On Fri, Jun 18, 1999 at 10:58:44PM +0000, Brian Buhrow wrote:
> And on another 905b we get the same probe, but  the mii_readreg never
> returns anyting other than 0, even when the phy value is 24, which appears
> to be a magic constant in /usr/src/sys/dev/ic/elinkxlvar.h

Have you tried the failing cards under other OSs for a longer period of
time? It looks like a hardware problem; you may just have a couple of
bad cards.

> 1.  What's magic about the phy value of 24 on an Elinkxl card with an
> internal phy?

The internal phy will always advertise itself as # 24 on the MII bus.
Why was this number picked? Ask 3com :-)

> 
> 2.  Is it possible this card, which is brand new, is one of the new
> 3c905b-M cards which is being mis-identified with a 3c905b-TX?

That would seem highly unlikely. If it is, then the card can be
considered broken, because it's returning a wrong PCI ID.
> 
> 3.  Are there online technical documents talking about this card and the
> phy?

Not online, no.

- Frank