> On Thu, Jan 17 2002 at 12:57:34 +0100, Xavier HUMBERT wrote:
> > dmesg says about the card :
> > 
> > > ex0 at pci1 dev 4 function 0: 3Com 3c905C-TX 10/100 Ethernet with mngmt
> > >(rev. 0x78)
> > > ex0: interrupting at irq 5
> > > ex0: MAC address 00:04:75:4c:19:11
> > > OUI 0x00085a model 0x0000 rev 0 at ex0 phy 24 not configured
> You neither have the correct phy nor ukphy configured in your kernel.
> 
> Most 3com ex NICs use the exphy but this thread
> http://mail-index.netbsd.org/current-users/2002/01/03/0015.html
> is about a 3c905C with a bmtphy
> 
> I'd suggest adding all phys to see a) what matches and b) if it works
> better with phy.

I'd suspect that there is a mismatch between the full duplex states of
the transceiver and the MAC unit.  Unfortunately the standard phy register
set (MII regs 0-7 are defined in 802.3) doesn't include the actual rate the
phy has negotiated with its peer, nor is there a standard method of
determining when renegotiation has happened.

With an unknown PHY, or a MAC engine that won't poll a PHY register, it
is probably safer to stop the FDX modes being negotiated (my writing to
the ANAR register).

(Bet this doesn't help...)

    David