Re: [PATCH] arm64: dts: qcom: qcs6490-rb3gen2: Adjust tsens thermal zone configuration

[Konrad Dybcio]

On 1/3/26 7:05 PM, Bjorn Andersson wrote:
> On Tue, Dec 16, 2025 at 01:01:01PM +0530, Manaf Meethalavalappu Pallikunhi wrote:
>> Hi Dmitry,
>>
>>
>> On 12/16/2025 1:21 AM, Dmitry Baryshkov wrote:
>>> On Mon, Dec 15, 2025 at 04:29:34PM +0530, Manaf Meethalavalappu Pallikunhi wrote:
>>>> The QCS6490 rb3gen2 board uses the same Qualcomm QCM6490 platform
>>> Is there a difference between QCS6490 and QCM6490 or between QC[SM]6490
>>> and SC7280?
>> Difference is only for qcs6490-rb3gen2 (IOT ).
>>>
>>>> but has a different thermal junction temperature specification
>>>> due to package-level differences.
>>>>
>>>> Update passive/hot trip thresholds to 105°C and critical trip
>>>> thresholds to 115°C for various subsystem TSENS sensors.
>>>>
>>>> Disable CPU cooling maps for CPU TSENS since CPU thermal mitigation
>>>> is handled automatically in hardware on this board.
>>> Is it a peculiarity of the RB3 Gen2 or is it that Chrome devices didn't
>>> do it? What about QCM6490 IDP or FairPhone FP5?
>>
>> Chrome devices do not perform automatic thermal mitigation, whereas all
>> other boards handle it automatically. I will push another patch to disable
>> cpu mitigation for all other boards other than chrome.
>>
> 
> This matches what we're doing everywhere else (i.e. rely on LMh/OSM/EPSS
> to do the fast throttling of cores), so I'd expect that this should
> apply to all non-Chrome boards...
> 
> Instead of doing this for every board, can we push the cpu/cooling-maps
> into e.g. sc7280-chrome-common.dtsi?

I think this is the way to go

> If I understand you correctly, we'd still have the desire to adjust the
> temperature values, but that's a smaller per-board change.

With the cooling out of the picture, the remaining delta is moving the
critical trip point from 110 to 115 degC, which sure, is useful if we
can use all that headroom, but it's also not a huge jump with a huge
amount of noise in the DT..

Could we get a confirmation about the non-IoT kodiak accepted
temperature ranges? Maybe those happen to be too high?

Konrad