Modules

  • ABCDE
  • FGHIL
  • MNOPS
  • TUX

Tools

perldebguts

Perl 5 version 18.2 documentation
Recently read

perldebguts

NAME

perldebguts - Guts of Perl debugging

DESCRIPTION

This is not perldebug, which tells you how to use the debugger. This manpage describes low-level details concerning the debugger's internals, which range from difficult to impossible to understand for anyone who isn't incredibly intimate with Perl's guts. Caveat lector.

Debugger Internals

Perl has special debugging hooks at compile-time and run-time used to create debugging environments. These hooks are not to be confused with the perl -Dxxx command described in perlrun, which is usable only if a special Perl is built per the instructions in the INSTALL podpage in the Perl source tree.

For example, whenever you call Perl's built-in caller function from the package DB , the arguments that the corresponding stack frame was called with are copied to the @DB::args array. These mechanisms are enabled by calling Perl with the -d switch. Specifically, the following additional features are enabled (cf. $^P in perlvar):

  • Perl inserts the contents of $ENV{PERL5DB} (or BEGIN {require 'perl5db.pl'} if not present) before the first line of your program.

  • Each array @{"_<$filename"} holds the lines of $filename for a file compiled by Perl. The same is also true for evaled strings that contain subroutines, or which are currently being executed. The $filename for evaled strings looks like (eval 34) .

    Values in this array are magical in numeric context: they compare equal to zero only if the line is not breakable.

  • Each hash %{"_<$filename"} contains breakpoints and actions keyed by line number. Individual entries (as opposed to the whole hash) are settable. Perl only cares about Boolean true here, although the values used by perl5db.pl have the form "$break_condition\0$action" .

    The same holds for evaluated strings that contain subroutines, or which are currently being executed. The $filename for evaled strings looks like (eval 34) .

  • Each scalar ${"_<$filename"} contains "_<$filename" . This is also the case for evaluated strings that contain subroutines, or which are currently being executed. The $filename for evaled strings looks like (eval 34) .

  • After each required file is compiled, but before it is executed, DB::postponed(*{"_<$filename"}) is called if the subroutine DB::postponed exists. Here, the $filename is the expanded name of the required file, as found in the values of %INC.

  • After each subroutine subname is compiled, the existence of $DB::postponed{subname} is checked. If this key exists, DB::postponed(subname) is called if the DB::postponed subroutine also exists.

  • A hash %DB::sub is maintained, whose keys are subroutine names and whose values have the form filename:startline-endline . filename has the form (eval 34) for subroutines defined inside evals.

  • When the execution of your program reaches a point that can hold a breakpoint, the DB::DB() subroutine is called if any of the variables $DB::trace , $DB::single , or $DB::signal is true. These variables are not localizable. This feature is disabled when executing inside DB::DB() , including functions called from it unless $^D & (1<<30) is true.

  • When execution of the program reaches a subroutine call, a call to &DB::sub (args) is made instead, with $DB::sub holding the name of the called subroutine. (This doesn't happen if the subroutine was compiled in the DB package.)

Note that if &DB::sub needs external data for it to work, no subroutine call is possible without it. As an example, the standard debugger's &DB::sub depends on the $DB::deep variable (it defines how many levels of recursion deep into the debugger you can go before a mandatory break). If $DB::deep is not defined, subroutine calls are not possible, even though &DB::sub exists.

Writing Your Own Debugger

Environment Variables

The PERL5DB environment variable can be used to define a debugger. For example, the minimal "working" debugger (it actually doesn't do anything) consists of one line:

  1. sub DB::DB {}

It can easily be defined like this:

  1. $ PERL5DB="sub DB::DB {}" perl -d your-script

Another brief debugger, slightly more useful, can be created with only the line:

  1. sub DB::DB {print ++$i; scalar <STDIN>}

This debugger prints a number which increments for each statement encountered and waits for you to hit a newline before continuing to the next statement.

The following debugger is actually useful:

  1. {
  2. package DB;
  3. sub DB {}
  4. sub sub {print ++$i, " $sub\n"; &$sub}
  5. }

It prints the sequence number of each subroutine call and the name of the called subroutine. Note that &DB::sub is being compiled into the package DB through the use of the package directive.

When it starts, the debugger reads your rc file (./.perldb or ~/.perldb under Unix), which can set important options. (A subroutine (&afterinit ) can be defined here as well; it is executed after the debugger completes its own initialization.)

After the rc file is read, the debugger reads the PERLDB_OPTS environment variable and uses it to set debugger options. The contents of this variable are treated as if they were the argument of an o ... debugger command (q.v. in Configurable Options in perldebug).

Debugger Internal Variables

In addition to the file and subroutine-related variables mentioned above, the debugger also maintains various magical internal variables.

  • @DB::dbline is an alias for @{"::_<current_file"} , which holds the lines of the currently-selected file (compiled by Perl), either explicitly chosen with the debugger's f command, or implicitly by flow of execution.

    Values in this array are magical in numeric context: they compare equal to zero only if the line is not breakable.

  • %DB::dbline is an alias for %{"::_<current_file"} , which contains breakpoints and actions keyed by line number in the currently-selected file, either explicitly chosen with the debugger's f command, or implicitly by flow of execution.

    As previously noted, individual entries (as opposed to the whole hash) are settable. Perl only cares about Boolean true here, although the values used by perl5db.pl have the form "$break_condition\0$action" .

Debugger Customization Functions

Some functions are provided to simplify customization.

  • See Configurable Options in perldebug for a description of options parsed by DB::parse_options(string) .

  • DB::dump_trace(skip[,count]) skips the specified number of frames and returns a list containing information about the calling frames (all of them, if count is missing). Each entry is reference to a hash with keys context (either ., $ , or @ ), sub (subroutine name, or info about eval), args (undef or a reference to an array), file , and line .

  • DB::print_trace(FH, skip[, count[, short]]) prints formatted info about caller frames. The last two functions may be convenient as arguments to < , << commands.

Note that any variables and functions that are not documented in this manpages (or in perldebug) are considered for internal use only, and as such are subject to change without notice.

Frame Listing Output Examples

The frame option can be used to control the output of frame information. For example, contrast this expression trace:

  1. $ perl -de 42
  2. Stack dump during die enabled outside of evals.
  3. Loading DB routines from perl5db.pl patch level 0.94
  4. Emacs support available.
  5. Enter h or 'h h' for help.
  6. main::(-e:1): 0
  7. DB<1> sub foo { 14 }
  8. DB<2> sub bar { 3 }
  9. DB<3> t print foo() * bar()
  10. main::((eval 172):3): print foo() + bar();
  11. main::foo((eval 168):2):
  12. main::bar((eval 170):2):
  13. 42

with this one, once the o ption frame=2 has been set:

  1. DB<4> o f=2
  2. frame = '2'
  3. DB<5> t print foo() * bar()
  4. 3: foo() * bar()
  5. entering main::foo
  6. 2: sub foo { 14 };
  7. exited main::foo
  8. entering main::bar
  9. 2: sub bar { 3 };
  10. exited main::bar
  11. 42

By way of demonstration, we present below a laborious listing resulting from setting your PERLDB_OPTS environment variable to the value f=n N , and running perl -d -V from the command line. Examples using various values of n are shown to give you a feel for the difference between settings. Long though it may be, this is not a complete listing, but only excerpts.

