#
✕ nick:kh
#asahi-dev · 2023-02-08 16:29
khmarcan: Are non-PSCI cpuidle drivers blocked on arm64? Sudeep doesn't seem to be against it
#asahi-dev · 2023-01-26 14:11
khmarcan: If you report it can you cc me too
#asahi-dev · 2023-01-26 13:59
khsystemd folks say otherwise https://github.com/systemd/systemd/issues/9538#issuecomment-405901335
#asahi-dev · 2023-01-26 13:57
khHere's my very poor attempt at fixing it https://github.com/Kaz205/linux/commit/6e19008c4eb51cdae4b5a1bb9f7f3cec7bfa78b3
#asahi-dev · 2023-01-26 13:49
khThat gets called in call_cpuidle_s2idle()
#asahi-dev · 2023-01-26 13:48
khsched_clock_suspend() is what stops the clock I think
#asahi-dev · 2023-01-26 13:48
khI don't think so, CLOCK_MONOTONIC only should stop during suspend, not idle
#asahi-dev · 2023-01-26 13:46
khThat causes programs like systemd to crash at times, and if I'm unlucky systemd-logind will crash and then I'm no longer able to login
#asahi-dev · 2023-01-26 13:43
khIt does but it's completely broken, for example CLOCK_MONOTONIC keeps ticking even though it has to be stopped because we ended up exiting early in cpuidle_idle_call() before we have a chance to stop it
#asahi-dev · 2023-01-26 13:40
khSeems like PSCI cpuidle has to be added before enabling suspend
#asahi-dev · 2023-01-26 13:39
khIt ended up being a really messy implementation since Arm removed support for the generic cpuidle driver
#asahi-dev · 2023-01-26 13:38
khI tried making it work without a cpuidle driver but I failed, the kernel doesn't boot anymore
#asahi-dev · 2023-01-26 13:36
khIt fails here https://github.com/torvalds/linux/blob/master/kernel/sched/idle.c#L188
#asahi-dev · 2023-01-26 13:36
khmarcan: Seems like s2idle was not working at all
#asahi-dev · 2023-01-12 13:34
kh(It's been a while since I last benchmarked it)
#asahi-dev · 2023-01-12 13:32
khchadmed: Sorry I forgot, I used coremark as well. Anyways the result was the exact same for me
#asahi-dev · 2023-01-12 08:35
khchadmed: Which benchmark did you use? Kernel docs recommend dhrystone hence why I used that
#asahi-dev · 2023-01-11 15:59
khchadmed
#asahi-dev · 2023-01-11 15:59
khWeird, because I got the exact same results when I ran dhrystone
#asahi-dev · 2023-01-10 23:34
khchadmed: Why drop the capacity, those are very much accurate
#asahi-dev · 2023-01-04 18:45
khI have the same issue on M1 MBP (j293) too
#asahi-dev · 2023-01-04 18:45
kh> but then its like disk is dont accessible anymore, if I was in a shell
#asahi-dev · 2022-12-23 07:51
khchadmed: You tested it on macos?
#asahi-dev · 2022-12-23 01:28
khI briefly remember you saying that you*
#asahi-dev · 2022-12-23 01:26
khchadmed: How did you find the power consumption for the energy model? I briefly remember saying you figured it out by reading some addresses through m1n1
#asahi-dev · 2022-12-17 02:44
khSame kernel same dtb same initramfs
#asahi-dev · 2022-12-17 02:44
khWeird...
#asahi-dev · 2022-12-17 02:44
khIt ended up booting when I used run_guest_kernel.sh
#asahi-dev · 2022-12-17 02:40
khOk I fixed it by adding UUID to the kernel cmdline but now it's stuck on "Run /init as init process"
#asahi-dev · 2022-12-17 02:33
kh"Can't open blockdev"*
#asahi-dev · 2022-12-17 02:33
khHas anyone faced something similar?
#asahi-dev · 2022-12-17 02:33
khI get "Can't get blockdev" when I use tools/linux.py
#asahi-dev · 2022-11-22 14:56
khsven: Oh, idk then
#asahi-dev · 2022-11-22 14:52
khYes that's a change to a userspace tool and I don't think there's any kernel changes required for it
#asahi-dev · 2022-11-22 14:48
khYeah but people previously were saying that f2fs should be blocked on 16k kernels altogether...
#asahi-dev · 2022-11-22 14:47
khsven: Hey, at least it works
#asahi-dev · 2022-11-22 14:44
khRegarding f2fs on 16k https://android-review.googlesource.com/c/platform/system/core/+/2244154