Messages in this thread | | | Date | Tue, 27 Jan 2015 15:18:04 +0100 | From | Peter Zijlstra <> | Subject | Re: [PATCH 4/5] sched/rt: Consider deadline tasks in cpupri_find() |
| |
On Tue, Jan 27, 2015 at 01:58:59PM +0100, Peter Zijlstra wrote: > > @@ -103,24 +104,11 @@ int cpupri_find(struct cpupri *cp, struct task_struct *p, > > if (skip) > > continue; > > > > - if (cpumask_any_and(&p->cpus_allowed, vec->mask) >= nr_cpu_ids) > > + cpumask_and(lowest_mask, &p->cpus_allowed, vec->mask); > > + cpumask_and(lowest_mask, lowest_mask, cp->cpudl->free_cpus); > > + if (cpumask_any(lowest_mask) >= nr_cpu_ids) > > continue; > >
> > +++ b/kernel/sched/rt.c > > @@ -1349,14 +1349,17 @@ out: > > return cpu; > > } > > > > +static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask); > > static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p) > > { > > + struct cpumask *lowest_mask = this_cpu_cpumask_var_ptr(local_cpu_mask); > > + > > /* > > * Current can't be migrated, useless to reschedule, > > * let's hope p can move out. > > */ > > if (rq->curr->nr_cpus_allowed == 1 || > > - !cpupri_find(&rq->rd->cpupri, rq->curr, NULL)) > > + !cpupri_find(&rq->rd->cpupri, rq->curr, lowest_mask)) > > return; > > > > /* > > > Again; should you not put something useful in the mask before you pass > it to cpupri_find()?
Similar to the other case; I can't read today. I only saw the _and(lowest_mask, lowest_mask, ...) and figured it ought to be populated before that.
| |