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 is the pumpking for the 5.8 release, and
42 Hugo van der Sanden will be the pumpking for the 5.10 release.
44 In addition, various people are pumpkings for different things. For
45 instance, Andy Dougherty and Jarkko Hietaniemi share the I<Configure>
48 Larry sees Perl development along the lines of the US government:
49 there's the Legislature (the porters), the Executive branch (the
50 pumpkings), and the Supreme Court (Larry). The legislature can
51 discuss and submit patches to the executive branch all they like, but
52 the executive branch is free to veto them. Rarely, the Supreme Court
53 will side with the executive branch over the legislature, or the
54 legislature over the executive branch. Mostly, however, the
55 legislature and the executive branch are supposed to get along and
56 work out their differences without impeachment or court cases.
58 You might sometimes see reference to Rule 1 and Rule 2. Larry's power
59 as Supreme Court is expressed in The Rules:
65 Larry is always by definition right about how Perl should behave.
66 This means he has final veto power on the core functionality.
70 Larry is allowed to change his mind about any matter at a later date,
71 regardless of whether he previously invoked Rule 1.
75 Got that? Larry is always right, even when he was wrong. It's rare
76 to see either Rule exercised, but they are often alluded to.
78 New features and extensions to the language are contentious, because
79 the criteria used by the pumpkings, Larry, and other porters to decide
80 which features should be implemented and incorporated are not codified
81 in a few small design goals as with some other languages. Instead,
82 the heuristics are flexible and often difficult to fathom. Here is
83 one person's list, roughly in decreasing order of importance, of
84 heuristics that new features have to be weighed against:
88 =item Does concept match the general goals of Perl?
90 These haven't been written anywhere in stone, but one approximation
93 1. Keep it fast, simple, and useful.
94 2. Keep features/concepts as orthogonal as possible.
95 3. No arbitrary limits (platforms, data sizes, cultures).
96 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
97 5. Either assimilate new technologies, or build bridges to them.
99 =item Where is the implementation?
101 All the talk in the world is useless without an implementation. In
102 almost every case, the person or people who argue for a new feature
103 will be expected to be the ones who implement it. Porters capable
104 of coding new features have their own agendas, and are not available
105 to implement your (possibly good) idea.
107 =item Backwards compatibility
109 It's a cardinal sin to break existing Perl programs. New warnings are
110 contentious--some say that a program that emits warnings is not
111 broken, while others say it is. Adding keywords has the potential to
112 break programs, changing the meaning of existing token sequences or
113 functions might break programs.
115 =item Could it be a module instead?
117 Perl 5 has extension mechanisms, modules and XS, specifically to avoid
118 the need to keep changing the Perl interpreter. You can write modules
119 that export functions, you can give those functions prototypes so they
120 can be called like built-in functions, you can even write XS code to
121 mess with the runtime data structures of the Perl interpreter if you
122 want to implement really complicated things. If it can be done in a
123 module instead of in the core, it's highly unlikely to be added.
125 =item Is the feature generic enough?
127 Is this something that only the submitter wants added to the language,
128 or would it be broadly useful? Sometimes, instead of adding a feature
129 with a tight focus, the porters might decide to wait until someone
130 implements the more generalized feature. For instance, instead of
131 implementing a ``delayed evaluation'' feature, the porters are waiting
132 for a macro system that would permit delayed evaluation and much more.
134 =item Does it potentially introduce new bugs?
136 Radical rewrites of large chunks of the Perl interpreter have the
137 potential to introduce new bugs. The smaller and more localized the
140 =item Does it preclude other desirable features?
142 A patch is likely to be rejected if it closes off future avenues of
143 development. For instance, a patch that placed a true and final
144 interpretation on prototypes is likely to be rejected because there
145 are still options for the future of prototypes that haven't been
148 =item Is the implementation robust?
150 Good patches (tight code, complete, correct) stand more chance of
151 going in. Sloppy or incorrect patches might be placed on the back
152 burner until the pumpking has time to fix, or might be discarded
153 altogether without further notice.
155 =item Is the implementation generic enough to be portable?
157 The worst patches make use of a system-specific features. It's highly
158 unlikely that nonportable additions to the Perl language will be
161 =item Is the implementation tested?
163 Patches which change behaviour (fixing bugs or introducing new features)
164 must include regression tests to verify that everything works as expected.
165 Without tests provided by the original author, how can anyone else changing
166 perl in the future be sure that they haven't unwittingly broken the behaviour
167 the patch implements? And without tests, how can the patch's author be
168 confident that his/her hard work put into the patch won't be accidentally
169 thrown away by someone in the future?
171 =item Is there enough documentation?
173 Patches without documentation are probably ill-thought out or
174 incomplete. Nothing can be added without documentation, so submitting
175 a patch for the appropriate manpages as well as the source code is
178 =item Is there another way to do it?
180 Larry said ``Although the Perl Slogan is I<There's More Than One Way
181 to Do It>, I hesitate to make 10 ways to do something''. This is a
182 tricky heuristic to navigate, though--one man's essential addition is
183 another man's pointless cruft.
185 =item Does it create too much work?
187 Work for the pumpking, work for Perl programmers, work for module
188 authors, ... Perl is supposed to be easy.
190 =item Patches speak louder than words
192 Working code is always preferred to pie-in-the-sky ideas. A patch to
193 add a feature stands a much higher chance of making it to the language
194 than does a random feature request, no matter how fervently argued the
195 request might be. This ties into ``Will it be useful?'', as the fact
196 that someone took the time to make the patch demonstrates a strong
197 desire for the feature.
201 If you're on the list, you might hear the word ``core'' bandied
202 around. It refers to the standard distribution. ``Hacking on the
203 core'' means you're changing the C source code to the Perl
204 interpreter. ``A core module'' is one that ships with Perl.
206 =head2 Keeping in sync
208 The source code to the Perl interpreter, in its different versions, is
209 kept in a repository managed by a revision control system ( which is
210 currently the Perforce program, see http://perforce.com/ ). The
211 pumpkings and a few others have access to the repository to check in
212 changes. Periodically the pumpking for the development version of Perl
213 will release a new version, so the rest of the porters can see what's
214 changed. The current state of the main trunk of repository, and patches
215 that describe the individual changes that have happened since the last
216 public release are available at this location:
218 http://public.activestate.com/gsar/APC/
219 ftp://ftp.linux.activestate.com/pub/staff/gsar/APC/
221 If you're looking for a particular change, or a change that affected
222 a particular set of files, you may find the B<Perl Repository Browser>
225 http://public.activestate.com/cgi-bin/perlbrowse
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 Keeping in sync with the most recent branch can be done in several ways,
243 but the most convenient and reliable way is using B<rsync>, available at
244 ftp://rsync.samba.org/pub/rsync/ . (You can also get the most recent
247 If you choose to keep in sync using rsync, there are two approaches
252 =item rsync'ing the source tree
254 Presuming you are in the directory where your perl source resides
255 and you have rsync installed and available, you can `upgrade' to
258 # rsync -avz rsync://ftp.linux.activestate.com/perl-current/ .
260 This takes care of updating every single item in the source tree to
261 the latest applied patch level, creating files that are new (to your
262 distribution) and setting date/time stamps of existing files to
263 reflect the bleadperl status.
265 Note that this will not delete any files that were in '.' before
266 the rsync. Once you are sure that the rsync is running correctly,
267 run it with the --delete and the --dry-run options like this:
269 # rsync -avz --delete --dry-run rsync://ftp.linux.activestate.com/perl-current/ .
271 This will I<simulate> an rsync run that also deletes files not
272 present in the bleadperl master copy. Observe the results from
273 this run closely. If you are sure that the actual run would delete
274 no files precious to you, you could remove the '--dry-run' option.
276 You can than check what patch was the latest that was applied by
277 looking in the file B<.patch>, which will show the number of the
280 If you have more than one machine to keep in sync, and not all of
281 them have access to the WAN (so you are not able to rsync all the
282 source trees to the real source), there are some ways to get around
287 =item Using rsync over the LAN
289 Set up a local rsync server which makes the rsynced source tree
290 available to the LAN and sync the other machines against this
293 From http://rsync.samba.org/README.html :
295 "Rsync uses rsh or ssh for communication. It does not need to be
296 setuid and requires no special privileges for installation. It
297 does not require an inetd entry or a daemon. You must, however,
298 have a working rsh or ssh system. Using ssh is recommended for
299 its security features."
301 =item Using pushing over the NFS
303 Having the other systems mounted over the NFS, you can take an
304 active pushing approach by checking the just updated tree against
305 the other not-yet synced trees. An example would be
314 $1 => [ (stat $1)[2, 7, 9] ]; # mode, size, mtime
317 my %remote = map { $_ => "/$_/pro/3gl/CPAN/perl-5.7.1" } qw(host1 host2);
319 foreach my $host (keys %remote) {
320 unless (-d $remote{$host}) {
321 print STDERR "Cannot Xsync for host $host\n";
324 foreach my $file (keys %MF) {
325 my $rfile = "$remote{$host}/$file";
326 my ($mode, $size, $mtime) = (stat $rfile)[2, 7, 9];
327 defined $size or ($mode, $size, $mtime) = (0, 0, 0);
328 $size == $MF{$file}[1] && $mtime == $MF{$file}[2] and next;
329 printf "%4s %-34s %8d %9d %8d %9d\n",
330 $host, $file, $MF{$file}[1], $MF{$file}[2], $size, $mtime;
332 copy ($file, $rfile);
333 utime time, $MF{$file}[2], $rfile;
334 chmod $MF{$file}[0], $rfile;
338 though this is not perfect. It could be improved with checking
339 file checksums before updating. Not all NFS systems support
340 reliable utime support (when used over the NFS).
