blob: 5d95fc007f1aa6244d500561ccc0324063cda221 [file] [log] [blame]
Andy Grover639b3212009-02-24 15:30:18 +00001/*
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
39/*
40 * XXX this probably still needs more work.. no INADDR_ANY, and rbtrees aren't
41 * particularly zippy.
42 *
43 * This is now called for every incoming frame so we arguably care much more
44 * about it than we used to.
45 */
46static DEFINE_SPINLOCK(rds_bind_lock);
47static struct rb_root rds_bind_tree = RB_ROOT;
48
49static struct rds_sock *rds_bind_tree_walk(__be32 addr, __be16 port,
50 struct rds_sock *insert)
51{
52 struct rb_node **p = &rds_bind_tree.rb_node;
53 struct rb_node *parent = NULL;
54 struct rds_sock *rs;
55 u64 cmp;
56 u64 needle = ((u64)be32_to_cpu(addr) << 32) | be16_to_cpu(port);
57
58 while (*p) {
59 parent = *p;
60 rs = rb_entry(parent, struct rds_sock, rs_bound_node);
61
62 cmp = ((u64)be32_to_cpu(rs->rs_bound_addr) << 32) |
63 be16_to_cpu(rs->rs_bound_port);
64
65 if (needle < cmp)
66 p = &(*p)->rb_left;
67 else if (needle > cmp)
68 p = &(*p)->rb_right;
69 else
70 return rs;
71 }
72
73 if (insert) {
74 rb_link_node(&insert->rs_bound_node, parent, p);
75 rb_insert_color(&insert->rs_bound_node, &rds_bind_tree);
76 }
77 return NULL;
78}
79
80/*
81 * Return the rds_sock bound at the given local address.
82 *
83 * The rx path can race with rds_release. We notice if rds_release() has
84 * marked this socket and don't return a rs ref to the rx path.
85 */
86struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
87{
88 struct rds_sock *rs;
89 unsigned long flags;
90
91 spin_lock_irqsave(&rds_bind_lock, flags);
92 rs = rds_bind_tree_walk(addr, port, NULL);
93 if (rs && !sock_flag(rds_rs_to_sk(rs), SOCK_DEAD))
94 rds_sock_addref(rs);
95 else
96 rs = NULL;
97 spin_unlock_irqrestore(&rds_bind_lock, flags);
98
99 rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
100 ntohs(port));
101 return rs;
102}
103
104/* returns -ve errno or +ve port */
105static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
106{
107 unsigned long flags;
108 int ret = -EADDRINUSE;
109 u16 rover, last;
110
111 if (*port != 0) {
112 rover = be16_to_cpu(*port);
113 last = rover;
114 } else {
115 rover = max_t(u16, net_random(), 2);
116 last = rover - 1;
117 }
118
119 spin_lock_irqsave(&rds_bind_lock, flags);
120
121 do {
122 if (rover == 0)
123 rover++;
124 if (rds_bind_tree_walk(addr, cpu_to_be16(rover), rs) == NULL) {
125 *port = cpu_to_be16(rover);
126 ret = 0;
127 break;
128 }
129 } while (rover++ != last);
130
131 if (ret == 0) {
132 rs->rs_bound_addr = addr;
133 rs->rs_bound_port = *port;
134 rds_sock_addref(rs);
135
136 rdsdebug("rs %p binding to %pI4:%d\n",
137 rs, &addr, (int)ntohs(*port));
138 }
139
140 spin_unlock_irqrestore(&rds_bind_lock, flags);
141
142 return ret;
143}
144
145void rds_remove_bound(struct rds_sock *rs)
146{
147 unsigned long flags;
148
149 spin_lock_irqsave(&rds_bind_lock, flags);
150
151 if (rs->rs_bound_addr) {
152 rdsdebug("rs %p unbinding from %pI4:%d\n",
153 rs, &rs->rs_bound_addr,
154 ntohs(rs->rs_bound_port));
155
156 rb_erase(&rs->rs_bound_node, &rds_bind_tree);
157 rds_sock_put(rs);
158 rs->rs_bound_addr = 0;
159 }
160
161 spin_unlock_irqrestore(&rds_bind_lock, flags);
162}
163
164int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
165{
166 struct sock *sk = sock->sk;
167 struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
168 struct rds_sock *rs = rds_sk_to_rs(sk);
169 struct rds_transport *trans;
170 int ret = 0;
171
172 lock_sock(sk);
173
174 if (addr_len != sizeof(struct sockaddr_in) ||
175 sin->sin_family != AF_INET ||
176 rs->rs_bound_addr ||
177 sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
178 ret = -EINVAL;
179 goto out;
180 }
181
182 ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
183 if (ret)
184 goto out;
185
186 trans = rds_trans_get_preferred(sin->sin_addr.s_addr);
187 if (trans == NULL) {
188 ret = -EADDRNOTAVAIL;
189 rds_remove_bound(rs);
Andy Groverf2c44932009-08-21 12:28:35 +0000190 if (printk_ratelimit())
191 printk(KERN_INFO "RDS: rds_bind() could not find a transport, "
192 "load rds_tcp or rds_rdma?\n");
Andy Grover639b3212009-02-24 15:30:18 +0000193 goto out;
194 }
195
196 rs->rs_transport = trans;
197 ret = 0;
198
199out:
200 release_sock(sk);
201 return ret;
202}