| [ Index ] |
PHP Cross Reference of Unnamed Project |
[Summary view] [Print] [Text view]
1 =head1 NAME 2 3 perldebguts - Guts of Perl debugging 4 5 =head1 DESCRIPTION 6 7 This is not the perldebug(1) manpage, which tells you how to use 8 the debugger. This manpage describes low-level details concerning 9 the debugger's internals, which range from difficult to impossible 10 to understand for anyone who isn't incredibly intimate with Perl's guts. 11 Caveat lector. 12 13 =head1 Debugger Internals 14 15 Perl has special debugging hooks at compile-time and run-time used 16 to create debugging environments. These hooks are not to be confused 17 with the I<perl -Dxxx> command described in L<perlrun>, which is 18 usable only if a special Perl is built per the instructions in the 19 F<INSTALL> podpage in the Perl source tree. 20 21 For example, whenever you call Perl's built-in C<caller> function 22 from the package C<DB>, the arguments that the corresponding stack 23 frame was called with are copied to the C<@DB::args> array. These 24 mechanisms are enabled by calling Perl with the B<-d> switch. 25 Specifically, the following additional features are enabled 26 (cf. L<perlvar/$^P>): 27 28 =over 4 29 30 =item * 31 32 Perl inserts the contents of C<$ENV{PERL5DB}> (or C<BEGIN {require 33 'perl5db.pl'}> if not present) before the first line of your program. 34 35 =item * 36 37 Each array C<@{"_<$filename"}> holds the lines of $filename for a 38 file compiled by Perl. The same is also true for C<eval>ed strings 39 that contain subroutines, or which are currently being executed. 40 The $filename for C<eval>ed strings looks like C<(eval 34)>. 41 Code assertions in regexes look like C<(re_eval 19)>. 42 43 Values in this array are magical in numeric context: they compare 44 equal to zero only if the line is not breakable. 45 46 =item * 47 48 Each hash C<%{"_<$filename"}> contains breakpoints and actions keyed 49 by line number. Individual entries (as opposed to the whole hash) 50 are settable. Perl only cares about Boolean true here, although 51 the values used by F<perl5db.pl> have the form 52 C<"$break_condition\0$action">. 53 54 The same holds for evaluated strings that contain subroutines, or 55 which are currently being executed. The $filename for C<eval>ed strings 56 looks like C<(eval 34)> or C<(re_eval 19)>. 57 58 =item * 59 60 Each scalar C<${"_<$filename"}> contains C<"_<$filename">. This is 61 also the case for evaluated strings that contain subroutines, or 62 which are currently being executed. The $filename for C<eval>ed 63 strings looks like C<(eval 34)> or C<(re_eval 19)>. 64 65 =item * 66 67 After each C<require>d file is compiled, but before it is executed, 68 C<DB::postponed(*{"_<$filename"})> is called if the subroutine 69 C<DB::postponed> exists. Here, the $filename is the expanded name of 70 the C<require>d file, as found in the values of %INC. 71 72 =item * 73 74 After each subroutine C<subname> is compiled, the existence of 75 C<$DB::postponed{subname}> is checked. If this key exists, 76 C<DB::postponed(subname)> is called if the C<DB::postponed> subroutine 77 also exists. 78 79 =item * 80 81 A hash C<%DB::sub> is maintained, whose keys are subroutine names 82 and whose values have the form C<filename:startline-endline>. 83 C<filename> has the form C<(eval 34)> for subroutines defined inside 84 C<eval>s, or C<(re_eval 19)> for those within regex code assertions. 85 86 =item * 87 88 When the execution of your program reaches a point that can hold a 89 breakpoint, the C<DB::DB()> subroutine is called if any of the variables 90 C<$DB::trace>, C<$DB::single>, or C<$DB::signal> is true. These variables 91 are not C<local>izable. This feature is disabled when executing 92 inside C<DB::DB()>, including functions called from it 93 unless C<< $^D & (1<<30) >> is true. 94 95 =item * 96 97 When execution of the program reaches a subroutine call, a call to 98 C<&DB::sub>(I<args>) is made instead, with C<$DB::sub> holding the 99 name of the called subroutine. (This doesn't happen if the subroutine 100 was compiled in the C<DB> package.) 