[PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Gabriele Monaco]

Commit 14a857056466 ("sched/deadline: Use revised wakeup rule for
dl_server") applies the revised wakeup rule to any server, as a result
servers that are not running (dl_defer_running == 0) and start with a
deadline overflow get enqueued and can boost tasks as if they were
running, invalidating the defer rule and the documented state model.

Apply the revised wakeup rule only for deferrable servers that are
marked as running.

Fixes: 14a857056466 ("sched/deadline: Use revised wakeup rule for dl_server")
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
---

Tested with ksched_football on a 16 CPUs machine: no regression (1s for
check in and game over in 10s).

Apparently the rt_stall sched_ext selftest does fail with tasks not
always getting above 4% (almost always above 3%). But that seems to be
the case also before this change on that machine.

---
 kernel/sched/deadline.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 7db4c87df83b..e5a7701a8af7 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1017,7 +1017,8 @@ static void update_dl_entity(struct sched_dl_entity *dl_se)
 	if (dl_time_before(dl_se->deadline, rq_clock(rq)) ||
 	    dl_entity_overflow(dl_se, rq_clock(rq))) {
 
-		if (unlikely((!dl_is_implicit(dl_se) || dl_se->dl_defer) &&
+		if (unlikely((!dl_is_implicit(dl_se) ||
+			      (dl_se->dl_defer && dl_se->dl_defer_running)) &&
 			     !dl_time_before(dl_se->deadline, rq_clock(rq)) &&
 			     !is_dl_boosted(dl_se))) {
 			update_dl_revised_wakeup(dl_se, rq);

base-commit: 8bc67e4db64aa72732c474b44ea8622062c903f0
-- 
2.54.0

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Andrea Righi]

Hi Garbiele,

On Fri, May 22, 2026 at 02:58:33PM +0200, Gabriele Monaco wrote:
> Commit 14a857056466 ("sched/deadline: Use revised wakeup rule for
> dl_server") applies the revised wakeup rule to any server, as a result
> servers that are not running (dl_defer_running == 0) and start with a
> deadline overflow get enqueued and can boost tasks as if they were
> running, invalidating the defer rule and the documented state model.
> 
> Apply the revised wakeup rule only for deferrable servers that are
> marked as running.
> 
> Fixes: 14a857056466 ("sched/deadline: Use revised wakeup rule for dl_server")
> Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
> ---
> 
> Tested with ksched_football on a 16 CPUs machine: no regression (1s for
> check in and game over in 10s).
> 
> Apparently the rt_stall sched_ext selftest does fail with tasks not
> always getting above 4% (almost always above 3%). But that seems to be
> the case also before this change on that machine.

Yeah the rt_stall selftest can be a bit flaky sometimes, as long as you get >3%
it's all good, maybe we should relax the min threshold a bit. In any case, I
tested this and it looks good om my side.

Tested-by: Andrea Righi <arighi@nvidia.com>

Thanks,
-Andrea

> 
> ---
>  kernel/sched/deadline.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 7db4c87df83b..e5a7701a8af7 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -1017,7 +1017,8 @@ static void update_dl_entity(struct sched_dl_entity *dl_se)
>  	if (dl_time_before(dl_se->deadline, rq_clock(rq)) ||
>  	    dl_entity_overflow(dl_se, rq_clock(rq))) {
>  
> -		if (unlikely((!dl_is_implicit(dl_se) || dl_se->dl_defer) &&
> +		if (unlikely((!dl_is_implicit(dl_se) ||
> +			      (dl_se->dl_defer && dl_se->dl_defer_running)) &&
>  			     !dl_time_before(dl_se->deadline, rq_clock(rq)) &&
>  			     !is_dl_boosted(dl_se))) {
>  			update_dl_revised_wakeup(dl_se, rq);
> 
> base-commit: 8bc67e4db64aa72732c474b44ea8622062c903f0
> -- 
> 2.54.0
> 

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Juri Lelli]

Hi!

On 22/05/26 14:58, Gabriele Monaco wrote:
> Commit 14a857056466 ("sched/deadline: Use revised wakeup rule for
> dl_server") applies the revised wakeup rule to any server, as a result
> servers that are not running (dl_defer_running == 0) and start with a
> deadline overflow get enqueued and can boost tasks as if they were
> running, invalidating the defer rule and the documented state model.
> 
> Apply the revised wakeup rule only for deferrable servers that are
> marked as running.
> 
> Fixes: 14a857056466 ("sched/deadline: Use revised wakeup rule for dl_server")
> Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
> ---

Makes sense to me.

