#
✕ nick:chadmed
#asahi-dev · 2026-07-19 10:38
chadmedaudio and edid stuff might not work
#asahi-dev · 2026-07-19 10:37
chadmedmost stuff seems to work, didnt test dcpext thoroughly but it boots fine and hpd works
#asahi-dev · 2026-07-19 10:36
chadmedi have updated the m1n1 MR with the latest developments, and pushed a rough kernel branch to https://github.com/chadmed/linux/tree/dcp/14.8.3
#asahi-dev · 2026-07-19 10:36
chadmedok i am unfortunately out of time
#asahi-dev · 2026-07-19 09:46
chadmedoh it does NOT like surfaces disappearing for some reason
#asahi-dev · 2026-07-19 09:24
chadmedlet me try that, one sec
#asahi-dev · 2026-07-19 09:22
chadmedguess i'll be rebasing all of this on top of that surface selection thing i did :P
#asahi-dev · 2026-07-19 09:21
chadmedso uhhh
#asahi-dev · 2026-07-19 09:21
chadmedok the apple hole goes away if you use surface 0
#asahi-dev · 2026-07-19 09:08
chadmedeyy brightness works
#asahi-dev · 2026-07-19 08:35
chadmedlooks like it, theres something up with the first swap getting rejected due to a 0x0 rectangle even though the first swap should null/clear all surfaces
#asahi-dev · 2026-07-19 08:34
chadmedthere's a hole in my framebuffer :(
#asahi-dev · 2026-07-18 14:30
chadmedkernel stuff is odd. struct sizes dont match by a significant amount and my eyes are glazing over, so problem for when i get back to brisbane perhaps
#asahi-dev · 2026-07-18 14:29
chadmedhttps://github.com/AsahiLinux/m1n1/pull/630
#asahi-dev · 2026-07-18 13:38
chadmedwe inch closer and closer :P
#asahi-dev · 2026-07-18 13:38
chadmed[ 69.579020] apple-dcp 38bc00000.dcp: swap failed! status 4
#asahi-dev · 2026-07-17 06:51
chadmed(this was not required with 13.5 but dcp asks you to do it and will crash immediately if you dont now)
#asahi-dev · 2026-07-17 06:51
chadmednot with an interchange surface, not with a standard rgba surface
#asahi-dev · 2026-07-17 06:51
chadmedthe m1n1 ipc client is nice now and the driver gets to trying to create the "default framebuffer surface" and then just... cant do that for some reason
#asahi-dev · 2026-07-17 06:50
chadmedfor this i may just have to push the mess ive already got and see if we cant snipe anyone else into looking at it which is unfortunate
#asahi-dev · 2026-07-17 06:49
chadmedsven: yknow id love to but unfortunately today and tomorrow are it and im back to work :(
#asahi-dev · 2026-07-17 06:47
chadmedand we're back to "the trace is identical why are you crashing" except slightly further along now :P
#asahi-dev · 2026-07-17 00:11
chadmed[ 3.208160] apple-dcp 38bc00000.dcp: DCP booted
#asahi-dev · 2026-07-17 00:09
chadmedwrong callback
#asahi-dev · 2026-07-17 00:09
chadmedHAHAHAHAHAHAHAHAHAa
#asahi-dev · 2026-07-16 14:21
chadmedi may be cooked if i cant figure it out in the time i have tomorrow though as im back to work next week :(
#asahi-dev · 2026-07-16 14:20
chadmedif however the issue with start_boot is what i now think it could be, we may not have to care
#asahi-dev · 2026-07-16 14:19
chadmedim starting everything xnu starts in the order it starts them
#asahi-dev · 2026-07-16 14:13
chadmedand i cannot figure out why because there is no material diff between the linux trace and xnu trace up to that point
#asahi-dev · 2026-07-16 14:12
chadmedproblem is on the linux side the message it sends after you do the notify ack is completely different to the one it sends xnu
#asahi-dev · 2026-07-16 14:12
chadmedafter xnu acks the notify by echoing it back, sac then responds with what looks like its current state or something like that, then xnu pushes the state that it wants
#asahi-dev · 2026-07-16 14:11
chadmedwe do all that fine
#asahi-dev · 2026-07-16 14:11
chadmedso with the sac endpoint, you start it with an empty command, which it then responds to with a NOTIFY when you then echo back to it
#asahi-dev · 2026-07-16 14:03
chadmedwhich.... huh?
#asahi-dev · 2026-07-16 14:03
chadmedthe issue is after you start it and ack the first notify, the responses are completely different to what it says to xnu
#asahi-dev · 2026-07-16 14:03
chadmedno theyre garbled there too in the exact same way
#asahi-dev · 2026-07-16 14:01
chadmedi am hoping we dont have to care about sac because its reports are all garbled and dont look like proper epic calls apart from the magic in the header
#asahi-dev · 2026-07-16 14:00
chadmedif we have to start more epic services then the driver's gonna need a bit of a refactor... again....
#asahi-dev · 2026-07-16 13:59
chadmedsome sort of clocking/state thing but this may not have been our issue
#asahi-dev · 2026-07-16 13:56
chadmedooh actually two more tricks
#asahi-dev · 2026-07-16 13:42
chadmedi have one more trick to try tomorrow and i may have to admit defeat :(
#asahi-dev · 2026-07-16 03:41
chadmedand im just about out of time :(
#asahi-dev · 2026-07-16 03:41
chadmed????
#asahi-dev · 2026-07-16 03:41
chadmedack the notify with the exact same thing as xnu, get different results
#asahi-dev · 2026-07-16 03:40
chadmedgreat the sac controller replies with completely different stuff (nonsense) when booted by linux
#asahi-dev · 2026-07-15 02:18
chadmedthe linux driver's trace looks exactly the same as the macos one and dcp still crashes :(
#asahi-dev · 2026-07-14 13:45
chadmedstill not sure about a few of the callbacks and i dont fancy running a certian java-based gui application on a high dpi laptop
#asahi-dev · 2026-07-14 13:44
chadmedhttps://github.com/chadmed/m1n1/tree/dcp/14.8.3 if anyone's curious
#asahi-dev · 2026-07-14 13:38
chadmedso we need to target a fw version that can be traced on m3
#asahi-dev · 2026-07-14 13:38
chadmedwell i dont need it to trace dcp on j314, but whoever does gpu will need to trace on an m3 machine so yes

plus ›