3 perlhack - How to hack at the Perl internals
7 This document attempts to explain how Perl development takes place,
8 and ends with some suggestions for people wanting to become bona fide
11 The perl5-porters mailing list is where the Perl standard distribution
12 is maintained and developed. The list can get anywhere from 10 to 150
13 messages a day, depending on the heatedness of the debate. Most days
14 there are two or three patches, extensions, features, or bugs being
17 A searchable archive of the list is at either:
19 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
23 http://archive.develooper.com/perl5-porters@perl.org/
25 List subscribers (the porters themselves) come in several flavours.
26 Some are quiet curious lurkers, who rarely pitch in and instead watch
27 the ongoing development to ensure they're forewarned of new changes or
28 features in Perl. Some are representatives of vendors, who are there
29 to make sure that Perl continues to compile and work on their
30 platforms. Some patch any reported bug that they know how to fix,
31 some are actively patching their pet area (threads, Win32, the regexp
32 engine), while others seem to do nothing but complain. In other
33 words, it's your usual mix of technical people.
35 Over this group of porters presides Larry Wall. He has the final word
36 in what does and does not change in the Perl language. Various
37 releases of Perl are shepherded by a "pumpking", a porter
38 responsible for gathering patches, deciding on a patch-by-patch,
39 feature-by-feature basis what will and will not go into the release.
40 For instance, Gurusamy Sarathy was the pumpking for the 5.6 release of
41 Perl, and Jarkko Hietaniemi was the pumpking for the 5.8 release, and
42 Rafael Garcia-Suarez holds the pumpking crown for the 5.10 release.
44 In addition, various people are pumpkings for different things. For
45 instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the
46 I<Configure> pumpkin up till the 5.8 release. For the 5.10 release
47 H.Merijn Brand took over.
49 Larry sees Perl development along the lines of the US government:
50 there's the Legislature (the porters), the Executive branch (the
51 pumpkings), and the Supreme Court (Larry). The legislature can
52 discuss and submit patches to the executive branch all they like, but
53 the executive branch is free to veto them. Rarely, the Supreme Court
54 will side with the executive branch over the legislature, or the
55 legislature over the executive branch. Mostly, however, the
56 legislature and the executive branch are supposed to get along and
57 work out their differences without impeachment or court cases.
59 You might sometimes see reference to Rule 1 and Rule 2. Larry's power
60 as Supreme Court is expressed in The Rules:
66 Larry is always by definition right about how Perl should behave.
67 This means he has final veto power on the core functionality.
71 Larry is allowed to change his mind about any matter at a later date,
72 regardless of whether he previously invoked Rule 1.
76 Got that? Larry is always right, even when he was wrong. It's rare
77 to see either Rule exercised, but they are often alluded to.
79 New features and extensions to the language are contentious, because
80 the criteria used by the pumpkings, Larry, and other porters to decide
81 which features should be implemented and incorporated are not codified
82 in a few small design goals as with some other languages. Instead,
83 the heuristics are flexible and often difficult to fathom. Here is
84 one person's list, roughly in decreasing order of importance, of
85 heuristics that new features have to be weighed against:
89 =item Does concept match the general goals of Perl?
91 These haven't been written anywhere in stone, but one approximation
94 1. Keep it fast, simple, and useful.
95 2. Keep features/concepts as orthogonal as possible.
96 3. No arbitrary limits (platforms, data sizes, cultures).
97 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
98 5. Either assimilate new technologies, or build bridges to them.
100 =item Where is the implementation?
102 All the talk in the world is useless without an implementation. In
103 almost every case, the person or people who argue for a new feature
104 will be expected to be the ones who implement it. Porters capable
105 of coding new features have their own agendas, and are not available
106 to implement your (possibly good) idea.
108 =item Backwards compatibility
110 It's a cardinal sin to break existing Perl programs. New warnings are
111 contentious--some say that a program that emits warnings is not
112 broken, while others say it is. Adding keywords has the potential to
113 break programs, changing the meaning of existing token sequences or
114 functions might break programs.
116 =item Could it be a module instead?
118 Perl 5 has extension mechanisms, modules and XS, specifically to avoid
119 the need to keep changing the Perl interpreter. You can write modules
120 that export functions, you can give those functions prototypes so they
121 can be called like built-in functions, you can even write XS code to
122 mess with the runtime data structures of the Perl interpreter if you
123 want to implement really complicated things. If it can be done in a
124 module instead of in the core, it's highly unlikely to be added.
126 =item Is the feature generic enough?
128 Is this something that only the submitter wants added to the language,
129 or would it be broadly useful? Sometimes, instead of adding a feature
130 with a tight focus, the porters might decide to wait until someone
131 implements the more generalized feature. For instance, instead of
132 implementing a "delayed evaluation" feature, the porters are waiting
133 for a macro system that would permit delayed evaluation and much more.
135 =item Does it potentially introduce new bugs?
137 Radical rewrites of large chunks of the Perl interpreter have the
138 potential to introduce new bugs. The smaller and more localized the
141 =item Does it preclude other desirable features?
143 A patch is likely to be rejected if it closes off future avenues of
144 development. For instance, a patch that placed a true and final
145 interpretation on prototypes is likely to be rejected because there
146 are still options for the future of prototypes that haven't been
149 =item Is the implementation robust?
151 Good patches (tight code, complete, correct) stand more chance of
152 going in. Sloppy or incorrect patches might be placed on the back
153 burner until the pumpking has time to fix, or might be discarded
154 altogether without further notice.
156 =item Is the implementation generic enough to be portable?
158 The worst patches make use of a system-specific features. It's highly
159 unlikely that non-portable additions to the Perl language will be
162 =item Is the implementation tested?
164 Patches which change behaviour (fixing bugs or introducing new features)
165 must include regression tests to verify that everything works as expected.
166 Without tests provided by the original author, how can anyone else changing
167 perl in the future be sure that they haven't unwittingly broken the behaviour
168 the patch implements? And without tests, how can the patch's author be
169 confident that his/her hard work put into the patch won't be accidentally
170 thrown away by someone in the future?
172 =item Is there enough documentation?
174 Patches without documentation are probably ill-thought out or
175 incomplete. Nothing can be added without documentation, so submitting
176 a patch for the appropriate manpages as well as the source code is
179 =item Is there another way to do it?
181 Larry said "Although the Perl Slogan is I<There's More Than One Way
182 to Do It>, I hesitate to make 10 ways to do something". This is a
183 tricky heuristic to navigate, though--one man's essential addition is
184 another man's pointless cruft.
186 =item Does it create too much work?
188 Work for the pumpking, work for Perl programmers, work for module
189 authors, ... Perl is supposed to be easy.
191 =item Patches speak louder than words
193 Working code is always preferred to pie-in-the-sky ideas. A patch to
194 add a feature stands a much higher chance of making it to the language
195 than does a random feature request, no matter how fervently argued the
196 request might be. This ties into "Will it be useful?", as the fact
197 that someone took the time to make the patch demonstrates a strong
198 desire for the feature.
202 If you're on the list, you might hear the word "core" bandied
203 around. It refers to the standard distribution. "Hacking on the
204 core" means you're changing the C source code to the Perl
205 interpreter. "A core module" is one that ships with Perl.
207 =head2 Keeping in sync
209 The source code to the Perl interpreter, in its different versions, is
210 kept in a repository managed by the git revision control system. The
211 pumpkings and a few others have write access to the repository to check in
214 How to clone and use the git perl repository is described in L<perlrepository>.
216 You can also choose to use rsync to get a copy of the current source tree
217 for the bleadperl branch and all maintenance branches:
219 $ rsync -avz rsync://perl5.git.perl.org/perl-current .
220 $ rsync -avz rsync://perl5.git.perl.org/perl-5.12.x .
221 $ rsync -avz rsync://perl5.git.perl.org/perl-5.10.x .
222 $ rsync -avz rsync://perl5.git.perl.org/perl-5.8.x .
223 $ rsync -avz rsync://perl5.git.perl.org/perl-5.6.x .
224 $ rsync -avz rsync://perl5.git.perl.org/perl-5.005xx .
226 (Add the C<--delete> option to remove leftover files)
228 To get a full list of the available sync points:
230 $ rsync perl5.git.perl.org::
232 You may also want to subscribe to the perl5-changes mailing list to
233 receive a copy of each patch that gets submitted to the maintenance
234 and development "branches" of the perl repository. See
235 http://lists.perl.org/ for subscription information.
237 If you are a member of the perl5-porters mailing list, it is a good
238 thing to keep in touch with the most recent changes. If not only to
239 verify if what you would have posted as a bug report isn't already
240 solved in the most recent available perl development branch, also
241 known as perl-current, bleading edge perl, bleedperl or bleadperl.
243 Needless to say, the source code in perl-current is usually in a perpetual
244 state of evolution. You should expect it to be very buggy. Do B<not> use
245 it for any purpose other than testing and development.
247 =head2 Perlbug administration
249 There is a single remote administrative interface for modifying bug status,
250 category, open issues etc. using the B<RT> bugtracker system, maintained
251 by Robert Spier. Become an administrator, and close any bugs you can get
252 your sticky mitts on:
254 http://bugs.perl.org/
256 To email the bug system administrators:
258 "perlbug-admin" <perlbug-admin@perl.org>
260 =head2 Submitting patches
262 Always submit patches to I<perl5-porters@perl.org>. If you're
263 patching a core module and there's an author listed, send the author a
264 copy (see L<Patching a core module>). This lets other porters review
265 your patch, which catches a surprising number of errors in patches.
266 Please patch against the latest B<development> version. (e.g., even if
267 you're fixing a bug in the 5.8 track, patch against the C<blead> branch in
270 If changes are accepted, they are applied to the development branch. Then
271 the maintenance pumpking decides which of those patches is to be
272 backported to the maint branch. Only patches that survive the heat of the
273 development branch get applied to maintenance versions.
275 Your patch should update the documentation and test suite. See
276 L<Writing a test>. If you have added or removed files in the distribution,
277 edit the MANIFEST file accordingly, sort the MANIFEST file using
278 C<make manisort>, and include those changes as part of your patch.
280 Patching documentation also follows the same order: if accepted, a patch
281 is first applied to B<development>, and if relevant then it's backported
282 to B<maintenance>. (With an exception for some patches that document
283 behaviour that only appears in the maintenance branch, but which has
284 changed in the development version.)
286 To report a bug in Perl, use the program I<perlbug> which comes with
287 Perl (if you can't get Perl to work, send mail to the address
288 I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
289 I<perlbug> feeds into the automated bug-tracking system, access to
290 which is provided through the web at http://rt.perl.org/rt3/ . It
291 often pays to check the archives of the perl5-porters mailing list to
292 see whether the bug you're reporting has been reported before, and if
293 so whether it was considered a bug. See above for the location of
294 the searchable archives.
296 The CPAN testers ( http://testers.cpan.org/ ) are a group of
297 volunteers who test CPAN modules on a variety of platforms. Perl
298 Smokers ( http://www.nntp.perl.org/group/perl.daily-build and
299 http://www.nntp.perl.org/group/perl.daily-build.reports/ )
300 automatically test Perl source releases on platforms with various
301 configurations. Both efforts welcome volunteers. In order to get
302 involved in smoke testing of the perl itself visit
303 L<http://search.cpan.org/dist/Test-Smoke>. In order to start smoke
304 testing CPAN modules visit L<http://search.cpan.org/dist/CPANPLUS-YACSmoke/>
305 or L<http://search.cpan.org/dist/minismokebox/> or
306 L<http://search.cpan.org/dist/CPAN-Reporter/>.
308 It's a good idea to read and lurk for a while before chipping in.
309 That way you'll get to see the dynamic of the conversations, learn the
310 personalities of the players, and hopefully be better prepared to make
311 a useful contribution when do you speak up.
313 If after all this you still think you want to join the perl5-porters
314 mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
315 unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
317 To hack on the Perl guts, you'll need to read the following things:
323 This is of paramount importance, since it's the documentation of what
324 goes where in the Perl source. Read it over a couple of times and it
325 might start to make sense - don't worry if it doesn't yet, because the
326 best way to study it is to read it in conjunction with poking at Perl
327 source, and we'll do that later on.
