Re: 2.6.18-1.2689.fc6PAE: oops in ext3_clear_inode+0x52/0x8b

[Eric Sandeen <sandeen@sandeen.net>]

Andrew Morton wrote:
> On Thu, 28 Sep 2006 13:42:26 -0700
> Jeremy Fitzhardinge <jeremy@goop.org> wrote:
> 
>> I just filed this in the Redhat bugzilla, since its from the FC6 distro 
>> kernel.  But since its fairly close to current kernel.org kernels, I 
>> thought it might be relevent.
>>
>> The bug is https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=208488
>>
>> Unfortunately this isn't a very useful report since it was a once-off, 
>> and there's a P-tainting module in there.  But if anyone sees anything 
>> else like this, it's interesting.
>>
>> The oops is:
>>
>> BUG: unable to handle kernel paging request at virtual address 756e6547
> 
> 756e6547 -> uneG.   Matches "GenuineIntel".
> 
> That'll get written into a temporary page by the /proc/cpuinfo handler, so
> it might just be a use-uninitialised.

But strangely enough, it's the second report we've seen with this exact 
backtrace, and the same "Genu" ascii string where the i_default_acl should be.

Both boxes had been through a suspend-to-ram recently, just in case that might 
matter.

Seems like something more than random chance...

-Eric