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/APC/perl-current .
220 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.10.x .
221 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.8.x .
222 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.6.x .
223 $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.005xx .
225 (Add the C<--delete> option to remove leftover files)
227 You may also want to subscribe to the perl5-changes mailing list to
228 receive a copy of each patch that gets submitted to the maintenance
229 and development "branches" of the perl repository. See
230 http://lists.perl.org/ for subscription information.
232 If you are a member of the perl5-porters mailing list, it is a good
233 thing to keep in touch with the most recent changes. If not only to
234 verify if what you would have posted as a bug report isn't already
235 solved in the most recent available perl development branch, also
236 known as perl-current, bleading edge perl, bleedperl or bleadperl.
238 Needless to say, the source code in perl-current is usually in a perpetual
239 state of evolution. You should expect it to be very buggy. Do B<not> use
240 it for any purpose other than testing and development.
242 =head2 Perlbug administration
244 There is a single remote administrative interface for modifying bug status,
245 category, open issues etc. using the B<RT> bugtracker system, maintained
246 by Robert Spier. Become an administrator, and close any bugs you can get
247 your sticky mitts on:
249 http://bugs.perl.org/
251 To email the bug system administrators:
253 "perlbug-admin" <perlbug-admin@perl.org>
255 =head2 Submitting patches
257 Always submit patches to I<perl5-porters@perl.org>. If you're
258 patching a core module and there's an author listed, send the author a
259 copy (see L<Patching a core module>). This lets other porters review
260 your patch, which catches a surprising number of errors in patches.
261 Please patch against the latest B<development> version. (e.g., even if
262 you're fixing a bug in the 5.8 track, patch against the C<blead> branch in
265 If changes are accepted, they are applied to the development branch. Then
266 the maintenance pumpking decides which of those patches is to be
267 backported to the maint branch. Only patches that survive the heat of the
268 development branch get applied to maintenance versions.
270 Your patch should update the documentation and test suite. See
271 L<Writing a test>. If you have added or removed files in the distribution,
272 edit the MANIFEST file accordingly, sort the MANIFEST file using
273 C<make manisort>, and include those changes as part of your patch.
275 Patching documentation also follows the same order: if accepted, a patch
276 is first applied to B<development>, and if relevant then it's backported
277 to B<maintenance>. (With an exception for some patches that document
278 behaviour that only appears in the maintenance branch, but which has
279 changed in the development version.)
281 To report a bug in Perl, use the program I<perlbug> which comes with
282 Perl (if you can't get Perl to work, send mail to the address
283 I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
284 I<perlbug> feeds into the automated bug-tracking system, access to
285 which is provided through the web at http://rt.perl.org/rt3/ . It
286 often pays to check the archives of the perl5-porters mailing list to
287 see whether the bug you're reporting has been reported before, and if
288 so whether it was considered a bug. See above for the location of
289 the searchable archives.
291 The CPAN testers ( http://testers.cpan.org/ ) are a group of
292 volunteers who test CPAN modules on a variety of platforms. Perl
293 Smokers ( http://www.nntp.perl.org/group/perl.daily-build and
294 http://www.nntp.perl.org/group/perl.daily-build.reports/ )
295 automatically test Perl source releases on platforms with various
296 configurations. Both efforts welcome volunteers. In order to get
297 involved in smoke testing of the perl itself visit
298 L<http://search.cpan.org/dist/Test-Smoke>. In order to start smoke
299 testing CPAN modules visit L<http://search.cpan.org/dist/CPAN-YACSmoke/>
300 or L<http://search.cpan.org/dist/POE-Component-CPAN-YACSmoke/> or
301 L<http://search.cpan.org/dist/CPAN-Reporter/>.
303 It's a good idea to read and lurk for a while before chipping in.
304 That way you'll get to see the dynamic of the conversations, learn the
305 personalities of the players, and hopefully be better prepared to make
306 a useful contribution when do you speak up.
308 If after all this you still think you want to join the perl5-porters
309 mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
310 unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
312 To hack on the Perl guts, you'll need to read the following things:
318 This is of paramount importance, since it's the documentation of what
319 goes where in the Perl source. Read it over a couple of times and it
320 might start to make sense - don't worry if it doesn't yet, because the
321 best way to study it is to read it in conjunction with poking at Perl
322 source, and we'll do that later on.
324 You might also want to look at Gisle Aas's illustrated perlguts -
325 there's no guarantee that this will be absolutely up-to-date with the
326 latest documentation in the Perl core, but the fundamentals will be
327 right. ( http://gisle.aas.no/perl/illguts/ )
329 =item L<perlxstut> and L<perlxs>
331 A working knowledge of XSUB programming is incredibly useful for core
332 hacking; XSUBs use techniques drawn from the PP code, the portion of the
333 guts that actually executes a Perl program. It's a lot gentler to learn
334 those techniques from simple examples and explanation than from the core
339 The documentation for the Perl API explains what some of the internal
340 functions do, as well as the many macros used in the source.
342 =item F<Porting/pumpkin.pod>
344 This is a collection of words of wisdom for a Perl porter; some of it is
345 only useful to the pumpkin holder, but most of it applies to anyone
346 wanting to go about Perl development.
348 =item The perl5-porters FAQ
350 This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html .
351 It contains hints on reading perl5-porters, information on how
352 perl5-porters works and how Perl development in general works.
356 =head2 Finding Your Way Around
358 Perl maintenance can be split into a number of areas, and certain people
359 (pumpkins) will have responsibility for each area. These areas sometimes
360 correspond to files or directories in the source kit. Among the areas are:
366 Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
367 subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
368 contains the core XS modules.
372 There are tests for nearly all the modules, built-ins and major bits
373 of functionality. Test files all have a .t suffix. Module tests live
374 in the F<lib/> and F<ext/> directories next to the module being
375 tested. Others live in F<t/>. See L<Writing a test>
379 Documentation maintenance includes looking after everything in the
380 F<pod/> directory, (as well as contributing new documentation) and
381 the documentation to the modules in core.
385 The configure process is the way we make Perl portable across the
386 myriad of operating systems it supports. Responsibility for the
387 configure, build and installation process, as well as the overall
388 portability of the core code rests with the configure pumpkin - others
389 help out with individual operating systems.
391 The files involved are the operating system directories, (F<win32/>,
392 F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
393 and F<Makefile>, as well as the metaconfig files which generate
394 F<Configure>. (metaconfig isn't included in the core distribution.)
398 And of course, there's the core of the Perl interpreter itself. Let's
399 have a look at that in a little more detail.
403 Before we leave looking at the layout, though, don't forget that
404 F<MANIFEST> contains not only the file names in the Perl distribution,
405 but short descriptions of what's in them, too. For an overview of the
406 important files, try this:
408 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
410 =head2 Elements of the interpreter
412 The work of the interpreter has two main stages: compiling the code
413 into the internal representation, or bytecode, and then executing it.
414 L<perlguts/Compiled code> explains exactly how the compilation stage
417 Here is a short breakdown of perl's operation:
423 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
424 This is very high-level code, enough to fit on a single screen, and it
425 resembles the code found in L<perlembed>; most of the real action takes
428 F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at
429 make time, so you should make perl to follow this along.
431 First, F<perlmain.c> allocates some memory and constructs a Perl
432 interpreter, along these lines:
434 1 PERL_SYS_INIT3(&argc,&argv,&env);
436 3 if (!PL_do_undump) {
437 4 my_perl = perl_alloc();
440 7 perl_construct(my_perl);
441 8 PL_perl_destruct_level = 0;
444 Line 1 is a macro, and its definition is dependent on your operating
445 system. Line 3 references C<PL_do_undump>, a global variable - all
446 global variables in Perl start with C<PL_>. This tells you whether the
447 current running program was created with the C<-u> flag to perl and then
448 F<undump>, which means it's going to be false in any sane context.
450 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
451 interpreter. It's quite a simple function, and the guts of it looks like
454 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
456 Here you see an example of Perl's system abstraction, which we'll see
457 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
458 own C<malloc> as defined in F<malloc.c> if you selected that option at
461 Next, in line 7, we construct the interpreter using perl_construct,
462 also in F<perl.c>; this sets up all the special variables that Perl
463 needs, the stacks, and so on.
465 Now we pass Perl the command line options, and tell it to go:
467 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
471 exitstatus = perl_destruct(my_perl);
475 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
476 in F<perl.c>, which processes the command line options, sets up any
477 statically linked XS modules, opens the program and calls C<yyparse> to
482 The aim of this stage is to take the Perl source, and turn it into an op
483 tree. We'll see what one of those looks like later. Strictly speaking,
484 there's three things going on here.
486 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
487 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
488 B<is> a YACC grammar for Perl!) The job of the parser is to take your
489 code and "understand" it, splitting it into sentences, deciding which
490 operands go with which operators and so on.
492 The parser is nobly assisted by the lexer, which chunks up your input
493 into tokens, and decides what type of thing each token is: a variable
494 name, an operator, a bareword, a subroutine, a core function, and so on.
495 The main point of entry to the lexer is C<yylex>, and that and its
496 associated routines can be found in F<toke.c>. Perl isn't much like
497 other computer languages; it's highly context sensitive at times, it can
498 be tricky to work out what sort of token something is, or where a token
499 ends. As such, there's a lot of interplay between the tokeniser and the
500 parser, which can get pretty frightening if you're not used to it.
502 As the parser understands a Perl program, it builds up a tree of
503 operations for the interpreter to perform during execution. The routines
504 which construct and link together the various operations are to be found
505 in F<op.c>, and will be examined later.
509 Now the parsing stage is complete, and the finished tree represents
510 the operations that the Perl interpreter needs to perform to execute our
511 program. Next, Perl does a dry run over the tree looking for
512 optimisations: constant expressions such as C<3 + 4> will be computed
513 now, and the optimizer will also see if any multiple operations can be
514 replaced with a single one. For instance, to fetch the variable C<$foo>,
515 instead of grabbing the glob C<*foo> and looking at the scalar
516 component, the optimizer fiddles the op tree to use a function which
517 directly looks up the scalar in question. The main optimizer is C<peep>
518 in F<op.c>, and many ops have their own optimizing functions.
522 Now we're finally ready to go: we have compiled Perl byte code, and all
523 that's left to do is run it. The actual execution is done by the
524 C<runops_standard> function in F<run.c>; more specifically, it's done by
525 these three innocent looking lines:
527 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
531 You may be more comfortable with the Perl version of that:
533 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
535 Well, maybe not. Anyway, each op contains a function pointer, which
536 stipulates the function which will actually carry out the operation.
537 This function will return the next op in the sequence - this allows for
538 things like C<if> which choose the next op dynamically at run time.
539 The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
540 execution if required.
542 The actual functions called are known as PP code, and they're spread
543 between four files: F<pp_hot.c> contains the "hot" code, which is most
544 often used and highly optimized, F<pp_sys.c> contains all the
545 system-specific functions, F<pp_ctl.c> contains the functions which
546 implement control structures (C<if>, C<while> and the like) and F<pp.c>
547 contains everything else. These are, if you like, the C code for Perl's
548 built-in functions and operators.
