#asahi-re · 2026-07-30 08:53
svenfffffe000c36a2f0: 90fddb50 adrp x16, 0xfffffe0007ed2000#asahi-re · 2026-07-30 08:53
svenfffffe000c36a2f8: 35000090 cbnz w16, 0xfffffe000c36a308#asahi-re · 2026-07-30 08:53
svenfffffe000c36a2ec: d503245f bti c#asahi-re · 2026-07-24 08:06
sveni'll be very very happy if you (or anyone else) figures out what's wrong :D#asahi-re · 2026-07-24 08:05
svenyeah#asahi-re · 2026-07-24 07:50
svenand then there’s also that weird tbt issue where the MMIO traces look similar enough and yet the tbt connection just won’t come up :(#asahi-re · 2026-07-23 08:13
svennow that could be extended to m4 where it’ll get horribly cursed because that require shadow pagetable to emulate SPRR then #asahi-re · 2026-07-23 08:13
sventhey are in academia so they have to hand in their thesis before the code etc will be public though :(#asahi-re · 2026-07-23 08:12
svensomeone has managed to virtualize SPTM and XNU using nested virt and some slightly cursed hacks on M2#asahi-re · 2026-07-22 08:38
sven:D#asahi-re · 2026-07-22 08:37
svenif you wanted Linux under SPTM we’d probably need to introduce a new architecture since it really wouldn’t be arm64 anymore #asahi-re · 2026-07-22 08:37
svenyes, Linux wants to manage its own pagetables #asahi-re · 2026-07-22 08:34
svenhttps://arxiv.org/abs/2510.09272 Is probably useful as well #asahi-re · 2026-07-22 08:34
svenalso make sure to take a look at open source XNU’s entry point. there’s some comments about how you need to talk to SPTM initially #asahi-re · 2026-07-22 08:33
svenmight make sense to start “from scratch”, i.e. only take basic proxyclient features and take it from there #asahi-re · 2026-07-22 06:28
svenyou may be able to convince XNU to just not use exclaves by messing with the device tree #asahi-re · 2026-07-22 06:27
svenkinda like TXM is SPTM’s user space #asahi-re · 2026-07-22 06:26
svenand the applets/exclaves are running in user space (GL0) then #asahi-re · 2026-07-22 06:26
svenyeah, the exclave kernel is essentially a VM running in GL1 (and maybe EL1) under SPTM as its hv #asahi-re · 2026-07-21 20:42
sventhey had to build new silicon for this, it’s not just iboot changes #asahi-re · 2026-07-21 20:41
svenwe’re pretty sure they locked down all these sysregs because arm lawyers complained about all apple extensions #asahi-re · 2026-07-21 20:41
svenwhat?#asahi-re · 2026-07-21 19:56
sveni think even GXR/SPRR but that doesn't help us because host-XNU already drives all the hardware and won't expose it to us#asahi-re · 2026-07-21 19:56
svenallows you to run EL1 guests from XNU with some additional apple features enabled#asahi-re · 2026-07-21 19:55
svenoh, and you need to write to HCR to enable EL1 anyway#asahi-re · 2026-07-21 19:54
svenand SPTM will probably make sure that you can't setup stage 2 translations, etc.#asahi-re · 2026-07-21 19:53
svenyou e.g. can't write to HCR from EL2 the moment GXF is enabled#asahi-re · 2026-07-21 19:45
sventhe latter#asahi-re · 2026-07-21 19:44
svenyeah#asahi-re · 2026-07-21 19:43
svenah#asahi-re · 2026-07-21 19:43
svenuh, what do you mean with sptm enabled m1n1?#asahi-re · 2026-07-21 19:42
svenmight be interesting either way for some experiments#asahi-re · 2026-07-21 19:41
svensure, but keep in mind that running m1n1 under SPTM is going to be quite annoying#asahi-re · 2026-07-21 19:40
svenyou're definitely not running under sptm unless you specifically adjusted m1n1 :)#asahi-re · 2026-07-21 19:39
svensptm owns the pagetables and you have to nicely ask it if you want to change any#asahi-re · 2026-07-21 19:39
svenwe can't run linux under sptm#asahi-re · 2026-07-21 19:38
svenmight even be easier to fork m1n1 and adjust it to be able to run under SPTM for experiments#asahi-re · 2026-07-21 19:38
svenand m1n1 will probably panic very early when it tries to access the uart. and even if that worked it'll definitely panic when trying to enable the MMU#asahi-re · 2026-07-21 19:37
svenit won't even allow the m1n1 macho to be installed right now because it doesn't look like xnu anymore :D#asahi-re · 2026-07-21 19:32
svenand then they didn't test anything 🙃#asahi-re · 2026-07-21 19:31
svenafaict on M4+ they at least have some way to configure which sysregs are read-only and i think they just disabled everything except what's in the arm architecture reference manual#asahi-re · 2026-07-21 19:30
sveni don't think the silicon is there on M3 to actually lock anything down#asahi-re · 2026-07-21 19:12
sveni hope i'm not right :D#asahi-re · 2026-07-21 19:12
svenif it's bugged all we can do is file a feedback assistant report and hope for the best essentially :(#asahi-re · 2026-07-21 19:08
svenand the fuos path needs to somehow lock down all sysregs, etc.#asahi-re · 2026-07-21 19:08
sventhe paths must look very different, the regular macos one needs to load SPTM and XNU to memory#asahi-re · 2026-07-21 19:05
svenand the fuos path is just not very well tested by apple#asahi-re · 2026-07-21 19:05
svenyes, because the boot path for fuos is different from the "regular" macos boot path#asahi-re · 2026-07-21 18:13
sveni still suspect that the smp issue is just some bug in iboot :(#asahi-re · 2026-07-21 07:13
sventhat change is more that trivial enough ;)