Front page | perl.perl5.porters |
Postings from October 2003
Nicholas Clark <nick@ccl4.org> writes:
> On Thu, Oct 09, 2003 at 12:32:53AM +0200, Slaven Rezic wrote:
> > Tatsuhiko Miyagawa <miyagawa@edge.co.jp> writes:
>
> > > Program received signal SIGSEGV, Segmentation fault.
> > > 0x281f1a95 in readdir_r () from /usr/lib/libc_r.so.4
> > > (gdb) bt
> > > #0 0x281f1a95 in readdir_r () from /usr/lib/libc_r.so.4
> > > #1 0x80e2139 in Perl_pp_readdir ()
> > > #2 0x80ab1a1 in Perl_runops_standard ()
> > > #3 0x8061d09 in S_run_body ()
> > > #4 0x80619d8 in perl_run ()
> > > #5 0x805e816 in main ()
> > > #6 0x805e6f5 in _start ()
> > > (gdb)
> > >
> > >
> > > should I rebuild perl to show full backtrace?
>
> I'm not sure if you need to. Given that the SEGV is in the libc_r code,
> I don't think that it's perl's fault, and I can't see how we can avoid
> tickling the bug.
>
The issue is mentioned in README.freebsd:
=head2 FreeBSD core dumps from readdir_r with ithreads
When perl is configured to use ithreads, it will use re-entrant library calls
in preference to non-re-entrant versions. There is a bug in FreeBSD's
C<readdir_r> function in versions 4.5 and earlier that can cause a SEGV when
reading large directories. A patch for FreeBSD libc is available
(see http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/30631 )
which has been integrated into FreeBSD 4.6.
Regards,
Slaven
--
Slaven Rezic - slaven@rezic.de
tknotes - A knotes clone, written in Perl/Tk.
http://ptktools.sourceforge.net/#tknotes
Thread Previous
|
Thread Next