Re: [PATCH 4/4] arm64: dts: qcom: sc8280xp: gaokun3: describe rear camera module information as musch as possible

[Konrad Dybcio]

On 5/5/26 8:55 PM, Vladimir Zapolskiy wrote:
> On 5/2/26 15:55, Pengyu Luo wrote:
>> On Thu, Apr 30, 2026 at 7:00 PM Vladimir Zapolskiy
>> <vladimir.zapolskiy@xxxxxxxxxx> wrote:
>>>
>>> On 4/25/26 13:53, Pengyu Luo wrote:
>>>> The rear sensor is S5K3L6, describing it but dropping compatible
>>>> string, since there is no upstream driver. A funcitonal downstream
>>>> driver is in comment.
>>>>
>>>> The VCM is dw9714, describe it.
>>>>
>>>> Signed-off-by: Pengyu Luo <mitltlatltl@xxxxxxxxx>
>>>> ---
>>>> Please take this patch as a RFC, I am not sure, how much I am allowed
>>>> to add without a sensor driver.
>>>> ---
>>>>    .../boot/dts/qcom/sc8280xp-huawei-gaokun3.dts | 129 +++++++++++++++++-
>>>>    1 file changed, 123 insertions(+), 6 deletions(-)
>>>>
>>>> diff --git a/arch/arm64/boot/dts/qcom/sc8280xp-huawei-gaokun3.dts b/arch/arm64/boot/dts/qcom/sc8280xp-huawei-gaokun3.dts
>>>> index 39e559e91289..76b1ecb3819d 100644
>>>> --- a/arch/arm64/boot/dts/qcom/sc8280xp-huawei-gaokun3.dts
>>>> +++ b/arch/arm64/boot/dts/qcom/sc8280xp-huawei-gaokun3.dts
>>>> @@ -22,12 +22,18 @@
>>>>    #include "sc8280xp.dtsi"
>>>>    #include "sc8280xp-pmics.dtsi"
>>>>
>>>> +/* remove due to gpio pins collision, skip 2nd instance won't break things */
>>>> +/delete-node/ &cci1_i2c1;
>>>> +/delete-node/ &cci1_i2c1_default;
>>>> +/delete-node/ &cci1_i2c1_sleep;
>>>
>>> Instead of removal 'cci1_default' and 'cci1_sleep' nodes shall be rewritten
>>> by excluding 'cci1_i2c1_default' and 'cci1_i2c1_sleep' from them.
>>>
>>
>> Should we register an unused node?
>>
> 
> I believe unused nodes are found everywhere marked by status = "disabled"
> or if they do not produce a conflict for resources. I understand that a CCI
> IP with two ports are special, but likely (not always though) it should be
> acceptable to keep its device tree node as is.

I wanted the pins to move to the sub-buses as well, but that doesn't seem
trivial - we would either have to alter some obscure core code, or manage
the pinctrl states manually from within the CCI driver

(contributions welcome)

Konrad