Parent Directory
|
Revision Log
|
Patch
revision 850 by zherczeg, Wed Jan 4 17:29:11 2012 UTC | revision 1275 by zherczeg, Sun Mar 10 05:32:10 2013 UTC | |
---|---|---|
# | Line 6 | Line 6 |
6 | and semantics are as close as possible to those of the Perl 5 language. | and semantics are as close as possible to those of the Perl 5 language. |
7 | ||
8 | Written by Philip Hazel | Written by Philip Hazel |
9 | Copyright (c) 1997-2012 University of Cambridge | Copyright (c) 1997-2013 University of Cambridge |
10 | ||
11 | The machine code generator part (this module) was written by Zoltan Herczeg | The machine code generator part (this module) was written by Zoltan Herczeg |
12 | Copyright (c) 2010-2012 | Copyright (c) 2010-2013 |
13 | ||
14 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
15 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
# | Line 46 POSSIBILITY OF SUCH DAMAGE. | Line 46 POSSIBILITY OF SUCH DAMAGE. |
46 | ||
47 | #include "pcre_internal.h" | #include "pcre_internal.h" |
48 | ||
49 | #ifdef SUPPORT_JIT | #if defined SUPPORT_JIT |
50 | ||
51 | /* All-in-one: Since we use the JIT compiler only from here, | /* All-in-one: Since we use the JIT compiler only from here, |
52 | we just include it. This way we don't need to touch the build | we just include it. This way we don't need to touch the build |
# | Line 65 system files. */ | Line 65 system files. */ |
65 | #error Unsupported architecture | #error Unsupported architecture |
66 | #endif | #endif |
67 | ||
68 | /* Allocate memory on the stack. Fast, but limited size. */ | /* Defines for debugging purposes. */ |
#define LOCAL_SPACE_SIZE 32768 | ||
69 | ||
70 | /* 1 - Use unoptimized capturing brackets. | |
71 | 2 - Enable capture_last_ptr (includes option 1). */ | |
72 | /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */ | |
73 | ||
74 | /* 1 - Always have a control head. */ | |
75 | /* #define DEBUG_FORCE_CONTROL_HEAD 1 */ | |
76 | ||
77 | /* Allocate memory for the regex stack on the real machine stack. | |
78 | Fast, but limited size. */ | |
79 | #define MACHINE_STACK_SIZE 32768 | |
80 | ||
81 | /* Growth rate for stack allocated by the OS. Should be the multiply | |
82 | of page size. */ | |
83 | #define STACK_GROWTH_RATE 8192 | #define STACK_GROWTH_RATE 8192 |
84 | ||
85 | /* 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 94 The code generator follows the recursive |
94 | expressions. The basic blocks of regular expressions are condition checkers | expressions. The basic blocks of regular expressions are condition checkers |
95 | whose execute different commands depending on the result of the condition check. | whose execute different commands depending on the result of the condition check. |
96 | The relationship between the operators can be horizontal (concatenation) and | The relationship between the operators can be horizontal (concatenation) and |
97 | vertical (sub-expression) (See struct fallback_common for more details). | vertical (sub-expression) (See struct backtrack_common for more details). |
98 | ||
99 | 'ab' - 'a' and 'b' regexps are concatenated | 'ab' - 'a' and 'b' regexps are concatenated |
100 | 'a+' - 'a' is the sub-expression of the '+' operator | 'a+' - 'a' is the sub-expression of the '+' operator |
101 | ||
102 | The condition checkers are boolean (true/false) checkers. Machine code is generated | The condition checkers are boolean (true/false) checkers. Machine code is generated |
103 | 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. |
104 | 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 |
105 | 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 |
106 | branches on the hot path. | branches on the matching path. |
107 | ||
108 | Greedy star operator (*) : | Greedy star operator (*) : |
109 | Hot path: match happens. | Matching path: match happens. |
110 | Fallback path: match failed. | Backtrack path: match failed. |
111 | Non-greedy star operator (*?) : | Non-greedy star operator (*?) : |
112 | Hot path: no need to perform a match. | Matching path: no need to perform a match. |
113 | Fallback path: match is required. | Backtrack path: match is required. |
114 | ||
115 | The following example shows how the code generated for a capturing bracket | The following example shows how the code generated for a capturing bracket |
116 | 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 120 we have the following regular expression |
120 | ||
121 | The generated code will be the following: | The generated code will be the following: |
122 | ||
123 | A hot path | A matching path |
124 | '(' hot path (pushing arguments to the stack) | '(' matching path (pushing arguments to the stack) |
125 | B hot path | B matching path |
126 | ')' hot path (pushing arguments to the stack) | ')' matching path (pushing arguments to the stack) |
127 | D hot path | D matching path |
128 | return with successful match | return with successful match |
129 | ||
130 | D fallback path | D backtrack path |
131 | ')' 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") |
132 | B fallback path | B backtrack path |
133 | C expected path | C expected path |
134 | jump to D hot path | jump to D matching path |
135 | C fallback path | C backtrack path |
136 | A fallback path | A backtrack path |
137 | ||
138 | 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 |
139 | 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 |
140 | to the current fallback code path. The fallback code path must check | to the current backtrack code path. The backtrack path must check |
141 | 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 |
142 | 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 |
143 | frame and continue the execution on the fallback code paths. | frame and continue the execution on the backtrack code paths. |
144 | */ | */ |
145 | ||
146 | /* | /* |
147 | Saved stack frames: | Saved stack frames: |
148 | ||
149 | Atomic blocks and asserts require reloading the values of local variables | Atomic blocks and asserts require reloading the values of private data |
150 | when the fallback mechanism performed. Because of OP_RECURSE, the locals | when the backtrack mechanism performed. Because of OP_RECURSE, the data |
151 | 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 |
152 | mechanism. | mechanism. |
153 | ||
154 | 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: |
155 | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] |
156 | ||
157 | 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. |
158 | */ | */ |
159 | ||
160 | typedef struct jit_arguments { | typedef struct jit_arguments { |
# | Line 152 typedef struct jit_arguments { | Line 164 typedef struct jit_arguments { |
164 | const pcre_uchar *begin; | const pcre_uchar *begin; |
165 | const pcre_uchar *end; | const pcre_uchar *end; |
166 | int *offsets; | int *offsets; |
167 | pcre_uchar *ptr; | pcre_uchar *uchar_ptr; |
168 | pcre_uchar *mark_ptr; | |
169 | void *callout_data; | |
170 | /* Everything else after. */ | /* Everything else after. */ |
171 | int offsetcount; | int real_offset_count; |
172 | int calllimit; | int offset_count; |
173 | int call_limit; | |
174 | pcre_uint8 notbol; | pcre_uint8 notbol; |
175 | pcre_uint8 noteol; | pcre_uint8 noteol; |
176 | pcre_uint8 notempty; | pcre_uint8 notempty; |
177 | pcre_uint8 notempty_atstart; | pcre_uint8 notempty_atstart; |
178 | } jit_arguments; | } jit_arguments; |
179 | ||
180 | typedef struct executable_function { | typedef struct executable_functions { |
181 | void *executable_func; | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; |
182 | pcre_jit_callback callback; | PUBL(jit_callback) callback; |
183 | void *userdata; | void *userdata; |
184 | sljit_uw executable_size; | pcre_uint32 top_bracket; |
185 | } executable_function; | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; |
186 | } executable_functions; | |
187 | ||
188 | typedef struct jump_list { | typedef struct jump_list { |
189 | struct sljit_jump *jump; | struct sljit_jump *jump; |
190 | struct jump_list *next; | struct jump_list *next; |
191 | } jump_list; | } jump_list; |
192 | ||
enum stub_types { stack_alloc }; | ||
193 | typedef struct stub_list { | typedef struct stub_list { |
enum stub_types type; | ||
int data; | ||
194 | struct sljit_jump *start; | struct sljit_jump *start; |
195 | struct sljit_label *leave; | struct sljit_label *quit; |
196 | struct stub_list *next; | struct stub_list *next; |
197 | } stub_list; | } stub_list; |
198 | ||
199 | enum frame_types { | |
200 | no_frame = -1, | |
201 | no_stack = -2 | |
202 | }; | |
203 | ||
204 | enum control_types { | |
205 | type_commit = 0, | |
206 | type_prune = 1, | |
207 | type_skip = 2 | |
208 | }; | |
209 | ||
210 | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); |
211 | ||
212 | /* The following structure is the key data type for the recursive | /* The following structure is the key data type for the recursive |
213 | code generator. It is allocated by compile_hotpath, and contains | code generator. It is allocated by compile_matchingpath, and contains |
214 | the aguments for compile_fallbackpath. Must be the first member | the aguments for compile_backtrackingpath. Must be the first member |
215 | of its descendants. */ | of its descendants. */ |
216 | typedef struct fallback_common { | typedef struct backtrack_common { |
217 | /* Concatenation stack. */ | /* Concatenation stack. */ |
218 | struct fallback_common *prev; | struct backtrack_common *prev; |
219 | jump_list *nextfallbacks; | jump_list *nextbacktracks; |
220 | /* Internal stack (for component operators). */ | /* Internal stack (for component operators). */ |
221 | struct fallback_common *top; | struct backtrack_common *top; |
222 | jump_list *topfallbacks; | jump_list *topbacktracks; |
223 | /* Opcode pointer. */ | /* Opcode pointer. */ |
224 | pcre_uchar *cc; | pcre_uchar *cc; |
225 | } fallback_common; | } backtrack_common; |
226 | ||
227 | typedef struct assert_fallback { | typedef struct assert_backtrack { |
228 | fallback_common common; | backtrack_common common; |
229 | jump_list *condfailed; | jump_list *condfailed; |
230 | /* Less than 0 (-1) if a frame is not needed. */ | /* Less than 0 (-1) if a frame is not needed. */ |
231 | int framesize; | int framesize; |
232 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
233 | int localptr; | int private_data_ptr; |
234 | /* For iterators. */ | /* For iterators. */ |
235 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
236 | } assert_fallback; | } assert_backtrack; |
237 | ||
238 | typedef struct bracket_fallback { | typedef struct bracket_backtrack { |
239 | fallback_common common; | backtrack_common common; |
240 | /* Where to coninue if an alternative is successfully matched. */ | /* Where to coninue if an alternative is successfully matched. */ |
241 | struct sljit_label *althotpath; | struct sljit_label *alternative_matchingpath; |
242 | /* For rmin and rmax iterators. */ | /* For rmin and rmax iterators. */ |
243 | struct sljit_label *recursivehotpath; | struct sljit_label *recursive_matchingpath; |
244 | /* For greedy ? operator. */ | /* For greedy ? operator. */ |
245 | struct sljit_label *zerohotpath; | struct sljit_label *zero_matchingpath; |
246 | /* Contains the branches of a failed condition. */ | /* Contains the branches of a failed condition. */ |
247 | union { | union { |
248 | /* Both for OP_COND, OP_SCOND. */ | /* Both for OP_COND, OP_SCOND. */ |
249 | jump_list *condfailed; | jump_list *condfailed; |
250 | assert_fallback *assert; | assert_backtrack *assert; |
251 | /* For OP_ONCE. -1 if not needed. */ | /* For OP_ONCE. -1 if not needed. */ |
252 | int framesize; | int framesize; |
253 | } u; | } u; |
254 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
255 | int localptr; | int private_data_ptr; |
256 | } bracket_fallback; | } bracket_backtrack; |
257 | ||
258 | typedef struct bracketpos_fallback { | typedef struct bracketpos_backtrack { |
259 | fallback_common common; | backtrack_common common; |
260 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
261 | int localptr; | int private_data_ptr; |
262 | /* Reverting stack is needed. */ | /* Reverting stack is needed. */ |
263 | int framesize; | int framesize; |
264 | /* Allocated stack size. */ | /* Allocated stack size. */ |
265 | int stacksize; | int stacksize; |
266 | } bracketpos_fallback; | } bracketpos_backtrack; |
267 | ||
268 | typedef struct braminzero_fallback { | typedef struct braminzero_backtrack { |
269 | fallback_common common; | backtrack_common common; |
270 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
271 | } braminzero_fallback; | } braminzero_backtrack; |
272 | ||
273 | typedef struct iterator_fallback { | typedef struct iterator_backtrack { |
274 | fallback_common common; | backtrack_common common; |
275 | /* Next iteration. */ | /* Next iteration. */ |
276 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
277 | } iterator_fallback; | } iterator_backtrack; |
278 | ||
279 | typedef struct recurse_entry { | typedef struct recurse_entry { |
280 | struct recurse_entry *next; | struct recurse_entry *next; |
# | Line 263 typedef struct recurse_entry { | Line 286 typedef struct recurse_entry { |
286 | int start; | int start; |
287 | } recurse_entry; | } recurse_entry; |
288 | ||
289 | typedef struct recurse_fallback { | typedef struct recurse_backtrack { |
290 | fallback_common common; | backtrack_common common; |
291 | } recurse_fallback; | BOOL inlined_pattern; |
292 | } recurse_backtrack; | |
293 | ||
294 | #define MAX_RANGE_SIZE 6 | |
295 | ||
296 | typedef struct compiler_common { | typedef struct compiler_common { |
297 | /* The sljit ceneric compiler. */ | |
298 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
299 | /* First byte code. */ | |
300 | pcre_uchar *start; | pcre_uchar *start; |
301 | int localsize; | /* Maps private data offset to each opcode. */ |
302 | int *localptrs; | int *private_data_ptrs; |
303 | /* Tells whether the capturing bracket is optimized. */ | |
304 | pcre_uint8 *optimized_cbracket; | |
305 | /* Starting offset of private data for capturing brackets. */ | |
306 | int cbra_ptr; | |
307 | /* Output vector starting point. Must be divisible by 2. */ | |
308 | int ovector_start; | |
309 | /* Last known position of the requested byte. */ | |
310 | int req_char_ptr; | |
311 | /* Head of the last recursion. */ | |
312 | int recursive_head_ptr; | |
313 | /* First inspected character for partial matching. */ | |
314 | int start_used_ptr; | |
315 | /* Starting pointer for partial soft matches. */ | |
316 | int hit_start; | |
317 | /* End pointer of the first line. */ | |
318 | int first_line_end; | |
319 | /* Points to the marked string. */ | |
320 | int mark_ptr; | |
321 | /* Recursive control verb management chain. */ | |
322 | int control_head_ptr; | |
323 | /* Points to the last matched capture block index. */ | |
324 | int capture_last_ptr; | |
325 | /* Points to the starting position of the current match. */ | |
326 | int start_ptr; | |
327 | ||
328 | /* Flipped and lower case tables. */ | |
329 | const pcre_uint8 *fcc; | const pcre_uint8 *fcc; |
330 | sljit_w lcc; | sljit_sw lcc; |
331 | int cbraptr; | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ |
332 | int mode; | |
333 | /* \K is in the pattern. */ | |
334 | BOOL has_set_som; | |
335 | /* Needs to know the start position anytime. */ | |
336 | BOOL needs_start_ptr; | |
337 | /* Currently in recurse or assert. */ | |
338 | BOOL local_exit; | |
339 | /* Newline control. */ | |
340 | int nltype; | int nltype; |
341 | int newline; | int newline; |
342 | int bsr_nltype; | int bsr_nltype; |
343 | /* Dollar endonly. */ | |
344 | int endonly; | int endonly; |
345 | sljit_w ctypes; | /* Tables. */ |
346 | sljit_sw ctypes; | |
347 | int digits[2 + MAX_RANGE_SIZE]; | |
348 | /* Named capturing brackets. */ | |
349 | sljit_uw name_table; | sljit_uw name_table; |
350 | sljit_w name_count; | sljit_sw name_count; |
351 | sljit_w name_entry_size; | sljit_sw name_entry_size; |
352 | struct sljit_label *acceptlabel; | |
353 | /* Labels and jump lists. */ | |
354 | struct sljit_label *partialmatchlabel; | |
355 | struct sljit_label *quit_label; | |
356 | struct sljit_label *forced_quit_label; | |
357 | struct sljit_label *accept_label; | |
358 | stub_list *stubs; | stub_list *stubs; |
359 | recurse_entry *entries; | recurse_entry *entries; |
360 | recurse_entry *currententry; | recurse_entry *currententry; |
361 | jump_list *partialmatch; | |
362 | jump_list *quit; | |
363 | jump_list *forced_quit; | |
364 | jump_list *accept; | jump_list *accept; |
365 | jump_list *calllimit; | jump_list *calllimit; |
366 | jump_list *stackalloc; | jump_list *stackalloc; |
# | Line 297 typedef struct compiler_common { | Line 371 typedef struct compiler_common { |
371 | jump_list *vspace; | jump_list *vspace; |
372 | jump_list *casefulcmp; | jump_list *casefulcmp; |
373 | jump_list *caselesscmp; | jump_list *caselesscmp; |
374 | jump_list *reset_match; | |
375 | BOOL jscript_compat; | BOOL jscript_compat; |
376 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
377 | BOOL utf; | BOOL utf; |
378 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
379 | BOOL use_ucp; | BOOL use_ucp; |
380 | #endif | #endif |
381 | #ifndef COMPILE_PCRE32 | |
382 | jump_list *utfreadchar; | jump_list *utfreadchar; |
383 | #endif | |
384 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
385 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
386 | #endif | #endif |
# | Line 321 typedef struct compare_context { | Line 398 typedef struct compare_context { |
398 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
399 | int ucharptr; | int ucharptr; |
400 | union { | union { |
401 | sljit_i asint; | sljit_si asint; |
