[patch 3/3] OCFS2 Configurable timeouts - Protocol changes

Modify the OCFS2 handshake to ensure essential timeouts are configured
identically on all nodes.

Only allow changes when there are no connected peers

Improves the logic in o2net_advance_rx() which broke now that
sizeof(struct o2net_handshake) is greater than sizeof(struct o2net_msg)

Included is the field for userspace-heartbeat timeout to avoid the need for
further protocol changes.

Uses a global spinlock to ensure the decisions to update configfs entries
are made on the correct value. The region covered by the spinlock when
incrementing the counter is much larger as this is the more critical case.

Small cleanup contributed by Adrian Bunk <[e-mail]>

Signed-off-by: Andrew Beekhof <[e-mail]>
Signed-off-by: Mark Fasheh <[e-mail]>

fs/ocfs2/cluster/nodemanager.c

···
 	ret =  o2nm_cluster_attr_write(page, count, &val);
 
 	if (ret > 0) {
-		if (val <= cluster->cl_keepalive_delay_ms) {
+		if (cluster->cl_idle_timeout_ms != val
+			&& o2net_num_connected_peers()) {
+			mlog(ML_NOTICE,
+			     "o2net: cannot change idle timeout after "
+			     "the first peer has agreed to it."
+			     "  %d connected peers\n",
+			     o2net_num_connected_peers());
+			ret = -EINVAL;
+		} else if (val <= cluster->cl_keepalive_delay_ms) {
 			mlog(ML_NOTICE, "o2net: idle timeout must be larger "
 			     "than keepalive delay\n");
-			return -EINVAL;
+			ret = -EINVAL;
+		} else {
+			cluster->cl_idle_timeout_ms = val;
 		}
-		cluster->cl_idle_timeout_ms = val;
 	}
 
 	return ret;
···
 	ret =  o2nm_cluster_attr_write(page, count, &val);
 
 	if (ret > 0) {
-		if (val >= cluster->cl_idle_timeout_ms) {
+		if (cluster->cl_keepalive_delay_ms != val
+		    && o2net_num_connected_peers()) {
+			mlog(ML_NOTICE,
+			     "o2net: cannot change keepalive delay after"
+			     " the first peer has agreed to it."
+			     "  %d connected peers\n",
+			     o2net_num_connected_peers());
+			ret = -EINVAL;
+		} else if (val >= cluster->cl_idle_timeout_ms) {
 			mlog(ML_NOTICE, "o2net: keepalive delay must be "
 			     "smaller than idle timeout\n");
-			return -EINVAL;
+			ret = -EINVAL;
+		} else {
+			cluster->cl_keepalive_delay_ms = val;
 		}
-		cluster->cl_keepalive_delay_ms = val;
 	}
 
 	return ret;

fs/ocfs2/cluster/tcp.c

···
 		sc_put(sc);
 }
 
+static atomic_t o2net_connected_peers = ATOMIC_INIT(0);
+
+int o2net_num_connected_peers(void)
+{
+	return atomic_read(&o2net_connected_peers);
+}
+
 static void o2net_set_nn_state(struct o2net_node *nn,
 			       struct o2net_sock_container *sc,
 			       unsigned valid, int err)
···
 	struct o2net_sock_container *old_sc = nn->nn_sc;
 
 	assert_spin_locked(&nn->nn_lock);
+
+	if (old_sc && !sc)
+		atomic_dec(&o2net_connected_peers);
+	else if (!old_sc && sc)
+		atomic_inc(&o2net_connected_peers);
 
 	/* the node num comparison and single connect/accept path should stop
 	 * an non-null sc from being overwritten with another */
···
 		return -1;
 	}
 
+	/*
+	 * Ensure timeouts are consistent with other nodes, otherwise
+	 * we can end up with one node thinking that the other must be down,
+	 * but isn't. This can ultimately cause corruption.
+	 */
+	if (be32_to_cpu(hand->o2net_idle_timeout_ms) !=
+				o2net_idle_timeout(sc->sc_node)) {
+		mlog(ML_NOTICE, SC_NODEF_FMT " uses a network idle timeout of "
+		     "%u ms, but we use %u ms locally.  disconnecting\n",
+		     SC_NODEF_ARGS(sc),
+		     be32_to_cpu(hand->o2net_idle_timeout_ms),
+		     o2net_idle_timeout(sc->sc_node));
+		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
+		return -1;
+	}
+
+	if (be32_to_cpu(hand->o2net_keepalive_delay_ms) !=
+			o2net_keepalive_delay(sc->sc_node)) {
+		mlog(ML_NOTICE, SC_NODEF_FMT " uses a keepalive delay of "
+		     "%u ms, but we use %u ms locally.  disconnecting\n",
+		     SC_NODEF_ARGS(sc),
+		     be32_to_cpu(hand->o2net_keepalive_delay_ms),
+		     o2net_keepalive_delay(sc->sc_node));
+		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
+		return -1;
+	}
+
+	if (be32_to_cpu(hand->o2hb_heartbeat_timeout_ms) !=
+			O2HB_MAX_WRITE_TIMEOUT_MS) {
+		mlog(ML_NOTICE, SC_NODEF_FMT " uses a heartbeat timeout of "
+		     "%u ms, but we use %u ms locally.  disconnecting\n",
+		     SC_NODEF_ARGS(sc),
+		     be32_to_cpu(hand->o2hb_heartbeat_timeout_ms),
+		     O2HB_MAX_WRITE_TIMEOUT_MS);
+		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
+		return -1;
+	}
+
 	sc->sc_handshake_ok = 1;
 
 	spin_lock(&nn->nn_lock);
···
 	sclog(sc, "receiving\n");
 	do_gettimeofday(&sc->sc_tv_advance_start);
 
+	if (unlikely(sc->sc_handshake_ok == 0)) {
+		if(sc->sc_page_off < sizeof(struct o2net_handshake)) {
+			data = page_address(sc->sc_page) + sc->sc_page_off;
+			datalen = sizeof(struct o2net_handshake) - sc->sc_page_off;
+			ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
+			if (ret > 0)
+				sc->sc_page_off += ret;
+		}
+
+		if (sc->sc_page_off == sizeof(struct o2net_handshake)) {
+			o2net_check_handshake(sc);
+			if (unlikely(sc->sc_handshake_ok == 0))
+				ret = -EPROTO;
+		}
+		goto out;
+	}
+
 	/* do we need more header? */
 	if (sc->sc_page_off < sizeof(struct o2net_msg)) {
 		data = page_address(sc->sc_page) + sc->sc_page_off;
···
 		ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
 		if (ret > 0) {
 			sc->sc_page_off += ret;
-
-			/* this working relies on the handshake being
-			 * smaller than the normal message header */
-			if (sc->sc_page_off >= sizeof(struct o2net_handshake)&&
-			    !sc->sc_handshake_ok && o2net_check_handshake(sc)) {
-				ret = -EPROTO;
-				goto out;
-			}
-
 			/* only swab incoming here.. we can
 			 * only get here once as we cross from
 			 * being under to over */
···
 	return ret;
 }
 
+static void o2net_initialize_handshake(void)
+{
+	o2net_hand->o2hb_heartbeat_timeout_ms = cpu_to_be32(
+		O2HB_MAX_WRITE_TIMEOUT_MS);
+	o2net_hand->o2net_idle_timeout_ms = cpu_to_be32(
+		o2net_idle_timeout(NULL));
+	o2net_hand->o2net_keepalive_delay_ms = cpu_to_be32(
+		o2net_keepalive_delay(NULL));
+	o2net_hand->o2net_reconnect_delay_ms = cpu_to_be32(
+		o2net_reconnect_delay(NULL));
+}
+
 /* ------------------------------------------------------------ */
 
 /* called when a connect completes and after a sock is accepted.  the
···
               (unsigned long long)O2NET_PROTOCOL_VERSION,
 	      (unsigned long long)be64_to_cpu(o2net_hand->connector_id));
 
+	o2net_initialize_handshake();
 	o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));
 	sc_put(sc);
 }
···
 
 	if (node_num != o2nm_this_node())
 		o2net_disconnect_node(node);
+
+	BUG_ON(atomic_read(&o2net_connected_peers) < 0);
 }
 
 static void o2net_hb_node_up_cb(struct o2nm_node *node, int node_num,
···
 	o2net_register_callbacks(sc->sc_sock->sk, sc);
 	o2net_sc_queue_work(sc, &sc->sc_rx_work);
 
+	o2net_initialize_handshake();
 	o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));
 
 out:

fs/ocfs2/cluster/tcp.h

···
 int o2net_start_listening(struct o2nm_node *node);
 void o2net_stop_listening(struct o2nm_node *node);
 void o2net_disconnect_node(struct o2nm_node *node);
+int o2net_num_connected_peers(void);
 
 int o2net_init(void);
 void o2net_exit(void);

fs/ocfs2/cluster/tcp_internal.h

···
  * locking semantics of the file system using the protocol.  It should 
  * be somewhere else, I'm sure, but right now it isn't.
  *
+ * New in version 5:
+ * 	- Network timeout checking protocol
+ *
  * New in version 4:
  * 	- Remove i_generation from lock names for better stat performance.
  *
···
  * 	- full 64 bit i_size in the metadata lock lvbs
  * 	- introduction of "rw" lock and pushing meta/data locking down
  */
-#define O2NET_PROTOCOL_VERSION 4ULL
+#define O2NET_PROTOCOL_VERSION 5ULL
 struct o2net_handshake {
 	__be64	protocol_version;
 	__be64	connector_id;
+	__be32  o2hb_heartbeat_timeout_ms;
+	__be32  o2net_idle_timeout_ms;
+	__be32  o2net_keepalive_delay_ms;
+	__be32  o2net_reconnect_delay_ms;
 };
 
 struct o2net_node {