On Sun, Jan 14, 2018 at 02:27:04PM -0700, Jason Gunthorpe wrote:
> On Thu, Jan 11, 2018 at 05:31:28PM +0200, Leon Romanovsky wrote:
> > @@ -1538,6 +1537,7 @@ int ib_destroy_qp(struct ib_qp *qp)
> > if (!qp->uobject)
> > rdma_rw_cleanup_mrs(qp);
> >
> > + rdma_restrack_del(&qp->res, RDMA_RESTRACK_QP);
> > ret = qp->device->destroy_qp(qp);
> > if (!ret) {
> > if (pd)
>
> This hunk belongs in a different patch right?
In this patch, I changed call from device->create_qp() to
the _ib_create_qp() and it is under the hood called to
the rdma_restrack_add(). There was a need to add
rdma_restrack_del() to balance refounts.
>
> Jason
Attachment:
signature.asc
Description: PGP signature
