In message <w06789rr6d.fsf@rupert.honors.montana.edu>, cjones wrote: 

-> This looks like a MI problem, though I'm really not sure.
-> 
-> The machine in question is a news server, NetBSD 1.3I, i386, source
-> probably grabbed around Jan 25.  The spool is on a ccd across a 6G IDE
-> disk and 2 1G SCSI disks.  It's been panicking on an irregular basis
-> for a while now, and this is the first time I've managed to get the
-> people who watch it to actually write down a stack trace.  I'm pretty
-> sure the process in question in expire:

[...trace snipped...]

Just as a further data point, I had my i386 desktop machine do the same
thing a few days back with 1.3I of similar provenance (don't have the exact
date, as I was trying to track down the NFS-related deadlock at the time and
in that process updated my whole source pool a few times)

I was running a 'make install' at the time the panic happened, but didn't
note if the process exiting was something invoked by make or whether it
was me exiting my login shell on another VT (I had hoped to get a dump,
but fudge-fingered the DDB command and as a result got neither dump nor
written stack trace, but from memory it was almost exactly what Chris
reports.

--rafal