550 Note that each C<pp_> function is expected to return a pointer to the next
551 op. Calls to perl subs (and eval blocks) are handled within the same
552 runops loop, and do not consume extra space on the C stack. For example,
553 C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block
554 struct onto the context stack which contain the address of the op
555 following the sub call or eval. They then return the first op of that sub
556 or eval block, and so execution continues of that sub or block. Later, a
557 C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>,
558 retrieves the return op from it, and returns it.
560 =item Exception handing
562 Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level
563 C<setjmp()>/C<longjmp()> C-library functions. These basically provide a
564 way to capture the current PC and SP registers and later restore them; i.e.
565 a C<longjmp()> continues at the point in code where a previous C<setjmp()>
566 was done, with anything further up on the C stack being lost. This is why
567 code should always save values using C<SAVE_FOO> rather than in auto
570 The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
571 C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
572 C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
573 C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
575 At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
576 C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
577 loop or whatever, and handle possible exception returns. For a 2 return,
578 final cleanup is performed, such as popping stacks and calling C<CHECK> or
579 C<END> blocks. Amongst other things, this is how scope cleanup still
580 occurs during an C<exit>.
582 If a C<die> can find a C<CxEVAL> block on the context stack, then the
583 stack is popped to that level and the return op in that block is assigned
584 to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally
585 passes control back to the guard. In the case of C<perl_run> and
586 C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops
587 loop. The is the normal way that C<die> or C<croak> is handled within an
590 Sometimes ops are executed within an inner runops loop, such as tie, sort
591 or overload code. In this case, something like
593 sub FETCH { eval { die } }
595 would cause a longjmp right back to the guard in C<perl_run>, popping both
596 runops loops, which is clearly incorrect. One way to avoid this is for the
597 tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner
598 runops loop, but for efficiency reasons, perl in fact just sets a flag,
599 using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and
600 C<pp_entertry> ops check this flag, and if true, they call C<docatch>,
601 which does a C<JMPENV_PUSH> and starts a new runops level to execute the
602 code, rather than doing it on the current loop.
604 As a further optimisation, on exit from the eval block in the C<FETCH>,
605 execution of the code following the block is still carried on in the inner
606 loop. When an exception is raised, C<docatch> compares the C<JMPENV>
607 level of the C<CxEVAL> with C<PL_top_env> and if they differ, just
608 re-throws the exception. In this way any inner loops get popped.
612 1: eval { tie @a, 'A' };
618 To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then
619 enters a runops loop. This loop executes the eval and tie ops on line 1,
620 with the eval pushing a C<CxEVAL> onto the context stack.
622 The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop
623 to execute the body of C<TIEARRAY>. When it executes the entertry op on
624 line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which
625 does a C<JMPENV_PUSH> and starts a third runops loop, which then executes
626 the die op. At this point the C call stack looks like this:
629 Perl_runops # third loop
633 Perl_runops # second loop
637 Perl_runops # first loop
642 and the context and data stacks, as shown by C<-Dstv>, look like:
646 CX 1: EVAL => AV() PV("A"\0)
654 The die pops the first C<CxEVAL> off the context stack, sets
655 C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to
656 the top C<docatch>. This then starts another third-level runops level,
657 which executes the nextstate, pushmark and die ops on line 4. At the point
658 that the second C<pp_die> is called, the C call stack looks exactly like
659 that above, even though we are no longer within an inner eval; this is
660 because of the optimization mentioned earlier. However, the context stack
661 now looks like this, ie with the top CxEVAL popped:
665 CX 1: EVAL => AV() PV("A"\0)
671 The die on line 4 pops the context stack back down to the CxEVAL, leaving
677 As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
678 C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
682 Perl_runops # second loop
686 Perl_runops # first loop
691 In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
692 differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
693 and the C stack unwinds to:
698 Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop
699 and execution continues.
703 =head2 Internal Variable Types
705 You should by now have had a look at L<perlguts>, which tells you about
706 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
709 These variables are used not only to represent Perl-space variables, but
710 also any constants in the code, as well as some structures completely
711 internal to Perl. The symbol table, for instance, is an ordinary Perl
712 hash. Your code is represented by an SV as it's read into the parser;
713 any program files you call are opened via ordinary Perl filehandles, and
716 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
717 Perl program. Let's see, for instance, how Perl treats the constant
720 % perl -MDevel::Peek -e 'Dump("hello")'
721 1 SV = PV(0xa041450) at 0xa04ecbc
723 3 FLAGS = (POK,READONLY,pPOK)
724 4 PV = 0xa0484e0 "hello"\0
728 Reading C<Devel::Peek> output takes a bit of practise, so let's go
729 through it line by line.
731 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
732 memory. SVs themselves are very simple structures, but they contain a
733 pointer to a more complex structure. In this case, it's a PV, a
734 structure which holds a string value, at location C<0xa041450>. Line 2
735 is the reference count; there are no other references to this data, so
738 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
739 read-only SV (because it's a constant) and the data is a PV internally.
740 Next we've got the contents of the string, starting at location
743 Line 5 gives us the current length of the string - note that this does
744 B<not> include the null terminator. Line 6 is not the length of the
745 string, but the length of the currently allocated buffer; as the string
746 grows, Perl automatically extends the available storage via a routine
749 You can get at any of these quantities from C very easily; just add
750 C<Sv> to the name of the field shown in the snippet, and you've got a
751 macro which will return the value: C<SvCUR(sv)> returns the current
752 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
753 C<SvPV(sv, len)> returns the string itself with its length, and so on.
754 More macros to manipulate these properties can be found in L<perlguts>.
756 Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
759 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
764 6 junk = SvPV_force(sv, tlen);
765 7 SvGROW(sv, tlen + len + 1);
768 10 Move(ptr,SvPVX(sv)+tlen,len,char);
770 12 *SvEND(sv) = '\0';
771 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
775 This is a function which adds a string, C<ptr>, of length C<len> onto
776 the end of the PV stored in C<sv>. The first thing we do in line 6 is
777 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
778 macro to force a PV. As a side effect, C<tlen> gets set to the current
779 value of the PV, and the PV itself is returned to C<junk>.
781 In line 7, we make sure that the SV will have enough room to accommodate
782 the old string, the new string and the null terminator. If C<LEN> isn't
783 big enough, C<SvGROW> will reallocate space for us.
785 Now, if C<junk> is the same as the string we're trying to add, we can
786 grab the string directly from the SV; C<SvPVX> is the address of the PV
789 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
790 memory around: we move the string C<ptr> to the end of the PV - that's
791 the start of the PV plus its current length. We're moving C<len> bytes
792 of type C<char>. After doing so, we need to tell Perl we've extended the
793 string, by altering C<CUR> to reflect the new length. C<SvEND> is a
794 macro which gives us the end of the string, so that needs to be a
797 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
798 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
799 want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
800 version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
801 and turns on POK. The final C<SvTAINT> is a macro which launders tainted
802 data if taint mode is turned on.
804 AVs and HVs are more complicated, but SVs are by far the most common
805 variable type being thrown around. Having seen something of how we
806 manipulate these, let's go on and look at how the op tree is
811 First, what is the op tree, anyway? The op tree is the parsed
812 representation of your program, as we saw in our section on parsing, and
813 it's the sequence of operations that Perl goes through to execute your
814 program, as we saw in L</Running>.
816 An op is a fundamental operation that Perl can perform: all the built-in
817 functions and operators are ops, and there are a series of ops which
818 deal with concepts the interpreter needs internally - entering and
819 leaving a block, ending a statement, fetching a variable, and so on.
821 The op tree is connected in two ways: you can imagine that there are two
822 "routes" through it, two orders in which you can traverse the tree.
823 First, parse order reflects how the parser understood the code, and
824 secondly, execution order tells perl what order to perform the
827 The easiest way to examine the op tree is to stop Perl after it has
828 finished parsing, and get it to dump out the tree. This is exactly what
829 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
830 and L<B::Debug|B::Debug> do.
832 Let's have a look at how Perl sees C<$a = $b + $c>:
834 % perl -MO=Terse -e '$a=$b+$c'
835 1 LISTOP (0x8179888) leave
836 2 OP (0x81798b0) enter
837 3 COP (0x8179850) nextstate
838 4 BINOP (0x8179828) sassign
839 5 BINOP (0x8179800) add [1]
840 6 UNOP (0x81796e0) null [15]
841 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
842 8 UNOP (0x81797e0) null [15]
843 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
844 10 UNOP (0x816b4f0) null [15]
845 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
847 Let's start in the middle, at line 4. This is a BINOP, a binary
848 operator, which is at location C<0x8179828>. The specific operator in
849 question is C<sassign> - scalar assignment - and you can find the code
850 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
851 binary operator, it has two children: the add operator, providing the
852 result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
855 Line 10 is the null op: this does exactly nothing. What is that doing
856 there? If you see the null op, it's a sign that something has been
857 optimized away after parsing. As we mentioned in L</Optimization>,
858 the optimization stage sometimes converts two operations into one, for
859 example when fetching a scalar variable. When this happens, instead of
860 rewriting the op tree and cleaning up the dangling pointers, it's easier
861 just to replace the redundant operation with the null op. Originally,
862 the tree would have looked like this:
864 10 SVOP (0x816b4f0) rv2sv [15]
865 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
867 That is, fetch the C<a> entry from the main symbol table, and then look
868 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
869 happens to do both these things.
871 The right hand side, starting at line 5 is similar to what we've just
872 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
875 Now, what's this about?
877 1 LISTOP (0x8179888) leave
878 2 OP (0x81798b0) enter
879 3 COP (0x8179850) nextstate
881 C<enter> and C<leave> are scoping ops, and their job is to perform any
882 housekeeping every time you enter and leave a block: lexical variables
883 are tidied up, unreferenced variables are destroyed, and so on. Every
884 program will have those first three lines: C<leave> is a list, and its
885 children are all the statements in the block. Statements are delimited
886 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
887 the ops to be performed for each statement being the children of
888 C<nextstate>. C<enter> is a single op which functions as a marker.
890 That's how Perl parsed the program, from top to bottom:
903 However, it's impossible to B<perform> the operations in this order:
904 you have to find the values of C<$b> and C<$c> before you add them
905 together, for instance. So, the other thread that runs through the op
906 tree is the execution order: each op has a field C<op_next> which points
907 to the next op to be run, so following these pointers tells us how perl
908 executes the code. We can traverse the tree in this order using
909 the C<exec> option to C<B::Terse>:
911 % perl -MO=Terse,exec -e '$a=$b+$c'
912 1 OP (0x8179928) enter
913 2 COP (0x81798c8) nextstate
914 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
915 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
916 5 BINOP (0x8179878) add [1]
917 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
918 7 BINOP (0x81798a0) sassign
919 8 LISTOP (0x8179900) leave
921 This probably makes more sense for a human: enter a block, start a
922 statement. Get the values of C<$b> and C<$c>, and add them together.
923 Find C<$a>, and assign one to the other. Then leave.
925 The way Perl builds up these op trees in the parsing process can be
926 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
927 piece we need to construct the tree for C<$a = $b + $c>
929 1 term : term ASSIGNOP term
930 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
932 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
934 If you're not used to reading BNF grammars, this is how it works: You're
935 fed certain things by the tokeniser, which generally end up in upper
936 case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
937 code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
938 "terminal symbols", because you can't get any simpler than them.
940 The grammar, lines one and three of the snippet above, tells you how to
941 build up more complex forms. These complex forms, "non-terminal symbols"
942 are generally placed in lower case. C<term> here is a non-terminal
943 symbol, representing a single expression.
945 The grammar gives you the following rule: you can make the thing on the
946 left of the colon if you see all the things on the right in sequence.
947 This is called a "reduction", and the aim of parsing is to completely
948 reduce the input. There are several different ways you can perform a
949 reduction, separated by vertical bars: so, C<term> followed by C<=>
950 followed by C<term> makes a C<term>, and C<term> followed by C<+>
951 followed by C<term> can also make a C<term>.
953 So, if you see two terms with an C<=> or C<+>, between them, you can
954 turn them into a single expression. When you do this, you execute the
955 code in the block on the next line: if you see C<=>, you'll do the code
956 in line 2. If you see C<+>, you'll do the code in line 4. It's this code
957 which contributes to the op tree.
960 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
962 What this does is creates a new binary op, and feeds it a number of
963 variables. The variables refer to the tokens: C<$1> is the first token in
964 the input, C<$2> the second, and so on - think regular expression
965 backreferences. C<$$> is the op returned from this reduction. So, we
966 call C<newBINOP> to create a new binary operator. The first parameter to
967 C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
968 operator, so we want the type to be C<ADDOP>. We could specify this
969 directly, but it's right there as the second token in the input, so we
970 use C<$2>. The second parameter is the op's flags: 0 means "nothing
971 special". Then the things to add: the left and right hand side of our
972 expression, in scalar context.
976 When perl executes something like C<addop>, how does it pass on its
977 results to the next op? The answer is, through the use of stacks. Perl
978 has a number of stacks to store things it's currently working on, and
979 we'll look at the three most important ones here.
985 Arguments are passed to PP code and returned from PP code using the
986 argument stack, C<ST>. The typical way to handle arguments is to pop
987 them off the stack, deal with them how you wish, and then push the result
988 back onto the stack. This is how, for instance, the cosine operator
993 value = Perl_cos(value);
996 We'll see a more tricky example of this when we consider Perl's macros
997 below. C<POPn> gives you the NV (floating point value) of the top SV on
998 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
999 the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1000 should be extended if necessary - it can't be necessary here, because we
1001 know there's room for one more item on the stack, since we've just
1002 removed one! The C<XPUSH*> macros at least guarantee safety.
1004 Alternatively, you can fiddle with the stack directly: C<SP> gives you
1005 the first element in your portion of the stack, and C<TOP*> gives you
1006 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1007 negation of an integer:
1011 Just set the integer value of the top stack entry to its negation.
1013 Argument stack manipulation in the core is exactly the same as it is in
1014 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1015 description of the macros used in stack manipulation.
1019 I say "your portion of the stack" above because PP code doesn't
1020 necessarily get the whole stack to itself: if your function calls
1021 another function, you'll only want to expose the arguments aimed for the
1022 called function, and not (necessarily) let it get at your own data. The
1023 way we do this is to have a "virtual" bottom-of-stack, exposed to each
1024 function. The mark stack keeps bookmarks to locations in the argument
1025 stack usable by each function. For instance, when dealing with a tied
1026 variable, (internally, something with "P" magic) Perl has to call
1027 methods for accesses to the tied variables. However, we need to separate
1028 the arguments exposed to the method to the argument exposed to the
1029 original function - the store or fetch or whatever it may be. Here's
1030 roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>:
1034 3 PUSHs(SvTIED_obj((SV*)av, mg));
1038 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1041 Let's examine the whole implementation, for practice:
1045 Push the current state of the stack pointer onto the mark stack. This is
1046 so that when we've finished adding items to the argument stack, Perl
1047 knows how many things we've added recently.
1050 3 PUSHs(SvTIED_obj((SV*)av, mg));
1053 We're going to add two more items onto the argument stack: when you have
1054 a tied array, the C<PUSH> subroutine receives the object and the value
1055 to be pushed, and that's exactly what we have here - the tied object,
1056 retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1060 Next we tell Perl to update the global stack pointer from our internal
1061 variable: C<dSP> only gave us a local copy, not a reference to the global.
1064 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1067 C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1068 variables are tidied up, everything that has been localised gets
1069 its previous value returned, and so on. Think of them as the C<{> and
1070 C<}> of a Perl block.
1072 To actually do the magic method call, we have to call a subroutine in
1073 Perl space: C<call_method> takes care of that, and it's described in
1074 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1075 going to discard its return value. The call_method() function
1076 removes the top element of the mark stack, so there is nothing for
1077 the caller to clean up.
1081 C doesn't have a concept of local scope, so perl provides one. We've
1082 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1083 stack implements the C equivalent of, for example:
1090 See L<perlguts/Localising Changes> for how to use the save stack.
1094 =head2 Millions of Macros
1096 One thing you'll notice about the Perl source is that it's full of
1097 macros. Some have called the pervasive use of macros the hardest thing
1098 to understand, others find it adds to clarity. Let's take an example,
1099 the code which implements the addition operator:
1103 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1106 6 SETn( left + right );
1111 Every line here (apart from the braces, of course) contains a macro. The
1112 first line sets up the function declaration as Perl expects for PP code;
1113 line 3 sets up variable declarations for the argument stack and the
1114 target, the return value of the operation. Finally, it tries to see if
1115 the addition operation is overloaded; if so, the appropriate subroutine
1118 Line 5 is another variable declaration - all variable declarations start
1119 with C<d> - which pops from the top of the argument stack two NVs (hence
1120 C<nn>) and puts them into the variables C<right> and C<left>, hence the
1121 C<rl>. These are the two operands to the addition operator. Next, we
1122 call C<SETn> to set the NV of the return value to the result of adding
1123 the two values. This done, we return - the C<RETURN> macro makes sure
1124 that our return value is properly handled, and we pass the next operator
1125 to run back to the main run loop.
1127 Most of these macros are explained in L<perlapi>, and some of the more
1128 important ones are explained in L<perlxs> as well. Pay special attention
1129 to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1130 the C<[pad]THX_?> macros.
1132 =head2 The .i Targets
1134 You can expand the macros in a F<foo.c> file by saying
1138 which will expand the macros using cpp. Don't be scared by the results.
1140 =head1 SOURCE CODE STATIC ANALYSIS
1142 Various tools exist for analysing C source code B<statically>, as
1143 opposed to B<dynamically>, that is, without executing the code.
1144 It is possible to detect resource leaks, undefined behaviour, type
1145 mismatches, portability problems, code paths that would cause illegal
1146 memory accesses, and other similar problems by just parsing the C code
1147 and looking at the resulting graph, what does it tell about the
1148 execution and data flows. As a matter of fact, this is exactly
1149 how C compilers know to give warnings about dubious code.
1153 The good old C code quality inspector, C<lint>, is available in
1154 several platforms, but please be aware that there are several
1155 different implementations of it by different vendors, which means that
1156 the flags are not identical across different platforms.
1158 There is a lint variant called C<splint> (Secure Programming Lint)
1159 available from http://www.splint.org/ that should compile on any
1162 There are C<lint> and <splint> targets in Makefile, but you may have
1163 to diddle with the flags (see above).
1167 Coverity (http://www.coverity.com/) is a product similar to lint and
1168 as a testbed for their product they periodically check several open
1169 source projects, and they give out accounts to open source developers
1170 to the defect databases.
1172 =head2 cpd (cut-and-paste detector)
1174 The cpd tool detects cut-and-paste coding. If one instance of the
1175 cut-and-pasted code changes, all the other spots should probably be
1176 changed, too. Therefore such code should probably be turned into a
1177 subroutine or a macro.
1179 cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1180 (http://pmd.sourceforge.net/). pmd was originally written for static
1181 analysis of Java code, but later the cpd part of it was extended to
1182 parse also C and C++.
1184 Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1185 pmd-X.Y.jar from it, and then run that on source code thusly:
1187 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1189 You may run into memory limits, in which case you should use the -Xmx option:
1195 Though much can be written about the inconsistency and coverage
1196 problems of gcc warnings (like C<-Wall> not meaning "all the
1197 warnings", or some common portability problems not being covered by
1198 C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined
1199 collection of warnings, and so forth), gcc is still a useful tool in
1200 keeping our coding nose clean.
1202 The C<-Wall> is by default on.
1204 The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1205 always, but unfortunately they are not safe on all platforms, they can
1206 for example cause fatal conflicts with the system headers (Solaris
1207 being a prime example). If Configure C<-Dgccansipedantic> is used,
1208 the C<cflags> frontend selects C<-ansi -pedantic> for the platforms
1209 where they are known to be safe.
1211 Starting from Perl 5.9.4 the following extra flags are added:
1225 C<-Wdeclaration-after-statement>
1229 The following flags would be nice to have but they would first need
1230 their own Augean stablemaster:
1244 C<-Wstrict-prototypes>
1248 The C<-Wtraditional> is another example of the annoying tendency of
1249 gcc to bundle a lot of warnings under one switch -- it would be
1250 impossible to deploy in practice because it would complain a lot -- but
1251 it does contain some warnings that would be beneficial to have available
1252 on their own, such as the warning about string constants inside macros
1253 containing the macro arguments: this behaved differently pre-ANSI
1254 than it does in ANSI, and some C compilers are still in transition,
1255 AIX being an example.
1257 =head2 Warnings of other C compilers
1259 Other C compilers (yes, there B<are> other C compilers than gcc) often
1260 have their "strict ANSI" or "strict ANSI with some portability extensions"
1261 modes on, like for example the Sun Workshop has its C<-Xa> mode on
1262 (though implicitly), or the DEC (these days, HP...) has its C<-std1>
1267 You can compile a special debugging version of Perl, which allows you
1268 to use the C<-D> option of Perl to tell more about what Perl is doing.
1269 But sometimes there is no alternative than to dive in with a debugger,
1270 either to see the stack trace of a core dump (very useful in a bug
1271 report), or trying to figure out what went wrong before the core dump
1272 happened, or how did we end up having wrong or unexpected results.
1274 =head2 Poking at Perl
1276 To really poke around with Perl, you'll probably want to build Perl for
1277 debugging, like this:
1279 ./Configure -d -D optimize=-g
1282 C<-g> is a flag to the C compiler to have it produce debugging
1283 information which will allow us to step through a running program,
1284 and to see in which C function we are at (without the debugging
1285 information we might see only the numerical addresses of the functions,
1286 which is not very helpful).
1288 F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1289 enables all the internal debugging code in Perl. There are a whole bunch
1290 of things you can debug with this: L<perlrun> lists them all, and the
1291 best way to find out about them is to play about with them. The most
1292 useful options are probably
1294 l Context (loop) stack processing
1296 o Method and overloading resolution
1297 c String/numeric conversions
1299 Some of the functionality of the debugging code can be achieved using XS
1302 -Dr => use re 'debug'
1303 -Dx => use O 'Debug'
1305 =head2 Using a source-level debugger
1307 If the debugging output of C<-D> doesn't help you, it's time to step
1308 through perl's execution with a source-level debugger.
1314 We'll use C<gdb> for our examples here; the principles will apply to
1315 any debugger (many vendors call their debugger C<dbx>), but check the
1316 manual of the one you're using.
