[PATCH 0/1] module: replace wq users and add WQ_PERCPU to alloc_workqueue() users

[PATCH 0/1] module: replace wq users and add WQ_PERCPU 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.

=== For Maintainers ===

There are prerequisites for this series, already merged in the master branch.
The commits are:

128ea9f6ccfb6960293ae4212f4f97165e42222d ("workqueue: Add system_percpu_wq and
system_dfl_wq")

930c2ea566aff59e962c50b2421d5fcc3b98b8be ("workqueue: Add new WQ_PERCPU flag")


Thanks!

Marco Crivellari (1):
  module: replace use of system_wq with system_percpu_wq

 kernel/module/dups.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

-- 
2.51.0

[PATCH 1/1] module: 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.

queue_work() / queue_delayed_work() mod_delayed_work() will now use the
new per-cpu wq: whether the user still stick on the old name a warn will
be printed along a wq redirect to the new one.

This patch add the new system_percpu_wq except for mm, fs and net
subsystem, whom are handled in separated patches.

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/module/dups.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/module/dups.c b/kernel/module/dups.c
index bd2149fbe117..e72fa393a2ec 100644
--- a/kernel/module/dups.c
+++ b/kernel/module/dups.c
@@ -113,7 +113,7 @@ static void kmod_dup_request_complete(struct work_struct *work)
 	 * let this linger forever as this is just a boot optimization for
 	 * possible abuses of vmalloc() incurred by finit_module() thrashing.
 	 */
-	queue_delayed_work(system_wq, &kmod_req->delete_work, 60 * HZ);
+	queue_delayed_work(system_percpu_wq, &kmod_req->delete_work, 60 * HZ);
 }
 
 bool kmod_dup_request_exists_wait(char *module_name, bool wait, int *dup_ret)
@@ -240,7 +240,7 @@ void kmod_dup_request_announce(char *module_name, int ret)
 	 * There is no rush. But we also don't want to hold the
 	 * caller up forever or introduce any boot delays.
 	 */
-	queue_work(system_wq, &kmod_req->complete_work);
+	queue_work(system_percpu_wq, &kmod_req->complete_work);
 
 out:
 	mutex_unlock(&kmod_dup_mutex);
-- 
2.51.0

Re: [PATCH 1/1] module: replace use of system_wq with system_percpu_wq

[Petr Pavlu]

On 9/5/25 11:01 AM, Marco Crivellari wrote:
> 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.
> 
> queue_work() / queue_delayed_work() mod_delayed_work() will now use the
> new per-cpu wq: whether the user still stick on the old name a warn will
> be printed along a wq redirect to the new one.
> 
> This patch add the new system_percpu_wq except for mm, fs and net
> subsystem, whom are handled in separated patches.
> 
> 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/module/dups.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/module/dups.c b/kernel/module/dups.c
> index bd2149fbe117..e72fa393a2ec 100644
> --- a/kernel/module/dups.c
> +++ b/kernel/module/dups.c
> @@ -113,7 +113,7 @@ static void kmod_dup_request_complete(struct work_struct *work)
>  	 * let this linger forever as this is just a boot optimization for
>  	 * possible abuses of vmalloc() incurred by finit_module() thrashing.
>  	 */
> -	queue_delayed_work(system_wq, &kmod_req->delete_work, 60 * HZ);
> +	queue_delayed_work(system_percpu_wq, &kmod_req->delete_work, 60 * HZ);
>  }
>  
>  bool kmod_dup_request_exists_wait(char *module_name, bool wait, int *dup_ret)
> @@ -240,7 +240,7 @@ void kmod_dup_request_announce(char *module_name, int ret)
>  	 * There is no rush. But we also don't want to hold the
>  	 * caller up forever or introduce any boot delays.
>  	 */
> -	queue_work(system_wq, &kmod_req->complete_work);
> +	queue_work(system_percpu_wq, &kmod_req->complete_work);
>  
>  out:
>  	mutex_unlock(&kmod_dup_mutex);

The two work items queued by the dups.c code can run anywhere. I don't
see a reason why they need to be bound to a specific CPU.

If I understand the cover letter and its linked discussion correctly,
the aim is to eventually move users to unbound workqueues unless they
really need to use per-CPU workqueues.

If it helps, I believe you can already update this code to use the new
system_dfl_wq.

-- 
Thanks,
Petr

Re: [PATCH 1/1] module: replace use of system_wq with system_percpu_wq

[Marco Crivellari]

On Tue, Sep 9, 2025 at 12:37 PM Petr Pavlu <petr.pavlu@suse.com> wrote:

Hi Petr,

> If I understand the cover letter and its linked discussion correctly,
> the aim is to eventually move users to unbound workqueues unless they
> really need to use per-CPU workqueues.

Yes, correct. This first round is just a 1:1 conversion keeping the
old behavior.
But later yes, the aim is to let per-cpu just who needs to be per-cpu.

> The two work items queued by the dups.c code can run anywhere. I don't
> see a reason why they need to be bound to a specific CPU.

> If it helps, I believe you can already update this code to use the new
> system_dfl_wq.

Cool, I will send the v2 converting directly from system_wq to system_dfl_wq.

Thank you!

-- 

Marco Crivellari

L3 Support Engineer, Technology & Product

marco.crivellari@suse.com