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

NAME

6       perlinterp - An overview of the Perl interpreter
7

DESCRIPTION

9       This document provides an overview of how the Perl interpreter works at
10       the level of C code, along with pointers to the relevant C source code
11       files.
12

ELEMENTS OF THE INTERPRETER

14       The work of the interpreter has two main stages: compiling the code
15       into the internal representation, or bytecode, and then executing it.
16       "Compiled code" in perlguts explains exactly how the compilation stage
17       happens.
18
19       Here is a short breakdown of perl's operation:
20
21   Startup
22       The action begins in perlmain.c. (or miniperlmain.c for miniperl) This
23       is very high-level code, enough to fit on a single screen, and it
24       resembles the code found in perlembed; most of the real action takes
25       place in perl.c
26
27       perlmain.c is generated by "ExtUtils::Miniperl" from miniperlmain.c at
28       make time, so you should make perl to follow this along.
29
30       First, perlmain.c allocates some memory and constructs a Perl
31       interpreter, along these lines:
32
33           1 PERL_SYS_INIT3(&argc,&argv,&env);
34           2
35           3 if (!PL_do_undump) {
36           4     my_perl = perl_alloc();
37           5     if (!my_perl)
38           6         exit(1);
39           7     perl_construct(my_perl);
40           8     PL_perl_destruct_level = 0;
41           9 }
42
43       Line 1 is a macro, and its definition is dependent on your operating
44       system. Line 3 references "PL_do_undump", a global variable - all
45       global variables in Perl start with "PL_". This tells you whether the
46       current running program was created with the "-u" flag to perl and then
47       undump, which means it's going to be false in any sane context.
48
49       Line 4 calls a function in perl.c to allocate memory for a Perl
50       interpreter. It's quite a simple function, and the guts of it looks
51       like this:
52
53        my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter));
54
55       Here you see an example of Perl's system abstraction, which we'll see
56       later: "PerlMem_malloc" is either your system's "malloc", or Perl's own
57       "malloc" as defined in malloc.c if you selected that option at
58       configure time.
59
60       Next, in line 7, we construct the interpreter using perl_construct,
61       also in perl.c; this sets up all the special variables that Perl needs,
62       the stacks, and so on.
63
64       Now we pass Perl the command line options, and tell it to go:
65
66        if (!perl_parse(my_perl, xs_init, argc, argv, (char **)NULL))
67            perl_run(my_perl);
68
69        exitstatus = perl_destruct(my_perl);
70
71        perl_free(my_perl);
72
73       "perl_parse" is actually a wrapper around "S_parse_body", as defined in
74       perl.c, which processes the command line options, sets up any
75       statically linked XS modules, opens the program and calls "yyparse" to
76       parse it.
77
78   Parsing
79       The aim of this stage is to take the Perl source, and turn it into an
80       op tree. We'll see what one of those looks like later. Strictly
81       speaking, there's three things going on here.
82
83       "yyparse", the parser, lives in perly.c, although you're better off
84       reading the original YACC input in perly.y. (Yes, Virginia, there is a
85       YACC grammar for Perl!) The job of the parser is to take your code and
86       "understand" it, splitting it into sentences, deciding which operands
87       go with which operators and so on.
88
89       The parser is nobly assisted by the lexer, which chunks up your input
90       into tokens, and decides what type of thing each token is: a variable
91       name, an operator, a bareword, a subroutine, a core function, and so
92       on. The main point of entry to the lexer is "yylex", and that and its
93       associated routines can be found in toke.c. Perl isn't much like other
94       computer languages; it's highly context sensitive at times, it can be
95       tricky to work out what sort of token something is, or where a token
96       ends. As such, there's a lot of interplay between the tokeniser and the
97       parser, which can get pretty frightening if you're not used to it.
98
99       As the parser understands a Perl program, it builds up a tree of
100       operations for the interpreter to perform during execution. The
101       routines which construct and link together the various operations are
102       to be found in op.c, and will be examined later.
