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, Perl's API exposes
157 several macros that coerce the actual scalar type into an IV, UV, double,
162 =item * C<SvIV(SV*)> (C<IV>) and C<SvUV(SV*)> (C<UV>)
164 =item * C<SvNV(SV*)> (C<double>)
166 =item * Strings are a bit complicated:
170 =item * Byte string: C<SvPVbyte(SV*, STRLEN len)> or C<SvPVbyte_nolen(SV*)>
172 If the Perl string is C<"\xff\xff">, then this returns a 2-byte C<char*>.
174 This is suitable for Perl strings that represent bytes.
176 =item * UTF-8 string: C<SvPVutf8(SV*, STRLEN len)> or C<SvPVutf8_nolen(SV*)>
178 If the Perl string is C<"\xff\xff">, then this returns a 4-byte C<char*>.
180 This is suitable for Perl strings that represent characters.
182 B<CAVEAT>: That C<char*> will be encoded via Perl's internal UTF-8 variant,
183 which means that if the SV contains non-Unicode code points (e.g.,
184 0x110000), then the result may contain extensions over valid UTF-8.
185 See L<perlapi/is_strict_utf8_string> for some methods Perl gives
186 you to check the UTF-8 validity of these macros' returns.
188 =item * You can also use C<SvPV(SV*, STRLEN len)> or C<SvPV_nolen(SV*)>
189 to fetch the SV's raw internal buffer. This is tricky, though; if your Perl
191 is C<"\xff\xff">, then depending on the SV's internal encoding you might get
192 back a 2-byte B<OR> a 4-byte C<char*>.
193 Moreover, if it's the 4-byte string, that could come from either Perl
194 C<"\xff\xff"> stored UTF-8 encoded, or Perl C<"\xc3\xbf\xc3\xbf"> stored
195 as raw octets. To differentiate between these you B<MUST> look up the
196 SV's UTF8 bit (cf. C<SvUTF8>) to know whether the source Perl string
197 is 2 characters (C<SvUTF8> would be on) or 4 characters (C<SvUTF8> would be
200 B<IMPORTANT:> Use of C<SvPV>, C<SvPV_nolen>, or
201 similarly-named macros I<without> looking up the SV's UTF8 bit is
202 almost certainly a bug if non-ASCII input is allowed.
204 When the UTF8 bit is on, the same B<CAVEAT> about UTF-8 validity applies
205 here as for C<SvPVutf8>.
209 (See L</How do I pass a Perl string to a C library?> for more details.)
211 In C<SvPVbyte>, C<SvPVutf8>, and C<SvPV>, the length of the C<char*> returned
213 variable C<len> (these are macros, so you do I<not> use C<&len>). If you do
214 not care what the length of the data is, use C<SvPVbyte_nolen>,
215 C<SvPVutf8_nolen>, or C<SvPV_nolen> instead.
216 The global variable C<PL_na> can also be given to
217 C<SvPVbyte>/C<SvPVutf8>/C<SvPV>
218 in this case. But that can be quite inefficient because C<PL_na> must
219 be accessed in thread-local storage in threaded Perl. In any case, remember
220 that Perl allows arbitrary strings of data that may both contain NULs and
221 might not be terminated by a C<NUL>.
223 Also remember that C doesn't allow you to safely say C<foo(SvPVbyte(s, len),
224 len);>. It might work with your
225 compiler, but it won't work for everyone.
226 Break this sort of statement up into separate assignments:
231 ptr = SvPVbyte(s, len);
236 If you want to know if the scalar value is TRUE, you can use:
240 Although Perl will automatically grow strings for you, if you need to force
241 Perl to allocate more memory for your SV, you can use the macro
243 SvGROW(SV*, STRLEN newlen)
245 which will determine if more memory needs to be allocated. If so, it will
246 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
247 decrease, the allocated memory of an SV and that it does not automatically
248 add space for the trailing C<NUL> byte (perl's own string functions typically do
249 C<SvGROW(sv, len + 1)>).
251 If you want to write to an existing SV's buffer and set its value to a
252 string, use SvPVbyte_force() or one of its variants to force the SV to be
253 a PV. This will remove any of various types of non-stringness from
254 the SV while preserving the content of the SV in the PV. This can be
255 used, for example, to append data from an API function to a buffer
256 without extra copying:
258 (void)SvPVbyte_force(sv, len);
259 s = SvGROW(sv, len + needlen + 1);
260 /* something that modifies up to needlen bytes at s+len, but
261 modifies newlen bytes
262 eg. newlen = read(fd, s + len, needlen);
263 ignoring errors for these examples
265 s[len + newlen] = '\0';
266 SvCUR_set(sv, len + newlen);
270 If you already have the data in memory or if you want to keep your
271 code simple, you can use one of the sv_cat*() variants, such as
272 sv_catpvn(). If you want to insert anywhere in the string you can use
273 sv_insert() or sv_insert_flags().
275 If you don't need the existing content of the SV, you can avoid some
279 s = SvGROW(sv, needlen + 1);
280 /* something that modifies up to needlen bytes at s, but modifies
282 eg. newlen = read(fd, s, needlen);
285 SvCUR_set(sv, newlen);
286 SvPOK_only(sv); /* also clears SVf_UTF8 */
289 Again, if you already have the data in memory or want to avoid the
290 complexity of the above, you can use sv_setpvn().
292 If you have a buffer allocated with Newx() and want to set that as the
293 SV's value, you can use sv_usepvn_flags(). That has some requirements
294 if you want to avoid perl re-allocating the buffer to fit the trailing
297 Newx(buf, somesize+1, char);
298 /* ... fill in buf ... */
299 buf[somesize] = '\0';
300 sv_usepvn_flags(sv, buf, somesize, SV_SMAGIC | SV_HAS_TRAILING_NUL);
301 /* buf now belongs to perl, don't release it */
303 If you have an SV and want to know what kind of data Perl thinks is stored
304 in it, you can use the following macros to check the type of SV you have.
310 You can get and set the current length of the string stored in an SV with
311 the following macros:
314 SvCUR_set(SV*, I32 val)
316 You can also get a pointer to the end of the string stored in the SV
321 But note that these last three macros are valid only if C<SvPOK()> is true.
323 If you want to append something to the end of string stored in an C<SV*>,
324 you can use the following functions:
326 void sv_catpv(SV*, const char*);
327 void sv_catpvn(SV*, const char*, STRLEN);
328 void sv_catpvf(SV*, const char*, ...);
329 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
331 void sv_catsv(SV*, SV*);
333 The first function calculates the length of the string to be appended by
334 using C<strlen>. In the second, you specify the length of the string
335 yourself. The third function processes its arguments like C<sprintf> and
336 appends the formatted output. The fourth function works like C<vsprintf>.
337 You can specify the address and length of an array of SVs instead of the
338 va_list argument. The fifth function
339 extends the string stored in the first
340 SV with the string stored in the second SV. It also forces the second SV
341 to be interpreted as a string.
343 The C<sv_cat*()> functions are not generic enough to operate on values that
344 have "magic". See L</Magic Virtual Tables> later in this document.
346 If you know the name of a scalar variable, you can get a pointer to its SV
347 by using the following:
349 SV* get_sv("package::varname", 0);
351 This returns NULL if the variable does not exist.
353 If you want to know if this variable (or any other SV) is actually C<defined>,
358 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
360 Its address can be used whenever an C<SV*> is needed. Make sure that
361 you don't try to compare a random sv with C<&PL_sv_undef>. For example
362 when interfacing Perl code, it'll work correctly for:
366 But won't work when called as:
371 So to repeat always use SvOK() to check whether an sv is defined.
373 Also you have to be careful when using C<&PL_sv_undef> as a value in
374 AVs or HVs (see L</AVs, HVs and undefined values>).
376 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
377 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
378 addresses can be used whenever an C<SV*> is needed.
380 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
384 if (I-am-to-return-a-real-value) {
385 sv = sv_2mortal(newSViv(42));
389 This code tries to return a new SV (which contains the value 42) if it should
390 return a real value, or undef otherwise. Instead it has returned a NULL
391 pointer which, somewhere down the line, will cause a segmentation violation,
392 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
393 first line and all will be well.
395 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
396 call is not necessary (see L</Reference Counts and Mortality>).
400 Perl provides the function C<sv_chop> to efficiently remove characters
401 from the beginning of a string; you give it an SV and a pointer to
402 somewhere inside the PV, and it discards everything before the
403 pointer. The efficiency comes by means of a little hack: instead of
404 actually removing the characters, C<sv_chop> sets the flag C<OOK>
405 (offset OK) to signal to other functions that the offset hack is in
406 effect, and it moves the PV pointer (called C<SvPVX>) forward
407 by the number of bytes chopped off, and adjusts C<SvCUR> and C<SvLEN>
408 accordingly. (A portion of the space between the old and new PV
409 pointers is used to store the count of chopped bytes.)
411 Hence, at this point, the start of the buffer that we allocated lives
412 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
413 into the middle of this allocated storage.
415 This is best demonstrated by example. Normally copy-on-write will prevent
416 the substitution from operator from using this hack, but if you can craft a
417 string for which copy-on-write is not possible, you can see it in play. In
418 the current implementation, the final byte of a string buffer is used as a
419 copy-on-write reference count. If the buffer is not big enough, then
420 copy-on-write is skipped. First have a look at an empty string:
422 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a .= ""; Dump $a'
423 SV = PV(0x7ffb7c008a70) at 0x7ffb7c030390
426 PV = 0x7ffb7bc05b50 ""\0
430 Notice here the LEN is 10. (It may differ on your platform.) Extend the
431 length of the string to one less than 10, and do a substitution:
433 % ./perl -Ilib -MDevel::Peek -le '$a=""; $a.="123456789"; $a=~s/.//; \
435 SV = PV(0x7ffa04008a70) at 0x7ffa04030390
437 FLAGS = (POK,OOK,pPOK)
439 PV = 0x7ffa03c05b61 ( "\1" . ) "23456789"\0
443 Here the number of bytes chopped off (1) is shown next as the OFFSET. The
444 portion of the string between the "real" and the "fake" beginnings is
445 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
446 the fake beginning, not the real one. (The first character of the string
447 buffer happens to have changed to "\1" here, not "1", because the current
448 implementation stores the offset count in the string buffer. This is
451 Something similar to the offset hack is performed on AVs to enable
452 efficient shifting and splicing off the beginning of the array; while
453 C<AvARRAY> points to the first element in the array that is visible from
454 Perl, C<AvALLOC> points to the real start of the C array. These are
455 usually the same, but a C<shift> operation can be carried out by
456 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
457 Again, the location of the real start of the C array only comes into
458 play when freeing the array. See C<av_shift> in F<av.c>.
460 =head2 What's Really Stored in an SV?
462 Recall that the usual method of determining the type of scalar you have is
463 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
464 usually these macros will always return TRUE and calling the C<Sv*V>
465 macros will do the appropriate conversion of string to integer/double or
466 integer/double to string.
468 If you I<really> need to know if you have an integer, double, or string
469 pointer in an SV, you can use the following three macros instead:
475 These will tell you if you truly have an integer, double, or string pointer
476 stored in your SV. The "p" stands for private.
478 There are various ways in which the private and public flags may differ.
479 For example, in perl 5.16 and earlier a tied SV may have a valid
480 underlying value in the IV slot (so SvIOKp is true), but the data
481 should be accessed via the FETCH routine rather than directly,
482 so SvIOK is false. (In perl 5.18 onwards, tied scalars use
483 the flags the same way as untied scalars.) Another is when
484 numeric conversion has occurred and precision has been lost: only the
485 private flag is set on 'lossy' values. So when an NV is converted to an
486 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
488 In general, though, it's best to use the C<Sv*V> macros.
490 =head2 Working with AVs
492 There are two ways to create and load an AV. The first method creates an
497 The second method both creates the AV and initially populates it with SVs:
499 AV* av_make(SSize_t num, SV **ptr);
501 The second argument points to an array containing C<num> C<SV*>'s. Once the
502 AV has been created, the SVs can be destroyed, if so desired.
504 Once the AV has been created, the following operations are possible on it:
506 void av_push(AV*, SV*);
509 void av_unshift(AV*, SSize_t num);
511 These should be familiar operations, with the exception of C<av_unshift>.
512 This routine adds C<num> elements at the front of the array with the C<undef>
513 value. You must then use C<av_store> (described below) to assign values
514 to these new elements.
516 Here are some other functions:
518 SSize_t av_top_index(AV*);
519 SV** av_fetch(AV*, SSize_t key, I32 lval);
520 SV** av_store(AV*, SSize_t key, SV* val);
522 The C<av_top_index> function returns the highest index value in an array (just
523 like $#array in Perl). If the array is empty, -1 is returned. The
524 C<av_fetch> function returns the value at index C<key>, but if C<lval>
525 is non-zero, then C<av_fetch> will store an undef value at that index.
526 The C<av_store> function stores the value C<val> at index C<key>, and does
527 not increment the reference count of C<val>. Thus the caller is responsible
528 for taking care of that, and if C<av_store> returns NULL, the caller will
529 have to decrement the reference count to avoid a memory leak. Note that
530 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
537 void av_extend(AV*, SSize_t key);
539 The C<av_clear> function deletes all the elements in the AV* array, but
540 does not actually delete the array itself. The C<av_undef> function will
541 delete all the elements in the array plus the array itself. The
542 C<av_extend> function extends the array so that it contains at least C<key+1>
543 elements. If C<key+1> is less than the currently allocated length of the array,
544 then nothing is done.
546 If you know the name of an array variable, you can get a pointer to its AV
547 by using the following:
549 AV* get_av("package::varname", 0);
551 This returns NULL if the variable does not exist.
553 See L</Understanding the Magic of Tied Hashes and Arrays> for more
554 information on how to use the array access functions on tied arrays.
556 =head2 Working with HVs
558 To create an HV, you use the following routine:
562 Once the HV has been created, the following operations are possible on it:
564 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
565 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
567 The C<klen> parameter is the length of the key being passed in (Note that
568 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
569 length of the key). The C<val> argument contains the SV pointer to the
570 scalar being stored, and C<hash> is the precomputed hash value (zero if
571 you want C<hv_store> to calculate it for you). The C<lval> parameter
572 indicates whether this fetch is actually a part of a store operation, in
573 which case a new undefined value will be added to the HV with the supplied
574 key and C<hv_fetch> will return as if the value had already existed.
576 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
577 C<SV*>. To access the scalar value, you must first dereference the return
578 value. However, you should check to make sure that the return value is
579 not NULL before dereferencing it.
581 The first of these two functions checks if a hash table entry exists, and the
584 bool hv_exists(HV*, const char* key, U32 klen);
585 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
587 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
588 create and return a mortal copy of the deleted value.
590 And more miscellaneous functions:
595 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
596 table but does not actually delete the hash table. The C<hv_undef> deletes
597 both the entries and the hash table itself.
599 Perl keeps the actual data in a linked list of structures with a typedef of HE.
600 These contain the actual key and value pointers (plus extra administrative
601 overhead). The key is a string pointer; the value is an C<SV*>. However,
602 once you have an C<HE*>, to get the actual key and value, use the routines
607 I32 hv_iterinit(HV*);
608 /* Prepares starting point to traverse hash table */
609 HE* hv_iternext(HV*);
610 /* Get the next entry, and return a pointer to a
611 structure that has both the key and value */
612 char* hv_iterkey(HE* entry, I32* retlen);
613 /* Get the key from an HE structure and also return
614 the length of the key string */
615 SV* hv_iterval(HV*, HE* entry);
616 /* Return an SV pointer to the value of the HE
618 SV* hv_iternextsv(HV*, char** key, I32* retlen);
619 /* This convenience routine combines hv_iternext,
620 hv_iterkey, and hv_iterval. The key and retlen
621 arguments are return values for the key and its
622 length. The value is returned in the SV* argument */
624 If you know the name of a hash variable, you can get a pointer to its HV
625 by using the following:
627 HV* get_hv("package::varname", 0);
629 This returns NULL if the variable does not exist.
631 The hash algorithm is defined in the C<PERL_HASH> macro:
633 PERL_HASH(hash, key, klen)
635 The exact implementation of this macro varies by architecture and version
636 of perl, and the return value may change per invocation, so the value
637 is only valid for the duration of a single perl process.
639 See L</Understanding the Magic of Tied Hashes and Arrays> for more
640 information on how to use the hash access functions on tied hashes.
642 =for apidoc_section $HV
643 =for apidoc Amh|void|PERL_HASH|U32 hash|char *key|STRLEN klen
645 =head2 Hash API Extensions
647 Beginning with version 5.004, the following functions are also supported:
649 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
650 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
652 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
653 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
655 SV* hv_iterkeysv (HE* entry);
657 Note that these functions take C<SV*> keys, which simplifies writing
658 of extension code that deals with hash structures. These functions
659 also allow passing of C<SV*> keys to C<tie> functions without forcing
660 you to stringify the keys (unlike the previous set of functions).
662 They also return and accept whole hash entries (C<HE*>), making their
663 use more efficient (since the hash number for a particular string
664 doesn't have to be recomputed every time). See L<perlapi> for detailed
667 The following macros must always be used to access the contents of hash
668 entries. Note that the arguments to these macros must be simple
669 variables, since they may get evaluated more than once. See
670 L<perlapi> for detailed descriptions of these macros.
672 HePV(HE* he, STRLEN len)
676 HeSVKEY_force(HE* he)
677 HeSVKEY_set(HE* he, SV* sv)
679 These two lower level macros are defined, but must only be used when
680 dealing with keys that are not C<SV*>s:
685 Note that both C<hv_store> and C<hv_store_ent> do not increment the
686 reference count of the stored C<val>, which is the caller's responsibility.
687 If these functions return a NULL value, the caller will usually have to
688 decrement the reference count of C<val> to avoid a memory leak.
690 =head2 AVs, HVs and undefined values
692 Sometimes you have to store undefined values in AVs or HVs. Although
693 this may be a rare case, it can be tricky. That's because you're
694 used to using C<&PL_sv_undef> if you need an undefined SV.
696 For example, intuition tells you that this XS code:
699 av_store( av, 0, &PL_sv_undef );
701 is equivalent to this Perl code:
706 Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
707 for indicating that an array element has not yet been initialized.
708 Thus, C<exists $av[0]> would be true for the above Perl code, but
709 false for the array generated by the XS code. In perl 5.20, storing
710 &PL_sv_undef will create a read-only element, because the scalar
711 &PL_sv_undef itself is stored, not a copy.
713 Similar problems can occur when storing C<&PL_sv_undef> in HVs:
715 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
717 This will indeed make the value C<undef>, but if you try to modify
718 the value of C<key>, you'll get the following error:
720 Modification of non-creatable hash value attempted
722 In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
723 in restricted hashes. This caused such hash entries not to appear
724 when iterating over the hash or when checking for the keys
725 with the C<hv_exists> function.
727 You can run into similar problems when you store C<&PL_sv_yes> or
728 C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
729 will give you the following error:
731 Modification of a read-only value attempted
733 To make a long story short, you can use the special variables
734 C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
735 HVs, but you have to make sure you know what you're doing.
737 Generally, if you want to store an undefined value in an AV
738 or HV, you should not use C<&PL_sv_undef>, but rather create a
739 new undefined value using the C<newSV> function, for example:
741 av_store( av, 42, newSV(0) );
742 hv_store( hv, "foo", 3, newSV(0), 0 );
746 References are a special type of scalar that point to other data types
747 (including other references).
749 To create a reference, use either of the following functions:
751 SV* newRV_inc((SV*) thing);
752 SV* newRV_noinc((SV*) thing);
754 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
755 functions are identical except that C<newRV_inc> increments the reference
756 count of the C<thing>, while C<newRV_noinc> does not. For historical
757 reasons, C<newRV> is a synonym for C<newRV_inc>.
759 Once you have a reference, you can use the following macro to dereference
764 then call the appropriate routines, casting the returned C<SV*> to either an
765 C<AV*> or C<HV*>, if required.
767 To determine if an SV is a reference, you can use the following macro:
771 To discover what type of value the reference refers to, use the following
772 macro and then check the return value.
776 The most useful types that will be returned are:
781 SVt_PVGV Glob (possibly a file handle)
783 Any numerical value returned which is less than SVt_PVAV will be a scalar
786 See L<perlapi/svtype> for more details.
788 =head2 Blessed References and Class Objects
790 References are also used to support object-oriented programming. In perl's
791 OO lexicon, an object is simply a reference that has been blessed into a
792 package (or class). Once blessed, the programmer may now use the reference
793 to access the various methods in the class.
795 A reference can be blessed into a package with the following function:
797 SV* sv_bless(SV* sv, HV* stash);
799 The C<sv> argument must be a reference value. The C<stash> argument
800 specifies which class the reference will belong to. See
801 L</Stashes and Globs> for information on converting class names into stashes.
803 /* Still under construction */
805 The following function upgrades rv to reference if not already one.
806 Creates a new SV for rv to point to. If C<classname> is non-null, the SV
807 is blessed into the specified class. SV is returned.
809 SV* newSVrv(SV* rv, const char* classname);
811 The following three functions copy integer, unsigned integer or double
812 into an SV whose reference is C<rv>. SV is blessed if C<classname> is
815 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
816 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
817 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
819 The following function copies the pointer value (I<the address, not the
820 string!>) into an SV whose reference is rv. SV is blessed if C<classname>
823 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
825 The following function copies a string into an SV whose reference is C<rv>.
826 Set length to 0 to let Perl calculate the string length. SV is blessed if
827 C<classname> is non-null.
829 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
832 The following function tests whether the SV is blessed into the specified
833 class. It does not check inheritance relationships.
835 int sv_isa(SV* sv, const char* name);
837 The following function tests whether the SV is a reference to a blessed object.
839 int sv_isobject(SV* sv);
841 The following function tests whether the SV is derived from the specified
842 class. SV can be either a reference to a blessed object or a string
843 containing a class name. This is the function implementing the
844 C<UNIVERSAL::isa> functionality.
846 bool sv_derived_from(SV* sv, const char* name);
848 To check if you've got an object derived from a specific class you have
851 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
853 =head2 Creating New Variables
855 To create a new Perl variable with an undef value which can be accessed from
856 your Perl script, use the following routines, depending on the variable type.
858 SV* get_sv("package::varname", GV_ADD);
859 AV* get_av("package::varname", GV_ADD);
860 HV* get_hv("package::varname", GV_ADD);
862 Notice the use of GV_ADD as the second parameter. The new variable can now
863 be set, using the routines appropriate to the data type.
865 There are additional macros whose values may be bitwise OR'ed with the
866 C<GV_ADD> argument to enable certain extra features. Those bits are:
872 Marks the variable as multiply defined, thus preventing the:
874 Name <varname> used only once: possible typo
882 Had to create <varname> unexpectedly
884 if the variable did not exist before the function was called.
888 If you do not specify a package name, the variable is created in the current
891 =head2 Reference Counts and Mortality
893 Perl uses a reference count-driven garbage collection mechanism. SVs,
894 AVs, or HVs (xV for short in the following) start their life with a
895 reference count of 1. If the reference count of an xV ever drops to 0,
896 then it will be destroyed and its memory made available for reuse.
897 At the most basic internal level, reference counts can be manipulated
898 with the following macros:
900 int SvREFCNT(SV* sv);
901 SV* SvREFCNT_inc(SV* sv);
902 void SvREFCNT_dec(SV* sv);
904 (There are also suffixed versions of the increment and decrement macros,
905 for situations where the full generality of these basic macros can be
906 exchanged for some performance.)
908 However, the way a programmer should think about references is not so
909 much in terms of the bare reference count, but in terms of I<ownership>
910 of references. A reference to an xV can be owned by any of a variety
911 of entities: another xV, the Perl interpreter, an XS data structure,
912 a piece of running code, or a dynamic scope. An xV generally does not
913 know what entities own the references to it; it only knows how many
914 references there are, which is the reference count.
916 To correctly maintain reference counts, it is essential to keep track
917 of what references the XS code is manipulating. The programmer should
918 always know where a reference has come from and who owns it, and be
919 aware of any creation or destruction of references, and any transfers
920 of ownership. Because ownership isn't represented explicitly in the xV
921 data structures, only the reference count need be actually maintained
922 by the code, and that means that this understanding of ownership is not
923 actually evident in the code. For example, transferring ownership of a
924 reference from one owner to another doesn't change the reference count
925 at all, so may be achieved with no actual code. (The transferring code
926 doesn't touch the referenced object, but does need to ensure that the
927 former owner knows that it no longer owns the reference, and that the
928 new owner knows that it now does.)
930 An xV that is visible at the Perl level should not become unreferenced
931 and thus be destroyed. Normally, an object will only become unreferenced
932 when it is no longer visible, often by the same means that makes it
933 invisible. For example, a Perl reference value (RV) owns a reference to
934 its referent, so if the RV is overwritten that reference gets destroyed,
935 and the no-longer-reachable referent may be destroyed as a result.
937 Many functions have some kind of reference manipulation as
938 part of their purpose. Sometimes this is documented in terms
939 of ownership of references, and sometimes it is (less helpfully)
940 documented in terms of changes to reference counts. For example, the
941 L<newRV_inc()|perlapi/newRV_inc> function is documented to create a new RV
942 (with reference count 1) and increment the reference count of the referent
943 that was supplied by the caller. This is best understood as creating
944 a new reference to the referent, which is owned by the created RV,
945 and returning to the caller ownership of the sole reference to the RV.
946 The L<newRV_noinc()|perlapi/newRV_noinc> function instead does not
947 increment the reference count of the referent, but the RV nevertheless
948 ends up owning a reference to the referent. It is therefore implied
949 that the caller of C<newRV_noinc()> is relinquishing a reference to the
950 referent, making this conceptually a more complicated operation even
951 though it does less to the data structures.
953 For example, imagine you want to return a reference from an XSUB
954 function. Inside the XSUB routine, you create an SV which initially
955 has just a single reference, owned by the XSUB routine. This reference
956 needs to be disposed of before the routine is complete, otherwise it
957 will leak, preventing the SV from ever being destroyed. So to create
958 an RV referencing the SV, it is most convenient to pass the SV to
959 C<newRV_noinc()>, which consumes that reference. Now the XSUB routine
960 no longer owns a reference to the SV, but does own a reference to the RV,
961 which in turn owns a reference to the SV. The ownership of the reference
962 to the RV is then transferred by the process of returning the RV from
965 There are some convenience functions available that can help with the
966 destruction of xVs. These functions introduce the concept of "mortality".
967 Much documentation speaks of an xV itself being mortal, but this is
968 misleading. It is really I<a reference to> an xV that is mortal, and it
969 is possible for there to be more than one mortal reference to a single xV.
970 For a reference to be mortal means that it is owned by the temps stack,
971 one of perl's many internal stacks, which will destroy that reference
972 "a short time later". Usually the "short time later" is the end of
973 the current Perl statement. However, it gets more complicated around
974 dynamic scopes: there can be multiple sets of mortal references hanging
975 around at the same time, with different death dates. Internally, the
976 actual determinant for when mortal xV references are destroyed depends
977 on two macros, SAVETMPS and FREETMPS. See L<perlcall> and L<perlxs>
978 and L</Temporaries Stack> below for more details on these macros.
980 Mortal references are mainly used for xVs that are placed on perl's
981 main stack. The stack is problematic for reference tracking, because it
982 contains a lot of xV references, but doesn't own those references: they
983 are not counted. Currently, there are many bugs resulting from xVs being
984 destroyed while referenced by the stack, because the stack's uncounted
985 references aren't enough to keep the xVs alive. So when putting an
986 (uncounted) reference on the stack, it is vitally important to ensure that
987 there will be a counted reference to the same xV that will last at least
988 as long as the uncounted reference. But it's also important that that
989 counted reference be cleaned up at an appropriate time, and not unduly
990 prolong the xV's life. For there to be a mortal reference is often the
991 best way to satisfy this requirement, especially if the xV was created
992 especially to be put on the stack and would otherwise be unreferenced.
994 To create a mortal reference, use the functions:
997 SV* sv_mortalcopy(SV*)
1000 C<sv_newmortal()> creates an SV (with the undefined value) whose sole
1001 reference is mortal. C<sv_mortalcopy()> creates an xV whose value is a
1002 copy of a supplied xV and whose sole reference is mortal. C<sv_2mortal()>
1003 mortalises an existing xV reference: it transfers ownership of a reference
1004 from the caller to the temps stack. Because C<sv_newmortal> gives the new
1005 SV no value, it must normally be given one via C<sv_setpv>, C<sv_setiv>,
1008 SV *tmp = sv_newmortal();
1009 sv_setiv(tmp, an_integer);
1011 As that is multiple C statements it is quite common so see this idiom instead:
1013 SV *tmp = sv_2mortal(newSViv(an_integer));
1015 The mortal routines are not just for SVs; AVs and HVs can be
1016 made mortal by passing their address (type-casted to C<SV*>) to the
1017 C<sv_2mortal> or C<sv_mortalcopy> routines.
1019 =head2 Stashes and Globs
1021 A B<stash> is a hash that contains all variables that are defined
1022 within a package. Each key of the stash is a symbol
1023 name (shared by all the different types of objects that have the same
1024 name), and each value in the hash table is a GV (Glob Value). This GV
1025 in turn contains references to the various objects of that name,
1026 including (but not limited to) the following:
1035 There is a single stash called C<PL_defstash> that holds the items that exist
1036 in the C<main> package. To get at the items in other packages, append the
1037 string "::" to the package name. The items in the C<Foo> package are in
1038 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
1039 in the stash C<Baz::> in C<Bar::>'s stash.
1041 To get the stash pointer for a particular package, use the function:
1043 HV* gv_stashpv(const char* name, I32 flags)
1044 HV* gv_stashsv(SV*, I32 flags)
1046 The first function takes a literal string, the second uses the string stored
1047 in the SV. Remember that a stash is just a hash table, so you get back an
1048 C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
1050 The name that C<gv_stash*v> wants is the name of the package whose symbol table
1051 you want. The default package is called C<main>. If you have multiply nested
1052 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
1055 Alternately, if you have an SV that is a blessed reference, you can find
1056 out the stash pointer by using:
1058 HV* SvSTASH(SvRV(SV*));
1060 then use the following to get the package name itself:
1062 char* HvNAME(HV* stash);
1064 If you need to bless or re-bless an object you can use the following
1067 SV* sv_bless(SV*, HV* stash)
1069 where the first argument, an C<SV*>, must be a reference, and the second
1070 argument is a stash. The returned C<SV*> can now be used in the same way
1073 For more information on references and blessings, consult L<perlref>.
1077 Like AVs and HVs, IO objects are another type of non-scalar SV which
1078 may contain input and output L<PerlIO|perlapio> objects or a C<DIR *>
1081 You can create a new IO object:
1085 Unlike other SVs, a new IO object is automatically blessed into the
1088 The IO object contains an input and output PerlIO handle:
1090 PerlIO *IoIFP(IO *io);
1091 PerlIO *IoOFP(IO *io);
1093 Typically if the IO object has been opened on a file, the input handle
1094 is always present, but the output handle is only present if the file
1095 is open for output. For a file, if both are present they will be the
1098 Distinct input and output PerlIO objects are created for sockets and
1101 The IO object also contains other data associated with Perl I/O
1104 IV IoLINES(io); /* $. */
1105 IV IoPAGE(io); /* $% */
1106 IV IoPAGE_LEN(io); /* $= */
1107 IV IoLINES_LEFT(io); /* $- */
1108 char *IoTOP_NAME(io); /* $^ */
1109 GV *IoTOP_GV(io); /* $^ */
1110 char *IoFMT_NAME(io); /* $~ */
1111 GV *IoFMT_GV(io); /* $~ */
1112 char *IoBOTTOM_NAME(io);
1113 GV *IoBOTTOM_GV(io);
1117 Most of these are involved with L<formats|perlform>.
1119 IoFLAGs() may contain a combination of flags, the most interesting of
1120 which are C<IOf_FLUSH> (C<$|>) for autoflush and C<IOf_UNTAINT>,
1121 settable with L<< IO::Handle's untaint() method|IO::Handle/"$io->untaint" >>.
1123 The IO object may also contains a directory handle:
1127 suitable for use with PerlDir_read() etc.
1129 All of these accessors macros are lvalues, there are no distinct
1130 C<_set()> macros to modify the members of the IO object.
1132 =head2 Double-Typed SVs
1134 Scalar variables normally contain only one type of value, an integer,
1135 double, pointer, or reference. Perl will automatically convert the
1136 actual scalar data from the stored type into the requested type.
1138 Some scalar variables contain more than one type of scalar data. For
1139 example, the variable C<$!> contains either the numeric value of C<errno>
1140 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
1142 To force multiple data values into an SV, you must do two things: use the
1143 C<sv_set*v> routines to add the additional scalar type, then set a flag
1144 so that Perl will believe it contains more than one type of data. The
1145 four macros to set the flags are:
1152 The particular macro you must use depends on which C<sv_set*v> routine
1153 you called first. This is because every C<sv_set*v> routine turns on
1154 only the bit for the particular type of data being set, and turns off
1157 For example, to create a new Perl variable called "dberror" that contains
1158 both the numeric and descriptive string error values, you could use the
1162 extern char *dberror_list;
1164 SV* sv = get_sv("dberror", GV_ADD);
1165 sv_setiv(sv, (IV) dberror);
1166 sv_setpv(sv, dberror_list[dberror]);
1169 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
1170 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
1172 =head2 Read-Only Values
1174 In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
1175 flag bit with read-only scalars. So the only way to test whether
1176 C<sv_setsv>, etc., will raise a "Modification of a read-only value" error
1177 in those versions is:
1179 SvREADONLY(sv) && !SvIsCOW(sv)
1181 Under Perl 5.18 and later, SvREADONLY only applies to read-only variables,
1182 and, under 5.20, copy-on-write scalars can also be read-only, so the above
1183 check is incorrect. You just want:
1187 If you need to do this check often, define your own macro like this:
1189 #if PERL_VERSION >= 18
1190 # define SvTRULYREADONLY(sv) SvREADONLY(sv)
1192 # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
1195 =head2 Copy on Write
1197 Perl implements a copy-on-write (COW) mechanism for scalars, in which
1198 string copies are not immediately made when requested, but are deferred
1199 until made necessary by one or the other scalar changing. This is mostly
1200 transparent, but one must take care not to modify string buffers that are
1201 shared by multiple SVs.
1203 You can test whether an SV is using copy-on-write with C<SvIsCOW(sv)>.
1205 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).
1207 If you want to make the SV drop its string buffer, use
1208 C<sv_force_normal_flags(sv, SV_COW_DROP_PV)> or simply
1209 C<sv_setsv(sv, NULL)>.
1211 All of these functions will croak on read-only scalars (see the previous
1212 section for more on those).
1214 To test that your code is behaving correctly and not modifying COW buffers,
1215 on systems that support L<mmap(2)> (i.e., Unix) you can configure perl with
1216 C<-Accflags=-DPERL_DEBUG_READONLY_COW> and it will turn buffer violations
1217 into crashes. You will find it to be marvellously slow, so you may want to
1218 skip perl's own tests.
