1PCRE2JIT(3) Library Functions Manual PCRE2JIT(3)
2
6 PCRE2 - Perl-compatible regular expressions (revised API)
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, so it is of most benefit when
13 the same pattern is going to be matched many times. This does not nec‐
14 essarily mean many calls of a matching function; if the pattern is not
15 anchored, matching attempts may take place many times at various posi‐
16 tions in the subject, even for a single call. Therefore, if the subject
17 string is very long, it may still pay to use JIT even for one-off
18 matches. JIT support is available for all of the 8-bit, 16-bit and
19 32-bit PCRE2 libraries.
20 JIT support applies only to the traditional Perl-compatible matching
22 function. It does not apply when the DFA matching function is being
23 used. The code for this support was written by Zoltan Herczeg.
24
26 JIT support is an optional feature of PCRE2. The "configure" option
28 --enable-jit (or equivalent CMake option) must be set when PCRE2 is
29 built if you want to use JIT. The support is limited to the following
30 hardware platforms:
31 ARM 32-bit (v5, v7, and Thumb2)
33 ARM 64-bit
34 IBM s390x 64 bit
35 Intel x86 32-bit and 64-bit
36 MIPS 32-bit and 64-bit
37 Power PC 32-bit and 64-bit
38 SPARC 32-bit
39 If --enable-jit is set on an unsupported platform, compilation fails.
41 A program can tell if JIT support is available by calling pcre2_con‐
43 fig() with the PCRE2_CONFIG_JIT option. The result is 1 when JIT is
44 available, and 0 otherwise. However, a simple program does not need to
45 check this in order to use JIT. The API is implemented in a way that
46 falls back to the interpretive code if JIT is not available. For pro‐
47 grams that need the best possible performance, there is also a "fast
48 path" API that is JIT-specific.
49
51 To make use of the JIT support in the simplest way, all you have to do
53 is to call pcre2_jit_compile() after successfully compiling a pattern
54 with pcre2_compile(). This function has two arguments: the first is the
55 compiled pattern pointer that was returned by pcre2_compile(), and the
56 second is zero or more of the following option bits: PCRE2_JIT_COM‐
57 PLETE, PCRE2_JIT_PARTIAL_HARD, or PCRE2_JIT_PARTIAL_SOFT.
58 If JIT support is not available, a call to pcre2_jit_compile() does
60 nothing and returns PCRE2_ERROR_JIT_BADOPTION. Otherwise, the compiled
61 pattern is passed to the JIT compiler, which turns it into machine code
62 that executes much faster than the normal interpretive code, but yields
63 exactly the same results. The returned value from pcre2_jit_compile()
64 is zero on success, or a negative error code.
65 There is a limit to the size of pattern that JIT supports, imposed by
67 the size of machine stack that it uses. The exact rules are not docu‐
68 mented because they may change at any time, in particular, when new op‐
69 timizations are introduced. If a pattern is too big, a call to
70 pcre2_jit_compile() returns PCRE2_ERROR_NOMEMORY.
71 PCRE2_JIT_COMPLETE requests the JIT compiler to generate code for com‐
73 plete matches. If you want to run partial matches using the PCRE2_PAR‐
74 TIAL_HARD or PCRE2_PARTIAL_SOFT options of pcre2_match(), you should
75 set one or both of the other options as well as, or instead of
76 PCRE2_JIT_COMPLETE. The JIT compiler generates different optimized code
77 for each of the three modes (normal, soft partial, hard partial). When
78 pcre2_match() is called, the appropriate code is run if it is avail‐
79 able. Otherwise, the pattern is matched using interpretive code.
80 You can call pcre2_jit_compile() multiple times for the same compiled
82 pattern. It does nothing if it has previously compiled code for any of
83 the option bits. For example, you can call it once with PCRE2_JIT_COM‐
84 PLETE and (perhaps later, when you find you need partial matching)
85 again with PCRE2_JIT_COMPLETE and PCRE2_JIT_PARTIAL_HARD. This time it
86 will ignore PCRE2_JIT_COMPLETE and just compile code for partial match‐
87 ing. If pcre2_jit_compile() is called with no option bits set, it imme‐
88 diately returns zero. This is an alternative way of testing whether JIT
89 is available.
90 At present, it is not possible to free JIT compiled code except when
92 the entire compiled pattern is freed by calling pcre2_code_free().
93 In some circumstances you may need to call additional functions. These
95 are described in the section entitled "Controlling the JIT stack" be‐
96 low.
