OK, so the issue persists. I'm going to try replicate this in a VM and report upstream. Thanks for the information so far.