#asahi-re · 2022-01-24 13:02
Jamie[m]1hehe#asahi-re · 2022-01-24 11:16
Jamie[m]1s/on it/was it/#asahi-re · 2022-01-24 11:16
Jamie[m]1which is odd, because iChat was never on iOS on it?#asahi-re · 2022-01-24 11:16
Jamie[m]1avd driver strings reference special handling for iChat streams#asahi-re · 2022-01-20 12:43
Jamie[m]1what is goblin the codename for lmao#asahi-re · 2022-01-20 12:43
Jamie[m]1"FieldDiagsGoblin" #asahi-re · 2022-01-20 12:00
Jamie[m]1lolwut#asahi-re · 2021-12-17 10:41
Jamie[m]1just making sure i'm not unnecessarily replicating what's already been done#asahi-re · 2021-12-17 10:41
Jamie[m]1cool#asahi-re · 2021-12-17 10:08
Jamie[m]1is there some prior art on REing this stuff that I've missed?#asahi-re · 2021-12-17 10:07
Jamie[m]1(just that I think this has equally as much right to be called "an M3 mailbox", not least because apple's kexts call it that)#asahi-re · 2021-12-17 10:07
Jamie[m]1(to be clear I'm not arguing that it should be a common codebase, that would indeed be messy)#asahi-re · 2021-12-17 10:06
Jamie[m]1the masks in the upper half of MSG1 also match up to a 32-bit register here#asahi-re · 2021-12-17 10:06
Jamie[m]1exactly as in the linux m3 mailbox driver#asahi-re · 2021-12-17 10:06
Jamie[m]1)#asahi-re · 2021-12-17 10:06
Jamie[m]1(e.g. the host writes 8 to the offset 0x48 to ack an IRQ#asahi-re · 2021-12-17 10:05
Jamie[m]1and some in exact register offsets#asahi-re · 2021-12-17 10:05
Jamie[m]1but there are a ton of similarities in register order#asahi-re · 2021-12-17 10:05
Jamie[m]1the key difference being that messages are just 32 bit#asahi-re · 2021-12-17 10:05
Jamie[m]1fwiw on further RE I am very confident that the mailbox mechanism in the AVD shares ancestry with the other M3 mailbox#asahi-re · 2021-12-16 12:15
Jamie[m]1we can try and reconcile it once i understand it enough, i guess#asahi-re · 2021-12-16 12:13
Jamie[m]1except that it's not identical#asahi-re · 2021-12-16 12:13
Jamie[m]1not confident enough in my RE yet to be able to say for sure#asahi-re · 2021-12-16 12:09
Jamie[m]1and is hardcoded into the kext#asahi-re · 2021-12-16 12:09
Jamie[m]1the avd is double-annoying in that its firmware is loaded by macOS#asahi-re · 2021-12-16 12:08
Jamie[m]1heh this one is also called wrap#asahi-re · 2021-12-16 12:07
Jamie[m]1but the kext does refer to it as an m3 mailbox#asahi-re · 2021-12-16 12:07
Jamie[m]1sven: no, it doesn't look like rtkit#asahi-re · 2021-12-16 07:30
Jamie[m]1aaa#asahi-re · 2021-12-16 07:29
Jamie[m]1oh no, is that a 4th open video decoding API I didn't know about#asahi-re · 2021-12-16 07:25
Jamie[m]1haha wtf#asahi-re · 2021-12-16 07:19
Jamie[m]1and prores only on the m1 pro/max right?#asahi-re · 2021-12-16 07:18
Jamie[m]1heh i just added APR to that page :P#asahi-re · 2021-12-16 07:18
Jamie[m]1s/notable/notably/#asahi-re · 2021-12-16 07:17
Jamie[m]1(APR)#asahi-re · 2021-12-16 07:17
Jamie[m]1Potentially, but notable ProRes has a completely separate accelerator and kext#asahi-re · 2021-12-16 07:16
Jamie[m]1but does anyone have a clue about LGH?#asahi-re · 2021-12-16 07:16
Jamie[m]1HEVC is obvious, and AVC what is commonly known as H.264#asahi-re · 2021-12-16 07:16
Jamie[m]1functions in the AVD kext seem to come in 3 flavours: HEVC, AVC, and LGH