Re: [PATCH v1 1/2] dt-bindings: arm: coresight: Update the pattern of ete node name

[James Clark]

On 26/12/2023 09:36, Krzysztof Kozlowski wrote:
> On 26/12/2023 02:50, Jinlong Mao wrote:
>>
>>
>> On 12/21/2023 4:44 PM, Krzysztof Kozlowski wrote:
>>> On 21/12/2023 09:36, Jinlong Mao wrote:
>>>>
>>>>
>>>> On 12/21/2023 4:17 PM, Krzysztof Kozlowski wrote:
>>>>> On 21/12/2023 09:15, Jinlong Mao wrote:
>>>>>>
>>>>>>
>>>>>> On 12/21/2023 4:12 PM, Krzysztof Kozlowski wrote:
>>>>>>> On 21/12/2023 04:28, Jinlong Mao wrote:
>>>>>>>>>> diff --git a/Documentation/devicetree/bindings/arm/arm,embedded-trace-extension.yaml b/Documentation/devicetree/bindings/arm/arm,embedded-trace-extension.yaml
>>>>>>>>>> index f725e6940993..cbf583d34029 100644
>>>>>>>>>> --- a/Documentation/devicetree/bindings/arm/arm,embedded-trace-extension.yaml
>>>>>>>>>> +++ b/Documentation/devicetree/bindings/arm/arm,embedded-trace-extension.yaml
>>>>>>>>>> @@ -23,7 +23,7 @@ description: |
>>>>>>>>>>      
>>>>>>>>>>      properties:
>>>>>>>>>>        $nodename:
>>>>>>>>>> -    pattern: "^ete([0-9a-f]+)$"
>>>>>>>>>> +    pattern: "^ete-([0-9a-f]+)$"
>>>>>>>>>
>>>>>>>>> My concerns are not resolved. Why is it here in the first place?
>>>>>>>>
>>>>>>>> Hi Krzysztof,
>>>>>>>>
>>>>>>>> ETE is acronym of embedded trace extension. The number of the name is
>>>>>>>> the same as the number of the CPU it belongs to.
>>>>>>>
>>>>>>> This is obvious and was not my question.
> 
> You already said it here...
> 
>>>>>>
>>>>>> Do you mean why the pattern match of the node name is added here ?
>>>>>
>>>>> Yes, especially that it is requiring a non-generic name.
>>>>>
>>>>>>
>>>>>> This node should not have the node name match, right ?
>>>>>
>>>>> Usually. For sure shouldn't be for non-generic names.
>>>>>
>>>> Hi Suzuki,
>>>>
>>>> Can we remove the pattern match of the node name and use a generic name
>>>> "ete" for the ete DT nodes ?
>>>
>>> "ete" is not a generic name. What is generic here? It's an acronym of
>>> some specific device name.
>>>
>>
>> The device full name is embedded trace extension. So use ETE as the name 
>> here.
> 
> That's obvious and my comment was not about it. Second time... This is
> my unlucky day... I said, why do you even want to enforce name which is
> not generic, since the names should be generic?
> 

I think we can just drop the enforced name if it's getting in the way.
It doesn't really do anything and other Coresight bindings don't have it
anyway.

> I assume you read the DT specification:
> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
> 
> 
> Best regards,
> Krzysztof
> 

I couldn't find anything in that list that would be a good fit for a
name, and it seems like all of the Coresight devices have already been
added with non generic names (like funnel and replicator etc), so it
might be a bit late now.

But if we drop the enforced name then it's probably fine.

James