3 perlguts - Introduction to the Perl API
7 This document attempts to describe how to use the Perl API, as well as
8 to provide some info on the basic workings of the Perl core. It is far
9 from complete and probably contains many errors. Please refer any
10 questions or comments to the author below.
16 Perl has three typedefs that handle Perl's three main data types:
22 Each typedef has specific routines that manipulate the various data types.
24 =for apidoc_section $AV
26 =for apidoc_section $HV
28 =for apidoc_section $SV
31 =head2 What is an "IV"?
33 Perl uses a special typedef IV which is a simple signed integer type that is
34 guaranteed to be large enough to hold a pointer (as well as an integer).
35 Additionally, there is the UV, which is simply an unsigned IV.
37 Perl also uses several special typedefs to declare variables to hold
38 integers of (at least) a given size.
39 Use I8, I16, I32, and I64 to declare a signed integer variable which has
40 at least as many bits as the number in its name. These all evaluate to
41 the native C type that is closest to the given number of bits, but no
42 smaller than that number. For example, on many platforms, a C<short> is
43 16 bits long, and if so, I16 will evaluate to a C<short>. But on
44 platforms where a C<short> isn't exactly 16 bits, Perl will use the
45 smallest type that contains 16 bits or more.
47 U8, U16, U32, and U64 are to declare the corresponding unsigned integer
50 If the platform doesn't support 64-bit integers, both I64 and U64 will
51 be undefined. Use IV and UV to declare the largest practicable, and
52 C<L<perlapi/WIDEST_UTYPE>> for the absolute maximum unsigned, but which
53 may not be usable in all circumstances.
55 A numeric constant can be specified with L<perlapi/C<INT16_C>>,
56 L<perlapi/C<UINTMAX_C>>, and similar.
58 =for apidoc_section $integer
60 =for apidoc_item ||I16
61 =for apidoc_item ||I32
62 =for apidoc_item ||I64
66 =for apidoc_item ||U16
67 =for apidoc_item ||U32
68 =for apidoc_item ||U64
71 =head2 Working with SVs
73 An SV can be created and loaded with one command. There are five types of
74 values that can be loaded: an integer value (IV), an unsigned integer
75 value (UV), a double (NV), a string (PV), and another scalar (SV).
76 ("PV" stands for "Pointer Value". You might think that it is misnamed
77 because it is described as pointing only to strings. However, it is
78 possible to have it point to other things. For example, it could point
79 to an array of UVs. But,
80 using it for non-strings requires care, as the underlying assumption of
81 much of the internals is that PVs are just for strings. Often, for
82 example, a trailing C<NUL> is tacked on automatically. The non-string use
83 is documented only in this paragraph.)
87 The seven routines are:
92 SV* newSVpv(const char*, STRLEN);
93 SV* newSVpvn(const char*, STRLEN);
94 SV* newSVpvf(const char*, ...);
97 C<STRLEN> is an integer type (C<Size_t>, usually defined as C<size_t> in
98 F<config.h>) guaranteed to be large enough to represent the size of
99 any string that perl can handle.
101 =for apidoc Ayh||STRLEN
103 In the unlikely case of a SV requiring more complex initialization, you
104 can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
105 type NULL is returned, else an SV of type PV is returned with len + 1 (for
106 the C<NUL>) bytes of storage allocated, accessible via SvPVX. In both cases
107 the SV has the undef value.
109 SV *sv = newSV(0); /* no storage allocated */
110 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage
113 To change the value of an I<already-existing> SV, there are eight routines:
115 void sv_setiv(SV*, IV);
116 void sv_setuv(SV*, UV);
117 void sv_setnv(SV*, double);
118 void sv_setpv(SV*, const char*);
119 void sv_setpvn(SV*, const char*, STRLEN)
120 void sv_setpvf(SV*, const char*, ...);
121 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
122 SV **, Size_t, bool *);
123 void sv_setsv(SV*, SV*);
125 Notice that you can choose to specify the length of the string to be
126 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
127 allow Perl to calculate the length by using C<sv_setpv> or by specifying
128 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
129 determine the string's length by using C<strlen>, which depends on the
130 string terminating with a C<NUL> character, and not otherwise containing
133 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
134 formatted output becomes the value.
136 C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
137 either a pointer to a variable argument list or the address and length of
138 an array of SVs. The last argument points to a boolean; on return, if that
139 boolean is true, then locale-specific information has been used to format
140 the string, and the string's contents are therefore untrustworthy (see
141 L<perlsec>). This pointer may be NULL if that information is not
142 important. Note that this function requires you to specify the length of
145 The C<sv_set*()> functions are not generic enough to operate on values
146 that have "magic". See L</Magic Virtual Tables> later in this document.
148 All SVs that contain strings should be terminated with a C<NUL> character.
149 If it is not C<NUL>-terminated there is a risk of
150 core dumps and corruptions from code which passes the string to C
151 functions or system calls which expect a C<NUL>-terminated string.
152 Perl's own functions typically add a trailing C<NUL> for this reason.
153 Nevertheless, you should be very careful when you pass a string stored
154 in an SV to a C function or system call.
156 To access the actual value that an SV points to, you can use the macros:
161 SvPV(SV*, STRLEN len)
164 which will automatically coerce the actual scalar type into an IV, UV, double,
167 In the C<SvPV> macro, the length of the string returned is placed into the
168 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
169 not care what the length of the data is, use the C<SvPV_nolen> macro.
170 Historically the C<SvPV> macro with the global variable C<PL_na> has been
171 used in this case. But that can be quite inefficient because C<PL_na> must
172 be accessed in thread-local storage in threaded Perl. In any case, remember
173 that Perl allows arbitrary strings of data that may both contain NULs and
174 might not be terminated by a C<NUL>.
176 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
177 len);>. It might work with your
178 compiler, but it won't work for everyone.
179 Break this sort of statement up into separate assignments:
187 If you want to know if the scalar value is TRUE, you can use:
191 Although Perl will automatically grow strings for you, if you need to force
192 Perl to allocate more memory for your SV, you can use the macro
194 SvGROW(SV*, STRLEN newlen)
196 which will determine if more memory needs to be allocated. If so, it will
197 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
198 decrease, the allocated memory of an SV and that it does not automatically
199 add space for the trailing C<NUL> byte (perl's own string functions typically do
200 C<SvGROW(sv, len + 1)>).
202 If you want to write to an existing SV's buffer and set its value to a
203 string, use SvPV_force() or one of its variants to force the SV to be
204 a PV. This will remove any of various types of non-stringness from
205 the SV while preserving the content of the SV in the PV. This can be
206 used, for example, to append data from an API function to a buffer
207 without extra copying:
209 (void)SvPVbyte_force(sv, len);
210 s = SvGROW(sv, len + needlen + 1);
211 /* something that modifies up to needlen bytes at s+len, but
212 modifies newlen bytes
213 eg. newlen = read(fd, s + len, needlen);
214 ignoring errors for these examples
216 s[len + newlen] = '\0';
217 SvCUR_set(sv, len + newlen);
221 If you already have the data in memory or if you want to keep your
222 code simple, you can use one of the sv_cat*() variants, such as
223 sv_catpvn(). If you want to insert anywhere in the string you can use
224 sv_insert() or sv_insert_flags().
226 If you don't need the existing content of the SV, you can avoid some
230 s = SvGROW(sv, needlen + 1);
231 /* something that modifies up to needlen bytes at s, but modifies
233 eg. newlen = read(fd, s, needlen);
236 SvCUR_set(sv, newlen);
237 SvPOK_only(sv); /* also clears SVf_UTF8 */
240 Again, if you already have the data in memory or want to avoid the
241 complexity of the above, you can use sv_setpvn().
243 If you have a buffer allocated with Newx() and want to set that as the
244 SV's value, you can use sv_usepvn_flags(). That has some requirements
245 if you want to avoid perl re-allocating the buffer to fit the trailing
248 Newx(buf, somesize+1, char);
249 /* ... fill in buf ... */
250 buf[somesize] = '\0';
251 sv_usepvn_flags(sv, buf, somesize, SV_SMAGIC | SV_HAS_TRAILING_NUL);
252 /* buf now belongs to perl, don't release it */
254 If you have an SV and want to know what kind of data Perl thinks is stored
255 in it, you can use the following macros to check the type of SV you have.
261 You can get and set the current length of the string stored in an SV with
262 the following macros:
265 SvCUR_set(SV*, I32 val)
267 You can also get a pointer to the end of the string stored in the SV
272 But note that these last three macros are valid only if C<SvPOK()> is true.
274 If you want to append something to the end of string stored in an C<SV*>,
275 you can use the following functions:
277 void sv_catpv(SV*, const char*);
278 void sv_catpvn(SV*, const char*, STRLEN);
279 void sv_catpvf(SV*, const char*, ...);
280 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
282 void sv_catsv(SV*, SV*);
284 The first function calculates the length of the string to be appended by
285 using C<strlen>. In the second, you specify the length of the string
286 yourself. The third function processes its arguments like C<sprintf> and
287 appends the formatted output. The fourth function works like C<vsprintf>.
288 You can specify the address and length of an array of SVs instead of the
289 va_list argument. The fifth function
290 extends the string stored in the first
291 SV with the string stored in the second SV. It also forces the second SV
292 to be interpreted as a string.
294 The C<sv_cat*()> functions are not generic enough to operate on values that
295 have "magic". See L</Magic Virtual Tables> later in this document.
297 If you know the name of a scalar variable, you can get a pointer to its SV
298 by using the following:
300 SV* get_sv("package::varname", 0);
302 This returns NULL if the variable does not exist.
304 If you want to know if this variable (or any other SV) is actually C<defined>,
309 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
311 Its address can be used whenever an C<SV*> is needed. Make sure that
312 you don't try to compare a random sv with C<&PL_sv_undef>. For example
313 when interfacing Perl code, it'll work correctly for:
317 But won't work when called as:
322 So to repeat always use SvOK() to check whether an sv is defined.
324 Also you have to be careful when using C<&PL_sv_undef> as a value in
325 AVs or HVs (see L</AVs, HVs and undefined values>).
327 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
328 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
329 addresses can be used whenever an C<SV*> is needed.
331 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
335 if (I-am-to-return-a-real-value) {
336 sv = sv_2mortal(newSViv(42));
340 This code tries to return a new SV (which contains the value 42) if it should
341 return a real value, or undef otherwise. Instead it has returned a NULL
342 pointer which, somewhere down the line, will cause a segmentation violation,
343 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
344 first line and all will be well.
346 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
347 call is not necessary (see L</Reference Counts and Mortality>).
351 Perl provides the function C<sv_chop> to efficiently remove characters
352 from the beginning of a string; you give it an SV and a pointer to
353 somewhere inside the PV, and it discards everything before the
354 pointer. The efficiency comes by means of a little hack: instead of
355 actually removing the characters, C<sv_chop> sets the flag C<OOK>
356 (offset OK) to signal to other functions that the offset hack is in
357 effect, and it moves the PV pointer (called C<SvPVX>) forward
358 by the number of bytes chopped off, and adjusts C<SvCUR> and C<SvLEN>
359 accordingly. (A portion of the space between the old and new PV
360 pointers is used to store the count of chopped bytes.)
362 Hence, at this point, the start of the buffer that we allocated lives
363 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
364 into the middle of this allocated storage.
366 This is best demonstrated by example. Normally copy-on-write will prevent
367 the substitution from operator from using this hack, but if you can craft a
368 string for which copy-on-write is not possible, you can see it in play. In
369 the current implementation, the final byte of a string buffer is used as a
370 copy-on-write reference count. If the buffer is not big enough, then
371 copy-on-write is skipped. First have a look at an empty string:
373 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a .= ""; Dump $a'
374 SV = PV(0x7ffb7c008a70) at 0x7ffb7c030390
377 PV = 0x7ffb7bc05b50 ""\0
381 Notice here the LEN is 10. (It may differ on your platform.) Extend the
382 length of the string to one less than 10, and do a substitution:
384 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a.="123456789"; $a=~s/.//; \
386 SV = PV(0x7ffa04008a70) at 0x7ffa04030390
388 FLAGS = (POK,OOK,pPOK)
390 PV = 0x7ffa03c05b61 ( "\1" . ) "23456789"\0
394 Here the number of bytes chopped off (1) is shown next as the OFFSET. The
395 portion of the string between the "real" and the "fake" beginnings is
396 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
397 the fake beginning, not the real one. (The first character of the string
398 buffer happens to have changed to "\1" here, not "1", because the current
399 implementation stores the offset count in the string buffer. This is
402 Something similar to the offset hack is performed on AVs to enable
403 efficient shifting and splicing off the beginning of the array; while
404 C<AvARRAY> points to the first element in the array that is visible from
405 Perl, C<AvALLOC> points to the real start of the C array. These are
406 usually the same, but a C<shift> operation can be carried out by
407 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
408 Again, the location of the real start of the C array only comes into
409 play when freeing the array. See C<av_shift> in F<av.c>.
411 =head2 What's Really Stored in an SV?
413 Recall that the usual method of determining the type of scalar you have is
414 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
415 usually these macros will always return TRUE and calling the C<Sv*V>
416 macros will do the appropriate conversion of string to integer/double or
417 integer/double to string.
419 If you I<really> need to know if you have an integer, double, or string
420 pointer in an SV, you can use the following three macros instead:
426 These will tell you if you truly have an integer, double, or string pointer
427 stored in your SV. The "p" stands for private.
429 There are various ways in which the private and public flags may differ.
430 For example, in perl 5.16 and earlier a tied SV may have a valid
431 underlying value in the IV slot (so SvIOKp is true), but the data
432 should be accessed via the FETCH routine rather than directly,
433 so SvIOK is false. (In perl 5.18 onwards, tied scalars use
434 the flags the same way as untied scalars.) Another is when
435 numeric conversion has occurred and precision has been lost: only the
436 private flag is set on 'lossy' values. So when an NV is converted to an
437 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
439 In general, though, it's best to use the C<Sv*V> macros.
441 =head2 Working with AVs
443 There are two ways to create and load an AV. The first method creates an
448 The second method both creates the AV and initially populates it with SVs:
450 AV* av_make(SSize_t num, SV **ptr);
452 The second argument points to an array containing C<num> C<SV*>'s. Once the
453 AV has been created, the SVs can be destroyed, if so desired.
455 Once the AV has been created, the following operations are possible on it:
457 void av_push(AV*, SV*);
460 void av_unshift(AV*, SSize_t num);
462 These should be familiar operations, with the exception of C<av_unshift>.
463 This routine adds C<num> elements at the front of the array with the C<undef>
464 value. You must then use C<av_store> (described below) to assign values
465 to these new elements.
467 Here are some other functions:
469 SSize_t av_top_index(AV*);
470 SV** av_fetch(AV*, SSize_t key, I32 lval);
471 SV** av_store(AV*, SSize_t key, SV* val);
473 The C<av_top_index> function returns the highest index value in an array (just
474 like $#array in Perl). If the array is empty, -1 is returned. The
475 C<av_fetch> function returns the value at index C<key>, but if C<lval>
476 is non-zero, then C<av_fetch> will store an undef value at that index.
477 The C<av_store> function stores the value C<val> at index C<key>, and does
478 not increment the reference count of C<val>. Thus the caller is responsible
479 for taking care of that, and if C<av_store> returns NULL, the caller will
480 have to decrement the reference count to avoid a memory leak. Note that
481 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
488 void av_extend(AV*, SSize_t key);
490 The C<av_clear> function deletes all the elements in the AV* array, but
491 does not actually delete the array itself. The C<av_undef> function will
492 delete all the elements in the array plus the array itself. The
493 C<av_extend> function extends the array so that it contains at least C<key+1>
494 elements. If C<key+1> is less than the currently allocated length of the array,
495 then nothing is done.
497 If you know the name of an array variable, you can get a pointer to its AV
498 by using the following:
500 AV* get_av("package::varname", 0);
502 This returns NULL if the variable does not exist.
504 See L</Understanding the Magic of Tied Hashes and Arrays> for more
505 information on how to use the array access functions on tied arrays.
507 =head2 Working with HVs
509 To create an HV, you use the following routine:
513 Once the HV has been created, the following operations are possible on it:
515 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
516 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
518 The C<klen> parameter is the length of the key being passed in (Note that
519 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
520 length of the key). The C<val> argument contains the SV pointer to the
521 scalar being stored, and C<hash> is the precomputed hash value (zero if
522 you want C<hv_store> to calculate it for you). The C<lval> parameter
523 indicates whether this fetch is actually a part of a store operation, in
524 which case a new undefined value will be added to the HV with the supplied
525 key and C<hv_fetch> will return as if the value had already existed.
527 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
528 C<SV*>. To access the scalar value, you must first dereference the return
529 value. However, you should check to make sure that the return value is
530 not NULL before dereferencing it.
532 The first of these two functions checks if a hash table entry exists, and the
535 bool hv_exists(HV*, const char* key, U32 klen);
536 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
538 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
539 create and return a mortal copy of the deleted value.
541 And more miscellaneous functions:
546 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
547 table but does not actually delete the hash table. The C<hv_undef> deletes
548 both the entries and the hash table itself.
