/*************************************************************************/
+#ifndef SPACE_SW_H
+#define SPACE_SW_H
+
+#include "typedefs.h"
+#include "hash_map.h"
+#include "body_sw.h"
+#include "area_sw.h"
+#include "body_pair_sw.h"
+#include "area_pair_sw.h"
+#include "broad_phase_sw.h"
+#include "collision_object_sw.h"
+
+
+class PhysicsDirectSpaceStateSW : public PhysicsDirectSpaceState {
+
+	OBJ_TYPE( PhysicsDirectSpaceStateSW, PhysicsDirectSpaceState );
+public:
+
+	SpaceSW *space;
+
+	virtual bool intersect_ray(const Vector3& p_from, const Vector3& p_to,RayResult &r_result,const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);
+	virtual int intersect_shape(const RID& p_shape, const Transform& p_xform,float p_margin,ShapeResult *r_results,int p_result_max,const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);
+	virtual bool cast_motion(const RID& p_shape, const Transform& p_xform,const Vector3& p_motion,float p_margin,float &p_closest_safe,float &p_closest_unsafe, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION,ShapeRestInfo *r_info=NULL);
+	virtual bool collide_shape(RID p_shape, const Transform& p_shape_xform,float p_margin,Vector3 *r_results,int p_result_max,int &r_result_count, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);
+	virtual bool rest_info(RID p_shape, const Transform& p_shape_xform,float p_margin,ShapeRestInfo *r_info, const Set<RID>& p_exclude=Set<RID>(),uint32_t p_layer_mask=0xFFFFFFFF,uint32_t p_object_type_mask=TYPE_MASK_COLLISION);
+
+	PhysicsDirectSpaceStateSW();
+};
+
+
+
+class SpaceSW {
+
+
+	PhysicsDirectSpaceStateSW *direct_access;
+	RID self;
+
+	BroadPhaseSW *broadphase;
+	SelfList<BodySW>::List active_list;
+	SelfList<BodySW>::List inertia_update_list;
+	SelfList<BodySW>::List state_query_list;
+	SelfList<AreaSW>::List monitor_query_list;
+	SelfList<AreaSW>::List area_moved_list;
+
+	static void* _broadphase_pair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_self);
+	static void _broadphase_unpair(CollisionObjectSW *A,int p_subindex_A,CollisionObjectSW *B,int p_subindex_B,void *p_data,void *p_self);
+
+	Set<CollisionObjectSW*> objects;
+
+	AreaSW *area;
+
+	real_t contact_recycle_radius;
+	real_t contact_max_separation;
+	real_t contact_max_allowed_penetration;
+	real_t constraint_bias;
+
+	enum {
+
+		INTERSECTION_QUERY_MAX=2048
+	};
+
+	CollisionObjectSW *intersection_query_results[INTERSECTION_QUERY_MAX];
+	int intersection_query_subindex_results[INTERSECTION_QUERY_MAX];
+
+	float body_linear_velocity_sleep_threshold;
+	float body_angular_velocity_sleep_threshold;
+	float body_time_to_sleep;
+	float body_angular_velocity_damp_ratio;
+
+	bool locked;
+
+	int island_count;
+	int active_objects;
+	int collision_pairs;
+
+	RID static_global_body;
+
+	Vector<Vector3> contact_debug;
+	int contact_debug_count;
+
+friend class PhysicsDirectSpaceStateSW;
+
+public:
+
+	_FORCE_INLINE_ void set_self(const RID& p_self) { self=p_self; }
+	_FORCE_INLINE_ RID get_self() const { return self; }
+
+	void set_default_area(AreaSW *p_area) { area=p_area; }
+	AreaSW *get_default_area() const { return area; }
+
+	const SelfList<BodySW>::List& get_active_body_list() const;
+	void body_add_to_active_list(SelfList<BodySW>* p_body);
+	void body_remove_from_active_list(SelfList<BodySW>* p_body);
+	void body_add_to_inertia_update_list(SelfList<BodySW>* p_body);
+	void body_remove_from_inertia_update_list(SelfList<BodySW>* p_body);
+
+	void body_add_to_state_query_list(SelfList<BodySW>* p_body);
+	void body_remove_from_state_query_list(SelfList<BodySW>* p_body);
+
+	void area_add_to_monitor_query_list(SelfList<AreaSW>* p_area);
+	void area_remove_from_monitor_query_list(SelfList<AreaSW>* p_area);
+	void area_add_to_moved_list(SelfList<AreaSW>* p_area);
+	void area_remove_from_moved_list(SelfList<AreaSW>* p_area);
+	const SelfList<AreaSW>::List& get_moved_area_list() const;
+
+	BroadPhaseSW *get_broadphase();
+
+	void add_object(CollisionObjectSW *p_object);
+	void remove_object(CollisionObjectSW *p_object);
+	const Set<CollisionObjectSW*> &get_objects() const;
+
+	_FORCE_INLINE_ real_t get_contact_recycle_radius() const { return contact_recycle_radius; }
+	_FORCE_INLINE_ real_t get_contact_max_separation() const { return contact_max_separation; }
+	_FORCE_INLINE_ real_t get_contact_max_allowed_penetration() const { return contact_max_allowed_penetration; }
+	_FORCE_INLINE_ real_t get_constraint_bias() const { return constraint_bias; }
+	_FORCE_INLINE_ real_t get_body_linear_velocity_sleep_treshold() const { return body_linear_velocity_sleep_threshold; }
+	_FORCE_INLINE_ real_t get_body_angular_velocity_sleep_treshold() const { return body_angular_velocity_sleep_threshold; }
+	_FORCE_INLINE_ real_t get_body_time_to_sleep() const { return body_time_to_sleep; }
+	_FORCE_INLINE_ real_t get_body_angular_velocity_damp_ratio() const { return body_angular_velocity_damp_ratio; }
+
+
+	void update();
+	void setup();
+	void call_queries();
+
+
+	bool is_locked() const;
+	void lock();
+	void unlock();
+
+	void set_param(PhysicsServer::SpaceParameter p_param, real_t p_value);
+	real_t get_param(PhysicsServer::SpaceParameter p_param) const;
+
+	void set_island_count(int p_island_count) { island_count=p_island_count; }
+	int get_island_count() const { return island_count; }
+
+	void set_active_objects(int p_active_objects) { active_objects=p_active_objects; }
+	int get_active_objects() const { return active_objects; }
+
+	int get_collision_pairs() const { return collision_pairs; }
+
+	PhysicsDirectSpaceStateSW *get_direct_state();
+
+	void set_debug_contacts(int p_amount) { contact_debug.resize(p_amount); }
+	_FORCE_INLINE_ bool is_debugging_contacts() const { return !contact_debug.empty(); }
+	_FORCE_INLINE_ void add_debug_contact(const Vector3& p_contact) { if (contact_debug_count<contact_debug.size()) contact_debug[contact_debug_count++]=p_contact; }
+	_FORCE_INLINE_ Vector<Vector3> get_debug_contacts() { return contact_debug; }
+	_FORCE_INLINE_ int get_debug_contact_count() { return contact_debug_count; }
+
+	void set_static_global_body(RID p_body) { static_global_body=p_body; }
+	RID get_static_global_body() { return static_global_body; }
+
+
+	SpaceSW();
+	~SpaceSW();
+};
+
+
+#endif // SPACE__SW_H
diff --git a/servers/physics_2d/joints_2d_sw.cpp b/servers/physics_2d/joints_2d_sw.cpp
index b4c149e7e0..7c12000084 100644
--- a/servers/physics_2d/joints_2d_sw.cpp
+++ b/servers/physics_2d/joints_2d_sw.cpp
@@ -279,6 +279,18 @@ void PinJoint2DSW::solve(float p_step){
 	P += impulse;
 }
 
+void PinJoint2DSW::set_param(Physics2DServer::PinJointParam p_param, real_t p_value) {
+
+	if(p_param == Physics2DServer::PIN_JOINT_SOFTNESS)
+		softness = p_value;
+}
+
+real_t PinJoint2DSW::get_param(Physics2DServer::PinJointParam p_param) const {
+
+	if(p_param == Physics2DServer::PIN_JOINT_SOFTNESS)
+		return softness;
+	ERR_FAIL_V(0);
+}
 
 PinJoint2DSW::PinJoint2DSW(const Vector2& p_pos,Body2DSW* p_body_a,Body2DSW* p_body_b) : Joint2DSW(_arr,p_body_b?2:1) {
 
diff --git a/servers/physics_2d/joints_2d_sw.h b/servers/physics_2d/joints_2d_sw.h
index 2093be88c9..e43f8eee33 100644
--- a/servers/physics_2d/joints_2d_sw.h
+++ b/servers/physics_2d/joints_2d_sw.h
@@ -121,6 +121,8 @@ public:
 	virtual bool setup(float p_step);
 	virtual void solve(float p_step);
 
+	void set_param(Physics2DServer::PinJointParam p_param, real_t p_value);
+	real_t get_param(Physics2DServer::PinJointParam p_param) const;
 
 	PinJoint2DSW(const Vector2& p_pos,Body2DSW* p_body_a,Body2DSW* p_body_b=NULL);
 	~PinJoint2DSW();
diff --git a/servers/physics_2d/physics_2d_server_sw.cpp b/servers/physics_2d/physics_2d_server_sw.cpp
index 14b4c09ebc..6a1c790da8 100644
--- a/servers/physics_2d/physics_2d_server_sw.cpp
+++ b/servers/physics_2d/physics_2d_server_sw.cpp
@@ -1096,6 +1096,25 @@ RID Physics2DServerSW::damped_spring_joint_create(const Vector2& p_anchor_a,cons
 
 }
 
+void Physics2DServerSW::pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) {
+
+	Joint2DSW *j = joint_owner.get(p_joint);
+	ERR_FAIL_COND(!j);
+	ERR_FAIL_COND(j->get_type()!=JOINT_PIN);
+
+	PinJoint2DSW *pin_joint = static_cast<PinJoint2DSW*>(j);
+	pin_joint->set_param(p_param, p_value);
+}
+
+real_t Physics2DServerSW::pin_joint_get_param(RID p_joint, PinJointParam p_param) const {
+	Joint2DSW *j = joint_owner.get(p_joint);
+	ERR_FAIL_COND_V(!j,0);
+	ERR_FAIL_COND_V(j->get_type()!=JOINT_PIN,0);
+
+	PinJoint2DSW *pin_joint = static_cast<PinJoint2DSW*>(j);
+	return pin_joint->get_param(p_param);
+}
+
 void Physics2DServerSW::damped_string_joint_set_param(RID p_joint, DampedStringParam p_param, real_t p_value) {
 
 
diff --git a/servers/physics_2d/physics_2d_server_sw.h b/servers/physics_2d/physics_2d_server_sw.h
index 605e04ead8..b2c58b788e 100644
--- a/servers/physics_2d/physics_2d_server_sw.h
+++ b/servers/physics_2d/physics_2d_server_sw.h
@@ -243,6 +243,8 @@ public:
 	virtual RID pin_joint_create(const Vector2& p_pos,RID p_body_a,RID p_body_b=RID());
 	virtual RID groove_joint_create(const Vector2& p_a_groove1,const Vector2& p_a_groove2, const Vector2& p_b_anchor, RID p_body_a,RID p_body_b);
 	virtual RID damped_spring_joint_create(const Vector2& p_anchor_a,const Vector2& p_anchor_b,RID p_body_a,RID p_body_b=RID());
+	virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value);
+	virtual real_t pin_joint_get_param(RID p_joint, PinJointParam p_param) const;
 	virtual void damped_string_joint_set_param(RID p_joint, DampedStringParam p_param, real_t p_value);
 	virtual real_t damped_string_joint_get_param(RID p_joint, DampedStringParam p_param) const;
 
diff --git a/servers/physics_2d/physics_2d_server_wrap_mt.h b/servers/physics_2d/physics_2d_server_wrap_mt.h
index 54af3eeb99..60f8a4c879 100644
--- a/servers/physics_2d/physics_2d_server_wrap_mt.h
+++ b/servers/physics_2d/physics_2d_server_wrap_mt.h
@@ -258,6 +258,9 @@ public:
 	FUNC5R(RID,groove_joint_create,const Vector2&,const Vector2&,const Vector2&,RID,RID);
 	FUNC4R(RID,damped_spring_joint_create,const Vector2&,const Vector2&,RID,RID);
 
+	FUNC3(pin_joint_set_param,RID,PinJointParam,real_t);
+	FUNC2RC(real_t,pin_joint_get_param,RID,PinJointParam);
+
 	FUNC3(damped_string_joint_set_param,RID,DampedStringParam,real_t);
 	FUNC2RC(real_t,damped_string_joint_get_param,RID,DampedStringParam);
 
diff --git a/servers/physics_2d/space_2d_sw.cpp b/servers/physics_2d/space_2d_sw.cpp
index a71e6c4bf5..9f2f03baec 100644
--- a/servers/physics_2d/space_2d_sw.cpp
+++ b/servers/physics_2d/space_2d_sw.cpp
@@ -36,8 +36,8 @@ _FORCE_INLINE_ static bool _match_object_type_query(CollisionObject2DSW *p_objec
 	if ((p_object->get_layer_mask()&p_layer_mask)==0)
 		return false;
 
-	if (p_object->get_type()==CollisionObject2DSW::TYPE_AREA && !(p_type_mask&Physics2DDirectSpaceState::TYPE_MASK_AREA))
-		return false;
+	if (p_object->get_type()==CollisionObject2DSW::TYPE_AREA)
+		return p_type_mask&Physics2DDirectSpaceState::TYPE_MASK_AREA;
 
 	Body2DSW *body = static_cast<Body2DSW*>(p_object);
 
diff --git a/servers/physics_2d_server.h b/servers/physics_2d_server.h
index 2d70337dc8..6845c7dfe1 100644
--- a/servers/physics_2d_server.h
+++ b/servers/physics_2d_server.h
@@ -516,6 +516,13 @@ public:
 	virtual RID groove_joint_create(const Vector2& p_a_groove1,const Vector2& p_a_groove2, const Vector2& p_b_anchor, RID p_body_a,RID p_body_b)=0;
 	virtual RID damped_spring_joint_create(const Vector2& p_anchor_a,const Vector2& p_anchor_b,RID p_body_a,RID p_body_b=RID())=0;
 
+	enum PinJointParam {
+		PIN_JOINT_SOFTNESS
+	};
+
+	virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value)=0;
+	virtual real_t pin_joint_get_param(RID p_joint, PinJointParam p_param) const=0;
+
 	enum DampedStringParam {
 		DAMPED_STRING_REST_LENGTH,
 		DAMPED_STRING_STIFFNESS,
diff --git a/servers/visual/visual_server_raster.cpp b/servers/visual/visual_server_raster.cpp
index fbea60c3a6..cf840c1be3 100644
--- a/servers/visual/visual_server_raster.cpp
+++ b/servers/visual/visual_server_raster.cpp
@@ -6824,7 +6824,11 @@ void VisualServerRaster::_render_canvas_item(CanvasItem *p_canvas_item,const Mat
 	copymem(child_items,ci->child_items.ptr(),child_item_count*sizeof(CanvasItem*));
 
 	if (ci->clip) {
-		ci->final_clip_rect=global_rect;
+		if (p_canvas_clip != NULL) {
+			ci->final_clip_rect=p_canvas_clip->final_clip_rect.clip(global_rect);
+		} else {
+			ci->final_clip_rect=global_rect;
+		}
 		ci->final_clip_owner=ci;
 
 	} else {
diff --git a/tools/editor/editor_node.cpp b/tools/editor/editor_node.cpp
index 4f5755bd3d..9137503e1b 100644
--- a/tools/editor/editor_node.cpp
+++ b/tools/editor/editor_node.cpp
@@ -1108,6 +1108,11 @@ void EditorNode::_dialog_action(String p_file) {
 
 			push_item(res.operator->() );
 		} break;			
+		case FILE_NEW_INHERITED_SCENE: {
+
+
+			load_scene(p_file,false,true);
+		} break;
 		case FILE_OPEN_SCENE: {
 
 
@@ -1931,6 +1936,7 @@ void EditorNode::_menu_option_confirm(int p_option,bool p_confirmed) {
 
 			
 		} break;
+		case FILE_NEW_INHERITED_SCENE:
 		case FILE_OPEN_SCENE: {
 			
 			
@@ -1951,7 +1957,7 @@ void EditorNode::_menu_option_confirm(int p_option,bool p_confirmed) {
 			if (scene) {
 				file->set_current_path(scene->get_filename());
 			};
-			file->set_title("Open Scene");
+			file->set_title(p_option==FILE_OPEN_SCENE?"Open Scene":"Open Base Scene");
 			file->popup_centered_ratio();
 			
 		} break;
@@ -3312,7 +3318,7 @@ void EditorNode::fix_dependencies(const String& p_for_file) {
 	dependency_fixer->edit(p_for_file);
 }
 
-Error EditorNode::load_scene(const String& p_scene, bool p_ignore_broken_deps) {
+Error EditorNode::load_scene(const String& p_scene, bool p_ignore_broken_deps,bool p_set_inherited) {
 
 	if (!is_inside_tree()) {
 		defer_load_scene = p_scene;
@@ -3442,6 +3448,13 @@ Error EditorNode::load_scene(const String& p_scene, bool p_ignore_broken_deps) {
 	}
 */
 
+	if (p_set_inherited) {
+		Ref<SceneState> state = sdata->get_state();
+		state->set_path(lpath);
+		new_scene->set_scene_inherited_state(state);
+	}
+
+
 	set_edited_scene(new_scene);
 	_get_scene_metadata();
 	/*
@@ -4404,6 +4417,7 @@ void EditorNode::_scene_tab_changed(int p_tab) {
 EditorNode::EditorNode() {
 
 	EditorHelp::generate_doc(); //before any editor classes are crated
+	SceneState::set_disable_placeholders(true);
 
 	InputDefault *id = Input::get_singleton()->cast_to<InputDefault>();
 
@@ -4794,6 +4808,7 @@ EditorNode::EditorNode() {
 	file_menu->set_tooltip("Operations with scene files.");
 	p=file_menu->get_popup();
 	p->add_item("New Scene",FILE_NEW_SCENE);
+	p->add_item("New Inherited Scene..",FILE_NEW_INHERITED_SCENE);
 	p->add_item("Open Scene..",FILE_OPEN_SCENE,KEY_MASK_CMD+KEY_O);
 	p->add_separator();
 	p->add_item("Save Scene",FILE_SAVE_SCENE,KEY_MASK_CMD+KEY_S);
diff --git a/tools/editor/editor_node.h b/tools/editor/editor_node.h
index 53e16d9c58..d52e08191d 100644
--- a/tools/editor/editor_node.h
+++ b/tools/editor/editor_node.h
@@ -107,6 +107,7 @@ class EditorNode : public Node {
 	enum MenuOptions {
 	
 		FILE_NEW_SCENE,
+		FILE_NEW_INHERITED_SCENE,
 		FILE_OPEN_SCENE,
 		FILE_SAVE_SCENE,
 		FILE_SAVE_AS_SCENE,
@@ -565,7 +566,7 @@ public:
 
 	void fix_dependencies(const String& p_for_file);
 	void clear_scene() { _cleanup_scene(); }
-	Error load_scene(const String& p_scene,bool p_ignore_broken_deps=false);
+	Error load_scene(const String& p_scene, bool p_ignore_broken_deps=false, bool p_set_inherited=false);
 	Error load_resource(const String& p_scene);
 
 	bool is_scene_open(const String& p_path);
diff --git a/tools/editor/plugins/path_editor_plugin.cpp b/tools/editor/plugins/path_editor_plugin.cpp
index 4af22e956f..f4bdf50fe9 100644
--- a/tools/editor/plugins/path_editor_plugin.cpp
+++ b/tools/editor/plugins/path_editor_plugin.cpp
@@ -1,597 +1,597 @@
-/*************************************************************************/

-/*  path_editor_plugin.cpp                                               */

-/*************************************************************************/

-/*                       This file is part of:                           */

-/*                           GODOT ENGINE                                */

-/*                    http://www.godotengine.org                         */

-/*************************************************************************/

-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */

-/*                                                                       */

-/* Permission is hereby granted, free of charge, to any person obtaining */

-/* a copy of this software and associated documentation files (the       */

-/* "Software"), to deal in the Software without restriction, including   */

-/* without limitation the rights to use, copy, modify, merge, publish,   */

-/* distribute, sublicense, and/or sell copies of the Software, and to    */

-/* permit persons to whom the Software is furnished to do so, subject to */

-/* the following conditions:                                             */

-/*                                                                       */

-/* The above copyright notice and this permission notice shall be        */

-/* included in all copies or substantial portions of the Software.       */

-/*                                                                       */

-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */

-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */

-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/

-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */

-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */

-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */

-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */

-/*************************************************************************/

-#include "path_editor_plugin.h"

-#include "spatial_editor_plugin.h"

-#include "scene/resources/curve.h"

-#include "os/keyboard.h"

-

-String PathSpatialGizmo::get_handle_name(int p_idx) const {

-

-	Ref<Curve3D> c = path->get_curve();

-	if (c.is_null())

-		return "";

-

-	if (p_idx<c->get_point_count()) {

-

-		return "Curve Point #"+itos(p_idx);

-	}

-

-	p_idx=p_idx-c->get_point_count()+1;

-

-	int idx=p_idx/2;

-	int t=p_idx%2;

-	String n = "Curve Point #"+itos(idx);

-	if (t==0)

-		n+=" In";

-	else

-		n+=" Out";

-

-	return n;

-

-

-}

-Variant PathSpatialGizmo::get_handle_value(int p_idx) const{

-

-	Ref<Curve3D> c = path->get_curve();

-	if (c.is_null())

-		return Variant();

-

-	if (p_idx<c->get_point_count()) {

-

-		original=c->get_point_pos(p_idx);

-		return original;

-	}

-

-	p_idx=p_idx-c->get_point_count()+1;

-

-	int idx=p_idx/2;

-	int t=p_idx%2;

-

-	Vector3 ofs;

-	if (t==0)

-		ofs=c->get_point_in(idx);

-	else

-		ofs= c->get_point_out(idx);

-

-	original=ofs+c->get_point_pos(idx);

-

-	return ofs;

-

-}

-void PathSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){

-

-	Ref<Curve3D> c = path->get_curve();

-	if (c.is_null())

-		return;

-

-	Transform gt = path->get_global_transform();

-	Transform gi = gt.affine_inverse();

-	Vector3 ray_from = p_camera->project_ray_origin(p_point);

-	Vector3 ray_dir = p_camera->project_ray_normal(p_point);

-

-	if (p_idx<c->get_point_count()) {

-

-		Plane p(gt.xform(original),p_camera->get_transform().basis.get_axis(2));

-

-		Vector3 inters;

-

-		if (p.intersects_ray(ray_from,ray_dir,&inters)) {

-

-			Vector3 local = gi.xform(inters);

-			c->set_point_pos(p_idx,local);

-		}

-

-		return;

-	}

-

-	p_idx=p_idx-c->get_point_count()+1;

-

-	int idx=p_idx/2;

-	int t=p_idx%2;

-

-	Vector3 base = c->get_point_pos(idx);

-

-	Plane p(gt.xform(original),p_camera->get_transform().basis.get_axis(2));

-

-	Vector3 inters;

-

-	if (p.intersects_ray(ray_from,ray_dir,&inters)) {

-

-		Vector3 local = gi.xform(inters)-base;

-		if (t==0) {

-			c->set_point_in(idx,local);

-		} else {

-			c->set_point_out(idx,local);

-		}

-	}

-

-}

-

-void PathSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){

-

-	Ref<Curve3D> c = path->get_curve();

-	if (c.is_null())

-		return;

-

-	UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-

-	if (p_idx<c->get_point_count()) {

-

-		if (p_cancel) {

-

-			c->set_point_pos(p_idx,p_restore);

-			return;

-		}

-		ur->create_action("Set Curve Point Pos");

-		ur->add_do_method(c.ptr(),"set_point_pos",p_idx,c->get_point_pos(p_idx));

-		ur->add_undo_method(c.ptr(),"set_point_pos",p_idx,p_restore);

-		ur->commit_action();

-

-		return;

-	}

-

-	p_idx=p_idx-c->get_point_count()+1;

-

-	int idx=p_idx/2;

-	int t=p_idx%2;

-

-	Vector3 ofs;

-

-	if (p_cancel) {

-

-

-

-		return;

-	}

-

-

-

-	if (t==0) {

-

-		if (p_cancel) {

-

-			c->set_point_in(p_idx,p_restore);

-			return;

-		}

-		ur->create_action("Set Curve In Pos");

-		ur->add_do_method(c.ptr(),"set_point_in",idx,c->get_point_in(idx));

-		ur->add_undo_method(c.ptr(),"set_point_in",idx,p_restore);

-		ur->commit_action();

-

-

-	} else {

-		if (p_cancel) {

-

-			c->set_point_out(idx,p_restore);

-			return;

-		}

-		ur->create_action("Set Curve Out Pos");

-		ur->add_do_method(c.ptr(),"set_point_out",idx,c->get_point_out(idx));

-		ur->add_undo_method(c.ptr(),"set_point_out",idx,p_restore);

-		ur->commit_action();

-

-	}

-

-}

-

-

-void PathSpatialGizmo::redraw(){

-

-	clear();

-

-	Ref<Curve3D> c = path->get_curve();

-	if (c.is_null())

-		return;

-

-	Vector3Array v3a=c->tesselate();

-	//Vector3Array v3a=c->get_baked_points();

-

-	int v3s = v3a.size();

-	if (v3s==0)

-		return;

-	Vector<Vector3> v3p;

-	Vector3Array::Read r = v3a.read();

-

-	for(int i=0;i<v3s-1;i++) {

-

-		v3p.push_back(r[i]);

-		v3p.push_back(r[i+1]);

-		//v3p.push_back(r[i]);

-		//v3p.push_back(r[i]+Vector3(0,0.2,0));

-	}

-

-	add_lines(v3p,PathEditorPlugin::singleton->path_material);

-	add_collision_segments(v3p);

-

-	if (PathEditorPlugin::singleton->get_edited_path()==path) {

-		v3p.clear();

-		Vector<Vector3> handles;

-		Vector<Vector3> sec_handles;

-

-		for(int i=0;i<c->get_point_count();i++) {

-

-			Vector3 p = c->get_point_pos(i);

-			handles.push_back(p);

-			if (i>0) {

-				v3p.push_back(p);

-				v3p.push_back(p+c->get_point_in(i));

-				sec_handles.push_back(p+c->get_point_in(i));

-			}

-

-			if (i<c->get_point_count()-1) {

-				v3p.push_back(p);

-				v3p.push_back(p+c->get_point_out(i));

-				sec_handles.push_back(p+c->get_point_out(i));

-			}

-		}

-

-		add_lines(v3p,PathEditorPlugin::singleton->path_thin_material);

-		add_handles(handles);

-		add_handles(sec_handles,false,true);

-	}

-

-}

-

-PathSpatialGizmo::PathSpatialGizmo(Path* p_path){

-

-	path=p_path;

-	set_spatial_node(p_path);

-

-

-

-}

-

-bool PathEditorPlugin::create_spatial_gizmo(Spatial* p_spatial) {

-

-	if (p_spatial->cast_to<Path>()) {

-

-

-		Ref<PathSpatialGizmo> psg = memnew( PathSpatialGizmo(p_spatial->cast_to<Path>()));

-		p_spatial->set_gizmo(psg);

-		return true;

-	}

-

-	return false;

-}

-

-bool PathEditorPlugin::forward_spatial_input_event(Camera* p_camera,const InputEvent& p_event) {

-

-	if (!path)

-		return false;

-	Ref<Curve3D> c=path->get_curve();

-	if (c.is_null())

-		return false;

-	Transform gt = path->get_global_transform();

-	Transform it = gt.affine_inverse();

-

-	static const int click_dist = 10; //should make global

-

-

-	if (p_event.type==InputEvent::MOUSE_BUTTON) {

-

-		const InputEventMouseButton &mb=p_event.mouse_button;

-		Point2 mbpos(mb.x,mb.y);

-

-		if (mb.pressed && mb.button_index==BUTTON_LEFT && (curve_create->is_pressed() || (curve_edit->is_pressed() && mb.mod.control))) {

-			//click into curve, break it down

-			Vector3Array v3a = c->tesselate();

-			int idx=0;

-			int rc=v3a.size();

-			int closest_seg=-1;

-			Vector3 closest_seg_point;

-			float closest_d=1e20;

-

-			if (rc>=2) {

-				Vector3Array::Read r = v3a.read();

-

-				if (p_camera->unproject_position(gt.xform(c->get_point_pos(0))).distance_to(mbpos)<click_dist)

-					return false; //nope, existing

-

-

-				for(int i=0;i<c->get_point_count()-1;i++) {

-					//find the offset and point index of the place to break up

-					int j=idx;

-					if (p_camera->unproject_position(gt.xform(c->get_point_pos(i+1))).distance_to(mbpos)<click_dist)

-						return false; //nope, existing

-

-

-					while(j<rc && c->get_point_pos(i+1)!=r[j]) {

-

-						Vector3 from =r[j];

-						Vector3 to =r[j+1];

-						real_t cdist = from.distance_to(to);

-						from=gt.xform(from);

-						to=gt.xform(to);

-						if (cdist>0) {

-							Vector2 s[2];

-							s[0] = p_camera->unproject_position(from);

-							s[1] = p_camera->unproject_position(to);

-							Vector2 inters = Geometry::get_closest_point_to_segment_2d(mbpos,s);

-							float d = inters.distance_to(mbpos);

-

-							if (d<10 && d<closest_d) {

-

-

-								closest_d=d;

-								closest_seg=i;

-								Vector3 ray_from=p_camera->project_ray_origin(mbpos);

-								Vector3 ray_dir=p_camera->project_ray_normal(mbpos);

-

-								Vector3 ra,rb;

-								Geometry::get_closest_points_between_segments(ray_from,ray_from+ray_dir*4096,from,to,ra,rb);

-

-								closest_seg_point=it.xform(rb);

-							}

-

-						}

-						j++;

-

-					}

-					if (idx==j)

-						idx++; //force next

-					else

-						idx=j; //swap

-

-

-					if (j==rc)

-						break;

-				}

-			}

-

-			UndoRedo *ur = editor->get_undo_redo();

-			if (closest_seg!=-1) {

-				//subdivide

-

-				ur->create_action("Split Path");

-				ur->add_do_method(c.ptr(),"add_point",closest_seg_point,Vector3(),Vector3(),closest_seg+1);

-				ur->add_undo_method(c.ptr(),"remove_point",closest_seg+1);

-				ur->commit_action();;

-				return true;

-

-			} else {

-

-				Vector3 org;

-				if (c->get_point_count()==0)

-					org=path->get_transform().get_origin();

-				else

-					org=gt.xform(c->get_point_pos(c->get_point_count()));

-				Plane p(org,p_camera->get_transform().basis.get_axis(2));

-				Vector3 ray_from=p_camera->project_ray_origin(mbpos);

-				Vector3 ray_dir=p_camera->project_ray_normal(mbpos);

-

-				Vector3 inters;

-				if (p.intersects_ray(ray_from,ray_dir,&inters)) {

-

-					ur->create_action("Add Point to Curve");

-					ur->add_do_method(c.ptr(),"add_point",it.xform(inters),Vector3(),Vector3(),-1);

-					ur->add_undo_method(c.ptr(),"remove_point",c->get_point_count());

-					ur->commit_action();;

-					return true;

-				}

-

-				//add new at pos

-			}

-

-		} else if (mb.pressed && ((mb.button_index==BUTTON_LEFT && curve_del->is_pressed()) || (mb.button_index==BUTTON_RIGHT && curve_edit->is_pressed()))) {

-

-			int erase_idx=-1;

-			for(int i=0;i<c->get_point_count();i++) {

-				//find the offset and point index of the place to break up

-				if (p_camera->unproject_position(gt.xform(c->get_point_pos(i))).distance_to(mbpos)<click_dist) {

-

-					erase_idx=i;

-					break;

-				}

-			}

-

-			if (erase_idx!=-1) {

-

-				UndoRedo *ur = editor->get_undo_redo();

-				ur->create_action("Remove Path Point");

-				ur->add_do_method(c.ptr(),"remove_point",erase_idx);

-				ur->add_undo_method(c.ptr(),"add_point",c->get_point_pos(erase_idx),c->get_point_in(erase_idx),c->get_point_out(erase_idx),erase_idx);

-				ur->commit_action();

-				return true;

-			}

-		}

-

-	}

-

-	return false;

-}

-

-

-void PathEditorPlugin::edit(Object *p_object) {

-

-	if (p_object) {

-		path=p_object->cast_to<Path>();

-		if (path) {

-

-			if (path->get_curve().is_valid()) {

-				path->get_curve()->emit_signal("changed");

-			}

-		}

-	} else {

-		Path *pre=path;

-		path=NULL;

-		if (pre) {

-			pre->get_curve()->emit_signal("changed");

-		}

-	}

-//	collision_polygon_editor->edit(p_object->cast_to<Node>());

-}

-

-bool PathEditorPlugin::handles(Object *p_object) const {

-

-	return p_object->is_type("Path");

-}

-

-void PathEditorPlugin::make_visible(bool p_visible) {

-

-	if (p_visible) {

-

-		curve_create->show();

-		curve_edit->show();

-		curve_del->show();

-        curve_close->show();

-        sep->show();

-	} else {

-

-		curve_create->hide();

-		curve_edit->hide();

-		curve_del->hide();

-        curve_close->hide();

-        sep->hide();

-

-		{

-			Path *pre=path;

-			path=NULL;

-			if (pre && pre->get_curve().is_valid()) {

-				pre->get_curve()->emit_signal("changed");

-			}

-		}

-	}

-

-}

-

-void PathEditorPlugin::_mode_changed(int p_idx) {

-

-	curve_create->set_pressed(p_idx==0);

-	curve_edit->set_pressed(p_idx==1);

-	curve_del->set_pressed(p_idx==2);

-}

-

-void PathEditorPlugin::_close_curve() {

-

-    Ref<Curve3D> c = path->get_curve();

-    if (c.is_null())

-        return ;

-    if (c->get_point_count()<2)

-        return;

-    c->add_point(c->get_point_pos(0),c->get_point_in(0),c->get_point_out(0));

-

-}

-

-void PathEditorPlugin::_notification(int p_what) {

-

-	if (p_what==NOTIFICATION_ENTER_TREE) {

-

-		curve_create->connect("pressed",this,"_mode_changed",make_binds(0));

-		curve_edit->connect("pressed",this,"_mode_changed",make_binds(1));

-		curve_del->connect("pressed",this,"_mode_changed",make_binds(2));

-        curve_close->connect("pressed",this,"_close_curve");

-    }

-}

-

-void PathEditorPlugin::_bind_methods() {

-

-	ObjectTypeDB::bind_method(_MD("_mode_changed"),&PathEditorPlugin::_mode_changed);

-    ObjectTypeDB::bind_method(_MD("_close_curve"),&PathEditorPlugin::_close_curve);

-}

-

-PathEditorPlugin* PathEditorPlugin::singleton=NULL;

-

-

-PathEditorPlugin::PathEditorPlugin(EditorNode *p_node) {

-

-	path=NULL;

-	editor=p_node;

-	singleton=this;

-

-	path_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	path_material->set_parameter( FixedMaterial::PARAM_DIFFUSE,Color(0.5,0.5,1.0,0.8) );

-	path_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	path_material->set_line_width(3);

-	path_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);

-	path_material->set_flag(Material::FLAG_UNSHADED,true);

-

-	path_thin_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	path_thin_material->set_parameter( FixedMaterial::PARAM_DIFFUSE,Color(0.5,0.5,1.0,0.4) );

-	path_thin_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	path_thin_material->set_line_width(1);

-	path_thin_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);

-	path_thin_material->set_flag(Material::FLAG_UNSHADED,true);

-

-	SpatialEditor::get_singleton()->add_gizmo_plugin(this);

-

-	sep = memnew( VSeparator);

-	sep->hide();

-	SpatialEditor::get_singleton()->add_control_to_menu_panel(sep);

-	curve_edit = memnew( ToolButton );

-	curve_edit->set_icon(SpatialEditor::get_singleton()->get_icon("CurveEdit","EditorIcons"));

-	curve_edit->set_toggle_mode(true);

-	curve_edit->hide();

-	curve_edit->set_focus_mode(Control::FOCUS_NONE);

-	curve_edit->set_tooltip("Select Points\nShift+Drag: Select Control Points\n"+keycode_get_string(KEY_MASK_CMD)+"Click: Add Point\nRight Click: Delete Point.");

-	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_edit);

-	curve_create = memnew( ToolButton );

-	curve_create->set_icon(SpatialEditor::get_singleton()->get_icon("CurveCreate","EditorIcons"));

-	curve_create->set_toggle_mode(true);

-	curve_create->hide();

-	curve_create->set_focus_mode(Control::FOCUS_NONE);

-	curve_create->set_tooltip("Add Point (in empty space)\nSplit Segment (in curve).");

-	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_create);

-	curve_del = memnew( ToolButton );

-	curve_del->set_icon(SpatialEditor::get_singleton()->get_icon("CurveDelete","EditorIcons"));

-	curve_del->set_toggle_mode(true);

-	curve_del->hide();

-	curve_del->set_focus_mode(Control::FOCUS_NONE);

-	curve_del->set_tooltip("Delete Point.");

-	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_del);

-	curve_close = memnew( ToolButton );

-	curve_close->set_icon(SpatialEditor::get_singleton()->get_icon("CurveClose","EditorIcons"));

-	curve_close->hide();

-	curve_close->set_focus_mode(Control::FOCUS_NONE);

-	curve_close->set_tooltip("Close Curve");

-	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_close);

-

-

-

-	curve_edit->set_pressed(true);

-	/*

-	collision_polygon_editor = memnew( PathEditor(p_node) );

-	editor->get_viewport()->add_child(collision_polygon_editor);

-

-	collision_polygon_editor->set_margin(MARGIN_LEFT,200);

-	collision_polygon_editor->set_margin(MARGIN_RIGHT,230);

-	collision_polygon_editor->set_margin(MARGIN_TOP,0);

-	collision_polygon_editor->set_margin(MARGIN_BOTTOM,10);

-

-

-	collision_polygon_editor->hide();

-	*/

-

-

-}

-

-

-PathEditorPlugin::~PathEditorPlugin()

