Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site umcp-cs.UUCP
Path: utzoo!linus!philabs!seismo!rlgvax!cvl!umcp-cs!chris
From: chris@umcp-cs.UUCP
Newsgroups: net.unix-wizards
Subject: Re: 4.1bsd vs. Versatec V80
Message-ID: <2763@umcp-cs.UUCP>
Date: Mon, 26-Sep-83 23:09:55 EDT
Article-I.D.: umcp-cs.2763
Posted: Mon Sep 26 23:09:55 1983
Date-Received: Wed, 28-Sep-83 23:18:30 EDT
References: <181@menlo70.UUCP>
Organization: Univ. of Maryland, Computer Science Dept.
Lines: 36

Ok, software solutions are pretty easy.  The main thing to remember
is that your V8 (11" Versatec) looks EXACTLY like the Varian, as
far as user-level code is concerned.  The easiest thing to do is hack
all the programs (vpr, vpd, etc) to think it's talking to the
Varian, but use the Versatec device (/dev/vp0 instead of /dev/va0).
(What I did for the eneevax versatec was a bit messier:  instead
of just changing device names, I changed everything to know that
the Versatec was 11" instead of 36", but that shouldn't be important.)

A REAL quick&dirty fix would be to make /dev/va0 be the Versatec.

Has anyone rewritten that mess of Versatec/Varian spooler?

DMA:  Well, we have the Versatec interface card, but the 4.1BSD
doesn't use DMA on it, just character-at-a-time.  If you have a
DMA interface I'd like a copy!  Concerning hanging:  William L.
Sebok (did I get your name right?), astrovax!wls, has posted fixes
to some of the VM management code for dealing with hung DMA transfers.
I'd like to see a "final" set of fixes myself....  Does anyone know
if 4.2 has all the VM bugs fixed?

Configuration:  Rather than cut up the Versatec card, we had our
LP11 interface moved and just plugged the thing right in.  Of
course, that left us with separate print and plot interrupts.  So,
I just changed the probe routine to subtract (or was it add?) 4 to
the value in cvec, and changed the config file to list a separate
"read" and "write" interrupt (both calling vpintr of course).
Poof, everything worked, and we didn't have to cut a single trace.
Of course, this way the board uses up some extra DMA vectors,
but....

Chris
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci
UUCP:	{seismo,allegra,brl-bmd}!umcp-cs!chris
CSNet:	chris@umcp-cs		ARPA:	chris.umcp-cs@UDel-Relay