Skip to content

Flake detect harder

This is my attempt at cutting out the ~1% spurious failure rate in Mesa CI from freedreno.

I'm also hoping to use it for enabling vulkan testing on cheza, which is still pretty flaky. As far as testing: after a vk run I found a chain of 4 matching caselists in a row suggesting that this actually helped us, but the diff of the .qpas didn't show changing Result StatusCodes (but did show changing pipeline caching?) so I'm a little confused.

Merge request reports