101 102 =back 103 104 Note that if C<&DB::sub> needs external data for it to work, no 105 subroutine call is possible without it. As an example, the standard 106 debugger's C<&DB::sub> depends on the C<$DB::deep> variable 107 (it defines how many levels of recursion deep into the debugger you can go 108 before a mandatory break). If C<$DB::deep> is not defined, subroutine 109 calls are not possible, even though C<&DB::sub> exists. 110 111 =head2 Writing Your Own Debugger 112 113 =head3 Environment Variables 114 115 The C<PERL5DB> environment variable can be used to define a debugger. 116 For example, the minimal "working" debugger (it actually doesn't do anything) 117 consists of one line: 118 119 sub DB::DB {} 120 121 It can easily be defined like this: 122 123 $ PERL5DB="sub DB::DB {}" perl -d your-script 124 125 Another brief debugger, slightly more useful, can be created 126 with only the line: 127 128 sub DB::DB {print ++$i; scalar <STDIN>} 129 130 This debugger prints a number which increments for each statement 131 encountered and waits for you to hit a newline before continuing 132 to the next statement. 133 134 The following debugger is actually useful: 135 136 { 137 package DB; 138 sub DB {} 139 sub sub {print ++$i, " $sub\n"; &$sub} 140 } 141 142 It prints the sequence number of each subroutine call and the name of the 143 called subroutine. Note that C<&DB::sub> is being compiled into the 144 package C<DB> through the use of the C<package> directive. 145 146 When it starts, the debugger reads your rc file (F<./.perldb> or 147 F<~/.perldb> under Unix), which can set important options. 148 (A subroutine (C<&afterinit>) can be defined here as well; it is executed 149 after the debugger completes its own initialization.) 150 151 After the rc file is read, the debugger reads the PERLDB_OPTS 152 environment variable and uses it to set debugger options. The 153 contents of this variable are treated as if they were the argument 154 of an C<o ...> debugger command (q.v. in L<perldebug/Options>). 155 156 =head3 Debugger internal variables 157 In addition to the file and subroutine-related variables mentioned above, 158 the debugger also maintains various magical internal variables. 159 160 =over 4 161 162 =item * 163 164 C<@DB::dbline> is an alias for C<@{"::_<current_file"}>, which 165 holds the lines of the currently-selected file (compiled by Perl), either 166 explicitly chosen with the debugger's C<f> command, or implicitly by flow 167 of execution. 168 169 Values in this array are magical in numeric context: they compare 170 equal to zero only if the line is not breakable. 171 172 =item * 173 174 C<%DB::dbline>, is an alias for C<%{"::_<current_file"}>, which 175 contains breakpoints and actions keyed by line number in 176 the currently-selected file, either explicitly chosen with the 177 debugger's C<f> command, or implicitly by flow of execution. 178 179 As previously noted, individual entries (as opposed to the whole hash) 180 are settable. Perl only cares about Boolean true here, although 181 the values used by F<perl5db.pl> have the form 182 C<"$break_condition\0$action">. 183 184 =back 185 186 =head3 Debugger customization functions 187 188 Some functions are provided to simplify customization. 189 190 =over 4 191 192 =item * 193 194 See L<perldebug/"Configurable Options"> for a description of options parsed by 195 C<DB::parse_options(string)>. 196 197 =item * 198 199 C<DB::dump_trace(skip[,count])> skips the specified number of frames 200 and returns a list containing information about the calling frames (all 201 of them, if C<count> is missing). Each entry is reference to a hash 202 with keys C<context> (either C<.>, C<$>, or C<@>), C<sub> (subroutine 203 name, or info about C<eval>), C<args> (C<undef> or a reference to 204 an array), C<file>, and C<line>. 205 206 =item * 207 208 C<DB::print_trace(FH, skip[, count[, short]])> prints 209 formatted info about caller frames. The last two functions may be 210 convenient as arguments to C<< < >>, C<< << >> commands. 211 212 =back 213 214 Note that any variables and functions that are not documented in 215 this manpages (or in L<perldebug>) are considered for internal 216 use only, and as such are subject to change without notice. 217 218 =head1 Frame Listing Output Examples 219 220 The C<frame> option can be used to control the output of frame 221 information. For example, contrast this expression trace: 222 223 $ perl -de 42 224 Stack dump during die enabled outside of evals. 225 226 Loading DB routines from perl5db.pl patch level 0.94 227 Emacs support available. 228 229 Enter h or `h h' for help. 230 231 main::(-e:1): 0 232 DB<1> sub foo { 14 } 233 234 DB<2> sub bar { 3 } 235 236 DB<3> t print foo() * bar() 237 main::((eval 172):3): print foo() + bar(); 238 main::foo((eval 168):2): 239 main::bar((eval 170):2): 240 42 241 242 with this one, once the C<o>ption C<frame=2> has been set: 243 244 DB<4> o f=2 245 frame = '2' 246 DB<5> t print foo() * bar() 247 3: foo() * bar() 248 entering main::foo 249 2: sub foo { 14 }; 250 exited main::foo 251 entering main::bar 252 2: sub bar { 3 }; 253 exited main::bar 254 42 255 256 By way of demonstration, we present below a laborious listing 257 resulting from setting your C<PERLDB_OPTS> environment variable to 258 the value C<f=n N>, and running I<perl -d -V> from the command line. 259 Examples use various values of C<n> are shown to give you a feel 260 for the difference between settings. Long those it may be, this 261 is not a complete listing, but only excerpts. 262 263 =over 4 264 265 =item 1 266 267 entering main::BEGIN 268 entering Config::BEGIN 269 Package lib/Exporter.pm. 270 Package lib/Carp.pm. 271 Package lib/Config.pm. 272 entering Config::TIEHASH 273 entering Exporter::import 274 entering Exporter::export 275 entering Config::myconfig 276 entering Config::FETCH 277 entering Config::FETCH 278 entering Config::FETCH 279 entering Config::FETCH 280 281 =item 2 282 283 entering main::BEGIN 284 entering Config::BEGIN 285 Package lib/Exporter.pm. 286 Package lib/Carp.pm. 287 exited Config::BEGIN 288 Package lib/Config.pm. 289 entering Config::TIEHASH 290 exited Config::TIEHASH 291 entering Exporter::import 292 entering Exporter::export 293 exited Exporter::export 294 exited Exporter::import 295 exited main::BEGIN 296 entering Config::myconfig 297 entering Config::FETCH 298 exited Config::FETCH 299 entering Config::FETCH 300 exited Config::FETCH 301 entering Config::FETCH 302 303 =item 3 304 305 in $=main::BEGIN() from /dev/null:0 306 in $=Config::BEGIN() from lib/Config.pm:2 307 Package lib/Exporter.pm. 308 Package lib/Carp.pm. 309 Package lib/Config.pm. 310 in $=Config::TIEHASH('Config') from lib/Config.pm:644 311 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 312 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from li 313 in @=Config::myconfig() from /dev/null:0 314 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574 315 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574 316 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574 317 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574 318 in $=Config::FETCH(ref(Config), 'osname') from lib/Config.pm:574 319 in $=Config::FETCH(ref(Config), 'osvers') from lib/Config.pm:574 320 321 =item 4 322 323 in $=main::BEGIN() from /dev/null:0 324 in $=Config::BEGIN() from lib/Config.pm:2 325 Package lib/Exporter.pm. 326 Package lib/Carp.pm. 327 out $=Config::BEGIN() from lib/Config.pm:0 328 Package lib/Config.pm. 329 in $=Config::TIEHASH('Config') from lib/Config.pm:644 330 out $=Config::TIEHASH('Config') from lib/Config.pm:644 331 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 332 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/ 333 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/ 334 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 335 out $=main::BEGIN() from /dev/null:0 336 in @=Config::myconfig() from /dev/null:0 337 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574 338 out $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574 339 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574 340 out $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574 341 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574 342 out $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574 343 