Merge tag 'nfs-for-4.8-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "Highlights include:

   - Stable patch from Olga to fix RPCSEC_GSS upcalls when the same user
     needs multiple different security services (e.g.  krb5i and krb5p).

   - Stable patch to fix a regression introduced by the use of
     SO_REUSEPORT, and that prevented the use of multiple different NFS
     versions to the same server.

   - TCP socket reconnection timer fixes.

   - Patch from Neil to disable the use of IPv6 temporary addresses"

* tag 'nfs-for-4.8-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4: Cap the transport reconnection timer at 1/2 lease period
  NFSv4: Cleanup the setting of the nfs4 lease period
  SUNRPC: Limit the reconnect backoff timer to the max RPC message timeout
  SUNRPC: Fix reconnection timeouts
  NFSv4.2: LAYOUTSTATS may return NFS4ERR_ADMIN/DELEG_REVOKED
  SUNRPC: disable the use of IPv6 temporary addresses.
  SUNRPC: allow for upcalls for same uid but different gss service
  SUNRPC: Fix up socket autodisconnect
  SUNRPC: Handle EADDRNOTAVAIL on connection failures

4 files changed, 94 insertions, 24 deletions

diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index 23c8e7c..976c781 100644
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -340,12 +340,14 @@ gss_release_msg(struct gss_upcall_msg *gss_msg)
 }
 
 static struct gss_upcall_msg *
-__gss_find_upcall(struct rpc_pipe *pipe, kuid_t uid)
+__gss_find_upcall(struct rpc_pipe *pipe, kuid_t uid, const struct gss_auth *auth)
 {
 	struct gss_upcall_msg *pos;
 	list_for_each_entry(pos, &pipe->in_downcall, list) {
 		if (!uid_eq(pos->uid, uid))
 			continue;
+		if (auth && pos->auth->service != auth->service)
+			continue;
 		atomic_inc(&pos->count);
 		dprintk("RPC:       %s found msg %p\n", __func__, pos);
 		return pos;
@@ -365,7 +367,7 @@ gss_add_msg(struct gss_upcall_msg *gss_msg)
 	struct gss_upcall_msg *old;
 
 	spin_lock(&pipe->lock);
-	old = __gss_find_upcall(pipe, gss_msg->uid);
+	old = __gss_find_upcall(pipe, gss_msg->uid, gss_msg->auth);
 	if (old == NULL) {
 		atomic_inc(&gss_msg->count);
 		list_add(&gss_msg->list, &pipe->in_downcall);
@@ -714,7 +716,7 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	err = -ENOENT;
 	/* Find a matching upcall */
 	spin_lock(&pipe->lock);
-	gss_msg = __gss_find_upcall(pipe, uid);
+	gss_msg = __gss_find_upcall(pipe, uid, NULL);
 	if (gss_msg == NULL) {
 		spin_unlock(&pipe->lock);
 		goto err_put_ctx;
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index cb49898..7f79fb7 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -2638,6 +2638,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
 {
 	struct rpc_xprt_switch *xps;
 	struct rpc_xprt *xprt;
+	unsigned long reconnect_timeout;
 	unsigned char resvport;
 	int ret = 0;
 
@@ -2649,6 +2650,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
 		return -EAGAIN;
 	}
 	resvport = xprt->resvport;
+	reconnect_timeout = xprt->max_reconnect_timeout;
 	rcu_read_unlock();
 
 	xprt = xprt_create_transport(xprtargs);
@@ -2657,6 +2659,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
 		goto out_put_switch;
 	}
 	xprt->resvport = resvport;
+	xprt->max_reconnect_timeout = reconnect_timeout;
 
 	rpc_xprt_switch_set_roundrobin(xps);
 	if (setup) {
@@ -2673,6 +2676,27 @@ out_put_switch:
 }
 EXPORT_SYMBOL_GPL(rpc_clnt_add_xprt);
 
+static int
+rpc_xprt_cap_max_reconnect_timeout(struct rpc_clnt *clnt,
+		struct rpc_xprt *xprt,
+		void *data)
+{
+	unsigned long timeout = *((unsigned long *)data);
+
+	if (timeout < xprt->max_reconnect_timeout)
+		xprt->max_reconnect_timeout = timeout;
+	return 0;
+}
+
+void
+rpc_cap_max_reconnect_timeout(struct rpc_clnt *clnt, unsigned long timeo)
+{
+	rpc_clnt_iterate_for_each_xprt(clnt,
+			rpc_xprt_cap_max_reconnect_timeout,
+			&timeo);
+}
+EXPORT_SYMBOL_GPL(rpc_cap_max_reconnect_timeout);
+
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 static void rpc_show_header(void)
 {
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 8313960..ea244b2 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -680,6 +680,20 @@ out:
 	spin_unlock_bh(&xprt->transport_lock);
 }
 
+static bool
+xprt_has_timer(const struct rpc_xprt *xprt)
+{
+	return xprt->idle_timeout != 0;
+}
+
+static void
+xprt_schedule_autodisconnect(struct rpc_xprt *xprt)
+	__must_hold(&xprt->transport_lock)
+{
+	if (list_empty(&xprt->recv) && xprt_has_timer(xprt))
+		mod_timer(&xprt->timer, xprt->last_used + xprt->idle_timeout);
+}
+
 static void
 xprt_init_autodisconnect(unsigned long data)
 {
@@ -688,6 +702,8 @@ xprt_init_autodisconnect(unsigned long data)
 	spin_lock(&xprt->transport_lock);
 	if (!list_empty(&xprt->recv))
 		goto out_abort;
+	/* Reset xprt->last_used to avoid connect/autodisconnect cycling */
+	xprt->last_used = jiffies;
 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 		goto out_abort;
 	spin_unlock(&xprt->transport_lock);
@@ -725,6 +741,7 @@ void xprt_unlock_connect(struct rpc_xprt *xprt, void *cookie)
 		goto out;
 	xprt->snd_task =NULL;
 	xprt->ops->release_xprt(xprt, NULL);
+	xprt_schedule_autodisconnect(xprt);
 out:
 	spin_unlock_bh(&xprt->transport_lock);
 	wake_up_bit(&xprt->state, XPRT_LOCKED);
@@ -888,11 +905,6 @@ static void xprt_timer(struct rpc_task *task)
 	spin_unlock_bh(&xprt->transport_lock);
 }
 
-static inline int xprt_has_timer(struct rpc_xprt *xprt)
-{
-	return xprt->idle_timeout != 0;
-}
-
 /**
  * xprt_prepare_transmit - reserve the transport before sending a request
  * @task: RPC task about to send a request
@@ -1280,9 +1292,7 @@ void xprt_release(struct rpc_task *task)
 	if (!list_empty(&req->rq_list))
 		list_del(&req->rq_list);
 	xprt->last_used = jiffies;
-	if (list_empty(&xprt->recv) && xprt_has_timer(xprt))
-		mod_timer(&xprt->timer,
-				xprt->last_used + xprt->idle_timeout);
+	xprt_schedule_autodisconnect(xprt);
 	spin_unlock_bh(&xprt->transport_lock);
 	if (req->rq_buffer)
 		xprt->ops->buf_free(req->rq_buffer);
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 111767a..8ede3bc 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -177,7 +177,6 @@ static struct ctl_table sunrpc_table[] = {
  * increase over time if the server is down or not responding.
  */
 #define XS_TCP_INIT_REEST_TO	(3U * HZ)
-#define XS_TCP_MAX_REEST_TO	(5U * 60 * HZ)
 
 /*
  * TCP idle timeout; client drops the transport socket if it is idle
@@ -2173,6 +2172,8 @@ static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 		write_unlock_bh(&sk->sk_callback_lock);
 	}
 	xs_udp_do_set_buffer_size(xprt);
+
+	xprt->stat.connect_start = jiffies;
 }
 
 static void xs_udp_setup_socket(struct work_struct *work)
@@ -2236,6 +2237,7 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 		unsigned int keepcnt = xprt->timeout->to_retries + 1;
 		unsigned int opt_on = 1;
 		unsigned int timeo;
+		unsigned int addr_pref = IPV6_PREFER_SRC_PUBLIC;
 
 		/* TCP Keepalive options */
 		kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
@@ -2247,6 +2249,16 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 		kernel_setsockopt(sock, SOL_TCP, TCP_KEEPCNT,
 				(char *)&keepcnt, sizeof(keepcnt));
 
+		/* Avoid temporary address, they are bad for long-lived
+		 * connections such as NFS mounts.
+		 * RFC4941, section 3.6 suggests that:
+		 *    Individual applications, which have specific
+		 *    knowledge about the normal duration of connections,
+		 *    MAY override this as appropriate.
+		 */
+		kernel_setsockopt(sock, SOL_IPV6, IPV6_ADDR_PREFERENCES,
+				(char *)&addr_pref, sizeof(addr_pref));
+
 		/* TCP user timeout (see RFC5482) */
 		timeo = jiffies_to_msecs(xprt->timeout->to_initval) *
 			(xprt->timeout->to_retries + 1);
@@ -2295,6 +2307,10 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 		/* SYN_SENT! */
 		if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
 			xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
+		break;
+	case -EADDRNOTAVAIL:
+		/* Source port number is unavailable. Try a new one! */
+		transport->srcport = 0;
 	}
 out:
 	return ret;
@@ -2369,6 +2385,25 @@ out:
 	xprt_wake_pending_tasks(xprt, status);
 }
 
+static unsigned long xs_reconnect_delay(const struct rpc_xprt *xprt)
+{
+	unsigned long start, now = jiffies;
+
+	start = xprt->stat.connect_start + xprt->reestablish_timeout;
+	if (time_after(start, now))
+		return start - now;
+	return 0;
+}
+
+static void xs_reconnect_backoff(struct rpc_xprt *xprt)
+{
+	xprt->reestablish_timeout <<= 1;
+	if (xprt->reestablish_timeout > xprt->max_reconnect_timeout)
+		xprt->reestablish_timeout = xprt->max_reconnect_timeout;
+	if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
+		xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
+}
+
 /**
  * xs_connect - connect a socket to a remote endpoint
  * @xprt: pointer to transport structure
@@ -2386,6 +2421,7 @@ out:
 static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task)
 {
 	struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
+	unsigned long delay = 0;
 
 	WARN_ON_ONCE(!xprt_lock_connect(xprt, task, transport));
 
@@ -2397,19 +2433,15 @@ static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task)
 		/* Start by resetting any existing state */
 		xs_reset_transport(transport);
 
-		queue_delayed_work(xprtiod_workqueue,
-				   &transport->connect_worker,
-				   xprt->reestablish_timeout);
-		xprt->reestablish_timeout <<= 1;
-		if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
-			xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
-		if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO)
-			xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO;
-	} else {
+		delay = xs_reconnect_delay(xprt);
+		xs_reconnect_backoff(xprt);
+
+	} else
 		dprintk("RPC:       xs_connect scheduled xprt %p\n", xprt);
-		queue_delayed_work(xprtiod_workqueue,
-				   &transport->connect_worker, 0);
-	}
+
+	queue_delayed_work(xprtiod_workqueue,
+			&transport->connect_worker,
+			delay);
 }
 
 /**
@@ -2961,6 +2993,8 @@ static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args)
 	xprt->ops = &xs_tcp_ops;
 	xprt->timeout = &xs_tcp_default_timeout;
 
+	xprt->max_reconnect_timeout = xprt->timeout->to_maxval;
+
 	INIT_WORK(&transport->recv_worker, xs_tcp_data_receive_workfn);
 	INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_setup_socket);