550 Perl keeps the actual data in a linked list of structures with a typedef of HE.
551 These contain the actual key and value pointers (plus extra administrative
552 overhead). The key is a string pointer; the value is an C<SV*>. However,
553 once you have an C<HE*>, to get the actual key and value, use the routines
558 I32 hv_iterinit(HV*);
559 /* Prepares starting point to traverse hash table */
560 HE* hv_iternext(HV*);
561 /* Get the next entry, and return a pointer to a
562 structure that has both the key and value */
563 char* hv_iterkey(HE* entry, I32* retlen);
564 /* Get the key from an HE structure and also return
565 the length of the key string */
566 SV* hv_iterval(HV*, HE* entry);
567 /* Return an SV pointer to the value of the HE
569 SV* hv_iternextsv(HV*, char** key, I32* retlen);
570 /* This convenience routine combines hv_iternext,
571 hv_iterkey, and hv_iterval. The key and retlen
572 arguments are return values for the key and its
573 length. The value is returned in the SV* argument */
575 If you know the name of a hash variable, you can get a pointer to its HV
576 by using the following:
578 HV* get_hv("package::varname", 0);
580 This returns NULL if the variable does not exist.
582 The hash algorithm is defined in the C<PERL_HASH> macro:
584 PERL_HASH(hash, key, klen)
586 The exact implementation of this macro varies by architecture and version
587 of perl, and the return value may change per invocation, so the value
588 is only valid for the duration of a single perl process.
590 See L</Understanding the Magic of Tied Hashes and Arrays> for more
591 information on how to use the hash access functions on tied hashes.
593 =for apidoc_section $HV
594 =for apidoc Amh|void|PERL_HASH|U32 hash|char *key|STRLEN klen
596 =head2 Hash API Extensions
598 Beginning with version 5.004, the following functions are also supported:
600 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
601 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
603 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
604 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
606 SV* hv_iterkeysv (HE* entry);
608 Note that these functions take C<SV*> keys, which simplifies writing
609 of extension code that deals with hash structures. These functions
610 also allow passing of C<SV*> keys to C<tie> functions without forcing
611 you to stringify the keys (unlike the previous set of functions).
613 They also return and accept whole hash entries (C<HE*>), making their
614 use more efficient (since the hash number for a particular string
615 doesn't have to be recomputed every time). See L<perlapi> for detailed
618 The following macros must always be used to access the contents of hash
619 entries. Note that the arguments to these macros must be simple
620 variables, since they may get evaluated more than once. See
621 L<perlapi> for detailed descriptions of these macros.
623 HePV(HE* he, STRLEN len)
627 HeSVKEY_force(HE* he)
628 HeSVKEY_set(HE* he, SV* sv)
630 These two lower level macros are defined, but must only be used when
631 dealing with keys that are not C<SV*>s:
636 Note that both C<hv_store> and C<hv_store_ent> do not increment the
637 reference count of the stored C<val>, which is the caller's responsibility.
638 If these functions return a NULL value, the caller will usually have to
639 decrement the reference count of C<val> to avoid a memory leak.
641 =head2 AVs, HVs and undefined values
643 Sometimes you have to store undefined values in AVs or HVs. Although
644 this may be a rare case, it can be tricky. That's because you're
645 used to using C<&PL_sv_undef> if you need an undefined SV.
647 For example, intuition tells you that this XS code:
650 av_store( av, 0, &PL_sv_undef );
652 is equivalent to this Perl code:
657 Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
658 for indicating that an array element has not yet been initialized.
659 Thus, C<exists $av[0]> would be true for the above Perl code, but
660 false for the array generated by the XS code. In perl 5.20, storing
661 &PL_sv_undef will create a read-only element, because the scalar
662 &PL_sv_undef itself is stored, not a copy.
664 Similar problems can occur when storing C<&PL_sv_undef> in HVs:
666 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
668 This will indeed make the value C<undef>, but if you try to modify
669 the value of C<key>, you'll get the following error:
671 Modification of non-creatable hash value attempted
673 In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
674 in restricted hashes. This caused such hash entries not to appear
675 when iterating over the hash or when checking for the keys
676 with the C<hv_exists> function.
678 You can run into similar problems when you store C<&PL_sv_yes> or
679 C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
680 will give you the following error:
682 Modification of a read-only value attempted
684 To make a long story short, you can use the special variables
685 C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
686 HVs, but you have to make sure you know what you're doing.
688 Generally, if you want to store an undefined value in an AV
689 or HV, you should not use C<&PL_sv_undef>, but rather create a
690 new undefined value using the C<newSV> function, for example:
692 av_store( av, 42, newSV(0) );
693 hv_store( hv, "foo", 3, newSV(0), 0 );
697 References are a special type of scalar that point to other data types
698 (including other references).
700 To create a reference, use either of the following functions:
702 SV* newRV_inc((SV*) thing);
703 SV* newRV_noinc((SV*) thing);
705 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
706 functions are identical except that C<newRV_inc> increments the reference
707 count of the C<thing>, while C<newRV_noinc> does not. For historical
708 reasons, C<newRV> is a synonym for C<newRV_inc>.
710 Once you have a reference, you can use the following macro to dereference
715 then call the appropriate routines, casting the returned C<SV*> to either an
716 C<AV*> or C<HV*>, if required.
718 To determine if an SV is a reference, you can use the following macro:
722 To discover what type of value the reference refers to, use the following
723 macro and then check the return value.
727 The most useful types that will be returned are:
732 SVt_PVGV Glob (possibly a file handle)
734 Any numerical value returned which is less than SVt_PVAV will be a scalar
737 See L<perlapi/svtype> for more details.
739 =head2 Blessed References and Class Objects
741 References are also used to support object-oriented programming. In perl's
742 OO lexicon, an object is simply a reference that has been blessed into a
743 package (or class). Once blessed, the programmer may now use the reference
744 to access the various methods in the class.
746 A reference can be blessed into a package with the following function:
748 SV* sv_bless(SV* sv, HV* stash);
750 The C<sv> argument must be a reference value. The C<stash> argument
751 specifies which class the reference will belong to. See
752 L</Stashes and Globs> for information on converting class names into stashes.
754 /* Still under construction */
756 The following function upgrades rv to reference if not already one.
757 Creates a new SV for rv to point to. If C<classname> is non-null, the SV
758 is blessed into the specified class. SV is returned.
760 SV* newSVrv(SV* rv, const char* classname);
762 The following three functions copy integer, unsigned integer or double
763 into an SV whose reference is C<rv>. SV is blessed if C<classname> is
766 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
767 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
768 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
770 The following function copies the pointer value (I<the address, not the
771 string!>) into an SV whose reference is rv. SV is blessed if C<classname>
774 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
776 The following function copies a string into an SV whose reference is C<rv>.
777 Set length to 0 to let Perl calculate the string length. SV is blessed if
778 C<classname> is non-null.
780 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
783 The following function tests whether the SV is blessed into the specified
784 class. It does not check inheritance relationships.
786 int sv_isa(SV* sv, const char* name);
788 The following function tests whether the SV is a reference to a blessed object.
790 int sv_isobject(SV* sv);
792 The following function tests whether the SV is derived from the specified
793 class. SV can be either a reference to a blessed object or a string
794 containing a class name. This is the function implementing the
795 C<UNIVERSAL::isa> functionality.
797 bool sv_derived_from(SV* sv, const char* name);
799 To check if you've got an object derived from a specific class you have
802 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
804 =head2 Creating New Variables
806 To create a new Perl variable with an undef value which can be accessed from
807 your Perl script, use the following routines, depending on the variable type.
809 SV* get_sv("package::varname", GV_ADD);
810 AV* get_av("package::varname", GV_ADD);
811 HV* get_hv("package::varname", GV_ADD);
813 Notice the use of GV_ADD as the second parameter. The new variable can now
814 be set, using the routines appropriate to the data type.
816 There are additional macros whose values may be bitwise OR'ed with the
817 C<GV_ADD> argument to enable certain extra features. Those bits are:
823 Marks the variable as multiply defined, thus preventing the:
825 Name <varname> used only once: possible typo
833 Had to create <varname> unexpectedly
835 if the variable did not exist before the function was called.
839 If you do not specify a package name, the variable is created in the current
842 =head2 Reference Counts and Mortality
844 Perl uses a reference count-driven garbage collection mechanism. SVs,
845 AVs, or HVs (xV for short in the following) start their life with a
846 reference count of 1. If the reference count of an xV ever drops to 0,
847 then it will be destroyed and its memory made available for reuse.
848 At the most basic internal level, reference counts can be manipulated
849 with the following macros:
851 int SvREFCNT(SV* sv);
852 SV* SvREFCNT_inc(SV* sv);
853 void SvREFCNT_dec(SV* sv);
855 (There are also suffixed versions of the increment and decrement macros,
856 for situations where the full generality of these basic macros can be
857 exchanged for some performance.)
859 However, the way a programmer should think about references is not so
860 much in terms of the bare reference count, but in terms of I<ownership>
861 of references. A reference to an xV can be owned by any of a variety
862 of entities: another xV, the Perl interpreter, an XS data structure,
863 a piece of running code, or a dynamic scope. An xV generally does not
864 know what entities own the references to it; it only knows how many
865 references there are, which is the reference count.
867 To correctly maintain reference counts, it is essential to keep track
868 of what references the XS code is manipulating. The programmer should
869 always know where a reference has come from and who owns it, and be
870 aware of any creation or destruction of references, and any transfers
871 of ownership. Because ownership isn't represented explicitly in the xV
872 data structures, only the reference count need be actually maintained
873 by the code, and that means that this understanding of ownership is not
874 actually evident in the code. For example, transferring ownership of a
875 reference from one owner to another doesn't change the reference count
876 at all, so may be achieved with no actual code. (The transferring code
877 doesn't touch the referenced object, but does need to ensure that the
878 former owner knows that it no longer owns the reference, and that the
879 new owner knows that it now does.)
881 An xV that is visible at the Perl level should not become unreferenced
882 and thus be destroyed. Normally, an object will only become unreferenced
883 when it is no longer visible, often by the same means that makes it
884 invisible. For example, a Perl reference value (RV) owns a reference to
885 its referent, so if the RV is overwritten that reference gets destroyed,
886 and the no-longer-reachable referent may be destroyed as a result.
888 Many functions have some kind of reference manipulation as
889 part of their purpose. Sometimes this is documented in terms
890 of ownership of references, and sometimes it is (less helpfully)
891 documented in terms of changes to reference counts. For example, the
892 L<newRV_inc()|perlapi/newRV_inc> function is documented to create a new RV
893 (with reference count 1) and increment the reference count of the referent
894 that was supplied by the caller. This is best understood as creating
895 a new reference to the referent, which is owned by the created RV,
896 and returning to the caller ownership of the sole reference to the RV.
897 The L<newRV_noinc()|perlapi/newRV_noinc> function instead does not
898 increment the reference count of the referent, but the RV nevertheless
899 ends up owning a reference to the referent. It is therefore implied
900 that the caller of C<newRV_noinc()> is relinquishing a reference to the
901 referent, making this conceptually a more complicated operation even
902 though it does less to the data structures.
904 For example, imagine you want to return a reference from an XSUB
905 function. Inside the XSUB routine, you create an SV which initially
906 has just a single reference, owned by the XSUB routine. This reference
907 needs to be disposed of before the routine is complete, otherwise it
908 will leak, preventing the SV from ever being destroyed. So to create
909 an RV referencing the SV, it is most convenient to pass the SV to
910 C<newRV_noinc()>, which consumes that reference. Now the XSUB routine
911 no longer owns a reference to the SV, but does own a reference to the RV,
912 which in turn owns a reference to the SV. The ownership of the reference
913 to the RV is then transferred by the process of returning the RV from
916 There are some convenience functions available that can help with the
917 destruction of xVs. These functions introduce the concept of "mortality".
918 Much documentation speaks of an xV itself being mortal, but this is
919 misleading. It is really I<a reference to> an xV that is mortal, and it
920 is possible for there to be more than one mortal reference to a single xV.
921 For a reference to be mortal means that it is owned by the temps stack,
922 one of perl's many internal stacks, which will destroy that reference
923 "a short time later". Usually the "short time later" is the end of
924 the current Perl statement. However, it gets more complicated around
925 dynamic scopes: there can be multiple sets of mortal references hanging
926 around at the same time, with different death dates. Internally, the
927 actual determinant for when mortal xV references are destroyed depends
928 on two macros, SAVETMPS and FREETMPS. See L<perlcall> and L<perlxs>
929 and L</Temporaries Stack> below for more details on these macros.
931 Mortal references are mainly used for xVs that are placed on perl's
932 main stack. The stack is problematic for reference tracking, because it
933 contains a lot of xV references, but doesn't own those references: they
934 are not counted. Currently, there are many bugs resulting from xVs being
935 destroyed while referenced by the stack, because the stack's uncounted
936 references aren't enough to keep the xVs alive. So when putting an
937 (uncounted) reference on the stack, it is vitally important to ensure that
938 there will be a counted reference to the same xV that will last at least
939 as long as the uncounted reference. But it's also important that that
940 counted reference be cleaned up at an appropriate time, and not unduly
941 prolong the xV's life. For there to be a mortal reference is often the
942 best way to satisfy this requirement, especially if the xV was created
943 especially to be put on the stack and would otherwise be unreferenced.
945 To create a mortal reference, use the functions:
948 SV* sv_mortalcopy(SV*)
951 C<sv_newmortal()> creates an SV (with the undefined value) whose sole
952 reference is mortal. C<sv_mortalcopy()> creates an xV whose value is a
953 copy of a supplied xV and whose sole reference is mortal. C<sv_2mortal()>
954 mortalises an existing xV reference: it transfers ownership of a reference
955 from the caller to the temps stack. Because C<sv_newmortal> gives the new
956 SV no value, it must normally be given one via C<sv_setpv>, C<sv_setiv>,
959 SV *tmp = sv_newmortal();
960 sv_setiv(tmp, an_integer);
962 As that is multiple C statements it is quite common so see this idiom instead:
964 SV *tmp = sv_2mortal(newSViv(an_integer));
966 The mortal routines are not just for SVs; AVs and HVs can be
967 made mortal by passing their address (type-casted to C<SV*>) to the
968 C<sv_2mortal> or C<sv_mortalcopy> routines.
970 =head2 Stashes and Globs
972 A B<stash> is a hash that contains all variables that are defined
973 within a package. Each key of the stash is a symbol
974 name (shared by all the different types of objects that have the same
975 name), and each value in the hash table is a GV (Glob Value). This GV
976 in turn contains references to the various objects of that name,
977 including (but not limited to) the following:
986 There is a single stash called C<PL_defstash> that holds the items that exist
987 in the C<main> package. To get at the items in other packages, append the
988 string "::" to the package name. The items in the C<Foo> package are in
989 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
990 in the stash C<Baz::> in C<Bar::>'s stash.
992 To get the stash pointer for a particular package, use the function:
994 HV* gv_stashpv(const char* name, I32 flags)
995 HV* gv_stashsv(SV*, I32 flags)
997 The first function takes a literal string, the second uses the string stored
998 in the SV. Remember that a stash is just a hash table, so you get back an
999 C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
1001 The name that C<gv_stash*v> wants is the name of the package whose symbol table
1002 you want. The default package is called C<main>. If you have multiply nested
1003 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
1006 Alternately, if you have an SV that is a blessed reference, you can find
1007 out the stash pointer by using:
1009 HV* SvSTASH(SvRV(SV*));
1011 then use the following to get the package name itself:
1013 char* HvNAME(HV* stash);
1015 If you need to bless or re-bless an object you can use the following
1018 SV* sv_bless(SV*, HV* stash)
1020 where the first argument, an C<SV*>, must be a reference, and the second
1021 argument is a stash. The returned C<SV*> can now be used in the same way
1024 For more information on references and blessings, consult L<perlref>.
1028 Like AVs and HVs, IO objects are another type of non-scalar SV which
1029 may contain input and output L<PerlIO|perlapio> objects or a C<DIR *>
1032 You can create a new IO object:
1036 Unlike other SVs, a new IO object is automatically blessed into the
1039 The IO object contains an input and output PerlIO handle:
1041 PerlIO *IoIFP(IO *io);
1042 PerlIO *IoOFP(IO *io);
1044 Typically if the IO object has been opened on a file, the input handle
1045 is always present, but the output handle is only present if the file
1046 is open for output. For a file, if both are present they will be the
1049 Distinct input and output PerlIO objects are created for sockets and
1052 The IO object also contains other data associated with Perl I/O
1055 IV IoLINES(io); /* $. */
1056 IV IoPAGE(io); /* $% */
1057 IV IoPAGE_LEN(io); /* $= */
1058 IV IoLINES_LEFT(io); /* $- */
1059 char *IoTOP_NAME(io); /* $^ */
1060 GV *IoTOP_GV(io); /* $^ */
1061 char *IoFMT_NAME(io); /* $~ */
1062 GV *IoFMT_GV(io); /* $~ */
1063 char *IoBOTTOM_NAME(io);
1064 GV *IoBOTTOM_GV(io);
1068 Most of these are involved with L<formats|perlform>.
1070 IoFLAGs() may contain a combination of flags, the most interesting of
1071 which are C<IOf_FLUSH> (C<$|>) for autoflush and C<IOf_UNTAINT>,
1072 settable with L<< IO::Handle's untaint() method|IO::Handle/"$io->untaint" >>.
1074 The IO object may also contains a directory handle:
1078 suitable for use with PerlDir_read() etc.
1080 All of these accessors macros are lvalues, there are no distinct
1081 C<_set()> macros to modify the members of the IO object.
1083 =head2 Double-Typed SVs
1085 Scalar variables normally contain only one type of value, an integer,
1086 double, pointer, or reference. Perl will automatically convert the
1087 actual scalar data from the stored type into the requested type.
1089 Some scalar variables contain more than one type of scalar data. For
1090 example, the variable C<$!> contains either the numeric value of C<errno>
1091 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
1093 To force multiple data values into an SV, you must do two things: use the
1094 C<sv_set*v> routines to add the additional scalar type, then set a flag
1095 so that Perl will believe it contains more than one type of data. The
1096 four macros to set the flags are:
1103 The particular macro you must use depends on which C<sv_set*v> routine
1104 you called first. This is because every C<sv_set*v> routine turns on
1105 only the bit for the particular type of data being set, and turns off
1108 For example, to create a new Perl variable called "dberror" that contains
1109 both the numeric and descriptive string error values, you could use the
1113 extern char *dberror_list;
1115 SV* sv = get_sv("dberror", GV_ADD);
1116 sv_setiv(sv, (IV) dberror);
1117 sv_setpv(sv, dberror_list[dberror]);
1120 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
1121 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
1123 =head2 Read-Only Values
1125 In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
1126 flag bit with read-only scalars. So the only way to test whether
1127 C<sv_setsv>, etc., will raise a "Modification of a read-only value" error
1128 in those versions is:
1130 SvREADONLY(sv) && !SvIsCOW(sv)
1132 Under Perl 5.18 and later, SvREADONLY only applies to read-only variables,
1133 and, under 5.20, copy-on-write scalars can also be read-only, so the above
1134 check is incorrect. You just want:
1138 If you need to do this check often, define your own macro like this:
1140 #if PERL_VERSION >= 18
1141 # define SvTRULYREADONLY(sv) SvREADONLY(sv)
1143 # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
1146 =head2 Copy on Write
1148 Perl implements a copy-on-write (COW) mechanism for scalars, in which
1149 string copies are not immediately made when requested, but are deferred
1150 until made necessary by one or the other scalar changing. This is mostly
1151 transparent, but one must take care not to modify string buffers that are
1152 shared by multiple SVs.
1154 You can test whether an SV is using copy-on-write with C<SvIsCOW(sv)>.
1156 You can force an SV to make its own copy of its string buffer by calling C<sv_force_normal(sv)> or SvPV_force_nolen(sv).
1158 If you want to make the SV drop its string buffer, use
1159 C<sv_force_normal_flags(sv, SV_COW_DROP_PV)> or simply
1160 C<sv_setsv(sv, NULL)>.
1162 All of these functions will croak on read-only scalars (see the previous
1163 section for more on those).
1165 To test that your code is behaving correctly and not modifying COW buffers,
1166 on systems that support L<mmap(2)> (i.e., Unix) you can configure perl with
1167 C<-Accflags=-DPERL_DEBUG_READONLY_COW> and it will turn buffer violations
1168 into crashes. You will find it to be marvellously slow, so you may want to
1169 skip perl's own tests.
1171 =head2 Magic Variables
1173 [This section still under construction. Ignore everything here. Post no
1174 bills. Everything not permitted is forbidden.]
1176 Any SV may be magical, that is, it has special features that a normal
1177 SV does not have. These features are stored in the SV structure in a
1178 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
1181 MAGIC* mg_moremagic;
1191 Note this is current as of patchlevel 0, and could change at any time.
1193 =head2 Assigning Magic
1195 Perl adds magic to an SV using the sv_magic function:
1197 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
1199 The C<sv> argument is a pointer to the SV that is to acquire a new magical
1202 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
1203 convert C<sv> to type C<SVt_PVMG>.
1204 Perl then continues by adding new magic
1205 to the beginning of the linked list of magical features. Any prior entry
1206 of the same type of magic is deleted. Note that this can be overridden,
1207 and multiple instances of the same type of magic can be associated with an
1210 The C<name> and C<namlen> arguments are used to associate a string with
1211 the magic, typically the name of a variable. C<namlen> is stored in the
1212 C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
1213 C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
1214 whether C<namlen> is greater than zero or equal to zero respectively. As a
1215 special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
1216 to contain an C<SV*> and is stored as-is with its REFCNT incremented.
1218 The sv_magic function uses C<how> to determine which, if any, predefined
1219 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
1220 See the L</Magic Virtual Tables> section below. The C<how> argument is also
1221 stored in the C<mg_type> field. The value of
1222 C<how> should be chosen from the set of macros
1223 C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
1224 these macros were added, Perl internals used to directly use character
1225 literals, so you may occasionally come across old code or documentation
1226 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
1228 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
1229 structure. If it is not the same as the C<sv> argument, the reference
1230 count of the C<obj> object is incremented. If it is the same, or if
1231 the C<how> argument is C<PERL_MAGIC_arylen>, C<PERL_MAGIC_regdatum>,
1232 C<PERL_MAGIC_regdata>, or if it is a NULL pointer, then C<obj> is merely
1233 stored, without the reference count being incremented.
1235 See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
1238 There is also a function to add magic to an C<HV>:
1240 void hv_magic(HV *hv, GV *gv, int how);
1242 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
1244 To remove the magic from an SV, call the function sv_unmagic:
1246 int sv_unmagic(SV *sv, int type);
1248 The C<type> argument should be equal to the C<how> value when the C<SV>
1249 was initially made magical.
1251 However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
1252 C<SV>. If you want to remove only certain
1253 magic of a C<type> based on the magic
1254 virtual table, use C<sv_unmagicext> instead:
1256 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1258 =head2 Magic Virtual Tables
1260 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
1261 C<MGVTBL>, which is a structure of function pointers and stands for
1262 "Magic Virtual Table" to handle the various operations that might be
1263 applied to that variable.
1265 =for apidoc Ayh||MGVTBL
1267 The C<MGVTBL> has five (or sometimes eight) pointers to the following
1270 int (*svt_get) (pTHX_ SV* sv, MAGIC* mg);
1271 int (*svt_set) (pTHX_ SV* sv, MAGIC* mg);
1272 U32 (*svt_len) (pTHX_ SV* sv, MAGIC* mg);
1273 int (*svt_clear)(pTHX_ SV* sv, MAGIC* mg);
1274 int (*svt_free) (pTHX_ SV* sv, MAGIC* mg);
1276 int (*svt_copy) (pTHX_ SV *sv, MAGIC* mg, SV *nsv,
1277 const char *name, I32 namlen);
1278 int (*svt_dup) (pTHX_ MAGIC *mg, CLONE_PARAMS *param);
1279 int (*svt_local)(pTHX_ SV *nsv, MAGIC *mg);
1282 This MGVTBL structure is set at compile-time in F<perl.h> and there are
1283 currently 32 types. These different structures contain pointers to various
1284 routines that perform additional actions depending on which function is
1287 Function pointer Action taken
1288 ---------------- ------------
1289 svt_get Do something before the value of the SV is
1291 svt_set Do something after the SV is assigned a value.
1292 svt_len Report on the SV's length.
1293 svt_clear Clear something the SV represents.
1294 svt_free Free any extra storage associated with the SV.
1296 svt_copy copy tied variable magic to a tied element
1297 svt_dup duplicate a magic structure during thread cloning
1298 svt_local copy magic to local value during 'local'
1300 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
1301 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
1303 { magic_get, magic_set, magic_len, 0, 0 }
1305 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1306 if a get operation is being performed, the routine C<magic_get> is
1307 called. All the various routines for the various magical types begin
1308 with C<magic_>. NOTE: the magic routines are not considered part of
1309 the Perl API, and may not be exported by the Perl library.
1311 The last three slots are a recent addition, and for source code
1312 compatibility they are only checked for if one of the three flags
1313 MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1314 This means that most code can continue declaring
1315 a vtable as a 5-element value. These three are
1316 currently used exclusively by the threading code, and are highly subject
1319 The current kinds of Magic Virtual Tables are:
1322 This table is generated by regen/mg_vtable.pl. Any changes made here
1325 =for mg_vtable.pl begin
1328 (old-style char and macro) MGVTBL Type of magic
1329 -------------------------- ------ -------------
1330 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1331 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1332 % PERL_MAGIC_rhash (none) Extra data for restricted
1334 * PERL_MAGIC_debugvar vtbl_debugvar $DB::single, signal, trace
1336 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1337 : PERL_MAGIC_symtab (none) Extra data for symbol
1339 < PERL_MAGIC_backref vtbl_backref For weak ref data
1340 @ PERL_MAGIC_arylen_p (none) To move arylen out of XPVAV
1341 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1342 (fast string search)
1343 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1345 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1347 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1349 E PERL_MAGIC_env vtbl_env %ENV hash
1350 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1351 f PERL_MAGIC_fm vtbl_regexp Formline
1353 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1354 H PERL_MAGIC_hints vtbl_hints %^H hash
1355 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1356 I PERL_MAGIC_isa vtbl_isa @ISA array
1357 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1358 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1359 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1360 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1362 N PERL_MAGIC_shared (none) Shared between threads
1363 n PERL_MAGIC_shared_scalar (none) Shared between threads
1364 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1365 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1366 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1367 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1368 r PERL_MAGIC_qr vtbl_regexp Precompiled qr// regex
1369 S PERL_MAGIC_sig (none) %SIG hash
1370 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1371 t PERL_MAGIC_taint vtbl_taint Taintedness
1372 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1374 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1376 V PERL_MAGIC_vstring (none) SV was vstring literal
1377 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1378 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1379 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1380 Y PERL_MAGIC_nonelem vtbl_nonelem Array element that does not
1382 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1383 variable / smart parameter
1385 \ PERL_MAGIC_lvref vtbl_lvref Lvalue reference
1387 ] PERL_MAGIC_checkcall vtbl_checkcall Inlining/mutation of call
1389 ~ PERL_MAGIC_ext (none) Available for use by
1393 =for apidoc AmnhU||PERL_MAGIC_arylen
1394 =for apidoc_item ||PERL_MAGIC_arylen_p
1395 =for apidoc_item ||PERL_MAGIC_backref
1396 =for apidoc_item ||PERL_MAGIC_bm
1397 =for apidoc_item ||PERL_MAGIC_checkcall
1398 =for apidoc_item ||PERL_MAGIC_collxfrm
1399 =for apidoc_item ||PERL_MAGIC_dbfile
1400 =for apidoc_item ||PERL_MAGIC_dbline
1401 =for apidoc_item ||PERL_MAGIC_debugvar
1402 =for apidoc_item ||PERL_MAGIC_defelem
1403 =for apidoc_item ||PERL_MAGIC_env
1404 =for apidoc_item ||PERL_MAGIC_envelem
1405 =for apidoc_item ||PERL_MAGIC_ext
1406 =for apidoc_item ||PERL_MAGIC_fm
1407 =for apidoc_item ||PERL_MAGIC_hints
1408 =for apidoc_item ||PERL_MAGIC_hintselem
1409 =for apidoc_item ||PERL_MAGIC_isa
1410 =for apidoc_item ||PERL_MAGIC_isaelem
1411 =for apidoc_item ||PERL_MAGIC_lvref
1412 =for apidoc_item ||PERL_MAGIC_nkeys
1413 =for apidoc_item ||PERL_MAGIC_nonelem
1414 =for apidoc_item ||PERL_MAGIC_overload_table
1415 =for apidoc_item ||PERL_MAGIC_pos
1416 =for apidoc_item ||PERL_MAGIC_qr
1417 =for apidoc_item ||PERL_MAGIC_regdata
1418 =for apidoc_item ||PERL_MAGIC_regdatum
1419 =for apidoc_item ||PERL_MAGIC_regex_global
1420 =for apidoc_item ||PERL_MAGIC_rhash
1421 =for apidoc_item ||PERL_MAGIC_shared
1422 =for apidoc_item ||PERL_MAGIC_shared_scalar
1423 =for apidoc_item ||PERL_MAGIC_sig
1424 =for apidoc_item ||PERL_MAGIC_sigelem
1425 =for apidoc_item ||PERL_MAGIC_substr
1426 =for apidoc_item ||PERL_MAGIC_sv
1427 =for apidoc_item ||PERL_MAGIC_symtab
1428 =for apidoc_item ||PERL_MAGIC_taint
1429 =for apidoc_item ||PERL_MAGIC_tied
1430 =for apidoc_item ||PERL_MAGIC_tiedelem
1431 =for apidoc_item ||PERL_MAGIC_tiedscalar
1432 =for apidoc_item ||PERL_MAGIC_utf8
1433 =for apidoc_item ||PERL_MAGIC_uvar
1434 =for apidoc_item ||PERL_MAGIC_uvar_elem
1435 =for apidoc_item ||PERL_MAGIC_vec
1436 =for apidoc_item ||PERL_MAGIC_vstring
1438 =for mg_vtable.pl end
1440 When an uppercase and lowercase letter both exist in the table, then the
1441 uppercase letter is typically used to represent some kind of composite type
1442 (a list or a hash), and the lowercase letter is used to represent an element
1443 of that composite type. Some internals code makes use of this case
1444 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
1446 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1447 specifically for use by extensions and will not be used by perl itself.
1448 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1449 to variables (typically objects). This is especially useful because
1450 there is no way for normal perl code to corrupt this private information
1451 (unlike using extra elements of a hash object).
1453 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1454 C function any time a scalar's value is used or changed. The C<MAGIC>'s
1455 C<mg_ptr> field points to a C<ufuncs> structure:
1458 I32 (*uf_val)(pTHX_ IV, SV*);
1459 I32 (*uf_set)(pTHX_ IV, SV*);
1463 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1464 function will be called with C<uf_index> as the first arg and a pointer to
1465 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1466 magic is shown below. Note that the ufuncs structure is copied by
1467 sv_magic, so you can safely allocate it on the stack.
1475 uf.uf_val = &my_get_fn;
1476 uf.uf_set = &my_set_fn;
1478 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1480 Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1482 For hashes there is a specialized hook that gives control over hash
1483 keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1484 if the "set" function in the C<ufuncs> structure is NULL. The hook
1485 is activated whenever the hash is accessed with a key specified as
1486 an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1487 C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1488 through the functions without the C<..._ent> suffix circumvents the
1489 hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1491 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1492 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1493 extra care to avoid conflict. Typically only using the magic on
1494 objects blessed into the same class as the extension is sufficient.
1495 For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1496 C<MGVTBL>, even if all its fields will be C<0>, so that individual
1497 C<MAGIC> pointers can be identified as a particular kind of magic
1498 using their magic virtual table. C<mg_findext> provides an easy way
1501 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1504 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1505 /* this is really ours, not another module's PERL_MAGIC_ext */
1506 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1510 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1511 earlier do B<not> invoke 'set' magic on their targets. This must
1512 be done by the user either by calling the C<SvSETMAGIC()> macro after
1513 calling these functions, or by using one of the C<sv_set*_mg()> or
1514 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1515 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1516 obtained from external sources in functions that don't handle magic.
1517 See L<perlapi> for a description of these functions.
1518 For example, calls to the C<sv_cat*()> functions typically need to be
1519 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1520 since their implementation handles 'get' magic.
1522 =head2 Finding Magic
1524 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1527 This routine returns a pointer to a C<MAGIC> structure stored in the SV.
1528 If the SV does not have that magical
1529 feature, C<NULL> is returned. If the
1530 SV has multiple instances of that magical feature, the first one will be
1531 returned. C<mg_findext> can be used
1532 to find a C<MAGIC> structure of an SV
1533 based on both its magic type and its magic virtual table:
1535 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
1537 Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1538 SVt_PVMG, Perl may core dump.
1540 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1542 This routine checks to see what types of magic C<sv> has. If the mg_type
1543 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1544 the mg_type field is changed to be the lowercase letter.
1546 =head2 Understanding the Magic of Tied Hashes and Arrays
1548 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1551 WARNING: As of the 5.004 release, proper usage of the array and hash
1552 access functions requires understanding a few caveats. Some
1553 of these caveats are actually considered bugs in the API, to be fixed
1554 in later releases, and are bracketed with [MAYCHANGE] below. If
1555 you find yourself actually applying such information in this section, be
1556 aware that the behavior may change in the future, umm, without warning.
1558 The perl tie function associates a variable with an object that implements
1559 the various GET, SET, etc methods. To perform the equivalent of the perl
1560 tie function from an XSUB, you must mimic this behaviour. The code below
1561 carries out the necessary steps -- firstly it creates a new hash, and then
1562 creates a second hash which it blesses into the class which will implement
1563 the tie methods. Lastly it ties the two hashes together, and returns a
1564 reference to the new tied hash. Note that the code below does NOT call the
1565 TIEHASH method in the MyTie class -
1566 see L</Calling Perl Routines from within C Programs> for details on how
1577 tie = newRV_noinc((SV*)newHV());
1578 stash = gv_stashpv("MyTie", GV_ADD);
1579 sv_bless(tie, stash);
1580 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1581 RETVAL = newRV_noinc(hash);
1585 The C<av_store> function, when given a tied array argument, merely
1586 copies the magic of the array onto the value to be "stored", using
1587 C<mg_copy>. It may also return NULL, indicating that the value did not
1588 actually need to be stored in the array. [MAYCHANGE] After a call to
1589 C<av_store> on a tied array, the caller will usually need to call
1590 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1591 TIEARRAY object. If C<av_store> did return NULL, a call to
1592 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1595 The previous paragraph is applicable verbatim to tied hash access using the
1596 C<hv_store> and C<hv_store_ent> functions as well.
1598 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1599 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1600 has been initialized using C<mg_copy>. Note the value so returned does not
1601 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1602 need to call C<mg_get()> on the returned value in order to actually invoke
1603 the perl level "FETCH" method on the underlying TIE object. Similarly,
1604 you may also call C<mg_set()> on the return value after possibly assigning
1605 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1606 method on the TIE object. [/MAYCHANGE]
1609 In other words, the array or hash fetch/store functions don't really
1610 fetch and store actual values in the case of tied arrays and hashes. They
1611 merely call C<mg_copy> to attach magic to the values that were meant to be
1612 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1613 do the job of invoking the TIE methods on the underlying objects. Thus
1614 the magic mechanism currently implements a kind of lazy access to arrays
1617 Currently (as of perl version 5.004), use of the hash and array access
1618 functions requires the user to be aware of whether they are operating on
1619 "normal" hashes and arrays, or on their tied variants. The API may be
1620 changed to provide more transparent access to both tied and normal data
1621 types in future versions.