329 Gisle Aas's "illustrated perlguts", also known as I<illguts>, has very
332 L<http://search.cpan.org/dist/illguts/>
334 =item L<perlxstut> and L<perlxs>
336 A working knowledge of XSUB programming is incredibly useful for core
337 hacking; XSUBs use techniques drawn from the PP code, the portion of the
338 guts that actually executes a Perl program. It's a lot gentler to learn
339 those techniques from simple examples and explanation than from the core
344 The documentation for the Perl API explains what some of the internal
345 functions do, as well as the many macros used in the source.
347 =item F<Porting/pumpkin.pod>
349 This is a collection of words of wisdom for a Perl porter; some of it is
350 only useful to the pumpkin holder, but most of it applies to anyone
351 wanting to go about Perl development.
353 =item The perl5-porters FAQ
355 This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html .
356 It contains hints on reading perl5-porters, information on how
357 perl5-porters works and how Perl development in general works.
361 =head2 Finding Your Way Around
363 Perl maintenance can be split into a number of areas, and certain people
364 (pumpkins) will have responsibility for each area. These areas sometimes
365 correspond to files or directories in the source kit. Among the areas are:
371 Modules shipped as part of the Perl core live in various subdirectories, where
372 two are dedicated to core-only modules, and two are for the dual-life modules
373 which live on CPAN and may be maintained separately with respect to the Perl
376 lib/ is for pure-Perl modules, which exist in the core only.
378 ext/ is for XS extensions, and modules with special Makefile.PL
379 requirements, which exist in the core only.
381 cpan/ is for dual-life modules, where the CPAN module is
382 canonical (should be patched first).
384 dist/ is for dual-life modules, where the blead source is
389 There are tests for nearly all the modules, built-ins and major bits
390 of functionality. Test files all have a .t suffix. Module tests live
391 in the F<lib/> and F<ext/> directories next to the module being
392 tested. Others live in F<t/>. See L<Writing a test>
396 Documentation maintenance includes looking after everything in the
397 F<pod/> directory, (as well as contributing new documentation) and
398 the documentation to the modules in core.
402 The Configure process is the way we make Perl portable across the
403 myriad of operating systems it supports. Responsibility for the
404 Configure, build and installation process, as well as the overall
405 portability of the core code rests with the Configure pumpkin -
406 others help out with individual operating systems.
408 The three files that fall under his/her responsibility are Configure,
409 config_h.SH, and Porting/Glossary (and a whole bunch of small related
410 files that are less important here). The Configure pumpkin decides how
411 patches to these are dealt with. Currently, the Configure pumpkin will
412 accept patches in most common formats, even directly to these files.
413 Other committers are allowed to commit to these files under the strict
414 condition that they will inform the Configure pumpkin, either on IRC
415 (if he/she happens to be around) or through (personal) e-mail.
417 The files involved are the operating system directories, (F<win32/>,
418 F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
419 and F<Makefile>, as well as the metaconfig files which generate
420 F<Configure>. (metaconfig isn't included in the core distribution.)
422 See http://perl5.git.perl.org/metaconfig.git/blob/HEAD:/README for a
423 description of the full process involved.
427 And of course, there's the core of the Perl interpreter itself. Let's
428 have a look at that in a little more detail.
432 Before we leave looking at the layout, though, don't forget that
433 F<MANIFEST> contains not only the file names in the Perl distribution,
434 but short descriptions of what's in them, too. For an overview of the
435 important files, try this:
437 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
439 =head2 Elements of the interpreter
441 The work of the interpreter has two main stages: compiling the code
442 into the internal representation, or bytecode, and then executing it.
443 L<perlguts/Compiled code> explains exactly how the compilation stage
446 Here is a short breakdown of perl's operation:
452 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
453 This is very high-level code, enough to fit on a single screen, and it
454 resembles the code found in L<perlembed>; most of the real action takes
457 F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at
458 make time, so you should make perl to follow this along.
460 First, F<perlmain.c> allocates some memory and constructs a Perl
461 interpreter, along these lines:
463 1 PERL_SYS_INIT3(&argc,&argv,&env);
465 3 if (!PL_do_undump) {
466 4 my_perl = perl_alloc();
469 7 perl_construct(my_perl);
470 8 PL_perl_destruct_level = 0;
473 Line 1 is a macro, and its definition is dependent on your operating
474 system. Line 3 references C<PL_do_undump>, a global variable - all
475 global variables in Perl start with C<PL_>. This tells you whether the
476 current running program was created with the C<-u> flag to perl and then
477 F<undump>, which means it's going to be false in any sane context.
479 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
480 interpreter. It's quite a simple function, and the guts of it looks like
483 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
485 Here you see an example of Perl's system abstraction, which we'll see
486 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
487 own C<malloc> as defined in F<malloc.c> if you selected that option at
490 Next, in line 7, we construct the interpreter using perl_construct,
491 also in F<perl.c>; this sets up all the special variables that Perl
492 needs, the stacks, and so on.
494 Now we pass Perl the command line options, and tell it to go:
496 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
500 exitstatus = perl_destruct(my_perl);
504 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
505 in F<perl.c>, which processes the command line options, sets up any
506 statically linked XS modules, opens the program and calls C<yyparse> to
511 The aim of this stage is to take the Perl source, and turn it into an op
512 tree. We'll see what one of those looks like later. Strictly speaking,
513 there's three things going on here.
515 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
516 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
517 B<is> a YACC grammar for Perl!) The job of the parser is to take your
518 code and "understand" it, splitting it into sentences, deciding which
519 operands go with which operators and so on.
521 The parser is nobly assisted by the lexer, which chunks up your input
522 into tokens, and decides what type of thing each token is: a variable
523 name, an operator, a bareword, a subroutine, a core function, and so on.
524 The main point of entry to the lexer is C<yylex>, and that and its
525 associated routines can be found in F<toke.c>. Perl isn't much like
526 other computer languages; it's highly context sensitive at times, it can
527 be tricky to work out what sort of token something is, or where a token
528 ends. As such, there's a lot of interplay between the tokeniser and the
529 parser, which can get pretty frightening if you're not used to it.
531 As the parser understands a Perl program, it builds up a tree of
532 operations for the interpreter to perform during execution. The routines
533 which construct and link together the various operations are to be found
534 in F<op.c>, and will be examined later.
538 Now the parsing stage is complete, and the finished tree represents
539 the operations that the Perl interpreter needs to perform to execute our
540 program. Next, Perl does a dry run over the tree looking for
541 optimisations: constant expressions such as C<3 + 4> will be computed
542 now, and the optimizer will also see if any multiple operations can be
543 replaced with a single one. For instance, to fetch the variable C<$foo>,
544 instead of grabbing the glob C<*foo> and looking at the scalar
545 component, the optimizer fiddles the op tree to use a function which
546 directly looks up the scalar in question. The main optimizer is C<peep>
547 in F<op.c>, and many ops have their own optimizing functions.
551 Now we're finally ready to go: we have compiled Perl byte code, and all
552 that's left to do is run it. The actual execution is done by the
553 C<runops_standard> function in F<run.c>; more specifically, it's done by
554 these three innocent looking lines:
556 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
560 You may be more comfortable with the Perl version of that:
562 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
564 Well, maybe not. Anyway, each op contains a function pointer, which
565 stipulates the function which will actually carry out the operation.
566 This function will return the next op in the sequence - this allows for
567 things like C<if> which choose the next op dynamically at run time.
568 The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
569 execution if required.
571 The actual functions called are known as PP code, and they're spread
572 between four files: F<pp_hot.c> contains the "hot" code, which is most
573 often used and highly optimized, F<pp_sys.c> contains all the
574 system-specific functions, F<pp_ctl.c> contains the functions which
575 implement control structures (C<if>, C<while> and the like) and F<pp.c>
576 contains everything else. These are, if you like, the C code for Perl's
577 built-in functions and operators.
579 Note that each C<pp_> function is expected to return a pointer to the next
580 op. Calls to perl subs (and eval blocks) are handled within the same
581 runops loop, and do not consume extra space on the C stack. For example,
582 C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
583 struct onto the context stack which contain the address of the op
584 following the sub call or eval. They then return the first op of that sub
585 or eval block, and so execution continues of that sub or block. Later, a
586 C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
587 retrieves the return op from it, and returns it.
589 =item Exception handing
591 Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level
592 C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
593 way to capture the current PC and SP registers and later restore them; i.e.
594 a C<longjmp()> continues at the point in code where a previous C<setjmp()>
595 was done, with anything further up on the C stack being lost. This is why
596 code should always save values using C<SAVE_FOO> rather than in auto
599 The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
600 C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
601 C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
602 C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
604 At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
605 C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
606 loop or whatever, and handle possible exception returns. For a 2 return,
607 final cleanup is performed, such as popping stacks and calling C<CHECK> or
608 C<END> blocks. Amongst other things, this is how scope cleanup still
609 occurs during an C<exit>.
611 If a C<die> can find a C<CxEVAL> block on the context stack, then the
612 stack is popped to that level and the return op in that block is assigned
613 to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
614 passes control back to the guard. In the case of C<perl_run> and
615 C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
616 loop. The is the normal way that C<die> or C<croak> is handled within an
619 Sometimes ops are executed within an inner runops loop, such as tie, sort
620 or overload code. In this case, something like
622 sub FETCH { eval { die } }
624 would cause a longjmp right back to the guard in C<perl_run>, popping both
625 runops loops, which is clearly incorrect. One way to avoid this is for the
626 tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
627 runops loop, but for efficiency reasons, perl in fact just sets a flag,
628 using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
629 C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
630 which does a C<JMPENV_PUSH> and starts a new runops level to execute the
631 code, rather than doing it on the current loop.
633 As a further optimisation, on exit from the eval block in the C<FETCH>,
634 execution of the code following the block is still carried on in the inner
635 loop. When an exception is raised, C<docatch> compares the C<JMPENV>
636 level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
637 re-throws the exception. In this way any inner loops get popped.
641 1: eval { tie @a, 'A' };
647 To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
648 enters a runops loop. This loop executes the eval and tie ops on line 1,
649 with the eval pushing a C<CxEVAL> onto the context stack.
651 The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
652 to execute the body of C<TIEARRAY>. When it executes the entertry op on
653 line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
654 does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
655 the die op. At this point the C call stack looks like this:
658 Perl_runops # third loop
662 Perl_runops # second loop
666 Perl_runops # first loop
671 and the context and data stacks, as shown by C<-Dstv>, look like:
675 CX 1: EVAL => AV() PV("A"\0)
683 The die pops the first C<CxEVAL> off the context stack, sets
684 C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
685 the top C<docatch>. This then starts another third-level runops level,
686 which executes the nextstate, pushmark and die ops on line 4. At the point
687 that the second C<pp_die> is called, the C call stack looks exactly like
688 that above, even though we are no longer within an inner eval; this is
689 because of the optimization mentioned earlier. However, the context stack
690 now looks like this, ie with the top CxEVAL popped:
694 CX 1: EVAL => AV() PV("A"\0)
700 The die on line 4 pops the context stack back down to the CxEVAL, leaving
706 As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
707 C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
711 Perl_runops # second loop
715 Perl_runops # first loop
720 In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
721 differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
722 and the C stack unwinds to:
727 Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
728 and execution continues.
732 =head2 Internal Variable Types
734 You should by now have had a look at L<perlguts>, which tells you about
735 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
738 These variables are used not only to represent Perl-space variables, but
739 also any constants in the code, as well as some structures completely
740 internal to Perl. The symbol table, for instance, is an ordinary Perl
741 hash. Your code is represented by an SV as it's read into the parser;
742 any program files you call are opened via ordinary Perl filehandles, and
745 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
746 Perl program. Let's see, for instance, how Perl treats the constant
749 % perl -MDevel::Peek -e 'Dump("hello")'
750 1 SV = PV(0xa041450) at 0xa04ecbc
752 3 FLAGS = (POK,READONLY,pPOK)
753 4 PV = 0xa0484e0 "hello"\0
757 Reading C<Devel::Peek> output takes a bit of practise, so let's go
758 through it line by line.