1
  1. entering main::BEGIN
  2. entering Config::BEGIN
  3. Package lib/Exporter.pm.
  4. Package lib/Carp.pm.
  5. Package lib/Config.pm.
  6. entering Config::TIEHASH
  7. entering Exporter::import
  8. entering Exporter::export
  9. entering Config::myconfig
  10. entering Config::FETCH
  11. entering Config::FETCH
  12. entering Config::FETCH
  13. entering Config::FETCH
2
  1. entering main::BEGIN
  2. entering Config::BEGIN
  3. Package lib/Exporter.pm.
  4. Package lib/Carp.pm.
  5. exited Config::BEGIN
  6. Package lib/Config.pm.
  7. entering Config::TIEHASH
  8. exited Config::TIEHASH
  9. entering Exporter::import
  10. entering Exporter::export
  11. exited Exporter::export
  12. exited Exporter::import
  13. exited main::BEGIN
  14. entering Config::myconfig
  15. entering Config::FETCH
  16. exited Config::FETCH
  17. entering Config::FETCH
  18. exited Config::FETCH
  19. entering Config::FETCH
3
  1. in $=main::BEGIN() from /dev/null:0
  2. in $=Config::BEGIN() from lib/Config.pm:2
  3. Package lib/Exporter.pm.
  4. Package lib/Carp.pm.
  5. Package lib/Config.pm.
  6. in $=Config::TIEHASH('Config') from lib/Config.pm:644
  7. in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  8. in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from li
  9. in @=Config::myconfig() from /dev/null:0
  10. in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
  11. in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
  12. in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
  13. in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
  14. in $=Config::FETCH(ref(Config), 'osname') from lib/Config.pm:574
  15. in $=Config::FETCH(ref(Config), 'osvers') from lib/Config.pm:574
4
  1. in $=main::BEGIN() from /dev/null:0
  2. in $=Config::BEGIN() from lib/Config.pm:2
  3. Package lib/Exporter.pm.
  4. Package lib/Carp.pm.
  5. out $=Config::BEGIN() from lib/Config.pm:0
  6. Package lib/Config.pm.
  7. in $=Config::TIEHASH('Config') from lib/Config.pm:644
  8. out $=Config::TIEHASH('Config') from lib/Config.pm:644
  9. in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  10. in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
  11. out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
  12. out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  13. out $=main::BEGIN() from /dev/null:0
  14. in @=Config::myconfig() from /dev/null:0
  15. in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
  16. out $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
  17. in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
  18. out $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
  19. in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
  20. out $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
  21. in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
5
  1. in $=main::BEGIN() from /dev/null:0
  2. in $=Config::BEGIN() from lib/Config.pm:2
  3. Package lib/Exporter.pm.
  4. Package lib/Carp.pm.
  5. out $=Config::BEGIN() from lib/Config.pm:0
  6. Package lib/Config.pm.
  7. in $=Config::TIEHASH('Config') from lib/Config.pm:644
  8. out $=Config::TIEHASH('Config') from lib/Config.pm:644
  9. in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  10. in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
  11. out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
  12. out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  13. out $=main::BEGIN() from /dev/null:0
  14. in @=Config::myconfig() from /dev/null:0
  15. in $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
  16. out $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
  17. in $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
  18. out $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
6
  1. in $=CODE(0x15eca4)() from /dev/null:0
  2. in $=CODE(0x182528)() from lib/Config.pm:2
  3. Package lib/Exporter.pm.
  4. out $=CODE(0x182528)() from lib/Config.pm:0
  5. scalar context return from CODE(0x182528): undef
  6. Package lib/Config.pm.
  7. in $=Config::TIEHASH('Config') from lib/Config.pm:628
  8. out $=Config::TIEHASH('Config') from lib/Config.pm:628
  9. scalar context return from Config::TIEHASH: empty hash
  10. in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  11. in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
  12. out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
  13. scalar context return from Exporter::export: ''
  14. out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
  15. scalar context return from Exporter::import: ''

In all cases shown above, the line indentation shows the call tree. If bit 2 of frame is set, a line is printed on exit from a subroutine as well. If bit 4 is set, the arguments are printed along with the caller info. If bit 8 is set, the arguments are printed even if they are tied or references. If bit 16 is set, the return value is printed, too.

When a package is compiled, a line like this

  1. Package lib/Carp.pm.

is printed with proper indentation.

Debugging Regular Expressions

There are two ways to enable debugging output for regular expressions.

If your perl is compiled with -DDEBUGGING , you may use the -Dr flag on the command line.

Otherwise, one can use re 'debug' , which has effects at compile time and run time. Since Perl 5.9.5, this pragma is lexically scoped.

