1PCREJIT(3) Library Functions Manual PCREJIT(3)
2
6 PCRE - Perl-compatible regular expressions
7
9 Just-in-time compiling is a heavyweight optimization that can greatly
11 speed up pattern matching. However, it comes at the cost of extra pro‐
12 cessing before the match is performed. Therefore, it is of most benefit
13 when the same pattern is going to be matched many times. This does not
14 necessarily mean many calls of a matching function; if the pattern is
15 not anchored, matching attempts may take place many times at various
16 positions in the subject, even for a single call. Therefore, if the
17 subject string is very long, it may still pay to use JIT for one-off
18 matches.
19 JIT support applies only to the traditional Perl-compatible matching
21 function. It does not apply when the DFA matching function is being
22 used. The code for this support was written by Zoltan Herczeg.
23
25 JIT support is available for all of the 8-bit, 16-bit and 32-bit PCRE
27 libraries. To keep this documentation simple, only the 8-bit interface
28 is described in what follows. If you are using the 16-bit library, sub‐
29 stitute the 16-bit functions and 16-bit structures (for example,
30 pcre16_jit_stack instead of pcre_jit_stack). If you are using the
31 32-bit library, substitute the 32-bit functions and 32-bit structures
32 (for example, pcre32_jit_stack instead of pcre_jit_stack).
33
35 JIT support is an optional feature of PCRE. The "configure" option
37 --enable-jit (or equivalent CMake option) must be set when PCRE is
38 built if you want to use JIT. The support is limited to the following
39 hardware platforms:
40 ARM v5, v7, and Thumb2
42 Intel x86 32-bit and 64-bit
43 MIPS 32-bit
44 Power PC 32-bit and 64-bit
45 SPARC 32-bit (experimental)
46 If --enable-jit is set on an unsupported platform, compilation fails.
48 A program that is linked with PCRE 8.20 or later can tell if JIT sup‐
50 port is available by calling pcre_config() with the PCRE_CONFIG_JIT
51 option. The result is 1 when JIT is available, and 0 otherwise. How‐
52 ever, a simple program does not need to check this in order to use JIT.
53 The normal API is implemented in a way that falls back to the interpre‐
54 tive code if JIT is not available. For programs that need the best pos‐
55 sible performance, there is also a "fast path" API that is JIT-spe‐
56 cific.
57 If your program may sometimes be linked with versions of PCRE that are
59 older than 8.20, but you want to use JIT when it is available, you can
60 test the values of PCRE_MAJOR and PCRE_MINOR, or the existence of a JIT
61 macro such as PCRE_CONFIG_JIT, for compile-time control of your code.
62 Also beware that the pcre_jit_exec() function was not available at all
63 before 8.32, and may not be available at all if PCRE isn't compiled
64 with --enable-jit. See the "JIT FAST PATH API" section below for
65 details.
66
68 You have to do two things to make use of the JIT support in the sim‐
70 plest way:
71 (1) Call pcre_study() with the PCRE_STUDY_JIT_COMPILE option for
73 each compiled pattern, and pass the resulting pcre_extra block to
74 pcre_exec().
75 (2) Use pcre_free_study() to free the pcre_extra block when it is
77 no longer needed, instead of just freeing it yourself. This
78 ensures that
79 any JIT data is also freed.
80 For a program that may be linked with pre-8.20 versions of PCRE, you
82 can insert
83 #ifndef PCRE_STUDY_JIT_COMPILE
85 #define PCRE_STUDY_JIT_COMPILE 0
86 #endif
87 so that no option is passed to pcre_study(), and then use something
89 like this to free the study data:
90 #ifdef PCRE_CONFIG_JIT
92 pcre_free_study(study_ptr);
93 #else
94 pcre_free(study_ptr);
95 #endif
96 PCRE_STUDY_JIT_COMPILE requests the JIT compiler to generate code for
98 complete matches. If you want to run partial matches using the
99 PCRE_PARTIAL_HARD or PCRE_PARTIAL_SOFT options of pcre_exec(), you
100 should set one or both of the following options in addition to, or
101 instead of, PCRE_STUDY_JIT_COMPILE when you call pcre_study():
102 PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
104 PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
105 If using pcre_jit_exec() and supporting a pre-8.32 version of PCRE, you
107 can insert:
108 #if PCRE_MAJOR >= 8 && PCRE_MINOR >= 32
110 pcre_jit_exec(...);
111 #else
112 pcre_exec(...)
113 #endif
114 but as described in the "JIT FAST PATH API" section below this assumes
116 version 8.32 and later are compiled with --enable-jit, which may break.
