void rds_connect_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_conn_w.work);
int ret;
clear_bit(RDS_RECONNECT_PENDING, &conn->c_flags);
if (rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
/*
* record the time we started trying to connect so that we can
* drop the connection if it doesn't work out after a while
*/
conn->c_connection_start = get_seconds();
ret = conn->c_trans->conn_connect(conn);
rdsdebug("conn %p for %pI4 to %pI4 dispatched, ret %d\n",
conn, &conn->c_laddr, &conn->c_faddr, ret);
if (ret) {
if (rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_DOWN)) {
rds_queue_reconnect(conn);
} else
rds_conn_error(conn, "RDS: connect failed\n");
}
}
}
void rds_shutdown_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_down_w);
/* shut it down unless it's down already */
if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_DOWN)) {
/*
* Quiesce the connection mgmt handlers before we start tearing
* things down. We don't hold the mutex for the entire
* duration of the shutdown operation, else we may be
* deadlocking with the CM handler. Instead, the CM event
* handler is supposed to check for state DISCONNECTING
*/
mutex_lock(&conn->c_cm_lock);
if (!rds_conn_transition(conn, RDS_CONN_UP, RDS_CONN_DISCONNECTING)
&& !rds_conn_transition(conn, RDS_CONN_ERROR, RDS_CONN_DISCONNECTING)) {
rds_conn_error(conn, "shutdown called in state %d\n",
atomic_read(&conn->c_state));
mutex_unlock(&conn->c_cm_lock);
return;
}
mutex_unlock(&conn->c_cm_lock);
mutex_lock(&conn->c_send_lock);
conn->c_trans->conn_shutdown(conn);
rds_conn_reset(conn);
mutex_unlock(&conn->c_send_lock);
if (!rds_conn_transition(conn, RDS_CONN_DISCONNECTING, RDS_CONN_DOWN)) {
/* This can happen - eg when we're in the middle of tearing
* down the connection, and someone unloads the rds module.
* Quite reproduceable with loopback connections.
* Mostly harmless.
*/
rds_conn_error(conn,
"%s: failed to transition to state DOWN, "
"current state is %d\n",
__func__,
atomic_read(&conn->c_state));
return;
}
}
/* Then reconnect if it's still live.
* The passive side of an IB loopback connection is never added
* to the conn hash, so we never trigger a reconnect on this
* conn - the reconnect is always triggered by the active peer. */
cancel_delayed_work(&conn->c_conn_w);
if (!hlist_unhashed(&conn->c_hash_node))
rds_queue_reconnect(conn);
}
void rds_connect_complete(struct rds_connection *conn)
{
if (!rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_UP)) {
printk(KERN_WARNING "%s: Cannot transition to state UP"
", current state is %d\n",
__func__,
atomic_read(&conn->c_state));
atomic_set(&conn->c_state, RDS_CONN_ERROR);
queue_work(rds_wq, &conn->c_down_w);
return;
}
rdsdebug("conn %p for %pI4 to %pI4 complete\n",
conn, &conn->c_laddr, &conn->c_faddr);
conn->c_reconnect_jiffies = 0;
set_bit(0, &conn->c_map_queued);
queue_delayed_work(rds_wq, &conn->c_send_w, 0);
queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
queue_delayed_work(rds_wq, &conn->c_hb_w, 0);
conn->c_hb_start = 0;
conn->c_connection_start = get_seconds();
conn->c_reconnect = 1;
conn->c_committed_version = conn->c_version;
conn->c_proposed_version = RDS_PROTOCOL_VERSION;
}
static int rds_tcp_accept_one(struct socket *sock)
{
struct socket *new_sock = NULL;
struct rds_connection *conn;
int ret;
struct inet_sock *inet;
ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
sock->sk->sk_protocol, &new_sock);
if (ret)
goto out;
new_sock->type = sock->type;
new_sock->ops = sock->ops;
ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
if (ret < 0)
goto out;
rds_tcp_tune(new_sock);
inet = inet_sk(new_sock->sk);
rdsdebug("accepted tcp %pI4:%u -> %pI4:%u\n",
&inet->inet_saddr, ntohs(inet->inet_sport),
&inet->inet_daddr, ntohs(inet->inet_dport));
conn = rds_conn_create(inet->inet_saddr, inet->inet_daddr,
&rds_tcp_transport, GFP_KERNEL);
if (IS_ERR(conn)) {
ret = PTR_ERR(conn);
goto out;
}
/*
* see the comment above rds_queue_delayed_reconnect()
*/
if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
