[tulip] Weird problem with Conexant

Donald Becker becker@scyld.com
Thu, 4 Oct 2001 19:23:35 -0400 (EDT)

On Wed, 3 Oct 2001, Felipe Contreras wrote:

> Subject: [tulip] Weird problem with Conexant
> I have a weird problem when conecting via an ethernet cross cable to a
> CNET PRO200, now running windows 2000.
> This is what ifconfig says:
> eth0      Link encap:Ethernet  HWaddr 00:50:8B:AA:F6:C1
>           TX packets:1 errors:3334 dropped:0 overruns:0 carrier:3817
>           collisions:0 txqueuelen:100

This is "normal" for a conexant in full duplex mode.
The Conexant chip reports constant carrier errors when set to FDX.  The
packets are sent correctly, as indicated by the '0' collision count.

> This is an extract of what tulip-diag says:
>  Port selection is MII, full-duplex.
>  Transmit started, Receive started, full-duplex.
>   The Rx process state is 'Waiting for packets'.
>   The Tx process state is 'Idle'.
>   The transmit threshold is 128.

Please send the whole output, although it doesn't matter here.

> The weird part is that I can actually use the driver as if it was
> working fine except that the windows machine keeps saying that the
> conection goes down, an then goes up, and that happends more often when
> I'm comsuming cpu power.

Hmmm, I haven't seen any problem with the implementation on the Compaq
1700 series.

> Also sometimes I get good pings, but most of
> the times I got them corrupted. This also happened with my old Davicom
> card, it was just coincide that my new CNET card use the same chipset
> now I wonder if it was really broken or was a driver problem.

Check your cable.
The early Davicom chips had a significant problem with generating and
detecting bad packets.  But I suspect a mispaired or otherwise bad cable.

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