On Mon, 2025-08-18 at 14:25 -0400, Olga Kornievskaia wrote:
> When a listener is added, a part of creation of transport also registers
> program/port with rpcbind. However, when the listener is removed,
> while transport goes away, rpcbind still has the entry for that
> port/type.
>
> When deleting the transport, unregister with rpcbind when appropriate.
>
> Fixes: d093c9089260 ("nfsd: fix management of listener transports")
> Signed-off-by: Olga Kornievskaia <okorniev@xxxxxxxxxx>
> ---
> net/sunrpc/svc_xprt.c | 17 +++++++++++++++++
> 1 file changed, 17 insertions(+)
>
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index 8b1837228799..223737fac95d 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -1014,6 +1014,23 @@ static void svc_delete_xprt(struct svc_xprt *xprt)
> struct svc_serv *serv = xprt->xpt_server;
> struct svc_deferred_req *dr;
>
> + /* unregister with rpcbind for when transport type is TCP or UDP.
> + * Only TCP and RDMA sockets are marked as LISTENER sockets, so
> + * check for UDP separately.
> + */
> + if ((test_bit(XPT_LISTENER, &xprt->xpt_flags) &&
> + xprt->xpt_class->xcl_ident != XPRT_TRANSPORT_RDMA) ||
> + xprt->xpt_class->xcl_ident == XPRT_TRANSPORT_UDP) {
> + struct svc_sock *svsk = container_of(xprt, struct svc_sock,
> + sk_xprt);
> + struct socket *sock = svsk->sk_sock;
> +
> + if (svc_register(serv, xprt->xpt_net, sock->sk->sk_family,
> + sock->sk->sk_protocol, 0) < 0)
> + pr_warn("failed to unregister %s with rpcbind\n",
> + xprt->xpt_class->xcl_name);
> + }
> +
> if (test_and_set_bit(XPT_DEAD, &xprt->xpt_flags))
> return;
>
This looks good to me. Doing it this way may be preferable if we ever
get around to allowing the removal of listeners while the server is
running.
Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxx>
