1PCRE2STACK(3) Library Functions Manual PCRE2STACK(3)
2
6 PCRE2 - Perl-compatible regular expressions (revised API)
7
9 When you call pcre2_match(), it makes use of an internal function
11 called match(). This calls itself recursively at branch points in the
12 pattern, in order to remember the state of the match so that it can
13 back up and try a different alternative after a failure. As matching
14 proceeds deeper and deeper into the tree of possibilities, the recur‐
15 sion depth increases. The match() function is also called in other cir‐
16 cumstances, for example, whenever a parenthesized sub-pattern is
17 entered, and in certain cases of repetition.
18 Not all calls of match() increase the recursion depth; for an item such
20 as a* it may be called several times at the same level, after matching
21 different numbers of a's. Furthermore, in a number of cases where the
22 result of the recursive call would immediately be passed back as the
23 result of the current call (a "tail recursion"), the function is just
24 restarted instead.
25 Each time the internal match() function is called recursively, it uses
27 memory from the process stack. For certain kinds of pattern and data,
28 very large amounts of stack may be needed, despite the recognition of
29 "tail recursion". Note that if PCRE2 is compiled with the -fsani‐
30 tize=address option of the GCC compiler, the stack requirements are
31 greatly increased.
32 The above comments apply when pcre2_match() is run in its normal inter‐
34 pretive manner. If the compiled pattern was processed by pcre2_jit_com‐
35 pile(), and just-in-time compiling was successful, and the options
36 passed to pcre2_match() were not incompatible, the matching process
37 uses the JIT-compiled code instead of the match() function. In this
38 case, the memory requirements are handled entirely differently. See the
39 pcre2jit documentation for details.
40 The pcre2_dfa_match() function operates in a different way to
42 pcre2_match(), and uses recursion only when there is a regular expres‐
43 sion recursion or subroutine call in the pattern. This includes the
44 processing of assertion and "once-only" subpatterns, which are handled
45 like subroutine calls. Normally, these are never very deep, and the
46 limit on the complexity of pcre2_dfa_match() is controlled by the
47 amount of workspace it is given. However, it is possible to write pat‐
48 terns with runaway infinite recursions; such patterns will cause
49 pcre2_dfa_match() to run out of stack unless a limit is applied (see
50 below).
51 The comments in the next three sections do not apply to
53 pcre2_dfa_match(); they are relevant only for pcre2_match() without the
54 JIT optimization.
55 Reducing pcre2_match()'s stack usage
57 You can often reduce the amount of recursion, and therefore the amount
59 of stack used, by modifying the pattern that is being matched. Con‐
60 sider, for example, this pattern:
61 ([^<]|<(?!inet))+
63 It matches from wherever it starts until it encounters "<inet" or the
65 end of the data, and is the kind of pattern that might be used when
66 processing an XML file. Each iteration of the outer parentheses matches
67 either one character that is not "<" or a "<" that is not followed by
68 "inet". However, each time a parenthesis is processed, a recursion
69 occurs, so this formulation uses a stack frame for each matched charac‐
70 ter. For a long string, a lot of stack is required. Consider now this
71 rewritten pattern, which matches exactly the same strings:
72 ([^<]++|<(?!inet))+
74 This uses very much less stack, because runs of characters that do not
76 contain "<" are "swallowed" in one item inside the parentheses. Recur‐
77 sion happens only when a "<" character that is not followed by "inet"
78 is encountered (and we assume this is relatively rare). A possessive
79 quantifier is used to stop any backtracking into the runs of non-"<"
80 characters, but that is not related to stack usage.
81 This example shows that one way of avoiding stack problems when match‐
83 ing long subject strings is to write repeated parenthesized subpatterns
84 to match more than one character whenever possible.
85 Compiling PCRE2 to use heap instead of stack for pcre2_match()
87 In environments where stack memory is constrained, you might want to
89 compile PCRE2 to use heap memory instead of stack for remembering back-
90 up points when pcre2_match() is running. This makes it run more slowly,
91 however. Details of how to do this are given in the pcre2build documen‐
92 tation. When built in this way, instead of using the stack, PCRE2 gets
93 memory for remembering backup points from the heap. By default, the
94 memory is obtained by calling the system malloc() function, but you can
95 arrange to supply your own memory management function. For details, see
96 the section entitled "The match context" in the pcre2api documentation.
97 Since the block sizes are always the same, it may be possible to imple‐
98 ment a customized memory handler that is more efficient than the stan‐
99 dard function. The memory blocks obtained for this purpose are retained
100 and re-used if possible while pcre2_match() is running. They are all
101 freed just before it exits.
102 Limiting pcre2_match()'s stack usage
104 You can set limits on the number of times the internal match() function
106 is called, both in total and recursively. If a limit is exceeded,
107 pcre2_match() returns an error code. Setting suitable limits should
108 prevent it from running out of stack. The default values of the limits
109 are very large, and unlikely ever to operate. They can be changed when
110 PCRE2 is built, and they can also be set when pcre2_match() is called.
111 For details of these interfaces, see the pcre2build documentation and
112 the section entitled "The match context" in the pcre2api documentation.
113 As a very rough rule of thumb, you should reckon on about 500 bytes per
115 recursion. Thus, if you want to limit your stack usage to 8Mb, you
116 should set the limit at 16000 recursions. A 64Mb stack, on the other
117 hand, can support around 128000 recursions.
118 The pcre2test test program has a modifier called "find_limits" which,
120 if applied to a subject line, causes it to find the smallest limits
121 that allow a a pattern to match. This is done by calling pcre2_match()
122 repeatedly with different limits.
123 Limiting pcre2_dfa_match()'s stack usage
125 The recursion limit, as described above for pcre2_match(), also applies
127 to pcre2_dfa_match(), whose use of recursive function calls for recur‐
128 sions in the pattern can lead to runaway stack usage. The non-recursive
129 match limit is not relevant for DFA matching, and is ignored.
130 Changing stack size in Unix-like systems
132 In Unix-like environments, there is not often a problem with the stack
134 unless very long strings are involved, though the default limit on
135 stack size varies from system to system. Values from 8Mb to 64Mb are
136 common. You can find your default limit by running the command:
137 ulimit -s
139 Unfortunately, the effect of running out of stack is often SIGSEGV,
141 though sometimes a more explicit error message is given. You can nor‐
142 mally increase the limit on stack size by code such as this:
143 struct rlimit rlim;
145 getrlimit(RLIMIT_STACK, &rlim);
146 rlim.rlim_cur = 100*1024*1024;
147 setrlimit(RLIMIT_STACK, &rlim);
148 This reads the current limits (soft and hard) using getrlimit(), then
150 attempts to increase the soft limit to 100Mb using setrlimit(). You
151 must do this before calling pcre2_match().
152 Changing stack size in Mac OS X
154 Using setrlimit(), as described above, should also work on Mac OS X. It
156 is also possible to set a stack size when linking a program. There is a
157 discussion about stack sizes in Mac OS X at this web site:
158 http://developer.apple.com/qa/qa2005/qa1419.html.
159
161 Philip Hazel
163 University Computing Service
164 Cambridge, England.
165
167 Last updated: 23 December 2016
169 Copyright (c) 1997-2016 University of Cambridge.
170PCRE2 10.23 23 December 2016 PCRE2STACK(3)