kernel BUG at fs/inode.c:250!

[u7l11ey@mail.lrz-muenchen.de]

Hi Debian community,

I have a Linksys NSLU2 running the latest Debian distribution (just
updated 2 days ago): Linux version 2.6.18-6-ixp4xx (Debian
2.6.18.dfsg.1-18etch1).

For a couple of days, the box hangs at least once a day, requiring a hard
reset by pulling the power chord. The kern.log file shows the following:

Feb 19 21:50:35 slug kernel: kernel BUG at fs/inode.c:250!
Feb 19 21:50:35 slug kernel: Unable to handle kernel NULL pointer
dereference at virtual address 00000000
Feb 19 21:50:35 slug kernel: pgd = c1778000
Feb 19 21:50:35 slug kernel: [00000000] *pgd=017b0031, *pte=00000000,
*ppte=00000000
Feb 19 21:50:35 slug kernel: Internal error: Oops: 817 [#1]

And so on. I can of course provide the complete error report including the
whole backtrace.

There are also other error messages:
Feb 20 02:50:30 slug kernel: Bad page state in process 'sh'
Feb 20 02:50:30 slug kernel: page:c02b1ac0 flags:0x00000000
mapping:00000040 mapcount:0 count:0
Feb 20 02:50:30 slug kernel: Trying to fix it up, but a reboot is needed
Feb 20 02:50:30 slug kernel: Backtrace:
Feb 20 03:20:03 slug kernel: Bad page state in process 'rrdtool'
Feb 20 03:20:03 slug kernel: page:c02b1ac0 flags:0x00000000
mapping:00000040 mapcount:0 count:0
Feb 20 03:20:03 slug kernel: Trying to fix it up, but a reboot is needed
Feb 20 03:20:03 slug kernel: Backtrace:

The box is running a cronjob every 10 minutes that processes a file with
sed, awk and rrdtool. There is no hard drive, just a 2GB USB stick with
plenty of free space.

I googled for this bug, but couldn't come up with anything helpful.

Any pointers about what I could do to get a stable system?

Regards, Richard