Parent Directory
|
Revision Log
|
Patch
| revision 669 by ph10, Tue Aug 23 09:56:11 2011 UTC | revision 953 by zherczeg, Thu Mar 29 17:41:57 2012 UTC | |
|---|---|---|
| # | Line 6 | Line 6 |
| 6 | and semantics are as close as possible to those of the Perl 5 language. | and semantics are as close as possible to those of the Perl 5 language. |
| 7 | ||
| 8 | Written by Philip Hazel | Written by Philip Hazel |
| 9 | Copyright (c) 1997-2008 University of Cambridge | Copyright (c) 1997-2012 University of Cambridge |
| 10 | ||
| 11 | The machine code generator part (this module) was written by Zoltan Herczeg | The machine code generator part (this module) was written by Zoltan Herczeg |
| 12 | Copyright (c) 2010-2011 | Copyright (c) 2010-2012 |
| 13 | ||
| 14 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
| 15 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
| # | Line 52 POSSIBILITY OF SUCH DAMAGE. | Line 52 POSSIBILITY OF SUCH DAMAGE. |
| 52 | we just include it. This way we don't need to touch the build | we just include it. This way we don't need to touch the build |
| 53 | system files. */ | system files. */ |
| 54 | ||
| 55 | #define SLJIT_MALLOC(size) (PUBL(malloc))(size) | |
| 56 | #define SLJIT_FREE(ptr) (PUBL(free))(ptr) | |
| 57 | #define SLJIT_CONFIG_AUTO 1 | #define SLJIT_CONFIG_AUTO 1 |
| 58 | #define SLJIT_CONFIG_STATIC 1 | |
| 59 | #define SLJIT_VERBOSE 0 | #define SLJIT_VERBOSE 0 |
| 60 | #define SLJIT_DEBUG 0 | #define SLJIT_DEBUG 0 |
| 61 | ||
| 62 | #include "sljit/sljitLir.c" | #include "sljit/sljitLir.c" |
| 63 | ||
| 64 | #if defined SLJIT_CONFIG_UNSUPPORTED && SLJIT_CONFIG_UNSUPPORTED | #if defined SLJIT_CONFIG_UNSUPPORTED && SLJIT_CONFIG_UNSUPPORTED |
| 65 | #error "Unsupported architecture" | #error Unsupported architecture |
| 66 | #endif | #endif |
| 67 | ||
| 68 | /* Allocate memory on the stack. Fast, but limited size. */ | /* Allocate memory on the stack. Fast, but limited size. */ |
| # | Line 119 The generated code will be the following | Line 122 The generated code will be the following |
| 122 | jump to D hot path | jump to D hot path |
| 123 | C fallback path | C fallback path |
| 124 | A fallback path | A fallback path |
| 125 | ||
| 126 | Notice, that the order of fallback code paths are the opposite of the fast | Notice, that the order of fallback code paths are the opposite of the fast |
| 127 | code paths. In this way the topmost value on the stack is always belong | code paths. In this way the topmost value on the stack is always belong |
| 128 | to the current fallback code path. The fallback code path must check | to the current fallback code path. The fallback code path must check |
| # | Line 145 Thus we can restore the locals to a part | Line 148 Thus we can restore the locals to a part |
| 148 | typedef struct jit_arguments { | typedef struct jit_arguments { |
| 149 | /* Pointers first. */ | /* Pointers first. */ |
| 150 | struct sljit_stack *stack; | struct sljit_stack *stack; |
| 151 | PCRE_SPTR str; | const pcre_uchar *str; |
| 152 | PCRE_SPTR begin; | const pcre_uchar *begin; |
| 153 | PCRE_SPTR end; | const pcre_uchar *end; |
| 154 | int *offsets; | int *offsets; |
| 155 | uschar *ptr; | pcre_uchar *uchar_ptr; |
| 156 | pcre_uchar *mark_ptr; | |
| 157 | /* Everything else after. */ | /* Everything else after. */ |
| 158 | int offsetcount; | int offsetcount; |
| 159 | uschar notbol; | int calllimit; |
| 160 | uschar noteol; | pcre_uint8 notbol; |
| 161 | uschar notempty; | pcre_uint8 noteol; |
| 162 | uschar notempty_atstart; | pcre_uint8 notempty; |
| 163 | pcre_uint8 notempty_atstart; | |
| 164 | } jit_arguments; | } jit_arguments; |
| 165 | ||
| 166 | typedef struct executable_function { | typedef struct executable_functions { |
| 167 | void *executable_func; | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; |
| 168 | pcre_jit_callback callback; | PUBL(jit_callback) callback; |
| 169 | void *userdata; | void *userdata; |
| 170 | } executable_function; | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; |
| 171 | } executable_functions; | |
| 172 | ||
| 173 | typedef struct jump_list { | typedef struct jump_list { |
| 174 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 175 | struct jump_list *next; | struct jump_list *next; |
| 176 | } jump_list; | } jump_list; |
| 177 | ||
| 178 | enum stub_types { stack_alloc, max_index }; | enum stub_types { stack_alloc }; |
| 179 | ||
| 180 | typedef struct stub_list { | typedef struct stub_list { |
| 181 | enum stub_types type; | enum stub_types type; |
| # | Line 193 typedef struct fallback_common { | Line 199 typedef struct fallback_common { |
| 199 | struct fallback_common *top; | struct fallback_common *top; |
| 200 | jump_list *topfallbacks; | jump_list *topfallbacks; |
| 201 | /* Opcode pointer. */ | /* Opcode pointer. */ |
| 202 | uschar *cc; | pcre_uchar *cc; |
| 203 | } fallback_common; | } fallback_common; |
| 204 | ||
| 205 | typedef struct assert_fallback { | typedef struct assert_fallback { |
| # | Line 264 typedef struct recurse_fallback { | Line 270 typedef struct recurse_fallback { |
| 270 | ||
| 271 | typedef struct compiler_common { | typedef struct compiler_common { |
| 272 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
| 273 | uschar *start; | pcre_uchar *start; |
| 274 | ||
| 275 | /* Local stack area size and variable pointers. */ | |
| 276 | int localsize; | int localsize; |
| 277 | int *localptrs; | int *localptrs; |
| const uschar *fcc; | ||
| sljit_w lcc; | ||
| 278 | int cbraptr; | int cbraptr; |
| 279 | /* OVector starting point. Must be divisible by 2. */ | |
| 280 | int ovector_start; | |
| 281 | /* Last known position of the requested byte. */ | |
| 282 | int req_char_ptr; | |
| 283 | /* Head of the last recursion. */ | |
| 284 | int recursive_head; | |
| 285 | /* First inspected character for partial matching. */ | |
| 286 | int start_used_ptr; | |
| 287 | /* Starting pointer for partial soft matches. */ | |
| 288 | int hit_start; | |
| 289 | /* End pointer of the first line. */ | |
| 290 | int first_line_end; | |
| 291 | /* Points to the marked string. */ | |
| 292 | int mark_ptr; | |
| 293 | ||
| 294 | /* Other */ | |
| 295 | const pcre_uint8 *fcc; | |
| 296 | sljit_w lcc; | |
| 297 | int mode; | |
| 298 | int nltype; | int nltype; |
| 299 | int newline; | int newline; |
| 300 | int bsr_nltype; | int bsr_nltype; |
| 301 | int endonly; | int endonly; |
| 302 | BOOL has_set_som; | |
| 303 | sljit_w ctypes; | sljit_w ctypes; |
| 304 | sljit_uw name_table; | |
| 305 | sljit_w name_count; | |
| 306 | sljit_w name_entry_size; | |
| 307 | ||
| 308 | /* Labels and jump lists. */ | |
| 309 | struct sljit_label *partialmatchlabel; | |
| 310 | struct sljit_label *leavelabel; | |
| 311 | struct sljit_label *acceptlabel; | struct sljit_label *acceptlabel; |
| 312 | stub_list *stubs; | stub_list *stubs; |
| 313 | recurse_entry *entries; | recurse_entry *entries; |
| 314 | recurse_entry *currententry; | recurse_entry *currententry; |
| 315 | jump_list *partialmatch; | |
| 316 | jump_list *leave; | |
| 317 | jump_list *accept; | jump_list *accept; |
| 318 | jump_list *calllimit; | |
| 319 | jump_list *stackalloc; | jump_list *stackalloc; |
| 320 | jump_list *revertframes; | jump_list *revertframes; |
| 321 | jump_list *wordboundary; | jump_list *wordboundary; |
| # | Line 289 typedef struct compiler_common { | Line 325 typedef struct compiler_common { |
| 325 | jump_list *casefulcmp; | jump_list *casefulcmp; |
| 326 | jump_list *caselesscmp; | jump_list *caselesscmp; |
| 327 | BOOL jscript_compat; | BOOL jscript_compat; |
| 328 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 329 | BOOL utf8; | BOOL utf; |
| 330 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 331 | BOOL useucp; | BOOL use_ucp; |
| 332 | #endif | #endif |
| 333 | jump_list *utf8readchar; | jump_list *utfreadchar; |
| 334 | jump_list *utf8readtype8; | #ifdef COMPILE_PCRE8 |
| 335 | jump_list *utfreadtype8; | |
| 336 | #endif | #endif |
| 337 | #endif /* SUPPORT_UTF */ | |
| 338 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 339 | jump_list *getucd; | jump_list *getucd; |
| 340 | #endif | #endif |
| # | Line 308 typedef struct compare_context { | Line 346 typedef struct compare_context { |
| 346 | int length; | int length; |
| 347 | int sourcereg; | int sourcereg; |
| 348 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
| 349 | int byteptr; | int ucharptr; |
| 350 | union { | union { |
| 351 | int asint; | sljit_i asint; |
| 352 | short asshort; | sljit_uh asushort; |
| 353 | #ifdef COMPILE_PCRE8 | |
| 354 | sljit_ub asbyte; | sljit_ub asbyte; |
| 355 | sljit_ub asbytes[4]; | sljit_ub asuchars[4]; |
| 356 | #else | |
| 357 | #ifdef COMPILE_PCRE16 | |
| 358 | sljit_uh asuchars[2]; | |
| 359 | #endif | |
| 360 | #endif | |
| 361 | } c; | } c; |
| 362 | union { | union { |
| 363 | int asint; | sljit_i asint; |
| 364 | short asshort; | sljit_uh asushort; |
| 365 | #ifdef COMPILE_PCRE8 | |
| 366 | sljit_ub asbyte; | sljit_ub asbyte; |
| 367 | sljit_ub asbytes[4]; | sljit_ub asuchars[4]; |
| 368 | #else | |
| 369 | #ifdef COMPILE_PCRE16 | |
| 370 | sljit_uh asuchars[2]; | |
| 371 | #endif | |
| 372 | #endif | |
| 373 | } oc; | } oc; |
| 374 | #endif | #endif |
| 375 | } compare_context; | } compare_context; |
| 376 | ||
| 377 | enum { | enum { |
| 378 | frame_end = 0, | frame_end = 0, |
| 379 | frame_setmaxindex = -1, | frame_setstrbegin = -1, |
| 380 | frame_setstrbegin = -2 | frame_setmark = -2 |
| 381 | }; | }; |
| 382 | ||
| 383 | /* Undefine sljit macros. */ | |
| 384 | #undef CMP | |
| 385 | ||
| 386 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
| 387 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) |
| 388 | ||
| 389 | #define TMP1 SLJIT_TEMPORARY_REG1 | #define TMP1 SLJIT_TEMPORARY_REG1 |
| 390 | #define TMP2 SLJIT_TEMPORARY_REG3 | #define TMP2 SLJIT_TEMPORARY_REG3 |
| 391 | #define TMP3 SLJIT_TEMPORARY_EREG2 | #define TMP3 SLJIT_TEMPORARY_EREG2 |
| 392 | #define STR_PTR SLJIT_GENERAL_REG1 | #define STR_PTR SLJIT_SAVED_REG1 |
| 393 | #define STR_END SLJIT_GENERAL_REG2 | #define STR_END SLJIT_SAVED_REG2 |
| 394 | #define STACK_TOP SLJIT_TEMPORARY_REG2 | #define STACK_TOP SLJIT_TEMPORARY_REG2 |
| 395 | #define STACK_LIMIT SLJIT_GENERAL_REG3 | #define STACK_LIMIT SLJIT_SAVED_REG3 |
| 396 | #define ARGUMENTS SLJIT_GENERAL_EREG1 | #define ARGUMENTS SLJIT_SAVED_EREG1 |
| 397 | #define MAX_INDEX SLJIT_GENERAL_EREG2 | #define CALL_COUNT SLJIT_SAVED_EREG2 |
| 398 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 |
| 399 | ||
| 400 | /* Locals layout. */ | /* Locals layout. */ |
| # | Line 351 enum { | Line 404 enum { |
| 404 | /* Two local variables for possessive quantifiers (char1 cannot use them). */ | /* Two local variables for possessive quantifiers (char1 cannot use them). */ |
| 405 | #define POSSESSIVE0 (2 * sizeof(sljit_w)) | #define POSSESSIVE0 (2 * sizeof(sljit_w)) |
| 406 | #define POSSESSIVE1 (3 * sizeof(sljit_w)) | #define POSSESSIVE1 (3 * sizeof(sljit_w)) |
| 407 | /* Head of the saved local variables */ | /* Max limit of recursions. */ |
| 408 | #define LOCALS_HEAD (4 * sizeof(sljit_w)) | #define CALL_LIMIT (4 * sizeof(sljit_w)) |
| /* Head of the last recursion. */ | ||
| #define RECURSIVE_HEAD (5 * sizeof(sljit_w)) | ||
| /* Last known position of the requested byte. */ | ||
| #define REQ_BYTE_PTR (6 * sizeof(sljit_w)) | ||
| /* End pointer of the first line. */ | ||
| #define FIRSTLINE_END (7 * sizeof(sljit_w)) | ||
| 409 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
| 410 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
| 411 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
| 412 | the start pointers when the end of the capturing group has not yet reached. */ | the start pointers when the end of the capturing group has not yet reached. */ |
| 413 | #define OVECTOR_START (8 * sizeof(sljit_w)) | #define OVECTOR_START (common->ovector_start) |
| 414 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) |
| 415 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) |
| 416 | #define PRIV(cc) (common->localptrs[(cc) - common->start]) | #define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) |
| 417 | ||
| 418 | #ifdef COMPILE_PCRE8 | |
| 419 | #define MOV_UCHAR SLJIT_MOV_UB | |
| 420 | #define MOVU_UCHAR SLJIT_MOVU_UB | |
| 421 | #else | |
| 422 | #ifdef COMPILE_PCRE16 | |
| 423 | #define MOV_UCHAR SLJIT_MOV_UH | |
| 424 | #define MOVU_UCHAR SLJIT_MOVU_UH | |
| 425 | #else | |
| 426 | #error Unsupported compiling mode | |
| 427 | #endif | |
| 428 | #endif | |
| 429 | ||
| 430 | /* Shortcuts. */ | /* Shortcuts. */ |
| 431 | #define DEFINE_COMPILER \ | #define DEFINE_COMPILER \ |
| # | Line 390 the start pointers when the end of the c | Line 449 the start pointers when the end of the c |
| 449 | #define COND_VALUE(op, dst, dstw, type) \ | #define COND_VALUE(op, dst, dstw, type) \ |
| 450 | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) |
| 451 | ||
| 452 | static uschar* bracketend(uschar* cc) | static pcre_uchar* bracketend(pcre_uchar* cc) |
| 453 | { | { |
| 454 | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); |
| 455 | do cc += GET(cc, 1); while (*cc == OP_ALT); | do cc += GET(cc, 1); while (*cc == OP_ALT); |
| # | Line 399 cc += 1 + LINK_SIZE; | Line 458 cc += 1 + LINK_SIZE; |
| 458 | return cc; | return cc; |
| 459 | } | } |
| 460 | ||
| 461 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
| 462 | next_opcode | next_opcode |
| 463 | get_localspace | get_localspace |
| 464 | set_localptrs | set_localptrs |
| # | Line 411 return cc; | Line 470 return cc; |
| 470 | compile_fallbackpath | compile_fallbackpath |
| 471 | */ | */ |
| 472 | ||
| 473 | static uschar *next_opcode(compiler_common *common, uschar *cc) | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) |
| 474 | { | { |
| 475 | SLJIT_UNUSED_ARG(common); | SLJIT_UNUSED_ARG(common); |
| 476 | switch(*cc) | switch(*cc) |
| # | Line 460 switch(*cc) | Line 519 switch(*cc) |
| 519 | case OP_BRAZERO: | case OP_BRAZERO: |
| 520 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
| 521 | case OP_BRAPOSZERO: | case OP_BRAPOSZERO: |
| 522 | case OP_COMMIT: | |
| 523 | case OP_FAIL: | case OP_FAIL: |
| 524 | case OP_ACCEPT: | case OP_ACCEPT: |
| 525 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
| 526 | case OP_SKIPZERO: | case OP_SKIPZERO: |
| 527 | return cc + 1; | return cc + 1; |
| 528 | ||
| 529 | case OP_ANYBYTE: | |
| 530 | #ifdef SUPPORT_UTF | |
| 531 | if (common->utf) return NULL; | |
| 532 | #endif | |
| 533 | return cc + 1; | |
| 534 | ||
| 535 | case OP_CHAR: | case OP_CHAR: |
| 536 | case OP_CHARI: | case OP_CHARI: |
| 537 | case OP_NOT: | case OP_NOT: |
| 538 | case OP_NOTI: | case OP_NOTI: |
| 539 | case OP_STAR: | case OP_STAR: |
| 540 | case OP_MINSTAR: | case OP_MINSTAR: |
| 541 | case OP_PLUS: | case OP_PLUS: |
| # | Line 508 switch(*cc) | Line 573 switch(*cc) |
| 573 | case OP_NOTPOSPLUSI: | case OP_NOTPOSPLUSI: |
| 574 | case OP_NOTPOSQUERYI: | case OP_NOTPOSQUERYI: |
| 575 | cc += 2; | cc += 2; |
| 576 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 577 | if (common->utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
| 578 | #endif | #endif |
| 579 | return cc; | return cc; |
| 580 | ||
| # | Line 529 switch(*cc) | Line 594 switch(*cc) |
| 594 | case OP_NOTMINUPTOI: | case OP_NOTMINUPTOI: |
| 595 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
| 596 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
| 597 | cc += 4; | cc += 2 + IMM2_SIZE; |
| 598 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 599 | if (common->utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
| 600 | #endif | #endif |
| 601 | return cc; | return cc; |
| 602 | ||
| 603 | case OP_NOTPROP: | case OP_NOTPROP: |
| 604 | case OP_PROP: | case OP_PROP: |
| 605 | return cc + 1 + 2; | |
| 606 | ||
| 607 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
| 608 | case OP_TYPEMINUPTO: | case OP_TYPEMINUPTO: |
| 609 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
| # | Line 544 switch(*cc) | Line 611 switch(*cc) |
| 611 | case OP_REF: | case OP_REF: |
| 612 | case OP_REFI: | case OP_REFI: |
| 613 | case OP_CREF: | case OP_CREF: |
| 614 | case OP_NCREF: | |
| 615 | case OP_RREF: | |
| 616 | case OP_NRREF: | |
| 617 | case OP_CLOSE: | case OP_CLOSE: |
| 618 | cc += 3; | cc += 1 + IMM2_SIZE; |
| 619 | return cc; | return cc; |
| 620 | ||
| 621 | case OP_CRRANGE: | case OP_CRRANGE: |
| 622 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 623 | return cc + 5; | return cc + 1 + 2 * IMM2_SIZE; |
| 624 | ||
| 625 | case OP_CLASS: | case OP_CLASS: |
| 626 | case OP_NCLASS: | case OP_NCLASS: |
| 627 | return cc + 33; | return cc + 1 + 32 / sizeof(pcre_uchar); |
| 628 | ||
| 629 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 630 | case OP_XCLASS: | case OP_XCLASS: |
| 631 | return cc + GET(cc, 1); | return cc + GET(cc, 1); |
| 632 | #endif | #endif |
| # | Line 568 switch(*cc) | Line 638 switch(*cc) |
| 638 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| 639 | case OP_REVERSE: | case OP_REVERSE: |
| 640 | case OP_ONCE: | case OP_ONCE: |
| 641 | case OP_ONCE_NC: | |
| 642 | case OP_BRA: | case OP_BRA: |
| 643 | case OP_BRAPOS: | case OP_BRAPOS: |
| 644 | case OP_COND: | case OP_COND: |
| # | Line 585 switch(*cc) | Line 656 switch(*cc) |
| 656 | case OP_CBRAPOS: | case OP_CBRAPOS: |
| 657 | case OP_SCBRA: | case OP_SCBRA: |
| 658 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| 659 | return cc + 1 + LINK_SIZE + 2; | return cc + 1 + LINK_SIZE + IMM2_SIZE; |
| 660 | ||
| 661 | case OP_MARK: | |
| 662 | return cc + 1 + 2 + cc[1]; | |
| 663 | ||
| 664 | default: | default: |
| 665 | return NULL; | return NULL; |
| 666 | } | } |
| 667 | } | } |
| 668 | ||
| 669 | static int get_localspace(compiler_common *common, uschar *cc, uschar *ccend) | static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
| 670 | { | { |
| 671 | int localspace = 0; | int localspace = 0; |
| 672 | uschar *alternative; | pcre_uchar *alternative; |
| 673 | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ |
| 674 | while (cc < ccend) | while (cc < ccend) |
| 675 | { | { |
| 676 | switch(*cc) | switch(*cc) |
| 677 | { | { |
| 678 | case OP_SET_SOM: | |
| 679 | common->has_set_som = TRUE; | |
| 680 | cc += 1; | |
| 681 | break; | |
| 682 | ||
| 683 | case OP_ASSERT: | case OP_ASSERT: |
| 684 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 685 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
| 686 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| 687 | case OP_ONCE: | case OP_ONCE: |
| 688 | case OP_ONCE_NC: | |
| 689 | case OP_BRAPOS: | case OP_BRAPOS: |
| 690 | case OP_SBRA: | case OP_SBRA: |
| 691 | case OP_SBRAPOS: | case OP_SBRAPOS: |
| # | Line 617 while (cc < ccend) | Line 697 while (cc < ccend) |
| 697 | case OP_CBRAPOS: | case OP_CBRAPOS: |
| 698 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| 699 | localspace += sizeof(sljit_w); | localspace += sizeof(sljit_w); |
| 700 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 701 | break; | break; |
| 702 | ||
| 703 | case OP_COND: | case OP_COND: |
| # | Line 628 while (cc < ccend) | Line 708 while (cc < ccend) |
| 708 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
| 709 | break; | break; |
| 710 | ||
| 711 | case OP_RECURSE: | |
| 712 | /* Set its value only once. */ | |
| 713 | if (common->recursive_head == 0) | |
| 714 | { | |
| 715 | common->recursive_head = common->ovector_start; | |
| 716 | common->ovector_start += sizeof(sljit_w); | |
| 717 | } | |
| 718 | cc += 1 + LINK_SIZE; | |
| 719 | break; | |
| 720 | ||
| 721 | case OP_MARK: | |
| 722 | if (common->mark_ptr == 0) | |
| 723 | { | |
| 724 | common->mark_ptr = common->ovector_start; | |
| 725 | common->ovector_start += sizeof(sljit_w); | |
| 726 | } | |
| 727 | cc += 1 + 2 + cc[1]; | |
| 728 | break; | |
| 729 | ||
| 730 | default: | default: |
| 731 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
| 732 | if (cc == NULL) | if (cc == NULL) |
| # | Line 638 while (cc < ccend) | Line 737 while (cc < ccend) |
| 737 | return localspace; | return localspace; |
| 738 | } | } |
| 739 | ||
| 740 | static void set_localptrs(compiler_common *common, int localptr, uschar *ccend) | static void set_localptrs(compiler_common *common, int localptr, pcre_uchar *ccend) |
| 741 | { | { |
| 742 | uschar *cc = common->start; | pcre_uchar *cc = common->start; |
| 743 | uschar *alternative; | pcre_uchar *alternative; |
| 744 | while (cc < ccend) | while (cc < ccend) |
| 745 | { | { |
| 746 | switch(*cc) | switch(*cc) |
| # | Line 651 while (cc < ccend) | Line 750 while (cc < ccend) |
| 750 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
| 751 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| 752 | case OP_ONCE: | case OP_ONCE: |
| 753 | case OP_ONCE_NC: | |
| 754 | case OP_BRAPOS: | case OP_BRAPOS: |
| 755 | case OP_SBRA: | case OP_SBRA: |
| 756 | case OP_SBRAPOS: | case OP_SBRAPOS: |
| # | Line 664 while (cc < ccend) | Line 764 while (cc < ccend) |
| 764 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| 765 | common->localptrs[cc - common->start] = localptr; | common->localptrs[cc - common->start] = localptr; |
| 766 | localptr += sizeof(sljit_w); | localptr += sizeof(sljit_w); |
| 767 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 768 | break; | break; |
| 769 | ||
| 770 | case OP_COND: | case OP_COND: |
| # | Line 687 while (cc < ccend) | Line 787 while (cc < ccend) |
| 787 | } | } |
| 788 | ||
| 789 | /* Returns with -1 if no need for frame. */ | /* Returns with -1 if no need for frame. */ |
| 790 | static int get_framesize(compiler_common *common, uschar *cc, BOOL recursive) | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) |
| 791 | { | { |
| 792 | uschar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc); |
| uschar *end; | ||
| 793 | int length = 0; | int length = 0; |
| 794 | BOOL possessive = FALSE; | BOOL possessive = FALSE; |
| 795 | BOOL needs_frame = FALSE; | BOOL setsom_found = recursive; |
