00:18Company: if I want my main loop to wake up when the GPU is finished, how do I best do that? Use a fence, spawn a thread, let the thread wait on the fence and when the fence signals it wakes up the main loop?
08:16MrCooper: Company: I'd get a sync file for the fence
08:17MrCooper: that's what mutter does
08:17MrCooper: i.e. it's battle-tested :)
08:28Company: that's Linux only I guess?
08:38MrCooper: should work on other platforms with the DRM subsystem, not Windows though
08:39Company: right
08:40Company: I'm usually thinking windows/mac/linux as the 3 platforms, sometimes I wonder about Android, too
09:00MrCooper: in this case the EGL feature happens to originate from Android
09:32jani: so this is off-topic, but there are people here who are bound to know something about this... chrome browser on debian started highlight-to-copy behaviour at some point. e.g. if I want to replace something in an input box with stuff I have in clipboard, simply selecting the existing text copies it, replacing whatever I have in the clipboard. and subsequent ctrl-V just pastes it back
09:33jani: I want to keep PRIMARY separate, and this works in firefox just fine. I can select and middle-button paste the PRIMARY, and that doesn't interfere with ctrl-C/ctrl-V stuff. except in chrome. how is that?
09:34jani: works in firefox and everywhere else, really
09:38MrCooper: sounds like a chrome issue then
09:42jani: yeah I'm sure it is, but I can't wrap my head around it, and the search results are less than helpful
09:48mvlad: https://chromestatus.com/feature/5085102657503232
12:10lucaceresoli: Hello, this bugfix
12:10lucaceresoli: https://lore.kernel.org/all/20260102155553.13243-1-osama.abdelkader@gmail.com/
12:10lucaceresoli: only has my R-by. Is it enough to apply is to drm-misc-fixes?
16:03glehmann: alyssa: can I get an rb for the brw commit here? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39180
16:03alyssa: glehmann: i'm in a good mood right now, so Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com>
16:04alyssa: now let me go read the code i guess :)
16:04alyssa: ok yes that's fine
16:05glehmann: thanks!
17:26mareko: FYI, nir_opt_algebraic_pattern_test is flaky: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/91379114
17:30alyssa: mareko: Oof, I have no idea how that can flake..
17:30alyssa: It isn't like.. randomized IIRC
17:32mareko: maybe clang/llvm is optimizing FP UB again
17:35alyssa: I mean maybe but I still dont see how it would *flake*
17:35alyssa: instead of reliably failing on only some platforms
17:35alyssa: which seems more straight up UB of the UBSAN/ASAN variety
18:30anholt: anything that changes float environment in nir_tests? wouldn't think so.
18:30anholt: also, I think we run our unit tests such that each gtest binary runs single-threaded? so it shouldn't be races on that front
18:31anholt: (I'm actually on vacation today, won't be available, but I've got the tab open for later)
19:29austriancoder: gfxstrand: if you have some spare minutes, would be great if you could look at this single commit from an already reviewed MR. https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38929/diffs?commit_id=635f22e8f252c49578d9e3281149be4f66781a7d