blob: 01d47674f7e5c802214565108e14a820ef7963fc [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 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070010 * Fixes:
11 * Hideaki YOSHIFUJI : sin6_scope_id support
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +090012 * YOSHIFUJI,H.@USAGI : raw checksum (RFC2292(bis) compliance)
Linus Torvalds1da177e2005-04-16 15:20:36 -070013 * Kazunori MIYAZAWA @USAGI: change process style to use ip6_append_data
14 *
15 * This program is free software; you can redistribute it and/or
16 * modify it under the terms of the GNU General Public License
17 * as published by the Free Software Foundation; either version
18 * 2 of the License, or (at your option) any later version.
19 */
20
21#include <linux/errno.h>
22#include <linux/types.h>
23#include <linux/socket.h>
24#include <linux/sockios.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070025#include <linux/net.h>
26#include <linux/in6.h>
27#include <linux/netdevice.h>
28#include <linux/if_arp.h>
29#include <linux/icmpv6.h>
30#include <linux/netfilter.h>
31#include <linux/netfilter_ipv6.h>
Herbert Xu3305b802005-12-13 23:16:37 -080032#include <linux/skbuff.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070033#include <asm/uaccess.h>
34#include <asm/ioctls.h>
35
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020036#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037#include <net/ip.h>
38#include <net/sock.h>
39#include <net/snmp.h>
40
41#include <net/ipv6.h>
42#include <net/ndisc.h>
43#include <net/protocol.h>
44#include <net/ip6_route.h>
45#include <net/ip6_checksum.h>
46#include <net/addrconf.h>
47#include <net/transp_v6.h>
48#include <net/udp.h>
49#include <net/inet_common.h>
Arnaldo Carvalho de Meloc752f072005-08-09 20:08:28 -070050#include <net/tcp_states.h>
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -070051#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -070052#include <net/mip6.h>
53#endif
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +090054#include <linux/mroute6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070055
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080056#include <net/raw.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070057#include <net/rawv6.h>
58#include <net/xfrm.h>
59
60#include <linux/proc_fs.h>
61#include <linux/seq_file.h>
62
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080063static struct raw_hashinfo raw_v6_hashinfo = {
Robert P. J. Day938b93a2008-03-18 00:59:23 -070064 .lock = __RW_LOCK_UNLOCKED(raw_v6_hashinfo.lock),
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080065};
Linus Torvalds1da177e2005-04-16 15:20:36 -070066
Pavel Emelyanovbe185882008-01-14 05:35:31 -080067static struct sock *__raw_v6_lookup(struct net *net, struct sock *sk,
68 unsigned short num, struct in6_addr *loc_addr,
69 struct in6_addr *rmt_addr, int dif)
Linus Torvalds1da177e2005-04-16 15:20:36 -070070{
71 struct hlist_node *node;
72 int is_multicast = ipv6_addr_is_multicast(loc_addr);
73
74 sk_for_each_from(sk, node)
75 if (inet_sk(sk)->num == num) {
76 struct ipv6_pinfo *np = inet6_sk(sk);
77
YOSHIFUJI Hideaki878628f2008-03-26 03:57:35 +090078 if (!net_eq(sock_net(sk), net))
Pavel Emelyanovbe185882008-01-14 05:35:31 -080079 continue;
80
Linus Torvalds1da177e2005-04-16 15:20:36 -070081 if (!ipv6_addr_any(&np->daddr) &&
82 !ipv6_addr_equal(&np->daddr, rmt_addr))
83 continue;
84
Andrew McDonald0bd1b592005-08-09 19:44:42 -070085 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
86 continue;
87
Linus Torvalds1da177e2005-04-16 15:20:36 -070088 if (!ipv6_addr_any(&np->rcv_saddr)) {
89 if (ipv6_addr_equal(&np->rcv_saddr, loc_addr))
90 goto found;
91 if (is_multicast &&
92 inet6_mc_check(sk, loc_addr, rmt_addr))
93 goto found;
94 continue;
95 }
96 goto found;
97 }
98 sk = NULL;
99found:
100 return sk;
101}
102
103/*
104 * 0 - deliver
105 * 1 - block
106 */
107static __inline__ int icmpv6_filter(struct sock *sk, struct sk_buff *skb)
108{
109 struct icmp6hdr *icmph;
110 struct raw6_sock *rp = raw6_sk(sk);
111
112 if (pskb_may_pull(skb, sizeof(struct icmp6hdr))) {
113 __u32 *data = &rp->filter.data[0];
114 int bit_nr;
115
116 icmph = (struct icmp6hdr *) skb->data;
117 bit_nr = icmph->icmp6_type;
118
119 return (data[bit_nr >> 5] & (1 << (bit_nr & 31))) != 0;
120 }
121 return 0;
122}
123
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700124#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
125static int (*mh_filter)(struct sock *sock, struct sk_buff *skb);
126
127int rawv6_mh_filter_register(int (*filter)(struct sock *sock,
128 struct sk_buff *skb))
129{
130 rcu_assign_pointer(mh_filter, filter);
131 return 0;
132}
133EXPORT_SYMBOL(rawv6_mh_filter_register);
134
135int rawv6_mh_filter_unregister(int (*filter)(struct sock *sock,
136 struct sk_buff *skb))
137{
138 rcu_assign_pointer(mh_filter, NULL);
139 synchronize_rcu();
140 return 0;
141}
142EXPORT_SYMBOL(rawv6_mh_filter_unregister);
143
144#endif
145
Linus Torvalds1da177e2005-04-16 15:20:36 -0700146/*
147 * demultiplex raw sockets.
148 * (should consider queueing the skb in the sock receive_queue
149 * without calling rawv6.c)
150 *
151 * Caller owns SKB so we must make clones.
152 */
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800153static int ipv6_raw_deliver(struct sk_buff *skb, int nexthdr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154{
155 struct in6_addr *saddr;
156 struct in6_addr *daddr;
157 struct sock *sk;
Patrick McHardyd13964f2005-08-09 19:45:02 -0700158 int delivered = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700159 __u8 hash;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800160 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700161
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700162 saddr = &ipv6_hdr(skb)->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700163 daddr = saddr + 1;
164
165 hash = nexthdr & (MAX_INET_PROTOS - 1);
166
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800167 read_lock(&raw_v6_hashinfo.lock);
168 sk = sk_head(&raw_v6_hashinfo.ht[hash]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170 if (sk == NULL)
171 goto out;
172
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900173 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800174 sk = __raw_v6_lookup(net, sk, nexthdr, daddr, saddr, IP6CB(skb)->iif);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700175
176 while (sk) {
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700177 int filtered;
178
Patrick McHardyd13964f2005-08-09 19:45:02 -0700179 delivered = 1;
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700180 switch (nexthdr) {
181 case IPPROTO_ICMPV6:
182 filtered = icmpv6_filter(sk, skb);
183 break;
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700184
185#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700186 case IPPROTO_MH:
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700187 {
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700188 /* XXX: To validate MH only once for each packet,
189 * this is placed here. It should be after checking
190 * xfrm policy, however it doesn't. The checking xfrm
191 * policy is placed in rawv6_rcv() because it is
192 * required for each socket.