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574 344 345 =item 5 346 347 in $=main::BEGIN() from /dev/null:0 348 in $=Config::BEGIN() from lib/Config.pm:2 349 Package lib/Exporter.pm. 350 Package lib/Carp.pm. 351 out $=Config::BEGIN() from lib/Config.pm:0 352 Package lib/Config.pm. 353 in $=Config::TIEHASH('Config') from lib/Config.pm:644 354 out $=Config::TIEHASH('Config') from lib/Config.pm:644 355 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 356 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E 357 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E 358 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 359 out $=main::BEGIN() from /dev/null:0 360 in @=Config::myconfig() from /dev/null:0 361 in $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574 362 out $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574 363 in $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574 364 out $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574 365 366 =item 6 367 368 in $=CODE(0x15eca4)() from /dev/null:0 369 in $=CODE(0x182528)() from lib/Config.pm:2 370 Package lib/Exporter.pm. 371 out $=CODE(0x182528)() from lib/Config.pm:0 372 scalar context return from CODE(0x182528): undef 373 Package lib/Config.pm. 374 in $=Config::TIEHASH('Config') from lib/Config.pm:628 375 out $=Config::TIEHASH('Config') from lib/Config.pm:628 376 scalar context return from Config::TIEHASH: empty hash 377 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 378 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171 379 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171 380 scalar context return from Exporter::export: '' 381 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0 382 scalar context return from Exporter::import: '' 383 384 =back 385 386 In all cases shown above, the line indentation shows the call tree. 387 If bit 2 of C<frame> is set, a line is printed on exit from a 388 subroutine as well. If bit 4 is set, the arguments are printed 389 along with the caller info. If bit 8 is set, the arguments are 390 printed even if they are tied or references. If bit 16 is set, the 391 return value is printed, too. 392 393 When a package is compiled, a line like this 394 395 Package lib/Carp.pm. 396 397 is printed with proper indentation. 398 399 =head1 Debugging regular expressions 400 401 There are two ways to enable debugging output for regular expressions. 402 403 If your perl is compiled with C<-DDEBUGGING>, you may use the 404 B<-Dr> flag on the command line. 405 406 Otherwise, one can C<use re 'debug'>, which has effects at 407 compile time and run time. It is not lexically scoped. 408 409 =head2 Compile-time output 410 411 The debugging output at compile time looks like this: 412 413 Compiling REx `[bc]d(ef*g)+h[ij]k$' 414 size 45 Got 364 bytes for offset annotations. 415 first at 1 416 rarest char g at 0 417 rarest char d at 0 418 1: ANYOF[bc](12) 419 12: EXACT <d>(14) 420 14: CURLYX[0] {1,32767}(28) 421 16: OPEN1(18) 422 18: EXACT <e>(20) 423 20: STAR(23) 424 21: EXACT <f>(0) 425 23: EXACT <g>(25) 426 25: CLOSE1(27) 427 27: WHILEM[1/1](0) 428 28: NOTHING(29) 429 29: EXACT <h>(31) 430 31: ANYOF[ij](42) 431 42: EXACT <k>(44) 432 44: EOL(45) 433 45: END(0) 434 anchored `de' at 1 floating `gh' at 3..2147483647 (checking floating) 435 stclass `ANYOF[bc]' minlen 7 436 Offsets: [45] 437 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1] 438 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0] 439 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0] 440 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0] 441 Omitting $` $& $' support. 442 443 The first line shows the pre-compiled form of the regex. The second 444 shows the size of the compiled form (in arbitrary units, usually 445 4-byte words) and the total number of bytes allocated for the 446 offset/length table, usually 4+C<size>*8. The next line shows the 447 label I<id> of the first node that does a match. 