blob: 603df76e052267f5882952e59e51296912143653 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * RAW sockets for IPv6
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/raw.c
9 *
10 * $Id: raw.c,v 1.51 2002/02/01 22:01:04 davem Exp $
11 *
12 * Fixes:
13 * Hideaki YOSHIFUJI : sin6_scope_id support
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +090014 * YOSHIFUJI,H.@USAGI : raw checksum (RFC2292(bis) compliance)
Linus Torvalds1da177e2005-04-16 15:20:36 -070015 * Kazunori MIYAZAWA @USAGI: change process style to use ip6_append_data
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#include <linux/errno.h>
24#include <linux/types.h>
25#include <linux/socket.h>
26#include <linux/sockios.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070027#include <linux/net.h>
28#include <linux/in6.h>
29#include <linux/netdevice.h>
30#include <linux/if_arp.h>
31#include <linux/icmpv6.h>
32#include <linux/netfilter.h>
33#include <linux/netfilter_ipv6.h>
Herbert Xu3305b802005-12-13 23:16:37 -080034#include <linux/skbuff.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070035#include <asm/uaccess.h>
36#include <asm/ioctls.h>
37
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020038#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070039#include <net/ip.h>
40#include <net/sock.h>
41#include <net/snmp.h>
42
43#include <net/ipv6.h>
44#include <net/ndisc.h>
45#include <net/protocol.h>
46#include <net/ip6_route.h>
47#include <net/ip6_checksum.h>
48#include <net/addrconf.h>
49#include <net/transp_v6.h>
50#include <net/udp.h>
51#include <net/inet_common.h>
Arnaldo Carvalho de Meloc752f072005-08-09 20:08:28 -070052#include <net/tcp_states.h>
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -070053#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -070054#include <net/mip6.h>
55#endif
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +090056#include <linux/mroute6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070057
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080058#include <net/raw.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070059#include <net/rawv6.h>
60#include <net/xfrm.h>
61
62#include <linux/proc_fs.h>
63#include <linux/seq_file.h>
64
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080065static struct raw_hashinfo raw_v6_hashinfo = {
Robert P. J. Day938b93a2008-03-18 00:59:23 -070066 .lock = __RW_LOCK_UNLOCKED(raw_v6_hashinfo.lock),
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080067};
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
Pavel Emelyanovbe185882008-01-14 05:35:31 -080069static struct sock *__raw_v6_lookup(struct net *net, struct sock *sk,
70 unsigned short num, struct in6_addr *loc_addr,
71 struct in6_addr *rmt_addr, int dif)
Linus Torvalds1da177e2005-04-16 15:20:36 -070072{
73 struct hlist_node *node;
74 int is_multicast = ipv6_addr_is_multicast(loc_addr);
75
76 sk_for_each_from(sk, node)
77 if (inet_sk(sk)->num == num) {
78 struct ipv6_pinfo *np = inet6_sk(sk);
79
YOSHIFUJI Hideaki878628f2008-03-26 03:57:35 +090080 if (!net_eq(sock_net(sk), net))
Pavel Emelyanovbe185882008-01-14 05:35:31 -080081 continue;
82
Linus Torvalds1da177e2005-04-16 15:20:36 -070083 if (!ipv6_addr_any(&np->daddr) &&
84 !ipv6_addr_equal(&np->daddr, rmt_addr))
85 continue;
86
Andrew McDonald0bd1b592005-08-09 19:44:42 -070087 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
88 continue;
89
Linus Torvalds1da177e2005-04-16 15:20:36 -070090 if (!ipv6_addr_any(&np->rcv_saddr)) {
91 if (ipv6_addr_equal(&np->rcv_saddr, loc_addr))
92 goto found;
93 if (is_multicast &&
94 inet6_mc_check(sk, loc_addr, rmt_addr))
95 goto found;
96 continue;
97 }
98 goto found;
99 }
100 sk = NULL;
101found:
102 return sk;
103}
104
105/*
106 * 0 - deliver
107 * 1 - block
108 */
109static __inline__ int icmpv6_filter(struct sock *sk, struct sk_buff *skb)
110{
111 struct icmp6hdr *icmph;
112 struct raw6_sock *rp = raw6_sk(sk);
113
114 if (pskb_may_pull(skb, sizeof(struct icmp6hdr))) {
115 __u32 *data = &rp->filter.data[0];
116 int bit_nr;
117
118 icmph = (struct icmp6hdr *) skb->data;
119 bit_nr = icmph->icmp6_type;
120
121 return (data[bit_nr >> 5] & (1 << (bit_nr & 31))) != 0;
122 }
123 return 0;
124}
125
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700126#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
127static int (*mh_filter)(struct sock *sock, struct sk_buff *skb);
128
129int rawv6_mh_filter_register(int (*filter)(struct sock *sock,
130 struct sk_buff *skb))
131{
132 rcu_assign_pointer(mh_filter, filter);
133 return 0;
134}
135EXPORT_SYMBOL(rawv6_mh_filter_register);
136
137int rawv6_mh_filter_unregister(int (*filter)(struct sock *sock,
138 struct sk_buff *skb))
139{
140 rcu_assign_pointer(mh_filter, NULL);
141 synchronize_rcu();
142 return 0;
143}
144EXPORT_SYMBOL(rawv6_mh_filter_unregister);
145
146#endif
147
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148/*
149 * demultiplex raw sockets.
150 * (should consider queueing the skb in the sock receive_queue
151 * without calling rawv6.c)
152 *
153 * Caller owns SKB so we must make clones.
154 */
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800155static int ipv6_raw_deliver(struct sk_buff *skb, int nexthdr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700156{
157 struct in6_addr *saddr;
158 struct in6_addr *daddr;
159 struct sock *sk;
Patrick McHardyd13964f2005-08-09 19:45:02 -0700160 int delivered = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700161 __u8 hash;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800162 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700163
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700164 saddr = &ipv6_hdr(skb)->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700165 daddr = saddr + 1;
166
167 hash = nexthdr & (MAX_INET_PROTOS - 1);
168
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800169 read_lock(&raw_v6_hashinfo.lock);
170 sk = sk_head(&raw_v6_hashinfo.ht[hash]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172 if (sk == NULL)
173 goto out;
174
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900175 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800176 sk = __raw_v6_lookup(net, sk, nexthdr, daddr, saddr, IP6CB(skb)->iif);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700177
178 while (sk) {
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700179 int filtered;
180
Patrick McHardyd13964f2005-08-09 19:45:02 -0700181 delivered = 1;
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700182 switch (nexthdr) {
183 case IPPROTO_ICMPV6:
184 filtered = icmpv6_filter(sk, skb);
185 break;
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700186
187#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700188 case IPPROTO_MH:
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700189 {
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700190 /* XXX: To validate MH only once for each packet,
191 * this is placed here. It should be after checking
192 * xfrm policy, however it doesn't. The checking xfrm
193 * policy is placed in rawv6_rcv() because it is
194 * required for each socket.
195 */
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700196 int (*filter)(struct sock *sock, struct sk_buff *skb);
197
198 filter = rcu_dereference(mh_filter);
199 filtered = filter ? filter(sk, skb) : 0;
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700200 break;
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700201 }
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700202#endif
203 default:
204 filtered = 0;
205 break;
206 }
207
208 if (filtered < 0)
209 break;
210 if (filtered == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700211 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
212
213 /* Not releasing hash table! */
Yasuyuki Kozakai9fb9cbb2005-11-09 16:38:16 -0800214 if (clone) {
215 nf_reset(clone);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700216 rawv6_rcv(sk, clone);
Yasuyuki Kozakai9fb9cbb2005-11-09 16:38:16 -0800217 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700218 }
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800219 sk = __raw_v6_lookup(net, sk_next(sk), nexthdr, daddr, saddr,
YOSHIFUJI Hideaki2dac4b92005-09-01 17:44:49 -0700220 IP6CB(skb)->iif);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700221 }
222out:
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800223 read_unlock(&raw_v6_hashinfo.lock);
Patrick McHardyd13964f2005-08-09 19:45:02 -0700224 return delivered;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225}
226
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800227int raw6_local_deliver(struct sk_buff *skb, int nexthdr)
228{
229 struct sock *raw_sk;
230
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800231 raw_sk = sk_head(&raw_v6_hashinfo.ht[nexthdr & (MAX_INET_PROTOS - 1)]);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800232 if (raw_sk && !ipv6_raw_deliver(skb, nexthdr))
233 raw_sk = NULL;
234
235 return raw_sk != NULL;
236}
237
Linus Torvalds1da177e2005-04-16 15:20:36 -0700238/* This cleans up af_inet6 a bit. -DaveM */
239static int rawv6_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
240{
241 struct inet_sock *inet = inet_sk(sk);
242 struct ipv6_pinfo *np = inet6_sk(sk);
243 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
Al Viroe69a4ad2006-11-14 20:56:00 -0800244 __be32 v4addr = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700245 int addr_type;
246 int err;
247
248 if (addr_len < SIN6_LEN_RFC2133)
249 return -EINVAL;
250 addr_type = ipv6_addr_type(&addr->sin6_addr);
251
252 /* Raw sockets are IPv6 only */
253 if (addr_type == IPV6_ADDR_MAPPED)
254 return(-EADDRNOTAVAIL);
255
256 lock_sock(sk);
257
258 err = -EINVAL;
259 if (sk->sk_state != TCP_CLOSE)
260 goto out;
261
262 /* Check if the address belongs to the host. */
263 if (addr_type != IPV6_ADDR_ANY) {
264 struct net_device *dev = NULL;
265
266 if (addr_type & IPV6_ADDR_LINKLOCAL) {
267 if (addr_len >= sizeof(struct sockaddr_in6) &&
268 addr->sin6_scope_id) {
269 /* Override any existing binding, if another
270 * one is supplied by user.
271 */
272 sk->sk_bound_dev_if = addr->sin6_scope_id;
273 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900274
Linus Torvalds1da177e2005-04-16 15:20:36 -0700275 /* Binding to link-local address requires an interface */
276 if (!sk->sk_bound_dev_if)
277 goto out;
278
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900279 dev = dev_get_by_index(sock_net(sk), sk->sk_bound_dev_if);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280 if (!dev) {
281 err = -ENODEV;
282 goto out;
283 }
284 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900285
Linus Torvalds1da177e2005-04-16 15:20:36 -0700286 /* ipv4 addr of the socket is invalid. Only the
287 * unspecified and mapped address have a v4 equivalent.
288 */
289 v4addr = LOOPBACK4_IPV6;
290 if (!(addr_type & IPV6_ADDR_MULTICAST)) {
291 err = -EADDRNOTAVAIL;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900292 if (!ipv6_chk_addr(sock_net(sk), &addr->sin6_addr,
Daniel Lezcanobfeade02008-01-10 22:43:18 -0800293 dev, 0)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700294 if (dev)
295 dev_put(dev);
296 goto out;
297 }
298 }
299 if (dev)
300 dev_put(dev);
301 }
302
303 inet->rcv_saddr = inet->saddr = v4addr;
304 ipv6_addr_copy(&np->rcv_saddr, &addr->sin6_addr);
305 if (!(addr_type & IPV6_ADDR_MULTICAST))
306 ipv6_addr_copy(&np->saddr, &addr->sin6_addr);
307 err = 0;
308out:
309 release_sock(sk);
310 return err;
311}
312
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800313static void rawv6_err(struct sock *sk, struct sk_buff *skb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314 struct inet6_skb_parm *opt,
Al Viro04ce6902006-11-08 00:21:01 -0800315 int type, int code, int offset, __be32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316{
317 struct inet_sock *inet = inet_sk(sk);
318 struct ipv6_pinfo *np = inet6_sk(sk);
319 int err;
320 int harderr;
321
322 /* Report error on raw socket, if:
323 1. User requested recverr.
324 2. Socket is connected (otherwise the error indication
325 is useless without recverr and error is hard.