1320 To fire up the debugger, type
1324 Or if you have a core dump:
1328 You'll want to do that in your Perl source tree so the debugger can read
1329 the source code. You should see the copyright message, followed by the
1334 C<help> will get you into the documentation, but here are the most
1341 Run the program with the given arguments.
1343 =item break function_name
1345 =item break source.c:xxx
1347 Tells the debugger that we'll want to pause execution when we reach
1348 either the named function (but see L<perlguts/Internal Functions>!) or the given
1349 line in the named source file.
1353 Steps through the program a line at a time.
1357 Steps through the program a line at a time, without descending into
1362 Run until the next breakpoint.
1366 Run until the end of the current function, then stop again.
1370 Just pressing Enter will do the most recent operation again - it's a
1371 blessing when stepping through miles of source code.
1375 Execute the given C code and print its results. B<WARNING>: Perl makes
1376 heavy use of macros, and F<gdb> does not necessarily support macros
1377 (see later L</"gdb macro support">). You'll have to substitute them
1378 yourself, or to invoke cpp on the source code files
1379 (see L</"The .i Targets">)
1380 So, for instance, you can't say
1382 print SvPV_nolen(sv)
1386 print Perl_sv_2pv_nolen(sv)
1390 You may find it helpful to have a "macro dictionary", which you can
1391 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1392 recursively apply those macros for you.
1394 =head2 gdb macro support
1396 Recent versions of F<gdb> have fairly good macro support, but
1397 in order to use it you'll need to compile perl with macro definitions
1398 included in the debugging information. Using F<gcc> version 3.1, this
1399 means configuring with C<-Doptimize=-g3>. Other compilers might use a
1400 different switch (if they support debugging macros at all).
1402 =head2 Dumping Perl Data Structures
1404 One way to get around this macro hell is to use the dumping functions in
1405 F<dump.c>; these work a little like an internal
1406 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1407 that you can't get at from Perl. Let's take an example. We'll use the
1408 C<$a = $b + $c> we used before, but give it a bit of context:
1409 C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1411 What about C<pp_add>, the function we examined earlier to implement the
1414 (gdb) break Perl_pp_add
1415 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1417 Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
1418 With the breakpoint in place, we can run our program:
1420 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1422 Lots of junk will go past as gdb reads in the relevant source files and
1423 libraries, and then:
1425 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
1426 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1431 We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1432 arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1435 #define dPOPTOPnnrl_ul NV right = POPn; \
1436 SV *leftsv = TOPs; \
1437 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1439 C<POPn> takes the SV from the top of the stack and obtains its NV either
1440 directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1441 C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1442 C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1443 C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1445 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1446 convert it. If we step again, we'll find ourselves there:
1448 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1452 We can now use C<Perl_sv_dump> to investigate the SV:
1454 SV = PV(0xa057cc0) at 0xa0675d0
1457 PV = 0xa06a510 "6XXXX"\0
1462 We know we're going to get C<6> from this, so let's finish the
1466 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1467 0x462669 in Perl_pp_add () at pp_hot.c:311
1470 We can also dump out this op: the current op is always stored in
1471 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1472 similar output to L<B::Debug|B::Debug>.
1475 13 TYPE = add ===> 14
1477 FLAGS = (SCALAR,KIDS)
1479 TYPE = null ===> (12)
1481 FLAGS = (SCALAR,KIDS)
1483 11 TYPE = gvsv ===> 12
1489 # finish this later #
1493 All right, we've now had a look at how to navigate the Perl sources and
1494 some things you'll need to know when fiddling with them. Let's now get
1495 on and create a simple patch. Here's something Larry suggested: if a
1496 C<U> is the first active format during a C<pack>, (for example,
1497 C<pack "U3C8", @stuff>) then the resulting string should be treated as
1500 How do we prepare to fix this up? First we locate the code in question -
1501 the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1502 files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1503 altering this file, let's copy it to F<pp.c~>.
1505 [Well, it was in F<pp.c> when this tutorial was written. It has now been
1506 split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1508 Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1509 loop over the pattern, taking each format character in turn into
1510 C<datum_type>. Then for each possible format character, we swallow up
1511 the other arguments in the pattern (a field width, an asterisk, and so
1512 on) and convert the next chunk input into the specified format, adding
1513 it onto the output SV C<cat>.
1515 How do we know if the C<U> is the first format in the C<pat>? Well, if
1516 we have a pointer to the start of C<pat> then, if we see a C<U> we can
1517 test whether we're still at the start of the string. So, here's where
1521 register char *pat = SvPVx(*++MARK, fromlen);
1522 register char *patend = pat + fromlen;
1527 We'll have another string pointer in there:
1530 register char *pat = SvPVx(*++MARK, fromlen);
1531 register char *patend = pat + fromlen;
1537 And just before we start the loop, we'll set C<patcopy> to be the start
1542 sv_setpvn(cat, "", 0);
1544 while (pat < patend) {
1546 Now if we see a C<U> which was at the start of the string, we turn on
1547 the C<UTF8> flag for the output SV, C<cat>:
1549 + if (datumtype == 'U' && pat==patcopy+1)
1551 if (datumtype == '#') {
1552 while (pat < patend && *pat != '\n')
1555 Remember that it has to be C<patcopy+1> because the first character of
1556 the string is the C<U> which has been swallowed into C<datumtype!>
1558 Oops, we forgot one thing: what if there are spaces at the start of the
1559 pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1560 character, even though it's not the first thing in the pattern. In this
1561 case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1563 if (isSPACE(datumtype))
1568 if (isSPACE(datumtype)) {
1573 OK. That's the C part done. Now we must do two additional things before
1574 this patch is ready to go: we've changed the behaviour of Perl, and so
1575 we must document that change. We must also provide some more regression
1576 tests to make sure our patch works and doesn't create a bug somewhere
1577 else along the line.
1579 The regression tests for each operator live in F<t/op/>, and so we
1580 make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1581 tests to the end. First, we'll test that the C<U> does indeed create
1584 t/op/pack.t has a sensible ok() function, but if it didn't we could
1585 use the one from t/test.pl.
1587 require './test.pl';
1588 plan( tests => 159 );
1592 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1593 print "ok $test\n"; $test++;
1595 we can write the more sensible (see L<Test::More> for a full
1596 explanation of is() and other testing functions).
1598 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
1599 "U* produces Unicode" );
1601 Now we'll test that we got that space-at-the-beginning business right:
1603 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
1604 " with spaces at the beginning" );
1606 And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1607 the first active format:
1609 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
1610 "U* not first isn't Unicode" );
1612 Mustn't forget to change the number of tests which appears at the top,
1613 or else the automated tester will get confused. This will either look
1620 plan( tests => 156 );
1622 We now compile up Perl, and run it through the test suite. Our new
1625 Finally, the documentation. The job is never done until the paperwork is
1626 over, so let's describe the change we've just made. The relevant place
1627 is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1628 this text in the description of C<pack>:
1632 If the pattern begins with a C<U>, the resulting string will be treated
1633 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1634 with an initial C<U0>, and the bytes that follow will be interpreted as
1635 Unicode characters. If you don't want this to happen, you can begin your
1636 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
1637 string, and then follow this with a C<U*> somewhere in your pattern.
1639 All done. Now let's create the patch. F<Porting/patching.pod> tells us
1640 that if we're making major changes, we should copy the entire directory
1641 to somewhere safe before we begin fiddling, and then do
1643 diff -ruN old new > patch
1645 However, we know which files we've changed, and we can simply do this:
1647 diff -u pp.c~ pp.c > patch
1648 diff -u t/op/pack.t~ t/op/pack.t >> patch
1649 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1651 We end up with a patch looking a little like this:
1653 --- pp.c~ Fri Jun 02 04:34:10 2000
1654 +++ pp.c Fri Jun 16 11:37:25 2000
1655 @@ -4375,6 +4375,7 @@
1658 register char *pat = SvPVx(*++MARK, fromlen);
1660 register char *patend = pat + fromlen;
1663 @@ -4405,6 +4406,7 @@
1666 And finally, we submit it, with our rationale, to perl5-porters. Job
1669 =head2 Patching a core module
1671 This works just like patching anything else, with an extra
1672 consideration. Many core modules also live on CPAN. If this is so,
1673 patch the CPAN version instead of the core and send the patch off to
1674 the module maintainer (with a copy to p5p). This will help the module
1675 maintainer keep the CPAN version in sync with the core version without
1676 constantly scanning p5p.
1678 The list of maintainers of core modules is usefully documented in
1679 F<Porting/Maintainers.pl>.
1681 =head2 Adding a new function to the core
1683 If, as part of a patch to fix a bug, or just because you have an
1684 especially good idea, you decide to add a new function to the core,
1685 discuss your ideas on p5p well before you start work. It may be that
1686 someone else has already attempted to do what you are considering and
1687 can give lots of good advice or even provide you with bits of code
1688 that they already started (but never finished).
1690 You have to follow all of the advice given above for patching. It is
1691 extremely important to test any addition thoroughly and add new tests
1692 to explore all boundary conditions that your new function is expected
1693 to handle. If your new function is used only by one module (e.g. toke),
1694 then it should probably be named S_your_function (for static); on the
1695 other hand, if you expect it to accessible from other functions in
1696 Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1699 The location of any new code is also an important consideration. Don't
1700 just create a new top level .c file and put your code there; you would
1701 have to make changes to Configure (so the Makefile is created properly),
1702 as well as possibly lots of include files. This is strictly pumpking
1705 It is better to add your function to one of the existing top level
1706 source code files, but your choice is complicated by the nature of
1707 the Perl distribution. Only the files that are marked as compiled
1708 static are located in the perl executable. Everything else is located
1709 in the shared library (or DLL if you are running under WIN32). So,
1710 for example, if a function was only used by functions located in
1711 toke.c, then your code can go in toke.c. If, however, you want to call
1712 the function from universal.c, then you should put your code in another
1713 location, for example util.c.
1715 In addition to writing your c-code, you will need to create an
1716 appropriate entry in embed.pl describing your function, then run
1717 'make regen_headers' to create the entries in the numerous header
1718 files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1719 for information on the various options that you can set in embed.pl.
1720 You will forget to do this a few (or many) times and you will get
1721 warnings during the compilation phase. Make sure that you mention
1722 this when you post your patch to P5P; the pumpking needs to know this.
1724 When you write your new code, please be conscious of existing code
1725 conventions used in the perl source files. See L<perlstyle> for
1726 details. Although most of the guidelines discussed seem to focus on
1727 Perl code, rather than c, they all apply (except when they don't ;).
1728 See also I<Porting/patching.pod> file in the Perl source distribution
1729 for lots of details about both formatting and submitting patches of
1732 Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1733 Test on as many platforms as you can find. Test as many perl
1734 Configure options as you can (e.g. MULTIPLICITY). If you have
1735 profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
1736 below for how to use them to further test your code. Remember that
1737 most of the people on P5P are doing this on their own time and
1738 don't have the time to debug your code.
1740 =head2 Writing a test
1742 Every module and built-in function has an associated test file (or
1743 should...). If you add or change functionality, you have to write a
1744 test. If you fix a bug, you have to write a test so that bug never
1745 comes back. If you alter the docs, it would be nice to test what the
1746 new documentation says.
