From: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
To: Linux Kernel <linux-kernel@vger.kernel.org>,
Suresh B Siddha <suresh.b.siddha@intel.com>,
Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
Dipankar Sarma <dipankar@in.ibm.com>,
Balbir Singh <balbir@linux.vnet.ibm.com>,
Vatsa <vatsa@linux.vnet.ibm.com>,
Gautham R Shenoy <ego@in.ibm.com>,
Andi Kleen <andi@firstfloor.org>,
David Collier-Brown <davecb@sun.com>,
Tim Connors <tconnors@astro.swin.edu.au>,
Max Krasnyansky <maxk@qualcomm.com>,
Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Subject: [RFC PATCH v2 6/7] sched: bias task wakeups to preferred semi-idle packages
Date: Mon, 08 Sep 2008 18:52:33 +0530 [thread overview]
Message-ID: <20080908132233.3221.18231.stgit@drishya.in.ibm.com> (raw)
In-Reply-To: <20080908131334.3221.61302.stgit@drishya.in.ibm.com>
Preferred wakeup cpu (from a semi idle package) has been
nominated in find_busiest_group() in the previous patch. Use
this information in sched_mc_preferred_wakeup_cpu in function
select_task_rq_fair() to bias task wakeups if the following
conditions are satisfied:
- The present cpu that is trying to wakeup the process is
idle and waking the target process on this cpu will
potentially wakeup a completely idle package
- The previous cpu on which the target process ran is
also idle and hence selecting the previous cpu may
wakeup a semi idle cpu package
- The task being woken up is allowed to run in the
nominated cpu (cpu affinity and restrictions)
Basically if both the current cpu and the previous cpu on
which the task ran is idle, select the nominated cpu from semi
idle cpu package for running the new task that is waking up.
Cache hotness and system utilisation could also be considered
before the decision is made. (Not currently implemented)
This technique will effectively move short running bursty jobs in
an mostly idle system.
Wakeup biasing for power savings gets automatically disabled if
system utilisation increases due to the fact that the probability
of finding both this_cpu and prev_cpu idle decreases.
Signed-off-by: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
---
kernel/sched_fair.c | 13 +++++++++++++
1 files changed, 13 insertions(+), 0 deletions(-)
diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index f1c96e3..1301521 100644
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -73,6 +73,9 @@ unsigned int sysctl_sched_wakeup_granularity = 5000000UL;
const_debug unsigned int sysctl_sched_migration_cost = 500000UL;
+/* Preferred wakeup cpu nominated by sched_mc load balancing logic */
+extern unsigned int sched_mc_preferred_wakeup_cpu;
+
/**************************************************************
* CFS operations on generic schedulable entities:
*/
@@ -1232,6 +1235,16 @@ static int select_task_rq_fair(struct task_struct *p, int sync)
}
}
+ /*
+ * 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
+ * allows it to be moved to preferred cpu, then just move!
+ */
+ if (sched_mc_power_savings >= POWERSAVINGS_BALANCE_WAKEUP &&
+ idle_cpu(prev_cpu) && idle_cpu(this_cpu) && p->mm &&
+ cpu_isset(sched_mc_preferred_wakeup_cpu, p->cpus_allowed))
+ return sched_mc_preferred_wakeup_cpu;
+
if (unlikely(!cpu_isset(this_cpu, p->cpus_allowed)))
goto out;
next prev parent reply other threads:[~2008-09-08 13:20 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-09-08 13:14 [RFC PATCH v2 0/7] Tunable sched_mc_power_savings=n Vaidyanathan Srinivasan
2008-09-08 13:16 ` [RFC PATCH v2 1/7] sched: arch_reinit_sched_domains() must destroy domains to force rebuild Vaidyanathan Srinivasan
2008-09-08 13:17 ` [RFC PATCH v2 2/7] sched: Fix __load_balance_iterator() for cfq with only one task Vaidyanathan Srinivasan
2008-09-08 13:18 ` [RFC PATCH v2 3/7] sched: Framework for sched_mc/smt_power_savings=N Vaidyanathan Srinivasan
2008-09-08 13:20 ` [RFC PATCH v2 4/7] sched: favour lower logical cpu number for sched_mc balance Vaidyanathan Srinivasan
2008-09-08 13:21 ` [RFC PATCH v2 5/7] sched: nominate preferred wakeup cpu Vaidyanathan Srinivasan
2008-09-08 13:21 ` Peter Zijlstra
2008-09-08 13:43 ` Vaidyanathan Srinivasan
2008-09-08 13:22 ` Vaidyanathan Srinivasan [this message]
2008-09-08 13:23 ` [RFC PATCH v2 7/7] sched: activate active load balancing in new idle cpus Vaidyanathan Srinivasan
2008-09-08 13:25 ` [RFC PATCH v2 0/7] Tunable sched_mc_power_savings=n Peter Zijlstra
2008-09-08 13:48 ` Vaidyanathan Srinivasan
2008-09-08 13:56 ` Peter Zijlstra
2008-09-09 1:20 ` Suresh Siddha
2008-09-09 6:18 ` Peter Zijlstra
2008-09-09 6:31 ` Nick Piggin
2008-09-09 6:54 ` Peter Zijlstra
2008-09-09 7:59 ` Nick Piggin
2008-09-09 8:25 ` Peter Zijlstra
2008-09-09 9:03 ` Nick Piggin
2008-09-08 13:58 ` Andi Kleen
2008-09-10 13:45 ` Vaidyanathan Srinivasan
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=20080908132233.3221.18231.stgit@drishya.in.ibm.com \
--to=svaidy@linux.vnet.ibm.com \
--cc=a.p.zijlstra@chello.nl \
--cc=andi@firstfloor.org \
--cc=balbir@linux.vnet.ibm.com \
--cc=davecb@sun.com \
--cc=dipankar@in.ibm.com \
--cc=ego@in.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=maxk@qualcomm.com \
--cc=mingo@elte.hu \
--cc=suresh.b.siddha@intel.com \
--cc=tconnors@astro.swin.edu.au \
--cc=vatsa@linux.vnet.ibm.com \
--cc=venkatesh.pallipadi@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;
as well as URLs for NNTP newsgroup(s).