public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Ingo Molnar <mingo@elte.hu>
Cc: linux-kernel@vger.kernel.org, Gautham R Shenoy <ego@in.ibm.com>,
	Andreas Herrmann <andreas.herrmann3@amd.com>,
	Balbir Singh <balbir@in.ibm.com>,
	Peter Zijlstra <a.p.zijlstra@chello.nl>
Subject: [RFC][PATCH 14/14] sched: cleanup wake_idle
Date: Thu, 03 Sep 2009 15:21:59 +0200	[thread overview]
Message-ID: <20090903132213.413056069@chello.nl> (raw)
In-Reply-To: 20090903132145.482814810@chello.nl

[-- Attachment #1: sched-lb-13.patch --]
[-- Type: text/plain, Size: 2694 bytes --]

A more readable version, with a few differences:

 - don't check against the root domain, but instead check
   SD_LOAD_BALANCE

 - don't re-iterate the cpus already iterated on the previous SD

 - use rcu_read_lock() around the sd iteration

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
 kernel/sched_fair.c |   45 +++++++++++++++++++++++++--------------------
 1 file changed, 25 insertions(+), 20 deletions(-)

Index: linux-2.6/kernel/sched_fair.c
===================================================================
--- linux-2.6.orig/kernel/sched_fair.c
+++ linux-2.6/kernel/sched_fair.c
@@ -1053,15 +1053,10 @@ static void yield_task_fair(struct rq *r
  * not idle and an idle cpu is available.  The span of cpus to
  * search starts with cpus closest then further out as needed,
  * so we always favor a closer, idle cpu.
- * Domains may include CPUs that are not usable for migration,
- * hence we need to mask them out (rq->rd->online)
  *
  * Returns the CPU we should wake onto.
  */
 #if defined(ARCH_HAS_SCHED_WAKE_IDLE)
-
-#define cpu_rd_active(cpu, rq) cpumask_test_cpu(cpu, rq->rd->online)
-
 /*
  * At POWERSAVINGS_BALANCE_WAKEUP level, if both this_cpu and prev_cpu
  * are idle and this is not a kernel thread and this task's affinity
@@ -1099,7 +1094,7 @@ static int wake_idle_power_save(int cpu,
 static int wake_idle(int cpu, struct task_struct *p)
 {
 	struct rq *task_rq = task_rq(p);
-	struct sched_domain *sd;
+	struct sched_domain *sd, *child = NULL;
 	int i;
 
 	i = wake_idle_power_save(cpu, p);
@@ -1118,24 +1113,34 @@ static int wake_idle(int cpu, struct tas
 	if (idle_cpu(cpu) || cpu_rq(cpu)->cfs.nr_running > 1)
 		return cpu;
 
+	rcu_read_lock();
 	for_each_domain(cpu, sd) {
-		if ((sd->flags & SD_WAKE_IDLE)
-		    || ((sd->flags & SD_WAKE_IDLE_FAR)
-			&& !task_hot(p, task_rq->clock, sd))) {
-			for_each_cpu_and(i, sched_domain_span(sd),
-					 &p->cpus_allowed) {
-				if (cpu_rd_active(i, task_rq) && idle_cpu(i)) {
-					if (i != task_cpu(p)) {
-						schedstat_inc(p,
-						       se.nr_wakeups_idle);
-					}
-					return i;
-				}
-			}
-		} else {
+		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;
+
+		if (!(sd->flags & SD_WAKE_IDLE) &&
+		    (task_hot(p, task_rq->clock, sd) || !(sd->flags & SD_WAKE_IDLE_FAR)))
+			break;
+
+		for_each_cpu_and(i, sched_domain_span(sd), &p->cpus_allowed) {
+			if (child && cpumask_test_cpu(i, sched_domain_span(child)))
+				continue;
+
+			if (!idle_cpu(i))
+				continue;
+
+			if (task_cpu(p) != i)
+				schedstat_inc(p, se.nr_wakeups_idle);
+
+			cpu = i;
+			goto unlock;
 		}
+
+		child = sd;
 	}
+unlock:
+	rcu_read_unlock();
+
 	return cpu;
 }
 #else /* !ARCH_HAS_SCHED_WAKE_IDLE*/

-- 


      parent reply	other threads:[~2009-09-03 13:23 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-09-03 13:21 [RFC][PATCH 00/14] load-balancing and cpu_power -v3 Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 01/14] sched: restore __cpu_power to a straight sum of power Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 02/14] sched: SD_PREFER_SIBLING Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 03/14] sched: update the cpu_power sum during load-balance Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 04/14] sched: add smt_gain Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 05/14] sched: dynamic cpu_power Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 06/14] sched: scale down cpu_power due to RT tasks Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 07/14] sched: try to deal with low capacity Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 08/14] sched: remove reciprocal for cpu_power Peter Zijlstra
2009-09-03 13:59   ` Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 09/14] x86: move APERF/MPERF into a X86_FEATURE Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 10/14] x86: generic aperf/mperf code Peter Zijlstra
2009-09-04  9:19   ` Thomas Renninger
2009-09-04  9:25     ` Peter Zijlstra
2009-09-04  9:27       ` Peter Zijlstra
2009-09-04  9:34         ` Thomas Renninger
2009-09-04 14:22         ` Dave Jones
2009-09-04 14:42           ` Peter Zijlstra
2009-09-04 17:45             ` H. Peter Anvin
2009-09-03 13:21 ` [RFC][PATCH 11/14] sched: provide arch_scale_freq_power Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 12/14] x86: sched: provide arch implementations using aperf/mperf Peter Zijlstra
2009-09-03 13:21 ` [RFC][PATCH 13/14] sched: cleanup wake_idle power saving Peter Zijlstra
2009-09-03 13:21 ` Peter Zijlstra [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20090903132213.413056069@chello.nl \
    --to=a.p.zijlstra@chello.nl \
    --cc=andreas.herrmann3@amd.com \
    --cc=balbir@in.ibm.com \
    --cc=ego@in.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox