Re: [PATCH v5 2/2] dt-bindings: rtc: add max313xx RTCs

[Krzysztof Kozlowski]

On 04/04/2023 14:29, Alexandre Belloni wrote:
> On 04/04/2023 14:26:18+0200, Alexandre Belloni wrote:
>> On 04/04/2023 10:35:53+0000, Tilki, Ibrahim wrote:
>>>>>>>> +  aux-voltage-chargeable:
>>>>>>>> +    enum: [0, 1, 2]
>>>>>>>> +    description: |
>>>>>>>> +      Enables trickle charger.
>>>>>>>> +      0: Charger is disabled (default)
>>>>>>>> +      1: Charger is enabled
>>>>>>>> +      2: Charger is enabled with a diode
>>>>>>>
>>>>>>> 2 is not an allowed value. I asked to drop this property. It is coming
>>>>>>> from rtc.yaml. I also do not understand "with a diode". So otherwise it
>>>>>>> is charging with, I don't know, FET?
>>>>>>
>>>>>> No, what is not explained here (and maybe not unsterstood by the
>>>>>> submitter) is that the RTC has an extra diode so, charging will always
>>>>>> enable a diode, select a resistor and then have or not an extra diode.
>>>>>> Figure2 of the MAX31329 datasheet is great.
>>>>>>
>>>>>
>>>>> That is exactly why I had "adi,trickle-diode-enable" property in previous patch.
>>>>> So if I can't have "adi,trickle-diode-enable" and can't add an additional value
>>>>> to "aux-voltage-chargeable", I am not sure how to add support for the extra
>>>>> diode at this point.
>>>>
>>>> Ask the person who asked you to remove adi,trickle-diode-enable...
>>>
>>> That was the purpose.
>>>
>>
>> If the earlier submission was clearer my answer would have been
>> different but note how I had to dig up the datasheet to understand there
>> were two diodes. All the trickle chargers have a schottky diode so
>> "adi,trickle-diode-enable" nor the commit log were explicit about the
>> second diode (which is a regular diode).
>>
>> aux-voltage-chargeable is enabling a diode on all the existing RTC
>> drivers so instead of trying to make me look like the bad guy you should
>> rather thank for taking the time trying to get better DT bindings.
>>
> 
> BTW, Krzysztof, you should focus more on how v5 of the driver is still
> abusing the #clock-cells property to do pinmuxing after I repeatedly
> explain to not do that.

Uh, I assumed #clock-cells in the binding is for its purpose - clocks. I
am rarely verifying driver implementation.

Best regards,
Krzysztof