402 | sljit_uh asushort; | sljit_uh asushort; |
403 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
404 | sljit_ub asbyte; | sljit_ub asbyte; |
405 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
406 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
407 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
408 | #endif | #elif defined COMPILE_PCRE32 |
409 | sljit_ui asuchars[1]; | |
410 | #endif | #endif |
411 | } c; | } c; |
412 | union { | union { |
413 | sljit_i asint; | sljit_si asint; |
414 | sljit_uh asushort; | sljit_uh asushort; |
415 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
416 | sljit_ub asbyte; | sljit_ub asbyte; |
417 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
418 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
419 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
420 | #endif | #elif defined COMPILE_PCRE32 |
421 | sljit_ui asuchars[1]; | |
422 | #endif | #endif |
423 | } oc; | } oc; |
424 | #endif | #endif |
425 | } compare_context; | } compare_context; |
426 | ||
427 | enum { | /* Undefine sljit macros. */ |
428 | frame_end = 0, | #undef CMP |
frame_setstrbegin = -1 | ||
}; | ||
429 | ||
430 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
431 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) |
432 | ||
433 | #define TMP1 SLJIT_TEMPORARY_REG1 | #define TMP1 SLJIT_SCRATCH_REG1 |
434 | #define TMP2 SLJIT_TEMPORARY_REG3 | #define TMP2 SLJIT_SCRATCH_REG3 |
435 | #define TMP3 SLJIT_TEMPORARY_EREG2 | #define TMP3 SLJIT_TEMPORARY_EREG2 |
436 | #define STR_PTR SLJIT_GENERAL_REG1 | #define STR_PTR SLJIT_SAVED_REG1 |
437 | #define STR_END SLJIT_GENERAL_REG2 | #define STR_END SLJIT_SAVED_REG2 |
438 | #define STACK_TOP SLJIT_TEMPORARY_REG2 | #define STACK_TOP SLJIT_SCRATCH_REG2 |
439 | #define STACK_LIMIT SLJIT_GENERAL_REG3 | #define STACK_LIMIT SLJIT_SAVED_REG3 |
440 | #define ARGUMENTS SLJIT_GENERAL_EREG1 | #define ARGUMENTS SLJIT_SAVED_EREG1 |
441 | #define CALL_COUNT SLJIT_GENERAL_EREG2 | #define CALL_COUNT SLJIT_SAVED_EREG2 |
442 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 |
443 | ||
444 | /* Locals layout. */ | /* Local space layout. */ |
445 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
446 | #define LOCALS0 (0 * sizeof(sljit_w)) | #define LOCALS0 (0 * sizeof(sljit_sw)) |
447 | #define LOCALS1 (1 * sizeof(sljit_w)) | #define LOCALS1 (1 * sizeof(sljit_sw)) |
448 | /* Two local variables for possessive quantifiers (char1 cannot use them). */ | /* Two local variables for possessive quantifiers (char1 cannot use them). */ |
449 | #define POSSESSIVE0 (2 * sizeof(sljit_w)) | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) |
450 | #define POSSESSIVE1 (3 * sizeof(sljit_w)) | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) |
/* Head of the last recursion. */ | ||
#define RECURSIVE_HEAD (4 * sizeof(sljit_w)) | ||
451 | /* Max limit of recursions. */ | /* Max limit of recursions. */ |
452 | #define CALL_LIMIT (5 * sizeof(sljit_w)) | #define CALL_LIMIT (4 * sizeof(sljit_sw)) |
/* Last known position of the requested byte. */ | ||
#define REQ_CHAR_PTR (6 * sizeof(sljit_w)) | ||
/* End pointer of the first line. */ | ||
#define FIRSTLINE_END (7 * sizeof(sljit_w)) | ||
453 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
454 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
455 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
456 | 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. */ |
457 | #define OVECTOR_START (8 * sizeof(sljit_w)) | #define OVECTOR_START (common->ovector_start) |
458 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_sw)) |
459 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) | #define OVECTOR_PRIV(i) (common->cbra_ptr + (i) * sizeof(sljit_sw)) |
460 | #define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
461 | ||
462 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
463 | #define MOV_UCHAR SLJIT_MOV_UB | #define MOV_UCHAR SLJIT_MOV_UB |
464 | #define MOVU_UCHAR SLJIT_MOVU_UB | #define MOVU_UCHAR SLJIT_MOVU_UB |
465 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
466 | #define MOV_UCHAR SLJIT_MOV_UH | #define MOV_UCHAR SLJIT_MOV_UH |
467 | #define MOVU_UCHAR SLJIT_MOVU_UH | #define MOVU_UCHAR SLJIT_MOVU_UH |
468 | #elif defined COMPILE_PCRE32 | |
469 | #define MOV_UCHAR SLJIT_MOV_UI | |
470 | #define MOVU_UCHAR SLJIT_MOVU_UI | |
471 | #else | #else |
472 | #error Unsupported compiling mode | #error Unsupported compiling mode |
473 | #endif | #endif |
#endif | ||
474 | ||
475 | /* Shortcuts. */ | /* Shortcuts. */ |
476 | #define DEFINE_COMPILER \ | #define DEFINE_COMPILER \ |
# | Line 417 the start pointers when the end of the c | Line 487 the start pointers when the end of the c |
487 | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) |
488 | #define JUMPHERE(jump) \ | #define JUMPHERE(jump) \ |
489 | sljit_set_label((jump), sljit_emit_label(compiler)) | sljit_set_label((jump), sljit_emit_label(compiler)) |
490 | #define SET_LABEL(jump, label) \ | |
491 | sljit_set_label((jump), (label)) | |
492 | #define CMP(type, src1, src1w, src2, src2w) \ | #define CMP(type, src1, src1w, src2, src2w) \ |
493 | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) |
494 | #define CMPTO(type, src1, src1w, src2, src2w, label) \ | #define CMPTO(type, src1, src1w, src2, src2w, label) \ |
495 | 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)) |
496 | #define COND_VALUE(op, dst, dstw, type) \ | #define OP_FLAGS(op, dst, dstw, src, srcw, type) \ |
497 | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) | sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type)) |
498 | #define GET_LOCAL_BASE(dst, dstw, offset) \ | |
499 | sljit_get_local_base(compiler, (dst), (dstw), (offset)) | |
500 | ||
501 | static pcre_uchar* bracketend(pcre_uchar* cc) | static pcre_uchar* bracketend(pcre_uchar* cc) |
502 | { | { |
# | Line 435 return cc; | Line 509 return cc; |
509 | ||
510 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
511 | next_opcode | next_opcode |
512 | get_localspace | get_private_data_length |
513 | set_localptrs | set_private_data_ptrs |
514 | get_framesize | get_framesize |
515 | init_frame | init_frame |
516 | get_localsize | get_private_data_length_for_copy |
517 | copy_locals | copy_private_data |
518 | compile_hotpath | compile_matchingpath |
519 | compile_fallbackpath | compile_backtrackingpath |
520 | */ | */ |
521 | ||
522 | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) |
# | Line 463 switch(*cc) | Line 537 switch(*cc) |
537 | case OP_WORDCHAR: | case OP_WORDCHAR: |
538 | case OP_ANY: | case OP_ANY: |
539 | case OP_ALLANY: | case OP_ALLANY: |
540 | case OP_NOTPROP: | |
541 | case OP_PROP: | |
542 | case OP_ANYNL: | case OP_ANYNL: |
543 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
544 | case OP_HSPACE: | case OP_HSPACE: |
# | Line 475 switch(*cc) | Line 551 switch(*cc) |
551 | case OP_CIRCM: | case OP_CIRCM: |
552 | case OP_DOLL: | case OP_DOLL: |
553 | case OP_DOLLM: | case OP_DOLLM: |
case OP_TYPESTAR: | ||
case OP_TYPEMINSTAR: | ||
case OP_TYPEPLUS: | ||
case OP_TYPEMINPLUS: | ||
case OP_TYPEQUERY: | ||
case OP_TYPEMINQUERY: | ||
case OP_TYPEPOSSTAR: | ||
case OP_TYPEPOSPLUS: | ||
case OP_TYPEPOSQUERY: | ||
554 | case OP_CRSTAR: | case OP_CRSTAR: |
555 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
556 | case OP_CRPLUS: | case OP_CRPLUS: |
557 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
558 | case OP_CRQUERY: | case OP_CRQUERY: |
559 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
560 | case OP_CRRANGE: | |
561 | case OP_CRMINRANGE: | |
562 | case OP_CLASS: | |
563 | case OP_NCLASS: | |
564 | case OP_REF: | |
565 | case OP_REFI: | |
566 | case OP_RECURSE: | |
567 | case OP_CALLOUT: | |
568 | case OP_ALT: | |
569 | case OP_KET: | |
570 | case OP_KETRMAX: | |
571 | case OP_KETRMIN: | |
572 | case OP_KETRPOS: | |
573 | case OP_REVERSE: | |
574 | case OP_ASSERT: | |
575 | case OP_ASSERT_NOT: | |
576 | case OP_ASSERTBACK: | |
577 | case OP_ASSERTBACK_NOT: | |
578 | case OP_ONCE: | |
579 | case OP_ONCE_NC: | |
580 | case OP_BRA: | |
581 | case OP_BRAPOS: | |
582 | case OP_CBRA: | |
583 | case OP_CBRAPOS: | |
584 | case OP_COND: | |
585 | case OP_SBRA: | |
586 | case OP_SBRAPOS: | |
587 | case OP_SCBRA: | |
588 | case OP_SCBRAPOS: | |
589 | case OP_SCOND: | |
590 | case OP_CREF: | |
591 | case OP_NCREF: | |
592 | case OP_RREF: | |
593 | case OP_NRREF: | |
594 | case OP_DEF: | case OP_DEF: |
595 | case OP_BRAZERO: | case OP_BRAZERO: |
596 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
597 | case OP_BRAPOSZERO: | case OP_BRAPOSZERO: |
598 | case OP_PRUNE: | |
599 | case OP_SKIP: | |
600 | case OP_COMMIT: | |
601 | case OP_FAIL: | case OP_FAIL: |
602 | case OP_ACCEPT: | case OP_ACCEPT: |
603 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
604 | case OP_CLOSE: | |
605 | case OP_SKIPZERO: | case OP_SKIPZERO: |
606 | return cc + 1; | return cc + PRIV(OP_lengths)[*cc]; |
case OP_ANYBYTE: | ||
#ifdef SUPPORT_UTF | ||
if (common->utf) return NULL; | ||
#endif | ||
return cc + 1; | ||
607 | ||
608 | case OP_CHAR: | case OP_CHAR: |
609 | case OP_CHARI: | case OP_CHARI: |
# | Line 516 switch(*cc) | Line 615 switch(*cc) |
615 | case OP_MINPLUS: | case OP_MINPLUS: |
616 | case OP_QUERY: | case OP_QUERY: |
617 | case OP_MINQUERY: | case OP_MINQUERY: |
618 | case OP_UPTO: | |
619 | case OP_MINUPTO: | |
620 | case OP_EXACT: | |
621 | case OP_POSSTAR: | case OP_POSSTAR: |
622 | case OP_POSPLUS: | case OP_POSPLUS: |
623 | case OP_POSQUERY: | case OP_POSQUERY: |
624 | case OP_POSUPTO: | |
625 | case OP_STARI: | case OP_STARI: |
626 | case OP_MINSTARI: | case OP_MINSTARI: |
627 | case OP_PLUSI: | case OP_PLUSI: |
628 | case OP_MINPLUSI: | case OP_MINPLUSI: |
629 | case OP_QUERYI: | case OP_QUERYI: |
630 | case OP_MINQUERYI: | case OP_MINQUERYI: |
631 | case OP_UPTOI: | |
632 | case OP_MINUPTOI: | |
633 | case OP_EXACTI: | |
634 | case OP_POSSTARI: | case OP_POSSTARI: |
635 | case OP_POSPLUSI: | case OP_POSPLUSI: |
636 | case OP_POSQUERYI: | case OP_POSQUERYI: |
637 | case OP_POSUPTOI: | |
638 | case OP_NOTSTAR: | case OP_NOTSTAR: |
639 | case OP_NOTMINSTAR: | case OP_NOTMINSTAR: |
640 | case OP_NOTPLUS: | case OP_NOTPLUS: |
641 | case OP_NOTMINPLUS: | case OP_NOTMINPLUS: |
642 | case OP_NOTQUERY: | case OP_NOTQUERY: |
643 | case OP_NOTMINQUERY: | case OP_NOTMINQUERY: |
644 | case OP_NOTUPTO: | |
645 | case OP_NOTMINUPTO: | |
646 | case OP_NOTEXACT: | |
647 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
648 | case OP_NOTPOSPLUS: | case OP_NOTPOSPLUS: |
649 | case OP_NOTPOSQUERY: | case OP_NOTPOSQUERY: |
650 | case OP_NOTPOSUPTO: | |
651 | case OP_NOTSTARI: | case OP_NOTSTARI: |
652 | case OP_NOTMINSTARI: | case OP_NOTMINSTARI: |
653 | case OP_NOTPLUSI: | case OP_NOTPLUSI: |
654 | case OP_NOTMINPLUSI: | case OP_NOTMINPLUSI: |
655 | case OP_NOTQUERYI: | case OP_NOTQUERYI: |
656 | case OP_NOTMINQUERYI: | case OP_NOTMINQUERYI: |
case OP_NOTPOSSTARI: | ||
case OP_NOTPOSPLUSI: | ||
case OP_NOTPOSQUERYI: | ||
cc += 2; | ||
#ifdef SUPPORT_UTF | ||
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | ||
#endif | ||
return cc; | ||
case OP_UPTO: | ||
case OP_MINUPTO: | ||
case OP_EXACT: | ||
case OP_POSUPTO: | ||
case OP_UPTOI: | ||
case OP_MINUPTOI: | ||
case OP_EXACTI: | ||
case OP_POSUPTOI: | ||
case OP_NOTUPTO: | ||
case OP_NOTMINUPTO: | ||
case OP_NOTEXACT: | ||
case OP_NOTPOSUPTO: | ||
657 | case OP_NOTUPTOI: | case OP_NOTUPTOI: |
658 | case OP_NOTMINUPTOI: | case OP_NOTMINUPTOI: |
659 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
660 | case OP_NOTPOSSTARI: | |
661 | case OP_NOTPOSPLUSI: | |
662 | case OP_NOTPOSQUERYI: | |
663 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
664 | cc += 2 + IMM2_SIZE; | cc += PRIV(OP_lengths)[*cc]; |
665 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
666 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
667 | #endif | #endif |
668 | return cc; | return cc; |
669 | ||
670 | case OP_NOTPROP: | /* Special cases. */ |
671 | case OP_PROP: | case OP_TYPESTAR: |
672 | return cc + 1 + 2; | case OP_TYPEMINSTAR: |
673 | case OP_TYPEPLUS: | |
674 | case OP_TYPEMINPLUS: | |
675 | case OP_TYPEQUERY: | |
676 | case OP_TYPEMINQUERY: | |
677 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
678 | case OP_TYPEMINUPTO: | case OP_TYPEMINUPTO: |
679 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
680 | case OP_TYPEPOSSTAR: | |
681 | case OP_TYPEPOSPLUS: | |
682 | case OP_TYPEPOSQUERY: | |
683 | case OP_TYPEPOSUPTO: | case OP_TYPEPOSUPTO: |
684 | case OP_REF: | return cc + PRIV(OP_lengths)[*cc] - 1; |
case OP_REFI: | ||
case OP_CREF: | ||
case OP_NCREF: | ||
case OP_RREF: | ||
case OP_NRREF: | ||
case OP_CLOSE: | ||
cc += 1 + IMM2_SIZE; | ||
return cc; | ||
685 | ||
686 | case OP_CRRANGE: | case OP_ANYBYTE: |
687 | case OP_CRMINRANGE: | #ifdef SUPPORT_UTF |
688 | return cc + 1 + 2 * IMM2_SIZE; | if (common->utf) return NULL; |
689 | #endif | |
690 | case OP_CLASS: | return cc + 1; |
case OP_NCLASS: | ||
return cc + 1 + 32 / sizeof(pcre_uchar); | ||
691 | ||
692 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
693 | case OP_XCLASS: | case OP_XCLASS: |
694 | return cc + GET(cc, 1); | return cc + GET(cc, 1); |
695 | #endif | #endif |
696 | ||
697 | case OP_RECURSE: | case OP_MARK: |
698 | case OP_ASSERT: | case OP_PRUNE_ARG: |
699 | case OP_ASSERT_NOT: | return cc + 1 + 2 + cc[1]; |
case OP_ASSERTBACK: | ||
case OP_ASSERTBACK_NOT: | ||
case OP_REVERSE: | ||
case OP_ONCE: | ||
case OP_ONCE_NC: | ||
case OP_BRA: | ||
case OP_BRAPOS: | ||
case OP_COND: | ||
case OP_SBRA: | ||
case OP_SBRAPOS: | ||
case OP_SCOND: | ||
case OP_ALT: | ||
case OP_KET: | ||
case OP_KETRMAX: | ||
case OP_KETRMIN: | ||
case OP_KETRPOS: | ||
return cc + 1 + LINK_SIZE; | ||
case OP_CBRA: | ||
case OP_CBRAPOS: | ||
case OP_SCBRA: | ||
case OP_SCBRAPOS: | ||
return cc + 1 + LINK_SIZE + IMM2_SIZE; | ||
700 | ||
701 | default: | default: |
702 | return NULL; | return NULL; |
703 | } | } |
704 | } | } |
705 | ||
706 | static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | #define CASE_ITERATOR_PRIVATE_DATA_1 \ |
707 | case OP_MINSTAR: \ | |
708 | case OP_MINPLUS: \ | |
709 | case OP_QUERY: \ | |
710 | case OP_MINQUERY: \ | |
711 | case OP_MINSTARI: \ | |
712 | case OP_MINPLUSI: \ | |
713 | case OP_QUERYI: \ | |
714 | case OP_MINQUERYI: \ | |
715 | case OP_NOTMINSTAR: \ | |
716 | case OP_NOTMINPLUS: \ | |
717 | case OP_NOTQUERY: \ | |
718 | case OP_NOTMINQUERY: \ | |
719 | case OP_NOTMINSTARI: \ | |
720 | case OP_NOTMINPLUSI: \ | |
721 | case OP_NOTQUERYI: \ | |
722 | case OP_NOTMINQUERYI: | |
723 | ||
724 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
725 | case OP_STAR: \ | |
726 | case OP_PLUS: \ | |
727 | case OP_STARI: \ | |
728 | case OP_PLUSI: \ | |
729 | case OP_NOTSTAR: \ | |
730 | case OP_NOTPLUS: \ | |
731 | case OP_NOTSTARI: \ | |
732 | case OP_NOTPLUSI: | |
733 | ||
734 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
735 | case OP_UPTO: \ | |
736 | case OP_MINUPTO: \ | |
737 | case OP_UPTOI: \ | |
738 | case OP_MINUPTOI: \ | |
739 | case OP_NOTUPTO: \ | |
740 | case OP_NOTMINUPTO: \ | |
741 | case OP_NOTUPTOI: \ | |
742 | case OP_NOTMINUPTOI: | |
743 | ||
744 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
745 | case OP_TYPEMINSTAR: \ | |
746 | case OP_TYPEMINPLUS: \ | |
747 | case OP_TYPEQUERY: \ | |
748 | case OP_TYPEMINQUERY: | |
749 | ||
750 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
751 | case OP_TYPESTAR: \ | |
752 | case OP_TYPEPLUS: | |
753 | ||
754 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
755 | case OP_TYPEUPTO: \ | |
756 | case OP_TYPEMINUPTO: | |
757 | ||
758 | static int get_class_iterator_size(pcre_uchar *cc) | |
759 | { | |
760 | switch(*cc) | |
761 | { | |
762 | case OP_CRSTAR: | |
763 | case OP_CRPLUS: | |
764 | return 2; | |
765 | ||
766 | case OP_CRMINSTAR: | |
767 | case OP_CRMINPLUS: | |
768 | case OP_CRQUERY: | |
769 | case OP_CRMINQUERY: | |
770 | return 1; | |
771 | ||
772 | case OP_CRRANGE: | |
773 | case OP_CRMINRANGE: | |
774 | if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | |
775 | return 0; | |
776 | return 2; | |
777 | ||
778 | default: | |
779 | return 0; | |
780 | } | |
781 | } | |
782 | ||
783 | static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | |
784 | { | { |
785 | int localspace = 0; | int private_data_length = 0; |
786 | pcre_uchar *alternative; | pcre_uchar *alternative; |
787 | pcre_uchar *name; | |
788 | pcre_uchar *end = NULL; | |
789 | int space, size, i; | |
790 | pcre_uint32 bracketlen; | |
791 | ||
792 | /* 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. */ |
793 | while (cc < ccend) | while (cc < ccend) |
794 | { | { |
795 | space = 0; | |
796 | size = 0; | |
797 | bracketlen = 0; | |
798 | switch(*cc) | switch(*cc) |
799 | { | { |
800 | case OP_SET_SOM: | |
801 | common->has_set_som = TRUE; | |
802 | cc += 1; | |
803 | break; | |
804 | ||
805 | case OP_REF: | |
806 | case OP_REFI: | |
807 | common->optimized_cbracket[GET2(cc, 1)] = 0; | |
808 | cc += 1 + IMM2_SIZE; | |
809 | break; | |
810 | ||
811 | case OP_ASSERT: | case OP_ASSERT: |
812 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
813 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 655 while (cc < ccend) | Line 817 while (cc < ccend) |
817 | case OP_BRAPOS: | case OP_BRAPOS: |
818 | case OP_SBRA: | case OP_SBRA: |
819 | case OP_SBRAPOS: | case OP_SBRAPOS: |
820 | case OP_SCOND: | private_data_length += sizeof(sljit_sw); |
821 | localspace += sizeof(sljit_w); | bracketlen = 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; | ||
822 | break; | break; |
823 | ||
824 | case OP_CBRAPOS: | case OP_CBRAPOS: |
825 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
826 | localspace += sizeof(sljit_w); | private_data_length += sizeof(sljit_sw); |
827 | cc += 1 + LINK_SIZE + IMM2_SIZE; | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
828 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
829 | break; | break; |
830 | ||
831 | case OP_COND: | case OP_COND: |
832 | /* Might be a hidden SCOND. */ | case OP_SCOND: |
833 | alternative = cc + GET(cc, 1); | /* Only AUTO_CALLOUT can insert this opcode. We do |
834 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | not intend to support this case. */ |
835 | localspace += sizeof(sljit_w); | if (cc[1 + LINK_SIZE] == OP_CALLOUT) |
836 | return -1; | |
837 | ||
838 | if (*cc == OP_COND) | |
839 | { | |
840 | /* Might be a hidden SCOND. */ | |
841 | alternative = cc + GET(cc, 1); | |
842 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | |
843 | private_data_length += sizeof(sljit_sw); | |
844 | } | |
845 | else | |
846 | private_data_length += sizeof(sljit_sw); | |
847 | bracketlen = 1 + LINK_SIZE; | |
848 | break; | |
849 | ||
850 | case OP_CREF: | |
851 | i = GET2(cc, 1); | |
852 | common->optimized_cbracket[i] = 0; | |
853 | cc += 1 + IMM2_SIZE; | |
854 | break; | |
855 | ||
856 | case OP_NCREF: | |
857 | bracketlen = GET2(cc, 1); | |
858 | name = (pcre_uchar *)common->name_table; | |
859 | alternative = name; | |
860 | for (i = 0; i < common->name_count; i++) | |
861 | { | |
862 | if (GET2(name, 0) == bracketlen) break; | |
863 | name += common->name_entry_size; | |
864 | } | |
865 | SLJIT_ASSERT(i != common->name_count); | |
866 | ||
867 | for (i = 0; i < common->name_count; i++) | |
868 | { | |
869 | if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | |
870 | common->optimized_cbracket[GET2(alternative, 0)] = 0; | |
871 | alternative += common->name_entry_size; | |
872 | } | |
873 | bracketlen = 0; | |
874 | cc += 1 + IMM2_SIZE; | |
875 | break; | |
876 | ||
877 | case OP_BRA: | |
878 | bracketlen = 1 + LINK_SIZE; | |
879 | break; | |
880 | ||
881 | case OP_CBRA: | |
882 | case OP_SCBRA: | |
883 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
884 | break; | |
885 | ||
886 | CASE_ITERATOR_PRIVATE_DATA_1 | |
887 | space = 1; | |
888 | size = -2; | |
889 | break; | |
890 | ||
891 | CASE_ITERATOR_PRIVATE_DATA_2A | |
892 | space = 2; | |
893 | size = -2; | |
894 | break; | |
895 | ||
896 | CASE_ITERATOR_PRIVATE_DATA_2B | |
897 | space = 2; | |
898 | size = -(2 + IMM2_SIZE); | |
899 | break; | |
900 | ||
901 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
902 | space = 1; | |
903 | size = 1; | |
904 | break; | |
905 | ||
906 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
907 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
908 | space = 2; | |
909 | size = 1; | |
910 | break; | |
911 | ||
912 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
913 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
914 | space = 2; | |
915 | size = 1 + IMM2_SIZE; | |
916 | break; | |
917 | ||
918 | case OP_CLASS: | |
919 | case OP_NCLASS: | |
920 | size += 1 + 32 / sizeof(pcre_uchar); | |
921 | space = get_class_iterator_size(cc + size); | |
922 | break; | |
923 | ||
924 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
925 | case OP_XCLASS: | |
926 | size = GET(cc, 1); | |
927 | space = get_class_iterator_size(cc + size); | |
928 | break; | |
929 | #endif | |
930 | ||
931 | case OP_RECURSE: | |
932 | /* Set its value only once. */ | |
933 | if (common->recursive_head_ptr == 0) | |
934 | { | |
935 | common->recursive_head_ptr = common->ovector_start; | |
936 | common->ovector_start += sizeof(sljit_sw); | |
937 | } | |
938 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
939 | break; | break; |
940 | ||
941 | case OP_CALLOUT: | |
942 | if (common->capture_last_ptr == 0) | |
943 | { | |
944 | common->capture_last_ptr = common->ovector_start; | |
945 | common->ovector_start += sizeof(sljit_sw); | |
946 | } | |
947 | cc += 2 + 2 * LINK_SIZE; | |
948 | break; | |
949 | ||
950 | case OP_PRUNE_ARG: | |
951 | common->needs_start_ptr = TRUE; | |
952 | common->control_head_ptr = 1; | |
953 | /* Fall through. */ | |
954 | ||
955 | case OP_MARK: | |
956 | if (common->mark_ptr == 0) | |
957 | { | |
958 | common->mark_ptr = common->ovector_start; | |
959 | common->ovector_start += sizeof(sljit_sw); | |
960 | } | |
961 | cc += 1 + 2 + cc[1]; | |
962 | break; | |
963 | ||
964 | case OP_PRUNE: | |
965 | case OP_SKIP: | |
966 | common->needs_start_ptr = TRUE; | |
967 | /* Fall through. */ | |
968 | ||
969 | case OP_COMMIT: | |
970 | common->control_head_ptr = 1; | |
971 | cc += 1; | |
972 | break; | |
973 | ||
974 | default: | default: |
975 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
976 | if (cc == NULL) | if (cc == NULL) |
977 | return -1; | return -1; |
978 | break; | break; |
979 | } | } |
980 | ||
981 | if (space > 0 && cc >= end) | |
982 | private_data_length += sizeof(sljit_sw) * space; | |
983 | ||
984 | if (size != 0) | |
985 | { | |
986 | if (size < 0) | |
987 | { | |
988 | cc += -size; | |
989 | #ifdef SUPPORT_UTF | |
990 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
991 | #endif | |
992 | } | |
993 | else | |
994 | cc += size; | |
995 | } | |
996 | ||
997 | if (bracketlen != 0) | |
998 | { | |
999 | if (cc >= end) | |
1000 | { | |
1001 | end = bracketend(cc); | |
1002 | if (end[-1 - LINK_SIZE] == OP_KET) | |
1003 | end = NULL; | |
1004 | } | |
1005 | cc += bracketlen; | |
1006 | } | |
1007 | } | } |
1008 | return localspace; | return private_data_length; |
1009 | } | } |
1010 | ||
1011 | 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) |
1012 | { | { |
1013 | pcre_uchar *cc = common->start; | pcre_uchar *cc = common->start; |
1014 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1015 | pcre_uchar *end = NULL; | |
1016 | int space, size, bracketlen; | |
1017 | ||
1018 | while (cc < ccend) | while (cc < ccend) |
1019 | { | { |
1020 | space = 0; | |
1021 | size = 0; | |
1022 | bracketlen = 0; | |
1023 | switch(*cc) | switch(*cc) |
1024 | { | { |
1025 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 702 while (cc < ccend) | Line 1032 while (cc < ccend) |
1032 | case OP_SBRA: | case OP_SBRA: |
1033 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1034 | case OP_SCOND: | case OP_SCOND: |
1035 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1036 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
1037 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1038 | break; | |
1039 | ||
1040 | case OP_CBRAPOS: | |
1041 | case OP_SCBRAPOS: | |
1042 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1043 | private_data_ptr += sizeof(sljit_sw); | |
1044 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
1045 | break; | |
1046 | ||
1047 | case OP_COND: | |
1048 | /* Might be a hidden SCOND. */ | |
1049 | alternative = cc + GET(cc, 1); | |
1050 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | |
1051 | { | |
1052 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1053 | private_data_ptr += sizeof(sljit_sw); | |
1054 | } | |
1055 | bracketlen = 1 + LINK_SIZE; | |
1056 | break; | |
1057 | ||
1058 | case OP_BRA: | |
1059 | bracketlen = 1 + LINK_SIZE; | |
1060 | break; | |
1061 | ||
1062 | case OP_CBRA: | |
1063 | case OP_SCBRA: | |
1064 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
1065 | break; | break; |
1066 | ||
1067 | case OP_CBRAPOS: | CASE_ITERATOR_PRIVATE_DATA_1 |
1068 | case OP_SCBRAPOS: | space = 1; |
1069 | common->localptrs[cc - common->start] = localptr; | size = -2; |
localptr += sizeof(sljit_w); | ||
cc += 1 + LINK_SIZE + IMM2_SIZE; | ||
1070 | break; | break; |
1071 | ||
1072 | case OP_COND: | CASE_ITERATOR_PRIVATE_DATA_2A |
1073 | /* Might be a hidden SCOND. */ | space = 2; |
1074 | alternative = cc + GET(cc, 1); | size = -2; |
1075 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | break; |
1076 | { | |
1077 | common->localptrs[cc - common->start] = localptr; | CASE_ITERATOR_PRIVATE_DATA_2B |
1078 | localptr += sizeof(sljit_w); | space = 2; |
1079 | } | size = -(2 + IMM2_SIZE); |
1080 | cc += 1 + LINK_SIZE; | break; |
1081 | ||
1082 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1083 | space = 1; | |
1084 | size = 1; | |
1085 | break; | break; |
1086 | ||
1087 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1088 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
1089 | space = 2; | |
1090 | size = 1; | |
1091 | break; | |
1092 | ||
1093 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1094 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
1095 | space = 2; | |
1096 | size = 1 + IMM2_SIZE; | |
1097 | break; | |
1098 | ||
1099 | case OP_CLASS: | |
1100 | case OP_NCLASS: | |
1101 | size += 1 + 32 / sizeof(pcre_uchar); | |
1102 | space = get_class_iterator_size(cc + size); | |
1103 | break; | |
1104 | ||
1105 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1106 | case OP_XCLASS: | |
1107 | size = GET(cc, 1); | |
1108 | space = get_class_iterator_size(cc + size); | |
1109 | break; | |
1110 | #endif | |
1111 | ||
1112 | default: | default: |
1113 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1114 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1115 | break; | break; |
1116 | } | } |
1117 | ||
1118 | if (space > 0 && cc >= end) | |
1119 | { | |
1120 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1121 | private_data_ptr += sizeof(sljit_sw) * space; | |
1122 | } | |
1123 | ||
1124 | if (size != 0) | |
1125 | { | |
1126 | if (size < 0) | |
1127 | { | |
1128 | cc += -size; | |
1129 | #ifdef SUPPORT_UTF | |
1130 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1131 | #endif | |
1132 | } | |
1133 | else | |
1134 | cc += size; | |
1135 | } | |
1136 | ||
1137 | if (bracketlen > 0) | |
1138 | { | |
1139 | if (cc >= end) | |
1140 | { | |
1141 | end = bracketend(cc); | |
1142 | if (end[-1 - LINK_SIZE] == OP_KET) | |
1143 | end = NULL; | |
1144 | } | |
1145 | cc += bracketlen; | |
1146 | } | |
1147 | } | } |
1148 | } | } |
1149 | ||
1150 | /* Returns with -1 if no need for frame. */ | /* Returns with a frame_types (always < 0) if no need for frame. */ |
1151 | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) |
1152 | { | { |
1153 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE); |
1154 | int length = 0; | int length = 0; |
1155 | BOOL possessive = FALSE; | int possessive = 0; |
1156 | BOOL setsom_found = FALSE; | BOOL stack_restore = FALSE; |
1157 | BOOL setsom_found = recursive; | |
1158 | BOOL setmark_found = recursive; | |
1159 | /* The last capture is a local variable even for recursions. */ | |
1160 | BOOL capture_last_found = FALSE; | |
1161 | ||
1162 | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
1163 | { | { |
1164 | length = 3; | possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; |
1165 | possessive = TRUE; | /* This is correct regardless of common->capture_last_ptr. */ |
1166 | capture_last_found = TRUE; | |
1167 | } | } |
1168 | ||
1169 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
# | Line 753 while (cc < ccend) | Line 1172 while (cc < ccend) |
1172 | switch(*cc) | switch(*cc) |
1173 | { | { |
1174 | case OP_SET_SOM: | case OP_SET_SOM: |
1175 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
1176 | stack_restore = TRUE; | |
1177 | if (!setsom_found) | if (!setsom_found) |
1178 | { | { |
1179 | length += 2; | length += 2; |
1180 | setsom_found = TRUE; | setsom_found = TRUE; |
1181 | } | } |
1182 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | cc += 1; |
1183 | break; | |
1184 | ||
1185 | case OP_MARK: | |
1186 | case OP_PRUNE_ARG: | |
1187 | SLJIT_ASSERT(common->mark_ptr != 0); | |
1188 | stack_restore = TRUE; | |
1189 | if (!setmark_found) | |
1190 | { | |
1191 | length += 2; | |
1192 | setmark_found = TRUE; | |
1193 | } | |
1194 | cc += 1 + 2 + cc[1]; | |
1195 | break; | |
1196 | ||
1197 | case OP_RECURSE: | |
1198 | stack_restore = TRUE; | |
1199 | if (common->has_set_som && !setsom_found) | |
1200 | { | |
1201 | length += 2; | |
1202 | setsom_found = TRUE; | |
1203 | } | |
1204 | if (common->mark_ptr != 0 && !setmark_found) | |
1205 | { | |
1206 | length += 2; | |
1207 | setmark_found = TRUE; | |
1208 | } | |
1209 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1210 | { | |
1211 | length += 2; | |
1212 | capture_last_found = TRUE; | |
1213 | } | |
1214 | cc += 1 + LINK_SIZE; | |
1215 | break; | break; |
1216 | ||
1217 | case OP_CBRA: | case OP_CBRA: |
1218 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1219 | case OP_SCBRA: | case OP_SCBRA: |
1220 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1221 | stack_restore = TRUE; | |
1222 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1223 | { | |
1224 | length += 2; | |
1225 | capture_last_found = TRUE; | |
1226 | } | |
1227 | length += 3; | length += 3; |
1228 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1229 | break; | break; |
1230 | ||
1231 | default: | default: |
1232 | stack_restore = TRUE; | |
1233 | /* Fall through. */ | |
1234 | ||
1235 | case OP_NOT_WORD_BOUNDARY: | |
1236 | case OP_WORD_BOUNDARY: | |
1237 | case OP_NOT_DIGIT: | |
1238 | case OP_DIGIT: | |
1239 | case OP_NOT_WHITESPACE: | |
1240 | case OP_WHITESPACE: | |
1241 | case OP_NOT_WORDCHAR: | |
1242 | case OP_WORDCHAR: | |
1243 | case OP_ANY: | |
1244 | case OP_ALLANY: | |
1245 | case OP_ANYBYTE: | |
1246 | case OP_NOTPROP: | |
1247 | case OP_PROP: | |
1248 | case OP_ANYNL: | |
1249 | case OP_NOT_HSPACE: | |
1250 | case OP_HSPACE: | |
1251 | case OP_NOT_VSPACE: | |
1252 | case OP_VSPACE: | |
1253 | case OP_EXTUNI: | |
1254 | case OP_EODN: | |
1255 | case OP_EOD: | |
1256 | case OP_CIRC: | |
1257 | case OP_CIRCM: | |
1258 | case OP_DOLL: | |
1259 | case OP_DOLLM: | |
1260 | case OP_CHAR: | |
1261 | case OP_CHARI: | |
1262 | case OP_NOT: | |
1263 | case OP_NOTI: | |
1264 | ||
1265 | case OP_EXACT: | |
1266 | case OP_POSSTAR: | |
1267 | case OP_POSPLUS: | |
1268 | case OP_POSQUERY: | |
1269 | case OP_POSUPTO: | |
1270 | ||
1271 | case OP_EXACTI: | |
1272 | case OP_POSSTARI: | |
1273 | case OP_POSPLUSI: | |
1274 | case OP_POSQUERYI: | |
1275 | case OP_POSUPTOI: | |
1276 | ||
1277 | case OP_NOTEXACT: | |
1278 | case OP_NOTPOSSTAR: | |
1279 | case OP_NOTPOSPLUS: | |
1280 | case OP_NOTPOSQUERY: | |
1281 | case OP_NOTPOSUPTO: | |
1282 | ||
1283 | case OP_NOTEXACTI: | |
1284 | case OP_NOTPOSSTARI: | |
1285 | case OP_NOTPOSPLUSI: | |
1286 | case OP_NOTPOSQUERYI: | |
1287 | case OP_NOTPOSUPTOI: | |
1288 | ||
1289 | case OP_TYPEEXACT: | |
1290 | case OP_TYPEPOSSTAR: | |
1291 | case OP_TYPEPOSPLUS: | |
1292 | case OP_TYPEPOSQUERY: | |
1293 | case OP_TYPEPOSUPTO: | |
1294 | ||
1295 | case OP_CLASS: | |
1296 | case OP_NCLASS: | |
1297 | case OP_XCLASS: | |
1298 | ||
1299 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1300 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1301 | break; | break; |
1302 | } | } |
1303 | ||
1304 | /* Possessive quantifiers can use a special case. */ | /* Possessive quantifiers can use a special case. */ |
1305 | if (SLJIT_UNLIKELY(possessive) && length == 3) | if (SLJIT_UNLIKELY(possessive == length)) |
1306 | return -1; | return stack_restore ? no_frame : no_stack; |
1307 | ||
1308 | if (length > 0) | if (length > 0) |
1309 | return length + 1; | return length + 1; |
1310 | return -1; | return stack_restore ? no_frame : no_stack; |
1311 | } | } |
1312 | ||
1313 | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) |
1314 | { | { |
1315 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1316 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE); |
1317 | BOOL setsom_found = FALSE; | BOOL setsom_found = recursive; |
1318 | BOOL setmark_found = recursive; | |
1319 | /* The last capture is a local variable even for recursions. */ | |
1320 | BOOL capture_last_found = FALSE; | |
1321 | int offset; | int offset; |
1322 | ||
1323 | /* >= 1 + shortest item size (2) */ | /* >= 1 + shortest item size (2) */ |
1324 | SLJIT_UNUSED_ARG(stacktop); | |
1325 | SLJIT_ASSERT(stackpos >= stacktop + 2); | SLJIT_ASSERT(stackpos >= stacktop + 2); |
1326 | ||
1327 | stackpos = STACK(stackpos); | stackpos = STACK(stackpos); |
# | Line 803 while (cc < ccend) | Line 1332 while (cc < ccend) |
1332 | switch(*cc) | switch(*cc) |
1333 | { | { |
1334 | case OP_SET_SOM: | case OP_SET_SOM: |
1335 | case OP_RECURSE: | SLJIT_ASSERT(common->has_set_som); |
1336 | if (!setsom_found) | if (!setsom_found) |
1337 | { | { |
1338 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
1339 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1340 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1341 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1342 | stackpos += (int)sizeof(sljit_sw); | |
1343 | setsom_found = TRUE; | |
1344 | } | |
1345 | cc += 1; | |
1346 | break; | |
1347 | ||
1348 | case OP_MARK: | |
1349 | case OP_PRUNE_ARG: | |
1350 | SLJIT_ASSERT(common->mark_ptr != 0); | |
1351 | if (!setmark_found) | |
1352 | { | |
1353 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
1354 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); | |
1355 | stackpos += (int)sizeof(sljit_sw); | |
1356 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1357 | stackpos += (int)sizeof(sljit_sw); | |
1358 | setmark_found = TRUE; | |
1359 | } | |
1360 | cc += 1 + 2 + cc[1]; | |
1361 | break; | |
1362 | ||
1363 | case OP_RECURSE: | |
1364 | if (common->has_set_som && !setsom_found) | |
1365 | { | |
1366 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | |
1367 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); | |
1368 | stackpos += (int)sizeof(sljit_sw); | |
1369 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1370 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1371 | setsom_found = TRUE; | setsom_found = TRUE; |
1372 | } | } |
1373 | cc += (*cc == OP_SET_SOM) ? 1 : 1 + LINK_SIZE; | if (common->mark_ptr != 0 && !setmark_found) |
1374 | { | |
1375 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | |
1376 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); | |
1377 | stackpos += (int)sizeof(sljit_sw); | |
1378 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1379 | stackpos += (int)sizeof(sljit_sw); | |
1380 | setmark_found = TRUE; | |
1381 | } | |
1382 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1383 | { | |
1384 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | |
1385 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1386 | stackpos += (int)sizeof(sljit_sw); | |
1387 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1388 | stackpos += (int)sizeof(sljit_sw); | |
1389 | capture_last_found = TRUE; | |
1390 | } | |
1391 | cc += 1 + LINK_SIZE; | |
1392 | break; | break; |
1393 | ||
1394 | case OP_CBRA: | case OP_CBRA: |
1395 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1396 | case OP_SCBRA: | case OP_SCBRA: |
1397 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1398 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1399 | { | |
1400 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | |
1401 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1402 | stackpos += (int)sizeof(sljit_sw); | |
1403 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1404 | stackpos += (int)sizeof(sljit_sw); | |
1405 | capture_last_found = TRUE; | |
1406 | } | |
1407 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
1408 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
1409 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1410 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
1411 | 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)); |
1412 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1413 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1414 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
1415 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1416 | ||
1417 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1418 | break; | break; |
# | Line 839 while (cc < ccend) | Line 1423 while (cc < ccend) |
1423 | break; | break; |
1424 | } | } |
1425 | ||
1426 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0); |
1427 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1428 | } | } |
1429 | ||
1430 | 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) |
1431 | { | { |
1432 | int localsize = 2; | int private_data_length = common->control_head_ptr ? 3 : 2; |
1433 | int size; | |
1434 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1435 | /* Calculate the sum of the local variables. */ | /* Calculate the sum of the private machine words. */ |
1436 | while (cc < ccend) | while (cc < ccend) |
1437 | { | { |
1438 | size = 0; | |
1439 | switch(*cc) | switch(*cc) |
1440 | { | { |
1441 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 862 while (cc < ccend) | Line 1448 while (cc < ccend) |
1448 | case OP_SBRA: | case OP_SBRA: |
1449 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1450 | case OP_SCOND: | case OP_SCOND: |
1451 | localsize++; | private_data_length++; |
1452 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1453 | break; | break; |
1454 | ||
1455 | case OP_CBRA: | case OP_CBRA: |
1456 | case OP_SCBRA: | case OP_SCBRA: |
1457 | localsize++; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1458 | private_data_length++; | |
1459 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1460 | break; | break; |
1461 | ||
1462 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1463 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1464 | localsize += 2; | private_data_length += 2; |
1465 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1466 | break; | break; |
1467 | ||
# | Line 882 while (cc < ccend) | Line 1469 while (cc < ccend) |
1469 | /* Might be a hidden SCOND. */ | /* Might be a hidden SCOND. */ |
1470 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1471 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1472 | localsize++; | private_data_length++; |
1473 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1474 | break; | break; |
1475 | ||
1476 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1477 | if (PRIVATE_DATA(cc)) | |
1478 | private_data_length++; | |
1479 | cc += 2; | |
1480 | #ifdef SUPPORT_UTF | |
1481 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1482 | #endif | |
1483 | break; | |
1484 | ||
1485 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1486 | if (PRIVATE_DATA(cc)) | |
1487 | private_data_length += 2; | |
1488 | cc += 2; | |
1489 | #ifdef SUPPORT_UTF | |
1490 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1491 | #endif | |
1492 | break; | |
1493 | ||
1494 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1495 | if (PRIVATE_DATA(cc)) | |
1496 | private_data_length += 2; | |
1497 | cc += 2 + IMM2_SIZE; | |
1498 | #ifdef SUPPORT_UTF | |
1499 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1500 | #endif | |
1501 | break; | |
1502 | ||
1503 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1504 | if (PRIVATE_DATA(cc)) | |
1505 | private_data_length++; | |
1506 | cc += 1; | |
1507 | break; | |
1508 | ||
1509 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1510 | if (PRIVATE_DATA(cc)) | |
1511 | private_data_length += 2; | |
1512 | cc += 1; | |
1513 | break; | |
1514 | ||
1515 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1516 | if (PRIVATE_DATA(cc)) | |
1517 | private_data_length += 2; | |
1518 | cc += 1 + IMM2_SIZE; | |
1519 | break; | |
1520 | ||
1521 | case OP_CLASS: | |
1522 | case OP_NCLASS: | |
1523 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1524 | case OP_XCLASS: | |
1525 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1526 | #else | |
1527 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1528 | #endif | |
1529 | if (PRIVATE_DATA(cc)) | |
1530 | private_data_length += get_class_iterator_size(cc + size); | |
1531 | cc += size; | |
1532 | break; | |
1533 | ||
1534 | default: | default: |
1535 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1536 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 893 while (cc < ccend) | Line 1538 while (cc < ccend) |
1538 | } | } |
1539 | } | } |
1540 | SLJIT_ASSERT(cc == ccend); | SLJIT_ASSERT(cc == ccend); |
1541 | return localsize; | return private_data_length; |
1542 | } | } |
1543 | ||
1544 | 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, |
1545 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop) |
1546 | { | { |
1547 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1548 | int srcw[2]; | int srcw[2]; |
1549 | int count; | int count, size; |
1550 | BOOL tmp1next = TRUE; | BOOL tmp1next = TRUE; |
1551 | BOOL tmp1empty = TRUE; | BOOL tmp1empty = TRUE; |
1552 | BOOL tmp2empty = TRUE; | BOOL tmp2empty = TRUE; |
# | Line 918 stacktop = STACK(stacktop - 1); | Line 1563 stacktop = STACK(stacktop - 1); |
1563 | ||
1564 | if (!save) | if (!save) |
1565 | { | { |
1566 | stackptr += sizeof(sljit_w); | stackptr += (common->control_head_ptr ? 2 : 1) * sizeof(sljit_sw); |
1567 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1568 | { | { |
1569 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1570 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1571 | tmp1empty = FALSE; | tmp1empty = FALSE; |
1572 | } | } |
1573 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1574 | { | { |
1575 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1576 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1577 | tmp2empty = FALSE; | tmp2empty = FALSE; |
1578 | } | } |
1579 | /* The tmp1next must be TRUE in either way. */ | /* The tmp1next must be TRUE in either way. */ |
1580 | } | } |
1581 | ||
1582 | while (status != end) | do |
1583 | { | { |
1584 | count = 0; | count = 0; |
1585 | switch(status) | switch(status) |
1586 | { | { |
1587 | case start: | case start: |
1588 | SLJIT_ASSERT(save); | SLJIT_ASSERT(save && common->recursive_head_ptr != 0); |
1589 | count = 1; | count = 1; |
1590 | srcw[0] = RECURSIVE_HEAD; | srcw[0] = common->recursive_head_ptr; |
1591 | if (common->control_head_ptr != 0) | |
1592 | { | |
1593 | count = 2; | |
1594 | srcw[1] = common->control_head_ptr; | |
1595 | } | |
1596 | status = loop; | status = loop; |
1597 | break; | break; |
1598 | ||
# | Line 966 while (status != end) | Line 1616 while (status != end) |
1616 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1617 | case OP_SCOND: | case OP_SCOND: |
1618 | count = 1; | count = 1; |
1619 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1620 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1621 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1622 | break; | break; |
1623 | ||
1624 | case OP_CBRA: | case OP_CBRA: |
1625 | case OP_SCBRA: | case OP_SCBRA: |
1626 | count = 1; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1627 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | { |
1628 | count = 1; | |
1629 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | |
1630 | } | |
1631 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1632 | break; | break; |
1633 | ||
1634 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1635 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1636 | count = 2; | count = 2; |
1637 | srcw[0] = PRIVATE_DATA(cc); | |
1638 | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
1639 | srcw[0] = PRIV_DATA(cc); | SLJIT_ASSERT(srcw[0] != 0 && srcw[1] != 0); |
SLJIT_ASSERT(srcw[0] != 0); | ||
1640 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1641 | break; | break; |
1642 | ||
# | Line 993 while (status != end) | Line 1646 while (status != end) |
1646 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1647 | { | { |
1648 | count = 1; | count = 1; |
1649 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1650 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1651 | } | } |
1652 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1653 | break; | break; |
1654 | ||
1655 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1656 | if (PRIVATE_DATA(cc)) | |
1657 | { | |
1658 | count = 1; | |
1659 | srcw[0] = PRIVATE_DATA(cc); | |
1660 | } | |
1661 | cc += 2; | |
1662 | #ifdef SUPPORT_UTF | |
1663 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1664 | #endif | |
1665 | break; | |
1666 | ||
1667 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1668 | if (PRIVATE_DATA(cc)) | |
1669 | { | |
1670 | count = 2; | |
1671 | srcw[0] = PRIVATE_DATA(cc); | |
1672 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1673 | } | |
1674 | cc += 2; | |
1675 | #ifdef SUPPORT_UTF | |
1676 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1677 | #endif | |
1678 | break; | |
1679 | ||
1680 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1681 | if (PRIVATE_DATA(cc)) | |
1682 | { | |
1683 | count = 2; | |
1684 | srcw[0] = PRIVATE_DATA(cc); | |
1685 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1686 | } | |
1687 | cc += 2 + IMM2_SIZE; | |
1688 | #ifdef SUPPORT_UTF | |
1689 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1690 | #endif | |
1691 | break; | |
1692 | ||
1693 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1694 | if (PRIVATE_DATA(cc)) | |
1695 | { | |
1696 | count = 1; | |
1697 | srcw[0] = PRIVATE_DATA(cc); | |
1698 | } | |
1699 | cc += 1; | |
1700 | break; | |
1701 | ||
1702 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1703 | if (PRIVATE_DATA(cc)) | |
1704 | { | |
1705 | count = 2; | |
1706 | srcw[0] = PRIVATE_DATA(cc); | |
1707 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1708 | } | |
1709 | cc += 1; | |
1710 | break; | |
1711 | ||
1712 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1713 | if (PRIVATE_DATA(cc)) | |
1714 | { | |
1715 | count = 2; | |
1716 | srcw[0] = PRIVATE_DATA(cc); | |
1717 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1718 | } | |
1719 | cc += 1 + IMM2_SIZE; | |
1720 | break; | |
1721 | ||
1722 | case OP_CLASS: | |
1723 | case OP_NCLASS: | |
1724 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1725 | case OP_XCLASS: | |
1726 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1727 | #else | |
1728 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1729 | #endif | |
1730 | if (PRIVATE_DATA(cc)) | |
1731 | switch(get_class_iterator_size(cc + size)) | |
1732 | { | |
1733 | case 1: | |
1734 | count = 1; | |
1735 | srcw[0] = PRIVATE_DATA(cc); | |
1736 | break; | |
1737 | ||
1738 | case 2: | |
1739 | count = 2; | |
1740 | srcw[0] = PRIVATE_DATA(cc); | |
1741 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1742 | break; | |
1743 | ||
1744 | default: | |
1745 | SLJIT_ASSERT_STOP(); | |
1746 | break; | |
1747 | } | |
1748 | cc += size; | |
1749 | break; | |
1750 | ||
1751 | default: | default: |
1752 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1753 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1021 while (status != end) | Line 1770 while (status != end) |
1770 | if (!tmp1empty) | if (!tmp1empty) |
1771 | { | { |
1772 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1773 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1774 | } | } |
1775 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1776 | tmp1empty = FALSE; | tmp1empty = FALSE; |
# | Line 1032 while (status != end) | Line 1781 while (status != end) |
1781 | if (!tmp2empty) | if (!tmp2empty) |
1782 | { | { |
1783 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1784 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1785 | } | } |
1786 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1787 | tmp2empty = FALSE; | tmp2empty = FALSE; |
# | Line 1049 while (status != end) | Line 1798 while (status != end) |
1798 | if (!tmp1empty) | if (!tmp1empty) |
1799 | { | { |
1800 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1801 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1802 | } | } |
1803 | tmp1next = FALSE; | tmp1next = FALSE; |
1804 | } | } |
# | Line 1061 while (status != end) | Line 1810 while (status != end) |
1810 | if (!tmp2empty) | if (!tmp2empty) |
1811 | { | { |
1812 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1813 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1814 | } | } |
1815 | tmp1next = TRUE; | tmp1next = TRUE; |
1816 | } | } |
1817 | } | } |
1818 | } | } |
1819 | } | } |
1820 | while (status != end); | |
1821 | ||
1822 | if (save) | if (save) |
1823 | { | { |
# | Line 1076 if (save) | Line 1826 if (save) |
1826 | if (!tmp1empty) | if (!tmp1empty) |
1827 | { | { |
1828 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1829 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1830 | } | } |
1831 | if (!tmp2empty) | if (!tmp2empty) |
1832 | { | { |
1833 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1834 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1835 | } | } |
1836 | } | } |
1837 | else | else |
# | Line 1089 if (save) | Line 1839 if (save) |
1839 | if (!tmp2empty) | if (!tmp2empty) |
1840 | { | { |
1841 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1842 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1843 | } | } |
1844 | if (!tmp1empty) | if (!tmp1empty) |
1845 | { | { |
1846 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1847 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1848 | } | } |
1849 | } | } |
1850 | } | } |
1851 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1852 | } | } |
1853 | ||
1854 | static SLJIT_INLINE BOOL ispowerof2(unsigned int value) | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
1855 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | |
1856 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | |
1857 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1858 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1859 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1860 | ||
1861 | static SLJIT_INLINE BOOL is_powerof2(unsigned int value) | |
1862 | { | { |
1863 | return (value & (value - 1)) == 0; | return (value & (value - 1)) == 0; |
1864 | } | } |
# | Line 1111 static SLJIT_INLINE void set_jumps(jump_ | Line 1868 static SLJIT_INLINE void set_jumps(jump_ |
1868 | while (list) | while (list) |
1869 | { | { |
1870 | /* sljit_set_label is clever enough to do nothing | /* sljit_set_label is clever enough to do nothing |
1871 | if either the jump or the label is NULL */ | if either the jump or the label is NULL. */ |
1872 | sljit_set_label(list->jump, label); | SET_LABEL(list->jump, label); |
1873 | list = list->next; | list = list->next; |
1874 | } | } |
1875 | } | } |
# | Line 1128 if (list_item) | Line 1885 if (list_item) |
1885 | } | } |
1886 | } | } |
1887 | ||
1888 | static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start) | static void add_stub(compiler_common *common, struct sljit_jump *start) |
1889 | { | { |
1890 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1891 | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); |
1892 | ||
1893 | if (list_item) | if (list_item) |
1894 | { | { |
list_item->type = type; | ||
list_item->data = data; | ||
1895 | list_item->start = start; | list_item->start = start; |
1896 | list_item->leave = LABEL(); | list_item->quit = LABEL(); |
1897 | list_item->next = common->stubs; | list_item->next = common->stubs; |
1898 | common->stubs = list_item; | common->stubs = list_item; |
1899 | } | } |
# | Line 1152 stub_list* list_item = common->stubs; | Line 1907 stub_list* list_item = common->stubs; |
1907 | while (list_item) | while (list_item) |
1908 | { | { |
1909 | JUMPHERE(list_item->start); | JUMPHERE(list_item->start); |
1910 | switch(list_item->type) | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
1911 | { | JUMPTO(SLJIT_JUMP, list_item->quit); |
case stack_alloc: | ||
add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | ||
break; | ||
} | ||
JUMPTO(SLJIT_JUMP, list_item->leave); | ||
1912 | list_item = list_item->next; | list_item = list_item->next; |
1913 | } | } |
1914 | common->stubs = NULL; | common->stubs = NULL; |
