[PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[Marco Crivellari]

Currently if a user enqueues 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.

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 continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:

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

This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.

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.

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

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/i3c/master.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
index d946db75df70..519b98c37ac7 100644
--- a/drivers/i3c/master.c
+++ b/drivers/i3c/master.c
@@ -2925,7 +2925,7 @@ int i3c_master_register(struct i3c_master_controller *master,
 	if (ret)
 		goto err_put_dev;
 
-	master->wq = alloc_workqueue("%s", 0, 0, dev_name(parent));
+	master->wq = alloc_workqueue("%s", WQ_PERCPU, 0, dev_name(parent));
 	if (!master->wq) {
 		ret = -ENOMEM;
 		goto err_put_dev;
-- 
2.51.1


-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c

Re: [PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[Frank Li]

On Fri, Nov 07, 2025 at 02:29:49PM +0100, Marco Crivellari wrote:
> Currently if a user enqueues 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.
>
> 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 continues the effort to refactor workqueue APIs, which began with
> the introduction of new workqueues and a new alloc_workqueue flag in:
>
> commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
> commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
>
> This change adds a new WQ_PERCPU flag to explicitly request
> alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
>
> 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.
>
> Once migration is complete, WQ_UNBOUND can be removed and unbound will
> become the implicit default.
>
> Suggested-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
> ---
>  drivers/i3c/master.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> index d946db75df70..519b98c37ac7 100644
> --- a/drivers/i3c/master.c
> +++ b/drivers/i3c/master.c
> @@ -2925,7 +2925,7 @@ int i3c_master_register(struct i3c_master_controller *master,
>  	if (ret)
>  		goto err_put_dev;
>
> -	master->wq = alloc_workqueue("%s", 0, 0, dev_name(parent));
> +	master->wq = alloc_workqueue("%s", WQ_PERCPU, 0, dev_name(parent));

Maybe off topic, I think it is not neccesary to create wq for IBI at all.
it can directly use system_bh_wq, or other wq.

Frank
>  	if (!master->wq) {
>  		ret = -ENOMEM;
>  		goto err_put_dev;
> --
> 2.51.1
>

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c

Re: [PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[Sebastian Andrzej Siewior]

On 2025-11-07 16:00:06 [-0500], Frank Li wrote:
> > diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> > index d946db75df70..519b98c37ac7 100644
> > --- a/drivers/i3c/master.c
> > +++ b/drivers/i3c/master.c
> > @@ -2925,7 +2925,7 @@ int i3c_master_register(struct i3c_master_controller *master,
> >  	if (ret)
> >  		goto err_put_dev;
> >
> > -	master->wq = alloc_workqueue("%s", 0, 0, dev_name(parent));
> > +	master->wq = alloc_workqueue("%s", WQ_PERCPU, 0, dev_name(parent));
> 
> Maybe off topic, I think it is not neccesary to create wq for IBI at all.
> it can directly use system_bh_wq, or other wq.

Do you really meant system_bh_wq or would system_unbound_wq do the job?
The _bh_ one would imply softirq usage, I don't think you want that.

> Frank

Sebastian

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c

Re: [PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[Frank Li]

On Fri, Nov 07, 2025 at 11:28:30PM +0100, Sebastian Andrzej Siewior wrote:
> On 2025-11-07 16:00:06 [-0500], Frank Li wrote:
> > > diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> > > index d946db75df70..519b98c37ac7 100644
> > > --- a/drivers/i3c/master.c
> > > +++ b/drivers/i3c/master.c
> > > @@ -2925,7 +2925,7 @@ int i3c_master_register(struct i3c_master_controller *master,
> > >  	if (ret)
> > >  		goto err_put_dev;
> > >
> > > -	master->wq = alloc_workqueue("%s", 0, 0, dev_name(parent));
> > > +	master->wq = alloc_workqueue("%s", WQ_PERCPU, 0, dev_name(parent));
> >
> > Maybe off topic, I think it is not neccesary to create wq for IBI at all.
> > it can directly use system_bh_wq, or other wq.
>
> Do you really meant system_bh_wq or would system_unbound_wq do the job?
> The _bh_ one would imply softirq usage, I don't think you want that.

My major means is that not neccessary to alloc a work queue for IBI. Use
one of system predefined should be enough. system_unbound_wq is good
candidate.

Your change is also good, let's wait for more reviewer's comments.

Frank

>
> > Frank
>
> Sebastian

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c

Re: [PATCH] i3c: master: add WQ_PERCPU to alloc_workqueue users

[Alexandre Belloni]

On 10/11/2025 12:00:48-0500, Frank Li wrote:
> On Fri, Nov 07, 2025 at 11:28:30PM +0100, Sebastian Andrzej Siewior wrote:
> > On 2025-11-07 16:00:06 [-0500], Frank Li wrote:
> > > > diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> > > > index d946db75df70..519b98c37ac7 100644
> > > > --- a/drivers/i3c/master.c
> > > > +++ b/drivers/i3c/master.c
> > > > @@ -2925,7 +2925,7 @@ int i3c_master_register(struct i3c_master_controller *master,
> > > >  	if (ret)
> > > >  		goto err_put_dev;
> > > >
> > > > -	master->wq = alloc_workqueue("%s", 0, 0, dev_name(parent));
> > > > +	master->wq = alloc_workqueue("%s", WQ_PERCPU, 0, dev_name(parent));
> > >
> > > Maybe off topic, I think it is not neccesary to create wq for IBI at all.
> > > it can directly use system_bh_wq, or other wq.
> >
> > Do you really meant system_bh_wq or would system_unbound_wq do the job?
> > The _bh_ one would imply softirq usage, I don't think you want that.
> 
> My major means is that not neccessary to alloc a work queue for IBI. Use
> one of system predefined should be enough. system_unbound_wq is good
> candidate.
> 

I don't agree because you may want to be able to prioritize handling
IBIs vs other work so I believe it makes sense to have a separate
workqueue.

> Your change is also good, let's wait for more reviewer's comments.
> 
> Frank
> 
> >
> > > Frank
> >
> > Sebastian

-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c