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.
.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
hardly uses recursion at all. The limit on its complexity is the amount of
workspace it is given. The comments that follow do NOT apply to
\fBpcre_dfa_exec()\fP; they are relevant only for \fBpcre_exec()\fP.
.P
You can set limits on the number of times that \fBmatch()\fP is called, both in
total and recursively. If the limit is exceeded, an error occurs. For details,
see the
.\" HTML <a href="pcreapi.html#extradata">
.\" </a>
section on extra data for \fBpcre_exec()\fP
.\"
in the
.\" HREF
\fBpcreapi\fP
.\"
documentation.
.P
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.
.P
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. 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.
.P
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.
.P
PCRE has an internal counter that can be used to limit the depth of recursion,
and thus cause \fBpcre_exec()\fP to give an error code before it runs out of
stack. By default, the limit is very large, and unlikely ever to operate. It
can be changed when PCRE is built, and it can also be set when
\fBpcre_exec()\fP is called. For details of these interfaces, see the
.\" HREF
\fBpcrebuild\fP
.\"
and
.\" 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. The \fBpcretest\fP test program has a command
line option (\fB-S\fP) that can be used to increase the size of its stack.
.
.
.SH AUTHOR
.rs
.sp
.nf
Philip Hazel
University Computing Service
Cambridge CB2 3QH, England.
.fi
.
.
.SH REVISION
.rs
.sp
.nf
Last updated: 05 June 2007
Copyright (c) 1997-2007 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			deeper into the tree of possibilities, the recursion depth increases.
12			.P
13			Not all calls of \fBmatch()\fP increase the recursion depth; for an item such
14			as a* it may be called several times at the same level, after matching
15			different numbers of a's. Furthermore, in a number of cases where the result of
16			the recursive call would immediately be passed back as the result of the
17			current call (a "tail recursion"), the function is just restarted instead.
18			.P
19			The \fBpcre_dfa_exec()\fP function operates in an entirely different way, and
20			hardly uses recursion at all. The limit on its complexity is the amount of
21			workspace it is given. The comments that follow do NOT apply to
22			\fBpcre_dfa_exec()\fP; they are relevant only for \fBpcre_exec()\fP.
23			.P
24			You can set limits on the number of times that \fBmatch()\fP is called, both in
25			total and recursively. If the limit is exceeded, an error occurs. For details,
26			see the
27			.\" HTML <a href="pcreapi.html#extradata">
28			.\" </a>
29			section on extra data for \fBpcre_exec()\fP
30			.\"
31			in the
32			.\" HREF
33			\fBpcreapi\fP
34			.\"
35			documentation.
36			.P
37			Each time that \fBmatch()\fP is actually called recursively, it uses memory
38			from the process stack. For certain kinds of pattern and data, very large
39			amounts of stack may be needed, despite the recognition of "tail recursion".
40			You can often reduce the amount of recursion, and therefore the amount of stack
41			used, by modifying the pattern that is being matched. Consider, for example,
42			this pattern:
43			.sp
44			([^<]\|<(?!inet))+
45			.sp
46			It matches from wherever it starts until it encounters "<inet" or the end of
47			the data, and is the kind of pattern that might be used when processing an XML
48			file. Each iteration of the outer parentheses matches either one character that
49			is not "<" or a "<" that is not followed by "inet". However, each time a
50			parenthesis is processed, a recursion occurs, so this formulation uses a stack
51			frame for each matched character. For a long string, a lot of stack is
52			required. Consider now this rewritten pattern, which matches exactly the same
53			strings:
54			.sp
55	ph10	122	([^<]++\|<(?!inet))+
56	nigel	91	.sp
57			This uses very much less stack, because runs of characters that do not contain
58			"<" are "swallowed" in one item inside the parentheses. Recursion happens only
59			when a "<" character that is not followed by "inet" is encountered (and we
60			assume this is relatively rare). A possessive quantifier is used to stop any
61			backtracking into the runs of non-"<" characters, but that is not related to
62			stack usage.
63			.P
64	nigel	93	This example shows that one way of avoiding stack problems when matching long
65			subject strings is to write repeated parenthesized subpatterns to match more
66			than one character whenever possible.
67			.P
68	nigel	91	In environments where stack memory is constrained, you might want to compile
69			PCRE to use heap memory instead of stack for remembering back-up points. This
70			makes it run a lot more slowly, however. Details of how to do this are given in
71			the
72			.\" HREF
73			\fBpcrebuild\fP
74			.\"
75	ph10	174	documentation. When built in this way, instead of using the stack, PCRE obtains
76			and frees memory by calling the functions that are pointed to by the
77			\fBpcre_stack_malloc\fP and \fBpcre_stack_free\fP variables. By default, these
78			point to \fBmalloc()\fP and \fBfree()\fP, but you can replace the pointers to
79	ph10	182	cause PCRE to use your own functions. Since the block sizes are always the
80			same, and are always freed in reverse order, it may be possible to implement
81	ph10	174	customized memory handlers that are more efficient than the standard functions.
82	nigel	91	.P
83	nigel	93	In Unix-like environments, there is not often a problem with the stack unless
84			very long strings are involved, though the default limit on stack size varies
85			from system to system. Values from 8Mb to 64Mb are common. You can find your
86			default limit by running the command:
87	nigel	91	.sp
88			ulimit -s
89			.sp
90	nigel	93	Unfortunately, the effect of running out of stack is often SIGSEGV, though
91			sometimes a more explicit error message is given. You can normally increase the
92			limit on stack size by code such as this:
93	nigel	91	.sp
94			struct rlimit rlim;
95			getrlimit(RLIMIT_STACK, &rlim);
96			rlim.rlim_cur = 10010241024;
97			setrlimit(RLIMIT_STACK, &rlim);
98			.sp
99			This reads the current limits (soft and hard) using \fBgetrlimit()\fP, then
100			attempts to increase the soft limit to 100Mb using \fBsetrlimit()\fP. You must
101			do this before calling \fBpcre_exec()\fP.
102			.P
103			PCRE has an internal counter that can be used to limit the depth of recursion,
104			and thus cause \fBpcre_exec()\fP to give an error code before it runs out of
105			stack. By default, the limit is very large, and unlikely ever to operate. It
106			can be changed when PCRE is built, and it can also be set when
107			\fBpcre_exec()\fP is called. For details of these interfaces, see the
108			.\" HREF
109			\fBpcrebuild\fP
110			.\"
111			and
112			.\" HREF
113			\fBpcreapi\fP
114			.\"
115			documentation.
116			.P
117			As a very rough rule of thumb, you should reckon on about 500 bytes per
118			recursion. Thus, if you want to limit your stack usage to 8Mb, you
119			should set the limit at 16000 recursions. A 64Mb stack, on the other hand, can
120			support around 128000 recursions. The \fBpcretest\fP test program has a command
121	nigel	93	line option (\fB-S\fP) that can be used to increase the size of its stack.
122	ph10	99	.
123			.
124			.SH AUTHOR
125			.rs
126			.sp
127			.nf
128			Philip Hazel
129			University Computing Service
130			Cambridge CB2 3QH, England.
131			.fi
132			.
133			.
134			.SH REVISION
135			.rs
136			.sp
137			.nf
138	ph10	174	Last updated: 05 June 2007
139	ph10	99	Copyright (c) 1997-2007 University of Cambridge.
140			.fi
Name	Value
svn:eol-style	native
svn:keywords	"Author Date Id Revision Url"