-{

-}

-

+/*************************************************************************/
+/*  path_editor_plugin.cpp                                               */
+/*************************************************************************/
+/*                       This file is part of:                           */
+/*                           GODOT ENGINE                                */
+/*                    http://www.godotengine.org                         */
+/*************************************************************************/
+/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
+/*                                                                       */
+/* Permission is hereby granted, free of charge, to any person obtaining */
+/* a copy of this software and associated documentation files (the       */
+/* "Software"), to deal in the Software without restriction, including   */
+/* without limitation the rights to use, copy, modify, merge, publish,   */
+/* distribute, sublicense, and/or sell copies of the Software, and to    */
+/* permit persons to whom the Software is furnished to do so, subject to */
+/* the following conditions:                                             */
+/*                                                                       */
+/* The above copyright notice and this permission notice shall be        */
+/* included in all copies or substantial portions of the Software.       */
+/*                                                                       */
+/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
+/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
+/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
+/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
+/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
+/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
+/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
+/*************************************************************************/
+#include "path_editor_plugin.h"
+#include "spatial_editor_plugin.h"
+#include "scene/resources/curve.h"
+#include "os/keyboard.h"
+
+String PathSpatialGizmo::get_handle_name(int p_idx) const {
+
+	Ref<Curve3D> c = path->get_curve();
+	if (c.is_null())
+		return "";
+
+	if (p_idx<c->get_point_count()) {
+
+		return "Curve Point #"+itos(p_idx);
+	}
+
+	p_idx=p_idx-c->get_point_count()+1;
+
+	int idx=p_idx/2;
+	int t=p_idx%2;
+	String n = "Curve Point #"+itos(idx);
+	if (t==0)
+		n+=" In";
+	else
+		n+=" Out";
+
+	return n;
+
+
+}
+Variant PathSpatialGizmo::get_handle_value(int p_idx) const{
+
+	Ref<Curve3D> c = path->get_curve();
+	if (c.is_null())
+		return Variant();
+
+	if (p_idx<c->get_point_count()) {
+
+		original=c->get_point_pos(p_idx);
+		return original;
+	}
+
+	p_idx=p_idx-c->get_point_count()+1;
+
+	int idx=p_idx/2;
+	int t=p_idx%2;
+
+	Vector3 ofs;
+	if (t==0)
+		ofs=c->get_point_in(idx);
+	else
+		ofs= c->get_point_out(idx);
+
+	original=ofs+c->get_point_pos(idx);
+
+	return ofs;
+
+}
+void PathSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){
+
+	Ref<Curve3D> c = path->get_curve();
+	if (c.is_null())
+		return;
+
+	Transform gt = path->get_global_transform();
+	Transform gi = gt.affine_inverse();
+	Vector3 ray_from = p_camera->project_ray_origin(p_point);
+	Vector3 ray_dir = p_camera->project_ray_normal(p_point);
+
+	if (p_idx<c->get_point_count()) {
+
+		Plane p(gt.xform(original),p_camera->get_transform().basis.get_axis(2));
+
+		Vector3 inters;
+
+		if (p.intersects_ray(ray_from,ray_dir,&inters)) {
+
+			Vector3 local = gi.xform(inters);
+			c->set_point_pos(p_idx,local);
+		}
+
+		return;
+	}
+
+	p_idx=p_idx-c->get_point_count()+1;
+
+	int idx=p_idx/2;
+	int t=p_idx%2;
+
+	Vector3 base = c->get_point_pos(idx);
+
+	Plane p(gt.xform(original),p_camera->get_transform().basis.get_axis(2));
+
+	Vector3 inters;
+
+	if (p.intersects_ray(ray_from,ray_dir,&inters)) {
+
+		Vector3 local = gi.xform(inters)-base;
+		if (t==0) {
+			c->set_point_in(idx,local);
+		} else {
+			c->set_point_out(idx,local);
+		}
+	}
+
+}
+
+void PathSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){
+
+	Ref<Curve3D> c = path->get_curve();
+	if (c.is_null())
+		return;
+
+	UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+
+	if (p_idx<c->get_point_count()) {
+
+		if (p_cancel) {
+
+			c->set_point_pos(p_idx,p_restore);
+			return;
+		}
+		ur->create_action("Set Curve Point Pos");
+		ur->add_do_method(c.ptr(),"set_point_pos",p_idx,c->get_point_pos(p_idx));
+		ur->add_undo_method(c.ptr(),"set_point_pos",p_idx,p_restore);
+		ur->commit_action();
+
+		return;
+	}
+
+	p_idx=p_idx-c->get_point_count()+1;
+
+	int idx=p_idx/2;
+	int t=p_idx%2;
+
+	Vector3 ofs;
+
+	if (p_cancel) {
+
+
+
+		return;
+	}
+
+
+
+	if (t==0) {
+
+		if (p_cancel) {
+
+			c->set_point_in(p_idx,p_restore);
+			return;
+		}
+		ur->create_action("Set Curve In Pos");
+		ur->add_do_method(c.ptr(),"set_point_in",idx,c->get_point_in(idx));
+		ur->add_undo_method(c.ptr(),"set_point_in",idx,p_restore);
+		ur->commit_action();
+
+
+	} else {
+		if (p_cancel) {
+
+			c->set_point_out(idx,p_restore);
+			return;
+		}
+		ur->create_action("Set Curve Out Pos");
+		ur->add_do_method(c.ptr(),"set_point_out",idx,c->get_point_out(idx));
+		ur->add_undo_method(c.ptr(),"set_point_out",idx,p_restore);
+		ur->commit_action();
+
+	}
+
+}
+
+
+void PathSpatialGizmo::redraw(){
+
+	clear();
+
+	Ref<Curve3D> c = path->get_curve();
+	if (c.is_null())
+		return;
+
+	Vector3Array v3a=c->tesselate();
+	//Vector3Array v3a=c->get_baked_points();
+
+	int v3s = v3a.size();
+	if (v3s==0)
+		return;
+	Vector<Vector3> v3p;
+	Vector3Array::Read r = v3a.read();
+
+	for(int i=0;i<v3s-1;i++) {
+
+		v3p.push_back(r[i]);
+		v3p.push_back(r[i+1]);
+		//v3p.push_back(r[i]);
+		//v3p.push_back(r[i]+Vector3(0,0.2,0));
+	}
+
+	add_lines(v3p,PathEditorPlugin::singleton->path_material);
+	add_collision_segments(v3p);
+
+	if (PathEditorPlugin::singleton->get_edited_path()==path) {
+		v3p.clear();
+		Vector<Vector3> handles;
+		Vector<Vector3> sec_handles;
+
+		for(int i=0;i<c->get_point_count();i++) {
+
+			Vector3 p = c->get_point_pos(i);
+			handles.push_back(p);
+			if (i>0) {
+				v3p.push_back(p);
+				v3p.push_back(p+c->get_point_in(i));
+				sec_handles.push_back(p+c->get_point_in(i));
+			}
+
+			if (i<c->get_point_count()-1) {
+				v3p.push_back(p);
+				v3p.push_back(p+c->get_point_out(i));
+				sec_handles.push_back(p+c->get_point_out(i));
+			}
+		}
+
+		add_lines(v3p,PathEditorPlugin::singleton->path_thin_material);
+		add_handles(handles);
+		add_handles(sec_handles,false,true);
+	}
+
+}
+
+PathSpatialGizmo::PathSpatialGizmo(Path* p_path){
+
+	path=p_path;
+	set_spatial_node(p_path);
+
+
+
+}
+
+bool PathEditorPlugin::create_spatial_gizmo(Spatial* p_spatial) {
+
+	if (p_spatial->cast_to<Path>()) {
+
+
+		Ref<PathSpatialGizmo> psg = memnew( PathSpatialGizmo(p_spatial->cast_to<Path>()));
+		p_spatial->set_gizmo(psg);
+		return true;
+	}
+
+	return false;
+}
+
+bool PathEditorPlugin::forward_spatial_input_event(Camera* p_camera,const InputEvent& p_event) {
+
+	if (!path)
+		return false;
+	Ref<Curve3D> c=path->get_curve();
+	if (c.is_null())
+		return false;
+	Transform gt = path->get_global_transform();
+	Transform it = gt.affine_inverse();
+
+	static const int click_dist = 10; //should make global
+
+
+	if (p_event.type==InputEvent::MOUSE_BUTTON) {
+
+		const InputEventMouseButton &mb=p_event.mouse_button;
+		Point2 mbpos(mb.x,mb.y);
+
+		if (mb.pressed && mb.button_index==BUTTON_LEFT && (curve_create->is_pressed() || (curve_edit->is_pressed() && mb.mod.control))) {
+			//click into curve, break it down
+			Vector3Array v3a = c->tesselate();
+			int idx=0;
+			int rc=v3a.size();
+			int closest_seg=-1;
+			Vector3 closest_seg_point;
+			float closest_d=1e20;
+
+			if (rc>=2) {
+				Vector3Array::Read r = v3a.read();
+
+				if (p_camera->unproject_position(gt.xform(c->get_point_pos(0))).distance_to(mbpos)<click_dist)
+					return false; //nope, existing
+
+
+				for(int i=0;i<c->get_point_count()-1;i++) {
+					//find the offset and point index of the place to break up
+					int j=idx;
+					if (p_camera->unproject_position(gt.xform(c->get_point_pos(i+1))).distance_to(mbpos)<click_dist)
+						return false; //nope, existing
+
+
+					while(j<rc && c->get_point_pos(i+1)!=r[j]) {
+
+						Vector3 from =r[j];
+						Vector3 to =r[j+1];
+						real_t cdist = from.distance_to(to);
+						from=gt.xform(from);
+						to=gt.xform(to);
+						if (cdist>0) {
+							Vector2 s[2];
+							s[0] = p_camera->unproject_position(from);
+							s[1] = p_camera->unproject_position(to);
+							Vector2 inters = Geometry::get_closest_point_to_segment_2d(mbpos,s);
+							float d = inters.distance_to(mbpos);
+
+							if (d<10 && d<closest_d) {
+
+
+								closest_d=d;
+								closest_seg=i;
+								Vector3 ray_from=p_camera->project_ray_origin(mbpos);
+								Vector3 ray_dir=p_camera->project_ray_normal(mbpos);
+
+								Vector3 ra,rb;
+								Geometry::get_closest_points_between_segments(ray_from,ray_from+ray_dir*4096,from,to,ra,rb);
+
+								closest_seg_point=it.xform(rb);
+							}
+
+						}
+						j++;
+
+					}
+					if (idx==j)
+						idx++; //force next
+					else
+						idx=j; //swap
+
+
+					if (j==rc)
+						break;
+				}
+			}
+
+			UndoRedo *ur = editor->get_undo_redo();
+			if (closest_seg!=-1) {
+				//subdivide
+
+				ur->create_action("Split Path");
+				ur->add_do_method(c.ptr(),"add_point",closest_seg_point,Vector3(),Vector3(),closest_seg+1);
+				ur->add_undo_method(c.ptr(),"remove_point",closest_seg+1);
+				ur->commit_action();;
+				return true;
+
+			} else {
+
+				Vector3 org;
+				if (c->get_point_count()==0)
+					org=path->get_transform().get_origin();
+				else
+					org=gt.xform(c->get_point_pos(c->get_point_count()));
+				Plane p(org,p_camera->get_transform().basis.get_axis(2));
+				Vector3 ray_from=p_camera->project_ray_origin(mbpos);
+				Vector3 ray_dir=p_camera->project_ray_normal(mbpos);
+
+				Vector3 inters;
+				if (p.intersects_ray(ray_from,ray_dir,&inters)) {
+
+					ur->create_action("Add Point to Curve");
+					ur->add_do_method(c.ptr(),"add_point",it.xform(inters),Vector3(),Vector3(),-1);
+					ur->add_undo_method(c.ptr(),"remove_point",c->get_point_count());
+					ur->commit_action();;
+					return true;
+				}
+
+				//add new at pos
+			}
+
+		} else if (mb.pressed && ((mb.button_index==BUTTON_LEFT && curve_del->is_pressed()) || (mb.button_index==BUTTON_RIGHT && curve_edit->is_pressed()))) {
+
+			int erase_idx=-1;
+			for(int i=0;i<c->get_point_count();i++) {
+				//find the offset and point index of the place to break up
+				if (p_camera->unproject_position(gt.xform(c->get_point_pos(i))).distance_to(mbpos)<click_dist) {
+
+					erase_idx=i;
+					break;
+				}
+			}
+
+			if (erase_idx!=-1) {
+
+				UndoRedo *ur = editor->get_undo_redo();
+				ur->create_action("Remove Path Point");
+				ur->add_do_method(c.ptr(),"remove_point",erase_idx);
+				ur->add_undo_method(c.ptr(),"add_point",c->get_point_pos(erase_idx),c->get_point_in(erase_idx),c->get_point_out(erase_idx),erase_idx);
+				ur->commit_action();
+				return true;
+			}
+		}
+
+	}
+
+	return false;
+}
+
+
+void PathEditorPlugin::edit(Object *p_object) {
+
+	if (p_object) {
+		path=p_object->cast_to<Path>();
+		if (path) {
+
+			if (path->get_curve().is_valid()) {
+				path->get_curve()->emit_signal("changed");
+			}
+		}
+	} else {
+		Path *pre=path;
+		path=NULL;
+		if (pre) {
+			pre->get_curve()->emit_signal("changed");
+		}
+	}
+//	collision_polygon_editor->edit(p_object->cast_to<Node>());
+}
+
+bool PathEditorPlugin::handles(Object *p_object) const {
+
+	return p_object->is_type("Path");
+}
+
+void PathEditorPlugin::make_visible(bool p_visible) {
+
+	if (p_visible) {
+
+		curve_create->show();
+		curve_edit->show();
+		curve_del->show();
+        curve_close->show();
+        sep->show();
+	} else {
+
+		curve_create->hide();
+		curve_edit->hide();
+		curve_del->hide();
+        curve_close->hide();
+        sep->hide();
+
+		{
+			Path *pre=path;
+			path=NULL;
+			if (pre && pre->get_curve().is_valid()) {
+				pre->get_curve()->emit_signal("changed");
+			}
+		}
+	}
+
+}
+
+void PathEditorPlugin::_mode_changed(int p_idx) {
+
+	curve_create->set_pressed(p_idx==0);
+	curve_edit->set_pressed(p_idx==1);
+	curve_del->set_pressed(p_idx==2);
+}
+
+void PathEditorPlugin::_close_curve() {
+
+    Ref<Curve3D> c = path->get_curve();
+    if (c.is_null())
+        return ;
+    if (c->get_point_count()<2)
+        return;
+    c->add_point(c->get_point_pos(0),c->get_point_in(0),c->get_point_out(0));
+
+}
+
+void PathEditorPlugin::_notification(int p_what) {
+
+	if (p_what==NOTIFICATION_ENTER_TREE) {
+
+		curve_create->connect("pressed",this,"_mode_changed",make_binds(0));
+		curve_edit->connect("pressed",this,"_mode_changed",make_binds(1));
+		curve_del->connect("pressed",this,"_mode_changed",make_binds(2));
+        curve_close->connect("pressed",this,"_close_curve");
+    }
+}
+
+void PathEditorPlugin::_bind_methods() {
+
+	ObjectTypeDB::bind_method(_MD("_mode_changed"),&PathEditorPlugin::_mode_changed);
+    ObjectTypeDB::bind_method(_MD("_close_curve"),&PathEditorPlugin::_close_curve);
+}
+
+PathEditorPlugin* PathEditorPlugin::singleton=NULL;
+
+
+PathEditorPlugin::PathEditorPlugin(EditorNode *p_node) {
+
+	path=NULL;
+	editor=p_node;
+	singleton=this;
+
+	path_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	path_material->set_parameter( FixedMaterial::PARAM_DIFFUSE,Color(0.5,0.5,1.0,0.8) );
+	path_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	path_material->set_line_width(3);
+	path_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);
+	path_material->set_flag(Material::FLAG_UNSHADED,true);
+
+	path_thin_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	path_thin_material->set_parameter( FixedMaterial::PARAM_DIFFUSE,Color(0.5,0.5,1.0,0.4) );
+	path_thin_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	path_thin_material->set_line_width(1);
+	path_thin_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);
+	path_thin_material->set_flag(Material::FLAG_UNSHADED,true);
+
+	SpatialEditor::get_singleton()->add_gizmo_plugin(this);
+
+	sep = memnew( VSeparator);
+	sep->hide();
+	SpatialEditor::get_singleton()->add_control_to_menu_panel(sep);
+	curve_edit = memnew( ToolButton );
+	curve_edit->set_icon(SpatialEditor::get_singleton()->get_icon("CurveEdit","EditorIcons"));
+	curve_edit->set_toggle_mode(true);
+	curve_edit->hide();
+	curve_edit->set_focus_mode(Control::FOCUS_NONE);
+	curve_edit->set_tooltip("Select Points\nShift+Drag: Select Control Points\n"+keycode_get_string(KEY_MASK_CMD)+"Click: Add Point\nRight Click: Delete Point.");
+	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_edit);
+	curve_create = memnew( ToolButton );
+	curve_create->set_icon(SpatialEditor::get_singleton()->get_icon("CurveCreate","EditorIcons"));
+	curve_create->set_toggle_mode(true);
+	curve_create->hide();
+	curve_create->set_focus_mode(Control::FOCUS_NONE);
+	curve_create->set_tooltip("Add Point (in empty space)\nSplit Segment (in curve).");
+	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_create);
+	curve_del = memnew( ToolButton );
+	curve_del->set_icon(SpatialEditor::get_singleton()->get_icon("CurveDelete","EditorIcons"));
+	curve_del->set_toggle_mode(true);
+	curve_del->hide();
+	curve_del->set_focus_mode(Control::FOCUS_NONE);
+	curve_del->set_tooltip("Delete Point.");
+	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_del);
+	curve_close = memnew( ToolButton );
+	curve_close->set_icon(SpatialEditor::get_singleton()->get_icon("CurveClose","EditorIcons"));
+	curve_close->hide();
+	curve_close->set_focus_mode(Control::FOCUS_NONE);
+	curve_close->set_tooltip("Close Curve");
+	SpatialEditor::get_singleton()->add_control_to_menu_panel(curve_close);
+
+
+
+	curve_edit->set_pressed(true);
+	/*
+	collision_polygon_editor = memnew( PathEditor(p_node) );
+	editor->get_viewport()->add_child(collision_polygon_editor);
+
+	collision_polygon_editor->set_margin(MARGIN_LEFT,200);
+	collision_polygon_editor->set_margin(MARGIN_RIGHT,230);
+	collision_polygon_editor->set_margin(MARGIN_TOP,0);
+	collision_polygon_editor->set_margin(MARGIN_BOTTOM,10);
+
+
+	collision_polygon_editor->hide();
+	*/
+
+
+}
+
+
+PathEditorPlugin::~PathEditorPlugin()
+{
+}
+
diff --git a/tools/editor/property_editor.cpp b/tools/editor/property_editor.cpp
index b513e32c13..1a07f30e4e 100644
--- a/tools/editor/property_editor.cpp
+++ b/tools/editor/property_editor.cpp
@@ -42,6 +42,8 @@
 #include "multi_node_edit.h"
 #include "array_property_edit.h"
 #include "editor_help.h"
+#include "scene/resources/packed_scene.h"
+
 
 void CustomPropertyEditor::_notification(int p_what) {
 	
@@ -1655,25 +1657,101 @@ CustomPropertyEditor::CustomPropertyEditor() {
 	menu->connect("item_pressed",this,"_menu_option");
 }
 
+bool PropertyEditor::_might_be_in_instance() {
 
-
-Node *PropertyEditor::get_instanced_node() {
-
-	//this sucks badly
 	if (!obj)
 		return NULL;
 
 	Node *node = obj->cast_to<Node>();
+
+	Node* edited_scene =EditorNode::get_singleton()->get_edited_scene();
+
+	bool might_be=false;
+
+	while(node) {
+
+		if (node->get_scene_instance_state().is_valid()) {
+			might_be=true;
+			break;
+		}
+		if (node==edited_scene) {
+			if (node->get_scene_inherited_state().is_valid()) {
+				might_be=true;
+				break;
+			}
+			might_be=false;
+			break;
+		}
+		node=node->get_owner();
+	}
+
+	return might_be;
+
+}
+
+bool PropertyEditor::_get_instanced_node_original_property(const StringName& p_prop,Variant& value) {
+
+	Node *node = obj->cast_to<Node>();
+
 	if (!node)
-		return NULL;
+		return false;
 
-	if (node->get_filename()=="")
-		return NULL;
+	Node *orig=node;
+
+	Node* edited_scene =EditorNode::get_singleton()->get_edited_scene();
+
+	bool found=false;
 
-	if (!node->get_owner())
-		return NULL; //scene root i guess
+//	print_line("for prop - "+String(p_prop));
 
-	return node;
+	while(node) {
+
+		Ref<SceneState> ss;
+
+		if (node==edited_scene) {
+			ss=node->get_scene_inherited_state();
+		} else {
+			ss=node->get_scene_instance_state();
+		}
+//		print_line("at - "+String(edited_scene->get_path_to(node)));
+
+		if (ss.is_valid()) {
+			NodePath np = node->get_path_to(orig);
+			int node_idx = ss->find_node_by_path(np);
+//			print_line("\t valid, nodeidx "+itos(node_idx));
+			if (node_idx>=0) {
+				bool lfound=false;
+				Variant lvar;
+				lvar=ss->get_property_value(node_idx,p_prop,lfound);
+				if (lfound) {
+					found=true;
+					value=lvar;
+//					print_line("\t found value "+String(value));
+				}
+			}
+		}
+		if (node==edited_scene) {
+			//just in case
+			break;
+		}
+		node=node->get_owner();
+
+	}
+
+	return found;
+}
+
+bool PropertyEditor::_is_property_different(const Variant& p_current, const Variant& p_orig,int p_usage) {
+
+	if (p_orig.get_type()==Variant::NIL) {
+		//special cases
+		if (p_current.is_zero() && p_usage&PROPERTY_USAGE_STORE_IF_NONZERO)
+			return false;
+		if (p_current.is_one() && p_usage&PROPERTY_USAGE_STORE_IF_NONONE)
+			return false;
+	}
+
+	return bool(Variant::evaluate(Variant::OP_NOT_EQUAL,p_current,p_orig));
 }
 
 TreeItem *PropertyEditor::find_item(TreeItem *p_item,const String& p_name) {
@@ -1910,6 +1988,8 @@ void PropertyEditor::set_item_text(TreeItem *p_item, int p_type, const String& p
 	
 }
 
+
+
 void PropertyEditor::_notification(int p_what) {
 
 	if (p_what==NOTIFICATION_ENTER_TREE) {
@@ -1950,12 +2030,16 @@ void PropertyEditor::_notification(int p_what) {
 				if (!item)
 					continue;
 				
-				if (get_instanced_node()) {
+				if (_might_be_in_instance()) {
+
+
+					Variant vorig;
+					Dictionary d=item->get_metadata(0);
+					int usage = d.has("usage")?int(int(d["usage"])&(PROPERTY_USAGE_STORE_IF_NONONE|PROPERTY_USAGE_STORE_IF_NONZERO)):0;
 
-					Dictionary d = get_instanced_node()->get_instance_state();
-					if (d.has(*k)) {
+
+					if (_get_instanced_node_original_property(*k,vorig) || usage) {
 						Variant v = obj->get(*k);
-						Variant vorig = d[*k];
 
 						int found=-1;
 						for(int i=0;i<item->get_button_count(1);i++) {
@@ -1966,7 +2050,7 @@ void PropertyEditor::_notification(int p_what) {
 							}
 						}
 
-						bool changed = ! (v==vorig);
+						bool changed = _is_property_different(v,vorig,usage);
 
 						if ((found!=-1)!=changed) {
 
@@ -2049,12 +2133,14 @@ void PropertyEditor::_refresh_item(TreeItem *p_item) {
 
 	if (name!=String()) {
 
-		if (get_instanced_node()) {
+		if (_might_be_in_instance()) {
+
+			Variant vorig;
+			Dictionary d=p_item->get_metadata(0);
+			int usage = d.has("usage")?int(int(d["usage"])&(PROPERTY_USAGE_STORE_IF_NONONE|PROPERTY_USAGE_STORE_IF_NONZERO)):0;
 
-			Dictionary d = get_instanced_node()->get_instance_state();
-			if (d.has(name)) {
+			if (_get_instanced_node_original_property(name,vorig) || usage) {
 				Variant v = obj->get(name);
-				Variant vorig = d[name];
 
 				int found=-1;
 				for(int i=0;i<p_item->get_button_count(1);i++) {
@@ -2065,7 +2151,7 @@ void PropertyEditor::_refresh_item(TreeItem *p_item) {
 					}
 				}
 
-				bool changed = ! (v==vorig);
+				bool changed = _is_property_different(v,vorig,usage);
 
 				if ((found!=-1)!=changed) {
 
@@ -2326,7 +2412,8 @@ void PropertyEditor::update_tree() {
 		d["type"]=(int)p.type;
 		d["hint"]=(int)p.hint;
 		d["hint_text"]=p.hint_string;
-					
+		d["usage"]=(int)p.usage;
+
 		item->set_metadata( 0, d );
 		item->set_metadata( 1, p.name );
 
@@ -2777,14 +2864,17 @@ void PropertyEditor::update_tree() {
 			}
 		}
 
-		if (get_instanced_node()) {
+		if (_might_be_in_instance()) {
 
-			Dictionary d = get_instanced_node()->get_instance_state();
-			if (d.has(p.name)) {
+			Variant vorig;
+			Dictionary d=item->get_metadata(0);
+			int usage = d.has("usage")?int(int(d["usage"])&(PROPERTY_USAGE_STORE_IF_NONONE|PROPERTY_USAGE_STORE_IF_NONZERO)):0;
+			if (_get_instanced_node_original_property(p.name,vorig) || usage) {
 				Variant v = obj->get(p.name);
-				Variant vorig = d[p.name];
-				if (! (v==vorig)) {
+				
 
+				if (_is_property_different(v,vorig,usage)) {
+					//print_line("FOR "+String(p.name)+" RELOAD WITH: "+String(v)+"("+Variant::get_type_name(v.get_type())+")=="+String(vorig)+"("+Variant::get_type_name(vorig.get_type())+")");
 					item->add_button(1,get_icon("Reload","EditorIcons"),3);
 				}
 			}
@@ -3081,17 +3171,18 @@ void PropertyEditor::_edit_button(Object *p_item, int p_column, int p_button) {
 		call_deferred("_set_range_def",ti,prop,ti->get_range(p_column)+1.0);
 	} else if (p_button==3) {
 
-		if (!get_instanced_node())
+		if (!_might_be_in_instance())
 			return;
 		if (!d.has("name"))
 			return;
 
 		String prop=d["name"];
 
-		Dictionary d2 = get_instanced_node()->get_instance_state();
-		if (d2.has(prop)) {
+		Variant vorig;
+
+		if (_get_instanced_node_original_property(prop,vorig)) {
 
-			_edit_set(prop,d2[prop]);
+			_edit_set(prop,vorig);
 		}
 
 	} else {
diff --git a/tools/editor/property_editor.h b/tools/editor/property_editor.h
index 36ecc794ed..9686bf0bd7 100644
--- a/tools/editor/property_editor.h
+++ b/tools/editor/property_editor.h
@@ -121,6 +121,7 @@ class CustomPropertyEditor : public Popup {
 	void show_value_editors(int p_amount);
 	void config_value_editors(int p_amount, int p_columns,int p_label_w,const List<String>& p_strings);
 	void config_action_buttons(const List<String>& p_strings);
+
 protected:
 
 	void _notification(int p_what);
@@ -184,13 +185,17 @@ class PropertyEditor : public Control {
 	virtual void _changed_callback(Object *p_changed,const char * p_what);
 	virtual void _changed_callbacks(Object *p_changed,const String& p_callback);
 
+
 	void _edit_button(Object *p_item, int p_column, int p_button);
 	
 	void _node_removed(Node *p_node);
 	void _edit_set(const String& p_name, const Variant& p_value);
 	void _draw_flags(Object *ti,const Rect2& p_rect);
 
-	Node *get_instanced_node();
+	bool _might_be_in_instance();
+	bool _get_instanced_node_original_property(const StringName& p_prop,Variant& value);
+	bool _is_property_different(const Variant& p_current, const Variant& p_orig,int p_usage=0);
+
 	void _refresh_item(TreeItem *p_item);
 	void _set_range_def(Object *p_item, String prop, float p_frame);
 
diff --git a/tools/editor/reparent_dialog.cpp b/tools/editor/reparent_dialog.cpp
index 6d0c5b867e..78ba47d54b 100644
--- a/tools/editor/reparent_dialog.cpp
+++ b/tools/editor/reparent_dialog.cpp
@@ -103,6 +103,7 @@ ReparentDialog::ReparentDialog() {
 	add_child(node_only);
 	node_only->hide();
 
+	tree->set_show_enabled_subscene(true);
 	//vbc->add_margin_child("Options:",node_only);;
 	
 
diff --git a/tools/editor/scene_tree_dock.cpp b/tools/editor/scene_tree_dock.cpp
index 276f2dea33..920ab599e9 100644
--- a/tools/editor/scene_tree_dock.cpp
+++ b/tools/editor/scene_tree_dock.cpp
@@ -36,8 +36,8 @@
 #include "script_editor_debugger.h"
 #include "tools/editor/plugins/script_editor_plugin.h"
 #include "multi_node_edit.h"
-void SceneTreeDock::_unhandled_key_input(InputEvent p_event) {
 
+void SceneTreeDock::_unhandled_key_input(InputEvent p_event) {
 
 	uint32_t sc = p_event.key.get_scancode_with_modifiers();
 	if (!p_event.key.pressed || p_event.key.echo)
@@ -71,7 +71,7 @@ Node* SceneTreeDock::instance(const String& p_file) {
 	Node*instanced_scene=NULL;
 	Ref<PackedScene> sdata = ResourceLoader::load(p_file);
 	if (sdata.is_valid())
-		instanced_scene=sdata->instance();
+		instanced_scene=sdata->instance(true);
 
 
 	if (!instanced_scene) {
@@ -96,7 +96,7 @@ Node* SceneTreeDock::instance(const String& p_file) {
 		}
 	}
 
-	instanced_scene->generate_instance_state();
+	//instanced_scene->generate_instance_state();
 	instanced_scene->set_filename( Globals::get_singleton()->localize_path(p_file) );
 
 	editor_data->get_undo_redo().create_action("Instance Scene");
@@ -158,8 +158,8 @@ void SceneTreeDock::_tool_selected(int p_tool, bool p_confirm_override) {
 		case TOOL_NEW: {
 
 
-			if (!_validate_no_foreign())
-				break;
+			//if (!_validate_no_foreign())
+			//	break;
 			create_dialog->popup_centered_ratio();
 		} break;
 		case TOOL_INSTANCE: {
@@ -176,8 +176,8 @@ void SceneTreeDock::_tool_selected(int p_tool, bool p_confirm_override) {
 				break;
 			}
 
-			if (!_validate_no_foreign())
-				break;
+			//if (!_validate_no_foreign())
+			//	break;
 
 			file->set_mode(EditorFileDialog::MODE_OPEN_FILE);
 			List<String> extensions;
@@ -202,8 +202,8 @@ void SceneTreeDock::_tool_selected(int p_tool, bool p_confirm_override) {
 			if (!current)
 				break;
 
-			if (!_validate_no_foreign())
-				break;
+			//if (!_validate_no_foreign())
+			//	break;
 			connect_dialog->popup_centered_ratio();
 			connect_dialog->set_node(current);
 
@@ -213,8 +213,8 @@ void SceneTreeDock::_tool_selected(int p_tool, bool p_confirm_override) {
 			Node *current = scene_tree->get_selected();
 			if (!current)
 				break;
-			if (!_validate_no_foreign())
-				break;
+			//if (!_validate_no_foreign())
+			//	break;
 			groups_editor->set_current(current);
 			groups_editor->popup_centered_ratio();
 		} break;
@@ -224,8 +224,8 @@ void SceneTreeDock::_tool_selected(int p_tool, bool p_confirm_override) {
 			if (!selected)
 				break;
 
-			if (!_validate_no_foreign())
-				break;
+			//if (!_validate_no_foreign())
+			//	break;
 
 			Ref<Script> existing = selected->get_script();
 			if (existing.is_valid())
@@ -573,9 +573,9 @@ Node *SceneTreeDock::_duplicate(Node *p_node, Map<Node*,Node*> &duplimap) {
 
 		Ref<PackedScene> sd = ResourceLoader::load( p_node->get_filename() );
 		ERR_FAIL_COND_V(!sd.is_valid(),NULL);
-		node = sd->instance();
+		node = sd->instance(true);
 		ERR_FAIL_COND_V(!node,NULL);
-		node->generate_instance_state();
+		//node->generate_instance_state();
 	} else {
 		Object *obj = ObjectTypeDB::instance(p_node->get_type());
 		ERR_FAIL_COND_V(!obj,NULL);
@@ -874,6 +874,16 @@ bool SceneTreeDock::_validate_no_foreign() {
 			return false;
 
 		}
+
+		if (edited_scene->get_scene_instance_state().is_valid() && edited_scene->get_scene_instance_state()->find_node_by_path(edited_scene->get_path_to(E->get()))>=0) {
+
+			accept->get_ok()->set_text("Makes Sense!");
+			accept->set_text("Can't operate on nodes the current scene inherits from!");
+			accept->popup_centered_minsize();
+			return false;
+
+		}
+
 	}
 
 	return true;
@@ -1078,14 +1088,15 @@ void SceneTreeDock::_delete_confirm() {
 void SceneTreeDock::_update_tool_buttons() {
 
 	Node *sel = scene_tree->get_selected();
-	bool disable = !sel || (sel!=edited_scene && sel->get_owner()!=edited_scene);
+	bool disable = !sel || (sel!=edited_scene && sel->get_owner()!=edited_scene) || (edited_scene->get_scene_instance_state().is_valid() && edited_scene->get_scene_instance_state()->find_node_by_path(edited_scene->get_path_to(sel))>=0);
 	bool disable_root = disable || sel->get_parent()==scene_root;
+	bool disable_edit = !sel;
 
-	tool_buttons[TOOL_INSTANCE]->set_disabled(disable);
+	tool_buttons[TOOL_INSTANCE]->set_disabled(disable_edit);
 	tool_buttons[TOOL_REPLACE]->set_disabled(disable);
-	tool_buttons[TOOL_CONNECT]->set_disabled(disable);
-	tool_buttons[TOOL_GROUP]->set_disabled(disable);
-	tool_buttons[TOOL_SCRIPT]->set_disabled(disable);
+	tool_buttons[TOOL_CONNECT]->set_disabled(disable_edit);
+	tool_buttons[TOOL_GROUP]->set_disabled(disable_edit);
+	tool_buttons[TOOL_SCRIPT]->set_disabled(disable_edit);
 	tool_buttons[TOOL_MOVE_UP]->set_disabled(disable_root);
 	tool_buttons[TOOL_MOVE_DOWN]->set_disabled(disable_root);
 	tool_buttons[TOOL_DUPLICATE]->set_disabled(disable_root);
diff --git a/tools/editor/scene_tree_editor.cpp b/tools/editor/scene_tree_editor.cpp
index fd841028a2..b746af7f0e 100644
--- a/tools/editor/scene_tree_editor.cpp
+++ b/tools/editor/scene_tree_editor.cpp
@@ -34,6 +34,8 @@
 #include "scene/main/viewport.h"
 #include "tools/editor/plugins/canvas_item_editor_plugin.h"
 
+#include "scene/resources/packed_scene.h"
+
 Node *SceneTreeEditor::get_scene_node() {
 
 	ERR_FAIL_COND_V(!is_inside_tree(),NULL);
@@ -57,22 +59,58 @@ void SceneTreeEditor::_subscene_option(int p_idx) {
 
 	switch(p_idx) {
 
-		case SCENE_MENU_SHOW_CHILDREN: {
+		case SCENE_MENU_EDITABLE_CHILDREN: {
 
-			if (node->has_meta("__editor_show_subtree")) {
-				instance_menu->set_item_checked(0,true);
-				node->set_meta("__editor_show_subtree",Variant());
-				_update_tree();
-			} else {
-				node->set_meta("__editor_show_subtree",true);
-				_update_tree();
+			bool editable = EditorNode::get_singleton()->get_edited_scene()->is_editable_instance(node);
+			editable = !editable;
+
+			//node->set_instance_children_editable(editable);
+			EditorNode::get_singleton()->get_edited_scene()->set_editable_instance(node,editable);
+			instance_menu->set_item_checked(0,editable);
+			if (editable) {
+				node->set_scene_instance_load_placeholder(false);
+				instance_menu->set_item_checked(1,false);
+			}
+
+			_update_tree();
+
+		} break;
+		case SCENE_MENU_USE_PLACEHOLDER: {
+
+			bool placeholder = node->get_scene_instance_load_placeholder();
+			placeholder = !placeholder;
+
+			//node->set_instance_children_editable(editable);
+			if (placeholder) {
+				EditorNode::get_singleton()->get_edited_scene()->set_editable_instance(node,false);
 			}
+			node->set_scene_instance_load_placeholder(placeholder);
+			instance_menu->set_item_checked(0,false);
+			instance_menu->set_item_checked(1,placeholder);
+
+			_update_tree();
 
 		} break;
 		case SCENE_MENU_OPEN: {
 
 			emit_signal("open",node->get_filename());
 		} break;
+		case SCENE_MENU_CLEAR_INHERITANCE: {
+			clear_inherit_confirm->popup_centered_minsize();
+		} break;
+		case SCENE_MENU_OPEN_INHERITED: {
+			if (node && node->get_scene_inherited_state().is_valid()) {
+				emit_signal("open",node->get_scene_inherited_state()->get_path());
+			}
+		} break;
+		case SCENE_MENU_CLEAR_INHERITANCE_CONFIRM: {
+			if (node && node->get_scene_inherited_state().is_valid()) {
+				node->set_scene_inherited_state(Ref<SceneState>());
+				update_tree();
+				EditorNode::get_singleton()->get_property_editor()->update_tree();
+			}
+
+		} break;
 
 	}
 
@@ -94,15 +132,33 @@ void SceneTreeEditor::_cell_button_pressed(Object *p_item,int p_column,int p_id)
 		Rect2 item_rect = tree->get_item_rect(item,0);
 		item_rect.pos.y-=tree->get_scroll().y;
 		item_rect.pos+=tree->get_global_pos();
-		instance_menu->set_pos(item_rect.pos+Vector2(0,item_rect.size.y));
-		instance_menu->set_size(Vector2(item_rect.size.x,0));
-		if (n->has_meta("__editor_show_subtree"))
-			instance_menu->set_item_checked(0,true);
-		else
-			instance_menu->set_item_checked(0,false);
 
-		instance_menu->popup();
-		instance_node=n->get_instance_ID();
+		if (n==get_scene_node()) {
+			inheritance_menu->set_pos(item_rect.pos+Vector2(0,item_rect.size.y));
+			inheritance_menu->set_size(Vector2(item_rect.size.x,0));
+			inheritance_menu->popup();
+			instance_node=n->get_instance_ID();
+
+		} else {
+			instance_menu->set_pos(item_rect.pos+Vector2(0,item_rect.size.y));
+			instance_menu->set_size(Vector2(item_rect.size.x,0));
+			if (EditorNode::get_singleton()->get_edited_scene()->is_editable_instance(n))
+				instance_menu->set_item_checked(0,true);
+			else
+				instance_menu->set_item_checked(0,false);
+
+			if (n->get_owner()==get_scene_node()) {
+				instance_menu->set_item_checked(1,n->get_scene_instance_load_placeholder());
+				instance_menu->set_item_disabled(1,false);
+			} else {
+
+				instance_menu->set_item_checked(1,false);
+				instance_menu->set_item_disabled(1,true);
+			}
+
+			instance_menu->popup();
+			instance_node=n->get_instance_ID();
+		}
 		//emit_signal("open",n->get_filename());
 	} else if (p_id==BUTTON_SCRIPT) {
 		RefPtr script=n->get_script();
@@ -168,15 +224,17 @@ void SceneTreeEditor::_add_nodes(Node *p_node,TreeItem *p_parent) {
 
 	bool part_of_subscene=false;
 
-	if (!display_foreign && p_node->get_owner()!=get_scene_node() && p_node!=get_scene_node()) {
+	if (!display_foreign && p_node->get_owner()!=get_scene_node()  && p_node!=get_scene_node()) {
 
-		if ((show_enabled_subscene || can_open_instance) && p_node->get_owner() && p_node->get_owner()->get_owner()==get_scene_node() && p_node->get_owner()->has_meta("__editor_show_subtree")) {
+		if ((show_enabled_subscene || can_open_instance) && p_node->get_owner() && (get_scene_node()->is_editable_instance(p_node->get_owner()))) {
 
 			part_of_subscene=true;
 			//allow
 		} else {
 			return;
 		}
+	} else {
+		part_of_subscene = get_scene_node()->get_scene_inherited_state().is_valid() && get_scene_node()->get_scene_inherited_state()->find_node_by_path(get_scene_node()->get_path_to(p_node))>=0;
 	}
 
 	TreeItem *item = tree->create_item(p_parent);
@@ -199,6 +257,7 @@ void SceneTreeEditor::_add_nodes(Node *p_node,TreeItem *p_parent) {
 		icon=get_icon( (has_icon(p_node->get_type(),"EditorIcons")?p_node->get_type():String("Object")),"EditorIcons");
 	item->set_icon(0, icon );
 	item->set_metadata( 0,p_node->get_path() );	
+
 	if (part_of_subscene) {
 
 		//item->set_selectable(0,marked_selectable);
@@ -221,7 +280,10 @@ void SceneTreeEditor::_add_nodes(Node *p_node,TreeItem *p_parent) {
 		}
 	}
 
-	if (p_node!=get_scene_node() && p_node->get_filename()!="" && can_open_instance) {
+	if (p_node==get_scene_node() && p_node->get_scene_inherited_state().is_valid()) {
+		item->add_button(0,get_icon("InstanceOptions","EditorIcons"),BUTTON_SUBSCENE);
+		item->set_tooltip(0,"Inherits: "+p_node->get_scene_inherited_state()->get_path()+"\nType: "+p_node->get_type());
+	} else if (p_node!=get_scene_node() && p_node->get_filename()!="" && can_open_instance) {
 
 		item->add_button(0,get_icon("InstanceOptions","EditorIcons"),BUTTON_SUBSCENE);
 		item->set_tooltip(0,"Instance: "+p_node->get_filename()+"\nType: "+p_node->get_type());
@@ -487,8 +549,11 @@ void SceneTreeEditor::_notification(int p_what) {
 
 		get_tree()->connect("tree_changed",this,"_tree_changed");
 		get_tree()->connect("node_removed",this,"_node_removed");
-		instance_menu->set_item_icon(2,get_icon("Load","EditorIcons"));
+		instance_menu->set_item_icon(3,get_icon("Load","EditorIcons"));
 		tree->connect("item_collapsed",this,"_cell_collapsed");
+		inheritance_menu->set_item_icon(2,get_icon("Load","EditorIcons"));
+		clear_inherit_confirm->connect("confirmed",this,"_subscene_option",varray(SCENE_MENU_CLEAR_INHERITANCE_CONFIRM));
+
 
 //		get_scene()->connect("tree_changed",this,"_tree_changed",Vector<Variant>(),CONNECT_DEFERRED);
 //		get_scene()->connect("node_removed",this,"_node_removed",Vector<Variant>(),CONNECT_DEFERRED);
@@ -499,6 +564,7 @@ void SceneTreeEditor::_notification(int p_what) {
 		get_tree()->disconnect("tree_changed",this,"_tree_changed");
 		get_tree()->disconnect("node_removed",this,"_node_removed");
 		tree->disconnect("item_collapsed",this,"_cell_collapsed");
+		clear_inherit_confirm->disconnect("confirmed",this,"_subscene_option");
 		_update_tree();
 	}
 
@@ -788,12 +854,27 @@ SceneTreeEditor::SceneTreeEditor(bool p_label,bool p_can_rename, bool p_can_open
 	blocked=0;
 
 	instance_menu = memnew( PopupMenu );
-	instance_menu->add_check_item("Show Children",SCENE_MENU_SHOW_CHILDREN);
+	instance_menu->add_check_item("Editable Children",SCENE_MENU_EDITABLE_CHILDREN);
+	instance_menu->add_check_item("Load As Placeholder",SCENE_MENU_USE_PLACEHOLDER);
 	instance_menu->add_separator();
 	instance_menu->add_item("Open in Editor",SCENE_MENU_OPEN);
 	instance_menu->connect("item_pressed",this,"_subscene_option");
 	add_child(instance_menu);
 
+	inheritance_menu = memnew( PopupMenu );
+	inheritance_menu->add_item("Clear Inheritance",SCENE_MENU_CLEAR_INHERITANCE);
+	inheritance_menu->add_separator();
+	inheritance_menu->add_item("Open in Editor",SCENE_MENU_OPEN_INHERITED);
+	inheritance_menu->connect("item_pressed",this,"_subscene_option");
+
+	add_child(inheritance_menu);
+
+	clear_inherit_confirm = memnew( ConfirmationDialog );
+	clear_inherit_confirm->set_text("Clear Inheritance? (No Undo!)");
+	clear_inherit_confirm->get_ok()->set_text("Clear!");
+	add_child(clear_inherit_confirm);
+
+
 }
 
 
diff --git a/tools/editor/scene_tree_editor.h b/tools/editor/scene_tree_editor.h
index b05a52a2da..50cca4e24b 100644
--- a/tools/editor/scene_tree_editor.h
+++ b/tools/editor/scene_tree_editor.h
@@ -52,16 +52,22 @@ class SceneTreeEditor : public Control {
 	};
 
 	enum {
-		SCENE_MENU_SHOW_CHILDREN,
+		SCENE_MENU_EDITABLE_CHILDREN,
+		SCENE_MENU_USE_PLACEHOLDER,
 		SCENE_MENU_OPEN,
+		SCENE_MENU_CLEAR_INHERITANCE,
+		SCENE_MENU_OPEN_INHERITED,
+		SCENE_MENU_CLEAR_INHERITANCE_CONFIRM,
 	};
 
 	Tree *tree;
 	Node *selected;
 	PopupMenu *instance_menu;
+	PopupMenu *inheritance_menu;
 	ObjectID instance_node;
 
 	AcceptDialog *error;
+	ConfirmationDialog *clear_inherit_confirm;
 
 	int blocked;
 	
diff --git a/tools/editor/spatial_editor_gizmos.cpp b/tools/editor/spatial_editor_gizmos.cpp
index 521a10bbd0..4dc9c4f43e 100644
--- a/tools/editor/spatial_editor_gizmos.cpp
+++ b/tools/editor/spatial_editor_gizmos.cpp
@@ -1,3191 +1,3191 @@
-/*************************************************************************/

-/*  spatial_editor_gizmos.cpp                                            */

-/*************************************************************************/

-/*                       This file is part of:                           */

-/*                           GODOT ENGINE                                */

-/*                    http://www.godotengine.org                         */

-/*************************************************************************/

-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */

-/*                                                                       */

-/* Permission is hereby granted, free of charge, to any person obtaining */

-/* a copy of this software and associated documentation files (the       */

-/* "Software"), to deal in the Software without restriction, including   */

-/* without limitation the rights to use, copy, modify, merge, publish,   */

-/* distribute, sublicense, and/or sell copies of the Software, and to    */

-/* permit persons to whom the Software is furnished to do so, subject to */

-/* the following conditions:                                             */

-/*                                                                       */

-/* The above copyright notice and this permission notice shall be        */

-/* included in all copies or substantial portions of the Software.       */

-/*                                                                       */

-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */

-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */

-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/

-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */

-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */

-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */

-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */

-/*************************************************************************/

-#include "spatial_editor_gizmos.h"

-#include "geometry.h"

-#include "scene/3d/camera.h"

-#include "scene/resources/surface_tool.h"

-#include "scene/resources/sphere_shape.h"

-#include "scene/resources/box_shape.h"

-#include "scene/resources/capsule_shape.h"

-#include "scene/resources/ray_shape.h"

-#include "scene/resources/convex_polygon_shape.h"

-#include "scene/resources/plane_shape.h"

-#include "quick_hull.h"

-

-// Keep small children away from this file.

