Bug #50454 for Class-XSAccessor: Serious issues in mod

Tue Oct 13 11:36:24 2009 frioux [...] gmail.com - Ticket created

Subject:

Serious issues in mod_perl + win32

We are using DBIx::Class, which uses Class::Accessor::Grouped, which in turn optionally uses this module. We found that in win32, if we hammered the server with a load testing tool we would get the following error: Free to wrong pool 2063268 not a378b0 at C:/Perl/site/lib/Class/Accessor/Grouped.pm line 102. The line of code that references is Class::XSAccessor::newxs_accessor("${class}::${name}", $field, 0); In CAG we disabled using this module (comment out line 21) and the issue went away. I really don't know how to help debug this issue, but I'd love to get it solved. I'm on IRC (Freenode and Perl) as frew almost all the time and of course you have my email. Look forward to talking with you!

Tue Oct 13 22:54:09 2009 smueller [...] cpan.org - Correspondence added

Hi, thanks for tracking this down! On Tue Oct 13 08:36:24 2009, frew wrote: Show quoted text

> We are using DBIx::Class, which uses Class::Accessor::Grouped, which in > turn optionally uses this module. We found that in win32, if we > hammered the server with a load testing tool we would get the following > error: > > Free to wrong pool 2063268 not a378b0 at > C:/Perl/site/lib/Class/Accessor/Grouped.pm line 102. > > The line of code that references is > > Class::XSAccessor::newxs_accessor("${class}::${name}", $field, 0); > > In CAG we disabled using this module (comment out line 21) and the issue > went away. I really don't know how to help debug this issue, but I'd > love to get it solved. I'm on IRC (Freenode and Perl) as frew almost > all the time and of course you have my email. > > Look forward to talking with you!

I'm currently travelling and will have access to a usable computer only in November, so I can't really debug this now. Either way, some comments: 1) This may be a nightmare to debug considering it happens when you hammer the server with requests. The "newxs_accessor"-using line you quote above should mostly be called *early* during the creation of a new process since it's installing a new method. It seems strange that the error happens due to high load unless your setup does funny stuff that spawns new threads or processes as load increases. Specifically threads may be an issue since the share the same C-level memory which Class::XSAccessor uses. Maybe my thread-safety hacking wasn't as good as I thought and I'd have to add more locking pain. The only case in which I can see this happening is when two threads try to extend the CXA global state at the same time. 2) Are these crashes reliably reproducable for you? Does this happen on another OS? Could you run this in a C debugger to see where the bad free happens? 3) Class::Accessor::Grouped has NO business calling newxs_accessor. The newxs* functions are NOT part of the API of Class::XSAccessor and considering they might be doing scary shit (i.e. produce segfaults) when used improperly, they're unlikely to become public. Besides, the public API of C::XSA can do the same thing (albeit with slightly higher overhead which doesn't really matter given it's a one-time cost). Don't expect much input besides this before November. I expect to have little connectivity. Best regards, Steffen

Tue Oct 13 22:54:10 2009 The RT System itself - Status changed from 'new' to 'open'

Wed Oct 14 10:42:36 2009 frioux [...] gmail.com - Correspondence added

Thanks for the fast response! to answer your questions in order: yes we can reproduce it fairly easily, and we haven't yet tried it in another OS. We will surely at some point, but the fire is put out now so we have less motivation :-) And as for a debugger, we'll probably have to do that in linux as I have no idea how to attach gdb in windows. I have commit access to Class::Accessor::Grouped, so I'll look at fixing it so that it doesn't use those functions. So before November I'll fix CAG to use the right functions, and then I'll wait till you get back to do more with this.

Mon Nov 02 12:56:00 2009 smueller [...] cpan.org - Correspondence added

Attached, simple test script demonstrates the (or *a*) crash. Reliably gives something like: Scalars leaked: -1800 Scalars leaked: 100 ... Occasionally also: Attempt to free non-existent shared string 'cCWeA', Perl interpreter: 0x251ee40 during global destruction. Attempt to free non-existent shared string 'LynVM', Perl interpreter: 0x251ee40 during global destruction. I'm experimenting with a potential fix, but it's currently colliding with other pending (complex) refactoring and optimization changes. Cheers, Steffen

#!/usr/bin/perl use strict; use warnings; use threads; use Class::XSAccessor constructor => 'new', class => 'Foo'; my @chars = ('a'..'z', 'A'..'Z'); my @t; foreach (1..20) { push @t, threads->new(sub { foreach (1..100) { my $foo = join '', map {$chars[rand(@chars)]} 1..5; Class::XSAccessor->import( class => 'Foo', getters => {$foo => $foo} ); } }); } $_->join for @t;

Mon Nov 02 12:57:31 2009 smueller [...] cpan.org - Correspondence added

Oops, simpler script attached.

#!/usr/bin/perl use strict; use warnings; use threads; require Class::XSAccessor; my @chars = ('a'..'z', 'A'..'Z'); my @t; foreach (1..20) { push @t, threads->new(sub { foreach (1..100) { my $foo = join '', map {$chars[rand(@chars)]} 1..5; Class::XSAccessor->import( class => 'Foo', getters => {$foo => $foo} ); } }); } $_->join for @t;

Mon Nov 02 12:57:53 2009 smueller [...] cpan.org - Broken in 1.03 added

Mon Nov 02 12:57:53 2009 smueller [...] cpan.org - Broken in 1.04_02 added

Mon Nov 02 12:57:53 2009 smueller [...] cpan.org - Severity Critical added

Tue Nov 03 16:37:24 2009 smueller [...] cpan.org - Correspondence added

Fixed in trunk according to reporter. Dev release 1.04_03 was uploaded to PAUSE. Stable release blocking for other changes. --Steffen

Tue Nov 03 16:37:27 2009 smueller [...] cpan.org - Status changed from 'open' to 'resolved'

Wed Sep 01 19:09:39 2010 ribasushi [...] leporine.io - Correspondence added

On Tue Nov 03 16:37:24 2009, SMUELLER wrote: Show quoted text

> Fixed in trunk according to reporter. Dev release 1.04_03 was uploaded > to PAUSE. Stable release blocking for other changes. --Steffen

Reopening this can of fat worms. Apparently there was a lapse in testing, as neither 1.04_03, nor any more recent versions solve the issue on win32. Testing was performed on a virgin install of strawberry perl 5.12.1, with cpanm downloaded to ease specific version installations. No other cpan modules were installed: FAIL log =============== C:\>perl cpanm http://backpan.perl.org/authors/id/S/SM/SMUELLER/Class-XSAccessor -1.04_03.tar.gz Fetching http://backpan.perl.org/authors/id/S/SM/SMUELLER/Class-XSAccessor-1.04_ 03.tar.gz ... OK Configuring Class-XSAccessor-1.04_03 ... OK Building and testing Class-XSAccessor-1.04_03 for http://backpan.perl.org/author s/id/S/SM/SMUELLER/Class-XSAccessor-1.04_03.tar.gz ... OK Successfully installed Class-XSAccessor-1.04_03 C:\>perl xsa_crash2.pl Free to wrong pool ac7e58 not a57230 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 81. C:\>perl xsa_crash2.pl Free to wrong pool ac7e58 not a57230 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 81. C:\>perl cpanm http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAcce ssor-1.07.tar.gz Fetching http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAccessor-1 .07.tar.gz ... OK Configuring Class-XSAccessor-1.07 ... OK Building and testing Class-XSAccessor-1.07 for http://search.cpan.org/CPAN/autho rs/id/S/SM/SMUELLER/Class-XSAccessor-1.07.tar.gz ... OK Successfully installed Class-XSAccessor-1.07 C:\>perl xsa_crash2.pl Free to wrong pool ac65d0 not a57238 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 84. C:\>perl xsa_crash2.pl Free to wrong pool ac65d0 not a57238 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 84. C:\>perl cpanm http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAcce ssor-1.07_02.tar.gz Fetching http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAccessor-1 .07_02.tar.gz ... OK Configuring Class-XSAccessor-1.07_02 ... OK Building and testing Class-XSAccessor-1.07_02 for http://search.cpan.org/CPAN/au thors/id/S/SM/SMUELLER/Class-XSAccessor-1.07_02.tar.gz ... OK Successfully installed Class-XSAccessor-1.07_02 C:\>perl xsa_crash2.pl Free to wrong pool ac8f78 not a58a38 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 87. C:\>perl xsa_crash2.pl Free to wrong pool ac8f78 not a58a38 at C:/strawberry/perl/site/lib/Class/XSAcce ssor.pm line 87.

