1 /* 2 * Copyright (c) 2006 Oracle. All rights reserved. 3 * 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: 9 * 10 * Redistribution and use in source and binary forms, with or 11 * without modification, are permitted provided that the following 12 * conditions are met: 13 * 14 * - Redistributions of source code must retain the above 15 * copyright notice, this list of conditions and the following 16 * disclaimer. 17 * 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. 22 * 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 30 * SOFTWARE. 31 * 32 */ 33 #include <linux/kernel.h> 34 #include <linux/in.h> 35 #include <net/tcp.h> 36 37 #include "rds.h" 38 #include "tcp.h" 39 40 void rds_tcp_state_change(struct sock *sk) 41 { 42 void (*state_change)(struct sock *sk); 43 struct rds_connection *conn; 44 struct rds_tcp_connection *tc; 45 46 read_lock_bh(&sk->sk_callback_lock); 47 conn = sk->sk_user_data; 48 if (!conn) { 49 state_change = sk->sk_state_change; 50 goto out; 51 } 52 tc = conn->c_transport_data; 53 state_change = tc->t_orig_state_change; 54 55 rdsdebug("sock %p state_change to %d\n", tc->t_sock, sk->sk_state); 56 57 switch(sk->sk_state) { 58 /* ignore connecting sockets as they make progress */ 59 case TCP_SYN_SENT: 60 case TCP_SYN_RECV: 61 break; 62 case TCP_ESTABLISHED: 63 rds_connect_complete(conn); 64 break; 65 case TCP_CLOSE_WAIT: 66 case TCP_CLOSE: 67 rds_conn_drop(conn); 68 default: 69 break; 70 } 71 out: 72 read_unlock_bh(&sk->sk_callback_lock); 73 state_change(sk); 74 } 75 76 int rds_tcp_conn_connect(struct rds_connection *conn) 77 { 78 struct socket *sock = NULL; 79 struct sockaddr_in src, dest; 80 int ret; 81 struct rds_tcp_connection *tc = conn->c_transport_data; 82 83 mutex_lock(&tc->t_conn_lock); 84 85 if (rds_conn_up(conn)) { 86 mutex_unlock(&tc->t_conn_lock); 87 return 0; 88 } 89 ret = sock_create_kern(rds_conn_net(conn), PF_INET, 90 SOCK_STREAM, IPPROTO_TCP, &sock); 91 if (ret < 0) 92 goto out; 93 94 rds_tcp_tune(sock); 95 96 src.sin_family = AF_INET; 97 src.sin_addr.s_addr = (__force u32)conn->c_laddr; 98 src.sin_port = (__force u16)htons(0); 99 100 ret = sock->ops->bind(sock, (struct sockaddr *)&src, sizeof(src)); 101 if (ret) { 102 rdsdebug("bind failed with %d at address %pI4\n", 103 ret, &conn->c_laddr); 104 goto out; 105 } 106 107 dest.sin_family = AF_INET; 108 dest.sin_addr.s_addr = (__force u32)conn->c_faddr; 109 dest.sin_port = (__force u16)htons(RDS_TCP_PORT); 110 111 /* 112 * once we call connect() we can start getting callbacks and they 113 * own the socket 114 */ 115 rds_tcp_set_callbacks(sock, conn); 116 ret = sock->ops->connect(sock, (struct sockaddr *)&dest, sizeof(dest), 117 O_NONBLOCK); 118 119 rdsdebug("connect to address %pI4 returned %d\n", &conn->c_faddr, ret); 120 if (ret == -EINPROGRESS) 121 ret = 0; 122 if (ret == 0) { 123 rds_tcp_keepalive(sock); 124 sock = NULL; 125 } else { 126 rds_tcp_restore_callbacks(sock, conn->c_transport_data); 127 } 128 129 out: 130 mutex_unlock(&tc->t_conn_lock); 131 if (sock) 132 sock_release(sock); 133 return ret; 134 } 135 136 /* 137 * Before killing the tcp socket this needs to serialize with callbacks. The 138 * caller has already grabbed the sending sem so we're serialized with other 139 * senders. 140 * 141 * TCP calls the callbacks with the sock lock so we hold it while we reset the 142 * callbacks to those set by TCP. Our callbacks won't execute again once we 143 * hold the sock lock. 144 */ 145 void rds_tcp_conn_shutdown(struct rds_connection *conn) 146 { 147 struct rds_tcp_connection *tc = conn->c_transport_data; 148 struct socket *sock = tc->t_sock; 149 150 rdsdebug("shutting down conn %p tc %p sock %p\n", conn, tc, sock); 151 152 if (sock) { 153 sock->ops->shutdown(sock, RCV_SHUTDOWN | SEND_SHUTDOWN); 154 lock_sock(sock->sk); 155 rds_tcp_restore_callbacks(sock, tc); /* tc->tc_sock = NULL */ 156 157 release_sock(sock->sk); 158 sock_release(sock); 159 } 160 161 if (tc->t_tinc) { 162 rds_inc_put(&tc->t_tinc->ti_inc); 163 tc->t_tinc = NULL; 164 } 165 tc->t_tinc_hdr_rem = sizeof(struct rds_header); 166 tc->t_tinc_data_rem = 0; 167 } 168