-// It's so ugly it will eat them alive

-

-#define HANDLE_HALF_SIZE 0.05

-

-void SpatialGizmoTool::clear() {

-

-	for(int i=0;i<instances.size();i++) {

-

-		if (instances[i].instance.is_valid())

-			VS::get_singleton()->free(instances[i].instance);

-

-

-	}

-

-	billboard_handle=false;

-	collision_segments.clear();

-	collision_mesh=Ref<TriangleMesh>();

-	instances.clear();

-	handles.clear();

-	secondary_handles.clear();

-}

-

-void SpatialGizmoTool::Instance::create_instance(Spatial *p_base) {

-

-	instance = VS::get_singleton()->instance_create2(mesh->get_rid(),p_base->get_world()->get_scenario());

-	VS::get_singleton()->instance_attach_object_instance_ID(instance,p_base->get_instance_ID());

-	if (billboard)

-		VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_BILLBOARD,true);

-	if (unscaled)

-		VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_DEPH_SCALE,true);

-	if (skeleton.is_valid())

-		VS::get_singleton()->instance_attach_skeleton(instance,skeleton);

-	if (extra_margin)

-		VS::get_singleton()->instance_set_extra_visibility_margin(instance,1);

-	VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_CAST_SHADOW,false);

-	VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_RECEIVE_SHADOWS,false);

-	VS::get_singleton()->instance_set_layer_mask(instance,1<<SpatialEditorViewport::GIZMO_EDIT_LAYER); //gizmos are 26

-}

-

-

-

-void SpatialGizmoTool::add_mesh(const Ref<Mesh>& p_mesh,bool p_billboard, const RID &p_skeleton) {

-

-	ERR_FAIL_COND(!spatial_node);

-	Instance ins;

-

-	ins.billboard=p_billboard;

-	ins.mesh=p_mesh;

-	ins.skeleton=p_skeleton;

-	if (valid) {

-		ins.create_instance(spatial_node);

-		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());

-	}

-

-	instances.push_back(ins);

-

-}

-

-void SpatialGizmoTool::add_lines(const Vector<Vector3> &p_lines, const Ref<Material> &p_material,bool p_billboard){

-

-	ERR_FAIL_COND(!spatial_node);

-	Instance ins;

-

-	Ref<Mesh> mesh = memnew( Mesh );

-	Array a;

-	a.resize(Mesh::ARRAY_MAX);

-

-	a[Mesh::ARRAY_VERTEX]=p_lines;

-

-	DVector<Color> color;

-	color.resize(p_lines.size());

-	{

-		DVector<Color>::Write w = color.write();

-		for(int i=0;i<p_lines.size();i++) {

-			if (is_selected())

-				w[i]=Color(1,1,1,0.6);

-			else

-				w[i]=Color(1,1,1,0.25);

-		}

-

-	}

-

-	a[Mesh::ARRAY_COLOR]=color;

-

-

-	mesh->add_surface(Mesh::PRIMITIVE_LINES,a);

-	mesh->surface_set_material(0,p_material);

-

-	if (p_billboard) {

-		float md=0;

-		for(int i=0;i<p_lines.size();i++) {

-

-			md=MAX(0,p_lines[i].length());

-

-		}

-		if (md) {

-			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));

-		}

-	}

-

-	ins.billboard=p_billboard;

-	ins.mesh=mesh;

-	if (valid) {

-		ins.create_instance(spatial_node);

-		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());

-	}

-

-	instances.push_back(ins);

-

-}

-

-void SpatialGizmoTool::add_unscaled_billboard(const Ref<Material>& p_material,float p_scale) {

-

-	ERR_FAIL_COND(!spatial_node);

-	Instance ins;

-

-	Vector<Vector3 > vs;

-	Vector<Vector2 > uv;

-

-	vs.push_back(Vector3(-p_scale,p_scale,0));

-	vs.push_back(Vector3(p_scale,p_scale,0));

-	vs.push_back(Vector3(p_scale,-p_scale,0));

-	vs.push_back(Vector3(-p_scale,-p_scale,0));

-

-	uv.push_back(Vector2(1,0));

-	uv.push_back(Vector2(0,0));

-	uv.push_back(Vector2(0,1));

-	uv.push_back(Vector2(1,1));

-

-	Ref<Mesh> mesh = memnew( Mesh );

-	Array a;

-	a.resize(Mesh::ARRAY_MAX);

-	a[Mesh::ARRAY_VERTEX]=vs;

-	a[Mesh::ARRAY_TEX_UV]=uv;

-	mesh->add_surface(Mesh::PRIMITIVE_TRIANGLE_FAN,a);

-	mesh->surface_set_material(0,p_material);

-

-	if (true) {

-		float md=0;

-		for(int i=0;i<vs.size();i++) {

-

-			md=MAX(0,vs[i].length());

-

-		}

-		if (md) {

-			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));

-		}

-	}

-

-	ins.mesh=mesh;

-	ins.unscaled=true;

-	ins.billboard=true;

-	if (valid) {

-		ins.create_instance(spatial_node);

-		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());

-	}

-

-	instances.push_back(ins);

-

-

-}

-

-void SpatialGizmoTool::add_collision_triangles(const Ref<TriangleMesh>& p_tmesh) {

-

-	collision_mesh=p_tmesh;

-}

-

-void SpatialGizmoTool::add_collision_segments(const Vector<Vector3> &p_lines) {

-

-	int from=collision_segments.size();

-	collision_segments.resize(from+p_lines.size());

-	for(int i=0;i<p_lines.size();i++) {

-

-		collision_segments[from+i]=p_lines[i];

-	}

-}

-

-

-void SpatialGizmoTool::add_handles(const Vector<Vector3> &p_handles, bool p_billboard,bool p_secondary){

-

-	billboard_handle=p_billboard;

-

-	if (!is_selected())

-		return;

-

-	ERR_FAIL_COND(!spatial_node);

-

-	ERR_FAIL_COND(!spatial_node);

-	Instance ins;

-

-

-	Ref<Mesh> mesh = memnew( Mesh );

-#if 1

-

-	Array a;

-	a.resize(VS::ARRAY_MAX);

-	a[VS::ARRAY_VERTEX]=p_handles;

-	DVector<Color> colors;

-	{

-		colors.resize(p_handles.size());

-		DVector<Color>::Write w=colors.write();

-		for(int i=0;i<p_handles.size();i++) {

-

-			Color col(1,1,1,1);

-			if (SpatialEditor::get_singleton()->get_over_gizmo_handle()!=i)

-				col=Color(0.9,0.9,0.9,0.9);

-			w[i]=col;

-		}

-

-	}

-	a[VS::ARRAY_COLOR]=colors;

-	mesh->add_surface(Mesh::PRIMITIVE_POINTS,a);

-	mesh->surface_set_material(0,SpatialEditorGizmos::singleton->handle2_material);

-

-	if (p_billboard) {

-		float md=0;

-		for(int i=0;i<p_handles.size();i++) {

-

-			md=MAX(0,p_handles[i].length());

-

-		}

-		if (md) {

-			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));

-		}

-	}

-

-

-

-#else

-	for(int ih=0;ih<p_handles.size();ih++) {

-

-

-		Vector<Vector3> vertices;

-		Vector<Vector3> normals;

-

-		int vtx_idx=0;

-#define ADD_VTX(m_idx);\

-	vertices.push_back( (face_points[m_idx]*HANDLE_HALF_SIZE+p_handles[ih]) );\

-	normals.push_back( normal_points[m_idx] );\

-	vtx_idx++;\

-

-		for (int i=0;i<6;i++) {

-

-

-			Vector3 face_points[4];

-			Vector3 normal_points[4];

-			float uv_points[8]={0,0,0,1,1,1,1,0};

-

-			for (int j=0;j<4;j++) {

-

-				float v[3];

-				v[0]=1.0;

-				v[1]=1-2*((j>>1)&1);

-				v[2]=v[1]*(1-2*(j&1));

-

-				for (int k=0;k<3;k++) {

-

-					if (i<3)

-						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);

-					else

-						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);

-				}

-				normal_points[j]=Vector3();

-				normal_points[j][i%3]=(i>=3?-1:1);

-			}

-			//tri 1

-			ADD_VTX(0);

-			ADD_VTX(1);

-			ADD_VTX(2);

-			//tri 2

-			ADD_VTX(2);

-			ADD_VTX(3);

-			ADD_VTX(0);

-

-		}

-

-

-		Array d;

-		d.resize(VS::ARRAY_MAX);

-		d[VisualServer::ARRAY_NORMAL]= normals ;

-		d[VisualServer::ARRAY_VERTEX]= vertices ;

-

-		mesh->add_surface(Mesh::PRIMITIVE_TRIANGLES,d);

-		mesh->surface_set_material(ih,SpatialEditorGizmos::singleton->handle_material);

-

-

-	}

-#endif

-	ins.mesh=mesh;

-	ins.billboard=p_billboard;

-	ins.extra_margin=true;

-	if (valid) {

-		ins.create_instance(spatial_node);

-		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());

-	}

-	instances.push_back(ins);

-	if (!p_secondary) {

-		int chs=handles.size();

-		handles.resize(chs+p_handles.size());

-		for(int i=0;i<p_handles.size();i++) {

-			handles[i+chs]=p_handles[i];

-		}

-	} else {

-

-		int chs=secondary_handles.size();

-		secondary_handles.resize(chs+p_handles.size());

-		for(int i=0;i<p_handles.size();i++) {

-			secondary_handles[i+chs]=p_handles[i];

-		}

-

-	}

-

-}

-

-

-void SpatialGizmoTool::set_spatial_node(Spatial *p_node){

-

-	spatial_node=p_node;

-

-}

-

-bool SpatialGizmoTool::intersect_frustum(const Camera *p_camera,const Vector<Plane> &p_frustum) {

-

-	ERR_FAIL_COND_V(!spatial_node,false);

-	ERR_FAIL_COND_V(!valid,false);

-

-	if (collision_segments.size()) {

-

-		const Plane *p=p_frustum.ptr();

-		int fc=p_frustum.size();

-

-		int vc=collision_segments.size();

-		const Vector3* vptr=collision_segments.ptr();

-		Transform t = spatial_node->get_global_transform();

-

-		for(int i=0;i<vc/2;i++) {

-

-

-			Vector3 a=t.xform(vptr[i*2+0]);

-			Vector3 b=t.xform(vptr[i*2+1]);

-

-			bool any_out=false;

-			for(int j=0;j<fc;j++) {

-

-				if (p[j].distance_to(a) > 0 && p[j].distance_to(b) >0) {

-

-					any_out=true;

-					break;

-				}

-			}

-

-			if (!any_out)

-				return true;

-		}

-

-		return false;

-	}

-

-	return false;

-}

-

-

-bool SpatialGizmoTool::intersect_ray(const Camera *p_camera,const Point2& p_point,  Vector3& r_pos, Vector3& r_normal,int *r_gizmo_handle,bool p_sec_first) {

-

-	ERR_FAIL_COND_V(!spatial_node,false);

-	ERR_FAIL_COND_V(!valid,false);

-

-	if (r_gizmo_handle) {

-

-		Transform t = spatial_node->get_global_transform();

-		t.orthonormalize();

-		if (billboard_handle) {

-			t.set_look_at(t.origin,t.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));

-		}

-		Transform ti=t.affine_inverse();

-

-		Vector3 ray_from=ti.xform(p_camera->project_ray_origin(p_point));

-		Vector3 ray_dir=t.basis.xform_inv(p_camera->project_ray_normal(p_point)).normalized();

-		Vector3 ray_to = ray_from+ray_dir*4096;

-

-		float min_d=1e20;

-		int idx=-1;

-

-		for(int i=0;i<secondary_handles.size();i++) {

-#if 1

-

-

-			Vector3 hpos = t.xform(secondary_handles[i]);

-			Vector2 p = p_camera->unproject_position(hpos);

-			if (p.distance_to(p_point)<SpatialEditorGizmos::singleton->handle_t->get_width()*0.6) {

-

-

-				real_t dp = p_camera->get_transform().origin.distance_to(hpos);

-				if (dp<min_d) {

-

-					r_pos=t.xform(hpos);

-					r_normal=p_camera->get_transform().basis.get_axis(2);

-					min_d=dp;

-					idx=i+handles.size();

-

-				}

-

-			}

-

-#else

-			AABB aabb;

-			aabb.pos=Vector3(-1,-1,-1)*HANDLE_HALF_SIZE;

-			aabb.size=aabb.pos*-2;

-			aabb.pos+=secondary_handles[i];

-

-

-			Vector3 rpos,rnorm;

-

-			if (aabb.intersects_segment(ray_from,ray_to,&rpos,&rnorm)) {

-

-				real_t dp = ray_dir.dot(rpos);

-				if (dp<min_d) {

-

-					r_pos=t.xform(rpos);

-					r_normal=ti.basis.xform_inv(rnorm).normalized();

-					min_d=dp;

-					idx=i+handles.size();

-

-				}

-			}

-#endif

-		}

-

-		if (p_sec_first && idx!=-1) {

-

-			*r_gizmo_handle=idx;

-			return true;

-		}

-

-		min_d=1e20;

-

-		for(int i=0;i<handles.size();i++) {

-

-#if 1

-

-

-			Vector3 hpos = t.xform(handles[i]);

-			Vector2 p = p_camera->unproject_position(hpos);

-			if (p.distance_to(p_point)<SpatialEditorGizmos::singleton->handle_t->get_width()*0.6) {

-

-

-				real_t dp = p_camera->get_transform().origin.distance_to(hpos);

-				if (dp<min_d) {

-

-					r_pos=t.xform(hpos);

-					r_normal=p_camera->get_transform().basis.get_axis(2);

-					min_d=dp;

-					idx=i;

-

-				}

-

-			}

-

-#else

-

-			AABB aabb;

-			aabb.pos=Vector3(-1,-1,-1)*HANDLE_HALF_SIZE;

-			aabb.size=aabb.pos*-2;

-			aabb.pos+=handles[i];

-

-

-			Vector3 rpos,rnorm;

-

-			if (aabb.intersects_segment(ray_from,ray_to,&rpos,&rnorm)) {

-

-				real_t dp = ray_dir.dot(rpos);

-				if (dp<min_d) {

-

-					r_pos=t.xform(rpos);

-					r_normal=ti.basis.xform_inv(rnorm).normalized();

-					min_d=dp;

-					idx=i;

-

-				}

-			}

-#endif

-		}

-

-		if (idx>=0) {

-			*r_gizmo_handle=idx;

-			return true;

-		}

-

-

-	}

-

-	if (collision_segments.size()) {

-

-		Plane camp(p_camera->get_transform().origin,(-p_camera->get_transform().basis.get_axis(2)).normalized());

-

-		int vc=collision_segments.size();

-		const Vector3* vptr=collision_segments.ptr();

-		Transform t = spatial_node->get_global_transform();

-		if (billboard_handle) {

-			t.set_look_at(t.origin,t.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));

-		}

-

-		Vector3 cp;

-		float cpd=1e20;

-

-		for(int i=0;i<vc/2;i++) {

-

-

-			Vector3 a=t.xform(vptr[i*2+0]);

-			Vector3 b=t.xform(vptr[i*2+1]);

-			Vector2 s[2];

-			s[0] = p_camera->unproject_position(a);

-			s[1] = p_camera->unproject_position(b);

-

-

-			Vector2 p = Geometry::get_closest_point_to_segment_2d(p_point,s);

-

-			float pd = p.distance_to(p_point);

-

-			if (pd<cpd) {

-

-

-				float d = s[0].distance_to(s[1]);

-				Vector3 tcp;

-				if (d>0) {

-

-					float d2=s[0].distance_to(p)/d;

-					tcp = a+(b-a)*d2;

-

-				} else {

-					tcp=a;

-

-				}

-

-				if (camp.distance_to(tcp)<p_camera->get_znear())

-					continue;

-				cp=tcp;

-				cpd=pd;

-			}

-		}

-

-		if (cpd<8) {

-

-			r_pos=cp;

-			r_normal=-p_camera->project_ray_normal(p_point);

-			return true;

-		}

-

-		return false;

-	}

-

-

-	if (collision_mesh.is_valid()) {

-		Transform gt = spatial_node->get_global_transform();

-

-		if (billboard_handle) {

-			gt.set_look_at(gt.origin,gt.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));

-		}

-

-		Transform ai=gt.affine_inverse();

-		Vector3 ray_from = ai.xform(p_camera->project_ray_origin(p_point));

-		Vector3 ray_dir=ai.basis.xform(p_camera->project_ray_normal(p_point)).normalized();

-		Vector3 rpos,rnorm;

-

-#if 1

-

-

-

-		if (collision_mesh->intersect_ray(ray_from,ray_dir,rpos,rnorm)) {

-

-			r_pos=gt.xform(rpos);

-			r_normal=gt.basis.xform(rnorm).normalized();

-			return true;

-		}

-#else

-

-		if (collision_mesh->intersect_segment(ray_from,ray_from+ray_dir*4906.0,rpos,rnorm)) {

-

-			r_pos=gt.xform(rpos);

-			r_normal=gt.basis.xform(rnorm).normalized();

-			return true;

-		}

-

-#endif

-	}

-

-	return false;

-

-}

-

-

-

-void SpatialGizmoTool::create() {

-

-	ERR_FAIL_COND(!spatial_node);

-	ERR_FAIL_COND(valid);

-	valid=true;

-

-	for(int i=0;i<instances.size();i++) {

-

-		instances[i].create_instance(spatial_node);

-	}

-

-	transform();

-

-}

-

-void SpatialGizmoTool::transform(){

-

-	ERR_FAIL_COND(!spatial_node);

-	ERR_FAIL_COND(!valid);

-	for(int i=0;i<instances.size();i++) {

-		VS::get_singleton()->instance_set_transform(instances[i].instance,spatial_node->get_global_transform());

-	}

-

-}

-

-

-void SpatialGizmoTool::free(){

-

-	ERR_FAIL_COND(!spatial_node);

-	ERR_FAIL_COND(!valid);

-

-	for(int i=0;i<instances.size();i++) {

-

-		if (instances[i].instance.is_valid())

-			VS::get_singleton()->free(instances[i].instance);

-		instances[i].instance=RID();

-	}

-

-	valid=false;

-

-

-}

-

-

-

-SpatialGizmoTool::SpatialGizmoTool() {

-	valid=false;

-	billboard_handle=false;

-

-}

-

-SpatialGizmoTool::~SpatialGizmoTool(){

-

-	clear();

-}

-

-Vector3 SpatialGizmoTool::get_handle_pos(int p_idx) const {

-

-	ERR_FAIL_INDEX_V(p_idx,handles.size(),Vector3());

-

-	return handles[p_idx];

-

-}

-

-//// light gizmo

-

-

-String LightSpatialGizmo::get_handle_name(int p_idx) const {

-

-	if (p_idx==0)

-		return "Radius";

-	else

-		return "Aperture";

-}

-

-

-Variant LightSpatialGizmo::get_handle_value(int p_idx) const{

-

-	if (p_idx==0)

-		return light->get_parameter(Light::PARAM_RADIUS);

-	if (p_idx==1)

-		return light->get_parameter(Light::PARAM_SPOT_ANGLE);

-

-	return Variant();

-}

-

-

-static float _find_closest_angle_to_half_pi_arc(const Vector3& p_from, const Vector3& p_to, float p_arc_radius,const Transform& p_arc_xform) {

-

-	//bleh, discrete is simpler

-	static const int arc_test_points=64;

-	float min_d = 1e20;

-	Vector3 min_p;

-

-

-	for(int i=0;i<arc_test_points;i++) {

-

-		float a = i*Math_PI*0.5/arc_test_points;

-		float an = (i+1)*Math_PI*0.5/arc_test_points;

-		Vector3 p=Vector3( Math::cos(a), 0, -Math::sin(a) )*p_arc_radius;

-		Vector3 n=Vector3( Math::cos(an), 0,- Math::sin(an) )*p_arc_radius;

-

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(p,n,p_from,p_to,ra,rb);

-

-		float d = ra.distance_to(rb);

-		if (d<min_d) {

-			min_d=d;

-			min_p=ra;

-		}

-

-	}

-

-	//min_p = p_arc_xform.affine_inverse().xform(min_p);

-	float a = Vector2(min_p.x,-min_p.z).atan2();

-	return a*180.0/Math_PI;

-}

-

-

-void LightSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point) {

-

-	Transform gt = light->get_global_transform();

-	gt.orthonormalize();

-	Transform gi = gt.affine_inverse();

-

-	Vector3 ray_from = p_camera->project_ray_origin(p_point);

-	Vector3 ray_dir = p_camera->project_ray_normal(p_point);

-

-	Vector3 s[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};

-	if (p_idx==0) {

-

-		if (light->cast_to<SpotLight>()) {

-			Vector3 ra,rb;

-			Geometry::get_closest_points_between_segments(Vector3(),Vector3(0,0,-4096),s[0],s[1],ra,rb);

-

-			float d = -ra.z;

-			if (d<0)

-				d=0;

-

-			light->set_parameter(Light::PARAM_RADIUS,d);

-		} else if (light->cast_to<OmniLight>()) {

-

-			Plane cp=Plane( gt.origin, p_camera->get_transform().basis.get_axis(2));

-

-			Vector3 inters;

-			if (cp.intersects_ray(ray_from,ray_dir,&inters)) {

-

-				float r = inters.distance_to(gt.origin);

-				light->set_parameter(Light::PARAM_RADIUS,r);

-			}

-

-		}

-

-	} else if (p_idx==1) {

-

-		float a = _find_closest_angle_to_half_pi_arc(s[0],s[1],light->get_parameter(Light::PARAM_RADIUS),gt);

-		light->set_parameter(Light::PARAM_SPOT_ANGLE,CLAMP(a,0.01,89.99));

-	}

-}

-

-void LightSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){

-

-	if (p_cancel) {

-

-		light->set_parameter(p_idx==0?Light::PARAM_RADIUS:Light::PARAM_SPOT_ANGLE,p_restore);

-

-	} else if (p_idx==0) {

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		ur->create_action("Change Light Radius");

-		ur->add_do_method(light,"set_parameter",Light::PARAM_RADIUS,light->get_parameter(Light::PARAM_RADIUS));

-		ur->add_undo_method(light,"set_parameter",Light::PARAM_RADIUS,p_restore);

-		ur->commit_action();

-	} else if (p_idx==1) {

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		ur->create_action("Change Light Radius");

-		ur->add_do_method(light,"set_parameter",Light::PARAM_SPOT_ANGLE,light->get_parameter(Light::PARAM_SPOT_ANGLE));

-		ur->add_undo_method(light,"set_parameter",Light::PARAM_SPOT_ANGLE,p_restore);

-		ur->commit_action();

-

-	}

-}

-

-

-

-void LightSpatialGizmo::redraw() {

-

-

-	if (light->cast_to<DirectionalLight>()) {

-

-

-

-		const int arrow_points=5;

-		Vector3 arrow[arrow_points]={

-			Vector3(0,0,2),

-			Vector3(1,1,2),

-			Vector3(1,1,-1),

-			Vector3(2,2,-1),

-			Vector3(0,0,-3)

-		};

-

-		int arrow_sides=4;

-

-		Vector<Vector3> lines;

-

-

-		for(int i = 0; i < arrow_sides ; i++) {

-

-

-			Matrix3 ma(Vector3(0,0,1),Math_PI*2*float(i)/arrow_sides);

-			Matrix3 mb(Vector3(0,0,1),Math_PI*2*float(i+1)/arrow_sides);

-

-

-			for(int j=1;j<arrow_points-1;j++) {

-

-				if (j!=2) {

-					lines.push_back(ma.xform(arrow[j]));

-					lines.push_back(ma.xform(arrow[j+1]));

-				}

-				if (j<arrow_points-1) {

-					lines.push_back(ma.xform(arrow[j]));

-					lines.push_back(mb.xform(arrow[j]));

-				}

-

-			}

-		}

-

-		add_lines(lines,SpatialEditorGizmos::singleton->light_material);

-		add_collision_segments(lines);

-		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_directional_icon,0.05);

-

-	}

-

-	if (light->cast_to<OmniLight>()) {

-

-		clear();

-

-

-		OmniLight *on = light->cast_to<OmniLight>();

-

-		float r = on->get_parameter(Light::PARAM_RADIUS);

-

-		Vector<Vector3> points;

-

-		for(int i=0;i<=360;i++) {

-

-			float ra=Math::deg2rad(i);

-			float rb=Math::deg2rad(i+1);

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;

-

-			/*points.push_back(Vector3(a.x,0,a.y));

-			points.push_back(Vector3(b.x,0,b.y));

-			points.push_back(Vector3(0,a.x,a.y));

-			points.push_back(Vector3(0,b.x,b.y));*/

-			points.push_back(Vector3(a.x,a.y,0));

-			points.push_back(Vector3(b.x,b.y,0));

-

-		}

-

-		add_lines(points,SpatialEditorGizmos::singleton->light_material,true);

-		add_collision_segments(points);

-

-		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_omni_icon,0.05);

-

-		Vector<Vector3> handles;

-		handles.push_back(Vector3(r,0,0));

-		add_handles(handles,true);

-

-

-	}

-

-

-	if (light->cast_to<SpotLight>()) {

-

-		clear();

-

-		Vector<Vector3> points;

-		SpotLight *on = light->cast_to<SpotLight>();

-

-		float r = on->get_parameter(Light::PARAM_RADIUS);

-		float w = r*Math::sin(Math::deg2rad(on->get_parameter(Light::PARAM_SPOT_ANGLE)));

-		float d = r*Math::cos(Math::deg2rad(on->get_parameter(Light::PARAM_SPOT_ANGLE)));

-

-

-

-		for(int i=0;i<360;i++) {

-

-			float ra=Math::deg2rad(i);

-			float rb=Math::deg2rad(i+1);

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;

-

-			/*points.push_back(Vector3(a.x,0,a.y));

-			points.push_back(Vector3(b.x,0,b.y));

-			points.push_back(Vector3(0,a.x,a.y));

-			points.push_back(Vector3(0,b.x,b.y));*/

-			points.push_back(Vector3(a.x,a.y,-d));

-			points.push_back(Vector3(b.x,b.y,-d));

-

-			if (i%90==0) {

-

-				points.push_back(Vector3(a.x,a.y,-d));

-				points.push_back(Vector3());

-

-			}

-

-

-		}

-

-		points.push_back(Vector3(0,0,-r));

-		points.push_back(Vector3());

-

-		add_lines(points,SpatialEditorGizmos::singleton->light_material);

-

-		Vector<Vector3> handles;

-		handles.push_back(Vector3(0,0,-r));

-

-		Vector<Vector3> collision_segments;

-

-		for(int i=0;i<64;i++) {

-

-			float ra=i*Math_PI*2.0/64.0;

-			float rb=(i+1)*Math_PI*2.0/64.0;

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;

-

-			collision_segments.push_back(Vector3(a.x,a.y,-d));

-			collision_segments.push_back(Vector3(b.x,b.y,-d));

-

-			if (i%16==0) {

-

-				collision_segments.push_back(Vector3(a.x,a.y,-d));

-				collision_segments.push_back(Vector3());

-

-			}

-

-			if (i==16) {

-

-				handles.push_back(Vector3(a.x,a.y,-d));

-			}

-

-		}

-

-		collision_segments.push_back(Vector3(0,0,-r));

-		collision_segments.push_back(Vector3());

-

-

-		add_handles(handles);

-		add_collision_segments(collision_segments);

-		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_omni_icon,0.05);

-

-	}

-

-}

-

-LightSpatialGizmo::LightSpatialGizmo(Light* p_light){

-

-	light=p_light;

-	set_spatial_node(p_light);

-

-}

-

-//////

-

-String CameraSpatialGizmo::get_handle_name(int p_idx) const {

-

-	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {

-		return "FOV";

-	} else {

-		return "Size";

-	}

-}

-Variant CameraSpatialGizmo::get_handle_value(int p_idx) const{

-

-	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {

-		return camera->get_fov();

-	} else {

-

-		return camera->get_size();

-	}

-}

-void CameraSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){

-

-	Transform gt = camera->get_global_transform();

-	gt.orthonormalize();

-	Transform gi = gt.affine_inverse();

-

-	Vector3 ray_from = p_camera->project_ray_origin(p_point);

-	Vector3 ray_dir = p_camera->project_ray_normal(p_point);

-

-	Vector3 s[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};

-

-	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {

-		Transform gt=camera->get_global_transform();

-		float a = _find_closest_angle_to_half_pi_arc(s[0],s[1],1.0,gt);

-		camera->set("fov",a);

-	} else {

-

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(Vector3(0,0,-1),Vector3(4096,0,-1),s[0],s[1],ra,rb);

-		float d = ra.x * 2.0;

-		if (d<0)

-			d=0;

-

-		camera->set("size",d);

-	}

-

-}

-void CameraSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){

-

-	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {

-

-		if (p_cancel) {

-

-			camera->set("fov",p_restore);

-		} else {

-			UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-			ur->create_action("Change Camera FOV");

-			ur->add_do_property(camera,"fov",camera->get_fov());

-			ur->add_undo_property(camera,"fov",p_restore);

-			ur->commit_action();

-		}

-

-	} else {

-

-		if (p_cancel) {

-

-			camera->set("size",p_restore);

-		} else {

-			UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-			ur->create_action("Change Camera Size");

-			ur->add_do_property(camera,"size",camera->get_size());

-			ur->add_undo_property(camera,"size",p_restore);

-			ur->commit_action();

-		}

-

-	}

-

-}

-

-void CameraSpatialGizmo::redraw(){

-

-	clear();

-

-	Vector<Vector3> lines;

-	Vector<Vector3> handles;

-

-

-	switch(camera->get_projection()) {

-

-		case Camera::PROJECTION_PERSPECTIVE: {

-

-			float fov = camera->get_fov();

-

-			Vector3 side=Vector3( Math::sin(Math::deg2rad(fov)), 0, -Math::cos(Math::deg2rad(fov)) );

-			Vector3 nside=side;

-			nside.x=-nside.x;

-			Vector3 up=Vector3(0,side.x,0);

-

-

-#define ADD_TRIANGLE( m_a, m_b, m_c)\

-{\

-	lines.push_back(m_a);\

-	lines.push_back(m_b);\

-	lines.push_back(m_b);\

-	lines.push_back(m_c);\

-	lines.push_back(m_c);\

-	lines.push_back(m_a);\

-}

-

-			ADD_TRIANGLE( Vector3(), side+up, side-up );

-			ADD_TRIANGLE( Vector3(), nside+up, nside-up );

-			ADD_TRIANGLE( Vector3(), side+up, nside+up );

-			ADD_TRIANGLE( Vector3(), side-up, nside-up );

-

-			handles.push_back(side);

-			side.x*=0.25;

-			nside.x*=0.25;

-			Vector3 tup( 0, up.y*3/2,side.z);

-			ADD_TRIANGLE( tup, side+up, nside+up );

-

-		} break;

-		case Camera::PROJECTION_ORTHOGONAL: {

-

-#define ADD_QUAD( m_a, m_b, m_c, m_d)\

-{\

-	lines.push_back(m_a);\

-	lines.push_back(m_b);\

-	lines.push_back(m_b);\

-	lines.push_back(m_c);\

-	lines.push_back(m_c);\

-	lines.push_back(m_d);\

-	lines.push_back(m_d);\

-	lines.push_back(m_a);\

-}

-			float size = camera->get_size();

-

-			float hsize=size*0.5;

-			Vector3 right(hsize,0,0);

-			Vector3 up(0,hsize,0);

-			Vector3 back(0,0,-1.0);

-			Vector3 front(0,0,0);

-

-			ADD_QUAD( -up-right,-up+right,up+right,up-right);

-			ADD_QUAD( -up-right+back,-up+right+back,up+right+back,up-right+back);

-			ADD_QUAD( up+right,up+right+back,up-right+back,up-right);

-			ADD_QUAD( -up+right,-up+right+back,-up-right+back,-up-right);

-			handles.push_back(right+back);

-

-			right.x*=0.25;

-			Vector3 tup( 0, up.y*3/2,back.z );

-			ADD_TRIANGLE( tup, right+up+back, -right+up+back );

-

-		} break;

-

-	}

-

-	add_lines(lines,SpatialEditorGizmos::singleton->camera_material);

-	add_collision_segments(lines);

-	add_handles(handles);

-}

-

-

-CameraSpatialGizmo::CameraSpatialGizmo(Camera* p_camera){

-

-	camera=p_camera;

-	set_spatial_node(camera);

-}

-

-

-

-

-//////

-

-void MeshInstanceSpatialGizmo::redraw() {

-

-	Ref<Mesh> m = mesh->get_mesh();

-	if (!m.is_valid())

-		return; //none

-

-	Ref<TriangleMesh> tm = m->generate_triangle_mesh();

-	if (tm.is_valid())

-		add_collision_triangles(tm);

-}

-

-MeshInstanceSpatialGizmo::MeshInstanceSpatialGizmo(MeshInstance* p_mesh) {

-

-	mesh=p_mesh;

-	set_spatial_node(p_mesh);

-}

-

-/////

-

-

-void Position3DSpatialGizmo::redraw() {

-

-	clear();

-	add_mesh(SpatialEditorGizmos::singleton->pos3d_mesh);

-	Vector<Vector3> cursor_points;

-	float cs = 0.25;

-	cursor_points.push_back(Vector3(+cs,0,0));

-	cursor_points.push_back(Vector3(-cs,0,0));

-	cursor_points.push_back(Vector3(0,+cs,0));

-	cursor_points.push_back(Vector3(0,-cs,0));

-	cursor_points.push_back(Vector3(0,0,+cs));

-	cursor_points.push_back(Vector3(0,0,-cs));

-	add_collision_segments(cursor_points);

-

-}

-

-

-Position3DSpatialGizmo::Position3DSpatialGizmo(Position3D* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-

-/////

-

-void SkeletonSpatialGizmo::redraw() {

-

-	clear();

-

-	Ref<SurfaceTool> surface_tool( memnew( SurfaceTool ));

-

-

-	surface_tool->begin(Mesh::PRIMITIVE_LINES);

-	surface_tool->set_material(SpatialEditorGizmos::singleton->skeleton_material);

-	Vector<Transform> grests;

-	grests.resize(skel->get_bone_count());

-

-	Vector<int> bones;

-	Vector<float> weights;

-	bones.resize(4);

-	weights.resize(4);

-

-	for(int i=0;i<4;i++) {

-		bones[i]=0;

-		weights[i]=0;

-	}

-

-	weights[0]=1;

-

-

-	AABB aabb;

-

-	Color bonecolor = Color(1.0,0.4,0.4,0.3);

-	Color rootcolor = Color(0.4,1.0,0.4,0.1);

-

-	for (int i=0;i<skel->get_bone_count();i++) {

-

-		int parent = skel->get_bone_parent(i);

-

-		if (parent>=0) {

-			grests[i]=grests[parent] * skel->get_bone_rest(i);

-

-			Vector3 v0 = grests[parent].origin;

-			Vector3 v1 = grests[i].origin;

-			Vector3 d = (v1-v0).normalized();

-			float dist = v0.distance_to(v1);

-

-			//find closest axis

-			int closest=-1;

-			float closest_d = 0.0;

-

-			for(int j=0;j<3;j++) {

-				float dp = Math::abs(grests[parent].basis[j].normalized().dot(d));

-				if (j==0 || dp>closest_d)

-					closest=j;

-			}

-

-			//find closest other

-			Vector3 first;

-			Vector3 points[4];

-			int pointidx=0;

-			for(int j=0;j<3;j++) {

-

-				bones[0]=parent;

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(rootcolor);

-				surface_tool->add_vertex(v0-grests[parent].basis[j].normalized()*dist*0.05);

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(rootcolor);

-				surface_tool->add_vertex(v0+grests[parent].basis[j].normalized()*dist*0.05);

-

-				if (j==closest)

-					continue;

-

-				Vector3 axis;

-				if (first==Vector3()) {

-					axis = d.cross(d.cross(grests[parent].basis[j])).normalized();

-					first=axis;

-				} else {

-					axis = d.cross(first).normalized();

-				}

-

-				for(int k=0;k<2;k++) {

-

-					if (k==1)

-						axis=-axis;

-					Vector3 point = v0+d*dist*0.2;

-					point+=axis*dist*0.1;

-

-

-					bones[0]=parent;

-					surface_tool->add_bones(bones);

-					surface_tool->add_weights(weights);

-					surface_tool->add_color(bonecolor);

-					surface_tool->add_vertex(v0);

-					surface_tool->add_bones(bones);

-					surface_tool->add_weights(weights);

-					surface_tool->add_color(bonecolor);

-					surface_tool->add_vertex(point);

-

-					bones[0]=parent;

-					surface_tool->add_bones(bones);

-					surface_tool->add_weights(weights);

-					surface_tool->add_color(bonecolor);

-					surface_tool->add_vertex(point);

-					bones[0]=i;

-					surface_tool->add_bones(bones);

-					surface_tool->add_weights(weights);

-					surface_tool->add_color(bonecolor);

-					surface_tool->add_vertex(v1);

-					points[pointidx++]=point;

-

-				}

-

-			}

-

-			SWAP( points[1],points[2] );

-			for(int j=0;j<4;j++) {

-

-

-				bones[0]=parent;

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(bonecolor);

-				surface_tool->add_vertex(points[j]);

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(bonecolor);

-				surface_tool->add_vertex(points[(j+1)%4]);

-			}

-

-

-/*

-			bones[0]=parent;

-			surface_tool->add_bones(bones);

-			surface_tool->add_weights(weights);

-			surface_tool->add_color(Color(0.4,1,0.4,0.4));

-			surface_tool->add_vertex(v0);

-			bones[0]=i;

-			surface_tool->add_bones(bones);

-			surface_tool->add_weights(weights);

-			surface_tool->add_color(Color(0.4,1,0.4,0.4));

-			surface_tool->add_vertex(v1);

-*/

-		} else {

-

-			grests[i]=skel->get_bone_rest(i);

-			bones[0]=i;

-		}

-/*

-		Transform  t = grests[i];

-		t.orthonormalize();

-

-		for (int i=0;i<6;i++) {

-

-

-			Vector3 face_points[4];

-

-			for (int j=0;j<4;j++) {

-

-				float v[3];

-				v[0]=1.0;

-				v[1]=1-2*((j>>1)&1);

-				v[2]=v[1]*(1-2*(j&1));

-

-				for (int k=0;k<3;k++) {

-

-					if (i<3)

-						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);

-					else

-						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);

-				}

-			}

-

-			for(int j=0;j<4;j++) {

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(Color(1.0,0.4,0.4,0.4));

-				surface_tool->add_vertex(t.xform(face_points[j]*0.04));

-				surface_tool->add_bones(bones);

-				surface_tool->add_weights(weights);

-				surface_tool->add_color(Color(1.0,0.4,0.4,0.4));

-				surface_tool->add_vertex(t.xform(face_points[(j+1)%4]*0.04));

-			}

-

-		}

-		*/

-	}

-

-	Ref<Mesh> m = surface_tool->commit();

-	add_mesh(m,false,skel->get_skeleton());

-

-}

-

-SkeletonSpatialGizmo::SkeletonSpatialGizmo(Skeleton* p_skel) {

-

-	skel=p_skel;

-	set_spatial_node(p_skel);

-}

-

-/////

-

-

-void SpatialPlayerSpatialGizmo::redraw() {

-

-	clear();

-	if (splayer->cast_to<SpatialStreamPlayer>()) {

-

-		add_unscaled_billboard(SpatialEditorGizmos::singleton->stream_player_icon,0.05);

-

-	} else if (splayer->cast_to<SpatialSamplePlayer>()) {

-

-		add_unscaled_billboard(SpatialEditorGizmos::singleton->sample_player_icon,0.05);

-

-	}

-

-}

-

-SpatialPlayerSpatialGizmo::SpatialPlayerSpatialGizmo(SpatialPlayer* p_splayer){

-

-	set_spatial_node(p_splayer);

-	splayer=p_splayer;

-}

-

-

-/////

-

-

-void RoomSpatialGizmo::redraw() {

-

-	clear();

-	Ref<RoomBounds> roomie = room->get_room();

-	if (roomie.is_null())

-		return;

-	DVector<Face3> faces = roomie->get_geometry_hint();

-

-	Vector<Vector3> lines;

-	int fc=faces.size();

-	DVector<Face3>::Read r =faces.read();

-

-	Map<_EdgeKey,Vector3> edge_map;

-

-	for(int i=0;i<fc;i++) {

-

-		Vector3 fn = r[i].get_plane().normal;

-

-		for(int j=0;j<3;j++) {

-

-			_EdgeKey ek;

-			ek.from=r[i].vertex[j].snapped(CMP_EPSILON);

-			ek.to=r[i].vertex[(j+1)%3].snapped(CMP_EPSILON);

-			if (ek.from<ek.to)

-				SWAP(ek.from,ek.to);

-

-			Map<_EdgeKey,Vector3>::Element *E=edge_map.find(ek);

-

-			if (E) {

-

-				if (E->get().dot(fn) >0.9) {

-

-					E->get()=Vector3();

-				}

-

-			} else {

-

-				edge_map[ek]=fn;

-			}

-

-		}

-	}

-

-	for(Map<_EdgeKey,Vector3>::Element *E=edge_map.front();E;E=E->next()) {

-

-		if (E->get()!=Vector3()) {

-			lines.push_back(E->key().from);

-			lines.push_back(E->key().to);

-		}

-	}

-

-	add_lines(lines,SpatialEditorGizmos::singleton->room_material);

-	add_collision_segments(lines);

-

-}

-

-RoomSpatialGizmo::RoomSpatialGizmo(Room* p_room){

-

-	set_spatial_node(p_room);

-	room=p_room;

-}

-

-/////

-

-

-void PortalSpatialGizmo::redraw() {

-

-	clear();

-

-	Vector<Point2> points = portal->get_shape();

-	if (points.size()==0) {

-		return;

-	}

-

-	Vector<Vector3> lines;

-

-	Vector3 center;

-	for(int i=0;i<points.size();i++) {

-

-		Vector3 f;

-		f.x=points[i].x;

-		f.y=points[i].y;

-		Vector3 fn;

-		fn.x=points[(i+1)%points.size()].x;

-		fn.y=points[(i+1)%points.size()].y;

-		center+=f;

-

-		lines.push_back(f);

-		lines.push_back(fn);

-	}

-

-	center/=points.size();

-	lines.push_back(center);

-	lines.push_back(center+Vector3(0,0,1));

-

-	add_lines(lines,SpatialEditorGizmos::singleton->portal_material);

-	add_collision_segments(lines);

-

-}

-

-PortalSpatialGizmo::PortalSpatialGizmo(Portal* p_portal){

-

-	set_spatial_node(p_portal);

-	portal=p_portal;

-}

-

-/////

-

-

-void RayCastSpatialGizmo::redraw() {

-

-	clear();

-

-

-	Vector<Vector3> lines;

-

-	lines.push_back(Vector3());

-	lines.push_back(raycast->get_cast_to());

-

-	add_lines(lines,SpatialEditorGizmos::singleton->raycast_material);

-	add_collision_segments(lines);

-

-}

-

-RayCastSpatialGizmo::RayCastSpatialGizmo(RayCast* p_raycast){

-

-	set_spatial_node(p_raycast);

-	raycast=p_raycast;

-}

-

-

-

-/////

-

-

-void VehicleWheelSpatialGizmo::redraw() {

-

-	clear();

-

-

-	Vector<Vector3> points;

-

-	float r = car_wheel->get_radius();

-	const int skip=10;

-	for(int i=0;i<=360;i+=skip) {

-

-		float ra=Math::deg2rad(i);

-		float rb=Math::deg2rad(i+skip);

-		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;

-		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;

-

-		points.push_back(Vector3(0,a.x,a.y));

-		points.push_back(Vector3(0,b.x,b.y));

-

-		const int springsec=4;

-

-		for(int j=0;j<springsec;j++) {

-			float t = car_wheel->get_suspension_rest_length()*5;

-			points.push_back(Vector3(a.x,i/360.0*t/springsec+j*(t/springsec),a.y)*0.2);

-			points.push_back(Vector3(b.x,(i+skip)/360.0*t/springsec+j*(t/springsec),b.y)*0.2);

-		}

-

-

-	}

-

-	//travel

-	points.push_back(Vector3(0,0,0));

-	points.push_back(Vector3(0,car_wheel->get_suspension_rest_length(),0));

-

-	//axis

-	points.push_back(Vector3(r*0.2,car_wheel->get_suspension_rest_length(),0));

-	points.push_back(Vector3(-r*0.2,car_wheel->get_suspension_rest_length(),0));

-	//axis

-	points.push_back(Vector3(r*0.2,0,0));

-	points.push_back(Vector3(-r*0.2,0,0));

-

-	//forward line

-	points.push_back(Vector3(0,-r,0));

-	points.push_back(Vector3(0,-r,r*2));

-	points.push_back(Vector3(0,-r,r*2));

-	points.push_back(Vector3(r*2*0.2,-r,r*2*0.8));

-	points.push_back(Vector3(0,-r,r*2));

-	points.push_back(Vector3(-r*2*0.2,-r,r*2*0.8));

-

-	add_lines(points,SpatialEditorGizmos::singleton->car_wheel_material);

-	add_collision_segments(points);

-

-}

-

-VehicleWheelSpatialGizmo::VehicleWheelSpatialGizmo(VehicleWheel* p_car_wheel){

-

-	set_spatial_node(p_car_wheel);

-	car_wheel=p_car_wheel;

-}

-

-

-

-///

-

-void TestCubeSpatialGizmo::redraw() {

-

-	clear();

-	add_collision_triangles(SpatialEditorGizmos::singleton->test_cube_tm);

-}

-

-TestCubeSpatialGizmo::TestCubeSpatialGizmo(TestCube* p_tc) {

-

-	tc=p_tc;

-	set_spatial_node(p_tc);

-}

-

-

-///////////

-

-

-

-

-

-

-String CollisionShapeSpatialGizmo::get_handle_name(int p_idx) const {

-

-	Ref<Shape> s = cs->get_shape();

-	if (s.is_null())

-		return "";

-

-	if (s->cast_to<SphereShape>()) {

-

-		return "Radius";

-	}

-

-	if (s->cast_to<BoxShape>()) {

-

-		return "Extents";

-	}

-

-	if (s->cast_to<CapsuleShape>()) {

-

-		return p_idx==0?"Radius":"Height";

-	}

-

-	if (s->cast_to<RayShape>()) {

-

-		return "Length";

-	}

-

-	return "";

-}

-Variant CollisionShapeSpatialGizmo::get_handle_value(int p_idx) const{

-

-	Ref<Shape> s = cs->get_shape();

-	if (s.is_null())

-		return Variant();

-

-	if (s->cast_to<SphereShape>()) {

-

-		Ref<SphereShape> ss = s;

-		return ss->get_radius();

-	}

-

-	if (s->cast_to<BoxShape>()) {

-

-		Ref<BoxShape> bs = s;

-		return bs->get_extents();

-	}

-

-	if (s->cast_to<CapsuleShape>()) {

-

-		Ref<CapsuleShape> cs = s;

-		return p_idx==0?cs->get_radius():cs->get_height();

-	}

-

-	if (s->cast_to<RayShape>()) {

-

-		Ref<RayShape> cs = s;

-		return cs->get_length();

-	}

-

-	return Variant();

-}

-void CollisionShapeSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){

-	Ref<Shape> s = cs->get_shape();

-	if (s.is_null())

-		return;

-

-	Transform gt = cs->get_global_transform();

-	gt.orthonormalize();

-	Transform gi = gt.affine_inverse();

-

-	Vector3 ray_from = p_camera->project_ray_origin(p_point);

-	Vector3 ray_dir = p_camera->project_ray_normal(p_point);

-

-	Vector3 sg[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};

-

-	if (s->cast_to<SphereShape>()) {

-

-		Ref<SphereShape> ss = s;

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(Vector3(),Vector3(4096,0,0),sg[0],sg[1],ra,rb);

-		float d = ra.x;

-		if (d<0.001)

-			d=0.001;

-

-		ss->set_radius(d);

-	}

-

-	if (s->cast_to<RayShape>()) {

-

-		Ref<RayShape> rs = s;

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(Vector3(),Vector3(0,0,4096),sg[0],sg[1],ra,rb);

-		float d = ra.z;

-		if (d<0.001)

-			d=0.001;

-

-		rs->set_length(d);

-	}

-

-

-	if (s->cast_to<BoxShape>()) {

-

-		Vector3 axis;

-		axis[p_idx]=1.0;

-		Ref<BoxShape> bs = s;

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(Vector3(),axis*4096,sg[0],sg[1],ra,rb);

-		float d = ra[p_idx];

-		if (d<0.001)

-			d=0.001;

-

-		Vector3 he = bs->get_extents();

-		he[p_idx]=d;

-		bs->set_extents(he);

-

-	}

-

-	if (s->cast_to<CapsuleShape>()) {

-

-		Vector3 axis;

-		axis[p_idx==0?0:2]=1.0;

-		Ref<CapsuleShape> cs = s;

-		Vector3 ra,rb;

-		Geometry::get_closest_points_between_segments(Vector3(),axis*4096,sg[0],sg[1],ra,rb);

-		float d = axis.dot(ra);

-		if (p_idx==1)

-			d-=cs->get_radius();

-		if (d<0.001)

-			d=0.001;

-

-		if (p_idx==0)

-			cs->set_radius(d);

-		else if (p_idx==1)

-			cs->set_height(d*2.0);

-

-	}

-

-}

-void CollisionShapeSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){

-	Ref<Shape> s = cs->get_shape();

-	if (s.is_null())

-		return;

-

-	if (s->cast_to<SphereShape>()) {

-

-		Ref<SphereShape> ss=s;

-		if (p_cancel) {

-			ss->set_radius(p_restore);

-			return;

-		}

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		ur->create_action("Change Sphere Shape Radius");

-		ur->add_do_method(ss.ptr(),"set_radius",ss->get_radius());

-		ur->add_undo_method(ss.ptr(),"set_radius",p_restore);

-		ur->commit_action();

-

-	}

-

-	if (s->cast_to<BoxShape>()) {

-

-		Ref<BoxShape> ss=s;

-		if (p_cancel) {

-			ss->set_extents(p_restore);

-			return;

-		}

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		ur->create_action("Change Box Shape Extents");

-		ur->add_do_method(ss.ptr(),"set_extents",ss->get_extents());

-		ur->add_undo_method(ss.ptr(),"set_extents",p_restore);

-		ur->commit_action();

-	}

-

-	if (s->cast_to<CapsuleShape>()) {

-

-		Ref<CapsuleShape> ss=s;

-		if (p_cancel) {

-			if (p_idx==0)

-				ss->set_radius(p_restore);

-			else

-				ss->set_height(p_restore);

-			return;

-		}

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		if (p_idx==0) {

-			ur->create_action("Change Capsule Shape Radius");

-			ur->add_do_method(ss.ptr(),"set_radius",ss->get_radius());

-			ur->add_undo_method(ss.ptr(),"set_radius",p_restore);

-		} else {

-			ur->create_action("Change Capsule Shape Height");

-			ur->add_do_method(ss.ptr(),"set_height",ss->get_height());

-			ur->add_undo_method(ss.ptr(),"set_height",p_restore);

-

-		}

-

-		ur->commit_action();

-

-	}

-

-	if (s->cast_to<RayShape>()) {

-

-		Ref<RayShape> ss=s;

-		if (p_cancel) {

-			ss->set_length(p_restore);

-			return;

-		}

-

-		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-		ur->create_action("Change Ray Shape Length");

-		ur->add_do_method(ss.ptr(),"set_length",ss->get_length());

-		ur->add_undo_method(ss.ptr(),"set_length",p_restore);

-		ur->commit_action();

-

-	}

-

-}

-void CollisionShapeSpatialGizmo::redraw(){

-

-	clear();

-

-	Ref<Shape> s = cs->get_shape();

-	if (s.is_null())

-		return;

-

-	if (s->cast_to<SphereShape>()) {

-

-		Ref<SphereShape> sp= s;

-		float r=sp->get_radius();

-

-		Vector<Vector3> points;

-

-		for(int i=0;i<=360;i++) {

-

-			float ra=Math::deg2rad(i);

-			float rb=Math::deg2rad(i+1);

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;

-

-			points.push_back(Vector3(a.x,0,a.y));

-			points.push_back(Vector3(b.x,0,b.y));

-			points.push_back(Vector3(0,a.x,a.y));

-			points.push_back(Vector3(0,b.x,b.y));

-			points.push_back(Vector3(a.x,a.y,0));

-			points.push_back(Vector3(b.x,b.y,0));

-

-		}

-

-		Vector<Vector3> collision_segments;

-

-		for(int i=0;i<64;i++) {

-

-			float ra=i*Math_PI*2.0/64.0;

-			float rb=(i+1)*Math_PI*2.0/64.0;

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;

-

-			collision_segments.push_back(Vector3(a.x,0,a.y));

-			collision_segments.push_back(Vector3(b.x,0,b.y));

-			collision_segments.push_back(Vector3(0,a.x,a.y));

-			collision_segments.push_back(Vector3(0,b.x,b.y));

-			collision_segments.push_back(Vector3(a.x,a.y,0));

-			collision_segments.push_back(Vector3(b.x,b.y,0));

-		}

-

-		add_lines(points,SpatialEditorGizmos::singleton->shape_material);

-		add_collision_segments(collision_segments);

-		Vector<Vector3> handles;

-		handles.push_back(Vector3(r,0,0));

-		add_handles(handles);

-

-	}

-

-	if (s->cast_to<BoxShape>()) {

-

-		Ref<BoxShape> bs=s;

-		Vector<Vector3> lines;

-		AABB aabb;

-		aabb.pos=-bs->get_extents();

-		aabb.size=aabb.pos*-2;

-

-		for(int i=0;i<12;i++) {

-			Vector3 a,b;

-			aabb.get_edge(i,a,b);

-			lines.push_back(a);

-			lines.push_back(b);

-		}

-

-		Vector<Vector3> handles;

-

-		for(int i=0;i<3;i++) {

-

-			Vector3 ax;

-			ax[i]=bs->get_extents()[i];

-			handles.push_back(ax);

-		}

-

-		add_lines(lines,SpatialEditorGizmos::singleton->shape_material);

-		add_collision_segments(lines);

-		add_handles(handles);

-

-	}

-

-	if (s->cast_to<CapsuleShape>()) {

-

-		Ref<CapsuleShape> cs=s;

-		float radius = cs->get_radius();

-		float height = cs->get_height();

-

-

-		Vector<Vector3> points;

-

-		Vector3 d(0,0,height*0.5);

-		for(int i=0;i<360;i++) {

-

-			float ra=Math::deg2rad(i);

-			float rb=Math::deg2rad(i+1);

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*radius;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*radius;

-

-			points.push_back(Vector3(a.x,a.y,0)+d);

-			points.push_back(Vector3(b.x,b.y,0)+d);

-

-			points.push_back(Vector3(a.x,a.y,0)-d);

-			points.push_back(Vector3(b.x,b.y,0)-d);

-

-			if (i%90==0) {

-

-				points.push_back(Vector3(a.x,a.y,0)+d);

-				points.push_back(Vector3(a.x,a.y,0)-d);

-			}

-

-			Vector3 dud = i<180?d:-d;

-

-			points.push_back(Vector3(0,a.y,a.x)+dud);

-			points.push_back(Vector3(0,b.y,b.x)+dud);

-			points.push_back(Vector3(a.y,0,a.x)+dud);

-			points.push_back(Vector3(b.y,0,b.x)+dud);

-

-		}

-

-		add_lines(points,SpatialEditorGizmos::singleton->shape_material);

-

-		Vector<Vector3> collision_segments;

-

-		for(int i=0;i<64;i++) {

-

-			float ra=i*Math_PI*2.0/64.0;

-			float rb=(i+1)*Math_PI*2.0/64.0;

-			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*radius;

-			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*radius;

-

-			collision_segments.push_back(Vector3(a.x,a.y,0)+d);

-			collision_segments.push_back(Vector3(b.x,b.y,0)+d);

-

-			collision_segments.push_back(Vector3(a.x,a.y,0)-d);

-			collision_segments.push_back(Vector3(b.x,b.y,0)-d);

-

-			if (i%16==0) {

-

-				collision_segments.push_back(Vector3(a.x,a.y,0)+d);

-				collision_segments.push_back(Vector3(a.x,a.y,0)-d);

-			}

-

-			Vector3 dud = i<32?d:-d;

-

-			collision_segments.push_back(Vector3(0,a.y,a.x)+dud);

-			collision_segments.push_back(Vector3(0,b.y,b.x)+dud);

-			collision_segments.push_back(Vector3(a.y,0,a.x)+dud);

-			collision_segments.push_back(Vector3(b.y,0,b.x)+dud);

-

-		}

-

-		add_collision_segments(collision_segments);

-

-		Vector<Vector3> handles;

-		handles.push_back(Vector3(cs->get_radius(),0,0));

-		handles.push_back(Vector3(0,0,cs->get_height()*0.5+cs->get_radius()));

-		add_handles(handles);

-

-

-	}

-

-	if (s->cast_to<PlaneShape>()) {

-

-		Ref<PlaneShape> ps=s;

-		Plane p = ps->get_plane();

-		Vector<Vector3> points;

-

-		Vector3 n1 = p.get_any_perpendicular_normal();

-		Vector3 n2 = p.normal.cross(n1).normalized();

-

-		Vector3 pface[4]={

-			p.normal*p.d+n1*10.0+n2*10.0,

-			p.normal*p.d+n1*10.0+n2*-10.0,

-			p.normal*p.d+n1*-10.0+n2*-10.0,

-			p.normal*p.d+n1*-10.0+n2*10.0,

-		};

-

-		points.push_back(pface[0]);

-		points.push_back(pface[1]);

-		points.push_back(pface[1]);

-		points.push_back(pface[2]);

-		points.push_back(pface[2]);

-		points.push_back(pface[3]);

-		points.push_back(pface[3]);

-		points.push_back(pface[0]);

-		points.push_back(p.normal*p.d);

-		points.push_back(p.normal*p.d+p.normal*3);

-

-		add_lines(points,SpatialEditorGizmos::singleton->shape_material);

-		add_collision_segments(points);

-

-	}

-

-

-	if (s->cast_to<ConvexPolygonShape>()) {

-

-		DVector<Vector3> points = s->cast_to<ConvexPolygonShape>()->get_points();

-

-		if (points.size()>3) {

-

-			QuickHull qh;

-			Vector<Vector3> varr = Variant(points);

-			Geometry::MeshData md;

-			Error err = qh.build(varr,md);

-			if (err==OK) {

-				Vector<Vector3> points;

-				points.resize(md.edges.size()*2);

-				for(int i=0;i<md.edges.size();i++) {

-					points[i*2+0]=md.vertices[md.edges[i].a];

-					points[i*2+1]=md.vertices[md.edges[i].b];

-				}

-

-

-				add_lines(points,SpatialEditorGizmos::singleton->shape_material);

-				add_collision_segments(points);

-

-			}

-		}

-

-	}

-

-

-	if (s->cast_to<RayShape>()) {

-

-		Ref<RayShape> rs=s;

-

-		Vector<Vector3> points;

-		points.push_back(Vector3());

-		points.push_back(Vector3(0,0,rs->get_length()));

-		add_lines(points,SpatialEditorGizmos::singleton->shape_material);

-		add_collision_segments(points);

-		Vector<Vector3> handles;

-		handles.push_back(Vector3(0,0,rs->get_length()));

-		add_handles(handles);

-

-

-	}

-

-}

-CollisionShapeSpatialGizmo::CollisionShapeSpatialGizmo(CollisionShape* p_cs) {

-

-	cs=p_cs;

-	set_spatial_node(p_cs);

-}

-

-

-

-/////

-

-

-void CollisionPolygonSpatialGizmo::redraw() {

-

-	clear();

-

-	Vector<Vector2> points = polygon->get_polygon();

-	float depth = polygon->get_depth()*0.5;

-

-	Vector<Vector3> lines;

-	for(int i=0;i<points.size();i++) {

-

-		int n = (i+1)%points.size();

-		lines.push_back(Vector3(points[i].x,points[i].y,depth));

-		lines.push_back(Vector3(points[n].x,points[n].y,depth));

-		lines.push_back(Vector3(points[i].x,points[i].y,-depth));

-		lines.push_back(Vector3(points[n].x,points[n].y,-depth));

-		lines.push_back(Vector3(points[i].x,points[i].y,depth));

-		lines.push_back(Vector3(points[i].x,points[i].y,-depth));

-

-	}

-

-	add_lines(lines,SpatialEditorGizmos::singleton->shape_material);

-	add_collision_segments(lines);

-}

-

-CollisionPolygonSpatialGizmo::CollisionPolygonSpatialGizmo(CollisionPolygon* p_polygon){

-

-	set_spatial_node(p_polygon);

-	polygon=p_polygon;

-}

-///

-

-

-String VisibilityNotifierGizmo::get_handle_name(int p_idx) const {

-

-	switch(p_idx) {

-		case 0: return "X";

-		case 1: return "Y";

-		case 2: return "Z";

-	}

-

-	return "";

-}

-Variant VisibilityNotifierGizmo::get_handle_value(int p_idx) const{

-

-	return notifier->get_aabb();

-}

-void VisibilityNotifierGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){

-

-

-	Transform gt = notifier->get_global_transform();

-	//gt.orthonormalize();

-	Transform gi = gt.affine_inverse();

-

-	AABB aabb = notifier->get_aabb();

-	Vector3 ray_from = p_camera->project_ray_origin(p_point);

-	Vector3 ray_dir = p_camera->project_ray_normal(p_point);

-

-	Vector3 sg[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};

-	Vector3 ofs = aabb.pos+aabb.size*0.5;;

-

-	Vector3 axis;

-	axis[p_idx]=1.0;

-

-	Vector3 ra,rb;

-	Geometry::get_closest_points_between_segments(ofs,ofs+axis*4096,sg[0],sg[1],ra,rb);

-	float d = ra[p_idx];

-	if (d<0.001)

-		d=0.001;

-

-	Vector3 he = aabb.size;

-	aabb.pos[p_idx]=(aabb.pos[p_idx]+aabb.size[p_idx]*0.5)-d;

-	aabb.size[p_idx]=d*2;

-	notifier->set_aabb(aabb);

-}

-

-void VisibilityNotifierGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){

-

-

-	if (p_cancel) {

-		notifier->set_aabb(p_restore);

-		return;

-	}

-

-	UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();

-	ur->create_action("Change Notifier Extents");

-	ur->add_do_method(notifier,"set_aabb",notifier->get_aabb());

-	ur->add_undo_method(notifier,"set_aabb",p_restore);

-	ur->commit_action();

-

-}

-

-void VisibilityNotifierGizmo::redraw(){

-

-	clear();

-

-	Vector<Vector3> lines;

-	AABB aabb = notifier->get_aabb();

-

-	for(int i=0;i<12;i++) {

-		Vector3 a,b;

-		aabb.get_edge(i,a,b);

-		lines.push_back(a);

-		lines.push_back(b);

-	}

-

-	Vector<Vector3> handles;

-

-

-	for(int i=0;i<3;i++) {

-

-		Vector3 ax;

-		ax[i]=aabb.pos[i]+aabb.size[i];

-		handles.push_back(ax);

-	}

-

-	add_lines(lines,SpatialEditorGizmos::singleton->visibility_notifier_material);

-	//add_unscaled_billboard(SpatialEditorGizmos::singleton->visi,0.05);

-	add_collision_segments(lines);

-	add_handles(handles);

-

-}

-VisibilityNotifierGizmo::VisibilityNotifierGizmo(VisibilityNotifier* p_notifier){

-

-	notifier=p_notifier;

-	set_spatial_node(p_notifier);

-}

-

-////////

-

-

-

-void NavigationMeshSpatialGizmo::redraw() {

-

-	clear();

-	Ref<NavigationMesh> navmeshie = navmesh->get_navigation_mesh();

-	if (navmeshie.is_null())

-		return;

-

-	DVector<Vector3> vertices = navmeshie->get_vertices();

-	DVector<Vector3>::Read vr=vertices.read();

-	List<Face3> faces;

-	for(int i=0;i<navmeshie->get_polygon_count();i++) {

-		Vector<int> p = navmeshie->get_polygon(i);

-

-		for(int j=2;j<p.size();j++) {

-			Face3 f;

-			f.vertex[0]=vr[p[0]];

-			f.vertex[1]=vr[p[j-1]];

-			f.vertex[2]=vr[p[j]];

-

-			faces.push_back(f);

-		}

-	}

-

-

-	Map<_EdgeKey,bool> edge_map;

-	DVector<Vector3> tmeshfaces;

-	tmeshfaces.resize(faces.size()*3);

-

-	{

-		DVector<Vector3>::Write tw=tmeshfaces.write();

-		int tidx=0;

-

-

-		for(List<Face3>::Element *E=faces.front();E;E=E->next()) {

-

-			const Face3 &f = E->get();

-

-			for(int j=0;j<3;j++) {

-

-				tw[tidx++]=f.vertex[j];

-				_EdgeKey ek;

-				ek.from=f.vertex[j].snapped(CMP_EPSILON);

-				ek.to=f.vertex[(j+1)%3].snapped(CMP_EPSILON);

-				if (ek.from<ek.to)

-					SWAP(ek.from,ek.to);

-

-				Map<_EdgeKey,bool>::Element *E=edge_map.find(ek);

-

-				if (E) {

-

-					E->get()=false;

-

-				} else {

-

-					edge_map[ek]=true;

-				}

-

-			}

-		}

-	}

-	Vector<Vector3> lines;

-

-	for(Map<_EdgeKey,bool>::Element *E=edge_map.front();E;E=E->next()) {

-

-		if (E->get()) {

-			lines.push_back(E->key().from);

-			lines.push_back(E->key().to);

-		}

-	}

-

-	Ref<TriangleMesh> tmesh = memnew( TriangleMesh);

-	tmesh->create(tmeshfaces);

-

-	if (lines.size())

-		add_lines(lines,navmesh->is_enabled()?SpatialEditorGizmos::singleton->navmesh_edge_material:SpatialEditorGizmos::singleton->navmesh_edge_material_disabled);

-	add_collision_triangles(tmesh);

-	Ref<Mesh> m = memnew( Mesh );

-	Array a;

-	a.resize(Mesh::ARRAY_MAX);

-	a[0]=tmeshfaces;

-	m->add_surface(Mesh::PRIMITIVE_TRIANGLES,a);

-	m->surface_set_material(0,navmesh->is_enabled()?SpatialEditorGizmos::singleton->navmesh_solid_material:SpatialEditorGizmos::singleton->navmesh_solid_material_disabled);

-	add_mesh(m);

-	add_collision_segments(lines);

-

-}

-

-NavigationMeshSpatialGizmo::NavigationMeshSpatialGizmo(NavigationMeshInstance *p_navmesh){

-

-	set_spatial_node(p_navmesh);

-	navmesh=p_navmesh;

-}

-

-//////

-///

-///

-

-

-

-void PinJointSpatialGizmo::redraw() {

-

-	clear();

-	Vector<Vector3> cursor_points;

-	float cs = 0.25;

-	cursor_points.push_back(Vector3(+cs,0,0));

-	cursor_points.push_back(Vector3(-cs,0,0));

-	cursor_points.push_back(Vector3(0,+cs,0));

-	cursor_points.push_back(Vector3(0,-cs,0));

-	cursor_points.push_back(Vector3(0,0,+cs));

-	cursor_points.push_back(Vector3(0,0,-cs));

-	add_collision_segments(cursor_points);

-	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);

-

-}

-

-

-PinJointSpatialGizmo::PinJointSpatialGizmo(PinJoint* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-////

-

-void HingeJointSpatialGizmo::redraw() {

-

-	clear();

-	Vector<Vector3> cursor_points;

-	float cs = 0.25;

-	/*cursor_points.push_back(Vector3(+cs,0,0));

-	cursor_points.push_back(Vector3(-cs,0,0));

-	cursor_points.push_back(Vector3(0,+cs,0));

-	cursor_points.push_back(Vector3(0,-cs,0));*/

-	cursor_points.push_back(Vector3(0,0,+cs*2));

-	cursor_points.push_back(Vector3(0,0,-cs*2));

-

-	float ll = p3d->get_param(HingeJoint::PARAM_LIMIT_LOWER);

-	float ul = p3d->get_param(HingeJoint::PARAM_LIMIT_UPPER);

-

-	if (p3d->get_flag(HingeJoint::FLAG_USE_LIMIT) && ll<ul) {

-

-		const int points = 32;

-		float step = (ul-ll)/points;

-

-

-		for(int i=0;i<points;i++) {

-

-			float s = ll+i*(ul-ll)/points;

-			float n = ll+(i+1)*(ul-ll)/points;

-

-			Vector3 from=Vector3( -Math::sin(s),Math::cos(s), 0 )*cs;

-			Vector3 to=Vector3( -Math::sin(n),Math::cos(n), 0 )*cs;

-

-			if (i==points-1) {

-				cursor_points.push_back(to);

-				cursor_points.push_back(Vector3());

-			}

-			if (i==0) {

-				cursor_points.push_back(from);

-				cursor_points.push_back(Vector3());

-			}

-

-

-			cursor_points.push_back(from);

-			cursor_points.push_back(to);

-

-

-		}

-

-		cursor_points.push_back(Vector3(0,cs*1.5,0));

-		cursor_points.push_back(Vector3());

-

-	} else {

-

-

-		const int points = 32;

-

-		for(int i=0;i<points;i++) {

-

-			float s = ll+i*(Math_PI*2.0)/points;

-			float n = ll+(i+1)*(Math_PI*2.0)/points;

-

-			Vector3 from=Vector3( -Math::sin(s),Math::cos(s), 0 )*cs;

-			Vector3 to=Vector3( -Math::sin(n),Math::cos(n), 0 )*cs;

-

-			cursor_points.push_back(from);

-			cursor_points.push_back(to);

-

-		}

-

-	}

-	add_collision_segments(cursor_points);

-	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);

-

-}

-

-

-HingeJointSpatialGizmo::HingeJointSpatialGizmo(HingeJoint* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-///////

-///

-////

-

-void SliderJointSpatialGizmo::redraw() {

-

-	clear();

-	Vector<Vector3> cursor_points;

-	float cs = 0.25;

-	/*cursor_points.push_back(Vector3(+cs,0,0));

-	cursor_points.push_back(Vector3(-cs,0,0));

-	cursor_points.push_back(Vector3(0,+cs,0));

-	cursor_points.push_back(Vector3(0,-cs,0));*/

-	cursor_points.push_back(Vector3(0,0,+cs*2));

-	cursor_points.push_back(Vector3(0,0,-cs*2));

-

-	float ll = p3d->get_param(SliderJoint::PARAM_ANGULAR_LIMIT_LOWER);

-	float ul = p3d->get_param(SliderJoint::PARAM_ANGULAR_LIMIT_UPPER);

-	float lll = -p3d->get_param(SliderJoint::PARAM_LINEAR_LIMIT_LOWER);

-	float lul = -p3d->get_param(SliderJoint::PARAM_LINEAR_LIMIT_UPPER);

-

-	if (lll>lul) {

-

-		cursor_points.push_back(Vector3(lul,0,0));

-		cursor_points.push_back(Vector3(lll,0,0));

-

-		cursor_points.push_back(Vector3(lul,-cs,-cs));

-		cursor_points.push_back(Vector3(lul,-cs,cs));

-		cursor_points.push_back(Vector3(lul,-cs,cs));

-		cursor_points.push_back(Vector3(lul,cs,cs));

-		cursor_points.push_back(Vector3(lul,cs,cs));

-		cursor_points.push_back(Vector3(lul,cs,-cs));

-		cursor_points.push_back(Vector3(lul,cs,-cs));

-		cursor_points.push_back(Vector3(lul,-cs,-cs));

-

-

-		cursor_points.push_back(Vector3(lll,-cs,-cs));

-		cursor_points.push_back(Vector3(lll,-cs,cs));

-		cursor_points.push_back(Vector3(lll,-cs,cs));

-		cursor_points.push_back(Vector3(lll,cs,cs));

-		cursor_points.push_back(Vector3(lll,cs,cs));

-		cursor_points.push_back(Vector3(lll,cs,-cs));

-		cursor_points.push_back(Vector3(lll,cs,-cs));

-		cursor_points.push_back(Vector3(lll,-cs,-cs));

-

-

-	} else {

-

-		cursor_points.push_back(Vector3(+cs*2,0,0));

-		cursor_points.push_back(Vector3(-cs*2,0,0));

-

-	}

-

-	if (ll<ul) {

-

-		const int points = 32;

-		float step = (ul-ll)/points;

-

-

-		for(int i=0;i<points;i++) {

-

-			float s = ll+i*(ul-ll)/points;

-			float n = ll+(i+1)*(ul-ll)/points;

-

-			Vector3 from=Vector3(0, Math::cos(s), -Math::sin(s) )*cs;

-			Vector3 to=Vector3(0,Math::cos(n),  -Math::sin(n) )*cs;

-

-			if (i==points-1) {

-				cursor_points.push_back(to);

-				cursor_points.push_back(Vector3());

-			}

-			if (i==0) {

-				cursor_points.push_back(from);

-				cursor_points.push_back(Vector3());

-			}

-

-

-			cursor_points.push_back(from);

-			cursor_points.push_back(to);

-

-

-		}

-

-		cursor_points.push_back(Vector3(0,cs*1.5,0));

-		cursor_points.push_back(Vector3());

-

-	} else {

-

-

-		const int points = 32;

-

-		for(int i=0;i<points;i++) {

-

-			float s = ll+i*(Math_PI*2.0)/points;

-			float n = ll+(i+1)*(Math_PI*2.0)/points;

-

-			Vector3 from=Vector3(0,Math::cos(s),-Math::sin(s) )*cs;

-			Vector3 to=Vector3( 0,Math::cos(n),-Math::sin(n) )*cs;

-

-			cursor_points.push_back(from);

-			cursor_points.push_back(to);

-

-		}

-

-	}

-	add_collision_segments(cursor_points);

-	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);

-

-}

-

-

-SliderJointSpatialGizmo::SliderJointSpatialGizmo(SliderJoint* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-///////

-///

-////

-

-void ConeTwistJointSpatialGizmo::redraw() {

-

-	clear();

-	float cs = 0.25;

-	Vector<Vector3> points;

-

-	float r = 1.0;

-	float w = r*Math::sin(p3d->get_param(ConeTwistJoint::PARAM_SWING_SPAN));

-	float d = r*Math::cos(p3d->get_param(ConeTwistJoint::PARAM_SWING_SPAN));

-

-

-	//swing

-	for(int i=0;i<360;i+=10) {

-

-		float ra=Math::deg2rad(i);

-		float rb=Math::deg2rad(i+10);

-		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;

-		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;

-

-		/*points.push_back(Vector3(a.x,0,a.y));

-		points.push_back(Vector3(b.x,0,b.y));

-		points.push_back(Vector3(0,a.x,a.y));

-		points.push_back(Vector3(0,b.x,b.y));*/

-		points.push_back(Vector3(d,a.x,a.y));

-		points.push_back(Vector3(d,b.x,b.y));

-

-		if (i%90==0) {

-

-			points.push_back(Vector3(d,a.x,a.y));

-			points.push_back(Vector3());

-

-		}

-	}

-

-	points.push_back(Vector3());

-	points.push_back(Vector3(1,0,0));

-

-	//twist

-	/*

-	 */

-	float ts=Math::rad2deg(p3d->get_param(ConeTwistJoint::PARAM_TWIST_SPAN));

-	ts=MIN(ts,720);

-

-

-	for(int i=0;i<int(ts);i+=5) {

-

-		float ra=Math::deg2rad(i);

-		float rb=Math::deg2rad(i+5);

-		float c = i/720.0;

-		float cn = (i+5)/720.0;

-		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w*c;

-		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w*cn;

-

-		/*points.push_back(Vector3(a.x,0,a.y));

-		points.push_back(Vector3(b.x,0,b.y));

-		points.push_back(Vector3(0,a.x,a.y));

-		points.push_back(Vector3(0,b.x,b.y));*/

-

-		points.push_back(Vector3(c,a.x,a.y));

-		points.push_back(Vector3(cn,b.x,b.y));

-

-	}

-

-

-	add_collision_segments(points);

-	add_lines(points,SpatialEditorGizmos::singleton->joint_material);

-

-}

-

-

-ConeTwistJointSpatialGizmo::ConeTwistJointSpatialGizmo(ConeTwistJoint* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-////////

-/// \brief SpatialEditorGizmos::singleton

-///

-///////

-///

-////

-

-void Generic6DOFJointSpatialGizmo::redraw() {

-

-	clear();

-	Vector<Vector3> cursor_points;

-	float cs = 0.25;

-

-	for(int ax=0;ax<3;ax++) {

-		/*cursor_points.push_back(Vector3(+cs,0,0));

-		cursor_points.push_back(Vector3(-cs,0,0));

-		cursor_points.push_back(Vector3(0,+cs,0));

-		cursor_points.push_back(Vector3(0,-cs,0));

-		cursor_points.push_back(Vector3(0,0,+cs*2));

-		cursor_points.push_back(Vector3(0,0,-cs*2)); */

-

-		float ll;

-		float ul;

-		float lll;

-		float lul;

-

-		int a1,a2,a3;

-		bool enable_ang;

-		bool enable_lin;

-

-		switch(ax) {

-			case 0:

-				ll = p3d->get_param_x(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);

-				ul = p3d->get_param_x(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);

-				lll = -p3d->get_param_x(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);

-				lul = -p3d->get_param_x(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);

-				enable_ang = p3d->get_flag_x(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);

-				enable_lin = p3d->get_flag_x(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);

-				a1=0;

-				a2=1;

-				a3=2;

-			break;

-			case 1:

-				ll = p3d->get_param_y(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);

-				ul = p3d->get_param_y(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);

-				lll = -p3d->get_param_y(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);

-				lul = -p3d->get_param_y(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);

-				enable_ang = p3d->get_flag_y(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);

-				enable_lin = p3d->get_flag_y(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);

-				a1=2;

-				a2=0;

-				a3=1;

-			break;

-			case 2:

-				ll = p3d->get_param_z(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);

-				ul = p3d->get_param_z(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);

-				lll = -p3d->get_param_z(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);

-				lul = -p3d->get_param_z(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);

-				enable_ang = p3d->get_flag_z(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);

-				enable_lin = p3d->get_flag_z(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);

-

-				a1=1;

-				a2=2;

-				a3=0;

-			break;

-		}

-

-#define ADD_VTX(x,y,z)\

-		{\

-			Vector3 v;\

-			v[a1]=(x);\

-			v[a2]=(y);\

-			v[a3]=(z);\

-			cursor_points.push_back(v);\

-		}

-

-#define SET_VTX(what,x,y,z)\

-		{\

-			Vector3 v;\

-			v[a1]=(x);\

-			v[a2]=(y);\

-			v[a3]=(z);\

-			what=v;\

-		}

-

-

-

-

-		if (enable_lin && lll>=lul) {

-

-			ADD_VTX(lul,0,0);

-			ADD_VTX(lll,0,0);

-

-			ADD_VTX(lul,-cs,-cs);

-			ADD_VTX(lul,-cs,cs);

-			ADD_VTX(lul,-cs,cs);

-			ADD_VTX(lul,cs,cs);

-			ADD_VTX(lul,cs,cs);

-			ADD_VTX(lul,cs,-cs);

-			ADD_VTX(lul,cs,-cs);

-			ADD_VTX(lul,-cs,-cs);

-

-

-			ADD_VTX(lll,-cs,-cs);

-			ADD_VTX(lll,-cs,cs);

-			ADD_VTX(lll,-cs,cs);

-			ADD_VTX(lll,cs,cs);

-			ADD_VTX(lll,cs,cs);

-			ADD_VTX(lll,cs,-cs);

-			ADD_VTX(lll,cs,-cs);

-			ADD_VTX(lll,-cs,-cs);

-

-

-		} else {

-

-			ADD_VTX(+cs*2,0,0);

-			ADD_VTX(-cs*2,0,0);

-

-		}

-

-		if (enable_ang && ll<=ul) {

-

-			const int points = 32;

-			float step = (ul-ll)/points;

-

-

-			for(int i=0;i<points;i++) {

-

-				float s = ll+i*(ul-ll)/points;

-				float n = ll+(i+1)*(ul-ll)/points;

-

-				Vector3 from;

-				SET_VTX(from,0, Math::cos(s), -Math::sin(s) );

-				from*=cs;

-				Vector3 to;

-				SET_VTX(to,0,Math::cos(n),  -Math::sin(n));

-				to*=cs;

-

-				if (i==points-1) {

-					cursor_points.push_back(to);

-					cursor_points.push_back(Vector3());

-				}

-				if (i==0) {

-					cursor_points.push_back(from);

-					cursor_points.push_back(Vector3());

-				}

-

-

-				cursor_points.push_back(from);

-				cursor_points.push_back(to);

-

-

-			}

-

-			ADD_VTX(0,cs*1.5,0);

-			cursor_points.push_back(Vector3());

-

-		} else {

-

-

-			const int points = 32;

-

-			for(int i=0;i<points;i++) {

-

-				float s = ll+i*(Math_PI*2.0)/points;

-				float n = ll+(i+1)*(Math_PI*2.0)/points;

-

-//				Vector3 from=Vector3(0,Math::cos(s),-Math::sin(s) )*cs;

-//				Vector3 to=Vector3( 0,Math::cos(n),-Math::sin(n) )*cs;

-

-				Vector3 from;

-				SET_VTX(from,0, Math::cos(s), -Math::sin(s) );

-				from*=cs;

-				Vector3 to;

-				SET_VTX(to,0,Math::cos(n),  -Math::sin(n));

-				to*=cs;

-

-				cursor_points.push_back(from);

-				cursor_points.push_back(to);

-

-			}

-

-		}

-	}

-

-#undef ADD_VTX

-#undef SET_VTX

-

-	add_collision_segments(cursor_points);

-	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);

-

-}

-

-

-Generic6DOFJointSpatialGizmo::Generic6DOFJointSpatialGizmo(Generic6DOFJoint* p_p3d) {

-

-	p3d=p_p3d;

-	set_spatial_node(p3d);

-}

-

-///////

-///

-////

-

-

-SpatialEditorGizmos *SpatialEditorGizmos::singleton=NULL;

-

-Ref<SpatialEditorGizmo> SpatialEditorGizmos::get_gizmo(Spatial *p_spatial) {

-

-	if (p_spatial->cast_to<Light>()) {

-

-		Ref<LightSpatialGizmo> lsg = memnew( LightSpatialGizmo(p_spatial->cast_to<Light>()) );

-		return lsg;

-	}

-

-	if (p_spatial->cast_to<Camera>()) {

-

-		Ref<CameraSpatialGizmo> lsg = memnew( CameraSpatialGizmo(p_spatial->cast_to<Camera>()) );

-		return lsg;

-	}

-

-	if (p_spatial->cast_to<Skeleton>()) {

-

-		Ref<SkeletonSpatialGizmo> lsg = memnew( SkeletonSpatialGizmo(p_spatial->cast_to<Skeleton>()) );

-		return lsg;

-	}

-

-

-	if (p_spatial->cast_to<Position3D>()) {

-

-		Ref<Position3DSpatialGizmo> lsg = memnew( Position3DSpatialGizmo(p_spatial->cast_to<Position3D>()) );

-		return lsg;

-	}

-

-	if (p_spatial->cast_to<MeshInstance>()) {

-

-		Ref<MeshInstanceSpatialGizmo> misg = memnew( MeshInstanceSpatialGizmo(p_spatial->cast_to<MeshInstance>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<Room>()) {

-

-		Ref<RoomSpatialGizmo> misg = memnew( RoomSpatialGizmo(p_spatial->cast_to<Room>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<NavigationMeshInstance>()) {

-

-		Ref<NavigationMeshSpatialGizmo> misg = memnew( NavigationMeshSpatialGizmo(p_spatial->cast_to<NavigationMeshInstance>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<RayCast>()) {

-

-		Ref<RayCastSpatialGizmo> misg = memnew( RayCastSpatialGizmo(p_spatial->cast_to<RayCast>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<Portal>()) {

-

-		Ref<PortalSpatialGizmo> misg = memnew( PortalSpatialGizmo(p_spatial->cast_to<Portal>()) );

-		return misg;

-	}

-

-

-	if (p_spatial->cast_to<TestCube>()) {

-

-		Ref<TestCubeSpatialGizmo> misg = memnew( TestCubeSpatialGizmo(p_spatial->cast_to<TestCube>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<SpatialPlayer>()) {

-

-		Ref<SpatialPlayerSpatialGizmo> misg = memnew( SpatialPlayerSpatialGizmo(p_spatial->cast_to<SpatialPlayer>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<CollisionShape>()) {

-

-		Ref<CollisionShapeSpatialGizmo> misg = memnew( CollisionShapeSpatialGizmo(p_spatial->cast_to<CollisionShape>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<VisibilityNotifier>()) {

-

-		Ref<VisibilityNotifierGizmo> misg = memnew( VisibilityNotifierGizmo(p_spatial->cast_to<VisibilityNotifier>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<VehicleWheel>()) {

