Re: [PATCH v1 1/3] dt-bindings: i3c: Add Qualcomm I3C master controller bindings

[Rob Herring <robh@kernel.org>]

On Mon, Feb 10, 2025 at 09:42:03PM +0530, Mukesh Kumar Savaliya wrote:
> Thanks Krzysztof !
> 
> On 2/9/2025 5:15 PM, Krzysztof Kozlowski wrote:
> > On 06/02/2025 14:43, Mukesh Kumar Savaliya wrote:
> > > Hi Krzysztof,  Thanks !
> > > 
> > > On 2/5/2025 8:12 PM, Krzysztof Kozlowski wrote:
> > > > On 05/02/2025 15:31, Mukesh Kumar Savaliya wrote:
> > > > > Add device tree bindings for the Qualcomm I3C master controller. This
> > > > > includes the necessary documentation and properties required to describe
> > > > > the hardware in the device tree.
> > > > 
> > > > A nit, subject: drop second/last, redundant "bindings". The
> > > > "dt-bindings" prefix is already stating that these are bindings.
> > > Sure
> > > > See also:
> > > > https://elixir.bootlin.com/linux/v6.7-rc8/source/Documentation/devicetree/bindings/submitting-patches.rst#L18
> > > > 
> > > > Use modern terminology, which means:
> > > > s/master/whatever else or even nothing/
> > > > See other recent bindings and discussions.
> > > > 
> > > Sure
> > > > 
> > > > > 
> > > > > Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com>
> > > > > ---
> > > > >    .../bindings/i3c/qcom,i3c-master.yaml         | 57 +++++++++++++++++++
> > > > >    1 file changed, 57 insertions(+)
> > > > >    create mode 100644 Documentation/devicetree/bindings/i3c/qcom,i3c-master.yaml
> > > > > 
> > > > > diff --git a/Documentation/devicetree/bindings/i3c/qcom,i3c-master.yaml b/Documentation/devicetree/bindings/i3c/qcom,i3c-master.yaml
> > > > > new file mode 100644
> > > > > index 000000000000..ad63ea779fd6
> > > > > --- /dev/null
> > > > > +++ b/Documentation/devicetree/bindings/i3c/qcom,i3c-master.yaml
> > > > 
> > > > Filename matching compatible.
> > > > 
> > > Changed compatible to "qcom,i3c-master"
> > > > > @@ -0,0 +1,57 @@
> > > > > +# SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause
> > > > > +%YAML 1.2
> > > > > +---
> > > > > +$id: http://devicetree.org/schemas/i3c/qcom,i3c-master.yaml#
> > > > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > > > +
> > > > > +title: Qualcomm I3C master controller
> > > > > +
> > > > > +maintainers:
> > > > > +  - Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com>
> > > > > +
> > > > > +allOf:
> > > > > +  - $ref: i3c.yaml#
> > > > > +
> > > > > +properties:
> > > > > +  compatible:
> > > > > +    const: qcom,geni-i3c
> > > > 
> > > > No SoC? So to be sure: you claim all future SoCs will be using exactly
> > > > the same interface. No new compatibles, no new properties will be added.
> > > > 
> > > I think i should remove const. kept it for now as no other compatible to
> > > be added as of now.
> > > 
> > > let me remove const.
> > 
> > No, it does not matter. Keep const.
> > 
> Sure. I reviewed other files and seems i should write as below. Please help
> confirm.
> 
>   compatible:
>     items:
>       - enum:
>           - qcom,sm8550-i3c-master
>       - const: qcom,i3c-master

No, that's even worse. I doubt there is some universal, never changing 
QCom I3C master.

> > > 
> > > SoC name is not required, as this compatible is generic to all the SOCs.
> > 
> > That's the statement you make. I accept it. I will bookmark this thread
> > and use it whenever you try to add any future property here (to be
> > clear: you agree you will not add new properties to fulfill *FUTURE* SoC
> > differences).
> > 
> Sorry, i am not saying there won't be any other compatible but i was saying
> base driver will use "qcom,i3c-master".
> After checking other files i realized there can be const compatible but
> other SOC specific can be added as enum.  Hope above given way is fine.

AIUI, "geni" is some firmware based multi-protocol serial i/o controller 
and we already have other "geni" bindings. So really, it's probably more 
coupled to firmware versions than SoC versions. If we haven't had 
problems with per SoC quirks with the other geni bindings, then I think 
using the same "geni" here is fine. But we won't be happy if we start 
seeing per SoC quirk properties.

Rob