1PERLCALL(1)            Perl Programmers Reference Guide            PERLCALL(1)
2
3
4

NAME

6       perlcall - Perl calling conventions from C
7

DESCRIPTION

9       The purpose of this document is to show you how to call Perl
10       subroutines directly from C, i.e., how to write callbacks.
11
12       Apart from discussing the C interface provided by Perl for writing
13       callbacks the document uses a series of examples to show how the
14       interface actually works in practice.  In addition some techniques for
15       coding callbacks are covered.
16
17       Examples where callbacks are necessary include
18
19       •    An Error Handler
20
21            You have created an XSUB interface to an application's C API.
22
23            A fairly common feature in applications is to allow you to define
24            a C function that will be called whenever something nasty occurs.
25            What we would like is to be able to specify a Perl subroutine that
26            will be called instead.
27
28       •    An Event-Driven Program
29
30            The classic example of where callbacks are used is when writing an
31            event driven program, such as for an X11 application.  In this
32            case you register functions to be called whenever specific events
33            occur, e.g., a mouse button is pressed, the cursor moves into a
34            window or a menu item is selected.
35
36       Although the techniques described here are applicable when embedding
37       Perl in a C program, this is not the primary goal of this document.
38       There are other details that must be considered and are specific to
39       embedding Perl. For details on embedding Perl in C refer to perlembed.
40
41       Before you launch yourself head first into the rest of this document,
42       it would be a good idea to have read the following two
43       documents--perlxs and perlguts.
44

THE CALL_ FUNCTIONS

46       Although this stuff is easier to explain using examples, you first need
47       be aware of a few important definitions.
48
49       Perl has a number of C functions that allow you to call Perl
50       subroutines.  They are
51
52           I32 call_sv(SV* sv, I32 flags);
53           I32 call_pv(char *subname, I32 flags);
54           I32 call_method(char *methname, I32 flags);
55           I32 call_argv(char *subname, I32 flags, char **argv);
56
57       The key function is call_sv.  All the other functions are fairly simple
58       wrappers which make it easier to call Perl subroutines in special
59       cases. At the end of the day they will all call call_sv to invoke the
60       Perl subroutine.
61
62       All the call_* functions have a "flags" parameter which is used to pass
63       a bit mask of options to Perl.  This bit mask operates identically for
64       each of the functions.  The settings available in the bit mask are
65       discussed in "FLAG VALUES".
66
67       Each of the functions will now be discussed in turn.
68
69       call_sv
70            call_sv takes two parameters. The first, "sv", is an SV*.  This
71            allows you to specify the Perl subroutine to be called either as a
72            C string (which has first been converted to an SV) or a reference
73            to a subroutine. The section, "Using call_sv", shows how you can
74            make use of call_sv.
75
76       call_pv
77            The function, call_pv, is similar to call_sv except it expects its
78            first parameter to be a C char* which identifies the Perl
79            subroutine you want to call, e.g., "call_pv("fred", 0)".  If the
80            subroutine you want to call is in another package, just include
81            the package name in the string, e.g., "pkg::fred".
82
83       call_method
84            The function call_method is used to call a method from a Perl
85            class.  The parameter "methname" corresponds to the name of the
86            method to be called.  Note that the class that the method belongs
87            to is passed on the Perl stack rather than in the parameter list.
88            This class can be either the name of the class (for a static
89            method) or a reference to an object (for a virtual method).  See
90            perlobj for more information on static and virtual methods and
91            "Using call_method" for an example of using call_method.
92
93       call_argv
94            call_argv calls the Perl subroutine specified by the C string
95            stored in the "subname" parameter. It also takes the usual "flags"
96            parameter.  The final parameter, "argv", consists of a NULL-
97            terminated list of C strings to be passed as parameters to the
98            Perl subroutine.  See "Using call_argv".
99
100       All the functions return an integer. This is a count of the number of
101       items returned by the Perl subroutine. The actual items returned by the
102       subroutine are stored on the Perl stack.
103
104       As a general rule you should always check the return value from these
105       functions.  Even if you are expecting only a particular number of
106       values to be returned from the Perl subroutine, there is nothing to
107       stop someone from doing something unexpected--don't say you haven't
108       been warned.
109

