2006-03-05 07:44:22

by Dave Jones

Subject: edac slab corruption.

rmmod e752x_edac edac_mc
Wait a few seconds...

EDAC MC0: Removed device 0 for "e752x_edac" E7525: PCI 0000:00:00.0
Slab corruption: start=ffff81003fc5a000, len=4096

Call Trace: <ffffffff8017bcab>{check_poison_obj+121}
<ffffffff802028a8>{kobject_uevent+676} <ffffffff8017be34>{cache_alloc_debugcheck_after+48}
<ffffffff802028a8>{kobject_uevent+676} <ffffffff8017dd52>{__kmalloc_track_caller+301}
<ffffffff802db731>{__alloc_skb+97} <ffffffff80202701>{kobject_uevent+253}
<ffffffff802028a8>{kobject_uevent+676} <ffffffff80201fd6>{kobject_unregister+14}
<ffffffff8026f273>{bus_remove_driver+126} <ffffffff8027004a>{driver_unregister+9}
<ffffffff8020f4a8>{pci_unregister_driver+16} <ffffffff8014ec37>{sys_delete_module+551}
<ffffffff8010dbdc>{syscall_trace_enter+156} <ffffffff8010a91c>{tracesys+209}
0f0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6a 6b 6b 6b

Dave



--
http://www.codemonkey.org.uk

2006-03-06 17:45:06

by Dave Peterson

Subject: Re: edac slab corruption.

On Saturday 04 March 2006 23:43, Dave Jones wrote:
> rmmod e752x_edac edac_mc
> Wait a few seconds...
>
> EDAC MC0: Removed device 0 for "e752x_edac" E7525: PCI 0000:00:00.0
> Slab corruption: start=ffff81003fc5a000, len=4096
>
> Call Trace: <ffffffff8017bcab>{check_poison_obj+121}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff8017be34>{cache_alloc_debugcheck_after+48}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff8017dd52>{__kmalloc_track_caller+301}
> <ffffffff802db731>{__alloc_skb+97} <ffffffff80202701>{kobject_uevent+253}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff80201fd6>{kobject_unregister+14}
> <ffffffff8026f273>{bus_remove_driver+126}
> <ffffffff8027004a>{driver_unregister+9}
> <ffffffff8020f4a8>{pci_unregister_driver+16}
> <ffffffff8014ec37>{sys_delete_module+551}
> <ffffffff8010dbdc>{syscall_trace_enter+156}
> <ffffffff8010a91c>{tracesys+209} 0f0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b
> 6a 6b 6b 6b
>
> Dave

Which version of the EDAC code was this observed with?

2006-03-06 17:53:39

by Dave Jones

Subject: Re: edac slab corruption.

On Mon, Mar 06, 2006 at 09:44:53AM -0800, Dave Peterson wrote:
> On Saturday 04 March 2006 23:43, Dave Jones wrote:
> > rmmod e752x_edac edac_mc
> > Wait a few seconds...
> >
> > EDAC MC0: Removed device 0 for "e752x_edac" E7525: PCI 0000:00:00.0
> > Slab corruption: start=ffff81003fc5a000, len=4096
> >
> > Call Trace: <ffffffff8017bcab>{check_poison_obj+121}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff8017be34>{cache_alloc_debugcheck_after+48}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff8017dd52>{__kmalloc_track_caller+301}
> > <ffffffff802db731>{__alloc_skb+97} <ffffffff80202701>{kobject_uevent+253}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff80201fd6>{kobject_unregister+14}
> > <ffffffff8026f273>{bus_remove_driver+126}
> > <ffffffff8027004a>{driver_unregister+9}
> > <ffffffff8020f4a8>{pci_unregister_driver+16}
> > <ffffffff8014ec37>{sys_delete_module+551}
> > <ffffffff8010dbdc>{syscall_trace_enter+156}
> > <ffffffff8010a91c>{tracesys+209} 0f0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b
> > 6a 6b 6b 6b
> >
> > Dave
>
> Which version of the EDAC code was this observed with?

2.6.16rc5-gitX (I forget the exact -git release, but it's
not relevant as edac hasn't changed recently).

Dave

--
http://www.codemonkey.org.uk

2006-03-07 16:53:16

by Dave Peterson

Subject: Re: edac slab corruption.

On Saturday 04 March 2006 23:43, Dave Jones wrote:
> rmmod e752x_edac edac_mc
> Wait a few seconds...
>
> EDAC MC0: Removed device 0 for "e752x_edac" E7525: PCI 0000:00:00.0
> Slab corruption: start=ffff81003fc5a000, len=4096
>
> Call Trace: <ffffffff8017bcab>{check_poison_obj+121}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff8017be34>{cache_alloc_debugcheck_after+48}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff8017dd52>{__kmalloc_track_caller+301}
> <ffffffff802db731>{__alloc_skb+97} <ffffffff80202701>{kobject_uevent+253}
> <ffffffff802028a8>{kobject_uevent+676}
> <ffffffff80201fd6>{kobject_unregister+14}
> <ffffffff8026f273>{bus_remove_driver+126}
> <ffffffff8027004a>{driver_unregister+9}
> <ffffffff8020f4a8>{pci_unregister_driver+16}
> <ffffffff8014ec37>{sys_delete_module+551}
> <ffffffff8010dbdc>{syscall_trace_enter+156}
> <ffffffff8010a91c>{tracesys+209} 0f0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b
> 6a 6b 6b 6b
>
> Dave

Can this behavior be reproduced relatively easily? One thing that
may be worth checking is whether this reproduces with a
2.6.16-rc5-mm2 kernel. Some EDAC bug fixes were added to that kernel

2006-03-07 16:57:25

by Dave Jones

Subject: Re: edac slab corruption.

On Tue, Mar 07, 2006 at 08:53:02AM -0800, Dave Peterson wrote:
> On Saturday 04 March 2006 23:43, Dave Jones wrote:
> > rmmod e752x_edac edac_mc
> > Wait a few seconds...
> >
> > EDAC MC0: Removed device 0 for "e752x_edac" E7525: PCI 0000:00:00.0
> > Slab corruption: start=ffff81003fc5a000, len=4096
> >
> > Call Trace: <ffffffff8017bcab>{check_poison_obj+121}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff8017be34>{cache_alloc_debugcheck_after+48}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff8017dd52>{__kmalloc_track_caller+301}
> > <ffffffff802db731>{__alloc_skb+97} <ffffffff80202701>{kobject_uevent+253}
> > <ffffffff802028a8>{kobject_uevent+676}
> > <ffffffff80201fd6>{kobject_unregister+14}
> > <ffffffff8026f273>{bus_remove_driver+126}
> > <ffffffff8027004a>{driver_unregister+9}
> > <ffffffff8020f4a8>{pci_unregister_driver+16}
> > <ffffffff8014ec37>{sys_delete_module+551}
> > <ffffffff8010dbdc>{syscall_trace_enter+156}
> > <ffffffff8010a91c>{tracesys+209} 0f0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b
> > 6a 6b 6b 6b
> >
> > Dave
>
> Can this behavior be reproduced relatively easily?

yep.

> One thing that
> may be worth checking is whether this reproduces with a
> 2.6.16-rc5-mm2 kernel. Some EDAC bug fixes were added to that kernel

I've got zero time to jump on this right now -- trying to get FC5 shippable :-/
I'll try and get back to it at some point.

Dave

--
http://www.codemonkey.org.uk