[vortex] Re: VP6 + 3c905 -- any working machines?

[Bogdan Costescu]

On Fri, 23 Feb 2001, Sean Tuling wrote:

> i get the problem in the phase 3 of the slave node boot process.  the
> error i get is
>
> eth0: PCI bus error, bust status 80000020, reset had 1999 tick left.
> eth0: Host error, status 8003, FIFO diagnostic register 0000.
>
> the documentation says that this error is fatal (which it is) and means
> there is some hardware problem - difficult to swallow since the systems
> do not seem to have a problem when running a full installation of linux
> (though they may be running 10Mb).

The difference in running at 10 or 100Mbit might be a big one, as the card
has to transfer between network and main memory ten times the data (when
fully loaded). You said that you have some Cyrix machines, I guess that
they are the all-integrated type. They might simply not be able to cope
with the higher traffic.

The error messages that you get are not very meaningfull: they say that
the card has detected a PCI error and the action that is taken is a full
reset of the card, trying to get over the error; AFAIK this is the only
possible action in this situation.

[ If you (Sean) reply to this message, please keep vortex@scyld.com on the
cc: list. ]

Sincerely,

Bogdan Costescu

IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
E-mail: Bogdan.Costescu@IWR.Uni-Heidelberg.De