Re: [PATCH v6 2/2] dt-bindings: mfd: mediatek: mt6397: Convert to DT schema format

[Krzysztof Kozlowski <krzk@kernel.org>]

On 27/09/2024 12:41, Macpaul Lin wrote:
> 
> 
> On 9/27/24 17:45, Krzysztof Kozlowski wrote:
>> 	
>>
>> External email : Please do not click links or open attachments until you 
>> have verified the sender or the content.
>>
>> On 18/09/2024 08:49, Macpaul Lin wrote:
>>> Convert the mfd: mediatek: mt6397 binding to DT schema format.
>>>
>>> MT6323, MT6358, and MT6397 are PMIC devices with multiple function
>>> subdevices. They share a common PMIC design but have variations in
>>> subdevice combinations.
>>>
>>> Key updates in this conversion:
>>>
>>> 1. RTC:
>>>    - Convert rtc-mt6397.txt and merge into parent MT6397 PMIC DT schema.
>>>
>>> 2. Regulators:
>>>    - Align to generic name "regulators".
>>>    - Update references from .txt to .yaml for mt6323, mt6358, and mt6397
>>>      regulators.
>>>    - Simplify regulator name labels in device tree examples.
>>>
>>> 3. Audio Codec:
>>>    - Convert sound/mt6358.txt and merge into parent MT6397 PMIC DT schema.
>>>    - Align to generic name "audio-codec" for codec and sound subdevices.
>>>    - Add "mediatek,dmic-mode" and "Avdd-supply" properties.
>>>
>>> 4. Clocks:
>>>    - Align to generic name "clocks" for clockbuffer subdevices.
>>>
>>> 5. LEDs:
>>>    - Convert leds-mt6323.txt and merge into parent MT6397 PMIC DT schema.
>>>    - Update LED binding.
>>>
>>> 6. Keys:
>>>    - Add detailed descriptions for power and home keys.
>>>    - Add compatible: mediatek,mt6358-keys.
>>>
>>> 7. Power Controller:
>>>    - Convert mt6323-poweroff.txt and merge into parent MT6397 PMIC DT
>>>      schema.
>>>    - Add #power-domain-cells property to fix dt-binding check error.
>>>    - Clarify "BBPU" as "Baseband power up".
>>>
>>> 8. Pinctrl:
>>>    - Align to generic name "pinctrl" instead of "pin-controller".
>>>
>>> 9. Compatible:
>>>    - Drop "mediatek,mt6357" since there is a separated DT Schema
>>>      for PMIC MT6357.
>>>
>>> 10. Examples:
>>>    - MT6323: Retain complete examples for this PMIC.
>>>    - MT6358 and MT6397: simplify settings in regulators.
>>>     - Preserve "audio-codec", "clocks", "pinctrl", "rtc", and "keys"
>>>       sections as they contain typical settings for different PMICs.
>>>
>>> Additional updates:
>>> - MAINTAINERS: Add co-maintainers and reference to
>>>   mfd/mediatek,mt6397.yaml for LED and power-controller drivers.
>>> - input/mediatek,pmic-keys.yaml: Update reference to
>>>   mfd/mediatek,mt6397.yaml.
>>>
>>> Signed-off-by: Sen Chu <sen.chu@mediatek.com>
>>> Signed-off-by: Macpaul Lin <macpaul.lin@mediatek.com>
>>
>> NAK
>>
>> Follow up patch pointed out this is incorrect. I don't understand this
>> concept of sending knowingly incorrect code, so please sort it out at v7.
>>
>> Best regards,
>> Krzysztof
>>
> 
> Sorry I didn't know mt6359-codec and mt6358-sound are similar device
> until Sept 26 (just yesterday) when I was reviewing the DT
> bindings. Until I've found the previous DT Schema of mt6359-codec 
> incorrect.
> See [1] 
> https://lore.kernel.org/linux-arm-kernel/20240926092519.6556-3-macpaul.lin@mediatek.com/T/
> and [2] 
> https://lore.kernel.org/all/20240926092519.6556-2-macpaul.lin@mediatek.com/
> 
> And.. I don't understand, the origin this patch v6 should be a text file
> conversion, shouldn't we keep the origin content just a conversion and
> trying not to broken other relate modules?

The conversion should be correct. Correct means each compatible has a
schema ending with additional/unevaluatedProperties: false.

If you need to change this binding to "true", it means above statement
was not fulfilled. You did not write proper conversion.

... or I do not understand what you wanted to achieve with the other patch.

> 
> Please help to indicates what is a better approach here:
> 
> Should I merge the properties of mt6359-codec, mt6358-sound, and 
> mt6397-codec
> together and put them all into to mt6359.yaml and redo the converting of 
> mfd/medaitek,mt6397.yaml?

Not sure, I don't know. Are these different devices? How are they
related? I expect you to know this, not me.

The problem is that you send a patch having a finished schema for child
node and then - in next patch in different patchset (!!!) - immediately
change it to false.

So the first code, which was not even merged, is incorrect. This is
heavily misleading.

> 
> That will be much simpler for the audio-codec node for mediatek,mt6397.yaml.
> However there will be a little mismatch from the origin text file.
> 
> Thanks
> Macpaul Lin

Best regards,
Krzysztof