193 */
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700194 int (*filter)(struct sock *sock, struct sk_buff *skb);
195
196 filter = rcu_dereference(mh_filter);
197 filtered = filter ? filter(sk, skb) : 0;
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700198 break;
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700199 }
Masahide NAKAMURA7be96f72006-08-23 20:35:31 -0700200#endif
201 default:
202 filtered = 0;
203 break;
204 }
205
206 if (filtered < 0)
207 break;
208 if (filtered == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700209 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
210
211 /* Not releasing hash table! */
Yasuyuki Kozakai9fb9cbb2005-11-09 16:38:16 -0800212 if (clone) {
213 nf_reset(clone);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700214 rawv6_rcv(sk, clone);
Yasuyuki Kozakai9fb9cbb2005-11-09 16:38:16 -0800215 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700216 }
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800217 sk = __raw_v6_lookup(net, sk_next(sk), nexthdr, daddr, saddr,
YOSHIFUJI Hideaki2dac4b92005-09-01 17:44:49 -0700218 IP6CB(skb)->iif);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219 }
220out:
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800221 read_unlock(&raw_v6_hashinfo.lock);
Patrick McHardyd13964f2005-08-09 19:45:02 -0700222 return delivered;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700223}
224
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800225int raw6_local_deliver(struct sk_buff *skb, int nexthdr)
226{
227 struct sock *raw_sk;
228
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800229 raw_sk = sk_head(&raw_v6_hashinfo.ht[nexthdr & (MAX_INET_PROTOS - 1)]);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800230 if (raw_sk && !ipv6_raw_deliver(skb, nexthdr))
231 raw_sk = NULL;
232
233 return raw_sk != NULL;
234}
235
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236/* This cleans up af_inet6 a bit. -DaveM */
237static int rawv6_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
238{
239 struct inet_sock *inet = inet_sk(sk);
240 struct ipv6_pinfo *np = inet6_sk(sk);
241 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
Al Viroe69a4adc2006-11-14 20:56:00 -0800242 __be32 v4addr = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243 int addr_type;
244 int err;
245
246 if (addr_len < SIN6_LEN_RFC2133)
247 return -EINVAL;
248 addr_type = ipv6_addr_type(&addr->sin6_addr);
249
250 /* Raw sockets are IPv6 only */
251 if (addr_type == IPV6_ADDR_MAPPED)
252 return(-EADDRNOTAVAIL);
253
254 lock_sock(sk);
255
256 err = -EINVAL;
257 if (sk->sk_state != TCP_CLOSE)
258 goto out;
259
260 /* Check if the address belongs to the host. */
261 if (addr_type != IPV6_ADDR_ANY) {
262 struct net_device *dev = NULL;
263
264 if (addr_type & IPV6_ADDR_LINKLOCAL) {
265 if (addr_len >= sizeof(struct sockaddr_in6) &&
266 addr->sin6_scope_id) {
267 /* Override any existing binding, if another
268 * one is supplied by user.
269 */
270 sk->sk_bound_dev_if = addr->sin6_scope_id;
271 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900272
Linus Torvalds1da177e2005-04-16 15:20:36 -0700273 /* Binding to link-local address requires an interface */
274 if (!sk->sk_bound_dev_if)
275 goto out;
276
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900277 dev = dev_get_by_index(sock_net(sk), sk->sk_bound_dev_if);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700278 if (!dev) {
279 err = -ENODEV;
280 goto out;
281 }
282 }
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900283
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284 /* ipv4 addr of the socket is invalid. Only the
285 * unspecified and mapped address have a v4 equivalent.
286 */
287 v4addr = LOOPBACK4_IPV6;
288 if (!(addr_type & IPV6_ADDR_MULTICAST)) {
289 err = -EADDRNOTAVAIL;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900290 if (!ipv6_chk_addr(sock_net(sk), &addr->sin6_addr,
Daniel Lezcanobfeade02008-01-10 22:43:18 -0800291 dev, 0)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700292 if (dev)
293 dev_put(dev);
294 goto out;
295 }
296 }
297 if (dev)
298 dev_put(dev);
299 }
300
301 inet->rcv_saddr = inet->saddr = v4addr;
302 ipv6_addr_copy(&np->rcv_saddr, &addr->sin6_addr);
303 if (!(addr_type & IPV6_ADDR_MULTICAST))
304 ipv6_addr_copy(&np->saddr, &addr->sin6_addr);
305 err = 0;
306out:
307 release_sock(sk);
308 return err;
309}
310
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800311static void rawv6_err(struct sock *sk, struct sk_buff *skb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312 struct inet6_skb_parm *opt,
Al Viro04ce6902006-11-08 00:21:01 -0800313 int type, int code, int offset, __be32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314{
315 struct inet_sock *inet = inet_sk(sk);
316 struct ipv6_pinfo *np = inet6_sk(sk);
317 int err;
318 int harderr;
319
320 /* Report error on raw socket, if:
321 1. User requested recverr.
322 2. Socket is connected (otherwise the error indication
323 is useless without recverr and error is hard.
324 */
325 if (!np->recverr && sk->sk_state != TCP_ESTABLISHED)
326 return;
327
328 harderr = icmpv6_err_convert(type, code, &err);
329 if (type == ICMPV6_PKT_TOOBIG)
330 harderr = (np->pmtudisc == IPV6_PMTUDISC_DO);
331
332 if (np->recverr) {
333 u8 *payload = skb->data;
334 if (!inet->hdrincl)
335 payload += offset;
336 ipv6_icmp_error(sk, skb, err, 0, ntohl(info), payload);
337 }
338
339 if (np->recverr || harderr) {
340 sk->sk_err = err;
341 sk->sk_error_report(sk);
342 }
343}
344
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800345void raw6_icmp_error(struct sk_buff *skb, int nexthdr,
346 int type, int code, int inner_offset, __be32 info)
347{
348 struct sock *sk;
349 int hash;
350 struct in6_addr *saddr, *daddr;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800351 struct net *net;
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800352
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800353 hash = nexthdr & (RAW_HTABLE_SIZE - 1);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800354
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800355 read_lock(&raw_v6_hashinfo.lock);
356 sk = sk_head(&raw_v6_hashinfo.ht[hash]);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800357 if (sk != NULL) {
YOSHIFUJI Hideaki05f175c2008-04-11 23:51:26 +0900358 /* Note: ipv6_hdr(skb) != skb->data */
359 struct ipv6hdr *ip6h = (struct ipv6hdr *)skb->data;
360 saddr = &ip6h->saddr;
361 daddr = &ip6h->daddr;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900362 net = dev_net(skb->dev);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800363
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800364 while ((sk = __raw_v6_lookup(net, sk, nexthdr, saddr, daddr,
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800365 IP6CB(skb)->iif))) {
366 rawv6_err(sk, skb, NULL, type, code,
367 inner_offset, info);
368 sk = sk_next(sk);
369 }
370 }
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800371 read_unlock(&raw_v6_hashinfo.lock);
Pavel Emelyanov69d6da02007-11-19 22:35:57 -0800372}
373
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374static inline int rawv6_rcv_skb(struct sock * sk, struct sk_buff * skb)
375{
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900376 if ((raw6_sk(sk)->checksum || sk->sk_filter) &&
Herbert Xufb286bb2005-11-10 13:01:24 -0800377 skb_checksum_complete(skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800378 atomic_inc(&sk->sk_drops);
Herbert Xufb286bb2005-11-10 13:01:24 -0800379 kfree_skb(skb);
380 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381 }
382
383 /* Charge it to the socket. */
384 if (sock_queue_rcv_skb(sk,skb)<0) {
Wang Chena92aa312007-11-13 20:31:14 -0800385 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386 kfree_skb(skb);
387 return 0;
388 }
389
390 return 0;
391}
392
393/*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900394 * This is next to useless...
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 * if we demultiplex in network layer we don't need the extra call
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900396 * just to queue the skb...
397 * maybe we could have the network decide upon a hint if it
Linus Torvalds1da177e2005-04-16 15:20:36 -0700398 * should call raw_rcv for demultiplexing
399 */
400int rawv6_rcv(struct sock *sk, struct sk_buff *skb)
401{
402 struct inet_sock *inet = inet_sk(sk);
403 struct raw6_sock *rp = raw6_sk(sk);
404
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900405 if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800406 atomic_inc(&sk->sk_drops);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900407 kfree_skb(skb);
408 return NET_RX_DROP;
409 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410
411 if (!rp->checksum)
412 skb->ip_summed = CHECKSUM_UNNECESSARY;
413
Patrick McHardy84fa7932006-08-29 16:44:56 -0700414 if (skb->ip_summed == CHECKSUM_COMPLETE) {
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700415 skb_postpull_rcsum(skb, skb_network_header(skb),
Arnaldo Carvalho de Melocfe1fc72007-03-16 17:26:39 -0300416 skb_network_header_len(skb));
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700417 if (!csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
418 &ipv6_hdr(skb)->daddr,
Herbert Xufb286bb2005-11-10 13:01:24 -0800419 skb->len, inet->num, skb->csum))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700420 skb->ip_summed = CHECKSUM_UNNECESSARY;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700421 }
Herbert Xu60476372007-04-09 11:59:39 -0700422 if (!skb_csum_unnecessary(skb))
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700423 skb->csum = ~csum_unfold(csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
424 &ipv6_hdr(skb)->daddr,
425 skb->len,
426 inet->num, 0));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700427
428 if (inet->hdrincl) {
Herbert Xufb286bb2005-11-10 13:01:24 -0800429 if (skb_checksum_complete(skb)) {
Wang Chena92aa312007-11-13 20:31:14 -0800430 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431 kfree_skb(skb);
432 return 0;
433 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700434 }
435
436 rawv6_rcv_skb(sk, skb);
437 return 0;
438}
439
440
441/*
442 * This should be easy, if there is something there
443 * we return it, otherwise we block.
444 */
445
446static int rawv6_recvmsg(struct kiocb *iocb, struct sock *sk,
447 struct msghdr *msg, size_t len,
448 int noblock, int flags, int *addr_len)
449{
450 struct ipv6_pinfo *np = inet6_sk(sk);
451 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)msg->msg_name;
452 struct sk_buff *skb;
453 size_t copied;
454 int err;
455
456 if (flags & MSG_OOB)
457 return -EOPNOTSUPP;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900458
459 if (addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460 *addr_len=sizeof(*sin6);
461
462 if (flags & MSG_ERRQUEUE)
463 return ipv6_recv_error(sk, msg, len);
464
465 skb = skb_recv_datagram(sk, flags, noblock, &err);
466 if (!skb)
467 goto out;
468
469 copied = skb->len;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900470 if (copied > len) {
471 copied = len;
472 msg->msg_flags |= MSG_TRUNC;
473 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700474
Herbert Xu60476372007-04-09 11:59:39 -0700475 if (skb_csum_unnecessary(skb)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700476 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
477 } else if (msg->msg_flags&MSG_TRUNC) {
Herbert Xufb286bb2005-11-10 13:01:24 -0800478 if (__skb_checksum_complete(skb))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700479 goto csum_copy_err;
480 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
481 } else {
482 err = skb_copy_and_csum_datagram_iovec(skb, 0, msg->msg_iov);
483 if (err == -EINVAL)
484 goto csum_copy_err;
485 }
486 if (err)
487 goto out_free;
488
489 /* Copy the address. */
490 if (sin6) {
491 sin6->sin6_family = AF_INET6;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700492 sin6->sin6_port = 0;
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700493 ipv6_addr_copy(&sin6->sin6_addr, &ipv6_hdr(skb)->saddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700494 sin6->sin6_flowinfo = 0;
495 sin6->sin6_scope_id = 0;
496 if (ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL)
497 sin6->sin6_scope_id = IP6CB(skb)->iif;
498 }
499
500 sock_recv_timestamp(msg, sk, skb);
501
502 if (np->rxopt.all)
503 datagram_recv_ctl(sk, msg, skb);
504
505 err = copied;
506 if (flags & MSG_TRUNC)
507 err = skb->len;
508
509out_free:
510 skb_free_datagram(sk, skb);
511out:
512 return err;
513
514csum_copy_err:
Herbert Xu3305b802005-12-13 23:16:37 -0800515 skb_kill_datagram(sk, skb, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700516
517 /* Error for blocking case is chosen to masquerade
518 as some normal condition.