Compile-time Output

The debugging output at compile time looks like this:

  1. Compiling REx '[bc]d(ef*g)+h[ij]k$'
  2. size 45 Got 364 bytes for offset annotations.
  3. first at 1
  4. rarest char g at 0
  5. rarest char d at 0
  6. 1: ANYOF[bc](12)
  7. 12: EXACT <d>(14)
  8. 14: CURLYX[0] {1,32767}(28)
  9. 16: OPEN1(18)
  10. 18: EXACT <e>(20)
  11. 20: STAR(23)
  12. 21: EXACT <f>(0)
  13. 23: EXACT <g>(25)
  14. 25: CLOSE1(27)
  15. 27: WHILEM[1/1](0)
  16. 28: NOTHING(29)
  17. 29: EXACT <h>(31)
  18. 31: ANYOF[ij](42)
  19. 42: EXACT <k>(44)
  20. 44: EOL(45)
  21. 45: END(0)
  22. anchored 'de' at 1 floating 'gh' at 3..2147483647 (checking floating)
  23. stclass 'ANYOF[bc]' minlen 7
  24. Offsets: [45]
  25. 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
  26. 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
  27. 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
  28. 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]
  29. Omitting $` $& $' support.

The first line shows the pre-compiled form of the regex. The second shows the size of the compiled form (in arbitrary units, usually 4-byte words) and the total number of bytes allocated for the offset/length table, usually 4+size *8. The next line shows the label id of the first node that does a match.

The

  1. anchored 'de' at 1 floating 'gh' at 3..2147483647 (checking floating)
  2. stclass 'ANYOF[bc]' minlen 7

line (split into two lines above) contains optimizer information. In the example shown, the optimizer found that the match should contain a substring de at offset 1, plus substring gh at some offset between 3 and infinity. Moreover, when checking for these substrings (to abandon impossible matches quickly), Perl will check for the substring gh before checking for the substring de . The optimizer may also use the knowledge that the match starts (at the first id) with a character class, and no string shorter than 7 characters can possibly match.

The fields of interest which may appear in this line are

  • anchored STRING at POS
  • floating STRING at POS1..POS2

    See above.

  • matching floating/anchored

    Which substring to check first.

  • minlen

    The minimal length of the match.

  • stclass TYPE

    Type of first matching node.

  • noscan

    Don't scan for the found substrings.

  • isall

    Means that the optimizer information is all that the regular expression contains, and thus one does not need to enter the regex engine at all.

  • GPOS

    Set if the pattern contains \G .

  • plus

    Set if the pattern starts with a repeated char (as in x+y).

  • implicit

    Set if the pattern starts with .*.

  • with eval

    Set if the pattern contain eval-groups, such as (?{ code }) and (??{ code }) .

  • anchored(TYPE)

    If the pattern may match only at a handful of places, with TYPE being BOL , MBOL , or GPOS . See the table below.

If a substring is known to match at end-of-line only, it may be followed by $ , as in floating 'k'$ .

The optimizer-specific information is used to avoid entering (a slow) regex engine on strings that will not definitely match. If the isall flag is set, a call to the regex engine may be avoided even when the optimizer found an appropriate place for the match.

Above the optimizer section is the list of nodes of the compiled form of the regex. Each line has format

id: TYPE OPTIONAL-INFO (next-id)

Types of Nodes

Here are the possible types, with short descriptions:

  1. # TYPE arg-description [num-args] [longjump-len] DESCRIPTION
  2. # Exit points
  3. END no End of program.
  4. SUCCEED no Return from a subroutine, basically.
  5. # Anchors:
  6. BOL no Match "" at beginning of line.
  7. MBOL no Same, assuming multiline.
  8. SBOL no Same, assuming singleline.
  9. EOS no Match "" at end of string.
  10. EOL no Match "" at end of line.
  11. MEOL no Same, assuming multiline.
  12. SEOL no Same, assuming singleline.
  13. BOUND no Match "" at any word boundary using
  14. native charset semantics for non-utf8
  15. BOUNDL no Match "" at any locale word boundary
  16. BOUNDU no Match "" at any word boundary using
  17. Unicode semantics
  18. BOUNDA no Match "" at any word boundary using ASCII
  19. semantics
  20. NBOUND no Match "" at any word non-boundary using
  21. native charset semantics for non-utf8
  22. NBOUNDL no Match "" at any locale word non-boundary
  23. NBOUNDU no Match "" at any word non-boundary using
  24. Unicode semantics
  25. NBOUNDA no Match "" at any word non-boundary using
  26. ASCII semantics
  27. GPOS no Matches where last m//g left off.
  28. # [Special] alternatives:
  29. REG_ANY no Match any one character (except newline).
  30. SANY no Match any one character.
  31. CANY no Match any one byte.
  32. ANYOF sv Match character in (or not in) this
  33. class, single char match only
  34. ANYOF_WARN_SUPER sv Match character in (or not in) this
  35. class, warn (if enabled) upon matching a
  36. char above Unicode max;
  37. ANYOF_SYNTHETIC sv Synthetic start class
  38. POSIXD none Some [[:class:]] under /d; the FLAGS
  39. field gives which one
  40. POSIXL none Some [[:class:]] under /l; the FLAGS
  41. field gives which one
  42. POSIXU none Some [[:class:]] under /u; the FLAGS
  43. field gives which one
  44. POSIXA none Some [[:class:]] under /a; the FLAGS
  45. field gives which one
  46. NPOSIXD none complement of POSIXD, [[:^class:]]
  47. NPOSIXL none complement of POSIXL, [[:^class:]]
  48. NPOSIXU none complement of POSIXU, [[:^class:]]
  49. NPOSIXA none complement of POSIXA, [[:^class:]]
  50. CLUMP no Match any extended grapheme cluster
  51. sequence
  52. # Alternation
  53. # BRANCH The set of branches constituting a single choice are
  54. # hooked together with their "next" pointers, since
  55. # precedence prevents anything being concatenated to
  56. # any individual branch. The "next" pointer of the last
  57. # BRANCH in a choice points to the thing following the
  58. # whole choice. This is also where the final "next"
  59. # pointer of each individual branch points; each branch
  60. # starts with the operand node of a BRANCH node.
  61. #
  62. BRANCH node Match this alternative, or the next...
  63. # Back pointer
  64. # BACK Normal "next" pointers all implicitly point forward;
  65. # BACK exists to make loop structures possible.
  66. # not used
  67. BACK no Match "", "next" ptr points backward.
  68. # Literals
  69. EXACT str Match this string (preceded by length).
  70. EXACTF str Match this non-UTF-8 string (not
  71. guaranteed to be folded) using /id rules
  72. (w/len).
  73. EXACTFL str Match this string (not guaranteed to be
  74. folded) using /il rules (w/len).
  75. EXACTFU str Match this string (folded iff in UTF-8,
  76. length in folding doesn't change if not
  77. in UTF-8) using /iu rules (w/len).
  78. EXACTFA str Match this string (not guaranteed to be
  79. folded) using /iaa rules (w/len).
  80. EXACTFU_SS str Match this string (folded iff in UTF-8,
  81. length in folding may change even if not
  82. in UTF-8) using /iu rules (w/len).
  83. EXACTFU_TRICKYFOLD str Match this folded UTF-8 string using /iu
  84. rules
  85. # Do nothing types
  86. NOTHING no Match empty string.
  87. # A variant of above which delimits a group, thus stops optimizations
  88. TAIL no Match empty string. Can jump here from
  89. outside.
  90. # Loops
  91. # STAR,PLUS '?', and complex '*' and '+', are implemented as
  92. # circular BRANCH structures using BACK. Simple cases
  93. # (one character per match) are implemented with STAR
  94. # and PLUS for speed and to minimize recursive plunges.
  95. #
  96. STAR node Match this (simple) thing 0 or more
  97. times.
  98. PLUS node Match this (simple) thing 1 or more
  99. times.
  100. CURLY sv 2 Match this simple thing {n,m} times.
  101. CURLYN no 2 Capture next-after-this simple thing
