diff options
Diffstat (limited to 'thirdparty/pcre2/src/pcre2_jit_compile.c')
| -rw-r--r-- | thirdparty/pcre2/src/pcre2_jit_compile.c | 928 |
1 files changed, 589 insertions, 339 deletions
diff --git a/thirdparty/pcre2/src/pcre2_jit_compile.c b/thirdparty/pcre2/src/pcre2_jit_compile.c index db2ce65598..d726c3ca04 100644 --- a/thirdparty/pcre2/src/pcre2_jit_compile.c +++ b/thirdparty/pcre2/src/pcre2_jit_compile.c @@ -8,7 +8,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel This module by Zoltan Herczeg Original API code Copyright (c) 1997-2012 University of Cambridge - New API code Copyright (c) 2016-2019 University of Cambridge + New API code Copyright (c) 2016-2021 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -413,6 +413,9 @@ typedef struct compiler_common { /* Locals used by fast fail optimization. */ sljit_s32 early_fail_start_ptr; sljit_s32 early_fail_end_ptr; + /* Variables used by recursive call generator. */ + sljit_s32 recurse_bitset_size; + uint8_t *recurse_bitset; /* Flipped and lower case tables. */ const sljit_u8 *fcc; @@ -613,6 +616,8 @@ the start pointers when the end of the capturing group has not yet reached. */ sljit_emit_op1(compiler, (op), (dst), (dstw), (src), (srcw)) #define OP2(op, dst, dstw, src1, src1w, src2, src2w) \ sljit_emit_op2(compiler, (op), (dst), (dstw), (src1), (src1w), (src2), (src2w)) +#define OP2U(op, src1, src1w, src2, src2w) \ + sljit_emit_op2u(compiler, (op), (src1), (src1w), (src2), (src2w)) #define OP_SRC(op, src, srcw) \ sljit_emit_op_src(compiler, (op), (src), (srcw)) #define LABEL() \ @@ -1621,7 +1626,7 @@ if (end[-(1 + LINK_SIZE)] != OP_KET || PRIVATE_DATA(begin) != 0) /* /(?:AB){4,6}/ is currently converted to /(?:AB){3}(?AB){1,3}/ * Skip the check of the second part. */ -if (PRIVATE_DATA(end - LINK_SIZE) == 0) +if (PRIVATE_DATA(end - LINK_SIZE) != 0) return TRUE; next = end; @@ -2315,22 +2320,47 @@ for (i = 0; i < RECURSE_TMP_REG_COUNT; i++) #undef RECURSE_TMP_REG_COUNT -static int get_recurse_data_length(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, - BOOL *needs_control_head, BOOL *has_quit, BOOL *has_accept) +static BOOL recurse_check_bit(compiler_common *common, sljit_sw bit_index) +{ +uint8_t *byte; +uint8_t mask; + +SLJIT_ASSERT((bit_index & (sizeof(sljit_sw) - 1)) == 0); + +bit_index >>= SLJIT_WORD_SHIFT; + +SLJIT_ASSERT((bit_index >> 3) < common->recurse_bitset_size); + +mask = 1 << (bit_index & 0x7); +byte = common->recurse_bitset + (bit_index >> 3); + +if (*byte & mask) + return FALSE; + +*byte |= mask; +return TRUE; +} + +enum get_recurse_flags { + recurse_flag_quit_found = (1 << 0), + recurse_flag_accept_found = (1 << 1), + recurse_flag_setsom_found = (1 << 2), + recurse_flag_setmark_found = (1 << 3), + recurse_flag_control_head_found = (1 << 4), +}; + +static int get_recurse_data_length(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, uint32_t *result_flags) { int length = 1; -int size; +int size, offset; PCRE2_SPTR alternative; -BOOL quit_found = FALSE; -BOOL accept_found = FALSE; -BOOL setsom_found = FALSE; -BOOL setmark_found = FALSE; -BOOL capture_last_found = FALSE; -BOOL control_head_found = FALSE; +uint32_t recurse_flags = 0; + +memset(common->recurse_bitset, 0, common->recurse_bitset_size); #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD SLJIT_ASSERT(common->control_head_ptr != 0); -control_head_found = TRUE; +recurse_flags |= recurse_flag_control_head_found; #endif /* Calculate the sum of the private machine words. */ @@ -2341,24 +2371,26 @@ while (cc < ccend) { case OP_SET_SOM: SLJIT_ASSERT(common->has_set_som); - setsom_found = TRUE; + recurse_flags |= recurse_flag_setsom_found; cc += 1; break; case OP_RECURSE: if (common->has_set_som) - setsom_found = TRUE; + recurse_flags |= recurse_flag_setsom_found; if (common->mark_ptr != 0) - setmark_found = TRUE; - if (common->capture_last_ptr != 0) - capture_last_found = TRUE; + recurse_flags |= recurse_flag_setmark_found; + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) + length++; cc += 1 + LINK_SIZE; break; case OP_KET: - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0) { - length++; + if (recurse_check_bit(common, offset)) + length++; SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0); cc += PRIVATE_DATA(cc + 1); } @@ -2377,39 +2409,55 @@ while (cc < ccend) case OP_SBRA: case OP_SBRAPOS: case OP_SCOND: - length++; SLJIT_ASSERT(PRIVATE_DATA(cc) != 0); + if (recurse_check_bit(common, PRIVATE_DATA(cc))) + length++; cc += 1 + LINK_SIZE; break; case OP_CBRA: case OP_SCBRA: - length += 2; - if (common->capture_last_ptr != 0) - capture_last_found = TRUE; - if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) + offset = GET2(cc, 1 + LINK_SIZE); + if (recurse_check_bit(common, OVECTOR(offset << 1))) + { + SLJIT_ASSERT(recurse_check_bit(common, OVECTOR((offset << 1) + 1))); + length += 2; + } + if (common->optimized_cbracket[offset] == 0 && recurse_check_bit(common, OVECTOR_PRIV(offset))) + length++; + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) length++; cc += 1 + LINK_SIZE + IMM2_SIZE; break; case OP_CBRAPOS: case OP_SCBRAPOS: - length += 2 + 2; - if (common->capture_last_ptr != 0) - capture_last_found = TRUE; + offset = GET2(cc, 1 + LINK_SIZE); + if (recurse_check_bit(common, OVECTOR(offset << 1))) + { + SLJIT_ASSERT(recurse_check_bit(common, OVECTOR((offset << 1) + 1))); + length += 2; + } + if (recurse_check_bit(common, OVECTOR_PRIV(offset))) + length++; + if (recurse_check_bit(common, PRIVATE_DATA(cc))) + length++; + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) + length++; cc += 1 + LINK_SIZE + IMM2_SIZE; break; case OP_COND: /* Might be a hidden SCOND. */ alternative = cc + GET(cc, 1); - if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) + if ((*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) && recurse_check_bit(common, PRIVATE_DATA(cc))) length++; cc += 1 + LINK_SIZE; break; CASE_ITERATOR_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) length++; cc += 2; #ifdef SUPPORT_UNICODE @@ -2418,8 +2466,12 @@ while (cc < ccend) break; CASE_ITERATOR_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) + { + SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw))); length += 2; + } cc += 2; #ifdef SUPPORT_UNICODE if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); @@ -2427,8 +2479,12 @@ while (cc < ccend) break; CASE_ITERATOR_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) + { + SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw))); length += 2; + } cc += 2 + IMM2_SIZE; #ifdef SUPPORT_UNICODE if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); @@ -2436,20 +2492,29 @@ while (cc < ccend) break; CASE_ITERATOR_TYPE_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) length++; cc += 1; break; CASE_ITERATOR_TYPE_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) + { + SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw))); length += 2; + } cc += 1; break; CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc) != 0) + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) + { + SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw))); length += 2; + } cc += 1 + IMM2_SIZE; break; @@ -2461,7 +2526,9 @@ while (cc < ccend) #else size = 1 + 32 / (int)sizeof(PCRE2_UCHAR); #endif - if (PRIVATE_DATA(cc) != 0) + + offset = PRIVATE_DATA(cc); + if (offset != 0 && recurse_check_bit(common, offset)) length += get_class_iterator_size(cc + size); cc += size; break; @@ -2471,12 +2538,11 @@ while (cc < ccend) case OP_PRUNE_ARG: case OP_THEN_ARG: SLJIT_ASSERT(common->mark_ptr != 0); - if (!setmark_found) - setmark_found = TRUE; + recurse_flags |= recurse_flag_setmark_found; if (common->control_head_ptr != 0) - control_head_found = TRUE; + recurse_flags |= recurse_flag_control_head_found; if (*cc != OP_MARK) - quit_found = TRUE; + recurse_flags |= recurse_flag_quit_found; cc += 1 + 2 + cc[1]; break; @@ -2484,26 +2550,24 @@ while (cc < ccend) case OP_PRUNE: case OP_SKIP: case OP_COMMIT: - quit_found = TRUE; + recurse_flags |= recurse_flag_quit_found; cc++; break; case OP_SKIP_ARG: - quit_found = TRUE; + recurse_flags |= recurse_flag_quit_found; cc += 1 + 2 + cc[1]; break; case OP_THEN: SLJIT_ASSERT(common->control_head_ptr != 0); - quit_found = TRUE; - if (!control_head_found) - control_head_found = TRUE; + recurse_flags |= recurse_flag_quit_found | recurse_flag_control_head_found; cc++; break; case OP_ACCEPT: case OP_ASSERT_ACCEPT: - accept_found = TRUE; + recurse_flags |= recurse_flag_accept_found; cc++; break; @@ -2515,21 +2579,17 @@ while (cc < ccend) } SLJIT_ASSERT(cc == ccend); -if (control_head_found) +if (recurse_flags & recurse_flag_control_head_found) length++; -if (capture_last_found) - length++; -if (quit_found) +if (recurse_flags & recurse_flag_quit_found) { - if (setsom_found) + if (recurse_flags & recurse_flag_setsom_found) length++; - if (setmark_found) + if (recurse_flags & recurse_flag_setmark_found) length++; } -*needs_control_head = control_head_found; -*has_quit = quit_found; -*has_accept = accept_found; +*result_flags = recurse_flags; return length; } @@ -2542,7 +2602,7 @@ enum copy_recurse_data_types { }; static void copy_recurse_data(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, - int type, int stackptr, int stacktop, BOOL has_quit) + int type, int stackptr, int stacktop, uint32_t recurse_flags) { delayed_mem_copy_status status; PCRE2_SPTR alternative; @@ -2551,14 +2611,12 @@ sljit_sw shared_srcw[3]; sljit_sw kept_shared_srcw[2]; int private_count, shared_count, kept_shared_count; int from_sp, base_reg, offset, i; -BOOL setsom_found = FALSE; -BOOL setmark_found = FALSE; -BOOL capture_last_found = FALSE; -BOOL control_head_found = FALSE; + +memset(common->recurse_bitset, 0, common->recurse_bitset_size); #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD SLJIT_ASSERT(common->control_head_ptr != 0); -control_head_found = TRUE; +recurse_check_bit(common, common->control_head_ptr); #endif switch (type) @@ -2646,45 +2704,42 @@ while (cc < ccend) { case OP_SET_SOM: SLJIT_ASSERT(common->has_set_som); - if (has_quit && !setsom_found) + if ((recurse_flags & recurse_flag_quit_found) && recurse_check_bit(common, OVECTOR(0))) { kept_shared_srcw[0] = OVECTOR(0); kept_shared_count = 1; - setsom_found = TRUE; } cc += 1; break; case OP_RECURSE: - if (has_quit) + if (recurse_flags & recurse_flag_quit_found) { - if (common->has_set_som && !setsom_found) + if (common->has_set_som && recurse_check_bit(common, OVECTOR(0))) { kept_shared_srcw[0] = OVECTOR(0); kept_shared_count = 1; - setsom_found = TRUE; } - if (common->mark_ptr != 0 && !setmark_found) + if (common->mark_ptr != 0 && recurse_check_bit(common, common->mark_ptr)) { kept_shared_srcw[kept_shared_count] = common->mark_ptr; kept_shared_count++; - setmark_found = TRUE; } } - if (common->capture_last_ptr != 0 && !capture_last_found) + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) { shared_srcw[0] = common->capture_last_ptr; shared_count = 1; - capture_last_found = TRUE; } cc += 1 + LINK_SIZE; break; case OP_KET: - if (PRIVATE_DATA(cc) != 0) + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0) { - private_count = 1; - private_srcw[0] = PRIVATE_DATA(cc); + if (recurse_check_bit(common, private_srcw[0])) + private_count = 1; SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0); cc += PRIVATE_DATA(cc + 1); } @@ -2703,50 +2758,66 @@ while (cc < ccend) case OP_SBRA: case OP_SBRAPOS: case OP_SCOND: - private_count = 1; private_srcw[0] = PRIVATE_DATA(cc); + if (recurse_check_bit(common, private_srcw[0])) + private_count = 1; cc += 1 + LINK_SIZE; break; case OP_CBRA: case OP_SCBRA: - offset = (GET2(cc, 1 + LINK_SIZE)) << 1; - shared_srcw[0] = OVECTOR(offset); - shared_srcw[1] = OVECTOR(offset + 1); - shared_count = 2; + offset = GET2(cc, 1 + LINK_SIZE); + shared_srcw[0] = OVECTOR(offset << 1); + if (recurse_check_bit(common, shared_srcw[0])) + { + shared_srcw[1] = shared_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, shared_srcw[1])); + shared_count = 2; + } - if (common->capture_last_ptr != 0 && !capture_last_found) + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) { - shared_srcw[2] = common->capture_last_ptr; - shared_count = 3; - capture_last_found = TRUE; + shared_srcw[shared_count] = common->capture_last_ptr; + shared_count++; } - if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) + if (common->optimized_cbracket[offset] == 0) { - private_count = 1; - private_srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); + private_srcw[0] = OVECTOR_PRIV(offset); + if (recurse_check_bit(common, private_srcw[0])) + private_count = 1; } + cc += 1 + LINK_SIZE + IMM2_SIZE; break; case OP_CBRAPOS: case OP_SCBRAPOS: - offset = (GET2(cc, 1 + LINK_SIZE)) << 1; - shared_srcw[0] = OVECTOR(offset); - shared_srcw[1] = OVECTOR(offset + 1); - shared_count = 2; + offset = GET2(cc, 1 + LINK_SIZE); + shared_srcw[0] = OVECTOR(offset << 1); + if (recurse_check_bit(common, shared_srcw[0])) + { + shared_srcw[1] = shared_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, shared_srcw[1])); + shared_count = 2; + } - if (common->capture_last_ptr != 0 && !capture_last_found) + if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr)) { - shared_srcw[2] = common->capture_last_ptr; - shared_count = 3; - capture_last_found = TRUE; + shared_srcw[shared_count] = common->capture_last_ptr; + shared_count++; } - private_count = 2; private_srcw[0] = PRIVATE_DATA(cc); - private_srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); + if (recurse_check_bit(common, private_srcw[0])) + private_count = 1; + + offset = OVECTOR_PRIV(offset); + if (recurse_check_bit(common, offset)) + { + private_srcw[private_count] = offset; + private_count++; + } cc += 1 + LINK_SIZE + IMM2_SIZE; break; @@ -2755,18 +2826,17 @@ while (cc < ccend) alternative = cc + GET(cc, 1); if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) { - private_count = 1; private_srcw[0] = PRIVATE_DATA(cc); + if (recurse_check_bit(common, private_srcw[0])) + private_count = 1; } cc += 1 + LINK_SIZE; break; CASE_ITERATOR_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - { + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) private_count = 1; - private_srcw[0] = PRIVATE_DATA(cc); - } cc += 2; #ifdef SUPPORT_UNICODE if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); @@ -2774,11 +2844,12 @@ while (cc < ccend) break; CASE_ITERATOR_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) { private_count = 2; - private_srcw[0] = PRIVATE_DATA(cc); - private_srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); + private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1])); } cc += 2; #ifdef SUPPORT_UNICODE @@ -2787,11 +2858,12 @@ while (cc < ccend) break; CASE_ITERATOR_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) { private_count = 2; - private_srcw[0] = PRIVATE_DATA(cc); - private_srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); + private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1])); } cc += 2 + IMM2_SIZE; #ifdef SUPPORT_UNICODE @@ -2800,30 +2872,30 @@ while (cc < ccend) break; CASE_ITERATOR_TYPE_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - { + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) private_count = 1; - private_srcw[0] = PRIVATE_DATA(cc); - } cc += 1; break; CASE_ITERATOR_TYPE_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) { private_count = 2; - private_srcw[0] = PRIVATE_DATA(cc); private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1])); } cc += 1; break; CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) + private_srcw[0] = PRIVATE_DATA(cc); + if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0])) { private_count = 2; - private_srcw[0] = PRIVATE_DATA(cc); private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1])); } cc += 1 + IMM2_SIZE; break; @@ -2837,23 +2909,28 @@ while (cc < ccend) i = 1 + 32 / (int)sizeof(PCRE2_UCHAR); #endif if (PRIVATE_DATA(cc) != 0) + { + private_count = 1; + private_srcw[0] = PRIVATE_DATA(cc); switch(get_class_iterator_size(cc + i)) { case 1: - private_count = 1; - private_srcw[0] = PRIVATE_DATA(cc); break; case 2: - private_count = 2; - private_srcw[0] = PRIVATE_DATA(cc); - private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + if (recurse_check_bit(common, private_srcw[0])) + { + private_count = 2; + private_srcw[1] = private_srcw[0] + sizeof(sljit_sw); + SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1])); + } break; default: SLJIT_UNREACHABLE(); break; } + } cc += i; break; @@ -2862,28 +2939,25 @@ while (cc < ccend) case OP_PRUNE_ARG: case OP_THEN_ARG: SLJIT_ASSERT(common->mark_ptr != 0); - if (has_quit && !setmark_found) + if ((recurse_flags & recurse_flag_quit_found) && recurse_check_bit(common, common->mark_ptr)) { kept_shared_srcw[0] = common->mark_ptr; kept_shared_count = 1; - setmark_found = TRUE; } - if (common->control_head_ptr != 0 && !control_head_found) + if (common->control_head_ptr != 0 && recurse_check_bit(common, common->control_head_ptr)) { private_srcw[0] = common->control_head_ptr; private_count = 1; - control_head_found = TRUE; } cc += 1 + 2 + cc[1]; break; case OP_THEN: SLJIT_ASSERT(common->control_head_ptr != 0); - if (!control_head_found) + if (recurse_check_bit(common, common->control_head_ptr)) { private_srcw[0] = common->control_head_ptr; private_count = 1; - control_head_found = TRUE; } cc++; break; @@ -2891,7 +2965,7 @@ while (cc < ccend) default: cc = next_opcode(common, cc); SLJIT_ASSERT(cc != NULL); - break; + continue; } if (type != recurse_copy_shared_to_global && type != recurse_copy_kept_shared_to_global) @@ -3743,9 +3817,9 @@ if (common->invalid_utf) else { OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x110000); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); } } @@ -3982,7 +4056,7 @@ if (common->utf) { if (options & READ_CHAR_UPDATE_STR_PTR) OP2(SLJIT_ADD, RETURN_ADDR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x400); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400); if (options & READ_CHAR_UPDATE_STR_PTR) CMOV(SLJIT_LESS, STR_PTR, RETURN_ADDR, 0); if (max >= 0xd800) @@ -4010,9 +4084,9 @@ if (common->invalid_utf) else { OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x110000); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); } } @@ -4167,7 +4241,7 @@ if (common->utf && negated) if (sljit_has_cpu_feature(SLJIT_HAS_CMOV) && !HAS_VIRTUAL_REGISTERS) { OP2(SLJIT_ADD, RETURN_ADDR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x400); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400); CMOV(SLJIT_LESS, STR_PTR, RETURN_ADDR, 0); } else @@ -4250,7 +4324,7 @@ if (common->utf) /* Skip low surrogate if necessary. */ OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xdc00); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL); OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); @@ -4267,7 +4341,7 @@ if (common->invalid_utf && !must_be_valid) return; } - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x110000); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x110000); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_LESS); OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); @@ -4332,7 +4406,7 @@ OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); /* Searching for the first zero. */ -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800); jump = JUMP(SLJIT_NOT_ZERO); /* Two byte sequence. */ OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3000); @@ -4345,7 +4419,7 @@ OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x10000); jump = JUMP(SLJIT_NOT_ZERO); /* Three byte sequence. */ OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0000); @@ -4373,7 +4447,7 @@ struct sljit_jump *compare; sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, 0x20); jump = JUMP(SLJIT_NOT_ZERO); /* Two byte sequence. */ OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); @@ -4432,7 +4506,7 @@ OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80); exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800); jump = JUMP(SLJIT_NOT_ZERO); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); @@ -4447,14 +4521,14 @@ OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x40); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, 0x20000); exit_invalid[2] = NULL; } else exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x10000); jump = JUMP(SLJIT_NOT_ZERO); three_byte_entry = LABEL(); @@ -4462,7 +4536,7 @@ three_byte_entry = LABEL(); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2d800); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0xd800); exit_invalid[3] = NULL; } @@ -4473,7 +4547,7 @@ OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); exit_invalid[4] = NULL; } @@ -4490,7 +4564,7 @@ OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x40); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, 0); exit_invalid[5] = NULL; } @@ -4500,7 +4574,7 @@ else OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc10000); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x100000); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0x10000); exit_invalid[6] = NULL; } @@ -4522,7 +4596,7 @@ OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80); exit_invalid[8] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800); jump = JUMP(SLJIT_NOT_ZERO); OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0); @@ -4537,7 +4611,7 @@ OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x40); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); exit_invalid[10] = NULL; } @@ -4830,7 +4904,7 @@ OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, -0xd800); exit_invalid[2] = NULL; } @@ -4840,7 +4914,7 @@ else OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800); CMOV(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR); exit_invalid[3] = NULL; } @@ -4865,7 +4939,7 @@ OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); if (has_cmov) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x100000); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000); CMOV(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0x10000); exit_invalid[5] = NULL; } @@ -4968,7 +5042,7 @@ OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xdc00); OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xdc00); -OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x400); +OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS); OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000); OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); @@ -5239,7 +5313,7 @@ if (newlinecheck) OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, common->newline & 0xff); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL); #if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); @@ -5304,12 +5378,12 @@ else if (common->utf) if (sljit_has_cpu_feature(SLJIT_HAS_CMOV)) { OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x400); CMOV(SLJIT_LESS, STR_PTR, TMP2, 0); } else { - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x400); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_LESS); OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); @@ -5860,7 +5934,7 @@ if (has_match_end) OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(offset + 1)); - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_END, 0, TMP1, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0); CMOV(SLJIT_GREATER, STR_END, TMP1, 0); } @@ -6063,7 +6137,7 @@ if (common->match_end_ptr != 0) OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); OP2(SLJIT_SUB | SLJIT_SET_LESS, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); add_jump(compiler, &common->failed_match, JUMP(SLJIT_LESS)); - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_END, 0, TMP1, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0); CMOV(SLJIT_GREATER, STR_END, TMP1, 0); } else @@ -6200,7 +6274,7 @@ if (common->nltype == NLTYPE_FIXED && common->newline > 255) firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); + OP2U(SLJIT_SUB | SLJIT_SET_Z, STR_PTR, 0, TMP1, 0); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_NOT_EQUAL); #if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); @@ -6228,7 +6302,7 @@ if (common->nltype == NLTYPE_FIXED && common->newline > 255) firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, STR_PTR, 0, TMP1, 0); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER_EQUAL); #if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); @@ -6293,7 +6367,7 @@ if (JIT_HAS_FAST_FORWARD_CHAR_SIMD && (common->nltype == NLTYPE_FIXED || common- OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); if (common->mode != PCRE2_JIT_COMPLETE) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_PTR, 0, STR_END, 0); CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0); } } @@ -6319,7 +6393,7 @@ if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) notfoundnl = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, CHAR_NL); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL); #if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); @@ -6355,7 +6429,7 @@ if (common->match_end_ptr != 0) OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr); OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_END, 0, TMP1, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0); CMOV(SLJIT_GREATER, STR_END, TMP1, 0); } @@ -6385,12 +6459,12 @@ if (!optimize_class(common, start_bits, (start_bits[31] & 0x80) != 0, FALSE, &ma if (!HAS_VIRTUAL_REGISTERS) { OP2(SLJIT_SHL, TMP3, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, TMP3, 0); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP3, 0); } else { OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP2, 0); } JUMPTO(SLJIT_ZERO, start); } @@ -6525,7 +6599,7 @@ jump = CMP(SLJIT_NOT_ZERO /* SIG_LESS */, TMP2, 0, SLJIT_IMM, 0); OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0); JUMPHERE(jump); -OP1(SLJIT_NEG, TMP2, 0, TMP2, 0); +OP2(SLJIT_SUB, TMP2, 0, SLJIT_IMM, 0, TMP2, 0); OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0); if (HAS_VIRTUAL_REGISTERS) { @@ -6600,10 +6674,10 @@ if (common->ucp) jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); add_jump(compiler, &common->getucdtype, JUMP(SLJIT_FAST_CALL)); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); JUMPHERE(jump); OP1(SLJIT_MOV, TMP3, 0, TMP2, 0); @@ -6646,10 +6720,10 @@ if (common->ucp) jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); add_jump(compiler, &common->getucdtype, JUMP(SLJIT_FAST_CALL)); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); JUMPHERE(jump); } @@ -6916,7 +6990,7 @@ j = 0; if (char_list[0] == 0) { i++; - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_ZERO); } else @@ -6928,7 +7002,7 @@ while (i < len) j++; else { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_list[i]); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, char_list[i]); CMOV(SLJIT_ZERO, TMP2, TMP1, 0); } i++; @@ -6942,7 +7016,7 @@ if (j != 0) if ((char_list[i] & 0x100) != 0) { j--; - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_list[i] & 0xff); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, char_list[i] & 0xff); CMOV(SLJIT_ZERO, TMP2, TMP1, 0); } } @@ -6971,9 +7045,9 @@ DEFINE_COMPILER; sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); -OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); +OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); +OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); #if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 #if PCRE2_CODE_UNIT_WIDTH == 8 if (common->utf) @@ -6981,7 +7055,7 @@ if (common->utf) #endif OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); #if PCRE2_CODE_UNIT_WIDTH == 8 } #endif @@ -6997,29 +7071,29 @@ DEFINE_COMPILER; sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09); +OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x09); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); +OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x20); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); +OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xa0); #if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 #if PCRE2_CODE_UNIT_WIDTH == 8 if (common->utf) { #endif OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x1680); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x180e); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); #if PCRE2_CODE_UNIT_WIDTH == 8 } #endif @@ -7037,9 +7111,9 @@ DEFINE_COMPILER; sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); -OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); +OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); +OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); #if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32 #if PCRE2_CODE_UNIT_WIDTH == 8 if (common->utf) @@ -7047,7 +7121,7 @@ if (common->utf) #endif OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); #if PCRE2_CODE_UNIT_WIDTH == 8 } #endif @@ -7412,6 +7486,21 @@ return cc; static PCRE2_SPTR compile_char1_matchingpath(compiler_common *common, PCRE2_UCHAR type, PCRE2_SPTR cc, jump_list **backtracks, BOOL check_str_ptr); +#ifdef SUPPORT_UNICODE +#define XCLASS_SAVE_CHAR 0x001 +#define XCLASS_CHAR_SAVED 0x002 +#define XCLASS_HAS_TYPE 0x004 +#define XCLASS_HAS_SCRIPT 0x008 +#define XCLASS_HAS_SCRIPT_EXTENSION 0x010 +#define XCLASS_HAS_BOOL 0x020 +#define XCLASS_HAS_BIDICL 0x040 +#define XCLASS_NEEDS_UCD (XCLASS_HAS_TYPE | XCLASS_HAS_SCRIPT | XCLASS_HAS_SCRIPT_EXTENSION | XCLASS_HAS_BOOL | XCLASS_HAS_BIDICL) +#define XCLASS_SCRIPT_EXTENSION_NOTPROP 0x080 +#define XCLASS_SCRIPT_EXTENSION_RESTORE_RETURN_ADDR 0x100 +#define XCLASS_SCRIPT_EXTENSION_RESTORE_LOCALS0 0x200 + +#endif /* SUPPORT_UNICODE */ + static void compile_xclass_matchingpath(compiler_common *common, PCRE2_SPTR cc, jump_list **backtracks) { DEFINE_COMPILER; @@ -7426,8 +7515,7 @@ BOOL utf = common->utf; #endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == [8|16] */ #ifdef SUPPORT_UNICODE -BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; -BOOL charsaved = FALSE; +sljit_u32 unicode_status = 0; int typereg = TMP1; const sljit_u32 *other_cases; sljit_uw typeoffset; @@ -7454,7 +7542,7 @@ while (*cc != XCL_END) if (c > max) max = c; if (c < min) min = c; #ifdef SUPPORT_UNICODE - needschar = TRUE; + unicode_status |= XCLASS_SAVE_CHAR; #endif /* SUPPORT_UNICODE */ } else if (*cc == XCL_RANGE) @@ -7465,7 +7553,7 @@ while (*cc != XCL_END) GETCHARINCTEST(c, cc); if (c > max) max = c; #ifdef SUPPORT_UNICODE - needschar = TRUE; + unicode_status |= XCLASS_SAVE_CHAR; #endif /* SUPPORT_UNICODE */ } #ifdef SUPPORT_UNICODE @@ -7473,7 +7561,7 @@ while (*cc != XCL_END) { SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); cc++; - if (*cc == PT_CLIST) + if (*cc == PT_CLIST && cc[-1] == XCL_PROP) { other_cases = PRIV(ucd_caseless_sets) + cc[1]; while (*other_cases != NOTACHAR) @@ -7506,11 +7594,21 @@ while (*cc != XCL_END) case PT_GC: case PT_PC: case PT_ALNUM: - needstype = TRUE; + unicode_status |= XCLASS_HAS_TYPE; break; + case PT_SCX: + unicode_status |= XCLASS_HAS_SCRIPT_EXTENSION; + if (cc[-1] == XCL_NOTPROP) + { + unicode_status |= XCLASS_SCRIPT_EXTENSION_NOTPROP; + break; + } + compares++; + /* Fall through */ + case PT_SC: - needsscript = TRUE; + unicode_status |= XCLASS_HAS_SCRIPT; break; case PT_SPACE: @@ -7519,13 +7617,20 @@ while (*cc != XCL_END) case PT_PXGRAPH: case PT_PXPRINT: case PT_PXPUNCT: - needstype = TRUE; - needschar = TRUE; + unicode_status |= XCLASS_SAVE_CHAR | XCLASS_HAS_TYPE; break; case PT_CLIST: case PT_UCNC: - needschar = TRUE; + unicode_status |= XCLASS_SAVE_CHAR; + break; + + case PT_BOOL: + unicode_status |= XCLASS_HAS_BOOL; + break; + + case PT_BIDICL: + unicode_status |= XCLASS_HAS_BIDICL; break; default: @@ -7545,7 +7650,7 @@ if ((cc[-1] & XCL_NOT) != 0) else { #ifdef SUPPORT_UNICODE - read_char(common, min, max, (needstype || needsscript) ? backtracks : NULL, 0); + read_char(common, min, max, (unicode_status & XCLASS_NEEDS_UCD) ? backtracks : NULL, 0); #else /* !SUPPORT_UNICODE */ read_char(common, min, max, NULL, 0); #endif /* SUPPORT_UNICODE */ @@ -7562,7 +7667,7 @@ if ((cc[-1] & XCL_HASPROP) == 0) OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP2, 0); add_jump(compiler, &found, JUMP(SLJIT_NOT_ZERO)); } @@ -7581,7 +7686,7 @@ else if ((cc[-1] & XCL_MAP) != 0) { OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); #ifdef SUPPORT_UNICODE - charsaved = TRUE; + unicode_status |= XCLASS_CHAR_SAVED; #endif /* SUPPORT_UNICODE */ if (!optimize_class(common, (const sljit_u8 *)cc, FALSE, TRUE, list)) { @@ -7595,7 +7700,7 @@ else if ((cc[-1] & XCL_MAP) != 0) OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP2, 0); add_jump(compiler, list, JUMP(SLJIT_NOT_ZERO)); #if PCRE2_CODE_UNIT_WIDTH == 8 @@ -7609,9 +7714,9 @@ else if ((cc[-1] & XCL_MAP) != 0) } #ifdef SUPPORT_UNICODE -if (needstype || needsscript) +if (unicode_status & XCLASS_NEEDS_UCD) { - if (needschar && !charsaved) + if ((unicode_status & (XCLASS_SAVE_CHAR | XCLASS_CHAR_SAVED)) == XCLASS_SAVE_CHAR) OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); #if PCRE2_CODE_UNIT_WIDTH == 32 @@ -7631,17 +7736,16 @@ if (needstype || needsscript) OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2)); OP1(SLJIT_MOV_U16, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); + OP2(SLJIT_SHL, TMP1, 0, TMP2, 0, SLJIT_IMM, 3); + OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 2); + OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0); - /* Before anything else, we deal with scripts. */ - if (needsscript) - { - OP2(SLJIT_SHL, TMP1, 0, TMP2, 0, SLJIT_IMM, 3); - OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 2); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); - - OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); + ccbegin = cc; - ccbegin = cc; + if (unicode_status & XCLASS_HAS_BIDICL) + { + OP1(SLJIT_MOV_U16, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, scriptx_bidiclass)); + OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BIDICLASS_SHIFT); while (*cc != XCL_END) { @@ -7660,7 +7764,7 @@ if (needstype || needsscript) { SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); cc++; - if (*cc == PT_SC) + if (*cc == PT_BIDICL) { compares--; invertcmp = (compares == 0 && list != backtracks); @@ -7674,52 +7778,176 @@ if (needstype || needsscript) } cc = ccbegin; + } + + if (unicode_status & XCLASS_HAS_BOOL) + { + OP1(SLJIT_MOV_U16, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, bprops)); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BPROPS_MASK); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 2); - if (needstype) + while (*cc != XCL_END) { - /* TMP2 has already been shifted by 2 */ - if (!needschar) + if (*cc == XCL_SINGLE) { - OP2(SLJIT_ADD, TMP1, 0, TMP2, 0, TMP2, 0); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); - - OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); + cc ++; + GETCHARINCTEST(c, cc); + } + else if (*cc == XCL_RANGE) + { + cc ++; + GETCHARINCTEST(c, cc); + GETCHARINCTEST(c, cc); } else { - OP2(SLJIT_ADD, TMP1, 0, TMP2, 0, TMP2, 0); - OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0); + SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); + cc++; + if (*cc == PT_BOOL) + { + compares--; + invertcmp = (compares == 0 && list != backtracks); + if (cc[-1] == XCL_NOTPROP) + invertcmp ^= 0x1; - OP1(SLJIT_MOV, TMP1, 0, RETURN_ADDR, 0); - OP1(SLJIT_MOV_U8, RETURN_ADDR, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); - typereg = RETURN_ADDR; + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP1), (sljit_sw)(PRIV(ucd_boolprop_sets) + (cc[1] >> 5)), SLJIT_IMM, (sljit_sw)1 << (cc[1] & 0x1f)); + add_jump(compiler, compares > 0 ? list : backtracks, JUMP(SLJIT_NOT_ZERO ^ invertcmp)); + } + cc += 2; } } - else if (needschar) - OP1(SLJIT_MOV, TMP1, 0, RETURN_ADDR, 0); + + cc = ccbegin; } - else if (needstype) + + if (unicode_status & XCLASS_HAS_SCRIPT) { - OP2(SLJIT_SHL, TMP1, 0, TMP2, 0, SLJIT_IMM, 3); - OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 2); + OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); - if (!needschar) + while (*cc != XCL_END) { - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); + if (*cc == XCL_SINGLE) + { + cc ++; + GETCHARINCTEST(c, cc); + } + else if (*cc == XCL_RANGE) + { + cc ++; + GETCHARINCTEST(c, cc); + GETCHARINCTEST(c, cc); + } + else + { + SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); + cc++; + switch (*cc) + { + case PT_SCX: + if (cc[-1] == XCL_NOTPROP) + break; + /* Fall through */ + + case PT_SC: + compares--; + invertcmp = (compares == 0 && list != backtracks); + if (cc[-1] == XCL_NOTPROP) + invertcmp ^= 0x1; - OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); + add_jump(compiler, compares > 0 ? list : backtracks, CMP(SLJIT_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, (int)cc[1])); + } + cc += 2; + } } - else + + cc = ccbegin; + } + + if (unicode_status & XCLASS_HAS_SCRIPT_EXTENSION) + { + OP1(SLJIT_MOV_U16, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, scriptx_bidiclass)); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_SCRIPTX_MASK); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 2); + + if (unicode_status & XCLASS_SCRIPT_EXTENSION_NOTPROP) { - OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0); + if (unicode_status & XCLASS_HAS_TYPE) + { + if (unicode_status & XCLASS_SAVE_CHAR) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS0, TMP2, 0); + unicode_status |= XCLASS_SCRIPT_EXTENSION_RESTORE_LOCALS0; + } + else + { + OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP2, 0); + unicode_status |= XCLASS_SCRIPT_EXTENSION_RESTORE_RETURN_ADDR; + } + } + OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); + } - OP1(SLJIT_MOV, TMP1, 0, RETURN_ADDR, 0); - OP1(SLJIT_MOV_U8, RETURN_ADDR, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); - typereg = RETURN_ADDR; + while (*cc != XCL_END) + { + if (*cc == XCL_SINGLE) + { + cc ++; + GETCHARINCTEST(c, cc); + } + else if (*cc == XCL_RANGE) + { + cc ++; + GETCHARINCTEST(c, cc); + GETCHARINCTEST(c, cc); + } + else + { + SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); + cc++; + if (*cc == PT_SCX) + { + compares--; + invertcmp = (compares == 0 && list != backtracks); + + jump = NULL; + if (cc[-1] == XCL_NOTPROP) + { + jump = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, (int)cc[1]); + if (invertcmp) + { + add_jump(compiler, backtracks, jump); + jump = NULL; + } + invertcmp ^= 0x1; + } + + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP1), (sljit_sw)(PRIV(ucd_script_sets) + (cc[1] >> 5)), SLJIT_IMM, (sljit_sw)1 << (cc[1] & 0x1f)); + add_jump(compiler, compares > 0 ? list : backtracks, JUMP(SLJIT_NOT_ZERO ^ invertcmp)); + + if (jump != NULL) + JUMPHERE(jump); + } + cc += 2; + } } + + if (unicode_status & XCLASS_SCRIPT_EXTENSION_RESTORE_LOCALS0) + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), LOCALS0); + else if (unicode_status & XCLASS_SCRIPT_EXTENSION_RESTORE_RETURN_ADDR) + OP1(SLJIT_MOV, TMP2, 0, RETURN_ADDR, 0); + cc = ccbegin; } - else if (needschar) + + if (unicode_status & XCLASS_SAVE_CHAR) OP1(SLJIT_MOV, TMP1, 0, RETURN_ADDR, 0); + + if (unicode_status & XCLASS_HAS_TYPE) + { + if (unicode_status & XCLASS_SAVE_CHAR) + typereg = RETURN_ADDR; + + OP1(SLJIT_MOV_U8, typereg, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); + } } #endif /* SUPPORT_UNICODE */ @@ -7743,13 +7971,13 @@ while (*cc != XCL_END) if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_EQUAL); numberofcmps++; } else if (numberofcmps > 0) { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); numberofcmps = 0; @@ -7769,13 +7997,13 @@ while (*cc != XCL_END) if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) { - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); numberofcmps++; } else if (numberofcmps > 0) { - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset)); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_LESS_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); numberofcmps = 0; @@ -7801,11 +8029,11 @@ while (*cc != XCL_END) break; case