#asahi-dev · 2026-10-05 11:32
yukaOdd. I have not seen problems with secondary CPU bringup on M4 Pro 26.6.2#asahi-dev · 2026-10-05 11:31
integralpilotill get upstream patch ready soon and ill put u Co-developed-by cause i used ur MIDRs plus the testing rly helped#asahi-dev · 2026-10-05 11:31
integralpilotno that's so fine dont worry about it! thank you so much for helping#asahi-dev · 2026-10-05 11:31
Hel3zi26.6.2 #asahi-dev · 2026-10-05 11:31
myriasorry it took so long to test#asahi-dev · 2026-10-05 11:30
myriaintegralpilot: can confirm kvm-arm.vgic_v3_group1_trap=1 fixes the boot on m3 max#asahi-dev · 2026-10-05 11:28
yukaInteresting. Which macos version did you use as base for the boot volume?#asahi-dev · 2026-10-05 11:28
Hel3ziyuka: To get it booting, the key changes were: 1. Stubbed out dapf and pcie_init in m1n1. 2. Used run_guest.py -C 1 (without -C 1, secondary CPU bringup crashes the hypervisor). 3. Fixed KeyError: 4 in hv/__init__.py for self.sysreg (boot CPU is cpu4). With that, the kernel boots cleanly all the way to VFS panic! code Text [ 0.670571] leds_pwm led-controller: CLIDR_EL1.LoC == 0, treating as coherent [ 0.674128] apple-pmgr-pwrstate ... PS dispext2_sys#asahi-dev · 2026-10-05 11:25
mmediouni[m]jannau: wow, better than the Android pile of hacks from before#asahi-dev · 2026-10-05 11:24
integralpilotbut i think with a little bit of work it should be fine#asahi-dev · 2026-10-05 11:24
integralpilotfs/binfmt_elf.c and mm/filemap.c#asahi-dev · 2026-10-05 11:23
integralpilotomg after fixing all the trivial things there's only 2 which don't apply cleanly#asahi-dev · 2026-10-05 11:22
myriaChecking the rest now#asahi-dev · 2026-10-05 11:22
integralpilotawesome thank you!!#asahi-dev · 2026-10-05 11:21
myriaintegralpilot its=off allows it to boot with uefi firmware#asahi-dev · 2026-10-05 11:20
integralpilotbtw asahi-6.18.15-1 base is much better, actually 60 apply cleanly and 16 don't (and most of those seem trivial)#asahi-dev · 2026-10-05 11:18
integralpilotjannau: thanks#asahi-dev · 2026-10-05 11:15
yuka* take a count#asahi-dev · 2026-10-05 11:14
yukaso -C 1 means running only on the second processor#asahi-dev · 2026-10-05 11:14
yukabtw the -C option does not take acount, but a list of cpus#asahi-dev · 2026-10-05 11:14
jannauper process page size talk should start in a minute or so https://www.youtube.com/watch?v=ksq7uqsg2kM#asahi-dev · 2026-10-05 11:13
yukajust using run_guest.py instead of linux.py?#asahi-dev · 2026-10-05 11:13
yukaHel3zi: what was the required change?#asahi-dev · 2026-10-05 11:10
integralpilot(the red one)#asahi-dev · 2026-10-05 11:10
integralpilothttps://alpinelinux.org/downloads/#asahi-dev · 2026-10-05 11:10
integralpilotcan you try a userspace? i personally use/like alpine mini rootfs for stuff like this#asahi-dev · 2026-10-05 11:10
integralpilotyay :)#asahi-dev · 2026-10-05 11:07
Hel3ziyuka: Using run_guest.py -C 1, the kernel successfully boots on the M4 Pro! Log via ttyACM1: code Text [ 0.000000] Booting Linux on physical CPU 0x0000010100 [0x611f0551] [ 0.000000] Machine model: Apple MacBook Pro (14-inch, M4 Pro, 2024) [ 0.000000] earlycon: s5l0 MMIO32:0x0000000429200000 (options '') [ 0.012056] smp: Brought up 1 node, 1 CPU [ 0.013396] CPU: All CPU(s) started at EL1 [ 0.040109] Memory: 22762848K/2304#asahi-dev · 2026-10-05 11:06
integralpilots/it/I#asahi-dev · 2026-10-05 11:04
integralpilotmaybe it can be a branch like fairydust but then it wouldnt put it past That Distro to ship it in prod like they did fairydust#asahi-dev · 2026-10-05 11:02
chadmedbuuuutttt i really want to play cs and i dont want to boot and then update my stupid x86 machine so #asahi-dev · 2026-10-05 11:02
integralpilottrue#asahi-dev · 2026-10-05 11:02
chadmeds/ir/it #asahi-dev · 2026-10-05 11:02
chadmedtbh id prefer we didnt just because i dont want people to start relying on it and then get mad if it never makes it upstream and we rugpull ir#asahi-dev · 2026-10-05 11:01
integralpilots/its/"is it"#asahi-dev · 2026-10-05 11:01
integralpilotwould it be possible for us to ship it downstream before it goes upstream, or its something way to messy and large to be carried downstream#asahi-dev · 2026-10-05 11:00
chadmed(thats us!)#asahi-dev · 2026-10-05 11:00
chadmedim hoping they do intend to either upstream or genericise this at some point given that "micro VMs" are called out as a specific pain point they are aiming to address on the first slide#asahi-dev · 2026-10-05 11:00
chadmedhopefully there arent a lot of androidisms #asahi-dev · 2026-10-05 11:00
integralpilotits not actually as bad as it looks because many of the "conflicts" are just added headers#asahi-dev · 2026-10-05 10:59
integralpilotthanks, maybe thats a better approach#asahi-dev · 2026-10-05 10:56
jannauI'd start with the latest asahi-6.18 tag and move forward from that. probabbly depends on things in the android kernel#asahi-dev · 2026-10-05 10:54
integralpilot23 files patches apply cleanly, 52 dont, 1 missing file (fs/userfaultfd.c)#asahi-dev · 2026-10-05 10:53
integralpilotalso im trying to rebase the android changes on 7.3-rc6 now cause i want to try them#asahi-dev · 2026-10-05 10:51
myriawilco#asahi-dev · 2026-10-05 10:49
integralpiloti'm pretty sure no firmware = hang is expected with -cdrom on arm#asahi-dev · 2026-10-05 10:49
integralpilotit might be in a different location for you but that's where it installs for me#asahi-dev · 2026-10-05 10:48
integralpilotcan you try that please? thanks#asahi-dev · 2026-10-05 10:48
integralpilotadd -bios /usr/share/edk2/aarch64/QEMU_EFI.silent.fd#asahi-dev · 2026-10-05 10:46
myriaahhhh, that would do something. how did you do that?