5 Consistent formatting of this file is achieved with:
6 perl ./Porting/podtidy pod/perlhacktips.pod
10 perlhacktips - Tips for Perl core C code hacking
14 This document will help you learn the best way to go about hacking on
15 the Perl core C code. It covers common problems, debugging, profiling,
18 If you haven't read L<perlhack> and L<perlhacktut> yet, you might want
21 =head1 COMMON PROBLEMS
23 Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In
24 some cases we have to take pre-ANSI requirements into consideration.
25 You don't care about some particular platform having broken Perl? I
26 hear there is still a strong demand for J2EE programmers.
28 =head2 Perl environment problems
34 Not compiling with threading
36 Compiling with threading (-Duseithreads) completely rewrites the
37 function prototypes of Perl. You better try your changes with that.
38 Related to this is the difference between "Perl_-less" and "Perl_-ly"
41 Perl_sv_setiv(aTHX_ ...);
44 The first one explicitly passes in the context, which is needed for
45 e.g. threaded builds. The second one does that implicitly; do not get
46 them mixed. If you are not passing in a aTHX_, you will need to do a
47 dTHX (or a dVAR) as the first thing in the function.
49 See L<perlguts/"How multiple interpreters and concurrency are
50 supported"> for further discussion about context.
54 Not compiling with -DDEBUGGING
56 The DEBUGGING define exposes more code to the compiler, therefore more
57 ways for things to go wrong. You should try it.
61 Introducing (non-read-only) globals
63 Do not introduce any modifiable globals, truly global or file static.
64 They are bad form and complicate multithreading and other forms of
65 concurrency. The right way is to introduce them as new interpreter
66 variables, see F<intrpvar.h> (at the very end for binary
69 Introducing read-only (const) globals is okay, as long as you verify
70 with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has
71 BSD-style output) that the data you added really is read-only. (If it
72 is, it shouldn't show up in the output of that command.)
74 If you want to have static strings, make them constant:
76 static const char etc[] = "...";
78 If you want to have arrays of constant strings, note carefully the
79 right combination of C<const>s:
81 static const char * const yippee[] =
82 {"hi", "ho", "silver"};
84 There is a way to completely hide any modifiable globals (they are all
85 moved to heap), the compilation setting
86 C<-DPERL_GLOBAL_STRUCT_PRIVATE>. It is not normally used, but can be
87 used for testing, read more about it in L<perlguts/"Background and
88 PERL_IMPLICIT_CONTEXT">.
92 Not exporting your new function
94 Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any
95 function that is part of the public API (the shared Perl library) to be
96 explicitly marked as exported. See the discussion about F<embed.pl> in
101 Exporting your new function
103 The new shiny result of either genuine new functionality or your
104 arduous refactoring is now ready and correctly exported. So what could
107 Maybe simply that your function did not need to be exported in the
108 first place. Perl has a long and not so glorious history of exporting
109 functions that it should not have.
111 If the function is used only inside one source code file, make it
112 static. See the discussion about F<embed.pl> in L<perlguts>.
114 If the function is used across several files, but intended only for
115 Perl's internal use (and this should be the common case), do not export
116 it to the public API. See the discussion about F<embed.pl> in
121 =head2 Portability problems
123 The following are common causes of compilation and/or execution
124 failures, not common to Perl as such. The C FAQ is good bedtime
125 reading. Please test your changes with as many C compilers and
126 platforms as possible; we will, anyway, and it's nice to save oneself
127 from public embarrassment.
129 If using gcc, you can add the C<-std=c89> option which will hopefully
130 catch most of these unportabilities. (However it might also catch
131 incompatibilities in your system's header files.)
133 Use the Configure C<-Dgccansipedantic> flag to enable the gcc C<-ansi
134 -pedantic> flags which enforce stricter ANSI rules.
136 If using the C<gcc -Wall> note that not all the possible warnings (like
137 C<-Wunitialized>) are given unless you also compile with C<-O>.
139 Note that if using gcc, starting from Perl 5.9.5 the Perl core source
140 code files (the ones at the top level of the source code distribution,
141 but not e.g. the extensions under ext/) are automatically compiled with
142 as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>, and a
143 selection of C<-W> flags (see cflags.SH).
145 Also study L<perlport> carefully to avoid any bad assumptions about the
146 operating system, filesystems, and so forth.
148 You may once in a while try a "make microperl" to see whether we can
149 still compile Perl with just the bare minimum of interfaces. (See
152 Do not assume an operating system indicates a certain compiler.
158 Casting pointers to integers or casting integers to pointers
170 Both are bad, and broken, and unportable. Use the PTR2IV() macro that
171 does it right. (Likewise, there are PTR2UV(), PTR2NV(), INT2PTR(), and
176 Casting between data function pointers and data pointers
178 Technically speaking casting between function pointers and data
179 pointers is unportable and undefined, but practically speaking it seems
180 to work, but you should use the FPTR2DPTR() and DPTR2FPTR() macros.
181 Sometimes you can also play games with unions.
185 Assuming sizeof(int) == sizeof(long)
187 There are platforms where longs are 64 bits, and platforms where ints
188 are 64 bits, and while we are out to shock you, even platforms where
189 shorts are 64 bits. This is all legal according to the C standard. (In
190 other words, "long long" is not a portable way to specify 64 bits, and
191 "long long" is not even guaranteed to be any wider than "long".)
193 Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth.
194 Avoid things like I32 because they are B<not> guaranteed to be
195 I<exactly> 32 bits, they are I<at least> 32 bits, nor are they
196 guaranteed to be B<int> or B<long>. If you really explicitly need
197 64-bit variables, use I64 and U64, but only if guarded by HAS_QUAD.
201 Assuming one can dereference any type of pointer for any type of data
204 long pony = *p; /* BAD */
206 Many platforms, quite rightly so, will give you a core dump instead of
207 a pony if the p happens not to be correctly aligned.
213 (int)*p = ...; /* BAD */
215 Simply not portable. Get your lvalue to be of the right type, or maybe
216 use temporary variables, or dirty tricks with unions.
220 Assume B<anything> about structs (especially the ones you don't
221 control, like the ones coming from the system headers)
227 That a certain field exists in a struct
231 That no other fields exist besides the ones you know of
235 That a field is of certain signedness, sizeof, or type
239 That the fields are in a certain order
245 While C guarantees the ordering specified in the struct definition,
246 between different platforms the definitions might differ
252 That the sizeof(struct) or the alignments are the same everywhere
258 There might be padding bytes between the fields to align the fields -
259 the bytes can be anything
263 Structs are required to be aligned to the maximum alignment required by
264 the fields - which for native types is for usually equivalent to
265 sizeof() of the field
273 Assuming the character set is ASCIIish
275 Perl can compile and run under EBCDIC platforms. See L<perlebcdic>.
276 This is transparent for the most part, but because the character sets
277 differ, you shouldn't use numeric (decimal, octal, nor hex) constants
278 to refer to characters. You can safely say 'A', but not 0x41. You can
279 safely say '\n', but not \012. If a character doesn't have a trivial
280 input form, you should add it to the list in
281 F<regen/unicode_constants.pl>, and have Perl create #defines for you,
282 based on the current platform.
284 Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper
285 case alphabetic characters. That is not true in EBCDIC. Nor for 'a' to
286 'z'. But '0' - '9' is an unbroken range in both systems. Don't assume
287 anything about other ranges.
289 Many of the comments in the existing code ignore the possibility of
290 EBCDIC, and may be wrong therefore, even if the code works. This is
291 actually a tribute to the successful transparent insertion of being
292 able to handle EBCDIC without having to change pre-existing code.
294 UTF-8 and UTF-EBCDIC are two different encodings used to represent
295 Unicode code points as sequences of bytes. Macros with the same names
296 (but different definitions) in C<utf8.h> and C<utfebcdic.h> are used to
297 allow the calling code to think that there is only one such encoding.
298 This is almost always referred to as C<utf8>, but it means the EBCDIC
299 version as well. Again, comments in the code may well be wrong even if
300 the code itself is right. For example, the concept of C<invariant
301 characters> differs between ASCII and EBCDIC. On ASCII platforms, only
302 characters that do not have the high-order bit set (i.e. whose ordinals
303 are strict ASCII, 0 - 127) are invariant, and the documentation and
304 comments in the code may assume that, often referring to something
305 like, say, C<hibit>. The situation differs and is not so simple on
306 EBCDIC machines, but as long as the code itself uses the
307 C<NATIVE_IS_INVARIANT()> macro appropriately, it works, even if the
312 Assuming the character set is just ASCII
314 ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
315 characters have different meanings depending on the locale. Absent a
316 locale, currently these extra characters are generally considered to be
317 unassigned, and this has presented some problems. This is being changed
318 starting in 5.12 so that these characters will be considered to be
319 Latin-1 (ISO-8859-1).
323 Mixing #define and #ifdef
325 #define BURGLE(x) ... \
326 #ifdef BURGLE_OLD_STYLE /* BAD */
327 ... do it the old way ... \
329 ... do it the new way ... \
332 You cannot portably "stack" cpp directives. For example in the above
333 you need two separate BURGLE() #defines, one for each #ifdef branch.
337 Adding non-comment stuff after #endif or #else
341 #else !SNOSH /* BAD */
343 #endif SNOSH /* BAD */
345 The #endif and #else cannot portably have anything non-comment after
346 them. If you want to document what is going (which is a good idea
347 especially if the branches are long), use (C) comments:
355 The gcc option C<-Wendif-labels> warns about the bad variant (by
356 default on starting from Perl 5.9.4).
360 Having a comma after the last element of an enum list
368 is not portable. Leave out the last comma.
370 Also note that whether enums are implicitly morphable to ints varies
371 between compilers, you might need to (int).
377 // This function bamfoodles the zorklator. /* BAD */
379 That is C99 or C++. Perl is C89. Using the //-comments is silently
380 allowed by many C compilers but cranking up the ANSI C89 strictness
381 (which we like to do) causes the compilation to fail.
385 Mixing declarations and code
390 set_zorkmids(n); /* BAD */
393 That is C99 or C++. Some C compilers allow that, but you shouldn't.
395 The gcc option C<-Wdeclaration-after-statements> scans for such
396 problems (by default on starting from Perl 5.9.4).
400 Introducing variables inside for()
402 for(int i = ...; ...; ...) { /* BAD */
404 That is C99 or C++. While it would indeed be awfully nice to have that
405 also in C89, to limit the scope of the loop variable, alas, we cannot.
409 Mixing signed char pointers with unsigned char pointers
411 int foo(char *s) { ... }
413 unsigned char *t = ...; /* Or U8* t = ... */
416 While this is legal practice, it is certainly dubious, and downright
417 fatal in at least one platform: for example VMS cc considers this a
418 fatal error. One cause for people often making this mistake is that a
419 "naked char" and therefore dereferencing a "naked char pointer" have an
420 undefined signedness: it depends on the compiler and the flags of the
421 compiler and the underlying platform whether the result is signed or
422 unsigned. For this very same reason using a 'char' as an array index is
427 Macros that have string constants and their arguments as substrings of
430 #define FOO(n) printf("number = %d\n", n) /* BAD */
433 Pre-ANSI semantics for that was equivalent to
435 printf("10umber = %d\10");
437 which is probably not what you were expecting. Unfortunately at least
438 one reasonably common and modern C compiler does "real backward
439 compatibility" here, in AIX that is what still happens even though the
440 rest of the AIX compiler is very happily C89.
444 Using printf formats for non-basic C types
447 printf("i = %d\n", i); /* BAD */
449 While this might by accident work in some platform (where IV happens to
450 be an C<int>), in general it cannot. IV might be something larger. Even
451 worse the situation is with more specific types (defined by Perl's
452 configuration step in F<config.h>):
455 printf("who = %d\n", who); /* BAD */
457 The problem here is that Uid_t might be not only not C<int>-wide but it
458 might also be unsigned, in which case large uids would be printed as
461 There is no simple solution to this because of printf()'s limited
462 intelligence, but for many types the right format is available as with
463 either 'f' or '_f' suffix, for example:
465 IVdf /* IV in decimal */
466 UVxf /* UV is hexadecimal */
468 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
470 Uid_t_f /* Uid_t in decimal */
472 printf("who = %"Uid_t_f"\n", who);
474 Or you can try casting to a "wide enough" type:
476 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
478 Also remember that the C<%p> format really does require a void pointer:
481 printf("p = %p\n", (void*)p);
483 The gcc option C<-Wformat> scans for such problems.
487 Blindly using variadic macros
489 gcc has had them for a while with its own syntax, and C99 brought them
490 with a standardized syntax. Don't use the former, and use the latter
491 only if the HAS_C99_VARIADIC_MACROS is defined.
495 Blindly passing va_list
497 Not all platforms support passing va_list to further varargs (stdarg)
498 functions. The right thing to do is to copy the va_list using the
499 Perl_va_copy() if the NEED_VA_COPY is defined.
503 Using gcc statement expressions
505 val = ({...;...;...}); /* BAD */
507 While a nice extension, it's not portable. The Perl code does
508 admittedly use them if available to gain some extra speed (essentially
509 as a funky form of inlining), but you shouldn't.
513 Binding together several statements in a macro
515 Use the macros STMT_START and STMT_END.
523 Testing for operating systems or versions when should be testing for
526 #ifdef __FOONIX__ /* BAD */
530 Unless you know with 100% certainty that quux() is only ever available
531 for the "Foonix" operating system B<and> that is available B<and>
532 correctly working for B<all> past, present, B<and> future versions of
533 "Foonix", the above is very wrong. This is more correct (though still
534 not perfect, because the below is a compile-time check):
540 How does the HAS_QUUX become defined where it needs to be? Well, if
541 Foonix happens to be Unixy enough to be able to run the Configure
542 script, and Configure has been taught about detecting and testing
543 quux(), the HAS_QUUX will be correctly defined. In other platforms, the
544 corresponding configuration step will hopefully do the same.
546 In a pinch, if you cannot wait for Configure to be educated, or if you
547 have a good hunch of where quux() might be available, you can
548 temporarily try the following:
550 #if (defined(__FOONIX__) || defined(__BARNIX__))
560 But in any case, try to keep the features and operating systems
565 =head2 Problematic System Interfaces
571 malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
572 allocate at least one byte. (In general you should rarely need to work
573 at this low level, but instead use the various malloc wrappers.)
577 snprintf() - the return type is unportable. Use my_snprintf() instead.
581 =head2 Security problems
583 Last but not least, here are various tips for safer coding.
584 See also L<perlclib> for libc/stdio replacements one should use.
592 Or we will publicly ridicule you. Seriously.
598 Use mkstemp() instead.
602 Do not use strcpy() or strcat() or strncpy() or strncat()
604 Use my_strlcpy() and my_strlcat() instead: they either use the native
605 implementation, or Perl's own implementation (borrowed from the public
606 domain implementation of INN).
610 Do not use sprintf() or vsprintf()
612 If you really want just plain byte strings, use my_snprintf() and
613 my_vsnprintf() instead, which will try to use snprintf() and
614 vsnprintf() if those safer APIs are available. If you want something
615 fancier than a plain byte string, use
616 L<C<Perl_form>()|perlapi/form> or SVs and
617 L<C<Perl_sv_catpvf()>|perlapi/sv_catpvf>.
619 Note that glibc C<printf()>, C<sprintf()>, etc. are buggy before glibc
620 version 2.17. They won't allow a C<%.s> format with a precision to
621 create a string that isn't valid UTF-8 if the current underlying locale
622 of the program is UTF-8. What happens is that the C<%s> and its operand are
623 simply skipped without any notice.
624 L<https://sourceware.org/bugzilla/show_bug.cgi?id=6530>.
630 Use grok_atou() instead. atoi() has ill-defined behavior on overflows,
631 and cannot be used for incremental parsing. It is also affected by locale,
636 Do not use strtol() or strtoul()
638 Use grok_atou() instead. strtol() or strtoul() (or their IV/UV-friendly
639 macro disguises, Strtol() and Strtoul(), or Atol() and Atoul() are
640 affected by locale, which is bad.
646 You can compile a special debugging version of Perl, which allows you
647 to use the C<-D> option of Perl to tell more about what Perl is doing.
648 But sometimes there is no alternative than to dive in with a debugger,
649 either to see the stack trace of a core dump (very useful in a bug
650 report), or trying to figure out what went wrong before the core dump
651 happened, or how did we end up having wrong or unexpected results.
653 =head2 Poking at Perl
655 To really poke around with Perl, you'll probably want to build Perl for
656 debugging, like this:
658 ./Configure -d -D optimize=-g
661 C<-g> is a flag to the C compiler to have it produce debugging
662 information which will allow us to step through a running program, and
663 to see in which C function we are at (without the debugging information
664 we might see only the numerical addresses of the functions, which is
667 F<Configure> will also turn on the C<DEBUGGING> compilation symbol
668 which enables all the internal debugging code in Perl. There are a
669 whole bunch of things you can debug with this: L<perlrun> lists them
670 all, and the best way to find out about them is to play about with
671 them. The most useful options are probably
673 l Context (loop) stack processing
675 o Method and overloading resolution
676 c String/numeric conversions
678 Some of the functionality of the debugging code can be achieved using
681 -Dr => use re 'debug'
684 =head2 Using a source-level debugger
686 If the debugging output of C<-D> doesn't help you, it's time to step
687 through perl's execution with a source-level debugger.
693 We'll use C<gdb> for our examples here; the principles will apply to
694 any debugger (many vendors call their debugger C<dbx>), but check the
695 manual of the one you're using.
699 To fire up the debugger, type
703 Or if you have a core dump:
707 You'll want to do that in your Perl source tree so the debugger can
708 read the source code. You should see the copyright message, followed by
713 C<help> will get you into the documentation, but here are the most
720 Run the program with the given arguments.
722 =item * break function_name
724 =item * break source.c:xxx
726 Tells the debugger that we'll want to pause execution when we reach
727 either the named function (but see L<perlguts/Internal Functions>!) or
728 the given line in the named source file.
732 Steps through the program a line at a time.
736 Steps through the program a line at a time, without descending into
741 Run until the next breakpoint.
745 Run until the end of the current function, then stop again.
749 Just pressing Enter will do the most recent operation again - it's a
750 blessing when stepping through miles of source code.
754 Prints the C definition of the argument given.
760 OP *(*op_ppaddr)(void);
762 unsigned int op_type : 9;
763 unsigned int op_opt : 1;
764 unsigned int op_slabbed : 1;
765 unsigned int op_savefree : 1;
766 unsigned int op_static : 1;
767 unsigned int op_folded : 1;
768 unsigned int op_spare : 2;
775 Execute the given C code and print its results. B<WARNING>: Perl makes
776 heavy use of macros, and F<gdb> does not necessarily support macros
777 (see later L</"gdb macro support">). You'll have to substitute them
778 yourself, or to invoke cpp on the source code files (see L</"The .i
779 Targets">) So, for instance, you can't say
785 print Perl_sv_2pv_nolen(sv)
789 You may find it helpful to have a "macro dictionary", which you can
790 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
791 recursively apply those macros for you.
793 =head2 gdb macro support
795 Recent versions of F<gdb> have fairly good macro support, but in order
796 to use it you'll need to compile perl with macro definitions included
797 in the debugging information. Using F<gcc> version 3.1, this means
798 configuring with C<-Doptimize=-g3>. Other compilers might use a
799 different switch (if they support debugging macros at all).
801 =head2 Dumping Perl Data Structures
803 One way to get around this macro hell is to use the dumping functions
804 in F<dump.c>; these work a little like an internal
805 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other
806 structures that you can't get at from Perl. Let's take an example.
807 We'll use the C<$a = $b + $c> we used before, but give it a bit of
808 context: C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and
811 What about C<pp_add>, the function we examined earlier to implement the
814 (gdb) break Perl_pp_add
815 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
817 Notice we use C<Perl_pp_add> and not C<pp_add> - see
818 L<perlguts/Internal Functions>. With the breakpoint in place, we can
821 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
823 Lots of junk will go past as gdb reads in the relevant source files and
826 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
827 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
832 We looked at this bit of code before, and we said that
833 C<dPOPTOPnnrl_ul> arranges for two C<NV>s to be placed into C<left> and
834 C<right> - let's slightly expand it:
836 #define dPOPTOPnnrl_ul NV right = POPn; \
838 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
840 C<POPn> takes the SV from the top of the stack and obtains its NV
841 either directly (if C<SvNOK> is set) or by calling the C<sv_2nv>
842 function. C<TOPs> takes the next SV from the top of the stack - yes,
843 C<POPn> uses C<TOPs> - but doesn't remove it. We then use C<SvNV> to
844 get the NV from C<leftsv> in the same way as before - yes, C<POPn> uses
847 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
848 convert it. If we step again, we'll find ourselves there:
851 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
855 We can now use C<Perl_sv_dump> to investigate the SV:
857 (gdb) print Perl_sv_dump(sv)
858 SV = PV(0xa057cc0) at 0xa0675d0
861 PV = 0xa06a510 "6XXXX"\0
866 We know we're going to get C<6> from this, so let's finish the
870 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
871 0x462669 in Perl_pp_add () at pp_hot.c:311
874 We can also dump out this op: the current op is always stored in
875 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
876 similar output to L<B::Debug|B::Debug>.
878 (gdb) print Perl_op_dump(PL_op)
880 13 TYPE = add ===> 14
882 FLAGS = (SCALAR,KIDS)
884 TYPE = null ===> (12)
886 FLAGS = (SCALAR,KIDS)
888 11 TYPE = gvsv ===> 12
894 # finish this later #
896 =head2 Using gdb to look at specific parts of a program
898 With the example above, you knew to look for C<Perl_pp_add>, but what if
899 there were multiple calls to it all over the place, or you didn't know what
900 the op was you were looking for?
902 One way to do this is to inject a rare call somewhere near what you're looking
903 for. For example, you could add C<study> before your method:
909 (gdb) break Perl_pp_study
911 And then step until you hit what you're
912 looking for. This works well in a loop
913 if you want to only break at certain iterations:
919 =head2 Using gdb to look at what the parser/lexer are doing
921 If you want to see what perl is doing when parsing/lexing your code, you can
930 (gdb) break Perl_pp_study
932 If you want to see what the parser/lexer is doing inside of C<if> blocks and
933 the like you need to be a little trickier:
935 if ($a && $b && do { BEGIN { study } 1 } && $c) { ... }
937 =head1 SOURCE CODE STATIC ANALYSIS
939 Various tools exist for analysing C source code B<statically>, as
940 opposed to B<dynamically>, that is, without executing the code. It is
941 possible to detect resource leaks, undefined behaviour, type
942 mismatches, portability problems, code paths that would cause illegal
943 memory accesses, and other similar problems by just parsing the C code
944 and looking at the resulting graph, what does it tell about the
945 execution and data flows. As a matter of fact, this is exactly how C
946 compilers know to give warnings about dubious code.
950 The good old C code quality inspector, C<lint>, is available in several
951 platforms, but please be aware that there are several different
952 implementations of it by different vendors, which means that the flags
953 are not identical across different platforms.
955 There is a lint variant called C<splint> (Secure Programming Lint)
956 available from http://www.splint.org/ that should compile on any
959 There are C<lint> and <splint> targets in Makefile, but you may have to
960 diddle with the flags (see above).
964 Coverity (http://www.coverity.com/) is a product similar to lint and as
965 a testbed for their product they periodically check several open source
966 projects, and they give out accounts to open source developers to the
969 =head2 cpd (cut-and-paste detector)
971 The cpd tool detects cut-and-paste coding. If one instance of the
972 cut-and-pasted code changes, all the other spots should probably be
973 changed, too. Therefore such code should probably be turned into a
974 subroutine or a macro.
976 cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
977 (http://pmd.sourceforge.net/). pmd was originally written for static
978 analysis of Java code, but later the cpd part of it was extended to
979 parse also C and C++.
981 Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
982 pmd-X.Y.jar from it, and then run that on source code thusly:
984 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD \
985 --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
987 You may run into memory limits, in which case you should use the -Xmx
994 Though much can be written about the inconsistency and coverage
995 problems of gcc warnings (like C<-Wall> not meaning "all the warnings",
996 or some common portability problems not being covered by C<-Wall>, or
997 C<-ansi> and C<-pedantic> both being a poorly defined collection of
998 warnings, and so forth), gcc is still a useful tool in keeping our
1001 The C<-Wall> is by default on.
1003 The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1004 always, but unfortunately they are not safe on all platforms, they can
1005 for example cause fatal conflicts with the system headers (Solaris
1006 being a prime example). If Configure C<-Dgccansipedantic> is used, the
1007 C<cflags> frontend selects C<-ansi -pedantic> for the platforms where
1008 they are known to be safe.
1010 Starting from Perl 5.9.4 the following extra flags are added:
1024 C<-Wdeclaration-after-statement>
1028 The following flags would be nice to have but they would first need
1029 their own Augean stablemaster:
1043 C<-Wstrict-prototypes>
1047 The C<-Wtraditional> is another example of the annoying tendency of gcc
1048 to bundle a lot of warnings under one switch (it would be impossible to
1049 deploy in practice because it would complain a lot) but it does contain
1050 some warnings that would be beneficial to have available on their own,
1051 such as the warning about string constants inside macros containing the
1052 macro arguments: this behaved differently pre-ANSI than it does in
1053 ANSI, and some C compilers are still in transition, AIX being an
1056 =head2 Warnings of other C compilers
1058 Other C compilers (yes, there B<are> other C compilers than gcc) often
1059 have their "strict ANSI" or "strict ANSI with some portability
1060 extensions" modes on, like for example the Sun Workshop has its C<-Xa>
1061 mode on (though implicitly), or the DEC (these days, HP...) has its
1064 =head1 MEMORY DEBUGGERS
1066 B<NOTE 1>: Running under older memory debuggers such as Purify,
1067 valgrind or Third Degree greatly slows down the execution: seconds
1068 become minutes, minutes become hours. For example as of Perl 5.8.1, the
1069 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
1070 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
1071 than six hours, even on a snappy computer. The said test must be doing
1072 something that is quite unfriendly for memory debuggers. If you don't
1073 feel like waiting, that you can simply kill away the perl process.
1074 Roughly valgrind slows down execution by factor 10, AddressSanitizer by
1077 B<NOTE 2>: To minimize the number of memory leak false alarms (see
1078 L</PERL_DESTRUCT_LEVEL> for more information), you have to set the
1079 environment variable PERL_DESTRUCT_LEVEL to 2. For example, like this:
1081 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
1083 B<NOTE 3>: There are known memory leaks when there are compile-time
1084 errors within eval or require, seeing C<S_doeval> in the call stack is
1085 a good sign of these. Fixing these leaks is non-trivial, unfortunately,
1086 but they must be fixed eventually.
