trunk/doc/pcrestack.3

.TH PCRESTACK 3
.SH NAME
PCRE - Perl-compatible regular expressions
.SH "PCRE DISCUSSION OF STACK USAGE"
.rs
.sp
When you call \fBpcre_exec()\fP, it makes use of an internal function called
\fBmatch()\fP. This calls itself recursively at branch points in the pattern,
in order to remember the state of the match so that it can back up and try a
different alternative if the first one fails. As matching proceeds deeper and
deeper into the tree of possibilities, the recursion depth increases. The
\fBmatch()\fP function is also called in other circumstances, for example,
whenever a parenthesized sub-pattern is entered, and in certain cases of
repetition.
.P
Not all calls of \fBmatch()\fP increase the recursion depth; for an item such
as a* it may be called several times at the same level, after matching
different numbers of a's. Furthermore, in a number of cases where the result of
the recursive call would immediately be passed back as the result of the
current call (a "tail recursion"), the function is just restarted instead.
.P
The \fBpcre_dfa_exec()\fP function operates in an entirely different way, and
uses recursion only when there is a regular expression recursion or subroutine
call in the pattern. This includes the processing of assertion and "once-only"
subpatterns, which are handled like subroutine calls. Normally, these are never
very deep, and the limit on the complexity of \fBpcre_dfa_exec()\fP is
controlled by the amount of workspace it is given. However, it is possible to
write patterns with runaway infinite recursions; such patterns will cause
\fBpcre_dfa_exec()\fP to run out of stack. At present, there is no protection
against this.
.P
The comments that follow do NOT apply to \fBpcre_dfa_exec()\fP; they are
relevant only for \fBpcre_exec()\fP.
.
.
.SS "Reducing \fBpcre_exec()\fP's stack usage"
.rs
.sp
Each time that \fBmatch()\fP is actually called recursively, it uses memory
from the process stack. For certain kinds of pattern and data, very large
amounts of stack may be needed, despite the recognition of "tail recursion".
You can often reduce the amount of recursion, and therefore the amount of stack
used, by modifying the pattern that is being matched. Consider, for example,
this pattern:
.sp
  ([^<]|<(?!inet))+
.sp
It matches from wherever it starts until it encounters "<inet" or the end of
the data, and is the kind of pattern that might be used when processing an XML
file. Each iteration of the outer parentheses matches either one character that
is not "<" or a "<" that is not followed by "inet". However, each time a
parenthesis is processed, a recursion occurs, so this formulation uses a stack
frame for each matched character. For a long string, a lot of stack is
required. Consider now this rewritten pattern, which matches exactly the same
strings:
.sp
  ([^<]++|<(?!inet))+
.sp
This uses very much less stack, because runs of characters that do not contain
"<" are "swallowed" in one item inside the parentheses. Recursion happens only
when a "<" character that is not followed by "inet" is encountered (and we
assume this is relatively rare). A possessive quantifier is used to stop any
backtracking into the runs of non-"<" characters, but that is not related to
stack usage.
.P
This example shows that one way of avoiding stack problems when matching long
subject strings is to write repeated parenthesized subpatterns to match more
than one character whenever possible.
.
.
.SS "Compiling PCRE to use heap instead of stack for \fBpcre_exec()\fP"
.rs
.sp
In environments where stack memory is constrained, you might want to compile
PCRE to use heap memory instead of stack for remembering back-up points when
\fBpcre_exec()\fP is running. This makes it run a lot more slowly, however.
Details of how to do this are given in the
.\" HREF
\fBpcrebuild\fP
.\"
documentation. When built in this way, instead of using the stack, PCRE obtains
and frees memory by calling the functions that are pointed to by the
\fBpcre_stack_malloc\fP and \fBpcre_stack_free\fP variables. By default, these
point to \fBmalloc()\fP and \fBfree()\fP, but you can replace the pointers to
cause PCRE to use your own functions. Since the block sizes are always the
same, and are always freed in reverse order, it may be possible to implement
customized memory handlers that are more efficient than the standard functions.