344 =item rsync'ing the patches
346 The source tree is maintained by the pumpking who applies patches to
347 the files in the tree. These patches are either created by the
348 pumpking himself using C<diff -c> after updating the file manually or
349 by applying patches sent in by posters on the perl5-porters list.
350 These patches are also saved and rsync'able, so you can apply them
351 yourself to the source files.
353 Presuming you are in a directory where your patches reside, you can
354 get them in sync with
356 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
358 This makes sure the latest available patch is downloaded to your
361 It's then up to you to apply these patches, using something like
363 # last=`ls -t *.gz | sed q`
364 # rsync -avz rsync://ftp.linux.activestate.com/perl-current-diffs/ .
365 # find . -name '*.gz' -newer $last -exec gzcat {} \; >blead.patch
367 # patch -p1 -N <../perl-current-diffs/blead.patch
369 or, since this is only a hint towards how it works, use CPAN-patchaperl
370 from Andreas König to have better control over the patching process.
374 =head2 Why rsync the source tree
378 =item It's easier to rsync the source tree
380 Since you don't have to apply the patches yourself, you are sure all
381 files in the source tree are in the right state.
383 =item It's more reliable
385 While both the rsync-able source and patch areas are automatically
386 updated every few minutes, keep in mind that applying patches may
387 sometimes mean careful hand-holding, especially if your version of
388 the C<patch> program does not understand how to deal with new files,
389 files with 8-bit characters, or files without trailing newlines.
393 =head2 Why rsync the patches
397 =item It's easier to rsync the patches
399 If you have more than one machine that you want to keep in track with
400 bleadperl, it's easier to rsync the patches only once and then apply
401 them to all the source trees on the different machines.
403 In case you try to keep in pace on 5 different machines, for which
404 only one of them has access to the WAN, rsync'ing all the source
405 trees should than be done 5 times over the NFS. Having
406 rsync'ed the patches only once, I can apply them to all the source
407 trees automatically. Need you say more ;-)
409 =item It's a good reference
411 If you do not only like to have the most recent development branch,
412 but also like to B<fix> bugs, or extend features, you want to dive
413 into the sources. If you are a seasoned perl core diver, you don't
414 need no manuals, tips, roadmaps, perlguts.pod or other aids to find
415 your way around. But if you are a starter, the patches may help you
416 in finding where you should start and how to change the bits that
419 The file B<Changes> is updated on occasions the pumpking sees as his
420 own little sync points. On those occasions, he releases a tar-ball of
421 the current source tree (i.e. perl@7582.tar.gz), which will be an
422 excellent point to start with when choosing to use the 'rsync the
423 patches' scheme. Starting with perl@7582, which means a set of source
424 files on which the latest applied patch is number 7582, you apply all
425 succeeding patches available from then on (7583, 7584, ...).
427 You can use the patches later as a kind of search archive.
431 =item Finding a start point
433 If you want to fix/change the behaviour of function/feature Foo, just
434 scan the patches for patches that mention Foo either in the subject,
435 the comments, or the body of the fix. A good chance the patch shows
436 you the files that are affected by that patch which are very likely
437 to be the starting point of your journey into the guts of perl.
439 =item Finding how to fix a bug
441 If you've found I<where> the function/feature Foo misbehaves, but you
442 don't know how to fix it (but you do know the change you want to
443 make), you can, again, peruse the patches for similar changes and
444 look how others apply the fix.
446 =item Finding the source of misbehaviour
448 When you keep in sync with bleadperl, the pumpking would love to
449 I<see> that the community efforts really work. So after each of his
450 sync points, you are to 'make test' to check if everything is still
451 in working order. If it is, you do 'make ok', which will send an OK
452 report to perlbug@perl.org. (If you do not have access to a mailer
453 from the system you just finished successfully 'make test', you can
454 do 'make okfile', which creates the file C<perl.ok>, which you can
455 than take to your favourite mailer and mail yourself).
457 But of course, as always, things will not always lead to a success
458 path, and one or more test do not pass the 'make test'. Before
459 sending in a bug report (using 'make nok' or 'make nokfile'), check
460 the mailing list if someone else has reported the bug already and if
461 so, confirm it by replying to that message. If not, you might want to
462 trace the source of that misbehaviour B<before> sending in the bug,
463 which will help all the other porters in finding the solution.
465 Here the saved patches come in very handy. You can check the list of
466 patches to see which patch changed what file and what change caused
467 the misbehaviour. If you note that in the bug report, it saves the
468 one trying to solve it, looking for that point.
472 If searching the patches is too bothersome, you might consider using
473 perl's bugtron to find more information about discussions and
474 ramblings on posted bugs.
476 If you want to get the best of both worlds, rsync both the source
477 tree for convenience, reliability and ease and rsync the patches
482 =head2 Working with the source
484 Because you cannot use the Perforce client, you cannot easily generate
485 diffs against the repository, nor will merges occur when you update
486 via rsync. If you edit a file locally and then rsync against the
487 latest source, changes made in the remote copy will I<overwrite> your
490 The best way to deal with this is to maintain a tree of symlinks to
491 the rsync'd source. Then, when you want to edit a file, you remove
492 the symlink, copy the real file into the other tree, and edit it. You
493 can then diff your edited file against the original to generate a
494 patch, and you can safely update the original tree.
496 Perl's F<Configure> script can generate this tree of symlinks for you.
497 The following example assumes that you have used rsync to pull a copy
498 of the Perl source into the F<perl-rsync> directory. In the directory
499 above that one, you can execute the following commands:
503 ../perl-rsync/Configure -Dmksymlinks -Dusedevel -D"optimize=-g"
505 This will start the Perl configuration process. After a few prompts,
506 you should see something like this:
508 Symbolic links are supported.
510 Checking how to test for symbolic links...
511 Your builtin 'test -h' may be broken.
512 Trying external '/usr/bin/test -h'.
513 You can test for symbolic links with '/usr/bin/test -h'.
515 Creating the symbolic links...
516 (First creating the subdirectories...)
517 (Then creating the symlinks...)
519 The specifics may vary based on your operating system, of course.
520 After you see this, you can abort the F<Configure> script, and you
521 will see that the directory you are in has a tree of symlinks to the
522 F<perl-rsync> directories and files.
524 If you plan to do a lot of work with the Perl source, here are some
525 Bourne shell script functions that can make your life easier:
538 if [ -L $1.orig ]; then
544 Replace "vi" with your favorite flavor of editor.
546 Here is another function which will quickly generate a patch for the
547 files which have been edited in your symlink tree:
551 for f in `find . -name '*.orig' | sed s,^\./,,`
553 case `echo $f | sed 's,.orig$,,;s,.*\.,,'` in
555 pod) diffopts='-F^=' ;;
558 diff -du $diffopts $f `echo $f | sed 's,.orig$,,'`
562 This function produces patches which include enough context to make
563 your changes obvious. This makes it easier for the Perl pumpking(s)
564 to review them when you send them to the perl5-porters list, and that
565 means they're more likely to get applied.
567 This function assumed a GNU diff, and may require some tweaking for
570 =head2 Perlbug administration
572 There is a single remote administrative interface for modifying bug status,
573 category, open issues etc. using the B<RT> I<bugtracker> system, maintained
574 by I<Robert Spier>. Become an administrator, and close any bugs you can get
575 your sticky mitts on:
579 The bugtracker mechanism for B<perl5> bugs in particular is at:
581 http://bugs6.perl.org/perlbug
583 To email the bug system administrators:
585 "perlbug-admin" <perlbug-admin@perl.org>
588 =head2 Submitting patches
590 Always submit patches to I<perl5-porters@perl.org>. If you're
591 patching a core module and there's an author listed, send the author a
592 copy (see L<Patching a core module>). This lets other porters review
593 your patch, which catches a surprising number of errors in patches.
594 Either use the diff program (available in source code form from
595 ftp://ftp.gnu.org/pub/gnu/ , or use Johan Vromans' I<makepatch>
596 (available from I<CPAN/authors/id/JV/>). Unified diffs are preferred,
597 but context diffs are accepted. Do not send RCS-style diffs or diffs
598 without context lines. More information is given in the
599 I<Porting/patching.pod> file in the Perl source distribution. Please
600 patch against the latest B<development> version (e.g., if you're
601 fixing a bug in the 5.005 track, patch against the latest 5.005_5x
602 version). Only patches that survive the heat of the development
603 branch get applied to maintenance versions.
605 Your patch should update the documentation and test suite. See
608 To report a bug in Perl, use the program I<perlbug> which comes with
609 Perl (if you can't get Perl to work, send mail to the address
610 I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through
611 I<perlbug> feeds into the automated bug-tracking system, access to
612 which is provided through the web at http://bugs.perl.org/ . It
613 often pays to check the archives of the perl5-porters mailing list to
614 see whether the bug you're reporting has been reported before, and if
615 so whether it was considered a bug. See above for the location of
616 the searchable archives.
618 The CPAN testers ( http://testers.cpan.org/ ) are a group of
619 volunteers who test CPAN modules on a variety of platforms. Perl
620 Smokers ( http://archives.develooper.com/daily-build@perl.org/ )
621 automatically tests Perl source releases on platforms with various
622 configurations. Both efforts welcome volunteers.
624 It's a good idea to read and lurk for a while before chipping in.
625 That way you'll get to see the dynamic of the conversations, learn the
626 personalities of the players, and hopefully be better prepared to make
627 a useful contribution when do you speak up.
629 If after all this you still think you want to join the perl5-porters
630 mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To
631 unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>.
633 To hack on the Perl guts, you'll need to read the following things:
639 This is of paramount importance, since it's the documentation of what
640 goes where in the Perl source. Read it over a couple of times and it
641 might start to make sense - don't worry if it doesn't yet, because the
642 best way to study it is to read it in conjunction with poking at Perl
643 source, and we'll do that later on.
645 You might also want to look at Gisle Aas's illustrated perlguts -
646 there's no guarantee that this will be absolutely up-to-date with the
647 latest documentation in the Perl core, but the fundamentals will be
648 right. ( http://gisle.aas.no/perl/illguts/ )
650 =item L<perlxstut> and L<perlxs>
652 A working knowledge of XSUB programming is incredibly useful for core
653 hacking; XSUBs use techniques drawn from the PP code, the portion of the
654 guts that actually executes a Perl program. It's a lot gentler to learn
655 those techniques from simple examples and explanation than from the core
660 The documentation for the Perl API explains what some of the internal
661 functions do, as well as the many macros used in the source.
663 =item F<Porting/pumpkin.pod>
665 This is a collection of words of wisdom for a Perl porter; some of it is
666 only useful to the pumpkin holder, but most of it applies to anyone
667 wanting to go about Perl development.
669 =item The perl5-porters FAQ
671 This should be available from http://simon-cozens.org/writings/p5p-faq ;
672 alternatively, you can get the FAQ emailed to you by sending mail to
673 C<perl5-porters-faq@perl.org>. It contains hints on reading perl5-porters,
674 information on how perl5-porters works and how Perl development in general
679 =head2 Finding Your Way Around
681 Perl maintenance can be split into a number of areas, and certain people
682 (pumpkins) will have responsibility for each area. These areas sometimes
683 correspond to files or directories in the source kit. Among the areas are:
689 Modules shipped as part of the Perl core live in the F<lib/> and F<ext/>
690 subdirectories: F<lib/> is for the pure-Perl modules, and F<ext/>
691 contains the core XS modules.
695 There are tests for nearly all the modules, built-ins and major bits
696 of functionality. Test files all have a .t suffix. Module tests live
697 in the F<lib/> and F<ext/> directories next to the module being
698 tested. Others live in F<t/>. See L<Writing a test>
702 Documentation maintenance includes looking after everything in the
703 F<pod/> directory, (as well as contributing new documentation) and
704 the documentation to the modules in core.
708 The configure process is the way we make Perl portable across the
709 myriad of operating systems it supports. Responsibility for the
710 configure, build and installation process, as well as the overall
711 portability of the core code rests with the configure pumpkin - others
712 help out with individual operating systems.
714 The files involved are the operating system directories, (F<win32/>,
715 F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h>
716 and F<Makefile>, as well as the metaconfig files which generate
717 F<Configure>. (metaconfig isn't included in the core distribution.)
721 And of course, there's the core of the Perl interpreter itself. Let's
722 have a look at that in a little more detail.
726 Before we leave looking at the layout, though, don't forget that
727 F<MANIFEST> contains not only the file names in the Perl distribution,
728 but short descriptions of what's in them, too. For an overview of the
729 important files, try this:
731 perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST
733 =head2 Elements of the interpreter
735 The work of the interpreter has two main stages: compiling the code
736 into the internal representation, or bytecode, and then executing it.
737 L<perlguts/Compiled code> explains exactly how the compilation stage
740 Here is a short breakdown of perl's operation:
746 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
747 This is very high-level code, enough to fit on a single screen, and it
748 resembles the code found in L<perlembed>; most of the real action takes
751 First, F<perlmain.c> allocates some memory and constructs a Perl
754 1 PERL_SYS_INIT3(&argc,&argv,&env);
756 3 if (!PL_do_undump) {
757 4 my_perl = perl_alloc();
760 7 perl_construct(my_perl);
761 8 PL_perl_destruct_level = 0;
764 Line 1 is a macro, and its definition is dependent on your operating
765 system. Line 3 references C<PL_do_undump>, a global variable - all
766 global variables in Perl start with C<PL_>. This tells you whether the
767 current running program was created with the C<-u> flag to perl and then
768 F<undump>, which means it's going to be false in any sane context.
770 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
771 interpreter. It's quite a simple function, and the guts of it looks like
774 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
776 Here you see an example of Perl's system abstraction, which we'll see
777 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
778 own C<malloc> as defined in F<malloc.c> if you selected that option at
781 Next, in line 7, we construct the interpreter; this sets up all the
782 special variables that Perl needs, the stacks, and so on.
784 Now we pass Perl the command line options, and tell it to go:
786 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
788 exitstatus = perl_run(my_perl);
792 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
793 in F<perl.c>, which processes the command line options, sets up any
794 statically linked XS modules, opens the program and calls C<yyparse> to
799 The aim of this stage is to take the Perl source, and turn it into an op
800 tree. We'll see what one of those looks like later. Strictly speaking,
801 there's three things going on here.
803 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
804 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
805 B<is> a YACC grammar for Perl!) The job of the parser is to take your
806 code and `understand' it, splitting it into sentences, deciding which
807 operands go with which operators and so on.
809 The parser is nobly assisted by the lexer, which chunks up your input
810 into tokens, and decides what type of thing each token is: a variable
811 name, an operator, a bareword, a subroutine, a core function, and so on.
812 The main point of entry to the lexer is C<yylex>, and that and its
813 associated routines can be found in F<toke.c>. Perl isn't much like
814 other computer languages; it's highly context sensitive at times, it can
815 be tricky to work out what sort of token something is, or where a token
816 ends. As such, there's a lot of interplay between the tokeniser and the
817 parser, which can get pretty frightening if you're not used to it.
819 As the parser understands a Perl program, it builds up a tree of
820 operations for the interpreter to perform during execution. The routines
821 which construct and link together the various operations are to be found
822 in F<op.c>, and will be examined later.
826 Now the parsing stage is complete, and the finished tree represents
827 the operations that the Perl interpreter needs to perform to execute our
828 program. Next, Perl does a dry run over the tree looking for
829 optimisations: constant expressions such as C<3 + 4> will be computed
830 now, and the optimizer will also see if any multiple operations can be
831 replaced with a single one. For instance, to fetch the variable C<$foo>,
832 instead of grabbing the glob C<*foo> and looking at the scalar
833 component, the optimizer fiddles the op tree to use a function which
834 directly looks up the scalar in question. The main optimizer is C<peep>
835 in F<op.c>, and many ops have their own optimizing functions.
839 Now we're finally ready to go: we have compiled Perl byte code, and all
840 that's left to do is run it. The actual execution is done by the
841 C<runops_standard> function in F<run.c>; more specifically, it's done by
842 these three innocent looking lines:
844 while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) {
848 You may be more comfortable with the Perl version of that:
850 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
852 Well, maybe not. Anyway, each op contains a function pointer, which
853 stipulates the function which will actually carry out the operation.
854 This function will return the next op in the sequence - this allows for
855 things like C<if> which choose the next op dynamically at run time.
856 The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
857 execution if required.
859 The actual functions called are known as PP code, and they're spread
860 between four files: F<pp_hot.c> contains the `hot' code, which is most
861 often used and highly optimized, F<pp_sys.c> contains all the
862 system-specific functions, F<pp_ctl.c> contains the functions which
863 implement control structures (C<if>, C<while> and the like) and F<pp.c>
864 contains everything else. These are, if you like, the C code for Perl's
865 built-in functions and operators.
869 =head2 Internal Variable Types
871 You should by now have had a look at L<perlguts>, which tells you about
872 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
875 These variables are used not only to represent Perl-space variables, but
876 also any constants in the code, as well as some structures completely
877 internal to Perl. The symbol table, for instance, is an ordinary Perl
878 hash. Your code is represented by an SV as it's read into the parser;
879 any program files you call are opened via ordinary Perl filehandles, and
882 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
883 Perl program. Let's see, for instance, how Perl treats the constant
886 % perl -MDevel::Peek -e 'Dump("hello")'
887 1 SV = PV(0xa041450) at 0xa04ecbc
889 3 FLAGS = (POK,READONLY,pPOK)
890 4 PV = 0xa0484e0 "hello"\0
894 Reading C<Devel::Peek> output takes a bit of practise, so let's go
895 through it line by line.
897 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
898 memory. SVs themselves are very simple structures, but they contain a
899 pointer to a more complex structure. In this case, it's a PV, a
900 structure which holds a string value, at location C<0xa041450>. Line 2
901 is the reference count; there are no other references to this data, so
904 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
905 read-only SV (because it's a constant) and the data is a PV internally.
906 Next we've got the contents of the string, starting at location
909 Line 5 gives us the current length of the string - note that this does
910 B<not> include the null terminator. Line 6 is not the length of the
911 string, but the length of the currently allocated buffer; as the string
912 grows, Perl automatically extends the available storage via a routine
915 You can get at any of these quantities from C very easily; just add
916 C<Sv> to the name of the field shown in the snippet, and you've got a
917 macro which will return the value: C<SvCUR(sv)> returns the current
918 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
919 C<SvPV(sv, len)> returns the string itself with its length, and so on.
920 More macros to manipulate these properties can be found in L<perlguts>.
922 Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c>
925 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len)
930 6 junk = SvPV_force(sv, tlen);
931 7 SvGROW(sv, tlen + len + 1);
934 10 Move(ptr,SvPVX(sv)+tlen,len,char);
936 12 *SvEND(sv) = '\0';
937 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
941 This is a function which adds a string, C<ptr>, of length C<len> onto
942 the end of the PV stored in C<sv>. The first thing we do in line 6 is
943 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
944 macro to force a PV. As a side effect, C<tlen> gets set to the current
945 value of the PV, and the PV itself is returned to C<junk>.
