sched: clean up new task placement
authorPeter Zijlstra <a.p.zijlstra@chello.nl>
Mon, 15 Oct 2007 15:00:05 +0000 (17:00 +0200)
committerIngo Molnar <mingo@elte.hu>
Mon, 15 Oct 2007 15:00:05 +0000 (17:00 +0200)
clean up new task placement.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Mike Galbraith <efault@gmx.de>
kernel/sched_fair.c

index a60b1da..cc447fb 100644 (file)
@@ -199,6 +199,21 @@ static struct sched_entity *__pick_next_entity(struct cfs_rq *cfs_rq)
        return rb_entry(first_fair(cfs_rq), struct sched_entity, run_node);
 }
 
+static inline struct sched_entity *__pick_last_entity(struct cfs_rq *cfs_rq)
+{
+       struct rb_node **link = &cfs_rq->tasks_timeline.rb_node;
+       struct sched_entity *se = NULL;
+       struct rb_node *parent;
+
+       while (*link) {
+               parent = *link;
+               se = rb_entry(parent, struct sched_entity, run_node);
+               link = &parent->rb_right;
+       }
+
+       return se;
+}
+
 /**************************************************************
  * Scheduling class statistics methods:
  */
@@ -530,6 +545,31 @@ static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se)
 }
 
 static void
+place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial)
+{
+       struct sched_entity *last = __pick_last_entity(cfs_rq);
+       u64 min_runtime, latency;
+
+       min_runtime = cfs_rq->min_vruntime;
+       if (last) {
+               min_runtime += last->vruntime;
+               min_runtime >>= 1;
+               if (initial && sched_feat(START_DEBIT))
+                       min_runtime += sysctl_sched_latency/2;
+       }
+
+       if (!initial && sched_feat(NEW_FAIR_SLEEPERS)) {
+               latency = sysctl_sched_latency;
+               if (min_runtime > latency)
+                       min_runtime -= latency;
+               else
+                       min_runtime = 0;
+       }
+
+       se->vruntime = max(se->vruntime, min_runtime);
+}
+
+static void
 enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int wakeup)
 {
        /*
@@ -538,19 +578,7 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int wakeup)
        update_curr(cfs_rq);
 
        if (wakeup) {
-               u64 min_runtime, latency;
-
-               min_runtime = cfs_rq->min_vruntime;
-               min_runtime += sysctl_sched_latency/2;
-
-               if (sched_feat(NEW_FAIR_SLEEPERS)) {
-                       latency = calc_weighted(sysctl_sched_latency, se);
-                       if (min_runtime > latency)
-                               min_runtime -= latency;
-               }
-
-               se->vruntime = max(se->vruntime, min_runtime);
-
+               place_entity(cfs_rq, se, 0);
                enqueue_sleeper(cfs_rq, se);
        }
 
@@ -1033,8 +1061,7 @@ static void task_new_fair(struct rq *rq, struct task_struct *p)
        sched_info_queued(p);
 
        update_curr(cfs_rq);
-       se->vruntime = cfs_rq->min_vruntime;
-       update_stats_enqueue(cfs_rq, se);
+       place_entity(cfs_rq, se, 1);
 
        /*
         * The first wait is dominated by the child-runs-first logic,