1748 In short, if you submit a patch you probably also have to patch the
1751 For modules, the test file is right next to the module itself.
1752 F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1753 so there are some snags (and it would be wonderful for you to brush
1754 them out), but it basically works that way. Everything else lives in
1761 Testing of the absolute basic functionality of Perl. Things like
1762 C<if>, basic file reads and writes, simple regexes, etc. These are
1763 run first in the test suite and if any of them fail, something is
1768 These test the basic control structures, C<if/else>, C<while>,
1773 Tests basic issues of how Perl parses and compiles itself.
1777 Tests for built-in IO functions, including command line arguments.
1781 The old home for the module tests, you shouldn't put anything new in
1782 here. There are still some bits and pieces hanging around in here
1783 that need to be moved. Perhaps you could move them? Thanks!
1787 Tests for perl's method resolution order implementations
1792 Tests for perl's built in functions that don't fit into any of the
1797 Tests for POD directives. There are still some tests for the Pod
1798 modules hanging around in here that need to be moved out into F<lib/>.
1802 Testing features of how perl actually runs, including exit codes and
1803 handling of PERL* environment variables.
1807 Tests for the core support of Unicode.
1811 Windows-specific tests.
1815 A test suite for the s2p converter.
1819 The core uses the same testing style as the rest of Perl, a simple
1820 "ok/not ok" run through Test::Harness, but there are a few special
1823 There are three ways to write a test in the core. Test::More,
1824 t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1825 decision of which to use depends on what part of the test suite you're
1826 working on. This is a measure to prevent a high-level failure (such
1827 as Config.pm breaking) from causing basic functionality tests to fail.
1833 Since we don't know if require works, or even subroutines, use ad hoc
1834 tests for these two. Step carefully to avoid using the feature being
1837 =item t/cmd t/run t/io t/op
1839 Now that basic require() and subroutines are tested, you can use the
1840 t/test.pl library which emulates the important features of Test::More
1841 while using a minimum of core features.
1843 You can also conditionally use certain libraries like Config, but be
1844 sure to skip the test gracefully if it's not there.
1848 Now that the core of Perl is tested, Test::More can be used. You can
1849 also use the full suite of core modules in the tests.
1853 When you say "make test" Perl uses the F<t/TEST> program to run the
1854 test suite (except under Win32 where it uses F<t/harness> instead.)
1855 All tests are run from the F<t/> directory, B<not> the directory
1856 which contains the test. This causes some problems with the tests
1857 in F<lib/>, so here's some opportunity for some patching.
1859 You must be triply conscious of cross-platform concerns. This usually
1860 boils down to using File::Spec and avoiding things like C<fork()> and
1861 C<system()> unless absolutely necessary.
1863 =head2 Special Make Test Targets
1865 There are various special make targets that can be used to test Perl
1866 slightly differently than the standard "test" target. Not all them
1867 are expected to give a 100% success rate. Many of them have several
1868 aliases, and many of them are not available on certain operating
1875 Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
1877 (Not available on Win32)
1881 Run all the tests through B::Deparse. Not all tests will succeed.
1883 (Not available on Win32)
1885 =item test.taintwarn
1887 Run all tests with the B<-t> command-line switch. Not all tests
1888 are expected to succeed (until they're specifically fixed, of course).
1890 (Not available on Win32)
1894 Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
1895 F<t/op>, F<t/uni> and F<t/mro> tests.
1897 =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1899 (Only in Linux) Run all the tests using the memory leak + naughty
1900 memory access tool "valgrind". The log files will be named
1901 F<testname.valgrind>.
1903 =item test.third check.third utest.third ucheck.third
1905 (Only in Tru64) Run all the tests using the memory leak + naughty
1906 memory access tool "Third Degree". The log files will be named
1907 F<perl.3log.testname>.
1909 =item test.torture torturetest
1911 Run all the usual tests and some extra tests. As of Perl 5.8.0 the
1912 only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
1914 You can also run the torture test with F<t/harness> by giving
1915 C<-torture> argument to F<t/harness>.
1917 =item utest ucheck test.utf8 check.utf8
1919 Run all the tests with -Mutf8. Not all tests will succeed.
1921 (Not available on Win32)
1923 =item minitest.utf16 test.utf16
1925 Runs the tests with UTF-16 encoded scripts, encoded with different
1926 versions of this encoding.
1928 C<make utest.utf16> runs the test suite with a combination of C<-utf8> and
1929 C<-utf16> arguments to F<t/TEST>.
1931 (Not available on Win32)
1935 Run the test suite with the F<t/harness> controlling program, instead of
1936 F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1937 L<Test::Harness> module, thus using this test target supposes that perl
1938 mostly works. The main advantage for our purposes is that it prints a
1939 detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1940 doesn't redirect stderr to stdout.
1942 Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so
1943 there is no special "test_harness" target.
1945 Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES
1946 environment variables to control the behaviour of F<t/harness>. This means
1949 nmake test TEST_FILES="op/*.t"
1950 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
1952 =item test-notty test_notty
1954 Sets PERL_SKIP_TTY_TEST to true before running normal test.
1958 =head2 Running tests by hand
1960 You can run part of the test suite by hand by using one the following
1961 commands from the F<t/> directory :
1963 ./perl -I../lib TEST list-of-.t-files
1967 ./perl -I../lib harness list-of-.t-files
1969 (if you don't specify test scripts, the whole test suite will be run.)
1971 =head3 Using t/harness for testing
1973 If you use C<harness> for testing you have several command line options
1974 available to you. The arguments are as follows, and are in the order
1975 that they must appear if used together.
1977 harness -v -torture -re=pattern LIST OF FILES TO TEST
1978 harness -v -torture -re LIST OF PATTERNS TO MATCH
1980 If C<LIST OF FILES TO TEST> is omitted the file list is obtained from
1981 the manifest. The file list may include shell wildcards which will be
1988 Run the tests under verbose mode so you can see what tests were run,
1993 Run the torture tests as well as the normal set.
1997 Filter the file list so that all the test files run match PATTERN.
1998 Note that this form is distinct from the B<-re LIST OF PATTERNS> form below
1999 in that it allows the file list to be provided as well.
2001 =item -re LIST OF PATTERNS
2003 Filter the file list so that all the test files run match
2004 /(LIST|OF|PATTERNS)/. Note that with this form the patterns
2005 are joined by '|' and you cannot supply a list of files, instead
2006 the test files are obtained from the MANIFEST.
2010 You can run an individual test by a command similar to
2012 ./perl -I../lib patho/to/foo.t
2014 except that the harnesses set up some environment variables that may
2015 affect the execution of the test :
2021 indicates that we're running this test part of the perl core test suite.
2022 This is useful for modules that have a dual life on CPAN.
2024 =item PERL_DESTRUCT_LEVEL=2
2026 is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
2030 (used only by F<t/TEST>) if set, overrides the path to the perl executable
2031 that should be used to run the tests (the default being F<./perl>).
2033 =item PERL_SKIP_TTY_TEST
2035 if set, tells to skip the tests that need a terminal. It's actually set
2036 automatically by the Makefile, but can also be forced artificially by
2037 running 'make test_notty'.
2041 =head3 Other environment variables that may influence tests
2045 =item PERL_TEST_Net_Ping
2047 Setting this variable runs all the Net::Ping modules tests,
2048 otherwise some tests that interact with the outside world are skipped.
2051 =item PERL_TEST_NOVREXX
2053 Setting this variable skips the vrexx.t tests for OS2::REXX.
2055 =item PERL_TEST_NUMCONVERTS
2057 This sets a variable in op/numconvert.t.
2061 See also the documentation for the Test and Test::Harness modules,
2062 for more environment variables that affect testing.
2064 =head2 Common problems when patching Perl source code
2066 Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
2067 some cases we have to take pre-ANSI requirements into consideration.
2068 You don't care about some particular platform having broken Perl?
2069 I hear there is still a strong demand for J2EE programmers.
2071 =head2 Perl environment problems
2077 Not compiling with threading
2079 Compiling with threading (-Duseithreads) completely rewrites
2080 the function prototypes of Perl. You better try your changes
2081 with that. Related to this is the difference between "Perl_-less"
2082 and "Perl_-ly" APIs, for example:
2084 Perl_sv_setiv(aTHX_ ...);
2087 The first one explicitly passes in the context, which is needed for e.g.
2088 threaded builds. The second one does that implicitly; do not get them
2089 mixed. If you are not passing in a aTHX_, you will need to do a dTHX
2090 (or a dVAR) as the first thing in the function.
2092 See L<perlguts/"How multiple interpreters and concurrency are supported">
2093 for further discussion about context.
2097 Not compiling with -DDEBUGGING
2099 The DEBUGGING define exposes more code to the compiler,
2100 therefore more ways for things to go wrong. You should try it.
2104 Introducing (non-read-only) globals
2106 Do not introduce any modifiable globals, truly global or file static.
2107 They are bad form and complicate multithreading and other forms of
2108 concurrency. The right way is to introduce them as new interpreter
2109 variables, see F<intrpvar.h> (at the very end for binary compatibility).
2111 Introducing read-only (const) globals is okay, as long as you verify
2112 with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
2113 BSD-style output) that the data you added really is read-only.
2114 (If it is, it shouldn't show up in the output of that command.)
2116 If you want to have static strings, make them constant:
2118 static const char etc[] = "...";
2120 If you want to have arrays of constant strings, note carefully
2121 the right combination of C<const>s:
2123 static const char * const yippee[] =
2124 {"hi", "ho", "silver"};
2126 There is a way to completely hide any modifiable globals (they are all
2127 moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>.
2128 It is not normally used, but can be used for testing, read more
2129 about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">.
2133 Not exporting your new function
2135 Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
2136 function that is part of the public API (the shared Perl library)
2137 to be explicitly marked as exported. See the discussion about
2138 F<embed.pl> in L<perlguts>.
2142 Exporting your new function
2144 The new shiny result of either genuine new functionality or your
2145 arduous refactoring is now ready and correctly exported. So what
2146 could possibly go wrong?
2148 Maybe simply that your function did not need to be exported in the
2149 first place. Perl has a long and not so glorious history of exporting
2150 functions that it should not have.
2152 If the function is used only inside one source code file, make it
2153 static. See the discussion about F<embed.pl> in L<perlguts>.
2155 If the function is used across several files, but intended only for
2156 Perl's internal use (and this should be the common case), do not
2157 export it to the public API. See the discussion about F<embed.pl>
2162 =head2 Portability problems
2164 The following are common causes of compilation and/or execution
2165 failures, not common to Perl as such. The C FAQ is good bedtime
2166 reading. Please test your changes with as many C compilers and
2167 platforms as possible -- we will, anyway, and it's nice to save
2168 oneself from public embarrassment.
2170 If using gcc, you can add the C<-std=c89> option which will hopefully
2171 catch most of these unportabilities. (However it might also catch
2172 incompatibilities in your system's header files.)
2174 Use the Configure C<-Dgccansipedantic> flag to enable the gcc
2175 C<-ansi -pedantic> flags which enforce stricter ANSI rules.
2177 If using the C<gcc -Wall> note that not all the possible warnings
2178 (like C<-Wunitialized>) are given unless you also compile with C<-O>.
