[PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Marco Crivellari]

Hi,

=== Current situation: problems ===

Let's consider a nohz_full system with isolated CPUs: wq_unbound_cpumask is
set to the housekeeping CPUs, for !WQ_UNBOUND the local CPU is selected.

This leads to different scenarios if a work item is scheduled on an
isolated CPU where "delay" value is 0 or greater then 0:
        schedule_delayed_work(, 0);

This will be handled by __queue_work() that will queue the work item on the
current local (isolated) CPU, while:

        schedule_delayed_work(, 1);

Will move the timer on an housekeeping CPU, and schedule the work there.

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 consistency cannot be addressed without refactoring the API.

=== Recent changes to the WQ API ===

The following, address the recent changes in the Workqueue API:

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

The old workqueues will be removed in a future release cycle.

=== Introduced Changes by this series ===

1) [P 1-2-3]  Replace uses of system_wq and system_unbound_wq

    system_wq is a per-CPU workqueue, but his name is not clear.
    system_unbound_wq is to be used when locality is not required.

    Because of that, system_wq has been replaced with system_percpu_wq, and
    system_unbound_wq has been replaced with system_dfl_wq.

Thanks!


Marco Crivellari (3):
  drm/atomic-helper: replace use of system_unbound_wq with system_dfl_wq
  drm/probe-helper: replace use of system_wq with system_percpu_wq
  drm/self_refresh: replace use of system_wq with system_percpu_wq

 drivers/gpu/drm/drm_atomic_helper.c       | 6 +++---
 drivers/gpu/drm/drm_probe_helper.c        | 2 +-
 drivers/gpu/drm/drm_self_refresh_helper.c | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)

-- 
2.51.0

[PATCH 1/3] drm/atomic-helper: replace use of system_unbound_wq with system_dfl_wq

[Marco Crivellari]

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 consistency cannot be addressed without refactoring the API.

system_unbound_wq should be the default workqueue so as not to enforce
locality constraints for random work whenever it's not required.

Adding system_dfl_wq to encourage its use when unbound work should be used.

The old system_unbound_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/drm_atomic_helper.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
index d5ebe6ea0acb..166085a7e681 100644
--- a/drivers/gpu/drm/drm_atomic_helper.c
+++ b/drivers/gpu/drm/drm_atomic_helper.c
@@ -2222,13 +2222,13 @@ int drm_atomic_helper_commit(struct drm_device *dev,
 	 * current layout.
 	 *
 	 * NOTE: Commit work has multiple phases, first hardware commit, then
-	 * cleanup. We want them to overlap, hence need system_unbound_wq to
+	 * cleanup. We want them to overlap, hence need system_dfl_wq to
 	 * make sure work items don't artificially stall on each another.
 	 */
 
 	drm_atomic_state_get(state);
 	if (nonblock)
-		queue_work(system_unbound_wq, &state->commit_work);
+		queue_work(system_dfl_wq, &state->commit_work);
 	else
 		commit_tail(state);
 
@@ -2261,7 +2261,7 @@ EXPORT_SYMBOL(drm_atomic_helper_commit);
  *
  * Asynchronous workers need to have sufficient parallelism to be able to run
  * different atomic commits on different CRTCs in parallel. The simplest way to
- * achieve this is by running them on the &system_unbound_wq work queue. Note
+ * achieve this is by running them on the &system_dfl_wq work queue. Note
  * that drivers are not required to split up atomic commits and run an
  * individual commit in parallel - userspace is supposed to do that if it cares.
  * But it might be beneficial to do that for modesets, since those necessarily
-- 
2.51.0

[PATCH 2/3] drm/probe-helper: replace use of system_wq with system_percpu_wq

[Marco Crivellari]

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 consistency cannot be addressed without refactoring the API.

