blob: 82572b507547e6bfff720a5e0d6dc8fdbe7c847a [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>
45#include <linux/smp_lock.h>
Harald Welte2cc7d572005-08-09 19:42:34 -070046#include <linux/netfilter_ipv6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070047
48#include <net/ip.h>
49#include <net/ipv6.h>
50#include <net/udp.h>
Gerrit Renkerba4e58e2006-11-27 11:10:57 -080051#include <net/udplite.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070052#include <net/tcp.h>
53#include <net/ipip.h>
54#include <net/protocol.h>
55#include <net/inet_common.h>
56#include <net/transp_v6.h>
57#include <net/ip6_route.h>
58#include <net/addrconf.h>
59#ifdef CONFIG_IPV6_TUNNEL
60#include <net/ip6_tunnel.h>
61#endif
Noriaki TAKAMIYA2c8d7ca2006-08-23 20:31:11 -070062#ifdef CONFIG_IPV6_MIP6
63#include <net/mip6.h>
64#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -070065
66#include <asm/uaccess.h>
67#include <asm/system.h>
68
69MODULE_AUTHOR("Cast of dozens");
70MODULE_DESCRIPTION("IPv6 protocol stack for Linux");
71MODULE_LICENSE("GPL");
72
Brian Haleyab32ea52006-09-22 14:15:41 -070073int sysctl_ipv6_bindv6only __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -070074
Linus Torvalds1da177e2005-04-16 15:20:36 -070075/* The inetsw table contains everything that inet_create needs to
76 * build a new socket.
77 */
78static struct list_head inetsw6[SOCK_MAX];
79static DEFINE_SPINLOCK(inetsw6_lock);
80
Linus Torvalds1da177e2005-04-16 15:20:36 -070081static __inline__ struct ipv6_pinfo *inet6_sk_generic(struct sock *sk)
82{
83 const int offset = sk->sk_prot->obj_size - sizeof(struct ipv6_pinfo);
84
85 return (struct ipv6_pinfo *)(((u8 *)sk) + offset);
86}
87
88static int inet6_create(struct socket *sock, int protocol)
89{
90 struct inet_sock *inet;
91 struct ipv6_pinfo *np;
92 struct sock *sk;
93 struct list_head *p;
94 struct inet_protosw *answer;
95 struct proto *answer_prot;
96 unsigned char answer_flags;
97 char answer_no_check;
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -080098 int try_loading_module = 0;
99 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100
101 /* Look for the requested type/protocol pair. */
102 answer = NULL;
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800103lookup_protocol:
104 err = -ESOCKTNOSUPPORT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105 rcu_read_lock();
106 list_for_each_rcu(p, &inetsw6[sock->type]) {
107 answer = list_entry(p, struct inet_protosw, list);
108
109 /* Check the non-wild match. */
110 if (protocol == answer->protocol) {
111 if (protocol != IPPROTO_IP)
112 break;
113 } else {
114 /* Check for the two wild cases. */
115 if (IPPROTO_IP == protocol) {
116 protocol = answer->protocol;
117 break;
118 }
119 if (IPPROTO_IP == answer->protocol)
120 break;
121 }
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800122 err = -EPROTONOSUPPORT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123 answer = NULL;
124 }
125
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800126 if (!answer) {
127 if (try_loading_module < 2) {
128 rcu_read_unlock();
129 /*
130 * Be more specific, e.g. net-pf-10-proto-132-type-1
131 * (net-pf-PF_INET6-proto-IPPROTO_SCTP-type-SOCK_STREAM)
132 */
133 if (++try_loading_module == 1)
134 request_module("net-pf-%d-proto-%d-type-%d",
135 PF_INET6, protocol, sock->type);
136 /*
137 * Fall back to generic, e.g. net-pf-10-proto-132
138 * (net-pf-PF_INET6-proto-IPPROTO_SCTP)
139 */
140 else
141 request_module("net-pf-%d-proto-%d",
142 PF_INET6, protocol);
143 goto lookup_protocol;
144 } else
145 goto out_rcu_unlock;
146 }
147
148 err = -EPERM;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149 if (answer->capability > 0 && !capable(answer->capability))
150 goto out_rcu_unlock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700151
152 sock->ops = answer->ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 answer_prot = answer->prot;
154 answer_no_check = answer->no_check;
155 answer_flags = answer->flags;
156 rcu_read_unlock();
157
158 BUG_TRAP(answer_prot->slab != NULL);
159
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800160 err = -ENOBUFS;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700161 sk = sk_alloc(PF_INET6, GFP_KERNEL, answer_prot, 1);
162 if (sk == NULL)
163 goto out;
164
165 sock_init_data(sock, sk);
166
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800167 err = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 sk->sk_no_check = answer_no_check;
169 if (INET_PROTOSW_REUSE & answer_flags)
170 sk->sk_reuse = 1;
171
172 inet = inet_sk(sk);
Paul Moore469de9b2007-01-09 14:37:06 -0800173 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174
175 if (SOCK_RAW == sock->type) {
176 inet->num = protocol;
177 if (IPPROTO_RAW == protocol)
178 inet->hdrincl = 1;
179 }
180
Arnaldo Carvalho de Meloe6848972005-08-09 19:45:38 -0700181 sk->sk_destruct = inet_sock_destruct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182 sk->sk_family = PF_INET6;
183 sk->sk_protocol = protocol;
184
185 sk->sk_backlog_rcv = answer->prot->backlog_rcv;
186
187 inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk);
188 np->hop_limit = -1;
189 np->mcast_hops = -1;
190 np->mc_loop = 1;
191 np->pmtudisc = IPV6_PMTUDISC_WANT;
192 np->ipv6only = sysctl_ipv6_bindv6only;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900193
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194 /* Init the ipv4 part of the socket since we can have sockets
195 * using v6 API for ipv4.
196 */
197 inet->uc_ttl = -1;
198
199 inet->mc_loop = 1;
200 inet->mc_ttl = 1;
201 inet->mc_index = 0;
202 inet->mc_list = NULL;
203
204 if (ipv4_config.no_pmtu_disc)
205 inet->pmtudisc = IP_PMTUDISC_DONT;
206 else
207 inet->pmtudisc = IP_PMTUDISC_WANT;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900208 /*
Arnaldo Carvalho de Meloe6848972005-08-09 19:45:38 -0700209 * Increment only the relevant sk_prot->socks debug field, this changes
210 * the previous behaviour of incrementing both the equivalent to
211 * answer->prot->socks (inet6_sock_nr) and inet_sock_nr.
212 *
213 * This allows better debug granularity as we'll know exactly how many
214 * UDPv6, TCPv6, etc socks were allocated, not the sum of all IPv6
215 * transport protocol socks. -acme
216 */
217 sk_refcnt_debug_inc(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700218
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219 if (inet->num) {
220 /* It assumes that any protocol which allows
221 * the user to assign a number at socket
222 * creation time automatically shares.
223 */
Al Viroe69a4adc2006-11-14 20:56:00 -0800224 inet->sport = htons(inet->num);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225 sk->sk_prot->hash(sk);
226 }
227 if (sk->sk_prot->init) {
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800228 err = sk->sk_prot->init(sk);
229 if (err) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230 sk_common_release(sk);
231 goto out;
232 }
233 }
234out:
YOSHIFUJI Hideakiaf1afe82005-12-02 20:56:57 -0800235 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236out_rcu_unlock:
237 rcu_read_unlock();
238 goto out;
239}
240
241
242/* bind for INET6 API */
243int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
244{
245 struct sockaddr_in6 *addr=(struct sockaddr_in6 *)uaddr;
246 struct sock *sk = sock->sk;
247 struct inet_sock *inet = inet_sk(sk);
248 struct ipv6_pinfo *np = inet6_sk(sk);
Al Virofd683222006-09-26 22:17:51 -0700249 __be32 v4addr = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250 unsigned short snum;
251 int addr_type = 0;
252 int err = 0;
253
254 /* If the socket has its own bind function then use it. */
255 if (sk->sk_prot->bind)
256 return sk->sk_prot->bind(sk, uaddr, addr_len);
257
258 if (addr_len < SIN6_LEN_RFC2133)
259 return -EINVAL;
260 addr_type = ipv6_addr_type(&addr->sin6_addr);
261 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
262 return -EINVAL;
263
264 snum = ntohs(addr->sin6_port);
265 if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
266 return -EACCES;
267
268 lock_sock(sk);
269
270 /* Check these errors (active socket, double bind). */
271 if (sk->sk_state != TCP_CLOSE || inet->num) {
272 err = -EINVAL;
273 goto out;
274 }
275
276 /* Check if the address belongs to the host. */
277 if (addr_type == IPV6_ADDR_MAPPED) {
278 v4addr = addr->sin6_addr.s6_addr32[3];
279 if (inet_addr_type(v4addr) != RTN_LOCAL) {
280 err = -EADDRNOTAVAIL;
281 goto out;
282 }
283 } else {
284 if (addr_type != IPV6_ADDR_ANY) {
285 struct net_device *dev = NULL;
286
287 if (addr_type & IPV6_ADDR_LINKLOCAL) {
288 if (addr_len >= sizeof(struct sockaddr_in6) &&
289 addr->sin6_scope_id) {
290 /* Override any existing binding, if another one
291 * is supplied by user.
292 */
293 sk->sk_bound_dev_if = addr->sin6_scope_id;
294 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900295
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296 /* Binding to link-local address requires an interface */
297 if (!sk->sk_bound_dev_if) {
298 err = -EINVAL;
299 goto out;
300 }
301 dev = dev_get_by_index(sk->sk_bound_dev_if);
302 if (!dev) {
303 err = -ENODEV;
304 goto out;
305 }
306 }
307
308 /* ipv4 addr of the socket is invalid. Only the
309 * unspecified and mapped address have a v4 equivalent.
310 */
311 v4addr = LOOPBACK4_IPV6;
312 if (!(addr_type & IPV6_ADDR_MULTICAST)) {
313 if (!ipv6_chk_addr(&addr->sin6_addr, dev, 0)) {
314 if (dev)
315 dev_put(dev);
316 err = -EADDRNOTAVAIL;
317 goto out;
318 }
319 }
320 if (dev)
321 dev_put(dev);
322 }
323 }
324
325 inet->rcv_saddr = v4addr;
326 inet->saddr = v4addr;
327
328 ipv6_addr_copy(&np->rcv_saddr, &addr->sin6_addr);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900329
Linus Torvalds1da177e2005-04-16 15:20:36 -0700330 if (!(addr_type & IPV6_ADDR_MULTICAST))
331 ipv6_addr_copy(&np->saddr, &addr->sin6_addr);
332
333 /* Make sure we are allowed to bind here. */
334 if (sk->sk_prot->get_port(sk, snum)) {
335 inet_reset_saddr(sk);
336 err = -EADDRINUSE;
337 goto out;
338 }
339
340 if (addr_type != IPV6_ADDR_ANY)
341 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
342 if (snum)
343 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
Al Viroe69a4adc2006-11-14 20:56:00 -0800344 inet->sport = htons(inet->num);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345 inet->dport = 0;
346 inet->daddr = 0;
347out:
348 release_sock(sk);
349 return err;
350}
351
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900352EXPORT_SYMBOL(inet6_bind);
353
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354int inet6_release(struct socket *sock)
355{
356 struct sock *sk = sock->sk;
357
358 if (sk == NULL)
359 return -EINVAL;
360
361 /* Free mc lists */
362 ipv6_sock_mc_close(sk);
363
364 /* Free ac lists */
365 ipv6_sock_ac_close(sk);
366
367 return inet_release(sock);
368}
369
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900370EXPORT_SYMBOL(inet6_release);
371
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372int inet6_destroy_sock(struct sock *sk)
373{
374 struct ipv6_pinfo *np = inet6_sk(sk);
375 struct sk_buff *skb;
376 struct ipv6_txoptions *opt;
377
Linus Torvalds1da177e2005-04-16 15:20:36 -0700378 /* Release rx options */
379
380 if ((skb = xchg(&np->pktoptions, NULL)) != NULL)
381 kfree_skb(skb);
382
383 /* Free flowlabels */
384 fl6_free_socklist(sk);
385
386 /* Free tx options */
387
388 if ((opt = xchg(&np->opt, NULL)) != NULL)
389 sock_kfree_s(sk, opt, opt->tot_len);
390
391 return 0;
392}
393
Arnaldo Carvalho de Melo3cf3dc62005-12-13 23:23:20 -0800394EXPORT_SYMBOL_GPL(inet6_destroy_sock);
395
Linus Torvalds1da177e2005-04-16 15:20:36 -0700396/*
397 * This does both peername and sockname.
398 */
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900399
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
401 int *uaddr_len, int peer)
402{
403 struct sockaddr_in6 *sin=(struct sockaddr_in6 *)uaddr;
404 struct sock *sk = sock->sk;
405 struct inet_sock *inet = inet_sk(sk);
406 struct ipv6_pinfo *np = inet6_sk(sk);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900407
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408 sin->sin6_family = AF_INET6;
409 sin->sin6_flowinfo = 0;
410 sin->sin6_scope_id = 0;
411 if (peer) {
412 if (!inet->dport)
413 return -ENOTCONN;
414 if (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) &&
415 peer == 1)
416 return -ENOTCONN;
417 sin->sin6_port = inet->dport;
418 ipv6_addr_copy(&sin->sin6_addr, &np->daddr);
419 if (np->sndflow)
420 sin->sin6_flowinfo = np->flow_label;
421 } else {
422 if (ipv6_addr_any(&np->rcv_saddr))
423 ipv6_addr_copy(&sin->sin6_addr, &np->saddr);
424 else
425 ipv6_addr_copy(&sin->sin6_addr, &np->rcv_saddr);
426
427 sin->sin6_port = inet->sport;
428 }
429 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL)
430 sin->sin6_scope_id = sk->sk_bound_dev_if;
431 *uaddr_len = sizeof(*sin);
432 return(0);
433}
434
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900435EXPORT_SYMBOL(inet6_getname);
436
Linus Torvalds1da177e2005-04-16 15:20:36 -0700437int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
438{
439 struct sock *sk = sock->sk;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700440
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900441 switch(cmd)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700442 {
443 case SIOCGSTAMP:
444 return sock_get_timestamp(sk, (struct timeval __user *)arg);
445
Eric Dumazetae40eb12007-03-18 17:33:16 -0700446 case SIOCGSTAMPNS:
447 return sock_get_timestampns(sk, (struct timespec __user *)arg);
448
Linus Torvalds1da177e2005-04-16 15:20:36 -0700449 case SIOCADDRT:
450 case SIOCDELRT:
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900451
Linus Torvalds1da177e2005-04-16 15:20:36 -0700452 return(ipv6_route_ioctl(cmd,(void __user *)arg));
453
454 case SIOCSIFADDR:
455 return addrconf_add_ifaddr((void __user *) arg);
456 case SIOCDIFADDR:
457 return addrconf_del_ifaddr((void __user *) arg);
458 case SIOCSIFDSTADDR:
459 return addrconf_set_dstaddr((void __user *) arg);
460 default:
Christoph Hellwigb5e5fa52006-01-03 14:18:33 -0800461 if (!sk->sk_prot->ioctl)
462 return -ENOIOCTLCMD;
463 return sk->sk_prot->ioctl(sk, cmd, arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464 }
465 /*NOTREACHED*/
466 return(0);
467}
468
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900469EXPORT_SYMBOL(inet6_ioctl);
470
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800471const struct proto_ops inet6_stream_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800472 .family = PF_INET6,
473 .owner = THIS_MODULE,
474 .release = inet6_release,
475 .bind = inet6_bind,
476 .connect = inet_stream_connect, /* ok */
477 .socketpair = sock_no_socketpair, /* a do nothing */
478 .accept = inet_accept, /* ok */
479 .getname = inet6_getname,
480 .poll = tcp_poll, /* ok */
481 .ioctl = inet6_ioctl, /* must change */
482 .listen = inet_listen, /* ok */
483 .shutdown = inet_shutdown, /* ok */
484 .setsockopt = sock_common_setsockopt, /* ok */
485 .getsockopt = sock_common_getsockopt, /* ok */
486 .sendmsg = inet_sendmsg, /* ok */
487 .recvmsg = sock_common_recvmsg, /* ok */
488 .mmap = sock_no_mmap,
489 .sendpage = tcp_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800490#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800491 .compat_setsockopt = compat_sock_common_setsockopt,
492 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800493#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700494};
495
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800496const struct proto_ops inet6_dgram_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800497 .family = PF_INET6,
498 .owner = THIS_MODULE,
