Since this had me looking at all that code, I did the below.
Holler if there's objections etc..
---
Subject: sched/fair: Remove SIS_PROP
From: Peter Zijlstra <[email protected]>
Date: Fri Oct 20 12:35:33 CEST 2023
SIS_UTIL seems to work well, lets remove the old thing.
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
---
include/linux/sched/topology.h | 2 -
kernel/sched/core.c | 5 ----
kernel/sched/fair.c | 48 -----------------------------------------
kernel/sched/features.h | 1
kernel/sched/sched.h | 3 --
5 files changed, 59 deletions(-)
--- a/include/linux/sched/topology.h
+++ b/include/linux/sched/topology.h
@@ -109,8 +109,6 @@ struct sched_domain {
u64 max_newidle_lb_cost;
unsigned long last_decay_max_lb_cost;
- u64 avg_scan_cost; /* select_idle_sibling */
-
#ifdef CONFIG_SCHEDSTATS
/* load_balance() stats */
unsigned int lb_count[CPU_MAX_IDLE_TYPES];
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3792,9 +3792,6 @@ ttwu_do_activate(struct rq *rq, struct t
if (rq->avg_idle > max)
rq->avg_idle = max;
- rq->wake_stamp = jiffies;
- rq->wake_avg_idle = rq->avg_idle / 2;
-
rq->idle_stamp = 0;
}
#endif
@@ -9991,8 +9988,6 @@ void __init sched_init(void)
rq->online = 0;
rq->idle_stamp = 0;
rq->avg_idle = 2*sysctl_sched_migration_cost;
- rq->wake_stamp = jiffies;
- rq->wake_avg_idle = rq->avg_idle;
rq->max_idle_balance_cost = sysctl_sched_migration_cost;
INIT_LIST_HEAD(&rq->cfs_tasks);
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7209,45 +7209,9 @@ static int select_idle_cpu(struct task_s
struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_rq_mask);
int i, cpu, idle_cpu = -1, nr = INT_MAX;
struct sched_domain_shared *sd_share;
- struct rq *this_rq = this_rq();
- int this = smp_processor_id();
- struct sched_domain *this_sd = NULL;
- u64 time = 0;
cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
- if (sched_feat(SIS_PROP) && !has_idle_core) {
- u64 avg_cost, avg_idle, span_avg;
- unsigned long now = jiffies;
-
- this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
- if (!this_sd)
- return -1;
-
- /*
- * If we're busy, the assumption that the last idle period
- * predicts the future is flawed; age away the remaining
- * predicted idle time.
- */
- if (unlikely(this_rq->wake_stamp < now)) {
- while (this_rq->wake_stamp < now && this_rq->wake_avg_idle) {
- this_rq->wake_stamp++;
- this_rq->wake_avg_idle >>= 1;
- }
- }
-
- avg_idle = this_rq->wake_avg_idle;
- avg_cost = this_sd->avg_scan_cost + 1;
-
- span_avg = sd->span_weight * avg_idle;
- if (span_avg > 4*avg_cost)
- nr = div_u64(span_avg, avg_cost);
- else
- nr = 4;
-
- time = cpu_clock(this);
- }
-
if (sched_feat(SIS_UTIL)) {
sd_share = rcu_dereference(per_cpu(sd_llc_shared, target));
if (sd_share) {
@@ -7301,18 +7265,6 @@ static int select_idle_cpu(struct task_s
if (has_idle_core)
set_idle_cores(target, false);
- if (sched_feat(SIS_PROP) && this_sd && !has_idle_core) {
- time = cpu_clock(this) - time;
-
- /*
- * Account for the scan cost of wakeups against the average
- * idle time.
- */
- this_rq->wake_avg_idle -= min(this_rq->wake_avg_idle, time);
-
- update_avg(&this_sd->avg_scan_cost, time);
- }
-
return idle_cpu;
}
--- a/kernel/sched/features.h
+++ b/kernel/sched/features.h
@@ -49,7 +49,6 @@ SCHED_FEAT(TTWU_QUEUE, true)
/*
* When doing wakeups, attempt to limit superfluous scans of the LLC domain.
*/
-SCHED_FEAT(SIS_PROP, false)
SCHED_FEAT(SIS_UTIL, true)
/*
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1059,9 +1059,6 @@ struct rq {
u64 idle_stamp;
u64 avg_idle;
- unsigned long wake_stamp;
- u64 wake_avg_idle;
-
/* This is used to determine avg_idle's max value */
u64 max_idle_balance_cost;
On Fri, 20 Oct 2023 at 15:44, Peter Zijlstra <[email protected]> wrote:
>
>
>
> Since this had me looking at all that code, I did the below.
>
> Holler if there's objections etc..
>
> ---
> Subject: sched/fair: Remove SIS_PROP
> From: Peter Zijlstra <[email protected]>
> Date: Fri Oct 20 12:35:33 CEST 2023
>
> SIS_UTIL seems to work well, lets remove the old thing.
