Subject: Re: Random hangs/crashes on 3100--anyone seen it?
To: Ben Cottrell <benco@ucsee.eecs.berkeley.edu>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
List: port-pmax
Date: 07/13/1997 19:08:57
>My 3100 is running a -current kernel from about June 8 or 9, and it's been
>experiencing some odd problems. With a frequency of about once every one or
>two weeks, it either hangs or reboots without explanation.

>When it hangs, it hangs in a rather odd, *incremental* fashion. What
>happens is I'll be sitting typing at a shell, and that shell will
f>reeze, but other shells will be fine. Then another shell will freeze,
>and then at a certain point, they all go at once.

>At first I suspected filesystem problems--I know we had problems a while
>ago with filesystems freezing up because a process was deadlocking while
>looking up a vnode--but after further analysis, I can be hitting return
>repeatedly at the shell, and not trying to do any file manipulation, and
>it will still hang. My current guess is a VM problem, but without any way
>to get a crash dump, it's a little hard to diagnose.

I think the first one sounds very much like a filesystem locking bug too. 
The second could be a VM bug or a VM locking bug.

Frank van der Linden committed locking changes (including the Lite-2
locking interface) last week.  And NetBSD/pmax kernels as of mid-June
include support for ddb, so if you can set up a serial console, you
can break into the debugger and at least get a hex stack traceback.
(no symbols or function names yet, sorry.)

>would try supping a new kernel and seeing Anyone experienced this?
>Should I just sup a new kernel and it will go away?

If you can build a new kernel with options DDB,and either see if it
goes away, or if not, figure out where the hang is, that would
be really great.