13:30 <macromorgan> not sure anyone is still looking at it, but I was finally able to get mainline A-TF working properly on my RK3566 based device. Turns out in addition to the SCMI clocking patch that's still pending (https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/31265) we also need to add the same regulator dependency for the GPU on the 3568 like we have to do on the 3588 (https://lore.kernel.org/linux-rockchip/20250220-rk3588-gpu-pwr-domain-

13:30 <macromorgan> regulator-v6-0-a4f9c24e5b81@kernel.org/)

13:30 <macromorgan> there's still a kernel error about unbalanced regulator disables, but that's probably a solveable problem and nothing more than a warning currently

14:00 <diederik> macromorgan: nice! I was planning to look at it "soon" (tm)

16:51 <Danct12> i have yet to try out mainline atf

16:52 <Danct12> so i'm still looking forward for improvements

16:52 <Danct12> especially with boot times because i've noticed that u-boot has this huge delay on boot

16:52 <Danct12> right after bl31 messages

18:22 <macromorgan> hmm... I notice for me that when I try to add the cpu-idle-states with mainline A-TF the device fails to boot. However if I use that on the binary A-TF it not only works but lowers idle power consumption by up to a half a watt

18:33 <diederik> What does 'SIP' stand for in f.e. ROCKCHIP_SIP_SUSPEND_MODE ?

18:49 <diederik> macromorgan: I just stumbled upon: https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7b7644831e7276f52 ("cpuidle: psci: Add trace for PSCI domain idle")