103
104   Optimization
105       Now the parsing stage is complete, and the finished tree represents the
106       operations that the Perl interpreter needs to perform to execute our
107       program. Next, Perl does a dry run over the tree looking for
108       optimisations: constant expressions such as "3 + 4" will be computed
109       now, and the optimizer will also see if any multiple operations can be
110       replaced with a single one. For instance, to fetch the variable $foo,
111       instead of grabbing the glob *foo and looking at the scalar component,
112       the optimizer fiddles the op tree to use a function which directly
113       looks up the scalar in question. The main optimizer is "peep" in op.c,
114       and many ops have their own optimizing functions.
115
116   Running
117       Now we're finally ready to go: we have compiled Perl byte code, and all
118       that's left to do is run it. The actual execution is done by the
119       "runops_standard" function in run.c; more specifically, it's done by
120       these three innocent looking lines:
121
122           while ((PL_op = PL_op->op_ppaddr(aTHX))) {
123               PERL_ASYNC_CHECK();
124           }
125
126       You may be more comfortable with the Perl version of that:
127
128           PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}};
129
130       Well, maybe not. Anyway, each op contains a function pointer, which
131       stipulates the function which will actually carry out the operation.
132       This function will return the next op in the sequence - this allows for
133       things like "if" which choose the next op dynamically at run time. The
134       "PERL_ASYNC_CHECK" makes sure that things like signals interrupt
135       execution if required.
136
137       The actual functions called are known as PP code, and they're spread
138       between four files: pp_hot.c contains the "hot" code, which is most
139       often used and highly optimized, pp_sys.c contains all the system-
140       specific functions, pp_ctl.c contains the functions which implement
141       control structures ("if", "while" and the like) and pp.c contains
142       everything else. These are, if you like, the C code for Perl's built-in
143       functions and operators.
144
145       Note that each "pp_" function is expected to return a pointer to the
146       next op. Calls to perl subs (and eval blocks) are handled within the
147       same runops loop, and do not consume extra space on the C stack. For
148       example, "pp_entersub" and "pp_entertry" just push a "CxSUB" or
149       "CxEVAL" block struct onto the context stack which contain the address
150       of the op following the sub call or eval. They then return the first op
151       of that sub or eval block, and so execution continues of that sub or
152       block. Later, a "pp_leavesub" or "pp_leavetry" op pops the "CxSUB" or
153       "CxEVAL", retrieves the return op from it, and returns it.
154
155   Exception handing
156       Perl's exception handing (i.e. "die" etc.) is built on top of the low-
157       level "setjmp()"/"longjmp()" C-library functions. These basically
158       provide a way to capture the current PC and SP registers and later
159       restore them; i.e. a "longjmp()" continues at the point in code where a
160       previous "setjmp()" was done, with anything further up on the C stack
161       being lost. This is why code should always save values using "SAVE_FOO"
162       rather than in auto variables.
163
164       The perl core wraps "setjmp()" etc in the macros "JMPENV_PUSH" and
165       "JMPENV_JUMP". The basic rule of perl exceptions is that "exit", and
166       "die" (in the absence of "eval") perform a JMPENV_JUMP(2), while "die"
167       within "eval" does a JMPENV_JUMP(3).
168
169       At entry points to perl, such as "perl_parse()", "perl_run()" and
170       "call_sv(cv, G_EVAL)" each does a "JMPENV_PUSH", then enter a runops
171       loop or whatever, and handle possible exception returns. For a 2
172       return, final cleanup is performed, such as popping stacks and calling
173       "CHECK" or "END" blocks. Amongst other things, this is how scope
174       cleanup still occurs during an "exit".
175
176       If a "die" can find a "CxEVAL" block on the context stack, then the
177       stack is popped to that level and the return op in that block is
178       assigned to "PL_restartop"; then a JMPENV_JUMP(3) is performed.  This
179       normally passes control back to the guard. In the case of "perl_run"
180       and "call_sv", a non-null "PL_restartop" triggers re-entry to the
181       runops loop. The is the normal way that "die" or "croak" is handled
182       within an "eval".
183
184       Sometimes ops are executed within an inner runops loop, such as tie,
185       sort or overload code. In this case, something like
186
187           sub FETCH { eval { die } }
188
189       would cause a longjmp right back to the guard in "perl_run", popping
190       both runops loops, which is clearly incorrect. One way to avoid this is
191       for the tie code to do a "JMPENV_PUSH" before executing "FETCH" in the
192       inner runops loop, but for efficiency reasons, perl in fact just sets a
193       flag, using "CATCH_SET(TRUE)". The "pp_require", "pp_entereval" and
194       "pp_entertry" ops check this flag, and if true, they call "docatch",
195       which does a "JMPENV_PUSH" and starts a new runops level to execute the
196       code, rather than doing it on the current loop.