2180 Note that if using gcc, starting from Perl 5.9.5 the Perl core source
2181 code files (the ones at the top level of the source code distribution,
2182 but not e.g. the extensions under ext/) are automatically compiled
2183 with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>,
2184 and a selection of C<-W> flags (see cflags.SH).
2186 Also study L<perlport> carefully to avoid any bad assumptions
2187 about the operating system, filesystems, and so forth.
2189 You may once in a while try a "make microperl" to see whether we
2190 can still compile Perl with just the bare minimum of interfaces.
2193 Do not assume an operating system indicates a certain compiler.
2199 Casting pointers to integers or casting integers to pointers
2201 void castaway(U8* p)
2207 void castaway(U8* p)
2211 Both are bad, and broken, and unportable. Use the PTR2IV()
2212 macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(),
2213 INT2PTR(), and NUM2PTR().)
2217 Casting between data function pointers and data pointers
2219 Technically speaking casting between function pointers and data
2220 pointers is unportable and undefined, but practically speaking
2221 it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR()
2222 macros. Sometimes you can also play games with unions.
2226 Assuming sizeof(int) == sizeof(long)
2228 There are platforms where longs are 64 bits, and platforms where ints
2229 are 64 bits, and while we are out to shock you, even platforms where
2230 shorts are 64 bits. This is all legal according to the C standard.
2231 (In other words, "long long" is not a portable way to specify 64 bits,
2232 and "long long" is not even guaranteed to be any wider than "long".)
2234 Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
2235 Avoid things like I32 because they are B<not> guaranteed to be
2236 I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
2237 guaranteed to be B<int> or B<long>. If you really explicitly need
2238 64-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
2242 Assuming one can dereference any type of pointer for any type of data
2245 long pony = *p; /* BAD */
2247 Many platforms, quite rightly so, will give you a core dump instead
2248 of a pony if the p happens not be correctly aligned.
2254 (int)*p = ...; /* BAD */
2256 Simply not portable. Get your lvalue to be of the right type,
2257 or maybe use temporary variables, or dirty tricks with unions.
2261 Assume B<anything> about structs (especially the ones you
2262 don't control, like the ones coming from the system headers)
2268 That a certain field exists in a struct
2272 That no other fields exist besides the ones you know of
2276 That a field is of certain signedness, sizeof, or type
2280 That the fields are in a certain order
2286 While C guarantees the ordering specified in the struct definition,
2287 between different platforms the definitions might differ
2293 That the sizeof(struct) or the alignments are the same everywhere
2299 There might be padding bytes between the fields to align the fields -
2300 the bytes can be anything
2304 Structs are required to be aligned to the maximum alignment required
2305 by the fields - which for native types is for usually equivalent to
2306 sizeof() of the field
2314 Assuming the character set is ASCIIish
2316 Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
2317 This is transparent for the most part, but because the character sets
2318 differ, you shouldn't use numeric (decimal, octal, nor hex) constants
2319 to refer to characters. You can safely say 'A', but not 0x41.
2320 You can safely say '\n', but not \012.
2321 If a character doesn't have a trivial input form, you can
2322 create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that
2323 it resolves to different values depending on the character set being used.
2324 (There are three different EBCDIC character sets defined in C<utfebcdic.h>,
2325 so it might be best to insert the #define three times in that file.)
2327 Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case
2328 alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'.
2329 But '0' - '9' is an unbroken range in both systems. Don't assume anything
2332 Many of the comments in the existing code ignore the possibility of EBCDIC,
2333 and may be wrong therefore, even if the code works.
2334 This is actually a tribute to the successful transparent insertion of being
2335 able to handle EBCDIC without having to change pre-existing code.
2337 UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode
2338 code points as sequences of bytes. Macros
2339 with the same names (but different definitions)
2340 in C<utf8.h> and C<utfebcdic.h>
2341 are used to allow the calling code to think that there is only one such
2343 This is almost always referred to as C<utf8>, but it means the EBCDIC version
2344 as well. Again, comments in the code may well be wrong even if the code itself
2346 For example, the concept of C<invariant characters> differs between ASCII and
2348 On ASCII platforms, only characters that do not have the high-order
2349 bit set (i.e. whose ordinals are strict ASCII, 0 - 127)
2350 are invariant, and the documentation and comments in the code
2352 often referring to something like, say, C<hibit>.
2353 The situation differs and is not so simple on EBCDIC machines, but as long as
2354 the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it
2355 works, even if the comments are wrong.
2359 Assuming the character set is just ASCII
2361 ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
2362 characters have different meanings depending on the locale. Absent a locale,
2363 currently these extra characters are generally considered to be unassigned,
2364 and this has presented some problems.
2365 This is scheduled to be changed in 5.12 so that these characters will
2366 be considered to be Latin-1 (ISO-8859-1).
2370 Mixing #define and #ifdef
2372 #define BURGLE(x) ... \
2373 #ifdef BURGLE_OLD_STYLE /* BAD */
2374 ... do it the old way ... \
2376 ... do it the new way ... \
2379 You cannot portably "stack" cpp directives. For example in the above
2380 you need two separate BURGLE() #defines, one for each #ifdef branch.
2384 Adding non-comment stuff after #endif or #else
2388 #else !SNOSH /* BAD */
2390 #endif SNOSH /* BAD */
2392 The #endif and #else cannot portably have anything non-comment after
2393 them. If you want to document what is going (which is a good idea
2394 especially if the branches are long), use (C) comments:
2402 The gcc option C<-Wendif-labels> warns about the bad variant
2403 (by default on starting from Perl 5.9.4).
2407 Having a comma after the last element of an enum list
2415 is not portable. Leave out the last comma.
2417 Also note that whether enums are implicitly morphable to ints
2418 varies between compilers, you might need to (int).
2424 // This function bamfoodles the zorklator. /* BAD */
2426 That is C99 or C++. Perl is C89. Using the //-comments is silently
2427 allowed by many C compilers but cranking up the ANSI C89 strictness
2428 (which we like to do) causes the compilation to fail.
2432 Mixing declarations and code
2437 set_zorkmids(n); /* BAD */
2440 That is C99 or C++. Some C compilers allow that, but you shouldn't.
2442 The gcc option C<-Wdeclaration-after-statements> scans for such problems
2443 (by default on starting from Perl 5.9.4).
2447 Introducing variables inside for()
2449 for(int i = ...; ...; ...) { /* BAD */
2451 That is C99 or C++. While it would indeed be awfully nice to have that
2452 also in C89, to limit the scope of the loop variable, alas, we cannot.
2456 Mixing signed char pointers with unsigned char pointers
2458 int foo(char *s) { ... }
2460 unsigned char *t = ...; /* Or U8* t = ... */
2463 While this is legal practice, it is certainly dubious, and downright
2464 fatal in at least one platform: for example VMS cc considers this a
2465 fatal error. One cause for people often making this mistake is that a
2466 "naked char" and therefore dereferencing a "naked char pointer" have
2467 an undefined signedness: it depends on the compiler and the flags of
2468 the compiler and the underlying platform whether the result is signed
2469 or unsigned. For this very same reason using a 'char' as an array
2474 Macros that have string constants and their arguments as substrings of
2475 the string constants
2477 #define FOO(n) printf("number = %d\n", n) /* BAD */
2480 Pre-ANSI semantics for that was equivalent to
2482 printf("10umber = %d\10");
2484 which is probably not what you were expecting. Unfortunately at least
2485 one reasonably common and modern C compiler does "real backward
2486 compatibility" here, in AIX that is what still happens even though the
2487 rest of the AIX compiler is very happily C89.
2491 Using printf formats for non-basic C types
2494 printf("i = %d\n", i); /* BAD */
2496 While this might by accident work in some platform (where IV happens
2497 to be an C<int>), in general it cannot. IV might be something larger.
2498 Even worse the situation is with more specific types (defined by Perl's
2499 configuration step in F<config.h>):
2502 printf("who = %d\n", who); /* BAD */
2504 The problem here is that Uid_t might be not only not C<int>-wide
2505 but it might also be unsigned, in which case large uids would be
2506 printed as negative values.
2508 There is no simple solution to this because of printf()'s limited
2509 intelligence, but for many types the right format is available as
2510 with either 'f' or '_f' suffix, for example:
2512 IVdf /* IV in decimal */
2513 UVxf /* UV is hexadecimal */
2515 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
2517 Uid_t_f /* Uid_t in decimal */
2519 printf("who = %"Uid_t_f"\n", who);
2521 Or you can try casting to a "wide enough" type:
2523 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
2525 Also remember that the C<%p> format really does require a void pointer:
2528 printf("p = %p\n", (void*)p);
2530 The gcc option C<-Wformat> scans for such problems.
2534 Blindly using variadic macros
2536 gcc has had them for a while with its own syntax, and C99 brought
2537 them with a standardized syntax. Don't use the former, and use
2538 the latter only if the HAS_C99_VARIADIC_MACROS is defined.
2542 Blindly passing va_list
2544 Not all platforms support passing va_list to further varargs (stdarg)
2545 functions. The right thing to do is to copy the va_list using the
2546 Perl_va_copy() if the NEED_VA_COPY is defined.
2550 Using gcc statement expressions
2552 val = ({...;...;...}); /* BAD */
2554 While a nice extension, it's not portable. The Perl code does
2555 admittedly use them if available to gain some extra speed
2556 (essentially as a funky form of inlining), but you shouldn't.
2560 Binding together several statements in a macro
2562 Use the macros STMT_START and STMT_END.
2570 Testing for operating systems or versions when should be testing for features
2572 #ifdef __FOONIX__ /* BAD */
2576 Unless you know with 100% certainty that quux() is only ever available
2577 for the "Foonix" operating system B<and> that is available B<and>
2578 correctly working for B<all> past, present, B<and> future versions of
2579 "Foonix", the above is very wrong. This is more correct (though still
2580 not perfect, because the below is a compile-time check):
2586 How does the HAS_QUUX become defined where it needs to be? Well, if
2587 Foonix happens to be UNIXy enough to be able to run the Configure
2588 script, and Configure has been taught about detecting and testing
2589 quux(), the HAS_QUUX will be correctly defined. In other platforms,
2590 the corresponding configuration step will hopefully do the same.
2592 In a pinch, if you cannot wait for Configure to be educated,
2593 or if you have a good hunch of where quux() might be available,
2594 you can temporarily try the following:
2596 #if (defined(__FOONIX__) || defined(__BARNIX__))
2606 But in any case, try to keep the features and operating systems separate.
2610 =head2 Problematic System Interfaces
2616 malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
2617 allocate at least one byte. (In general you should rarely need to
2618 work at this low level, but instead use the various malloc wrappers.)
2622 snprintf() - the return type is unportable. Use my_snprintf() instead.
2626 =head2 Security problems
2628 Last but not least, here are various tips for safer coding.
2636 Or we will publicly ridicule you. Seriously.
2640 Do not use strcpy() or strcat() or strncpy() or strncat()
2642 Use my_strlcpy() and my_strlcat() instead: they either use the native
2643 implementation, or Perl's own implementation (borrowed from the public
2644 domain implementation of INN).
