Parent Directory
|
Revision Log
|
Patch
revision 915 by zherczeg, Tue Feb 14 13:05:39 2012 UTC | revision 1209 by zherczeg, Wed Nov 7 07:25:07 2012 UTC | |
---|---|---|
# | 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 | /* Allocate memory on the stack. Fast, but limited size. */ | /* Allocate memory for the regex stack on the real machine stack. |
69 | #define LOCAL_SPACE_SIZE 32768 | Fast, but limited size. */ |
70 | #define MACHINE_STACK_SIZE 32768 | |
71 | ||
72 | /* Growth rate for stack allocated by the OS. Should be the multiply | |
73 | of page size. */ | |
74 | #define STACK_GROWTH_RATE 8192 | #define STACK_GROWTH_RATE 8192 |
75 | ||
76 | /* Enable to check that the allocation could destroy temporaries. */ | /* Enable to check that the allocation could destroy temporaries. */ |
# | Line 82 The code generator follows the recursive | Line 85 The code generator follows the recursive |
85 | expressions. The basic blocks of regular expressions are condition checkers | expressions. The basic blocks of regular expressions are condition checkers |
86 | whose execute different commands depending on the result of the condition check. | whose execute different commands depending on the result of the condition check. |
87 | The relationship between the operators can be horizontal (concatenation) and | The relationship between the operators can be horizontal (concatenation) and |
88 | vertical (sub-expression) (See struct fallback_common for more details). | vertical (sub-expression) (See struct backtrack_common for more details). |
89 | ||
90 | 'ab' - 'a' and 'b' regexps are concatenated | 'ab' - 'a' and 'b' regexps are concatenated |
91 | 'a+' - 'a' is the sub-expression of the '+' operator | 'a+' - 'a' is the sub-expression of the '+' operator |
92 | ||
93 | The condition checkers are boolean (true/false) checkers. Machine code is generated | The condition checkers are boolean (true/false) checkers. Machine code is generated |
94 | for the checker itself and for the actions depending on the result of the checker. | for the checker itself and for the actions depending on the result of the checker. |
95 | The 'true' case is called as the hot path (expected path), and the other is called as | The 'true' case is called as the matching path (expected path), and the other is called as |
96 | the 'fallback' path. Branch instructions are expesive for all CPUs, so we avoid taken | the 'backtrack' path. Branch instructions are expesive for all CPUs, so we avoid taken |
97 | branches on the hot path. | branches on the matching path. |
98 | ||
99 | Greedy star operator (*) : | Greedy star operator (*) : |
100 | Hot path: match happens. | Matching path: match happens. |
101 | Fallback path: match failed. | Backtrack path: match failed. |
102 | Non-greedy star operator (*?) : | Non-greedy star operator (*?) : |
103 | Hot path: no need to perform a match. | Matching path: no need to perform a match. |
104 | Fallback path: match is required. | Backtrack path: match is required. |
105 | ||
106 | The following example shows how the code generated for a capturing bracket | The following example shows how the code generated for a capturing bracket |
107 | with two alternatives. Let A, B, C, D are arbirary regular expressions, and | with two alternatives. Let A, B, C, D are arbirary regular expressions, and |
# | Line 108 we have the following regular expression | Line 111 we have the following regular expression |
111 | ||
112 | The generated code will be the following: | The generated code will be the following: |
113 | ||
114 | A hot path | A matching path |
115 | '(' hot path (pushing arguments to the stack) | '(' matching path (pushing arguments to the stack) |
116 | B hot path | B matching path |
117 | ')' hot path (pushing arguments to the stack) | ')' matching path (pushing arguments to the stack) |
118 | D hot path | D matching path |
119 | return with successful match | return with successful match |
120 | ||
121 | D fallback path | D backtrack path |
122 | ')' fallback path (If we arrived from "C" jump to the fallback of "C") | ')' backtrack path (If we arrived from "C" jump to the backtrack of "C") |
123 | B fallback path | B backtrack path |
124 | C expected path | C expected path |
125 | jump to D hot path | jump to D matching path |
126 | C fallback path | C backtrack path |
127 | A fallback path | A backtrack path |
128 | ||
129 | Notice, that the order of fallback code paths are the opposite of the fast | Notice, that the order of backtrack code paths are the opposite of the fast |
130 | code paths. In this way the topmost value on the stack is always belong | code paths. In this way the topmost value on the stack is always belong |
131 | to the current fallback code path. The fallback code path must check | to the current backtrack code path. The backtrack path must check |
132 | whether there is a next alternative. If so, it needs to jump back to | whether there is a next alternative. If so, it needs to jump back to |
133 | the hot path eventually. Otherwise it needs to clear out its own stack | the matching path eventually. Otherwise it needs to clear out its own stack |
134 | frame and continue the execution on the fallback code paths. | frame and continue the execution on the backtrack code paths. |
135 | */ | */ |
136 | ||
137 | /* | /* |
138 | Saved stack frames: | Saved stack frames: |
139 | ||
140 | Atomic blocks and asserts require reloading the values of local variables | Atomic blocks and asserts require reloading the values of private data |
141 | when the fallback mechanism performed. Because of OP_RECURSE, the locals | when the backtrack mechanism performed. Because of OP_RECURSE, the data |
142 | are not necessarly known in compile time, thus we need a dynamic restore | are not necessarly known in compile time, thus we need a dynamic restore |
143 | mechanism. | mechanism. |
144 | ||
145 | The stack frames are stored in a chain list, and have the following format: | The stack frames are stored in a chain list, and have the following format: |
146 | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] |
147 | ||
148 | Thus we can restore the locals to a particular point in the stack. | Thus we can restore the private data to a particular point in the stack. |
149 | */ | */ |
150 | ||
151 | typedef struct jit_arguments { | typedef struct jit_arguments { |
# | Line 152 typedef struct jit_arguments { | Line 155 typedef struct jit_arguments { |
155 | const pcre_uchar *begin; | const pcre_uchar *begin; |
156 | const pcre_uchar *end; | const pcre_uchar *end; |
157 | int *offsets; | int *offsets; |
158 | pcre_uchar *ptr; | pcre_uchar *uchar_ptr; |
159 | pcre_uchar *mark_ptr; | |
160 | /* Everything else after. */ | /* Everything else after. */ |
161 | int offsetcount; | int offsetcount; |
162 | int calllimit; | int calllimit; |
# | Line 166 typedef struct executable_functions { | Line 170 typedef struct executable_functions { |
170 | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; |
171 | PUBL(jit_callback) callback; | PUBL(jit_callback) callback; |
172 | void *userdata; | void *userdata; |
173 | pcre_uint32 top_bracket; | |
174 | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; |
175 | } executable_functions; | } executable_functions; |
176 | ||
# | Line 180 typedef struct stub_list { | Line 185 typedef struct stub_list { |
185 | enum stub_types type; | enum stub_types type; |
186 | int data; | int data; |
187 | struct sljit_jump *start; | struct sljit_jump *start; |
188 | struct sljit_label *leave; | struct sljit_label *quit; |
189 | struct stub_list *next; | struct stub_list *next; |
190 | } stub_list; | } stub_list; |
191 | ||
192 | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); |
193 | ||
194 | /* The following structure is the key data type for the recursive | /* The following structure is the key data type for the recursive |
195 | code generator. It is allocated by compile_hotpath, and contains | code generator. It is allocated by compile_matchingpath, and contains |
196 | the aguments for compile_fallbackpath. Must be the first member | the aguments for compile_backtrackingpath. Must be the first member |
197 | of its descendants. */ | of its descendants. */ |
198 | typedef struct fallback_common { | typedef struct backtrack_common { |
199 | /* Concatenation stack. */ | /* Concatenation stack. */ |
200 | struct fallback_common *prev; | struct backtrack_common *prev; |
201 | jump_list *nextfallbacks; | jump_list *nextbacktracks; |
202 | /* Internal stack (for component operators). */ | /* Internal stack (for component operators). */ |
203 | struct fallback_common *top; | struct backtrack_common *top; |
204 | jump_list *topfallbacks; | jump_list *topbacktracks; |
205 | /* Opcode pointer. */ | /* Opcode pointer. */ |
206 | pcre_uchar *cc; | pcre_uchar *cc; |
207 | } fallback_common; | } backtrack_common; |
208 | ||
209 | typedef struct assert_fallback { | typedef struct assert_backtrack { |
210 | fallback_common common; | backtrack_common common; |
211 | jump_list *condfailed; | jump_list *condfailed; |
212 | /* Less than 0 (-1) if a frame is not needed. */ | /* Less than 0 (-1) if a frame is not needed. */ |
213 | int framesize; | int framesize; |
214 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
215 | int localptr; | int private_data_ptr; |
216 | /* For iterators. */ | /* For iterators. */ |
217 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
218 | } assert_fallback; | } assert_backtrack; |
219 | ||
220 | typedef struct bracket_fallback { | typedef struct bracket_backtrack { |
221 | fallback_common common; | backtrack_common common; |
222 | /* Where to coninue if an alternative is successfully matched. */ | /* Where to coninue if an alternative is successfully matched. */ |
223 | struct sljit_label *althotpath; | struct sljit_label *alternative_matchingpath; |
224 | /* For rmin and rmax iterators. */ | /* For rmin and rmax iterators. */ |
225 | struct sljit_label *recursivehotpath; | struct sljit_label *recursive_matchingpath; |
226 | /* For greedy ? operator. */ | /* For greedy ? operator. */ |
227 | struct sljit_label *zerohotpath; | struct sljit_label *zero_matchingpath; |
228 | /* Contains the branches of a failed condition. */ | /* Contains the branches of a failed condition. */ |
229 | union { | union { |
230 | /* Both for OP_COND, OP_SCOND. */ | /* Both for OP_COND, OP_SCOND. */ |
231 | jump_list *condfailed; | jump_list *condfailed; |
232 | assert_fallback *assert; | assert_backtrack *assert; |
233 | /* For OP_ONCE. -1 if not needed. */ | /* For OP_ONCE. -1 if not needed. */ |
234 | int framesize; | int framesize; |
235 | } u; | } u; |
236 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
237 | int localptr; | int private_data_ptr; |
238 | } bracket_fallback; | } bracket_backtrack; |
239 | ||
240 | typedef struct bracketpos_fallback { | typedef struct bracketpos_backtrack { |
241 | fallback_common common; | backtrack_common common; |
242 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
243 | int localptr; | int private_data_ptr; |
244 | /* Reverting stack is needed. */ | /* Reverting stack is needed. */ |
245 | int framesize; | int framesize; |
246 | /* Allocated stack size. */ | /* Allocated stack size. */ |
247 | int stacksize; | int stacksize; |
248 | } bracketpos_fallback; | } bracketpos_backtrack; |
249 | ||
250 | typedef struct braminzero_fallback { | typedef struct braminzero_backtrack { |
251 | fallback_common common; | backtrack_common common; |
252 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
253 | } braminzero_fallback; | } braminzero_backtrack; |
254 | ||
255 | typedef struct iterator_fallback { | typedef struct iterator_backtrack { |
256 | fallback_common common; | backtrack_common common; |
257 | /* Next iteration. */ | /* Next iteration. */ |
258 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
259 | } iterator_fallback; | } iterator_backtrack; |
260 | ||
261 | typedef struct recurse_entry { | typedef struct recurse_entry { |
262 | struct recurse_entry *next; | struct recurse_entry *next; |
# | Line 263 typedef struct recurse_entry { | Line 268 typedef struct recurse_entry { |
268 | int start; | int start; |
269 | } recurse_entry; | } recurse_entry; |
270 | ||
271 | typedef struct recurse_fallback { | typedef struct recurse_backtrack { |
272 | fallback_common common; | backtrack_common common; |
273 | } recurse_fallback; | } recurse_backtrack; |
274 | ||
275 | #define MAX_RANGE_SIZE 6 | |
276 | ||
277 | typedef struct compiler_common { | typedef struct compiler_common { |
278 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
279 | pcre_uchar *start; | pcre_uchar *start; |
280 | int localsize; | |
281 | int *localptrs; | /* Maps private data offset to each opcode. */ |
282 | const pcre_uint8 *fcc; | int *private_data_ptrs; |
283 | sljit_w lcc; | /* Tells whether the capturing bracket is optimized. */ |
284 | pcre_uint8 *optimized_cbracket; | |
285 | /* Starting offset of private data for capturing brackets. */ | |
286 | int cbraptr; | int cbraptr; |
287 | /* OVector starting point. Must be divisible by 2. */ | |
288 | int ovector_start; | |
289 | /* Last known position of the requested byte. */ | |
290 | int req_char_ptr; | |
291 | /* Head of the last recursion. */ | |
292 | int recursive_head; | |
293 | /* First inspected character for partial matching. */ | |
294 | int start_used_ptr; | |
295 | /* Starting pointer for partial soft matches. */ | |
296 | int hit_start; | |
297 | /* End pointer of the first line. */ | |
298 | int first_line_end; | |
299 | /* Points to the marked string. */ | |
300 | int mark_ptr; | |
301 | ||
302 | /* Flipped and lower case tables. */ | |
303 | const pcre_uint8 *fcc; | |
304 | sljit_sw lcc; | |
305 | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ | |
306 | int mode; | int mode; |
307 | /* Newline control. */ | |
308 | int nltype; | int nltype; |
309 | int newline; | int newline; |
310 | int bsr_nltype; | int bsr_nltype; |
311 | /* Dollar endonly. */ | |
312 | int endonly; | int endonly; |
313 | sljit_w ctypes; | BOOL has_set_som; |
314 | /* Tables. */ | |
315 | sljit_sw ctypes; | |
316 | int digits[2 + MAX_RANGE_SIZE]; | |
317 | /* Named capturing brackets. */ | |
318 | sljit_uw name_table; | sljit_uw name_table; |
319 | sljit_w name_count; | sljit_sw name_count; |
320 | sljit_w name_entry_size; | sljit_sw name_entry_size; |
321 | ||
322 | /* Labels and jump lists. */ | |
323 | struct sljit_label *partialmatchlabel; | struct sljit_label *partialmatchlabel; |
324 | struct sljit_label *quitlabel; | |
325 | struct sljit_label *acceptlabel; | struct sljit_label *acceptlabel; |
326 | stub_list *stubs; | stub_list *stubs; |
327 | recurse_entry *entries; | recurse_entry *entries; |
328 | recurse_entry *currententry; | recurse_entry *currententry; |
329 | jump_list *partialmatch; | jump_list *partialmatch; |
330 | jump_list *quit; | |
331 | jump_list *accept; | jump_list *accept; |
332 | jump_list *calllimit; | jump_list *calllimit; |
333 | jump_list *stackalloc; | jump_list *stackalloc; |
# | Line 306 typedef struct compiler_common { | Line 344 typedef struct compiler_common { |
344 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
345 | BOOL use_ucp; | BOOL use_ucp; |
346 | #endif | #endif |
347 | #ifndef COMPILE_PCRE32 | |
348 | jump_list *utfreadchar; | jump_list *utfreadchar; |
349 | #endif | |
350 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
351 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
352 | #endif | #endif |
# | Line 324 typedef struct compare_context { | Line 364 typedef struct compare_context { |
364 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
365 | int ucharptr; | int ucharptr; |
366 | union { | union { |
367 | sljit_i asint; | sljit_si asint; |
368 | sljit_uh asushort; | sljit_uh asushort; |
369 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
370 | sljit_ub asbyte; | sljit_ub asbyte; |
371 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
372 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
373 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
374 | #endif | #elif defined COMPILE_PCRE32 |
375 | sljit_ui asuchars[1]; | |
376 | #endif | #endif |
377 | } c; | } c; |
378 | union { | union { |
379 | sljit_i asint; | sljit_si asint; |
380 | sljit_uh asushort; | sljit_uh asushort; |
381 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
382 | sljit_ub asbyte; | sljit_ub asbyte; |
383 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
384 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
385 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
386 | #endif | #elif defined COMPILE_PCRE32 |
387 | sljit_ui asuchars[1]; | |
388 | #endif | #endif |
389 | } oc; | } oc; |
390 | #endif | #endif |
# | Line 352 typedef struct compare_context { | Line 392 typedef struct compare_context { |
392 | ||
393 | enum { | enum { |
394 | frame_end = 0, | frame_end = 0, |
395 | frame_setstrbegin = -1 | frame_setstrbegin = -1, |
396 | frame_setmark = -2 | |
397 | }; | }; |
398 | ||
399 | /* Undefine sljit macros. */ | /* Undefine sljit macros. */ |
400 | #undef CMP | #undef CMP |
401 | ||
402 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
403 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) |
404 | ||
405 | #define TMP1 SLJIT_TEMPORARY_REG1 | #define TMP1 SLJIT_TEMPORARY_REG1 |
406 | #define TMP2 SLJIT_TEMPORARY_REG3 | #define TMP2 SLJIT_TEMPORARY_REG3 |
# | Line 372 enum { | Line 413 enum { |
413 | #define CALL_COUNT SLJIT_SAVED_EREG2 | #define CALL_COUNT SLJIT_SAVED_EREG2 |
414 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 |
415 | ||
416 | /* Locals layout. */ | /* Local space layout. */ |
417 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
418 | #define LOCALS0 (0 * sizeof(sljit_w)) | #define LOCALS0 (0 * sizeof(sljit_sw)) |
419 | #define LOCALS1 (1 * sizeof(sljit_w)) | #define LOCALS1 (1 * sizeof(sljit_sw)) |
420 | /* Two local variables for possessive quantifiers (char1 cannot use them). */ | /* Two local variables for possessive quantifiers (char1 cannot use them). */ |
421 | #define POSSESSIVE0 (2 * sizeof(sljit_w)) | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) |
422 | #define POSSESSIVE1 (3 * sizeof(sljit_w)) | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) |
/* Head of the last recursion. */ | ||
#define RECURSIVE_HEAD (4 * sizeof(sljit_w)) | ||
423 | /* Max limit of recursions. */ | /* Max limit of recursions. */ |
424 | #define CALL_LIMIT (5 * sizeof(sljit_w)) | #define CALL_LIMIT (4 * sizeof(sljit_sw)) |
/* Last known position of the requested byte. | ||
Same as START_USED_PTR. (Partial matching and req_char are exclusive) */ | ||
#define REQ_CHAR_PTR (6 * sizeof(sljit_w)) | ||
/* First inspected character for partial matching. | ||
Same as REQ_CHAR_PTR. (Partial matching and req_char are exclusive) */ | ||
#define START_USED_PTR (6 * sizeof(sljit_w)) | ||
/* Starting pointer for partial soft matches. */ | ||
#define HIT_START (8 * sizeof(sljit_w)) | ||
/* End pointer of the first line. */ | ||
#define FIRSTLINE_END (9 * sizeof(sljit_w)) | ||
425 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
426 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
427 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
428 | 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. */ |
429 | #define OVECTOR_START (10 * sizeof(sljit_w)) | #define OVECTOR_START (common->ovector_start) |
430 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_sw)) |
431 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_sw)) |
432 | #define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
433 | ||
434 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
435 | #define MOV_UCHAR SLJIT_MOV_UB | #define MOV_UCHAR SLJIT_MOV_UB |
436 | #define MOVU_UCHAR SLJIT_MOVU_UB | #define MOVU_UCHAR SLJIT_MOVU_UB |
437 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
438 | #define MOV_UCHAR SLJIT_MOV_UH | #define MOV_UCHAR SLJIT_MOV_UH |
439 | #define MOVU_UCHAR SLJIT_MOVU_UH | #define MOVU_UCHAR SLJIT_MOVU_UH |
440 | #elif defined COMPILE_PCRE32 | |
441 | #define MOV_UCHAR SLJIT_MOV_UI | |
442 | #define MOVU_UCHAR SLJIT_MOVU_UI | |
443 | #else | #else |
444 | #error Unsupported compiling mode | #error Unsupported compiling mode |
445 | #endif | #endif |
#endif | ||
446 | ||
447 | /* Shortcuts. */ | /* Shortcuts. */ |
448 | #define DEFINE_COMPILER \ | #define DEFINE_COMPILER \ |
# | Line 433 the start pointers when the end of the c | Line 463 the start pointers when the end of the c |
463 | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) |
464 | #define CMPTO(type, src1, src1w, src2, src2w, label) \ | #define CMPTO(type, src1, src1w, src2, src2w, label) \ |
465 | 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)) |
466 | #define COND_VALUE(op, dst, dstw, type) \ | #define OP_FLAGS(op, dst, dstw, src, srcw, type) \ |
467 | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) | sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type)) |
468 | #define GET_LOCAL_BASE(dst, dstw, offset) \ | |
469 | sljit_get_local_base(compiler, (dst), (dstw), (offset)) | |
470 | ||
471 | static pcre_uchar* bracketend(pcre_uchar* cc) | static pcre_uchar* bracketend(pcre_uchar* cc) |
472 | { | { |
# | Line 447 return cc; | Line 479 return cc; |
479 | ||
480 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
481 | next_opcode | next_opcode |
482 | get_localspace | get_private_data_length |
483 | set_localptrs | set_private_data_ptrs |
484 | get_framesize | get_framesize |
485 | init_frame | init_frame |
486 | get_localsize | get_private_data_length_for_copy |
487 | copy_locals | copy_private_data |
488 | compile_hotpath | compile_matchingpath |
489 | compile_fallbackpath | compile_backtrackingpath |
490 | */ | */ |
491 | ||
492 | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) |
# | Line 506 switch(*cc) | Line 538 switch(*cc) |
538 | case OP_BRAZERO: | case OP_BRAZERO: |
539 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
540 | case OP_BRAPOSZERO: | case OP_BRAPOSZERO: |
541 | case OP_COMMIT: | |
542 | case OP_FAIL: | case OP_FAIL: |
543 | case OP_ACCEPT: | case OP_ACCEPT: |
544 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
# | Line 644 switch(*cc) | Line 677 switch(*cc) |
677 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
678 | return cc + 1 + LINK_SIZE + IMM2_SIZE; | return cc + 1 + LINK_SIZE + IMM2_SIZE; |
679 | ||
680 | case OP_MARK: | |
681 | return cc + 1 + 2 + cc[1]; | |
682 | ||
683 | default: | default: |
684 | return NULL; | return NULL; |
685 | } | } |
686 | } | } |
687 | ||
688 | static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | #define CASE_ITERATOR_PRIVATE_DATA_1 \ |
689 | case OP_MINSTAR: \ | |
690 | case OP_MINPLUS: \ | |
691 | case OP_QUERY: \ | |
692 | case OP_MINQUERY: \ | |
693 | case OP_MINSTARI: \ | |
694 | case OP_MINPLUSI: \ | |
695 | case OP_QUERYI: \ | |
696 | case OP_MINQUERYI: \ | |
697 | case OP_NOTMINSTAR: \ | |
698 | case OP_NOTMINPLUS: \ | |
699 | case OP_NOTQUERY: \ | |
700 | case OP_NOTMINQUERY: \ | |
701 | case OP_NOTMINSTARI: \ | |
702 | case OP_NOTMINPLUSI: \ | |
703 | case OP_NOTQUERYI: \ | |
704 | case OP_NOTMINQUERYI: | |
705 | ||
706 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
707 | case OP_STAR: \ | |
708 | case OP_PLUS: \ | |
709 | case OP_STARI: \ | |
710 | case OP_PLUSI: \ | |
711 | case OP_NOTSTAR: \ | |
712 | case OP_NOTPLUS: \ | |
713 | case OP_NOTSTARI: \ | |
714 | case OP_NOTPLUSI: | |
715 | ||
716 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
717 | case OP_UPTO: \ | |
718 | case OP_MINUPTO: \ | |
719 | case OP_UPTOI: \ | |
720 | case OP_MINUPTOI: \ | |
721 | case OP_NOTUPTO: \ | |
722 | case OP_NOTMINUPTO: \ | |
723 | case OP_NOTUPTOI: \ | |
724 | case OP_NOTMINUPTOI: | |
725 | ||
726 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
727 | case OP_TYPEMINSTAR: \ | |
728 | case OP_TYPEMINPLUS: \ | |
729 | case OP_TYPEQUERY: \ | |
730 | case OP_TYPEMINQUERY: | |
731 | ||
732 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
733 | case OP_TYPESTAR: \ | |
734 | case OP_TYPEPLUS: | |
735 | ||
736 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
737 | case OP_TYPEUPTO: \ | |
738 | case OP_TYPEMINUPTO: | |
739 | ||
740 | static int get_class_iterator_size(pcre_uchar *cc) | |
741 | { | |
742 | switch(*cc) | |
743 | { | |
744 | case OP_CRSTAR: | |
745 | case OP_CRPLUS: | |
746 | return 2; | |
747 | ||
748 | case OP_CRMINSTAR: | |
749 | case OP_CRMINPLUS: | |
750 | case OP_CRQUERY: | |
751 | case OP_CRMINQUERY: | |
752 | return 1; | |
753 | ||
754 | case OP_CRRANGE: | |
755 | case OP_CRMINRANGE: | |
756 | if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | |
757 | return 0; | |
758 | return 2; | |
759 | ||
760 | default: | |
761 | return 0; | |
762 | } | |
763 | } | |
764 | ||
765 | static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | |
766 | { | { |
767 | int localspace = 0; | int private_data_length = 0; |
768 | pcre_uchar *alternative; | pcre_uchar *alternative; |
769 | pcre_uchar *name; | |
770 | pcre_uchar *end = NULL; | |
771 | int space, size, i; | |
772 | pcre_uint32 bracketlen; | |
773 | ||
774 | /* 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. */ |
775 | while (cc < ccend) | while (cc < ccend) |
776 | { | { |
777 | space = 0; | |
778 | size = 0; | |
779 | bracketlen = 0; | |
780 | switch(*cc) | switch(*cc) |
781 | { | { |
782 | case OP_SET_SOM: | |
783 | common->has_set_som = TRUE; | |
784 | cc += 1; | |
785 | break; | |
786 | ||
787 | case OP_REF: | |
788 | case OP_REFI: | |
789 | common->optimized_cbracket[GET2(cc, 1)] = 0; | |
790 | cc += 1 + IMM2_SIZE; | |
791 | break; | |
792 | ||
793 | case OP_ASSERT: | case OP_ASSERT: |
794 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
795 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 667 while (cc < ccend) | Line 799 while (cc < ccend) |
799 | case OP_BRAPOS: | case OP_BRAPOS: |
800 | case OP_SBRA: | case OP_SBRA: |
801 | case OP_SBRAPOS: | case OP_SBRAPOS: |
802 | case OP_SCOND: | private_data_length += sizeof(sljit_sw); |
803 | localspace += sizeof(sljit_w); | bracketlen = 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; | ||
804 | break; | break; |
805 | ||
806 | case OP_CBRAPOS: | case OP_CBRAPOS: |
807 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
808 | localspace += sizeof(sljit_w); | private_data_length += sizeof(sljit_sw); |
809 | cc += 1 + LINK_SIZE + IMM2_SIZE; | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
810 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
811 | break; | break; |
812 | ||
813 | case OP_COND: | case OP_COND: |
814 | /* Might be a hidden SCOND. */ | case OP_SCOND: |
815 | alternative = cc + GET(cc, 1); | bracketlen = cc[1 + LINK_SIZE]; |
816 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (bracketlen == OP_CREF) |
817 | localspace += sizeof(sljit_w); | { |
818 | bracketlen = GET2(cc, 1 + LINK_SIZE + 1); | |
819 | common->optimized_cbracket[bracketlen] = 0; | |
820 | } | |
821 | else if (bracketlen == OP_NCREF) | |
822 | { | |
823 | bracketlen = GET2(cc, 1 + LINK_SIZE + 1); | |
824 | name = (pcre_uchar *)common->name_table; | |
825 | alternative = name; | |
826 | for (i = 0; i < common->name_count; i++) | |
827 | { | |
828 | if (GET2(name, 0) == bracketlen) break; | |
829 | name += common->name_entry_size; | |
830 | } | |
831 | SLJIT_ASSERT(i != common->name_count); | |
832 | ||
833 | for (i = 0; i < common->name_count; i++) | |
834 | { | |
835 | if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | |
836 | common->optimized_cbracket[GET2(alternative, 0)] = 0; | |
837 | alternative += common->name_entry_size; | |
838 | } | |
839 | } | |
840 | ||
841 | if (*cc == OP_COND) | |
842 | { | |
843 | /* Might be a hidden SCOND. */ | |
844 | alternative = cc + GET(cc, 1); | |
845 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | |
846 | private_data_length += sizeof(sljit_sw); | |
847 | } | |
848 | else | |
849 | private_data_length += sizeof(sljit_sw); | |
850 | bracketlen = 1 + LINK_SIZE; | |
851 | break; | |
852 | ||
853 | case OP_BRA: | |
854 | bracketlen = 1 + LINK_SIZE; | |
855 | break; | |
856 | ||
857 | case OP_CBRA: | |
858 | case OP_SCBRA: | |
859 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
860 | break; | |
861 | ||
862 | CASE_ITERATOR_PRIVATE_DATA_1 | |
863 | space = 1; | |
864 | size = -2; | |
865 | break; | |
866 | ||
867 | CASE_ITERATOR_PRIVATE_DATA_2A | |
868 | space = 2; | |
869 | size = -2; | |
870 | break; | |
871 | ||
872 | CASE_ITERATOR_PRIVATE_DATA_2B | |
873 | space = 2; | |
874 | size = -(2 + IMM2_SIZE); | |
875 | break; | |
876 | ||
877 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
878 | space = 1; | |
879 | size = 1; | |
880 | break; | |
881 | ||
882 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
883 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
884 | space = 2; | |
885 | size = 1; | |
886 | break; | |
887 | ||
888 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
889 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
890 | space = 2; | |
891 | size = 1 + IMM2_SIZE; | |
892 | break; | |
893 | ||
894 | case OP_CLASS: | |
895 | case OP_NCLASS: | |
896 | size += 1 + 32 / sizeof(pcre_uchar); | |
897 | space = get_class_iterator_size(cc + size); | |
898 | break; | |
899 | ||
900 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
901 | case OP_XCLASS: | |
902 | size = GET(cc, 1); | |
903 | space = get_class_iterator_size(cc + size); | |
904 | break; | |
905 | #endif | |
906 | ||
907 | case OP_RECURSE: | |
908 | /* Set its value only once. */ | |
909 | if (common->recursive_head == 0) | |
910 | { | |
911 | common->recursive_head = common->ovector_start; | |
912 | common->ovector_start += sizeof(sljit_sw); | |
913 | } | |
914 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
915 | break; | break; |
916 | ||
917 | case OP_MARK: | |
918 | if (common->mark_ptr == 0) | |
919 | { | |
920 | common->mark_ptr = common->ovector_start; | |
921 | common->ovector_start += sizeof(sljit_sw); | |
922 | } | |
923 | cc += 1 + 2 + cc[1]; | |
924 | break; | |
925 | ||
926 | default: | default: |
927 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
928 | if (cc == NULL) | if (cc == NULL) |
929 | return -1; | return -1; |
930 | break; | break; |
931 | } | } |
932 | ||
933 | if (space > 0 && cc >= end) | |
934 | private_data_length += sizeof(sljit_sw) * space; | |
935 | ||
936 | if (size != 0) | |
937 | { | |
938 | if (size < 0) | |
939 | { | |
940 | cc += -size; | |
941 | #ifdef SUPPORT_UTF | |
942 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
943 | #endif | |
944 | } | |
945 | else | |
946 | cc += size; | |
947 | } | |
948 | ||
949 | if (bracketlen != 0) | |
950 | { | |
951 | if (cc >= end) | |
952 | { | |
953 | end = bracketend(cc); | |
954 | if (end[-1 - LINK_SIZE] == OP_KET) | |
955 | end = NULL; | |
956 | } | |
957 | cc += bracketlen; | |
958 | } | |
959 | } | } |
960 | return localspace; | return private_data_length; |
961 | } | } |
962 | ||
963 | static void set_localptrs(compiler_common *common, int localptr, pcre_uchar *ccend) | static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend) |
964 | { | { |
965 | pcre_uchar *cc = common->start; | pcre_uchar *cc = common->start; |
966 | pcre_uchar *alternative; | pcre_uchar *alternative; |
967 | pcre_uchar *end = NULL; | |
968 | int space, size, bracketlen; | |
969 | ||
970 | while (cc < ccend) | while (cc < ccend) |
971 | { | { |
972 | space = 0; | |
973 | size = 0; | |
974 | bracketlen = 0; | |
975 | switch(*cc) | switch(*cc) |
976 | { | { |
977 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 714 while (cc < ccend) | Line 984 while (cc < ccend) |
984 | case OP_SBRA: | case OP_SBRA: |
985 | case OP_SBRAPOS: | case OP_SBRAPOS: |
986 | case OP_SCOND: | case OP_SCOND: |
987 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
988 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
989 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
990 | break; | break; |
991 | ||
992 | case OP_CBRAPOS: | case OP_CBRAPOS: |
993 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
994 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
995 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
996 | cc += 1 + LINK_SIZE + IMM2_SIZE; | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; |
997 | break; | break; |
998 | ||
999 | case OP_COND: | case OP_COND: |
# | Line 731 while (cc < ccend) | Line 1001 while (cc < ccend) |
1001 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1002 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1003 | { | { |
1004 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1005 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
1006 | } | } |
1007 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1008 | break; | |
1009 | ||
1010 | case OP_BRA: | |
1011 | bracketlen = 1 + LINK_SIZE; | |
1012 | break; | |
1013 | ||
1014 | case OP_CBRA: | |
1015 | case OP_SCBRA: | |
1016 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
1017 | break; | |
1018 | ||
1019 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1020 | space = 1; | |
1021 | size = -2; | |
1022 | break; | |
1023 | ||
1024 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1025 | space = 2; | |
1026 | size = -2; | |
1027 | break; | |
1028 | ||
1029 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1030 | space = 2; | |
1031 | size = -(2 + IMM2_SIZE); | |
1032 | break; | |
1033 | ||
1034 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1035 | space = 1; | |
1036 | size = 1; | |
1037 | break; | |
1038 | ||
1039 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1040 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
1041 | space = 2; | |
1042 | size = 1; | |
1043 | break; | |
1044 | ||
1045 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1046 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
1047 | space = 2; | |
1048 | size = 1 + IMM2_SIZE; | |
1049 | break; | |
1050 | ||
1051 | case OP_CLASS: | |
1052 | case OP_NCLASS: | |
1053 | size += 1 + 32 / sizeof(pcre_uchar); | |
1054 | space = get_class_iterator_size(cc + size); | |
1055 | break; | |
1056 | ||
1057 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1058 | case OP_XCLASS: | |
1059 | size = GET(cc, 1); | |
1060 | space = get_class_iterator_size(cc + size); | |
1061 | break; | break; |
1062 | #endif | |
1063 | ||
1064 | default: | default: |
1065 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1066 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1067 | break; | break; |
1068 | } | } |
1069 | ||
1070 | if (space > 0 && cc >= end) | |
1071 | { | |
1072 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1073 | private_data_ptr += sizeof(sljit_sw) * space; | |
1074 | } | |
1075 | ||
1076 | if (size != 0) | |
1077 | { | |
1078 | if (size < 0) | |
1079 | { | |
1080 | cc += -size; | |
1081 | #ifdef SUPPORT_UTF | |
1082 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1083 | #endif | |
1084 | } | |
1085 | else | |
1086 | cc += size; | |
1087 | } | |
1088 | ||
1089 | if (bracketlen > 0) | |
1090 | { | |
1091 | if (cc >= end) | |
1092 | { | |
1093 | end = bracketend(cc); | |
1094 | if (end[-1 - LINK_SIZE] == OP_KET) | |
1095 | end = NULL; | |
1096 | } | |
1097 | cc += bracketlen; | |
1098 | } | |
1099 | } | } |
1100 | } | } |
1101 | ||
# | Line 751 static int get_framesize(compiler_common | Line 1105 static int get_framesize(compiler_common |
1105 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc); |
1106 | int length = 0; | int length = 0; |
1107 | BOOL possessive = FALSE; | BOOL possessive = FALSE; |
1108 | BOOL setsom_found = FALSE; | BOOL setsom_found = recursive; |
1109 | BOOL setmark_found = recursive; | |
1110 | ||
1111 | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
1112 | { | { |
# | Line 765 while (cc < ccend) | Line 1120 while (cc < ccend) |
1120 | switch(*cc) | switch(*cc) |
1121 | { | { |
1122 | case OP_SET_SOM: | case OP_SET_SOM: |
1123 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
1124 | if (!setsom_found) | if (!setsom_found) |
1125 | { | { |
1126 | length += 2; | length += 2; |
1127 | setsom_found = TRUE; | setsom_found = TRUE; |
1128 | } | } |
1129 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | cc += 1; |
1130 | break; | |
1131 | ||
1132 | case OP_MARK: | |
1133 | SLJIT_ASSERT(common->mark_ptr != 0); | |
1134 | if (!setmark_found) | |
1135 | { | |
1136 | length += 2; | |
1137 | setmark_found = TRUE; | |
1138 | } | |
1139 | cc += 1 + 2 + cc[1]; | |
1140 | break; | |
1141 | ||
1142 | case OP_RECURSE: | |
1143 | if (common->has_set_som && !setsom_found) | |
1144 | { | |
1145 | length += 2; | |
1146 | setsom_found = TRUE; | |
1147 | } | |
1148 | if (common->mark_ptr != 0 && !setmark_found) | |
1149 | { | |
1150 | length += 2; | |
1151 | setmark_found = TRUE; | |
1152 | } | |
1153 | cc += 1 + LINK_SIZE; | |
1154 | break; | break; |
1155 | ||
1156 | case OP_CBRA: | case OP_CBRA: |
# | Line 801 static void init_frame(compiler_common * | Line 1180 static void init_frame(compiler_common * |
1180 | { | { |
1181 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1182 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc); |
1183 | BOOL setsom_found = FALSE; | BOOL setsom_found = recursive; |
1184 | BOOL setmark_found = recursive; | |
1185 | int offset; | int offset; |
1186 | ||
1187 | /* >= 1 + shortest item size (2) */ | /* >= 1 + shortest item size (2) */ |
# | Line 816 while (cc < ccend) | Line 1196 while (cc < ccend) |
1196 | switch(*cc) | switch(*cc) |
1197 | { | { |
1198 | case OP_SET_SOM: | case OP_SET_SOM: |
1199 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
1200 | if (!setsom_found) | if (!setsom_found) |
1201 | { | { |
1202 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
1203 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); |
1204 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1205 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1206 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1207 | setsom_found = TRUE; | setsom_found = TRUE; |
1208 | } | } |
1209 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | cc += 1; |
1210 | break; | |
1211 | ||
1212 | case OP_MARK: | |
1213 | SLJIT_ASSERT(common->mark_ptr != 0); | |
1214 | if (!setmark_found) | |
1215 | { | |
1216 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
1217 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | |
1218 | stackpos += (int)sizeof(sljit_sw); | |
1219 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1220 | stackpos += (int)sizeof(sljit_sw); | |
1221 | setmark_found = TRUE; | |
1222 | } | |
1223 | cc += 1 + 2 + cc[1]; | |
1224 | break; | |
1225 | ||
1226 | case OP_RECURSE: | |
1227 | if (common->has_set_som && !setsom_found) | |
1228 | { | |
1229 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | |
1230 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | |
1231 | stackpos += (int)sizeof(sljit_sw); | |
1232 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1233 | stackpos += (int)sizeof(sljit_sw); | |
1234 | setsom_found = TRUE; | |
1235 | } | |
1236 | if (common->mark_ptr != 0 && !setmark_found) | |
1237 | { | |
1238 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
1239 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | |
1240 | stackpos += (int)sizeof(sljit_sw); | |
1241 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1242 | stackpos += (int)sizeof(sljit_sw); | |
1243 | setmark_found = TRUE; | |
1244 | } | |
1245 | cc += 1 + LINK_SIZE; | |
1246 | break; | break; |
1247 | ||
1248 | case OP_CBRA: | case OP_CBRA: |
# | Line 835 while (cc < ccend) | Line 1251 while (cc < ccend) |
1251 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1252 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
1253 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
1254 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1255 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
1256 | 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)); |
1257 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1258 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1259 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
1260 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1261 | ||
1262 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1263 | break; | break; |
# | Line 856 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st | Line 1272 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st |
1272 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1273 | } | } |
1274 | ||
1275 | static SLJIT_INLINE int get_localsize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
1276 | { | { |
1277 | int localsize = 2; | int private_data_length = 2; |
1278 | int size; | |
1279 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1280 | /* Calculate the sum of the local variables. */ | /* Calculate the sum of the private machine words. */ |
1281 | while (cc < ccend) | while (cc < ccend) |
1282 | { | { |
1283 | size = 0; | |
1284 | switch(*cc) | switch(*cc) |
1285 | { | { |
1286 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 875 while (cc < ccend) | Line 1293 while (cc < ccend) |
1293 | case OP_SBRA: | case OP_SBRA: |
1294 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1295 | case OP_SCOND: | case OP_SCOND: |
1296 | localsize++; | private_data_length++; |
1297 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1298 | break; | break; |
1299 | ||
1300 | case OP_CBRA: | case OP_CBRA: |
1301 | case OP_SCBRA: | case OP_SCBRA: |
1302 | localsize++; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1303 | private_data_length++; | |
1304 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1305 | break; | break; |
1306 | ||
1307 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1308 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1309 | localsize += 2; | private_data_length += 2; |
1310 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1311 | break; | break; |
1312 | ||
# | Line 895 while (cc < ccend) | Line 1314 while (cc < ccend) |
1314 | /* Might be a hidden SCOND. */ | /* Might be a hidden SCOND. */ |
1315 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1316 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1317 | localsize++; | private_data_length++; |
1318 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1319 | break; | break; |
1320 | ||
1321 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1322 | if (PRIVATE_DATA(cc)) | |
1323 | private_data_length++; | |
1324 | cc += 2; | |
1325 | #ifdef SUPPORT_UTF | |
1326 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1327 | #endif | |
1328 | break; | |
1329 | ||
1330 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1331 | if (PRIVATE_DATA(cc)) | |
1332 | private_data_length += 2; | |
1333 | cc += 2; | |
1334 | #ifdef SUPPORT_UTF | |
1335 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1336 | #endif | |
1337 | break; | |
1338 | ||
1339 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1340 | if (PRIVATE_DATA(cc)) | |
1341 | private_data_length += 2; | |
1342 | cc += 2 + IMM2_SIZE; | |
1343 | #ifdef SUPPORT_UTF | |
1344 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1345 | #endif | |
1346 | break; | |
1347 | ||
1348 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1349 | if (PRIVATE_DATA(cc)) | |
1350 | private_data_length++; | |
1351 | cc += 1; | |
1352 | break; | |
1353 | ||
1354 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1355 | if (PRIVATE_DATA(cc)) | |
1356 | private_data_length += 2; | |
1357 | cc += 1; | |
1358 | break; | |
1359 | ||
1360 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1361 | if (PRIVATE_DATA(cc)) | |
1362 | private_data_length += 2; | |
1363 | cc += 1 + IMM2_SIZE; | |
1364 | break; | |
1365 | ||
1366 | case OP_CLASS: | |
1367 | case OP_NCLASS: | |
1368 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1369 | case OP_XCLASS: | |
1370 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1371 | #else | |
1372 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1373 | #endif | |
1374 | if (PRIVATE_DATA(cc)) | |
1375 | private_data_length += get_class_iterator_size(cc + size); | |
1376 | cc += size; | |
1377 | break; | |
1378 | ||
1379 | default: | default: |
1380 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1381 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 906 while (cc < ccend) | Line 1383 while (cc < ccend) |
1383 | } | } |
1384 | } | } |
1385 | SLJIT_ASSERT(cc == ccend); | SLJIT_ASSERT(cc == ccend); |
1386 | return localsize; | return private_data_length; |
1387 | } | } |
1388 | ||
1389 | static void copy_locals(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, | static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, |
1390 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop) |
1391 | { | { |
1392 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1393 | int srcw[2]; | int srcw[2]; |
1394 | int count; | int count, size; |
1395 | BOOL tmp1next = TRUE; | BOOL tmp1next = TRUE; |
1396 | BOOL tmp1empty = TRUE; | BOOL tmp1empty = TRUE; |
1397 | BOOL tmp2empty = TRUE; | BOOL tmp2empty = TRUE; |
# | Line 931 stacktop = STACK(stacktop - 1); | Line 1408 stacktop = STACK(stacktop - 1); |
1408 | ||
1409 | if (!save) | if (!save) |
1410 | { | { |
1411 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1412 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1413 | { | { |
1414 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1415 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1416 | tmp1empty = FALSE; | tmp1empty = FALSE; |
1417 | } | } |
1418 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1419 | { | { |
1420 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1421 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1422 | tmp2empty = FALSE; | tmp2empty = FALSE; |
1423 | } | } |
1424 | /* The tmp1next must be TRUE in either way. */ | /* The tmp1next must be TRUE in either way. */ |
# | Line 953 while (status != end) | Line 1430 while (status != end) |
1430 | switch(status) | switch(status) |
1431 | { | { |
1432 | case start: | case start: |
1433 | SLJIT_ASSERT(save); | SLJIT_ASSERT(save && common->recursive_head != 0); |
1434 | count = 1; | count = 1; |
1435 | srcw[0] = RECURSIVE_HEAD; | srcw[0] = common->recursive_head; |
1436 | status = loop; | status = loop; |
1437 | break; | break; |
1438 | ||
# | Line 979 while (status != end) | Line 1456 while (status != end) |
1456 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1457 | case OP_SCOND: | case OP_SCOND: |
1458 | count = 1; | count = 1; |
1459 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1460 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1461 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1462 | break; | break; |
1463 | ||
1464 | case OP_CBRA: | case OP_CBRA: |
1465 | case OP_SCBRA: | case OP_SCBRA: |
1466 | count = 1; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1467 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | { |
1468 | count = 1; | |
1469 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | |
1470 | } | |
1471 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1472 | break; | break; |
1473 | ||
1474 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1475 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1476 | count = 2; | count = 2; |
1477 | srcw[0] = PRIVATE_DATA(cc); | |
1478 | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
1479 | srcw[0] = PRIV_DATA(cc); | SLJIT_ASSERT(srcw[0] != 0 && srcw[1] != 0); |
SLJIT_ASSERT(srcw[0] != 0); | ||
1480 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1481 | break; | break; |
1482 | ||
# | Line 1006 while (status != end) | Line 1486 while (status != end) |
1486 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1487 | { | { |
1488 | count = 1; | count = 1; |
1489 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1490 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1491 | } | } |
1492 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1493 | break; | break; |
1494 | ||
1495 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1496 | if (PRIVATE_DATA(cc)) | |
1497 | { | |
1498 | count = 1; | |
1499 | srcw[0] = PRIVATE_DATA(cc); | |
1500 | } | |
1501 | cc += 2; | |
1502 | #ifdef SUPPORT_UTF | |
1503 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1504 | #endif | |
1505 | break; | |
1506 | ||
1507 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1508 | if (PRIVATE_DATA(cc)) | |
1509 | { | |
1510 | count = 2; | |
1511 | srcw[0] = PRIVATE_DATA(cc); | |
1512 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1513 | } | |
1514 | cc += 2; | |
1515 | #ifdef SUPPORT_UTF | |
1516 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1517 | #endif | |
1518 | break; | |
1519 | ||
1520 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1521 | if (PRIVATE_DATA(cc)) | |
1522 | { | |
1523 | count = 2; | |
1524 | srcw[0] = PRIVATE_DATA(cc); | |
1525 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1526 | } | |
1527 | cc += 2 + IMM2_SIZE; | |
1528 | #ifdef SUPPORT_UTF | |
1529 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1530 | #endif | |
1531 | break; | |
1532 | ||
1533 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1534 | if (PRIVATE_DATA(cc)) | |
1535 | { | |
1536 | count = 1; | |
1537 | srcw[0] = PRIVATE_DATA(cc); | |
1538 | } | |
1539 | cc += 1; | |
1540 | break; | |
1541 | ||
1542 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1543 | if (PRIVATE_DATA(cc)) | |
1544 | { | |
1545 | count = 2; | |
1546 | srcw[0] = PRIVATE_DATA(cc); | |
1547 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1548 | } | |
1549 | cc += 1; | |
1550 | break; | |
1551 | ||
1552 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1553 | if (PRIVATE_DATA(cc)) | |
1554 | { | |
1555 | count = 2; | |
1556 | srcw[0] = PRIVATE_DATA(cc); | |
1557 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1558 | } | |
1559 | cc += 1 + IMM2_SIZE; | |
1560 | break; | |
1561 | ||
1562 | case OP_CLASS: | |
1563 | case OP_NCLASS: | |
1564 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1565 | case OP_XCLASS: | |
1566 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1567 | #else | |
1568 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1569 | #endif | |
1570 | if (PRIVATE_DATA(cc)) | |
1571 | switch(get_class_iterator_size(cc + size)) | |
1572 | { | |
1573 | case 1: | |
1574 | count = 1; | |
1575 | srcw[0] = PRIVATE_DATA(cc); | |
1576 | break; | |
1577 | ||
1578 | case 2: | |
1579 | count = 2; | |
1580 | srcw[0] = PRIVATE_DATA(cc); | |
1581 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1582 | break; | |
1583 | ||
1584 | default: | |
1585 | SLJIT_ASSERT_STOP(); | |
1586 | break; | |
1587 | } | |
1588 | cc += size; | |
1589 | break; | |
1590 | ||
1591 | default: | default: |
1592 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1593 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1034 while (status != end) | Line 1610 while (status != end) |
1610 | if (!tmp1empty) | if (!tmp1empty) |
1611 | { | { |
1612 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1613 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1614 | } | } |
1615 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1616 | tmp1empty = FALSE; | tmp1empty = FALSE; |
# | Line 1045 while (status != end) | Line 1621 while (status != end) |
1621 | if (!tmp2empty) | if (!tmp2empty) |
1622 | { | { |
1623 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1624 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1625 | } | } |
1626 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1627 | tmp2empty = FALSE; | tmp2empty = FALSE; |
# | Line 1062 while (status != end) | Line 1638 while (status != end) |
1638 | if (!tmp1empty) | if (!tmp1empty) |
1639 | { | { |
1640 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1641 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1642 | } | } |
1643 | tmp1next = FALSE; | tmp1next = FALSE; |
1644 | } | } |
# | Line 1074 while (status != end) | Line 1650 while (status != end) |
1650 | if (!tmp2empty) | if (!tmp2empty) |
1651 | { | { |
1652 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1653 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1654 | } | } |
1655 | tmp1next = TRUE; | tmp1next = TRUE; |
1656 | } | } |
# | Line 1089 if (save) | Line 1665 if (save) |
1665 | if (!tmp1empty) | if (!tmp1empty) |
1666 | { | { |
1667 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1668 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1669 | } | } |
1670 | if (!tmp2empty) | if (!tmp2empty) |
1671 | { | { |
1672 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1673 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1674 | } | } |
1675 | } | } |
1676 | else | else |
# | Line 1102 if (save) | Line 1678 if (save) |
1678 | if (!tmp2empty) | if (!tmp2empty) |
1679 | { | { |
1680 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1681 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1682 | } | } |
1683 | if (!tmp1empty) | if (!tmp1empty) |
1684 | { | { |
1685 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1686 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1687 | } | } |
1688 | } | } |
1689 | } | } |
1690 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1691 | } | } |
1692 | ||
1693 | static SLJIT_INLINE BOOL ispowerof2(unsigned int value) | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
1694 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | |
1695 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | |
1696 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1697 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1698 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1699 | ||
1700 | static SLJIT_INLINE BOOL is_powerof2(unsigned int value) | |
1701 | { | { |
1702 | return (value & (value - 1)) == 0; | return (value & (value - 1)) == 0; |
1703 | } | } |
# | Line 1124 static SLJIT_INLINE void set_jumps(jump_ | Line 1707 static SLJIT_INLINE void set_jumps(jump_ |
1707 | while (list) | while (list) |
1708 | { | { |
1709 | /* sljit_set_label is clever enough to do nothing | /* sljit_set_label is clever enough to do nothing |
1710 | if either the jump or the label is NULL */ | if either the jump or the label is NULL. */ |
1711 | sljit_set_label(list->jump, label); | sljit_set_label(list->jump, label); |
1712 | list = list->next; | list = list->next; |
1713 | } | } |
# | Line 1151 if (list_item) | Line 1734 if (list_item) |
1734 | list_item->type = type; | list_item->type = type; |
1735 | list_item->data = data; | list_item->data = data; |
1736 | list_item->start = start; | list_item->start = start; |
1737 | list_item->leave = LABEL(); | list_item->quit = LABEL(); |