1220 =head2 Magic Variables
1222 [This section still under construction. Ignore everything here. Post no
1223 bills. Everything not permitted is forbidden.]
1225 Any SV may be magical, that is, it has special features that a normal
1226 SV does not have. These features are stored in the SV structure in a
1227 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
1230 MAGIC* mg_moremagic;
1240 Note this is current as of patchlevel 0, and could change at any time.
1242 =head2 Assigning Magic
1244 Perl adds magic to an SV using the sv_magic function:
1246 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
1248 The C<sv> argument is a pointer to the SV that is to acquire a new magical
1251 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
1252 convert C<sv> to type C<SVt_PVMG>.
1253 Perl then continues by adding new magic
1254 to the beginning of the linked list of magical features. Any prior entry
1255 of the same type of magic is deleted. Note that this can be overridden,
1256 and multiple instances of the same type of magic can be associated with an
1259 The C<name> and C<namlen> arguments are used to associate a string with
1260 the magic, typically the name of a variable. C<namlen> is stored in the
1261 C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
1262 C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
1263 whether C<namlen> is greater than zero or equal to zero respectively. As a
1264 special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
1265 to contain an C<SV*> and is stored as-is with its REFCNT incremented.
1267 The sv_magic function uses C<how> to determine which, if any, predefined
1268 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
1269 See the L</Magic Virtual Tables> section below. The C<how> argument is also
1270 stored in the C<mg_type> field. The value of
1271 C<how> should be chosen from the set of macros
1272 C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
1273 these macros were added, Perl internals used to directly use character
1274 literals, so you may occasionally come across old code or documentation
1275 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
1277 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
1278 structure. If it is not the same as the C<sv> argument, the reference
1279 count of the C<obj> object is incremented. If it is the same, or if
1280 the C<how> argument is C<PERL_MAGIC_arylen>, C<PERL_MAGIC_regdatum>,
1281 C<PERL_MAGIC_regdata>, or if it is a NULL pointer, then C<obj> is merely
1282 stored, without the reference count being incremented.
1284 See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
1287 There is also a function to add magic to an C<HV>:
1289 void hv_magic(HV *hv, GV *gv, int how);
1291 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
1293 To remove the magic from an SV, call the function sv_unmagic:
1295 int sv_unmagic(SV *sv, int type);
1297 The C<type> argument should be equal to the C<how> value when the C<SV>
1298 was initially made magical.
1300 However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
1301 C<SV>. If you want to remove only certain
1302 magic of a C<type> based on the magic
1303 virtual table, use C<sv_unmagicext> instead:
1305 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1307 =head2 Magic Virtual Tables
1309 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
1310 C<MGVTBL>, which is a structure of function pointers and stands for
1311 "Magic Virtual Table" to handle the various operations that might be
1312 applied to that variable.
1314 =for apidoc Ayh||MGVTBL
1316 The C<MGVTBL> has five (or sometimes eight) pointers to the following
1319 int (*svt_get) (pTHX_ SV* sv, MAGIC* mg);
1320 int (*svt_set) (pTHX_ SV* sv, MAGIC* mg);
1321 U32 (*svt_len) (pTHX_ SV* sv, MAGIC* mg);
1322 int (*svt_clear)(pTHX_ SV* sv, MAGIC* mg);
1323 int (*svt_free) (pTHX_ SV* sv, MAGIC* mg);
1325 int (*svt_copy) (pTHX_ SV *sv, MAGIC* mg, SV *nsv,
1326 const char *name, I32 namlen);
1327 int (*svt_dup) (pTHX_ MAGIC *mg, CLONE_PARAMS *param);
1328 int (*svt_local)(pTHX_ SV *nsv, MAGIC *mg);
1331 This MGVTBL structure is set at compile-time in F<perl.h> and there are
1332 currently 32 types. These different structures contain pointers to various
1333 routines that perform additional actions depending on which function is
1336 Function pointer Action taken
1337 ---------------- ------------
1338 svt_get Do something before the value of the SV is
1340 svt_set Do something after the SV is assigned a value.
1341 svt_len Report on the SV's length.
1342 svt_clear Clear something the SV represents.
1343 svt_free Free any extra storage associated with the SV.
1345 svt_copy copy tied variable magic to a tied element
1346 svt_dup duplicate a magic structure during thread cloning
1347 svt_local copy magic to local value during 'local'
1349 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
1350 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
1352 { magic_get, magic_set, magic_len, 0, 0 }
1354 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1355 if a get operation is being performed, the routine C<magic_get> is
1356 called. All the various routines for the various magical types begin
1357 with C<magic_>. NOTE: the magic routines are not considered part of
1358 the Perl API, and may not be exported by the Perl library.
1360 The last three slots are a recent addition, and for source code
1361 compatibility they are only checked for if one of the three flags
1362 MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1363 This means that most code can continue declaring
1364 a vtable as a 5-element value. These three are
1365 currently used exclusively by the threading code, and are highly subject
1368 The current kinds of Magic Virtual Tables are:
1371 This table is generated by regen/mg_vtable.pl. Any changes made here
1374 =for mg_vtable.pl begin
1377 (old-style char and macro) MGVTBL Type of magic
1378 -------------------------- ------ -------------
1379 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1380 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1381 % PERL_MAGIC_rhash (none) Extra data for restricted
1383 * PERL_MAGIC_debugvar vtbl_debugvar $DB::single, signal, trace
1385 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1386 : PERL_MAGIC_symtab (none) Extra data for symbol
1388 < PERL_MAGIC_backref vtbl_backref For weak ref data
1389 @ PERL_MAGIC_arylen_p (none) To move arylen out of XPVAV
1390 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1391 (fast string search)
1392 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1394 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1396 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1398 E PERL_MAGIC_env vtbl_env %ENV hash
1399 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1400 f PERL_MAGIC_fm vtbl_regexp Formline
1402 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1403 H PERL_MAGIC_hints vtbl_hints %^H hash
1404 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1405 I PERL_MAGIC_isa vtbl_isa @ISA array
1406 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1407 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1408 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1409 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1411 N PERL_MAGIC_shared (none) Shared between threads
1412 n PERL_MAGIC_shared_scalar (none) Shared between threads
1413 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1414 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1415 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1416 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1417 r PERL_MAGIC_qr vtbl_regexp Precompiled qr// regex
1418 S PERL_MAGIC_sig vtbl_sig %SIG hash
1419 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1420 t PERL_MAGIC_taint vtbl_taint Taintedness
1421 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1423 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1425 V PERL_MAGIC_vstring (none) SV was vstring literal
1426 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1427 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1428 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1429 Y PERL_MAGIC_nonelem vtbl_nonelem Array element that does not
1431 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1432 variable / smart parameter
1434 \ PERL_MAGIC_lvref vtbl_lvref Lvalue reference
1436 ] PERL_MAGIC_checkcall vtbl_checkcall Inlining/mutation of call
1438 ~ PERL_MAGIC_ext (none) Available for use by
1442 =for apidoc AmnhU||PERL_MAGIC_arylen
1443 =for apidoc_item ||PERL_MAGIC_arylen_p
1444 =for apidoc_item ||PERL_MAGIC_backref
1445 =for apidoc_item ||PERL_MAGIC_bm
1446 =for apidoc_item ||PERL_MAGIC_checkcall
1447 =for apidoc_item ||PERL_MAGIC_collxfrm
1448 =for apidoc_item ||PERL_MAGIC_dbfile
1449 =for apidoc_item ||PERL_MAGIC_dbline
1450 =for apidoc_item ||PERL_MAGIC_debugvar
1451 =for apidoc_item ||PERL_MAGIC_defelem
1452 =for apidoc_item ||PERL_MAGIC_env
1453 =for apidoc_item ||PERL_MAGIC_envelem
1454 =for apidoc_item ||PERL_MAGIC_ext
1455 =for apidoc_item ||PERL_MAGIC_fm
1456 =for apidoc_item ||PERL_MAGIC_hints
1457 =for apidoc_item ||PERL_MAGIC_hintselem
1458 =for apidoc_item ||PERL_MAGIC_isa
1459 =for apidoc_item ||PERL_MAGIC_isaelem
1460 =for apidoc_item ||PERL_MAGIC_lvref
1461 =for apidoc_item ||PERL_MAGIC_nkeys
1462 =for apidoc_item ||PERL_MAGIC_nonelem
1463 =for apidoc_item ||PERL_MAGIC_overload_table
1464 =for apidoc_item ||PERL_MAGIC_pos
1465 =for apidoc_item ||PERL_MAGIC_qr
1466 =for apidoc_item ||PERL_MAGIC_regdata
1467 =for apidoc_item ||PERL_MAGIC_regdatum
1468 =for apidoc_item ||PERL_MAGIC_regex_global
1469 =for apidoc_item ||PERL_MAGIC_rhash
1470 =for apidoc_item ||PERL_MAGIC_shared
1471 =for apidoc_item ||PERL_MAGIC_shared_scalar
1472 =for apidoc_item ||PERL_MAGIC_sig
1473 =for apidoc_item ||PERL_MAGIC_sigelem
1474 =for apidoc_item ||PERL_MAGIC_substr
1475 =for apidoc_item ||PERL_MAGIC_sv
1476 =for apidoc_item ||PERL_MAGIC_symtab
1477 =for apidoc_item ||PERL_MAGIC_taint
1478 =for apidoc_item ||PERL_MAGIC_tied
1479 =for apidoc_item ||PERL_MAGIC_tiedelem
1480 =for apidoc_item ||PERL_MAGIC_tiedscalar
1481 =for apidoc_item ||PERL_MAGIC_utf8
1482 =for apidoc_item ||PERL_MAGIC_uvar
1483 =for apidoc_item ||PERL_MAGIC_uvar_elem
1484 =for apidoc_item ||PERL_MAGIC_vec
1485 =for apidoc_item ||PERL_MAGIC_vstring
1487 =for mg_vtable.pl end
1489 When an uppercase and lowercase letter both exist in the table, then the
1490 uppercase letter is typically used to represent some kind of composite type
1491 (a list or a hash), and the lowercase letter is used to represent an element
1492 of that composite type. Some internals code makes use of this case
1493 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
1495 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1496 specifically for use by extensions and will not be used by perl itself.
1497 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1498 to variables (typically objects). This is especially useful because
1499 there is no way for normal perl code to corrupt this private information
1500 (unlike using extra elements of a hash object).
1502 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1503 C function any time a scalar's value is used or changed. The C<MAGIC>'s
1504 C<mg_ptr> field points to a C<ufuncs> structure:
1507 I32 (*uf_val)(pTHX_ IV, SV*);
1508 I32 (*uf_set)(pTHX_ IV, SV*);
1512 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1513 function will be called with C<uf_index> as the first arg and a pointer to
1514 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1515 magic is shown below. Note that the ufuncs structure is copied by
1516 sv_magic, so you can safely allocate it on the stack.
1524 uf.uf_val = &my_get_fn;
1525 uf.uf_set = &my_set_fn;
1527 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1529 Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1531 For hashes there is a specialized hook that gives control over hash
1532 keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1533 if the "set" function in the C<ufuncs> structure is NULL. The hook
1534 is activated whenever the hash is accessed with a key specified as
1535 an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1536 C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1537 through the functions without the C<..._ent> suffix circumvents the
1538 hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1540 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1541 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1542 extra care to avoid conflict. Typically only using the magic on
1543 objects blessed into the same class as the extension is sufficient.
1544 For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1545 C<MGVTBL>, even if all its fields will be C<0>, so that individual
1546 C<MAGIC> pointers can be identified as a particular kind of magic
1547 using their magic virtual table. C<mg_findext> provides an easy way
1550 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1553 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1554 /* this is really ours, not another module's PERL_MAGIC_ext */
1555 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1559 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1560 earlier do B<not> invoke 'set' magic on their targets. This must
1561 be done by the user either by calling the C<SvSETMAGIC()> macro after
1562 calling these functions, or by using one of the C<sv_set*_mg()> or
1563 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1564 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1565 obtained from external sources in functions that don't handle magic.
1566 See L<perlapi> for a description of these functions.
1567 For example, calls to the C<sv_cat*()> functions typically need to be
1568 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1569 since their implementation handles 'get' magic.
1571 =head2 Finding Magic
1573 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1576 This routine returns a pointer to a C<MAGIC> structure stored in the SV.
1577 If the SV does not have that magical
1578 feature, C<NULL> is returned. If the
1579 SV has multiple instances of that magical feature, the first one will be
1580 returned. C<mg_findext> can be used
1581 to find a C<MAGIC> structure of an SV
1582 based on both its magic type and its magic virtual table:
1584 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
1586 Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1587 SVt_PVMG, Perl may core dump.
1589 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1591 This routine checks to see what types of magic C<sv> has. If the mg_type
1592 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1593 the mg_type field is changed to be the lowercase letter.
1595 =head2 Understanding the Magic of Tied Hashes and Arrays
1597 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1600 WARNING: As of the 5.004 release, proper usage of the array and hash
1601 access functions requires understanding a few caveats. Some
1602 of these caveats are actually considered bugs in the API, to be fixed
1603 in later releases, and are bracketed with [MAYCHANGE] below. If
1604 you find yourself actually applying such information in this section, be
1605 aware that the behavior may change in the future, umm, without warning.
1607 The perl tie function associates a variable with an object that implements
1608 the various GET, SET, etc methods. To perform the equivalent of the perl
1609 tie function from an XSUB, you must mimic this behaviour. The code below
1610 carries out the necessary steps -- firstly it creates a new hash, and then
1611 creates a second hash which it blesses into the class which will implement
1612 the tie methods. Lastly it ties the two hashes together, and returns a
1613 reference to the new tied hash. Note that the code below does NOT call the
1614 TIEHASH method in the MyTie class -
1615 see L</Calling Perl Routines from within C Programs> for details on how
1626 tie = newRV_noinc((SV*)newHV());
1627 stash = gv_stashpv("MyTie", GV_ADD);
1628 sv_bless(tie, stash);
1629 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1630 RETVAL = newRV_noinc(hash);
1634 The C<av_store> function, when given a tied array argument, merely
1635 copies the magic of the array onto the value to be "stored", using
1636 C<mg_copy>. It may also return NULL, indicating that the value did not
1637 actually need to be stored in the array. [MAYCHANGE] After a call to
1638 C<av_store> on a tied array, the caller will usually need to call
1639 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1640 TIEARRAY object. If C<av_store> did return NULL, a call to
1641 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1644 The previous paragraph is applicable verbatim to tied hash access using the
1645 C<hv_store> and C<hv_store_ent> functions as well.
1647 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1648 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1649 has been initialized using C<mg_copy>. Note the value so returned does not
1650 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1651 need to call C<mg_get()> on the returned value in order to actually invoke
1652 the perl level "FETCH" method on the underlying TIE object. Similarly,
1653 you may also call C<mg_set()> on the return value after possibly assigning
1654 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1655 method on the TIE object. [/MAYCHANGE]
1658 In other words, the array or hash fetch/store functions don't really
1659 fetch and store actual values in the case of tied arrays and hashes. They
1660 merely call C<mg_copy> to attach magic to the values that were meant to be
1661 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1662 do the job of invoking the TIE methods on the underlying objects. Thus
1663 the magic mechanism currently implements a kind of lazy access to arrays
1666 Currently (as of perl version 5.004), use of the hash and array access
1667 functions requires the user to be aware of whether they are operating on
1668 "normal" hashes and arrays, or on their tied variants. The API may be
1669 changed to provide more transparent access to both tied and normal data
1670 types in future versions.
1673 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1674 are mere sugar to invoke some perl method calls while using the uniform hash
1675 and array syntax. The use of this sugar imposes some overhead (typically
1676 about two to four extra opcodes per FETCH/STORE operation, in addition to
1677 the creation of all the mortal variables required to invoke the methods).
1678 This overhead will be comparatively small if the TIE methods are themselves
1679 substantial, but if they are only a few statements long, the overhead
1680 will not be insignificant.
1682 =head2 Localizing changes
1684 Perl has a very handy construction
1691 This construction is I<approximately> equivalent to
1700 The biggest difference is that the first construction would
1701 reinstate the initial value of $var, irrespective of how control exits
1702 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1703 more efficient as well.
1705 There is a way to achieve a similar task from C via Perl API: create a
1706 I<pseudo-block>, and arrange for some changes to be automatically
1707 undone at the end of it, either explicit, or via a non-local exit (via
1708 die()). A I<block>-like construct is created by a pair of
1709 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1710 Such a construct may be created specially for some important localized
1711 task, or an existing one (like boundaries of enclosing Perl
1712 subroutine/block, or an existing pair for freeing TMPs) may be
1713 used. (In the second case the overhead of additional localization must
1714 be almost negligible.) Note that any XSUB is automatically enclosed in
1715 an C<ENTER>/C<LEAVE> pair.