Acked-by: Juri Lelli <juri.lelli@redhat.com>

Thanks,
Juri

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Gabriele Monaco]

Peter,

could you take this patch?

Thanks,
Gabriele

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Gabriele Monaco]

On Wed, 2026-07-29 at 15:27 +0200, Gabriele Monaco wrote:
> Peter,
> 
> could you take this patch?
> 
> Thanks,
> Gabriele

I somehow expected the reply command in lkml to give a bit more context, of
course it does not. [1] is the link to the mentioned patch.

[1] - https://lore.kernel.org/lkml/20260522125833.264145-1-gmonaco@redhat.com/

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Peter Zijlstra]

On Wed, Jul 29, 2026 at 03:27:22PM +0200, Gabriele Monaco wrote:
> Peter,
> 
> could you take this patch?

Yes, just back from holidays, so immense backlog. sched/urgent this?

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[Gabriele Monaco]

On Wed, 2026-07-29 at 16:52 +0200, Peter Zijlstra wrote:
> On Wed, Jul 29, 2026 at 03:27:22PM +0200, Gabriele Monaco wrote:
> > Peter,
> > 
> > could you take this patch?
> 
> Yes, just back from holidays, so immense backlog. sched/urgent this?

I'd say so, thanks!

Re: [PATCH] sched/deadline: Use revised wakeup rule only for running dl_server

[John Stultz]

On Fri, May 22, 2026 at 5:59 AM Gabriele Monaco <gmonaco@redhat.com> wrote:
>
> Commit 14a857056466 ("sched/deadline: Use revised wakeup rule for
> dl_server") applies the revised wakeup rule to any server, as a result
> servers that are not running (dl_defer_running == 0) and start with a
> deadline overflow get enqueued and can boost tasks as if they were
> running, invalidating the defer rule and the documented state model.
>
> Apply the revised wakeup rule only for deferrable servers that are
> marked as running.
>
> Fixes: 14a857056466 ("sched/deadline: Use revised wakeup rule for dl_server")
> Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
> ---
>
> Tested with ksched_football on a 16 CPUs machine: no regression (1s for
> check in and game over in 10s).
>

Yeah, I double checked in my environment and I'm not seeing any
trouble with the ksched_football startup on my 64 vcpu guest I test
with.
[   61.150605] ksched_football: Started referee, game_time: 10 secs !
[   62.407743] ksched_football: All players checked in! Starting game.

Tested-by: John Stultz <jstultz@google.com>

thanks
-john

[tip: sched/urgent] sched/deadline: Use revised wakeup rule only for running dl_server

[tip-bot2 for Gabriele Monaco]

The following commit has been merged into the sched/urgent branch of tip:

Commit-ID:     1842bf97af109f5ebf830175c9725bf81ebb78b1
Gitweb:        https://git.kernel.org/tip/1842bf97af109f5ebf830175c9725bf81ebb78b1
Author:        Gabriele Monaco <gmonaco@redhat.com>
AuthorDate:    Fri, 22 May 2026 14:58:33 +02:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Thu, 30 Jul 2026 12:59:23 +02:00

sched/deadline: Use revised wakeup rule only for running dl_server

Commit 14a857056466 ("sched/deadline: Use revised wakeup rule for
dl_server") applies the revised wakeup rule to any server, as a result
servers that are not running (dl_defer_running == 0) and start with a
deadline overflow get enqueued and can boost tasks as if they were
running, invalidating the defer rule and the documented state model.

Apply the revised wakeup rule only for deferrable servers that are
marked as running.

Fixes: 14a857056466 ("sched/deadline: Use revised wakeup rule for dl_server")
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Juri Lelli <juri.lelli@redhat.com>
Tested-by: Andrea Righi <arighi@nvidia.com>
Link: https://patch.msgid.link/20260522125833.264145-1-gmonaco@redhat.com
---
 kernel/sched/deadline.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 0f858b9..2003000 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1017,7 +1017,8 @@ static void update_dl_entity(struct sched_dl_entity *dl_se)
 	if (dl_time_before(dl_se->deadline, rq_clock(rq)) ||
 	    dl_entity_overflow(dl_se, rq_clock(rq))) {
 
-		if (unlikely((!dl_is_implicit(dl_se) || dl_se->dl_defer) &&
+		if (unlikely((!dl_is_implicit(dl_se) ||
+			      (dl_se->dl_defer && dl_se->dl_defer_running)) &&
 			     !dl_time_before(dl_se->deadline, rq_clock(rq)) &&
 			     !is_dl_boosted(dl_se))) {
 			update_dl_revised_wakeup(dl_se, rq);