[vortex] 3c595 'Transmit error, Tx status register 90' revisited

[Andrew Morton]

Koos van den Hout wrote:
> 
> After I ran into the aforementioned error again (bigtime) I decided to
> google for other reports of this error. I have mentioned it to this list
> before, we never got anything better out of it then 'yes, this is a
> problem'.

hmm...

So basically your're getting a storm of Tx underrun errors,
and resetting the transmitter in the Tx interrupt routine is
causing some sort of problem?  I don't understand *why* it
causes a problem - the transmitter has been stopped.

Deferring the reset until start_xmit in this way will
mean that any additional queued frames will sit in
memory for an arbitrary amount of time - they won't
be kicked off until higher-level code tries another transmit.
Normally not a problem, but it could cause long delays in
some workloads.

The core question is: why are you getting Tx underrun errors?
Any ideas?