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, character set, 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 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 C<'A'>, but not C<0x41>.
279 You can safely say C<'\n'>, but not C<\012>. However, you can use
280 macros defined in F<utf8.h> to specify any code point portably.
281 C<LATIN1_TO_NATIVE(0xDF)> is going to be the code point that means
282 LATIN SMALL LETTER SHARP S on whatever platform you are running on (on
283 ASCII platforms it compiles without adding any extra code, so there is
284 zero performance hit on those). The acceptable inputs to
285 C<LATIN1_TO_NATIVE> are from C<0x00> through C<0xFF>. If your input
286 isn't guaranteed to be in that range, use C<UNICODE_TO_NATIVE> instead.
287 C<NATIVE_TO_LATIN1> and C<NATIVE_TO_UNICODE> translate the opposite
290 If you need the string representation of a character that doesn't have a
291 mnemonic name in C, you should add it to the list in
292 F<regen/unicode_constants.pl>, and have Perl create C<#define>'s for you,
293 based on the current platform.
295 Note that the C<isI<FOO>> and C<toI<FOO>> macros in F<handy.h> work
296 properly on native code points and strings.
298 Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper
299 case alphabetic characters. That is not true in EBCDIC. Nor for 'a' to
300 'z'. But '0' - '9' is an unbroken range in both systems. Don't assume
301 anything about other ranges. (Note that special handling of ranges in
302 regular expression patterns and transliterations makes it appear to Perl
303 code that the aforementioned ranges are all unbroken.)
305 Many of the comments in the existing code ignore the possibility of
306 EBCDIC, and may be wrong therefore, even if the code works. This is
307 actually a tribute to the successful transparent insertion of being
308 able to handle EBCDIC without having to change pre-existing code.
310 UTF-8 and UTF-EBCDIC are two different encodings used to represent
311 Unicode code points as sequences of bytes. Macros with the same names
312 (but different definitions) in F<utf8.h> and F<utfebcdic.h> are used to
313 allow the calling code to think that there is only one such encoding.
314 This is almost always referred to as C<utf8>, but it means the EBCDIC
315 version as well. Again, comments in the code may well be wrong even if
316 the code itself is right. For example, the concept of UTF-8 C<invariant
317 characters> differs between ASCII and EBCDIC. On ASCII platforms, only
318 characters that do not have the high-order bit set (i.e. whose ordinals
319 are strict ASCII, 0 - 127) are invariant, and the documentation and
320 comments in the code may assume that, often referring to something
321 like, say, C<hibit>. The situation differs and is not so simple on
322 EBCDIC machines, but as long as the code itself uses the
323 C<NATIVE_IS_INVARIANT()> macro appropriately, it works, even if the
326 As noted in L<perlhack/TESTING>, when writing test scripts, the file
327 F<t/charset_tools.pl> contains some helpful functions for writing tests
328 valid on both ASCII and EBCDIC platforms. Sometimes, though, a test
329 can't use a function and it's inconvenient to have different test
330 versions depending on the platform. There are 20 code points that are
331 the same in all 4 character sets currently recognized by Perl (the 3
332 EBCDIC code pages plus ISO 8859-1 (ASCII/Latin1)). These can be used in
333 such tests, though there is a small possibility that Perl will become
334 available in yet another character set, breaking your test. All but one
335 of these code points are C0 control characters. The most significant
336 controls that are the same are C<\0>, C<\r>, and C<\N{VT}> (also
337 specifiable as C<\cK>, C<\x0B>, C<\N{U+0B}>, or C<\013>). The single
338 non-control is U+00B6 PILCROW SIGN. The controls that are the same have
339 the same bit pattern in all 4 character sets, regardless of the UTF8ness
340 of the string containing them. The bit pattern for U+B6 is the same in
341 all 4 for non-UTF8 strings, but differs in each when its containing
342 string is UTF-8 encoded. The only other code points that have some sort
343 of sameness across all 4 character sets are the pair 0xDC and 0xFC.
344 Together these represent upper- and lowercase LATIN LETTER U WITH
345 DIAERESIS, but which is upper and which is lower may be reversed: 0xDC
346 is the capital in Latin1 and 0xFC is the small letter, while 0xFC is the
347 capital in EBCDIC and 0xDC is the small one. This factoid may be
348 exploited in writing case insensitive tests that are the same across all
353 Assuming the character set is just ASCII
355 ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra
356 characters have different meanings depending on the locale. Absent a
357 locale, currently these extra characters are generally considered to be
358 unassigned, and this has presented some problems. This has being
359 changed starting in 5.12 so that these characters can be considered to
360 be Latin-1 (ISO-8859-1).
364 Mixing #define and #ifdef
366 #define BURGLE(x) ... \
367 #ifdef BURGLE_OLD_STYLE /* BAD */
368 ... do it the old way ... \
370 ... do it the new way ... \
373 You cannot portably "stack" cpp directives. For example in the above
374 you need two separate BURGLE() #defines, one for each #ifdef branch.
378 Adding non-comment stuff after #endif or #else
382 #else !SNOSH /* BAD */
384 #endif SNOSH /* BAD */
386 The #endif and #else cannot portably have anything non-comment after
387 them. If you want to document what is going (which is a good idea
388 especially if the branches are long), use (C) comments:
396 The gcc option C<-Wendif-labels> warns about the bad variant (by
397 default on starting from Perl 5.9.4).
401 Having a comma after the last element of an enum list
409 is not portable. Leave out the last comma.
411 Also note that whether enums are implicitly morphable to ints varies
412 between compilers, you might need to (int).
418 // This function bamfoodles the zorklator. /* BAD */
420 That is C99 or C++. Perl is C89. Using the //-comments is silently
421 allowed by many C compilers but cranking up the ANSI C89 strictness
422 (which we like to do) causes the compilation to fail.
426 Mixing declarations and code
431 set_zorkmids(n); /* BAD */
434 That is C99 or C++. Some C compilers allow that, but you shouldn't.
436 The gcc option C<-Wdeclaration-after-statements> scans for such
437 problems (by default on starting from Perl 5.9.4).
441 Introducing variables inside for()
443 for(int i = ...; ...; ...) { /* BAD */
445 That is C99 or C++. While it would indeed be awfully nice to have that
446 also in C89, to limit the scope of the loop variable, alas, we cannot.
450 Mixing signed char pointers with unsigned char pointers
452 int foo(char *s) { ... }
454 unsigned char *t = ...; /* Or U8* t = ... */
457 While this is legal practice, it is certainly dubious, and downright
458 fatal in at least one platform: for example VMS cc considers this a
459 fatal error. One cause for people often making this mistake is that a
460 "naked char" and therefore dereferencing a "naked char pointer" have an
461 undefined signedness: it depends on the compiler and the flags of the
462 compiler and the underlying platform whether the result is signed or
463 unsigned. For this very same reason using a 'char' as an array index is
468 Macros that have string constants and their arguments as substrings of
471 #define FOO(n) printf("number = %d\n", n) /* BAD */
474 Pre-ANSI semantics for that was equivalent to
476 printf("10umber = %d\10");
478 which is probably not what you were expecting. Unfortunately at least
479 one reasonably common and modern C compiler does "real backward
480 compatibility" here, in AIX that is what still happens even though the
481 rest of the AIX compiler is very happily C89.
485 Using printf formats for non-basic C types
488 printf("i = %d\n", i); /* BAD */
490 While this might by accident work in some platform (where IV happens to
491 be an C<int>), in general it cannot. IV might be something larger. Even
492 worse the situation is with more specific types (defined by Perl's
493 configuration step in F<config.h>):
496 printf("who = %d\n", who); /* BAD */
498 The problem here is that Uid_t might be not only not C<int>-wide but it
499 might also be unsigned, in which case large uids would be printed as
502 There is no simple solution to this because of printf()'s limited
503 intelligence, but for many types the right format is available as with
504 either 'f' or '_f' suffix, for example:
506 IVdf /* IV in decimal */
507 UVxf /* UV is hexadecimal */
509 printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */
511 Uid_t_f /* Uid_t in decimal */
513 printf("who = %"Uid_t_f"\n", who);
515 Or you can try casting to a "wide enough" type:
517 printf("i = %"IVdf"\n", (IV)something_very_small_and_signed);
519 Also remember that the C<%p> format really does require a void pointer:
522 printf("p = %p\n", (void*)p);
524 The gcc option C<-Wformat> scans for such problems.
528 Blindly using variadic macros
530 gcc has had them for a while with its own syntax, and C99 brought them
531 with a standardized syntax. Don't use the former, and use the latter
532 only if the HAS_C99_VARIADIC_MACROS is defined.
536 Blindly passing va_list
538 Not all platforms support passing va_list to further varargs (stdarg)
539 functions. The right thing to do is to copy the va_list using the
540 Perl_va_copy() if the NEED_VA_COPY is defined.
544 Using gcc statement expressions
546 val = ({...;...;...}); /* BAD */
548 While a nice extension, it's not portable. The Perl code does
549 admittedly use them if available to gain some extra speed (essentially
550 as a funky form of inlining), but you shouldn't.
554 Binding together several statements in a macro
556 Use the macros STMT_START and STMT_END.
564 Testing for operating systems or versions when should be testing for
567 #ifdef __FOONIX__ /* BAD */
571 Unless you know with 100% certainty that quux() is only ever available
572 for the "Foonix" operating system B<and> that is available B<and>
573 correctly working for B<all> past, present, B<and> future versions of
574 "Foonix", the above is very wrong. This is more correct (though still
575 not perfect, because the below is a compile-time check):
581 How does the HAS_QUUX become defined where it needs to be? Well, if
582 Foonix happens to be Unixy enough to be able to run the Configure
583 script, and Configure has been taught about detecting and testing
584 quux(), the HAS_QUUX will be correctly defined. In other platforms, the
585 corresponding configuration step will hopefully do the same.
587 In a pinch, if you cannot wait for Configure to be educated, or if you
588 have a good hunch of where quux() might be available, you can
589 temporarily try the following:
591 #if (defined(__FOONIX__) || defined(__BARNIX__))
601 But in any case, try to keep the features and operating systems
606 Assuming the contents of static memory pointed to by the return values
607 of Perl wrappers for C library functions doesn't change. Many C library
608 functions return pointers to static storage that can be overwritten by
609 subsequent calls to the same or related functions. Perl has
610 light-weight wrappers for some of these functions, and which don't make
611 copies of the static memory. A good example is the interface to the
612 environment variables that are in effect for the program. Perl has
613 C<PerlEnv_getenv> to get values from the environment. But the return is
614 a pointer to static memory in the C library. If you are using the value
615 to immediately test for something, that's fine, but if you save the
616 value and expect it to be unchanged by later processing, you would be
617 wrong, but perhaps you wouldn't know it because different C library
618 implementations behave differently, and the one on the platform you're
619 testing on might work for your situation. But on some platforms, a
620 subsequent call to C<PerlEnv_getenv> or related function WILL overwrite
621 the memory that your first call points to. This has led to some
622 hard-to-debug problems. Do a L<perlapi/savepv> to make a copy, thus
623 avoiding these problems. You will have to free the copy when you're
624 done to avoid memory leaks. If you don't have control over when it gets
625 freed, you'll need to make the copy in a mortal scalar, like so:
627 if ((s = PerlEnv_getenv("foo") == NULL) {
628 ... /* handle NULL case */
631 s = SvPVX(sv_2mortal(newSVpv(s, 0)));
634 The above example works only if C<"s"> is C<NUL>-terminated; otherwise
635 you have to pass its length to C<newSVpv>.
639 =head2 Problematic System Interfaces
645 malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable
646 allocate at least one byte. (In general you should rarely need to work
647 at this low level, but instead use the various malloc wrappers.)
651 snprintf() - the return type is unportable. Use my_snprintf() instead.
655 =head2 Security problems
657 Last but not least, here are various tips for safer coding.
658 See also L<perlclib> for libc/stdio replacements one should use.
666 Or we will publicly ridicule you. Seriously.
672 Use mkstemp() instead.
676 Do not use strcpy() or strcat() or strncpy() or strncat()
678 Use my_strlcpy() and my_strlcat() instead: they either use the native
679 implementation, or Perl's own implementation (borrowed from the public
680 domain implementation of INN).
684 Do not use sprintf() or vsprintf()
686 If you really want just plain byte strings, use my_snprintf() and
687 my_vsnprintf() instead, which will try to use snprintf() and
688 vsnprintf() if those safer APIs are available. If you want something
689 fancier than a plain byte string, use
690 L<C<Perl_form>()|perlapi/form> or SVs and
691 L<C<Perl_sv_catpvf()>|perlapi/sv_catpvf>.
693 Note that glibc C<printf()>, C<sprintf()>, etc. are buggy before glibc
694 version 2.17. They won't allow a C<%.s> format with a precision to
695 create a string that isn't valid UTF-8 if the current underlying locale
696 of the program is UTF-8. What happens is that the C<%s> and its operand are
697 simply skipped without any notice.
698 L<https://sourceware.org/bugzilla/show_bug.cgi?id=6530>.
704 Use grok_atoUV() instead. atoi() has ill-defined behavior on overflows,
705 and cannot be used for incremental parsing. It is also affected by locale,
710 Do not use strtol() or strtoul()
712 Use grok_atoUV() instead. strtol() or strtoul() (or their IV/UV-friendly
713 macro disguises, Strtol() and Strtoul(), or Atol() and Atoul() are
714 affected by locale, which is bad.
720 You can compile a special debugging version of Perl, which allows you
721 to use the C<-D> option of Perl to tell more about what Perl is doing.
722 But sometimes there is no alternative than to dive in with a debugger,
723 either to see the stack trace of a core dump (very useful in a bug
724 report), or trying to figure out what went wrong before the core dump
725 happened, or how did we end up having wrong or unexpected results.
727 =head2 Poking at Perl
729 To really poke around with Perl, you'll probably want to build Perl for
730 debugging, like this:
732 ./Configure -d -D optimize=-g
735 C<-g> is a flag to the C compiler to have it produce debugging
736 information which will allow us to step through a running program, and
737 to see in which C function we are at (without the debugging information
738 we might see only the numerical addresses of the functions, which is
741 F<Configure> will also turn on the C<DEBUGGING> compilation symbol
742 which enables all the internal debugging code in Perl. There are a
743 whole bunch of things you can debug with this: L<perlrun> lists them
744 all, and the best way to find out about them is to play about with
745 them. The most useful options are probably
747 l Context (loop) stack processing
749 o Method and overloading resolution
750 c String/numeric conversions
752 Some of the functionality of the debugging code can be achieved using
755 -Dr => use re 'debug'
758 =head2 Using a source-level debugger
760 If the debugging output of C<-D> doesn't help you, it's time to step
761 through perl's execution with a source-level debugger.
767 We'll use C<gdb> for our examples here; the principles will apply to
768 any debugger (many vendors call their debugger C<dbx>), but check the
769 manual of the one you're using.
773 To fire up the debugger, type
777 Or if you have a core dump:
781 You'll want to do that in your Perl source tree so the debugger can
782 read the source code. You should see the copyright message, followed by
787 C<help> will get you into the documentation, but here are the most
794 Run the program with the given arguments.
796 =item * break function_name
798 =item * break source.c:xxx
800 Tells the debugger that we'll want to pause execution when we reach
801 either the named function (but see L<perlguts/Internal Functions>!) or
802 the given line in the named source file.
