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#define BIND_HASH_SIZE 1024
42static struct hlist_head bind_hash_table[BIND_HASH_SIZE];
43static DEFINE_SPINLOCK(rds_bind_lock);
44
45static struct hlist_head *hash_to_bucket(__be32 addr, __be16 port)
46{
47	return bind_hash_table + (jhash_2words((u32)addr, (u32)port, 0) &
48				  (BIND_HASH_SIZE - 1));
49}
50
51static struct rds_sock *rds_bind_lookup(__be32 addr, __be16 port,
52					struct rds_sock *insert)
53{
54	struct rds_sock *rs;
55	struct hlist_head *head = hash_to_bucket(addr, port);
56	u64 cmp;
57	u64 needle = ((u64)be32_to_cpu(addr) << 32) | be16_to_cpu(port);
58
59	rcu_read_lock();
60	hlist_for_each_entry_rcu(rs, head, rs_bound_node) {
61		cmp = ((u64)be32_to_cpu(rs->rs_bound_addr) << 32) |
62		      be16_to_cpu(rs->rs_bound_port);
63
64		if (cmp == needle) {
65			rcu_read_unlock();
66			return rs;
67		}
68	}
69	rcu_read_unlock();
70
71	if (insert) {
72		/*
73		 * make sure our addr and port are set before
74		 * we are added to the list, other people
75		 * in rcu will find us as soon as the
76		 * hlist_add_head_rcu is done
77		 */
78		insert->rs_bound_addr = addr;
79		insert->rs_bound_port = port;
80		rds_sock_addref(insert);
81
82		hlist_add_head_rcu(&insert->rs_bound_node, head);
83	}
84	return NULL;
85}
86
87/*
88 * Return the rds_sock bound at the given local address.
89 *
90 * The rx path can race with rds_release.  We notice if rds_release() has
91 * marked this socket and don't return a rs ref to the rx path.
92 */
93struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
94{
95	struct rds_sock *rs;
96
97	rs = rds_bind_lookup(addr, port, NULL);
98
99	if (rs && !sock_flag(rds_rs_to_sk(rs), SOCK_DEAD))
100		rds_sock_addref(rs);
101	else
102		rs = NULL;
103
104	rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
105		ntohs(port));
106	return rs;
107}
108
109/* returns -ve errno or +ve port */
110static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
111{
112	unsigned long flags;
113	int ret = -EADDRINUSE;
114	u16 rover, last;
115
116	if (*port != 0) {
117		rover = be16_to_cpu(*port);
118		last = rover;
119	} else {
120		rover = max_t(u16, prandom_u32(), 2);
121		last = rover - 1;
122	}
123
124	spin_lock_irqsave(&rds_bind_lock, flags);
125
126	do {
127		if (rover == 0)
128			rover++;
129		if (!rds_bind_lookup(addr, cpu_to_be16(rover), rs)) {
130			*port = rs->rs_bound_port;
131			ret = 0;
132			rdsdebug("rs %p binding to %pI4:%d\n",
133			  rs, &addr, (int)ntohs(*port));
134			break;
135		}
136	} while (rover++ != last);
137
138	spin_unlock_irqrestore(&rds_bind_lock, flags);
139
140	return ret;
141}
142
143void rds_remove_bound(struct rds_sock *rs)
144{
145	unsigned long flags;
146
147	spin_lock_irqsave(&rds_bind_lock, flags);
148
149	if (rs->rs_bound_addr) {
150		rdsdebug("rs %p unbinding from %pI4:%d\n",
151		  rs, &rs->rs_bound_addr,
152		  ntohs(rs->rs_bound_port));
153
154		hlist_del_init_rcu(&rs->rs_bound_node);
155		rds_sock_put(rs);
156		rs->rs_bound_addr = 0;
157	}
158
159	spin_unlock_irqrestore(&rds_bind_lock, flags);
160}
161
162int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
163{
164	struct sock *sk = sock->sk;
165	struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
166	struct rds_sock *rs = rds_sk_to_rs(sk);
167	struct rds_transport *trans;
168	int ret = 0;
169
170	lock_sock(sk);
171
172	if (addr_len != sizeof(struct sockaddr_in) ||
173	    sin->sin_family != AF_INET ||
174	    rs->rs_bound_addr ||
175	    sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
176		ret = -EINVAL;
177		goto out;
178	}
179
180	ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
181	if (ret)
182		goto out;
183
184	trans = rds_trans_get_preferred(sin->sin_addr.s_addr);
185	if (!trans) {
186		ret = -EADDRNOTAVAIL;
187		rds_remove_bound(rs);
188		printk_ratelimited(KERN_INFO "RDS: rds_bind() could not find a transport, "
189				"load rds_tcp or rds_rdma?\n");
190		goto out;
191	}
192
193	rs->rs_transport = trans;
194	ret = 0;
195
196out:
197	release_sock(sk);
198
199	/* we might have called rds_remove_bound on error */
200	if (ret)
201		synchronize_rcu();
202	return ret;
203}
204