ARM: dts: sunxi: Add mdio bus sub-node to GMAC

The DWMAC binding never supported having the Ethernet PHY node as a
direct child to the controller, nor did it support the "phy" property
as a way to specify which Ethernet PHY to use. What seemed to work
was simply the implementation ignoring the "phy" property and instead
probing all addresses on the MDIO bus and using the first available
one.

The recent switch from "phy" to "phy-handle" breaks the assumptions
of the implementation, and does not match what the binding requires.
The binding requires that if an MDIO bus is described, it shall be
a sub-node with the "snps,dwmac-mdio" compatible string.

Add a device node for the MDIO bus, and move the Ethernet PHY node
under it. Also fix up the #address-cells and #size-cells properties
where needed.

Fixes: de332de26d19 ("ARM: dts: sunxi: Switch from phy to phy-handle")
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>

1 files changed, 6 insertions, 4 deletions

diff --git a/arch/arm/boot/dts/sun7i-a20-olinuxino-lime2.dts b/arch/arm/boot/dts/sun7i-a20-olinuxino-lime2.dts
index 0fe657e062a7..9ba62774e89a 100644
--- a/arch/arm/boot/dts/sun7i-a20-olinuxino-lime2.dts
+++ b/arch/arm/boot/dts/sun7i-a20-olinuxino-lime2.dts
@@ -114,10 +114,6 @@
 	phy-handle = <&phy1>;
 	phy-mode = "rgmii";
 	status = "okay";
-
-	phy1: ethernet-phy@1 {
-		reg = <1>;
-	};
 };
 
 &hdmi {
@@ -154,6 +150,12 @@
 	vref-supply = <&reg_vcc3v0>;
 };
 
+&gmac_mdio {
+	phy1: ethernet-phy@1 {
+		reg = <1>;
+	};
+};
+
 &mmc0 {
 	vmmc-supply = <&reg_vcc3v3>;
 	bus-width = <4>;