519 */
520 err = (flags&MSG_DONTWAIT) ? -EAGAIN : -EHOSTUNREACH;
Wang Chena92aa312007-11-13 20:31:14 -0800521 atomic_inc(&sk->sk_drops);
Herbert Xu3305b802005-12-13 23:16:37 -0800522 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700523}
524
525static int rawv6_push_pending_frames(struct sock *sk, struct flowi *fl,
Herbert Xu357b40a2005-04-19 22:30:14 -0700526 struct raw6_sock *rp)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700527{
528 struct sk_buff *skb;
529 int err = 0;
Herbert Xu357b40a2005-04-19 22:30:14 -0700530 int offset;
531 int len;
Herbert Xu679a8732005-05-03 14:24:36 -0700532 int total_len;
Al Viro868c86b2006-11-14 21:35:48 -0800533 __wsum tmp_csum;
534 __sum16 csum;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700535
536 if (!rp->checksum)
537 goto send;
538
539 if ((skb = skb_peek(&sk->sk_write_queue)) == NULL)
540 goto out;
541
Herbert Xu357b40a2005-04-19 22:30:14 -0700542 offset = rp->offset;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700543 total_len = inet_sk(sk)->cork.length - (skb_network_header(skb) -
544 skb->data);
Herbert Xu679a8732005-05-03 14:24:36 -0700545 if (offset >= total_len - 1) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546 err = -EINVAL;
Herbert Xu357b40a2005-04-19 22:30:14 -0700547 ip6_flush_pending_frames(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700548 goto out;
549 }
550
551 /* should be check HW csum miyazawa */
552 if (skb_queue_len(&sk->sk_write_queue) == 1) {
553 /*
554 * Only one fragment on the socket.
555 */
556 tmp_csum = skb->csum;
557 } else {
Herbert Xu357b40a2005-04-19 22:30:14 -0700558 struct sk_buff *csum_skb = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700559 tmp_csum = 0;
560
561 skb_queue_walk(&sk->sk_write_queue, skb) {
562 tmp_csum = csum_add(tmp_csum, skb->csum);
Herbert Xu357b40a2005-04-19 22:30:14 -0700563
564 if (csum_skb)
565 continue;
566
Arnaldo Carvalho de Meloea2ae172007-04-25 17:55:53 -0700567 len = skb->len - skb_transport_offset(skb);
Herbert Xu357b40a2005-04-19 22:30:14 -0700568 if (offset >= len) {
569 offset -= len;
570 continue;
571 }
572
573 csum_skb = skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700574 }
Herbert Xu357b40a2005-04-19 22:30:14 -0700575
576 skb = csum_skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700577 }
578
Arnaldo Carvalho de Meloea2ae172007-04-25 17:55:53 -0700579 offset += skb_transport_offset(skb);
Herbert Xu357b40a2005-04-19 22:30:14 -0700580 if (skb_copy_bits(skb, offset, &csum, 2))
581 BUG();
582
Linus Torvalds1da177e2005-04-16 15:20:36 -0700583 /* in case cksum was not initialized */
Herbert Xu357b40a2005-04-19 22:30:14 -0700584 if (unlikely(csum))
Al Viro5f92a732006-11-14 21:36:54 -0800585 tmp_csum = csum_sub(tmp_csum, csum_unfold(csum));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700586
Al Viro868c86b2006-11-14 21:35:48 -0800587 csum = csum_ipv6_magic(&fl->fl6_src,
Herbert Xu357b40a2005-04-19 22:30:14 -0700588 &fl->fl6_dst,
Herbert Xu679a8732005-05-03 14:24:36 -0700589 total_len, fl->proto, tmp_csum);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590
Al Virof6ab0282006-11-16 02:36:50 -0800591 if (csum == 0 && fl->proto == IPPROTO_UDP)
592 csum = CSUM_MANGLED_0;
Herbert Xu357b40a2005-04-19 22:30:14 -0700593
Herbert Xu357b40a2005-04-19 22:30:14 -0700594 if (skb_store_bits(skb, offset, &csum, 2))
595 BUG();
596
Linus Torvalds1da177e2005-04-16 15:20:36 -0700597send:
598 err = ip6_push_pending_frames(sk);
599out:
600 return err;
601}
602
603static int rawv6_send_hdrinc(struct sock *sk, void *from, int length,
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900604 struct flowi *fl, struct rt6_info *rt,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700605 unsigned int flags)
606{
Herbert Xu3320da82005-04-19 22:32:22 -0700607 struct ipv6_pinfo *np = inet6_sk(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700608 struct ipv6hdr *iph;
609 struct sk_buff *skb;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 int err;
611
612 if (length > rt->u.dst.dev->mtu) {
613 ipv6_local_error(sk, EMSGSIZE, fl, rt->u.dst.dev->mtu);
614 return -EMSGSIZE;
615 }
616 if (flags&MSG_PROBE)
617 goto out;
618
Johannes Bergf5184d22008-05-12 20:48:31 -0700619 skb = sock_alloc_send_skb(sk,
620 length + LL_ALLOCATED_SPACE(rt->u.dst.dev) + 15,
621 flags & MSG_DONTWAIT, &err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700622 if (skb == NULL)
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900623 goto error;
Johannes Bergf5184d22008-05-12 20:48:31 -0700624 skb_reserve(skb, LL_RESERVED_SPACE(rt->u.dst.dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700625
626 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800627 skb->mark = sk->sk_mark;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700628 skb->dst = dst_clone(&rt->u.dst);
629
Arnaldo Carvalho de Melo1ced98e2007-03-10 19:57:15 -0300630 skb_put(skb, length);
631 skb_reset_network_header(skb);
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700632 iph = ipv6_hdr(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633
634 skb->ip_summed = CHECKSUM_NONE;
635
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700636 skb->transport_header = skb->network_header;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637 err = memcpy_fromiovecend((void *)iph, from, 0, length);
638 if (err)
639 goto error_fault;
640
YOSHIFUJI Hideakia11d2062006-11-04 20:11:37 +0900641 IP6_INC_STATS(rt->rt6i_idev, IPSTATS_MIB_OUTREQUESTS);
Patrick McHardy6e23ae22007-11-19 18:53:30 -0800642 err = NF_HOOK(PF_INET6, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 dst_output);
644 if (err > 0)
Herbert Xu3320da82005-04-19 22:32:22 -0700645 err = np->recverr ? net_xmit_errno(err) : 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700646 if (err)
647 goto error;
648out:
649 return 0;
650
651error_fault:
652 err = -EFAULT;
653 kfree_skb(skb);
654error:
