ViewVC logotype

Contents of /code/trunk/doc/pcrejit.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 859 - (show annotations)
Mon Jan 9 17:43:54 2012 UTC (8 years, 1 month ago) by ph10
File size: 14577 byte(s)
Error occurred while calculating annotation data.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 Just-in-time compiling is a heavyweight optimization that can greatly speed up
8 pattern matching. However, it comes at the cost of extra processing before the
9 match is performed. Therefore, it is of most benefit when the same pattern is
10 going to be matched many times. This does not necessarily mean many calls of a
11 matching function; if the pattern is not anchored, matching attempts may take
12 place many times at various positions in the subject, even for a single call.
13 Therefore, if the subject string is very long, it may still pay to use JIT for
14 one-off matches.
15 .P
16 JIT support applies only to the traditional Perl-compatible matching function.
17 It does not apply when the DFA matching function is being used. The code for
18 this support was written by Zoltan Herczeg.
19 .
20 .
21 .SH "8-BIT and 16-BIT SUPPORT"
22 .rs
23 .sp
24 JIT support is available for both the 8-bit and 16-bit PCRE libraries. To keep
25 this documentation simple, only the 8-bit interface is described in what
26 follows. If you are using the 16-bit library, substitute the 16-bit functions
27 and 16-bit structures (for example, \fIpcre16_jit_stack\fP instead of
28 \fIpcre_jit_stack\fP).
29 .
30 .
32 .rs
33 .sp
34 JIT support is an optional feature of PCRE. The "configure" option --enable-jit
35 (or equivalent CMake option) must be set when PCRE is built if you want to use
36 JIT. The support is limited to the following hardware platforms:
37 .sp
38 ARM v5, v7, and Thumb2
39 Intel x86 32-bit and 64-bit
40 MIPS 32-bit
41 Power PC 32-bit and 64-bit
42 .sp
43 The Power PC support is designated as experimental because it has not been
44 fully tested. If --enable-jit is set on an unsupported platform, compilation
45 fails.
46 .P
47 A program that is linked with PCRE 8.20 or later can tell if JIT support is
48 available by calling \fBpcre_config()\fP with the PCRE_CONFIG_JIT option. The
49 result is 1 when JIT is available, and 0 otherwise. However, a simple program
50 does not need to check this in order to use JIT. The API is implemented in a
51 way that falls back to the ordinary PCRE code if JIT is not available.
52 .P
53 If your program may sometimes be linked with versions of PCRE that are older
54 than 8.20, but you want to use JIT when it is available, you can test
55 the values of PCRE_MAJOR and PCRE_MINOR, or the existence of a JIT macro such
56 as PCRE_CONFIG_JIT, for compile-time control of your code.
57 .
58 .
60 .rs
61 .sp
62 You have to do two things to make use of the JIT support in the simplest way:
63 .sp
64 (1) Call \fBpcre_study()\fP with the PCRE_STUDY_JIT_COMPILE option for
65 each compiled pattern, and pass the resulting \fBpcre_extra\fP block to
66 \fBpcre_exec()\fP.
67 .sp
68 (2) Use \fBpcre_free_study()\fP to free the \fBpcre_extra\fP block when it is
69 no longer needed instead of just freeing it yourself. This
70 ensures that any JIT data is also freed.
71 .sp
72 For a program that may be linked with pre-8.20 versions of PCRE, you can insert
73 .sp
76 #endif
77 .sp
78 so that no option is passed to \fBpcre_study()\fP, and then use something like
79 this to free the study data:
80 .sp
82 pcre_free_study(study_ptr);
83 #else
84 pcre_free(study_ptr);
85 #endif
86 .sp
87 In some circumstances you may need to call additional functions. These are
88 described in the section entitled
89 .\" HTML <a href="#stackcontrol">
90 .\" </a>
91 "Controlling the JIT stack"
92 .\"
93 below.
94 .P
95 If JIT support is not available, PCRE_STUDY_JIT_COMPILE is ignored, and no JIT
96 data is set up. Otherwise, the compiled pattern is passed to the JIT compiler,
97 which turns it into machine code that executes much faster than the normal
98 interpretive code. When \fBpcre_exec()\fP is passed a \fBpcre_extra\fP block
99 containing a pointer to JIT code, it obeys that instead of the normal code. The
100 result is identical, but the code runs much faster.
101 .P
102 There are some \fBpcre_exec()\fP options that are not supported for JIT
103 execution. There are also some pattern items that JIT cannot handle. Details
104 are given below. In both cases, execution automatically falls back to the
105 interpretive code.
106 .P
107 If the JIT compiler finds an unsupported item, no JIT data is generated. You
108 can find out if JIT execution is available after studying a pattern by calling
109 \fBpcre_fullinfo()\fP with the PCRE_INFO_JIT option. A result of 1 means that
110 JIT compilation was successful. A result of 0 means that JIT support is not
111 available, or the pattern was not studied with PCRE_STUDY_JIT_COMPILE, or the
112 JIT compiler was not able to handle the pattern.
113 .P
114 Once a pattern has been studied, with or without JIT, it can be used as many
115 times as you like for matching different subject strings.
116 .
117 .
119 .rs
120 .sp
121 The only \fBpcre_exec()\fP options that are supported for JIT execution are
123 PCRE_NOTEMPTY_ATSTART. Note in particular that partial matching is not
124 supported.
125 .P
126 The unsupported pattern items are:
127 .sp
128 \eC match a single byte; not supported in UTF-8 mode
129 (?Cn) callouts
130 (*COMMIT) )
131 (*MARK) )
132 (*PRUNE) ) the backtracking control verbs
133 (*SKIP) )
134 (*THEN) )
135 .sp
136 Support for some of these may be added in future.
137 .
138 .
140 .rs
141 .sp
142 When a pattern is matched using JIT execution, the return values are the same
143 as those given by the interpretive \fBpcre_exec()\fP code, with the addition of
144 one new error code: PCRE_ERROR_JIT_STACKLIMIT. This means that the memory used
145 for the JIT stack was insufficient. See
146 .\" HTML <a href="#stackcontrol">
147 .\" </a>
148 "Controlling the JIT stack"
149 .\"
150 below for a discussion of JIT stack usage. For compatibility with the
151 interpretive \fBpcre_exec()\fP code, no more than two-thirds of the
152 \fIovector\fP argument is used for passing back captured substrings.
153 .P
154 The error code PCRE_ERROR_MATCHLIMIT is returned by the JIT code if searching a
155 very large pattern tree goes on for too long, as it is in the same circumstance
156 when JIT is not used, but the details of exactly what is counted are not the
157 same. The PCRE_ERROR_RECURSIONLIMIT error code is never returned by JIT
158 execution.
159 .
160 .
162 .rs
163 .sp
164 The code that is generated by the JIT compiler is architecture-specific, and is
165 also position dependent. For those reasons it cannot be saved (in a file or
166 database) and restored later like the bytecode and other data of a compiled
167 pattern. Saving and restoring compiled patterns is not something many people
168 do. More detail about this facility is given in the
169 .\" HREF
170 \fBpcreprecompile\fP
171 .\"
172 documentation. It should be possible to run \fBpcre_study()\fP on a saved and
173 restored pattern, and thereby recreate the JIT data, but because JIT
174 compilation uses significant resources, it is probably not worth doing this;
175 you might as well recompile the original pattern.
176 .
177 .
178 .\" HTML <a name="stackcontrol"></a>
180 .rs
181 .sp
182 When the compiled JIT code runs, it needs a block of memory to use as a stack.
183 By default, it uses 32K on the machine stack. However, some large or
184 complicated patterns need more than this. The error PCRE_ERROR_JIT_STACKLIMIT
185 is given when there is not enough stack. Three functions are provided for
186 managing blocks of memory for use as JIT stacks. There is further discussion
187 about the use of JIT stacks in the section entitled
188 .\" HTML <a href="#stackcontrol">
189 .\" </a>
190 "JIT stack FAQ"
191 .\"
192 below.
193 .P
194 The \fBpcre_jit_stack_alloc()\fP function creates a JIT stack. Its arguments
195 are a starting size and a maximum size, and it returns a pointer to an opaque
196 structure of type \fBpcre_jit_stack\fP, or NULL if there is an error. The
197 \fBpcre_jit_stack_free()\fP function can be used to free a stack that is no
198 longer needed. (For the technically minded: the address space is allocated by
199 mmap or VirtualAlloc.)
200 .P
201 JIT uses far less memory for recursion than the interpretive code,
202 and a maximum stack size of 512K to 1M should be more than enough for any
203 pattern.
204 .P
205 The \fBpcre_assign_jit_stack()\fP function specifies which stack JIT code
206 should use. Its arguments are as follows:
207 .sp
208 pcre_extra *extra
209 pcre_jit_callback callback
210 void *data
211 .sp
212 The \fIextra\fP argument must be the result of studying a pattern with
213 PCRE_STUDY_JIT_COMPILE. There are three cases for the values of the other two
214 options:
215 .sp
216 (1) If \fIcallback\fP is NULL and \fIdata\fP is NULL, an internal 32K block
217 on the machine stack is used.
218 .sp
219 (2) If \fIcallback\fP is NULL and \fIdata\fP is not NULL, \fIdata\fP must be
220 a valid JIT stack, the result of calling \fBpcre_jit_stack_alloc()\fP.
221 .sp
222 (3) If \fIcallback\fP not NULL, it must point to a function that is called
223 with \fIdata\fP as an argument at the start of matching, in order to
224 set up a JIT stack. If the result is NULL, the internal 32K stack
225 is used; otherwise the return value must be a valid JIT stack,
226 the result of calling \fBpcre_jit_stack_alloc()\fP.
227 .sp
228 You may safely assign the same JIT stack to more than one pattern, as long as
229 they are all matched sequentially in the same thread. In a multithread
230 application, each thread must use its own JIT stack.
231 .P
232 Strictly speaking, even more is allowed. You can assign the same stack to any
233 number of patterns as long as they are not used for matching by multiple
234 threads at the same time. For example, you can assign the same stack to all
235 compiled patterns, and use a global mutex in the callback to wait until the
236 stack is available for use. However, this is an inefficient solution, and
237 not recommended.
238 .P
239 This is a suggestion for how a typical multithreaded program might operate:
240 .sp
241 During thread initalization
242 thread_local_var = pcre_jit_stack_alloc(...)
243 .sp
244 During thread exit
245 pcre_jit_stack_free(thread_local_var)
246 .sp
247 Use a one-line callback function
248 return thread_local_var
249 .sp
250 All the functions described in this section do nothing if JIT is not available,
251 and \fBpcre_assign_jit_stack()\fP does nothing unless the \fBextra\fP argument
252 is non-NULL and points to a \fBpcre_extra\fP block that is the result of a
253 successful study with PCRE_STUDY_JIT_COMPILE.
254 .
255 .
256 .\" HTML <a name="stackfaq"></a>
258 .rs
259 .sp
260 (1) Why do we need JIT stacks?
261 .sp
262 PCRE (and JIT) is a recursive, depth-first engine, so it needs a stack where
263 the local data of the current node is pushed before checking its child nodes.
264 Allocating real machine stack on some platforms is difficult. For example, the
265 stack chain needs to be updated every time if we extend the stack on PowerPC.
266 Although it is possible, its updating time overhead decreases performance. So
267 we do the recursion in memory.
268 .P
269 (2) Why don't we simply allocate blocks of memory with \fBmalloc()\fP?
270 .sp
271 Modern operating systems have a nice feature: they can reserve an address space
272 instead of allocating memory. We can safely allocate memory pages inside this
273 address space, so the stack could grow without moving memory data (this is
274 important because of pointers). Thus we can allocate 1M address space, and use
275 only a single memory page (usually 4K) if that is enough. However, we can still
276 grow up to 1M anytime if needed.
277 .P
278 (3) Who "owns" a JIT stack?
279 .sp
280 The owner of the stack is the user program, not the JIT studied pattern or
281 anything else. The user program must ensure that if a stack is used by
282 \fBpcre_exec()\fP, (that is, it is assigned to the pattern currently running),
283 that stack must not be used by any other threads (to avoid overwriting the same
284 memory area). The best practice for multithreaded programs is to allocate a
285 stack for each thread, and return this stack through the JIT callback function.
286 .P
287 (4) When should a JIT stack be freed?
288 .sp
289 You can free a JIT stack at any time, as long as it will not be used by
290 \fBpcre_exec()\fP again. When you assign the stack to a pattern, only a pointer
291 is set. There is no reference counting or any other magic. You can free the
292 patterns and stacks in any order, anytime. Just \fIdo not\fP call
293 \fBpcre_exec()\fP with a pattern pointing to an already freed stack, as that
294 will cause SEGFAULT. (Also, do not free a stack currently used by
295 \fBpcre_exec()\fP in another thread). You can also replace the stack for a
296 pattern at any time. You can even free the previous stack before assigning a
297 replacement.
298 .P
299 (5) Should I allocate/free a stack every time before/after calling
300 \fBpcre_exec()\fP?
301 .sp
302 No, because this is too costly in terms of resources. However, you could
303 implement some clever idea which release the stack if it is not used in let's
304 say two minutes. The JIT callback can help to achive this without keeping a
305 list of the currently JIT studied patterns.
306 .P
307 (6) OK, the stack is for long term memory allocation. But what happens if a
308 pattern causes stack overflow with a stack of 1M? Is that 1M kept until the
309 stack is freed?
310 .sp
311 Especially on embedded sytems, it might be a good idea to release
312 memory sometimes without freeing the stack. There is no API for this at the
313 moment. Probably a function call which returns with the currently allocated
314 memory for any stack and another which allows releasing memory (shrinking the
315 stack) would be a good idea if someone needs this.
316 .P
317 (7) This is too much of a headache. Isn't there any better solution for JIT
318 stack handling?
319 .sp
320 No, thanks to Windows. If POSIX threads were used everywhere, we could throw
321 out this complicated API.
322 .
323 .
325 .rs
326 .sp
327 This is a single-threaded example that specifies a JIT stack without using a
328 callback.
329 .sp
330 int rc;
331 int ovector[30];
332 pcre *re;
333 pcre_extra *extra;
334 pcre_jit_stack *jit_stack;
335 .sp
336 re = pcre_compile(pattern, 0, &error, &erroffset, NULL);
337 /* Check for errors */
338 extra = pcre_study(re, PCRE_STUDY_JIT_COMPILE, &error);
339 jit_stack = pcre_jit_stack_alloc(32*1024, 512*1024);
340 /* Check for error (NULL) */
341 pcre_assign_jit_stack(extra, NULL, jit_stack);
342 rc = pcre_exec(re, extra, subject, length, 0, 0, ovector, 30);
343 /* Check results */
344 pcre_free(re);
345 pcre_free_study(extra);
346 pcre_jit_stack_free(jit_stack);
347 .sp
348 .
349 .
350 .SH "SEE ALSO"
351 .rs
352 .sp
353 \fBpcreapi\fP(3)
354 .
355 .
357 .rs
358 .sp
359 .nf
360 Philip Hazel (FAQ by Zoltan Herczeg)
361 University Computing Service
362 Cambridge CB2 3QH, England.
363 .fi
364 .
365 .
367 .rs
368 .sp
369 .nf
370 Last updated: 08 January 2012
371 Copyright (c) 1997-2012 University of Cambridge.
372 .fi

  ViewVC Help
Powered by ViewVC 1.1.5