499 .release = inet6_release,
500 .bind = inet6_bind,
501 .connect = inet_dgram_connect, /* ok */
502 .socketpair = sock_no_socketpair, /* a do nothing */
503 .accept = sock_no_accept, /* a do nothing */
504 .getname = inet6_getname,
505 .poll = udp_poll, /* ok */
506 .ioctl = inet6_ioctl, /* must change */
507 .listen = sock_no_listen, /* ok */
508 .shutdown = inet_shutdown, /* ok */
509 .setsockopt = sock_common_setsockopt, /* ok */
510 .getsockopt = sock_common_getsockopt, /* ok */
511 .sendmsg = inet_sendmsg, /* ok */
512 .recvmsg = sock_common_recvmsg, /* ok */
513 .mmap = sock_no_mmap,
514 .sendpage = sock_no_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800515#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800516 .compat_setsockopt = compat_sock_common_setsockopt,
517 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800518#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519};
520
521static struct net_proto_family inet6_family_ops = {
522 .family = PF_INET6,
523 .create = inet6_create,
524 .owner = THIS_MODULE,
525};
526
Linus Torvalds1da177e2005-04-16 15:20:36 -0700527/* Same as inet6_dgram_ops, sans udp_poll. */
Eric Dumazet90ddc4f2005-12-22 12:49:22 -0800528static const struct proto_ops inet6_sockraw_ops = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800529 .family = PF_INET6,
530 .owner = THIS_MODULE,
531 .release = inet6_release,
532 .bind = inet6_bind,
533 .connect = inet_dgram_connect, /* ok */
534 .socketpair = sock_no_socketpair, /* a do nothing */
535 .accept = sock_no_accept, /* a do nothing */
536 .getname = inet6_getname,
537 .poll = datagram_poll, /* ok */
538 .ioctl = inet6_ioctl, /* must change */
539 .listen = sock_no_listen, /* ok */
540 .shutdown = inet_shutdown, /* ok */
541 .setsockopt = sock_common_setsockopt, /* ok */
542 .getsockopt = sock_common_getsockopt, /* ok */
543 .sendmsg = inet_sendmsg, /* ok */
544 .recvmsg = sock_common_recvmsg, /* ok */
545 .mmap = sock_no_mmap,
546 .sendpage = sock_no_sendpage,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800547#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800548 .compat_setsockopt = compat_sock_common_setsockopt,
549 .compat_getsockopt = compat_sock_common_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800550#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700551};
552
553static struct inet_protosw rawv6_protosw = {
554 .type = SOCK_RAW,
555 .protocol = IPPROTO_IP, /* wild card */
556 .prot = &rawv6_prot,
557 .ops = &inet6_sockraw_ops,
558 .capability = CAP_NET_RAW,
559 .no_check = UDP_CSUM_DEFAULT,
560 .flags = INET_PROTOSW_REUSE,
561};
562
563void
564inet6_register_protosw(struct inet_protosw *p)
565{
566 struct list_head *lh;
567 struct inet_protosw *answer;
568 int protocol = p->protocol;
569 struct list_head *last_perm;
570
571 spin_lock_bh(&inetsw6_lock);
572
573 if (p->type >= SOCK_MAX)
574 goto out_illegal;
575
576 /* If we are trying to override a permanent protocol, bail. */
577 answer = NULL;
578 last_perm = &inetsw6[p->type];
579 list_for_each(lh, &inetsw6[p->type]) {
580 answer = list_entry(lh, struct inet_protosw, list);
581
582 /* Check only the non-wild match. */
583 if (INET_PROTOSW_PERMANENT & answer->flags) {
584 if (protocol == answer->protocol)
585 break;
586 last_perm = lh;
587 }
588
589 answer = NULL;
590 }
591 if (answer)
592 goto out_permanent;
593
594 /* Add the new entry after the last permanent entry if any, so that
595 * the new entry does not override a permanent entry when matched with
596 * a wild-card protocol. But it is allowed to override any existing
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900597 * non-permanent entry. This means that when we remove this entry, the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700598 * system automatically returns to the old behavior.
599 */
600 list_add_rcu(&p->list, last_perm);
601out:
602 spin_unlock_bh(&inetsw6_lock);
603 return;
604
605out_permanent:
606 printk(KERN_ERR "Attempt to override permanent protocol %d.\n",
