[U-Boot-Users] Configuration of the PHY address

[Mirco Fuchs <mircofuchs@web.de>]

Hi,
I'm operating on an at91rm9200-based board. The AT91RM9200 is 
connected to an ethernet PHY via MII interface. The current 
implementations of the MII-access functions in ether.c assume, that 
the phy's address is 00000. But the address of the PHY that I use 
(KS8721) is 00001 by default, so i cannot access the MII-registers 
with the this implementation. Apart from that a user could configure 
the address of the PHY on his board with pull-ups. I think we need to 
set the PHY address explicite when accessing it.

I suggest to introduce a new macro (e. g. CFG_PHY_ADDRESS) that can 
be set by the user in the board specific header file.

There are two ways to implement the usage of this macro.
The first is that we use the macro directly in the functions 
at91rm9200_EmacWritePhy() and at91rm9200_EmacReadPhy() without 
changing any parameters. The advantage is that we can implement it 
very easy and very fast. The disadvantage is that we cannot change 
the address after compilation (which shouldn't be a problem in most 
cases). 

The second way is to add a new parameter (e. g. phyadd) to the 
EmacWrite and EmacRead function. The PHY driver would be responsible 
to set the PHY address when calling these functions. The PHY address 
could be taken from CFG_PHY_ADDRESS (like above) or could be set 
dynamically (e. g. if we have more than one PHY connected). In 
addition to the EmacRead and EmacWrite interface itself, we need to 
change all drivers using this function and implement the new 
interface in this case.

Which of the suggested ways would you prefer? 


Best Regards
Mirco