YOSHIFUJI Hideakia11d2062006-11-04 20:11:37 +0900655 IP6_INC_STATS(rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900656 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700657}
658
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800659static int rawv6_probe_proto_opt(struct flowi *fl, struct msghdr *msg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660{
661 struct iovec *iov;
662 u8 __user *type = NULL;
663 u8 __user *code = NULL;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700664 u8 len = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665 int probed = 0;
666 int i;
667
668 if (!msg->msg_iov)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800669 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700670
671 for (i = 0; i < msg->msg_iovlen; i++) {
672 iov = &msg->msg_iov[i];
673 if (!iov)
674 continue;
675
676 switch (fl->proto) {
677 case IPPROTO_ICMPV6:
678 /* check if one-byte field is readable or not. */
679 if (iov->iov_base && iov->iov_len < 1)
680 break;
681
682 if (!type) {
683 type = iov->iov_base;
684 /* check if code field is readable or not. */
685 if (iov->iov_len > 1)
686 code = type + 1;
687 } else if (!code)
688 code = iov->iov_base;
689
690 if (type && code) {
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800691 if (get_user(fl->fl_icmp_type, type) ||
692 get_user(fl->fl_icmp_code, code))
693 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700694 probed = 1;
695 }
696 break;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700697 case IPPROTO_MH:
698 if (iov->iov_base && iov->iov_len < 1)
699 break;
700 /* check if type field is readable or not. */
701 if (iov->iov_len > 2 - len) {
702 u8 __user *p = iov->iov_base;
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800703 if (get_user(fl->fl_mh_type, &p[2 - len]))
704 return -EFAULT;
Masahide NAKAMURA6e8f4d42006-08-23 20:36:47 -0700705 probed = 1;
706 } else
707 len += iov->iov_len;
708
709 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700710 default:
711 probed = 1;
712 break;
713 }
714 if (probed)
715 break;
716 }
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800717 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718}
719
720static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
721 struct msghdr *msg, size_t len)
722{
723 struct ipv6_txoptions opt_space;
724 struct sockaddr_in6 * sin6 = (struct sockaddr_in6 *) msg->msg_name;
725 struct in6_addr *daddr, *final_p = NULL, final;
726 struct inet_sock *inet = inet_sk(sk);
727 struct ipv6_pinfo *np = inet6_sk(sk);
728 struct raw6_sock *rp = raw6_sk(sk);
729 struct ipv6_txoptions *opt = NULL;
730 struct ip6_flowlabel *flowlabel = NULL;
731 struct dst_entry *dst = NULL;
732 struct flowi fl;
733 int addr_len = msg->msg_namelen;
734 int hlimit = -1;
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900735 int tclass = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736 u16 proto;
737 int err;
738
739 /* Rough check on arithmetic overflow,
YOSHIFUJI Hideakib59e1392007-03-30 14:45:35 -0700740 better check is made in ip6_append_data().
Linus Torvalds1da177e2005-04-16 15:20:36 -0700741 */
YOSHIFUJI Hideakib59e1392007-03-30 14:45:35 -0700742 if (len > INT_MAX)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743 return -EMSGSIZE;
744
745 /* Mirror BSD error message compatibility */
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900746 if (msg->msg_flags & MSG_OOB)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747 return -EOPNOTSUPP;
748
749 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900750 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700751 */
752 memset(&fl, 0, sizeof(fl));
753
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800754 fl.mark = sk->sk_mark;
755
Linus Torvalds1da177e2005-04-16 15:20:36 -0700756 if (sin6) {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900757 if (addr_len < SIN6_LEN_RFC2133)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700758 return -EINVAL;
759
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900760 if (sin6->sin6_family && sin6->sin6_family != AF_INET6)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700761 return(-EAFNOSUPPORT);
762
763 /* port is the proto value [0..255] carried in nexthdr */
764 proto = ntohs(sin6->sin6_port);
765
766 if (!proto)
767 proto = inet->num;
768 else if (proto != inet->num)
769 return(-EINVAL);
770
771 if (proto > 255)
772 return(-EINVAL);
773
774 daddr = &sin6->sin6_addr;
775 if (np->sndflow) {
776 fl.fl6_flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
777 if (fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) {
778 flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
779 if (flowlabel == NULL)
780 return -EINVAL;
781 daddr = &flowlabel->dst;
782 }
783 }
784
785 /*
786 * Otherwise it will be difficult to maintain
787 * sk->sk_dst_cache.
788 */
789 if (sk->sk_state == TCP_ESTABLISHED &&
790 ipv6_addr_equal(daddr, &np->daddr))
791 daddr = &np->daddr;
792
793 if (addr_len >= sizeof(struct sockaddr_in6) &&
794 sin6->sin6_scope_id &&
795 ipv6_addr_type(daddr)&IPV6_ADDR_LINKLOCAL)
796 fl.oif = sin6->sin6_scope_id;
797 } else {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900798 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700799 return -EDESTADDRREQ;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900800
Linus Torvalds1da177e2005-04-16 15:20:36 -0700801 proto = inet->num;
802 daddr = &np->daddr;
803 fl.fl6_flowlabel = np->flow_label;
804 }
805
Linus Torvalds1da177e2005-04-16 15:20:36 -0700806 if (fl.oif == 0)
807 fl.oif = sk->sk_bound_dev_if;
808
809 if (msg->msg_controllen) {
810 opt = &opt_space;
811 memset(opt, 0, sizeof(struct ipv6_txoptions));
812 opt->tot_len = sizeof(struct ipv6_txoptions);
813
YOSHIFUJI Hideaki91e19082008-06-04 13:02:49 +0900814 err = datagram_send_ctl(sock_net(sk), msg, &fl, opt, &hlimit, &tclass);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700815 if (err < 0) {
816 fl6_sock_release(flowlabel);
817 return err;
818 }
819 if ((fl.fl6_flowlabel&IPV6_FLOWLABEL_MASK) && !flowlabel) {
