[SOLVED]Seems to crash … perl, what?

[bluefoxicy]

Like it

Up

0

Down

Drop it

::

OK I think I figured this out:

https://bugzilla.redhat.com/show_bug.cgi?id=839788

Bug in Perl related to threading.

[tpalacios]

Like it

Up

0

Down

Drop it

::

Thanks for the bug report.

>It looks like it should be included as an agent requirement to have Thread module version above 1.72, in order to have it working properly.

Much appreciated.

Regards.

[bluefoxicy]

Like it

Up

0

Down

Drop it

::

Thanks for the bug report.

>It looks like it should be included as an agent requirement to have Thread module version above 1.72, in order to have it working properly.

Much appreciated.

Regards.

I had a perl core dev test the one-liner test case, he says it’s not in Perl Threads and it’s fine on perl 5.6.2-5.10.0 and 5.12.0+.  perl 5.10.1 and perl 5.11.0 both segfault on the bug.  His non-threaded perl release builds also segfault on the same versions.

So likely not actually in perl threads, just a perl bug.

[bluefoxicy]

Like it

Up

0

Down

Drop it

::

I got it!

http://www.nntp.perl.org/group/perl.perl5.porters/2010/01/msg155286.html

Also the work-around is to disable threading (set agent threads to 1).  Odd, since this isn’t a threading bug; however it’s possible that threading causes a destructor call on a non-defined destructor.

The check for this is to fork off a new copy of perl (don’t try to exception catch this; it segfaults libperl) and see if it can exit safely.  If this throws an error (exit status 137 rather than exit status 0), then perl is broken:

perl -e "sub M::DESTROY; bless {}, M;"

In which case you should disable threading.

system("perl", "-e", "sub M::DESTROY; bless {}, M;") == 0 or agent_threads = 1;

[Nacho1988]

Like it

Up

0

Down

Drop it

::

Nice job! 🙂 I will edit the post as solved.

Regards