[PATCH v2 0/3] rcu: replace system_wq and add WQ_PERCPU or WQ_UNBOUND to alloc_workqueue users

[PATCH v2 0/3] rcu: replace system_wq and add WQ_PERCPU or WQ_UNBOUND to alloc_workqueue users

[Marco Crivellari]

Hi!

Below is a summary of a discussion about the Workqueue API and cpu isolation
considerations. Details and more information are available here:

        "workqueue: Always use wq_select_unbound_cpu() for WORK_CPU_UNBOUND."
        https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/

=== 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 consistentcy cannot be addressed without refactoring the API.

=== Plan and future plans ===

This patchset is the first stone on a refactoring needed in order to
address the points aforementioned; it will have a positive impact also
on the cpu isolation, in the long term, moving away percpu workqueue in
favor to an unbound model.

These are the main steps:
1)  API refactoring (that this patch is introducing)
    -   Make more clear and uniform the system wq names, both per-cpu and
        unbound. This to avoid any possible confusion on what should be
        used.

    -   Introduction of WQ_PERCPU: this flag is the complement of WQ_UNBOUND,
        introduced in this patchset and used on all the callers that are not
        currently using WQ_UNBOUND.

        WQ_UNBOUND will be removed in a future release cycle.

        Most users don't need to be per-cpu, because they don't have
        locality requirements, because of that, a next future step will be
        make "unbound" the default behavior.

2)  Check who really needs to be per-cpu
    -   Remove the WQ_PERCPU flag when is not strictly required.

3)  Add a new API (prefer local cpu)
    -   There are users that don't require a local execution, like mentioned
        above; despite that, local execution yeld to performance gain.

        This new API will prefer the local execution, without requiring it.

=== Introduced Changes by this series ===

1) [P 1] Replace use of system_wq

        system_wq is a per-CPU workqueue, but his name is not clear.

        Because of that, system_wq has been renamed in system_percpu_wq.

2) [P 2-3] add WQ_PERCPU / WQ_UNBOUND to alloc_workqueue() users

        Every alloc_workqueue() caller should use one among WQ_PERCPU or
        WQ_UNBOUND.

        WQ_UNBOUND will be removed in a next release cycle.


Thanks!

---
Changes in v2:
- removed WQ_PERCPU from sync_wq

- added patch 3/3: explicitly use WQ_UNBOUND if WQ_PERCPU is not needed


Marco Crivellari (3):
  rcu: replace use of system_wq with system_percpu_wq
  rcu: WQ_PERCPU added to alloc_workqueue users
  rcu: WQ_UNBOUND added to sync_wq workqueue

 kernel/rcu/tasks.h | 4 ++--
 kernel/rcu/tree.c  | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

-- 
2.51.0

[PATCH v2 1/3] rcu: 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 consistentcy cannot be addressed without refactoring the API.

system_wq is a per-CPU worqueue, yet nothing in its name tells about that
CPU affinity constraint, which is very often not required by users. Make
it clear by adding a system_percpu_wq.

The old wq will be kept for a few release cylces.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 kernel/rcu/tasks.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h
index f92443561d36..2dc044fd126e 100644
--- a/kernel/rcu/tasks.h
+++ b/kernel/rcu/tasks.h
@@ -553,13 +553,13 @@ static void rcu_tasks_invoke_cbs(struct rcu_tasks *rtp, struct rcu_tasks_percpu
 		rtpcp_next = rtp->rtpcp_array[index];
 		if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) {
 			cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND;
-			queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work);
+			queue_work_on(cpuwq, system_percpu_wq, &rtpcp_next->rtp_work);
 			index++;
 			if (index < num_possible_cpus()) {
 				rtpcp_next = rtp->rtpcp_array[index];
 				if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) {
 					cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND;
-					queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work);
+					queue_work_on(cpuwq, system_percpu_wq, &rtpcp_next->rtp_work);
 				}
 			}
 		}
-- 
2.51.0

[PATCH v2 2/3] rcu: WQ_PERCPU added to alloc_workqueue users

[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 consistentcy cannot be addressed without refactoring the API.

alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.

This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.

This patch adds a new WQ_PERCPU flag to explicitly request the use of
the per-CPU behavior. Both flags coexist for one release cycle to allow
callers to transition their calls.

Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.

With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.

All existing users have been updated accordingly.

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

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 8eff357b0436..4f3175df5999 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -4885,7 +4885,7 @@ void __init rcu_init(void)
 	rcutree_online_cpu(cpu);
 
 	/* Create workqueue for Tree SRCU and for expedited GPs. */
-	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM, 0);
+	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	WARN_ON(!rcu_gp_wq);
 
 	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
-- 
2.51.0

[PATCH v2 3/3] rcu: WQ_UNBOUND added to sync_wq workqueue

[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 consistentcy cannot be addressed without refactoring the API.

alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.

This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.

This change add the WQ_UNBOUND flag to sync_wq, to make explicit this
workqueue can be unbound and that it does not benefit from per-cpu work.

Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.

With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.

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

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 4f3175df5999..7137723f8f95 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -4888,7 +4888,7 @@ void __init rcu_init(void)
 	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	WARN_ON(!rcu_gp_wq);
 
-	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
+	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM | WQ_UNBOUND, 0);
 	WARN_ON(!sync_wq);
 
 	/* Respect if explicitly disabled via a boot parameter. */
-- 
2.51.0

Re: [PATCH v2 1/3] rcu: replace use of system_wq with system_percpu_wq

[Frederic Weisbecker]

Le Fri, Sep 19, 2025 at 04:50:37PM +0200, Marco Crivellari a écrit :
> 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.
> 
> system_wq is a per-CPU worqueue, yet nothing in its name tells about that
> CPU affinity constraint, which is very often not required by users. Make
> it clear by adding a system_percpu_wq.
> 
> The old wq will be kept for a few release cylces.
> 
> Suggested-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>

Reviewed-by: Frederic Weisbecker <frederic@kernel.org>


> ---
>  kernel/rcu/tasks.h | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h
> index f92443561d36..2dc044fd126e 100644
> --- a/kernel/rcu/tasks.h
> +++ b/kernel/rcu/tasks.h
> @@ -553,13 +553,13 @@ static void rcu_tasks_invoke_cbs(struct rcu_tasks *rtp, struct rcu_tasks_percpu
>  		rtpcp_next = rtp->rtpcp_array[index];
>  		if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) {
>  			cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND;
> -			queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work);
> +			queue_work_on(cpuwq, system_percpu_wq, &rtpcp_next->rtp_work);
>  			index++;
>  			if (index < num_possible_cpus()) {
>  				rtpcp_next = rtp->rtpcp_array[index];
>  				if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) {
>  					cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND;
> -					queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work);
> +					queue_work_on(cpuwq, system_percpu_wq, &rtpcp_next->rtp_work);
>  				}
>  			}
>  		}
> -- 
> 2.51.0
> 

-- 
Frederic Weisbecker
SUSE Labs

Re: [PATCH v2 2/3] rcu: WQ_PERCPU added to alloc_workqueue users

[Frederic Weisbecker]

Le Fri, Sep 19, 2025 at 04:50:38PM +0200, Marco Crivellari a écrit :
> 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.
> 
> alloc_workqueue() treats all queues as per-CPU by default, while unbound
> workqueues must opt-in via WQ_UNBOUND.
> 
> This default is suboptimal: most workloads benefit from unbound queues,
> allowing the scheduler to place worker threads where they’re needed and
> reducing noise when CPUs are isolated.
> 
> This patch adds a new WQ_PERCPU flag to explicitly request the use of
> the per-CPU behavior. Both flags coexist for one release cycle to allow
> callers to transition their calls.
> 
> Once migration is complete, WQ_UNBOUND can be removed and unbound will
> become the implicit default.
> 
> With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
> any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
> must now use WQ_PERCPU.
> 
> All existing users have been updated accordingly.
> 
> Suggested-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>

Reviewed-by: Frederic Weisbecker <frederic@kernel.org>


> ---
>  kernel/rcu/tree.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 8eff357b0436..4f3175df5999 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -4885,7 +4885,7 @@ void __init rcu_init(void)
>  	rcutree_online_cpu(cpu);
>  
>  	/* Create workqueue for Tree SRCU and for expedited GPs. */
> -	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM, 0);
> +	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM | WQ_PERCPU, 0);
>  	WARN_ON(!rcu_gp_wq);
>  
>  	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
> -- 
> 2.51.0
> 
> 

