I've been trying to get my KA650 up to -current, with little luck.
It's running a snapshot of -current from the beginning of June, at
which time I was able to get it stable by compiling without UVM.

Now I've been trying to use the source tar balls from 1998-10-03 to
upgrade (having first done so on my Sparc and i386 systems).  The
system is a simple setup, with MSCP disk, TMSCP tape, and DEQNA
network interface.  The only oddity is the MSCP disk controller, a
CMD CQD-220 SCSI controller.  According to its documentation, some
systems may have trouble with it because it's so much faster than
typical DEC MSCP controllers.  That may be just hype, I don't know.

Trying various kernel configurations, I always end up with the same
type of crash: once the system gets loaded down somewhat (such as
running a "make <anything>" in /usr/src/lib/libc), it will bomb out
with this message:

	panic: access fault without process

Then it reboots, and early in the hardware detection phase goes:

	panic: segv in kernel mode

It then says it's syncing the disk, but immediately gets:

	panic: ptelen fault in system space

The automatic reboot after this will get it running again.  Note that
this is with the KA650.  If I replace the CPU and memory with KA630
and MS630 boards, I can seldom get to multiuser before it crashes...

Incidentally, I've seen the "automatic reboot after crash doesn't get
far, but the next one works" thing before with NetBSD/vax.  Looks like
something doesn't get properly reset at boot.

-tih
-- 
Popularity is the hallmark of mediocrity.  --Niles Crane, "Frasier"