[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <DEW09WJ1U1FU.IQ9B68XD9VAS@etsalapatis.com>
Date: Fri, 12 Dec 2025 00:57:42 -0500
From: "Emil Tsalapatis" <emil@...alapatis.com>
To: "Tejun Heo" <tj@...nel.org>, "Andrea Righi" <arighi@...dia.com>,
"Changwoo Min" <changwoo@...lia.com>, "David Vernet" <void@...ifault.com>
Cc: <linux-kernel@...r.kernel.org>, <sched-ext@...ts.linux.dev>,
<stable@...r.kernel.org>
Subject: Re: [PATCH 2/2] sched_ext: Fix missing post-enqueue handling in
move_local_task_to_local_dsq()
On Thu Dec 11, 2025 at 8:45 PM EST, Tejun Heo wrote:
> move_local_task_to_local_dsq() is used when moving a task from a non-local
> DSQ to a local DSQ on the same CPU. It directly manipulates the local DSQ
> without going through dispatch_enqueue() and was missing the post-enqueue
> handling that triggers preemption when SCX_ENQ_PREEMPT is set or the idle
> task is running.
>
> The function is used by move_task_between_dsqs() which backs
> scx_bpf_dsq_move() and may be called while the CPU is busy.
>
> Add local_dsq_post_enq() call to move_local_task_to_local_dsq(). As the
> dispatch path doesn't need post-enqueue handling, add SCX_RQ_IN_BALANCE
> early exit to keep consume_dispatch_q() behavior unchanged and avoid
> triggering unnecessary resched when scx_bpf_dsq_move() is used from the
> dispatch path.
>
> Fixes: 4c30f5ce4f7a ("sched_ext: Implement scx_bpf_dispatch[_vtime]_from_dsq()")
> Cc: stable@...r.kernel.org # v6.12+
> Reviewed-by: Andrea Righi <arighi@...dia.com>
> Signed-off-by: Tejun Heo <tj@...nel.org>
Reviewed-by: Emil Tsalapatis <emil@...alapatis.com>
> ---
> kernel/sched/ext.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
> index c78efa99406f..695503a2f7d1 100644
> --- a/kernel/sched/ext.c
> +++ b/kernel/sched/ext.c
> @@ -988,6 +988,14 @@ static void local_dsq_post_enq(struct scx_dispatch_q *dsq, struct task_struct *p
> struct rq *rq = container_of(dsq, struct rq, scx.local_dsq);
> bool preempt = false;
>
> + /*
> + * If @rq is in balance, the CPU is already vacant and looking for the
> + * next task to run. No need to preempt or trigger resched after moving
> + * @p into its local DSQ.
> + */
> + if (rq->scx.flags & SCX_RQ_IN_BALANCE)
> + return;
> +
> if ((enq_flags & SCX_ENQ_PREEMPT) && p != rq->curr &&
> rq->curr->sched_class == &ext_sched_class) {
> rq->curr->scx.slice = 0;
> @@ -1636,6 +1644,8 @@ static void move_local_task_to_local_dsq(struct task_struct *p, u64 enq_flags,
>
> dsq_mod_nr(dst_dsq, 1);
> p->scx.dsq = dst_dsq;
> +
> + local_dsq_post_enq(dst_dsq, p, enq_flags);
> }
>
> /**
Powered by blists - more mailing lists