326 */
327 if (!np->recverr && sk->sk_state != TCP_ESTABLISHED)
328 return;
329
330 harderr = icmpv6_err_convert(type, code, &err);
331 if (type == ICMPV6_PKT_TOOBIG)
332 harderr = (np->pmtudisc == IPV6_PMTUDISC_DO);
333
334 if (np->recverr) {
335 u8 *payload = skb->data;
336 if (!inet->hdrincl)
337 payload += offset;
338 ipv6_icmp_error(sk, skb, err, 0, ntohl(info), payload);
339 }
340
341 if (np->recverr || harderr) {
342 sk->sk_err = err;
343 sk->sk_error_report(sk);
344 }
345}
346
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800347void raw6_icmp_error(struct sk_buff *skb, int nexthdr,
348 int type, int code, int inner_offset, __be32 info)
349{
350 struct sock *sk;
351 int hash;
352 struct in6_addr *saddr, *daddr;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800353 struct net *net;
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800354
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800355 hash = nexthdr & (RAW_HTABLE_SIZE - 1);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800356
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800357 read_lock(&raw_v6_hashinfo.lock);
358 sk = sk_head(&raw_v6_hashinfo.ht[hash]);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800359 if (sk != NULL) {
YOSHIFUJI Hideaki05f175c2008-04-11 23:51:26 +0900360 /* Note: ipv6_hdr(skb) != skb->data */
361 struct ipv6hdr *ip6h = (struct ipv6hdr *)skb->data;
362 saddr = &ip6h->saddr;
363 daddr = &ip6h->daddr;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900364 net = dev_net(skb->dev);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800365
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800366 while ((sk = __raw_v6_lookup(net, sk, nexthdr, saddr, daddr,
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800367 IP6CB(skb)->iif))) {
368 rawv6_err(sk, skb, NULL, type, code,
369 inner_offset, info);
370 sk = sk_next(sk);
371 }
372 }
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800373 read_unlock(&raw_v6_hashinfo.lock);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800374}
375
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376static inline int rawv6_rcv_skb(struct sock * sk, struct sk_buff * skb)
377{
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900378 if ((raw6_sk(sk)->checksum || sk->sk_filter) &&
Herbert Xufb286bb2005-11-10 13:01:24 -0800379 skb_checksum_complete(skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800380 atomic_inc(&sk->sk_drops);
Herbert Xufb286bb2005-11-10 13:01:24 -0800381 kfree_skb(skb);
382 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383 }
384
385 /* Charge it to the socket. */
386 if (sock_queue_rcv_skb(sk,skb)<0) {
Wang Chena92aa312007-11-13 20:31:14 -0800387 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700388 kfree_skb(skb);
389 return 0;
390 }
391
392 return 0;
393}
394
395/*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900396 * This is next to useless...
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397 * if we demultiplex in network layer we don't need the extra call
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900398 * just to queue the skb...
399 * maybe we could have the network decide upon a hint if it
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400 * should call raw_rcv for demultiplexing
401 */
402int rawv6_rcv(struct sock *sk, struct sk_buff *skb)
403{
404 struct inet_sock *inet = inet_sk(sk);
405 struct raw6_sock *rp = raw6_sk(sk);
406
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900407 if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800408 atomic_inc(&sk->sk_drops);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900409 kfree_skb(skb);
410 return NET_RX_DROP;
411 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412
413 if (!rp->checksum)
414 skb->ip_summed = CHECKSUM_UNNECESSARY;
415
Patrick McHardy84fa7932006-08-29 16:44:56 -0700416 if (skb->ip_summed == CHECKSUM_COMPLETE) {
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700417 skb_postpull_rcsum(skb, skb_network_header(skb),
Arnaldo Carvalho de Melocfe1fc72007-03-16 17:26:39 -0300418 skb_network_header_len(skb));
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700419 if (!csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
420 &ipv6_hdr(skb)->daddr,
Herbert Xufb286bb2005-11-10 13:01:24 -0800421 skb->len, inet->num, skb->csum))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700422 skb->ip_summed = CHECKSUM_UNNECESSARY;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423 }
Herbert Xu60476372007-04-09 11:59:39 -0700424 if (!skb_csum_unnecessary(skb))
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700425 skb->csum = ~csum_unfold(csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
426 &ipv6_hdr(skb)->daddr,
427 skb->len,
428 inet->num, 0));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700429
430 if (inet->hdrincl) {
Herbert Xufb286bb2005-11-10 13:01:24 -0800431 if (skb_checksum_complete(skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800432 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700433 kfree_skb(skb);
434 return 0;
435 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436 }
437
438 rawv6_rcv_skb(sk, skb);
439 return 0;
440}
441
442
443/*
444 * This should be easy, if there is something there
445 * we return it, otherwise we block.
446 */
447
448static int rawv6_recvmsg(struct kiocb *iocb, struct sock *sk,
449 struct msghdr *msg, size_t len,
450 int noblock, int flags, int *addr_len)
451{
452 struct ipv6_pinfo *np = inet6_sk(sk);
453 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)msg->msg_name;
454 struct sk_buff *skb;
455 size_t copied;
456 int err;
457
458 if (flags & MSG_OOB)
459 return -EOPNOTSUPP;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900460
461 if (addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700462 *addr_len=sizeof(*sin6);
463
464 if (flags & MSG_ERRQUEUE)
465 return ipv6_recv_error(sk, msg, len);
466
467 skb = skb_recv_datagram(sk, flags, noblock, &err);
468 if (!skb)
469 goto out;
470
471 copied = skb->len;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900472 if (copied > len) {
473 copied = len;
474 msg->msg_flags |= MSG_TRUNC;
475 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700476
Herbert Xu60476372007-04-09 11:59:39 -0700477 if (skb_csum_unnecessary(skb)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700478 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
479 } else if (msg->msg_flags&MSG_TRUNC) {
Herbert Xufb286bb2005-11-10 13:01:24 -0800480 if (__skb_checksum_complete(skb))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700481 goto csum_copy_err;
482 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
483 } else {
484 err = skb_copy_and_csum_datagram_iovec(skb, 0, msg->msg_iov);
485 if (err == -EINVAL)
486 goto csum_copy_err;
487 }
488 if (err)
489 goto out_free;
490
491 /* Copy the address. */
492 if (sin6) {
493 sin6->sin6_family = AF_INET6;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700494 sin6->sin6_port = 0;
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700495 ipv6_addr_copy(&sin6->sin6_addr, &ipv6_hdr(skb)->saddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700496 sin6->sin6_flowinfo = 0;
497 sin6->sin6_scope_id = 0;
498 if (ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL)
499 sin6->sin6_scope_id = IP6CB(skb)->iif;
500 }
501
502 sock_recv_timestamp(msg, sk, skb);
503
504 if (np->rxopt.all)
505 datagram_recv_ctl(sk, msg, skb);
506
507 err = copied;
508 if (flags & MSG_TRUNC)
509 err = skb->len;
510
511out_free:
512 skb_free_datagram(sk, skb);
513out:
514 return err;
515
516csum_copy_err:
Herbert Xu3305b802005-12-13 23:16:37 -0800517 skb_kill_datagram(sk, skb, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700518
519 /* Error for blocking case is chosen to masquerade
520 as some normal condition.
