From: "Steven Rostedt (Red Hat)" <rost...@goodmis.org> Now that the idle_balance is called from the post_schedule of the idle task sched class, it is safe to enable interrupts. This allows for better interaction of tasks waking up and other interrupts that are triggered while the idle balance is in process.
Preemption is still disabled, but perhaps that can change as well. That may need some more investigation. It may be safe to also enable preemption, but we'll leave that change for another time. Cc: Peter Zijlstra <pet...@infradead.org> Signed-off-by: Steven Rostedt <rost...@goodmis.org> --- kernel/sched/fair.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index ed18c74..0fcdbff 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5222,9 +5222,10 @@ void idle_balance(int this_cpu, struct rq *this_rq) update_rq_runnable_avg(this_rq, 1); /* - * Drop the rq->lock, but keep IRQ/preempt disabled. + * Drop the rq->lock, but keep preempt disabled. */ - raw_spin_unlock(&this_rq->lock); + preempt_disable(); + raw_spin_unlock_irq(&this_rq->lock); update_blocked_averages(this_cpu); rcu_read_lock(); @@ -5251,7 +5252,8 @@ void idle_balance(int this_cpu, struct rq *this_rq) } rcu_read_unlock(); - raw_spin_lock(&this_rq->lock); + raw_spin_lock_irq(&this_rq->lock); + preempt_enable(); if (pulled_task || time_after(jiffies, this_rq->next_balance)) { /* -- 1.7.10.4
signature.asc
Description: This is a digitally signed message part