607 protocol);
608 goto out;
609
610out_illegal:
611 printk(KERN_ERR
612 "Ignoring attempt to register invalid socket type %d.\n",
613 p->type);
614 goto out;
615}
616
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900617EXPORT_SYMBOL(inet6_register_protosw);
618
Linus Torvalds1da177e2005-04-16 15:20:36 -0700619void
620inet6_unregister_protosw(struct inet_protosw *p)
621{
622 if (INET_PROTOSW_PERMANENT & p->flags) {
623 printk(KERN_ERR
624 "Attempt to unregister permanent protocol %d.\n",
625 p->protocol);
626 } else {
627 spin_lock_bh(&inetsw6_lock);
628 list_del_rcu(&p->list);
629 spin_unlock_bh(&inetsw6_lock);
630
631 synchronize_net();
632 }
633}
634
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900635EXPORT_SYMBOL(inet6_unregister_protosw);
636
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800637int inet6_sk_rebuild_header(struct sock *sk)
638{
639 int err;
640 struct dst_entry *dst;
641 struct ipv6_pinfo *np = inet6_sk(sk);
642
643 dst = __sk_dst_check(sk, np->dst_cookie);
644
645 if (dst == NULL) {
646 struct inet_sock *inet = inet_sk(sk);
647 struct in6_addr *final_p = NULL, final;
648 struct flowi fl;
649
650 memset(&fl, 0, sizeof(fl));
651 fl.proto = sk->sk_protocol;
652 ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
653 ipv6_addr_copy(&fl.fl6_src, &np->saddr);
654 fl.fl6_flowlabel = np->flow_label;
655 fl.oif = sk->sk_bound_dev_if;
656 fl.fl_ip_dport = inet->dport;
657 fl.fl_ip_sport = inet->sport;
Venkat Yekkiralabeb8d132006-08-04 23:12:42 -0700658 security_sk_classify_flow(sk, &fl);
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800659
660 if (np->opt && np->opt->srcrt) {
661 struct rt0_hdr *rt0 = (struct rt0_hdr *) np->opt->srcrt;
662 ipv6_addr_copy(&final, &fl.fl6_dst);
663 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
664 final_p = &final;
665 }
666
667 err = ip6_dst_lookup(sk, &dst, &fl);
668 if (err) {
669 sk->sk_route_caps = 0;
670 return err;
671 }
672 if (final_p)
673 ipv6_addr_copy(&fl.fl6_dst, final_p);
674
675 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) {
676 sk->sk_err_soft = -err;
677 return err;
678 }
679
YOSHIFUJI Hideaki8e1ef0a2006-08-29 17:15:09 -0700680 __ip6_dst_store(sk, dst, NULL, NULL);
Arnaldo Carvalho de Melob9750ce2005-12-13 23:22:54 -0800681 }
682
683 return 0;
684}
685
686EXPORT_SYMBOL_GPL(inet6_sk_rebuild_header);
687
Arnaldo Carvalho de Melo399c07d2005-12-13 23:24:28 -0800688int ipv6_opt_accepted(struct sock *sk, struct sk_buff *skb)
689{
690 struct ipv6_pinfo *np = inet6_sk(sk);
691 struct inet6_skb_parm *opt = IP6CB(skb);
692
693 if (np->rxopt.all) {
694 if ((opt->hop && (np->rxopt.bits.hopopts ||
695 np->rxopt.bits.ohopopts)) ||
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700696 ((IPV6_FLOWINFO_MASK &
697 *(__be32 *)skb_network_header(skb)) &&
Arnaldo Carvalho de Melo399c07d2005-12-13 23:24:28 -0800698 np->rxopt.bits.rxflow) ||
699 (opt->srcrt && (np->rxopt.bits.srcrt ||
700 np->rxopt.bits.osrcrt)) ||
701 ((opt->dst1 || opt->dst0) &&
702 (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts)))
703 return 1;
704 }
705 return 0;
706}
707
708EXPORT_SYMBOL_GPL(ipv6_opt_accepted);
709
Linus Torvalds1da177e2005-04-16 15:20:36 -0700710int
711snmp6_mib_init(void *ptr[2], size_t mibsize, size_t mibalign)
712{
713 if (ptr == NULL)
714 return -EINVAL;
715
Pekka Enbergf9f75002006-01-08 01:00:33 -0800716 ptr[0] = __alloc_percpu(mibsize);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700717 if (!ptr[0])
718 goto err0;
719
Pekka Enbergf9f75002006-01-08 01:00:33 -0800720 ptr[1] = __alloc_percpu(mibsize);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700721 if (!ptr[1])
722 goto err1;
723
724 return 0;
725
726err1:
727 free_percpu(ptr[0]);
728 ptr[0] = NULL;
729err0:
730 return -ENOMEM;
731}
732
733void
734snmp6_mib_free(void *ptr[2])
735{
736 if (ptr == NULL)
737 return;
Alan Sterna1205862006-12-06 20:32:37 -0800738 free_percpu(ptr[0]);
739 free_percpu(ptr[1]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740 ptr[0] = ptr[1] = NULL;
741}
742
743static int __init init_ipv6_mibs(void)
744{
745 if (snmp6_mib_init((void **)ipv6_statistics, sizeof (struct ipstats_mib),
746 __alignof__(struct ipstats_mib)) < 0)
747 goto err_ip_mib;
748 if (snmp6_mib_init((void **)icmpv6_statistics, sizeof (struct icmpv6_mib),
749 __alignof__(struct icmpv6_mib)) < 0)
750 goto err_icmp_mib;
751 if (snmp6_mib_init((void **)udp_stats_in6, sizeof (struct udp_mib),
752 __alignof__(struct udp_mib)) < 0)
753 goto err_udp_mib;
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800754 if (snmp6_mib_init((void **)udplite_stats_in6, sizeof (struct udp_mib),
755 __alignof__(struct udp_mib)) < 0)
756 goto err_udplite_mib;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757 return 0;
758
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800759err_udplite_mib:
760 snmp6_mib_free((void **)udp_stats_in6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700761err_udp_mib:
762 snmp6_mib_free((void **)icmpv6_statistics);
763err_icmp_mib:
764 snmp6_mib_free((void **)ipv6_statistics);
765err_ip_mib:
766 return -ENOMEM;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900767
Linus Torvalds1da177e2005-04-16 15:20:36 -0700768}
769
770static void cleanup_ipv6_mibs(void)
771{
772 snmp6_mib_free((void **)ipv6_statistics);
773 snmp6_mib_free((void **)icmpv6_statistics);
774 snmp6_mib_free((void **)udp_stats_in6);
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800775 snmp6_mib_free((void **)udplite_stats_in6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776}
777
Linus Torvalds1da177e2005-04-16 15:20:36 -0700778static int __init inet6_init(void)
779{
780 struct sk_buff *dummy_skb;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900781 struct list_head *r;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700782 int err;
783
YOSHIFUJI Hideakief047f52006-09-01 00:29:06 -0700784 BUILD_BUG_ON(sizeof(struct inet6_skb_parm) > sizeof(dummy_skb->cb));
785
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786#ifdef MODULE
787#if 0 /* FIXME --RR */
788 if (!mod_member_present(&__this_module, can_unload))
789 return -EINVAL;
790
791 __this_module.can_unload = &ipv6_unload;
792#endif
793#endif
794
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795 err = proto_register(&tcpv6_prot, 1);
796 if (err)
797 goto out;
798
799 err = proto_register(&udpv6_prot, 1);
800 if (err)
801 goto out_unregister_tcp_proto;
802
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800803 err = proto_register(&udplitev6_prot, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700804 if (err)
805 goto out_unregister_udp_proto;
806
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800807 err = proto_register(&rawv6_prot, 1);
808 if (err)
809 goto out_unregister_udplite_proto;
810
Linus Torvalds1da177e2005-04-16 15:20:36 -0700811
812 /* Register the socket-side information for inet6_create. */
813 for(r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r)
814 INIT_LIST_HEAD(r);
815
816 /* We MUST register RAW sockets before we create the ICMP6,
817 * IGMP6, or NDISC control sockets.
818 */
819 inet6_register_protosw(&rawv6_protosw);
820
821 /* Register the family here so that the init calls below will
822 * be able to create sockets. (?? is this dangerous ??)
823 */
David S. Miller8eb55912005-11-11 15:05:47 -0800824 err = sock_register(&inet6_family_ops);
825 if (err)
826 goto out_unregister_raw_proto;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700827
828 /* Initialise ipv6 mibs */
829 err = init_ipv6_mibs();
830 if (err)
David S. Miller8eb55912005-11-11 15:05:47 -0800831 goto out_unregister_sock;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900832
Linus Torvalds1da177e2005-04-16 15:20:36 -0700833 /*
834 * ipngwg API draft makes clear that the correct semantics
835 * for TCP and UDP is to consider one TCP and UDP instance
836 * in a host availiable by both INET and INET6 APIs and
837 * able to communicate via both network protocols.