.
.
.SS "Limiting \fBpcre_exec()\fP's stack usage"
.rs
.sp
You can set limits on the number of times that \fBmatch()\fP is called, both in
total and recursively. If a limit is exceeded, \fBpcre_exec()\fP returns an
error code. Setting suitable limits should prevent it from running out of
stack. The default values of the limits are very large, and unlikely ever to
operate. They can be changed when PCRE is built, and they can also be set when
\fBpcre_exec()\fP is called. For details of these interfaces, see the
.\" HREF
\fBpcrebuild\fP
.\"
documentation and the
.\" HTML <a href="pcreapi.html#extradata">
.\" </a>
section on extra data for \fBpcre_exec()\fP
.\"
in the
.\" HREF
\fBpcreapi\fP
.\"
documentation.
.P
As a very rough rule of thumb, you should reckon on about 500 bytes per
recursion. Thus, if you want to limit your stack usage to 8Mb, you
should set the limit at 16000 recursions. A 64Mb stack, on the other hand, can
support around 128000 recursions.
.P
In Unix-like environments, the \fBpcretest\fP test program has a command line
option (\fB-S\fP) that can be used to increase the size of its stack. As long
as the stack is large enough, another option (\fB-M\fP) can be used to find the
smallest limits that allow a particular pattern to match a given subject
string. This is done by calling \fBpcre_exec()\fP repeatedly with different
limits.
.
.
.SS "Changing stack size in Unix-like systems"
.rs
.sp
In Unix-like environments, there is not often a problem with the stack unless
very long strings are involved, though the default limit on stack size varies
from system to system. Values from 8Mb to 64Mb are common. You can find your
default limit by running the command:
.sp
  ulimit -s
.sp
Unfortunately, the effect of running out of stack is often SIGSEGV, though
sometimes a more explicit error message is given. You can normally increase the
limit on stack size by code such as this:
.sp
  struct rlimit rlim;
  getrlimit(RLIMIT_STACK, &rlim);
  rlim.rlim_cur = 100*1024*1024;
  setrlimit(RLIMIT_STACK, &rlim);
.sp
This reads the current limits (soft and hard) using \fBgetrlimit()\fP, then
attempts to increase the soft limit to 100Mb using \fBsetrlimit()\fP. You must
do this before calling \fBpcre_exec()\fP.
.
.
.SS "Changing stack size in Mac OS X"
.rs
.sp
Using \fBsetrlimit()\fP, as described above, should also work on Mac OS X. It
is also possible to set a stack size when linking a program. There is a
discussion about stack sizes in Mac OS X at this web site:
.\" HTML <a href="http://developer.apple.com/qa/qa2005/qa1419.html">
.\" </a>
http://developer.apple.com/qa/qa2005/qa1419.html.
.\"
.
.
.SH AUTHOR
.rs
.sp
.nf
Philip Hazel
University Computing Service
Cambridge CB2 3QH, England.
.fi
.
.
.SH REVISION
.rs
.sp
.nf
Last updated: 22 July 2011
Copyright (c) 1997-2011 University of Cambridge.
.fi
1	nigel	91	.TH PCRESTACK 3
2			.SH NAME
3			PCRE - Perl-compatible regular expressions
4			.SH "PCRE DISCUSSION OF STACK USAGE"
5			.rs
6			.sp
7			When you call \fBpcre_exec()\fP, it makes use of an internal function called
8			\fBmatch()\fP. This calls itself recursively at branch points in the pattern,
9			in order to remember the state of the match so that it can back up and try a
10			different alternative if the first one fails. As matching proceeds deeper and
11	ph10	631	deeper into the tree of possibilities, the recursion depth increases. The
12	ph10	654	\fBmatch()\fP function is also called in other circumstances, for example,
13	ph10	631	whenever a parenthesized sub-pattern is entered, and in certain cases of
14			repetition.