806 Steps through the program a line at a time.
810 Steps through the program a line at a time, without descending into
815 Run until the next breakpoint.
819 Run until the end of the current function, then stop again.
823 Just pressing Enter will do the most recent operation again - it's a
824 blessing when stepping through miles of source code.
828 Prints the C definition of the argument given.
834 OP *(*op_ppaddr)(void);
836 unsigned int op_type : 9;
837 unsigned int op_opt : 1;
838 unsigned int op_slabbed : 1;
839 unsigned int op_savefree : 1;
840 unsigned int op_static : 1;
841 unsigned int op_folded : 1;
842 unsigned int op_spare : 2;
849 Execute the given C code and print its results. B<WARNING>: Perl makes
850 heavy use of macros, and F<gdb> does not necessarily support macros
851 (see later L</"gdb macro support">). You'll have to substitute them
852 yourself, or to invoke cpp on the source code files (see L</"The .i
853 Targets">) So, for instance, you can't say
859 print Perl_sv_2pv_nolen(sv)
863 You may find it helpful to have a "macro dictionary", which you can
864 produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't
865 recursively apply those macros for you.
867 =head2 gdb macro support
869 Recent versions of F<gdb> have fairly good macro support, but in order
870 to use it you'll need to compile perl with macro definitions included
871 in the debugging information. Using F<gcc> version 3.1, this means
872 configuring with C<-Doptimize=-g3>. Other compilers might use a
873 different switch (if they support debugging macros at all).
875 =head2 Dumping Perl Data Structures
877 One way to get around this macro hell is to use the dumping functions
878 in F<dump.c>; these work a little like an internal
879 L<Devel::Peek|Devel::Peek>, but they also cover OPs and other
880 structures that you can't get at from Perl. Let's take an example.
881 We'll use the C<$a = $b + $c> we used before, but give it a bit of
882 context: C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and
885 What about C<pp_add>, the function we examined earlier to implement the
888 (gdb) break Perl_pp_add
889 Breakpoint 1 at 0x46249f: file pp_hot.c, line 309.
891 Notice we use C<Perl_pp_add> and not C<pp_add> - see
892 L<perlguts/Internal Functions>. With the breakpoint in place, we can
895 (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c'
897 Lots of junk will go past as gdb reads in the relevant source files and
900 Breakpoint 1, Perl_pp_add () at pp_hot.c:309
901 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
906 We looked at this bit of code before, and we said that
907 C<dPOPTOPnnrl_ul> arranges for two C<NV>s to be placed into C<left> and
908 C<right> - let's slightly expand it:
910 #define dPOPTOPnnrl_ul NV right = POPn; \
912 NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0
914 C<POPn> takes the SV from the top of the stack and obtains its NV
915 either directly (if C<SvNOK> is set) or by calling the C<sv_2nv>
916 function. C<TOPs> takes the next SV from the top of the stack - yes,
917 C<POPn> uses C<TOPs> - but doesn't remove it. We then use C<SvNV> to
918 get the NV from C<leftsv> in the same way as before - yes, C<POPn> uses
921 Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to
922 convert it. If we step again, we'll find ourselves there:
925 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669
929 We can now use C<Perl_sv_dump> to investigate the SV:
931 (gdb) print Perl_sv_dump(sv)
932 SV = PV(0xa057cc0) at 0xa0675d0
935 PV = 0xa06a510 "6XXXX"\0
940 We know we're going to get C<6> from this, so let's finish the
944 Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671
945 0x462669 in Perl_pp_add () at pp_hot.c:311
948 We can also dump out this op: the current op is always stored in
949 C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us
950 similar output to L<B::Debug|B::Debug>.
952 (gdb) print Perl_op_dump(PL_op)
954 13 TYPE = add ===> 14
956 FLAGS = (SCALAR,KIDS)
958 TYPE = null ===> (12)
960 FLAGS = (SCALAR,KIDS)
962 11 TYPE = gvsv ===> 12
968 # finish this later #
970 =head2 Using gdb to look at specific parts of a program
972 With the example above, you knew to look for C<Perl_pp_add>, but what if
973 there were multiple calls to it all over the place, or you didn't know what
974 the op was you were looking for?
976 One way to do this is to inject a rare call somewhere near what you're looking
977 for. For example, you could add C<study> before your method:
983 (gdb) break Perl_pp_study
985 And then step until you hit what you're
986 looking for. This works well in a loop
987 if you want to only break at certain iterations:
993 =head2 Using gdb to look at what the parser/lexer are doing
995 If you want to see what perl is doing when parsing/lexing your code, you can
1004 (gdb) break Perl_pp_study
1006 If you want to see what the parser/lexer is doing inside of C<if> blocks and
1007 the like you need to be a little trickier:
1009 if ($a && $b && do { BEGIN { study } 1 } && $c) { ... }
1011 =head1 SOURCE CODE STATIC ANALYSIS
1013 Various tools exist for analysing C source code B<statically>, as
1014 opposed to B<dynamically>, that is, without executing the code. It is
1015 possible to detect resource leaks, undefined behaviour, type
1016 mismatches, portability problems, code paths that would cause illegal
1017 memory accesses, and other similar problems by just parsing the C code
1018 and looking at the resulting graph, what does it tell about the
1019 execution and data flows. As a matter of fact, this is exactly how C
1020 compilers know to give warnings about dubious code.
