Re: [PATCH v2 1/3] media: dt-bindings: Add qcom,sm6150-camss

[Wenmeng Liu <wenmeng.liu@oss.qualcomm.com>]

On 12/24/2025 1:03 PM, Dmitry Baryshkov wrote:
> On Wed, Dec 24, 2025 at 12:31:33PM +0800, Wenmeng Liu wrote:
>>
>>
>> On 12/24/2025 12:21 PM, Dmitry Baryshkov wrote:
>>> On Wed, Dec 24, 2025 at 11:18:02AM +0800, Wenmeng Liu wrote:
>>>>
>>>>
>>>> On 12/23/2025 9:38 PM, Krzysztof Kozlowski wrote:
>>>>> On Mon, Dec 22, 2025 at 04:28:39PM +0800, Wenmeng Liu wrote:
>>>>>> +  interconnects:
>>>>>> +    maxItems: 4
>>>>>> +
>>>>>> +  interconnect-names:
>>>>>> +    items:
>>>>>> +      - const: ahb
>>>>>> +      - const: hf0_mnoc
>>>>>> +      - const: hf1_mnoc
>>>>>
>>>>> Same comments as before, do not invent names.
>>>>
>>>> <&mmss_noc MASTER_CAMNOC_HF0 QCOM_ICC_TAG_ALWAYS
>>>> &mc_virt SLAVE_EBI1 QCOM_ICC_TAG_ALWAYS>,
>>>> <&mmss_noc MASTER_CAMNOC_HF1 QCOM_ICC_TAG_ALWAYS
>>>> &mc_virt SLAVE_EBI1 QCOM_ICC_TAG_ALWAYS>,
>>>>
>>>> This platform(qcs615) is different from others. It has two types of sf,
>>>> namely sf0 and sf1.
>>>> The same as it is:
>>>> sc7180 sc8180x sdm670 sdm845 sm8150
>>>> Do you have any suggestions about this?
>>>
>>> Which _names_ are used on other platforms? This question is quite clear
>>> from Krzysztof's comment.
>>
>> The platform mentioned above either has no camss ICC node or no support for
>> CAMSS on the upstream, so this is a new one.
> 
> I did a quick look for you.
> 
> kodiak, lemans, monaco: ahb, hf_0
> 
> x1e80100: ahb, hf_mnoc, sf_mnoc, sf_icp_mnoc
> sm8650: ahb, hf_mnoc
> agatti: ahb, hf_mnoc, sf_mnoc
> sm8550: ahb, hf_0_mnoc
> 
> sc8280xp: cam_ahb, cam_hf_mnoc, cam_sf_mnoc, cam_sf_icp_mnoc
> sm8250: cam_ahb, cam_hf_0_mnoc, cam_sf_0_mnoc, cam_sf_icp_mnoc
> sdm660: vfe-mem
> 
> I'd obviously hope for some unification here. Other than that, we have
> two clean winners: KLM and X Elite+SM8650+Agatti. Yours proposal is
> different from either of the options. In fact, none of the platforms
> have the same _approach_ as yours. Why?
> 

Yes, you're right.
But none of the above cases involved having two hf_mnoc simultaneously, 
so do you have any good suggestions for handling such a situation?


Thanks,
Wenmeng