Re: [PATCH 1/2] media: dt-bindings: Remove assigned-clock-* from various schema

[Sakari Ailus]

Hi Krzysztof,

On Mon, Oct 14, 2024 at 12:34:51PM +0200, Krzysztof Kozlowski wrote:
> On 14/10/2024 12:20, Sakari Ailus wrote:
> > Hi Krzysztof,
> > 
> > On Mon, Oct 14, 2024 at 09:43:07AM +0200, Krzysztof Kozlowski wrote:
> >> On Sat, Oct 12, 2024 at 04:02:50PM +0100, Bryan O'Donoghue wrote:
> >>> diff --git a/Documentation/devicetree/bindings/media/i2c/hynix,hi846.yaml b/Documentation/devicetree/bindings/media/i2c/hynix,hi846.yaml
> >>> index 60f19e1152b33128cf3baa15b8c70a874ca6d52e..d18ead8f7fc43bfacc291aed85b5ca9166c46edb 100644
> >>> --- a/Documentation/devicetree/bindings/media/i2c/hynix,hi846.yaml
> >>> +++ b/Documentation/devicetree/bindings/media/i2c/hynix,hi846.yaml
> >>> @@ -28,12 +28,6 @@ properties:
> >>>      items:
> >>>        - description: Reference to the mclk clock.
> >>>  
> >>> -  assigned-clocks:
> >>> -    maxItems: 1
> >>> -
> >>> -  assigned-clock-rates:
> >>> -    maxItems: 1
> >>> -
> >>>    reset-gpios:
> >>>      description: Reference to the GPIO connected to the RESETB pin. Active low.
> >>>      maxItems: 1
> >>> @@ -82,8 +76,6 @@ required:
> >>>    - compatible
> >>>    - reg
> >>>    - clocks
> >>> -  - assigned-clocks
> >>> -  - assigned-clock-rates
> >>
> >> That's not extraneous, but has a meaning that without assigned-clocks
> >> this device or driver will not operate.
> >>
> >> File should rather stay as is.
> > 
> > ...
> > 
> >>> diff --git a/Documentation/devicetree/bindings/media/i2c/sony,imx258.yaml b/Documentation/devicetree/bindings/media/i2c/sony,imx258.yaml
> >>> index c978abc0cdb35cfe2b85069946cf1be435a58cb8..f0f9726a2add89492b8c56e17ed607841baa3a0d 100644
> >>> --- a/Documentation/devicetree/bindings/media/i2c/sony,imx258.yaml
> >>> +++ b/Documentation/devicetree/bindings/media/i2c/sony,imx258.yaml
> >>> @@ -24,10 +24,6 @@ properties:
> >>>        - sony,imx258
> >>>        - sony,imx258-pdaf
> >>>  
> >>> -  assigned-clocks: true
> >>> -  assigned-clock-parents: true
> >>> -  assigned-clock-rates: true
> >>> -
> >>
> >> This is ok.
> > 
> > Basically the clock related requirements for these devices are the same:
> > they all need a clock configured at a board specific frequency. Shouldn't
> > we treat them the same way?
> 
> I don't know these devices, but binding did not express such
> requirement, so according to current binding the properties are 100%
> redundant.

It sounds like to me we should have both assigned-clocks and
assigned-clock-rates on all of them. I recall from the earlier discussion
setting the rate could have been a problem but based on the schema it looks
like providing zero for the rate ot indicate not setting it is possible.

Maybe we could have "csi2-sensor" and "parallel-sensor" bindings that would
require, among other things, just this.

-- 
Kind regards,

Sakari Ailus