On Mon, Jul 29, 2019 at 03:17:01PM +0200, Juri Lelli wrote: > On 29/07/19 15:04, Peter Zijlstra wrote:
> > Now, looking at it, this also doesn't do push balancing when we > > re-select the same task, even though we really should be doing it. So I > > suppose not adding the condition, and always doing the push balance, > > while wasteful, is not wrong. > > Right, also because deadline_queue_push_tasks() already checks if there > are tasks to potentially push around before queuing the balance > callback. Yes, but in the overloaded case, where there is always a task to push, but nowhere to push it to, we can waste a 'lot' of time looking for naught in case of extra pushes. So in that regard the check you reference is not sufficient. Anyway, let me change this for now.