Received-SPF: pass (google.com: domain of linux-kernel+bounces-88414-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
From: bala <kumaran.4353@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: mingo@redhat.com,
	peterz@infradead.org,
	Bala <kumaran.4353@gmail.com>
Subject: [PATCH] Check whether the wakeup task is eligible first
Date: Fri,  1 Mar 2024 12:45:56 +0000
Message-Id: <20240301124556.248820-1-kumaran.4353@gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

From: Bala <kumaran.4353@gmail.com>

check_prempt_wakeup_fair is checking whether the current task has to be
preempted by the newly wake-up task. To avoid multiple swaps the logic
is decided as if the wake-up task is the best task then only preempt the
current task.

In CFS time, getting the best task was as simple as piking the left-most
node. But in EEVDF non-fast path, the entire tree has to be traversed in
a worst case scenario.

So, IMHO let's first check the newly wake-up task is eligible before
searching for the best task. At least until the comment becomes
effective to preempt the curr irrespective of the waking task.
---
 kernel/sched/fair.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 533547e3c90a..8d810d6a2cb7 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8334,7 +8334,7 @@ static void check_preempt_wakeup_fair(struct rq *rq, struct task_struct *p, int
 	/*
 	 * XXX pick_eevdf(cfs_rq) != se ?
 	 */
-	if (pick_eevdf(cfs_rq) == pse)
+	if (entity_eligible(cfs_rq, pse) && (pick_eevdf(cfs_rq) == pse))
 		goto preempt;
 
 	return;
-- 
2.33.8