448 449 The 450 451 anchored `de' at 1 floating `gh' at 3..2147483647 (checking floating) 452 stclass `ANYOF[bc]' minlen 7 453 454 line (split into two lines above) contains optimizer 455 information. In the example shown, the optimizer found that the match 456 should contain a substring C<de> at offset 1, plus substring C<gh> 457 at some offset between 3 and infinity. Moreover, when checking for 458 these substrings (to abandon impossible matches quickly), Perl will check 459 for the substring C<gh> before checking for the substring C<de>. The 460 optimizer may also use the knowledge that the match starts (at the 461 C<first> I<id>) with a character class, and no string 462 shorter than 7 characters can possibly match. 463 464 The fields of interest which may appear in this line are 465 466 =over 4 467 468 =item C<anchored> I<STRING> C<at> I<POS> 469 470 =item C<floating> I<STRING> C<at> I<POS1..POS2> 471 472 See above. 473 474 =item C<matching floating/anchored> 475 476 Which substring to check first. 477 478 =item C<minlen> 479 480 The minimal length of the match. 481 482 =item C<stclass> I<TYPE> 483 484 Type of first matching node. 485 486 =item C<noscan> 487 488 Don't scan for the found substrings. 489 490 =item C<isall> 491 492 Means that the optimizer information is all that the regular 493 expression contains, and thus one does not need to enter the regex engine at 494 all. 495 496 =item C<GPOS> 497 498 Set if the pattern contains C<\G>. 499 500 =item C<plus> 501 502 Set if the pattern starts with a repeated char (as in C<x+y>). 503 504 =item C<implicit> 505 506 Set if the pattern starts with C<.*>. 507 508 =item C<with eval> 509 510 Set if the pattern contain eval-groups, such as C<(?{ code })> and 511 C<(??{ code })>. 512 513 =item C<anchored(TYPE)> 514 515 If the pattern may match only at a handful of places, (with C<TYPE> 516 being C<BOL>, C<MBOL>, or C<GPOS>. See the table below. 517 518 =back 519 520 If a substring is known to match at end-of-line only, it may be 521 followed by C<$>, as in C<floating `k'$>. 522 523 The optimizer-specific information is used to avoid entering (a slow) regex 524 engine on strings that will not definitely match. If the C<isall> flag 525 is set, a call to the regex engine may be avoided even when the optimizer 526 found an appropriate place for the match. 527 528 Above the optimizer section is the list of I<nodes> of the compiled 529 form of the regex. Each line has format 530 531 C< >I<id>: I<TYPE> I<OPTIONAL-INFO> (I<next-id>) 532 533 =head2 Types of nodes 534 535 Here are the possible types, with short descriptions: 536 537 # TYPE arg-description [num-args] [longjump-len] DESCRIPTION 538 539 # Exit points 540 END no End of program. 541 SUCCEED no Return from a subroutine, basically. 542 543 # Anchors: 544 BOL no Match "" at beginning of line. 545 MBOL no Same, assuming multiline. 546 SBOL no Same, assuming singleline. 547 EOS no Match "" at end of string. 548 EOL no Match "" at end of line. 549 MEOL no Same, assuming multiline. 550 SEOL no Same, assuming singleline. 551 BOUND no Match "" at any word boundary 552 BOUNDL no Match "" at any word boundary 553 NBOUND no Match "" at any word non-boundary 554 NBOUNDL no Match "" at any word non-boundary 555 GPOS no Matches where last m//g left off. 556 557 # [Special] alternatives 558 ANY no Match any one character (except newline). 559 SANY no Match any one character. 560 ANYOF sv Match character in (or not in) this class. 561 ALNUM no Match any alphanumeric character 562 ALNUML no Match any alphanumeric char in locale 563 NALNUM no Match any non-alphanumeric character 564 NALNUML no Match any non-alphanumeric char in locale 565 SPACE no Match any whitespace character 566 SPACEL no Match any whitespace char in locale 567 NSPACE no Match any non-whitespace character 568 NSPACEL no Match any non-whitespace char in locale 569 DIGIT no Match any numeric character 570 NDIGIT no Match any non-numeric character 571 572 # BRANCH The set of branches constituting a single choice are hooked 573 # together with their "next" pointers, since precedence prevents 574 # anything being concatenated to any individual branch. The 575 # "next" pointer of the last BRANCH in a choice points to the 576 # thing following the whole choice. This is also where the 577 # final "next" pointer of each individual branch points; each 578 # branch starts with the operand node of a BRANCH node. 579 # 580 BRANCH node Match this alternative, or the next... 581 582 # BACK Normal "next" pointers all implicitly point forward; BACK 583 # exists to make loop structures possible. 