On Tue, Sep 23, 2014 at 04:33:48PM +0200, Arnd Bergmann wrote:

On Tuesday 23 September 2014 16:01:13 Antoine Tenart wrote:

quoted

On Tue, Sep 23, 2014 at 02:38:39PM +0200, Arnd Bergmann wrote:

quoted

On Tuesday 23 September 2014 14:15:30 Antoine Tenart wrote:

quoted

+Optional properties:
+- port-id: should be '0','1' or '2'.
+- phy-addr: MDIO address of the PHY.
+- local-mac-address: see ethernet.txt file in the same directory.
+

I believe new bindings should not use "phy-addr" properties, but
instead use a phy-handle property pointing to a phy device instead.

That's the MDIO address of the PHY. It's a bit difficult to use a phy
device as the driver uses the libphy. And that would break the platform
using it.

Or maybe you have something in mind?

I don't understand what you mean. Doesn't libphy imply using a phy_device?
What I meant was that you should be able to just call the standard
of_phy_connect() and other interfaces to deal with phylib.

Yes sure, I read again my answer and... well I should have had a better
look on the matter.

Anyway, as the proposed binding seems to be OK with you, I'll update the
series accordingly.

Antoine

-- 
Antoine T?nart, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com