1088 B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
1089 unless Perl is built with the Configure option
1090 C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
1094 The valgrind tool can be used to find out both memory leaks and illegal
1095 heap memory accesses. As of version 3.3.0, Valgrind only supports Linux
1096 on x86, x86-64 and PowerPC and Darwin (OS X) on x86 and x86-64). The
1097 special "test.valgrind" target can be used to run the tests under
1098 valgrind. Found errors and memory leaks are logged in files named
1099 F<testfile.valgrind> and by default output is displayed inline.
1105 Since valgrind adds significant overhead, tests will take much longer to
1106 run. The valgrind tests support being run in parallel to help with this:
1108 TEST_JOBS=9 make test.valgrind
1110 Note that the above two invocations will be very verbose as reachable
1111 memory and leak-checking is enabled by default. If you want to just see
1114 VG_OPTS='-q --leak-check=no --show-reachable=no' TEST_JOBS=9 \
1117 Valgrind also provides a cachegrind tool, invoked on perl as:
1119 VG_OPTS=--tool=cachegrind make test.valgrind
1121 As system libraries (most notably glibc) are also triggering errors,
1122 valgrind allows to suppress such errors using suppression files. The
1123 default suppression file that comes with valgrind already catches a lot
1124 of them. Some additional suppressions are defined in F<t/perl.supp>.
1126 To get valgrind and for more information see
1128 http://valgrind.org/
1130 =head2 AddressSanitizer
1132 AddressSanitizer is a clang and gcc extension, included in clang since
1133 v3.1 and gcc since v4.8. It checks illegal heap pointers, global
1134 pointers, stack pointers and use after free errors, and is fast enough
1135 that you can easily compile your debugging or optimized perl with it.
1136 It does not check memory leaks though. AddressSanitizer is available
1137 for Linux, Mac OS X and soon on Windows.
1139 To build perl with AddressSanitizer, your Configure invocation should
1142 sh Configure -des -Dcc=clang \
1143 -Accflags=-faddress-sanitizer -Aldflags=-faddress-sanitizer \
1144 -Alddlflags=-shared\ -faddress-sanitizer
1146 where these arguments mean:
1152 This should be replaced by the full path to your clang executable if it
1153 is not in your path.
1155 =item * -Accflags=-faddress-sanitizer
1157 Compile perl and extensions sources with AddressSanitizer.
1159 =item * -Aldflags=-faddress-sanitizer
1161 Link the perl executable with AddressSanitizer.
1163 =item * -Alddlflags=-shared\ -faddress-sanitizer
1165 Link dynamic extensions with AddressSanitizer. You must manually
1166 specify C<-shared> because using C<-Alddlflags=-shared> will prevent
1167 Configure from setting a default value for C<lddlflags>, which usually
1168 contains C<-shared> (at least on Linux).
1173 L<http://code.google.com/p/address-sanitizer/wiki/AddressSanitizer>.
1178 Depending on your platform there are various ways of profiling Perl.
1180 There are two commonly used techniques of profiling executables:
1181 I<statistical time-sampling> and I<basic-block counting>.
1183 The first method takes periodically samples of the CPU program counter,
1184 and since the program counter can be correlated with the code generated
1185 for functions, we get a statistical view of in which functions the
1186 program is spending its time. The caveats are that very small/fast
1187 functions have lower probability of showing up in the profile, and that
1188 periodically interrupting the program (this is usually done rather
1189 frequently, in the scale of milliseconds) imposes an additional
1190 overhead that may skew the results. The first problem can be alleviated
1191 by running the code for longer (in general this is a good idea for
1192 profiling), the second problem is usually kept in guard by the
1193 profiling tools themselves.
1195 The second method divides up the generated code into I<basic blocks>.
1196 Basic blocks are sections of code that are entered only in the
1197 beginning and exited only at the end. For example, a conditional jump
1198 starts a basic block. Basic block profiling usually works by
1199 I<instrumenting> the code by adding I<enter basic block #nnnn>
1200 book-keeping code to the generated code. During the execution of the
1201 code the basic block counters are then updated appropriately. The
1202 caveat is that the added extra code can skew the results: again, the
1203 profiling tools usually try to factor their own effects out of the
1206 =head2 Gprof Profiling
1208 I<gprof> is a profiling tool available in many Unix platforms which
1209 uses I<statistical time-sampling>. You can build a profiled version of
1210 F<perl> by compiling using gcc with the flag C<-pg>. Either edit
1211 F<config.sh> or re-run F<Configure>. Running the profiled version of
1212 Perl will create an output file called F<gmon.out> which contains the
1213 profiling data collected during the execution.
1217 $ sh Configure -des -Dusedevel -Accflags='-pg' \
1218 -Aldflags='-pg' -Alddlflags='-pg -shared' \
1220 $ ./perl ... # creates gmon.out in current directory
1221 $ gprof ./perl > out
1224 (you probably need to add C<-shared> to the <-Alddlflags> line until RT
1225 #118199 is resolved)
1227 The F<gprof> tool can then display the collected data in various ways.
1228 Usually F<gprof> understands the following options:
1234 Suppress statically defined functions from the profile.
1238 Suppress the verbose descriptions in the profile.
1242 Exclude the given routine and its descendants from the profile.
1246 Display only the given routine and its descendants in the profile.
1250 Generate a summary file called F<gmon.sum> which then may be given to
1251 subsequent gprof runs to accumulate data over several runs.