1624 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1625 are mere sugar to invoke some perl method calls while using the uniform hash
1626 and array syntax. The use of this sugar imposes some overhead (typically
1627 about two to four extra opcodes per FETCH/STORE operation, in addition to
1628 the creation of all the mortal variables required to invoke the methods).
1629 This overhead will be comparatively small if the TIE methods are themselves
1630 substantial, but if they are only a few statements long, the overhead
1631 will not be insignificant.
1633 =head2 Localizing changes
1635 Perl has a very handy construction
1642 This construction is I<approximately> equivalent to
1651 The biggest difference is that the first construction would
1652 reinstate the initial value of $var, irrespective of how control exits
1653 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1654 more efficient as well.
1656 There is a way to achieve a similar task from C via Perl API: create a
1657 I<pseudo-block>, and arrange for some changes to be automatically
1658 undone at the end of it, either explicit, or via a non-local exit (via
1659 die()). A I<block>-like construct is created by a pair of
1660 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1661 Such a construct may be created specially for some important localized
1662 task, or an existing one (like boundaries of enclosing Perl
1663 subroutine/block, or an existing pair for freeing TMPs) may be
1664 used. (In the second case the overhead of additional localization must
1665 be almost negligible.) Note that any XSUB is automatically enclosed in
1666 an C<ENTER>/C<LEAVE> pair.
1668 Inside such a I<pseudo-block> the following service is available:
1672 =item C<SAVEINT(int i)>
1674 =item C<SAVEIV(IV i)>
1676 =item C<SAVEI32(I32 i)>
1678 =item C<SAVELONG(long i)>
1680 =item C<SAVEI8(I8 i)>
1682 =item C<SAVEI16(I16 i)>
1684 =item C<SAVEBOOL(int i)>
1686 These macros arrange things to restore the value of integer variable
1687 C<i> at the end of the enclosing I<pseudo-block>.
1689 =for apidoc_section $stack
1690 =for apidoc Amh||SAVEINT|int i
1691 =for apidoc Amh||SAVEIV|IV i
1692 =for apidoc Amh||SAVEI32|I32 i
1693 =for apidoc Amh||SAVELONG|long i
1694 =for apidoc Amh||SAVEI8|I8 i
1695 =for apidoc Amh||SAVEI16|I16 i
1696 =for apidoc Amh||SAVEBOOL|bool i
1698 =item C<SAVESPTR(s)>
1700 =item C<SAVEPPTR(p)>
1702 These macros arrange things to restore the value of pointers C<s> and
1703 C<p>. C<s> must be a pointer of a type which survives conversion to
1704 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1707 =for apidoc Amh||SAVESPTR|SV * s
1708 =for apidoc Amh||SAVEPPTR|char * p
1710 =item C<SAVEFREESV(SV *sv)>
1712 The refcount of C<sv> will be decremented at the end of
1713 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1714 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1715 extends the lifetime of C<sv> until the beginning of the next statement,
1716 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1717 lifetimes can be wildly different.
1719 Also compare C<SAVEMORTALIZESV>.
1721 =for apidoc Amh||SAVEFREESV|SV* sv
1723 =item C<SAVEMORTALIZESV(SV *sv)>
1725 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1726 scope instead of decrementing its reference count. This usually has the
1727 effect of keeping C<sv> alive until the statement that called the currently
1728 live scope has finished executing.
1730 =for apidoc Amh||SAVEMORTALIZESV|SV* sv
1732 =item C<SAVEFREEOP(OP *op)>
1734 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1736 =for apidoc Amh||SAVEFREEOP|OP *op
1738 =item C<SAVEFREEPV(p)>
1740 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1741 end of I<pseudo-block>.
1743 =for apidoc Amh||SAVEFREEPV|void * p
1745 =item C<SAVECLEARSV(SV *sv)>
1747 Clears a slot in the current scratchpad which corresponds to C<sv> at
1748 the end of I<pseudo-block>.
1750 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1752 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1753 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1754 short-lived storage, the corresponding string may be reallocated like
1757 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1759 =for apidoc Amh||SAVEDELETE|HV * hv|char * key|I32 length
1761 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1763 At the end of I<pseudo-block> the function C<f> is called with the
1766 =for apidoc Ayh||DESTRUCTORFUNC_NOCONTEXT_t
1767 =for apidoc Amh||SAVEDESTRUCTOR|DESTRUCTORFUNC_NOCONTEXT_t f|void *p
1769 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1771 At the end of I<pseudo-block> the function C<f> is called with the
1772 implicit context argument (if any), and C<p>.
1774 =for apidoc Ayh||DESTRUCTORFUNC_t
1775 =for apidoc Amh||SAVEDESTRUCTOR_X|DESTRUCTORFUNC_t f|void *p
1777 =item C<SAVESTACK_POS()>
1779 The current offset on the Perl internal stack (cf. C<SP>) is restored
1780 at the end of I<pseudo-block>.
1782 =for apidoc Amh||SAVESTACK_POS
1786 The following API list contains functions, thus one needs to
1787 provide pointers to the modifiable data explicitly (either C pointers,
1788 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1789 function takes C<int *>.
1791 Other macros above have functions implementing them, but its probably
1792 best to just use the macro, and not those or the ones below.
1796 =item C<SV* save_scalar(GV *gv)>
1798 =for apidoc save_scalar
1800 Equivalent to Perl code C<local $gv>.
1802 =item C<AV* save_ary(GV *gv)>
1804 =for apidoc save_ary
1806 =item C<HV* save_hash(GV *gv)>
1808 =for apidoc save_hash
1810 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1812 =item C<void save_item(SV *item)>
1814 =for apidoc save_item
1816 Duplicates the current value of C<SV>. On the exit from the current
1817 C<ENTER>/C<LEAVE> I<pseudo-block> the value of C<SV> will be restored
1818 using the stored value. It doesn't handle magic. Use C<save_scalar> if
1821 =item C<void save_list(SV **sarg, I32 maxsarg)>
1823 =for apidoc save_list
1825 A variant of C<save_item> which takes multiple arguments via an array
1826 C<sarg> of C<SV*> of length C<maxsarg>.
1828 =item C<SV* save_svref(SV **sptr)>
1830 =for apidoc save_svref
1832 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1834 =item C<void save_aptr(AV **aptr)>
1836 =item C<void save_hptr(HV **hptr)>
1838 =for apidoc save_aptr
1839 =for apidoc save_hptr
1841 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1845 The C<Alias> module implements localization of the basic types within the
1846 I<caller's scope>. People who are interested in how to localize things in
1847 the containing scope should take a look there too.
1851 =head2 XSUBs and the Argument Stack
1853 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1854 An XSUB routine will have a stack that contains the arguments from the Perl
1855 program, and a way to map from the Perl data structures to a C equivalent.
1857 The stack arguments are accessible through the C<ST(n)> macro, which returns
1858 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1859 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1862 Most of the time, output from the C routine can be handled through use of
1863 the RETVAL and OUTPUT directives. However, there are some cases where the
1864 argument stack is not already long enough to handle all the return values.
1865 An example is the POSIX tzname() call, which takes no arguments, but returns
1866 two, the local time zone's standard and summer time abbreviations.
1868 To handle this situation, the PPCODE directive is used and the stack is
1869 extended using the macro:
1873 where C<SP> is the macro that represents the local copy of the stack pointer,
1874 and C<num> is the number of elements the stack should be extended by.
1876 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1877 macro. The pushed values will often need to be "mortal" (See
1878 L</Reference Counts and Mortality>):
1880 PUSHs(sv_2mortal(newSViv(an_integer)))
1881 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1882 PUSHs(sv_2mortal(newSVnv(a_double)))
1883 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1884 /* Although the last example is better written as the more
1886 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
1888 And now the Perl program calling C<tzname>, the two values will be assigned
1891 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1893 An alternate (and possibly simpler) method to pushing values on the stack is
1898 This macro automatically adjusts the stack for you, if needed. Thus, you
1899 do not need to call C<EXTEND> to extend the stack.
1901 Despite their suggestions in earlier versions of this document the macros
1902 C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1903 For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1904 C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
1906 For more information, consult L<perlxs> and L<perlxstut>.
1908 =head2 Autoloading with XSUBs
1910 If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1911 fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1912 of the XSUB's package.
1914 But it also puts the same information in certain fields of the XSUB itself:
1916 HV *stash = CvSTASH(cv);
1917 const char *subname = SvPVX(cv);
1918 STRLEN name_length = SvCUR(cv); /* in bytes */
1919 U32 is_utf8 = SvUTF8(cv);
1921 C<SvPVX(cv)> contains just the sub name itself, not including the package.
1922 For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1923 C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
1925 B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1926 XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1927 XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1928 to support 5.8-5.14, use the XSUB's fields.
1930 =head2 Calling Perl Routines from within C Programs
1932 There are four routines that can be used to call a Perl subroutine from
1933 within a C program. These four are:
1935 I32 call_sv(SV*, I32);
1936 I32 call_pv(const char*, I32);
1937 I32 call_method(const char*, I32);
1938 I32 call_argv(const char*, I32, char**);
1940 The routine most often used is C<call_sv>. The C<SV*> argument
1941 contains either the name of the Perl subroutine to be called, or a
1942 reference to the subroutine. The second argument consists of flags
1943 that control the context in which the subroutine is called, whether
1944 or not the subroutine is being passed arguments, how errors should be
1945 trapped, and how to treat return values.
1947 All four routines return the number of arguments that the subroutine returned
1950 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1951 but those names are now deprecated; macros of the same name are provided for
1954 When using any of these routines (except C<call_argv>), the programmer
1955 must manipulate the Perl stack. These include the following macros and
1970 For a detailed description of calling conventions from C to Perl,
1971 consult L<perlcall>.
1973 =head2 Putting a C value on Perl stack
1975 A lot of opcodes (this is an elementary operation in the internal perl
1976 stack machine) put an SV* on the stack. However, as an optimization
1977 the corresponding SV is (usually) not recreated each time. The opcodes
1978 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1979 not constantly freed/created.
1981 Each of the targets is created only once (but see
1982 L</Scratchpads and recursion> below), and when an opcode needs to put
1983 an integer, a double, or a string on stack, it just sets the
1984 corresponding parts of its I<target> and puts the I<target> on stack.
1986 The macro to put this target on stack is C<PUSHTARG>, and it is
1987 directly used in some opcodes, as well as indirectly in zillions of
1988 others, which use it via C<(X)PUSH[iunp]>.
1990 Because the target is reused, you must be careful when pushing multiple
1991 values on the stack. The following code will not do what you think:
1996 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1997 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1998 At the end of the operation, the stack does not contain the values 10
1999 and 20, but actually contains two pointers to C<TARG>, which we have set
2002 If you need to push multiple different values then you should either use
2003 the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
2004 none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
2005 SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
2006 will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
2007 this a little easier to achieve by creating a new mortal for you (via
2008 C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
2009 in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
2010 Thus, instead of writing this to "fix" the example above:
2012 XPUSHs(sv_2mortal(newSViv(10)))
2013 XPUSHs(sv_2mortal(newSViv(20)))
2015 you can simply write:
2020 On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
2021 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
2022 macros can make use of the local variable C<TARG>. See also C<dTARGET>
2027 The question remains on when the SVs which are I<target>s for opcodes
2028 are created. The answer is that they are created when the current
2029 unit--a subroutine or a file (for opcodes for statements outside of
2030 subroutines)--is compiled. During this time a special anonymous Perl
2031 array is created, which is called a scratchpad for the current unit.
2033 A scratchpad keeps SVs which are lexicals for the current unit and are
2034 targets for opcodes. A previous version of this document
2035 stated that one can deduce that an SV lives on a scratchpad
2036 by looking on its flags: lexicals have C<SVs_PADMY> set, and
2037 I<target>s have C<SVs_PADTMP> set. But this has never been fully true.
2038 C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
2039 While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
2040 that have never resided in a pad, but nonetheless act like I<target>s. As
2041 of perl 5.21.5, the C<SVs_PADMY> flag is no longer used and is defined as
2042 0. C<SvPADMY()> now returns true for anything without C<SVs_PADTMP>.
2044 The correspondence between OPs and I<target>s is not 1-to-1. Different
2045 OPs in the compile tree of the unit can use the same target, if this
2046 would not conflict with the expected life of the temporary.
2048 =head2 Scratchpads and recursion
2050 In fact it is not 100% true that a compiled unit contains a pointer to
2051 the scratchpad AV. In fact it contains a pointer to an AV of
2052 (initially) one element, and this element is the scratchpad AV. Why do
2053 we need an extra level of indirection?
2055 The answer is B<recursion>, and maybe B<threads>. Both
2056 these can create several execution pointers going into the same
2057 subroutine. For the subroutine-child not write over the temporaries
2058 for the subroutine-parent (lifespan of which covers the call to the
2059 child), the parent and the child should have different
2060 scratchpads. (I<And> the lexicals should be separate anyway!)
2062 So each subroutine is born with an array of scratchpads (of length 1).
2063 On each entry to the subroutine it is checked that the current
2064 depth of the recursion is not more than the length of this array, and
2065 if it is, new scratchpad is created and pushed into the array.
2067 The I<target>s on this scratchpad are C<undef>s, but they are already
2068 marked with correct flags.
2070 =head1 Memory Allocation
2074 All memory meant to be used with the Perl API functions should be manipulated
2075 using the macros described in this section. The macros provide the necessary
2076 transparency between differences in the actual malloc implementation that is
2079 The following three macros are used to initially allocate memory :
2081 Newx(pointer, number, type);
2082 Newxc(pointer, number, type, cast);
2083 Newxz(pointer, number, type);
2085 The first argument C<pointer> should be the name of a variable that will
2086 point to the newly allocated memory.
2088 The second and third arguments C<number> and C<type> specify how many of
2089 the specified type of data structure should be allocated. The argument
2090 C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
2091 should be used if the C<pointer> argument is different from the C<type>
2094 Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
2095 to zero out all the newly allocated memory.
2099 Renew(pointer, number, type);
2100 Renewc(pointer, number, type, cast);
2103 These three macros are used to change a memory buffer size or to free a
2104 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
2105 match those of C<New> and C<Newc> with the exception of not needing the
2106 "magic cookie" argument.
2110 Move(source, dest, number, type);
2111 Copy(source, dest, number, type);
2112 Zero(dest, number, type);
2114 These three macros are used to move, copy, or zero out previously allocated
2115 memory. The C<source> and C<dest> arguments point to the source and
2116 destination starting points. Perl will move, copy, or zero out C<number>
2117 instances of the size of the C<type> data structure (using the C<sizeof>
2122 The most recent development releases of Perl have been experimenting with
2123 removing Perl's dependency on the "normal" standard I/O suite and allowing
2124 other stdio implementations to be used. This involves creating a new
2125 abstraction layer that then calls whichever implementation of stdio Perl
2126 was compiled with. All XSUBs should now use the functions in the PerlIO
2127 abstraction layer and not make any assumptions about what kind of stdio
2130 For a complete description of the PerlIO abstraction, consult L<perlapio>.
2132 =head1 Compiled code
2136 Here we describe the internal form your code is converted to by
2137 Perl. Start with a simple example:
2141 This is converted to a tree similar to this one:
2149 (but slightly more complicated). This tree reflects the way Perl
2150 parsed your code, but has nothing to do with the execution order.
2151 There is an additional "thread" going through the nodes of the tree
2152 which shows the order of execution of the nodes. In our simplified
2153 example above it looks like:
2155 $b ---> $c ---> + ---> $a ---> assign-to
2157 But with the actual compile tree for C<$a = $b + $c> it is different:
2158 some nodes I<optimized away>. As a corollary, though the actual tree
2159 contains more nodes than our simplified example, the execution order
2160 is the same as in our example.
2162 =head2 Examining the tree
2164 If you have your perl compiled for debugging (usually done with
2165 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
2166 compiled tree by specifying C<-Dx> on the Perl command line. The
2167 output takes several lines per node, and for C<$b+$c> it looks like
2172 FLAGS = (SCALAR,KIDS)
2174 TYPE = null ===> (4)
2176 FLAGS = (SCALAR,KIDS)
2178 3 TYPE = gvsv ===> 4
2184 TYPE = null ===> (5)
2186 FLAGS = (SCALAR,KIDS)
2188 4 TYPE = gvsv ===> 5
2194 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
2195 not optimized away (one per number in the left column). The immediate
2196 children of the given node correspond to C<{}> pairs on the same level
2197 of indentation, thus this listing corresponds to the tree:
2205 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
2206 4 5 6> (node C<6> is not included into above listing), i.e.,
2207 C<gvsv gvsv add whatever>.
2209 Each of these nodes represents an op, a fundamental operation inside the
2210 Perl core. The code which implements each operation can be found in the
2211 F<pp*.c> files; the function which implements the op with type C<gvsv>
2212 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
2213 different numbers of children: C<add> is a binary operator, as one would
2214 expect, and so has two children. To accommodate the various different
2215 numbers of children, there are various types of op data structure, and
2216 they link together in different ways.
