Re: [net-next,PATCH 2/3] dt-bindings: net: realtek,rtl82xx: Document realtek,ssc-enable property

[Krzysztof Kozlowski <krzk@kernel.org>]

On 03/12/2025 02:30, Marek Vasut wrote:
> On 12/1/25 8:20 AM, Krzysztof Kozlowski wrote:
>> On 30/11/2025 14:41, Marek Vasut wrote:
>>> On 11/30/25 9:20 AM, Krzysztof Kozlowski wrote:
>>>
>>> Hello Krzysztof,
>>>
>>>>> diff --git a/Documentation/devicetree/bindings/net/realtek,rtl82xx.yaml b/Documentation/devicetree/bindings/net/realtek,rtl82xx.yaml
>>>>> index eafcc2f3e3d66..f1bd0095026be 100644
>>>>> --- a/Documentation/devicetree/bindings/net/realtek,rtl82xx.yaml
>>>>> +++ b/Documentation/devicetree/bindings/net/realtek,rtl82xx.yaml
>>>>> @@ -50,6 +50,11 @@ properties:
>>>>>        description:
>>>>>          Disable CLKOUT clock, CLKOUT clock default is enabled after hardware reset.
>>>>>    
>>>>> +  realtek,ssc-enable:
>>>>> +    type: boolean
>>>>> +    description:
>>>>> +      Enable SSC mode, SSC mode default is disabled after hardware reset.
>>>>
>>>> I don't want more SSC properties. We already had a big discussions about
>>>> it - one person pushing vendor property and only shortly after we learnt
>>>> that more vendors want it and they are actually working on this.
>>> What kind of a property would you propose I use for this ?
>>
>> I don't know, please look at existing work around SSC from Peng. If
>> nothing is applicable, this should be explained somewhere.
> 
> The work from Peng you refer to (I guess) is this "assigned-clock-sscs" 
> property ? This is not applicable, because this is a boolean property of 
> the PHY here, the clock does not expose those clock via the clock API.

OK, please mention this in the commit msg - that assigned-clock-sscs is
not applicable, because these are clocks not exposed outside.

I saw already brcm,enable-ssc property, so use rather "realtek,enable-ssc".

> 
> However, I can call the property "ssc-enable" without the realtek, 
> vendor prefix ?

I think no, I am not so sure how generic it would be to cover all
existing cases. Some devices, e.g. cdns, defines the mode of SSC, so
uses an enum.

> 
> The remaining question is, should I have one property "ssc-enable" to 
> control all SSC in the PHY or one for each bit "realtek,ssc-enable-rxc" 
> / "realtek,ssc-enable-clkout" ?

I don't know. Can they be enabled independently? Does it make sense for
the hardware to have different choices?


Best regards,
Krzysztof