#
#asahi-re · 2025-04-03 21:03
chaos_princessthats gonna be annoying to reverse via m1n1 on future machines
#asahi-re · 2025-04-03 21:02
sveni currently think so as well
#asahi-re · 2025-04-03 21:02
chaos_princessdo we need to do it from raw boot object? i assume yes
#asahi-re · 2025-04-03 21:02
chaos_princessinteresting
#asahi-re · 2025-04-03 21:02
svenso the sptm for t8112 inside UniversalMac_15.4_24E248_Restore.ipsw contains the chicken sequence
#asahi-re · 2025-04-03 21:01
chaos_princessi've grepped disassembly of kernel (15.something, don't remember which) for the msrs to chicken registers and they are still present on m3 and down kernels and are only missing on m4 (though could be dead code)
#asahi-re · 2025-04-03 20:59
svenxnu is already loaded with MMU enabled afaict
#asahi-re · 2025-04-03 20:58
svenokay, so on t8112 the chickens are done inside SPTM now so that's probably true for t8132 as well now
#asahi-re · 2025-04-03 17:30
svenyeah, at that point hacking xnu itself might be the easier approach
#asahi-re · 2025-04-03 17:18
nickchanwhich I may do on a11 soontm
#asahi-re · 2025-04-03 17:16
nickchanlike, at that point i am not sure if proper hypervisor is worth it compared to just hijack xnu vbar handling and tracing by disabling access in pagetables (all under sptm)
#asahi-re · 2025-04-03 17:13
nickchanwhich is ultra complex, of course
#asahi-re · 2025-04-03 17:12
nickchansuppose that cannot be worked around without bugs the way I see is run m1n1 in EL2 under sptm in GL2 and then have (possibly NV if sptm & xnu really does not like it) GL1 sptm and EL1 xnu
#asahi-re · 2025-04-03 17:10
sventhat EL2 only issue still makes the hypervisor very painful though
#asahi-re · 2025-04-03 17:07
yukaPossible
#asahi-re · 2025-04-03 17:07
svenmaybe they just broke raw boot objects in 15.2 and fixed that later
#asahi-re · 2025-04-03 17:06
yukaI don't have access to my m4 until monday but I will certainly try 15.4 and maybe some other stuffs
#asahi-re · 2025-04-03 17:05
yukaJust m1n1
#asahi-re · 2025-04-03 17:05
yukaHaven't done anything like that yet
#asahi-re · 2025-04-03 17:05
yukaOooh I see
#asahi-re · 2025-04-03 17:05
nickchanon 15.2
#asahi-re · 2025-04-03 17:05
nickchanright and after unwrapping the macOS kernel macho from the img4 try setting the macho as the custom boot object
#asahi-re · 2025-04-03 17:04
yukaEverything works the same in mac os 15.1 and 15.2 except that on 15.2 it bootloops
#asahi-re · 2025-04-03 17:03
yukaMy process has been: install second mac os, sth sth csrutil (disable SIP), set custom boot object
#asahi-re · 2025-04-03 17:01
nickchanneed to disable ctrr too
#asahi-re · 2025-04-03 17:01
nickchanfrom the ipsw
#asahi-re · 2025-04-03 17:01
nickchanyuka: does setting the kernel for the ipsw as the custom boot object work at least
#asahi-re · 2025-04-03 17:00
jannau_kmutil does not complain but reboot before early debug writing on the uart
#asahi-re · 2025-04-03 16:59
yukadid not test 15.4 yet
#asahi-re · 2025-04-03 16:59
yukaafter updating the base system to mac os 15.2 I didn't get any uart output, just instant bootloop and eventually goes into the dfu mode
#asahi-re · 2025-04-03 16:59
jannau_did you already test 15.4?
#asahi-re · 2025-04-03 16:59
svenweird
#asahi-re · 2025-04-03 16:59
yukaunclear
#asahi-re · 2025-04-03 16:58
svenso 15.2 just has no support for custom boot objects / fuOS anymore?
#asahi-re · 2025-04-03 16:58
svenI haven’t looked at iboot at all fwiw
#asahi-re · 2025-04-03 16:58
yukayup
#asahi-re · 2025-04-03 16:58
svenhuh, really?
#asahi-re · 2025-04-03 16:58
svennot yet
#asahi-re · 2025-04-03 16:58
yuka15.2+ does not boot custom code at all
#asahi-re · 2025-04-03 16:57
yukaI assume you have noticed the difference between 15.1 and 15.2+ iboot (at least on base M4)?
#asahi-re · 2025-04-03 16:56
svennot looking forward to that either
#asahi-re · 2025-04-03 16:55
sventime to RE iboot then I guess :(
#asahi-re · 2025-03-05 05:52
chadmedahhh
#asahi-re · 2025-03-05 05:52
nickchanchadmed: drm-misc
#asahi-re · 2025-03-05 05:51
chadmedor did it go in through drm?
#asahi-re · 2025-03-05 05:51
chadmedoh yeah having both would suck. if we havent sent a PR up to torvalds yet then there may be time to get it changed
#asahi-re · 2025-03-05 05:50
nickchanregardless of compatibility it's dodgy
#asahi-re · 2025-03-05 05:50
nickchanchadmed: the dodgy thing is apple,txxxx-something followed by apple,hxx-something
#asahi-re · 2025-03-05 04:02
chadmedwe do that for all the other soc blocks
#asahi-re · 2025-03-05 04:01
chadmedapple,txxxx-adp is dodgy?

plus ›