blob: b5f96372ad7344b3b815b85a5e29a0b08ac8ad13 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * PF_INET6 socket protocol family
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003 * Linux INET6 implementation
Linus Torvalds1da177e2005-04-16 15:20:36 -07004 *
5 * Authors:
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09006 * Pedro Roque <roque@di.fc.ul.pt>
Linus Torvalds1da177e2005-04-16 15:20:36 -07007 *
8 * Adapted from linux/net/ipv4/af_inet.c
9 *
10 * $Id: af_inet6.c,v 1.66 2002/02/01 22:01:04 davem Exp $
11 *
12 * Fixes:
13 * piggy, Karl Knutson : Socket protocol table
14 * Hideaki YOSHIFUJI : sin6_scope_id support
15 * Arnaldo Melo : check proc_net_create return, cleanups
16 *
17 * This program is free software; you can redistribute it and/or
18 * modify it under the terms of the GNU General Public License
19 * as published by the Free Software Foundation; either version
20 * 2 of the License, or (at your option) any later version.
21 */
22
23
24#include <linux/module.h>
Randy Dunlap4fc268d2006-01-11 12:17:47 -080025#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070026#include <linux/errno.h>
27#include <linux/types.h>
28#include <linux/socket.h>
29#include <linux/in.h>
30#include <linux/kernel.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070031#include <linux/timer.h>
32#include <linux/string.h>
33#include <linux/sockios.h>
34#include <linux/net.h>
35#include <linux/fcntl.h>
36#include <linux/mm.h>
37#include <linux/interrupt.h>
38#include <linux/proc_fs.h>
39#include <linux/stat.h>
40#include <linux/init.h>
41
42#include <linux/inet.h>
43#include <linux/netdevice.h>
44#include <linux/icmpv6.h>
Harald Welte2cc7d572005-08-09 19:42:34 -070045#include <linux/netfilter_ipv6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070046
47#include <net/ip.h>
48#include <net/ipv6.h>
49#include <net/udp.h>
Gerrit Renkerba4e58e2006-11-27 11:10:57 -080050#include <net/udplite.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070051#include <net/tcp.h>
52#include <net/ipip.h>
53#include <net/protocol.h>
54#include <net/inet_common.h>
55#include <net/transp_v6.h>
56#include <net/ip6_route.h>
57#include <net/addrconf.h>
58#ifdef CONFIG_IPV6_TUNNEL
59#include <net/ip6_tunnel.h>
60#endif
61
62#include <asm/uaccess.h>
63#include <asm/system.h>
64
65MODULE_AUTHOR("Cast of dozens");
66MODULE_DESCRIPTION("IPv6 protocol stack for Linux");
67MODULE_LICENSE("GPL");
68
Brian Haleyab32ea52006-09-22 14:15:41 -070069int sysctl_ipv6_bindv6only __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -070070
Linus Torvalds1da177e2005-04-16 15:20:36 -070071/* The inetsw table contains everything that inet_create needs to
72 * build a new socket.
73 */
74static struct list_head inetsw6[SOCK_MAX];
75static DEFINE_SPINLOCK(inetsw6_lock);
76
Linus Torvalds1da177e2005-04-16 15:20:36 -070077static __inline__ struct ipv6_pinfo *inet6_sk_generic(struct sock *sk)
78{
79 const int offset = sk->sk_prot->obj_size - sizeof(struct ipv6_pinfo);
80
81 return (struct ipv6_pinfo *)(((u8 *)sk) + offset);
82}
83
84static int inet6_create(struct socket *sock, int protocol)
85{
86 struct inet_sock *inet;
87 struct ipv6_pinfo *np;
88 struct sock *sk;
89 struct list_head *p;
90 struct inet_protosw *answer;
91 struct proto *answer_prot;
92 unsigned char answer_flags;
93 char answer_no_check;
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -080094 int try_loading_module = 0;
95 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -070096
David S. Millerb3da2cf2007-03-23 11:40:27 -070097 if (sock->type != SOCK_RAW &&
98 sock->type != SOCK_DGRAM &&
99 !inet_ehash_secret)
100 build_ehash_secret();
101
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102 /* Look for the requested type/protocol pair. */
103 answer = NULL;
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800104lookup_protocol:
105 err = -ESOCKTNOSUPPORT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106 rcu_read_lock();
107 list_for_each_rcu(p, &inetsw6[sock->type]) {
108 answer = list_entry(p, struct inet_protosw, list);