760 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
761 memory. SVs themselves are very simple structures, but they contain a
762 pointer to a more complex structure. In this case, it's a PV, a
763 structure which holds a string value, at location C<0xa041450>. Line 2
764 is the reference count; there are no other references to this data, so
767 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
768 read-only SV (because it's a constant) and the data is a PV internally.
769 Next we've got the contents of the string, starting at location
772 Line 5 gives us the current length of the string - note that this does
773 B<not> include the null terminator. Line 6 is not the length of the
774 string, but the length of the currently allocated buffer; as the string
775 grows, Perl automatically extends the available storage via a routine
778 You can get at any of these quantities from C very easily; just add
779 C<Sv> to the name of the field shown in the snippet, and you've got a
780 macro which will return the value: C<SvCUR(sv)> returns the current
781 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
782 C<SvPV(sv, len)> returns the string itself with its length, and so on.
783 More macros to manipulate these properties can be found in L<perlguts>.
785 Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
788 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
793 6 junk = SvPV_force(sv, tlen);
794 7 SvGROW(sv, tlen + len + 1);
797 10 Move(ptr,SvPVX(sv)+tlen,len,char);
799 12 *SvEND(sv) = '\0';
800 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
804 This is a function which adds a string, C<ptr>, of length C<len> onto
805 the end of the PV stored in C<sv>. The first thing we do in line 6 is
806 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
807 macro to force a PV. As a side effect, C<tlen> gets set to the current
808 value of the PV, and the PV itself is returned to C<junk>.
810 In line 7, we make sure that the SV will have enough room to accommodate
811 the old string, the new string and the null terminator. If C<LEN> isn't
812 big enough, C<SvGROW> will reallocate space for us.
814 Now, if C<junk> is the same as the string we're trying to add, we can
815 grab the string directly from the SV; C<SvPVX> is the address of the PV
818 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
819 memory around: we move the string C<ptr> to the end of the PV - that's
820 the start of the PV plus its current length. We're moving C<len> bytes
821 of type C<char>. After doing so, we need to tell Perl we've extended the
822 string, by altering C<CUR> to reflect the new length. C<SvEND> is a
823 macro which gives us the end of the string, so that needs to be a
826 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
827 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
828 want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
829 version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
830 and turns on POK. The final C<SvTAINT> is a macro which launders tainted
831 data if taint mode is turned on.
833 AVs and HVs are more complicated, but SVs are by far the most common
834 variable type being thrown around. Having seen something of how we
835 manipulate these, let's go on and look at how the op tree is
840 First, what is the op tree, anyway? The op tree is the parsed
841 representation of your program, as we saw in our section on parsing, and
842 it's the sequence of operations that Perl goes through to execute your
843 program, as we saw in L</Running>.
845 An op is a fundamental operation that Perl can perform: all the built-in
846 functions and operators are ops, and there are a series of ops which
847 deal with concepts the interpreter needs internally - entering and
848 leaving a block, ending a statement, fetching a variable, and so on.
850 The op tree is connected in two ways: you can imagine that there are two
851 "routes" through it, two orders in which you can traverse the tree.
852 First, parse order reflects how the parser understood the code, and
853 secondly, execution order tells perl what order to perform the
856 The easiest way to examine the op tree is to stop Perl after it has
857 finished parsing, and get it to dump out the tree. This is exactly what
858 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
859 and L<B::Debug|B::Debug> do.
861 Let's have a look at how Perl sees C<$a = $b + $c>:
863 % perl -MO=Terse -e '$a=$b+$c'
864 1 LISTOP (0x8179888) leave
865 2 OP (0x81798b0) enter
866 3 COP (0x8179850) nextstate
867 4 BINOP (0x8179828) sassign
868 5 BINOP (0x8179800) add [1]
869 6 UNOP (0x81796e0) null [15]
870 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
871 8 UNOP (0x81797e0) null [15]
872 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
873 10 UNOP (0x816b4f0) null [15]
874 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
876 Let's start in the middle, at line 4. This is a BINOP, a binary
877 operator, which is at location C<0x8179828>. The specific operator in
878 question is C<sassign> - scalar assignment - and you can find the code
879 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
880 binary operator, it has two children: the add operator, providing the
881 result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
884 Line 10 is the null op: this does exactly nothing. What is that doing
885 there? If you see the null op, it's a sign that something has been
886 optimized away after parsing. As we mentioned in L</Optimization>,
887 the optimization stage sometimes converts two operations into one, for
888 example when fetching a scalar variable. When this happens, instead of
889 rewriting the op tree and cleaning up the dangling pointers, it's easier
890 just to replace the redundant operation with the null op. Originally,
891 the tree would have looked like this:
893 10 SVOP (0x816b4f0) rv2sv [15]
894 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
896 That is, fetch the C<a> entry from the main symbol table, and then look
897 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
898 happens to do both these things.
900 The right hand side, starting at line 5 is similar to what we've just
901 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
904 Now, what's this about?
906 1 LISTOP (0x8179888) leave
907 2 OP (0x81798b0) enter
908 3 COP (0x8179850) nextstate
910 C<enter> and C<leave> are scoping ops, and their job is to perform any
911 housekeeping every time you enter and leave a block: lexical variables
912 are tidied up, unreferenced variables are destroyed, and so on. Every
913 program will have those first three lines: C<leave> is a list, and its
914 children are all the statements in the block. Statements are delimited
915 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
916 the ops to be performed for each statement being the children of
917 C<nextstate>. C<enter> is a single op which functions as a marker.
919 That's how Perl parsed the program, from top to bottom:
932 However, it's impossible to B<perform> the operations in this order:
933 you have to find the values of C<$b> and C<$c> before you add them
934 together, for instance. So, the other thread that runs through the op
935 tree is the execution order: each op has a field C<op_next> which points
936 to the next op to be run, so following these pointers tells us how perl
937 executes the code. We can traverse the tree in this order using
938 the C<exec> option to C<B::Terse>:
940 % perl -MO=Terse,exec -e '$a=$b+$c'
941 1 OP (0x8179928) enter
942 2 COP (0x81798c8) nextstate
943 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
944 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
945 5 BINOP (0x8179878) add [1]
946 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
947 7 BINOP (0x81798a0) sassign
948 8 LISTOP (0x8179900) leave
950 This probably makes more sense for a human: enter a block, start a
951 statement. Get the values of C<$b> and C<$c>, and add them together.
952 Find C<$a>, and assign one to the other. Then leave.
954 The way Perl builds up these op trees in the parsing process can be
955 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
956 piece we need to construct the tree for C<$a = $b + $c>
958 1 term : term ASSIGNOP term
959 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
961 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
963 If you're not used to reading BNF grammars, this is how it works: You're
964 fed certain things by the tokeniser, which generally end up in upper
965 case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
966 code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
967 "terminal symbols", because you can't get any simpler than them.
969 The grammar, lines one and three of the snippet above, tells you how to
970 build up more complex forms. These complex forms, "non-terminal symbols"
971 are generally placed in lower case. C<term> here is a non-terminal
972 symbol, representing a single expression.
974 The grammar gives you the following rule: you can make the thing on the
975 left of the colon if you see all the things on the right in sequence.
976 This is called a "reduction", and the aim of parsing is to completely
977 reduce the input. There are several different ways you can perform a
978 reduction, separated by vertical bars: so, C<term> followed by C<=>
979 followed by C<term> makes a C<term>, and C<term> followed by C<+>
980 followed by C<term> can also make a C<term>.
982 So, if you see two terms with an C<=> or C<+>, between them, you can
983 turn them into a single expression. When you do this, you execute the
984 code in the block on the next line: if you see C<=>, you'll do the code
985 in line 2. If you see C<+>, you'll do the code in line 4. It's this code
986 which contributes to the op tree.
989 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
991 What this does is creates a new binary op, and feeds it a number of
992 variables. The variables refer to the tokens: C<$1> is the first token in
993 the input, C<$2> the second, and so on - think regular expression
994 backreferences. C<$$> is the op returned from this reduction. So, we
995 call C<newBINOP> to create a new binary operator. The first parameter to
996 C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
997 operator, so we want the type to be C<ADDOP>. We could specify this
998 directly, but it's right there as the second token in the input, so we
999 use C<$2>. The second parameter is the op's flags: 0 means "nothing
1000 special". Then the things to add: the left and right hand side of our
1001 expression, in scalar context.
1005 When perl executes something like C<addop>, how does it pass on its
1006 results to the next op? The answer is, through the use of stacks. Perl
1007 has a number of stacks to store things it's currently working on, and
1008 we'll look at the three most important ones here.
1012 =item Argument stack
1014 Arguments are passed to PP code and returned from PP code using the
1015 argument stack, C<ST>. The typical way to handle arguments is to pop
1016 them off the stack, deal with them how you wish, and then push the result
1017 back onto the stack. This is how, for instance, the cosine operator
1022 value = Perl_cos(value);
1025 We'll see a more tricky example of this when we consider Perl's macros
1026 below. C<POPn> gives you the NV (floating point value) of the top SV on
1027 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1028 the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1029 should be extended if necessary - it can't be necessary here, because we
1030 know there's room for one more item on the stack, since we've just
1031 removed one! The C<XPUSH*> macros at least guarantee safety.
1033 Alternatively, you can fiddle with the stack directly: C<SP> gives you
1034 the first element in your portion of the stack, and C<TOP*> gives you
1035 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1036 negation of an integer:
1040 Just set the integer value of the top stack entry to its negation.
1042 Argument stack manipulation in the core is exactly the same as it is in
1043 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1044 description of the macros used in stack manipulation.
1048 I say "your portion of the stack" above because PP code doesn't
1049 necessarily get the whole stack to itself: if your function calls
1050 another function, you'll only want to expose the arguments aimed for the
1051 called function, and not (necessarily) let it get at your own data. The
1052 way we do this is to have a "virtual" bottom-of-stack, exposed to each
1053 function. The mark stack keeps bookmarks to locations in the argument
1054 stack usable by each function. For instance, when dealing with a tied
1055 variable, (internally, something with "P" magic) Perl has to call
1056 methods for accesses to the tied variables. However, we need to separate
1057 the arguments exposed to the method to the argument exposed to the
1058 original function - the store or fetch or whatever it may be. Here's
1059 roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
1063 3 PUSHs(SvTIED_obj((SV*)av, mg));
1067 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1070 Let's examine the whole implementation, for practice:
1074 Push the current state of the stack pointer onto the mark stack. This is
1075 so that when we've finished adding items to the argument stack, Perl
1076 knows how many things we've added recently.
1079 3 PUSHs(SvTIED_obj((SV*)av, mg));
1082 We're going to add two more items onto the argument stack: when you have
1083 a tied array, the C<PUSH> subroutine receives the object and the value
1084 to be pushed, and that's exactly what we have here - the tied object,
1085 retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1089 Next we tell Perl to update the global stack pointer from our internal
1090 variable: C<dSP> only gave us a local copy, not a reference to the global.
1093 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1096 C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1097 variables are tidied up, everything that has been localised gets
1098 its previous value returned, and so on. Think of them as the C<{> and
1099 C<}> of a Perl block.
1101 To actually do the magic method call, we have to call a subroutine in
1102 Perl space: C<call_method> takes care of that, and it's described in
1103 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1104 going to discard its return value. The call_method() function
1105 removes the top element of the mark stack, so there is nothing for
1106 the caller to clean up.
1110 C doesn't have a concept of local scope, so perl provides one. We've
1111 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1112 stack implements the C equivalent of, for example:
1119 See L<perlguts/Localising Changes> for how to use the save stack.
1123 =head2 Millions of Macros
1125 One thing you'll notice about the Perl source is that it's full of
1126 macros. Some have called the pervasive use of macros the hardest thing
1127 to understand, others find it adds to clarity. Let's take an example,
1128 the code which implements the addition operator:
1132 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1135 6 SETn( left + right );
1140 Every line here (apart from the braces, of course) contains a macro. The
1141 first line sets up the function declaration as Perl expects for PP code;
1142 line 3 sets up variable declarations for the argument stack and the
1143 target, the return value of the operation. Finally, it tries to see if
1144 the addition operation is overloaded; if so, the appropriate subroutine
1147 Line 5 is another variable declaration - all variable declarations start
1148 with C<d> - which pops from the top of the argument stack two NVs (hence
1149 C<nn>) and puts them into the variables C<right> and C<left>, hence the
1150 C<rl>. These are the two operands to the addition operator. Next, we
1151 call C<SETn> to set the NV of the return value to the result of adding
1152 the two values. This done, we return - the C<RETURN> macro makes sure
1153 that our return value is properly handled, and we pass the next operator
1154 to run back to the main run loop.
1156 Most of these macros are explained in L<perlapi>, and some of the more
1157 important ones are explained in L<perlxs> as well. Pay special attention
1158 to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1159 the C<[pad]THX_?> macros.
1161 =head2 The .i Targets
1163 You can expand the macros in a F<foo.c> file by saying
1167 which will expand the macros using cpp. Don't be scared by the results.
1169 =head1 SOURCE CODE STATIC ANALYSIS
1171 Various tools exist for analysing C source code B<statically>, as
1172 opposed to B<dynamically>, that is, without executing the code.
1173 It is possible to detect resource leaks, undefined behaviour, type
1174 mismatches, portability problems, code paths that would cause illegal
1175 memory accesses, and other similar problems by just parsing the C code
1176 and looking at the resulting graph, what does it tell about the
1177 execution and data flows. As a matter of fact, this is exactly
1178 how C compilers know to give warnings about dubious code.
1182 The good old C code quality inspector, C<lint>, is available in
1183 several platforms, but please be aware that there are several
1184 different implementations of it by different vendors, which means that
1185 the flags are not identical across different platforms.
1187 There is a lint variant called C<splint> (Secure Programming Lint)
1188 available from http://www.splint.org/ that should compile on any
1191 There are C<lint> and <splint> targets in Makefile, but you may have
1192 to diddle with the flags (see above).
1196 Coverity (http://www.coverity.com/) is a product similar to lint and
1197 as a testbed for their product they periodically check several open
1198 source projects, and they give out accounts to open source developers
1199 to the defect databases.
1201 =head2 cpd (cut-and-paste detector)
1203 The cpd tool detects cut-and-paste coding. If one instance of the
1204 cut-and-pasted code changes, all the other spots should probably be
1205 changed, too. Therefore such code should probably be turned into a
1206 subroutine or a macro.
1208 cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1209 (http://pmd.sourceforge.net/). pmd was originally written for static
1210 analysis of Java code, but later the cpd part of it was extended to
1211 parse also C and C++.
1213 Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1214 pmd-X.Y.jar from it, and then run that on source code thusly:
1216 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1218 You may run into memory limits, in which case you should use the -Xmx option:
1224 Though much can be written about the inconsistency and coverage
1225 problems of gcc warnings (like C<-Wall> not meaning "all the
1226 warnings", or some common portability problems not being covered by
1227 C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1228 collection of warnings, and so forth), gcc is still a useful tool in
1229 keeping our coding nose clean.
1231 The C<-Wall> is by default on.
1233 The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1234 always, but unfortunately they are not safe on all platforms, they can
1235 for example cause fatal conflicts with the system headers (Solaris
1236 being a prime example). If Configure C<-Dgccansipedantic> is used,
1237 the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1238 where they are known to be safe.
1240 Starting from Perl 5.9.4 the following extra flags are added:
1254 C<-Wdeclaration-after-statement>
1258 The following flags would be nice to have but they would first need
1259 their own Augean stablemaster:
1273 C<-Wstrict-prototypes>
1277 The C<-Wtraditional> is another example of the annoying tendency of
1278 gcc to bundle a lot of warnings under one switch (it would be
1279 impossible to deploy in practice because it would complain a lot) but
1280 it does contain some warnings that would be beneficial to have available
1281 on their own, such as the warning about string constants inside macros
1282 containing the macro arguments: this behaved differently pre-ANSI
1283 than it does in ANSI, and some C compilers are still in transition,
1284 AIX being an example.
1286 =head2 Warnings of other C compilers
1288 Other C compilers (yes, there B<are> other C compilers than gcc) often
1289 have their "strict ANSI" or "strict ANSI with some portability extensions"
1290 modes on, like for example the Sun Workshop has its C<-Xa> mode on
1291 (though implicitly), or the DEC (these days, HP...) has its C<-std1>
1296 You can compile a special debugging version of Perl, which allows you
1297 to use the C<-D> option of Perl to tell more about what Perl is doing.
1298 But sometimes there is no alternative than to dive in with a debugger,
1299 either to see the stack trace of a core dump (very useful in a bug
1300 report), or trying to figure out what went wrong before the core dump
1301 happened, or how did we end up having wrong or unexpected results.
1303 =head2 Poking at Perl
1305 To really poke around with Perl, you'll probably want to build Perl for
1306 debugging, like this:
1308 ./Configure -d -D optimize=-g
1311 C<-g> is a flag to the C compiler to have it produce debugging
1312 information which will allow us to step through a running program,
1313 and to see in which C function we are at (without the debugging
1314 information we might see only the numerical addresses of the functions,
1315 which is not very helpful).
1317 F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1318 enables all the internal debugging code in Perl. There are a whole bunch
1319 of things you can debug with this: L<perlrun> lists them all, and the
1320 best way to find out about them is to play about with them. The most
1321 useful options are probably
1323 l Context (loop) stack processing
1325 o Method and overloading resolution
1326 c String/numeric conversions
1328 Some of the functionality of the debugging code can be achieved using XS
1331 -Dr => use re 'debug'
1332 -Dx => use O 'Debug'
1334 =head2 Using a source-level debugger
1336 If the debugging output of C<-D> doesn't help you, it's time to step
1337 through perl's execution with a source-level debugger.
1343 We'll use C<gdb> for our examples here; the principles will apply to
1344 any debugger (many vendors call their debugger C<dbx>), but check the
1345 manual of the one you're using.
1349 To fire up the debugger, type
1353 Or if you have a core dump:
1357 You'll want to do that in your Perl source tree so the debugger can read
1358 the source code. You should see the copyright message, followed by the
1363 C<help> will get you into the documentation, but here are the most
1370 Run the program with the given arguments.
1372 =item break function_name
1374 =item break source.c:xxx
1376 Tells the debugger that we'll want to pause execution when we reach
1377 either the named function (but see L<perlguts/Internal Functions>!) or the given
1378 line in the named source file.
1382 Steps through the program a line at a time.
1386 Steps through the program a line at a time, without descending into
1391 Run until the next breakpoint.
1395 Run until the end of the current function, then stop again.
1399 Just pressing Enter will do the most recent operation again - it's a
1400 blessing when stepping through miles of source code.
1404 Execute the given C code and print its results. B<WARNING>: Perl makes
1405 heavy use of macros, and F<gdb> does not necessarily support macros
1406 (see later L</"gdb macro support">). You'll have to substitute them
1407 yourself, or to invoke cpp on the source code files
1408 (see L</"The .i Targets">)
1409 So, for instance, you can't say
1411 print SvPV_nolen(sv)
1415 print Perl_sv_2pv_nolen(sv)
1419 You may find it helpful to have a "macro dictionary", which you can
1420 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1421 recursively apply those macros for you.
1423 =head2 gdb macro support
1425 Recent versions of F<gdb> have fairly good macro support, but
1426 in order to use it you'll need to compile perl with macro definitions
1427 included in the debugging information. Using F<gcc> version 3.1, this
1428 means configuring with C<-Doptimize=-g3>. Other compilers might use a
1429 different switch (if they support debugging macros at all).
1431 =head2 Dumping Perl Data Structures
1433 One way to get around this macro hell is to use the dumping functions in
1434 F<dump.c>; these work a little like an internal
1435 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1436 that you can't get at from Perl. Let's take an example. We'll use the
1437 C<$a = $b + $c> we used before, but give it a bit of context:
1438 C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1440 What about C<pp_add>, the function we examined earlier to implement the
1443 (gdb) break Perl_pp_add
1444 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1446 Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
1447 With the breakpoint in place, we can run our program:
1449 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1451 Lots of junk will go past as gdb reads in the relevant source files and
1452 libraries, and then:
1454 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
1455 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1460 We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1461 arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1464 #define dPOPTOPnnrl_ul NV right = POPn; \
1465 SV *leftsv = TOPs; \
1466 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1468 C<POPn> takes the SV from the top of the stack and obtains its NV either
1469 directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1470 C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1471 C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1472 C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1474 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1475 convert it. If we step again, we'll find ourselves there:
1477 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1481 We can now use C<Perl_sv_dump> to investigate the SV:
1483 SV = PV(0xa057cc0) at 0xa0675d0
1486 PV = 0xa06a510 "6XXXX"\0
1491 We know we're going to get C<6> from this, so let's finish the
1495 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1496 0x462669 in Perl_pp_add () at pp_hot.c:311
1499 We can also dump out this op: the current op is always stored in
1500 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1501 similar output to L<B::Debug|B::Debug>.
1504 13 TYPE = add ===> 14
1506 FLAGS = (SCALAR,KIDS)
1508 TYPE = null ===> (12)
1510 FLAGS = (SCALAR,KIDS)
1512 11 TYPE = gvsv ===> 12
1518 # finish this later #
1522 All right, we've now had a look at how to navigate the Perl sources and
1523 some things you'll need to know when fiddling with them. Let's now get
1524 on and create a simple patch. Here's something Larry suggested: if a
1525 C<U> is the first active format during a C<pack>, (for example,
1526 C<pack "U3C8", @stuff>) then the resulting string should be treated as
1529 If you are working with a git clone of the Perl repository, you will want to
1530 create a branch for your changes. This will make creating a proper patch much
1531 simpler. See the L<perlrepository> for details on how to do this.
1533 How do we prepare to fix this up? First we locate the code in question -
1534 the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1535 files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1536 altering this file, let's copy it to F<pp.c~>.
1538 [Well, it was in F<pp.c> when this tutorial was written. It has now been
1539 split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1541 Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1542 loop over the pattern, taking each format character in turn into
1543 C<datum_type>. Then for each possible format character, we swallow up
1544 the other arguments in the pattern (a field width, an asterisk, and so
1545 on) and convert the next chunk input into the specified format, adding
1546 it onto the output SV C<cat>.
1548 How do we know if the C<U> is the first format in the C<pat>? Well, if
1549 we have a pointer to the start of C<pat> then, if we see a C<U> we can
1550 test whether we're still at the start of the string. So, here's where
1554 register char *pat = SvPVx(*++MARK, fromlen);
1555 register char *patend = pat + fromlen;
1560 We'll have another string pointer in there:
1563 register char *pat = SvPVx(*++MARK, fromlen);
1564 register char *patend = pat + fromlen;
1570 And just before we start the loop, we'll set C<patcopy> to be the start
1575 sv_setpvn(cat, "", 0);
1577 while (pat < patend) {
1579 Now if we see a C<U> which was at the start of the string, we turn on
1580 the C<UTF8> flag for the output SV, C<cat>:
1582 + if (datumtype == 'U' && pat==patcopy+1)
1584 if (datumtype == '#') {
1585 while (pat < patend && *pat != '\n')
1588 Remember that it has to be C<patcopy+1> because the first character of
1589 the string is the C<U> which has been swallowed into C<datumtype!>
1591 Oops, we forgot one thing: what if there are spaces at the start of the
1592 pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1593 character, even though it's not the first thing in the pattern. In this
1594 case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1596 if (isSPACE(datumtype))
1601 if (isSPACE(datumtype)) {
1606 OK. That's the C part done. Now we must do two additional things before
1607 this patch is ready to go: we've changed the behaviour of Perl, and so
1608 we must document that change. We must also provide some more regression
1609 tests to make sure our patch works and doesn't create a bug somewhere
1610 else along the line.
1612 The regression tests for each operator live in F<t/op/>, and so we
1613 make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1614 tests to the end. First, we'll test that the C<U> does indeed create
1617 t/op/pack.t has a sensible ok() function, but if it didn't we could
1618 use the one from t/test.pl.
1620 require './test.pl';
1621 plan( tests => 159 );
1625 print 'not ' unless "1.20.300.4000" eq sprintf "%vd",
1626 pack("U*",1,20,300,4000);
1627 print "ok $test\n"; $test++;
1629 we can write the more sensible (see L<Test::More> for a full
1630 explanation of is() and other testing functions).