584 # not used 585 BACK no Match "", "next" ptr points backward. 586 587 # Literals 588 EXACT sv Match this string (preceded by length). 589 EXACTF sv Match this string, folded (prec. by length). 590 EXACTFL sv Match this string, folded in locale (w/len). 591 592 # Do nothing 593 NOTHING no Match empty string. 594 # A variant of above which delimits a group, thus stops optimizations 595 TAIL no Match empty string. Can jump here from outside. 596 597 # STAR,PLUS '?', and complex '*' and '+', are implemented as circular 598 # BRANCH structures using BACK. Simple cases (one character 599 # per match) are implemented with STAR and PLUS for speed 600 # and to minimize recursive plunges. 601 # 602 STAR node Match this (simple) thing 0 or more times. 603 PLUS node Match this (simple) thing 1 or more times. 604 605 CURLY sv 2 Match this simple thing {n,m} times. 606 CURLYN no 2 Match next-after-this simple thing 607 # {n,m} times, set parens. 608 CURLYM no 2 Match this medium-complex thing {n,m} times. 609 CURLYX sv 2 Match this complex thing {n,m} times. 610 611 # This terminator creates a loop structure for CURLYX 612 WHILEM no Do curly processing and see if rest matches. 613 614 # OPEN,CLOSE,GROUPP ...are numbered at compile time. 615 OPEN num 1 Mark this point in input as start of #n. 616 CLOSE num 1 Analogous to OPEN. 617 618 REF num 1 Match some already matched string 619 REFF num 1 Match already matched string, folded 620 REFFL num 1 Match already matched string, folded in loc. 621 622 # grouping assertions 623 IFMATCH off 1 2 Succeeds if the following matches. 624 UNLESSM off 1 2 Fails if the following matches. 625 SUSPEND off 1 1 "Independent" sub-regex. 626 IFTHEN off 1 1 Switch, should be preceded by switcher . 627 GROUPP num 1 Whether the group matched. 628 629 # Support for long regex 630 LONGJMP off 1 1 Jump far away. 631 BRANCHJ off 1 1 BRANCH with long offset. 632 633 # The heavy worker 634 EVAL evl 1 Execute some Perl code. 635 636 # Modifiers 637 MINMOD no Next operator is not greedy. 638 LOGICAL no Next opcode should set the flag only. 639 640 # This is not used yet 641 RENUM off 1 1 Group with independently numbered parens. 642 643 # This is not really a node, but an optimized away piece of a "long" node. 644 # To simplify debugging output, we mark it as if it were a node 645 OPTIMIZED off Placeholder for dump. 646 647 =for unprinted-credits 648 Next section M-J. Dominus (mjd-perl-patch+@plover.com) 20010421 649 650 Following the optimizer information is a dump of the offset/length 651 table, here split across several lines: 652 653 Offsets: [45] 654 1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1] 655 0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0] 656 11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0] 657 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0] 658 659 The first line here indicates that the offset/length table contains 45 660 entries. Each entry is a pair of integers, denoted by C<offset[length]>. 661 Entries are numbered starting with 1, so entry #1 here is C<1[4]> and 662 entry #12 is C<5[1]>. C<1[4]> indicates that the node labeled C<1:> 663 (the C<1: ANYOF[bc]>) begins at character position 1 in the 664 pre-compiled form of the regex, and has a length of 4 characters. 665 C<5[1]> in position 12 666 indicates that the node labeled C<12:> 667 (the C<< 12: EXACT <d> >>) begins at character position 5 in the 668 pre-compiled form of the regex, and has a length of 1 character. 669 C<12[1]> in position 14 670 indicates that the node labeled C<14:> 671 (the C<< 14: CURLYX[0] {1,32767} >>) begins at character position 12 in the 672 pre-compiled form of the regex, and has a length of 1 character---that 673 is, it corresponds to the C<+> symbol in the precompiled regex. 