521 */
522 err = (flags&MSG_DONTWAIT) ? -EAGAIN : -EHOSTUNREACH;
Wang Chena92aa312007-11-13 20:31:14 -0800523 atomic_inc(&sk->sk_drops);
Herbert Xu3305b802005-12-13 23:16:37 -0800524 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700525}
526
527static int rawv6_push_pending_frames(struct sock *sk, struct flowi *fl,
Herbert Xu357b40a2005-04-19 22:30:14 -0700528 struct raw6_sock *rp)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700529{
530 struct sk_buff *skb;
531 int err = 0;
Herbert Xu357b40a2005-04-19 22:30:14 -0700532 int offset;
533 int len;
Herbert Xu679a8732005-05-03 14:24:36 -0700534 int total_len;
Al Viro868c86b2006-11-14 21:35:48 -0800535 __wsum tmp_csum;
536 __sum16 csum;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700537
538 if (!rp->checksum)
539 goto send;
540
541 if ((skb = skb_peek(&sk->sk_write_queue)) == NULL)
542 goto out;
543
Herbert Xu357b40a2005-04-19 22:30:14 -0700544 offset = rp->offset;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700545 total_len = inet_sk(sk)->cork.length - (skb_network_header(skb) -
546 skb->data);
Herbert Xu679a8732005-05-03 14:24:36 -0700547 if (offset >= total_len - 1) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700548 err = -EINVAL;
Herbert Xu357b40a2005-04-19 22:30:14 -0700549 ip6_flush_pending_frames(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700550 goto out;
551 }
552
553 /* should be check HW csum miyazawa */
554 if (skb_queue_len(&sk->sk_write_queue) == 1) {
555 /*
556 * Only one fragment on the socket.
557 */
558 tmp_csum = skb->csum;
559 } else {
Herbert Xu357b40a2005-04-19 22:30:14 -0700560 struct sk_buff *csum_skb = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700561 tmp_csum = 0;
562
563 skb_queue_walk(&sk->sk_write_queue, skb) {
564 tmp_csum = csum_add(tmp_csum, skb->csum);
Herbert Xu357b40a2005-04-19 22:30:14 -0700565
566 if (csum_skb)
567 continue;
568
Arnaldo Carvalho de Meloea2ae172007-04-25 17:55:53 -0700569 len = skb->len - skb_transport_offset(skb);
Herbert Xu357b40a2005-04-19 22:30:14 -0700570 if (offset >= len) {
571 offset -= len;
572 continue;
573 }
574
575 csum_skb = skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700576 }
Herbert Xu357b40a2005-04-19 22:30:14 -0700577
578 skb = csum_skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700579 }
580
Arnaldo Carvalho de Meloea2ae172007-04-25 17:55:53 -0700581 offset += skb_transport_offset(skb);
Herbert Xu357b40a2005-04-19 22:30:14 -0700582 if (skb_copy_bits(skb, offset, &csum, 2))
583 BUG();
584
Linus Torvalds1da177e2005-04-16 15:20:36 -0700585 /* in case cksum was not initialized */
Herbert Xu357b40a2005-04-19 22:30:14 -0700586 if (unlikely(csum))
Al Viro5f92a732006-11-14 21:36:54 -0800587 tmp_csum = csum_sub(tmp_csum, csum_unfold(csum));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588
Al Viro868c86b2006-11-14 21:35:48 -0800589 csum = csum_ipv6_magic(&fl->fl6_src,
Herbert Xu357b40a2005-04-19 22:30:14 -0700590 &fl->fl6_dst,
Herbert Xu679a8732005-05-03 14:24:36 -0700591 total_len, fl->proto, tmp_csum);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700592
Al Virof6ab0282006-11-16 02:36:50 -0800593 if (csum == 0 && fl->proto == IPPROTO_UDP)
594 csum = CSUM_MANGLED_0;
Herbert Xu357b40a2005-04-19 22:30:14 -0700595
Herbert Xu357b40a2005-04-19 22:30:14 -0700596 if (skb_store_bits(skb, offset, &csum, 2))
597 BUG();
598
Linus Torvalds1da177e2005-04-16 15:20:36 -0700599send:
600 err = ip6_push_pending_frames(sk);
601out:
602 return err;
603}
604
605static int rawv6_send_hdrinc(struct sock *sk, void *from, int length,
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900606 struct flowi *fl, struct rt6_info *rt,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700607 unsigned int flags)
608{
Herbert Xu3320da82005-04-19 22:32:22 -0700609 struct ipv6_pinfo *np = inet6_sk(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 struct ipv6hdr *iph;
611 struct sk_buff *skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700612 int err;
613
614 if (length > rt->u.dst.dev->mtu) {
615 ipv6_local_error(sk, EMSGSIZE, fl, rt->u.dst.dev->mtu);
616 return -EMSGSIZE;
617 }
618 if (flags&MSG_PROBE)
619 goto out;
620
Johannes Bergf5184d22008-05-12 20:48:31 -0700621 skb = sock_alloc_send_skb(sk,
622 length + LL_ALLOCATED_SPACE(rt->u.dst.dev) + 15,
623 flags & MSG_DONTWAIT, &err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700624 if (skb == NULL)
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900625 goto error;
Johannes Bergf5184d22008-05-12 20:48:31 -0700626 skb_reserve(skb, LL_RESERVED_SPACE(rt->u.dst.dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700627
628 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800629 skb->mark = sk->sk_mark;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700630 skb->dst = dst_clone(&rt->u.dst);
631
Arnaldo Carvalho de Melo1ced98e2007-03-10 19:57:15 -0300632 skb_put(skb, length);
633 skb_reset_network_header(skb);
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700634 iph = ipv6_hdr(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700635
636 skb->ip_summed = CHECKSUM_NONE;
637
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700638 skb->transport_header = skb->network_header;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 err = memcpy_fromiovecend((void *)iph, from, 0, length);
640 if (err)
641 goto error_fault;
642
YOSHIFUJI Hideakia11d2062006-11-04 20:11:37 +0900643 IP6_INC_STATS(rt->rt6i_idev, IPSTATS_MIB_OUTREQUESTS);
Patrick McHardy6e23ae22007-11-19 18:53:30 -0800644 err = NF_HOOK(PF_INET6, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700645 dst_output);
646 if (err > 0)
Herbert Xu3320da82005-04-19 22:32:22 -0700647 err = np->recverr ? net_xmit_errno(err) : 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700648 if (err)
649 goto error;
650out:
651 return 0;
652
653error_fault:
654 err = -EFAULT;
655 kfree_skb(skb);
656error:
YOSHIFUJI Hideakia11d2062006-11-04 20:11:37 +0900657 IP6_INC_STATS(rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900658 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659}
660
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800661static int rawv6_probe_proto_opt(struct flowi *fl, struct msghdr *msg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700662{
663 struct iovec *iov;
664 u8 __user *type = NULL;
665 u8 __user *code = NULL;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700666 u8 len = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700667 int probed = 0;
668 int i;
669
670 if (!msg->msg_iov)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800671 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700672
673 for (i = 0; i < msg->msg_iovlen; i++) {
674 iov = &msg->msg_iov[i];
675 if (!iov)
676 continue;
677
678 switch (fl->proto) {
679 case IPPROTO_ICMPV6:
680 /* check if one-byte field is readable or not. */
681 if (iov->iov_base && iov->iov_len < 1)
682 break;
683
684 if (!type) {
685 type = iov->iov_base;
686 /* check if code field is readable or not. */
687 if (iov->iov_len > 1)
688 code = type + 1;
689 } else if (!code)
690 code = iov->iov_base;
691
692 if (type && code) {
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800693 if (get_user(fl->fl_icmp_type, type) ||
694 get_user(fl->fl_icmp_code, code))
695 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700696 probed = 1;
697 }
698 break;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700699 case IPPROTO_MH:
700 if (iov->iov_base && iov->iov_len < 1)
701 break;
702 /* check if type field is readable or not. */
703 if (iov->iov_len > 2 - len) {
704 u8 __user *p = iov->iov_base;
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800705 if (get_user(fl->fl_mh_type, &p[2 - len]))
706 return -EFAULT;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700707 probed = 1;
708 } else
709 len += iov->iov_len;
710
711 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712 default:
713 probed = 1;
714 break;
715 }
716 if (probed)
717 break;
718 }
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800719 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700720}
721
722static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
723 struct msghdr *msg, size_t len)
724{
725 struct ipv6_txoptions opt_space;
726 struct sockaddr_in6 * sin6 = (struct sockaddr_in6 *) msg->msg_name;
727 struct in6_addr *daddr, *final_p = NULL, final;
728 struct inet_sock *inet = inet_sk(sk);
729 struct ipv6_pinfo *np = inet6_sk(sk);
730 struct raw6_sock *rp = raw6_sk(sk);
731 struct ipv6_txoptions *opt = NULL;
732 struct ip6_flowlabel *flowlabel = NULL;
733 struct dst_entry *dst = NULL;
734 struct flowi fl;
735 int addr_len = msg->msg_namelen;
736 int hlimit = -1;
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900737 int tclass = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700738 u16 proto;
739 int err;
740
741 /* Rough check on arithmetic overflow,
YOSHIFUJI Hideakib59e1392007-03-30 14:45:35 -0700742 better check is made in ip6_append_data().
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743 */
YOSHIFUJI Hideakib59e1392007-03-30 14:45:35 -0700744 if (len > INT_MAX)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700745 return -EMSGSIZE;
746
747 /* Mirror BSD error message compatibility */
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900748 if (msg->msg_flags & MSG_OOB)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749 return -EOPNOTSUPP;
750
751 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900752 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700753 */
754 memset(&fl, 0, sizeof(fl));
755
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800756 fl.mark = sk->sk_mark;
757
Linus Torvalds1da177e2005-04-16 15:20:36 -0700758 if (sin6) {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900759 if (addr_len < SIN6_LEN_RFC2133)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700760 return -EINVAL;
761
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900762 if (sin6->sin6_family && sin6->sin6_family != AF_INET6)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763 return(-EAFNOSUPPORT);
764
765 /* port is the proto value [0..255] carried in nexthdr */
766 proto = ntohs(sin6->sin6_port);
767
768 if (!proto)
769 proto = inet->num;
770 else if (proto != inet->num)
771 return(-EINVAL);
772
773 if (proto > 255)
774 return(-EINVAL);
775
776 daddr = &sin6->sin6_addr;
777 if (np->sndflow) {
778 fl.fl6_flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
779 if (fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) {
780 flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
781 if (flowlabel == NULL)
782 return -EINVAL;
783 daddr = &flowlabel->dst;
784 }
785 }
786
787 /*
788 * Otherwise it will be difficult to maintain
789 * sk->sk_dst_cache.
790 */
791 if (sk->sk_state == TCP_ESTABLISHED &&
792 ipv6_addr_equal(daddr, &np->daddr))
793 daddr = &np->daddr;
794
795 if (addr_len >= sizeof(struct sockaddr_in6) &&
796 sin6->sin6_scope_id &&
797 ipv6_addr_type(daddr)&IPV6_ADDR_LINKLOCAL)
798 fl.oif = sin6->sin6_scope_id;
799 } else {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900800 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700801 return -EDESTADDRREQ;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900802
Linus Torvalds1da177e2005-04-16 15:20:36 -0700803 proto = inet->num;
804 daddr = &np->daddr;
805 fl.fl6_flowlabel = np->flow_label;
806 }
807
Linus Torvalds1da177e2005-04-16 15:20:36 -0700808 if (fl.oif == 0)
809 fl.oif = sk->sk_bound_dev_if;
810
811 if (msg->msg_controllen) {
812 opt = &opt_space;
813 memset(opt, 0, sizeof(struct ipv6_txoptions));
814 opt->tot_len = sizeof(struct ipv6_txoptions);
815
YOSHIFUJI Hideaki91e19082008-06-04 13:02:49 +0900816 err = datagram_send_ctl(sock_net(sk), msg, &fl, opt, &hlimit, &tclass);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817 if (err < 0) {
818 fl6_sock_release(flowlabel);
819 return err;
820 }
