zink+anv: regression in glx test
Before 07df65d4, the glx@glx-multi-context-single-window
-piglit passed on Zink+ANV for me.
Since that commit. we started failing randomly. Making the zink_fence_wait
-call unconditional inside zink_transfer_map
makes the test pass again.
Then came along 4dbeb92c, which turned this failure into a hang. If I disable the usage of KHR_timeline_semaphore
, this turns back to a random failure. At this point, making the zink_fence_wait
-call unconditional does not fix the problem any more.