109
110 /* Check the non-wild match. */
111 if (protocol == answer->protocol) {
112 if (protocol != IPPROTO_IP)
113 break;
114 } else {
115 /* Check for the two wild cases. */
116 if (IPPROTO_IP == protocol) {
117 protocol = answer->protocol;
118 break;
119 }
120 if (IPPROTO_IP == answer->protocol)
121 break;
122 }
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800123 err = -EPROTONOSUPPORT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700124 answer = NULL;
125 }
126
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800127 if (!answer) {
128 if (try_loading_module < 2) {
129 rcu_read_unlock();
130 /*
131 * Be more specific, e.g. net-pf-10-proto-132-type-1
132 * (net-pf-PF_INET6-proto-IPPROTO_SCTP-type-SOCK_STREAM)
133 */
134 if (++try_loading_module == 1)
135 request_module("net-pf-%d-proto-%d-type-%d",
136 PF_INET6, protocol, sock->type);
137 /*
138 * Fall back to generic, e.g. net-pf-10-proto-132
139 * (net-pf-PF_INET6-proto-IPPROTO_SCTP)
140 */
141 else
142 request_module("net-pf-%d-proto-%d",
143 PF_INET6, protocol);
144 goto lookup_protocol;
145 } else
146 goto out_rcu_unlock;
147 }
148
149 err = -EPERM;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150 if (answer->capability > 0 && !capable(answer->capability))
151 goto out_rcu_unlock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700152
153 sock->ops = answer->ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154 answer_prot = answer->prot;
155 answer_no_check = answer->no_check;
156 answer_flags = answer->flags;
157 rcu_read_unlock();
158
159 BUG_TRAP(answer_prot->slab != NULL);
160
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800161 err = -ENOBUFS;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700162 sk = sk_alloc(PF_INET6, GFP_KERNEL, answer_prot, 1);
163 if (sk == NULL)
164 goto out;
165
166 sock_init_data(sock, sk);
167
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800168 err = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169 sk->sk_no_check = answer_no_check;
170 if (INET_PROTOSW_REUSE & answer_flags)
171 sk->sk_reuse = 1;
172
173 inet = inet_sk(sk);
Paul Moore469de9b2007-01-09 14:37:06 -0800174 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700175
176 if (SOCK_RAW == sock->type) {
177 inet->num = protocol;
178 if (IPPROTO_RAW == protocol)
179 inet->hdrincl = 1;
180 }
181
Arnaldo Carvalho de Meloe6848972005-08-09 19:45:38 -0700182 sk->sk_destruct = inet_sock_destruct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183 sk->sk_family = PF_INET6;
184 sk->sk_protocol = protocol;
185
186 sk->sk_backlog_rcv = answer->prot->backlog_rcv;
187
188 inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk);
189 np->hop_limit = -1;
190 np->mcast_hops = -1;
191 np->mc_loop = 1;
192 np->pmtudisc = IPV6_PMTUDISC_WANT;
193 np->ipv6only = sysctl_ipv6_bindv6only;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900194
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195 /* Init the ipv4 part of the socket since we can have sockets
196 * using v6 API for ipv4.
197 */
198 inet->uc_ttl = -1;
199
200 inet->mc_loop = 1;
201 inet->mc_ttl = 1;
202 inet->mc_index = 0;
203 inet->mc_list = NULL;
204
205 if (ipv4_config.no_pmtu_disc)
206 inet->pmtudisc = IP_PMTUDISC_DONT;
207 else
208 inet->pmtudisc = IP_PMTUDISC_WANT;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900209 /*
Arnaldo Carvalho de Meloe6848972005-08-09 19:45:38 -0700210 * Increment only the relevant sk_prot->socks debug field, this changes
211 * the previous behaviour of incrementing both the equivalent to
212 * answer->prot->socks (inet6_sock_nr) and inet_sock_nr.
213 *
214 * This allows better debug granularity as we'll know exactly how many
215 * UDPv6, TCPv6, etc socks were allocated, not the sum of all IPv6
216 * transport protocol socks. -acme
217 */
218 sk_refcnt_debug_inc(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 if (inet->num) {
221 /* It assumes that any protocol which allows
222 * the user to assign a number at socket
223 * creation time automatically shares.
