#
✕ nick:Shiz
#asahi-dev · 2026-01-20 09:25
Shiznickchan: right, that tracks
#asahi-dev · 2026-01-20 08:42
Shizthe xnu source straight up sets the other bits to 0 too, wonder if that's meaningful
#asahi-dev · 2026-01-20 08:41
Shizright, okay
#asahi-dev · 2026-01-20 08:16
Shizsurely one of them is related to the architectural timer clock scaling
#asahi-dev · 2026-01-20 08:16
Shiznickchan: do we know what bit[1] does? :P
#asahi-dev · 2026-01-20 08:16
Shizyeah i was wondering, "slunds like you're describing smp" :p
#asahi-dev · 2026-01-18 22:01
Shizraise*
#asahi-dev · 2026-01-18 13:31
Shiz-rw------- 1 io staff 13 18 jan. 14:30 /Users/io/.m1n1-history
#asahi-dev · 2026-01-18 13:31
Shizoverriding readline with the `gnureadline` modules 'fixes' it but uhh
#asahi-dev · 2026-01-18 13:31
Shizhas anyone gotten similar readline errors like this? https://txt.shiz.me/YzhmMDdkOT
#asahi-dev · 2026-01-18 12:10
Shizso if that's the alternative give me MLs
#asahi-dev · 2026-01-18 12:10
Shizgitlab definitely takes longer than two weeks to load any page
#asahi-dev · 2026-01-17 20:46
ShizI can test it tomorrow
#asahi-dev · 2026-01-17 18:57
Shizdefinitely, can't hurt to take it one at a time :)
#asahi-dev · 2026-01-17 18:53
Shizi2cdev could be one backend, the IOKit HPM interface for macOS could be another
#asahi-dev · 2026-01-17 18:53
ShizBjarne[m]: probably some kind of standard 'backend API' (rust trait) along with a way to enumerate them
#asahi-dev · 2026-01-17 18:36
Shiz:(
#asahi-dev · 2026-01-17 18:28
Shizor derp, is this from when you're already booted into the usb
#asahi-dev · 2026-01-17 18:27
Shizhave you tried using Mist?
#asahi-dev · 2026-01-17 18:27
Shizhmm
#asahi-dev · 2026-01-17 14:40
Shizm3 max does have 3 clusters
#asahi-dev · 2026-01-17 14:38
Shizcluster 0 is E, cluster 1 is P
#asahi-dev · 2026-01-17 14:37
Shizactually, can just check the devtree -- yep, two clusters
#asahi-dev · 2026-01-17 14:36
Shiz(xnu will crash early if not at least one core in all clusters is activated, so easy to check)
#asahi-dev · 2026-01-17 14:36
ShizI wanna say yes, but I'll verify tomorrow :P
#asahi-dev · 2026-01-17 14:35
Shizwhich of course also have board IDs :)
#asahi-dev · 2026-01-17 14:35
ShizI found myself diving into weird apple hardware for a bit and found out about something called cloudOS which they run on their private compute nodes...
#asahi-dev · 2026-01-17 14:35
Shizm3 max/ultra i guess
#asahi-dev · 2026-01-17 14:34
Shizahh ;p
#asahi-dev · 2026-01-17 14:31
Shiz(one P-core, one E-core)
#asahi-dev · 2026-01-17 14:31
Shizunless I misunderstood, that is what we did
#asahi-dev · 2026-01-17 14:31
Shiz nickchan : maybe check if it works with 1 core per cluster
#asahi-dev · 2026-01-17 12:43
Shizoh, I thought that was what the micro-usb on the actual CS board was for :x
#asahi-dev · 2026-01-17 05:20
Shiz(no shade of course, you can never foresee the entire future)
#asahi-dev · 2026-01-17 05:19
Shizdirectly or indirectly
#asahi-dev · 2026-01-17 05:19
Shizat least half of it seems to come down to "whoops the m3 has a very different memory/mmio map and maybe m1n1 was making a few assumptions related to how m1/m2 were mapping"
#asahi-dev · 2026-01-17 05:16
Shizmessy but at least a good base for further research :D
#asahi-dev · 2026-01-17 05:16
Shizit seems `self.add_tracer(zone, "CPU_START", TraceMode.RESERVED)` is also messed up, needs to be commented-out
#asahi-dev · 2026-01-17 05:16
Shizomitting any of those seems to make it return to unstable boot, so there's def still work to be done :p (mostly the last one)
#asahi-dev · 2026-01-17 05:16
Shiz3) booting with only one P-core and only one E-core enabled
#asahi-dev · 2026-01-17 05:15
Shiz2) adding some MSRs and marking some more memory as reserved using the memory-map ADT node
#asahi-dev · 2026-01-17 05:15
Shiz1) the PMGR power-gating hacks were definitely poking addresses they shouldn't be (and entirely differently mapped on M3+)
#asahi-dev · 2026-01-17 05:14
ShizI think it was a combination of three things:
#asahi-dev · 2026-01-17 05:12
Shizmuch thanks to your earlier efforts too :)
#asahi-dev · 2026-01-16 21:01
Shiznop sled? try AAAA-sled
#asahi-dev · 2026-01-16 21:01
Shiz(but it does work without)
#asahi-dev · 2026-01-16 21:01
Shizat this point the padding is fully placebo and thus makes us feel better
#asahi-dev · 2026-01-16 20:37
Shizthere may or may not be still weird memory corruption
#asahi-dev · 2026-01-16 20:37
Shizwell
#asahi-dev · 2026-01-16 20:37
Shiz\o/

plus ›