#asahi-re · 2026-07-18 17:44
amarioguyi'm trying to dump some of that area for context#asahi-re · 2026-07-18 17:44
amarioguyok definitely not the iorvbar region...but seems like trying to just read other stuff outside of that can result in an serror loop#asahi-re · 2026-07-18 17:43
amarioguynow why tf when i try to read from cpu-impl-reg (IORVBAR region) does it cause an serror loop...#asahi-re · 2026-07-18 17:42
amarioguyyep, now the proxy shell works#asahi-re · 2026-07-18 17:42
amarioguyacked#asahi-re · 2026-07-18 17:17
amarioguythis is on t8142, j704ap#asahi-re · 2026-07-18 17:17
amarioguyyuka: erm....seems like when i try to connect to the uartproxy the shell just spams '> 0'#asahi-re · 2026-07-17 19:02
amarioguyyuka: acked, that's good, all i really need is to read/write memory regions here#asahi-re · 2026-07-17 19:00
amarioguybtw just to ask, is m1n1 able to boot to the proxyconsole uart on m5?#asahi-re · 2026-07-16 01:34
amarioguymy hypothesis here is that these are "tunables" like mentioned before that are applied in the fuOS case, as the same functions seem to also apply other things too that are unrelated to cpu lockouts#asahi-re · 2026-07-16 01:33
amarioguyfor j616c/25G5043d, the base address is loaded from 0x10081D9A290, the tables are 0x10081D12728, 0x10081D126F0, 0x10081D12080 (last one i think is dependent on that flag)#asahi-re · 2026-07-16 01:29
amarioguyi *think* one of these might be conditioned on another flag that seemed lpol related too?#asahi-re · 2026-07-16 01:29
amarioguythe whole routine seems to be applying some values from the tables across the board like chaos_princess implied, critically these specific tables *only* seem to be accessed in the case where an fuos is being booted#asahi-re · 2026-07-16 01:28
amarioguythe actual tables i'm noting as "of interest" are 0x10081D116D8, 0x10081D11388, 0x10081D119F8#asahi-re · 2026-07-16 01:27
amarioguyfor j616c/25G5028f of note, before the functions that use these tables access them, it seems it loads the value 0x211000000 from 0x10081D9A290 (this is why i have the gut feeling these might be related as the effects of the lockout primarily affect cpu features and msrs and such, the CPU-local address being loaded here seems a bit important as it might imply it's a "base" address of sorts)#asahi-re · 2026-07-16 01:23
amarioguyif anyone wants to follow along, (the specific iBoots i'm diffing are 25G5043d and 25G5028f) will post some addresses below of some of the things i found, if someone wants to act as verifier but i have a bit of a gut feeling these might be it#asahi-re · 2026-07-16 01:19
amarioguy(as in might've stumbled across the table...)#asahi-re · 2026-07-16 01:19
amarioguyuhhhhhhhhhhhh i might have hit eureka?#asahi-re · 2026-07-15 23:03
amarioguyprobably some very opaque, firebloom'ed format that would be exceptionally unfun to RE....tho i guess i should do it just for the science aspect...#asahi-re · 2026-07-15 23:02
amarioguy(tl;dr my current suspicion is that it seems like the list of locked out registers is a data table of some kind)#asahi-re · 2026-07-15 23:02
amarioguyi *really* wish diaphora for data sections existed...#asahi-re · 2026-07-15 23:02
amarioguyi'm *guessing* here (strong emphasis on guessing) that the sysreg stuff might be pure data and there's a routine that iBoot runs to read a table of sorts to decide which stuff gets locked out#asahi-re · 2026-07-15 23:01
amarioguyidk how i forgot to post this the other day but i diffed the two iboots with the single change, didn't see anything immediately evident that would imply a code-path change#asahi-re · 2026-07-10 23:52
amarioguyso erm....back to shadow paging cursed stuff for >=m4 i presume?#asahi-re · 2026-06-30 03:16
amarioguywouldn't have linked it if i found that out...i really need to stop giving the benefit of the doubt in this situation#asahi-re · 2026-06-30 03:15
amarioguybruh....so sorry for not realizing the paper was slop assisted #asahi-re · 2026-06-29 22:03
amarioguynickchan: yep....kinda disappointed and puts a good bit of the paper's credibility under question...#asahi-re · 2026-06-29 17:14
amarioguynote that part of the methodology did involve binary RE hence why it's here#asahi-re · 2026-06-29 17:14
amarioguyhttps://arxiv.org/pdf/2606.22283#asahi-re · 2026-06-29 17:14
amarioguybut interesting paper came out recently on ANE, someone wrote a 300 page research paper on RE'ing the ANE#asahi-re · 2026-06-29 17:13
amarioguyjust realized i was disconnected before sending last time....#asahi-re · 2026-06-17 15:07
amarioguyyea that's what i meant lol#asahi-re · 2026-06-17 14:53
amarioguyyea i'm diffing 26.2b1 and b2 iboots right now as well#asahi-re · 2026-06-17 14:51
amarioguy0x210F80000, 0x211F80000, 0x212F80000 region it seems (this isn't in the outline hell i was mentioning months ago)#asahi-re · 2026-06-17 14:43
amarioguyidk how i completely missed it last time (i guess i wasn't diffing against pre-m4 then...the more you know)#asahi-re · 2026-06-17 14:41
amarioguyso i decided to look at iboot again this time also comparing against pre-m4 for the same ish version, seems like there's a unique register access only on m4 and later socs#asahi-re · 2026-06-11 19:48
amarioguyiirc back in jan/feb this year i was looking at how it was done, but the code i'd suspected to be responsible was.....about 500 outlined functions which was erm not fun#asahi-re · 2026-06-11 19:46
amarioguyStanfordAppliedCyber[m]: was the mechanism behind sysreg locking in iboot figured out?#asahi-re · 2026-05-13 16:10
amarioguyi found it strangely quite beneficial actually to learn like that tbh, because it made sure my mental model was closer to how hardware actually operates, rather than having to dig through many abstractions#asahi-re · 2026-05-13 16:09
amarioguyhabiby: don't worry, i'm the same, i learned how to reverse engineer before i could forward engineer :P#asahi-re · 2026-03-14 21:26
amarioguy(not the person who was testing on m2 btw, rather was the one looking into "m1n1 as mach-o")#asahi-re · 2026-03-14 21:06
amarioguytbh based on this i don't think the payload properties are the determining factor for whether or not something is going down the mach-o path or raw binary path#asahi-re · 2026-03-14 21:06
amarioguysome cross-referencing...seems lpol related#asahi-re · 2026-03-14 20:39
amarioguy(the logic to decide whether or not to boot via the fuOS path or not)#asahi-re · 2026-03-14 20:39
amarioguythere's some tag it's checking somehow called `hsec`#asahi-re · 2026-03-14 20:38
amarioguyre the SPTM things, decided to peek at iBoot again#asahi-re · 2026-02-19 20:20
amarioguynicolas17: no, more so just accounting for "my reversing of iboot a few weeks ago"#asahi-re · 2026-02-19 18:32
amarioguyunsure if i'm actually allowed to contribute the code, if not i'll list out what changes would need to be made#asahi-re · 2026-02-19 18:32
amarioguyi'll look into resurrecting building m1n1 as mach-o#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