Commit | Line | Data |
---|---|---|
1da177e4 | 1 | /* |
c4d3efaf | 2 | * IPv6 tunneling device |
1da177e4 LT |
3 | * Linux INET6 implementation |
4 | * | |
5 | * Authors: | |
1ab1457c | 6 | * Ville Nuorvala <vnuorval@tcs.hut.fi> |
c4d3efaf | 7 | * Yasuyuki Kozakai <kozakai@linux-ipv6.org> |
1da177e4 LT |
8 | * |
9 | * $Id$ | |
10 | * | |
11 | * Based on: | |
c4d3efaf | 12 | * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c |
1da177e4 LT |
13 | * |
14 | * RFC 2473 | |
15 | * | |
16 | * This program is free software; you can redistribute it and/or | |
17 | * modify it under the terms of the GNU General Public License | |
18 | * as published by the Free Software Foundation; either version | |
19 | * 2 of the License, or (at your option) any later version. | |
20 | * | |
21 | */ | |
22 | ||
1da177e4 | 23 | #include <linux/module.h> |
4fc268d2 | 24 | #include <linux/capability.h> |
1da177e4 LT |
25 | #include <linux/errno.h> |
26 | #include <linux/types.h> | |
27 | #include <linux/sockios.h> | |
c4d3efaf | 28 | #include <linux/icmp.h> |
1da177e4 LT |
29 | #include <linux/if.h> |
30 | #include <linux/in.h> | |
31 | #include <linux/ip.h> | |
32 | #include <linux/if_tunnel.h> | |
33 | #include <linux/net.h> | |
34 | #include <linux/in6.h> | |
35 | #include <linux/netdevice.h> | |
36 | #include <linux/if_arp.h> | |
37 | #include <linux/icmpv6.h> | |
38 | #include <linux/init.h> | |
39 | #include <linux/route.h> | |
40 | #include <linux/rtnetlink.h> | |
41 | #include <linux/netfilter_ipv6.h> | |
42 | ||
43 | #include <asm/uaccess.h> | |
44 | #include <asm/atomic.h> | |
45 | ||
c4d3efaf | 46 | #include <net/icmp.h> |
1da177e4 LT |
47 | #include <net/ip.h> |
48 | #include <net/ipv6.h> | |
1da177e4 LT |
49 | #include <net/ip6_route.h> |
50 | #include <net/addrconf.h> | |
51 | #include <net/ip6_tunnel.h> | |
52 | #include <net/xfrm.h> | |
53 | #include <net/dsfield.h> | |
54 | #include <net/inet_ecn.h> | |
55 | ||
56 | MODULE_AUTHOR("Ville Nuorvala"); | |
c4d3efaf | 57 | MODULE_DESCRIPTION("IPv6 tunneling device"); |
1da177e4 LT |
58 | MODULE_LICENSE("GPL"); |
59 | ||
60 | #define IPV6_TLV_TEL_DST_SIZE 8 | |
61 | ||
62 | #ifdef IP6_TNL_DEBUG | |
63 | #define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __FUNCTION__) | |
64 | #else | |
65 | #define IP6_TNL_TRACE(x...) do {;} while(0) | |
66 | #endif | |
67 | ||
68 | #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK) | |
c4d3efaf | 69 | #define IPV6_TCLASS_SHIFT 20 |
1da177e4 LT |
70 | |
71 | #define HASH_SIZE 32 | |
72 | ||
e69a4adc | 73 | #define HASH(addr) ((__force u32)((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \ |
1ab1457c YH |
74 | (addr)->s6_addr32[2] ^ (addr)->s6_addr32[3]) & \ |
75 | (HASH_SIZE - 1)) | |
1da177e4 | 76 | |
3144581c YK |
77 | static int ip6_fb_tnl_dev_init(struct net_device *dev); |
78 | static int ip6_tnl_dev_init(struct net_device *dev); | |
79 | static void ip6_tnl_dev_setup(struct net_device *dev); | |
1da177e4 LT |
80 | |
81 | /* the IPv6 tunnel fallback device */ | |
3144581c | 82 | static struct net_device *ip6_fb_tnl_dev; |
1da177e4 LT |
83 | |
84 | ||
85 | /* lists for storing tunnels in use */ | |
86 | static struct ip6_tnl *tnls_r_l[HASH_SIZE]; | |
87 | static struct ip6_tnl *tnls_wc[1]; | |
88 | static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l }; | |
89 | ||
90 | /* lock for the tunnel lists */ | |
3144581c | 91 | static DEFINE_RWLOCK(ip6_tnl_lock); |
1da177e4 LT |
92 | |
93 | static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t) | |
94 | { | |
95 | struct dst_entry *dst = t->dst_cache; | |
96 | ||
1ab1457c | 97 | if (dst && dst->obsolete && |