1255 Display routines that have zero usage.
1259 For more detailed explanation of the available commands and output
1260 formats, see your own local documentation of F<gprof>.
1262 =head2 GCC gcov Profiling
1264 I<basic block profiling> is officially available in gcc 3.0 and later.
1265 You can build a profiled version of F<perl> by compiling using gcc with
1266 the flags C<-fprofile-arcs -ftest-coverage>. Either edit F<config.sh>
1267 or re-run F<Configure>.
1271 $ sh Configure -des -Dusedevel -Doptimize='-g' \
1272 -Accflags='-fprofile-arcs -ftest-coverage' \
1273 -Aldflags='-fprofile-arcs -ftest-coverage' \
1274 -Alddlflags='-fprofile-arcs -ftest-coverage -shared' \
1276 $ rm -f regexec.c.gcov regexec.gcda
1279 $ less regexec.c.gcov
1281 (you probably need to add C<-shared> to the <-Alddlflags> line until RT
1282 #118199 is resolved)
1284 Running the profiled version of Perl will cause profile output to be
1285 generated. For each source file an accompanying F<.gcda> file will be
1288 To display the results you use the I<gcov> utility (which should be
1289 installed if you have gcc 3.0 or newer installed). F<gcov> is run on
1290 source code files, like this
1294 which will cause F<sv.c.gcov> to be created. The F<.gcov> files contain
1295 the source code annotated with relative frequencies of execution
1296 indicated by "#" markers. If you want to generate F<.gcov> files for
1297 all profiled object files, you can run something like this:
1299 for file in `find . -name \*.gcno`
1300 do sh -c "cd `dirname $file` && gcov `basename $file .gcno`"
1303 Useful options of F<gcov> include C<-b> which will summarise the basic
1304 block, branch, and function call coverage, and C<-c> which instead of
1305 relative frequencies will use the actual counts. For more information
1306 on the use of F<gcov> and basic block profiling with gcc, see the
1307 latest GNU CC manual. As of gcc 4.8, this is at
1308 L<http://gcc.gnu.org/onlinedocs/gcc/Gcov-Intro.html#Gcov-Intro>
1310 =head1 MISCELLANEOUS TRICKS
1312 =head2 PERL_DESTRUCT_LEVEL
1314 If you want to run any of the tests yourself manually using e.g.
1315 valgrind, please note that by default perl B<does not> explicitly
1316 cleanup all the memory it has allocated (such as global memory arenas)
1317 but instead lets the exit() of the whole program "take care" of such
1318 allocations, also known as "global destruction of objects".
1320 There is a way to tell perl to do complete cleanup: set the environment
1321 variable PERL_DESTRUCT_LEVEL to a non-zero value. The t/TEST wrapper
1322 does set this to 2, and this is what you need to do too, if you don't
1323 want to see the "global leaks": For example, for running under valgrind
1325 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib t/foo/bar.t
1327 (Note: the mod_perl apache module uses also this environment variable
1328 for its own purposes and extended its semantics. Refer to the mod_perl
1329 documentation for more information. Also, spawned threads do the
1330 equivalent of setting this variable to the value 1.)
1332 If, at the end of a run you get the message I<N scalars leaked>, you
1333 can recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the
1334 addresses of all those leaked SVs to be dumped along with details as to
1335 where each SV was originally allocated. This information is also
1336 displayed by Devel::Peek. Note that the extra details recorded with
1337 each SV increases memory usage, so it shouldn't be used in production
1338 environments. It also converts C<new_SV()> from a macro into a real
1339 function, so you can use your favourite debugger to discover where
1340 those pesky SVs were allocated.
1342 If you see that you're leaking memory at runtime, but neither valgrind
1343 nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
1344 leaking SVs that are still reachable and will be properly cleaned up
1345 during destruction of the interpreter. In such cases, using the C<-Dm>
1346 switch can point you to the source of the leak. If the executable was
1347 built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV
1348 allocations in addition to memory allocations. Each SV allocation has a
1349 distinct serial number that will be written on creation and destruction
1350 of the SV. So if you're executing the leaking code in a loop, you need
1351 to look for SVs that are created, but never destroyed between each
1352 cycle. If such an SV is found, set a conditional breakpoint within
1353 C<new_SV()> and make it break only when C<PL_sv_serial> is equal to the
1354 serial number of the leaking SV. Then you will catch the interpreter in
1355 exactly the state where the leaking SV is allocated, which is
1356 sufficient in many cases to find the source of the leak.
1358 As C<-Dm> is using the PerlIO layer for output, it will by itself
1359 allocate quite a bunch of SVs, which are hidden to avoid recursion. You
1360 can bypass the PerlIO layer if you use the SV logging provided by
1361 C<-DPERL_MEM_LOG> instead.
1365 If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go
1366 through logging functions, which is handy for breakpoint setting.
1368 Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging functions
1369 read $ENV{PERL_MEM_LOG} to determine whether to log the event, and if
1372 $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops
1373 $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops
1374 $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log
1375 $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2)
1377 Memory logging is somewhat similar to C<-Dm> but is independent of
1378 C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(), and
1379 Safefree() are logged with the caller's source code file and line
1380 number (and C function name, if supported by the C compiler). In
1381 contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging is
1384 Since the logging doesn't use PerlIO, all SV allocations are logged and
1385 no extra SV allocations are introduced by enabling the logging. If
1386 compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for each SV
1387 allocation is also logged.
