[tulip-bug] tulip not correctly identifying LNE100TX

Donald Becker becker@scyld.com
Tue Jul 9 15:02:01 2002

On Tue, 9 Jul 2002, Dan Hollis wrote:
> On Tue, 9 Jul 2002, Donald Becker wrote:
> > > Is there any way to tell the driver, that it is indeed an LNE100TX and not an
> > > ADMtek Centaur-P?
> > Uhhmmm, it's a ADMtek Centaur-P chip.  There is likely a PCI subsystem
> > ID programmed into the EEPROM that identifies the board as having a
> > Linksys label, but that has no impact on operation.
> 00:04:5A is listed to linksys, it should be possible to identify as 
> LNE100TX if required.

Using the station address to identify boards is not a good idea.
Linksys has sold Ethernet adapters based on many different chips.  The
previous LNE100TX v2 and v3 boards were based on the PNIC and PNIC-2.

The only place the driver uses the station address is very old 21140 boards
without transceiver/media description tables.  It was very painful to
develop those tables, and the approach only works because vendors stopped
making new boards without media tables.

> IIRC 00:04:5A is only on newer LNE100TX's, older 
> model LNE100TX's with Centaur-P use 00:20:78.

... making my point about station addresses not identifying boards.

> It may be possible to use the MAC to specifically identify other cards 
> that have unique tweaking requirements.

The ADMtek Centaur is an all-in-one chip.  There is no way for a board
designer to screw it up, so there no reason to identify the specific

If the board subsystem ID is unique, it's easy enough to add a new table
entry to match only that ID.  The only reason to do this is for a
friendly detection message, not to fix a bug.  Thus I won't add a table
entry unless I'm very certain of the information.

Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993