1632 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
1633 "U* produces Unicode" );
1635 Now we'll test that we got that space-at-the-beginning business right:
1637 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
1638 " with spaces at the beginning" );
1640 And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1641 the first active format:
1643 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
1644 "U* not first isn't Unicode" );
1646 Mustn't forget to change the number of tests which appears at the top,
1647 or else the automated tester will get confused. This will either look
1654 plan( tests => 156 );
1656 We now compile up Perl, and run it through the test suite. Our new
1659 Finally, the documentation. The job is never done until the paperwork is
1660 over, so let's describe the change we've just made. The relevant place
1661 is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1662 this text in the description of C<pack>:
1666 If the pattern begins with a C<U>, the resulting string will be treated
1667 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1668 with an initial C<U0>, and the bytes that follow will be interpreted as
1669 Unicode characters. If you don't want this to happen, you can begin
1670 your pattern with C<C0> (or anything else) to force Perl not to UTF-8
1671 encode your string, and then follow this with a C<U*> somewhere in your
1674 =head2 Patching a core module
1676 This works just like patching anything else, with an extra
1677 consideration. Many core modules also live on CPAN. If this is so,
1678 patch the CPAN version instead of the core and send the patch off to
1679 the module maintainer (with a copy to p5p). This will help the module
1680 maintainer keep the CPAN version in sync with the core version without
1681 constantly scanning p5p.
1683 The list of maintainers of core modules is usefully documented in
1684 F<Porting/Maintainers.pl>.
1686 =head2 Adding a new function to the core
1688 If, as part of a patch to fix a bug, or just because you have an
1689 especially good idea, you decide to add a new function to the core,
1690 discuss your ideas on p5p well before you start work. It may be that
1691 someone else has already attempted to do what you are considering and
1692 can give lots of good advice or even provide you with bits of code
1693 that they already started (but never finished).
1695 You have to follow all of the advice given above for patching. It is
1696 extremely important to test any addition thoroughly and add new tests
1697 to explore all boundary conditions that your new function is expected
1698 to handle. If your new function is used only by one module (e.g. toke),
1699 then it should probably be named S_your_function (for static); on the
1700 other hand, if you expect it to accessible from other functions in
1701 Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1704 The location of any new code is also an important consideration. Don't
1705 just create a new top level .c file and put your code there; you would
1706 have to make changes to Configure (so the Makefile is created properly),
1707 as well as possibly lots of include files. This is strictly pumpking
1710 It is better to add your function to one of the existing top level
1711 source code files, but your choice is complicated by the nature of
1712 the Perl distribution. Only the files that are marked as compiled
1713 static are located in the perl executable. Everything else is located
1714 in the shared library (or DLL if you are running under WIN32). So,
1715 for example, if a function was only used by functions located in
1716 toke.c, then your code can go in toke.c. If, however, you want to call
1717 the function from universal.c, then you should put your code in another
1718 location, for example util.c.
1720 In addition to writing your c-code, you will need to create an
1721 appropriate entry in embed.pl describing your function, then run
1722 'make regen_headers' to create the entries in the numerous header
1723 files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1724 for information on the various options that you can set in embed.pl.
1725 You will forget to do this a few (or many) times and you will get
1726 warnings during the compilation phase. Make sure that you mention
1727 this when you post your patch to P5P; the pumpking needs to know this.
1729 When you write your new code, please be conscious of existing code
1730 conventions used in the perl source files. See L<perlstyle> for
1731 details. Although most of the guidelines discussed seem to focus on
1732 Perl code, rather than c, they all apply (except when they don't ;).
1733 Also see L<perlrepository> for lots of details about both formatting and
1734 submitting patches of your changes.
1736 Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1737 Test on as many platforms as you can find. Test as many perl
1738 Configure options as you can (e.g. MULTIPLICITY). If you have
1739 profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
1740 below for how to use them to further test your code. Remember that
1741 most of the people on P5P are doing this on their own time and
1742 don't have the time to debug your code.
1744 =head2 Writing a test
1746 Every module and built-in function has an associated test file (or
1747 should...). If you add or change functionality, you have to write a
1748 test. If you fix a bug, you have to write a test so that bug never
1749 comes back. If you alter the docs, it would be nice to test what the
1750 new documentation says.
1752 In short, if you submit a patch you probably also have to patch the
1755 For modules, the test file is right next to the module itself.
1756 F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1757 so there are some snags (and it would be wonderful for you to brush
1758 them out), but it basically works that way. Everything else lives in
1761 If you add a new test directory under F<t/>, it is imperative that you
1762 add that directory to F<t/HARNESS> and F<t/TEST>.
1768 Testing of the absolute basic functionality of Perl. Things like
1769 C<if>, basic file reads and writes, simple regexes, etc. These are
1770 run first in the test suite and if any of them fail, something is
1775 These test the basic control structures, C<if/else>, C<while>,
1780 Tests basic issues of how Perl parses and compiles itself.
1784 Tests for built-in IO functions, including command line arguments.
1788 The old home for the module tests, you shouldn't put anything new in
1789 here. There are still some bits and pieces hanging around in here
1790 that need to be moved. Perhaps you could move them? Thanks!
1794 Tests for perl's method resolution order implementations
1799 Tests for perl's built in functions that don't fit into any of the
1804 Tests for regex related functions or behaviour. (These used to live
1809 Testing features of how perl actually runs, including exit codes and
1810 handling of PERL* environment variables.
1814 Tests for the core support of Unicode.
1818 Windows-specific tests.
1822 A test suite for the s2p converter.
1826 The core uses the same testing style as the rest of Perl, a simple
1827 "ok/not ok" run through Test::Harness, but there are a few special
1830 There are three ways to write a test in the core. Test::More,
1831 t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1832 decision of which to use depends on what part of the test suite you're
1833 working on. This is a measure to prevent a high-level failure (such
1834 as Config.pm breaking) from causing basic functionality tests to fail.
1835 If you write your own test, use the L<Test Anything Protocol|TAP>.
1841 Since we don't know if require works, or even subroutines, use ad hoc
1842 tests for these two. Step carefully to avoid using the feature being
1845 =item t/cmd t/run t/io t/op
1847 Now that basic require() and subroutines are tested, you can use the
1848 t/test.pl library which emulates the important features of Test::More
1849 while using a minimum of core features.
1851 You can also conditionally use certain libraries like Config, but be
1852 sure to skip the test gracefully if it's not there.
1856 Now that the core of Perl is tested, Test::More can be used. You can
1857 also use the full suite of core modules in the tests.
1861 When you say "make test" Perl uses the F<t/TEST> program to run the
1862 test suite (except under Win32 where it uses F<t/harness> instead.)
1863 All tests are run from the F<t/> directory, B<not> the directory
1864 which contains the test. This causes some problems with the tests
1865 in F<lib/>, so here's some opportunity for some patching.
1867 You must be triply conscious of cross-platform concerns. This usually
1868 boils down to using File::Spec and avoiding things like C<fork()> and
1869 C<system()> unless absolutely necessary.
1871 =head2 Special Make Test Targets
1873 There are various special make targets that can be used to test Perl
1874 slightly differently than the standard "test" target. Not all them
1875 are expected to give a 100% success rate. Many of them have several
1876 aliases, and many of them are not available on certain operating
1883 Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
1885 (Not available on Win32)
1889 Run all the tests through B::Deparse. Not all tests will succeed.
1891 (Not available on Win32)
1893 =item test.taintwarn
1895 Run all tests with the B<-t> command-line switch. Not all tests
1896 are expected to succeed (until they're specifically fixed, of course).
1898 (Not available on Win32)
1902 Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
1903 F<t/op>, F<t/uni> and F<t/mro> tests.
1905 =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1907 (Only in Linux) Run all the tests using the memory leak + naughty
1908 memory access tool "valgrind". The log files will be named
1909 F<testname.valgrind>.
1911 =item test.third check.third utest.third ucheck.third
1913 (Only in Tru64) Run all the tests using the memory leak + naughty
1914 memory access tool "Third Degree". The log files will be named
1915 F<perl.3log.testname>.
1917 =item test.torture torturetest
1919 Run all the usual tests and some extra tests. As of Perl 5.8.0 the
1920 only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
1922 You can also run the torture test with F<t/harness> by giving
1923 C<-torture> argument to F<t/harness>.
1925 =item utest ucheck test.utf8 check.utf8
1927 Run all the tests with -Mutf8. Not all tests will succeed.
1929 (Not available on Win32)
1931 =item minitest.utf16 test.utf16
1933 Runs the tests with UTF-16 encoded scripts, encoded with different
1934 versions of this encoding.
1936 C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
1937 C<-utf16> arguments to F<t/TEST>.
1939 (Not available on Win32)
1943 Run the test suite with the F<t/harness> controlling program, instead of
1944 F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1945 L<Test::Harness> module, thus using this test target supposes that perl
1946 mostly works. The main advantage for our purposes is that it prints a
1947 detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1948 doesn't redirect stderr to stdout.
1950 Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
1951 there is no special "test_harness" target.
1953 Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
1954 environment variables to control the behaviour of F<t/harness>. This means
1957 nmake test TEST_FILES="op/*.t"
1958 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
1960 =item Parallel tests
1962 The core distribution can now run its regression tests in parallel on
1963 Unix-like platforms. Instead of running C<make test>, set C<TEST_JOBS> in
1964 your environment to the number of tests to run in parallel, and run
1965 C<make test_harness>. On a Bourne-like shell, this can be done as
1967 TEST_JOBS=3 make test_harness # Run 3 tests in parallel
1969 An environment variable is used, rather than parallel make itself, because
1970 L<TAP::Harness> needs to be able to schedule individual non-conflicting test
1971 scripts itself, and there is no standard interface to C<make> utilities to
1972 interact with their job schedulers.
1974 Note that currently some test scripts may fail when run in parallel (most
1975 notably C<ext/IO/t/io_dir.t>). If necessary run just the failing scripts
1976 again sequentially and see if the failures go away.
1977 =item test-notty test_notty
1979 Sets PERL_SKIP_TTY_TEST to true before running normal test.
1983 =head2 Running tests by hand
1985 You can run part of the test suite by hand by using one the following
1986 commands from the F<t/> directory :
1988 ./perl -I../lib TEST list-of-.t-files
1992 ./perl -I../lib harness list-of-.t-files
1994 (if you don't specify test scripts, the whole test suite will be run.)
1996 =head3 Using t/harness for testing
1998 If you use C<harness> for testing you have several command line options
1999 available to you. The arguments are as follows, and are in the order
2000 that they must appear if used together.
2002 harness -v -torture -re=pattern LIST OF FILES TO TEST
2003 harness -v -torture -re LIST OF PATTERNS TO MATCH
2005 If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
2006 the manifest. The file list may include shell wildcards which will be
2013 Run the tests under verbose mode so you can see what tests were run,
2018 Run the torture tests as well as the normal set.
2022 Filter the file list so that all the test files run match PATTERN.
2023 Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
2024 in that it allows the file list to be provided as well.
2026 =item -re LIST OF PATTERNS
2028 Filter the file list so that all the test files run match
2029 /(LIST|OF|PATTERNS)/. Note that with this form the patterns
2030 are joined by '|' and you cannot supply a list of files, instead
2031 the test files are obtained from the MANIFEST.
2035 You can run an individual test by a command similar to
2037 ./perl -I../lib patho/to/foo.t
2039 except that the harnesses set up some environment variables that may
2040 affect the execution of the test :
2046 indicates that we're running this test part of the perl core test suite.
2047 This is useful for modules that have a dual life on CPAN.
2049 =item PERL_DESTRUCT_LEVEL=2
2051 is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2055 (used only by F<t/TEST>) if set, overrides the path to the perl executable
2056 that should be used to run the tests (the default being F<./perl>).
2058 =item PERL_SKIP_TTY_TEST
2060 if set, tells to skip the tests that need a terminal. It's actually set
2061 automatically by the Makefile, but can also be forced artificially by
2062 running 'make test_notty'.
