#asahi-gpu · 2022-02-14 11:25
marcanso these should be ARM format page tables#asahi-gpu · 2022-02-14 11:25
marcanlooking at the firmware init a bit#asahi-gpu · 2022-02-14 11:25
marcanphire: yeah, the ARM's TTBR gets pointed straight at gfx-shared-region-base#asahi-gpu · 2022-02-14 11:14
povikexciting#asahi-gpu · 2022-02-14 11:14
povik11:38 < phire> I understand close to enough to launch a compute kernel#asahi-gpu · 2022-02-14 10:46
marcanyou sign extend at some bit#asahi-gpu · 2022-02-14 10:46
marcanand yes, that's how the PAC truncation works#asahi-gpu · 2022-02-14 10:46
marcanalso I see it going to f8..fa#asahi-gpu · 2022-02-14 10:46
phireit just seems to be sign extention of bit 39#asahi-gpu · 2022-02-14 10:46
marcanthat could be PAC?#asahi-gpu · 2022-02-14 10:45
phireit gets truncated to 40 bits#asahi-gpu · 2022-02-14 10:45
marcan0xffffff8000000000#asahi-gpu · 2022-02-14 10:45
marcanmore Fs#asahi-gpu · 2022-02-14 10:45
phireyeah, rtkit-private-vm-region-base is 0xf80_00000000 to 0xf8f_ffffffff#asahi-gpu · 2022-02-14 10:45
phireEverything I've seen points to a unified address space#asahi-gpu · 2022-02-14 10:44
marcanwe have rtkit-private-vm-region-base which points to what very much looks like a typical ARM top half VM address space base#asahi-gpu · 2022-02-14 10:44
marcanand then it's possible the ARM never actually has to R/W data in the lower half of the address space directly#asahi-gpu · 2022-02-14 10:43
marcanyeah, like maybe the ARM's actual sysreg TTBR1 gets pointed to the same L1[0] and that all uses the standard VMSA lockdown stuff#asahi-gpu · 2022-02-14 10:43
phirewell, same pagetables#asahi-gpu · 2022-02-14 10:43
phireponting at the same data?#asahi-gpu · 2022-02-14 10:43
marcanmaybe they run in parallel?#asahi-gpu · 2022-02-14 10:42
phireI think it might be a modified ARM MMU#asahi-gpu · 2022-02-14 10:42
marcani.e. does this sit entirely outside of the ARM's own MMU?#asahi-gpu · 2022-02-14 10:42
marcanalso I wonder how this works together with the ARM's actual page tables#asahi-gpu · 2022-02-14 10:41
marcanmakes sense#asahi-gpu · 2022-02-14 10:41
marcanyeah#asahi-gpu · 2022-02-14 10:41
phirebut they got combined#asahi-gpu · 2022-02-14 10:41
phireyes. UAT seems to mean Unified Address Translater, So I'm guessing there used to be a seperate GART and DART#asahi-gpu · 2022-02-14 10:40
marcanI'm guessing UAT is the current generation of GART or so#asahi-gpu · 2022-02-14 10:40
marcanbtw, stringsing the kexts, they seem to call the PTs UAT and GART somewhat interchangeably#asahi-gpu · 2022-02-14 10:40
marcanthat would be awesome#asahi-gpu · 2022-02-14 10:40
phireyes#asahi-gpu · 2022-02-14 10:40
marcanin python right?#asahi-gpu · 2022-02-14 10:38
phireI understand close to enough to launch a compute kernel#asahi-gpu · 2022-02-14 10:37
phireat this point I'm going to have to start on a m1n1 agx driver#asahi-gpu · 2022-02-14 10:35
marcanah, right#asahi-gpu · 2022-02-14 10:35
phireeach L0 is 16 bytes, so they sit next to eachother at gpu-region-base#asahi-gpu · 2022-02-14 10:35
marcanand doing it that way is quite Apple-like#asahi-gpu · 2022-02-14 10:35
marcanbut yeah, if the AGX firmware maps its own PTs read-only, and then VMSA locks itself, early in init, that's all the code we need to audit to say "no backdoor here"#asahi-gpu · 2022-02-14 10:34
phirethey all share the same L0[1] ptr#asahi-gpu · 2022-02-14 10:34
phireyeah, there is an L0 table for every context#asahi-gpu · 2022-02-14 10:34
marcanthat would make sense#asahi-gpu · 2022-02-14 10:34
marcanpossibly a per-context VA prefix is involved?#asahi-gpu · 2022-02-14 10:33
phireit lives at VA 0x1100018000#asahi-gpu · 2022-02-14 10:33
phireAh... found the Pipeline data I was missing#asahi-gpu · 2022-02-14 10:32
phireinteresting. I get the feeling we are going to have to RE this blob just so we know it's security model#asahi-gpu · 2022-02-14 10:32
marcantheir cores already have the ability to lock translation on#asahi-gpu · 2022-02-14 10:32
marcanthere's also the existing VMSA lockdown stuff#asahi-gpu · 2022-02-14 10:32
phiresome of the log messages I've seen talk about a "microPPL"#asahi-gpu · 2022-02-14 10:30
marcanso only the initial mechanism to populate the PTs would have to be locked off