> 3. Now that we have utrace->resume, can't we kill report->resume_action ?
I thought this initially when making the change and then decided against it. I don't recall exactly what was in my mind at the time. It would take some more thought now to be sure whether there is a semantic problem. But at any rate, keeping it reduces the number of cases in which we take the utrace lock in finish_callback_report, which seems like a good thing. Still, if you propose a patch to remove it with thorough rationale on why it's good or better, then that will be fine. Thanks, Roland