lkml.org 
[lkml]   [2011]   [Jan]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[34-longterm 094/260] sched: Pre-compute cpumask_weight(sched_domain_span(sd))
Date
From: Peter Zijlstra <a.p.zijlstra@chello.nl>

commit 669c55e9f99b90e46eaa0f98a67ec53d46dc969a upstream.

Dave reported that his large SPARC machines spend lots of time in
hweight64(), try and optimize some of those needless cpumask_weight()
invocations (esp. with the large offstack cpumasks these are very
expensive indeed).

Reported-by: David Miller <davem@davemloft.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
include/linux/sched.h | 1 +
kernel/sched.c | 3 +++
kernel/sched_fair.c | 12 +++++-------
3 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 25f9669..89e0fee 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -954,6 +954,7 @@ struct sched_domain {
char *name;
#endif

+ unsigned int span_weight;
/*
* Span of all CPUs in this domain.
*
diff --git a/kernel/sched.c b/kernel/sched.c
index 074c4d8..b741de5 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -6296,6 +6296,9 @@ cpu_attach_domain(struct sched_domain *sd, struct root_domain *rd, int cpu)
struct rq *rq = cpu_rq(cpu);
struct sched_domain *tmp;

+ for (tmp = sd; tmp; tmp = tmp->parent)
+ tmp->span_weight = cpumask_weight(sched_domain_span(tmp));
+
/* Remove the sched domains which do not contribute to scheduling. */
for (tmp = sd; tmp; ) {
struct sched_domain *parent = tmp->parent;
diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index ae4d842..a29df86 100644
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -1542,9 +1542,7 @@ select_task_rq_fair(struct rq *rq, struct task_struct *p, int sd_flag, int wake_
* Pick the largest domain to update shares over
*/
tmp = sd;
- if (affine_sd && (!tmp ||
- cpumask_weight(sched_domain_span(affine_sd)) >
- cpumask_weight(sched_domain_span(sd))))
+ if (affine_sd && (!tmp || affine_sd->span_weight > sd->span_weight))
tmp = affine_sd;

if (tmp) {
@@ -1588,10 +1586,10 @@ select_task_rq_fair(struct rq *rq, struct task_struct *p, int sd_flag, int wake_

/* Now try balancing at a lower domain level of new_cpu */
cpu = new_cpu;
- weight = cpumask_weight(sched_domain_span(sd));
+ weight = sd->span_weight;
sd = NULL;
for_each_domain(cpu, tmp) {
- if (weight <= cpumask_weight(sched_domain_span(tmp)))
+ if (weight <= tmp->span_weight)
break;
if (tmp->flags & sd_flag)
sd = tmp;
@@ -2323,7 +2321,7 @@ unsigned long __weak arch_scale_freq_power(struct sched_domain *sd, int cpu)

unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu)
{
- unsigned long weight = cpumask_weight(sched_domain_span(sd));
+ unsigned long weight = sd->span_weight;
unsigned long smt_gain = sd->smt_gain;

smt_gain /= weight;
@@ -2356,7 +2354,7 @@ unsigned long scale_rt_power(int cpu)

static void update_cpu_power(struct sched_domain *sd, int cpu)
{
- unsigned long weight = cpumask_weight(sched_domain_span(sd));
+ unsigned long weight = sd->span_weight;
unsigned long power = SCHED_LOAD_SCALE;
struct sched_group *sdg = sd->groups;

--
1.7.3.3


\
 
 \ /
  Last update: 2011-01-02 09:01    [W:0.858 / U:0.244 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site