FLAG VALUES

111       The "flags" parameter in all the call_* functions is one of "G_VOID",
112       "G_SCALAR", or "G_LIST", which indicate the call context, OR'ed
113       together with a bit mask of any combination of the other G_* symbols
114       defined below.
115
116   G_VOID
117       Calls the Perl subroutine in a void context.
118
119       This flag has 2 effects:
120
121       1.   It indicates to the subroutine being called that it is executing
122            in a void context (if it executes wantarray the result will be the
123            undefined value).
124
125       2.   It ensures that nothing is actually returned from the subroutine.
126
127       The value returned by the call_* function indicates how many items have
128       been returned by the Perl subroutine--in this case it will be 0.
129
130   G_SCALAR
131       Calls the Perl subroutine in a scalar context.  This is the default
132       context flag setting for all the call_* functions.
133
134       This flag has 2 effects:
135
136       1.   It indicates to the subroutine being called that it is executing
137            in a scalar context (if it executes wantarray the result will be
138            false).
139
140       2.   It ensures that only a scalar is actually returned from the
141            subroutine.  The subroutine can, of course,  ignore the wantarray
142            and return a list anyway. If so, then only the last element of the
143            list will be returned.
144
145       The value returned by the call_* function indicates how many items have
146       been returned by the Perl subroutine - in this case it will be either 0
147       or 1.
148
149       If 0, then you have specified the G_DISCARD flag.
150
151       If 1, then the item actually returned by the Perl subroutine will be
152       stored on the Perl stack - the section "Returning a Scalar" shows how
153       to access this value on the stack.  Remember that regardless of how
154       many items the Perl subroutine returns, only the last one will be
155       accessible from the stack - think of the case where only one value is
156       returned as being a list with only one element.  Any other items that
157       were returned will not exist by the time control returns from the
158       call_* function.  The section "Returning a List in Scalar Context"
159       shows an example of this behavior.
160
161   G_LIST
162       Calls the Perl subroutine in a list context. Prior to Perl version
163       5.35.1 this was called "G_ARRAY".
164
165       As with G_SCALAR, this flag has 2 effects:
166
167       1.   It indicates to the subroutine being called that it is executing
168            in a list context (if it executes wantarray the result will be
169            true).
170
171       2.   It ensures that all items returned from the subroutine will be
172            accessible when control returns from the call_* function.
173
174       The value returned by the call_* function indicates how many items have
175       been returned by the Perl subroutine.
176
177       If 0, then you have specified the G_DISCARD flag.
178
179       If not 0, then it will be a count of the number of items returned by
180       the subroutine. These items will be stored on the Perl stack.  The
181       section "Returning a List of Values" gives an example of using the
182       G_LIST flag and the mechanics of accessing the returned items from the
183       Perl stack.
184
185   G_DISCARD
186       By default, the call_* functions place the items returned from by the
187       Perl subroutine on the stack.  If you are not interested in these
188       items, then setting this flag will make Perl get rid of them
189       automatically for you.  Note that it is still possible to indicate a
190       context to the Perl subroutine by using either G_SCALAR or G_LIST.
191
192       If you do not set this flag then it is very important that you make
193       sure that any temporaries (i.e., parameters passed to the Perl
194       subroutine and values returned from the subroutine) are disposed of
195       yourself.  The section "Returning a Scalar" gives details of how to
196       dispose of these temporaries explicitly and the section "Using Perl to
197       Dispose of Temporaries" discusses the specific circumstances where you
198       can ignore the problem and let Perl deal with it for you.
199
200   G_NOARGS
201       Whenever a Perl subroutine is called using one of the call_* functions,
202       it is assumed by default that parameters are to be passed to the
203       subroutine.  If you are not passing any parameters to the Perl
204       subroutine, you can save a bit of time by setting this flag.  It has
205       the effect of not creating the @_ array for the Perl subroutine.
206
207       Although the functionality provided by this flag may seem
208       straightforward, it should be used only if there is a good reason to do
209       so.  The reason for being cautious is that, even if you have specified
210       the G_NOARGS flag, it is still possible for the Perl subroutine that
211       has been called to think that you have passed it parameters.
212
213       In fact, what can happen is that the Perl subroutine you have called
214       can access the @_ array from a previous Perl subroutine.  This will
215       occur when the code that is executing the call_* function has itself
216       been called from another Perl subroutine. The code below illustrates
217       this
218
219           sub fred
220             { print "@_\n"  }
221
222           sub joe
223             { &fred }
224
225           &joe(1,2,3);
226
227       This will print
228
229           1 2 3
230
231       What has happened is that "fred" accesses the @_ array which belongs to
232       "joe".
233
234   G_EVAL
235       It is possible for the Perl subroutine you are calling to terminate
236       abnormally, e.g., by calling die explicitly or by not actually
237       existing.  By default, when either of these events occurs, the process
238       will terminate immediately.  If you want to trap this type of event,
239       specify the G_EVAL flag.  It will put an eval { } around the subroutine
240       call.
241
242       Whenever control returns from the call_* function you need to check the
243       $@ variable as you would in a normal Perl script.
244
245       The value returned from the call_* function is dependent on what other
246       flags have been specified and whether an error has occurred.  Here are
247       all the different cases that can occur:
248
249       •    If the call_* function returns normally, then the value returned
250            is as specified in the previous sections.
251
252       •    If G_DISCARD is specified, the return value will always be 0.
253
254       •    If G_LIST is specified and an error has occurred, the return value
255            will always be 0.
256
257       •    If G_SCALAR is specified and an error has occurred, the return
258            value will be 1 and the value on the top of the stack will be
259            undef. This means that if you have already detected the error by
260            checking $@ and you want the program to continue, you must
261            remember to pop the undef from the stack.
262
263       See "Using G_EVAL" for details on using G_EVAL.
264
265   G_KEEPERR
266       Using the G_EVAL flag described above will always set $@: clearing it
267       if there was no error, and setting it to describe the error if there
268       was an error in the called code.  This is what you want if your
269       intention is to handle possible errors, but sometimes you just want to
270       trap errors and stop them interfering with the rest of the program.
271
272       This scenario will mostly be applicable to code that is meant to be
273       called from within destructors, asynchronous callbacks, and signal
274       handlers.  In such situations, where the code being called has little
275       relation to the surrounding dynamic context, the main program needs to
276       be insulated from errors in the called code, even if they can't be
277       handled intelligently.  It may also be useful to do this with code for
278       "__DIE__" or "__WARN__" hooks, and "tie" functions.
279
280       The G_KEEPERR flag is meant to be used in conjunction with G_EVAL in
281       call_* functions that are used to implement such code, or with
282       "eval_sv".  This flag has no effect on the "call_*" functions when
283       G_EVAL is not used.
284
285       When G_KEEPERR is used, any error in the called code will terminate the
286       call as usual, and the error will not propagate beyond the call (as
287       usual for G_EVAL), but it will not go into $@.  Instead the error will
288       be converted into a warning, prefixed with the string "\t(in cleanup)".
289       This can be disabled using "no warnings 'misc'".  If there is no error,
290       $@ will not be cleared.
291
292       Note that the G_KEEPERR flag does not propagate into inner evals; these
293       may still set $@.
294
295       The G_KEEPERR flag was introduced in Perl version 5.002.
296
297       See "Using G_KEEPERR" for an example of a situation that warrants the
298       use of this flag.
299
300   Determining the Context
301       As mentioned above, you can determine the context of the currently
302       executing subroutine in Perl with wantarray.  The equivalent test can
303       be made in C by using the "GIMME_V" macro, which returns "G_LIST" if
304       you have been called in a list context, "G_SCALAR" if in a scalar
305       context, or "G_VOID" if in a void context (i.e., the return value will
306       not be used).  An older version of this macro is called "GIMME"; in a
307       void context it returns "G_SCALAR" instead of "G_VOID".  An example of
308       using the "GIMME_V" macro is shown in section "Using GIMME_V".
309