947 In line 7, we make sure that the SV will have enough room to accommodate
948 the old string, the new string and the null terminator. If C<LEN> isn't
949 big enough, C<SvGROW> will reallocate space for us.
951 Now, if C<junk> is the same as the string we're trying to add, we can
952 grab the string directly from the SV; C<SvPVX> is the address of the PV
955 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
956 memory around: we move the string C<ptr> to the end of the PV - that's
957 the start of the PV plus its current length. We're moving C<len> bytes
958 of type C<char>. After doing so, we need to tell Perl we've extended the
959 string, by altering C<CUR> to reflect the new length. C<SvEND> is a
960 macro which gives us the end of the string, so that needs to be a
963 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
964 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
965 want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware
966 version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags
967 and turns on POK. The final C<SvTAINT> is a macro which launders tainted
968 data if taint mode is turned on.
970 AVs and HVs are more complicated, but SVs are by far the most common
971 variable type being thrown around. Having seen something of how we
972 manipulate these, let's go on and look at how the op tree is
977 First, what is the op tree, anyway? The op tree is the parsed
978 representation of your program, as we saw in our section on parsing, and
979 it's the sequence of operations that Perl goes through to execute your
980 program, as we saw in L</Running>.
982 An op is a fundamental operation that Perl can perform: all the built-in
983 functions and operators are ops, and there are a series of ops which
984 deal with concepts the interpreter needs internally - entering and
985 leaving a block, ending a statement, fetching a variable, and so on.
987 The op tree is connected in two ways: you can imagine that there are two
988 "routes" through it, two orders in which you can traverse the tree.
989 First, parse order reflects how the parser understood the code, and
990 secondly, execution order tells perl what order to perform the
993 The easiest way to examine the op tree is to stop Perl after it has
994 finished parsing, and get it to dump out the tree. This is exactly what
995 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
996 and L<B::Debug|B::Debug> do.
998 Let's have a look at how Perl sees C<$a = $b + $c>:
1000 % perl -MO=Terse -e '$a=$b+$c'
1001 1 LISTOP (0x8179888) leave
1002 2 OP (0x81798b0) enter
1003 3 COP (0x8179850) nextstate
1004 4 BINOP (0x8179828) sassign
1005 5 BINOP (0x8179800) add [1]
1006 6 UNOP (0x81796e0) null [15]
1007 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
1008 8 UNOP (0x81797e0) null [15]
1009 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
1010 10 UNOP (0x816b4f0) null [15]
1011 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
1013 Let's start in the middle, at line 4. This is a BINOP, a binary
1014 operator, which is at location C<0x8179828>. The specific operator in
1015 question is C<sassign> - scalar assignment - and you can find the code
1016 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
1017 binary operator, it has two children: the add operator, providing the
1018 result of C<$b+$c>, is uppermost on line 5, and the left hand side is on
1021 Line 10 is the null op: this does exactly nothing. What is that doing
1022 there? If you see the null op, it's a sign that something has been
1023 optimized away after parsing. As we mentioned in L</Optimization>,
1024 the optimization stage sometimes converts two operations into one, for
1025 example when fetching a scalar variable. When this happens, instead of
1026 rewriting the op tree and cleaning up the dangling pointers, it's easier
1027 just to replace the redundant operation with the null op. Originally,
1028 the tree would have looked like this:
1030 10 SVOP (0x816b4f0) rv2sv [15]
1031 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
1033 That is, fetch the C<a> entry from the main symbol table, and then look
1034 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
1035 happens to do both these things.
1037 The right hand side, starting at line 5 is similar to what we've just
1038 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together
1041 Now, what's this about?
1043 1 LISTOP (0x8179888) leave
1044 2 OP (0x81798b0) enter
1045 3 COP (0x8179850) nextstate
1047 C<enter> and C<leave> are scoping ops, and their job is to perform any
1048 housekeeping every time you enter and leave a block: lexical variables
1049 are tidied up, unreferenced variables are destroyed, and so on. Every
1050 program will have those first three lines: C<leave> is a list, and its
1051 children are all the statements in the block. Statements are delimited
1052 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
1053 the ops to be performed for each statement being the children of
1054 C<nextstate>. C<enter> is a single op which functions as a marker.
1056 That's how Perl parsed the program, from top to bottom:
1069 However, it's impossible to B<perform> the operations in this order:
1070 you have to find the values of C<$b> and C<$c> before you add them
1071 together, for instance. So, the other thread that runs through the op
1072 tree is the execution order: each op has a field C<op_next> which points
1073 to the next op to be run, so following these pointers tells us how perl
1074 executes the code. We can traverse the tree in this order using
1075 the C<exec> option to C<B::Terse>:
1077 % perl -MO=Terse,exec -e '$a=$b+$c'
1078 1 OP (0x8179928) enter
1079 2 COP (0x81798c8) nextstate
1080 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
1081 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
1082 5 BINOP (0x8179878) add [1]
1083 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
1084 7 BINOP (0x81798a0) sassign
1085 8 LISTOP (0x8179900) leave
1087 This probably makes more sense for a human: enter a block, start a
1088 statement. Get the values of C<$b> and C<$c>, and add them together.
1089 Find C<$a>, and assign one to the other. Then leave.
1091 The way Perl builds up these op trees in the parsing process can be
1092 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
1093 piece we need to construct the tree for C<$a = $b + $c>
1095 1 term : term ASSIGNOP term
1096 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
1098 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1100 If you're not used to reading BNF grammars, this is how it works: You're
1101 fed certain things by the tokeniser, which generally end up in upper
1102 case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your
1103 code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are
1104 `terminal symbols', because you can't get any simpler than them.
1106 The grammar, lines one and three of the snippet above, tells you how to
1107 build up more complex forms. These complex forms, `non-terminal symbols'
1108 are generally placed in lower case. C<term> here is a non-terminal
1109 symbol, representing a single expression.
1111 The grammar gives you the following rule: you can make the thing on the
1112 left of the colon if you see all the things on the right in sequence.
1113 This is called a "reduction", and the aim of parsing is to completely
1114 reduce the input. There are several different ways you can perform a
1115 reduction, separated by vertical bars: so, C<term> followed by C<=>
1116 followed by C<term> makes a C<term>, and C<term> followed by C<+>
1117 followed by C<term> can also make a C<term>.
1119 So, if you see two terms with an C<=> or C<+>, between them, you can
1120 turn them into a single expression. When you do this, you execute the
1121 code in the block on the next line: if you see C<=>, you'll do the code
1122 in line 2. If you see C<+>, you'll do the code in line 4. It's this code
1123 which contributes to the op tree.
1126 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
1128 What this does is creates a new binary op, and feeds it a number of
1129 variables. The variables refer to the tokens: C<$1> is the first token in
1130 the input, C<$2> the second, and so on - think regular expression
1131 backreferences. C<$$> is the op returned from this reduction. So, we
1132 call C<newBINOP> to create a new binary operator. The first parameter to
1133 C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
1134 operator, so we want the type to be C<ADDOP>. We could specify this
1135 directly, but it's right there as the second token in the input, so we
1136 use C<$2>. The second parameter is the op's flags: 0 means `nothing
1137 special'. Then the things to add: the left and right hand side of our
1138 expression, in scalar context.
1142 When perl executes something like C<addop>, how does it pass on its
1143 results to the next op? The answer is, through the use of stacks. Perl
1144 has a number of stacks to store things it's currently working on, and
1145 we'll look at the three most important ones here.
1149 =item Argument stack
1151 Arguments are passed to PP code and returned from PP code using the
1152 argument stack, C<ST>. The typical way to handle arguments is to pop
1153 them off the stack, deal with them how you wish, and then push the result
1154 back onto the stack. This is how, for instance, the cosine operator
1159 value = Perl_cos(value);
1162 We'll see a more tricky example of this when we consider Perl's macros
1163 below. C<POPn> gives you the NV (floating point value) of the top SV on
1164 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push
1165 the result back as an NV. The C<X> in C<XPUSHn> means that the stack
1166 should be extended if necessary - it can't be necessary here, because we
1167 know there's room for one more item on the stack, since we've just
1168 removed one! The C<XPUSH*> macros at least guarantee safety.
1170 Alternatively, you can fiddle with the stack directly: C<SP> gives you
1171 the first element in your portion of the stack, and C<TOP*> gives you
1172 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
1173 negation of an integer:
1177 Just set the integer value of the top stack entry to its negation.
1179 Argument stack manipulation in the core is exactly the same as it is in
1180 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
1181 description of the macros used in stack manipulation.
1185 I say `your portion of the stack' above because PP code doesn't
1186 necessarily get the whole stack to itself: if your function calls
1187 another function, you'll only want to expose the arguments aimed for the
1188 called function, and not (necessarily) let it get at your own data. The
1189 way we do this is to have a `virtual' bottom-of-stack, exposed to each
1190 function. The mark stack keeps bookmarks to locations in the argument
1191 stack usable by each function. For instance, when dealing with a tied
1192 variable, (internally, something with `P' magic) Perl has to call
1193 methods for accesses to the tied variables. However, we need to separate
1194 the arguments exposed to the method to the argument exposed to the
1195 original function - the store or fetch or whatever it may be. Here's how
1196 the tied C<push> is implemented; see C<av_push> in F<av.c>:
1200 3 PUSHs(SvTIED_obj((SV*)av, mg));
1204 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1208 The lines which concern the mark stack are the first, fifth and last
1209 lines: they save away, restore and remove the current position of the
1212 Let's examine the whole implementation, for practice:
1216 Push the current state of the stack pointer onto the mark stack. This is
1217 so that when we've finished adding items to the argument stack, Perl
1218 knows how many things we've added recently.
