Bug#255175: kernel-image-2.4.26-1-686: system crash due to kernel bug
On Mon, Jul 26, 2004 at 08:57:19PM +0200, Javier wrote:
> severity 255175 critical
> thanks
>
> I upgraded some time ago the kernel-image to kernel-image-2.4.26-1-686,
> version 2.4.26-2 (in testing at the moment) and these oops appear
> frequently (I would even say that with a higher frequency that with
> previous version 2.4.26-1, but I have not investigated in depth). My system
> still crashes from time to time (it will not stay up for more than a week).
> It usually will start first spouting some 'Unable to handle kernel paging
> request at virtual address' oopses but it will keep running, however, at
> some point, and after some time (see attached syslog) it will start to send
> oops: 'kernel BUG at page_alloc.c:221!' (sometimes page_alloc.c:113)
> whenever a process is spawned, rendering the system unusable. In this
> situation only a hard reboot can make it recover.
>
> I'm consequently raising this bug's priority, because this kernel image
> makes the system quite unstable. I don't know if this affects version
> 2.4.26-4 (in sid at the moment) too, it's not easy to reproduce, but I will
> install that kernel version and see what happens.
>
> Attached is a capture of the different oops I've encountered at different
> points before the system crash, these were copy&pasted directly from the
> screen. Also attached are the kernel oops that syslog recorded (as you can
> see, not all of them are included). Also, in case it's useful, I'm
> providing the output of ksymoops using this oops file. Most of the tasks
> associated with the oops are usually cron tasks and the system has plenty
> of swap space.
>
> If there is anything I can do to provide additional info, feel free to ask.
> Since I don't really know what causes it I'm at a loss in order to
> reproduce it.
Hi Javier,
thanks for the feed back. Do you have a reliable way to reproduce this.
What hardware (CPU) are you running this on?
I am wondering if what you are seeing relates to the clear_fpu() bug,
also known as CAN-2004-0554, which effects signal handling. It was
fixed in 2.4.26-3 (kernel-source and kernel-image-2.4.26-1-*) and
the fix is also present in 2.4.26-4.
http://www.ultramonkey.org/bugs/cve/CAN-2004-0554.shtml
http://marc.theaimsgroup.com/?l=linux-kernel&m=108704809114434&w=2
--
Horms
Reply to: