1PERLDEBGUTS(1) Perl Programmers Reference Guide PERLDEBGUTS(1)
2
6 perldebguts - Guts of Perl debugging
7
9 This is not the perldebug(1) manpage, which tells you how to use the
10 debugger. This manpage describes low-level details concerning the
11 debugger's internals, which range from difficult to impossible to
12 understand for anyone who isn't incredibly intimate with Perl's guts.
13 Caveat lector.
14
16 Perl has special debugging hooks at compile-time and run-time used to
17 create debugging environments. These hooks are not to be confused with
18 the perl -Dxxx command described in perlrun, which is usable only if a
19 special Perl is built per the instructions in the INSTALL podpage in
20 the Perl source tree.
21 For example, whenever you call Perl's built-in "caller" function from
23 the package "DB", the arguments that the corresponding stack frame was
24 called with are copied to the @DB::args array. These mechanisms are
25 enabled by calling Perl with the -d switch. Specifically, the
26 following additional features are enabled (cf. "$^P" in perlvar):
27 · Perl inserts the contents of $ENV{PERL5DB} (or "BEGIN {require
29 'perl5db.pl'}" if not present) before the first line of your
30 program.
31 · Each array "@{"_<$filename"}" holds the lines of $filename for a
33 file compiled by Perl. The same is also true for "eval"ed strings
34 that contain subroutines, or which are currently being executed.
35 The $filename for "eval"ed strings looks like "(eval 34)". Code
36 assertions in regexes look like "(re_eval 19)".
37 Values in this array are magical in numeric context: they compare
39 equal to zero only if the line is not breakable.
40 · Each hash "%{"_<$filename"}" contains breakpoints and actions keyed
42 by line number. Individual entries (as opposed to the whole hash)
43 are settable. Perl only cares about Boolean true here, although
44 the values used by perl5db.pl have the form
45 "$break_condition\0$action".
46 The same holds for evaluated strings that contain subroutines, or
48 which are currently being executed. The $filename for "eval"ed
49 strings looks like "(eval 34)" or "(re_eval 19)".
50 · Each scalar "${"_<$filename"}" contains "_<$filename". This is
52 also the case for evaluated strings that contain subroutines, or
53 which are currently being executed. The $filename for "eval"ed
54 strings looks like "(eval 34)" or "(re_eval 19)".
55 · After each "require"d file is compiled, but before it is executed,
57 "DB::postponed(*{"_<$filename"})" is called if the subroutine
58 "DB::postponed" exists. Here, the $filename is the expanded name
59 of the "require"d file, as found in the values of %INC.
60 · After each subroutine "subname" is compiled, the existence of
62 $DB::postponed{subname} is checked. If this key exists,
63 "DB::postponed(subname)" is called if the "DB::postponed"
64 subroutine also exists.
65 · A hash %DB::sub is maintained, whose keys are subroutine names and
67 whose values have the form "filename:startline-endline".
68 "filename" has the form "(eval 34)" for subroutines defined inside
69 "eval"s, or "(re_eval 19)" for those within regex code assertions.
70 · When the execution of your program reaches a point that can hold a
72 breakpoint, the "DB::DB()" subroutine is called if any of the
73 variables $DB::trace, $DB::single, or $DB::signal is true. These
74 variables are not "local"izable. This feature is disabled when
75 executing inside "DB::DB()", including functions called from it
76 unless "$^D & (1<<30)" is true.
77 · When execution of the program reaches a subroutine call, a call to
79 &DB::sub(args) is made instead, with $DB::sub holding the name of
80 the called subroutine. (This doesn't happen if the subroutine was
81 compiled in the "DB" package.)
82 Note that if &DB::sub needs external data for it to work, no subroutine
84 call is possible without it. As an example, the standard debugger's
85 &DB::sub depends on the $DB::deep variable (it defines how many levels
86 of recursion deep into the debugger you can go before a mandatory
87 break). If $DB::deep is not defined, subroutine calls are not
88 possible, even though &DB::sub exists.
89 Writing Your Own Debugger
91 Environment Variables
92 The "PERL5DB" environment variable can be used to define a debugger.
94 For example, the minimal "working" debugger (it actually doesn't do
95 anything) consists of one line:
96 sub DB::DB {}
98 It can easily be defined like this:
100 $ PERL5DB="sub DB::DB {}" perl -d your-script
102 Another brief debugger, slightly more useful, can be created with only
104 the line:
105 sub DB::DB {print ++$i; scalar <STDIN>}
107 This debugger prints a number which increments for each statement
109 encountered and waits for you to hit a newline before continuing to the
110 next statement.
