#asahi-dev · 2026-06-27 08:01
JamesCalligeros[m]forgor to push the notes, there should be some brainfart notes in there now#asahi-dev · 2026-06-27 08:00
JamesCalligeros[m]i dont think we need one right now but if we push more stuff into firmware it could be useful#asahi-dev · 2026-06-27 07:58
JamesCalligeros[m]basically i just think reprogramming the decoders on every slice is cpu grunt work that we shouldnt really be waking up the big app cores to do#asahi-dev · 2026-06-27 07:58
JamesCalligeros[m]but the latter can just be iovas to something on host memory probably#asahi-dev · 2026-06-27 07:57
JamesCalligeros[m]the cm3 would only need to keep track of in-flight streams and the state of each decoder and maaaybe the dpb for each stream#asahi-dev · 2026-06-27 07:57
JamesCalligeros[m]https://github.com/chadmed/avd-fw/tree/work/interface i put some notes and the crap logger here, but im sure you have better ideas :P#asahi-dev · 2026-06-27 07:56
JamesCalligeros[m]yeah... it chokes easily and i think the cm3 is better poised to track streams in progress and dispatch arriving frames to the decoder units#asahi-dev · 2026-06-27 07:54
JamesCalligeros[m]anyways i wrote a terrible janked up syslog thingy for the firmware and it "works" in the sense that it will write a log to a ringbuffer and kick the application core to read it, but its awful and idk where in sram we can put it without exploding stuff the driver is doing #asahi-dev · 2026-06-27 07:53
JamesCalligeros[m]they come from -bad #asahi-dev · 2026-06-27 07:53
JamesCalligeros[m]the stateless decoders dont come from that#asahi-dev · 2026-06-27 07:52
JamesCalligeros[m]most of the garbage has to move to firmware anyway so we can support multiple decode streams at once, so its going to happen regardless#asahi-dev · 2026-06-27 07:52
JamesCalligeros[m]including gentoo, which i guess could be fixed with a patch to gst-plugins-bad ebuild, but i think its one of those "do not build this it has external deps" ones#asahi-dev · 2026-06-27 07:52
JamesCalligeros[m]yeah except some distros dont build v4l2codecs#asahi-dev · 2026-06-27 07:51
JamesCalligeros[m]meh doesnt matter right now, todays project has been brainstorming moving most of the garbage to firmware#asahi-dev · 2026-06-27 07:50
JamesCalligeros[m]apparently it works with stateless if you have the ffmpeg request api patches applied, but i could not get it to work at all #asahi-dev · 2026-06-27 07:48
JamesCalligeros[m]"oh yeah just install gst-plugins-bad but make sure your distro builds it with v4l2codecs so you get the v4l2slh264dec component and make sure your pipeline uses it and...." #asahi-dev · 2026-06-27 07:47
JamesCalligeros[m]gstreamer worked fine, my point was more that we cannot ship something that requires users to do some magic pipeline incantation just to have working video decode.#asahi-dev · 2026-06-27 07:43
JamesCalligeros[m]hasnt worked at all in the past week of testing on any of my devices#asahi-dev · 2026-06-27 07:36
JamesCalligeros[m]this is not a situation that we alone can fix, so the path of least resistance is to just do what everyone else already supports, which is vaapi or increasingly vk video#asahi-dev · 2026-06-27 07:35
JamesCalligeros[m]and even though mpv uses it "well" it still wont actually pick it unless you force it to with -hwdec, which is also a problem because it wont automatically pick dmabuf-wayland as the output driver either, so you dont even get zerocopy by default #asahi-dev · 2026-06-27 07:34
JamesCalligeros[m]yes, external patches.... that have not been merged.... and nothing that consumes ffmpeg other than mpv effectively uses....#asahi-dev · 2026-06-27 07:32
JamesCalligeros[m]firefox requires a "proper" v4l2-m2m, not request aiui#asahi-dev · 2026-06-27 07:30
JamesCalligeros[m]everything on the desktop assumes va-api or whatever the nvidia one is is present, and stuff that doesnt is going to very quickly move to vk video. look at how quick firefox was to adopt it vs v4l2.#asahi-dev · 2026-06-27 07:29