821 if ((fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
822 flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
823 if (flowlabel == NULL)
824 return -EINVAL;
825 }
826 if (!(opt->opt_nflen|opt->opt_flen))
827 opt = NULL;
828 }
829 if (opt == NULL)
830 opt = np->opt;
YOSHIFUJI Hideakidf9890c2005-11-20 12:23:18 +0900831 if (flowlabel)
832 opt = fl6_merge_options(&opt_space, flowlabel, opt);
833 opt = ipv6_fixup_options(&opt_space, opt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700834
835 fl.proto = proto;
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800836 err = rawv6_probe_proto_opt(&fl, msg);
837 if (err)
838 goto out;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900839
Brian Haley876c7f42008-04-11 00:38:24 -0400840 if (!ipv6_addr_any(daddr))
841 ipv6_addr_copy(&fl.fl6_dst, daddr);
842 else
843 fl.fl6_dst.s6_addr[15] = 0x1; /* :: means loopback (BSD'ism) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700844 if (ipv6_addr_any(&fl.fl6_src) && !ipv6_addr_any(&np->saddr))
845 ipv6_addr_copy(&fl.fl6_src, &np->saddr);
846
847 /* merge ip6_build_xmit from ip6_output */
848 if (opt && opt->srcrt) {
849 struct rt0_hdr *rt0 = (struct rt0_hdr *) opt->srcrt;
850 ipv6_addr_copy(&final, &fl.fl6_dst);
851 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
852 final_p = &final;
853 }
854
855 if (!fl.oif && ipv6_addr_is_multicast(&fl.fl6_dst))
856 fl.oif = np->mcast_oif;
Venkat Yekkiralabeb8d132006-08-04 23:12:42 -0700857 security_sk_classify_flow(sk, &fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700858
859 err = ip6_dst_lookup(sk, &dst, &fl);
860 if (err)
861 goto out;
862 if (final_p)
863 ipv6_addr_copy(&fl.fl6_dst, final_p);
864
Herbert Xubb728452007-12-12 18:48:58 -0800865 if ((err = __xfrm_lookup(&dst, &fl, sk, XFRM_LOOKUP_WAIT)) < 0) {
David S. Miller14e50e52007-05-24 18:17:54 -0700866 if (err == -EREMOTE)
867 err = ip6_dst_blackhole(sk, &dst, &fl);
868 if (err < 0)
869 goto out;
870 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700871
872 if (hlimit < 0) {
873 if (ipv6_addr_is_multicast(&fl.fl6_dst))
874 hlimit = np->mcast_hops;
875 else
876 hlimit = np->hop_limit;
877 if (hlimit < 0)
YOSHIFUJI Hideaki6b75d092008-03-10 06:00:30 -0400878 hlimit = ip6_dst_hoplimit(dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700879 }
880
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900881 if (tclass < 0) {
YOSHIFUJI Hideakie012d512006-09-13 20:01:28 -0700882 tclass = np->tclass;
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900883 if (tclass < 0)
884 tclass = 0;
885 }
886
Linus Torvalds1da177e2005-04-16 15:20:36 -0700887 if (msg->msg_flags&MSG_CONFIRM)
888 goto do_confirm;
889
890back_from_confirm:
891 if (inet->hdrincl) {
892 err = rawv6_send_hdrinc(sk, msg->msg_iov, len, &fl, (struct rt6_info*)dst, msg->msg_flags);
893 } else {
894 lock_sock(sk);
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900895 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov,
896 len, 0, hlimit, tclass, opt, &fl, (struct rt6_info*)dst,
897 msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898
899 if (err)
900 ip6_flush_pending_frames(sk);
901 else if (!(msg->msg_flags & MSG_MORE))
Herbert Xu357b40a2005-04-19 22:30:14 -0700902 err = rawv6_push_pending_frames(sk, &fl, rp);
YOSHIFUJI Hideaki3ef9d942007-09-14 16:45:40 -0700903 release_sock(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904 }
905done:
Nicolas DICHTEL6d3e85e2006-02-13 15:56:13 -0800906 dst_release(dst);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900907out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908 fl6_sock_release(flowlabel);
909 return err<0?err:len;
910do_confirm:
911 dst_confirm(dst);
912 if (!(msg->msg_flags & MSG_PROBE) || len)
913 goto back_from_confirm;
914 err = 0;
915 goto done;
916}
917
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900918static int rawv6_seticmpfilter(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700919 char __user *optval, int optlen)
920{
921 switch (optname) {
922 case ICMPV6_FILTER:
923 if (optlen > sizeof(struct icmp6_filter))
924 optlen = sizeof(struct icmp6_filter);
925 if (copy_from_user(&raw6_sk(sk)->filter, optval, optlen))
926 return -EFAULT;
927 return 0;
928 default:
929 return -ENOPROTOOPT;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700930 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700931
932 return 0;
933}
934
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900935static int rawv6_geticmpfilter(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936 char __user *optval, int __user *optlen)
937{
938 int len;
939
940 switch (optname) {
941 case ICMPV6_FILTER:
942 if (get_user(len, optlen))
943 return -EFAULT;
944 if (len < 0)
945 return -EINVAL;
946 if (len > sizeof(struct icmp6_filter))
947 len = sizeof(struct icmp6_filter);
948 if (put_user(len, optlen))
949 return -EFAULT;
950 if (copy_to_user(optval, &raw6_sk(sk)->filter, len))
951 return -EFAULT;
952 return 0;
953 default:
954 return -ENOPROTOOPT;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700955 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700956
957 return 0;
958}
959
960
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800961static int do_rawv6_setsockopt(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700962 char __user *optval, int optlen)
963{
964 struct raw6_sock *rp = raw6_sk(sk);
965 int val;
966
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900967 if (get_user(val, (int __user *)optval))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968 return -EFAULT;
969
970 switch (optname) {
971 case IPV6_CHECKSUM:
YOSHIFUJI Hideaki1a98d052008-04-24 21:30:38 -0700972 if (inet_sk(sk)->num == IPPROTO_ICMPV6 &&
973 level == IPPROTO_IPV6) {
974 /*
975 * RFC3542 tells that IPV6_CHECKSUM socket
976 * option in the IPPROTO_IPV6 level is not
977 * allowed on ICMPv6 sockets.
978 * If you want to set it, use IPPROTO_RAW
979 * level IPV6_CHECKSUM socket option
980 * (Linux extension).
