linux-kernel - [RFC PATCH 2/3] sched/fair: Sync se with root cfs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed,  1 Jun 2016 20:39:21 +0100
From:	Dietmar Eggemann <dietmar.eggemann@....com>
To:	Peter Zijlstra <peterz@...radead.org>, linux-kernel@...r.kernel.org
Cc:	Vincent Guittot <vincent.guittot@...aro.org>,
	Ben Segall <bsegall@...gle.com>,
	Morten Rasmussen <morten.rasmussen@....com>,
	Yuyang Du <yuyang.du@...el.com>
Subject: [RFC PATCH 2/3] sched/fair: Sync se with root cfs_rq

Since task utilization is accrued only on the root cfs_rq, there are a
couple of places where the se has to be synced with the root cfs_rq:

(1) The root cfs_rq has to be updated in attach_entity_load_avg() for
    an se representing a task in a tg other than the root tg before
    the se utilization can be added to it.

(2) The last_update_time value of the root cfs_rq can be higher
    than the one of the cfs_rq the se is enqueued in. Call
    __update_load_avg() on the se with the last_update_time value of
    the root cfs_rq before removing se's utilization from the root
    cfs_rq in [remove|detach]_entity_load_avg().

In case the difference between the last_update_time value of the cfs_rq
and the root cfs_rq is smaller than 1024ns, the additional calls to
__update_load_avg() will bail early.

Signed-off-by: Dietmar Eggemann <dietmar.eggemann@....com>
---
 kernel/sched/fair.c | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 212becd3708f..3ae8e79fb687 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -2970,6 +2970,8 @@ static inline void update_load_avg(struct sched_entity *se, int update_tg)
 
 static void attach_entity_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se)
 {
+	struct cfs_rq* root_cfs_rq;
+
 	if (!sched_feat(ATTACH_AGE_LOAD))
 		goto skip_aging;
 
@@ -2995,8 +2997,16 @@ static void attach_entity_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *s
 	if (!entity_is_task(se))
 		return;
 
-	rq_of(cfs_rq)->cfs.avg.util_avg += se->avg.util_avg;
-	rq_of(cfs_rq)->cfs.avg.util_sum += se->avg.util_sum;
+	root_cfs_rq = &rq_of(cfs_rq)->cfs;
+
+	if (parent_entity(se))
+		__update_load_avg(cfs_rq_clock_task(root_cfs_rq),
+				  cpu_of(rq_of(root_cfs_rq)), &root_cfs_rq->avg,
+				  scale_load_down(root_cfs_rq->load.weight),
+				  upd_util_cfs_rq(root_cfs_rq), root_cfs_rq);
+
+	root_cfs_rq->avg.util_avg += se->avg.util_avg;
+	root_cfs_rq->avg.util_sum += se->avg.util_sum;
 
 	cfs_rq_util_change(cfs_rq);
 }
@@ -3013,6 +3023,10 @@ static void detach_entity_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *s
 	if (!entity_is_task(se))
 		return;
 
+	__update_load_avg(rq_of(cfs_rq)->cfs.avg.last_update_time, cpu_of(rq_of(cfs_rq)),
+			  &se->avg, se->on_rq * scale_load_down(se->load.weight),
+			  cfs_rq->curr == se, NULL);
+
 	rq_of(cfs_rq)->cfs.avg.util_avg =
 	    max_t(long, rq_of(cfs_rq)->cfs.avg.util_avg - se->avg.util_avg, 0);
 	rq_of(cfs_rq)->cfs.avg.util_sum =
@@ -3105,6 +3119,9 @@ void remove_entity_load_avg(struct sched_entity *se)
 	if (!entity_is_task(se))
 		return;
 
+	last_update_time = cfs_rq_last_update_time(&rq_of(cfs_rq)->cfs);
+
+	__update_load_avg(last_update_time, cpu_of(rq_of(cfs_rq)), &se->avg, 0, 0, NULL);
 	atomic_long_add(se->avg.util_avg, &rq_of(cfs_rq)->cfs.removed_util_avg);
 }
 
-- 
1.9.1