111 The following debugger is actually useful:
113 {
115 package DB;
116 sub DB {}
117 sub sub {print ++$i, " $sub\n"; &$sub}
118 }
119 It prints the sequence number of each subroutine call and the name of
121 the called subroutine. Note that &DB::sub is being compiled into the
122 package "DB" through the use of the "package" directive.
123 When it starts, the debugger reads your rc file (./.perldb or ~/.perldb
125 under Unix), which can set important options. (A subroutine
126 (&afterinit) can be defined here as well; it is executed after the
127 debugger completes its own initialization.)
128 After the rc file is read, the debugger reads the PERLDB_OPTS
130 environment variable and uses it to set debugger options. The contents
131 of this variable are treated as if they were the argument of an "o ..."
132 debugger command (q.v. in "Options" in perldebug).
133 Debugger internal variables
135 In addition to the file and subroutine-related variables mentioned
137 above, the debugger also maintains various magical internal variables.
138 · @DB::dbline is an alias for "@{"::_<current_file"}", which holds
140 the lines of the currently-selected file (compiled by Perl), either
141 explicitly chosen with the debugger's "f" command, or implicitly by
142 flow of execution.
143 Values in this array are magical in numeric context: they compare
145 equal to zero only if the line is not breakable.
146 · %DB::dbline, is an alias for "%{"::_<current_file"}", which
148 contains breakpoints and actions keyed by line number in the
149 currently-selected file, either explicitly chosen with the
150 debugger's "f" command, or implicitly by flow of execution.
151 As previously noted, individual entries (as opposed to the whole
153 hash) are settable. Perl only cares about Boolean true here,
154 although the values used by perl5db.pl have the form
155 "$break_condition\0$action".
156 Debugger customization functions
158 Some functions are provided to simplify customization.
160 · See "Configurable Options" in perldebug for a description of
162 options parsed by "DB::parse_options(string)".
163 · "DB::dump_trace(skip[,count])" skips the specified number of frames
165 and returns a list containing information about the calling frames
166 (all of them, if "count" is missing). Each entry is reference to a
167 hash with keys "context" (either ".", "$", or "@"), "sub"
168 (subroutine name, or info about "eval"), "args" ("undef" or a
169 reference to an array), "file", and "line".
170 · "DB::print_trace(FH, skip[, count[, short]])" prints formatted info
172 about caller frames. The last two functions may be convenient as
173 arguments to "<", "<<" commands.
174 Note that any variables and functions that are not documented in this
176 manpages (or in perldebug) are considered for internal use only, and as
177 such are subject to change without notice.
178
180 The "frame" option can be used to control the output of frame
181 information. For example, contrast this expression trace:
182 $ perl -de 42
184 Stack dump during die enabled outside of evals.
185 Loading DB routines from perl5db.pl patch level 0.94
187 Emacs support available.
188 Enter h or `h h' for help.
190 main::(-e:1): 0
192 DB<1> sub foo { 14 }
193 DB<2> sub bar { 3 }
195 DB<3> t print foo() * bar()
197 main::((eval 172):3): print foo() + bar();
198 main::foo((eval 168):2):
199 main::bar((eval 170):2):
200 42
201 with this one, once the "o"ption "frame=2" has been set:
203 DB<4> o f=2
205 frame = '2'
206 DB<5> t print foo() * bar()
207 3: foo() * bar()
208 entering main::foo
209 2: sub foo { 14 };
210 exited main::foo
211 entering main::bar
212 2: sub bar { 3 };
213 exited main::bar
214 42
215 By way of demonstration, we present below a laborious listing resulting
217 from setting your "PERLDB_OPTS" environment variable to the value "f=n
218 N", and running perl -d -V from the command line. Examples use various
219 values of "n" are shown to give you a feel for the difference between
220 settings. Long those it may be, this is not a complete listing, but
221 only excerpts.
222 1.
224 entering main::BEGIN
225 entering Config::BEGIN
226 Package lib/Exporter.pm.
227 Package lib/Carp.pm.
228 Package lib/Config.pm.
229 entering Config::TIEHASH
230 entering Exporter::import
231 entering Exporter::export
232 entering Config::myconfig
233 entering Config::FETCH
234 entering Config::FETCH
235 entering Config::FETCH
236 entering Config::FETCH
237 2.
