[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 <bunk@stusta.de>
Signed-off-by: Andrew Beekhof <abeekhof@suse.de>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index ebbaee6..457753d 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -380,6 +380,13 @@
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)
@@ -390,6 +397,11 @@
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 */
BUG_ON(sc && nn->nn_sc && nn->nn_sc != sc);
@@ -1123,6 +1135,44 @@
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);
@@ -1155,6 +1205,23 @@
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;
@@ -1162,15 +1229,6 @@
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 */
@@ -1272,6 +1330,18 @@
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
@@ -1286,6 +1356,7 @@
(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);
}
@@ -1514,6 +1585,8 @@
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,
@@ -1677,6 +1750,7 @@
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: