From: Rafael J. Wysocki <[email protected]>
Subject: [PATCH] cpuidle: menu: Skip tick_nohz_get_sleep_length() call in some cases
Because the cost of calling tick_nohz_get_sleep_length() may increase
in the future, reorder the code in menu_select() so it first uses the
statistics to determine the expected idle duration. If that value is
higher than RESIDENCY_THRESHOLD_NS, tick_nohz_get_sleep_length() will
be called to obtain the time till the closest timer and refine the
idle duration prediction if necessary.
This causes the governor to always take the full overhead of
get_typical_interval() with the assumption that the cost will be
amortized by skipping the tick_nohz_get_sleep_length() call in the
cases when the predicted idle duration is relatively very small.
Signed-off-by: Rafael J. Wysocki <[email protected]>
---
v1 -> v2: Add missing max check to get_typical_interval().
---
drivers/cpuidle/governors/gov.h | 14 ++++++++
drivers/cpuidle/governors/menu.c | 65 ++++++++++++++++++++++-----------------
drivers/cpuidle/governors/teo.c | 9 +----
3 files changed, 54 insertions(+), 34 deletions(-)
Index: linux-pm/drivers/cpuidle/governors/menu.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/governors/menu.c
+++ linux-pm/drivers/cpuidle/governors/menu.c
@@ -19,6 +19,8 @@
#include <linux/sched/stat.h>
#include <linux/math64.h>
+#include "gov.h"
+
#define BUCKETS 12
#define INTERVAL_SHIFT 3
#define INTERVALS (1UL << INTERVAL_SHIFT)
@@ -166,8 +168,7 @@ static void menu_update(struct cpuidle_d
* of points is below a threshold. If it is... then use the
* average of these 8 points as the estimated value.
*/
-static unsigned int get_typical_interval(struct menu_device *data,
- unsigned int predicted_us)
+static unsigned int get_typical_interval(struct menu_device *data)
{
int i, divisor;
unsigned int min, max, thresh, avg;
@@ -195,11 +196,7 @@ again:
}
}
- /*
- * If the result of the computation is going to be discarded anyway,
- * avoid the computation altogether.
- */
- if (min >= predicted_us)
+ if (!max)
return UINT_MAX;
if (divisor == INTERVALS)
@@ -267,7 +264,6 @@ static int menu_select(struct cpuidle_dr
{
struct menu_device *data = this_cpu_ptr(&menu_devices);
s64 latency_req = cpuidle_governor_latency_req(dev->cpu);
- unsigned int predicted_us;
u64 predicted_ns;
u64 interactivity_req;
unsigned int nr_iowaiters;
@@ -279,16 +275,41 @@ static int menu_select(struct cpuidle_dr
data->needs_update = 0;
}
- /* determine the expected residency time, round up */
- delta = tick_nohz_get_sleep_length(&delta_tick);
- if (unlikely(delta < 0)) {
- delta = 0;
- delta_tick = 0;
- }
- data->next_timer_ns = delta;
-
nr_iowaiters = nr_iowait_cpu(dev->cpu);
- data->bucket = which_bucket(data->next_timer_ns, nr_iowaiters);
+
+ /* Find the shortest expected idle interval. */
+ predicted_ns = get_typical_interval(data) * NSEC_PER_USEC;
+ if (predicted_ns > RESIDENCY_THRESHOLD_NS) {
+ unsigned int timer_us;
+
+ /* Determine the time till the closest timer. */
+ delta = tick_nohz_get_sleep_length(&delta_tick);
+ if (unlikely(delta < 0)) {
+ delta = 0;
+ delta_tick = 0;
+ }
+
+ data->next_timer_ns = delta;
+ data->bucket = which_bucket(data->next_timer_ns, nr_iowaiters);
+
+ /* Round up the result for half microseconds. */
+ timer_us = div_u64((RESOLUTION * DECAY * NSEC_PER_USEC) / 2 +
+ data->next_timer_ns *
+ data->correction_factor[data->bucket],
+ RESOLUTION * DECAY * NSEC_PER_USEC);
+ /* Use the lowest expected idle interval to pick the idle state. */
+ predicted_ns = min((u64)timer_us * NSEC_PER_USEC, predicted_ns);
+ } else {
+ /*
+ * Because the next timer event is not going to be determined
+ * in this case, assume that without the tick the closest timer
+ * will be in distant future and that the closest tick will occur
+ * after 1/2 of the tick period.
