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

[sfrench/cifs-2.6.git] / net / sunrpc / xprtsock.c

diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 111767ab124aa4037dfe8c7040866d7196343292..8ede3bc52481b73c82834aa684111013c6d40cad 100644 (file)
--- 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);