# | Line 1177 static SLJIT_INLINE void allocate_stack( | Line 1927 static SLJIT_INLINE void allocate_stack( |
1927 | /* May destroy all locals and registers except TMP2. */ | /* May destroy all locals and registers except TMP2. */ |
1928 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1929 | ||
1930 | 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)); |
1931 | #ifdef DESTROY_REGISTERS | #ifdef DESTROY_REGISTERS |
1932 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); |
1933 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
# | Line 1185 OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); | Line 1935 OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); |
1935 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); |
1936 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); |
1937 | #endif | #endif |
1938 | add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); | add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); |
1939 | } | } |
1940 | ||
1941 | static SLJIT_INLINE void free_stack(compiler_common *common, int size) | static SLJIT_INLINE void free_stack(compiler_common *common, int size) |
1942 | { | { |
1943 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1944 | 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)); |
1945 | } | } |
1946 | ||
1947 | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) |
# | Line 1199 static SLJIT_INLINE void reset_ovector(c | Line 1949 static SLJIT_INLINE void reset_ovector(c |
1949 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1950 | struct sljit_label *loop; | struct sljit_label *loop; |
1951 | int i; | int i; |
1952 | ||
1953 | /* At this point we can freely use all temporary registers. */ | /* At this point we can freely use all temporary registers. */ |
1954 | SLJIT_ASSERT(length > 1); | |
1955 | /* TMP1 returns with begin - 1. */ | /* TMP1 returns with begin - 1. */ |
1956 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_GENERAL_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); |
1957 | if (length < 8) | |
1958 | { | |
1959 | for (i = 1; i < length; i++) | |
1960 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0); | |
1961 | } | |
1962 | else | |
1963 | { | |
1964 | GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START); | |
1965 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length - 1); | |
1966 | loop = LABEL(); | |
1967 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0); | |
1968 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1); | |
1969 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | |
1970 | } | |
1971 | } | |
1972 | ||
1973 | static SLJIT_INLINE void do_reset_match(compiler_common *common, int length) | |
1974 | { | |
1975 | DEFINE_COMPILER; | |
1976 | struct sljit_label *loop; | |
1977 | int i; | |
1978 | ||
1979 | SLJIT_ASSERT(length > 1); | |
1980 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | |
1981 | if (length > 2) | |
1982 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | |
1983 | if (length < 8) | if (length < 8) |
1984 | { | { |
1985 | for (i = 0; i < length; i++) | for (i = 2; i < length; i++) |
1986 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), TMP1, 0); |
1987 | } | } |
1988 | else | else |
1989 | { | { |
1990 | OP2(SLJIT_ADD, SLJIT_TEMPORARY_REG2, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START - sizeof(sljit_w)); | GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw)); |
1991 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2); |
1992 | loop = LABEL(); | loop = LABEL(); |
1993 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_w), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0); |
1994 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1); |
1995 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
1996 | } | } |
1997 | ||
1998 | OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0); | |
1999 | if (common->mark_ptr != 0) | |
2000 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0); | |
2001 | SLJIT_ASSERT(common->control_head_ptr != 0); | |
2002 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); | |
2003 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack)); | |
2004 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr); | |
2005 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base)); | |
2006 | } | |
2007 | ||
2008 | static sljit_sw do_check_control_chain(sljit_sw *current) | |
2009 | { | |
2010 | sljit_sw return_value = 0; | |
2011 | ||
2012 | SLJIT_ASSERT(current != NULL); | |
2013 | do | |
2014 | { | |
2015 | switch (current[-2]) | |
2016 | { | |
2017 | case type_commit: | |
2018 | /* Commit overwrites all. */ | |
2019 | return -1; | |
2020 | ||
2021 | case type_prune: | |
2022 | break; | |
2023 | ||
2024 | case type_skip: | |
2025 | /* Overwrites prune, but not other skips. */ | |
2026 | if (return_value == 0) | |
2027 | return_value = current[-3]; | |
2028 | break; | |
2029 | ||
2030 | default: | |
2031 | SLJIT_ASSERT_STOP(); | |
2032 | break; | |
2033 | } | |
2034 | current = (sljit_sw*)current[-1]; | |
2035 | } | |
2036 | while (current != NULL); | |
2037 | return return_value; | |
2038 | } | } |
2039 | ||
2040 | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) |
2041 | { | { |
2042 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2043 | struct sljit_label *loop; | struct sljit_label *loop; |
2044 | struct sljit_jump *earlyexit; | struct sljit_jump *early_quit; |
2045 | ||
2046 | /* At this point we can freely use all registers. */ | /* At this point we can freely use all registers. */ |
2047 | OP1(SLJIT_MOV, SLJIT_GENERAL_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); |
2048 | 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); |
2049 | ||
2050 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, ARGUMENTS, 0); |
2051 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | if (common->mark_ptr != 0) |
2052 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); |
2053 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offset_count)); |
2054 | OP2(SLJIT_ADD, SLJIT_GENERAL_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START); | if (common->mark_ptr != 0) |
2055 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_SCRATCH_REG3, 0); | |
2056 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | |
2057 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | |
2058 | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); | |
2059 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
2060 | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); | early_quit = CMP(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 0); |
2061 | loop = LABEL(); | loop = LABEL(); |
2062 | OP2(SLJIT_SUB, SLJIT_GENERAL_REG2, 0, SLJIT_MEM1(SLJIT_GENERAL_REG1), 0, SLJIT_TEMPORARY_REG1, 0); | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_SCRATCH_REG1, 0); |
2063 | OP2(SLJIT_ADD, SLJIT_GENERAL_REG1, 0, SLJIT_GENERAL_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_sw)); |
2064 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
2065 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2066 | OP2(SLJIT_ASHR, SLJIT_GENERAL_REG2, 0, SLJIT_GENERAL_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2067 | #endif | #endif |
2068 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_GENERAL_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); |
2069 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); |
2070 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
2071 | JUMPHERE(earlyexit); | JUMPHERE(early_quit); |
2072 | ||
2073 | /* Calculate the return value, which is the maximum ovector value. */ | /* Calculate the return value, which is the maximum ovector value. */ |
2074 | if (topbracket > 1) | if (topbracket > 1) |
2075 | { | { |
2076 | OP2(SLJIT_ADD, SLJIT_TEMPORARY_REG1, 0, SLJIT_LOCALS_REG, 0, SLJIT_IMM, OVECTOR_START + topbracket * 2 * sizeof(sljit_w)); | GET_LOCAL_BASE(SLJIT_SCRATCH_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); |
2077 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, topbracket + 1); |
2078 | ||
2079 | /* OVECTOR(0) is never equal to SLJIT_GENERAL_REG3. */ | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ |
2080 | loop = LABEL(); | loop = LABEL(); |
2081 | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w))); | OP1(SLJIT_MOVU, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); |
2082 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); |
2083 | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_GENERAL_REG3, 0, loop); | CMPTO(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG3, 0, loop); |
2084 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_SCRATCH_REG2, 0); |
2085 | } | } |
2086 | else | else |
2087 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
2088 | } | } |
2089 | ||
2090 | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) | |
2091 | { | |
2092 | DEFINE_COMPILER; | |
2093 | struct sljit_jump *jump; | |
2094 | ||
2095 | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); | |
2096 | SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0 | |
2097 | && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | |
2098 | ||
2099 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0); | |
2100 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | |
2101 | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count)); | |
2102 | CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit); | |
2103 | ||
2104 | /* Store match begin and end. */ | |
2105 | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); | |
2106 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | |
2107 | ||
2108 | jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3); | |
2109 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0); | |
2110 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2111 | OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2112 | #endif | |
2113 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0); | |
2114 | JUMPHERE(jump); | |
2115 | ||
2116 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
2117 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); | |
2118 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2119 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2120 | #endif | |
2121 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | |
2122 | ||
2123 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG1, 0); | |
2124 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2125 | OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2126 | #endif | |
2127 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 0, SLJIT_SCRATCH_REG3, 0); | |
2128 | ||
2129 | JUMPTO(SLJIT_JUMP, quit); | |
2130 | } | |
2131 | ||
2132 | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) | |
2133 | { | |
2134 | /* May destroy TMP1. */ | |
2135 | DEFINE_COMPILER; | |
2136 | struct sljit_jump *jump; | |
2137 | ||
2138 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2139 | { | |
2140 | /* The value of -1 must be kept for start_used_ptr! */ | |
2141 | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); | |
2142 | /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting | |
2143 | is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ | |
2144 | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); | |
2145 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
2146 | JUMPHERE(jump); | |
2147 | } | |
2148 | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) | |
2149 | { | |
2150 | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
2151 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
2152 | JUMPHERE(jump); | |
2153 | } | |
2154 | } | |
2155 | ||
2156 | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) |
2157 | { | { |
2158 | /* Detects if the character has an othercase. */ | /* Detects if the character has an othercase. */ |
# | Line 1348 if (c <= 127 && bit == 0x20) | Line 2237 if (c <= 127 && bit == 0x20) |
2237 | return (0 << 8) | 0x20; | return (0 << 8) | 0x20; |
2238 | ||
2239 | /* Since c != oc, they must have at least 1 bit difference. */ | /* Since c != oc, they must have at least 1 bit difference. */ |
2240 | if (!ispowerof2(bit)) | if (!is_powerof2(bit)) |
2241 | return 0; | return 0; |
2242 | ||
2243 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2244 | ||
2245 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2246 | if (common->utf && c > 127) | if (common->utf && c > 127) |
# | Line 1367 if (common->utf && c > 127) | Line 2256 if (common->utf && c > 127) |
2256 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2257 | return (0 << 8) | bit; | return (0 << 8) | bit; |
2258 | ||
2259 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2260 | ||
#ifdef COMPILE_PCRE16 | ||
2261 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2262 | if (common->utf && c > 65535) | if (common->utf && c > 65535) |
2263 | { | { |
# | Line 1380 if (common->utf && c > 65535) | Line 2268 if (common->utf && c > 65535) |
2268 | } | } |
2269 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2270 | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); |
#endif /* COMPILE_PCRE16 */ | ||
2271 | ||
2272 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16|32] */ |
2273 | } | |
2274 | ||
2275 | static void check_partial(compiler_common *common, BOOL force) | |
2276 | { | |
2277 | /* Checks whether a partial matching is occured. Does not modify registers. */ | |
2278 | DEFINE_COMPILER; | |
2279 | struct sljit_jump *jump = NULL; | |
2280 | ||
2281 | SLJIT_ASSERT(!force || common->mode != JIT_COMPILE); | |
2282 | ||
2283 | if (common->mode == JIT_COMPILE) | |
2284 | return; | |
2285 | ||
2286 | if (!force) | |
2287 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | |
2288 | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2289 | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); | |
2290 | ||
2291 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2292 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); | |
2293 | else | |
2294 | { | |
2295 | if (common->partialmatchlabel != NULL) | |
2296 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
2297 | else | |
2298 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
2299 | } | |
2300 | ||
2301 | if (jump != NULL) | |
2302 | JUMPHERE(jump); | |
2303 | } | |
2304 | ||
2305 | static void check_str_end(compiler_common *common, jump_list **end_reached) | |
2306 | { | |
2307 | /* Does not affect registers. Usually used in a tight spot. */ | |
2308 | DEFINE_COMPILER; | |
2309 | struct sljit_jump *jump; | |
2310 | ||
2311 | if (common->mode == JIT_COMPILE) | |
2312 | { | |
2313 | add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | |
2314 | return; | |
2315 | } | |
2316 | ||
2317 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | |
2318 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2319 | { | |
2320 | add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); | |
2321 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); | |
2322 | add_jump(compiler, end_reached, JUMP(SLJIT_JUMP)); | |
2323 | } | |
2324 | else | |
2325 | { | |
2326 | add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); | |
2327 | if (common->partialmatchlabel != NULL) | |
2328 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
2329 | else | |
2330 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
2331 | } | |
2332 | JUMPHERE(jump); | |
2333 | } | } |
2334 | ||
2335 | static SLJIT_INLINE void check_input_end(compiler_common *common, jump_list **fallbacks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
2336 | { | { |
2337 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2338 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | struct sljit_jump *jump; |
2339 | ||
2340 | if (common->mode == JIT_COMPILE) | |
2341 | { | |
2342 | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | |
2343 | return; | |
2344 | } | |
2345 | ||
2346 | /* Partial matching mode. */ | |
2347 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | |
2348 | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); | |
2349 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | |
2350 | { | |
2351 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); | |
2352 | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); | |
2353 | } | |
2354 | else | |
2355 | { | |
2356 | if (common->partialmatchlabel != NULL) | |
2357 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | |
2358 | else | |
2359 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | |
2360 | } | |
2361 | JUMPHERE(jump); | |
2362 | } | } |
2363 | ||
2364 | static void read_char(compiler_common *common) | static void read_char(compiler_common *common) |
# | Line 1396 static void read_char(compiler_common *c | Line 2366 static void read_char(compiler_common *c |
2366 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, updates STR_PTR. |
2367 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2368 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2369 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2370 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2371 | #endif | #endif |
2372 | ||
2373 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2374 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2375 | if (common->utf) | if (common->utf) |
2376 | { | { |
2377 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2378 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2379 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2380 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2381 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2382 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2383 | JUMPHERE(jump); | JUMPHERE(jump); |
2384 | } | } |
2385 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2386 | 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)); |
2387 | } | } |