97 There are some pcre2_match() options that are not supported by JIT, and
99 there are also some pattern items that JIT cannot handle. Details are
100 given below. In both cases, matching automatically falls back to the
101 interpretive code. If you want to know whether JIT was actually used
102 for a particular match, you should arrange for a JIT callback function
103 to be set up as described in the section entitled "Controlling the JIT
104 stack" below, even if you do not need to supply a non-default JIT
105 stack. Such a callback function is called whenever JIT code is about to
106 be obeyed. If the match-time options are not right for JIT execution,
107 the callback function is not obeyed.
108 If the JIT compiler finds an unsupported item, no JIT data is gener‐
110 ated. You can find out if JIT matching is available after compiling a
111 pattern by calling pcre2_pattern_info() with the PCRE2_INFO_JITSIZE op‐
112 tion. A non-zero result means that JIT compilation was successful. A
113 result of 0 means that JIT support is not available, or the pattern was
114 not processed by pcre2_jit_compile(), or the JIT compiler was not able
115 to handle the pattern.
116
118 When a pattern is compiled with the PCRE2_UTF option, subject strings
120 are normally expected to be a valid sequence of UTF code units. By de‐
121 fault, this is checked at the start of matching and an error is gener‐
122 ated if invalid UTF is detected. The PCRE2_NO_UTF_CHECK option can be
123 passed to pcre2_match() to skip the check (for improved performance) if
124 you are sure that a subject string is valid. If this option is used
125 with an invalid string, the result is undefined.
126 However, a way of running matches on strings that may contain invalid
128 UTF sequences is available. Calling pcre2_compile() with the
129 PCRE2_MATCH_INVALID_UTF option has two effects: it tells the inter‐
130 preter in pcre2_match() to support invalid UTF, and, if pcre2_jit_com‐
131 pile() is called, the compiled JIT code also supports invalid UTF. De‐
132 tails of how this support works, in both the JIT and the interpretive
133 cases, is given in the pcre2unicode documentation.
134 There is also an obsolete option for pcre2_jit_compile() called
136 PCRE2_JIT_INVALID_UTF, which currently exists only for backward compat‐
137 ibility. It is superseded by the pcre2_compile() option
138 PCRE2_MATCH_INVALID_UTF and should no longer be used. It may be removed
139 in future.
140
142 The pcre2_match() options that are supported for JIT matching are
144 PCRE2_COPY_MATCHED_SUBJECT, PCRE2_NOTBOL, PCRE2_NOTEOL, PCRE2_NOTEMPTY,
145 PCRE2_NOTEMPTY_ATSTART, PCRE2_NO_UTF_CHECK, PCRE2_PARTIAL_HARD, and
146 PCRE2_PARTIAL_SOFT. The PCRE2_ANCHORED and PCRE2_ENDANCHORED options
147 are not supported at match time.
148 If the PCRE2_NO_JIT option is passed to pcre2_match() it disables the
150 use of JIT, forcing matching by the interpreter code.
151 The only unsupported pattern items are \C (match a single data unit)
153 when running in a UTF mode, and a callout immediately before an asser‐
154 tion condition in a conditional group.
155
157 When a pattern is matched using JIT matching, the return values are the
159 same as those given by the interpretive pcre2_match() code, with the
160 addition of one new error code: PCRE2_ERROR_JIT_STACKLIMIT. This means
161 that the memory used for the JIT stack was insufficient. See "Control‐
162 ling the JIT stack" below for a discussion of JIT stack usage.
163 The error code PCRE2_ERROR_MATCHLIMIT is returned by the JIT code if
165 searching a very large pattern tree goes on for too long, as it is in
166 the same circumstance when JIT is not used, but the details of exactly
167 what is counted are not the same. The PCRE2_ERROR_DEPTHLIMIT error code
168 is never returned when JIT matching is used.
169
171 When the compiled JIT code runs, it needs a block of memory to use as a
173 stack. By default, it uses 32KiB on the machine stack. However, some
174 large or complicated patterns need more than this. The error PCRE2_ER‐
175 ROR_JIT_STACKLIMIT is given when there is not enough stack. Three func‐
176 tions are provided for managing blocks of memory for use as JIT stacks.
177 There is further discussion about the use of JIT stacks in the section
178 entitled "JIT stack FAQ" below.
179 The pcre2_jit_stack_create() function creates a JIT stack. Its argu‐
181 ments are a starting size, a maximum size, and a general context (for
182 memory allocation functions, or NULL for standard memory allocation).
183 It returns a pointer to an opaque structure of type pcre2_jit_stack, or
184 NULL if there is an error. The pcre2_jit_stack_free() function is used
185 to free a stack that is no longer needed. If its argument is NULL, this
186 function returns immediately, without doing anything. (For the techni‐
187 cally minded: the address space is allocated by mmap or VirtualAlloc.)
188 A maximum stack size of 512KiB to 1MiB should be more than enough for
189 any pattern.
190 The pcre2_jit_stack_assign() function specifies which stack JIT code
192 should use. Its arguments are as follows:
193 pcre2_match_context *mcontext
195 pcre2_jit_callback callback
196 void *data
197 The first argument is a pointer to a match context. When this is subse‐
199 quently passed to a matching function, its information determines which
200 JIT stack is used. If this argument is NULL, the function returns imme‐
201 diately, without doing anything. There are three cases for the values
202 of the other two options:
203 (1) If callback is NULL and data is NULL, an internal 32KiB block
205 on the machine stack is used. This is the default when a match
206 context is created.
207 (2) If callback is NULL and data is not NULL, data must be
209 a pointer to a valid JIT stack, the result of calling
210 pcre2_jit_stack_create().
211 (3) If callback is not NULL, it must point to a function that is
213 called with data as an argument at the start of matching, in
214 order to set up a JIT stack. If the return from the callback
215 function is NULL, the internal 32KiB stack is used; otherwise the
216 return value must be a valid JIT stack, the result of calling
217 pcre2_jit_stack_create().
218 A callback function is obeyed whenever JIT code is about to be run; it
220 is not obeyed when pcre2_match() is called with options that are incom‐
221 patible for JIT matching. A callback function can therefore be used to
222 determine whether a match operation was executed by JIT or by the in‐
223 terpreter.
224 You may safely use the same JIT stack for more than one pattern (either
226 by assigning directly or by callback), as long as the patterns are
227 matched sequentially in the same thread. Currently, the only way to set
228 up non-sequential matches in one thread is to use callouts: if a call‐
229 out function starts another match, that match must use a different JIT
230 stack to the one used for currently suspended match(es).
231 In a multithread application, if you do not specify a JIT stack, or if
233 you assign or pass back NULL from a callback, that is thread-safe, be‐
234 cause each thread has its own machine stack. However, if you assign or
235 pass back a non-NULL JIT stack, this must be a different stack for each
236 thread so that the application is thread-safe.
237 Strictly speaking, even more is allowed. You can assign the same non-
239 NULL stack to a match context that is used by any number of patterns,
240 as long as they are not used for matching by multiple threads at the
241 same time. For example, you could use the same stack in all compiled
242 patterns, with a global mutex in the callback to wait until the stack
243 is available for use. However, this is an inefficient solution, and not
244 recommended.
245 This is a suggestion for how a multithreaded program that needs to set
247 up non-default JIT stacks might operate:
248 During thread initialization
250 thread_local_var = pcre2_jit_stack_create(...)
251 During thread exit
253 pcre2_jit_stack_free(thread_local_var)
254 Use a one-line callback function
256 return thread_local_var
257 All the functions described in this section do nothing if JIT is not
259 available.
260
262 (1) Why do we need JIT stacks?
264 PCRE2 (and JIT) is a recursive, depth-first engine, so it needs a stack
266 where the local data of the current node is pushed before checking its
267 child nodes. Allocating real machine stack on some platforms is diffi‐
268 cult. For example, the stack chain needs to be updated every time if we
269 extend the stack on PowerPC. Although it is possible, its updating
270 time overhead decreases performance. So we do the recursion in memory.
271 (2) Why don't we simply allocate blocks of memory with malloc()?
273 Modern operating systems have a nice feature: they can reserve an ad‐
275 dress space instead of allocating memory. We can safely allocate memory
276 pages inside this address space, so the stack could grow without moving
277 memory data (this is important because of pointers). Thus we can allo‐
278 cate 1MiB address space, and use only a single memory page (usually
279 4KiB) if that is enough. However, we can still grow up to 1MiB anytime
280 if needed.
281 (3) Who "owns" a JIT stack?
283 The owner of the stack is the user program, not the JIT studied pattern
285 or anything else. The user program must ensure that if a stack is being
286 used by pcre2_match(), (that is, it is assigned to a match context that
287 is passed to the pattern currently running), that stack must not be
288 used by any other threads (to avoid overwriting the same memory area).
289 The best practice for multithreaded programs is to allocate a stack for
290 each thread, and return this stack through the JIT callback function.
291 (4) When should a JIT stack be freed?
293 You can free a JIT stack at any time, as long as it will not be used by
295 pcre2_match() again. When you assign the stack to a match context, only
296 a pointer is set. There is no reference counting or any other magic.
297 You can free compiled patterns, contexts, and stacks in any order, any‐
298 time. Just do not call pcre2_match() with a match context pointing to
299 an already freed stack, as that will cause SEGFAULT. (Also, do not free
300 a stack currently used by pcre2_match() in another thread). You can
301 also replace the stack in a context at any time when it is not in use.
