[PATCH next 1/2] sched: Init new task's vruntime after select cpu

From: Zhang Qiao <zhangqiao22@huawei.com>
To: <mingo@redhat.com>, <peterz@infradead.org>,
	<juri.lelli@redhat.com>, <vincent.guittot@linaro.org>
Cc: <zhangqiao22@huawei.com>, <dietmar.eggemann@arm.com>,
	<rostedt@goodmis.org>, <bsegall@google.com>, <mgorman@suse.de>,
	<bristot@redhat.com>, <vschneid@redhat.com>, <brauner@kernel.org>,
	<linux-kernel@vger.kernel.org>
Subject: [PATCH next 1/2] sched: Init new task's vruntime after select cpu
Date: Mon, 31 Oct 2022 20:51:12 +0800	[thread overview]
Message-ID: <20221031125113.72980-2-zhangqiao22@huawei.com> (raw)
In-Reply-To: <20221031125113.72980-1-zhangqiao22@huawei.com>

When create a new task, we initialize vruntime of the new task
at sched_cgroup_fork(). However, this action is executed too
early and may be incorrect, because it use current cpu to
init the vruntime, but the new task actually runs on the
cpu assigned at wake_up_new_task().

So the patch call task_fork() after select fork cpu and use
the ready cpu(the child will run on it) init the new task.

Signed-off-by: Zhang Qiao <zhangqiao22@huawei.com>
---
 kernel/sched/core.c | 7 ++++++-
 kernel/sched/fair.c | 7 +------
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index e4ce124ec701..ca5677206efd 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -4627,9 +4627,13 @@ void sched_cgroup_fork(struct task_struct *p, struct kernel_clone_args *kargs)
 	 * so use __set_task_cpu().
 	 */
 	__set_task_cpu(p, smp_processor_id());
+	raw_spin_unlock_irqrestore(&p->pi_lock, flags);
+}
+
+void sched_task_fork(struct task_struct *p)
+{
 	if (p->sched_class->task_fork)
 		p->sched_class->task_fork(p);
-	raw_spin_unlock_irqrestore(&p->pi_lock, flags);
 }
 
 void sched_post_fork(struct task_struct *p)
@@ -4682,6 +4686,7 @@ void wake_up_new_task(struct task_struct *p)
 #endif
 	rq = __task_rq_lock(p, &rf);
 	update_rq_clock(rq);
+	sched_task_fork(p);
 	post_init_entity_util_avg(p);
 
 	activate_task(rq, p, ENQUEUE_NOCLOCK);
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index e4a0b8bd941c..34845d425180 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -11603,12 +11603,8 @@ static void task_fork_fair(struct task_struct *p)
 	struct cfs_rq *cfs_rq;
 	struct sched_entity *se = &p->se, *curr;
 	struct rq *rq = this_rq();
-	struct rq_flags rf;
 
-	rq_lock(rq, &rf);
-	update_rq_clock(rq);
-
-	cfs_rq = task_cfs_rq(current);
+	cfs_rq = task_cfs_rq(p);
 	curr = cfs_rq->curr;
 	if (curr) {
 		update_curr(cfs_rq);
@@ -11626,7 +11622,6 @@ static void task_fork_fair(struct task_struct *p)
 	}
 
 	se->vruntime -= cfs_rq->min_vruntime;
-	rq_unlock(rq, &rf);
 }
 
 /*
-- 
2.17.1

