For some BPF-schedulers, they should do something when
task is doing migration, such as updating per-cpu map.
If SCX_OPS_TRACK_MIGRATION is set, runnable/quiescent
would be called whether task is doing migration or not.

Signed-off-by: Henry Huang <henry.hj@antgroup.com>
---
 kernel/sched/ext.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index b498d86..9b05bb9 100644
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -161,6 +161,12 @@ enum scx_ops_flags {
 	SCX_OPS_BUILTIN_IDLE_PER_NODE	= 1LLU << 6,
 
 	/*
+	 * If set, runnable/quiescent ops would be called whether the task is
+	 * doing migration or not.
+	 */
+	SCX_OPS_TRACK_MIGRATION		= 1LLU << 7,
+
+	/*
 	 * CPU cgroup support flags
 	 */
 	SCX_OPS_HAS_CGROUP_WEIGHT	= 1LLU << 16,	/* DEPRECATED, will be removed on 6.18 */
@@ -172,6 +178,7 @@ enum scx_ops_flags {
 					  SCX_OPS_ALLOW_QUEUED_WAKEUP |
 					  SCX_OPS_SWITCH_PARTIAL |
 					  SCX_OPS_BUILTIN_IDLE_PER_NODE |
+					  SCX_OPS_TRACK_MIGRATION |
 					  SCX_OPS_HAS_CGROUP_WEIGHT,
 
 	/* high 8 bits are internal, don't include in SCX_OPS_ALL_FLAGS */
@@ -2390,7 +2397,8 @@ static void enqueue_task_scx(struct rq *rq, struct task_struct *p, int enq_flags
 	rq->scx.nr_running++;
 	add_nr_running(rq, 1);
 
-	if (SCX_HAS_OP(sch, runnable) && !task_on_rq_migrating(p))
+	if (SCX_HAS_OP(sch, runnable) &&
+	    ((sch->ops.flags & SCX_OPS_TRACK_MIGRATION) || !task_on_rq_migrating(p)))
 		SCX_CALL_OP_TASK(sch, SCX_KF_REST, runnable, rq, p, enq_flags);
 
 	if (enq_flags & SCX_ENQ_WAKEUP)
@@ -2482,7 +2490,8 @@ static bool dequeue_task_scx(struct rq *rq, struct task_struct *p, int deq_flags
 		SCX_CALL_OP_TASK(sch, SCX_KF_REST, stopping, rq, p, false);
 	}
 
-	if (SCX_HAS_OP(sch, quiescent) && !task_on_rq_migrating(p))
+	if (SCX_HAS_OP(sch, quiescent) &&
+	    ((sch->ops.flags & SCX_OPS_TRACK_MIGRATION) || !task_on_rq_migrating(p)))
 		SCX_CALL_OP_TASK(sch, SCX_KF_REST, quiescent, rq, p, deq_flags);
 
 	if (deq_flags & SCX_DEQ_SLEEP)
@@ -5495,6 +5504,11 @@ static int validate_ops(struct scx_sched *sch, const struct sched_ext_ops *ops)
 		return -EINVAL;
 	}
 
+	if ((ops->flags & SCX_OPS_TRACK_MIGRATION) && (!ops->runnable || !ops->quiescent)) {
+		scx_error(sch, "SCX_OPS_TRACK_MIGRATION requires ops.runnable() and ops.quiescent() to be implemented");
+		return -EINVAL;
+	}
+
 	if (ops->flags & SCX_OPS_HAS_CGROUP_WEIGHT)
 		pr_warn("SCX_OPS_HAS_CGROUP_WEIGHT is deprecated and a noop\n");
 
-- 
Henry

Hi Henry,

On Sun, Jun 22, 2025 at 05:36:21PM +0800, Henry Huang wrote:
> For some BPF-schedulers, they should do something when
> task is doing migration, such as updating per-cpu map.
> If SCX_OPS_TRACK_MIGRATION is set, runnable/quiescent
> would be called whether task is doing migration or not.
> 
> Signed-off-by: Henry Huang <henry.hj@antgroup.com>
> ---
>  kernel/sched/ext.c | 18 ++++++++++++++++--
>  1 file changed, 16 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
> index b498d86..9b05bb9 100644
> --- a/kernel/sched/ext.c
> +++ b/kernel/sched/ext.c
> @@ -161,6 +161,12 @@ enum scx_ops_flags {
>  	SCX_OPS_BUILTIN_IDLE_PER_NODE	= 1LLU << 6,
>  
>  	/*
> +	 * If set, runnable/quiescent ops would be called whether the task is
> +	 * doing migration or not.
> +	 */
> +	SCX_OPS_TRACK_MIGRATION		= 1LLU << 7,
> +
> +	/*
>  	 * CPU cgroup support flags
>  	 */
>  	SCX_OPS_HAS_CGROUP_WEIGHT	= 1LLU << 16,	/* DEPRECATED, will be removed on 6.18 */
> @@ -172,6 +178,7 @@ enum scx_ops_flags {
>  					  SCX_OPS_ALLOW_QUEUED_WAKEUP |
>  					  SCX_OPS_SWITCH_PARTIAL |
>  					  SCX_OPS_BUILTIN_IDLE_PER_NODE |
> +					  SCX_OPS_TRACK_MIGRATION |
>  					  SCX_OPS_HAS_CGROUP_WEIGHT,
>  
>  	/* high 8 bits are internal, don't include in SCX_OPS_ALL_FLAGS */
> @@ -2390,7 +2397,8 @@ static void enqueue_task_scx(struct rq *rq, struct task_struct *p, int enq_flags
>  	rq->scx.nr_running++;
>  	add_nr_running(rq, 1);
>  
> -	if (SCX_HAS_OP(sch, runnable) && !task_on_rq_migrating(p))
> +	if (SCX_HAS_OP(sch, runnable) &&
> +	    ((sch->ops.flags & SCX_OPS_TRACK_MIGRATION) || !task_on_rq_migrating(p)))
>  		SCX_CALL_OP_TASK(sch, SCX_KF_REST, runnable, rq, p, enq_flags);
>  
>  	if (enq_flags & SCX_ENQ_WAKEUP)
> @@ -2482,7 +2490,8 @@ static bool dequeue_task_scx(struct rq *rq, struct task_struct *p, int deq_flags
>  		SCX_CALL_OP_TASK(sch, SCX_KF_REST, stopping, rq, p, false);
>  	}
>  
> -	if (SCX_HAS_OP(sch, quiescent) && !task_on_rq_migrating(p))
> +	if (SCX_HAS_OP(sch, quiescent) &&
> +	    ((sch->ops.flags & SCX_OPS_TRACK_MIGRATION) || !task_on_rq_migrating(p)))
>  		SCX_CALL_OP_TASK(sch, SCX_KF_REST, quiescent, rq, p, deq_flags);

The overall change makes sense to me. I'm wondering if we should set
DEQUEUE_MIGRATING here when task_on_rq_migrating(p) == true (apparently the
sched core doesn't set this flag).

In this way we can use ENQUEUE_MIGRATING and DEQUEUE_MIGRATING to
distinguish between a migration ops.runnable/quiescent() call vs a
"regular" one.

Thanks,
-Andrea

>  
>  	if (deq_flags & SCX_DEQ_SLEEP)
> @@ -5495,6 +5504,11 @@ static int validate_ops(struct scx_sched *sch, const struct sched_ext_ops *ops)
>  		return -EINVAL;
>  	}
>  
> +	if ((ops->flags & SCX_OPS_TRACK_MIGRATION) && (!ops->runnable || !ops->quiescent)) {
> +		scx_error(sch, "SCX_OPS_TRACK_MIGRATION requires ops.runnable() and ops.quiescent() to be implemented");
> +		return -EINVAL;
> +	}
> +
>  	if (ops->flags & SCX_OPS_HAS_CGROUP_WEIGHT)
>  		pr_warn("SCX_OPS_HAS_CGROUP_WEIGHT is deprecated and a noop\n");
>  
> -- 
> Henry
>

On Sun, 22 Jun 2025 14:41:25 +0200, Andrea Righi wrote:
> The overall change makes sense to me. I'm wondering if we should set
> DEQUEUE_MIGRATING here when task_on_rq_migrating(p) == true (apparently the
> sched core doesn't set this flag).
> 
> In this way we can use ENQUEUE_MIGRATING and DEQUEUE_MIGRATING to
> distinguish between a migration ops.runnable/quiescent() call vs a
> "regular" one.

Thanks, I'll implement this in patch v2.

-- 
Henry