Re: [PATCH] arm64: qcom: dts: sm8750: add coresight nodes

[Jie Gan <jie.gan@oss.qualcomm.com>]

On 11/18/2025 6:03 PM, Konrad Dybcio wrote:
> On 11/17/25 10:31 AM, Jie Gan wrote:
>> Add CoreSight DT nodes for AOSS, QDSS, Turing, and Modem blocks to enable
>> the STM and TPDM sources to route trace data to the ETF for debugging.
>>
>> Signed-off-by: Jie Gan <jie.gan@oss.qualcomm.com>
>> ---
>>   arch/arm64/boot/dts/qcom/sm8750.dtsi | 981 +++++++++++++++++++++++++++++++++++
>>   1 file changed, 981 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/qcom/sm8750.dtsi b/arch/arm64/boot/dts/qcom/sm8750.dtsi
>> index 3f0b57f428bb..56c2605f3e0d 100644
>> --- a/arch/arm64/boot/dts/qcom/sm8750.dtsi
>> +++ b/arch/arm64/boot/dts/qcom/sm8750.dtsi
>> @@ -3313,6 +3313,948 @@ tcsrcc: clock-controller@f204008 {
>>   			#reset-cells = <1>;
>>   		};
>>   
>> +		stm@10002000 {
>> +			compatible = "arm,coresight-stm", "arm,primecell";
>> +			reg = <0x0 0x10002000 0x0 0x1000>,
>> +			      <0x0 0x37280000 0x0 0x180000>;
> 
> This region is a little bigger but it's not described clearly. Is there
> a reason to use this slice of it and not the whole thing?

This region is about the STM channels which are allocated for APSS/HLOS. 
The channel 10240-20479 is allocated for APSS/HLOS, each occupied 256. 
So the start address is 10240 * 256 = 2,621,440 (0x28000). The length is 
0x180000 because we only use part of these channels so far.

> 
> [...]
> 
>> +					funnel_in0_in0: endpoint {
>> +						remote-endpoint =
>> +						<&tn_ag_out>;
> 
> Please unwrap this line

Will fix it.

> 
> [...]
> 
> 
>> +		tn@109ab000 {
>> +			compatible = "qcom,coresight-tnoc", "arm,primecell";
> 
> "tnoc@"?

In dt-binding, we suggested tn, it also could be tnoc here.

> 
> [...]
> 
> 
>> +	tpdm_llm {
> 
> Node names still can't include underscores ;)

Sure, missed this part, will fix.

> 
> Please also rename 'llm' and 'turing' to 'cdsp' unless there's a good
> reason not to
> 

Will fix.

> I confirmed the coresight components addresses match their claimed function

Thanks for checking, it's a huge effort.

Thanks,
Jie

> 
> Konrad