Scientific Operations Bellum Gratia Artis

testing - 1998 - July - this message

[ previous , next ]

Subject: Re: [9fans] 100base ether card

From: Eric Dorman <edorman@[REDACTED]>
Date: Fri, 3 Jul 1998 12:28:13 -0700

On Thu, Jul 02, 1998 at 09:09:13AM +0000, Kenji Arisawa wrote:
> Hello 9fans!
> Does anyone use 100-base ether cards?
> I would like to buy the cards for my plan9 system.
> Please advise me.
> Kenji Arisawa
> E-mail:

I've done some of the work in getting a dec 21140-based
card to work, a Linksys LNE100TX (sadly NLA) though the
BayNetworks Netgear FA310TX is also based on this chip.
The driver is based on a brazil driver kindly supplied by jmk,
which is supposed to work with with some other dec 21140-based
cards as well.

The driver compiles cleanly now and I'm going through and
checking out all the functions. I hope to have it working
relatively soon. If you'd like to play with it I can forward
the code to you.

One problem I've yet to decide how to solve is how to
have the fs and everything else use precisely the same
driver for ether. While the differences are relatively
minor, the decision of which target to compile for tends
to be decided using #ifdefs. I feel this hides structural
weakness (fs uses different buffering structure from
everything else) and is just plain hard to read.. a more
complete solution would be to rewrite the interface layer
on the fs (which I have to do anyway to support multiple
interfaces in the fs) to use the same structure as the
other codes, then just use the usual set of ether drivers.
Damn sight more work but I think more rewarding in the
long run :)

I've been delayed on this work since my auth server died
an unnatural death and now I have to replace it :<


Eric Dorman