302 You should free the previous stack before assigning a replacement.
303 (5) Should I allocate/free a stack every time before/after calling
305 pcre2_match()?
306 No, because this is too costly in terms of resources. However, you
308 could implement some clever idea which release the stack if it is not
309 used in let's say two minutes. The JIT callback can help to achieve
310 this without keeping a list of patterns.
311 (6) OK, the stack is for long term memory allocation. But what happens
313 if a pattern causes stack overflow with a stack of 1MiB? Is that 1MiB
314 kept until the stack is freed?
315 Especially on embedded sytems, it might be a good idea to release mem‐
317 ory sometimes without freeing the stack. There is no API for this at
318 the moment. Probably a function call which returns with the currently
319 allocated memory for any stack and another which allows releasing mem‐
320 ory (shrinking the stack) would be a good idea if someone needs this.
321 (7) This is too much of a headache. Isn't there any better solution for
323 JIT stack handling?
324 No, thanks to Windows. If POSIX threads were used everywhere, we could
326 throw out this complicated API.
327
329 void pcre2_jit_free_unused_memory(pcre2_general_context *gcontext);
331 The JIT executable allocator does not free all memory when it is possi‐
333 ble. It expects new allocations, and keeps some free memory around to
334 improve allocation speed. However, in low memory conditions, it might
335 be better to free all possible memory. You can cause this to happen by
336 calling pcre2_jit_free_unused_memory(). Its argument is a general con‐
337 text, for custom memory management, or NULL for standard memory manage‐
338 ment.
339
341 This is a single-threaded example that specifies a JIT stack without
343 using a callback. A real program should include error checking after
344 all the function calls.
345 int rc;
347 pcre2_code *re;
348 pcre2_match_data *match_data;
349 pcre2_match_context *mcontext;
350 pcre2_jit_stack *jit_stack;
351 re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0,
353 &errornumber, &erroffset, NULL);
354 rc = pcre2_jit_compile(re, PCRE2_JIT_COMPLETE);
355 mcontext = pcre2_match_context_create(NULL);
356 jit_stack = pcre2_jit_stack_create(32*1024, 512*1024, NULL);
357 pcre2_jit_stack_assign(mcontext, NULL, jit_stack);
358 match_data = pcre2_match_data_create(re, 10);
359 rc = pcre2_match(re, subject, length, 0, 0, match_data, mcontext);
360 /* Process result */
361 pcre2_code_free(re);
363 pcre2_match_data_free(match_data);
364 pcre2_match_context_free(mcontext);
365 pcre2_jit_stack_free(jit_stack);
366
369 Because the API described above falls back to interpreted matching when
371 JIT is not available, it is convenient for programs that are written
372 for general use in many environments. However, calling JIT via
373 pcre2_match() does have a performance impact. Programs that are written
374 for use where JIT is known to be available, and which need the best
375 possible performance, can instead use a "fast path" API to call JIT
376 matching directly instead of calling pcre2_match() (obviously only for
377 patterns that have been successfully processed by pcre2_jit_compile()).
378 The fast path function is called pcre2_jit_match(), and it takes ex‐
380 actly the same arguments as pcre2_match(). However, the subject string
381 must be specified with a length; PCRE2_ZERO_TERMINATED is not sup‐
382 ported. Unsupported option bits (for example, PCRE2_ANCHORED, PCRE2_EN‐
383 DANCHORED and PCRE2_COPY_MATCHED_SUBJECT) are ignored, as is the
384 PCRE2_NO_JIT option. The return values are also the same as for
385 pcre2_match(), plus PCRE2_ERROR_JIT_BADOPTION if a matching mode (par‐
386 tial or complete) is requested that was not compiled.
387 When you call pcre2_match(), as well as testing for invalid options, a
389 number of other sanity checks are performed on the arguments. For exam‐
390 ple, if the subject pointer is NULL but the length is non-zero, an im‐
391 mediate error is given. Also, unless PCRE2_NO_UTF_CHECK is set, a UTF
392 subject string is tested for validity. In the interests of speed, these
393 checks do not happen on the JIT fast path, and if invalid data is
394 passed, the result is undefined.
395 Bypassing the sanity checks and the pcre2_match() wrapping can give
397 speedups of more than 10%.
398
400 pcre2api(3)
402
404 Philip Hazel (FAQ by Zoltan Herczeg)
406 University Computing Service
407 Cambridge, England.
408
410 Last updated: 30 November 2021
412 Copyright (c) 1997-2021 University of Cambridge.
413PCRE2 10.40 30 November 2021 PCRE2JIT(3)