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 "rds.h"
38
39static DEFINE_SPINLOCK(rds_bind_lock);
40static struct rb_root rds_bind_tree = RB_ROOT;
41
42static struct rds_sock *rds_bind_tree_walk(__be32 addr, __be16 port,
43					   struct rds_sock *insert)
44{
45	struct rb_node **p = &rds_bind_tree.rb_node;
46	struct rb_node *parent = NULL;
47	struct rds_sock *rs;
48	u64 cmp;
49	u64 needle = ((u64)be32_to_cpu(addr) << 32) | be16_to_cpu(port);
50
51	while (*p) {
52		parent = *p;
53		rs = rb_entry(parent, struct rds_sock, rs_bound_node);
54
55		cmp = ((u64)be32_to_cpu(rs->rs_bound_addr) << 32) |
56		      be16_to_cpu(rs->rs_bound_port);
57
58		if (needle < cmp)
59			p = &(*p)->rb_left;
60		else if (needle > cmp)
61			p = &(*p)->rb_right;
62		else
63			return rs;
64	}
65
66	if (insert) {
67		rb_link_node(&insert->rs_bound_node, parent, p);
68		rb_insert_color(&insert->rs_bound_node, &rds_bind_tree);
69	}
70	return NULL;
71}
72
73/*
74 * Return the rds_sock bound at the given local address.
75 *
76 * The rx path can race with rds_release.  We notice if rds_release() has
77 * marked this socket and don't return a rs ref to the rx path.
78 */
79struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
80{
81	struct rds_sock *rs;
82	unsigned long flags;
83
84	spin_lock_irqsave(&rds_bind_lock, flags);
85	rs = rds_bind_tree_walk(addr, port, NULL);
86	if (rs && !sock_flag(rds_rs_to_sk(rs), SOCK_DEAD))
87		rds_sock_addref(rs);
88	else
89		rs = NULL;
90	spin_unlock_irqrestore(&rds_bind_lock, flags);
91
92	rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
93		ntohs(port));
94	return rs;
95}
96
97/* returns -ve errno or +ve port */
98static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
99{
100	unsigned long flags;
101	int ret = -EADDRINUSE;
102	u16 rover, last;
103
104	if (*port != 0) {
105		rover = be16_to_cpu(*port);
106		last = rover;
107	} else {
108		rover = max_t(u16, net_random(), 2);
109		last = rover - 1;
110	}
111
112	spin_lock_irqsave(&rds_bind_lock, flags);
113
114	do {
115		if (rover == 0)
116			rover++;
117		if (rds_bind_tree_walk(addr, cpu_to_be16(rover), rs) == NULL) {
118			*port = cpu_to_be16(rover);
119			ret = 0;
120			break;
121		}
122	} while (rover++ != last);
123
124	if (ret == 0)  {
125		rs->rs_bound_addr = addr;
126		rs->rs_bound_port = *port;
127		rds_sock_addref(rs);
128
129		rdsdebug("rs %p binding to %pI4:%d\n",
130		  rs, &addr, (int)ntohs(*port));
131	}
132
133	spin_unlock_irqrestore(&rds_bind_lock, flags);
134
135	return ret;
136}
137
138void rds_remove_bound(struct rds_sock *rs)
139{
140	unsigned long flags;
141
142	spin_lock_irqsave(&rds_bind_lock, flags);
143
144	if (rs->rs_bound_addr) {
145		rdsdebug("rs %p unbinding from %pI4:%d\n",
146		  rs, &rs->rs_bound_addr,
147		  ntohs(rs->rs_bound_port));
148
149		rb_erase(&rs->rs_bound_node, &rds_bind_tree);
150		rds_sock_put(rs);
151		rs->rs_bound_addr = 0;
152	}
153
154	spin_unlock_irqrestore(&rds_bind_lock, flags);
155}
156
157int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
158{
159	struct sock *sk = sock->sk;
160	struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
161	struct rds_sock *rs = rds_sk_to_rs(sk);
162	struct rds_transport *trans;
163	int ret = 0;
164
165	lock_sock(sk);
166
167	if (addr_len != sizeof(struct sockaddr_in) ||
168	    sin->sin_family != AF_INET ||
169	    rs->rs_bound_addr ||
170	    sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
171		ret = -EINVAL;
172		goto out;
173	}
174
175	ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
176	if (ret)
177		goto out;
178
179	trans = rds_trans_get_preferred(sin->sin_addr.s_addr);
180	if (trans == NULL) {
181		ret = -EADDRNOTAVAIL;
182		rds_remove_bound(rs);
183		if (printk_ratelimit())
184			printk(KERN_INFO "RDS: rds_bind() could not find a transport, "
185				"load rds_tcp or rds_rdma?\n");
186		goto out;
187	}
188
189	rs->rs_transport = trans;
190	ret = 0;
191
192out:
193	release_sock(sk);
194	return ret;
195}
196