2648 Do not use sprintf() or vsprintf()
2650 If you really want just plain byte strings, use my_snprintf()
2651 and my_vsnprintf() instead, which will try to use snprintf() and
2652 vsnprintf() if those safer APIs are available. If you want something
2653 fancier than a plain byte string, use SVs and Perl_sv_catpvf().
2657 =head1 EXTERNAL TOOLS FOR DEBUGGING PERL
2659 Sometimes it helps to use external tools while debugging and
2660 testing Perl. This section tries to guide you through using
2661 some common testing and debugging tools with Perl. This is
2662 meant as a guide to interfacing these tools with Perl, not
2663 as any kind of guide to the use of the tools themselves.
2665 B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2666 Third Degree greatly slows down the execution: seconds become minutes,
2667 minutes become hours. For example as of Perl 5.8.1, the
2668 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2669 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2670 than six hours, even on a snappy computer-- the said test must be
2671 doing something that is quite unfriendly for memory debuggers. If you
2672 don't feel like waiting, that you can simply kill away the perl
2675 B<NOTE 2>: To minimize the number of memory leak false alarms (see
2676 L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2677 environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2678 and harness scripts do that automatically. But if you are running
2679 some of the tests manually-- for csh-like shells:
2681 setenv PERL_DESTRUCT_LEVEL 2
2683 and for Bourne-type shells:
2685 PERL_DESTRUCT_LEVEL=2
2686 export PERL_DESTRUCT_LEVEL
2688 or in UNIXy environments you can also use the C<env> command:
2690 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
2692 B<NOTE 3>: There are known memory leaks when there are compile-time
2693 errors within eval or require, seeing C<S_doeval> in the call stack
2694 is a good sign of these. Fixing these leaks is non-trivial,
2695 unfortunately, but they must be fixed eventually.
2697 B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
2698 unless Perl is built with the Configure option
2699 C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
2701 =head2 Rational Software's Purify
2703 Purify is a commercial tool that is helpful in identifying
2704 memory overruns, wild pointers, memory leaks and other such
2705 badness. Perl must be compiled in a specific way for
2706 optimal testing with Purify. Purify is available under
2707 Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2709 =head2 Purify on Unix
2711 On Unix, Purify creates a new Perl binary. To get the most
2712 benefit out of Purify, you should create the perl to Purify
2715 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2716 -Uusemymalloc -Dusemultiplicity
2718 where these arguments mean:
2722 =item -Accflags=-DPURIFY
2724 Disables Perl's arena memory allocation functions, as well as
2725 forcing use of memory allocation functions derived from the
2728 =item -Doptimize='-g'
2730 Adds debugging information so that you see the exact source
2731 statements where the problem occurs. Without this flag, all
2732 you will see is the source filename of where the error occurred.
2736 Disable Perl's malloc so that Purify can more closely monitor
2737 allocations and leaks. Using Perl's malloc will make Purify
2738 report most leaks in the "potential" leaks category.
2740 =item -Dusemultiplicity
2742 Enabling the multiplicity option allows perl to clean up
2743 thoroughly when the interpreter shuts down, which reduces the
2744 number of bogus leak reports from Purify.
2748 Once you've compiled a perl suitable for Purify'ing, then you
2753 which creates a binary named 'pureperl' that has been Purify'ed.
2754 This binary is used in place of the standard 'perl' binary
2755 when you want to debug Perl memory problems.
2757 As an example, to show any memory leaks produced during the
2758 standard Perl testset you would create and run the Purify'ed
2763 ../pureperl -I../lib harness
2765 which would run Perl on test.pl and report any memory problems.
2767 Purify outputs messages in "Viewer" windows by default. If
2768 you don't have a windowing environment or if you simply
2769 want the Purify output to unobtrusively go to a log file
2770 instead of to the interactive window, use these following
2771 options to output to the log file "perl.log":
2773 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2774 -log-file=perl.log -append-logfile=yes"
2776 If you plan to use the "Viewer" windows, then you only need this option:
2778 setenv PURIFYOPTIONS "-chain-length=25"
2780 In Bourne-type shells:
2783 export PURIFYOPTIONS
2785 or if you have the "env" utility:
2787 env PURIFYOPTIONS="..." ../pureperl ...
2791 Purify on Windows NT instruments the Perl binary 'perl.exe'
2792 on the fly. There are several options in the makefile you
2793 should change to get the most use out of Purify:
2799 You should add -DPURIFY to the DEFINES line so the DEFINES
2800 line looks something like:
2802 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
2804 to disable Perl's arena memory allocation functions, as
2805 well as to force use of memory allocation functions derived
2806 from the system malloc.
2808 =item USE_MULTI = define
2810 Enabling the multiplicity option allows perl to clean up
2811 thoroughly when the interpreter shuts down, which reduces the
2812 number of bogus leak reports from Purify.
2814 =item #PERL_MALLOC = define
2816 Disable Perl's malloc so that Purify can more closely monitor
2817 allocations and leaks. Using Perl's malloc will make Purify
2818 report most leaks in the "potential" leaks category.
2822 Adds debugging information so that you see the exact source
2823 statements where the problem occurs. Without this flag, all
2824 you will see is the source filename of where the error occurred.
2828 As an example, to show any memory leaks produced during the
2829 standard Perl testset you would create and run Purify as:
2834 purify ../perl -I../lib harness
2836 which would instrument Perl in memory, run Perl on test.pl,
2837 then finally report any memory problems.
2841 The excellent valgrind tool can be used to find out both memory leaks
2842 and illegal memory accesses. As of version 3.3.0, Valgrind only
2843 supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind"
2844 target can be used to run the tests under valgrind. Found errors
2845 and memory leaks are logged in files named F<testfile.valgrind>.
2847 Valgrind also provides a cachegrind tool, invoked on perl as:
2849 VG_OPTS=--tool=cachegrind make test.valgrind
2851 As system libraries (most notably glibc) are also triggering errors,
2852 valgrind allows to suppress such errors using suppression files. The
2853 default suppression file that comes with valgrind already catches a lot
2854 of them. Some additional suppressions are defined in F<t/perl.supp>.
2856 To get valgrind and for more information see
2858 http://developer.kde.org/~sewardj/
2860 =head2 Compaq's/Digital's/HP's Third Degree
2862 Third Degree is a tool for memory leak detection and memory access checks.
2863 It is one of the many tools in the ATOM toolkit. The toolkit is only
2864 available on Tru64 (formerly known as Digital UNIX formerly known as
2867 When building Perl, you must first run Configure with -Doptimize=-g
2868 and -Uusemymalloc flags, after that you can use the make targets
2869 "perl.third" and "test.third". (What is required is that Perl must be
2870 compiled using the C<-g> flag, you may need to re-Configure.)
2872 The short story is that with "atom" you can instrument the Perl
2873 executable to create a new executable called F<perl.third>. When the
2874 instrumented executable is run, it creates a log of dubious memory
2875 traffic in file called F<perl.3log>. See the manual pages of atom and
2876 third for more information. The most extensive Third Degree
2877 documentation is available in the Compaq "Tru64 UNIX Programmer's
2878 Guide", chapter "Debugging Programs with Third Degree".
2880 The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
2881 subdirectory. There is a problem with these files: Third Degree is so
2882 effective that it finds problems also in the system libraries.
2883 Therefore you should used the Porting/thirdclean script to cleanup
2884 the F<*.3log> files.
2886 There are also leaks that for given certain definition of a leak,
2887 aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2889 =head2 PERL_DESTRUCT_LEVEL
2891 If you want to run any of the tests yourself manually using e.g.
2892 valgrind, or the pureperl or perl.third executables, please note that
2893 by default perl B<does not> explicitly cleanup all the memory it has
2894 allocated (such as global memory arenas) but instead lets the exit()
2895 of the whole program "take care" of such allocations, also known as
2896 "global destruction of objects".
2898 There is a way to tell perl to do complete cleanup: set the
2899 environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2900 The t/TEST wrapper does set this to 2, and this is what you
2901 need to do too, if you don't want to see the "global leaks":
2902 For example, for "third-degreed" Perl:
2904 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
2906 (Note: the mod_perl apache module uses also this environment variable
2907 for its own purposes and extended its semantics. Refer to the mod_perl
2908 documentation for more information. Also, spawned threads do the
2909 equivalent of setting this variable to the value 1.)
2911 If, at the end of a run you get the message I<N scalars leaked>, you can
2912 recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses
2913 of all those leaked SVs to be dumped along with details as to where each
2914 SV was originally allocated. This information is also displayed by
2915 Devel::Peek. Note that the extra details recorded with each SV increases
2916 memory usage, so it shouldn't be used in production environments. It also
2917 converts C<new_SV()> from a macro into a real function, so you can use
2918 your favourite debugger to discover where those pesky SVs were allocated.
2920 If you see that you're leaking memory at runtime, but neither valgrind
2921 nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
2922 leaking SVs that are still reachable and will be properly cleaned up
2923 during destruction of the interpreter. In such cases, using the C<-Dm>
2924 switch can point you to the source of the leak. If the executable was
2925 built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations
2926 in addition to memory allocations. Each SV allocation has a distinct
2927 serial number that will be written on creation and destruction of the SV.
2928 So if you're executing the leaking code in a loop, you need to look for
2929 SVs that are created, but never destroyed between each cycle. If such an
2930 SV is found, set a conditional breakpoint within C<new_SV()> and make it
2931 break only when C<PL_sv_serial> is equal to the serial number of the
2932 leaking SV. Then you will catch the interpreter in exactly the state
2933 where the leaking SV is allocated, which is sufficient in many cases to
2934 find the source of the leak.
2936 As C<-Dm> is using the PerlIO layer for output, it will by itself
2937 allocate quite a bunch of SVs, which are hidden to avoid recursion.
2938 You can bypass the PerlIO layer if you use the SV logging provided
2939 by C<-DPERL_MEM_LOG> instead.
2943 If compiled with C<-DPERL_MEM_LOG>, all Newx() and Renew() allocations
2944 and Safefree() in the Perl core go through logging functions, which is
2945 handy for breakpoint setting. If also compiled with C<-DPERL_MEM_LOG_STDERR>,
2946 the allocations and frees are logged to STDERR (or more precisely, to the
2947 file descriptor 2) in these logging functions, with the calling source code
2948 file and line number (and C function name, if supported by the C compiler).
2950 This logging is somewhat similar to C<-Dm> but independent of C<-DDEBUGGING>,
2951 and at a higher level (the C<-Dm> is directly at the point of C<malloc()>,
2952 while the C<PERL_MEM_LOG> is at the level of C<New()>).
2954 In addition to memory allocations, SV allocations will be logged, just as
2955 with C<-Dm>. However, since the logging doesn't use PerlIO, all SV allocations
2956 are logged and no extra SV allocations are introduced by enabling the logging.
2957 If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for each SV
2958 allocation is also logged.
