#asahi-dev · 2026-11-09 17:54
svenit is!#asahi-dev · 2026-11-09 17:52
sventhat’s already in exclave userspace #asahi-dev · 2026-11-09 08:54
sventhat entire hv is a big hack so go ahead :D#asahi-dev · 2026-11-09 08:54
svenhttps://asahilinux.org/docs/sw/m1n1-hypervisor/#running-the-hypervisor-with-sptmtxmcl4-on-m5#asahi-dev · 2026-11-09 08:53
svenyou’ll still need that CTRR hack #asahi-dev · 2026-11-09 08:53
svenyes#asahi-dev · 2026-11-09 07:41
svenmight be similar for m5#asahi-dev · 2026-11-09 07:41
svenbase m4 has the old atcphy, m4 pro the new one due to tbt5#asahi-dev · 2026-11-09 07:38
svenif they don’t have tbt5 that’s what I’d assume #asahi-dev · 2026-11-08 20:59
svendon't think so. hv is still missing m5 pro and a18 pro support but that's not blocking a release#asahi-dev · 2026-11-08 20:33
svenit just has bonus fun like that co-processor now and changed the registers around I think#asahi-dev · 2026-11-08 20:33
svenso I *think* most of the sequencing for the m4pro atcphy stays the same#asahi-dev · 2026-11-08 20:04
svenm4 pro atcphy looks quite different I think but the good news is that XNU is still very chatty about register names :D#asahi-dev · 2026-11-08 13:10
svenchaos_princess: I was just checking who the spmi maintainers are and if there's any activity on the tree and it looks like they have a new one you haven't cced: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8f70c729e68c3e02ccca6d13c733175489fdc069#asahi-dev · 2026-11-07 23:18
sveni think I got them all#asahi-dev · 2026-11-07 20:09
sventhey changed that protocol layer now? :/#asahi-dev · 2026-11-07 14:34
svennice#asahi-dev · 2026-11-07 14:33
svenevery exceptions goes up to EL2 now anyway#asahi-dev · 2026-11-07 14:33
sventhough I guess that's not all that hard with the page I inject for vectors anyway#asahi-dev · 2026-11-07 14:32
svenyuuuup, and also UNDEF instructions please#asahi-dev · 2026-11-07 14:31
sventhat alone was like 80% of all traps to EL2#asahi-dev · 2026-11-07 14:30
sventurns out that trapping TPIDR_{GL,EL} for every single read is not a good idea...#asahi-dev · 2026-11-07 14:30
svenI thought I made it fast enough to not run into the timeouts for co-procs anymore :D#asahi-dev · 2026-11-07 14:30
sveneither that or m1n1 broke DCP#asahi-dev · 2026-11-07 11:07
svenyeah, should be fine then. i'll pick the dts changes up after the rest has been merged#asahi-dev · 2026-11-07 11:02
svenstill need to run dt-check-style but that should be fine#asahi-dev · 2026-11-07 11:01
sventhe dts changes at least look good to me#asahi-dev · 2026-11-07 11:00
svenif there are no driver changes it's easier for everyone if they go through apple-soc as well but that requires at least a mention in the cover letter#asahi-dev · 2026-11-07 11:00
svenmyria: fwiw, dts changes always go through apple-soc, dt-bindings always go through the driver tree if there are any driver changes. no need to mention that in the cover letter :-)#asahi-dev · 2026-11-07 00:05
svenwe should add if that's actually EL0 or fake-GL0-running-in-EL0#asahi-dev · 2026-11-07 00:04
sven== Exception taken from EL0t == <-- huh, it's from EL0?!#asahi-dev · 2026-11-07 00:04
svenyeah. 'night!#asahi-dev · 2026-11-07 00:03
svenugh... 1am.. I should go to bed instead :d#asahi-dev · 2026-11-07 00:02
svensometimes it's easy to see wtf is going on#asahi-dev · 2026-11-07 00:02
svenwhat instruction did it fault on? can you paste context() and bt() somewhere?#asahi-dev · 2026-11-07 00:02
svenbut usually only once you're past the login screen#asahi-dev · 2026-11-07 00:01
svenxnu will complain about UNEXP_RT_HIT_DIR otherwise at some point#asahi-dev · 2026-11-07 00:01
sventhe AMCC hook just prevents panics #asahi-dev · 2026-11-07 00:01
svendoes unmapped IPA drop you to a hv shell? could try context and bt then#asahi-dev · 2026-11-07 00:00
svenhrm#asahi-dev · 2026-11-07 00:00
svenm5 pro is still single die, isn't it?#asahi-dev · 2026-11-06 23:59
sventhat would mean shadow pageable corruption but I think I still map those as r/o to the guest#asahi-dev · 2026-11-06 23:59
svenhuh, I assume that address is just invalid?#asahi-dev · 2026-11-06 23:56
svenno SError or anything similar?#asahi-dev · 2026-11-06 23:56
sventhat's odd#asahi-dev · 2026-11-06 23:56
svenhuh#asahi-dev · 2026-11-06 23:56
svenyup :D#asahi-dev · 2026-11-06 23:48
svenand then just add them to the vreg list in the .c and .py file and see if that's enough to get it to boot :D#asahi-dev · 2026-11-06 23:48
svenand then see if they are impdef#asahi-dev · 2026-11-06 23:48
svenso what I usually did when I hit more than two of those in a row is to just scan the executable sections of SPTM and XNU for any sysregs