Hi Peter, first of all, thanks for all the reviews!
On Tue, 5 Apr 2016 14:23:27 +0200 Peter Zijlstra <[email protected]> wrote: [...] > > @@ -1218,6 +1242,9 @@ static void task_dead_dl(struct task_struct *p) > > /* XXX we should retain the bw until 0-lag */ > > dl_b->total_bw -= p->dl.dl_bw; > > raw_spin_unlock_irq(&dl_b->lock); > > + > > + if (task_on_rq_queued(p)) > > + sub_running_bw(&p->dl, &rq->dl); > > } > > > > A dead task cannot be running, not be queued, right? ISTR you remove > this hunk in a later patch as well. I suspect this is some "better safe than sorry" code I added trying to solve some issue that then I solved in a different way... But I forgot to remove it. I'll fix the patch and re-test. Thanks, Luca
