Well, I feel rather sheepish about this. A few hours after sending
off the cry for help, it occured to me that it would be good to
cycle power on the sick machine. That fixed the problem.
Anyway, I got several educational replies from:
eeikhey@eeiua.ericsson.se (Kevin Heagney)
ups!uniq.com.au!kevin@warrane.connect.com.au (Glenn Satchell)
pas@medicine.wustl.edu (Paul Schoening)
stern@sunne.East.Sun.COM (Hal Stern)
david@camscan.co.uk (David Sansom)
poffen@sj.ate.slb.com (Russ Poffenberger)
djh@igor.rational.com
derekt@pwcs.uucp
sms!roy!cadence.ccas.msk.su!ganesh@sequent.kiae.su (R. Ganesh)
They all agreed that it had something to do with flaky hardware.
------- begin brief quotes from replies ------------------------
SOLUTION SUMMARY : The problem is that 4mb simms are mixed with 1mb
simms.
Did you upgrade the ss1+ with any simm modules?
If so pls check the following
1.4mb and 1mb memory simms shouldn't be mixed in one bank.
2.4mb simms should be in bank 0, if there is a mix of 1 & 4mb simms in
the
system.
I seem to recall that being an old 4.1.1/4.1 bug - one thing to try for
sure is to upgrade to 4.1.3, or wait a week and do 4.1.3u1 to see if it
make a difference.
known bug -- fixed in 4.1.2
We had the same problem on an IPC. Removed a Pioneer CD changer
from the SCSI chain and all went well. Sun documentation wasn't
too clear on what the error meant. Hope you work it out.
a watchdog reset occurs when the processor gets stuck, usually due to a
duff bit of hardware. maybe you ethernet chip(s) are on the fritz.
(if
you didn't know, le0 is the ethernet :-)
At that point, it is probably trying to probe the video card. Check
that
it is properly seated in the SBUS slot. If so, it is probably bad.
1. Take out the ethernet card (if possible) and try to boot without
it.
2. If that's not possible, look at the output of the "dmesg" command
from another Sparc 1, or some other sparc to see what the next device
is that it polls. Maybe it's not the ethernet but the one that comes
after that's causing the problem.
could it be an anaemic power supply?
----------- end brief quotes ------------------------------------
Glenn Satchell actually explained some about what's happening:
> Generally, the window underflow error is produced when the Current
Window
> pointer of the SPARC IU rotates past it's starting point or a
specific window
> which has been defined (usually by the kernel) in the window mask
register.
>
> What probably happened is that your machine got a double-parity hit
which it
> couldn't recover from or some other nasty error which produced the
watchdog
> reset. The window underflow occured when the rest of the watchdog
reset
> routine tried to finish but ran into difficulties.
>
> My advice is to take the following errors after the first error with
a grain
> of salt. Usually, it doesn't mean a whole lot since the first error
was the
> problem that caused things to crash.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:19 CDT