820 flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
821 if (flowlabel == NULL)
822 return -EINVAL;
823 }
824 if (!(opt->opt_nflen|opt->opt_flen))
825 opt = NULL;
826 }
827 if (opt == NULL)
828 opt = np->opt;
YOSHIFUJI Hideakidf9890c2005-11-20 12:23:18 +0900829 if (flowlabel)
830 opt = fl6_merge_options(&opt_space, flowlabel, opt);
831 opt = ipv6_fixup_options(&opt_space, opt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700832
833 fl.proto = proto;
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800834 err = rawv6_probe_proto_opt(&fl, msg);
835 if (err)
836 goto out;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900837
Brian Haley876c7f42008-04-11 00:38:24 -0400838 if (!ipv6_addr_any(daddr))
839 ipv6_addr_copy(&fl.fl6_dst, daddr);
840 else
841 fl.fl6_dst.s6_addr[15] = 0x1; /* :: means loopback (BSD'ism) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700842 if (ipv6_addr_any(&fl.fl6_src) && !ipv6_addr_any(&np->saddr))
843 ipv6_addr_copy(&fl.fl6_src, &np->saddr);
844
845 /* merge ip6_build_xmit from ip6_output */
846 if (opt && opt->srcrt) {
847 struct rt0_hdr *rt0 = (struct rt0_hdr *) opt->srcrt;
848 ipv6_addr_copy(&final, &fl.fl6_dst);
849 ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
850 final_p = &final;
851 }
852
853 if (!fl.oif && ipv6_addr_is_multicast(&fl.fl6_dst))
854 fl.oif = np->mcast_oif;
Venkat Yekkiralabeb8d132006-08-04 23:12:42 -0700855 security_sk_classify_flow(sk, &fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700856
857 err = ip6_dst_lookup(sk, &dst, &fl);
858 if (err)
859 goto out;
860 if (final_p)
861 ipv6_addr_copy(&fl.fl6_dst, final_p);
862
Herbert Xubb728452007-12-12 18:48:58 -0800863 if ((err = __xfrm_lookup(&dst, &fl, sk, XFRM_LOOKUP_WAIT)) < 0) {
David S. Miller14e50e52007-05-24 18:17:54 -0700864 if (err == -EREMOTE)
865 err = ip6_dst_blackhole(sk, &dst, &fl);
866 if (err < 0)
867 goto out;
868 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869
870 if (hlimit < 0) {
871 if (ipv6_addr_is_multicast(&fl.fl6_dst))
872 hlimit = np->mcast_hops;
873 else
874 hlimit = np->hop_limit;
875 if (hlimit < 0)
YOSHIFUJI Hideaki6b75d092008-03-10 06:00:30 -0400876 hlimit = ip6_dst_hoplimit(dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877 }
878
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900879 if (tclass < 0) {
YOSHIFUJI Hideakie012d512006-09-13 20:01:28 -0700880 tclass = np->tclass;
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900881 if (tclass < 0)
882 tclass = 0;
883 }
884
Linus Torvalds1da177e2005-04-16 15:20:36 -0700885 if (msg->msg_flags&MSG_CONFIRM)
886 goto do_confirm;
887
888back_from_confirm:
889 if (inet->hdrincl) {
890 err = rawv6_send_hdrinc(sk, msg->msg_iov, len, &fl, (struct rt6_info*)dst, msg->msg_flags);
891 } else {
892 lock_sock(sk);
YOSHIFUJI Hideaki41a1f8e2005-09-08 10:19:03 +0900893 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov,
894 len, 0, hlimit, tclass, opt, &fl, (struct rt6_info*)dst,
895 msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896
897 if (err)
898 ip6_flush_pending_frames(sk);
899 else if (!(msg->msg_flags & MSG_MORE))
Herbert Xu357b40a2005-04-19 22:30:14 -0700900 err = rawv6_push_pending_frames(sk, &fl, rp);
YOSHIFUJI Hideaki3ef9d942007-09-14 16:45:40 -0700901 release_sock(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902 }
903done:
Nicolas DICHTEL6d3e85e2006-02-13 15:56:13 -0800904 dst_release(dst);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900905out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700906 fl6_sock_release(flowlabel);
907 return err<0?err:len;
908do_confirm:
909 dst_confirm(dst);
910 if (!(msg->msg_flags & MSG_PROBE) || len)
911 goto back_from_confirm;
912 err = 0;
913 goto done;
914}
915
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900916static int rawv6_seticmpfilter(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700917 char __user *optval, int optlen)
918{
919 switch (optname) {
920 case ICMPV6_FILTER:
921 if (optlen > sizeof(struct icmp6_filter))
922 optlen = sizeof(struct icmp6_filter);
923 if (copy_from_user(&raw6_sk(sk)->filter, optval, optlen))
924 return -EFAULT;
925 return 0;
926 default:
927 return -ENOPROTOOPT;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700928 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700929
930 return 0;
931}
932
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900933static int rawv6_geticmpfilter(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934 char __user *optval, int __user *optlen)
935{
936 int len;
937
938 switch (optname) {
939 case ICMPV6_FILTER:
940 if (get_user(len, optlen))
941 return -EFAULT;
942 if (len < 0)
943 return -EINVAL;
944 if (len > sizeof(struct icmp6_filter))
945 len = sizeof(struct icmp6_filter);
946 if (put_user(len, optlen))
947 return -EFAULT;
948 if (copy_to_user(optval, &raw6_sk(sk)->filter, len))
949 return -EFAULT;
950 return 0;
951 default:
952 return -ENOPROTOOPT;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700953 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700954
955 return 0;
956}
957
958
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800959static int do_rawv6_setsockopt(struct sock *sk, int level, int optname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700960 char __user *optval, int optlen)
961{
962 struct raw6_sock *rp = raw6_sk(sk);
963 int val;
964
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900965 if (get_user(val, (int __user *)optval))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966 return -EFAULT;
967
968 switch (optname) {
969 case IPV6_CHECKSUM:
YOSHIFUJI Hideaki1a98d052008-04-24 21:30:38 -0700970 if (inet_sk(sk)->num == IPPROTO_ICMPV6 &&
971 level == IPPROTO_IPV6) {
972 /*
973 * RFC3542 tells that IPV6_CHECKSUM socket
974 * option in the IPPROTO_IPV6 level is not
975 * allowed on ICMPv6 sockets.
976 * If you want to set it, use IPPROTO_RAW
977 * level IPV6_CHECKSUM socket option
978 * (Linux extension).
979 */
980 return -EINVAL;
981 }
982
Linus Torvalds1da177e2005-04-16 15:20:36 -0700983 /* You may get strange result with a positive odd offset;
984 RFC2292bis agrees with me. */
985 if (val > 0 && (val&1))
986 return(-EINVAL);
987 if (val < 0) {
988 rp->checksum = 0;
989 } else {
990 rp->checksum = 1;
991 rp->offset = val;
992 }
993
994 return 0;
995 break;
996
997 default:
998 return(-ENOPROTOOPT);
999 }
1000}
1001
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001002static int rawv6_setsockopt(struct sock *sk, int level, int optname,
1003 char __user *optval, int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005 switch(level) {
1006 case SOL_RAW:
1007 break;
1008
1009 case SOL_ICMPV6:
1010 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1011 return -EOPNOTSUPP;
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001012 return rawv6_seticmpfilter(sk, level, optname, optval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013 optlen);
1014 case SOL_IPV6:
1015 if (optname == IPV6_CHECKSUM)
1016 break;
1017 default:
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001018 return ipv6_setsockopt(sk, level, optname, optval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019 optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001020 }
1021
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001022 return do_rawv6_setsockopt(sk, level, optname, optval, optlen);
1023}
1024
1025#ifdef CONFIG_COMPAT
1026static int compat_rawv6_setsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001027 char __user *optval, int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001028{
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001029 switch (level) {
1030 case SOL_RAW:
1031 break;
1032 case SOL_ICMPV6:
1033 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1034 return -EOPNOTSUPP;
1035 return rawv6_seticmpfilter(sk, level, optname, optval, optlen);
1036 case SOL_IPV6:
1037 if (optname == IPV6_CHECKSUM)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001038 break;
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001039 default:
1040 return compat_ipv6_setsockopt(sk, level, optname,
1041 optval, optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001042 }
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001043 return do_rawv6_setsockopt(sk, level, optname, optval, optlen);
1044}
1045#endif
1046
1047static int do_rawv6_getsockopt(struct sock *sk, int level, int optname,
1048 char __user *optval, int __user *optlen)
1049{
1050 struct raw6_sock *rp = raw6_sk(sk);
1051 int val, len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001052
1053 if (get_user(len,optlen))
1054 return -EFAULT;
1055
1056 switch (optname) {
1057 case IPV6_CHECKSUM:
YOSHIFUJI Hideaki1a98d052008-04-24 21:30:38 -07001058 /*
1059 * We allow getsockopt() for IPPROTO_IPV6-level
1060 * IPV6_CHECKSUM socket option on ICMPv6 sockets
1061 * since RFC3542 is silent about it.
1062 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001063 if (rp->checksum == 0)
1064 val = -1;
1065 else
1066 val = rp->offset;
1067 break;
1068
1069 default:
1070 return -ENOPROTOOPT;
1071 }
1072
1073 len = min_t(unsigned int, sizeof(int), len);
1074
1075 if (put_user(len, optlen))
1076 return -EFAULT;
1077 if (copy_to_user(optval,&val,len))
1078 return -EFAULT;
1079 return 0;
1080}
1081
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001082static int rawv6_getsockopt(struct sock *sk, int level, int optname,
1083 char __user *optval, int __user *optlen)
1084{
1085 switch(level) {
1086 case SOL_RAW:
1087 break;
1088
1089 case SOL_ICMPV6:
1090 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1091 return -EOPNOTSUPP;
1092 return rawv6_geticmpfilter(sk, level, optname, optval,
1093 optlen);
1094 case SOL_IPV6:
1095 if (optname == IPV6_CHECKSUM)
1096 break;
1097 default:
1098 return ipv6_getsockopt(sk, level, optname, optval,
1099 optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001100 }
1101
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001102 return do_rawv6_getsockopt(sk, level, optname, optval, optlen);
1103}
1104
1105#ifdef CONFIG_COMPAT
1106static int compat_rawv6_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001107 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001108{
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001109 switch (level) {
1110 case SOL_RAW:
1111 break;
1112 case SOL_ICMPV6:
1113 if (inet_sk(sk)->num != IPPROTO_ICMPV6)
1114 return -EOPNOTSUPP;
1115 return rawv6_geticmpfilter(sk, level, optname, optval, optlen);
1116 case SOL_IPV6:
1117 if (optname == IPV6_CHECKSUM)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001118 break;
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001119 default:
1120 return compat_ipv6_getsockopt(sk, level, optname,
1121 optval, optlen);
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001122 }
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001123 return do_rawv6_getsockopt(sk, level, optname, optval, optlen);
1124}
1125#endif
1126
Linus Torvalds1da177e2005-04-16 15:20:36 -07001127static int rawv6_ioctl(struct sock *sk, int cmd, unsigned long arg)
1128{
1129 switch(cmd) {
1130 case SIOCOUTQ:
1131 {
1132 int amount = atomic_read(&sk->sk_wmem_alloc);
1133 return put_user(amount, (int __user *)arg);
1134 }
1135 case SIOCINQ:
1136 {
1137 struct sk_buff *skb;
1138 int amount = 0;
1139
Herbert Xue0f9f852005-06-18 22:56:18 -07001140 spin_lock_bh(&sk->sk_receive_queue.lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001141 skb = skb_peek(&sk->sk_receive_queue);
1142 if (skb != NULL)
Arnaldo Carvalho de Melo27a884d2007-04-19 20:29:13 -07001143 amount = skb->tail - skb->transport_header;
Herbert Xue0f9f852005-06-18 22:56:18 -07001144 spin_unlock_bh(&sk->sk_receive_queue.lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001145 return put_user(amount, (int __user *)arg);
1146 }
1147
1148 default:
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001149#ifdef CONFIG_IPV6_MROUTE
1150 return ip6mr_ioctl(sk, cmd, (void __user *)arg);
1151#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152 return -ENOIOCTLCMD;
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001153#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001154 }
1155}
1156
1157static void rawv6_close(struct sock *sk, long timeout)
1158{
1159 if (inet_sk(sk)->num == IPPROTO_RAW)
Denis V. Lunev725a8ff2008-07-19 00:28:58 -07001160 ip6_ra_control(sk, -1);
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09001161 ip6mr_sk_done(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001162 sk_common_release(sk);
1163}
1164
Brian Haley7d06b2e2008-06-14 17:04:49 -07001165static void raw6_destroy(struct sock *sk)
Denis V. Lunev22dd4852008-06-04 15:16:12 -07001166{
1167 lock_sock(sk);
1168 ip6_flush_pending_frames(sk);
1169 release_sock(sk);
David S. Millerf23d60d2008-06-12 14:47:58 -07001170
Brian Haley7d06b2e2008-06-14 17:04:49 -07001171 inet6_destroy_sock(sk);
Denis V. Lunev22dd4852008-06-04 15:16:12 -07001172}
1173