2066 =head3 Other environment variables that may influence tests
2070 =item PERL_TEST_Net_Ping
2072 Setting this variable runs all the Net::Ping modules tests,
2073 otherwise some tests that interact with the outside world are skipped.
2076 =item PERL_TEST_NOVREXX
2078 Setting this variable skips the vrexx.t tests for OS2::REXX.
2080 =item PERL_TEST_NUMCONVERTS
2082 This sets a variable in op/numconvert.t.
2086 See also the documentation for the Test and Test::Harness modules,
2087 for more environment variables that affect testing.
2089 =head2 Common problems when patching Perl source code
2091 Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2092 some cases we have to take pre-ANSI requirements into consideration.
2093 You don't care about some particular platform having broken Perl?
2094 I hear there is still a strong demand for J2EE programmers.
2096 =head2 Perl environment problems
2102 Not compiling with threading
2104 Compiling with threading (-Duseithreads) completely rewrites
2105 the function prototypes of Perl. You better try your changes
2106 with that. Related to this is the difference between "Perl_-less"
2107 and "Perl_-ly" APIs, for example:
2109 Perl_sv_setiv(aTHX_ ...);
2112 The first one explicitly passes in the context, which is needed for e.g.
2113 threaded builds. The second one does that implicitly; do not get them
2114 mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2115 (or a dVAR) as the first thing in the function.
2117 See L<perlguts/"How multiple interpreters and concurrency are supported">
2118 for further discussion about context.
2122 Not compiling with -DDEBUGGING
2124 The DEBUGGING define exposes more code to the compiler,
2125 therefore more ways for things to go wrong. You should try it.
2129 Introducing (non-read-only) globals
2131 Do not introduce any modifiable globals, truly global or file static.
2132 They are bad form and complicate multithreading and other forms of
2133 concurrency. The right way is to introduce them as new interpreter
2134 variables, see F<intrpvar.h> (at the very end for binary compatibility).
2136 Introducing read-only (const) globals is okay, as long as you verify
2137 with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2138 BSD-style output) that the data you added really is read-only.
2139 (If it is, it shouldn't show up in the output of that command.)
2141 If you want to have static strings, make them constant:
2143 static const char etc[] = "...";
2145 If you want to have arrays of constant strings, note carefully
2146 the right combination of C<const>s:
2148 static const char * const yippee[] =
2149 {"hi", "ho", "silver"};
2151 There is a way to completely hide any modifiable globals (they are all
2152 moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2153 It is not normally used, but can be used for testing, read more
2154 about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
2158 Not exporting your new function
2160 Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2161 function that is part of the public API (the shared Perl library)
2162 to be explicitly marked as exported. See the discussion about
2163 F<embed.pl> in L<perlguts>.
2167 Exporting your new function
2169 The new shiny result of either genuine new functionality or your
2170 arduous refactoring is now ready and correctly exported. So what
2171 could possibly go wrong?
2173 Maybe simply that your function did not need to be exported in the
2174 first place. Perl has a long and not so glorious history of exporting
2175 functions that it should not have.
2177 If the function is used only inside one source code file, make it
2178 static. See the discussion about F<embed.pl> in L<perlguts>.
2180 If the function is used across several files, but intended only for
2181 Perl's internal use (and this should be the common case), do not
2182 export it to the public API. See the discussion about F<embed.pl>
2187 =head2 Portability problems
2189 The following are common causes of compilation and/or execution
2190 failures, not common to Perl as such. The C FAQ is good bedtime
2191 reading. Please test your changes with as many C compilers and
2192 platforms as possible; we will, anyway, and it's nice to save
2193 oneself from public embarrassment.
2195 If using gcc, you can add the C<-std=c89> option which will hopefully
2196 catch most of these unportabilities. (However it might also catch
2197 incompatibilities in your system's header files.)
2199 Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2200 C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2202 If using the C<gcc -Wall> note that not all the possible warnings
2203 (like C<-Wunitialized>) are given unless you also compile with C<-O>.
2205 Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2206 code files (the ones at the top level of the source code distribution,
2207 but not e.g. the extensions under ext/) are automatically compiled
2208 with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2209 and a selection of C<-W> flags (see cflags.SH).
2211 Also study L<perlport> carefully to avoid any bad assumptions
2212 about the operating system, filesystems, and so forth.
2214 You may once in a while try a "make microperl" to see whether we
2215 can still compile Perl with just the bare minimum of interfaces.
2218 Do not assume an operating system indicates a certain compiler.
2224 Casting pointers to integers or casting integers to pointers
2226 void castaway(U8* p)
2232 void castaway(U8* p)
2236 Both are bad, and broken, and unportable. Use the PTR2IV()
2237 macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2238 INT2PTR(), and NUM2PTR().)
2242 Casting between data function pointers and data pointers
2244 Technically speaking casting between function pointers and data
2245 pointers is unportable and undefined, but practically speaking
2246 it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
2247 macros. Sometimes you can also play games with unions.
2251 Assuming sizeof(int) == sizeof(long)
2253 There are platforms where longs are 64 bits, and platforms where ints
2254 are 64 bits, and while we are out to shock you, even platforms where
2255 shorts are 64 bits. This is all legal according to the C standard.
2256 (In other words, "long long" is not a portable way to specify 64 bits,
2257 and "long long" is not even guaranteed to be any wider than "long".)
2259 Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2260 Avoid things like I32 because they are B<not> guaranteed to be
2261 I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2262 guaranteed to be B<int> or B<long>. If you really explicitly need
2263 64-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
2267 Assuming one can dereference any type of pointer for any type of data
2270 long pony = *p; /* BAD */
2272 Many platforms, quite rightly so, will give you a core dump instead
2273 of a pony if the p happens not be correctly aligned.
2279 (int)*p = ...; /* BAD */
2281 Simply not portable. Get your lvalue to be of the right type,
2282 or maybe use temporary variables, or dirty tricks with unions.
2286 Assume B<anything> about structs (especially the ones you
2287 don't control, like the ones coming from the system headers)
2293 That a certain field exists in a struct
2297 That no other fields exist besides the ones you know of
2301 That a field is of certain signedness, sizeof, or type
2305 That the fields are in a certain order
2311 While C guarantees the ordering specified in the struct definition,
2312 between different platforms the definitions might differ
2318 That the sizeof(struct) or the alignments are the same everywhere
2324 There might be padding bytes between the fields to align the fields -
2325 the bytes can be anything
2329 Structs are required to be aligned to the maximum alignment required
2330 by the fields - which for native types is for usually equivalent to
2331 sizeof() of the field
2339 Assuming the character set is ASCIIish
2341 Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
2342 This is transparent for the most part, but because the character sets
2343 differ, you shouldn't use numeric (decimal, octal, nor hex) constants
2344 to refer to characters. You can safely say 'A', but not 0x41.
2345 You can safely say '\n', but not \012.
2346 If a character doesn't have a trivial input form, you can
2347 create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that
2348 it resolves to different values depending on the character set being used.
2349 (There are three different EBCDIC character sets defined in C<utfebcdic.h>,
2350 so it might be best to insert the #define three times in that file.)
2352 Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case
2353 alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'.
2354 But '0' - '9' is an unbroken range in both systems. Don't assume anything
2357 Many of the comments in the existing code ignore the possibility of EBCDIC,
2358 and may be wrong therefore, even if the code works.
2359 This is actually a tribute to the successful transparent insertion of being
2360 able to handle EBCDIC without having to change pre-existing code.
2362 UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode
2363 code points as sequences of bytes. Macros
2364 with the same names (but different definitions)
2365 in C<utf8.h> and C<utfebcdic.h>
2366 are used to allow the calling code to think that there is only one such
2368 This is almost always referred to as C<utf8>, but it means the EBCDIC version
2369 as well. Again, comments in the code may well be wrong even if the code itself
2371 For example, the concept of C<invariant characters> differs between ASCII and
2373 On ASCII platforms, only characters that do not have the high-order
2374 bit set (i.e. whose ordinals are strict ASCII, 0 - 127)
2375 are invariant, and the documentation and comments in the code
2377 often referring to something like, say, C<hibit>.
2378 The situation differs and is not so simple on EBCDIC machines, but as long as
2379 the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it
2380 works, even if the comments are wrong.
2384 Assuming the character set is just ASCII
2386 ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
2387 characters have different meanings depending on the locale. Absent a locale,
2388 currently these extra characters are generally considered to be unassigned,
2389 and this has presented some problems.
2390 This is being changed starting in 5.12 so that these characters will
2391 be considered to be Latin-1 (ISO-8859-1).
2395 Mixing #define and #ifdef
2397 #define BURGLE(x) ... \
2398 #ifdef BURGLE_OLD_STYLE /* BAD */
2399 ... do it the old way ... \
2401 ... do it the new way ... \
2404 You cannot portably "stack" cpp directives. For example in the above
2405 you need two separate BURGLE() #defines, one for each #ifdef branch.
2409 Adding non-comment stuff after #endif or #else
2413 #else !SNOSH /* BAD */
2415 #endif SNOSH /* BAD */
2417 The #endif and #else cannot portably have anything non-comment after
2418 them. If you want to document what is going (which is a good idea
2419 especially if the branches are long), use (C) comments:
2427 The gcc option C<-Wendif-labels> warns about the bad variant
2428 (by default on starting from Perl 5.9.4).
2432 Having a comma after the last element of an enum list
2440 is not portable. Leave out the last comma.
2442 Also note that whether enums are implicitly morphable to ints
2443 varies between compilers, you might need to (int).
2449 // This function bamfoodles the zorklator. /* BAD */
2451 That is C99 or C++. Perl is C89. Using the //-comments is silently
2452 allowed by many C compilers but cranking up the ANSI C89 strictness
2453 (which we like to do) causes the compilation to fail.
2457 Mixing declarations and code
2462 set_zorkmids(n); /* BAD */
2465 That is C99 or C++. Some C compilers allow that, but you shouldn't.
2467 The gcc option C<-Wdeclaration-after-statements> scans for such problems
2468 (by default on starting from Perl 5.9.4).
2472 Introducing variables inside for()
2474 for(int i = ...; ...; ...) { /* BAD */
2476 That is C99 or C++. While it would indeed be awfully nice to have that
2477 also in C89, to limit the scope of the loop variable, alas, we cannot.
2481 Mixing signed char pointers with unsigned char pointers
2483 int foo(char *s) { ... }
2485 unsigned char *t = ...; /* Or U8* t = ... */
2488 While this is legal practice, it is certainly dubious, and downright
2489 fatal in at least one platform: for example VMS cc considers this a
2490 fatal error. One cause for people often making this mistake is that a
2491 "naked char" and therefore dereferencing a "naked char pointer" have
2492 an undefined signedness: it depends on the compiler and the flags of
2493 the compiler and the underlying platform whether the result is signed
2494 or unsigned. For this very same reason using a 'char' as an array
2499 Macros that have string constants and their arguments as substrings of
2500 the string constants
2502 #define FOO(n) printf("number = %d\n", n) /* BAD */
2505 Pre-ANSI semantics for that was equivalent to
2507 printf("10umber = %d\10");
2509 which is probably not what you were expecting. Unfortunately at least
2510 one reasonably common and modern C compiler does "real backward
2511 compatibility" here, in AIX that is what still happens even though the
2512 rest of the AIX compiler is very happily C89.
2516 Using printf formats for non-basic C types
2519 printf("i = %d\n", i); /* BAD */
2521 While this might by accident work in some platform (where IV happens
2522 to be an C<int>), in general it cannot. IV might be something larger.
2523 Even worse the situation is with more specific types (defined by Perl's
2524 configuration step in F<config.h>):
2527 printf("who = %d\n", who); /* BAD */
2529 The problem here is that Uid_t might be not only not C<int>-wide
2530 but it might also be unsigned, in which case large uids would be
2531 printed as negative values.
