Re: [PATCH 1/2] dt-bindings: iio: adc: rtq6056: add support for the whole RTQ6056 family

[Krzysztof Kozlowski]

On 28/12/2023 08:58, ChiYuan Huang wrote:
> On Thu, Dec 28, 2023 at 08:09:35AM +0100, Krzysztof Kozlowski wrote:
>> On 28/12/2023 04:19, ChiYuan Huang wrote:
>>> On Tue, Dec 26, 2023 at 01:12:50PM +0100, Krzysztof Kozlowski wrote:
>>>> On 26/12/2023 12:19, ChiYuan Huang wrote:
>>>>> On Tue, Dec 26, 2023 at 10:18:47AM +0100, Krzysztof Kozlowski wrote:
>>>>>> On 26/12/2023 04:47, cy_huang@xxxxxxxxxxx wrote:
>>>>>>> From: ChiYuan Huang <cy_huang@xxxxxxxxxxx>
>>>>>>>
>>>>>>> Add compatible support for RTQ6053 and RTQ6059.
>>>>>>>
>>>>>>> Signed-off-by: ChiYuan Huang <cy_huang@xxxxxxxxxxx>
>>>>>>> ---
>>>>>>>  .../devicetree/bindings/iio/adc/richtek,rtq6056.yaml         | 5 ++++-
>>>>>>>  1 file changed, 4 insertions(+), 1 deletion(-)
>>>>>>>
>>>>>>> diff --git a/Documentation/devicetree/bindings/iio/adc/richtek,rtq6056.yaml b/Documentation/devicetree/bindings/iio/adc/richtek,rtq6056.yaml
>>>>>>> index 88e008629ea8..d1e1f36d1972 100644
>>>>>>> --- a/Documentation/devicetree/bindings/iio/adc/richtek,rtq6056.yaml
>>>>>>> +++ b/Documentation/devicetree/bindings/iio/adc/richtek,rtq6056.yaml
>>>>>>> @@ -25,7 +25,10 @@ description: |
>>>>>>>  
>>>>>>>  properties:
>>>>>>>    compatible:
>>>>>>> -    const: richtek,rtq6056
>>>>>>> +    enum:
>>>>>>> +      - richtek,rtq6053
>>>>>>> +      - richtek,rtq6056
>>>>>>
>>>>>> Aren't these devices compatible? Your driver change says they are, so
>>>>>> express compatibility with list here (and oneOf).
>>>>>>
>>>>> Thanks, I try to take other binding as the reference. One more question.
>>>>> If rtq6053 is compatible with rtq6056, there's only chip package type difference.
>>>>> Do I need to seperate it into a dedicated enum element?
>>>>> Or just put it into one item and said this part number is compatible with rtq6056? 
>>>>
>>>> See example-schema. You need enum and items, both in oneOf:.
>>>>
>>> After reading the 'example-schema', I Still cannot understand what the special case items
>>> means.
>>
>> What is "special case items"?
>>
> I may misunderstand something. The special case is the 'fallback' that you mentaioned in
> the last.
>>>
>>> But in my case, is the below change correct?
>>> [Diff]
>>> properties:
>>>    compatible:
>>> -    enum:
>>> -      - richtek,rtq6053
>>> -      - richtek,rtq6056
>>> -      - richtek,rtq6059
>>> +    oneOf:
>>> +      - items:
>>> +          - enum:
>>> +              - richtek,rtq6053
>>> +              - richtek,rtq6056
>>> +              - richtek,rtq6059
>>
>> This changes nothing, you still have just one item. The example-schema
>> has exactly that case, so why you are coding it differently?
>>
>> Anyway, test your DTS with the fallback, you will see that above does
>> not work.
>>
> I rewrite the below one and tested. it seems correct.
> 
> [Diff]
>    compatible:
> -    enum:
> -      - richtek,rtq6053
> -      - richtek,rtq6056
> -      - richtek,rtq6059
> +    oneOf:
> +      - enum:
> +          - richtek,rtq6053
> +          - richtek,rtq6059
> +      - items:
> +          - const: richtek,rtq6056

You still need two items here to express compatibility. What is
compatible with what? It must be rtq6053 compatible with rtq6056,
because you cannot break the ABI, can you?

> 
> Just one more question. If rtq6053 is fully compatibie with rtq6056, does it need to be put
> into oneOf enum or be just put into items enum and use the 'fallback' mechanism?

The fallback is just a term. The point is to have a list of two
compatibles. See: Devicetree specification, writing-bindings and
numerous presentations about writing DTS.

> 
> If so, the 'richtek,rtq6053' in of_device_id match can be removed. Though it just remove
> one line, but less-changed would be better.

The device using fallback should be removed the driver of_device_id, so
6053 goes away.


Best regards,
Krzysztof