[PATCH] cpuidle: poll_state: Avoid invoking local_clock() too often

[PATCH] cpuidle: poll_state: Avoid invoking local_clock() too often

[Rafael J. Wysocki]

From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Rik reports that he sees an increase in CPU use in one benchmark
due to commit 612f1a22f067 "cpuidle: poll_state: Add time limit to
poll_idle()" that caused poll_idle() to call local_clock() in every
iteration of the loop.  Utilization increase generally means more
non-idle time with respect to total CPU time (on the average) which
implies reduced CPU frequency.

Doug reports that limiting the rate of local_clock() invocations
in there causes much less power to be drawn during a CPU-intensive
parallel workload (with idle states 1 and 2 disabled to enforce more
state 0 residency).

These two reports together suggest that executing local_clock() on
multiple CPUs in parallel at a high rate may cause chips to get hot
and trigger thermal/power limits on them to kick in, so reduce the
rate of local_clock() invocations in poll_idle() to avoid that issue.

Fixes: 612f1a22f067 "cpuidle: poll_state: Add time limit to poll_idle()"
Reported-by: Rik van Riel <riel@surriel.com>
Reported-by: Doug Smythies <dsmythies@telus.net>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

I've settled for POLL_IDLE_RELAX_COUNT = 200 after quite a bit of
back-and-forth and a number of test runs.

It may need to be refined going forward if somebody has a problem with
the current one.

---
 drivers/cpuidle/poll_state.c |    6 ++++++
 1 file changed, 6 insertions(+)

Index: linux-pm/drivers/cpuidle/poll_state.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/poll_state.c
+++ linux-pm/drivers/cpuidle/poll_state.c
@@ -10,6 +10,7 @@
 #include <linux/sched/idle.h>
 
 #define POLL_IDLE_TIME_LIMIT	(TICK_NSEC / 16)
+#define POLL_IDLE_RELAX_COUNT	200
 
 static int __cpuidle poll_idle(struct cpuidle_device *dev,
 			       struct cpuidle_driver *drv, int index)
@@ -18,9 +19,14 @@ static int __cpuidle poll_idle(struct cp
 
 	local_irq_enable();
 	if (!current_set_polling_and_test()) {
+		unsigned int loop_count = 0;
+
 		while (!need_resched()) {
 			cpu_relax();
+			if (loop_count++ < POLL_IDLE_RELAX_COUNT)
+				continue;
 
+			loop_count = 0;
 			if (local_clock() - time_start > POLL_IDLE_TIME_LIMIT)
 				break;
 		}

Re: [PATCH] cpuidle: poll_state: Avoid invoking local_clock() too often

[Rik van Riel]

[-- Attachment #1: Type: text/plain, Size: 1414 bytes --]

On Tue, 2018-03-27 at 23:58 +0200, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> Rik reports that he sees an increase in CPU use in one benchmark
> due to commit 612f1a22f067 "cpuidle: poll_state: Add time limit to
> poll_idle()" that caused poll_idle() to call local_clock() in every
> iteration of the loop.  Utilization increase generally means more
> non-idle time with respect to total CPU time (on the average) which
> implies reduced CPU frequency.
> 
> Doug reports that limiting the rate of local_clock() invocations
> in there causes much less power to be drawn during a CPU-intensive
> parallel workload (with idle states 1 and 2 disabled to enforce more
> state 0 residency).
> 
> These two reports together suggest that executing local_clock() on
> multiple CPUs in parallel at a high rate may cause chips to get hot
> and trigger thermal/power limits on them to kick in, so reduce the
> rate of local_clock() invocations in poll_idle() to avoid that issue.
> 
> Fixes: 612f1a22f067 "cpuidle: poll_state: Add time limit to
> poll_idle()"
> Reported-by: Rik van Riel <riel@surriel.com>
> Reported-by: Doug Smythies <dsmythies@telus.net>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Thanks Rafael!

Tested-by: Rik van Riel <riel@surriel.com>
Reviewed-by: Rik van Riel <riel@surriel.com>

-- 
All Rights Reversed.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 488 bytes --]