2533 There is no simple solution to this because of printf()'s limited
2534 intelligence, but for many types the right format is available as
2535 with either 'f' or '_f' suffix, for example:
2537 IVdf /* IV in decimal */
2538 UVxf /* UV is hexadecimal */
2540 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2542 Uid_t_f /* Uid_t in decimal */
2544 printf("who = %"Uid_t_f"\n", who);
2546 Or you can try casting to a "wide enough" type:
2548 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2550 Also remember that the C<%p> format really does require a void pointer:
2553 printf("p = %p\n", (void*)p);
2555 The gcc option C<-Wformat> scans for such problems.
2559 Blindly using variadic macros
2561 gcc has had them for a while with its own syntax, and C99 brought
2562 them with a standardized syntax. Don't use the former, and use
2563 the latter only if the HAS_C99_VARIADIC_MACROS is defined.
2567 Blindly passing va_list
2569 Not all platforms support passing va_list to further varargs (stdarg)
2570 functions. The right thing to do is to copy the va_list using the
2571 Perl_va_copy() if the NEED_VA_COPY is defined.
2575 Using gcc statement expressions
2577 val = ({...;...;...}); /* BAD */
2579 While a nice extension, it's not portable. The Perl code does
2580 admittedly use them if available to gain some extra speed
2581 (essentially as a funky form of inlining), but you shouldn't.
2585 Binding together several statements in a macro
2587 Use the macros STMT_START and STMT_END.
2595 Testing for operating systems or versions when should be testing for features
2597 #ifdef __FOONIX__ /* BAD */
2601 Unless you know with 100% certainty that quux() is only ever available
2602 for the "Foonix" operating system B<and> that is available B<and>
2603 correctly working for B<all> past, present, B<and> future versions of
2604 "Foonix", the above is very wrong. This is more correct (though still
2605 not perfect, because the below is a compile-time check):
2611 How does the HAS_QUUX become defined where it needs to be? Well, if
2612 Foonix happens to be Unixy enough to be able to run the Configure
2613 script, and Configure has been taught about detecting and testing
2614 quux(), the HAS_QUUX will be correctly defined. In other platforms,
2615 the corresponding configuration step will hopefully do the same.
2617 In a pinch, if you cannot wait for Configure to be educated,
2618 or if you have a good hunch of where quux() might be available,
2619 you can temporarily try the following:
2621 #if (defined(__FOONIX__) || defined(__BARNIX__))
2631 But in any case, try to keep the features and operating systems separate.
2635 =head2 Problematic System Interfaces
2641 malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
2642 allocate at least one byte. (In general you should rarely need to
2643 work at this low level, but instead use the various malloc wrappers.)
2647 snprintf() - the return type is unportable. Use my_snprintf() instead.
2651 =head2 Security problems
2653 Last but not least, here are various tips for safer coding.
2661 Or we will publicly ridicule you. Seriously.
2665 Do not use strcpy() or strcat() or strncpy() or strncat()
2667 Use my_strlcpy() and my_strlcat() instead: they either use the native
2668 implementation, or Perl's own implementation (borrowed from the public
2669 domain implementation of INN).
2673 Do not use sprintf() or vsprintf()
2675 If you really want just plain byte strings, use my_snprintf()
2676 and my_vsnprintf() instead, which will try to use snprintf() and
2677 vsnprintf() if those safer APIs are available. If you want something
2678 fancier than a plain byte string, use SVs and Perl_sv_catpvf().
2682 =head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2684 Sometimes it helps to use external tools while debugging and
2685 testing Perl. This section tries to guide you through using
2686 some common testing and debugging tools with Perl. This is
2687 meant as a guide to interfacing these tools with Perl, not
2688 as any kind of guide to the use of the tools themselves.
2690 B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2691 Third Degree greatly slows down the execution: seconds become minutes,
2692 minutes become hours. For example as of Perl 5.8.1, the
2693 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2694 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2695 than six hours, even on a snappy computer. The said test must be
2696 doing something that is quite unfriendly for memory debuggers. If you
2697 don't feel like waiting, that you can simply kill away the perl
2700 B<NOTE 2>: To minimize the number of memory leak false alarms (see
2701 L</PERL_DESTRUCT_LEVEL> for more information), you have to set the
2702 environment variable PERL_DESTRUCT_LEVEL to 2.
2704 For csh-like shells:
2706 setenv PERL_DESTRUCT_LEVEL 2
2708 For Bourne-type shells:
2710 PERL_DESTRUCT_LEVEL=2
2711 export PERL_DESTRUCT_LEVEL
2713 In Unixy environments you can also use the C<env> command:
2715 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
2717 B<NOTE 3>: There are known memory leaks when there are compile-time
2718 errors within eval or require, seeing C<S_doeval> in the call stack
2719 is a good sign of these. Fixing these leaks is non-trivial,
2720 unfortunately, but they must be fixed eventually.
2722 B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2723 unless Perl is built with the Configure option
2724 C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2726 =head2 Rational Software's Purify
2728 Purify is a commercial tool that is helpful in identifying
2729 memory overruns, wild pointers, memory leaks and other such
2730 badness. Perl must be compiled in a specific way for
2731 optimal testing with Purify. Purify is available under
2732 Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2734 =head2 Purify on Unix
2736 On Unix, Purify creates a new Perl binary. To get the most
2737 benefit out of Purify, you should create the perl to Purify
2740 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2741 -Uusemymalloc -Dusemultiplicity
2743 where these arguments mean:
2747 =item -Accflags=-DPURIFY
2749 Disables Perl's arena memory allocation functions, as well as
2750 forcing use of memory allocation functions derived from the
2753 =item -Doptimize='-g'
2755 Adds debugging information so that you see the exact source
2756 statements where the problem occurs. Without this flag, all
2757 you will see is the source filename of where the error occurred.
2761 Disable Perl's malloc so that Purify can more closely monitor
2762 allocations and leaks. Using Perl's malloc will make Purify
2763 report most leaks in the "potential" leaks category.
2765 =item -Dusemultiplicity
2767 Enabling the multiplicity option allows perl to clean up
2768 thoroughly when the interpreter shuts down, which reduces the
2769 number of bogus leak reports from Purify.
2773 Once you've compiled a perl suitable for Purify'ing, then you
2778 which creates a binary named 'pureperl' that has been Purify'ed.
2779 This binary is used in place of the standard 'perl' binary
2780 when you want to debug Perl memory problems.
2782 As an example, to show any memory leaks produced during the
2783 standard Perl testset you would create and run the Purify'ed
2788 ../pureperl -I../lib harness
2790 which would run Perl on test.pl and report any memory problems.
2792 Purify outputs messages in "Viewer" windows by default. If
2793 you don't have a windowing environment or if you simply
2794 want the Purify output to unobtrusively go to a log file
2795 instead of to the interactive window, use these following
2796 options to output to the log file "perl.log":
2798 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2799 -log-file=perl.log -append-logfile=yes"
2801 If you plan to use the "Viewer" windows, then you only need this option:
2803 setenv PURIFYOPTIONS "-chain-length=25"
2805 In Bourne-type shells:
2808 export PURIFYOPTIONS
2810 or if you have the "env" utility:
2812 env PURIFYOPTIONS="..." ../pureperl ...
2816 Purify on Windows NT instruments the Perl binary 'perl.exe'
2817 on the fly. There are several options in the makefile you
2818 should change to get the most use out of Purify:
2824 You should add -DPURIFY to the DEFINES line so the DEFINES
2825 line looks something like:
2827 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
2829 to disable Perl's arena memory allocation functions, as
2830 well as to force use of memory allocation functions derived
2831 from the system malloc.
2833 =item USE_MULTI = define
2835 Enabling the multiplicity option allows perl to clean up
2836 thoroughly when the interpreter shuts down, which reduces the
2837 number of bogus leak reports from Purify.
2839 =item #PERL_MALLOC = define
2841 Disable Perl's malloc so that Purify can more closely monitor
2842 allocations and leaks. Using Perl's malloc will make Purify
2843 report most leaks in the "potential" leaks category.
2847 Adds debugging information so that you see the exact source
2848 statements where the problem occurs. Without this flag, all
2849 you will see is the source filename of where the error occurred.
2853 As an example, to show any memory leaks produced during the
2854 standard Perl testset you would create and run Purify as:
2859 purify ../perl -I../lib harness
2861 which would instrument Perl in memory, run Perl on test.pl,
2862 then finally report any memory problems.
2866 The excellent valgrind tool can be used to find out both memory leaks
2867 and illegal memory accesses. As of version 3.3.0, Valgrind only
2868 supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind"
2869 target can be used to run the tests under valgrind. Found errors
2870 and memory leaks are logged in files named F<testfile.valgrind>.
2872 Valgrind also provides a cachegrind tool, invoked on perl as:
2874 VG_OPTS=--tool=cachegrind make test.valgrind
2876 As system libraries (most notably glibc) are also triggering errors,
2877 valgrind allows to suppress such errors using suppression files. The
2878 default suppression file that comes with valgrind already catches a lot
2879 of them. Some additional suppressions are defined in F<t/perl.supp>.
2881 To get valgrind and for more information see
2883 http://developer.kde.org/~sewardj/
2885 =head2 Compaq's/Digital's/HP's Third Degree
2887 Third Degree is a tool for memory leak detection and memory access checks.
2888 It is one of the many tools in the ATOM toolkit. The toolkit is only
2889 available on Tru64 (formerly known as Digital UNIX formerly known as
2892 When building Perl, you must first run Configure with -Doptimize=-g
2893 and -Uusemymalloc flags, after that you can use the make targets
2894 "perl.third" and "test.third". (What is required is that Perl must be
2895 compiled using the C<-g> flag, you may need to re-Configure.)
2897 The short story is that with "atom" you can instrument the Perl
2898 executable to create a new executable called F<perl.third>. When the
2899 instrumented executable is run, it creates a log of dubious memory
2900 traffic in file called F<perl.3log>. See the manual pages of atom and
2901 third for more information. The most extensive Third Degree
2902 documentation is available in the Compaq "Tru64 UNIX Programmer's
2903 Guide", chapter "Debugging Programs with Third Degree".
2905 The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
2906 subdirectory. There is a problem with these files: Third Degree is so
2907 effective that it finds problems also in the system libraries.
2908 Therefore you should used the Porting/thirdclean script to cleanup
2909 the F<*.3log> files.
2911 There are also leaks that for given certain definition of a leak,
2912 aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2914 =head2 PERL_DESTRUCT_LEVEL
2916 If you want to run any of the tests yourself manually using e.g.
2917 valgrind, or the pureperl or perl.third executables, please note that
2918 by default perl B<does not> explicitly cleanup all the memory it has
2919 allocated (such as global memory arenas) but instead lets the exit()
2920 of the whole program "take care" of such allocations, also known as
2921 "global destruction of objects".
2923 There is a way to tell perl to do complete cleanup: set the
2924 environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2925 The t/TEST wrapper does set this to 2, and this is what you
2926 need to do too, if you don't want to see the "global leaks":
2927 For example, for "third-degreed" Perl:
2929 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
2931 (Note: the mod_perl apache module uses also this environment variable
2932 for its own purposes and extended its semantics. Refer to the mod_perl
2933 documentation for more information. Also, spawned threads do the
2934 equivalent of setting this variable to the value 1.)
2936 If, at the end of a run you get the message I<N scalars leaked>, you can
2937 recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2938 of all those leaked SVs to be dumped along with details as to where each
2939 SV was originally allocated. This information is also displayed by
2940 Devel::Peek. Note that the extra details recorded with each SV increases
2941 memory usage, so it shouldn't be used in production environments. It also
2942 converts C<new_SV()> from a macro into a real function, so you can use
2943 your favourite debugger to discover where those pesky SVs were allocated.
2945 If you see that you're leaking memory at runtime, but neither valgrind
2946 nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
2947 leaking SVs that are still reachable and will be properly cleaned up
2948 during destruction of the interpreter. In such cases, using the C<-Dm>
2949 switch can point you to the source of the leak. If the executable was
2950 built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations
2951 in addition to memory allocations. Each SV allocation has a distinct
2952 serial number that will be written on creation and destruction of the SV.
2953 So if you're executing the leaking code in a loop, you need to look for
2954 SVs that are created, but never destroyed between each cycle. If such an
2955 SV is found, set a conditional breakpoint within C<new_SV()> and make it
2956 break only when C<PL_sv_serial> is equal to the serial number of the
2957 leaking SV. Then you will catch the interpreter in exactly the state
2958 where the leaking SV is allocated, which is sufficient in many cases to
2959 find the source of the leak.
