Re: [PATCH v4 1/3] dt-bindings: PCI: mediatek-gen3: Support mt8195

[Rob Herring]

On Mon, Oct 31, 2022 at 02:28:54PM +0100, AngeloGioacchino Del Regno wrote:
> Il 28/10/22 11:43, Tinghan Shen ha scritto:
> > From: Jianjun Wang <jianjun.wang@xxxxxxxxxxxx>
> > 
> > In order to support mt8195 pcie node, update the yaml to support new
> > properties of iommu and power-domain, and update the reset-names
> > property to allow only one 'mac' name.
> > 
> > Signed-off-by: Jianjun Wang <jianjun.wang@xxxxxxxxxxxx>
> > Signed-off-by: TingHan Shen <tinghan.shen@xxxxxxxxxxxx>
> > ---
> >   .../bindings/pci/mediatek-pcie-gen3.yaml         | 16 +++++++++++++---
> >   1 file changed, 13 insertions(+), 3 deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml b/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml
> > index c00be39af64e..a9013c10131a 100644
> > --- a/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml
> > +++ b/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml
> > @@ -70,14 +70,21 @@ properties:
> >       minItems: 1
> >       maxItems: 8
> > +  iommu-map:
> > +    maxItems: 1
> > +
> > +  iommu-map-mask:
> > +    const: 0
> > +
> >     resets:
> >       minItems: 1
> >       maxItems: 2
> >     reset-names:
> > -    minItems: 1
> > -    items:
> > -      - const: phy
> > +    oneOf:
> > +      - items:
> > +          - const: phy
> > +          - const: mac
> >         - const: mac
> 
> Sorry, this looks a bit messy.
> 
> I can propose two solutions, either:
> 
> reset-names:
>   minItems: 1
>   items:
>     - const: mac
>     - const: phy
> ... and change the order in mt8192.dtsi;

That doesn't work because originally 'phy' alone was allowed.

You could do:

minItems: 1
maxItems: 2
items:
  enum: [ phy, mac ]

But for 2 items, either order would be allowed. Not great, but not a 
huge deal.

> otherwise, something like
> allOf:
>   - if:
>      properties:
>        compatible:
>          contains:
>            - const: mediatek,mt8195-pcie
>     ... invert reset-names
> 
> I think that the first solution makes a bit more sense though, as the
> PHY reset may be moved in a PCIE-PHY specific driver instead of being
> managed here in pci-mediatek-gen3.

That either makes sense or it doesn't. If it does, then you should go 
ahead and separate the phy rather than change the ABI twice.

Rob