On Wed, 5 Sep 2001, Stas Bekman wrote:
> I was tryint to trap SEGV from the child process that segfaults, but I
> cannot seem to catch it. I've added this to t/conf/modperl_startup.pl (of
> course via TestConfigPerl.pm, since the former is autogenerated):
>
> use Carp;
> $SIG{SEGV} = sub { Carp::confess("caught SIGSEGV") };
> $SIG{ABRT} = sub { Carp::confess("caught SIGABRT") };
>
> But the child segfaults with this message in t/logs/error_log:
>
> [Wed Sep 05 12:20:07 2001] [notice] child pid 21228 exit signal
> Segmentation fault (11), possible coredump in
> /home/stas/apache.org/modperl-2.0/t
>
> It seems like Apache handles the SEGV and doesn't propogate it to Perl. I
> was searching on the google to learn about this issue, but couldn't find
> anything useful. Your help is appreciated.
well, if apache core dumps before it reaches mod_perl land, that could be
one reason. and if it is Perl that segvs, it is likely the perl runtime
is in a corrupt state where calling back into Perl simply isn't going to
work.
> BTW, Doug, Sys::Signal doesn't compile with 5.7.1. (I've tried to use this
> module also to see if it can help me to catch SEGV).
i don't think it would help here.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