-

-		Ref<VehicleWheelSpatialGizmo> misg = memnew( VehicleWheelSpatialGizmo(p_spatial->cast_to<VehicleWheel>()) );

-		return misg;

-	}

-	if (p_spatial->cast_to<PinJoint>()) {

-

-		Ref<PinJointSpatialGizmo> misg = memnew( PinJointSpatialGizmo(p_spatial->cast_to<PinJoint>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<HingeJoint>()) {

-

-		Ref<HingeJointSpatialGizmo> misg = memnew( HingeJointSpatialGizmo(p_spatial->cast_to<HingeJoint>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<SliderJoint>()) {

-

-		Ref<SliderJointSpatialGizmo> misg = memnew( SliderJointSpatialGizmo(p_spatial->cast_to<SliderJoint>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<ConeTwistJoint>()) {

-

-		Ref<ConeTwistJointSpatialGizmo> misg = memnew( ConeTwistJointSpatialGizmo(p_spatial->cast_to<ConeTwistJoint>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<Generic6DOFJoint>()) {

-

-		Ref<Generic6DOFJointSpatialGizmo> misg = memnew( Generic6DOFJointSpatialGizmo(p_spatial->cast_to<Generic6DOFJoint>()) );

-		return misg;

-	}

-

-	if (p_spatial->cast_to<CollisionPolygon>()) {

-

-		Ref<CollisionPolygonSpatialGizmo> misg = memnew( CollisionPolygonSpatialGizmo(p_spatial->cast_to<CollisionPolygon>()) );

-		return misg;

-	}

-

-

-	return Ref<SpatialEditorGizmo>();

-}

-

-

-Ref<FixedMaterial> SpatialEditorGizmos::create_line_material(const Color& p_base_color) {

-

-	Ref<FixedMaterial> line_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	line_material->set_flag(Material::FLAG_UNSHADED, true);

-	line_material->set_line_width(3.0);

-	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, true);

-	line_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,p_base_color);

-

-	return line_material;

-

-}

-

-Ref<FixedMaterial> SpatialEditorGizmos::create_solid_material(const Color& p_base_color) {

-

-	Ref<FixedMaterial> line_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	line_material->set_flag(Material::FLAG_UNSHADED, true);

-	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	line_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,p_base_color);

-

-	return line_material;

-

-}

-

-SpatialEditorGizmos::SpatialEditorGizmos() {

-

-	singleton=this;

-

-	handle_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	handle_material->set_flag(Material::FLAG_UNSHADED, true);

-	handle_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(0.8,0.8,0.8));

-

-	handle2_material = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	handle2_material->set_flag(Material::FLAG_UNSHADED, true);

-	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_POINT_SIZE, true);

-	handle_t = SpatialEditor::get_singleton()->get_icon("Editor3DHandle","EditorIcons");

-	handle2_material->set_point_size(handle_t->get_width());

-	handle2_material->set_texture(FixedMaterial::PARAM_DIFFUSE,handle_t);

-	handle2_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1));

-	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, true);

-

-	light_material = create_line_material(Color(1,1,0.2));

-

-	light_material_omni_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	light_material_omni_icon->set_flag(Material::FLAG_UNSHADED, true);

-	light_material_omni_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);

-	light_material_omni_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-	light_material_omni_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	light_material_omni_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));

-	light_material_omni_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoLight","EditorIcons"));

-

-

-	light_material_directional_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	light_material_directional_icon->set_flag(Material::FLAG_UNSHADED, true);

-	light_material_directional_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);

-	light_material_directional_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-	light_material_directional_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	light_material_directional_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));

-	light_material_directional_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoDirectionalLight","EditorIcons"));

-

-	camera_material = create_line_material(Color(1.0,0.5,1.0));

-

-

-	navmesh_edge_material = create_line_material(Color(0.1,0.8,1.0));

-	navmesh_solid_material = create_solid_material(Color(0.1,0.8,1.0,0.4));

-	navmesh_edge_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, false);

-	navmesh_solid_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);

-

-	navmesh_edge_material_disabled = create_line_material(Color(1.0,0.8,0.1));

-	navmesh_solid_material_disabled = create_solid_material(Color(1.0,0.8,0.1,0.4));

-	navmesh_edge_material_disabled->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, false);

-	navmesh_solid_material_disabled->set_flag(Material::FLAG_DOUBLE_SIDED,true);

-

-	skeleton_material = create_line_material(Color(0.6,1.0,0.3));

-	skeleton_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);

-	skeleton_material->set_flag(Material::FLAG_UNSHADED,true);

-	skeleton_material->set_flag(Material::FLAG_ONTOP,true);

-	skeleton_material->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-

-	//position 3D Shared mesh

-

-	pos3d_mesh = Ref<Mesh>( memnew( Mesh ) );

-	{

-

-		DVector<Vector3> cursor_points;

-		DVector<Color> cursor_colors;

-		float cs = 0.25;

-		cursor_points.push_back(Vector3(+cs,0,0));

-		cursor_points.push_back(Vector3(-cs,0,0));

-		cursor_points.push_back(Vector3(0,+cs,0));

-		cursor_points.push_back(Vector3(0,-cs,0));

-		cursor_points.push_back(Vector3(0,0,+cs));

-		cursor_points.push_back(Vector3(0,0,-cs));

-		cursor_colors.push_back(Color(1,0.5,0.5,0.7));

-		cursor_colors.push_back(Color(1,0.5,0.5,0.7));

-		cursor_colors.push_back(Color(0.5,1,0.5,0.7));

-		cursor_colors.push_back(Color(0.5,1,0.5,0.7));

-		cursor_colors.push_back(Color(0.5,0.5,1,0.7));

-		cursor_colors.push_back(Color(0.5,0.5,1,0.7));

-

-		Ref<FixedMaterial> mat = memnew( FixedMaterial );

-		mat->set_flag(Material::FLAG_UNSHADED,true);

-		mat->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY,true);

-		mat->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA,true);

-		mat->set_line_width(3);

-		Array d;

-		d.resize(VS::ARRAY_MAX);

-		d[Mesh::ARRAY_VERTEX]=cursor_points;

-		d[Mesh::ARRAY_COLOR]=cursor_colors;

-		pos3d_mesh->add_surface(Mesh::PRIMITIVE_LINES,d);

-		pos3d_mesh->surface_set_material(0,mat);

-	}

-

-

-	sample_player_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	sample_player_icon->set_flag(Material::FLAG_UNSHADED, true);

-	sample_player_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);

-	sample_player_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-	sample_player_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	sample_player_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));

-	sample_player_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoSpatialSamplePlayer","EditorIcons"));

-

-	room_material = create_line_material(Color(1.0,0.6,0.9));

-	portal_material = create_line_material(Color(1.0,0.8,0.6));

-	raycast_material = create_line_material(Color(1.0,0.8,0.6));

-	car_wheel_material = create_line_material(Color(0.6,0.8,1.0));

-	visibility_notifier_material = create_line_material(Color(1.0,0.5,1.0));

-	joint_material = create_line_material(Color(0.6,0.8,1.0));

-

-	stream_player_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	stream_player_icon->set_flag(Material::FLAG_UNSHADED, true);

-	stream_player_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);

-	stream_player_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-	stream_player_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	stream_player_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));

-	stream_player_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoSpatialStreamPlayer","EditorIcons"));

-

-	visibility_notifier_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));

-	visibility_notifier_icon->set_flag(Material::FLAG_UNSHADED, true);

-	visibility_notifier_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);

-	visibility_notifier_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);

-	visibility_notifier_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);

-	visibility_notifier_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));

-	visibility_notifier_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("Visible","EditorIcons"));

-

-	{

-

-		DVector<Vector3> vertices;

-

-#undef ADD_VTX

-#define ADD_VTX(m_idx);\

-	vertices.push_back( face_points[m_idx] );

-

-		for (int i=0;i<6;i++) {

-

-

-			Vector3 face_points[4];

-

-			for (int j=0;j<4;j++) {

-

-				float v[3];

-				v[0]=1.0;

-				v[1]=1-2*((j>>1)&1);

-				v[2]=v[1]*(1-2*(j&1));

-

-				for (int k=0;k<3;k++) {

-

-					if (i<3)

-						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);

-					else

-						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);

-				}

-			}

-			//tri 1

-			ADD_VTX(0);

-			ADD_VTX(1);

-			ADD_VTX(2);

-			//tri 2

-			ADD_VTX(2);

-			ADD_VTX(3);

-			ADD_VTX(0);

-

-		}

-

-		test_cube_tm = Ref<TriangleMesh>( memnew( TriangleMesh ) );

-		test_cube_tm->create(vertices);

-	}

-

-	shape_material = create_line_material(Color(0.2,1,1.0));