2218 The simplest type of op structure is C<OP>: this has no children. Unary
2219 operators, C<UNOP>s, have one child, and this is pointed to by the
2220 C<op_first> field. Binary operators (C<BINOP>s) have not only an
2221 C<op_first> field but also an C<op_last> field. The most complex type of
2222 op is a C<LISTOP>, which has any number of children. In this case, the
2223 first child is pointed to by C<op_first> and the last child by
2224 C<op_last>. The children in between can be found by iteratively
2225 following the C<OpSIBLING> pointer from the first child to the last (but
2229 =for apidoc Ayh||BINOP
2230 =for apidoc Ayh||LISTOP
2231 =for apidoc Ayh||UNOP
2233 There are also some other op types: a C<PMOP> holds a regular expression,
2234 and has no children, and a C<LOOP> may or may not have children. If the
2235 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
2236 complicate matters, if a C<UNOP> is actually a C<null> op after
2237 optimization (see L</Compile pass 2: context propagation>) it will still
2238 have children in accordance with its former type.
2240 =for apidoc Ayh||LOOP
2241 =for apidoc Ayh||PMOP
2243 Finally, there is a C<LOGOP>, or logic op. Like a C<LISTOP>, this has one
2244 or more children, but it doesn't have an C<op_last> field: so you have to
2245 follow C<op_first> and then the C<OpSIBLING> chain itself to find the
2246 last child. Instead it has an C<op_other> field, which is comparable to
2247 the C<op_next> field described below, and represents an alternate
2248 execution path. Operators like C<and>, C<or> and C<?> are C<LOGOP>s. Note
2249 that in general, C<op_other> may not point to any of the direct children
2252 =for apidoc Ayh||LOGOP
2254 Starting in version 5.21.2, perls built with the experimental
2255 define C<-DPERL_OP_PARENT> add an extra boolean flag for each op,
2256 C<op_moresib>. When not set, this indicates that this is the last op in an
2257 C<OpSIBLING> chain. This frees up the C<op_sibling> field on the last
2258 sibling to point back to the parent op. Under this build, that field is
2259 also renamed C<op_sibparent> to reflect its joint role. The macro
2260 C<OpSIBLING(o)> wraps this special behaviour, and always returns NULL on
2261 the last sibling. With this build the C<op_parent(o)> function can be
2262 used to find the parent of any op. Thus for forward compatibility, you
2263 should always use the C<OpSIBLING(o)> macro rather than accessing
2264 C<op_sibling> directly.
2266 Another way to examine the tree is to use a compiler back-end module, such
2269 =head2 Compile pass 1: check routines
2271 The tree is created by the compiler while I<yacc> code feeds it
2272 the constructions it recognizes. Since I<yacc> works bottom-up, so does
2273 the first pass of perl compilation.
2275 What makes this pass interesting for perl developers is that some
2276 optimization may be performed on this pass. This is optimization by
2277 so-called "check routines". The correspondence between node names
2278 and corresponding check routines is described in F<opcode.pl> (do not
2279 forget to run C<make regen_headers> if you modify this file).
2281 A check routine is called when the node is fully constructed except
2282 for the execution-order thread. Since at this time there are no
2283 back-links to the currently constructed node, one can do most any
2284 operation to the top-level node, including freeing it and/or creating
2285 new nodes above/below it.
2287 The check routine returns the node which should be inserted into the
2288 tree (if the top-level node was not modified, check routine returns
2291 By convention, check routines have names C<ck_*>. They are usually
2292 called from C<new*OP> subroutines (or C<convert>) (which in turn are
2293 called from F<perly.y>).
2295 =head2 Compile pass 1a: constant folding
2297 Immediately after the check routine is called the returned node is
2298 checked for being compile-time executable. If it is (the value is
2299 judged to be constant) it is immediately executed, and a I<constant>
2300 node with the "return value" of the corresponding subtree is
2301 substituted instead. The subtree is deleted.
2303 If constant folding was not performed, the execution-order thread is
2306 =head2 Compile pass 2: context propagation
2308 When a context for a part of compile tree is known, it is propagated
2309 down through the tree. At this time the context can have 5 values
2310 (instead of 2 for runtime context): void, boolean, scalar, list, and
2311 lvalue. In contrast with the pass 1 this pass is processed from top
2312 to bottom: a node's context determines the context for its children.
2314 Additional context-dependent optimizations are performed at this time.
2315 Since at this moment the compile tree contains back-references (via
2316 "thread" pointers), nodes cannot be free()d now. To allow
2317 optimized-away nodes at this stage, such nodes are null()ified instead
2318 of free()ing (i.e. their type is changed to OP_NULL).
2320 =head2 Compile pass 3: peephole optimization
2322 After the compile tree for a subroutine (or for an C<eval> or a file)
2323 is created, an additional pass over the code is performed. This pass
2324 is neither top-down or bottom-up, but in the execution order (with
2325 additional complications for conditionals). Optimizations performed
2326 at this stage are subject to the same restrictions as in the pass 2.
2328 Peephole optimizations are done by calling the function pointed to
2329 by the global variable C<PL_peepp>. By default, C<PL_peepp> just
2330 calls the function pointed to by the global variable C<PL_rpeepp>.
2331 By default, that performs some basic op fixups and optimisations along
2332 the execution-order op chain, and recursively calls C<PL_rpeepp> for
2333 each side chain of ops (resulting from conditionals). Extensions may
2334 provide additional optimisations or fixups, hooking into either the
2335 per-subroutine or recursive stage, like this:
2337 static peep_t prev_peepp;
2338 static void my_peep(pTHX_ OP *o)
2340 /* custom per-subroutine optimisation goes here */
2341 prev_peepp(aTHX_ o);
2342 /* custom per-subroutine optimisation may also go here */
2345 prev_peepp = PL_peepp;
2348 static peep_t prev_rpeepp;
2349 static void my_rpeep(pTHX_ OP *first)
2351 OP *o = first, *t = first;
2352 for(; o = o->op_next, t = t->op_next) {
2353 /* custom per-op optimisation goes here */
2355 if (!o || o == t) break;
2356 /* custom per-op optimisation goes AND here */
2358 prev_rpeepp(aTHX_ orig_o);
2361 prev_rpeepp = PL_rpeepp;
2362 PL_rpeepp = my_rpeep;
2364 =for apidoc Ayh||peep_t
2366 =head2 Pluggable runops
2368 The compile tree is executed in a runops function. There are two runops
2369 functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
2370 with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
2371 control over the execution of the compile tree it is possible to provide
2372 your own runops function.
2374 It's probably best to copy one of the existing runops functions and
2375 change it to suit your needs. Then, in the BOOT section of your XS
2378 PL_runops = my_runops;
2380 =for apidoc Amnh|runops_proc_t|PL_runops
2382 This function should be as efficient as possible to keep your programs
2383 running as fast as possible.
2385 =head2 Compile-time scope hooks
2387 As of perl 5.14 it is possible to hook into the compile-time lexical
2388 scope mechanism using C<Perl_blockhook_register>. This is used like
2391 STATIC void my_start_hook(pTHX_ int full);
2392 STATIC BHK my_hooks;
2395 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
2396 Perl_blockhook_register(aTHX_ &my_hooks);
2398 This will arrange to have C<my_start_hook> called at the start of
2399 compiling every lexical scope. The available hooks are:
2401 =for apidoc Ayh||BHK
2405 =item C<void bhk_start(pTHX_ int full)>
2407 This is called just after starting a new lexical scope. Note that Perl
2412 creates two scopes: the first starts at the C<(> and has C<full == 1>,
2413 the second starts at the C<{> and has C<full == 0>. Both end at the
2414 C<}>, so calls to C<start> and C<pre>/C<post_end> will match. Anything
2415 pushed onto the save stack by this hook will be popped just before the
2416 scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2418 =item C<void bhk_pre_end(pTHX_ OP **o)>
2420 This is called at the end of a lexical scope, just before unwinding the
2421 stack. I<o> is the root of the optree representing the scope; it is a
2422 double pointer so you can replace the OP if you need to.
2424 =item C<void bhk_post_end(pTHX_ OP **o)>
2426 This is called at the end of a lexical scope, just after unwinding the
2427 stack. I<o> is as above. Note that it is possible for calls to C<pre_>
2428 and C<post_end> to nest, if there is something on the save stack that
2431 =item C<void bhk_eval(pTHX_ OP *const o)>
2433 This is called just before starting to compile an C<eval STRING>, C<do
2434 FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
2435 OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2436 C<OP_DOFILE> or C<OP_REQUIRE>.
2440 Once you have your hook functions, you need a C<BHK> structure to put
2441 them in. It's best to allocate it statically, since there is no way to
2442 free it once it's registered. The function pointers should be inserted
2443 into this structure using the C<BhkENTRY_set> macro, which will also set
2444 flags indicating which entries are valid. If you do need to allocate
2445 your C<BHK> dynamically for some reason, be sure to zero it before you
2448 Once registered, there is no mechanism to switch these hooks off, so if
2449 that is necessary you will need to do this yourself. An entry in C<%^H>
2450 is probably the best way, so the effect is lexically scoped; however it
2451 is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
2452 temporarily switch entries on and off. You should also be aware that
2453 generally speaking at least one scope will have opened before your
2454 extension is loaded, so you will see some C<pre>/C<post_end> pairs that
2455 didn't have a matching C<start>.
2457 =head1 Examining internal data structures with the C<dump> functions
2459 To aid debugging, the source file F<dump.c> contains a number of
2460 functions which produce formatted output of internal data structures.
2462 The most commonly used of these functions is C<Perl_sv_dump>; it's used
2463 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
2464 C<sv_dump> to produce debugging output from Perl-space, so users of that
2465 module should already be familiar with its format.
2467 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
2468 derivatives, and produces output similar to C<perl -Dx>; in fact,
2469 C<Perl_dump_eval> will dump the main root of the code being evaluated,
2470 exactly like C<-Dx>.
2472 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2473 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2474 subroutines in a package like so: (Thankfully, these are all xsubs, so
2475 there is no op tree)
2477 (gdb) print Perl_dump_packsubs(PL_defstash)
2479 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2481 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2483 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2485 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2487 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2489 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2490 the op tree of the main root.
2492 =head1 How multiple interpreters and concurrency are supported
2494 =head2 Background and PERL_IMPLICIT_CONTEXT
2496 The Perl interpreter can be regarded as a closed box: it has an API
2497 for feeding it code or otherwise making it do things, but it also has
2498 functions for its own use. This smells a lot like an object, and
2499 there is a way for you to build Perl so that you can have multiple
2500 interpreters, with one interpreter represented either as a C structure,
2501 or inside a thread-specific structure. These structures contain all
2502 the context, the state of that interpreter.
2504 The macro that controls the major Perl build flavor is MULTIPLICITY. The
2505 MULTIPLICITY build has a C structure that packages all the interpreter
2506 state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2507 normally defined, and enables the support for passing in a "hidden" first
2508 argument that represents all three data structures. MULTIPLICITY makes
2509 multi-threaded perls possible (with the ithreads threading model, related
2510 to the macro USE_ITHREADS.)
2512 To see whether you have non-const data you can use a BSD (or GNU)
2515 nm libperl.a | grep -v ' [TURtr] '
2517 If this displays any C<D> or C<d> symbols (or possibly C<C> or C<c>),
2518 you have non-const data. The symbols the C<grep> removed are as follows:
2519 C<Tt> are I<text>, or code, the C<Rr> are I<read-only> (const) data,
2520 and the C<U> is <undefined>, external symbols referred to.
2522 The test F<t/porting/libperl.t> does this kind of symbol sanity
2523 checking on C<libperl.a>.
2525 All this obviously requires a way for the Perl internal functions to be
2526 either subroutines taking some kind of structure as the first
2527 argument, or subroutines taking nothing as the first argument. To
2528 enable these two very different ways of building the interpreter,
2529 the Perl source (as it does in so many other situations) makes heavy
2530 use of macros and subroutine naming conventions.
2532 First problem: deciding which functions will be public API functions and
2533 which will be private. All functions whose names begin C<S_> are private
2534 (think "S" for "secret" or "static"). All other functions begin with
2535 "Perl_", but just because a function begins with "Perl_" does not mean it is
2536 part of the API. (See L</Internal
2537 Functions>.) The easiest way to be B<sure> a
2538 function is part of the API is to find its entry in L<perlapi>.
2539 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2540 think it should be (i.e., you need it for your extension), submit an issue at
2541 L<https://github.com/Perl/perl5/issues> explaining why you think it should be.
2543 Second problem: there must be a syntax so that the same subroutine
2544 declarations and calls can pass a structure as their first argument,
2545 or pass nothing. To solve this, the subroutines are named and
2546 declared in a particular way. Here's a typical start of a static
2547 function used within the Perl guts:
2550 S_incline(pTHX_ char *s)
2552 STATIC becomes "static" in C, and may be #define'd to nothing in some
2553 configurations in the future.
2555 =for apidoc_section $directives
2556 =for apidoc Ayh||STATIC
2558 A public function (i.e. part of the internal API, but not necessarily
2559 sanctioned for use in extensions) begins like this:
2562 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
2564 C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
2565 details of the interpreter's context. THX stands for "thread", "this",
2566 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2567 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
2568 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2571 =for apidoc_section $concurrency
2572 =for apidoc Amnh||aTHX
2573 =for apidoc Amnh||aTHX_
2574 =for apidoc Amnh||dTHX
2575 =for apidoc Amnh||pTHX
2576 =for apidoc Amnh||pTHX_
2578 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2579 first argument containing the interpreter's context. The trailing underscore
2580 in the pTHX_ macro indicates that the macro expansion needs a comma
2581 after the context argument because other arguments follow it. If
2582 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
2583 subroutine is not prototyped to take the extra argument. The form of the
2584 macro without the trailing underscore is used when there are no additional
2587 When a core function calls another, it must pass the context. This
2588 is normally hidden via macros. Consider C<sv_setiv>. It expands into
2589 something like this:
2591 #ifdef PERL_IMPLICIT_CONTEXT
2592 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
2593 /* can't do this for vararg functions, see below */
2595 #define sv_setiv Perl_sv_setiv
2598 This works well, and means that XS authors can gleefully write:
2602 and still have it work under all the modes Perl could have been
2605 This doesn't work so cleanly for varargs functions, though, as macros
2606 imply that the number of arguments is known in advance. Instead we
2607 either need to spell them out fully, passing C<aTHX_> as the first
2608 argument (the Perl core tends to do this with functions like
2609 Perl_warner), or use a context-free version.
2611 The context-free version of Perl_warner is called
2612 Perl_warner_nocontext, and does not take the extra argument. Instead
2613 it does C<dTHX;> to get the context from thread-local storage. We
2614 C<#define warner Perl_warner_nocontext> so that extensions get source
2615 compatibility at the expense of performance. (Passing an arg is
2616 cheaper than grabbing it from thread-local storage.)
2618 You can ignore [pad]THXx when browsing the Perl headers/sources.
2619 Those are strictly for use within the core. Extensions and embedders
2620 need only be aware of [pad]THX.
2622 =head2 So what happened to dTHR?
2624 =for apidoc Amnh||dTHR
2626 C<dTHR> was introduced in perl 5.005 to support the older thread model.
2627 The older thread model now uses the C<THX> mechanism to pass context
2628 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2629 later still have it for backward source compatibility, but it is defined
2632 =head2 How do I use all this in extensions?
2634 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2635 any functions in the Perl API will need to pass the initial context
2636 argument somehow. The kicker is that you will need to write it in
2637 such a way that the extension still compiles when Perl hasn't been
2638 built with PERL_IMPLICIT_CONTEXT enabled.
2640 There are three ways to do this. First, the easy but inefficient way,
2641 which is also the default, in order to maintain source compatibility
2642 with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
2643 and aTHX_ macros to call a function that will return the context.
2644 Thus, something like:
2648 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
2651 Perl_sv_setiv(Perl_get_context(), sv, num);
2653 or to this otherwise:
2655 Perl_sv_setiv(sv, num);
2657 You don't have to do anything new in your extension to get this; since
2658 the Perl library provides Perl_get_context(), it will all just
2661 The second, more efficient way is to use the following template for
2664 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2669 STATIC void my_private_function(int arg1, int arg2);
2672 my_private_function(int arg1, int arg2)
2674 dTHX; /* fetch context */
2675 ... call many Perl API functions ...
2680 MODULE = Foo PACKAGE = Foo
2688 my_private_function(arg, 10);
2690 Note that the only two changes from the normal way of writing an
2691 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2692 including the Perl headers, followed by a C<dTHX;> declaration at
2693 the start of every function that will call the Perl API. (You'll
2694 know which functions need this, because the C compiler will complain
2695 that there's an undeclared identifier in those functions.) No changes
2696 are needed for the XSUBs themselves, because the XS() macro is
2697 correctly defined to pass in the implicit context if needed.
2699 The third, even more efficient way is to ape how it is done within
2703 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2708 /* pTHX_ only needed for functions that call Perl API */
2709 STATIC void my_private_function(pTHX_ int arg1, int arg2);
2712 my_private_function(pTHX_ int arg1, int arg2)
2714 /* dTHX; not needed here, because THX is an argument */
2715 ... call Perl API functions ...
2720 MODULE = Foo PACKAGE = Foo
2728 my_private_function(aTHX_ arg, 10);
2730 This implementation never has to fetch the context using a function
2731 call, since it is always passed as an extra argument. Depending on
2732 your needs for simplicity or efficiency, you may mix the previous
2733 two approaches freely.
2735 Never add a comma after C<pTHX> yourself--always use the form of the
2736 macro with the underscore for functions that take explicit arguments,
2737 or the form without the argument for functions with no explicit arguments.
2739 =head2 Should I do anything special if I call perl from multiple threads?
2741 If you create interpreters in one thread and then proceed to call them in
2742 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2743 initialized correctly in each of those threads.
2745 The C<perl_alloc> and C<perl_clone> API functions will automatically set
2746 the TLS slot to the interpreter they created, so that there is no need to do
2747 anything special if the interpreter is always accessed in the same thread that
2748 created it, and that thread did not create or call any other interpreters
2749 afterwards. If that is not the case, you have to set the TLS slot of the
2750 thread before calling any functions in the Perl API on that particular
2751 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2752 thread as the first thing you do:
2754 /* do this before doing anything else with some_perl */
2755 PERL_SET_CONTEXT(some_perl);
2757 ... other Perl API calls on some_perl go here ...
2759 =head2 Future Plans and PERL_IMPLICIT_SYS
2761 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2762 that the interpreter knows about itself and pass it around, so too are
2763 there plans to allow the interpreter to bundle up everything it knows
2764 about the environment it's running on. This is enabled with the
2765 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2768 This allows the ability to provide an extra pointer (called the "host"
2769 environment) for all the system calls. This makes it possible for
2770 all the system stuff to maintain their own state, broken down into
2771 seven C structures. These are thin wrappers around the usual system
2772 calls (see F<win32/perllib.c>) for the default perl executable, but for a
2773 more ambitious host (like the one that would do fork() emulation) all
2774 the extra work needed to pretend that different interpreters are
2775 actually different "processes", would be done here.
2777 The Perl engine/interpreter and the host are orthogonal entities.
2778 There could be one or more interpreters in a process, and one or
2779 more "hosts", with free association between them.
2781 =head1 Internal Functions
2783 All of Perl's internal functions which will be exposed to the outside
2784 world are prefixed by C<Perl_> so that they will not conflict with XS
2785 functions or functions used in a program in which Perl is embedded.
2786 Similarly, all global variables begin with C<PL_>. (By convention,
2787 static functions start with C<S_>.)
2789 Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2790 either with or without the C<Perl_> prefix, thanks to a bunch of defines
2791 that live in F<embed.h>. Note that extension code should I<not> set
2792 C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2793 breakage of the XS in each new perl release.
2795 The file F<embed.h> is generated automatically from
2796 F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2797 header files for the internal functions, generates the documentation
2798 and a lot of other bits and pieces. It's important that when you add
2799 a new function to the core or change an existing one, you change the
2800 data in the table in F<embed.fnc> as well. Here's a sample entry from
2803 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2805 The first column is a set of flags, the second column the return type,
2806 the third column the name. Columns after that are the arguments.
2807 The flags are documented at the top of F<embed.fnc>.
2809 If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2810 C<make regen_headers> to force a rebuild of F<embed.h> and other
2811 auto-generated files.
2813 =head2 Formatted Printing of IVs, UVs, and NVs
2815 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2816 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2817 following macros for portability
2822 UVxf UV in hexadecimal
2827 These will take care of 64-bit integers and long doubles.
2830 printf("IV is %" IVdf "\n", iv);
2832 The C<IVdf> will expand to whatever is the correct format for the IVs.
2833 Note that the spaces are required around the format in case the code is
2834 compiled with C++, to maintain compliance with its standard.
2836 Note that there are different "long doubles": Perl will use
2837 whatever the compiler has.
2839 If you are printing addresses of pointers, use %p or UVxf combined
2842 =head2 Formatted Printing of SVs
2844 The contents of SVs may be printed using the C<SVf> format, like so:
2846 Perl_croak(aTHX_ "This croaked because: %" SVf "\n", SVfARG(err_msg))
2848 where C<err_msg> is an SV.
2850 =for apidoc Amnh||SVf
2851 =for apidoc Amh||SVfARG|SV *sv
2853 Not all scalar types are printable. Simple values certainly are: one of
2854 IV, UV, NV, or PV. Also, if the SV is a reference to some value,
2855 either it will be dereferenced and the value printed, or information
2856 about the type of that value and its address are displayed. The results
2857 of printing any other type of SV are undefined and likely to lead to an
2858 interpreter crash. NVs are printed using a C<%g>-ish format.
2860 Note that the spaces are required around the C<SVf> in case the code is
2861 compiled with C++, to maintain compliance with its standard.
2863 Note that any filehandle being printed to under UTF-8 must be expecting
2864 UTF-8 in order to get good results and avoid Wide-character warnings.
2865 One way to do this for typical filehandles is to invoke perl with the
2866 C<-C>> parameter. (See L<perlrun/-C [numberE<sol>list]>.
2868 You can use this to concatenate two scalars:
2870 SV *var1 = get_sv("var1", GV_ADD);
2871 SV *var2 = get_sv("var2", GV_ADD);
2872 SV *var3 = newSVpvf("var1=%" SVf " and var2=%" SVf,
2873 SVfARG(var1), SVfARG(var2));
2875 =head2 Formatted Printing of Strings
2877 If you just want the bytes printed in a 7bit NUL-terminated string, you can
2878 just use C<%s> (assuming they are all really only 7bit). But if there is a
2879 possibility the value will be encoded as UTF-8 or contains bytes above
2880 C<0x7F> (and therefore 8bit), you should instead use the C<UTF8f> format.
2881 And as its parameter, use the C<UTF8fARG()> macro:
2885 /* U+2018: \xE2\x80\x98 LEFT SINGLE QUOTATION MARK
2886 U+2019: \xE2\x80\x99 RIGHT SINGLE QUOTATION MARK */
2888 msg = "\xE2\x80\x98Uses fancy quotes\xE2\x80\x99";
2890 msg = "'Uses simple quotes'";
2892 Perl_croak(aTHX_ "The message is: %" UTF8f "\n",
2893 UTF8fARG(can_utf8, strlen(msg), msg));
2895 The first parameter to C<UTF8fARG> is a boolean: 1 if the string is in
2896 UTF-8; 0 if string is in native byte encoding (Latin1).
2897 The second parameter is the number of bytes in the string to print.
2898 And the third and final parameter is a pointer to the first byte in the
2901 Note that any filehandle being printed to under UTF-8 must be expecting
2902 UTF-8 in order to get good results and avoid Wide-character warnings.
2903 One way to do this for typical filehandles is to invoke perl with the
2904 C<-C>> parameter. (See L<perlrun/-C [numberE<sol>list]>.
2906 =for apidoc_section $formats
2907 =for apidoc Amnh||UTF8f
2908 =for apidoc Amh||UTF8fARG|bool is_utf8|Size_t byte_len|char *str
2912 =head2 Formatted Printing of C<Size_t> and C<SSize_t>
2914 The most general way to do this is to cast them to a UV or IV, and
2916 L<previous section|/Formatted Printing of IVs, UVs, and NVs>.
2918 But if you're using C<PerlIO_printf()>, it's less typing and visual
2919 clutter to use the C<%z> length modifier (for I<siZe>):
2921 PerlIO_printf("STRLEN is %zu\n", len);
2923 This modifier is not portable, so its use should be restricted to
2926 =head2 Formatted Printing of C<Ptrdiff_t>, C<intmax_t>, C<short> and other special sizes
2928 There are modifiers for these special situations if you are using
2929 C<PerlIO_printf()>. See L<perlfunc/size>.
2931 =head2 Pointer-To-Integer and Integer-To-Pointer
2933 Because pointer size does not necessarily equal integer size,
2934 use the follow macros to do it right.
2939 INT2PTR(pointertotype, integer)
2941 =for apidoc_section $casting
2942 =for apidoc Amh|type|INT2PTR|type|int value
2943 =for apidoc Amh|UV|PTR2UV|void * ptr
2944 =for apidoc Amh|IV|PTR2IV|void * ptr
2945 =for apidoc Amh|NV|PTR2NV|void * ptr
2950 SV *sv = INT2PTR(SV*, iv);
2959 PTR2nat(pointer) /* pointer to integer of PTRSIZE */
2960 PTR2ul(pointer) /* pointer to unsigned long */
2962 =for apidoc Amh|IV|PTR2nat|void *
2963 =for apidoc Amh|unsigned long|PTR2ul|void *
2965 And C<PTRV> which gives the native type for an integer the same size as
2966 pointers, such as C<unsigned> or C<unsigned long>.
2968 =for apidoc Ayh|type|PTRV
2970 =head2 Exception Handling
2972 There are a couple of macros to do very basic exception handling in XS
2973 modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
2974 be able to use these macros:
2979 You can use these macros if you call code that may croak, but you need
2980 to do some cleanup before giving control back to Perl. For example:
2982 dXCPT; /* set up necessary variables */
2985 code_that_may_croak();
2990 /* do cleanup here */
2994 Note that you always have to rethrow an exception that has been
2995 caught. Using these macros, it is not possible to just catch the
2996 exception and ignore it. If you have to ignore the exception, you
2997 have to use the C<call_*> function.
2999 The advantage of using the above macros is that you don't have
3000 to setup an extra function for C<call_*>, and that using these
3001 macros is faster than using C<call_*>.
3003 =head2 Source Documentation
3005 There's an effort going on to document the internal functions and
3006 automatically produce reference manuals from them -- L<perlapi> is one
3007 such manual which details all the functions which are available to XS
3008 writers. L<perlintern> is the autogenerated manual for the functions
3009 which are not part of the API and are supposedly for internal use only.
3011 Source documentation is created by putting POD comments into the C
3015 =for apidoc sv_setiv
3017 Copies an integer into the given SV. Does not handle 'set' magic. See
3018 L<perlapi/sv_setiv_mg>.
3023 Please try and supply some documentation if you add functions to the
3026 =head2 Backwards compatibility
3028 The Perl API changes over time. New functions are
3029 added or the interfaces of existing functions are
3030 changed. The C<Devel::PPPort> module tries to
3031 provide compatibility code for some of these changes, so XS writers don't
3032 have to code it themselves when supporting multiple versions of Perl.
3034 C<Devel::PPPort> generates a C header file F<ppport.h> that can also
3035 be run as a Perl script. To generate F<ppport.h>, run:
3037 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
3039 Besides checking existing XS code, the script can also be used to retrieve
3040 compatibility information for various API calls using the C<--api-info>
3041 command line switch. For example:
3043 % perl ppport.h --api-info=sv_magicext
3045 For details, see C<perldoc ppport.h>.
3047 =head1 Unicode Support
3049 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
3050 writers to understand this support and make sure that the code they
3051 write does not corrupt Unicode data.
3053 =head2 What B<is> Unicode, anyway?
3055 In the olden, less enlightened times, we all used to use ASCII. Most of
3056 us did, anyway. The big problem with ASCII is that it's American. Well,
3057 no, that's not actually the problem; the problem is that it's not
3058 particularly useful for people who don't use the Roman alphabet. What
3059 used to happen was that particular languages would stick their own
3060 alphabet in the upper range of the sequence, between 128 and 255. Of
3061 course, we then ended up with plenty of variants that weren't quite
3062 ASCII, and the whole point of it being a standard was lost.
3064 Worse still, if you've got a language like Chinese or
3065 Japanese that has hundreds or thousands of characters, then you really
3066 can't fit them into a mere 256, so they had to forget about ASCII
3067 altogether, and build their own systems using pairs of numbers to refer
3070 To fix this, some people formed Unicode, Inc. and
3071 produced a new character set containing all the characters you can
3072 possibly think of and more. There are several ways of representing these
3073 characters, and the one Perl uses is called UTF-8. UTF-8 uses
3074 a variable number of bytes to represent a character. You can learn more
3075 about Unicode and Perl's Unicode model in L<perlunicode>.
3077 (On EBCDIC platforms, Perl uses instead UTF-EBCDIC, which is a form of
3078 UTF-8 adapted for EBCDIC platforms. Below, we just talk about UTF-8.
3079 UTF-EBCDIC is like UTF-8, but the details are different. The macros
3080 hide the differences from you, just remember that the particular numbers
3081 and bit patterns presented below will differ in UTF-EBCDIC.)
3083 =head2 How can I recognise a UTF-8 string?
3085 You can't. This is because UTF-8 data is stored in bytes just like
3086 non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
3087 capital E with a grave accent, is represented by the two bytes
3088 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
3089 has that byte sequence as well. So you can't tell just by looking -- this
3090 is what makes Unicode input an interesting problem.
3092 In general, you either have to know what you're dealing with, or you
3093 have to guess. The API function C<is_utf8_string> can help; it'll tell
3094 you if a string contains only valid UTF-8 characters, and the chances
3095 of a non-UTF-8 string looking like valid UTF-8 become very small very
3096 quickly with increasing string length. On a character-by-character
3097 basis, C<isUTF8_CHAR>
3098 will tell you whether the current character in a string is valid UTF-8.
3100 =head2 How does UTF-8 represent Unicode characters?
3102 As mentioned above, UTF-8 uses a variable number of bytes to store a
3103 character. Characters with values 0...127 are stored in one
3104 byte, just like good ol' ASCII. Character 128 is stored as
3105 C<v194.128>; this continues up to character 191, which is
3106 C<v194.191>. Now we've run out of bits (191 is binary
3107 C<10111111>) so we move on; character 192 is C<v195.128>. And
3108 so it goes on, moving to three bytes at character 2048.
3109 L<perlunicode/Unicode Encodings> has pictures of how this works.
3111 Assuming you know you're dealing with a UTF-8 string, you can find out
3112 how long the first character in it is with the C<UTF8SKIP> macro:
3114 char *utf = "\305\233\340\240\201";
3117 len = UTF8SKIP(utf); /* len is 2 here */
3119 len = UTF8SKIP(utf); /* len is 3 here */
3121 Another way to skip over characters in a UTF-8 string is to use
3122 C<utf8_hop>, which takes a string and a number of characters to skip
3123 over. You're on your own about bounds checking, though, so don't use it
3126 All bytes in a multi-byte UTF-8 character will have the high bit set,
3127 so you can test if you need to do something special with this
3128 character like this (the C<UTF8_IS_INVARIANT()> is a macro that tests
3129 whether the byte is encoded as a single byte even in UTF-8):
3131 U8 *utf; /* Initialize this to point to the beginning of the
3132 sequence to convert */
3133 U8 *utf_end; /* Initialize this to 1 beyond the end of the sequence
3134 pointed to by 'utf' */
3135 UV uv; /* Returned code point; note: a UV, not a U8, not a
3137 STRLEN len; /* Returned length of character in bytes */
3139 if (!UTF8_IS_INVARIANT(*utf))
3140 /* Must treat this as UTF-8 */
3141 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
3143 /* OK to treat this character as a byte */
3146 You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
3147 value of the character; the inverse function C<uvchr_to_utf8> is available
3148 for putting a UV into UTF-8:
3150 if (!UVCHR_IS_INVARIANT(uv))
3151 /* Must treat this as UTF8 */
3152 utf8 = uvchr_to_utf8(utf8, uv);
3154 /* OK to treat this character as a byte */
3157 You B<must> convert characters to UVs using the above functions if
3158 you're ever in a situation where you have to match UTF-8 and non-UTF-8
3159 characters. You may not skip over UTF-8 characters in this case. If you
3160 do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
3161 for instance, if your UTF-8 string contains C<v196.172>, and you skip
3162 that character, you can never match a C<chr(200)> in a non-UTF-8 string.
3165 (Note that we don't have to test for invariant characters in the
3166 examples above. The functions work on any well-formed UTF-8 input.
3167 It's just that its faster to avoid the function overhead when it's not
3170 =head2 How does Perl store UTF-8 strings?
3172 Currently, Perl deals with UTF-8 strings and non-UTF-8 strings
3173 slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
3174 string is internally encoded as UTF-8. Without it, the byte value is the
3175 codepoint number and vice versa. This flag is only meaningful if the SV
3176 is C<SvPOK> or immediately after stringification via C<SvPV> or a
3177 similar macro. You can check and manipulate this flag with the
3184 This flag has an important effect on Perl's treatment of the string: if
3185 UTF-8 data is not properly distinguished, regular expressions,
3186 C<length>, C<substr> and other string handling operations will have
3187 undesirable (wrong) results.
3189 The problem comes when you have, for instance, a string that isn't
3190 flagged as UTF-8, and contains a byte sequence that could be UTF-8 --
3191 especially when combining non-UTF-8 and UTF-8 strings.
3193 Never forget that the C<SVf_UTF8> flag is separate from the PV value; you
3194 need to be sure you don't accidentally knock it off while you're
3195 manipulating SVs. More specifically, you cannot expect to do this:
3204 nsv = newSVpvn(p, len);
3206 The C<char*> string does not tell you the whole story, and you can't
3207 copy or reconstruct an SV just by copying the string value. Check if the
3208 old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
3212 is_utf8 = SvUTF8(sv);
3213 frobnicate(p, is_utf8);
3214 nsv = newSVpvn(p, len);
3218 In the above, your C<frobnicate> function has been changed to be made
3219 aware of whether or not it's dealing with UTF-8 data, so that it can
3220 handle the string appropriately.
3222 Since just passing an SV to an XS function and copying the data of
3223 the SV is not enough to copy the UTF8 flags, even less right is just
3224 passing a S<C<char *>> to an XS function.