2388 | ||
# | Line 1423 static void peek_char(compiler_common *c | Line 2391 static void peek_char(compiler_common *c |
2391 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2392 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2393 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2394 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2395 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2396 | #endif | #endif |
2397 | ||
2398 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2399 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2400 | if (common->utf) | if (common->utf) |
2401 | { | { |
2402 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2403 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2404 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2405 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2406 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2407 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2408 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2409 | JUMPHERE(jump); | JUMPHERE(jump); |
2410 | } | } |
2411 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2412 | } | } |
2413 | ||
2414 | static void read_char8_type(compiler_common *common) | static void read_char8_type(compiler_common *common) |
2415 | { | { |
2416 | /* 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. */ |
2417 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2418 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2419 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2420 | #endif | #endif |
2421 | ||
# | Line 1458 if (common->utf) | Line 2424 if (common->utf) |
2424 | { | { |
2425 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2426 | 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)); |
2427 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2428 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
2429 | it is needed in most cases. */ | it is needed in most cases. */ |
2430 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2431 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2432 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
2433 | JUMPHERE(jump); | JUMPHERE(jump); |
2434 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2435 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2436 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2437 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
# | Line 1474 if (common->utf) | Line 2439 if (common->utf) |
2439 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2440 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); |
2441 | 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); |
2442 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2443 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
2444 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2445 | #endif | #elif defined COMPILE_PCRE32 |
2446 | #endif /* COMPILE_PCRE8 */ | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2447 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
2448 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2449 | JUMPHERE(jump); | |
2450 | #endif /* COMPILE_PCRE[8|16|32] */ | |
2451 | return; | return; |
2452 | } | } |
2453 | #endif | #endif /* SUPPORT_UTF */ |
2454 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2455 | 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)); |
2456 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2457 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2458 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2459 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2460 | #endif | #endif |
2461 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2462 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2463 | JUMPHERE(jump); | JUMPHERE(jump); |
2464 | #endif | #endif |
2465 | } | } |
# | Line 1499 static void skip_char_back(compiler_comm | Line 2468 static void skip_char_back(compiler_comm |
2468 | { | { |
2469 | /* 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. */ |
2470 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2471 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2472 | #if defined COMPILE_PCRE8 | |
2473 | struct sljit_label *label; | struct sljit_label *label; |
2474 | ||
2475 | if (common->utf) | if (common->utf) |
# | Line 1511 if (common->utf) | Line 2481 if (common->utf) |
2481 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
2482 | return; | return; |
2483 | } | } |
2484 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2485 | if (common->utf) | if (common->utf) |
2486 | { | { |
2487 | 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 1520 if (common->utf) | Line 2489 if (common->utf) |
2489 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2490 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2491 | 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); |
2492 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2493 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2494 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2495 | return; | return; |
2496 | } | } |
2497 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2498 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2499 | 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)); |
2500 | } | } |
2501 | ||
2502 | 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) |
2503 | { | { |
2504 | /* 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. */ |
2505 | DEFINE_COMPILER; | DEFINE_COMPILER; |
# | Line 1537 DEFINE_COMPILER; | Line 2507 DEFINE_COMPILER; |
2507 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2508 | { | { |
2509 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2510 | 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)); |
2511 | } | } |
2512 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2513 | { | { |
2514 | 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); |
2515 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2516 | 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); |
2517 | 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); |
2518 | 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)); |
2519 | } | } |
2520 | else | else |
2521 | { | { |
2522 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2523 | 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)); |
2524 | } | } |
2525 | } | } |
2526 | ||
2527 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2528 | ||
2529 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2530 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2531 | { | { |
2532 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
# | Line 1564 of the character (>= 0xc0). Return char | Line 2534 of the character (>= 0xc0). Return char |
2534 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2535 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2536 | ||
2537 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2538 | /* Searching for the first zero. */ | /* Searching for the first zero. */ |
2539 | 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); |
2540 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
# | Line 1623 DEFINE_COMPILER; | Line 2593 DEFINE_COMPILER; |
2593 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2594 | struct sljit_jump *compare; | struct sljit_jump *compare; |
2595 | ||
2596 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2597 | ||
2598 | 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); |
2599 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
# | Line 1644 sljit_emit_fast_return(compiler, RETURN_ | Line 2614 sljit_emit_fast_return(compiler, RETURN_ |
2614 | JUMPHERE(jump); | JUMPHERE(jump); |
2615 | ||
2616 | /* We only have types for characters less than 256. */ | /* We only have types for characters less than 256. */ |
2617 | 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); |
2618 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2619 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2620 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2621 | } | } |
2622 | ||
2623 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
2624 | ||
#ifdef COMPILE_PCRE16 | ||
2625 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2626 | { | { |
2627 | /* 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 1660 of the character (>= 0xd800). Return cha | Line 2629 of the character (>= 0xd800). Return cha |
2629 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2630 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2631 | ||
2632 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2633 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); |
2634 | /* Do nothing, only return. */ | /* Do nothing, only return. */ |
2635 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
# | Line 1677 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC | Line 2646 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC |
2646 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2647 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2648 | } | } |
#endif /* COMPILE_PCRE16 */ | ||
2649 | ||
2650 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16] */ |
2651 | ||
2652 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2653 | ||
# | Line 1697 DEFINE_COMPILER; | Line 2665 DEFINE_COMPILER; |
2665 | ||
2666 | SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); | SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); |
2667 | ||
2668 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2669 | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2670 | 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)); |
2671 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); |
2672 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2673 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); |
2674 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2)); |
2675 | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); |
2676 | 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)); |
2677 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
2678 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2679 | } | } |
# | Line 1719 struct sljit_label *newlinelabel = NULL; | Line 2687 struct sljit_label *newlinelabel = NULL; |
2687 | struct sljit_jump *start; | struct sljit_jump *start; |
2688 | struct sljit_jump *end = NULL; | struct sljit_jump *end = NULL; |
2689 | struct sljit_jump *nl = NULL; | struct sljit_jump *nl = NULL; |
2690 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2691 | struct sljit_jump *singlechar; | struct sljit_jump *singlechar; |
2692 | #endif | #endif |
2693 | jump_list *newline = NULL; | jump_list *newline = NULL; |
# | Line 1733 if (!(hascrorlf || firstline) && (common | Line 2701 if (!(hascrorlf || firstline) && (common |
2701 | if (firstline) | if (firstline) |
2702 | { | { |
2703 | /* Search for the end of the first line. */ | /* Search for the end of the first line. */ |
2704 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2705 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); |
2706 | ||
2707 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
2708 | { | { |
# | Line 1745 if (firstline) | Line 2713 if (firstline) |
2713 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2714 | 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); |
2715 | 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); |
2716 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | JUMPHERE(end); |
2717 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2718 | } | |
2719 | else | |
2720 | { | |
2721 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
2722 | mainloop = LABEL(); | |
2723 | /* Continual stores does not cause data dependency. */ | |
2724 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | |
2725 | read_char(common); | |
2726 | check_newlinechar(common, common->nltype, &newline, TRUE); | |
2727 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | |
2728 | JUMPHERE(end); | |
2729 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | |
2730 | set_jumps(newline, LABEL()); | |
2731 | } | |
2732 | ||
2733 | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); | |
2734 | } | |
2735 | ||
2736 | start = JUMP(SLJIT_JUMP); | |
2737 | ||
2738 | if (newlinecheck) | |
2739 | { | |
2740 | newlinelabel = LABEL(); | |
2741 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2742 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
2743 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
2744 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); | |
2745 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
2746 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2747 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2748 | #endif | |
2749 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
2750 | nl = JUMP(SLJIT_JUMP); | |
2751 | } | |
2752 | ||
2753 | mainloop = LABEL(); | |
2754 | ||
2755 | /* Increasing the STR_PTR here requires one less jump in the most common case. */ | |
2756 | #ifdef SUPPORT_UTF | |
2757 | if (common->utf) readuchar = TRUE; | |
2758 | #endif | |
2759 | if (newlinecheck) readuchar = TRUE; | |
2760 | ||
2761 | if (readuchar) | |
2762 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
2763 | ||
2764 | if (newlinecheck) | |
2765 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | |
2766 | ||
2767 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2768 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
2769 | #if defined COMPILE_PCRE8 | |
2770 | if (common->utf) | |
2771 | { | |
2772 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
2773 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2774 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
2775 | JUMPHERE(singlechar); | |
2776 | } | |
2777 | #elif defined COMPILE_PCRE16 | |
2778 | if (common->utf) | |
2779 | { | |
2780 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
2781 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
2782 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2783 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
2784 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
2785 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
2786 | JUMPHERE(singlechar); | |
2787 | } | |
2788 | #endif /* COMPILE_PCRE[8|16] */ | |
2789 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2790 | JUMPHERE(start); | |
2791 | ||
2792 | if (newlinecheck) | |
2793 | { | |
2794 | JUMPHERE(end); | |
2795 | JUMPHERE(nl); | |
2796 | } | |
2797 | ||
2798 | return mainloop; | |
2799 | } | |
2800 | ||
2801 | #define MAX_N_CHARS 3 | |
2802 | ||
2803 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | |
2804 | { | |
2805 | DEFINE_COMPILER; | |
2806 | struct sljit_label *start; | |
2807 | struct sljit_jump *quit; | |
2808 | pcre_uint32 chars[MAX_N_CHARS * 2]; | |
2809 | pcre_uchar *cc = common->start + 1 + LINK_SIZE; | |
2810 | int location = 0; | |
2811 | pcre_int32 len, c, bit, caseless; | |
2812 | int must_stop; | |
2813 | ||
2814 | /* We do not support alternatives now. */ | |
2815 | if (*(common->start + GET(common->start, 1)) == OP_ALT) | |
2816 | return FALSE; | |
2817 | ||
2818 | while (TRUE) | |
2819 | { | |
2820 | caseless = 0; | |
2821 | must_stop = 1; | |
2822 | switch(*cc) | |
2823 | { | |
2824 | case OP_CHAR: | |
2825 | must_stop = 0; | |
2826 | cc++; | |
2827 | break; | |
2828 | ||
2829 | case OP_CHARI: | |
2830 | caseless = 1; | |
2831 | must_stop = 0; | |
2832 | cc++; | |
2833 | break; | |
2834 | ||
2835 | case OP_SOD: | |
2836 | case OP_SOM: | |
2837 | case OP_SET_SOM: | |
2838 | case OP_NOT_WORD_BOUNDARY: | |
2839 | case OP_WORD_BOUNDARY: | |
2840 | case OP_EODN: | |
2841 | case OP_EOD: | |
2842 | case OP_CIRC: | |
2843 | case OP_CIRCM: | |
2844 | case OP_DOLL: | |
2845 | case OP_DOLLM: | |
2846 | /* Zero width assertions. */ | |
2847 | cc++; | |
2848 | continue; | |
2849 | ||
2850 | case OP_PLUS: | |
2851 | case OP_MINPLUS: | |
2852 | case OP_POSPLUS: | |
2853 | cc++; | |
2854 | break; | |
2855 | ||
2856 | case OP_EXACT: | |
2857 | cc += 1 + IMM2_SIZE; | |
2858 | break; | |
2859 | ||
2860 | case OP_PLUSI: | |
2861 | case OP_MINPLUSI: | |
2862 | case OP_POSPLUSI: | |
2863 | caseless = 1; | |
2864 | cc++; | |
2865 | break; | |
2866 | ||
2867 | case OP_EXACTI: | |
2868 | caseless = 1; | |
2869 | cc += 1 + IMM2_SIZE; | |
2870 | break; | |
2871 | ||
2872 | default: | |
2873 | must_stop = 2; | |
2874 | break; | |
2875 | } | |
2876 | ||
2877 | if (must_stop == 2) | |
2878 | break; | |
2879 | ||
2880 | len = 1; | |
2881 | #ifdef SUPPORT_UTF | |
2882 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | |
2883 | #endif | |
2884 | ||
2885 | if (caseless && char_has_othercase(common, cc)) | |
2886 | { | |
2887 | caseless = char_get_othercase_bit(common, cc); | |
2888 | if (caseless == 0) | |
2889 | return FALSE; | |
2890 | #ifdef COMPILE_PCRE8 | |
2891 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | |
2892 | #else | |
2893 | if ((caseless & 0x100) != 0) | |
2894 | caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | |
2895 | else | |
2896 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | |
2897 | #endif | |
2898 | } | } |
2899 | else | else |
2900 | caseless = 0; | |
2901 | ||
2902 | while (len > 0 && location < MAX_N_CHARS * 2) | |
2903 | { | { |
2904 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | c = *cc; |
2905 | mainloop = LABEL(); | bit = 0; |
2906 | /* Continual stores does not cause data dependency. */ | if (len == (caseless & 0xff)) |
2907 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | { |
2908 | read_char(common); | bit = caseless >> 8; |
2909 | check_newlinechar(common, common->nltype, &newline, TRUE); | c |= bit; |
2910 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | } |
2911 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END, STR_PTR, 0); | |
2912 | set_jumps(newline, LABEL()); | chars[location] = c; |
2913 | chars[location + 1] = bit; | |
2914 | ||
2915 | len--; | |
2916 | location += 2; | |
2917 | cc++; | |
2918 | } | } |
2919 | ||
2920 | JUMPHERE(end); | if (location >= MAX_N_CHARS * 2 || must_stop != 0) |
2921 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | break; |
2922 | } | } |
2923 | ||
2924 | start = JUMP(SLJIT_JUMP); | /* At least two characters are required. */ |
2925 | if (location < 2 * 2) | |
2926 | return FALSE; | |
2927 | ||
2928 | if (newlinecheck) | if (firstline) |
2929 | { | { |
2930 | newlinelabel = LABEL(); | SLJIT_ASSERT(common->first_line_end != 0); |
2931 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
2932 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); |
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | ||
OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); | ||
COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | ||
#ifdef COMPILE_PCRE16 | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | ||
#endif | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
nl = JUMP(SLJIT_JUMP); | ||
2933 | } | } |
2934 | else | |
2935 | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | |
2936 | ||
2937 | mainloop = LABEL(); | start = LABEL(); |
2938 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
/* Increasing the STR_PTR here requires one less jump in the most common case. */ | ||
#ifdef SUPPORT_UTF | ||
if (common->utf) readuchar = TRUE; | ||
#endif | ||
if (newlinecheck) readuchar = TRUE; | ||
if (readuchar) | ||
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | ||
if (newlinecheck) | ||
CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | ||
2939 | ||
2940 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2941 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2942 | 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)); |
2943 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | if (chars[1] != 0) |
2944 | if (common->utf) | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); |
2945 | { | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); |
2946 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | if (location > 2 * 2) |
2947 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
2948 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | if (chars[3] != 0) |
2949 | JUMPHERE(singlechar); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); |
2950 | } | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); |
2951 | #endif | if (location > 2 * 2) |
2952 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | { |
2953 | if (common->utf) | if (chars[5] != 0) |
2954 | { | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); |
2955 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], 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); | ||
JUMPHERE(singlechar); | ||
2956 | } | } |
2957 | #endif | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
JUMPHERE(start); | ||
2958 | ||
2959 | if (newlinecheck) | JUMPHERE(quit); |
{ | ||
JUMPHERE(end); | ||
JUMPHERE(nl); | ||
} | ||
2960 | ||
2961 | return mainloop; | if (firstline) |
2962 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | |
2963 | else | |
2964 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | |
2965 | return TRUE; | |
2966 | } | } |
2967 | ||
2968 | #undef MAX_N_CHARS | |
2969 | ||
2970 | 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) |
2971 | { | { |
2972 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2973 | struct sljit_label *start; | struct sljit_label *start; |
2974 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2975 | struct sljit_jump *found; | struct sljit_jump *found; |
2976 | pcre_uchar oc, bit; | pcre_uchar oc, bit; |
2977 | ||
2978 | if (firstline) | if (firstline) |
2979 | { | { |
2980 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2981 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
2982 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
2983 | } | } |
2984 | ||
2985 | start = LABEL(); | start = LABEL(); |
2986 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2987 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2988 | ||
2989 | oc = first_char; | oc = first_char; |
# | Line 1860 if (first_char == oc) | Line 3000 if (first_char == oc) |
3000 | else | else |
3001 | { | { |
3002 | bit = first_char ^ oc; | bit = first_char ^ oc; |
3003 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
3004 | { | { |
3005 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
3006 | 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 1868 else | Line 3008 else |
3008 | else | else |
3009 | { | { |
3010 | 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); |
3011 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3012 | 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); |
3013 | 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); |
3014 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
3015 | } | } |
3016 | } | } |
3017 | ||
3018 | 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 | ||
3019 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
3020 | JUMPHERE(found); | JUMPHERE(found); |
3021 | JUMPHERE(leave); | JUMPHERE(quit); |
3022 | ||
3023 | if (firstline) | if (firstline) |
3024 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3025 | } | } |
3026 | ||
3027 | 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 1909 DEFINE_COMPILER; | Line 3030 DEFINE_COMPILER; |
3030 | struct sljit_label *loop; | struct sljit_label *loop; |
3031 | struct sljit_jump *lastchar; | struct sljit_jump *lastchar; |
3032 | struct sljit_jump *firstchar; | struct sljit_jump *firstchar; |
3033 | struct sljit_jump *leave; | struct sljit_jump *quit; |
3034 | struct sljit_jump *foundcr = NULL; | struct sljit_jump *foundcr = NULL; |
3035 | struct sljit_jump *notfoundnl; | struct sljit_jump *notfoundnl; |
3036 | jump_list *newline = NULL; | jump_list *newline = NULL; |
3037 | ||
3038 | if (firstline) | if (firstline) |
3039 | { | { |
3040 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3041 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3042 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
3043 | } | } |
3044 | ||
3045 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
# | Line 1930 if (common->nltype == NLTYPE_FIXED && co | Line 3052 if (common->nltype == NLTYPE_FIXED && co |
3052 | ||
3053 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
3054 | 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); |
3055 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); |
3056 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3057 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
3058 | #endif | #endif |
3059 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
3060 | ||
3061 | loop = LABEL(); | loop = LABEL(); |
3062 | 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)); |
3063 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3064 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
3065 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
3066 | 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); |
3067 | 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); |
3068 | ||
3069 | JUMPHERE(leave); | JUMPHERE(quit); |
3070 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
3071 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
3072 | ||
# | Line 1968 set_jumps(newline, loop); | Line 3090 set_jumps(newline, loop); |
3090 | ||
3091 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
3092 | { | { |
3093 | leave = JUMP(SLJIT_JUMP); | quit = JUMP(SLJIT_JUMP); |
3094 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
3095 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3096 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3097 | 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); |
3098 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3099 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3100 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
3101 | #endif | #endif |
3102 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3103 | JUMPHERE(notfoundnl); | JUMPHERE(notfoundnl); |
3104 | JUMPHERE(leave); | JUMPHERE(quit); |
3105 | } | } |
3106 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
3107 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
3108 | ||
3109 | if (firstline) | if (firstline) |
3110 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3111 | } | } |
3112 | ||
3113 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks); | |
3114 | ||
3115 | 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) |
3116 | { | { |
3117 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3118 | struct sljit_label *start; | struct sljit_label *start; |
3119 | struct sljit_jump *leave; | struct sljit_jump *quit; |
3120 | struct sljit_jump *found; | struct sljit_jump *found = NULL; |
3121 | jump_list *matches = NULL; | |
3122 | pcre_uint8 inverted_start_bits[32]; | |
3123 | int i; | |
3124 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3125 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3126 | #endif | #endif |
3127 | ||
3128 | for (i = 0; i < 32; ++i) | |
3129 | inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]); | |
3130 | ||
3131 | if (firstline) | if (firstline) |
3132 | { | { |
3133 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3134 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), FIRSTLINE_END); | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); |
3135 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | |
3136 | } | } |
3137 | ||
3138 | start = LABEL(); | start = LABEL(); |
3139 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3140 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3141 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3142 | if (common->utf) | if (common->utf) |
3143 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
3144 | #endif | #endif |
3145 | ||
3146 | if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches)) | |
3147 | { | |
3148 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3149 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); |
3150 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); |
3151 | JUMPHERE(jump); | JUMPHERE(jump); |
3152 | #endif | #endif |
3153 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3154 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3155 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); |
3156 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3157 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); |
3158 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
3159 | } | |
3160 | ||
3161 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3162 | if (common->utf) | if (common->utf) |
3163 | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
3164 | #endif | #endif |
3165 | 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)); |
3166 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #ifdef SUPPORT_UTF |
3167 | #if defined COMPILE_PCRE8 | |
3168 | if (common->utf) | if (common->utf) |
3169 | { | { |
3170 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
3171 | 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); |
3172 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3173 | } | } |
3174 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
3175 | if (common->utf) | if (common->utf) |
3176 | { | { |
3177 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
3178 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
3179 | 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); |
3180 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3181 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
3182 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3183 | } | } |
3184 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
3185 | #endif /* SUPPORT_UTF */ | |
3186 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
3187 | JUMPHERE(found); | if (found != NULL) |
3188 | JUMPHERE(leave); | JUMPHERE(found); |
3189 | if (matches != NULL) | |
3190 | set_jumps(matches, LABEL()); | |
3191 | JUMPHERE(quit); | |
3192 | ||
3193 | if (firstline) | if (firstline) |
3194 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0); |
3195 | } | } |
3196 | ||
3197 | 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 2064 struct sljit_jump *alreadyfound; | Line 3203 struct sljit_jump *alreadyfound; |
3203 | struct sljit_jump *found; | struct sljit_jump *found; |
3204 | struct sljit_jump *foundoc = NULL; | struct sljit_jump *foundoc = NULL; |
3205 | struct sljit_jump *notfound; | struct sljit_jump *notfound; |
3206 | pcre_uchar oc, bit; | pcre_uint32 oc, bit; |
3207 | ||
3208 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), REQ_CHAR_PTR); | SLJIT_ASSERT(common->req_char_ptr != 0); |
3209 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); | |
3210 | 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); |
3211 | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); |
3212 | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); |
# | Line 2094 if (req_char == oc) | Line 3234 if (req_char == oc) |
3234 | else | else |
3235 | { | { |
3236 | bit = req_char ^ oc; | bit = req_char ^ oc; |
3237 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
3238 | { | { |
3239 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
3240 | 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 2111 JUMPTO(SLJIT_JUMP, loop); | Line 3251 JUMPTO(SLJIT_JUMP, loop); |
3251 | JUMPHERE(found); | JUMPHERE(found); |
3252 | if (foundoc) | if (foundoc) |
3253 | JUMPHERE(foundoc); | JUMPHERE(foundoc); |
3254 | 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); |
3255 | JUMPHERE(alreadyfound); | JUMPHERE(alreadyfound); |
3256 | JUMPHERE(toolong); | JUMPHERE(toolong); |
3257 | return notfound; | return notfound; |
# | Line 2123 DEFINE_COMPILER; | Line 3263 DEFINE_COMPILER; |
3263 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3264 | struct sljit_label *mainloop; | struct sljit_label *mainloop; |
3265 | ||
3266 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3267 | OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); | OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); |
3268 | GET_LOCAL_BASE(TMP3, 0, 0); | |
3269 | ||
3270 | /* Drop frames until we reach STACK_TOP. */ | /* Drop frames until we reach STACK_TOP. */ |
3271 | mainloop = LABEL(); | mainloop = LABEL(); |
3272 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
3273 | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); |
3274 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_LOCALS_REG, 0); | jump = JUMP(SLJIT_C_SIG_LESS_EQUAL); |
3275 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | |
3276 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_w), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
3277 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
3278 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw)); | |
3279 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_sw)); | |
3280 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3281 | ||
3282 | JUMPHERE(jump); | JUMPHERE(jump); |
3283 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = JUMP(SLJIT_C_SIG_LESS); |
3284 | /* End of dropping frames. */ | /* End of dropping frames. */ |
3285 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3286 | ||
3287 | JUMPHERE(jump); | JUMPHERE(jump); |
3288 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_NEG, TMP2, 0, TMP2, 0); |
3289 | /* Set string begin. */ | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
3290 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
3291 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); |
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); | ||
JUMPTO(SLJIT_JUMP, mainloop); | ||
JUMPHERE(jump); | ||
/* Unknown command. */ | ||
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | ||
3292 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3293 | } | } |
3294 | ||
3295 | static void check_wordboundary(compiler_common *common) | static void check_wordboundary(compiler_common *common) |
3296 | { | { |
3297 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3298 | struct sljit_jump *beginend; | struct sljit_jump *skipread; |
3299 | jump_list *skipread_list = NULL; | |
3300 | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF |
3301 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3302 | #endif | #endif |
3303 | ||
3304 | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); |
3305 | ||
3306 | 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); |
3307 | /* Get type of the previous char, and put it to LOCALS1. */ | /* Get type of the previous char, and put it to LOCALS1. */ |
3308 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
3309 | 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)); |
3310 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); |
3311 | beginend = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); |
3312 | skip_char_back(common); | skip_char_back(common); |
3313 | check_start_used_ptr(common); | |
3314 | read_char(common); | read_char(common); |
3315 | ||
3316 | /* Testing char type. */ | /* Testing char type. */ |
# | Line 2183 if (common->use_ucp) | Line 3322 if (common->use_ucp) |
3322 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3323 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3324 | 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); |
3325 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3326 | 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); |
3327 | 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); |
3328 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3329 | JUMPHERE(jump); | JUMPHERE(jump); |
3330 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); |
3331 | } | } |
# | Line 2212 else | Line 3351 else |
3351 | JUMPHERE(jump); | JUMPHERE(jump); |
3352 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
3353 | } | } |
3354 | JUMPHERE(beginend); | JUMPHERE(skipread); |
3355 | ||
3356 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
3357 | beginend = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | check_str_end(common, &skipread_list); |
3358 | peek_char(common); | peek_char(common); |
3359 | ||
3360 | /* Testing char type. This is a code duplication. */ | /* Testing char type. This is a code duplication. */ |
# | Line 2227 if (common->use_ucp) | Line 3366 if (common->use_ucp) |
3366 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3367 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3368 | 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); |
3369 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3370 | 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); |
3371 | 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); |
3372 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3373 | JUMPHERE(jump); | JUMPHERE(jump); |
3374 | } | } |
3375 | else | else |
# | Line 2256 else | Line 3395 else |
3395 | JUMPHERE(jump); | JUMPHERE(jump); |
3396 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
3397 | } | } |
3398 | JUMPHERE(beginend); | set_jumps(skipread_list, LABEL()); |
3399 | ||
3400 | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); |
3401 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
3402 | } | } |
3403 | ||
3404 | /* | |
3405 | range format: | |
3406 | ||
3407 | ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). | |
3408 | ranges[1] = first bit (0 or 1) | |
3409 | ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) | |
3410 | */ | |
3411 | ||
3412 | static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) | |
3413 | { | |
3414 | DEFINE_COMPILER; | |
3415 | struct sljit_jump *jump; | |
3416 | ||
3417 | if (ranges[0] < 0) | |
3418 | return FALSE; | |
3419 | ||
3420 | switch(ranges[0]) | |
3421 | { | |
3422 | case 1: | |
3423 | if (readch) | |
3424 | read_char(common); | |
3425 | add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3426 | return TRUE; | |
3427 | ||
3428 | case 2: | |
3429 | if (readch) | |
3430 | read_char(common); | |
3431 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3432 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); | |
3433 | return TRUE; | |
3434 | ||
3435 | case 4: | |
3436 | if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) | |
3437 | { | |
3438 | if (readch) | |
3439 | read_char(common); | |
3440 | if (ranges[1] != 0) | |
3441 | { | |
3442 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3443 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3444 | } | |
3445 | else | |
3446 | { | |
3447 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3448 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3449 | JUMPHERE(jump); | |
3450 | } | |
3451 | return TRUE; | |
3452 | } | |
3453 | if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) | |
3454 | { | |
3455 | if (readch) | |
3456 | read_char(common); | |
3457 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); | |
3458 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); | |
3459 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); | |
3460 | return TRUE; | |
3461 | } | |
3462 | return FALSE; | |
3463 | ||
3464 | default: | |
3465 | return FALSE; | |
3466 | } | |
3467 | } | |
3468 | ||
3469 | static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) | |
3470 | { | |
3471 | int i, bit, length; | |
3472 | const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; | |
3473 | ||
3474 | bit = ctypes[0] & flag; | |
3475 | ranges[0] = -1; | |
3476 | ranges[1] = bit != 0 ? 1 : 0; | |
3477 | length = 0; | |
3478 | ||
3479 | for (i = 1; i < 256; i++) | |
3480 | if ((ctypes[i] & flag) != bit) | |
3481 | { | |
3482 | if (length >= MAX_RANGE_SIZE) | |
3483 | return; | |
3484 | ranges[2 + length] = i; | |
3485 | length++; | |
3486 | bit ^= flag; | |
3487 | } | |
3488 | ||
3489 | if (bit != 0) | |
3490 | { | |
3491 | if (length >= MAX_RANGE_SIZE) | |
3492 | return; | |
3493 | ranges[2 + length] = 256; | |
3494 | length++; | |
3495 | } | |
3496 | ranges[0] = length; | |
3497 | } | |
3498 | ||
3499 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks) | |
3500 | { | |
3501 | int ranges[2 + MAX_RANGE_SIZE]; | |
3502 | pcre_uint8 bit, cbit, all; | |
3503 | int i, byte, length = 0; | |
3504 | ||
3505 | bit = bits[0] & 0x1; | |
3506 | ranges[1] = bit; | |
3507 | /* Can be 0 or 255. */ | |
3508 | all = -bit; | |
3509 | ||
3510 | for (i = 0; i < 256; ) | |
3511 | { | |
3512 | byte = i >> 3; | |
3513 | if ((i & 0x7) == 0 && bits[byte] == all) | |
3514 | i += 8; | |
3515 | else | |
3516 | { | |
3517 | cbit = (bits[byte] >> (i & 0x7)) & 0x1; | |
3518 | if (cbit != bit) | |
3519 | { | |
3520 | if (length >= MAX_RANGE_SIZE) | |
3521 | return FALSE; | |
3522 | ranges[2 + length] = i; | |
3523 | length++; | |
3524 | bit = cbit; | |
3525 | all = -cbit; | |
3526 | } | |
3527 | i++; | |
3528 | } | |
3529 | } | |
3530 | ||
3531 | if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) | |
3532 | { | |
3533 | if (length >= MAX_RANGE_SIZE) | |
3534 | return FALSE; | |
3535 | ranges[2 + length] = 256; | |
3536 | length++; | |
3537 | } | |
3538 | ranges[0] = length; | |
3539 | ||
3540 | return check_ranges(common, ranges, backtracks, FALSE); | |
3541 | } | |
3542 | ||
3543 | static void check_anynewline(compiler_common *common) | static void check_anynewline(compiler_common *common) |
3544 | { | { |
3545 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3546 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3547 | ||
3548 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3549 | ||
3550 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3551 | 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); |
3552 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3553 | 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); |
3554 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3555 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3556 | if (common->utf) | if (common->utf) |
3557 | { | { |
3558 | #endif | #endif |
3559 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3560 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3561 | 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); |
3562 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3563 | } | } |
3564 | #endif | #endif |
3565 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3566 | 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); |
3567 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3568 | } | } |
3569 | ||
# | Line 2294 static void check_hspace(compiler_common | Line 3572 static void check_hspace(compiler_common |
3572 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3573 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3574 | ||
3575 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3576 | ||
3577 | 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); |
3578 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3579 | 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); |
3580 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3581 | 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); |
3582 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3583 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3584 | if (common->utf) | if (common->utf) |
3585 | { | { |
3586 | #endif | #endif |
3587 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3588 | 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); |
3589 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3590 | 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); |
3591 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3592 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); |
3593 | 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); |
3594 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3595 | 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); |
3596 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3597 | 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); |
3598 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3599 | 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); |
3600 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3601 | } | } |
3602 | #endif | #endif |
3603 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3604 | 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); |
3605 | ||
3606 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3607 | } | } |
# | Line 2333 static void check_vspace(compiler_common | Line 3611 static void check_vspace(compiler_common |
3611 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
3612 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3613 | ||
3614 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3615 | ||
3616 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3617 | 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); |
3618 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3619 | 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); |
3620 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3621 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3622 | if (common->utf) | if (common->utf) |
3623 | { | { |
3624 | #endif | #endif |
3625 | 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); |
3626 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3627 | 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); |
3628 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3629 | } | } |
3630 | #endif | #endif |
3631 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3632 | 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); |
3633 | ||
3634 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3635 | } | } |
# | Line 2365 DEFINE_COMPILER; | Line 3643 DEFINE_COMPILER; |
3643 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3644 | struct sljit_label *label; | struct sljit_label *label; |
3645 | ||
3646 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3647 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
3648 | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); | OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); |
3649 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); |
# | Line 2394 DEFINE_COMPILER; | Line 3672 DEFINE_COMPILER; |
3672 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3673 | struct sljit_label *label; | struct sljit_label *label; |
3674 | ||
3675 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0, 1, 5, 5, common->localsize); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3676 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
3677 | ||
3678 | OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); | OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); |
# | Line 2437 sljit_emit_fast_return(compiler, RETURN_ | Line 3715 sljit_emit_fast_return(compiler, RETURN_ |
3715 | ||
3716 | #if defined SUPPORT_UTF && defined SUPPORT_UCP | #if defined SUPPORT_UTF && defined SUPPORT_UCP |
3717 | ||
3718 | 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) |
3719 | { | { |
3720 | /* This function would be ineffective to do in JIT level. */ | /* This function would be ineffective to do in JIT level. */ |
3721 | int c1, c2; | pcre_uint32 c1, c2; |
3722 | const pcre_uchar *src2 = args->ptr; | const pcre_uchar *src2 = args->uchar_ptr; |
3723 | const pcre_uchar *end2 = args->end; | const pcre_uchar *end2 = args->end; |
3724 | const ucd_record *ur; | |
3725 | const pcre_uint32 *pp; | |
3726 | ||
3727 | while (src1 < end1) | while (src1 < end1) |
3728 | { | { |
3729 | if (src2 >= end2) | if (src2 >= end2) |
3730 | return 0; | return (pcre_uchar*)1; |
3731 | GETCHARINC(c1, src1); | GETCHARINC(c1, src1); |
3732 | GETCHARINC(c2, src2); | GETCHARINC(c2, src2); |
3733 | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return 0; | ur = GET_UCD(c2); |
3734 | if (c1 != c2 && c1 != c2 + ur->other_case) | |
3735 | { | |
3736 | pp = PRIV(ucd_caseless_sets) + ur->caseset; | |
3737 | for (;;) | |
3738 | { | |
3739 | if (c1 < *pp) return NULL; | |
3740 | if (c1 == *pp++) break; | |
3741 | } | |
3742 | } | |
3743 | } | } |
3744 | return src2; | return src2; |
3745 | } | } |
# | Line 2458 return src2; | Line 3747 return src2; |
3747 | #endif /* SUPPORT_UTF && SUPPORT_UCP */ | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
3748 | ||
3749 | 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, |
3750 | compare_context* context, jump_list **fallbacks) | compare_context* context, jump_list **backtracks) |
3751 | { | { |
3752 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3753 | unsigned int othercasebit = 0; | unsigned int othercasebit = 0; |
# | Line 2472 if (caseless && char_has_othercase(commo | Line 3761 if (caseless && char_has_othercase(commo |
3761 | othercasebit = char_get_othercase_bit(common, cc); | othercasebit = char_get_othercase_bit(common, cc); |
3762 | SLJIT_ASSERT(othercasebit); | SLJIT_ASSERT(othercasebit); |
3763 | /* Extracting bit difference info. */ | /* Extracting bit difference info. */ |
3764 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3765 | othercasechar = cc + (othercasebit >> 8); | othercasechar = cc + (othercasebit >> 8); |
3766 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3767 | #else | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3768 | #ifdef COMPILE_PCRE16 | /* Note that this code only handles characters in the BMP. If there |
3769 | ever are characters outside the BMP whose othercase differs in only one | |
3770 | bit from itself (there currently are none), this code will need to be | |
3771 | revised for COMPILE_PCRE32. */ | |
3772 | othercasechar = cc + (othercasebit >> 9); | othercasechar = cc + (othercasebit >> 9); |
3773 | if ((othercasebit & 0x100) != 0) | if ((othercasebit & 0x100) != 0) |
3774 | othercasebit = (othercasebit & 0xff) << 8; | othercasebit = (othercasebit & 0xff) << 8; |
3775 | else | else |
3776 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3777 | #endif | #endif /* COMPILE_PCRE[8|16|32] */ |
#endif | ||
3778 | } | } |
3779 | ||
3780 | if (context->sourcereg == -1) | if (context->sourcereg == -1) |
3781 | { | { |
3782 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3783 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3784 | if (context->length >= 4) | if (context->length >= 4) |
3785 | 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 2497 if (context->sourcereg == -1) | Line 3788 if (context->sourcereg == -1) |
3788 | else | else |
3789 | #endif | #endif |
3790 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3791 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
3792 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3793 | if (context->length >= 4) | if (context->length >= 4) |
3794 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3795 | else | else |
3796 | #endif | #endif |
3797 | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3798 | #endif | #elif defined COMPILE_PCRE32 |
3799 | #endif /* COMPILE_PCRE8 */ | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3800 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3801 | context->sourcereg = TMP2; | context->sourcereg = TMP2; |
3802 | } | } |
3803 | ||
# | Line 2520 do | Line 3811 do |
3811 | #endif | #endif |
3812 | ||
3813 | context->length -= IN_UCHARS(1); | context->length -= IN_UCHARS(1); |
3814 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if (defined SLJIT_UNALIGNED && SLJIT_UNALIGNED) && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16) |
3815 | ||
3816 | /* Unaligned read is supported. */ | /* Unaligned read is supported. */ |
3817 | if (othercasebit != 0 && othercasechar == cc) | if (othercasebit != 0 && othercasechar == cc) |
# | Line 2535 do | Line 3826 do |
3826 | } | } |
3827 | context->ucharptr++; | context->ucharptr++; |
3828 | ||
3829 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3830 | 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)) |
3831 | #else | #else |
3832 | if (context->ucharptr >= 2 || context->length == 0) | if (context->ucharptr >= 2 || context->length == 0) |
# | Line 2543 do | Line 3834 do |
3834 | { | { |
3835 | if (context->length >= 4) | if (context->length >= 4) |
3836 | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
#ifdef COMPILE_PCRE8 | ||
3837 | else if (context->length >= 2) | else if (context->length >= 2) |
3838 | 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); |
3839 | #if defined COMPILE_PCRE8 | |
3840 | else if (context->length >= 1) | else if (context->length >= 1) |
3841 | 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); |
3842 | #else | #endif /* COMPILE_PCRE8 */ |
else if (context->length >= 2) | ||
OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | ||
#endif | ||
3843 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3844 | ||
3845 | switch(context->ucharptr) | switch(context->ucharptr) |
# | Line 2559 do | Line 3847 do |
3847 | case 4 / sizeof(pcre_uchar): | case 4 / sizeof(pcre_uchar): |
3848 | if (context->oc.asint != 0) | if (context->oc.asint != 0) |
3849 | 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); |
3850 | 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)); |
3851 | break; | break; |
3852 | ||
3853 | case 2 / sizeof(pcre_uchar): | case 2 / sizeof(pcre_uchar): |
3854 | if (context->oc.asushort != 0) | if (context->oc.asushort != 0) |
3855 | 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); |
3856 | 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)); |
3857 | break; | break; |
3858 | ||
3859 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |