/[pcre]/code/trunk/doc/pcrejit.3
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 691 by ph10, Sun Sep 11 14:31:21 2011 UTC revision 926 by ph10, Wed Feb 22 15:01:32 2012 UTC
# Line 7  PCRE - Perl-compatible regular expressio Line 7  PCRE - Perl-compatible regular expressio
7  Just-in-time compiling is a heavyweight optimization that can greatly speed up  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  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  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  going to be matched many times. This does not necessarily mean many calls of a
11  \fPpcre_exec()\fP; if the pattern is not anchored, matching attempts may take  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 to  place many times at various positions in the subject, even for a single call.
13  \fBpcre_exec()\fP. If the subject string is very long, it may still pay to use  Therefore, if the subject string is very long, it may still pay to use JIT for
14  JIT for one-off matches.  one-off matches.
15  .P  .P
16  JIT support applies only to the traditional matching function,  JIT support applies only to the traditional Perl-compatible matching function.
17  \fBpcre_exec()\fP. It does not apply when \fBpcre_dfa_exec()\fP is being used.  It does not apply when the DFA matching function is being used. The code for
18  The code for this support was written by Zoltan Herczeg.  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  .  .
31  .SH "AVAILABILITY OF JIT SUPPORT"  .SH "AVAILABILITY OF JIT SUPPORT"
# Line 32  JIT. The support is limited to the follo Line 42  JIT. The support is limited to the follo
42  .sp  .sp
43  If --enable-jit is set on an unsupported platform, compilation fails.  If --enable-jit is set on an unsupported platform, compilation fails.
44  .P  .P
45  A program can tell if JIT support is available by calling \fBpcre_config()\fP  A program that is linked with PCRE 8.20 or later can tell if JIT support is
46  with the PCRE_CONFIG_JIT option. The result is 1 when JIT is available, and 0  available by calling \fBpcre_config()\fP with the PCRE_CONFIG_JIT option. The
47  otherwise. However, a simple program does not need to check this in order to  result is 1 when JIT is available, and 0 otherwise. However, a simple program
48  use JIT. The API is implemented in a way that falls back to the ordinary PCRE  does not need to check this in order to use JIT. The API is implemented in a
49  code if JIT is not available.  way that falls back to the interpretive code if JIT is not available.
50    .P
51    If your program may sometimes be linked with versions of PCRE that are older
52    than 8.20, but you want to use JIT when it is available, you can test
53    the values of PCRE_MAJOR and PCRE_MINOR, or the existence of a JIT macro such
54    as PCRE_CONFIG_JIT, for compile-time control of your code.
55  .  .
56  .  .
57  .SH "SIMPLE USE OF JIT"  .SH "SIMPLE USE OF JIT"
# Line 49  You have to do two things to make use of Line 64  You have to do two things to make use of
64        \fBpcre_exec()\fP.        \fBpcre_exec()\fP.
65  .sp  .sp
66    (2) Use \fBpcre_free_study()\fP to free the \fBpcre_extra\fP block when it is    (2) Use \fBpcre_free_study()\fP to free the \fBpcre_extra\fP block when it is
67        no longer needed instead of just freeing it yourself. This        no longer needed, instead of just freeing it yourself. This
68        ensures that any JIT data is also freed.        ensures that any JIT data is also freed.
69  .sp  .sp
70    For a program that may be linked with pre-8.20 versions of PCRE, you can insert
71    .sp
72      #ifndef PCRE_STUDY_JIT_COMPILE
73      #define PCRE_STUDY_JIT_COMPILE 0
74      #endif
75    .sp
76    so that no option is passed to \fBpcre_study()\fP, and then use something like
77    this to free the study data:
78    .sp
79      #ifdef PCRE_CONFIG_JIT
80          pcre_free_study(study_ptr);
81      #else
82          pcre_free(study_ptr);
83      #endif
84    .sp
85    PCRE_STUDY_JIT_COMPILE requests the JIT compiler to generate code for complete
86    matches. If you want to run partial matches using the PCRE_PARTIAL_HARD or
87    PCRE_PARTIAL_SOFT options of \fBpcre_exec()\fP, you should set one or both of
88    the following options in addition to, or instead of, PCRE_STUDY_JIT_COMPILE
89    when you call \fBpcre_study()\fP:
90    .sp
91      PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
92      PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
93    .sp
94    The JIT compiler generates different optimized code for each of the three
95    modes (normal, soft partial, hard partial). When \fBpcre_exec()\fP is called,
96    the appropriate code is run if it is available. Otherwise, the pattern is
97    matched using interpretive code.
98    .P
99  In some circumstances you may need to call additional functions. These are  In some circumstances you may need to call additional functions. These are
100  described in the section entitled  described in the section entitled
101  .\" HTML <a href="#stackcontrol">  .\" HTML <a href="#stackcontrol">
# Line 60  described in the section entitled Line 104  described in the section entitled
104  .\"  .\"
105  below.  below.
106  .P  .P
107  If JIT support is not available, PCRE_STUDY_JIT_COMPILE is ignored, and no JIT  If JIT support is not available, PCRE_STUDY_JIT_COMPILE etc. are ignored, and
108  data is set up. Otherwise, the compiled pattern is passed to the JIT compiler,  no JIT data is created. Otherwise, the compiled pattern is passed to the JIT
109  which turns it into machine code that executes much faster than the normal  compiler, which turns it into machine code that executes much faster than the
110  interpretive code. When \fBpcre_exec()\fP is passed a \fBpcre_extra\fP block  normal interpretive code. When \fBpcre_exec()\fP is passed a \fBpcre_extra\fP
111  containing a pointer to JIT code, it obeys that instead of the normal code. The  block containing a pointer to JIT code of the appropriate mode (normal or
112  result is identical, but the code runs much faster.  hard/soft partial), it obeys that code instead of running the interpreter. The
113    result is identical, but the compiled JIT code runs much faster.
114  .P  .P
115  There are some \fBpcre_exec()\fP options that are not supported for JIT  There are some \fBpcre_exec()\fP options that are not supported for JIT
116  execution. There are also some pattern items that JIT cannot handle. Details  execution. There are also some pattern items that JIT cannot handle. Details
117  are given below. In both cases, execution automatically falls back to the  are given below. In both cases, execution automatically falls back to the
118  interpretive code.  interpretive code. If you want to know whether JIT was actually used for a
119    particular match, you should arrange for a JIT callback function to be set up
120    as described in the section entitled
121    .\" HTML <a href="#stackcontrol">
122    .\" </a>
123    "Controlling the JIT stack"
124    .\"
125    below, even if you do not need to supply a non-default JIT stack. Such a
126    callback function is called whenever JIT code is about to be obeyed. If the
127    execution options are not right for JIT execution, the callback function is not
128    obeyed.
129  .P  .P
130  If the JIT compiler finds an unsupported item, no JIT data is generated. You  If the JIT compiler finds an unsupported item, no JIT data is generated. You
131  can find out if JIT execution is available after studying a pattern by calling  can find out if JIT execution is available after studying a pattern by calling
132  \fBpcre_fullinfo()\fP with the PCRE_INFO_JIT option. A result of 1 means that  \fBpcre_fullinfo()\fP with the PCRE_INFO_JIT option. A result of 1 means that
133  JIT compilationw was successful. A result of 0 means that JIT support is not  JIT compilation was successful. A result of 0 means that JIT support is not
134  available, or the pattern was not studied with PCRE_STUDY_JIT_COMPILE, or the  available, or the pattern was not studied with PCRE_STUDY_JIT_COMPILE etc., or
135  JIT compiler was not able to handle the pattern.  the JIT compiler was not able to handle the pattern.
136    .P
137    Once a pattern has been studied, with or without JIT, it can be used as many
138    times as you like for matching different subject strings.
139  .  .
140  .  .
141  .SH "UNSUPPORTED OPTIONS AND PATTERN ITEMS"  .SH "UNSUPPORTED OPTIONS AND PATTERN ITEMS"
142  .rs  .rs
143  .sp  .sp
144  The only \fBpcre_exec()\fP options that are supported for JIT execution are  The only \fBpcre_exec()\fP options that are supported for JIT execution are
145  PCRE_NO_UTF8_CHECK, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and  PCRE_NO_UTF8_CHECK, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
146  PCRE_NOTEMPTY_ATSTART. Note in particular that partial matching is not  PCRE_NOTEMPTY_ATSTART, PCRE_PARTIAL_HARD, and PCRE_PARTIAL_SOFT.
 supported.  
147  .P  .P
148  The unsupported pattern items are:  The unsupported pattern items are:
149  .sp  .sp
150    \eC            match a single byte, even in UTF-8 mode    \eC             match a single byte; not supported in UTF-8 mode
151    (?Cn)          callouts    (?Cn)          callouts
   (?(<name>)...  conditional test on setting of a named subpattern  
   (?(R)...       conditional test on whole pattern recursion  
   (?(Rn)...      conditional test on recursion, by number  
   (?(R&name)...  conditional test on recursion, by name  
152    (*COMMIT)      )    (*COMMIT)      )
153    (*MARK)        )    (*MARK)        )
154    (*PRUNE)       ) the backtracking control verbs    (*PRUNE)       ) the backtracking control verbs