838 */
839
840#ifdef CONFIG_SYSCTL
841 ipv6_sysctl_register();
842#endif
843 err = icmpv6_init(&inet6_family_ops);
844 if (err)
845 goto icmp_fail;
846 err = ndisc_init(&inet6_family_ops);
847 if (err)
848 goto ndisc_fail;
849 err = igmp6_init(&inet6_family_ops);
850 if (err)
851 goto igmp_fail;
Harald Welte2cc7d572005-08-09 19:42:34 -0700852 err = ipv6_netfilter_init();
853 if (err)
854 goto netfilter_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700855 /* Create /proc/foo6 entries. */
856#ifdef CONFIG_PROC_FS
857 err = -ENOMEM;
858 if (raw6_proc_init())
859 goto proc_raw6_fail;
860 if (tcp6_proc_init())
861 goto proc_tcp6_fail;
862 if (udp6_proc_init())
863 goto proc_udp6_fail;
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800864 if (udplite6_proc_init())
865 goto proc_udplite6_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866 if (ipv6_misc_proc_init())
867 goto proc_misc6_fail;
868
869 if (ac6_proc_init())
870 goto proc_anycast6_fail;
871 if (if6_proc_init())
872 goto proc_if6_fail;
873#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700874 ip6_route_init();
875 ip6_flowlabel_init();
876 err = addrconf_init();
877 if (err)
878 goto addrconf_fail;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700879
880 /* Init v6 extension headers. */
881 ipv6_rthdr_init();
882 ipv6_frag_init();
883 ipv6_nodata_init();
884 ipv6_destopt_init();
Noriaki TAKAMIYA2c8d7ca2006-08-23 20:31:11 -0700885#ifdef CONFIG_IPV6_MIP6
886 mip6_init();
887#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888
889 /* Init v6 transport protocols. */
890 udpv6_init();
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800891 udplitev6_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700892 tcpv6_init();
Herbert Xue2ed4052005-07-05 14:41:20 -0700893
894 ipv6_packet_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700895 err = 0;
896out:
897 return err;
898
899addrconf_fail:
900 ip6_flowlabel_cleanup();
901 ip6_route_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902#ifdef CONFIG_PROC_FS
903 if6_proc_exit();
904proc_if6_fail:
905 ac6_proc_exit();
906proc_anycast6_fail:
907 ipv6_misc_proc_exit();
908proc_misc6_fail:
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800909 udplite6_proc_exit();
910proc_udplite6_fail:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700911 udp6_proc_exit();
912proc_udp6_fail:
913 tcp6_proc_exit();
914proc_tcp6_fail:
915 raw6_proc_exit();
916proc_raw6_fail:
917#endif
Harald Welte2cc7d572005-08-09 19:42:34 -0700918 ipv6_netfilter_fini();
919netfilter_fail:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700920 igmp6_cleanup();
921igmp_fail:
922 ndisc_cleanup();
923ndisc_fail:
924 icmpv6_cleanup();
925icmp_fail:
926#ifdef CONFIG_SYSCTL
927 ipv6_sysctl_unregister();
928#endif
929 cleanup_ipv6_mibs();
David S. Miller8eb55912005-11-11 15:05:47 -0800930out_unregister_sock:
931 sock_unregister(PF_INET6);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700932out_unregister_raw_proto:
933 proto_unregister(&rawv6_prot);
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800934out_unregister_udplite_proto:
935 proto_unregister(&udplitev6_prot);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936out_unregister_udp_proto:
937 proto_unregister(&udpv6_prot);
938out_unregister_tcp_proto:
939 proto_unregister(&tcpv6_prot);
940 goto out;
941}
942module_init(inet6_init);
943
944static void __exit inet6_exit(void)
945{
946 /* First of all disallow new sockets creation. */
947 sock_unregister(PF_INET6);
Thomas Grafc127ea22007-03-22 11:58:32 -0700948 /* Disallow any further netlink messages */
949 rtnl_unregister_all(PF_INET6);
Joe Jinca17c232007-02-20 01:30:15 -0800950
951 /* Cleanup code parts. */
952 ipv6_packet_cleanup();
Noriaki TAKAMIYA2c8d7ca2006-08-23 20:31:11 -0700953#ifdef CONFIG_IPV6_MIP6
954 mip6_fini();
955#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700956 addrconf_cleanup();
Joe Jinca17c232007-02-20 01:30:15 -0800957 ip6_flowlabel_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700958 ip6_route_cleanup();
Joe Jinca17c232007-02-20 01:30:15 -0800959#ifdef CONFIG_PROC_FS
960
961 /* Cleanup code parts. */
962 if6_proc_exit();
963 ac6_proc_exit();
964 ipv6_misc_proc_exit();
965 udplite6_proc_exit();
966 udp6_proc_exit();
967 tcp6_proc_exit();
968 raw6_proc_exit();
969#endif
Harald Welte2cc7d572005-08-09 19:42:34 -0700970 ipv6_netfilter_fini();
Joe Jinca17c232007-02-20 01:30:15 -0800971 igmp6_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972 ndisc_cleanup();
973 icmpv6_cleanup();
974#ifdef CONFIG_SYSCTL
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900975 ipv6_sysctl_unregister();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700976#endif
977 cleanup_ipv6_mibs();
978 proto_unregister(&rawv6_prot);
Joe Jinca17c232007-02-20 01:30:15 -0800979 proto_unregister(&udplitev6_prot);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980 proto_unregister(&udpv6_prot);
981 proto_unregister(&tcpv6_prot);
982}
983module_exit(inet6_exit);
984
985MODULE_ALIAS_NETPROTO(PF_INET6);