2961 As C<-Dm> is using the PerlIO layer for output, it will by itself
2962 allocate quite a bunch of SVs, which are hidden to avoid recursion.
2963 You can bypass the PerlIO layer if you use the SV logging provided
2964 by C<-DPERL_MEM_LOG> instead.
2968 If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go
2969 through logging functions, which is handy for breakpoint setting.
2971 Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging
2972 functions read $ENV{PERL_MEM_LOG} to determine whether to log the
2973 event, and if so how:
2975 $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops
2976 $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops
2977 $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log
2978 $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2)
2980 Memory logging is somewhat similar to C<-Dm> but is independent of
2981 C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(),
2982 and Safefree() are logged with the caller's source code file and line
2983 number (and C function name, if supported by the C compiler). In
2984 contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging
2987 Since the logging doesn't use PerlIO, all SV allocations are logged
2988 and no extra SV allocations are introduced by enabling the logging.
2989 If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for
2990 each SV allocation is also logged.
2994 Depending on your platform there are various of profiling Perl.
2996 There are two commonly used techniques of profiling executables:
2997 I<statistical time-sampling> and I<basic-block counting>.
2999 The first method takes periodically samples of the CPU program
3000 counter, and since the program counter can be correlated with the code
3001 generated for functions, we get a statistical view of in which
3002 functions the program is spending its time. The caveats are that very
3003 small/fast functions have lower probability of showing up in the
3004 profile, and that periodically interrupting the program (this is
3005 usually done rather frequently, in the scale of milliseconds) imposes
3006 an additional overhead that may skew the results. The first problem
3007 can be alleviated by running the code for longer (in general this is a
3008 good idea for profiling), the second problem is usually kept in guard
3009 by the profiling tools themselves.
3011 The second method divides up the generated code into I<basic blocks>.
3012 Basic blocks are sections of code that are entered only in the
3013 beginning and exited only at the end. For example, a conditional jump
3014 starts a basic block. Basic block profiling usually works by
3015 I<instrumenting> the code by adding I<enter basic block #nnnn>
3016 book-keeping code to the generated code. During the execution of the
3017 code the basic block counters are then updated appropriately. The
3018 caveat is that the added extra code can skew the results: again, the
3019 profiling tools usually try to factor their own effects out of the
3022 =head2 Gprof Profiling
3024 gprof is a profiling tool available in many Unix platforms,
3025 it uses F<statistical time-sampling>.
3027 You can build a profiled version of perl called "perl.gprof" by
3028 invoking the make target "perl.gprof" (What is required is that Perl
3029 must be compiled using the C<-pg> flag, you may need to re-Configure).
3030 Running the profiled version of Perl will create an output file called
3031 F<gmon.out> is created which contains the profiling data collected
3032 during the execution.
3034 The gprof tool can then display the collected data in various ways.
3035 Usually gprof understands the following options:
3041 Suppress statically defined functions from the profile.
3045 Suppress the verbose descriptions in the profile.
3049 Exclude the given routine and its descendants from the profile.
3053 Display only the given routine and its descendants in the profile.
3057 Generate a summary file called F<gmon.sum> which then may be given
3058 to subsequent gprof runs to accumulate data over several runs.
3062 Display routines that have zero usage.
3066 For more detailed explanation of the available commands and output
3067 formats, see your own local documentation of gprof.
3071 $ sh Configure -des -Dusedevel -Doptimize='-pg' && make perl.gprof
3072 $ ./perl.gprof someprog # creates gmon.out in current directory
3073 $ gprof ./perl.gprof > out
3076 =head2 GCC gcov Profiling
3078 Starting from GCC 3.0 I<basic block profiling> is officially available
3081 You can build a profiled version of perl called F<perl.gcov> by
3082 invoking the make target "perl.gcov" (what is required that Perl must
3083 be compiled using gcc with the flags C<-fprofile-arcs
3084 -ftest-coverage>, you may need to re-Configure).
3086 Running the profiled version of Perl will cause profile output to be
3087 generated. For each source file an accompanying ".da" file will be
3090 To display the results you use the "gcov" utility (which should
3091 be installed if you have gcc 3.0 or newer installed). F<gcov> is
3092 run on source code files, like this
3096 which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3097 contain the source code annotated with relative frequencies of
3098 execution indicated by "#" markers.
3100 Useful options of F<gcov> include C<-b> which will summarise the
3101 basic block, branch, and function call coverage, and C<-c> which
3102 instead of relative frequencies will use the actual counts. For
3103 more information on the use of F<gcov> and basic block profiling
3104 with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3106 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3108 and its section titled "8. gcov: a Test Coverage Program"
3110 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3114 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3115 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3116 $ rm -f regexec.c.gcov regexec.gcda
3119 $ view regexec.c.gcov
3121 =head2 Pixie Profiling
3123 Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3124 UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
3125 I<basic-block counting>.
3127 You can build a profiled version of perl called F<perl.pixie> by
3128 invoking the make target "perl.pixie" (what is required is that Perl
3129 must be compiled using the C<-g> flag, you may need to re-Configure).
3131 In Tru64 a file called F<perl.Addrs> will also be silently created,
3132 this file contains the addresses of the basic blocks. Running the
3133 profiled version of Perl will create a new file called "perl.Counts"
3134 which contains the counts for the basic block for that particular
3137 To display the results you use the F<prof> utility. The exact
3138 incantation depends on your operating system, "prof perl.Counts" in
3139 IRIX, and "prof -pixie -all -L. perl" in Tru64.
3141 In IRIX the following prof options are available:
3147 Reports the most heavily used lines in descending order of use.
3148 Useful for finding the hotspot lines.
3152 Groups lines by procedure, with procedures sorted in descending order of use.
3153 Within a procedure, lines are listed in source order.
3154 Useful for finding the hotspots of procedures.
3158 In Tru64 the following options are available:
3164 Procedures sorted in descending order by the number of cycles executed
3165 in each procedure. Useful for finding the hotspot procedures.
3166 (This is the default option.)
3170 Lines sorted in descending order by the number of cycles executed in
3171 each line. Useful for finding the hotspot lines.
3173 =item -i[nvocations]
3175 The called procedures are sorted in descending order by number of calls
3176 made to the procedures. Useful for finding the most used procedures.
3180 Grouped by procedure, sorted by cycles executed per procedure.
3181 Useful for finding the hotspots of procedures.
3185 The compiler emitted code for these lines, but the code was unexecuted.
3189 Unexecuted procedures.
3193 For further information, see your system's manual pages for pixie and prof.
3195 =head2 Miscellaneous tricks
3201 Those debugging perl with the DDD frontend over gdb may find the
3204 You can extend the data conversion shortcuts menu, so for example you
3205 can display an SV's IV value with one click, without doing any typing.
3206 To do that simply edit ~/.ddd/init file and add after:
3208 ! Display shortcuts.
3209 Ddd*gdbDisplayShortcuts: \
3210 /t () // Convert to Bin\n\
3211 /d () // Convert to Dec\n\
3212 /x () // Convert to Hex\n\
3213 /o () // Convert to Oct(\n\
3215 the following two lines:
3217 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3218 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3220 so now you can do ivx and pvx lookups or you can plug there the
3221 sv_peek "conversion":
3223 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3225 (The my_perl is for threaded builds.)
3226 Just remember that every line, but the last one, should end with \n\
3228 Alternatively edit the init file interactively via:
3229 3rd mouse button -> New Display -> Edit Menu
3231 Note: you can define up to 20 conversion shortcuts in the gdb
3236 If you see in a debugger a memory area mysteriously full of 0xABABABAB
3237 or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3242 Under ithreads the optree is read only. If you want to enforce this, to check
3243 for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3244 enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3245 that allocates op memory via C<mmap>, and sets it read-only at run time.
3246 Any write access to an op results in a C<SIGBUS> and abort.
3248 This code is intended for development only, and may not be portable even to
3249 all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3250 all ops read only. Specifically it
3256 Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3257 later via C<require> or C<eval> will be re-write
3261 Turns an entire slab of ops read-write if the refcount of any op in the slab
3262 needs to be decreased.
3266 Turns an entire slab of ops read-write if any op from the slab is freed.
3270 It's not possible to turn the slabs to read-only after an action requiring
3271 read-write access, as either can happen during op tree building time, so
3272 there may still be legitimate write access.
3274 However, as an 80% solution it is still effective, as currently it catches
3275 a write access during the generation of F<Config.pm>, which means that we
3276 can't yet build F<perl> with this enabled.
3283 We've had a brief look around the Perl source, how to maintain quality
3284 of the source code, an overview of the stages F<perl> goes through
3285 when it's running your code, how to use debuggers to poke at the Perl
3286 guts, and finally how to analyse the execution of Perl. We took a very
3287 simple problem and demonstrated how to solve it fully - with
3288 documentation, regression tests, and finally a patch for submission to
3289 p5p. Finally, we talked about how to use external tools to debug and
3292 I'd now suggest you read over those references again, and then, as soon
3293 as possible, get your hands dirty. The best way to learn is by doing,
3300 Subscribe to perl5-porters, follow the patches and try and understand
3301 them; don't be afraid to ask if there's a portion you're not clear on -
3302 who knows, you may unearth a bug in the patch...
3306 Keep up to date with the bleeding edge Perl distributions and get
3307 familiar with the changes. Try and get an idea of what areas people are
3308 working on and the changes they're making.
3312 Do read the README associated with your operating system, e.g. README.aix
3313 on the IBM AIX OS. Don't hesitate to supply patches to that README if
3314 you find anything missing or changed over a new OS release.
3318 Find an area of Perl that seems interesting to you, and see if you can
3319 work out how it works. Scan through the source, and step over it in the
3320 debugger. Play, poke, investigate, fiddle! You'll probably get to
3321 understand not just your chosen area but a much wider range of F<perl>'s
3322 activity as well, and probably sooner than you'd think.
3328 =item I<The Road goes ever on and on, down from the door where it began.>
3332 If you can do these things, you've started on the long road to Perl porting.
3333 Thanks for wanting to help make Perl better - and happy hacking!
3335 =head2 Metaphoric Quotations
3337 If you recognized the quote about the Road above, you're in luck.
3339 Most software projects begin each file with a literal description of each
3340 file's purpose. Perl instead begins each with a literary allusion to that
3343 Like chapters in many books, all top-level Perl source files (along with a
3344 few others here and there) begin with an epigramic inscription that alludes,
3345 indirectly and metaphorically, to the material you're about to read.
3347 Quotations are taken from writings of J.R.R Tolkien pertaining to his
3348 Legendarium, almost always from I<The Lord of the Rings>. Chapters and
3349 page numbers are given using the following editions:
3355 I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary
3356 edition of 2007 was used, published in the UK by Harper Collins Publishers
3357 and in the US by the Houghton Mifflin Company.
3361 I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover,
3362 50th-anniversary edition of 2004 was used, published in the UK by Harper
3363 Collins Publishers and in the US by the Houghton Mifflin Company.
3367 I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his
3368 son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes
3369 in Christopher's mammoth I<History of Middle Earth>. Page numbers derive
3370 from the hardcover edition, first published in 1983 by George Allen &
3371 Unwin; no page numbers changed for the special 3-volume omnibus edition of
3372 2002 or the various trade-paper editions, all again now by Harper Collins
3373 or Houghton Mifflin.
3377 Other JRRT books fair game for quotes would thus include I<The Adventures of
3378 Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of
3379 the Children of Hurin>, all but the first posthumously assembled by CJRT.
3380 But I<The Lord of the Rings> itself is perfectly fine and probably best to
3381 quote from, provided you can find a suitable quote there.
3383 So if you were to supply a new, complete, top-level source file to add to
3384 Perl, you should conform to this peculiar practice by yourself selecting an
3385 appropriate quotation from Tolkien, retaining the original spelling and
3386 punctuation and using the same format the rest of the quotes are in.
3387 Indirect and oblique is just fine; remember, it's a metaphor, so being meta
3388 is, after all, what it's for.
3392 This document was written by Nathan Torkington, and is maintained by
3393 the perl5-porters mailing list.