| 796 | BOOL needs_maxindex = FALSE; | BOOL setmark_found = recursive; |
| BOOL setsom_found = FALSE; | ||
| 797 | ||
| 798 | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
| 799 | { | { |
| 800 | length = 3 + 2; | length = 3; |
| needs_maxindex = TRUE; | ||
| 801 | possessive = TRUE; | possessive = TRUE; |
| 802 | } | } |
| 803 | ||
| # | Line 710 while (cc < ccend) | Line 807 while (cc < ccend) |
| 807 | switch(*cc) | switch(*cc) |
| 808 | { | { |
| 809 | case OP_SET_SOM: | case OP_SET_SOM: |
| 810 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
| 811 | if (!setsom_found) | if (!setsom_found) |
| 812 | { | { |
| 813 | length += 2; | length += 2; |
| 814 | setsom_found = TRUE; | setsom_found = TRUE; |
| 815 | } | } |
| 816 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | cc += 1; |
| 817 | break; | break; |
| 818 | ||
| 819 | case OP_ASSERT: | case OP_MARK: |
| 820 | case OP_ASSERT_NOT: | SLJIT_ASSERT(common->mark_ptr != 0); |
| 821 | case OP_ASSERTBACK: | if (!setmark_found) |
| case OP_ASSERTBACK_NOT: | ||
| case OP_ONCE: | ||
| if (needs_frame || length > 0) | ||
| 822 | { | { |
| 823 | cc = bracketend(cc); | length += 2; |
| 824 | break; | setmark_found = TRUE; |
| 825 | } | } |
| 826 | /* Check whether a frame must be created. */ | cc += 1 + 2 + cc[1]; |
| 827 | end = bracketend(cc); | break; |
| 828 | while (cc < end) | |
| 829 | { | case OP_RECURSE: |
| 830 | if (*cc == OP_SET_SOM || *cc == OP_CBRA || *cc == OP_CBRAPOS | if (common->has_set_som && !setsom_found) |
| 831 | || *cc == OP_SCBRA || *cc == OP_SCBRAPOS || *cc == OP_RECURSE) | { |
| 832 | needs_frame = TRUE; | length += 2; |
| 833 | cc = next_opcode(common, cc); | setsom_found = TRUE; |
| 834 | SLJIT_ASSERT(cc != NULL); | } |
| 835 | if (common->mark_ptr != 0 && !setmark_found) | |
| 836 | { | |
| 837 | length += 2; | |
| 838 | setmark_found = TRUE; | |
| 839 | } | } |
| 840 | cc += 1 + LINK_SIZE; | |
| 841 | break; | break; |
| 842 | ||
| 843 | case OP_CBRA: | case OP_CBRA: |
| 844 | case OP_CBRAPOS: | case OP_CBRAPOS: |
| 845 | case OP_SCBRA: | case OP_SCBRA: |
| 846 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| if (!needs_maxindex) | ||
| { | ||
| needs_maxindex = TRUE; | ||
| length += 2; | ||
| } | ||
| 847 | length += 3; | length += 3; |
| 848 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 849 | break; | break; |
| 850 | ||
| 851 | default: | default: |
| # | Line 761 while (cc < ccend) | Line 855 while (cc < ccend) |
| 855 | } | } |
| 856 | ||
| 857 | /* Possessive quantifiers can use a special case. */ | /* Possessive quantifiers can use a special case. */ |
| 858 | if (SLJIT_UNLIKELY(possessive) && !needs_frame && length == 3 + 2) | if (SLJIT_UNLIKELY(possessive) && length == 3) |
| 859 | return -1; | return -1; |
| 860 | ||
| 861 | if (length > 0) | if (length > 0) |
| 862 | return length + 2; | return length + 1; |
| 863 | return needs_frame ? 0 : -1; | return -1; |
| 864 | } | } |
| 865 | ||
| 866 | static void init_frame(compiler_common *common, uschar *cc, int stackpos, int stacktop, BOOL recursive) | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) |
| 867 | { | { |
| /* TMP2 must contain STACK_TOP - (-STACK(stackpos)) */ | ||
| 868 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 869 | uschar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc); |
| 870 | BOOL needs_maxindex = FALSE; | BOOL setsom_found = recursive; |
| 871 | BOOL setsom_found = FALSE; | BOOL setmark_found = recursive; |
| 872 | int offset; | int offset; |
| 873 | ||
| 874 | if (stackpos < stacktop) | /* >= 1 + shortest item size (2) */ |
| 875 | { | SLJIT_UNUSED_ARG(stacktop); |
| 876 | SLJIT_ASSERT(stackpos + 1 == stacktop); | SLJIT_ASSERT(stackpos >= stacktop + 2); |
| return; | ||
| } | ||
| 877 | ||
| 878 | stackpos = STACK(stackpos); | stackpos = STACK(stackpos); |
| OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD); | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD, TMP2, 0); | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacktop), TMP1, 0); | ||
| 879 | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) |
| 880 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
| 881 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
| # | Line 796 while (cc < ccend) | Line 883 while (cc < ccend) |
| 883 | switch(*cc) | switch(*cc) |
| 884 | { | { |
| 885 | case OP_SET_SOM: | case OP_SET_SOM: |
| 886 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
| 887 | if (!setsom_found) | if (!setsom_found) |
| 888 | { | { |
| 889 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
| # | Line 806 while (cc < ccend) | Line 893 while (cc < ccend) |
| 893 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_w); |
| 894 | setsom_found = TRUE; | setsom_found = TRUE; |
| 895 | } | } |
| 896 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | cc += 1; |
| 897 | break; | break; |
| 898 | ||
| 899 | case OP_ASSERT: | case OP_MARK: |
| 900 | case OP_ASSERT_NOT: | SLJIT_ASSERT(common->mark_ptr != 0); |
| 901 | case OP_ASSERTBACK: | if (!setmark_found) |
| 902 | case OP_ASSERTBACK_NOT: | { |
| 903 | case OP_ONCE: | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); |
| 904 | cc = bracketend(cc); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); |
| 905 | stackpos += (int)sizeof(sljit_w); | |
| 906 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
| 907 | stackpos += (int)sizeof(sljit_w); | |
| 908 | setmark_found = TRUE; | |
| 909 | } | |
| 910 | cc += 1 + 2 + cc[1]; | |
| 911 | break; | break; |
| 912 | ||
| 913 | case OP_CBRA: | case OP_RECURSE: |
| 914 | case OP_CBRAPOS: | if (common->has_set_som && !setsom_found) |
| case OP_SCBRA: | ||
| case OP_SCBRAPOS: | ||
| if (!needs_maxindex) | ||
| 915 | { | { |
| 916 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmaxindex); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
| 917 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | |
| 918 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_w); |
| 919 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, MAX_INDEX, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
| 920 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_w); |
| 921 | needs_maxindex = TRUE; | setsom_found = TRUE; |
| 922 | } | } |
| 923 | if (common->mark_ptr != 0 && !setmark_found) | |
| 924 | { | |
| 925 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
| 926 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | |
| 927 | stackpos += (int)sizeof(sljit_w); | |
| 928 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
| 929 | stackpos += (int)sizeof(sljit_w); | |
| 930 | setmark_found = TRUE; | |
| 931 | } | |
| 932 | cc += 1 + LINK_SIZE; | |
| 933 | break; | |
| 934 | ||
| 935 | case OP_CBRA: | |
| 936 | case OP_CBRAPOS: | |
| 937 | case OP_SCBRA: | |
| 938 | case OP_SCBRAPOS: | |
| 939 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
| 940 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
| 941 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_w); |
| # | Line 839 while (cc < ccend) | Line 946 while (cc < ccend) |
| 946 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
| 947 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_w); |
| 948 | ||
| 949 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 950 | break; | break; |
| 951 | ||
| 952 | default: | default: |
| # | Line 849 while (cc < ccend) | Line 956 while (cc < ccend) |
| 956 | } | } |
| 957 | ||
| 958 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); |
| 959 | SLJIT_ASSERT(stackpos == STACK(stacktop + 1)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
| 960 | } | } |
| 961 | ||
| 962 | static SLJIT_INLINE int get_localsize(compiler_common *common, uschar *cc, uschar *ccend) | static SLJIT_INLINE int get_localsize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
| 963 | { | { |
| 964 | int localsize = 2; | int localsize = 2; |
| 965 | uschar *alternative; | pcre_uchar *alternative; |
| 966 | /* Calculate the sum of the local variables. */ | /* Calculate the sum of the local variables. */ |
| 967 | while (cc < ccend) | while (cc < ccend) |
| 968 | { | { |
| # | Line 866 while (cc < ccend) | Line 973 while (cc < ccend) |
| 973 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
| 974 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| 975 | case OP_ONCE: | case OP_ONCE: |
| 976 | case OP_ONCE_NC: | |
| 977 | case OP_BRAPOS: | case OP_BRAPOS: |
| 978 | case OP_SBRA: | case OP_SBRA: |
| 979 | case OP_SBRAPOS: | case OP_SBRAPOS: |
| # | Line 877 while (cc < ccend) | Line 985 while (cc < ccend) |
| 985 | case OP_CBRA: | case OP_CBRA: |
| 986 | case OP_SCBRA: | case OP_SCBRA: |
| 987 | localsize++; | localsize++; |
| 988 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 989 | break; | break; |
| 990 | ||
| 991 | case OP_CBRAPOS: | case OP_CBRAPOS: |
| 992 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| 993 | localsize += 2; | localsize += 2; |
| 994 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 995 | break; | break; |
| 996 | ||
| 997 | case OP_COND: | case OP_COND: |
| # | Line 904 SLJIT_ASSERT(cc == ccend); | Line 1012 SLJIT_ASSERT(cc == ccend); |
| 1012 | return localsize; | return localsize; |
| 1013 | } | } |
| 1014 | ||
| 1015 | static void copy_locals(compiler_common *common, uschar *cc, uschar *ccend, | static void copy_locals(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, |
| 1016 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop) |
| 1017 | { | { |
| 1018 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| # | Line 913 int count; | Line 1021 int count; |
| 1021 | BOOL tmp1next = TRUE; | BOOL tmp1next = TRUE; |
| 1022 | BOOL tmp1empty = TRUE; | BOOL tmp1empty = TRUE; |
| 1023 | BOOL tmp2empty = TRUE; | BOOL tmp2empty = TRUE; |
| 1024 | uschar *alternative; | pcre_uchar *alternative; |
| 1025 | enum { | enum { |
| 1026 | start, | start, |
| 1027 | loop, | loop, |
| # | Line 948 while (status != end) | Line 1056 while (status != end) |
| 1056 | switch(status) | switch(status) |
| 1057 | { | { |
| 1058 | case start: | case start: |
| 1059 | SLJIT_ASSERT(save); | SLJIT_ASSERT(save && common->recursive_head != 0); |
| 1060 | count = 1; | count = 1; |
| 1061 | srcw[0] = RECURSIVE_HEAD; | srcw[0] = common->recursive_head; |
| 1062 | status = loop; | status = loop; |
| 1063 | break; | break; |
| 1064 | ||
| # | Line 968 while (status != end) | Line 1076 while (status != end) |
| 1076 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
| 1077 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| 1078 | case OP_ONCE: | case OP_ONCE: |
| 1079 | case OP_ONCE_NC: | |
| 1080 | case OP_BRAPOS: | case OP_BRAPOS: |
| 1081 | case OP_SBRA: | case OP_SBRA: |
| 1082 | case OP_SBRAPOS: | case OP_SBRAPOS: |
| 1083 | case OP_SCOND: | case OP_SCOND: |
| 1084 | count = 1; | count = 1; |
| 1085 | srcw[0] = PRIV(cc); | srcw[0] = PRIV_DATA(cc); |
| 1086 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
| 1087 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
| 1088 | break; | break; |
| # | Line 982 while (status != end) | Line 1091 while (status != end) |
| 1091 | case OP_SCBRA: | case OP_SCBRA: |
| 1092 | count = 1; | count = 1; |
| 1093 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
| 1094 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 1095 | break; | break; |
| 1096 | ||
| 1097 | case OP_CBRAPOS: | case OP_CBRAPOS: |
| 1098 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
| 1099 | count = 2; | count = 2; |
| 1100 | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
| 1101 | srcw[0] = PRIV(cc); | srcw[0] = PRIV_DATA(cc); |
| 1102 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
| 1103 | cc += 1 + LINK_SIZE + 2; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
| 1104 | break; | break; |
| 1105 | ||
| 1106 | case OP_COND: | case OP_COND: |
| # | Line 1000 while (status != end) | Line 1109 while (status != end) |
| 1109 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
| 1110 | { | { |
| 1111 | count = 1; | count = 1; |
| 1112 | srcw[0] = PRIV(cc); | srcw[0] = PRIV_DATA(cc); |
| 1113 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
| 1114 | } | } |
| 1115 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
| # | Line 1164 while (list_item) | Line 1273 while (list_item) |
| 1273 | case stack_alloc: | case stack_alloc: |
| 1274 | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
| 1275 | break; | break; |
| case max_index: | ||
| OP1(SLJIT_MOV, MAX_INDEX, 0, SLJIT_IMM, list_item->data); | ||
| break; | ||
| 1276 | } | } |
| 1277 | JUMPTO(SLJIT_JUMP, list_item->leave); | JUMPTO(SLJIT_JUMP, list_item->leave); |
| 1278 | list_item = list_item->next; | list_item = list_item->next; |
| # | Line 1175 while (list_item) | Line 1280 while (list_item) |
| 1280 | common->stubs = NULL; | common->stubs = NULL; |
| 1281 | } | } |
| 1282 | ||
| 1283 | static SLJIT_INLINE void decrease_call_count(compiler_common *common) | |
| 1284 | { | |
| 1285 | DEFINE_COMPILER; | |
| 1286 | ||
| 1287 | OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1); | |
| 1288 | add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); | |
| 1289 | } | |
| 1290 | ||
| 1291 | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) |
| 1292 | { | { |
| 1293 | /* May destroy all locals and registers except TMP2. */ | /* May destroy all locals and registers except TMP2. */ |
| # | Line 1204 struct sljit_label *loop; | Line 1317 struct sljit_label *loop; |
| 1317 | int i; | int i; |
| 1318 | /* At this point we can freely use all temporary registers. */ | /* At this point we can freely use all temporary registers. */ |
| 1319 | /* TMP1 returns with begin - 1. */ | /* TMP1 returns with begin - 1. */ |
| 1320 | OP1(SLJIT_MOV, MAX_INDEX, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); |
| OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_GENERAL_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, 1); | ||
| 1321 | if (length < 8) | if (length < 8) |
| 1322 | { | { |
| 1323 | for (i = 0; i < length; i++) | for (i = 0; i < length; i++) |
| # | Line 1222 else | Line 1334 else |
| 1334 | } | } |
| 1335 | } | } |
| 1336 | ||
| 1337 | static SLJIT_INLINE void copy_ovector(compiler_common *common) | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) |
| 1338 | { | { |
| 1339 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1340 | struct sljit_label *loop; | struct sljit_label *loop; |
| 1341 | struct sljit_jump *earlyexit; | struct sljit_jump *earlyexit; |
| 1342 | ||
| 1343 | /* At this point we can freely use all registers. */ | /* At this point we can freely use all registers. */ |
| 1344 | OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | |
| 1345 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); | |
| 1346 | ||
| 1347 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); |
| 1348 | if (common->mark_ptr != 0) | |
| 1349 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
| 1350 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); |
| 1351 | if (common->mark_ptr != 0) | |
| 1352 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0); | |
| 1353 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); |
| 1354 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); |
| 1355 | OP2(SLJIT_ADD, SLJIT_GENERAL_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START); | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START); |
| 1356 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
| 1357 | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); |
| 1358 | loop = LABEL(); | loop = LABEL(); |
| 1359 | OP2(SLJIT_SUB, SLJIT_GENERAL_REG2, 0, SLJIT_MEM1(SLJIT_GENERAL_REG1), 0, SLJIT_TEMPORARY_REG1, 0); | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0); |
| 1360 | OP2(SLJIT_ADD, SLJIT_GENERAL_REG1, 0, SLJIT_GENERAL_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); |
| 1361 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
| 1362 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_GENERAL_REG2, 0); | #ifdef COMPILE_PCRE16 |
| 1363 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); | |
| 1364 | #endif | |
| 1365 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | |
| 1366 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); |
| 1367 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
| 1368 | JUMPHERE(earlyexit); | JUMPHERE(earlyexit); |
| 1369 | ||
| 1370 | /* Calculate the return value, which is the maximum ovector value. */ | |
| 1371 | if (topbracket > 1) | |
| 1372 | { | |
| 1373 | OP2(SLJIT_ADD, SLJIT_TEMPORARY_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START + topbracket * 2 * sizeof(sljit_w)); | |
| 1374 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); | |
| 1375 | ||
| 1376 | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ | |
| 1377 | loop = LABEL(); | |
| 1378 | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w))); | |
| 1379 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | |
| 1380 | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); | |
| 1381 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); | |
| 1382 | } | |
| 1383 | else | |
| 1384 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | |
| 1385 | } | |
| 1386 | ||
| 1387 | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *leave) | |
| 1388 | { | |
| 1389 | DEFINE_COMPILER; | |
| 1390 | ||
| 1391 | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); | |
| 1392 | SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | |
| 1393 | ||
| 1394 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | |
| 1395 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | |
| 1396 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | |
| 1397 | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, leave); | |
| 1398 | ||
| 1399 | /* Store match begin and end. */ | |
| 1400 | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); | |
| 1401 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | |
| 1402 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
| 1403 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); | |
| 1404 | #ifdef COMPILE_PCRE16 | |
| 1405 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); | |
| 1406 | #endif | |
| 1407 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | |
| 1408 | ||
| 1409 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); | |
| 1410 | #ifdef COMPILE_PCRE16 | |
| 1411 | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | |
| 1412 | #endif | |
| 1413 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); | |
| 1414 | ||
| 1415 | JUMPTO(SLJIT_JUMP, leave); | |
| 1416 | } | |
| 1417 | ||
| 1418 | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) | |
| 1419 | { | |
| 1420 | /* May destroy TMP1. */ | |
| 1421 | DEFINE_COMPILER; | |
| 1422 | struct sljit_jump *jump; | |
| 1423 | ||
| 1424 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
| 1425 | { | |
| 1426 | /* The value of -1 must be kept for start_used_ptr! */ | |
| 1427 | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); | |
| 1428 | /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting | |
| 1429 | is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ | |
| 1430 | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); | |
| 1431 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1432 | JUMPHERE(jump); | |
| 1433 | } | |
| 1434 | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) | |
| 1435 | { | |
| 1436 | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1437 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1438 | JUMPHERE(jump); | |
| 1439 | } | |
| 1440 | } | } |
| 1441 | ||
| 1442 | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, uschar* cc) | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) |
| 1443 | { | { |
| 1444 | /* Detects if the character has an othercase. */ | /* Detects if the character has an othercase. */ |
| 1445 | unsigned int c; | unsigned int c; |
| 1446 | ||
| 1447 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1448 | if (common->utf8) | if (common->utf) |
| 1449 | { | { |
| 1450 | GETCHAR(c, cc); | GETCHAR(c, cc); |
| 1451 | if (c > 127) | if (c > 127) |
| # | Line 1263 if (common->utf8) | Line 1456 if (common->utf8) |
| 1456 | return FALSE; | return FALSE; |
| 1457 | #endif | #endif |
| 1458 | } | } |
| 1459 | #ifndef COMPILE_PCRE8 | |
| 1460 | return common->fcc[c] != c; | |
| 1461 | #endif | |
| 1462 | } | } |
| 1463 | else | else |
| 1464 | #endif | #endif |
| 1465 | c = *cc; | c = *cc; |
| 1466 | return common->fcc[c] != c; | return MAX_255(c) ? common->fcc[c] != c : FALSE; |
| 1467 | } | } |
| 1468 | ||
| 1469 | static SLJIT_INLINE unsigned int char_othercase(compiler_common *common, unsigned int c) | static SLJIT_INLINE unsigned int char_othercase(compiler_common *common, unsigned int c) |
| 1470 | { | { |
| 1471 | /* Returns with the othercase. */ | /* Returns with the othercase. */ |
| 1472 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1473 | if (common->utf8 && c > 127) | if (common->utf && c > 127) |
| 1474 | { | { |
| 1475 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 1476 | return UCD_OTHERCASE(c); | return UCD_OTHERCASE(c); |
| # | Line 1283 if (common->utf8 && c > 127) | Line 1479 if (common->utf8 && c > 127) |
| 1479 | #endif | #endif |
| 1480 | } | } |
| 1481 | #endif | #endif |
| 1482 | return common->fcc[c]; | return TABLE_GET(c, common->fcc, c); |
| 1483 | } | } |
| 1484 | ||
| 1485 | static unsigned int char_get_othercase_bit(compiler_common *common, uschar* cc) | static unsigned int char_get_othercase_bit(compiler_common *common, pcre_uchar* cc) |
| 1486 | { | { |
| 1487 | /* Detects if the character and its othercase has only 1 bit difference. */ | /* Detects if the character and its othercase has only 1 bit difference. */ |
| 1488 | unsigned int c, oc, bit; | unsigned int c, oc, bit; |
| 1489 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
| 1490 | int n; | int n; |
| 1491 | #endif | #endif |
| 1492 | ||
| 1493 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1494 | if (common->utf8) | if (common->utf) |
| 1495 | { | { |
| 1496 | GETCHAR(c, cc); | GETCHAR(c, cc); |
| 1497 | if (c <= 127) | if (c <= 127) |
| # | Line 1312 if (common->utf8) | Line 1508 if (common->utf8) |
| 1508 | else | else |
| 1509 | { | { |
| 1510 | c = *cc; | c = *cc; |
| 1511 | oc = common->fcc[c]; | oc = TABLE_GET(c, common->fcc, c); |
| 1512 | } | } |
| 1513 | #else | #else |
| 1514 | c = *cc; | c = *cc; |
| 1515 | oc = common->fcc[c]; | oc = TABLE_GET(c, common->fcc, c); |
| 1516 | #endif | #endif |
| 1517 | ||
| 1518 | SLJIT_ASSERT(c != oc); | SLJIT_ASSERT(c != oc); |
| # | Line 1330 if (c <= 127 && bit == 0x20) | Line 1526 if (c <= 127 && bit == 0x20) |
| 1526 | if (!ispowerof2(bit)) | if (!ispowerof2(bit)) |
| 1527 | return 0; | return 0; |
| 1528 | ||
| 1529 | #ifdef SUPPORT_UTF8 | #ifdef COMPILE_PCRE8 |
| 1530 | if (common->utf8 && c > 127) | |
| 1531 | #ifdef SUPPORT_UTF | |
| 1532 | if (common->utf && c > 127) | |
| 1533 | { | { |
| 1534 | n = _pcre_utf8_table4[*cc & 0x3f]; | n = GET_EXTRALEN(*cc); |
| 1535 | while ((bit & 0x3f) == 0) | while ((bit & 0x3f) == 0) |
| 1536 | { | { |
| 1537 | n--; | n--; |
| # | Line 1341 if (common->utf8 && c > 127) | Line 1539 if (common->utf8 && c > 127) |
| 1539 | } | } |
| 1540 | return (n << 8) | bit; | return (n << 8) | bit; |
| 1541 | } | } |
| 1542 | #endif | #endif /* SUPPORT_UTF */ |
| 1543 | return (0 << 8) | bit; | return (0 << 8) | bit; |
| 1544 | ||
| 1545 | #else /* COMPILE_PCRE8 */ | |
| 1546 | ||
| 1547 | #ifdef COMPILE_PCRE16 | |
| 1548 | #ifdef SUPPORT_UTF | |
| 1549 | if (common->utf && c > 65535) | |
| 1550 | { | |
| 1551 | if (bit >= (1 << 10)) | |
| 1552 | bit >>= 10; | |
| 1553 | else | |
| 1554 | return (bit < 256) ? ((2 << 8) | bit) : ((3 << 8) | (bit >> 8)); | |
| 1555 | } | |
| 1556 | #endif /* SUPPORT_UTF */ | |
| 1557 | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); | |
| 1558 | #endif /* COMPILE_PCRE16 */ | |
| 1559 | ||
| 1560 | #endif /* COMPILE_PCRE8 */ | |
| 1561 | } | |
| 1562 | ||
| 1563 | static void check_partial(compiler_common *common, BOOL force) | |
| 1564 | { | |
| 1565 | /* Checks whether a partial matching is occured. Does not modify registers. */ | |
| 1566 | DEFINE_COMPILER; | |
| 1567 | struct sljit_jump *jump = NULL; | |
| 1568 | ||
| 1569 | SLJIT_ASSERT(!force || common->mode != JIT_COMPILE); | |
| 1570 | ||
| 1571 | if (common->mode == JIT_COMPILE) | |
| 1572 | return; | |
| 1573 | ||
| 1574 | if (!force) | |
| 1575 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1576 | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
| 1577 | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); | |
| 1578 | ||
| 1579 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
| 1580 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | |
| 1581 | else | |
| 1582 | { | |
| 1583 | if (common->partialmatchlabel != NULL) | |
| 1584 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
| 1585 | else | |
| 1586 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
| 1587 | } | |
| 1588 | ||
| 1589 | if (jump != NULL) | |
| 1590 | JUMPHERE(jump); | |
| 1591 | } | |
| 1592 | ||
| 1593 | static struct sljit_jump *check_str_end(compiler_common *common) | |
| 1594 | { | |
| 1595 | /* Does not affect registers. Usually used in a tight spot. */ | |
| 1596 | DEFINE_COMPILER; | |
| 1597 | struct sljit_jump *jump; | |
| 1598 | struct sljit_jump *nohit; | |
| 1599 | struct sljit_jump *return_value; | |
| 1600 | ||
| 1601 | if (common->mode == JIT_COMPILE) | |
| 1602 | return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
| 1603 | ||
| 1604 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | |
| 1605 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
| 1606 | { | |
| 1607 | nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1608 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | |
| 1609 | JUMPHERE(nohit); | |
| 1610 | return_value = JUMP(SLJIT_JUMP); | |
| 1611 | } | |
| 1612 | else | |
| 1613 | { | |
| 1614 | return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
| 1615 | if (common->partialmatchlabel != NULL) | |
| 1616 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
| 1617 | else | |
| 1618 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
| 1619 | } | |
| 1620 | JUMPHERE(jump); | |
| 1621 | return return_value; | |
| 1622 | } | } |
| 1623 | ||
| 1624 | static SLJIT_INLINE void check_input_end(compiler_common *common, jump_list **fallbacks) | static void fallback_at_str_end(compiler_common *common, jump_list **fallbacks) |
| 1625 | { | { |
| 1626 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1627 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | struct sljit_jump *jump; |
| 1628 | ||
| 1629 | if (common->mode == JIT_COMPILE) | |
| 1630 | { | |
| 1631 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | |
| 1632 | return; | |
| 1633 | } | |
| 1634 | ||
| 1635 | /* Partial matching mode. */ | |
| 1636 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | |
| 1637 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); | |
| 1638 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
| 1639 | { | |
| 1640 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | |
| 1641 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | |
| 1642 | } | |
| 1643 | else | |
| 1644 | { | |
| 1645 | if (common->partialmatchlabel != NULL) | |
| 1646 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
| 1647 | else | |
| 1648 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
| 1649 | } | |
| 1650 | JUMPHERE(jump); | |
| 1651 | } | } |
| 1652 | ||
| 1653 | static void read_char(compiler_common *common) | static void read_char(compiler_common *common) |
| # | Line 1356 static void read_char(compiler_common *c | Line 1655 static void read_char(compiler_common *c |
| 1655 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, updates STR_PTR. |
| 1656 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
| 1657 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1658 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1659 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 1660 | #endif | #endif |
| 1661 | ||
| 1662 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 1663 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1664 | if (common->utf8) | if (common->utf) |
| 1665 | { | { |
| 1666 | /* Should not found a value between 128 and 192 here. */ | #ifdef COMPILE_PCRE8 |
| 1667 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 192); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
| 1668 | add_jump(compiler, &common->utf8readchar, JUMP(SLJIT_FAST_CALL)); | #else |
| 1669 | #ifdef COMPILE_PCRE16 | |
| 1670 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 1671 | #endif | |
| 1672 | #endif /* COMPILE_PCRE8 */ | |
| 1673 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | |
| 1674 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1675 | } | } |
| 1676 | #endif | #endif |
| 1677 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1678 | } | } |
| 1679 | ||
| 1680 | static void peek_char(compiler_common *common) | static void peek_char(compiler_common *common) |
| # | Line 1378 static void peek_char(compiler_common *c | Line 1682 static void peek_char(compiler_common *c |
| 1682 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
| 1683 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
| 1684 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1685 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1686 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 1687 | #endif | #endif |
| 1688 | ||
| 1689 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 1690 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1691 | if (common->utf8) | if (common->utf) |
| 1692 | { | { |
| 1693 | /* Should not found a value between 128 and 192 here. */ | #ifdef COMPILE_PCRE8 |
| 1694 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 192); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
| 1695 | add_jump(compiler, &common->utf8readchar, JUMP(SLJIT_FAST_CALL)); | #else |
| 1696 | #ifdef COMPILE_PCRE16 | |
| 1697 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 1698 | #endif | |
| 1699 | #endif /* COMPILE_PCRE8 */ | |
| 1700 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | |
| 1701 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
| 1702 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1703 | } | } |
| # | Line 1399 static void read_char8_type(compiler_com | Line 1708 static void read_char8_type(compiler_com |
| 1708 | { | { |
| 1709 | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ |
| 1710 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1711 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 |
| 1712 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 1713 | #endif | #endif |
| 1714 | ||
| 1715 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1716 | if (common->utf8) | if (common->utf) |
| 1717 | { | { |
| 1718 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
| 1719 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1720 | #ifdef COMPILE_PCRE8 | |
| 1721 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
| 1722 | it is a clever early read in most cases. */ | it is needed in most cases. */ |
| 1723 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
| 1724 | /* Should not found a value between 128 and 192 here. */ | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
| 1725 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 192); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
| add_jump(compiler, &common->utf8readtype8, JUMP(SLJIT_FAST_CALL)); | ||
| 1726 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1727 | #else | |
| 1728 | #ifdef COMPILE_PCRE16 | |
| 1729 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | |
| 1730 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
| 1731 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
| 1732 | JUMPHERE(jump); | |
| 1733 | /* Skip low surrogate if necessary. */ | |
| 1734 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | |
| 1735 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); | |
| 1736 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | |
| 1737 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | |
| 1738 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
| 1739 | #endif | |
| 1740 | #endif /* COMPILE_PCRE8 */ | |
| 1741 | return; | return; |
| 1742 | } | } |
| 1743 | #endif | #endif |
| 1744 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
| 1745 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1746 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); | #ifdef COMPILE_PCRE16 |
| 1747 | /* The ctypes array contains only 256 values. */ | |
| 1748 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | |
| 1749 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
| 1750 | #endif | |
| 1751 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
| 1752 | #ifdef COMPILE_PCRE16 | |
| 1753 | JUMPHERE(jump); | |
| 1754 | #endif | |
| 1755 | } | } |
| 1756 | ||
| 1757 | static void skip_char_back(compiler_common *common) | static void skip_char_back(compiler_common *common) |
| 1758 | { | { |
| 1759 | /* Goes one character back. Only affects STR_PTR. Does not check begin. */ | /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */ |
| 1760 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1761 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
| 1762 | struct sljit_label *label; | struct sljit_label *label; |
| 1763 | ||
| 1764 | if (common->utf8) | if (common->utf) |
| 1765 | { | { |
| 1766 | label = LABEL(); | label = LABEL(); |
| 1767 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); |
| 1768 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1769 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); |
| 1770 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
| 1771 | return; | return; |
| 1772 | } | } |
| 1773 | #endif | #endif |
| 1774 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 |
| 1775 | if (common->utf) | |
| 1776 | { | |
| 1777 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); | |
| 1778 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
| 1779 | /* Skip low surrogate if necessary. */ | |
| 1780 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
| 1781 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); | |
| 1782 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | |
| 1783 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 1784 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 1785 | return; | |
| 1786 | } | |
| 1787 | #endif | |
| 1788 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
| 1789 | } | } |
| 1790 | ||
| 1791 | static void check_newlinechar(compiler_common *common, int nltype, jump_list **fallbacks, BOOL jumpiftrue) | static void check_newlinechar(compiler_common *common, int nltype, jump_list **fallbacks, BOOL jumpiftrue) |
| # | Line 1463 else if (nltype == NLTYPE_ANYCRLF) | Line 1808 else if (nltype == NLTYPE_ANYCRLF) |
| 1808 | } | } |
| 1809 | else | else |
| 1810 | { | { |
| 1811 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline <= 255); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
| 1812 | add_jump(compiler, fallbacks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, fallbacks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
| 1813 | } | } |
| 1814 | } | } |
| 1815 | ||
| 1816 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1817 | static void do_utf8readchar(compiler_common *common) | |
| 1818 | #ifdef COMPILE_PCRE8 | |
| 1819 | static void do_utfreadchar(compiler_common *common) | |
| 1820 | { | { |
| 1821 | /* Fast decoding an utf8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
| 1822 | of the character (>= 192). Return char value in TMP1, length - 1 in TMP2. */ | of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */ |
| 1823 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1824 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 1825 | ||
| # | Line 1480 sljit_emit_fast_enter(compiler, RETURN_A | Line 1827 sljit_emit_fast_enter(compiler, RETURN_A |
| 1827 | /* Searching for the first zero. */ | /* Searching for the first zero. */ |
| 1828 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); |
| 1829 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
| 1830 | /* 2 byte sequence */ | /* Two byte sequence. */ |
| 1831 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
| 1832 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1833 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); |
| 1834 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
| 1835 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1836 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1837 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1838 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 1839 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1840 | ||
| 1841 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); |
| 1842 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
| 1843 | /* 3 byte sequence */ | /* Three byte sequence. */ |
| 1844 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
| 1845 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); |
| 1846 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); |
| 1847 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1848 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
| 1849 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1850 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 2); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); |
| 1851 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 2); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 1852 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1853 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1854 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 2); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 1855 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 1856 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1857 | ||
| 1858 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x08); | /* Four byte sequence. */ |
| 1859 | jump = JUMP(SLJIT_C_NOT_ZERO); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
| /* 4 byte sequence */ | ||
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | ||
| 1860 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); |
| 1861 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); |
| 1862 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1863 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); |
| 1864 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1865 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 2); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); |
| 1866 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1867 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
| 1868 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1869 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 3); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3)); |
| 1870 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 3); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); |
| 1871 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
| 1872 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1873 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 3); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); |
| sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | ||
| JUMPHERE(jump); | ||
| /* 5 byte sequence */ | ||
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | ||
| OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x03); | ||
| OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 24); | ||
| OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | ||
| OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 18); | ||
| OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 2); | ||
| OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | ||
| OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); | ||
| OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 3); | ||
| OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | ||
| OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | ||
| OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 4); | ||
| OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 4); | ||
| OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | ||
| OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
| OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 4); | ||
| 1874 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 1875 | } | } |
| 1876 | ||
| 1877 | static void do_utf8readtype8(compiler_common *common) | static void do_utfreadtype8(compiler_common *common) |
| 1878 | { | { |
| 1879 | /* Fast decoding an utf8 character type. TMP2 contains the first byte | /* Fast decoding a UTF-8 character type. TMP2 contains the first byte |
| 1880 | of the character (>= 192) and TMP1 is destroyed. Return value in TMP1. */ | of the character (>= 0xc0). Return value in TMP1. */ |
| 1881 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 1882 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 1883 | struct sljit_jump *compare; | struct sljit_jump *compare; |
| # | Line 1564 sljit_emit_fast_enter(compiler, RETURN_A | Line 1886 sljit_emit_fast_enter(compiler, RETURN_A |
| 1886 | ||
| 1887 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); |
| 1888 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
| 1889 | /* 2 byte sequence */ | /* Two byte sequence. */ |
| 1890 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 1891 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 1892 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); |
| 1893 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
| 1894 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
| # | Line 1581 sljit_emit_fast_return(compiler, RETURN_ | Line 1903 sljit_emit_fast_return(compiler, RETURN_ |
| 1903 | JUMPHERE(jump); | JUMPHERE(jump); |
| 1904 | ||
| 1905 | /* We only have types for characters less than 256. */ | /* We only have types for characters less than 256. */ |
| 1906 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_w)_pcre_utf8_char_sizes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(utf8_table4) - 0xc0); |
| OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | ||
| 1907 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 1908 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
| 1909 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 1910 | } | } |
| 1911 | ||
| 1912 | #endif | #else /* COMPILE_PCRE8 */ |
| 1913 | ||
| 1914 | #ifdef COMPILE_PCRE16 | |
| 1915 | static void do_utfreadchar(compiler_common *common) | |
| 1916 | { | |
| 1917 | /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char | |
| 1918 | of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */ | |
| 1919 | DEFINE_COMPILER; | |
| 1920 | struct sljit_jump *jump; | |
| 1921 | ||
| 1922 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | |
| 1923 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); | |
| 1924 | /* Do nothing, only return. */ | |
| 1925 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
| 1926 | ||
| 1927 | JUMPHERE(jump); | |
| 1928 | /* Combine two 16 bit characters. */ | |
| 1929 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
| 1930 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
| 1931 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | |
| 1932 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10); | |
| 1933 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff); | |
| 1934 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
| 1935 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); | |
| 1936 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | |
| 1937 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
| 1938 | } | |
| 1939 | #endif /* COMPILE_PCRE16 */ | |
| 1940 | ||
| 1941 | #endif /* COMPILE_PCRE8 */ | |
| 1942 | ||
| 1943 | #endif /* SUPPORT_UTF */ | |
| 1944 | ||
| 1945 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 1946 | ||
| # | Line 1606 SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si | Line 1958 SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si |
| 1958 | ||
| 1959 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); |
| 1960 | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
| 1961 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_w)_pcre_ucd_stage1); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(ucd_stage1)); |
| 1962 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); |
| 1963 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
| 1964 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); |
| 1965 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)_pcre_ucd_stage2); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2)); |
| 1966 | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); |
| 1967 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)_pcre_ucd_records + SLJIT_OFFSETOF(ucd_record, chartype)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); |
| 1968 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
| 1969 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 1970 | } | } |
| # | Line 1626 struct sljit_label *newlinelabel = NULL; | Line 1978 struct sljit_label *newlinelabel = NULL; |
| 1978 | struct sljit_jump *start; | struct sljit_jump *start; |
| 1979 | struct sljit_jump *end = NULL; | struct sljit_jump *end = NULL; |
| 1980 | struct sljit_jump *nl = NULL; | struct sljit_jump *nl = NULL; |
| 1981 | #ifdef SUPPORT_UTF | |
| 1982 | struct sljit_jump *singlechar; | |
| 1983 | #endif | |
| 1984 | jump_list *newline = NULL; | jump_list *newline = NULL; |
| 1985 | BOOL newlinecheck = FALSE; | BOOL newlinecheck = FALSE; |
| 1986 | BOOL readbyte = FALSE; | BOOL readuchar = FALSE; |
| 1987 | ||
| 1988 | if (!(hascrorlf || firstline) && (common->nltype == NLTYPE_ANY || | if (!(hascrorlf || firstline) && (common->nltype == NLTYPE_ANY || |
| 1989 | common->nltype == NLTYPE_ANYCRLF || common->newline > 255)) | common->nltype == NLTYPE_ANYCRLF || common->newline > 255)) |
| # | Line 1637 if (!(hascrorlf || firstline) && (common | Line 1992 if (!(hascrorlf || firstline) && (common |
| 1992 | if (firstline) | if (firstline) |
| 1993 | { | { |
| 1994 | /* Search for the end of the first line. */ | /* Search for the end of the first line. */ |
| 1995 | SLJIT_ASSERT(common->first_line_end != 0); | |
| 1996 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); |
| 1997 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_END, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_END, 0); |
| 1998 | ||
| 1999 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| 2000 | { | { |
| 2001 | mainloop = LABEL(); | mainloop = LABEL(); |
| 2002 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2003 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2004 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -1); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
| 2005 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 2006 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); |
| 2007 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); |
| 2008 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2009 | } | } |
| 2010 | else | else |
| 2011 | { | { |
| 2012 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2013 | mainloop = LABEL(); | mainloop = LABEL(); |
| 2014 | /* Continual stores does not cause data dependency. */ | /* Continual stores does not cause data dependency. */ |
| 2015 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); |
| 2016 | read_char(common); | read_char(common); |
| 2017 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
| 2018 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
| 2019 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); |
| 2020 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
| 2021 | } | } |
| 2022 | ||
| # | Line 1673 start = JUMP(SLJIT_JUMP); | Line 2029 start = JUMP(SLJIT_JUMP); |
| 2029 | if (newlinecheck) | if (newlinecheck) |
| 2030 | { | { |
| 2031 | newlinelabel = LABEL(); | newlinelabel = LABEL(); |
| 2032 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2033 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2034 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 2035 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); |
| 2036 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); |
| 2037 | #ifdef COMPILE_PCRE16 | |
| 2038 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 2039 | #endif | |
| 2040 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 2041 | nl = JUMP(SLJIT_JUMP); | nl = JUMP(SLJIT_JUMP); |
| 2042 | } | } |
| # | Line 1685 if (newlinecheck) | Line 2044 if (newlinecheck) |
| 2044 | mainloop = LABEL(); | mainloop = LABEL(); |
| 2045 | ||
| 2046 | /* Increasing the STR_PTR here requires one less jump in the most common case. */ | /* Increasing the STR_PTR here requires one less jump in the most common case. */ |
| 2047 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2048 | if (common->utf8) readbyte = TRUE; | if (common->utf) readuchar = TRUE; |
| 2049 | #endif | #endif |
| 2050 | if (newlinecheck) readbyte = TRUE; | if (newlinecheck) readuchar = TRUE; |
| 2051 | ||
| 2052 | if (readbyte) | if (readuchar) |
| 2053 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 2054 | ||
| 2055 | if (newlinecheck) | if (newlinecheck) |
| 2056 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); |
| 2057 | ||
| 2058 | #ifdef SUPPORT_UTF8 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2059 | if (common->utf8) | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
| 2060 | if (common->utf) | |
| 2061 | { | { |
| 2062 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)_pcre_utf8_char_sizes); | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
| 2063 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | |
| 2064 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 2065 | JUMPHERE(singlechar); | |
| 2066 | } | |
| 2067 | #endif | |
| 2068 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
| 2069 | if (common->utf) | |
| 2070 | { | |
| 2071 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 2072 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
| 2073 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 2074 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | |
| 2075 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 2076 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 2077 | JUMPHERE(singlechar); | |
| 2078 | } | } |
| else | ||
| OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | ||
| #else | ||
| OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | ||
| 2079 | #endif | #endif |
| 2080 | JUMPHERE(start); | JUMPHERE(start); |
| 2081 | ||
| # | Line 1718 if (newlinecheck) | Line 2088 if (newlinecheck) |
| 2088 | return mainloop; | return mainloop; |
| 2089 | } | } |
| 2090 | ||
| 2091 | static SLJIT_INLINE void fast_forward_first_byte(compiler_common *common, pcre_uint16 firstbyte, BOOL firstline) | static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline) |
| 2092 | { | { |
| 2093 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 2094 | struct sljit_label *start; | struct sljit_label *start; |
| 2095 | struct sljit_jump *leave; | struct sljit_jump *leave; |
| 2096 | struct sljit_jump *found; | struct sljit_jump *found; |
| 2097 | pcre_uint16 oc, bit; | pcre_uchar oc, bit; |
| 2098 | ||
| 2099 | if (firstline) | if (firstline) |
| 2100 | { | { |
| 2101 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); |
| 2102 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
| 2103 | } | } |
| 2104 | ||
| 2105 | start = LABEL(); | start = LABEL(); |
| 2106 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2107 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 2108 | ||
| 2109 | if ((firstbyte & REQ_CASELESS) == 0) | oc = first_char; |
| 2110 | found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, firstbyte & 0xff); | if (caseless) |
| 2111 | { | |
| 2112 | oc = TABLE_GET(first_char, common->fcc, first_char); | |
| 2113 | #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) | |
| 2114 | if (first_char > 127 && common->utf) | |
| 2115 | oc = UCD_OTHERCASE(first_char); | |
| 2116 | #endif | |
| 2117 | } | |
| 2118 | if (first_char == oc) | |
| 2119 | found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, first_char); | |
| 2120 | else | else |
| 2121 | { | { |
| 2122 | firstbyte &= 0xff; | bit = first_char ^ oc; |
| oc = common->fcc[firstbyte]; | ||
| bit = firstbyte ^ oc; | ||
| 2123 | if (ispowerof2(bit)) | if (ispowerof2(bit)) |
| 2124 | { | { |
| 2125 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
| 2126 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, firstbyte | bit); | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); |
| 2127 | } | } |
| 2128 | else | else |
| 2129 | { | { |
| 2130 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, firstbyte); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); |
| 2131 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
| 2132 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); |
| 2133 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
| # | Line 1758 else | Line 2135 else |
| 2135 | } | } |
| 2136 | } | } |
| 2137 | ||
| 2138 | #ifdef SUPPORT_UTF8 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2139 | if (common->utf8) | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
| 2140 | if (common->utf) | |
| 2141 | { | { |
| 2142 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)_pcre_utf8_char_sizes); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
| 2143 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | |
| 2144 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 2145 | } | |
| 2146 | #endif | |
| 2147 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
| 2148 | if (common->utf) | |
| 2149 | { | |
| 2150 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | |
| 2151 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
| 2152 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 2153 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | |
| 2154 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 2155 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 2156 | } | } |
| else | ||
| OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | ||
| #else | ||
| OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | ||
| 2157 | #endif | #endif |
| 2158 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
| 2159 | JUMPHERE(found); | JUMPHERE(found); |
| # | Line 1791 jump_list *newline = NULL; | Line 2177 jump_list *newline = NULL; |
| 2177 | if (firstline) | if (firstline) |
| 2178 | { | { |
| 2179 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); |
| 2180 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
| 2181 | } | } |
| 2182 | ||
| 2183 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| # | Line 1802 if (common->nltype == NLTYPE_FIXED && co | Line 2188 if (common->nltype == NLTYPE_FIXED && co |
| 2188 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
| 2189 | firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); | firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); |
| 2190 | ||
| 2191 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 2192 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); |
| 2193 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); |
| 2194 | #ifdef COMPILE_PCRE16 | |
| 2195 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | |
| 2196 | #endif | |
| 2197 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
| 2198 | ||
| 2199 | loop = LABEL(); | loop = LABEL(); |
| 2200 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2201 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2202 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -2); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
| 2203 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), -1); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
| 2204 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); |
| 2205 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); |
| 2206 | ||
| # | Line 1842 if (common->nltype == NLTYPE_ANY || comm | Line 2231 if (common->nltype == NLTYPE_ANY || comm |
| 2231 | leave = JUMP(SLJIT_JUMP); | leave = JUMP(SLJIT_JUMP); |
| 2232 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
| 2233 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2234 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 2235 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); |
| 2236 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); |
| 2237 | #ifdef COMPILE_PCRE16 | |
| 2238 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 2239 | #endif | |
| 2240 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 2241 | JUMPHERE(notfoundnl); | JUMPHERE(notfoundnl); |
| 2242 | JUMPHERE(leave); | JUMPHERE(leave); |
| # | Line 1862 DEFINE_COMPILER; | Line 2254 DEFINE_COMPILER; |
| 2254 | struct sljit_label *start; | struct sljit_label *start; |
| 2255 | struct sljit_jump *leave; | struct sljit_jump *leave; |
| 2256 | struct sljit_jump *found; | struct sljit_jump *found; |
| 2257 | #ifndef COMPILE_PCRE8 | |
| 2258 | struct sljit_jump *jump; | |
| 2259 | #endif | |
| 2260 | ||
| 2261 | if (firstline) | if (firstline) |
| 2262 | { | { |
| 2263 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); |
| 2264 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
| 2265 | } | } |
| 2266 | ||
| 2267 | start = LABEL(); | start = LABEL(); |
| 2268 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 2269 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 2270 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2271 | if (common->utf8) | if (common->utf) |
| 2272 | OP1(SLJIT_MOV_UB, TMP3, 0, SLJIT_MEM1(TMP1), (sljit_w)_pcre_utf8_char_sizes); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
| 2273 | #endif | |
| 2274 | #ifndef COMPILE_PCRE8 | |
| 2275 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); | |
| 2276 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); | |
| 2277 | JUMPHERE(jump); | |
| 2278 | #endif | #endif |
| 2279 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
| 2280 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
| # | Line 1883 OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TM | Line 2283 OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TM |
| 2283 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); |
| 2284 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
| 2285 | ||
| 2286 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2287 | if (common->utf8) | if (common->utf) |
| 2288 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
| 2289 | else | #endif |
| 2290 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2291 | #else | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
| 2292 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | if (common->utf) |
| 2293 | { | |
| 2294 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | |
| 2295 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | |
| 2296 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 2297 | } | |
| 2298 | #endif | |
| 2299 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
| 2300 | if (common->utf) | |
| 2301 | { | |
| 2302 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | |
| 2303 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
| 2304 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 2305 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | |
| 2306 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 2307 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 2308 | } | |
| 2309 | #endif | #endif |
| 2310 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
| 2311 | JUMPHERE(found); | JUMPHERE(found); |
| # | Line 1899 if (firstline) | Line 2315 if (firstline) |
| 2315 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); |
| 2316 | } | } |
| 2317 | ||
| 2318 | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uint16 reqbyte, BOOL has_firstbyte) | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) |
| 2319 | { | { |
| 2320 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 2321 | struct sljit_label *loop; | struct sljit_label *loop; |
| # | Line 1908 struct sljit_jump *alreadyfound; | Line 2324 struct sljit_jump *alreadyfound; |
| 2324 | struct sljit_jump *found; | struct sljit_jump *found; |
| 2325 | struct sljit_jump *foundoc = NULL; | struct sljit_jump *foundoc = NULL; |
| 2326 | struct sljit_jump *notfound; | struct sljit_jump *notfound; |
| 2327 | pcre_uint16 oc, bit; | pcre_uchar oc, bit; |
| 2328 | ||
| 2329 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), REQ_BYTE_PTR); | SLJIT_ASSERT(common->req_char_ptr != 0); |
| 2330 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); | |
| 2331 | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); |
| 2332 | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); |
| 2333 | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); |
| 2334 | ||
| 2335 | if (has_firstbyte) | if (has_firstchar) |
| 2336 | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2337 | else | else |
| 2338 | OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); | OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); |
| 2339 | ||
| 2340 | loop = LABEL(); | loop = LABEL(); |
| 2341 | notfound = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, STR_END, 0); | notfound = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, STR_END, 0); |
| 2342 | ||
| 2343 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
| 2344 | if ((reqbyte & REQ_CASELESS) == 0) | oc = req_char; |
| 2345 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, reqbyte & 0xff); | if (caseless) |
| 2346 | { | |
| 2347 | oc = TABLE_GET(req_char, common->fcc, req_char); | |
| 2348 | #if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) | |
| 2349 | if (req_char > 127 && common->utf) | |
| 2350 | oc = UCD_OTHERCASE(req_char); | |
| 2351 | #endif | |
| 2352 | } | |
| 2353 | if (req_char == oc) | |
| 2354 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); | |
| 2355 | else | else |
| 2356 | { | { |
| 2357 | reqbyte &= 0xff; | bit = req_char ^ oc; |
| oc = common->fcc[reqbyte]; | ||
| bit = reqbyte ^ oc; | ||
| 2358 | if (ispowerof2(bit)) | if (ispowerof2(bit)) |
| 2359 | { | { |
| 2360 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
| 2361 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, reqbyte | bit); | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); |
| 2362 | } | } |
| 2363 | else | else |
| 2364 | { | { |
| 2365 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, reqbyte); | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); |
| 2366 | foundoc = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, oc); | foundoc = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, oc); |
| 2367 | } | } |
| 2368 | } | } |
| 2369 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2370 | JUMPTO(SLJIT_JUMP, loop); | JUMPTO(SLJIT_JUMP, loop); |
| 2371 | ||
| 2372 | JUMPHERE(found); | JUMPHERE(found); |
| 2373 | if (foundoc) | if (foundoc) |
| 2374 | JUMPHERE(foundoc); | JUMPHERE(foundoc); |
| 2375 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), REQ_BYTE_PTR, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, TMP1, 0); |
| 2376 | JUMPHERE(alreadyfound); | JUMPHERE(alreadyfound); |
| 2377 | JUMPHERE(toolong); | JUMPHERE(toolong); |
| 2378 | return notfound; | return notfound; |
| # | Line 1957 return notfound; | Line 2381 return notfound; |
| 2381 | static void do_revertframes(compiler_common *common) | static void do_revertframes(compiler_common *common) |
| 2382 | { | { |
| 2383 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| struct sljit_jump *earlyexit; | ||
| 2384 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 2385 | struct sljit_label *mainloop; | struct sljit_label *mainloop; |
| 2386 | ||
| 2387 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); |
| 2388 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD); | OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); |
| 2389 | ||
| 2390 | /* Drop frames until we reach STACK_TOP. */ | /* Drop frames until we reach STACK_TOP. */ |
| earlyexit = CMP(SLJIT_C_LESS, TMP1, 0, STACK_TOP, 0); | ||
| 2391 | mainloop = LABEL(); | mainloop = LABEL(); |
| 2392 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
| 2393 | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); |
| # | Line 1978 JUMPTO(SLJIT_JUMP, mainloop); | Line 2400 JUMPTO(SLJIT_JUMP, mainloop); |
| 2400 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2401 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); |
| 2402 | /* End of dropping frames. */ | /* End of dropping frames. */ |
| OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD, TMP1, 0); | ||
| CMPTO(SLJIT_C_GREATER_EQUAL, TMP1, 0, STACK_TOP, 0, mainloop); | ||
| JUMPHERE(earlyexit); | ||
| 2403 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 2404 | ||
| 2405 | JUMPHERE(jump); | JUMPHERE(jump); |
| jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmaxindex); | ||
| /* Set max index. */ | ||
| OP1(SLJIT_MOV, MAX_INDEX, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | ||
| OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | ||
| JUMPTO(SLJIT_JUMP, mainloop); | ||
| JUMPHERE(jump); | ||
| 2406 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); |
| 2407 | /* Set max index. */ | /* Set string begin. */ |
| 2408 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); |
| 2409 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); |
| 2410 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); |
| 2411 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
| 2412 | ||
| 2413 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2414 | if (common->mark_ptr != 0) | |
| 2415 | { | |
| 2416 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark); | |