239 entering main::BEGIN
240 entering Config::BEGIN
241 Package lib/Exporter.pm.
242 Package lib/Carp.pm.
243 exited Config::BEGIN
244 Package lib/Config.pm.
245 entering Config::TIEHASH
246 exited Config::TIEHASH
247 entering Exporter::import
248 entering Exporter::export
249 exited Exporter::export
250 exited Exporter::import
251 exited main::BEGIN
252 entering Config::myconfig
253 entering Config::FETCH
254 exited Config::FETCH
255 entering Config::FETCH
256 exited Config::FETCH
257 entering Config::FETCH
258 3.
260 in $=main::BEGIN() from /dev/null:0
261 in $=Config::BEGIN() from lib/Config.pm:2
262 Package lib/Exporter.pm.
263 Package lib/Carp.pm.
264 Package lib/Config.pm.
265 in $=Config::TIEHASH('Config') from lib/Config.pm:644
266 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
267 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from li
268 in @=Config::myconfig() from /dev/null:0
269 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
270 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
271 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
272 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
273 in $=Config::FETCH(ref(Config), 'osname') from lib/Config.pm:574
274 in $=Config::FETCH(ref(Config), 'osvers') from lib/Config.pm:574
275 4.
277 in $=main::BEGIN() from /dev/null:0
278 in $=Config::BEGIN() from lib/Config.pm:2
279 Package lib/Exporter.pm.
280 Package lib/Carp.pm.
281 out $=Config::BEGIN() from lib/Config.pm:0
282 Package lib/Config.pm.
283 in $=Config::TIEHASH('Config') from lib/Config.pm:644
284 out $=Config::TIEHASH('Config') from lib/Config.pm:644
285 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
286 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
287 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
288 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
289 out $=main::BEGIN() from /dev/null:0
290 in @=Config::myconfig() from /dev/null:0
291 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
292 out $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
293 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
294 out $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
295 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
296 out $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
297 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
298 5.
300 in $=main::BEGIN() from /dev/null:0
301 in $=Config::BEGIN() from lib/Config.pm:2
302 Package lib/Exporter.pm.
303 Package lib/Carp.pm.
304 out $=Config::BEGIN() from lib/Config.pm:0
305 Package lib/Config.pm.
306 in $=Config::TIEHASH('Config') from lib/Config.pm:644
307 out $=Config::TIEHASH('Config') from lib/Config.pm:644
308 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
309 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
310 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
311 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
312 out $=main::BEGIN() from /dev/null:0
313 in @=Config::myconfig() from /dev/null:0
314 in $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
315 out $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
316 in $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
317 out $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
318 6.
320 in $=CODE(0x15eca4)() from /dev/null:0
321 in $=CODE(0x182528)() from lib/Config.pm:2
322 Package lib/Exporter.pm.
323 out $=CODE(0x182528)() from lib/Config.pm:0
324 scalar context return from CODE(0x182528): undef
325 Package lib/Config.pm.
326 in $=Config::TIEHASH('Config') from lib/Config.pm:628
327 out $=Config::TIEHASH('Config') from lib/Config.pm:628
328 scalar context return from Config::TIEHASH: empty hash
329 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
330 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
331 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
332 scalar context return from Exporter::export: ''
333 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
334 scalar context return from Exporter::import: ''
335 In all cases shown above, the line indentation shows the call tree. If
337 bit 2 of "frame" is set, a line is printed on exit from a subroutine as
338 well. If bit 4 is set, the arguments are printed along with the caller
339 info. If bit 8 is set, the arguments are printed even if they are tied
340 or references. If bit 16 is set, the return value is printed, too.
341 When a package is compiled, a line like this
343 Package lib/Carp.pm.
345 is printed with proper indentation.
347
349 There are two ways to enable debugging output for regular expressions.
350 If your perl is compiled with "-DDEBUGGING", you may use the -Dr flag
352 on the command line.
353 Otherwise, one can "use re 'debug'", which has effects at compile time
355 and run time. It is not lexically scoped.
356 Compile-time output
358 The debugging output at compile time looks like this:
359 Compiling REx `[bc]d(ef*g)+h[ij]k$'
361 size 45 Got 364 bytes for offset annotations.
