OK, the Rev C LANCE has a bug that can be worked around in software by what
I would describe as a high cost check (more instructions in the networking
data path to get packets in the door are a bad thing to have), and we can't
tell whether we have the Rev C chip or not at config time.

This is a hardware bug. Has anyone looked into what a later (correct and
pin compatible, one hopes) revision of the LANCE chip would cost from AMD,
so that we can simply replace the buggered chips in those Sun 3's that have
the Rev C? Call it a NetBSD Engineering Change Order (ECO) or Field Change
Order (FCO). Pity that we probably can't get AMD to replace these chips
under warantee.

While we're at it, we could ask Motorola for pricing on 68882's to replace
the 68881's in our Suns for a little zippier FPU performance, as suggested
in:

http://www.unix-ag.uni-kl.de/~hardware/Sun3/sun3main.html

hack hardware? me? Naaah,

Erik Fair