Re: [PATCH] soc: qcom: icc-bwmon: drop the bandwidth vote across system suspend

[Konrad Dybcio]

On 10/19/26 5:48 PM, Ramón Jesús Miranda Hernández wrote:
> Hi Konrad,
> 
> Thanks for asking. I have now run the negative test on X1E80100 (Surface
> Pro 11): kernel 7.2.8-arch1-10 with the icc-bwmon PM ops removed (patch
> 0002 reverted), then s2idle.
> 
> - Duration: 10308 s (2 h 52 min), suspended 14:23 to 17:15 local.
> - Residency from qcom_stats: CXSD 100 %, AOSD 99.8 % (S0ix), DDR 100 %.
> - With the patch, earlier runs of 849 s and 880 s: S0ix 99.7 % and 99.9 %.
> - No CDSP crash (sleep_statsi) during this boot.
> 
> This contradicts the claim in my commit message that the qcom_stats "ddr"
> counter never advances without the patch. In this configuration it does
> advance, so reverting the PM ops alone does not break DDR collapse on this
> platform. It is one sample, and the runs differ in duration and conditions,
> so it is not a controlled comparison. I have not yet read the bwmon votes
> from the interconnect debugfs before suspend; I can add that if you want it.

No, this is conclusive enough that we can drop this patch.

Konrad