117 The JIT compiler generates different optimized code for each of the
119 three modes (normal, soft partial, hard partial). When pcre_exec() is
120 called, the appropriate code is run if it is available. Otherwise, the
121 pattern is matched using interpretive code.
122 In some circumstances you may need to call additional functions. These
124 are described in the section entitled "Controlling the JIT stack"
125 below.
126 If JIT support is not available, PCRE_STUDY_JIT_COMPILE etc. are
128 ignored, and no JIT data is created. Otherwise, the compiled pattern is
129 passed to the JIT compiler, which turns it into machine code that exe‐
130 cutes much faster than the normal interpretive code. When pcre_exec()
131 is passed a pcre_extra block containing a pointer to JIT code of the
132 appropriate mode (normal or hard/soft partial), it obeys that code
133 instead of running the interpreter. The result is identical, but the
134 compiled JIT code runs much faster.
135 There are some pcre_exec() options that are not supported for JIT exe‐
137 cution. There are also some pattern items that JIT cannot handle.
138 Details are given below. In both cases, execution automatically falls
139 back to the interpretive code. If you want to know whether JIT was
140 actually used for a particular match, you should arrange for a JIT
141 callback function to be set up as described in the section entitled
142 "Controlling the JIT stack" below, even if you do not need to supply a
143 non-default JIT stack. Such a callback function is called whenever JIT
144 code is about to be obeyed. If the execution options are not right for
145 JIT execution, the callback function is not obeyed.
146 If the JIT compiler finds an unsupported item, no JIT data is gener‐
148 ated. You can find out if JIT execution is available after studying a
149 pattern by calling pcre_fullinfo() with the PCRE_INFO_JIT option. A
150 result of 1 means that JIT compilation was successful. A result of 0
151 means that JIT support is not available, or the pattern was not studied
152 with PCRE_STUDY_JIT_COMPILE etc., or the JIT compiler was not able to
153 handle the pattern.
154 Once a pattern has been studied, with or without JIT, it can be used as
156 many times as you like for matching different subject strings.
157
159 The only pcre_exec() options that are supported for JIT execution are
161 PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK, PCRE_NO_UTF32_CHECK, PCRE_NOT‐
162 BOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART, PCRE_PAR‐
163 TIAL_HARD, and PCRE_PARTIAL_SOFT.
164 The only unsupported pattern items are \C (match a single data unit)
166 when running in a UTF mode, and a callout immediately before an asser‐
167 tion condition in a conditional group.
168
170 When a pattern is matched using JIT execution, the return values are
172 the same as those given by the interpretive pcre_exec() code, with the
173 addition of one new error code: PCRE_ERROR_JIT_STACKLIMIT. This means
174 that the memory used for the JIT stack was insufficient. See "Control‐
175 ling the JIT stack" below for a discussion of JIT stack usage. For com‐
176 patibility with the interpretive pcre_exec() code, no more than two-
177 thirds of the ovector argument is used for passing back captured sub‐
178 strings.
179 The error code PCRE_ERROR_MATCHLIMIT is returned by the JIT code if
181 searching a very large pattern tree goes on for too long, as it is in
182 the same circumstance when JIT is not used, but the details of exactly
183 what is counted are not the same. The PCRE_ERROR_RECURSIONLIMIT error
184 code is never returned by JIT execution.
185
187 The code that is generated by the JIT compiler is architecture-spe‐
189 cific, and is also position dependent. For those reasons it cannot be
190 saved (in a file or database) and restored later like the bytecode and
191 other data of a compiled pattern. Saving and restoring compiled pat‐
192 terns is not something many people do. More detail about this facility
193 is given in the pcreprecompile documentation. It should be possible to
194 run pcre_study() on a saved and restored pattern, and thereby recreate
195 the JIT data, but because JIT compilation uses significant resources,
196 it is probably not worth doing this; you might as well recompile the
197 original pattern.
198
200 When the compiled JIT code runs, it needs a block of memory to use as a
202 stack. By default, it uses 32K on the machine stack. However, some
203 large or complicated patterns need more than this. The error
204 PCRE_ERROR_JIT_STACKLIMIT is given when there is not enough stack.
205 Three functions are provided for managing blocks of memory for use as
206 JIT stacks. There is further discussion about the use of JIT stacks in
207 the section entitled "JIT stack FAQ" below.
208 The pcre_jit_stack_alloc() function creates a JIT stack. Its arguments
210 are a starting size and a maximum size, and it returns a pointer to an
211 opaque structure of type pcre_jit_stack, or NULL if there is an error.
212 The pcre_jit_stack_free() function can be used to free a stack that is
213 no longer needed. (For the technically minded: the address space is
214 allocated by mmap or VirtualAlloc.)
215 JIT uses far less memory for recursion than the interpretive code, and
217 a maximum stack size of 512K to 1M should be more than enough for any
218 pattern.
219 The pcre_assign_jit_stack() function specifies which stack JIT code
221 should use. Its arguments are as follows:
222 pcre_extra *extra
224 pcre_jit_callback callback
225 void *data
226 The extra argument must be the result of studying a pattern with
228 PCRE_STUDY_JIT_COMPILE etc. There are three cases for the values of the
229 other two options:
230 (1) If callback is NULL and data is NULL, an internal 32K block
232 on the machine stack is used.
233 (2) If callback is NULL and data is not NULL, data must be
235 a valid JIT stack, the result of calling pcre_jit_stack_alloc().
236 (3) If callback is not NULL, it must point to a function that is
238 called with data as an argument at the start of matching, in
239 order to set up a JIT stack. If the return from the callback
240 function is NULL, the internal 32K stack is used; otherwise the
241 return value must be a valid JIT stack, the result of calling
242 pcre_jit_stack_alloc().
243 A callback function is obeyed whenever JIT code is about to be run; it
245 is not obeyed when pcre_exec() is called with options that are incom‐
246 patible for JIT execution. A callback function can therefore be used to
247 determine whether a match operation was executed by JIT or by the
248 interpreter.
249 You may safely use the same JIT stack for more than one pattern (either
251 by assigning directly or by callback), as long as the patterns are all
252 matched sequentially in the same thread. In a multithread application,
253 if you do not specify a JIT stack, or if you assign or pass back NULL
254 from a callback, that is thread-safe, because each thread has its own
255 machine stack. However, if you assign or pass back a non-NULL JIT
256 stack, this must be a different stack for each thread so that the
257 application is thread-safe.
258 Strictly speaking, even more is allowed. You can assign the same non-
260 NULL stack to any number of patterns as long as they are not used for
261 matching by multiple threads at the same time. For example, you can
262 assign the same stack to all compiled patterns, and use a global mutex
263 in the callback to wait until the stack is available for use. However,
264 this is an inefficient solution, and not recommended.
265 This is a suggestion for how a multithreaded program that needs to set
267 up non-default JIT stacks might operate:
268 During thread initalization
270 thread_local_var = pcre_jit_stack_alloc(...)
271 During thread exit
273 pcre_jit_stack_free(thread_local_var)
274 Use a one-line callback function
276 return thread_local_var
277 All the functions described in this section do nothing if JIT is not
279 available, and pcre_assign_jit_stack() does nothing unless the extra
280 argument is non-NULL and points to a pcre_extra block that is the
281 result of a successful study with PCRE_STUDY_JIT_COMPILE etc.
282
284 (1) Why do we need JIT stacks?
286 PCRE (and JIT) is a recursive, depth-first engine, so it needs a stack
288 where the local data of the current node is pushed before checking its
289 child nodes. Allocating real machine stack on some platforms is diffi‐
290 cult. For example, the stack chain needs to be updated every time if we
291 extend the stack on PowerPC. Although it is possible, its updating
292 time overhead decreases performance. So we do the recursion in memory.
293 (2) Why don't we simply allocate blocks of memory with malloc()?
295 Modern operating systems have a nice feature: they can reserve an
297 address space instead of allocating memory. We can safely allocate mem‐
298 ory pages inside this address space, so the stack could grow without
299 moving memory data (this is important because of pointers). Thus we can
300 allocate 1M address space, and use only a single memory page (usually
301 4K) if that is enough. However, we can still grow up to 1M anytime if
302 needed.
303 (3) Who "owns" a JIT stack?
305 The owner of the stack is the user program, not the JIT studied pattern
307 or anything else. The user program must ensure that if a stack is used
308 by pcre_exec(), (that is, it is assigned to the pattern currently run‐
309 ning), that stack must not be used by any other threads (to avoid over‐
310 writing the same memory area). The best practice for multithreaded pro‐
311 grams is to allocate a stack for each thread, and return this stack
312 through the JIT callback function.
313 (4) When should a JIT stack be freed?
315 You can free a JIT stack at any time, as long as it will not be used by
317 pcre_exec() again. When you assign the stack to a pattern, only a
318 pointer is set. There is no reference counting or any other magic. You
319 can free the patterns and stacks in any order, anytime. Just do not
320 call pcre_exec() with a pattern pointing to an already freed stack, as
321 that will cause SEGFAULT. (Also, do not free a stack currently used by
322 pcre_exec() in another thread). You can also replace the stack for a
323 pattern at any time. You can even free the previous stack before
324 assigning a replacement.
325 (5) Should I allocate/free a stack every time before/after calling
327 pcre_exec()?
328 No, because this is too costly in terms of resources. However, you
330 could implement some clever idea which release the stack if it is not
331 used in let's say two minutes. The JIT callback can help to achieve
332 this without keeping a list of the currently JIT studied patterns.
333 (6) OK, the stack is for long term memory allocation. But what happens
335 if a pattern causes stack overflow with a stack of 1M? Is that 1M kept
336 until the stack is freed?
337 Especially on embedded sytems, it might be a good idea to release mem‐
339 ory sometimes without freeing the stack. There is no API for this at
340 the moment. Probably a function call which returns with the currently
341 allocated memory for any stack and another which allows releasing mem‐
342 ory (shrinking the stack) would be a good idea if someone needs this.
343 (7) This is too much of a headache. Isn't there any better solution for
345 JIT stack handling?
346 No, thanks to Windows. If POSIX threads were used everywhere, we could
348 throw out this complicated API.
349
351 This is a single-threaded example that specifies a JIT stack without
353 using a callback.
354 int rc;
356 int ovector[30];
357 pcre *re;
358 pcre_extra *extra;
359 pcre_jit_stack *jit_stack;
360 re = pcre_compile(pattern, 0, &error, &erroffset, NULL);
362 /* Check for errors */
363 extra = pcre_study(re, PCRE_STUDY_JIT_COMPILE, &error);
364 jit_stack = pcre_jit_stack_alloc(32*1024, 512*1024);
365 /* Check for error (NULL) */
366 pcre_assign_jit_stack(extra, NULL, jit_stack);
367 rc = pcre_exec(re, extra, subject, length, 0, 0, ovector, 30);
368 /* Check results */
369 pcre_free(re);
370 pcre_free_study(extra);
371 pcre_jit_stack_free(jit_stack);
372
375 Because the API described above falls back to interpreted execution
377 when JIT is not available, it is convenient for programs that are writ‐
378 ten for general use in many environments. However, calling JIT via
379 pcre_exec() does have a performance impact. Programs that are written
380 for use where JIT is known to be available, and which need the best
381 possible performance, can instead use a "fast path" API to call JIT
382 execution directly instead of calling pcre_exec() (obviously only for
383 patterns that have been successfully studied by JIT).
384 The fast path function is called pcre_jit_exec(), and it takes exactly
386 the same arguments as pcre_exec(), plus one additional argument that
387 must point to a JIT stack. The JIT stack arrangements described above
388 do not apply. The return values are the same as for pcre_exec().
389 When you call pcre_exec(), as well as testing for invalid options, a
391 number of other sanity checks are performed on the arguments. For exam‐
392 ple, if the subject pointer is NULL, or its length is negative, an
393 immediate error is given. Also, unless PCRE_NO_UTF[8|16|32] is set, a
394 UTF subject string is tested for validity. In the interests of speed,
395 these checks do not happen on the JIT fast path, and if invalid data is
396 passed, the result is undefined.
397 Bypassing the sanity checks and the pcre_exec() wrapping can give
399 speedups of more than 10%.
400 Note that the pcre_jit_exec() function is not available in versions of
402 PCRE before 8.32 (released in November 2012). If you need to support
403 versions that old you must either use the slower pcre_exec(), or switch
404 between the two codepaths by checking the values of PCRE_MAJOR and
405 PCRE_MINOR.
406 Due to an unfortunate implementation oversight, even in versions 8.32
408 and later there will be no pcre_jit_exec() stub function defined when
409 PCRE is compiled with --disable-jit, which is the default, and there's
410 no way to detect whether PCRE was compiled with --enable-jit via a
411 macro.
412 If you need to support versions older than 8.32, or versions that may
414 not build with --enable-jit, you must either use the slower
415 pcre_exec(), or switch between the two codepaths by checking the values
416 of PCRE_MAJOR and PCRE_MINOR.
417 Switching between the two by checking the version assumes that all the
419 versions being targeted are built with --enable-jit. To also support
420 builds that may use --disable-jit either pcre_exec() must be used, or a
421 compile-time check for JIT via pcre_config() (which assumes the runtime
422 environment will be the same), or as the Git project decided to do,
423 simply assume that pcre_jit_exec() is present in 8.32 or later unless a
424 compile-time flag is provided, see the "grep: un-break building with
425 PCRE >= 8.32 without --enable-jit" commit in git.git for an example of
426 that.
427
429 pcreapi(3)
431
433 Philip Hazel (FAQ by Zoltan Herczeg)
435 University Computing Service
436 Cambridge CB2 3QH, England.
437
439 Last updated: 05 July 2017
441 Copyright (c) 1997-2017 University of Cambridge.
442PCRE 8.41 05 July 2017 PCREJIT(3)