-

-

-}

-

+/*************************************************************************/
+/*  spatial_editor_gizmos.cpp                                            */
+/*************************************************************************/
+/*                       This file is part of:                           */
+/*                           GODOT ENGINE                                */
+/*                    http://www.godotengine.org                         */
+/*************************************************************************/
+/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
+/*                                                                       */
+/* Permission is hereby granted, free of charge, to any person obtaining */
+/* a copy of this software and associated documentation files (the       */
+/* "Software"), to deal in the Software without restriction, including   */
+/* without limitation the rights to use, copy, modify, merge, publish,   */
+/* distribute, sublicense, and/or sell copies of the Software, and to    */
+/* permit persons to whom the Software is furnished to do so, subject to */
+/* the following conditions:                                             */
+/*                                                                       */
+/* The above copyright notice and this permission notice shall be        */
+/* included in all copies or substantial portions of the Software.       */
+/*                                                                       */
+/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
+/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
+/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
+/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
+/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
+/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
+/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
+/*************************************************************************/
+#include "spatial_editor_gizmos.h"
+#include "geometry.h"
+#include "scene/3d/camera.h"
+#include "scene/resources/surface_tool.h"
+#include "scene/resources/sphere_shape.h"
+#include "scene/resources/box_shape.h"
+#include "scene/resources/capsule_shape.h"
+#include "scene/resources/ray_shape.h"
+#include "scene/resources/convex_polygon_shape.h"
+#include "scene/resources/plane_shape.h"
+#include "quick_hull.h"
+
+// Keep small children away from this file.
+// It's so ugly it will eat them alive
+
+#define HANDLE_HALF_SIZE 0.05
+
+void SpatialGizmoTool::clear() {
+
+	for(int i=0;i<instances.size();i++) {
+
+		if (instances[i].instance.is_valid())
+			VS::get_singleton()->free(instances[i].instance);
+
+
+	}
+
+	billboard_handle=false;
+	collision_segments.clear();
+	collision_mesh=Ref<TriangleMesh>();
+	instances.clear();
+	handles.clear();
+	secondary_handles.clear();
+}
+
+void SpatialGizmoTool::Instance::create_instance(Spatial *p_base) {
+
+	instance = VS::get_singleton()->instance_create2(mesh->get_rid(),p_base->get_world()->get_scenario());
+	VS::get_singleton()->instance_attach_object_instance_ID(instance,p_base->get_instance_ID());
+	if (billboard)
+		VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_BILLBOARD,true);
+	if (unscaled)
+		VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_DEPH_SCALE,true);
+	if (skeleton.is_valid())
+		VS::get_singleton()->instance_attach_skeleton(instance,skeleton);
+	if (extra_margin)
+		VS::get_singleton()->instance_set_extra_visibility_margin(instance,1);
+	VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_CAST_SHADOW,false);
+	VS::get_singleton()->instance_geometry_set_flag(instance,VS::INSTANCE_FLAG_RECEIVE_SHADOWS,false);
+	VS::get_singleton()->instance_set_layer_mask(instance,1<<SpatialEditorViewport::GIZMO_EDIT_LAYER); //gizmos are 26
+}
+
+
+
+void SpatialGizmoTool::add_mesh(const Ref<Mesh>& p_mesh,bool p_billboard, const RID &p_skeleton) {
+
+	ERR_FAIL_COND(!spatial_node);
+	Instance ins;
+
+	ins.billboard=p_billboard;
+	ins.mesh=p_mesh;
+	ins.skeleton=p_skeleton;
+	if (valid) {
+		ins.create_instance(spatial_node);
+		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());
+	}
+
+	instances.push_back(ins);
+
+}
+
+void SpatialGizmoTool::add_lines(const Vector<Vector3> &p_lines, const Ref<Material> &p_material,bool p_billboard){
+
+	ERR_FAIL_COND(!spatial_node);
+	Instance ins;
+
+	Ref<Mesh> mesh = memnew( Mesh );
+	Array a;
+	a.resize(Mesh::ARRAY_MAX);
+
+	a[Mesh::ARRAY_VERTEX]=p_lines;
+
+	DVector<Color> color;
+	color.resize(p_lines.size());
+	{
+		DVector<Color>::Write w = color.write();
+		for(int i=0;i<p_lines.size();i++) {
+			if (is_selected())
+				w[i]=Color(1,1,1,0.6);
+			else
+				w[i]=Color(1,1,1,0.25);
+		}
+
+	}
+
+	a[Mesh::ARRAY_COLOR]=color;
+
+
+	mesh->add_surface(Mesh::PRIMITIVE_LINES,a);
+	mesh->surface_set_material(0,p_material);
+
+	if (p_billboard) {
+		float md=0;
+		for(int i=0;i<p_lines.size();i++) {
+
+			md=MAX(0,p_lines[i].length());
+
+		}
+		if (md) {
+			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));
+		}
+	}
+
+	ins.billboard=p_billboard;
+	ins.mesh=mesh;
+	if (valid) {
+		ins.create_instance(spatial_node);
+		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());
+	}
+
+	instances.push_back(ins);
+
+}
+
+void SpatialGizmoTool::add_unscaled_billboard(const Ref<Material>& p_material,float p_scale) {
+
+	ERR_FAIL_COND(!spatial_node);
+	Instance ins;
+
+	Vector<Vector3 > vs;
+	Vector<Vector2 > uv;
+
+	vs.push_back(Vector3(-p_scale,p_scale,0));
+	vs.push_back(Vector3(p_scale,p_scale,0));
+	vs.push_back(Vector3(p_scale,-p_scale,0));
+	vs.push_back(Vector3(-p_scale,-p_scale,0));
+
+	uv.push_back(Vector2(1,0));
+	uv.push_back(Vector2(0,0));
+	uv.push_back(Vector2(0,1));
+	uv.push_back(Vector2(1,1));
+
+	Ref<Mesh> mesh = memnew( Mesh );
+	Array a;
+	a.resize(Mesh::ARRAY_MAX);
+	a[Mesh::ARRAY_VERTEX]=vs;
+	a[Mesh::ARRAY_TEX_UV]=uv;
+	mesh->add_surface(Mesh::PRIMITIVE_TRIANGLE_FAN,a);
+	mesh->surface_set_material(0,p_material);
+
+	if (true) {
+		float md=0;
+		for(int i=0;i<vs.size();i++) {
+
+			md=MAX(0,vs[i].length());
+
+		}
+		if (md) {
+			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));
+		}
+	}
+
+	ins.mesh=mesh;
+	ins.unscaled=true;
+	ins.billboard=true;
+	if (valid) {
+		ins.create_instance(spatial_node);
+		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());
+	}
+
+	instances.push_back(ins);
+
+
+}
+
+void SpatialGizmoTool::add_collision_triangles(const Ref<TriangleMesh>& p_tmesh) {
+
+	collision_mesh=p_tmesh;
+}
+
+void SpatialGizmoTool::add_collision_segments(const Vector<Vector3> &p_lines) {
+
+	int from=collision_segments.size();
+	collision_segments.resize(from+p_lines.size());
+	for(int i=0;i<p_lines.size();i++) {
+
+		collision_segments[from+i]=p_lines[i];
+	}
+}
+
+
+void SpatialGizmoTool::add_handles(const Vector<Vector3> &p_handles, bool p_billboard,bool p_secondary){
+
+	billboard_handle=p_billboard;
+
+	if (!is_selected())
+		return;
+
+	ERR_FAIL_COND(!spatial_node);
+
+	ERR_FAIL_COND(!spatial_node);
+	Instance ins;
+
+
+	Ref<Mesh> mesh = memnew( Mesh );
+#if 1
+
+	Array a;
+	a.resize(VS::ARRAY_MAX);
+	a[VS::ARRAY_VERTEX]=p_handles;
+	DVector<Color> colors;
+	{
+		colors.resize(p_handles.size());
+		DVector<Color>::Write w=colors.write();
+		for(int i=0;i<p_handles.size();i++) {
+
+			Color col(1,1,1,1);
+			if (SpatialEditor::get_singleton()->get_over_gizmo_handle()!=i)
+				col=Color(0.9,0.9,0.9,0.9);
+			w[i]=col;
+		}
+
+	}
+	a[VS::ARRAY_COLOR]=colors;
+	mesh->add_surface(Mesh::PRIMITIVE_POINTS,a);
+	mesh->surface_set_material(0,SpatialEditorGizmos::singleton->handle2_material);
+
+	if (p_billboard) {
+		float md=0;
+		for(int i=0;i<p_handles.size();i++) {
+
+			md=MAX(0,p_handles[i].length());
+
+		}
+		if (md) {
+			mesh->set_custom_aabb(AABB(Vector3(-md,-md,-md),Vector3(md,md,md)*2.0));
+		}
+	}
+
+
+
+#else
+	for(int ih=0;ih<p_handles.size();ih++) {
+
+
+		Vector<Vector3> vertices;
+		Vector<Vector3> normals;
+
+		int vtx_idx=0;
+#define ADD_VTX(m_idx);\
+	vertices.push_back( (face_points[m_idx]*HANDLE_HALF_SIZE+p_handles[ih]) );\
+	normals.push_back( normal_points[m_idx] );\
+	vtx_idx++;\
+
+		for (int i=0;i<6;i++) {
+
+
+			Vector3 face_points[4];
+			Vector3 normal_points[4];
+			float uv_points[8]={0,0,0,1,1,1,1,0};
+
+			for (int j=0;j<4;j++) {
+
+				float v[3];
+				v[0]=1.0;
+				v[1]=1-2*((j>>1)&1);
+				v[2]=v[1]*(1-2*(j&1));
+
+				for (int k=0;k<3;k++) {
+
+					if (i<3)
+						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);
+					else
+						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);
+				}
+				normal_points[j]=Vector3();
+				normal_points[j][i%3]=(i>=3?-1:1);
+			}
+			//tri 1
+			ADD_VTX(0);
+			ADD_VTX(1);
+			ADD_VTX(2);
+			//tri 2
+			ADD_VTX(2);
+			ADD_VTX(3);
+			ADD_VTX(0);
+
+		}
+
+
+		Array d;
+		d.resize(VS::ARRAY_MAX);
+		d[VisualServer::ARRAY_NORMAL]= normals ;
+		d[VisualServer::ARRAY_VERTEX]= vertices ;
+
+		mesh->add_surface(Mesh::PRIMITIVE_TRIANGLES,d);
+		mesh->surface_set_material(ih,SpatialEditorGizmos::singleton->handle_material);
+
+
+	}
+#endif
+	ins.mesh=mesh;
+	ins.billboard=p_billboard;
+	ins.extra_margin=true;
+	if (valid) {
+		ins.create_instance(spatial_node);
+		VS::get_singleton()->instance_set_transform(ins.instance,spatial_node->get_global_transform());
+	}
+	instances.push_back(ins);
+	if (!p_secondary) {
+		int chs=handles.size();
+		handles.resize(chs+p_handles.size());
+		for(int i=0;i<p_handles.size();i++) {
+			handles[i+chs]=p_handles[i];
+		}
+	} else {
+
+		int chs=secondary_handles.size();
+		secondary_handles.resize(chs+p_handles.size());
+		for(int i=0;i<p_handles.size();i++) {
+			secondary_handles[i+chs]=p_handles[i];
+		}
+
+	}
+
+}
+
+
+void SpatialGizmoTool::set_spatial_node(Spatial *p_node){
+
+	spatial_node=p_node;
+
+}
+
+bool SpatialGizmoTool::intersect_frustum(const Camera *p_camera,const Vector<Plane> &p_frustum) {
+
+	ERR_FAIL_COND_V(!spatial_node,false);
+	ERR_FAIL_COND_V(!valid,false);
+
+	if (collision_segments.size()) {
+
+		const Plane *p=p_frustum.ptr();
+		int fc=p_frustum.size();
+
+		int vc=collision_segments.size();
+		const Vector3* vptr=collision_segments.ptr();
+		Transform t = spatial_node->get_global_transform();
+
+		for(int i=0;i<vc/2;i++) {
+
+
+			Vector3 a=t.xform(vptr[i*2+0]);
+			Vector3 b=t.xform(vptr[i*2+1]);
+
+			bool any_out=false;
+			for(int j=0;j<fc;j++) {
+
+				if (p[j].distance_to(a) > 0 && p[j].distance_to(b) >0) {
+
+					any_out=true;
+					break;
+				}
+			}
+
+			if (!any_out)
+				return true;
+		}
+
+		return false;
+	}
+
+	return false;
+}
+
+
+bool SpatialGizmoTool::intersect_ray(const Camera *p_camera,const Point2& p_point,  Vector3& r_pos, Vector3& r_normal,int *r_gizmo_handle,bool p_sec_first) {
+
+	ERR_FAIL_COND_V(!spatial_node,false);
+	ERR_FAIL_COND_V(!valid,false);
+
+	if (r_gizmo_handle) {
+
+		Transform t = spatial_node->get_global_transform();
+		t.orthonormalize();
+		if (billboard_handle) {
+			t.set_look_at(t.origin,t.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));
+		}
+		Transform ti=t.affine_inverse();
+
+		Vector3 ray_from=ti.xform(p_camera->project_ray_origin(p_point));
+		Vector3 ray_dir=t.basis.xform_inv(p_camera->project_ray_normal(p_point)).normalized();
+		Vector3 ray_to = ray_from+ray_dir*4096;
+
+		float min_d=1e20;
+		int idx=-1;
+
+		for(int i=0;i<secondary_handles.size();i++) {
+#if 1
+
+
+			Vector3 hpos = t.xform(secondary_handles[i]);
+			Vector2 p = p_camera->unproject_position(hpos);
+			if (p.distance_to(p_point)<SpatialEditorGizmos::singleton->handle_t->get_width()*0.6) {
+
+
+				real_t dp = p_camera->get_transform().origin.distance_to(hpos);
+				if (dp<min_d) {
+
+					r_pos=t.xform(hpos);
+					r_normal=p_camera->get_transform().basis.get_axis(2);
+					min_d=dp;
+					idx=i+handles.size();
+
+				}
+
+			}
+
+#else
+			AABB aabb;
+			aabb.pos=Vector3(-1,-1,-1)*HANDLE_HALF_SIZE;
+			aabb.size=aabb.pos*-2;
+			aabb.pos+=secondary_handles[i];
+
+
+			Vector3 rpos,rnorm;
+
+			if (aabb.intersects_segment(ray_from,ray_to,&rpos,&rnorm)) {
+
+				real_t dp = ray_dir.dot(rpos);
+				if (dp<min_d) {
+
+					r_pos=t.xform(rpos);
+					r_normal=ti.basis.xform_inv(rnorm).normalized();
+					min_d=dp;
+					idx=i+handles.size();
+
+				}
+			}
+#endif
+		}
+
+		if (p_sec_first && idx!=-1) {
+
+			*r_gizmo_handle=idx;
+			return true;
+		}
+
+		min_d=1e20;
+
+		for(int i=0;i<handles.size();i++) {
+
+#if 1
+
+
+			Vector3 hpos = t.xform(handles[i]);
+			Vector2 p = p_camera->unproject_position(hpos);
+			if (p.distance_to(p_point)<SpatialEditorGizmos::singleton->handle_t->get_width()*0.6) {
+
+
+				real_t dp = p_camera->get_transform().origin.distance_to(hpos);
+				if (dp<min_d) {
+
+					r_pos=t.xform(hpos);
+					r_normal=p_camera->get_transform().basis.get_axis(2);
+					min_d=dp;
+					idx=i;
+
+				}
+
+			}
+
+#else
+
+			AABB aabb;
+			aabb.pos=Vector3(-1,-1,-1)*HANDLE_HALF_SIZE;
+			aabb.size=aabb.pos*-2;
+			aabb.pos+=handles[i];
+
+
+			Vector3 rpos,rnorm;
+
+			if (aabb.intersects_segment(ray_from,ray_to,&rpos,&rnorm)) {
+
+				real_t dp = ray_dir.dot(rpos);
+				if (dp<min_d) {
+
+					r_pos=t.xform(rpos);
+					r_normal=ti.basis.xform_inv(rnorm).normalized();
+					min_d=dp;
+					idx=i;
+
+				}
+			}
+#endif
+		}
+
+		if (idx>=0) {
+			*r_gizmo_handle=idx;
+			return true;
+		}
+
+
+	}
+
+	if (collision_segments.size()) {
+
+		Plane camp(p_camera->get_transform().origin,(-p_camera->get_transform().basis.get_axis(2)).normalized());
+
+		int vc=collision_segments.size();
+		const Vector3* vptr=collision_segments.ptr();
+		Transform t = spatial_node->get_global_transform();
+		if (billboard_handle) {
+			t.set_look_at(t.origin,t.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));
+		}
+
+		Vector3 cp;
+		float cpd=1e20;
+
+		for(int i=0;i<vc/2;i++) {
+
+
+			Vector3 a=t.xform(vptr[i*2+0]);
+			Vector3 b=t.xform(vptr[i*2+1]);
+			Vector2 s[2];
+			s[0] = p_camera->unproject_position(a);
+			s[1] = p_camera->unproject_position(b);
+
+
+			Vector2 p = Geometry::get_closest_point_to_segment_2d(p_point,s);
+
+			float pd = p.distance_to(p_point);
+
+			if (pd<cpd) {
+
+
+				float d = s[0].distance_to(s[1]);
+				Vector3 tcp;
+				if (d>0) {
+
+					float d2=s[0].distance_to(p)/d;
+					tcp = a+(b-a)*d2;
+
+				} else {
+					tcp=a;
+
+				}
+
+				if (camp.distance_to(tcp)<p_camera->get_znear())
+					continue;
+				cp=tcp;
+				cpd=pd;
+			}
+		}
+
+		if (cpd<8) {
+
+			r_pos=cp;
+			r_normal=-p_camera->project_ray_normal(p_point);
+			return true;
+		}
+
+		return false;
+	}
+
+
+	if (collision_mesh.is_valid()) {
+		Transform gt = spatial_node->get_global_transform();
+
+		if (billboard_handle) {
+			gt.set_look_at(gt.origin,gt.origin+p_camera->get_transform().basis.get_axis(2),p_camera->get_transform().basis.get_axis(1));
+		}
+
+		Transform ai=gt.affine_inverse();
+		Vector3 ray_from = ai.xform(p_camera->project_ray_origin(p_point));
+		Vector3 ray_dir=ai.basis.xform(p_camera->project_ray_normal(p_point)).normalized();
+		Vector3 rpos,rnorm;
+
+#if 1
+
+
+
+		if (collision_mesh->intersect_ray(ray_from,ray_dir,rpos,rnorm)) {
+
+			r_pos=gt.xform(rpos);
+			r_normal=gt.basis.xform(rnorm).normalized();
+			return true;
+		}
+#else
+
+		if (collision_mesh->intersect_segment(ray_from,ray_from+ray_dir*4906.0,rpos,rnorm)) {
+
+			r_pos=gt.xform(rpos);
+			r_normal=gt.basis.xform(rnorm).normalized();
+			return true;
+		}
+
+#endif
+	}
+
+	return false;
+
+}
+
+
+
+void SpatialGizmoTool::create() {
+
+	ERR_FAIL_COND(!spatial_node);
+	ERR_FAIL_COND(valid);
+	valid=true;
+
+	for(int i=0;i<instances.size();i++) {
+
+		instances[i].create_instance(spatial_node);
+	}
+
+	transform();
+
+}
+
+void SpatialGizmoTool::transform(){
+
+	ERR_FAIL_COND(!spatial_node);
+	ERR_FAIL_COND(!valid);
+	for(int i=0;i<instances.size();i++) {
+		VS::get_singleton()->instance_set_transform(instances[i].instance,spatial_node->get_global_transform());
+	}
+
+}
+
+
+void SpatialGizmoTool::free(){
+
+	ERR_FAIL_COND(!spatial_node);
+	ERR_FAIL_COND(!valid);
+
+	for(int i=0;i<instances.size();i++) {
+
+		if (instances[i].instance.is_valid())
+			VS::get_singleton()->free(instances[i].instance);
+		instances[i].instance=RID();
+	}
+
+	valid=false;
+
+
+}
+
+
+
+SpatialGizmoTool::SpatialGizmoTool() {
+	valid=false;
+	billboard_handle=false;
+
+}
+
+SpatialGizmoTool::~SpatialGizmoTool(){
+
+	clear();
+}
+
+Vector3 SpatialGizmoTool::get_handle_pos(int p_idx) const {
+
+	ERR_FAIL_INDEX_V(p_idx,handles.size(),Vector3());
+
+	return handles[p_idx];
+
+}
+
+//// light gizmo
+
+
+String LightSpatialGizmo::get_handle_name(int p_idx) const {
+
+	if (p_idx==0)
+		return "Radius";
+	else
+		return "Aperture";
+}
+
+
+Variant LightSpatialGizmo::get_handle_value(int p_idx) const{
+
+	if (p_idx==0)
+		return light->get_parameter(Light::PARAM_RADIUS);
+	if (p_idx==1)
+		return light->get_parameter(Light::PARAM_SPOT_ANGLE);
+
+	return Variant();
+}
+
+
+static float _find_closest_angle_to_half_pi_arc(const Vector3& p_from, const Vector3& p_to, float p_arc_radius,const Transform& p_arc_xform) {
+
+	//bleh, discrete is simpler
+	static const int arc_test_points=64;
+	float min_d = 1e20;
+	Vector3 min_p;
+
+
+	for(int i=0;i<arc_test_points;i++) {
+
+		float a = i*Math_PI*0.5/arc_test_points;
+		float an = (i+1)*Math_PI*0.5/arc_test_points;
+		Vector3 p=Vector3( Math::cos(a), 0, -Math::sin(a) )*p_arc_radius;
+		Vector3 n=Vector3( Math::cos(an), 0,- Math::sin(an) )*p_arc_radius;
+
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(p,n,p_from,p_to,ra,rb);
+
+		float d = ra.distance_to(rb);
+		if (d<min_d) {
+			min_d=d;
+			min_p=ra;
+		}
+
+	}
+
+	//min_p = p_arc_xform.affine_inverse().xform(min_p);
+	float a = Vector2(min_p.x,-min_p.z).atan2();
+	return a*180.0/Math_PI;
+}
+
+
+void LightSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point) {
+
+	Transform gt = light->get_global_transform();
+	gt.orthonormalize();
+	Transform gi = gt.affine_inverse();
+
+	Vector3 ray_from = p_camera->project_ray_origin(p_point);
+	Vector3 ray_dir = p_camera->project_ray_normal(p_point);
+
+	Vector3 s[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};
+	if (p_idx==0) {
+
+		if (light->cast_to<SpotLight>()) {
+			Vector3 ra,rb;
+			Geometry::get_closest_points_between_segments(Vector3(),Vector3(0,0,-4096),s[0],s[1],ra,rb);
+
+			float d = -ra.z;
+			if (d<0)
+				d=0;
+
+			light->set_parameter(Light::PARAM_RADIUS,d);
+		} else if (light->cast_to<OmniLight>()) {
+
+			Plane cp=Plane( gt.origin, p_camera->get_transform().basis.get_axis(2));
+
+			Vector3 inters;
+			if (cp.intersects_ray(ray_from,ray_dir,&inters)) {
+
+				float r = inters.distance_to(gt.origin);
+				light->set_parameter(Light::PARAM_RADIUS,r);
+			}
+
+		}
+
+	} else if (p_idx==1) {
+
+		float a = _find_closest_angle_to_half_pi_arc(s[0],s[1],light->get_parameter(Light::PARAM_RADIUS),gt);
+		light->set_parameter(Light::PARAM_SPOT_ANGLE,CLAMP(a,0.01,89.99));
+	}
+}
+
+void LightSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){
+
+	if (p_cancel) {
+
+		light->set_parameter(p_idx==0?Light::PARAM_RADIUS:Light::PARAM_SPOT_ANGLE,p_restore);
+
+	} else if (p_idx==0) {
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		ur->create_action("Change Light Radius");
+		ur->add_do_method(light,"set_parameter",Light::PARAM_RADIUS,light->get_parameter(Light::PARAM_RADIUS));
+		ur->add_undo_method(light,"set_parameter",Light::PARAM_RADIUS,p_restore);
+		ur->commit_action();
+	} else if (p_idx==1) {
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		ur->create_action("Change Light Radius");
+		ur->add_do_method(light,"set_parameter",Light::PARAM_SPOT_ANGLE,light->get_parameter(Light::PARAM_SPOT_ANGLE));
+		ur->add_undo_method(light,"set_parameter",Light::PARAM_SPOT_ANGLE,p_restore);
+		ur->commit_action();
+
+	}
+}
+
+
+
+void LightSpatialGizmo::redraw() {
+
+
+	if (light->cast_to<DirectionalLight>()) {
+
+
+
+		const int arrow_points=5;
+		Vector3 arrow[arrow_points]={
+			Vector3(0,0,2),
+			Vector3(1,1,2),
+			Vector3(1,1,-1),
+			Vector3(2,2,-1),
+			Vector3(0,0,-3)
+		};
+
+		int arrow_sides=4;
+
+		Vector<Vector3> lines;
+
+
+		for(int i = 0; i < arrow_sides ; i++) {
+
+
+			Matrix3 ma(Vector3(0,0,1),Math_PI*2*float(i)/arrow_sides);
+			Matrix3 mb(Vector3(0,0,1),Math_PI*2*float(i+1)/arrow_sides);
+
+
+			for(int j=1;j<arrow_points-1;j++) {
+
+				if (j!=2) {
+					lines.push_back(ma.xform(arrow[j]));
+					lines.push_back(ma.xform(arrow[j+1]));
+				}
+				if (j<arrow_points-1) {
+					lines.push_back(ma.xform(arrow[j]));
+					lines.push_back(mb.xform(arrow[j]));
+				}
+
+			}
+		}
+
+		add_lines(lines,SpatialEditorGizmos::singleton->light_material);
+		add_collision_segments(lines);
+		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_directional_icon,0.05);
+
+	}
+
+	if (light->cast_to<OmniLight>()) {
+
+		clear();
+
+
+		OmniLight *on = light->cast_to<OmniLight>();
+
+		float r = on->get_parameter(Light::PARAM_RADIUS);
+
+		Vector<Vector3> points;
+
+		for(int i=0;i<=360;i++) {
+
+			float ra=Math::deg2rad(i);
+			float rb=Math::deg2rad(i+1);
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;
+
+			/*points.push_back(Vector3(a.x,0,a.y));
+			points.push_back(Vector3(b.x,0,b.y));
+			points.push_back(Vector3(0,a.x,a.y));
+			points.push_back(Vector3(0,b.x,b.y));*/
+			points.push_back(Vector3(a.x,a.y,0));
+			points.push_back(Vector3(b.x,b.y,0));
+
+		}
+
+		add_lines(points,SpatialEditorGizmos::singleton->light_material,true);
+		add_collision_segments(points);
+
+		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_omni_icon,0.05);
+
+		Vector<Vector3> handles;
+		handles.push_back(Vector3(r,0,0));
+		add_handles(handles,true);
+
+
+	}
+
+
+	if (light->cast_to<SpotLight>()) {
+
+		clear();
+
+		Vector<Vector3> points;
+		SpotLight *on = light->cast_to<SpotLight>();
+
+		float r = on->get_parameter(Light::PARAM_RADIUS);
+		float w = r*Math::sin(Math::deg2rad(on->get_parameter(Light::PARAM_SPOT_ANGLE)));
+		float d = r*Math::cos(Math::deg2rad(on->get_parameter(Light::PARAM_SPOT_ANGLE)));
+
+
+
+		for(int i=0;i<360;i++) {
+
+			float ra=Math::deg2rad(i);
+			float rb=Math::deg2rad(i+1);
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;
+
+			/*points.push_back(Vector3(a.x,0,a.y));
+			points.push_back(Vector3(b.x,0,b.y));
+			points.push_back(Vector3(0,a.x,a.y));
+			points.push_back(Vector3(0,b.x,b.y));*/
+			points.push_back(Vector3(a.x,a.y,-d));
+			points.push_back(Vector3(b.x,b.y,-d));
+
+			if (i%90==0) {
+
+				points.push_back(Vector3(a.x,a.y,-d));
+				points.push_back(Vector3());
+
+			}
+
+
+		}
+
+		points.push_back(Vector3(0,0,-r));
+		points.push_back(Vector3());
+
+		add_lines(points,SpatialEditorGizmos::singleton->light_material);
+
+		Vector<Vector3> handles;
+		handles.push_back(Vector3(0,0,-r));
+
+		Vector<Vector3> collision_segments;
+
+		for(int i=0;i<64;i++) {
+
+			float ra=i*Math_PI*2.0/64.0;
+			float rb=(i+1)*Math_PI*2.0/64.0;
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;
+
+			collision_segments.push_back(Vector3(a.x,a.y,-d));
+			collision_segments.push_back(Vector3(b.x,b.y,-d));
+
+			if (i%16==0) {
+
+				collision_segments.push_back(Vector3(a.x,a.y,-d));
+				collision_segments.push_back(Vector3());
+
+			}
+
+			if (i==16) {
+
+				handles.push_back(Vector3(a.x,a.y,-d));
+			}
+
+		}
+
+		collision_segments.push_back(Vector3(0,0,-r));
+		collision_segments.push_back(Vector3());
+
+
+		add_handles(handles);
+		add_collision_segments(collision_segments);
+		add_unscaled_billboard(SpatialEditorGizmos::singleton->light_material_omni_icon,0.05);
+
+	}
+
+}
+
+LightSpatialGizmo::LightSpatialGizmo(Light* p_light){
+
+	light=p_light;
+	set_spatial_node(p_light);
+
+}
+
+//////
+
+String CameraSpatialGizmo::get_handle_name(int p_idx) const {
+
+	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {
+		return "FOV";
+	} else {
+		return "Size";
+	}
+}
+Variant CameraSpatialGizmo::get_handle_value(int p_idx) const{
+
+	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {
+		return camera->get_fov();
+	} else {
+
+		return camera->get_size();
+	}
+}
+void CameraSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){
+
+	Transform gt = camera->get_global_transform();
+	gt.orthonormalize();
+	Transform gi = gt.affine_inverse();
+
+	Vector3 ray_from = p_camera->project_ray_origin(p_point);
+	Vector3 ray_dir = p_camera->project_ray_normal(p_point);
+
+	Vector3 s[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};
+
+	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {
+		Transform gt=camera->get_global_transform();
+		float a = _find_closest_angle_to_half_pi_arc(s[0],s[1],1.0,gt);
+		camera->set("fov",a);
+	} else {
+
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(Vector3(0,0,-1),Vector3(4096,0,-1),s[0],s[1],ra,rb);
+		float d = ra.x * 2.0;
+		if (d<0)
+			d=0;
+
+		camera->set("size",d);
+	}
+
+}
+void CameraSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){
+
+	if (camera->get_projection()==Camera::PROJECTION_PERSPECTIVE) {
+
+		if (p_cancel) {
+
+			camera->set("fov",p_restore);
+		} else {
+			UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+			ur->create_action("Change Camera FOV");
+			ur->add_do_property(camera,"fov",camera->get_fov());
+			ur->add_undo_property(camera,"fov",p_restore);
+			ur->commit_action();
+		}
+
+	} else {
+
+		if (p_cancel) {
+
+			camera->set("size",p_restore);
+		} else {
+			UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+			ur->create_action("Change Camera Size");
+			ur->add_do_property(camera,"size",camera->get_size());
+			ur->add_undo_property(camera,"size",p_restore);
+			ur->commit_action();
+		}
+
+	}
+
+}
+
+void CameraSpatialGizmo::redraw(){
+
+	clear();
+
+	Vector<Vector3> lines;
+	Vector<Vector3> handles;
+
+
+	switch(camera->get_projection()) {
+
+		case Camera::PROJECTION_PERSPECTIVE: {
+
+			float fov = camera->get_fov();
+
+			Vector3 side=Vector3( Math::sin(Math::deg2rad(fov)), 0, -Math::cos(Math::deg2rad(fov)) );
+			Vector3 nside=side;
+			nside.x=-nside.x;
+			Vector3 up=Vector3(0,side.x,0);
+
+
+#define ADD_TRIANGLE( m_a, m_b, m_c)\
+{\
+	lines.push_back(m_a);\
+	lines.push_back(m_b);\
+	lines.push_back(m_b);\
+	lines.push_back(m_c);\
+	lines.push_back(m_c);\
+	lines.push_back(m_a);\
+}
+
+			ADD_TRIANGLE( Vector3(), side+up, side-up );
+			ADD_TRIANGLE( Vector3(), nside+up, nside-up );
+			ADD_TRIANGLE( Vector3(), side+up, nside+up );
+			ADD_TRIANGLE( Vector3(), side-up, nside-up );
+
+			handles.push_back(side);
+			side.x*=0.25;
+			nside.x*=0.25;
+			Vector3 tup( 0, up.y*3/2,side.z);
+			ADD_TRIANGLE( tup, side+up, nside+up );
+
+		} break;
+		case Camera::PROJECTION_ORTHOGONAL: {
+
+#define ADD_QUAD( m_a, m_b, m_c, m_d)\
+{\
+	lines.push_back(m_a);\
+	lines.push_back(m_b);\
+	lines.push_back(m_b);\
+	lines.push_back(m_c);\
+	lines.push_back(m_c);\
+	lines.push_back(m_d);\
+	lines.push_back(m_d);\
+	lines.push_back(m_a);\
+}
+			float size = camera->get_size();
+
+			float hsize=size*0.5;
+			Vector3 right(hsize,0,0);
+			Vector3 up(0,hsize,0);
+			Vector3 back(0,0,-1.0);
+			Vector3 front(0,0,0);
+
+			ADD_QUAD( -up-right,-up+right,up+right,up-right);
+			ADD_QUAD( -up-right+back,-up+right+back,up+right+back,up-right+back);
+			ADD_QUAD( up+right,up+right+back,up-right+back,up-right);
+			ADD_QUAD( -up+right,-up+right+back,-up-right+back,-up-right);
+			handles.push_back(right+back);
+
+			right.x*=0.25;
+			Vector3 tup( 0, up.y*3/2,back.z );
+			ADD_TRIANGLE( tup, right+up+back, -right+up+back );
+
+		} break;
+
+	}
+
+	add_lines(lines,SpatialEditorGizmos::singleton->camera_material);
+	add_collision_segments(lines);
+	add_handles(handles);
+}
+
+
+CameraSpatialGizmo::CameraSpatialGizmo(Camera* p_camera){
+
+	camera=p_camera;
+	set_spatial_node(camera);
+}
+
+
+
+
+//////
+
+void MeshInstanceSpatialGizmo::redraw() {
+
+	Ref<Mesh> m = mesh->get_mesh();
+	if (!m.is_valid())
+		return; //none
+
+	Ref<TriangleMesh> tm = m->generate_triangle_mesh();
+	if (tm.is_valid())
+		add_collision_triangles(tm);
+}
+
+MeshInstanceSpatialGizmo::MeshInstanceSpatialGizmo(MeshInstance* p_mesh) {
+
+	mesh=p_mesh;
+	set_spatial_node(p_mesh);
+}
+
+/////
+
+
+void Position3DSpatialGizmo::redraw() {
+
+	clear();
+	add_mesh(SpatialEditorGizmos::singleton->pos3d_mesh);
+	Vector<Vector3> cursor_points;
+	float cs = 0.25;
+	cursor_points.push_back(Vector3(+cs,0,0));
+	cursor_points.push_back(Vector3(-cs,0,0));
+	cursor_points.push_back(Vector3(0,+cs,0));
+	cursor_points.push_back(Vector3(0,-cs,0));
+	cursor_points.push_back(Vector3(0,0,+cs));
+	cursor_points.push_back(Vector3(0,0,-cs));
+	add_collision_segments(cursor_points);
+
+}
+
+
+Position3DSpatialGizmo::Position3DSpatialGizmo(Position3D* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+
+/////
+
+void SkeletonSpatialGizmo::redraw() {
+
+	clear();
+
+	Ref<SurfaceTool> surface_tool( memnew( SurfaceTool ));
+
+
+	surface_tool->begin(Mesh::PRIMITIVE_LINES);
+	surface_tool->set_material(SpatialEditorGizmos::singleton->skeleton_material);
+	Vector<Transform> grests;
+	grests.resize(skel->get_bone_count());
+
+	Vector<int> bones;
+	Vector<float> weights;
+	bones.resize(4);
+	weights.resize(4);
+
+	for(int i=0;i<4;i++) {
+		bones[i]=0;
+		weights[i]=0;
+	}
+
+	weights[0]=1;
+
+
+	AABB aabb;
+
+	Color bonecolor = Color(1.0,0.4,0.4,0.3);
+	Color rootcolor = Color(0.4,1.0,0.4,0.1);
+
+	for (int i=0;i<skel->get_bone_count();i++) {
+
+		int parent = skel->get_bone_parent(i);
+
+		if (parent>=0) {
+			grests[i]=grests[parent] * skel->get_bone_rest(i);
+
+			Vector3 v0 = grests[parent].origin;
+			Vector3 v1 = grests[i].origin;
+			Vector3 d = (v1-v0).normalized();
+			float dist = v0.distance_to(v1);
+
+			//find closest axis
+			int closest=-1;
+			float closest_d = 0.0;
+
+			for(int j=0;j<3;j++) {
+				float dp = Math::abs(grests[parent].basis[j].normalized().dot(d));
+				if (j==0 || dp>closest_d)
+					closest=j;
+			}
+
+			//find closest other
+			Vector3 first;
+			Vector3 points[4];
+			int pointidx=0;
+			for(int j=0;j<3;j++) {
+
+				bones[0]=parent;
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(rootcolor);
+				surface_tool->add_vertex(v0-grests[parent].basis[j].normalized()*dist*0.05);
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(rootcolor);
+				surface_tool->add_vertex(v0+grests[parent].basis[j].normalized()*dist*0.05);
+
+				if (j==closest)
+					continue;
+
+				Vector3 axis;
+				if (first==Vector3()) {
+					axis = d.cross(d.cross(grests[parent].basis[j])).normalized();
+					first=axis;
+				} else {
+					axis = d.cross(first).normalized();
+				}
+
+				for(int k=0;k<2;k++) {
+
+					if (k==1)
+						axis=-axis;
+					Vector3 point = v0+d*dist*0.2;
+					point+=axis*dist*0.1;
+
+
+					bones[0]=parent;
+					surface_tool->add_bones(bones);
+					surface_tool->add_weights(weights);
+					surface_tool->add_color(bonecolor);
+					surface_tool->add_vertex(v0);
+					surface_tool->add_bones(bones);
+					surface_tool->add_weights(weights);
+					surface_tool->add_color(bonecolor);
+					surface_tool->add_vertex(point);
+
+					bones[0]=parent;
+					surface_tool->add_bones(bones);
+					surface_tool->add_weights(weights);
+					surface_tool->add_color(bonecolor);
+					surface_tool->add_vertex(point);
+					bones[0]=i;
+					surface_tool->add_bones(bones);
+					surface_tool->add_weights(weights);
+					surface_tool->add_color(bonecolor);
+					surface_tool->add_vertex(v1);
+					points[pointidx++]=point;
+
+				}
+
+			}
+
+			SWAP( points[1],points[2] );
+			for(int j=0;j<4;j++) {
+
+
+				bones[0]=parent;
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(bonecolor);
+				surface_tool->add_vertex(points[j]);
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(bonecolor);
+				surface_tool->add_vertex(points[(j+1)%4]);
+			}
+
+
+/*
+			bones[0]=parent;
+			surface_tool->add_bones(bones);
+			surface_tool->add_weights(weights);
+			surface_tool->add_color(Color(0.4,1,0.4,0.4));
+			surface_tool->add_vertex(v0);
+			bones[0]=i;
+			surface_tool->add_bones(bones);
+			surface_tool->add_weights(weights);
+			surface_tool->add_color(Color(0.4,1,0.4,0.4));
+			surface_tool->add_vertex(v1);
+*/
+		} else {
+
+			grests[i]=skel->get_bone_rest(i);
+			bones[0]=i;
+		}
+/*
+		Transform  t = grests[i];
+		t.orthonormalize();
+
+		for (int i=0;i<6;i++) {
+
+
+			Vector3 face_points[4];
+
+			for (int j=0;j<4;j++) {
+
+				float v[3];
+				v[0]=1.0;
+				v[1]=1-2*((j>>1)&1);
+				v[2]=v[1]*(1-2*(j&1));
+
+				for (int k=0;k<3;k++) {
+
+					if (i<3)
+						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);
+					else
+						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);
+				}
+			}
+
+			for(int j=0;j<4;j++) {
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(Color(1.0,0.4,0.4,0.4));
+				surface_tool->add_vertex(t.xform(face_points[j]*0.04));
+				surface_tool->add_bones(bones);
+				surface_tool->add_weights(weights);
+				surface_tool->add_color(Color(1.0,0.4,0.4,0.4));
+				surface_tool->add_vertex(t.xform(face_points[(j+1)%4]*0.04));
+			}
+
+		}
+		*/
+	}
+
+	Ref<Mesh> m = surface_tool->commit();
+	add_mesh(m,false,skel->get_skeleton());
+
+}
+
+SkeletonSpatialGizmo::SkeletonSpatialGizmo(Skeleton* p_skel) {
+
+	skel=p_skel;
+	set_spatial_node(p_skel);
+}
+
+/////
+
+
+void SpatialPlayerSpatialGizmo::redraw() {
+
+	clear();
+	if (splayer->cast_to<SpatialStreamPlayer>()) {
+
+		add_unscaled_billboard(SpatialEditorGizmos::singleton->stream_player_icon,0.05);
+
+	} else if (splayer->cast_to<SpatialSamplePlayer>()) {
+
+		add_unscaled_billboard(SpatialEditorGizmos::singleton->sample_player_icon,0.05);
+
+	}
+
+}
+
+SpatialPlayerSpatialGizmo::SpatialPlayerSpatialGizmo(SpatialPlayer* p_splayer){
+
+	set_spatial_node(p_splayer);
+	splayer=p_splayer;
+}
+
+
+/////
+
+
+void RoomSpatialGizmo::redraw() {
+
+	clear();
+	Ref<RoomBounds> roomie = room->get_room();
+	if (roomie.is_null())
+		return;
+	DVector<Face3> faces = roomie->get_geometry_hint();
+
+	Vector<Vector3> lines;
+	int fc=faces.size();
+	DVector<Face3>::Read r =faces.read();
+
+	Map<_EdgeKey,Vector3> edge_map;
+
+	for(int i=0;i<fc;i++) {
+
+		Vector3 fn = r[i].get_plane().normal;
+
+		for(int j=0;j<3;j++) {
+
+			_EdgeKey ek;
+			ek.from=r[i].vertex[j].snapped(CMP_EPSILON);
+			ek.to=r[i].vertex[(j+1)%3].snapped(CMP_EPSILON);
+			if (ek.from<ek.to)
+				SWAP(ek.from,ek.to);
+
+			Map<_EdgeKey,Vector3>::Element *E=edge_map.find(ek);
+
+			if (E) {
+
+				if (E->get().dot(fn) >0.9) {
+
+					E->get()=Vector3();
+				}
+
+			} else {
+
+				edge_map[ek]=fn;
+			}
+
+		}
+	}
+
+	for(Map<_EdgeKey,Vector3>::Element *E=edge_map.front();E;E=E->next()) {
+
+		if (E->get()!=Vector3()) {
+			lines.push_back(E->key().from);
+			lines.push_back(E->key().to);
+		}
+	}
+
+	add_lines(lines,SpatialEditorGizmos::singleton->room_material);
+	add_collision_segments(lines);
+
+}
+
+RoomSpatialGizmo::RoomSpatialGizmo(Room* p_room){
+
+	set_spatial_node(p_room);
+	room=p_room;
+}
+
+/////
+
+
+void PortalSpatialGizmo::redraw() {
+
+	clear();
+
+	Vector<Point2> points = portal->get_shape();
+	if (points.size()==0) {
+		return;
+	}
+
+	Vector<Vector3> lines;
+
+	Vector3 center;
+	for(int i=0;i<points.size();i++) {
+
+		Vector3 f;
+		f.x=points[i].x;
+		f.y=points[i].y;
+		Vector3 fn;
+		fn.x=points[(i+1)%points.size()].x;
+		fn.y=points[(i+1)%points.size()].y;
+		center+=f;
+
+		lines.push_back(f);
+		lines.push_back(fn);
+	}
+
+	center/=points.size();
+	lines.push_back(center);
+	lines.push_back(center+Vector3(0,0,1));
+
+	add_lines(lines,SpatialEditorGizmos::singleton->portal_material);
+	add_collision_segments(lines);
+
+}
+
+PortalSpatialGizmo::PortalSpatialGizmo(Portal* p_portal){
+
+	set_spatial_node(p_portal);
+	portal=p_portal;
+}
+
+/////
+
+
+void RayCastSpatialGizmo::redraw() {
+
+	clear();
+
+
+	Vector<Vector3> lines;
+
+	lines.push_back(Vector3());
+	lines.push_back(raycast->get_cast_to());
+
+	add_lines(lines,SpatialEditorGizmos::singleton->raycast_material);
+	add_collision_segments(lines);
+
+}
+
+RayCastSpatialGizmo::RayCastSpatialGizmo(RayCast* p_raycast){
+
+	set_spatial_node(p_raycast);
+	raycast=p_raycast;
+}
+
+
+
+/////
+
+
+void VehicleWheelSpatialGizmo::redraw() {
+
+	clear();
+
+
+	Vector<Vector3> points;
+
+	float r = car_wheel->get_radius();
+	const int skip=10;
+	for(int i=0;i<=360;i+=skip) {
+
+		float ra=Math::deg2rad(i);
+		float rb=Math::deg2rad(i+skip);
+		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;
+		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;
+
+		points.push_back(Vector3(0,a.x,a.y));
+		points.push_back(Vector3(0,b.x,b.y));
+
+		const int springsec=4;
+
+		for(int j=0;j<springsec;j++) {
+			float t = car_wheel->get_suspension_rest_length()*5;
+			points.push_back(Vector3(a.x,i/360.0*t/springsec+j*(t/springsec),a.y)*0.2);
+			points.push_back(Vector3(b.x,(i+skip)/360.0*t/springsec+j*(t/springsec),b.y)*0.2);
+		}
+
+
+	}
+
+	//travel
+	points.push_back(Vector3(0,0,0));
+	points.push_back(Vector3(0,car_wheel->get_suspension_rest_length(),0));
+
+	//axis
+	points.push_back(Vector3(r*0.2,car_wheel->get_suspension_rest_length(),0));
+	points.push_back(Vector3(-r*0.2,car_wheel->get_suspension_rest_length(),0));
+	//axis
+	points.push_back(Vector3(r*0.2,0,0));
+	points.push_back(Vector3(-r*0.2,0,0));
+
+	//forward line
+	points.push_back(Vector3(0,-r,0));
+	points.push_back(Vector3(0,-r,r*2));
+	points.push_back(Vector3(0,-r,r*2));
+	points.push_back(Vector3(r*2*0.2,-r,r*2*0.8));
+	points.push_back(Vector3(0,-r,r*2));
+	points.push_back(Vector3(-r*2*0.2,-r,r*2*0.8));
+
+	add_lines(points,SpatialEditorGizmos::singleton->car_wheel_material);
+	add_collision_segments(points);
+
+}
+
+VehicleWheelSpatialGizmo::VehicleWheelSpatialGizmo(VehicleWheel* p_car_wheel){
+
+	set_spatial_node(p_car_wheel);
+	car_wheel=p_car_wheel;
+}
+
+
+
+///
+
+void TestCubeSpatialGizmo::redraw() {
+
+	clear();
+	add_collision_triangles(SpatialEditorGizmos::singleton->test_cube_tm);
+}
+
+TestCubeSpatialGizmo::TestCubeSpatialGizmo(TestCube* p_tc) {
+
+	tc=p_tc;
+	set_spatial_node(p_tc);
+}
+
+
+///////////
+
+
+
+
+
+
+String CollisionShapeSpatialGizmo::get_handle_name(int p_idx) const {
+
+	Ref<Shape> s = cs->get_shape();
+	if (s.is_null())
+		return "";
+
+	if (s->cast_to<SphereShape>()) {
+
+		return "Radius";
+	}
+
+	if (s->cast_to<BoxShape>()) {
+
+		return "Extents";
+	}
+
+	if (s->cast_to<CapsuleShape>()) {
+
+		return p_idx==0?"Radius":"Height";
+	}
+
+	if (s->cast_to<RayShape>()) {
+
+		return "Length";
+	}
+
+	return "";
+}
+Variant CollisionShapeSpatialGizmo::get_handle_value(int p_idx) const{
+
+	Ref<Shape> s = cs->get_shape();
+	if (s.is_null())
+		return Variant();
+
+	if (s->cast_to<SphereShape>()) {
+
+		Ref<SphereShape> ss = s;
+		return ss->get_radius();
+	}
+
+	if (s->cast_to<BoxShape>()) {
+
+		Ref<BoxShape> bs = s;
+		return bs->get_extents();
+	}
+
+	if (s->cast_to<CapsuleShape>()) {
+
+		Ref<CapsuleShape> cs = s;
+		return p_idx==0?cs->get_radius():cs->get_height();
+	}
+
+	if (s->cast_to<RayShape>()) {
+
+		Ref<RayShape> cs = s;
+		return cs->get_length();
+	}
+
+	return Variant();
+}
+void CollisionShapeSpatialGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){
+	Ref<Shape> s = cs->get_shape();
+	if (s.is_null())
+		return;
+
+	Transform gt = cs->get_global_transform();
+	gt.orthonormalize();
+	Transform gi = gt.affine_inverse();
+
+	Vector3 ray_from = p_camera->project_ray_origin(p_point);
+	Vector3 ray_dir = p_camera->project_ray_normal(p_point);
+
+	Vector3 sg[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};
+
+	if (s->cast_to<SphereShape>()) {
+
+		Ref<SphereShape> ss = s;
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(Vector3(),Vector3(4096,0,0),sg[0],sg[1],ra,rb);
+		float d = ra.x;
+		if (d<0.001)
+			d=0.001;
+
+		ss->set_radius(d);
+	}
+
+	if (s->cast_to<RayShape>()) {
+
+		Ref<RayShape> rs = s;
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(Vector3(),Vector3(0,0,4096),sg[0],sg[1],ra,rb);
+		float d = ra.z;
+		if (d<0.001)
+			d=0.001;
+
+		rs->set_length(d);
+	}
+
+
+	if (s->cast_to<BoxShape>()) {
+
+		Vector3 axis;
+		axis[p_idx]=1.0;
+		Ref<BoxShape> bs = s;
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(Vector3(),axis*4096,sg[0],sg[1],ra,rb);
+		float d = ra[p_idx];
+		if (d<0.001)
+			d=0.001;
+
+		Vector3 he = bs->get_extents();
+		he[p_idx]=d;
+		bs->set_extents(he);
+
+	}
+
+	if (s->cast_to<CapsuleShape>()) {
+
+		Vector3 axis;
+		axis[p_idx==0?0:2]=1.0;
+		Ref<CapsuleShape> cs = s;
+		Vector3 ra,rb;
+		Geometry::get_closest_points_between_segments(Vector3(),axis*4096,sg[0],sg[1],ra,rb);
+		float d = axis.dot(ra);
+		if (p_idx==1)
+			d-=cs->get_radius();
+		if (d<0.001)
+			d=0.001;
+
+		if (p_idx==0)
+			cs->set_radius(d);
+		else if (p_idx==1)
+			cs->set_height(d*2.0);
+
+	}
+
+}
+void CollisionShapeSpatialGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){
+	Ref<Shape> s = cs->get_shape();
+	if (s.is_null())
+		return;
+
+	if (s->cast_to<SphereShape>()) {
+
+		Ref<SphereShape> ss=s;
+		if (p_cancel) {
+			ss->set_radius(p_restore);
+			return;
+		}
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		ur->create_action("Change Sphere Shape Radius");
+		ur->add_do_method(ss.ptr(),"set_radius",ss->get_radius());
+		ur->add_undo_method(ss.ptr(),"set_radius",p_restore);
+		ur->commit_action();
+
+	}
+
+	if (s->cast_to<BoxShape>()) {
+
+		Ref<BoxShape> ss=s;
+		if (p_cancel) {
+			ss->set_extents(p_restore);
+			return;
+		}
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		ur->create_action("Change Box Shape Extents");
+		ur->add_do_method(ss.ptr(),"set_extents",ss->get_extents());
+		ur->add_undo_method(ss.ptr(),"set_extents",p_restore);
+		ur->commit_action();
+	}
+
+	if (s->cast_to<CapsuleShape>()) {
+
+		Ref<CapsuleShape> ss=s;
+		if (p_cancel) {
+			if (p_idx==0)
+				ss->set_radius(p_restore);
+			else
+				ss->set_height(p_restore);
+			return;
+		}
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		if (p_idx==0) {
+			ur->create_action("Change Capsule Shape Radius");
+			ur->add_do_method(ss.ptr(),"set_radius",ss->get_radius());
+			ur->add_undo_method(ss.ptr(),"set_radius",p_restore);
+		} else {
+			ur->create_action("Change Capsule Shape Height");
+			ur->add_do_method(ss.ptr(),"set_height",ss->get_height());
+			ur->add_undo_method(ss.ptr(),"set_height",p_restore);
+
+		}
+
+		ur->commit_action();
+
+	}
+
+	if (s->cast_to<RayShape>()) {
+
+		Ref<RayShape> ss=s;
+		if (p_cancel) {
+			ss->set_length(p_restore);
+			return;
+		}
+
+		UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+		ur->create_action("Change Ray Shape Length");
+		ur->add_do_method(ss.ptr(),"set_length",ss->get_length());
+		ur->add_undo_method(ss.ptr(),"set_length",p_restore);
+		ur->commit_action();
+
+	}
+
+}
+void CollisionShapeSpatialGizmo::redraw(){
+
+	clear();
+
+	Ref<Shape> s = cs->get_shape();
+	if (s.is_null())
+		return;
+
+	if (s->cast_to<SphereShape>()) {
+
+		Ref<SphereShape> sp= s;
+		float r=sp->get_radius();
+
+		Vector<Vector3> points;
+
+		for(int i=0;i<=360;i++) {
+
+			float ra=Math::deg2rad(i);
+			float rb=Math::deg2rad(i+1);
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;
+
+			points.push_back(Vector3(a.x,0,a.y));
+			points.push_back(Vector3(b.x,0,b.y));
+			points.push_back(Vector3(0,a.x,a.y));
+			points.push_back(Vector3(0,b.x,b.y));
+			points.push_back(Vector3(a.x,a.y,0));
+			points.push_back(Vector3(b.x,b.y,0));
+
+		}
+
+		Vector<Vector3> collision_segments;
+
+		for(int i=0;i<64;i++) {
+
+			float ra=i*Math_PI*2.0/64.0;
+			float rb=(i+1)*Math_PI*2.0/64.0;
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*r;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*r;
+
+			collision_segments.push_back(Vector3(a.x,0,a.y));
+			collision_segments.push_back(Vector3(b.x,0,b.y));
+			collision_segments.push_back(Vector3(0,a.x,a.y));
+			collision_segments.push_back(Vector3(0,b.x,b.y));
+			collision_segments.push_back(Vector3(a.x,a.y,0));
+			collision_segments.push_back(Vector3(b.x,b.y,0));
+		}
+
+		add_lines(points,SpatialEditorGizmos::singleton->shape_material);
+		add_collision_segments(collision_segments);
+		Vector<Vector3> handles;
+		handles.push_back(Vector3(r,0,0));
+		add_handles(handles);
+
+	}
+
+	if (s->cast_to<BoxShape>()) {
+
+		Ref<BoxShape> bs=s;
+		Vector<Vector3> lines;
+		AABB aabb;
+		aabb.pos=-bs->get_extents();
+		aabb.size=aabb.pos*-2;
+
+		for(int i=0;i<12;i++) {
+			Vector3 a,b;
+			aabb.get_edge(i,a,b);
+			lines.push_back(a);
+			lines.push_back(b);
+		}
+
+		Vector<Vector3> handles;
+
+		for(int i=0;i<3;i++) {
+
+			Vector3 ax;
+			ax[i]=bs->get_extents()[i];
+			handles.push_back(ax);
+		}
+
+		add_lines(lines,SpatialEditorGizmos::singleton->shape_material);
+		add_collision_segments(lines);
+		add_handles(handles);
+
+	}
+
+	if (s->cast_to<CapsuleShape>()) {
+
+		Ref<CapsuleShape> cs=s;
+		float radius = cs->get_radius();
+		float height = cs->get_height();
+
+
+		Vector<Vector3> points;
+
+		Vector3 d(0,0,height*0.5);
+		for(int i=0;i<360;i++) {
+
+			float ra=Math::deg2rad(i);
+			float rb=Math::deg2rad(i+1);
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*radius;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*radius;
+
+			points.push_back(Vector3(a.x,a.y,0)+d);
+			points.push_back(Vector3(b.x,b.y,0)+d);
+
+			points.push_back(Vector3(a.x,a.y,0)-d);
+			points.push_back(Vector3(b.x,b.y,0)-d);
+
+			if (i%90==0) {
+
+				points.push_back(Vector3(a.x,a.y,0)+d);
+				points.push_back(Vector3(a.x,a.y,0)-d);
+			}
+
+			Vector3 dud = i<180?d:-d;
+
+			points.push_back(Vector3(0,a.y,a.x)+dud);
+			points.push_back(Vector3(0,b.y,b.x)+dud);
+			points.push_back(Vector3(a.y,0,a.x)+dud);
+			points.push_back(Vector3(b.y,0,b.x)+dud);
+
+		}
+
+		add_lines(points,SpatialEditorGizmos::singleton->shape_material);
+
+		Vector<Vector3> collision_segments;
+
+		for(int i=0;i<64;i++) {
+
+			float ra=i*Math_PI*2.0/64.0;
+			float rb=(i+1)*Math_PI*2.0/64.0;
+			Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*radius;
+			Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*radius;
+
+			collision_segments.push_back(Vector3(a.x,a.y,0)+d);
+			collision_segments.push_back(Vector3(b.x,b.y,0)+d);
+
+			collision_segments.push_back(Vector3(a.x,a.y,0)-d);
+			collision_segments.push_back(Vector3(b.x,b.y,0)-d);
+
+			if (i%16==0) {
+
+				collision_segments.push_back(Vector3(a.x,a.y,0)+d);
+				collision_segments.push_back(Vector3(a.x,a.y,0)-d);
+			}
+
+			Vector3 dud = i<32?d:-d;
+
+			collision_segments.push_back(Vector3(0,a.y,a.x)+dud);
+			collision_segments.push_back(Vector3(0,b.y,b.x)+dud);
+			collision_segments.push_back(Vector3(a.y,0,a.x)+dud);
+			collision_segments.push_back(Vector3(b.y,0,b.x)+dud);
+
+		}
+
+		add_collision_segments(collision_segments);
+
+		Vector<Vector3> handles;
+		handles.push_back(Vector3(cs->get_radius(),0,0));
+		handles.push_back(Vector3(0,0,cs->get_height()*0.5+cs->get_radius()));
+		add_handles(handles);
+
+
+	}
+
+	if (s->cast_to<PlaneShape>()) {
+
+		Ref<PlaneShape> ps=s;
+		Plane p = ps->get_plane();
+		Vector<Vector3> points;
+
+		Vector3 n1 = p.get_any_perpendicular_normal();
+		Vector3 n2 = p.normal.cross(n1).normalized();
+
+		Vector3 pface[4]={
+			p.normal*p.d+n1*10.0+n2*10.0,
+			p.normal*p.d+n1*10.0+n2*-10.0,
+			p.normal*p.d+n1*-10.0+n2*-10.0,
+			p.normal*p.d+n1*-10.0+n2*10.0,
+		};
+
+		points.push_back(pface[0]);
+		points.push_back(pface[1]);
+		points.push_back(pface[1]);
+		points.push_back(pface[2]);
+		points.push_back(pface[2]);
+		points.push_back(pface[3]);
+		points.push_back(pface[3]);
+		points.push_back(pface[0]);
+		points.push_back(p.normal*p.d);
+		points.push_back(p.normal*p.d+p.normal*3);
+
+		add_lines(points,SpatialEditorGizmos::singleton->shape_material);
+		add_collision_segments(points);
+
+	}
+
+
+	if (s->cast_to<ConvexPolygonShape>()) {
+
+		DVector<Vector3> points = s->cast_to<ConvexPolygonShape>()->get_points();
+
+		if (points.size()>3) {
+
+			QuickHull qh;
+			Vector<Vector3> varr = Variant(points);
+			Geometry::MeshData md;
+			Error err = qh.build(varr,md);
+			if (err==OK) {
+				Vector<Vector3> points;
+				points.resize(md.edges.size()*2);
+				for(int i=0;i<md.edges.size();i++) {
+					points[i*2+0]=md.vertices[md.edges[i].a];
+					points[i*2+1]=md.vertices[md.edges[i].b];
+				}
+
+
+				add_lines(points,SpatialEditorGizmos::singleton->shape_material);
+				add_collision_segments(points);
+
+			}
+		}
+
+	}
+
+
+	if (s->cast_to<RayShape>()) {
+
+		Ref<RayShape> rs=s;
+
+		Vector<Vector3> points;
+		points.push_back(Vector3());
+		points.push_back(Vector3(0,0,rs->get_length()));
+		add_lines(points,SpatialEditorGizmos::singleton->shape_material);
+		add_collision_segments(points);
+		Vector<Vector3> handles;
+		handles.push_back(Vector3(0,0,rs->get_length()));
+		add_handles(handles);
+
+
+	}
+
+}
+CollisionShapeSpatialGizmo::CollisionShapeSpatialGizmo(CollisionShape* p_cs) {
+
+	cs=p_cs;
+	set_spatial_node(p_cs);
+}
+
+
+
+/////
+
+
+void CollisionPolygonSpatialGizmo::redraw() {
+
+	clear();
+
+	Vector<Vector2> points = polygon->get_polygon();
+	float depth = polygon->get_depth()*0.5;
+
+	Vector<Vector3> lines;
+	for(int i=0;i<points.size();i++) {
+
+		int n = (i+1)%points.size();
+		lines.push_back(Vector3(points[i].x,points[i].y,depth));
+		lines.push_back(Vector3(points[n].x,points[n].y,depth));
+		lines.push_back(Vector3(points[i].x,points[i].y,-depth));
+		lines.push_back(Vector3(points[n].x,points[n].y,-depth));
+		lines.push_back(Vector3(points[i].x,points[i].y,depth));
+		lines.push_back(Vector3(points[i].x,points[i].y,-depth));
+
+	}
+
+	add_lines(lines,SpatialEditorGizmos::singleton->shape_material);
+	add_collision_segments(lines);
+}
+
+CollisionPolygonSpatialGizmo::CollisionPolygonSpatialGizmo(CollisionPolygon* p_polygon){
+
+	set_spatial_node(p_polygon);
+	polygon=p_polygon;
+}
+///
+
+
+String VisibilityNotifierGizmo::get_handle_name(int p_idx) const {
+
+	switch(p_idx) {
+		case 0: return "X";
+		case 1: return "Y";
+		case 2: return "Z";
+	}
+
+	return "";
+}
+Variant VisibilityNotifierGizmo::get_handle_value(int p_idx) const{
+
+	return notifier->get_aabb();
+}
+void VisibilityNotifierGizmo::set_handle(int p_idx,Camera *p_camera, const Point2& p_point){
+
+
+	Transform gt = notifier->get_global_transform();
+	//gt.orthonormalize();
+	Transform gi = gt.affine_inverse();
+
+	AABB aabb = notifier->get_aabb();
+	Vector3 ray_from = p_camera->project_ray_origin(p_point);
+	Vector3 ray_dir = p_camera->project_ray_normal(p_point);
+
+	Vector3 sg[2]={gi.xform(ray_from),gi.xform(ray_from+ray_dir*4096)};
+	Vector3 ofs = aabb.pos+aabb.size*0.5;;
+
+	Vector3 axis;
+	axis[p_idx]=1.0;
+
+	Vector3 ra,rb;
+	Geometry::get_closest_points_between_segments(ofs,ofs+axis*4096,sg[0],sg[1],ra,rb);
+	float d = ra[p_idx];
+	if (d<0.001)
+		d=0.001;
+
+	Vector3 he = aabb.size;
+	aabb.pos[p_idx]=(aabb.pos[p_idx]+aabb.size[p_idx]*0.5)-d;
+	aabb.size[p_idx]=d*2;
+	notifier->set_aabb(aabb);
+}
+
+void VisibilityNotifierGizmo::commit_handle(int p_idx,const Variant& p_restore,bool p_cancel){
+
+
+	if (p_cancel) {
+		notifier->set_aabb(p_restore);
+		return;
+	}
+
+	UndoRedo *ur = SpatialEditor::get_singleton()->get_undo_redo();
+	ur->create_action("Change Notifier Extents");
+	ur->add_do_method(notifier,"set_aabb",notifier->get_aabb());
+	ur->add_undo_method(notifier,"set_aabb",p_restore);
+	ur->commit_action();
+
+}
+
+void VisibilityNotifierGizmo::redraw(){
+
+	clear();
+
+	Vector<Vector3> lines;
+	AABB aabb = notifier->get_aabb();
+
+	for(int i=0;i<12;i++) {
+		Vector3 a,b;
+		aabb.get_edge(i,a,b);
+		lines.push_back(a);
+		lines.push_back(b);
+	}
+
+	Vector<Vector3> handles;
+
+
+	for(int i=0;i<3;i++) {
+
+		Vector3 ax;
+		ax[i]=aabb.pos[i]+aabb.size[i];
+		handles.push_back(ax);
+	}
+
+	add_lines(lines,SpatialEditorGizmos::singleton->visibility_notifier_material);
+	//add_unscaled_billboard(SpatialEditorGizmos::singleton->visi,0.05);
+	add_collision_segments(lines);
+	add_handles(handles);
+
+}
+VisibilityNotifierGizmo::VisibilityNotifierGizmo(VisibilityNotifier* p_notifier){
+
+	notifier=p_notifier;
+	set_spatial_node(p_notifier);
+}
+
+////////
+
+
+
+void NavigationMeshSpatialGizmo::redraw() {
+
+	clear();
+	Ref<NavigationMesh> navmeshie = navmesh->get_navigation_mesh();
+	if (navmeshie.is_null())
+		return;
+
+	DVector<Vector3> vertices = navmeshie->get_vertices();
+	DVector<Vector3>::Read vr=vertices.read();
+	List<Face3> faces;
+	for(int i=0;i<navmeshie->get_polygon_count();i++) {
+		Vector<int> p = navmeshie->get_polygon(i);
+
+		for(int j=2;j<p.size();j++) {
+			Face3 f;
+			f.vertex[0]=vr[p[0]];
+			f.vertex[1]=vr[p[j-1]];
+			f.vertex[2]=vr[p[j]];
+
+			faces.push_back(f);
+		}
+	}
+
+
+	Map<_EdgeKey,bool> edge_map;
+	DVector<Vector3> tmeshfaces;
+	tmeshfaces.resize(faces.size()*3);
+
+	{
+		DVector<Vector3>::Write tw=tmeshfaces.write();
+		int tidx=0;
+
+
+		for(List<Face3>::Element *E=faces.front();E;E=E->next()) {
+
+			const Face3 &f = E->get();
+
+			for(int j=0;j<3;j++) {
+
+				tw[tidx++]=f.vertex[j];
+				_EdgeKey ek;
+				ek.from=f.vertex[j].snapped(CMP_EPSILON);
+				ek.to=f.vertex[(j+1)%3].snapped(CMP_EPSILON);
+				if (ek.from<ek.to)
+					SWAP(ek.from,ek.to);
+
+				Map<_EdgeKey,bool>::Element *E=edge_map.find(ek);
+
+				if (E) {
+
+					E->get()=false;
+
+				} else {
+
+					edge_map[ek]=true;
+				}
+
+			}
+		}
+	}
+	Vector<Vector3> lines;
+
+	for(Map<_EdgeKey,bool>::Element *E=edge_map.front();E;E=E->next()) {
+
+		if (E->get()) {
+			lines.push_back(E->key().from);
+			lines.push_back(E->key().to);
+		}
+	}
+
+	Ref<TriangleMesh> tmesh = memnew( TriangleMesh);
+	tmesh->create(tmeshfaces);
+
+	if (lines.size())
+		add_lines(lines,navmesh->is_enabled()?SpatialEditorGizmos::singleton->navmesh_edge_material:SpatialEditorGizmos::singleton->navmesh_edge_material_disabled);
+	add_collision_triangles(tmesh);
+	Ref<Mesh> m = memnew( Mesh );
+	Array a;
+	a.resize(Mesh::ARRAY_MAX);
+	a[0]=tmeshfaces;
+	m->add_surface(Mesh::PRIMITIVE_TRIANGLES,a);
+	m->surface_set_material(0,navmesh->is_enabled()?SpatialEditorGizmos::singleton->navmesh_solid_material:SpatialEditorGizmos::singleton->navmesh_solid_material_disabled);
+	add_mesh(m);
+	add_collision_segments(lines);
+
+}
+
+NavigationMeshSpatialGizmo::NavigationMeshSpatialGizmo(NavigationMeshInstance *p_navmesh){
+
+	set_spatial_node(p_navmesh);
+	navmesh=p_navmesh;
+}
+
+//////
+///
+///
+
+
+
+void PinJointSpatialGizmo::redraw() {
+
+	clear();
+	Vector<Vector3> cursor_points;
+	float cs = 0.25;
+	cursor_points.push_back(Vector3(+cs,0,0));
+	cursor_points.push_back(Vector3(-cs,0,0));
+	cursor_points.push_back(Vector3(0,+cs,0));
+	cursor_points.push_back(Vector3(0,-cs,0));
+	cursor_points.push_back(Vector3(0,0,+cs));
+	cursor_points.push_back(Vector3(0,0,-cs));
+	add_collision_segments(cursor_points);
+	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);
+
+}
+
+
+PinJointSpatialGizmo::PinJointSpatialGizmo(PinJoint* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+////
+
+void HingeJointSpatialGizmo::redraw() {
+
+	clear();
+	Vector<Vector3> cursor_points;
+	float cs = 0.25;
+	/*cursor_points.push_back(Vector3(+cs,0,0));
+	cursor_points.push_back(Vector3(-cs,0,0));
+	cursor_points.push_back(Vector3(0,+cs,0));
+	cursor_points.push_back(Vector3(0,-cs,0));*/
+	cursor_points.push_back(Vector3(0,0,+cs*2));
+	cursor_points.push_back(Vector3(0,0,-cs*2));
+
+	float ll = p3d->get_param(HingeJoint::PARAM_LIMIT_LOWER);
+	float ul = p3d->get_param(HingeJoint::PARAM_LIMIT_UPPER);
+
+	if (p3d->get_flag(HingeJoint::FLAG_USE_LIMIT) && ll<ul) {
+
+		const int points = 32;
+		float step = (ul-ll)/points;
+
+
+		for(int i=0;i<points;i++) {
+
+			float s = ll+i*(ul-ll)/points;
+			float n = ll+(i+1)*(ul-ll)/points;
+
+			Vector3 from=Vector3( -Math::sin(s),Math::cos(s), 0 )*cs;
+			Vector3 to=Vector3( -Math::sin(n),Math::cos(n), 0 )*cs;
+
+			if (i==points-1) {
+				cursor_points.push_back(to);
+				cursor_points.push_back(Vector3());
+			}
+			if (i==0) {
+				cursor_points.push_back(from);
+				cursor_points.push_back(Vector3());
+			}
+
+
+			cursor_points.push_back(from);
+			cursor_points.push_back(to);
+
+
+		}
+
+		cursor_points.push_back(Vector3(0,cs*1.5,0));
+		cursor_points.push_back(Vector3());
+
+	} else {
+
+
+		const int points = 32;
+
+		for(int i=0;i<points;i++) {
+
+			float s = ll+i*(Math_PI*2.0)/points;
+			float n = ll+(i+1)*(Math_PI*2.0)/points;
+
+			Vector3 from=Vector3( -Math::sin(s),Math::cos(s), 0 )*cs;
+			Vector3 to=Vector3( -Math::sin(n),Math::cos(n), 0 )*cs;
+
+			cursor_points.push_back(from);
+			cursor_points.push_back(to);
+
+		}
+
+	}
+	add_collision_segments(cursor_points);
+	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);
+
+}
+
+
+HingeJointSpatialGizmo::HingeJointSpatialGizmo(HingeJoint* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+///////
+///
+////
+
+void SliderJointSpatialGizmo::redraw() {
+
+	clear();
+	Vector<Vector3> cursor_points;
+	float cs = 0.25;
+	/*cursor_points.push_back(Vector3(+cs,0,0));
+	cursor_points.push_back(Vector3(-cs,0,0));
+	cursor_points.push_back(Vector3(0,+cs,0));
+	cursor_points.push_back(Vector3(0,-cs,0));*/
+	cursor_points.push_back(Vector3(0,0,+cs*2));
+	cursor_points.push_back(Vector3(0,0,-cs*2));
+
+	float ll = p3d->get_param(SliderJoint::PARAM_ANGULAR_LIMIT_LOWER);
+	float ul = p3d->get_param(SliderJoint::PARAM_ANGULAR_LIMIT_UPPER);
+	float lll = -p3d->get_param(SliderJoint::PARAM_LINEAR_LIMIT_LOWER);
+	float lul = -p3d->get_param(SliderJoint::PARAM_LINEAR_LIMIT_UPPER);
+
+	if (lll>lul) {
+
+		cursor_points.push_back(Vector3(lul,0,0));
+		cursor_points.push_back(Vector3(lll,0,0));
+
+		cursor_points.push_back(Vector3(lul,-cs,-cs));
+		cursor_points.push_back(Vector3(lul,-cs,cs));
+		cursor_points.push_back(Vector3(lul,-cs,cs));
+		cursor_points.push_back(Vector3(lul,cs,cs));
+		cursor_points.push_back(Vector3(lul,cs,cs));
+		cursor_points.push_back(Vector3(lul,cs,-cs));
+		cursor_points.push_back(Vector3(lul,cs,-cs));
+		cursor_points.push_back(Vector3(lul,-cs,-cs));
+
+
+		cursor_points.push_back(Vector3(lll,-cs,-cs));
+		cursor_points.push_back(Vector3(lll,-cs,cs));
+		cursor_points.push_back(Vector3(lll,-cs,cs));
+		cursor_points.push_back(Vector3(lll,cs,cs));
+		cursor_points.push_back(Vector3(lll,cs,cs));
+		cursor_points.push_back(Vector3(lll,cs,-cs));
+		cursor_points.push_back(Vector3(lll,cs,-cs));
+		cursor_points.push_back(Vector3(lll,-cs,-cs));
+
+
+	} else {
+
+		cursor_points.push_back(Vector3(+cs*2,0,0));
+		cursor_points.push_back(Vector3(-cs*2,0,0));
+
+	}
+
+	if (ll<ul) {
+
+		const int points = 32;
+		float step = (ul-ll)/points;
+
+
+		for(int i=0;i<points;i++) {
+
+			float s = ll+i*(ul-ll)/points;
+			float n = ll+(i+1)*(ul-ll)/points;
+
+			Vector3 from=Vector3(0, Math::cos(s), -Math::sin(s) )*cs;
+			Vector3 to=Vector3(0,Math::cos(n),  -Math::sin(n) )*cs;
+
+			if (i==points-1) {
+				cursor_points.push_back(to);
+				cursor_points.push_back(Vector3());
+			}
+			if (i==0) {
+				cursor_points.push_back(from);
+				cursor_points.push_back(Vector3());
+			}
+
+
+			cursor_points.push_back(from);
+			cursor_points.push_back(to);
+
+
+		}
+
+		cursor_points.push_back(Vector3(0,cs*1.5,0));
+		cursor_points.push_back(Vector3());
+
+	} else {
+
+
+		const int points = 32;
+
+		for(int i=0;i<points;i++) {
+
+			float s = ll+i*(Math_PI*2.0)/points;
+			float n = ll+(i+1)*(Math_PI*2.0)/points;
+
+			Vector3 from=Vector3(0,Math::cos(s),-Math::sin(s) )*cs;
+			Vector3 to=Vector3( 0,Math::cos(n),-Math::sin(n) )*cs;
+
+			cursor_points.push_back(from);
+			cursor_points.push_back(to);
+
+		}
+
+	}
+	add_collision_segments(cursor_points);
+	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);
+
+}
+
+
+SliderJointSpatialGizmo::SliderJointSpatialGizmo(SliderJoint* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+///////
+///
+////
+
+void ConeTwistJointSpatialGizmo::redraw() {
+
+	clear();
+	float cs = 0.25;
+	Vector<Vector3> points;
+
+	float r = 1.0;
+	float w = r*Math::sin(p3d->get_param(ConeTwistJoint::PARAM_SWING_SPAN));
+	float d = r*Math::cos(p3d->get_param(ConeTwistJoint::PARAM_SWING_SPAN));
+
+
+	//swing
+	for(int i=0;i<360;i+=10) {
+
+		float ra=Math::deg2rad(i);
+		float rb=Math::deg2rad(i+10);
+		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w;
+		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w;
+
+		/*points.push_back(Vector3(a.x,0,a.y));
+		points.push_back(Vector3(b.x,0,b.y));
+		points.push_back(Vector3(0,a.x,a.y));
+		points.push_back(Vector3(0,b.x,b.y));*/
+		points.push_back(Vector3(d,a.x,a.y));
+		points.push_back(Vector3(d,b.x,b.y));
+
+		if (i%90==0) {
+
+			points.push_back(Vector3(d,a.x,a.y));
+			points.push_back(Vector3());
+
+		}
+	}
+
+	points.push_back(Vector3());
+	points.push_back(Vector3(1,0,0));
+
+	//twist
+	/*
+	 */
+	float ts=Math::rad2deg(p3d->get_param(ConeTwistJoint::PARAM_TWIST_SPAN));
+	ts=MIN(ts,720);
+
+
+	for(int i=0;i<int(ts);i+=5) {
+
+		float ra=Math::deg2rad(i);
+		float rb=Math::deg2rad(i+5);
+		float c = i/720.0;
+		float cn = (i+5)/720.0;
+		Point2 a = Vector2(Math::sin(ra),Math::cos(ra))*w*c;
+		Point2 b = Vector2(Math::sin(rb),Math::cos(rb))*w*cn;
+
+		/*points.push_back(Vector3(a.x,0,a.y));
+		points.push_back(Vector3(b.x,0,b.y));
+		points.push_back(Vector3(0,a.x,a.y));
+		points.push_back(Vector3(0,b.x,b.y));*/
+
+		points.push_back(Vector3(c,a.x,a.y));
+		points.push_back(Vector3(cn,b.x,b.y));
+
+	}
+
+
+	add_collision_segments(points);
+	add_lines(points,SpatialEditorGizmos::singleton->joint_material);
+
+}
+
+
+ConeTwistJointSpatialGizmo::ConeTwistJointSpatialGizmo(ConeTwistJoint* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+////////
+/// \brief SpatialEditorGizmos::singleton
+///
+///////
+///
+////
+
+void Generic6DOFJointSpatialGizmo::redraw() {
+
+	clear();
+	Vector<Vector3> cursor_points;
+	float cs = 0.25;
+
+	for(int ax=0;ax<3;ax++) {
+		/*cursor_points.push_back(Vector3(+cs,0,0));
+		cursor_points.push_back(Vector3(-cs,0,0));
+		cursor_points.push_back(Vector3(0,+cs,0));
+		cursor_points.push_back(Vector3(0,-cs,0));
+		cursor_points.push_back(Vector3(0,0,+cs*2));
+		cursor_points.push_back(Vector3(0,0,-cs*2)); */
+
+		float ll;
+		float ul;
+		float lll;
+		float lul;
+
+		int a1,a2,a3;
+		bool enable_ang;
+		bool enable_lin;
+
+		switch(ax) {
+			case 0:
+				ll = p3d->get_param_x(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);
+				ul = p3d->get_param_x(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);
+				lll = -p3d->get_param_x(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);
+				lul = -p3d->get_param_x(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);
+				enable_ang = p3d->get_flag_x(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);
+				enable_lin = p3d->get_flag_x(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);
+				a1=0;
+				a2=1;
+				a3=2;
+			break;
+			case 1:
+				ll = p3d->get_param_y(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);
+				ul = p3d->get_param_y(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);
+				lll = -p3d->get_param_y(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);
+				lul = -p3d->get_param_y(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);
+				enable_ang = p3d->get_flag_y(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);
+				enable_lin = p3d->get_flag_y(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);
+				a1=2;
+				a2=0;
+				a3=1;
+			break;
+			case 2:
+				ll = p3d->get_param_z(Generic6DOFJoint::PARAM_ANGULAR_LOWER_LIMIT);
+				ul = p3d->get_param_z(Generic6DOFJoint::PARAM_ANGULAR_UPPER_LIMIT);
+				lll = -p3d->get_param_z(Generic6DOFJoint::PARAM_LINEAR_LOWER_LIMIT);
+				lul = -p3d->get_param_z(Generic6DOFJoint::PARAM_LINEAR_UPPER_LIMIT);
+				enable_ang = p3d->get_flag_z(Generic6DOFJoint::FLAG_ENABLE_ANGULAR_LIMIT);
+				enable_lin = p3d->get_flag_z(Generic6DOFJoint::FLAG_ENABLE_LINEAR_LIMIT);
+
+				a1=1;
+				a2=2;
+				a3=0;
+			break;
+		}
+
+#define ADD_VTX(x,y,z)\
+		{\
+			Vector3 v;\
+			v[a1]=(x);\
+			v[a2]=(y);\
+			v[a3]=(z);\
+			cursor_points.push_back(v);\
+		}
+
+#define SET_VTX(what,x,y,z)\
+		{\
+			Vector3 v;\
+			v[a1]=(x);\
+			v[a2]=(y);\
+			v[a3]=(z);\
+			what=v;\
+		}
+
+
+
+
+		if (enable_lin && lll>=lul) {
+
+			ADD_VTX(lul,0,0);
+			ADD_VTX(lll,0,0);
+
+			ADD_VTX(lul,-cs,-cs);
+			ADD_VTX(lul,-cs,cs);
+			ADD_VTX(lul,-cs,cs);
+			ADD_VTX(lul,cs,cs);
+			ADD_VTX(lul,cs,cs);
+			ADD_VTX(lul,cs,-cs);
+			ADD_VTX(lul,cs,-cs);
+			ADD_VTX(lul,-cs,-cs);
+
+
+			ADD_VTX(lll,-cs,-cs);
+			ADD_VTX(lll,-cs,cs);
+			ADD_VTX(lll,-cs,cs);
+			ADD_VTX(lll,cs,cs);
+			ADD_VTX(lll,cs,cs);
+			ADD_VTX(lll,cs,-cs);
+			ADD_VTX(lll,cs,-cs);
+			ADD_VTX(lll,-cs,-cs);
+
+
+		} else {
+
+			ADD_VTX(+cs*2,0,0);
+			ADD_VTX(-cs*2,0,0);
+
+		}
+
+		if (enable_ang && ll<=ul) {
+
+			const int points = 32;
+			float step = (ul-ll)/points;
+
+
+			for(int i=0;i<points;i++) {
+
+				float s = ll+i*(ul-ll)/points;
+				float n = ll+(i+1)*(ul-ll)/points;
+
+				Vector3 from;
+				SET_VTX(from,0, Math::cos(s), -Math::sin(s) );
+				from*=cs;
+				Vector3 to;
+				SET_VTX(to,0,Math::cos(n),  -Math::sin(n));
+				to*=cs;
+
+				if (i==points-1) {
+					cursor_points.push_back(to);
+					cursor_points.push_back(Vector3());
+				}
+				if (i==0) {
+					cursor_points.push_back(from);
+					cursor_points.push_back(Vector3());
+				}
+
+
+				cursor_points.push_back(from);
+				cursor_points.push_back(to);
+
+
+			}
+
+			ADD_VTX(0,cs*1.5,0);
+			cursor_points.push_back(Vector3());
+
+		} else {
+
+
+			const int points = 32;
+
+			for(int i=0;i<points;i++) {
+
+				float s = ll+i*(Math_PI*2.0)/points;
+				float n = ll+(i+1)*(Math_PI*2.0)/points;
+
+//				Vector3 from=Vector3(0,Math::cos(s),-Math::sin(s) )*cs;
+//				Vector3 to=Vector3( 0,Math::cos(n),-Math::sin(n) )*cs;
+
+				Vector3 from;
+				SET_VTX(from,0, Math::cos(s), -Math::sin(s) );
+				from*=cs;
+				Vector3 to;
+				SET_VTX(to,0,Math::cos(n),  -Math::sin(n));
+				to*=cs;
+
+				cursor_points.push_back(from);
+				cursor_points.push_back(to);
+
+			}
+
+		}
+	}
+
+#undef ADD_VTX
+#undef SET_VTX
+
+	add_collision_segments(cursor_points);
+	add_lines(cursor_points,SpatialEditorGizmos::singleton->joint_material);
+
+}
+
+
+Generic6DOFJointSpatialGizmo::Generic6DOFJointSpatialGizmo(Generic6DOFJoint* p_p3d) {
+
+	p3d=p_p3d;
+	set_spatial_node(p3d);
+}
+
+///////
+///
+////
+
+
+SpatialEditorGizmos *SpatialEditorGizmos::singleton=NULL;
+
+Ref<SpatialEditorGizmo> SpatialEditorGizmos::get_gizmo(Spatial *p_spatial) {
+
+	if (p_spatial->cast_to<Light>()) {
+
+		Ref<LightSpatialGizmo> lsg = memnew( LightSpatialGizmo(p_spatial->cast_to<Light>()) );
+		return lsg;
+	}
+
+	if (p_spatial->cast_to<Camera>()) {
+
+		Ref<CameraSpatialGizmo> lsg = memnew( CameraSpatialGizmo(p_spatial->cast_to<Camera>()) );
+		return lsg;
+	}
+
+	if (p_spatial->cast_to<Skeleton>()) {
+
+		Ref<SkeletonSpatialGizmo> lsg = memnew( SkeletonSpatialGizmo(p_spatial->cast_to<Skeleton>()) );
+		return lsg;
+	}
+
+
+	if (p_spatial->cast_to<Position3D>()) {
+
+		Ref<Position3DSpatialGizmo> lsg = memnew( Position3DSpatialGizmo(p_spatial->cast_to<Position3D>()) );
+		return lsg;
+	}
+
+	if (p_spatial->cast_to<MeshInstance>()) {
+
+		Ref<MeshInstanceSpatialGizmo> misg = memnew( MeshInstanceSpatialGizmo(p_spatial->cast_to<MeshInstance>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<Room>()) {
+
+		Ref<RoomSpatialGizmo> misg = memnew( RoomSpatialGizmo(p_spatial->cast_to<Room>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<NavigationMeshInstance>()) {
+
+		Ref<NavigationMeshSpatialGizmo> misg = memnew( NavigationMeshSpatialGizmo(p_spatial->cast_to<NavigationMeshInstance>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<RayCast>()) {
+
+		Ref<RayCastSpatialGizmo> misg = memnew( RayCastSpatialGizmo(p_spatial->cast_to<RayCast>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<Portal>()) {
+
+		Ref<PortalSpatialGizmo> misg = memnew( PortalSpatialGizmo(p_spatial->cast_to<Portal>()) );
+		return misg;
+	}
+
+
+	if (p_spatial->cast_to<TestCube>()) {
+
+		Ref<TestCubeSpatialGizmo> misg = memnew( TestCubeSpatialGizmo(p_spatial->cast_to<TestCube>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<SpatialPlayer>()) {
+
+		Ref<SpatialPlayerSpatialGizmo> misg = memnew( SpatialPlayerSpatialGizmo(p_spatial->cast_to<SpatialPlayer>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<CollisionShape>()) {
+
+		Ref<CollisionShapeSpatialGizmo> misg = memnew( CollisionShapeSpatialGizmo(p_spatial->cast_to<CollisionShape>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<VisibilityNotifier>()) {
+
+		Ref<VisibilityNotifierGizmo> misg = memnew( VisibilityNotifierGizmo(p_spatial->cast_to<VisibilityNotifier>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<VehicleWheel>()) {
+
+		Ref<VehicleWheelSpatialGizmo> misg = memnew( VehicleWheelSpatialGizmo(p_spatial->cast_to<VehicleWheel>()) );
+		return misg;
+	}
+	if (p_spatial->cast_to<PinJoint>()) {
+
+		Ref<PinJointSpatialGizmo> misg = memnew( PinJointSpatialGizmo(p_spatial->cast_to<PinJoint>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<HingeJoint>()) {
+
+		Ref<HingeJointSpatialGizmo> misg = memnew( HingeJointSpatialGizmo(p_spatial->cast_to<HingeJoint>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<SliderJoint>()) {
+
+		Ref<SliderJointSpatialGizmo> misg = memnew( SliderJointSpatialGizmo(p_spatial->cast_to<SliderJoint>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<ConeTwistJoint>()) {
+
+		Ref<ConeTwistJointSpatialGizmo> misg = memnew( ConeTwistJointSpatialGizmo(p_spatial->cast_to<ConeTwistJoint>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<Generic6DOFJoint>()) {
+
+		Ref<Generic6DOFJointSpatialGizmo> misg = memnew( Generic6DOFJointSpatialGizmo(p_spatial->cast_to<Generic6DOFJoint>()) );
+		return misg;
+	}
+
+	if (p_spatial->cast_to<CollisionPolygon>()) {
+
+		Ref<CollisionPolygonSpatialGizmo> misg = memnew( CollisionPolygonSpatialGizmo(p_spatial->cast_to<CollisionPolygon>()) );
+		return misg;
+	}
+
+
+	return Ref<SpatialEditorGizmo>();
+}
+
+
+Ref<FixedMaterial> SpatialEditorGizmos::create_line_material(const Color& p_base_color) {
+
+	Ref<FixedMaterial> line_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	line_material->set_flag(Material::FLAG_UNSHADED, true);
+	line_material->set_line_width(3.0);
+	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, true);
+	line_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,p_base_color);
+
+	return line_material;
+
+}
+
+Ref<FixedMaterial> SpatialEditorGizmos::create_solid_material(const Color& p_base_color) {
+
+	Ref<FixedMaterial> line_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	line_material->set_flag(Material::FLAG_UNSHADED, true);
+	line_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	line_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,p_base_color);
+
+	return line_material;
+
+}
+
+SpatialEditorGizmos::SpatialEditorGizmos() {
+
+	singleton=this;
+
+	handle_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	handle_material->set_flag(Material::FLAG_UNSHADED, true);
+	handle_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(0.8,0.8,0.8));
+
+	handle2_material = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	handle2_material->set_flag(Material::FLAG_UNSHADED, true);
+	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_POINT_SIZE, true);
+	handle_t = SpatialEditor::get_singleton()->get_icon("Editor3DHandle","EditorIcons");
+	handle2_material->set_point_size(handle_t->get_width());
+	handle2_material->set_texture(FixedMaterial::PARAM_DIFFUSE,handle_t);
+	handle2_material->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1));
+	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	handle2_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, true);
+
+	light_material = create_line_material(Color(1,1,0.2));
+
+	light_material_omni_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	light_material_omni_icon->set_flag(Material::FLAG_UNSHADED, true);
+	light_material_omni_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);
+	light_material_omni_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+	light_material_omni_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	light_material_omni_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));
+	light_material_omni_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoLight","EditorIcons"));
+
+
+	light_material_directional_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	light_material_directional_icon->set_flag(Material::FLAG_UNSHADED, true);
+	light_material_directional_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);
+	light_material_directional_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+	light_material_directional_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	light_material_directional_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));
+	light_material_directional_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoDirectionalLight","EditorIcons"));
+
+	camera_material = create_line_material(Color(1.0,0.5,1.0));
+
+
+	navmesh_edge_material = create_line_material(Color(0.1,0.8,1.0));
+	navmesh_solid_material = create_solid_material(Color(0.1,0.8,1.0,0.4));
+	navmesh_edge_material->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, false);
+	navmesh_solid_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);
+
+	navmesh_edge_material_disabled = create_line_material(Color(1.0,0.8,0.1));
+	navmesh_solid_material_disabled = create_solid_material(Color(1.0,0.8,0.1,0.4));
+	navmesh_edge_material_disabled->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY, false);
+	navmesh_solid_material_disabled->set_flag(Material::FLAG_DOUBLE_SIDED,true);
+
+	skeleton_material = create_line_material(Color(0.6,1.0,0.3));
+	skeleton_material->set_flag(Material::FLAG_DOUBLE_SIDED,true);
+	skeleton_material->set_flag(Material::FLAG_UNSHADED,true);
+	skeleton_material->set_flag(Material::FLAG_ONTOP,true);
+	skeleton_material->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+
+	//position 3D Shared mesh
+
+	pos3d_mesh = Ref<Mesh>( memnew( Mesh ) );
+	{
+
+		DVector<Vector3> cursor_points;
+		DVector<Color> cursor_colors;
+		float cs = 0.25;
+		cursor_points.push_back(Vector3(+cs,0,0));
+		cursor_points.push_back(Vector3(-cs,0,0));
+		cursor_points.push_back(Vector3(0,+cs,0));
+		cursor_points.push_back(Vector3(0,-cs,0));
+		cursor_points.push_back(Vector3(0,0,+cs));
+		cursor_points.push_back(Vector3(0,0,-cs));
+		cursor_colors.push_back(Color(1,0.5,0.5,0.7));
+		cursor_colors.push_back(Color(1,0.5,0.5,0.7));
+		cursor_colors.push_back(Color(0.5,1,0.5,0.7));
+		cursor_colors.push_back(Color(0.5,1,0.5,0.7));
+		cursor_colors.push_back(Color(0.5,0.5,1,0.7));
+		cursor_colors.push_back(Color(0.5,0.5,1,0.7));
+
+		Ref<FixedMaterial> mat = memnew( FixedMaterial );
+		mat->set_flag(Material::FLAG_UNSHADED,true);
+		mat->set_fixed_flag(FixedMaterial::FLAG_USE_COLOR_ARRAY,true);
+		mat->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA,true);
+		mat->set_line_width(3);
+		Array d;
+		d.resize(VS::ARRAY_MAX);
+		d[Mesh::ARRAY_VERTEX]=cursor_points;
+		d[Mesh::ARRAY_COLOR]=cursor_colors;
+		pos3d_mesh->add_surface(Mesh::PRIMITIVE_LINES,d);
+		pos3d_mesh->surface_set_material(0,mat);
+	}
+
+
+	sample_player_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	sample_player_icon->set_flag(Material::FLAG_UNSHADED, true);
+	sample_player_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);
+	sample_player_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+	sample_player_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	sample_player_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));
+	sample_player_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoSpatialSamplePlayer","EditorIcons"));
+
+	room_material = create_line_material(Color(1.0,0.6,0.9));
+	portal_material = create_line_material(Color(1.0,0.8,0.6));
+	raycast_material = create_line_material(Color(1.0,0.8,0.6));
+	car_wheel_material = create_line_material(Color(0.6,0.8,1.0));
+	visibility_notifier_material = create_line_material(Color(1.0,0.5,1.0));
+	joint_material = create_line_material(Color(0.6,0.8,1.0));
+
+	stream_player_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	stream_player_icon->set_flag(Material::FLAG_UNSHADED, true);
+	stream_player_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);
+	stream_player_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+	stream_player_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	stream_player_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));
+	stream_player_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("GizmoSpatialStreamPlayer","EditorIcons"));
+
+	visibility_notifier_icon = Ref<FixedMaterial>( memnew( FixedMaterial ));
+	visibility_notifier_icon->set_flag(Material::FLAG_UNSHADED, true);
+	visibility_notifier_icon->set_flag(Material::FLAG_DOUBLE_SIDED, true);
+	visibility_notifier_icon->set_depth_draw_mode(Material::DEPTH_DRAW_NEVER);
+	visibility_notifier_icon->set_fixed_flag(FixedMaterial::FLAG_USE_ALPHA, true);
+	visibility_notifier_icon->set_parameter(FixedMaterial::PARAM_DIFFUSE,Color(1,1,1,0.9));
+	visibility_notifier_icon->set_texture(FixedMaterial::PARAM_DIFFUSE,SpatialEditor::get_singleton()->get_icon("Visible","EditorIcons"));
+
+	{
+
+		DVector<Vector3> vertices;
+
+#undef ADD_VTX
+#define ADD_VTX(m_idx);\
+	vertices.push_back( face_points[m_idx] );
+
+		for (int i=0;i<6;i++) {
+
+
+			Vector3 face_points[4];
+
+			for (int j=0;j<4;j++) {
+
+				float v[3];
+				v[0]=1.0;
+				v[1]=1-2*((j>>1)&1);
+				v[2]=v[1]*(1-2*(j&1));
+
+				for (int k=0;k<3;k++) {
+
+					if (i<3)
+						face_points[j][(i+k)%3]=v[k]*(i>=3?-1:1);
+					else
+						face_points[3-j][(i+k)%3]=v[k]*(i>=3?-1:1);
+				}
+			}
+			//tri 1
+			ADD_VTX(0);
+			ADD_VTX(1);
+			ADD_VTX(2);
+			//tri 2
+			ADD_VTX(2);
+			ADD_VTX(3);
+			ADD_VTX(0);
+
+		}
+
+		test_cube_tm = Ref<TriangleMesh>( memnew( TriangleMesh ) );
+		test_cube_tm->create(vertices);
+	}
+
+	shape_material = create_line_material(Color(0.2,1,1.0));
+
+
+}
+
diff --git a/tools/editor/spatial_editor_gizmos.h b/tools/editor/spatial_editor_gizmos.h
index 02f60db7f8..bc7e8ad21d 100644
--- a/tools/editor/spatial_editor_gizmos.h
+++ b/tools/editor/spatial_editor_gizmos.h
@@ -1,491 +1,491 @@
-/*************************************************************************/

