sched: Use rq->clock_task instead of rq->clock for correctly maintaining load averages

[linux-flexiantxendom0-natty.git] / kernel / sched_fair.c

diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index 1997383..0c26e2d 100644 (file)
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -725,7 +725,7 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
        if (cfs_rq->tg == &root_task_group)
                return;
 
-       now = rq_of(cfs_rq)->clock;
+       now = rq_of(cfs_rq)->clock_task;
        delta = now - cfs_rq->load_stamp;
 
        /* truncate load history at 4 idle periods */