Wed Sep 01 19:09:40 2010 The RT System itself - Status changed from 'resolved' to 'open'

Wed Sep 01 19:15:48 2010 ribasushi [...] leporine.io - Broken in 1.04_03 added

Wed Sep 01 19:15:48 2010 ribasushi [...] leporine.io - Broken in 1.07 added

Wed Sep 01 19:15:48 2010 ribasushi [...] leporine.io - Broken in 1.07_02 added

Thu Sep 02 04:05:31 2010 smueller [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #50454] Serious issues in mod_perl + win32
Date:	Thu, 02 Sep 2010 10:05:18 +0200
To:	bug-Class-XSAccessor [...] rt.cpan.org
From:	Steffen Mueller <smueller [...] cpan.org>

Hi Peter, thanks for the report. Peter Rabbitson via RT wrote: Show quoted text

> Queue: Class-XSAccessor > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=50454 > > > On Tue Nov 03 16:37:24 2009, SMUELLER wrote:

>> Fixed in trunk according to reporter. Dev release 1.04_03 was uploaded >> to PAUSE. Stable release blocking for other changes. --Steffen

> > Reopening this can of fat worms. Apparently there was a lapse in > testing, as neither 1.04_03, nor any more recent versions solve the > issue on win32.

Damn. I can't reproduce this on my linux even when hammering it with 20 threads, each installing 100k methods in parallel. That's about where I run ou t of memory. Show quoted text

> Testing was performed on a virgin install of strawberry perl 5.12.1, > with cpanm downloaded to ease specific version installations. No other > cpan modules were installed:

Still no win32 here. Not sure when or whether I'll be able to set one up. But even if I do, I will possess no win32 debugging skills at all. :( [...] Show quoted text

> C:\>perl cpanm > http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAcce > ssor-1.07_02.tar.gz > Fetching > http://search.cpan.org/CPAN/authors/id/S/SM/SMUELLER/Class-XSAccessor-1 > .07_02.tar.gz ... OK > Configuring Class-XSAccessor-1.07_02 ... OK > Building and testing Class-XSAccessor-1.07_02 for > http://search.cpan.org/CPAN/au > thors/id/S/SM/SMUELLER/Class-XSAccessor-1.07_02.tar.gz ... OK > Successfully installed Class-XSAccessor-1.07_02 > > C:\>perl xsa_crash2.pl > Free to wrong pool ac8f78 not a58a38 at > C:/strawberry/perl/site/lib/Class/XSAcce > ssor.pm line 87. > > C:\>perl xsa_crash2.pl > Free to wrong pool ac8f78 not a58a38 at > C:/strawberry/perl/site/lib/Class/XSAcce > ssor.pm line 87.

I'll try staring at the locking code a bit. But I don't expect enlightenment. --Steffen

Thu Sep 02 10:10:30 2010 smueller [...] cpan.org - Correspondence added

After serious hacking and much cursing, I have high hope that it might be fixed on this branch: http://github.com/tsee/Class-XSAccessor/tree/memoryCraziness Or directly for a git clone command: git://github.com/tsee/Class-XSAccessor.git Feedback welcome!

Thu Sep 02 14:18:25 2010 frioux [...] gmail.com - Correspondence added

On Thu Sep 02 10:10:30 2010, SMUELLER wrote: Show quoted text

> After serious hacking and much cursing, I have high hope that it might > be fixed on this branch: > > http://github.com/tsee/Class-XSAccessor/tree/memoryCraziness > > Or directly for a git clone command: > > git://github.com/tsee/Class-XSAccessor.git > > Feedback welcome!

Sorry, we tried it, and although we don't get the errors we got before, we now get new errors. Here are our new errors: [Dispatch] ERROR for request '/dispatch/AssignmentRule/single_mgmtid?_dc=1283451393808&mgmtid=756': Error executing run mode 'single_mgmtid': DBIx::Class::Schema::resultset(): Can't find source for at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 435 CGI::Application::Dispatch::http_error('EPMS::Dispatcher', 'Exception::Class::Base=HASH(0x9fd6574)', undef) called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 409 CGI::Application::Dispatch::dispatch('EPMS::Dispatcher', 'args_to_new', 'HASH(0x18fc2a04)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 809 CGI::Application::Dispatch::handler('EPMS::Dispatcher', 'Apache2::RequestRec=SCALAR(0x18fc2444)') called at -e line 0 eval {...} called at -e line 0 Use of uninitialized value $moniker in exists at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 593 DBIx::Class::Schema::source('EPMS::Schema=HASH(0xb3aa44c)', undef) called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSourceHandle.pm line 68 DBIx::Class::ResultSourceHandle::resolve('DBIx::Class::ResultSourceHandle=HASH(0x95d79b4)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSet.pm line 208 DBIx::Class::ResultSet::new('DBIx::Class::ResultSet', 'DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)', 'HASH(0xb3aa4fc)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSource.pm line 849 DBIx::Class::ResultSource::resultset('DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)') called at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 554 DBIx::Class::Schema::resultset('EPMS::Schema=HASH(0xb3aa44c)', 'Mgmt') called at /Inetpub/epms/Controller/Mgmt.pm line 107 Controller::Mgmt::single('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 143 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application.pm line 142 CGI::Application::__get_body('Controller::Mgmt=HASH(0xef154c)', 'single') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 191 CGI::Application::run('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 704 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 701 CGI::Application::Dispatch::_run_app('EPMS::Dispatcher', 'Controller::Mgmt', 'single', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 406 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 368 CGI::Application::Dispatch::dispatch('EPMS::Dispatcher', 'args_to_new', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 809 CGI::Application::Dispatch::handler('EPMS::Dispatcher', 'Apache2::RequestRec=SCALAR(0x47877f4)') called at -e line 0 eval {...} called at -e line 0 Use of uninitialized value $moniker in hash element at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 596 DBIx::Class::Schema::source('EPMS::Schema=HASH(0xb3aa44c)', undef) called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSourceHandle.pm line 68 DBIx::Class::ResultSourceHandle::resolve('DBIx::Class::ResultSourceHandle=HASH(0x95d79b4)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSet.pm line 208 DBIx::Class::ResultSet::new('DBIx::Class::ResultSet', 'DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)', 'HASH(0xb3aa4fc)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSource.pm line 849 DBIx::Class::ResultSource::resultset('DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)') called at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 554 DBIx::Class::Schema::resultset('EPMS::Schema=HASH(0xb3aa44c)', 'Mgmt') called at /Inetpub/epms/Controller/Mgmt.pm line 107 Controller::Mgmt::single('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 143 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application.pm line 142 CGI::Application::__get_body('Controller::Mgmt=HASH(0xef154c)', 'single') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 191 CGI::Application::run('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 704 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 701 CGI::Application::Dispatch::_run_app('EPMS::Dispatcher', 'Controller::Mgmt', 'single', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 406 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 368 CGI::Application::Dispatch::dispatch('EPMS::Dispatcher', 'args_to_new', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 809 CGI::Application::Dispatch::handler('EPMS::Dispatcher', 'Apache2::RequestRec=SCALAR(0x47877f4)') called at -e line 0 eval {...} called at -e line 0 Use of uninitialized value $moniker in concatenation (.) or string at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 597 DBIx::Class::Schema::source('EPMS::Schema=HASH(0xb3aa44c)', undef) called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSourceHandle.pm line 68 DBIx::Class::ResultSourceHandle::resolve('DBIx::Class::ResultSourceHandle=HASH(0x95d79b4)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSet.pm line 208 DBIx::Class::ResultSet::new('DBIx::Class::ResultSet', 'DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)', 'HASH(0xb3aa4fc)') called at C:/strawberry/perl/site/lib/DBIx/Class/ResultSource.pm line 849 DBIx::Class::ResultSource::resultset('DBIx::Class::ResultSource::Table=HASH(0xbbe3d3c)') called at C:/strawberry/perl/site/lib/DBIx/Class/Schema.pm line 554 DBIx::Class::Schema::resultset('EPMS::Schema=HASH(0xb3aa44c)', 'Mgmt') called at /Inetpub/epms/Controller/Mgmt.pm line 107 Controller::Mgmt::single('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 143 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application.pm line 142 CGI::Application::__get_body('Controller::Mgmt=HASH(0xef154c)', 'single') called at C:/strawberry/perl/site/lib/CGI/Application.pm line 191 CGI::Application::run('Controller::Mgmt=HASH(0xef154c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 704 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 701 CGI::Application::Dispatch::_run_app('EPMS::Dispatcher', 'Controller::Mgmt', 'single', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 406 eval {...} called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 368 CGI::Application::Dispatch::dispatch('EPMS::Dispatcher', 'args_to_new', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 809 CGI::Application::Dispatch::handler('EPMS::Dispatcher', 'Apache2::RequestRec=SCALAR(0x47877f4)') called at -e line 0 eval {...} called at -e line 0 [Dispatch] ERROR for request '/dispatch/mgmt/single?_dc=1283451394491&mgmtid=756': Error executing run mode 'single': DBIx::Class::Schema::resultset(): Can't find source for at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 435 CGI::Application::Dispatch::http_error('EPMS::Dispatcher', 'Exception::Class::Base=HASH(0xb88eeb4)', undef) called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 409 CGI::Application::Dispatch::dispatch('EPMS::Dispatcher', 'args_to_new', 'HASH(0x942e50c)') called at C:/strawberry/perl/site/lib/CGI/Application/Dispatch.pm line 809 CGI::Application::Dispatch::handler('EPMS::Dispatcher', 'Apache2::RequestRec=SCALAR(0x47877f4)') called at -e line 0 eval {...} called at -e line 0 As you may be able to tell they are nearly meaningless and all over the board and probably useless. But we consistently get issues when using CXA and do not when we don't.

Thu Sep 02 14:49:42 2010 smueller [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #50454] Serious issues in mod_perl + win32
Date:	Thu, 02 Sep 2010 20:48:24 +0200
To:	bug-Class-XSAccessor [...] rt.cpan.org
From:	Steffen Müller <smueller [...] cpan.org>

Hi! Thanks for trying again. Did you try with the memory... branch or master? Could you run the test scripts that are attached to this ticket? Best regards, Steffen -- This message was sent from a phone. Apologies for brevity.

Thu Sep 02 16:42:54 2010 frioux [...] gmail.com - Correspondence added

On Thu Sep 02 14:49:42 2010, SMUELLER wrote: Show quoted text

> Hi! > > Thanks for trying again. Did you try with the memory... branch or > master? Could you run the test scripts that are attached to this > ticket? > > Best regards, > Steffen

We used the memoryCraziness branch. Yeah, we ran the test in the thread (which if I get a chance I'll make a *real* test that can run on install) and it works fine. It seems like something to do with the way mod_perl does things in win32?

Fri Sep 03 13:33:53 2010 frioux [...] gmail.com - Correspondence added

Latest version of the memoryCraziness branch still causes useless errors :-/

Sat Sep 04 03:54:16 2010 smueller [...] cpan.org - Correspondence added

On Fri Sep 03 13:33:53 2010, frew wrote: Show quoted text

> Latest version of the memoryCraziness branch still causes useless

errors :-/ I don't enjoy saying this, but I guess I'm officially out of clue then. Sorry, Steffen

Mon Nov 29 21:16:21 2010 TJC [...] cpan.org - Correspondence added

I'm unsure if it's related to this ticket.. but it looks similar. I'm getting segfaults due to bad memory access from Class::XSAccessor, on perl 5.12.2 on linux/amd64 architecture. It seems the issues aren't limited to mod_perl or win32, as I'm not using either of thos. ie. *** glibc detected *** perl: realloc(): invalid pointer: 0x0000000001674838 *** ======= Backtrace: ========= /lib/libc.so.6(+0x775b6)[0x7f46229f65b6] /lib/libc.so.6(realloc+0x352)[0x7f46229fd312] /usr/local/strategic/perl/lib/site_perl/5.12.2/x86_64-linux-thread- multi/auto/Class/XSAccessor/XSAccessor.so(get_hashkey_index+0x2a7) [0x7f46210dcb77] /usr/local/strategic/perl/lib/site_perl/5.12.2/x86_64-linux-thread- multi/auto/Class/XSAccessor/XSAccessor.so(XS_Class__XSAccessor_newxs_acc essor+0x1db)[0x7f46210dd5cb] perl(Perl_pp_entersub+0x530)[0x495c90] perl(Perl_runops_standard+0x16)[0x4947d6] perl(Perl_call_sv+0x4b7)[0x439a47] perl(Perl_call_list+0x2ad)[0x439f1d] perl(perl_destruct+0x170c)[0x43c3bc] perl(main+0xcb)[0x421e7b] /lib/libc.so.6(__libc_start_main+0xfd)[0x7f462299dc4d] perl[0x421ce9]

Wed Dec 01 15:08:33 2010 smueller [...] cpan.org - Correspondence added

On Mon Nov 29 21:16:21 2010, TJC wrote: Show quoted text

> I'm unsure if it's related to this ticket.. but it looks similar. > > I'm getting segfaults due to bad memory access from Class::XSAccessor, > on perl 5.12.2 on linux/amd64 architecture. It seems the issues aren't > limited to mod_perl or win32, as I'm not using either of thos. > > ie. > > *** glibc detected *** perl: realloc(): invalid pointer: > 0x0000000001674838 *** > ======= Backtrace: ========= > /lib/libc.so.6(+0x775b6)[0x7f46229f65b6] > /lib/libc.so.6(realloc+0x352)[0x7f46229fd312] > /usr/local/strategic/perl/lib/site_perl/5.12.2/x86_64-linux-thread- > multi/auto/Class/XSAccessor/XSAccessor.so(get_hashkey_index+0x2a7) > [0x7f46210dcb77] > /usr/local/strategic/perl/lib/site_perl/5.12.2/x86_64-linux-thread- > multi/auto/Class/XSAccessor/XSAccessor.so(XS_Class__XSAccessor_newxs_acc > essor+0x1db)[0x7f46210dd5cb] > perl(Perl_pp_entersub+0x530)[0x495c90] > perl(Perl_runops_standard+0x16)[0x4947d6] > perl(Perl_call_sv+0x4b7)[0x439a47] > perl(Perl_call_list+0x2ad)[0x439f1d] > perl(perl_destruct+0x170c)[0x43c3bc] > perl(main+0xcb)[0x421e7b] > /lib/libc.so.6(__libc_start_main+0xfd)[0x7f462299dc4d] > perl[0x421ce9]

... and based on this, I have uploaded Class::XSAccessor 1.10. Please retest with that version. --Steffen

Thu Dec 02 13:28:04 2010 smueller [...] cpan.org - Correspondence added

According to Peter Rabbitson, this issue is now resolved.

Thu Dec 02 13:28:04 2010 smueller [...] cpan.org - Status changed from 'open' to 'resolved'

Bug #50454 for Class-XSAccessor: Serious issues in mod_perl + win32