12:31 glehmann: anholt: debian-riscv64 seems to timeout in nir_tests quite often, either the time limit should be increased or algebraic testing must be faster
12:57 lumag: mripard, tzimmermann, mlankhorst: what is the current tree for fixes for the next? Should I push to drm-misc-next or drm-misc-next-fixes?
12:58 lumag: I expected that at -rc7 the drm-misc-next would be closed, but dim says that we are still out of the merge window..
13:07 mripard: lumag: yeah, I haven't opened drm-misc-next-fixes right now
13:07 mripard: I can't right now, but I'll do it sometime this afternoon
13:29 mlankhorst: mripard: can take over if you want?
14:52 mripard: mlankhorst: if you have some time, with great pleasure :)
15:26 mlankhorst: mripard: ok will do so
15:28 mlankhorst: lumag: You should be able to push to drm-misc-next-fixes now, enjoy!
17:02 mripard: mlankhorst: thanks :)
17:28 anholt: glehmann: wish we had a shared libnir so we didn't have to link all our gtest tests together as one -- makes timeout management suck. Bumping the timeout to 90-120 seems like a good choice given the current limitations.
18:16 jenatali: Anybody able to help debug why https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39501 breaks ambient occlusion in Age of Empires on v3d?
18:54 lumag: mlankhorst: thanks!
20:51 daniels: glehmann: or set RUN_MESON_TESTS: "false" for debian-riscv64
20:51 daniels: it would be nice to run tests on RISC-V, but it would be much nicer to not do loads of nonsense under qemu
22:22 anholt: Anyone have any other tests that cause havoc on memory on your test devices? https://gitlab.freedesktop.org/anholt/mesa/-/commit/a5b3630732dfb8cf719abde0cacf36056fb3990c
22:25 daniels: anholt: <3
22:28 zmike: oooo
22:36 austriancoder: anholt: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/etnaviv/ci/etnaviv-gc2000-r5108-skips.txt#L2
23:06 anholt: austriancoder: already in the linked commit
23:11 austriancoder: anholt: are you sure? a5b3630732dfb8cf719abde0cacf36056fb3990c does not touch etnaviv-gc2000-r5108-skips.txt
23:12 anholt: austriancoder: I didn't remove them from all the devices, because I don't know for sure if your device will work once it's moved to single-thread. (e.g. a530 will definitely *not* work for some, even single-threaded)
23:13 anholt: getting touches-all-ci changes like this through is hard enough to begin wtih.
23:13 austriancoder: anholt: ahh .. okay, will test it once you work lands
23:38 Kayden: pendingchaos: thanks a lot for your review on https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39250 - I've fixed up all the feedback you had. think it's probably good to land now, unless you want to take another look
23:41 Kayden: ngcortes: you might want to check out anholt's parallel-deqp-runner change above - and if we've got any memory-intensive tests that we'd like to avoid running concurrently with others, it might be nice to add them to said list
23:41 Kayden: (it'll basically ensure those tests are run alone)
23:42 ngcortes: Kayden, anholt roger!
23:42 ngcortes: tnx :)
23:44 Kayden: (intel's ci moved to using parallel-deqp-runner as well recently, and I know we've hit some annoying tests. I see some of them listed there already, but not sure if there are others)
23:44 Kayden: good idea to centralize it! :)