#asahi-dev · 2025-12-08 08:49
nickchanintegralpilot: i have code execution in bootrom :) but only having EL0&1 is annoying#asahi-dev · 2025-12-08 08:48
integralpilotdidn't know #asahi-re existed, thanks for pointing it out, i'll use that next time for anything involving looking at what xnu does#asahi-dev · 2025-12-08 08:48
integralpilotooh, a11's cool - do you do iPhone RE (that must be really challenging/fun since you can't just run a fuOS)?#asahi-dev · 2025-12-08 08:42
nickchanthough i feel like some of this content should have been in #asahi-re#asahi-dev · 2025-12-08 08:38
nickchana11 soc is still my favourite #asahi-dev · 2025-12-08 08:38
nickchaneh im just taking others info#asahi-dev · 2025-12-08 08:37
integralpilotthank you so much for finding out all this info, it's super helpful#asahi-dev · 2025-12-08 08:34
nickchans3_4_c15_c10_7, you do not want to touch that so keep on with the nopping#asahi-dev · 2025-12-08 08:33
nickchanthey have a VNCR-like like register but for all their custom registers in NV#asahi-dev · 2025-12-08 08:32
nickchan(well, other than enabling NV, but then you would have to save/restore ALL the secret regs or otherwise everythiung blows up)#asahi-dev · 2025-12-08 08:32
nickchani think it's the correct solution#asahi-dev · 2025-12-08 08:32
integralpilotthat's been my current "solution#asahi-dev · 2025-12-08 08:32
nickchani think you need to code patch xnu that to nop#asahi-dev · 2025-12-08 08:31
nickchanintegralpilot: s3_4_c15_c14_6 = AGTCNTRDIR_EL12 so it's just incorrect behavior for el1#asahi-dev · 2025-12-08 08:31
integralpilot*away#asahi-dev · 2025-12-08 08:31
integralpilotdoesn't look trapped, causes an exception straight way in EL1 when xnu tries to read it#asahi-dev · 2025-12-08 08:29
nickchanintegralpilot: is S3_4_C15_C12_1 trapped at all or does the exception go right into EL1#asahi-dev · 2025-12-08 08:27
integralpiloti'll run it now, thanks#asahi-dev · 2025-12-08 08:27
nickchanit's crucial when handling unknown regs#asahi-dev · 2025-12-08 08:27
nickchanm1n1/find_regs.py#asahi-dev · 2025-12-08 08:26
nickchanright#asahi-dev · 2025-12-08 08:26
integralpilotit's _configure_misc_apple_regs#asahi-dev · 2025-12-08 08:26
nickchanintegralpilot: there's a bruteforce sysreg script in m1n1#asahi-dev · 2025-12-08 08:23
integralpilotnickchan: xnu tries to read, apply bitmask, and then write to it in a function that's called by _arm64_init, i'll check the exact name now#asahi-dev · 2025-12-08 08:20
nickchancause im pretty sure that's an EL2 reg#asahi-dev · 2025-12-08 08:19
nickchan integralpilot: context for S3_4_C15_C12_1?#asahi-dev · 2025-12-08 07:30
integralpilothopefully soon m1n1 hypervisor for XNU on t6030 will be able to boot the whole way#asahi-dev · 2025-12-08 07:30
integralpilotso i'm going to debug that next, but like starting IOKit is massive progress i'm so happy!!#asahi-dev · 2025-12-08 07:29
integralpilotit's successfully finishing _arm64_init now, and gets in the process of spinning up IOKit when it panics, then panics in panic handeller, then infinite loops as fallback#asahi-dev · 2025-12-08 07:29
svennice!!#asahi-dev · 2025-12-08 07:28
integralpilotooooh... did some fixing up for that register, and also S3_4_C15_C12_1 needed some fixing up and now we're making a lot more progress with XNU boot#asahi-dev · 2025-12-08 05:38
integralpiloti need to set HACR_EL2 right, but is there anyway to figure out what i need to set it to in order to trap S3_4_C15_C14_6?#asahi-dev · 2025-12-08 05:38
integralpilotokay did some more digging and it turns out we're not trapping S3_4_C15_C14_6 at all, and it can't be accessed by EL1 if EL2 doesn't have a trap configured#asahi-dev · 2025-12-08 05:07
integralpilotnoticed t6030 is missing the chickens for these timer registers that are causing issues, i'll see if added the t6031 setup logic for these registers fixes that#asahi-dev · 2025-12-08 05:07
integralpilotthanks everyone, i think i'll get it working on t6030 first then try t8122#asahi-dev · 2025-12-07 13:52
nickchans3_0_c15_c2_4 is HID31#asahi-dev · 2025-12-07 13:49
jannauhttps://github.com/AsahiLinux/m1n1/pull/518#asahi-dev · 2025-12-07 13:26
nickchan(kdk's libT8122.os.DEVELOPMENT.a's start.o has all the startup sequence because of course it is hidden from xnu source code)#asahi-dev · 2025-12-07 13:24
nickchanah#asahi-dev · 2025-12-07 13:24
nickchannow being startup code, there is no way around it; disassemble xnu's startup code and libT8122.os.DEVELOPMENT.a should be the best place to look#asahi-dev · 2025-12-07 13:23
jannauI have t8122 chickens but never pushed them#asahi-dev · 2025-12-07 13:22
nickchanespecially on the p cores#asahi-dev · 2025-12-07 13:22
nickchanintegralpilot: you will need to setup chickens for m3 otherwise errata will cause all sorts of weird behavior#asahi-dev · 2025-12-07 12:30
integralpiloti'll try and work on getting these registers working, i NOP-ed out the first accesses but then they just get used again a bit later so i think i proper solution is needed since the look like they are a bit in very early XNU boot#asahi-dev · 2025-12-07 12:29
integralpilotoh, that makes a lot more sense than it just being an m3 pro thing, sorry i misunderstood & thanks for explaining#asahi-dev · 2025-12-07 12:26
jannauthat's because m1n1 hasn't chickens for t8122 (m3)#asahi-dev · 2025-12-07 12:26
integralpiloti'll try and get them sorted out though, seems like the best way forward#asahi-dev · 2025-12-07 12:23
integralpiloti don't remember the exact plain m3 number#asahi-dev · 2025-12-07 12:23
integralpilotin the chickens they only look like they are set on t6030, not t80-whatever on plain m3