PROBLEM: 2.6.18.2 kernel BUG at mm/vmscan.c:606!
Hi all,
second try via the public list since my first post was rejected due to containing much more than 100.000 characters...
Almost daily, my IBM ThinkPad T23 (Modell 2647-9RG) with 1GB of RAM and normally running the latest of Debian/unstable is getting an Oops displaying "kernel BUG at mm/vmscan.c:606!" in the syslog. This happened both with 2.6.18.1 and now 2.6.18.2 various times. Simply going back to 2.6.17.11 or even former kernel versions usually resumes rock stable operation, but this is no resolution for the bug in 2.6.18.*. Two times i experienced the machine to simply stop responding after/while over night without using it. The last two times this had happened, i was doing an "apt-get dist-upgrade" for Debian/unstable and the oops happened in the middle of package installation.
The kernel was compiled using Manoj Srivasta's kernelpackage, using the command line 'MAKEFLAGS="CC=gcc-3.4" make-kpkg kernel_image' to create a Debian kernel package with gcc-3.4.6 for compilation.
I followed the instructions at www.tux.org/lkml/reporting-bugs.html as closely as i could and have attached the requested information in separate files attached in Oops.tgz. I'm not subscribed to the list and would therefore
appreciate CC'ing for possible followups, but try to follow it via www.lkml.org nonetheless. I would be delighted to further provide any information and further debugging in resolving the issue at hand.
Thanks for Linux and all
P. *8^)
Quoted 4 lines · On Sun, 12 Nov 2006 20:10:12 +0100 Paul Seelig <[email protected]> wrote:
Almost daily, my IBM ThinkPad T23 (Modell 2647-9RG) with 1GB of RAM and normally running the latest of Debian/unstable is getting an Oops displaying "kernel BUG at
mm/vmscan.c:606!" in the syslog. This happened both with 2.6.18.1 and now 2.6.18.2 various times.
1) you don't have to run ksymoops with 2.6.x kernels (only with <=2.4.x)
2) your kernel is tainted:ath_hal: module license 'Proprietary' taints kernel.
http://www.tux.org/lkml/#s1-18
3) your kernel is also patched (Suspend2 isn't in vanilla kernels)
Suspend2 Core.
Suspend2 Compression Driver loading.
Suspend2 Encryption Driver loading.
So I suggest you to reproduce the problem with a vanilla/not tainted 2.6.18.2:
http://kernel.org/pub/linux/kernel/v2.6/linux-2.6.18.2.tar.bz2
Signature · 2 lines
Paolo Ornati Linux 2.6.19-rc5 on x86_64
Quoted 9 lines · Paolo Ornati
3) your kernel is also patched (Suspend2 isn't in vanilla kernels)
Suspend2 Core.
Suspend2 Compression Driver loading.
Suspend2 Encryption Driver loading.So I suggest you to reproduce the problem with a vanilla/not tainted 2.6.18.2:
Thanks for pointing this out! I've been running a selfcompiled vanilla kernel for the last three days without any of the reported problems reappearing! :-)
As it looks like, the culprit seemed to be the suspend2 patch. I was not aware of newer suspend2 patches for 2.6.18.2 and kept using the ones for 2.6.18 instead, which have applied perfectly.
Today i fetched the newest suspend2 patches and rebooted the machine with a 2.6.18.2 kernel containing them. Now let's see if the problem i reported persists. If yes, i'll report them to Nigel Cunningham instead.
Thanks for the help!
Quoted 3 lines · On Thu, 16 Nov 2006 02:43:30 +0100 Paul Seelig <[email protected]> wrote:
Today i fetched the newest suspend2 patches and rebooted the machine with a 2.6.18.2 kernel containing them. Now let's see if the problem i reported persists. If yes, i'll report them to Nigel Cunningham instead.
You can also do suspend without external patches... for example with
http://suspend.sourceforge.net/
it's also packaged by Debian:
http://packages.debian.org/testing/admin/uswsusp
:)
Signature · 2 lines
Paolo Ornati Linux 2.6.19-rc4-g2de6c39f on x86_64