15	nigel	91	.P
16			Not all calls of \fBmatch()\fP increase the recursion depth; for an item such
17			as a* it may be called several times at the same level, after matching
18			different numbers of a's. Furthermore, in a number of cases where the result of
19			the recursive call would immediately be passed back as the result of the
20			current call (a "tail recursion"), the function is just restarted instead.
21			.P
22			The \fBpcre_dfa_exec()\fP function operates in an entirely different way, and
23	ph10	480	uses recursion only when there is a regular expression recursion or subroutine
24			call in the pattern. This includes the processing of assertion and "once-only"
25			subpatterns, which are handled like subroutine calls. Normally, these are never
26			very deep, and the limit on the complexity of \fBpcre_dfa_exec()\fP is
27			controlled by the amount of workspace it is given. However, it is possible to
28			write patterns with runaway infinite recursions; such patterns will cause
29			\fBpcre_dfa_exec()\fP to run out of stack. At present, there is no protection
30			against this.
31	nigel	91	.P
32	ph10	480	The comments that follow do NOT apply to \fBpcre_dfa_exec()\fP; they are
33			relevant only for \fBpcre_exec()\fP.
34			.
35			.
36			.SS "Reducing \fBpcre_exec()\fP's stack usage"
37			.rs
38			.sp
39	nigel	91	Each time that \fBmatch()\fP is actually called recursively, it uses memory
40			from the process stack. For certain kinds of pattern and data, very large
41			amounts of stack may be needed, despite the recognition of "tail recursion".
42			You can often reduce the amount of recursion, and therefore the amount of stack
43			used, by modifying the pattern that is being matched. Consider, for example,
44			this pattern:
45			.sp
46			([^<]\|<(?!inet))+
47			.sp
48			It matches from wherever it starts until it encounters "<inet" or the end of
49			the data, and is the kind of pattern that might be used when processing an XML
50			file. Each iteration of the outer parentheses matches either one character that
51			is not "<" or a "<" that is not followed by "inet". However, each time a
52			parenthesis is processed, a recursion occurs, so this formulation uses a stack
53			frame for each matched character. For a long string, a lot of stack is
54			required. Consider now this rewritten pattern, which matches exactly the same
55			strings:
56			.sp
57	ph10	122	([^<]++\|<(?!inet))+
58	nigel	91	.sp
59			This uses very much less stack, because runs of characters that do not contain
60			"<" are "swallowed" in one item inside the parentheses. Recursion happens only
61			when a "<" character that is not followed by "inet" is encountered (and we
62			assume this is relatively rare). A possessive quantifier is used to stop any
63			backtracking into the runs of non-"<" characters, but that is not related to
64			stack usage.
65			.P
66	nigel	93	This example shows that one way of avoiding stack problems when matching long
67			subject strings is to write repeated parenthesized subpatterns to match more
68			than one character whenever possible.
69	ph10	358	.
70	ph10	480	.
71			.SS "Compiling PCRE to use heap instead of stack for \fBpcre_exec()\fP"
72	ph10	358	.rs
73			.sp
74	nigel	91	In environments where stack memory is constrained, you might want to compile
75	ph10	480	PCRE to use heap memory instead of stack for remembering back-up points when
76			\fBpcre_exec()\fP is running. This makes it run a lot more slowly, however.
77			Details of how to do this are given in the
78	nigel	91	.\" HREF
79			\fBpcrebuild\fP
80			.\"
81	ph10	174	documentation. When built in this way, instead of using the stack, PCRE obtains
82			and frees memory by calling the functions that are pointed to by the
83			\fBpcre_stack_malloc\fP and \fBpcre_stack_free\fP variables. By default, these
84			point to \fBmalloc()\fP and \fBfree()\fP, but you can replace the pointers to
85	ph10	182	cause PCRE to use your own functions. Since the block sizes are always the
86			same, and are always freed in reverse order, it may be possible to implement
87	ph10	174	customized memory handlers that are more efficient than the standard functions.
88	ph10	358	.
89	ph10	480	.
90			.SS "Limiting \fBpcre_exec()\fP's stack usage"
91	ph10	358	.rs
92			.sp
93	ph10	480	You can set limits on the number of times that \fBmatch()\fP is called, both in
94			total and recursively. If a limit is exceeded, \fBpcre_exec()\fP returns an
95			error code. Setting suitable limits should prevent it from running out of
96			stack. The default values of the limits are very large, and unlikely ever to
97			operate. They can be changed when PCRE is built, and they can also be set when
98	ph10	358	\fBpcre_exec()\fP is called. For details of these interfaces, see the
99			.\" HREF
100			\fBpcrebuild\fP
101			.\"
102	ph10	480	documentation and the
103			.\" HTML <a href="pcreapi.html#extradata">
104			.\" </a>
105			section on extra data for \fBpcre_exec()\fP
106			.\"
107			in the
108	ph10	358	.\" HREF
109			\fBpcreapi\fP
110			.\"
111			documentation.
112	nigel	91	.P
113	ph10	358	As a very rough rule of thumb, you should reckon on about 500 bytes per
114			recursion. Thus, if you want to limit your stack usage to 8Mb, you
115			should set the limit at 16000 recursions. A 64Mb stack, on the other hand, can
116	ph10	487	support around 128000 recursions.
117	ph10	480	.P
118			In Unix-like environments, the \fBpcretest\fP test program has a command line
119			option (\fB-S\fP) that can be used to increase the size of its stack. As long
120			as the stack is large enough, another option (\fB-M\fP) can be used to find the
121			smallest limits that allow a particular pattern to match a given subject
122			string. This is done by calling \fBpcre_exec()\fP repeatedly with different
123			limits.
124	ph10	358	.
125	ph10	480	.
126	ph10	358	.SS "Changing stack size in Unix-like systems"
127			.rs
128			.sp
129	nigel	93	In Unix-like environments, there is not often a problem with the stack unless
130			very long strings are involved, though the default limit on stack size varies
131			from system to system. Values from 8Mb to 64Mb are common. You can find your
132			default limit by running the command:
133	nigel	91	.sp
134			ulimit -s
135			.sp
136	nigel	93	Unfortunately, the effect of running out of stack is often SIGSEGV, though
137			sometimes a more explicit error message is given. You can normally increase the
138			limit on stack size by code such as this:
139	nigel	91	.sp
140			struct rlimit rlim;
141			getrlimit(RLIMIT_STACK, &rlim);
142			rlim.rlim_cur = 10010241024;
143			setrlimit(RLIMIT_STACK, &rlim);
144			.sp
145			This reads the current limits (soft and hard) using \fBgetrlimit()\fP, then
146			attempts to increase the soft limit to 100Mb using \fBsetrlimit()\fP. You must
147			do this before calling \fBpcre_exec()\fP.
148	ph10	358	.
149	ph10	480	.
150	ph10	358	.SS "Changing stack size in Mac OS X"
151			.rs
152			.sp
153			Using \fBsetrlimit()\fP, as described above, should also work on Mac OS X. It
154			is also possible to set a stack size when linking a program. There is a
155			discussion about stack sizes in Mac OS X at this web site:
156			.\" HTML <a href="http://developer.apple.com/qa/qa2005/qa1419.html">
157			.\" </a>
158			http://developer.apple.com/qa/qa2005/qa1419.html.
159	nigel	91	.\"
160	ph10	99	.
161			.
162			.SH AUTHOR
163			.rs
164			.sp
165			.nf
166			Philip Hazel
167			University Computing Service
168			Cambridge CB2 3QH, England.
169			.fi
170			.
171			.
172			.SH REVISION
173			.rs
174			.sp
175			.nf
176	ph10	631	Last updated: 22 July 2011
177			Copyright (c) 1997-2011 University of Cambridge.
178	ph10	99	.fi
Name	Value
svn:eol-style	native
svn:keywords	"Author Date Id Revision Url"