John Hawkinson <jhawk@MIT.EDU> writes:

> | 
> | In the last three days, I've had two system crashes that didn't leave
> | a coredump. (My system is setup properly to do this, it had done it
> | before, and there was enough room for it). I don't have any clue what
> | caused these crashes, although the machine was busy doing a build of
> | something in both cases.
> 
> That's not good. Do you have a persistent message buffer across
> reboots?

I don't think that's possible on a PC - the BIOS clears all of memory
during the POST memory test (at least it used to 10 years ago, and I
assume they haven't changed the behavior since then...)

> If you have DDB compiled in, the system will print a stack
> trace to the console if entry to ddb is not enabled, before rebooting.

Hmm, I didn't know it was posible to have DDB enable without enabling
"entry to ddb". What should my options be for that?

> If you have a persistent message buffer, you can find this with dmesg.
> If not, you could hook up a serial console and and log the output.

That sounds like a good idea - I'll hook something up and hope that
it crashes again.