LXR linux/net/rds/tcp

   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
  40void 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(&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        }
  71out:
  72        read_unlock(&sk->sk_callback_lock);
  73        state_change(sk);
  74}
  75
  76int rds_tcp_conn_connect(struct rds_connection *conn)
  77{
  78        struct socket *sock = NULL;
  79        struct sockaddr_in src, dest;
  80        int ret;
  81
  82        ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
  83        if (ret < 0)
  84                goto out;
  85
  86        rds_tcp_tune(sock);
  87
  88        src.sin_family = AF_INET;
  89        src.sin_addr.s_addr = (__force u32)conn->c_laddr;
  90        src.sin_port = (__force u16)htons(0);
  91
  92        ret = sock->ops->bind(sock, (struct sockaddr *)&src, sizeof(src));
  93        if (ret) {
  94                rdsdebug("bind failed with %d at address %pI4\n",
  95                         ret, &conn->c_laddr);
  96                goto out;
  97        }
  98
  99        dest.sin_family = AF_INET;
 100        dest.sin_addr.s_addr = (__force u32)conn->c_faddr;
 101        dest.sin_port = (__force u16)htons(RDS_TCP_PORT);
 102
 103        /*
 104         * once we call connect() we can start getting callbacks and they
 105         * own the socket
 106         */
 107        rds_tcp_set_callbacks(sock, conn);
 108        ret = sock->ops->connect(sock, (struct sockaddr *)&dest, sizeof(dest),
 109                                 O_NONBLOCK);
 110
 111        rdsdebug("connect to address %pI4 returned %d\n", &conn->c_faddr, ret);
 112        if (ret == -EINPROGRESS)
 113                ret = 0;
 114        if (ret == 0)
 115                sock = NULL;
 116        else
 117                rds_tcp_restore_callbacks(sock, conn->c_transport_data);
 118
 119out:
 120        if (sock)
 121                sock_release(sock);
 122        return ret;
 123}
 124
 125/*
 126 * Before killing the tcp socket this needs to serialize with callbacks.  The
 127 * caller has already grabbed the sending sem so we're serialized with other
 128 * senders.
 129 *
 130 * TCP calls the callbacks with the sock lock so we hold it while we reset the
 131 * callbacks to those set by TCP.  Our callbacks won't execute again once we
 132 * hold the sock lock.
 133 */
 134void rds_tcp_conn_shutdown(struct rds_connection *conn)
 135{
 136        struct rds_tcp_connection *tc = conn->c_transport_data;
 137        struct socket *sock = tc->t_sock;
 138
 139        rdsdebug("shutting down conn %p tc %p sock %p\n", conn, tc, sock);
 140
 141        if (sock) {
 142                sock->ops->shutdown(sock, RCV_SHUTDOWN | SEND_SHUTDOWN);
 143                lock_sock(sock->sk);
 144                rds_tcp_restore_callbacks(sock, tc); /* tc->tc_sock = NULL */
 145
 146                release_sock(sock->sk);
 147                sock_release(sock);
 148        }
 149
 150        if (tc->t_tinc) {
 151                rds_inc_put(&tc->t_tinc->ti_inc);
 152                tc->t_tinc = NULL;
 153        }
 154        tc->t_tinc_hdr_rem = sizeof(struct rds_header);
 155        tc->t_tinc_data_rem = 0;
 156}
 157