674 675 C<0[0]> items indicate that there is no corresponding node. 676 677 =head2 Run-time output 678 679 First of all, when doing a match, one may get no run-time output even 680 if debugging is enabled. This means that the regex engine was never 681 entered and that all of the job was therefore done by the optimizer. 682 683 If the regex engine was entered, the output may look like this: 684 685 Matching `[bc]d(ef*g)+h[ij]k$' against `abcdefg__gh__' 686 Setting an EVAL scope, savestack=3 687 2 <ab> <cdefg__gh_> | 1: ANYOF 688 3 <abc> <defg__gh_> | 11: EXACT <d> 689 4 <abcd> <efg__gh_> | 13: CURLYX {1,32767} 690 4 <abcd> <efg__gh_> | 26: WHILEM 691 0 out of 1..32767 cc=effff31c 692 4 <abcd> <efg__gh_> | 15: OPEN1 693 4 <abcd> <efg__gh_> | 17: EXACT <e> 694 5 <abcde> <fg__gh_> | 19: STAR 695 EXACT <f> can match 1 times out of 32767... 696 Setting an EVAL scope, savestack=3 697 6 <bcdef> <g__gh__> | 22: EXACT <g> 698 7 <bcdefg> <__gh__> | 24: CLOSE1 699 7 <bcdefg> <__gh__> | 26: WHILEM 700 1 out of 1..32767 cc=effff31c 701 Setting an EVAL scope, savestack=12 702 7 <bcdefg> <__gh__> | 15: OPEN1 703 7 <bcdefg> <__gh__> | 17: EXACT <e> 704 restoring \1 to 4(4)..7 705 failed, try continuation... 706 7 <bcdefg> <__gh__> | 27: NOTHING 707 7 <bcdefg> <__gh__> | 28: EXACT <h> 708 failed... 709 failed... 710 711 The most significant information in the output is about the particular I<node> 712 of the compiled regex that is currently being tested against the target string. 713 The format of these lines is 714 715 C< >I<STRING-OFFSET> <I<PRE-STRING>> <I<POST-STRING>> |I<ID>: I<TYPE> 716 717 The I<TYPE> info is indented with respect to the backtracking level. 718 Other incidental information appears interspersed within. 719 720 =head1 Debugging Perl memory usage 721 722 Perl is a profligate wastrel when it comes to memory use. There 723 is a saying that to estimate memory usage of Perl, assume a reasonable 724 algorithm for memory allocation, multiply that estimate by 10, and 725 while you still may miss the mark, at least you won't be quite so 726 astonished. This is not absolutely true, but may provide a good 727 grasp of what happens. 728 729 Assume that an integer cannot take less than 20 bytes of memory, a 730 float cannot take less than 24 bytes, a string cannot take less 731 than 32 bytes (all these examples assume 32-bit architectures, the 732 result are quite a bit worse on 64-bit architectures). If a variable 733 is accessed in two of three different ways (which require an integer, 734 a float, or a string), the memory footprint may increase yet another 735 20 bytes. A sloppy malloc(3) implementation can inflate these 736 numbers dramatically. 737 738 On the opposite end of the scale, a declaration like 739 740 sub foo; 741 742 may take up to 500 bytes of memory, depending on which release of Perl 743 you're running. 744 745 Anecdotal estimates of source-to-compiled code bloat suggest an 746 eightfold increase. This means that the compiled form of reasonable 747 (normally commented, properly indented etc.) code will take 748 about eight times more space in memory than the code took 749 on disk. 750 751 The B<-DL> command-line switch is obsolete since circa Perl 5.6.0 752 (it was available only if Perl was built with C<-DDEBUGGING>). 753 The switch was used to track Perl's memory allocations and possible 754 memory leaks. These days the use of malloc debugging tools like 755 F<Purify> or F<valgrind> is suggested instead. See also 756 L<perlhack/PERL_MEM_LOG>. 757 758 One way to find out how much memory is being used by Perl data 759 structures is to install the Devel::Size module from CPAN: it gives 760 you the minimum number of bytes required to store a particular data 761 structure. Please be mindful of the difference between the size() 762 and total_size(). 763 764 If Perl has been compiled using Perl's malloc you can analyze Perl 765 memory usage by setting the $ENV{PERL_DEBUG_MSTATS}. 