Bug #39592 for Data-Serializer: Attempt to free .... during global destruction.

Wed Sep 24 14:11:47 2008 jmartin [...] gsihosting.com - Ticket created

Subject:	Attempt to free .... during global destruction.
Date:	Wed, 24 Sep 2008 13:11:04 -0500
To:	<bug-Data-Serializer [...] rt.cpan.org>
From:	"Martin, Jeremy" <jmartin [...] gsihosting.com>

Starting with Data::Serializer 0.48 an existing script now scrolls error messages like these for a very long time. Not sure if it will ever stop, I aborted the script after over 3 million lines were printed out. Reverting to Data::Serializer 0.46 seems to fix this. (Didn't try 0.47 since 0.48 came out only a day later) Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '(Zú*' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '(Zú*' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'Ð' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'Ð' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'àXC' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'àXC' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '¢' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '¢' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '°}¨' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string '°}¨' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free unreferenced scalar at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'ý' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction. Attempt to free non-existent shared string 'ý' at /usr/lib/perl5/site_perl/5.8.0/Data/Serializer.pm line 198 during global destruction.

Wed Dec 29 17:50:06 2010 neil [...] neely.cx - Taken

Wed Dec 29 17:52:38 2010 neil [...] neely.cx - Correspondence added 10 min

The relevant code that references is the following: sub DESTROY { my $self = (shift); undef %_internal; } Which essentially is just ensuring that an internal data structure is cleaned out between invocations to ensure we don't slowly leak memory in very long memory resident runs. I cannot reproduce the problem you are reporting, can you provide me any means of reproducing this problem?

Wed Dec 29 17:52:39 2010 The RT System itself - Status changed from 'new' to 'open'

Wed Dec 29 17:52:39 2010 neil [...] neely.cx - Status changed from 'open' to 'stalled'

Wed Dec 29 18:06:12 2010 jmartin [...] gsihosting.com - Correspondence added

Subject:	RE: [rt.cpan.org #39592] Attempt to free .... during global destruction.
Date:	Wed, 29 Dec 2010 17:06:01 -0600
To:	"bug-Data-Serializer [...] rt.cpan.org" <bug-Data-Serializer [...] rt.cpan.org>
From:	"Martin, Jeremy" <jmartin [...] gsihosting.com>

Well, that was from 2 years ago, and that OS (RedHat Enterprise Linux 3 aka CentOS 3) is now End Of Life'd. :-) I have been sticking with v0.46 on RHEL 4 this whole time, but I just let it install v0.49 via CPAN and I was unable to reproduce this now. RHEL 5 now include a perl-Data-Serializer RPM (v0.49) which has also not caused us any issues. I don't have the exact code that was making RHEL 3 wig out back then, but I'm pretty sure it was something very close to this which is about all I do with Data::Serializer currently. It would have either been writing a cache file... my $serializer = Data::Serializer->new( cipher => 'Rijndael', digester => 'MD5', secret => 'change me, ); $serializer->store($structure, $cache_file); Or reading a cache file... my $serializer = Data::Serializer->new( cipher => 'Rijndael', digester => 'MD5', secret => 'change me', ); my $ref_structure = $serializer->retrieve($cache_file); Best wishes Jeremy Show quoted text

-----Original Message----- From: Neil A. Neely via RT [mailto:bug-Data-Serializer@rt.cpan.org] Sent: Wednesday, December 29, 2010 4:53 PM To: Martin, Jeremy Subject: [rt.cpan.org #39592] Attempt to free .... during global destruction. <URL: https://rt.cpan.org/Ticket/Display.html?id=39592 > The relevant code that references is the following: sub DESTROY { my $self = (shift); undef %_internal; } Which essentially is just ensuring that an internal data structure is cleaned out between invocations to ensure we don't slowly leak memory in very long memory resident runs. I cannot reproduce the problem you are reporting, can you provide me any means of reproducing this problem?

Wed Dec 29 18:06:13 2010 The RT System itself - Status changed from 'stalled' to 'open'

Mon Jan 03 14:18:06 2011 neil [...] neely.cx - Correspondence added

On Wed Dec 29 18:06:12 2010, jmartin@gsihosting.com wrote: Show quoted text

> Well, that was from 2 years ago, and that OS (RedHat Enterprise Linux > 3 aka CentOS 3) is now End Of Life'd. :-)

Dove much deeper into this problem - the override to DESTROY was only put in place to try and stomp a memory leak issue related towards bloat in the %_internal data structure over time. The much simpler solution was to eliminate %_internal completely and no longer require an override to DESTROY, and we should no longer have any risk of confusing perl's Garbage collector. I've uploaded this as 0.52.

Mon Jan 03 14:18:08 2011 neil [...] neely.cx - Status changed from 'open' to 'resolved'