1391 Those debugging perl with the DDD frontend over gdb may find the
1394 You can extend the data conversion shortcuts menu, so for example you
1395 can display an SV's IV value with one click, without doing any typing.
1396 To do that simply edit ~/.ddd/init file and add after:
1398 ! Display shortcuts.
1399 Ddd*gdbDisplayShortcuts: \
1400 /t () // Convert to Bin\n\
1401 /d () // Convert to Dec\n\
1402 /x () // Convert to Hex\n\
1403 /o () // Convert to Oct(\n\
1405 the following two lines:
1407 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
1408 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
1410 so now you can do ivx and pvx lookups or you can plug there the sv_peek
1413 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
1415 (The my_perl is for threaded builds.) Just remember that every line,
1416 but the last one, should end with \n\
1418 Alternatively edit the init file interactively via: 3rd mouse button ->
1419 New Display -> Edit Menu
1421 Note: you can define up to 20 conversion shortcuts in the gdb section.
1425 On some platforms Perl supports retrieving the C level backtrace
1426 (similar to what symbolic debuggers like gdb do).
1428 The backtrace returns the stack trace of the C call frames,
1429 with the symbol names (function names), the object names (like "perl"),
1430 and if it can, also the source code locations (file:line).
1432 The supported platforms are Linux, and OS X (some *BSD might
1433 work at least partly, but they have not yet been tested).
1435 This feature hasn't been tested with multiple threads, but it will
1436 only show the backtrace of the thread doing the backtracing.
1438 The feature needs to be enabled with C<Configure -Dusecbacktrace>.
1440 The C<-Dusecbacktrace> also enables keeping the debug information when
1441 compiling/linking (often: C<-g>). Many compilers/linkers do support
1442 having both optimization and keeping the debug information. The debug
1443 information is needed for the symbol names and the source locations.
1445 Static functions might not be visible for the backtrace.
1447 Source code locations, even if available, can often be missing or
1448 misleading if the compiler has e.g. inlined code. Optimizer can
1449 make matching the source code and the object code quite challenging.
1455 You B<must> have the BFD (-lbfd) library installed, otherwise C<perl> will
1456 fail to link. The BFD is usually distributed as part of the GNU binutils.
1458 Summary: C<Configure ... -Dusecbacktrace>
1459 and you need C<-lbfd>.
1463 The source code locations are supported B<only> if you have
1464 the Developer Tools installed. (BFD is B<not> needed.)
1466 Summary: C<Configure ... -Dusecbacktrace>
1467 and installing the Developer Tools would be good.
1471 Optionally, for trying out the feature, you may want to enable
1472 automatic dumping of the backtrace just before a warning or croak (die)
1473 message is emitted, by adding C<-Accflags=-DUSE_C_BACKTRACE_ON_ERROR>
1476 Unless the above additional feature is enabled, nothing about the
1477 backtrace functionality is visible, except for the Perl/XS level.
1479 Furthermore, even if you have enabled this feature to be compiled,
1480 you need to enable it in runtime with an environment variable:
1481 C<PERL_C_BACKTRACE_ON_ERROR=10>. It must be an integer higher
1482 than zero, telling the desired frame count.
1484 Retrieving the backtrace from Perl level (using for example an XS
1485 extension) would be much less exciting than one would hope: normally
1486 you would see C<runops>, C<entersub>, and not much else. This API is
1487 intended to be called B<from within> the Perl implementation, not from
1488 Perl level execution.
1490 The C API for the backtrace is as follows:
1494 =item get_c_backtrace
1496 =item free_c_backtrace
1498 =item get_c_backtrace_dump
1500 =item dump_c_backtrace
1506 If you see in a debugger a memory area mysteriously full of 0xABABABAB
1507 or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros, see
1510 =head2 Read-only optrees
1512 Under ithreads the optree is read only. If you want to enforce this, to
1513 check for write accesses from buggy code, compile with
1514 C<-Accflags=-DPERL_DEBUG_READONLY_OPS>
1515 to enable code that allocates op memory
1516 via C<mmap>, and sets it read-only when it is attached to a subroutine.
1517 Any write access to an op results in a C<SIGBUS> and abort.
1519 This code is intended for development only, and may not be portable
1520 even to all Unix variants. Also, it is an 80% solution, in that it
1521 isn't able to make all ops read only. Specifically it does not apply to
1522 op slabs belonging to C<BEGIN> blocks.
1524 However, as an 80% solution it is still effective, as it has caught
1527 =head2 When is a bool not a bool?
1529 On pre-C99 compilers, C<bool> is defined as equivalent to C<char>.
1530 Consequently assignment of any larger type to a C<bool> is unsafe and may
1531 be truncated. The C<cBOOL> macro exists to cast it correctly.
1533 On those platforms and compilers where C<bool> really is a boolean (C++,
1534 C99), it is easy to forget the cast. You can force C<bool> to be a C<char>
1535 by compiling with C<-Accflags=-DPERL_BOOL_AS_CHAR>. You may also wish to
1536 run C<Configure> with something like
1538 -Accflags='-Wconversion -Wno-sign-conversion -Wno-shorten-64-to-32'
1540 or your compiler's equivalent to make it easier to spot any unsafe truncations
1543 =head2 The .i Targets
1545 You can expand the macros in a F<foo.c> file by saying
1549 which will expand the macros using cpp. Don't be scared by the
1554 This document was originally written by Nathan Torkington, and is
1555 maintained by the perl5-porters mailing list.