Linus Torvalds1da177e2005-04-16 15:20:36 -07001174static int rawv6_init_sk(struct sock *sk)
1175{
Masahide NAKAMURAf48d5ff2007-02-07 00:07:39 -08001176 struct raw6_sock *rp = raw6_sk(sk);
1177
1178 switch (inet_sk(sk)->num) {
1179 case IPPROTO_ICMPV6:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001180 rp->checksum = 1;
1181 rp->offset = 2;
Masahide NAKAMURAf48d5ff2007-02-07 00:07:39 -08001182 break;
1183 case IPPROTO_MH:
1184 rp->checksum = 1;
1185 rp->offset = 4;
1186 break;
1187 default:
1188 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001189 }
1190 return(0);
1191}
1192
1193struct proto rawv6_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001194 .name = "RAWv6",
1195 .owner = THIS_MODULE,
1196 .close = rawv6_close,
Denis V. Lunev22dd4852008-06-04 15:16:12 -07001197 .destroy = raw6_destroy,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001198 .connect = ip6_datagram_connect,
1199 .disconnect = udp_disconnect,
1200 .ioctl = rawv6_ioctl,
1201 .init = rawv6_init_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001202 .setsockopt = rawv6_setsockopt,
1203 .getsockopt = rawv6_getsockopt,
1204 .sendmsg = rawv6_sendmsg,
1205 .recvmsg = rawv6_recvmsg,
1206 .bind = rawv6_bind,
1207 .backlog_rcv = rawv6_rcv_skb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -07001208 .hash = raw_hash_sk,
1209 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001210 .obj_size = sizeof(struct raw6_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -07001211 .h.raw_hash = &raw_v6_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001212#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -08001213 .compat_setsockopt = compat_rawv6_setsockopt,
1214 .compat_getsockopt = compat_rawv6_getsockopt,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -08001215#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001216};
1217
1218#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219static void raw6_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
1220{
1221 struct ipv6_pinfo *np = inet6_sk(sp);
1222 struct in6_addr *dest, *src;
1223 __u16 destp, srcp;
1224
1225 dest = &np->daddr;
1226 src = &np->rcv_saddr;
1227 destp = 0;
1228 srcp = inet_sk(sp)->num;
1229 seq_printf(seq,
1230 "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
Wang Chena92aa312007-11-13 20:31:14 -08001231 "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
Linus Torvalds1da177e2005-04-16 15:20:36 -07001232 i,
1233 src->s6_addr32[0], src->s6_addr32[1],
1234 src->s6_addr32[2], src->s6_addr32[3], srcp,
1235 dest->s6_addr32[0], dest->s6_addr32[1],
1236 dest->s6_addr32[2], dest->s6_addr32[3], destp,
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001237 sp->sk_state,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001238 atomic_read(&sp->sk_wmem_alloc),
1239 atomic_read(&sp->sk_rmem_alloc),
1240 0, 0L, 0,
1241 sock_i_uid(sp), 0,
1242 sock_i_ino(sp),
Wang Chena92aa312007-11-13 20:31:14 -08001243 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001244}
1245
1246static int raw6_seq_show(struct seq_file *seq, void *v)
1247{
1248 if (v == SEQ_START_TOKEN)
1249 seq_printf(seq,
1250 " sl "
1251 "local_address "
1252 "remote_address "
1253 "st tx_queue rx_queue tr tm->when retrnsmt"
Eric Dumazetcb61cb92008-06-17 21:04:56 -07001254 " uid timeout inode ref pointer drops\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -07001255 else
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001256 raw6_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001257 return 0;
1258}
1259
Philippe De Muyter56b3d972007-07-10 23:07:31 -07001260static const struct seq_operations raw6_seq_ops = {
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001261 .start = raw_seq_start,
1262 .next = raw_seq_next,
1263 .stop = raw_seq_stop,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001264 .show = raw6_seq_show,
1265};
1266
1267static int raw6_seq_open(struct inode *inode, struct file *file)
1268{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001269 return raw_seq_open(inode, file, &raw_v6_hashinfo, &raw6_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001270}
1271
Arjan van de Ven9a321442007-02-12 00:55:35 -08001272static const struct file_operations raw6_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273 .owner = THIS_MODULE,
1274 .open = raw6_seq_open,
1275 .read = seq_read,
1276 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001277 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278};
1279
Pavel Emelyanova308da12008-01-14 05:36:50 -08001280static int raw6_init_net(struct net *net)
1281{
1282 if (!proc_net_fops_create(net, "raw6", S_IRUGO, &raw6_seq_fops))
1283 return -ENOMEM;
1284
1285 return 0;
1286}
1287
1288static void raw6_exit_net(struct net *net)
1289{
1290 proc_net_remove(net, "raw6");
1291}
1292
1293static struct pernet_operations raw6_net_ops = {
1294 .init = raw6_init_net,
1295 .exit = raw6_exit_net,
1296};
1297
Linus Torvalds1da177e2005-04-16 15:20:36 -07001298int __init raw6_proc_init(void)
1299{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001300 return register_pernet_subsys(&raw6_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301}
1302
1303void raw6_proc_exit(void)
1304{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001305 unregister_pernet_subsys(&raw6_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001306}
1307#endif /* CONFIG_PROC_FS */
Daniel Lezcano7f4e4862007-12-11 02:25:35 -08001308
1309/* Same as inet6_dgram_ops, sans udp_poll. */
1310static const struct proto_ops inet6_sockraw_ops = {
1311 .family = PF_INET6,
1312 .owner = THIS_MODULE,
1313 .release = inet6_release,
1314 .bind = inet6_bind,
1315 .connect = inet_dgram_connect, /* ok */
1316 .socketpair = sock_no_socketpair, /* a do nothing */
1317 .accept = sock_no_accept, /* a do nothing */
1318 .getname = inet6_getname,
1319 .poll = datagram_poll, /* ok */
1320 .ioctl = inet6_ioctl, /* must change */
1321 .listen = sock_no_listen, /* ok */
1322 .shutdown = inet_shutdown, /* ok */
1323 .setsockopt = sock_common_setsockopt, /* ok */
1324 .getsockopt = sock_common_getsockopt, /* ok */
1325 .sendmsg = inet_sendmsg, /* ok */
1326 .recvmsg = sock_common_recvmsg, /* ok */
1327 .mmap = sock_no_mmap,
1328 .sendpage = sock_no_sendpage,
1329#ifdef CONFIG_COMPAT
1330 .compat_setsockopt = compat_sock_common_setsockopt,
1331 .compat_getsockopt = compat_sock_common_getsockopt,
1332#endif
1333};
1334
1335static struct inet_protosw rawv6_protosw = {
1336 .type = SOCK_RAW,
1337 .protocol = IPPROTO_IP, /* wild card */
1338 .prot = &rawv6_prot,
1339 .ops = &inet6_sockraw_ops,
1340 .capability = CAP_NET_RAW,
1341 .no_check = UDP_CSUM_DEFAULT,
1342 .flags = INET_PROTOSW_REUSE,
1343};
1344
1345int __init rawv6_init(void)
1346{
1347 int ret;
1348
1349 ret = inet6_register_protosw(&rawv6_protosw);
1350 if (ret)
1351 goto out;
1352out:
1353 return ret;
1354}
1355
Daniel Lezcano09f77092007-12-13 05:34:58 -08001356void rawv6_exit(void)
Daniel Lezcano7f4e4862007-12-11 02:25:35 -08001357{
1358 inet6_unregister_protosw(&rawv6_protosw);
1359}