On Mon, May 11, 2020 at 03:04:57PM +0200, Andrew Lunn wrote: > > NXP's LX2160ARDB platform currently has the following MDIO-PHY connection. > > > > MDIO-1 ==> one 40G PHY, two 1G PHYs(C45), two 10G PHYs(C22) > > MDIO-2 ==> one 25G PHY > > It has been suggested that ACPI only support a one to one > mapping. Each MAC has one MDIO bus, with one PHY on it. KISS. > > This clearly does not work for your hardware. So not only do we need > to solve how PHY properties are described, we also need an equivalent > of phy-handle, so a MAC can indicate which PHY it is connected to. > Right. I had introduced fwnode_get_phy_node() to take care of phy-handle. > So in effect, you seem to be heading towards a pretty full > reproduction of the DT binding. Before you go too much further and > waste too much of your time, you might want confirmation from the ACPI > people this is not too advanced for what ACPI can do and they tell you > to forget ACPI for this hardware and stick with DT. I've copied the patchset to Rafael and acpi ML. Waiting for their response. Thanks Calvin
