12:07sima: hwentlan__, agd5f, airlied, rodrigovivi, jani there's going to be a conflict between -rc2 and drm-intel-next in amdgpu due to 7707dd6022593f3edd8e182e793
12:07sima: sorting it out for drm-tip rn
12:12sima: mripard, I'll do the drm-misc-next pull on Monday so it can go into -rc2 (or airlied will)
12:12mripard: sima: sounds good, thanks
13:29haasn: is .cache/mesa_shader_cache size limited or regularly cleaned up at all?
13:48alyssa: any galliumistas want to review a 1 line comment?
14:06ishitatsuyuki: haasn, one of the buckets gets evicted if the cache exceeds a size limit
14:07ishitatsuyuki: bucketing is by first two character of hash iirc
16:33Company: so I am testing with my embedded GPU (that's my secondary one just for that purpose)
16:34Company: and I noticed that when I use GL, nvtop reports a maximum of 85% usage of the GPU
16:34Company: while with Vulkan, I get to 98%
16:35Company: is GL stalling somewhere when transferring the buffer to the main gpu?
16:35Company: I just used DRI_PRIME=1 to switch GPUs btw
16:36Company: (also, the first frame looks like https://i.imgur.com/5hww6Yi.png so I suppose Mesa (or mutter?) is screwing up modifiers somewhere, but that goes away for the 2nd frame)
17:03jenatali: gfxstrand: I was trying to avoid bugging you for every nir change but I'm not getting any bites... any chance you can check the one nir patch in !26210?
17:17jenatali: Nvm, got a bite
18:04jenatali: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/51725768 - 46 flakes seems like a lot...
18:05gfxstrand: Yeah it does
18:06gfxstrand: Either freedreno CI is having problems or there's something funky going on with that patch.
18:06gfxstrand: The patch seems pretty harmless, though.
18:06jenatali: The re-run had 1 fail and 26 flakes
18:06jenatali: I think it's just a CI issue
18:12jenatali: Yeah, a re-re-run passed with 4 flakes
18:28gfxstrand: robclark: ^^
18:30robclark: either that MR or something that landed recently made things flakey? That isn't normal..
18:30gfxstrand: I'm surprised someone was able to land something that makes it flake that bad.
18:31gfxstrand: But it's possible, I guess.
18:31robclark: hmm, but also seems like collabora runners aren't reporting to our flakes channel
18:31robclark: yeah, if someone mashed retry enough times and eventually got lucky...
18:32robclark: oh, but wait.. that is cheza..
18:56daniels: robclark: that's the way flakes generally land, yeah
18:57jenatali: The re-re-re-run in Marge's pipeline had 7 flakes. Looks like a GPU fault hit and took out all the tests that were running in parallel
19:02robclark: hmm, assuming no one managed to silently break per-process pgtables on sdm845, parallel processes should be reasonably well isolated
19:03robclark: but that does seem like a lot more flakes than normal, unless what is logged to the flake channel is truncated
19:24jenatali: robclark: Is this something you would want to look at too? https://gitlab.freedesktop.org/mesa/mesa/-/jobs/51730159
19:24jenatali: Really seems like something should've timed out
19:26robclark: daniels: ^^^ I guess the lava stuff doesn't look for patterns like that? I think we have something like that for the bm jobs
19:31daniels: robclark: yep, it's on the list, was just stuck behind a bunch of other rework and cleanup of the various job submitters/monitors
19:31robclark: gotcha
19:36kisak: Hey folks, I'm gearing up to do an irregular bump of the mesa build in my PPA. Here's the tenative patchlist on top of the previous build of mesa 23.2.1. https://dpaste.com/2PLCCHP58
19:37kisak: my notes are primarily oriented to game fixes, but if anyone wants other things fast tracked to my build, please pass me a note in the next couple hours.
19:38kisak: if I was paying attention, I'm not allowed to do a minor revision bump without changing VERSION because there will be cache conflicts?
20:40jenatali: Ooh, MSVC finally got support for C11 threads
20:40gfxstrand: \o/