Re: [PATCH v7 2/9] dt-bindings: display/msm: dp-controller: fix fallback for SM6350

[Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>]

On 02/09/2025 12:01, Dmitry Baryshkov wrote:
> On Tue, Sep 02, 2025 at 08:04:54AM +0200, Krzysztof Kozlowski wrote:
>> On 02/09/2025 06:03, Dmitry Baryshkov wrote:
>>> On Mon, Sep 01, 2025 at 05:45:49AM +0200, Krzysztof Kozlowski wrote:
>>>> On Fri, Aug 29, 2025 at 01:48:15AM +0300, Dmitry Baryshkov wrote:
>>>>> The SM6350 doesn't have MST support, as such it is not compatible with
>>>>> the SM8350 platform. Add new entry for SM6350 with fallback to SC7180
>>>>> (which belongs to the same generation and also doesn't have MST
>>>>> support).
>>>>>
>>>>> Fixes: 39086151593a ("dt-bindings: display: msm: dp-controller: document SM6350 compatible")
>>>>> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
>>>>> ---
>>>>>  .../devicetree/bindings/display/msm/dp-controller.yaml     | 14 +++++++++++++-
>>>>>  1 file changed, 13 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/Documentation/devicetree/bindings/display/msm/dp-controller.yaml b/Documentation/devicetree/bindings/display/msm/dp-controller.yaml
>>>>> index aed3bafa67e3c24d2a876acd29660378b367603a..0f814aa6f51406fdbdd7386027f88dfbacb24392 100644
>>>>> --- a/Documentation/devicetree/bindings/display/msm/dp-controller.yaml
>>>>> +++ b/Documentation/devicetree/bindings/display/msm/dp-controller.yaml
>>>>> @@ -31,13 +31,25 @@ properties:
>>>>>            - qcom,sm8650-dp
>>>>>        - items:
>>>>>            - enum:
>>>>> -              - qcom,sar2130p-dp
>>>>>                - qcom,sm6350-dp
>>>>> +          - const: qcom,sc7180-dp
>>>>> +
>>>>> +      # deprecated entry for compatibility with old DT
>>>>> +      - items:
>>>>> +          - enum:
>>>>> +              - qcom,sm6350-dp
>>>>> +          - const: qcom,sm8350-dp
>>>>> +        deprecated: true
>>>>
>>>> If it is only about bindings then there is little benefit in keeping
>>>> this, just drop this case.  However you cannot drop it from DTS, so this
>>>> is a bit pointless.
>>>
>>> Our plan is:
>>> - land updated DT bindings, describing MST clocks on MST-enabled
>>>   platforms,
>>> - land updated DTS, adding MST clocks where applicable,
>>
>> This part breaks all out-of-tree users of DTS.
> 
> User of which one? SM6350 or all DTS?

SM6350.

> 
> It extends the ABI, so no, it shouldn't. We add regions and clocks,

No, it does not extend the ABI. You are changing the fallback, so you
are changing the ABI.


> existing users can use the previously defined feature set.
> 
> Anyway, is it about adding more explicit note to the cover letter and/or
> commit message or would you have any other proposal on how to handle the
> issue?

Please be explicit if Linux was working with the old fallback or not. In
the DTS change be explicit about impact, e.g. that sc7180 fallback is
there since forever.

> 
>>
>>> - land driver changes, keeping legacy support for non-MST DTs on
>>>   MST-enabled SoCs
>>>
>>>>
>>>> Lack of MST support is not informative enough to claim it is not
>>>> compatible with 8350. For example if it was working fine via fallback,
>>>> then that statement is simply not correct.
>>>>
>>>> And it HAD to work fine, because there is nothing binding to
>>>> qcom,sm6350-dp.
>>>
>>> It is working fine since currently we don't have MST support on the
>>> driver side (nor do we describe MST clocks in DT). It's true that the
>>> driver will have to handle non-MST DT for SM8350. However I definitely
>>> don't want to describe both cases in the bindings. SM6350 is not going
>>> to be compatible with the MST-enabled SM8350 schema.
>>
>> The question is rather: is SM6350 going to be compatible (working) with
>> MST-enabled drivers.
> 
> If somebody implements e.g. U-Boot DP MST support after we land DP MST
> bindings, I would not guarantee that. SM8350 will already have second
> stream clock.


Having second stream clock does not mean device stops working with Linux
kernel. I can easily imagine both cases after adding MST to the drivers,
the SM6350 stops working or keeps working...

Best regards,
Krzysztof