RE: VirtualPC tulip bug, more info

On Fri, 30 Jan 2004, Schneider, Andrew wrote:
> I've also got this one:
> eth0: Too much work during and interrupt, csr5=0xfc6900c4

Which is indicating that a Tx-complete interrupt has also occured.

> Right now RX_RING_SIZE is set to 128 and max_interrupt_work is 75. I'm going
> to try bumping max_interrupt_work to 100. Would increasing RX_RING_SIZE help
> here?

Those are already big numbers.

If you are using my tulip driver release, the max_interrupt_work unit
is Rx packets, not loops through the interrupt handler (or Tx packets
scavenged or errors handled).  This is unlike most other drivers which do
count loops.

This means you should increase max_interrupt work much more than with
other drivers.  You can do that through either the module parameter or
using the ioctl(SIOCSPARAMS) call.  I think that mii-diag can do this
with
   mii-diag eth0 -G ,,1000
The values mean
   mii-diag eth0 -G <msg_level>,<multicast_filter_limit>,<max_interrupt_work>,<tx_copybreak>

If you were on a physical machine, I would suggest leaving RX_RING_SIZE
at its original value.  In this case, a virtual hardware emulator,
increasing it may help throughput by minimizing context switches.


-- 
Donald Becker				[email protected]
Scyld Computing Corporation		http://www.scyld.com
914 Bay Ridge Road, Suite 220		Scyld Beowulf cluster systems
Annapolis MD 21403			410-990-9993

_______________________________________________
tulip mailing list, [email protected]
To change to digest mode or unsubscribe visit
http://www.scyld.com/mailman/listinfo/tulip