| 2417 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | |
| 2418 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | |
| 2419 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); | |
| 2420 | JUMPTO(SLJIT_JUMP, mainloop); | |
| 2421 | ||
| 2422 | JUMPHERE(jump); | |
| 2423 | } | |
| 2424 | ||
| 2425 | /* Unknown command. */ | /* Unknown command. */ |
| 2426 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); |
| 2427 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
| # | Line 2008 JUMPTO(SLJIT_JUMP, mainloop); | Line 2430 JUMPTO(SLJIT_JUMP, mainloop); |
| 2430 | static void check_wordboundary(compiler_common *common) | static void check_wordboundary(compiler_common *common) |
| 2431 | { | { |
| 2432 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 2433 | struct sljit_jump *beginend; | struct sljit_jump *skipread; |
| 2434 | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF | |
| 2435 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 2436 | #endif | |
| 2437 | ||
| 2438 | SLJIT_ASSERT(ctype_word == 0x10); | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); |
| 2439 | ||
| 2440 | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, 1, 5, 5, common->localsize); |
| 2441 | /* Get type of the previous char, and put it to LOCALS1. */ | /* Get type of the previous char, and put it to LOCALS1. */ |
| 2442 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
| 2443 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
| 2444 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); |
| 2445 | beginend = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); |
| 2446 | skip_char_back(common); | skip_char_back(common); |
| 2447 | check_start_used_ptr(common); | |
| 2448 | read_char(common); | read_char(common); |
| 2449 | ||
| 2450 | /* Testing char type. */ | /* Testing char type. */ |
| 2451 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2452 | if (common->useucp) | if (common->use_ucp) |
| 2453 | { | { |
| 2454 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); |
| 2455 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); |
| # | Line 2041 if (common->useucp) | Line 2466 if (common->useucp) |
| 2466 | else | else |
| 2467 | #endif | #endif |
| 2468 | { | { |
| 2469 | #ifdef SUPPORT_UTF8 | #ifndef COMPILE_PCRE8 |
| 2470 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | |
| 2471 | #elif defined SUPPORT_UTF | |
| 2472 | /* Here LOCALS1 has already been zeroed. */ | /* Here LOCALS1 has already been zeroed. */ |
| 2473 | jump = NULL; | jump = NULL; |
| 2474 | if (common->utf8) | if (common->utf) |
| 2475 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
| 2476 | #endif | #endif /* COMPILE_PCRE8 */ |
| 2477 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); |
| 2478 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); |
| 2479 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
| 2480 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); |
| 2481 | #ifdef SUPPORT_UTF8 | #ifndef COMPILE_PCRE8 |
| 2482 | JUMPHERE(jump); | |
| 2483 | #elif defined SUPPORT_UTF | |
| 2484 | if (jump != NULL) | if (jump != NULL) |
| 2485 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2486 | #endif | #endif /* COMPILE_PCRE8 */ |
| 2487 | } | } |
| 2488 | JUMPHERE(beginend); | JUMPHERE(skipread); |
| 2489 | ||
| 2490 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
| 2491 | beginend = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | skipread = check_str_end(common); |
| 2492 | peek_char(common); | peek_char(common); |
| 2493 | ||
| 2494 | /* Testing char type. This is a code duplication. */ | /* Testing char type. This is a code duplication. */ |
| 2495 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2496 | if (common->useucp) | if (common->use_ucp) |
| 2497 | { | { |
| 2498 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); |
| 2499 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); |
| # | Line 2080 if (common->useucp) | Line 2509 if (common->useucp) |
| 2509 | else | else |
| 2510 | #endif | #endif |
| 2511 | { | { |
| 2512 | #ifdef SUPPORT_UTF8 | #ifndef COMPILE_PCRE8 |
| 2513 | /* TMP2 may be destroyed by peek_char. */ | |
| 2514 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | |
| 2515 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | |
| 2516 | #elif defined SUPPORT_UTF | |
| 2517 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
| 2518 | jump = NULL; | jump = NULL; |
| 2519 | if (common->utf8) | if (common->utf) |
| 2520 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
| 2521 | #endif | #endif |
| 2522 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); |
| 2523 | OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); | OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); |
| 2524 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
| 2525 | #ifdef SUPPORT_UTF8 | #ifndef COMPILE_PCRE8 |
| 2526 | JUMPHERE(jump); | |
| 2527 | #elif defined SUPPORT_UTF | |
| 2528 | if (jump != NULL) | if (jump != NULL) |
| 2529 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2530 | #endif | #endif /* COMPILE_PCRE8 */ |
| 2531 | } | } |
| 2532 | JUMPHERE(beginend); | JUMPHERE(skipread); |
| 2533 | ||
| 2534 | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); |
| 2535 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
| # | Line 2111 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I | Line 2546 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I |
| 2546 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); |
| 2547 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); |
| 2548 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); |
| 2549 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 |
| 2550 | if (common->utf8) | #ifdef COMPILE_PCRE8 |
| 2551 | if (common->utf) | |
| 2552 | { | { |
| 2553 | #endif | |
| 2554 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
| 2555 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
| 2556 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); |
| 2557 | #ifdef COMPILE_PCRE8 | |
| 2558 | } | } |
| 2559 | #endif | #endif |
| 2560 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | |
| 2561 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
| 2562 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| 2563 | } | } |
| # | Line 2135 COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_E | Line 2574 COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_E |
| 2574 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); |
| 2575 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
| 2576 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); |
| 2577 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 |
| 2578 | if (common->utf8) | #ifdef COMPILE_PCRE8 |
| 2579 | if (common->utf) | |
| 2580 | { | { |
| 2581 | #endif | |
| 2582 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
| 2583 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); |
| 2584 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
| # | Line 2151 if (common->utf8) | Line 2592 if (common->utf8) |
| 2592 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); |
| 2593 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
| 2594 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); |
| 2595 | #ifdef COMPILE_PCRE8 | |
| 2596 | } | } |
| 2597 | #endif | #endif |
| 2598 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | |
| 2599 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
| 2600 | ||
| 2601 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| # | Line 2169 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I | Line 2612 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I |
| 2612 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); |
| 2613 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); |
| 2614 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); |
| 2615 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 |
| 2616 | if (common->utf8) | #ifdef COMPILE_PCRE8 |
| 2617 | if (common->utf) | |
| 2618 | { | { |
| 2619 | #endif | |
| 2620 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
| 2621 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
| 2622 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); |
| 2623 | #ifdef COMPILE_PCRE8 | |
| 2624 | } | } |
| 2625 | #endif | #endif |
| 2626 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | |
| 2627 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
| 2628 | ||
| 2629 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| # | Line 2195 sljit_emit_fast_enter(compiler, RETURN_A | Line 2642 sljit_emit_fast_enter(compiler, RETURN_A |
| 2642 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
| 2643 | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); |
| 2644 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); |
| 2645 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2646 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2647 | ||
| 2648 | label = LABEL(); | label = LABEL(); |
| 2649 | OP1(SLJIT_MOVU_UB, CHAR1, 0, SLJIT_MEM1(TMP1), 1); | OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); |
| 2650 | OP1(SLJIT_MOVU_UB, CHAR2, 0, SLJIT_MEM1(STR_PTR), 1); | OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
| 2651 | jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); | jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); |
| 2652 | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2653 | JUMPTO(SLJIT_C_NOT_ZERO, label); | JUMPTO(SLJIT_C_NOT_ZERO, label); |
| 2654 | ||
| 2655 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2656 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2657 | OP1(SLJIT_MOV, CHAR1, 0, TMP3, 0); | OP1(SLJIT_MOV, CHAR1, 0, TMP3, 0); |
| 2658 | OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
| 2659 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
| # | Line 2227 OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); | Line 2674 OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); |
| 2674 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR1, 0); |
| 2675 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, CHAR2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, CHAR2, 0); |
| 2676 | OP1(SLJIT_MOV, LCC_TABLE, 0, SLJIT_IMM, common->lcc); | OP1(SLJIT_MOV, LCC_TABLE, 0, SLJIT_IMM, common->lcc); |
| 2677 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2678 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2679 | ||
| 2680 | label = LABEL(); | label = LABEL(); |
| 2681 | OP1(SLJIT_MOVU_UB, CHAR1, 0, SLJIT_MEM1(TMP1), 1); | OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); |
| 2682 | OP1(SLJIT_MOVU_UB, CHAR2, 0, SLJIT_MEM1(STR_PTR), 1); | OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
| 2683 | #ifndef COMPILE_PCRE8 | |
| 2684 | jump = CMP(SLJIT_C_GREATER, CHAR1, 0, SLJIT_IMM, 255); | |
| 2685 | #endif | |
| 2686 | OP1(SLJIT_MOV_UB, CHAR1, 0, SLJIT_MEM2(LCC_TABLE, CHAR1), 0); | OP1(SLJIT_MOV_UB, CHAR1, 0, SLJIT_MEM2(LCC_TABLE, CHAR1), 0); |
| 2687 | #ifndef COMPILE_PCRE8 | |
| 2688 | JUMPHERE(jump); | |
| 2689 | jump = CMP(SLJIT_C_GREATER, CHAR2, 0, SLJIT_IMM, 255); | |
| 2690 | #endif | |
| 2691 | OP1(SLJIT_MOV_UB, CHAR2, 0, SLJIT_MEM2(LCC_TABLE, CHAR2), 0); | OP1(SLJIT_MOV_UB, CHAR2, 0, SLJIT_MEM2(LCC_TABLE, CHAR2), 0); |
| 2692 | #ifndef COMPILE_PCRE8 | |
| 2693 | JUMPHERE(jump); | |
| 2694 | #endif | |
| 2695 | jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); | jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); |
| 2696 | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2697 | JUMPTO(SLJIT_C_NOT_ZERO, label); | JUMPTO(SLJIT_C_NOT_ZERO, label); |
| 2698 | ||
| 2699 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2700 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 2701 | OP1(SLJIT_MOV, LCC_TABLE, 0, TMP3, 0); | OP1(SLJIT_MOV, LCC_TABLE, 0, TMP3, 0); |
| 2702 | OP1(SLJIT_MOV, CHAR1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | OP1(SLJIT_MOV, CHAR1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
| 2703 | OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); |
| # | Line 2251 sljit_emit_fast_return(compiler, RETURN_ | Line 2708 sljit_emit_fast_return(compiler, RETURN_ |
| 2708 | #undef CHAR1 | #undef CHAR1 |
| 2709 | #undef CHAR2 | #undef CHAR2 |
| 2710 | ||
| 2711 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF && defined SUPPORT_UCP |
| #ifdef SUPPORT_UCP | ||
| 2712 | ||
| 2713 | static uschar * SLJIT_CALL do_utf8caselesscmp(uschar *src1, jit_arguments *args, uschar *end1) | static const pcre_uchar *SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1) |
| 2714 | { | { |
| 2715 | /* This function would be ineffective to do in JIT level. */ | /* This function would be ineffective to do in JIT level. */ |
| 2716 | int c1, c2; | int c1, c2; |
| 2717 | uschar *src2 = args->ptr; | const pcre_uchar *src2 = args->uchar_ptr; |
| 2718 | uschar *end2 = (uschar*)args->end; | const pcre_uchar *end2 = args->end; |
| 2719 | ||
| 2720 | while (src1 < end1) | while (src1 < end1) |
| 2721 | { | { |
| 2722 | if (src2 >= end2) | if (src2 >= end2) |
| 2723 | return 0; | return (pcre_uchar*)1; |
| 2724 | GETCHARINC(c1, src1); | GETCHARINC(c1, src1); |
| 2725 | GETCHARINC(c2, src2); | GETCHARINC(c2, src2); |
| 2726 | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return 0; | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL; |
| 2727 | } | } |
| 2728 | return src2; | return src2; |
| 2729 | } | } |
| 2730 | ||
| 2731 | #endif | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
| #endif | ||
| 2732 | ||
| 2733 | static uschar *byte_sequence_compare(compiler_common *common, BOOL caseless, uschar *cc, | static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, |
| 2734 | compare_context* context, jump_list **fallbacks) | compare_context* context, jump_list **fallbacks) |
| 2735 | { | { |
| 2736 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 2737 | unsigned int othercasebit = 0; | unsigned int othercasebit = 0; |
| 2738 | uschar *othercasebyte = NULL; | pcre_uchar *othercasechar = NULL; |
| 2739 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2740 | int utf8length; | int utflength; |
| 2741 | #endif | #endif |
| 2742 | ||
| 2743 | if (caseless && char_has_othercase(common, cc)) | if (caseless && char_has_othercase(common, cc)) |
| # | Line 2290 if (caseless && char_has_othercase(commo | Line 2745 if (caseless && char_has_othercase(commo |
| 2745 | othercasebit = char_get_othercase_bit(common, cc); | othercasebit = char_get_othercase_bit(common, cc); |
| 2746 | SLJIT_ASSERT(othercasebit); | SLJIT_ASSERT(othercasebit); |
| 2747 | /* Extracting bit difference info. */ | /* Extracting bit difference info. */ |
| 2748 | othercasebyte = cc + (othercasebit >> 8); | #ifdef COMPILE_PCRE8 |
| 2749 | othercasechar = cc + (othercasebit >> 8); | |
| 2750 | othercasebit &= 0xff; | othercasebit &= 0xff; |
| 2751 | #else | |
| 2752 | #ifdef COMPILE_PCRE16 | |
| 2753 | othercasechar = cc + (othercasebit >> 9); | |
| 2754 | if ((othercasebit & 0x100) != 0) | |
| 2755 | othercasebit = (othercasebit & 0xff) << 8; | |
| 2756 | else | |
| 2757 | othercasebit &= 0xff; | |
| 2758 | #endif | |
| 2759 | #endif | |
| 2760 | } | } |
| 2761 | ||
| 2762 | if (context->sourcereg == -1) | if (context->sourcereg == -1) |
| 2763 | { | { |
| 2764 | #ifdef COMPILE_PCRE8 | |
| 2765 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
| 2766 | if (context->length >= 4) | if (context->length >= 4) |
| 2767 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2768 | else if (context->length >= 2) | else if (context->length >= 2) |
| 2769 | OP1(SLJIT_MOV_SH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2770 | else | else |
| 2771 | #endif | #endif |
| 2772 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2773 | #else | |
| 2774 | #ifdef COMPILE_PCRE16 | |
| 2775 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | |
| 2776 | if (context->length >= 4) | |
| 2777 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
| 2778 | else | |
| 2779 | #endif | |
| 2780 | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
| 2781 | #endif | |
| 2782 | #endif /* COMPILE_PCRE8 */ | |
| 2783 | context->sourcereg = TMP2; | context->sourcereg = TMP2; |
| 2784 | } | } |
| 2785 | ||
| 2786 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2787 | utf8length = 1; | utflength = 1; |
| 2788 | if (common->utf8 && *cc >= 0xc0) | if (common->utf && HAS_EXTRALEN(*cc)) |
| 2789 | utf8length += _pcre_utf8_table4[*cc & 0x3f]; | utflength += GET_EXTRALEN(*cc); |
| 2790 | ||
| 2791 | do | do |
| 2792 | { | { |
| 2793 | #endif | #endif |
| 2794 | ||
| 2795 | context->length--; | context->length -= IN_UCHARS(1); |
| 2796 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
| 2797 | ||
| 2798 | /* Unaligned read is supported. */ | /* Unaligned read is supported. */ |
| 2799 | if (othercasebit != 0 && othercasebyte == cc) | if (othercasebit != 0 && othercasechar == cc) |
| 2800 | { | { |
| 2801 | context->c.asbytes[context->byteptr] = *cc | othercasebit; | context->c.asuchars[context->ucharptr] = *cc | othercasebit; |
| 2802 | context->oc.asbytes[context->byteptr] = othercasebit; | context->oc.asuchars[context->ucharptr] = othercasebit; |
| 2803 | } | } |
| 2804 | else | else |
| 2805 | { | { |
| 2806 | context->c.asbytes[context->byteptr] = *cc; | context->c.asuchars[context->ucharptr] = *cc; |
| 2807 | context->oc.asbytes[context->byteptr] = 0; | context->oc.asuchars[context->ucharptr] = 0; |
| 2808 | } | } |
| 2809 | context->byteptr++; | context->ucharptr++; |
| 2810 | ||
| 2811 | if (context->byteptr >= 4 || context->length == 0 || (context->byteptr == 2 && context->length == 1)) | #ifdef COMPILE_PCRE8 |
| 2812 | if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1)) | |
| 2813 | #else | |
| 2814 | if (context->ucharptr >= 2 || context->length == 0) | |
| 2815 | #endif | |
| 2816 | { | { |
| 2817 | if (context->length >= 4) | if (context->length >= 4) |
| 2818 | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2819 | #ifdef COMPILE_PCRE8 | |
| 2820 | else if (context->length >= 2) | else if (context->length >= 2) |
| 2821 | OP1(SLJIT_MOV_SH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2822 | else if (context->length >= 1) | else if (context->length >= 1) |
| 2823 | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2824 | #else | |
| 2825 | else if (context->length >= 2) | |
| 2826 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
| 2827 | #endif | |
| 2828 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
| 2829 | ||
| 2830 | switch(context->byteptr) | switch(context->ucharptr) |
| 2831 | { | { |
| 2832 | case 4: | case 4 / sizeof(pcre_uchar): |
| 2833 | if (context->oc.asint != 0) | if (context->oc.asint != 0) |
| 2834 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); |
| 2835 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); |
| 2836 | break; | break; |
| 2837 | ||
| 2838 | case 2: | case 2 / sizeof(pcre_uchar): |
| 2839 | if (context->oc.asshort != 0) | if (context->oc.asushort != 0) |
| 2840 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asshort); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); |
| 2841 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asshort | context->oc.asshort)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); |
| 2842 | break; | break; |
| 2843 | ||
| 2844 | #ifdef COMPILE_PCRE8 | |
| 2845 | case 1: | case 1: |
| 2846 | if (context->oc.asbyte != 0) | if (context->oc.asbyte != 0) |
| 2847 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); |
| 2848 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); |
| 2849 | break; | break; |
| 2850 | #endif | |
| 2851 | ||
| 2852 | default: | default: |
| 2853 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
| 2854 | break; | break; |
| 2855 | } | } |
| 2856 | context->byteptr = 0; | context->ucharptr = 0; |
| 2857 | } | } |
| 2858 | ||
| 2859 | #else | #else |
| 2860 | ||
| 2861 | /* Unaligned read is unsupported. */ | /* Unaligned read is unsupported. */ |
| 2862 | #ifdef COMPILE_PCRE8 | |
| 2863 | if (context->length > 0) | if (context->length > 0) |
| 2864 | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
| 2865 | #else | |
| 2866 | if (context->length > 0) | |
| 2867 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
| 2868 | #endif | |
| 2869 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
| 2870 | ||
| 2871 | if (othercasebit != 0 && othercasebyte == cc) | if (othercasebit != 0 && othercasechar == cc) |
| 2872 | { | { |
| 2873 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); |
| 2874 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); |
| # | Line 2387 do | Line 2879 do |
| 2879 | #endif | #endif |
| 2880 | ||
| 2881 | cc++; | cc++; |
| 2882 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2883 | utf8length--; | utflength--; |
| 2884 | } | } |
| 2885 | while (utf8length > 0); | while (utflength > 0); |
| 2886 | #endif | #endif |
| 2887 | ||
| 2888 | return cc; | return cc; |
| 2889 | } | } |
| 2890 | ||
| 2891 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 2892 | ||
| 2893 | #define SET_TYPE_OFFSET(value) \ | #define SET_TYPE_OFFSET(value) \ |
| 2894 | if ((value) != typeoffset) \ | if ((value) != typeoffset) \ |
| # | Line 2418 return cc; | Line 2910 return cc; |
| 2910 | } \ | } \ |
| 2911 | charoffset = (value); | charoffset = (value); |
| 2912 | ||
| 2913 | static void compile_xclass_hotpath(compiler_common *common, uschar *cc, jump_list **fallbacks) | static void compile_xclass_hotpath(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks) |
| 2914 | { | { |
| 2915 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 2916 | jump_list *found = NULL; | jump_list *found = NULL; |
| # | Line 2426 jump_list **list = (*cc & XCL_NOT) == 0 | Line 2918 jump_list **list = (*cc & XCL_NOT) == 0 |
| 2918 | unsigned int c; | unsigned int c; |
| 2919 | int compares; | int compares; |
| 2920 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
| 2921 | uschar *ccbegin; | pcre_uchar *ccbegin; |
