Hard lockup on 5.3-STABLE

[Kris Kennaway]

On Tue, Feb 08, 2005 at 10:56:28PM -0500, James Welsh wrote:
> Hi,
> 
> Our AMD Opteron server is locking up every few hours - no core dumps being
> made, and by the look of the output in /var/log/messages, whatever is
> occurring is fairly severe:
> 
> Feb  9 03:29:26 pollux kernel:
> Feb  9 03:29:26 pollux kernel:
> Feb  9 03:29:26 pollux kernel: Fatal trap 12: page fault while in kernel
> mode
> Feb  9 03:29:26 pollux kernel: cpuid = 1; apic id = 01
> Feb  9 03:29:26 pollux kernel: fault virtual address    = 0x28
> Feb  9 03:29:26 pollux kernel: fault code               = supervis
> Feb  9 03:29:26 pollux kernel: or
> Feb  9 03:29:26 pollux kernel: wr
> Feb  9 03:29:26 pollux kernel: it
> Feb  9 03:29:26 pollux kernel: e
> Feb  9 03:29:26 pollux kernel: , page
> Feb  9 03:29:26 pollux kernel: n
> Feb  9 03:29:26 pollux kernel: ot pr
> Feb  9 03:29:26 pollux kernel: esent
> Feb  9 03:29:26 pollux kernel:
> Feb  9 03:29:26 pollux kernel: in
> Feb  9 03:29:26 pollux kernel: st
> 
> The machine then reboots.

Are you certain you have crashdumps configured properly?  If you break
to DDB and 'call doadump', does it dump and savecore?

If yes, then you'll at least need to transcribe the instruction
pointer from the system console when the system panics (e.g. remove
KDB_UNATTENDED so that it doesn't automatically reboot after the panic
so you can copy it down, or set up a serial console and log the panic
output there).

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-amd64/attachments/20050208/bdc9da85/attachment.bin