Request for Information

Donald Becker becker@cesdis1.gsfc.nasa.gov
Tue Sep 22 13:06:42 1998


On Mon, 21 Sep 1998, Jon wrote:

> I have used Linux now for quite some time, and have a good deal
> of networking experience as well.  However, I have a question which
> I believe only you can answer.  First, the DEC cards are rumoured to be
> discontinued.  Second, they are being replaced by a Macronix or a Lite-On.
> In addition, the Winbond 840 is DEC-like.
> 
> What I would like to know is what you would recommend in the 10/100 line
> in terms of chipset, either one of the above or perhaps others.

The PNIC (from LiteOn) is a not-quite-clone that's working OK.  There are
already three versions, and the middle generation boards from LinkSys
currently do not work.  Two of those boards arrived today, so that problem
will likely be fixed soon.  (Thanks to Greg LaPolla at LinkSys for sending
samples all three versions.  Apparently the middle version required
searching for working returns.)

The MXIC (Macronix) chips are closer clones, but they have had problems with
duplicated packets (!?).  There is a suggested fix -- a initialization
change in an configuration register -- from Macronix that's in the next
driver version.

The Winbond 840 is Tulip-like, but in a strange way.  The first twelve
registers -- the common Tulip register set -- are mostly similar, but they
are densely packed in I/O space.  That immediately rules out code sharing
with the Tulip driver.  The station address and multicast filter are stored
in the the four media control registers, instead of being loaded by a
configuration frame.  The register-based scheme is easier to use, but there
is no 16 element perfect filter.

The reason I don't recommend the '840 is that my board corrupts packets
longer than about 1100 bytes.  I've heard that there is a bug in descriptors
longer than 1024, but even having a packet span two descriptor doesn't avoid
the problem.


Donald Becker					  becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center,  Greenbelt, MD.  20771
301-286-0882	     http://cesdis.gsfc.nasa.gov/people/becker/whoiam.html