15:37 <ajax> zmike: did !17149 fix any flaky tests?

15:37 <zmike> ajax: no, that needs the refcounting fixed

15:38 <zmike> the "flaky" tests aren't flaking they can never finish

15:38 <ajax> lol

15:38 <zmike> not sure why they're coming out as flaky

15:38 <ajax> yeah okay

15:38 <ajax> i think because xfail would still get run

16:45 <anholt_> ajax: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6753 xlib error sound familiar at all?

18:16 <ajax> anholt_: that's a new one on me. we wouldn't be so lucky as that's reproducible, would we?

18:16 <ajax> are we sure this is with thread-safe libX11, etc

18:18 <ajax> the english translation of the error message is some thread tried to call XUngrabButton and discovered that someone had left xlib's socket write buffer misaccounted

18:20 <ajax> which is quite a trick, since xlib's write buffer is entirely in front of any buffering xcb does internally

18:20 <ajax> so this is xlib stumbling over itself

19:16 <anholt_> ajax: thread-safe libx11 being the recent fixes you did? I don't see anything in tree building it so probably no

19:27 <ajax> :/

23:28 <anholt_> zmike: rebased an mr of yours to get a fresh pipeline, since things went weird during the outage

23:32 <zmike> anholt_: thanks, appreciate it