> Well, I think you are right... But I just can't convince myself I > _really_ understand why the "set new bits" case is not interesting. > Otoh, I can't imagine why the caller should worry about -EINPROGRESS > if it doesn't clear any events.
If you set a new bit and utrace_set_events() returns 0, that very instant, before you even examine the return value, a just-enabled report_* callback could be in progress. So indeed it is hard to see a value in even claiming any such behavior. > But I think the new code should have the comment to explain this > "old_flags & ~events" condition. Yes, I intended to change/add comments if I put it in. Thanks, Roland
