#asahi-dev · 2026-06-15 16:44
nickchanwith bit [56] set access to ARM PMUv3 registers in EL1 will be trapped to EL2 with EC==0x3f#asahi-dev · 2026-06-15 16:44
nickchanthe contents of this register is implementation defined#asahi-dev · 2026-06-15 16:42
nickchanchaos_princess: pmu emulation uses hacr_el2 [56]#asahi-dev · 2026-06-15 16:00
nickchanmicrosoft Windows needs it but clearly you are not the target audience for that kind of stuff#asahi-dev · 2026-06-15 15:58
nickchanreplacement only if one does not need the arm performance counters inside the vm#asahi-dev · 2026-06-15 15:57
nickchansimply clear cyc_ovrd's wfi_ret and setting wfi force start does not allow boot cpu to restart via wfi at least on t6000#asahi-dev · 2026-06-15 15:19
nickchanhm actually let me try to make it happen on t6000#asahi-dev · 2026-06-15 15:19
nickchani will use my devices then #asahi-dev · 2026-06-15 15:17
nickchanthere's also the issue of fb initialized while mmu is off which is very slow#asahi-dev · 2026-06-15 15:02
nickchanand m1n1 needs to remember where its pagetables are#asahi-dev · 2026-06-15 15:00
nickchanright, it definitely shouldn't init the heapblock again#asahi-dev · 2026-06-15 14:59
nickchanafter exercise that code path two times, would result in `dlmalloc: internal error` at least on a7#asahi-dev · 2026-06-15 14:59
nickchanin m1n1#asahi-dev · 2026-06-15 14:59
nickchani think when the rvbar entry on primary cpu code path is exercised, the malloc entries will get corrupted#asahi-dev · 2026-06-15 13:36
nickchanwasn't pstate 0 usually like, 24 MHz#asahi-dev · 2026-06-15 13:35
nickchando we keep dcp on? we have dcp_poweroff() and dcp_poweron() in firmware logs when going to s2idle#asahi-dev · 2026-06-15 13:32
nickchanspeaking of trampoline m1n1 entry point kind of has a trampoline now to drop to EL1 from EL3#asahi-dev · 2026-06-15 13:28
nickchanbut macos does the same thing and power consumption is perfectly fine there#asahi-dev · 2026-06-15 13:28
nickchani mean, wasn't there an issue where the uncore is not powering down on pro/max/ultra on deep wfi#asahi-dev · 2026-06-15 13:27
nickchanpmp :3#asahi-dev · 2026-06-15 13:25
nickchannow you need to restore the state of a bunch of system registers#asahi-dev · 2026-06-15 13:24
nickchanthe rvbar is still inside m1n1 while in linux#asahi-dev · 2026-06-15 13:23
nickchaneven the non retention wfi is already annoying enough#asahi-dev · 2026-06-15 13:06
nickchani hope upstream can reply soon#asahi-dev · 2026-06-15 12:59
nickchansent#asahi-dev · 2026-06-15 12:56
nickchani will write a reply on list#asahi-dev · 2026-06-15 12:52
nickchan...actually, where is "must use psci" even mentioned?#asahi-dev · 2026-06-15 12:51
nickchanthe check is still there and your list comes after that check#asahi-dev · 2026-06-15 12:50
nickchanso it works well enough for bare metal cpu behavior#asahi-dev · 2026-06-15 12:49
nickchanchaos_princess: apple,arm-platform is what cpuidle-apple does#asahi-dev · 2026-06-15 12:30
nickchanthis also allows the workaround to be enabled under m1n1 hypervisor#asahi-dev · 2026-06-15 12:29
nickchanonly real reliable indicator of running bare metal is checking "apple,arm-platform" in the root node fdt compatible#asahi-dev · 2026-06-15 12:28
nickchanwell, currentEL would be 2 in a VM with nested enabled#asahi-dev · 2026-06-15 12:28
nickchancurrentEL check don't work for nested case :melting_face:#asahi-dev · 2026-06-15 10:53
nickchanm3 does not have sptm in 15#asahi-dev · 2026-06-15 10:33
nickchansome leaf executables like launchd is gone#asahi-dev · 2026-06-15 10:06
nickchanboth of which suffers from cursed core switching#asahi-dev · 2026-06-15 10:06
nickchanyuka: Only on T2 and Apple TV 4K#asahi-dev · 2026-06-15 10:05
nickchanin any case it seems i should keep all pre-m1 cpu pstate experiments to myself for longer since none of them work without usb dwc2#asahi-dev · 2026-06-15 10:03
nickchan(albeit with an error rate that cause adt fetch to explode 1/3 of the time)#asahi-dev · 2026-06-15 10:02
nickchanfine#asahi-dev · 2026-06-15 10:02
nickchanfor me 1.5M baud don't seem to work at <600 MHz, but on 1152 MHz it seems to be find#asahi-dev · 2026-06-15 10:00
nickchanin my case once the cpu switch to the lower state, it can't keep up with 1.5M baud anymore#asahi-dev · 2026-06-15 10:00
nickchanyuka: how did you run cpufreq experiments on neo anyways#asahi-dev · 2026-06-15 09:57
nickchani think i will just keep the workarounds to myself and add experiments for other socs#asahi-dev · 2026-06-15 09:57
nickchanthat is if both init and end freq is higher than around 450 MHz#asahi-dev · 2026-06-15 09:56
nickchanif switching from one core type to another then the frequency always decrease to around 450 MHz, change MPIDR, and start to increase again#asahi-dev · 2026-06-15 09:56
nickchanno, it's cursed core switching on A10#asahi-dev · 2026-06-15 09:55
nickchanactually no, what it doesn't like is frequency starting to decrease when a switch is initiated, only to increase again halfway through the switch#asahi-dev · 2026-06-15 09:50
nickchanthe cpu_pstate_latencies seems to not like higher pstate slower than low pstate