224 */
Al Viroe69a4ad2006-11-14 20:56:00 -0800225 inet->sport = htons(inet->num);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226 sk->sk_prot->hash(sk);
227 }
228 if (sk->sk_prot->init) {
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800229 err = sk->sk_prot->init(sk);
230 if (err) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700231 sk_common_release(sk);
232 goto out;
233 }
234 }
235out:
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800236 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237out_rcu_unlock:
238 rcu_read_unlock();
239 goto out;
240}
241
242
243/* bind for INET6 API */
244int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
245{
246 struct sockaddr_in6 *addr=(struct sockaddr_in6 *)uaddr;
247 struct sock *sk = sock->sk;
248 struct inet_sock *inet = inet_sk(sk);
249 struct ipv6_pinfo *np = inet6_sk(sk);
Al Virofd683222006-09-26 22:17:51 -0700250 __be32 v4addr = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700251 unsigned short snum;
252 int addr_type = 0;
253 int err = 0;
254
255 /* If the socket has its own bind function then use it. */
256 if (sk->sk_prot->bind)
257 return sk->sk_prot->bind(sk, uaddr, addr_len);
258
259 if (addr_len < SIN6_LEN_RFC2133)
260 return -EINVAL;
261 addr_type = ipv6_addr_type(&addr->sin6_addr);
262 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
263 return -EINVAL;
264
265 snum = ntohs(addr->sin6_port);
266 if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
267 return -EACCES;
268
269 lock_sock(sk);
270
271 /* Check these errors (active socket, double bind). */
272 if (sk->sk_state != TCP_CLOSE || inet->num) {
273 err = -EINVAL;
274 goto out;
275 }
276
277 /* Check if the address belongs to the host. */
278 if (addr_type == IPV6_ADDR_MAPPED) {
279 v4addr = addr->sin6_addr.s6_addr32[3];
280 if (inet_addr_type(v4addr) != RTN_LOCAL) {
281 err = -EADDRNOTAVAIL;
282 goto out;
283 }
284 } else {
285 if (addr_type != IPV6_ADDR_ANY) {
286 struct net_device *dev = NULL;
287
288 if (addr_type & IPV6_ADDR_LINKLOCAL) {
289 if (addr_len >= sizeof(struct sockaddr_in6) &&
290 addr->sin6_scope_id) {
291 /* Override any existing binding, if another one
292 * is supplied by user.
293 */
294 sk->sk_bound_dev_if = addr->sin6_scope_id;
295 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900296
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297 /* Binding to link-local address requires an interface */
298 if (!sk->sk_bound_dev_if) {
299 err = -EINVAL;
300 goto out;
301 }
302 dev = dev_get_by_index(sk->sk_bound_dev_if);
303 if (!dev) {
304 err = -ENODEV;
305 goto out;
306 }
307 }
308
309 /* ipv4 addr of the socket is invalid. Only the
310 * unspecified and mapped address have a v4 equivalent.
311 */
312 v4addr = LOOPBACK4_IPV6;
313 if (!(addr_type & IPV6_ADDR_MULTICAST)) {
314 if (!ipv6_chk_addr(&addr->sin6_addr, dev, 0)) {
315 if (dev)
316 dev_put(dev);
317 err = -EADDRNOTAVAIL;
318 goto out;
319 }
320 }
321 if (dev)
322 dev_put(dev);
323 }
324 }
325
326 inet->rcv_saddr = v4addr;
327 inet->saddr = v4addr;
328
329 ipv6_addr_copy(&np->rcv_saddr, &addr->sin6_addr);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900330
Linus Torvalds1da177e2005-04-16 15:20:36 -0700331 if (!(addr_type & IPV6_ADDR_MULTICAST))
332 ipv6_addr_copy(&np->saddr, &addr->sin6_addr);
333
334 /* Make sure we are allowed to bind here. */
335 if (sk->sk_prot->get_port(sk, snum)) {
336 inet_reset_saddr(sk);
337 err = -EADDRINUSE;
338 goto out;
339 }
340
341 if (addr_type != IPV6_ADDR_ANY)
342 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
343 if (snum)
344 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
Al Viroe69a4ad2006-11-14 20:56:00 -0800345 inet->sport = htons(inet->num);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700346 inet->dport = 0;
347 inet->daddr = 0;
348out:
349 release_sock(sk);
350 return err;
351}
352
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900353EXPORT_SYMBOL(inet6_bind);
354
Linus Torvalds1da177e2005-04-16 15:20:36 -0700355int inet6_release(struct socket *sock)
356{
357 struct sock *sk = sock->sk;
358
359 if (sk == NULL)
360 return -EINVAL;
361
362 /* Free mc lists */
363 ipv6_sock_mc_close(sk);
364
365 /* Free ac lists */
366 ipv6_sock_ac_close(sk);
367
368 return inet_release(sock);
369}
370
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900371EXPORT_SYMBOL(inet6_release);
372
Linus Torvalds1da177e2005-04-16 15:20:36 -0700373int inet6_destroy_sock(struct sock *sk)
374{
375 struct ipv6_pinfo *np = inet6_sk(sk);
376 struct sk_buff *skb;
377 struct ipv6_txoptions *opt;
378
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379 /* Release rx options */
380
381 if ((skb = xchg(&np->pktoptions, NULL)) != NULL)
382 kfree_skb(skb);
383
384 /* Free flowlabels */
385 fl6_free_socklist(sk);
386
387 /* Free tx options */
388
389 if ((opt = xchg(&np->opt, NULL)) != NULL)
390 sock_kfree_s(sk, opt, opt->tot_len);
391
392 return 0;
393}
394
Arnaldo Carvalho de Melo3cf3dc62005-12-13 23:23:20 -0800395EXPORT_SYMBOL_GPL(inet6_destroy_sock);
396
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397/*
398 * This does both peername and sockname.
399 */
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900400
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
402 int *uaddr_len, int peer)
403{
404 struct sockaddr_in6 *sin=(struct sockaddr_in6 *)uaddr;
405 struct sock *sk = sock->sk;
406 struct inet_sock *inet = inet_sk(sk);
407 struct ipv6_pinfo *np = inet6_sk(sk);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900408
Linus Torvalds1da177e2005-04-16 15:20:36 -0700409 sin->sin6_family = AF_INET6;
410 sin->sin6_flowinfo = 0;
411 sin->sin6_scope_id = 0;
412 if (peer) {
413 if (!inet->dport)
414 return -ENOTCONN;
415 if (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) &&
416 peer == 1)
417 return -ENOTCONN;
418 sin->sin6_port = inet->dport;
419 ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
420 if (np->sndflow)
421 sin->sin6_flowinfo = np->flow_label;
422 } else {
423 if (ipv6_addr_any(&np->rcv_saddr))
424 ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
425 else
426 ipv6_addr_copy(&sin->sin6_addr, &np->rcv_saddr);
427
428 sin->sin6_port = inet->sport;
429 }
430 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL)
431 sin->sin6_scope_id = sk->sk_bound_dev_if;
432 *uaddr_len = sizeof(*sin);
433 return(0);
434}
435
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900436EXPORT_SYMBOL(inet6_getname);
437
Linus Torvalds1da177e2005-04-16 15:20:36 -0700438int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
439{
440 struct sock *sk = sock->sk;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900442 switch(cmd)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700443 {
444 case SIOCGSTAMP:
445 return sock_get_timestamp(sk, (struct timeval __user *)arg);
446
Eric Dumazetae40eb12007-03-18 17:33:16 -0700447 case SIOCGSTAMPNS:
448 return sock_get_timestampns(sk, (struct timespec __user *)arg);
449
Linus Torvalds1da177e2005-04-16 15:20:36 -0700450 case SIOCADDRT:
451 case SIOCDELRT:
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900452
Linus Torvalds1da177e2005-04-16 15:20:36 -0700453 return(ipv6_route_ioctl(cmd,(void __user *)arg));
454
455 case SIOCSIFADDR:
456 return addrconf_add_ifaddr((void __user *) arg);
457 case SIOCDIFADDR:
458 return addrconf_del_ifaddr((void __user *) arg);
459 case SIOCSIFDSTADDR:
460 return addrconf_set_dstaddr((void __user *) arg);
461 default:
Christoph Hellwigb5e5fa52006-01-03 14:18:33 -0800462 if (!sk->sk_prot->ioctl)
463 return -ENOIOCTLCMD;
464 return sk->sk_prot->ioctl(sk, cmd, arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465 }
466 /*NOTREACHED*/
467 return(0);
468}
469
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900470EXPORT_SYMBOL(inet6_ioctl);
471
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800472const struct proto_ops inet6_stream_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800473 .family = PF_INET6,
474 .owner = THIS_MODULE,
475 .release = inet6_release,
476 .bind = inet6_bind,
477 .connect = inet_stream_connect, /* ok */
478 .socketpair = sock_no_socketpair, /* a do nothing */
479 .accept = inet_accept, /* ok */
480 .getname = inet6_getname,
481 .poll = tcp_poll, /* ok */
482 .ioctl = inet6_ioctl, /* must change */
483 .listen = inet_listen, /* ok */
484 .shutdown = inet_shutdown, /* ok */
485 .setsockopt = sock_common_setsockopt, /* ok */
486 .getsockopt = sock_common_getsockopt, /* ok */
David S. Miller3516ffb2007-08-02 19:23:56 -0700487 .sendmsg = tcp_sendmsg, /* ok */
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800488 .recvmsg = sock_common_recvmsg, /* ok */
489 .mmap = sock_no_mmap,
490 .sendpage = tcp_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800491#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800492 .compat_setsockopt = compat_sock_common_setsockopt,
493 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800494#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700495};
496
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800497const struct proto_ops inet6_dgram_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800498 .family = PF_INET6,
499 .owner = THIS_MODULE,
500 .release = inet6_release,
501 .bind = inet6_bind,
502 .connect = inet_dgram_connect, /* ok */
503 .socketpair = sock_no_socketpair, /* a do nothing */
504 .accept = sock_no_accept, /* a do nothing */
505 .getname = inet6_getname,
506 .poll = udp_poll, /* ok */
507 .ioctl = inet6_ioctl, /* must change */
508 .listen = sock_no_listen, /* ok */
509 .shutdown = inet_shutdown, /* ok */
510 .setsockopt = sock_common_setsockopt, /* ok */
511 .getsockopt = sock_common_getsockopt, /* ok */
512 .sendmsg = inet_sendmsg, /* ok */
513 .recvmsg = sock_common_recvmsg, /* ok */
514 .mmap = sock_no_mmap,
515 .sendpage = sock_no_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800516#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800517 .compat_setsockopt = compat_sock_common_setsockopt,
518 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800519#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520};
521
522static struct net_proto_family inet6_family_ops = {
523 .family = PF_INET6,
524 .create = inet6_create,
525 .owner = THIS_MODULE,
526};
527
Linus Torvalds1da177e2005-04-16 15:20:36 -0700528/* Same as inet6_dgram_ops, sans udp_poll. */
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800529static const struct proto_ops inet6_sockraw_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800530 .family = PF_INET6,
531 .owner = THIS_MODULE,
532 .release = inet6_release,
533 .bind = inet6_bind,
534 .connect = inet_dgram_connect, /* ok */
535 .socketpair = sock_no_socketpair, /* a do nothing */
536 .accept = sock_no_accept, /* a do nothing */
537 .getname = inet6_getname,
538 .poll = datagram_poll, /* ok */
539 .ioctl = inet6_ioctl, /* must change */
540 .listen = sock_no_listen, /* ok */
541 .shutdown = inet_shutdown, /* ok */
542 .setsockopt = sock_common_setsockopt, /* ok */
543 .getsockopt = sock_common_getsockopt, /* ok */
544 .sendmsg = inet_sendmsg, /* ok */
545 .recvmsg = sock_common_recvmsg, /* ok */
546 .mmap = sock_no_mmap,
547 .sendpage = sock_no_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800548#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800549 .compat_setsockopt = compat_sock_common_setsockopt,
550 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800551#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700552};
553
554static struct inet_protosw rawv6_protosw = {
555 .type = SOCK_RAW,
556 .protocol = IPPROTO_IP, /* wild card */
557 .prot = &rawv6_prot,
558 .ops = &inet6_sockraw_ops,
559 .capability = CAP_NET_RAW,
560 .no_check = UDP_CSUM_DEFAULT,
561 .flags = INET_PROTOSW_REUSE,
562};
563
564void
565inet6_register_protosw(struct inet_protosw *p)
566{
567 struct list_head *lh;
568 struct inet_protosw *answer;
569 int protocol = p->protocol;
570 struct list_head *last_perm;
571
572 spin_lock_bh(&inetsw6_lock);
573
574 if (p->type >= SOCK_MAX)
575 goto out_illegal;
576
577 /* If we are trying to override a permanent protocol, bail. */
578 answer = NULL;
579 last_perm = &inetsw6[p->type];
580 list_for_each(lh, &inetsw6[p->type]) {
581 answer = list_entry(lh, struct inet_protosw, list);
582
583 /* Check only the non-wild match. */
584 if (INET_PROTOSW_PERMANENT & answer->flags) {
585 if (protocol == answer->protocol)
586 break;
587 last_perm = lh;
588 }
589
590 answer = NULL;
591 }
592 if (answer)
593 goto out_permanent;
594
595 /* Add the new entry after the last permanent entry if any, so that
596 * the new entry does not override a permanent entry when matched with
597 * a wild-card protocol. But it is allowed to override any existing
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900598 * non-permanent entry. This means that when we remove this entry, the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700599 * system automatically returns to the old behavior.
600 */
601 list_add_rcu(&p->list, last_perm);
602out:
603 spin_unlock_bh(&inetsw6_lock);
604 return;
605
606out_permanent:
607 printk(KERN_ERR "Attempt to override permanent protocol %d.\n",
608 protocol);
609 goto out;
610
611out_illegal:
612 printk(KERN_ERR
613 "Ignoring attempt to register invalid socket type %d.\n",
614 p->type);
615 goto out;
616}
617
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900618EXPORT_SYMBOL(inet6_register_protosw);
619
Linus Torvalds1da177e2005-04-16 15:20:36 -0700620void
621inet6_unregister_protosw(struct inet_protosw *p)
622{
623 if (INET_PROTOSW_PERMANENT & p->flags) {
624 printk(KERN_ERR
625 "Attempt to unregister permanent protocol %d.\n",
626 p->protocol);
627 } else {
628 spin_lock_bh(&inetsw6_lock);
629 list_del_rcu(&p->list);
630 spin_unlock_bh(&inetsw6_lock);
631
632 synchronize_net();
633 }
634}
635
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900636EXPORT_SYMBOL(inet6_unregister_protosw);
637
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800638int inet6_sk_rebuild_header(struct sock *sk)
639{
640 int err;
641 struct dst_entry *dst;
642 struct ipv6_pinfo *np = inet6_sk(sk);
643
644 dst = __sk_dst_check(sk, np->dst_cookie);
645
646 if (dst == NULL) {
647 struct inet_sock *inet = inet_sk(sk);
648 struct in6_addr *final_p = NULL, final;
649 struct flowi fl;
650
651 memset(&fl, 0, sizeof(fl));
652 fl.proto = sk->sk_protocol;
653 ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
654 ipv6_addr_copy(&fl.fl6_src, &np->saddr);
655 fl.fl6_flowlabel = np->flow_label;
656 fl.oif = sk->sk_bound_dev_if;
657 fl.fl_ip_dport = inet->dport;
658 fl.fl_ip_sport = inet->sport;
Venkat Yekkiralabeb8d132006-08-04 23:12:42 -0700659 security_sk_classify_flow(sk, &fl);
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800660
661 if (np->opt && np->opt->srcrt) {
662 struct rt0_hdr *rt0 = (struct rt0_hdr *) np->opt->srcrt;
663 ipv6_addr_copy(&final, &fl.fl6_dst);
664 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
665 final_p = &final;
666 }
667
668 err = ip6_dst_lookup(sk, &dst, &fl);
669 if (err) {
670 sk->sk_route_caps = 0;
671 return err;
672 }
673 if (final_p)
674 ipv6_addr_copy(&fl.fl6_dst, final_p);
675
676 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) {
677 sk->sk_err_soft = -err;
678 return err;
679 }
680
YOSHIFUJI Hideaki8e1ef0a2006-08-29 17:15:09 -0700681 __ip6_dst_store(sk, dst, NULL, NULL);
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800682 }
683
684 return 0;
685}
686
687EXPORT_SYMBOL_GPL(inet6_sk_rebuild_header);
688
Arnaldo Carvalho de Melo399c07d2005-12-13 23:24:28 -0800689int ipv6_opt_accepted(struct sock *sk, struct sk_buff *skb)
690{
691 struct ipv6_pinfo *np = inet6_sk(sk);
692 struct inet6_skb_parm *opt = IP6CB(skb);
693
694 if (np->rxopt.all) {
695 if ((opt->hop && (np->rxopt.bits.hopopts ||
696 np->rxopt.bits.ohopopts)) ||
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700697 ((IPV6_FLOWINFO_MASK &
698 *(__be32 *)skb_network_header(skb)) &&
Arnaldo Carvalho de Melo399c07d2005-12-13 23:24:28 -0800699 np->rxopt.bits.rxflow) ||
700 (opt->srcrt && (np->rxopt.bits.srcrt ||
701 np->rxopt.bits.osrcrt)) ||
702 ((opt->dst1 || opt->dst0) &&
703 (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts)))
704 return 1;
705 }
706 return 0;
707}
708
709EXPORT_SYMBOL_GPL(ipv6_opt_accepted);
710
Linus Torvalds1da177e2005-04-16 15:20:36 -0700711static int __init init_ipv6_mibs(void)
712{
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700713 if (snmp_mib_init((void **)ipv6_statistics, sizeof (struct ipstats_mib),
714 __alignof__(struct ipstats_mib)) < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715 goto err_ip_mib;
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700716 if (snmp_mib_init((void **)icmpv6_statistics, sizeof (struct icmpv6_mib),
717 __alignof__(struct icmpv6_mib)) < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718 goto err_icmp_mib;