1738 | list_item->next = common->stubs; | list_item->next = common->stubs; |
1739 | common->stubs = list_item; | common->stubs = list_item; |
1740 | } | } |
# | Line 1171 while (list_item) | Line 1754 while (list_item) |
1754 | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
1755 | break; | break; |
1756 | } | } |
1757 | JUMPTO(SLJIT_JUMP, list_item->leave); | JUMPTO(SLJIT_JUMP, list_item->quit); |
1758 | list_item = list_item->next; | list_item = list_item->next; |
1759 | } | } |
1760 | common->stubs = NULL; | common->stubs = NULL; |
# | Line 1190 static SLJIT_INLINE void allocate_stack( | Line 1773 static SLJIT_INLINE void allocate_stack( |
1773 | /* May destroy all locals and registers except TMP2. */ | /* May destroy all locals and registers except TMP2. */ |
1774 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1775 | ||
1776 | 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)); |
1777 | #ifdef DESTROY_REGISTERS | #ifdef DESTROY_REGISTERS |
1778 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); |
1779 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
# | Line 1204 add_stub(common, stack_alloc, 0, CMP(SLJ | Line 1787 add_stub(common, stack_alloc, 0, CMP(SLJ |
1787 | static SLJIT_INLINE void free_stack(compiler_common *common, int size) | static SLJIT_INLINE void free_stack(compiler_common *common, int size) |
1788 | { | { |
1789 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1790 | 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)); |
1791 | } | } |
1792 | ||
1793 | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) |
# | Line 1222 if (length < 8) | Line 1805 if (length < 8) |
1805 | } | } |
1806 | else | else |
1807 | { | { |
1808 | OP2(SLJIT_ADD, SLJIT_TEMPORARY_REG2, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START - sizeof(sljit_w)); | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, OVECTOR_START - sizeof(sljit_sw)); |
1809 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); |
1810 | loop = LABEL(); | loop = LABEL(); |
1811 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_w), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_sw), SLJIT_TEMPORARY_REG1, 0); |
1812 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); |
1813 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
1814 | } | } |
# | Line 1242 OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJI | Line 1825 OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJI |
1825 | 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); |
1826 | ||
1827 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); |
1828 | if (common->mark_ptr != 0) | |
1829 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
1830 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); |
1831 | if (common->mark_ptr != 0) | |
1832 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0); | |
1833 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); |
1834 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); |
1835 | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START); | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); |
1836 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
1837 | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); |
1838 | loop = LABEL(); | loop = LABEL(); |
1839 | 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_TEMPORARY_REG1, 0); |
1840 | 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)); |
1841 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
1842 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1843 | 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); |
1844 | #endif | #endif |
1845 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); |
1846 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); |
# | Line 1263 JUMPHERE(earlyexit); | Line 1850 JUMPHERE(earlyexit); |
1850 | /* Calculate the return value, which is the maximum ovector value. */ | /* Calculate the return value, which is the maximum ovector value. */ |
1851 | if (topbracket > 1) | if (topbracket > 1) |
1852 | { | { |
1853 | OP2(SLJIT_ADD, SLJIT_TEMPORARY_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START + topbracket * 2 * sizeof(sljit_w)); | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); |
1854 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); |
1855 | ||
1856 | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ |
1857 | loop = LABEL(); | loop = LABEL(); |
1858 | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w))); | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); |
1859 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); |
1860 | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); |
1861 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); |
# | Line 1277 else | Line 1864 else |
1864 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
1865 | } | } |
1866 | ||
1867 | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *leave) | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) |
1868 | { | { |
1869 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1870 | ||
1871 | 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); |
1872 | SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | |
1873 | ||
1874 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); |
1875 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); |
1876 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); |
1877 | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, leave); | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, quit); |
1878 | ||
1879 | /* Store match begin and end. */ | /* Store match begin and end. */ |
1880 | 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_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); |
1881 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); |
1882 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? START_USED_PTR : HIT_START); | 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); |
1883 | 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); |
1884 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1885 | 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); |
1886 | #endif | #endif |
1887 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); |
1888 | ||
1889 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); |
1890 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1891 | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); |
1892 | #endif | #endif |
1893 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); |
1894 | ||
1895 | JUMPTO(SLJIT_JUMP, leave); | JUMPTO(SLJIT_JUMP, quit); |
1896 | } | } |
1897 | ||
1898 | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) |
# | Line 1315 struct sljit_jump *jump; | Line 1903 struct sljit_jump *jump; |
1903 | ||
1904 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
1905 | { | { |
1906 | /* The value of -1 must be kept for START_USED_PTR! */ | /* The value of -1 must be kept for start_used_ptr! */ |
1907 | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); |
1908 | /* Jumps if START_USED_PTR < STR_PTR, or START_USED_PTR == -1. Although overwriting | /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting |
1909 | is not necessary if START_USED_PTR == STR_PTR, it does not hurt as well. */ | is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ |
1910 | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); |
1911 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); |
1912 | JUMPHERE(jump); | JUMPHERE(jump); |
1913 | } | } |
1914 | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) |
1915 | { | { |
1916 | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); |
1917 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); |
1918 | JUMPHERE(jump); | JUMPHERE(jump); |
1919 | } | } |
1920 | } | } |
# | Line 1415 if (c <= 127 && bit == 0x20) | Line 2003 if (c <= 127 && bit == 0x20) |
2003 | return (0 << 8) | 0x20; | return (0 << 8) | 0x20; |
2004 | ||
2005 | /* Since c != oc, they must have at least 1 bit difference. */ | /* Since c != oc, they must have at least 1 bit difference. */ |
2006 | if (!ispowerof2(bit)) | if (!is_powerof2(bit)) |
2007 | return 0; | return 0; |
2008 | ||
2009 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2010 | ||
2011 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2012 | if (common->utf && c > 127) | if (common->utf && c > 127) |
# | Line 1434 if (common->utf && c > 127) | Line 2022 if (common->utf && c > 127) |
2022 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2023 | return (0 << 8) | bit; | return (0 << 8) | bit; |
2024 | ||
2025 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2026 | ||
#ifdef COMPILE_PCRE16 | ||
2027 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2028 | if (common->utf && c > 65535) | if (common->utf && c > 65535) |
2029 | { | { |
# | Line 1447 if (common->utf && c > 65535) | Line 2034 if (common->utf && c > 65535) |
2034 | } | } |
2035 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2036 | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); |
#endif /* COMPILE_PCRE16 */ | ||
2037 | ||
2038 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16|32] */ |
2039 | } | } |
2040 | ||
2041 | static void check_partial(compiler_common *common) | static void check_partial(compiler_common *common, BOOL force) |
2042 | { | { |
2043 | /* Checks whether a partial matching is occured. Does not modify registers. */ | |
2044 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2045 | struct sljit_jump *jump; | struct sljit_jump *jump = NULL; |
2046 | ||
2047 | SLJIT_ASSERT(!force || common->mode != JIT_COMPILE); | |
2048 | ||
2049 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2050 | return; | return; |
2051 | ||
2052 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | if (!force) |
2053 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
2054 | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2055 | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); | |
2056 | ||
2057 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2058 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), HIT_START, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); |
2059 | else | else |
2060 | { | { |
2061 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
# | Line 1470 else | Line 2063 else |
2063 | else | else |
2064 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); |
2065 | } | } |
2066 | JUMPHERE(jump); | |
2067 | if (jump != NULL) | |
2068 | JUMPHERE(jump); | |
2069 | } | } |
2070 | ||
2071 | static struct sljit_jump *check_str_end(compiler_common *common) | static struct sljit_jump *check_str_end(compiler_common *common) |
# | Line 1487 if (common->mode == JIT_COMPILE) | Line 2082 if (common->mode == JIT_COMPILE) |
2082 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2083 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2084 | { | { |
2085 | nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); |
2086 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), HIT_START, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); |
2087 | JUMPHERE(nohit); | JUMPHERE(nohit); |
2088 | return_value = JUMP(SLJIT_JUMP); | return_value = JUMP(SLJIT_JUMP); |
2089 | } | } |
2090 | else | else |
2091 | { | { |
2092 | return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), START_USED_PTR, STR_PTR, 0); | return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); |
2093 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
2094 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); |
2095 | else | else |
# | Line 1504 JUMPHERE(jump); | Line 2099 JUMPHERE(jump); |
2099 | return return_value; | return return_value; |
2100 | } | } |
2101 | ||
2102 | static void fallback_at_str_end(compiler_common *common, jump_list **fallbacks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
2103 | { | { |
2104 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2105 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2106 | ||
2107 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2108 | { | { |
2109 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
2110 | return; | return; |
2111 | } | } |
2112 | ||
2113 | /* Partial matching mode. */ | /* Partial matching mode. */ |
2114 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2115 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), 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)); |
2116 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2117 | { | { |
2118 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), HIT_START, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); |
2119 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
2120 | } | } |
2121 | else | else |
2122 | { | { |
# | Line 1538 static void read_char(compiler_common *c | Line 2133 static void read_char(compiler_common *c |
2133 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, updates STR_PTR. |
2134 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2135 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2136 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2137 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2138 | #endif | #endif |
2139 | ||
2140 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2141 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2142 | if (common->utf) | if (common->utf) |
2143 | { | { |
2144 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2145 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2146 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2147 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2148 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2149 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2150 | JUMPHERE(jump); | JUMPHERE(jump); |
2151 | } | } |
2152 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2153 | 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)); |
2154 | } | } |
2155 | ||
# | Line 1565 static void peek_char(compiler_common *c | Line 2158 static void peek_char(compiler_common *c |
2158 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2159 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2160 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2161 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2162 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2163 | #endif | #endif |
2164 | ||
2165 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2166 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2167 | if (common->utf) | if (common->utf) |
2168 | { | { |
2169 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2170 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2171 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2172 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2173 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2174 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2175 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2176 | JUMPHERE(jump); | JUMPHERE(jump); |
2177 | } | } |
2178 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2179 | } | } |
2180 | ||
2181 | static void read_char8_type(compiler_common *common) | static void read_char8_type(compiler_common *common) |
2182 | { | { |
2183 | /* 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. */ |
2184 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2185 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2186 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2187 | #endif | #endif |
2188 | ||
# | Line 1600 if (common->utf) | Line 2191 if (common->utf) |
2191 | { | { |
2192 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2193 | 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)); |
2194 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2195 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
2196 | it is needed in most cases. */ | it is needed in most cases. */ |
2197 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2198 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2199 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
2200 | JUMPHERE(jump); | JUMPHERE(jump); |
2201 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2202 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2203 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2204 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
# | Line 1616 if (common->utf) | Line 2206 if (common->utf) |
2206 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2207 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); |
2208 | 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); |
2209 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2210 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
2211 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2212 | #endif | #elif defined COMPILE_PCRE32 |
2213 | #endif /* COMPILE_PCRE8 */ | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2214 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
2215 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2216 | JUMPHERE(jump); | |
2217 | #endif /* COMPILE_PCRE[8|16|32] */ | |
2218 | return; | return; |
2219 | } | } |
2220 | #endif | #endif /* SUPPORT_UTF */ |
2221 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2222 | 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)); |
2223 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2224 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2225 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2226 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2227 | #endif | #endif |
2228 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2229 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2230 | JUMPHERE(jump); | JUMPHERE(jump); |
2231 | #endif | #endif |
2232 | } | } |
# | Line 1641 static void skip_char_back(compiler_comm | Line 2235 static void skip_char_back(compiler_comm |
2235 | { | { |
2236 | /* 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. */ |
2237 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2238 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2239 | #if defined COMPILE_PCRE8 | |
2240 | struct sljit_label *label; | struct sljit_label *label; |
2241 | ||
2242 | if (common->utf) | if (common->utf) |
# | Line 1653 if (common->utf) | Line 2248 if (common->utf) |
2248 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
