sched, nohz: Change rq->nr_running to always use wrappers

Sometimes ->nr_running may cross 2 but interrupt is not being
sent to rq's cpu. In this case we don't reenable the timer.
Looks like this may be the reason for rare unexpected effects,
if nohz is enabled.

Patch replaces all places of direct changing of nr_running
and makes add_nr_running() caring about crossing border.

Signed-off-by: Kirill Tkhai <tkhai@yandex.ru>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20140508225830.2469.97461.stgit@localhost
Signed-off-by: Ingo Molnar <mingo@kernel.org>

5 files changed, 17 insertions, 15 deletions

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 800e99b99075..e0a04ae1e0dd 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -741,7 +741,7 @@ void inc_dl_tasks(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 
         WARN_ON(!dl_prio(prio));
         dl_rq->dl_nr_running++;
-        inc_nr_running(rq_of_dl_rq(dl_rq));
+        add_nr_running(rq_of_dl_rq(dl_rq), 1);
 
         inc_dl_deadline(dl_rq, deadline);
         inc_dl_migration(dl_se, dl_rq);
@@ -755,7 +755,7 @@ void dec_dl_tasks(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
         WARN_ON(!dl_prio(prio));
         WARN_ON(!dl_rq->dl_nr_running);
         dl_rq->dl_nr_running--;
-        dec_nr_running(rq_of_dl_rq(dl_rq));
+        sub_nr_running(rq_of_dl_rq(dl_rq), 1);
 
         dec_dl_deadline(dl_rq, dl_se->deadline);
         dec_dl_migration(dl_se, dl_rq);
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 26ec6686a00b..f7cac2ba62ea 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -3325,7 +3325,7 @@ static void throttle_cfs_rq(struct cfs_rq *cfs_rq)
         }
 
         if (!se)
-                rq->nr_running -= task_delta;
+                sub_nr_running(rq, task_delta);
 
         cfs_rq->throttled = 1;
         cfs_rq->throttled_clock = rq_clock(rq);
@@ -3376,7 +3376,7 @@ void unthrottle_cfs_rq(struct cfs_rq *cfs_rq)
         }
 
         if (!se)
-                rq->nr_running += task_delta;
+                add_nr_running(rq, task_delta);
 
         /* determine whether we need to wake up potentially idle cpu */
         if (rq->curr == rq->idle && rq->cfs.nr_running)
@@ -3908,7 +3908,7 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
 
         if (!se) {
                 update_rq_runnable_avg(rq, rq->nr_running);
-                inc_nr_running(rq);
+                add_nr_running(rq, 1);
         }
         hrtick_update(rq);
 }
@@ -3968,7 +3968,7 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
         }
 
         if (!se) {
-                dec_nr_running(rq);
+                sub_nr_running(rq, 1);
                 update_rq_runnable_avg(rq, 1);
         }
         hrtick_update(rq);
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index 7795e292f4c9..0ebfd7a29472 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -973,7 +973,7 @@ dequeue_top_rt_rq(struct rt_rq *rt_rq)
 
         BUG_ON(!rq->nr_running);
 
-        rq->nr_running -= rt_rq->rt_nr_running;
+        sub_nr_running(rq, rt_rq->rt_nr_running);
         rt_rq->rt_queued = 0;
 }
 
@@ -989,7 +989,7 @@ enqueue_top_rt_rq(struct rt_rq *rt_rq)
         if (rt_rq_throttled(rt_rq) || !rt_rq->rt_nr_running)
                 return;
 
-        rq->nr_running += rt_rq->rt_nr_running;
+        add_nr_running(rq, rt_rq->rt_nr_running);
         rt_rq->rt_queued = 1;
 }
 
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index b2cbe81308af..600e2291a75c 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1206,12 +1206,14 @@ extern void update_idle_cpu_load(struct rq *this_rq);
 
 extern void init_task_runnable_average(struct task_struct *p);
 
-static inline void inc_nr_running(struct rq *rq)
+static inline void add_nr_running(struct rq *rq, unsigned count)
 {
-        rq->nr_running++;
+        unsigned prev_nr = rq->nr_running;
+
+        rq->nr_running = prev_nr + count;
 
 #ifdef CONFIG_NO_HZ_FULL
-        if (rq->nr_running == 2) {
+        if (prev_nr < 2 && rq->nr_running >= 2) {
                 if (tick_nohz_full_cpu(rq->cpu)) {
                         /* Order rq->nr_running write against the IPI */
                         smp_wmb();
@@ -1221,9 +1223,9 @@ static inline void inc_nr_running(struct rq *rq)
 #endif
 }
 
-static inline void dec_nr_running(struct rq *rq)
+static inline void sub_nr_running(struct rq *rq, unsigned count)
 {
-        rq->nr_running--;
+        rq->nr_running -= count;
 }
 
 static inline void rq_last_tick_reset(struct rq *rq)
diff --git a/kernel/sched/stop_task.c b/kernel/sched/stop_task.c
index d6ce65dde541..bfe0edadbfbb 100644
--- a/kernel/sched/stop_task.c
+++ b/kernel/sched/stop_task.c
@@ -41,13 +41,13 @@ pick_next_task_stop(struct rq *rq, struct task_struct *prev)
 static void
 enqueue_task_stop(struct rq *rq, struct task_struct *p, int flags)
 {
-        inc_nr_running(rq);
+        add_nr_running(rq, 1);
 }
 
 static void
 dequeue_task_stop(struct rq *rq, struct task_struct *p, int flags)
 {
-        dec_nr_running(rq);
+        sub_nr_running(rq, 1);
 }
 
 static void yield_task_stop(struct rq *rq)