2023-04-20 19:53:48

by Tejun Heo

Subject: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call

Workqueue is in the process of cleaning up the distinction between unbound
workqueues w/ @nr_active==1 and ordered workqueues. Explicit WQ_UNBOUND
isn't needed for alloc_ordered_workqueue() and will trigger a warning in the
future. Let's remove it. This doesn't cause any functional changes.

Signed-off-by: Tejun Heo <[email protected]>
---
drivers/infiniband/hw/irdma/hw.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

--- a/drivers/infiniband/hw/irdma/hw.c
+++ b/drivers/infiniband/hw/irdma/hw.c
@@ -1901,8 +1901,8 @@ int irdma_ctrl_init_hw(struct irdma_pci_
break;
rf->init_state = CEQ0_CREATED;
/* Handles processing of CQP completions */
- rf->cqp_cmpl_wq = alloc_ordered_workqueue("cqp_cmpl_wq",
- WQ_HIGHPRI | WQ_UNBOUND);
+ rf->cqp_cmpl_wq =
+ alloc_ordered_workqueue("cqp_cmpl_wq", WQ_HIGHPRI);
if (!rf->cqp_cmpl_wq) {
status = -ENOMEM;
break;

2023-04-21 14:31:08

by Shiraz Saleem

Subject: RE: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call

> Subject: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from
> alloc_ordered_workqueue() call
>
> Workqueue is in the process of cleaning up the distinction between unbound
> workqueues w/ @nr_active==1 and ordered workqueues. Explicit
> WQ_UNBOUND isn't needed for alloc_ordered_workqueue() and will trigger a
> warning in the future. Let's remove it. This doesn't cause any functional
> changes.
>
> Signed-off-by: Tejun Heo <[email protected]>
> ---
> drivers/infiniband/hw/irdma/hw.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> --- a/drivers/infiniband/hw/irdma/hw.c
> +++ b/drivers/infiniband/hw/irdma/hw.c
> @@ -1901,8 +1901,8 @@ int irdma_ctrl_init_hw(struct irdma_pci_
> break;
> rf->init_state = CEQ0_CREATED;
> /* Handles processing of CQP completions */
> - rf->cqp_cmpl_wq = alloc_ordered_workqueue("cqp_cmpl_wq",
> - WQ_HIGHPRI |
> WQ_UNBOUND);
> + rf->cqp_cmpl_wq =
> + alloc_ordered_workqueue("cqp_cmpl_wq",
> WQ_HIGHPRI);
> if (!rf->cqp_cmpl_wq) {
> status = -ENOMEM;
> break;

Acked-by: Shiraz Saleem <[email protected]>

2023-04-21 14:54:39

by Tejun Heo

Subject: Re: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call

Hello,

On Fri, Apr 21, 2023 at 11:39:24AM -0300, Jason Gunthorpe wrote:
> On Fri, Apr 21, 2023 at 02:29:06PM +0000, Saleem, Shiraz wrote:
> > > Subject: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from
> > > alloc_ordered_workqueue() call
> > >
> > > Workqueue is in the process of cleaning up the distinction between unbound
> > > workqueues w/ @nr_active==1 and ordered workqueues. Explicit
> > > WQ_UNBOUND isn't needed for alloc_ordered_workqueue() and will trigger a
> > > warning in the future. Let's remove it. This doesn't cause any functional
> > > changes.
> > >
> > > Signed-off-by: Tejun Heo <[email protected]>
> > > ---
> > > drivers/infiniband/hw/irdma/hw.c | 4 ++--
> > > 1 file changed, 2 insertions(+), 2 deletions(-)
> > >
> > > --- a/drivers/infiniband/hw/irdma/hw.c
> > > +++ b/drivers/infiniband/hw/irdma/hw.c
> > > @@ -1901,8 +1901,8 @@ int irdma_ctrl_init_hw(struct irdma_pci_
> > > break;
> > > rf->init_state = CEQ0_CREATED;
> > > /* Handles processing of CQP completions */
> > > - rf->cqp_cmpl_wq = alloc_ordered_workqueue("cqp_cmpl_wq",
> > > - WQ_HIGHPRI |
> > > WQ_UNBOUND);
> > > + rf->cqp_cmpl_wq =
> > > + alloc_ordered_workqueue("cqp_cmpl_wq",
> > > WQ_HIGHPRI);
> > > if (!rf->cqp_cmpl_wq) {
> > > status = -ENOMEM;
> > > break;
> >
> > Acked-by: Shiraz Saleem <[email protected]>
>
> Tejun, do you want this in your tree too?

This one doesn't create any dependency and can go through any tree. So, I
guess routing through RDMA is better?

Thanks.

--
tejun

2023-04-21 15:00:08

by Jason Gunthorpe

Subject: Re: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call

On Fri, Apr 21, 2023 at 02:29:06PM +0000, Saleem, Shiraz wrote:
> > Subject: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from
> > alloc_ordered_workqueue() call
> >
> > Workqueue is in the process of cleaning up the distinction between unbound
> > workqueues w/ @nr_active==1 and ordered workqueues. Explicit
> > WQ_UNBOUND isn't needed for alloc_ordered_workqueue() and will trigger a
> > warning in the future. Let's remove it. This doesn't cause any functional
> > changes.
> >
> > Signed-off-by: Tejun Heo <[email protected]>
> > ---
> > drivers/infiniband/hw/irdma/hw.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > --- a/drivers/infiniband/hw/irdma/hw.c
> > +++ b/drivers/infiniband/hw/irdma/hw.c
> > @@ -1901,8 +1901,8 @@ int irdma_ctrl_init_hw(struct irdma_pci_
> > break;
> > rf->init_state = CEQ0_CREATED;
> > /* Handles processing of CQP completions */
> > - rf->cqp_cmpl_wq = alloc_ordered_workqueue("cqp_cmpl_wq",
> > - WQ_HIGHPRI |
> > WQ_UNBOUND);
> > + rf->cqp_cmpl_wq =
> > + alloc_ordered_workqueue("cqp_cmpl_wq",
> > WQ_HIGHPRI);
> > if (!rf->cqp_cmpl_wq) {
> > status = -ENOMEM;
> > break;
>
> Acked-by: Shiraz Saleem <[email protected]>

Tejun, do you want this in your tree too?

Thanks
Jason

2023-04-21 15:45:07

by Jason Gunthorpe

Subject: Re: [PATCH] RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call

On Thu, Apr 20, 2023 at 09:48:08AM -1000, Tejun Heo wrote:
> Workqueue is in the process of cleaning up the distinction between unbound
> workqueues w/ @nr_active==1 and ordered workqueues. Explicit WQ_UNBOUND
> isn't needed for alloc_ordered_workqueue() and will trigger a warning in the
> future. Let's remove it. This doesn't cause any functional changes.
>
> Signed-off-by: Tejun Heo <[email protected]>
> Acked-by: Shiraz Saleem <[email protected]>
> ---
> drivers/infiniband/hw/irdma/hw.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)

Applied to for-next, thanks

Jason