sched: make the scheduler converge to the ideal latency
de-HZ-ification of the granularity defaults unearthed a pre-existing
property of CFS: while it correctly converges to the granularity goal,
it does not prevent run-time fluctuations in the range of
[-gran ... 0 ... +gran].
With the increase of the granularity due to the removal of HZ
dependencies, this becomes visible in chew-max output (with 5 tasks
running):
out: 28 . 27. 32 | flu: 0 . 0 | ran: 9 . 13 | per: 37 . 40
out: 27 . 27. 32 | flu: 0 . 0 | ran: 17 . 13 | per: 44 . 40
out: 27 . 27. 32 | flu: 0 . 0 | ran: 9 . 13 | per: 36 . 40
out: 29 . 27. 32 | flu: 2 . 0 | ran: 17 . 13 | per: 46 . 40
out: 28 . 27. 32 | flu: 0 . 0 | ran: 9 . 13 | per: 37 . 40
out: 29 . 27. 32 | flu: 0 . 0 | ran: 18 . 13 | per: 47 . 40
out: 28 . 27. 32 | flu: 0 . 0 | ran: 9 . 13 | per: 37 . 40
average slice is the ideal 13 msecs and the period is picture-perfect 40
msecs. But the 'ran' field fluctuates around 13.33 msecs and there's no
mechanism in CFS to keep that from happening: it's a perfectly valid
solution that CFS finds.
to fix this we add a granularity/preemption rule that knows about
the "target latency", which makes tasks that run longer than the ideal
latency run a bit less. The simplest approach is to simply decrease the
preemption granularity when a task overruns its ideal latency. For this
we have to track how much the task executed since its last preemption.
( this adds a new field to task_struct, but we can eliminate that
overhead in 2.6.24 by putting all the scheduler timestamps into an
anonymous union. )
with this change in place, chew-max output is fluctuation-less all
around:
out: 28 . 27. 39 | flu: 0 . 2 | ran: 13 . 13 | per: 41 . 40
out: 28 . 27. 39 | flu: 0 . 2 | ran: 13 . 13 | per: 41 . 40
out: 28 . 27. 39 | flu: 0 . 2 | ran: 13 . 13 | per: 41 . 40
out: 28 . 27. 39 | flu: 0 . 2 | ran: 13 . 13 | per: 41 . 40
out: 28 . 27. 39 | flu: 0 . 1 | ran: 13 . 13 | per: 41 . 40
out: 28 . 27. 39 | flu: 0 . 1 | ran: 13 . 13 | per: 41 . 40
this patch has no impact on any fastpath or on any globally observable
scheduling property. (unless you have sharp enough eyes to see
millisecond-level ruckles in glxgears smoothness :-)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Mike Galbraith <efault@gmx.de>
diff --git a/include/linux/sched.h b/include/linux/sched.h
index bd6a032..f4e324e 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -904,6 +904,7 @@
u64 exec_start;
u64 sum_exec_runtime;
+ u64 prev_sum_exec_runtime;
u64 wait_start_fair;
u64 sleep_start_fair;
diff --git a/kernel/sched.c b/kernel/sched.c
index 9fe473a..b533d6d 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -1587,6 +1587,7 @@
p->se.wait_start_fair = 0;
p->se.exec_start = 0;
p->se.sum_exec_runtime = 0;
+ p->se.prev_sum_exec_runtime = 0;
p->se.delta_exec = 0;
p->se.delta_fair_run = 0;
p->se.delta_fair_sleep = 0;
diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index 9f53d49..721fe77 100644
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -668,7 +668,7 @@
/*
* Preempt the current task with a newly woken task if needed:
*/
-static void
+static int
__check_preempt_curr_fair(struct cfs_rq *cfs_rq, struct sched_entity *se,
struct sched_entity *curr, unsigned long granularity)
{
@@ -679,8 +679,11 @@
* preempt the current task unless the best task has
* a larger than sched_granularity fairness advantage:
*/
- if (__delta > niced_granularity(curr, granularity))
+ if (__delta > niced_granularity(curr, granularity)) {
resched_task(rq_of(cfs_rq)->curr);
+ return 1;
+ }
+ return 0;
}
static inline void
@@ -725,6 +728,7 @@
static void entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr)
{
+ unsigned long gran, ideal_runtime, delta_exec;
struct sched_entity *next;
/*
@@ -741,8 +745,22 @@
if (next == curr)
return;
- __check_preempt_curr_fair(cfs_rq, next, curr,
- sched_granularity(cfs_rq));
+ gran = sched_granularity(cfs_rq);
+ ideal_runtime = niced_granularity(curr,
+ max(sysctl_sched_latency / cfs_rq->nr_running,
+ (unsigned long)sysctl_sched_min_granularity));
+ /*
+ * If we executed more than what the latency constraint suggests,
+ * reduce the rescheduling granularity. This way the total latency
+ * of how much a task is not scheduled converges to
+ * sysctl_sched_latency:
+ */
+ delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime;
+ if (delta_exec > ideal_runtime)
+ gran = 0;
+
+ if (__check_preempt_curr_fair(cfs_rq, next, curr, gran))
+ curr->prev_sum_exec_runtime = curr->sum_exec_runtime;
}
/**************************************************
