Front page | perl.perl5.porters |
Postings from October 2013
On Sat Sep 28 15:39:15 2013, plokinom@gmail.com wrote: > On 28.09.2013 19:05, Father Chrysostomos via RT wrote: > > > > More random guessing: > > > > Is constant folding causing the stack to be reallocated when coresubs > > are vivified? Does PUSHSTACKi/POPSTACK in gv.c:S_maybe_add_coresub make > > the problem go away? > > I wasn't sure what you meant by that so I did this: > > diff --git a/gv.c b/gv.c > index 9f0b57e..baf2ea7 100644 > --- a/gv.c > +++ b/gv.c > @@ -536,6 +536,8 @@ S_maybe_add_coresub(pTHX_ HV * const stash, GV *gv, > #ifdef DEBUGGING > CV *orig_cv = cv; > #endif > + dSP; > + PUSHSTACK; > CvLVALUE_on(cv); > /* newATTRSUB will free the CV and return NULL if we're still > compiling after a syntax error */ > @@ -555,6 +557,7 @@ S_maybe_add_coresub(pTHX_ HV * const stash, GV *gv, > && opnum != OP_UNDEF) > CvLVALUE_off(cv); /* Now *that* was a neat trick. */ > } > + POPSTACK; > LEAVE; > PL_parser = oldparser; > PL_curcop = oldcurcop; > > > Not only does this compile, it also seems to fix the problem: No more > panics or segfaults. > In order to move the discussion forward, I created a branch, manually made the changes indicated, tested it (PASS) and pushed it to the following smoke branch: smoke-me/jkeenan/119993-gv-c Since the area being patched is outside of my expertise, I myself take no position on whether the patch should be merged to blead. Thank you very much. Jim Keenan --- via perlbug: queue: perl5 status: open https://rt.perl.org:443/rt3/Ticket/Display.html?id=119993Thread Previous | Thread Next