1221 3 PUSHs(SvTIED_obj((SV*)av, mg));
1224 We're going to add two more items onto the argument stack: when you have
1225 a tied array, the C<PUSH> subroutine receives the object and the value
1226 to be pushed, and that's exactly what we have here - the tied object,
1227 retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
1231 Next we tell Perl to make the change to the global stack pointer: C<dSP>
1232 only gave us a local copy, not a reference to the global.
1235 7 call_method("PUSH", G_SCALAR|G_DISCARD);
1238 C<ENTER> and C<LEAVE> localise a block of code - they make sure that all
1239 variables are tidied up, everything that has been localised gets
1240 its previous value returned, and so on. Think of them as the C<{> and
1241 C<}> of a Perl block.
1243 To actually do the magic method call, we have to call a subroutine in
1244 Perl space: C<call_method> takes care of that, and it's described in
1245 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
1246 going to discard its return value.
1250 Finally, we remove the value we placed on the mark stack, since we
1251 don't need it any more.
1255 C doesn't have a concept of local scope, so perl provides one. We've
1256 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
1257 stack implements the C equivalent of, for example:
1264 See L<perlguts/Localising Changes> for how to use the save stack.
1268 =head2 Millions of Macros
1270 One thing you'll notice about the Perl source is that it's full of
1271 macros. Some have called the pervasive use of macros the hardest thing
1272 to understand, others find it adds to clarity. Let's take an example,
1273 the code which implements the addition operator:
1277 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1280 6 SETn( left + right );
1285 Every line here (apart from the braces, of course) contains a macro. The
1286 first line sets up the function declaration as Perl expects for PP code;
1287 line 3 sets up variable declarations for the argument stack and the
1288 target, the return value of the operation. Finally, it tries to see if
1289 the addition operation is overloaded; if so, the appropriate subroutine
1292 Line 5 is another variable declaration - all variable declarations start
1293 with C<d> - which pops from the top of the argument stack two NVs (hence
1294 C<nn>) and puts them into the variables C<right> and C<left>, hence the
1295 C<rl>. These are the two operands to the addition operator. Next, we
1296 call C<SETn> to set the NV of the return value to the result of adding
1297 the two values. This done, we return - the C<RETURN> macro makes sure
1298 that our return value is properly handled, and we pass the next operator
1299 to run back to the main run loop.
1301 Most of these macros are explained in L<perlapi>, and some of the more
1302 important ones are explained in L<perlxs> as well. Pay special attention
1303 to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on
1304 the C<[pad]THX_?> macros.
1306 =head2 The .i Targets
1308 You can expand the macros in a F<foo.c> file by saying
1312 which will expand the macros using cpp. Don't be scared by the results.
1314 =head2 Poking at Perl
1316 To really poke around with Perl, you'll probably want to build Perl for
1317 debugging, like this:
1319 ./Configure -d -D optimize=-g
1322 C<-g> is a flag to the C compiler to have it produce debugging
1323 information which will allow us to step through a running program.
1324 F<Configure> will also turn on the C<DEBUGGING> compilation symbol which
1325 enables all the internal debugging code in Perl. There are a whole bunch
1326 of things you can debug with this: L<perlrun> lists them all, and the
1327 best way to find out about them is to play about with them. The most
1328 useful options are probably
1330 l Context (loop) stack processing
1332 o Method and overloading resolution
1333 c String/numeric conversions
1335 Some of the functionality of the debugging code can be achieved using XS
1338 -Dr => use re 'debug'
1339 -Dx => use O 'Debug'
1341 =head2 Using a source-level debugger
1343 If the debugging output of C<-D> doesn't help you, it's time to step
1344 through perl's execution with a source-level debugger.
1350 We'll use C<gdb> for our examples here; the principles will apply to any
1351 debugger, but check the manual of the one you're using.
1355 To fire up the debugger, type
1359 You'll want to do that in your Perl source tree so the debugger can read
1360 the source code. You should see the copyright message, followed by the
1365 C<help> will get you into the documentation, but here are the most
1372 Run the program with the given arguments.
1374 =item break function_name
1376 =item break source.c:xxx
1378 Tells the debugger that we'll want to pause execution when we reach
1379 either the named function (but see L<perlguts/Internal Functions>!) or the given
1380 line in the named source file.
1384 Steps through the program a line at a time.
1388 Steps through the program a line at a time, without descending into
1393 Run until the next breakpoint.
1397 Run until the end of the current function, then stop again.
1401 Just pressing Enter will do the most recent operation again - it's a
1402 blessing when stepping through miles of source code.
1406 Execute the given C code and print its results. B<WARNING>: Perl makes
1407 heavy use of macros, and F<gdb> does not necessarily support macros
1408 (see later L</"gdb macro support">). You'll have to substitute them
1409 yourself, or to invoke cpp on the source code files
1410 (see L</"The .i Targets">)
1411 So, for instance, you can't say
1413 print SvPV_nolen(sv)
1417 print Perl_sv_2pv_nolen(sv)
1421 You may find it helpful to have a "macro dictionary", which you can
1422 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
1423 recursively apply those macros for you.
1425 =head2 gdb macro support
1427 Recent versions of F<gdb> have fairly good macro support, but
1428 in order to use it you'll need to compile perl with macro definitions
1429 included in the debugging information. Using F<gcc> version 3.1, this
1430 means configuring with C<-Doptimize=-g3>. Other compilers might use a
1431 different switch (if they support debugging macros at all).
1433 =head2 Dumping Perl Data Structures
1435 One way to get around this macro hell is to use the dumping functions in
1436 F<dump.c>; these work a little like an internal
1437 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures
1438 that you can't get at from Perl. Let's take an example. We'll use the
1439 C<$a = $b + $c> we used before, but give it a bit of context:
1440 C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around?
1442 What about C<pp_add>, the function we examined earlier to implement the
1445 (gdb) break Perl_pp_add
1446 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
1448 Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>.
1449 With the breakpoint in place, we can run our program:
1451 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
1453 Lots of junk will go past as gdb reads in the relevant source files and
1454 libraries, and then:
1456 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
1457 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
1462 We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul>
1463 arranges for two C<NV>s to be placed into C<left> and C<right> - let's
1466 #define dPOPTOPnnrl_ul NV right = POPn; \
1467 SV *leftsv = TOPs; \
1468 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
1470 C<POPn> takes the SV from the top of the stack and obtains its NV either
1471 directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function.
1472 C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses
1473 C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from
1474 C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>.
1476 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
1477 convert it. If we step again, we'll find ourselves there:
1479 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
1483 We can now use C<Perl_sv_dump> to investigate the SV:
1485 SV = PV(0xa057cc0) at 0xa0675d0
1488 PV = 0xa06a510 "6XXXX"\0
1493 We know we're going to get C<6> from this, so let's finish the
1497 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
1498 0x462669 in Perl_pp_add () at pp_hot.c:311
1501 We can also dump out this op: the current op is always stored in
1502 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
1503 similar output to L<B::Debug|B::Debug>.
1506 13 TYPE = add ===> 14
1508 FLAGS = (SCALAR,KIDS)
1510 TYPE = null ===> (12)
1512 FLAGS = (SCALAR,KIDS)
1514 11 TYPE = gvsv ===> 12
1520 # finish this later #
1524 All right, we've now had a look at how to navigate the Perl sources and
1525 some things you'll need to know when fiddling with them. Let's now get
1526 on and create a simple patch. Here's something Larry suggested: if a
1527 C<U> is the first active format during a C<pack>, (for example,
1528 C<pack "U3C8", @stuff>) then the resulting string should be treated as
1531 How do we prepare to fix this up? First we locate the code in question -
1532 the C<pack> happens at runtime, so it's going to be in one of the F<pp>
1533 files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be
1534 altering this file, let's copy it to F<pp.c~>.
1536 [Well, it was in F<pp.c> when this tutorial was written. It has now been
1537 split off with C<pp_unpack> to its own file, F<pp_pack.c>]
1539 Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then
1540 loop over the pattern, taking each format character in turn into
1541 C<datum_type>. Then for each possible format character, we swallow up
1542 the other arguments in the pattern (a field width, an asterisk, and so
1543 on) and convert the next chunk input into the specified format, adding
1544 it onto the output SV C<cat>.
1546 How do we know if the C<U> is the first format in the C<pat>? Well, if
1547 we have a pointer to the start of C<pat> then, if we see a C<U> we can
1548 test whether we're still at the start of the string. So, here's where
1552 register char *pat = SvPVx(*++MARK, fromlen);
1553 register char *patend = pat + fromlen;
1558 We'll have another string pointer in there:
1561 register char *pat = SvPVx(*++MARK, fromlen);
1562 register char *patend = pat + fromlen;
1568 And just before we start the loop, we'll set C<patcopy> to be the start
1573 sv_setpvn(cat, "", 0);
1575 while (pat < patend) {
1577 Now if we see a C<U> which was at the start of the string, we turn on
1578 the C<UTF8> flag for the output SV, C<cat>:
1580 + if (datumtype == 'U' && pat==patcopy+1)
1582 if (datumtype == '#') {
1583 while (pat < patend && *pat != '\n')
1586 Remember that it has to be C<patcopy+1> because the first character of
1587 the string is the C<U> which has been swallowed into C<datumtype!>
1589 Oops, we forgot one thing: what if there are spaces at the start of the
1590 pattern? C<pack(" U*", @stuff)> will have C<U> as the first active
1591 character, even though it's not the first thing in the pattern. In this
1592 case, we have to advance C<patcopy> along with C<pat> when we see spaces:
1594 if (isSPACE(datumtype))
1599 if (isSPACE(datumtype)) {
1604 OK. That's the C part done. Now we must do two additional things before
1605 this patch is ready to go: we've changed the behaviour of Perl, and so
1606 we must document that change. We must also provide some more regression
1607 tests to make sure our patch works and doesn't create a bug somewhere
1608 else along the line.