1717 Inside such a I<pseudo-block> the following service is available:
1721 =item C<SAVEINT(int i)>
1723 =item C<SAVEIV(IV i)>
1725 =item C<SAVEI32(I32 i)>
1727 =item C<SAVELONG(long i)>
1729 =item C<SAVEI8(I8 i)>
1731 =item C<SAVEI16(I16 i)>
1733 =item C<SAVEBOOL(int i)>
1735 =item C<SAVESTRLEN(STRLEN i)>
1737 These macros arrange things to restore the value of integer variable
1738 C<i> at the end of the enclosing I<pseudo-block>.
1740 =for apidoc_section $stack
1741 =for apidoc Amh||SAVEINT|int i
1742 =for apidoc Amh||SAVEIV|IV i
1743 =for apidoc Amh||SAVEI32|I32 i
1744 =for apidoc Amh||SAVELONG|long i
1745 =for apidoc Amh||SAVEI8|I8 i
1746 =for apidoc Amh||SAVEI16|I16 i
1747 =for apidoc Amh||SAVEBOOL|bool i
1748 =for apidoc Amh||SAVESTRLEN|STRLEN i
1750 =item C<SAVESPTR(s)>
1752 =item C<SAVEPPTR(p)>
1754 These macros arrange things to restore the value of pointers C<s> and
1755 C<p>. C<s> must be a pointer of a type which survives conversion to
1756 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1759 =for apidoc Amh||SAVESPTR|SV * s
1760 =for apidoc Amh||SAVEPPTR|char * p
1762 =item C<SAVEFREESV(SV *sv)>
1764 The refcount of C<sv> will be decremented at the end of
1765 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1766 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1767 extends the lifetime of C<sv> until the beginning of the next statement,
1768 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1769 lifetimes can be wildly different.
1771 Also compare C<SAVEMORTALIZESV>.
1773 =for apidoc Amh||SAVEFREESV|SV* sv
1775 =item C<SAVEMORTALIZESV(SV *sv)>
1777 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1778 scope instead of decrementing its reference count. This usually has the
1779 effect of keeping C<sv> alive until the statement that called the currently
1780 live scope has finished executing.
1782 =for apidoc Amh||SAVEMORTALIZESV|SV* sv
1784 =item C<SAVEFREEOP(OP *op)>
1786 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1788 =for apidoc Amh||SAVEFREEOP|OP *op
1790 =item C<SAVEFREEPV(p)>
1792 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1793 end of I<pseudo-block>.
1795 =for apidoc Amh||SAVEFREEPV|void * p
1797 =item C<SAVECLEARSV(SV *sv)>
1799 Clears a slot in the current scratchpad which corresponds to C<sv> at
1800 the end of I<pseudo-block>.
1802 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1804 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1805 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1806 short-lived storage, the corresponding string may be reallocated like
1809 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1811 =for apidoc Amh||SAVEDELETE|HV * hv|char * key|I32 length
1813 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1815 At the end of I<pseudo-block> the function C<f> is called with the
1818 =for apidoc Ayh||DESTRUCTORFUNC_NOCONTEXT_t
1819 =for apidoc Amh||SAVEDESTRUCTOR|DESTRUCTORFUNC_NOCONTEXT_t f|void *p
1821 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1823 At the end of I<pseudo-block> the function C<f> is called with the
1824 implicit context argument (if any), and C<p>.
1826 =for apidoc Ayh||DESTRUCTORFUNC_t
1827 =for apidoc Amh||SAVEDESTRUCTOR_X|DESTRUCTORFUNC_t f|void *p
1829 =item C<SAVESTACK_POS()>
1831 The current offset on the Perl internal stack (cf. C<SP>) is restored
1832 at the end of I<pseudo-block>.
1834 =for apidoc Amh||SAVESTACK_POS
1838 The following API list contains functions, thus one needs to
1839 provide pointers to the modifiable data explicitly (either C pointers,
1840 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1841 function takes C<int *>.
1843 Other macros above have functions implementing them, but its probably
1844 best to just use the macro, and not those or the ones below.
1848 =item C<SV* save_scalar(GV *gv)>
1850 =for apidoc save_scalar
1852 Equivalent to Perl code C<local $gv>.
1854 =item C<AV* save_ary(GV *gv)>
1856 =for apidoc save_ary
1858 =item C<HV* save_hash(GV *gv)>
1860 =for apidoc save_hash
1862 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1864 =item C<void save_item(SV *item)>
1866 =for apidoc save_item
1868 Duplicates the current value of C<SV>. On the exit from the current
1869 C<ENTER>/C<LEAVE> I<pseudo-block> the value of C<SV> will be restored
1870 using the stored value. It doesn't handle magic. Use C<save_scalar> if
1873 =item C<void save_list(SV **sarg, I32 maxsarg)>
1875 =for apidoc save_list
1877 A variant of C<save_item> which takes multiple arguments via an array
1878 C<sarg> of C<SV*> of length C<maxsarg>.
1880 =item C<SV* save_svref(SV **sptr)>
1882 =for apidoc save_svref
1884 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1886 =item C<void save_aptr(AV **aptr)>
1888 =item C<void save_hptr(HV **hptr)>
1890 =for apidoc save_aptr
1891 =for apidoc save_hptr
1893 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1897 The C<Alias> module implements localization of the basic types within the
1898 I<caller's scope>. People who are interested in how to localize things in
1899 the containing scope should take a look there too.
1903 =head2 XSUBs and the Argument Stack
1905 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1906 An XSUB routine will have a stack that contains the arguments from the Perl
1907 program, and a way to map from the Perl data structures to a C equivalent.
1909 The stack arguments are accessible through the C<ST(n)> macro, which returns
1910 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1911 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1914 Most of the time, output from the C routine can be handled through use of
1915 the RETVAL and OUTPUT directives. However, there are some cases where the
1916 argument stack is not already long enough to handle all the return values.
1917 An example is the POSIX tzname() call, which takes no arguments, but returns
1918 two, the local time zone's standard and summer time abbreviations.
1920 To handle this situation, the PPCODE directive is used and the stack is
1921 extended using the macro:
1925 where C<SP> is the macro that represents the local copy of the stack pointer,
1926 and C<num> is the number of elements the stack should be extended by.
1928 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1929 macro. The pushed values will often need to be "mortal" (See
1930 L</Reference Counts and Mortality>):
1932 PUSHs(sv_2mortal(newSViv(an_integer)))
1933 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1934 PUSHs(sv_2mortal(newSVnv(a_double)))
1935 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1936 /* Although the last example is better written as the more
1938 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
1940 And now the Perl program calling C<tzname>, the two values will be assigned
1943 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1945 An alternate (and possibly simpler) method to pushing values on the stack is
1950 This macro automatically adjusts the stack for you, if needed. Thus, you
1951 do not need to call C<EXTEND> to extend the stack.
1953 Despite their suggestions in earlier versions of this document the macros
1954 C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1955 For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1956 C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
1958 For more information, consult L<perlxs> and L<perlxstut>.
1960 =head2 Autoloading with XSUBs
1962 If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1963 fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1964 of the XSUB's package.
1966 But it also puts the same information in certain fields of the XSUB itself:
1968 HV *stash = CvSTASH(cv);
1969 const char *subname = SvPVX(cv);
1970 STRLEN name_length = SvCUR(cv); /* in bytes */
1971 U32 is_utf8 = SvUTF8(cv);
1973 C<SvPVX(cv)> contains just the sub name itself, not including the package.
1974 For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1975 C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
1977 B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1978 XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1979 XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1980 to support 5.8-5.14, use the XSUB's fields.
1982 =head2 Calling Perl Routines from within C Programs
1984 There are four routines that can be used to call a Perl subroutine from
1985 within a C program. These four are:
1987 I32 call_sv(SV*, I32);
1988 I32 call_pv(const char*, I32);
1989 I32 call_method(const char*, I32);
1990 I32 call_argv(const char*, I32, char**);
1992 The routine most often used is C<call_sv>. The C<SV*> argument
1993 contains either the name of the Perl subroutine to be called, or a
1994 reference to the subroutine. The second argument consists of flags
1995 that control the context in which the subroutine is called, whether
1996 or not the subroutine is being passed arguments, how errors should be
1997 trapped, and how to treat return values.
1999 All four routines return the number of arguments that the subroutine returned
2002 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
2003 but those names are now deprecated; macros of the same name are provided for
2006 When using any of these routines (except C<call_argv>), the programmer
2007 must manipulate the Perl stack. These include the following macros and
2022 For a detailed description of calling conventions from C to Perl,
2023 consult L<perlcall>.
2025 =head2 Putting a C value on Perl stack
2027 A lot of opcodes (this is an elementary operation in the internal perl
2028 stack machine) put an SV* on the stack. However, as an optimization
2029 the corresponding SV is (usually) not recreated each time. The opcodes
2030 reuse specially assigned SVs (I<target>s) which are (as a corollary)
2031 not constantly freed/created.
2033 Each of the targets is created only once (but see
2034 L</Scratchpads and recursion> below), and when an opcode needs to put
2035 an integer, a double, or a string on stack, it just sets the
2036 corresponding parts of its I<target> and puts the I<target> on stack.
2038 The macro to put this target on stack is C<PUSHTARG>, and it is
2039 directly used in some opcodes, as well as indirectly in zillions of
2040 others, which use it via C<(X)PUSH[iunp]>.
2042 Because the target is reused, you must be careful when pushing multiple
2043 values on the stack. The following code will not do what you think:
2048 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
2049 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
2050 At the end of the operation, the stack does not contain the values 10
2051 and 20, but actually contains two pointers to C<TARG>, which we have set
2054 If you need to push multiple different values then you should either use
2055 the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
2056 none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
2057 SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
2058 will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
2059 this a little easier to achieve by creating a new mortal for you (via
2060 C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
2061 in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
2062 Thus, instead of writing this to "fix" the example above:
2064 XPUSHs(sv_2mortal(newSViv(10)))
2065 XPUSHs(sv_2mortal(newSViv(20)))
2067 you can simply write:
2072 On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
2073 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
2074 macros can make use of the local variable C<TARG>. See also C<dTARGET>
2079 The question remains on when the SVs which are I<target>s for opcodes
2080 are created. The answer is that they are created when the current
2081 unit--a subroutine or a file (for opcodes for statements outside of
2082 subroutines)--is compiled. During this time a special anonymous Perl
2083 array is created, which is called a scratchpad for the current unit.
2085 A scratchpad keeps SVs which are lexicals for the current unit and are
2086 targets for opcodes. A previous version of this document
2087 stated that one can deduce that an SV lives on a scratchpad
2088 by looking on its flags: lexicals have C<SVs_PADMY> set, and
2089 I<target>s have C<SVs_PADTMP> set. But this has never been fully true.
2090 C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
2091 While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
2092 that have never resided in a pad, but nonetheless act like I<target>s. As
2093 of perl 5.21.5, the C<SVs_PADMY> flag is no longer used and is defined as
2094 0. C<SvPADMY()> now returns true for anything without C<SVs_PADTMP>.
2096 The correspondence between OPs and I<target>s is not 1-to-1. Different
2097 OPs in the compile tree of the unit can use the same target, if this
2098 would not conflict with the expected life of the temporary.
2100 =head2 Scratchpads and recursion
2102 In fact it is not 100% true that a compiled unit contains a pointer to
2103 the scratchpad AV. In fact it contains a pointer to an AV of
2104 (initially) one element, and this element is the scratchpad AV. Why do
2105 we need an extra level of indirection?
2107 The answer is B<recursion>, and maybe B<threads>. Both
2108 these can create several execution pointers going into the same
2109 subroutine. For the subroutine-child not write over the temporaries
2110 for the subroutine-parent (lifespan of which covers the call to the
2111 child), the parent and the child should have different
2112 scratchpads. (I<And> the lexicals should be separate anyway!)
2114 So each subroutine is born with an array of scratchpads (of length 1).
2115 On each entry to the subroutine it is checked that the current
2116 depth of the recursion is not more than the length of this array, and
2117 if it is, new scratchpad is created and pushed into the array.
2119 The I<target>s on this scratchpad are C<undef>s, but they are already
2120 marked with correct flags.
2122 =head1 Memory Allocation
2126 All memory meant to be used with the Perl API functions should be manipulated
2127 using the macros described in this section. The macros provide the necessary
2128 transparency between differences in the actual malloc implementation that is
2131 The following three macros are used to initially allocate memory :
2133 Newx(pointer, number, type);
2134 Newxc(pointer, number, type, cast);
2135 Newxz(pointer, number, type);
2137 The first argument C<pointer> should be the name of a variable that will
2138 point to the newly allocated memory.
2140 The second and third arguments C<number> and C<type> specify how many of
2141 the specified type of data structure should be allocated. The argument
2142 C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
2143 should be used if the C<pointer> argument is different from the C<type>
2146 Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
2147 to zero out all the newly allocated memory.
2151 Renew(pointer, number, type);
2152 Renewc(pointer, number, type, cast);
2155 These three macros are used to change a memory buffer size or to free a
2156 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
2157 match those of C<New> and C<Newc> with the exception of not needing the
2158 "magic cookie" argument.
2162 Move(source, dest, number, type);
2163 Copy(source, dest, number, type);
2164 Zero(dest, number, type);
2166 These three macros are used to move, copy, or zero out previously allocated
2167 memory. The C<source> and C<dest> arguments point to the source and
2168 destination starting points. Perl will move, copy, or zero out C<number>
2169 instances of the size of the C<type> data structure (using the C<sizeof>
2174 The most recent development releases of Perl have been experimenting with
2175 removing Perl's dependency on the "normal" standard I/O suite and allowing
2176 other stdio implementations to be used. This involves creating a new
2177 abstraction layer that then calls whichever implementation of stdio Perl
2178 was compiled with. All XSUBs should now use the functions in the PerlIO
2179 abstraction layer and not make any assumptions about what kind of stdio
2182 For a complete description of the PerlIO abstraction, consult L<perlapio>.
2184 =head1 Compiled code
2188 Here we describe the internal form your code is converted to by
2189 Perl. Start with a simple example:
2193 This is converted to a tree similar to this one:
2201 (but slightly more complicated). This tree reflects the way Perl
2202 parsed your code, but has nothing to do with the execution order.
2203 There is an additional "thread" going through the nodes of the tree
2204 which shows the order of execution of the nodes. In our simplified
2205 example above it looks like:
2207 $b ---> $c ---> + ---> $a ---> assign-to
2209 But with the actual compile tree for C<$a = $b + $c> it is different:
2210 some nodes I<optimized away>. As a corollary, though the actual tree
2211 contains more nodes than our simplified example, the execution order
2212 is the same as in our example.
2214 =head2 Examining the tree
2216 If you have your perl compiled for debugging (usually done with
2217 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
2218 compiled tree by specifying C<-Dx> on the Perl command line. The
2219 output takes several lines per node, and for C<$b+$c> it looks like
2224 FLAGS = (SCALAR,KIDS)
2226 TYPE = null ===> (4)
2228 FLAGS = (SCALAR,KIDS)
2230 3 TYPE = gvsv ===> 4
2236 TYPE = null ===> (5)
2238 FLAGS = (SCALAR,KIDS)
2240 4 TYPE = gvsv ===> 5
2246 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
2247 not optimized away (one per number in the left column). The immediate
2248 children of the given node correspond to C<{}> pairs on the same level
2249 of indentation, thus this listing corresponds to the tree:
2257 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
2258 4 5 6> (node C<6> is not included into above listing), i.e.,
2259 C<gvsv gvsv add whatever>.
2261 Each of these nodes represents an op, a fundamental operation inside the
2262 Perl core. The code which implements each operation can be found in the
2263 F<pp*.c> files; the function which implements the op with type C<gvsv>
2264 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
2265 different numbers of children: C<add> is a binary operator, as one would
2266 expect, and so has two children. To accommodate the various different
2267 numbers of children, there are various types of op data structure, and
2268 they link together in different ways.
2270 The simplest type of op structure is C<OP>: this has no children. Unary
2271 operators, C<UNOP>s, have one child, and this is pointed to by the
2272 C<op_first> field. Binary operators (C<BINOP>s) have not only an
2273 C<op_first> field but also an C<op_last> field. The most complex type of
2274 op is a C<LISTOP>, which has any number of children. In this case, the
2275 first child is pointed to by C<op_first> and the last child by
2276 C<op_last>. The children in between can be found by iteratively
2277 following the C<OpSIBLING> pointer from the first child to the last (but
2281 =for apidoc Ayh||BINOP
2282 =for apidoc Ayh||LISTOP
2283 =for apidoc Ayh||UNOP
2285 There are also some other op types: a C<PMOP> holds a regular expression,
2286 and has no children, and a C<LOOP> may or may not have children. If the
2287 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
2288 complicate matters, if a C<UNOP> is actually a C<null> op after
2289 optimization (see L</Compile pass 2: context propagation>) it will still
2290 have children in accordance with its former type.
2292 =for apidoc Ayh||LOOP
2293 =for apidoc Ayh||PMOP
2295 Finally, there is a C<LOGOP>, or logic op. Like a C<LISTOP>, this has one
2296 or more children, but it doesn't have an C<op_last> field: so you have to
2297 follow C<op_first> and then the C<OpSIBLING> chain itself to find the
2298 last child. Instead it has an C<op_other> field, which is comparable to
2299 the C<op_next> field described below, and represents an alternate
2300 execution path. Operators like C<and>, C<or> and C<?> are C<LOGOP>s. Note
2301 that in general, C<op_other> may not point to any of the direct children
2304 =for apidoc Ayh||LOGOP
2306 Starting in version 5.21.2, perls built with the experimental
2307 define C<-DPERL_OP_PARENT> add an extra boolean flag for each op,
2308 C<op_moresib>. When not set, this indicates that this is the last op in an
2309 C<OpSIBLING> chain. This frees up the C<op_sibling> field on the last
2310 sibling to point back to the parent op. Under this build, that field is
2311 also renamed C<op_sibparent> to reflect its joint role. The macro
2312 C<OpSIBLING(o)> wraps this special behaviour, and always returns NULL on
2313 the last sibling. With this build the C<op_parent(o)> function can be
2314 used to find the parent of any op. Thus for forward compatibility, you
2315 should always use the C<OpSIBLING(o)> macro rather than accessing
2316 C<op_sibling> directly.
2318 Another way to examine the tree is to use a compiler back-end module, such
2321 =head2 Compile pass 1: check routines
2323 The tree is created by the compiler while I<yacc> code feeds it
2324 the constructions it recognizes. Since I<yacc> works bottom-up, so does
2325 the first pass of perl compilation.
