Re: [PATCH net-next v4 14/17] dt-bindings: net: pse-pd: Add bindings for PD692x0 PSE controller

[Rob Herring]

On Sat, Feb 17, 2024 at 01:14:29PM +0100, Oleksij Rempel wrote:
> On Thu, Feb 15, 2024 at 05:02:55PM +0100, Kory Maincent wrote:
> > Add the PD692x0 I2C Power Sourcing Equipment controller device tree
> > bindings documentation.
> > 
> > This patch is sponsored by Dent Project <dentproject@xxxxxxxxxxxxxxxxxxx>.
> > 
> > Signed-off-by: Kory Maincent <kory.maincent@xxxxxxxxxxx>
> > ---
> ...
> > +        pse_pis {
> > +          #address-cells = <1>;
> > +          #size-cells = <0>;
> > +
> > +          pse_pi0: pse_pi@0 {
> > +            reg = <0>;
> > +            #pse-cells = <0>;
> > +            pairset-names = "alternative-a", "alternative-b";
> > +            pairsets = <&phys0>, <&phys1>;
> > +          };
> > +          pse_pi1: pse_pi@1 {
> > +            reg = <1>;
> > +            #pse-cells = <0>;
> > +            pairset-names = "alternative-a";
> > +            pairsets = <&phys2>;
> 
> According to latest discussions, PSE PI nodes will need some
> additional, board specific, information:
> - this controller do not implements polarity switching, we need to know
>   what polarity is implemented on this board. The 802.3 spec provide not
>   really consistent names for polarity configurations:
>   - Alternative A MDI-X
>   - Alternative A MDI
>   - Alternative B X
>   - Alternative B S
>   The board may implement one of polarity configurations per alternative
>   or have additional helpers to switch them without using PSE
>   controller.
>   Even if specification explicitly say:
>   "The PD shall be implemented to be insensitive to the polarity of the power
>    supply and shall be able to operate per the PD Mode A column and the PD
>    Mode B column in Table 33–13"
>   it is possible to find reports like this:
>   https://community.ui.com/questions/M5-cant-take-reversed-power-polarity-/d834d9a8-579d-4f08-80b1-623806cc5070
> 
>   Probably this kind of property is a good fit:
>   polarity-supported = "MDI-X", "MDI", "X", "S";

Where does that live? Looks like a property of the consumers defined in 
the provider. Generally, that's not the right way for DT. I'll say it 
again, I think you should be expanding #pse-cells (>1), not getting rid 
of them (==0).

Rob