[PATCH 1/5 v3] powerpc: DTS file for the C2K

[David Gibson]

On Tue, May 20, 2008 at 09:38:08AM -0700, Remi Machet wrote:
> On Tue, 2008-05-20 at 11:13 +1000, David Gibson wrote:
> > On Mon, May 19, 2008 at 05:00:23PM -0700, Remi Machet wrote:
> > > Support for the C2K cPCI Single Board Computer from GEFanuc
> > > (PowerPC MPC7448 with a Marvell MV64460 chipset)
> > > All features of the board are not supported yet, but the board
> > > boots, flash works, all Ethernet ports are working and PCI 
> > > devices are all found (USB and SATA on PCI1 do not work yet).
> > > 
> > > Part 1 of 5: DTS file describing the board peripherals. As far as I know
> > > all peripherals except the FPGA are listed in there (I did not included
> > > the FPGA because a lot of work is needed there).
> > 
> > [snip]
> > > +		ethernet-group at 2000 {
> > > +			#address-cells = <1>;
> > > +			#size-cells = <0>;
> > > +			compatible = "marvell,mv64360-eth-group";
> > > +			reg = <0x2000 0x2000>;
> > > +			PHY0: ethernet-phy at 0 {
> > > +				device_type = "ethernet-phy";
> > > +				interrupts = <76>;	/* GPP 12 */
> > > +				interrupt-parent = <&PIC>;
> > > +				reg = <0>;
> > > +			};
> > > +			ethernet at 0 {
> > > +				device_type = "network";
> > > +				compatible = "marvell,mv64360-eth";
> > > +				reg = <0>;
> > > +				interrupts = <32>;
> > > +				interrupt-parent = <&PIC>;
> > > +				phy = <&PHY0>;
> > > +				local-mac-address = [ 00 00 00 00 00 00 ];
> > > +			};
> > 
> > Not sure if this is new, or I just didn't notice it before.  Also not
> > sure if this is your mistake, or a mistake in the binding for the
> > marvell device here.
> > 
> > The address in "reg" should be unique within the bus the device
> > appears on.  So having both the PHY and the MAC devices with the same
> > "reg" and same unit address here is bad.
> > 
> > Having both the PHYs and the MACs as children of the ethernet-group
> > node is possible, but the addresses must be encoded to distinguish
> > them (e.g. MACs are 0x0, 0x1, 0x2, PHYS are 0x1000 0x1001 0x1002).
> > 
> It was an attempt at getting rid of the mdio group since the PHY is
> effectively managed by the Ethernet hardware. The problem is that the
> reg field must contain those values ... if I change this, I must change
> the driver and other boards DTS files that depend on it which I would
> rather not do at the same time I am pushing this patch through.

Right.  Getting rid of the mdio node isn't an inherently bad idea, but
it constitutes a change in the binding, so as you've noticed it would
need the drivers to change, and can't just be done for one board
unilaterally.

> I will put back the mdio group and add a reg field to it as you
> suggested previously.

Ok.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson