#asahi-re · 2021-01-29 07:39
marcanthe macho file inside the img4 kernelcache#asahi-re · 2021-01-29 07:39
amwok - I haven't found that on my macbook air so far.#asahi-re · 2021-01-29 07:38
amwIs this stuff worth documenting/capturing?#asahi-re · 2021-01-29 07:38
marcanyou need to extract it yourself#asahi-re · 2021-01-29 07:38
marcanthat's macOS, that is not redistributable#asahi-re · 2021-01-29 07:38
amwmarcan: Is the kernelcache.macho file uploaded? or Should I make my own some how?#asahi-re · 2021-01-29 07:38
marcanoh that#asahi-re · 2021-01-29 07:37
amwThis was then dumped out with: aarch64-linux-gnu-objdump -D -b binary -a aarch64 -m aarch64 init.bin#asahi-re · 2021-01-29 07:35
amwWhere the 0xc04 was computed from the macho header format.#asahi-re · 2021-01-29 07:35
amwmarcan: Then it was dd if=kernelcache.macho bs=4k skip=$((0xc04)) count=16 of=init.bin#asahi-re · 2021-01-29 07:35
amwmarcan: The script I was refering to was used as: python machodump.py kernelcache.macho#asahi-re · 2021-01-29 07:34
amwmarcan: Sorry I was confused - you actually objdump a kernelcache#asahi-re · 2021-01-29 06:04
amwI thought you actually worked from the iBoot file off disk?#asahi-re · 2021-01-29 06:01
marcanI usually commit generally useful stuff#asahi-re · 2021-01-29 06:01
marcanamw: https://github.com/AsahiLinux/m1n1/blob/main/proxyclient/memdump.py#asahi-re · 2021-01-29 06:00
amwmarcan: I was wondering if the python script you used to dump the iBoot header at 3:46:00 in the last live stream was available some where?#asahi-re · 2021-01-27 21:43
never_releaseddavidrysk[m]: No.#asahi-re · 2021-01-27 20:45
davidrysk[m]mainly asking out of curiosity -- I expect that to be handled by iBoot et al#asahi-re · 2021-01-27 20:45
davidrysk[m]never_released: do these Apple chips have/use loadable CPU microcode?#asahi-re · 2021-01-27 19:04
koen_Sweet, thanks for that pointer =)!#asahi-re · 2021-01-27 19:03
jn__for off-topic stuff, there's #asahi-offtopic, too :)#asahi-re · 2021-01-27 19:01
koen_n spot what's going in a second, my messing around doesn't to bring me any further...#asahi-re · 2021-01-27 19:01
koen_Hey there! A bit off topic, not sure if there are better channels to ask this, but here it goes anyways, I am trying to get "Yabai" (a tiling WM for macOS) to work on the M1, but it throws an error on this line: https://github.com/koekeishiya/yabai/blob/master/src/osax/mach_loader.c#L103 saying "could not spawn remote thread: (os/kern) protection failure", the error code is 2, which is "KERN_PROTECTION_FAILURE", maybe someone here ca#asahi-re · 2021-01-27 18:37
davidrysk[m]or shadows it in some way? uh.#asahi-re · 2021-01-27 18:37
davidrysk[m]okay, S3_4_C15_C10_6 on t8101 seems to be equiv to CNTVCT_EL0 on t8020#asahi-re · 2021-01-27 18:29
davidrysk[m]I told the zfs people that they need to determine the minimal set of APIs they need and then file a support incident with apple asking for new APIs#asahi-re · 2021-01-27 18:23
j`eysounds like the Nvidia driver problem with 5.9#asahi-re · 2021-01-27 18:21
davidrysk[m]j`ey: it's causing problems for zfs because they use not officially exported symbols that were being accidentally exported before :P#asahi-re · 2021-01-27 18:18
j`eyget dat perf#asahi-re · 2021-01-27 18:17
davidrysk[m]apple compiled the arm kernels with the maximum amount of inlining/LTO and the minimum amount of symbols :P#asahi-re · 2021-01-27 18:17
davidrysk[m]hopefully t8101 xnu source / kdk will have more insights#asahi-re · 2021-01-27 18:17
davidrysk[m]and there are a whole load of undocumented but utilized Apple specific registers that might just be used for virtualization, idk#asahi-re · 2021-01-27 18:17
davidrysk[m]looks like A14/M1 has a ton of features that A12 doesn't e.g. DBGCLAIM#asahi-re · 2021-01-27 16:45
marcan(this is from the p-core init)#asahi-re · 2021-01-27 16:45
marcanyes, they used the spare bits.#asahi-re · 2021-01-27 16:45
marcan HID16_ENABLE_MPX_PICK_45 | HID16_ENABLE_MP_CYCLONE_7);#asahi-re · 2021-01-27 16:45
marcan reg_set(SYS_HID16, HID16_SPAREBIT0 | HID16_SPAREBIT3 |#asahi-re · 2021-01-27 16:45
marcandavidrysk[m]: btw#asahi-re · 2021-01-27 16:43
davidrysk[m]I think this is related to smp init but I need to load xnu into a disasm that supports xrefs :P#asahi-re · 2021-01-27 16:03
marcanso I only dumped out the start code into objdump#asahi-re · 2021-01-27 16:03
marcanI was only interested in the chicken bits#asahi-re · 2021-01-27 16:03
marcanI did not load all of xnu into a disasm#asahi-re · 2021-01-27 16:03
marcanand which version?#asahi-re · 2021-01-27 16:02
marcandavidrysk[m]: is that in the init path?#asahi-re · 2021-01-27 16:00
davidrysk[m]there's a fair amount of poking at HID registers that are not documented in the xnu we have#asahi-re · 2021-01-27 16:00
davidrysk[m]or rather, disasm#asahi-re · 2021-01-27 15:59
davidrysk[m]marcan: xnu#asahi-re · 2021-01-27 15:59
marcandavidrysk[m]: ^#asahi-re · 2021-01-27 15:59
marcanhttps://github.com/corellium/preloader-m1/blob/main/preboot.S#L288#asahi-re · 2021-01-25 19:07
j`eymaybe if other CPUs implement it too