766 767 =head2 Using C<$ENV{PERL_DEBUG_MSTATS}> 768 769 If your perl is using Perl's malloc() and was compiled with the 770 necessary switches (this is the default), then it will print memory 771 usage statistics after compiling your code when C<< $ENV{PERL_DEBUG_MSTATS} 772 > 1 >>, and before termination of the program when C<< 773 $ENV{PERL_DEBUG_MSTATS} >= 1 >>. The report format is similar to 774 the following example: 775 776 $ PERL_DEBUG_MSTATS=2 perl -e "require Carp" 777 Memory allocation statistics after compilation: (buckets 4(4)..8188(8192) 778 14216 free: 130 117 28 7 9 0 2 2 1 0 0 779 437 61 36 0 5 780 60924 used: 125 137 161 55 7 8 6 16 2 0 1 781 74 109 304 84 20 782 Total sbrk(): 77824/21:119. Odd ends: pad+heads+chain+tail: 0+636+0+2048. 783 Memory allocation statistics after execution: (buckets 4(4)..8188(8192) 784 30888 free: 245 78 85 13 6 2 1 3 2 0 1 785 315 162 39 42 11 786 175816 used: 265 176 1112 111 26 22 11 27 2 1 1 787 196 178 1066 798 39 788 Total sbrk(): 215040/47:145. Odd ends: pad+heads+chain+tail: 0+2192+0+6144. 789 790 It is possible to ask for such a statistic at arbitrary points in 791 your execution using the mstat() function out of the standard 792 Devel::Peek module. 793 794 Here is some explanation of that format: 795 796 =over 4 797 798 =item C<buckets SMALLEST(APPROX)..GREATEST(APPROX)> 799 800 Perl's malloc() uses bucketed allocations. Every request is rounded 801 up to the closest bucket size available, and a bucket is taken from 802 the pool of buckets of that size. 803 804 The line above describes the limits of buckets currently in use. 805 Each bucket has two sizes: memory footprint and the maximal size 806 of user data that can fit into this bucket. Suppose in the above 807 example that the smallest bucket were size 4. The biggest bucket 808 would have usable size 8188, and the memory footprint would be 8192. 809 810 In a Perl built for debugging, some buckets may have negative usable 811 size. This means that these buckets cannot (and will not) be used. 812 For larger buckets, the memory footprint may be one page greater 813 than a power of 2. If so, case the corresponding power of two is 814 printed in the C<APPROX> field above. 815 816 =item Free/Used 817 818 The 1 or 2 rows of numbers following that correspond to the number 819 of buckets of each size between C<SMALLEST> and C<GREATEST>. In 820 the first row, the sizes (memory footprints) of buckets are powers 821 of two--or possibly one page greater. In the second row, if present, 822 the memory footprints of the buckets are between the memory footprints 823 of two buckets "above". 824 825 For example, suppose under the previous example, the memory footprints 826 were 827 828 free: 8 16 32 64 128 256 512 1024 2048 4096 8192 829 4 12 24 48 80 830 831 With non-C<DEBUGGING> perl, the buckets starting from C<128> have 832 a 4-byte overhead, and thus an 8192-long bucket may take up to 833 8188-byte allocations. 834 835 =item C<Total sbrk(): SBRKed/SBRKs:CONTINUOUS> 836 837 The first two fields give the total amount of memory perl sbrk(2)ed 838 (ess-broken? :-) and number of sbrk(2)s used. The third number is 839 what perl thinks about continuity of returned chunks. So long as 840 this number is positive, malloc() will assume that it is probable 841 that sbrk(2) will provide continuous memory. 842 843 Memory allocated by external libraries is not counted. 844 845 =item C<pad: 0> 846 847 The amount of sbrk(2)ed memory needed to keep buckets aligned. 848 849 =item C<heads: 2192> 850 851 Although memory overhead of bigger buckets is kept inside the bucket, for 852 smaller buckets, it is kept in separate areas. This field gives the 853 total size of these areas. 854 855 =item C<chain: 0> 856 857 malloc() may want to subdivide a bigger bucket into smaller buckets. 858 If only a part of the deceased bucket is left unsubdivided, the rest 859 is kept as an element of a linked list. This field gives the total 860 size of these chunks. 861 862 =item C<tail: 6144> 863 864 To minimize the number of sbrk(2)s, malloc() asks for more memory. This 865 field gives the size of the yet unused part, which is sbrk(2)ed, but 866 never touched. 867 868 =back 869 870 =head1 SEE ALSO 871 872 L<perldebug>, 873 L<perlguts>, 874 L<perlrun> 875 L<re>, 876 and 877 L<Devel::DProf>.
title
Description
Body
title
Description
Body
title
Description
Body
title
Body
| Generated: Tue Mar 17 22:47:18 2015 | Cross-referenced by PHPXref 0.7.1 |