+ */
+ data->next_timer_ns = KTIME_MAX;
+ delta_tick = TICK_NSEC / 2;
+ data->bucket = which_bucket(KTIME_MAX, nr_iowaiters);
+ }
if (unlikely(drv->state_count <= 1 || latency_req == 0) ||
((data->next_timer_ns < drv->states[1].target_residency_ns ||
@@ -303,16 +324,6 @@ static int menu_select(struct cpuidle_dr
return 0;
}
- /* Round up the result for half microseconds. */
- predicted_us = div_u64(data->next_timer_ns *
- data->correction_factor[data->bucket] +
- (RESOLUTION * DECAY * NSEC_PER_USEC) / 2,
- RESOLUTION * DECAY * NSEC_PER_USEC);
- /* Use the lowest expected idle interval to pick the idle state. */
- predicted_ns = (u64)min(predicted_us,
- get_typical_interval(data, predicted_us)) *
- NSEC_PER_USEC;
-
if (tick_nohz_tick_stopped()) {
/*
* If the tick is already stopped, the cost of possible short
Index: linux-pm/drivers/cpuidle/governors/gov.h
===================================================================
--- /dev/null
+++ linux-pm/drivers/cpuidle/governors/gov.h
@@ -0,0 +1,14 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+/* Common definitions for cpuidle governors. */
+
+#ifndef __CPUIDLE_GOVERNOR_H
+#define __CPUIDLE_GOVERNOR_H
+
+/*
+ * Idle state target residency threshold used for deciding whether or not to
+ * check the time till the closest expected timer event.
+ */
+#define RESIDENCY_THRESHOLD_NS (15 * NSEC_PER_USEC)
+
+#endif /* __CPUIDLE_GOVERNOR_H */
Index: linux-pm/drivers/cpuidle/governors/teo.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/governors/teo.c
+++ linux-pm/drivers/cpuidle/governors/teo.c
@@ -140,6 +140,8 @@
#include <linux/sched/topology.h>
#include <linux/tick.h>
+#include "gov.h"
+
/*
* The number of bits to shift the CPU's capacity by in order to determine
* the utilized threshold.
@@ -152,7 +154,6 @@
*/
#define UTIL_THRESHOLD_SHIFT 6
-
/*
* The PULSE value is added to metrics when they grow and the DECAY_SHIFT value
* is used for decreasing metrics on a regular basis.
@@ -166,12 +167,6 @@
*/
#define NR_RECENT 9
-/*
- * Idle state target residency threshold used for deciding whether or not to
- * check the time till the closest expected timer event.
- */
-#define RESIDENCY_THRESHOLD_NS (15 * NSEC_PER_USEC)
-
/**
* struct teo_bin - Metrics used by the TEO cpuidle governor.
* @intercepts: The "intercepts" metric.
On Thu, Aug 10, 2023 at 11:38 AM Rafael J. Wysocki <[email protected]> wrote:
>
> From: Rafael J. Wysocki <[email protected]>
> Subject: [PATCH] cpuidle: menu: Skip tick_nohz_get_sleep_length() call in some cases
>
> Because the cost of calling tick_nohz_get_sleep_length() may increase
> in the future, reorder the code in menu_select() so it first uses the
> statistics to determine the expected idle duration. If that value is
> higher than RESIDENCY_THRESHOLD_NS, tick_nohz_get_sleep_length() will
> be called to obtain the time till the closest timer and refine the
> idle duration prediction if necessary.
>
> This causes the governor to always take the full overhead of
> get_typical_interval() with the assumption that the cost will be
> amortized by skipping the tick_nohz_get_sleep_length() call in the
> cases when the predicted idle duration is relatively very small.
>
> Signed-off-by: Rafael J. Wysocki <[email protected]>
I ran the same tests as I did for the teo governor changes.
Nothing of significance to report (well, some minor improvements
in records per second at the long interval end of the sleeping
ebizzy test, with a corresponding slight increase in processor
package power).
Tested-by: Doug Smythies <[email protected]>
> ---
>
> v1 -> v2: Add missing max check to get_typical_interval().
>
... Doug
On Mon, Aug 14, 2023 at 4:52 PM Doug Smythies <[email protected]> wrote:
>
> On Thu, Aug 10, 2023 at 11:38 AM Rafael J. Wysocki <[email protected]> wrote:
> >
> > From: Rafael J. Wysocki <[email protected]>
> > Subject: [PATCH] cpuidle: menu: Skip tick_nohz_get_sleep_length() call in some cases
> >
> > Because the cost of calling tick_nohz_get_sleep_length() may increase
> > in the future, reorder the code in menu_select() so it first uses the
> > statistics to determine the expected idle duration. If that value is
> > higher than RESIDENCY_THRESHOLD_NS, tick_nohz_get_sleep_length() will
> > be called to obtain the time till the closest timer and refine the
> > idle duration prediction if necessary.
> >
> > This causes the governor to always take the full overhead of
> > get_typical_interval() with the assumption that the cost will be
> > amortized by skipping the tick_nohz_get_sleep_length() call in the
> > cases when the predicted idle duration is relatively very small.
> >
> > Signed-off-by: Rafael J. Wysocki <[email protected]>
>
> I ran the same tests as I did for the teo governor changes.
> Nothing of significance to report (well, some minor improvements
> in records per second at the long interval end of the sleeping
> ebizzy test, with a corresponding slight increase in processor
> package power).
>
> Tested-by: Doug Smythies <[email protected]>
Thank you!