eth0: Host Error, FIFO diagnostic register 8000

[Donald Becker]

On Fri, 15 Oct 1999, Jim Dethlefsen wrote:

> The machine is an HP Vectra workstation with the 2.2.6 (Slackware 4.0)
> kernel. The card is a 3C905B (4/11/99 manufac. date)
> 
> It seems that every other reboot (I may be wrong here) the machine loads
> the OS up to about multiuser time and then the console gets swamped with
> the noted message... Every so often it throws out a "Too many tries,

This message covers many error types, most unusual and card type specific.
In the case of the 3c905B the description is

____
hostError [1]: This bit is set when a catastrophic error related to the bus
interface occurs. The errors which set hostError are: PCI target abort and
PCI master abort. hostError is cleared by issuing GlobalReset with the
upDownReset mask bit cleared.
____

Rereading the v0.99L driver code, I think that it's nominally doing the
correct reset to recover from a PCI bus error.  But since I don't have a
system that generates PCI errors I can't test the recovery code.

It would be useful to try another type of *PCI bus master* ethercard and see
if it also reports a PCI bus error.

Donald Becker
Scyld Computing Corporation, and
USRA-CESDIS,   becker@cesdis.gsfc.nasa.gov