362 first at 1
363 rarest char g at 0
364 rarest char d at 0
365 1: ANYOF[bc](12)
366 12: EXACT <d>(14)
367 14: CURLYX[0] {1,32767}(28)
368 16: OPEN1(18)
369 18: EXACT <e>(20)
370 20: STAR(23)
371 21: EXACT <f>(0)
372 23: EXACT <g>(25)
373 25: CLOSE1(27)
374 27: WHILEM[1/1](0)
375 28: NOTHING(29)
376 29: EXACT <h>(31)
377 31: ANYOF[ij](42)
378 42: EXACT <k>(44)
379 44: EOL(45)
380 45: END(0)
381 anchored `de' at 1 floating `gh' at 3..2147483647 (checking floating)
382 stclass `ANYOF[bc]' minlen 7
383 Offsets: [45]
384 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
385 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
386 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
387 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]
388 Omitting $` $& $' support.
389 The first line shows the pre-compiled form of the regex. The second
391 shows the size of the compiled form (in arbitrary units, usually 4-byte
392 words) and the total number of bytes allocated for the offset/length
393 table, usually 4+"size"*8. The next line shows the label id of the
394 first node that does a match.
395 The
397 anchored `de' at 1 floating `gh' at 3..2147483647 (checking floating)
399 stclass `ANYOF[bc]' minlen 7
400 line (split into two lines above) contains optimizer information. In
402 the example shown, the optimizer found that the match should contain a
403 substring "de" at offset 1, plus substring "gh" at some offset between
404 3 and infinity. Moreover, when checking for these substrings (to
405 abandon impossible matches quickly), Perl will check for the substring
406 "gh" before checking for the substring "de". The optimizer may also
407 use the knowledge that the match starts (at the "first" id) with a
408 character class, and no string shorter than 7 characters can possibly
409 match.
410 The fields of interest which may appear in this line are
412 "anchored" STRING "at" POS
414 "floating" STRING "at" POS1..POS2
415 See above.
416 "matching floating/anchored"
418 Which substring to check first.
419 "minlen"
421 The minimal length of the match.
422 "stclass" TYPE
424 Type of first matching node.
425 "noscan"
427 Don't scan for the found substrings.
428 "isall"
430 Means that the optimizer information is all that the regular
431 expression contains, and thus one does not need to enter the regex
432 engine at all.
433 "GPOS"
435 Set if the pattern contains "\G".
436 "plus"
438 Set if the pattern starts with a repeated char (as in "x+y").
439 "implicit"
441 Set if the pattern starts with ".*".
442 "with eval"
444 Set if the pattern contain eval-groups, such as "(?{ code })" and
445 "(??{ code })".
446 "anchored(TYPE)"
448 If the pattern may match only at a handful of places, (with "TYPE"
449 being "BOL", "MBOL", or "GPOS". See the table below.
450 If a substring is known to match at end-of-line only, it may be
452 followed by "$", as in "floating `k'$".
453 The optimizer-specific information is used to avoid entering (a slow)
455 regex engine on strings that will not definitely match. If the "isall"
456 flag is set, a call to the regex engine may be avoided even when the
457 optimizer found an appropriate place for the match.
458 Above the optimizer section is the list of nodes of the compiled form
460 of the regex. Each line has format
461 " "id: TYPE OPTIONAL-INFO (next-id)
463 Types of nodes
465 Here are the possible types, with short descriptions:
466 # TYPE arg-description [num-args] [longjump-len] DESCRIPTION
468 # Exit points
470 END no End of program.
471 SUCCEED no Return from a subroutine, basically.
472 # Anchors:
