/[pcre]/code/trunk/pcre_jit_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_jit_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1015 by ph10, Sun Aug 26 16:07:14 2012 UTC revision 1414 by zherczeg, Sun Dec 22 16:27:35 2013 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) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 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) 2010-2012                        Copyright (c) 2010-2013
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 46  POSSIBILITY OF SUCH DAMAGE. Line 46  POSSIBILITY OF SUCH DAMAGE.
46    
47  #include "pcre_internal.h"  #include "pcre_internal.h"
48    
49  #ifdef SUPPORT_JIT  #if defined SUPPORT_JIT
50    
51  /* All-in-one: Since we use the JIT compiler only from here,  /* All-in-one: Since we use the JIT compiler only from here,
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
# Line 65  system files. */ Line 65  system files. */
65  #error Unsupported architecture  #error Unsupported architecture
66  #endif  #endif
67    
68    /* Defines for debugging purposes. */
69    
70    /* 1 - Use unoptimized capturing brackets.
71       2 - Enable capture_last_ptr (includes option 1). */
72    /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */
73    
74    /* 1 - Always have a control head. */
75    /* #define DEBUG_FORCE_CONTROL_HEAD 1 */
76    
77  /* Allocate memory for the regex stack on the real machine stack.  /* Allocate memory for the regex stack on the real machine stack.
78  Fast, but limited size. */  Fast, but limited size. */
79  #define MACHINE_STACK_SIZE 32768  #define MACHINE_STACK_SIZE 32768
# Line 157  typedef struct jit_arguments { Line 166  typedef struct jit_arguments {
166    int *offsets;    int *offsets;
167    pcre_uchar *uchar_ptr;    pcre_uchar *uchar_ptr;
168    pcre_uchar *mark_ptr;    pcre_uchar *mark_ptr;
169      void *callout_data;
170    /* Everything else after. */    /* Everything else after. */
171    int offsetcount;    pcre_uint32 limit_match;
172    int calllimit;    int real_offset_count;
173      int offset_count;
174    pcre_uint8 notbol;    pcre_uint8 notbol;
175    pcre_uint8 noteol;    pcre_uint8 noteol;
176    pcre_uint8 notempty;    pcre_uint8 notempty;
# Line 170  typedef struct executable_functions { Line 181  typedef struct executable_functions {
181    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];
182    PUBL(jit_callback) callback;    PUBL(jit_callback) callback;
183    void *userdata;    void *userdata;
184      pcre_uint32 top_bracket;
185      pcre_uint32 limit_match;
186    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];
187  } executable_functions;  } executable_functions;
188    
# Line 178  typedef struct jump_list { Line 191  typedef struct jump_list {
191    struct jump_list *next;    struct jump_list *next;
192  } jump_list;  } jump_list;
193    
 enum stub_types { stack_alloc };  
   
194  typedef struct stub_list {  typedef struct stub_list {
   enum stub_types type;  
   int data;  
195    struct sljit_jump *start;    struct sljit_jump *start;
196    struct sljit_label *quit;    struct sljit_label *quit;
197    struct stub_list *next;    struct stub_list *next;
198  } stub_list;  } stub_list;
199    
200    enum frame_types {
201      no_frame = -1,
202      no_stack = -2
203    };
204    
205    enum control_types {
206      type_mark = 0,
207      type_then_trap = 1
208    };
209    
210  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);
211    
212  /* The following structure is the key data type for the recursive  /* The following structure is the key data type for the recursive
213  code generator. It is allocated by compile_matchingpath, and contains  code generator. It is allocated by compile_matchingpath, and contains
214  the aguments for compile_backtrackingpath. Must be the first member  the arguments for compile_backtrackingpath. Must be the first member
215  of its descendants. */  of its descendants. */
216  typedef struct backtrack_common {  typedef struct backtrack_common {
217    /* Concatenation stack. */    /* Concatenation stack. */
# Line 208  typedef struct backtrack_common { Line 227  typedef struct backtrack_common {
227  typedef struct assert_backtrack {  typedef struct assert_backtrack {
228    backtrack_common common;    backtrack_common common;
229    jump_list *condfailed;    jump_list *condfailed;
230    /* Less than 0 (-1) if a frame is not needed. */    /* Less than 0 if a frame is not needed. */
231    int framesize;    int framesize;
232    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
233    int private_data_ptr;    int private_data_ptr;
# Line 229  typedef struct bracket_backtrack { Line 248  typedef struct bracket_backtrack {
248      /* Both for OP_COND, OP_SCOND. */      /* Both for OP_COND, OP_SCOND. */
249      jump_list *condfailed;      jump_list *condfailed;
250      assert_backtrack *assert;      assert_backtrack *assert;
251      /* For OP_ONCE. -1 if not needed. */      /* For OP_ONCE. Less than 0 if not needed. */
252      int framesize;      int framesize;
253    } u;    } u;
254    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
# Line 264  typedef struct recurse_entry { Line 283  typedef struct recurse_entry {
283    /* Collects the calls until the function is not created. */    /* Collects the calls until the function is not created. */
284    jump_list *calls;    jump_list *calls;
285    /* Points to the starting opcode. */    /* Points to the starting opcode. */
286    int start;    sljit_sw start;
287  } recurse_entry;  } recurse_entry;
288    
289  typedef struct recurse_backtrack {  typedef struct recurse_backtrack {
290    backtrack_common common;    backtrack_common common;
291      BOOL inlined_pattern;
292  } recurse_backtrack;  } recurse_backtrack;
293    
294    #define OP_THEN_TRAP OP_TABLE_LENGTH
295    
296    typedef struct then_trap_backtrack {
297      backtrack_common common;
298      /* If then_trap is not NULL, this structure contains the real
299      then_trap for the backtracking path. */
300      struct then_trap_backtrack *then_trap;
301      /* Points to the starting opcode. */
302      sljit_sw start;
303      /* Exit point for the then opcodes of this alternative. */
304      jump_list *quit;
305      /* Frame size of the current alternative. */
306      int framesize;
307    } then_trap_backtrack;
308    
309  #define MAX_RANGE_SIZE 6  #define MAX_RANGE_SIZE 6
310    
311  typedef struct compiler_common {  typedef struct compiler_common {
312      /* The sljit ceneric compiler. */
313    struct sljit_compiler *compiler;    struct sljit_compiler *compiler;
314      /* First byte code. */
315    pcre_uchar *start;    pcre_uchar *start;
   
316    /* Maps private data offset to each opcode. */    /* Maps private data offset to each opcode. */
317    int *private_data_ptrs;    sljit_si *private_data_ptrs;
318    /* Tells whether the capturing bracket is optimized. */    /* Tells whether the capturing bracket is optimized. */
319    pcre_uint8 *optimized_cbracket;    pcre_uint8 *optimized_cbracket;
320      /* Tells whether the starting offset is a target of then. */
321      pcre_uint8 *then_offsets;
322      /* Current position where a THEN must jump. */
323      then_trap_backtrack *then_trap;
324    /* Starting offset of private data for capturing brackets. */    /* Starting offset of private data for capturing brackets. */
325    int cbraptr;    int cbra_ptr;
326    /* OVector starting point. Must be divisible by 2. */    /* Output vector starting point. Must be divisible by 2. */
327    int ovector_start;    int ovector_start;
328    /* Last known position of the requested byte. */    /* Last known position of the requested byte. */
329    int req_char_ptr;    int req_char_ptr;
330    /* Head of the last recursion. */    /* Head of the last recursion. */
331    int recursive_head;    int recursive_head_ptr;
332    /* First inspected character for partial matching. */    /* First inspected character for partial matching. */
333    int start_used_ptr;    int start_used_ptr;
334    /* Starting pointer for partial soft matches. */    /* Starting pointer for partial soft matches. */
# Line 297  typedef struct compiler_common { Line 337  typedef struct compiler_common {
337    int first_line_end;    int first_line_end;
338    /* Points to the marked string. */    /* Points to the marked string. */
339    int mark_ptr;    int mark_ptr;
340      /* Recursive control verb management chain. */
341      int control_head_ptr;
342      /* Points to the last matched capture block index. */
343      int capture_last_ptr;
344      /* Points to the starting position of the current match. */
345      int start_ptr;
346    
347    /* Flipped and lower case tables. */    /* Flipped and lower case tables. */
348    const pcre_uint8 *fcc;    const pcre_uint8 *fcc;
349    sljit_w lcc;    sljit_sw lcc;
350    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */
351    int mode;    int mode;
352      /* \K is found in the pattern. */
353      BOOL has_set_som;
354      /* (*SKIP:arg) is found in the pattern. */
355      BOOL has_skip_arg;
356      /* (*THEN) is found in the pattern. */
357      BOOL has_then;
358      /* Needs to know the start position anytime. */
359      BOOL needs_start_ptr;
360      /* Currently in recurse or negative assert. */
361      BOOL local_exit;
362      /* Currently in a positive assert. */
363      BOOL positive_assert;
364    /* Newline control. */    /* Newline control. */
365    int nltype;    int nltype;
366    int newline;    int newline;
367    int bsr_nltype;    int bsr_nltype;
368    /* Dollar endonly. */    /* Dollar endonly. */
369    int endonly;    int endonly;
   BOOL has_set_som;  
370    /* Tables. */    /* Tables. */
371    sljit_w ctypes;    sljit_sw ctypes;
372    int digits[2 + MAX_RANGE_SIZE];    int digits[2 + MAX_RANGE_SIZE];
373    /* Named capturing brackets. */    /* Named capturing brackets. */
374    sljit_uw name_table;    pcre_uchar *name_table;
375    sljit_w name_count;    sljit_sw name_count;
376    sljit_w name_entry_size;    sljit_sw name_entry_size;
377    
378    /* Labels and jump lists. */    /* Labels and jump lists. */
379    struct sljit_label *partialmatchlabel;    struct sljit_label *partialmatchlabel;
380    struct sljit_label *quitlabel;    struct sljit_label *quit_label;
381    struct sljit_label *acceptlabel;    struct sljit_label *forced_quit_label;
382      struct sljit_label *accept_label;
383    stub_list *stubs;    stub_list *stubs;
384    recurse_entry *entries;    recurse_entry *entries;
385    recurse_entry *currententry;    recurse_entry *currententry;
386    jump_list *partialmatch;    jump_list *partialmatch;
387    jump_list *quit;    jump_list *quit;
388      jump_list *positive_assert_quit;
389      jump_list *forced_quit;
390    jump_list *accept;    jump_list *accept;
391    jump_list *calllimit;    jump_list *calllimit;
392    jump_list *stackalloc;    jump_list *stackalloc;
# Line 337  typedef struct compiler_common { Line 397  typedef struct compiler_common {
397    jump_list *vspace;    jump_list *vspace;
398    jump_list *casefulcmp;    jump_list *casefulcmp;
399    jump_list *caselesscmp;    jump_list *caselesscmp;
400      jump_list *reset_match;
401    BOOL jscript_compat;    BOOL jscript_compat;
402  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
403    BOOL utf;    BOOL utf;
404  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
405    BOOL use_ucp;    BOOL use_ucp;
406  #endif  #endif
407    #ifndef COMPILE_PCRE32
408    jump_list *utfreadchar;    jump_list *utfreadchar;
409    #endif
410  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
411    jump_list *utfreadtype8;    jump_list *utfreadtype8;
412  #endif  #endif
# Line 361  typedef struct compare_context { Line 424  typedef struct compare_context {
424  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED
425    int ucharptr;    int ucharptr;
426    union {    union {
427      sljit_i asint;      sljit_si asint;
428      sljit_uh asushort;      sljit_uh asushort;
429  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
430      sljit_ub asbyte;      sljit_ub asbyte;
431      sljit_ub asuchars[4];      sljit_ub asuchars[4];
432  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
433      sljit_uh asuchars[2];      sljit_uh asuchars[2];
434  #endif  #elif defined COMPILE_PCRE32
435        sljit_ui asuchars[1];
436  #endif  #endif
437    } c;    } c;
438    union {    union {
439      sljit_i asint;      sljit_si asint;
440      sljit_uh asushort;      sljit_uh asushort;
441  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
442      sljit_ub asbyte;      sljit_ub asbyte;
443      sljit_ub asuchars[4];      sljit_ub asuchars[4];
444  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
445      sljit_uh asuchars[2];      sljit_uh asuchars[2];
446  #endif  #elif defined COMPILE_PCRE32
447        sljit_ui asuchars[1];
448  #endif  #endif
449    } oc;    } oc;
450  #endif  #endif
451  } compare_context;  } compare_context;
452    
 enum {  
   frame_end = 0,  
   frame_setstrbegin = -1,  
   frame_setmark = -2  
 };  
   
453  /* Undefine sljit macros. */  /* Undefine sljit macros. */
454  #undef CMP  #undef CMP
455    
456  /* Used for accessing the elements of the stack. */  /* Used for accessing the elements of the stack. */
457  #define STACK(i)      ((-(i) - 1) * (int)sizeof(sljit_w))  #define STACK(i)      ((-(i) - 1) * (int)sizeof(sljit_sw))
458    
459  #define TMP1          SLJIT_TEMPORARY_REG1  #define TMP1          SLJIT_SCRATCH_REG1
460  #define TMP2          SLJIT_TEMPORARY_REG3  #define TMP2          SLJIT_SCRATCH_REG3
461  #define TMP3          SLJIT_TEMPORARY_EREG2  #define TMP3          SLJIT_TEMPORARY_EREG2
462  #define STR_PTR       SLJIT_SAVED_REG1  #define STR_PTR       SLJIT_SAVED_REG1
463  #define STR_END       SLJIT_SAVED_REG2  #define STR_END       SLJIT_SAVED_REG2
464  #define STACK_TOP     SLJIT_TEMPORARY_REG2  #define STACK_TOP     SLJIT_SCRATCH_REG2
465  #define STACK_LIMIT   SLJIT_SAVED_REG3  #define STACK_LIMIT   SLJIT_SAVED_REG3
466  #define ARGUMENTS     SLJIT_SAVED_EREG1  #define ARGUMENTS     SLJIT_SAVED_EREG1
467  #define CALL_COUNT    SLJIT_SAVED_EREG2  #define COUNT_MATCH   SLJIT_SAVED_EREG2
468  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1
469    
470  /* Local space layout. */  /* Local space layout. */
471  /* These two locals can be used by the current opcode. */  /* These two locals can be used by the current opcode. */
472  #define LOCALS0          (0 * sizeof(sljit_w))  #define LOCALS0          (0 * sizeof(sljit_sw))
473  #define LOCALS1          (1 * sizeof(sljit_w))  #define LOCALS1          (1 * sizeof(sljit_sw))
474  /* Two local variables for possessive quantifiers (char1 cannot use them). */  /* Two local variables for possessive quantifiers (char1 cannot use them). */
475  #define POSSESSIVE0      (2 * sizeof(sljit_w))  #define POSSESSIVE0      (2 * sizeof(sljit_sw))
476  #define POSSESSIVE1      (3 * sizeof(sljit_w))  #define POSSESSIVE1      (3 * sizeof(sljit_sw))
477  /* Max limit of recursions. */  /* Max limit of recursions. */
478  #define CALL_LIMIT       (4 * sizeof(sljit_w))  #define LIMIT_MATCH      (4 * sizeof(sljit_sw))
479  /* The output vector is stored on the stack, and contains pointers  /* The output vector is stored on the stack, and contains pointers
480  to characters. The vector data is divided into two groups: the first  to characters. The vector data is divided into two groups: the first
481  group contains the start / end character pointers, and the second is  group contains the start / end character pointers, and the second is
482  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. */
483  #define OVECTOR_START    (common->ovector_start)  #define OVECTOR_START    (common->ovector_start)
484  #define OVECTOR(i)       (OVECTOR_START + (i) * sizeof(sljit_w))  #define OVECTOR(i)       (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw))
485  #define OVECTOR_PRIV(i)  (common->cbraptr + (i) * sizeof(sljit_w))  #define OVECTOR_PRIV(i)  (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw))
486  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])
487    
488  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
489  #define MOV_UCHAR  SLJIT_MOV_UB  #define MOV_UCHAR  SLJIT_MOV_UB
490  #define MOVU_UCHAR SLJIT_MOVU_UB  #define MOVU_UCHAR SLJIT_MOVU_UB
491  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
492  #define MOV_UCHAR  SLJIT_MOV_UH  #define MOV_UCHAR  SLJIT_MOV_UH
493  #define MOVU_UCHAR SLJIT_MOVU_UH  #define MOVU_UCHAR SLJIT_MOVU_UH
494    #elif defined COMPILE_PCRE32
495    #define MOV_UCHAR  SLJIT_MOV_UI
496    #define MOVU_UCHAR SLJIT_MOVU_UI
497  #else  #else
498  #error Unsupported compiling mode  #error Unsupported compiling mode
499  #endif  #endif
 #endif  
500    
501  /* Shortcuts. */  /* Shortcuts. */
502  #define DEFINE_COMPILER \  #define DEFINE_COMPILER \
# Line 455  the start pointers when the end of the c Line 513  the start pointers when the end of the c
513    sljit_set_label(sljit_emit_jump(compiler, (type)), (label))    sljit_set_label(sljit_emit_jump(compiler, (type)), (label))
514  #define JUMPHERE(jump) \  #define JUMPHERE(jump) \
515    sljit_set_label((jump), sljit_emit_label(compiler))    sljit_set_label((jump), sljit_emit_label(compiler))
516    #define SET_LABEL(jump, label) \
517      sljit_set_label((jump), (label))
518  #define CMP(type, src1, src1w, src2, src2w) \  #define CMP(type, src1, src1w, src2, src2w) \
519    sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w))    sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w))
520  #define CMPTO(type, src1, src1w, src2, src2w, label) \  #define CMPTO(type, src1, src1w, src2, src2w, label) \
521    sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label))    sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label))
522  #define COND_VALUE(op, dst, dstw, type) \  #define OP_FLAGS(op, dst, dstw, src, srcw, type) \
523    sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type))    sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type))
524  #define GET_LOCAL_BASE(dst, dstw, offset) \  #define GET_LOCAL_BASE(dst, dstw, offset) \
525    sljit_get_local_base(compiler, (dst), (dstw), (offset))    sljit_get_local_base(compiler, (dst), (dstw), (offset))
526    
# Line 475  return cc; Line 535  return cc;
535    
536  /* Functions whose might need modification for all new supported opcodes:  /* Functions whose might need modification for all new supported opcodes:
537   next_opcode   next_opcode
538   get_private_data_length   check_opcode_types
539   set_private_data_ptrs   set_private_data_ptrs
540   get_framesize   get_framesize
541   init_frame   init_frame
542   get_private_data_length_for_copy   get_private_data_copy_length
543   copy_private_data   copy_private_data
544   compile_matchingpath   compile_matchingpath
545   compile_backtrackingpath   compile_backtrackingpath
# Line 503  switch(*cc) Line 563  switch(*cc)
563    case OP_WORDCHAR:    case OP_WORDCHAR:
564    case OP_ANY:    case OP_ANY:
565    case OP_ALLANY:    case OP_ALLANY:
566      case OP_NOTPROP:
567      case OP_PROP:
568    case OP_ANYNL:    case OP_ANYNL:
569    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
570    case OP_HSPACE:    case OP_HSPACE:
# Line 515  switch(*cc) Line 577  switch(*cc)
577    case OP_CIRCM:    case OP_CIRCM:
578    case OP_DOLL:    case OP_DOLL:
579    case OP_DOLLM:    case OP_DOLLM:
   case OP_TYPESTAR:  
   case OP_TYPEMINSTAR:  
   case OP_TYPEPLUS:  
   case OP_TYPEMINPLUS:  
   case OP_TYPEQUERY:  
   case OP_TYPEMINQUERY:  
   case OP_TYPEPOSSTAR:  
   case OP_TYPEPOSPLUS:  
   case OP_TYPEPOSQUERY:  
580    case OP_CRSTAR:    case OP_CRSTAR:
581    case OP_CRMINSTAR:    case OP_CRMINSTAR:
582    case OP_CRPLUS:    case OP_CRPLUS:
583    case OP_CRMINPLUS:    case OP_CRMINPLUS:
584    case OP_CRQUERY:    case OP_CRQUERY:
585    case OP_CRMINQUERY:    case OP_CRMINQUERY:
586      case OP_CRRANGE:
587      case OP_CRMINRANGE:
588      case OP_CRPOSSTAR:
589      case OP_CRPOSPLUS:
590      case OP_CRPOSQUERY:
591      case OP_CRPOSRANGE:
592      case OP_CLASS:
593      case OP_NCLASS:
594      case OP_REF:
595      case OP_REFI:
596      case OP_DNREF:
597      case OP_DNREFI:
598      case OP_RECURSE:
599      case OP_CALLOUT:
600      case OP_ALT:
601      case OP_KET:
602      case OP_KETRMAX:
603      case OP_KETRMIN:
604      case OP_KETRPOS:
605      case OP_REVERSE:
606      case OP_ASSERT:
607      case OP_ASSERT_NOT:
608      case OP_ASSERTBACK:
609      case OP_ASSERTBACK_NOT:
610      case OP_ONCE:
611      case OP_ONCE_NC:
612      case OP_BRA:
613      case OP_BRAPOS:
614      case OP_CBRA:
615      case OP_CBRAPOS:
616      case OP_COND:
617      case OP_SBRA:
618      case OP_SBRAPOS:
619      case OP_SCBRA:
620      case OP_SCBRAPOS:
621      case OP_SCOND:
622      case OP_CREF:
623      case OP_DNCREF:
624      case OP_RREF:
625      case OP_DNRREF:
626    case OP_DEF:    case OP_DEF:
627    case OP_BRAZERO:    case OP_BRAZERO:
628    case OP_BRAMINZERO:    case OP_BRAMINZERO:
629    case OP_BRAPOSZERO:    case OP_BRAPOSZERO:
630      case OP_PRUNE:
631      case OP_SKIP:
632      case OP_THEN:
633    case OP_COMMIT:    case OP_COMMIT:
634    case OP_FAIL:    case OP_FAIL:
635    case OP_ACCEPT:    case OP_ACCEPT:
636    case OP_ASSERT_ACCEPT:    case OP_ASSERT_ACCEPT:
637      case OP_CLOSE:
638    case OP_SKIPZERO:    case OP_SKIPZERO:
639    return cc + 1;    return cc + PRIV(OP_lengths)[*cc];
   
   case OP_ANYBYTE:  
 #ifdef SUPPORT_UTF  
   if (common->utf) return NULL;  
 #endif  
   return cc + 1;  
640    
641    case OP_CHAR:    case OP_CHAR:
642    case OP_CHARI:    case OP_CHARI:
# Line 557  switch(*cc) Line 648  switch(*cc)
648    case OP_MINPLUS:    case OP_MINPLUS:
649    case OP_QUERY:    case OP_QUERY:
650    case OP_MINQUERY:    case OP_MINQUERY:
651      case OP_UPTO:
652      case OP_MINUPTO:
653      case OP_EXACT:
654    case OP_POSSTAR:    case OP_POSSTAR:
655    case OP_POSPLUS:    case OP_POSPLUS:
656    case OP_POSQUERY:    case OP_POSQUERY:
657      case OP_POSUPTO:
658    case OP_STARI:    case OP_STARI:
659    case OP_MINSTARI:    case OP_MINSTARI:
660    case OP_PLUSI:    case OP_PLUSI:
661    case OP_MINPLUSI:    case OP_MINPLUSI:
662    case OP_QUERYI:    case OP_QUERYI:
663    case OP_MINQUERYI:    case OP_MINQUERYI:
664      case OP_UPTOI:
665      case OP_MINUPTOI:
666      case OP_EXACTI:
667    case OP_POSSTARI:    case OP_POSSTARI:
668    case OP_POSPLUSI:    case OP_POSPLUSI:
669    case OP_POSQUERYI:    case OP_POSQUERYI:
670      case OP_POSUPTOI:
671    case OP_NOTSTAR:    case OP_NOTSTAR:
672    case OP_NOTMINSTAR:    case OP_NOTMINSTAR:
673    case OP_NOTPLUS:    case OP_NOTPLUS:
674    case OP_NOTMINPLUS:    case OP_NOTMINPLUS:
675    case OP_NOTQUERY:    case OP_NOTQUERY:
676    case OP_NOTMINQUERY:    case OP_NOTMINQUERY:
677      case OP_NOTUPTO:
678      case OP_NOTMINUPTO:
679      case OP_NOTEXACT:
680    case OP_NOTPOSSTAR:    case OP_NOTPOSSTAR:
681    case OP_NOTPOSPLUS:    case OP_NOTPOSPLUS:
682    case OP_NOTPOSQUERY:    case OP_NOTPOSQUERY:
683      case OP_NOTPOSUPTO:
684    case OP_NOTSTARI:    case OP_NOTSTARI:
685    case OP_NOTMINSTARI:    case OP_NOTMINSTARI:
686    case OP_NOTPLUSI:    case OP_NOTPLUSI:
687    case OP_NOTMINPLUSI:    case OP_NOTMINPLUSI:
688    case OP_NOTQUERYI:    case OP_NOTQUERYI:
689    case OP_NOTMINQUERYI:    case OP_NOTMINQUERYI:
   case OP_NOTPOSSTARI:  
   case OP_NOTPOSPLUSI:  
   case OP_NOTPOSQUERYI:  
   cc += 2;  
 #ifdef SUPPORT_UTF  
   if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
   return cc;  
   
   case OP_UPTO:  
   case OP_MINUPTO:  
   case OP_EXACT:  
   case OP_POSUPTO:  
   case OP_UPTOI:  
   case OP_MINUPTOI:  
   case OP_EXACTI:  
   case OP_POSUPTOI:  
   case OP_NOTUPTO:  
   case OP_NOTMINUPTO:  
   case OP_NOTEXACT:  
   case OP_NOTPOSUPTO:  
690    case OP_NOTUPTOI:    case OP_NOTUPTOI:
691    case OP_NOTMINUPTOI:    case OP_NOTMINUPTOI:
692    case OP_NOTEXACTI:    case OP_NOTEXACTI:
693      case OP_NOTPOSSTARI:
694      case OP_NOTPOSPLUSI:
695      case OP_NOTPOSQUERYI:
696    case OP_NOTPOSUPTOI:    case OP_NOTPOSUPTOI:
697    cc += 2 + IMM2_SIZE;    cc += PRIV(OP_lengths)[*cc];
698  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
699    if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);    if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
700  #endif  #endif
701    return cc;    return cc;
702    
703    case OP_NOTPROP:    /* Special cases. */
704    case OP_PROP:    case OP_TYPESTAR:
705    return cc + 1 + 2;    case OP_TYPEMINSTAR:
706      case OP_TYPEPLUS:
707      case OP_TYPEMINPLUS:
708      case OP_TYPEQUERY:
709      case OP_TYPEMINQUERY:
710    case OP_TYPEUPTO:    case OP_TYPEUPTO:
711    case OP_TYPEMINUPTO:    case OP_TYPEMINUPTO:
712    case OP_TYPEEXACT:    case OP_TYPEEXACT:
713      case OP_TYPEPOSSTAR:
714      case OP_TYPEPOSPLUS:
715      case OP_TYPEPOSQUERY:
716    case OP_TYPEPOSUPTO:    case OP_TYPEPOSUPTO:
717    case OP_REF:    return cc + PRIV(OP_lengths)[*cc] - 1;
   case OP_REFI:  
   case OP_CREF:  
   case OP_NCREF:  
   case OP_RREF:  
   case OP_NRREF:  
   case OP_CLOSE:  
   cc += 1 + IMM2_SIZE;  
   return cc;  
718    
719    case OP_CRRANGE:    case OP_ANYBYTE:
720    case OP_CRMINRANGE:  #ifdef SUPPORT_UTF
721    return cc + 1 + 2 * IMM2_SIZE;    if (common->utf) return NULL;
722    #endif
723    case OP_CLASS:    return cc + 1;
   case OP_NCLASS:  
   return cc + 1 + 32 / sizeof(pcre_uchar);  
724    
725  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
726    case OP_XCLASS:    case OP_XCLASS:
727    return cc + GET(cc, 1);    return cc + GET(cc, 1);
728  #endif  #endif
729    
   case OP_RECURSE:  
   case OP_ASSERT:  
   case OP_ASSERT_NOT:  
   case OP_ASSERTBACK:  
   case OP_ASSERTBACK_NOT:  
   case OP_REVERSE:  
   case OP_ONCE:  
   case OP_ONCE_NC:  
   case OP_BRA:  
   case OP_BRAPOS:  
   case OP_COND:  
   case OP_SBRA:  
   case OP_SBRAPOS:  
   case OP_SCOND:  
   case OP_ALT:  
   case OP_KET:  
   case OP_KETRMAX:  
   case OP_KETRMIN:  
   case OP_KETRPOS:  
   return cc + 1 + LINK_SIZE;  
   
   case OP_CBRA:  
   case OP_CBRAPOS:  
   case OP_SCBRA:  
   case OP_SCBRAPOS:  
   return cc + 1 + LINK_SIZE + IMM2_SIZE;  
   
730    case OP_MARK:    case OP_MARK:
731      case OP_PRUNE_ARG:
732      case OP_SKIP_ARG:
733      case OP_THEN_ARG:
734    return cc + 1 + 2 + cc[1];    return cc + 1 + 2 + cc[1];
735    
736    default:    default:
737      /* All opcodes are supported now! */
738      SLJIT_ASSERT_STOP();
739    return NULL;    return NULL;
740    }    }
741  }  }
742    
743  #define CASE_ITERATOR_PRIVATE_DATA_1 \  static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)
     case OP_MINSTAR: \  
     case OP_MINPLUS: \  
     case OP_QUERY: \  
     case OP_MINQUERY: \  
     case OP_MINSTARI: \  
     case OP_MINPLUSI: \  
     case OP_QUERYI: \  
     case OP_MINQUERYI: \  
     case OP_NOTMINSTAR: \  
     case OP_NOTMINPLUS: \  
     case OP_NOTQUERY: \  
     case OP_NOTMINQUERY: \  
     case OP_NOTMINSTARI: \  
     case OP_NOTMINPLUSI: \  
     case OP_NOTQUERYI: \  
     case OP_NOTMINQUERYI:  
   
 #define CASE_ITERATOR_PRIVATE_DATA_2A \  
     case OP_STAR: \  
     case OP_PLUS: \  
     case OP_STARI: \  
     case OP_PLUSI: \  
     case OP_NOTSTAR: \  
     case OP_NOTPLUS: \  
     case OP_NOTSTARI: \  
     case OP_NOTPLUSI:  
   
 #define CASE_ITERATOR_PRIVATE_DATA_2B \  
     case OP_UPTO: \  
     case OP_MINUPTO: \  
     case OP_UPTOI: \  
     case OP_MINUPTOI: \  
     case OP_NOTUPTO: \  
     case OP_NOTMINUPTO: \  
     case OP_NOTUPTOI: \  
     case OP_NOTMINUPTOI:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \  
     case OP_TYPEMINSTAR: \  
     case OP_TYPEMINPLUS: \  
     case OP_TYPEQUERY: \  
     case OP_TYPEMINQUERY:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \  
     case OP_TYPESTAR: \  
     case OP_TYPEPLUS:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \  
     case OP_TYPEUPTO: \  
     case OP_TYPEMINUPTO:  
   
 static int get_class_iterator_size(pcre_uchar *cc)  
744  {  {
745  switch(*cc)  int count;
746    {  pcre_uchar *slot;
   case OP_CRSTAR:  
   case OP_CRPLUS:  
   return 2;  
   
   case OP_CRMINSTAR:  
   case OP_CRMINPLUS:  
   case OP_CRQUERY:  
   case OP_CRMINQUERY:  
   return 1;  
   
   case OP_CRRANGE:  
   case OP_CRMINRANGE:  
   if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))  
     return 0;  
   return 2;  
   
   default:  
   return 0;  
   }  
 }  
   
 static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  
 {  
 int private_data_length = 0;  
 pcre_uchar *alternative;  
 pcre_uchar *name;  
 pcre_uchar *end = NULL;  
 int space, size, bracketlen, i;  
747    
748  /* 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. */
749  while (cc < ccend)  while (cc < ccend)
750    {    {
   space = 0;  
   size = 0;  
   bracketlen = 0;  
751    switch(*cc)    switch(*cc)
752      {      {
753      case OP_SET_SOM:      case OP_SET_SOM:
# Line 785  while (cc < ccend) Line 761  while (cc < ccend)
761      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
762      break;      break;
763    
     case OP_ASSERT:  
     case OP_ASSERT_NOT:  
     case OP_ASSERTBACK:  
     case OP_ASSERTBACK_NOT:  
     case OP_ONCE:  
     case OP_ONCE_NC:  
     case OP_BRAPOS:  
     case OP_SBRA:  
     case OP_SBRAPOS:  
     private_data_length += sizeof(sljit_w);  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
764      case OP_CBRAPOS:      case OP_CBRAPOS:
765      case OP_SCBRAPOS:      case OP_SCBRAPOS:
     private_data_length += sizeof(sljit_w);  
766      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;
767      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;      cc += 1 + LINK_SIZE + IMM2_SIZE;
768      break;      break;
769    
770      case OP_COND:      case OP_COND:
771      case OP_SCOND:      case OP_SCOND:
772      bracketlen = cc[1 + LINK_SIZE];      /* Only AUTO_CALLOUT can insert this opcode. We do
773      if (bracketlen == OP_CREF)         not intend to support this case. */
774        {      if (cc[1 + LINK_SIZE] == OP_CALLOUT)
775        bracketlen = GET2(cc, 1 + LINK_SIZE + 1);        return FALSE;
776        common->optimized_cbracket[bracketlen] = 0;      cc += 1 + LINK_SIZE;
       }  
     else if (bracketlen == OP_NCREF)  
       {  
       bracketlen = GET2(cc, 1 + LINK_SIZE + 1);  
       name = (pcre_uchar *)common->name_table;  
       alternative = name;  
       for (i = 0; i < common->name_count; i++)  
         {  
         if (GET2(name, 0) == bracketlen) break;  
         name += common->name_entry_size;  
         }  
       SLJIT_ASSERT(i != common->name_count);  
   
       for (i = 0; i < common->name_count; i++)  
         {  
         if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0)  
           common->optimized_cbracket[GET2(alternative, 0)] = 0;  
         alternative += common->name_entry_size;  
         }  
       }  
   
     if (*cc == OP_COND)  
       {  
       /* Might be a hidden SCOND. */  
       alternative = cc + GET(cc, 1);  
       if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)  
         private_data_length += sizeof(sljit_w);  
       }  
     else  
       private_data_length += sizeof(sljit_w);  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_BRA:  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_CBRA:  
     case OP_SCBRA:  
     bracketlen = 1 + LINK_SIZE + IMM2_SIZE;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_1  
     space = 1;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2A  
     space = 2;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2B  
     space = 2;  
     size = -(2 + IMM2_SIZE);  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_1  
     space = 1;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2A  
     if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI)  
       space = 2;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2B  
     if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI)  
       space = 2;  
     size = 1 + IMM2_SIZE;  
777      break;      break;
778    
779      case OP_CLASS:      case OP_CREF:
780      case OP_NCLASS:      common->optimized_cbracket[GET2(cc, 1)] = 0;
781      size += 1 + 32 / sizeof(pcre_uchar);      cc += 1 + IMM2_SIZE;
     space = get_class_iterator_size(cc + size);  
782      break;      break;
783    
784  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8      case OP_DNREF:
785      case OP_XCLASS:      case OP_DNREFI:
786      size = GET(cc, 1);      case OP_DNCREF:
787      space = get_class_iterator_size(cc + size);      count = GET2(cc, 1 + IMM2_SIZE);
788        slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
789        while (count-- > 0)
790          {
791          common->optimized_cbracket[GET2(slot, 0)] = 0;
792          slot += common->name_entry_size;
793          }
794        cc += 1 + 2 * IMM2_SIZE;
795      break;      break;
 #endif  
796    
797      case OP_RECURSE:      case OP_RECURSE:
798      /* Set its value only once. */      /* Set its value only once. */
799      if (common->recursive_head == 0)      if (common->recursive_head_ptr == 0)
800        {        {
801        common->recursive_head = common->ovector_start;        common->recursive_head_ptr = common->ovector_start;
802        common->ovector_start += sizeof(sljit_w);        common->ovector_start += sizeof(sljit_sw);
803        }        }
804      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
805      break;      break;
806    
807        case OP_CALLOUT:
808        if (common->capture_last_ptr == 0)
809          {
810          common->capture_last_ptr = common->ovector_start;
811          common->ovector_start += sizeof(sljit_sw);
812          }
813        cc += 2 + 2 * LINK_SIZE;
814        break;
815    
816        case OP_THEN_ARG:
817        common->has_then = TRUE;
818        common->control_head_ptr = 1;
819        /* Fall through. */
820    
821        case OP_PRUNE_ARG:
822        common->needs_start_ptr = TRUE;
823        /* Fall through. */
824    
825      case OP_MARK:      case OP_MARK:
826      if (common->mark_ptr == 0)      if (common->mark_ptr == 0)
827        {        {
828        common->mark_ptr = common->ovector_start;        common->mark_ptr = common->ovector_start;
829        common->ovector_start += sizeof(sljit_w);        common->ovector_start += sizeof(sljit_sw);
830        }        }
831      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
832      break;      break;
833    
834        case OP_THEN:
835        common->has_then = TRUE;
836        common->control_head_ptr = 1;
837        /* Fall through. */
838    
839        case OP_PRUNE:
840        case OP_SKIP:
841        common->needs_start_ptr = TRUE;
842        cc += 1;
843        break;
844    
845        case OP_SKIP_ARG:
846        common->control_head_ptr = 1;
847        common->has_skip_arg = TRUE;
848        cc += 1 + 2 + cc[1];
849        break;
850    
851      default:      default:
852      cc = next_opcode(common, cc);      cc = next_opcode(common, cc);
853      if (cc == NULL)      if (cc == NULL)
854        return -1;        return FALSE;
855      break;      break;
856      }      }
857      }
858    return TRUE;
859    }
860    
861    if (space > 0 && cc >= end)  static int get_class_iterator_size(pcre_uchar *cc)
862      private_data_length += sizeof(sljit_w) * space;  {
863    switch(*cc)
864      {
865      case OP_CRSTAR:
866      case OP_CRPLUS:
867      return 2;
868    
869    if (size != 0)    case OP_CRMINSTAR:
870      {    case OP_CRMINPLUS:
871      if (size < 0)    case OP_CRQUERY:
872        {    case OP_CRMINQUERY:
873        cc += -size;    return 1;
874  #ifdef SUPPORT_UTF  
875        if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);    case OP_CRRANGE:
876  #endif    case OP_CRMINRANGE:
877        }    if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))
878      else      return 0;
879        cc += size;    return 2;
880    
881      default:
882      return 0;
883      }
884    }
885    
886    static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin)
887    {
888    pcre_uchar *end = bracketend(begin);
889    pcre_uchar *next;
890    pcre_uchar *next_end;
891    pcre_uchar *max_end;
892    pcre_uchar type;
893    sljit_sw length = end - begin;
894    int min, max, i;
895    
896    /* Detect fixed iterations first. */
897    if (end[-(1 + LINK_SIZE)] != OP_KET)
898      return FALSE;
899    
900    /* Already detected repeat. */
901    if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0)
902      return TRUE;
903    
904    next = end;
905    min = 1;
906    while (1)
907      {
908      if (*next != *begin)
909        break;
910      next_end = bracketend(next);
911      if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0)
912        break;
913      next = next_end;
914      min++;
915      }
916    
917    if (min == 2)
918      return FALSE;
919    
920    max = 0;
921    max_end = next;
922    if (*next == OP_BRAZERO || *next == OP_BRAMINZERO)
923      {
924      type = *next;
925      while (1)
926        {
927        if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin)
928          break;
929        next_end = bracketend(next + 2 + LINK_SIZE);
930        if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0)
931          break;
932        next = next_end;
933        max++;
934      }      }
935    
936    if (bracketlen > 0)    if (next[0] == type && next[1] == *begin && max >= 1)
937      {      {
938      if (cc >= end)      next_end = bracketend(next + 1);
939        if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0)
940        {        {
941        end = bracketend(cc);        for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE)
942        if (end[-1 - LINK_SIZE] == OP_KET)          if (*next_end != OP_KET)
943          end = NULL;            break;
944    
945          if (i == max)
946            {
947            common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end;
948            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO;
949            /* +2 the original and the last. */
950            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2;
951            if (min == 1)
952              return TRUE;
953            min--;
954            max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE);
955            }
956        }        }
     cc += bracketlen;  
957      }      }
958    }    }
959  return private_data_length;  
960    if (min >= 3)
961      {
962      common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end;
963      common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT;
964      common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min;
965      return TRUE;
966      }
967    
968    return FALSE;
969  }  }
970    
971  static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend)  #define CASE_ITERATOR_PRIVATE_DATA_1 \
972        case OP_MINSTAR: \
973        case OP_MINPLUS: \
974        case OP_QUERY: \
975        case OP_MINQUERY: \
976        case OP_MINSTARI: \
977        case OP_MINPLUSI: \
978        case OP_QUERYI: \
979        case OP_MINQUERYI: \
980        case OP_NOTMINSTAR: \
981        case OP_NOTMINPLUS: \
982        case OP_NOTQUERY: \
983        case OP_NOTMINQUERY: \
984        case OP_NOTMINSTARI: \
985        case OP_NOTMINPLUSI: \
986        case OP_NOTQUERYI: \
987        case OP_NOTMINQUERYI:
988    
989    #define CASE_ITERATOR_PRIVATE_DATA_2A \
990        case OP_STAR: \
991        case OP_PLUS: \
992        case OP_STARI: \
993        case OP_PLUSI: \
994        case OP_NOTSTAR: \
995        case OP_NOTPLUS: \
996        case OP_NOTSTARI: \
997        case OP_NOTPLUSI:
998    
999    #define CASE_ITERATOR_PRIVATE_DATA_2B \
1000        case OP_UPTO: \
1001        case OP_MINUPTO: \
1002        case OP_UPTOI: \
1003        case OP_MINUPTOI: \
1004        case OP_NOTUPTO: \
1005        case OP_NOTMINUPTO: \
1006        case OP_NOTUPTOI: \
1007        case OP_NOTMINUPTOI:
1008    
1009    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \
1010        case OP_TYPEMINSTAR: \
1011        case OP_TYPEMINPLUS: \
1012        case OP_TYPEQUERY: \
1013        case OP_TYPEMINQUERY:
1014    
1015    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \
1016        case OP_TYPESTAR: \
1017        case OP_TYPEPLUS:
1018    
1019    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \
1020        case OP_TYPEUPTO: \
1021        case OP_TYPEMINUPTO:
1022    
1023    static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend)
1024  {  {
1025  pcre_uchar *cc = common->start;  pcre_uchar *cc = common->start;
1026  pcre_uchar *alternative;  pcre_uchar *alternative;
1027  pcre_uchar *end = NULL;  pcre_uchar *end = NULL;
1028    int private_data_ptr = *private_data_start;
1029  int space, size, bracketlen;  int space, size, bracketlen;
1030    
1031  while (cc < ccend)  while (cc < ccend)
# Line 967  while (cc < ccend) Line 1033  while (cc < ccend)
1033    space = 0;    space = 0;
1034    size = 0;    size = 0;
1035    bracketlen = 0;    bracketlen = 0;
1036      if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE)
1037        return;
1038    
1039      if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND)
1040        if (detect_repeat(common, cc))
1041          {
1042          /* These brackets are converted to repeats, so no global
1043          based single character repeat is allowed. */
1044          if (cc >= end)
1045            end = bracketend(cc);
1046          }
1047    
1048    switch(*cc)    switch(*cc)
1049      {      {
1050        case OP_KET:
1051        if (common->private_data_ptrs[cc + 1 - common->start] != 0)
1052          {
1053          common->private_data_ptrs[cc - common->start] = private_data_ptr;
1054          private_data_ptr += sizeof(sljit_sw);
1055          cc += common->private_data_ptrs[cc + 1 - common->start];
1056          }
1057        cc += 1 + LINK_SIZE;
1058        break;
1059    
1060      case OP_ASSERT:      case OP_ASSERT:
1061      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1062      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 980  while (cc < ccend) Line 1068  while (cc < ccend)
1068      case OP_SBRAPOS:      case OP_SBRAPOS:
1069      case OP_SCOND:      case OP_SCOND:
1070      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
1071      private_data_ptr += sizeof(sljit_w);      private_data_ptr += sizeof(sljit_sw);
1072      bracketlen = 1 + LINK_SIZE;      bracketlen = 1 + LINK_SIZE;
1073      break;      break;
1074    
1075      case OP_CBRAPOS:      case OP_CBRAPOS:
1076      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1077      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
1078      private_data_ptr += sizeof(sljit_w);      private_data_ptr += sizeof(sljit_sw);
1079      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;
1080      break;      break;
1081    
# Line 997  while (cc < ccend) Line 1085  while (cc < ccend)
1085      if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)      if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)
1086        {        {
1087        common->private_data_ptrs[cc - common->start] = private_data_ptr;        common->private_data_ptrs[cc - common->start] = private_data_ptr;
1088        private_data_ptr += sizeof(sljit_w);        private_data_ptr += sizeof(sljit_sw);
1089        }        }
1090      bracketlen = 1 + LINK_SIZE;      bracketlen = 1 + LINK_SIZE;
1091      break;      break;
# Line 1062  while (cc < ccend) Line 1150  while (cc < ccend)
1150      break;      break;
1151      }      }
1152    
1153      /* Character iterators, which are not inside a repeated bracket,
1154         gets a private slot instead of allocating it on the stack. */
1155    if (space > 0 && cc >= end)    if (space > 0 && cc >= end)
1156      {      {
1157      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
1158      private_data_ptr += sizeof(sljit_w) * space;      private_data_ptr += sizeof(sljit_sw) * space;
1159      }      }
1160    
1161    if (size != 0)    if (size != 0)
# Line 1092  while (cc < ccend) Line 1182  while (cc < ccend)
1182      cc += bracketlen;      cc += bracketlen;
1183      }      }
1184    }    }
1185    *private_data_start = private_data_ptr;
1186  }  }
1187    
1188  /* Returns with -1 if no need for frame. */  /* Returns with a frame_types (always < 0) if no need for frame. */
1189  static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive)  static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL* needs_control_head)
1190  {  {
 pcre_uchar *ccend = bracketend(cc);  
1191  int length = 0;  int length = 0;
1192  BOOL possessive = FALSE;  int possessive = 0;
1193    BOOL stack_restore = FALSE;
1194  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1195  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1196    /* The last capture is a local variable even for recursions. */
1197    BOOL capture_last_found = FALSE;
1198    
1199    #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
1200    SLJIT_ASSERT(common->control_head_ptr != 0);
1201    *needs_control_head = TRUE;
1202    #else
1203    *needs_control_head = FALSE;
1204    #endif
1205    
1206  if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))  if (ccend == NULL)
1207    {    {
1208    length = 3;    ccend = bracketend(cc) - (1 + LINK_SIZE);
1209    possessive = TRUE;    if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))
1210        {
1211        possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;
1212        /* This is correct regardless of common->capture_last_ptr. */
1213        capture_last_found = TRUE;
1214        }
1215      cc = next_opcode(common, cc);
1216    }    }
1217    
 cc = next_opcode(common, cc);  
1218  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1219  while (cc < ccend)  while (cc < ccend)
1220    switch(*cc)    switch(*cc)
1221      {      {
1222      case OP_SET_SOM:      case OP_SET_SOM:
1223      SLJIT_ASSERT(common->has_set_som);      SLJIT_ASSERT(common->has_set_som);
1224        stack_restore = TRUE;
1225      if (!setsom_found)      if (!setsom_found)
1226        {        {
1227        length += 2;        length += 2;
# Line 1125  while (cc < ccend) Line 1231  while (cc < ccend)
1231      break;      break;
1232    
1233      case OP_MARK:      case OP_MARK:
1234        case OP_PRUNE_ARG:
1235        case OP_THEN_ARG:
1236      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1237        stack_restore = TRUE;
1238      if (!setmark_found)      if (!setmark_found)
1239        {        {
1240        length += 2;        length += 2;
1241        setmark_found = TRUE;        setmark_found = TRUE;
1242        }        }
1243        if (common->control_head_ptr != 0)
1244          *needs_control_head = TRUE;
1245      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
1246      break;      break;
1247    
1248      case OP_RECURSE:      case OP_RECURSE:
1249        stack_restore = TRUE;
1250      if (common->has_set_som && !setsom_found)      if (common->has_set_som && !setsom_found)
1251        {        {
1252        length += 2;        length += 2;
# Line 1145  while (cc < ccend) Line 1257  while (cc < ccend)
1257        length += 2;        length += 2;
1258        setmark_found = TRUE;        setmark_found = TRUE;
1259        }        }
1260        if (common->capture_last_ptr != 0 && !capture_last_found)
1261          {
1262          length += 2;
1263          capture_last_found = TRUE;
1264          }
1265      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1266      break;      break;
1267    
# Line 1152  while (cc < ccend) Line 1269  while (cc < ccend)
1269      case OP_CBRAPOS:      case OP_CBRAPOS:
1270      case OP_SCBRA:      case OP_SCBRA:
1271      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1272        stack_restore = TRUE;
1273        if (common->capture_last_ptr != 0 && !capture_last_found)
1274          {
1275          length += 2;
1276          capture_last_found = TRUE;
1277          }
1278      length += 3;      length += 3;
1279      cc += 1 + LINK_SIZE + IMM2_SIZE;      cc += 1 + LINK_SIZE + IMM2_SIZE;
1280      break;      break;
1281    
1282      default:      default:
1283        stack_restore = TRUE;
1284        /* Fall through. */
1285    
1286        case OP_NOT_WORD_BOUNDARY:
1287        case OP_WORD_BOUNDARY:
1288        case OP_NOT_DIGIT:
1289        case OP_DIGIT:
1290        case OP_NOT_WHITESPACE:
1291        case OP_WHITESPACE:
1292        case OP_NOT_WORDCHAR:
1293        case OP_WORDCHAR:
1294        case OP_ANY:
1295        case OP_ALLANY:
1296        case OP_ANYBYTE:
1297        case OP_NOTPROP:
1298        case OP_PROP:
1299        case OP_ANYNL:
1300        case OP_NOT_HSPACE:
1301        case OP_HSPACE:
1302        case OP_NOT_VSPACE:
1303        case OP_VSPACE:
1304        case OP_EXTUNI:
1305        case OP_EODN:
1306        case OP_EOD:
1307        case OP_CIRC:
1308        case OP_CIRCM:
1309        case OP_DOLL:
1310        case OP_DOLLM:
1311        case OP_CHAR:
1312        case OP_CHARI:
1313        case OP_NOT:
1314        case OP_NOTI:
1315    
1316        case OP_EXACT:
1317        case OP_POSSTAR:
1318        case OP_POSPLUS:
1319        case OP_POSQUERY:
1320        case OP_POSUPTO:
1321    
1322        case OP_EXACTI:
1323        case OP_POSSTARI:
1324        case OP_POSPLUSI:
1325        case OP_POSQUERYI:
1326        case OP_POSUPTOI:
1327    
1328        case OP_NOTEXACT:
1329        case OP_NOTPOSSTAR:
1330        case OP_NOTPOSPLUS:
1331        case OP_NOTPOSQUERY:
1332        case OP_NOTPOSUPTO:
1333    
1334        case OP_NOTEXACTI:
1335        case OP_NOTPOSSTARI:
1336        case OP_NOTPOSPLUSI:
1337        case OP_NOTPOSQUERYI:
1338        case OP_NOTPOSUPTOI:
1339    
1340        case OP_TYPEEXACT:
1341        case OP_TYPEPOSSTAR:
1342        case OP_TYPEPOSPLUS:
1343        case OP_TYPEPOSQUERY:
1344        case OP_TYPEPOSUPTO:
1345    
1346        case OP_CLASS:
1347        case OP_NCLASS:
1348        case OP_XCLASS:
1349    
1350      cc = next_opcode(common, cc);      cc = next_opcode(common, cc);
1351      SLJIT_ASSERT(cc != NULL);      SLJIT_ASSERT(cc != NULL);
1352      break;      break;
1353      }      }
1354    
1355  /* Possessive quantifiers can use a special case. */  /* Possessive quantifiers can use a special case. */
1356  if (SLJIT_UNLIKELY(possessive) && length == 3)  if (SLJIT_UNLIKELY(possessive == length))
1357    return -1;    return stack_restore ? no_frame : no_stack;
1358    
1359  if (length > 0)  if (length > 0)
1360    return length + 1;    return length + 1;
1361  return -1;  return stack_restore ? no_frame : no_stack;
1362  }  }
1363    
1364  static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive)  static void init_frame(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, int stackpos, int stacktop, BOOL recursive)
1365  {  {
1366  DEFINE_COMPILER;  DEFINE_COMPILER;
 pcre_uchar *ccend = bracketend(cc);  
1367  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1368  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1369    /* The last capture is a local variable even for recursions. */
1370    BOOL capture_last_found = FALSE;
1371  int offset;  int offset;
1372    
1373  /* >= 1 + shortest item size (2) */  /* >= 1 + shortest item size (2) */
# Line 1184  SLJIT_UNUSED_ARG(stacktop); Line 1375  SLJIT_UNUSED_ARG(stacktop);
1375  SLJIT_ASSERT(stackpos >= stacktop + 2);  SLJIT_ASSERT(stackpos >= stacktop + 2);
1376    
1377  stackpos = STACK(stackpos);  stackpos = STACK(stackpos);
1378  if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))  if (ccend == NULL)
1379    cc = next_opcode(common, cc);    {
1380      ccend = bracketend(cc) - (1 + LINK_SIZE);
1381      if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))
1382        cc = next_opcode(common, cc);
1383      }
1384    
1385  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1386  while (cc < ccend)  while (cc < ccend)
1387    switch(*cc)    switch(*cc)
# Line 1195  while (cc < ccend) Line 1391  while (cc < ccend)
1391      if (!setsom_found)      if (!setsom_found)
1392        {        {
1393        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));
1394        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0));
1395        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1396        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1397        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1398        setsom_found = TRUE;        setsom_found = TRUE;
1399        }        }
1400      cc += 1;      cc += 1;
1401      break;      break;
1402    
1403      case OP_MARK:      case OP_MARK:
1404        case OP_PRUNE_ARG:
1405        case OP_THEN_ARG:
1406      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1407      if (!setmark_found)      if (!setmark_found)
1408        {        {
1409        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);
1410        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr);
1411        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1412        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1413        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1414        setmark_found = TRUE;        setmark_found = TRUE;
1415        }        }
1416      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
# Line 1222  while (cc < ccend) Line 1420  while (cc < ccend)
1420      if (common->has_set_som && !setsom_found)      if (common->has_set_som && !setsom_found)
1421        {        {
1422        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));
1423        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0));
1424        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1425        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1426        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1427        setsom_found = TRUE;        setsom_found = TRUE;
1428        }        }
1429      if (common->mark_ptr != 0 && !setmark_found)      if (common->mark_ptr != 0 && !setmark_found)
1430        {        {
1431        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);
1432        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr);
1433        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1434        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1435        stackpos += (int)sizeof(sljit_w);        stackpos += (int)sizeof(sljit_sw);
1436        setmark_found = TRUE;        setmark_found = TRUE;
1437        }        }
1438        if (common->capture_last_ptr != 0 && !capture_last_found)
1439          {
1440          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr);
1441          OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr);
1442          stackpos += (int)sizeof(sljit_sw);
1443          OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1444          stackpos += (int)sizeof(sljit_sw);
1445          capture_last_found = TRUE;
1446          }
1447      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1448      break;      break;
1449    
# Line 1244  while (cc < ccend) Line 1451  while (cc < ccend)
1451      case OP_CBRAPOS:      case OP_CBRAPOS:
1452      case OP_SCBRA:      case OP_SCBRA:
1453      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1454        if (common->capture_last_ptr != 0 && !capture_last_found)
1455          {
1456          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr);
1457          OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr);
1458          stackpos += (int)sizeof(sljit_sw);
1459          OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1460          stackpos += (int)sizeof(sljit_sw);
1461          capture_last_found = TRUE;
1462          }
1463      offset = (GET2(cc, 1 + LINK_SIZE)) << 1;      offset = (GET2(cc, 1 + LINK_SIZE)) << 1;
1464      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset));      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset));
1465      stackpos += (int)sizeof(sljit_w);      stackpos += (int)sizeof(sljit_sw);
1466      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
1467      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));
1468      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
1469      stackpos += (int)sizeof(sljit_w);      stackpos += (int)sizeof(sljit_sw);
1470      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0);
1471      stackpos += (int)sizeof(sljit_w);      stackpos += (int)sizeof(sljit_sw);
1472    
1473      cc += 1 + LINK_SIZE + IMM2_SIZE;      cc += 1 + LINK_SIZE + IMM2_SIZE;
1474      break;      break;
# Line 1263  while (cc < ccend) Line 1479  while (cc < ccend)
1479      break;      break;
1480      }      }
1481    
1482  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0);
1483  SLJIT_ASSERT(stackpos == STACK(stacktop));  SLJIT_ASSERT(stackpos == STACK(stacktop));
1484  }  }
1485    
1486  static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  static SLJIT_INLINE int get_private_data_copy_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL needs_control_head)
1487  {  {
1488  int private_data_length = 2;  int private_data_length = needs_control_head ? 3 : 2;
1489  int size;  int size;
1490  pcre_uchar *alternative;  pcre_uchar *alternative;
1491  /* Calculate the sum of the private machine words. */  /* Calculate the sum of the private machine words. */
# Line 1278  while (cc < ccend) Line 1494  while (cc < ccend)
1494    size = 0;    size = 0;
1495    switch(*cc)    switch(*cc)
1496      {      {
1497        case OP_KET:
1498        if (PRIVATE_DATA(cc) != 0)
1499          private_data_length++;
1500        cc += 1 + LINK_SIZE;
1501        break;
1502    
1503      case OP_ASSERT:      case OP_ASSERT:
1504      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1505      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1382  return private_data_length; Line 1604  return private_data_length;
1604  }  }
1605    
1606  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,
1607    BOOL save, int stackptr, int stacktop)    BOOL save, int stackptr, int stacktop, BOOL needs_control_head)
1608  {  {
1609  DEFINE_COMPILER;  DEFINE_COMPILER;
1610  int srcw[2];  int srcw[2];
# Line 1403  stacktop = STACK(stacktop - 1); Line 1625  stacktop = STACK(stacktop - 1);
1625    
1626  if (!save)  if (!save)
1627    {    {
1628    stackptr += sizeof(sljit_w);    stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw);
1629    if (stackptr < stacktop)    if (stackptr < stacktop)
1630      {      {
1631      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);
1632      stackptr += sizeof(sljit_w);      stackptr += sizeof(sljit_sw);
1633      tmp1empty = FALSE;      tmp1empty = FALSE;
1634      }      }
1635    if (stackptr < stacktop)    if (stackptr < stacktop)
1636      {      {
1637      OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr);      OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr);
1638      stackptr += sizeof(sljit_w);      stackptr += sizeof(sljit_sw);
1639      tmp2empty = FALSE;      tmp2empty = FALSE;
1640      }      }
1641    /* The tmp1next must be TRUE in either way. */    /* The tmp1next must be TRUE in either way. */
1642    }    }
1643    
1644  while (status != end)  do
1645    {    {
1646    count = 0;    count = 0;
1647    switch(status)    switch(status)
1648      {      {
1649      case start:      case start:
1650      SLJIT_ASSERT(save && common->recursive_head != 0);      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);
1651      count = 1;      count = 1;
1652      srcw[0] = common->recursive_head;      srcw[0] = common->recursive_head_ptr;
1653        if (needs_control_head)
1654          {
1655          SLJIT_ASSERT(common->control_head_ptr != 0);
1656          count = 2;
1657          srcw[1] = common->control_head_ptr;
1658          }
1659      status = loop;      status = loop;
1660      break;      break;
1661    
# Line 1440  while (status != end) Line 1668  while (status != end)
1668    
1669      switch(*cc)      switch(*cc)
1670        {        {
1671          case OP_KET:
1672          if (PRIVATE_DATA(cc) != 0)
1673            {
1674            count = 1;
1675            srcw[0] = PRIVATE_DATA(cc);
1676            }
1677          cc += 1 + LINK_SIZE;
1678          break;
1679    
1680        case OP_ASSERT:        case OP_ASSERT:
1681        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
1682        case OP_ASSERTBACK:        case OP_ASSERTBACK:
# Line 1504  while (status != end) Line 1741  while (status != end)
1741          {          {
1742          count = 2;          count = 2;
1743          srcw[0] = PRIVATE_DATA(cc);          srcw[0] = PRIVATE_DATA(cc);
1744          srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_w);          srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw);
1745          }          }
1746        cc += 2;        cc += 2;
1747  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
# Line 1517  while (status != end) Line 1754  while (status != end)
1754          {          {
1755          count = 2;          count = 2;
1756          srcw[0] = PRIVATE_DATA(cc);          srcw[0] = PRIVATE_DATA(cc);
1757          srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_w);          srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw);
1758          }          }
1759        cc += 2 + IMM2_SIZE;        cc += 2 + IMM2_SIZE;
1760  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
# Line 1539  while (status != end) Line 1776  while (status != end)
1776          {          {
1777          count = 2;          count = 2;
1778          srcw[0] = PRIVATE_DATA(cc);          srcw[0] = PRIVATE_DATA(cc);
1779          srcw[1] = srcw[0] + sizeof(sljit_w);          srcw[1] = srcw[0] + sizeof(sljit_sw);
1780          }          }
1781        cc += 1;        cc += 1;
1782        break;        break;
# Line 1549  while (status != end) Line 1786  while (status != end)
1786          {          {
1787          count = 2;          count = 2;
1788          srcw[0] = PRIVATE_DATA(cc);          srcw[0] = PRIVATE_DATA(cc);
1789          srcw[1] = srcw[0] + sizeof(sljit_w);          srcw[1] = srcw[0] + sizeof(sljit_sw);
1790          }          }
1791        cc += 1 + IMM2_SIZE;        cc += 1 + IMM2_SIZE;
1792        break;        break;
# Line 1573  while (status != end) Line 1810  while (status != end)
1810            case 2:            case 2:
1811            count = 2;            count = 2;
1812            srcw[0] = PRIVATE_DATA(cc);            srcw[0] = PRIVATE_DATA(cc);
1813            srcw[1] = srcw[0] + sizeof(sljit_w);            srcw[1] = srcw[0] + sizeof(sljit_sw);
1814            break;            break;
1815    
1816            default:            default:
# Line 1605  while (status != end) Line 1842  while (status != end)
1842          if (!tmp1empty)          if (!tmp1empty)
1843            {            {
1844            OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);            OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);
1845            stackptr += sizeof(sljit_w);            stackptr += sizeof(sljit_sw);
1846            }            }
1847          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]);          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]);
1848          tmp1empty = FALSE;          tmp1empty = FALSE;
# Line 1616  while (status != end) Line 1853  while (status != end)
1853          if (!tmp2empty)          if (!tmp2empty)
1854            {            {
1855            OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);            OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);
1856            stackptr += sizeof(sljit_w);            stackptr += sizeof(sljit_sw);
1857            }            }
1858          OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]);          OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]);
1859          tmp2empty = FALSE;          tmp2empty = FALSE;
# Line 1633  while (status != end) Line 1870  while (status != end)
1870          if (!tmp1empty)          if (!tmp1empty)
1871            {            {
1872            OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);            OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);
1873            stackptr += sizeof(sljit_w);            stackptr += sizeof(sljit_sw);
1874            }            }
1875          tmp1next = FALSE;          tmp1next = FALSE;
1876          }          }
# Line 1645  while (status != end) Line 1882  while (status != end)
1882          if (!tmp2empty)          if (!tmp2empty)
1883            {            {
1884            OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr);            OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr);
1885            stackptr += sizeof(sljit_w);            stackptr += sizeof(sljit_sw);
1886            }            }
1887          tmp1next = TRUE;          tmp1next = TRUE;
1888          }          }
1889        }        }
1890      }      }
1891    }    }
1892    while (status != end);
1893    
1894  if (save)  if (save)
1895    {    {
# Line 1660  if (save) Line 1898  if (save)
1898      if (!tmp1empty)      if (!tmp1empty)
1899        {        {
1900        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);
1901        stackptr += sizeof(sljit_w);        stackptr += sizeof(sljit_sw);
1902        }        }
1903      if (!tmp2empty)      if (!tmp2empty)
1904        {        {
1905        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);
1906        stackptr += sizeof(sljit_w);        stackptr += sizeof(sljit_sw);
1907        }        }
1908      }      }
1909    else    else
# Line 1673  if (save) Line 1911  if (save)
1911      if (!tmp2empty)      if (!tmp2empty)
1912        {        {
1913        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0);
1914        stackptr += sizeof(sljit_w);        stackptr += sizeof(sljit_sw);
1915        }        }
1916      if (!tmp1empty)      if (!tmp1empty)
1917        {        {
1918        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0);
1919        stackptr += sizeof(sljit_w);        stackptr += sizeof(sljit_sw);
1920        }        }
1921      }      }
1922    }    }
1923  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));
1924  }  }
1925    
1926    static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset)
1927    {
1928    pcre_uchar *end = bracketend(cc);
1929    BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT;
1930    
1931    /* Assert captures then. */
1932    if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT)
1933      current_offset = NULL;
1934    /* Conditional block does not. */
1935    if (*cc == OP_COND || *cc == OP_SCOND)
1936      has_alternatives = FALSE;
1937    
1938    cc = next_opcode(common, cc);
1939    if (has_alternatives)
1940      current_offset = common->then_offsets + (cc - common->start);
1941    
1942    while (cc < end)
1943      {
1944      if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND))
1945        cc = set_then_offsets(common, cc, current_offset);
1946      else
1947        {
1948        if (*cc == OP_ALT && has_alternatives)
1949          current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start);
1950        if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL)
1951          *current_offset = 1;
1952        cc = next_opcode(common, cc);
1953        }
1954      }
1955    
1956    return end;
1957    }
1958    
1959  #undef CASE_ITERATOR_PRIVATE_DATA_1  #undef CASE_ITERATOR_PRIVATE_DATA_1
1960  #undef CASE_ITERATOR_PRIVATE_DATA_2A  #undef CASE_ITERATOR_PRIVATE_DATA_2A
1961  #undef CASE_ITERATOR_PRIVATE_DATA_2B  #undef CASE_ITERATOR_PRIVATE_DATA_2B
# Line 1692  SLJIT_ASSERT(cc == ccend && stackptr == Line 1963  SLJIT_ASSERT(cc == ccend && stackptr ==
1963  #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A  #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A
1964  #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B  #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B
1965    
1966  static SLJIT_INLINE BOOL ispowerof2(unsigned int value)  static SLJIT_INLINE BOOL is_powerof2(unsigned int value)
1967  {  {
1968  return (value & (value - 1)) == 0;  return (value & (value - 1)) == 0;
1969  }  }
# Line 1703  while (list) Line 1974  while (list)
1974    {    {
1975    /* sljit_set_label is clever enough to do nothing    /* sljit_set_label is clever enough to do nothing
1976    if either the jump or the label is NULL. */    if either the jump or the label is NULL. */
1977    sljit_set_label(list->jump, label);    SET_LABEL(list->jump, label);
1978    list = list->next;    list = list->next;
1979    }    }
1980  }  }
# Line 1719  if (list_item) Line 1990  if (list_item)
1990    }    }
1991  }  }
1992    
1993  static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start)  static void add_stub(compiler_common *common, struct sljit_jump *start)
1994  {  {
1995  DEFINE_COMPILER;  DEFINE_COMPILER;
1996  stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list));  stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list));
1997    
1998  if (list_item)  if (list_item)
1999    {    {
   list_item->type = type;  
   list_item->data = data;  
2000    list_item->start = start;    list_item->start = start;
2001    list_item->quit = LABEL();    list_item->quit = LABEL();
2002    list_item->next = common->stubs;    list_item->next = common->stubs;
# Line 1743  stub_list* list_item = common->stubs; Line 2012  stub_list* list_item = common->stubs;
2012  while (list_item)  while (list_item)
2013    {    {
2014    JUMPHERE(list_item->start);    JUMPHERE(list_item->start);
2015    switch(list_item->type)    add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL));
     {  
     case stack_alloc:  
     add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL));  
     break;  
     }  
2016    JUMPTO(SLJIT_JUMP, list_item->quit);    JUMPTO(SLJIT_JUMP, list_item->quit);
2017    list_item = list_item->next;    list_item = list_item->next;
2018    }    }
2019  common->stubs = NULL;  common->stubs = NULL;
2020  }  }
2021    
2022  static SLJIT_INLINE void decrease_call_count(compiler_common *common)  static SLJIT_INLINE void count_match(compiler_common *common)
2023  {  {
2024  DEFINE_COMPILER;  DEFINE_COMPILER;
2025    
2026  OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1);  OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
2027  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));
2028  }  }
2029    
# Line 1768  static SLJIT_INLINE void allocate_stack( Line 2032  static SLJIT_INLINE void allocate_stack(
2032  /* May destroy all locals and registers except TMP2. */  /* May destroy all locals and registers except TMP2. */
2033  DEFINE_COMPILER;  DEFINE_COMPILER;
2034    
2035  OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w));  OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw));
2036  #ifdef DESTROY_REGISTERS  #ifdef DESTROY_REGISTERS
2037  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345);
2038  OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);  OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
# Line 1776  OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); Line 2040  OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0);
2040  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0);
2041  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0);
2042  #endif  #endif
2043  add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0));  add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0));
2044  }  }
2045    
2046  static SLJIT_INLINE void free_stack(compiler_common *common, int size)  static SLJIT_INLINE void free_stack(compiler_common *common, int size)
2047  {  {
2048  DEFINE_COMPILER;  DEFINE_COMPILER;
2049  OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w));  OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw));
2050  }  }
2051    
2052  static SLJIT_INLINE void reset_ovector(compiler_common *common, int length)  static SLJIT_INLINE void reset_ovector(compiler_common *common, int length)
# Line 1790  static SLJIT_INLINE void reset_ovector(c Line 2054  static SLJIT_INLINE void reset_ovector(c
2054  DEFINE_COMPILER;  DEFINE_COMPILER;
2055  struct sljit_label *loop;  struct sljit_label *loop;
2056  int i;  int i;
2057    
2058  /* At this point we can freely use all temporary registers. */  /* At this point we can freely use all temporary registers. */
2059    SLJIT_ASSERT(length > 1);
2060  /* TMP1 returns with begin - 1. */  /* TMP1 returns with begin - 1. */
2061  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_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));
2062    if (length < 8)
2063      {
2064      for (i = 1; i < length; i++)
2065        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0);
2066      }
2067    else
2068      {
2069      GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START);
2070      OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length - 1);
2071      loop = LABEL();
2072      OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0);
2073      OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1);
2074      JUMPTO(SLJIT_C_NOT_ZERO, loop);
2075      }
2076    }
2077    
2078    static SLJIT_INLINE void do_reset_match(compiler_common *common, int length)
2079    {
2080    DEFINE_COMPILER;
2081    struct sljit_label *loop;
2082    int i;
2083    
2084    SLJIT_ASSERT(length > 1);
2085    /* OVECTOR(1) contains the "string begin - 1" constant. */
2086    if (length > 2)
2087      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
2088  if (length < 8)  if (length < 8)
2089    {    {
2090    for (i = 0; i < length; i++)    for (i = 2; i < length; i++)
2091      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_TEMPORARY_REG1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), TMP1, 0);
2092    }    }
2093  else  else
2094    {    {
2095    GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, OVECTOR_START - sizeof(sljit_w));    GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw));
2096    OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length);    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2);
2097    loop = LABEL();    loop = LABEL();
2098    OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_w), SLJIT_TEMPORARY_REG1, 0);    OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
2099    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1);    OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1);
2100    JUMPTO(SLJIT_C_NOT_ZERO, loop);    JUMPTO(SLJIT_C_NOT_ZERO, loop);
2101    }    }
2102    
2103    OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);
2104    if (common->mark_ptr != 0)
2105      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0);
2106    if (common->control_head_ptr != 0)
2107      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0);
2108    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));
2109    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr);
2110    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base));
2111    }
2112    
2113    static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg)
2114    {
2115    while (current != NULL)
2116      {
2117      switch (current[-2])
2118        {
2119        case type_then_trap:
2120        break;
2121    
2122        case type_mark:
2123        if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0)
2124          return current[-4];
2125        break;
2126    
2127        default:
2128        SLJIT_ASSERT_STOP();
2129        break;
2130        }
2131      current = (sljit_sw*)current[-1];
2132      }
2133    return -1;
2134  }  }
2135    
2136  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)
2137  {  {
2138  DEFINE_COMPILER;  DEFINE_COMPILER;
2139  struct sljit_label *loop;  struct sljit_label *loop;
2140  struct sljit_jump *earlyexit;  struct sljit_jump *early_quit;
2141    
2142  /* At this point we can freely use all registers. */  /* At this point we can freely use all registers. */
2143  OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));  OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
2144  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0);
2145    
2146  OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, ARGUMENTS, 0);
2147  if (common->mark_ptr != 0)  if (common->mark_ptr != 0)
2148    OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr);
2149  OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount));  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offset_count));
2150  if (common->mark_ptr != 0)  if (common->mark_ptr != 0)
2151    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_SCRATCH_REG3, 0);
2152  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_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int));
2153  OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin));  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, begin));
2154  GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START);  GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START);
2155  /* Unlikely, but possible */  /* Unlikely, but possible */
2156  earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0);  early_quit = CMP(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 0);
2157  loop = LABEL();  loop = LABEL();
2158  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_SCRATCH_REG1, 0);
2159  OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_w));  OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_sw));
2160  /* Copy the integer value to the output buffer */  /* Copy the integer value to the output buffer */
2161  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2162  OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1);  OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT);
2163  #endif  #endif
2164  OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0);  OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG3), sizeof(int), SLJIT_SAVED_REG2, 0);
2165  OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1);  OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1);
2166  JUMPTO(SLJIT_C_NOT_ZERO, loop);  JUMPTO(SLJIT_C_NOT_ZERO, loop);
2167  JUMPHERE(earlyexit);  JUMPHERE(early_quit);
2168    
2169  /* Calculate the return value, which is the maximum ovector value. */  /* Calculate the return value, which is the maximum ovector value. */
2170  if (topbracket > 1)  if (topbracket > 1)
2171    {    {
2172    GET_LOCAL_BASE(SLJIT_TEMPORARY_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_w));    GET_LOCAL_BASE(SLJIT_SCRATCH_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw));
2173    OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1);    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, topbracket + 1);
2174    
2175    /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */    /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */
2176    loop = LABEL();    loop = LABEL();
2177    OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w)));    OP1(SLJIT_MOVU, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), -(2 * (sljit_sw)sizeof(sljit_sw)));
2178    OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1);    OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1);
2179    CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop);    CMPTO(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG3, 0, loop);
2180    OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0);    OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_SCRATCH_REG2, 0);
2181    }    }
2182  else  else
2183    OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1);    OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1);
# Line 1862  else Line 2186  else
2186  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)
2187  {  {
2188  DEFINE_COMPILER;  DEFINE_COMPILER;
2189    struct sljit_jump *jump;
2190    
2191  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);
2192  SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));  SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0
2193      && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));
2194    
2195  OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0);
2196  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);
2197  OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount));  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count));
2198  CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, quit);  CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit);
2199    
2200  /* Store match begin and end. */  /* Store match begin and end. */
2201  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin));  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin));
2202  OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));
2203  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);  
2204    jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3);
2205    OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + (int)sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0);
2206    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2207    OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);
2208    #endif
2209    OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0);
2210    JUMPHERE(jump);
2211    
2212    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start);
2213  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);
2214  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2215  OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1);  OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT);
2216  #endif  #endif
2217  OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0);  OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(int), SLJIT_SAVED_REG2, 0);
2218    
2219  OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG1, 0);
2220  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2221  OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1);  OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);
2222  #endif  #endif
2223  OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0);  OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 0, SLJIT_SCRATCH_REG3, 0);
2224    
2225  JUMPTO(SLJIT_JUMP, quit);  JUMPTO(SLJIT_JUMP, quit);
2226  }  }
# Line 1998  if (c <= 127 && bit == 0x20) Line 2333  if (c <= 127 && bit == 0x20)
2333    return (0 << 8) | 0x20;    return (0 << 8) | 0x20;
2334    
2335  /* Since c != oc, they must have at least 1 bit difference. */  /* Since c != oc, they must have at least 1 bit difference. */
2336  if (!ispowerof2(bit))  if (!is_powerof2(bit))
2337    return 0;    return 0;
2338    
2339  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2340    
2341  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
2342  if (common->utf && c > 127)  if (common->utf && c > 127)
# Line 2017  if (common->utf && c > 127) Line 2352  if (common->utf && c > 127)
2352  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
2353  return (0 << 8) | bit;  return (0 << 8) | bit;
2354    
2355  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2356    
 #ifdef COMPILE_PCRE16  
2357  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
2358  if (common->utf && c > 65535)  if (common->utf && c > 65535)
2359    {    {
# Line 2030  if (common->utf && c > 65535) Line 2364  if (common->utf && c > 65535)
2364    }    }
2365  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
2366  return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8));  return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8));
 #endif /* COMPILE_PCRE16 */  
2367    
2368  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE[8|16|32] */
2369  }  }
2370    
2371  static void check_partial(compiler_common *common, BOOL force)  static void check_partial(compiler_common *common, BOOL force)
2372  {  {
2373  /* Checks whether a partial matching is occured. Does not modify registers. */  /* Checks whether a partial matching is occurred. Does not modify registers. */
2374  DEFINE_COMPILER;  DEFINE_COMPILER;
2375  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
2376    
# Line 2052  else if (common->mode == JIT_PARTIAL_SOF Line 2385  else if (common->mode == JIT_PARTIAL_SOF
2385    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);
2386    
2387  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2388    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2389  else  else
2390    {    {
2391    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
# Line 2065  if (jump != NULL) Line 2398  if (jump != NULL)
2398    JUMPHERE(jump);    JUMPHERE(jump);
2399  }  }
2400    
2401  static struct sljit_jump *check_str_end(compiler_common *common)  static void check_str_end(compiler_common *common, jump_list **end_reached)
2402  {  {
2403  /* Does not affect registers. Usually used in a tight spot. */  /* Does not affect registers. Usually used in a tight spot. */
2404  DEFINE_COMPILER;  DEFINE_COMPILER;
2405  struct sljit_jump *jump;  struct sljit_jump *jump;
 struct sljit_jump *nohit;  
 struct sljit_jump *return_value;  
2406    
2407  if (common->mode == JIT_COMPILE)  if (common->mode == JIT_COMPILE)
2408    return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);    {
2409      add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
2410      return;
2411      }
2412    
2413  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);
2414  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2415    {    {
2416    nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2417    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2418    JUMPHERE(nohit);    add_jump(compiler, end_reached, JUMP(SLJIT_JUMP));
   return_value = JUMP(SLJIT_JUMP);  
2419    }    }
2420  else  else
2421    {    {
2422    return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2423    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
2424      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
2425    else    else
2426      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
2427    }    }
2428  JUMPHERE(jump);  JUMPHERE(jump);
 return return_value;  
2429  }  }
2430    
2431  static void detect_partial_match(compiler_common *common, jump_list **backtracks)  static void detect_partial_match(compiler_common *common, jump_list **backtracks)
# Line 2112  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR Line 2444  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR
2444  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2445  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2446    {    {
2447    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2448    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
2449    }    }
2450  else  else
# Line 2130  static void read_char(compiler_common *c Line 2462  static void read_char(compiler_common *c
2462  /* Reads the character into TMP1, updates STR_PTR.  /* Reads the character into TMP1, updates STR_PTR.
2463  Does not check STR_END. TMP2 Destroyed. */  Does not check STR_END. TMP2 Destroyed. */
2464  DEFINE_COMPILER;  DEFINE_COMPILER;
2465  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2466  struct sljit_jump *jump;  struct sljit_jump *jump;
2467  #endif  #endif
2468    
2469  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2470  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2471  if (common->utf)  if (common->utf)
2472    {    {
2473  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2474    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2475  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
2476    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
2477  #endif  #endif /* COMPILE_PCRE[8|16] */
 #endif /* COMPILE_PCRE8 */  
2478    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2479    JUMPHERE(jump);    JUMPHERE(jump);
2480    }    }
2481  #endif  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2482  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2483  }  }
2484    
# Line 2157  static void peek_char(compiler_common *c Line 2487  static void peek_char(compiler_common *c
2487  /* Reads the character into TMP1, keeps STR_PTR.  /* Reads the character into TMP1, keeps STR_PTR.
2488  Does not check STR_END. TMP2 Destroyed. */  Does not check STR_END. TMP2 Destroyed. */
2489  DEFINE_COMPILER;  DEFINE_COMPILER;
2490  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2491  struct sljit_jump *jump;  struct sljit_jump *jump;
2492  #endif  #endif
2493    
2494  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2495  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2496  if (common->utf)  if (common->utf)
2497    {    {
2498  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2499    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2500  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
2501    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
2502  #endif  #endif /* COMPILE_PCRE[8|16] */
 #endif /* COMPILE_PCRE8 */  
2503    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2504    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2505    JUMPHERE(jump);    JUMPHERE(jump);
2506    }    }
2507  #endif  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2508  }  }
2509    
2510  static void read_char8_type(compiler_common *common)  static void read_char8_type(compiler_common *common)
2511  {  {
2512  /* 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. */
2513  DEFINE_COMPILER;  DEFINE_COMPILER;
2514  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2515  struct sljit_jump *jump;  struct sljit_jump *jump;
2516  #endif  #endif
2517    
# Line 2192  if (common->utf) Line 2520  if (common->utf)
2520    {    {
2521    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2522    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2523  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2524    /* This can be an extra read in some situations, but hopefully    /* This can be an extra read in some situations, but hopefully
2525    it is needed in most cases. */    it is needed in most cases. */
2526    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2527    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2528    add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));
2529    JUMPHERE(jump);    JUMPHERE(jump);
2530  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
2531    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2532    jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);    jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2533    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
# Line 2208  if (common->utf) Line 2535  if (common->utf)
2535    /* Skip low surrogate if necessary. */    /* Skip low surrogate if necessary. */
2536    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00);    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00);
2537    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800);
2538    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
2539    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);
2540    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2541  #endif  #elif defined COMPILE_PCRE32
2542  #endif /* COMPILE_PCRE8 */    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2543      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2544      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2545      JUMPHERE(jump);
2546    #endif /* COMPILE_PCRE[8|16|32] */
2547    return;    return;
2548    }    }
2549  #endif  #endif /* SUPPORT_UTF */
2550  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2551  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2552  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2553  /* The ctypes array contains only 256 values. */  /* The ctypes array contains only 256 values. */
2554  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2555  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2556  #endif  #endif
2557  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2558  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2559  JUMPHERE(jump);  JUMPHERE(jump);
2560  #endif  #endif
2561  }  }
# Line 2233  static void skip_char_back(compiler_comm Line 2564  static void skip_char_back(compiler_comm
2564  {  {
2565  /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */  /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */
2566  DEFINE_COMPILER;  DEFINE_COMPILER;
2567  #if defined SUPPORT_UTF && defined COMPILE_PCRE8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2568    #if defined COMPILE_PCRE8
2569  struct sljit_label *label;  struct sljit_label *label;
2570    
2571  if (common->utf)  if (common->utf)
# Line 2245  if (common->utf) Line 2577  if (common->utf)
2577    CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label);    CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label);
2578    return;    return;
2579    }    }
2580  #endif  #elif defined COMPILE_PCRE16
 #if defined SUPPORT_UTF && defined COMPILE_PCRE16  
2581  if (common->utf)  if (common->utf)
2582    {    {
2583    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
# Line 2254  if (common->utf) Line 2585  if (common->utf)
2585    /* Skip low surrogate if necessary. */    /* Skip low surrogate if necessary. */
2586    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
2587    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00);
2588    COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
2589    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
2590    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
2591    return;    return;
2592    }    }
2593  #endif  #endif /* COMPILE_PCRE[8|16] */
2594    #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2595  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2596  }  }
2597    
# Line 2276  if (nltype == NLTYPE_ANY) Line 2608  if (nltype == NLTYPE_ANY)
2608  else if (nltype == NLTYPE_ANYCRLF)  else if (nltype == NLTYPE_ANYCRLF)
2609    {    {
2610    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR);
2611    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
2612    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);
2613    COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
2614    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
2615    }    }
2616  else  else
# Line 2290  else Line 2622  else
2622    
2623  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
2624    
2625  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
2626  static void do_utfreadchar(compiler_common *common)  static void do_utfreadchar(compiler_common *common)
2627  {  {
2628  /* Fast decoding a UTF-8 character. TMP1 contains the first byte  /* Fast decoding a UTF-8 character. TMP1 contains the first byte
# Line 2378  sljit_emit_fast_return(compiler, RETURN_ Line 2710  sljit_emit_fast_return(compiler, RETURN_
2710  JUMPHERE(jump);  JUMPHERE(jump);
2711    
2712  /* We only have types for characters less than 256. */  /* We only have types for characters less than 256. */
2713  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(utf8_table4) - 0xc0);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2714  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
2715  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2716  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2717  }  }
2718    
2719  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
2720    
 #ifdef COMPILE_PCRE16  
2721  static void do_utfreadchar(compiler_common *common)  static void do_utfreadchar(compiler_common *common)
2722  {  {
2723  /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char  /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char
# Line 2411  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC Line 2742  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC
2742  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2743  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2744  }  }
 #endif /* COMPILE_PCRE16 */  
2745    
2746  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE[8|16] */
2747    
2748  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
2749    
# Line 2433  SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si Line 2763  SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si
2763    
2764  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2765  OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);  OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
2766  OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(ucd_stage1));  OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_stage1));
2767  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK);  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK);
2768  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
2769  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
2770  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2));  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2));
2771  OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1);  OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1);
2772  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype));  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype));
2773  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3);
2774  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2775  }  }
# Line 2453  struct sljit_label *newlinelabel = NULL; Line 2783  struct sljit_label *newlinelabel = NULL;
2783  struct sljit_jump *start;  struct sljit_jump *start;
2784  struct sljit_jump *end = NULL;  struct sljit_jump *end = NULL;
2785  struct sljit_jump *nl = NULL;  struct sljit_jump *nl = NULL;
2786  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2787  struct sljit_jump *singlechar;  struct sljit_jump *singlechar;
2788  #endif  #endif
2789  jump_list *newline = NULL;  jump_list *newline = NULL;
# Line 2508  if (newlinecheck) Line 2838  if (newlinecheck)
2838    end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);    end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
2839    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2840    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);
2841    COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
2842  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2843    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
2844  #endif  #endif
2845    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
2846    nl = JUMP(SLJIT_JUMP);    nl = JUMP(SLJIT_JUMP);
# Line 2531  if (newlinecheck) Line 2861  if (newlinecheck)
2861    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);
2862    
2863  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2864  #if defined SUPPORT_UTF && defined COMPILE_PCRE8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2865    #if defined COMPILE_PCRE8
2866  if (common->utf)  if (common->utf)
2867    {    {
2868    singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2869    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2870    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
2871    JUMPHERE(singlechar);    JUMPHERE(singlechar);
2872    }    }
2873  #endif  #elif defined COMPILE_PCRE16
 #if defined SUPPORT_UTF && defined COMPILE_PCRE16  
2874  if (common->utf)  if (common->utf)
2875    {    {
2876    singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);    singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
2877    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
2878    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2879    COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
2880    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
2881    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
2882    JUMPHERE(singlechar);    JUMPHERE(singlechar);
2883    }    }
2884  #endif  #endif /* COMPILE_PCRE[8|16] */
2885    #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2886  JUMPHERE(start);  JUMPHERE(start);
2887    
2888  if (newlinecheck)  if (newlinecheck)
# Line 2563  if (newlinecheck) Line 2894  if (newlinecheck)
2894  return mainloop;  return mainloop;
2895  }  }
2896    
2897  static SLJIT_INLINE BOOL fast_forward_first_two_chars(compiler_common *common, BOOL firstline)  #define MAX_N_CHARS 3
2898    
2899    static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)
2900  {  {
2901  DEFINE_COMPILER;  DEFINE_COMPILER;
2902  struct sljit_label *start;  struct sljit_label *start;
2903  struct sljit_jump *quit;  struct sljit_jump *quit;
2904  struct sljit_jump *found;  pcre_uint32 chars[MAX_N_CHARS * 2];
2905  pcre_int32 chars[4];  pcre_uchar *cc = common->start + 1 + LINK_SIZE;
 pcre_uchar *cc = common->start + 1 + IMM2_SIZE;  
2906  int location = 0;  int location = 0;
2907  pcre_int32 len, c, bit, caseless;  pcre_int32 len, c, bit, caseless;
2908  BOOL must_end;  int must_stop;
   
