#asahi-dev · 2026-07-25 01:16
chadmedthe next target is 14.8.5#asahi-dev · 2026-07-24 09:00
chadmedthanks for fixing#asahi-dev · 2026-07-24 09:00
chadmedyuka: yeah it should, C version never did and I overlooked "fixing" it as i was more concerned about the load-bearing UB in another part of the offset stuff :P#asahi-dev · 2026-07-24 08:58
chadmedoh makes sense#asahi-dev · 2026-07-24 08:58
chadmedwhat panicked#asahi-dev · 2026-07-23 11:26
chadmedsofus: its the headphone jack codec#asahi-dev · 2026-07-22 11:07
chadmedi wonder if theyre just doing something stupid like swapping a transparent framebuffer into the surface before disabling swaps for it#asahi-dev · 2026-07-22 11:07
chadmedthe only thing i can see that macos does differently to us on hiding a surface is setting a bunch of the flags, but i just tried copying the flags they set and the surface still sticks around#asahi-dev · 2026-07-22 09:19
chadmedand transcoding for jellyfin and friends#asahi-dev · 2026-07-22 09:19
chadmedalso useful for intermediates/rushes/previews for effects compositing shots and stuff like that#asahi-dev · 2026-07-21 13:36
chadmedmore useless than <NT 6.0 file copy ETA#asahi-dev · 2026-07-21 13:33
chadmedif you open the installer log and set it to max verbosity you can see that the ETA is completely made up as each step just returns -1 for the ETA #asahi-dev · 2026-07-21 13:33
chadmedit doesnt actually take that long, the ETA is completely useless#asahi-dev · 2026-07-21 13:24
chadmedive had the most success with an apple charging cable between two apple silicon macs. that setup hasnt failed me once yet. i recall using a random x86 host being a bit flaky at times but that was years ago #asahi-dev · 2026-07-21 13:20
chadmedyeah you can dfu it with the ipsw, but since youve got 14.6 already just bputil -f the apfs volume and use the installassistant#asahi-dev · 2026-07-21 13:18
chadmedgrab the 14.8.3 installassistant and its easy to upgrade#asahi-dev · 2026-07-21 13:18
chadmedplease use 14.8.3, the ABIs all changed with 14.7 #asahi-dev · 2026-07-21 13:17
chadmeddo you have it plugged in on boot to the correct port?#asahi-dev · 2026-07-21 13:16
chadmedor is this m1n1#asahi-dev · 2026-07-21 13:16
chadmedare you trying to use idevicerestore? do you have usbmuxd running? #asahi-dev · 2026-07-21 12:49
chadmedaudio service seems to start fine too with just the 13.5 commands. i will try plugging into a display with speakers tomorrow #asahi-dev · 2026-07-21 12:40
chadmedeverything seems to just work apart from surface clearing though, so im happy #asahi-dev · 2026-07-21 12:39
chadmedwe cannot ship this without gpu though it's horrible#asahi-dev · 2026-07-21 12:38
chadmedslightly less concerning than i thought yesterday :P #asahi-dev · 2026-07-21 12:37
chadmedso it thinks there should be something there when there shouldnt be#asahi-dev · 2026-07-21 12:37
chadmedhmm actually i think this is a red herring. we need to do something else to make dcp clear a surface now, and i think it's just still trying to read stale/dead surfaces. cursor images dont go away when theyre supposed to and the swap fails due to dcp trying to paint a null rectangle#asahi-dev · 2026-07-21 12:32
chadmedeven the swap complete callback is too soon? wtf? #asahi-dev · 2026-07-21 12:24
chadmedaaaaaaaa#asahi-dev · 2026-07-21 12:24
chadmedyeah thats what it is#asahi-dev · 2026-07-21 12:15
chadmedthe dcp crashes are iova read errors when a framebuffer disappears. i wonder if swapping has somehow gotten even slower and we need to keep the gem references around even longer now....#asahi-dev · 2026-07-21 10:36
chadmedanyone struggling with low mic volume is invited to tesst https://github.com/AsahiLinux/asahi-audio/tree/mic-fixes#asahi-dev · 2026-07-20 11:42
chadmedwe would not have to do or maintain anything else after that#asahi-dev · 2026-07-20 11:42
chadmedwhich is why i would very much like an actual vk video driver#asahi-dev · 2026-07-20 11:42
chadmedthis works in the opposite direction#asahi-dev · 2026-07-20 11:42
chadmedhttps://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31517#asahi-dev · 2026-07-20 11:38
chadmedyeah im not concerned about that, its the kind of workload the scheduler will force on to the ecores anyway#asahi-dev · 2026-07-20 11:35
chadmedits already shipped in fedora even though it's abandoned. we may need to fork it#asahi-dev · 2026-07-20 11:35
chadmedyeah, the bootlin one might be nicer idk#asahi-dev · 2026-07-20 11:35
chadmedand im sure some vaapi-to-vkv thing will end up in mesa or some such however we are then relying on two translation layers to DTRT #asahi-dev · 2026-07-20 11:33
chadmedbut vk video is where desktop video offload is going so it's gonna have to happen at some point#asahi-dev · 2026-07-20 11:33
chadmedits probably not as urgent, especially if the bridge can be made seamless and not janky#asahi-dev · 2026-07-20 11:31
chadmedsaves some cycles by having some of the stuff prefilled (e.g. statically allocate the ops for each variant, then once youve devm_kzalloced your driver data struct just avd->ops = &avd_v*_ops or whatever which is the usual pattern)#asahi-dev · 2026-07-20 11:29
chadmedyeah ultimately it does not _really_ matter as long as it's readable and reasonably easy to add new variants#asahi-dev · 2026-07-20 11:13
chadmedmy preference is typically toward generic/extensible solutions in most cases especially when we cannot know what horrors we will be subjected to by apple in the future#asahi-dev · 2026-07-20 11:12
chadmedmaybe just switch (avd->variant): case AVD_VARIANT_V2:... where necessary then?#asahi-dev · 2026-07-20 11:11
chadmednot necessarily, the approach can be extended to cover whatever#asahi-dev · 2026-07-20 11:07
chadmedavd->ops->generate_stream(...) where you then just set .generate_stream=avd_v*_generate_stream on probe#asahi-dev · 2026-07-20 11:07
chadmedi think i suggested that a while ago, and i think it makes the most sense and makes it easier to reason about the code and add new variants#asahi-dev · 2026-07-19 10:40
chadmedalso you will need to patch the gpu initdata stuff in m1n1 to get this kernel to boot on a 14.x-ed machine. this has been left as an exercise for the reader#asahi-dev · 2026-07-19 10:39
chadmedi think also the surface skipping/clearing stuff has changed in a breaking way. setting surf_null and friends to true no longer seems to be enough to get dcp to ignore a surface. macos also never sets these anymore