Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site brl-tgr.ARPA
Path: utzoo!linus!philabs!cmcl2!seismo!brl-tgr!tgr!cak@PURDUE.ARPA
From: cak@PURDUE.ARPA (Christopher A Kent)
Newsgroups: net.unix-wizards
Subject: Re: Need a DEQNA driver
Message-ID: <10733@brl-tgr.ARPA>
Date: Thu, 16-May-85 10:53:32 EDT
Article-I.D.: brl-tgr.10733
Posted: Thu May 16 10:53:32 1985
Date-Received: Fri, 17-May-85 03:28:49 EDT
Sender: news@brl-tgr.ARPA
Lines: 23

The DEQNA is a real can of worms ... even DEC admits it sucks. We don't
have a Unix driver, but I have a "skeletal" standalone driver for the
pdp11 that I used to figure out how to make the damn thing work.
There's one problem with it; in the protocol for dealing with who owns
net buffers, there's a provision for marking a packets "used"; the
manual seems to imply that you make a ring of unused buffers, hand them
to the device, he fills them in, marks them used, interrupts you, and
waits for more. It's not quite true; the device will happily overwrite
a buffer that it has used and that you haven't cleared out yet. My
driver/program does NOT handle this; I had finished it before we
discovered this, and never got interested enough to fix it (especially
since the "right" way to fix it isn't at all clear.) 

For what it's worth, the Ultrix driver ignores the overwriting problem,
too.

I'll happily send my program out to anyone that wants a copy; I'm
putting together a note summarizing the horror stories that we've had
trying to make this thing talk to the world.

Cheers,
chris
----------