4 Consistent formatting of this file is achieved with:
5 perl ./Porting/podtidy pod/perlinterp.pod
9 perlinterp - An overview of the Perl interpreter
13 This document provides an overview of how the Perl interpreter works at
14 the level of C code, along with pointers to the relevant C source code
17 =head1 ELEMENTS OF THE INTERPRETER
19 The work of the interpreter has two main stages: compiling the code
20 into the internal representation, or bytecode, and then executing it.
21 L<perlguts/Compiled code> explains exactly how the compilation stage
24 Here is a short breakdown of perl's operation:
28 The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl)
29 This is very high-level code, enough to fit on a single screen, and it
30 resembles the code found in L<perlembed>; most of the real action takes
33 F<perlmain.c> is generated by C<ExtUtils::Miniperl> from
34 F<miniperlmain.c> at make time, so you should make perl to follow this
37 First, F<perlmain.c> allocates some memory and constructs a Perl
38 interpreter, along these lines:
40 1 PERL_SYS_INIT3(&argc,&argv,&env);
42 3 if (!PL_do_undump) {
43 4 my_perl = perl_alloc();
46 7 perl_construct(my_perl);
47 8 PL_perl_destruct_level = 0;
50 Line 1 is a macro, and its definition is dependent on your operating
51 system. Line 3 references C<PL_do_undump>, a global variable - all
52 global variables in Perl start with C<PL_>. This tells you whether the
53 current running program was created with the C<-u> flag to perl and
54 then F<undump>, which means it's going to be false in any sane context.
56 Line 4 calls a function in F<perl.c> to allocate memory for a Perl
57 interpreter. It's quite a simple function, and the guts of it looks
60 my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
62 Here you see an example of Perl's system abstraction, which we'll see
63 later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's
64 own C<malloc> as defined in F<malloc.c> if you selected that option at
67 Next, in line 7, we construct the interpreter using perl_construct,
68 also in F<perl.c>; this sets up all the special variables that Perl
69 needs, the stacks, and so on.
71 Now we pass Perl the command line options, and tell it to go:
73 exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL);
77 exitstatus = perl_destruct(my_perl);
81 C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined
82 in F<perl.c>, which processes the command line options, sets up any
83 statically linked XS modules, opens the program and calls C<yyparse> to
88 The aim of this stage is to take the Perl source, and turn it into an
89 op tree. We'll see what one of those looks like later. Strictly
90 speaking, there's three things going on here.
92 C<yyparse>, the parser, lives in F<perly.c>, although you're better off
93 reading the original YACC input in F<perly.y>. (Yes, Virginia, there
94 B<is> a YACC grammar for Perl!) The job of the parser is to take your
95 code and "understand" it, splitting it into sentences, deciding which
96 operands go with which operators and so on.
98 The parser is nobly assisted by the lexer, which chunks up your input
99 into tokens, and decides what type of thing each token is: a variable
100 name, an operator, a bareword, a subroutine, a core function, and so
101 on. The main point of entry to the lexer is C<yylex>, and that and its
102 associated routines can be found in F<toke.c>. Perl isn't much like
103 other computer languages; it's highly context sensitive at times, it
104 can be tricky to work out what sort of token something is, or where a
105 token ends. As such, there's a lot of interplay between the tokeniser
106 and the parser, which can get pretty frightening if you're not used to
109 As the parser understands a Perl program, it builds up a tree of
110 operations for the interpreter to perform during execution. The
111 routines which construct and link together the various operations are
112 to be found in F<op.c>, and will be examined later.
116 Now the parsing stage is complete, and the finished tree represents the
117 operations that the Perl interpreter needs to perform to execute our
118 program. Next, Perl does a dry run over the tree looking for
119 optimisations: constant expressions such as C<3 + 4> will be computed
120 now, and the optimizer will also see if any multiple operations can be
121 replaced with a single one. For instance, to fetch the variable
122 C<$foo>, instead of grabbing the glob C<*foo> and looking at the scalar
123 component, the optimizer fiddles the op tree to use a function which
124 directly looks up the scalar in question. The main optimizer is C<peep>
125 in F<op.c>, and many ops have their own optimizing functions.
