[vortex-bug] 3c556 TX bug ???

Bogdan Costescu bogdan.costescu@iwr.uni-heidelberg.de
Wed Apr 17 13:53:01 2002

On Tue, 16 Apr 2002, Felix Lindner wrote:

> The later (Tx status register 90), I see pretty often with this system,
> but I never experienced the issue with no sending before. 

That's quite curious... The way I see things happening is the following:

The card is not fed fast enough with packets for whatever reason, so it's 
complaining by raising the interrupt with this error status. In the 
meantime no packets are transmitted, so they are queued. When the error 
condition appears often, the probability of a Tx timeout (Tx queue being 
full past some time limit) increases. So any bulk data transfer or "ping 
-f" might be able to trigger this situation.

My knowledge about mobile cards is very limited, so maybe Donald could 
share his view on this issue. I do have one ideea though 8-):
if the transfer across the bus is slow, would an increase in PCI latency 
for this device also increase chances that the PCI burst with a full 
packet would finish before the underrun occurs ?

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