Re: [PATCH 2/6] arm64: dts: qcom: sm8550-qrd: remove data-lanes property of image sensor

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

On 12/3/25 12:57 PM, Vladimir Zapolskiy wrote:
> On 12/3/25 13:50, Konrad Dybcio wrote:
>> On 12/3/25 12:41 PM, Vladimir Zapolskiy wrote:
>>> On 12/3/25 13:11, Konrad Dybcio wrote:
>>>> On 12/3/25 5:05 AM, Vladimir Zapolskiy wrote:
>>>>> Samsung S5K3M5 image sensor supports only 4-lane MIPI CSI-2 interface,
>>>>> which makes the sensor data-lanes property redundant, and it can be
>>>>> safely removed from the board dts file. Noteworthy that the property
>>>>> value is incorrect, because conventionally lanes enumeration of image
>>>>> sensors starts from index 1.
>>>>
>>>> This seems to vary between sensors
>>>>
>>>
>>> It should not, from Documentation/devicetree/bindings/media/video-interfaces.yaml
>>> the clock lane has index 0, data lanes enumeration starts from 1:
>>>
>>>    data-lanes:
>>>    ....
>>>      description:
>>>        An array of physical data lane indexes. Position of an entry determines
>>>        the logical lane number, while the value of an entry indicates physical
>>>        lane, e.g. for 2-lane MIPI CSI-2 bus we could have "data-lanes = <1 2>;",
>>>        assuming the clock lane is on hardware lane 0.
>>
>> I interpret the last sentence as ">>if<< the clock lane is 0, then data lanes
>> start at 1. I'm not sure if the spec mandates that's the case though.
>>
> 
> Well, that's the absolute common practice for media devices, which I'd like
> to follow without a deviation from it or introducing a novation.
> 
> The dt bindings documentation of this S5K3M5 sensor device describes the value
> of the optional data-lanes property accordingly:
> 
> https://lore.kernel.org/linux-media/20251203040241.71018-2-vladimir.zapolskiy@linaro.org/
> 
> So, I belive, this change under review should be fine as is.

What I'm trying to get an answer for, is that whether the lanes
are physically connected like this. You can find an answer for that
in the board schematics, specifically looking at the connection between
the PHY and the sensor.

Konrad