Unable to handle kernel paging request

[Philippe Clérié]

Binary package hint: linux-source-2.6.22

Kernel Oops whenever memory used goes over 1GB.

System description:
Gigabyte GA-MA69GM-S2H
Athlon X2 6000+
2GB memory
320GB SATA Hard Drive +
linux 2.6.22-14.46 with vserver patch (from multivax.ca)
and 2.6.22-14.47 generic.

I have attached the relevant portions of /var/log/kern.log. I've included the info from the vserver kernel, even though it's not a supported version, because of the trace output, which I did not get from the generic kernel. I'm sure I could have produced the trace from the generic kernel, but at this point my root partition seems to be gone plus I'm fairly certain I've correctly identify the problem.

I'm also running the same kernel on a Core 2 Duo laptop with 1.5GB of memory and there's no problem there. I also have a system with a 32 bit Athlon and 2GB of RAM running just fine. So, the problem appears to be specific to the Athlon X2.

I can reproduce the problem consistently by using tar to pack or unpack a large file such as a vserver; by using rsync or cp on large directories with large files (I used /var/cache/apt/archives); when packaging using pbuilder.

[Philippe Clérié (pclerie)]

I recompiled hardy kernel 2.6.24-11.17 for gutsy on my system and I'm pleased to report that this problem appears to have been fixed. So far every operation that Oopsed now terminates normally. I also watched as free -s1 reached and crossed the 1GB threshold several times. So I'm happy that hardy is fixed.

Hopefully there'll be a fix for gutsy, because upgrading to 2.6.24 is problematic for me unless I can find linux-vserver patches for that kernel. Officially, there aren't any.

Cheers!

[Leann Ogasawara (leannogasawara)]

Hi Philippe,

Thanks for testing Hardy. I'm glad to hear things are are at least working there. I'm going to mark this "Fix Released" against the Hardy kernel. I'm not positive if this will qualify for a stable release update to Gutsy, but I'll nudge the kernel team and ask. Thanks.

[Leann Ogasawara (leannogasawara)]

Hi Philippe,

Just chatted with the kernel team. Unfortunately you've caught us at a rather busy time in our release schedule. We are almost at feature freeze for the Hardy kernel and are focusing most of our efforts to make sure it is stable. As a result it's unsure how much time can be spent researching the exact patch to backport to Gutsy for this issue. Would you be willing to help narrow down the specific patch to backport? Using something like git-bisect can help with this: http://www.kernel.org/doc/local/git-quick.html#bisect . Also please be sure to do this on the stock Ubuntu kernel and not the vserver one. Thanks!

[Philippe Clérié (pclerie)]

Will try over the week-end. Might take longer though. I don't know anything about git.

Wish me luck! :-)

Cheers

[Philippe Clérié (pclerie)]

Quick question: are we looking for the patch that corrected the problem
or the patch that caused the problem in the first place?

Still trying to figure out how to do this.

Philippe

------
The trouble with common sense is that it is so uncommon.
<Anonymous>

On Thursday 06 March 2008 12:47:20 Leann Ogasawara wrote:
> Hi Philippe,
>
> Just chatted with the kernel team. Unfortunately you've caught us at
> a rather busy time in our release schedule. We are almost at feature
> freeze for the Hardy kernel and are focusing most of our efforts to
> make sure it is stable. As a result it's unsure how much time can be
> spent researching the exact patch to backport to Gutsy for this
> issue. Would you be willing to help narrow down the specific patch
> to backport? Using something like git-bisect can help with this:
> http://www.kernel.org/doc/local/git-quick.html#bisect . Also please
> be sure to do this on the stock Ubuntu kernel and not the vserver
> one. Thanks!
>
> ** Changed in: linux-source-2.6.22 (Ubuntu)
> Status: New => Incomplete

[Philippe Clérié (pclerie)]

I reported earlier that kernel 2.6.24-11.17 was not affected. That turns out to be incorrect. With a default initrd it works just fine. Adding vesafb and fbcon to the initrd appears to trigger the bug. At least it looks that way to me.

kern.log attached.

[Sergio Zanchetta (primes2h)]

The 18 month support period for Gutsy Gibbon 7.10 has reached its end of life -
http://www.ubuntu.com/news/ubuntu-7.10-eol . As a result, we are closing the
linux-source-2.6.22 kernel task.