PT_LAMP: - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); + OP2U(SLJIT_SUB | SLJIT_SET_Z, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); + OP2U(SLJIT_SUB | SLJIT_SET_Z, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); + OP2U(SLJIT_SUB | SLJIT_SET_Z, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; @@ -7821,6 +8049,9 @@ while (*cc != XCL_END) break; case PT_SC: + case PT_SCX: + case PT_BOOL: + case PT_BIDICL: compares++; /* Do nothing. */ break; @@ -7828,32 +8059,32 @@ while (*cc != XCL_END) case PT_SPACE: case PT_PXSPACE: SET_CHAR_OFFSET(9); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd - 0x9); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0xd - 0x9); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x85 - 0x9); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x180e - 0x9); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); SET_TYPE_OFFSET(ucp_Zl); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_LESS_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; case PT_WORD: - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_UNDERSCORE - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_UNDERSCORE - charoffset)); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); /* Fall through. */ case PT_ALNUM: SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); OP_FLAGS((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); SET_TYPE_OFFSET(ucp_Nd); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_LESS_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; @@ -7876,7 +8107,7 @@ while (*cc != XCL_END) OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); } - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, other_cases[1]); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); other_cases += 2; } @@ -7889,41 +8120,41 @@ while (*cc != XCL_END) OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); } - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, other_cases[2]); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(other_cases[0] - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(other_cases[0] - charoffset)); OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_Z : 0), TMP2, 0, SLJIT_EQUAL); other_cases += 3; } else { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset)); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); } while (*other_cases != NOTACHAR) { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset)); OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_Z : 0), TMP2, 0, SLJIT_EQUAL); } jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; case PT_UCNC: - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_DOLLAR_SIGN - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_DOLLAR_SIGN - charoffset)); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_COMMERCIAL_AT - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_COMMERCIAL_AT - charoffset)); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_GRAVE_ACCENT - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_GRAVE_ACCENT - charoffset)); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); SET_CHAR_OFFSET(0xa0); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(0xd7ff - charoffset)); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, (sljit_sw)(0xd7ff - charoffset)); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL); SET_CHAR_OFFSET(0); - OP2(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xe000 - 0); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_GREATER_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; @@ -7931,20 +8162,20 @@ while (*cc != XCL_END) case PT_PXGRAPH: /* C and Z groups are the farthest two groups. */ SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER); jump = CMP(SLJIT_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll); /* In case of ucp_Cf, we overwrite the result. */ SET_CHAR_OFFSET(0x2066); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); JUMPHERE(jump); @@ -7954,20 +8185,20 @@ while (*cc != XCL_END) case PT_PXPRINT: /* C and Z groups are the farthest two groups. */ SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll); + OP2U(SLJIT_SUB | SLJIT_SET_Z, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll); OP_FLAGS(SLJIT_AND, TMP2, 0, SLJIT_NOT_EQUAL); jump = CMP(SLJIT_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll); /* In case of ucp_Cf, we overwrite the result. */ SET_CHAR_OFFSET(0x2066); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL); JUMPHERE(jump); @@ -7976,15 +8207,15 @@ while (*cc != XCL_END) case PT_PXPUNCT: SET_TYPE_OFFSET(ucp_Sc); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL); SET_CHAR_OFFSET(0); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x7f); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x7f); OP_FLAGS(SLJIT_AND, TMP2, 0, SLJIT_LESS_EQUAL); SET_TYPE_OFFSET(ucp_Pc); - OP2(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc); + OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_LESS_EQUAL); jump = JUMP(SLJIT_NOT_ZERO ^ invertcmp); break; @@ -8069,9 +8300,9 @@ switch(type) else { jump[1] = CMP(SLJIT_EQUAL, TMP2, 0, STR_END, 0); - OP2(SLJIT_SUB | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); + OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, STR_END, 0); OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_NOT_EQUAL); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_EQUAL)); check_partial(common, TRUE); @@ -8094,7 +8325,7 @@ switch(type) OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); jump[1] = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); - OP2(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); + OP2U(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_GREATER, TMP2, 0, STR_END, 0); jump[2] = JUMP(SLJIT_GREATER); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_EQUAL) /* LESS */); /* Equal. */ @@ -8137,10 +8368,10 @@ switch(type) if (HAS_VIRTUAL_REGISTERS) { OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); } else - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO)); if (!common->endonly) @@ -8157,10 +8388,10 @@ switch(type) if (HAS_VIRTUAL_REGISTERS) { OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); } else - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO)); check_partial(common, FALSE); jump[0] = JUMP(SLJIT_JUMP); @@ -8200,14 +8431,14 @@ switch(type) OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); add_jump(compiler, backtracks, CMP(SLJIT_GREATER, STR_PTR, 0, TMP1, 0)); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO)); } else { OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin)); add_jump(compiler, backtracks, CMP(SLJIT_GREATER, STR_PTR, 0, TMP1, 0)); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO)); } return cc; @@ -8219,13 +8450,13 @@ switch(type) OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); jump[1] = CMP(SLJIT_GREATER, STR_PTR, 0, TMP2, 0); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); } else { OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin)); jump[1] = CMP(SLJIT_GREATER, STR_PTR, 0, TMP2, 0); - OP2(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_UNUSED, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); + OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL); } add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO)); jump[0] = JUMP(SLJIT_JUMP); @@ -8319,7 +8550,7 @@ do /* Not breaking between Regional Indicators is allowed only if there are an even number of preceding RIs. */ - if (lgb == ucp_gbRegionalIndicator && rgb == ucp_gbRegionalIndicator) + if (lgb == ucp_gbRegional_Indicator && rgb == ucp_gbRegional_Indicator) { ricount = 0; bptr = prevcc; @@ -8331,7 +8562,7 @@ do BACKCHAR(bptr); GETCHAR(c, bptr); - if (UCD_GRAPHBREAK(c) != ucp_gbRegionalIndicator) + if (UCD_GRAPHBREAK(c) != ucp_gbRegional_Indicator) break; ricount++; @@ -8387,7 +8618,7 @@ do /* Not breaking between Regional Indicators is allowed only if there are an even number of preceding RIs. */ - if (lgb == ucp_gbRegionalIndicator && rgb == ucp_gbRegionalIndicator) + if (lgb == ucp_gbRegional_Indicator && rgb == ucp_gbRegional_Indicator) { ricount = 0; bptr = prevcc; @@ -8397,7 +8628,7 @@ do { GETCHARBACK_INVALID(c, bptr, start_subject, break); - if (UCD_GRAPHBREAK(c) != ucp_gbRegionalIndicator) + if (UCD_GRAPHBREAK(c) != ucp_gbRegional_Indicator) break; ricount++; @@ -8455,7 +8686,7 @@ while (cc < end_subject) /* Not breaking between Regional Indicators is allowed only if there are an even number of preceding RIs. */ - if (lgb == ucp_gbRegionalIndicator && rgb == ucp_gbRegionalIndicator) + if (lgb == ucp_gbRegional_Indicator && rgb == ucp_gbRegional_Indicator) { ricount = 0; bptr = cc - 1; @@ -8470,7 +8701,7 @@ while (cc < end_subject) break; #endif /* PCRE2_CODE_UNIT_WIDTH == 32 */ - if (UCD_GRAPHBREAK(c) != ucp_gbRegionalIndicator) break; + if (UCD_GRAPHBREAK(c) != ucp_gbRegional_Indicator) break; ricount++; } @@ -8520,7 +8751,7 @@ switch(type) #endif read_char8_type(common, backtracks, type == OP_NOT_DIGIT); /* Flip the starting bit in the negative case. */ - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, ctype_digit); add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_ZERO : SLJIT_NOT_ZERO)); return cc; @@ -8534,7 +8765,7 @@ switch(type) else #endif read_char8_type(common, backtracks, type == OP_NOT_WHITESPACE); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, ctype_space); add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_ZERO : SLJIT_NOT_ZERO)); return cc; @@ -8548,7 +8779,7 @@ switch(type) else #endif read_char8_type(common, backtracks, type == OP_NOT_WORDCHAR); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, ctype_word); add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_ZERO : SLJIT_NOT_ZERO)); return cc; @@ -8596,7 +8827,7 @@ switch(type) #elif PCRE2_CODE_UNIT_WIDTH == 16 jump[0] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xd800); OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xd800); OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL); OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); @@ -8690,13 +8921,13 @@ switch(type) OP1(SLJIT_MOV, SLJIT_R0, 0, ARGUMENTS, 0); #if PCRE2_CODE_UNIT_WIDTH != 32 - sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, - common->utf ? (common->invalid_utf ? SLJIT_FUNC_OFFSET(do_extuni_utf_invalid) : SLJIT_FUNC_OFFSET(do_extuni_utf)) : SLJIT_FUNC_OFFSET(do_extuni_no_utf)); + sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, + common->utf ? (common->invalid_utf ? SLJIT_FUNC_ADDR(do_extuni_utf_invalid) : SLJIT_FUNC_ADDR(do_extuni_utf)) : SLJIT_FUNC_ADDR(do_extuni_no_utf)); if (common->invalid_utf) add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); #else - sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, - common->invalid_utf ? SLJIT_FUNC_OFFSET(do_extuni_utf_invalid) : SLJIT_FUNC_OFFSET(do_extuni_no_utf)); + sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, + common->invalid_utf ? SLJIT_FUNC_ADDR(do_extuni_utf_invalid) : SLJIT_FUNC_ADDR(do_extuni_no_utf)); if (!common->utf || common->invalid_utf) add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); #endif @@ -8758,7 +8989,7 @@ switch(type) if (sljit_has_cpu_feature(SLJIT_HAS_CMOV)) { - OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); + OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, oc); CMOV(SLJIT_EQUAL, TMP1, SLJIT_IMM, c); add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); } @@ -8878,7 +9109,7 @@ switch(type) OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP2, 0); add_jump(compiler, backtracks, JUMP(SLJIT_ZERO)); #if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8 @@ -9116,7 +9347,7 @@ if (common->utf && *cc == OP_REFI) caseless_loop = LABEL(); OP1(SLJIT_MOV_U32, TMP1, 0, SLJIT_MEM1(TMP2), 0); OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, sizeof(uint32_t)); - OP2(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_LESS, SLJIT_UNUSED, 0, TMP1, 0, char1_reg, 0); + OP2U(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_LESS, TMP1, 0, char1_reg, 0); JUMPTO(SLJIT_EQUAL, loop); JUMPTO(SLJIT_LESS, caseless_loop); @@ -9575,12 +9806,12 @@ OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS0, STR_PTR, 0); /* SLJIT_R0 = arguments */ OP1(SLJIT_MOV, SLJIT_R1, 0, STACK_TOP, 0); GET_LOCAL_BASE(SLJIT_R2, 0, OVECTOR_START); -sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(S32) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW) | SLJIT_ARG3(SW), SLJIT_IMM, SLJIT_FUNC_OFFSET(do_callout)); +sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS3(32, W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_callout)); OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), LOCALS0); free_stack(common, callout_arg_size); /* Check return value. */ -OP2(SLJIT_SUB32 | SLJIT_SET_Z | SLJIT_SET_SIG_GREATER, SLJIT_UNUSED, 0, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); +OP2U(SLJIT_SUB32 | SLJIT_SET_Z | SLJIT_SET_SIG_GREATER, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_SIG_GREATER)); if (common->abort_label == NULL) add_jump(compiler, &common->abort, JUMP(SLJIT_NOT_EQUAL) /* SIG_LESS */); @@ -10148,10 +10379,10 @@ SLJIT_ASSERT(TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1); OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr); #ifdef SUPPORT_UNICODE -sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, - common->utf ? SLJIT_FUNC_OFFSET(do_script_run_utf) : SLJIT_FUNC_OFFSET(do_script_run)); +sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, + common->utf ? SLJIT_FUNC_ADDR(do_script_run_utf) : SLJIT_FUNC_ADDR(do_script_run)); #else -sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, SLJIT_FUNC_OFFSET(do_script_run)); +sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_script_run)); #endif OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); @@ -11374,7 +11605,7 @@ switch(opcode) if (common->mode == PCRE2_JIT_COMPLETE) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_PTR, 0, STR_END, 0); CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0); } else @@ -11667,7 +11898,7 @@ switch(opcode) if (common->mode == PCRE2_JIT_COMPLETE) { - OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0); + OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_PTR, 0, STR_END, 0); CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0); } else @@ -11751,9 +11982,9 @@ if (HAS_VIRTUAL_REGISTERS) else OP1(SLJIT_MOV_U32, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options)); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY); add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_NOT_ZERO)); -OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART); +OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART); if (common->accept_label == NULL) add_jump(compiler, &common->accept, JUMP(SLJIT_ZERO)); else @@ -13004,7 +13235,7 @@ if (opcode == OP_SKIP_ARG) SLJIT_ASSERT(common->control_head_ptr != 0 && TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1); OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr); OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, (sljit_sw)(current->cc + 2)); - sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, SLJIT_FUNC_OFFSET(do_search_mark)); + sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_search_mark)); OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_R0, 0); add_jump(compiler, &common->reset_match, CMP(SLJIT_NOT_EQUAL, SLJIT_R0, 0, SLJIT_IMM, 0)); @@ -13248,10 +13479,8 @@ DEFINE_COMPILER; PCRE2_SPTR cc = common->start + common->currententry->start; PCRE2_SPTR ccbegin = cc + 1 + LINK_SIZE + (*cc == OP_BRA ? 