Alan Cox wrote:
> 
> > Peter pointed out that the contents of the CSR12-14 registers are
> > initialized from the EEPROM, so reading the EEPROM is superfluous--we
> > should just read the CSRs and not read the EEPROM.  I think he has a
> > point, so I'll make that change and submit yet another patch pair.
> 
> I'd rather keep the existing initialisation behaviour of using the eeprom
> for 2.2. There are also some power management cases where I am not sure
> the values are restored on the pcnet/pci.
> 
> For 2.2 conservatism is the key. For 2.4 by all means default to CSR12-14 and
> print a warning if they dont match the eeprom value and we'll see what it
> shows
> 
> > Alan, do you want me to put your inline version in <linux/etherdevice.h>
> > while I'm at it, or what?
> 
> Sure

Ok, here is the 2.4.1 version.  Note that it is slightly different from
the 2.2.18 version.  This one uses the CSRs and complains if the PROM is
different (but uses the CSRs anyway since those can be overridden by a
bootloader, etc.).

I don't have a 2.4.x tree handy that I can compile at the moment, but I
tested this change in a 2.2.x kernel, so it should be ok.

Is this patch satisfactory?

Eli
--------------------.              Rule of Accuracy: When working toward
Eli Carter          |               the solution of a problem, it always 
eli.carter@inet.com `--------------------- helps if you know the answer.