From: Yuyang Du <yuyang.du@intel.com>
To: mingo@redhat.com, peterz@infradead.org,
rafael.j.wysocki@intel.com, linux-kernel@vger.kernel.org,
linux-pm@vger.kernel.org
Cc: arjan.van.de.ven@intel.com, len.brown@intel.com,
alan.cox@intel.com, mark.gross@intel.com,
morten.rasmussen@arm.com, vincent.guittot@linaro.org,
rajeev.d.muralidhar@intel.com, vishwesh.m.rudramuni@intel.com,
nicole.chalhoub@intel.com, ajaya.durg@intel.com,
harinarayanan.seshadri@intel.com, jacob.jun.pan@linux.intel.com,
fengguang.wu@intel.com, yuyang.du@intel.com
Subject: [RFC PATCH 12/12 v2] Intercept RT scheduler
Date: Mon, 12 May 2014 02:17:01 +0800 [thread overview]
Message-ID: <1399832221-8314-13-git-send-email-yuyang.du@intel.com> (raw)
In-Reply-To: <1399832221-8314-1-git-send-email-yuyang.du@intel.com>
We intercept load balancing to contain the load and load balancing in
the consolidated CPUs according to our consolidating mechanism.
In RT scheduler, we also skip pulling/selecting task to the idle
non-consolidated CPUs. This is pretty provocative.
Signed-off-by: Yuyang Du <yuyang.du@intel.com>
---
kernel/sched/rt.c | 25 +++++++++++++++++++++++++
1 file changed, 25 insertions(+)
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index bd2267a..f8141fb 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -1217,6 +1217,9 @@ select_task_rq_rt(struct task_struct *p, int cpu, int sd_flag, int flags)
{
struct task_struct *curr;
struct rq *rq;
+#ifdef CONFIG_WORKLOAD_CONSOLIDATION
+ int do_find = 0;
+#endif
if (p->nr_cpus_allowed == 1)
goto out;
@@ -1230,6 +1233,11 @@ select_task_rq_rt(struct task_struct *p, int cpu, int sd_flag, int flags)
rcu_read_lock();
curr = ACCESS_ONCE(rq->curr); /* unlocked access */
+#ifdef CONFIG_WORKLOAD_CONSOLIDATION
+ if (workload_consolidation_cpu_shielded(cpu))
+ do_find = 1;
+#endif
+
/*
* If the current task on @p's runqueue is an RT task, then
* try to see if we can wake this RT task up on another
@@ -1252,9 +1260,15 @@ select_task_rq_rt(struct task_struct *p, int cpu, int sd_flag, int flags)
* This test is optimistic, if we get it wrong the load-balancer
* will have to sort it out.
*/
+#ifdef CONFIG_WORKLOAD_CONSOLIDATION
+ if (do_find || (curr && unlikely(rt_task(curr)) &&
+ (curr->nr_cpus_allowed < 2 ||
+ curr->prio <= p->prio))) {
+#else
if (curr && unlikely(rt_task(curr)) &&
(curr->nr_cpus_allowed < 2 ||
curr->prio <= p->prio)) {
+#endif
int target = find_lowest_rq(p);
if (target != -1)
@@ -1460,6 +1474,12 @@ static int find_lowest_rq(struct task_struct *task)
if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask))
return -1; /* No targets found */
+#ifdef CONFIG_WORKLOAD_CONSOLIDATION
+ workload_consolidation_nonshielded_mask(this_cpu, lowest_mask);
+ if (!cpumask_weight(lowest_mask))
+ return -1;
+#endif
+
/*
* At this point we have built a mask of cpus representing the
* lowest priority tasks in the system. Now we want to elect
@@ -1687,6 +1707,11 @@ static int pull_rt_task(struct rq *this_rq)
if (likely(!rt_overloaded(this_rq)))
return 0;
+#ifdef CONFIG_WORKLOAD_CONSOLIDATION
+ if (workload_consolidation_cpu_shielded(this_cpu))
+ return 0;
+#endif
+
/*
* Match the barrier from rt_set_overloaded; this guarantees that if we
* see overloaded we must also see the rto_mask bit.
--
1.7.9.5
next prev parent reply other threads:[~2014-05-11 18:17 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-05-11 18:16 [RFC PATCH 00/12 v2] A new CPU load metric for power-efficient scheduler: CPU ConCurrency Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 01/12 v2] CONFIG for " Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 02/12 v2] Init " Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 03/12 v2] CPU ConCurrency calculation Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 04/12 v2] CPU ConCurrency tracking Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 05/12 v2] CONFIG for Workload Consolidation Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 06/12 v2] Attach CPU topology to specify each sched_domain's workload consolidation Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 07/12 v2] CPU ConCurrency API for Workload Consolidation Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 08/12 v2] Intercept wakeup/fork/exec load balancing Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 09/12 v2] Intercept idle balancing Yuyang Du
2014-05-11 18:16 ` [RFC PATCH 10/12 v2] Intercept periodic nohz " Yuyang Du
2014-05-11 18:17 ` [RFC PATCH 11/12 v2] Intercept periodic load balancing Yuyang Du
2014-05-11 18:17 ` Yuyang Du [this message]
2014-05-12 6:45 ` [RFC PATCH 00/12 v2] A new CPU load metric for power-efficient scheduler: CPU ConCurrency Peter Zijlstra
2014-05-12 1:28 ` Yuyang Du
2014-05-12 15:15 ` Peter Zijlstra
2014-05-13 13:23 ` Krzysztof Kozlowski
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=1399832221-8314-13-git-send-email-yuyang.du@intel.com \
--to=yuyang.du@intel.com \
--cc=ajaya.durg@intel.com \
--cc=alan.cox@intel.com \
--cc=arjan.van.de.ven@intel.com \
--cc=fengguang.wu@intel.com \
--cc=harinarayanan.seshadri@intel.com \
--cc=jacob.jun.pan@linux.intel.com \
--cc=len.brown@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=mark.gross@intel.com \
--cc=mingo@redhat.com \
--cc=morten.rasmussen@arm.com \
--cc=nicole.chalhoub@intel.com \
--cc=peterz@infradead.org \
--cc=rafael.j.wysocki@intel.com \
--cc=rajeev.d.muralidhar@intel.com \
--cc=vincent.guittot@linaro.org \
--cc=vishwesh.m.rudramuni@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).