From: "Siddha, Suresh B" Periodic load balancing in recent kernels happen in the softirq. In certain -rt configurations, these softirqs are handled in softirqd context. And hence the check for idle processor was always returning busy (as nr_running > 1). This patch captures the idle information at the tick and passes this info to softirq context through an element 'idle_at_tick' in rq. Signed-off-by: Suresh Siddha Cc: Ingo Molnar Cc: Thomas Gleixner Cc: Nick Piggin Signed-off-by: Andrew Morton --- kernel/sched.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) diff -puN kernel/sched.c~sched-fix-idle-load-balancing-in-softirqd-context kernel/sched.c --- a/kernel/sched.c~sched-fix-idle-load-balancing-in-softirqd-context +++ a/kernel/sched.c @@ -223,6 +223,7 @@ struct rq { unsigned long raw_weighted_load; #ifdef CONFIG_SMP unsigned long cpu_load[3]; + unsigned char idle_at_tick; #endif unsigned long long nr_switches; @@ -2944,12 +2945,7 @@ static void run_rebalance_domains(struct struct rq *this_rq = cpu_rq(this_cpu); unsigned long interval; struct sched_domain *sd; - /* - * We are idle if there are no processes running. This - * is valid even if we are the idle process (SMT). - */ - enum idle_type idle = !this_rq->nr_running ? - SCHED_IDLE : NOT_IDLE; + enum idle_type idle = this_rq->idle_at_tick ? SCHED_IDLE : NOT_IDLE; /* Earliest time when we have to call run_rebalance_domains again */ unsigned long next_balance = jiffies + 60*HZ; @@ -3219,14 +3215,16 @@ void scheduler_tick(void) unsigned long long now = sched_clock(); struct task_struct *p = current; int cpu = smp_processor_id(); + int idle_at_tick = idle_cpu(cpu); struct rq *rq = cpu_rq(cpu); update_cpu_clock(p, rq, now); - if (p != rq->idle) + if (idle_at_tick) task_running_tick(rq, p); #ifdef CONFIG_SMP update_load(rq); + rq->idle_at_tick = idle_at_tick; if (time_after_eq(jiffies, rq->next_balance)) raise_softirq(SCHED_SOFTIRQ); #endif _