#
#asahi-re · 2026-02-13 11:49
nickchanand of course iBootStage2 as well
#asahi-re · 2026-02-13 11:49
nickchan"Firmware/all_flash/LLB.*.RELEASE.im4p"
#asahi-re · 2026-02-13 11:48
nickchanI can locate iBootStage1 just fine in macOS 26.2 ipsw
#asahi-re · 2026-02-13 08:56
chaos_princessno. iboot1 does get updated
#asahi-re · 2026-02-13 08:55
nnymThe macOS 26.2 IPSW seems to contain only iBootStage2. Does this mean that iBoot1 never changes?
#asahi-re · 2026-02-10 14:35
chaos_princesssptm-less machines have a different entry point, yes
#asahi-re · 2026-02-10 14:30
nnymThe entry point with SPTM enabled seems to be at `__TEXT_BOOT_EXEC` in the 26.2 kernels for T6020 and T6041; in the kernel for T6000 (without SPTM) there is no `__TEXT_BOOT_EXEC`. Has anybody looked at this?
#asahi-re · 2026-02-10 13:29
nnymGood point; thanks.
#asahi-re · 2026-02-10 13:28
svenhttps://github.com/apple-oss-distributions/xnu/blob/f6217f891ac0bb64f3d375211650a4c1ff8ca1ea/osfmk/arm64/sptm/start_sptm.s#L34
#asahi-re · 2026-02-10 13:27
svenif you look at the open source xnu repo it'll contain some comments regarding what's going on
#asahi-re · 2026-02-10 13:27
sventhe kernel is running under sptm there
#asahi-re · 2026-02-10 13:27
svenon m4?
#asahi-re · 2026-02-10 13:18
nnym`entry` (`__start`) looks like the kernel's entry point but the first parameter—which is supposed to be a pointer to the boot arguments—can be 4. What does this mean?
#asahi-re · 2026-02-10 10:05
nnymI found many occurrences of 2 address adding patterns in ApplePMGR that confuse me a little. I can guess that one adds an offset to a base address and produces an intentionally invalid address if the offset is bigger than 4 GiB; the other checks whether the top 10 bits changed in the sum and sets the top bits to `0xc8a2` but I don't know what that's supposed to do. Can somebody explain? I've ignored them until now as just error
#asahi-re · 2026-02-10 07:05
svenmaybe next weekend i'll actually have time for debugusb for real this time....
#asahi-re · 2026-02-10 07:03
yukaIt's nice to see it working, sure, but since the hypervisor is not working on M4 it's a lot less useful. As soon as you boot into a kernel baremetal, you will lose the virtual m1n1 UART.
#asahi-re · 2026-02-10 07:03
nnymVirtual UART is exactly what I wanted.
#asahi-re · 2026-02-10 06:42
yukathis also has nothing to do with -re though ;)
#asahi-re · 2026-02-10 06:41
yukaI would be curious how much of the hacks I did are necessary on t6041, but I also never got anywhere near run_guest working
#asahi-re · 2026-02-10 06:40
svensame ;)
#asahi-re · 2026-02-10 06:39
chadmedi have a feeling youll never need to :P
#asahi-re · 2026-02-10 06:39
sventhough i never tested windows I think :D
#asahi-re · 2026-02-10 06:39
svenessentially m1n1 just emulates a usb-to-uart converter using a USB device class that has built-in drivers on windows, mac and linux
#asahi-re · 2026-02-10 06:32
nnymOh.
#asahi-re · 2026-02-10 06:32
sveni precisely wrote that thing years ago to make this work with non-apple computers
#asahi-re · 2026-02-10 06:31
svenyeah, that's cdc acm and not thunderbolt
#asahi-re · 2026-02-10 06:31
nnymThe diff in `lsusb` is "Bus 005 Device 034: ID 1209:316d Generic m1n1 uartproxy v1.5.0-150-g4562f406-dirty".
#asahi-re · 2026-02-10 06:28
JamesCalligeros[m]Sounds like its just CDC ACM given the “evidence” is /dev/ttyACM
#asahi-re · 2026-02-10 06:27
svenor do you mean there's just a CDC ACM connection when m1n1 is running? because that's expected
#asahi-re · 2026-02-10 06:27
svencan you show the lsusb output? while m1n1 is runnning there shouldn't be any thunderbolt connection possible
#asahi-re · 2026-02-10 06:26
sventhat sounds very odd
#asahi-re · 2026-02-10 06:24
nnym_After enabling the USB stuff in my T6041, I can get a pure USB (Thunderbolt) connection between it and my non-Apple computer: I have `/dev/ttyACM{0,1}` and `/dev/m1n1`. I thought that it wasn't supposed to work without 2 Macs and I didn't see any recent commit about DebugUSB. `run_guest.py` almost works but there's some error message about reading less than some specified amount.
#asahi-re · 2026-02-07 12:43
nickchanfrom vphone600ap dt included in some pcc firmwares
#asahi-re · 2026-02-07 12:41
nickchan"device-tree-tag" property in ADT => EmbeddedDeviceTrees-11156.42.1 (example)
#asahi-re · 2026-02-07 12:25
jannauiirc dcp has messages with edt in context there it doesn't seem likely to refer to the main adt. maybe iboot has a partial devicetree embedded
#asahi-re · 2026-02-07 12:22
nickchannnym: I think it's just another name of ADT
#asahi-re · 2026-02-07 11:55
jannauI think device trees for some of the embedded coprocessors
#asahi-re · 2026-02-07 11:49
nnymWhat's that? A devicetree in a devicetree?
#asahi-re · 2026-02-06 17:31
nickchanedt = EmbeddedDeviceTree
#asahi-re · 2026-02-04 17:28
nnyms/ps-groups`\zs/ seems/
#asahi-re · 2026-02-04 17:27
nnymApplePMGR seems to determine whether to use `ps-regs` or `ps-groups` (and more) based on the presence and value of the property `edt-property-ver` apparently introduced in M4. The property seems to specify the schema of `/arm-io/pmgr` or some part of it but I don't know what the first "e" means. Like `ps-regs`, `ps-groups` to comprise 12-byte groups (only 4 in T6041).
#asahi-re · 2026-01-29 19:11
svenwould be very helpful if you figure out how pmgr works on m4 though
#asahi-re · 2026-01-29 19:11
svenNot yet!
#asahi-re · 2026-01-29 16:26
nnym_I see `ps-groups` used as a substitute for `ps-regs` in macOS 26.2 ApplePMGR. Has anybody looked into this (for M4)?
#asahi-re · 2026-01-29 06:51
oliveryea, that was the first issue i had, but i adjusted the offsets. i also had to adjust mesa to put it on a page boundary. when i did all that i got the triangle to look like in my picture. before that it was even more garbage. but likei said, there is just not enough color data in the first 20mb of my framebuffer.
#asahi-re · 2026-01-29 06:49
chadmedthe tile sizes also change depending on the bit depth aiui
#asahi-re · 2026-01-29 06:49
chadmedyeah i had all the tile arithmetic right in my spaghetti branch
#asahi-re · 2026-01-29 06:48
oliveryou then have to adjust the sizes a little. agx compression uses 64px "macro tiles", so the image is actually 192 16px tiles large, not only 189 as the interchange one, that uses 16px tiles directly.
#asahi-re · 2026-01-29 06:48
chadmedi initially thought that those artefacts in your screenshots might be caused by bad metadata offsets or bad stride in one or both of the "planes" in the surface
#asahi-re · 2026-01-29 06:47
chadmedevery time :P

plus ›