2249 | return; | return; |
2250 | } | } |
2251 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2252 | if (common->utf) | if (common->utf) |
2253 | { | { |
2254 | 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 1662 if (common->utf) | Line 2256 if (common->utf) |
2256 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2257 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2258 | 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); |
2259 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2260 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2261 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2262 | return; | return; |
2263 | } | } |
2264 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2265 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2266 | 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)); |
2267 | } | } |
2268 | ||
2269 | static void check_newlinechar(compiler_common *common, int nltype, jump_list **fallbacks, BOOL jumpiftrue) | static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue) |
2270 | { | { |
2271 | /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ | /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ |
2272 | DEFINE_COMPILER; | DEFINE_COMPILER; |
# | Line 1679 DEFINE_COMPILER; | Line 2274 DEFINE_COMPILER; |
2274 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2275 | { | { |
2276 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2277 | add_jump(compiler, fallbacks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
2278 | } | } |
2279 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2280 | { | { |
2281 | 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); |
2282 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2283 | 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); |
2284 | 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); |
2285 | add_jump(compiler, fallbacks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
2286 | } | } |
2287 | else | else |
2288 | { | { |
2289 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2290 | add_jump(compiler, fallbacks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
2291 | } | } |
2292 | } | } |
2293 | ||
2294 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2295 | ||
2296 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2297 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2298 | { | { |
2299 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
# | Line 1706 of the character (>= 0xc0). Return char | Line 2301 of the character (>= 0xc0). Return char |
2301 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2302 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2303 | ||
2304 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2305 | /* Searching for the first zero. */ | /* Searching for the first zero. */ |
2306 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); |
2307 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
# | Line 1765 DEFINE_COMPILER; | Line 2360 DEFINE_COMPILER; |
2360 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2361 | struct sljit_jump *compare; | struct sljit_jump *compare; |
2362 | ||
2363 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2364 | ||
2365 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); |
2366 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
# | Line 1786 sljit_emit_fast_return(compiler, RETURN_ | Line 2381 sljit_emit_fast_return(compiler, RETURN_ |
2381 | JUMPHERE(jump); | JUMPHERE(jump); |
2382 | ||
2383 | /* We only have types for characters less than 256. */ | /* We only have types for characters less than 256. */ |
2384 | 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); |
2385 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2386 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2387 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2388 | } | } |
2389 | ||
2390 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
2391 | ||
#ifdef COMPILE_PCRE16 | ||
2392 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2393 | { | { |
2394 | /* 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 1802 of the character (>= 0xd800). Return cha | Line 2396 of the character (>= 0xd800). Return cha |
2396 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2397 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2398 | ||
2399 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2400 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); |
2401 | /* Do nothing, only return. */ | /* Do nothing, only return. */ |
2402 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
# | Line 1819 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC | Line 2413 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC |
2413 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2414 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2415 | } | } |
#endif /* COMPILE_PCRE16 */ | ||
2416 | ||
2417 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16] */ |
2418 | ||
2419 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2420 | ||
# | Line 1839 DEFINE_COMPILER; | Line 2432 DEFINE_COMPILER; |
2432 | ||
2433 | SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); | SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); |
2434 | ||
2435 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2436 | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2437 | 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)); |
2438 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); |
2439 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2440 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); |
2441 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2)); |
2442 | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); |
2443 | 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)); |
2444 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
2445 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2446 | } | } |
# | Line 1861 struct sljit_label *newlinelabel = NULL; | Line 2454 struct sljit_label *newlinelabel = NULL; |
2454 | struct sljit_jump *start; | struct sljit_jump *start; |
2455 | struct sljit_jump *end = NULL; | struct sljit_jump *end = NULL; |
2456 | struct sljit_jump *nl = NULL; | struct sljit_jump *nl = NULL; |
2457 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2458 | struct sljit_jump *singlechar; | struct sljit_jump *singlechar; |
2459 | #endif | #endif |
2460 | jump_list *newline = NULL; | jump_list *newline = NULL; |
# | Line 1875 if (!(hascrorlf || firstline) && (common | Line 2468 if (!(hascrorlf || firstline) && (common |
2468 | if (firstline) | if (firstline) |
2469 | { | { |
2470 | /* Search for the end of the first line. */ | /* Search for the end of the first line. */ |
2471 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2472 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); |
2473 | ||
2474 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
2475 | { | { |
# | Line 1887 if (firstline) | Line 2480 if (firstline) |
2480 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2481 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); |
2482 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); |
2483 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | JUMPHERE(end); |
2484 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2485 | } | } |
2486 | else | else |
2487 | { | { |
2488 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2489 | mainloop = LABEL(); | mainloop = LABEL(); |
2490 | /* Continual stores does not cause data dependency. */ | /* Continual stores does not cause data dependency. */ |
2491 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); |
2492 | read_char(common); | read_char(common); |
2493 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
2494 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
2495 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | JUMPHERE(end); |
2496 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | |
2497 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
2498 | } | } |
2499 | ||
2500 | JUMPHERE(end); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | ||
2501 | } | } |
2502 | ||
2503 | start = JUMP(SLJIT_JUMP); | start = JUMP(SLJIT_JUMP); |
# | Line 1915 if (newlinecheck) | Line 2509 if (newlinecheck) |
2509 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2510 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2511 | 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); |
2512 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2513 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2514 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2515 | #endif | #endif |
2516 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2517 | nl = JUMP(SLJIT_JUMP); | nl = JUMP(SLJIT_JUMP); |
# | Line 1929 mainloop = LABEL(); | Line 2523 mainloop = LABEL(); |
2523 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2524 | if (common->utf) readuchar = TRUE; | if (common->utf) readuchar = TRUE; |
2525 | #endif | #endif |
2526 | if (newlinecheck) readuchar = TRUE; | if (newlinecheck) readuchar = TRUE; |
2527 | ||
2528 | if (readuchar) | |
2529 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
2530 | ||
2531 | if (newlinecheck) | |
2532 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | |
2533 | ||
2534 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2535 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
2536 | #if defined COMPILE_PCRE8 | |
2537 | if (common->utf) | |
2538 | { | |
2539 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
2540 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2541 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
2542 | JUMPHERE(singlechar); | |
2543 | } | |
2544 | #elif defined COMPILE_PCRE16 | |
2545 | if (common->utf) | |
2546 | { | |
2547 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
2548 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
2549 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2550 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
2551 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
2552 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
2553 | JUMPHERE(singlechar); | |
2554 | } | |
2555 | #endif /* COMPILE_PCRE[8|16] */ | |
2556 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2557 | JUMPHERE(start); | |
2558 | ||
2559 | if (newlinecheck) | |
2560 | { | |
2561 | JUMPHERE(end); | |
2562 | JUMPHERE(nl); | |
2563 | } | |
2564 | ||
2565 | return mainloop; | |
2566 | } | |
2567 | ||
2568 | #define MAX_N_CHARS 3 | |
2569 | ||
2570 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | |
2571 | { | |
2572 | DEFINE_COMPILER; | |
2573 | struct sljit_label *start; | |
2574 | struct sljit_jump *quit; | |
2575 | pcre_uint32 chars[MAX_N_CHARS * 2]; | |
2576 | pcre_uchar *cc = common->start + 1 + IMM2_SIZE; | |
2577 | int location = 0; | |
2578 | pcre_int32 len, c, bit, caseless; | |
2579 | int must_stop; | |
2580 | ||
2581 | /* We do not support alternatives now. */ | |
2582 | if (*(common->start + GET(common->start, 1)) == OP_ALT) | |
2583 | return FALSE; | |
2584 | ||
2585 | while (TRUE) | |
2586 | { | |
2587 | caseless = 0; | |
2588 | must_stop = 1; | |
2589 | switch(*cc) | |
2590 | { | |
2591 | case OP_CHAR: | |
2592 | must_stop = 0; | |
2593 | cc++; | |
2594 | break; | |
2595 | ||
2596 | case OP_CHARI: | |
2597 | caseless = 1; | |
2598 | must_stop = 0; | |
2599 | cc++; | |
2600 | break; | |
2601 | ||
2602 | case OP_SOD: | |
2603 | case OP_SOM: | |
2604 | case OP_SET_SOM: | |
2605 | case OP_NOT_WORD_BOUNDARY: | |
2606 | case OP_WORD_BOUNDARY: | |
2607 | case OP_EODN: | |
2608 | case OP_EOD: | |
2609 | case OP_CIRC: | |
2610 | case OP_CIRCM: | |
2611 | case OP_DOLL: | |
2612 | case OP_DOLLM: | |
2613 | /* Zero width assertions. */ | |
2614 | cc++; | |
2615 | continue; | |
2616 | ||
2617 | case OP_PLUS: | |
2618 | case OP_MINPLUS: | |
2619 | case OP_POSPLUS: | |
2620 | cc++; | |
2621 | break; | |
2622 | ||
2623 | case OP_EXACT: | |
2624 | cc += 1 + IMM2_SIZE; | |
2625 | break; | |
2626 | ||
2627 | case OP_PLUSI: | |
2628 | case OP_MINPLUSI: | |
2629 | case OP_POSPLUSI: | |
2630 | caseless = 1; | |
2631 | cc++; | |
2632 | break; | |
2633 | ||
2634 | case OP_EXACTI: | |
2635 | caseless = 1; | |
2636 | cc += 1 + IMM2_SIZE; | |
2637 | break; | |
2638 | ||
2639 | default: | |
2640 | must_stop = 2; | |
2641 | break; | |
2642 | } | |
2643 | ||
2644 | if (must_stop == 2) | |
2645 | break; | |
2646 | ||
2647 | len = 1; | |
2648 | #ifdef SUPPORT_UTF | |
2649 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | |
2650 | #endif | |
2651 | ||
2652 | if (caseless && char_has_othercase(common, cc)) | |
2653 | { | |
2654 | caseless = char_get_othercase_bit(common, cc); | |
2655 | if (caseless == 0) | |
2656 | return FALSE; | |
2657 | #ifdef COMPILE_PCRE8 | |
2658 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | |
2659 | #else | |
2660 | if ((caseless & 0x100) != 0) | |
2661 | caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | |
2662 | else | |
2663 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | |
2664 | #endif | |
2665 | } | |
2666 | else | |
2667 | caseless = 0; | |
2668 | ||
2669 | if (readuchar) | while (len > 0 && location < MAX_N_CHARS * 2) |
2670 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | { |
2671 | c = *cc; | |
2672 | bit = 0; | |
2673 | if (len == (caseless & 0xff)) | |
2674 | { | |
2675 | bit = caseless >> 8; | |
2676 | c |= bit; | |
2677 | } | |
2678 | ||
2679 | if (newlinecheck) | chars[location] = c; |
2680 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | chars[location + 1] = bit; |
2681 | ||
2682 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | len--; |
2683 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | location += 2; |
2684 | if (common->utf) | cc++; |
2685 | { | } |
2686 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
2687 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | if (location >= MAX_N_CHARS * 2 || must_stop != 0) |
2688 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | break; |
JUMPHERE(singlechar); | ||
2689 | } | } |
2690 | #endif | |
2691 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | /* At least two characters are required. */ |
2692 | if (common->utf) | if (location < 2 * 2) |
2693 | return FALSE; | |
2694 | ||
2695 | if (firstline) | |
2696 | { | { |
2697 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | SLJIT_ASSERT(common->first_line_end != 0); |
2698 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
2699 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, (location >> 1) - 1); |
COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
JUMPHERE(singlechar); | ||
2700 | } | } |
2701 | #endif | else |
2702 | JUMPHERE(start); | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); |
2703 | ||
2704 | if (newlinecheck) | start = LABEL(); |
2705 | { | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2706 | JUMPHERE(end); | |
2707 | JUMPHERE(nl); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2708 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2709 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2710 | if (chars[1] != 0) | |
2711 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | |
2712 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | |
2713 | if (location > 2 * 2) | |
2714 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2715 | if (chars[3] != 0) | |
2716 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); | |
2717 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); | |
2718 | if (location > 2 * 2) | |
2719 | { | |
2720 | if (chars[5] != 0) | |
2721 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); | |
2722 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); | |
2723 | } | } |
2724 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2725 | ||
2726 | return mainloop; | JUMPHERE(quit); |
2727 | ||
2728 | if (firstline) | |
2729 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | |
2730 | else | |
2731 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); | |
2732 | return TRUE; | |
2733 | } | } |
2734 | ||
2735 | #undef MAX_N_CHARS | |
2736 | ||
2737 | 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) |
2738 | { | { |
2739 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2740 | struct sljit_label *start; | struct sljit_label *start; |
2741 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2742 | struct sljit_jump *found; | struct sljit_jump *found; |
2743 | pcre_uchar oc, bit; | pcre_uchar oc, bit; |
2744 | ||
2745 | if (firstline) | if (firstline) |
2746 | { | { |
2747 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2748 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
2749 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
2750 | } | } |
2751 | ||
2752 | start = LABEL(); | start = LABEL(); |
2753 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2754 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2755 | ||
2756 | oc = first_char; | oc = first_char; |
# | Line 2002 if (first_char == oc) | Line 2767 if (first_char == oc) |
2767 | else | else |
2768 | { | { |
2769 | bit = first_char ^ oc; | bit = first_char ^ oc; |
2770 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
2771 | { | { |
2772 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
2773 | 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 2010 else | Line 2775 else |
2775 | else | else |
2776 | { | { |
2777 | 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); |
2778 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2779 | 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); |
2780 | 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); |
2781 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
2782 | } | } |
2783 | } | } |
2784 | ||
2785 | 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)); |
#if defined SUPPORT_UTF && defined COMPILE_PCRE8 | ||
if (common->utf) | ||
{ | ||
CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | ||
OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
} | ||
#endif | ||
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
if (common->utf) | ||
{ | ||
CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | ||
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | ||
OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | ||
COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
} | ||
#endif | ||
2786 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
2787 | JUMPHERE(found); | JUMPHERE(found); |
2788 | JUMPHERE(leave); | JUMPHERE(quit); |
2789 | ||
2790 | if (firstline) | if (firstline) |
2791 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2792 | } | } |
2793 | ||
2794 | static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) | static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) |
# | Line 2051 DEFINE_COMPILER; | Line 2797 DEFINE_COMPILER; |
2797 | struct sljit_label *loop; | struct sljit_label *loop; |
2798 | struct sljit_jump *lastchar; | struct sljit_jump *lastchar; |
2799 | struct sljit_jump *firstchar; | struct sljit_jump *firstchar; |
2800 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2801 | struct sljit_jump *foundcr = NULL; | struct sljit_jump *foundcr = NULL; |
2802 | struct sljit_jump *notfoundnl; | struct sljit_jump *notfoundnl; |
2803 | jump_list *newline = NULL; | jump_list *newline = NULL; |
2804 | ||
2805 | if (firstline) | if (firstline) |
2806 | { | { |
2807 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2808 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
2809 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
2810 | } | } |
2811 | ||
2812 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
# | Line 2072 if (common->nltype == NLTYPE_FIXED && co | Line 2819 if (common->nltype == NLTYPE_FIXED && co |
2819 | ||
2820 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
2821 | 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); |
2822 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); |
2823 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2824 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2825 | #endif | #endif |
2826 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2827 | ||
2828 | loop = LABEL(); | loop = LABEL(); |
2829 | 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)); |
2830 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2831 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
2832 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
2833 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); |
2834 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); |
2835 | ||
2836 | JUMPHERE(leave); | JUMPHERE(quit); |
2837 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2838 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2839 | ||
# | Line 2110 set_jumps(newline, loop); | Line 2857 set_jumps(newline, loop); |
2857 | ||
2858 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
2859 | { | { |
2860 | leave = JUMP(SLJIT_JUMP); | quit = JUMP(SLJIT_JUMP); |
2861 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
2862 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2863 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2864 | 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); |
2865 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2866 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2867 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2868 | #endif | #endif |
2869 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2870 | JUMPHERE(notfoundnl); | JUMPHERE(notfoundnl); |
2871 | JUMPHERE(leave); | JUMPHERE(quit); |
2872 | } | } |
2873 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2874 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2875 | ||
2876 | if (firstline) | if (firstline) |
2877 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2878 | } | } |
2879 | ||
2880 | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) |
2881 | { | { |
2882 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2883 | struct sljit_label *start; | struct sljit_label *start; |
2884 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2885 | struct sljit_jump *found; | struct sljit_jump *found; |
2886 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
2887 | struct sljit_jump *jump; | struct sljit_jump *jump; |
# | Line 2142 struct sljit_jump *jump; | Line 2889 struct sljit_jump *jump; |
2889 | ||
2890 | if (firstline) | if (firstline) |
2891 | { | { |
2892 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2893 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); |
2894 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
2895 | } | } |
2896 | ||
2897 | start = LABEL(); | start = LABEL(); |
2898 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2899 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2900 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2901 | if (common->utf) | if (common->utf) |
# | Line 2170 if (common->utf) | Line 2918 if (common->utf) |
2918 | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
2919 | #endif | #endif |
2920 | 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)); |
2921 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #ifdef SUPPORT_UTF |
2922 | #if defined COMPILE_PCRE8 | |
2923 | if (common->utf) | if (common->utf) |
2924 | { | { |
2925 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
2926 | 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); |
2927 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2928 | } | } |
2929 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2930 | if (common->utf) | if (common->utf) |
2931 | { | { |
2932 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
2933 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2934 | 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); |
2935 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2936 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2937 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2938 | } | } |
2939 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2940 | #endif /* SUPPORT_UTF */ | |
2941 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
2942 | JUMPHERE(found); | JUMPHERE(found); |
2943 | JUMPHERE(leave); | JUMPHERE(quit); |
2944 | ||
2945 | if (firstline) | if (firstline) |
2946 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0); |
2947 | } | } |
2948 | ||
2949 | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) |
# | Line 2206 struct sljit_jump *alreadyfound; | Line 2955 struct sljit_jump *alreadyfound; |
2955 | struct sljit_jump *found; | struct sljit_jump *found; |
2956 | struct sljit_jump *foundoc = NULL; | struct sljit_jump *foundoc = NULL; |
2957 | struct sljit_jump *notfound; | struct sljit_jump *notfound; |
2958 | pcre_uchar oc, bit; | pcre_uint32 oc, bit; |
2959 | ||
2960 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), REQ_CHAR_PTR); | SLJIT_ASSERT(common->req_char_ptr != 0); |
2961 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); | |
2962 | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); |
2963 | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); |
2964 | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); |
# | Line 2236 if (req_char == oc) | Line 2986 if (req_char == oc) |
2986 | else | else |
2987 | { | { |
2988 | bit = req_char ^ oc; | bit = req_char ^ oc; |
2989 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
2990 | { | { |
2991 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
2992 | 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 2253 JUMPTO(SLJIT_JUMP, loop); | Line 3003 JUMPTO(SLJIT_JUMP, loop); |
3003 | JUMPHERE(found); | JUMPHERE(found); |
3004 | if (foundoc) | if (foundoc) |
3005 | JUMPHERE(foundoc); | JUMPHERE(foundoc); |
3006 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), REQ_CHAR_PTR, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, TMP1, 0); |
3007 | JUMPHERE(alreadyfound); | JUMPHERE(alreadyfound); |
3008 | JUMPHERE(toolong); | JUMPHERE(toolong); |
3009 | return notfound; | return notfound; |
# | Line 2265 DEFINE_COMPILER; | Line 3015 DEFINE_COMPILER; |
3015 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3016 | struct sljit_label *mainloop; | struct sljit_label *mainloop; |
3017 | ||
3018 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3019 | OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); | OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); |
3020 | GET_LOCAL_BASE(TMP3, 0, 0); | |
3021 | ||
3022 | /* Drop frames until we reach STACK_TOP. */ | /* Drop frames until we reach STACK_TOP. */ |
3023 | mainloop = LABEL(); | mainloop = LABEL(); |
3024 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
3025 | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); |
3026 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_LOCALS_REG, 0); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
3027 | 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)); |
3028 | 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)); |
3029 | 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)); |
3030 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3031 | ||
3032 | JUMPHERE(jump); | JUMPHERE(jump); |
# | Line 2286 sljit_emit_fast_return(compiler, RETURN_ | Line 3037 sljit_emit_fast_return(compiler, RETURN_ |
3037 | JUMPHERE(jump); | JUMPHERE(jump); |
3038 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); |
3039 | /* Set string begin. */ | /* Set string begin. */ |
3040 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
3041 | 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)); |
3042 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); |
3043 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3044 | ||
3045 | JUMPHERE(jump); | JUMPHERE(jump); |
3046 | if (common->mark_ptr != 0) | |
3047 | { | |
3048 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark); | |
3049 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); | |
3050 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); | |
3051 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); | |
3052 | JUMPTO(SLJIT_JUMP, mainloop); | |
3053 | ||
3054 | JUMPHERE(jump); | |
3055 | } | |
3056 | ||
3057 | /* Unknown command. */ | /* Unknown command. */ |
3058 | 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)); |
3059 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3060 | } | } |
3061 | ||
# | Line 2307 struct sljit_jump *jump; | Line 3069 struct sljit_jump *jump; |
3069 | ||
3070 | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); |
3071 | ||
3072 | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
3073 | /* Get type of the previous char, and put it to LOCALS1. */ | /* Get type of the previous char, and put it to LOCALS1. */ |
3074 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
3075 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
# | Line 2326 if (common->use_ucp) | Line 3088 if (common->use_ucp) |
3088 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3089 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3090 | 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); |
3091 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3092 | 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); |
3093 | 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); |
3094 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3095 | JUMPHERE(jump); | JUMPHERE(jump); |
3096 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); |
3097 | } | } |
# | Line 2370 if (common->use_ucp) | Line 3132 if (common->use_ucp) |
3132 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3133 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3134 | 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); |
3135 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3136 | 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); |
3137 | 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); |
3138 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3139 | JUMPHERE(jump); | JUMPHERE(jump); |
3140 | } | } |
3141 | else | else |
# | Line 2405 OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE | Line 3167 OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE |
3167 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
3168 | } | } |
3169 | ||
3170 | /* | |
3171 | range format: | |
3172 | ||
3173 | ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). | |
3174 | ranges[1] = first bit (0 or 1) | |
3175 | ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) | |
3176 | */ | |
3177 | ||
3178 | static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) | |
3179 | { | |
3180 | DEFINE_COMPILER; | |
3181 | struct sljit_jump *jump; | |
3182 | ||
3183 | if (ranges[0] < 0) | |
3184 | return FALSE; | |
3185 | ||
3186 | switch(ranges[0]) | |
3187 | { | |
3188 | case 1: | |
3189 | if (readch) | |
3190 | read_char(common); | |
3191 | add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3192 | return TRUE; | |
3193 | ||
3194 | case 2: | |
3195 | if (readch) | |
3196 | read_char(common); | |
3197 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3198 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); | |
3199 | return TRUE; | |
3200 | ||
3201 | case 4: | |
3202 | if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) | |
3203 | { | |
3204 | if (readch) | |
3205 | read_char(common); | |
3206 | if (ranges[1] != 0) | |
3207 | { | |
3208 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3209 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3210 | } | |
3211 | else | |
3212 | { | |
3213 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3214 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3215 | JUMPHERE(jump); | |
3216 | } | |
3217 | return TRUE; | |
3218 | } | |
3219 | if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) | |
3220 | { | |
3221 | if (readch) | |
3222 | read_char(common); | |
3223 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); | |
3224 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); | |
3225 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); | |
3226 | return TRUE; | |
3227 | } | |
3228 | return FALSE; | |
3229 | ||
3230 | default: | |
3231 | return FALSE; | |
3232 | } | |
3233 | } | |
3234 | ||
3235 | static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) | |
3236 | { | |
3237 | int i, bit, length; | |
3238 | const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; | |
3239 | ||
3240 | bit = ctypes[0] & flag; | |
3241 | ranges[0] = -1; | |
3242 | ranges[1] = bit != 0 ? 1 : 0; | |
3243 | length = 0; | |
3244 | ||
3245 | for (i = 1; i < 256; i++) | |
3246 | if ((ctypes[i] & flag) != bit) | |
3247 | { | |
3248 | if (length >= MAX_RANGE_SIZE) | |
3249 | return; | |
3250 | ranges[2 + length] = i; | |
3251 | length++; | |
3252 | bit ^= flag; | |
3253 | } | |
3254 | ||
3255 | if (bit != 0) | |
3256 | { | |
3257 | if (length >= MAX_RANGE_SIZE) | |
3258 | return; | |
3259 | ranges[2 + length] = 256; | |
3260 | length++; | |
3261 | } | |
3262 | ranges[0] = length; | |
3263 | } | |
3264 | ||
3265 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks) | |
3266 | { | |
3267 | int ranges[2 + MAX_RANGE_SIZE]; | |
3268 | pcre_uint8 bit, cbit, all; | |
3269 | int i, byte, length = 0; | |
3270 | ||
3271 | bit = bits[0] & 0x1; | |
3272 | ranges[1] = bit; | |
3273 | /* Can be 0 or 255. */ | |
3274 | all = -bit; | |
3275 | ||
3276 | for (i = 0; i < 256; ) | |
3277 | { | |
3278 | byte = i >> 3; | |
3279 | if ((i & 0x7) == 0 && bits[byte] == all) | |
3280 | i += 8; | |
3281 | else | |
3282 | { | |
3283 | cbit = (bits[byte] >> (i & 0x7)) & 0x1; | |
3284 | if (cbit != bit) | |
3285 | { | |
3286 | if (length >= MAX_RANGE_SIZE) | |
3287 | return FALSE; | |
3288 | ranges[2 + length] = i; | |
3289 | length++; | |
3290 | bit = cbit; | |
3291 | all = -cbit; | |
3292 | } | |
3293 | i++; | |
3294 | } | |
3295 | } | |
3296 | ||
3297 | if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) | |
3298 | { | |
3299 | if (length >= MAX_RANGE_SIZE) | |
3300 | return FALSE; | |
3301 | ranges[2 + length] = 256; | |
3302 | length++; | |
3303 | } | |
3304 | ranges[0] = length; | |
3305 | ||
3306 | return check_ranges(common, ranges, backtracks, FALSE); | |
3307 | } | |
3308 | ||
3309 | static void check_anynewline(compiler_common *common) | static void check_anynewline(compiler_common *common) |
3310 | { | { |
3311 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3312 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3313 | ||
3314 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3315 | ||
3316 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3317 | 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); |
3318 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3319 | 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); |
3320 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3321 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3322 | if (common->utf) | if (common->utf) |
3323 | { | { |
3324 | #endif | #endif |
3325 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3326 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3327 | 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); |
3328 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3329 | } | } |
3330 | #endif | #endif |
3331 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3332 | 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); |
3333 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3334 | } | } |
3335 | ||
# | Line 2437 static void check_hspace(compiler_common | Line 3338 static void check_hspace(compiler_common |
3338 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3339 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3340 | ||
3341 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3342 | ||
3343 | 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); |
3344 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3345 | 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); |
3346 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3347 | 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); |
3348 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3349 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3350 | if (common->utf) | if (common->utf) |
3351 | { | { |
3352 | #endif | #endif |
3353 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3354 | 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); |
3355 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3356 | 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); |
3357 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3358 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); |
3359 | 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); |
3360 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3361 | 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); |
3362 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3363 | 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); |
3364 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3365 | 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); |
3366 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3367 | } | } |
3368 | #endif | #endif |
3369 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3370 | 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); |
3371 | ||
3372 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3373 | } | } |
# | Line 2476 static void check_vspace(compiler_common | Line 3377 static void check_vspace(compiler_common |
3377 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3378 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3379 | ||
3380 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3381 | ||
3382 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3383 | 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); |
3384 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3385 | 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); |
3386 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3387 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3388 | if (common->utf) | if (common->utf) |
3389 | { | { |
3390 | #endif | #endif |
3391 | 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); |
3392 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3393 | 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); |
3394 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3395 | } | } |
3396 | #endif | #endif |
3397 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3398 | 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); |
3399 | ||
3400 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3401 | } | } |
# | Line 2508 DEFINE_COMPILER; | Line 3409 DEFINE_COMPILER; |
3409 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3410 | struct sljit_label *label; | struct sljit_label *label; |
3411 | ||
3412 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3413 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
3414 | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); |
3415 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); |
# | Line 2537 DEFINE_COMPILER; | Line 3438 DEFINE_COMPILER; |
3438 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3439 | struct sljit_label *label; | struct sljit_label *label; |
3440 | ||
3441 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3442 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
3443 | ||
3444 | OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); | OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); |
# | Line 2583 sljit_emit_fast_return(compiler, RETURN_ | Line 3484 sljit_emit_fast_return(compiler, RETURN_ |
3484 | 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) |
3485 | { | { |
3486 | /* This function would be ineffective to do in JIT level. */ | /* This function would be ineffective to do in JIT level. */ |
3487 | int c1, c2; | pcre_uint32 c1, c2; |
3488 | const pcre_uchar *src2 = args->ptr; | const pcre_uchar *src2 = args->uchar_ptr; |
3489 | const pcre_uchar *end2 = args->end; | const pcre_uchar *end2 = args->end; |
3490 | const ucd_record *ur; | |
3491 | const pcre_uint32 *pp; | |
3492 | ||
3493 | while (src1 < end1) | while (src1 < end1) |
3494 | { | { |
# | Line 2593 while (src1 < end1) | Line 3496 while (src1 < end1) |
3496 | return (pcre_uchar*)1; | return (pcre_uchar*)1; |
3497 | GETCHARINC(c1, src1); | GETCHARINC(c1, src1); |
3498 | GETCHARINC(c2, src2); | GETCHARINC(c2, src2); |
3499 | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL; | ur = GET_UCD(c2); |
3500 | if (c1 != c2 && c1 != c2 + ur->other_case) | |
3501 | { | |
3502 | pp = PRIV(ucd_caseless_sets) + ur->caseset; | |
3503 | for (;;) | |
3504 | { | |
3505 | if (c1 < *pp) return NULL; | |
3506 | if (c1 == *pp++) break; | |
3507 | } | |
3508 | } | |
3509 | } | } |
3510 | return src2; | return src2; |
3511 | } | } |
# | Line 2601 return src2; | Line 3513 return src2; |
3513 | #endif /* SUPPORT_UTF && SUPPORT_UCP */ | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
3514 | ||
3515 | static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, | static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, |
3516 | compare_context* context, jump_list **fallbacks) | compare_context* context, jump_list **backtracks) |
3517 | { | { |
3518 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3519 | unsigned int othercasebit = 0; | unsigned int othercasebit = 0; |
# | Line 2615 if (caseless && char_has_othercase(commo | Line 3527 if (caseless && char_has_othercase(commo |
3527 | othercasebit = char_get_othercase_bit(common, cc); | othercasebit = char_get_othercase_bit(common, cc); |
3528 | SLJIT_ASSERT(othercasebit); | SLJIT_ASSERT(othercasebit); |
3529 | /* Extracting bit difference info. */ | /* Extracting bit difference info. */ |
3530 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3531 | othercasechar = cc + (othercasebit >> 8); | othercasechar = cc + (othercasebit >> 8); |
3532 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3533 | #else | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3534 | #ifdef COMPILE_PCRE16 | /* Note that this code only handles characters in the BMP. If there |
3535 | ever are characters outside the BMP whose othercase differs in only one | |
3536 | bit from itself (there currently are none), this code will need to be | |
3537 | revised for COMPILE_PCRE32. */ | |
3538 | othercasechar = cc + (othercasebit >> 9); | othercasechar = cc + (othercasebit >> 9); |
3539 | if ((othercasebit & 0x100) != 0) | if ((othercasebit & 0x100) != 0) |
3540 | othercasebit = (othercasebit & 0xff) << 8; | othercasebit = (othercasebit & 0xff) << 8; |
3541 | else | else |
3542 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3543 | #endif | #endif /* COMPILE_PCRE[8|16|32] */ |
#endif | ||
3544 | } | } |
3545 | ||
3546 | if (context->sourcereg == -1) | if (context->sourcereg == -1) |
3547 | { | { |
3548 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3549 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3550 | if (context->length >= 4) | if (context->length >= 4) |
3551 | 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 2640 if (context->sourcereg == -1) | Line 3554 if (context->sourcereg == -1) |
3554 | else | else |
3555 | #endif | #endif |
3556 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3557 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
3558 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3559 | if (context->length >= 4) | if (context->length >= 4) |
3560 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3561 | else | else |
3562 | #endif | #endif |
3563 | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3564 | #endif | #elif defined COMPILE_PCRE32 |
3565 | #endif /* COMPILE_PCRE8 */ | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3566 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3567 | context->sourcereg = TMP2; | context->sourcereg = TMP2; |
3568 | } | } |
3569 | ||
# | Line 2678 do | Line 3592 do |
3592 | } | } |
3593 | context->ucharptr++; | context->ucharptr++; |
3594 | ||
3595 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3596 | 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)) |
3597 | #else | #elif defined COMPILE_PCRE16 |
3598 | if (context->ucharptr >= 2 || context->length == 0) | if (context->ucharptr >= 2 || context->length == 0) |
3599 | #elif defined COMPILE_PCRE32 | |
3600 | if (1 /* context->ucharptr >= 1 || context->length == 0 */) | |
3601 | #endif | #endif |
3602 | { | { |
3603 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 | |
3604 | if (context->length >= 4) | if (context->length >= 4) |
3605 | 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); |
3606 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3607 | else if (context->length >= 2) | else if (context->length >= 2) |
3608 | 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); |
3609 | else if (context->length >= 1) | else if (context->length >= 1) |
3610 | 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); |
3611 | #else | #elif defined COMPILE_PCRE16 |
3612 | else if (context->length >= 2) | else if (context->length >= 2) |
3613 | 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); |
3614 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
3615 | #elif defined COMPILE_PCRE32 | |
3616 | OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
3617 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3618 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3619 | ||
3620 | switch(context->ucharptr) | switch(context->ucharptr) |
# | Line 2702 do | Line 3622 do |
3622 | case 4 / sizeof(pcre_uchar): | case 4 / sizeof(pcre_uchar): |
3623 | if (context->oc.asint != 0) | if (context->oc.asint != 0) |
3624 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); |
3625 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); |
3626 | break; | break; |
3627 | ||
3628 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 | |
3629 | case 2 / sizeof(pcre_uchar): | case 2 / sizeof(pcre_uchar): |
3630 | if (context->oc.asushort != 0) | if (context->oc.asushort != 0) |
3631 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); |
3632 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); |
3633 | break; | break; |
3634 | ||
3635 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3636 | case 1: | case 1: |
3637 | if (context->oc.asbyte != 0) | if (context->oc.asbyte != 0) |
3638 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); |
3639 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); |
3640 | break; | break; |
3641 | #endif | #endif |
3642 | ||
3643 | #endif /* COMPILE_PCRE[8|16] */ | |
3644 | ||
3645 | default: | default: |
3646 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
3647 | break; | break; |
# | Line 2729 do | Line 3652 do |
3652 | #else | #else |
3653 | ||
3654 | /* Unaligned read is unsupported. */ | /* Unaligned read is unsupported. */ |
#ifdef COMPILE_PCRE8 | ||
if (context->length > 0) | ||
OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | ||
#else | ||
3655 | if (context->length > 0) | if (context->length > 0) |
3656 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3657 | #endif | |
3658 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3659 | ||
3660 | if (othercasebit != 0 && othercasechar == cc) | if (othercasebit != 0 && othercasechar == cc) |
3661 | { | { |
3662 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); |
3663 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); |
3664 | } | } |
3665 | else | else |
3666 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); |
3667 | ||
3668 | #endif | #endif |
3669 | ||
# | Line 2780 return cc; | Line 3699 return cc; |
3699 | } \ | } \ |
3700 | charoffset = (value); | charoffset = (value); |
3701 | ||
3702 | static void compile_xclass_hotpath(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks) | static void compile_xclass_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) |
3703 | { | { |
3704 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3705 | jump_list *found = NULL; | jump_list *found = NULL; |
3706 | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : fallbacks; | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks; |
3707 | unsigned int c; | pcre_int32 c, charoffset; |
3708 | int compares; | const pcre_uint32 *other_cases; |
3709 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
3710 | pcre_uchar *ccbegin; | pcre_uchar *ccbegin; |
3711 | int compares, invertcmp, numberofcmps; | |
3712 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3713 | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; |
3714 | BOOL charsaved = FALSE; | BOOL charsaved = FALSE; |
3715 | int typereg = TMP1, scriptreg = TMP1; | int typereg = TMP1, scriptreg = TMP1; |
3716 | unsigned int typeoffset; | pcre_int32 typeoffset; |
3717 | #endif | #endif |
int invertcmp, numberofcmps; | ||
unsigned int charoffset; | ||
3718 | ||
3719 | /* Although SUPPORT_UTF must be defined, we are not necessary in utf mode. */ | /* Although SUPPORT_UTF must be defined, we are |
3720 | fallback_at_str_end(common, fallbacks); | not necessary in utf mode even in 8 bit mode. */ |
3721 | detect_partial_match(common, backtracks); | |
3722 | read_char(common); | read_char(common); |
3723 | ||
3724 | if ((*cc++ & XCL_MAP) != 0) | if ((*cc++ & XCL_MAP) != 0) |
# | Line 2812 if ((*cc++ & XCL_MAP) != 0) | Line 3731 if ((*cc++ & XCL_MAP) != 0) |
3731 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
3732 | #endif | #endif |
3733 | ||
3734 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list)) |
3735 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | { |
3736 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3737 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3738 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); |
3739 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3740 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | |
3741 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | |
3742 | } | |
3743 | ||
3744 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3745 | JUMPHERE(jump); | JUMPHERE(jump); |
# | Line 2890 while (*cc != XCL_END) | Line 3812 while (*cc != XCL_END) |
3812 | needschar = TRUE; | needschar = TRUE; |
3813 | break; | break; |
3814 | ||
3815 | case PT_CLIST: | |
3816 | needschar = TRUE; | |
3817 | break; | |
3818 | ||
3819 | default: | default: |
3820 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
3821 | break; | break; |
# | Line 2926 if (needstype || needsscript) | Line 3852 if (needstype || needsscript) |
3852 | { | { |
3853 | if (scriptreg == TMP1) | if (scriptreg == TMP1) |
3854 | { | { |
3855 | 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)); |
3856 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); |
3857 | } | } |
3858 | else | else |
3859 | { | { |
3860 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); |
3861 | 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)); |
3862 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); |
3863 | } | } |
3864 | } | } |
# | Line 2950 typeoffset = 0; | Line 3876 typeoffset = 0; |
3876 | while (*cc != XCL_END) | while (*cc != XCL_END) |
3877 | { | { |
3878 | compares--; | compares--; |
3879 | invertcmp = (compares == 0 && list != fallbacks); | invertcmp = (compares == 0 && list != backtracks); |
3880 | jump = NULL; | jump = NULL; |
3881 | ||
3882 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
# | Line 2968 while (*cc != XCL_END) | Line 3894 while (*cc != XCL_END) |
3894 | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) |
3895 | { | { |
3896 | 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); |
3897 | 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); |
3898 | numberofcmps++; | numberofcmps++; |
3899 | } | } |
3900 | else if (numberofcmps > 0) | else if (numberofcmps > 0) |
3901 | { | { |
3902 | 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); |
3903 | 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); |
3904 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
3905 | numberofcmps = 0; | numberofcmps = 0; |
3906 | } | } |
# | Line 3007 while (*cc != XCL_END) | Line 3933 while (*cc != XCL_END) |
3933 | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) |
3934 | { | { |
3935 | 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); |
3936 | 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); |
3937 | numberofcmps++; | numberofcmps++; |
3938 | } | } |
3939 | else if (numberofcmps > 0) | else if (numberofcmps > 0) |
3940 | { | { |
3941 | 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); |
3942 | 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); |
3943 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
3944 | numberofcmps = 0; | numberofcmps = 0; |
3945 | } | } |
# | Line 3032 while (*cc != XCL_END) | Line 3958 while (*cc != XCL_END) |
3958 | switch(*cc) | switch(*cc) |
3959 | { | { |
3960 | case PT_ANY: | case PT_ANY: |
3961 | if (list != fallbacks) | if (list != backtracks) |
3962 | { | { |
3963 | if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) | if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) |
3964 | continue; | continue; |
# | Line 3044 while (*cc != XCL_END) | Line 3970 while (*cc != XCL_END) |
3970 | ||
3971 | case PT_LAMP: | case PT_LAMP: |
3972 | 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); |
3973 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3974 | 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); |
3975 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3976 | 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); |
3977 | 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); |
3978 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
3979 | break; | break; |
3980 | ||
# | Line 3075 while (*cc != XCL_END) | Line 4001 while (*cc != XCL_END) |
4001 | } | } |
4002 | SET_CHAR_OFFSET(9); | SET_CHAR_OFFSET(9); |
4003 | 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, 13 - 9); |
4004 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
4005 | if (*cc == PT_SPACE) | if (*cc == PT_SPACE) |
4006 | JUMPHERE(jump); | JUMPHERE(jump); |
4007 | ||
4008 | SET_TYPE_OFFSET(ucp_Zl); | SET_TYPE_OFFSET(ucp_Zl); |
4009 | 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); |
4010 | 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); |
4011 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4012 | break; | break; |
4013 | ||
4014 | case PT_WORD: | case PT_WORD: |
4015 | 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); |
4016 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
4017 | /* ... fall through */ | /* ... fall through */ |
4018 | ||
4019 | case PT_ALNUM: | case PT_ALNUM: |
4020 | SET_TYPE_OFFSET(ucp_Ll); | SET_TYPE_OFFSET(ucp_Ll); |
4021 | 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_Lu - ucp_Ll); |
