On Sun, 2011-06-05 at 13:51 +0100, Richard Kettlewell wrote: > > It did not happen with the same machine's previous kernel, which my > > backups indicate was 2.6.32-30 (and to which I'm planning to revert if > > it continues crashing). > > Instead I tried the corresponding 486 kernel (i.e. > linux-image-2.6.32-5-486 2.6.32-34squeeze1). The system has now stayed > up for over a week. Going back to your original report: I'm afraid I can't make any sense of the 'oops'. The faulting address doesn't seem to relate to the code or register values. The message 'Thread overran stack, or stack corrupted' suggests the task information structure might have been corrupted (as that is immediately below the stack canary) though I still don't see why that would cause an oops at this location. But then I don't think an actual stack overrun is likely - as I understand it, current concern about the possibility of stack overrun mostly involves complex stacked storage configurations, not networking. Can you provide a kernel boot log from the older kernel version (2.6.32-30)? Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse.
Attachment:
signature.asc
Description: This is a digitally signed message part