Re: [PATCH 1/2] workqueue: Make alloc_workqueue() unbound by default

[Marco Crivellari]

Hi,

On Thu, Sep 3, 2026 at 5:22 PM Marco Crivellari
<marco.crivellari@xxxxxxxx> wrote:
> [...]
>         if (unlikely(!(flags & (WQ_UNBOUND | WQ_PERCPU)))) {
> -               WARN_ONCE(1, "workqueue: %s is using neither WQ_PERCPU or WQ_UNBOUND. "
> -                         "Setting WQ_PERCPU.\n", wq->name);
> -               flags |= WQ_PERCPU;
> +               pr_warn_once("workqueue: %s is using neither WQ_PERCPU or WQ_UNBOUND. "
> +                         "Setting WQ_UNBOUND.\n", wq->name);
> +               flags |= WQ_UNBOUND;
>         } else if (unlikely((flags & WQ_PERCPU) && (flags & WQ_UNBOUND))) {
>                 WARN_ONCE(1, "workqueue: %s uses both WQ_PERCPU and WQ_UNBOUND. "
>                           "Dropped WQ_PERCPU, keeping WQ_UNBOUND.\n", wq->name);

Please ignore the patch, I will send a v2.
This will cause issues because wq_size is calculated before:

5911     if (flags & WQ_UNBOUND)
5912         wq_size = struct_size(wq, node_nr_active, nr_node_ids + 1);
5913     else
5914         wq_size = sizeof(*wq);
5915
5916     wq = kzalloc_noprof(wq_size, GFP_KERNEL);
5917     if (!wq)
5918         return NULL;

Thanks, and sorry for the noise!

--

Marco Crivellari

SUSE Labs