129 Now we're finally ready to go: we have compiled Perl byte code, and all
130 that's left to do is run it. The actual execution is done by the
131 C<runops_standard> function in F<run.c>; more specifically, it's done
132 by these three innocent looking lines:
134 while ((PL_op = PL_op->op_ppaddr(aTHX))) {
138 You may be more comfortable with the Perl version of that:
140 PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
142 Well, maybe not. Anyway, each op contains a function pointer, which
143 stipulates the function which will actually carry out the operation.
144 This function will return the next op in the sequence - this allows for
145 things like C<if> which choose the next op dynamically at run time. The
146 C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt
147 execution if required.
149 The actual functions called are known as PP code, and they're spread
150 between four files: F<pp_hot.c> contains the "hot" code, which is most
151 often used and highly optimized, F<pp_sys.c> contains all the
152 system-specific functions, F<pp_ctl.c> contains the functions which
153 implement control structures (C<if>, C<while> and the like) and F<pp.c>
154 contains everything else. These are, if you like, the C code for Perl's
155 built-in functions and operators.
157 Note that each C<pp_> function is expected to return a pointer to the
158 next op. Calls to perl subs (and eval blocks) are handled within the
159 same runops loop, and do not consume extra space on the C stack. For
160 example, C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or
161 C<CxEVAL> block struct onto the context stack which contain the address
162 of the op following the sub call or eval. They then return the first op
163 of that sub or eval block, and so execution continues of that sub or
164 block. Later, a C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB>
165 or C<CxEVAL>, retrieves the return op from it, and returns it.
167 =head2 Exception handing
169 Perl's exception handing (i.e. C<die> etc.) is built on top of the
170 low-level C<setjmp()>/C<longjmp()> C-library functions. These basically
171 provide a way to capture the current PC and SP registers and later
172 restore them; i.e. a C<longjmp()> continues at the point in code where
173 a previous C<setjmp()> was done, with anything further up on the C
174 stack being lost. This is why code should always save values using
175 C<SAVE_FOO> rather than in auto variables.
177 The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and
178 C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and
179 C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while
180 C<die> within C<eval> does a C<JMPENV_JUMP(3)>.
182 At entry points to perl, such as C<perl_parse()>, C<perl_run()> and
183 C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops
184 loop or whatever, and handle possible exception returns. For a 2
185 return, final cleanup is performed, such as popping stacks and calling
186 C<CHECK> or C<END> blocks. Amongst other things, this is how scope
187 cleanup still occurs during an C<exit>.
189 If a C<die> can find a C<CxEVAL> block on the context stack, then the
190 stack is popped to that level and the return op in that block is
191 assigned to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed.
192 This normally passes control back to the guard. In the case of
193 C<perl_run> and C<call_sv>, a non-null C<PL_restartop> triggers
194 re-entry to the runops loop. The is the normal way that C<die> or
195 C<croak> is handled within an C<eval>.
197 Sometimes ops are executed within an inner runops loop, such as tie,
198 sort or overload code. In this case, something like
200 sub FETCH { eval { die } }
202 would cause a longjmp right back to the guard in C<perl_run>, popping
203 both runops loops, which is clearly incorrect. One way to avoid this is
204 for the tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in
205 the inner runops loop, but for efficiency reasons, perl in fact just
206 sets a flag, using C<CATCH_SET(TRUE)>. The C<pp_require>,
207 C<pp_entereval> and C<pp_entertry> ops check this flag, and if true,
208 they call C<docatch>, which does a C<JMPENV_PUSH> and starts a new
209 runops level to execute the code, rather than doing it on the current
212 As a further optimisation, on exit from the eval block in the C<FETCH>,
213 execution of the code following the block is still carried on in the
214 inner loop. When an exception is raised, C<docatch> compares the
215 C<JMPENV> level of the C<CxEVAL> with C<PL_top_env> and if they differ,
216 just re-throws the exception. In this way any inner loops get popped.
220 1: eval { tie @a, 'A' };
226 To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH>
227 then enters a runops loop. This loop executes the eval and tie ops on
228 line 1, with the eval pushing a C<CxEVAL> onto the context stack.
