[vortex] 3C905C-TX-M asics

[Bogdan Costescu]

On Wed, 26 Sep 2001, John Kutil wrote:

> Running the same driver on a 3C905C with a Broadcom 40-0579-005 asic
> has problems with upComplete receive interrupts not being
> generated, resulting in overruns.
> ...
> Have the linux developers had basic functionality problems on 3com
> cards with different asic vendors and revision levels?  If if so,
> do you tend to use the safest driver functionality?

The 3c59x driver doesn't make any difference between ASICs. The docs that
I have (the PDF file downloaded from 3Com's site) doesn't document any
difference between ASICs in that area.
The only difference that we've seen recently which might be related is the
introduction of the 905CX (note the X after C) which had a problem with
resetting the Rx part - the time needed for completing this operation is
much larger than for the older chips. However, after we found the
workaround, the cards worked well without any other modification in the
driver (well, there was another one needed in the MII transceiver
detection, but this is unrelated). I have no idea if the Broadcom ASIC
that you mention is the one used on the CX cards.

OTOH, the 3c59x driver can process several uploaded packets at once and I
don't think that a missed upComplete interrupt will be noticed in this
case... Is it possible for you to detect whether the card is setting the
InterruptLatch bit after uploading the packet ? I have no idea how to do
it as you would have to not check this from the interrupt handler for the
card, but maybe somebody else...

Sincerely,

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