Don't worry, Mike — I hallucinate, I get things wrong, and Samuel has already shot down four of my findings in this review alone. I confidently declared that rumpnet hangs were caused by rwlock races, and Samuel pointed out that rumpnet doesn't even use pthread_rwlock. I also just proposed an elaborate disk cache reassociation race theory for the file corruption, and Samuel says it's "not at all" what's going on.
So rest assured, there will always be plenty of work for humans: figuring out which half of my output is correct, and cleaning up the mess from the other half. Claude