230 The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops
231 loop to execute the body of C<TIEARRAY>. When it executes the entertry
232 op on line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch>
233 which does a C<JMPENV_PUSH> and starts a third runops loop, which then
234 executes the die op. At this point the C call stack looks like this:
237 Perl_runops # third loop
241 Perl_runops # second loop
245 Perl_runops # first loop
250 and the context and data stacks, as shown by C<-Dstv>, look like:
254 CX 1: EVAL => AV() PV("A"\0)
262 The die pops the first C<CxEVAL> off the context stack, sets
263 C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns
264 to the top C<docatch>. This then starts another third-level runops
265 level, which executes the nextstate, pushmark and die ops on line 4. At
266 the point that the second C<pp_die> is called, the C call stack looks
267 exactly like that above, even though we are no longer within an inner
268 eval; this is because of the optimization mentioned earlier. However,
269 the context stack now looks like this, ie with the top CxEVAL popped:
273 CX 1: EVAL => AV() PV("A"\0)
279 The die on line 4 pops the context stack back down to the CxEVAL,
285 As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a
286 C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch:
290 Perl_runops # second loop
294 Perl_runops # first loop
299 In this case, because the C<JMPENV> level recorded in the C<CxEVAL>
300 differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)>
301 and the C stack unwinds to:
306 Because C<PL_restartop> is non-null, C<run_body> starts a new runops
307 loop and execution continues.
309 =head2 INTERNAL VARIABLE TYPES
311 You should by now have had a look at L<perlguts>, which tells you about
312 Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
315 These variables are used not only to represent Perl-space variables,
316 but also any constants in the code, as well as some structures
317 completely internal to Perl. The symbol table, for instance, is an
318 ordinary Perl hash. Your code is represented by an SV as it's read into
319 the parser; any program files you call are opened via ordinary Perl
320 filehandles, and so on.
322 The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a
323 Perl program. Let's see, for instance, how Perl treats the constant
326 % perl -MDevel::Peek -e 'Dump("hello")'
327 1 SV = PV(0xa041450) at 0xa04ecbc
329 3 FLAGS = (POK,READONLY,pPOK)
330 4 PV = 0xa0484e0 "hello"\0
334 Reading C<Devel::Peek> output takes a bit of practise, so let's go
335 through it line by line.
337 Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in
338 memory. SVs themselves are very simple structures, but they contain a
339 pointer to a more complex structure. In this case, it's a PV, a
340 structure which holds a string value, at location C<0xa041450>. Line 2
341 is the reference count; there are no other references to this data, so
344 Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
345 read-only SV (because it's a constant) and the data is a PV internally.
346 Next we've got the contents of the string, starting at location
349 Line 5 gives us the current length of the string - note that this does
350 B<not> include the null terminator. Line 6 is not the length of the
351 string, but the length of the currently allocated buffer; as the string
352 grows, Perl automatically extends the available storage via a routine
355 You can get at any of these quantities from C very easily; just add
356 C<Sv> to the name of the field shown in the snippet, and you've got a
357 macro which will return the value: C<SvCUR(sv)> returns the current
358 length of the string, C<SvREFCOUNT(sv)> returns the reference count,
359 C<SvPV(sv, len)> returns the string itself with its length, and so on.
360 More macros to manipulate these properties can be found in L<perlguts>.
362 Let's take an example of manipulating a PV, from C<sv_catpvn>, in
366 2 Perl_sv_catpvn(pTHX_ SV *sv, const char *ptr, STRLEN len)
371 6 junk = SvPV_force(sv, tlen);
372 7 SvGROW(sv, tlen + len + 1);
375 10 Move(ptr,SvPVX(sv)+tlen,len,char);
377 12 *SvEND(sv) = '\0';
378 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */
382 This is a function which adds a string, C<ptr>, of length C<len> onto
383 the end of the PV stored in C<sv>. The first thing we do in line 6 is
384 make sure that the SV B<has> a valid PV, by calling the C<SvPV_force>
385 macro to force a PV. As a side effect, C<tlen> gets set to the current
386 value of the PV, and the PV itself is returned to C<junk>.
