diff options
Diffstat (limited to 'drivers/builtin_openssl2/crypto/modes')
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/asm/ghash-armv4.pl | 6 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/asm/ghash-x86.pl | 2 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/cbc128.c | 250 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/ccm128.c | 682 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/cfb128.c | 292 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/ctr128.c | 347 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/cts128.c | 707 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/gcm128.c | 3400 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/modes_lcl.h | 159 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/ofb128.c | 105 | ||||
-rw-r--r-- | drivers/builtin_openssl2/crypto/modes/xts128.c | 243 |
11 files changed, 3378 insertions, 2815 deletions
diff --git a/drivers/builtin_openssl2/crypto/modes/asm/ghash-armv4.pl b/drivers/builtin_openssl2/crypto/modes/asm/ghash-armv4.pl index d91586ee29..e46f8e34da 100644 --- a/drivers/builtin_openssl2/crypto/modes/asm/ghash-armv4.pl +++ b/drivers/builtin_openssl2/crypto/modes/asm/ghash-armv4.pl @@ -374,8 +374,8 @@ gcm_ghash_neon: vdup.8 $xi,`&Dlo("$IN")`[0] @ broadcast lowest byte .Linner_neon: subs $cnt,$cnt,#1 - vmull.p8 $Qlo,$Hlo,$xi @ H.lo·Xi[i] - vmull.p8 $Qhi,$Hhi,$xi @ H.hi·Xi[i] + vmull.p8 $Qlo,$Hlo,$xi @ H.lo·Xi[i] + vmull.p8 $Qhi,$Hhi,$xi @ H.hi·Xi[i] vext.8 $IN,$zero,#1 @ IN>>=8 veor $Z,$Qpost @ modulo-scheduled part @@ -388,7 +388,7 @@ gcm_ghash_neon: vsli.8 $Zo,$T,#1 @ compose the "carry" byte vext.8 $Z,$zero,#1 @ Z>>=8 - vmull.p8 $R,$Zo,$mod @ "carry"·0xe1 + vmull.p8 $R,$Zo,$mod @ "carry"·0xe1 vshr.u8 $Zo,$T,#7 @ save Z's bottom bit vext.8 $Qpost,$Qlo,$zero,#1 @ Qlo>>=8 veor $Z,$Qhi diff --git a/drivers/builtin_openssl2/crypto/modes/asm/ghash-x86.pl b/drivers/builtin_openssl2/crypto/modes/asm/ghash-x86.pl index 83c727e07f..2426cd0c8a 100644 --- a/drivers/builtin_openssl2/crypto/modes/asm/ghash-x86.pl +++ b/drivers/builtin_openssl2/crypto/modes/asm/ghash-x86.pl @@ -346,7 +346,7 @@ $S=12; # shift factor for rem_4bit # effective address calculation and finally merge of value to Z.hi. # Reference to rem_4bit is scheduled so late that I had to >>4 # rem_4bit elements. This resulted in 20-45% procent improvement -# on contemporary µ-archs. +# on contemporary µ-archs. { my $cnt; my $rem_4bit = "eax"; diff --git a/drivers/builtin_openssl2/crypto/modes/cbc128.c b/drivers/builtin_openssl2/crypto/modes/cbc128.c index 0e54f75470..1ed7967274 100644 --- a/drivers/builtin_openssl2/crypto/modes/cbc128.c +++ b/drivers/builtin_openssl2/crypto/modes/cbc128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -60,146 +60,148 @@ #include <assert.h> #ifndef STRICT_ALIGNMENT -# define STRICT_ALIGNMENT 0 +# define STRICT_ALIGNMENT 0 #endif void CRYPTO_cbc128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) + size_t len, const void *key, + unsigned char ivec[16], block128_f block) { - size_t n; - const unsigned char *iv = ivec; + size_t n; + const unsigned char *iv = ivec; - assert(in && out && key && ivec); + assert(in && out && key && ivec); #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (STRICT_ALIGNMENT && - ((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) { - while (len>=16) { - for(n=0; n<16; ++n) - out[n] = in[n] ^ iv[n]; - (*block)(out, out, key); - iv = out; - len -= 16; - in += 16; - out += 16; - } - } else { - while (len>=16) { - for(n=0; n<16; n+=sizeof(size_t)) - *(size_t*)(out+n) = - *(size_t*)(in+n) ^ *(size_t*)(iv+n); - (*block)(out, out, key); - iv = out; - len -= 16; - in += 16; - out += 16; - } - } + if (STRICT_ALIGNMENT && + ((size_t)in | (size_t)out | (size_t)ivec) % sizeof(size_t) != 0) { + while (len >= 16) { + for (n = 0; n < 16; ++n) + out[n] = in[n] ^ iv[n]; + (*block) (out, out, key); + iv = out; + len -= 16; + in += 16; + out += 16; + } + } else { + while (len >= 16) { + for (n = 0; n < 16; n += sizeof(size_t)) + *(size_t *)(out + n) = + *(size_t *)(in + n) ^ *(size_t *)(iv + n); + (*block) (out, out, key); + iv = out; + len -= 16; + in += 16; + out += 16; + } + } #endif - while (len) { - for(n=0; n<16 && n<len; ++n) - out[n] = in[n] ^ iv[n]; - for(; n<16; ++n) - out[n] = iv[n]; - (*block)(out, out, key); - iv = out; - if (len<=16) break; - len -= 16; - in += 16; - out += 16; - } - memcpy(ivec,iv,16); + while (len) { + for (n = 0; n < 16 && n < len; ++n) + out[n] = in[n] ^ iv[n]; + for (; n < 16; ++n) + out[n] = iv[n]; + (*block) (out, out, key); + iv = out; + if (len <= 16) + break; + len -= 16; + in += 16; + out += 16; + } + memcpy(ivec, iv, 16); } void CRYPTO_cbc128_decrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) + size_t len, const void *key, + unsigned char ivec[16], block128_f block) { - size_t n; - union { size_t t[16/sizeof(size_t)]; unsigned char c[16]; } tmp; + size_t n; + union { + size_t t[16 / sizeof(size_t)]; + unsigned char c[16]; + } tmp; - assert(in && out && key && ivec); + assert(in && out && key && ivec); #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (in != out) { - const unsigned char *iv = ivec; + if (in != out) { + const unsigned char *iv = ivec; - if (STRICT_ALIGNMENT && - ((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) { - while (len>=16) { - (*block)(in, out, key); - for(n=0; n<16; ++n) - out[n] ^= iv[n]; - iv = in; - len -= 16; - in += 16; - out += 16; - } - } - else if (16%sizeof(size_t) == 0) { /* always true */ - while (len>=16) { - size_t *out_t=(size_t *)out, *iv_t=(size_t *)iv; + if (STRICT_ALIGNMENT && + ((size_t)in | (size_t)out | (size_t)ivec) % sizeof(size_t) != 0) { + while (len >= 16) { + (*block) (in, out, key); + for (n = 0; n < 16; ++n) + out[n] ^= iv[n]; + iv = in; + len -= 16; + in += 16; + out += 16; + } + } else if (16 % sizeof(size_t) == 0) { /* always true */ + while (len >= 16) { + size_t *out_t = (size_t *)out, *iv_t = (size_t *)iv; - (*block)(in, out, key); - for(n=0; n<16/sizeof(size_t); n++) - out_t[n] ^= iv_t[n]; - iv = in; - len -= 16; - in += 16; - out += 16; - } - } - memcpy(ivec,iv,16); - } else { - if (STRICT_ALIGNMENT && - ((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) { - unsigned char c; - while (len>=16) { - (*block)(in, tmp.c, key); - for(n=0; n<16; ++n) { - c = in[n]; - out[n] = tmp.c[n] ^ ivec[n]; - ivec[n] = c; - } - len -= 16; - in += 16; - out += 16; - } - } - else if (16%sizeof(size_t) == 0) { /* always true */ - while (len>=16) { - size_t c, *out_t=(size_t *)out, *ivec_t=(size_t *)ivec; - const size_t *in_t=(const size_t *)in; + (*block) (in, out, key); + for (n = 0; n < 16 / sizeof(size_t); n++) + out_t[n] ^= iv_t[n]; + iv = in; + len -= 16; + in += 16; + out += 16; + } + } + memcpy(ivec, iv, 16); + } else { + if (STRICT_ALIGNMENT && + ((size_t)in | (size_t)out | (size_t)ivec) % sizeof(size_t) != 0) { + unsigned char c; + while (len >= 16) { + (*block) (in, tmp.c, key); + for (n = 0; n < 16; ++n) { + c = in[n]; + out[n] = tmp.c[n] ^ ivec[n]; + ivec[n] = c; + } + len -= 16; + in += 16; + out += 16; + } + } else if (16 % sizeof(size_t) == 0) { /* always true */ + while (len >= 16) { + size_t c, *out_t = (size_t *)out, *ivec_t = (size_t *)ivec; + const size_t *in_t = (const size_t *)in; - (*block)(in, tmp.c, key); - for(n=0; n<16/sizeof(size_t); n++) { - c = in_t[n]; - out_t[n] = tmp.t[n] ^ ivec_t[n]; - ivec_t[n] = c; - } - len -= 16; - in += 16; - out += 16; - } - } - } + (*block) (in, tmp.c, key); + for (n = 0; n < 16 / sizeof(size_t); n++) { + c = in_t[n]; + out_t[n] = tmp.t[n] ^ ivec_t[n]; + ivec_t[n] = c; + } + len -= 16; + in += 16; + out += 16; + } + } + } #endif - while (len) { - unsigned char c; - (*block)(in, tmp.c, key); - for(n=0; n<16 && n<len; ++n) { - c = in[n]; - out[n] = tmp.c[n] ^ ivec[n]; - ivec[n] = c; - } - if (len<=16) { - for (; n<16; ++n) - ivec[n] = in[n]; - break; - } - len -= 16; - in += 16; - out += 16; - } + while (len) { + unsigned char c; + (*block) (in, tmp.c, key); + for (n = 0; n < 16 && n < len; ++n) { + c = in[n]; + out[n] = tmp.c[n] ^ ivec[n]; + ivec[n] = c; + } + if (len <= 16) { + for (; n < 16; ++n) + ivec[n] = in[n]; + break; + } + len -= 16; + in += 16; + out += 16; + } } diff --git a/drivers/builtin_openssl2/crypto/modes/ccm128.c b/drivers/builtin_openssl2/crypto/modes/ccm128.c index 3ce11d0d98..c1ded0f914 100644 --- a/drivers/builtin_openssl2/crypto/modes/ccm128.c +++ b/drivers/builtin_openssl2/crypto/modes/ccm128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -58,384 +58,422 @@ #endif #include <assert.h> -/* First you setup M and L parameters and pass the key schedule. - * This is called once per session setup... */ +/* + * First you setup M and L parameters and pass the key schedule. This is + * called once per session setup... + */ void CRYPTO_ccm128_init(CCM128_CONTEXT *ctx, - unsigned int M,unsigned int L,void *key,block128_f block) + unsigned int M, unsigned int L, void *key, + block128_f block) { - memset(ctx->nonce.c,0,sizeof(ctx->nonce.c)); - ctx->nonce.c[0] = ((u8)(L-1)&7) | (u8)(((M-2)/2)&7)<<3; - ctx->blocks = 0; - ctx->block = block; - ctx->key = key; + memset(ctx->nonce.c, 0, sizeof(ctx->nonce.c)); + ctx->nonce.c[0] = ((u8)(L - 1) & 7) | (u8)(((M - 2) / 2) & 7) << 3; + ctx->blocks = 0; + ctx->block = block; + ctx->key = key; } /* !!! Following interfaces are to be called *once* per packet !!! */ /* Then you setup per-message nonce and pass the length of the message */ int CRYPTO_ccm128_setiv(CCM128_CONTEXT *ctx, - const unsigned char *nonce,size_t nlen,size_t mlen) + const unsigned char *nonce, size_t nlen, size_t mlen) { - unsigned int L = ctx->nonce.c[0]&7; /* the L parameter */ + unsigned int L = ctx->nonce.c[0] & 7; /* the L parameter */ - if (nlen<(14-L)) return -1; /* nonce is too short */ + if (nlen < (14 - L)) + return -1; /* nonce is too short */ - if (sizeof(mlen)==8 && L>=3) { - ctx->nonce.c[8] = (u8)(mlen>>(56%(sizeof(mlen)*8))); - ctx->nonce.c[9] = (u8)(mlen>>(48%(sizeof(mlen)*8))); - ctx->nonce.c[10] = (u8)(mlen>>(40%(sizeof(mlen)*8))); - ctx->nonce.c[11] = (u8)(mlen>>(32%(sizeof(mlen)*8))); - } - else - ctx->nonce.u[1] = 0; + if (sizeof(mlen) == 8 && L >= 3) { + ctx->nonce.c[8] = (u8)(mlen >> (56 % (sizeof(mlen) * 8))); + ctx->nonce.c[9] = (u8)(mlen >> (48 % (sizeof(mlen) * 8))); + ctx->nonce.c[10] = (u8)(mlen >> (40 % (sizeof(mlen) * 8))); + ctx->nonce.c[11] = (u8)(mlen >> (32 % (sizeof(mlen) * 8))); + } else + ctx->nonce.u[1] = 0; - ctx->nonce.c[12] = (u8)(mlen>>24); - ctx->nonce.c[13] = (u8)(mlen>>16); - ctx->nonce.c[14] = (u8)(mlen>>8); - ctx->nonce.c[15] = (u8)mlen; + ctx->nonce.c[12] = (u8)(mlen >> 24); + ctx->nonce.c[13] = (u8)(mlen >> 16); + ctx->nonce.c[14] = (u8)(mlen >> 8); + ctx->nonce.c[15] = (u8)mlen; - ctx->nonce.c[0] &= ~0x40; /* clear Adata flag */ - memcpy(&ctx->nonce.c[1],nonce,14-L); + ctx->nonce.c[0] &= ~0x40; /* clear Adata flag */ + memcpy(&ctx->nonce.c[1], nonce, 14 - L); - return 0; + return 0; } /* Then you pass additional authentication data, this is optional */ void CRYPTO_ccm128_aad(CCM128_CONTEXT *ctx, - const unsigned char *aad,size_t alen) -{ unsigned int i; - block128_f block = ctx->block; - - if (alen==0) return; - - ctx->nonce.c[0] |= 0x40; /* set Adata flag */ - (*block)(ctx->nonce.c,ctx->cmac.c,ctx->key), - ctx->blocks++; - - if (alen<(0x10000-0x100)) { - ctx->cmac.c[0] ^= (u8)(alen>>8); - ctx->cmac.c[1] ^= (u8)alen; - i=2; - } - else if (sizeof(alen)==8 && alen>=(size_t)1<<(32%(sizeof(alen)*8))) { - ctx->cmac.c[0] ^= 0xFF; - ctx->cmac.c[1] ^= 0xFF; - ctx->cmac.c[2] ^= (u8)(alen>>(56%(sizeof(alen)*8))); - ctx->cmac.c[3] ^= (u8)(alen>>(48%(sizeof(alen)*8))); - ctx->cmac.c[4] ^= (u8)(alen>>(40%(sizeof(alen)*8))); - ctx->cmac.c[5] ^= (u8)(alen>>(32%(sizeof(alen)*8))); - ctx->cmac.c[6] ^= (u8)(alen>>24); - ctx->cmac.c[7] ^= (u8)(alen>>16); - ctx->cmac.c[8] ^= (u8)(alen>>8); - ctx->cmac.c[9] ^= (u8)alen; - i=10; - } - else { - ctx->cmac.c[0] ^= 0xFF; - ctx->cmac.c[1] ^= 0xFE; - ctx->cmac.c[2] ^= (u8)(alen>>24); - ctx->cmac.c[3] ^= (u8)(alen>>16); - ctx->cmac.c[4] ^= (u8)(alen>>8); - ctx->cmac.c[5] ^= (u8)alen; - i=6; - } - - do { - for(;i<16 && alen;++i,++aad,--alen) - ctx->cmac.c[i] ^= *aad; - (*block)(ctx->cmac.c,ctx->cmac.c,ctx->key), - ctx->blocks++; - i=0; - } while (alen); + const unsigned char *aad, size_t alen) +{ + unsigned int i; + block128_f block = ctx->block; + + if (alen == 0) + return; + + ctx->nonce.c[0] |= 0x40; /* set Adata flag */ + (*block) (ctx->nonce.c, ctx->cmac.c, ctx->key), ctx->blocks++; + + if (alen < (0x10000 - 0x100)) { + ctx->cmac.c[0] ^= (u8)(alen >> 8); + ctx->cmac.c[1] ^= (u8)alen; + i = 2; + } else if (sizeof(alen) == 8 + && alen >= (size_t)1 << (32 % (sizeof(alen) * 8))) { + ctx->cmac.c[0] ^= 0xFF; + ctx->cmac.c[1] ^= 0xFF; + ctx->cmac.c[2] ^= (u8)(alen >> (56 % (sizeof(alen) * 8))); + ctx->cmac.c[3] ^= (u8)(alen >> (48 % (sizeof(alen) * 8))); + ctx->cmac.c[4] ^= (u8)(alen >> (40 % (sizeof(alen) * 8))); + ctx->cmac.c[5] ^= (u8)(alen >> (32 % (sizeof(alen) * 8))); + ctx->cmac.c[6] ^= (u8)(alen >> 24); + ctx->cmac.c[7] ^= (u8)(alen >> 16); + ctx->cmac.c[8] ^= (u8)(alen >> 8); + ctx->cmac.c[9] ^= (u8)alen; + i = 10; + } else { + ctx->cmac.c[0] ^= 0xFF; + ctx->cmac.c[1] ^= 0xFE; + ctx->cmac.c[2] ^= (u8)(alen >> 24); + ctx->cmac.c[3] ^= (u8)(alen >> 16); + ctx->cmac.c[4] ^= (u8)(alen >> 8); + ctx->cmac.c[5] ^= (u8)alen; + i = 6; + } + + do { + for (; i < 16 && alen; ++i, ++aad, --alen) + ctx->cmac.c[i] ^= *aad; + (*block) (ctx->cmac.c, ctx->cmac.c, ctx->key), ctx->blocks++; + i = 0; + } while (alen); } /* Finally you encrypt or decrypt the message */ -/* counter part of nonce may not be larger than L*8 bits, - * L is not larger than 8, therefore 64-bit counter... */ -static void ctr64_inc(unsigned char *counter) { - unsigned int n=8; - u8 c; - - counter += 8; - do { - --n; - c = counter[n]; - ++c; - counter[n] = c; - if (c) return; - } while (n); +/* + * counter part of nonce may not be larger than L*8 bits, L is not larger + * than 8, therefore 64-bit counter... + */ +static void ctr64_inc(unsigned char *counter) +{ + unsigned int n = 8; + u8 c; + + counter += 8; + do { + --n; + c = counter[n]; + ++c; + counter[n] = c; + if (c) + return; + } while (n); } int CRYPTO_ccm128_encrypt(CCM128_CONTEXT *ctx, - const unsigned char *inp, unsigned char *out, - size_t len) + const unsigned char *inp, unsigned char *out, + size_t len) { - size_t n; - unsigned int i,L; - unsigned char flags0 = ctx->nonce.c[0]; - block128_f block = ctx->block; - void * key = ctx->key; - union { u64 u[2]; u8 c[16]; } scratch; - - if (!(flags0&0x40)) - (*block)(ctx->nonce.c,ctx->cmac.c,key), - ctx->blocks++; - - ctx->nonce.c[0] = L = flags0&7; - for (n=0,i=15-L;i<15;++i) { - n |= ctx->nonce.c[i]; - ctx->nonce.c[i]=0; - n <<= 8; - } - n |= ctx->nonce.c[15]; /* reconstructed length */ - ctx->nonce.c[15]=1; - - if (n!=len) return -1; /* length mismatch */ - - ctx->blocks += ((len+15)>>3)|1; - if (ctx->blocks > (U64(1)<<61)) return -2; /* too much data */ - - while (len>=16) { + size_t n; + unsigned int i, L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void *key = ctx->key; + union { + u64 u[2]; + u8 c[16]; + } scratch; + + if (!(flags0 & 0x40)) + (*block) (ctx->nonce.c, ctx->cmac.c, key), ctx->blocks++; + + ctx->nonce.c[0] = L = flags0 & 7; + for (n = 0, i = 15 - L; i < 15; ++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i] = 0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15] = 1; + + if (n != len) + return -1; /* length mismatch */ + + ctx->blocks += ((len + 15) >> 3) | 1; + if (ctx->blocks > (U64(1) << 61)) + return -2; /* too much data */ + + while (len >= 16) { #if defined(STRICT_ALIGNMENT) - union { u64 u[2]; u8 c[16]; } temp; - - memcpy (temp.c,inp,16); - ctx->cmac.u[0] ^= temp.u[0]; - ctx->cmac.u[1] ^= temp.u[1]; + union { + u64 u[2]; + u8 c[16]; + } temp; + + memcpy(temp.c, inp, 16); + ctx->cmac.u[0] ^= temp.u[0]; + ctx->cmac.u[1] ^= temp.u[1]; #else - ctx->cmac.u[0] ^= ((u64*)inp)[0]; - ctx->cmac.u[1] ^= ((u64*)inp)[1]; + ctx->cmac.u[0] ^= ((u64 *)inp)[0]; + ctx->cmac.u[1] ^= ((u64 *)inp)[1]; #endif - (*block)(ctx->cmac.c,ctx->cmac.c,key); - (*block)(ctx->nonce.c,scratch.c,key); - ctr64_inc(ctx->nonce.c); + (*block) (ctx->cmac.c, ctx->cmac.c, key); + (*block) (ctx->nonce.c, scratch.c, key); + ctr64_inc(ctx->nonce.c); #if defined(STRICT_ALIGNMENT) - temp.u[0] ^= scratch.u[0]; - temp.u[1] ^= scratch.u[1]; - memcpy(out,temp.c,16); + temp.u[0] ^= scratch.u[0]; + temp.u[1] ^= scratch.u[1]; + memcpy(out, temp.c, 16); #else - ((u64*)out)[0] = scratch.u[0]^((u64*)inp)[0]; - ((u64*)out)[1] = scratch.u[1]^((u64*)inp)[1]; + ((u64 *)out)[0] = scratch.u[0] ^ ((u64 *)inp)[0]; + ((u64 *)out)[1] = scratch.u[1] ^ ((u64 *)inp)[1]; #endif - inp += 16; - out += 16; - len -= 16; - } - - if (len) { - for (i=0; i<len; ++i) ctx->cmac.c[i] ^= inp[i]; - (*block)(ctx->cmac.c,ctx->cmac.c,key); - (*block)(ctx->nonce.c,scratch.c,key); - for (i=0; i<len; ++i) out[i] = scratch.c[i]^inp[i]; - } - - for (i=15-L;i<16;++i) - ctx->nonce.c[i]=0; - - (*block)(ctx->nonce.c,scratch.c,key); - ctx->cmac.u[0] ^= scratch.u[0]; - ctx->cmac.u[1] ^= scratch.u[1]; - - ctx->nonce.c[0] = flags0; - - return 0; + inp += 16; + out += 16; + len -= 16; + } + + if (len) { + for (i = 0; i < len; ++i) + ctx->cmac.c[i] ^= inp[i]; + (*block) (ctx->cmac.c, ctx->cmac.c, key); + (*block) (ctx->nonce.c, scratch.c, key); + for (i = 0; i < len; ++i) + out[i] = scratch.c[i] ^ inp[i]; + } + + for (i = 15 - L; i < 16; ++i) + ctx->nonce.c[i] = 0; + + (*block) (ctx->nonce.c, scratch.c, key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; } int CRYPTO_ccm128_decrypt(CCM128_CONTEXT *ctx, - const unsigned char *inp, unsigned char *out, - size_t len) + const unsigned char *inp, unsigned char *out, + size_t len) { - size_t n; - unsigned int i,L; - unsigned char flags0 = ctx->nonce.c[0]; - block128_f block = ctx->block; - void * key = ctx->key; - union { u64 u[2]; u8 c[16]; } scratch; - - if (!(flags0&0x40)) - (*block)(ctx->nonce.c,ctx->cmac.c,key); - - ctx->nonce.c[0] = L = flags0&7; - for (n=0,i=15-L;i<15;++i) { - n |= ctx->nonce.c[i]; - ctx->nonce.c[i]=0; - n <<= 8; - } - n |= ctx->nonce.c[15]; /* reconstructed length */ - ctx->nonce.c[15]=1; - - if (n!=len) return -1; - - while (len>=16) { + size_t n; + unsigned int i, L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void *key = ctx->key; + union { + u64 u[2]; + u8 c[16]; + } scratch; + + if (!(flags0 & 0x40)) + (*block) (ctx->nonce.c, ctx->cmac.c, key); + + ctx->nonce.c[0] = L = flags0 & 7; + for (n = 0, i = 15 - L; i < 15; ++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i] = 0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15] = 1; + + if (n != len) + return -1; + + while (len >= 16) { #if defined(STRICT_ALIGNMENT) - union { u64 u[2]; u8 c[16]; } temp; + union { + u64 u[2]; + u8 c[16]; + } temp; #endif - (*block)(ctx->nonce.c,scratch.c,key); - ctr64_inc(ctx->nonce.c); + (*block) (ctx->nonce.c, scratch.c, key); + ctr64_inc(ctx->nonce.c); #if defined(STRICT_ALIGNMENT) - memcpy (temp.c,inp,16); - ctx->cmac.u[0] ^= (scratch.u[0] ^= temp.u[0]); - ctx->cmac.u[1] ^= (scratch.u[1] ^= temp.u[1]); - memcpy (out,scratch.c,16); + memcpy(temp.c, inp, 16); + ctx->cmac.u[0] ^= (scratch.u[0] ^= temp.u[0]); + ctx->cmac.u[1] ^= (scratch.u[1] ^= temp.u[1]); + memcpy(out, scratch.c, 16); #else - ctx->cmac.u[0] ^= (((u64*)out)[0] = scratch.u[0]^((u64*)inp)[0]); - ctx->cmac.u[1] ^= (((u64*)out)[1] = scratch.u[1]^((u64*)inp)[1]); + ctx->cmac.u[0] ^= (((u64 *)out)[0] = scratch.u[0] ^ ((u64 *)inp)[0]); + ctx->cmac.u[1] ^= (((u64 *)out)[1] = scratch.u[1] ^ ((u64 *)inp)[1]); #endif - (*block)(ctx->cmac.c,ctx->cmac.c,key); + (*block) (ctx->cmac.c, ctx->cmac.c, key); - inp += 16; - out += 16; - len -= 16; - } + inp += 16; + out += 16; + len -= 16; + } - if (len) { - (*block)(ctx->nonce.c,scratch.c,key); - for (i=0; i<len; ++i) - ctx->cmac.c[i] ^= (out[i] = scratch.c[i]^inp[i]); - (*block)(ctx->cmac.c,ctx->cmac.c,key); - } + if (len) { + (*block) (ctx->nonce.c, scratch.c, key); + for (i = 0; i < len; ++i) + ctx->cmac.c[i] ^= (out[i] = scratch.c[i] ^ inp[i]); + (*block) (ctx->cmac.c, ctx->cmac.c, key); + } - for (i=15-L;i<16;++i) - ctx->nonce.c[i]=0; + for (i = 15 - L; i < 16; ++i) + ctx->nonce.c[i] = 0; - (*block)(ctx->nonce.c,scratch.c,key); - ctx->cmac.u[0] ^= scratch.u[0]; - ctx->cmac.u[1] ^= scratch.u[1]; + (*block) (ctx->nonce.c, scratch.c, key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; - ctx->nonce.c[0] = flags0; + ctx->nonce.c[0] = flags0; - return 0; + return 0; } -static void ctr64_add (unsigned char *counter,size_t inc) -{ size_t n=8, val=0; - - counter += 8; - do { - --n; - val += counter[n] + (inc&0xff); - counter[n] = (unsigned char)val; - val >>= 8; /* carry bit */ - inc >>= 8; - } while(n && (inc || val)); +static void ctr64_add(unsigned char *counter, size_t inc) +{ + size_t n = 8, val = 0; + + counter += 8; + do { + --n; + val += counter[n] + (inc & 0xff); + counter[n] = (unsigned char)val; + val >>= 8; /* carry bit */ + inc >>= 8; + } while (n && (inc || val)); } int CRYPTO_ccm128_encrypt_ccm64(CCM128_CONTEXT *ctx, - const unsigned char *inp, unsigned char *out, - size_t len,ccm128_f stream) + const unsigned char *inp, unsigned char *out, + size_t len, ccm128_f stream) { - size_t n; - unsigned int i,L; - unsigned char flags0 = ctx->nonce.c[0]; - block128_f block = ctx->block; - void * key = ctx->key; - union { u64 u[2]; u8 c[16]; } scratch; - - if (!(flags0&0x40)) - (*block)(ctx->nonce.c,ctx->cmac.c,key), - ctx->blocks++; - - ctx->nonce.c[0] = L = flags0&7; - for (n=0,i=15-L;i<15;++i) { - n |= ctx->nonce.c[i]; - ctx->nonce.c[i]=0; - n <<= 8; - } - n |= ctx->nonce.c[15]; /* reconstructed length */ - ctx->nonce.c[15]=1; - - if (n!=len) return -1; /* length mismatch */ - - ctx->blocks += ((len+15)>>3)|1; - if (ctx->blocks > (U64(1)<<61)) return -2; /* too much data */ - - if ((n=len/16)) { - (*stream)(inp,out,n,key,ctx->nonce.c,ctx->cmac.c); - n *= 16; - inp += n; - out += n; - len -= n; - if (len) ctr64_add(ctx->nonce.c,n/16); - } - - if (len) { - for (i=0; i<len; ++i) ctx->cmac.c[i] ^= inp[i]; - (*block)(ctx->cmac.c,ctx->cmac.c,key); - (*block)(ctx->nonce.c,scratch.c,key); - for (i=0; i<len; ++i) out[i] = scratch.c[i]^inp[i]; - } - - for (i=15-L;i<16;++i) - ctx->nonce.c[i]=0; - - (*block)(ctx->nonce.c,scratch.c,key); - ctx->cmac.u[0] ^= scratch.u[0]; - ctx->cmac.u[1] ^= scratch.u[1]; - - ctx->nonce.c[0] = flags0; - - return 0; + size_t n; + unsigned int i, L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void *key = ctx->key; + union { + u64 u[2]; + u8 c[16]; + } scratch; + + if (!(flags0 & 0x40)) + (*block) (ctx->nonce.c, ctx->cmac.c, key), ctx->blocks++; + + ctx->nonce.c[0] = L = flags0 & 7; + for (n = 0, i = 15 - L; i < 15; ++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i] = 0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15] = 1; + + if (n != len) + return -1; /* length mismatch */ + + ctx->blocks += ((len + 15) >> 3) | 1; + if (ctx->blocks > (U64(1) << 61)) + return -2; /* too much data */ + + if ((n = len / 16)) { + (*stream) (inp, out, n, key, ctx->nonce.c, ctx->cmac.c); + n *= 16; + inp += n; + out += n; + len -= n; + if (len) + ctr64_add(ctx->nonce.c, n / 16); + } + + if (len) { + for (i = 0; i < len; ++i) + ctx->cmac.c[i] ^= inp[i]; + (*block) (ctx->cmac.c, ctx->cmac.c, key); + (*block) (ctx->nonce.c, scratch.c, key); + for (i = 0; i < len; ++i) + out[i] = scratch.c[i] ^ inp[i]; + } + + for (i = 15 - L; i < 16; ++i) + ctx->nonce.c[i] = 0; + + (*block) (ctx->nonce.c, scratch.c, key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; } int CRYPTO_ccm128_decrypt_ccm64(CCM128_CONTEXT *ctx, - const unsigned char *inp, unsigned char *out, - size_t len,ccm128_f stream) + const unsigned char *inp, unsigned char *out, + size_t len, ccm128_f stream) { - size_t n; - unsigned int i,L; - unsigned char flags0 = ctx->nonce.c[0]; - block128_f block = ctx->block; - void * key = ctx->key; - union { u64 u[2]; u8 c[16]; } scratch; - - if (!(flags0&0x40)) - (*block)(ctx->nonce.c,ctx->cmac.c,key); - - ctx->nonce.c[0] = L = flags0&7; - for (n=0,i=15-L;i<15;++i) { - n |= ctx->nonce.c[i]; - ctx->nonce.c[i]=0; - n <<= 8; - } - n |= ctx->nonce.c[15]; /* reconstructed length */ - ctx->nonce.c[15]=1; - - if (n!=len) return -1; - - if ((n=len/16)) { - (*stream)(inp,out,n,key,ctx->nonce.c,ctx->cmac.c); - n *= 16; - inp += n; - out += n; - len -= n; - if (len) ctr64_add(ctx->nonce.c,n/16); - } - - if (len) { - (*block)(ctx->nonce.c,scratch.c,key); - for (i=0; i<len; ++i) - ctx->cmac.c[i] ^= (out[i] = scratch.c[i]^inp[i]); - (*block)(ctx->cmac.c,ctx->cmac.c,key); - } - - for (i=15-L;i<16;++i) - ctx->nonce.c[i]=0; - - (*block)(ctx->nonce.c,scratch.c,key); - ctx->cmac.u[0] ^= scratch.u[0]; - ctx->cmac.u[1] ^= scratch.u[1]; - - ctx->nonce.c[0] = flags0; - - return 0; + size_t n; + unsigned int i, L; + unsigned char flags0 = ctx->nonce.c[0]; + block128_f block = ctx->block; + void *key = ctx->key; + union { + u64 u[2]; + u8 c[16]; + } scratch; + + if (!(flags0 & 0x40)) + (*block) (ctx->nonce.c, ctx->cmac.c, key); + + ctx->nonce.c[0] = L = flags0 & 7; + for (n = 0, i = 15 - L; i < 15; ++i) { + n |= ctx->nonce.c[i]; + ctx->nonce.c[i] = 0; + n <<= 8; + } + n |= ctx->nonce.c[15]; /* reconstructed length */ + ctx->nonce.c[15] = 1; + + if (n != len) + return -1; + + if ((n = len / 16)) { + (*stream) (inp, out, n, key, ctx->nonce.c, ctx->cmac.c); + n *= 16; + inp += n; + out += n; + len -= n; + if (len) + ctr64_add(ctx->nonce.c, n / 16); + } + + if (len) { + (*block) (ctx->nonce.c, scratch.c, key); + for (i = 0; i < len; ++i) + ctx->cmac.c[i] ^= (out[i] = scratch.c[i] ^ inp[i]); + (*block) (ctx->cmac.c, ctx->cmac.c, key); + } + + for (i = 15 - L; i < 16; ++i) + ctx->nonce.c[i] = 0; + + (*block) (ctx->nonce.c, scratch.c, key); + ctx->cmac.u[0] ^= scratch.u[0]; + ctx->cmac.u[1] ^= scratch.u[1]; + + ctx->nonce.c[0] = flags0; + + return 0; } -size_t CRYPTO_ccm128_tag(CCM128_CONTEXT *ctx,unsigned char *tag,size_t len) -{ unsigned int M = (ctx->nonce.c[0]>>3)&7; /* the M parameter */ - - M *= 2; M += 2; - if (len<M) return 0; - memcpy(tag,ctx->cmac.c,M); - return M; +size_t CRYPTO_ccm128_tag(CCM128_CONTEXT *ctx, unsigned char *tag, size_t len) +{ + unsigned int M = (ctx->nonce.c[0] >> 3) & 7; /* the M parameter */ + + M *= 2; + M += 2; + if (len < M) + return 0; + memcpy(tag, ctx->cmac.c, M); + return M; } diff --git a/drivers/builtin_openssl2/crypto/modes/cfb128.c b/drivers/builtin_openssl2/crypto/modes/cfb128.c index 4e6f5d35e1..d4ecbd08ee 100644 --- a/drivers/builtin_openssl2/crypto/modes/cfb128.c +++ b/drivers/builtin_openssl2/crypto/modes/cfb128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -59,14 +59,15 @@ #endif #include <assert.h> -/* The input and output encrypted as though 128bit cfb mode is being - * used. The extra state information to record how much of the - * 128bit block we have used is contained in *num; +/* + * The input and output encrypted as though 128bit cfb mode is being used. + * The extra state information to record how much of the 128bit block we have + * used is contained in *num; */ void CRYPTO_cfb128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], int *num, - int enc, block128_f block) + size_t len, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) { unsigned int n; size_t l = 0; @@ -77,166 +78,177 @@ void CRYPTO_cfb128_encrypt(const unsigned char *in, unsigned char *out, if (enc) { #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - while (n && len) { - *(out++) = ivec[n] ^= *(in++); - --len; - n = (n+1) % 16; - } -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) - break; -#endif - while (len>=16) { - (*block)(ivec, ivec, key); - for (; n<16; n+=sizeof(size_t)) { - *(size_t*)(out+n) = - *(size_t*)(ivec+n) ^= *(size_t*)(in+n); - } - len -= 16; - out += 16; - in += 16; - n = 0; - } - if (len) { - (*block)(ivec, ivec, key); - while (len--) { - out[n] = ivec[n] ^= in[n]; - ++n; - } - } - *num = n; - return; - } while (0); - /* the rest would be commonly eliminated by x86* compiler */ + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + while (n && len) { + *(out++) = ivec[n] ^= *(in++); + --len; + n = (n + 1) % 16; + } +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out | (size_t)ivec) % + sizeof(size_t) != 0) + break; +# endif + while (len >= 16) { + (*block) (ivec, ivec, key); + for (; n < 16; n += sizeof(size_t)) { + *(size_t *)(out + n) = + *(size_t *)(ivec + n) ^= *(size_t *)(in + n); + } + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block) (ivec, ivec, key); + while (len--) { + out[n] = ivec[n] ^= in[n]; + ++n; + } + } + *num = n; + return; + } while (0); + } + /* the rest would be commonly eliminated by x86* compiler */ #endif - while (l<len) { - if (n == 0) { - (*block)(ivec, ivec, key); - } - out[l] = ivec[n] ^= in[l]; - ++l; - n = (n+1) % 16; - } - *num = n; + while (l < len) { + if (n == 0) { + (*block) (ivec, ivec, key); + } + out[l] = ivec[n] ^= in[l]; + ++l; + n = (n + 1) % 16; + } + *num = n; } else { #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - while (n && len) { - unsigned char c; - *(out++) = ivec[n] ^ (c = *(in++)); ivec[n] = c; - --len; - n = (n+1) % 16; - } -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) - break; -#endif - while (len>=16) { - (*block)(ivec, ivec, key); - for (; n<16; n+=sizeof(size_t)) { - size_t t = *(size_t*)(in+n); - *(size_t*)(out+n) = *(size_t*)(ivec+n) ^ t; - *(size_t*)(ivec+n) = t; - } - len -= 16; - out += 16; - in += 16; - n = 0; - } - if (len) { - (*block)(ivec, ivec, key); - while (len--) { - unsigned char c; - out[n] = ivec[n] ^ (c = in[n]); ivec[n] = c; - ++n; - } - } - *num = n; - return; - } while (0); - /* the rest would be commonly eliminated by x86* compiler */ + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + while (n && len) { + unsigned char c; + *(out++) = ivec[n] ^ (c = *(in++)); + ivec[n] = c; + --len; + n = (n + 1) % 16; + } +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out | (size_t)ivec) % + sizeof(size_t) != 0) + break; +# endif + while (len >= 16) { + (*block) (ivec, ivec, key); + for (; n < 16; n += sizeof(size_t)) { + size_t t = *(size_t *)(in + n); + *(size_t *)(out + n) = *(size_t *)(ivec + n) ^ t; + *(size_t *)(ivec + n) = t; + } + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block) (ivec, ivec, key); + while (len--) { + unsigned char c; + out[n] = ivec[n] ^ (c = in[n]); + ivec[n] = c; + ++n; + } + } + *num = n; + return; + } while (0); + } + /* the rest would be commonly eliminated by x86* compiler */ #endif - while (l<len) { - unsigned char c; - if (n == 0) { - (*block)(ivec, ivec, key); - } - out[l] = ivec[n] ^ (c = in[l]); ivec[n] = c; - ++l; - n = (n+1) % 16; - } - *num=n; + while (l < len) { + unsigned char c; + if (n == 0) { + (*block) (ivec, ivec, key); + } + out[l] = ivec[n] ^ (c = in[l]); + ivec[n] = c; + ++l; + n = (n + 1) % 16; + } + *num = n; } } -/* This expects a single block of size nbits for both in and out. Note that - it corrupts any extra bits in the last byte of out */ -static void cfbr_encrypt_block(const unsigned char *in,unsigned char *out, - int nbits,const void *key, - unsigned char ivec[16],int enc, - block128_f block) +/* + * This expects a single block of size nbits for both in and out. Note that + * it corrupts any extra bits in the last byte of out + */ +static void cfbr_encrypt_block(const unsigned char *in, unsigned char *out, + int nbits, const void *key, + unsigned char ivec[16], int enc, + block128_f block) { - int n,rem,num; - unsigned char ovec[16*2 + 1]; /* +1 because we dererefence (but don't use) one byte off the end */ - - if (nbits<=0 || nbits>128) return; - - /* fill in the first half of the new IV with the current IV */ - memcpy(ovec,ivec,16); - /* construct the new IV */ - (*block)(ivec,ivec,key); - num = (nbits+7)/8; - if (enc) /* encrypt the input */ - for(n=0 ; n < num ; ++n) - out[n] = (ovec[16+n] = in[n] ^ ivec[n]); - else /* decrypt the input */ - for(n=0 ; n < num ; ++n) - out[n] = (ovec[16+n] = in[n]) ^ ivec[n]; - /* shift ovec left... */ - rem = nbits%8; - num = nbits/8; - if(rem==0) - memcpy(ivec,ovec+num,16); - else - for(n=0 ; n < 16 ; ++n) - ivec[n] = ovec[n+num]<<rem | ovec[n+num+1]>>(8-rem); + int n, rem, num; + unsigned char ovec[16 * 2 + 1]; /* +1 because we dererefence (but don't + * use) one byte off the end */ + + if (nbits <= 0 || nbits > 128) + return; + + /* fill in the first half of the new IV with the current IV */ + memcpy(ovec, ivec, 16); + /* construct the new IV */ + (*block) (ivec, ivec, key); + num = (nbits + 7) / 8; + if (enc) /* encrypt the input */ + for (n = 0; n < num; ++n) + out[n] = (ovec[16 + n] = in[n] ^ ivec[n]); + else /* decrypt the input */ + for (n = 0; n < num; ++n) + out[n] = (ovec[16 + n] = in[n]) ^ ivec[n]; + /* shift ovec left... */ + rem = nbits % 8; + num = nbits / 8; + if (rem == 0) + memcpy(ivec, ovec + num, 16); + else + for (n = 0; n < 16; ++n) + ivec[n] = ovec[n + num] << rem | ovec[n + num + 1] >> (8 - rem); /* it is not necessary to cleanse ovec, since the IV is not secret */ } /* N.B. This expects the input to be packed, MS bit first */ void CRYPTO_cfb128_1_encrypt(const unsigned char *in, unsigned char *out, - size_t bits, const void *key, - unsigned char ivec[16], int *num, - int enc, block128_f block) + size_t bits, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) { size_t n; - unsigned char c[1],d[1]; + unsigned char c[1], d[1]; assert(in && out && key && ivec && num); assert(*num == 0); - for(n=0 ; n<bits ; ++n) - { - c[0]=(in[n/8]&(1 << (7-n%8))) ? 0x80 : 0; - cfbr_encrypt_block(c,d,1,key,ivec,enc,block); - out[n/8]=(out[n/8]&~(1 << (unsigned int)(7-n%8))) | - ((d[0]&0x80) >> (unsigned int)(n%8)); - } + for (n = 0; n < bits; ++n) { + c[0] = (in[n / 8] & (1 << (7 - n % 8))) ? 0x80 : 0; + cfbr_encrypt_block(c, d, 1, key, ivec, enc, block); + out[n / 8] = (out[n / 8] & ~(1 << (unsigned int)(7 - n % 8))) | + ((d[0] & 0x80) >> (unsigned int)(n % 8)); + } } void CRYPTO_cfb128_8_encrypt(const unsigned char *in, unsigned char *out, - size_t length, const void *key, - unsigned char ivec[16], int *num, - int enc, block128_f block) + size_t length, const void *key, + unsigned char ivec[16], int *num, + int enc, block128_f block) { size_t n; assert(in && out && key && ivec && num); assert(*num == 0); - for(n=0 ; n<length ; ++n) - cfbr_encrypt_block(&in[n],&out[n],8,key,ivec,enc,block); + for (n = 0; n < length; ++n) + cfbr_encrypt_block(&in[n], &out[n], 8, key, ivec, enc, block); } - diff --git a/drivers/builtin_openssl2/crypto/modes/ctr128.c b/drivers/builtin_openssl2/crypto/modes/ctr128.c index ee642c5863..bcafd6b6bf 100644 --- a/drivers/builtin_openssl2/crypto/modes/ctr128.c +++ b/drivers/builtin_openssl2/crypto/modes/ctr128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -59,194 +59,205 @@ #endif #include <assert.h> -/* NOTE: the IV/counter CTR mode is big-endian. The code itself - * is endian-neutral. */ +/* + * NOTE: the IV/counter CTR mode is big-endian. The code itself is + * endian-neutral. + */ /* increment counter (128-bit int) by 1 */ -static void ctr128_inc(unsigned char *counter) { - u32 n=16; - u8 c; - - do { - --n; - c = counter[n]; - ++c; - counter[n] = c; - if (c) return; - } while (n); +static void ctr128_inc(unsigned char *counter) +{ + u32 n = 16, c = 1; + + do { + --n; + c += counter[n]; + counter[n] = (u8)c; + c >>= 8; + } while (n); } #if !defined(OPENSSL_SMALL_FOOTPRINT) -static void ctr128_inc_aligned(unsigned char *counter) { - size_t *data,c,n; - const union { long one; char little; } is_endian = {1}; - - if (is_endian.little) { - ctr128_inc(counter); - return; - } - - data = (size_t *)counter; - n = 16/sizeof(size_t); - do { - --n; - c = data[n]; - ++c; - data[n] = c; - if (c) return; - } while (n); +static void ctr128_inc_aligned(unsigned char *counter) +{ + size_t *data, c, d, n; + const union { + long one; + char little; + } is_endian = { + 1 + }; + + if (is_endian.little || ((size_t)counter % sizeof(size_t)) != 0) { + ctr128_inc(counter); + return; + } + + data = (size_t *)counter; + c = 1; + n = 16 / sizeof(size_t); + do { + --n; + d = data[n] += c; + /* did addition carry? */ + c = ((d - c) ^ d) >> (sizeof(size_t) * 8 - 1); + } while (n); } #endif -/* The input encrypted as though 128bit counter mode is being - * used. The extra state information to record how much of the - * 128bit block we have used is contained in *num, and the - * encrypted counter is kept in ecount_buf. Both *num and - * ecount_buf must be initialised with zeros before the first - * call to CRYPTO_ctr128_encrypt(). - * - * This algorithm assumes that the counter is in the x lower bits - * of the IV (ivec), and that the application has full control over - * overflow and the rest of the IV. This implementation takes NO - * responsability for checking that the counter doesn't overflow - * into the rest of the IV when incremented. +/* + * The input encrypted as though 128bit counter mode is being used. The + * extra state information to record how much of the 128bit block we have + * used is contained in *num, and the encrypted counter is kept in + * ecount_buf. Both *num and ecount_buf must be initialised with zeros + * before the first call to CRYPTO_ctr128_encrypt(). This algorithm assumes + * that the counter is in the x lower bits of the IV (ivec), and that the + * application has full control over overflow and the rest of the IV. This + * implementation takes NO responsability for checking that the counter + * doesn't overflow into the rest of the IV when incremented. */ void CRYPTO_ctr128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], unsigned char ecount_buf[16], - unsigned int *num, block128_f block) + size_t len, const void *key, + unsigned char ivec[16], + unsigned char ecount_buf[16], unsigned int *num, + block128_f block) { - unsigned int n; - size_t l=0; + unsigned int n; + size_t l = 0; - assert(in && out && key && ecount_buf && num); - assert(*num < 16); + assert(in && out && key && ecount_buf && num); + assert(*num < 16); - n = *num; + n = *num; #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - while (n && len) { - *(out++) = *(in++) ^ ecount_buf[n]; - --len; - n = (n+1) % 16; - } - -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) - break; -#endif - while (len>=16) { - (*block)(ivec, ecount_buf, key); - ctr128_inc_aligned(ivec); - for (; n<16; n+=sizeof(size_t)) - *(size_t *)(out+n) = - *(size_t *)(in+n) ^ *(size_t *)(ecount_buf+n); - len -= 16; - out += 16; - in += 16; - n = 0; - } - if (len) { - (*block)(ivec, ecount_buf, key); - ctr128_inc_aligned(ivec); - while (len--) { - out[n] = in[n] ^ ecount_buf[n]; - ++n; - } - } - *num = n; - return; - } while(0); - /* the rest would be commonly eliminated by x86* compiler */ + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + while (n && len) { + *(out++) = *(in++) ^ ecount_buf[n]; + --len; + n = (n + 1) % 16; + } + +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out | (size_t)ecount_buf) + % sizeof(size_t) != 0) + break; +# endif + while (len >= 16) { + (*block) (ivec, ecount_buf, key); + ctr128_inc_aligned(ivec); + for (n = 0; n < 16; n += sizeof(size_t)) + *(size_t *)(out + n) = + *(size_t *)(in + n) ^ *(size_t *)(ecount_buf + n); + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block) (ivec, ecount_buf, key); + ctr128_inc_aligned(ivec); + while (len--) { + out[n] = in[n] ^ ecount_buf[n]; + ++n; + } + } + *num = n; + return; + } while (0); + } + /* the rest would be commonly eliminated by x86* compiler */ #endif - while (l<len) { - if (n==0) { - (*block)(ivec, ecount_buf, key); - ctr128_inc(ivec); - } - out[l] = in[l] ^ ecount_buf[n]; - ++l; - n = (n+1) % 16; - } - - *num=n; + while (l < len) { + if (n == 0) { + (*block) (ivec, ecount_buf, key); + ctr128_inc(ivec); + } + out[l] = in[l] ^ ecount_buf[n]; + ++l; + n = (n + 1) % 16; + } + + *num = n; } /* increment upper 96 bits of 128-bit counter by 1 */ -static void ctr96_inc(unsigned char *counter) { - u32 n=12; - u8 c; - - do { - --n; - c = counter[n]; - ++c; - counter[n] = c; - if (c) return; - } while (n); +static void ctr96_inc(unsigned char *counter) +{ + u32 n = 12, c = 1; + + do { + --n; + c += counter[n]; + counter[n] = (u8)c; + c >>= 8; + } while (n); } void CRYPTO_ctr128_encrypt_ctr32(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], unsigned char ecount_buf[16], - unsigned int *num, ctr128_f func) + size_t len, const void *key, + unsigned char ivec[16], + unsigned char ecount_buf[16], + unsigned int *num, ctr128_f func) { - unsigned int n,ctr32; - - assert(in && out && key && ecount_buf && num); - assert(*num < 16); - - n = *num; - - while (n && len) { - *(out++) = *(in++) ^ ecount_buf[n]; - --len; - n = (n+1) % 16; - } - - ctr32 = GETU32(ivec+12); - while (len>=16) { - size_t blocks = len/16; - /* - * 1<<28 is just a not-so-small yet not-so-large number... - * Below condition is practically never met, but it has to - * be checked for code correctness. - */ - if (sizeof(size_t)>sizeof(unsigned int) && blocks>(1U<<28)) - blocks = (1U<<28); - /* - * As (*func) operates on 32-bit counter, caller - * has to handle overflow. 'if' below detects the - * overflow, which is then handled by limiting the - * amount of blocks to the exact overflow point... - */ - ctr32 += (u32)blocks; - if (ctr32 < blocks) { - blocks -= ctr32; - ctr32 = 0; - } - (*func)(in,out,blocks,key,ivec); - /* (*ctr) does not update ivec, caller does: */ - PUTU32(ivec+12,ctr32); - /* ... overflow was detected, propogate carry. */ - if (ctr32 == 0) ctr96_inc(ivec); - blocks *= 16; - len -= blocks; - out += blocks; - in += blocks; - } - if (len) { - memset(ecount_buf,0,16); - (*func)(ecount_buf,ecount_buf,1,key,ivec); - ++ctr32; - PUTU32(ivec+12,ctr32); - if (ctr32 == 0) ctr96_inc(ivec); - while (len--) { - out[n] = in[n] ^ ecount_buf[n]; - ++n; - } - } - - *num=n; + unsigned int n, ctr32; + + assert(in && out && key && ecount_buf && num); + assert(*num < 16); + + n = *num; + + while (n && len) { + *(out++) = *(in++) ^ ecount_buf[n]; + --len; + n = (n + 1) % 16; + } + + ctr32 = GETU32(ivec + 12); + while (len >= 16) { + size_t blocks = len / 16; + /* + * 1<<28 is just a not-so-small yet not-so-large number... + * Below condition is practically never met, but it has to + * be checked for code correctness. + */ + if (sizeof(size_t) > sizeof(unsigned int) && blocks > (1U << 28)) + blocks = (1U << 28); + /* + * As (*func) operates on 32-bit counter, caller + * has to handle overflow. 'if' below detects the + * overflow, which is then handled by limiting the + * amount of blocks to the exact overflow point... + */ + ctr32 += (u32)blocks; + if (ctr32 < blocks) { + blocks -= ctr32; + ctr32 = 0; + } + (*func) (in, out, blocks, key, ivec); + /* (*ctr) does not update ivec, caller does: */ + PUTU32(ivec + 12, ctr32); + /* ... overflow was detected, propogate carry. */ + if (ctr32 == 0) + ctr96_inc(ivec); + blocks *= 16; + len -= blocks; + out += blocks; + in += blocks; + } + if (len) { + memset(ecount_buf, 0, 16); + (*func) (ecount_buf, ecount_buf, 1, key, ivec); + ++ctr32; + PUTU32(ivec + 12, ctr32); + if (ctr32 == 0) + ctr96_inc(ivec); + while (len--) { + out[n] = in[n] ^ ecount_buf[n]; + ++n; + } + } + + *num = n; } diff --git a/drivers/builtin_openssl2/crypto/modes/cts128.c b/drivers/builtin_openssl2/crypto/modes/cts128.c index 2d583de6f6..137be595a1 100644 --- a/drivers/builtin_openssl2/crypto/modes/cts128.c +++ b/drivers/builtin_openssl2/crypto/modes/cts128.c @@ -29,425 +29,516 @@ * compliant with the NIST proposal, both extending CBC mode. */ -size_t CRYPTO_cts128_encrypt_block(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) -{ size_t residue, n; +size_t CRYPTO_cts128_encrypt_block(const unsigned char *in, + unsigned char *out, size_t len, + const void *key, unsigned char ivec[16], + block128_f block) +{ + size_t residue, n; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len <= 16) return 0; + if (len <= 16) + return 0; - if ((residue=len%16) == 0) residue = 16; + if ((residue = len % 16) == 0) + residue = 16; - len -= residue; + len -= residue; - CRYPTO_cbc128_encrypt(in,out,len,key,ivec,block); + CRYPTO_cbc128_encrypt(in, out, len, key, ivec, block); - in += len; - out += len; + in += len; + out += len; - for (n=0; n<residue; ++n) - ivec[n] ^= in[n]; - (*block)(ivec,ivec,key); - memcpy(out,out-16,residue); - memcpy(out-16,ivec,16); + for (n = 0; n < residue; ++n) + ivec[n] ^= in[n]; + (*block) (ivec, ivec, key); + memcpy(out, out - 16, residue); + memcpy(out - 16, ivec, 16); - return len+residue; + return len + residue; } -size_t CRYPTO_nistcts128_encrypt_block(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) -{ size_t residue, n; +size_t CRYPTO_nistcts128_encrypt_block(const unsigned char *in, + unsigned char *out, size_t len, + const void *key, + unsigned char ivec[16], + block128_f block) +{ + size_t residue, n; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len < 16) return 0; + if (len < 16) + return 0; - residue=len%16; + residue = len % 16; - len -= residue; + len -= residue; - CRYPTO_cbc128_encrypt(in,out,len,key,ivec,block); + CRYPTO_cbc128_encrypt(in, out, len, key, ivec, block); - if (residue==0) return len; + if (residue == 0) + return len; - in += len; - out += len; + in += len; + out += len; - for (n=0; n<residue; ++n) - ivec[n] ^= in[n]; - (*block)(ivec,ivec,key); - memcpy(out-16+residue,ivec,16); + for (n = 0; n < residue; ++n) + ivec[n] ^= in[n]; + (*block) (ivec, ivec, key); + memcpy(out - 16 + residue, ivec, 16); - return len+residue; + return len + residue; } size_t CRYPTO_cts128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], cbc128_f cbc) -{ size_t residue; - union { size_t align; unsigned char c[16]; } tmp; + size_t len, const void *key, + unsigned char ivec[16], cbc128_f cbc) +{ + size_t residue; + union { + size_t align; + unsigned char c[16]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len <= 16) return 0; + if (len <= 16) + return 0; - if ((residue=len%16) == 0) residue = 16; + if ((residue = len % 16) == 0) + residue = 16; - len -= residue; + len -= residue; - (*cbc)(in,out,len,key,ivec,1); + (*cbc) (in, out, len, key, ivec, 1); - in += len; - out += len; + in += len; + out += len; #if defined(CBC_HANDLES_TRUNCATED_IO) - memcpy(tmp.c,out-16,16); - (*cbc)(in,out-16,residue,key,ivec,1); - memcpy(out,tmp.c,residue); + memcpy(tmp.c, out - 16, 16); + (*cbc) (in, out - 16, residue, key, ivec, 1); + memcpy(out, tmp.c, residue); #else - memset(tmp.c,0,sizeof(tmp)); - memcpy(tmp.c,in,residue); - memcpy(out,out-16,residue); - (*cbc)(tmp.c,out-16,16,key,ivec,1); + memset(tmp.c, 0, sizeof(tmp)); + memcpy(tmp.c, in, residue); + memcpy(out, out - 16, residue); + (*cbc) (tmp.c, out - 16, 16, key, ivec, 1); #endif - return len+residue; + return len + residue; } size_t CRYPTO_nistcts128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], cbc128_f cbc) -{ size_t residue; - union { size_t align; unsigned char c[16]; } tmp; + size_t len, const void *key, + unsigned char ivec[16], cbc128_f cbc) +{ + size_t residue; + union { + size_t align; + unsigned char c[16]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len < 16) return 0; + if (len < 16) + return 0; - residue=len%16; + residue = len % 16; - len -= residue; + len -= residue; - (*cbc)(in,out,len,key,ivec,1); + (*cbc) (in, out, len, key, ivec, 1); - if (residue==0) return len; + if (residue == 0) + return len; - in += len; - out += len; + in += len; + out += len; #if defined(CBC_HANDLES_TRUNCATED_IO) - (*cbc)(in,out-16+residue,residue,key,ivec,1); + (*cbc) (in, out - 16 + residue, residue, key, ivec, 1); #else - memset(tmp.c,0,sizeof(tmp)); - memcpy(tmp.c,in,residue); - (*cbc)(tmp.c,out-16+residue,16,key,ivec,1); + memset(tmp.c, 0, sizeof(tmp)); + memcpy(tmp.c, in, residue); + (*cbc) (tmp.c, out - 16 + residue, 16, key, ivec, 1); #endif - return len+residue; + return len + residue; } -size_t CRYPTO_cts128_decrypt_block(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) -{ size_t residue, n; - union { size_t align; unsigned char c[32]; } tmp; +size_t CRYPTO_cts128_decrypt_block(const unsigned char *in, + unsigned char *out, size_t len, + const void *key, unsigned char ivec[16], + block128_f block) +{ + size_t residue, n; + union { + size_t align; + unsigned char c[32]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len<=16) return 0; + if (len <= 16) + return 0; - if ((residue=len%16) == 0) residue = 16; + if ((residue = len % 16) == 0) + residue = 16; - len -= 16+residue; + len -= 16 + residue; - if (len) { - CRYPTO_cbc128_decrypt(in,out,len,key,ivec,block); - in += len; - out += len; - } + if (len) { + CRYPTO_cbc128_decrypt(in, out, len, key, ivec, block); + in += len; + out += len; + } - (*block)(in,tmp.c+16,key); + (*block) (in, tmp.c + 16, key); - memcpy(tmp.c,tmp.c+16,16); - memcpy(tmp.c,in+16,residue); - (*block)(tmp.c,tmp.c,key); + memcpy(tmp.c, tmp.c + 16, 16); + memcpy(tmp.c, in + 16, residue); + (*block) (tmp.c, tmp.c, key); - for(n=0; n<16; ++n) { - unsigned char c = in[n]; - out[n] = tmp.c[n] ^ ivec[n]; - ivec[n] = c; - } - for(residue+=16; n<residue; ++n) - out[n] = tmp.c[n] ^ in[n]; + for (n = 0; n < 16; ++n) { + unsigned char c = in[n]; + out[n] = tmp.c[n] ^ ivec[n]; + ivec[n] = c; + } + for (residue += 16; n < residue; ++n) + out[n] = tmp.c[n] ^ in[n]; - return 16+len+residue; + return 16 + len + residue; } -size_t CRYPTO_nistcts128_decrypt_block(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], block128_f block) -{ size_t residue, n; - union { size_t align; unsigned char c[32]; } tmp; +size_t CRYPTO_nistcts128_decrypt_block(const unsigned char *in, + unsigned char *out, size_t len, + const void *key, + unsigned char ivec[16], + block128_f block) +{ + size_t residue, n; + union { + size_t align; + unsigned char c[32]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len<16) return 0; + if (len < 16) + return 0; - residue=len%16; + residue = len % 16; - if (residue==0) { - CRYPTO_cbc128_decrypt(in,out,len,key,ivec,block); - return len; - } + if (residue == 0) { + CRYPTO_cbc128_decrypt(in, out, len, key, ivec, block); + return len; + } - len -= 16+residue; + len -= 16 + residue; - if (len) { - CRYPTO_cbc128_decrypt(in,out,len,key,ivec,block); - in += len; - out += len; - } + if (len) { + CRYPTO_cbc128_decrypt(in, out, len, key, ivec, block); + in += len; + out += len; + } - (*block)(in+residue,tmp.c+16,key); + (*block) (in + residue, tmp.c + 16, key); - memcpy(tmp.c,tmp.c+16,16); - memcpy(tmp.c,in,residue); - (*block)(tmp.c,tmp.c,key); + memcpy(tmp.c, tmp.c + 16, 16); + memcpy(tmp.c, in, residue); + (*block) (tmp.c, tmp.c, key); - for(n=0; n<16; ++n) { - unsigned char c = in[n]; - out[n] = tmp.c[n] ^ ivec[n]; - ivec[n] = in[n+residue]; - tmp.c[n] = c; - } - for(residue+=16; n<residue; ++n) - out[n] = tmp.c[n] ^ tmp.c[n-16]; + for (n = 0; n < 16; ++n) { + unsigned char c = in[n]; + out[n] = tmp.c[n] ^ ivec[n]; + ivec[n] = in[n + residue]; + tmp.c[n] = c; + } + for (residue += 16; n < residue; ++n) + out[n] = tmp.c[n] ^ tmp.c[n - 16]; - return 16+len+residue; + return 16 + len + residue; } size_t CRYPTO_cts128_decrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], cbc128_f cbc) -{ size_t residue; - union { size_t align; unsigned char c[32]; } tmp; + size_t len, const void *key, + unsigned char ivec[16], cbc128_f cbc) +{ + size_t residue; + union { + size_t align; + unsigned char c[32]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len<=16) return 0; + if (len <= 16) + return 0; - if ((residue=len%16) == 0) residue = 16; + if ((residue = len % 16) == 0) + residue = 16; - len -= 16+residue; + len -= 16 + residue; - if (len) { - (*cbc)(in,out,len,key,ivec,0); - in += len; - out += len; - } + if (len) { + (*cbc) (in, out, len, key, ivec, 0); + in += len; + out += len; + } - memset(tmp.c,0,sizeof(tmp)); - /* this places in[16] at &tmp.c[16] and decrypted block at &tmp.c[0] */ - (*cbc)(in,tmp.c,16,key,tmp.c+16,0); + memset(tmp.c, 0, sizeof(tmp)); + /* + * this places in[16] at &tmp.c[16] and decrypted block at &tmp.c[0] + */ + (*cbc) (in, tmp.c, 16, key, tmp.c + 16, 0); - memcpy(tmp.c,in+16,residue); + memcpy(tmp.c, in + 16, residue); #if defined(CBC_HANDLES_TRUNCATED_IO) - (*cbc)(tmp.c,out,16+residue,key,ivec,0); + (*cbc) (tmp.c, out, 16 + residue, key, ivec, 0); #else - (*cbc)(tmp.c,tmp.c,32,key,ivec,0); - memcpy(out,tmp.c,16+residue); + (*cbc) (tmp.c, tmp.c, 32, key, ivec, 0); + memcpy(out, tmp.c, 16 + residue); #endif - return 16+len+residue; + return 16 + len + residue; } size_t CRYPTO_nistcts128_decrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], cbc128_f cbc) -{ size_t residue; - union { size_t align; unsigned char c[32]; } tmp; + size_t len, const void *key, + unsigned char ivec[16], cbc128_f cbc) +{ + size_t residue; + union { + size_t align; + unsigned char c[32]; + } tmp; - assert (in && out && key && ivec); + assert(in && out && key && ivec); - if (len<16) return 0; + if (len < 16) + return 0; - residue=len%16; + residue = len % 16; - if (residue==0) { - (*cbc)(in,out,len,key,ivec,0); - return len; - } + if (residue == 0) { + (*cbc) (in, out, len, key, ivec, 0); + return len; + } - len -= 16+residue; + len -= 16 + residue; - if (len) { - (*cbc)(in,out,len,key,ivec,0); - in += len; - out += len; - } + if (len) { + (*cbc) (in, out, len, key, ivec, 0); + in += len; + out += len; + } - memset(tmp.c,0,sizeof(tmp)); - /* this places in[16] at &tmp.c[16] and decrypted block at &tmp.c[0] */ - (*cbc)(in+residue,tmp.c,16,key,tmp.c+16,0); + memset(tmp.c, 0, sizeof(tmp)); + /* + * this places in[16] at &tmp.c[16] and decrypted block at &tmp.c[0] + */ + (*cbc) (in + residue, tmp.c, 16, key, tmp.c + 16, 0); - memcpy(tmp.c,in,residue); + memcpy(tmp.c, in, residue); #if defined(CBC_HANDLES_TRUNCATED_IO) - (*cbc)(tmp.c,out,16+residue,key,ivec,0); + (*cbc) (tmp.c, out, 16 + residue, key, ivec, 0); #else - (*cbc)(tmp.c,tmp.c,32,key,ivec,0); - memcpy(out,tmp.c,16+residue); + (*cbc) (tmp.c, tmp.c, 32, key, ivec, 0); + memcpy(out, tmp.c, 16 + residue); #endif - return 16+len+residue; + return 16 + len + residue; } #if defined(SELFTEST) -#include <stdio.h> -#include <openssl/aes.h> +# include <stdio.h> +# include <openssl/aes.h> /* test vectors from RFC 3962 */ static const unsigned char test_key[16] = "chicken teriyaki"; static const unsigned char test_input[64] = - "I would like the" " General Gau's C" - "hicken, please, " "and wonton soup."; -static const unsigned char test_iv[16] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0}; - -static const unsigned char vector_17[17] = -{0xc6,0x35,0x35,0x68,0xf2,0xbf,0x8c,0xb4, 0xd8,0xa5,0x80,0x36,0x2d,0xa7,0xff,0x7f, - 0x97}; -static const unsigned char vector_31[31] = -{0xfc,0x00,0x78,0x3e,0x0e,0xfd,0xb2,0xc1, 0xd4,0x45,0xd4,0xc8,0xef,0xf7,0xed,0x22, - 0x97,0x68,0x72,0x68,0xd6,0xec,0xcc,0xc0, 0xc0,0x7b,0x25,0xe2,0x5e,0xcf,0xe5}; -static const unsigned char vector_32[32] = -{0x39,0x31,0x25,0x23,0xa7,0x86,0x62,0xd5, 0xbe,0x7f,0xcb,0xcc,0x98,0xeb,0xf5,0xa8, - 0x97,0x68,0x72,0x68,0xd6,0xec,0xcc,0xc0, 0xc0,0x7b,0x25,0xe2,0x5e,0xcf,0xe5,0x84}; -static const unsigned char vector_47[47] = -{0x97,0x68,0x72,0x68,0xd6,0xec,0xcc,0xc0, 0xc0,0x7b,0x25,0xe2,0x5e,0xcf,0xe5,0x84, - 0xb3,0xff,0xfd,0x94,0x0c,0x16,0xa1,0x8c, 0x1b,0x55,0x49,0xd2,0xf8,0x38,0x02,0x9e, - 0x39,0x31,0x25,0x23,0xa7,0x86,0x62,0xd5, 0xbe,0x7f,0xcb,0xcc,0x98,0xeb,0xf5}; -static const unsigned char vector_48[48] = -{0x97,0x68,0x72,0x68,0xd6,0xec,0xcc,0xc0, 0xc0,0x7b,0x25,0xe2,0x5e,0xcf,0xe5,0x84, - 0x9d,0xad,0x8b,0xbb,0x96,0xc4,0xcd,0xc0, 0x3b,0xc1,0x03,0xe1,0xa1,0x94,0xbb,0xd8, - 0x39,0x31,0x25,0x23,0xa7,0x86,0x62,0xd5, 0xbe,0x7f,0xcb,0xcc,0x98,0xeb,0xf5,0xa8}; -static const unsigned char vector_64[64] = -{0x97,0x68,0x72,0x68,0xd6,0xec,0xcc,0xc0, 0xc0,0x7b,0x25,0xe2,0x5e,0xcf,0xe5,0x84, - 0x39,0x31,0x25,0x23,0xa7,0x86,0x62,0xd5, 0xbe,0x7f,0xcb,0xcc,0x98,0xeb,0xf5,0xa8, - 0x48,0x07,0xef,0xe8,0x36,0xee,0x89,0xa5, 0x26,0x73,0x0d,0xbc,0x2f,0x7b,0xc8,0x40, - 0x9d,0xad,0x8b,0xbb,0x96,0xc4,0xcd,0xc0, 0x3b,0xc1,0x03,0xe1,0xa1,0x94,0xbb,0xd8}; + "I would like the" " General Gau's C" + "hicken, please, " "and wonton soup."; +static const unsigned char test_iv[16] = + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + +static const unsigned char vector_17[17] = { + 0xc6, 0x35, 0x35, 0x68, 0xf2, 0xbf, 0x8c, 0xb4, + 0xd8, 0xa5, 0x80, 0x36, 0x2d, 0xa7, 0xff, 0x7f, + 0x97 +}; + +static const unsigned char vector_31[31] = { + 0xfc, 0x00, 0x78, 0x3e, 0x0e, 0xfd, 0xb2, 0xc1, + 0xd4, 0x45, 0xd4, 0xc8, 0xef, 0xf7, 0xed, 0x22, + 0x97, 0x68, 0x72, 0x68, 0xd6, 0xec, 0xcc, 0xc0, + 0xc0, 0x7b, 0x25, 0xe2, 0x5e, 0xcf, 0xe5 +}; + +static const unsigned char vector_32[32] = { + 0x39, 0x31, 0x25, 0x23, 0xa7, 0x86, 0x62, 0xd5, + 0xbe, 0x7f, 0xcb, 0xcc, 0x98, 0xeb, 0xf5, 0xa8, + 0x97, 0x68, 0x72, 0x68, 0xd6, 0xec, 0xcc, 0xc0, + 0xc0, 0x7b, 0x25, 0xe2, 0x5e, 0xcf, 0xe5, 0x84 +}; + +static const unsigned char vector_47[47] = { + 0x97, 0x68, 0x72, 0x68, 0xd6, 0xec, 0xcc, 0xc0, + 0xc0, 0x7b, 0x25, 0xe2, 0x5e, 0xcf, 0xe5, 0x84, + 0xb3, 0xff, 0xfd, 0x94, 0x0c, 0x16, 0xa1, 0x8c, + 0x1b, 0x55, 0x49, 0xd2, 0xf8, 0x38, 0x02, 0x9e, + 0x39, 0x31, 0x25, 0x23, 0xa7, 0x86, 0x62, 0xd5, + 0xbe, 0x7f, 0xcb, 0xcc, 0x98, 0xeb, 0xf5 +}; + +static const unsigned char vector_48[48] = { + 0x97, 0x68, 0x72, 0x68, 0xd6, 0xec, 0xcc, 0xc0, + 0xc0, 0x7b, 0x25, 0xe2, 0x5e, 0xcf, 0xe5, 0x84, + 0x9d, 0xad, 0x8b, 0xbb, 0x96, 0xc4, 0xcd, 0xc0, + 0x3b, 0xc1, 0x03, 0xe1, 0xa1, 0x94, 0xbb, 0xd8, + 0x39, 0x31, 0x25, 0x23, 0xa7, 0x86, 0x62, 0xd5, + 0xbe, 0x7f, 0xcb, 0xcc, 0x98, 0xeb, 0xf5, 0xa8 +}; + +static const unsigned char vector_64[64] = { + 0x97, 0x68, 0x72, 0x68, 0xd6, 0xec, 0xcc, 0xc0, + 0xc0, 0x7b, 0x25, 0xe2, 0x5e, 0xcf, 0xe5, 0x84, + 0x39, 0x31, 0x25, 0x23, 0xa7, 0x86, 0x62, 0xd5, + 0xbe, 0x7f, 0xcb, 0xcc, 0x98, 0xeb, 0xf5, 0xa8, + 0x48, 0x07, 0xef, 0xe8, 0x36, 0xee, 0x89, 0xa5, + 0x26, 0x73, 0x0d, 0xbc, 0x2f, 0x7b, 0xc8, 0x40, + 0x9d, 0xad, 0x8b, 0xbb, 0x96, 0xc4, 0xcd, 0xc0, + 0x3b, 0xc1, 0x03, 0xe1, 0xa1, 0x94, 0xbb, 0xd8 +}; static AES_KEY encks, decks; -void test_vector(const unsigned char *vector,size_t len) -{ unsigned char iv[sizeof(test_iv)]; - unsigned char cleartext[64],ciphertext[64]; - size_t tail; - - printf("vector_%d\n",len); fflush(stdout); - - if ((tail=len%16) == 0) tail = 16; - tail += 16; - - /* test block-based encryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_cts128_encrypt_block(test_input,ciphertext,len,&encks,iv,(block128_f)AES_encrypt); - if (memcmp(ciphertext,vector,len)) - fprintf(stderr,"output_%d mismatch\n",len), exit(1); - if (memcmp(iv,vector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(1); - - /* test block-based decryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_cts128_decrypt_block(ciphertext,cleartext,len,&decks,iv,(block128_f)AES_decrypt); - if (memcmp(cleartext,test_input,len)) - fprintf(stderr,"input_%d mismatch\n",len), exit(2); - if (memcmp(iv,vector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(2); - - /* test streamed encryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_cts128_encrypt(test_input,ciphertext,len,&encks,iv,(cbc128_f)AES_cbc_encrypt); - if (memcmp(ciphertext,vector,len)) - fprintf(stderr,"output_%d mismatch\n",len), exit(3); - if (memcmp(iv,vector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(3); - - /* test streamed decryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_cts128_decrypt(ciphertext,cleartext,len,&decks,iv,(cbc128_f)AES_cbc_encrypt); - if (memcmp(cleartext,test_input,len)) - fprintf(stderr,"input_%d mismatch\n",len), exit(4); - if (memcmp(iv,vector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(4); +void test_vector(const unsigned char *vector, size_t len) +{ + unsigned char iv[sizeof(test_iv)]; + unsigned char cleartext[64], ciphertext[64]; + size_t tail; + + printf("vector_%d\n", len); + fflush(stdout); + + if ((tail = len % 16) == 0) + tail = 16; + tail += 16; + + /* test block-based encryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_cts128_encrypt_block(test_input, ciphertext, len, &encks, iv, + (block128_f) AES_encrypt); + if (memcmp(ciphertext, vector, len)) + fprintf(stderr, "output_%d mismatch\n", len), exit(1); + if (memcmp(iv, vector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(1); + + /* test block-based decryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_cts128_decrypt_block(ciphertext, cleartext, len, &decks, iv, + (block128_f) AES_decrypt); + if (memcmp(cleartext, test_input, len)) + fprintf(stderr, "input_%d mismatch\n", len), exit(2); + if (memcmp(iv, vector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(2); + + /* test streamed encryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_cts128_encrypt(test_input, ciphertext, len, &encks, iv, + (cbc128_f) AES_cbc_encrypt); + if (memcmp(ciphertext, vector, len)) + fprintf(stderr, "output_%d mismatch\n", len), exit(3); + if (memcmp(iv, vector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(3); + + /* test streamed decryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_cts128_decrypt(ciphertext, cleartext, len, &decks, iv, + (cbc128_f) AES_cbc_encrypt); + if (memcmp(cleartext, test_input, len)) + fprintf(stderr, "input_%d mismatch\n", len), exit(4); + if (memcmp(iv, vector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(4); } -void test_nistvector(const unsigned char *vector,size_t len) -{ unsigned char iv[sizeof(test_iv)]; - unsigned char cleartext[64],ciphertext[64],nistvector[64]; - size_t tail; - - printf("nistvector_%d\n",len); fflush(stdout); - - if ((tail=len%16) == 0) tail = 16; - - len -= 16 + tail; - memcpy(nistvector,vector,len); - /* flip two last blocks */ - memcpy(nistvector+len,vector+len+16,tail); - memcpy(nistvector+len+tail,vector+len,16); - len += 16 + tail; - tail = 16; - - /* test block-based encryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_nistcts128_encrypt_block(test_input,ciphertext,len,&encks,iv,(block128_f)AES_encrypt); - if (memcmp(ciphertext,nistvector,len)) - fprintf(stderr,"output_%d mismatch\n",len), exit(1); - if (memcmp(iv,nistvector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(1); - - /* test block-based decryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_nistcts128_decrypt_block(ciphertext,cleartext,len,&decks,iv,(block128_f)AES_decrypt); - if (memcmp(cleartext,test_input,len)) - fprintf(stderr,"input_%d mismatch\n",len), exit(2); - if (memcmp(iv,nistvector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(2); - - /* test streamed encryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_nistcts128_encrypt(test_input,ciphertext,len,&encks,iv,(cbc128_f)AES_cbc_encrypt); - if (memcmp(ciphertext,nistvector,len)) - fprintf(stderr,"output_%d mismatch\n",len), exit(3); - if (memcmp(iv,nistvector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(3); - - /* test streamed decryption */ - memcpy(iv,test_iv,sizeof(test_iv)); - CRYPTO_nistcts128_decrypt(ciphertext,cleartext,len,&decks,iv,(cbc128_f)AES_cbc_encrypt); - if (memcmp(cleartext,test_input,len)) - fprintf(stderr,"input_%d mismatch\n",len), exit(4); - if (memcmp(iv,nistvector+len-tail,sizeof(iv))) - fprintf(stderr,"iv_%d mismatch\n",len), exit(4); +void test_nistvector(const unsigned char *vector, size_t len) +{ + unsigned char iv[sizeof(test_iv)]; + unsigned char cleartext[64], ciphertext[64], nistvector[64]; + size_t tail; + + printf("nistvector_%d\n", len); + fflush(stdout); + + if ((tail = len % 16) == 0) + tail = 16; + + len -= 16 + tail; + memcpy(nistvector, vector, len); + /* flip two last blocks */ + memcpy(nistvector + len, vector + len + 16, tail); + memcpy(nistvector + len + tail, vector + len, 16); + len += 16 + tail; + tail = 16; + + /* test block-based encryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_nistcts128_encrypt_block(test_input, ciphertext, len, &encks, iv, + (block128_f) AES_encrypt); + if (memcmp(ciphertext, nistvector, len)) + fprintf(stderr, "output_%d mismatch\n", len), exit(1); + if (memcmp(iv, nistvector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(1); + + /* test block-based decryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_nistcts128_decrypt_block(ciphertext, cleartext, len, &decks, iv, + (block128_f) AES_decrypt); + if (memcmp(cleartext, test_input, len)) + fprintf(stderr, "input_%d mismatch\n", len), exit(2); + if (memcmp(iv, nistvector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(2); + + /* test streamed encryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_nistcts128_encrypt(test_input, ciphertext, len, &encks, iv, + (cbc128_f) AES_cbc_encrypt); + if (memcmp(ciphertext, nistvector, len)) + fprintf(stderr, "output_%d mismatch\n", len), exit(3); + if (memcmp(iv, nistvector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(3); + + /* test streamed decryption */ + memcpy(iv, test_iv, sizeof(test_iv)); + CRYPTO_nistcts128_decrypt(ciphertext, cleartext, len, &decks, iv, + (cbc128_f) AES_cbc_encrypt); + if (memcmp(cleartext, test_input, len)) + fprintf(stderr, "input_%d mismatch\n", len), exit(4); + if (memcmp(iv, nistvector + len - tail, sizeof(iv))) + fprintf(stderr, "iv_%d mismatch\n", len), exit(4); } int main() { - AES_set_encrypt_key(test_key,128,&encks); - AES_set_decrypt_key(test_key,128,&decks); - - test_vector(vector_17,sizeof(vector_17)); - test_vector(vector_31,sizeof(vector_31)); - test_vector(vector_32,sizeof(vector_32)); - test_vector(vector_47,sizeof(vector_47)); - test_vector(vector_48,sizeof(vector_48)); - test_vector(vector_64,sizeof(vector_64)); - - test_nistvector(vector_17,sizeof(vector_17)); - test_nistvector(vector_31,sizeof(vector_31)); - test_nistvector(vector_32,sizeof(vector_32)); - test_nistvector(vector_47,sizeof(vector_47)); - test_nistvector(vector_48,sizeof(vector_48)); - test_nistvector(vector_64,sizeof(vector_64)); - - return 0; + AES_set_encrypt_key(test_key, 128, &encks); + AES_set_decrypt_key(test_key, 128, &decks); + + test_vector(vector_17, sizeof(vector_17)); + test_vector(vector_31, sizeof(vector_31)); + test_vector(vector_32, sizeof(vector_32)); + test_vector(vector_47, sizeof(vector_47)); + test_vector(vector_48, sizeof(vector_48)); + test_vector(vector_64, sizeof(vector_64)); + + test_nistvector(vector_17, sizeof(vector_17)); + test_nistvector(vector_31, sizeof(vector_31)); + test_nistvector(vector_32, sizeof(vector_32)); + test_nistvector(vector_47, sizeof(vector_47)); + test_nistvector(vector_48, sizeof(vector_48)); + test_nistvector(vector_64, sizeof(vector_64)); + + return 0; } #endif diff --git a/drivers/builtin_openssl2/crypto/modes/gcm128.c b/drivers/builtin_openssl2/crypto/modes/gcm128.c index e1dc2b0f47..0ee569fb7a 100644 --- a/drivers/builtin_openssl2/crypto/modes/gcm128.c +++ b/drivers/builtin_openssl2/crypto/modes/gcm128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -62,27 +62,27 @@ #if defined(BSWAP4) && defined(STRICT_ALIGNMENT) /* redefine, because alignment is ensured */ -#undef GETU32 -#define GETU32(p) BSWAP4(*(const u32 *)(p)) -#undef PUTU32 -#define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) -#endif - -#define PACK(s) ((size_t)(s)<<(sizeof(size_t)*8-16)) -#define REDUCE1BIT(V) do { \ - if (sizeof(size_t)==8) { \ - u64 T = U64(0xe100000000000000) & (0-(V.lo&1)); \ - V.lo = (V.hi<<63)|(V.lo>>1); \ - V.hi = (V.hi>>1 )^T; \ - } \ - else { \ - u32 T = 0xe1000000U & (0-(u32)(V.lo&1)); \ - V.lo = (V.hi<<63)|(V.lo>>1); \ - V.hi = (V.hi>>1 )^((u64)T<<32); \ - } \ +# undef GETU32 +# define GETU32(p) BSWAP4(*(const u32 *)(p)) +# undef PUTU32 +# define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) +#endif + +#define PACK(s) ((size_t)(s)<<(sizeof(size_t)*8-16)) +#define REDUCE1BIT(V) do { \ + if (sizeof(size_t)==8) { \ + u64 T = U64(0xe100000000000000) & (0-(V.lo&1)); \ + V.lo = (V.hi<<63)|(V.lo>>1); \ + V.hi = (V.hi>>1 )^T; \ + } \ + else { \ + u32 T = 0xe1000000U & (0-(u32)(V.lo&1)); \ + V.lo = (V.hi<<63)|(V.lo>>1); \ + V.hi = (V.hi>>1 )^((u64)T<<32); \ + } \ } while(0) -/* +/*- * Even though permitted values for TABLE_BITS are 8, 4 and 1, it should * never be set to 8. 8 is effectively reserved for testing purposes. * TABLE_BITS>1 are lookup-table-driven implementations referred to as @@ -116,286 +116,311 @@ * * Value of 1 is not appropriate for performance reasons. */ -#if TABLE_BITS==8 +#if TABLE_BITS==8 static void gcm_init_8bit(u128 Htable[256], u64 H[2]) { - int i, j; - u128 V; - - Htable[0].hi = 0; - Htable[0].lo = 0; - V.hi = H[0]; - V.lo = H[1]; - - for (Htable[128]=V, i=64; i>0; i>>=1) { - REDUCE1BIT(V); - Htable[i] = V; - } - - for (i=2; i<256; i<<=1) { - u128 *Hi = Htable+i, H0 = *Hi; - for (j=1; j<i; ++j) { - Hi[j].hi = H0.hi^Htable[j].hi; - Hi[j].lo = H0.lo^Htable[j].lo; - } - } + int i, j; + u128 V; + + Htable[0].hi = 0; + Htable[0].lo = 0; + V.hi = H[0]; + V.lo = H[1]; + + for (Htable[128] = V, i = 64; i > 0; i >>= 1) { + REDUCE1BIT(V); + Htable[i] = V; + } + + for (i = 2; i < 256; i <<= 1) { + u128 *Hi = Htable + i, H0 = *Hi; + for (j = 1; j < i; ++j) { + Hi[j].hi = H0.hi ^ Htable[j].hi; + Hi[j].lo = H0.lo ^ Htable[j].lo; + } + } } static void gcm_gmult_8bit(u64 Xi[2], const u128 Htable[256]) { - u128 Z = { 0, 0}; - const u8 *xi = (const u8 *)Xi+15; - size_t rem, n = *xi; - const union { long one; char little; } is_endian = {1}; - static const size_t rem_8bit[256] = { - PACK(0x0000), PACK(0x01C2), PACK(0x0384), PACK(0x0246), - PACK(0x0708), PACK(0x06CA), PACK(0x048C), PACK(0x054E), - PACK(0x0E10), PACK(0x0FD2), PACK(0x0D94), PACK(0x0C56), - PACK(0x0918), PACK(0x08DA), PACK(0x0A9C), PACK(0x0B5E), - PACK(0x1C20), PACK(0x1DE2), PACK(0x1FA4), PACK(0x1E66), - PACK(0x1B28), PACK(0x1AEA), PACK(0x18AC), PACK(0x196E), - PACK(0x1230), PACK(0x13F2), PACK(0x11B4), PACK(0x1076), - PACK(0x1538), PACK(0x14FA), PACK(0x16BC), PACK(0x177E), - PACK(0x3840), PACK(0x3982), PACK(0x3BC4), PACK(0x3A06), - PACK(0x3F48), PACK(0x3E8A), PACK(0x3CCC), PACK(0x3D0E), - PACK(0x3650), PACK(0x3792), PACK(0x35D4), PACK(0x3416), - PACK(0x3158), PACK(0x309A), PACK(0x32DC), PACK(0x331E), - PACK(0x2460), PACK(0x25A2), PACK(0x27E4), PACK(0x2626), - PACK(0x2368), PACK(0x22AA), PACK(0x20EC), PACK(0x212E), - PACK(0x2A70), PACK(0x2BB2), PACK(0x29F4), PACK(0x2836), - PACK(0x2D78), PACK(0x2CBA), PACK(0x2EFC), PACK(0x2F3E), - PACK(0x7080), PACK(0x7142), PACK(0x7304), PACK(0x72C6), - PACK(0x7788), PACK(0x764A), PACK(0x740C), PACK(0x75CE), - PACK(0x7E90), PACK(0x7F52), PACK(0x7D14), PACK(0x7CD6), - PACK(0x7998), PACK(0x785A), PACK(0x7A1C), PACK(0x7BDE), - PACK(0x6CA0), PACK(0x6D62), PACK(0x6F24), PACK(0x6EE6), - PACK(0x6BA8), PACK(0x6A6A), PACK(0x682C), PACK(0x69EE), - PACK(0x62B0), PACK(0x6372), PACK(0x6134), PACK(0x60F6), - PACK(0x65B8), PACK(0x647A), PACK(0x663C), PACK(0x67FE), - PACK(0x48C0), PACK(0x4902), PACK(0x4B44), PACK(0x4A86), - PACK(0x4FC8), PACK(0x4E0A), PACK(0x4C4C), PACK(0x4D8E), - PACK(0x46D0), PACK(0x4712), PACK(0x4554), PACK(0x4496), - PACK(0x41D8), PACK(0x401A), PACK(0x425C), PACK(0x439E), - PACK(0x54E0), PACK(0x5522), PACK(0x5764), PACK(0x56A6), - PACK(0x53E8), PACK(0x522A), PACK(0x506C), PACK(0x51AE), - PACK(0x5AF0), PACK(0x5B32), PACK(0x5974), PACK(0x58B6), - PACK(0x5DF8), PACK(0x5C3A), PACK(0x5E7C), PACK(0x5FBE), - PACK(0xE100), PACK(0xE0C2), PACK(0xE284), PACK(0xE346), - PACK(0xE608), PACK(0xE7CA), PACK(0xE58C), PACK(0xE44E), - PACK(0xEF10), PACK(0xEED2), PACK(0xEC94), PACK(0xED56), - PACK(0xE818), PACK(0xE9DA), PACK(0xEB9C), PACK(0xEA5E), - PACK(0xFD20), PACK(0xFCE2), PACK(0xFEA4), PACK(0xFF66), - PACK(0xFA28), PACK(0xFBEA), PACK(0xF9AC), PACK(0xF86E), - PACK(0xF330), PACK(0xF2F2), PACK(0xF0B4), PACK(0xF176), - PACK(0xF438), PACK(0xF5FA), PACK(0xF7BC), PACK(0xF67E), - PACK(0xD940), PACK(0xD882), PACK(0xDAC4), PACK(0xDB06), - PACK(0xDE48), PACK(0xDF8A), PACK(0xDDCC), PACK(0xDC0E), - PACK(0xD750), PACK(0xD692), PACK(0xD4D4), PACK(0xD516), - PACK(0xD058), PACK(0xD19A), PACK(0xD3DC), PACK(0xD21E), - PACK(0xC560), PACK(0xC4A2), PACK(0xC6E4), PACK(0xC726), - PACK(0xC268), PACK(0xC3AA), PACK(0xC1EC), PACK(0xC02E), - PACK(0xCB70), PACK(0xCAB2), PACK(0xC8F4), PACK(0xC936), - PACK(0xCC78), PACK(0xCDBA), PACK(0xCFFC), PACK(0xCE3E), - PACK(0x9180), PACK(0x9042), PACK(0x9204), PACK(0x93C6), - PACK(0x9688), PACK(0x974A), PACK(0x950C), PACK(0x94CE), - PACK(0x9F90), PACK(0x9E52), PACK(0x9C14), PACK(0x9DD6), - PACK(0x9898), PACK(0x995A), PACK(0x9B1C), PACK(0x9ADE), - PACK(0x8DA0), PACK(0x8C62), PACK(0x8E24), PACK(0x8FE6), - PACK(0x8AA8), PACK(0x8B6A), PACK(0x892C), PACK(0x88EE), - PACK(0x83B0), PACK(0x8272), PACK(0x8034), PACK(0x81F6), - PACK(0x84B8), PACK(0x857A), PACK(0x873C), PACK(0x86FE), - PACK(0xA9C0), PACK(0xA802), PACK(0xAA44), PACK(0xAB86), - PACK(0xAEC8), PACK(0xAF0A), PACK(0xAD4C), PACK(0xAC8E), - PACK(0xA7D0), PACK(0xA612), PACK(0xA454), PACK(0xA596), - PACK(0xA0D8), PACK(0xA11A), PACK(0xA35C), PACK(0xA29E), - PACK(0xB5E0), PACK(0xB422), PACK(0xB664), PACK(0xB7A6), - PACK(0xB2E8), PACK(0xB32A), PACK(0xB16C), PACK(0xB0AE), - PACK(0xBBF0), PACK(0xBA32), PACK(0xB874), PACK(0xB9B6), - PACK(0xBCF8), PACK(0xBD3A), PACK(0xBF7C), PACK(0xBEBE) }; - - while (1) { - Z.hi ^= Htable[n].hi; - Z.lo ^= Htable[n].lo; - - if ((u8 *)Xi==xi) break; - - n = *(--xi); - - rem = (size_t)Z.lo&0xff; - Z.lo = (Z.hi<<56)|(Z.lo>>8); - Z.hi = (Z.hi>>8); - if (sizeof(size_t)==8) - Z.hi ^= rem_8bit[rem]; - else - Z.hi ^= (u64)rem_8bit[rem]<<32; - } - - if (is_endian.little) { -#ifdef BSWAP8 - Xi[0] = BSWAP8(Z.hi); - Xi[1] = BSWAP8(Z.lo); -#else - u8 *p = (u8 *)Xi; - u32 v; - v = (u32)(Z.hi>>32); PUTU32(p,v); - v = (u32)(Z.hi); PUTU32(p+4,v); - v = (u32)(Z.lo>>32); PUTU32(p+8,v); - v = (u32)(Z.lo); PUTU32(p+12,v); -#endif - } - else { - Xi[0] = Z.hi; - Xi[1] = Z.lo; - } + u128 Z = { 0, 0 }; + const u8 *xi = (const u8 *)Xi + 15; + size_t rem, n = *xi; + const union { + long one; + char little; + } is_endian = { + 1 + }; + static const size_t rem_8bit[256] = { + PACK(0x0000), PACK(0x01C2), PACK(0x0384), PACK(0x0246), + PACK(0x0708), PACK(0x06CA), PACK(0x048C), PACK(0x054E), + PACK(0x0E10), PACK(0x0FD2), PACK(0x0D94), PACK(0x0C56), + PACK(0x0918), PACK(0x08DA), PACK(0x0A9C), PACK(0x0B5E), + PACK(0x1C20), PACK(0x1DE2), PACK(0x1FA4), PACK(0x1E66), + PACK(0x1B28), PACK(0x1AEA), PACK(0x18AC), PACK(0x196E), + PACK(0x1230), PACK(0x13F2), PACK(0x11B4), PACK(0x1076), + PACK(0x1538), PACK(0x14FA), PACK(0x16BC), PACK(0x177E), + PACK(0x3840), PACK(0x3982), PACK(0x3BC4), PACK(0x3A06), + PACK(0x3F48), PACK(0x3E8A), PACK(0x3CCC), PACK(0x3D0E), + PACK(0x3650), PACK(0x3792), PACK(0x35D4), PACK(0x3416), + PACK(0x3158), PACK(0x309A), PACK(0x32DC), PACK(0x331E), + PACK(0x2460), PACK(0x25A2), PACK(0x27E4), PACK(0x2626), + PACK(0x2368), PACK(0x22AA), PACK(0x20EC), PACK(0x212E), + PACK(0x2A70), PACK(0x2BB2), PACK(0x29F4), PACK(0x2836), + PACK(0x2D78), PACK(0x2CBA), PACK(0x2EFC), PACK(0x2F3E), + PACK(0x7080), PACK(0x7142), PACK(0x7304), PACK(0x72C6), + PACK(0x7788), PACK(0x764A), PACK(0x740C), PACK(0x75CE), + PACK(0x7E90), PACK(0x7F52), PACK(0x7D14), PACK(0x7CD6), + PACK(0x7998), PACK(0x785A), PACK(0x7A1C), PACK(0x7BDE), + PACK(0x6CA0), PACK(0x6D62), PACK(0x6F24), PACK(0x6EE6), + PACK(0x6BA8), PACK(0x6A6A), PACK(0x682C), PACK(0x69EE), + PACK(0x62B0), PACK(0x6372), PACK(0x6134), PACK(0x60F6), + PACK(0x65B8), PACK(0x647A), PACK(0x663C), PACK(0x67FE), + PACK(0x48C0), PACK(0x4902), PACK(0x4B44), PACK(0x4A86), + PACK(0x4FC8), PACK(0x4E0A), PACK(0x4C4C), PACK(0x4D8E), + PACK(0x46D0), PACK(0x4712), PACK(0x4554), PACK(0x4496), + PACK(0x41D8), PACK(0x401A), PACK(0x425C), PACK(0x439E), + PACK(0x54E0), PACK(0x5522), PACK(0x5764), PACK(0x56A6), + PACK(0x53E8), PACK(0x522A), PACK(0x506C), PACK(0x51AE), + PACK(0x5AF0), PACK(0x5B32), PACK(0x5974), PACK(0x58B6), + PACK(0x5DF8), PACK(0x5C3A), PACK(0x5E7C), PACK(0x5FBE), + PACK(0xE100), PACK(0xE0C2), PACK(0xE284), PACK(0xE346), + PACK(0xE608), PACK(0xE7CA), PACK(0xE58C), PACK(0xE44E), + PACK(0xEF10), PACK(0xEED2), PACK(0xEC94), PACK(0xED56), + PACK(0xE818), PACK(0xE9DA), PACK(0xEB9C), PACK(0xEA5E), + PACK(0xFD20), PACK(0xFCE2), PACK(0xFEA4), PACK(0xFF66), + PACK(0xFA28), PACK(0xFBEA), PACK(0xF9AC), PACK(0xF86E), + PACK(0xF330), PACK(0xF2F2), PACK(0xF0B4), PACK(0xF176), + PACK(0xF438), PACK(0xF5FA), PACK(0xF7BC), PACK(0xF67E), + PACK(0xD940), PACK(0xD882), PACK(0xDAC4), PACK(0xDB06), + PACK(0xDE48), PACK(0xDF8A), PACK(0xDDCC), PACK(0xDC0E), + PACK(0xD750), PACK(0xD692), PACK(0xD4D4), PACK(0xD516), + PACK(0xD058), PACK(0xD19A), PACK(0xD3DC), PACK(0xD21E), + PACK(0xC560), PACK(0xC4A2), PACK(0xC6E4), PACK(0xC726), + PACK(0xC268), PACK(0xC3AA), PACK(0xC1EC), PACK(0xC02E), + PACK(0xCB70), PACK(0xCAB2), PACK(0xC8F4), PACK(0xC936), + PACK(0xCC78), PACK(0xCDBA), PACK(0xCFFC), PACK(0xCE3E), + PACK(0x9180), PACK(0x9042), PACK(0x9204), PACK(0x93C6), + PACK(0x9688), PACK(0x974A), PACK(0x950C), PACK(0x94CE), + PACK(0x9F90), PACK(0x9E52), PACK(0x9C14), PACK(0x9DD6), + PACK(0x9898), PACK(0x995A), PACK(0x9B1C), PACK(0x9ADE), + PACK(0x8DA0), PACK(0x8C62), PACK(0x8E24), PACK(0x8FE6), + PACK(0x8AA8), PACK(0x8B6A), PACK(0x892C), PACK(0x88EE), + PACK(0x83B0), PACK(0x8272), PACK(0x8034), PACK(0x81F6), + PACK(0x84B8), PACK(0x857A), PACK(0x873C), PACK(0x86FE), + PACK(0xA9C0), PACK(0xA802), PACK(0xAA44), PACK(0xAB86), + PACK(0xAEC8), PACK(0xAF0A), PACK(0xAD4C), PACK(0xAC8E), + PACK(0xA7D0), PACK(0xA612), PACK(0xA454), PACK(0xA596), + PACK(0xA0D8), PACK(0xA11A), PACK(0xA35C), PACK(0xA29E), + PACK(0xB5E0), PACK(0xB422), PACK(0xB664), PACK(0xB7A6), + PACK(0xB2E8), PACK(0xB32A), PACK(0xB16C), PACK(0xB0AE), + PACK(0xBBF0), PACK(0xBA32), PACK(0xB874), PACK(0xB9B6), + PACK(0xBCF8), PACK(0xBD3A), PACK(0xBF7C), PACK(0xBEBE) + }; + + while (1) { + Z.hi ^= Htable[n].hi; + Z.lo ^= Htable[n].lo; + + if ((u8 *)Xi == xi) + break; + + n = *(--xi); + + rem = (size_t)Z.lo & 0xff; + Z.lo = (Z.hi << 56) | (Z.lo >> 8); + Z.hi = (Z.hi >> 8); + if (sizeof(size_t) == 8) + Z.hi ^= rem_8bit[rem]; + else + Z.hi ^= (u64)rem_8bit[rem] << 32; + } + + if (is_endian.little) { +# ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +# else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi >> 32); + PUTU32(p, v); + v = (u32)(Z.hi); + PUTU32(p + 4, v); + v = (u32)(Z.lo >> 32); + PUTU32(p + 8, v); + v = (u32)(Z.lo); + PUTU32(p + 12, v); +# endif + } else { + Xi[0] = Z.hi; + Xi[1] = Z.lo; + } } -#define GCM_MUL(ctx,Xi) gcm_gmult_8bit(ctx->Xi.u,ctx->Htable) -#elif TABLE_BITS==4 +# define GCM_MUL(ctx,Xi) gcm_gmult_8bit(ctx->Xi.u,ctx->Htable) + +#elif TABLE_BITS==4 static void gcm_init_4bit(u128 Htable[16], u64 H[2]) { - u128 V; -#if defined(OPENSSL_SMALL_FOOTPRINT) - int i; -#endif + u128 V; +# if defined(OPENSSL_SMALL_FOOTPRINT) + int i; +# endif - Htable[0].hi = 0; - Htable[0].lo = 0; - V.hi = H[0]; - V.lo = H[1]; - -#if defined(OPENSSL_SMALL_FOOTPRINT) - for (Htable[8]=V, i=4; i>0; i>>=1) { - REDUCE1BIT(V); - Htable[i] = V; - } - - for (i=2; i<16; i<<=1) { - u128 *Hi = Htable+i; - int j; - for (V=*Hi, j=1; j<i; ++j) { - Hi[j].hi = V.hi^Htable[j].hi; - Hi[j].lo = V.lo^Htable[j].lo; - } - } -#else - Htable[8] = V; - REDUCE1BIT(V); - Htable[4] = V; - REDUCE1BIT(V); - Htable[2] = V; - REDUCE1BIT(V); - Htable[1] = V; - Htable[3].hi = V.hi^Htable[2].hi, Htable[3].lo = V.lo^Htable[2].lo; - V=Htable[4]; - Htable[5].hi = V.hi^Htable[1].hi, Htable[5].lo = V.lo^Htable[1].lo; - Htable[6].hi = V.hi^Htable[2].hi, Htable[6].lo = V.lo^Htable[2].lo; - Htable[7].hi = V.hi^Htable[3].hi, Htable[7].lo = V.lo^Htable[3].lo; - V=Htable[8]; - Htable[9].hi = V.hi^Htable[1].hi, Htable[9].lo = V.lo^Htable[1].lo; - Htable[10].hi = V.hi^Htable[2].hi, Htable[10].lo = V.lo^Htable[2].lo; - Htable[11].hi = V.hi^Htable[3].hi, Htable[11].lo = V.lo^Htable[3].lo; - Htable[12].hi = V.hi^Htable[4].hi, Htable[12].lo = V.lo^Htable[4].lo; - Htable[13].hi = V.hi^Htable[5].hi, Htable[13].lo = V.lo^Htable[5].lo; - Htable[14].hi = V.hi^Htable[6].hi, Htable[14].lo = V.lo^Htable[6].lo; - Htable[15].hi = V.hi^Htable[7].hi, Htable[15].lo = V.lo^Htable[7].lo; -#endif -#if defined(GHASH_ASM) && (defined(__arm__) || defined(__arm)) - /* - * ARM assembler expects specific dword order in Htable. - */ - { - int j; - const union { long one; char little; } is_endian = {1}; - - if (is_endian.little) - for (j=0;j<16;++j) { - V = Htable[j]; - Htable[j].hi = V.lo; - Htable[j].lo = V.hi; - } - else - for (j=0;j<16;++j) { - V = Htable[j]; - Htable[j].hi = V.lo<<32|V.lo>>32; - Htable[j].lo = V.hi<<32|V.hi>>32; - } - } -#endif + Htable[0].hi = 0; + Htable[0].lo = 0; + V.hi = H[0]; + V.lo = H[1]; + +# if defined(OPENSSL_SMALL_FOOTPRINT) + for (Htable[8] = V, i = 4; i > 0; i >>= 1) { + REDUCE1BIT(V); + Htable[i] = V; + } + + for (i = 2; i < 16; i <<= 1) { + u128 *Hi = Htable + i; + int j; + for (V = *Hi, j = 1; j < i; ++j) { + Hi[j].hi = V.hi ^ Htable[j].hi; + Hi[j].lo = V.lo ^ Htable[j].lo; + } + } +# else + Htable[8] = V; + REDUCE1BIT(V); + Htable[4] = V; + REDUCE1BIT(V); + Htable[2] = V; + REDUCE1BIT(V); + Htable[1] = V; + Htable[3].hi = V.hi ^ Htable[2].hi, Htable[3].lo = V.lo ^ Htable[2].lo; + V = Htable[4]; + Htable[5].hi = V.hi ^ Htable[1].hi, Htable[5].lo = V.lo ^ Htable[1].lo; + Htable[6].hi = V.hi ^ Htable[2].hi, Htable[6].lo = V.lo ^ Htable[2].lo; + Htable[7].hi = V.hi ^ Htable[3].hi, Htable[7].lo = V.lo ^ Htable[3].lo; + V = Htable[8]; + Htable[9].hi = V.hi ^ Htable[1].hi, Htable[9].lo = V.lo ^ Htable[1].lo; + Htable[10].hi = V.hi ^ Htable[2].hi, Htable[10].lo = V.lo ^ Htable[2].lo; + Htable[11].hi = V.hi ^ Htable[3].hi, Htable[11].lo = V.lo ^ Htable[3].lo; + Htable[12].hi = V.hi ^ Htable[4].hi, Htable[12].lo = V.lo ^ Htable[4].lo; + Htable[13].hi = V.hi ^ Htable[5].hi, Htable[13].lo = V.lo ^ Htable[5].lo; + Htable[14].hi = V.hi ^ Htable[6].hi, Htable[14].lo = V.lo ^ Htable[6].lo; + Htable[15].hi = V.hi ^ Htable[7].hi, Htable[15].lo = V.lo ^ Htable[7].lo; +# endif +# if defined(GHASH_ASM) && (defined(__arm__) || defined(__arm)) + /* + * ARM assembler expects specific dword order in Htable. + */ + { + int j; + const union { + long one; + char little; + } is_endian = { + 1 + }; + + if (is_endian.little) + for (j = 0; j < 16; ++j) { + V = Htable[j]; + Htable[j].hi = V.lo; + Htable[j].lo = V.hi; + } else + for (j = 0; j < 16; ++j) { + V = Htable[j]; + Htable[j].hi = V.lo << 32 | V.lo >> 32; + Htable[j].lo = V.hi << 32 | V.hi >> 32; + } + } +# endif } -#ifndef GHASH_ASM +# ifndef GHASH_ASM static const size_t rem_4bit[16] = { - PACK(0x0000), PACK(0x1C20), PACK(0x3840), PACK(0x2460), - PACK(0x7080), PACK(0x6CA0), PACK(0x48C0), PACK(0x54E0), - PACK(0xE100), PACK(0xFD20), PACK(0xD940), PACK(0xC560), - PACK(0x9180), PACK(0x8DA0), PACK(0xA9C0), PACK(0xB5E0) }; + PACK(0x0000), PACK(0x1C20), PACK(0x3840), PACK(0x2460), + PACK(0x7080), PACK(0x6CA0), PACK(0x48C0), PACK(0x54E0), + PACK(0xE100), PACK(0xFD20), PACK(0xD940), PACK(0xC560), + PACK(0x9180), PACK(0x8DA0), PACK(0xA9C0), PACK(0xB5E0) +}; static void gcm_gmult_4bit(u64 Xi[2], const u128 Htable[16]) { - u128 Z; - int cnt = 15; - size_t rem, nlo, nhi; - const union { long one; char little; } is_endian = {1}; - - nlo = ((const u8 *)Xi)[15]; - nhi = nlo>>4; - nlo &= 0xf; - - Z.hi = Htable[nlo].hi; - Z.lo = Htable[nlo].lo; - - while (1) { - rem = (size_t)Z.lo&0xf; - Z.lo = (Z.hi<<60)|(Z.lo>>4); - Z.hi = (Z.hi>>4); - if (sizeof(size_t)==8) - Z.hi ^= rem_4bit[rem]; - else - Z.hi ^= (u64)rem_4bit[rem]<<32; - - Z.hi ^= Htable[nhi].hi; - Z.lo ^= Htable[nhi].lo; - - if (--cnt<0) break; - - nlo = ((const u8 *)Xi)[cnt]; - nhi = nlo>>4; - nlo &= 0xf; - - rem = (size_t)Z.lo&0xf; - Z.lo = (Z.hi<<60)|(Z.lo>>4); - Z.hi = (Z.hi>>4); - if (sizeof(size_t)==8) - Z.hi ^= rem_4bit[rem]; - else - Z.hi ^= (u64)rem_4bit[rem]<<32; - - Z.hi ^= Htable[nlo].hi; - Z.lo ^= Htable[nlo].lo; - } - - if (is_endian.little) { -#ifdef BSWAP8 - Xi[0] = BSWAP8(Z.hi); - Xi[1] = BSWAP8(Z.lo); -#else - u8 *p = (u8 *)Xi; - u32 v; - v = (u32)(Z.hi>>32); PUTU32(p,v); - v = (u32)(Z.hi); PUTU32(p+4,v); - v = (u32)(Z.lo>>32); PUTU32(p+8,v); - v = (u32)(Z.lo); PUTU32(p+12,v); -#endif - } - else { - Xi[0] = Z.hi; - Xi[1] = Z.lo; - } + u128 Z; + int cnt = 15; + size_t rem, nlo, nhi; + const union { + long one; + char little; + } is_endian = { + 1 + }; + + nlo = ((const u8 *)Xi)[15]; + nhi = nlo >> 4; + nlo &= 0xf; + + Z.hi = Htable[nlo].hi; + Z.lo = Htable[nlo].lo; + + while (1) { + rem = (size_t)Z.lo & 0xf; + Z.lo = (Z.hi << 60) | (Z.lo >> 4); + Z.hi = (Z.hi >> 4); + if (sizeof(size_t) == 8) + Z.hi ^= rem_4bit[rem]; + else + Z.hi ^= (u64)rem_4bit[rem] << 32; + + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + + if (--cnt < 0) + break; + + nlo = ((const u8 *)Xi)[cnt]; + nhi = nlo >> 4; + nlo &= 0xf; + + rem = (size_t)Z.lo & 0xf; + Z.lo = (Z.hi << 60) | (Z.lo >> 4); + Z.hi = (Z.hi >> 4); + if (sizeof(size_t) == 8) + Z.hi ^= rem_4bit[rem]; + else + Z.hi ^= (u64)rem_4bit[rem] << 32; + + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + } + + if (is_endian.little) { +# ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +# else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi >> 32); + PUTU32(p, v); + v = (u32)(Z.hi); + PUTU32(p + 4, v); + v = (u32)(Z.lo >> 32); + PUTU32(p + 8, v); + v = (u32)(Z.lo); + PUTU32(p + 12, v); +# endif + } else { + Xi[0] = Z.hi; + Xi[1] = Z.lo; + } } -#if !defined(OPENSSL_SMALL_FOOTPRINT) +# if !defined(OPENSSL_SMALL_FOOTPRINT) /* * Streamed gcm_mult_4bit, see CRYPTO_gcm128_[en|de]crypt for * details... Compiler-generated code doesn't seem to give any @@ -403,1503 +428,1862 @@ static void gcm_gmult_4bit(u64 Xi[2], const u128 Htable[16]) * mostly as reference and a placeholder for possible future * non-trivial optimization[s]... */ -static void gcm_ghash_4bit(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) +static void gcm_ghash_4bit(u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) { u128 Z; int cnt; size_t rem, nlo, nhi; - const union { long one; char little; } is_endian = {1}; - -#if 1 + const union { + long one; + char little; + } is_endian = { + 1 + }; + +# if 1 do { - cnt = 15; - nlo = ((const u8 *)Xi)[15]; - nlo ^= inp[15]; - nhi = nlo>>4; - nlo &= 0xf; - - Z.hi = Htable[nlo].hi; - Z.lo = Htable[nlo].lo; - - while (1) { - rem = (size_t)Z.lo&0xf; - Z.lo = (Z.hi<<60)|(Z.lo>>4); - Z.hi = (Z.hi>>4); - if (sizeof(size_t)==8) - Z.hi ^= rem_4bit[rem]; - else - Z.hi ^= (u64)rem_4bit[rem]<<32; - - Z.hi ^= Htable[nhi].hi; - Z.lo ^= Htable[nhi].lo; - - if (--cnt<0) break; - - nlo = ((const u8 *)Xi)[cnt]; - nlo ^= inp[cnt]; - nhi = nlo>>4; - nlo &= 0xf; - - rem = (size_t)Z.lo&0xf; - Z.lo = (Z.hi<<60)|(Z.lo>>4); - Z.hi = (Z.hi>>4); - if (sizeof(size_t)==8) - Z.hi ^= rem_4bit[rem]; - else - Z.hi ^= (u64)rem_4bit[rem]<<32; - - Z.hi ^= Htable[nlo].hi; - Z.lo ^= Htable[nlo].lo; - } -#else + cnt = 15; + nlo = ((const u8 *)Xi)[15]; + nlo ^= inp[15]; + nhi = nlo >> 4; + nlo &= 0xf; + + Z.hi = Htable[nlo].hi; + Z.lo = Htable[nlo].lo; + + while (1) { + rem = (size_t)Z.lo & 0xf; + Z.lo = (Z.hi << 60) | (Z.lo >> 4); + Z.hi = (Z.hi >> 4); + if (sizeof(size_t) == 8) + Z.hi ^= rem_4bit[rem]; + else + Z.hi ^= (u64)rem_4bit[rem] << 32; + + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + + if (--cnt < 0) + break; + + nlo = ((const u8 *)Xi)[cnt]; + nlo ^= inp[cnt]; + nhi = nlo >> 4; + nlo &= 0xf; + + rem = (size_t)Z.lo & 0xf; + Z.lo = (Z.hi << 60) | (Z.lo >> 4); + Z.hi = (Z.hi >> 4); + if (sizeof(size_t) == 8) + Z.hi ^= rem_4bit[rem]; + else + Z.hi ^= (u64)rem_4bit[rem] << 32; + + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; + } +# else /* * Extra 256+16 bytes per-key plus 512 bytes shared tables * [should] give ~50% improvement... One could have PACK()-ed * the rem_8bit even here, but the priority is to minimize * cache footprint... - */ - u128 Hshr4[16]; /* Htable shifted right by 4 bits */ - u8 Hshl4[16]; /* Htable shifted left by 4 bits */ + */ + u128 Hshr4[16]; /* Htable shifted right by 4 bits */ + u8 Hshl4[16]; /* Htable shifted left by 4 bits */ static const unsigned short rem_8bit[256] = { - 0x0000, 0x01C2, 0x0384, 0x0246, 0x0708, 0x06CA, 0x048C, 0x054E, - 0x0E10, 0x0FD2, 0x0D94, 0x0C56, 0x0918, 0x08DA, 0x0A9C, 0x0B5E, - 0x1C20, 0x1DE2, 0x1FA4, 0x1E66, 0x1B28, 0x1AEA, 0x18AC, 0x196E, - 0x1230, 0x13F2, 0x11B4, 0x1076, 0x1538, 0x14FA, 0x16BC, 0x177E, - 0x3840, 0x3982, 0x3BC4, 0x3A06, 0x3F48, 0x3E8A, 0x3CCC, 0x3D0E, - 0x3650, 0x3792, 0x35D4, 0x3416, 0x3158, 0x309A, 0x32DC, 0x331E, - 0x2460, 0x25A2, 0x27E4, 0x2626, 0x2368, 0x22AA, 0x20EC, 0x212E, - 0x2A70, 0x2BB2, 0x29F4, 0x2836, 0x2D78, 0x2CBA, 0x2EFC, 0x2F3E, - 0x7080, 0x7142, 0x7304, 0x72C6, 0x7788, 0x764A, 0x740C, 0x75CE, - 0x7E90, 0x7F52, 0x7D14, 0x7CD6, 0x7998, 0x785A, 0x7A1C, 0x7BDE, - 0x6CA0, 0x6D62, 0x6F24, 0x6EE6, 0x6BA8, 0x6A6A, 0x682C, 0x69EE, - 0x62B0, 0x6372, 0x6134, 0x60F6, 0x65B8, 0x647A, 0x663C, 0x67FE, - 0x48C0, 0x4902, 0x4B44, 0x4A86, 0x4FC8, 0x4E0A, 0x4C4C, 0x4D8E, - 0x46D0, 0x4712, 0x4554, 0x4496, 0x41D8, 0x401A, 0x425C, 0x439E, - 0x54E0, 0x5522, 0x5764, 0x56A6, 0x53E8, 0x522A, 0x506C, 0x51AE, - 0x5AF0, 0x5B32, 0x5974, 0x58B6, 0x5DF8, 0x5C3A, 0x5E7C, 0x5FBE, - 0xE100, 0xE0C2, 0xE284, 0xE346, 0xE608, 0xE7CA, 0xE58C, 0xE44E, - 0xEF10, 0xEED2, 0xEC94, 0xED56, 0xE818, 0xE9DA, 0xEB9C, 0xEA5E, - 0xFD20, 0xFCE2, 0xFEA4, 0xFF66, 0xFA28, 0xFBEA, 0xF9AC, 0xF86E, - 0xF330, 0xF2F2, 0xF0B4, 0xF176, 0xF438, 0xF5FA, 0xF7BC, 0xF67E, - 0xD940, 0xD882, 0xDAC4, 0xDB06, 0xDE48, 0xDF8A, 0xDDCC, 0xDC0E, - 0xD750, 0xD692, 0xD4D4, 0xD516, 0xD058, 0xD19A, 0xD3DC, 0xD21E, - 0xC560, 0xC4A2, 0xC6E4, 0xC726, 0xC268, 0xC3AA, 0xC1EC, 0xC02E, - 0xCB70, 0xCAB2, 0xC8F4, 0xC936, 0xCC78, 0xCDBA, 0xCFFC, 0xCE3E, - 0x9180, 0x9042, 0x9204, 0x93C6, 0x9688, 0x974A, 0x950C, 0x94CE, - 0x9F90, 0x9E52, 0x9C14, 0x9DD6, 0x9898, 0x995A, 0x9B1C, 0x9ADE, - 0x8DA0, 0x8C62, 0x8E24, 0x8FE6, 0x8AA8, 0x8B6A, 0x892C, 0x88EE, - 0x83B0, 0x8272, 0x8034, 0x81F6, 0x84B8, 0x857A, 0x873C, 0x86FE, - 0xA9C0, 0xA802, 0xAA44, 0xAB86, 0xAEC8, 0xAF0A, 0xAD4C, 0xAC8E, - 0xA7D0, 0xA612, 0xA454, 0xA596, 0xA0D8, 0xA11A, 0xA35C, 0xA29E, - 0xB5E0, 0xB422, 0xB664, 0xB7A6, 0xB2E8, 0xB32A, 0xB16C, 0xB0AE, - 0xBBF0, 0xBA32, 0xB874, 0xB9B6, 0xBCF8, 0xBD3A, 0xBF7C, 0xBEBE }; + 0x0000, 0x01C2, 0x0384, 0x0246, 0x0708, 0x06CA, 0x048C, 0x054E, + 0x0E10, 0x0FD2, 0x0D94, 0x0C56, 0x0918, 0x08DA, 0x0A9C, 0x0B5E, + 0x1C20, 0x1DE2, 0x1FA4, 0x1E66, 0x1B28, 0x1AEA, 0x18AC, 0x196E, + 0x1230, 0x13F2, 0x11B4, 0x1076, 0x1538, 0x14FA, 0x16BC, 0x177E, + 0x3840, 0x3982, 0x3BC4, 0x3A06, 0x3F48, 0x3E8A, 0x3CCC, 0x3D0E, + 0x3650, 0x3792, 0x35D4, 0x3416, 0x3158, 0x309A, 0x32DC, 0x331E, + 0x2460, 0x25A2, 0x27E4, 0x2626, 0x2368, 0x22AA, 0x20EC, 0x212E, + 0x2A70, 0x2BB2, 0x29F4, 0x2836, 0x2D78, 0x2CBA, 0x2EFC, 0x2F3E, + 0x7080, 0x7142, 0x7304, 0x72C6, 0x7788, 0x764A, 0x740C, 0x75CE, + 0x7E90, 0x7F52, 0x7D14, 0x7CD6, 0x7998, 0x785A, 0x7A1C, 0x7BDE, + 0x6CA0, 0x6D62, 0x6F24, 0x6EE6, 0x6BA8, 0x6A6A, 0x682C, 0x69EE, + 0x62B0, 0x6372, 0x6134, 0x60F6, 0x65B8, 0x647A, 0x663C, 0x67FE, + 0x48C0, 0x4902, 0x4B44, 0x4A86, 0x4FC8, 0x4E0A, 0x4C4C, 0x4D8E, + 0x46D0, 0x4712, 0x4554, 0x4496, 0x41D8, 0x401A, 0x425C, 0x439E, + 0x54E0, 0x5522, 0x5764, 0x56A6, 0x53E8, 0x522A, 0x506C, 0x51AE, + 0x5AF0, 0x5B32, 0x5974, 0x58B6, 0x5DF8, 0x5C3A, 0x5E7C, 0x5FBE, + 0xE100, 0xE0C2, 0xE284, 0xE346, 0xE608, 0xE7CA, 0xE58C, 0xE44E, + 0xEF10, 0xEED2, 0xEC94, 0xED56, 0xE818, 0xE9DA, 0xEB9C, 0xEA5E, + 0xFD20, 0xFCE2, 0xFEA4, 0xFF66, 0xFA28, 0xFBEA, 0xF9AC, 0xF86E, + 0xF330, 0xF2F2, 0xF0B4, 0xF176, 0xF438, 0xF5FA, 0xF7BC, 0xF67E, + 0xD940, 0xD882, 0xDAC4, 0xDB06, 0xDE48, 0xDF8A, 0xDDCC, 0xDC0E, + 0xD750, 0xD692, 0xD4D4, 0xD516, 0xD058, 0xD19A, 0xD3DC, 0xD21E, + 0xC560, 0xC4A2, 0xC6E4, 0xC726, 0xC268, 0xC3AA, 0xC1EC, 0xC02E, + 0xCB70, 0xCAB2, 0xC8F4, 0xC936, 0xCC78, 0xCDBA, 0xCFFC, 0xCE3E, + 0x9180, 0x9042, 0x9204, 0x93C6, 0x9688, 0x974A, 0x950C, 0x94CE, + 0x9F90, 0x9E52, 0x9C14, 0x9DD6, 0x9898, 0x995A, 0x9B1C, 0x9ADE, + 0x8DA0, 0x8C62, 0x8E24, 0x8FE6, 0x8AA8, 0x8B6A, 0x892C, 0x88EE, + 0x83B0, 0x8272, 0x8034, 0x81F6, 0x84B8, 0x857A, 0x873C, 0x86FE, + 0xA9C0, 0xA802, 0xAA44, 0xAB86, 0xAEC8, 0xAF0A, 0xAD4C, 0xAC8E, + 0xA7D0, 0xA612, 0xA454, 0xA596, 0xA0D8, 0xA11A, 0xA35C, 0xA29E, + 0xB5E0, 0xB422, 0xB664, 0xB7A6, 0xB2E8, 0xB32A, 0xB16C, 0xB0AE, + 0xBBF0, 0xBA32, 0xB874, 0xB9B6, 0xBCF8, 0xBD3A, 0xBF7C, 0xBEBE + }; /* * This pre-processing phase slows down procedure by approximately * same time as it makes each loop spin faster. In other words * single block performance is approximately same as straightforward * "4-bit" implementation, and then it goes only faster... */ - for (cnt=0; cnt<16; ++cnt) { - Z.hi = Htable[cnt].hi; - Z.lo = Htable[cnt].lo; - Hshr4[cnt].lo = (Z.hi<<60)|(Z.lo>>4); - Hshr4[cnt].hi = (Z.hi>>4); - Hshl4[cnt] = (u8)(Z.lo<<4); + for (cnt = 0; cnt < 16; ++cnt) { + Z.hi = Htable[cnt].hi; + Z.lo = Htable[cnt].lo; + Hshr4[cnt].lo = (Z.hi << 60) | (Z.lo >> 4); + Hshr4[cnt].hi = (Z.hi >> 4); + Hshl4[cnt] = (u8)(Z.lo << 4); } do { - for (Z.lo=0, Z.hi=0, cnt=15; cnt; --cnt) { - nlo = ((const u8 *)Xi)[cnt]; - nlo ^= inp[cnt]; - nhi = nlo>>4; - nlo &= 0xf; + for (Z.lo = 0, Z.hi = 0, cnt = 15; cnt; --cnt) { + nlo = ((const u8 *)Xi)[cnt]; + nlo ^= inp[cnt]; + nhi = nlo >> 4; + nlo &= 0xf; - Z.hi ^= Htable[nlo].hi; - Z.lo ^= Htable[nlo].lo; + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; - rem = (size_t)Z.lo&0xff; + rem = (size_t)Z.lo & 0xff; - Z.lo = (Z.hi<<56)|(Z.lo>>8); - Z.hi = (Z.hi>>8); + Z.lo = (Z.hi << 56) | (Z.lo >> 8); + Z.hi = (Z.hi >> 8); - Z.hi ^= Hshr4[nhi].hi; - Z.lo ^= Hshr4[nhi].lo; - Z.hi ^= (u64)rem_8bit[rem^Hshl4[nhi]]<<48; - } + Z.hi ^= Hshr4[nhi].hi; + Z.lo ^= Hshr4[nhi].lo; + Z.hi ^= (u64)rem_8bit[rem ^ Hshl4[nhi]] << 48; + } - nlo = ((const u8 *)Xi)[0]; - nlo ^= inp[0]; - nhi = nlo>>4; - nlo &= 0xf; + nlo = ((const u8 *)Xi)[0]; + nlo ^= inp[0]; + nhi = nlo >> 4; + nlo &= 0xf; - Z.hi ^= Htable[nlo].hi; - Z.lo ^= Htable[nlo].lo; + Z.hi ^= Htable[nlo].hi; + Z.lo ^= Htable[nlo].lo; - rem = (size_t)Z.lo&0xf; + rem = (size_t)Z.lo & 0xf; - Z.lo = (Z.hi<<60)|(Z.lo>>4); - Z.hi = (Z.hi>>4); + Z.lo = (Z.hi << 60) | (Z.lo >> 4); + Z.hi = (Z.hi >> 4); - Z.hi ^= Htable[nhi].hi; - Z.lo ^= Htable[nhi].lo; - Z.hi ^= ((u64)rem_8bit[rem<<4])<<48; -#endif + Z.hi ^= Htable[nhi].hi; + Z.lo ^= Htable[nhi].lo; + Z.hi ^= ((u64)rem_8bit[rem << 4]) << 48; +# endif - if (is_endian.little) { -#ifdef BSWAP8 - Xi[0] = BSWAP8(Z.hi); - Xi[1] = BSWAP8(Z.lo); -#else - u8 *p = (u8 *)Xi; - u32 v; - v = (u32)(Z.hi>>32); PUTU32(p,v); - v = (u32)(Z.hi); PUTU32(p+4,v); - v = (u32)(Z.lo>>32); PUTU32(p+8,v); - v = (u32)(Z.lo); PUTU32(p+12,v); -#endif - } - else { - Xi[0] = Z.hi; - Xi[1] = Z.lo; - } - } while (inp+=16, len-=16); + if (is_endian.little) { +# ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +# else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi >> 32); + PUTU32(p, v); + v = (u32)(Z.hi); + PUTU32(p + 4, v); + v = (u32)(Z.lo >> 32); + PUTU32(p + 8, v); + v = (u32)(Z.lo); + PUTU32(p + 12, v); +# endif + } else { + Xi[0] = Z.hi; + Xi[1] = Z.lo; + } + } while (inp += 16, len -= 16); } -#endif -#else -void gcm_gmult_4bit(u64 Xi[2],const u128 Htable[16]); -void gcm_ghash_4bit(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); -#endif +# endif +# else +void gcm_gmult_4bit(u64 Xi[2], const u128 Htable[16]); +void gcm_ghash_4bit(u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); +# endif -#define GCM_MUL(ctx,Xi) gcm_gmult_4bit(ctx->Xi.u,ctx->Htable) -#if defined(GHASH_ASM) || !defined(OPENSSL_SMALL_FOOTPRINT) -#define GHASH(ctx,in,len) gcm_ghash_4bit((ctx)->Xi.u,(ctx)->Htable,in,len) -/* GHASH_CHUNK is "stride parameter" missioned to mitigate cache - * trashing effect. In other words idea is to hash data while it's - * still in L1 cache after encryption pass... */ -#define GHASH_CHUNK (3*1024) -#endif +# define GCM_MUL(ctx,Xi) gcm_gmult_4bit(ctx->Xi.u,ctx->Htable) +# if defined(GHASH_ASM) || !defined(OPENSSL_SMALL_FOOTPRINT) +# define GHASH(ctx,in,len) gcm_ghash_4bit((ctx)->Xi.u,(ctx)->Htable,in,len) +/* + * GHASH_CHUNK is "stride parameter" missioned to mitigate cache trashing + * effect. In other words idea is to hash data while it's still in L1 cache + * after encryption pass... + */ +# define GHASH_CHUNK (3*1024) +# endif -#else /* TABLE_BITS */ +#else /* TABLE_BITS */ -static void gcm_gmult_1bit(u64 Xi[2],const u64 H[2]) +static void gcm_gmult_1bit(u64 Xi[2], const u64 H[2]) { - u128 V,Z = { 0,0 }; - long X; - int i,j; - const long *xi = (const long *)Xi; - const union { long one; char little; } is_endian = {1}; - - V.hi = H[0]; /* H is in host byte order, no byte swapping */ - V.lo = H[1]; - - for (j=0; j<16/sizeof(long); ++j) { - if (is_endian.little) { - if (sizeof(long)==8) { -#ifdef BSWAP8 - X = (long)(BSWAP8(xi[j])); -#else - const u8 *p = (const u8 *)(xi+j); - X = (long)((u64)GETU32(p)<<32|GETU32(p+4)); -#endif - } - else { - const u8 *p = (const u8 *)(xi+j); - X = (long)GETU32(p); - } - } - else - X = xi[j]; - - for (i=0; i<8*sizeof(long); ++i, X<<=1) { - u64 M = (u64)(X>>(8*sizeof(long)-1)); - Z.hi ^= V.hi&M; - Z.lo ^= V.lo&M; - - REDUCE1BIT(V); - } - } - - if (is_endian.little) { -#ifdef BSWAP8 - Xi[0] = BSWAP8(Z.hi); - Xi[1] = BSWAP8(Z.lo); -#else - u8 *p = (u8 *)Xi; - u32 v; - v = (u32)(Z.hi>>32); PUTU32(p,v); - v = (u32)(Z.hi); PUTU32(p+4,v); - v = (u32)(Z.lo>>32); PUTU32(p+8,v); - v = (u32)(Z.lo); PUTU32(p+12,v); -#endif - } - else { - Xi[0] = Z.hi; - Xi[1] = Z.lo; - } + u128 V, Z = { 0, 0 }; + long X; + int i, j; + const long *xi = (const long *)Xi; + const union { + long one; + char little; + } is_endian = { + 1 + }; + + V.hi = H[0]; /* H is in host byte order, no byte swapping */ + V.lo = H[1]; + + for (j = 0; j < 16 / sizeof(long); ++j) { + if (is_endian.little) { + if (sizeof(long) == 8) { +# ifdef BSWAP8 + X = (long)(BSWAP8(xi[j])); +# else + const u8 *p = (const u8 *)(xi + j); + X = (long)((u64)GETU32(p) << 32 | GETU32(p + 4)); +# endif + } else { + const u8 *p = (const u8 *)(xi + j); + X = (long)GETU32(p); + } + } else + X = xi[j]; + + for (i = 0; i < 8 * sizeof(long); ++i, X <<= 1) { + u64 M = (u64)(X >> (8 * sizeof(long) - 1)); + Z.hi ^= V.hi & M; + Z.lo ^= V.lo & M; + + REDUCE1BIT(V); + } + } + + if (is_endian.little) { +# ifdef BSWAP8 + Xi[0] = BSWAP8(Z.hi); + Xi[1] = BSWAP8(Z.lo); +# else + u8 *p = (u8 *)Xi; + u32 v; + v = (u32)(Z.hi >> 32); + PUTU32(p, v); + v = (u32)(Z.hi); + PUTU32(p + 4, v); + v = (u32)(Z.lo >> 32); + PUTU32(p + 8, v); + v = (u32)(Z.lo); + PUTU32(p + 12, v); +# endif + } else { + Xi[0] = Z.hi; + Xi[1] = Z.lo; + } } -#define GCM_MUL(ctx,Xi) gcm_gmult_1bit(ctx->Xi.u,ctx->H.u) + +# define GCM_MUL(ctx,Xi) gcm_gmult_1bit(ctx->Xi.u,ctx->H.u) #endif -#if TABLE_BITS==4 && defined(GHASH_ASM) -# if !defined(I386_ONLY) && \ - (defined(__i386) || defined(__i386__) || \ - defined(__x86_64) || defined(__x86_64__) || \ - defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64)) +#if TABLE_BITS==4 && defined(GHASH_ASM) +# if !defined(I386_ONLY) && \ + (defined(__i386) || defined(__i386__) || \ + defined(__x86_64) || defined(__x86_64__) || \ + defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64)) # define GHASH_ASM_X86_OR_64 # define GCM_FUNCREF_4BIT extern unsigned int OPENSSL_ia32cap_P[2]; -void gcm_init_clmul(u128 Htable[16],const u64 Xi[2]); -void gcm_gmult_clmul(u64 Xi[2],const u128 Htable[16]); -void gcm_ghash_clmul(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +void gcm_init_clmul(u128 Htable[16], const u64 Xi[2]); +void gcm_gmult_clmul(u64 Xi[2], const u128 Htable[16]); +void gcm_ghash_clmul(u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); -# if defined(__i386) || defined(__i386__) || defined(_M_IX86) +# if defined(__i386) || defined(__i386__) || defined(_M_IX86) # define GHASH_ASM_X86 -void gcm_gmult_4bit_mmx(u64 Xi[2],const u128 Htable[16]); -void gcm_ghash_4bit_mmx(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +void gcm_gmult_4bit_mmx(u64 Xi[2], const u128 Htable[16]); +void gcm_ghash_4bit_mmx(u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); -void gcm_gmult_4bit_x86(u64 Xi[2],const u128 Htable[16]); -void gcm_ghash_4bit_x86(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +void gcm_gmult_4bit_x86(u64 Xi[2], const u128 Htable[16]); +void gcm_ghash_4bit_x86(u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); # endif # elif defined(__arm__) || defined(__arm) # include "arm_arch.h" # if __ARM_ARCH__>=7 # define GHASH_ASM_ARM # define GCM_FUNCREF_4BIT -void gcm_gmult_neon(u64 Xi[2],const u128 Htable[16]); -void gcm_ghash_neon(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); +void gcm_gmult_neon(u64 Xi[2], const u128 Htable[16]); +void gcm_ghash_neon(u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); # endif # endif #endif #ifdef GCM_FUNCREF_4BIT # undef GCM_MUL -# define GCM_MUL(ctx,Xi) (*gcm_gmult_p)(ctx->Xi.u,ctx->Htable) +# define GCM_MUL(ctx,Xi) (*gcm_gmult_p)(ctx->Xi.u,ctx->Htable) # ifdef GHASH # undef GHASH -# define GHASH(ctx,in,len) (*gcm_ghash_p)(ctx->Xi.u,ctx->Htable,in,len) +# define GHASH(ctx,in,len) (*gcm_ghash_p)(ctx->Xi.u,ctx->Htable,in,len) # endif #endif -void CRYPTO_gcm128_init(GCM128_CONTEXT *ctx,void *key,block128_f block) +void CRYPTO_gcm128_init(GCM128_CONTEXT *ctx, void *key, block128_f block) { - const union { long one; char little; } is_endian = {1}; + const union { + long one; + char little; + } is_endian = { + 1 + }; - memset(ctx,0,sizeof(*ctx)); - ctx->block = block; - ctx->key = key; + memset(ctx, 0, sizeof(*ctx)); + ctx->block = block; + ctx->key = key; - (*block)(ctx->H.c,ctx->H.c,key); + (*block) (ctx->H.c, ctx->H.c, key); - if (is_endian.little) { - /* H is stored in host byte order */ + if (is_endian.little) { + /* H is stored in host byte order */ #ifdef BSWAP8 - ctx->H.u[0] = BSWAP8(ctx->H.u[0]); - ctx->H.u[1] = BSWAP8(ctx->H.u[1]); + ctx->H.u[0] = BSWAP8(ctx->H.u[0]); + ctx->H.u[1] = BSWAP8(ctx->H.u[1]); #else - u8 *p = ctx->H.c; - u64 hi,lo; - hi = (u64)GETU32(p) <<32|GETU32(p+4); - lo = (u64)GETU32(p+8)<<32|GETU32(p+12); - ctx->H.u[0] = hi; - ctx->H.u[1] = lo; + u8 *p = ctx->H.c; + u64 hi, lo; + hi = (u64)GETU32(p) << 32 | GETU32(p + 4); + lo = (u64)GETU32(p + 8) << 32 | GETU32(p + 12); + ctx->H.u[0] = hi; + ctx->H.u[1] = lo; #endif - } - -#if TABLE_BITS==8 - gcm_init_8bit(ctx->Htable,ctx->H.u); -#elif TABLE_BITS==4 -# if defined(GHASH_ASM_X86_OR_64) -# if !defined(GHASH_ASM_X86) || defined(OPENSSL_IA32_SSE2) - if (OPENSSL_ia32cap_P[0]&(1<<24) && /* check FXSR bit */ - OPENSSL_ia32cap_P[1]&(1<<1) ) { /* check PCLMULQDQ bit */ - gcm_init_clmul(ctx->Htable,ctx->H.u); - ctx->gmult = gcm_gmult_clmul; - ctx->ghash = gcm_ghash_clmul; - return; - } + } +#if TABLE_BITS==8 + gcm_init_8bit(ctx->Htable, ctx->H.u); +#elif TABLE_BITS==4 +# if defined(GHASH_ASM_X86_OR_64) +# if !defined(GHASH_ASM_X86) || defined(OPENSSL_IA32_SSE2) + if (OPENSSL_ia32cap_P[0] & (1 << 24) && /* check FXSR bit */ + OPENSSL_ia32cap_P[1] & (1 << 1)) { /* check PCLMULQDQ bit */ + gcm_init_clmul(ctx->Htable, ctx->H.u); + ctx->gmult = gcm_gmult_clmul; + ctx->ghash = gcm_ghash_clmul; + return; + } # endif - gcm_init_4bit(ctx->Htable,ctx->H.u); -# if defined(GHASH_ASM_X86) /* x86 only */ -# if defined(OPENSSL_IA32_SSE2) - if (OPENSSL_ia32cap_P[0]&(1<<25)) { /* check SSE bit */ + gcm_init_4bit(ctx->Htable, ctx->H.u); +# if defined(GHASH_ASM_X86) /* x86 only */ +# if defined(OPENSSL_IA32_SSE2) + if (OPENSSL_ia32cap_P[0] & (1 << 25)) { /* check SSE bit */ # else - if (OPENSSL_ia32cap_P[0]&(1<<23)) { /* check MMX bit */ + if (OPENSSL_ia32cap_P[0] & (1 << 23)) { /* check MMX bit */ # endif - ctx->gmult = gcm_gmult_4bit_mmx; - ctx->ghash = gcm_ghash_4bit_mmx; - } else { - ctx->gmult = gcm_gmult_4bit_x86; - ctx->ghash = gcm_ghash_4bit_x86; - } + ctx->gmult = gcm_gmult_4bit_mmx; + ctx->ghash = gcm_ghash_4bit_mmx; + } else { + ctx->gmult = gcm_gmult_4bit_x86; + ctx->ghash = gcm_ghash_4bit_x86; + } # else - ctx->gmult = gcm_gmult_4bit; - ctx->ghash = gcm_ghash_4bit; + ctx->gmult = gcm_gmult_4bit; + ctx->ghash = gcm_ghash_4bit; # endif -# elif defined(GHASH_ASM_ARM) - if (OPENSSL_armcap_P & ARMV7_NEON) { - ctx->gmult = gcm_gmult_neon; - ctx->ghash = gcm_ghash_neon; - } else { - gcm_init_4bit(ctx->Htable,ctx->H.u); - ctx->gmult = gcm_gmult_4bit; - ctx->ghash = gcm_ghash_4bit; - } +# elif defined(GHASH_ASM_ARM) + if (OPENSSL_armcap_P & ARMV7_NEON) { + ctx->gmult = gcm_gmult_neon; + ctx->ghash = gcm_ghash_neon; + } else { + gcm_init_4bit(ctx->Htable, ctx->H.u); + ctx->gmult = gcm_gmult_4bit; + ctx->ghash = gcm_ghash_4bit; + } # else - gcm_init_4bit(ctx->Htable,ctx->H.u); + gcm_init_4bit(ctx->Htable, ctx->H.u); # endif #endif } -void CRYPTO_gcm128_setiv(GCM128_CONTEXT *ctx,const unsigned char *iv,size_t len) +void CRYPTO_gcm128_setiv(GCM128_CONTEXT *ctx, const unsigned char *iv, + size_t len) { - const union { long one; char little; } is_endian = {1}; - unsigned int ctr; + const union { + long one; + char little; + } is_endian = { + 1 + }; + unsigned int ctr; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; -#endif - - ctx->Yi.u[0] = 0; - ctx->Yi.u[1] = 0; - ctx->Xi.u[0] = 0; - ctx->Xi.u[1] = 0; - ctx->len.u[0] = 0; /* AAD length */ - ctx->len.u[1] = 0; /* message length */ - ctx->ares = 0; - ctx->mres = 0; - - if (len==12) { - memcpy(ctx->Yi.c,iv,12); - ctx->Yi.c[15]=1; - ctr=1; - } - else { - size_t i; - u64 len0 = len; - - while (len>=16) { - for (i=0; i<16; ++i) ctx->Yi.c[i] ^= iv[i]; - GCM_MUL(ctx,Yi); - iv += 16; - len -= 16; - } - if (len) { - for (i=0; i<len; ++i) ctx->Yi.c[i] ^= iv[i]; - GCM_MUL(ctx,Yi); - } - len0 <<= 3; - if (is_endian.little) { + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; +#endif + + ctx->Yi.u[0] = 0; + ctx->Yi.u[1] = 0; + ctx->Xi.u[0] = 0; + ctx->Xi.u[1] = 0; + ctx->len.u[0] = 0; /* AAD length */ + ctx->len.u[1] = 0; /* message length */ + ctx->ares = 0; + ctx->mres = 0; + + if (len == 12) { + memcpy(ctx->Yi.c, iv, 12); + ctx->Yi.c[15] = 1; + ctr = 1; + } else { + size_t i; + u64 len0 = len; + + while (len >= 16) { + for (i = 0; i < 16; ++i) + ctx->Yi.c[i] ^= iv[i]; + GCM_MUL(ctx, Yi); + iv += 16; + len -= 16; + } + if (len) { + for (i = 0; i < len; ++i) + ctx->Yi.c[i] ^= iv[i]; + GCM_MUL(ctx, Yi); + } + len0 <<= 3; + if (is_endian.little) { #ifdef BSWAP8 - ctx->Yi.u[1] ^= BSWAP8(len0); + ctx->Yi.u[1] ^= BSWAP8(len0); #else - ctx->Yi.c[8] ^= (u8)(len0>>56); - ctx->Yi.c[9] ^= (u8)(len0>>48); - ctx->Yi.c[10] ^= (u8)(len0>>40); - ctx->Yi.c[11] ^= (u8)(len0>>32); - ctx->Yi.c[12] ^= (u8)(len0>>24); - ctx->Yi.c[13] ^= (u8)(len0>>16); - ctx->Yi.c[14] ^= (u8)(len0>>8); - ctx->Yi.c[15] ^= (u8)(len0); + ctx->Yi.c[8] ^= (u8)(len0 >> 56); + ctx->Yi.c[9] ^= (u8)(len0 >> 48); + ctx->Yi.c[10] ^= (u8)(len0 >> 40); + ctx->Yi.c[11] ^= (u8)(len0 >> 32); + ctx->Yi.c[12] ^= (u8)(len0 >> 24); + ctx->Yi.c[13] ^= (u8)(len0 >> 16); + ctx->Yi.c[14] ^= (u8)(len0 >> 8); + ctx->Yi.c[15] ^= (u8)(len0); #endif - } - else - ctx->Yi.u[1] ^= len0; + } else + ctx->Yi.u[1] ^= len0; - GCM_MUL(ctx,Yi); + GCM_MUL(ctx, Yi); - if (is_endian.little) + if (is_endian.little) #ifdef BSWAP4 - ctr = BSWAP4(ctx->Yi.d[3]); + ctr = BSWAP4(ctx->Yi.d[3]); #else - ctr = GETU32(ctx->Yi.c+12); + ctr = GETU32(ctx->Yi.c + 12); #endif - else - ctr = ctx->Yi.d[3]; - } + else + ctr = ctx->Yi.d[3]; + } - (*ctx->block)(ctx->Yi.c,ctx->EK0.c,ctx->key); - ++ctr; - if (is_endian.little) + (*ctx->block) (ctx->Yi.c, ctx->EK0.c, ctx->key); + ++ctr; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); + PUTU32(ctx->Yi.c + 12, ctr); #endif - else - ctx->Yi.d[3] = ctr; + else + ctx->Yi.d[3] = ctr; } -int CRYPTO_gcm128_aad(GCM128_CONTEXT *ctx,const unsigned char *aad,size_t len) +int CRYPTO_gcm128_aad(GCM128_CONTEXT *ctx, const unsigned char *aad, + size_t len) { - size_t i; - unsigned int n; - u64 alen = ctx->len.u[0]; + size_t i; + unsigned int n; + u64 alen = ctx->len.u[0]; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; # ifdef GHASH - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx->ghash; + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx->ghash; # endif #endif - if (ctx->len.u[1]) return -2; - - alen += len; - if (alen>(U64(1)<<61) || (sizeof(len)==8 && alen<len)) - return -1; - ctx->len.u[0] = alen; - - n = ctx->ares; - if (n) { - while (n && len) { - ctx->Xi.c[n] ^= *(aad++); - --len; - n = (n+1)%16; - } - if (n==0) GCM_MUL(ctx,Xi); - else { - ctx->ares = n; - return 0; - } - } - + if (ctx->len.u[1]) + return -2; + + alen += len; + if (alen > (U64(1) << 61) || (sizeof(len) == 8 && alen < len)) + return -1; + ctx->len.u[0] = alen; + + n = ctx->ares; + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(aad++); + --len; + n = (n + 1) % 16; + } + if (n == 0) + GCM_MUL(ctx, Xi); + else { + ctx->ares = n; + return 0; + } + } #ifdef GHASH - if ((i = (len&(size_t)-16))) { - GHASH(ctx,aad,i); - aad += i; - len -= i; - } + if ((i = (len & (size_t)-16))) { + GHASH(ctx, aad, i); + aad += i; + len -= i; + } #else - while (len>=16) { - for (i=0; i<16; ++i) ctx->Xi.c[i] ^= aad[i]; - GCM_MUL(ctx,Xi); - aad += 16; - len -= 16; - } + while (len >= 16) { + for (i = 0; i < 16; ++i) + ctx->Xi.c[i] ^= aad[i]; + GCM_MUL(ctx, Xi); + aad += 16; + len -= 16; + } #endif - if (len) { - n = (unsigned int)len; - for (i=0; i<len; ++i) ctx->Xi.c[i] ^= aad[i]; - } + if (len) { + n = (unsigned int)len; + for (i = 0; i < len; ++i) + ctx->Xi.c[i] ^= aad[i]; + } - ctx->ares = n; - return 0; + ctx->ares = n; + return 0; } int CRYPTO_gcm128_encrypt(GCM128_CONTEXT *ctx, - const unsigned char *in, unsigned char *out, - size_t len) + const unsigned char *in, unsigned char *out, + size_t len) { - const union { long one; char little; } is_endian = {1}; - unsigned int n, ctr; - size_t i; - u64 mlen = ctx->len.u[1]; - block128_f block = ctx->block; - void *key = ctx->key; + const union { + long one; + char little; + } is_endian = { + 1 + }; + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + block128_f block = ctx->block; + void *key = ctx->key; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; # ifdef GHASH - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx->ghash; + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx->ghash; # endif #endif #if 0 - n = (unsigned int)mlen%16; /* alternative to ctx->mres */ + n = (unsigned int)mlen % 16; /* alternative to ctx->mres */ #endif - mlen += len; - if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlen<len)) - return -1; - ctx->len.u[1] = mlen; - - if (ctx->ares) { - /* First call to encrypt finalizes GHASH(AAD) */ - GCM_MUL(ctx,Xi); - ctx->ares = 0; - } - - if (is_endian.little) -#ifdef BSWAP4 - ctr = BSWAP4(ctx->Yi.d[3]); -#else - ctr = GETU32(ctx->Yi.c+12); -#endif - else - ctr = ctx->Yi.d[3]; + mlen += len; + if (mlen > ((U64(1) << 36) - 32) || (sizeof(len) == 8 && mlen < len)) + return -1; + ctx->len.u[1] = mlen; - n = ctx->mres; -#if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - if (n) { - while (n && len) { - ctx->Xi.c[n] ^= *(out++) = *(in++)^ctx->EKi.c[n]; - --len; - n = (n+1)%16; - } - if (n==0) GCM_MUL(ctx,Xi); - else { - ctx->mres = n; - return 0; - } - } -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out)%sizeof(size_t) != 0) - break; -#endif -#if defined(GHASH) && defined(GHASH_CHUNK) - while (len>=GHASH_CHUNK) { - size_t j=GHASH_CHUNK; - - while (j) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; + if (ctx->ares) { + /* First call to encrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx, Xi); + ctx->ares = 0; + } - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) - out_t[i] = in_t[i] ^ ctx->EKi.t[i]; - out += 16; - in += 16; - j -= 16; - } - GHASH(ctx,out-GHASH_CHUNK,GHASH_CHUNK); - len -= GHASH_CHUNK; - } - if ((i = (len&(size_t)-16))) { - size_t j=i; - - while (len>=16) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; - - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctr = BSWAP4(ctx->Yi.d[3]); #else - PUTU32(ctx->Yi.c+12,ctr); + ctr = GETU32(ctx->Yi.c + 12); #endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) - out_t[i] = in_t[i] ^ ctx->EKi.t[i]; - out += 16; - in += 16; - len -= 16; - } - GHASH(ctx,out-j,j); - } -#else - while (len>=16) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; + else + ctr = ctx->Yi.d[3]; - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) - ctx->Xi.t[i] ^= - out_t[i] = in_t[i]^ctx->EKi.t[i]; - GCM_MUL(ctx,Xi); - out += 16; - in += 16; - len -= 16; - } -#endif - if (len) { - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - while (len--) { - ctx->Xi.c[n] ^= out[n] = in[n]^ctx->EKi.c[n]; - ++n; - } - } - - ctx->mres = n; - return 0; - } while(0); + n = ctx->mres; +#if !defined(OPENSSL_SMALL_FOOTPRINT) + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(out++) = *(in++) ^ ctx->EKi.c[n]; + --len; + n = (n + 1) % 16; + } + if (n == 0) + GCM_MUL(ctx, Xi); + else { + ctx->mres = n; + return 0; + } + } +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out) % sizeof(size_t) != 0) + break; +# endif +# if defined(GHASH) && defined(GHASH_CHUNK) + while (len >= GHASH_CHUNK) { + size_t j = GHASH_CHUNK; + + while (j) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + j -= 16; + } + GHASH(ctx, out - GHASH_CHUNK, GHASH_CHUNK); + len -= GHASH_CHUNK; + } + if ((i = (len & (size_t)-16))) { + size_t j = i; + + while (len >= 16) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + len -= 16; + } + GHASH(ctx, out - j, j); + } +# else + while (len >= 16) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) + ctx->Xi.t[i] ^= out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + GCM_MUL(ctx, Xi); + out += 16; + in += 16; + len -= 16; + } +# endif + if (len) { + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + while (len--) { + ctx->Xi.c[n] ^= out[n] = in[n] ^ ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; + } while (0); + } #endif - for (i=0;i<len;++i) { - if (n==0) { - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) + for (i = 0; i < len; ++i) { + if (n == 0) { + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - } - ctx->Xi.c[n] ^= out[i] = in[i]^ctx->EKi.c[n]; - n = (n+1)%16; - if (n==0) - GCM_MUL(ctx,Xi); - } - - ctx->mres = n; - return 0; + PUTU32(ctx->Yi.c + 12, ctr); +#endif + else + ctx->Yi.d[3] = ctr; + } + ctx->Xi.c[n] ^= out[i] = in[i] ^ ctx->EKi.c[n]; + n = (n + 1) % 16; + if (n == 0) + GCM_MUL(ctx, Xi); + } + + ctx->mres = n; + return 0; } int CRYPTO_gcm128_decrypt(GCM128_CONTEXT *ctx, - const unsigned char *in, unsigned char *out, - size_t len) + const unsigned char *in, unsigned char *out, + size_t len) { - const union { long one; char little; } is_endian = {1}; - unsigned int n, ctr; - size_t i; - u64 mlen = ctx->len.u[1]; - block128_f block = ctx->block; - void *key = ctx->key; + const union { + long one; + char little; + } is_endian = { + 1 + }; + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + block128_f block = ctx->block; + void *key = ctx->key; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; # ifdef GHASH - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx->ghash; + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx->ghash; # endif #endif - mlen += len; - if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlen<len)) - return -1; - ctx->len.u[1] = mlen; + mlen += len; + if (mlen > ((U64(1) << 36) - 32) || (sizeof(len) == 8 && mlen < len)) + return -1; + ctx->len.u[1] = mlen; - if (ctx->ares) { - /* First call to decrypt finalizes GHASH(AAD) */ - GCM_MUL(ctx,Xi); - ctx->ares = 0; - } + if (ctx->ares) { + /* First call to decrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx, Xi); + ctx->ares = 0; + } - if (is_endian.little) + if (is_endian.little) #ifdef BSWAP4 - ctr = BSWAP4(ctx->Yi.d[3]); + ctr = BSWAP4(ctx->Yi.d[3]); #else - ctr = GETU32(ctx->Yi.c+12); + ctr = GETU32(ctx->Yi.c + 12); #endif - else - ctr = ctx->Yi.d[3]; + else + ctr = ctx->Yi.d[3]; - n = ctx->mres; + n = ctx->mres; #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - if (n) { - while (n && len) { - u8 c = *(in++); - *(out++) = c^ctx->EKi.c[n]; - ctx->Xi.c[n] ^= c; - --len; - n = (n+1)%16; - } - if (n==0) GCM_MUL (ctx,Xi); - else { - ctx->mres = n; - return 0; - } - } -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out)%sizeof(size_t) != 0) - break; -#endif -#if defined(GHASH) && defined(GHASH_CHUNK) - while (len>=GHASH_CHUNK) { - size_t j=GHASH_CHUNK; - - GHASH(ctx,in,GHASH_CHUNK); - while (j) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; - - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + if (n) { + while (n && len) { + u8 c = *(in++); + *(out++) = c ^ ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + --len; + n = (n + 1) % 16; + } + if (n == 0) + GCM_MUL(ctx, Xi); + else { + ctx->mres = n; + return 0; + } + } +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out) % sizeof(size_t) != 0) + break; +# endif +# if defined(GHASH) && defined(GHASH_CHUNK) + while (len >= GHASH_CHUNK) { + size_t j = GHASH_CHUNK; + + GHASH(ctx, in, GHASH_CHUNK); + while (j) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + j -= 16; + } + len -= GHASH_CHUNK; + } + if ((i = (len & (size_t)-16))) { + GHASH(ctx, in, i); + while (len >= 16) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) + out_t[i] = in_t[i] ^ ctx->EKi.t[i]; + out += 16; + in += 16; + len -= 16; + } + } +# else + while (len >= 16) { + size_t *out_t = (size_t *)out; + const size_t *in_t = (const size_t *)in; + + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + for (i = 0; i < 16 / sizeof(size_t); ++i) { + size_t c = in[i]; + out[i] = c ^ ctx->EKi.t[i]; + ctx->Xi.t[i] ^= c; + } + GCM_MUL(ctx, Xi); + out += 16; + in += 16; + len -= 16; + } +# endif + if (len) { + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + while (len--) { + u8 c = in[n]; + ctx->Xi.c[n] ^= c; + out[n] = c ^ ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; + } while (0); + } #endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) - out_t[i] = in_t[i]^ctx->EKi.t[i]; - out += 16; - in += 16; - j -= 16; - } - len -= GHASH_CHUNK; - } - if ((i = (len&(size_t)-16))) { - GHASH(ctx,in,i); - while (len>=16) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; - - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) + for (i = 0; i < len; ++i) { + u8 c; + if (n == 0) { + (*block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) - out_t[i] = in_t[i]^ctx->EKi.t[i]; - out += 16; - in += 16; - len -= 16; - } - } + ctx->Yi.d[3] = BSWAP4(ctr); #else - while (len>=16) { - size_t *out_t=(size_t *)out; - const size_t *in_t=(const size_t *)in; + PUTU32(ctx->Yi.c + 12, ctr); +#endif + else + ctx->Yi.d[3] = ctr; + } + c = in[i]; + out[i] = c ^ ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + n = (n + 1) % 16; + if (n == 0) + GCM_MUL(ctx, Xi); + } - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - for (i=0; i<16/sizeof(size_t); ++i) { - size_t c = in[i]; - out[i] = c^ctx->EKi.t[i]; - ctx->Xi.t[i] ^= c; - } - GCM_MUL(ctx,Xi); - out += 16; - in += 16; - len -= 16; - } -#endif - if (len) { - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - while (len--) { - u8 c = in[n]; - ctx->Xi.c[n] ^= c; - out[n] = c^ctx->EKi.c[n]; - ++n; - } - } - - ctx->mres = n; - return 0; - } while(0); -#endif - for (i=0;i<len;++i) { - u8 c; - if (n==0) { - (*block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - } - c = in[i]; - out[i] = c^ctx->EKi.c[n]; - ctx->Xi.c[n] ^= c; - n = (n+1)%16; - if (n==0) - GCM_MUL(ctx,Xi); - } - - ctx->mres = n; - return 0; + ctx->mres = n; + return 0; } int CRYPTO_gcm128_encrypt_ctr32(GCM128_CONTEXT *ctx, - const unsigned char *in, unsigned char *out, - size_t len, ctr128_f stream) + const unsigned char *in, unsigned char *out, + size_t len, ctr128_f stream) { - const union { long one; char little; } is_endian = {1}; - unsigned int n, ctr; - size_t i; - u64 mlen = ctx->len.u[1]; - void *key = ctx->key; + const union { + long one; + char little; + } is_endian = { + 1 + }; + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + void *key = ctx->key; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; # ifdef GHASH - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx->ghash; + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx->ghash; # endif #endif - mlen += len; - if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlen<len)) - return -1; - ctx->len.u[1] = mlen; + mlen += len; + if (mlen > ((U64(1) << 36) - 32) || (sizeof(len) == 8 && mlen < len)) + return -1; + ctx->len.u[1] = mlen; - if (ctx->ares) { - /* First call to encrypt finalizes GHASH(AAD) */ - GCM_MUL(ctx,Xi); - ctx->ares = 0; - } + if (ctx->ares) { + /* First call to encrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx, Xi); + ctx->ares = 0; + } - if (is_endian.little) + if (is_endian.little) #ifdef BSWAP4 - ctr = BSWAP4(ctx->Yi.d[3]); + ctr = BSWAP4(ctx->Yi.d[3]); #else - ctr = GETU32(ctx->Yi.c+12); -#endif - else - ctr = ctx->Yi.d[3]; - - n = ctx->mres; - if (n) { - while (n && len) { - ctx->Xi.c[n] ^= *(out++) = *(in++)^ctx->EKi.c[n]; - --len; - n = (n+1)%16; - } - if (n==0) GCM_MUL(ctx,Xi); - else { - ctx->mres = n; - return 0; - } - } + ctr = GETU32(ctx->Yi.c + 12); +#endif + else + ctr = ctx->Yi.d[3]; + + n = ctx->mres; + if (n) { + while (n && len) { + ctx->Xi.c[n] ^= *(out++) = *(in++) ^ ctx->EKi.c[n]; + --len; + n = (n + 1) % 16; + } + if (n == 0) + GCM_MUL(ctx, Xi); + else { + ctx->mres = n; + return 0; + } + } #if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT) - while (len>=GHASH_CHUNK) { - (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c); - ctr += GHASH_CHUNK/16; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - GHASH(ctx,out,GHASH_CHUNK); - out += GHASH_CHUNK; - in += GHASH_CHUNK; - len -= GHASH_CHUNK; - } + while (len >= GHASH_CHUNK) { + (*stream) (in, out, GHASH_CHUNK / 16, key, ctx->Yi.c); + ctr += GHASH_CHUNK / 16; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + GHASH(ctx, out, GHASH_CHUNK); + out += GHASH_CHUNK; + in += GHASH_CHUNK; + len -= GHASH_CHUNK; + } #endif - if ((i = (len&(size_t)-16))) { - size_t j=i/16; + if ((i = (len & (size_t)-16))) { + size_t j = i / 16; - (*stream)(in,out,j,key,ctx->Yi.c); - ctr += (unsigned int)j; - if (is_endian.little) + (*stream) (in, out, j, key, ctx->Yi.c); + ctr += (unsigned int)j; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); + PUTU32(ctx->Yi.c + 12, ctr); #endif - else - ctx->Yi.d[3] = ctr; - in += i; - len -= i; + else + ctx->Yi.d[3] = ctr; + in += i; + len -= i; #if defined(GHASH) - GHASH(ctx,out,i); - out += i; + GHASH(ctx, out, i); + out += i; #else - while (j--) { - for (i=0;i<16;++i) ctx->Xi.c[i] ^= out[i]; - GCM_MUL(ctx,Xi); - out += 16; - } + while (j--) { + for (i = 0; i < 16; ++i) + ctx->Xi.c[i] ^= out[i]; + GCM_MUL(ctx, Xi); + out += 16; + } #endif - } - if (len) { - (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) + } + if (len) { + (*ctx->block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - while (len--) { - ctx->Xi.c[n] ^= out[n] = in[n]^ctx->EKi.c[n]; - ++n; - } - } - - ctx->mres = n; - return 0; + PUTU32(ctx->Yi.c + 12, ctr); +#endif + else + ctx->Yi.d[3] = ctr; + while (len--) { + ctx->Xi.c[n] ^= out[n] = in[n] ^ ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; } int CRYPTO_gcm128_decrypt_ctr32(GCM128_CONTEXT *ctx, - const unsigned char *in, unsigned char *out, - size_t len,ctr128_f stream) + const unsigned char *in, unsigned char *out, + size_t len, ctr128_f stream) { - const union { long one; char little; } is_endian = {1}; - unsigned int n, ctr; - size_t i; - u64 mlen = ctx->len.u[1]; - void *key = ctx->key; + const union { + long one; + char little; + } is_endian = { + 1 + }; + unsigned int n, ctr; + size_t i; + u64 mlen = ctx->len.u[1]; + void *key = ctx->key; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; # ifdef GHASH - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx->ghash; + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx->ghash; # endif #endif - mlen += len; - if (mlen>((U64(1)<<36)-32) || (sizeof(len)==8 && mlen<len)) - return -1; - ctx->len.u[1] = mlen; + mlen += len; + if (mlen > ((U64(1) << 36) - 32) || (sizeof(len) == 8 && mlen < len)) + return -1; + ctx->len.u[1] = mlen; - if (ctx->ares) { - /* First call to decrypt finalizes GHASH(AAD) */ - GCM_MUL(ctx,Xi); - ctx->ares = 0; - } + if (ctx->ares) { + /* First call to decrypt finalizes GHASH(AAD) */ + GCM_MUL(ctx, Xi); + ctx->ares = 0; + } - if (is_endian.little) + if (is_endian.little) #ifdef BSWAP4 - ctr = BSWAP4(ctx->Yi.d[3]); + ctr = BSWAP4(ctx->Yi.d[3]); #else - ctr = GETU32(ctx->Yi.c+12); -#endif - else - ctr = ctx->Yi.d[3]; - - n = ctx->mres; - if (n) { - while (n && len) { - u8 c = *(in++); - *(out++) = c^ctx->EKi.c[n]; - ctx->Xi.c[n] ^= c; - --len; - n = (n+1)%16; - } - if (n==0) GCM_MUL (ctx,Xi); - else { - ctx->mres = n; - return 0; - } - } + ctr = GETU32(ctx->Yi.c + 12); +#endif + else + ctr = ctx->Yi.d[3]; + + n = ctx->mres; + if (n) { + while (n && len) { + u8 c = *(in++); + *(out++) = c ^ ctx->EKi.c[n]; + ctx->Xi.c[n] ^= c; + --len; + n = (n + 1) % 16; + } + if (n == 0) + GCM_MUL(ctx, Xi); + else { + ctx->mres = n; + return 0; + } + } #if defined(GHASH) && !defined(OPENSSL_SMALL_FOOTPRINT) - while (len>=GHASH_CHUNK) { - GHASH(ctx,in,GHASH_CHUNK); - (*stream)(in,out,GHASH_CHUNK/16,key,ctx->Yi.c); - ctr += GHASH_CHUNK/16; - if (is_endian.little) -#ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); -#else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - out += GHASH_CHUNK; - in += GHASH_CHUNK; - len -= GHASH_CHUNK; - } + while (len >= GHASH_CHUNK) { + GHASH(ctx, in, GHASH_CHUNK); + (*stream) (in, out, GHASH_CHUNK / 16, key, ctx->Yi.c); + ctr += GHASH_CHUNK / 16; + if (is_endian.little) +# ifdef BSWAP4 + ctx->Yi.d[3] = BSWAP4(ctr); +# else + PUTU32(ctx->Yi.c + 12, ctr); +# endif + else + ctx->Yi.d[3] = ctr; + out += GHASH_CHUNK; + in += GHASH_CHUNK; + len -= GHASH_CHUNK; + } #endif - if ((i = (len&(size_t)-16))) { - size_t j=i/16; + if ((i = (len & (size_t)-16))) { + size_t j = i / 16; #if defined(GHASH) - GHASH(ctx,in,i); + GHASH(ctx, in, i); #else - while (j--) { - size_t k; - for (k=0;k<16;++k) ctx->Xi.c[k] ^= in[k]; - GCM_MUL(ctx,Xi); - in += 16; - } - j = i/16; - in -= i; -#endif - (*stream)(in,out,j,key,ctx->Yi.c); - ctr += (unsigned int)j; - if (is_endian.little) + while (j--) { + size_t k; + for (k = 0; k < 16; ++k) + ctx->Xi.c[k] ^= in[k]; + GCM_MUL(ctx, Xi); + in += 16; + } + j = i / 16; + in -= i; +#endif + (*stream) (in, out, j, key, ctx->Yi.c); + ctr += (unsigned int)j; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); + PUTU32(ctx->Yi.c + 12, ctr); #endif - else - ctx->Yi.d[3] = ctr; - out += i; - in += i; - len -= i; - } - if (len) { - (*ctx->block)(ctx->Yi.c,ctx->EKi.c,key); - ++ctr; - if (is_endian.little) + else + ctx->Yi.d[3] = ctr; + out += i; + in += i; + len -= i; + } + if (len) { + (*ctx->block) (ctx->Yi.c, ctx->EKi.c, key); + ++ctr; + if (is_endian.little) #ifdef BSWAP4 - ctx->Yi.d[3] = BSWAP4(ctr); + ctx->Yi.d[3] = BSWAP4(ctr); #else - PUTU32(ctx->Yi.c+12,ctr); -#endif - else - ctx->Yi.d[3] = ctr; - while (len--) { - u8 c = in[n]; - ctx->Xi.c[n] ^= c; - out[n] = c^ctx->EKi.c[n]; - ++n; - } - } - - ctx->mres = n; - return 0; + PUTU32(ctx->Yi.c + 12, ctr); +#endif + else + ctx->Yi.d[3] = ctr; + while (len--) { + u8 c = in[n]; + ctx->Xi.c[n] ^= c; + out[n] = c ^ ctx->EKi.c[n]; + ++n; + } + } + + ctx->mres = n; + return 0; } -int CRYPTO_gcm128_finish(GCM128_CONTEXT *ctx,const unsigned char *tag, - size_t len) +int CRYPTO_gcm128_finish(GCM128_CONTEXT *ctx, const unsigned char *tag, + size_t len) { - const union { long one; char little; } is_endian = {1}; - u64 alen = ctx->len.u[0]<<3; - u64 clen = ctx->len.u[1]<<3; + const union { + long one; + char little; + } is_endian = { + 1 + }; + u64 alen = ctx->len.u[0] << 3; + u64 clen = ctx->len.u[1] << 3; #ifdef GCM_FUNCREF_4BIT - void (*gcm_gmult_p)(u64 Xi[2],const u128 Htable[16]) = ctx->gmult; + void (*gcm_gmult_p) (u64 Xi[2], const u128 Htable[16]) = ctx->gmult; #endif - if (ctx->mres || ctx->ares) - GCM_MUL(ctx,Xi); + if (ctx->mres || ctx->ares) + GCM_MUL(ctx, Xi); - if (is_endian.little) { + if (is_endian.little) { #ifdef BSWAP8 - alen = BSWAP8(alen); - clen = BSWAP8(clen); + alen = BSWAP8(alen); + clen = BSWAP8(clen); #else - u8 *p = ctx->len.c; + u8 *p = ctx->len.c; - ctx->len.u[0] = alen; - ctx->len.u[1] = clen; + ctx->len.u[0] = alen; + ctx->len.u[1] = clen; - alen = (u64)GETU32(p) <<32|GETU32(p+4); - clen = (u64)GETU32(p+8)<<32|GETU32(p+12); + alen = (u64)GETU32(p) << 32 | GETU32(p + 4); + clen = (u64)GETU32(p + 8) << 32 | GETU32(p + 12); #endif - } + } - ctx->Xi.u[0] ^= alen; - ctx->Xi.u[1] ^= clen; - GCM_MUL(ctx,Xi); + ctx->Xi.u[0] ^= alen; + ctx->Xi.u[1] ^= clen; + GCM_MUL(ctx, Xi); - ctx->Xi.u[0] ^= ctx->EK0.u[0]; - ctx->Xi.u[1] ^= ctx->EK0.u[1]; + ctx->Xi.u[0] ^= ctx->EK0.u[0]; + ctx->Xi.u[1] ^= ctx->EK0.u[1]; - if (tag && len<=sizeof(ctx->Xi)) - return memcmp(ctx->Xi.c,tag,len); - else - return -1; + if (tag && len <= sizeof(ctx->Xi)) + return CRYPTO_memcmp(ctx->Xi.c, tag, len); + else + return -1; } void CRYPTO_gcm128_tag(GCM128_CONTEXT *ctx, unsigned char *tag, size_t len) { - CRYPTO_gcm128_finish(ctx, NULL, 0); - memcpy(tag, ctx->Xi.c, len<=sizeof(ctx->Xi.c)?len:sizeof(ctx->Xi.c)); + CRYPTO_gcm128_finish(ctx, NULL, 0); + memcpy(tag, ctx->Xi.c, + len <= sizeof(ctx->Xi.c) ? len : sizeof(ctx->Xi.c)); } GCM128_CONTEXT *CRYPTO_gcm128_new(void *key, block128_f block) { - GCM128_CONTEXT *ret; + GCM128_CONTEXT *ret; - if ((ret = (GCM128_CONTEXT *)OPENSSL_malloc(sizeof(GCM128_CONTEXT)))) - CRYPTO_gcm128_init(ret,key,block); + if ((ret = (GCM128_CONTEXT *)OPENSSL_malloc(sizeof(GCM128_CONTEXT)))) + CRYPTO_gcm128_init(ret, key, block); - return ret; + return ret; } void CRYPTO_gcm128_release(GCM128_CONTEXT *ctx) { - if (ctx) { - OPENSSL_cleanse(ctx,sizeof(*ctx)); - OPENSSL_free(ctx); - } + if (ctx) { + OPENSSL_cleanse(ctx, sizeof(*ctx)); + OPENSSL_free(ctx); + } } #if defined(SELFTEST) -#include <stdio.h> -#include <openssl/aes.h> +# include <stdio.h> +# include <openssl/aes.h> /* Test Case 1 */ -static const u8 K1[16], - *P1=NULL, - *A1=NULL, - IV1[12], - *C1=NULL, - T1[]= {0x58,0xe2,0xfc,0xce,0xfa,0x7e,0x30,0x61,0x36,0x7f,0x1d,0x57,0xa4,0xe7,0x45,0x5a}; +static const u8 K1[16], *P1 = NULL, *A1 = NULL, IV1[12], *C1 = NULL; +static const u8 T1[] = { + 0x58, 0xe2, 0xfc, 0xce, 0xfa, 0x7e, 0x30, 0x61, + 0x36, 0x7f, 0x1d, 0x57, 0xa4, 0xe7, 0x45, 0x5a +}; /* Test Case 2 */ -#define K2 K1 -#define A2 A1 -#define IV2 IV1 -static const u8 P2[16], - C2[]= {0x03,0x88,0xda,0xce,0x60,0xb6,0xa3,0x92,0xf3,0x28,0xc2,0xb9,0x71,0xb2,0xfe,0x78}, - T2[]= {0xab,0x6e,0x47,0xd4,0x2c,0xec,0x13,0xbd,0xf5,0x3a,0x67,0xb2,0x12,0x57,0xbd,0xdf}; +# define K2 K1 +# define A2 A1 +# define IV2 IV1 +static const u8 P2[16]; +static const u8 C2[] = { + 0x03, 0x88, 0xda, 0xce, 0x60, 0xb6, 0xa3, 0x92, + 0xf3, 0x28, 0xc2, 0xb9, 0x71, 0xb2, 0xfe, 0x78 +}; + +static const u8 T2[] = { + 0xab, 0x6e, 0x47, 0xd4, 0x2c, 0xec, 0x13, 0xbd, + 0xf5, 0x3a, 0x67, 0xb2, 0x12, 0x57, 0xbd, 0xdf +}; /* Test Case 3 */ -#define A3 A2 -static const u8 K3[]= {0xfe,0xff,0xe9,0x92,0x86,0x65,0x73,0x1c,0x6d,0x6a,0x8f,0x94,0x67,0x30,0x83,0x08}, - P3[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39,0x1a,0xaf,0xd2,0x55}, - IV3[]= {0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad,0xde,0xca,0xf8,0x88}, - C3[]= {0x42,0x83,0x1e,0xc2,0x21,0x77,0x74,0x24,0x4b,0x72,0x21,0xb7,0x84,0xd0,0xd4,0x9c, - 0xe3,0xaa,0x21,0x2f,0x2c,0x02,0xa4,0xe0,0x35,0xc1,0x7e,0x23,0x29,0xac,0xa1,0x2e, - 0x21,0xd5,0x14,0xb2,0x54,0x66,0x93,0x1c,0x7d,0x8f,0x6a,0x5a,0xac,0x84,0xaa,0x05, - 0x1b,0xa3,0x0b,0x39,0x6a,0x0a,0xac,0x97,0x3d,0x58,0xe0,0x91,0x47,0x3f,0x59,0x85}, - T3[]= {0x4d,0x5c,0x2a,0xf3,0x27,0xcd,0x64,0xa6,0x2c,0xf3,0x5a,0xbd,0x2b,0xa6,0xfa,0xb4}; +# define A3 A2 +static const u8 K3[] = { + 0xfe, 0xff, 0xe9, 0x92, 0x86, 0x65, 0x73, 0x1c, + 0x6d, 0x6a, 0x8f, 0x94, 0x67, 0x30, 0x83, 0x08 +}; + +static const u8 P3[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39, 0x1a, 0xaf, 0xd2, 0x55 +}; + +static const u8 IV3[] = { + 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad, + 0xde, 0xca, 0xf8, 0x88 +}; + +static const u8 C3[] = { + 0x42, 0x83, 0x1e, 0xc2, 0x21, 0x77, 0x74, 0x24, + 0x4b, 0x72, 0x21, 0xb7, 0x84, 0xd0, 0xd4, 0x9c, + 0xe3, 0xaa, 0x21, 0x2f, 0x2c, 0x02, 0xa4, 0xe0, + 0x35, 0xc1, 0x7e, 0x23, 0x29, 0xac, 0xa1, 0x2e, + 0x21, 0xd5, 0x14, 0xb2, 0x54, 0x66, 0x93, 0x1c, + 0x7d, 0x8f, 0x6a, 0x5a, 0xac, 0x84, 0xaa, 0x05, + 0x1b, 0xa3, 0x0b, 0x39, 0x6a, 0x0a, 0xac, 0x97, + 0x3d, 0x58, 0xe0, 0x91, 0x47, 0x3f, 0x59, 0x85 +}; + +static const u8 T3[] = { + 0x4d, 0x5c, 0x2a, 0xf3, 0x27, 0xcd, 0x64, 0xa6, + 0x2c, 0xf3, 0x5a, 0xbd, 0x2b, 0xa6, 0xfa, 0xb4 +}; /* Test Case 4 */ -#define K4 K3 -#define IV4 IV3 -static const u8 P4[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39}, - A4[]= {0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef,0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef, - 0xab,0xad,0xda,0xd2}, - C4[]= {0x42,0x83,0x1e,0xc2,0x21,0x77,0x74,0x24,0x4b,0x72,0x21,0xb7,0x84,0xd0,0xd4,0x9c, - 0xe3,0xaa,0x21,0x2f,0x2c,0x02,0xa4,0xe0,0x35,0xc1,0x7e,0x23,0x29,0xac,0xa1,0x2e, - 0x21,0xd5,0x14,0xb2,0x54,0x66,0x93,0x1c,0x7d,0x8f,0x6a,0x5a,0xac,0x84,0xaa,0x05, - 0x1b,0xa3,0x0b,0x39,0x6a,0x0a,0xac,0x97,0x3d,0x58,0xe0,0x91}, - T4[]= {0x5b,0xc9,0x4f,0xbc,0x32,0x21,0xa5,0xdb,0x94,0xfa,0xe9,0x5a,0xe7,0x12,0x1a,0x47}; +# define K4 K3 +# define IV4 IV3 +static const u8 P4[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39 +}; + +static const u8 A4[] = { + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xab, 0xad, 0xda, 0xd2 +}; + +static const u8 C4[] = { + 0x42, 0x83, 0x1e, 0xc2, 0x21, 0x77, 0x74, 0x24, + 0x4b, 0x72, 0x21, 0xb7, 0x84, 0xd0, 0xd4, 0x9c, + 0xe3, 0xaa, 0x21, 0x2f, 0x2c, 0x02, 0xa4, 0xe0, + 0x35, 0xc1, 0x7e, 0x23, 0x29, 0xac, 0xa1, 0x2e, + 0x21, 0xd5, 0x14, 0xb2, 0x54, 0x66, 0x93, 0x1c, + 0x7d, 0x8f, 0x6a, 0x5a, 0xac, 0x84, 0xaa, 0x05, + 0x1b, 0xa3, 0x0b, 0x39, 0x6a, 0x0a, 0xac, 0x97, + 0x3d, 0x58, 0xe0, 0x91 +}; + +static const u8 T4[] = { + 0x5b, 0xc9, 0x4f, 0xbc, 0x32, 0x21, 0xa5, 0xdb, + 0x94, 0xfa, 0xe9, 0x5a, 0xe7, 0x12, 0x1a, 0x47 +}; /* Test Case 5 */ -#define K5 K4 -#define P5 P4 -#define A5 A4 -static const u8 IV5[]= {0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad}, - C5[]= {0x61,0x35,0x3b,0x4c,0x28,0x06,0x93,0x4a,0x77,0x7f,0xf5,0x1f,0xa2,0x2a,0x47,0x55, - 0x69,0x9b,0x2a,0x71,0x4f,0xcd,0xc6,0xf8,0x37,0x66,0xe5,0xf9,0x7b,0x6c,0x74,0x23, - 0x73,0x80,0x69,0x00,0xe4,0x9f,0x24,0xb2,0x2b,0x09,0x75,0x44,0xd4,0x89,0x6b,0x42, - 0x49,0x89,0xb5,0xe1,0xeb,0xac,0x0f,0x07,0xc2,0x3f,0x45,0x98}, - T5[]= {0x36,0x12,0xd2,0xe7,0x9e,0x3b,0x07,0x85,0x56,0x1b,0xe1,0x4a,0xac,0xa2,0xfc,0xcb}; +# define K5 K4 +# define P5 P4 +# define A5 A4 +static const u8 IV5[] = { + 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad +}; + +static const u8 C5[] = { + 0x61, 0x35, 0x3b, 0x4c, 0x28, 0x06, 0x93, 0x4a, + 0x77, 0x7f, 0xf5, 0x1f, 0xa2, 0x2a, 0x47, 0x55, + 0x69, 0x9b, 0x2a, 0x71, 0x4f, 0xcd, 0xc6, 0xf8, + 0x37, 0x66, 0xe5, 0xf9, 0x7b, 0x6c, 0x74, 0x23, + 0x73, 0x80, 0x69, 0x00, 0xe4, 0x9f, 0x24, 0xb2, + 0x2b, 0x09, 0x75, 0x44, 0xd4, 0x89, 0x6b, 0x42, + 0x49, 0x89, 0xb5, 0xe1, 0xeb, 0xac, 0x0f, 0x07, + 0xc2, 0x3f, 0x45, 0x98 +}; + +static const u8 T5[] = { + 0x36, 0x12, 0xd2, 0xe7, 0x9e, 0x3b, 0x07, 0x85, + 0x56, 0x1b, 0xe1, 0x4a, 0xac, 0xa2, 0xfc, 0xcb +}; /* Test Case 6 */ -#define K6 K5 -#define P6 P5 -#define A6 A5 -static const u8 IV6[]= {0x93,0x13,0x22,0x5d,0xf8,0x84,0x06,0xe5,0x55,0x90,0x9c,0x5a,0xff,0x52,0x69,0xaa, - 0x6a,0x7a,0x95,0x38,0x53,0x4f,0x7d,0xa1,0xe4,0xc3,0x03,0xd2,0xa3,0x18,0xa7,0x28, - 0xc3,0xc0,0xc9,0x51,0x56,0x80,0x95,0x39,0xfc,0xf0,0xe2,0x42,0x9a,0x6b,0x52,0x54, - 0x16,0xae,0xdb,0xf5,0xa0,0xde,0x6a,0x57,0xa6,0x37,0xb3,0x9b}, - C6[]= {0x8c,0xe2,0x49,0x98,0x62,0x56,0x15,0xb6,0x03,0xa0,0x33,0xac,0xa1,0x3f,0xb8,0x94, - 0xbe,0x91,0x12,0xa5,0xc3,0xa2,0x11,0xa8,0xba,0x26,0x2a,0x3c,0xca,0x7e,0x2c,0xa7, - 0x01,0xe4,0xa9,0xa4,0xfb,0xa4,0x3c,0x90,0xcc,0xdc,0xb2,0x81,0xd4,0x8c,0x7c,0x6f, - 0xd6,0x28,0x75,0xd2,0xac,0xa4,0x17,0x03,0x4c,0x34,0xae,0xe5}, - T6[]= {0x61,0x9c,0xc5,0xae,0xff,0xfe,0x0b,0xfa,0x46,0x2a,0xf4,0x3c,0x16,0x99,0xd0,0x50}; +# define K6 K5 +# define P6 P5 +# define A6 A5 +static const u8 IV6[] = { + 0x93, 0x13, 0x22, 0x5d, 0xf8, 0x84, 0x06, 0xe5, + 0x55, 0x90, 0x9c, 0x5a, 0xff, 0x52, 0x69, 0xaa, + 0x6a, 0x7a, 0x95, 0x38, 0x53, 0x4f, 0x7d, 0xa1, + 0xe4, 0xc3, 0x03, 0xd2, 0xa3, 0x18, 0xa7, 0x28, + 0xc3, 0xc0, 0xc9, 0x51, 0x56, 0x80, 0x95, 0x39, + 0xfc, 0xf0, 0xe2, 0x42, 0x9a, 0x6b, 0x52, 0x54, + 0x16, 0xae, 0xdb, 0xf5, 0xa0, 0xde, 0x6a, 0x57, + 0xa6, 0x37, 0xb3, 0x9b +}; + +static const u8 C6[] = { + 0x8c, 0xe2, 0x49, 0x98, 0x62, 0x56, 0x15, 0xb6, + 0x03, 0xa0, 0x33, 0xac, 0xa1, 0x3f, 0xb8, 0x94, + 0xbe, 0x91, 0x12, 0xa5, 0xc3, 0xa2, 0x11, 0xa8, + 0xba, 0x26, 0x2a, 0x3c, 0xca, 0x7e, 0x2c, 0xa7, + 0x01, 0xe4, 0xa9, 0xa4, 0xfb, 0xa4, 0x3c, 0x90, + 0xcc, 0xdc, 0xb2, 0x81, 0xd4, 0x8c, 0x7c, 0x6f, + 0xd6, 0x28, 0x75, 0xd2, 0xac, 0xa4, 0x17, 0x03, + 0x4c, 0x34, 0xae, 0xe5 +}; + +static const u8 T6[] = { + 0x61, 0x9c, 0xc5, 0xae, 0xff, 0xfe, 0x0b, 0xfa, + 0x46, 0x2a, 0xf4, 0x3c, 0x16, 0x99, 0xd0, 0x50 +}; /* Test Case 7 */ -static const u8 K7[24], - *P7=NULL, - *A7=NULL, - IV7[12], - *C7=NULL, - T7[]= {0xcd,0x33,0xb2,0x8a,0xc7,0x73,0xf7,0x4b,0xa0,0x0e,0xd1,0xf3,0x12,0x57,0x24,0x35}; +static const u8 K7[24], *P7 = NULL, *A7 = NULL, IV7[12], *C7 = NULL; +static const u8 T7[] = { + 0xcd, 0x33, 0xb2, 0x8a, 0xc7, 0x73, 0xf7, 0x4b, + 0xa0, 0x0e, 0xd1, 0xf3, 0x12, 0x57, 0x24, 0x35 +}; /* Test Case 8 */ -#define K8 K7 -#define IV8 IV7 -#define A8 A7 -static const u8 P8[16], - C8[]= {0x98,0xe7,0x24,0x7c,0x07,0xf0,0xfe,0x41,0x1c,0x26,0x7e,0x43,0x84,0xb0,0xf6,0x00}, - T8[]= {0x2f,0xf5,0x8d,0x80,0x03,0x39,0x27,0xab,0x8e,0xf4,0xd4,0x58,0x75,0x14,0xf0,0xfb}; +# define K8 K7 +# define IV8 IV7 +# define A8 A7 +static const u8 P8[16]; +static const u8 C8[] = { + 0x98, 0xe7, 0x24, 0x7c, 0x07, 0xf0, 0xfe, 0x41, + 0x1c, 0x26, 0x7e, 0x43, 0x84, 0xb0, 0xf6, 0x00 +}; + +static const u8 T8[] = { + 0x2f, 0xf5, 0x8d, 0x80, 0x03, 0x39, 0x27, 0xab, + 0x8e, 0xf4, 0xd4, 0x58, 0x75, 0x14, 0xf0, 0xfb +}; /* Test Case 9 */ -#define A9 A8 -static const u8 K9[]= {0xfe,0xff,0xe9,0x92,0x86,0x65,0x73,0x1c,0x6d,0x6a,0x8f,0x94,0x67,0x30,0x83,0x08, - 0xfe,0xff,0xe9,0x92,0x86,0x65,0x73,0x1c}, - P9[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39,0x1a,0xaf,0xd2,0x55}, - IV9[]= {0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad,0xde,0xca,0xf8,0x88}, - C9[]= {0x39,0x80,0xca,0x0b,0x3c,0x00,0xe8,0x41,0xeb,0x06,0xfa,0xc4,0x87,0x2a,0x27,0x57, - 0x85,0x9e,0x1c,0xea,0xa6,0xef,0xd9,0x84,0x62,0x85,0x93,0xb4,0x0c,0xa1,0xe1,0x9c, - 0x7d,0x77,0x3d,0x00,0xc1,0x44,0xc5,0x25,0xac,0x61,0x9d,0x18,0xc8,0x4a,0x3f,0x47, - 0x18,0xe2,0x44,0x8b,0x2f,0xe3,0x24,0xd9,0xcc,0xda,0x27,0x10,0xac,0xad,0xe2,0x56}, - T9[]= {0x99,0x24,0xa7,0xc8,0x58,0x73,0x36,0xbf,0xb1,0x18,0x02,0x4d,0xb8,0x67,0x4a,0x14}; +# define A9 A8 +static const u8 K9[] = { + 0xfe, 0xff, 0xe9, 0x92, 0x86, 0x65, 0x73, 0x1c, + 0x6d, 0x6a, 0x8f, 0x94, 0x67, 0x30, 0x83, 0x08, + 0xfe, 0xff, 0xe9, 0x92, 0x86, 0x65, 0x73, 0x1c +}; + +static const u8 P9[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39, 0x1a, 0xaf, 0xd2, 0x55 +}; + +static const u8 IV9[] = { + 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad, + 0xde, 0xca, 0xf8, 0x88 +}; + +static const u8 C9[] = { + 0x39, 0x80, 0xca, 0x0b, 0x3c, 0x00, 0xe8, 0x41, + 0xeb, 0x06, 0xfa, 0xc4, 0x87, 0x2a, 0x27, 0x57, + 0x85, 0x9e, 0x1c, 0xea, 0xa6, 0xef, 0xd9, 0x84, + 0x62, 0x85, 0x93, 0xb4, 0x0c, 0xa1, 0xe1, 0x9c, + 0x7d, 0x77, 0x3d, 0x00, 0xc1, 0x44, 0xc5, 0x25, + 0xac, 0x61, 0x9d, 0x18, 0xc8, 0x4a, 0x3f, 0x47, + 0x18, 0xe2, 0x44, 0x8b, 0x2f, 0xe3, 0x24, 0xd9, + 0xcc, 0xda, 0x27, 0x10, 0xac, 0xad, 0xe2, 0x56 +}; + +static const u8 T9[] = { + 0x99, 0x24, 0xa7, 0xc8, 0x58, 0x73, 0x36, 0xbf, + 0xb1, 0x18, 0x02, 0x4d, 0xb8, 0x67, 0x4a, 0x14 +}; /* Test Case 10 */ -#define K10 K9 -#define IV10 IV9 -static const u8 P10[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39}, - A10[]= {0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef,0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef, - 0xab,0xad,0xda,0xd2}, - C10[]= {0x39,0x80,0xca,0x0b,0x3c,0x00,0xe8,0x41,0xeb,0x06,0xfa,0xc4,0x87,0x2a,0x27,0x57, - 0x85,0x9e,0x1c,0xea,0xa6,0xef,0xd9,0x84,0x62,0x85,0x93,0xb4,0x0c,0xa1,0xe1,0x9c, - 0x7d,0x77,0x3d,0x00,0xc1,0x44,0xc5,0x25,0xac,0x61,0x9d,0x18,0xc8,0x4a,0x3f,0x47, - 0x18,0xe2,0x44,0x8b,0x2f,0xe3,0x24,0xd9,0xcc,0xda,0x27,0x10}, - T10[]= {0x25,0x19,0x49,0x8e,0x80,0xf1,0x47,0x8f,0x37,0xba,0x55,0xbd,0x6d,0x27,0x61,0x8c}; +# define K10 K9 +# define IV10 IV9 +static const u8 P10[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39 +}; + +static const u8 A10[] = { + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xab, 0xad, 0xda, 0xd2 +}; + +static const u8 C10[] = { + 0x39, 0x80, 0xca, 0x0b, 0x3c, 0x00, 0xe8, 0x41, + 0xeb, 0x06, 0xfa, 0xc4, 0x87, 0x2a, 0x27, 0x57, + 0x85, 0x9e, 0x1c, 0xea, 0xa6, 0xef, 0xd9, 0x84, + 0x62, 0x85, 0x93, 0xb4, 0x0c, 0xa1, 0xe1, 0x9c, + 0x7d, 0x77, 0x3d, 0x00, 0xc1, 0x44, 0xc5, 0x25, + 0xac, 0x61, 0x9d, 0x18, 0xc8, 0x4a, 0x3f, 0x47, + 0x18, 0xe2, 0x44, 0x8b, 0x2f, 0xe3, 0x24, 0xd9, + 0xcc, 0xda, 0x27, 0x10 +}; + +static const u8 T10[] = { + 0x25, 0x19, 0x49, 0x8e, 0x80, 0xf1, 0x47, 0x8f, + 0x37, 0xba, 0x55, 0xbd, 0x6d, 0x27, 0x61, 0x8c +}; /* Test Case 11 */ -#define K11 K10 -#define P11 P10 -#define A11 A10 -static const u8 IV11[]={0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad}, - C11[]= {0x0f,0x10,0xf5,0x99,0xae,0x14,0xa1,0x54,0xed,0x24,0xb3,0x6e,0x25,0x32,0x4d,0xb8, - 0xc5,0x66,0x63,0x2e,0xf2,0xbb,0xb3,0x4f,0x83,0x47,0x28,0x0f,0xc4,0x50,0x70,0x57, - 0xfd,0xdc,0x29,0xdf,0x9a,0x47,0x1f,0x75,0xc6,0x65,0x41,0xd4,0xd4,0xda,0xd1,0xc9, - 0xe9,0x3a,0x19,0xa5,0x8e,0x8b,0x47,0x3f,0xa0,0xf0,0x62,0xf7}, - T11[]= {0x65,0xdc,0xc5,0x7f,0xcf,0x62,0x3a,0x24,0x09,0x4f,0xcc,0xa4,0x0d,0x35,0x33,0xf8}; +# define K11 K10 +# define P11 P10 +# define A11 A10 +static const u8 IV11[] = { 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad }; + +static const u8 C11[] = { + 0x0f, 0x10, 0xf5, 0x99, 0xae, 0x14, 0xa1, 0x54, + 0xed, 0x24, 0xb3, 0x6e, 0x25, 0x32, 0x4d, 0xb8, + 0xc5, 0x66, 0x63, 0x2e, 0xf2, 0xbb, 0xb3, 0x4f, + 0x83, 0x47, 0x28, 0x0f, 0xc4, 0x50, 0x70, 0x57, + 0xfd, 0xdc, 0x29, 0xdf, 0x9a, 0x47, 0x1f, 0x75, + 0xc6, 0x65, 0x41, 0xd4, 0xd4, 0xda, 0xd1, 0xc9, + 0xe9, 0x3a, 0x19, 0xa5, 0x8e, 0x8b, 0x47, 0x3f, + 0xa0, 0xf0, 0x62, 0xf7 +}; + +static const u8 T11[] = { + 0x65, 0xdc, 0xc5, 0x7f, 0xcf, 0x62, 0x3a, 0x24, + 0x09, 0x4f, 0xcc, 0xa4, 0x0d, 0x35, 0x33, 0xf8 +}; /* Test Case 12 */ -#define K12 K11 -#define P12 P11 -#define A12 A11 -static const u8 IV12[]={0x93,0x13,0x22,0x5d,0xf8,0x84,0x06,0xe5,0x55,0x90,0x9c,0x5a,0xff,0x52,0x69,0xaa, - 0x6a,0x7a,0x95,0x38,0x53,0x4f,0x7d,0xa1,0xe4,0xc3,0x03,0xd2,0xa3,0x18,0xa7,0x28, - 0xc3,0xc0,0xc9,0x51,0x56,0x80,0x95,0x39,0xfc,0xf0,0xe2,0x42,0x9a,0x6b,0x52,0x54, - 0x16,0xae,0xdb,0xf5,0xa0,0xde,0x6a,0x57,0xa6,0x37,0xb3,0x9b}, - C12[]= {0xd2,0x7e,0x88,0x68,0x1c,0xe3,0x24,0x3c,0x48,0x30,0x16,0x5a,0x8f,0xdc,0xf9,0xff, - 0x1d,0xe9,0xa1,0xd8,0xe6,0xb4,0x47,0xef,0x6e,0xf7,0xb7,0x98,0x28,0x66,0x6e,0x45, - 0x81,0xe7,0x90,0x12,0xaf,0x34,0xdd,0xd9,0xe2,0xf0,0x37,0x58,0x9b,0x29,0x2d,0xb3, - 0xe6,0x7c,0x03,0x67,0x45,0xfa,0x22,0xe7,0xe9,0xb7,0x37,0x3b}, - T12[]= {0xdc,0xf5,0x66,0xff,0x29,0x1c,0x25,0xbb,0xb8,0x56,0x8f,0xc3,0xd3,0x76,0xa6,0xd9}; +# define K12 K11 +# define P12 P11 +# define A12 A11 +static const u8 IV12[] = { + 0x93, 0x13, 0x22, 0x5d, 0xf8, 0x84, 0x06, 0xe5, + 0x55, 0x90, 0x9c, 0x5a, 0xff, 0x52, 0x69, 0xaa, + 0x6a, 0x7a, 0x95, 0x38, 0x53, 0x4f, 0x7d, 0xa1, + 0xe4, 0xc3, 0x03, 0xd2, 0xa3, 0x18, 0xa7, 0x28, + 0xc3, 0xc0, 0xc9, 0x51, 0x56, 0x80, 0x95, 0x39, + 0xfc, 0xf0, 0xe2, 0x42, 0x9a, 0x6b, 0x52, 0x54, + 0x16, 0xae, 0xdb, 0xf5, 0xa0, 0xde, 0x6a, 0x57, + 0xa6, 0x37, 0xb3, 0x9b +}; + +static const u8 C12[] = { + 0xd2, 0x7e, 0x88, 0x68, 0x1c, 0xe3, 0x24, 0x3c, + 0x48, 0x30, 0x16, 0x5a, 0x8f, 0xdc, 0xf9, 0xff, + 0x1d, 0xe9, 0xa1, 0xd8, 0xe6, 0xb4, 0x47, 0xef, + 0x6e, 0xf7, 0xb7, 0x98, 0x28, 0x66, 0x6e, 0x45, + 0x81, 0xe7, 0x90, 0x12, 0xaf, 0x34, 0xdd, 0xd9, + 0xe2, 0xf0, 0x37, 0x58, 0x9b, 0x29, 0x2d, 0xb3, + 0xe6, 0x7c, 0x03, 0x67, 0x45, 0xfa, 0x22, 0xe7, + 0xe9, 0xb7, 0x37, 0x3b +}; + +static const u8 T12[] = { + 0xdc, 0xf5, 0x66, 0xff, 0x29, 0x1c, 0x25, 0xbb, + 0xb8, 0x56, 0x8f, 0xc3, 0xd3, 0x76, 0xa6, 0xd9 +}; /* Test Case 13 */ -static const u8 K13[32], - *P13=NULL, - *A13=NULL, - IV13[12], - *C13=NULL, - T13[]={0x53,0x0f,0x8a,0xfb,0xc7,0x45,0x36,0xb9,0xa9,0x63,0xb4,0xf1,0xc4,0xcb,0x73,0x8b}; +static const u8 K13[32], *P13 = NULL, *A13 = NULL, IV13[12], *C13 = NULL; +static const u8 T13[] = { + 0x53, 0x0f, 0x8a, 0xfb, 0xc7, 0x45, 0x36, 0xb9, + 0xa9, 0x63, 0xb4, 0xf1, 0xc4, 0xcb, 0x73, 0x8b +}; /* Test Case 14 */ -#define K14 K13 -#define A14 A13 -static const u8 P14[16], - IV14[12], - C14[]= {0xce,0xa7,0x40,0x3d,0x4d,0x60,0x6b,0x6e,0x07,0x4e,0xc5,0xd3,0xba,0xf3,0x9d,0x18}, - T14[]= {0xd0,0xd1,0xc8,0xa7,0x99,0x99,0x6b,0xf0,0x26,0x5b,0x98,0xb5,0xd4,0x8a,0xb9,0x19}; +# define K14 K13 +# define A14 A13 +static const u8 P14[16], IV14[12]; +static const u8 C14[] = { + 0xce, 0xa7, 0x40, 0x3d, 0x4d, 0x60, 0x6b, 0x6e, + 0x07, 0x4e, 0xc5, 0xd3, 0xba, 0xf3, 0x9d, 0x18 +}; + +static const u8 T14[] = { + 0xd0, 0xd1, 0xc8, 0xa7, 0x99, 0x99, 0x6b, 0xf0, + 0x26, 0x5b, 0x98, 0xb5, 0xd4, 0x8a, 0xb9, 0x19 +}; /* Test Case 15 */ -#define A15 A14 -static const u8 K15[]= {0xfe,0xff,0xe9,0x92,0x86,0x65,0x73,0x1c,0x6d,0x6a,0x8f,0x94,0x67,0x30,0x83,0x08, - 0xfe,0xff,0xe9,0x92,0x86,0x65,0x73,0x1c,0x6d,0x6a,0x8f,0x94,0x67,0x30,0x83,0x08}, - P15[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39,0x1a,0xaf,0xd2,0x55}, - IV15[]={0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad,0xde,0xca,0xf8,0x88}, - C15[]= {0x52,0x2d,0xc1,0xf0,0x99,0x56,0x7d,0x07,0xf4,0x7f,0x37,0xa3,0x2a,0x84,0x42,0x7d, - 0x64,0x3a,0x8c,0xdc,0xbf,0xe5,0xc0,0xc9,0x75,0x98,0xa2,0xbd,0x25,0x55,0xd1,0xaa, - 0x8c,0xb0,0x8e,0x48,0x59,0x0d,0xbb,0x3d,0xa7,0xb0,0x8b,0x10,0x56,0x82,0x88,0x38, - 0xc5,0xf6,0x1e,0x63,0x93,0xba,0x7a,0x0a,0xbc,0xc9,0xf6,0x62,0x89,0x80,0x15,0xad}, - T15[]= {0xb0,0x94,0xda,0xc5,0xd9,0x34,0x71,0xbd,0xec,0x1a,0x50,0x22,0x70,0xe3,0xcc,0x6c}; +# define A15 A14 +static const u8 K15[] = { + 0xfe, 0xff, 0xe9, 0x92, 0x86, 0x65, 0x73, 0x1c, + 0x6d, 0x6a, 0x8f, 0x94, 0x67, 0x30, 0x83, 0x08, + 0xfe, 0xff, 0xe9, 0x92, 0x86, 0x65, 0x73, 0x1c, + 0x6d, 0x6a, 0x8f, 0x94, 0x67, 0x30, 0x83, 0x08 +}; + +static const u8 P15[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39, 0x1a, 0xaf, 0xd2, 0x55 +}; + +static const u8 IV15[] = { + 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad, + 0xde, 0xca, 0xf8, 0x88 +}; + +static const u8 C15[] = { + 0x52, 0x2d, 0xc1, 0xf0, 0x99, 0x56, 0x7d, 0x07, + 0xf4, 0x7f, 0x37, 0xa3, 0x2a, 0x84, 0x42, 0x7d, + 0x64, 0x3a, 0x8c, 0xdc, 0xbf, 0xe5, 0xc0, 0xc9, + 0x75, 0x98, 0xa2, 0xbd, 0x25, 0x55, 0xd1, 0xaa, + 0x8c, 0xb0, 0x8e, 0x48, 0x59, 0x0d, 0xbb, 0x3d, + 0xa7, 0xb0, 0x8b, 0x10, 0x56, 0x82, 0x88, 0x38, + 0xc5, 0xf6, 0x1e, 0x63, 0x93, 0xba, 0x7a, 0x0a, + 0xbc, 0xc9, 0xf6, 0x62, 0x89, 0x80, 0x15, 0xad +}; + +static const u8 T15[] = { + 0xb0, 0x94, 0xda, 0xc5, 0xd9, 0x34, 0x71, 0xbd, + 0xec, 0x1a, 0x50, 0x22, 0x70, 0xe3, 0xcc, 0x6c +}; /* Test Case 16 */ -#define K16 K15 -#define IV16 IV15 -static const u8 P16[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39}, - A16[]= {0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef,0xfe,0xed,0xfa,0xce,0xde,0xad,0xbe,0xef, - 0xab,0xad,0xda,0xd2}, - C16[]= {0x52,0x2d,0xc1,0xf0,0x99,0x56,0x7d,0x07,0xf4,0x7f,0x37,0xa3,0x2a,0x84,0x42,0x7d, - 0x64,0x3a,0x8c,0xdc,0xbf,0xe5,0xc0,0xc9,0x75,0x98,0xa2,0xbd,0x25,0x55,0xd1,0xaa, - 0x8c,0xb0,0x8e,0x48,0x59,0x0d,0xbb,0x3d,0xa7,0xb0,0x8b,0x10,0x56,0x82,0x88,0x38, - 0xc5,0xf6,0x1e,0x63,0x93,0xba,0x7a,0x0a,0xbc,0xc9,0xf6,0x62}, - T16[]= {0x76,0xfc,0x6e,0xce,0x0f,0x4e,0x17,0x68,0xcd,0xdf,0x88,0x53,0xbb,0x2d,0x55,0x1b}; +# define K16 K15 +# define IV16 IV15 +static const u8 P16[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39 +}; + +static const u8 A16[] = { + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xfe, 0xed, 0xfa, 0xce, 0xde, 0xad, 0xbe, 0xef, + 0xab, 0xad, 0xda, 0xd2 +}; + +static const u8 C16[] = { + 0x52, 0x2d, 0xc1, 0xf0, 0x99, 0x56, 0x7d, 0x07, + 0xf4, 0x7f, 0x37, 0xa3, 0x2a, 0x84, 0x42, 0x7d, + 0x64, 0x3a, 0x8c, 0xdc, 0xbf, 0xe5, 0xc0, 0xc9, + 0x75, 0x98, 0xa2, 0xbd, 0x25, 0x55, 0xd1, 0xaa, + 0x8c, 0xb0, 0x8e, 0x48, 0x59, 0x0d, 0xbb, 0x3d, + 0xa7, 0xb0, 0x8b, 0x10, 0x56, 0x82, 0x88, 0x38, + 0xc5, 0xf6, 0x1e, 0x63, 0x93, 0xba, 0x7a, 0x0a, + 0xbc, 0xc9, 0xf6, 0x62 +}; + +static const u8 T16[] = { + 0x76, 0xfc, 0x6e, 0xce, 0x0f, 0x4e, 0x17, 0x68, + 0xcd, 0xdf, 0x88, 0x53, 0xbb, 0x2d, 0x55, 0x1b +}; /* Test Case 17 */ -#define K17 K16 -#define P17 P16 -#define A17 A16 -static const u8 IV17[]={0xca,0xfe,0xba,0xbe,0xfa,0xce,0xdb,0xad}, - C17[]= {0xc3,0x76,0x2d,0xf1,0xca,0x78,0x7d,0x32,0xae,0x47,0xc1,0x3b,0xf1,0x98,0x44,0xcb, - 0xaf,0x1a,0xe1,0x4d,0x0b,0x97,0x6a,0xfa,0xc5,0x2f,0xf7,0xd7,0x9b,0xba,0x9d,0xe0, - 0xfe,0xb5,0x82,0xd3,0x39,0x34,0xa4,0xf0,0x95,0x4c,0xc2,0x36,0x3b,0xc7,0x3f,0x78, - 0x62,0xac,0x43,0x0e,0x64,0xab,0xe4,0x99,0xf4,0x7c,0x9b,0x1f}, - T17[]= {0x3a,0x33,0x7d,0xbf,0x46,0xa7,0x92,0xc4,0x5e,0x45,0x49,0x13,0xfe,0x2e,0xa8,0xf2}; +# define K17 K16 +# define P17 P16 +# define A17 A16 +static const u8 IV17[] = { 0xca, 0xfe, 0xba, 0xbe, 0xfa, 0xce, 0xdb, 0xad }; + +static const u8 C17[] = { + 0xc3, 0x76, 0x2d, 0xf1, 0xca, 0x78, 0x7d, 0x32, + 0xae, 0x47, 0xc1, 0x3b, 0xf1, 0x98, 0x44, 0xcb, + 0xaf, 0x1a, 0xe1, 0x4d, 0x0b, 0x97, 0x6a, 0xfa, + 0xc5, 0x2f, 0xf7, 0xd7, 0x9b, 0xba, 0x9d, 0xe0, + 0xfe, 0xb5, 0x82, 0xd3, 0x39, 0x34, 0xa4, 0xf0, + 0x95, 0x4c, 0xc2, 0x36, 0x3b, 0xc7, 0x3f, 0x78, + 0x62, 0xac, 0x43, 0x0e, 0x64, 0xab, 0xe4, 0x99, + 0xf4, 0x7c, 0x9b, 0x1f +}; + +static const u8 T17[] = { + 0x3a, 0x33, 0x7d, 0xbf, 0x46, 0xa7, 0x92, 0xc4, + 0x5e, 0x45, 0x49, 0x13, 0xfe, 0x2e, 0xa8, 0xf2 +}; /* Test Case 18 */ -#define K18 K17 -#define P18 P17 -#define A18 A17 -static const u8 IV18[]={0x93,0x13,0x22,0x5d,0xf8,0x84,0x06,0xe5,0x55,0x90,0x9c,0x5a,0xff,0x52,0x69,0xaa, - 0x6a,0x7a,0x95,0x38,0x53,0x4f,0x7d,0xa1,0xe4,0xc3,0x03,0xd2,0xa3,0x18,0xa7,0x28, - 0xc3,0xc0,0xc9,0x51,0x56,0x80,0x95,0x39,0xfc,0xf0,0xe2,0x42,0x9a,0x6b,0x52,0x54, - 0x16,0xae,0xdb,0xf5,0xa0,0xde,0x6a,0x57,0xa6,0x37,0xb3,0x9b}, - C18[]= {0x5a,0x8d,0xef,0x2f,0x0c,0x9e,0x53,0xf1,0xf7,0x5d,0x78,0x53,0x65,0x9e,0x2a,0x20, - 0xee,0xb2,0xb2,0x2a,0xaf,0xde,0x64,0x19,0xa0,0x58,0xab,0x4f,0x6f,0x74,0x6b,0xf4, - 0x0f,0xc0,0xc3,0xb7,0x80,0xf2,0x44,0x45,0x2d,0xa3,0xeb,0xf1,0xc5,0xd8,0x2c,0xde, - 0xa2,0x41,0x89,0x97,0x20,0x0e,0xf8,0x2e,0x44,0xae,0x7e,0x3f}, - T18[]= {0xa4,0x4a,0x82,0x66,0xee,0x1c,0x8e,0xb0,0xc8,0xb5,0xd4,0xcf,0x5a,0xe9,0xf1,0x9a}; +# define K18 K17 +# define P18 P17 +# define A18 A17 +static const u8 IV18[] = { + 0x93, 0x13, 0x22, 0x5d, 0xf8, 0x84, 0x06, 0xe5, + 0x55, 0x90, 0x9c, 0x5a, 0xff, 0x52, 0x69, 0xaa, + 0x6a, 0x7a, 0x95, 0x38, 0x53, 0x4f, 0x7d, 0xa1, + 0xe4, 0xc3, 0x03, 0xd2, 0xa3, 0x18, 0xa7, 0x28, + 0xc3, 0xc0, 0xc9, 0x51, 0x56, 0x80, 0x95, 0x39, + 0xfc, 0xf0, 0xe2, 0x42, 0x9a, 0x6b, 0x52, 0x54, + 0x16, 0xae, 0xdb, 0xf5, 0xa0, 0xde, 0x6a, 0x57, + 0xa6, 0x37, 0xb3, 0x9b +}; + +static const u8 C18[] = { + 0x5a, 0x8d, 0xef, 0x2f, 0x0c, 0x9e, 0x53, 0xf1, + 0xf7, 0x5d, 0x78, 0x53, 0x65, 0x9e, 0x2a, 0x20, + 0xee, 0xb2, 0xb2, 0x2a, 0xaf, 0xde, 0x64, 0x19, + 0xa0, 0x58, 0xab, 0x4f, 0x6f, 0x74, 0x6b, 0xf4, + 0x0f, 0xc0, 0xc3, 0xb7, 0x80, 0xf2, 0x44, 0x45, + 0x2d, 0xa3, 0xeb, 0xf1, 0xc5, 0xd8, 0x2c, 0xde, + 0xa2, 0x41, 0x89, 0x97, 0x20, 0x0e, 0xf8, 0x2e, + 0x44, 0xae, 0x7e, 0x3f +}; + +static const u8 T18[] = { + 0xa4, 0x4a, 0x82, 0x66, 0xee, 0x1c, 0x8e, 0xb0, + 0xc8, 0xb5, 0xd4, 0xcf, 0x5a, 0xe9, 0xf1, 0x9a +}; /* Test Case 19 */ -#define K19 K1 -#define P19 P1 -#define IV19 IV1 -#define C19 C1 -static const u8 A19[]= {0xd9,0x31,0x32,0x25,0xf8,0x84,0x06,0xe5,0xa5,0x59,0x09,0xc5,0xaf,0xf5,0x26,0x9a, - 0x86,0xa7,0xa9,0x53,0x15,0x34,0xf7,0xda,0x2e,0x4c,0x30,0x3d,0x8a,0x31,0x8a,0x72, - 0x1c,0x3c,0x0c,0x95,0x95,0x68,0x09,0x53,0x2f,0xcf,0x0e,0x24,0x49,0xa6,0xb5,0x25, - 0xb1,0x6a,0xed,0xf5,0xaa,0x0d,0xe6,0x57,0xba,0x63,0x7b,0x39,0x1a,0xaf,0xd2,0x55, - 0x52,0x2d,0xc1,0xf0,0x99,0x56,0x7d,0x07,0xf4,0x7f,0x37,0xa3,0x2a,0x84,0x42,0x7d, - 0x64,0x3a,0x8c,0xdc,0xbf,0xe5,0xc0,0xc9,0x75,0x98,0xa2,0xbd,0x25,0x55,0xd1,0xaa, - 0x8c,0xb0,0x8e,0x48,0x59,0x0d,0xbb,0x3d,0xa7,0xb0,0x8b,0x10,0x56,0x82,0x88,0x38, - 0xc5,0xf6,0x1e,0x63,0x93,0xba,0x7a,0x0a,0xbc,0xc9,0xf6,0x62,0x89,0x80,0x15,0xad}, - T19[]= {0x5f,0xea,0x79,0x3a,0x2d,0x6f,0x97,0x4d,0x37,0xe6,0x8e,0x0c,0xb8,0xff,0x94,0x92}; +# define K19 K1 +# define P19 P1 +# define IV19 IV1 +# define C19 C1 +static const u8 A19[] = { + 0xd9, 0x31, 0x32, 0x25, 0xf8, 0x84, 0x06, 0xe5, + 0xa5, 0x59, 0x09, 0xc5, 0xaf, 0xf5, 0x26, 0x9a, + 0x86, 0xa7, 0xa9, 0x53, 0x15, 0x34, 0xf7, 0xda, + 0x2e, 0x4c, 0x30, 0x3d, 0x8a, 0x31, 0x8a, 0x72, + 0x1c, 0x3c, 0x0c, 0x95, 0x95, 0x68, 0x09, 0x53, + 0x2f, 0xcf, 0x0e, 0x24, 0x49, 0xa6, 0xb5, 0x25, + 0xb1, 0x6a, 0xed, 0xf5, 0xaa, 0x0d, 0xe6, 0x57, + 0xba, 0x63, 0x7b, 0x39, 0x1a, 0xaf, 0xd2, 0x55, + 0x52, 0x2d, 0xc1, 0xf0, 0x99, 0x56, 0x7d, 0x07, + 0xf4, 0x7f, 0x37, 0xa3, 0x2a, 0x84, 0x42, 0x7d, + 0x64, 0x3a, 0x8c, 0xdc, 0xbf, 0xe5, 0xc0, 0xc9, + 0x75, 0x98, 0xa2, 0xbd, 0x25, 0x55, 0xd1, 0xaa, + 0x8c, 0xb0, 0x8e, 0x48, 0x59, 0x0d, 0xbb, 0x3d, + 0xa7, 0xb0, 0x8b, 0x10, 0x56, 0x82, 0x88, 0x38, + 0xc5, 0xf6, 0x1e, 0x63, 0x93, 0xba, 0x7a, 0x0a, + 0xbc, 0xc9, 0xf6, 0x62, 0x89, 0x80, 0x15, 0xad +}; + +static const u8 T19[] = { + 0x5f, 0xea, 0x79, 0x3a, 0x2d, 0x6f, 0x97, 0x4d, + 0x37, 0xe6, 0x8e, 0x0c, 0xb8, 0xff, 0x94, 0x92 +}; /* Test Case 20 */ -#define K20 K1 -#define A20 A1 -static const u8 IV20[64]={0xff,0xff,0xff,0xff}, /* this results in 0xff in counter LSB */ - P20[288], - C20[]= {0x56,0xb3,0x37,0x3c,0xa9,0xef,0x6e,0x4a,0x2b,0x64,0xfe,0x1e,0x9a,0x17,0xb6,0x14, - 0x25,0xf1,0x0d,0x47,0xa7,0x5a,0x5f,0xce,0x13,0xef,0xc6,0xbc,0x78,0x4a,0xf2,0x4f, - 0x41,0x41,0xbd,0xd4,0x8c,0xf7,0xc7,0x70,0x88,0x7a,0xfd,0x57,0x3c,0xca,0x54,0x18, - 0xa9,0xae,0xff,0xcd,0x7c,0x5c,0xed,0xdf,0xc6,0xa7,0x83,0x97,0xb9,0xa8,0x5b,0x49, - 0x9d,0xa5,0x58,0x25,0x72,0x67,0xca,0xab,0x2a,0xd0,0xb2,0x3c,0xa4,0x76,0xa5,0x3c, - 0xb1,0x7f,0xb4,0x1c,0x4b,0x8b,0x47,0x5c,0xb4,0xf3,0xf7,0x16,0x50,0x94,0xc2,0x29, - 0xc9,0xe8,0xc4,0xdc,0x0a,0x2a,0x5f,0xf1,0x90,0x3e,0x50,0x15,0x11,0x22,0x13,0x76, - 0xa1,0xcd,0xb8,0x36,0x4c,0x50,0x61,0xa2,0x0c,0xae,0x74,0xbc,0x4a,0xcd,0x76,0xce, - 0xb0,0xab,0xc9,0xfd,0x32,0x17,0xef,0x9f,0x8c,0x90,0xbe,0x40,0x2d,0xdf,0x6d,0x86, - 0x97,0xf4,0xf8,0x80,0xdf,0xf1,0x5b,0xfb,0x7a,0x6b,0x28,0x24,0x1e,0xc8,0xfe,0x18, - 0x3c,0x2d,0x59,0xe3,0xf9,0xdf,0xff,0x65,0x3c,0x71,0x26,0xf0,0xac,0xb9,0xe6,0x42, - 0x11,0xf4,0x2b,0xae,0x12,0xaf,0x46,0x2b,0x10,0x70,0xbe,0xf1,0xab,0x5e,0x36,0x06, - 0x87,0x2c,0xa1,0x0d,0xee,0x15,0xb3,0x24,0x9b,0x1a,0x1b,0x95,0x8f,0x23,0x13,0x4c, - 0x4b,0xcc,0xb7,0xd0,0x32,0x00,0xbc,0xe4,0x20,0xa2,0xf8,0xeb,0x66,0xdc,0xf3,0x64, - 0x4d,0x14,0x23,0xc1,0xb5,0x69,0x90,0x03,0xc1,0x3e,0xce,0xf4,0xbf,0x38,0xa3,0xb6, - 0x0e,0xed,0xc3,0x40,0x33,0xba,0xc1,0x90,0x27,0x83,0xdc,0x6d,0x89,0xe2,0xe7,0x74, - 0x18,0x8a,0x43,0x9c,0x7e,0xbc,0xc0,0x67,0x2d,0xbd,0xa4,0xdd,0xcf,0xb2,0x79,0x46, - 0x13,0xb0,0xbe,0x41,0x31,0x5e,0xf7,0x78,0x70,0x8a,0x70,0xee,0x7d,0x75,0x16,0x5c}, - T20[]= {0x8b,0x30,0x7f,0x6b,0x33,0x28,0x6d,0x0a,0xb0,0x26,0xa9,0xed,0x3f,0xe1,0xe8,0x5f}; - -#define TEST_CASE(n) do { \ - u8 out[sizeof(P##n)]; \ - AES_set_encrypt_key(K##n,sizeof(K##n)*8,&key); \ - CRYPTO_gcm128_init(&ctx,&key,(block128_f)AES_encrypt); \ - CRYPTO_gcm128_setiv(&ctx,IV##n,sizeof(IV##n)); \ - memset(out,0,sizeof(out)); \ - if (A##n) CRYPTO_gcm128_aad(&ctx,A##n,sizeof(A##n)); \ - if (P##n) CRYPTO_gcm128_encrypt(&ctx,P##n,out,sizeof(out)); \ - if (CRYPTO_gcm128_finish(&ctx,T##n,16) || \ - (C##n && memcmp(out,C##n,sizeof(out)))) \ - ret++, printf ("encrypt test#%d failed.\n",n); \ - CRYPTO_gcm128_setiv(&ctx,IV##n,sizeof(IV##n)); \ - memset(out,0,sizeof(out)); \ - if (A##n) CRYPTO_gcm128_aad(&ctx,A##n,sizeof(A##n)); \ - if (C##n) CRYPTO_gcm128_decrypt(&ctx,C##n,out,sizeof(out)); \ - if (CRYPTO_gcm128_finish(&ctx,T##n,16) || \ - (P##n && memcmp(out,P##n,sizeof(out)))) \ - ret++, printf ("decrypt test#%d failed.\n",n); \ - } while(0) +# define K20 K1 +# define A20 A1 +/* this results in 0xff in counter LSB */ +static const u8 IV20[64] = { 0xff, 0xff, 0xff, 0xff }; + +static const u8 P20[288]; +static const u8 C20[] = { + 0x56, 0xb3, 0x37, 0x3c, 0xa9, 0xef, 0x6e, 0x4a, + 0x2b, 0x64, 0xfe, 0x1e, 0x9a, 0x17, 0xb6, 0x14, + 0x25, 0xf1, 0x0d, 0x47, 0xa7, 0x5a, 0x5f, 0xce, + 0x13, 0xef, 0xc6, 0xbc, 0x78, 0x4a, 0xf2, 0x4f, + 0x41, 0x41, 0xbd, 0xd4, 0x8c, 0xf7, 0xc7, 0x70, + 0x88, 0x7a, 0xfd, 0x57, 0x3c, 0xca, 0x54, 0x18, + 0xa9, 0xae, 0xff, 0xcd, 0x7c, 0x5c, 0xed, 0xdf, + 0xc6, 0xa7, 0x83, 0x97, 0xb9, 0xa8, 0x5b, 0x49, + 0x9d, 0xa5, 0x58, 0x25, 0x72, 0x67, 0xca, 0xab, + 0x2a, 0xd0, 0xb2, 0x3c, 0xa4, 0x76, 0xa5, 0x3c, + 0xb1, 0x7f, 0xb4, 0x1c, 0x4b, 0x8b, 0x47, 0x5c, + 0xb4, 0xf3, 0xf7, 0x16, 0x50, 0x94, 0xc2, 0x29, + 0xc9, 0xe8, 0xc4, 0xdc, 0x0a, 0x2a, 0x5f, 0xf1, + 0x90, 0x3e, 0x50, 0x15, 0x11, 0x22, 0x13, 0x76, + 0xa1, 0xcd, 0xb8, 0x36, 0x4c, 0x50, 0x61, 0xa2, + 0x0c, 0xae, 0x74, 0xbc, 0x4a, 0xcd, 0x76, 0xce, + 0xb0, 0xab, 0xc9, 0xfd, 0x32, 0x17, 0xef, 0x9f, + 0x8c, 0x90, 0xbe, 0x40, 0x2d, 0xdf, 0x6d, 0x86, + 0x97, 0xf4, 0xf8, 0x80, 0xdf, 0xf1, 0x5b, 0xfb, + 0x7a, 0x6b, 0x28, 0x24, 0x1e, 0xc8, 0xfe, 0x18, + 0x3c, 0x2d, 0x59, 0xe3, 0xf9, 0xdf, 0xff, 0x65, + 0x3c, 0x71, 0x26, 0xf0, 0xac, 0xb9, 0xe6, 0x42, + 0x11, 0xf4, 0x2b, 0xae, 0x12, 0xaf, 0x46, 0x2b, + 0x10, 0x70, 0xbe, 0xf1, 0xab, 0x5e, 0x36, 0x06, + 0x87, 0x2c, 0xa1, 0x0d, 0xee, 0x15, 0xb3, 0x24, + 0x9b, 0x1a, 0x1b, 0x95, 0x8f, 0x23, 0x13, 0x4c, + 0x4b, 0xcc, 0xb7, 0xd0, 0x32, 0x00, 0xbc, 0xe4, + 0x20, 0xa2, 0xf8, 0xeb, 0x66, 0xdc, 0xf3, 0x64, + 0x4d, 0x14, 0x23, 0xc1, 0xb5, 0x69, 0x90, 0x03, + 0xc1, 0x3e, 0xce, 0xf4, 0xbf, 0x38, 0xa3, 0xb6, + 0x0e, 0xed, 0xc3, 0x40, 0x33, 0xba, 0xc1, 0x90, + 0x27, 0x83, 0xdc, 0x6d, 0x89, 0xe2, 0xe7, 0x74, + 0x18, 0x8a, 0x43, 0x9c, 0x7e, 0xbc, 0xc0, 0x67, + 0x2d, 0xbd, 0xa4, 0xdd, 0xcf, 0xb2, 0x79, 0x46, + 0x13, 0xb0, 0xbe, 0x41, 0x31, 0x5e, 0xf7, 0x78, + 0x70, 0x8a, 0x70, 0xee, 0x7d, 0x75, 0x16, 0x5c +}; + +static const u8 T20[] = { + 0x8b, 0x30, 0x7f, 0x6b, 0x33, 0x28, 0x6d, 0x0a, + 0xb0, 0x26, 0xa9, 0xed, 0x3f, 0xe1, 0xe8, 0x5f +}; + +# define TEST_CASE(n) do { \ + u8 out[sizeof(P##n)]; \ + AES_set_encrypt_key(K##n,sizeof(K##n)*8,&key); \ + CRYPTO_gcm128_init(&ctx,&key,(block128_f)AES_encrypt); \ + CRYPTO_gcm128_setiv(&ctx,IV##n,sizeof(IV##n)); \ + memset(out,0,sizeof(out)); \ + if (A##n) CRYPTO_gcm128_aad(&ctx,A##n,sizeof(A##n)); \ + if (P##n) CRYPTO_gcm128_encrypt(&ctx,P##n,out,sizeof(out)); \ + if (CRYPTO_gcm128_finish(&ctx,T##n,16) || \ + (C##n && memcmp(out,C##n,sizeof(out)))) \ + ret++, printf ("encrypt test#%d failed.\n",n); \ + CRYPTO_gcm128_setiv(&ctx,IV##n,sizeof(IV##n)); \ + memset(out,0,sizeof(out)); \ + if (A##n) CRYPTO_gcm128_aad(&ctx,A##n,sizeof(A##n)); \ + if (C##n) CRYPTO_gcm128_decrypt(&ctx,C##n,out,sizeof(out)); \ + if (CRYPTO_gcm128_finish(&ctx,T##n,16) || \ + (P##n && memcmp(out,P##n,sizeof(out)))) \ + ret++, printf ("decrypt test#%d failed.\n",n); \ + } while(0) int main() { - GCM128_CONTEXT ctx; - AES_KEY key; - int ret=0; - - TEST_CASE(1); - TEST_CASE(2); - TEST_CASE(3); - TEST_CASE(4); - TEST_CASE(5); - TEST_CASE(6); - TEST_CASE(7); - TEST_CASE(8); - TEST_CASE(9); - TEST_CASE(10); - TEST_CASE(11); - TEST_CASE(12); - TEST_CASE(13); - TEST_CASE(14); - TEST_CASE(15); - TEST_CASE(16); - TEST_CASE(17); - TEST_CASE(18); - TEST_CASE(19); - TEST_CASE(20); - -#ifdef OPENSSL_CPUID_OBJ - { - size_t start,stop,gcm_t,ctr_t,OPENSSL_rdtsc(); - union { u64 u; u8 c[1024]; } buf; - int i; - - AES_set_encrypt_key(K1,sizeof(K1)*8,&key); - CRYPTO_gcm128_init(&ctx,&key,(block128_f)AES_encrypt); - CRYPTO_gcm128_setiv(&ctx,IV1,sizeof(IV1)); - - CRYPTO_gcm128_encrypt(&ctx,buf.c,buf.c,sizeof(buf)); - start = OPENSSL_rdtsc(); - CRYPTO_gcm128_encrypt(&ctx,buf.c,buf.c,sizeof(buf)); - gcm_t = OPENSSL_rdtsc() - start; - - CRYPTO_ctr128_encrypt(buf.c,buf.c,sizeof(buf), - &key,ctx.Yi.c,ctx.EKi.c,&ctx.mres, - (block128_f)AES_encrypt); - start = OPENSSL_rdtsc(); - CRYPTO_ctr128_encrypt(buf.c,buf.c,sizeof(buf), - &key,ctx.Yi.c,ctx.EKi.c,&ctx.mres, - (block128_f)AES_encrypt); - ctr_t = OPENSSL_rdtsc() - start; - - printf("%.2f-%.2f=%.2f\n", - gcm_t/(double)sizeof(buf), - ctr_t/(double)sizeof(buf), - (gcm_t-ctr_t)/(double)sizeof(buf)); -#ifdef GHASH - { - void (*gcm_ghash_p)(u64 Xi[2],const u128 Htable[16], - const u8 *inp,size_t len) = ctx.ghash; - - GHASH((&ctx),buf.c,sizeof(buf)); - start = OPENSSL_rdtsc(); - for (i=0;i<100;++i) GHASH((&ctx),buf.c,sizeof(buf)); - gcm_t = OPENSSL_rdtsc() - start; - printf("%.2f\n",gcm_t/(double)sizeof(buf)/(double)i); - } -#endif - } -#endif + GCM128_CONTEXT ctx; + AES_KEY key; + int ret = 0; + + TEST_CASE(1); + TEST_CASE(2); + TEST_CASE(3); + TEST_CASE(4); + TEST_CASE(5); + TEST_CASE(6); + TEST_CASE(7); + TEST_CASE(8); + TEST_CASE(9); + TEST_CASE(10); + TEST_CASE(11); + TEST_CASE(12); + TEST_CASE(13); + TEST_CASE(14); + TEST_CASE(15); + TEST_CASE(16); + TEST_CASE(17); + TEST_CASE(18); + TEST_CASE(19); + TEST_CASE(20); + +# ifdef OPENSSL_CPUID_OBJ + { + size_t start, stop, gcm_t, ctr_t, OPENSSL_rdtsc(); + union { + u64 u; + u8 c[1024]; + } buf; + int i; + + AES_set_encrypt_key(K1, sizeof(K1) * 8, &key); + CRYPTO_gcm128_init(&ctx, &key, (block128_f) AES_encrypt); + CRYPTO_gcm128_setiv(&ctx, IV1, sizeof(IV1)); + + CRYPTO_gcm128_encrypt(&ctx, buf.c, buf.c, sizeof(buf)); + start = OPENSSL_rdtsc(); + CRYPTO_gcm128_encrypt(&ctx, buf.c, buf.c, sizeof(buf)); + gcm_t = OPENSSL_rdtsc() - start; + + CRYPTO_ctr128_encrypt(buf.c, buf.c, sizeof(buf), + &key, ctx.Yi.c, ctx.EKi.c, &ctx.mres, + (block128_f) AES_encrypt); + start = OPENSSL_rdtsc(); + CRYPTO_ctr128_encrypt(buf.c, buf.c, sizeof(buf), + &key, ctx.Yi.c, ctx.EKi.c, &ctx.mres, + (block128_f) AES_encrypt); + ctr_t = OPENSSL_rdtsc() - start; + + printf("%.2f-%.2f=%.2f\n", + gcm_t / (double)sizeof(buf), + ctr_t / (double)sizeof(buf), + (gcm_t - ctr_t) / (double)sizeof(buf)); +# ifdef GHASH + { + void (*gcm_ghash_p) (u64 Xi[2], const u128 Htable[16], + const u8 *inp, size_t len) = ctx.ghash; + + GHASH((&ctx), buf.c, sizeof(buf)); + start = OPENSSL_rdtsc(); + for (i = 0; i < 100; ++i) + GHASH((&ctx), buf.c, sizeof(buf)); + gcm_t = OPENSSL_rdtsc() - start; + printf("%.2f\n", gcm_t / (double)sizeof(buf) / (double)i); + } +# endif + } +# endif - return ret; + return ret; } #endif diff --git a/drivers/builtin_openssl2/crypto/modes/modes_lcl.h b/drivers/builtin_openssl2/crypto/modes/modes_lcl.h index 9d83e12844..296849b867 100644 --- a/drivers/builtin_openssl2/crypto/modes/modes_lcl.h +++ b/drivers/builtin_openssl2/crypto/modes/modes_lcl.h @@ -7,122 +7,127 @@ #include <openssl/modes.h> - #if (defined(_WIN32) || defined(_WIN64)) && !defined(__MINGW32__) typedef __int64 i64; typedef unsigned __int64 u64; -#define U64(C) C##UI64 +# define U64(C) C##UI64 #elif defined(__arch64__) typedef long i64; typedef unsigned long u64; -#define U64(C) C##UL +# define U64(C) C##UL #else typedef long long i64; typedef unsigned long long u64; -#define U64(C) C##ULL +# define U64(C) C##ULL #endif typedef unsigned int u32; typedef unsigned char u8; #define STRICT_ALIGNMENT 1 -#if defined(__i386) || defined(__i386__) || \ - defined(__x86_64) || defined(__x86_64__) || \ - defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64) || \ - defined(__s390__) || defined(__s390x__) +#if defined(__i386) || defined(__i386__) || \ + defined(__x86_64) || defined(__x86_64__) || \ + defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64) || \ + defined(__s390__) || defined(__s390x__) # undef STRICT_ALIGNMENT #endif #if !defined(PEDANTIC) && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) -#if defined(__GNUC__) && __GNUC__>=2 -# if defined(__x86_64) || defined(__x86_64__) -# define BSWAP8(x) ({ u64 ret=(x); \ - asm ("bswapq %0" \ - : "+r"(ret)); ret; }) -# define BSWAP4(x) ({ u32 ret=(x); \ - asm ("bswapl %0" \ - : "+r"(ret)); ret; }) -# elif (defined(__i386) || defined(__i386__)) && !defined(I386_ONLY) -# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ - asm ("bswapl %0; bswapl %1" \ - : "+r"(hi),"+r"(lo)); \ - (u64)hi<<32|lo; }) -# define BSWAP4(x) ({ u32 ret=(x); \ - asm ("bswapl %0" \ - : "+r"(ret)); ret; }) -# elif (defined(__arm__) || defined(__arm)) && !defined(STRICT_ALIGNMENT) -# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ - asm ("rev %0,%0; rev %1,%1" \ - : "+r"(hi),"+r"(lo)); \ - (u64)hi<<32|lo; }) -# define BSWAP4(x) ({ u32 ret; \ - asm ("rev %0,%1" \ - : "=r"(ret) : "r"((u32)(x))); \ - ret; }) -# endif -#elif defined(_MSC_VER) -# if _MSC_VER>=1300 -# pragma intrinsic(_byteswap_uint64,_byteswap_ulong) -# define BSWAP8(x) _byteswap_uint64((u64)(x)) -# define BSWAP4(x) _byteswap_ulong((u32)(x)) -# elif defined(_M_IX86) - __inline u32 _bswap4(u32 val) { - _asm mov eax,val - _asm bswap eax - } -# define BSWAP4(x) _bswap4(x) +# if defined(__GNUC__) && __GNUC__>=2 +# if defined(__x86_64) || defined(__x86_64__) +# define BSWAP8(x) ({ u64 ret=(x); \ + asm ("bswapq %0" \ + : "+r"(ret)); ret; }) +# define BSWAP4(x) ({ u32 ret=(x); \ + asm ("bswapl %0" \ + : "+r"(ret)); ret; }) +# elif (defined(__i386) || defined(__i386__)) && !defined(I386_ONLY) +# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ + asm ("bswapl %0; bswapl %1" \ + : "+r"(hi),"+r"(lo)); \ + (u64)hi<<32|lo; }) +# define BSWAP4(x) ({ u32 ret=(x); \ + asm ("bswapl %0" \ + : "+r"(ret)); ret; }) +# elif (defined(__arm__) || defined(__arm)) && !defined(STRICT_ALIGNMENT) +# define BSWAP8(x) ({ u32 lo=(u64)(x)>>32,hi=(x); \ + asm ("rev %0,%0; rev %1,%1" \ + : "+r"(hi),"+r"(lo)); \ + (u64)hi<<32|lo; }) +# define BSWAP4(x) ({ u32 ret; \ + asm ("rev %0,%1" \ + : "=r"(ret) : "r"((u32)(x))); \ + ret; }) +# endif +# elif defined(_MSC_VER) +# if _MSC_VER>=1300 +# pragma intrinsic(_byteswap_uint64,_byteswap_ulong) +# define BSWAP8(x) _byteswap_uint64((u64)(x)) +# define BSWAP4(x) _byteswap_ulong((u32)(x)) +# elif defined(_M_IX86) +__inline u32 _bswap4(u32 val) +{ +_asm mov eax, val _asm bswap eax} +# define BSWAP4(x) _bswap4(x) +# endif # endif #endif -#endif - #if defined(BSWAP4) && !defined(STRICT_ALIGNMENT) -#define GETU32(p) BSWAP4(*(const u32 *)(p)) -#define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) +# define GETU32(p) BSWAP4(*(const u32 *)(p)) +# define PUTU32(p,v) *(u32 *)(p) = BSWAP4(v) #else -#define GETU32(p) ((u32)(p)[0]<<24|(u32)(p)[1]<<16|(u32)(p)[2]<<8|(u32)(p)[3]) -#define PUTU32(p,v) ((p)[0]=(u8)((v)>>24),(p)[1]=(u8)((v)>>16),(p)[2]=(u8)((v)>>8),(p)[3]=(u8)(v)) +# define GETU32(p) ((u32)(p)[0]<<24|(u32)(p)[1]<<16|(u32)(p)[2]<<8|(u32)(p)[3]) +# define PUTU32(p,v) ((p)[0]=(u8)((v)>>24),(p)[1]=(u8)((v)>>16),(p)[2]=(u8)((v)>>8),(p)[3]=(u8)(v)) #endif +/*- GCM definitions */ typedef struct { + u64 hi, lo; +} u128; -/* GCM definitions */ - -typedef struct { u64 hi,lo; } u128; - -#ifdef TABLE_BITS -#undef TABLE_BITS +#ifdef TABLE_BITS +# undef TABLE_BITS #endif /* * Even though permitted values for TABLE_BITS are 8, 4 and 1, it should * never be set to 8 [or 1]. For further information see gcm128.c. */ -#define TABLE_BITS 4 +#define TABLE_BITS 4 struct gcm128_context { - /* Following 6 names follow names in GCM specification */ - union { u64 u[2]; u32 d[4]; u8 c[16]; size_t t[16/sizeof(size_t)]; } - Yi,EKi,EK0,len,Xi,H; - /* Relative position of Xi, H and pre-computed Htable is used - * in some assembler modules, i.e. don't change the order! */ + /* Following 6 names follow names in GCM specification */ + union { + u64 u[2]; + u32 d[4]; + u8 c[16]; + size_t t[16 / sizeof(size_t)]; + } Yi, EKi, EK0, len, Xi, H; + /* + * Relative position of Xi, H and pre-computed Htable is used in some + * assembler modules, i.e. don't change the order! + */ #if TABLE_BITS==8 - u128 Htable[256]; + u128 Htable[256]; #else - u128 Htable[16]; - void (*gmult)(u64 Xi[2],const u128 Htable[16]); - void (*ghash)(u64 Xi[2],const u128 Htable[16],const u8 *inp,size_t len); + u128 Htable[16]; + void (*gmult) (u64 Xi[2], const u128 Htable[16]); + void (*ghash) (u64 Xi[2], const u128 Htable[16], const u8 *inp, + size_t len); #endif - unsigned int mres, ares; - block128_f block; - void *key; + unsigned int mres, ares; + block128_f block; + void *key; }; struct xts128_context { - void *key1, *key2; - block128_f block1,block2; + void *key1, *key2; + block128_f block1, block2; }; struct ccm128_context { - union { u64 u[2]; u8 c[16]; } nonce, cmac; - u64 blocks; - block128_f block; - void *key; + union { + u64 u[2]; + u8 c[16]; + } nonce, cmac; + u64 blocks; + block128_f block; + void *key; }; - diff --git a/drivers/builtin_openssl2/crypto/modes/ofb128.c b/drivers/builtin_openssl2/crypto/modes/ofb128.c index 01c01702c4..4dbaccd7a6 100644 --- a/drivers/builtin_openssl2/crypto/modes/ofb128.c +++ b/drivers/builtin_openssl2/crypto/modes/ofb128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -59,63 +59,66 @@ #endif #include <assert.h> -/* The input and output encrypted as though 128bit ofb mode is being - * used. The extra state information to record how much of the - * 128bit block we have used is contained in *num; +/* + * The input and output encrypted as though 128bit ofb mode is being used. + * The extra state information to record how much of the 128bit block we have + * used is contained in *num; */ void CRYPTO_ofb128_encrypt(const unsigned char *in, unsigned char *out, - size_t len, const void *key, - unsigned char ivec[16], int *num, - block128_f block) + size_t len, const void *key, + unsigned char ivec[16], int *num, block128_f block) { - unsigned int n; - size_t l=0; + unsigned int n; + size_t l = 0; - assert(in && out && key && ivec && num); + assert(in && out && key && ivec && num); - n = *num; + n = *num; #if !defined(OPENSSL_SMALL_FOOTPRINT) - if (16%sizeof(size_t) == 0) do { /* always true actually */ - while (n && len) { - *(out++) = *(in++) ^ ivec[n]; - --len; - n = (n+1) % 16; - } -#if defined(STRICT_ALIGNMENT) - if (((size_t)in|(size_t)out|(size_t)ivec)%sizeof(size_t) != 0) - break; -#endif - while (len>=16) { - (*block)(ivec, ivec, key); - for (; n<16; n+=sizeof(size_t)) - *(size_t*)(out+n) = - *(size_t*)(in+n) ^ *(size_t*)(ivec+n); - len -= 16; - out += 16; - in += 16; - n = 0; - } - if (len) { - (*block)(ivec, ivec, key); - while (len--) { - out[n] = in[n] ^ ivec[n]; - ++n; - } - } - *num = n; - return; - } while(0); - /* the rest would be commonly eliminated by x86* compiler */ + if (16 % sizeof(size_t) == 0) { /* always true actually */ + do { + while (n && len) { + *(out++) = *(in++) ^ ivec[n]; + --len; + n = (n + 1) % 16; + } +# if defined(STRICT_ALIGNMENT) + if (((size_t)in | (size_t)out | (size_t)ivec) % sizeof(size_t) != + 0) + break; +# endif + while (len >= 16) { + (*block) (ivec, ivec, key); + for (; n < 16; n += sizeof(size_t)) + *(size_t *)(out + n) = + *(size_t *)(in + n) ^ *(size_t *)(ivec + n); + len -= 16; + out += 16; + in += 16; + n = 0; + } + if (len) { + (*block) (ivec, ivec, key); + while (len--) { + out[n] = in[n] ^ ivec[n]; + ++n; + } + } + *num = n; + return; + } while (0); + } + /* the rest would be commonly eliminated by x86* compiler */ #endif - while (l<len) { - if (n==0) { - (*block)(ivec, ivec, key); - } - out[l] = in[l] ^ ivec[n]; - ++l; - n = (n+1) % 16; - } + while (l < len) { + if (n == 0) { + (*block) (ivec, ivec, key); + } + out[l] = in[l] ^ ivec[n]; + ++l; + n = (n + 1) % 16; + } - *num=n; + *num = n; } diff --git a/drivers/builtin_openssl2/crypto/modes/xts128.c b/drivers/builtin_openssl2/crypto/modes/xts128.c index 9cf27a25e9..8f2af588b2 100644 --- a/drivers/builtin_openssl2/crypto/modes/xts128.c +++ b/drivers/builtin_openssl2/crypto/modes/xts128.c @@ -6,7 +6,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -58,130 +58,147 @@ #endif #include <assert.h> -int CRYPTO_xts128_encrypt(const XTS128_CONTEXT *ctx, const unsigned char iv[16], - const unsigned char *inp, unsigned char *out, - size_t len, int enc) +int CRYPTO_xts128_encrypt(const XTS128_CONTEXT *ctx, + const unsigned char iv[16], + const unsigned char *inp, unsigned char *out, + size_t len, int enc) { - const union { long one; char little; } is_endian = {1}; - union { u64 u[2]; u32 d[4]; u8 c[16]; } tweak, scratch; - unsigned int i; - - if (len<16) return -1; - - memcpy(tweak.c, iv, 16); - - (*ctx->block2)(tweak.c,tweak.c,ctx->key2); - - if (!enc && (len%16)) len-=16; - - while (len>=16) { + const union { + long one; + char little; + } is_endian = { + 1 + }; + union { + u64 u[2]; + u32 d[4]; + u8 c[16]; + } tweak, scratch; + unsigned int i; + + if (len < 16) + return -1; + + memcpy(tweak.c, iv, 16); + + (*ctx->block2) (tweak.c, tweak.c, ctx->key2); + + if (!enc && (len % 16)) + len -= 16; + + while (len >= 16) { #if defined(STRICT_ALIGNMENT) - memcpy(scratch.c,inp,16); - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; + memcpy(scratch.c, inp, 16); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; #else - scratch.u[0] = ((u64*)inp)[0]^tweak.u[0]; - scratch.u[1] = ((u64*)inp)[1]^tweak.u[1]; + scratch.u[0] = ((u64 *)inp)[0] ^ tweak.u[0]; + scratch.u[1] = ((u64 *)inp)[1] ^ tweak.u[1]; #endif - (*ctx->block1)(scratch.c,scratch.c,ctx->key1); + (*ctx->block1) (scratch.c, scratch.c, ctx->key1); #if defined(STRICT_ALIGNMENT) - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; - memcpy(out,scratch.c,16); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy(out, scratch.c, 16); #else - ((u64*)out)[0] = scratch.u[0]^=tweak.u[0]; - ((u64*)out)[1] = scratch.u[1]^=tweak.u[1]; + ((u64 *)out)[0] = scratch.u[0] ^= tweak.u[0]; + ((u64 *)out)[1] = scratch.u[1] ^= tweak.u[1]; #endif - inp += 16; - out += 16; - len -= 16; - - if (len==0) return 0; - - if (is_endian.little) { - unsigned int carry,res; - - res = 0x87&(((int)tweak.d[3])>>31); - carry = (unsigned int)(tweak.u[0]>>63); - tweak.u[0] = (tweak.u[0]<<1)^res; - tweak.u[1] = (tweak.u[1]<<1)|carry; - } - else { - size_t c; - - for (c=0,i=0;i<16;++i) { - /*+ substitutes for |, because c is 1 bit */ - c += ((size_t)tweak.c[i])<<1; - tweak.c[i] = (u8)c; - c = c>>8; - } - tweak.c[0] ^= (u8)(0x87&(0-c)); - } - } - if (enc) { - for (i=0;i<len;++i) { - u8 c = inp[i]; - out[i] = scratch.c[i]; - scratch.c[i] = c; - } - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; - (*ctx->block1)(scratch.c,scratch.c,ctx->key1); - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; - memcpy(out-16,scratch.c,16); - } - else { - union { u64 u[2]; u8 c[16]; } tweak1; - - if (is_endian.little) { - unsigned int carry,res; - - res = 0x87&(((int)tweak.d[3])>>31); - carry = (unsigned int)(tweak.u[0]>>63); - tweak1.u[0] = (tweak.u[0]<<1)^res; - tweak1.u[1] = (tweak.u[1]<<1)|carry; - } - else { - size_t c; - - for (c=0,i=0;i<16;++i) { - /*+ substitutes for |, because c is 1 bit */ - c += ((size_t)tweak.c[i])<<1; - tweak1.c[i] = (u8)c; - c = c>>8; - } - tweak1.c[0] ^= (u8)(0x87&(0-c)); - } + inp += 16; + out += 16; + len -= 16; + + if (len == 0) + return 0; + + if (is_endian.little) { + unsigned int carry, res; + + res = 0x87 & (((int)tweak.d[3]) >> 31); + carry = (unsigned int)(tweak.u[0] >> 63); + tweak.u[0] = (tweak.u[0] << 1) ^ res; + tweak.u[1] = (tweak.u[1] << 1) | carry; + } else { + size_t c; + + for (c = 0, i = 0; i < 16; ++i) { + /* + * + substitutes for |, because c is 1 bit + */ + c += ((size_t)tweak.c[i]) << 1; + tweak.c[i] = (u8)c; + c = c >> 8; + } + tweak.c[0] ^= (u8)(0x87 & (0 - c)); + } + } + if (enc) { + for (i = 0; i < len; ++i) { + u8 c = inp[i]; + out[i] = scratch.c[i]; + scratch.c[i] = c; + } + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + (*ctx->block1) (scratch.c, scratch.c, ctx->key1); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy(out - 16, scratch.c, 16); + } else { + union { + u64 u[2]; + u8 c[16]; + } tweak1; + + if (is_endian.little) { + unsigned int carry, res; + + res = 0x87 & (((int)tweak.d[3]) >> 31); + carry = (unsigned int)(tweak.u[0] >> 63); + tweak1.u[0] = (tweak.u[0] << 1) ^ res; + tweak1.u[1] = (tweak.u[1] << 1) | carry; + } else { + size_t c; + + for (c = 0, i = 0; i < 16; ++i) { + /* + * + substitutes for |, because c is 1 bit + */ + c += ((size_t)tweak.c[i]) << 1; + tweak1.c[i] = (u8)c; + c = c >> 8; + } + tweak1.c[0] ^= (u8)(0x87 & (0 - c)); + } #if defined(STRICT_ALIGNMENT) - memcpy(scratch.c,inp,16); - scratch.u[0] ^= tweak1.u[0]; - scratch.u[1] ^= tweak1.u[1]; + memcpy(scratch.c, inp, 16); + scratch.u[0] ^= tweak1.u[0]; + scratch.u[1] ^= tweak1.u[1]; #else - scratch.u[0] = ((u64*)inp)[0]^tweak1.u[0]; - scratch.u[1] = ((u64*)inp)[1]^tweak1.u[1]; + scratch.u[0] = ((u64 *)inp)[0] ^ tweak1.u[0]; + scratch.u[1] = ((u64 *)inp)[1] ^ tweak1.u[1]; #endif - (*ctx->block1)(scratch.c,scratch.c,ctx->key1); - scratch.u[0] ^= tweak1.u[0]; - scratch.u[1] ^= tweak1.u[1]; - - for (i=0;i<len;++i) { - u8 c = inp[16+i]; - out[16+i] = scratch.c[i]; - scratch.c[i] = c; - } - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; - (*ctx->block1)(scratch.c,scratch.c,ctx->key1); + (*ctx->block1) (scratch.c, scratch.c, ctx->key1); + scratch.u[0] ^= tweak1.u[0]; + scratch.u[1] ^= tweak1.u[1]; + + for (i = 0; i < len; ++i) { + u8 c = inp[16 + i]; + out[16 + i] = scratch.c[i]; + scratch.c[i] = c; + } + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + (*ctx->block1) (scratch.c, scratch.c, ctx->key1); #if defined(STRICT_ALIGNMENT) - scratch.u[0] ^= tweak.u[0]; - scratch.u[1] ^= tweak.u[1]; - memcpy (out,scratch.c,16); + scratch.u[0] ^= tweak.u[0]; + scratch.u[1] ^= tweak.u[1]; + memcpy(out, scratch.c, 16); #else - ((u64*)out)[0] = scratch.u[0]^tweak.u[0]; - ((u64*)out)[1] = scratch.u[1]^tweak.u[1]; + ((u64 *)out)[0] = scratch.u[0] ^ tweak.u[0]; + ((u64 *)out)[1] = scratch.u[1] ^ tweak.u[1]; #endif - } + } - return 0; + return 0; } |