if (rds_conn_state(conn) == RDS_CONN_UP)
rds_tcp_stats_inc(s_tcp_listen_closed_stale);
else
rds_tcp_stats_inc(s_tcp_connect_raced);
rds_conn_drop(conn);
ret = 0;
goto out;
}
rds_tcp_set_callbacks(new_sock, conn);
rds_connect_complete(conn);
new_sock = NULL;
ret = 0;
out:
if (new_sock)
sock_release(new_sock);
return ret;
}
void rds_connect_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_conn_w.work);
int ret;
clear_bit(RDS_RECONNECT_PENDING, &conn->c_flags);
if (rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
ret = conn->c_trans->conn_connect(conn);
rdsdebug("conn %p for %pI4 to %pI4 dispatched, ret %d\n",
conn, &conn->c_laddr, &conn->c_faddr, ret);
if (ret) {
if (rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_DOWN))
rds_queue_reconnect(conn);
else
rds_conn_error(conn, "RDS: connect failed\n");
}
}
}
void rds_shutdown_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_down_w);
if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_DOWN)) {
mutex_lock(&conn->c_cm_lock);
if (!rds_conn_transition(conn, RDS_CONN_UP, RDS_CONN_DISCONNECTING)
&& !rds_conn_transition(conn, RDS_CONN_ERROR, RDS_CONN_DISCONNECTING)) {
rds_conn_error(conn, "shutdown called in state %d\n",
atomic_read(&conn->c_state));
mutex_unlock(&conn->c_cm_lock);
return;
}
mutex_unlock(&conn->c_cm_lock);
mutex_lock(&conn->c_send_lock);
conn->c_trans->conn_shutdown(conn);
rds_conn_reset(conn);
mutex_unlock(&conn->c_send_lock);
if (!rds_conn_transition(conn, RDS_CONN_DISCONNECTING, RDS_CONN_DOWN)) {
rds_conn_error(conn,
"%s: failed to transition to state DOWN, "
"current state is %d\n",
__func__,
atomic_read(&conn->c_state));
return;
}
}
cancel_delayed_work(&conn->c_conn_w);
if (!hlist_unhashed(&conn->c_hash_node))
rds_queue_reconnect(conn);
}
void rds_connect_complete(struct rds_connection *conn)
{
if (!rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_UP)) {
printk(KERN_WARNING "%s: Cannot transition to state UP, "
"current state is %d\n",
__func__,
atomic_read(&conn->c_state));
atomic_set(&conn->c_state, RDS_CONN_ERROR);
queue_work(rds_wq, &conn->c_down_w);
return;
}
rdsdebug("conn %p for %pI4 to %pI4 complete\n",
conn, &conn->c_laddr, &conn->c_faddr);
conn->c_reconnect_jiffies = 0;
set_bit(0, &conn->c_map_queued);
queue_delayed_work(rds_wq, &conn->c_send_w, 0);
queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
}
int rds_tcp_accept_one(struct socket *sock)
{
struct socket *new_sock = NULL;
struct rds_connection *conn;
int ret;
struct inet_sock *inet;
struct rds_tcp_connection *rs_tcp;
ret = sock_create_kern(sock_net(sock->sk), sock->sk->sk_family,
sock->sk->sk_type, sock->sk->sk_protocol,
&new_sock);
if (ret)
goto out;
new_sock->type = sock->type;
new_sock->ops = sock->ops;
ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
if (ret < 0)
goto out;
ret = rds_tcp_keepalive(new_sock);
if (ret < 0)
goto out;
rds_tcp_tune(new_sock);
inet = inet_sk(new_sock->sk);
rdsdebug("accepted tcp %pI4:%u -> %pI4:%u\n",
&inet->inet_saddr, ntohs(inet->inet_sport),
&inet->inet_daddr, ntohs(inet->inet_dport));
conn = rds_conn_create(sock_net(sock->sk),
inet->inet_saddr, inet->inet_daddr,
&rds_tcp_transport, GFP_KERNEL);
if (IS_ERR(conn)) {
ret = PTR_ERR(conn);
goto out;
}
/* An incoming SYN request came in, and TCP just accepted it.
*
* If the client reboots, this conn will need to be cleaned up.
* rds_tcp_state_change() will do that cleanup
*/
rs_tcp = (struct rds_tcp_connection *)conn->c_transport_data;
if (rs_tcp->t_sock &&
ntohl(inet->inet_saddr) < ntohl(inet->inet_daddr)) {
struct sock *nsk = new_sock->sk;
nsk->sk_user_data = NULL;
nsk->sk_prot->disconnect(nsk, 0);
tcp_done(nsk);
new_sock = NULL;
ret = 0;
goto out;
} else if (rs_tcp->t_sock) {
rds_tcp_restore_callbacks(rs_tcp->t_sock, rs_tcp);
conn->c_outgoing = 0;
}
rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING);
rds_tcp_set_callbacks(new_sock, conn);
rds_connect_complete(conn);
new_sock = NULL;
ret = 0;
out:
if (new_sock)
sock_release(new_sock);
return ret;
}