388 In line 7, we make sure that the SV will have enough room to
389 accommodate the old string, the new string and the null terminator. If
390 C<LEN> isn't big enough, C<SvGROW> will reallocate space for us.
392 Now, if C<junk> is the same as the string we're trying to add, we can
393 grab the string directly from the SV; C<SvPVX> is the address of the PV
396 Line 10 does the actual catenation: the C<Move> macro moves a chunk of
397 memory around: we move the string C<ptr> to the end of the PV - that's
398 the start of the PV plus its current length. We're moving C<len> bytes
399 of type C<char>. After doing so, we need to tell Perl we've extended
400 the string, by altering C<CUR> to reflect the new length. C<SvEND> is a
401 macro which gives us the end of the string, so that needs to be a
404 Line 13 manipulates the flags; since we've changed the PV, any IV or NV
405 values will no longer be valid: if we have C<$a=10; $a.="6";> we don't
406 want to use the old IV of 10. C<SvPOK_only_utf8> is a special
407 UTF-8-aware version of C<SvPOK_only>, a macro which turns off the IOK
408 and NOK flags and turns on POK. The final C<SvTAINT> is a macro which
409 launders tainted data if taint mode is turned on.
411 AVs and HVs are more complicated, but SVs are by far the most common
412 variable type being thrown around. Having seen something of how we
413 manipulate these, let's go on and look at how the op tree is
418 First, what is the op tree, anyway? The op tree is the parsed
419 representation of your program, as we saw in our section on parsing,
420 and it's the sequence of operations that Perl goes through to execute
421 your program, as we saw in L</Running>.
423 An op is a fundamental operation that Perl can perform: all the
424 built-in functions and operators are ops, and there are a series of ops
425 which deal with concepts the interpreter needs internally - entering
426 and leaving a block, ending a statement, fetching a variable, and so
429 The op tree is connected in two ways: you can imagine that there are
430 two "routes" through it, two orders in which you can traverse the tree.
431 First, parse order reflects how the parser understood the code, and
432 secondly, execution order tells perl what order to perform the
435 The easiest way to examine the op tree is to stop Perl after it has
436 finished parsing, and get it to dump out the tree. This is exactly what
437 the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise>
438 and L<B::Debug|B::Debug> do.
440 Let's have a look at how Perl sees C<$a = $b + $c>:
442 % perl -MO=Terse -e '$a=$b+$c'
443 1 LISTOP (0x8179888) leave
444 2 OP (0x81798b0) enter
445 3 COP (0x8179850) nextstate
446 4 BINOP (0x8179828) sassign
447 5 BINOP (0x8179800) add [1]
448 6 UNOP (0x81796e0) null [15]
449 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b
450 8 UNOP (0x81797e0) null [15]
451 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c
452 10 UNOP (0x816b4f0) null [15]
453 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a
455 Let's start in the middle, at line 4. This is a BINOP, a binary
456 operator, which is at location C<0x8179828>. The specific operator in
457 question is C<sassign> - scalar assignment - and you can find the code
458 which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a
459 binary operator, it has two children: the add operator, providing the
460 result of C<$b+$c>, is uppermost on line 5, and the left hand side is
463 Line 10 is the null op: this does exactly nothing. What is that doing
464 there? If you see the null op, it's a sign that something has been
465 optimized away after parsing. As we mentioned in L</Optimization>, the
466 optimization stage sometimes converts two operations into one, for
467 example when fetching a scalar variable. When this happens, instead of
468 rewriting the op tree and cleaning up the dangling pointers, it's
469 easier just to replace the redundant operation with the null op.
470 Originally, the tree would have looked like this:
472 10 SVOP (0x816b4f0) rv2sv [15]
473 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a
475 That is, fetch the C<a> entry from the main symbol table, and then look
476 at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>)
477 happens to do both these things.
479 The right hand side, starting at line 5 is similar to what we've just
480 seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add
481 together two C<gvsv>s.
483 Now, what's this about?
485 1 LISTOP (0x8179888) leave
486 2 OP (0x81798b0) enter
487 3 COP (0x8179850) nextstate
489 C<enter> and C<leave> are scoping ops, and their job is to perform any
490 housekeeping every time you enter and leave a block: lexical variables
491 are tidied up, unreferenced variables are destroyed, and so on. Every
492 program will have those first three lines: C<leave> is a list, and its
493 children are all the statements in the block. Statements are delimited
494 by C<nextstate>, so a block is a collection of C<nextstate> ops, with
495 the ops to be performed for each statement being the children of
496 C<nextstate>. C<enter> is a single op which functions as a marker.
498 That's how Perl parsed the program, from top to bottom:
511 However, it's impossible to B<perform> the operations in this order:
512 you have to find the values of C<$b> and C<$c> before you add them
513 together, for instance. So, the other thread that runs through the op
514 tree is the execution order: each op has a field C<op_next> which
515 points to the next op to be run, so following these pointers tells us
516 how perl executes the code. We can traverse the tree in this order
517 using the C<exec> option to C<B::Terse>:
519 % perl -MO=Terse,exec -e '$a=$b+$c'
520 1 OP (0x8179928) enter
521 2 COP (0x81798c8) nextstate
522 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b
523 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c
524 5 BINOP (0x8179878) add [1]
525 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a
526 7 BINOP (0x81798a0) sassign
527 8 LISTOP (0x8179900) leave
529 This probably makes more sense for a human: enter a block, start a
530 statement. Get the values of C<$b> and C<$c>, and add them together.
531 Find C<$a>, and assign one to the other. Then leave.
533 The way Perl builds up these op trees in the parsing process can be
534 unravelled by examining F<perly.y>, the YACC grammar. Let's take the
535 piece we need to construct the tree for C<$a = $b + $c>
537 1 term : term ASSIGNOP term
538 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
540 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
542 If you're not used to reading BNF grammars, this is how it works:
543 You're fed certain things by the tokeniser, which generally end up in
544 upper case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in
545 your code. C<ASSIGNOP> is provided when C<=> is used for assigning.
546 These are "terminal symbols", because you can't get any simpler than
549 The grammar, lines one and three of the snippet above, tells you how to
550 build up more complex forms. These complex forms, "non-terminal
551 symbols" are generally placed in lower case. C<term> here is a
552 non-terminal symbol, representing a single expression.
554 The grammar gives you the following rule: you can make the thing on the
555 left of the colon if you see all the things on the right in sequence.
556 This is called a "reduction", and the aim of parsing is to completely
557 reduce the input. There are several different ways you can perform a
558 reduction, separated by vertical bars: so, C<term> followed by C<=>
559 followed by C<term> makes a C<term>, and C<term> followed by C<+>
560 followed by C<term> can also make a C<term>.
562 So, if you see two terms with an C<=> or C<+>, between them, you can
563 turn them into a single expression. When you do this, you execute the
564 code in the block on the next line: if you see C<=>, you'll do the code
565 in line 2. If you see C<+>, you'll do the code in line 4. It's this
566 code which contributes to the op tree.
569 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
571 What this does is creates a new binary op, and feeds it a number of
572 variables. The variables refer to the tokens: C<$1> is the first token
573 in the input, C<$2> the second, and so on - think regular expression
574 backreferences. C<$$> is the op returned from this reduction. So, we
575 call C<newBINOP> to create a new binary operator. The first parameter
576 to C<newBINOP>, a function in F<op.c>, is the op type. It's an addition
577 operator, so we want the type to be C<ADDOP>. We could specify this
578 directly, but it's right there as the second token in the input, so we
579 use C<$2>. The second parameter is the op's flags: 0 means "nothing
580 special". Then the things to add: the left and right hand side of our
581 expression, in scalar context.
585 When perl executes something like C<addop>, how does it pass on its
586 results to the next op? The answer is, through the use of stacks. Perl
587 has a number of stacks to store things it's currently working on, and
588 we'll look at the three most important ones here.
590 =head2 Argument stack
592 Arguments are passed to PP code and returned from PP code using the
593 argument stack, C<ST>. The typical way to handle arguments is to pop
594 them off the stack, deal with them how you wish, and then push the
595 result back onto the stack. This is how, for instance, the cosine
600 value = Perl_cos(value);
603 We'll see a more tricky example of this when we consider Perl's macros
604 below. C<POPn> gives you the NV (floating point value) of the top SV on
605 the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and
606 push the result back as an NV. The C<X> in C<XPUSHn> means that the
607 stack should be extended if necessary - it can't be necessary here,
608 because we know there's room for one more item on the stack, since
609 we've just removed one! The C<XPUSH*> macros at least guarantee safety.
611 Alternatively, you can fiddle with the stack directly: C<SP> gives you
612 the first element in your portion of the stack, and C<TOP*> gives you
613 the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
614 negation of an integer:
618 Just set the integer value of the top stack entry to its negation.
620 Argument stack manipulation in the core is exactly the same as it is in
621 XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer
622 description of the macros used in stack manipulation.
626 I say "your portion of the stack" above because PP code doesn't
627 necessarily get the whole stack to itself: if your function calls
628 another function, you'll only want to expose the arguments aimed for
629 the called function, and not (necessarily) let it get at your own data.
630 The way we do this is to have a "virtual" bottom-of-stack, exposed to
631 each function. The mark stack keeps bookmarks to locations in the
632 argument stack usable by each function. For instance, when dealing with
633 a tied variable, (internally, something with "P" magic) Perl has to
634 call methods for accesses to the tied variables. However, we need to
635 separate the arguments exposed to the method to the argument exposed to
636 the original function - the store or fetch or whatever it may be.
637 Here's roughly how the tied C<push> is implemented; see C<av_push> in
642 3 PUSHs(SvTIED_obj((SV*)av, mg));
646 7 call_method("PUSH", G_SCALAR|G_DISCARD);
649 Let's examine the whole implementation, for practice:
653 Push the current state of the stack pointer onto the mark stack. This
654 is so that when we've finished adding items to the argument stack, Perl
655 knows how many things we've added recently.
658 3 PUSHs(SvTIED_obj((SV*)av, mg));
661 We're going to add two more items onto the argument stack: when you
662 have a tied array, the C<PUSH> subroutine receives the object and the
663 value to be pushed, and that's exactly what we have here - the tied
664 object, retrieved with C<SvTIED_obj>, and the value, the SV C<val>.
668 Next we tell Perl to update the global stack pointer from our internal
669 variable: C<dSP> only gave us a local copy, not a reference to the
673 7 call_method("PUSH", G_SCALAR|G_DISCARD);
676 C<ENTER> and C<LEAVE> localise a block of code - they make sure that
677 all variables are tidied up, everything that has been localised gets
678 its previous value returned, and so on. Think of them as the C<{> and
679 C<}> of a Perl block.
681 To actually do the magic method call, we have to call a subroutine in
682 Perl space: C<call_method> takes care of that, and it's described in
683 L<perlcall>. We call the C<PUSH> method in scalar context, and we're
684 going to discard its return value. The call_method() function removes
685 the top element of the mark stack, so there is nothing for the caller
690 C doesn't have a concept of local scope, so perl provides one. We've
691 seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save
692 stack implements the C equivalent of, for example:
699 See L<perlguts/"Localizing changes"> for how to use the save stack.
701 =head1 MILLIONS OF MACROS
703 One thing you'll notice about the Perl source is that it's full of
704 macros. Some have called the pervasive use of macros the hardest thing
705 to understand, others find it adds to clarity. Let's take an example,
706 the code which implements the addition operator:
710 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN);
713 6 SETn( left + right );
718 Every line here (apart from the braces, of course) contains a macro.
719 The first line sets up the function declaration as Perl expects for PP
720 code; line 3 sets up variable declarations for the argument stack and
721 the target, the return value of the operation. Finally, it tries to see
722 if the addition operation is overloaded; if so, the appropriate
723 subroutine is called.
725 Line 5 is another variable declaration - all variable declarations
726 start with C<d> - which pops from the top of the argument stack two NVs
727 (hence C<nn>) and puts them into the variables C<right> and C<left>,
728 hence the C<rl>. These are the two operands to the addition operator.
729 Next, we call C<SETn> to set the NV of the return value to the result
730 of adding the two values. This done, we return - the C<RETURN> macro
731 makes sure that our return value is properly handled, and we pass the
732 next operator to run back to the main run loop.
734 Most of these macros are explained in L<perlapi>, and some of the more
735 important ones are explained in L<perlxs> as well. Pay special
736 attention to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for
737 information on the C<[pad]THX_?> macros.
739 =head1 FURTHER READING
741 For more information on the Perl internals, please see the documents
742 listed at L<perl/Internals and C Language Interface>.