-/*  spatial_editor_gizmos.h                                              */

-/*************************************************************************/

-/*                       This file is part of:                           */

-/*                           GODOT ENGINE                                */

-/*                    http://www.godotengine.org                         */

-/*************************************************************************/

-/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */

-/*                                                                       */

-/* Permission is hereby granted, free of charge, to any person obtaining */

-/* a copy of this software and associated documentation files (the       */

-/* "Software"), to deal in the Software without restriction, including   */

-/* without limitation the rights to use, copy, modify, merge, publish,   */

-/* distribute, sublicense, and/or sell copies of the Software, and to    */

-/* permit persons to whom the Software is furnished to do so, subject to */

-/* the following conditions:                                             */

-/*                                                                       */

-/* The above copyright notice and this permission notice shall be        */

-/* included in all copies or substantial portions of the Software.       */

-/*                                                                       */

-/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */

-/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */

-/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/

-/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */

-/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */

-/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */

-/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */

-/*************************************************************************/

-#ifndef SPATIAL_EDITOR_GIZMOS_H

-#define SPATIAL_EDITOR_GIZMOS_H

-

-

-#include "tools/editor/plugins/spatial_editor_plugin.h"

-#include "scene/3d/light.h"

-#include "scene/3d/camera.h"

-#include "scene/3d/position_3d.h"

-#include "scene/3d/spatial_sample_player.h"

-#include "scene/3d/spatial_stream_player.h"

-#include "scene/3d/test_cube.h"

-#include "scene/3d/mesh_instance.h"

-#include "scene/3d/body_shape.h"

-#include "scene/3d/room_instance.h"

-#include "scene/3d/visibility_notifier.h"

-#include "scene/3d/portal.h"

-#include "scene/3d/ray_cast.h"

-#include "scene/3d/navigation_mesh.h"

-

-#include "scene/3d/vehicle_body.h"

-#include "scene/3d/collision_polygon.h"

-#include "scene/3d/physics_joint.h"

-

-

-class Camera;

-

-class SpatialGizmoTool  : public SpatialEditorGizmo {

-

-	OBJ_TYPE(SpatialGizmoTool,SpatialGizmo);

-

-	struct Instance{

-

-		RID instance;

-		Ref<Mesh> mesh;

-		RID skeleton;

-		bool billboard;

-		bool unscaled;

-		bool can_intersect;

-		bool extra_margin;

-		Instance() {

-

-			billboard=false;

-			unscaled=false;

-			can_intersect=false;

-			extra_margin=false;

-		}

-

-		void create_instance(Spatial *p_base);

-

-	};

-

-	Vector<Vector3> collision_segments;

-	Ref<TriangleMesh> collision_mesh;

-

-	struct Handle {

-		Vector3 pos;

-		bool billboard;

-	};

-

-	Vector<Vector3> handles;

-	Vector<Vector3> secondary_handles;

-	bool billboard_handle;

-

-	bool valid;

-	Spatial *base;

-	Vector<Instance> instances;

-	Spatial *spatial_node;

-protected:

-	void add_lines(const Vector<Vector3> &p_lines,const Ref<Material>& p_material,bool p_billboard=false);

-	void add_mesh(const Ref<Mesh>& p_mesh,bool p_billboard=false,const RID& p_skeleton=RID());

-	void add_collision_segments(const Vector<Vector3> &p_lines);

-	void add_collision_triangles(const Ref<TriangleMesh>& p_tmesh);

-	void add_unscaled_billboard(const Ref<Material>& p_material,float p_scale=1);

-	void add_handles(const Vector<Vector3> &p_handles,bool p_billboard=false,bool p_secondary=false);

-

-	void set_spatial_node(Spatial *p_node);

-

-public:

-

-	virtual Vector3 get_handle_pos(int p_idx) const;

-	virtual bool intersect_frustum(const Camera *p_camera,const Vector<Plane> &p_frustum);

-	virtual bool intersect_ray(const Camera *p_camera,const Point2& p_point,  Vector3& r_pos, Vector3& r_normal,int *r_gizmo_handle=NULL,bool p_sec_first=false);

-

-	void clear();

-	void create();

-	void transform();

-	//void redraw();

-	void free();

-

-	SpatialGizmoTool();

-	~SpatialGizmoTool();

-};

-

-

-

-class LightSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(LightSpatialGizmo,SpatialGizmoTool);

-

-	Light* light;

-

-public:

-

-

-	virtual String get_handle_name(int p_idx) const;

-	virtual Variant get_handle_value(int p_idx) const;

-	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);

-	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);

-

-	void redraw();

-	LightSpatialGizmo(Light* p_light=NULL);

-

-};

-

-class CameraSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(CameraSpatialGizmo,SpatialGizmoTool);

-

-	Camera* camera;

-

-public:

-

-

-	virtual String get_handle_name(int p_idx) const;

-	virtual Variant get_handle_value(int p_idx) const;

-	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);

-	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);

-

-	void redraw();

-	CameraSpatialGizmo(Camera* p_camera=NULL);

-

-};

-

-

-

-class MeshInstanceSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(MeshInstanceSpatialGizmo,SpatialGizmoTool);

-

-	MeshInstance* mesh;

-

-public:

-

-	void redraw();

-	MeshInstanceSpatialGizmo(MeshInstance* p_mesh=NULL);

-

-};

-

-class Position3DSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(Position3DSpatialGizmo,SpatialGizmoTool);

-

-	Position3D* p3d;

-

-public:

-

-	void redraw();

-	Position3DSpatialGizmo(Position3D* p_p3d=NULL);

-

-};

-

-class SkeletonSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(SkeletonSpatialGizmo,SpatialGizmoTool);

-

-	Skeleton* skel;

-

-public:

-

-	void redraw();

-	SkeletonSpatialGizmo(Skeleton* p_skel=NULL);

-

-};

-

-

-

-

-class SpatialPlayerSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(SpatialPlayerSpatialGizmo,SpatialGizmoTool);

-

-	SpatialPlayer* splayer;

-

-public:

-

-	void redraw();

-	SpatialPlayerSpatialGizmo(SpatialPlayer* p_splayer=NULL);

-

-};

-

-

-

-class TestCubeSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(TestCubeSpatialGizmo,SpatialGizmoTool);

-

-	TestCube* tc;

-

-public:

-	void redraw();

-	TestCubeSpatialGizmo(TestCube* p_tc=NULL);

-

-};

-

-

-class RoomSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(RoomSpatialGizmo,SpatialGizmoTool);

-

-

-	struct _EdgeKey {

-

-		Vector3 from;

-		Vector3 to;

-

-		bool operator<(const _EdgeKey& p_with) const { return from==p_with.from ? to < p_with.to : from < p_with.from; }

-	};

-

-

-

-	Room* room;

-

-public:

-

-	void redraw();

-	RoomSpatialGizmo(Room* p_room=NULL);

-

-};

-

-

-class PortalSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(PortalSpatialGizmo,SpatialGizmoTool);

-

-	Portal* portal;

-

-public:

-

-	void redraw();

-	PortalSpatialGizmo(Portal* p_portal=NULL);

-

-};

-

-

-class VisibilityNotifierGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(VisibilityNotifierGizmo ,SpatialGizmoTool);

-

-

-	VisibilityNotifier* notifier;

-

-public:

-

-	virtual String get_handle_name(int p_idx) const;

-	virtual Variant get_handle_value(int p_idx) const;

-	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);

-	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);

-

-	void redraw();

-	VisibilityNotifierGizmo(VisibilityNotifier* p_notifier=NULL);

-

-};

-

-

-

-class CollisionShapeSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(CollisionShapeSpatialGizmo,SpatialGizmoTool);

-

-	CollisionShape* cs;

-

-public:

-	virtual String get_handle_name(int p_idx) const;

-	virtual Variant get_handle_value(int p_idx) const;

-	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);

-	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);

-	void redraw();

-	CollisionShapeSpatialGizmo(CollisionShape* p_cs=NULL);

-

-};

-

-

-class CollisionPolygonSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(CollisionPolygonSpatialGizmo,SpatialGizmoTool);

-

-	CollisionPolygon* polygon;

-

-public:

-

-	void redraw();

-	CollisionPolygonSpatialGizmo(CollisionPolygon* p_polygon=NULL);

-

-};

-

-

-class RayCastSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(RayCastSpatialGizmo,SpatialGizmoTool);

-

-	RayCast* raycast;

-

-public:

-

-	void redraw();

-	RayCastSpatialGizmo(RayCast* p_raycast=NULL);

-

-};

-

-

-

-class VehicleWheelSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(VehicleWheelSpatialGizmo,SpatialGizmoTool);

-

-	VehicleWheel* car_wheel;

-

-public:

-

-	void redraw();

-	VehicleWheelSpatialGizmo(VehicleWheel* p_car_wheel=NULL);

-

-};

-

-

-class NavigationMeshSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(NavigationMeshSpatialGizmo,SpatialGizmoTool);

-

-

-	struct _EdgeKey {

-

-		Vector3 from;

-		Vector3 to;

-

-		bool operator<(const _EdgeKey& p_with) const { return from==p_with.from ? to < p_with.to : from < p_with.from; }

-	};

-

-

-

-	NavigationMeshInstance* navmesh;

-

-public:

-

-	void redraw();

-	NavigationMeshSpatialGizmo(NavigationMeshInstance* p_navmesh=NULL);

-

-};

-

-

-class PinJointSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(PinJointSpatialGizmo,SpatialGizmoTool);

-

-	PinJoint* p3d;

-

-public:

-

-	void redraw();

-	PinJointSpatialGizmo(PinJoint* p_p3d=NULL);

-

-};

-

-

-class HingeJointSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(HingeJointSpatialGizmo,SpatialGizmoTool);

-

-	HingeJoint* p3d;

-

-public:

-

-	void redraw();

-	HingeJointSpatialGizmo(HingeJoint* p_p3d=NULL);

-

-};

-

-class SliderJointSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(SliderJointSpatialGizmo,SpatialGizmoTool);

-

-	SliderJoint* p3d;

-

-public:

-

-	void redraw();

-	SliderJointSpatialGizmo(SliderJoint* p_p3d=NULL);

-

-};

-

-class ConeTwistJointSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(ConeTwistJointSpatialGizmo,SpatialGizmoTool);

-

-	ConeTwistJoint* p3d;

-

-public:

-

-	void redraw();

-	ConeTwistJointSpatialGizmo(ConeTwistJoint* p_p3d=NULL);

-

-};

-

-

-class Generic6DOFJointSpatialGizmo  : public SpatialGizmoTool {

-

-	OBJ_TYPE(Generic6DOFJointSpatialGizmo,SpatialGizmoTool);

-

-	Generic6DOFJoint* p3d;

-

-public:

-

-	void redraw();

-	Generic6DOFJointSpatialGizmo(Generic6DOFJoint* p_p3d=NULL);

-

-};

-

-

-class SpatialEditorGizmos  {

-public:

-

-	Ref<FixedMaterial> create_line_material(const Color& p_base_color);

-	Ref<FixedMaterial> create_solid_material(const Color& p_base_color);

-	Ref<FixedMaterial> handle2_material;

-	Ref<FixedMaterial> handle_material;

-	Ref<FixedMaterial> light_material;

-	Ref<FixedMaterial> light_material_omni_icon;

-	Ref<FixedMaterial> light_material_directional_icon;

-	Ref<FixedMaterial> camera_material;

-	Ref<FixedMaterial> skeleton_material;

-	Ref<FixedMaterial> room_material;

-	Ref<FixedMaterial> portal_material;

-	Ref<FixedMaterial> raycast_material;

-	Ref<FixedMaterial> visibility_notifier_material;

-	Ref<FixedMaterial> car_wheel_material;

-	Ref<FixedMaterial> joint_material;

-

-	Ref<FixedMaterial> navmesh_edge_material;

-	Ref<FixedMaterial> navmesh_solid_material;

-	Ref<FixedMaterial> navmesh_edge_material_disabled;

-	Ref<FixedMaterial> navmesh_solid_material_disabled;

-

-

-	Ref<FixedMaterial> sample_player_icon;

-	Ref<FixedMaterial> stream_player_icon;

-	Ref<FixedMaterial> visibility_notifier_icon;

-

-	Ref<FixedMaterial> shape_material;

-	Ref<Texture> handle_t;

-

-	Ref<Mesh> pos3d_mesh;

-	static SpatialEditorGizmos *singleton;

-

-	Ref<TriangleMesh> test_cube_tm;

-

-

-	Ref<SpatialEditorGizmo> get_gizmo(Spatial *p_spatial);

-

-	SpatialEditorGizmos();

-};

-

-#endif // SPATIAL_EDITOR_GIZMOS_H

-

+/*************************************************************************/
+/*  spatial_editor_gizmos.h                                              */
+/*************************************************************************/
+/*                       This file is part of:                           */
+/*                           GODOT ENGINE                                */
+/*                    http://www.godotengine.org                         */
+/*************************************************************************/
+/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur.                 */
+/*                                                                       */
+/* Permission is hereby granted, free of charge, to any person obtaining */
+/* a copy of this software and associated documentation files (the       */
+/* "Software"), to deal in the Software without restriction, including   */
+/* without limitation the rights to use, copy, modify, merge, publish,   */
+/* distribute, sublicense, and/or sell copies of the Software, and to    */
+/* permit persons to whom the Software is furnished to do so, subject to */
+/* the following conditions:                                             */
+/*                                                                       */
+/* The above copyright notice and this permission notice shall be        */
+/* included in all copies or substantial portions of the Software.       */
+/*                                                                       */
+/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
+/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
+/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
+/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
+/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
+/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
+/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
+/*************************************************************************/
+#ifndef SPATIAL_EDITOR_GIZMOS_H
+#define SPATIAL_EDITOR_GIZMOS_H
+
+
+#include "tools/editor/plugins/spatial_editor_plugin.h"
+#include "scene/3d/light.h"
+#include "scene/3d/camera.h"
+#include "scene/3d/position_3d.h"
+#include "scene/3d/spatial_sample_player.h"
+#include "scene/3d/spatial_stream_player.h"
+#include "scene/3d/test_cube.h"
+#include "scene/3d/mesh_instance.h"
+#include "scene/3d/body_shape.h"
+#include "scene/3d/room_instance.h"
+#include "scene/3d/visibility_notifier.h"
+#include "scene/3d/portal.h"
+#include "scene/3d/ray_cast.h"
+#include "scene/3d/navigation_mesh.h"
+
+#include "scene/3d/vehicle_body.h"
+#include "scene/3d/collision_polygon.h"
+#include "scene/3d/physics_joint.h"
+
+
+class Camera;
+
+class SpatialGizmoTool  : public SpatialEditorGizmo {
+
+	OBJ_TYPE(SpatialGizmoTool,SpatialGizmo);
+
+	struct Instance{
+
+		RID instance;
+		Ref<Mesh> mesh;
+		RID skeleton;
+		bool billboard;
+		bool unscaled;
+		bool can_intersect;
+		bool extra_margin;
+		Instance() {
+
+			billboard=false;
+			unscaled=false;
+			can_intersect=false;
+			extra_margin=false;
+		}
+
+		void create_instance(Spatial *p_base);
+
+	};
+
+	Vector<Vector3> collision_segments;
+	Ref<TriangleMesh> collision_mesh;
+
+	struct Handle {
+		Vector3 pos;
+		bool billboard;
+	};
+
+	Vector<Vector3> handles;
+	Vector<Vector3> secondary_handles;
+	bool billboard_handle;
+
+	bool valid;
+	Spatial *base;
+	Vector<Instance> instances;
+	Spatial *spatial_node;
+protected:
+	void add_lines(const Vector<Vector3> &p_lines,const Ref<Material>& p_material,bool p_billboard=false);
+	void add_mesh(const Ref<Mesh>& p_mesh,bool p_billboard=false,const RID& p_skeleton=RID());
+	void add_collision_segments(const Vector<Vector3> &p_lines);
+	void add_collision_triangles(const Ref<TriangleMesh>& p_tmesh);
+	void add_unscaled_billboard(const Ref<Material>& p_material,float p_scale=1);
+	void add_handles(const Vector<Vector3> &p_handles,bool p_billboard=false,bool p_secondary=false);
+
+	void set_spatial_node(Spatial *p_node);
+
+public:
+
+	virtual Vector3 get_handle_pos(int p_idx) const;
+	virtual bool intersect_frustum(const Camera *p_camera,const Vector<Plane> &p_frustum);
+	virtual bool intersect_ray(const Camera *p_camera,const Point2& p_point,  Vector3& r_pos, Vector3& r_normal,int *r_gizmo_handle=NULL,bool p_sec_first=false);
+
+	void clear();
+	void create();
+	void transform();
+	//void redraw();
+	void free();
+
+	SpatialGizmoTool();
+	~SpatialGizmoTool();
+};
+
+
+
+class LightSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(LightSpatialGizmo,SpatialGizmoTool);
+
+	Light* light;
+
+public:
+
+
+	virtual String get_handle_name(int p_idx) const;
+	virtual Variant get_handle_value(int p_idx) const;
+	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);
+	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);
+
+	void redraw();
+	LightSpatialGizmo(Light* p_light=NULL);
+
+};
+
+class CameraSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(CameraSpatialGizmo,SpatialGizmoTool);
+
+	Camera* camera;
+
+public:
+
+
+	virtual String get_handle_name(int p_idx) const;
+	virtual Variant get_handle_value(int p_idx) const;
+	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);
+	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);
+
+	void redraw();
+	CameraSpatialGizmo(Camera* p_camera=NULL);
+
+};
+
+
+
+class MeshInstanceSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(MeshInstanceSpatialGizmo,SpatialGizmoTool);
+
+	MeshInstance* mesh;
+
+public:
+
+	void redraw();
+	MeshInstanceSpatialGizmo(MeshInstance* p_mesh=NULL);
+
+};
+
+class Position3DSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(Position3DSpatialGizmo,SpatialGizmoTool);
+
+	Position3D* p3d;
+
+public:
+
+	void redraw();
+	Position3DSpatialGizmo(Position3D* p_p3d=NULL);
+
+};
+
+class SkeletonSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(SkeletonSpatialGizmo,SpatialGizmoTool);
+
+	Skeleton* skel;
+
+public:
+
+	void redraw();
+	SkeletonSpatialGizmo(Skeleton* p_skel=NULL);
+
+};
+
+
+
+
+class SpatialPlayerSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(SpatialPlayerSpatialGizmo,SpatialGizmoTool);
+
+	SpatialPlayer* splayer;
+
+public:
+
+	void redraw();
+	SpatialPlayerSpatialGizmo(SpatialPlayer* p_splayer=NULL);
+
+};
+
+
+
+class TestCubeSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(TestCubeSpatialGizmo,SpatialGizmoTool);
+
+	TestCube* tc;
+
+public:
+	void redraw();
+	TestCubeSpatialGizmo(TestCube* p_tc=NULL);
+
+};
+
+
+class RoomSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(RoomSpatialGizmo,SpatialGizmoTool);
+
+
+	struct _EdgeKey {
+
+		Vector3 from;
+		Vector3 to;
+
+		bool operator<(const _EdgeKey& p_with) const { return from==p_with.from ? to < p_with.to : from < p_with.from; }
+	};
+
+
+
+	Room* room;
+
+public:
+
+	void redraw();
+	RoomSpatialGizmo(Room* p_room=NULL);
+
+};
+
+
+class PortalSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(PortalSpatialGizmo,SpatialGizmoTool);
+
+	Portal* portal;
+
+public:
+
+	void redraw();
+	PortalSpatialGizmo(Portal* p_portal=NULL);
+
+};
+
+
+class VisibilityNotifierGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(VisibilityNotifierGizmo ,SpatialGizmoTool);
+
+
+	VisibilityNotifier* notifier;
+
+public:
+
+	virtual String get_handle_name(int p_idx) const;
+	virtual Variant get_handle_value(int p_idx) const;
+	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);
+	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);
+
+	void redraw();
+	VisibilityNotifierGizmo(VisibilityNotifier* p_notifier=NULL);
+
+};
+
+
+
+class CollisionShapeSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(CollisionShapeSpatialGizmo,SpatialGizmoTool);
+
+	CollisionShape* cs;
+
+public:
+	virtual String get_handle_name(int p_idx) const;
+	virtual Variant get_handle_value(int p_idx) const;
+	virtual void set_handle(int p_idx,Camera *p_camera, const Point2& p_point);
+	virtual void commit_handle(int p_idx,const Variant& p_restore,bool p_cancel=false);
+	void redraw();
+	CollisionShapeSpatialGizmo(CollisionShape* p_cs=NULL);
+
+};
+
+
+class CollisionPolygonSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(CollisionPolygonSpatialGizmo,SpatialGizmoTool);
+
+	CollisionPolygon* polygon;
+
+public:
+
+	void redraw();
+	CollisionPolygonSpatialGizmo(CollisionPolygon* p_polygon=NULL);
+
+};
+
+
+class RayCastSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(RayCastSpatialGizmo,SpatialGizmoTool);
+
+	RayCast* raycast;
+
+public:
+
+	void redraw();
+	RayCastSpatialGizmo(RayCast* p_raycast=NULL);
+
+};
+
+
+
+class VehicleWheelSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(VehicleWheelSpatialGizmo,SpatialGizmoTool);
+
+	VehicleWheel* car_wheel;
+
+public:
+
+	void redraw();
+	VehicleWheelSpatialGizmo(VehicleWheel* p_car_wheel=NULL);
+
+};
+
+
+class NavigationMeshSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(NavigationMeshSpatialGizmo,SpatialGizmoTool);
+
+
+	struct _EdgeKey {
+
+		Vector3 from;
+		Vector3 to;
+
+		bool operator<(const _EdgeKey& p_with) const { return from==p_with.from ? to < p_with.to : from < p_with.from; }
+	};
+
+
+
+	NavigationMeshInstance* navmesh;
+
+public:
+
+	void redraw();
+	NavigationMeshSpatialGizmo(NavigationMeshInstance* p_navmesh=NULL);
+
+};
+
+
+class PinJointSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(PinJointSpatialGizmo,SpatialGizmoTool);
+
+	PinJoint* p3d;
+
+public:
+
+	void redraw();
+	PinJointSpatialGizmo(PinJoint* p_p3d=NULL);
+
+};
+
+
+class HingeJointSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(HingeJointSpatialGizmo,SpatialGizmoTool);
+
+	HingeJoint* p3d;
+
+public:
+
+	void redraw();
+	HingeJointSpatialGizmo(HingeJoint* p_p3d=NULL);
+
+};
+
+class SliderJointSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(SliderJointSpatialGizmo,SpatialGizmoTool);
+
+	SliderJoint* p3d;
+
+public:
+
+	void redraw();
+	SliderJointSpatialGizmo(SliderJoint* p_p3d=NULL);
+
+};
+
+class ConeTwistJointSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(ConeTwistJointSpatialGizmo,SpatialGizmoTool);
+
+	ConeTwistJoint* p3d;
+
+public:
+
+	void redraw();
+	ConeTwistJointSpatialGizmo(ConeTwistJoint* p_p3d=NULL);
+
+};
+
+
+class Generic6DOFJointSpatialGizmo  : public SpatialGizmoTool {
+
+	OBJ_TYPE(Generic6DOFJointSpatialGizmo,SpatialGizmoTool);
+
+	Generic6DOFJoint* p3d;
+
+public:
+
+	void redraw();
+	Generic6DOFJointSpatialGizmo(Generic6DOFJoint* p_p3d=NULL);
+
+};
+
+
+class SpatialEditorGizmos  {
+public:
+
+	Ref<FixedMaterial> create_line_material(const Color& p_base_color);
+	Ref<FixedMaterial> create_solid_material(const Color& p_base_color);
+	Ref<FixedMaterial> handle2_material;
+	Ref<FixedMaterial> handle_material;
+	Ref<FixedMaterial> light_material;
+	Ref<FixedMaterial> light_material_omni_icon;
+	Ref<FixedMaterial> light_material_directional_icon;
+	Ref<FixedMaterial> camera_material;
+	Ref<FixedMaterial> skeleton_material;
+	Ref<FixedMaterial> room_material;
+	Ref<FixedMaterial> portal_material;
+	Ref<FixedMaterial> raycast_material;
+	Ref<FixedMaterial> visibility_notifier_material;
+	Ref<FixedMaterial> car_wheel_material;
+	Ref<FixedMaterial> joint_material;
+
+	Ref<FixedMaterial> navmesh_edge_material;
+	Ref<FixedMaterial> navmesh_solid_material;
+	Ref<FixedMaterial> navmesh_edge_material_disabled;
+	Ref<FixedMaterial> navmesh_solid_material_disabled;
+
+
+	Ref<FixedMaterial> sample_player_icon;
+	Ref<FixedMaterial> stream_player_icon;
+	Ref<FixedMaterial> visibility_notifier_icon;
+
+	Ref<FixedMaterial> shape_material;
+	Ref<Texture> handle_t;
+
+	Ref<Mesh> pos3d_mesh;
+	static SpatialEditorGizmos *singleton;
+
+	Ref<TriangleMesh> test_cube_tm;
+
+
+	Ref<SpatialEditorGizmo> get_gizmo(Spatial *p_spatial);
+
+	SpatialEditorGizmos();
+};
+
+#endif // SPATIAL_EDITOR_GIZMOS_H
+
diff --git a/tools/export/blender25/godot_export_manager.py b/tools/export/blender25/godot_export_manager.py
index 31db2c9e94..e390ae6ce3 100644
--- a/tools/export/blender25/godot_export_manager.py
+++ b/tools/export/blender25/godot_export_manager.py
@@ -1,474 +1,474 @@
-# ##### BEGIN GPL LICENSE BLOCK #####

-#

-# This program is free software; you can redistribute it and/or

-# modify it under the terms of the GNU General Public License

-# as published by the Free Software Foundation; either version 2

-# of the License, or (at your option) any later version.

-#

-# This program is distributed in the hope that it will be useful,

-# but WITHOUT ANY WARRANTY; without even the implied warranty of

-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

-# GNU General Public License for more details.

-#

-# You should have received a copy of the GNU General Public License

-# along with this program; if not, write to the Free Software Foundation,

-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.

-#

-# ##### END GPL LICENSE BLOCK #####

-

-# Script copyright (c) Andreas Esau

-

-bl_info = {

-    "name": "Godot Export Manager",

-    "author": "Andreas Esau",

-    "version": (1, 0),

-    "blender": (2, 7, 0),

-    "location": "Scene Properties > Godot Export Manager",

-    "description": "Godot Export Manager uses the Better Collada Exporter to manage Export Groups and automatically export the objects groups to Collada Files.",

-    "warning": "",

-    "wiki_url": ("http://www.godotengine.org"),

-    "tracker_url": "",

-    "category": "Import-Export"}

-

-import bpy

-from bpy.props import StringProperty, BoolProperty, EnumProperty, FloatProperty, FloatVectorProperty, IntProperty, CollectionProperty, PointerProperty

-import os

-from bpy.app.handlers import persistent

-from mathutils import Vector, Matrix

-

-class godot_export_manager(bpy.types.Panel):

-    bl_label = "Godot Export Manager"

-    bl_space_type = 'PROPERTIES'

-    bl_region_type = 'WINDOW'

-    bl_context = "scene"

-    

-    bpy.types.Scene.godot_export_on_save = BoolProperty(default=False)

-        

-    ### draw function for all ui elements

-    def draw(self, context):

-        layout = self.layout

-        split = self.layout.split()

-        scene = bpy.data.scenes[0]

-        ob = context.object

-        scene = context.scene

-        

-        row = layout.row()

-        col = row.column()

-        col.prop(scene,"godot_export_on_save",text="Export Groups on save")

-        

-        row = layout.row()

-        col = row.column(align=True)

-        op = col.operator("scene.godot_add_objects_to_group",text="Add selected objects to Group",icon="COPYDOWN")

-        

-        op = col.operator("scene.godot_delete_objects_from_group",text="Delete selected objects from Group",icon="PASTEDOWN")

-        

-        

-        

-        row = layout.row()

-        col = row.column()

-        col.label(text="Export Groups:")

-        

-        

-        row = layout.row()

-        col = row.column()

-        

-        col.template_list("UI_List_Godot","dummy",scene, "godot_export_groups", scene, "godot_export_groups_index",rows=1,maxrows=10,type='DEFAULT')

-        

-        col = row.column(align=True)

-        col.operator("scene.godot_add_export_group",text="",icon="ZOOMIN")

-        col.operator("scene.godot_delete_export_group",text="",icon="ZOOMOUT")

-        col.operator("scene.godot_export_all_groups",text="",icon="EXPORT")

-        

-        if len(scene.godot_export_groups) > 0:        

-            row = layout.row()

-            col = row.column()

-            group = scene.godot_export_groups[scene.godot_export_groups_index]

-            col.prop(group,"name",text="Group Name")

-            col.prop(group,"export_name",text="Export Name")

-            col.prop(group,"export_path",text="Export Filepath")

-            

-            row = layout.row()

-            col = row.column()

-            row = layout.row()

-            col = row.column()

-            col.label(text="Export Settings:")

-            

-            col = col.row(align=True)

-            col.prop(group,"apply_loc",toggle=True,icon="MAN_TRANS")

-            col.prop(group,"apply_rot",toggle=True,icon="MAN_ROT")

-            col.prop(group,"apply_scale",toggle=True,icon="MAN_SCALE")

-            

-            row = layout.row()

-            col = row.column()

-            

-            col.prop(group,"use_include_particle_duplicates")

-            col.prop(group,"use_mesh_modifiers")

-            col.prop(group,"use_tangent_arrays")

-            col.prop(group,"use_triangles")

-            col.prop(group,"use_copy_images")

-            col.prop(group,"use_active_layers")

-            col.prop(group,"use_exclude_ctrl_bones")

-            col.prop(group,"use_anim")

-            col.prop(group,"use_anim_action_all")

-            col.prop(group,"use_anim_skip_noexp")

-            col.prop(group,"use_anim_optimize")

-            col.prop(group,"anim_optimize_precision")

-            col.prop(group,"use_metadata")

-

-### Custom template_list look

-class UI_List_Godot(bpy.types.UIList):

-    def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index):

-        ob = data

-        slot = item

-        col = layout.row(align=True)

-        

-        col.label(text=item.name,icon="GROUP")

-        col.prop(item,"active",text="")

-        

-        op = col.operator("scene.godot_select_group_objects",text="",emboss=False,icon="RESTRICT_SELECT_OFF")

-        op.idx = index

-        op = col.operator("scene.godot_export_group",text="",emboss=False,icon="EXPORT")

-        op.idx = index

-              

-class add_objects_to_group(bpy.types.Operator):

-    bl_idname = "scene.godot_add_objects_to_group"

-    bl_label = "Add Objects to Group"

-    bl_description = "Adds the selected Objects to the active group below."

-    

-    undo = BoolProperty(default=True)

-    

-    def execute(self,context):

-        scene = context.scene

-        

-        objects_str = ""

-        if len(scene.godot_export_groups) > 0:

-            for i,object in enumerate(context.selected_objects):

-                if object.name not in scene.godot_export_groups[scene.godot_export_groups_index].nodes:

-                    node = scene.godot_export_groups[scene.godot_export_groups_index].nodes.add()

-                    node.name = object.name

-                    if i == 0:

-                        objects_str += object.name

-                    else:

-                        objects_str += ", "+object.name    

-        

-        

-            self.report({'INFO'}, objects_str + " added to group." )

-            if self.undo:

-                bpy.ops.ed.undo_push(message="Objects added to group")

-        else:

-            self.report({'WARNING'}, "Create a group first." )   

-        return{'FINISHED'}

-

-class del_objects_from_group(bpy.types.Operator):

-    bl_idname = "scene.godot_delete_objects_from_group"

-    bl_label = "Delete Objects from Group"

-    bl_description = "Delets the selected Objects from the active group below."

-    

-    def execute(self,context):

-        scene = context.scene

-        

-        if len(scene.godot_export_groups) > 0:

-                

-            selected_objects = []

-            for object in context.selected_objects:

-                selected_objects.append(object.name)

-            

-            objects_str = ""

-            j = 0

-            for i,node in enumerate(scene.godot_export_groups[scene.godot_export_groups_index].nodes):

-                if node.name in selected_objects:

-                    scene.godot_export_groups[scene.godot_export_groups_index].nodes.remove(i)

-

-                    

-                    if j == 0:

-                            objects_str += object.name

-                    else:

-                        objects_str += ", "+object.name 

-                    j+=1

-                    

-                    

-            self.report({'INFO'}, objects_str + " deleted from group." )                    

-            bpy.ops.ed.undo_push(message="Objects deleted from group")

-        else:

-            self.report({'WARNING'}, "There is no group to delete from." ) 

-        return{'FINISHED'}

-

-class select_group_objects(bpy.types.Operator):

-    bl_idname = "scene.godot_select_group_objects"

-    bl_label = "Select Group Objects"

-    bl_description = "Will select all group Objects in the scene."

-    

-    idx = IntProperty()

-    

-    def execute(self,context):

-        scene = context.scene

-        for object in context.scene.objects:

-            object.select = False

-        for node in scene.godot_export_groups[self.idx].nodes:

-            if node.name in bpy.data.objects:

-                bpy.data.objects[node.name].select = True

-                context.scene.objects.active = bpy.data.objects[node.name]

-        return{'FINISHED'}

-      

-class export_groups_autosave(bpy.types.Operator):

-    bl_idname = "scene.godot_export_groups_autosave"

-    bl_label = "Export All Groups"

-    bl_description = "Exports all groups to Collada."

-    

-    def execute(self,context):

-        scene = context.scene

-        if scene.godot_export_on_save:

-            for i in range(len(scene.godot_export_groups)):

-                if scene.godot_export_groups[i].active:

-                    bpy.ops.scene.godot_export_group(idx=i)

-        self.report({'INFO'}, "All Groups exported." )

-        bpy.ops.ed.undo_push(message="Export all Groups")   

-        return{'FINISHED'}

-    

-class export_all_groups(bpy.types.Operator):

-    bl_idname = "scene.godot_export_all_groups"

-    bl_label = "Export All Groups"

-    bl_description = "Exports all groups to Collada."

-    

-    def execute(self,context):

-        scene = context.scene

-        

-        for i in range(0,len(scene.godot_export_groups)):

-            bpy.ops.scene.godot_export_group(idx=i,export_all=True)

-                    

-        self.report({'INFO'}, "All Groups exported." )

-        return{'FINISHED'}    

-        

-        

-class export_group(bpy.types.Operator):

-    bl_idname = "scene.godot_export_group"

-    bl_label = "Export Group"

-    bl_description = "Exports the active group to destination folder as Collada file."

-    

-    idx = IntProperty(default=0)

-    export_all = BoolProperty(default=False)

-

-    

-    def copy_object_recursive(self,ob,parent,single_user = True):

-        new_ob = bpy.data.objects[ob.name].copy()

-        if single_user or ob.type=="ARMATURE":

-            new_mesh_data = new_ob.data.copy()

-            new_ob.data = new_mesh_data

-        bpy.context.scene.objects.link(new_ob)

-        

-        if ob != parent:

-           new_ob.parent = parent

-        else:

-            new_ob.parent = None   

-             

-        for child in ob.children:        

-            self.copy_object_recursive(child,new_ob,single_user)

-        new_ob.select = True    

-        return new_ob

-    

-    def delete_object(self,ob):

-        if ob != None:

-            for child in ob.children:

-                self.delete_object(child)

-            bpy.context.scene.objects.unlink(ob)

-            bpy.data.objects.remove(ob) 

-    

-    def convert_group_to_node(self,group):

-        if group.dupli_group != None:

-            for object in group.dupli_group.objects:

-                if object.parent == None:

-                    object = self.copy_object_recursive(object,object,True)

-                    matrix = Matrix(object.matrix_local)

-                    object.matrix_local = Matrix()

-                    object.matrix_local *= group.matrix_local

-                    object.matrix_local *= matrix

-        

-            self.delete_object(group)   

-    

-    def execute(self,context):

-        

-        scene = context.scene

-        group = context.scene.godot_export_groups

-        

-        if not group[self.idx].active and self.export_all:

-            return{'FINISHED'}

-        

-        for i,object in enumerate(group[self.idx].nodes):

-            if object.name in bpy.data.objects:

-                pass

-            else:

-                group[self.idx].nodes.remove(i)

-        bpy.ops.ed.undo_push(message="Clear not existent Group Nodes.")

-        

-        path = group[self.idx].export_path

-        if (path.find("//")==0 or path.find("\\\\")==0):

-            #if relative, convert to absolute

-            path = bpy.path.abspath(path)

-            path = path.replace("\\","/")

-        

-        ### if path exists and group export name is set the group will be exported  

-        if os.path.exists(path) and  group[self.idx].export_name != "":

-            

-            context.scene.layers = [True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True]

-            

-            

-            if group[self.idx].export_name.endswith(".dae"):

-                path = os.path.join(path,group[self.idx].export_name)

-            else:    

-                path = os.path.join(path,group[self.idx].export_name+".dae")

-            

-            hide_select = []    

-            for object in context.scene.objects:

-                hide_select.append(object.hide_select)

-                object.hide_select = False

-                object.select = False

-            context.scene.objects.active = None

-            

-            ### make particle duplicates, parent and select them

-            nodes_to_be_added = []

-            if group[self.idx].use_include_particle_duplicates:

-                for i,object in enumerate(group[self.idx].nodes):

-                    if bpy.data.objects[object.name].type != "EMPTY":

-                        context.scene.objects.active = bpy.data.objects[object.name]

-                        bpy.data.objects[object.name].select = True

-                        bpy.ops.object.duplicates_make_real()

-                        for object in context.selected_objects:

-                            nodes_to_be_added.append(object)

-                        bpy.ops.object.parent_set(type="OBJECT", keep_transform=False)

-

-                        for object in context.selected_objects:

-                            object.select = False

-                        bpy.data.objects[object.name].select = False

-                        context.scene.objects.active = None

-            for object in nodes_to_be_added:

-                object.select = True

-            

-            ### select all other nodes from the group

-            for i,object in enumerate(group[self.idx].nodes):

-                if bpy.data.objects[object.name].type == "EMPTY":

-                    self.convert_group_to_node(bpy.data.objects[object.name])

-                else:    

-                    bpy.data.objects[object.name].select = True

-                    

-            bpy.ops.object.transform_apply(location=group[self.idx].apply_loc, rotation=group[self.idx].apply_rot, scale=group[self.idx].apply_scale)

-            bpy.ops.export_scene.dae(check_existing=True, filepath=path, filter_glob="*.dae", object_types=group[self.idx].object_types, use_export_selected=group[self.idx].use_export_selected, use_mesh_modifiers=group[self.idx].use_mesh_modifiers, use_tangent_arrays=group[self.idx].use_tangent_arrays, use_triangles=group[self.idx].use_triangles, use_copy_images=group[self.idx].use_copy_images, use_active_layers=group[self.idx].use_active_layers, use_exclude_ctrl_bones=group[self.idx].use_exclude_ctrl_bones, use_anim=group[self.idx].use_anim, use_anim_action_all=group[self.idx].use_anim_action_all, use_anim_skip_noexp=group[self.idx].use_anim_skip_noexp, use_anim_optimize=group[self.idx].use_anim_optimize, anim_optimize_precision=group[self.idx].anim_optimize_precision, use_metadata=group[self.idx].use_metadata)    

-          

-            self.report({'INFO'}, '"'+group[self.idx].name+'"' + " Group exported." )  

-            msg = "Export Group "+group[self.idx].name

-            

-            bpy.ops.ed.undo_push(message="")

-            bpy.ops.ed.undo()

-            bpy.ops.ed.undo_push(message=msg)

-                

-        else:

-            self.report({'INFO'}, "Define Export Name and Export Path." )  

-        return{'FINISHED'}

-

-class add_export_group(bpy.types.Operator):

-    bl_idname = "scene.godot_add_export_group"

-    bl_label = "Adds a new export Group"

-    bl_description = "Creates a new Export Group with the selected Objects assigned to it."

-    

-    def execute(self,context):

-        scene = context.scene

-        

-        item = scene.godot_export_groups.add()

-        item.name = "New Group"

-        for object in context.selected_objects:

-            node = item.nodes.add()

-            node.name = object.name

-        scene.godot_export_groups_index = len(scene.godot_export_groups)-1    

-        bpy.ops.ed.undo_push(message="Create New Export Group")

-        return{'FINISHED'}

-    

-class del_export_group(bpy.types.Operator):

-    bl_idname = "scene.godot_delete_export_group"

-    bl_label = "Delets the selected export Group"

-    bl_description = "Delets the active Export Group."

-    

-    def invoke(self, context, event):

-        wm = context.window_manager 

-        return wm.invoke_confirm(self,event)

-    

-    def execute(self,context):

-        scene = context.scene

-        

-        scene.godot_export_groups.remove(scene.godot_export_groups_index)

-        if scene.godot_export_groups_index > 0:

-            scene.godot_export_groups_index -= 1

-        bpy.ops.ed.undo_push(message="Delete Export Group")

-        return{'FINISHED'}    

-

-class godot_node_list(bpy.types.PropertyGroup):

-    name = StringProperty()

-    

-class godot_export_groups(bpy.types.PropertyGroup):

-    name = StringProperty(name="Group Name")

-    export_name = StringProperty(name="scene_name")

-    nodes = CollectionProperty(type=godot_node_list)

-    export_path = StringProperty(subtype="DIR_PATH")

-    active = BoolProperty(default=True,description="Export Group")

-    

-    object_types = EnumProperty(name="Object Types",options={'ENUM_FLAG'},items=(('EMPTY', "Empty", ""),('CAMERA', "Camera", ""),('LAMP', "Lamp", ""),('ARMATURE', "Armature", ""),('MESH', "Mesh", ""),('CURVE', "Curve", ""),),default={'EMPTY', 'CAMERA', 'LAMP', 'ARMATURE', 'MESH','CURVE'})

-    

-    apply_scale = BoolProperty(name="Apply Scale",description="Apply Scale before export.",default=False)

-    apply_rot = BoolProperty(name="Apply Rotation",description="Apply Rotation before export.",default=False)

-    apply_loc = BoolProperty(name="Apply Location",description="Apply Location before export.",default=False)

-    

-    use_export_selected = BoolProperty(name="Selected Objects",description="Export only selected objects (and visible in active layers if that applies).",default=True)

-    use_mesh_modifiers = BoolProperty(name="Apply Modifiers",description="Apply modifiers to mesh objects (on a copy!).",default=True)

-    use_tangent_arrays = BoolProperty(name="Tangent Arrays",description="Export Tangent and Binormal arrays (for normalmapping).",default=False)

-    use_triangles = BoolProperty(name="Triangulate",description="Export Triangles instead of Polygons.",default=False)

-

-    use_copy_images = BoolProperty(name="Copy Images",description="Copy Images (create images/ subfolder)",default=False)

-    use_active_layers = BoolProperty(name="Active Layers",description="Export only objects on the active layers.",default=True)

-    use_exclude_ctrl_bones = BoolProperty(name="Exclude Control Bones",description="Exclude skeleton bones with names that begin with 'ctrl'.",default=True)

-    use_anim = BoolProperty(name="Export Animation",description="Export keyframe animation",default=False)

-    use_anim_action_all = BoolProperty(name="All Actions",description=("Export all actions for the first armature found in separate DAE files"),default=False)

-    use_anim_skip_noexp = BoolProperty(name="Skip (-noexp) Actions",description="Skip exporting of actions whose name end in (-noexp). Useful to skip control animations.",default=True)

-    use_anim_optimize = BoolProperty(name="Optimize Keyframes",description="Remove double keyframes",default=True)

-

-    anim_optimize_precision = FloatProperty(name="Precision",description=("Tolerence for comparing double keyframes (higher for greater accuracy)"),min=1, max=16,soft_min=1, soft_max=16,default=6.0)

-

-    use_metadata = BoolProperty(name="Use Metadata",default=True,options={'HIDDEN'})

-    use_include_particle_duplicates = BoolProperty(name="Include Particle Duplicates",default=True)

-

-def register():        

-    bpy.utils.register_class(godot_export_manager)

-    bpy.utils.register_class(godot_node_list)

-    bpy.utils.register_class(godot_export_groups)

-    bpy.utils.register_class(add_export_group)

-    bpy.utils.register_class(del_export_group)

-    bpy.utils.register_class(export_all_groups)

-    bpy.utils.register_class(export_groups_autosave)

-    bpy.utils.register_class(export_group)

-    bpy.utils.register_class(add_objects_to_group)

-    bpy.utils.register_class(del_objects_from_group)

-    bpy.utils.register_class(select_group_objects)

-    bpy.utils.register_class(UI_List_Godot)

-

-    bpy.types.Scene.godot_export_groups = CollectionProperty(type=godot_export_groups)

-    bpy.types.Scene.godot_export_groups_index = IntProperty(default=0,min=0)

-

-def unregister():

-    bpy.utils.unregister_class(godot_export_manager)

-    bpy.utils.unregister_class(godot_node_list)

-    bpy.utils.unregister_class(godot_export_groups)

-    bpy.utils.unregister_class(export_groups_autosave)

-    bpy.utils.unregister_class(add_export_group)

-    bpy.utils.unregister_class(del_export_group)

-    bpy.utils.unregister_class(export_all_groups)

-    bpy.utils.unregister_class(export_group)

-    bpy.utils.unregister_class(add_objects_to_group)

-    bpy.utils.unregister_class(del_objects_from_group)

-    bpy.utils.unregister_class(select_group_objects)

-    bpy.utils.unregister_class(UI_List_Godot)

-

-@persistent

-def auto_export(dummy):

-    bpy.ops.scene.godot_export_groups_autosave()

-        

-bpy.app.handlers.save_post.append(auto_export)

-

-if __name__ == "__main__":

-    register()

+# ##### BEGIN GPL LICENSE BLOCK #####
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation; either version 2
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+#
+# ##### END GPL LICENSE BLOCK #####
+
+# Script copyright (c) Andreas Esau
+
+bl_info = {
+    "name": "Godot Export Manager",
+    "author": "Andreas Esau",
+    "version": (1, 0),
+    "blender": (2, 7, 0),
+    "location": "Scene Properties > Godot Export Manager",
+    "description": "Godot Export Manager uses the Better Collada Exporter to manage Export Groups and automatically export the objects groups to Collada Files.",
+    "warning": "",
+    "wiki_url": ("http://www.godotengine.org"),
+    "tracker_url": "",
+    "category": "Import-Export"}
+
+import bpy
+from bpy.props import StringProperty, BoolProperty, EnumProperty, FloatProperty, FloatVectorProperty, IntProperty, CollectionProperty, PointerProperty
+import os
+from bpy.app.handlers import persistent
+from mathutils import Vector, Matrix
+
+class godot_export_manager(bpy.types.Panel):
+    bl_label = "Godot Export Manager"
+    bl_space_type = 'PROPERTIES'
+    bl_region_type = 'WINDOW'
+    bl_context = "scene"
+    
+    bpy.types.Scene.godot_export_on_save = BoolProperty(default=False)
+        
+    ### draw function for all ui elements
+    def draw(self, context):
+        layout = self.layout
+        split = self.layout.split()
+        scene = bpy.data.scenes[0]
+        ob = context.object
+        scene = context.scene
+        
+        row = layout.row()
+        col = row.column()
+        col.prop(scene,"godot_export_on_save",text="Export Groups on save")
+        
+        row = layout.row()
+        col = row.column(align=True)
+        op = col.operator("scene.godot_add_objects_to_group",text="Add selected objects to Group",icon="COPYDOWN")
+        
+        op = col.operator("scene.godot_delete_objects_from_group",text="Delete selected objects from Group",icon="PASTEDOWN")
+        
+        
+        
+        row = layout.row()
+        col = row.column()
+        col.label(text="Export Groups:")
+        
+        
+        row = layout.row()
+        col = row.column()
+        
+        col.template_list("UI_List_Godot","dummy",scene, "godot_export_groups", scene, "godot_export_groups_index",rows=1,maxrows=10,type='DEFAULT')
+        
+        col = row.column(align=True)
+        col.operator("scene.godot_add_export_group",text="",icon="ZOOMIN")
+        col.operator("scene.godot_delete_export_group",text="",icon="ZOOMOUT")
+        col.operator("scene.godot_export_all_groups",text="",icon="EXPORT")
+        
+        if len(scene.godot_export_groups) > 0:        
+            row = layout.row()
+            col = row.column()
+            group = scene.godot_export_groups[scene.godot_export_groups_index]
+            col.prop(group,"name",text="Group Name")
+            col.prop(group,"export_name",text="Export Name")
+            col.prop(group,"export_path",text="Export Filepath")
+            
+            row = layout.row()
+            col = row.column()
+            row = layout.row()
+            col = row.column()
+            col.label(text="Export Settings:")
+            
+            col = col.row(align=True)
+            col.prop(group,"apply_loc",toggle=True,icon="MAN_TRANS")
+            col.prop(group,"apply_rot",toggle=True,icon="MAN_ROT")
+            col.prop(group,"apply_scale",toggle=True,icon="MAN_SCALE")
+            
+            row = layout.row()
+            col = row.column()
+            
+            col.prop(group,"use_include_particle_duplicates")
+            col.prop(group,"use_mesh_modifiers")
+            col.prop(group,"use_tangent_arrays")
+            col.prop(group,"use_triangles")
+            col.prop(group,"use_copy_images")
+            col.prop(group,"use_active_layers")
+            col.prop(group,"use_exclude_ctrl_bones")
+            col.prop(group,"use_anim")
+            col.prop(group,"use_anim_action_all")
+            col.prop(group,"use_anim_skip_noexp")
+            col.prop(group,"use_anim_optimize")
+            col.prop(group,"anim_optimize_precision")
+            col.prop(group,"use_metadata")
+
+### Custom template_list look
+class UI_List_Godot(bpy.types.UIList):
+    def draw_item(self, context, layout, data, item, icon, active_data, active_propname, index):
+        ob = data
+        slot = item
+        col = layout.row(align=True)
+        
+        col.label(text=item.name,icon="GROUP")
+        col.prop(item,"active",text="")
+        
+        op = col.operator("scene.godot_select_group_objects",text="",emboss=False,icon="RESTRICT_SELECT_OFF")
+        op.idx = index
+        op = col.operator("scene.godot_export_group",text="",emboss=False,icon="EXPORT")
+        op.idx = index
+              
+class add_objects_to_group(bpy.types.Operator):
+    bl_idname = "scene.godot_add_objects_to_group"
+    bl_label = "Add Objects to Group"
+    bl_description = "Adds the selected Objects to the active group below."
+    
+    undo = BoolProperty(default=True)
+    
+    def execute(self,context):
+        scene = context.scene
+        
+        objects_str = ""
+        if len(scene.godot_export_groups) > 0:
+            for i,object in enumerate(context.selected_objects):
+                if object.name not in scene.godot_export_groups[scene.godot_export_groups_index].nodes:
+                    node = scene.godot_export_groups[scene.godot_export_groups_index].nodes.add()
+                    node.name = object.name
+                    if i == 0:
+                        objects_str += object.name
+                    else:
+                        objects_str += ", "+object.name    
+        
+        
+            self.report({'INFO'}, objects_str + " added to group." )
+            if self.undo:
+                bpy.ops.ed.undo_push(message="Objects added to group")
+        else:
+            self.report({'WARNING'}, "Create a group first." )   
+        return{'FINISHED'}
+
+class del_objects_from_group(bpy.types.Operator):
+    bl_idname = "scene.godot_delete_objects_from_group"
+    bl_label = "Delete Objects from Group"
+    bl_description = "Delets the selected Objects from the active group below."
+    
+    def execute(self,context):
+        scene = context.scene
+        
+        if len(scene.godot_export_groups) > 0:
+                
+            selected_objects = []
+            for object in context.selected_objects:
+                selected_objects.append(object.name)
+            
+            objects_str = ""
+            j = 0
+            for i,node in enumerate(scene.godot_export_groups[scene.godot_export_groups_index].nodes):
+                if node.name in selected_objects:
+                    scene.godot_export_groups[scene.godot_export_groups_index].nodes.remove(i)
+
+                    
+                    if j == 0:
+                            objects_str += object.name
+                    else:
+                        objects_str += ", "+object.name 
+                    j+=1
+                    
+                    
+            self.report({'INFO'}, objects_str + " deleted from group." )                    
+            bpy.ops.ed.undo_push(message="Objects deleted from group")
+        else:
+            self.report({'WARNING'}, "There is no group to delete from." ) 
+        return{'FINISHED'}
+
+class select_group_objects(bpy.types.Operator):
+    bl_idname = "scene.godot_select_group_objects"
+    bl_label = "Select Group Objects"
+    bl_description = "Will select all group Objects in the scene."
+    
+    idx = IntProperty()
+    
+    def execute(self,context):
+        scene = context.scene
+        for object in context.scene.objects:
+            object.select = False
+        for node in scene.godot_export_groups[self.idx].nodes:
+            if node.name in bpy.data.objects:
+                bpy.data.objects[node.name].select = True
+                context.scene.objects.active = bpy.data.objects[node.name]
+        return{'FINISHED'}
+      
+class export_groups_autosave(bpy.types.Operator):
+    bl_idname = "scene.godot_export_groups_autosave"
+    bl_label = "Export All Groups"
+    bl_description = "Exports all groups to Collada."
+    
+    def execute(self,context):
+        scene = context.scene
+        if scene.godot_export_on_save:
+            for i in range(len(scene.godot_export_groups)):
+                if scene.godot_export_groups[i].active:
+                    bpy.ops.scene.godot_export_group(idx=i)
+        self.report({'INFO'}, "All Groups exported." )
+        bpy.ops.ed.undo_push(message="Export all Groups")   
+        return{'FINISHED'}
+    
+class export_all_groups(bpy.types.Operator):
+    bl_idname = "scene.godot_export_all_groups"
+    bl_label = "Export All Groups"
+    bl_description = "Exports all groups to Collada."
+    
+    def execute(self,context):
+        scene = context.scene
+        
+        for i in range(0,len(scene.godot_export_groups)):
+            bpy.ops.scene.godot_export_group(idx=i,export_all=True)
+                    
+        self.report({'INFO'}, "All Groups exported." )
+        return{'FINISHED'}    
+        
+        
+class export_group(bpy.types.Operator):
+    bl_idname = "scene.godot_export_group"
+    bl_label = "Export Group"
+    bl_description = "Exports the active group to destination folder as Collada file."
+    
+    idx = IntProperty(default=0)
+    export_all = BoolProperty(default=False)
+
+    
+    def copy_object_recursive(self,ob,parent,single_user = True):
+        new_ob = bpy.data.objects[ob.name].copy()
+        if single_user or ob.type=="ARMATURE":
+            new_mesh_data = new_ob.data.copy()
+            new_ob.data = new_mesh_data
+        bpy.context.scene.objects.link(new_ob)
+        
+        if ob != parent:
+           new_ob.parent = parent
+        else:
+            new_ob.parent = None   
+             
+        for child in ob.children:        
+            self.copy_object_recursive(child,new_ob,single_user)
+        new_ob.select = True    
+        return new_ob
+    
+    def delete_object(self,ob):
+        if ob != None:
+            for child in ob.children:
+                self.delete_object(child)
+            bpy.context.scene.objects.unlink(ob)
+            bpy.data.objects.remove(ob) 
+    
+    def convert_group_to_node(self,group):
+        if group.dupli_group != None:
+            for object in group.dupli_group.objects:
+                if object.parent == None:
+                    object = self.copy_object_recursive(object,object,True)
+                    matrix = Matrix(object.matrix_local)
+                    object.matrix_local = Matrix()
+                    object.matrix_local *= group.matrix_local
+                    object.matrix_local *= matrix
+        
+            self.delete_object(group)   
+    
+    def execute(self,context):
+        
+        scene = context.scene
+        group = context.scene.godot_export_groups
+        
+        if not group[self.idx].active and self.export_all:
+            return{'FINISHED'}
+        
+        for i,object in enumerate(group[self.idx].nodes):
+            if object.name in bpy.data.objects:
+                pass
+            else:
+                group[self.idx].nodes.remove(i)
+        bpy.ops.ed.undo_push(message="Clear not existent Group Nodes.")
+        
+        path = group[self.idx].export_path
+        if (path.find("//")==0 or path.find("\\\\")==0):
+            #if relative, convert to absolute
+            path = bpy.path.abspath(path)
+            path = path.replace("\\","/")
+        
+        ### if path exists and group export name is set the group will be exported  
+        if os.path.exists(path) and  group[self.idx].export_name != "":
+            
+            context.scene.layers = [True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True,True]
+            
+            
+            if group[self.idx].export_name.endswith(".dae"):
+                path = os.path.join(path,group[self.idx].export_name)
+            else:    
+                path = os.path.join(path,group[self.idx].export_name+".dae")
+            
+            hide_select = []    
+            for object in context.scene.objects:
+                hide_select.append(object.hide_select)
+                object.hide_select = False
+                object.select = False
+            context.scene.objects.active = None
+            
+            ### make particle duplicates, parent and select them
+            nodes_to_be_added = []
+            if group[self.idx].use_include_particle_duplicates:
+                for i,object in enumerate(group[self.idx].nodes):
+                    if bpy.data.objects[object.name].type != "EMPTY":
+                        context.scene.objects.active = bpy.data.objects[object.name]
+                        bpy.data.objects[object.name].select = True
+                        bpy.ops.object.duplicates_make_real()
+                        for object in context.selected_objects:
+                            nodes_to_be_added.append(object)
+                        bpy.ops.object.parent_set(type="OBJECT", keep_transform=False)
+
+                        for object in context.selected_objects:
+                            object.select = False
+                        bpy.data.objects[object.name].select = False
+                        context.scene.objects.active = None
+            for object in nodes_to_be_added:
+                object.select = True
+            
+            ### select all other nodes from the group
+            for i,object in enumerate(group[self.idx].nodes):
+                if bpy.data.objects[object.name].type == "EMPTY":
+                    self.convert_group_to_node(bpy.data.objects[object.name])
+                else:    
+                    bpy.data.objects[object.name].select = True
+                    
+            bpy.ops.object.transform_apply(location=group[self.idx].apply_loc, rotation=group[self.idx].apply_rot, scale=group[self.idx].apply_scale)
+            bpy.ops.export_scene.dae(check_existing=True, filepath=path, filter_glob="*.dae", object_types=group[self.idx].object_types, use_export_selected=group[self.idx].use_export_selected, use_mesh_modifiers=group[self.idx].use_mesh_modifiers, use_tangent_arrays=group[self.idx].use_tangent_arrays, use_triangles=group[self.idx].use_triangles, use_copy_images=group[self.idx].use_copy_images, use_active_layers=group[self.idx].use_active_layers, use_exclude_ctrl_bones=group[self.idx].use_exclude_ctrl_bones, use_anim=group[self.idx].use_anim, use_anim_action_all=group[self.idx].use_anim_action_all, use_anim_skip_noexp=group[self.idx].use_anim_skip_noexp, use_anim_optimize=group[self.idx].use_anim_optimize, anim_optimize_precision=group[self.idx].anim_optimize_precision, use_metadata=group[self.idx].use_metadata)    
+          
+            self.report({'INFO'}, '"'+group[self.idx].name+'"' + " Group exported." )  
+            msg = "Export Group "+group[self.idx].name
+            
+            bpy.ops.ed.undo_push(message="")
+            bpy.ops.ed.undo()
+            bpy.ops.ed.undo_push(message=msg)
+                
+        else:
+            self.report({'INFO'}, "Define Export Name and Export Path." )  
+        return{'FINISHED'}
+
+class add_export_group(bpy.types.Operator):
+    bl_idname = "scene.godot_add_export_group"
+    bl_label = "Adds a new export Group"
+    bl_description = "Creates a new Export Group with the selected Objects assigned to it."
+    
+    def execute(self,context):
+        scene = context.scene
+        
+        item = scene.godot_export_groups.add()
+        item.name = "New Group"
+        for object in context.selected_objects:
+            node = item.nodes.add()
+            node.name = object.name
+        scene.godot_export_groups_index = len(scene.godot_export_groups)-1    
+        bpy.ops.ed.undo_push(message="Create New Export Group")
+        return{'FINISHED'}
+    
+class del_export_group(bpy.types.Operator):
+    bl_idname = "scene.godot_delete_export_group"
+    bl_label = "Delets the selected export Group"
+    bl_description = "Delets the active Export Group."
+    
+    def invoke(self, context, event):
+        wm = context.window_manager 
+        return wm.invoke_confirm(self,event)
+    
+    def execute(self,context):
+        scene = context.scene
+        
+        scene.godot_export_groups.remove(scene.godot_export_groups_index)
+        if scene.godot_export_groups_index > 0:
+            scene.godot_export_groups_index -= 1
+        bpy.ops.ed.undo_push(message="Delete Export Group")
+        return{'FINISHED'}    
+
+class godot_node_list(bpy.types.PropertyGroup):
+    name = StringProperty()
+    
+class godot_export_groups(bpy.types.PropertyGroup):
+    name = StringProperty(name="Group Name")
+    export_name = StringProperty(name="scene_name")
+    nodes = CollectionProperty(type=godot_node_list)
+    export_path = StringProperty(subtype="DIR_PATH")
+    active = BoolProperty(default=True,description="Export Group")
+    
+    object_types = EnumProperty(name="Object Types",options={'ENUM_FLAG'},items=(('EMPTY', "Empty", ""),('CAMERA', "Camera", ""),('LAMP', "Lamp", ""),('ARMATURE', "Armature", ""),('MESH', "Mesh", ""),('CURVE', "Curve", ""),),default={'EMPTY', 'CAMERA', 'LAMP', 'ARMATURE', 'MESH','CURVE'})
+    
+    apply_scale = BoolProperty(name="Apply Scale",description="Apply Scale before export.",default=False)
+    apply_rot = BoolProperty(name="Apply Rotation",description="Apply Rotation before export.",default=False)
+    apply_loc = BoolProperty(name="Apply Location",description="Apply Location before export.",default=False)
+    
+    use_export_selected = BoolProperty(name="Selected Objects",description="Export only selected objects (and visible in active layers if that applies).",default=True)
+    use_mesh_modifiers = BoolProperty(name="Apply Modifiers",description="Apply modifiers to mesh objects (on a copy!).",default=True)
+    use_tangent_arrays = BoolProperty(name="Tangent Arrays",description="Export Tangent and Binormal arrays (for normalmapping).",default=False)
+    use_triangles = BoolProperty(name="Triangulate",description="Export Triangles instead of Polygons.",default=False)
+
+    use_copy_images = BoolProperty(name="Copy Images",description="Copy Images (create images/ subfolder)",default=False)
+    use_active_layers = BoolProperty(name="Active Layers",description="Export only objects on the active layers.",default=True)
+    use_exclude_ctrl_bones = BoolProperty(name="Exclude Control Bones",description="Exclude skeleton bones with names that begin with 'ctrl'.",default=True)
+    use_anim = BoolProperty(name="Export Animation",description="Export keyframe animation",default=False)
+    use_anim_action_all = BoolProperty(name="All Actions",description=("Export all actions for the first armature found in separate DAE files"),default=False)
+    use_anim_skip_noexp = BoolProperty(name="Skip (-noexp) Actions",description="Skip exporting of actions whose name end in (-noexp). Useful to skip control animations.",default=True)
+    use_anim_optimize = BoolProperty(name="Optimize Keyframes",description="Remove double keyframes",default=True)
+
+    anim_optimize_precision = FloatProperty(name="Precision",description=("Tolerence for comparing double keyframes (higher for greater accuracy)"),min=1, max=16,soft_min=1, soft_max=16,default=6.0)
+
+    use_metadata = BoolProperty(name="Use Metadata",default=True,options={'HIDDEN'})
+    use_include_particle_duplicates = BoolProperty(name="Include Particle Duplicates",default=True)
+
+def register():        
+    bpy.utils.register_class(godot_export_manager)
+    bpy.utils.register_class(godot_node_list)
+    bpy.utils.register_class(godot_export_groups)
+    bpy.utils.register_class(add_export_group)
+    bpy.utils.register_class(del_export_group)
+    bpy.utils.register_class(export_all_groups)
+    bpy.utils.register_class(export_groups_autosave)
+    bpy.utils.register_class(export_group)
+    bpy.utils.register_class(add_objects_to_group)
+    bpy.utils.register_class(del_objects_from_group)
+    bpy.utils.register_class(select_group_objects)
+    bpy.utils.register_class(UI_List_Godot)
+
+    bpy.types.Scene.godot_export_groups = CollectionProperty(type=godot_export_groups)
+    bpy.types.Scene.godot_export_groups_index = IntProperty(default=0,min=0)
+
+def unregister():
+    bpy.utils.unregister_class(godot_export_manager)
+    bpy.utils.unregister_class(godot_node_list)
+    bpy.utils.unregister_class(godot_export_groups)
+    bpy.utils.unregister_class(export_groups_autosave)
+    bpy.utils.unregister_class(add_export_group)
+    bpy.utils.unregister_class(del_export_group)
+    bpy.utils.unregister_class(export_all_groups)
+    bpy.utils.unregister_class(export_group)
+    bpy.utils.unregister_class(add_objects_to_group)
+    bpy.utils.unregister_class(del_objects_from_group)
+    bpy.utils.unregister_class(select_group_objects)
+    bpy.utils.unregister_class(UI_List_Godot)
+
+@persistent
+def auto_export(dummy):
+    bpy.ops.scene.godot_export_groups_autosave()
+        
+bpy.app.handlers.save_post.append(auto_export)
+
+if __name__ == "__main__":
+    register()