17:41 <minecrell> lumag: FWIW you literally need a ~4 line change to use SPM on arm64. It is abstracted through the standard cpuidle framework, so all that is necessary is to adjust the Kconfig and fix a small compile error in cpuidle-qcom-spm.c on arm64

17:42 <minecrell> I don't think there are good technical arguments to forbid it, other than forcing vendors (the ones in control of the firmware, so usually not us) to use PSCI

17:43 <minecrell> This is actively used for msm8916 as well since DB410c is literally the only msm8916 device with PSCI support

17:44 <minecrell> The preparation for using SPM for MSM8916 is in mainline already but I never posted the 4 line patch to enable it on ARM64 because I didn't really feel like going through that discussion

17:45 <Mis012[m]> arguably PSCI is not particularly pretty, it's just a standardized way to abstract away potentially ugly stuff

17:46 <Mis012[m]> why make it sane when you can put it in a codebase that nobody expects to be sane

18:22 <z3ntu> bamse: Could you pick up the patches from https://lore.kernel.org/linux-arm-msm/20220612092218.424809-1-luca@z3ntu.xyz/ (apart from "perf data" commit which is already applied)?

18:23 <z3ntu> Ah actually some of the "core" clock commits from Dmitry seems to have been picked up so I'll respin to add core clock to msm8953.dtsi also

22:27 <lumag> minecrell, I think it's all about standards and forcing vendors.

23:48 <steev> forcing vendors doesn't help on eol devices