1610 The regression tests for each operator live in F<t/op/>, and so we
1611 make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our
1612 tests to the end. First, we'll test that the C<U> does indeed create
1615 t/op/pack.t has a sensible ok() function, but if it didn't we could
1616 use the one from t/test.pl.
1618 require './test.pl';
1619 plan( tests => 159 );
1623 print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000);
1624 print "ok $test\n"; $test++;
1626 we can write the more sensible (see L<Test::More> for a full
1627 explanation of is() and other testing functions).
1629 is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000),
1630 "U* produces unicode" );
1632 Now we'll test that we got that space-at-the-beginning business right:
1634 is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000),
1635 " with spaces at the beginning" );
1637 And finally we'll test that we don't make Unicode strings if C<U> is B<not>
1638 the first active format:
1640 isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000),
1641 "U* not first isn't unicode" );
1643 Mustn't forget to change the number of tests which appears at the top,
1644 or else the automated tester will get confused. This will either look
1651 plan( tests => 156 );
1653 We now compile up Perl, and run it through the test suite. Our new
1656 Finally, the documentation. The job is never done until the paperwork is
1657 over, so let's describe the change we've just made. The relevant place
1658 is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert
1659 this text in the description of C<pack>:
1663 If the pattern begins with a C<U>, the resulting string will be treated
1664 as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string
1665 with an initial C<U0>, and the bytes that follow will be interpreted as
1666 Unicode characters. If you don't want this to happen, you can begin your
1667 pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your
1668 string, and then follow this with a C<U*> somewhere in your pattern.
1670 All done. Now let's create the patch. F<Porting/patching.pod> tells us
1671 that if we're making major changes, we should copy the entire directory
1672 to somewhere safe before we begin fiddling, and then do
1674 diff -ruN old new > patch
1676 However, we know which files we've changed, and we can simply do this:
1678 diff -u pp.c~ pp.c > patch
1679 diff -u t/op/pack.t~ t/op/pack.t >> patch
1680 diff -u pod/perlfunc.pod~ pod/perlfunc.pod >> patch
1682 We end up with a patch looking a little like this:
1684 --- pp.c~ Fri Jun 02 04:34:10 2000
1685 +++ pp.c Fri Jun 16 11:37:25 2000
1686 @@ -4375,6 +4375,7 @@
1689 register char *pat = SvPVx(*++MARK, fromlen);
1691 register char *patend = pat + fromlen;
1694 @@ -4405,6 +4406,7 @@
1697 And finally, we submit it, with our rationale, to perl5-porters. Job
1700 =head2 Patching a core module
1702 This works just like patching anything else, with an extra
1703 consideration. Many core modules also live on CPAN. If this is so,
1704 patch the CPAN version instead of the core and send the patch off to
1705 the module maintainer (with a copy to p5p). This will help the module
1706 maintainer keep the CPAN version in sync with the core version without
1707 constantly scanning p5p.
1709 =head2 Adding a new function to the core
1711 If, as part of a patch to fix a bug, or just because you have an
1712 especially good idea, you decide to add a new function to the core,
1713 discuss your ideas on p5p well before you start work. It may be that
1714 someone else has already attempted to do what you are considering and
1715 can give lots of good advice or even provide you with bits of code
1716 that they already started (but never finished).
1718 You have to follow all of the advice given above for patching. It is
1719 extremely important to test any addition thoroughly and add new tests
1720 to explore all boundary conditions that your new function is expected
1721 to handle. If your new function is used only by one module (e.g. toke),
1722 then it should probably be named S_your_function (for static); on the
1723 other hand, if you expect it to accessible from other functions in
1724 Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions>
1727 The location of any new code is also an important consideration. Don't
1728 just create a new top level .c file and put your code there; you would
1729 have to make changes to Configure (so the Makefile is created properly),
1730 as well as possibly lots of include files. This is strictly pumpking
1733 It is better to add your function to one of the existing top level
1734 source code files, but your choice is complicated by the nature of
1735 the Perl distribution. Only the files that are marked as compiled
1736 static are located in the perl executable. Everything else is located
1737 in the shared library (or DLL if you are running under WIN32). So,
1738 for example, if a function was only used by functions located in
1739 toke.c, then your code can go in toke.c. If, however, you want to call
1740 the function from universal.c, then you should put your code in another
1741 location, for example util.c.
1743 In addition to writing your c-code, you will need to create an
1744 appropriate entry in embed.pl describing your function, then run
1745 'make regen_headers' to create the entries in the numerous header
1746 files that perl needs to compile correctly. See L<perlguts/Internal Functions>
1747 for information on the various options that you can set in embed.pl.
1748 You will forget to do this a few (or many) times and you will get
1749 warnings during the compilation phase. Make sure that you mention
1750 this when you post your patch to P5P; the pumpking needs to know this.
1752 When you write your new code, please be conscious of existing code
1753 conventions used in the perl source files. See L<perlstyle> for
1754 details. Although most of the guidelines discussed seem to focus on
1755 Perl code, rather than c, they all apply (except when they don't ;).
1756 See also I<Porting/patching.pod> file in the Perl source distribution
1757 for lots of details about both formatting and submitting patches of
1760 Lastly, TEST TEST TEST TEST TEST any code before posting to p5p.
1761 Test on as many platforms as you can find. Test as many perl
1762 Configure options as you can (e.g. MULTIPLICITY). If you have
1763 profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL>
1764 below for how to use them to further test your code. Remember that
1765 most of the people on P5P are doing this on their own time and
1766 don't have the time to debug your code.
1768 =head2 Writing a test
1770 Every module and built-in function has an associated test file (or
1771 should...). If you add or change functionality, you have to write a
1772 test. If you fix a bug, you have to write a test so that bug never
1773 comes back. If you alter the docs, it would be nice to test what the
1774 new documentation says.
1776 In short, if you submit a patch you probably also have to patch the
1779 For modules, the test file is right next to the module itself.
1780 F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation,
1781 so there are some snags (and it would be wonderful for you to brush
1782 them out), but it basically works that way. Everything else lives in
1789 Testing of the absolute basic functionality of Perl. Things like
1790 C<if>, basic file reads and writes, simple regexes, etc. These are
1791 run first in the test suite and if any of them fail, something is
1796 These test the basic control structures, C<if/else>, C<while>,
1801 Tests basic issues of how Perl parses and compiles itself.
1805 Tests for built-in IO functions, including command line arguments.
1809 The old home for the module tests, you shouldn't put anything new in
1810 here. There are still some bits and pieces hanging around in here
1811 that need to be moved. Perhaps you could move them? Thanks!
1815 Tests for perl's built in functions that don't fit into any of the
1820 Tests for POD directives. There are still some tests for the Pod
1821 modules hanging around in here that need to be moved out into F<lib/>.
1825 Testing features of how perl actually runs, including exit codes and
1826 handling of PERL* environment variables.
1830 Tests for the core support of Unicode.
1834 Windows-specific tests.
1838 A test suite for the s2p converter.
1842 The core uses the same testing style as the rest of Perl, a simple
1843 "ok/not ok" run through Test::Harness, but there are a few special
1846 There are three ways to write a test in the core. Test::More,
1847 t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The
1848 decision of which to use depends on what part of the test suite you're
1849 working on. This is a measure to prevent a high-level failure (such
1850 as Config.pm breaking) from causing basic functionality tests to fail.
1856 Since we don't know if require works, or even subroutines, use ad hoc
1857 tests for these two. Step carefully to avoid using the feature being
1860 =item t/cmd t/run t/io t/op
1862 Now that basic require() and subroutines are tested, you can use the
1863 t/test.pl library which emulates the important features of Test::More
1864 while using a minimum of core features.
1866 You can also conditionally use certain libraries like Config, but be
1867 sure to skip the test gracefully if it's not there.
1871 Now that the core of Perl is tested, Test::More can be used. You can
1872 also use the full suite of core modules in the tests.
1876 When you say "make test" Perl uses the F<t/TEST> program to run the
1877 test suite. All tests are run from the F<t/> directory, B<not> the
1878 directory which contains the test. This causes some problems with the
1879 tests in F<lib/>, so here's some opportunity for some patching.
1881 You must be triply conscious of cross-platform concerns. This usually
1882 boils down to using File::Spec and avoiding things like C<fork()> and
1883 C<system()> unless absolutely necessary.
1885 =head2 Special Make Test Targets
1887 There are various special make targets that can be used to test Perl
1888 slightly differently than the standard "test" target. Not all them
1889 are expected to give a 100% success rate. Many of them have several
1896 Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests).
1900 Run all the tests through B::Deparse. Not all tests will succeed.
1902 =item test.taintwarn
1904 Run all tests with the B<-t> command-line switch. Not all tests
1905 are expected to succeed (until they're specifically fixed, of course).