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700719 if (snmp_mib_init((void **)udp_stats_in6, sizeof (struct udp_mib),
720 __alignof__(struct udp_mib)) < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700721 goto err_udp_mib;
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700722 if (snmp_mib_init((void **)udplite_stats_in6, sizeof (struct udp_mib),
723 __alignof__(struct udp_mib)) < 0)
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800724 goto err_udplite_mib;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700725 return 0;
726
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800727err_udplite_mib:
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700728 snmp_mib_free((void **)udp_stats_in6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700729err_udp_mib:
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700730 snmp_mib_free((void **)icmpv6_statistics);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731err_icmp_mib:
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700732 snmp_mib_free((void **)ipv6_statistics);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700733err_ip_mib:
734 return -ENOMEM;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900735
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736}
737
738static void cleanup_ipv6_mibs(void)
739{
Herbert Xu7f7d9a62007-04-24 21:54:09 -0700740 snmp_mib_free((void **)ipv6_statistics);
741 snmp_mib_free((void **)icmpv6_statistics);
742 snmp_mib_free((void **)udp_stats_in6);
743 snmp_mib_free((void **)udplite_stats_in6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700744}
745
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746static int __init inet6_init(void)
747{
748 struct sk_buff *dummy_skb;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900749 struct list_head *r;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700750 int err;
751
YOSHIFUJI Hideakief047f52006-09-01 00:29:06 -0700752 BUILD_BUG_ON(sizeof(struct inet6_skb_parm) > sizeof(dummy_skb->cb));
753
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754#ifdef MODULE
755#if 0 /* FIXME --RR */
756 if (!mod_member_present(&__this_module, can_unload))
757 return -EINVAL;
758
759 __this_module.can_unload = &ipv6_unload;
760#endif
761#endif
762
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763 err = proto_register(&tcpv6_prot, 1);
764 if (err)
765 goto out;
766
767 err = proto_register(&udpv6_prot, 1);
768 if (err)
769 goto out_unregister_tcp_proto;
770
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800771 err = proto_register(&udplitev6_prot, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700772 if (err)
773 goto out_unregister_udp_proto;
774
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800775 err = proto_register(&rawv6_prot, 1);
776 if (err)
777 goto out_unregister_udplite_proto;
778
Linus Torvalds1da177e2005-04-16 15:20:36 -0700779
780 /* Register the socket-side information for inet6_create. */
781 for(r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r)
782 INIT_LIST_HEAD(r);
783
784 /* We MUST register RAW sockets before we create the ICMP6,
785 * IGMP6, or NDISC control sockets.
786 */
787 inet6_register_protosw(&rawv6_protosw);
788
789 /* Register the family here so that the init calls below will
790 * be able to create sockets. (?? is this dangerous ??)
791 */
David S. Miller8eb55912005-11-11 15:05:47 -0800792 err = sock_register(&inet6_family_ops);
793 if (err)
794 goto out_unregister_raw_proto;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795
796 /* Initialise ipv6 mibs */
797 err = init_ipv6_mibs();
798 if (err)
David S. Miller8eb55912005-11-11 15:05:47 -0800799 goto out_unregister_sock;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900800
Linus Torvalds1da177e2005-04-16 15:20:36 -0700801 /*
802 * ipngwg API draft makes clear that the correct semantics
803 * for TCP and UDP is to consider one TCP and UDP instance
804 * in a host availiable by both INET and INET6 APIs and
805 * able to communicate via both network protocols.
806 */
807
808#ifdef CONFIG_SYSCTL
809 ipv6_sysctl_register();
810#endif
811 err = icmpv6_init(&inet6_family_ops);
812 if (err)
813 goto icmp_fail;
814 err = ndisc_init(&inet6_family_ops);
815 if (err)
816 goto ndisc_fail;
817 err = igmp6_init(&inet6_family_ops);
818 if (err)
819 goto igmp_fail;
Harald Welte2cc7d572005-08-09 19:42:34 -0700820 err = ipv6_netfilter_init();
821 if (err)
822 goto netfilter_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700823 /* Create /proc/foo6 entries. */
824#ifdef CONFIG_PROC_FS
825 err = -ENOMEM;
826 if (raw6_proc_init())
827 goto proc_raw6_fail;
828 if (tcp6_proc_init())
829 goto proc_tcp6_fail;
830 if (udp6_proc_init())
831 goto proc_udp6_fail;
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800832 if (udplite6_proc_init())
833 goto proc_udplite6_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700834 if (ipv6_misc_proc_init())
835 goto proc_misc6_fail;
836
837 if (ac6_proc_init())
838 goto proc_anycast6_fail;
839 if (if6_proc_init())
840 goto proc_if6_fail;
841#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700842 ip6_route_init();
843 ip6_flowlabel_init();
844 err = addrconf_init();
845 if (err)
846 goto addrconf_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700847
848 /* Init v6 extension headers. */
849 ipv6_rthdr_init();
850 ipv6_frag_init();
851 ipv6_nodata_init();
852 ipv6_destopt_init();
853
854 /* Init v6 transport protocols. */
855 udpv6_init();
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800856 udplitev6_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857 tcpv6_init();
Herbert Xue2ed4052005-07-05 14:41:20 -0700858
859 ipv6_packet_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700860 err = 0;
861out:
862 return err;
863
864addrconf_fail:
865 ip6_flowlabel_cleanup();
866 ip6_route_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700867#ifdef CONFIG_PROC_FS
868 if6_proc_exit();
869proc_if6_fail:
870 ac6_proc_exit();
871proc_anycast6_fail:
872 ipv6_misc_proc_exit();
873proc_misc6_fail:
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800874 udplite6_proc_exit();
875proc_udplite6_fail:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876 udp6_proc_exit();
877proc_udp6_fail:
878 tcp6_proc_exit();
879proc_tcp6_fail:
880 raw6_proc_exit();
881proc_raw6_fail:
882#endif
Harald Welte2cc7d572005-08-09 19:42:34 -0700883 ipv6_netfilter_fini();
884netfilter_fail:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700885 igmp6_cleanup();
886igmp_fail:
887 ndisc_cleanup();
888ndisc_fail:
889 icmpv6_cleanup();
890icmp_fail:
891#ifdef CONFIG_SYSCTL
892 ipv6_sysctl_unregister();
893#endif
894 cleanup_ipv6_mibs();
David S. Miller8eb55912005-11-11 15:05:47 -0800895out_unregister_sock:
896 sock_unregister(PF_INET6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897out_unregister_raw_proto:
898 proto_unregister(&rawv6_prot);
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800899out_unregister_udplite_proto:
900 proto_unregister(&udplitev6_prot);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901out_unregister_udp_proto:
902 proto_unregister(&udpv6_prot);
903out_unregister_tcp_proto:
904 proto_unregister(&tcpv6_prot);
905 goto out;
906}
907module_init(inet6_init);
908
909static void __exit inet6_exit(void)
910{
911 /* First of all disallow new sockets creation. */
912 sock_unregister(PF_INET6);
Thomas Grafc127ea22007-03-22 11:58:32 -0700913 /* Disallow any further netlink messages */
914 rtnl_unregister_all(PF_INET6);
Joe Jinca17c232007-02-20 01:30:15 -0800915
916 /* Cleanup code parts. */
917 ipv6_packet_cleanup();
Masahide NAKAMURA59fbb3a2007-06-26 23:56:32 -0700918
Linus Torvalds1da177e2005-04-16 15:20:36 -0700919 addrconf_cleanup();
Joe Jinca17c232007-02-20 01:30:15 -0800920 ip6_flowlabel_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700921 ip6_route_cleanup();
Joe Jinca17c232007-02-20 01:30:15 -0800922#ifdef CONFIG_PROC_FS
923
924 /* Cleanup code parts. */
925 if6_proc_exit();
926 ac6_proc_exit();
927 ipv6_misc_proc_exit();
928 udplite6_proc_exit();
929 udp6_proc_exit();
930 tcp6_proc_exit();
931 raw6_proc_exit();
932#endif
Harald Welte2cc7d572005-08-09 19:42:34 -0700933 ipv6_netfilter_fini();
Joe Jinca17c232007-02-20 01:30:15 -0800934 igmp6_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700935 ndisc_cleanup();
936 icmpv6_cleanup();
937#ifdef CONFIG_SYSCTL
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900938 ipv6_sysctl_unregister();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700939#endif
940 cleanup_ipv6_mibs();
941 proto_unregister(&rawv6_prot);
Joe Jinca17c232007-02-20 01:30:15 -0800942 proto_unregister(&udplitev6_prot);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700943 proto_unregister(&udpv6_prot);
944 proto_unregister(&tcpv6_prot);
945}
946module_exit(inet6_exit);
947
948MODULE_ALIAS_NETPROTO(PF_INET6);