EXAMPLES

311       Enough of the definition talk! Let's have a few examples.
312
313       Perl provides many macros to assist in accessing the Perl stack.
314       Wherever possible, these macros should always be used when interfacing
315       to Perl internals.  We hope this should make the code less vulnerable
316       to any changes made to Perl in the future.
317
318       Another point worth noting is that in the first series of examples I
319       have made use of only the call_pv function.  This has been done to keep
320       the code simpler and ease you into the topic.  Wherever possible, if
321       the choice is between using call_pv and call_sv, you should always try
322       to use call_sv.  See "Using call_sv" for details.
323
324   No Parameters, Nothing Returned
325       This first trivial example will call a Perl subroutine, PrintUID, to
326       print out the UID of the process.
327
328           sub PrintUID
329           {
330               print "UID is $<\n";
331           }
332
333       and here is a C function to call it
334
335           static void
336           call_PrintUID()
337           {
338               dSP;
339
340               PUSHMARK(SP);
341               call_pv("PrintUID", G_DISCARD|G_NOARGS);
342           }
343
344       Simple, eh?
345
346       A few points to note about this example:
347
348       1.   Ignore "dSP" and "PUSHMARK(SP)" for now. They will be discussed in
349            the next example.
350
351       2.   We aren't passing any parameters to PrintUID so G_NOARGS can be
352            specified.
353
354       3.   We aren't interested in anything returned from PrintUID, so
355            G_DISCARD is specified. Even if PrintUID was changed to return
356            some value(s), having specified G_DISCARD will mean that they will
357            be wiped by the time control returns from call_pv.
358
359       4.   As call_pv is being used, the Perl subroutine is specified as a C
360            string. In this case the subroutine name has been 'hard-wired'
361            into the code.
362
363       5.   Because we specified G_DISCARD, it is not necessary to check the
364            value returned from call_pv. It will always be 0.
365
366   Passing Parameters
367       Now let's make a slightly more complex example. This time we want to
368       call a Perl subroutine, "LeftString", which will take 2 parameters--a
369       string ($s) and an integer ($n).  The subroutine will simply print the
370       first $n characters of the string.
371
372       So the Perl subroutine would look like this:
373
374           sub LeftString
375           {
376               my($s, $n) = @_;
377               print substr($s, 0, $n), "\n";
378           }
379
380       The C function required to call LeftString would look like this:
381
382           static void
383           call_LeftString(a, b)
384           char * a;
385           int b;
386           {
387               dSP;
388
389               ENTER;
390               SAVETMPS;
391
392               PUSHMARK(SP);
393               EXTEND(SP, 2);
394               PUSHs(sv_2mortal(newSVpv(a, 0)));
395               PUSHs(sv_2mortal(newSViv(b)));
396               PUTBACK;
397
398               call_pv("LeftString", G_DISCARD);
399
400               FREETMPS;
401               LEAVE;
402           }
403
404       Here are a few notes on the C function call_LeftString.
405
406       1.   Parameters are passed to the Perl subroutine using the Perl stack.
407            This is the purpose of the code beginning with the line "dSP" and
408            ending with the line "PUTBACK".  The "dSP" declares a local copy
409            of the stack pointer.  This local copy should always be accessed
410            as "SP".
411
412       2.   If you are going to put something onto the Perl stack, you need to
413            know where to put it. This is the purpose of the macro "dSP"--it
414            declares and initializes a local copy of the Perl stack pointer.
415
416            All the other macros which will be used in this example require
417            you to have used this macro.
418
419            The exception to this rule is if you are calling a Perl subroutine
420            directly from an XSUB function. In this case it is not necessary
421            to use the "dSP" macro explicitly--it will be declared for you
422            automatically.
423
424       3.   Any parameters to be pushed onto the stack should be bracketed by
425            the "PUSHMARK" and "PUTBACK" macros.  The purpose of these two
426            macros, in this context, is to count the number of parameters you
427            are pushing automatically.  Then whenever Perl is creating the @_
428            array for the subroutine, it knows how big to make it.
429
430            The "PUSHMARK" macro tells Perl to make a mental note of the
431            current stack pointer. Even if you aren't passing any parameters
432            (like the example shown in the section "No Parameters, Nothing
433            Returned") you must still call the "PUSHMARK" macro before you can
434            call any of the call_* functions--Perl still needs to know that
435            there are no parameters.
436
437            The "PUTBACK" macro sets the global copy of the stack pointer to
438            be the same as our local copy. If we didn't do this, call_pv
439            wouldn't know where the two parameters we pushed were--remember
440            that up to now all the stack pointer manipulation we have done is
441            with our local copy, not the global copy.
442
443       4.   Next, we come to EXTEND and PUSHs. This is where the parameters
444            actually get pushed onto the stack. In this case we are pushing a
445            string and an integer.
446
447            Alternatively you can use the XPUSHs() macro, which combines a
448            "EXTEND(SP, 1)" and "PUSHs()".  This is less efficient if you're
449            pushing multiple values.
450
451            See "XSUBs and the Argument Stack" in perlguts for details on how
452            the PUSH macros work.
453
454       5.   Because we created temporary values (by means of sv_2mortal()
455            calls) we will have to tidy up the Perl stack and dispose of
456            mortal SVs.
457
458            This is the purpose of
459
460                ENTER;
461                SAVETMPS;
462
463            at the start of the function, and
464
465                FREETMPS;
466                LEAVE;
467
468            at the end. The "ENTER"/"SAVETMPS" pair creates a boundary for any
469            temporaries we create.  This means that the temporaries we get rid
470            of will be limited to those which were created after these calls.
471
472            The "FREETMPS"/"LEAVE" pair will get rid of any values returned by
473            the Perl subroutine (see next example), plus it will also dump the
474            mortal SVs we have created.  Having "ENTER"/"SAVETMPS" at the
475            beginning of the code makes sure that no other mortals are
476            destroyed.
477
478            Think of these macros as working a bit like "{" and "}" in Perl to
479            limit the scope of local variables.
480
481            See the section "Using Perl to Dispose of Temporaries" for details
482            of an alternative to using these macros.
483
484       6.   Finally, LeftString can now be called via the call_pv function.
485            The only flag specified this time is G_DISCARD. Because we are
486            passing 2 parameters to the Perl subroutine this time, we have not
487            specified G_NOARGS.
488
489   Returning a Scalar
490       Now for an example of dealing with the items returned from a Perl
491       subroutine.
492
493       Here is a Perl subroutine, Adder, that takes 2 integer parameters and
494       simply returns their sum.
495
496           sub Adder
497           {
498               my($a, $b) = @_;
499               $a + $b;
500           }
501
502       Because we are now concerned with the return value from Adder, the C
503       function required to call it is now a bit more complex.
504
505           static void
506           call_Adder(a, b)
507           int a;
508           int b;
509           {
510               dSP;
511               int count;
512
513               ENTER;
514               SAVETMPS;
515
516               PUSHMARK(SP);
517               EXTEND(SP, 2);
518               PUSHs(sv_2mortal(newSViv(a)));
519               PUSHs(sv_2mortal(newSViv(b)));
520               PUTBACK;
521
522               count = call_pv("Adder", G_SCALAR);
523
524               SPAGAIN;
525
526               if (count != 1)
527                   croak("Big trouble\n");
528
529               printf ("The sum of %d and %d is %d\n", a, b, POPi);
530
531               PUTBACK;
532               FREETMPS;
533               LEAVE;
534           }
535
536       Points to note this time are
537
538       1.   The only flag specified this time was G_SCALAR. That means that
539            the @_ array will be created and that the value returned by Adder
540            will still exist after the call to call_pv.
541
542       2.   The purpose of the macro "SPAGAIN" is to refresh the local copy of
543            the stack pointer. This is necessary because it is possible that
544            the memory allocated to the Perl stack has been reallocated during
545            the call_pv call.
546
547            If you are making use of the Perl stack pointer in your code you
548            must always refresh the local copy using SPAGAIN whenever you make
549            use of the call_* functions or any other Perl internal function.
550
551       3.   Although only a single value was expected to be returned from
552            Adder, it is still good practice to check the return code from
553            call_pv anyway.
554
555            Expecting a single value is not quite the same as knowing that
556            there will be one. If someone modified Adder to return a list and
557            we didn't check for that possibility and take appropriate action
558            the Perl stack would end up in an inconsistent state. That is
559            something you really don't want to happen ever.
560
561       4.   The "POPi" macro is used here to pop the return value from the
562            stack.  In this case we wanted an integer, so "POPi" was used.
563
564            Here is the complete list of POP macros available, along with the
565            types they return.
566
567                POPs        SV
568                POPp        pointer (PV)
569                POPpbytex   pointer to bytes (PV)
570                POPn        double (NV)
571                POPi        integer (IV)
572                POPu        unsigned integer (UV)
573                POPl        long
574                POPul       unsigned long
575
576            Since these macros have side-effects don't use them as arguments
577            to macros that may evaluate their argument several times, for
578            example:
579
580              /* Bad idea, don't do this */
581              STRLEN len;
582              const char *s = SvPV(POPs, len);
583
584            Instead, use a temporary:
585
586              STRLEN len;
587              SV *sv = POPs;
588              const char *s = SvPV(sv, len);
589
590            or a macro that guarantees it will evaluate its arguments only
591            once:
592
593              STRLEN len;
594              const char *s = SvPVx(POPs, len);
595
596       5.   The final "PUTBACK" is used to leave the Perl stack in a
597            consistent state before exiting the function.  This is necessary
598            because when we popped the return value from the stack with "POPi"
599            it updated only our local copy of the stack pointer.  Remember,
600            "PUTBACK" sets the global stack pointer to be the same as our
601            local copy.
602
603   Returning a List of Values
604       Now, let's extend the previous example to return both the sum of the
605       parameters and the difference.
606
607       Here is the Perl subroutine
608
609           sub AddSubtract
610           {
611              my($a, $b) = @_;
612              ($a+$b, $a-$b);
613           }
614
615       and this is the C function
616
617           static void
618           call_AddSubtract(a, b)
619           int a;
620           int b;
621           {
622               dSP;
623               int count;
624
625               ENTER;
626               SAVETMPS;
627
628               PUSHMARK(SP);
629               EXTEND(SP, 2);
630               PUSHs(sv_2mortal(newSViv(a)));
631               PUSHs(sv_2mortal(newSViv(b)));
632               PUTBACK;
633
634               count = call_pv("AddSubtract", G_LIST);
635
636               SPAGAIN;
637
638               if (count != 2)
639                   croak("Big trouble\n");
640
641               printf ("%d - %d = %d\n", a, b, POPi);
642               printf ("%d + %d = %d\n", a, b, POPi);
643
644               PUTBACK;
645               FREETMPS;
646               LEAVE;
647           }
648
649       If call_AddSubtract is called like this
650
651           call_AddSubtract(7, 4);
652
653       then here is the output
654
655           7 - 4 = 3
656           7 + 4 = 11
657
658       Notes
659
660       1.   We wanted list context, so G_LIST was used.
661
662       2.   Not surprisingly "POPi" is used twice this time because we were
663            retrieving 2 values from the stack. The important thing to note is
664            that when using the "POP*" macros they come off the stack in
665            reverse order.
666
667   Returning a List in Scalar Context
668       Say the Perl subroutine in the previous section was called in a scalar
669       context, like this
670
671           static void
672           call_AddSubScalar(a, b)
673           int a;
674           int b;
675           {
676               dSP;
677               int count;
678               int i;
679
680               ENTER;
681               SAVETMPS;
682
683               PUSHMARK(SP);
684               EXTEND(SP, 2);
685               PUSHs(sv_2mortal(newSViv(a)));
686               PUSHs(sv_2mortal(newSViv(b)));
687               PUTBACK;
688
689               count = call_pv("AddSubtract", G_SCALAR);
690
691               SPAGAIN;
692
693               printf ("Items Returned = %d\n", count);
694
695               for (i = 1; i <= count; ++i)
696                   printf ("Value %d = %d\n", i, POPi);
697
698               PUTBACK;
699               FREETMPS;
700               LEAVE;
701           }
702
703       The other modification made is that call_AddSubScalar will print the
704       number of items returned from the Perl subroutine and their value (for
705       simplicity it assumes that they are integer).  So if call_AddSubScalar
706       is called
707
708           call_AddSubScalar(7, 4);
709
710       then the output will be
711
712           Items Returned = 1
713           Value 1 = 3
714
715       In this case the main point to note is that only the last item in the
716       list is returned from the subroutine. AddSubtract actually made it back
717       to call_AddSubScalar.
718
719   Returning Data from Perl via the Parameter List
720       It is also possible to return values directly via the parameter
721       list--whether it is actually desirable to do it is another matter
722       entirely.
723
724       The Perl subroutine, Inc, below takes 2 parameters and increments each
725       directly.
726
727           sub Inc
728           {
729               ++ $_[0];
730               ++ $_[1];
731           }
732
733       and here is a C function to call it.
734
735           static void
736           call_Inc(a, b)
737           int a;
738           int b;
739           {
740               dSP;
741               int count;
742               SV * sva;
743               SV * svb;
744
745               ENTER;
746               SAVETMPS;
747
748               sva = sv_2mortal(newSViv(a));
749               svb = sv_2mortal(newSViv(b));
750
751               PUSHMARK(SP);
752               EXTEND(SP, 2);
753               PUSHs(sva);
754               PUSHs(svb);
755               PUTBACK;
756
757               count = call_pv("Inc", G_DISCARD);
758
759               if (count != 0)
760                   croak ("call_Inc: expected 0 values from 'Inc', got %d\n",
761                          count);
762
763               printf ("%d + 1 = %d\n", a, SvIV(sva));
764               printf ("%d + 1 = %d\n", b, SvIV(svb));
765
766               FREETMPS;
767               LEAVE;
768           }
769
770       To be able to access the two parameters that were pushed onto the stack
771       after they return from call_pv it is necessary to make a note of their
772       addresses--thus the two variables "sva" and "svb".
773
774       The reason this is necessary is that the area of the Perl stack which
775       held them will very likely have been overwritten by something else by
776       the time control returns from call_pv.
777
778   Using G_EVAL
779       Now an example using G_EVAL. Below is a Perl subroutine which computes
780       the difference of its 2 parameters. If this would result in a negative
781       result, the subroutine calls die.
782
783           sub Subtract
784           {
785               my ($a, $b) = @_;
786
787               die "death can be fatal\n" if $a < $b;
788
789               $a - $b;
790           }
791
792       and some C to call it
793
794        static void
795        call_Subtract(a, b)
796        int a;
797        int b;
798        {
799            dSP;
800            int count;
801            SV *err_tmp;
802
803            ENTER;
804            SAVETMPS;
805
806            PUSHMARK(SP);
807            EXTEND(SP, 2);
808            PUSHs(sv_2mortal(newSViv(a)));
809            PUSHs(sv_2mortal(newSViv(b)));
810            PUTBACK;
811
812            count = call_pv("Subtract", G_EVAL|G_SCALAR);
813
814            SPAGAIN;
815
816            /* Check the eval first */
817            err_tmp = ERRSV;
818            if (SvTRUE(err_tmp))
819            {
820                printf ("Uh oh - %s\n", SvPV_nolen(err_tmp));
821                POPs;
822            }
823            else
824            {
825              if (count != 1)
826               croak("call_Subtract: wanted 1 value from 'Subtract', got %d\n",
827                     count);
828
829                printf ("%d - %d = %d\n", a, b, POPi);
830            }
831
832            PUTBACK;
833            FREETMPS;
834            LEAVE;
835        }
836
837       If call_Subtract is called thus
838
839           call_Subtract(4, 5)
840
841       the following will be printed
842
843           Uh oh - death can be fatal
844
845       Notes
846
847       1.   We want to be able to catch the die so we have used the G_EVAL
848            flag.  Not specifying this flag would mean that the program would
849            terminate immediately at the die statement in the subroutine
850            Subtract.
851
852       2.   The code
853
854                err_tmp = ERRSV;
855                if (SvTRUE(err_tmp))
856                {
857                    printf ("Uh oh - %s\n", SvPV_nolen(err_tmp));
858                    POPs;
859                }
860
861            is the direct equivalent of this bit of Perl
862
863                print "Uh oh - $@\n" if $@;
864
865            "PL_errgv" is a perl global of type "GV *" that points to the
866            symbol table entry containing the error.  "ERRSV" therefore refers
867            to the C equivalent of $@.  We use a local temporary, "err_tmp",
868            since "ERRSV" is a macro that calls a function, and
869            "SvTRUE(ERRSV)" would end up calling that function multiple times.
870
871       3.   Note that the stack is popped using "POPs" in the block where
872            "SvTRUE(err_tmp)" is true.  This is necessary because whenever a
873            call_* function invoked with G_EVAL|G_SCALAR returns an error, the
874            top of the stack holds the value undef. Because we want the
875            program to continue after detecting this error, it is essential
876            that the stack be tidied up by removing the undef.
877
878   Using G_KEEPERR
879       Consider this rather facetious example, where we have used an XS
880       version of the call_Subtract example above inside a destructor:
881
882           package Foo;
883           sub new { bless {}, $_[0] }
884           sub Subtract {
885               my($a,$b) = @_;
886               die "death can be fatal" if $a < $b;
887               $a - $b;
888           }
889           sub DESTROY { call_Subtract(5, 4); }
890           sub foo { die "foo dies"; }
891
892           package main;
893           {
894               my $foo = Foo->new;
895               eval { $foo->foo };
896           }
897           print "Saw: $@" if $@;             # should be, but isn't
898
899       This example will fail to recognize that an error occurred inside the
900       "eval {}".  Here's why: the call_Subtract code got executed while perl
901       was cleaning up temporaries when exiting the outer braced block, and
902       because call_Subtract is implemented with call_pv using the G_EVAL
903       flag, it promptly reset $@.  This results in the failure of the
904       outermost test for $@, and thereby the failure of the error trap.
905
906       Appending the G_KEEPERR flag, so that the call_pv call in call_Subtract
907       reads:
908
909               count = call_pv("Subtract", G_EVAL|G_SCALAR|G_KEEPERR);
910
911       will preserve the error and restore reliable error handling.
912
913   Using call_sv
914       In all the previous examples I have 'hard-wired' the name of the Perl
915       subroutine to be called from C.  Most of the time though, it is more
916       convenient to be able to specify the name of the Perl subroutine from
917       within the Perl script, and you'll want to use call_sv.
918
919       Consider the Perl code below
920
921           sub fred
922           {
923               print "Hello there\n";
924           }
925
926           CallSubPV("fred");
927
928       Here is a snippet of XSUB which defines CallSubPV.
929
930           void
931           CallSubPV(name)
932               char *  name
933               CODE:
934               PUSHMARK(SP);
935               call_pv(name, G_DISCARD|G_NOARGS);
936
937       That is fine as far as it goes. The thing is, the Perl subroutine can
938       be specified as only a string, however, Perl allows references to
939       subroutines and anonymous subroutines.  This is where call_sv is
940       useful.
941
942       The code below for CallSubSV is identical to CallSubPV except that the
943       "name" parameter is now defined as an SV* and we use call_sv instead of
944       call_pv.
945
946           void
947           CallSubSV(name)
948               SV *    name
949               CODE:
950               PUSHMARK(SP);
951               call_sv(name, G_DISCARD|G_NOARGS);
952
953       Because we are using an SV to call fred the following can all be used:
954
955           CallSubSV("fred");
956           CallSubSV(\&fred);
957           $ref = \&fred;
958           CallSubSV($ref);
959           CallSubSV( sub { print "Hello there\n" } );
960
961       As you can see, call_sv gives you much greater flexibility in how you
962       can specify the Perl subroutine.
963
964       You should note that, if it is necessary to store the SV ("name" in the
965       example above) which corresponds to the Perl subroutine so that it can
966       be used later in the program, it not enough just to store a copy of the
967       pointer to the SV. Say the code above had been like this:
968
969           static SV * rememberSub;
970
971           void
972           SaveSub1(name)
973               SV *    name
974               CODE:
975               rememberSub = name;
976
977           void
978           CallSavedSub1()
979               CODE:
980               PUSHMARK(SP);
981               call_sv(rememberSub, G_DISCARD|G_NOARGS);
982
983       The reason this is wrong is that, by the time you come to use the
984       pointer "rememberSub" in "CallSavedSub1", it may or may not still refer
985       to the Perl subroutine that was recorded in "SaveSub1".  This is
986       particularly true for these cases:
987
988           SaveSub1(\&fred);
989           CallSavedSub1();
990
991           SaveSub1( sub { print "Hello there\n" } );
992           CallSavedSub1();
993
994       By the time each of the "SaveSub1" statements above has been executed,
995       the SV*s which corresponded to the parameters will no longer exist.
996       Expect an error message from Perl of the form
997
998           Can't use an undefined value as a subroutine reference at ...
999
1000       for each of the "CallSavedSub1" lines.
1001
1002       Similarly, with this code
1003
1004           $ref = \&fred;
1005           SaveSub1($ref);
1006           $ref = 47;
1007           CallSavedSub1();
1008
1009       you can expect one of these messages (which you actually get is
1010       dependent on the version of Perl you are using)
1011
1012           Not a CODE reference at ...
1013           Undefined subroutine &main::47 called ...
1014
1015       The variable $ref may have referred to the subroutine "fred" whenever
1016       the call to "SaveSub1" was made but by the time "CallSavedSub1" gets
1017       called it now holds the number 47. Because we saved only a pointer to
1018       the original SV in "SaveSub1", any changes to $ref will be tracked by
1019       the pointer "rememberSub". This means that whenever "CallSavedSub1"
1020       gets called, it will attempt to execute the code which is referenced by
1021       the SV* "rememberSub".  In this case though, it now refers to the
1022       integer 47, so expect Perl to complain loudly.
1023
1024       A similar but more subtle problem is illustrated with this code:
1025
1026           $ref = \&fred;
1027           SaveSub1($ref);
1028           $ref = \&joe;
1029           CallSavedSub1();
1030
1031       This time whenever "CallSavedSub1" gets called it will execute the Perl
1032       subroutine "joe" (assuming it exists) rather than "fred" as was
1033       originally requested in the call to "SaveSub1".
1034
1035       To get around these problems it is necessary to take a full copy of the
1036       SV.  The code below shows "SaveSub2" modified to do that.
1037
1038           /* this isn't thread-safe */
1039           static SV * keepSub = (SV*)NULL;
1040
1041           void
1042           SaveSub2(name)
1043               SV *    name
1044               CODE:
1045               /* Take a copy of the callback */
1046               if (keepSub == (SV*)NULL)
1047                   /* First time, so create a new SV */
1048                   keepSub = newSVsv(name);
1049               else
1050                   /* Been here before, so overwrite */
1051                   SvSetSV(keepSub, name);
1052
1053           void
1054           CallSavedSub2()
1055               CODE:
1056               PUSHMARK(SP);
1057               call_sv(keepSub, G_DISCARD|G_NOARGS);
1058
1059       To avoid creating a new SV every time "SaveSub2" is called, the
1060       function first checks to see if it has been called before.  If not,
1061       then space for a new SV is allocated and the reference to the Perl
1062       subroutine "name" is copied to the variable "keepSub" in one operation
1063       using "newSVsv".  Thereafter, whenever "SaveSub2" is called, the
1064       existing SV, "keepSub", is overwritten with the new value using
1065       "SvSetSV".
1066
1067       Note: using a static or global variable to store the SV isn't thread-
1068       safe.  You can either use the "MY_CXT" mechanism documented in "Safely
1069       Storing Static Data in XS" in perlxs which is fast, or store the values
1070       in perl global variables, using get_sv(), which is much slower.
1071
1072   Using call_argv
1073       Here is a Perl subroutine which prints whatever parameters are passed
1074       to it.
1075
1076           sub PrintList
1077           {
1078               my(@list) = @_;
1079
1080               foreach (@list) { print "$_\n" }
1081           }
1082
1083       And here is an example of call_argv which will call PrintList.
1084
1085           static char * words[] = {"alpha", "beta", "gamma", "delta", NULL};
1086
1087           static void
1088           call_PrintList()
1089           {
1090               call_argv("PrintList", G_DISCARD, words);
1091           }
1092
1093       Note that it is not necessary to call "PUSHMARK" in this instance.
1094       This is because call_argv will do it for you.
1095
1096   Using call_method
1097       Consider the following Perl code:
1098
1099           {
1100               package Mine;
1101
1102               sub new
1103               {
1104                   my($type) = shift;
1105                   bless [@_]
1106               }
1107
1108               sub Display
1109               {
1110                   my ($self, $index) = @_;
1111                   print "$index: $$self[$index]\n";
1112               }
1113
1114               sub PrintID
1115               {
1116                   my($class) = @_;
1117                   print "This is Class $class version 1.0\n";
1118               }
1119           }
1120
1121       It implements just a very simple class to manage an array.  Apart from
1122       the constructor, "new", it declares methods, one static and one
1123       virtual. The static method, "PrintID", prints out simply the class name
1124       and a version number. The virtual method, "Display", prints out a
1125       single element of the array.  Here is an all-Perl example of using it.
1126
1127           $a = Mine->new('red', 'green', 'blue');
1128           $a->Display(1);
1129           Mine->PrintID;
1130
1131       will print
1132
1133           1: green
1134           This is Class Mine version 1.0
1135
1136       Calling a Perl method from C is fairly straightforward. The following
1137       things are required:
1138
1139       •    A reference to the object for a virtual method or the name of the
1140            class for a static method
1141
1142       •    The name of the method
1143
1144       •    Any other parameters specific to the method
1145
1146       Here is a simple XSUB which illustrates the mechanics of calling both
1147       the "PrintID" and "Display" methods from C.
1148
1149           void
1150           call_Method(ref, method, index)
1151               SV *    ref
1152               char *  method
1153               int             index
1154               CODE:
1155               PUSHMARK(SP);
1156               EXTEND(SP, 2);
1157               PUSHs(ref);
1158               PUSHs(sv_2mortal(newSViv(index)));
1159               PUTBACK;
1160
1161               call_method(method, G_DISCARD);
1162
1163           void
1164           call_PrintID(class, method)
1165               char *  class
1166               char *  method
1167               CODE:
1168               PUSHMARK(SP);
1169               XPUSHs(sv_2mortal(newSVpv(class, 0)));
1170               PUTBACK;
1171
1172               call_method(method, G_DISCARD);
1173
1174       So the methods "PrintID" and "Display" can be invoked like this:
1175
1176           $a = Mine->new('red', 'green', 'blue');
1177           call_Method($a, 'Display', 1);
1178           call_PrintID('Mine', 'PrintID');
1179
1180       The only thing to note is that, in both the static and virtual methods,
1181       the method name is not passed via the stack--it is used as the first
1182       parameter to call_method.
1183
1184   Using GIMME_V
1185       Here is a trivial XSUB which prints the context in which it is
1186       currently executing.
1187
1188           void
1189           PrintContext()
1190               CODE:
1191               U8 gimme = GIMME_V;
1192               if (gimme == G_VOID)
1193                   printf ("Context is Void\n");
1194               else if (gimme == G_SCALAR)
1195                   printf ("Context is Scalar\n");
1196               else
1197                   printf ("Context is Array\n");
1198
1199       And here is some Perl to test it.
1200
1201           PrintContext;
1202           $a = PrintContext;
1203           @a = PrintContext;
1204
1205       The output from that will be
1206
1207           Context is Void
1208           Context is Scalar
1209           Context is Array
1210
1211   Using Perl to Dispose of Temporaries
1212       In the examples given to date, any temporaries created in the callback
1213       (i.e., parameters passed on the stack to the call_* function or values
1214       returned via the stack) have been freed by one of these methods:
1215
1216       •    Specifying the G_DISCARD flag with call_*
1217
1218       •    Explicitly using the "ENTER"/"SAVETMPS"--"FREETMPS"/"LEAVE"
1219            pairing
1220
1221       There is another method which can be used, namely letting Perl do it
1222       for you automatically whenever it regains control after the callback
1223       has terminated.  This is done by simply not using the
1224
1225           ENTER;
1226           SAVETMPS;
1227           ...
1228           FREETMPS;
1229           LEAVE;
1230
1231       sequence in the callback (and not, of course, specifying the G_DISCARD
1232       flag).
1233
1234       If you are going to use this method you have to be aware of a possible
1235       memory leak which can arise under very specific circumstances.  To
1236       explain these circumstances you need to know a bit about the flow of
1237       control between Perl and the callback routine.
1238
1239       The examples given at the start of the document (an error handler and
1240       an event driven program) are typical of the two main sorts of flow
1241       control that you are likely to encounter with callbacks.  There is a
1242       very important distinction between them, so pay attention.
1243
1244       In the first example, an error handler, the flow of control could be as
1245       follows.  You have created an interface to an external library.
1246       Control can reach the external library like this
1247
1248           perl --> XSUB --> external library
1249
1250       Whilst control is in the library, an error condition occurs. You have
1251       previously set up a Perl callback to handle this situation, so it will
1252       get executed. Once the callback has finished, control will drop back to
1253       Perl again.  Here is what the flow of control will be like in that
1254       situation
1255
1256           perl --> XSUB --> external library
1257                             ...
1258                             error occurs
1259                             ...
1260                             external library --> call_* --> perl
1261                                                                 |
1262           perl <-- XSUB <-- external library <-- call_* <----+
1263
1264       After processing of the error using call_* is completed, control
1265       reverts back to Perl more or less immediately.
1266
1267       In the diagram, the further right you go the more deeply nested the
1268       scope is.  It is only when control is back with perl on the extreme
1269       left of the diagram that you will have dropped back to the enclosing
1270       scope and any temporaries you have left hanging around will be freed.
1271
1272       In the second example, an event driven program, the flow of control
1273       will be more like this
1274
1275           perl --> XSUB --> event handler
1276                             ...
1277                             event handler --> call_* --> perl
1278                                                              |
1279                             event handler <-- call_* <----+
1280                             ...
1281                             event handler --> call_* --> perl
1282                                                              |
1283                             event handler <-- call_* <----+
1284                             ...
1285                             event handler --> call_* --> perl
1286                                                              |
1287                             event handler <-- call_* <----+
1288
1289       In this case the flow of control can consist of only the repeated
1290       sequence
1291
1292           event handler --> call_* --> perl
1293
1294       for practically the complete duration of the program.  This means that
1295       control may never drop back to the surrounding scope in Perl at the
1296       extreme left.
1297
1298       So what is the big problem? Well, if you are expecting Perl to tidy up
1299       those temporaries for you, you might be in for a long wait.  For Perl
1300       to dispose of your temporaries, control must drop back to the enclosing
1301       scope at some stage.  In the event driven scenario that may never
1302       happen.  This means that, as time goes on, your program will create
1303       more and more temporaries, none of which will ever be freed. As each of
1304       these temporaries consumes some memory your program will eventually
1305       consume all the available memory in your system--kapow!
1306
1307       So here is the bottom line--if you are sure that control will revert
1308       back to the enclosing Perl scope fairly quickly after the end of your
1309       callback, then it isn't absolutely necessary to dispose explicitly of
1310       any temporaries you may have created. Mind you, if you are at all
1311       uncertain about what to do, it doesn't do any harm to tidy up anyway.
1312
1313   Strategies for Storing Callback Context Information
1314       Potentially one of the trickiest problems to overcome when designing a
1315       callback interface can be figuring out how to store the mapping between
1316       the C callback function and the Perl equivalent.
1317
1318       To help understand why this can be a real problem first consider how a
1319       callback is set up in an all C environment.  Typically a C API will
1320       provide a function to register a callback.  This will expect a pointer
1321       to a function as one of its parameters.  Below is a call to a
1322       hypothetical function "register_fatal" which registers the C function
1323       to get called when a fatal error occurs.
1324
1325           register_fatal(cb1);
1326
1327       The single parameter "cb1" is a pointer to a function, so you must have
1328       defined "cb1" in your code, say something like this
1329
1330           static void
1331           cb1()
1332           {
1333               printf ("Fatal Error\n");
1334               exit(1);
1335           }
1336
1337       Now change that to call a Perl subroutine instead
1338
1339           static SV * callback = (SV*)NULL;
1340
1341           static void
1342           cb1()
1343           {
1344               dSP;
1345
1346               PUSHMARK(SP);
1347
1348               /* Call the Perl sub to process the callback */
1349               call_sv(callback, G_DISCARD);
1350           }
1351
1352
1353           void
1354           register_fatal(fn)
1355               SV *    fn
1356               CODE:
1357               /* Remember the Perl sub */
1358               if (callback == (SV*)NULL)
1359                   callback = newSVsv(fn);
1360               else
1361                   SvSetSV(callback, fn);
1362
1363               /* register the callback with the external library */
1364               register_fatal(cb1);
1365
1366       where the Perl equivalent of "register_fatal" and the callback it
1367       registers, "pcb1", might look like this
1368
1369           # Register the sub pcb1
1370           register_fatal(\&pcb1);
1371
1372           sub pcb1
1373           {
1374               die "I'm dying...\n";
1375           }
1376
1377       The mapping between the C callback and the Perl equivalent is stored in
1378       the global variable "callback".
1379
1380       This will be adequate if you ever need to have only one callback
1381       registered at any time. An example could be an error handler like the
1382       code sketched out above. Remember though, repeated calls to
1383       "register_fatal" will replace the previously registered callback
1384       function with the new one.
1385
1386       Say for example you want to interface to a library which allows
1387       asynchronous file i/o.  In this case you may be able to register a
1388       callback whenever a read operation has completed. To be of any use we
1389       want to be able to call separate Perl subroutines for each file that is
1390       opened.  As it stands, the error handler example above would not be
1391       adequate as it allows only a single callback to be defined at any time.
1392       What we require is a means of storing the mapping between the opened
1393       file and the Perl subroutine we want to be called for that file.
1394
1395       Say the i/o library has a function "asynch_read" which associates a C
1396       function "ProcessRead" with a file handle "fh"--this assumes that it
1397       has also provided some routine to open the file and so obtain the file
1398       handle.
1399
1400           asynch_read(fh, ProcessRead)
1401
1402       This may expect the C ProcessRead function of this form
1403
1404           void
1405           ProcessRead(fh, buffer)
1406           int fh;
1407           char *      buffer;
1408           {
1409                ...
1410           }
1411
1412       To provide a Perl interface to this library we need to be able to map
1413       between the "fh" parameter and the Perl subroutine we want called.  A
1414       hash is a convenient mechanism for storing this mapping.  The code
1415       below shows a possible implementation
1416
1417           static HV * Mapping = (HV*)NULL;
1418
1419           void
1420           asynch_read(fh, callback)
1421               int     fh
1422               SV *    callback
1423               CODE:
1424               /* If the hash doesn't already exist, create it */
1425               if (Mapping == (HV*)NULL)
1426                   Mapping = newHV();
1427
1428               /* Save the fh -> callback mapping */
1429               hv_store(Mapping, (char*)&fh, sizeof(fh), newSVsv(callback), 0);
1430
1431               /* Register with the C Library */
1432               asynch_read(fh, asynch_read_if);
1433
1434       and "asynch_read_if" could look like this
1435
1436           static void
1437           asynch_read_if(fh, buffer)
1438           int fh;
1439           char *      buffer;
1440           {
1441               dSP;
1442               SV ** sv;
1443
1444               /* Get the callback associated with fh */
1445               sv =  hv_fetch(Mapping, (char*)&fh , sizeof(fh), FALSE);
1446               if (sv == (SV**)NULL)
1447                   croak("Internal error...\n");
1448
1449               PUSHMARK(SP);
1450               EXTEND(SP, 2);
1451               PUSHs(sv_2mortal(newSViv(fh)));
1452               PUSHs(sv_2mortal(newSVpv(buffer, 0)));
1453               PUTBACK;
1454
1455               /* Call the Perl sub */
1456               call_sv(*sv, G_DISCARD);
1457           }
1458
1459       For completeness, here is "asynch_close".  This shows how to remove the
1460       entry from the hash "Mapping".
1461
1462           void
1463           asynch_close(fh)
1464               int     fh
1465               CODE:
1466               /* Remove the entry from the hash */
1467               (void) hv_delete(Mapping, (char*)&fh, sizeof(fh), G_DISCARD);
1468
1469               /* Now call the real asynch_close */
1470               asynch_close(fh);
1471
1472       So the Perl interface would look like this
1473
1474           sub callback1
1475           {
1476               my($handle, $buffer) = @_;
1477           }
1478
1479           # Register the Perl callback
1480           asynch_read($fh, \&callback1);
1481
1482           asynch_close($fh);
1483
1484       The mapping between the C callback and Perl is stored in the global
1485       hash "Mapping" this time. Using a hash has the distinct advantage that
1486       it allows an unlimited number of callbacks to be registered.
1487
1488       What if the interface provided by the C callback doesn't contain a
1489       parameter which allows the file handle to Perl subroutine mapping?  Say
1490       in the asynchronous i/o package, the callback function gets passed only
1491       the "buffer" parameter like this
1492
1493           void
1494           ProcessRead(buffer)
1495           char *      buffer;
1496           {
1497               ...
1498           }
1499
1500       Without the file handle there is no straightforward way to map from the
1501       C callback to the Perl subroutine.
1502
1503       In this case a possible way around this problem is to predefine a
1504       series of C functions to act as the interface to Perl, thus
1505
1506           #define MAX_CB              3
1507           #define NULL_HANDLE -1
1508           typedef void (*FnMap)();
1509
1510           struct MapStruct {
1511               FnMap    Function;
1512               SV *     PerlSub;
1513               int      Handle;
1514             };
1515
1516           static void  fn1();
1517           static void  fn2();
1518           static void  fn3();
1519
1520           static struct MapStruct Map [MAX_CB] =
1521               {
1522                   { fn1, NULL, NULL_HANDLE },
1523                   { fn2, NULL, NULL_HANDLE },
1524                   { fn3, NULL, NULL_HANDLE }
1525               };
1526
1527           static void
1528           Pcb(index, buffer)
1529           int index;
1530           char * buffer;
1531           {
1532               dSP;
1533
1534               PUSHMARK(SP);
1535               XPUSHs(sv_2mortal(newSVpv(buffer, 0)));
1536               PUTBACK;
1537
1538               /* Call the Perl sub */
1539               call_sv(Map[index].PerlSub, G_DISCARD);
1540           }
1541
1542           static void
1543           fn1(buffer)
1544           char * buffer;
1545           {
1546               Pcb(0, buffer);
1547           }
1548
1549           static void
1550           fn2(buffer)
1551           char * buffer;
1552           {
1553               Pcb(1, buffer);
1554           }
1555
1556           static void
1557           fn3(buffer)
1558           char * buffer;
1559           {
1560               Pcb(2, buffer);
1561           }
1562
1563           void
1564           array_asynch_read(fh, callback)
1565               int             fh
1566               SV *    callback
1567               CODE:
1568               int index;
1569               int null_index = MAX_CB;
1570
1571               /* Find the same handle or an empty entry */
1572               for (index = 0; index < MAX_CB; ++index)
1573               {
1574                   if (Map[index].Handle == fh)
1575                       break;
1576
1577                   if (Map[index].Handle == NULL_HANDLE)
1578                       null_index = index;
1579               }
1580
1581               if (index == MAX_CB && null_index == MAX_CB)
1582                   croak ("Too many callback functions registered\n");
1583
1584               if (index == MAX_CB)
1585                   index = null_index;
1586
1587               /* Save the file handle */
1588               Map[index].Handle = fh;
1589
1590               /* Remember the Perl sub */
1591               if (Map[index].PerlSub == (SV*)NULL)
1592                   Map[index].PerlSub = newSVsv(callback);
1593               else
1594                   SvSetSV(Map[index].PerlSub, callback);
1595
1596               asynch_read(fh, Map[index].Function);
1597
1598           void
1599           array_asynch_close(fh)
1600               int     fh
1601               CODE:
1602               int index;
1603
1604               /* Find the file handle */
1605               for (index = 0; index < MAX_CB; ++ index)
1606                   if (Map[index].Handle == fh)
1607                       break;
1608
1609               if (index == MAX_CB)
1610                   croak ("could not close fh %d\n", fh);
1611
1612               Map[index].Handle = NULL_HANDLE;
1613               SvREFCNT_dec(Map[index].PerlSub);
1614               Map[index].PerlSub = (SV*)NULL;
1615
1616               asynch_close(fh);
1617
1618       In this case the functions "fn1", "fn2", and "fn3" are used to remember
1619       the Perl subroutine to be called. Each of the functions holds a
1620       separate hard-wired index which is used in the function "Pcb" to access
1621       the "Map" array and actually call the Perl subroutine.
1622
1623       There are some obvious disadvantages with this technique.
1624
1625       Firstly, the code is considerably more complex than with the previous
1626       example.
1627
1628       Secondly, there is a hard-wired limit (in this case 3) to the number of
1629       callbacks that can exist simultaneously. The only way to increase the
1630       limit is by modifying the code to add more functions and then
1631       recompiling.  None the less, as long as the number of functions is
1632       chosen with some care, it is still a workable solution and in some
1633       cases is the only one available.
1634
1635       To summarize, here are a number of possible methods for you to consider
1636       for storing the mapping between C and the Perl callback
1637
1638       1. Ignore the problem - Allow only 1 callback
1639            For a lot of situations, like interfacing to an error handler,
1640            this may be a perfectly adequate solution.
1641
1642       2. Create a sequence of callbacks - hard wired limit
1643            If it is impossible to tell from the parameters passed back from
1644            the C callback what the context is, then you may need to create a
1645            sequence of C callback interface functions, and store pointers to
1646            each in an array.
1647
1648       3. Use a parameter to map to the Perl callback
1649            A hash is an ideal mechanism to store the mapping between C and
1650            Perl.
1651
1652   Alternate Stack Manipulation
1653       Although I have made use of only the "POP*" macros to access values
1654       returned from Perl subroutines, it is also possible to bypass these
1655       macros and read the stack using the "ST" macro (See perlxs for a full
1656       description of the "ST" macro).
1657
1658       Most of the time the "POP*" macros should be adequate; the main problem
1659       with them is that they force you to process the returned values in
1660       sequence. This may not be the most suitable way to process the values
1661       in some cases. What we want is to be able to access the stack in a
1662       random order. The "ST" macro as used when coding an XSUB is ideal for
1663       this purpose.
1664
1665       The code below is the example given in the section "Returning a List of
1666       Values" recoded to use "ST" instead of "POP*".
1667
1668           static void
1669           call_AddSubtract2(a, b)
1670           int a;
1671           int b;
1672           {
1673               dSP;
1674               I32 ax;
1675               int count;
1676
1677               ENTER;
1678               SAVETMPS;
1679
1680               PUSHMARK(SP);
1681               EXTEND(SP, 2);
1682               PUSHs(sv_2mortal(newSViv(a)));
1683               PUSHs(sv_2mortal(newSViv(b)));
1684               PUTBACK;
1685
1686               count = call_pv("AddSubtract", G_LIST);
1687
1688               SPAGAIN;
1689               SP -= count;
1690               ax = (SP - PL_stack_base) + 1;
1691
1692               if (count != 2)
1693                   croak("Big trouble\n");
1694
1695               printf ("%d + %d = %d\n", a, b, SvIV(ST(0)));
1696               printf ("%d - %d = %d\n", a, b, SvIV(ST(1)));
1697
1698               PUTBACK;
1699               FREETMPS;
1700               LEAVE;
1701           }
1702
1703       Notes
1704
1705       1.   Notice that it was necessary to define the variable "ax".  This is
1706            because the "ST" macro expects it to exist.  If we were in an XSUB
1707            it would not be necessary to define "ax" as it is already defined
1708            for us.
1709
1710       2.   The code
1711
1712                    SPAGAIN;
1713                    SP -= count;
1714                    ax = (SP - PL_stack_base) + 1;
1715
1716            sets the stack up so that we can use the "ST" macro.
1717
1718       3.   Unlike the original coding of this example, the returned values
1719            are not accessed in reverse order.  So ST(0) refers to the first
1720            value returned by the Perl subroutine and "ST(count-1)" refers to
1721            the last.
1722
1723   Creating and Calling an Anonymous Subroutine in C
1724       As we've already shown, "call_sv" can be used to invoke an anonymous
1725       subroutine.  However, our example showed a Perl script invoking an XSUB
1726       to perform this operation.  Let's see how it can be done inside our C
1727       code:
1728
1729        ...
1730
1731        SV *cvrv
1732           = eval_pv("sub {
1733                       print 'You will not find me cluttering any namespace!'
1734                      }", TRUE);
1735
1736        ...
1737
1738        call_sv(cvrv, G_VOID|G_NOARGS);
1739
1740       "eval_pv" is used to compile the anonymous subroutine, which will be
1741       the return value as well (read more about "eval_pv" in "eval_pv" in
1742       perlapi).  Once this code reference is in hand, it can be mixed in with
1743       all the previous examples we've shown.
1744

LIGHTWEIGHT CALLBACKS

1746       Sometimes you need to invoke the same subroutine repeatedly.  This
1747       usually happens with a function that acts on a list of values, such as
1748       Perl's built-in sort(). You can pass a comparison function to sort(),
1749       which will then be invoked for every pair of values that needs to be
1750       compared. The first() and reduce() functions from List::Util follow a
1751       similar pattern.
1752
1753       In this case it is possible to speed up the routine (often quite
1754       substantially) by using the lightweight callback API.  The idea is that
1755       the calling context only needs to be created and destroyed once, and
1756       the sub can be called arbitrarily many times in between.
1757
1758       It is usual to pass parameters using global variables (typically $_ for
1759       one parameter, or $a and $b for two parameters) rather than via @_. (It
1760       is possible to use the @_ mechanism if you know what you're doing,
1761       though there is as yet no supported API for it. It's also inherently
1762       slower.)
1763
1764       The pattern of macro calls is like this:
1765
1766           dMULTICALL;                 /* Declare local variables */
1767           U8 gimme = G_SCALAR;        /* context of the call: G_SCALAR,
1768                                        * G_LIST, or G_VOID */
1769
1770           PUSH_MULTICALL(cv);         /* Set up the context for calling cv,
1771                                          and set local vars appropriately */
1772
1773           /* loop */ {
1774               /* set the value(s) af your parameter variables */
1775               MULTICALL;              /* Make the actual call */
1776           } /* end of loop */
1777
1778           POP_MULTICALL;              /* Tear down the calling context */
1779
1780       For some concrete examples, see the implementation of the first() and
1781       reduce() functions of List::Util 1.18. There you will also find a
1782       header file that emulates the multicall API on older versions of perl.
1783

SEE ALSO

1785       perlxs, perlguts, perlembed
1786

AUTHOR

1788       Paul Marquess
1789
1790       Special thanks to the following people who assisted in the creation of
1791       the document.
1792
1793       Jeff Okamoto, Tim Bunce, Nick Gianniotis, Steve Kelem, Gurusamy Sarathy
1794       and Larry Wall.
1795

DATE

1797       Last updated for perl 5.23.1.
1798
1799
1800
1801perl v5.36.0                      2022-08-30                       PERLCALL(1)
Impressum