2 * Copyright (c) 2006, 2017 Oracle and/or its affiliates. All rights reserved.
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
33 #include <linux/kernel.h>
40 void rds_tcp_state_change(struct sock *sk)
42 void (*state_change)(struct sock *sk);
43 struct rds_conn_path *cp;
44 struct rds_tcp_connection *tc;
46 read_lock_bh(&sk->sk_callback_lock);
47 cp = sk->sk_user_data;
49 state_change = sk->sk_state_change;
52 tc = cp->cp_transport_data;
53 state_change = tc->t_orig_state_change;
55 rdsdebug("sock %p state_change to %d\n", tc->t_sock, sk->sk_state);
57 switch (sk->sk_state) {
58 /* ignore connecting sockets as they make progress */
63 /* Force the peer to reconnect so that we have the
64 * TCP ports going from <smaller-ip>.<transient> to
65 * <larger-ip>.<RDS_TCP_PORT>. We avoid marking the
66 * RDS connection as RDS_CONN_UP until the reconnect,
67 * to avoid RDS datagram loss.
69 if (rds_addr_cmp(&cp->cp_conn->c_laddr,
70 &cp->cp_conn->c_faddr) >= 0 &&
71 rds_conn_path_transition(cp, RDS_CONN_CONNECTING,
73 rds_conn_path_drop(cp, false);
75 rds_connect_path_complete(cp, RDS_CONN_CONNECTING);
80 rds_conn_path_drop(cp, false);
86 read_unlock_bh(&sk->sk_callback_lock);
90 int rds_tcp_conn_path_connect(struct rds_conn_path *cp)
92 struct socket *sock = NULL;
93 struct sockaddr_in6 sin6;
94 struct sockaddr_in sin;
95 struct sockaddr *addr;
99 struct rds_connection *conn = cp->cp_conn;
100 struct rds_tcp_connection *tc = cp->cp_transport_data;
102 /* for multipath rds,we only trigger the connection after
103 * the handshake probe has determined the number of paths.
105 if (cp->cp_index > 0 && cp->cp_conn->c_npaths < 2)
108 mutex_lock(&tc->t_conn_path_lock);
110 if (rds_conn_path_up(cp)) {
111 mutex_unlock(&tc->t_conn_path_lock);
114 if (ipv6_addr_v4mapped(&conn->c_laddr)) {
115 ret = sock_create_kern(rds_conn_net(conn), PF_INET,
116 SOCK_STREAM, IPPROTO_TCP, &sock);
119 ret = sock_create_kern(rds_conn_net(conn), PF_INET6,
120 SOCK_STREAM, IPPROTO_TCP, &sock);
127 if (!rds_tcp_tune(sock)) {
133 sin6.sin6_family = AF_INET6;
134 sin6.sin6_addr = conn->c_laddr;
136 sin6.sin6_flowinfo = 0;
137 sin6.sin6_scope_id = conn->c_dev_if;
138 addr = (struct sockaddr *)&sin6;
139 addrlen = sizeof(sin6);
141 sin.sin_family = AF_INET;
142 sin.sin_addr.s_addr = conn->c_laddr.s6_addr32[3];
144 addr = (struct sockaddr *)&sin;
145 addrlen = sizeof(sin);
148 ret = sock->ops->bind(sock, addr, addrlen);
150 rdsdebug("bind failed with %d at address %pI6c\n",
151 ret, &conn->c_laddr);
156 sin6.sin6_family = AF_INET6;
157 sin6.sin6_addr = conn->c_faddr;
158 sin6.sin6_port = htons(RDS_TCP_PORT);
159 sin6.sin6_flowinfo = 0;
160 sin6.sin6_scope_id = conn->c_dev_if;
161 addr = (struct sockaddr *)&sin6;
162 addrlen = sizeof(sin6);
164 sin.sin_family = AF_INET;
165 sin.sin_addr.s_addr = conn->c_faddr.s6_addr32[3];
166 sin.sin_port = htons(RDS_TCP_PORT);
167 addr = (struct sockaddr *)&sin;
168 addrlen = sizeof(sin);
172 * once we call connect() we can start getting callbacks and they
175 rds_tcp_set_callbacks(sock, cp);
176 ret = sock->ops->connect(sock, addr, addrlen, O_NONBLOCK);
178 rdsdebug("connect to address %pI6c returned %d\n", &conn->c_faddr, ret);
179 if (ret == -EINPROGRESS)
182 rds_tcp_keepalive(sock);
185 rds_tcp_restore_callbacks(sock, cp->cp_transport_data);
189 mutex_unlock(&tc->t_conn_path_lock);
196 * Before killing the tcp socket this needs to serialize with callbacks. The
197 * caller has already grabbed the sending sem so we're serialized with other
200 * TCP calls the callbacks with the sock lock so we hold it while we reset the
201 * callbacks to those set by TCP. Our callbacks won't execute again once we
202 * hold the sock lock.
204 void rds_tcp_conn_path_shutdown(struct rds_conn_path *cp)
206 struct rds_tcp_connection *tc = cp->cp_transport_data;
207 struct socket *sock = tc->t_sock;
209 rdsdebug("shutting down conn %p tc %p sock %p\n",
210 cp->cp_conn, tc, sock);
213 if (rds_destroy_pending(cp->cp_conn))
214 sock_no_linger(sock->sk);
215 sock->ops->shutdown(sock, RCV_SHUTDOWN | SEND_SHUTDOWN);
217 rds_tcp_restore_callbacks(sock, tc); /* tc->tc_sock = NULL */
219 release_sock(sock->sk);
224 rds_inc_put(&tc->t_tinc->ti_inc);
227 tc->t_tinc_hdr_rem = sizeof(struct rds_header);
228 tc->t_tinc_data_rem = 0;