-- 
Frederic Weisbecker
SUSE Labs

Re: [PATCH v2 3/3] rcu: WQ_UNBOUND added to sync_wq workqueue

[Frederic Weisbecker]

Le Fri, Sep 19, 2025 at 04:50:39PM +0200, Marco Crivellari a écrit :
> 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.
> 
> alloc_workqueue() treats all queues as per-CPU by default, while unbound
> workqueues must opt-in via WQ_UNBOUND.
> 
> This default is suboptimal: most workloads benefit from unbound queues,
> allowing the scheduler to place worker threads where they’re needed and
> reducing noise when CPUs are isolated.
> 
> This change add the WQ_UNBOUND flag to sync_wq, to make explicit this
> workqueue can be unbound and that it does not benefit from per-cpu work.
> 
> Once migration is complete, WQ_UNBOUND can be removed and unbound will
> become the implicit default.
> 
> With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
> any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
> must now use WQ_PERCPU.
> 
> Suggested-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>

Reviewed-by: Frederic Weisbecker <frederic@kernel.org>


> ---
>  kernel/rcu/tree.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 4f3175df5999..7137723f8f95 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -4888,7 +4888,7 @@ void __init rcu_init(void)
>  	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM | WQ_PERCPU, 0);
>  	WARN_ON(!rcu_gp_wq);
>  
> -	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
> +	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM | WQ_UNBOUND, 0);
>  	WARN_ON(!sync_wq);
>  
>  	/* Respect if explicitly disabled via a boot parameter. */
> -- 
> 2.51.0
> 

-- 
Frederic Weisbecker
SUSE Labs

Re: [PATCH v2 3/3] rcu: WQ_UNBOUND added to sync_wq workqueue

[Paul E. McKenney]

On Mon, Sep 22, 2025 at 04:14:46PM +0200, Frederic Weisbecker wrote:
> Le Fri, Sep 19, 2025 at 04:50:39PM +0200, Marco Crivellari a écrit :
> > 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.
> > 
> > alloc_workqueue() treats all queues as per-CPU by default, while unbound
> > workqueues must opt-in via WQ_UNBOUND.
> > 
> > This default is suboptimal: most workloads benefit from unbound queues,
> > allowing the scheduler to place worker threads where they’re needed and
> > reducing noise when CPUs are isolated.
> > 
> > This change add the WQ_UNBOUND flag to sync_wq, to make explicit this
> > workqueue can be unbound and that it does not benefit from per-cpu work.
> > 
> > Once migration is complete, WQ_UNBOUND can be removed and unbound will
> > become the implicit default.
> > 
> > With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
> > any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
> > must now use WQ_PERCPU.
> > 
> > Suggested-by: Tejun Heo <tj@kernel.org>
> > Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
> 
> Reviewed-by: Frederic Weisbecker <frederic@kernel.org>

Applied, thank you both!

I will push these out on my next rebase.

							Thanx, Paul

> > ---
> >  kernel/rcu/tree.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> > index 4f3175df5999..7137723f8f95 100644
> > --- a/kernel/rcu/tree.c
> > +++ b/kernel/rcu/tree.c
> > @@ -4888,7 +4888,7 @@ void __init rcu_init(void)
> >  	rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM | WQ_PERCPU, 0);
> >  	WARN_ON(!rcu_gp_wq);
> >  
> > -	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
> > +	sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM | WQ_UNBOUND, 0);
> >  	WARN_ON(!sync_wq);
> >  
> >  	/* Respect if explicitly disabled via a boot parameter. */
> > -- 
> > 2.51.0
> > 
> 
> -- 
> Frederic Weisbecker
> SUSE Labs

Re: [PATCH v2 3/3] rcu: WQ_UNBOUND added to sync_wq workqueue

[Marco Crivellari]

On Mon, Sep 22, 2025 at 4:27 PM Paul E. McKenney <paulmck@kernel.org> wrote:
> Applied, thank you both!
>
> I will push these out on my next rebase.

Many thanks, Paul!

-- 

Marco Crivellari

L3 Support Engineer, Technology & Product