From: Mike Galbraith <efault@gmx.de>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: lkml <linux-kernel@vger.kernel.org>,
Suresh Siddha <suresh.b.siddha@intel.com>,
Paul Turner <pjt@google.com>,
Arjan Van De Ven <arjan@linux.intel.com>
Subject: Re: [rfc][patch] select_idle_sibling() inducing bouncing on westmere
Date: Tue, 05 Jun 2012 16:30:21 +0200 [thread overview]
Message-ID: <1338906621.6110.30.camel@marge.simpson.net> (raw)
In-Reply-To: <1337865431.9783.148.camel@laptop>
On Thu, 2012-05-24 at 15:17 +0200, Peter Zijlstra wrote:
> For power7 with 4 smt it would end up as 1230 I guess.
Ok, so I still don't have a POWER7 w. SMT enabled to play with, so just
wired up all buddies cross wise. What's good for the goose is good for
the gander and such.
Oh yeah, alert: barrier impaired^Wchallenged^Wmoron :)
sched: fix select_idle_sibling() induced bouncing
Traversing an entire package is not only expensive, it also leads to tasks
bouncing all over a partially idle and possible quite large package. Fix
that up by assigning a 'buddy' CPU to try to motivate. Each buddy may try
to motivate that one other CPU, if it's busy, tough, it may then try it's
SMT sibling, but that's all this optimization is allowed to cost.
Sibling cache buddies are cross-wired to prevent bouncing.
Signed-off-by: Mike Galbraith <efault@gmx.de>
---
include/linux/sched.h | 1 +
kernel/sched/core.c | 35 ++++++++++++++++++++++++++++++++++-
kernel/sched/fair.c | 27 +++++++--------------------
3 files changed, 42 insertions(+), 21 deletions(-)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -944,6 +944,7 @@ struct sched_domain {
unsigned int smt_gain;
int flags; /* See SD_* */
int level;
+ int idle_buddy; /* cpu assigned to select_idle_sibling() */
/* Runtime fields. */
unsigned long last_balance; /* init to jiffies. units in jiffies */
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5914,6 +5914,11 @@ static void destroy_sched_domains(struct
* SD_SHARE_PKG_RESOURCE set (Last Level Cache Domain) for this
* allows us to avoid some pointer chasing select_idle_sibling().
*
+ * Iterate domains and sched_groups upward, assigning CPUs to be
+ * select_idle_sibling() hw buddy. Cross-wiring hw makes bouncing
+ * due to random perturbation self canceling, ie sw buddies pull
+ * their counterpart to their CPU's hw counterpart.
+ *
* Also keep a unique ID per domain (we use the first cpu number in
* the cpumask of the domain), this allows us to quickly tell if
* two cpus are in the same cache domain, see cpus_share_cache().
@@ -5929,8 +5934,36 @@ static void update_domain_cache(int cpu)
int id = cpu;
sd = highest_flag_domain(cpu, SD_SHARE_PKG_RESOURCES);
- if (sd)
+ if (sd) {
+ struct sched_domain *tmp = sd;
+ struct sched_group *sg, *prev;
+ int right;
+
+ do {
+ id = cpumask_first(sched_domain_span(tmp));
+ prev = sg = tmp->groups;
+ right = 1;
+
+ while (cpumask_first(sched_group_cpus(sg)) != id)
+ sg = sg->next;
+
+ while (!cpumask_test_cpu(cpu, sched_group_cpus(sg))) {
+ prev = sg;
+ sg = sg->next;
+ right = !right;
+ }
+
+ /* A CPU went down, never point back to domain start. */
+ if (right && cpumask_first(sched_group_cpus(sg->next)) == id)
+ right = 0;
+
+ sg = right? sg->next : prev;
+ tmp->idle_buddy = cpumask_first(sched_group_cpus(sg));
+ smp_wmb();
+ } while ((tmp = tmp->child));
+
id = cpumask_first(sched_domain_span(sd));
+ }
rcu_assign_pointer(per_cpu(sd_llc, cpu), sd);
per_cpu(sd_llc_id, cpu) = id;
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -2642,7 +2642,6 @@ static int select_idle_sibling(struct ta
int cpu = smp_processor_id();
int prev_cpu = task_cpu(p);
struct sched_domain *sd;
- struct sched_group *sg;
int i;
/*
@@ -2660,29 +2659,17 @@ static int select_idle_sibling(struct ta
return prev_cpu;
/*
- * Otherwise, iterate the domains and find an elegible idle cpu.
+ * Otherwise, check assigned siblings to find an elegible idle cpu.
*/
sd = rcu_dereference(per_cpu(sd_llc, target));
+
for_each_lower_domain(sd) {
- sg = sd->groups;
- do {
- if (!cpumask_intersects(sched_group_cpus(sg),
- tsk_cpus_allowed(p)))
- goto next;
-
- for_each_cpu(i, sched_group_cpus(sg)) {
- if (!idle_cpu(i))
- goto next;
- }
-
- target = cpumask_first_and(sched_group_cpus(sg),
- tsk_cpus_allowed(p));
- goto done;
-next:
- sg = sg->next;
- } while (sg != sd->groups);
+ smp_rmb();
+ i = sd->idle_buddy;
+ if (idle_cpu(i))
+ return i;
}
-done:
+
return target;
}
next prev parent reply other threads:[~2012-06-05 14:30 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-05-24 11:04 Mike Galbraith
2012-05-24 13:17 ` Peter Zijlstra
2012-05-24 13:20 ` Peter Zijlstra
2012-05-25 6:14 ` Mike Galbraith
2012-05-26 6:37 ` Mike Galbraith
2012-05-26 7:29 ` Peter Zijlstra
2012-05-26 8:27 ` Mike Galbraith
2012-05-27 9:17 ` Mike Galbraith
2012-05-27 11:02 ` Mike Galbraith
2012-05-27 11:12 ` Mike Galbraith
2012-05-27 14:11 ` Arjan van de Ven
2012-05-27 14:29 ` Mike Galbraith
2012-05-27 14:32 ` Mike Galbraith
2012-05-29 18:58 ` Andreas Herrmann
2012-05-25 6:08 ` Mike Galbraith
2012-05-25 8:06 ` Mike Galbraith
2012-06-05 14:30 ` Mike Galbraith [this message]
2012-06-11 16:57 ` [patch v3] sched: fix select_idle_sibling() induced bouncing Mike Galbraith
2012-06-11 17:22 ` Peter Zijlstra
2012-06-11 17:55 ` Mike Galbraith
2012-06-11 18:53 ` Suresh Siddha
2012-06-12 3:18 ` Mike Galbraith
2012-06-20 10:48 ` [tip:sched/core] sched: Improve scalability via 'CPU buddies', which withstand random perturbations tip-bot for Mike Galbraith
2012-07-24 14:18 ` tip-bot for Mike Galbraith
2012-06-19 8:47 ` [patch v3] sched: fix select_idle_sibling() induced bouncing Paul Turner
2012-06-06 10:17 ` [rfc][patch] select_idle_sibling() inducing bouncing on westmere Mike Galbraith
2012-06-06 10:38 ` Mike Galbraith
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=1338906621.6110.30.camel@marge.simpson.net \
--to=efault@gmx.de \
--cc=a.p.zijlstra@chello.nl \
--cc=arjan@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=pjt@google.com \
--cc=suresh.b.siddha@intel.com \
/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
all inboxes | Powered by JetHome®