00:18 DavidHeidelberg: zmike: can you run "./modules/egl/deqp-egl -n "dEQP-EGL.functional.choose_config.simple.selection_and_sort.buffer_size"" (or the whole group of the tests)?
00:25 zmike: DavidHeidelberg: isn't that from the group in the ticket?
00:25 zmike: I ran all of those
00:25 DavidHeidelberg: so far for me the 2nd group formats (when sort is done) are before these 1st group ones
00:26 DavidHeidelberg: just rechecking if by any chance the test could be skipped
00:26 zmike: I'm not actually running it right now
00:26 zmike: I'll look monday
00:27 zmike: but I don't recall any tests skipping
00:27 DavidHeidelberg: my branch is opengl-cts-4.6.4 + the patch
00:27 DavidHeidelberg: kk
00:28 zmike: just run khronos main?
00:28 DavidHeidelberg: let me retest
00:38 DavidHeidelberg: fails for me on main too
00:38 zmike: intel?
00:39 DavidHeidelberg: yup TGL
00:40 zmike: ok I only tested amd
00:40 zmike: I can test icl or skl
00:40 zmike: ...on monday
00:40 DavidHeidelberg: that would be helpful, since I think the Intel CI gives this on some recent XE cards
00:40 DavidHeidelberg: the EGL tests in MesaCI also passing
00:42 DavidHeidelberg: but the problem I'm seeing after deqp doing the sort the 2nd group IDs aren't put on the end of the lists but mix with the 1st group ones
00:42 DavidHeidelberg: which is weird, because I would expect this will be broken everywhere
11:46 daniels: Company: it’s not an error condition - if it was the swapchain would be in an error state, not ‘I can continue but it would be better if you reallocated’ …
11:47 zamundaaa[m]: oeuaoeuoeu: there's no way to do it with the proprietary NVidia driver
11:47 daniels: which can happen because the optimal modifier for rendering and the optimal modifier for scanout are rarely the same thing
11:53 Company: daniels: I am aware of all of that - but it's still strange that it does happen
11:54 Company: daniels: also, nvidia apparently returns OUT_OF_DATE in the same situation, which is an error
19:41 DemiMarie: With virtio-GPU native contexts, does the guest kernel need to be aware of what GPU is in use, or does only guest userspace need to be aware of that? (Resending because I don't think previous post made it to IRC.)
20:14 kode54: karolherbst: I understood your hex number reference to be a floating point number, but my brain forgot that it was 1.0
20:14 kode54: hopefully this doesn't spoil your survey
20:15 kode54: I honestly voted no because I didn't remember the exact reference off-hand
20:24 karolherbst: knowing 1.0 is the only valid reason to press yes tho :D
23:11 DavidHeidelberg: kode54: u ruined everything