Re: [PATCH 1/3] dt-binding: document QCOM platforms for CTCU device

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

On 2/3/26 10:44 AM, Suzuki K Poulose wrote:
> On 03/02/2026 09:36, Konrad Dybcio wrote:
>> On 2/3/26 10:31 AM, Suzuki K Poulose wrote:
>>> On 03/02/2026 09:00, Jie Gan wrote:
>>>>
>>>>
>>>> On 2/3/2026 4:50 PM, Konrad Dybcio wrote:
>>>>> On 2/3/26 9:08 AM, Jie Gan wrote:
>>>>>> Document the platforms that fallback to using the qcom,sa8775p-ctcu
>>>>>> compatible for probing.
>>>>>>
>>>>>> Signed-off-by: Jie Gan <jie.gan@oss.qualcomm.com>
>>>>>> ---
>>>>>>    Documentation/devicetree/bindings/arm/qcom,coresight-ctcu.yaml | 4 ++++
>>>>>>    1 file changed, 4 insertions(+)
>>>>>>
>>>>>> diff --git a/Documentation/devicetree/bindings/arm/qcom,coresight- ctcu.yaml b/Documentation/devicetree/bindings/arm/qcom,coresight- ctcu.yaml
>>>>>> index e002f87361ad..68853db52bef 100644
>>>>>> --- a/Documentation/devicetree/bindings/arm/qcom,coresight-ctcu.yaml
>>>>>> +++ b/Documentation/devicetree/bindings/arm/qcom,coresight-ctcu.yaml
>>>>>> @@ -29,6 +29,10 @@ properties:
>>>>>>        oneOf:
>>>>>>          - items:
>>>>>>              - enum:
>>>>>> +              - qcom,glymur-ctcu
>>>>>> +              - qcom,hamoa-ctcu
>>>>>> +              - qcom,kaanapali-ctcu
>>>>>> +              - qcom,pakala-ctcu
>>>>>
>>>>> Platforms with existing numeric compatibles should continue to use them,
>>>>> so that the mess is somewhat containable
>>>>
>>>> Sure Konrad. So for Pakala, I will change it back to qcom,sm8750-ctcu
>>>
>>> Why do we need different compatibles for the others ? Are they not all compliant to the CTCU programming model ? i.e., sa8775p-ctcu ? or even,
>>> a generic,
>>>
>>> qcom,coresight-ctcu
>>
>> It's a huge anti-pattern with the DT maintainers, since a compatible is
>> the only way to effectively differentiate different implementations (i.e.
>> instances on different SoCs) of an IP block
> 
> Do you mean, same IP block integrated to different SoC ? Or are they
> different implementations altogether ? Why are these not applicable for
> other components ? (e.g., Tnoc, I-Tnoc, TPDA, TPDM etc ?)

The former.

I think coresight bindings are fully generic, since they have been in
place for about as long as the arm64 port, which precedes actual bindings
validation (yaml dt-bindings) and in those times the requirements were way
more lax

Konrad

>> This is important for the case where a DTB is shipped as part of firmware
>> and can not be replaced - if some quirk needs to be applied retroactively,
>> we can look for "qcom,glymur-ctcu" without affecting all the 50 other'
>> users of the effectively-identical IP block
> 
> Fair enough, thank for the explanation.
> 
> Kind regards
> Suzuki
> 
>>
>> In this case, we're already reducing the impact on the driver, as that
>> only looks for the single fallback compatible (qcom,sa8775p-ctcu)
>>
>> Konrad
>