#asahi-dev · 2026-06-15 09:35
nickchancause apple signs every released version still#asahi-dev · 2026-06-15 09:34
nickchanin practice this only means no old beta downgrades#asahi-dev · 2026-06-15 08:49
nickchanwell yes I just thought it's not on the list because it's not on asahi#asahi-dev · 2026-06-15 08:46
nickchanfwiw yuka's patch works well on pre-M1 with cyc_ovrd workaround disabled or force down so I will give R-b when it hits the list#asahi-dev · 2026-06-15 07:37
nickchanSobek[m]: M5 Ultra probably#asahi-dev · 2026-06-15 07:36
nickchanconfirmation from... somewhere?#asahi-dev · 2026-06-15 07:34
nickchanbugs aside the only one that could really know is Apple#asahi-dev · 2026-06-15 07:32
nickchanthere's other explainations for that NO_CPU_OVRD, like the job is done by sptm or preconfigured by iBoot#asahi-dev · 2026-06-15 07:31
nickchanthe register still exists at least#asahi-dev · 2026-06-15 07:24
nickchani feel like for cpu if some sysreg workaround ugly enough is posted on the list then the people at Arm Limited may bother apple to unlock them and that could be much faster#asahi-dev · 2026-06-15 07:21
nickchane bogus and minimum design voltage is 0 mV#asahi-dev · 2026-06-15 07:21
nickchanseems like some recent T2 firmware have the SMC gain enough keys to "support" the macsmc-battery driver which is not something I want to support because half of the keys ar#asahi-dev · 2026-06-15 07:16
nickchanthe m2 ultra has 24 cpu so it looks like not even apple is escaping their own 2 VM limit#asahi-dev · 2026-06-15 07:15
nickchanand the xcode cloud 27 gives you 12 cpus per CI runner#asahi-dev · 2026-06-15 07:15
nickchanxcode cloud 27 anyways 26 is on intel xeon gold#asahi-dev · 2026-06-15 07:14
nickchanxcode cloud runs on M2 Ultra#asahi-dev · 2026-06-15 07:13
nickchanthe kind of demand good for those ultra would need to require loads of VRAM, but not fast VRAM#asahi-dev · 2026-06-15 07:12
nickchani really think most people that bought the high ram ultras are doing ai inference of sorts#asahi-dev · 2026-06-15 07:11
nickchanwith the current DRAM pricing#asahi-dev · 2026-06-15 07:11
nickchanchadmed: enough to run out of stock at least#asahi-dev · 2026-06-15 07:10
nickchanallegedly Nevis is some Apple Watch chip, and Borneo should be A20 Pro#asahi-dev · 2026-06-15 07:03
nickchani did not fill Sotra D but what it is could not be more obvious#asahi-dev · 2026-06-15 07:01
nickchanupdated soc codenames now M5 is, in fact, t8142#asahi-dev · 2026-06-14 17:56
nickchannohlt should work with m4 too shouldn't it#asahi-dev · 2026-06-14 17:55
nickchani do know that but the workaround is still as valid as on m1,m2,m3#asahi-dev · 2026-06-14 17:54
nickchanenable it for cpu implementer 0x61 specifically#asahi-dev · 2026-06-14 17:51
nickchanyuka: just in case here are the midrs https://github.com/HoolockLinux/linux/commits/wfi-state-pre-m1/#asahi-dev · 2026-06-14 17:51
nickchani think that can work#asahi-dev · 2026-06-14 17:44
nickchanbut on m1 the cpu really only looses gpr and simd state#asahi-dev · 2026-06-14 17:43
nickchanas in, lose all state and restart from rvbar and highest EL on older#asahi-dev · 2026-06-14 17:43
nickchanthe only difference is on older socs the consequence of losing state is more severe#asahi-dev · 2026-06-14 17:43
nickchani think everything with apple midr has the wfi lose state behavior#asahi-dev · 2026-06-14 17:31
nickchanyuka: i am preparing it#asahi-dev · 2026-06-14 17:29
nickchanand by adding a workaround for all the apple silicon cpus which could all lose state on wfi, I can upstream that midr patch#asahi-dev · 2026-06-14 17:28
nickchan(i am using those midrs to ban executing aarch32 on a10 but that's besides the point)#asahi-dev · 2026-06-14 17:28
nickchani am looked at chance to upstream the patch that added all the pre-M1 midrs#asahi-dev · 2026-06-14 17:27
nickchanit is not#asahi-dev · 2026-06-14 17:26
nickchanwell, the server is accessible now#asahi-dev · 2026-06-14 17:25
nickchanenough that the pmu patches that had little to do with that is affected#asahi-dev · 2026-06-14 17:25
nickchanand let's say people really do not like these kinds of behavior#asahi-dev · 2026-06-14 17:24
nickchanas well as aarch32 can only be executed in performance core, attempting to do so on e-core or switching to p-core while executing results in illegal instruction#asahi-dev · 2026-06-14 17:24
nickchanon A10/A10X/T2 the MPIDR_EL1 can appear to change while switching pstate because it's switching physical cores#asahi-dev · 2026-06-14 17:23
nickchanI suppose I could summarize:#asahi-dev · 2026-06-14 17:22
nickchanback to "where..." it's a combination of things: https://lore.kernel.org/asahi/aHUeUMmn_19EayL1@willie-the-truck/#asahi-dev · 2026-06-14 17:21
nickchanwell then that's fine it's not like having unused midr will do much#asahi-dev · 2026-06-14 17:18
nickchanwfi force up#asahi-dev · 2026-06-14 17:18
nickchansven: uses same workaround as m1#asahi-dev · 2026-06-14 17:18
nickchanbased on previous interactions the only way to get cpu midr into the kernel is to lump them together with M-series...#asahi-dev · 2026-06-14 17:18
nickchansven: nah not really#asahi-dev · 2026-06-14 17:15
nickchan(and i do not have th patches downloaded now)