Re: [PATCH] dt-bindings: mfd: Update pattern property case

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

On 06/02/2024 12:11, Krzysztof Kozlowski wrote:
> On 06/02/2024 12:07, Naresh Solanki wrote:
>> Hi Krzysztof,
>>
>>
>> On Tue, 6 Feb 2024 at 13:06, Krzysztof Kozlowski
>> <krzysztof.kozlowski@linaro.org> wrote:
>>>
>>> On 05/02/2024 15:00, Naresh Solanki wrote:
>>>> Hi Krzysztof,
>>>>
>>>>
>>>> On Mon, 5 Feb 2024 at 18:38, Krzysztof Kozlowski
>>>> <krzysztof.kozlowski@linaro.org> wrote:
>>>>>
>>>>> On 05/02/2024 12:02, Naresh Solanki wrote:
>>>>>> Driver expects regulator child node in upper case.
>>>>>> Hence align with the same.
>>>>>
>>>>> Did the driver have DT support before? I think no, so why aligning that
>>>>> way? I would argue that driver should be aligned with bindings, the
>>>>> moment you add DT for the first time.
>>>> Yes the driver has DT support already.
>>>> This patch is to align with driver:
>>>> https://github.com/torvalds/linux/blob/master/drivers/regulator/max5970-regulator.c#L381
>>>
>>> That's not support for DT, but just naming regulators.
>> I'm not sure if I get your perspective right & please correct me if I'm wrong,
>> I'm just trying to adjust the dt binding of max5970 because found
>> errors related to
>> regulator child node SW0/SW1.
>> Regulator driver expects to have regulators child nodes as SW0/1(upper case)
>> But dt binding expects it to be lower case.
>> Thus there is misalignment & due to which, when running CHECK_DTBS for my
>> mainboard DTS, I see some errors.
> 
> I understood that, no need to clarify, it is kind of obvious. I still
> claim the same: When the binding was added? Mid 2023. When the driver
> was added? Much earlier. So someone posted driver bypassing DT
> documentation and review and now you claim we need to accept the driver
> choice. Do you think this is correct process?
> 
> If so, what stops people from sending all the driver changes without DT
> and documenting whatever they had post-factum?

I would like to add that OF probing for MFD driver was added the same
time as bindings, so that's the time when OF was added. Therefore you
cannot use excuse "there is ABI", because your commits added both OF
support and non-matching documentation.

My stance is: fix the driver, not the binding.

Best regards,
Krzysztof