<anholt>
no reproducing it on adl, even with the driver build from ci. no asan fails, no validation fails. cmd line copied from the trace. my apitrace isn't exactly the same build, and my window system doesn't match, but that's all I've got I can think of.
<anholt>
do we have any idea when it started?
<zmike>
I'd guess with one of the MRs merged over the past few days
<zmike>
I know what's causing it, I just have to figure out how
<zmike>
some weird corner case of rp_changed not being flagged