3226 For full generality, use the L<C<DO_UTF8>|perlapi/DO_UTF8> macro to see if the
3227 string in an SV is to be I<treated> as UTF-8. This takes into account
3228 if the call to the XS function is being made from within the scope of
3229 L<S<C<use bytes>>|bytes>. If so, the underlying bytes that comprise the
3230 UTF-8 string are to be exposed, rather than the character they
3231 represent. But this pragma should only really be used for debugging and
3232 perhaps low-level testing at the byte level. Hence most XS code need
3233 not concern itself with this, but various areas of the perl core do need
3236 And this isn't the whole story. Starting in Perl v5.12, strings that
3237 aren't encoded in UTF-8 may also be treated as Unicode under various
3238 conditions (see L<perlunicode/ASCII Rules versus Unicode Rules>).
3239 This is only really a problem for characters whose ordinals are between
3240 128 and 255, and their behavior varies under ASCII versus Unicode rules
3241 in ways that your code cares about (see L<perlunicode/The "Unicode Bug">).
3242 There is no published API for dealing with this, as it is subject to
3243 change, but you can look at the code for C<pp_lc> in F<pp.c> for an
3244 example as to how it's currently done.
3246 =head2 How do I convert a string to UTF-8?
3248 If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
3249 the non-UTF-8 strings to UTF-8. If you've got an SV, the easiest way to do
3252 sv_utf8_upgrade(sv);
3254 However, you must not do this, for example:
3257 sv_utf8_upgrade(left);
3259 If you do this in a binary operator, you will actually change one of the
3260 strings that came into the operator, and, while it shouldn't be noticeable
3261 by the end user, it can cause problems in deficient code.
3263 Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
3264 string argument. This is useful for having the data available for
3265 comparisons and so on, without harming the original SV. There's also
3266 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
3267 the string contains any characters above 255 that can't be represented
3270 =head2 How do I compare strings?
3272 L<perlapi/sv_cmp> and L<perlapi/sv_cmp_flags> do a lexigraphic
3273 comparison of two SV's, and handle UTF-8ness properly. Note, however,
3274 that Unicode specifies a much fancier mechanism for collation, available
3275 via the L<Unicode::Collate> module.
3277 To just compare two strings for equality/non-equality, you can just use
3278 L<C<memEQ()>|perlapi/memEQ> and L<C<memNE()>|perlapi/memEQ> as usual,
3279 except the strings must be both UTF-8 or not UTF-8 encoded.
3281 To compare two strings case-insensitively, use
3282 L<C<foldEQ_utf8()>|perlapi/foldEQ_utf8> (the strings don't have to have
3283 the same UTF-8ness).
3285 =head2 Is there anything else I need to know?
3287 Not really. Just remember these things:
3293 There's no way to tell if a S<C<char *>> or S<C<U8 *>> string is UTF-8
3294 or not. But you can tell if an SV is to be treated as UTF-8 by calling
3295 C<DO_UTF8> on it, after stringifying it with C<SvPV> or a similar
3296 macro. And, you can tell if SV is actually UTF-8 (even if it is not to
3297 be treated as such) by looking at its C<SvUTF8> flag (again after
3298 stringifying it). Don't forget to set the flag if something should be
3300 Treat the flag as part of the PV, even though it's not -- if you pass on
3301 the PV to somewhere, pass on the flag too.
3305 If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3306 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
3310 When writing a character UV to a UTF-8 string, B<always> use
3311 C<uvchr_to_utf8>, unless C<UVCHR_IS_INVARIANT(uv))> in which case
3312 you can use C<*s = uv>.
3316 Mixing UTF-8 and non-UTF-8 strings is
3317 tricky. Use C<bytes_to_utf8> to get
3318 a new string which is UTF-8 encoded, and then combine them.
3322 =head1 Custom Operators
3324 Custom operator support is an experimental feature that allows you to
3325 define your own ops. This is primarily to allow the building of
3326 interpreters for other languages in the Perl core, but it also allows
3327 optimizations through the creation of "macro-ops" (ops which perform the
3328 functions of multiple ops which are usually executed together, such as
3329 C<gvsv, gvsv, add>.)
3331 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
3332 core does not "know" anything special about this op type, and so it will
3333 not be involved in any optimizations. This also means that you can
3334 define your custom ops to be any op structure -- unary, binary, list and
3337 It's important to know what custom operators won't do for you. They
3338 won't let you add new syntax to Perl, directly. They won't even let you
3339 add new keywords, directly. In fact, they won't change the way Perl
3340 compiles a program at all. You have to do those changes yourself, after
3341 Perl has compiled the program. You do this either by manipulating the op
3342 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
3343 a custom peephole optimizer with the C<optimize> module.
3345 When you do this, you replace ordinary Perl ops with custom ops by
3346 creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
3347 PP function. This should be defined in XS code, and should look like
3348 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
3349 takes the appropriate number of values from the stack, and you are
3350 responsible for adding stack marks if necessary.
3352 You should also "register" your op with the Perl interpreter so that it
3353 can produce sensible error and warning messages. Since it is possible to
3354 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
3355 Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
3356 it is dealing with. You should create an C<XOP> structure for each
3357 ppaddr you use, set the properties of the custom op with
3358 C<XopENTRY_set>, and register the structure against the ppaddr using
3359 C<Perl_custom_op_register>. A trivial example might look like:
3361 =for apidoc Ayh||XOP
3364 static OP *my_pp(pTHX);
3367 XopENTRY_set(&my_xop, xop_name, "myxop");
3368 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
3369 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
3371 The available fields in the structure are:
3377 A short name for your op. This will be included in some error messages,
3378 and will also be returned as C<< $op->name >> by the L<B|B> module, so
3379 it will appear in the output of module like L<B::Concise|B::Concise>.
3383 A short description of the function of the op.
3387 Which of the various C<*OP> structures this op uses. This should be one of
3388 the C<OA_*> constants from F<op.h>, namely
3408 =item OA_PVOP_OR_SVOP
3410 This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
3411 the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
3419 The other C<OA_*> constants should not be used.
3423 This member is of type C<Perl_cpeep_t>, which expands to C<void
3424 (*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
3425 will be called from C<Perl_rpeep> when ops of this type are encountered
3426 by the peephole optimizer. I<o> is the OP that needs optimizing;
3427 I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
3429 =for apidoc Ayh||Perl_cpeep_t
3433 C<B::Generate> directly supports the creation of custom ops by name.
3437 Descriptions above occasionally refer to "the stack", but there are in fact
3438 many stack-like data structures within the perl interpreter. When otherwise
3439 unqualified, "the stack" usually refers to the value stack.
3441 The various stacks have different purposes, and operate in slightly different
3442 ways. Their differences are noted below.
3446 This stack stores the values that regular perl code is operating on, usually
3447 intermediate values of expressions within a statement. The stack itself is
3448 formed of an array of SV pointers.
3450 The base of this stack is pointed to by the interpreter variable
3451 C<PL_stack_base>, of type C<SV **>.
3453 The head of the stack is C<PL_stack_sp>, and points to the most
3454 recently-pushed item.
3456 Items are pushed to the stack by using the C<PUSHs()> macro or its variants
3457 described above; C<XPUSHs()>, C<mPUSHs()>, C<mXPUSHs()> and the typed
3458 versions. Note carefully that the non-C<X> versions of these macros do not
3459 check the size of the stack and assume it to be big enough. These must be
3460 paired with a suitable check of the stack's size, such as the C<EXTEND> macro
3461 to ensure it is large enough. For example
3469 This is slightly more performant than making four separate checks in four
3470 separate C<mXPUSHi()> calls.
3472 As a further performance optimisation, the various C<PUSH> macros all operate
3473 using a local variable C<SP>, rather than the interpreter-global variable
3474 C<PL_stack_sp>. This variable is declared by the C<dSP> macro - though it is
3475 normally implied by XSUBs and similar so it is rare you have to consider it
3476 directly. Once declared, the C<PUSH> macros will operate only on this local
3477 variable, so before invoking any other perl core functions you must use the
3478 C<PUTBACK> macro to return the value from the local C<SP> variable back to
3479 the interpreter variable. Similarly, after calling a perl core function which
3480 may have had reason to move the stack or push/pop values to it, you must use
3481 the C<SPAGAIN> macro which refreshes the local C<SP> value back from the
3484 Items are popped from the stack by using the C<POPs> macro or its typed
3485 versions, There is also a macro C<TOPs> that inspects the topmost item without
3488 Note specifically that SV pointers on the value stack do not contribute to the
3489 overall reference count of the xVs being referred to. If newly-created xVs are
3490 being pushed to the stack you must arrange for them to be destroyed at a
3491 suitable time; usually by using one of the C<mPUSH*> macros or C<sv_2mortal()>
3492 to mortalise the xV.
3496 The value stack stores individual perl scalar values as temporaries between
3497 expressions. Some perl expressions operate on entire lists; for that purpose
3498 we need to know where on the stack each list begins. This is the purpose of the
3501 The mark stack stores integers as I32 values, which are the height of the
3502 value stack at the time before the list began; thus the mark itself actually
3503 points to the value stack entry one before the list. The list itself starts at
3506 The base of this stack is pointed to by the interpreter variable
3507 C<PL_markstack>, of type C<I32 *>.
3509 The head of the stack is C<PL_markstack_ptr>, and points to the most
3510 recently-pushed item.
3512 Items are pushed to the stack by using the C<PUSHMARK()> macro. Even though
3513 the stack itself stores (value) stack indices as integers, the C<PUSHMARK>
3514 macro should be given a stack pointer directly; it will calculate the index
3515 offset by comparing to the C<PL_stack_sp> variable. Thus almost always the
3516 code to perform this is
3520 Items are popped from the stack by the C<POPMARK> macro. There is also a macro
3521 C<TOPMARK> that inspects the topmost item without removing it. These macros
3522 return I32 index values directly. There is also the C<dMARK> macro which
3523 declares a new SV double-pointer variable, called C<mark>, which points at the
3524 marked stack slot; this is the usual macro that C code will use when operating
3525 on lists given on the stack.
3527 As noted above, the C<mark> variable itself will point at the most recently
3528 pushed value on the value stack before the list begins, and so the list itself
3529 starts at C<mark + 1>. The values of the list may be iterated by code such as
3531 for(SV **svp = mark + 1; svp <= PL_stack_sp; svp++) {
3536 Note specifically in the case that the list is already empty, C<mark> will
3537 equal C<PL_stack_sp>.
3539 Because the C<mark> variable is converted to a pointer on the value stack,
3540 extra care must be taken if C<EXTEND> or any of the C<XPUSH> macros are
3541 invoked within the function, because the stack may need to be moved to
3542 extend it and so the existing pointer will now be invalid. If this may be a
3543 problem, a possible solution is to track the mark offset as an integer and
3544 track the mark itself later on after the stack had been moved.
3546 I32 markoff = POPMARK;
3550 SP **mark = PL_stack_base + markoff;
3552 =head2 Temporaries Stack
3554 As noted above, xV references on the main value stack do not contribute to the
3555 reference count of an xV, and so another mechanism is used to track when
3556 temporary values which live on the stack must be released. This is the job of
3557 the temporaries stack.
3559 The temporaries stack stores pointers to xVs whose reference counts will be
3562 The base of this stack is pointed to by the interpreter variable
3563 C<PL_tmps_stack>, of type C<SV **>.
3565 The head of the stack is indexed by C<PL_tmps_ix>, an integer which stores the
3566 index in the array of the most recently-pushed item.
3568 There is no public API to directly push items to the temporaries stack. Instead,
3569 the API function C<sv_2mortal()> is used to mortalize an xV, adding its
3570 address to the temporaries stack.
3572 Likewise, there is no public API to read values from the temporaries stack.
3573 Instead. the macros C<SAVETMPS> and C<FREETPMS> are used. The C<SAVETMPS>
3574 macro establishes the base levels of the temporaries stack, by capturing the
3575 current value of C<PL_tmps_ix> into C<PL_tmps_floor> and saving the previous
3576 value to the save stack. Thereafter, whenever C<FREETMPS> is invoked all of
3577 the temporaries that have been pushed since that level are reclaimed.
3579 While it is common to see these two macros in pairs within an C<ENTER>/
3580 C<LEAVE> pair, it is not necessary to match them. It is permitted to invoke
3581 C<FREETMPS> multiple times since the most recent C<SAVETMPS>; for example in a
3582 loop iterating over elements of a list. While you can invoke C<SAVETMPS>
3583 multiple times within a scope pair, it is unlikely to be useful. Subsequent
3584 invocations will move the temporaries floor further up, thus effectively
3585 trapping the existing temporaries to only be released at the end of the scope.
3589 The save stack is used by perl to implement the C<local> keyword and other
3590 similar behaviours; any cleanup operations that need to be performed when
3591 leaving the current scope. Items pushed to this stack generally capture the
3592 current value of some internal variable or state, which will be restored when
3593 the scope is unwound due to leaving, C<return>, C<die>, C<goto> or other
3596 Whereas other perl internal stacks store individual items all of the same type
3597 (usually SV pointers or integers), the items pushed to the save stack are
3598 formed of many different types, having multiple fields to them. For example,
3599 the C<SAVEt_INT> type needs to store both the address of the C<int> variable
3600 to restore, and the value to restore it to. This information could have been
3601 stored using fields of a C<struct>, but would have to be large enough to store
3602 three pointers in the largest case, which would waste a lot of space in most
3603 of the smaller cases.
3605 Instead, the stack stores information in a variable-length encoding of C<ANY>
3606 structures. The final value pushed is stored in the C<UV> field which encodes
3607 the kind of item held by the preceeding items; the count and types of which
3608 will depend on what kind of item is being stored. The kind field is pushed
3609 last because that will be the first field to be popped when unwinding items
3612 The base of this stack is pointed to by the interpreter variable
3613 C<PL_savestack>, of type C<ANY *>.
3615 The head of the stack is indexed by C<PL_savestack_ix>, an integer which
3616 stores the index in the array at which the next item should be pushed. (Note
3617 that this is different to most other stacks, which reference the most
3618 recently-pushed item).
3620 Items are pushed to the save stack by using the various C<SAVE...()> macros.
3621 Many of these macros take a variable and store both its address and current
3622 value on the save stack, ensuring that value gets restored on scope exit.
3630 There are also a variety of other special-purpose macros which save particular
3631 types or values of interest. C<SAVETMPS> has already been mentioned above.
3632 Others include C<SAVEFREEPV> which arranges for a PV (i.e. a string buffer) to
3633 be freed, or C<SAVEDESTRUCTOR> which arranges for a given function pointer to
3634 be invoked on scope exit. A full list of such macros can be found in
3637 There is no public API for popping individual values or items from the save
3638 stack. Instead, via the scope stack, the C<ENTER> and C<LEAVE> pair form a way
3639 to start and stop nested scopes. Leaving a nested scope via C<LEAVE> will
3640 restore all of the saved values that had been pushed since the most recent
3645 As with the mark stack to the value stack, the scope stack forms a pair with
3646 the save stack. The scope stack stores the height of the save stack at which
3647 nested scopes begin, and allows the save stack to be unwound back to that
3648 point when the scope is left.
3650 When perl is built with debugging enabled, there is a second part to this
3651 stack storing human-readable string names describing the type of stack
3652 context. Each push operation saves the name as well as the height of the save
3653 stack, and each pop operation checks the topmost name with what is expected,
3654 causing an assertion failure if the name does not match.
3656 The base of this stack is pointed to by the interpreter variable
3657 C<PL_scopestack>, of type C<I32 *>. If enabled, the scope stack names are
3658 stored in a separate array pointed to by C<PL_scopestack_name>, of type
3661 The head of the stack is indexed by C<PL_scopestack_ix>, an integer which
3662 stores the index of the array or arrays at which the next item should be
3663 pushed. (Note that this is different to most other stacks, which reference the
3664 most recently-pushed item).
3666 Values are pushed to the scope stack using the C<ENTER> macro, which begins a
3667 new nested scope. Any items pushed to the save stack are then restored at the
3668 next nested invocation of the C<LEAVE> macro.
3670 =head1 Dynamic Scope and the Context Stack
3672 B<Note:> this section describes a non-public internal API that is subject
3673 to change without notice.
3675 =head2 Introduction to the context stack
3677 In Perl, dynamic scoping refers to the runtime nesting of things like
3678 subroutine calls, evals etc, as well as the entering and exiting of block
3679 scopes. For example, the restoring of a C<local>ised variable is
3680 determined by the dynamic scope.
3682 Perl tracks the dynamic scope by a data structure called the context
3683 stack, which is an array of C<PERL_CONTEXT> structures, and which is
3684 itself a big union for all the types of context. Whenever a new scope is
3685 entered (such as a block, a C<for> loop, or a subroutine call), a new
3686 context entry is pushed onto the stack. Similarly when leaving a block or
3687 returning from a subroutine call etc. a context is popped. Since the
3688 context stack represents the current dynamic scope, it can be searched.
3689 For example, C<next LABEL> searches back through the stack looking for a
3690 loop context that matches the label; C<return> pops contexts until it
3691 finds a sub or eval context or similar; C<caller> examines sub contexts on
3694 Each context entry is labelled with a context type, C<cx_type>. Typical
3695 context types are C<CXt_SUB>, C<CXt_EVAL> etc., as well as C<CXt_BLOCK>
3696 and C<CXt_NULL> which represent a basic scope (as pushed by C<pp_enter>)
3697 and a sort block. The type determines which part of the context union are
3700 The main division in the context struct is between a substitution scope
3701 (C<CXt_SUBST>) and block scopes, which are everything else. The former is
3702 just used while executing C<s///e>, and won't be discussed further
3705 All the block scope types share a common base, which corresponds to
3706 C<CXt_BLOCK>. This stores the old values of various scope-related
3707 variables like C<PL_curpm>, as well as information about the current
3708 scope, such as C<gimme>. On scope exit, the old variables are restored.
