Commit-ID: 0e5b5337f0da073e1f17aec3c322ea7826975d0d
Gitweb: http://git.kernel.org/tip/0e5b5337f0da073e1f17aec3c322ea7826975d0d
Author: Jason Low <jason.low2@xxxxxx>
AuthorDate: Mon, 28 Apr 2014 15:45:54 -0700
Committer: Ingo Molnar <mingo@xxxxxxxxxx>
CommitDate: Wed, 7 May 2014 11:51:36 +0200
sched: Fix updating rq->max_idle_balance_cost and rq->next_balance in idle_balance()
The following commit:
e5fc66119ec9 ("sched: Fix race in idle_balance()")
can potentially cause rq->max_idle_balance_cost to not be updated,
even when load_balance(NEWLY_IDLE) is attempted and the per-sd
max cost value is updated.
Preeti noticed a similar issue with updating rq->next_balance.
In this patch, we fix this by making sure we still check/update those values
even if a task gets enqueued while browsing the domains.
Signed-off-by: Jason Low <jason.low2@xxxxxx>
Reviewed-by: Preeti U Murthy <preeti@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: morten.rasmussen@xxxxxxx
Cc: aswin@xxxxxx
Cc: daniel.lezcano@xxxxxxxxxx
Cc: alex.shi@xxxxxxxxxx
Cc: efault@xxxxxx
Cc: vincent.guittot@xxxxxxxxxx
Link: http://lkml.kernel.org/r/1398725155-7591-2-git-send-email-jason.low2@xxxxxx
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
---
kernel/sched/fair.c | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 7570dd9..0fdb96d 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6653,6 +6653,7 @@ static int idle_balance(struct rq *this_rq)
int this_cpu = this_rq->cpu;
idle_enter_fair(this_rq);
+
/*
* We must set idle_stamp _before_ calling idle_balance(), such that we
* measure the duration of idle_balance() as idle time.
@@ -6705,14 +6706,16 @@ static int idle_balance(struct rq *this_rq)
raw_spin_lock(&this_rq->lock);
+ if (curr_cost > this_rq->max_idle_balance_cost)
+ this_rq->max_idle_balance_cost = curr_cost;
+
/*
- * While browsing the domains, we released the rq lock.
- * A task could have be enqueued in the meantime
+ * While browsing the domains, we released the rq lock, a task could
+ * have been enqueued in the meantime. Since we're not going idle,
+ * pretend we pulled a task.
*/
- if (this_rq->cfs.h_nr_running && !pulled_task) {
+ if (this_rq->cfs.h_nr_running && !pulled_task)
pulled_task = 1;
- goto out;
- }
if (pulled_task || time_after(jiffies, this_rq->next_balance)) {
/*
@@ -6722,9 +6725,6 @@ static int idle_balance(struct rq *this_rq)
this_rq->next_balance = next_balance;
}
- if (curr_cost > this_rq->max_idle_balance_cost)
- this_rq->max_idle_balance_cost = curr_cost;
-
out:
/* Is there a task of a high priority class? */
if (this_rq->nr_running != this_rq->cfs.h_nr_running &&
--
To unsubscribe from this list: send the line "unsubscribe linux-tip-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
|