#
#asahi-dev · 2026-01-09 21:12
Shizand without the _recount_transition one it hangs in... _recount_transition :p
#asahi-dev · 2026-01-09 21:11
Shizdabort_lower without the two msr patches
#asahi-dev · 2026-01-09 21:11
svenShiz: what does it run into without them? illegal instruction because the msr doesn't exist in el1?
#asahi-dev · 2026-01-09 21:06
Shizsven: looks like all 3 patches are atm actually needed :p
#asahi-dev · 2026-01-09 21:02
jannaufairydust is rebased onto asahi-6.18.4-1
#asahi-dev · 2026-01-09 20:48
svenand ideally debug any usb device that breaks and directly submit a patch with the fix to atcphy or dwc3 :P
#asahi-dev · 2026-01-09 20:44
jannauasahi-6.18.4-1 tagged and pushed. test USB devices
#asahi-dev · 2026-01-09 20:40
fl0_idchaos_princess hvc bisect looks promising btw, at least now I know it reache init_el in head.S. haven't had time to bisect further yet
#asahi-dev · 2026-01-09 20:21
Shizhttps://txt.shiz.me/ZGY0Y2Y3OD
#asahi-dev · 2026-01-09 20:21
Shizokay, so booting xnu under the m3 now gets us to roughly the same point as with the m3 pro
#asahi-dev · 2026-01-09 20:06
Shizokay, it mysteriously worked now and we're very confused
#asahi-dev · 2026-01-09 19:49
noopwafeltime to play bisect
#asahi-dev · 2026-01-09 19:44
Shizwell, at least it does work as chainloaded payload
#asahi-dev · 2026-01-09 19:43
Shizah, gotcha
#asahi-dev · 2026-01-09 19:40
jannauShiz: I didn't test after rebasing and changing things so it's very possible that it is just broken
#asahi-dev · 2026-01-09 19:23
Shizchainloading also works
#asahi-dev · 2026-01-09 19:22
Shizupstream m1n1 does work
#asahi-dev · 2026-01-09 19:09
Shiz(this one: https://github.com/AsahiLinux/m1n1/pull/518)
#asahi-dev · 2026-01-09 19:09
Shiz(trying to get serial working in the meantime)
#asahi-dev · 2026-01-09 19:09
Shizjannau: we're trying to boot your t8122 m1n1 branch on noopwafel's m3, but it seems to be stuck in a bootloop -- any ideas?
#asahi-dev · 2026-01-09 16:54
jannaukettenis: unmodified u-boot v2026.01 boots for me on a m2 mac mini
#asahi-dev · 2026-01-09 16:51
sven(trying to improve usb2 phy init in m1n1 right now)
#asahi-dev · 2026-01-09 16:51
svenhrm, the entire usb.c code with the i2c vs. spmi and some other model-specific stuff could use some refactoring as well
#asahi-dev · 2026-01-09 16:45
svennice, usb dts changes on the mailing list :)
#asahi-dev · 2026-01-09 14:40
fl0_idty
#asahi-dev · 2026-01-09 14:40
fl0_idright
#asahi-dev · 2026-01-09 14:40
chaos_princessor multiple hvcs and use m1n1 to release it after each one
#asahi-dev · 2026-01-09 14:40
fl0_idyes
#asahi-dev · 2026-01-09 14:39
chaos_princessso, binary search with hvc
#asahi-dev · 2026-01-09 14:38
fl0_id*right now
#asahi-dev · 2026-01-09 14:38
fl0_idright. ok. looking at that right. I'm not interested in a particular function atm, just trying to figure out where it reaches before exception/crashing
#asahi-dev · 2026-01-09 14:36
chaos_princessyes. just inline asm with a hypervisor call
#asahi-dev · 2026-01-09 14:36
fl0_idchaos_princess you mean hypervisor calls, not hvc tty driver, right?
#asahi-dev · 2026-01-09 14:24
chaos_princessso, idea - sprinkle hvc instructions into the function you are debugging
#asahi-dev · 2026-01-09 14:20
fl0_idthis is through EL1, when running run_guest_kernel.sh, as integralpilot said debugging/output would be better/easier than when just running with linux.py directly
#asahi-dev · 2026-01-09 14:18
chaos_princessare you running linux at el1 or el2?
#asahi-dev · 2026-01-09 14:15
fl0_idI did also try to put some assembly in head.S, which had the intended result of rebooting, also when starting cpu 3 or 4, but not as that output is from m1n1, not sure if that is strictly related.
#asahi-dev · 2026-01-09 14:14
fl0_idf.e. I put some printk(KERNEL_EMERG "Some text") into start_kernel() and I assume these should come through on the second serial port. (they don't) The do seem to do something, just not sure what. (it seemed to crash/reboot earlier than ususal, but then that also happened 2-3 times with the same setup as before (exception somewhere on starting cpu 3 or 4)
#asahi-dev · 2026-01-09 14:12
fl0_idhey. does anybody have a good link about early boot kernel debugging? reading kernel code and blog posts etc that I could find were ok. but still not sure if I'm doing things right
#asahi-dev · 2026-01-09 14:08
jannau"23 files changed, 2245 insertions(+)" usb type-c device tree changes sent
#asahi-dev · 2026-01-09 10:26
nickchanon a10x there's a 0x5533300a at 0x20c30c120 though - have not managed to bring that one up
#asahi-dev · 2026-01-09 09:08
nickchanso for a lot of stuff i still use the uart but the signal integrity at baudrate 1500000 is not good
#asahi-dev · 2026-01-09 09:07
nickchanI still don't have a dwc2 stack in m1n1 that doesn't break after a few transactions 🙃
#asahi-dev · 2026-01-09 09:06
nickchanI hate dwc2 with a passion
#asahi-dev · 2026-01-09 00:08
fl0_idjannau yeah that did it. thanks
#asahi-dev · 2026-01-08 23:54
jannautry disabling CONFIG_EFI_ZBOOT
#asahi-dev · 2026-01-08 23:50
fl0_idyeah, that's what I get for vmlinuz. there is just no Image.gz. Image is there, but that's uncompressed. I thought it should be there as I though I just used the asahi fedora config originally
#asahi-dev · 2026-01-08 23:42
jannaupossibly a kernel config option
#asahi-dev · 2026-01-08 23:41
jannau_typec/arch/arm64/boot/Image.gz: gzip compressed data, max compression, from Unix, original size modulo 2^32 20630016
#asahi-dev · 2026-01-08 23:41
jannaufile ${KBUILD_OUTPUT}/arch/arm64/boot/Image.gz

plus ›