 #ifdef COMPILE_PCRE8  
 union {  
     sljit_uh ascombined;  
     sljit_ub asuchars[2];  
 } pair;  
 #else  
 union {  
     sljit_ui ascombined;  
     sljit_uh asuchars[2];  
 } pair;  
 #endif  
2909    
2910    /* We do not support alternatives now. */
2911  if (*(common->start + GET(common->start, 1)) == OP_ALT)  if (*(common->start + GET(common->start, 1)) == OP_ALT)
2912    return FALSE;    return FALSE;
2913    
2914  while (TRUE)  while (TRUE)
2915    {    {
2916    caseless = 0;    caseless = 0;
2917    must_end = TRUE;    must_stop = 1;
2918    switch(*cc)    switch(*cc)
2919      {      {
2920      case OP_CHAR:      case OP_CHAR:
2921      must_end = FALSE;      must_stop = 0;
2922      cc++;      cc++;
2923      break;      break;
2924    
2925      case OP_CHARI:      case OP_CHARI:
2926      caseless = 1;      caseless = 1;
2927      must_end = FALSE;      must_stop = 0;
2928      cc++;      cc++;
2929      break;      break;
2930    
# Line 2645  while (TRUE) Line 2966  while (TRUE)
2966      break;      break;
2967    
2968      default:      default:
2969      return FALSE;      must_stop = 2;
2970        break;
2971      }      }
2972    
2973      if (must_stop == 2)
2974          break;
2975    
2976    len = 1;    len = 1;
2977  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
2978    if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);    if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);
# Line 2670  while (TRUE) Line 2995  while (TRUE)
2995    else    else
2996      caseless = 0;      caseless = 0;
2997    
2998    while (len > 0 && location < 2 * 2)    while (len > 0 && location < MAX_N_CHARS * 2)
2999      {      {
3000      c = *cc;      c = *cc;
3001      bit = 0;      bit = 0;
# Line 2688  while (TRUE) Line 3013  while (TRUE)
3013      cc++;      cc++;
3014      }      }
3015    
3016    if (location == 2 * 2)    if (location >= MAX_N_CHARS * 2 || must_stop != 0)
3017      break;      break;
   else if (must_end)  
     return FALSE;  
3018    }    }
3019    
3020    /* At least two characters are required. */
3021    if (location < 2 * 2)
3022        return FALSE;
3023    
3024  if (firstline)  if (firstline)
3025    {    {
3026    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
3027    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
3028    OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, 1);    OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));
3029    }    }
3030  else  else
3031    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, 1);    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));
3032    
3033  start = LABEL();  start = LABEL();
3034  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
 #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED  
 #ifdef COMPILE_PCRE8  
 OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  
 #else /* COMPILE_PCRE8 */  
 OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  
 #endif  
   
