Re: [PATCH 2/2] dt-bindings: PCI: mediatek-gen3: Add support for controlling power and reset

[Jian Yang (杨戬)]

Dear Rob,

Thanks for your comment.

On Fri, 2023-02-03 at 10:07 -0600, Rob Herring wrote:
> On Tue, Jan 10, 2023 at 9:28 PM Jian Yang <jian.yang@xxxxxxxxxxxx>
> wrote:
> > 
> > From: "jian.yang" <jian.yang@xxxxxxxxxxxx>
> > 
> > Add new properties to support control power supplies and reset pin
> > of
> > a downstream component.
> > 
> > Signed-off-by: jian.yang <jian.yang@xxxxxxxxxxxx>
> > ---
> >  .../bindings/pci/mediatek-pcie-gen3.yaml      | 23
> > +++++++++++++++++++
> >  1 file changed, 23 insertions(+)
> > 
> > diff --git a/Documentation/devicetree/bindings/pci/mediatek-pcie-
> > gen3.yaml b/Documentation/devicetree/bindings/pci/mediatek-pcie-
> > gen3.yaml
> > index 7e8c7a2a5f9b..46149cc63989 100644
> > --- a/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml
> > +++ b/Documentation/devicetree/bindings/pci/mediatek-pcie-gen3.yaml
> > @@ -84,6 +84,29 @@ properties:
> >      items:
> >        enum: [ phy, mac ]
> > 
> > +  pcie1v8-supply:
> > +    description:
> > +      The regulator phandle that provides 1.8V power to downstream
> > component.
> > +
> > +  pcie3v3-supply:
> > +    description:
> > +      The regulator phandle that provides 3.3V power to downstream
> > component.
> > +
> > +  pcie12v-supply:
> > +    description:
> > +      The regulator phandle that provides 12V power to downstream
> > component.
> 
> While in some bindings we've allowed these in the host bridge node,
> that is a mistake. These should be in the root port node. You
> probably
> don't have one in DT, so add one.

In Mediatek's PCIe structure, there is only one root port under a host
bridge. I am wondering if you think it's necessary to add a root port
node in our host bridge node based on that structure?

And I'm a bit confused about how to declare a property which should be
added in a root port node. I would be grateful if you could provide me
some good example about it.

> > +
> > +  dsc-reset-gpios:
> > +    description:
> > +      The reset GPIO of a downstream component.
> > +    maxItems: 1
> > +
> > +  dsc-reset-msleep:
> 
> Doesn't the PCI spec define this time? We're talking about PERST#,
> right?

The "dsc-reset-gpios" represents an extra reset pin other than PERST#
required by a downstream component. I tried to add a property here so
that users can control the delay time between the assertion and
deassertion according to their requirement, but Krzysztof mentioned
that there is an ongoing discussion about a "GPIO delay" driver can
handle this. So I will remove the "dsc-reset-msleep" in V2 patch.

> > +    description:
> > +      The delay time between assertion and de-assertion of a
> > downstream
> > +      component's reset GPIO.
> > +    maxItems: 1
> > +
> >    clocks:
> >      minItems: 4
> >      maxItems: 6
> > --
> > 2.18.0
> > 

Best regards,
Jian Yang