Re: [PATCH 1/4] dt-bindings: net: Add Realtek MDIO controller

[Chris Packham]

Hi Andrew,

On 14/12/2024 00:58, Andrew Lunn wrote:
> On Fri, Dec 13, 2024 at 01:56:52PM +1300, Chris Packham wrote:
> > Hi Andrew,
> >
> > On 13/12/2024 06:13, Andrew Lunn wrote:
> > > > +      realtek,smi-address:
> > > > +        $ref: /schemas/types.yaml#/definitions/uint32-array
> > > > +        description: SMI interface and address for the connected PHY
> > > > +        items:
> > > > +          - description: SMI interface number associated with the port.
> > > > +          - description: SMI address of the PHY for the port.
> > > Is the hardware really random here with its mapping of PHYs?
> > > Generally, hardware is pretty predictable, logical. I'm just wondering
> > > if this property is really required, or if it can be replaced with
> > > some logic, avoiding typ0s.
> > The hardware is pretty flexible in this respect. You can pretty arbitrarily
> > arrange your PHYs. Not all ports support the same kind of SERDES interface
> > but as far as the MDIO interface goes you can arrange things however you
> > like.
> Is there a reason to not just statically define it, maybe with a table
> in the driver? Does a DT developer actually need this flexibility?

It's a part of the board design so I think it's something that should go 
into the device tree. For example I've got 3 different boards right now 
that use different arrangements of MDIO connections to the PHYs mostly 
for ease of board layout (and one because it has different PHYs).