loop.c 7.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254
  1. /*
  2. * Copyright (c) 2006, 2017 Oracle and/or its affiliates. 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/slab.h>
  35. #include <linux/in.h>
  36. #include <net/net_namespace.h>
  37. #include <net/netns/generic.h>
  38. #include <linux/ipv6.h>
  39. #include "rds_single_path.h"
  40. #include "rds.h"
  41. #include "loop.h"
  42. static DEFINE_SPINLOCK(loop_conns_lock);
  43. static LIST_HEAD(loop_conns);
  44. static atomic_t rds_loop_unloading = ATOMIC_INIT(0);
  45. static void rds_loop_set_unloading(void)
  46. {
  47. atomic_set(&rds_loop_unloading, 1);
  48. }
  49. static bool rds_loop_is_unloading(struct rds_connection *conn)
  50. {
  51. return atomic_read(&rds_loop_unloading) != 0;
  52. }
  53. /*
  54. * This 'loopback' transport is a special case for flows that originate
  55. * and terminate on the same machine.
  56. *
  57. * Connection build-up notices if the destination address is thought of
  58. * as a local address by a transport. At that time it decides to use the
  59. * loopback transport instead of the bound transport of the sending socket.
  60. *
  61. * The loopback transport's sending path just hands the sent rds_message
  62. * straight to the receiving path via an embedded rds_incoming.
  63. */
  64. /*
  65. * Usually a message transits both the sender and receiver's conns as it
  66. * flows to the receiver. In the loopback case, though, the receive path
  67. * is handed the sending conn so the sense of the addresses is reversed.
  68. */
  69. static int rds_loop_xmit(struct rds_connection *conn, struct rds_message *rm,
  70. unsigned int hdr_off, unsigned int sg,
  71. unsigned int off)
  72. {
  73. struct scatterlist *sgp = &rm->data.op_sg[sg];
  74. int ret = sizeof(struct rds_header) +
  75. be32_to_cpu(rm->m_inc.i_hdr.h_len);
  76. /* Do not send cong updates to loopback */
  77. if (rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
  78. rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
  79. ret = min_t(int, ret, sgp->length - conn->c_xmit_data_off);
  80. goto out;
  81. }
  82. BUG_ON(hdr_off || sg || off);
  83. rds_inc_init(&rm->m_inc, conn, &conn->c_laddr);
  84. /* For the embedded inc. Matching put is in loop_inc_free() */
  85. rds_message_addref(rm);
  86. rds_recv_incoming(conn, &conn->c_laddr, &conn->c_faddr, &rm->m_inc,
  87. GFP_KERNEL);
  88. rds_send_drop_acked(conn, be64_to_cpu(rm->m_inc.i_hdr.h_sequence),
  89. NULL);
  90. rds_inc_put(&rm->m_inc);
  91. out:
  92. return ret;
  93. }
  94. /*
  95. * See rds_loop_xmit(). Since our inc is embedded in the rm, we
  96. * make sure the rm lives at least until the inc is done.
  97. */
  98. static void rds_loop_inc_free(struct rds_incoming *inc)
  99. {
  100. struct rds_message *rm = container_of(inc, struct rds_message, m_inc);
  101. rds_message_put(rm);
  102. }
  103. /* we need to at least give the thread something to succeed */
  104. static int rds_loop_recv_path(struct rds_conn_path *cp)
  105. {
  106. return 0;
  107. }
  108. struct rds_loop_connection {
  109. struct list_head loop_node;
  110. struct rds_connection *conn;
  111. };
  112. /*
  113. * Even the loopback transport needs to keep track of its connections,
  114. * so it can call rds_conn_destroy() on them on exit. N.B. there are
  115. * 1+ loopback addresses (127.*.*.*) so it's not a bug to have
  116. * multiple loopback conns allocated, although rather useless.