197
198       As a further optimisation, on exit from the eval block in the "FETCH",
199       execution of the code following the block is still carried on in the
200       inner loop. When an exception is raised, "docatch" compares the
201       "JMPENV" level of the "CxEVAL" with "PL_top_env" and if they differ,
202       just re-throws the exception. In this way any inner loops get popped.
203
204       Here's an example.
205
206           1: eval { tie @a, 'A' };
207           2: sub A::TIEARRAY {
208           3:     eval { die };
209           4:     die;
210           5: }
211
212       To run this code, "perl_run" is called, which does a "JMPENV_PUSH" then
213       enters a runops loop. This loop executes the eval and tie ops on line
214       1, with the eval pushing a "CxEVAL" onto the context stack.
215
216       The "pp_tie" does a "CATCH_SET(TRUE)", then starts a second runops loop
217       to execute the body of "TIEARRAY". When it executes the entertry op on
218       line 3, "CATCH_GET" is true, so "pp_entertry" calls "docatch" which
219       does a "JMPENV_PUSH" and starts a third runops loop, which then
220       executes the die op. At this point the C call stack looks like this:
221
222           Perl_pp_die
223           Perl_runops      # third loop
224           S_docatch_body
225           S_docatch
226           Perl_pp_entertry
227           Perl_runops      # second loop
228           S_call_body
229           Perl_call_sv
230           Perl_pp_tie
231           Perl_runops      # first loop
232           S_run_body
233           perl_run
234           main
235
236       and the context and data stacks, as shown by "-Dstv", look like:
237
238           STACK 0: MAIN
239             CX 0: BLOCK  =>
240             CX 1: EVAL   => AV()  PV("A"\0)
241             retop=leave
242           STACK 1: MAGIC
243             CX 0: SUB    =>
244             retop=(null)
245             CX 1: EVAL   => *
246           retop=nextstate
247
248       The die pops the first "CxEVAL" off the context stack, sets
249       "PL_restartop" from it, does a JMPENV_JUMP(3), and control returns to
250       the top "docatch". This then starts another third-level runops level,
251       which executes the nextstate, pushmark and die ops on line 4. At the
252       point that the second "pp_die" is called, the C call stack looks
253       exactly like that above, even though we are no longer within an inner
254       eval; this is because of the optimization mentioned earlier. However,
255       the context stack now looks like this, ie with the top CxEVAL popped:
256
257           STACK 0: MAIN
258             CX 0: BLOCK  =>
259             CX 1: EVAL   => AV()  PV("A"\0)
260             retop=leave
261           STACK 1: MAGIC
262             CX 0: SUB    =>
263             retop=(null)
264
265       The die on line 4 pops the context stack back down to the CxEVAL,
266       leaving it as:
267
268           STACK 0: MAIN
269             CX 0: BLOCK  =>
270
271       As usual, "PL_restartop" is extracted from the "CxEVAL", and a
272       JMPENV_JUMP(3) done, which pops the C stack back to the docatch:
273
274           S_docatch
275           Perl_pp_entertry
276           Perl_runops      # second loop
277           S_call_body
278           Perl_call_sv
279           Perl_pp_tie
280           Perl_runops      # first loop
281           S_run_body
282           perl_run
283           main
284
285       In  this case, because the "JMPENV" level recorded in the "CxEVAL"
286       differs from the current one, "docatch" just does a JMPENV_JUMP(3) and
287       the C stack unwinds to:
288
289           perl_run
290           main
291
292       Because "PL_restartop" is non-null, "run_body" starts a new runops loop
293       and execution continues.
294
295   INTERNAL VARIABLE TYPES
296       You should by now have had a look at perlguts, which tells you about
297       Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do
298       that now.
299
300       These variables are used not only to represent Perl-space variables,
301       but also any constants in the code, as well as some structures
302       completely internal to Perl. The symbol table, for instance, is an
303       ordinary Perl hash. Your code is represented by an SV as it's read into
304       the parser; any program files you call are opened via ordinary Perl
305       filehandles, and so on.
