#asahi-dev · 2026-05-17 08:22
yukaand still get output on endpoint 4#asahi-dev · 2026-05-17 08:22
yukaI reduced the number of commands to just 2#asahi-dev · 2026-05-17 08:22
yukaHmm are you sure that's required to get serial?#asahi-dev · 2026-05-17 08:21
svenand the m1 address comes from somewhere inside par(seq=0, portal=0, offset=13, addr=0, len_=0x10000) but when i used the address that m3 returned it didn't work#asahi-dev · 2026-05-17 08:21
svena = par(seq=0x1337, portal=KISPortal.PAM, offset=0, addr=0x23d134014, len_=4) <-- iirc that address was different on m3#asahi-dev · 2026-05-17 08:19
svenhttps://f.svpe.de/e4731ec11516d79b9c675329ebf55b38134a22d30fa2251318b01867e9e61aee_enter-kis.patch i think that's what i used to enter debugusb from m1n1#asahi-dev · 2026-05-17 08:18
svenhrm, m3 had a different address in the PCR or whatever thing#asahi-dev · 2026-05-17 08:18
yukasven: for M1 Pro, M4 and A18 Pro, the same commands worked, only that M1 Pro needed different usb endpoint ids#asahi-dev · 2026-05-17 08:17
yukaAs in I put \x05\x05\x05\x05\x05 into the cu.kis-...0 device and on wire I only see 4 bytes#asahi-dev · 2026-05-17 08:16
svendid you figure out where the addresses you have to read from come from? i thought they were in that initial 0x8000 or whatever block but those didn't work for m3 then#asahi-dev · 2026-05-17 08:16
yukanot only that, it also cuts off longer messages which are not multiples of 32-bit#asahi-dev · 2026-05-17 08:15
svenyeah, the macos driver is a bit weird and you can it stuck very easily#asahi-dev · 2026-05-17 08:14
yukasome time today I will be able to check, once get kisd into a state usable for this#asahi-dev · 2026-05-17 08:13
yukayes but I think it might be due to the mac os kis/debugusb driver#asahi-dev · 2026-05-17 08:13
flokliMmh, debugusb chainloading still is unreliable on the m1n1 side, no?#asahi-dev · 2026-05-17 08:10
yukaIs the question whether this is at all interesting now with debugusb#asahi-dev · 2026-05-17 08:09
jannauyuka: no, not sure whether it's that valueable. it's based on your+lisa's splitting. the command and output changes are intended for easier CI integration#asahi-dev · 2026-05-17 08:08
jannauyes, I dropped the the other two changes from v2#asahi-dev · 2026-05-17 08:02
sveni.e. pick up power, pwm and arm#asahi-dev · 2026-05-17 08:02
sveni just have to skip watchdog, don't i?#asahi-dev · 2026-05-17 08:01
jannausven: yes, no known/comments issues. be careful about which dt-bindings patches to pick#asahi-dev · 2026-05-17 07:59
svenjannau: can i pick up v3 of the m3 device tree series?#asahi-dev · 2026-05-17 07:51
yukathe one with a shell prompt where I could type "reboot"#asahi-dev · 2026-05-17 07:50
yukajannau you showed me a new cs-sw, is this public anywhere?#asahi-dev · 2026-05-17 02:13
StanfordAppliedCyber[m]we take that#asahi-dev · 2026-05-17 02:13
StanfordAppliedCyber[m]i tossed the 26.4.1 container, made a new volume, threw 26.5 in it, installed the exact same m1n1 image into that container, boom, smp works#asahi-dev · 2026-05-17 02:12
StanfordAppliedCyber[m]literally nothing else#asahi-dev · 2026-05-17 02:12
StanfordAppliedCyber[m]yes#asahi-dev · 2026-05-17 02:07
nicolas17wait what fixed the smp issues? running your code by 26.5 iBoot?#asahi-dev · 2026-05-17 01:52
StanfordAppliedCyber[m]it looks like 26.5 changed a bunch of sysreg behavior-- looks like a lot more are locked by iboot now#asahi-dev · 2026-05-17 00:53
StanfordAppliedCyber[m]hypervisor development continuing on 26.5 in that case#asahi-dev · 2026-05-17 00:53
StanfordAppliedCyber[m]I shall now say a prayer to steve jobs#asahi-dev · 2026-05-17 00:52
StanfordAppliedCyber[m]:)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))#asahi-dev · 2026-05-17 00:52
StanfordAppliedCyber[m]MACOS 26.5 FIXED MY SMP ISSUES#asahi-dev · 2026-05-16 21:48
tryhard[m]in asc mmio#asahi-dev · 2026-05-16 21:47
tryhard[m]which index is OUTBOX_CTRL?#asahi-dev · 2026-05-16 19:17
flokli🎉#asahi-dev · 2026-05-16 17:31
yukaI was able to make the macbook neo spit out its boot logs including m1n1 dockchannel uart to my Linux host#asahi-dev · 2026-05-16 17:25
yukaSome success with debugusb on linux#asahi-dev · 2026-05-16 14:01
yukaFYI I tested the new 26.5 mBoot on t8132 and both SYS_IMP_APL_VM_TMR_FIQ_ENA_EL2 and deep wfi behave the same as before#asahi-dev · 2026-05-16 09:19
fl0_id__thx#asahi-dev · 2026-05-16 09:08
JamesCalligeros[m]Trap the condition you want in a tracer and then do hv.run_shell()#asahi-dev · 2026-05-16 09:07
fl0_id__is there a way to stop run_guest at a later time automatically? With -S it drops to a hv shell immediately which is too early. Would be awesome to do that after loading the kernel so I don't have to spam ctrl+c#asahi-dev · 2026-05-16 03:58
chadmeddcp doesnt set anything that is obviously HDR static metadata... this is slightly worrying #asahi-dev · 2026-05-16 03:00
StanfordAppliedCyber[m]anyways, i must be going, thanks for the assistance!#asahi-dev · 2026-05-16 03:00
StanfordAppliedCyber[m]if i can't --> that means there's something we dont understand because a vanilla macos kernel with sptm enabled works 100% fine#asahi-dev · 2026-05-16 02:59
StanfordAppliedCyber[m]if i can --> there's a chance i cna't fix this myself because sptm is involved#asahi-dev · 2026-05-16 02:58
StanfordAppliedCyber[m]i feel the last resort method is to figure out how sptm does this and see if i can make it happen from inside macho boot mode from m1n1#asahi-dev · 2026-05-16 02:58
StanfordAppliedCyber[m]idk, very puzzling, endlessly mysterious these machines#asahi-dev · 2026-05-16 02:57
StanfordAppliedCyber[m]they did not