#asahi-dev · 2023-09-23 16:06
linasorry, I had totally forgotten about the tracer pastes... I meant to do this earlier but stuff happened and I completely forgot that bit...#asahi-dev · 2023-09-23 16:05
linaI need to get some sleep, but is there some time that would be convenient for you next time? ^^#asahi-dev · 2023-09-23 16:04
linaon both repos#asahi-dev · 2023-09-23 16:04
linaI just pushed what I have to isp/t602x#asahi-dev · 2023-09-23 16:04
linaohhhh#asahi-dev · 2023-09-23 12:41
linaI also tried getting the tracer to trace IO stuff to have a better idea of what macOS is doing, but I'm not sure if it's working right, or I don't understand how to compare it to what Linux does...#asahi-dev · 2023-09-23 12:40
linaeiln: It seems things have changed in this SoC... I got it booting and initializing with boring changes, but it's asserting on some commands. I tried making changes without knowing what I'm doing to make progress but I wound up at an assert when starting streaming. Help... ^^;;#asahi-dev · 2023-09-23 09:06
linaeiln: If you're around, I'm starting now ^^#asahi-dev · 2023-09-22 15:14
linaI mean saturday#asahi-dev · 2023-09-22 15:13
linaeiln: I was planning to do it today, but I had a weirdly busy day again... maybe tomorrow (sunday) 3PM JST?#asahi-dev · 2023-09-19 13:03
linabut also for testing I'll just point it at a wall...#asahi-dev · 2023-09-19 13:03
linaj`ey: I mean, I hope one day I can use eiln's ANE stuff to run motion capture instead of an iPhone ^^#asahi-dev · 2023-09-19 11:08
linaeiln: Got it! I'll give it a try tomorrow ^^#asahi-dev · 2023-09-19 03:39
linaeiln: Hi! When you're around, I heard you want more people helping get the ISP stuff working on other machines... do you think I could try to get t802x working in a stream tomorrow? I have this M2 Max lying around... ^^#asahi-dev · 2023-06-02 08:28
linasid-linux: The dmabuf leak would affect any app that doesn't explicitly close all FDs#asahi-dev · 2023-06-02 08:06
linasid-linux: Fixed the dmabuf leak ^^. All of this is leaking from plasmashell, so it depends on how you start konsole whether you see the leaks or not. I think the other fd leaks have always been there...#asahi-dev · 2023-06-02 07:43
linaI think I used plain dup without CLOEXEC in a change recently... that's probably where those come from#asahi-dev · 2023-05-18 18:49
lina(By that I mean I'm too self-conscious about my piano skills so the piano stuff goes on Twitch... ^^; )#asahi-dev · 2023-05-18 18:49
linaI use Twitch for ephemeral stuff and YouTube for more persistent stuff ^^#asahi-dev · 2023-05-18 18:39
linaThey just reduced it I think#asahi-dev · 2023-05-18 18:38
linaI mean for us plebs ^^;;#asahi-dev · 2023-05-18 18:38
linaI think Twitch is one week now? I got an email about that...#asahi-dev · 2023-05-18 18:24
linaI sometimes watch another vtuber that uses it and well... I think the player doesn't pipeline properly, which means when you're more than halfway across the world it all kind of breaks down.#asahi-dev · 2023-05-18 18:23
linaOwncast is... not super great.#asahi-dev · 2023-05-08 13:24
linaI'll ask her about it!#asahi-dev · 2023-05-08 13:22
linaSoooo... https://github.com/hikalium/wasabi/tree/main/os/src/xhci ^^#asahi-dev · 2023-05-08 13:21
linaLet me see if I can find it...#asahi-dev · 2023-05-08 13:21
linaAh! Yeah, that was hikalium ^^#asahi-dev · 2023-04-30 10:45
linaI wonder if I should just do what DART did and reimplement it (except directly in Rust and not use io-pgtable at all...)#asahi-dev · 2023-04-30 10:44
linasven: And I never got a reply from Robin about what to do with the UAT IOMMU code... ;;#asahi-dev · 2023-04-28 10:59
linamarcan: Umm, your cpuidle stuff is making lockdep very angry ^^;;#asahi-dev · 2023-04-04 03:12
linakettenis_: I already mentioned this to robher at some point ^^#asahi-dev · 2023-03-21 12:46
linaThe more interesting thing will be how hard it is to get shmem to do 16K allocations...#asahi-dev · 2023-03-21 12:46
linaWhat sven said ^^#asahi-dev · 2023-03-12 15:32
linachadmed: I didn't even know a glmark score that low was even possible, even at low pstates...#asahi-dev · 2023-03-12 15:32
linaBetween that and the kernel oopses I think next week is going to be another debugging week... ;;#asahi-dev · 2023-02-21 16:46
linaI sent it a month ago, got a reply 3 days later, replied to that and after that nothing...#asahi-dev · 2023-02-21 16:46
linaI think I might be getting the DART2 treatment from Robin for my UAT patch... ;;#asahi-dev · 2023-02-17 08:02
linaOtherwise we're just going to get Phoronix putting out bad benchmarks again...#asahi-dev · 2023-02-17 08:02
linaPeople are starting to try to run glmark & friends and are really running into this, and reviewers aren't going to know to pin tasks so... Even if it's just a hack at first, I think we really should do this to avoid confusion.#asahi-dev · 2023-02-16 16:23
linaSomething something cgroups?#asahi-dev · 2023-02-16 16:23
linaIt basically doesn't make any sense not to throw UI tasks immediately on the P cores, in my opinion. Anything less and you risk reduced responsiveness, since the scheduler has no idea what the upcoming CPU requirement will be and anything UI is inherently bursty.#asahi-dev · 2023-02-16 16:22
linaI think we're going to have to investigate the scheduler more, possibly make core changes. This is why Apple has fancy usecase-type scheduling in macOS...#asahi-dev · 2023-02-16 16:22
linaalyssa: It'll still happen for many GPU workloads... Anything where the workload isn't overall fully CPU-bound will end up having that problem.#asahi-dev · 2023-02-10 11:13
linaarnd: Yeah, this is a fatal error path anyway so __packed makes perfect sense ^^#asahi-dev · 2023-02-09 23:39
linaarnd: I think I did that to avoid alignment issues, but at least on arm64 that shouldn't matter... I can replace it with a pointer if you want#asahi-dev · 2023-01-14 05:37
lina(also I got pinged by "linalg" www)