clone 537996 -1
reassign -1 perl 5.10.0-19
retitle -1 segfault calling SvIV() after perl_run()
tag 537996 patch
thanks
On Sun, Jul 26, 2009 at 05:01:39PM +0200, Tomaž Šolc wrote:
> >> Somebody with deeper knowledge of Perl guts will have to tell if it just
> >> fixes the symptoms or the cause of the bug.
> With some trial-and-error I came up with the attached script - that's
> the shortest script I can get to crash.
I'm attaching a core-only version of the test script and a short C program
that embeds Perl and crashes in the same way on i386 when running the
test script with '-w'.
I'll apply your SvTRUE workaround for speedy-cgi-perl, but I'm cloning
a separate bug against perl. If the C program is doing something wrong,
I suppose perlembed.pod should have a note about it.
FWIW, I see the crashing code in the Perl core was briefly changed not
to crash as a workaround for another bug.
Resetting PL_op works here too, but should it be needed?
http://perl5.git.perl.org/perl.git/commitdiff/d4f41c0196a61a9d6f38ee42fc192ebf28ef6420
http://perl5.git.perl.org/perl.git/commitdiff/cecbe0108c1c5a1c170a53b486232d7a592e3312
http://perl5.git.perl.org/perl.git/commitdiff/30c8d9e411f7375ddf81a32aa625d6c0e0c39091
Cheers,
--
Niko Tyni [email protected]
my $q;
sub f {}
sub g {
$q = f();
exit 0;
}
g();
#include <EXTERN.h> /* from the Perl distribution */
#include <perl.h> /* from the Perl distribution */
static PerlInterpreter *my_perl; /*** The Perl interpreter ***/
int main(int argc, char **argv, char **env)
{
SV *t;
PERL_SYS_INIT3(&argc,&argv,&env);
my_perl = perl_alloc();
perl_construct(my_perl);
perl_parse(my_perl, NULL, argc, argv, (char **)NULL);
t = get_sv("v", 1); /* create the variable */
perl_run(my_perl);
printf("value: %d\n", SvIV(t));
perl_destruct(my_perl);
perl_free(my_perl);
PERL_SYS_TERM();
}