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/drm_probe_helper.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/drm_probe_helper.c b/drivers/gpu/drm/drm_probe_helper.c
index 09b12c30df69..d4dc8cb45bce 100644
--- a/drivers/gpu/drm/drm_probe_helper.c
+++ b/drivers/gpu/drm/drm_probe_helper.c
@@ -626,7 +626,7 @@ int drm_helper_probe_single_connector_modes(struct drm_connector *connector,
 		 */
 		dev->mode_config.delayed_event = true;
 		if (dev->mode_config.poll_enabled)
-			mod_delayed_work(system_wq,
+			mod_delayed_work(system_percpu_wq,
 					 &dev->mode_config.output_poll_work,
 					 0);
 	}
-- 
2.51.0

[PATCH 3/3] drm/self_refresh: replace use of system_wq with system_percpu_wq

[Marco Crivellari]

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 consistency cannot be addressed without refactoring the API.

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/drm_self_refresh_helper.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/drm_self_refresh_helper.c b/drivers/gpu/drm/drm_self_refresh_helper.c
index c0948586b7fd..36f4237efd4d 100644
--- a/drivers/gpu/drm/drm_self_refresh_helper.c
+++ b/drivers/gpu/drm/drm_self_refresh_helper.c
@@ -218,7 +218,7 @@ void drm_self_refresh_helper_alter_state(struct drm_atomic_state *state)
 			 ewma_psr_time_read(&sr_data->exit_avg_ms)) * 2;
 		mutex_unlock(&sr_data->avg_mutex);
 
-		mod_delayed_work(system_wq, &sr_data->entry_work,
+		mod_delayed_work(system_percpu_wq, &sr_data->entry_work,
 				 msecs_to_jiffies(delay));
 	}
 }
-- 
2.51.0

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Marco Crivellari]

Hi,

On Thu, Oct 30, 2025 at 5:20 PM Marco Crivellari
<marco.crivellari@suse.com> wrote:
> Marco Crivellari (3):
>   drm/atomic-helper: replace use of system_unbound_wq with system_dfl_wq
>   drm/probe-helper: replace use of system_wq with system_percpu_wq
>   drm/self_refresh: replace use of system_wq with system_percpu_wq
>
>  drivers/gpu/drm/drm_atomic_helper.c       | 6 +++---
>  drivers/gpu/drm/drm_probe_helper.c        | 2 +-
>  drivers/gpu/drm/drm_self_refresh_helper.c | 2 +-
>  3 files changed, 5 insertions(+), 5 deletions(-)

Gentle ping.

Thanks!

-- 

Marco Crivellari

L3 Support Engineer, Technology & Product

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Marco Crivellari]

On Tue, Dec 2, 2025 at 2:21 PM Marco Crivellari
<marco.crivellari@suse.com> wrote:
> On Thu, Oct 30, 2025 at 5:20 PM Marco Crivellari
> <marco.crivellari@suse.com> wrote:
> > Marco Crivellari (3):
> >   drm/atomic-helper: replace use of system_unbound_wq with system_dfl_wq
> >   drm/probe-helper: replace use of system_wq with system_percpu_wq
> >   drm/self_refresh: replace use of system_wq with system_percpu_wq
> >
> >  drivers/gpu/drm/drm_atomic_helper.c       | 6 +++---
> >  drivers/gpu/drm/drm_probe_helper.c        | 2 +-
> >  drivers/gpu/drm/drm_self_refresh_helper.c | 2 +-
> >  3 files changed, 5 insertions(+), 5 deletions(-)
>
> Gentle ping.

Hi,

Gentle ping.

Thanks.