474 BOL no Match "" at beginning of line.
475 MBOL no Same, assuming multiline.
476 SBOL no Same, assuming singleline.
477 EOS no Match "" at end of string.
478 EOL no Match "" at end of line.
479 MEOL no Same, assuming multiline.
480 SEOL no Same, assuming singleline.
481 BOUND no Match "" at any word boundary
482 BOUNDL no Match "" at any word boundary
483 NBOUND no Match "" at any word non-boundary
484 NBOUNDL no Match "" at any word non-boundary
485 GPOS no Matches where last m//g left off.
486 # [Special] alternatives
488 ANY no Match any one character (except newline).
489 SANY no Match any one character.
490 ANYOF sv Match character in (or not in) this class.
491 ALNUM no Match any alphanumeric character
492 ALNUML no Match any alphanumeric char in locale
493 NALNUM no Match any non-alphanumeric character
494 NALNUML no Match any non-alphanumeric char in locale
495 SPACE no Match any whitespace character
496 SPACEL no Match any whitespace char in locale
497 NSPACE no Match any non-whitespace character
498 NSPACEL no Match any non-whitespace char in locale
499 DIGIT no Match any numeric character
500 NDIGIT no Match any non-numeric character
501 # BRANCH The set of branches constituting a single choice are hooked
503 # together with their "next" pointers, since precedence prevents
504 # anything being concatenated to any individual branch. The
505 # "next" pointer of the last BRANCH in a choice points to the
506 # thing following the whole choice. This is also where the
507 # final "next" pointer of each individual branch points; each
508 # branch starts with the operand node of a BRANCH node.
509 #
510 BRANCH node Match this alternative, or the next...
511 # BACK Normal "next" pointers all implicitly point forward; BACK
513 # exists to make loop structures possible.
514 # not used
515 BACK no Match "", "next" ptr points backward.
516 # Literals
518 EXACT sv Match this string (preceded by length).
519 EXACTF sv Match this string, folded (prec. by length).
520 EXACTFL sv Match this string, folded in locale (w/len).
521 # Do nothing
523 NOTHING no Match empty string.
524 # A variant of above which delimits a group, thus stops optimizations
525 TAIL no Match empty string. Can jump here from outside.
526 # STAR,PLUS '?', and complex '*' and '+', are implemented as circular
528 # BRANCH structures using BACK. Simple cases (one character
529 # per match) are implemented with STAR and PLUS for speed
530 # and to minimize recursive plunges.
531 #
532 STAR node Match this (simple) thing 0 or more times.
533 PLUS node Match this (simple) thing 1 or more times.
534 CURLY sv 2 Match this simple thing {n,m} times.
536 CURLYN no 2 Match next-after-this simple thing
537 # {n,m} times, set parens.
538 CURLYM no 2 Match this medium-complex thing {n,m} times.
539 CURLYX sv 2 Match this complex thing {n,m} times.
540 # This terminator creates a loop structure for CURLYX
542 WHILEM no Do curly processing and see if rest matches.
543 # OPEN,CLOSE,GROUPP ...are numbered at compile time.
545 OPEN num 1 Mark this point in input as start of #n.
546 CLOSE num 1 Analogous to OPEN.
547 REF num 1 Match some already matched string
549 REFF num 1 Match already matched string, folded
550 REFFL num 1 Match already matched string, folded in loc.
551 # grouping assertions
553 IFMATCH off 1 2 Succeeds if the following matches.
554 UNLESSM off 1 2 Fails if the following matches.
555 SUSPEND off 1 1 "Independent" sub-regex.
556 IFTHEN off 1 1 Switch, should be preceded by switcher .
557 GROUPP num 1 Whether the group matched.
558 # Support for long regex
560 LONGJMP off 1 1 Jump far away.
561 BRANCHJ off 1 1 BRANCH with long offset.
562 # The heavy worker
564 EVAL evl 1 Execute some Perl code.
565 # Modifiers
567 MINMOD no Next operator is not greedy.
568 LOGICAL no Next opcode should set the flag only.
569 # This is not used yet
571 RENUM off 1 1 Group with independently numbered parens.
572 # This is not really a node, but an optimized away piece of a "long" node.
574 # To simplify debugging output, we mark it as if it were a node
575 OPTIMIZED off Placeholder for dump.
576 Following the optimizer information is a dump of the offset/length
578 table, here split across several lines:
579 Offsets: [45]
581 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
582 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
583 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
584 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]
585 The first line here indicates that the offset/length table contains 45
587 entries. Each entry is a pair of integers, denoted by
588 "offset[length]". Entries are numbered starting with 1, so entry #1
589 here is "1[4]" and entry #12 is "5[1]". "1[4]" indicates that the node
590 labeled "1:" (the "1: ANYOF[bc]") begins at character position 1 in the
591 pre-compiled form of the regex, and has a length of 4 characters.