1024 The good old C code quality inspector, C<lint>, is available in several
1025 platforms, but please be aware that there are several different
1026 implementations of it by different vendors, which means that the flags
1027 are not identical across different platforms.
1029 There is a lint variant called C<splint> (Secure Programming Lint)
1030 available from http://www.splint.org/ that should compile on any
1033 There are C<lint> and <splint> targets in Makefile, but you may have to
1034 diddle with the flags (see above).
1038 Coverity (http://www.coverity.com/) is a product similar to lint and as
1039 a testbed for their product they periodically check several open source
1040 projects, and they give out accounts to open source developers to the
1043 =head2 cpd (cut-and-paste detector)
1045 The cpd tool detects cut-and-paste coding. If one instance of the
1046 cut-and-pasted code changes, all the other spots should probably be
1047 changed, too. Therefore such code should probably be turned into a
1048 subroutine or a macro.
1050 cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project
1051 (http://pmd.sourceforge.net/). pmd was originally written for static
1052 analysis of Java code, but later the cpd part of it was extended to
1053 parse also C and C++.
1055 Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the
1056 pmd-X.Y.jar from it, and then run that on source code thusly:
1058 java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD \
1059 --minimum-tokens 100 --files /some/where/src --language c > cpd.txt
1061 You may run into memory limits, in which case you should use the -Xmx
1068 Though much can be written about the inconsistency and coverage
1069 problems of gcc warnings (like C<-Wall> not meaning "all the warnings",
1070 or some common portability problems not being covered by C<-Wall>, or
1071 C<-ansi> and C<-pedantic> both being a poorly defined collection of
1072 warnings, and so forth), gcc is still a useful tool in keeping our
1075 The C<-Wall> is by default on.
1077 The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on
1078 always, but unfortunately they are not safe on all platforms, they can
1079 for example cause fatal conflicts with the system headers (Solaris
1080 being a prime example). If Configure C<-Dgccansipedantic> is used, the
1081 C<cflags> frontend selects C<-ansi -pedantic> for the platforms where
1082 they are known to be safe.
1084 Starting from Perl 5.9.4 the following extra flags are added:
1098 C<-Wdeclaration-after-statement>
1102 The following flags would be nice to have but they would first need
1103 their own Augean stablemaster:
1117 C<-Wstrict-prototypes>
1121 The C<-Wtraditional> is another example of the annoying tendency of gcc
1122 to bundle a lot of warnings under one switch (it would be impossible to
1123 deploy in practice because it would complain a lot) but it does contain
1124 some warnings that would be beneficial to have available on their own,
1125 such as the warning about string constants inside macros containing the
1126 macro arguments: this behaved differently pre-ANSI than it does in
1127 ANSI, and some C compilers are still in transition, AIX being an
1130 =head2 Warnings of other C compilers
1132 Other C compilers (yes, there B<are> other C compilers than gcc) often
1133 have their "strict ANSI" or "strict ANSI with some portability
1134 extensions" modes on, like for example the Sun Workshop has its C<-Xa>
1135 mode on (though implicitly), or the DEC (these days, HP...) has its
1138 =head1 MEMORY DEBUGGERS
1140 B<NOTE 1>: Running under older memory debuggers such as Purify,
1141 valgrind or Third Degree greatly slows down the execution: seconds
1142 become minutes, minutes become hours. For example as of Perl 5.8.1, the
1143 ext/Encode/t/Unicode.t takes extraordinarily long to complete under
1144 e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more
1145 than six hours, even on a snappy computer. The said test must be doing
1146 something that is quite unfriendly for memory debuggers. If you don't
1147 feel like waiting, that you can simply kill away the perl process.
1148 Roughly valgrind slows down execution by factor 10, AddressSanitizer by
1151 B<NOTE 2>: To minimize the number of memory leak false alarms (see
1152 L</PERL_DESTRUCT_LEVEL> for more information), you have to set the
1153 environment variable PERL_DESTRUCT_LEVEL to 2. For example, like this:
1155 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ...
1157 B<NOTE 3>: There are known memory leaks when there are compile-time
1158 errors within eval or require, seeing C<S_doeval> in the call stack is
1159 a good sign of these. Fixing these leaks is non-trivial, unfortunately,
1160 but they must be fixed eventually.
1162 B<NOTE 4>: L<DynaLoader> will not clean up after itself completely
1163 unless Perl is built with the Configure option
1164 C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>.
1168 The valgrind tool can be used to find out both memory leaks and illegal
1169 heap memory accesses. As of version 3.3.0, Valgrind only supports Linux
1170 on x86, x86-64 and PowerPC and Darwin (OS X) on x86 and x86-64. The
1171 special "test.valgrind" target can be used to run the tests under
1172 valgrind. Found errors and memory leaks are logged in files named
1173 F<testfile.valgrind> and by default output is displayed inline.
1179 Since valgrind adds significant overhead, tests will take much longer to
1180 run. The valgrind tests support being run in parallel to help with this:
1182 TEST_JOBS=9 make test.valgrind
1184 Note that the above two invocations will be very verbose as reachable
1185 memory and leak-checking is enabled by default. If you want to just see
1188 VG_OPTS='-q --leak-check=no --show-reachable=no' TEST_JOBS=9 \
1191 Valgrind also provides a cachegrind tool, invoked on perl as:
1193 VG_OPTS=--tool=cachegrind make test.valgrind
1195 As system libraries (most notably glibc) are also triggering errors,
1196 valgrind allows to suppress such errors using suppression files. The
1197 default suppression file that comes with valgrind already catches a lot
1198 of them. Some additional suppressions are defined in F<t/perl.supp>.