1da177e4 LT |
98 | dst->ops->check(dst, t->dst_cookie) == NULL) { |
99 | t->dst_cache = NULL; | |
100 | dst_release(dst); | |
101 | return NULL; | |
102 | } | |
103 | ||
104 | return dst; | |
105 | } | |
106 | ||
107 | static inline void ip6_tnl_dst_reset(struct ip6_tnl *t) | |
108 | { | |
109 | dst_release(t->dst_cache); | |
110 | t->dst_cache = NULL; | |
111 | } | |
112 | ||
113 | static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst) | |
114 | { | |
115 | struct rt6_info *rt = (struct rt6_info *) dst; | |
116 | t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0; | |
117 | dst_release(t->dst_cache); | |
118 | t->dst_cache = dst; | |
119 | } | |
120 | ||
121 | /** | |
3144581c | 122 | * ip6_tnl_lookup - fetch tunnel matching the end-point addresses |
1ab1457c YH |
123 | * @remote: the address of the tunnel exit-point |
124 | * @local: the address of the tunnel entry-point | |
1da177e4 | 125 | * |
1ab1457c | 126 | * Return: |
1da177e4 | 127 | * tunnel matching given end-points if found, |
1ab1457c | 128 | * else fallback tunnel if its device is up, |
1da177e4 LT |
129 | * else %NULL |
130 | **/ | |
131 | ||
132 | static struct ip6_tnl * | |
3144581c | 133 | ip6_tnl_lookup(struct in6_addr *remote, struct in6_addr *local) |
1da177e4 LT |
134 | { |
135 | unsigned h0 = HASH(remote); | |
136 | unsigned h1 = HASH(local); | |
137 | struct ip6_tnl *t; | |
138 | ||
139 | for (t = tnls_r_l[h0 ^ h1]; t; t = t->next) { | |
140 | if (ipv6_addr_equal(local, &t->parms.laddr) && | |
141 | ipv6_addr_equal(remote, &t->parms.raddr) && | |
142 | (t->dev->flags & IFF_UP)) | |
143 | return t; | |
144 | } | |
145 | if ((t = tnls_wc[0]) != NULL && (t->dev->flags & IFF_UP)) | |
146 | return t; | |
147 | ||
148 | return NULL; | |
149 | } | |
150 | ||
151 | /** | |
3144581c | 152 | * ip6_tnl_bucket - get head of list matching given tunnel parameters |
1ab1457c | 153 | * @p: parameters containing tunnel end-points |
1da177e4 LT |
154 | * |
155 | * Description: | |
3144581c | 156 | * ip6_tnl_bucket() returns the head of the list matching the |
1da177e4 LT |
157 | * &struct in6_addr entries laddr and raddr in @p. |
158 | * | |
1ab1457c | 159 | * Return: head of IPv6 tunnel list |
1da177e4 LT |
160 | **/ |
161 | ||
162 | static struct ip6_tnl ** | |
3144581c | 163 | ip6_tnl_bucket(struct ip6_tnl_parm *p) |
1da177e4 LT |
164 | { |
165 | struct in6_addr *remote = &p->raddr; | |
166 | struct in6_addr *local = &p->laddr; | |
167 | unsigned h = 0; | |
168 | int prio = 0; | |
169 | ||
170 | if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) { | |
171 | prio = 1; | |
172 | h = HASH(remote) ^ HASH(local); | |
173 | } | |
174 | return &tnls[prio][h]; | |
175 | } | |
176 | ||
177 | /** | |
3144581c | 178 | * ip6_tnl_link - add tunnel to hash table |
1da177e4 LT |
179 | * @t: tunnel to be added |
180 | **/ | |
181 | ||
182 | static void | |
3144581c | 183 | ip6_tnl_link(struct ip6_tnl *t) |
1da177e4 | 184 | { |
3144581c | 185 | struct ip6_tnl **tp = ip6_tnl_bucket(&t->parms); |
1da177e4 LT |
186 | |
187 | t->next = *tp; | |
3144581c | 188 | write_lock_bh(&ip6_tnl_lock); |
1da177e4 | 189 | *tp = t; |
3144581c | 190 | write_unlock_bh(&ip6_tnl_lock); |
1da177e4 LT |
191 | } |
192 | ||
193 | /** | |
3144581c | 194 | * ip6_tnl_unlink - remove tunnel from hash table |
1da177e4 LT |
195 | * @t: tunnel to be removed |
196 | **/ | |
197 | ||
198 | static void | |
3144581c | 199 | ip6_tnl_unlink(struct ip6_tnl *t) |
1da177e4 LT |
200 | { |
201 | struct ip6_tnl **tp; | |
202 | ||
3144581c | 203 | for (tp = ip6_tnl_bucket(&t->parms); *tp; tp = &(*tp)->next) { |
1da177e4 | 204 | if (t == *tp) { |
