[realtek] on-board rtl8139 fails.

[Donald Becker]

On 7 Dec 2002, Jarl Friis wrote:
> > IRQ 9, the driver is using the same IRQ as reported in /proc/pci.
> So that is as it should be, right?

Yes, if that IRQ works...

> > Do CardBus cards work?
> 
> I have only tried with the FA510 (DECchip) which, as you know does not
> work completely, but I guess that is for another reason. It seems like the
> CardBus system itself does work in general.

I suspect that the Tulip problem is related to this one.

> > > > Try running this while the interface is "up" and check for interrupt
> > > > sources.  My guess is that you will see unhandled interrupts.
...
>   Interrupt sources are pending.
>    Rx Complete indication.

The card is reporting that packets have been received, but the interrupt
handler has not run.

> Meanhwile at the DHCP server I get:
> ec  7 18:13:36 server dhcpd: DHCPDISCOVER from 00:40:ca:bb:ae:ac via eth0
> Dec  7 18:13:36 server dhcpd: DHCPOFFER on 192.168.0.95 to 00:40:ca:bb:ae:ac via
> 
> which indicates that it does transmit OK, it just doesn't
> receive. The ping also gives "Destination Host Unreachable"

You can transmit four packets before the interrupt handler must run to
recover the entries.  So you can get the initial discover out, but you
never see the offer nor could you reply with a request.

I'm not certain of the details of your problem, but it's a IRQ mapping
problem,  not a driver problem.

-- 
Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Scyld Beowulf cluster system
Annapolis MD 21403			410-990-9993