2327 What makes this pass interesting for perl developers is that some
2328 optimization may be performed on this pass. This is optimization by
2329 so-called "check routines". The correspondence between node names
2330 and corresponding check routines is described in F<opcode.pl> (do not
2331 forget to run C<make regen_headers> if you modify this file).
2333 A check routine is called when the node is fully constructed except
2334 for the execution-order thread. Since at this time there are no
2335 back-links to the currently constructed node, one can do most any
2336 operation to the top-level node, including freeing it and/or creating
2337 new nodes above/below it.
2339 The check routine returns the node which should be inserted into the
2340 tree (if the top-level node was not modified, check routine returns
2343 By convention, check routines have names C<ck_*>. They are usually
2344 called from C<new*OP> subroutines (or C<convert>) (which in turn are
2345 called from F<perly.y>).
2347 =head2 Compile pass 1a: constant folding
2349 Immediately after the check routine is called the returned node is
2350 checked for being compile-time executable. If it is (the value is
2351 judged to be constant) it is immediately executed, and a I<constant>
2352 node with the "return value" of the corresponding subtree is
2353 substituted instead. The subtree is deleted.
2355 If constant folding was not performed, the execution-order thread is
2358 =head2 Compile pass 2: context propagation
2360 When a context for a part of compile tree is known, it is propagated
2361 down through the tree. At this time the context can have 5 values
2362 (instead of 2 for runtime context): void, boolean, scalar, list, and
2363 lvalue. In contrast with the pass 1 this pass is processed from top
2364 to bottom: a node's context determines the context for its children.
2366 Additional context-dependent optimizations are performed at this time.
2367 Since at this moment the compile tree contains back-references (via
2368 "thread" pointers), nodes cannot be free()d now. To allow
2369 optimized-away nodes at this stage, such nodes are null()ified instead
2370 of free()ing (i.e. their type is changed to OP_NULL).
2372 =head2 Compile pass 3: peephole optimization
2374 After the compile tree for a subroutine (or for an C<eval> or a file)
2375 is created, an additional pass over the code is performed. This pass
2376 is neither top-down or bottom-up, but in the execution order (with
2377 additional complications for conditionals). Optimizations performed
2378 at this stage are subject to the same restrictions as in the pass 2.
2380 Peephole optimizations are done by calling the function pointed to
2381 by the global variable C<PL_peepp>. By default, C<PL_peepp> just
2382 calls the function pointed to by the global variable C<PL_rpeepp>.
2383 By default, that performs some basic op fixups and optimisations along
2384 the execution-order op chain, and recursively calls C<PL_rpeepp> for
2385 each side chain of ops (resulting from conditionals). Extensions may
2386 provide additional optimisations or fixups, hooking into either the
2387 per-subroutine or recursive stage, like this:
2389 static peep_t prev_peepp;
2390 static void my_peep(pTHX_ OP *o)
2392 /* custom per-subroutine optimisation goes here */
2393 prev_peepp(aTHX_ o);
2394 /* custom per-subroutine optimisation may also go here */
2397 prev_peepp = PL_peepp;
2400 static peep_t prev_rpeepp;
2401 static void my_rpeep(pTHX_ OP *first)
2403 OP *o = first, *t = first;
2404 for(; o = o->op_next, t = t->op_next) {
2405 /* custom per-op optimisation goes here */
2407 if (!o || o == t) break;
2408 /* custom per-op optimisation goes AND here */
2410 prev_rpeepp(aTHX_ orig_o);
2413 prev_rpeepp = PL_rpeepp;
2414 PL_rpeepp = my_rpeep;
2416 =for apidoc Ayh||peep_t
2418 =head2 Pluggable runops
2420 The compile tree is executed in a runops function. There are two runops
2421 functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
2422 with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
2423 control over the execution of the compile tree it is possible to provide
2424 your own runops function.
2426 It's probably best to copy one of the existing runops functions and
2427 change it to suit your needs. Then, in the BOOT section of your XS
2430 PL_runops = my_runops;
2432 =for apidoc Amnh|runops_proc_t|PL_runops
2434 This function should be as efficient as possible to keep your programs
2435 running as fast as possible.
2437 =head2 Compile-time scope hooks
2439 As of perl 5.14 it is possible to hook into the compile-time lexical
2440 scope mechanism using C<Perl_blockhook_register>. This is used like
2443 STATIC void my_start_hook(pTHX_ int full);
2444 STATIC BHK my_hooks;
2447 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
2448 Perl_blockhook_register(aTHX_ &my_hooks);
2450 This will arrange to have C<my_start_hook> called at the start of
2451 compiling every lexical scope. The available hooks are:
2453 =for apidoc Ayh||BHK
2457 =item C<void bhk_start(pTHX_ int full)>
2459 This is called just after starting a new lexical scope. Note that Perl
2464 creates two scopes: the first starts at the C<(> and has C<full == 1>,
2465 the second starts at the C<{> and has C<full == 0>. Both end at the
2466 C<}>, so calls to C<start> and C<pre>/C<post_end> will match. Anything
2467 pushed onto the save stack by this hook will be popped just before the
2468 scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2470 =item C<void bhk_pre_end(pTHX_ OP **o)>
2472 This is called at the end of a lexical scope, just before unwinding the
2473 stack. I<o> is the root of the optree representing the scope; it is a
2474 double pointer so you can replace the OP if you need to.
2476 =item C<void bhk_post_end(pTHX_ OP **o)>
2478 This is called at the end of a lexical scope, just after unwinding the
2479 stack. I<o> is as above. Note that it is possible for calls to C<pre_>
2480 and C<post_end> to nest, if there is something on the save stack that
2483 =item C<void bhk_eval(pTHX_ OP *const o)>
2485 This is called just before starting to compile an C<eval STRING>, C<do
2486 FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
2487 OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2488 C<OP_DOFILE> or C<OP_REQUIRE>.
2492 Once you have your hook functions, you need a C<BHK> structure to put
2493 them in. It's best to allocate it statically, since there is no way to
2494 free it once it's registered. The function pointers should be inserted
2495 into this structure using the C<BhkENTRY_set> macro, which will also set
2496 flags indicating which entries are valid. If you do need to allocate
2497 your C<BHK> dynamically for some reason, be sure to zero it before you
2500 Once registered, there is no mechanism to switch these hooks off, so if
2501 that is necessary you will need to do this yourself. An entry in C<%^H>
2502 is probably the best way, so the effect is lexically scoped; however it
2503 is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
2504 temporarily switch entries on and off. You should also be aware that
2505 generally speaking at least one scope will have opened before your
2506 extension is loaded, so you will see some C<pre>/C<post_end> pairs that
2507 didn't have a matching C<start>.
2509 =head1 Examining internal data structures with the C<dump> functions
2511 To aid debugging, the source file F<dump.c> contains a number of
2512 functions which produce formatted output of internal data structures.
2514 The most commonly used of these functions is C<Perl_sv_dump>; it's used
2515 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
2516 C<sv_dump> to produce debugging output from Perl-space, so users of that
2517 module should already be familiar with its format.
2519 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
2520 derivatives, and produces output similar to C<perl -Dx>; in fact,
2521 C<Perl_dump_eval> will dump the main root of the code being evaluated,
2522 exactly like C<-Dx>.
2524 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2525 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2526 subroutines in a package like so: (Thankfully, these are all xsubs, so
2527 there is no op tree)
2529 (gdb) print Perl_dump_packsubs(PL_defstash)
2531 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2533 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2535 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2537 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2539 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2541 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2542 the op tree of the main root.
2544 =head1 How multiple interpreters and concurrency are supported
2546 =head2 Background and MULTIPLICITY
2548 The Perl interpreter can be regarded as a closed box: it has an API
2549 for feeding it code or otherwise making it do things, but it also has
2550 functions for its own use. This smells a lot like an object, and
2551 there is a way for you to build Perl so that you can have multiple
2552 interpreters, with one interpreter represented either as a C structure,
2553 or inside a thread-specific structure. These structures contain all
2554 the context, the state of that interpreter.
2556 The macro that controls the major Perl build flavor is MULTIPLICITY. The
2557 MULTIPLICITY build has a C structure that packages all the interpreter
2558 state, which is being passed to various perl functions as a "hidden"
2559 first argument. MULTIPLICITY makes multi-threaded perls possible (with the
2560 ithreads threading model, related to the macro USE_ITHREADS.)
2562 PERL_IMPLICIT_CONTEXT is a legacy synonym for MULTIPLICITY.
2564 To see whether you have non-const data you can use a BSD (or GNU)
2567 nm libperl.a | grep -v ' [TURtr] '
2569 If this displays any C<D> or C<d> symbols (or possibly C<C> or C<c>),
2570 you have non-const data. The symbols the C<grep> removed are as follows:
2571 C<Tt> are I<text>, or code, the C<Rr> are I<read-only> (const) data,
2572 and the C<U> is <undefined>, external symbols referred to.
2574 The test F<t/porting/libperl.t> does this kind of symbol sanity
2575 checking on C<libperl.a>.
2577 All this obviously requires a way for the Perl internal functions to be
2578 either subroutines taking some kind of structure as the first
2579 argument, or subroutines taking nothing as the first argument. To
2580 enable these two very different ways of building the interpreter,
2581 the Perl source (as it does in so many other situations) makes heavy
2582 use of macros and subroutine naming conventions.
2584 First problem: deciding which functions will be public API functions and
2585 which will be private. All functions whose names begin C<S_> are private
2586 (think "S" for "secret" or "static"). All other functions begin with
2587 "Perl_", but just because a function begins with "Perl_" does not mean it is
2588 part of the API. (See L</Internal
2589 Functions>.) The easiest way to be B<sure> a
2590 function is part of the API is to find its entry in L<perlapi>.
2591 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2592 think it should be (i.e., you need it for your extension), submit an issue at
2593 L<https://github.com/Perl/perl5/issues> explaining why you think it should be.
2595 Second problem: there must be a syntax so that the same subroutine
2596 declarations and calls can pass a structure as their first argument,
2597 or pass nothing. To solve this, the subroutines are named and
2598 declared in a particular way. Here's a typical start of a static
2599 function used within the Perl guts:
2602 S_incline(pTHX_ char *s)
2604 STATIC becomes "static" in C, and may be #define'd to nothing in some
2605 configurations in the future.
2607 =for apidoc_section $directives
2608 =for apidoc Ayh||STATIC
2610 A public function (i.e. part of the internal API, but not necessarily
2611 sanctioned for use in extensions) begins like this:
2614 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
2616 C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
2617 details of the interpreter's context. THX stands for "thread", "this",
2618 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2619 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
2620 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2623 =for apidoc_section $concurrency
2624 =for apidoc Amnh||aTHX
2625 =for apidoc Amnh||aTHX_
2626 =for apidoc Amnh||dTHX
2627 =for apidoc Amnh||pTHX
2628 =for apidoc Amnh||pTHX_
2630 When Perl is built without options that set MULTIPLICITY, there is no
2631 first argument containing the interpreter's context. The trailing underscore
2632 in the pTHX_ macro indicates that the macro expansion needs a comma
2633 after the context argument because other arguments follow it. If
2634 MULTIPLICITY is not defined, pTHX_ will be ignored, and the
2635 subroutine is not prototyped to take the extra argument. The form of the
2636 macro without the trailing underscore is used when there are no additional
2639 When a core function calls another, it must pass the context. This
2640 is normally hidden via macros. Consider C<sv_setiv>. It expands into
2641 something like this:
2644 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
2645 /* can't do this for vararg functions, see below */
2647 #define sv_setiv Perl_sv_setiv
2650 This works well, and means that XS authors can gleefully write:
2654 and still have it work under all the modes Perl could have been
2657 This doesn't work so cleanly for varargs functions, though, as macros
2658 imply that the number of arguments is known in advance. Instead we
2659 either need to spell them out fully, passing C<aTHX_> as the first
2660 argument (the Perl core tends to do this with functions like
2661 Perl_warner), or use a context-free version.
2663 The context-free version of Perl_warner is called
2664 Perl_warner_nocontext, and does not take the extra argument. Instead
2665 it does C<dTHX;> to get the context from thread-local storage. We
2666 C<#define warner Perl_warner_nocontext> so that extensions get source
2667 compatibility at the expense of performance. (Passing an arg is
2668 cheaper than grabbing it from thread-local storage.)
2670 You can ignore [pad]THXx when browsing the Perl headers/sources.
2671 Those are strictly for use within the core. Extensions and embedders
2672 need only be aware of [pad]THX.
2674 =head2 So what happened to dTHR?
2676 =for apidoc Amnh||dTHR
2678 C<dTHR> was introduced in perl 5.005 to support the older thread model.
2679 The older thread model now uses the C<THX> mechanism to pass context
2680 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2681 later still have it for backward source compatibility, but it is defined
2684 =head2 How do I use all this in extensions?
2686 When Perl is built with MULTIPLICITY, extensions that call
2687 any functions in the Perl API will need to pass the initial context
2688 argument somehow. The kicker is that you will need to write it in
2689 such a way that the extension still compiles when Perl hasn't been
2690 built with MULTIPLICITY enabled.
2692 There are three ways to do this. First, the easy but inefficient way,
2693 which is also the default, in order to maintain source compatibility
2694 with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
2695 and aTHX_ macros to call a function that will return the context.
2696 Thus, something like:
2700 in your extension will translate to this when MULTIPLICITY is
2703 Perl_sv_setiv(Perl_get_context(), sv, num);
2705 or to this otherwise:
2707 Perl_sv_setiv(sv, num);
2709 You don't have to do anything new in your extension to get this; since
2710 the Perl library provides Perl_get_context(), it will all just
2713 The second, more efficient way is to use the following template for
2716 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2721 STATIC void my_private_function(int arg1, int arg2);
2724 my_private_function(int arg1, int arg2)
2726 dTHX; /* fetch context */
2727 ... call many Perl API functions ...
2732 MODULE = Foo PACKAGE = Foo
2740 my_private_function(arg, 10);
2742 Note that the only two changes from the normal way of writing an
2743 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2744 including the Perl headers, followed by a C<dTHX;> declaration at
2745 the start of every function that will call the Perl API. (You'll
2746 know which functions need this, because the C compiler will complain
2747 that there's an undeclared identifier in those functions.) No changes
2748 are needed for the XSUBs themselves, because the XS() macro is
2749 correctly defined to pass in the implicit context if needed.
2751 The third, even more efficient way is to ape how it is done within
2755 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2760 /* pTHX_ only needed for functions that call Perl API */
2761 STATIC void my_private_function(pTHX_ int arg1, int arg2);
2764 my_private_function(pTHX_ int arg1, int arg2)
2766 /* dTHX; not needed here, because THX is an argument */
2767 ... call Perl API functions ...
2772 MODULE = Foo PACKAGE = Foo
2780 my_private_function(aTHX_ arg, 10);
2782 This implementation never has to fetch the context using a function
2783 call, since it is always passed as an extra argument. Depending on
2784 your needs for simplicity or efficiency, you may mix the previous
2785 two approaches freely.
2787 Never add a comma after C<pTHX> yourself--always use the form of the
2788 macro with the underscore for functions that take explicit arguments,
2789 or the form without the argument for functions with no explicit arguments.
2791 =head2 Should I do anything special if I call perl from multiple threads?
2793 If you create interpreters in one thread and then proceed to call them in
2794 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2795 initialized correctly in each of those threads.
2797 The C<perl_alloc> and C<perl_clone> API functions will automatically set
2798 the TLS slot to the interpreter they created, so that there is no need to do
2799 anything special if the interpreter is always accessed in the same thread that
2800 created it, and that thread did not create or call any other interpreters
2801 afterwards. If that is not the case, you have to set the TLS slot of the
2802 thread before calling any functions in the Perl API on that particular
2803 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2804 thread as the first thing you do:
2806 /* do this before doing anything else with some_perl */
2807 PERL_SET_CONTEXT(some_perl);
2809 ... other Perl API calls on some_perl go here ...
2811 =head2 Future Plans and PERL_IMPLICIT_SYS
2813 Just as MULTIPLICITY provides a way to bundle up everything
2814 that the interpreter knows about itself and pass it around, so too are
2815 there plans to allow the interpreter to bundle up everything it knows
2816 about the environment it's running on. This is enabled with the
2817 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2820 This allows the ability to provide an extra pointer (called the "host"
2821 environment) for all the system calls. This makes it possible for
2822 all the system stuff to maintain their own state, broken down into
2823 seven C structures. These are thin wrappers around the usual system
2824 calls (see F<win32/perllib.c>) for the default perl executable, but for a
2825 more ambitious host (like the one that would do fork() emulation) all
2826 the extra work needed to pretend that different interpreters are
2827 actually different "processes", would be done here.
2829 The Perl engine/interpreter and the host are orthogonal entities.
2830 There could be one or more interpreters in a process, and one or
2831 more "hosts", with free association between them.
2833 =head1 Internal Functions
2835 All of Perl's internal functions which will be exposed to the outside
2836 world are prefixed by C<Perl_> so that they will not conflict with XS
2837 functions or functions used in a program in which Perl is embedded.
2838 Similarly, all global variables begin with C<PL_>. (By convention,
2839 static functions start with C<S_>.)
