22:46 <anholt> has zink flaking in random deqp tests recently been noticed?

22:46 <anholt> https://gitlab.freedesktop.org/mesa/mesa/-/jobs/19117249

22:47 <zmike> no

22:47 <anholt> https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14962 got hit by a couple of flakes, but I'm thinking it's not that MR

22:48 <zmike> not sure I've ever seen that test fail for any reason

22:49 <anholt> my MR was a different set of tests

22:49 <anholt> one of the deqp fails was heap corruption

22:50 <anholt> https://gitlab.freedesktop.org/mesa/mesa/-/jobs/19034659

22:50 <zmike> haven't seen that either

22:51 <anholt> ignore the specific testcase, I'm looking for a pattern of when flakiness started happening.

22:51 <zmike> ah

22:52 <anholt> would be nice if we had a ci flakes channel for zink that we could just look at the history of:(

22:52 <zmike> it's been on my todo, but it's not very high up atm

22:53 <anholt> 2 zink pipelines just before overlapping vs/fs landed were clean, increasing my suspicion.

22:53 <anholt> anyway, I'll try to get ci-swrast in and then I'll see if I can get you some asan in the followup work, which may reveal the problem anyway.

22:54 <airlied> oh overlapping might cause flakes indeed

22:54 <anholt> airlied: nothing conclusive here, just "huh, we sure got flaky recently"