On Tue, May 25, 2021 at 03:16:52PM +0200, Andrew Lunn wrote:
quoted
quoted
It however sounds like you have the two busses one level deeper?
It would be good if you document this as part of the binding.
Yes, it looks like this:
ethernet-switch@2 {
compatible = "nxp,sja1110a";
ethernet-ports {
...
};
mdio {
#address-cells = <1>;
#size-cells = <0>;
mdio@0 {
reg = <0>;
compatible = "nxp,sja1110-base-t1-mdio";
#address-cells = <1>;
#size-cells = <0>;
sw2_port5_base_t1_phy: ethernet-phy@1 {
compatible = "ethernet-phy-ieee802.3-c45";
reg = <0x1>;
};
...
};
We should run this by Rob.
That is probably not the intention of
Documentation/devicetree/bindings/net/mdio.yaml, it works because of
additionalProperties: true
What meaning does reg have, in mdio@0?
None apart from "not the other MDIO bus".
I haven't run this through the DT validator yet, I am doing that right
now and will copy Rob to the next patch series.
Just to be clear, what is your suggestion for this? I am not a great fan
of 2 internal MDIO buses, but as mentioned, the PHY access procedure is
different for the 100base-TX and the 100base-T1 PHYs.