  117. */
  118. static int rds_loop_conn_alloc(struct rds_connection *conn, gfp_t gfp)
  119. {
  120. struct rds_loop_connection *lc;
  121. unsigned long flags;
  122. lc = kzalloc(sizeof(struct rds_loop_connection), gfp);
  123. if (!lc)
  124. return -ENOMEM;
  125. INIT_LIST_HEAD(&lc->loop_node);
  126. lc->conn = conn;
  127. conn->c_transport_data = lc;
  128. spin_lock_irqsave(&loop_conns_lock, flags);
  129. list_add_tail(&lc->loop_node, &loop_conns);
  130. spin_unlock_irqrestore(&loop_conns_lock, flags);
  131. return 0;
  132. }
  133. static void rds_loop_conn_free(void *arg)
  134. {
  135. struct rds_loop_connection *lc = arg;
  136. unsigned long flags;
  137. rdsdebug("lc %p\n", lc);
  138. spin_lock_irqsave(&loop_conns_lock, flags);
  139. list_del(&lc->loop_node);
  140. spin_unlock_irqrestore(&loop_conns_lock, flags);
  141. kfree(lc);
  142. }
  143. static int rds_loop_conn_path_connect(struct rds_conn_path *cp)
  144. {
  145. rds_connect_complete(cp->cp_conn);
  146. return 0;
  147. }
  148. static void rds_loop_conn_path_shutdown(struct rds_conn_path *cp)
  149. {
  150. }
  151. void rds_loop_exit(void)
  152. {
  153. struct rds_loop_connection *lc, *_lc;
  154. LIST_HEAD(tmp_list);
  155. rds_loop_set_unloading();
  156. synchronize_rcu();
  157. /* avoid calling conn_destroy with irqs off */
  158. spin_lock_irq(&loop_conns_lock);
  159. list_splice(&loop_conns, &tmp_list);
  160. INIT_LIST_HEAD(&loop_conns);
  161. spin_unlock_irq(&loop_conns_lock);
  162. list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
  163. WARN_ON(lc->conn->c_passive);
  164. rds_conn_destroy(lc->conn);
  165. }
  166. }
  167. static void rds_loop_kill_conns(struct net *net)
  168. {
  169. struct rds_loop_connection *lc, *_lc;
  170. LIST_HEAD(tmp_list);
  171. spin_lock_irq(&loop_conns_lock);
  172. list_for_each_entry_safe(lc, _lc, &loop_conns, loop_node) {
  173. struct net *c_net = read_pnet(&lc->conn->c_net);
  174. if (net != c_net)
  175. continue;
  176. list_move_tail(&lc->loop_node, &tmp_list);
  177. }
  178. spin_unlock_irq(&loop_conns_lock);
  179. list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
  180. WARN_ON(lc->conn->c_passive);
  181. rds_conn_destroy(lc->conn);
  182. }
  183. }
  184. static void __net_exit rds_loop_exit_net(struct net *net)
  185. {
  186. rds_loop_kill_conns(net);
  187. }
  188. static struct pernet_operations rds_loop_net_ops = {
  189. .exit = rds_loop_exit_net,
  190. };
  191. int rds_loop_net_init(void)
  192. {
  193. return register_pernet_device(&rds_loop_net_ops);
  194. }
  195. void rds_loop_net_exit(void)
  196. {
  197. unregister_pernet_device(&rds_loop_net_ops);
  198. }
  199. /*
  200. * This is missing .xmit_* because loop doesn't go through generic
  201. * rds_send_xmit() and doesn't call rds_recv_incoming(). .listen_stop and
  202. * .laddr_check are missing because transport.c doesn't iterate over
  203. * rds_loop_transport.
  204. */
  205. struct rds_transport rds_loop_transport = {
  206. .xmit = rds_loop_xmit,
  207. .recv_path = rds_loop_recv_path,
  208. .conn_alloc = rds_loop_conn_alloc,
  209. .conn_free = rds_loop_conn_free,
  210. .conn_path_connect = rds_loop_conn_path_connect,
  211. .conn_path_shutdown = rds_loop_conn_path_shutdown,
  212. .inc_copy_to_user = rds_message_inc_copy_to_user,
  213. .inc_free = rds_loop_inc_free,
  214. .t_name = "loopback",
  215. .t_type = RDS_TRANS_LOOP,
  216. .t_unloading = rds_loop_is_unloading,
  217. };