#asahi-re · 2026-08-11 15:21
svenIOThunderboltSwitchType5(%x@%llx)::configureRouter - writing acio_fw_debug 0x%08x <-- now that sounds interesting#asahi-re · 2026-08-11 08:56
BrandiWhite[m]https://t.me/+pa-CiKYv9-5lNWM8#asahi-re · 2026-08-11 08:56
BrandiWhite[m]Get in touch with this platform for greatness you’ll definitely thank me later
ℹ️❤️
👇🏻👇🏻👇🏻#asahi-re · 2026-07-31 21:51
amarioguywell guess i'm reading elr no matter what then for this#asahi-re · 2026-07-31 21:48
amarioguysighs a bit#asahi-re · 2026-07-31 21:48
amarioguyso undefined instructions have the same exception information....yep#asahi-re · 2026-07-31 21:40
amarioguybecause the imm is cleared for some reason after m1n1 spills the exception info#asahi-re · 2026-07-31 21:40
amarioguynot 0#asahi-re · 2026-07-31 21:40
amarioguywait oh dear...the imm is 0x14 actually#asahi-re · 2026-07-31 21:35
amarioguyofc i could just be dumb and that's expected behavior for HVC traps for FAR to read as 0, but i'm not the hv expert here#asahi-re · 2026-07-31 21:35
amarioguyoption 2 is treat FAR = 0 as the PMUv3 case always#asahi-re · 2026-07-31 21:35
amarioguyoption 1 is 32-bit read ELR to see if it's one of the PMU msr insn's and handle in that case, else treat as normal hvc#asahi-re · 2026-07-31 21:33
amarioguywhat's not helping is it *seems* like FAR_EL{2, 1} is reading as 0 so i basically have two options here, neither of which i like#asahi-re · 2026-07-31 21:29
amarioguyfor the purposes of trapping, that's sufficient so the main concern i had for win on m4/m5 is not an issue...the question is how to differentiate between this and an actual hvc#asahi-re · 2026-07-31 21:28
amarioguy(this is on m5 btw)#asahi-re · 2026-07-31 21:28
amarioguyoddly enough resolves to HVC invocation with imm = 0#asahi-re · 2026-07-31 21:27
amarioguyESR_EL1 regardless of tge value = 0x5a000000#asahi-re · 2026-07-31 21:26
amarioguywas testing PMUv3 in raw boot mode#asahi-re · 2026-07-31 21:26
amarioguynickchan: yep, just remembered...#asahi-re · 2026-07-31 20:54
NickChan[m]trying to access not exist sysreg or UDF instruction would do that#asahi-re · 2026-07-31 20:53
NickChan[m]Isn't that just undefined instructions #asahi-re · 2026-07-31 20:34
amarioguyi'm *guessing* it's some bizarro impdef thing?#asahi-re · 2026-07-31 20:32
amarioguybtw does anyone know what ESR syndrome 0x2000000 means?#asahi-re · 2026-07-30 17:48
nicolas17it takes weeks between a build being in the PCC transparency log and the binaries actually being downloadable, but 27 is not in the log at all#asahi-re · 2026-07-30 17:45
nicolas17would make sense to not put a beta kernel in their production servers :P#asahi-re · 2026-07-30 17:45
nicolas17NickChan[m]: I think 27 is not being used in PCC yet#asahi-re · 2026-07-30 09:42
NickChan[m]27 kdks are available #asahi-re · 2026-07-30 09:42
svensame for kdks i guess, or is there a 27 kdk yet?#asahi-re · 2026-07-30 09:42
svenah#asahi-re · 2026-07-30 09:41
NickChan[m]pcc releases are very delayed so version 27 is not available yet#asahi-re · 2026-07-30 09:41
sventexit is 0xd6ff03e0 according to my binutil if anyone wants to look#asahi-re · 2026-07-30 09:40
NickChan[m]assuming they want it to possibly work in the future #asahi-re · 2026-07-30 09:39
NickChan[m]i thibk the interesting thing would be the vresearch101 sptm#asahi-re · 2026-07-30 09:38
svens/either/just/#asahi-re · 2026-07-30 09:38
svenno texit in SPTM, so i guess it's either #ifdefed out there#asahi-re · 2026-07-30 09:31
j`eychadmed: FEAT_TEV etc aren't about different exception levels, so different from GXF but I guess allows to accomplish similar things#asahi-re · 2026-07-30 09:29
sveni think my cursed "shift SPTM down one level by patching _EL12/_EL2 to _EL1 or hvc" might just work though#asahi-re · 2026-07-30 09:26
svenyeah, it's handed down by SPTM i think#asahi-re · 2026-07-30 09:16
sveni think that flag just comes from a bootarg in some handover struct so probably just from iboot#asahi-re · 2026-07-30 09:16
yukaand if there's a chicken bit, we probably can't change it#asahi-re · 2026-07-30 09:15
jannauso not before m6 (unless there's a chicken bit)#asahi-re · 2026-07-30 09:14
yukajannau: m5pro ID_AA64MMFR4_EL1: 0x0 (https://termbin.com/ki1q)#asahi-re · 2026-07-30 09:14
svenhahahaha :D#asahi-re · 2026-07-30 09:13
jannauespecially when you forget to update your ARMs regularly#asahi-re · 2026-07-30 09:12
svenvery happy it was added though because this could've sent me down another rabbit hole :D#asahi-re · 2026-07-30 09:12
jannauadded in december: https://sourceware.org/pipermail/binutils/2025-December/146143.html#asahi-re · 2026-07-30 09:06
svenjannau: weird that binutils already knows about tenter though#asahi-re · 2026-07-30 09:05
Sobek[m]Hmme MTE / EMTE are hardening, but iirc definitely weaker than say Cheri (but I guess I really need to get reading the spec for MTE / EMTE at some point).#asahi-re · 2026-07-30 09:04
chaos_princesssobek[m]: they are already using capabilities, FEAT_MTE#asahi-re · 2026-07-30 09:03
Sobek[m]chaos_princess: What if Apple decide to have hardware accelarated capabilities (since iirc they have a software version of capabilities, inspired by Cheri), and use them to further hardened some privileged parts) ?