Re: [PATCH] dt-bindings: display: renesas,du: Provide bindings for r8a779a0

[Kieran Bingham]

Hi Laurent,

On 01/09/2021 23:13, Laurent Pinchart wrote:
> Hi Kieran,
> 
> On Wed, Sep 01, 2021 at 11:01:11PM +0100, Kieran Bingham wrote:
>> On 23/06/2021 13:53, Geert Uytterhoeven wrote:
>>> On Wed, Jun 23, 2021 at 1:11 AM Kieran Bingham wrote:
>>>> From: Kieran Bingham <kieran.bingham+renesas@xxxxxxxxxxxxxxxx>
>>>>
>>>> Extend the Renesas DU display bindings to support the r8a779a0 V3U.
>>>>
>>>> Signed-off-by: Kieran Bingham <kieran.bingham+renesas@xxxxxxxxxxxxxxxx>
>>>
>>> Thanks for your patch!
>>>
>>>> --- a/Documentation/devicetree/bindings/display/renesas,du.yaml
>>>> +++ b/Documentation/devicetree/bindings/display/renesas,du.yaml
>>>> @@ -39,6 +39,7 @@ properties:
>>>>        - renesas,du-r8a77980 # for R-Car V3H compatible DU
>>>>        - renesas,du-r8a77990 # for R-Car E3 compatible DU
>>>>        - renesas,du-r8a77995 # for R-Car D3 compatible DU
>>>> +      - renesas,du-r8a779a0 # for R-Car V3U compatible DU
>>>>
>>>>    reg:
>>>>      maxItems: 1
>>>> @@ -774,6 +775,57 @@ allOf:
>>>>          - reset-names
>>>>          - renesas,vsps
>>>>
>>>> +  - if:
>>>> +      properties:
>>>> +        compatible:
>>>> +          contains:
>>>> +            enum:
>>>> +              - renesas,du-r8a779a0
>>>> +    then:
>>>> +      properties:
>>>> +        clocks:
>>>> +          items:
>>>> +            - description: Functional clock for DU0
>>>> +            - description: Functional clock for DU1
>>>> +
>>>> +        clock-names:
>>>> +          items:
>>>> +            - const: du.0
>>>> +            - const: du.1
>>>
>>> The hardware block has only a single function clock for both channels,
>>> like on R-Car H1.
>>
>> Indeed, but I believe both channels still need to set them, if they can
>> be operated independently, the driver looks up the clock based on the
>> du.%d, and so for DU1, it is simply expressed as the same clock in DT.
>>
>> Is this acceptable? or is there further issues there?
> 
> Could we handle that on the driver side, like we do for H1 by not
> setting RCAR_DU_FEATURE_CRTC_IRQ_CLOCK ? We would probably need to split
> that flag in two, as there are two interrupts.

Ok, that's not so bad to split, Done.


> It's a bit annoying not knowing what the MSTP bits do exactly, we've
> modelled them as gates for the functional clock, but maybe in cases like
> this one the mapping isn't fully correct, I'm not sure.
> 
>>> And what about DU_DOTCLKIN?
>>
>> This thread has already discussed this with Laurent, and I concur -
>> There doesn't appear to be any relevant reference to DU_DOTCLKIN on the
>> DU side.
>