Re: [PATCH rdma-next 3/5] RDMA/irdma: Use ib_respond_empty_udata where applicable

[Leon Romanovsky <leon@kernel.org>]

On Sat, Jun 27, 2026 at 02:56:40AM +0000, Jacob Moroni wrote:
> Methods that do not provide any user response should use
> the ib_respond_empty_udata() helper to ensure that user
> response buffers are cleared.
> 
> Signed-off-by: Jacob Moroni <jmoroni@google.com>
> ---
>  drivers/infiniband/hw/irdma/verbs.c | 14 +++++++-------
>  1 file changed, 7 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/infiniband/hw/irdma/verbs.c b/drivers/infiniband/hw/irdma/verbs.c
> index 19dcc475c355..d06df520d9be 100644
> --- a/drivers/infiniband/hw/irdma/verbs.c
> +++ b/drivers/infiniband/hw/irdma/verbs.c
> @@ -457,7 +457,7 @@ static int irdma_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
>  
>  	irdma_free_rsrc(iwdev->rf, iwdev->rf->allocated_pds, iwpd->sc_pd.pd_id);
>  
> -	return 0;
> +	return ib_respond_empty_udata(udata);

I wonder how this is supposed to work. What should the user do
after hitting this error? Retry the operation? The object has
already been deallocated.

Thanks.