Re: [PATCH v2 1/1] dt-bindings: nvmem: mediatek: Convert mtk-efuse binding to YAML

[Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>]

On 26/04/2022 12:02, allen-kh.cheng wrote:
> Hi Krzysztof,
> 
> On Tue, 2022-04-26 at 08:31 +0200, Krzysztof Kozlowski wrote:
>> On 26/04/2022 08:23, allen-kh.cheng wrote:
>>>>> +properties:
>>>>> +  compatible:
>>>>> +    oneOf:
>>>>> +      - enum:
>>>>> +          - mediatek,mt8173-efuse
>>>>> +          - mediatek,efuse
>>>>
>>>> Still no changes...
>>>>
>>>
>>> I just want to confirm again.
>>>
>>> "Generic compatibles should not be used standalone"
>>>
>>> It seems we should remove mediatek,efuse and keep "mediatek,mt8173-
>>> efuse"in binding. have I got that right?
>>
>> You should comment for which chipsets this compatible is and add a
>> deprecated:true. In such case it cannot be part of enum but separate
>> item in this oneOf.
>>
>>
>> Best regards,
>> Krzysztof
> 
> Thanks for your suggestions, I would plan to send PATCHs as below,
> 
> We have a PATCH 01 for current accepted dts
> 
> properties:
>   compatible:
>     oneOf:
>       - enum:
>         - mediatek,efuse
> 
>         - mediatek,mt8173-efuse
>         description: Only mt8173-efuse
> with generic fallback should be used
>       - items:
>           - enum:
>     
> - mediatek,mt7622-efuse
> 			  ...
>           - const: mediatek,efuse
> 
> Then add PATCH 02 to deprecate it
> 
> properties:
>   compatible:
>     oneOf:
>       - enum:
>         - mediatek,efuse
>         - mediatek,mt8173-efuse
>         deprecated: true
>         description: The mediatek,efuse is a generic fallback for other
> Chipset. Do not use the single compatible such as mediatek,efuse
> or mediatek,mt8173-efuse. It is deprecated.
>       - items:
>           - enum:
>               - mediatek,mt7622-efuse
>               ...
>           - const: mediatek,efuse
> 
> 
> PATCH 03 for 8173
> 
> update mt8173.dtsi 
> 
> change compatible from "mediatek,mt8173-efuse" to "mediatek,mt8173-
> efuse", "mediatek,efuse";
> 
> 
> Do you think it'd be okay ?

The idea is correct, but as I said it cannot be part of enum, but
separate item in oneOf. You should see an error when testing your patch.


Best regards,
Krzysztof