Jukka Salmi <j+nbsd@2006.salmi.ch> writes:

> Maybe this is related to the problem I [1]reported two days ago?

Hmm... I'll look into that, but when I tried to recompile the whole
system from scratch (that is, renamed the object directory), I could
not remove the object directory for the toolchain due to some strange
reason.  This is from memory, so the exact errors are probably
different, but `rm -rf /usr/obj/tool*' complained that the directories
are not empty, and a `rm -rf' in the appropriate directory complained
`no such file or directory'.  A `fsck -f /usr' (LFS) gave a lot of
errors (sorry, I did not write up in which stages, but I *think* it
was in stages >= 6).  Now I switched back to a two weeks older kernel
and check if I can reproduce the error with a complete rebuilt of
kernel+userland; I'll report back if I find something interesting.

> Are you running X when the system locks up? If yes, you probably don't
> see whether it drops into ddb or not. Type `sync' and watch your hd
> led... After reboot examine the crash dump to see what happened.

Ahaa! I was indeed running X, but I do not have a visible hd led.
Anyway, I'll check if the machine responds to any ddb commands next
time.

Thanks,

-- Johan