 #else /* SLJIT_UNALIGNED */  
3035    
3036  #if defined SLJIT_BIG_ENDIAN && SLJIT_BIG_ENDIAN  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
 OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
 #else /* SLJIT_BIG_ENDIAN */  
3037  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
3038  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3039  #endif /* SLJIT_BIG_ENDIAN */  if (chars[1] != 0)
3040      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);
3041  #ifdef COMPILE_PCRE8  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);
3042  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 8);  if (location > 2 * 2)
3043  #else /* COMPILE_PCRE8 */    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
3044  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 16);  if (chars[3] != 0)
3045  #endif    OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]);
3046  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start);
3047    if (location > 2 * 2)
3048  #endif    {
3049      if (chars[5] != 0)
3050  if (chars[1] != 0 || chars[3] != 0)      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]);
3051    {    CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start);
   pair.asuchars[0] = chars[1];  
   pair.asuchars[1] = chars[3];  
   OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, pair.ascombined);  
3052    }    }
3053    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3054    
 pair.asuchars[0] = chars[0];  
 pair.asuchars[1] = chars[2];  
 found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, pair.ascombined);  
   
 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
 JUMPTO(SLJIT_JUMP, start);  
 JUMPHERE(found);  
3055  JUMPHERE(quit);  JUMPHERE(quit);
3056    
3057  if (firstline)  if (firstline)
3058    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3059  else  else
3060    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, 1);    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));
3061  return TRUE;  return TRUE;
3062  }  }
3063    
3064    #undef MAX_N_CHARS
3065    
3066  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)
3067  {  {
3068  DEFINE_COMPILER;  DEFINE_COMPILER;
# Line 2787  if (first_char == oc) Line 3096  if (first_char == oc)
3096  else  else
3097    {    {
3098    bit = first_char ^ oc;    bit = first_char ^ oc;
3099    if (ispowerof2(bit))    if (is_powerof2(bit))
3100      {      {
3101      OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit);      OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit);
3102      found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit);      found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit);
# Line 2795  else Line 3104  else
3104    else    else
3105      {      {
3106      OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char);      OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char);
3107      COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);      OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
3108      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);
3109      COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);      OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3110      found = JUMP(SLJIT_C_NOT_ZERO);      found = JUMP(SLJIT_C_NOT_ZERO);
3111      }      }
3112    }    }
# Line 2839  if (common->nltype == NLTYPE_FIXED && co Line 3148  if (common->nltype == NLTYPE_FIXED && co
3148    
3149    OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2));    OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2));
3150    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);
3151    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL);
3152  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3153    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT);
3154  #endif  #endif
3155    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
3156    
# Line 2882  if (common->nltype == NLTYPE_ANY || comm Line 3191  if (common->nltype == NLTYPE_ANY || comm
3191    notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);    notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3192    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
3193    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);
3194    COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
3195  #ifdef COMPILE_PCRE16  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3196    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
3197  #endif  #endif
3198    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3199    JUMPHERE(notfoundnl);    JUMPHERE(notfoundnl);
# Line 2897  if (firstline) Line 3206  if (firstline)
3206    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3207  }  }
3208    
3209  static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline)  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks);
3210    
3211    static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, pcre_uint8 *start_bits, BOOL firstline)
3212  {  {
3213  DEFINE_COMPILER;  DEFINE_COMPILER;
3214  struct sljit_label *start;  struct sljit_label *start;
3215  struct sljit_jump *quit;  struct sljit_jump *quit;
3216  struct sljit_jump *found;  struct sljit_jump *found = NULL;
3217    jump_list *matches = NULL;
3218  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3219  struct sljit_jump *jump;  struct sljit_jump *jump;
3220  #endif  #endif
# Line 2921  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_P Line 3233  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_P
3233  if (common->utf)  if (common->utf)
3234    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
3235  #endif  #endif
3236    
3237    if (!check_class_ranges(common, start_bits, (start_bits[31] & 0x80) != 0, TRUE, &matches))
3238      {
3239  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3240  jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);
3241  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255);    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255);
3242  JUMPHERE(jump);    JUMPHERE(jump);
3243  #endif  #endif
3244  OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
3245  OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
3246  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
3247  OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
3248  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);
3249  found = JUMP(SLJIT_C_NOT_ZERO);    found = JUMP(SLJIT_C_NOT_ZERO);
3250      }
3251    
3252  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3253  if (common->utf)  if (common->utf)
3254    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
3255  #endif  #endif
3256  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3257  #if defined SUPPORT_UTF && defined COMPILE_PCRE8  #ifdef SUPPORT_UTF
3258    #if defined COMPILE_PCRE8
3259  if (common->utf)  if (common->utf)
3260    {    {
3261    CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start);    CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start);
3262    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
3263    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3264    }    }
3265  #endif  #elif defined COMPILE_PCRE16
 #if defined SUPPORT_UTF && defined COMPILE_PCRE16  
3266  if (common->utf)  if (common->utf)
3267    {    {
3268    CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start);    CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start);
3269    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
3270    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);
3271    COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
3272    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
3273    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3274    }    }
3275  #endif  #endif /* COMPILE_PCRE[8|16] */
3276    #endif /* SUPPORT_UTF */
3277  JUMPTO(SLJIT_JUMP, start);  JUMPTO(SLJIT_JUMP, start);
3278  JUMPHERE(found);  if (found != NULL)
3279      JUMPHERE(found);
3280    if (matches != NULL)
3281      set_jumps(matches, LABEL());
3282  JUMPHERE(quit);  JUMPHERE(quit);
3283    
3284  if (firstline)  if (firstline)
# Line 2974  struct sljit_jump *alreadyfound; Line 3294  struct sljit_jump *alreadyfound;
3294  struct sljit_jump *found;  struct sljit_jump *found;
3295  struct sljit_jump *foundoc = NULL;  struct sljit_jump *foundoc = NULL;
3296  struct sljit_jump *notfound;  struct sljit_jump *notfound;
3297  pcre_uchar oc, bit;  pcre_uint32 oc, bit;
3298    
3299  SLJIT_ASSERT(common->req_char_ptr != 0);  SLJIT_ASSERT(common->req_char_ptr != 0);
3300  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr);
# Line 3005  if (req_char == oc) Line 3325  if (req_char == oc)
3325  else  else
3326    {    {
3327    bit = req_char ^ oc;    bit = req_char ^ oc;
3328    if (ispowerof2(bit))    if (is_powerof2(bit))
3329      {      {
3330      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit);      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit);
3331      found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit);      found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit);
# Line 3041  GET_LOCAL_BASE(TMP3, 0, 0); Line 3361  GET_LOCAL_BASE(TMP3, 0, 0);
3361  /* Drop frames until we reach STACK_TOP. */  /* Drop frames until we reach STACK_TOP. */
3362  mainloop = LABEL();  mainloop = LABEL();
3363  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0);
3364  jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end);  OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0);
3365    jump = JUMP(SLJIT_C_SIG_LESS_EQUAL);
3366    
3367  OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0);  OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0);
3368  OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_w));  OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw));
3369  OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_w), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_w));  OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw));
3370  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_w));  OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_sw));
3371  JUMPTO(SLJIT_JUMP, mainloop);  JUMPTO(SLJIT_JUMP, mainloop);
3372    
3373  JUMPHERE(jump);  JUMPHERE(jump);
3374  jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end);  jump = JUMP(SLJIT_C_SIG_LESS);
3375  /* End of dropping frames. */  /* End of dropping frames. */
3376  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
3377    
3378  JUMPHERE(jump);  JUMPHERE(jump);
3379  jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin);  OP1(SLJIT_NEG, TMP2, 0, TMP2, 0);
3380  /* Set string begin. */  OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0);
3381  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w));  OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw));
3382  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_sw));
 OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0);  
 JUMPTO(SLJIT_JUMP, mainloop);  
   
 JUMPHERE(jump);  
 if (common->mark_ptr != 0)  
   {  
   jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark);  
   OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w));  
   OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w));  
   OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0);  
   JUMPTO(SLJIT_JUMP, mainloop);  
   
   JUMPHERE(jump);  
   }  
   
 /* Unknown command. */  
 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w));  
3383  JUMPTO(SLJIT_JUMP, mainloop);  JUMPTO(SLJIT_JUMP, mainloop);
3384  }  }
3385    
# Line 3082  static void check_wordboundary(compiler_ Line 3387  static void check_wordboundary(compiler_
3387  {  {
3388  DEFINE_COMPILER;  DEFINE_COMPILER;
3389  struct sljit_jump *skipread;  struct sljit_jump *skipread;
3390    jump_list *skipread_list = NULL;
3391  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF
3392  struct sljit_jump *jump;  struct sljit_jump *jump;
3393  #endif  #endif
# Line 3107  if (common->use_ucp) Line 3413  if (common->use_ucp)
3413    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));
3414    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll);    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll);
3415    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);
3416    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
3417    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll);    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll);
3418    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd);    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd);
3419    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
3420    JUMPHERE(jump);    JUMPHERE(jump);
3421    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0);
3422    }    }
# Line 3139  else Line 3445  else
3445  JUMPHERE(skipread);  JUMPHERE(skipread);
3446    
3447  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
3448  skipread = check_str_end(common);  check_str_end(common, &skipread_list);
3449  peek_char(common);  peek_char(common);
3450    
3451  /* Testing char type. This is a code duplication. */  /* Testing char type. This is a code duplication. */
# Line 3151  if (common->use_ucp) Line 3457  if (common->use_ucp)
3457    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));
3458    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll);    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll);
3459    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);
3460    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
3461    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll);    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll);
3462    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd);    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd);
3463    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
3464    JUMPHERE(jump);    JUMPHERE(jump);
3465    }    }
3466  else  else
# Line 3180  else Line 3486  else
3486      JUMPHERE(jump);      JUMPHERE(jump);
3487  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
3488    }    }
3489  JUMPHERE(skipread);  set_jumps(skipread_list, LABEL());
3490    
3491  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);
3492  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);
# Line 3213  switch(ranges[0]) Line 3519  switch(ranges[0])
3519    case 2:    case 2:
3520    if (readch)    if (readch)
3521      read_char(common);      read_char(common);
3522    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);    if (ranges[2] + 1 != ranges[3])
3523    add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));      {
3524        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
3525        add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
3526        }
3527      else
3528        add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
3529      return TRUE;
3530    
3531      case 3:
3532      if (readch)
3533        read_char(common);
3534      if (ranges[1] != 0)
3535        {
3536        add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4]));
3537        if (ranges[2] + 1 != ranges[3])
3538          {
3539          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
3540          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
3541          }
3542        else
3543          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
3544        }
3545      else
3546        {
3547        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2]));
3548        if (ranges[3] + 1 != ranges[4])
3549          {
3550          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[3]);
3551          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[3]));
3552          }
3553        else
3554          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3]));
3555        }
3556    return TRUE;    return TRUE;
3557    
3558    case 4:    case 4:
# Line 3235  switch(ranges[0]) Line 3573  switch(ranges[0])
3573        }        }
3574      return TRUE;      return TRUE;
3575      }      }
3576    if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && ispowerof2(ranges[4] - ranges[2]))    if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2]))
3577      {      {
3578      if (readch)      if (readch)
3579        read_char(common);        read_char(common);
# Line 3281  if (bit != 0) Line 3619  if (bit != 0)
3619  ranges[0] = length;  ranges[0] = length;
3620  }  }
3621    
3622  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks)  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
3623  {  {
3624  int ranges[2 + MAX_RANGE_SIZE];  int ranges[2 + MAX_RANGE_SIZE];
3625  pcre_uint8 bit, cbit, all;  pcre_uint8 bit, cbit, all;
3626  int i, byte, length = 0;  int i, byte, length = 0;
3627    
3628  bit = bits[0] & 0x1;  bit = bits[0] & 0x1;
3629  ranges[1] = bit;  ranges[1] = !invert ? bit : (bit ^ 0x1);
3630  /* Can be 0 or 255. */  /* All bits will be zero or one (since bit is zero or one). */
3631  all = -bit;  all = -bit;
3632    
3633  for (i = 0; i < 256; )  for (i = 0; i < 256; )
# Line 3334  sljit_emit_fast_enter(compiler, RETURN_A Line 3672  sljit_emit_fast_enter(compiler, RETURN_A
3672    
3673  OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);  OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);
3674  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);
3675  COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL);  OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
3676  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);
3677  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3678  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3679  if (common->utf)  if (common->utf)
3680    {    {
3681  #endif  #endif
3682    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3683    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);
3684    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);
3685  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3686    }    }
3687  #endif  #endif
3688  #endif /* SUPPORT_UTF || COMPILE_PCRE16 */  #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */
3689  COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);  OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3690  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
3691  }  }
3692    
# Line 3360  DEFINE_COMPILER; Line 3698  DEFINE_COMPILER;
3698  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
3699    
3700  OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09);  OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09);
3701  COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);  OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
3702  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);
3703  COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);  OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3704  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);
3705  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3706  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3707  if (common->utf)  if (common->utf)
3708    {    {
3709  #endif  #endif
3710    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3711    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);
3712    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3713    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e);
3714    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3715    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000);    OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000);
3716    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000);    OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000);
3717    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
3718    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000);
3719    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3720    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);
3721    COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3722    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);
3723  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3724    }    }
3725  #endif  #endif
3726  #endif /* SUPPORT_UTF || COMPILE_PCRE16 */  #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */
3727  COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);  OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3728    
3729  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
3730  }  }
# Line 3400  sljit_emit_fast_enter(compiler, RETURN_A Line 3738  sljit_emit_fast_enter(compiler, RETURN_A
3738    
3739  OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);  OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);
3740  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);
3741  COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL);  OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
3742  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);
3743  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3744  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3745  if (common->utf)  if (common->utf)
3746    {    {
3747  #endif  #endif
3748    COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3749    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);
3750    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);
3751  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3752    }    }
3753  #endif  #endif
3754  #endif /* SUPPORT_UTF || COMPILE_PCRE16 */  #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */
3755  COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);  OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
3756    
3757  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
3758  }  }
# Line 3500  sljit_emit_fast_return(compiler, RETURN_ Line 3838  sljit_emit_fast_return(compiler, RETURN_
3838    
3839  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
3840    
3841  static const pcre_uchar *SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1)  static const pcre_uchar * SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1)
3842  {  {
3843  /* This function would be ineffective to do in JIT level. */  /* This function would be ineffective to do in JIT level. */
3844  int c1, c2;  pcre_uint32 c1, c2;
3845  const pcre_uchar *src2 = args->uchar_ptr;  const pcre_uchar *src2 = args->uchar_ptr;
3846  const pcre_uchar *end2 = args->end;  const pcre_uchar *end2 = args->end;
3847    const ucd_record *ur;
3848    const pcre_uint32 *pp;
3849    
3850  while (src1 < end1)  while (src1 < end1)
3851    {    {
# Line 3513  while (src1 < end1) Line 3853  while (src1 < end1)
3853      return (pcre_uchar*)1;      return (pcre_uchar*)1;
3854    GETCHARINC(c1, src1);    GETCHARINC(c1, src1);
3855    GETCHARINC(c2, src2);    GETCHARINC(c2, src2);
3856    if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL;    ur = GET_UCD(c2);
3857      if (c1 != c2 && c1 != c2 + ur->other_case)
3858        {
3859        pp = PRIV(ucd_caseless_sets) + ur->caseset;
3860        for (;;)
3861          {
3862          if (c1 < *pp) return NULL;
3863          if (c1 == *pp++) break;
3864          }
3865        }
3866    }    }
3867  return src2;  return src2;
3868  }  }
# Line 3535  if (caseless && char_has_othercase(commo Line 3884  if (caseless && char_has_othercase(commo
3884    othercasebit = char_get_othercase_bit(common, cc);    othercasebit = char_get_othercase_bit(common, cc);
3885    SLJIT_ASSERT(othercasebit);    SLJIT_ASSERT(othercasebit);
3886    /* Extracting bit difference info. */    /* Extracting bit difference info. */
3887  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
3888    othercasechar = cc + (othercasebit >> 8);    othercasechar = cc + (othercasebit >> 8);
3889    othercasebit &= 0xff;    othercasebit &= 0xff;
3890  #else  #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
3891  #ifdef COMPILE_PCRE16    /* Note that this code only handles characters in the BMP. If there
3892      ever are characters outside the BMP whose othercase differs in only one
3893      bit from itself (there currently are none), this code will need to be
3894      revised for COMPILE_PCRE32. */
3895    othercasechar = cc + (othercasebit >> 9);    othercasechar = cc + (othercasebit >> 9);
3896    if ((othercasebit & 0x100) != 0)    if ((othercasebit & 0x100) != 0)
3897      othercasebit = (othercasebit & 0xff) << 8;      othercasebit = (othercasebit & 0xff) << 8;
3898    else    else
3899      othercasebit &= 0xff;      othercasebit &= 0xff;
3900  #endif  #endif /* COMPILE_PCRE[8|16|32] */
 #endif  
3901    }    }
3902    
3903  if (context->sourcereg == -1)  if (context->sourcereg == -1)
3904    {    {
3905  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
3906  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED
3907    if (context->length >= 4)    if (context->length >= 4)
3908      OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);      OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);
# Line 3560  if (context->sourcereg == -1) Line 3911  if (context->sourcereg == -1)
3911    else    else
3912  #endif  #endif
3913      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);
3914  #else  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
3915  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED
3916    if (context->length >= 4)    if (context->length >= 4)
3917      OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);      OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);
3918    else    else
3919  #endif  #endif
3920      OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);
3921  #endif  #elif defined COMPILE_PCRE32
3922  #endif /* COMPILE_PCRE8 */    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length);
3923    #endif /* COMPILE_PCRE[8|16|32] */
3924    context->sourcereg = TMP2;    context->sourcereg = TMP2;
3925    }    }
3926    
# Line 3583  do Line 3934  do
3934  #endif  #endif
3935    
3936    context->length -= IN_UCHARS(1);    context->length -= IN_UCHARS(1);
3937  #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED  #if (defined SLJIT_UNALIGNED && SLJIT_UNALIGNED) && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16)
3938    
3939    /* Unaligned read is supported. */    /* Unaligned read is supported. */
3940    if (othercasebit != 0 && othercasechar == cc)    if (othercasebit != 0 && othercasechar == cc)
# Line 3598  do Line 3949  do
3949      }      }
3950    context->ucharptr++;    context->ucharptr++;
3951    
3952  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
3953    if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1))    if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1))
3954  #else  #else
3955    if (context->ucharptr >= 2 || context->length == 0)    if (context->ucharptr >= 2 || context->length == 0)
# Line 3606  do Line 3957  do
3957      {      {
3958      if (context->length >= 4)      if (context->length >= 4)
3959        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);
 #ifdef COMPILE_PCRE8  
3960      else if (context->length >= 2)      else if (context->length >= 2)
3961        OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);        OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);
3962    #if defined COMPILE_PCRE8
3963      else if (context->length >= 1)      else if (context->length >= 1)
3964        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);
3965  #else  #endif /* COMPILE_PCRE8 */
     else if (context->length >= 2)  
       OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);  
 #endif  
3966      context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1;      context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1;
3967    
3968      switch(context->ucharptr)      switch(context->ucharptr)
# Line 3648  do Line 3996  do
3996    
3997  #else  #else
3998    
3999    /* Unaligned read is unsupported. */    /* Unaligned read is unsupported or in 32 bit mode. */
4000  #ifdef COMPILE_PCRE8    if (context->length >= 1)
4001    if (context->length > 0)      OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);
4002      OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);  
 #else  
   if (context->length > 0)  
     OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length);  
 #endif  
4003    context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1;    context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1;
4004    
4005    if (othercasebit != 0 && othercasechar == cc)    if (othercasebit != 0 && othercasechar == cc)
# Line 3704  static void compile_xclass_matchingpath( Line 4048  static void compile_xclass_matchingpath(
4048  {  {
4049  DEFINE_COMPILER;  DEFINE_COMPILER;
4050  jump_list *found = NULL;  jump_list *found = NULL;
4051  jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks;  jump_list **list = (cc[0] & XCL_NOT) == 0 ? &found : backtracks;
4052  unsigned int c;  pcre_int32 c, charoffset;
 int compares;  
4053  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
4054  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
4055    int compares, invertcmp, numberofcmps;
4056    
4057  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4058  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4059  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4060  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4061  unsigned int typeoffset;  const pcre_uint32 *other_cases;
4062    pcre_int32 typeoffset;
4063  #endif  #endif
 int invertcmp, numberofcmps;  
 unsigned int charoffset;  
4064    
4065  /* Although SUPPORT_UTF must be defined, we are  /* Although SUPPORT_UTF must be defined, we are
4066     not necessary in utf mode even in 8 bit mode. */     not necessary in utf mode even in 8 bit mode. */
4067  detect_partial_match(common, backtracks);  detect_partial_match(common, backtracks);
4068  read_char(common);  read_char(common);
4069    
4070  if ((*cc++ & XCL_MAP) != 0)  cc++;
4071    if ((cc[-1] & XCL_HASPROP) == 0)
4072    {    {
4073    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    if ((cc[-1] & XCL_MAP) != 0)
4074  #ifndef COMPILE_PCRE8      {
4075    jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4076  #elif defined SUPPORT_UTF  #ifdef SUPPORT_UCP
4077    if (common->utf)      charsaved = TRUE;
     jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
4078  #endif  #endif
4079        if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, FALSE, backtracks))
4080          {
4081          jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4082    
4083    if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list))        OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4084          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4085          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4086          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4087          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4088          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4089          add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4090    
4091          JUMPHERE(jump);
4092          }
4093        else
4094          add_jump(compiler, &found, CMP(SLJIT_C_LESS_EQUAL, TMP3, 0, SLJIT_IMM, 0xff));
4095    
4096        OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4097        cc += 32 / sizeof(pcre_uchar);
4098        }
4099      else
4100        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0xff));
4101      }
4102    else if ((cc[-1] & XCL_MAP) != 0)
4103      {
4104      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4105    #ifdef SUPPORT_UCP
4106      charsaved = TRUE;
4107    #endif
4108      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4109      {      {
4110    #ifdef COMPILE_PCRE8
4111        SLJIT_ASSERT(common->utf);
4112    #endif
4113        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4114    
4115      OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);      OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4116      OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);      OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4117      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4118      OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);      OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4119      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);
4120      add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));      add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
     }  
4121    
 #ifndef COMPILE_PCRE8  
   JUMPHERE(jump);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
4122      JUMPHERE(jump);      JUMPHERE(jump);
4123  #endif      }
4124    
4125    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
 #ifdef SUPPORT_UCP  
   charsaved = TRUE;  
 #endif  
4126    cc += 32 / sizeof(pcre_uchar);    cc += 32 / sizeof(pcre_uchar);
4127    }    }
4128    
# Line 3810  while (*cc != XCL_END) Line 4180  while (*cc != XCL_END)
4180        case PT_SPACE:        case PT_SPACE:
4181        case PT_PXSPACE:        case PT_PXSPACE:
4182        case PT_WORD:        case PT_WORD:
4183          case PT_PXGRAPH:
4184          case PT_PXPRINT:
4185          case PT_PXPUNCT:
4186        needstype = TRUE;        needstype = TRUE;
4187        needschar = TRUE;        needschar = TRUE;
4188        break;        break;
4189    
4190          case PT_CLIST:
4191          case PT_UCNC:
4192          needschar = TRUE;
4193          break;
4194    
4195        default:        default:
4196        SLJIT_ASSERT_STOP();        SLJIT_ASSERT_STOP();
4197        break;        break;
# Line 3850  if (needstype || needsscript) Line 4228  if (needstype || needsscript)
4228      {      {
4229      if (scriptreg == TMP1)      if (scriptreg == TMP1)
4230        {        {
4231        OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script));        OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script));
4232        OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3);        OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3);
4233        }        }
4234      else      else
4235        {        {
4236        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3);        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3);
4237        OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script));        OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script));
4238        OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0);        OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0);
4239        }        }
4240      }      }
# Line 3892  while (*cc != XCL_END) Line 4270  while (*cc != XCL_END)
4270      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4271        {        {
4272        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);
4273        COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL);
4274        numberofcmps++;        numberofcmps++;
4275        }        }
4276      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4277        {        {
4278        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);
4279        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4280        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4281        numberofcmps = 0;        numberofcmps = 0;
4282        }        }
# Line 3931  while (*cc != XCL_END) Line 4309  while (*cc != XCL_END)
4309      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4310        {        {
4311        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);
4312        COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);
4313        numberofcmps++;        numberofcmps++;
4314        }        }
4315      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4316        {        {
4317        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);
4318        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4319        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4320        numberofcmps = 0;        numberofcmps = 0;
4321        }        }
# Line 3968  while (*cc != XCL_END) Line 4346  while (*cc != XCL_END)
4346    
4347        case PT_LAMP:        case PT_LAMP:
4348        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset);
4349        COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4350        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset);
4351        COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4352        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset);
4353        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4354        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4355        break;        break;
4356    
# Line 3992  while (*cc != XCL_END) Line 4370  while (*cc != XCL_END)
4370    
4371        case PT_SPACE:        case PT_SPACE:
4372        case PT_PXSPACE:        case PT_PXSPACE:
       if (*cc == PT_SPACE)  
         {  
         OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  
         jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 11 - charoffset);  
         }  
4373        SET_CHAR_OFFSET(9);        SET_CHAR_OFFSET(9);
4374        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd - 0x9);
4375        COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4376        if (*cc == PT_SPACE)  
4377          JUMPHERE(jump);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9);
4378          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4379    
4380          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9);
4381          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4382    
4383        SET_TYPE_OFFSET(ucp_Zl);        SET_TYPE_OFFSET(ucp_Zl);
4384        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);
4385        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4386        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4387        break;        break;
4388    
4389        case PT_WORD:        case PT_WORD:
4390        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset);
4391        COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4392        /* ... fall through */        /* Fall through. */
4393    
4394          case PT_ALNUM:
4395          SET_TYPE_OFFSET(ucp_Ll);
4396          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);
4397          OP_FLAGS((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, (*cc == PT_ALNUM) ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);
4398          SET_TYPE_OFFSET(ucp_Nd);
4399          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd);
4400          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4401          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4402          break;
4403    
4404          case PT_CLIST:
4405          other_cases = PRIV(ucd_caseless_sets) + cc[1];
4406    
4407          /* At least three characters are required.
4408             Otherwise this case would be handled by the normal code path. */
4409          SLJIT_ASSERT(other_cases[0] != NOTACHAR && other_cases[1] != NOTACHAR && other_cases[2] != NOTACHAR);
4410          SLJIT_ASSERT(other_cases[0] < other_cases[1] && other_cases[1] < other_cases[2]);
4411    
4412          /* Optimizing character pairs, if their difference is power of 2. */
4413          if (is_powerof2(other_cases[1] ^ other_cases[0]))
4414            {
4415            if (charoffset == 0)
4416              OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]);
4417            else
4418              {
4419              OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset);
4420              OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]);
4421              }
4422            OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]);
4423            OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4424            other_cases += 2;
4425            }
4426          else if (is_powerof2(other_cases[2] ^ other_cases[1]))
4427            {
4428            if (charoffset == 0)
4429              OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[2] ^ other_cases[1]);
4430            else
4431              {
4432              OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset);
4433              OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]);
4434              }
4435            OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]);
4436            OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4437    
4438            OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset);
4439            OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4440    
4441            other_cases += 3;
4442            }
4443          else
4444            {
4445            OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);
4446            OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4447            }
4448    
4449          while (*other_cases != NOTACHAR)
4450            {
4451            OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);
4452            OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4453            }
4454          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4455          break;
4456    
4457          case PT_UCNC:
4458          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_DOLLAR_SIGN - charoffset);
4459          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4460          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_COMMERCIAL_AT - charoffset);
4461          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4462          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_GRAVE_ACCENT - charoffset);
4463          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4464    
4465          SET_CHAR_OFFSET(0xa0);
4466          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd7ff - charoffset);
4467          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4468          SET_CHAR_OFFSET(0);
4469          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0);
4470          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL);
4471          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4472          break;
4473    
4474          case PT_PXGRAPH:
4475          /* C and Z groups are the farthest two groups. */
4476          SET_TYPE_OFFSET(ucp_Ll);
4477          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
4478          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
4479    
4480          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
4481    
4482          /* In case of ucp_Cf, we overwrite the result. */
4483          SET_CHAR_OFFSET(0x2066);
4484          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4485          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4486    
4487          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4488          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4489    
4490          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
4491          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4492    
4493          JUMPHERE(jump);
4494          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
4495          break;
4496    
4497          case PT_PXPRINT:
4498          /* C and Z groups are the farthest two groups. */
4499          SET_TYPE_OFFSET(ucp_Ll);
4500          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
4501          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
4502    
4503          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
4504          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
4505    
4506          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
4507    
4508          /* In case of ucp_Cf, we overwrite the result. */
4509          SET_CHAR_OFFSET(0x2066);
4510          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4511          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4512    
4513          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4514          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4515    
4516          JUMPHERE(jump);
4517          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
4518          break;
4519    
4520        case PT_ALNUM:        case PT_PXPUNCT:
4521        SET_TYPE_OFFSET(ucp_Ll);        SET_TYPE_OFFSET(ucp_Sc);
4522        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc);
4523        COND_VALUE((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4524        SET_TYPE_OFFSET(ucp_Nd);  
4525        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd);        SET_CHAR_OFFSET(0);
4526        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff);
4527          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4528    
4529          SET_TYPE_OFFSET(ucp_Pc);
4530          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc);
4531          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4532        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4533        break;        break;
4534        }        }
# Line 4048  int length; Line 4556  int length;
4556  unsigned int c, oc, bit;  unsigned int c, oc, bit;
4557  compare_context context;  compare_context context;
4558  struct sljit_jump *jump[4];  struct sljit_jump *jump[4];
4559    jump_list *end_list;
4560  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4561  struct sljit_label *label;  struct sljit_label *label;
4562  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4116  switch(type) Line 4625  switch(type)
4625    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
4626      {      {
4627      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);
4628        end_list = NULL;
4629      if (common->mode != JIT_PARTIAL_HARD_COMPILE)      if (common->mode != JIT_PARTIAL_HARD_COMPILE)
4630        jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);        add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
4631      else      else
4632        jump[1] = check_str_end(common);        check_str_end(common, &end_list);
4633    
4634      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
4635      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));
4636      if (jump[1] != NULL)      set_jumps(end_list, LABEL());
       JUMPHERE(jump[1]);  
4637      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
4638      }      }
4639    else    else
# Line 4138  switch(type) Line 4647  switch(type)
4647      {      {
4648      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
4649      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
4650  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
4651    #if defined COMPILE_PCRE8
4652      jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);      jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
4653      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
4654      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
4655  #else /* COMPILE_PCRE8 */  #elif defined COMPILE_PCRE16
 #ifdef COMPILE_PCRE16  
4656      jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);      jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
4657      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
4658      OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);      OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800);
4659      COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL);      OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4660      OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);      OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
4661      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
4662  #endif /* COMPILE_PCRE16 */  #endif
 #endif /* COMPILE_PCRE8 */  
4663      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
4664    #endif /* COMPILE_PCRE[8|16] */
4665      return cc;      return cc;
4666      }      }
4667  #endif  #endif
# Line 4168  switch(type) Line 4677  switch(type)
4677  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4678    case OP_NOTPROP:    case OP_NOTPROP:
4679    case OP_PROP:    case OP_PROP:
4680    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
4681    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
4682    propdata[2] = cc[0];    propdata[2] = cc[0];
4683    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4183  switch(type) Line 4692  switch(type)
4692    read_char(common);    read_char(common);
4693    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);
4694    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
4695      end_list = NULL;
4696    if (common->mode != JIT_PARTIAL_HARD_COMPILE)    if (common->mode != JIT_PARTIAL_HARD_COMPILE)
4697      jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);      add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
4698    else    else
4699      jump[1] = check_str_end(common);      check_str_end(common, &end_list);
4700    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
4701    jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);    jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);
4702    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
4703    jump[3] = JUMP(SLJIT_JUMP);    jump[2] = JUMP(SLJIT_JUMP);
4704    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
4705    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);
4706      set_jumps(end_list, LABEL());
4707    JUMPHERE(jump[1]);    JUMPHERE(jump[1]);
4708    JUMPHERE(jump[2]);    JUMPHERE(jump[2]);
   JUMPHERE(jump[3]);  
4709    return cc;    return cc;
4710    
4711    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
# Line 4219  switch(type) Line 4729  switch(type)
4729    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4730    read_char(common);    read_char(common);
4731    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));
4732    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop));    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop));
4733    /* Optimize register allocation: use a real register. */    /* Optimize register allocation: use a real register. */
4734    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0);
4735    OP1(SLJIT_MOV_UB, STACK_TOP, 0, SLJIT_MEM2(TMP1, TMP2), 3);    OP1(SLJIT_MOV_UB, STACK_TOP, 0, SLJIT_MEM2(TMP1, TMP2), 3);
# Line 4229  switch(type) Line 4739  switch(type)
4739    OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
4740    read_char(common);    read_char(common);
4741    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));
4742    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop));    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop));
4743    OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM2(TMP1, TMP2), 3);    OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM2(TMP1, TMP2), 3);
4744    
4745    OP2(SLJIT_SHL, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2);    OP2(SLJIT_SHL, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2);
4746    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), (sljit_w)PRIV(ucp_gbtable));    OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(STACK_TOP), (sljit_sw)PRIV(ucp_gbtable));
4747    OP1(SLJIT_MOV, STACK_TOP, 0, TMP2, 0);    OP1(SLJIT_MOV, STACK_TOP, 0, TMP2, 0);
4748    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4749    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);
# Line 4266  switch(type) Line 4776  switch(type)
4776        {        {
4777        jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0);        jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0);
4778        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);
4779        COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS);
4780        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);
4781        COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_NOT_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
4782        add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL));        add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL));
4783        check_partial(common, TRUE);        check_partial(common, TRUE);
4784        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
# Line 4448  switch(type) Line 4958  switch(type)
4958      }      }
4959    oc = char_othercase(common, c);    oc = char_othercase(common, c);
4960    bit = c ^ oc;    bit = c ^ oc;
4961    if (ispowerof2(bit))    if (is_powerof2(bit))
4962      {      {
4963      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit);      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit);
4964      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));
4965      return cc + length;      return cc + length;
4966      }      }
4967    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);
4968    COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4969    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);
4970    COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);    OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4971    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
4972    return cc + length;    return cc + length;
4973    
# Line 4484  switch(type) Line 4994  switch(type)
4994        /* Skip the variable-length character. */        /* Skip the variable-length character. */
4995        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
4996        jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);        jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
4997        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0);        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
4998        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
4999        JUMPHERE(jump[0]);        JUMPHERE(jump[0]);
5000        return cc + 1;        return cc + 1;
# Line 4509  switch(type) Line 5019  switch(type)
5019      {      {
5020      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5021      bit = c ^ oc;      bit = c ^ oc;
5022      if (ispowerof2(bit))      if (is_powerof2(bit))
5023        {        {
5024        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit);        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit);
5025        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));
# Line 4526  switch(type) Line 5036  switch(type)
5036    case OP_NCLASS:    case OP_NCLASS:
5037    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5038    read_char(common);    read_char(common);
5039    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5040      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5041    
5042  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
# Line 4547  switch(type) Line 5057  switch(type)
5057  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */
5058    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5059    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5060    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5061    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5062    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);
5063    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4557  switch(type) Line 5067  switch(type)
5067  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */
5068    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5069    
5070  #if defined SUPPORT_UTF || defined COMPILE_PCRE16  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
5071    case OP_XCLASS:    case OP_XCLASS:
5072    compile_xclass_matchingpath(common, cc + LINK_SIZE, backtracks);    compile_xclass_matchingpath(common, cc + LINK_SIZE, backtracks);
5073    return cc + GET(cc, 0) - 1;    return cc + GET(cc, 0) - 1;
# Line 4659  if (context.length > 0) Line 5169  if (context.length > 0)
5169  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);
5170  }  }
5171    
 static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)  
 {  
 DEFINE_COMPILER;  
 int offset = GET2(cc, 1) << 1;  
   
