tcp_listen.c 5.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201
  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/gfp.h>
  35. #include <linux/in.h>
  36. #include <net/tcp.h>
  37. #include "rds.h"
  38. #include "tcp.h"
  39. /*
  40. * cheesy, but simple..
  41. */
  42. static void rds_tcp_accept_worker(struct work_struct *work);
  43. static DECLARE_WORK(rds_tcp_listen_work, rds_tcp_accept_worker);
  44. static struct socket *rds_tcp_listen_sock;
  45. static int rds_tcp_accept_one(struct socket *sock)
  46. {
  47. struct socket *new_sock = NULL;
  48. struct rds_connection *conn;
  49. int ret;
  50. struct inet_sock *inet;
  51. ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
  52. sock->sk->sk_protocol, &new_sock);
  53. if (ret)
  54. goto out;
  55. new_sock->type = sock->type;
  56. new_sock->ops = sock->ops;
  57. ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
  58. if (ret < 0)
  59. goto out;
  60. rds_tcp_tune(new_sock);
  61. inet = inet_sk(new_sock->sk);
  62. rdsdebug("accepted tcp %pI4:%u -> %pI4:%u\n",
  63. &inet->inet_saddr, ntohs(inet->inet_sport),
  64. &inet->inet_daddr, ntohs(inet->inet_dport));
  65. conn = rds_conn_create(inet->inet_saddr, inet->inet_daddr,
  66. &rds_tcp_transport, GFP_KERNEL);
  67. if (IS_ERR(conn)) {
  68. ret = PTR_ERR(conn);
  69. goto out;
  70. }
  71. /*
  72. * see the comment above rds_queue_delayed_reconnect()
  73. */
  74. if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
  75. if (rds_conn_state(conn) == RDS_CONN_UP)
  76. rds_tcp_stats_inc(s_tcp_listen_closed_stale);
  77. else
  78. rds_tcp_stats_inc(s_tcp_connect_raced);
  79. rds_conn_drop(conn);
  80. ret = 0;
  81. goto out;
  82. }
  83. rds_tcp_set_callbacks(new_sock, conn);
  84. rds_connect_complete(conn);
  85. new_sock = NULL;
  86. ret = 0;
  87. out:
  88. if (new_sock)
  89. sock_release(new_sock);
  90. return ret;
  91. }
  92. static void rds_tcp_accept_worker(struct work_struct *work)
  93. {
  94. while (rds_tcp_accept_one(rds_tcp_listen_sock) == 0)
  95. cond_resched();
  96. }
  97. void rds_tcp_listen_data_ready(struct sock *sk, int bytes)
  98. {
  99. void (*ready)(struct sock *sk, int bytes);
  100. rdsdebug("listen data ready sk %p\n", sk);
  101. read_lock_bh(&sk->sk_callback_lock);
  102. ready = sk->sk_user_data;
  103. if (!ready) { /* check for teardown race */
  104. ready = sk->sk_data_ready;
  105. goto out;
  106. }
  107. /*
  108. * ->sk_data_ready is also called for a newly established child socket
  109. * before it has been accepted and the accepter has set up their
  110. * data_ready.. we only want to queue listen work for our listening
  111. * socket
  112. */
  113. if (sk->sk_state == TCP_LISTEN)
  114. queue_work(rds_wq, &rds_tcp_listen_work);
  115. out:
  116. read_unlock_bh(&sk->sk_callback_lock);
  117. ready(sk, bytes);
  118. }
  119. int rds_tcp_listen_init(void)
  120. {
  121. struct sockaddr_in sin;
  122. struct socket *sock = NULL;
  123. int ret;
  124. ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
  125. if (ret < 0)
  126. goto out;
  127. sock->sk->sk_reuse = 1;
  128. rds_tcp_nonagle(sock);
  129. write_lock_bh(&sock->sk->sk_callback_lock);
  130. sock->sk->sk_user_data = sock->sk->sk_data_ready;
  131. sock->sk->sk_data_ready = rds_tcp_listen_data_ready;
  132. write_unlock_bh(&sock->sk->sk_callback_lock);
  133. sin.sin_family = PF_INET,
  134. sin.sin_addr.s_addr = (__force u32)htonl(INADDR_ANY);
  135. sin.sin_port = (__force u16)htons(RDS_TCP_PORT);
  136. ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin));
  137. if (ret < 0)
  138. goto out;
  139. ret = sock->ops->listen(sock, 64);
  140. if (ret < 0)
  141. goto out;
  142. rds_tcp_listen_sock = sock;
  143. sock = NULL;
  144. out:
  145. if (sock)
  146. sock_release(sock);
  147. return ret;
  148. }
  149. void rds_tcp_listen_stop(void)
  150. {
  151. struct socket *sock = rds_tcp_listen_sock;
  152. struct sock *sk;
  153. if (!sock)
  154. return;
  155. sk = sock->sk;
  156. /* serialize with and prevent further callbacks */
  157. lock_sock(sk);
  158. write_lock_bh(&sk->sk_callback_lock);
  159. if (sk->sk_user_data) {
  160. sk->sk_data_ready = sk->sk_user_data;
  161. sk->sk_user_data = NULL;
  162. }
  163. write_unlock_bh(&sk->sk_callback_lock);
  164. release_sock(sk);
  165. /* wait for accepts to stop and close the socket */
  166. flush_workqueue(rds_wq);
  167. sock_release(sock);
  168. rds_tcp_listen_sock = NULL;
  169. }