0 : IMM2_SIZE); PCRE2_SPTR ccend = bracketend(cc) - (1 + LINK_SIZE); -BOOL needs_control_head; -BOOL has_quit; -BOOL has_accept; -int private_data_size = get_recurse_data_length(common, ccbegin, ccend, &needs_control_head, &has_quit, &has_accept); +uint32_t recurse_flags = 0; +int private_data_size = get_recurse_data_length(common, ccbegin, ccend, &recurse_flags); int alt_count, alt_max, local_size; backtrack_common altbacktrack; jump_list *match = NULL; @@ -13285,12 +13514,12 @@ allocate_stack(common, private_data_size + local_size); /* Save return address. */ OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1), TMP2, 0); -copy_recurse_data(common, ccbegin, ccend, recurse_copy_from_global, local_size, private_data_size + local_size, has_quit); +copy_recurse_data(common, ccbegin, ccend, recurse_copy_from_global, local_size, private_data_size + local_size, recurse_flags); /* This variable is saved and restored all time when we enter or exit from a recursive context. */ OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr, STACK_TOP, 0); -if (needs_control_head) +if (recurse_flags & recurse_flag_control_head_found) OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0); if (alt_max > 1) @@ -13315,10 +13544,10 @@ while (1) if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) return; - allocate_stack(common, (alt_max > 1 || has_accept) ? 2 : 1); + allocate_stack(common, (alt_max > 1 || (recurse_flags & recurse_flag_accept_found)) ? 2 : 1); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr); - if (alt_max > 1 || has_accept) + if (alt_max > 1 || (recurse_flags & recurse_flag_accept_found)) { if (alt_max > 3) put_label = sljit_emit_put_label(compiler, SLJIT_MEM1(STACK_TOP), STACK(1)); @@ -13337,14 +13566,14 @@ while (1) sljit_emit_fast_enter(compiler, TMP1, 0); - if (has_accept) + if (recurse_flags & recurse_flag_accept_found) accept_exit = CMP(SLJIT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, -1); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); /* Save return address. */ OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), STACK(local_size - 1), TMP1, 0); - copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, has_quit); + copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, recurse_flags); if (alt_max > 1) { @@ -13361,7 +13590,7 @@ while (1) next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0); } else - free_stack(common, has_accept ? 2 : 1); + free_stack(common, (recurse_flags & recurse_flag_accept_found) ? 2 : 1); } else if (alt_max > 3) { @@ -13396,7 +13625,7 @@ while (1) quit = LABEL(); -copy_recurse_data(common, ccbegin, ccend, recurse_copy_private_to_global, local_size, private_data_size + local_size, has_quit); +copy_recurse_data(common, ccbegin, ccend, recurse_copy_private_to_global, local_size, private_data_size + local_size, recurse_flags); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1)); free_stack(common, private_data_size + local_size); @@ -13405,15 +13634,15 @@ OP_SRC(SLJIT_FAST_RETURN, TMP2, 0); if (common->quit != NULL) { - SLJIT_ASSERT(has_quit); + SLJIT_ASSERT(recurse_flags & recurse_flag_quit_found); set_jumps(common->quit, LABEL()); OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr); - copy_recurse_data(common, ccbegin, ccend, recurse_copy_shared_to_global, local_size, private_data_size + local_size, has_quit); + copy_recurse_data(common, ccbegin, ccend, recurse_copy_shared_to_global, local_size, private_data_size + local_size, recurse_flags); JUMPTO(SLJIT_JUMP, quit); } -if (has_accept) +if (recurse_flags & recurse_flag_accept_found) { JUMPHERE(accept_exit); free_stack(common, 2); @@ -13421,7 +13650,7 @@ if (has_accept) /* Save return address. */ OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1), TMP1, 0); - copy_recurse_data(common, ccbegin, ccend, recurse_copy_kept_shared_to_global, local_size, private_data_size + local_size, has_quit); + copy_recurse_data(common, ccbegin, ccend, recurse_copy_kept_shared_to_global, local_size, private_data_size + local_size, recurse_flags); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1)); free_stack(common, private_data_size + local_size); @@ -13431,7 +13660,7 @@ if (has_accept) if (common->accept != NULL) { - SLJIT_ASSERT(has_accept); + SLJIT_ASSERT(recurse_flags & recurse_flag_accept_found); set_jumps(common->accept, LABEL()); @@ -13446,7 +13675,7 @@ set_jumps(match, LABEL()); OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); -copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, has_quit); +copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, recurse_flags); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), STACK(local_size - 1)); OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1); @@ -13652,7 +13881,7 @@ SLJIT_ASSERT(!(common->req_char_ptr != 0 && common->start_used_ptr != 0)); common->cbra_ptr = OVECTOR_START + (re->top_bracket + 1) * 2 * sizeof(sljit_sw); total_length = ccend - common->start; -common->private_data_ptrs = (sljit_s32 *)SLJIT_MALLOC(total_length * (sizeof(sljit_s32) + (common->has_then ? 1 : 0)), allocator_data); +common->private_data_ptrs = (sljit_s32*)SLJIT_MALLOC(total_length * (sizeof(sljit_s32) + (common->has_then ? 1 : 0)), allocator_data); if (!common->private_data_ptrs) { SLJIT_FREE(common->optimized_cbracket, allocator_data); @@ -13692,8 +13921,9 @@ if (!compiler) } common->compiler = compiler; -/* Main pcre_jit_exec entry. */ -sljit_emit_enter(compiler, 0, SLJIT_ARG1(SW), 5, 5, 0, 0, private_data_size); +/* Main pcre2_jit_exec entry. */ +SLJIT_ASSERT((private_data_size & (sizeof(sljit_sw) - 1)) == 0); +sljit_emit_enter(compiler, 0, SLJIT_ARGS1(W, W), 5, 5, 0, 0, private_data_size); /* Register init. */ reset_ovector(common, (re->top_bracket + 1) * 2); @@ -13900,9 +14130,9 @@ if (common->might_be_empty) JUMPHERE(empty_match); OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); OP1(SLJIT_MOV_U32, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options)); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY); JUMPTO(SLJIT_NOT_ZERO, empty_match_backtrack_label); - OP2(SLJIT_AND | SLJIT_SET_Z, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART); + OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART); JUMPTO(SLJIT_ZERO, empty_match_found_label); OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, STR_PTR, 0, empty_match_found_label); @@ -13915,20 +14145,40 @@ common->early_fail_end_ptr = 0; common->currententry = common->entries; common->local_quit_available = TRUE; quit_label = common->quit_label; -while (common->currententry != NULL) +if (common->currententry != NULL) { - /* Might add new entries. */ - compile_recurse(common); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + /* A free bit for each private data. */ + common->recurse_bitset_size = ((private_data_size / (int)sizeof(sljit_sw)) + 7) >> 3; + SLJIT_ASSERT(common->recurse_bitset_size > 0); + common->recurse_bitset = (sljit_u8*)SLJIT_MALLOC(common->recurse_bitset_size, allocator_data);; + + if (common->recurse_bitset != NULL) + { + do + { + /* Might add new entries. */ + compile_recurse(common); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + break; + flush_stubs(common); + common->currententry = common->currententry->next; + } + while (common->currententry != NULL); + + SLJIT_FREE(common->recurse_bitset, allocator_data); + } + + if (common->currententry != NULL) { + /* The common->recurse_bitset has been freed. */ + SLJIT_ASSERT(sljit_get_compiler_error(compiler) || common->recurse_bitset == NULL); + sljit_free_compiler(compiler); SLJIT_FREE(common->optimized_cbracket, allocator_data); SLJIT_FREE(common->private_data_ptrs, allocator_data); PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data); return PCRE2_ERROR_NOMEMORY; } - flush_stubs(common); - common->currententry = common->currententry->next; } common->local_quit_available = FALSE; common->quit_label = quit_label; @@ -13947,7 +14197,7 @@ OP2(SLJIT_SUB, SLJIT_R1, 0, STACK_LIMIT, 0, SLJIT_IMM, STACK_GROWTH_RATE); OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, stack)); OP1(SLJIT_MOV, STACK_LIMIT, 0, TMP2, 0); -sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_RET(SW) | SLJIT_ARG1(SW) | SLJIT_ARG2(SW), SLJIT_IMM, SLJIT_FUNC_OFFSET(sljit_stack_resize)); +sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(sljit_stack_resize)); jump = CMP(SLJIT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); OP1(SLJIT_MOV, TMP2, 0, STACK_LIMIT, 0); |