07:35 MrCooper: emersion lina: FWIW, not handling implicit sync across certain user-space API boundaries isn't really an option for upstream drivers (it's not hard to achieve even while using explicit sync internally in the drivers); even for nvidia, not handling implicit sync has been causing trouble for their users for many years, and will continue so for years to come
10:24 DodoGTA: Why are all Gallium drivers collapsed into one massive library (libgallium)? This makes it near impossible to prevent individual drivers from loading
10:35 MrCooper: why do you want to prevent them from loading? Their code pages won't be faulted in while they're unused
10:41 Lynne: gfxstrand: any plans for implementing VK_EXT_shader_atomic_float for nvk?
11:16 lina: MrCooper: Right, and it all works already for the native/host case (and it also works, but inefficiently, between apps within a VM), that's been implemented in Mesa for a long time. It's just the VM-host case that is broken right now for pure explicit sync kernel drivers (it looks like drm/asahi is the first such driver to be used with virtgpu
11:16 lina: native context on top, so nobody ran into this before...).
11:38 DodoGTA: MrCooper: I mean I don't want a problematic Gallium driver to be running on one of my GPUs
12:05 alyssa: can confirm, my galium drivers are .. problematic :p
12:06 karolherbst: problematic gallium drivers? What about problematic gallium frontends?
12:06 alyssa: so true bestie
12:10 karolherbst: (wait until they hear that frontends also ship a copy of all drivers)
12:10 karolherbst: ("all")
12:11 Ermine: what is a gallium frontend them?
12:11 karolherbst: API implementation
12:50 sghuge: Could someone confirm if they can turn on Ray tracing option for Spiderman remastered?
15:17 Company: Do I need to do anything for https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31289 and https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31293 ? Find somebody who can assign them to Marge? Wait for the maintainers to do that?
15:21 karolherbst: Company: probably want to add the review tags just to be sure, but yeah, ping the maintainer or on the MR. The default is mostly that submitters can assign to marge and get it merged
15:22 Company: so the real fix is to get me the permission to assign to marge :)
15:26 karolherbst: that would certainly be one way to do it :) I think there are some rough rules around having 25 commits already or something like that, it's maybe even documented somewhere, but I honestly don't know
15:27 karolherbst: (and then finding people acking such a request)
15:32 Company: yeah, I'd also want to feel like I know what I'm doing first
15:53 zmike: are there any fourcc utils in mesa to get the modifier name?
15:56 daniels: drmGetFormatName(), drmGetFormatModifierVendor(), drmGetFormatModifierName()
15:56 zmike: well yeah, sure I COULD use libdrm
15:56 zmike: I wanted some of that hand-rolled, bespoke 2004-era enum matching code though
15:58 MrCooper: would need a time machine to make modifiers a thing in 2004 first
15:59 zmike: you know what I meant
16:00 MrCooper: yeah, I don't know of Mesa ever having anything like that though
16:00 zmike: I guess I'll have to settle
18:31 mareko: zmike: you would need a modifier parser, libdrm might have it
18:32 mareko: I don't think modifiers use fourcc, that's for formats
18:32 zmike: yeah, I realized after I said it
18:33 zmike: it's a tough start to the week.
18:34 mareko: drmGetFormatModifierName
18:34 mareko: is the parser
18:42 Mis012[m]: @_oftc_JoshuaAshton:matrix.org: Hi, seeing as Valve is doing some stuff with aarch64/android/VR, can we expect stuff like OVR_multiview to be implemented in freedreno? (we have gotten it to work anyway in [atl](https://gitlab.com/android_translation_layer/android_translation_layer), but the performance would presumably be quite a bit better with multiview)
18:42 Mis012[m]: hm, does the mention always look this sad on the irc side...
18:45 Mis012[m]: JoshuaAshton: iirc there's a nice vulkan extension for multiview, but a lot of Oculus games don't even bother to use OpenXR, so it would be wishful thinking to hope they use OpenXR AND vk
19:15 haagch: well the zink support for GL_OVR_multiview *just* landed
19:16 haagch: godot 4's opengl renderer for openxr requires the extension, that might be the most straightforward argument for driver support
19:48 mareko: alyssa: hi, can you please have a look at? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29384 Thanks.
20:07 alyssa: sure
20:08 alyssa: mareko: I'm confused why you want it merged, given the comment there?
20:09 alyssa: but sure
20:15 alyssa: oh, for Intel. OK
20:44 mareko: alyssa: it's a good change for the community even though AMD drivers won't likely use it
20:46 alyssa: sounds good