Re: [PATCH rdma-next v8 3/4] RDMA/bnxt_re: Direct Verbs: Support DBR verbs

[Jason Gunthorpe <jgg@ziepe.ca>]

On Sat, Jan 17, 2026 at 01:30:51PM +0530, Sriharsha Basavapatna wrote:
> diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
> index a11f56730a31..33e0f66b39eb 100644
> --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
> +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
> @@ -164,6 +164,13 @@ struct bnxt_re_user_mmap_entry {
>  	u8 mmap_flag;
>  };
>  
> +struct bnxt_re_dbr_obj {
> +	struct bnxt_re_dev *rdev;
> +	struct bnxt_qplib_dpi dpi;
> +	struct bnxt_re_user_mmap_entry *entry;
> +	atomic_t usecnt; /* QPs using this dbr */
> +};

This added usecnt but it doesn't use it

It is supposed to prevent destroying the object by the user out of sequence.

> +static int bnxt_re_dv_dbr_cleanup(struct ib_uobject *uobject,
> +				  enum rdma_remove_reason why,
> +				  struct uverbs_attr_bundle *attrs)
> +{
> +	struct bnxt_re_dbr_obj *obj = uobject->object;
> +	struct bnxt_re_dev *rdev = obj->rdev;
> +
> +	rdma_user_mmap_entry_remove(&obj->entry->rdma_entry);
> +	bnxt_qplib_free_uc_dpi(&rdev->qplib_res, &obj->dpi);
> +	return 0;
> +}

For instance CQ did it in the above function:

static int uverbs_free_cq(struct ib_uobject *uobject,
			  enum rdma_remove_reason why,
			  struct uverbs_attr_bundle *attrs)
{
	ret = ib_destroy_cq_user(cq, &attrs->driver_udata);
	if (ret)
		return ret;

int ib_destroy_cq_user(struct ib_cq *cq, struct ib_udata *udata)
{
	if (atomic_read(&cq->usecnt))
		return -EBUSY;

So this patch should be doing the usecnt check as well, otherwise it
won't work right.

Jason