  102. CURLYM no 2 Capture this medium-complex thing {n,m}
  103. times.
  104. CURLYX sv 2 Match this complex thing {n,m} times.
  105. # This terminator creates a loop structure for CURLYX
  106. WHILEM no Do curly processing and see if rest
  107. matches.
  108. # Buffer related
  109. # OPEN,CLOSE,GROUPP ...are numbered at compile time.
  110. OPEN num 1 Mark this point in input as start of #n.
  111. CLOSE num 1 Analogous to OPEN.
  112. REF num 1 Match some already matched string
  113. REFF num 1 Match already matched string, folded
  114. using native charset semantics for non-
  115. utf8
  116. REFFL num 1 Match already matched string, folded in
  117. loc.
  118. REFFU num 1 Match already matched string, folded
  119. using unicode semantics for non-utf8
  120. REFFA num 1 Match already matched string, folded
  121. using unicode semantics for non-utf8, no
  122. mixing ASCII, non-ASCII
  123. # Named references. Code in regcomp.c assumes that these all are after
  124. # the numbered references
  125. NREF no-sv 1 Match some already matched string
  126. NREFF no-sv 1 Match already matched string, folded
  127. using native charset semantics for non-
  128. utf8
  129. NREFFL no-sv 1 Match already matched string, folded in
  130. loc.
  131. NREFFU num 1 Match already matched string, folded
  132. using unicode semantics for non-utf8
  133. NREFFA num 1 Match already matched string, folded
  134. using unicode semantics for non-utf8, no
  135. mixing ASCII, non-ASCII
  136. IFMATCH off 1 2 Succeeds if the following matches.
  137. UNLESSM off 1 2 Fails if the following matches.
  138. SUSPEND off 1 1 "Independent" sub-RE.
  139. IFTHEN off 1 1 Switch, should be preceded by switcher.
  140. GROUPP num 1 Whether the group matched.
  141. # Support for long RE
  142. LONGJMP off 1 1 Jump far away.
  143. BRANCHJ off 1 1 BRANCH with long offset.
  144. # The heavy worker
  145. EVAL evl 1 Execute some Perl code.
  146. # Modifiers
  147. MINMOD no Next operator is not greedy.
  148. LOGICAL no Next opcode should set the flag only.
  149. # This is not used yet
  150. RENUM off 1 1 Group with independently numbered parens.
  151. # Trie Related
  152. # Behave the same as A|LIST|OF|WORDS would. The '..C' variants
  153. # have inline charclass data (ascii only), the 'C' store it in the
  154. # structure.
  155. TRIE trie 1 Match many EXACT(F[ALU]?)? at once.
  156. flags==type
  157. TRIEC trie Same as TRIE, but with embedded charclass
  158. charclass data
  159. AHOCORASICK trie 1 Aho Corasick stclass. flags==type
  160. AHOCORASICKC trie Same as AHOCORASICK, but with embedded
  161. charclass charclass data
  162. # Regex Subroutines
  163. GOSUB num/ofs 2L recurse to paren arg1 at (signed) ofs
  164. arg2
  165. GOSTART no recurse to start of pattern
  166. # Special conditionals
  167. NGROUPP no-sv 1 Whether the group matched.
  168. INSUBP num 1 Whether we are in a specific recurse.
  169. DEFINEP none 1 Never execute directly.
  170. # Backtracking Verbs
  171. ENDLIKE none Used only for the type field of verbs
  172. OPFAIL none Same as (?!)
  173. ACCEPT parno 1 Accepts the current matched string.
  174. # Verbs With Arguments
  175. VERB no-sv 1 Used only for the type field of verbs
  176. PRUNE no-sv 1 Pattern fails at this startpoint if no-
  177. backtracking through this
  178. MARKPOINT no-sv 1 Push the current location for rollback by
  179. cut.
  180. SKIP no-sv 1 On failure skip forward (to the mark)
  181. before retrying
  182. COMMIT no-sv 1 Pattern fails outright if backtracking
  183. through this
  184. CUTGROUP no-sv 1 On failure go to the next alternation in
  185. the group
  186. # Control what to keep in $&.
  187. KEEPS no $& begins here.
  188. # New charclass like patterns
  189. LNBREAK none generic newline pattern
  190. # SPECIAL REGOPS
  191. # This is not really a node, but an optimized away piece of a "long"
  192. # node. To simplify debugging output, we mark it as if it were a node
  193. OPTIMIZED off Placeholder for dump.
  194. # Special opcode with the property that no opcode in a compiled program
  195. # will ever be of this type. Thus it can be used as a flag value that
  196. # no other opcode has been seen. END is used similarly, in that an END
  197. # node cant be optimized. So END implies "unoptimizable" and PSEUDO
  198. # mean "not seen anything to optimize yet".
  199. PSEUDO off Pseudo opcode for internal use.

Following the optimizer information is a dump of the offset/length table, here split across several lines:

  1. Offsets: [45]
  2. 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
  3. 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
  4. 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
  5. 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]

The first line here indicates that the offset/length table contains 45 entries. Each entry is a pair of integers, denoted by offset[length] . Entries are numbered starting with 1, so entry #1 here is 1[4] and entry #12 is 5[1] . 1[4] indicates that the node labeled 1: (the 1: ANYOF[bc]) begins at character position 1 in the pre-compiled form of the regex, and has a length of 4 characters. 5[1] in position 12 indicates that the node labeled 12: (the 12: EXACT <d>) begins at character position 5 in the pre-compiled form of the regex, and has a length of 1 character. 12[1] in position 14 indicates that the node labeled 14: (the 14: CURLYX[0] {1,32767}) begins at character position 12 in the pre-compiled form of the regex, and has a length of 1 character---that is, it corresponds to the + symbol in the precompiled regex.

0[0] items indicate that there is no corresponding node.

Run-time Output

First of all, when doing a match, one may get no run-time output even if debugging is enabled. This means that the regex engine was never entered and that all of the job was therefore done by the optimizer.

If the regex engine was entered, the output may look like this:

  1. Matching '[bc]d(ef*g)+h[ij]k$' against 'abcdefg__gh__'
  2. Setting an EVAL scope, savestack=3
  3. 2 <ab> <cdefg__gh_> | 1: ANYOF
  4. 3 <abc> <defg__gh_> | 11: EXACT <d>
  5. 4 <abcd> <efg__gh_> | 13: CURLYX {1,32767}
  6. 4 <abcd> <efg__gh_> | 26: WHILEM
  7. 0 out of 1..32767 cc=effff31c
  8. 4 <abcd> <efg__gh_> | 15: OPEN1
  9. 4 <abcd> <efg__gh_> | 17: EXACT <e>
  10. 5 <abcde> <fg__gh_> | 19: STAR
  11. EXACT <f> can match 1 times out of 32767...
  12. Setting an EVAL scope, savestack=3
  13. 6 <bcdef> <g__gh__> | 22: EXACT <g>
  14. 7 <bcdefg> <__gh__> | 24: CLOSE1
  15. 7 <bcdefg> <__gh__> | 26: WHILEM
  16. 1 out of 1..32767 cc=effff31c
  17. Setting an EVAL scope, savestack=12
  18. 7 <bcdefg> <__gh__> | 15: OPEN1
  19. 7 <bcdefg> <__gh__> | 17: EXACT <e>
  20. restoring \1 to 4(4)..7
  21. failed, try continuation...
  22. 7 <bcdefg> <__gh__> | 27: NOTHING
  23. 7 <bcdefg> <__gh__> | 28: EXACT <h>
  24. failed...
  25. failed...

The most significant information in the output is about the particular node of the compiled regex that is currently being tested against the target string. The format of these lines is

STRING-OFFSET <PRE-STRING> <POST-STRING> |ID: TYPE

The TYPE info is indented with respect to the backtracking level. Other incidental information appears interspersed within.

Debugging Perl Memory Usage

Perl is a profligate wastrel when it comes to memory use. There is a saying that to estimate memory usage of Perl, assume a reasonable algorithm for memory allocation, multiply that estimate by 10, and while you still may miss the mark, at least you won't be quite so astonished. This is not absolutely true, but may provide a good grasp of what happens.

Assume that an integer cannot take less than 20 bytes of memory, a float cannot take less than 24 bytes, a string cannot take less than 32 bytes (all these examples assume 32-bit architectures, the result are quite a bit worse on 64-bit architectures). If a variable is accessed in two of three different ways (which require an integer, a float, or a string), the memory footprint may increase yet another 20 bytes. A sloppy malloc(3) implementation can inflate these numbers dramatically.

On the opposite end of the scale, a declaration like

  1. sub foo;

may take up to 500 bytes of memory, depending on which release of Perl you're running.

Anecdotal estimates of source-to-compiled code bloat suggest an eightfold increase. This means that the compiled form of reasonable (normally commented, properly indented etc.) code will take about eight times more space in memory than the code took on disk.

The -DL command-line switch is obsolete since circa Perl 5.6.0 (it was available only if Perl was built with -DDEBUGGING ). The switch was used to track Perl's memory allocations and possible memory leaks. These days the use of malloc debugging tools like Purify or valgrind is suggested instead. See also PERL_MEM_LOG in perlhacktips.

