#asahi-dev · 2026-10-03 03:57
chadmedwe will not be undoing the licence change for new code however just to be very clear. new code will not be dual-licenced.#asahi-dev · 2026-10-03 03:53
chadmedand no you do not need to get rid of the fuchsia thing. our intention is _not_ to revoke MIT for existing code. it is silly to presume that we would do that#asahi-dev · 2026-10-03 03:53
chadmedwe are going to put the licence text back yes but no the spdx licence identifier does not need to specify dual licence. it is obvious which code was checked in under mit and which will not be#asahi-dev · 2026-10-03 00:55
chadmedyessss lets go#asahi-dev · 2026-10-03 00:34
chadmedmaybe i can just refactor the called function so that it only acquires the semaphore when called from the workqueue #asahi-dev · 2026-10-03 00:34
chadmedhmm it seems to be deadlocking when called from slk_put/get#asahi-dev · 2026-10-03 00:04
chadmedso even nonblocking causes deadlocks if i take the rw semaphore from the volume limit workqueues...#asahi-dev · 2026-10-03 00:04
chadmedfrom the alsa crate docs: "Open does not support async mode (it’s not very Rustic anyway)"#asahi-dev · 2026-10-01 11:32
chadmedi might have time to look into it soon but it's gonna be mesa regardless in that case#asahi-dev · 2026-10-01 11:05
chadmedand it looks fine when compositing is bypassed and its scanned out directly to a dcp surface right#asahi-dev · 2026-09-30 20:19
chadmedprobably worth just shelving the interchange stuff until we sort that out otherwise there will be issues when the compositor decides to do gpu compositing#asahi-dev · 2026-09-30 20:19
chadmedsofus: do you have the patched interchange mesa installed#asahi-dev · 2026-09-27 12:37
chadmedwe cant just move fast and break things all the time, thats how we end up with enterprise codebases#asahi-dev · 2026-09-27 08:36
chadmedits worked with no issues for like 5 years now#asahi-dev · 2026-09-27 08:36
chadmedi dont know if we should change how speakersafetyd works (i dont think we even can?) or just justify the relaxed locking#asahi-dev · 2026-09-27 08:35
chadmedauuughhh another deadlock when stopping pipewire/wireplumber#asahi-dev · 2026-09-26 12:41
chadmedhopefully a few of the other things mark has nacked previously also turn out to be unnecessary :P#asahi-dev · 2026-09-26 12:40
chadmedsome of the macaudio patches were rather superfluous as it turns out. not sure why we needed to do the flakey wildcard matching thing when strstr exists#asahi-dev · 2026-09-26 12:03
chadmedmmediouni[m]: people are complaining about that because its no longer device-local, cant win :P #asahi-dev · 2026-09-26 12:00
chadmedthen again if i were a PM at apple i would say it seems unreasonable for the OS to eat 60 GB of storage so that you can make siri into a dommy mummy but whatever what do i know my company isnt worth 6 trillion dollars#asahi-dev · 2026-09-26 11:59
chadmedit seems unreasonable to expect users to do all this #asahi-dev · 2026-09-26 11:56
chadmed"AI has made everything 500% more expensive so we're cutting back on our ram and storage also you have to dedicate most of whats left to AI" #asahi-dev · 2026-09-26 11:56
chadmedsend the invoice directly to john apple #asahi-dev · 2026-09-26 10:46
chadmedwhat the hell have they done :/ #asahi-dev · 2026-09-26 10:36
chadmedlets go macaudio v2 works on every class of device#asahi-dev · 2026-09-25 13:03
chadmedfrom what i can see the format should describe iomfb's 420p#asahi-dev · 2026-09-25 12:58
chadmed(that it would be p030)#asahi-dev · 2026-09-25 12:58
chadmedthats what i thought#asahi-dev · 2026-09-25 12:41
chadmeddoes it output correctly #asahi-dev · 2026-09-24 12:56
chadmedyou can look at the oss xnu for facts about the hardware, which are not copyrightable#asahi-dev · 2026-09-24 12:41
chadmedwhich is why lscpu and kde system info will tell you which soc the core is on#asahi-dev · 2026-09-24 12:40
chadmednot at any user-facing level but they had different chickens and core IDs #asahi-dev · 2026-09-24 12:40
chadmedtbf this was the case even for m1#asahi-dev · 2026-09-24 12:36
chadmedthese were all decisions that have to have been made ~3 years ago before everything became expensive and on fire and evil#asahi-dev · 2026-09-24 12:35
chadmedi feel like theyll converge again #asahi-dev · 2026-09-24 12:33
chadmedor even just M4 Series Feature Support (incl. A18 Pro) #asahi-dev · 2026-09-24 12:33
chadmedA18 & M4 Series Feature Support or something like that#asahi-dev · 2026-09-24 12:33
chadmed^ #asahi-dev · 2026-09-24 12:29
chadmedthis is pretty much the only (major) issue ive encountered and like i said it was just because i was doing stuff we would never ship anyway#asahi-dev · 2026-09-24 12:28
chadmedsaying that it will crash if a strong breeze is blowing or the planets are orbiting a certain way is saying that it's extremely flakey and unreliable and fussy#asahi-dev · 2026-09-24 12:27
chadmedlmao sorry im being cheeky, im saying that dcp is extremely temperamental and will crash if something is even slightly outside of what it expects#asahi-dev · 2026-09-24 12:24
chadmedweston wont put a cursor on an overlay plane so you need to explicitly expose the plane as a cursor plane (which we dont do in production) #asahi-dev · 2026-09-24 12:24
chadmedno i havent opened an issue for it because its a silly edge case#asahi-dev · 2026-09-24 12:23
chadmedkwin at least will immediately fall back to software compositing if it tries to do something dcp wouldnt like (e.g. putting a cursor too close to the edge of the screen lmao this display controller suuucks) #asahi-dev · 2026-09-24 12:23
chadmedndufresne: Weston is not suitable for us as it does not gracefully handle us rejecting swaps when the breeze is blowing or mercury is in retrograde#asahi-dev · 2026-09-24 11:40
chadmedkwin for example still has issues with direct scanout on machines where the gpu and display controller are not the same thing, as it treats this as a multi-gpu setup#asahi-dev · 2026-09-24 11:39
chadmedit is still useful for testing stuff like this without the interference of a wayland compositor doing unexpected things#asahi-dev · 2026-09-24 11:21
chadmedbut we need to negotiate the correct pixel format and drm format modifier to make both the codec and display controller work properly so we cant rely on create_dumb and kludging things in the kms driver#asahi-dev · 2026-09-24 11:20
chadmedthe tricky thing is that interchange/agx are similar to afbc in this regard. software is supposed to treat them as opaque/single-plane formats but the interface to the display controller requires the compression buffer metadata to be programmed in by us#asahi-dev · 2026-09-23 06:53
chadmedmykelmoss: if you read the logs, yes a couple of people have contacted me about working on it but none have seen it though to completion