# Line 131  execution. Line 184  execution.
184  .rs  .rs
185  .sp  .sp
186  The code that is generated by the JIT compiler is architecture-specific, and is  The code that is generated by the JIT compiler is architecture-specific, and is
187  also position dependent. For those reasons it cannot be saved and restored like  also position dependent. For those reasons it cannot be saved (in a file or
188  the bytecode and other data of a compiled pattern. You should be able run  database) and restored later like the bytecode and other data of a compiled
189  \fBpcre_study()\fP on a saved and restored pattern, and thereby recreate the  pattern. Saving and restoring compiled patterns is not something many people
190  JIT data, but because JIT compilation uses significant resources, it is  do. More detail about this facility is given in the
191  probably not worth doing this.  .\" HREF
192    \fBpcreprecompile\fP
193    .\"
194    documentation. It should be possible to run \fBpcre_study()\fP on a saved and
195    restored pattern, and thereby recreate the JIT data, but because JIT
196    compilation uses significant resources, it is probably not worth doing this;
197    you might as well recompile the original pattern.
198  .  .
199  .  .
200  .\" HTML <a name="stackcontrol"></a>  .\" HTML <a name="stackcontrol"></a>
# Line 146  When the compiled JIT code runs, it need Line 205  When the compiled JIT code runs, it need
205  By default, it uses 32K on the machine stack. However, some large or  By default, it uses 32K on the machine stack. However, some large or
206  complicated patterns need more than this. The error PCRE_ERROR_JIT_STACKLIMIT  complicated patterns need more than this. The error PCRE_ERROR_JIT_STACKLIMIT
207  is given when there is not enough stack. Three functions are provided for  is given when there is not enough stack. Three functions are provided for
208  managing blocks of memory for use as JIT stacks.  managing blocks of memory for use as JIT stacks. There is further discussion
209    about the use of JIT stacks in the section entitled
210    .\" HTML <a href="#stackcontrol">
211    .\" </a>
212    "JIT stack FAQ"
213    .\"
214    below.
215  .P  .P
216  The \fBpcre_jit_stack_alloc()\fP function creates a JIT stack. Its arguments  The \fBpcre_jit_stack_alloc()\fP function creates a JIT stack. Its arguments
217  are a starting size and a maximum size, and it returns a pointer to an opaque  are a starting size and a maximum size, and it returns a pointer to an opaque
# Line 167  should use. Its arguments are as follows Line 232  should use. Its arguments are as follows
232    void               *data    void               *data
233  .sp  .sp
234  The \fIextra\fP argument must be the result of studying a pattern with  The \fIextra\fP argument must be the result of studying a pattern with
235  PCRE_STUDY_JIT_COMPILE. There are three cases for the values of the other two  PCRE_STUDY_JIT_COMPILE etc. There are three cases for the values of the other
236  options:  two options:
237  .sp  .sp
238    (1) If \fIcallback\fP is NULL and \fIdata\fP is NULL, an internal 32K block    (1) If \fIcallback\fP is NULL and \fIdata\fP is NULL, an internal 32K block
239        on the machine stack is used.        on the machine stack is used.
# Line 176  options: Line 241  options:
241    (2) If \fIcallback\fP is NULL and \fIdata\fP is not NULL, \fIdata\fP must be    (2) If \fIcallback\fP is NULL and \fIdata\fP is not NULL, \fIdata\fP must be
242        a valid JIT stack, the result of calling \fBpcre_jit_stack_alloc()\fP.        a valid JIT stack, the result of calling \fBpcre_jit_stack_alloc()\fP.
243  .sp  .sp
244    (3) If \fIcallback\fP not NULL, it must point to a function that is called    (3) If \fIcallback\fP is not NULL, it must point to a function that is
245        with \fIdata\fP as an argument at the start of matching, in order to        called with \fIdata\fP as an argument at the start of matching, in
246        set up a JIT stack. If the result is NULL, the internal 32K stack        order to set up a JIT stack. If the return from the callback
247        is used; otherwise the return value must be a valid JIT stack,        function is NULL, the internal 32K stack is used; otherwise the
248        the result of calling \fBpcre_jit_stack_alloc()\fP.        return value must be a valid JIT stack, the result of calling
249  .sp        \fBpcre_jit_stack_alloc()\fP.
250    .sp
251    A callback function is obeyed whenever JIT code is about to be run; it is not
252    obeyed when \fBpcre_exec()\fP is called with options that are incompatible for
253    JIT execution. A callback function can therefore be used to determine whether a
254    match operation was executed by JIT or by the interpreter.
255    .P
256  You may safely assign the same JIT stack to more than one pattern, as long as  You may safely assign the same JIT stack to more than one pattern, as long as
257  they are all matched sequentially in the same thread. In a multithread  they are all matched sequentially in the same thread. In a multithread
258  application, each thread must use its own JIT stack.  application, each thread must use its own JIT stack.