One way to find out how much memory is being used by Perl data structures is to install the Devel::Size module from CPAN: it gives you the minimum number of bytes required to store a particular data structure. Please be mindful of the difference between the size() and total_size().

If Perl has been compiled using Perl's malloc you can analyze Perl memory usage by setting $ENV{PERL_DEBUG_MSTATS}.

Using $ENV{PERL_DEBUG_MSTATS}

If your perl is using Perl's malloc() and was compiled with the necessary switches (this is the default), then it will print memory usage statistics after compiling your code when $ENV{PERL_DEBUG_MSTATS} > 1 , and before termination of the program when $ENV{PERL_DEBUG_MSTATS} >= 1 . The report format is similar to the following example:

  1. $ PERL_DEBUG_MSTATS=2 perl -e "require Carp"
  2. Memory allocation statistics after compilation: (buckets 4(4)..8188(8192)
  3. 14216 free: 130 117 28 7 9 0 2 2 1 0 0
  4. 437 61 36 0 5
  5. 60924 used: 125 137 161 55 7 8 6 16 2 0 1
  6. 74 109 304 84 20
  7. Total sbrk(): 77824/21:119. Odd ends: pad+heads+chain+tail: 0+636+0+2048.
  8. Memory allocation statistics after execution: (buckets 4(4)..8188(8192)
  9. 30888 free: 245 78 85 13 6 2 1 3 2 0 1
  10. 315 162 39 42 11
  11. 175816 used: 265 176 1112 111 26 22 11 27 2 1 1
  12. 196 178 1066 798 39
  13. Total sbrk(): 215040/47:145. Odd ends: pad+heads+chain+tail: 0+2192+0+6144.

It is possible to ask for such a statistic at arbitrary points in your execution using the mstat() function out of the standard Devel::Peek module.

Here is some explanation of that format:

  • buckets SMALLEST(APPROX)..GREATEST(APPROX)

    Perl's malloc() uses bucketed allocations. Every request is rounded up to the closest bucket size available, and a bucket is taken from the pool of buckets of that size.

    The line above describes the limits of buckets currently in use. Each bucket has two sizes: memory footprint and the maximal size of user data that can fit into this bucket. Suppose in the above example that the smallest bucket were size 4. The biggest bucket would have usable size 8188, and the memory footprint would be 8192.

    In a Perl built for debugging, some buckets may have negative usable size. This means that these buckets cannot (and will not) be used. For larger buckets, the memory footprint may be one page greater than a power of 2. If so, the corresponding power of two is printed in the APPROX field above.

  • Free/Used

    The 1 or 2 rows of numbers following that correspond to the number of buckets of each size between SMALLEST and GREATEST . In the first row, the sizes (memory footprints) of buckets are powers of two--or possibly one page greater. In the second row, if present, the memory footprints of the buckets are between the memory footprints of two buckets "above".

    For example, suppose under the previous example, the memory footprints were

    1. free: 8 16 32 64 128 256 512 1024 2048 4096 8192
    2. 4 12 24 48 80

    With a non-DEBUGGING perl, the buckets starting from 128 have a 4-byte overhead, and thus an 8192-long bucket may take up to 8188-byte allocations.

  • Total sbrk(): SBRKed/SBRKs:CONTINUOUS

    The first two fields give the total amount of memory perl sbrk(2)ed (ess-broken? :-) and number of sbrk(2)s used. The third number is what perl thinks about continuity of returned chunks. So long as this number is positive, malloc() will assume that it is probable that sbrk(2) will provide continuous memory.

    Memory allocated by external libraries is not counted.

  • pad: 0

    The amount of sbrk(2)ed memory needed to keep buckets aligned.

  • heads: 2192

    Although memory overhead of bigger buckets is kept inside the bucket, for smaller buckets, it is kept in separate areas. This field gives the total size of these areas.

  • chain: 0

    malloc() may want to subdivide a bigger bucket into smaller buckets. If only a part of the deceased bucket is left unsubdivided, the rest is kept as an element of a linked list. This field gives the total size of these chunks.

  • tail: 6144

    To minimize the number of sbrk(2)s, malloc() asks for more memory. This field gives the size of the yet unused part, which is sbrk(2)ed, but never touched.

SEE ALSO

perldebug, perlguts, perlrun re, and Devel::DProf.