2 * Copyright (c) 2006 Oracle. All rights reserved.
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:
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
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.
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
33 #include <linux/kernel.h>
36 #include <linux/if_arp.h>
37 #include <linux/jhash.h>
38 #include <linux/ratelimit.h>
43 struct hlist_head head;
46 #define BIND_HASH_SIZE 1024
47 static struct bind_bucket bind_hash_table[BIND_HASH_SIZE];
49 static struct bind_bucket *hash_to_bucket(__be32 addr, __be16 port)
51 return bind_hash_table + (jhash_2words((u32)addr, (u32)port, 0) &
52 (BIND_HASH_SIZE - 1));
55 /* must hold either read or write lock (write lock for insert != NULL) */
56 static struct rds_sock *rds_bind_lookup(struct bind_bucket *bucket,
57 __be32 addr, __be16 port,
58 struct rds_sock *insert)
61 struct hlist_head *head = &bucket->head;
63 u64 needle = ((u64)be32_to_cpu(addr) << 32) | be16_to_cpu(port);
65 hlist_for_each_entry(rs, head, rs_bound_node) {
66 cmp = ((u64)be32_to_cpu(rs->rs_bound_addr) << 32) |
67 be16_to_cpu(rs->rs_bound_port);
77 * make sure our addr and port are set before
78 * we are added to the list.
80 insert->rs_bound_addr = addr;
81 insert->rs_bound_port = port;
82 rds_sock_addref(insert);
84 hlist_add_head(&insert->rs_bound_node, head);
90 * Return the rds_sock bound at the given local address.
92 * The rx path can race with rds_release. We notice if rds_release() has
93 * marked this socket and don't return a rs ref to the rx path.
95 struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
99 struct bind_bucket *bucket = hash_to_bucket(addr, port);
101 read_lock_irqsave(&bucket->lock, flags);
102 rs = rds_bind_lookup(bucket, addr, port, NULL);
103 read_unlock_irqrestore(&bucket->lock, flags);
105 if (rs && sock_flag(rds_rs_to_sk(rs), SOCK_DEAD)) {
110 rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
116 /* returns -ve errno or +ve port */
117 static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
120 int ret = -EADDRINUSE;
122 struct bind_bucket *bucket;
125 rover = be16_to_cpu(*port);
128 rover = max_t(u16, prandom_u32(), 2);
133 struct rds_sock *rrs;
137 bucket = hash_to_bucket(addr, cpu_to_be16(rover));
138 write_lock_irqsave(&bucket->lock, flags);
139 rrs = rds_bind_lookup(bucket, addr, cpu_to_be16(rover), rs);
140 write_unlock_irqrestore(&bucket->lock, flags);
142 *port = rs->rs_bound_port;
144 rdsdebug("rs %p binding to %pI4:%d\n",
145 rs, &addr, (int)ntohs(*port));
150 } while (rover++ != last);
155 void rds_remove_bound(struct rds_sock *rs)
158 struct bind_bucket *bucket =
159 hash_to_bucket(rs->rs_bound_addr, rs->rs_bound_port);
161 write_lock_irqsave(&bucket->lock, flags);
163 if (rs->rs_bound_addr) {
164 rdsdebug("rs %p unbinding from %pI4:%d\n",
165 rs, &rs->rs_bound_addr,
166 ntohs(rs->rs_bound_port));
168 hlist_del_init(&rs->rs_bound_node);
170 rs->rs_bound_addr = 0;
173 write_unlock_irqrestore(&bucket->lock, flags);
176 int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
178 struct sock *sk = sock->sk;
179 struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
180 struct rds_sock *rs = rds_sk_to_rs(sk);
181 struct rds_transport *trans;
186 if (addr_len != sizeof(struct sockaddr_in) ||
187 sin->sin_family != AF_INET ||
189 sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
194 ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
198 if (rs->rs_transport) { /* previously bound */
202 trans = rds_trans_get_preferred(sock_net(sock->sk),
203 sin->sin_addr.s_addr);
205 ret = -EADDRNOTAVAIL;
206 rds_remove_bound(rs);
207 printk_ratelimited(KERN_INFO "RDS: rds_bind() could not find a transport, "
208 "load rds_tcp or rds_rdma?\n");
212 rs->rs_transport = trans;
220 void rds_bind_lock_init(void)
224 for (i = 0; i < BIND_HASH_SIZE; i++)
225 rwlock_init(&bind_hash_table[i].lock);