*BSD News Article 9195


Return to BSD News archive

Received: by minnie.vk1xwt.ampr.org with NNTP
	id AA5331 ; Wed, 23 Dec 92 13:00:43 EST
Newsgroups: comp.unix.bsd
Path: sserve!manuel.anu.edu.au!munnari.oz.au!spool.mu.edu!agate!linus!alliant!rdr
From: rdr@cl4.alliant.com (Roger Roles)
Subject: Re: WD8003E ethernet card compatibility
Message-ID: <1992Dec22.001805.29535@alliant.com>
Keywords: AKCL Common LISP Kyoto Austin 386BSD
Sender: usenet@alliant.com
Organization: Alliant Computer Systems, Littleton, MA
References: <zxmsd01.724542361@studserv> <1gorbsINN8ku@neuro.usc.edu>
Date: Tue, 22 Dec 1992 00:18:05 GMT
Lines: 32

In article <1gorbsINN8ku@neuro.usc.edu> merlin@neuro.usc.edu (merlin) writes:
>Has anyone managed to get 386BSD to work correctly with the WD8003E ethernet
>card?

I managed (after long debugging sessions :-) to get a WD8003E ethernet card
working. The first step was to configure a kernel without the Isolan driver,
which is configured to the same I/O address in the standard config. The symptom
before this was done was a lot of "reject 32..." (I can't remember the exact
number) messages whenever the interface was configured up.

After getting past that problem, I discovered that most of the transmissions
were being garbled, as evidenced by checksum errors on remote machines which
were receiving the packets, and no replies being received. I am not sure what
the cause of this was, unless the machine (a 486/33) was trying to load data
into the board's buffer too fast, but the symptom was that immediately after
doing a bcopy of the data onto the board, a comparison between the mbuf and
the buffer showed numerous differences. I "fixed" this (a truly gross hack,
probably with significant speed impact, but it works fast enough for my
purposes) by comparing every byte and rewriting it until the correct value
appeared in the board buffer. I limited the attempts to store the correct
value to the board to 10, so as to avoid an infinite loop if the board was
truly unable to receive the correct value. So far, the count of 10 from somewhere in austrailia.  I dont have it online right now but I will
post the location of it as soon as I get some time as I am right in the middle
of finals.  I hope to find time by friday.  The name of the file was xmfm????
something.  I cannot remember exactly.  It is somewhere in .oz.

-- 
Chris
		tjon@plains.nodak.edu	 Christopher C. Tjon
		uunet!plains!tjon (UUCP) B