#asahi-dev · 2025-11-05 20:56
cy8aercam and cheese is still not working...#asahi-dev · 2025-11-05 19:53
jannauI've pushed asahi-6.17.7-1 to github. fixes booting on M1 Ultra with v6.17.x, will eventually fix webcam in GNOME snapshot and cheese, and fixes sanitizer builds with rust 1.91#asahi-dev · 2025-11-03 07:52
kaazooThe kernel has support for tracking page ownership, but that's not compiled in by default and also has some overhead: https://docs.kernel.org/6.16/mm/page_owner.html#asahi-dev · 2025-11-02 22:20
kaazooHi, is there a way to measure VRAM usage / allocated pages by the asahi kernel module? I tried tracing with BPF to count allocated pages, but that won't count pages that have been allocated before tracing has been started.#asahi-dev · 2025-11-02 18:55
chaos_princessseems to be working alright#asahi-dev · 2025-11-02 13:03
jannauthanks#asahi-dev · 2025-11-02 12:58
chaos_princessack, will try later today#asahi-dev · 2025-11-02 12:57
jannaugeneric GPU things and keeping an eye on memory use. I'll run the vulkan/opengl CTS#asahi-dev · 2025-11-02 12:55
chaos_princessany specific things to try?#asahi-dev · 2025-11-02 12:52
jannaunot yet tagged because I want to do more GPU testing. additional testing from someone else would be appreciated#asahi-dev · 2025-11-02 12:50
jannauI've pushed a fix for the kernel deadlock as rebase onto v6.17.6 to asahi-wip #asahi-dev · 2025-11-01 14:34
svenfound a few more and fixed them, thanks!#asahi-dev · 2025-11-01 12:54
svenyup#asahi-dev · 2025-11-01 12:54
kettenisI suppose that D was supposed to be an R too#asahi-dev · 2025-11-01 12:54
kettenis#define PIPEHANDLED_MUX_CTRL_CLK GENMASK(5, 3)#asahi-dev · 2025-11-01 12:54
kettenisbut#asahi-dev · 2025-11-01 12:54
kettenis#define PIPEHANDLER_MUX_CTRL 0x0c#asahi-dev · 2025-11-01 12:53
kettenissven: just noticed some interesting typos/pastos in the atcphy driver diff#asahi-dev · 2025-11-01 01:10
chadmedif you move the cursor out of the sw region quick enough, a ghost of it will stay there until kwin tracks damage and repaints that part of the screen#asahi-dev · 2025-11-01 01:09
chadmedkwin's software cursor fallback has regressed with 6.5, but only cosmetically #asahi-dev · 2025-10-28 10:03
svenok. i'll need something similar for tbt (tl;dr: can only touch the iommu MMIO region after the tbt nhi is powered on, last time i did the fake child device hack as well iirc)#asahi-dev · 2025-10-28 10:01
jannausven: still the fake platform child device just for the iommu, I still need to finish and send the email asking about that setup#asahi-dev · 2025-10-28 09:27
sventhat probably needs some improvements to the spmi controller and then a small change to tipd itself to use a spmi regmap#asahi-dev · 2025-10-28 09:27
chaos_princesswhich is probably a good thing to poke at if someone wants to help out, the RE is all done for you already :P#asahi-dev · 2025-10-28 09:26
svenyeah#asahi-dev · 2025-10-28 09:26
chaos_princessmodulo spmi i guess#asahi-dev · 2025-10-28 09:26
svenbut i'm fairly confident usb3 on m3 will essentially just work#asahi-dev · 2025-10-28 09:26
svenmaybe tbt5 changes something, haven't looked into that at all#asahi-dev · 2025-10-28 09:25
chadmedwhen the team's output is measured in commits :P#asahi-dev · 2025-10-28 09:25
sveni think m3 will just need a new dt-compatible and a change in m1n1 because the renamed some tunable in the adt#asahi-dev · 2025-10-28 09:25
sveni checked for m3 and it seems very similar#asahi-dev · 2025-10-28 09:24
chaos_princesson more serious note, do we know if tbt/usb-c/that entire mess is mostly same on newer SoCs?#asahi-dev · 2025-10-28 09:22
svenlol#asahi-dev · 2025-10-28 09:22
chaos_princesssve n: m5 wen :P#asahi-dev · 2025-10-28 09:21
mazah, now you're talking! :)#asahi-dev · 2025-10-28 09:21
sveni need some excuse to shift back the release date... ;)#asahi-dev · 2025-10-28 09:20
mazsven: don't say that in public -- you know how it goes... :-/#asahi-dev · 2025-10-28 09:17
svenis finally getting back to tbt#asahi-dev · 2025-10-28 09:17
svenjannau: didn't dcp have some weird iommu setup as well? have you found a sane solution for that?#asahi-dev · 2025-10-27 15:48
jnnok, thanks#asahi-dev · 2025-10-27 15:48
chaos_princessi mean in general, there is no document per se, just read the code that works with rtkit and do by example#asahi-dev · 2025-10-27 15:47
iretqthe linux one is much more complete, you can find it under drivers/soc/apple#asahi-dev · 2025-10-27 15:43
jnnin m1n1?#asahi-dev · 2025-10-27 15:34
chaos_princessrtkit.c#asahi-dev · 2025-10-27 15:33
jnnis there a document somewhat that summarizes the current knowledge about RTKit and its protocols? #asahi-dev · 2025-10-26 22:07
iretqyup, stupid copy paste mistake :P fortunately i found this out quick enough and now i'm working on hooking it up to the rest of the uefi firmware#asahi-dev · 2025-10-26 22:06
nicolas17iretq: so you had both sq and cq on the same piece of memory? :D#asahi-dev · 2025-10-26 20:42
iretqwow, i found out what the issue was - i was passing the address of the sq ring buffer when creating the io cq... really dumb :) it turns out i was right that it was going to be the dumbest mistake and i'll be stuck on it for the longest time, but now i've got nvme read working 🎉#asahi-dev · 2025-10-26 19:45
iretqany pointers