2960 You can control the logging from your environment if you compile with
2961 C<-DPERL_MEM_LOG_ENV>. Then you need to explicitly set C<PERL_MEM_LOG> and/or
2962 C<PERL_SV_LOG> to a non-zero value to enable logging of memory and/or SV
2967 Depending on your platform there are various of profiling Perl.
2969 There are two commonly used techniques of profiling executables:
2970 I<statistical time-sampling> and I<basic-block counting>.
2972 The first method takes periodically samples of the CPU program
2973 counter, and since the program counter can be correlated with the code
2974 generated for functions, we get a statistical view of in which
2975 functions the program is spending its time. The caveats are that very
2976 small/fast functions have lower probability of showing up in the
2977 profile, and that periodically interrupting the program (this is
2978 usually done rather frequently, in the scale of milliseconds) imposes
2979 an additional overhead that may skew the results. The first problem
2980 can be alleviated by running the code for longer (in general this is a
2981 good idea for profiling), the second problem is usually kept in guard
2982 by the profiling tools themselves.
2984 The second method divides up the generated code into I<basic blocks>.
2985 Basic blocks are sections of code that are entered only in the
2986 beginning and exited only at the end. For example, a conditional jump
2987 starts a basic block. Basic block profiling usually works by
2988 I<instrumenting> the code by adding I<enter basic block #nnnn>
2989 book-keeping code to the generated code. During the execution of the
2990 code the basic block counters are then updated appropriately. The
2991 caveat is that the added extra code can skew the results: again, the
2992 profiling tools usually try to factor their own effects out of the
2995 =head2 Gprof Profiling
2997 gprof is a profiling tool available in many UNIX platforms,
2998 it uses F<statistical time-sampling>.
3000 You can build a profiled version of perl called "perl.gprof" by
3001 invoking the make target "perl.gprof" (What is required is that Perl
3002 must be compiled using the C<-pg> flag, you may need to re-Configure).
3003 Running the profiled version of Perl will create an output file called
3004 F<gmon.out> is created which contains the profiling data collected
3005 during the execution.
3007 The gprof tool can then display the collected data in various ways.
3008 Usually gprof understands the following options:
3014 Suppress statically defined functions from the profile.
3018 Suppress the verbose descriptions in the profile.
3022 Exclude the given routine and its descendants from the profile.
3026 Display only the given routine and its descendants in the profile.
3030 Generate a summary file called F<gmon.sum> which then may be given
3031 to subsequent gprof runs to accumulate data over several runs.
3035 Display routines that have zero usage.
3039 For more detailed explanation of the available commands and output
3040 formats, see your own local documentation of gprof.
3044 $ sh Configure -des -Dusedevel -Doptimize='-g' -Accflags='-pg' -Aldflags='-pg' && make
3045 $ ./perl someprog # creates gmon.out in current directory
3049 =head2 GCC gcov Profiling
3051 Starting from GCC 3.0 I<basic block profiling> is officially available
3054 You can build a profiled version of perl called F<perl.gcov> by
3055 invoking the make target "perl.gcov" (what is required that Perl must
3056 be compiled using gcc with the flags C<-fprofile-arcs
3057 -ftest-coverage>, you may need to re-Configure).
3059 Running the profiled version of Perl will cause profile output to be
3060 generated. For each source file an accompanying ".da" file will be
3063 To display the results you use the "gcov" utility (which should
3064 be installed if you have gcc 3.0 or newer installed). F<gcov> is
3065 run on source code files, like this
3069 which will cause F<sv.c.gcov> to be created. The F<.gcov> files
3070 contain the source code annotated with relative frequencies of
3071 execution indicated by "#" markers.
3073 Useful options of F<gcov> include C<-b> which will summarise the
3074 basic block, branch, and function call coverage, and C<-c> which
3075 instead of relative frequencies will use the actual counts. For
3076 more information on the use of F<gcov> and basic block profiling
3077 with gcc, see the latest GNU CC manual, as of GCC 3.0 see
3079 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
3081 and its section titled "8. gcov: a Test Coverage Program"
3083 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
3087 $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \
3088 -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov
3089 $ rm -f regexec.c.gcov regexec.gcda
3092 $ view regexec.c.gcov
3094 =head2 Pixie Profiling
3096 Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
3097 UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
3098 I<basic-block counting>.
3100 You can build a profiled version of perl called F<perl.pixie> by
3101 invoking the make target "perl.pixie" (what is required is that Perl
3102 must be compiled using the C<-g> flag, you may need to re-Configure).
3104 In Tru64 a file called F<perl.Addrs> will also be silently created,
3105 this file contains the addresses of the basic blocks. Running the
3106 profiled version of Perl will create a new file called "perl.Counts"
3107 which contains the counts for the basic block for that particular
3110 To display the results you use the F<prof> utility. The exact
3111 incantation depends on your operating system, "prof perl.Counts" in
3112 IRIX, and "prof -pixie -all -L. perl" in Tru64.
3114 In IRIX the following prof options are available:
3120 Reports the most heavily used lines in descending order of use.
3121 Useful for finding the hotspot lines.
3125 Groups lines by procedure, with procedures sorted in descending order of use.
3126 Within a procedure, lines are listed in source order.
3127 Useful for finding the hotspots of procedures.
3131 In Tru64 the following options are available:
3137 Procedures sorted in descending order by the number of cycles executed
3138 in each procedure. Useful for finding the hotspot procedures.
3139 (This is the default option.)
3143 Lines sorted in descending order by the number of cycles executed in
3144 each line. Useful for finding the hotspot lines.
3146 =item -i[nvocations]
3148 The called procedures are sorted in descending order by number of calls
3149 made to the procedures. Useful for finding the most used procedures.
3153 Grouped by procedure, sorted by cycles executed per procedure.
3154 Useful for finding the hotspots of procedures.
3158 The compiler emitted code for these lines, but the code was unexecuted.
3162 Unexecuted procedures.
3166 For further information, see your system's manual pages for pixie and prof.
3168 =head2 Miscellaneous tricks
3174 Those debugging perl with the DDD frontend over gdb may find the
3177 You can extend the data conversion shortcuts menu, so for example you
3178 can display an SV's IV value with one click, without doing any typing.
3179 To do that simply edit ~/.ddd/init file and add after:
3181 ! Display shortcuts.
3182 Ddd*gdbDisplayShortcuts: \
3183 /t () // Convert to Bin\n\
3184 /d () // Convert to Dec\n\
3185 /x () // Convert to Hex\n\
3186 /o () // Convert to Oct(\n\
3188 the following two lines:
3190 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
3191 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
3193 so now you can do ivx and pvx lookups or you can plug there the
3194 sv_peek "conversion":
3196 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
3198 (The my_perl is for threaded builds.)
3199 Just remember that every line, but the last one, should end with \n\
3201 Alternatively edit the init file interactively via:
3202 3rd mouse button -> New Display -> Edit Menu
3204 Note: you can define up to 20 conversion shortcuts in the gdb
3209 If you see in a debugger a memory area mysteriously full of 0xABABABAB
3210 or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros,
3215 Under ithreads the optree is read only. If you want to enforce this, to check
3216 for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to
3217 enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code
3218 that allocates op memory via C<mmap>, and sets it read-only at run time.
3219 Any write access to an op results in a C<SIGBUS> and abort.
3221 This code is intended for development only, and may not be portable even to
3222 all Unix variants. Also, it is an 80% solution, in that it isn't able to make
3223 all ops read only. Specifically it
3229 Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated
3230 later via C<require> or C<eval> will be re-write
3234 Turns an entire slab of ops read-write if the refcount of any op in the slab
3235 needs to be decreased.
3239 Turns an entire slab of ops read-write if any op from the slab is freed.
3243 It's not possible to turn the slabs to read-only after an action requiring
3244 read-write access, as either can happen during op tree building time, so
3245 there may still be legitimate write access.
3247 However, as an 80% solution it is still effective, as currently it catches
3248 a write access during the generation of F<Config.pm>, which means that we
3249 can't yet build F<perl> with this enabled.
3256 We've had a brief look around the Perl source, how to maintain quality
3257 of the source code, an overview of the stages F<perl> goes through
3258 when it's running your code, how to use debuggers to poke at the Perl
3259 guts, and finally how to analyse the execution of Perl. We took a very
3260 simple problem and demonstrated how to solve it fully - with
3261 documentation, regression tests, and finally a patch for submission to
3262 p5p. Finally, we talked about how to use external tools to debug and
3265 I'd now suggest you read over those references again, and then, as soon
3266 as possible, get your hands dirty. The best way to learn is by doing,
3273 Subscribe to perl5-porters, follow the patches and try and understand
3274 them; don't be afraid to ask if there's a portion you're not clear on -
3275 who knows, you may unearth a bug in the patch...
3279 Keep up to date with the bleeding edge Perl distributions and get
3280 familiar with the changes. Try and get an idea of what areas people are
3281 working on and the changes they're making.
3285 Do read the README associated with your operating system, e.g. README.aix
3286 on the IBM AIX OS. Don't hesitate to supply patches to that README if
3287 you find anything missing or changed over a new OS release.
3291 Find an area of Perl that seems interesting to you, and see if you can
3292 work out how it works. Scan through the source, and step over it in the
3293 debugger. Play, poke, investigate, fiddle! You'll probably get to
3294 understand not just your chosen area but a much wider range of F<perl>'s
3295 activity as well, and probably sooner than you'd think.
3301 =item I<The Road goes ever on and on, down from the door where it began.>
3305 If you can do these things, you've started on the long road to Perl porting.
3306 Thanks for wanting to help make Perl better - and happy hacking!
3308 =head2 Metaphoric Quotations
3310 If you recognized the quote about the Road above, you're in luck.
3312 Most software projects begin each file with a literal description of each
3313 file's purpose. Perl instead begins each with a literary allusion to that
3316 Like chapters in many books, all top-level Perl source files (along with a
3317 few others here and there) begin with an epigramic inscription that alludes,
3318 indirectly and metaphorically, to the material you're about to read.
3320 Quotations are taken from writings of J.R.R Tolkien pertaining to his
3321 Legendarium, almost always from I<The Lord of the Rings>. Chapters and
3322 page numbers are given using the following editions:
3328 I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary
3329 edition of 2007 was used, published in the UK by Harper Collins Publishers
3330 and in the US by the Houghton Mifflin Company.
3334 I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover,
3335 50th-anniversary edition of 2004 was used, published in the UK by Harper
3336 Collins Publishers and in the US by the Houghton Mifflin Company.
3340 I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his
3341 son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes
3342 in Christopher's mammoth I<History of Middle Earth>. Page numbers derive
3343 from the hardcover edition, first published in 1983 by George Allen &
3344 Unwin; no page numbers changed for the special 3-volume omnibus edition of
3345 2002 or the various trade-paper editions, all again now by Harper Collins
3346 or Houghton Mifflin.
3350 Other JRRT books fair game for quotes would thus include I<The Adventures of
3351 Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of
3352 the Children of Hurin>, all but the first posthumously assembled by CJRT.
3353 But I<The Lord of the Rings> itself is perfectly fine and probably best to
3354 quote from, provided you can find a suitable quote there.
3356 So if you were to supply a new, complete, top-level source file to add to
3357 Perl, you should conform to this peculiar practice by yourself selecting an
3358 appropriate quotation from Tolkien, retaining the original spelling and
3359 punctuation and using the same format the rest of the quotes are in.
3360 Indirect and oblique is just fine; remember, it's a metaphor, so being meta
3361 is, after all, what it's for.
3365 This document was written by Nathan Torkington, and is maintained by
3366 the perl5-porters mailing list.