Re: [PATCH v4 01/19] dt-bindings: mfd: mediatek: mt6397: Add accdet subnode

[Krzysztof Kozlowski <krzk@kernel.org>]

On 07/03/2025 14:22, Nícolas F. R. A. Prado wrote:
> On Fri, Mar 07, 2025 at 08:11:26AM +0100, Krzysztof Kozlowski wrote:
>> On 06/03/2025 13:19, Nícolas F. R. A. Prado wrote:
>>>>>    It is interfaced to host controller using SPI interface by a proprietary hardware
>>>>>    called PMIC wrapper or pwrap. MT6397/MT6323 PMIC is a child device of pwrap.
>>>>> @@ -224,6 +225,30 @@ properties:
>>>>>      description:
>>>>>        Pin controller
>>>>>  
>>>>> +  accdet:
>>>>> +    type: object
>>>>> +    additionalProperties: false
>>>>> +    description:
>>>>> +      The Accessory Detection module found on the PMIC allows detecting audio
>>>>> +      jack insertion and removal, as well as identifying the type of events
>>>>> +      connected to the jack.
>>>>> +
>>>>> +    properties:
>>>>> +      compatible:
>>>>> +        const: mediatek,mt6359-accdet
>>>>
>>>> You just removed the other file, no folding happened here. Drop the
>>>> accdet node and fold this into parent.
>>>
>>> Sorry, I'm still not sure what you mean by folding here then. Right now the
>>> accdet is a subnode of the PMIC. If you want me to remove the accdet node, where
>>
>> Yes
>>
>>> would its compatible and property go?
>>
>> compatible: nowhere, because it is close to redundancy.
>>
>> property: to the parent pmic node.
>>
>>     pmic {
>>         compatible = "mediatek,mt6359";
>>         interrupt-controller;
>>         #interrupt-cells = <2>;
>>
>>         mediatek,hp-eint-high;
>>     };
> 
> I'm not sure that's right. The ACCDET submodule does have some resources, IRQs,
> that it registers in its mfd cell, see patch 2 of this series [1]. It also has

Binding is supposed to be complete, so why suddenly we have here some
resources which you did not add?

Post complete binding, so you will get proper review.

> its own driver (sound/soc/codecs/mt6359-accdet.c) that probes based on this

Drivers do not define bindings.

> compatible and handles those interrupts. Why would it not get its own node like

Sorry, cannot go. You cannot document binding post factum and claim "I
have a driver which uses that compatible".

This would be a nice way to bypass review.

> the other MFD cells?

I explained why. I gave you the exact reason.

Best regards,
Krzysztof