592 "5[1]" in position 12 indicates that the node labeled "12:" (the "12:
593 EXACT <d>") begins at character position 5 in the pre-compiled form of
594 the regex, and has a length of 1 character. "12[1]" in position 14
595 indicates that the node labeled "14:" (the "14: CURLYX[0] {1,32767}")
596 begins at character position 12 in the pre-compiled form of the regex,
597 and has a length of 1 character---that is, it corresponds to the "+"
598 symbol in the precompiled regex.
599 "0[0]" items indicate that there is no corresponding node.
601 Run-time output
603 First of all, when doing a match, one may get no run-time output even
604 if debugging is enabled. This means that the regex engine was never
605 entered and that all of the job was therefore done by the optimizer.
606 If the regex engine was entered, the output may look like this:
608 Matching `[bc]d(ef*g)+h[ij]k$' against `abcdefg__gh__'
610 Setting an EVAL scope, savestack=3
611 2 <ab> <cdefg__gh_> | 1: ANYOF
612 3 <abc> <defg__gh_> | 11: EXACT <d>
613 4 <abcd> <efg__gh_> | 13: CURLYX {1,32767}
614 4 <abcd> <efg__gh_> | 26: WHILEM
615 0 out of 1..32767 cc=effff31c
616 4 <abcd> <efg__gh_> | 15: OPEN1
617 4 <abcd> <efg__gh_> | 17: EXACT <e>
618 5 <abcde> <fg__gh_> | 19: STAR
619 EXACT <f> can match 1 times out of 32767...
620 Setting an EVAL scope, savestack=3
621 6 <bcdef> <g__gh__> | 22: EXACT <g>
622 7 <bcdefg> <__gh__> | 24: CLOSE1
623 7 <bcdefg> <__gh__> | 26: WHILEM
624 1 out of 1..32767 cc=effff31c
625 Setting an EVAL scope, savestack=12
626 7 <bcdefg> <__gh__> | 15: OPEN1
627 7 <bcdefg> <__gh__> | 17: EXACT <e>
628 restoring \1 to 4(4)..7
629 failed, try continuation...
630 7 <bcdefg> <__gh__> | 27: NOTHING
631 7 <bcdefg> <__gh__> | 28: EXACT <h>
632 failed...
633 failed...
634 The most significant information in the output is about the particular
636 node of the compiled regex that is currently being tested against the
637 target string. The format of these lines is
638 " "STRING-OFFSET <PRE-STRING> <POST-STRING> |ID: TYPE
640 The TYPE info is indented with respect to the backtracking level.
642 Other incidental information appears interspersed within.
643
645 Perl is a profligate wastrel when it comes to memory use. There is a
646 saying that to estimate memory usage of Perl, assume a reasonable
647 algorithm for memory allocation, multiply that estimate by 10, and
648 while you still may miss the mark, at least you won't be quite so
649 astonished. This is not absolutely true, but may provide a good grasp
650 of what happens.
651 Assume that an integer cannot take less than 20 bytes of memory, a
653 float cannot take less than 24 bytes, a string cannot take less than 32
654 bytes (all these examples assume 32-bit architectures, the result are
655 quite a bit worse on 64-bit architectures). If a variable is accessed
656 in two of three different ways (which require an integer, a float, or a
657 string), the memory footprint may increase yet another 20 bytes. A
658 sloppy malloc(3) implementation can inflate these numbers dramatically.
659 On the opposite end of the scale, a declaration like
661 sub foo;
663 may take up to 500 bytes of memory, depending on which release of Perl
665 you're running.
666 Anecdotal estimates of source-to-compiled code bloat suggest an
668 eightfold increase. This means that the compiled form of reasonable
669 (normally commented, properly indented etc.) code will take about eight
670 times more space in memory than the code took on disk.
671 The -DL command-line switch is obsolete since circa Perl 5.6.0 (it was
673 available only if Perl was built with "-DDEBUGGING"). The switch was
674 used to track Perl's memory allocations and possible memory leaks.
675 These days the use of malloc debugging tools like Purify or valgrind is
676 suggested instead. See also "PERL_MEM_LOG" in perlhack.
677 One way to find out how much memory is being used by Perl data
679 structures is to install the Devel::Size module from CPAN: it gives you
680 the minimum number of bytes required to store a particular data
681 structure. Please be mindful of the difference between the size() and
682 total_size().
683 If Perl has been compiled using Perl's malloc you can analyze Perl
685 memory usage by setting the $ENV{PERL_DEBUG_MSTATS}.
