Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 4 Apr 2018 09:15:39 -0700
From:   Davidlohr Bueso <dave@stgolabs.net>
To:     peterz@infradead.org, mingo@kernel.org
Cc:     efault@gmx.de, rostedt@goodmis.org, matt@codeblueprint.co.uk,
        linux-kernel@vger.kernel.org, Davidlohr Bueso <dbueso@suse.de>
Subject: Re: [PATCH] sched/rt: Fix rq->clock_update_flags < RQCF_ACT_SKIP
 warning
Message-ID: <20180404161539.nhadkff2aats74jh@linux-n805>
References: <20180402164954.16255-1-dave@stgolabs.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: <20180402164954.16255-1-dave@stgolabs.net>
User-Agent: NeoMutt/20170421 (1.8.2)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, 02 Apr 2018, Davidlohr Bueso wrote:

>The case for the rt task throttling (which this workload also hits) can be ignored in
>that the skip_update call is actually bogus and quite the contrary (the request bits
>are removed/reverted).

While at it, how about this trivial patch?

----8<--------------------------------------------------------
[PATCH 2/1] sched: Simplify helpers for rq clock update skip requests

By renaming the functions we can get rid of the skip parameter
and have better code redability. It makes zero sense to have
things such as:

rq_clock_skip_update(rq, false)

When the skip request is in fact not going to happen. Ever. Rename
things such that we end up with:

rq_clock_skip_update(rq)
rq_clock_cancel_skipupdate(rq)

Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
---
 kernel/sched/core.c     |  2 +-
 kernel/sched/deadline.c |  2 +-
 kernel/sched/fair.c     |  2 +-
 kernel/sched/rt.c       |  2 +-
 kernel/sched/sched.h    | 17 ++++++++++++-----
 5 files changed, 16 insertions(+), 9 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index de440456f15c..c4334a3350c5 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -874,7 +874,7 @@ void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags)
 	 * this case, we can save a useless back to back clock update.
 	 */
 	if (task_on_rq_queued(rq->curr) && test_tsk_need_resched(rq->curr))
-		rq_clock_skip_update(rq, true);
+		rq_clock_skip_update(rq);
 }
 
 #ifdef CONFIG_SMP
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index d1c7bf7c7e5b..e7b3008b85bb 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1560,7 +1560,7 @@ static void yield_task_dl(struct rq *rq)
 	 * so we don't do microscopic update in schedule()
 	 * and double the fastpath cost.
 	 */
-	rq_clock_skip_update(rq, true);
+	rq_clock_skip_update(rq);
 }
 
 #ifdef CONFIG_SMP
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 0951d1c58d2f..54dc31e7ab9b 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7089,7 +7089,7 @@ static void yield_task_fair(struct rq *rq)
 		 * so we don't do microscopic update in schedule()
 		 * and double the fastpath cost.
 		 */
-		rq_clock_skip_update(rq, true);
+		rq_clock_skip_update(rq);
 	}
 
 	set_skip_buddy(se);
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index ad13e6242481..7aef6b4e885a 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -861,7 +861,7 @@ static int do_sched_rt_period_timer(struct rt_bandwidth *rt_b, int overrun)
 				 * 'runtime'.
 				 */
 				if (rt_rq->rt_nr_running && rq->curr == rq->idle)
-					rq_clock_skip_update(rq, false);
+					rq_clock_cancel_skipupdate(rq);
 			}
 			if (rt_rq->rt_time || rt_rq->rt_nr_running)
 				idle = 0;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index c3deaee7a7a2..15750c222ca2 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -976,13 +976,20 @@ static inline u64 rq_clock_task(struct rq *rq)
 	return rq->clock_task;
 }
 
-static inline void rq_clock_skip_update(struct rq *rq, bool skip)
+static inline void rq_clock_skip_update(struct rq *rq)
 {
 	lockdep_assert_held(&rq->lock);
-	if (skip)
-		rq->clock_update_flags |= RQCF_REQ_SKIP;
-	else
-		rq->clock_update_flags &= ~RQCF_REQ_SKIP;
+	rq->clock_update_flags |= RQCF_REQ_SKIP;
+}
+
+/*
+ * See rt task throttoling, which is the only time a skip
+ * request is cancelled.
+ */
+static inline void rq_clock_cancel_skipupdate(struct rq *rq)
+{
+	lockdep_assert_held(&rq->lock);
+	rq->clock_update_flags &= ~RQCF_REQ_SKIP;
 }
 
 struct rq_flags {
-- 
2.13.6