Re: [PATCH v2 1/4] dt-bindings: firmware: scm: Add QDU1000/QRU1000 compatibles

[Melody Olvera]

On 10/20/2022 5:35 AM, Krzysztof Kozlowski wrote:
> On 19/10/2022 14:08, Melody Olvera wrote:
>>
>> On 10/15/2022 6:34 AM, Krzysztof Kozlowski wrote:
>>> On 14/10/2022 18:11, Melody Olvera wrote:
>>>> Add compatibles for scm driver for QDU1000 and QRU1000 platforms.
>>>>
>>>> Signed-off-by: Melody Olvera <quic_molvera@xxxxxxxxxxx>
>>>> ---
>>>>  .../devicetree/bindings/firmware/qcom,scm.yaml   | 16 ++++++++++++++++
>>>>  1 file changed, 16 insertions(+)
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/firmware/qcom,scm.yaml b/Documentation/devicetree/bindings/firmware/qcom,scm.yaml
>>>> index c5b76c9f7ad0..47083f47f109 100644
>>>> --- a/Documentation/devicetree/bindings/firmware/qcom,scm.yaml
>>>> +++ b/Documentation/devicetree/bindings/firmware/qcom,scm.yaml
>>>> @@ -38,6 +38,8 @@ properties:
>>>>            - qcom,scm-msm8994
>>>>            - qcom,scm-msm8996
>>>>            - qcom,scm-msm8998
>>>> +          - qcom,scm-qdu1000
>>>> +          - qcom,scm-qru1000
>>> Why exactly we are no using qdu1000 as fallback? That was the
>>> recommendation in previous discussion.
>> Will use only qdu; I think I misunderstood the outcome of that discussion.
> Actually, I think I commented about this in wrong patch. I think the
> outcome was to use two compatibles for most of the cases, but as a
> fallback, so:
>
> QDU: "qcom,qdu1000-rpmhpd"
> QRU: "qcom,qru1000-rpmhpd", "qcom,qdu1000-rpmhpd"
> (or skip entirely second if you do not customize QRU in DTSI)
>
> However here we already have a fallback, so these are fine:
>
> "qcom,scm-qdu1000", "qcom,scm"
> "qcom,scm-qru1000", "qcom,scm"
>
> Still assuming you customize them in DTSI... which does not seem the
> case, right?
Yeah dtsi is largely shared between RU and DU. It probably makes more sense to
drop RU compat string all together unless there is a significant difference.
>>> Patch is still incomplete - you still do no have proper changes in allOf
>>> for the clocks. If you want to say that this SoC does not take any
>>> clocks as input, then they should not be allowed.
>> That is what I'm trying to say; it seems most of our most recent SoCs (sm8*) don't have any
>> clocks associated with the scm. Does it make sense to remove the minItems earlier
>> in the binding, or is there something else that would communicate this in allOf better?
>>
>
> Then disallow clocks for your variant:
>
>   - if:
>      ....
>     then:
>      ...
>       clocks: false
>       clock-names: false
Got it; thanks.

Thanks,
Melody