686 Using $ENV{PERL_DEBUG_MSTATS}
688 If your perl is using Perl's malloc() and was compiled with the
689 necessary switches (this is the default), then it will print memory
690 usage statistics after compiling your code when
691 "$ENV{PERL_DEBUG_MSTATS} > 1", and before termination of the program
692 when "$ENV{PERL_DEBUG_MSTATS} >= 1". The report format is similar to
693 the following example:
694 $ PERL_DEBUG_MSTATS=2 perl -e "require Carp"
696 Memory allocation statistics after compilation: (buckets 4(4)..8188(8192)
697 14216 free: 130 117 28 7 9 0 2 2 1 0 0
698 437 61 36 0 5
699 60924 used: 125 137 161 55 7 8 6 16 2 0 1
700 74 109 304 84 20
701 Total sbrk(): 77824/21:119. Odd ends: pad+heads+chain+tail: 0+636+0+2048.
702 Memory allocation statistics after execution: (buckets 4(4)..8188(8192)
703 30888 free: 245 78 85 13 6 2 1 3 2 0 1
704 315 162 39 42 11
705 175816 used: 265 176 1112 111 26 22 11 27 2 1 1
706 196 178 1066 798 39
707 Total sbrk(): 215040/47:145. Odd ends: pad+heads+chain+tail: 0+2192+0+6144.
708 It is possible to ask for such a statistic at arbitrary points in your
710 execution using the mstat() function out of the standard Devel::Peek
711 module.
712 Here is some explanation of that format:
714 "buckets SMALLEST(APPROX)..GREATEST(APPROX)"
716 Perl's malloc() uses bucketed allocations. Every request is
717 rounded up to the closest bucket size available, and a bucket is
718 taken from the pool of buckets of that size.
719 The line above describes the limits of buckets currently in use.
721 Each bucket has two sizes: memory footprint and the maximal size of
722 user data that can fit into this bucket. Suppose in the above
723 example that the smallest bucket were size 4. The biggest bucket
724 would have usable size 8188, and the memory footprint would be
725 8192.
726 In a Perl built for debugging, some buckets may have negative
728 usable size. This means that these buckets cannot (and will not)
729 be used. For larger buckets, the memory footprint may be one page
730 greater than a power of 2. If so, case the corresponding power of
731 two is printed in the "APPROX" field above.
732 Free/Used
734 The 1 or 2 rows of numbers following that correspond to the number
735 of buckets of each size between "SMALLEST" and "GREATEST". In the
736 first row, the sizes (memory footprints) of buckets are powers of
737 two--or possibly one page greater. In the second row, if present,
738 the memory footprints of the buckets are between the memory
739 footprints of two buckets "above".
740 For example, suppose under the previous example, the memory
742 footprints were
743 free: 8 16 32 64 128 256 512 1024 2048 4096 8192
745 4 12 24 48 80
746 With non-"DEBUGGING" perl, the buckets starting from 128 have a
748 4-byte overhead, and thus an 8192-long bucket may take up to
749 8188-byte allocations.
750 "Total sbrk(): SBRKed/SBRKs:CONTINUOUS"
752 The first two fields give the total amount of memory perl sbrk(2)ed
753 (ess-broken? :-) and number of sbrk(2)s used. The third number is
754 what perl thinks about continuity of returned chunks. So long as
755 this number is positive, malloc() will assume that it is probable
756 that sbrk(2) will provide continuous memory.
757 Memory allocated by external libraries is not counted.
759 "pad: 0"
761 The amount of sbrk(2)ed memory needed to keep buckets aligned.
762 "heads: 2192"
764 Although memory overhead of bigger buckets is kept inside the
765 bucket, for smaller buckets, it is kept in separate areas. This
766 field gives the total size of these areas.
767 "chain: 0"
769 malloc() may want to subdivide a bigger bucket into smaller
770 buckets. If only a part of the deceased bucket is left
771 unsubdivided, the rest is kept as an element of a linked list.
772 This field gives the total size of these chunks.
773 "tail: 6144"
775 To minimize the number of sbrk(2)s, malloc() asks for more memory.
776 This field gives the size of the yet unused part, which is
777 sbrk(2)ed, but never touched.
778
780 perldebug, perlguts, perlrun re, and Devel::DProf.
781perl v5.10.1 2009-02-12 PERLDEBGUTS(1)