# Line 207  This is a suggestion for how a typical m Line 278  This is a suggestion for how a typical m
278  All the functions described in this section do nothing if JIT is not available,  All the functions described in this section do nothing if JIT is not available,
279  and \fBpcre_assign_jit_stack()\fP does nothing unless the \fBextra\fP argument  and \fBpcre_assign_jit_stack()\fP does nothing unless the \fBextra\fP argument
280  is non-NULL and points to a \fBpcre_extra\fP block that is the result of a  is non-NULL and points to a \fBpcre_extra\fP block that is the result of a
281  successful study with PCRE_STUDY_JIT_COMPILE.  successful study with PCRE_STUDY_JIT_COMPILE etc.
282    .
283    .
284    .\" HTML <a name="stackfaq"></a>
285    .SH "JIT STACK FAQ"
286    .rs
287    .sp
288    (1) Why do we need JIT stacks?
289    .sp
290    PCRE (and JIT) is a recursive, depth-first engine, so it needs a stack where
291    the local data of the current node is pushed before checking its child nodes.
292    Allocating real machine stack on some platforms is difficult. For example, the
293    stack chain needs to be updated every time if we extend the stack on PowerPC.
294    Although it is possible, its updating time overhead decreases performance. So
295    we do the recursion in memory.
296    .P
297    (2) Why don't we simply allocate blocks of memory with \fBmalloc()\fP?
298    .sp
299    Modern operating systems have a nice feature: they can reserve an address space
300    instead of allocating memory. We can safely allocate memory pages inside this
301    address space, so the stack could grow without moving memory data (this is
302    important because of pointers). Thus we can allocate 1M address space, and use
303    only a single memory page (usually 4K) if that is enough. However, we can still
304    grow up to 1M anytime if needed.
305    .P
306    (3) Who "owns" a JIT stack?
307    .sp
308    The owner of the stack is the user program, not the JIT studied pattern or
309    anything else. The user program must ensure that if a stack is used by
310    \fBpcre_exec()\fP, (that is, it is assigned to the pattern currently running),
311    that stack must not be used by any other threads (to avoid overwriting the same
312    memory area). The best practice for multithreaded programs is to allocate a
313    stack for each thread, and return this stack through the JIT callback function.
314    .P
315    (4) When should a JIT stack be freed?
316    .sp
317    You can free a JIT stack at any time, as long as it will not be used by
318    \fBpcre_exec()\fP again. When you assign the stack to a pattern, only a pointer
319    is set. There is no reference counting or any other magic. You can free the
320    patterns and stacks in any order, anytime. Just \fIdo not\fP call
321    \fBpcre_exec()\fP with a pattern pointing to an already freed stack, as that
322    will cause SEGFAULT. (Also, do not free a stack currently used by
323    \fBpcre_exec()\fP in another thread). You can also replace the stack for a
324    pattern at any time. You can even free the previous stack before assigning a
325    replacement.
326    .P
327    (5) Should I allocate/free a stack every time before/after calling
328    \fBpcre_exec()\fP?
329    .sp
330    No, because this is too costly in terms of resources. However, you could
331    implement some clever idea which release the stack if it is not used in let's
332    say two minutes. The JIT callback can help to achive this without keeping a
333    list of the currently JIT studied patterns.
334    .P
335    (6) OK, the stack is for long term memory allocation. But what happens if a
336    pattern causes stack overflow with a stack of 1M? Is that 1M kept until the
337    stack is freed?
338    .sp
339    Especially on embedded sytems, it might be a good idea to release
340    memory sometimes without freeing the stack. There is no API for this at the
341    moment. Probably a function call which returns with the currently allocated
342    memory for any stack and another which allows releasing memory (shrinking the
343    stack) would be a good idea if someone needs this.
344    .P
345    (7) This is too much of a headache. Isn't there any better solution for JIT
346    stack handling?
347    .sp
348    No, thanks to Windows. If POSIX threads were used everywhere, we could throw
349    out this complicated API.
350  .  .
351  .  .
352  .SH "EXAMPLE CODE"  .SH "EXAMPLE CODE"
# Line 246  callback. Line 385  callback.
385  .rs  .rs
386  .sp  .sp
387  .nf  .nf
388  Philip Hazel  Philip Hazel (FAQ by Zoltan Herczeg)
389  University Computing Service  University Computing Service
390  Cambridge CB2 3QH, England.  Cambridge CB2 3QH, England.
391  .fi  .fi
# Line 256  Cambridge CB2 3QH, England. Line 395  Cambridge CB2 3QH, England.
395  .rs  .rs
396  .sp  .sp
397  .nf  .nf
398  Last updated: 06 September 2011  Last updated: 22 February 2012
399  Copyright (c) 1997-2011 University of Cambridge.  Copyright (c) 1997-2012 University of Cambridge.
400  .fi  .fi
Legend:
Removed from v.691  
changed lines
  Added in v.926
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12