[U-Boot-Users] [PATCH] m68k fix for Ethernet Buffers at address 0

[Wilson Callan]

Dear Wolfgang,

> First, a formal issue: you missed to include the Signed-off-by: line
> which is mandatory for each patch.
>
Apologies.  I didnt know i should "sign off" my own patches, but now  
i see that is what i'm supposed to do.


> This is a problem with the Linux driver then, and should be fixed in
> Linux. Please see the FAQ entry about this, and then let's get rid of
> this eth_init call in U-Boot. Philosophy here is that devices shall be
> only initialized when really used by U-Boot.
that makes sense.  in this case its u-boot that knows the Ethernet  
Address since its stored in its environment.  Possibly the solution  
is to create a routine in fec.c (the fast ethernet driver) which only  
sets the MAC in the coldfire, instead of starting up the ethernet  
port before the buffers are ready.  This change would keep things  
working as they are, but fix the bug where packets are received  
before the buffer address is set.

should i go ahead with this fix, or instead remove the eth_init()  
call and risk breaking all the m68k systems?

thanks, wilson