[PATCH 0/2] sched: misc. fix and cleanup for sched-devel

[PATCH 0/2] sched: misc. fix and cleanup for sched-devel

[Gregory Haskins]

Hi Ingo,
   Here are a few more small patches for consideration in sched-devel.

The second patch should be Ack'd by Steven before accepting to make sure I
didn't misunderstand here...but I believe that logic is now defunct since he
moved away from the overlapped cpuset work some time ago.

Regards,
-Greg

[PATCH 1/2] Subject: SCHED - Only adjust overload state when changing

[Gregory Haskins]

The overload set/clears were originally idempotent when this logic was first
implemented.  But that is no longer true due to the addition of the atomic
counter and this logic was never updated to work properly with that change.
So only adjust the overload state if it is actually changing to avoid
getting out of sync.

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 kernel/sched_rt.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c
index 4cbde83..53cd9e8 100644
--- a/kernel/sched_rt.c
+++ b/kernel/sched_rt.c
@@ -34,9 +34,11 @@ static inline void rt_clear_overload(struct rq *rq)
 static void update_rt_migration(struct rq *rq)
 {
 	if (rq->rt.rt_nr_migratory && (rq->rt.rt_nr_running > 1)) {
-		rt_set_overload(rq);
-		rq->rt.overloaded = 1;
-	} else {
+		if (!rq->rt.overloaded) {
+			rt_set_overload(rq);
+			rq->rt.overloaded = 1;
+		}
+	} else if (rq->rt.overloaded) {
 		rt_clear_overload(rq);
 		rq->rt.overloaded = 0;
 	}

[PATCH 2/2] Subject: SCHED - Clean up some old cpuset logic

[Gregory Haskins]

We had support for overlapping cpuset based rto logic in early prototypes that
is no longer used, so clean it up.

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 kernel/sched_rt.c |   32 --------------------------------
 1 files changed, 0 insertions(+), 32 deletions(-)

diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c
index 53cd9e8..65cbb78 100644
--- a/kernel/sched_rt.c
+++ b/kernel/sched_rt.c
@@ -586,37 +586,6 @@ static int pull_rt_task(struct rq *this_rq)
 			continue;
 
 		src_rq = cpu_rq(cpu);
-		if (unlikely(src_rq->rt.rt_nr_running <= 1)) {
-			/*
-			 * It is possible that overlapping cpusets
-			 * will miss clearing a non overloaded runqueue.
-			 * Clear it now.
-			 */
-			if (double_lock_balance(this_rq, src_rq)) {
-				/* unlocked our runqueue lock */
-				struct task_struct *old_next = next;
-
-				next = pick_next_task_rt(this_rq);
-				if (next != old_next)
-					ret = 1;
-			}
-			if (likely(src_rq->rt.rt_nr_running <= 1)) {
-				/*
-				 * Small chance that this_rq->curr changed
-				 * but it's really harmless here.
-				 */
-				rt_clear_overload(this_rq);
-			} else {
-				/*
-				 * Heh, the src_rq is now overloaded, since
-				 * we already have the src_rq lock, go straight
-				 * to pulling tasks from it.
-				 */
-				goto try_pulling;
-			}
-			spin_unlock(&src_rq->lock);
-			continue;
-		}
 
 		/*
 		 * We can potentially drop this_rq's lock in
@@ -641,7 +610,6 @@ static int pull_rt_task(struct rq *this_rq)
 			continue;
 		}
 
- try_pulling:
 		p = pick_next_highest_task_rt(src_rq, this_cpu);
 
 		/*