2841 Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2842 either with or without the C<Perl_> prefix, thanks to a bunch of defines
2843 that live in F<embed.h>. Note that extension code should I<not> set
2844 C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2845 breakage of the XS in each new perl release.
2847 The file F<embed.h> is generated automatically from
2848 F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2849 header files for the internal functions, generates the documentation
2850 and a lot of other bits and pieces. It's important that when you add
2851 a new function to the core or change an existing one, you change the
2852 data in the table in F<embed.fnc> as well. Here's a sample entry from
2855 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2857 The first column is a set of flags, the second column the return type,
2858 the third column the name. Columns after that are the arguments.
2859 The flags are documented at the top of F<embed.fnc>.
2861 If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2862 C<make regen_headers> to force a rebuild of F<embed.h> and other
2863 auto-generated files.
2865 =head2 Formatted Printing of IVs, UVs, and NVs
2867 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2868 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2869 following macros for portability
2874 UVxf UV in hexadecimal
2879 These will take care of 64-bit integers and long doubles.
2882 printf("IV is %" IVdf "\n", iv);
2884 The C<IVdf> will expand to whatever is the correct format for the IVs.
2885 Note that the spaces are required around the format in case the code is
2886 compiled with C++, to maintain compliance with its standard.
2888 Note that there are different "long doubles": Perl will use
2889 whatever the compiler has.
2891 If you are printing addresses of pointers, use %p or UVxf combined
2894 =head2 Formatted Printing of SVs
2896 The contents of SVs may be printed using the C<SVf> format, like so:
2898 Perl_croak(aTHX_ "This croaked because: %" SVf "\n", SVfARG(err_msg))
2900 where C<err_msg> is an SV.
2902 =for apidoc Amnh||SVf
2903 =for apidoc Amh||SVfARG|SV *sv
2905 Not all scalar types are printable. Simple values certainly are: one of
2906 IV, UV, NV, or PV. Also, if the SV is a reference to some value,
2907 either it will be dereferenced and the value printed, or information
2908 about the type of that value and its address are displayed. The results
2909 of printing any other type of SV are undefined and likely to lead to an
2910 interpreter crash. NVs are printed using a C<%g>-ish format.
2912 Note that the spaces are required around the C<SVf> in case the code is
2913 compiled with C++, to maintain compliance with its standard.
2915 Note that any filehandle being printed to under UTF-8 must be expecting
2916 UTF-8 in order to get good results and avoid Wide-character warnings.
2917 One way to do this for typical filehandles is to invoke perl with the
2918 C<-C>> parameter. (See L<perlrun/-C [numberE<sol>list]>.
2920 You can use this to concatenate two scalars:
2922 SV *var1 = get_sv("var1", GV_ADD);
2923 SV *var2 = get_sv("var2", GV_ADD);
2924 SV *var3 = newSVpvf("var1=%" SVf " and var2=%" SVf,
2925 SVfARG(var1), SVfARG(var2));
2927 =head2 Formatted Printing of Strings
2929 If you just want the bytes printed in a 7bit NUL-terminated string, you can
2930 just use C<%s> (assuming they are all really only 7bit). But if there is a
2931 possibility the value will be encoded as UTF-8 or contains bytes above
2932 C<0x7F> (and therefore 8bit), you should instead use the C<UTF8f> format.
2933 And as its parameter, use the C<UTF8fARG()> macro:
2937 /* U+2018: \xE2\x80\x98 LEFT SINGLE QUOTATION MARK
2938 U+2019: \xE2\x80\x99 RIGHT SINGLE QUOTATION MARK */
2940 msg = "\xE2\x80\x98Uses fancy quotes\xE2\x80\x99";
2942 msg = "'Uses simple quotes'";
2944 Perl_croak(aTHX_ "The message is: %" UTF8f "\n",
2945 UTF8fARG(can_utf8, strlen(msg), msg));
2947 The first parameter to C<UTF8fARG> is a boolean: 1 if the string is in
2948 UTF-8; 0 if string is in native byte encoding (Latin1).
2949 The second parameter is the number of bytes in the string to print.
2950 And the third and final parameter is a pointer to the first byte in the
2953 Note that any filehandle being printed to under UTF-8 must be expecting
2954 UTF-8 in order to get good results and avoid Wide-character warnings.
2955 One way to do this for typical filehandles is to invoke perl with the
2956 C<-C>> parameter. (See L<perlrun/-C [numberE<sol>list]>.
2958 =for apidoc_section $formats
2959 =for apidoc Amnh||UTF8f
2960 =for apidoc Amh||UTF8fARG|bool is_utf8|Size_t byte_len|char *str
2964 =head2 Formatted Printing of C<Size_t> and C<SSize_t>
2966 The most general way to do this is to cast them to a UV or IV, and
2968 L<previous section|/Formatted Printing of IVs, UVs, and NVs>.
2970 But if you're using C<PerlIO_printf()>, it's less typing and visual
2971 clutter to use the C<%z> length modifier (for I<siZe>):
2973 PerlIO_printf("STRLEN is %zu\n", len);
2975 This modifier is not portable, so its use should be restricted to
2978 =head2 Formatted Printing of C<Ptrdiff_t>, C<intmax_t>, C<short> and other special sizes
2980 There are modifiers for these special situations if you are using
2981 C<PerlIO_printf()>. See L<perlfunc/size>.
2983 =head2 Pointer-To-Integer and Integer-To-Pointer
2985 Because pointer size does not necessarily equal integer size,
2986 use the follow macros to do it right.
2991 INT2PTR(pointertotype, integer)
2993 =for apidoc_section $casting
2994 =for apidoc Amh|type|INT2PTR|type|int value
2995 =for apidoc Amh|UV|PTR2UV|void * ptr
2996 =for apidoc Amh|IV|PTR2IV|void * ptr
2997 =for apidoc Amh|NV|PTR2NV|void * ptr
3002 SV *sv = INT2PTR(SV*, iv);
3011 PTR2nat(pointer) /* pointer to integer of PTRSIZE */
3012 PTR2ul(pointer) /* pointer to unsigned long */
3014 =for apidoc Amh|IV|PTR2nat|void *
3015 =for apidoc Amh|unsigned long|PTR2ul|void *
3017 And C<PTRV> which gives the native type for an integer the same size as
3018 pointers, such as C<unsigned> or C<unsigned long>.
3020 =for apidoc Ayh|type|PTRV
3022 =head2 Exception Handling
3024 There are a couple of macros to do very basic exception handling in XS
3025 modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
3026 be able to use these macros:
3031 You can use these macros if you call code that may croak, but you need
3032 to do some cleanup before giving control back to Perl. For example:
3034 dXCPT; /* set up necessary variables */
3037 code_that_may_croak();
3042 /* do cleanup here */
3046 Note that you always have to rethrow an exception that has been
3047 caught. Using these macros, it is not possible to just catch the
3048 exception and ignore it. If you have to ignore the exception, you
3049 have to use the C<call_*> function.
3051 The advantage of using the above macros is that you don't have
3052 to setup an extra function for C<call_*>, and that using these
3053 macros is faster than using C<call_*>.
3055 =head2 Source Documentation
3057 There's an effort going on to document the internal functions and
3058 automatically produce reference manuals from them -- L<perlapi> is one
3059 such manual which details all the functions which are available to XS
3060 writers. L<perlintern> is the autogenerated manual for the functions
3061 which are not part of the API and are supposedly for internal use only.
3063 Source documentation is created by putting POD comments into the C
3067 =for apidoc sv_setiv
3069 Copies an integer into the given SV. Does not handle 'set' magic. See
3070 L<perlapi/sv_setiv_mg>.
3075 Please try and supply some documentation if you add functions to the
3078 =head2 Backwards compatibility
3080 The Perl API changes over time. New functions are
3081 added or the interfaces of existing functions are
3082 changed. The C<Devel::PPPort> module tries to
3083 provide compatibility code for some of these changes, so XS writers don't
3084 have to code it themselves when supporting multiple versions of Perl.
3086 C<Devel::PPPort> generates a C header file F<ppport.h> that can also
3087 be run as a Perl script. To generate F<ppport.h>, run:
3089 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
3091 Besides checking existing XS code, the script can also be used to retrieve
3092 compatibility information for various API calls using the C<--api-info>
3093 command line switch. For example:
3095 % perl ppport.h --api-info=sv_magicext
3097 For details, see C<perldoc ppport.h>.
3099 =head1 Unicode Support
3101 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
3102 writers to understand this support and make sure that the code they
3103 write does not corrupt Unicode data.
3105 =head2 What B<is> Unicode, anyway?
3107 In the olden, less enlightened times, we all used to use ASCII. Most of
3108 us did, anyway. The big problem with ASCII is that it's American. Well,
3109 no, that's not actually the problem; the problem is that it's not
3110 particularly useful for people who don't use the Roman alphabet. What
3111 used to happen was that particular languages would stick their own
3112 alphabet in the upper range of the sequence, between 128 and 255. Of
3113 course, we then ended up with plenty of variants that weren't quite
3114 ASCII, and the whole point of it being a standard was lost.
3116 Worse still, if you've got a language like Chinese or
3117 Japanese that has hundreds or thousands of characters, then you really
3118 can't fit them into a mere 256, so they had to forget about ASCII
3119 altogether, and build their own systems using pairs of numbers to refer
3122 To fix this, some people formed Unicode, Inc. and
3123 produced a new character set containing all the characters you can
3124 possibly think of and more. There are several ways of representing these
3125 characters, and the one Perl uses is called UTF-8. UTF-8 uses
3126 a variable number of bytes to represent a character. You can learn more
3127 about Unicode and Perl's Unicode model in L<perlunicode>.
3129 (On EBCDIC platforms, Perl uses instead UTF-EBCDIC, which is a form of
3130 UTF-8 adapted for EBCDIC platforms. Below, we just talk about UTF-8.
3131 UTF-EBCDIC is like UTF-8, but the details are different. The macros
3132 hide the differences from you, just remember that the particular numbers
3133 and bit patterns presented below will differ in UTF-EBCDIC.)
3135 =head2 How can I recognise a UTF-8 string?
3137 You can't. This is because UTF-8 data is stored in bytes just like
3138 non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
3139 capital E with a grave accent, is represented by the two bytes
3140 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
3141 has that byte sequence as well. So you can't tell just by looking -- this
3142 is what makes Unicode input an interesting problem.
3144 In general, you either have to know what you're dealing with, or you
3145 have to guess. The API function C<is_utf8_string> can help; it'll tell
3146 you if a string contains only valid UTF-8 characters, and the chances
3147 of a non-UTF-8 string looking like valid UTF-8 become very small very
3148 quickly with increasing string length. On a character-by-character
3149 basis, C<isUTF8_CHAR>
3150 will tell you whether the current character in a string is valid UTF-8.
3152 =head2 How does UTF-8 represent Unicode characters?
3154 As mentioned above, UTF-8 uses a variable number of bytes to store a
3155 character. Characters with values 0...127 are stored in one
3156 byte, just like good ol' ASCII. Character 128 is stored as
3157 C<v194.128>; this continues up to character 191, which is
3158 C<v194.191>. Now we've run out of bits (191 is binary
3159 C<10111111>) so we move on; character 192 is C<v195.128>. And
3160 so it goes on, moving to three bytes at character 2048.
3161 L<perlunicode/Unicode Encodings> has pictures of how this works.
3163 Assuming you know you're dealing with a UTF-8 string, you can find out
3164 how long the first character in it is with the C<UTF8SKIP> macro:
3166 char *utf = "\305\233\340\240\201";
3169 len = UTF8SKIP(utf); /* len is 2 here */
3171 len = UTF8SKIP(utf); /* len is 3 here */
3173 Another way to skip over characters in a UTF-8 string is to use
3174 C<utf8_hop>, which takes a string and a number of characters to skip
3175 over. You're on your own about bounds checking, though, so don't use it
3178 All bytes in a multi-byte UTF-8 character will have the high bit set,
3179 so you can test if you need to do something special with this
3180 character like this (the C<UTF8_IS_INVARIANT()> is a macro that tests
3181 whether the byte is encoded as a single byte even in UTF-8):
3183 U8 *utf; /* Initialize this to point to the beginning of the
3184 sequence to convert */
3185 U8 *utf_end; /* Initialize this to 1 beyond the end of the sequence
3186 pointed to by 'utf' */
3187 UV uv; /* Returned code point; note: a UV, not a U8, not a
3189 STRLEN len; /* Returned length of character in bytes */
3191 if (!UTF8_IS_INVARIANT(*utf))
3192 /* Must treat this as UTF-8 */
3193 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
3195 /* OK to treat this character as a byte */
3198 You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
3199 value of the character; the inverse function C<uvchr_to_utf8> is available
3200 for putting a UV into UTF-8:
3202 if (!UVCHR_IS_INVARIANT(uv))
3203 /* Must treat this as UTF8 */
3204 utf8 = uvchr_to_utf8(utf8, uv);
3206 /* OK to treat this character as a byte */
3209 You B<must> convert characters to UVs using the above functions if
3210 you're ever in a situation where you have to match UTF-8 and non-UTF-8
3211 characters. You may not skip over UTF-8 characters in this case. If you
3212 do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
3213 for instance, if your UTF-8 string contains C<v196.172>, and you skip
3214 that character, you can never match a C<chr(200)> in a non-UTF-8 string.
3217 (Note that we don't have to test for invariant characters in the
3218 examples above. The functions work on any well-formed UTF-8 input.
3219 It's just that its faster to avoid the function overhead when it's not
3222 =head2 How does Perl store UTF-8 strings?
3224 Currently, Perl deals with UTF-8 strings and non-UTF-8 strings
3225 slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
3226 string is internally encoded as UTF-8. Without it, the byte value is the
3227 codepoint number and vice versa. This flag is only meaningful if the SV
3228 is C<SvPOK> or immediately after stringification via C<SvPV> or a
3229 similar macro. You can check and manipulate this flag with the
3236 This flag has an important effect on Perl's treatment of the string: if
3237 UTF-8 data is not properly distinguished, regular expressions,
3238 C<length>, C<substr> and other string handling operations will have
3239 undesirable (wrong) results.
3241 The problem comes when you have, for instance, a string that isn't
3242 flagged as UTF-8, and contains a byte sequence that could be UTF-8 --
3243 especially when combining non-UTF-8 and UTF-8 strings.
3245 Never forget that the C<SVf_UTF8> flag is separate from the PV value; you
3246 need to be sure you don't accidentally knock it off while you're
3247 manipulating SVs. More specifically, you cannot expect to do this:
3256 nsv = newSVpvn(p, len);
3258 The C<char*> string does not tell you the whole story, and you can't
3259 copy or reconstruct an SV just by copying the string value. Check if the
3260 old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
3264 is_utf8 = SvUTF8(sv);
3265 frobnicate(p, is_utf8);
3266 nsv = newSVpvn(p, len);
3270 In the above, your C<frobnicate> function has been changed to be made
3271 aware of whether or not it's dealing with UTF-8 data, so that it can
3272 handle the string appropriately.
3274 Since just passing an SV to an XS function and copying the data of
3275 the SV is not enough to copy the UTF8 flags, even less right is just
3276 passing a S<C<char *>> to an XS function.
3278 For full generality, use the L<C<DO_UTF8>|perlapi/DO_UTF8> macro to see if the
3279 string in an SV is to be I<treated> as UTF-8. This takes into account
3280 if the call to the XS function is being made from within the scope of
3281 L<S<C<use bytes>>|bytes>. If so, the underlying bytes that comprise the
3282 UTF-8 string are to be exposed, rather than the character they
3283 represent. But this pragma should only really be used for debugging and
3284 perhaps low-level testing at the byte level. Hence most XS code need
3285 not concern itself with this, but various areas of the perl core do need
3288 And this isn't the whole story. Starting in Perl v5.12, strings that
3289 aren't encoded in UTF-8 may also be treated as Unicode under various
3290 conditions (see L<perlunicode/ASCII Rules versus Unicode Rules>).
3291 This is only really a problem for characters whose ordinals are between
3292 128 and 255, and their behavior varies under ASCII versus Unicode rules
3293 in ways that your code cares about (see L<perlunicode/The "Unicode Bug">).
3294 There is no published API for dealing with this, as it is subject to
3295 change, but you can look at the code for C<pp_lc> in F<pp.c> for an
3296 example as to how it's currently done.
3298 =head2 How do I pass a Perl string to a C library?
3300 A Perl string, conceptually, is an opaque sequence of code points.
3301 Many C libraries expect their inputs to be "classical" C strings, which are
3302 arrays of octets 1-255, terminated with a NUL byte. Your job when writing
3303 an interface between Perl and a C library is to define the mapping between
3304 Perl and that library.
3306 Generally speaking, C<SvPVbyte> and related macros suit this task well.
3307 These assume that your Perl string is a "byte string", i.e., is either
3308 raw, undecoded input into Perl or is pre-encoded to, e.g., UTF-8.
3310 Alternatively, if your C library expects UTF-8 text, you can use
3311 C<SvPVutf8> and related macros. This has the same effect as encoding
3312 to UTF-8 then calling the corresponding C<SvPVbyte>-related macro.
3314 Some C libraries may expect other encodings (e.g., UTF-16LE). To give
3315 Perl strings to such libraries
3316 you must either do that encoding in Perl then use C<SvPVbyte>, or
3317 use an intermediary C library to convert from however Perl stores the
3318 string to the desired encoding.
3320 Take care also that NULs in your Perl string don't confuse the C
3321 library. If possible, give the string's length to the C library; if that's
3322 not possible, consider rejecting strings that contain NUL bytes.
3324 =head3 What about C<SvPV>, C<SvPV_nolen>, etc.?
