On Fri, Nov 07, 2025 at 03:33:35PM +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.
Thanks for these patches. We have:
[PATCH] PCI: endpoint: add WQ_PERCPU to alloc_workqueue users
[PATCH] PCI: endpoint: epf-mhi: add WQ_PERCPU to alloc_workqueue users
[PATCH] PCI: endpoint: pci-epf-test: add WQ_PERCPU to alloc_workqueue users
[PATCH] PCI: pnv_php: add WQ_PERCPU to alloc_workqueue users
[PATCH] PCI: shpchp: add WQ_PERCPU to alloc_workqueue users
[PATCH] PCI: endpoint: replace use of system_wq with system_percpu_wq
IIUC these are all part of the same effort to refactor the workqueue
API and don't really have anything to do with the endpoint or hotplug
drivers themselves.
So I put these all on pci/workqueue and squashed the WQ_PERCPU patches
together because they do the same thing, they have the same commit
log, and there's not really any point in reviewing them separately.
> Suggested-by: Tejun Heo <[email protected]>
> Signed-off-by: Marco Crivellari <[email protected]>
> ---
> drivers/pci/hotplug/pnv_php.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/pci/hotplug/pnv_php.c b/drivers/pci/hotplug/pnv_php.c
> index c5345bff9a55..35f1758126c6 100644
> --- a/drivers/pci/hotplug/pnv_php.c
> +++ b/drivers/pci/hotplug/pnv_php.c
> @@ -802,7 +802,7 @@ static struct pnv_php_slot *pnv_php_alloc_slot(struct
> device_node *dn)
> }
>
> /* Allocate workqueue for this slot's interrupt handling */
> - php_slot->wq = alloc_workqueue("pciehp-%s", 0, 0, php_slot->name);
> + php_slot->wq = alloc_workqueue("pciehp-%s", WQ_PERCPU, 0,
> php_slot->name);
> if (!php_slot->wq) {
> SLOT_WARN(php_slot, "Cannot alloc workqueue\n");
> kfree(php_slot->name);
> --
> 2.51.1
>
