svcrdma: Reserve an extra WQE for ib_drain_rq()

Do as other ULPs already do: ensure there is an extra Receive WQE
reserved for the tear-down drain WR. I haven't heard reports of
problems but it can't hurt.

Note that rq_depth is used to compute the Send Queue depth as well,
so this fix should affect both the SQ and RQ.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index 4f27325ace4a4f6eaf5944d39dae072b88cad4ab..4a038c7e86f988b7469f015a42057cbb25bda385 100644 (file)
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -415,7 +415,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
        if (newxprt->sc_max_send_sges > dev->attrs.max_send_sge)
                newxprt->sc_max_send_sges = dev->attrs.max_send_sge;
        rq_depth = newxprt->sc_max_requests + newxprt->sc_max_bc_requests +
-                  newxprt->sc_recv_batch;
+                  newxprt->sc_recv_batch + 1 /* drain */;
        if (rq_depth > dev->attrs.max_qp_wr) {
                rq_depth = dev->attrs.max_qp_wr;
                newxprt->sc_recv_batch = 1;