#
#asahi-re · 2026-01-29 06:47
chadmedah of course its the one i didnt try
#asahi-re · 2026-01-29 06:47
oliverregarding the issue you saw: you need to set "unk2" in PlaneInfo (dcp/ipc.py) to 3 (kIOSurfaceAddressFormatTiled). the field is not yet part of the struct in the kernel.
#asahi-re · 2026-01-29 06:47
oliverchadmed: i was using the m1n1 dcp experiment.py. i first got it to work in scanning out a dumped compressed framebuffer from macos to verify that i understand the numbers.
#asahi-re · 2026-01-28 09:03
nickchanive needed to set cdw12 to 0 on t8015
#asahi-re · 2026-01-28 08:04
svenmight make sense to dump the entire ans3 region and see if there's something that looks like a version
#asahi-re · 2026-01-27 22:53
Artingli can get fresh adt from tahoe if it would help
#asahi-re · 2026-01-27 22:49
jannauyes, iBoot 11881.61.3
#asahi-re · 2026-01-27 22:48
jannau15.2 if I can believe my file naming
#asahi-re · 2026-01-27 22:48
jannauhere neither but /defaults has 'nvme-iboot-sptm-security = None'
#asahi-re · 2026-01-27 22:48
Artingli actually just realized the adt is 6 months old, when i still was on sonoma. maybe that's also the issue
#asahi-re · 2026-01-27 22:46
Artingldumped from m1n1 with a proxy script
#asahi-re · 2026-01-27 22:45
chaos_princessare you looking at the template adt or live adt?
#asahi-re · 2026-01-27 22:45
Artingli dont understand what nvme-secure-reg-layout is. i just went through the RE again to fresh memory, it seems like it gets this from the adt. and in adt i cannot find anything like that
#asahi-re · 2026-01-27 22:44
jannauthe new compatible should get a "firmware_abi" or so property so that we don't run into the same issue again
#asahi-re · 2026-01-27 22:42
jannaubut if we do that we could just do the right thing and modify the compatible
#asahi-re · 2026-01-27 22:41
jannauthat doesn't sound like it would help. we would need something which is contained inside ans. there are various ways of how we could signal this externally
#asahi-re · 2026-01-27 22:30
Artingli only could find that it referenced nvme-secure-reg-layout, passed that to external kext, and based result of that either chose the old reg address or a new one
#asahi-re · 2026-01-27 22:28
Artingli am not sure what the condition checked, as it referenced other kexts and i didnt want to look into that
#asahi-re · 2026-01-27 22:27
Artinglwhile i was looking at init of ans during RE, there ware some conditions checking to use either 0x24008 or 0x28130 for unknown_ctrl and so on. and that happens on newer macos versions as well
#asahi-re · 2026-01-27 22:25
jannaunot just m3 and later
#asahi-re · 2026-01-27 22:25
jannauas fair as I'm aware it affects every device using macOS 15 for system fw
#asahi-re · 2026-01-27 22:25
Artingloh, well yeah
#asahi-re · 2026-01-27 22:24
chaos_princesswe know the fw version in m1n1
#asahi-re · 2026-01-27 22:24
Artingland m1n1 needs a way to know that this is ans for M3 and so on, right? because role is ANS2 still there in adt
#asahi-re · 2026-01-27 22:23
chaos_princessadt processing is done in m1n1, and at this point, it can change the compatible anyway
#asahi-re · 2026-01-27 22:22
Artinglm1n1 can pass values
#asahi-re · 2026-01-27 22:22
chaos_princesslinux doesn't have access to adt
#asahi-re · 2026-01-27 22:22
Artingli thought about that. in M3 adt the ans has compatible to ascwrap-v6. i don't have adt for other devices except m1, and there it is ascwrap-v4. maybe that can be used?
#asahi-re · 2026-01-27 22:19
svenand I don’t think there’s any version register or similar we could use to change the offsets with some patch cc‘ed to stable@
#asahi-re · 2026-01-27 22:18
jannaualso apple's devicetree have up M5 still have "role = ANS2" in the ans node
#asahi-re · 2026-01-27 22:18
svenit’s very annoying they moved registers :/
#asahi-re · 2026-01-27 22:16
Artinglnow it makes more sense. thank you'all
#asahi-re · 2026-01-27 22:15
chaos_princessArtingl: pretty much, SART is ans-specific, everything else uses DART that can also do remapping
#asahi-re · 2026-01-27 22:15
sven*others
#asahi-re · 2026-01-27 22:15
sventhe other usually have a much more powerful iommu called dart
#asahi-re · 2026-01-27 22:14
svenjust for the nvme coprocessor
#asahi-re · 2026-01-27 22:14
sventhe m1 co-processors already is ans3 fwiw, I messed the name up because I trusted some string the firmware printed
#asahi-re · 2026-01-27 22:13
Artinglso that's a way for cpu to restrict/allow memory regions for co-processors to access?
#asahi-re · 2026-01-27 22:13
sven*ans
#asahi-re · 2026-01-27 22:12
svenonly things listed there can be reached by the and co-processor
#asahi-re · 2026-01-27 22:12
Artingli see
#asahi-re · 2026-01-27 22:12
svenit’s a very simple memory protection unit
#asahi-re · 2026-01-27 22:11
Artinglthe sart_add_allowed_region i placed before rtkit_init in nvme.c m1n1, and before booting ans
#asahi-re · 2026-01-27 22:08
Artingljannau: for sart i did only this value, only worked with that. i am not really into arm64/apple silicon, just started learning in december. is sart something like mmu that allows certain memory regions?
#asahi-re · 2026-01-27 22:06
jannauhave you tested other values for sart_add_allowed_region()? 0x100_0000_0000 is the start of the physical ram on M3. That should be be mostly empty. I suspect it tries to access some MMIO. its own, pmp or pmgr would be my first guesses
#asahi-re · 2026-01-27 22:05
chaos_princessand now it must be non-zero
#asahi-re · 2026-01-27 22:05
chaos_princesswasn't one of the tunables for allowing cdw12=0?
#asahi-re · 2026-01-27 22:04
jannauthanks. so APPLE_ANS_UNKNOWN_CTRL and APPLE_ANS_LINEAR_SQ_CTRL moved. not sure what we can do except using new compatibles. I guess "apple,t*-ans2-15" and add code to m1n1 to modify compitbles when iboot is >= 15
#asahi-re · 2026-01-27 21:45
Artinglhello everyone. wanted to share findings I did last week over ans initialization on M3/new firmware. hope I didn't forget to comment on all important things https://gist.github.com/Artingl/c5e7f9fbb0307f7c236e29e4f1fc1f8e
#asahi-re · 2026-01-26 21:24
amarioguyi can probably snag access to an m5 machine soon as well so i can test it out when the time comes

plus ›