02:26 Company: das Vulkan encode stride requirements somewhere?
02:26 Company: I'm trying to query physical device properties for dmabuf import, but can't find any mention of strides anywhere
02:37 i509vcb: Company: I don't think you can query those before creating a device. vkGetImageSubresourceLayout requires a VkImage to exist
02:37 Company: hrm, device would be okay
02:37 Company: vkImage is a bit overkill
02:37 i509vcb: Yeah you'd have to import to learn the strides from what I see
02:38 Company: because I want to enumerate all supported formats/modifiers
02:38 i509vcb: Oh you want modifiers for formats
02:38 Company: and was wondering if I could get info about required strides there
02:38 Company: no, I want strides for each format:modifier combo
02:39 i509vcb: yeah you can't seem to query strides with the current api
02:40 i509vcb: I guess you could have Vulkan allocate some images to query the strides, but that might not be a strong guarantee
02:41 Company: and there's like 200 format:modifier combos and allocating 200 images at startup just to query strides seems a bit excessive
02:43 Company:still on his quest to get enough info to hand off to v4l or hw decoders so they generate dmabufs that don't fail when creating Vulkan/EGL resources from them
02:49 i509vcb: I think a new Vulkan extension might be needed for that info
02:51 i509vcb: Although the Vulkan video extension would know how to allocate it's own video buffers, that won't guarantee it works when handing over a vulkan video buffer to v4l
05:50 daniels: (there’s also not a single ‘align to $n’ answer)
06:33 emersion: the stride is just a small part of the whole puzzle
06:34 emersion: wouldn't be nearly enough
07:10 pq: robclark, agreed, no worries. :-)
08:16 Company: swick[m]: good news! the mutter hdr hack that didn't work on F38 for me does work on F39 and I've seen HDR output for the first time ever on Linux
08:17 swick[m]: Company: kernel update I guess
08:17 Company: this is the desktop with the external monitor (I don't expect the laptop to work)
08:22 MrCooper: Company: a first glimpse of our glorious future :)
08:23 Company: I got a light panel, this is already a glorious present
09:57 aravind: Hi agd5f, it would help if you could comment on this series https://lore.kernel.org/dri-devel/20231008092616.1457995-1-aravind.iddamsetty@linux.intel.com/
10:21 tzimmermann: javierm, can i talk you into reviewing https://patchwork.freedesktop.org/series/124314/ ?
12:09 javierm: tzimmermann: sure thing. I'll try to do it today
12:09 tzimmermann: ok :)
12:09 tzimmermann: thank you
12:09 tzimmermann: the patches are all very similar
12:09 javierm: Ok
12:19 daniels: hwentlan_: you should have it
12:19 daniels: Melissa should have two tils
12:19 daniels: *we can see ourselves)
15:15 ccr:boogies.
16:41 _DOOM_: I am looking through some linux drm stuff, and I keep seeing references to planes with dma-bufs. But I can't find any reference, in detail, in documentation. What are these planes supposed to be?
16:53 tnt: _DOOM_: it's just if the color data isn't stored "interleaved" but instead in different buffer. Like you have all the Y data in one buffer and then all the UV data in another buffer for instance.
17:58 mareko: jenatali: are you OK with this change to CI fails? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25391/diffs
17:59 jenatali: mareko: Yeah, ack
18:09 mareko: enunes: are you also OK with this change to CI fails for lima? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25391/diffs
19:24 _DOOM_: tnt: Is there any documentation on how I should handle this?
19:29 mareko: what are GTF-GL46 tests and are they important?
19:30 zmike: those are the confidential tests
19:31 zmike: they are required for conformance until 4.6.3 and will be ported into new (non-confidential) tests soon
19:32 zmike: CTS readme has more info
19:32 airlied: they cover a lot of legacy GL features
19:43 tnt: _DOOM_: sort of depends what you're trying to do.
19:44 tnt: (also bear in mind I'm not expert, I just delved into that stuff a bit recently because of some dmabuf debugging ...)
19:58 _DOOM_: Oh ok
19:58 _DOOM_: Thank you
23:19 mareko: where do I get VkPhysicalDeviceVulkanSC10Properties? it's required for building dEQP and CTS
23:22 Sachiel: should be there already
23:55 mareko: Sachiel: deqp/external/vulkancts/modules/vulkan/vktTestCase.hpp:97:19: error: ‘VkPhysicalDeviceVulkanSC10Features’ in namespace ‘vk’ does not name a type
23:57 Sachiel: I built main today just fine. Did you do all the fetch_sources.py dance? Though the type is in the generated framework stuff already