08:28MrCooper: there's no way from user space to get a timestamp when the kernel signalled a dma-fence, is there?
08:50daniels: MrCooper: https://gitlab.freedesktop.org/wayland/weston/-/blob/main/libweston/linux-sync-file.c?ref_type=heads#L60-L83
08:54MrCooper: wow, thanks!
08:57daniels: np :)
08:58MrCooper: that makes the GPU timestamp gymnastics mutter used to do to reconstruct that timestamp all the more hilarious
10:17DragoonAethis: FYI: Intel CI will be down for the weekend (queues stop 26.06 9:30 CEST, resume 29.06 10:00) due to a lab building maintenance - please plan accordingly :)
10:34dolphin: "Branch setup for the integration repo is borked" <- I'm getting the following error almost every time I push something with dim
10:34dolphin: are we quite sure the check actually works out?
10:35dolphin: drm-tip appears in detached state
10:35dolphin: have to do "git reset --hard drm-tip/drm-tip"
11:54YaLTeR[m]: should be good now; i replied with "interesting"
14:20zmike: mareko: maybe dumb question, but is there any reason why set_constant_buffer still only passes a single cb instead of multiple? seems like tc at least could be batching them
15:49lucaceresoli: mripard: about https://lore.kernel.org/all/20260624-delectable-sexy-crayfish-e15c1f@houat/
15:49lucaceresoli: mripard: thanks for the discussion, I am going to reply
15:50lucaceresoli: mripard: but it's still not clear to me whether the panel bridge allocation changes you mention are on your TODO list
15:51lucaceresoli: mripard: if they aren't, I'll take them; just I'd better avoid duplicating the effort
20:36alyssa: glehmann: uff, right. good point
21:10austriancoder: is there a tool that makes use of EXT_device_address_binding_report?
21:38FireBurn: Now I know why Lyude's branch is called rust-stuck :P
21:38Lyude: oh haha
21:39Lyude: yeah that was a branch specifically intended to show off a bug I found that hands rust's compiler
21:39Lyude: *hangs
22:01FireBurn: I think I've worked past it
22:01FireBurn: I've rebasing vino onto it
22:02Lyude: on that branch? also, what is vino and what's the reasoning
22:02Lyude: if it's KMS related there's a lot more KMS work past what's already there that I still have yet to rebase lol
22:03FireBurn: Vino is a rust reverse engineered driver for DisplayLink USB Docks - it replaces evdi + DLM (well if I ever get it to work right)
22:03Lyude: and also like, I don't think the KMS work on that branch even goes that far
22:04Lyude: rvkms-slim is closer to what I had last, but needs the rebasing
22:04FireBurn: Dave suggested I base it on your work - and that was your last updated branch - I did email
22:05Lyude: FireBurn: how far are you with this so far?
22:06Lyude: rvkms-slim is probably the best bet but I can prioritize trying to get something newer rebased sooner then later
22:06FireBurn: Core stays on the old solver and kernel-cate moves onto the old solver, it prevents the ICE or the OOM
22:07FireBurn: Let me finish off what's been done and you can decide if it's useful or helpful
22:07Lyude: yeah np
22:07FireBurn: I don't suppose you know anyone with a DisplayLink dock do you?
22:09Lyude: I can probably get my hands on one or expense one
22:09Lyude: i bet my girlfriend probably has a spare one lying around somewhere in the ewaste repo tho
22:11FireBurn: Every test I've done, whether it's timing, ordering, or resolving the keys, suggests it's idential to what evdi+DLM does, but when DLM starts doing video, vino doesn't :(
22:12Lyude: is DLM some userspace portion of this? I remember DisplayLink being quite wonky
22:13FireBurn: Yeah, evdi drives the kms part, the DLM uses libusb to talk to the dock
22:13Lyude: wild
22:14FireBurn: It works so badly on linux
22:14FireBurn: So decided to do something about it
22:14Lyude: I'm just kind of amazed you got anywhere on that branch since like, we don't even have the basic Crtc/Plane/Connector objects there just yet
22:14Lyude: https://gitlab.freedesktop.org/lyudess/linux/-/commits/rvkms-slim?ref_type=heads this was the other branch jfyi
22:15Lyude: oh god it's already been 2 years. ouch.
22:56FireBurn: Lyude: I need some sleep, is this moving things in the right direction? https://gitlab.freedesktop.org/FireBurn/linux/-/commits/vino-v2-base
22:56Lyude: oh
22:56Lyude: that is a lot of claude
23:02Lyude: FireBurn: on top of the fact it's incredibly difficult to tell what is going on here, and a lot of this is markdown junk. also, my co-developed-by is placed on these patches despite the fact I haven't had any involvement with writing this. Please remove it.
23:03FireBurn: Yeah I've not had a chance to clean any of it up yet
23:04Lyude: I'm more then happy to guide you through how things work so you can write up some proper code but I'm really, really not at all comfortable with the development of these bindings being driven by generative AI.
23:06Lyude: there's not really anything here I can use
23:09Lyude: on top of that - it's really difficult to tell what's going on in here in general
23:15Lyude: If you're really serious about this please let me know and I'm happy to guide you through anything you need clarification if you want to move forward with this. But most of this is random markdown files, a huge amount of it is just large copy pastes of the code that I wrote ithout any of the actual commits
23:15Lyude: https://gitlab.freedesktop.org/FireBurn/linux/-/commit/d543f17bb7aeab73592ec70e2fffcb99087ad775
23:16tarceri: mareko: Can you take a look at https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/42108 when you have a free moment? Thank you
23:25Lyude: FireBurn: let me know if you're up to giving this another attempt in the future, without the claude code, I'm more then happy to do whatever I can to help. and I promise, you are probably capable of doing this on your own and you'll save yourself a lot of time. I say the last part because honestly speaking, this would take me ages to actually clean up into something that can be
23:25Lyude: submitted.
23:33Lyude: and also, sorry for the response. Genuinely really trying not to make people feel bad here
23:59mareko: zmike: yes it's not efficient