* [PATCH v1] include SCX_OPS_TRACK_MIGRATION
@ 2025-06-22 9:36 Henry Huang
2025-06-22 9:36 ` [PATCH v1] sched_ext: " Henry Huang
0 siblings, 1 reply; 4+ messages in thread
From: Henry Huang @ 2025-06-22 9:36 UTC (permalink / raw)
To: changwoo, arighi, tj, void
Cc: 谈鉴锋, Yan Yan(cailing), linux-kernel,
sched-ext, Henry Huang
In our environment, we need track task migrations to update per-cpu
map.
Implementing fentry(on enqueue_task_scx & dequeue_task_scx)
is a feasible solution. But there are some limitations:
1. Can't modify p->scx.xxx
2. enqueue_task_scx & dequeue_task_scx can't have some
special compilation optimizations.
3. Has more overhead compared to struct_ops
So we include SCX_OPS_TRACK_MIGRATION to support tracking task
migrations.
If SCX_OPS_TRACK_MIGRATION is set, runnable/quiescent
would be called whether task is doing migration or not.
Henry Huang (1):
sched_ext: include SCX_OPS_TRACK_MIGRATION
kernel/sched/ext.c | 18 ++++++++++++++++--
1 file changed, 16 insertions(+), 2 deletions(-)
--
Henry
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH v1] sched_ext: include SCX_OPS_TRACK_MIGRATION
2025-06-22 9:36 [PATCH v1] include SCX_OPS_TRACK_MIGRATION Henry Huang
@ 2025-06-22 9:36 ` Henry Huang
2025-06-22 12:41 ` Andrea Righi
0 siblings, 1 reply; 4+ messages in thread
From: Henry Huang @ 2025-06-22 9:36 UTC (permalink / raw)
To: changwoo, arighi, tj, void
Cc: 谈鉴锋, Yan Yan(cailing), linux-kernel,
sched-ext, Henry Huang
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
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH v1] sched_ext: include SCX_OPS_TRACK_MIGRATION
2025-06-22 9:36 ` [PATCH v1] sched_ext: " Henry Huang
@ 2025-06-22 12:41 ` Andrea Righi
2025-06-22 14:55 ` Henry Huang
0 siblings, 1 reply; 4+ messages in thread
From: Andrea Righi @ 2025-06-22 12:41 UTC (permalink / raw)
To: Henry Huang
Cc: changwoo, tj, void, 谈鉴锋, Yan Yan(cailing),
linux-kernel, sched-ext
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
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v1] sched_ext: include SCX_OPS_TRACK_MIGRATION
2025-06-22 12:41 ` Andrea Righi
@ 2025-06-22 14:55 ` Henry Huang
0 siblings, 0 replies; 4+ messages in thread
From: Henry Huang @ 2025-06-22 14:55 UTC (permalink / raw)
To: arighi
Cc: changwoo, Henry Huang, 谈鉴锋, linux-kernel,
sched-ext, tj, void, Yan Yan(cailing)
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
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2025-06-22 14:55 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-06-22 9:36 [PATCH v1] include SCX_OPS_TRACK_MIGRATION Henry Huang
2025-06-22 9:36 ` [PATCH v1] sched_ext: " Henry Huang
2025-06-22 12:41 ` Andrea Righi
2025-06-22 14:55 ` Henry Huang
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.