JamesCalligeros[m]the v4l2 request api is not well supported by anything. ffmpeg requires patches that have not yet been merged (and it still doesnt handle it properly), firefox is not the be-all-end-all of video decode software, and no one is interested in constructing gstreamer pipelines by hand for desktop use#asahi-dev · 2026-06-26 11:42
JamesCalligeros[m](your invitation to install gentoo :p)#asahi-dev · 2026-06-26 11:42
JamesCalligeros[m]The FAR repos may not yet have the latest#asahi-dev · 2026-06-26 11:42
JamesCalligeros[m]Actually no it’s there. Can you install from HEAD?#asahi-dev · 2026-06-26 11:41
JamesCalligeros[m]Lol whoops#asahi-dev · 2026-06-26 11:40
JamesCalligeros[m]alsa is useless on these machines, dont use it directly. Use pipewire/wireplumber only#asahi-dev · 2026-06-26 11:40
JamesCalligeros[m]Yeah one PCM is the speakers and the other is V/ISENSE#asahi-dev · 2026-06-26 11:39
JamesCalligeros[m]Which output do you get? Only AppleJ6xx or do you see the virtual pipewire sink too?#asahi-dev · 2026-06-25 09:01
JamesCalligeros[m]are the two interrupt lines to aic just mbox0 not empty and mbox1 not empty?#asahi-dev · 2026-06-25 08:34
JamesCalligeros[m](or any of the other cm3 emulators)#asahi-dev · 2026-06-25 08:34
JamesCalligeros[m]sofus: can you chuck the apple firmware in qemu and see what it does differently for a given input?#asahi-dev · 2026-06-25 07:44
JamesCalligeros[m]im out rn but i can test later tonight or tomorrow#asahi-dev · 2026-06-24 11:19
JamesCalligeros[m]6.6k*#asahi-dev · 2026-06-24 11:17
JamesCalligeros[m]Oh my 66kloc of slop#asahi-dev · 2026-06-24 09:58
JamesCalligeros[m]No more dcp work until the surface caps stuff lands anyway, i have too many dcp-related m1n1 and kernel branches#asahi-dev · 2026-06-24 08:26
JamesCalligeros[m]And id still spin our own because pulling apple’s out of the kext will suck#asahi-dev · 2026-06-24 08:26
JamesCalligeros[m]Ideally the cm3 runs fw that exposes a unified interface to linux (or at least an interface for each codec) and we deal with instruction dispatch to the ffu on the cm3 i agree#asahi-dev · 2026-06-24 06:21
JamesCalligeros[m]And then colourspaces when those land in the api#asahi-dev · 2026-06-24 06:20
JamesCalligeros[m]Either way we need to hook up the eotfs#asahi-dev · 2026-06-24 06:20
JamesCalligeros[m]I need to have another crack at colour pipelines. I think iosurface might have space for a per-surface clut but im not sure#asahi-dev · 2026-06-24 06:07
JamesCalligeros[m]Upstream will likely be fine with it if there’s a decent justification. Having to handle an irq on every single byte if we use mailbox is a fine justification.#asahi-dev · 2026-06-24 06:01
JamesCalligeros[m]They dont need to share infrastructure, it would just be Proper if they do. Upstream might grumble about it though.#asahi-dev · 2026-06-24 05:52
JamesCalligeros[m]I dont think the markdown stuff works from the irc bridge at all lmao#asahi-dev · 2026-06-24 05:38
JamesCalligeros[m]if mailbox isnt gonna fly we can do what we did for rtk mailbox and just make it a not-mailbox mailbox driver :P#asahi-dev · 2026-06-24 05:37
JamesCalligeros[m]yuka wanted to work on a serial driver for dockchannel like nowish#asahi-dev · 2026-06-22 12:42
JamesCalligeros[m]night light is different and uses the crtc-level colour transformation matrix. the per-plane colour operation pipeline api is very new. i had a branch with it implemented lying around somewhere but the api still doesnt have the features we support (amd wrote it for their display controllers and intel extended it for theirs)#asahi-dev · 2026-06-22 11:52
JamesCalligeros[m]avd can share nv12 dmabufs with the gpu with no additional changes though so thats nice. not sure if we're doing interchange buffers on avd though