3326 Consider a 3-character Perl string C<$foo = "\x64\x78\x8c">.
3327 Perl can store these 3 characters either of two ways:
3331 =item * bytes: 0x64 0x78 0x8c
3333 =item * UTF-8: 0x64 0x78 0xc2 0x8c
3337 Now let's say you convert C<$foo> to a C string thus:
3340 char *str = SvPV(foo_sv, strlen);
3342 At this point C<str> could point to a 3-byte C string or a 4-byte one.
3344 Generally speaking, we want C<str> to be the same regardless of how
3345 Perl stores C<$foo>, so the ambiguity here is undesirable. C<SvPVbyte>
3346 and C<SvPVutf8> solve that by giving predictable output: use
3347 C<SvPVbyte> if your C library expects byte strings, or C<SvPVutf8>
3348 if it expects UTF-8.
3350 If your C library happens to support both encodings, then C<SvPV>--always
3351 in tandem with lookups to C<SvUTF8>!--may be safe and (slightly) more
3354 B<TESTING> B<TIP:> Use L<utf8>'s C<upgrade> and C<downgrade> functions
3355 in your tests to ensure consistent handling regardless of Perl's
3358 =head2 How do I convert a string to UTF-8?
3360 If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
3361 the non-UTF-8 strings to UTF-8. If you've got an SV, the easiest way to do
3364 sv_utf8_upgrade(sv);
3366 However, you must not do this, for example:
3369 sv_utf8_upgrade(left);
3371 If you do this in a binary operator, you will actually change one of the
3372 strings that came into the operator, and, while it shouldn't be noticeable
3373 by the end user, it can cause problems in deficient code.
3375 Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
3376 string argument. This is useful for having the data available for
3377 comparisons and so on, without harming the original SV. There's also
3378 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
3379 the string contains any characters above 255 that can't be represented
3382 =head2 How do I compare strings?
3384 L<perlapi/sv_cmp> and L<perlapi/sv_cmp_flags> do a lexigraphic
3385 comparison of two SV's, and handle UTF-8ness properly. Note, however,
3386 that Unicode specifies a much fancier mechanism for collation, available
3387 via the L<Unicode::Collate> module.
3389 To just compare two strings for equality/non-equality, you can just use
3390 L<C<memEQ()>|perlapi/memEQ> and L<C<memNE()>|perlapi/memEQ> as usual,
3391 except the strings must be both UTF-8 or not UTF-8 encoded.
3393 To compare two strings case-insensitively, use
3394 L<C<foldEQ_utf8()>|perlapi/foldEQ_utf8> (the strings don't have to have
3395 the same UTF-8ness).
3397 =head2 Is there anything else I need to know?
3399 Not really. Just remember these things:
3405 There's no way to tell if a S<C<char *>> or S<C<U8 *>> string is UTF-8
3406 or not. But you can tell if an SV is to be treated as UTF-8 by calling
3407 C<DO_UTF8> on it, after stringifying it with C<SvPV> or a similar
3408 macro. And, you can tell if SV is actually UTF-8 (even if it is not to
3409 be treated as such) by looking at its C<SvUTF8> flag (again after
3410 stringifying it). Don't forget to set the flag if something should be
3412 Treat the flag as part of the PV, even though it's not -- if you pass on
3413 the PV to somewhere, pass on the flag too.
3417 If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3418 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
3422 When writing a character UV to a UTF-8 string, B<always> use
3423 C<uvchr_to_utf8>, unless C<UVCHR_IS_INVARIANT(uv))> in which case
3424 you can use C<*s = uv>.
3428 Mixing UTF-8 and non-UTF-8 strings is
3429 tricky. Use C<bytes_to_utf8> to get
3430 a new string which is UTF-8 encoded, and then combine them.
3434 =head1 Custom Operators
3436 Custom operator support is an experimental feature that allows you to
3437 define your own ops. This is primarily to allow the building of
3438 interpreters for other languages in the Perl core, but it also allows
3439 optimizations through the creation of "macro-ops" (ops which perform the
3440 functions of multiple ops which are usually executed together, such as
3441 C<gvsv, gvsv, add>.)
3443 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
3444 core does not "know" anything special about this op type, and so it will
3445 not be involved in any optimizations. This also means that you can
3446 define your custom ops to be any op structure -- unary, binary, list and
3449 It's important to know what custom operators won't do for you. They
3450 won't let you add new syntax to Perl, directly. They won't even let you
3451 add new keywords, directly. In fact, they won't change the way Perl
3452 compiles a program at all. You have to do those changes yourself, after
3453 Perl has compiled the program. You do this either by manipulating the op
3454 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
3455 a custom peephole optimizer with the C<optimize> module.
3457 When you do this, you replace ordinary Perl ops with custom ops by
3458 creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
3459 PP function. This should be defined in XS code, and should look like
3460 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
3461 takes the appropriate number of values from the stack, and you are
3462 responsible for adding stack marks if necessary.
3464 You should also "register" your op with the Perl interpreter so that it
3465 can produce sensible error and warning messages. Since it is possible to
3466 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
3467 Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
3468 it is dealing with. You should create an C<XOP> structure for each
3469 ppaddr you use, set the properties of the custom op with
3470 C<XopENTRY_set>, and register the structure against the ppaddr using
3471 C<Perl_custom_op_register>. A trivial example might look like:
3473 =for apidoc Ayh||XOP
3476 static OP *my_pp(pTHX);
3479 XopENTRY_set(&my_xop, xop_name, "myxop");
3480 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
3481 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
3483 The available fields in the structure are:
3489 A short name for your op. This will be included in some error messages,
3490 and will also be returned as C<< $op->name >> by the L<B|B> module, so
3491 it will appear in the output of module like L<B::Concise|B::Concise>.
3495 A short description of the function of the op.
3499 Which of the various C<*OP> structures this op uses. This should be one of
3500 the C<OA_*> constants from F<op.h>, namely
3520 =item OA_PVOP_OR_SVOP
3522 This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
3523 the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
3531 The other C<OA_*> constants should not be used.
3535 This member is of type C<Perl_cpeep_t>, which expands to C<void
3536 (*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
3537 will be called from C<Perl_rpeep> when ops of this type are encountered
3538 by the peephole optimizer. I<o> is the OP that needs optimizing;
3539 I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
3541 =for apidoc Ayh||Perl_cpeep_t
3545 C<B::Generate> directly supports the creation of custom ops by name.
3549 Descriptions above occasionally refer to "the stack", but there are in fact
3550 many stack-like data structures within the perl interpreter. When otherwise
3551 unqualified, "the stack" usually refers to the value stack.
3553 The various stacks have different purposes, and operate in slightly different
3554 ways. Their differences are noted below.
3558 This stack stores the values that regular perl code is operating on, usually
3559 intermediate values of expressions within a statement. The stack itself is
3560 formed of an array of SV pointers.
3562 The base of this stack is pointed to by the interpreter variable
3563 C<PL_stack_base>, of type C<SV **>.
3565 The head of the stack is C<PL_stack_sp>, and points to the most
3566 recently-pushed item.
3568 Items are pushed to the stack by using the C<PUSHs()> macro or its variants
3569 described above; C<XPUSHs()>, C<mPUSHs()>, C<mXPUSHs()> and the typed
3570 versions. Note carefully that the non-C<X> versions of these macros do not
3571 check the size of the stack and assume it to be big enough. These must be
3572 paired with a suitable check of the stack's size, such as the C<EXTEND> macro
3573 to ensure it is large enough. For example
3581 This is slightly more performant than making four separate checks in four
3582 separate C<mXPUSHi()> calls.
3584 As a further performance optimisation, the various C<PUSH> macros all operate
3585 using a local variable C<SP>, rather than the interpreter-global variable
3586 C<PL_stack_sp>. This variable is declared by the C<dSP> macro - though it is
3587 normally implied by XSUBs and similar so it is rare you have to consider it
3588 directly. Once declared, the C<PUSH> macros will operate only on this local
3589 variable, so before invoking any other perl core functions you must use the
3590 C<PUTBACK> macro to return the value from the local C<SP> variable back to
3591 the interpreter variable. Similarly, after calling a perl core function which
3592 may have had reason to move the stack or push/pop values to it, you must use
3593 the C<SPAGAIN> macro which refreshes the local C<SP> value back from the
3596 Items are popped from the stack by using the C<POPs> macro or its typed
3597 versions, There is also a macro C<TOPs> that inspects the topmost item without
3600 Note specifically that SV pointers on the value stack do not contribute to the
3601 overall reference count of the xVs being referred to. If newly-created xVs are
3602 being pushed to the stack you must arrange for them to be destroyed at a
3603 suitable time; usually by using one of the C<mPUSH*> macros or C<sv_2mortal()>
3604 to mortalise the xV.
3608 The value stack stores individual perl scalar values as temporaries between
3609 expressions. Some perl expressions operate on entire lists; for that purpose
3610 we need to know where on the stack each list begins. This is the purpose of the
3613 The mark stack stores integers as I32 values, which are the height of the
3614 value stack at the time before the list began; thus the mark itself actually
3615 points to the value stack entry one before the list. The list itself starts at
3618 The base of this stack is pointed to by the interpreter variable
3619 C<PL_markstack>, of type C<I32 *>.
3621 The head of the stack is C<PL_markstack_ptr>, and points to the most
3622 recently-pushed item.
3624 Items are pushed to the stack by using the C<PUSHMARK()> macro. Even though
3625 the stack itself stores (value) stack indices as integers, the C<PUSHMARK>
3626 macro should be given a stack pointer directly; it will calculate the index
3627 offset by comparing to the C<PL_stack_sp> variable. Thus almost always the
3628 code to perform this is
3632 Items are popped from the stack by the C<POPMARK> macro. There is also a macro
3633 C<TOPMARK> that inspects the topmost item without removing it. These macros
3634 return I32 index values directly. There is also the C<dMARK> macro which
3635 declares a new SV double-pointer variable, called C<mark>, which points at the
3636 marked stack slot; this is the usual macro that C code will use when operating
3637 on lists given on the stack.
3639 As noted above, the C<mark> variable itself will point at the most recently
3640 pushed value on the value stack before the list begins, and so the list itself
3641 starts at C<mark + 1>. The values of the list may be iterated by code such as
3643 for(SV **svp = mark + 1; svp <= PL_stack_sp; svp++) {
3648 Note specifically in the case that the list is already empty, C<mark> will
3649 equal C<PL_stack_sp>.
3651 Because the C<mark> variable is converted to a pointer on the value stack,
3652 extra care must be taken if C<EXTEND> or any of the C<XPUSH> macros are
3653 invoked within the function, because the stack may need to be moved to
3654 extend it and so the existing pointer will now be invalid. If this may be a
3655 problem, a possible solution is to track the mark offset as an integer and
3656 track the mark itself later on after the stack had been moved.
3658 I32 markoff = POPMARK;
3662 SP **mark = PL_stack_base + markoff;
3664 =head2 Temporaries Stack
3666 As noted above, xV references on the main value stack do not contribute to the
3667 reference count of an xV, and so another mechanism is used to track when
3668 temporary values which live on the stack must be released. This is the job of
3669 the temporaries stack.
3671 The temporaries stack stores pointers to xVs whose reference counts will be
3674 The base of this stack is pointed to by the interpreter variable
3675 C<PL_tmps_stack>, of type C<SV **>.
3677 The head of the stack is indexed by C<PL_tmps_ix>, an integer which stores the
3678 index in the array of the most recently-pushed item.
3680 There is no public API to directly push items to the temporaries stack. Instead,
3681 the API function C<sv_2mortal()> is used to mortalize an xV, adding its
3682 address to the temporaries stack.
3684 Likewise, there is no public API to read values from the temporaries stack.
3685 Instead, the macros C<SAVETMPS> and C<FREETMPS> are used. The C<SAVETMPS>
3686 macro establishes the base levels of the temporaries stack, by capturing the
3687 current value of C<PL_tmps_ix> into C<PL_tmps_floor> and saving the previous
3688 value to the save stack. Thereafter, whenever C<FREETMPS> is invoked all of
3689 the temporaries that have been pushed since that level are reclaimed.
3691 While it is common to see these two macros in pairs within an C<ENTER>/
3692 C<LEAVE> pair, it is not necessary to match them. It is permitted to invoke
3693 C<FREETMPS> multiple times since the most recent C<SAVETMPS>; for example in a
3694 loop iterating over elements of a list. While you can invoke C<SAVETMPS>
3695 multiple times within a scope pair, it is unlikely to be useful. Subsequent
3696 invocations will move the temporaries floor further up, thus effectively
3697 trapping the existing temporaries to only be released at the end of the scope.
3701 The save stack is used by perl to implement the C<local> keyword and other
3702 similar behaviours; any cleanup operations that need to be performed when
3703 leaving the current scope. Items pushed to this stack generally capture the
3704 current value of some internal variable or state, which will be restored when
3705 the scope is unwound due to leaving, C<return>, C<die>, C<goto> or other
3708 Whereas other perl internal stacks store individual items all of the same type
3709 (usually SV pointers or integers), the items pushed to the save stack are
3710 formed of many different types, having multiple fields to them. For example,
3711 the C<SAVEt_INT> type needs to store both the address of the C<int> variable
3712 to restore, and the value to restore it to. This information could have been
3713 stored using fields of a C<struct>, but would have to be large enough to store
3714 three pointers in the largest case, which would waste a lot of space in most
3715 of the smaller cases.
3717 Instead, the stack stores information in a variable-length encoding of C<ANY>
3718 structures. The final value pushed is stored in the C<UV> field which encodes
3719 the kind of item held by the preceding items; the count and types of which
3720 will depend on what kind of item is being stored. The kind field is pushed
3721 last because that will be the first field to be popped when unwinding items
3724 The base of this stack is pointed to by the interpreter variable
3725 C<PL_savestack>, of type C<ANY *>.
3727 The head of the stack is indexed by C<PL_savestack_ix>, an integer which
3728 stores the index in the array at which the next item should be pushed. (Note
3729 that this is different to most other stacks, which reference the most
3730 recently-pushed item).
3732 Items are pushed to the save stack by using the various C<SAVE...()> macros.
3733 Many of these macros take a variable and store both its address and current
3734 value on the save stack, ensuring that value gets restored on scope exit.
3742 There are also a variety of other special-purpose macros which save particular
3743 types or values of interest. C<SAVETMPS> has already been mentioned above.
3744 Others include C<SAVEFREEPV> which arranges for a PV (i.e. a string buffer) to
3745 be freed, or C<SAVEDESTRUCTOR> which arranges for a given function pointer to
3746 be invoked on scope exit. A full list of such macros can be found in
3749 There is no public API for popping individual values or items from the save
3750 stack. Instead, via the scope stack, the C<ENTER> and C<LEAVE> pair form a way
3751 to start and stop nested scopes. Leaving a nested scope via C<LEAVE> will
3752 restore all of the saved values that had been pushed since the most recent
3757 As with the mark stack to the value stack, the scope stack forms a pair with
3758 the save stack. The scope stack stores the height of the save stack at which
3759 nested scopes begin, and allows the save stack to be unwound back to that
3760 point when the scope is left.
3762 When perl is built with debugging enabled, there is a second part to this
3763 stack storing human-readable string names describing the type of stack
3764 context. Each push operation saves the name as well as the height of the save
3765 stack, and each pop operation checks the topmost name with what is expected,
3766 causing an assertion failure if the name does not match.
3768 The base of this stack is pointed to by the interpreter variable
3769 C<PL_scopestack>, of type C<I32 *>. If enabled, the scope stack names are
3770 stored in a separate array pointed to by C<PL_scopestack_name>, of type
3773 The head of the stack is indexed by C<PL_scopestack_ix>, an integer which
3774 stores the index of the array or arrays at which the next item should be
3775 pushed. (Note that this is different to most other stacks, which reference the
3776 most recently-pushed item).
3778 Values are pushed to the scope stack using the C<ENTER> macro, which begins a
3779 new nested scope. Any items pushed to the save stack are then restored at the
3780 next nested invocation of the C<LEAVE> macro.
3782 =head1 Dynamic Scope and the Context Stack
3784 B<Note:> this section describes a non-public internal API that is subject
3785 to change without notice.
3787 =head2 Introduction to the context stack
3789 In Perl, dynamic scoping refers to the runtime nesting of things like
3790 subroutine calls, evals etc, as well as the entering and exiting of block
3791 scopes. For example, the restoring of a C<local>ised variable is
3792 determined by the dynamic scope.
3794 Perl tracks the dynamic scope by a data structure called the context
3795 stack, which is an array of C<PERL_CONTEXT> structures, and which is
3796 itself a big union for all the types of context. Whenever a new scope is
3797 entered (such as a block, a C<for> loop, or a subroutine call), a new
3798 context entry is pushed onto the stack. Similarly when leaving a block or
3799 returning from a subroutine call etc. a context is popped. Since the
3800 context stack represents the current dynamic scope, it can be searched.
3801 For example, C<next LABEL> searches back through the stack looking for a
3802 loop context that matches the label; C<return> pops contexts until it
3803 finds a sub or eval context or similar; C<caller> examines sub contexts on
3806 Each context entry is labelled with a context type, C<cx_type>. Typical
3807 context types are C<CXt_SUB>, C<CXt_EVAL> etc., as well as C<CXt_BLOCK>
3808 and C<CXt_NULL> which represent a basic scope (as pushed by C<pp_enter>)
3809 and a sort block. The type determines which part of the context union are
3812 The main division in the context struct is between a substitution scope
3813 (C<CXt_SUBST>) and block scopes, which are everything else. The former is
3814 just used while executing C<s///e>, and won't be discussed further
3817 All the block scope types share a common base, which corresponds to
3818 C<CXt_BLOCK>. This stores the old values of various scope-related
3819 variables like C<PL_curpm>, as well as information about the current
3820 scope, such as C<gimme>. On scope exit, the old variables are restored.
