From 9ef845f894c93416a1cbcbc6ec42525fb06aaf4e Mon Sep 17 00:00:00 2001 From: Rao Shoaib Date: Wed, 1 Jul 2020 12:23:38 -0700 Subject: rds: If one path needs re-connection, check all and re-connect In testing with mprds enabled, Oracle Cluster nodes after reboot were not able to communicate with others nodes and so failed to rejoin the cluster. Peers with lower IP address initiated connection but the node could not respond as it choose a different path and could not initiate a connection as it had a higher IP address. With this patch, when a node sends out a packet and the selected path is down, all other paths are also checked and any down paths are re-connected. Reviewed-by: Ka-cheong Poon Reviewed-by: David Edmondson Signed-off-by: Somasundaram Krishnasamy Signed-off-by: Rao Shoaib Signed-off-by: David S. Miller --- net/rds/connection.c | 11 +++++++++++ net/rds/rds.h | 7 +++++++ net/rds/send.c | 3 ++- 3 files changed, 20 insertions(+), 1 deletion(-) (limited to 'net/rds') diff --git a/net/rds/connection.c b/net/rds/connection.c index ed7f2133acc2..f2fcab182095 100644 --- a/net/rds/connection.c +++ b/net/rds/connection.c @@ -905,6 +905,17 @@ void rds_conn_path_connect_if_down(struct rds_conn_path *cp) } EXPORT_SYMBOL_GPL(rds_conn_path_connect_if_down); +/* Check connectivity of all paths + */ +void rds_check_all_paths(struct rds_connection *conn) +{ + int i = 0; + + do { + rds_conn_path_connect_if_down(&conn->c_path[i]); + } while (++i < conn->c_npaths); +} + void rds_conn_connect_if_down(struct rds_connection *conn) { WARN_ON(conn->c_trans->t_mp_capable); diff --git a/net/rds/rds.h b/net/rds/rds.h index 6019b0c004a9..106e862996b9 100644 --- a/net/rds/rds.h +++ b/net/rds/rds.h @@ -778,6 +778,7 @@ void rds_conn_drop(struct rds_connection *conn); void rds_conn_path_drop(struct rds_conn_path *cpath, bool destroy); void rds_conn_connect_if_down(struct rds_connection *conn); void rds_conn_path_connect_if_down(struct rds_conn_path *cp); +void rds_check_all_paths(struct rds_connection *conn); void rds_for_each_conn_info(struct socket *sock, unsigned int len, struct rds_info_iterator *iter, struct rds_info_lengths *lens, @@ -822,6 +823,12 @@ rds_conn_path_up(struct rds_conn_path *cp) return atomic_read(&cp->cp_state) == RDS_CONN_UP; } +static inline int +rds_conn_path_down(struct rds_conn_path *cp) +{ + return atomic_read(&cp->cp_state) == RDS_CONN_DOWN; +} + static inline int rds_conn_up(struct rds_connection *conn) { diff --git a/net/rds/send.c b/net/rds/send.c index 68e2bdb08fd0..9a529a01cdc6 100644 --- a/net/rds/send.c +++ b/net/rds/send.c @@ -1340,7 +1340,8 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len) goto out; } - rds_conn_path_connect_if_down(cpath); + if (rds_conn_path_down(cpath)) + rds_check_all_paths(conn); ret = rds_cong_wait(conn->c_fcong, dport, nonblock, rs); if (ret) { -- cgit v1.2.1