1909 Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>,
1910 F<t/op>, and F<t/uni> tests.
1912 =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind
1914 (Only in Linux) Run all the tests using the memory leak + naughty
1915 memory access tool "valgrind". The log files will be named
1916 F<testname.valgrind>.
1918 =item test.third check.third utest.third ucheck.third
1920 (Only in Tru64) Run all the tests using the memory leak + naughty
1921 memory access tool "Third Degree". The log files will be named
1922 F<perl3.log.testname>.
1924 =item test.torture torturetest
1926 Run all the usual tests and some extra tests. As of Perl 5.8.0 the
1927 only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>.
1929 You can also run the torture test with F<t/harness> by giving
1930 C<-torture> argument to F<t/harness>.
1932 =item utest ucheck test.utf8 check.utf8
1934 Run all the tests with -Mutf8. Not all tests will succeed.
1938 Run the test suite with the F<t/harness> controlling program, instead of
1939 F<t/TEST>. F<t/harness> is more sophisticated, and uses the
1940 L<Test::Harness> module, thus using this test target supposes that perl
1941 mostly works. The main advantage for our purposes is that it prints a
1942 detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it
1943 doesn't redirect stderr to stdout.
1947 =head2 Running tests by hand
1949 You can run part of the test suite by hand by using one the following
1950 commands from the F<t/> directory :
1952 ./perl -I../lib TEST list-of-.t-files
1956 ./perl -I../lib harness list-of-.t-files
1958 (if you don't specify test scripts, the whole test suite will be run.)
1960 You can run an individual test by a command similar to
1962 ./perl -I../lib patho/to/foo.t
1964 except that the harnesses set up some environment variables that may
1965 affect the execution of the test :
1971 indicates that we're running this test part of the perl core test suite.
1972 This is useful for modules that have a dual life on CPAN.
1974 =item PERL_DESTRUCT_LEVEL=2
1976 is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>)
1980 (used only by F<t/TEST>) if set, overrides the path to the perl executable
1981 that should be used to run the tests (the default being F<./perl>).
1983 =item PERL_SKIP_TTY_TEST
1985 if set, tells to skip the tests that need a terminal. It's actually set
1986 automatically by the Makefile, but can also be forced artificially by
1987 running 'make test_notty'.
1991 =head1 EXTERNAL TOOLS FOR DEBUGGING PERL
1993 Sometimes it helps to use external tools while debugging and
1994 testing Perl. This section tries to guide you through using
1995 some common testing and debugging tools with Perl. This is
1996 meant as a guide to interfacing these tools with Perl, not
1997 as any kind of guide to the use of the tools themselves.
1999 B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or
2000 Third Degree greatly slows down the execution: seconds become minutes,
2001 minutes become hours. For example as of Perl 5.8.1, the
2002 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
2003 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
2004 than six hours, even on a snappy computer-- the said test must be
2005 doing something that is quite unfriendly for memory debuggers. If you
2006 don't feel like waiting, that you can simply kill away the perl
2009 B<NOTE 2>: To minimize the number of memory leak false alarms (see
2010 L</PERL_DESTRUCT_LEVEL> for more information), you have to have
2011 environment variable PERL_DESTRUCT_LEVEL set to 2. The F<TEST>
2012 and harness scripts do that automatically. But if you are running
2013 some of the tests manually-- for csh-like shells:
2015 setenv PERL_DESTRUCT_LEVEL 2
2017 and for Bourne-type shells:
2019 PERL_DESTRUCT_LEVEL=2
2020 export PERL_DESTRUCT_LEVEL
2022 or in UNIXy environments you can also use the C<env> command:
2024 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
2026 B<NOTE 3>: There are known memory leaks when there are compile-time
2027 errors within eval or require, seeing C<S_doeval> in the call stack
2028 is a good sign of these. Fixing these leaks is non-trivial,
2029 unfortunately, but they must be fixed eventually.
2031 =head2 Rational Software's Purify
2033 Purify is a commercial tool that is helpful in identifying
2034 memory overruns, wild pointers, memory leaks and other such
2035 badness. Perl must be compiled in a specific way for
2036 optimal testing with Purify. Purify is available under
2037 Windows NT, Solaris, HP-UX, SGI, and Siemens Unix.
2039 =head2 Purify on Unix
2041 On Unix, Purify creates a new Perl binary. To get the most
2042 benefit out of Purify, you should create the perl to Purify
2045 sh Configure -Accflags=-DPURIFY -Doptimize='-g' \
2046 -Uusemymalloc -Dusemultiplicity
2048 where these arguments mean:
2052 =item -Accflags=-DPURIFY
2054 Disables Perl's arena memory allocation functions, as well as
2055 forcing use of memory allocation functions derived from the
2058 =item -Doptimize='-g'
2060 Adds debugging information so that you see the exact source
2061 statements where the problem occurs. Without this flag, all
2062 you will see is the source filename of where the error occurred.
2066 Disable Perl's malloc so that Purify can more closely monitor
2067 allocations and leaks. Using Perl's malloc will make Purify
2068 report most leaks in the "potential" leaks category.
2070 =item -Dusemultiplicity
2072 Enabling the multiplicity option allows perl to clean up
2073 thoroughly when the interpreter shuts down, which reduces the
2074 number of bogus leak reports from Purify.
2078 Once you've compiled a perl suitable for Purify'ing, then you
2083 which creates a binary named 'pureperl' that has been Purify'ed.
2084 This binary is used in place of the standard 'perl' binary
2085 when you want to debug Perl memory problems.
2087 As an example, to show any memory leaks produced during the
2088 standard Perl testset you would create and run the Purify'ed
2093 ../pureperl -I../lib harness
2095 which would run Perl on test.pl and report any memory problems.
2097 Purify outputs messages in "Viewer" windows by default. If
2098 you don't have a windowing environment or if you simply
2099 want the Purify output to unobtrusively go to a log file
2100 instead of to the interactive window, use these following
2101 options to output to the log file "perl.log":
2103 setenv PURIFYOPTIONS "-chain-length=25 -windows=no \
2104 -log-file=perl.log -append-logfile=yes"
2106 If you plan to use the "Viewer" windows, then you only need this option:
2108 setenv PURIFYOPTIONS "-chain-length=25"
2110 In Bourne-type shells:
2113 export PURIFYOPTIONS
2115 or if you have the "env" utility:
2117 env PURIFYOPTIONS="..." ../pureperl ...
2121 Purify on Windows NT instruments the Perl binary 'perl.exe'
2122 on the fly. There are several options in the makefile you
2123 should change to get the most use out of Purify:
2129 You should add -DPURIFY to the DEFINES line so the DEFINES
2130 line looks something like:
2132 DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1
2134 to disable Perl's arena memory allocation functions, as
2135 well as to force use of memory allocation functions derived
2136 from the system malloc.
2138 =item USE_MULTI = define
2140 Enabling the multiplicity option allows perl to clean up
2141 thoroughly when the interpreter shuts down, which reduces the
2142 number of bogus leak reports from Purify.
2144 =item #PERL_MALLOC = define
2146 Disable Perl's malloc so that Purify can more closely monitor
2147 allocations and leaks. Using Perl's malloc will make Purify
2148 report most leaks in the "potential" leaks category.
2152 Adds debugging information so that you see the exact source
2153 statements where the problem occurs. Without this flag, all
2154 you will see is the source filename of where the error occurred.
2158 As an example, to show any memory leaks produced during the
2159 standard Perl testset you would create and run Purify as:
2164 purify ../perl -I../lib harness
2166 which would instrument Perl in memory, run Perl on test.pl,
2167 then finally report any memory problems.
2171 The excellent valgrind tool can be used to find out both memory leaks
2172 and illegal memory accesses. As of August 2003 it unfortunately works
2173 only on x86 (ELF) Linux. The special "test.valgrind" target can be used
2174 to run the tests under valgrind. Found errors and memory leaks are
2175 logged in files named F<test.valgrind>.
2177 As system libraries (most notably glibc) are also triggering errors,
2178 valgrind allows to suppress such errors using suppression files. The
2179 default suppression file that comes with valgrind already catches a lot
2180 of them. Some additional suppressions are defined in F<t/perl.supp>.
2182 To get valgrind and for more information see
2184 http://developer.kde.org/~sewardj/
2186 =head2 Compaq's/Digital's/HP's Third Degree
2188 Third Degree is a tool for memory leak detection and memory access checks.
2189 It is one of the many tools in the ATOM toolkit. The toolkit is only
2190 available on Tru64 (formerly known as Digital UNIX formerly known as
2193 When building Perl, you must first run Configure with -Doptimize=-g
2194 and -Uusemymalloc flags, after that you can use the make targets
2195 "perl.third" and "test.third". (What is required is that Perl must be
2196 compiled using the C<-g> flag, you may need to re-Configure.)
2198 The short story is that with "atom" you can instrument the Perl
2199 executable to create a new executable called F<perl.third>. When the
2200 instrumented executable is run, it creates a log of dubious memory
2201 traffic in file called F<perl.3log>. See the manual pages of atom and
2202 third for more information. The most extensive Third Degree
2203 documentation is available in the Compaq "Tru64 UNIX Programmer's
2204 Guide", chapter "Debugging Programs with Third Degree".
2206 The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/
2207 subdirectory. There is a problem with these files: Third Degree is so
2208 effective that it finds problems also in the system libraries.
2209 Therefore you should used the Porting/thirdclean script to cleanup
2210 the F<*.3log> files.
2212 There are also leaks that for given certain definition of a leak,
2213 aren't. See L</PERL_DESTRUCT_LEVEL> for more information.
