On Fri, Jun 09, 2000 at 12:38:41AM +0100, Alan Cox wrote:
> > I'm not sure exactly what we are trying to test with here, but i have a
> > real world machine serving cgi's and graphics which has died 4 out of 5
> > days in a row with a variety of eepro100 drivers. All are as new or newer
> > than the one included in 2.2.14. Given that, I think there IS a problem.
> > Any idea on solutions?(other than freeBSD)
> Ok what errors do you see logged when it fails and what revision of the
> board do you have. Also what is it plugged into and does it have an errors
> logged when it fails
The points important for diagnostic are:
- Do you see messages "can't fill rx buffer"?
These messages mean that the driver experience allocation failures because
of high load creating high memory pressure.
- Do you see messages "card reports no RX buffers (no resources)"?
These messages mean that the card realize receive buffer shortage, too.
- When the driver manages to allocate buffers again, it issues message
"restart the receiver after a possible hang". It should appear after the
The things may be complicated by the fact that the receiver hang may be
followed by an unreasonable transmitter hang.
A few months ago I performed stress tests and the driver worked well.
I'm going to repeat the tests this weekend.
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to email@example.com
This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:40 EST