| 2922 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2923 | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; |
| 2924 | BOOL charsaved = FALSE; | BOOL charsaved = FALSE; |
| 2925 | int typereg = TMP1, scriptreg = TMP1, typeoffset; | int typereg = TMP1, scriptreg = TMP1; |
| 2926 | unsigned int typeoffset; | |
| 2927 | #endif | #endif |
| 2928 | int charoffset, invertcmp, numberofcmps; | int invertcmp, numberofcmps; |
| 2929 | unsigned int charoffset; | |
| 2930 | ||
| 2931 | /* Although SUPPORT_UTF8 must be defined, we are not necessary in utf8 mode. */ | /* Although SUPPORT_UTF must be defined, we are not necessary in utf mode. */ |
| 2932 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 2933 | read_char(common); | read_char(common); |
| 2934 | ||
| 2935 | if ((*cc++ & XCL_MAP) != 0) | if ((*cc++ & XCL_MAP) != 0) |
| 2936 | { | { |
| 2937 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
| 2938 | if (common->utf8) | #ifndef COMPILE_PCRE8 |
| 2939 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | |
| 2940 | #elif defined SUPPORT_UTF | |
| 2941 | if (common->utf) | |
| 2942 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
| 2943 | #endif | |
| 2944 | ||
| 2945 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
| 2946 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
| # | Line 2451 if ((*cc++ & XCL_MAP) != 0) | Line 2949 if ((*cc++ & XCL_MAP) != 0) |
| 2949 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); |
| 2950 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); |
| 2951 | ||
| 2952 | if (common->utf8) | #ifndef COMPILE_PCRE8 |
| 2953 | JUMPHERE(jump); | |
| 2954 | #elif defined SUPPORT_UTF | |
| 2955 | if (common->utf) | |
| 2956 | JUMPHERE(jump); | JUMPHERE(jump); |
| 2957 | #endif | |
| 2958 | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
| 2959 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2960 | charsaved = TRUE; | charsaved = TRUE; |
| 2961 | #endif | #endif |
| 2962 | cc += 32; | cc += 32 / sizeof(pcre_uchar); |
| 2963 | } | } |
| 2964 | ||
| 2965 | /* Scanning the necessary info. */ | /* Scanning the necessary info. */ |
| # | Line 2469 while (*cc != XCL_END) | Line 2971 while (*cc != XCL_END) |
| 2971 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
| 2972 | { | { |
| 2973 | cc += 2; | cc += 2; |
| 2974 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2975 | if (common->utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
| 2976 | #endif | #endif |
| 2977 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2978 | needschar = TRUE; | needschar = TRUE; |
| # | Line 2479 while (*cc != XCL_END) | Line 2981 while (*cc != XCL_END) |
| 2981 | else if (*cc == XCL_RANGE) | else if (*cc == XCL_RANGE) |
| 2982 | { | { |
| 2983 | cc += 2; | cc += 2; |
| 2984 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2985 | if (common->utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
| 2986 | #endif | #endif |
| 2987 | cc++; | cc++; |
| 2988 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2989 | if (common->utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
| 2990 | #endif | #endif |
| 2991 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2992 | needschar = TRUE; | needschar = TRUE; |
| # | Line 2554 if (needstype || needsscript) | Line 3056 if (needstype || needsscript) |
| 3056 | { | { |
| 3057 | if (scriptreg == TMP1) | if (scriptreg == TMP1) |
| 3058 | { | { |
| 3059 | OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_w)_pcre_ucd_records + SLJIT_OFFSETOF(ucd_record, script)); | OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); |
| 3060 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); |
| 3061 | } | } |
| 3062 | else | else |
| 3063 | { | { |
| 3064 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); |
| 3065 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_w)_pcre_ucd_records + SLJIT_OFFSETOF(ucd_record, script)); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); |
| 3066 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); |
| 3067 | } | } |
| 3068 | } | } |
| # | Line 2584 while (*cc != XCL_END) | Line 3086 while (*cc != XCL_END) |
| 3086 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
| 3087 | { | { |
| 3088 | cc ++; | cc ++; |
| 3089 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3090 | if (common->utf8) | if (common->utf) |
| 3091 | { | { |
| 3092 | GETCHARINC(c, cc); | GETCHARINC(c, cc); |
| 3093 | } | } |
| # | Line 2615 while (*cc != XCL_END) | Line 3117 while (*cc != XCL_END) |
| 3117 | else if (*cc == XCL_RANGE) | else if (*cc == XCL_RANGE) |
| 3118 | { | { |
| 3119 | cc ++; | cc ++; |
| 3120 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3121 | if (common->utf8) | if (common->utf) |
| 3122 | { | { |
| 3123 | GETCHARINC(c, cc); | GETCHARINC(c, cc); |
| 3124 | } | } |
| # | Line 2624 while (*cc != XCL_END) | Line 3126 while (*cc != XCL_END) |
| 3126 | #endif | #endif |
| 3127 | c = *cc++; | c = *cc++; |
| 3128 | SET_CHAR_OFFSET(c); | SET_CHAR_OFFSET(c); |
| 3129 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3130 | if (common->utf8) | if (common->utf) |
| 3131 | { | { |
| 3132 | GETCHARINC(c, cc); | GETCHARINC(c, cc); |
| 3133 | } | } |
| # | Line 2681 while (*cc != XCL_END) | Line 3183 while (*cc != XCL_END) |
| 3183 | break; | break; |
| 3184 | ||
| 3185 | case PT_GC: | case PT_GC: |
| 3186 | c = _pcre_ucp_typerange[(int)cc[1] * 2]; | c = PRIV(ucp_typerange)[(int)cc[1] * 2]; |
| 3187 | SET_TYPE_OFFSET(c); | SET_TYPE_OFFSET(c); |
| 3188 | jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, _pcre_ucp_typerange[(int)cc[1] * 2 + 1] - c); | jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, PRIV(ucp_typerange)[(int)cc[1] * 2 + 1] - c); |
| 3189 | break; | break; |
| 3190 | ||
| 3191 | case PT_PC: | case PT_PC: |
| # | Line 2745 if (found != NULL) | Line 3247 if (found != NULL) |
| 3247 | ||
| 3248 | #endif | #endif |
| 3249 | ||
| 3250 | static uschar *compile_char1_hotpath(compiler_common *common, uschar type, uschar *cc, jump_list **fallbacks) | static pcre_uchar *compile_char1_hotpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **fallbacks) |
| 3251 | { | { |
| 3252 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 3253 | int length; | int length; |
| 3254 | unsigned int c, oc, bit; | unsigned int c, oc, bit; |
| 3255 | compare_context context; | compare_context context; |
| struct sljit_label *label; | ||
| 3256 | struct sljit_jump *jump[4]; | struct sljit_jump *jump[4]; |
| 3257 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3258 | struct sljit_label *label; | |
| 3259 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3260 | uschar propdata[5]; | pcre_uchar propdata[5]; |
| 3261 | #endif | #endif |
| 3262 | #endif | #endif |
| 3263 | ||
| # | Line 2781 switch(type) | Line 3283 switch(type) |
| 3283 | ||
| 3284 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3285 | case OP_DIGIT: | case OP_DIGIT: |
| 3286 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3287 | read_char8_type(common); | read_char8_type(common); |
| 3288 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); |
| 3289 | add_jump(compiler, fallbacks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, fallbacks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
| # | Line 2789 switch(type) | Line 3291 switch(type) |
| 3291 | ||
| 3292 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3293 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3294 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3295 | read_char8_type(common); | read_char8_type(common); |
| 3296 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); |
| 3297 | add_jump(compiler, fallbacks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, fallbacks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
| # | Line 2797 switch(type) | Line 3299 switch(type) |
| 3299 | ||
| 3300 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3301 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 3302 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3303 | read_char8_type(common); | read_char8_type(common); |
| 3304 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); |
| 3305 | add_jump(compiler, fallbacks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, fallbacks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
| 3306 | return cc; | return cc; |
| 3307 | ||
| 3308 | case OP_ANY: | case OP_ANY: |
| 3309 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3310 | read_char(common); | read_char(common); |
| 3311 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| 3312 | { | { |
| 3313 | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); |
| 3314 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | if (common->mode != JIT_PARTIAL_HARD_COMPILE) |
| 3315 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 3316 | else | |
| 3317 | jump[1] = check_str_end(common); | |
| 3318 | ||
| 3319 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
| 3320 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); |
| 3321 | JUMPHERE(jump[1]); | if (jump[1] != NULL) |
| 3322 | JUMPHERE(jump[1]); | |
| 3323 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
| 3324 | } | } |
| 3325 | else | else |
| # | Line 2820 switch(type) | Line 3327 switch(type) |
| 3327 | return cc; | return cc; |
| 3328 | ||
| 3329 | case OP_ALLANY: | case OP_ALLANY: |
| 3330 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3331 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3332 | if (common->utf8) | if (common->utf) |
| 3333 | { | { |
| 3334 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| 3335 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)_pcre_utf8_char_sizes); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3336 | #ifdef COMPILE_PCRE8 | |
| 3337 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
| 3338 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | |
| 3339 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
| 3340 | #else /* COMPILE_PCRE8 */ | |
| 3341 | #ifdef COMPILE_PCRE16 | |
| 3342 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 3343 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
| 3344 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
| 3345 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | |
| 3346 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
| 3347 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 3348 | #endif /* COMPILE_PCRE16 */ | |
| 3349 | #endif /* COMPILE_PCRE8 */ | |
| 3350 | JUMPHERE(jump[0]); | |
| 3351 | return cc; | return cc; |
| 3352 | } | } |
| 3353 | #endif | #endif |
| 3354 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3355 | return cc; | |
| 3356 | ||
| 3357 | case OP_ANYBYTE: | |
| 3358 | fallback_at_str_end(common, fallbacks); | |
| 3359 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
| 3360 | return cc; | return cc; |
| 3361 | ||
| 3362 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3363 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3364 | case OP_NOTPROP: | case OP_NOTPROP: |
| 3365 | case OP_PROP: | case OP_PROP: |
| # | Line 2848 switch(type) | Line 3374 switch(type) |
| 3374 | #endif | #endif |
| 3375 | ||
| 3376 | case OP_ANYNL: | case OP_ANYNL: |
| 3377 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3378 | read_char(common); | read_char(common); |
| 3379 | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); |
| 3380 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | /* We don't need to handle soft partial matching case. */ |
| 3381 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | if (common->mode != JIT_PARTIAL_HARD_COMPILE) |
| 3382 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
| 3383 | else | |
| 3384 | jump[1] = check_str_end(common); | |
| 3385 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
| 3386 | jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); | jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); |
| 3387 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3388 | jump[3] = JUMP(SLJIT_JUMP); | jump[3] = JUMP(SLJIT_JUMP); |
| 3389 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
| 3390 | check_newlinechar(common, common->bsr_nltype, fallbacks, FALSE); | check_newlinechar(common, common->bsr_nltype, fallbacks, FALSE); |
| # | Line 2865 switch(type) | Line 3395 switch(type) |
| 3395 | ||
| 3396 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3397 | case OP_HSPACE: | case OP_HSPACE: |
| 3398 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3399 | read_char(common); | read_char(common); |
| 3400 | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); |
| 3401 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
| # | Line 2873 switch(type) | Line 3403 switch(type) |
| 3403 | ||
| 3404 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3405 | case OP_VSPACE: | case OP_VSPACE: |
| 3406 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3407 | read_char(common); | read_char(common); |
| 3408 | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); |
| 3409 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
| # | Line 2881 switch(type) | Line 3411 switch(type) |
| 3411 | ||
| 3412 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3413 | case OP_EXTUNI: | case OP_EXTUNI: |
| 3414 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3415 | read_char(common); | read_char(common); |
| 3416 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
| 3417 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); |
| # | Line 2897 switch(type) | Line 3427 switch(type) |
| 3427 | ||
| 3428 | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
| 3429 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
| 3430 | if (common->mode == JIT_PARTIAL_HARD_COMPILE) | |
| 3431 | { | |
| 3432 | jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | |
| 3433 | /* Since we successfully read a char above, partial matching must occure. */ | |
| 3434 | check_partial(common, TRUE); | |
| 3435 | JUMPHERE(jump[0]); | |
| 3436 | } | |
| 3437 | return cc; | return cc; |
| 3438 | #endif | #endif |
| 3439 | ||
| 3440 | case OP_EODN: | case OP_EODN: |
| 3441 | /* Requires rather complex checks. */ | |
| 3442 | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
| 3443 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| 3444 | { | { |
| 3445 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 2); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 3446 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 3447 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | if (common->mode == JIT_COMPILE) |
| 3448 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); |
| 3449 | else | |
| 3450 | { | |
| 3451 | jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); | |
| 3452 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); | |
| 3453 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS); | |
| 3454 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); | |
| 3455 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_NOT_EQUAL); | |
| 3456 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_NOT_EQUAL)); | |
| 3457 | check_partial(common, TRUE); | |
| 3458 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | |
| 3459 | JUMPHERE(jump[1]); | |
| 3460 | } | |
| 3461 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
| 3462 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
| 3463 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
| 3464 | } | } |
| 3465 | else if (common->nltype == NLTYPE_FIXED) | else if (common->nltype == NLTYPE_FIXED) |
| 3466 | { | { |
| 3467 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3468 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 3469 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); |
| 3470 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
| 3471 | } | } |
| 3472 | else | else |
| 3473 | { | { |
| 3474 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 3475 | jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); |
| 3476 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 2); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 3477 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); |
| 3478 | jump[2] = JUMP(SLJIT_C_GREATER); | jump[2] = JUMP(SLJIT_C_GREATER); |
| 3479 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_LESS)); | add_jump(compiler, fallbacks, JUMP(SLJIT_C_LESS)); |
| 3480 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 1); | /* Equal. */ |
| 3481 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
| 3482 | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); |
| 3483 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); |
| 3484 | ||
| 3485 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
| 3486 | if (common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANYCRLF) |
| 3487 | { | { |
| 3488 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3489 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); |
| 3490 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); |
| 3491 | } | } |
| # | Line 2950 switch(type) | Line 3502 switch(type) |
| 3502 | JUMPHERE(jump[3]); | JUMPHERE(jump[3]); |
| 3503 | } | } |
| 3504 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
| 3505 | check_partial(common, FALSE); | |
| 3506 | return cc; | return cc; |
| 3507 | ||
| 3508 | case OP_EOD: | case OP_EOD: |
| 3509 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); |
| 3510 | check_partial(common, FALSE); | |
| 3511 | return cc; | return cc; |
| 3512 | ||
| 3513 | case OP_CIRC: | case OP_CIRC: |
| # | Line 2973 switch(type) | Line 3527 switch(type) |
| 3527 | jump[0] = JUMP(SLJIT_JUMP); | jump[0] = JUMP(SLJIT_JUMP); |
| 3528 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
| 3529 | ||
| 3530 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, end)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
| add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP2, 0, STR_PTR, 0)); | ||
| 3531 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| 3532 | { | { |
| 3533 | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 2); | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 3534 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); |
| 3535 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -2); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
| 3536 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), -1); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
| 3537 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
| 3538 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
| 3539 | } | } |
| # | Line 3002 switch(type) | Line 3554 switch(type) |
| 3554 | if (!common->endonly) | if (!common->endonly) |
| 3555 | compile_char1_hotpath(common, OP_EODN, cc, fallbacks); | compile_char1_hotpath(common, OP_EODN, cc, fallbacks); |
| 3556 | else | else |
| 3557 | { | |
| 3558 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); |
| 3559 | check_partial(common, FALSE); | |
| 3560 | } | |
| 3561 | return cc; | return cc; |
| 3562 | ||
| 3563 | case OP_DOLLM: | case OP_DOLLM: |
| # | Line 3010 switch(type) | Line 3565 switch(type) |
| 3565 | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); |
| 3566 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); |
| 3567 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
| 3568 | check_partial(common, FALSE); | |
| 3569 | jump[0] = JUMP(SLJIT_JUMP); | jump[0] = JUMP(SLJIT_JUMP); |
| 3570 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
| 3571 | ||
| 3572 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
| 3573 | { | { |
| 3574 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, 2); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
| 3575 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
| 3576 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | if (common->mode == JIT_COMPILE) |
| 3577 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(STR_PTR), 1); | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); |
| 3578 | else | |
| 3579 | { | |
| 3580 | jump[1] = CMP(SLJIT_C_LESS_EQUAL, TMP2, 0, STR_END, 0); | |
| 3581 | /* STR_PTR = STR_END - IN_UCHARS(1) */ | |
| 3582 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | |
| 3583 | check_partial(common, TRUE); | |
| 3584 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | |
| 3585 | JUMPHERE(jump[1]); | |
| 3586 | } | |
| 3587 | ||
| 3588 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
| 3589 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
| 3590 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
| 3591 | } | } |
| # | Line 3033 switch(type) | Line 3600 switch(type) |
| 3600 | case OP_CHAR: | case OP_CHAR: |
| 3601 | case OP_CHARI: | case OP_CHARI: |
| 3602 | length = 1; | length = 1; |
| 3603 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3604 | if (common->utf8 && *cc >= 0xc0) length += _pcre_utf8_table4[*cc & 0x3f]; | if (common->utf && HAS_EXTRALEN(*cc)) length += GET_EXTRALEN(*cc); |
| 3605 | #endif | #endif |
| 3606 | if (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0) | if (common->mode == JIT_COMPILE && (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0)) |
| 3607 | { | { |
| 3608 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, length); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); |
| 3609 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); |
| 3610 | ||
| 3611 | context.length = length; | context.length = IN_UCHARS(length); |
| 3612 | context.sourcereg = -1; | context.sourcereg = -1; |
| 3613 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
| 3614 | context.byteptr = 0; | context.ucharptr = 0; |
| 3615 | #endif | #endif |
| 3616 | return byte_sequence_compare(common, type == OP_CHARI, cc, &context, fallbacks); | return byte_sequence_compare(common, type == OP_CHARI, cc, &context, fallbacks); |
| 3617 | } | } |
| 3618 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | fallback_at_str_end(common, fallbacks); |
| 3619 | read_char(common); | read_char(common); |
| 3620 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3621 | if (common->utf8) | if (common->utf) |
| 3622 | { | { |
| 3623 | GETCHAR(c, cc); | GETCHAR(c, cc); |
| 3624 | } | } |
| 3625 | else | else |
| 3626 | #endif | #endif |
| 3627 | c = *cc; | c = *cc; |
| 3628 | if (type == OP_CHAR || !char_has_othercase(common, cc)) | |
| 3629 | { | |
| 3630 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); | |
| 3631 | return cc + length; | |
| 3632 | } | |
| 3633 | oc = char_othercase(common, c); | |
| 3634 | bit = c ^ oc; | |
| 3635 | if (ispowerof2(bit)) | |
| 3636 | { | |
| 3637 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); | |
| 3638 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); | |
| 3639 | return cc + length; | |
| 3640 | } | |
| 3641 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); |
| 3642 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
| 3643 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_othercase(common, c)); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_othercase(common, c)); |
| # | Line 3067 switch(type) | Line 3647 switch(type) |
| 3647 | ||
| 3648 | case OP_NOT: | case OP_NOT: |
| 3649 | case OP_NOTI: | case OP_NOTI: |
| 3650 | fallback_at_str_end(common, fallbacks); | |
| 3651 | length = 1; | length = 1; |
| 3652 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3653 | if (common->utf8) | if (common->utf) |
| 3654 | { | { |
| 3655 | if (*cc >= 0xc0) length += _pcre_utf8_table4[*cc & 0x3f]; | #ifdef COMPILE_PCRE8 |
| 3656 | c = *cc; | |
| 3657 | check_input_end(common, fallbacks); | if (c < 128) |
| GETCHAR(c, cc); | ||
| if (c <= 127) | ||
| 3658 | { | { |
| 3659 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
| OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_w)_pcre_utf8_char_sizes); | ||
| 3660 | if (type == OP_NOT || !char_has_othercase(common, cc)) | if (type == OP_NOT || !char_has_othercase(common, cc)) |
| 3661 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); |
| 3662 | else | else |
| 3663 | { | { |
| 3664 | /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ | /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ |
| 3665 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x20); |
| 3666 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | 0x20)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, c | 0x20)); |
| 3667 | } | } |
| 3668 | /* Skip the variable-length character. */ | /* Skip the variable-length character. */ |
| 3669 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
| 3670 | return cc + length; | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
| 3671 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | |
| 3672 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
| 3673 | JUMPHERE(jump[0]); | |
| 3674 | return cc + 1; | |
| 3675 | } | } |
| 3676 | else | else |
| 3677 | #endif /* COMPILE_PCRE8 */ | |
| 3678 | { | |
| 3679 | GETCHARLEN(c, cc, length); | |
| 3680 | read_char(common); | read_char(common); |
| 3681 | } | |
| 3682 | } | } |
| 3683 | else | else |
| 3684 | #endif | #endif /* SUPPORT_UTF */ |
| 3685 | { | { |
| 3686 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, 1); | read_char(common); |
| add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); | ||
| OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -1); | ||
| 3687 | c = *cc; | c = *cc; |
| 3688 | } | } |
| 3689 | ||
| # | Line 3125 switch(type) | Line 3708 switch(type) |
| 3708 | ||
| 3709 | case OP_CLASS: | case OP_CLASS: |
| 3710 | case OP_NCLASS: | case OP_NCLASS: |
| 3711 | check_input_end(common, fallbacks); | fallback_at_str_end(common, fallbacks); |
| 3712 | read_char(common); | read_char(common); |
| 3713 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 3714 | jump[0] = NULL; | jump[0] = NULL; |
| 3715 | if (common->utf8) | #ifdef COMPILE_PCRE8 |
| 3716 | /* This check only affects 8 bit mode. In other modes, we | |
| 3717 | always need to compare the value with 255. */ | |
| 3718 | if (common->utf) | |
| 3719 | #endif /* COMPILE_PCRE8 */ | |
| 3720 | { | { |
| 3721 | jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
| 3722 | if (type == OP_CLASS) | if (type == OP_CLASS) |
| # | Line 3138 switch(type) | Line 3725 switch(type) |
| 3725 | jump[0] = NULL; | jump[0] = NULL; |
| 3726 | } | } |
| 3727 | } | } |
| 3728 | #endif | #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ |
| 3729 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
| 3730 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
| 3731 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); |
| 3732 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
| 3733 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); |
| 3734 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); |
| 3735 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 3736 | if (jump[0] != NULL) | if (jump[0] != NULL) |
| 3737 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
| 3738 | #endif | #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ |
| 3739 | return cc + 32; | return cc + 32 / sizeof(pcre_uchar); |
| 3740 | ||
| 3741 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 |
| 3742 | case OP_XCLASS: | case OP_XCLASS: |
| 3743 | compile_xclass_hotpath(common, cc + LINK_SIZE, fallbacks); | compile_xclass_hotpath(common, cc + LINK_SIZE, fallbacks); |
| 3744 | return cc + GET(cc, 0) - 1; | return cc + GET(cc, 0) - 1; |
| # | Line 3159 switch(type) | Line 3746 switch(type) |
| 3746 | ||
| 3747 | case OP_REVERSE: | case OP_REVERSE: |
| 3748 | length = GET(cc, 0); | length = GET(cc, 0); |
| 3749 | SLJIT_ASSERT(length > 0); | if (length == 0) |
| 3750 | return cc + LINK_SIZE; | |
| 3751 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
| 3752 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | #ifdef SUPPORT_UTF |
| 3753 | #ifdef SUPPORT_UTF8 | if (common->utf) |
| if (common->utf8) | ||
| 3754 | { | { |
| 3755 | OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | |
| 3756 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); |
| 3757 | label = LABEL(); | label = LABEL(); |
| 3758 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP3, 0)); |
| 3759 | skip_char_back(common); | skip_char_back(common); |
| 3760 | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
| 3761 | JUMPTO(SLJIT_C_NOT_ZERO, label); | JUMPTO(SLJIT_C_NOT_ZERO, label); |
| return cc + LINK_SIZE; | ||
| 3762 | } | } |
| 3763 | else | |
| 3764 | #endif | #endif |
| 3765 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, length); | { |
| 3766 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
| 3767 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); | |
| 3768 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); | |
| 3769 | } | |
| 3770 | check_start_used_ptr(common); | |
| 3771 | return cc + LINK_SIZE; | return cc + LINK_SIZE; |
| 3772 | } | } |
| 3773 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
| 3774 | return cc; | return cc; |
| 3775 | } | } |
| 3776 | ||
| 3777 | static SLJIT_INLINE uschar *compile_charn_hotpath(compiler_common *common, uschar *cc, uschar *ccend, jump_list **fallbacks) | static SLJIT_INLINE pcre_uchar *compile_charn_hotpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, jump_list **fallbacks) |
| 3778 | { | { |
| 3779 | /* This function consumes at least one input character. */ | /* This function consumes at least one input character. */ |
| 3780 | /* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ | /* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ |
| 3781 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 3782 | uschar *ccbegin = cc; | pcre_uchar *ccbegin = cc; |
| 3783 | compare_context context; | compare_context context; |
| 3784 | int size; | int size; |
| 3785 | ||
| # | Line 3200 do | Line 3792 do |
| 3792 | if (*cc == OP_CHAR) | if (*cc == OP_CHAR) |
| 3793 | { | { |
| 3794 | size = 1; | size = 1; |
| 3795 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3796 | if (common->utf8 && cc[1] >= 0xc0) | if (common->utf && HAS_EXTRALEN(cc[1])) |
| 3797 | size += _pcre_utf8_table4[cc[1] & 0x3f]; | size += GET_EXTRALEN(cc[1]); |
| 3798 | #endif | #endif |
| 3799 | } | } |
| 3800 | else if (*cc == OP_CHARI) | else if (*cc == OP_CHARI) |
| 3801 | { | { |
| 3802 | size = 1; | size = 1; |
| 3803 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3804 | if (common->utf8) | if (common->utf) |
| 3805 | { | { |
| 3806 | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) |
| 3807 | size = 0; | size = 0; |
| 3808 | else if (cc[1] >= 0xc0) | else if (HAS_EXTRALEN(cc[1])) |
| 3809 | size += _pcre_utf8_table4[cc[1] & 0x3f]; | size += GET_EXTRALEN(cc[1]); |
| 3810 | } | } |
| 3811 | else | else |
| 3812 | #endif | #endif |
| 3813 | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) |
| 3814 | size = 0; | size = 0; |
| # | Line 3225 do | Line 3817 do |
| 3817 | size = 0; | size = 0; |
| 3818 | ||
| 3819 | cc += 1 + size; | cc += 1 + size; |
| 3820 | context.length += size; | context.length += IN_UCHARS(size); |
| 3821 | } | } |
| 3822 | while (size > 0 && context.length <= 128); | while (size > 0 && context.length <= 128); |
| 3823 | ||
| # | Line 3238 if (context.length > 0) | Line 3830 if (context.length > 0) |
| 3830 | ||
| 3831 | context.sourcereg = -1; | context.sourcereg = -1; |
| 3832 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
| 3833 | context.byteptr = 0; | context.ucharptr = 0; |
| 3834 | #endif | #endif |
| 3835 | do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, fallbacks); while (context.length > 0); | do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, fallbacks); while (context.length > 0); |
| 3836 | return cc; | return cc; |
| # | Line 3248 if (context.length > 0) | Line 3840 if (context.length > 0) |
| 3840 | return compile_char1_hotpath(common, *cc, cc + 1, fallbacks); | return compile_char1_hotpath(common, *cc, cc + 1, fallbacks); |
| 3841 | } | } |
| 3842 | ||
| 3843 | static struct sljit_jump *compile_ref_checks(compiler_common *common, uschar *cc, jump_list **fallbacks) | static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks) |
| 3844 | { | { |
| 3845 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 3846 | int offset = GET2(cc, 1) << 1; | int offset = GET2(cc, 1) << 1; |
| # | Line 3258 if (!common->jscript_compat) | Line 3850 if (!common->jscript_compat) |
| 3850 | { | { |
| 3851 | if (fallbacks == NULL) | if (fallbacks == NULL) |
| 3852 | { | { |
| 3853 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | |
| 3854 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); |
| 3855 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
| 3856 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); |
| # | Line 3270 return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT | Line 3863 return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT |
| 3863 | } | } |
| 3864 | ||
| 3865 | /* Forward definitions. */ | /* Forward definitions. */ |
| 3866 | static void compile_hotpath(compiler_common *, uschar *, uschar *, fallback_common *); | static void compile_hotpath(compiler_common *, pcre_uchar *, pcre_uchar *, fallback_common *); |
| 3867 | static void compile_fallbackpath(compiler_common *, struct fallback_common *); | static void compile_fallbackpath(compiler_common *, struct fallback_common *); |
| 3868 | ||
| 3869 | #define PUSH_FALLBACK(size, ccstart, error) \ | #define PUSH_FALLBACK(size, ccstart, error) \ |
| # | Line 3299 static void compile_fallbackpath(compile | Line 3892 static void compile_fallbackpath(compile |
| 3892 | } \ | } \ |
| 3893 | while (0) | while (0) |
| 3894 | ||
| 3895 | #define FALLBACK_AS(type) ((type*)fallback) | #define FALLBACK_AS(type) ((type *)fallback) |
| 3896 | ||
| 3897 | static uschar *compile_ref_hotpath(compiler_common *common, uschar *cc, jump_list **fallbacks, BOOL withchecks, BOOL emptyfail) | static pcre_uchar *compile_ref_hotpath(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks, BOOL withchecks, BOOL emptyfail) |
| 3898 | { | { |
| 3899 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 3900 | int offset = GET2(cc, 1) << 1; | int offset = GET2(cc, 1) << 1; |
| 3901 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
| 3902 | struct sljit_jump *partial; | |
| 3903 | struct sljit_jump *nopartial; | |
| 3904 | ||
| 3905 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
| 3906 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | |
| 3907 | if (withchecks && !common->jscript_compat) | if (withchecks && !common->jscript_compat) |
| 3908 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); |
| 3909 | ||
| 3910 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF && defined SUPPORT_UCP |
| 3911 | #ifdef SUPPORT_UCP | if (common->utf && *cc == OP_REFI) |
| if (common->utf8 && *cc == OP_REFI) | ||
| 3912 | { | { |
| 3913 | SLJIT_ASSERT(TMP1 == SLJIT_TEMPORARY_REG1 && STACK_TOP == SLJIT_TEMPORARY_REG2 && TMP2 == SLJIT_TEMPORARY_REG3); | SLJIT_ASSERT(TMP1 == SLJIT_TEMPORARY_REG1 && STACK_TOP == SLJIT_TEMPORARY_REG2 && TMP2 == SLJIT_TEMPORARY_REG3); |
| 3914 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); |
| # | Line 3323 if (common->utf8 && *cc == OP_REFI) | Line 3918 if (common->utf8 && *cc == OP_REFI) |
| 3918 | /* Needed to save important temporary registers. */ | /* Needed to save important temporary registers. */ |
| 3919 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); |
| 3920 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); |
| 3921 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, ptr), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, uchar_ptr), STR_PTR, 0); |
| 3922 | sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf8caselesscmp)); | sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf_caselesscmp)); |
| 3923 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
| 3924 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); | if (common->mode == JIT_COMPILE) |
| 3925 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1)); | |
| 3926 | else | |
| 3927 | { | |
| 3928 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); | |
| 3929 | nopartial = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | |
| 3930 | check_partial(common, FALSE); | |
| 3931 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | |
| 3932 | JUMPHERE(nopartial); | |
| 3933 | } | |
| 3934 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); |
| 3935 | } | } |
| 3936 | else | else |
| 3937 | #endif | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
| #endif | ||
| 3938 | { | { |
| 3939 | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0); | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0); |
| 3940 | if (withchecks) | if (withchecks) |
| 3941 | jump = JUMP(SLJIT_C_ZERO); | jump = JUMP(SLJIT_C_ZERO); |
| 3942 | ||
| 3943 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
| 3944 | partial = CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0); | |
| 3945 | if (common->mode == JIT_COMPILE) | |
| 3946 | add_jump(compiler, fallbacks, partial); | |
| 3947 | ||
| add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); | ||
| 3948 | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); |
| 3949 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
| 3950 | ||
| 3951 | if (common->mode != JIT_COMPILE) | |
| 3952 | { | |
| 3953 | nopartial = JUMP(SLJIT_JUMP); | |
| 3954 | JUMPHERE(partial); | |
| 3955 | /* TMP2 -= STR_END - STR_PTR */ | |
| 3956 | OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, STR_PTR, 0); | |
| 3957 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, STR_END, 0); | |
| 3958 | partial = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0); | |
| 3959 | OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0); | |
| 3960 | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); | |
| 3961 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | |
| 3962 | JUMPHERE(partial); | |
| 3963 | check_partial(common, FALSE); | |
| 3964 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | |
| 3965 | JUMPHERE(nopartial); | |
| 3966 | } | |
| 3967 | } | } |
| 3968 | ||
| 3969 | if (jump != NULL) | if (jump != NULL) |
| # | Line 3350 if (jump != NULL) | Line 3973 if (jump != NULL) |
| 3973 | else | else |
| 3974 | JUMPHERE(jump); | JUMPHERE(jump); |
| 3975 | } | } |
| 3976 | return cc + 3; | return cc + 1 + IMM2_SIZE; |
| 3977 | } | } |
| 3978 | ||
| 3979 | static SLJIT_INLINE uschar *compile_ref_iterator_hotpath(compiler_common *common, uschar *cc, fallback_common *parent) | static SLJIT_INLINE pcre_uchar *compile_ref_iterator_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) |
| 3980 | { | { |
| 3981 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 3982 | fallback_common *fallback; | fallback_common *fallback; |
| 3983 | uschar type; | pcre_uchar type; |
| 3984 | struct sljit_label *label; | struct sljit_label *label; |
| 3985 | struct sljit_jump *zerolength; | struct sljit_jump *zerolength; |
| 3986 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
| 3987 | uschar *ccbegin = cc; | pcre_uchar *ccbegin = cc; |
| 3988 | int min = 0, max = 0; | int min = 0, max = 0; |
| 3989 | BOOL minimize; | BOOL minimize; |
| 3990 | ||
| 3991 | PUSH_FALLBACK(sizeof(iterator_fallback), cc, NULL); | PUSH_FALLBACK(sizeof(iterator_fallback), cc, NULL); |
| 3992 | ||
| 3993 | type = cc[3]; | type = cc[1 + IMM2_SIZE]; |
| 3994 | minimize = (type & 0x1) != 0; | minimize = (type & 0x1) != 0; |
| 3995 | switch(type) | switch(type) |
| 3996 | { | { |
| # | Line 3375 switch(type) | Line 3998 switch(type) |
| 3998 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
| 3999 | min = 0; | min = 0; |
| 4000 | max = 0; | max = 0; |
| 4001 | cc += 4; | cc += 1 + IMM2_SIZE + 1; |
| 4002 | break; | break; |
| 4003 | case OP_CRPLUS: | case OP_CRPLUS: |
| 4004 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
| 4005 | min = 1; | min = 1; |
| 4006 | max = 0; | max = 0; |
| 4007 | cc += 4; | cc += 1 + IMM2_SIZE + 1; |
| 4008 | break; | break; |
| 4009 | case OP_CRQUERY: | case OP_CRQUERY: |
| 4010 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
| 4011 | min = 0; | min = 0; |
| 4012 | max = 1; | max = 1; |
| 4013 | cc += 4; | cc += 1 + IMM2_SIZE + 1; |
| 4014 | break; | break; |
| 4015 | case OP_CRRANGE: | case OP_CRRANGE: |
| 4016 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 4017 | min = GET2(cc, 3 + 1); | min = GET2(cc, 1 + IMM2_SIZE + 1); |
| 4018 | max = GET2(cc, 3 + 3); | max = GET2(cc, 1 + IMM2_SIZE + 1 + IMM2_SIZE); |
| 4019 | cc += 8; | cc += 1 + IMM2_SIZE + 1 + 2 * IMM2_SIZE; |
| 4020 | break; | break; |
| 4021 | default: | default: |
| 4022 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
| # | Line 3453 if (!minimize) | Line 4076 if (!minimize) |
| 4076 | ||
| 4077 | JUMPHERE(zerolength); | JUMPHERE(zerolength); |
| 4078 | FALLBACK_AS(iterator_fallback)->hotpath = LABEL(); | FALLBACK_AS(iterator_fallback)->hotpath = LABEL(); |
| 4079 | ||
| 4080 | decrease_call_count(common); | |
| 4081 | return cc; | return cc; |
| 4082 | } | } |
| 4083 | ||
| # | Line 3490 else if (max > 0) | Line 4115 else if (max > 0) |
| 4115 | if (jump != NULL) | if (jump != NULL) |
| 4116 | JUMPHERE(jump); | JUMPHERE(jump); |
| 4117 | JUMPHERE(zerolength); | JUMPHERE(zerolength); |
| 4118 | ||
| 4119 | decrease_call_count(common); | |
| 4120 | return cc; | return cc; |
| 4121 | } | } |
| 4122 | ||
| 4123 | static SLJIT_INLINE uschar *compile_recurse_hotpath(compiler_common *common, uschar *cc, fallback_common *parent) | static SLJIT_INLINE pcre_uchar *compile_recurse_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) |
| 4124 | { | { |
| 4125 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 4126 | fallback_common *fallback; | fallback_common *fallback; |
| # | Line 3526 if (entry == NULL) | Line 4153 if (entry == NULL) |
| 4153 | common->entries = entry; | common->entries = entry; |
| 4154 | } | } |
| 4155 | ||
| 4156 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | if (common->has_set_som && common->mark_ptr != 0) |
| 4157 | allocate_stack(common, 1); | { |
| 4158 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
| 4159 | allocate_stack(common, 2); | |
| 4160 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
| 4161 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); | |
| 4162 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); | |
| 4163 | } | |
| 4164 | else if (common->has_set_som || common->mark_ptr != 0) | |
| 4165 | { | |
| 4166 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->has_set_som ? (int)(OVECTOR(0)) : common->mark_ptr); | |
| 4167 | allocate_stack(common, 1); | |
| 4168 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); | |
| 4169 | } | |
| 4170 | ||
| 4171 | if (entry->entry == NULL) | if (entry->entry == NULL) |
| 4172 | add_jump(compiler, &entry->calls, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &entry->calls, JUMP(SLJIT_FAST_CALL)); |
| # | Line 3539 add_jump(compiler, &fallback->topfallbac | Line 4177 add_jump(compiler, &fallback->topfallbac |
| 4177 | return cc + 1 + LINK_SIZE; | return cc + 1 + LINK_SIZE; |
| 4178 | } | } |
| 4179 | ||
| 4180 | static uschar *compile_assert_hotpath(compiler_common *common, uschar *cc, assert_fallback *fallback, BOOL conditional) | static pcre_uchar *compile_assert_hotpath(compiler_common *common, pcre_uchar *cc, assert_fallback *fallback, BOOL conditional) |
| 4181 | { | { |
| 4182 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 4183 | int framesize; | int framesize; |
| 4184 | int localptr; | int localptr; |
| 4185 | fallback_common altfallback; | fallback_common altfallback; |
| 4186 | uschar *ccbegin; | pcre_uchar *ccbegin; |
| 4187 | uschar opcode; | pcre_uchar opcode; |
| 4188 | uschar bra = OP_BRA; | pcre_uchar bra = OP_BRA; |
| 4189 | jump_list *tmp = NULL; | jump_list *tmp = NULL; |
| 4190 | jump_list **target = (conditional) ? &fallback->condfailed : &fallback->common.topfallbacks; | jump_list **target = (conditional) ? &fallback->condfailed : &fallback->common.topfallbacks; |
| 4191 | jump_list **found; | jump_list **found; |
| 4192 | /* Saving previous accept variables. */ | /* Saving previous accept variables. */ |
| 4193 | struct sljit_label *save_leavelabel = common->leavelabel; | |
| 4194 | struct sljit_label *save_acceptlabel = common->acceptlabel; | struct sljit_label *save_acceptlabel = common->acceptlabel; |
| 4195 | jump_list *save_leave = common->leave; | |
| 4196 | jump_list *save_accept = common->accept; | |
| 4197 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| 4198 | struct sljit_jump *brajump = NULL; | struct sljit_jump *brajump = NULL; |
| jump_list *save_accept = common->accept; | ||
| 4199 | ||
| 4200 | if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) | if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) |
| 4201 | { | { |
| # | Line 3563 if (*cc == OP_BRAZERO || *cc == OP_BRAMI | Line 4203 if (*cc == OP_BRAZERO || *cc == OP_BRAMI |
| 4203 | bra = *cc; | bra = *cc; |
| 4204 | cc++; | cc++; |
| 4205 | } | } |
| 4206 | localptr = PRIV(cc); | localptr = PRIV_DATA(cc); |
| 4207 | SLJIT_ASSERT(localptr != 0); | SLJIT_ASSERT(localptr != 0); |
| 4208 | framesize = get_framesize(common, cc, FALSE); | framesize = get_framesize(common, cc, FALSE); |
| 4209 | fallback->framesize = framesize; | fallback->framesize = framesize; |
| # | Line 3600 else | Line 4240 else |
| 4240 | } | } |
| 4241 | ||
| 4242 | memset(&altfallback, 0, sizeof(fallback_common)); | memset(&altfallback, 0, sizeof(fallback_common)); |
| 4243 | common->leavelabel = NULL; | |
| 4244 | common->leave = NULL; | |
| 4245 | while (1) | while (1) |
| 4246 | { | { |
| 4247 | common->acceptlabel = NULL; | common->acceptlabel = NULL; |
| # | Line 3614 while (1) | Line 4256 while (1) |
| 4256 | compile_hotpath(common, ccbegin + 1 + LINK_SIZE, cc, &altfallback); | compile_hotpath(common, ccbegin + 1 + LINK_SIZE, cc, &altfallback); |
| 4257 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) |
| 4258 | { | { |
| 4259 | common->leavelabel = save_leavelabel; | |
| 4260 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
| 4261 | common->leave = save_leave; | |
| 4262 | common->accept = save_accept; | common->accept = save_accept; |
| 4263 | return NULL; | return NULL; |
| 4264 | } | } |
| # | Line 3622 while (1) | Line 4266 while (1) |
| 4266 | if (common->accept != NULL) | if (common->accept != NULL) |
| 4267 | set_jumps(common->accept, common->acceptlabel); | set_jumps(common->accept, common->acceptlabel); |
| 4268 | ||
| 4269 | /* Reset stack. */ | |
| 4270 | if (framesize < 0) | if (framesize < 0) |
| 4271 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); |
| 4272 | else { | |
| 4273 | if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional) | |
| 4274 | { | |
| 4275 | /* We don't need to keep the STR_PTR, only the previous localptr. */ | |
| 4276 | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); | |
| 4277 | } | |
| 4278 | else | |
| 4279 | { | |
| 4280 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | |
| 4281 | add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); | |
| 4282 | } | |
| 4283 | } | |
| 4284 | ||
| 4285 | if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) | if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) |
| 4286 | { | { |
| 4287 | /* We know that STR_PTR was stored on the top of the stack. */ | /* We know that STR_PTR was stored on the top of the stack. */ |
| 4288 | if (conditional) | if (conditional) |
| 4289 | { | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); |
| if (framesize < 0) | ||
| OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); | ||
| else | ||
| { | ||
| OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | ||
| OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP1), (framesize + 1) * sizeof(sljit_w)); | ||
| } | ||
| } | ||
| 4290 | else if (bra == OP_BRAZERO) | else if (bra == OP_BRAZERO) |
| 4291 | { | { |
| 4292 | if (framesize < 0) | if (framesize < 0) |
| 4293 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); |
| 4294 | else | else |
| 4295 | { | { |
| OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | ||
| add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); | ||
| 4296 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); |
| 4297 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_w)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_w)); |
| 4298 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); |
| # | Line 3653 while (1) | Line 4300 while (1) |
| 4300 | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); |
| 4301 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); |
| 4302 | } | } |
| 4303 | else if (bra == OP_BRAMINZERO) | else if (framesize >= 0) |
| 4304 | { | { |
| 4305 | if (framesize >= 0) | /* For OP_BRA and OP_BRAMINZERO. */ |
| 4306 | { | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); |
| OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | ||
| add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); | ||
| } | ||
| 4307 | } | } |
| 4308 | } | } |
| 4309 | add_jump(compiler, found, JUMP(SLJIT_JUMP)); | add_jump(compiler, found, JUMP(SLJIT_JUMP)); |
| # | Line 3668 while (1) | Line 4311 while (1) |
| 4311 | compile_fallbackpath(common, altfallback.top); | compile_fallbackpath(common, altfallback.top); |
| 4312 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) |
| 4313 | { | { |
| 4314 | common->leavelabel = save_leavelabel; | |
| 4315 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
| 4316 | common->leave = save_leave; | |
| 4317 | common->accept = save_accept; | common->accept = save_accept; |
| 4318 | return NULL; | return NULL; |
| 4319 | } | } |
| # | Line 3681 while (1) | Line 4326 while (1) |
| 4326 | cc += GET(cc, 1); | cc += GET(cc, 1); |
| 4327 | } | } |
| 4328 | /* None of them matched. */ | /* None of them matched. */ |
| 4329 | if (common->leave != NULL) | |
| 4330 | set_jumps(common->leave, LABEL()); | |
| 4331 | ||
| 4332 | if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) | if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) |
| 4333 | { | { |
| 4334 | /* Assert is failed. */ | /* Assert is failed. */ |
| 4335 | if (conditional || bra == OP_BRAZERO) | if (conditional || bra == OP_BRAZERO) |
| 4336 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); |
| 4337 | ||
| 4338 | if (framesize < 0) | if (framesize < 0) |
| 4339 | { | { |
| 4340 | /* The topmost item should be 0. */ | /* The topmost item should be 0. */ |
| # | Line 3698 if (opcode == OP_ASSERT || opcode == OP_ | Line 4346 if (opcode == OP_ASSERT || opcode == OP_ |
| 4346 | else | else |
| 4347 | { | { |
| 4348 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); |
| if (framesize > 0) | ||
| OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2)); | ||
| 4349 | /* The topmost item should be 0. */ | /* The topmost item should be 0. */ |
| 4350 | if (bra == OP_BRAZERO) | if (bra == OP_BRAZERO) |
| 4351 | { | { |
| # | Line 3709 if (opcode == OP_ASSERT || opcode == OP_ | Line 4355 if (opcode == OP_ASSERT || opcode == OP_ |
| 4355 | else | else |
| 4356 | free_stack(common, framesize + 2); | free_stack(common, framesize + 2); |
| 4357 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); |
| if (framesize > 0) | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD, TMP2, 0); | ||
| 4358 | } | } |
| 4359 | jump = JUMP(SLJIT_JUMP); | jump = JUMP(SLJIT_JUMP); |
| 4360 | if (bra != OP_BRAZERO) | if (bra != OP_BRAZERO) |
| # | Line 3733 if (opcode == OP_ASSERT || opcode == OP_ | Line 4377 if (opcode == OP_ASSERT || opcode == OP_ |
| 4377 | } | } |
| 4378 | else | else |
| 4379 | { | { |
| 4380 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | if (bra == OP_BRA) |
| 4381 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP1), (framesize + 1) * sizeof(sljit_w)); | { |
| 4382 | if (bra == OP_BRAZERO) | /* We don't need to keep the STR_PTR, only the previous localptr. */ |
| 4383 | { | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); |
| 4384 | allocate_stack(common, 1); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); |
| OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); | ||
| 4385 | } | } |
| 4386 | else if (bra == OP_BRAMINZERO) | else |
| 4387 | { | { |
| 4388 | allocate_stack(common, 1); | /* We don't need to keep the STR_PTR, only the previous localptr. */ |
| 4389 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 2) * sizeof(sljit_w)); |
| 4390 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | |
| 4391 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0); | |
| 4392 | } | } |
| 4393 | } | } |
| 4394 | ||
| # | Line 3780 else | Line 4425 else |
| 4425 | { | { |
| 4426 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); |
| 4427 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); |
| if (framesize > 0) | ||
| OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2)); | ||
| 4428 | /* The topmost item should be 0. */ | /* The topmost item should be 0. */ |
| 4429 | if (bra != OP_BRA) | if (bra != OP_BRA) |
| 4430 | { | { |
| # | Line 3791 else | Line 4434 else |
| 4434 | else | else |
| 4435 | free_stack(common, framesize + 2); | free_stack(common, framesize + 2); |
| 4436 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); |
| if (framesize > 0) | ||
| OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS_HEAD, TMP2, 0); | ||
| 4437 | } | } |
| 4438 | ||
| 4439 | if (bra == OP_BRAZERO) | if (bra == OP_BRAZERO) |
| # | Line 3811 else | Line 4452 else |
| 4452 | } | } |
| 4453 | } | } |
| 4454 | ||
| 4455 | common->leavelabel = save_leavelabel; | |
| 4456 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
| 4457 | common->leave = save_leave; | |
| 4458 | common->accept = save_accept; | common->accept = save_accept; |
| 4459 | return cc + 1 + LINK_SIZE; | return cc + 1 + LINK_SIZE; |
| 4460 | } | } |
| 4461 | ||
| 4462 | static sljit_w SLJIT_CALL do_searchovector(sljit_w refno, sljit_w* locals, pcre_uchar *name_table) | |
| 4463 | { | |
| 4464 | int condition = FALSE; | |
| 4465 | pcre_uchar *slotA = name_table; | |
| 4466 | pcre_uchar *slotB; | |
| 4467 | sljit_w name_count = locals[LOCALS0 / sizeof(sljit_w)]; | |
| 4468 | sljit_w name_entry_size = locals[LOCALS1 / sizeof(sljit_w)]; | |
| 4469 | sljit_w no_capture; | |
| 4470 | int i; | |
| 4471 | ||
| 4472 | locals += refno & 0xff; | |
| 4473 | refno >>= 8; | |
| 4474 | no_capture = locals[1]; | |
| 4475 | ||
| 4476 | for (i = 0; i < name_count; i++) | |
| 4477 | { | |
| 4478 | if (GET2(slotA, 0) == refno) break; | |
| 4479 | slotA += name_entry_size; | |
| 4480 | } | |
| 4481 | ||
| 4482 | if (i < name_count) | |
| 4483 | { | |
| 4484 | /* Found a name for the number - there can be only one; duplicate names | |
| 4485 | for different numbers are allowed, but not vice versa. First scan down | |
| 4486 | for duplicates. */ | |
| 4487 | ||
| 4488 | slotB = slotA; | |
| 4489 | while (slotB > name_table) | |
| 4490 | { | |
| 4491 | slotB -= name_entry_size; | |
| 4492 | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) | |
| 4493 | { | |
| 4494 | condition = locals[GET2(slotB, 0) << 1] != no_capture; | |
| 4495 | if (condition) break; | |
| 4496 | } | |
| 4497 | else break; | |
| 4498 | } | |
| 4499 | ||
| 4500 | /* Scan up for duplicates */ | |
| 4501 | if (!condition) | |
| 4502 | { | |
| 4503 | slotB = slotA; | |
| 4504 | for (i++; i < name_count; i++) | |
| 4505 | { | |
| 4506 | slotB += name_entry_size; | |
| 4507 | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) | |
| 4508 | { | |
| 4509 | condition = locals[GET2(slotB, 0) << 1] != no_capture; | |
| 4510 | if (condition) break; | |
| 4511 | } | |
| 4512 | else break; | |
| 4513 | } | |
| 4514 | } | |
| 4515 | } | |
| 4516 | return condition; | |
| 4517 | } | |
| 4518 | ||
| 4519 | static sljit_w SLJIT_CALL do_searchgroups(sljit_w recno, sljit_w* locals, pcre_uchar *name_table) | |
| 4520 | { | |
| 4521 | int condition = FALSE; | |
| 4522 | pcre_uchar *slotA = name_table; | |
| 4523 | pcre_uchar *slotB; | |
| 4524 | sljit_w name_count = locals[LOCALS0 / sizeof(sljit_w)]; | |
| 4525 | sljit_w name_entry_size = locals[LOCALS1 / sizeof(sljit_w)]; | |
| 4526 | sljit_w group_num = locals[POSSESSIVE0 / sizeof(sljit_w)]; | |
| 4527 | int i; | |
| 4528 | ||
| 4529 | for (i = 0; i < name_count; i++) | |
| 4530 | { | |
| 4531 | if (GET2(slotA, 0) == recno) break; | |
| 4532 | slotA += name_entry_size; | |
| 4533 | } | |
| 4534 | ||
| 4535 | if (i < name_count) | |
| 4536 | { | |
| 4537 | /* Found a name for the number - there can be only one; duplicate | |
| 4538 | names for different numbers are allowed, but not vice versa. First | |
| 4539 | scan down for duplicates. */ | |
| 4540 | ||
| 4541 | slotB = slotA; | |
| 4542 | while (slotB > name_table) | |
| 4543 | { | |
| 4544 | slotB -= name_entry_size; | |
| 4545 | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) | |
| 4546 | { | |
| 4547 | condition = GET2(slotB, 0) == group_num; | |
| 4548 | if (condition) break; | |
| 4549 | } | |
| 4550 | else break; | |
| 4551 | } | |
| 4552 | ||
| 4553 | /* Scan up for duplicates */ | |
| 4554 | if (!condition) | |
| 4555 | { | |
| 4556 | slotB = slotA; | |
| 4557 | for (i++; i < name_count; i++) | |
| 4558 | { | |
| 4559 | slotB += name_entry_size; | |
| 4560 | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) | |
| 4561 | { | |
| 4562 | condition = GET2(slotB, 0) == group_num; | |
| 4563 | if (condition) break; | |
| 4564 | } | |
| 4565 | else break; | |
| 4566 | } | |
| 4567 | } | |
| 4568 | } | |
| 4569 | return condition; | |
| 4570 | } | |
| 4571 | ||
| 4572 | /* | /* |
| 4573 | Handling bracketed expressions is probably the most complex part. | Handling bracketed expressions is probably the most complex part. |
| 4574 | ||
| # | Line 3825 return cc + 1 + LINK_SIZE; | Line 4578 return cc + 1 + LINK_SIZE; |
| 4578 | A - Push the current STR_PTR. Needed for restoring the STR_PTR | A - Push the current STR_PTR. Needed for restoring the STR_PTR |
| 4579 | before the next alternative. Not pushed if there are no alternatives. | before the next alternative. Not pushed if there are no alternatives. |
| 4580 | M - Any values pushed by the current alternative. Can be empty, or anything. | M - Any values pushed by the current alternative. Can be empty, or anything. |
| 4581 | C - Push the previous OVECTOR(i), OVECTOR(i+1), MAX_INDEX and OVECTOR_PRIV(i) to the stack. | C - Push the previous OVECTOR(i), OVECTOR(i+1) and OVECTOR_PRIV(i) to the stack. |
| 4582 | L - Push the previous local (pointed by localptr) to the stack | L - Push the previous local (pointed by localptr) to the stack |
| 4583 | () - opional values stored on the stack | () - opional values stored on the stack |
| 4584 | ()* - optonal, can be stored multiple times | ()* - optonal, can be stored multiple times |
| # | Line 3863 return cc + 1 + LINK_SIZE; | Line 4616 return cc + 1 + LINK_SIZE; |
| 4616 | M - Any values pushed by the current alternative. Can be empty, or anything. | M - Any values pushed by the current alternative. Can be empty, or anything. |
| 4617 | ||
| 4618 | The next list shows the possible content of a bracket: | The next list shows the possible content of a bracket: |
| 4619 | (|) OP_*BRA | OP_ALT ... M A | (|) OP_*BRA | OP_ALT ... M A |
| 4620 | (?()|) OP_*COND | OP_ALT M A | (?()|) OP_*COND | OP_ALT M A |
| 4621 | (?>|) OP_ONCE | OP_ALT ... [stack trace] M A | (?>|) OP_ONCE | OP_ALT ... [stack trace] M A |
| 4622 | Or nothing, if trace is unnecessary | (?>|) OP_ONCE_NC | OP_ALT ... [stack trace] M A |
| 4623 | Or nothing, if trace is unnecessary | |
| 4624 | */ | */ |
| 4625 | ||
| 4626 | static uschar *compile_bracket_hotpath(compiler_common *common, uschar *cc, fallback_common *parent) | static pcre_uchar *compile_bracket_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) |
| 4627 | { | { |
| 4628 | DEFINE_COMPILER; | DEFINE_COMPILER; |
| 4629 | fallback_common *fallback; | fallback_common *fallback; |
| 4630 | uschar opcode; | pcre_uchar opcode; |
| 4631 | int localptr = 0; | int localptr = 0; |
| 4632 | int offset = 0; | int offset = 0; |
| 4633 | int stacksize; | int stacksize; |
| 4634 | uschar *ccbegin; | pcre_uchar *ccbegin; |
| 4635 | uschar *hotpath; | pcre_uchar *hotpath; |
| 4636 | uschar bra = OP_BRA; | pcre_uchar bra = OP_BRA; |
| 4637 | uschar ket; | pcre_uchar ket; |
| 4638 | assert_fallback *assert; | assert_fallback *assert; |
| 4639 | BOOL has_alternatives; | BOOL has_alternatives; |
| 4640 | struct sljit_jump *jump; | struct sljit_jump *jump; |
| # | Line 3899 if (*cc == OP_BRAZERO || *cc == OP_BRAMI | Line 4653 if (*cc == OP_BRAZERO || *cc == OP_BRAMI |
| 4653 | ||
| 4654 | opcode = *cc; | opcode = *cc; |
| 4655 | ccbegin = cc; | ccbegin = cc; |
| 4656 | hotpath = ccbegin + 1 + LINK_SIZE; | |
| 4657 | ||
| 4658 | if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) | if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) |
| 4659 | { | { |
| 4660 | /* Drop this bracket_fallback. */ | /* Drop this bracket_fallback. */ |
| # | Line 3910 ket = *(bracketend(cc) - 1 - LINK_SIZE); | Line 4666 ket = *(bracketend(cc) - 1 - LINK_SIZE); |
| 4666 | SLJIT_ASSERT(ket == OP_KET || ket == OP_KETRMAX || ket == OP_KETRMIN); | SLJIT_ASSERT(ket == OP_KET || ket == OP_KETRMAX || ket == OP_KETRMIN); |
| 4667 | SLJIT_ASSERT(!((bra == OP_BRAZERO && ket == OP_KETRMIN) || (bra == OP_BRAMINZERO && ket == OP_KETRMAX))); | SLJIT_ASSERT(!((bra == OP_BRAZERO && ket == OP_KETRMIN) || (bra == OP_BRAMINZERO && ket == OP_KETRMAX))); |
| 4668 | cc += GET(cc, 1); | cc += GET(cc, 1); |
| 4669 | has_alternatives = *cc == OP_ALT || opcode == OP_COND || opcode == OP_SCOND; | |
| 4670 | has_alternatives = *cc == OP_ALT; | |
| 4671 | if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) | |
| 4672 | { | |
| 4673 | has_alternatives = (*hotpath == OP_RREF) ? FALSE : TRUE; | |
| 4674 | if (*hotpath == OP_NRREF) | |
| 4675 | { | |
| 4676 | stacksize = GET2(hotpath, 1); | |
| 4677 | if (common->currententry == NULL || stacksize == RREF_ANY) | |
| 4678 | has_alternatives = FALSE; | |
| 4679 | else if (common->currententry->start == 0) | |
| 4680 | has_alternatives = stacksize != 0; | |
| 4681 | else | |
| 4682 | has_alternatives = stacksize != GET2(common->start, common->currententry->start + 1 + LINK_SIZE); | |
| 4683 | } | |
| 4684 | } | |
| 4685 | ||
| 4686 | if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) | if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) |
| 4687 | opcode = OP_SCOND; | opcode = OP_SCOND; |
| 4688 | if (SLJIT_UNLIKELY(opcode == OP_ONCE_NC)) | |
| 4689 | opcode = OP_ONCE; | |
| 4690 | ||
| 4691 | if (opcode == OP_CBRA || opcode == OP_SCBRA) | if (opcode == OP_CBRA || opcode == OP_SCBRA) |
| 4692 | { | { |
| # | Line 3921 if (opcode == OP_CBRA || opcode == OP_SC | Line 4695 if (opcode == OP_CBRA || opcode == OP_SC |
| 4695 | localptr = OVECTOR_PRIV(offset); | localptr = OVECTOR_PRIV(offset); |
| 4696 | offset <<= 1; | offset <<= 1; |
| 4697 | FALLBACK_AS(bracket_fallback)->localptr = localptr; | FALLBACK_AS(bracket_fallback)->localptr = localptr; |
| 4698 | hotpath += IMM2_SIZE; | |
| 4699 | } | } |
| 4700 | else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) | else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) |
| 4701 | { | { |
| 4702 | /* Other brackets simply allocate the next entry. */ | /* Other brackets simply allocate the next entry. */ |
| 4703 | localptr = PRIV(ccbegin); | localptr = PRIV_DATA(ccbegin); |
| 4704 | SLJIT_ASSERT(localptr != 0); | SLJIT_ASSERT(localptr != 0); |
| 4705 | FALLBACK_AS(bracket_fallback)->localptr = localptr; | FALLBACK_AS(bracket_fallback)->localptr = localptr; |
| 4706 | if (opcode == OP_ONCE) | if (opcode == OP_ONCE) |
| # | Line 4052 if (opcode == OP_ONCE) | Line 4827 if (opcode == OP_ONCE) |
| 4827 | else if (opcode == OP_CBRA || opcode == OP_SCBRA) | else if (opcode == OP_CBRA || opcode == OP_SCBRA) |
| 4828 | { | { |
| 4829 | /* Saving the previous values. */ | /* Saving the previous values. */ |
| 4830 | allocate_stack(common, 4); | allocate_stack(common, 3); |
| 4831 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
| 4832 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); |
| 4833 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); |
| 4834 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); |
| 4835 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); |
| OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), MAX_INDEX, 0); | ||
| 4836 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STR_PTR, 0); |
| 4837 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(3), TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP1, 0); |
| /* Update MAX_INDEX if necessary. */ | ||
| add_stub(common, max_index, (offset >> 1) + 1, CMP(SLJIT_C_LESS, MAX_INDEX, 0, SLJIT_IMM, (offset >> 1) + 1)); | ||
| 4838 | } | } |
| 4839 | else if (opcode == OP_SBRA || opcode == OP_SCOND) | else if (opcode == OP_SBRA || opcode == OP_SCOND) |
| 4840 | { | { |
| # | Line 4080 else if (has_alternatives) | Line 4852 else if (has_alternatives) |
| 4852 | } | } |
| 4853 | ||
| 4854 | /* Generating code for the first alternative. */ | /* Generating code for the first alternative. */ |
| hotpath = ccbegin + 1 + LINK_SIZE; | ||
| if (offset != 0) | ||
| hotpath += 2; | ||
| 4855 | if (opcode == OP_COND || opcode == OP_SCOND) | if (opcode == OP_COND || opcode == OP_SCOND) |
| 4856 | { | { |
| 4857 | if (*hotpath == OP_CREF) | if (*hotpath == OP_CREF) |
| 4858 | { | { |
| 4859 | SLJIT_ASSERT(has_alternatives); | |
| 4860 | add_jump(compiler, &(FALLBACK_AS(bracket_fallback)->u.condfailed), | add_jump(compiler, &(FALLBACK_AS(bracket_fallback)->u.condfailed), |
| 4861 | CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR((GET2(hotpath, 1) << 1)), SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); | CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(GET2(hotpath, 1) << 1), SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); |
| 4862 | hotpath += 3; | hotpath += 1 + IMM2_SIZE; |
| 4863 | } | |
| 4864 | else if (*hotpath == OP_NCREF) | |
| 4865 | { | |
| 4866 | SLJIT_ASSERT(has_alternatives); | |
| 4867 | stacksize = GET2(hotpath, 1); | |
| 4868 |