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 <net/sock.h>
35 #include <linux/in.h>
36 #include <linux/if_arp.h>
37 #include <linux/jhash.h>
38 #include <linux/ratelimit.h>
39 #include "rds.h"
40
41 static struct rhashtable bind_hash_table;
42
43 static const struct rhashtable_params ht_parms = {
44 .nelem_hint = 768,
45 .key_len = sizeof(u64),
46 .key_offset = offsetof(struct rds_sock, rs_bound_key),
47 .head_offset = offsetof(struct rds_sock, rs_bound_node),
48 .max_size = 16384,
49 .min_size = 1024,
50 };
51
52 /*
53 * Return the rds_sock bound at the given local address.
54 *
55 * The rx path can race with rds_release. We notice if rds_release() has
56 * marked this socket and don't return a rs ref to the rx path.
57 */
rds_find_bound(__be32 addr,__be16 port)58 struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
59 {
60 u64 key = ((u64)addr << 32) | port;
61 struct rds_sock *rs;
62
63 rcu_read_lock();
64 rs = rhashtable_lookup(&bind_hash_table, &key, ht_parms);
65 if (rs && (sock_flag(rds_rs_to_sk(rs), SOCK_DEAD) ||
66 !refcount_inc_not_zero(&rds_rs_to_sk(rs)->sk_refcnt)))
67 rs = NULL;
68
69 rcu_read_unlock();
70
71 rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
72 ntohs(port));
73
74 return rs;
75 }
76
77 /* returns -ve errno or +ve port */
rds_add_bound(struct rds_sock * rs,__be32 addr,__be16 * port)78 static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
79 {
80 int ret = -EADDRINUSE;
81 u16 rover, last;
82 u64 key;
83
84 if (*port != 0) {
85 rover = be16_to_cpu(*port);
86 if (rover == RDS_FLAG_PROBE_PORT)
87 return -EINVAL;
88 last = rover;
89 } else {
90 rover = max_t(u16, prandom_u32(), 2);
91 last = rover - 1;
92 }
93
94 do {
95 if (rover == 0)
96 rover++;
97
98 if (rover == RDS_FLAG_PROBE_PORT)
99 continue;
100 key = ((u64)addr << 32) | cpu_to_be16(rover);
101 if (rhashtable_lookup_fast(&bind_hash_table, &key, ht_parms))
102 continue;
103
104 rs->rs_bound_key = key;
105 rs->rs_bound_addr = addr;
106 net_get_random_once(&rs->rs_hash_initval,
107 sizeof(rs->rs_hash_initval));
108 rs->rs_bound_port = cpu_to_be16(rover);
109 rs->rs_bound_node.next = NULL;
110 rds_sock_addref(rs);
111 if (!rhashtable_insert_fast(&bind_hash_table,
112 &rs->rs_bound_node, ht_parms)) {
113 *port = rs->rs_bound_port;
114 ret = 0;
115 rdsdebug("rs %p binding to %pI4:%d\n",
116 rs, &addr, (int)ntohs(*port));
117 break;
118 } else {
119 rs->rs_bound_addr = 0;
120 rds_sock_put(rs);
121 ret = -ENOMEM;
122 break;
123 }
124 } while (rover++ != last);
125
126 return ret;
127 }
128
rds_remove_bound(struct rds_sock * rs)129 void rds_remove_bound(struct rds_sock *rs)
130 {
131
132 if (!rs->rs_bound_addr)
133 return;
134
135 rdsdebug("rs %p unbinding from %pI4:%d\n",
136 rs, &rs->rs_bound_addr,
137 ntohs(rs->rs_bound_port));
138
139 rhashtable_remove_fast(&bind_hash_table, &rs->rs_bound_node, ht_parms);
140 rds_sock_put(rs);
141 rs->rs_bound_addr = 0;
142 }
143
rds_bind(struct socket * sock,struct sockaddr * uaddr,int addr_len)144 int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
145 {
146 struct sock *sk = sock->sk;
147 struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
148 struct rds_sock *rs = rds_sk_to_rs(sk);
149 struct rds_transport *trans;
150 int ret = 0;
151
152 lock_sock(sk);
153
154 if (addr_len != sizeof(struct sockaddr_in) ||
155 sin->sin_family != AF_INET ||
156 rs->rs_bound_addr ||
157 sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
158 ret = -EINVAL;
159 goto out;
160 }
161
162 sock_set_flag(sk, SOCK_RCU_FREE);
163 ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
164 if (ret)
165 goto out;
166
167 if (rs->rs_transport) { /* previously bound */
168 trans = rs->rs_transport;
169 if (trans->laddr_check(sock_net(sock->sk),
170 sin->sin_addr.s_addr) != 0) {
171 ret = -ENOPROTOOPT;
172 rds_remove_bound(rs);
173 } else {
174 ret = 0;
175 }
176 goto out;
177 }
178 trans = rds_trans_get_preferred(sock_net(sock->sk),
179 sin->sin_addr.s_addr);
180 if (!trans) {
181 ret = -EADDRNOTAVAIL;
182 rds_remove_bound(rs);
183 pr_info_ratelimited("RDS: %s could not find a transport for %pI4, load rds_tcp or rds_rdma?\n",
184 __func__, &sin->sin_addr.s_addr);
185 goto out;
186 }
187
188 rs->rs_transport = trans;
189 ret = 0;
190
191 out:
192 release_sock(sk);
193 return ret;
194 }
195
rds_bind_lock_destroy(void)196 void rds_bind_lock_destroy(void)
197 {
198 rhashtable_destroy(&bind_hash_table);
199 }
200
rds_bind_lock_init(void)201 int rds_bind_lock_init(void)
202 {
203 return rhashtable_init(&bind_hash_table, &ht_parms);
204 }
205