Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.

This lack of consistentcy cannot be addressed without refactoring the API.

This patch continues the effort to refactor worqueue APIs, which has begun
with the change introducing new workqueues and a new alloc_workqueue flag:

commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

system_wq should be the per-cpu workqueue, yet in this name nothing makes
that clear, so replace system_wq with system_percpu_wq.

The old wq (system_wq) will be kept for a few release cycles.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/gpu/drm/scheduler/sched_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index c39f0245e3a9..13192e99637a 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -1315,7 +1315,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched, const struct drm_sched_init_
 	sched->name = args->name;
 	sched->timeout = args->timeout;
 	sched->hang_limit = args->hang_limit;
-	sched->timeout_wq = args->timeout_wq ? args->timeout_wq : system_wq;
+	sched->timeout_wq = args->timeout_wq ? args->timeout_wq : system_percpu_wq;
 	sched->score = args->score ? args->score : &sched->_score;
 	sched->dev = args->dev;
 
-- 
2.51.1

nit: s/replace/Replace

On Tue, 2025-11-04 at 17:52 +0100, Marco Crivellari wrote:
> Currently if a user enqueue a work item using schedule_delayed_work() the

s/enqueue/enqueues

Also: maybe start the sentence with something like "In the general
workqueue implementation, if a user […]". Otherwise it at first reads
as if we're talking about a drm/sched user here.

In general, the commit message should focus more on drm/sched. See
below, too.

> used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
> WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
> schedule_work() that is using system_wq and queue_work(), that makes use
> again of WORK_CPU_UNBOUND.
> 
> This lack of consistentcy cannot be addressed without refactoring the API.
> 
> This patch continues the effort to refactor worqueue APIs, which has begun
> with the change introducing new workqueues and a new alloc_workqueue flag:
> 
> commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
> commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

From my POV it would be enough if you provide these commits in this
commit message and maybe a Link: to the overall discussion in the
workqueue subsystem / implementation.

You can give the details above if you want, but I think drm/sched
doesn't care too much about them. The drm/sched users who really care
about the timeout_wq's exact behavior use one they allocate themselves
anyways.

> 
> system_wq should be the per-cpu workqueue, yet in this name nothing makes
> that clear, so replace system_wq with system_percpu_wq.
> 
> The old wq (system_wq) will be kept for a few release cycles.

Please state in your commit message what you're actually doing to
drm/sched. Like:

"Use the successor of system_wq, system_percpu_wq, for the scheduler's
default timeout_wq. system_wq will be removed in a few release cycles."

> 
> Suggested-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
> ---
>  drivers/gpu/drm/scheduler/sched_main.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
> index c39f0245e3a9..13192e99637a 100644
> --- a/drivers/gpu/drm/scheduler/sched_main.c
> +++ b/drivers/gpu/drm/scheduler/sched_main.c
> @@ -1315,7 +1315,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched, const struct drm_sched_init_
>  	sched->name = args->name;
>  	sched->timeout = args->timeout;
>  	sched->hang_limit = args->hang_limit;
> -	sched->timeout_wq = args->timeout_wq ? args->timeout_wq : system_wq;
> +	sched->timeout_wq = args->timeout_wq ? args->timeout_wq : system_percpu_wq;

Actual change looks fine by me.


Thanks for your patch,
P.

>  	sched->score = args->score ? args->score : &sched->_score;
>  	sched->dev = args->dev;
>  

On Wed, Nov 5, 2025 at 9:39 AM Philipp Stanner <phasta@mailbox.org> wrote:
>
> nit: s/replace/Replace
>
> On Tue, 2025-11-04 at 17:52 +0100, Marco Crivellari wrote:
> > Currently if a user enqueue a work item using schedule_delayed_work() the
>
> s/enqueue/enqueues
>
> Also: maybe start the sentence with something like "In the general
> workqueue implementation, if a user […]". Otherwise it at first reads
> as if we're talking about a drm/sched user here.
>
> In general, the commit message should focus more on drm/sched. See
> below, too.
>
> > used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
> > WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
> > schedule_work() that is using system_wq and queue_work(), that makes use
> > again of WORK_CPU_UNBOUND.
> >
> > This lack of consistentcy cannot be addressed without refactoring the API.
> >
> > This patch continues the effort to refactor worqueue APIs, which has begun
> > with the change introducing new workqueues and a new alloc_workqueue flag:
> >
> > commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
> > commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
>
> From my POV it would be enough if you provide these commits in this
> commit message and maybe a Link: to the overall discussion in the
> workqueue subsystem / implementation.
>
> You can give the details above if you want, but I think drm/sched
> doesn't care too much about them. The drm/sched users who really care
> about the timeout_wq's exact behavior use one they allocate themselves
> anyways.
>
> >
> > system_wq should be the per-cpu workqueue, yet in this name nothing makes
> > that clear, so replace system_wq with system_percpu_wq.
> >
> > The old wq (system_wq) will be kept for a few release cycles.
>
> Please state in your commit message what you're actually doing to
> drm/sched. Like:
>
> "Use the successor of system_wq, system_percpu_wq, for the scheduler's
> default timeout_wq. system_wq will be removed in a few release cycles."
>[...]
> Actual change looks fine by me.
>
>
> Thanks for your patch,
> P.

Many thanks for all your feedback Philipp.

-- 

Marco Crivellari

L3 Support Engineer, Technology & Product