1200 To get valgrind and for more information see
1202 http://valgrind.org/
1204 =head2 AddressSanitizer
1206 AddressSanitizer is a clang and gcc extension, included in clang since
1207 v3.1 and gcc since v4.8. It checks illegal heap pointers, global
1208 pointers, stack pointers and use after free errors, and is fast enough
1209 that you can easily compile your debugging or optimized perl with it.
1210 It does not check memory leaks though. AddressSanitizer is available
1211 for Linux, Mac OS X and soon on Windows.
1213 To build perl with AddressSanitizer, your Configure invocation should
1216 sh Configure -des -Dcc=clang \
1217 -Accflags=-faddress-sanitizer -Aldflags=-faddress-sanitizer \
1218 -Alddlflags=-shared\ -faddress-sanitizer
1220 where these arguments mean:
1226 This should be replaced by the full path to your clang executable if it
1227 is not in your path.
1229 =item * -Accflags=-faddress-sanitizer
1231 Compile perl and extensions sources with AddressSanitizer.
1233 =item * -Aldflags=-faddress-sanitizer
1235 Link the perl executable with AddressSanitizer.
1237 =item * -Alddlflags=-shared\ -faddress-sanitizer
1239 Link dynamic extensions with AddressSanitizer. You must manually
1240 specify C<-shared> because using C<-Alddlflags=-shared> will prevent
1241 Configure from setting a default value for C<lddlflags>, which usually
1242 contains C<-shared> (at least on Linux).
1247 L<http://code.google.com/p/address-sanitizer/wiki/AddressSanitizer>.
1252 Depending on your platform there are various ways of profiling Perl.
1254 There are two commonly used techniques of profiling executables:
1255 I<statistical time-sampling> and I<basic-block counting>.
1257 The first method takes periodically samples of the CPU program counter,
1258 and since the program counter can be correlated with the code generated
1259 for functions, we get a statistical view of in which functions the
1260 program is spending its time. The caveats are that very small/fast
1261 functions have lower probability of showing up in the profile, and that
1262 periodically interrupting the program (this is usually done rather
1263 frequently, in the scale of milliseconds) imposes an additional
1264 overhead that may skew the results. The first problem can be alleviated
1265 by running the code for longer (in general this is a good idea for
1266 profiling), the second problem is usually kept in guard by the
1267 profiling tools themselves.
1269 The second method divides up the generated code into I<basic blocks>.
1270 Basic blocks are sections of code that are entered only in the
1271 beginning and exited only at the end. For example, a conditional jump
1272 starts a basic block. Basic block profiling usually works by
1273 I<instrumenting> the code by adding I<enter basic block #nnnn>
1274 book-keeping code to the generated code. During the execution of the
1275 code the basic block counters are then updated appropriately. The
1276 caveat is that the added extra code can skew the results: again, the
1277 profiling tools usually try to factor their own effects out of the
1280 =head2 Gprof Profiling
1282 I<gprof> is a profiling tool available in many Unix platforms which
1283 uses I<statistical time-sampling>. You can build a profiled version of
1284 F<perl> by compiling using gcc with the flag C<-pg>. Either edit
1285 F<config.sh> or re-run F<Configure>. Running the profiled version of
1286 Perl will create an output file called F<gmon.out> which contains the
1287 profiling data collected during the execution.
1291 $ sh Configure -des -Dusedevel -Accflags='-pg' \
1292 -Aldflags='-pg' -Alddlflags='-pg -shared' \
1294 $ ./perl ... # creates gmon.out in current directory
1295 $ gprof ./perl > out
1298 (you probably need to add C<-shared> to the <-Alddlflags> line until RT
1299 #118199 is resolved)
1301 The F<gprof> tool can then display the collected data in various ways.
1302 Usually F<gprof> understands the following options:
1308 Suppress statically defined functions from the profile.
1312 Suppress the verbose descriptions in the profile.
1316 Exclude the given routine and its descendants from the profile.
1320 Display only the given routine and its descendants in the profile.
1324 Generate a summary file called F<gmon.sum> which then may be given to
1325 subsequent gprof runs to accumulate data over several runs.
1329 Display routines that have zero usage.
1333 For more detailed explanation of the available commands and output
1334 formats, see your own local documentation of F<gprof>.
1336 =head2 GCC gcov Profiling
1338 I<basic block profiling> is officially available in gcc 3.0 and later.
1339 You can build a profiled version of F<perl> by compiling using gcc with
1340 the flags C<-fprofile-arcs -ftest-coverage>. Either edit F<config.sh>
1341 or re-run F<Configure>.
1345 $ sh Configure -des -Dusedevel -Doptimize='-g' \
1346 -Accflags='-fprofile-arcs -ftest-coverage' \
1347 -Aldflags='-fprofile-arcs -ftest-coverage' \
1348 -Alddlflags='-fprofile-arcs -ftest-coverage -shared' \
1350 $ rm -f regexec.c.gcov regexec.gcda
1353 $ less regexec.c.gcov
1355 (you probably need to add C<-shared> to the <-Alddlflags> line until RT
1356 #118199 is resolved)
1358 Running the profiled version of Perl will cause profile output to be
1359 generated. For each source file an accompanying F<.gcda> file will be
1362 To display the results you use the I<gcov> utility (which should be
1363 installed if you have gcc 3.0 or newer installed). F<gcov> is run on
1364 source code files, like this
1368 which will cause F<sv.c.gcov> to be created. The F<.gcov> files contain
1369 the source code annotated with relative frequencies of execution
1370 indicated by "#" markers. If you want to generate F<.gcov> files for
1371 all profiled object files, you can run something like this:
1373 for file in `find . -name \*.gcno`
1374 do sh -c "cd `dirname $file` && gcov `basename $file .gcno`"
1377 Useful options of F<gcov> include C<-b> which will summarise the basic
1378 block, branch, and function call coverage, and C<-c> which instead of
1379 relative frequencies will use the actual counts. For more information
1380 on the use of F<gcov> and basic block profiling with gcc, see the
1381 latest GNU CC manual. As of gcc 4.8, this is at
1382 L<http://gcc.gnu.org/onlinedocs/gcc/Gcov-Intro.html#Gcov-Intro>
1384 =head1 MISCELLANEOUS TRICKS
1386 =head2 PERL_DESTRUCT_LEVEL
1388 If you want to run any of the tests yourself manually using e.g.
1389 valgrind, please note that by default perl B<does not> explicitly
1390 cleanup all the memory it has allocated (such as global memory arenas)
1391 but instead lets the exit() of the whole program "take care" of such
1392 allocations, also known as "global destruction of objects".
1394 There is a way to tell perl to do complete cleanup: set the environment
1395 variable PERL_DESTRUCT_LEVEL to a non-zero value. The t/TEST wrapper
1396 does set this to 2, and this is what you need to do too, if you don't
1397 want to see the "global leaks": For example, for running under valgrind
1399 env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib t/foo/bar.t
1401 (Note: the mod_perl apache module uses also this environment variable
1402 for its own purposes and extended its semantics. Refer to the mod_perl
1403 documentation for more information. Also, spawned threads do the
1404 equivalent of setting this variable to the value 1.)
1406 If, at the end of a run you get the message I<N scalars leaked>, you
1407 can recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the
1408 addresses of all those leaked SVs to be dumped along with details as to
1409 where each SV was originally allocated. This information is also
1410 displayed by Devel::Peek. Note that the extra details recorded with
1411 each SV increases memory usage, so it shouldn't be used in production
1412 environments. It also converts C<new_SV()> from a macro into a real
1413 function, so you can use your favourite debugger to discover where
1414 those pesky SVs were allocated.
1416 If you see that you're leaking memory at runtime, but neither valgrind
1417 nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably
1418 leaking SVs that are still reachable and will be properly cleaned up
1419 during destruction of the interpreter. In such cases, using the C<-Dm>
1420 switch can point you to the source of the leak. If the executable was
1421 built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV
1422 allocations in addition to memory allocations. Each SV allocation has a
1423 distinct serial number that will be written on creation and destruction
1424 of the SV. So if you're executing the leaking code in a loop, you need
1425 to look for SVs that are created, but never destroyed between each
1426 cycle. If such an SV is found, set a conditional breakpoint within
1427 C<new_SV()> and make it break only when C<PL_sv_serial> is equal to the
1428 serial number of the leaking SV. Then you will catch the interpreter in
1429 exactly the state where the leaking SV is allocated, which is
1430 sufficient in many cases to find the source of the leak.
1432 As C<-Dm> is using the PerlIO layer for output, it will by itself
1433 allocate quite a bunch of SVs, which are hidden to avoid recursion. You
1434 can bypass the PerlIO layer if you use the SV logging provided by
1435 C<-DPERL_MEM_LOG> instead.
1439 If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go
1440 through logging functions, which is handy for breakpoint setting.
1442 Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging functions
1443 read $ENV{PERL_MEM_LOG} to determine whether to log the event, and if
1446 $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops
1447 $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops
1448 $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log
1449 $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2)
1451 Memory logging is somewhat similar to C<-Dm> but is independent of
1452 C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(), and
1453 Safefree() are logged with the caller's source code file and line
1454 number (and C function name, if supported by the C compiler). In
1455 contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging is
1458 Since the logging doesn't use PerlIO, all SV allocations are logged and
1459 no extra SV allocations are introduced by enabling the logging. If
1460 compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for each SV
1461 allocation is also logged.
1465 Those debugging perl with the DDD frontend over gdb may find the
1468 You can extend the data conversion shortcuts menu, so for example you
1469 can display an SV's IV value with one click, without doing any typing.
1470 To do that simply edit ~/.ddd/init file and add after:
1472 ! Display shortcuts.
1473 Ddd*gdbDisplayShortcuts: \
1474 /t () // Convert to Bin\n\
1475 /d () // Convert to Dec\n\
1476 /x () // Convert to Hex\n\
1477 /o () // Convert to Oct(\n\
1479 the following two lines:
1481 ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\
1482 ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx
1484 so now you can do ivx and pvx lookups or you can plug there the sv_peek
1487 Perl_sv_peek(my_perl, (SV*)()) // sv_peek
1489 (The my_perl is for threaded builds.) Just remember that every line,
1490 but the last one, should end with \n\
1492 Alternatively edit the init file interactively via: 3rd mouse button ->
1493 New Display -> Edit Menu
1495 Note: you can define up to 20 conversion shortcuts in the gdb section.
1499 On some platforms Perl supports retrieving the C level backtrace
1500 (similar to what symbolic debuggers like gdb do).
1502 The backtrace returns the stack trace of the C call frames,
1503 with the symbol names (function names), the object names (like "perl"),
1504 and if it can, also the source code locations (file:line).
1506 The supported platforms are Linux, and OS X (some *BSD might
1507 work at least partly, but they have not yet been tested).
1509 This feature hasn't been tested with multiple threads, but it will
1510 only show the backtrace of the thread doing the backtracing.
1512 The feature needs to be enabled with C<Configure -Dusecbacktrace>.
1514 The C<-Dusecbacktrace> also enables keeping the debug information when
1515 compiling/linking (often: C<-g>). Many compilers/linkers do support
1516 having both optimization and keeping the debug information. The debug
1517 information is needed for the symbol names and the source locations.
1519 Static functions might not be visible for the backtrace.
1521 Source code locations, even if available, can often be missing or
1522 misleading if the compiler has e.g. inlined code. Optimizer can
1523 make matching the source code and the object code quite challenging.
1529 You B<must> have the BFD (-lbfd) library installed, otherwise C<perl> will
1530 fail to link. The BFD is usually distributed as part of the GNU binutils.
1532 Summary: C<Configure ... -Dusecbacktrace>
1533 and you need C<-lbfd>.
1537 The source code locations are supported B<only> if you have
1538 the Developer Tools installed. (BFD is B<not> needed.)
1540 Summary: C<Configure ... -Dusecbacktrace>
1541 and installing the Developer Tools would be good.
1545 Optionally, for trying out the feature, you may want to enable
1546 automatic dumping of the backtrace just before a warning or croak (die)
1547 message is emitted, by adding C<-Accflags=-DUSE_C_BACKTRACE_ON_ERROR>
1550 Unless the above additional feature is enabled, nothing about the
1551 backtrace functionality is visible, except for the Perl/XS level.
1553 Furthermore, even if you have enabled this feature to be compiled,
1554 you need to enable it in runtime with an environment variable:
1555 C<PERL_C_BACKTRACE_ON_ERROR=10>. It must be an integer higher
1556 than zero, telling the desired frame count.
1558 Retrieving the backtrace from Perl level (using for example an XS
1559 extension) would be much less exciting than one would hope: normally
1560 you would see C<runops>, C<entersub>, and not much else. This API is
1561 intended to be called B<from within> the Perl implementation, not from
1562 Perl level execution.
1564 The C API for the backtrace is as follows:
1568 =item get_c_backtrace
1570 =item free_c_backtrace
1572 =item get_c_backtrace_dump
1574 =item dump_c_backtrace
1580 If you see in a debugger a memory area mysteriously full of 0xABABABAB
1581 or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros, see
1584 =head2 Read-only optrees
1586 Under ithreads the optree is read only. If you want to enforce this, to
1587 check for write accesses from buggy code, compile with
1588 C<-Accflags=-DPERL_DEBUG_READONLY_OPS>
1589 to enable code that allocates op memory
1590 via C<mmap>, and sets it read-only when it is attached to a subroutine.
1591 Any write access to an op results in a C<SIGBUS> and abort.
1593 This code is intended for development only, and may not be portable
1594 even to all Unix variants. Also, it is an 80% solution, in that it
1595 isn't able to make all ops read only. Specifically it does not apply to
1596 op slabs belonging to C<BEGIN> blocks.
1598 However, as an 80% solution it is still effective, as it has caught
1601 =head2 When is a bool not a bool?
1603 On pre-C99 compilers, C<bool> is defined as equivalent to C<char>.
1604 Consequently assignment of any larger type to a C<bool> is unsafe and may
1605 be truncated. The C<cBOOL> macro exists to cast it correctly.
1607 On those platforms and compilers where C<bool> really is a boolean (C++,
1608 C99), it is easy to forget the cast. You can force C<bool> to be a C<char>
1609 by compiling with C<-Accflags=-DPERL_BOOL_AS_CHAR>. You may also wish to
1610 run C<Configure> with something like
1612 -Accflags='-Wconversion -Wno-sign-conversion -Wno-shorten-64-to-32'
1614 or your compiler's equivalent to make it easier to spot any unsafe truncations
1617 =head2 The .i Targets
1619 You can expand the macros in a F<foo.c> file by saying
1623 which will expand the macros using cpp. Don't be scared by the
1628 This document was originally written by Nathan Torkington, and is
1629 maintained by the perl5-porters mailing list.