Subject: More on first port MVII
To: None <port-vax@netbsd.org>
From: None <rick@snowhite.cis.uoguelph.ca>
List: port-vax
Date: 12/21/2000 14:17:09
By any chance is this in the stock distribution tape masters? I will
have to unroll them and have a lookiesee.
The MVII port should be in the sources for the stock 4.3BSD Vax distribution,
although Berkeley did not have the resources to cut TK50s.
Most of my problems, so far have centered around my SQ706A scsi controller.
It must deviate in some critical places from what the RQDXx things that
were original on the MVII's did. I can get most of the kernel to load
and go, but when it polls the mscp disks, it shells out to a reboot
on that SQ706A controller. Close, but not quite.
There was a long standing bug in many of the variants of uda.c (the MSCP
disk driver) where it didn't conform to the spec. during initialization.
(Since it worked for the DEC hardware, most people didn't care, but I
did run into it when I tried to use a third party MSCP disk controller.)
If I recall correctly, the fix was a 100usec delay followed by a while loop
waiting for the Sn bit to be set instead of expecting the Sn bit to be set.
(The 100usec delay is specified in the manual, which I just cracked open for
the first time in ages and wasn't done in the driver, if I recall.) Sn are
bits 11-14 of the SA register and refer to "start initialization n".
Unfortunately that's all I can
remember about that one (which sounds like what you are getting bit by:-).
Ohhh, for sure, for sure, I would love to have that tape. May I
also pass the bits along to the Unix Historical Society (TUHS)/PDP-11
Unix Preservation Society (PUPS) archives that Warren Toomey is running
for SCO, down under in Australia? That would make a nice addition
to the archives. Write me a paragraph of any info/history you have
about that or its development, and I will pass that too along to Warren.
We all got to keep the mystical and magickal Unix bits preserved...(:+}}.
I'll toss it in the mail. It is just 4.3BSD Reno stuck on a TK50, so it
shouldn't be any different than the generic distribution (except media),
so I don't think it's historically significant, but do with it as you
please.