>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Vincent Guittot <[email protected]>
> ---
> include/linux/sched/topology.h | 2 -
> kernel/sched/core.c | 5 ----
> kernel/sched/fair.c | 48 -----------------------------------------
> kernel/sched/features.h | 1
> kernel/sched/sched.h | 3 --
> 5 files changed, 59 deletions(-)
>
> --- a/include/linux/sched/topology.h
> +++ b/include/linux/sched/topology.h
> @@ -109,8 +109,6 @@ struct sched_domain {
> u64 max_newidle_lb_cost;
> unsigned long last_decay_max_lb_cost;
>
> - u64 avg_scan_cost; /* select_idle_sibling */
> -
> #ifdef CONFIG_SCHEDSTATS
> /* load_balance() stats */
> unsigned int lb_count[CPU_MAX_IDLE_TYPES];
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -3792,9 +3792,6 @@ ttwu_do_activate(struct rq *rq, struct t
> if (rq->avg_idle > max)
> rq->avg_idle = max;
>
> - rq->wake_stamp = jiffies;
> - rq->wake_avg_idle = rq->avg_idle / 2;
> -
> rq->idle_stamp = 0;
> }
> #endif
> @@ -9991,8 +9988,6 @@ void __init sched_init(void)
> rq->online = 0;
> rq->idle_stamp = 0;
> rq->avg_idle = 2*sysctl_sched_migration_cost;
> - rq->wake_stamp = jiffies;
> - rq->wake_avg_idle = rq->avg_idle;
> rq->max_idle_balance_cost = sysctl_sched_migration_cost;
>
> INIT_LIST_HEAD(&rq->cfs_tasks);
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -7209,45 +7209,9 @@ static int select_idle_cpu(struct task_s
> struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_rq_mask);
> int i, cpu, idle_cpu = -1, nr = INT_MAX;
> struct sched_domain_shared *sd_share;
> - struct rq *this_rq = this_rq();
> - int this = smp_processor_id();
> - struct sched_domain *this_sd = NULL;
> - u64 time = 0;
>
> cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
>
> - if (sched_feat(SIS_PROP) && !has_idle_core) {
> - u64 avg_cost, avg_idle, span_avg;
> - unsigned long now = jiffies;
> -
> - this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
> - if (!this_sd)
> - return -1;
> -
> - /*
> - * If we're busy, the assumption that the last idle period
> - * predicts the future is flawed; age away the remaining
> - * predicted idle time.
> - */
> - if (unlikely(this_rq->wake_stamp < now)) {
> - while (this_rq->wake_stamp < now && this_rq->wake_avg_idle) {
> - this_rq->wake_stamp++;
> - this_rq->wake_avg_idle >>= 1;
> - }
> - }
> -
> - avg_idle = this_rq->wake_avg_idle;
> - avg_cost = this_sd->avg_scan_cost + 1;
> -
> - span_avg = sd->span_weight * avg_idle;
> - if (span_avg > 4*avg_cost)
> - nr = div_u64(span_avg, avg_cost);
> - else
> - nr = 4;
> -
> - time = cpu_clock(this);
> - }
> -
> if (sched_feat(SIS_UTIL)) {
> sd_share = rcu_dereference(per_cpu(sd_llc_shared, target));
> if (sd_share) {
> @@ -7301,18 +7265,6 @@ static int select_idle_cpu(struct task_s
> if (has_idle_core)
> set_idle_cores(target, false);
>
> - if (sched_feat(SIS_PROP) && this_sd && !has_idle_core) {
> - time = cpu_clock(this) - time;
> -
> - /*
> - * Account for the scan cost of wakeups against the average
> - * idle time.
> - */
> - this_rq->wake_avg_idle -= min(this_rq->wake_avg_idle, time);
> -
> - update_avg(&this_sd->avg_scan_cost, time);
> - }
> -
> return idle_cpu;
> }
>
> --- a/kernel/sched/features.h
> +++ b/kernel/sched/features.h
> @@ -49,7 +49,6 @@ SCHED_FEAT(TTWU_QUEUE, true)
> /*
> * When doing wakeups, attempt to limit superfluous scans of the LLC domain.
> */
> -SCHED_FEAT(SIS_PROP, false)
> SCHED_FEAT(SIS_UTIL, true)
>
> /*
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -1059,9 +1059,6 @@ struct rq {
> u64 idle_stamp;
> u64 avg_idle;
>
> - unsigned long wake_stamp;
> - u64 wake_avg_idle;
> -
> /* This is used to determine avg_idle's max value */
> u64 max_idle_balance_cost;
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Fri, Oct 20, 2023 at 03:43:37PM +0200, Peter Zijlstra wrote:
>
>
> Since this had me looking at all that code, I did the below.
>
> Holler if there's objections etc..
>
> ---
> Subject: sched/fair: Remove SIS_PROP
> From: Peter Zijlstra <[email protected]>
> Date: Fri Oct 20 12:35:33 CEST 2023
>
> SIS_UTIL seems to work well, lets remove the old thing.
>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Mel Gorman <[email protected]>
--
Mel Gorman
SUSE Labs
The following commit has been merged into the sched/core branch of tip:
Commit-ID: 984ffb6a4366752c949f7b39640aecdce222607f
Gitweb: https://git.kernel.org/tip/984ffb6a4366752c949f7b39640aecdce222607f
Author: Peter Zijlstra <[email protected]>
AuthorDate: Fri, 20 Oct 2023 12:35:33 +02:00
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Tue, 24 Oct 2023 10:38:44 +02:00
sched/fair: Remove SIS_PROP
SIS_UTIL seems to work well, lets remove the old thing.
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Vincent Guittot <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
include/linux/sched/topology.h | 2 +-
kernel/sched/core.c | 5 +---
kernel/sched/fair.c | 48 +---------------------------------
kernel/sched/features.h | 1 +-
kernel/sched/sched.h | 3 +--
5 files changed, 59 deletions(-)
diff --git a/include/linux/sched/topology.h b/include/linux/sched/topology.h
index 4c14fe1..de545ba 100644
--- a/include/linux/sched/topology.h
+++ b/include/linux/sched/topology.h
@@ -109,8 +109,6 @@ struct sched_domain {
u64 max_newidle_lb_cost;
unsigned long last_decay_max_lb_cost;
- u64 avg_scan_cost; /* select_idle_sibling */
-
#ifdef CONFIG_SCHEDSTATS
/* load_balance() stats */
unsigned int lb_count[CPU_MAX_IDLE_TYPES];
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 5e1fb8a..7a0c161 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3792,9 +3792,6 @@ ttwu_do_activate(struct rq *rq, struct task_struct *p, int wake_flags,
if (rq->avg_idle > max)
rq->avg_idle = max;
- rq->wake_stamp = jiffies;
- rq->wake_avg_idle = rq->avg_idle / 2;
-
rq->idle_stamp = 0;
}
#endif
@@ -9953,8 +9950,6 @@ void __init sched_init(void)
rq->online = 0;
rq->idle_stamp = 0;
rq->avg_idle = 2*sysctl_sched_migration_cost;
- rq->wake_stamp = jiffies;
- rq->wake_avg_idle = rq->avg_idle;
rq->max_idle_balance_cost = sysctl_sched_migration_cost;
INIT_LIST_HEAD(&rq->cfs_tasks);
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 523b5ae..8767988 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7209,45 +7209,9 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, bool
struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_rq_mask);
int i, cpu, idle_cpu = -1, nr = INT_MAX;
struct sched_domain_shared *sd_share;
- struct rq *this_rq = this_rq();
- int this = smp_processor_id();
- struct sched_domain *this_sd = NULL;
- u64 time = 0;
cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
- if (sched_feat(SIS_PROP) && !has_idle_core) {
- u64 avg_cost, avg_idle, span_avg;
- unsigned long now = jiffies;
-
- this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
- if (!this_sd)
- return -1;
-
- /*
- * If we're busy, the assumption that the last idle period
- * predicts the future is flawed; age away the remaining
- * predicted idle time.
- */
- if (unlikely(this_rq->wake_stamp < now)) {
- while (this_rq->wake_stamp < now && this_rq->wake_avg_idle) {
- this_rq->wake_stamp++;
- this_rq->wake_avg_idle >>= 1;
- }
- }
-
- avg_idle = this_rq->wake_avg_idle;
- avg_cost = this_sd->avg_scan_cost + 1;
-
- span_avg = sd->span_weight * avg_idle;
- if (span_avg > 4*avg_cost)
- nr = div_u64(span_avg, avg_cost);
- else
- nr = 4;
-
- time = cpu_clock(this);
- }
-
if (sched_feat(SIS_UTIL)) {
sd_share = rcu_dereference(per_cpu(sd_llc_shared, target));
if (sd_share) {
@@ -7301,18 +7265,6 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, bool
if (has_idle_core)
set_idle_cores(target, false);
- if (sched_feat(SIS_PROP) && this_sd && !has_idle_core) {
- time = cpu_clock(this) - time;
-
- /*
- * Account for the scan cost of wakeups against the average
- * idle time.
- */
- this_rq->wake_avg_idle -= min(this_rq->wake_avg_idle, time);
-
- update_avg(&this_sd->avg_scan_cost, time);
- }
-
return idle_cpu;
}
diff --git a/kernel/sched/features.h b/kernel/sched/features.h
index f770168..a3ddf84 100644
--- a/kernel/sched/features.h
+++ b/kernel/sched/features.h
@@ -49,7 +49,6 @@ SCHED_FEAT(TTWU_QUEUE, true)
/*
* When doing wakeups, attempt to limit superfluous scans of the LLC domain.
*/
-SCHED_FEAT(SIS_PROP, false)
SCHED_FEAT(SIS_UTIL, true)
/*
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index ef4fe7b..2e5a954 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1059,9 +1059,6 @@ struct rq {
u64 idle_stamp;
u64 avg_idle;
- unsigned long wake_stamp;
- u64 wake_avg_idle;
-
/* This is used to determine avg_idle's max value */
u64 max_idle_balance_cost;