2215 =head2 PERL_DESTRUCT_LEVEL
2217 If you want to run any of the tests yourself manually using e.g.
2218 valgrind, or the pureperl or perl.third executables, please note that
2219 by default perl B<does not> explicitly cleanup all the memory it has
2220 allocated (such as global memory arenas) but instead lets the exit()
2221 of the whole program "take care" of such allocations, also known as
2222 "global destruction of objects".
2224 There is a way to tell perl to do complete cleanup: set the
2225 environment variable PERL_DESTRUCT_LEVEL to a non-zero value.
2226 The t/TEST wrapper does set this to 2, and this is what you
2227 need to do too, if you don't want to see the "global leaks":
2228 For example, for "third-degreed" Perl:
2230 env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t
2232 (Note: the mod_perl apache module uses also this environment variable
2233 for its own purposes and extended its semantics. Refer to the mod_perl
2234 documentation for more information. Also, spawned threads do the
2235 equivalent of setting this variable to the value 1.)
2237 If, at the end of a run you get the message I<N scalars leaked>, you can
2238 recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause
2239 the addresses of all those leaked SVs to be dumped; it also converts
2240 C<new_SV()> from a macro into a real function, so you can use your
2241 favourite debugger to discover where those pesky SVs were allocated.
2245 Depending on your platform there are various of profiling Perl.
2247 There are two commonly used techniques of profiling executables:
2248 I<statistical time-sampling> and I<basic-block counting>.
2250 The first method takes periodically samples of the CPU program
2251 counter, and since the program counter can be correlated with the code
2252 generated for functions, we get a statistical view of in which
2253 functions the program is spending its time. The caveats are that very
2254 small/fast functions have lower probability of showing up in the
2255 profile, and that periodically interrupting the program (this is
2256 usually done rather frequently, in the scale of milliseconds) imposes
2257 an additional overhead that may skew the results. The first problem
2258 can be alleviated by running the code for longer (in general this is a
2259 good idea for profiling), the second problem is usually kept in guard
2260 by the profiling tools themselves.
2262 The second method divides up the generated code into I<basic blocks>.
2263 Basic blocks are sections of code that are entered only in the
2264 beginning and exited only at the end. For example, a conditional jump
2265 starts a basic block. Basic block profiling usually works by
2266 I<instrumenting> the code by adding I<enter basic block #nnnn>
2267 book-keeping code to the generated code. During the execution of the
2268 code the basic block counters are then updated appropriately. The
2269 caveat is that the added extra code can skew the results: again, the
2270 profiling tools usually try to factor their own effects out of the
2273 =head2 Gprof Profiling
2275 gprof is a profiling tool available in many UNIX platforms,
2276 it uses F<statistical time-sampling>.
2278 You can build a profiled version of perl called "perl.gprof" by
2279 invoking the make target "perl.gprof" (What is required is that Perl
2280 must be compiled using the C<-pg> flag, you may need to re-Configure).
2281 Running the profiled version of Perl will create an output file called
2282 F<gmon.out> is created which contains the profiling data collected
2283 during the execution.
2285 The gprof tool can then display the collected data in various ways.
2286 Usually gprof understands the following options:
2292 Suppress statically defined functions from the profile.
2296 Suppress the verbose descriptions in the profile.
2300 Exclude the given routine and its descendants from the profile.
2304 Display only the given routine and its descendants in the profile.
2308 Generate a summary file called F<gmon.sum> which then may be given
2309 to subsequent gprof runs to accumulate data over several runs.
2313 Display routines that have zero usage.
2317 For more detailed explanation of the available commands and output
2318 formats, see your own local documentation of gprof.
2320 =head2 GCC gcov Profiling
2322 Starting from GCC 3.0 I<basic block profiling> is officially available
2325 You can build a profiled version of perl called F<perl.gcov> by
2326 invoking the make target "perl.gcov" (what is required that Perl must
2327 be compiled using gcc with the flags C<-fprofile-arcs
2328 -ftest-coverage>, you may need to re-Configure).
2330 Running the profiled version of Perl will cause profile output to be
2331 generated. For each source file an accompanying ".da" file will be
2334 To display the results you use the "gcov" utility (which should
2335 be installed if you have gcc 3.0 or newer installed). F<gcov> is
2336 run on source code files, like this
2340 which will cause F<sv.c.gcov> to be created. The F<.gcov> files
2341 contain the source code annotated with relative frequencies of
2342 execution indicated by "#" markers.
2344 Useful options of F<gcov> include C<-b> which will summarise the
2345 basic block, branch, and function call coverage, and C<-c> which
2346 instead of relative frequencies will use the actual counts. For
2347 more information on the use of F<gcov> and basic block profiling
2348 with gcc, see the latest GNU CC manual, as of GCC 3.0 see
2350 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html
2352 and its section titled "8. gcov: a Test Coverage Program"
2354 http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132
2356 =head2 Pixie Profiling
2358 Pixie is a profiling tool available on IRIX and Tru64 (aka Digital
2359 UNIX aka DEC OSF/1) platforms. Pixie does its profiling using
2360 I<basic-block counting>.
2362 You can build a profiled version of perl called F<perl.pixie> by
2363 invoking the make target "perl.pixie" (what is required is that Perl
2364 must be compiled using the C<-g> flag, you may need to re-Configure).
2366 In Tru64 a file called F<perl.Addrs> will also be silently created,
2367 this file contains the addresses of the basic blocks. Running the
2368 profiled version of Perl will create a new file called "perl.Counts"
2369 which contains the counts for the basic block for that particular
2372 To display the results you use the F<prof> utility. The exact
2373 incantation depends on your operating system, "prof perl.Counts" in
2374 IRIX, and "prof -pixie -all -L. perl" in Tru64.
2376 In IRIX the following prof options are available:
2382 Reports the most heavily used lines in descending order of use.
2383 Useful for finding the hotspot lines.
2387 Groups lines by procedure, with procedures sorted in descending order of use.
2388 Within a procedure, lines are listed in source order.
2389 Useful for finding the hotspots of procedures.
2393 In Tru64 the following options are available:
2399 Procedures sorted in descending order by the number of cycles executed
2400 in each procedure. Useful for finding the hotspot procedures.
2401 (This is the default option.)
2405 Lines sorted in descending order by the number of cycles executed in
2406 each line. Useful for finding the hotspot lines.
2408 =item -i[nvocations]
2410 The called procedures are sorted in descending order by number of calls
2411 made to the procedures. Useful for finding the most used procedures.
2415 Grouped by procedure, sorted by cycles executed per procedure.
2416 Useful for finding the hotspots of procedures.
2420 The compiler emitted code for these lines, but the code was unexecuted.
2424 Unexecuted procedures.
2428 For further information, see your system's manual pages for pixie and prof.
2430 =head2 Miscellaneous tricks
2436 Those debugging perl with the DDD frontend over gdb may find the
2439 You can extend the data conversion shortcuts menu, so for example you
2440 can display an SV's IV value with one click, without doing any typing.
2441 To do that simply edit ~/.ddd/init file and add after:
2443 ! Display shortcuts.
2444 Ddd*gdbDisplayShortcuts: \
2445 /t () // Convert to Bin\n\
2446 /d () // Convert to Dec\n\
2447 /x () // Convert to Hex\n\
2448 /o () // Convert to Oct(\n\
2450 the following two lines:
2452 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
2453 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
2455 so now you can do ivx and pvx lookups or you can plug there the
2456 sv_peek "conversion":
2458 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
2460 (The my_perl is for threaded builds.)
2461 Just remember that every line, but the last one, should end with \n\
2463 Alternatively edit the init file interactively via:
2464 3rd mouse button -> New Display -> Edit Menu
2466 Note: you can define up to 20 conversion shortcuts in the gdb
2471 If you see in a debugger a memory area mysteriously full of 0xabababab,
2472 you may be seeing the effect of the Poison() macro, see L<perlclib>.
2478 We've had a brief look around the Perl source, an overview of the stages
2479 F<perl> goes through when it's running your code, and how to use a
2480 debugger to poke at the Perl guts. We took a very simple problem and
2481 demonstrated how to solve it fully - with documentation, regression
2482 tests, and finally a patch for submission to p5p. Finally, we talked
2483 about how to use external tools to debug and test Perl.
2485 I'd now suggest you read over those references again, and then, as soon
2486 as possible, get your hands dirty. The best way to learn is by doing,
2493 Subscribe to perl5-porters, follow the patches and try and understand
2494 them; don't be afraid to ask if there's a portion you're not clear on -
2495 who knows, you may unearth a bug in the patch...
2499 Keep up to date with the bleeding edge Perl distributions and get
2500 familiar with the changes. Try and get an idea of what areas people are
2501 working on and the changes they're making.
2505 Do read the README associated with your operating system, e.g. README.aix
2506 on the IBM AIX OS. Don't hesitate to supply patches to that README if
2507 you find anything missing or changed over a new OS release.
2511 Find an area of Perl that seems interesting to you, and see if you can
2512 work out how it works. Scan through the source, and step over it in the
2513 debugger. Play, poke, investigate, fiddle! You'll probably get to
2514 understand not just your chosen area but a much wider range of F<perl>'s
2515 activity as well, and probably sooner than you'd think.
2521 =item I<The Road goes ever on and on, down from the door where it began.>
2525 If you can do these things, you've started on the long road to Perl porting.
2526 Thanks for wanting to help make Perl better - and happy hacking!
2530 This document was written by Nathan Torkington, and is maintained by
2531 the perl5-porters mailing list.