Bug #49878 for forks: make test fails with assertion

Mon Sep 21 11:29:28 2009 bjohnson [...] symetrix.com - Ticket created

Subject:

make test fails with assertion

$ perl -v This is perl, v5.10.0 built for i386-linux-thread-multi Copyright 1987-2007, Larry Wall Perl may be copied only under the terms of either the Artistic License or the GNU General Public License, which may be found in the Perl 5 source kit. Complete documentation for Perl, including FAQ lists, should be found on this system using "man perl" or "perldoc perl". If you have access to the Internet, point your browser at http://www.perl.org/, the Perl Home Page. + make test PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t t/forks00-sigtrap.t .. ok Assertion ((sv)->sv_flags & (0x00200000|0x00400000|0x00800000)) failed: file "mg.c", line 88 at /usr/lib/perl5/vendor_perl/5.10.0/Sys/SigAction.pm line 132. Compilation failed in require at t/forks01.t line 14. BEGIN failed--compilation aborted at t/forks01.t line 14. ^C

Thu Oct 15 16:56:41 2009 mibarra [...] msg.com.mx - Cc mibarra added

Mon Jun 14 01:10:42 2010 RYBSKEJ [...] cpan.org - Status changed from 'new' to 'open'

Mon Jun 14 01:15:03 2010 RYBSKEJ [...] cpan.org - Correspondence added

This appears to be a Perl core issue, documented in RT#75150: http://rt.perl.org/rt3/Public/Bug/Display.html?id=75150

Tue Nov 15 04:36:41 2011 http://was.id.ly/ - Correspondence added

A possible workaround is this patch from Tuukka. The comment is not very reassuring though: "This little stupid patch fixes things but I don't know why!" :) Also, seems like hes wrong when saying it's a forks problem. https://bugzilla.novell.com/show_bug.cgi?id=527537#c3

Tue Nov 15 16:30:23 2011 RYBSKEJ [...] cpan.org - Correspondence added

On Tue Nov 15 04:36:41 2011, http://was.id.ly/ wrote: Show quoted text

> > A possible workaround is this patch from Tuukka. The comment is not very > reassuring though: "This little stupid patch fixes things but I > don't know why!" :) > > Also, seems like hes wrong when saying it's a forks problem. > > https://bugzilla.novell.com/show_bug.cgi?id=527537#c3 >

The bug is unfortunately in Perl, as noted earlier in this thread. This is fixed in Perl 5.14. The specific Perl patch is here: http://perl5.git.perl.org/perl.git/commitdiff/17cffb3718693d45499c002f0f50df9196858f28 ?hp=8df6b97c1de8326d50ac9c8cae4bf716393b45bb The core issue is sigaction would crash/assert with a replaced %SIG, e.g.: *SIG = {}; POSIX::sigaction(SIGHUP, POSIX::SigAction->new); # <-- crash without patch So commenting out that block in forks::signals only prevents forks from using sigaction. But any user code can still call sigaction, leading to a crash. So perhaps I need a Perl 5.10-5.12 "mitigation" patch: install forks.pm CHLD early (the only signal forks.pm requres with flags via sigaction); then warn users not to use POSIX::sigaction in their code, which is quite uncommon anyway, especially in ithread-aware code where you shouldn't be fork'ing anymore. Or vendors can apply the above mentioned bleed patch to their Perl 5.10 or 5.12 distribution.

Tue Nov 15 16:30:34 2011 RYBSKEJ [...] cpan.org - Taken