-- 

Marco Crivellari

L3 Support Engineer

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Thomas Zimmermann]

Hi

Am 30.10.25 um 17:20 schrieb Marco Crivellari:
> Hi,
>
> === Current situation: problems ===
>
> Let's consider a nohz_full system with isolated CPUs: wq_unbound_cpumask is
> set to the housekeeping CPUs, for !WQ_UNBOUND the local CPU is selected.
>
> This leads to different scenarios if a work item is scheduled on an
> isolated CPU where "delay" value is 0 or greater then 0:
>          schedule_delayed_work(, 0);
>
> This will be handled by __queue_work() that will queue the work item on the
> current local (isolated) CPU, while:
>
>          schedule_delayed_work(, 1);
>
> Will move the timer on an housekeeping CPU, and schedule the work there.
>
> 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 consistency cannot be addressed without refactoring the API.
>
> === Recent changes to the WQ API ===
>
> The following, address the recent changes in the Workqueue API:
>
> - commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
> - commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
>
> The old workqueues will be removed in a future release cycle.
>
> === Introduced Changes by this series ===
>
> 1) [P 1-2-3]  Replace uses of system_wq and system_unbound_wq
>
>      system_wq is a per-CPU workqueue, but his name is not clear.
>      system_unbound_wq is to be used when locality is not required.
>
>      Because of that, system_wq has been replaced with system_percpu_wq, and
>      system_unbound_wq has been replaced with system_dfl_wq.

 From the description, I've found it hard to see if there's a change in 
semantics here. But this series is effectively about renaming AFAICT. If so,

Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>

for all patches.

Best regards
Thomas

>
> Thanks!
>
>
> Marco Crivellari (3):
>    drm/atomic-helper: replace use of system_unbound_wq with system_dfl_wq
>    drm/probe-helper: replace use of system_wq with system_percpu_wq
>    drm/self_refresh: replace use of system_wq with system_percpu_wq
>
>   drivers/gpu/drm/drm_atomic_helper.c       | 6 +++---
>   drivers/gpu/drm/drm_probe_helper.c        | 2 +-
>   drivers/gpu/drm/drm_self_refresh_helper.c | 2 +-
>   3 files changed, 5 insertions(+), 5 deletions(-)
>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Marco Crivellari]

On Wed, Feb 4, 2026 at 12:58 PM Thomas Zimmermann <tzimmermann@suse.de> wrote:
> [...]
>  From the description, I've found it hard to see if there's a change in
> semantics here. But this series is effectively about renaming AFAICT. If so,
>
> Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
>
> for all patches.

Hi Thomas,

The new version of the changelog is more clear then this one. In case you
want a cleaner version I can submit a new version.

Anyhow yes, in short the change is the introduction of system_percpu_wq
and system_dfl_wq without changing the behavior:

   system_wq -> system_percpu_wq
   system_unbound_wq -> system_dfl_wq


Many thanks!

-- 

Marco Crivellari

L3 Support Engineer

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Thomas Zimmermann]

Hi

Am 04.02.26 um 14:36 schrieb Marco Crivellari:
> On Wed, Feb 4, 2026 at 12:58 PM Thomas Zimmermann <tzimmermann@suse.de> wrote:
>> [...]
>>   From the description, I've found it hard to see if there's a change in
>> semantics here. But this series is effectively about renaming AFAICT. If so,
>>
>> Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
>>
>> for all patches.
> Hi Thomas,
>
> The new version of the changelog is more clear then this one. In case you
> want a cleaner version I can submit a new version.

No need for an update.

>
> Anyhow yes, in short the change is the introduction of system_percpu_wq
> and system_dfl_wq without changing the behavior:
>
>     system_wq -> system_percpu_wq
>     system_unbound_wq -> system_dfl_wq
>
>
> Many thanks!

These patches go through DRM trees, right? I can merge them if no other 
reviews come in. Ping me if they get lost again.

Best regards
Thomas

>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)

Re: [PATCH 0/3] replace system_unbound_wq and system_wq with the new wqs

[Marco Crivellari]

Hi,

On Thu, Feb 5, 2026 at 8:22 AM Thomas Zimmermann <tzimmermann@suse.de> wrote:
>[...]
> These patches go through DRM trees, right? I can merge them if no other
> reviews come in. Ping me if they get lost again.

Yes, sure I will, thank you!


-- 

Marco Crivellari

L3 Support Engineer