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  
 if (!common->jscript_compat)  
   {  
   if (backtracks == NULL)  
     {  
     /* OVECTOR(1) contains the "string begin - 1" constant. */  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));  
     COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL);  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
     COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL);  
     return JUMP(SLJIT_C_NOT_ZERO);  
     }  
   add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));  
   }  
 return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
 }  
   
5172  /* Forward definitions. */  /* Forward definitions. */
5173  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);
5174  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);
# Line 4713  static void compile_backtrackingpath(com Line 5201  static void compile_backtrackingpath(com
5201    
5202  #define BACKTRACK_AS(type) ((type *)backtrack)  #define BACKTRACK_AS(type) ((type *)backtrack)
5203    
5204  static pcre_uchar *compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)  static void compile_dnref_search(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)
5205    {
5206    /* The OVECTOR offset goes to TMP2. */
5207    DEFINE_COMPILER;
5208    int count = GET2(cc, 1 + IMM2_SIZE);
5209    pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
5210    unsigned int offset;
5211    jump_list *found = NULL;
5212    
5213    SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
5214    
5215    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
5216    
5217    count--;
5218    while (count-- > 0)
5219      {
5220      offset = GET2(slot, 0) << 1;
5221      GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5222      add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5223      slot += common->name_entry_size;
5224      }
5225    
5226    offset = GET2(slot, 0) << 1;
5227    GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5228    if (backtracks != NULL && !common->jscript_compat)
5229      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5230    
5231    set_jumps(found, LABEL());
5232    }
5233    
5234    static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
5235  {  {
5236  DEFINE_COMPILER;  DEFINE_COMPILER;
5237  int offset = GET2(cc, 1) << 1;  BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5238    int offset = 0;
5239  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
5240  struct sljit_jump *partial;  struct sljit_jump *partial;
5241  struct sljit_jump *nopartial;  struct sljit_jump *nopartial;
5242    
5243  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  if (ref)
5244  /* OVECTOR(1) contains the "string begin - 1" constant. */    {
5245  if (withchecks && !common->jscript_compat)    offset = GET2(cc, 1) << 1;
5246    add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5247      /* OVECTOR(1) contains the "string begin - 1" constant. */
5248      if (withchecks && !common->jscript_compat)
5249        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5250      }
5251    else
5252      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5253    
5254  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
5255  if (common->utf && *cc == OP_REFI)  if (common->utf && *cc == OP_REFI)
5256    {    {
5257    SLJIT_ASSERT(TMP1 == SLJIT_TEMPORARY_REG1 && STACK_TOP == SLJIT_TEMPORARY_REG2 && TMP2 == SLJIT_TEMPORARY_REG3);    SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3);
5258    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));    if (ref)
5259        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5260      else
5261        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5262    
5263    if (withchecks)    if (withchecks)
5264      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);
5265