3710 Particular block scope types store extra per-type information. For
3711 example, C<CXt_SUB> stores the currently executing CV, while the various
3712 for loop types might hold the original loop variable SV. On scope exit,
3713 the per-type data is processed; for example the CV has its reference count
3714 decremented, and the original loop variable is restored.
3716 The macro C<cxstack> returns the base of the current context stack, while
3717 C<cxstack_ix> is the index of the current frame within that stack.
3719 In fact, the context stack is actually part of a stack-of-stacks system;
3720 whenever something unusual is done such as calling a C<DESTROY> or tie
3721 handler, a new stack is pushed, then popped at the end.
3723 Note that the API described here changed considerably in perl 5.24; prior
3724 to that, big macros like C<PUSHBLOCK> and C<POPSUB> were used; in 5.24
3725 they were replaced by the inline static functions described below. In
3726 addition, the ordering and detail of how these macros/function work
3727 changed in many ways, often subtly. In particular they didn't handle
3728 saving the savestack and temps stack positions, and required additional
3729 C<ENTER>, C<SAVETMPS> and C<LEAVE> compared to the new functions. The
3730 old-style macros will not be described further.
3733 =head2 Pushing contexts
3735 For pushing a new context, the two basic functions are
3736 C<cx = cx_pushblock()>, which pushes a new basic context block and returns
3737 its address, and a family of similar functions with names like
3738 C<cx_pushsub(cx)> which populate the additional type-dependent fields in
3739 the C<cx> struct. Note that C<CXt_NULL> and C<CXt_BLOCK> don't have their
3740 own push functions, as they don't store any data beyond that pushed by
3743 The fields of the context struct and the arguments to the C<cx_*>
3744 functions are subject to change between perl releases, representing
3745 whatever is convenient or efficient for that release.
3747 A typical context stack pushing can be found in C<pp_entersub>; the
3748 following shows a simplified and stripped-down example of a non-XS call,
3749 along with comments showing roughly what each function does.
3753 bool hasargs = cBOOL(PL_op->op_flags & OPf_STACKED);
3754 OP *retop = PL_op->op_next;
3755 I32 old_ss_ix = PL_savestack_ix;
3758 /* ... make mortal copies of stack args which are PADTMPs here ... */
3760 /* ... do any additional savestack pushes here ... */
3762 /* Now push a new context entry of type 'CXt_SUB'; initially just
3763 * doing the actions common to all block types: */
3765 cx = cx_pushblock(CXt_SUB, gimme, MARK, old_ss_ix);
3767 /* this does (approximately):
3768 CXINC; /* cxstack_ix++ (grow if necessary) */
3769 cx = CX_CUR(); /* and get the address of new frame */
3770 cx->cx_type = CXt_SUB;
3771 cx->blk_gimme = gimme;
3772 cx->blk_oldsp = MARK - PL_stack_base;
3773 cx->blk_oldsaveix = old_ss_ix;
3774 cx->blk_oldcop = PL_curcop;
3775 cx->blk_oldmarksp = PL_markstack_ptr - PL_markstack;
3776 cx->blk_oldscopesp = PL_scopestack_ix;
3777 cx->blk_oldpm = PL_curpm;
3778 cx->blk_old_tmpsfloor = PL_tmps_floor;
3780 PL_tmps_floor = PL_tmps_ix;
3784 /* then update the new context frame with subroutine-specific info,
3785 * such as the CV about to be executed: */
3787 cx_pushsub(cx, cv, retop, hasargs);
3789 /* this does (approximately):
3790 cx->blk_sub.cv = cv;
3791 cx->blk_sub.olddepth = CvDEPTH(cv);
3792 cx->blk_sub.prevcomppad = PL_comppad;
3793 cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
3794 cx->blk_sub.retop = retop;
3795 SvREFCNT_inc_simple_void_NN(cv);
3798 Note that C<cx_pushblock()> sets two new floors: for the args stack (to
3799 C<MARK>) and the temps stack (to C<PL_tmps_ix>). While executing at this
3800 scope level, every C<nextstate> (amongst others) will reset the args and
3801 tmps stack levels to these floors. Note that since C<cx_pushblock> uses
3802 the current value of C<PL_tmps_ix> rather than it being passed as an arg,
3803 this dictates at what point C<cx_pushblock> should be called. In
3804 particular, any new mortals which should be freed only on scope exit
3805 (rather than at the next C<nextstate>) should be created first.
3807 Most callers of C<cx_pushblock> simply set the new args stack floor to the
3808 top of the previous stack frame, but for C<CXt_LOOP_LIST> it stores the
3809 items being iterated over on the stack, and so sets C<blk_oldsp> to the
3810 top of these items instead. Note that, contrary to its name, C<blk_oldsp>
3811 doesn't always represent the value to restore C<PL_stack_sp> to on scope
3814 Note the early capture of C<PL_savestack_ix> to C<old_ss_ix>, which is
3815 later passed as an arg to C<cx_pushblock>. In the case of C<pp_entersub>,
3816 this is because, although most values needing saving are stored in fields
3817 of the context struct, an extra value needs saving only when the debugger
3818 is running, and it doesn't make sense to bloat the struct for this rare
3819 case. So instead it is saved on the savestack. Since this value gets
3820 calculated and saved before the context is pushed, it is necessary to pass
3821 the old value of C<PL_savestack_ix> to C<cx_pushblock>, to ensure that the
3822 saved value gets freed during scope exit. For most users of
3823 C<cx_pushblock>, where nothing needs pushing on the save stack,
3824 C<PL_savestack_ix> is just passed directly as an arg to C<cx_pushblock>.
3826 Note that where possible, values should be saved in the context struct
3827 rather than on the save stack; it's much faster that way.
3829 Normally C<cx_pushblock> should be immediately followed by the appropriate
3830 C<cx_pushfoo>, with nothing between them; this is because if code
3831 in-between could die (e.g. a warning upgraded to fatal), then the context
3832 stack unwinding code in C<dounwind> would see (in the example above) a
3833 C<CXt_SUB> context frame, but without all the subroutine-specific fields
3834 set, and crashes would soon ensue.
3836 Where the two must be separate, initially set the type to C<CXt_NULL> or
3837 C<CXt_BLOCK>, and later change it to C<CXt_foo> when doing the
3838 C<cx_pushfoo>. This is exactly what C<pp_enteriter> does, once it's
3839 determined which type of loop it's pushing.
3841 =head2 Popping contexts
3843 Contexts are popped using C<cx_popsub()> etc. and C<cx_popblock()>. Note
3844 however, that unlike C<cx_pushblock>, neither of these functions actually
3845 decrement the current context stack index; this is done separately using
3848 There are two main ways that contexts are popped. During normal execution
3849 as scopes are exited, functions like C<pp_leave>, C<pp_leaveloop> and
3850 C<pp_leavesub> process and pop just one context using C<cx_popfoo> and
3851 C<cx_popblock>. On the other hand, things like C<pp_return> and C<next>
3852 may have to pop back several scopes until a sub or loop context is found,
3853 and exceptions (such as C<die>) need to pop back contexts until an eval
3854 context is found. Both of these are accomplished by C<dounwind()>, which
3855 is capable of processing and popping all contexts above the target one.
3857 Here is a typical example of context popping, as found in C<pp_leavesub>
3858 (simplified slightly):
3867 gimme = cx->blk_gimme;
3868 oldsp = PL_stack_base + cx->blk_oldsp; /* last arg of previous frame */
3870 if (gimme == G_VOID)
3871 PL_stack_sp = oldsp;
3873 leave_adjust_stacks(oldsp, oldsp, gimme, 0);
3878 retop = cx->blk_sub.retop;
3883 The steps above are in a very specific order, designed to be the reverse
3884 order of when the context was pushed. The first thing to do is to copy
3885 and/or protect any return arguments and free any temps in the current
3886 scope. Scope exits like an rvalue sub normally return a mortal copy of
3887 their return args (as opposed to lvalue subs). It is important to make
3888 this copy before the save stack is popped or variables are restored, or
3889 bad things like the following can happen:
3891 sub f { my $x =...; $x } # $x freed before we get to copy it
3892 sub f { /(...)/; $1 } # PL_curpm restored before $1 copied
3894 Although we wish to free any temps at the same time, we have to be careful
3895 not to free any temps which are keeping return args alive; nor to free the
3896 temps we have just created while mortal copying return args. Fortunately,
3897 C<leave_adjust_stacks()> is capable of making mortal copies of return args,
3898 shifting args down the stack, and only processing those entries on the
3899 temps stack that are safe to do so.
3901 In void context no args are returned, so it's more efficient to skip
3902 calling C<leave_adjust_stacks()>. Also in void context, a C<nextstate> op
3903 is likely to be imminently called which will do a C<FREETMPS>, so there's
3904 no need to do that either.
3906 The next step is to pop savestack entries: C<CX_LEAVE_SCOPE(cx)> is just
3907 defined as C<< LEAVE_SCOPE(cx->blk_oldsaveix) >>. Note that during the
3908 popping, it's possible for perl to call destructors, call C<STORE> to undo
3909 localisations of tied vars, and so on. Any of these can die or call
3910 C<exit()>. In this case, C<dounwind()> will be called, and the current
3911 context stack frame will be re-processed. Thus it is vital that all steps
3912 in popping a context are done in such a way to support reentrancy. The
3913 other alternative, of decrementing C<cxstack_ix> I<before> processing the
3914 frame, would lead to leaks and the like if something died halfway through,
3915 or overwriting of the current frame.
3917 C<CX_LEAVE_SCOPE> itself is safely re-entrant: if only half the savestack
3918 items have been popped before dying and getting trapped by eval, then the
3919 C<CX_LEAVE_SCOPE>s in C<dounwind> or C<pp_leaveeval> will continue where
3920 the first one left off.
3922 The next step is the type-specific context processing; in this case
3923 C<cx_popsub>. In part, this looks like:
3925 cv = cx->blk_sub.cv;
3926 CvDEPTH(cv) = cx->blk_sub.olddepth;
3927 cx->blk_sub.cv = NULL;
3930 where its processing the just-executed CV. Note that before it decrements
3931 the CV's reference count, it nulls the C<blk_sub.cv>. This means that if
3932 it re-enters, the CV won't be freed twice. It also means that you can't
3933 rely on such type-specific fields having useful values after the return
3936 Next, C<cx_popblock> restores all the various interpreter vars to their
3937 previous values or previous high water marks; it expands to:
3939 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
3940 PL_scopestack_ix = cx->blk_oldscopesp;
3941 PL_curpm = cx->blk_oldpm;
3942 PL_curcop = cx->blk_oldcop;
3943 PL_tmps_floor = cx->blk_old_tmpsfloor;
3945 Note that it I<doesn't> restore C<PL_stack_sp>; as mentioned earlier,
3946 which value to restore it to depends on the context type (specifically
3947 C<for (list) {}>), and what args (if any) it returns; and that will
3948 already have been sorted out earlier by C<leave_adjust_stacks()>.
3950 Finally, the context stack pointer is actually decremented by C<CX_POP(cx)>.
3951 After this point, it's possible that that the current context frame could
3952 be overwritten by other contexts being pushed. Although things like ties
3953 and C<DESTROY> are supposed to work within a new context stack, it's best
3954 not to assume this. Indeed on debugging builds, C<CX_POP(cx)> deliberately
3955 sets C<cx> to null to detect code that is still relying on the field
3956 values in that context frame. Note in the C<pp_leavesub()> example above,
3957 we grab C<blk_sub.retop> I<before> calling C<CX_POP>.
3959 =head2 Redoing contexts
3961 Finally, there is C<cx_topblock(cx)>, which acts like a super-C<nextstate>
3962 as regards to resetting various vars to their base values. It is used in
3963 places like C<pp_next>, C<pp_redo> and C<pp_goto> where rather than
3964 exiting a scope, we want to re-initialise the scope. As well as resetting
3965 C<PL_stack_sp> like C<nextstate>, it also resets C<PL_markstack_ptr>,
3966 C<PL_scopestack_ix> and C<PL_curpm>. Note that it doesn't do a
3970 =head1 Slab-based operator allocation
3972 B<Note:> this section describes a non-public internal API that is subject
3973 to change without notice.
3975 Perl's internal error-handling mechanisms implement C<die> (and its internal
3976 equivalents) using longjmp. If this occurs during lexing, parsing or
3977 compilation, we must ensure that any ops allocated as part of the compilation
3978 process are freed. (Older Perl versions did not adequately handle this
3979 situation: when failing a parse, they would leak ops that were stored in
3980 C C<auto> variables and not linked anywhere else.)
3982 To handle this situation, Perl uses I<op slabs> that are attached to the
3983 currently-compiling CV. A slab is a chunk of allocated memory. New ops are
3984 allocated as regions of the slab. If the slab fills up, a new one is created
3985 (and linked from the previous one). When an error occurs and the CV is freed,
3986 any ops remaining are freed.
3988 Each op is preceded by two pointers: one points to the next op in the slab, and
3989 the other points to the slab that owns it. The next-op pointer is needed so
3990 that Perl can iterate over a slab and free all its ops. (Op structures are of
3991 different sizes, so the slab's ops can't merely be treated as a dense array.)
3992 The slab pointer is needed for accessing a reference count on the slab: when
3993 the last op on a slab is freed, the slab itself is freed.
3995 The slab allocator puts the ops at the end of the slab first. This will tend to
3996 allocate the leaves of the op tree first, and the layout will therefore
3997 hopefully be cache-friendly. In addition, this means that there's no need to
3998 store the size of the slab (see below on why slabs vary in size), because Perl
3999 can follow pointers to find the last op.
4001 It might seem possible to eliminate slab reference counts altogether, by having
4002 all ops implicitly attached to C<PL_compcv> when allocated and freed when the
4003 CV is freed. That would also allow C<op_free> to skip C<FreeOp> altogether, and
4004 thus free ops faster. But that doesn't work in those cases where ops need to
4005 survive beyond their CVs, such as re-evals.
4007 The CV also has to have a reference count on the slab. Sometimes the first op
4008 created is immediately freed. If the reference count of the slab reaches 0,
4009 then it will be freed with the CV still pointing to it.
4011 CVs use the C<CVf_SLABBED> flag to indicate that the CV has a reference count
4012 on the slab. When this flag is set, the slab is accessible via C<CvSTART> when
4013 C<CvROOT> is not set, or by subtracting two pointers C<(2*sizeof(I32 *))> from
4014 C<CvROOT> when it is set. The alternative to this approach of sneaking the slab
4015 into C<CvSTART> during compilation would be to enlarge the C<xpvcv> struct by
4016 another pointer. But that would make all CVs larger, even though slab-based op
4017 freeing is typically of benefit only for programs that make significant use of
4020 When the C<CVf_SLABBED> flag is set, the CV takes responsibility for freeing
4021 the slab. If C<CvROOT> is not set when the CV is freed or undeffed, it is
4022 assumed that a compilation error has occurred, so the op slab is traversed and
4023 all the ops are freed.
4025 Under normal circumstances, the CV forgets about its slab (decrementing the
4026 reference count) when the root is attached. So the slab reference counting that
4027 happens when ops are freed takes care of freeing the slab. In some cases, the
4028 CV is told to forget about the slab (C<cv_forget_slab>) precisely so that the
4029 ops can survive after the CV is done away with.
4031 Forgetting the slab when the root is attached is not strictly necessary, but
4032 avoids potential problems with C<CvROOT> being written over. There is code all
4033 over the place, both in core and on CPAN, that does things with C<CvROOT>, so
4034 forgetting the slab makes things more robust and avoids potential problems.
4036 Since the CV takes ownership of its slab when flagged, that flag is never
4037 copied when a CV is cloned, as one CV could free a slab that another CV still
4038 points to, since forced freeing of ops ignores the reference count (but asserts
4039 that it looks right).
4041 To avoid slab fragmentation, freed ops are marked as freed and attached to the
4042 slab's freed chain (an idea stolen from DBM::Deep). Those freed ops are reused
4043 when possible. Not reusing freed ops would be simpler, but it would result in
4044 significantly higher memory usage for programs with large C<if (DEBUG) {...}>
4047 C<SAVEFREEOP> is slightly problematic under this scheme. Sometimes it can cause
4048 an op to be freed after its CV. If the CV has forcibly freed the ops on its
4049 slab and the slab itself, then we will be fiddling with a freed slab. Making
4050 C<SAVEFREEOP> a no-op doesn't help, as sometimes an op can be savefreed when
4051 there is no compilation error, so the op would never be freed. It holds
4052 a reference count on the slab, so the whole slab would leak. So C<SAVEFREEOP>
4053 now sets a special flag on the op (C<< ->op_savefree >>). The forced freeing of
4054 ops after a compilation error won't free any ops thus marked.
4056 Since many pieces of code create tiny subroutines consisting of only a few ops,
4057 and since a huge slab would be quite a bit of baggage for those to carry
4058 around, the first slab is always very small. To avoid allocating too many
4059 slabs for a single CV, each subsequent slab is twice the size of the previous.
4061 Smartmatch expects to be able to allocate an op at run time, run it, and then
4062 throw it away. For that to work the op is simply malloced when PL_compcv hasn't
4063 been set up. So all slab-allocated ops are marked as such (C<< ->op_slabbed >>),
4064 to distinguish them from malloced ops.
4069 Until May 1997, this document was maintained by Jeff Okamoto
4070 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
4071 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
4073 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
4074 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
4075 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
4076 Stephen McCamant, and Gurusamy Sarathy.
4080 L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>