306
307       The core Devel::Peek module lets us examine SVs from a Perl program.
308       Let's see, for instance, how Perl treats the constant "hello".
309
310             % perl -MDevel::Peek -e 'Dump("hello")'
311           1 SV = PV(0xa041450) at 0xa04ecbc
312           2   REFCNT = 1
313           3   FLAGS = (POK,READONLY,pPOK)
314           4   PV = 0xa0484e0 "hello"\0
315           5   CUR = 5
316           6   LEN = 6
317
318       Reading "Devel::Peek" output takes a bit of practise, so let's go
319       through it line by line.
320
321       Line 1 tells us we're looking at an SV which lives at 0xa04ecbc in
322       memory. SVs themselves are very simple structures, but they contain a
323       pointer to a more complex structure. In this case, it's a PV, a
324       structure which holds a string value, at location 0xa041450. Line 2 is
325       the reference count; there are no other references to this data, so
326       it's 1.
327
328       Line 3 are the flags for this SV - it's OK to use it as a PV, it's a
329       read-only SV (because it's a constant) and the data is a PV internally.
330       Next we've got the contents of the string, starting at location
331       0xa0484e0.
332
333       Line 5 gives us the current length of the string - note that this does
334       not include the null terminator. Line 6 is not the length of the
335       string, but the length of the currently allocated buffer; as the string
336       grows, Perl automatically extends the available storage via a routine
337       called "SvGROW".
338
339       You can get at any of these quantities from C very easily; just add
340       "Sv" to the name of the field shown in the snippet, and you've got a
341       macro which will return the value: "SvCUR(sv)" returns the current
342       length of the string, "SvREFCOUNT(sv)" returns the reference count,
343       "SvPV(sv, len)" returns the string itself with its length, and so on.
344       More macros to manipulate these properties can be found in perlguts.
345
346       Let's take an example of manipulating a PV, from "sv_catpvn", in sv.c
347
348            1  void
349            2  Perl_sv_catpvn(pTHX_ SV *sv, const char *ptr, STRLEN len)
350            3  {
351            4      STRLEN tlen;
352            5      char *junk;
353
354            6      junk = SvPV_force(sv, tlen);
355            7      SvGROW(sv, tlen + len + 1);
356            8      if (ptr == junk)
357            9          ptr = SvPVX(sv);
358           10      Move(ptr,SvPVX(sv)+tlen,len,char);
359           11      SvCUR(sv) += len;
360           12      *SvEND(sv) = '\0';
361           13      (void)SvPOK_only_UTF8(sv);          /* validate pointer */
362           14      SvTAINT(sv);
363           15  }
364
365       This is a function which adds a string, "ptr", of length "len" onto the
366       end of the PV stored in "sv". The first thing we do in line 6 is make
367       sure that the SV has a valid PV, by calling the "SvPV_force" macro to
368       force a PV. As a side effect, "tlen" gets set to the current value of
369       the PV, and the PV itself is returned to "junk".
370
371       In line 7, we make sure that the SV will have enough room to
372       accommodate the old string, the new string and the null terminator. If
373       "LEN" isn't big enough, "SvGROW" will reallocate space for us.
374
375       Now, if "junk" is the same as the string we're trying to add, we can
376       grab the string directly from the SV; "SvPVX" is the address of the PV
377       in the SV.
378
379       Line 10 does the actual catenation: the "Move" macro moves a chunk of
380       memory around: we move the string "ptr" to the end of the PV - that's
381       the start of the PV plus its current length. We're moving "len" bytes
382       of type "char". After doing so, we need to tell Perl we've extended the
383       string, by altering "CUR" to reflect the new length. "SvEND" is a macro
384       which gives us the end of the string, so that needs to be a "\0".
385
386       Line 13 manipulates the flags; since we've changed the PV, any IV or NV
387       values will no longer be valid: if we have "$a=10; $a.="6";" we don't
388       want to use the old IV of 10. "SvPOK_only_utf8" is a special
389       UTF-8-aware version of "SvPOK_only", a macro which turns off the IOK
390       and NOK flags and turns on POK. The final "SvTAINT" is a macro which
391       launders tainted data if taint mode is turned on.
392
393       AVs and HVs are more complicated, but SVs are by far the most common
394       variable type being thrown around. Having seen something of how we
395       manipulate these, let's go on and look at how the op tree is
396       constructed.
397

OP TREES

399       First, what is the op tree, anyway? The op tree is the parsed
400       representation of your program, as we saw in our section on parsing,
401       and it's the sequence of operations that Perl goes through to execute
402       your program, as we saw in "Running".
403
404       An op is a fundamental operation that Perl can perform: all the built-
405       in functions and operators are ops, and there are a series of ops which
406       deal with concepts the interpreter needs internally - entering and
407       leaving a block, ending a statement, fetching a variable, and so on.
408
409       The op tree is connected in two ways: you can imagine that there are
410       two "routes" through it, two orders in which you can traverse the tree.
411       First, parse order reflects how the parser understood the code, and
412       secondly, execution order tells perl what order to perform the
413       operations in.
414
415       The easiest way to examine the op tree is to stop Perl after it has
416       finished parsing, and get it to dump out the tree. This is exactly what
417       the compiler backends B::Terse, B::Concise and CPAN module <B::Debug
418       do.
419
420       Let's have a look at how Perl sees "$a = $b + $c":
421
422            % perl -MO=Terse -e '$a=$b+$c'
423            1  LISTOP (0x8179888) leave
424            2      OP (0x81798b0) enter
425            3      COP (0x8179850) nextstate
426            4      BINOP (0x8179828) sassign
427            5          BINOP (0x8179800) add [1]
428            6              UNOP (0x81796e0) null [15]
429            7                  SVOP (0x80fafe0) gvsv  GV (0x80fa4cc) *b
430            8              UNOP (0x81797e0) null [15]
431            9                  SVOP (0x8179700) gvsv  GV (0x80efeb0) *c
432           10          UNOP (0x816b4f0) null [15]
433           11              SVOP (0x816dcf0) gvsv  GV (0x80fa460) *a
434
435       Let's start in the middle, at line 4. This is a BINOP, a binary
436       operator, which is at location 0x8179828. The specific operator in
437       question is "sassign" - scalar assignment - and you can find the code
438       which implements it in the function "pp_sassign" in pp_hot.c. As a
439       binary operator, it has two children: the add operator, providing the
440       result of "$b+$c", is uppermost on line 5, and the left hand side is on
441       line 10.
442
443       Line 10 is the null op: this does exactly nothing. What is that doing
444       there? If you see the null op, it's a sign that something has been
445       optimized away after parsing. As we mentioned in "Optimization", the
446       optimization stage sometimes converts two operations into one, for
447       example when fetching a scalar variable. When this happens, instead of
448       rewriting the op tree and cleaning up the dangling pointers, it's
449       easier just to replace the redundant operation with the null op.
450       Originally, the tree would have looked like this:
451
452           10          SVOP (0x816b4f0) rv2sv [15]
453           11              SVOP (0x816dcf0) gv  GV (0x80fa460) *a
454
455       That is, fetch the "a" entry from the main symbol table, and then look
456       at the scalar component of it: "gvsv" ("pp_gvsv" in pp_hot.c) happens
457       to do both these things.
458
459       The right hand side, starting at line 5 is similar to what we've just
460       seen: we have the "add" op ("pp_add", also in pp_hot.c) add together
461       two "gvsv"s.
462
463       Now, what's this about?
464
465            1  LISTOP (0x8179888) leave
466            2      OP (0x81798b0) enter
467            3      COP (0x8179850) nextstate
468
469       "enter" and "leave" are scoping ops, and their job is to perform any
470       housekeeping every time you enter and leave a block: lexical variables
471       are tidied up, unreferenced variables are destroyed, and so on. Every
472       program will have those first three lines: "leave" is a list, and its
473       children are all the statements in the block. Statements are delimited
474       by "nextstate", so a block is a collection of "nextstate" ops, with the
475       ops to be performed for each statement being the children of
476       "nextstate". "enter" is a single op which functions as a marker.
477
478       That's how Perl parsed the program, from top to bottom:
479
480                               Program
481                                  |
482                              Statement
483                                  |
484                                  =
485                                 / \
486                                /   \
487                               $a   +
488                                   / \
489                                 $b   $c
490
491       However, it's impossible to perform the operations in this order: you
492       have to find the values of $b and $c before you add them together, for
493       instance. So, the other thread that runs through the op tree is the
494       execution order: each op has a field "op_next" which points to the next
495       op to be run, so following these pointers tells us how perl executes
496       the code. We can traverse the tree in this order using the "exec"
497       option to "B::Terse":
498
499            % perl -MO=Terse,exec -e '$a=$b+$c'
500            1  OP (0x8179928) enter
501            2  COP (0x81798c8) nextstate
502            3  SVOP (0x81796c8) gvsv  GV (0x80fa4d4) *b
503            4  SVOP (0x8179798) gvsv  GV (0x80efeb0) *c
504            5  BINOP (0x8179878) add [1]
505            6  SVOP (0x816dd38) gvsv  GV (0x80fa468) *a
506            7  BINOP (0x81798a0) sassign
507            8  LISTOP (0x8179900) leave
508
509       This probably makes more sense for a human: enter a block, start a
510       statement. Get the values of $b and $c, and add them together.  Find
511       $a, and assign one to the other. Then leave.
512
513       The way Perl builds up these op trees in the parsing process can be
514       unravelled by examining toke.c, the lexer, and perly.y, the YACC
515       grammar. Let's look at the code that constructs the tree for "$a = $b +
516       $c".
517
518       First, we'll look at the "Perl_yylex" function in the lexer. We want to
519       look for "case 'x'", where x is the first character of the operator.
520       (Incidentally, when looking for the code that handles a keyword, you'll
521       want to search for "KEY_foo" where "foo" is the keyword.) Here is the
522       code that handles assignment (there are quite a few operators beginning
523       with "=", so most of it is omitted for brevity):
524
525            1    case '=':
526            2        s++;
527                     ... code that handles == => etc. and pod ...
528            3        pl_yylval.ival = 0;
529            4        OPERATOR(ASSIGNOP);
530
531       We can see on line 4 that our token type is "ASSIGNOP" ("OPERATOR" is a
532       macro, defined in toke.c, that returns the token type, among other
533       things). And "+":
534
535            1     case '+':
536            2         {
537            3             const char tmp = *s++;
538                          ... code for ++ ...
539            4             if (PL_expect == XOPERATOR) {
540                              ...
541            5                 Aop(OP_ADD);
542            6             }
543                          ...
544            7         }
545
546       Line 4 checks what type of token we are expecting. "Aop" returns a
547       token.  If you search for "Aop" elsewhere in toke.c, you will see that
548       it returns an "ADDOP" token.
549
550       Now that we know the two token types we want to look for in the parser,
551       let's take the piece of perly.y we need to construct the tree for "$a =
552       $b + $c"
553
554           1 term    :   term ASSIGNOP term
555           2                { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); }
556           3         |   term ADDOP term
557           4                { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
558
559       If you're not used to reading BNF grammars, this is how it works:
560       You're fed certain things by the tokeniser, which generally end up in
561       upper case. "ADDOP" and "ASSIGNOP" are examples of "terminal symbols",
562       because you can't get any simpler than them.
563
564       The grammar, lines one and three of the snippet above, tells you how to
565       build up more complex forms. These complex forms, "non-terminal
566       symbols" are generally placed in lower case. "term" here is a non-
567       terminal symbol, representing a single expression.
568
569       The grammar gives you the following rule: you can make the thing on the
570       left of the colon if you see all the things on the right in sequence.
571       This is called a "reduction", and the aim of parsing is to completely
572       reduce the input. There are several different ways you can perform a
573       reduction, separated by vertical bars: so, "term" followed by "="
574       followed by "term" makes a "term", and "term" followed by "+" followed
575       by "term" can also make a "term".
576
577       So, if you see two terms with an "=" or "+", between them, you can turn
578       them into a single expression. When you do this, you execute the code
579       in the block on the next line: if you see "=", you'll do the code in
580       line 2. If you see "+", you'll do the code in line 4. It's this code
581       which contributes to the op tree.
582
583                   |   term ADDOP term
584                   { $$ = newBINOP($2, 0, scalar($1), scalar($3)); }
585
586       What this does is creates a new binary op, and feeds it a number of
587       variables. The variables refer to the tokens: $1 is the first token in
588       the input, $2 the second, and so on - think regular expression
589       backreferences. $$ is the op returned from this reduction. So, we call
590       "newBINOP" to create a new binary operator. The first parameter to
591       "newBINOP", a function in op.c, is the op type. It's an addition
592       operator, so we want the type to be "ADDOP". We could specify this
593       directly, but it's right there as the second token in the input, so we
594       use $2. The second parameter is the op's flags: 0 means "nothing
595       special". Then the things to add: the left and right hand side of our
596       expression, in scalar context.
597
598       The functions that create ops, which have names like "newUNOP" and
599       "newBINOP", call a "check" function associated with each op type,
600       before returning the op. The check functions can mangle the op as they
601       see fit, and even replace it with an entirely new one. These functions
602       are defined in op.c, and have a "Perl_ck_" prefix. You can find out
603       which check function is used for a particular op type by looking in
604       regen/opcodes.  Take "OP_ADD", for example. ("OP_ADD" is the token
605       value from the "Aop(OP_ADD)" in toke.c which the parser passes to
606       "newBINOP" as its first argument.) Here is the relevant line:
607
608           add             addition (+)            ck_null         IfsT2   S S
609
610       The check function in this case is "Perl_ck_null", which does nothing.
611       Let's look at a more interesting case:
612
613           readline        <HANDLE>                ck_readline     t%      F?
614
615       And here is the function from op.c:
616
617            1 OP *
618            2 Perl_ck_readline(pTHX_ OP *o)
619            3 {
620            4     PERL_ARGS_ASSERT_CK_READLINE;
621            5
622            6     if (o->op_flags & OPf_KIDS) {
623            7          OP *kid = cLISTOPo->op_first;
624            8          if (kid->op_type == OP_RV2GV)
625            9              kid->op_private |= OPpALLOW_FAKE;
626           10     }
627           11     else {
628           12         OP * const newop
629           13             = newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0,
630           14                                               PL_argvgv));
631           15         op_free(o);
632           16         return newop;
633           17     }
634           18     return o;
635           19 }
636
637       One particularly interesting aspect is that if the op has no kids
638       (i.e., "readline()" or "<>") the op is freed and replaced with an
639       entirely new one that references *ARGV (lines 12-16).
640

STACKS

642       When perl executes something like "addop", how does it pass on its
643       results to the next op? The answer is, through the use of stacks. Perl
644       has a number of stacks to store things it's currently working on, and
645       we'll look at the three most important ones here.
646
647   Argument stack
648       Arguments are passed to PP code and returned from PP code using the
649       argument stack, "ST". The typical way to handle arguments is to pop
650       them off the stack, deal with them how you wish, and then push the
651       result back onto the stack. This is how, for instance, the cosine
652       operator works:
653
654             NV value;
655             value = POPn;
656             value = Perl_cos(value);
657             XPUSHn(value);
658
659       We'll see a more tricky example of this when we consider Perl's macros
660       below. "POPn" gives you the NV (floating point value) of the top SV on
661       the stack: the $x in "cos($x)". Then we compute the cosine, and push
662       the result back as an NV. The "X" in "XPUSHn" means that the stack
663       should be extended if necessary - it can't be necessary here, because
664       we know there's room for one more item on the stack, since we've just
665       removed one! The "XPUSH*" macros at least guarantee safety.
666
667       Alternatively, you can fiddle with the stack directly: "SP" gives you
668       the first element in your portion of the stack, and "TOP*" gives you
669       the top SV/IV/NV/etc. on the stack. So, for instance, to do unary
670       negation of an integer:
671
672            SETi(-TOPi);
673
674       Just set the integer value of the top stack entry to its negation.
675
676       Argument stack manipulation in the core is exactly the same as it is in
677       XSUBs - see perlxstut, perlxs and perlguts for a longer description of
678       the macros used in stack manipulation.
679
680   Mark stack
681       I say "your portion of the stack" above because PP code doesn't
682       necessarily get the whole stack to itself: if your function calls
683       another function, you'll only want to expose the arguments aimed for
684       the called function, and not (necessarily) let it get at your own data.
685       The way we do this is to have a "virtual" bottom-of-stack, exposed to
686       each function. The mark stack keeps bookmarks to locations in the
687       argument stack usable by each function. For instance, when dealing with
688       a tied variable, (internally, something with "P" magic) Perl has to
689       call methods for accesses to the tied variables. However, we need to
690       separate the arguments exposed to the method to the argument exposed to
691       the original function - the store or fetch or whatever it may be.
692       Here's roughly how the tied "push" is implemented; see "av_push" in
693       av.c:
694
695            1  PUSHMARK(SP);
696            2  EXTEND(SP,2);
697            3  PUSHs(SvTIED_obj((SV*)av, mg));
698            4  PUSHs(val);
699            5  PUTBACK;
700            6  ENTER;
701            7  call_method("PUSH", G_SCALAR|G_DISCARD);
702            8  LEAVE;
703
704       Let's examine the whole implementation, for practice:
705
706            1  PUSHMARK(SP);
707
708       Push the current state of the stack pointer onto the mark stack. This
709       is so that when we've finished adding items to the argument stack, Perl
710       knows how many things we've added recently.
711
712            2  EXTEND(SP,2);
713            3  PUSHs(SvTIED_obj((SV*)av, mg));
714            4  PUSHs(val);
715
716       We're going to add two more items onto the argument stack: when you
717       have a tied array, the "PUSH" subroutine receives the object and the
718       value to be pushed, and that's exactly what we have here - the tied
719       object, retrieved with "SvTIED_obj", and the value, the SV "val".
720
721            5  PUTBACK;
722
723       Next we tell Perl to update the global stack pointer from our internal
724       variable: "dSP" only gave us a local copy, not a reference to the
725       global.
726
727            6  ENTER;
728            7  call_method("PUSH", G_SCALAR|G_DISCARD);
729            8  LEAVE;
730
731       "ENTER" and "LEAVE" localise a block of code - they make sure that all
732       variables are tidied up, everything that has been localised gets its
733       previous value returned, and so on. Think of them as the "{" and "}" of
734       a Perl block.
735
736       To actually do the magic method call, we have to call a subroutine in
737       Perl space: "call_method" takes care of that, and it's described in
738       perlcall. We call the "PUSH" method in scalar context, and we're going
739       to discard its return value. The call_method() function removes the top
740       element of the mark stack, so there is nothing for the caller to clean
741       up.
742
743   Save stack
744       C doesn't have a concept of local scope, so perl provides one. We've
745       seen that "ENTER" and "LEAVE" are used as scoping braces; the save
746       stack implements the C equivalent of, for example:
747
748           {
749               local $foo = 42;
750               ...
751           }
752
753       See "Localizing changes" in perlguts for how to use the save stack.
754

MILLIONS OF MACROS

756       One thing you'll notice about the Perl source is that it's full of
757       macros. Some have called the pervasive use of macros the hardest thing
758       to understand, others find it adds to clarity. Let's take an example, a
759       stripped-down version the code which implements the addition operator:
760
761          1  PP(pp_add)
762          2  {
763          3      dSP; dATARGET;
764          4      tryAMAGICbin_MG(add_amg, AMGf_assign|AMGf_numeric);
765          5      {
766          6        dPOPTOPnnrl_ul;
767          7        SETn( left + right );
768          8        RETURN;
769          9      }
770         10  }
771
772       Every line here (apart from the braces, of course) contains a macro.
773       The first line sets up the function declaration as Perl expects for PP
774       code; line 3 sets up variable declarations for the argument stack and
775       the target, the return value of the operation. Line 4 tries to see if
776       the addition operation is overloaded; if so, the appropriate subroutine
777       is called.
778
779       Line 6 is another variable declaration - all variable declarations
780       start with "d" - which pops from the top of the argument stack two NVs
781       (hence "nn") and puts them into the variables "right" and "left", hence
782       the "rl". These are the two operands to the addition operator.  Next,
783       we call "SETn" to set the NV of the return value to the result of
784       adding the two values. This done, we return - the "RETURN" macro makes
785       sure that our return value is properly handled, and we pass the next
786       operator to run back to the main run loop.
787
788       Most of these macros are explained in perlapi, and some of the more
789       important ones are explained in perlxs as well. Pay special attention
790       to "Background and PERL_IMPLICIT_CONTEXT" in perlguts for information
791       on the "[pad]THX_?" macros.
792

FURTHER READING

794       For more information on the Perl internals, please see the documents
795       listed at "Internals and C Language Interface" in perl.
796
797
798
799perl v5.30.2                      2020-03-27                     PERLINTERP(1)
Impressum