3822 Particular block scope types store extra per-type information. For
3823 example, C<CXt_SUB> stores the currently executing CV, while the various
3824 for loop types might hold the original loop variable SV. On scope exit,
3825 the per-type data is processed; for example the CV has its reference count
3826 decremented, and the original loop variable is restored.
3828 The macro C<cxstack> returns the base of the current context stack, while
3829 C<cxstack_ix> is the index of the current frame within that stack.
3831 In fact, the context stack is actually part of a stack-of-stacks system;
3832 whenever something unusual is done such as calling a C<DESTROY> or tie
3833 handler, a new stack is pushed, then popped at the end.
3835 Note that the API described here changed considerably in perl 5.24; prior
3836 to that, big macros like C<PUSHBLOCK> and C<POPSUB> were used; in 5.24
3837 they were replaced by the inline static functions described below. In
3838 addition, the ordering and detail of how these macros/function work
3839 changed in many ways, often subtly. In particular they didn't handle
3840 saving the savestack and temps stack positions, and required additional
3841 C<ENTER>, C<SAVETMPS> and C<LEAVE> compared to the new functions. The
3842 old-style macros will not be described further.
3845 =head2 Pushing contexts
3847 For pushing a new context, the two basic functions are
3848 C<cx = cx_pushblock()>, which pushes a new basic context block and returns
3849 its address, and a family of similar functions with names like
3850 C<cx_pushsub(cx)> which populate the additional type-dependent fields in
3851 the C<cx> struct. Note that C<CXt_NULL> and C<CXt_BLOCK> don't have their
3852 own push functions, as they don't store any data beyond that pushed by
3855 The fields of the context struct and the arguments to the C<cx_*>
3856 functions are subject to change between perl releases, representing
3857 whatever is convenient or efficient for that release.
3859 A typical context stack pushing can be found in C<pp_entersub>; the
3860 following shows a simplified and stripped-down example of a non-XS call,
3861 along with comments showing roughly what each function does.
3865 bool hasargs = cBOOL(PL_op->op_flags & OPf_STACKED);
3866 OP *retop = PL_op->op_next;
3867 I32 old_ss_ix = PL_savestack_ix;
3870 /* ... make mortal copies of stack args which are PADTMPs here ... */
3872 /* ... do any additional savestack pushes here ... */
3874 /* Now push a new context entry of type 'CXt_SUB'; initially just
3875 * doing the actions common to all block types: */
3877 cx = cx_pushblock(CXt_SUB, gimme, MARK, old_ss_ix);
3879 /* this does (approximately):
3880 CXINC; /* cxstack_ix++ (grow if necessary) */
3881 cx = CX_CUR(); /* and get the address of new frame */
3882 cx->cx_type = CXt_SUB;
3883 cx->blk_gimme = gimme;
3884 cx->blk_oldsp = MARK - PL_stack_base;
3885 cx->blk_oldsaveix = old_ss_ix;
3886 cx->blk_oldcop = PL_curcop;
3887 cx->blk_oldmarksp = PL_markstack_ptr - PL_markstack;
3888 cx->blk_oldscopesp = PL_scopestack_ix;
3889 cx->blk_oldpm = PL_curpm;
3890 cx->blk_old_tmpsfloor = PL_tmps_floor;
3892 PL_tmps_floor = PL_tmps_ix;
3896 /* then update the new context frame with subroutine-specific info,
3897 * such as the CV about to be executed: */
3899 cx_pushsub(cx, cv, retop, hasargs);
3901 /* this does (approximately):
3902 cx->blk_sub.cv = cv;
3903 cx->blk_sub.olddepth = CvDEPTH(cv);
3904 cx->blk_sub.prevcomppad = PL_comppad;
3905 cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
3906 cx->blk_sub.retop = retop;
3907 SvREFCNT_inc_simple_void_NN(cv);
3910 Note that C<cx_pushblock()> sets two new floors: for the args stack (to
3911 C<MARK>) and the temps stack (to C<PL_tmps_ix>). While executing at this
3912 scope level, every C<nextstate> (amongst others) will reset the args and
3913 tmps stack levels to these floors. Note that since C<cx_pushblock> uses
3914 the current value of C<PL_tmps_ix> rather than it being passed as an arg,
3915 this dictates at what point C<cx_pushblock> should be called. In
3916 particular, any new mortals which should be freed only on scope exit
3917 (rather than at the next C<nextstate>) should be created first.
3919 Most callers of C<cx_pushblock> simply set the new args stack floor to the
3920 top of the previous stack frame, but for C<CXt_LOOP_LIST> it stores the
3921 items being iterated over on the stack, and so sets C<blk_oldsp> to the
3922 top of these items instead. Note that, contrary to its name, C<blk_oldsp>
3923 doesn't always represent the value to restore C<PL_stack_sp> to on scope
3926 Note the early capture of C<PL_savestack_ix> to C<old_ss_ix>, which is
3927 later passed as an arg to C<cx_pushblock>. In the case of C<pp_entersub>,
3928 this is because, although most values needing saving are stored in fields
3929 of the context struct, an extra value needs saving only when the debugger
3930 is running, and it doesn't make sense to bloat the struct for this rare
3931 case. So instead it is saved on the savestack. Since this value gets
3932 calculated and saved before the context is pushed, it is necessary to pass
3933 the old value of C<PL_savestack_ix> to C<cx_pushblock>, to ensure that the
3934 saved value gets freed during scope exit. For most users of
3935 C<cx_pushblock>, where nothing needs pushing on the save stack,
3936 C<PL_savestack_ix> is just passed directly as an arg to C<cx_pushblock>.
3938 Note that where possible, values should be saved in the context struct
3939 rather than on the save stack; it's much faster that way.
3941 Normally C<cx_pushblock> should be immediately followed by the appropriate
3942 C<cx_pushfoo>, with nothing between them; this is because if code
3943 in-between could die (e.g. a warning upgraded to fatal), then the context
3944 stack unwinding code in C<dounwind> would see (in the example above) a
3945 C<CXt_SUB> context frame, but without all the subroutine-specific fields
3946 set, and crashes would soon ensue.
3948 Where the two must be separate, initially set the type to C<CXt_NULL> or
3949 C<CXt_BLOCK>, and later change it to C<CXt_foo> when doing the
3950 C<cx_pushfoo>. This is exactly what C<pp_enteriter> does, once it's
3951 determined which type of loop it's pushing.
3953 =head2 Popping contexts
3955 Contexts are popped using C<cx_popsub()> etc. and C<cx_popblock()>. Note
3956 however, that unlike C<cx_pushblock>, neither of these functions actually
3957 decrement the current context stack index; this is done separately using
3960 There are two main ways that contexts are popped. During normal execution
3961 as scopes are exited, functions like C<pp_leave>, C<pp_leaveloop> and
3962 C<pp_leavesub> process and pop just one context using C<cx_popfoo> and
3963 C<cx_popblock>. On the other hand, things like C<pp_return> and C<next>
3964 may have to pop back several scopes until a sub or loop context is found,
3965 and exceptions (such as C<die>) need to pop back contexts until an eval
3966 context is found. Both of these are accomplished by C<dounwind()>, which
3967 is capable of processing and popping all contexts above the target one.
3969 Here is a typical example of context popping, as found in C<pp_leavesub>
3970 (simplified slightly):
3979 gimme = cx->blk_gimme;
3980 oldsp = PL_stack_base + cx->blk_oldsp; /* last arg of previous frame */
3982 if (gimme == G_VOID)
3983 PL_stack_sp = oldsp;
3985 leave_adjust_stacks(oldsp, oldsp, gimme, 0);
3990 retop = cx->blk_sub.retop;
3995 The steps above are in a very specific order, designed to be the reverse
3996 order of when the context was pushed. The first thing to do is to copy
3997 and/or protect any return arguments and free any temps in the current
3998 scope. Scope exits like an rvalue sub normally return a mortal copy of
3999 their return args (as opposed to lvalue subs). It is important to make
4000 this copy before the save stack is popped or variables are restored, or
4001 bad things like the following can happen:
4003 sub f { my $x =...; $x } # $x freed before we get to copy it
4004 sub f { /(...)/; $1 } # PL_curpm restored before $1 copied
4006 Although we wish to free any temps at the same time, we have to be careful
4007 not to free any temps which are keeping return args alive; nor to free the
4008 temps we have just created while mortal copying return args. Fortunately,
4009 C<leave_adjust_stacks()> is capable of making mortal copies of return args,
4010 shifting args down the stack, and only processing those entries on the
4011 temps stack that are safe to do so.
4013 In void context no args are returned, so it's more efficient to skip
4014 calling C<leave_adjust_stacks()>. Also in void context, a C<nextstate> op
4015 is likely to be imminently called which will do a C<FREETMPS>, so there's
4016 no need to do that either.
4018 The next step is to pop savestack entries: C<CX_LEAVE_SCOPE(cx)> is just
4019 defined as C<< LEAVE_SCOPE(cx->blk_oldsaveix) >>. Note that during the
4020 popping, it's possible for perl to call destructors, call C<STORE> to undo
4021 localisations of tied vars, and so on. Any of these can die or call
4022 C<exit()>. In this case, C<dounwind()> will be called, and the current
4023 context stack frame will be re-processed. Thus it is vital that all steps
4024 in popping a context are done in such a way to support reentrancy. The
4025 other alternative, of decrementing C<cxstack_ix> I<before> processing the
4026 frame, would lead to leaks and the like if something died halfway through,
4027 or overwriting of the current frame.
4029 C<CX_LEAVE_SCOPE> itself is safely re-entrant: if only half the savestack
4030 items have been popped before dying and getting trapped by eval, then the
4031 C<CX_LEAVE_SCOPE>s in C<dounwind> or C<pp_leaveeval> will continue where
4032 the first one left off.
4034 The next step is the type-specific context processing; in this case
4035 C<cx_popsub>. In part, this looks like:
4037 cv = cx->blk_sub.cv;
4038 CvDEPTH(cv) = cx->blk_sub.olddepth;
4039 cx->blk_sub.cv = NULL;
4042 where its processing the just-executed CV. Note that before it decrements
4043 the CV's reference count, it nulls the C<blk_sub.cv>. This means that if
4044 it re-enters, the CV won't be freed twice. It also means that you can't
4045 rely on such type-specific fields having useful values after the return
4048 Next, C<cx_popblock> restores all the various interpreter vars to their
4049 previous values or previous high water marks; it expands to:
4051 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
4052 PL_scopestack_ix = cx->blk_oldscopesp;
4053 PL_curpm = cx->blk_oldpm;
4054 PL_curcop = cx->blk_oldcop;
4055 PL_tmps_floor = cx->blk_old_tmpsfloor;
4057 Note that it I<doesn't> restore C<PL_stack_sp>; as mentioned earlier,
4058 which value to restore it to depends on the context type (specifically
4059 C<for (list) {}>), and what args (if any) it returns; and that will
4060 already have been sorted out earlier by C<leave_adjust_stacks()>.
4062 Finally, the context stack pointer is actually decremented by C<CX_POP(cx)>.
4063 After this point, it's possible that that the current context frame could
4064 be overwritten by other contexts being pushed. Although things like ties
4065 and C<DESTROY> are supposed to work within a new context stack, it's best
4066 not to assume this. Indeed on debugging builds, C<CX_POP(cx)> deliberately
4067 sets C<cx> to null to detect code that is still relying on the field
4068 values in that context frame. Note in the C<pp_leavesub()> example above,
4069 we grab C<blk_sub.retop> I<before> calling C<CX_POP>.
4071 =head2 Redoing contexts
4073 Finally, there is C<cx_topblock(cx)>, which acts like a super-C<nextstate>
4074 as regards to resetting various vars to their base values. It is used in
4075 places like C<pp_next>, C<pp_redo> and C<pp_goto> where rather than
4076 exiting a scope, we want to re-initialise the scope. As well as resetting
4077 C<PL_stack_sp> like C<nextstate>, it also resets C<PL_markstack_ptr>,
4078 C<PL_scopestack_ix> and C<PL_curpm>. Note that it doesn't do a
4082 =head1 Slab-based operator allocation
4084 B<Note:> this section describes a non-public internal API that is subject
4085 to change without notice.
4087 Perl's internal error-handling mechanisms implement C<die> (and its internal
4088 equivalents) using longjmp. If this occurs during lexing, parsing or
4089 compilation, we must ensure that any ops allocated as part of the compilation
4090 process are freed. (Older Perl versions did not adequately handle this
4091 situation: when failing a parse, they would leak ops that were stored in
4092 C C<auto> variables and not linked anywhere else.)
4094 To handle this situation, Perl uses I<op slabs> that are attached to the
4095 currently-compiling CV. A slab is a chunk of allocated memory. New ops are
4096 allocated as regions of the slab. If the slab fills up, a new one is created
4097 (and linked from the previous one). When an error occurs and the CV is freed,
4098 any ops remaining are freed.
4100 Each op is preceded by two pointers: one points to the next op in the slab, and
4101 the other points to the slab that owns it. The next-op pointer is needed so
4102 that Perl can iterate over a slab and free all its ops. (Op structures are of
4103 different sizes, so the slab's ops can't merely be treated as a dense array.)
4104 The slab pointer is needed for accessing a reference count on the slab: when
4105 the last op on a slab is freed, the slab itself is freed.
4107 The slab allocator puts the ops at the end of the slab first. This will tend to
4108 allocate the leaves of the op tree first, and the layout will therefore
4109 hopefully be cache-friendly. In addition, this means that there's no need to
4110 store the size of the slab (see below on why slabs vary in size), because Perl
4111 can follow pointers to find the last op.
4113 It might seem possible to eliminate slab reference counts altogether, by having
4114 all ops implicitly attached to C<PL_compcv> when allocated and freed when the
4115 CV is freed. That would also allow C<op_free> to skip C<FreeOp> altogether, and
4116 thus free ops faster. But that doesn't work in those cases where ops need to
4117 survive beyond their CVs, such as re-evals.
4119 The CV also has to have a reference count on the slab. Sometimes the first op
4120 created is immediately freed. If the reference count of the slab reaches 0,
4121 then it will be freed with the CV still pointing to it.
4123 CVs use the C<CVf_SLABBED> flag to indicate that the CV has a reference count
4124 on the slab. When this flag is set, the slab is accessible via C<CvSTART> when
4125 C<CvROOT> is not set, or by subtracting two pointers C<(2*sizeof(I32 *))> from
4126 C<CvROOT> when it is set. The alternative to this approach of sneaking the slab
4127 into C<CvSTART> during compilation would be to enlarge the C<xpvcv> struct by
4128 another pointer. But that would make all CVs larger, even though slab-based op
4129 freeing is typically of benefit only for programs that make significant use of
4132 When the C<CVf_SLABBED> flag is set, the CV takes responsibility for freeing
4133 the slab. If C<CvROOT> is not set when the CV is freed or undeffed, it is
4134 assumed that a compilation error has occurred, so the op slab is traversed and
4135 all the ops are freed.
4137 Under normal circumstances, the CV forgets about its slab (decrementing the
4138 reference count) when the root is attached. So the slab reference counting that
4139 happens when ops are freed takes care of freeing the slab. In some cases, the
4140 CV is told to forget about the slab (C<cv_forget_slab>) precisely so that the
4141 ops can survive after the CV is done away with.
4143 Forgetting the slab when the root is attached is not strictly necessary, but
4144 avoids potential problems with C<CvROOT> being written over. There is code all
4145 over the place, both in core and on CPAN, that does things with C<CvROOT>, so
4146 forgetting the slab makes things more robust and avoids potential problems.
4148 Since the CV takes ownership of its slab when flagged, that flag is never
4149 copied when a CV is cloned, as one CV could free a slab that another CV still
4150 points to, since forced freeing of ops ignores the reference count (but asserts
4151 that it looks right).
4153 To avoid slab fragmentation, freed ops are marked as freed and attached to the
4154 slab's freed chain (an idea stolen from DBM::Deep). Those freed ops are reused
4155 when possible. Not reusing freed ops would be simpler, but it would result in
4156 significantly higher memory usage for programs with large C<if (DEBUG) {...}>
4159 C<SAVEFREEOP> is slightly problematic under this scheme. Sometimes it can cause
4160 an op to be freed after its CV. If the CV has forcibly freed the ops on its
4161 slab and the slab itself, then we will be fiddling with a freed slab. Making
4162 C<SAVEFREEOP> a no-op doesn't help, as sometimes an op can be savefreed when
4163 there is no compilation error, so the op would never be freed. It holds
4164 a reference count on the slab, so the whole slab would leak. So C<SAVEFREEOP>
4165 now sets a special flag on the op (C<< ->op_savefree >>). The forced freeing of
4166 ops after a compilation error won't free any ops thus marked.
4168 Since many pieces of code create tiny subroutines consisting of only a few ops,
4169 and since a huge slab would be quite a bit of baggage for those to carry
4170 around, the first slab is always very small. To avoid allocating too many
4171 slabs for a single CV, each subsequent slab is twice the size of the previous.
4173 Smartmatch expects to be able to allocate an op at run time, run it, and then
4174 throw it away. For that to work the op is simply malloced when PL_compcv hasn't
4175 been set up. So all slab-allocated ops are marked as such (C<< ->op_slabbed >>),
4176 to distinguish them from malloced ops.
4181 Until May 1997, this document was maintained by Jeff Okamoto
4182 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
4183 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
4185 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
4186 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
4187 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
4188 Stephen McCamant, and Gurusamy Sarathy.
4192 L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>