Document OP_NEXT's use of OPf_SPECIAL in op.h

[wolfsage]

See cd97dc8 for its meaning

All ops that use the OPf_SPECIAL flag should be documented in op.h where OPf_SPECIAL is defined. This one is missing and should be pretty easy to add.

[leonerd]

Hrm; your link shows the code where the flag is tested by pp_next. I've been hunting around for the code which sets it but so far I've come up blank. OP_NEXT doesn't have an opchecker, nor does peep.c have much to do with an OP_NEXT at all.

[leonerd]

Ah, I see it. It's directly in the parser as a direct consequence of next vs next LABEL:

	|	LOOPEX  /* loop exiting command (goto, last, dump, etc) */
			{ $$ = newOP($LOOPEX, OPf_SPECIAL);
			    PL_hints |= HINT_BLOCK_SCOPE; }
	|	LOOPEX term[operand]
			{ $$ = newLOOPEX($LOOPEX,$operand); }

[iabyn]

On Fri, Dec 16, 2022 at 02:39:27AM -0800, Paul Evans wrote: Hrm; your link shows the code where the flag is tested by `pp_next`. I've been hunting around for the code which sets it but so far I've come up blank. `OP_NEXT` doesn't have an opchecker, nor does `peep.c` have much to do with an `OP_NEXT` at all.

perly.y: for the LOOPEX without arg case, it calls newOP($LOOPEX, OPf_SPECIAL) There's a relatively easy way to answer 'where does this flag get set' in gdb. Break at a point where you know the flag has been set, set a watchpoint on the flag's storage location, then rerun and continue until the flag changes: $ PERL_HASH_SEED=0 gdb ./perl (gdb) b Perl_pp_next (gdb) run -e'for (1) { next; }' Breakpoint 1, Perl_pp_next (my_perl=0xa8f2a0) at pp_ctl.c:2715 2715 cx = CX_CUR(); (gdb) p my_perl->Iop->op_flags $1 = 0x81 (gdb) wa -l my_perl->Iop->op_flags (gdb) r Hardware watchpoint 2: -location my_perl->Iop->op_flags Old value = <unreadable> New value = 0x0 0x00007ffff7ccb25b in __memset_evex_unaligned_erms () from /lib64/libc.so.6 (gdb) c Hardware watchpoint 2: -location my_perl->Iop->op_flags Old value = 0x0 New value = 0x80 Perl_newOP (my_perl=0xa8f2a0, type=0xd2, flags=0x80) at op.c:5418 5418 o->op_next = o; (gdb) up #1 0x00000000004f5d9f in Perl_yyparse (my_perl=0xa8f2a0, gramtype=0x102) at /home/davem/perl5/git/bleed/perly.y:1356 1356 { $$ = newOP($LOOPEX, OPf_SPECIAL);

…

-- A major Starfleet emergency breaks out near the Enterprise, but fortunately some other ships in the area are able to deal with it to everyone's satisfaction. -- Things That Never Happen in "Star Trek" #13

[leonerd]

Ah interesting. I've never had much luck with rerunning programs in gdb like that because often it gets different memory addresses. But that hint of PERL_HASH_SEED=0 might do the trick.

[leonerd]

I believe #20614 fixes this now. @wolfsage Can this be closed?

[wolfsage]

Yes thank you.

I've never had much luck with rerunning programs in gdb like that because often it gets different memory addresses.

I think that's what https://rr-project.org/ aims to solve?