>> I was doing a "mt /dev/rst0 offline", and an ls hung [...]

>> Note: the controlling terminal is hung (console), but I can telnet
>> into the box.

> This is what has been happening to me lately.  I have often observed
> it on one or more of the overview process of cnews, pppd, uucico and
> cc.

Something similar happened to me.  Yesterday, I installed a new kernel,
the first new kernel in about two weeks.  I left a build of the world
going and went home.  Today I get in and find that the build has hung.
But I could log in on the conjsole and run ps and such just fine; it
was just that process, not all access to the disk, that jammed.

I crashed the machine with L1-A to DDB, then "mach prom" and "sync"
(which produces "panic: PROM sync command") and then telling ddb to
continue and take a coredump.  So I have a kernel coredump; ps axlk on
it (ok, ok, ps alx with -M and -N) reports the hung process as

    0  5780 -107802624  17  -5  0  1784    0 vflush DW   ??    0:00.25 (cc1)

Is there anything useful I could do with the coredump to help anyone
figure out what's wrong?  Does NetBSD's kernel postmortem gdb have any
way to look at things like non-curproc process's kernel stack?

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu