Re: [PATCH 1/2] dt-bindings: clock: qcom,sm8450-videocc: Add minItems property

[Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>]

On 6/20/25 7:56 AM, Krzysztof Kozlowski wrote:
> On 19/06/2025 12:20, Jagadeesh Kona wrote:
>>
>>
>> On 6/18/2025 11:56 AM, Krzysztof Kozlowski wrote:
>>> On 17/06/2025 21:07, Jagadeesh Kona wrote:
>>>> Add minItems as 1 for power-domains and required-opps properties
>>>> to allow this binding to be compatible with both single and multiple
>>>> power domains.
>>>
>>> This is your hardware, so you know how it works thus I expect here
>>> arguments why this is correct from the hardware point of view. Without
>>> this, it is impossible to judge whether this is a correct change.
>>>
>>> If I overlook this now, it will be used in discussions by other qcom
>>> engineers, so unfortunately you see, you need to prepare perfect commits
>>> now...
>>>
>>
>> These clk controllers mainly require MMCX power domain to be enabled to access
>> the clock registers. But to configure the cam & video PLLs in probe, an additional
>> MXC power domain also needs to be enabled.
> 
> 
> Then your patch is not correct. Anyway, you should explain the hardware
> in commit msg, why this domain is optional in the hardware.
> 
>>
>> Since the initial DTS changes only added MMCX power domain, this change is required
>> to be backward compatible with older DTS and avoid ABI breakage as discussed in below
>> thread.
> 
> 
> So you send incorrect hardware description allowing something which will
> not work? Or how exactly?

So I think there's a mistake in understanding the backwards compatibility
paradigm here.

There exists a single, objectively correct and represented in hardware,
list of required power-domains and the commit that caused the schema
validation errors was essentially "align YAML with reality" which should
be coupled with an immediate DT update to match and we forget about the
incomplete past

Konrad