981 */
982 return -EINVAL;
983 }
984
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985 /* You may get strange result with a positive odd offset;
986 RFC2292bis agrees with me. */
987 if (val > 0 && (val&1))
988 return(-EINVAL);
989 if (val < 0) {
990 rp->checksum = 0;
991 } else {
992 rp->checksum = 1;
993 rp->offset = val;
994 }
995
996 return 0;
997 break;
998
999 default:
1000 return(-ENOPROTOOPT);
1001 }
1002}
1003
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001004static int rawv6_setsockopt(struct sock *sk, int level, int optname,
1005 char __user *optval, int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007 switch(level) {
1008 case SOL_RAW:
1009 break;
1010
1011 case SOL_ICMPV6:
1012 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1013 return -EOPNOTSUPP;
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001014 return rawv6_seticmpfilter(sk, level, optname, optval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001015 optlen);
1016 case SOL_IPV6:
1017 if (optname == IPV6_CHECKSUM)
1018 break;
1019 default:
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001020 return ipv6_setsockopt(sk, level, optname, optval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001021 optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001022 }
1023
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001024 return do_rawv6_setsockopt(sk, level, optname, optval, optlen);
1025}
1026
1027#ifdef CONFIG_COMPAT
1028static int compat_rawv6_setsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001029 char __user *optval, int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001030{
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001031 switch (level) {
1032 case SOL_RAW:
1033 break;
1034 case SOL_ICMPV6:
1035 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1036 return -EOPNOTSUPP;
1037 return rawv6_seticmpfilter(sk, level, optname, optval, optlen);
1038 case SOL_IPV6:
1039 if (optname == IPV6_CHECKSUM)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001040 break;
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001041 default:
1042 return compat_ipv6_setsockopt(sk, level, optname,
1043 optval, optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001044 }
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001045 return do_rawv6_setsockopt(sk, level, optname, optval, optlen);
1046}
1047#endif
1048
1049static int do_rawv6_getsockopt(struct sock *sk, int level, int optname,
1050 char __user *optval, int __user *optlen)
1051{
1052 struct raw6_sock *rp = raw6_sk(sk);
1053 int val, len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001054
1055 if (get_user(len,optlen))
1056 return -EFAULT;
1057
1058 switch (optname) {
1059 case IPV6_CHECKSUM:
YOSHIFUJI Hideaki1a98d052008-04-24 21:30:38 -07001060 /*
1061 * We allow getsockopt() for IPPROTO_IPV6-level
1062 * IPV6_CHECKSUM socket option on ICMPv6 sockets
1063 * since RFC3542 is silent about it.
1064 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001065 if (rp->checksum == 0)
1066 val = -1;
1067 else
1068 val = rp->offset;
1069 break;
1070
1071 default:
1072 return -ENOPROTOOPT;
1073 }
1074
1075 len = min_t(unsigned int, sizeof(int), len);
1076
1077 if (put_user(len, optlen))
1078 return -EFAULT;
1079 if (copy_to_user(optval,&val,len))
1080 return -EFAULT;
1081 return 0;
1082}
1083
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001084static int rawv6_getsockopt(struct sock *sk, int level, int optname,
1085 char __user *optval, int __user *optlen)
1086{
1087 switch(level) {
1088 case SOL_RAW:
1089 break;
1090
1091 case SOL_ICMPV6:
1092 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1093 return -EOPNOTSUPP;
1094 return rawv6_geticmpfilter(sk, level, optname, optval,
1095 optlen);
1096 case SOL_IPV6:
1097 if (optname == IPV6_CHECKSUM)
1098 break;
1099 default:
1100 return ipv6_getsockopt(sk, level, optname, optval,
1101 optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001102 }
1103
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001104 return do_rawv6_getsockopt(sk, level, optname, optval, optlen);
1105}
1106
1107#ifdef CONFIG_COMPAT
1108static int compat_rawv6_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001109 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001110{
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001111 switch (level) {
1112 case SOL_RAW:
1113 break;
1114 case SOL_ICMPV6:
1115 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1116 return -EOPNOTSUPP;
1117 return rawv6_geticmpfilter(sk, level, optname, optval, optlen);
1118 case SOL_IPV6:
1119 if (optname == IPV6_CHECKSUM)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001120 break;
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001121 default:
1122 return compat_ipv6_getsockopt(sk, level, optname,
1123 optval, optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001124 }
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001125 return do_rawv6_getsockopt(sk, level, optname, optval, optlen);
1126}
1127#endif
1128
Linus Torvalds1da177e2005-04-16 15:20:36 -07001129static int rawv6_ioctl(struct sock *sk, int cmd, unsigned long arg)
1130{
1131 switch(cmd) {
1132 case SIOCOUTQ:
1133 {
1134 int amount = atomic_read(&sk->sk_wmem_alloc);
1135 return put_user(amount, (int __user *)arg);
1136 }
1137 case SIOCINQ:
1138 {
1139 struct sk_buff *skb;
1140 int amount = 0;
1141
Herbert Xue0f9f852005-06-18 22:56:18 -07001142 spin_lock_bh(&sk->sk_receive_queue.lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001143 skb = skb_peek(&sk->sk_receive_queue);
1144 if (skb != NULL)
Arnaldo Carvalho de Melo27a884d2007-04-19 20:29:13 -07001145 amount = skb->tail - skb->transport_header;
Herbert Xue0f9f852005-06-18 22:56:18 -07001146 spin_unlock_bh(&sk->sk_receive_queue.lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001147 return put_user(amount, (int __user *)arg);
1148 }
1149
1150 default:
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001151#ifdef CONFIG_IPV6_MROUTE
1152 return ip6mr_ioctl(sk, cmd, (void __user *)arg);
1153#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001154 return -ENOIOCTLCMD;
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001155#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001156 }
1157}
1158
1159static void rawv6_close(struct sock *sk, long timeout)
1160{
1161 if (inet_sk(sk)->num == IPPROTO_RAW)
1162 ip6_ra_control(sk, -1, NULL);
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001163 ip6mr_sk_done(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001164 sk_common_release(sk);
1165}
1166
1167static int rawv6_init_sk(struct sock *sk)
1168{
Masahide NAKAMURAf48d5ff2007-02-07 00:07:39 -08001169 struct raw6_sock *rp = raw6_sk(sk);
1170
1171 switch (inet_sk(sk)->num) {
1172 case IPPROTO_ICMPV6:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001173 rp->checksum = 1;
1174 rp->offset = 2;
Masahide NAKAMURAf48d5ff2007-02-07 00:07:39 -08001175 break;
1176 case IPPROTO_MH:
1177 rp->checksum = 1;
1178 rp->offset = 4;
1179 break;
1180 default:
1181 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001182 }
1183 return(0);
1184}
1185
1186struct proto rawv6_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001187 .name = "RAWv6",
1188 .owner = THIS_MODULE,
1189 .close = rawv6_close,
1190 .connect = ip6_datagram_connect,
1191 .disconnect = udp_disconnect,
1192 .ioctl = rawv6_ioctl,
1193 .init = rawv6_init_sk,
1194 .destroy = inet6_destroy_sock,
1195 .setsockopt = rawv6_setsockopt,
1196 .getsockopt = rawv6_getsockopt,
1197 .sendmsg = rawv6_sendmsg,
1198 .recvmsg = rawv6_recvmsg,
1199 .bind = rawv6_bind,
1200 .backlog_rcv = rawv6_rcv_skb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -07001201 .hash = raw_hash_sk,
1202 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001203 .obj_size = sizeof(struct raw6_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -07001204 .h.raw_hash = &raw_v6_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001205#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001206 .compat_setsockopt = compat_rawv6_setsockopt,
1207 .compat_getsockopt = compat_rawv6_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001208#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001209};
1210
1211#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212static void raw6_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
1213{
1214 struct ipv6_pinfo *np = inet6_sk(sp);
1215 struct in6_addr *dest, *src;
1216 __u16 destp, srcp;
1217
1218 dest = &np->daddr;
1219 src = &np->rcv_saddr;
1220 destp = 0;
1221 srcp = inet_sk(sp)->num;
1222 seq_printf(seq,
1223 "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
Wang Chena92aa312007-11-13 20:31:14 -08001224 "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
Linus Torvalds1da177e2005-04-16 15:20:36 -07001225 i,
1226 src->s6_addr32[0], src->s6_addr32[1],
1227 src->s6_addr32[2], src->s6_addr32[3], srcp,
1228 dest->s6_addr32[0], dest->s6_addr32[1],
1229 dest->s6_addr32[2], dest->s6_addr32[3], destp,
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001230 sp->sk_state,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001231 atomic_read(&sp->sk_wmem_alloc),
1232 atomic_read(&sp->sk_rmem_alloc),
1233 0, 0L, 0,
1234 sock_i_uid(sp), 0,
1235 sock_i_ino(sp),
Wang Chena92aa312007-11-13 20:31:14 -08001236 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001237}
1238
1239static int raw6_seq_show(struct seq_file *seq, void *v)
1240{
1241 if (v == SEQ_START_TOKEN)
1242 seq_printf(seq,
1243 " sl "
1244 "local_address "
1245 "remote_address "
1246 "st tx_queue rx_queue tr tm->when retrnsmt"
Wang Chena92aa312007-11-13 20:31:14 -08001247 " uid timeout inode drops\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -07001248 else
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001249 raw6_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001250 return 0;
1251}
1252
Philippe De Muyter56b3d972007-07-10 23:07:31 -07001253static const struct seq_operations raw6_seq_ops = {
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001254 .start = raw_seq_start,
1255 .next = raw_seq_next,
1256 .stop = raw_seq_stop,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001257 .show = raw6_seq_show,
1258};
1259
1260static int raw6_seq_open(struct inode *inode, struct file *file)
1261{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001262 return raw_seq_open(inode, file, &raw_v6_hashinfo, &raw6_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001263}
1264
Arjan van de Ven9a321442007-02-12 00:55:35 -08001265static const struct file_operations raw6_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001266 .owner = THIS_MODULE,
1267 .open = raw6_seq_open,
1268 .read = seq_read,
1269 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001270 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001271};
1272
Pavel Emelyanova308da12008-01-14 05:36:50 -08001273static int raw6_init_net(struct net *net)
1274{
1275 if (!proc_net_fops_create(net, "raw6", S_IRUGO, &raw6_seq_fops))
1276 return -ENOMEM;
1277
1278 return 0;
1279}
1280
1281static void raw6_exit_net(struct net *net)
1282{
1283 proc_net_remove(net, "raw6");
1284}
1285
1286static struct pernet_operations raw6_net_ops = {
1287 .init = raw6_init_net,
1288 .exit = raw6_exit_net,
1289};
1290
Linus Torvalds1da177e2005-04-16 15:20:36 -07001291int __init raw6_proc_init(void)
1292{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001293 return register_pernet_subsys(&raw6_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001294}
1295
1296void raw6_proc_exit(void)
1297{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001298 unregister_pernet_subsys(&raw6_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299}
1300#endif /* CONFIG_PROC_FS */
Daniel Lezcano7f4e4862007-12-11 02:25:35 -08001301
1302/* Same as inet6_dgram_ops, sans udp_poll. */
1303static const struct proto_ops inet6_sockraw_ops = {
1304 .family = PF_INET6,
1305 .owner = THIS_MODULE,
1306 .release = inet6_release,
1307 .bind = inet6_bind,
1308 .connect = inet_dgram_connect, /* ok */
1309 .socketpair = sock_no_socketpair, /* a do nothing */
1310 .accept = sock_no_accept, /* a do nothing */
1311 .getname = inet6_getname,
1312 .poll = datagram_poll, /* ok */
1313 .ioctl = inet6_ioctl, /* must change */
1314 .listen = sock_no_listen, /* ok */
1315 .shutdown = inet_shutdown, /* ok */
1316 .setsockopt = sock_common_setsockopt, /* ok */
1317 .getsockopt = sock_common_getsockopt, /* ok */
1318 .sendmsg = inet_sendmsg, /* ok */
1319 .recvmsg = sock_common_recvmsg, /* ok */
1320 .mmap = sock_no_mmap,
1321 .sendpage = sock_no_sendpage,
1322#ifdef CONFIG_COMPAT
1323 .compat_setsockopt = compat_sock_common_setsockopt,
1324 .compat_getsockopt = compat_sock_common_getsockopt,
1325#endif
1326};
1327
1328static struct inet_protosw rawv6_protosw = {
1329 .type = SOCK_RAW,
1330 .protocol = IPPROTO_IP, /* wild card */
1331 .prot = &rawv6_prot,
1332 .ops = &inet6_sockraw_ops,
1333 .capability = CAP_NET_RAW,
1334 .no_check = UDP_CSUM_DEFAULT,
1335 .flags = INET_PROTOSW_REUSE,
1336};
1337
1338int __init rawv6_init(void)
1339{
1340 int ret;
1341
1342 ret = inet6_register_protosw(&rawv6_protosw);
1343 if (ret)
1344 goto out;
1345out:
1346 return ret;
1347}
1348
Daniel Lezcano09f77092007-12-13 05:34:58 -08001349void rawv6_exit(void)
Daniel Lezcano7f4e4862007-12-11 02:25:35 -08001350{
1351 inet6_unregister_protosw(&rawv6_protosw);
1352}