#asahi-dev · 2026-01-06 11:01
svenyeah, and really give it a day or two before submitting#asahi-dev · 2026-01-06 11:01
integralpilotI made DT bindings, probably a good idea to post them here first thanks#asahi-dev · 2026-01-06 11:00
chadmedim ~80% sure the mute/unmute popping and undrained buffer thing can be fixed with proper operation sequencing#asahi-dev · 2026-01-06 10:59
chadmedand we still get spurious OCP interrupts and other dodgy things #asahi-dev · 2026-01-06 10:59
chadmedthe RE was incomplete and there are lots and lots of 'UNKx' all over the place#asahi-dev · 2026-01-06 10:58
chadmedthe kernel drivers do need some love though so id rather attention be focused there#asahi-dev · 2026-01-06 10:58
chadmedapple cant really claim IP over the concept of software eq#asahi-dev · 2026-01-06 10:57
sven*give it#asahi-dev · 2026-01-06 10:57
chadmednor do i have any intention to _let_ anyone do that#asahi-dev · 2026-01-06 10:57
sveni'd strongly recommend to post them here and give a day or two before submitting to lkml#asahi-dev · 2026-01-06 10:57
chadmedand in terms of looking at coreaudio, i dont think even looking at userspace is off limits at this point because we're not copying apples algorithms or filter chains#asahi-dev · 2026-01-06 10:57
integralpilotoh if I might be able to help with audio that’s great, yeah mainly core audio and not kernel side, just want to be careful#asahi-dev · 2026-01-06 10:57
svenyeah, dt-bindings take a while to get used to. i'm sure you can do them but there's a lot of pitfalls to avoid#asahi-dev · 2026-01-06 10:56
chadmedim pretty sure theres a codec driver for them now, but no userspace integration #asahi-dev · 2026-01-06 10:56
chaos_princessright, for dt-bindings post here before going to mailing list#asahi-dev · 2026-01-06 10:56
chaos_princesschadmed: do we support audio on them? i thought we didn't#asahi-dev · 2026-01-06 10:56
sventhese are, uh, a steep learning curve#asahi-dev · 2026-01-06 10:55
svenintegralpilot: do you have dt-bindings for mtp?#asahi-dev · 2026-01-06 10:55
chadmedactually no we have three types of speaker codec, i keep forgetting about the damn imacs#asahi-dev · 2026-01-06 10:55
chadmedand then we have two types of speaker codec and two types of headphone codec#asahi-dev · 2026-01-06 10:55
chadmedwe have MCA which ties the PCM frontends to I2S ports#asahi-dev · 2026-01-06 10:54
chaos_princessif you only looked at coreaudio, you should be fine with touching kernel stuff. what you shouldn't probably touch is the pipewire filter chain#asahi-dev · 2026-01-06 10:54
chadmedwe have the platform driver which instantiates the PCMs and handles all the DPCM routing #asahi-dev · 2026-01-06 10:54
integralpilotoh good luck with that!#asahi-dev · 2026-01-06 10:53
integralpilothave done a lot of static analysis of coreaudiod and stuff so probably tainted#asahi-dev · 2026-01-06 10:53
chadmedbut all of this stuff is split across 6 drivers and i cant figure out where we're supposed to do that :P #asahi-dev · 2026-01-06 10:53
chadmednow the real fix is to simply fix the sequencing and mute the codecs _before_ we try to free the hardware and gate the clocks #asahi-dev · 2026-01-06 10:53
integralpilotI’d love to touch audio stuff seems super interested but don’t think I can cause of my macOS CVEs in that space :(#asahi-dev · 2026-01-06 10:52
chadmedtheres one patch in particular, where we move the codec clock gating to some place its not really supposed to be to work around a sequencing issue#asahi-dev · 2026-01-06 10:52
chadmedim looking at whats left of the audio stuff and oh boy #asahi-dev · 2026-01-06 10:51
integralpilotspeaking of that, MTP is almost ready, i should be all to send it off before I go to bed tonight#asahi-dev · 2026-01-06 10:51
chadmedyup, one bite at a time etc#asahi-dev · 2026-01-06 10:51
integralpilotI guess we’re all doing a bit (and some people a lot, thank you!) and more bits are going upstream#asahi-dev · 2026-01-06 10:51
integralpilotthat’s really good!!#asahi-dev · 2026-01-06 09:57
chadmedwe're down to 81 kLOC changed across 816 patches#asahi-dev · 2026-01-06 09:49
sveninterestingly, tbt seems very reliable even as coldplug :D#asahi-dev · 2026-01-06 09:46
svenSounds about right #asahi-dev · 2026-01-06 09:46
chadmedon m2 max studio#asahi-dev · 2026-01-06 09:46
chadmedit is intermittent, probably ~35% of boots#asahi-dev · 2026-01-06 09:45
jannauyes, same here#asahi-dev · 2026-01-06 09:44
svenI can only sometimes reproduce it here #asahi-dev · 2026-01-06 09:44
svenNot yet#asahi-dev · 2026-01-06 09:44
jannausven: have you started looking into the unreliable coldpug / initial hotplug of the usb-c ports in asahi-wip?#asahi-dev · 2026-01-06 09:39
jannauit's unlikely that asahi kernel driver just works. see the amount of fw version and hw variant dependent code in the driver. so if the driver loaded and tried to use the gpu I would expect that the firmware crashes immediately#asahi-dev · 2026-01-06 08:27
svenyeah :)#asahi-dev · 2026-01-06 08:26
integralpilotsorry pressed enter too soon#asahi-dev · 2026-01-06 08:26
integralpilotdef merge it then, thank you!#asahi-dev · 2026-01-06 08:26
integralpilotoh we should#asahi-dev · 2026-01-06 08:26
svenall the others have the logic in their driver like NVMe or SMC do#asahi-dev · 2026-01-06 08:25
svenjust mtp