4022 | COND_VALUE((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, (*cc == PT_ALNUM) ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL); |
4023 | SET_TYPE_OFFSET(ucp_Nd); | SET_TYPE_OFFSET(ucp_Nd); |
4024 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
4025 | 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); |
4026 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | |
4027 | break; | |
4028 | ||
4029 | case PT_CLIST: | |
4030 | other_cases = PRIV(ucd_caseless_sets) + cc[1]; | |
4031 | ||
4032 | /* At least three characters are required. | |
4033 | Otherwise this case would be handled by the normal code path. */ | |
4034 | SLJIT_ASSERT(other_cases[0] != NOTACHAR && other_cases[1] != NOTACHAR && other_cases[2] != NOTACHAR); | |
4035 | SLJIT_ASSERT(other_cases[0] < other_cases[1] && other_cases[1] < other_cases[2]); | |
4036 | ||
4037 | /* Optimizing character pairs, if their difference is power of 2. */ | |
4038 | if (is_powerof2(other_cases[1] ^ other_cases[0])) | |
4039 | { | |
4040 | if (charoffset == 0) | |
4041 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4042 | else | |
4043 | { | |
4044 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); | |
4045 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4046 | } | |
4047 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]); | |
4048 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4049 | other_cases += 2; | |
4050 | } | |
4051 | else if (is_powerof2(other_cases[2] ^ other_cases[1])) | |
4052 | { | |
4053 | if (charoffset == 0) | |
4054 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[2] ^ other_cases[1]); | |
4055 | else | |
4056 | { | |
4057 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); | |
4058 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4059 | } | |
4060 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]); | |
4061 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4062 | ||
4063 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset); | |
4064 | OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | |
4065 | ||
4066 | other_cases += 3; | |
4067 | } | |
4068 | else | |
4069 | { | |
4070 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4071 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4072 | } | |
4073 | ||
4074 | while (*other_cases != NOTACHAR) | |
4075 | { | |
4076 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4077 | OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | |
4078 | } | |
4079 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4080 | break; | break; |
4081 | } | } |
# | Line 3105 while (*cc != XCL_END) | Line 4084 while (*cc != XCL_END) |
4084 | #endif | #endif |
4085 | ||
4086 | if (jump != NULL) | if (jump != NULL) |
4087 | add_jump(compiler, compares > 0 ? list : fallbacks, jump); | add_jump(compiler, compares > 0 ? list : backtracks, jump); |
4088 | } | } |
4089 | ||
4090 | if (found != NULL) | if (found != NULL) |
# | Line 3117 if (found != NULL) | Line 4096 if (found != NULL) |
4096 | ||
4097 | #endif | #endif |
4098 | ||
4099 | static pcre_uchar *compile_char1_hotpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **fallbacks) | static pcre_uchar *compile_char1_matchingpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **backtracks) |
4100 | { | { |
4101 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4102 | int length; | int length; |
# | Line 3136 switch(type) | Line 4115 switch(type) |
4115 | case OP_SOD: | case OP_SOD: |
4116 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4117 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
4118 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); |
4119 | return cc; | return cc; |
4120 | ||
4121 | case OP_SOM: | case OP_SOM: |
4122 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4123 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); |
4124 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); |
4125 | return cc; | return cc; |
4126 | ||
4127 | case OP_NOT_WORD_BOUNDARY: | case OP_NOT_WORD_BOUNDARY: |
4128 | case OP_WORD_BOUNDARY: | case OP_WORD_BOUNDARY: |
4129 | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); |
4130 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4131 | return cc; | return cc; |
4132 | ||
4133 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
4134 | case OP_DIGIT: | case OP_DIGIT: |
4135 | fallback_at_str_end(common, fallbacks); | /* Digits are usually 0-9, so it is worth to optimize them. */ |
4136 | read_char8_type(common); | if (common->digits[0] == -2) |
4137 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | get_ctype_ranges(common, ctype_digit, common->digits); |
4138 | add_jump(compiler, fallbacks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | detect_partial_match(common, backtracks); |
4139 | /* Flip the starting bit in the negative case. */ | |
4140 | if (type == OP_NOT_DIGIT) | |
4141 | common->digits[1] ^= 1; | |
4142 | if (!check_ranges(common, common->digits, backtracks, TRUE)) | |
4143 | { | |
4144 | read_char8_type(common); | |
4145 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | |
4146 | add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | |
4147 | } | |
4148 | if (type == OP_NOT_DIGIT) | |
4149 | common->digits[1] ^= 1; | |
4150 | return cc; | return cc; |
4151 | ||
4152 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
4153 | case OP_WHITESPACE: | case OP_WHITESPACE: |
4154 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4155 | read_char8_type(common); | read_char8_type(common); |
4156 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); |
4157 | add_jump(compiler, fallbacks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
4158 | return cc; | return cc; |
4159 | ||
4160 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
4161 | case OP_WORDCHAR: | case OP_WORDCHAR: |
4162 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4163 | read_char8_type(common); | read_char8_type(common); |
4164 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); |
4165 | add_jump(compiler, fallbacks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
4166 | return cc; | return cc; |
4167 | ||
4168 | case OP_ANY: | case OP_ANY: |
4169 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4170 | read_char(common); | read_char(common); |
4171 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4172 | { | { |
4173 | 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); |
4174 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | if (common->mode != JIT_PARTIAL_HARD_COMPILE) |
4175 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
4176 | else | |
4177 | jump[1] = check_str_end(common); | |
4178 | ||
4179 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4180 | add_jump(compiler, fallbacks, 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)); |
4181 | JUMPHERE(jump[1]); | if (jump[1] != NULL) |
4182 | JUMPHERE(jump[1]); | |
4183 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4184 | } | } |
4185 | else | else |
4186 | check_newlinechar(common, common->nltype, fallbacks, TRUE); | check_newlinechar(common, common->nltype, backtracks, TRUE); |
4187 | return cc; | return cc; |
4188 | ||
4189 | case OP_ALLANY: | case OP_ALLANY: |
4190 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4191 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
4192 | if (common->utf) | if (common->utf) |
4193 | { | { |
4194 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4195 | 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)); |
4196 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 |
4197 | #if defined COMPILE_PCRE8 | |
4198 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
4199 | 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); |
4200 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4201 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
4202 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
4203 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
4204 | 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); |
4205 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
4206 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4207 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4208 | #endif /* COMPILE_PCRE16 */ | #endif |
#endif /* COMPILE_PCRE8 */ | ||
4209 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4210 | #endif /* COMPILE_PCRE[8|16] */ | |
4211 | return cc; | return cc; |
4212 | } | } |
4213 | #endif | #endif |
# | Line 3220 switch(type) | Line 4215 switch(type) |
4215 | return cc; | return cc; |
4216 | ||
4217 | case OP_ANYBYTE: | case OP_ANYBYTE: |
4218 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4219 | 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)); |
4220 | return cc; | return cc; |
4221 | ||
# | Line 3233 switch(type) | Line 4228 switch(type) |
4228 | propdata[2] = cc[0]; | propdata[2] = cc[0]; |
4229 | propdata[3] = cc[1]; | propdata[3] = cc[1]; |
4230 | propdata[4] = XCL_END; | propdata[4] = XCL_END; |
4231 | compile_xclass_hotpath(common, propdata, fallbacks); | compile_xclass_matchingpath(common, propdata, backtracks); |
4232 | return cc + 2; | return cc + 2; |
4233 | #endif | #endif |
4234 | #endif | #endif |
4235 | ||
4236 | case OP_ANYNL: | case OP_ANYNL: |
4237 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4238 | read_char(common); | read_char(common); |
4239 | 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); |
4240 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | /* We don't need to handle soft partial matching case. */ |
4241 | if (common->mode != JIT_PARTIAL_HARD_COMPILE) | |
4242 | jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
4243 | else | |
4244 | jump[1] = check_str_end(common); | |
4245 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4246 | jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); | jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); |
4247 | 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)); |
4248 | jump[3] = JUMP(SLJIT_JUMP); | jump[3] = JUMP(SLJIT_JUMP); |
4249 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4250 | check_newlinechar(common, common->bsr_nltype, fallbacks, FALSE); | check_newlinechar(common, common->bsr_nltype, backtracks, FALSE); |
4251 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4252 | JUMPHERE(jump[2]); | JUMPHERE(jump[2]); |
4253 | JUMPHERE(jump[3]); | JUMPHERE(jump[3]); |
# | Line 3256 switch(type) | Line 4255 switch(type) |
4255 | ||
4256 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
4257 | case OP_HSPACE: | case OP_HSPACE: |
4258 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4259 | read_char(common); | read_char(common); |
4260 | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); |
4261 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4262 | return cc; | return cc; |
4263 | ||
4264 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
4265 | case OP_VSPACE: | case OP_VSPACE: |
4266 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4267 | read_char(common); | read_char(common); |
4268 | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); |
4269 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4270 | return cc; | return cc; |
4271 | ||
4272 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
4273 | case OP_EXTUNI: | case OP_EXTUNI: |
4274 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4275 | read_char(common); | read_char(common); |
4276 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4277 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); |
4278 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc)); | /* Optimize register allocation: use a real register. */ |
4279 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); | |
4280 | OP1(SLJIT_MOV_UB, STACK_TOP, 0, SLJIT_MEM2(TMP1, TMP2), 3); | |
4281 | ||
4282 | label = LABEL(); | label = LABEL(); |
4283 | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
4284 | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); |
4285 | read_char(common); | read_char(common); |
4286 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4287 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); |
4288 | CMPTO(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc, label); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
4289 | ||
4290 | OP2(SLJIT_SHL, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2); | |
4291 | OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(STACK_TOP), (sljit_sw)PRIV(ucp_gbtable)); | |
4292 | OP1(SLJIT_MOV, STACK_TOP, 0, TMP2, 0); | |
4293 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | |
4294 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | |
4295 | JUMPTO(SLJIT_C_NOT_ZERO, label); | |
4296 | ||
4297 | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
4298 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4299 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | |
4300 | ||
4301 | if (common->mode == JIT_PARTIAL_HARD_COMPILE) | if (common->mode == JIT_PARTIAL_HARD_COMPILE) |
4302 | { | { |
4303 | jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
4304 | check_partial(common); | /* Since we successfully read a char above, partial matching must occure. */ |
4305 | check_partial(common, TRUE); | |
4306 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4307 | } | } |
4308 | return cc; | return cc; |
4309 | #endif | #endif |
4310 | ||
4311 | case OP_EODN: | case OP_EODN: |
4312 | /* Requires rather complex checks. */ | |
4313 | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
4314 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4315 | { | { |
4316 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4317 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
4318 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | if (common->mode == JIT_COMPILE) |
4319 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | |
4320 | else | |
4321 | { | |
4322 | jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); | |
4323 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); | |
4324 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS); | |
4325 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); | |
4326 | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL); | |
4327 | add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL)); | |
4328 | check_partial(common, TRUE); | |
4329 | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); | |
4330 | JUMPHERE(jump[1]); | |
4331 | } | |
4332 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
4333 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
4334 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
4335 | } | } |
4336 | else if (common->nltype == NLTYPE_FIXED) | else if (common->nltype == NLTYPE_FIXED) |
4337 | { | { |
4338 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
4339 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
4340 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); |
4341 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
4342 | } | } |
4343 | else | else |
4344 | { | { |
# | Line 3322 switch(type) | Line 4347 switch(type) |
4347 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4348 | 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); |
4349 | jump[2] = JUMP(SLJIT_C_GREATER); | jump[2] = JUMP(SLJIT_C_GREATER); |
4350 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_LESS)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_LESS)); |
4351 | /* Equal. */ | /* Equal. */ |
4352 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
4353 | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); |
4354 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4355 | ||
4356 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4357 | if (common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANYCRLF) |
4358 | { | { |
4359 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
4360 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); |
4361 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); |
4362 | } | } |
4363 | else | else |
4364 | { | { |
4365 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); |
4366 | read_char(common); | read_char(common); |
4367 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); |
4368 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
4369 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); |
4370 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); |
4371 | } | } |
4372 | JUMPHERE(jump[2]); | JUMPHERE(jump[2]); |
4373 | JUMPHERE(jump[3]); | JUMPHERE(jump[3]); |
4374 | } | } |
4375 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4376 | check_partial(common); | check_partial(common, FALSE); |
4377 | return cc; | return cc; |
4378 | ||
4379 | case OP_EOD: | case OP_EOD: |
4380 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); |
4381 | check_partial(common); | check_partial(common, FALSE); |
4382 | return cc; | return cc; |
4383 | ||
4384 | case OP_CIRC: | case OP_CIRC: |
4385 | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); |
4386 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); |
4387 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); |
4388 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); |
4389 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4390 | return cc; | return cc; |
4391 | ||
4392 | case OP_CIRCM: | case OP_CIRCM: |
# | Line 3369 switch(type) | Line 4394 switch(type) |
4394 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); |
4395 | jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); | jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); |
4396 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); |
4397 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4398 | jump[0] = JUMP(SLJIT_JUMP); | jump[0] = JUMP(SLJIT_JUMP); |
4399 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4400 | ||
4401 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
4402 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4403 | { | { |
4404 | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4405 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); |
4406 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
4407 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
4408 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUA |