Secondary Memory corruption during destruction?

Still on the trail of warnings during exit from certain Tk apps
I turned on PERL_DESTRUCT_LEVEL 3.

Now some of them core-dump instead ;-)

Code that causes that is (perl5.005_03) sv.c line 2891:

    if (SvTYPE(sv) >= SVt_PVMG && SvMAGIC(sv))
	mg_free(sv);

Now the sv in question has SvTYPE(sv) == SVTYPEMASK - which is 
certainly bigger than SVt_PVMG but does not mean that SvANY points
to something which will lead to a sane chain of magics to free!

The root of the problem is probably still elsewhere, 
but should sv_clear() just return if SvTYPE(sv) == SVTYPEMASK ?


-- 
Nick Ing-Simmons <nik@tiuk.ti.com>
Via, but not speaking for: Texas Instruments Ltd.

• Secondary Memory corruption during destruction? by Nick Ing-Simmons
• Re: Secondary Memory corruption during destruction? by Gurusamy Sarathy
• Re: Secondary Memory corruption during destruction? by Nick Ing-Simmons