3144581c | 205 | write_lock_bh(&ip6_tnl_lock); |
1da177e4 | 206 | *tp = t->next; |
3144581c | 207 | write_unlock_bh(&ip6_tnl_lock); |
1da177e4 LT |
208 | break; |
209 | } | |
210 | } | |
211 | } | |
212 | ||
213 | /** | |
214 | * ip6_tnl_create() - create a new tunnel | |
215 | * @p: tunnel parameters | |
216 | * @pt: pointer to new tunnel | |
217 | * | |
218 | * Description: | |
219 | * Create tunnel matching given parameters. | |
1ab1457c YH |
220 | * |
221 | * Return: | |
567131a7 | 222 | * created tunnel or NULL |
1da177e4 LT |
223 | **/ |
224 | ||
567131a7 | 225 | static struct ip6_tnl *ip6_tnl_create(struct ip6_tnl_parm *p) |
1da177e4 LT |
226 | { |
227 | struct net_device *dev; | |
228 | struct ip6_tnl *t; | |
229 | char name[IFNAMSIZ]; | |
230 | int err; | |
231 | ||
232 | if (p->name[0]) { | |
233 | strlcpy(name, p->name, IFNAMSIZ); | |
234 | } else { | |
235 | int i; | |
236 | for (i = 1; i < IP6_TNL_MAX; i++) { | |
237 | sprintf(name, "ip6tnl%d", i); | |
238 | if (__dev_get_by_name(name) == NULL) | |
239 | break; | |
240 | } | |
1ab1457c | 241 | if (i == IP6_TNL_MAX) |
567131a7 | 242 | goto failed; |
1da177e4 | 243 | } |
3144581c | 244 | dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup); |
1da177e4 | 245 | if (dev == NULL) |
567131a7 | 246 | goto failed; |
1da177e4 | 247 | |
2941a486 | 248 | t = netdev_priv(dev); |
3144581c | 249 | dev->init = ip6_tnl_dev_init; |
1da177e4 LT |
250 | t->parms = *p; |
251 | ||
252 | if ((err = register_netdevice(dev)) < 0) { | |
253 | free_netdev(dev); | |
567131a7 | 254 | goto failed; |
1da177e4 LT |
255 | } |
256 | dev_hold(dev); | |
3144581c | 257 | ip6_tnl_link(t); |
567131a7 VN |
258 | return t; |
259 | failed: | |
260 | return NULL; | |
1da177e4 LT |
261 | } |
262 | ||
263 | /** | |
3144581c | 264 | * ip6_tnl_locate - find or create tunnel matching given parameters |
1ab1457c | 265 | * @p: tunnel parameters |
1da177e4 LT |
266 | * @create: != 0 if allowed to create new tunnel if no match found |
267 | * | |
268 | * Description: | |
3144581c | 269 | * ip6_tnl_locate() first tries to locate an existing tunnel |
1da177e4 LT |
270 | * based on @parms. If this is unsuccessful, but @create is set a new |
271 | * tunnel device is created and registered for use. | |
272 | * | |
273 | * Return: | |
567131a7 | 274 | * matching tunnel or NULL |
1da177e4 LT |
275 | **/ |
276 | ||
3144581c | 277 | static struct ip6_tnl *ip6_tnl_locate(struct ip6_tnl_parm *p, int create) |
1da177e4 LT |
278 | { |
279 | struct in6_addr *remote = &p->raddr; | |
280 | struct in6_addr *local = &p->laddr; | |
281 | struct ip6_tnl *t; | |
282 | ||
3144581c | 283 | for (t = *ip6_tnl_bucket(p); t; t = t->next) { |
1da177e4 | 284 | if (ipv6_addr_equal(local, &t->parms.laddr) && |
567131a7 VN |
285 | ipv6_addr_equal(remote, &t->parms.raddr)) |
286 | return t; | |
1da177e4 LT |
287 | } |
288 | if (!create) | |
567131a7 VN |
289 | return NULL; |
290 | return ip6_tnl_create(p); | |
1da177e4 LT |
291 | } |
292 | ||
293 | /** | |
3144581c | 294 | * ip6_tnl_dev_uninit - tunnel device uninitializer |
1da177e4 | 295 | * @dev: the device to be destroyed |
1ab1457c | 296 | * |
1da177e4 | 297 | * Description: |
3144581c | 298 | * ip6_tnl_dev_uninit() removes tunnel from its list |
1da177e4 LT |
299 | **/ |
300 | ||
301 | static void | |
3144581c | 302 | ip6_tnl_dev_uninit(struct net_device *dev) |
1da177e4 | 303 | { |
2941a486 | 304 | struct ip6_tnl *t = netdev_priv(dev); |
1da177e4 | 305 | |
3144581c YK |
306 | if (dev == ip6_fb_tnl_dev) { |
307 | write_lock_bh(&ip6_tnl_lock); | |
1da177e4 | 308 | tnls_wc[0] = NULL; |
3144581c | 309 | write_unlock_bh(&ip6_tnl_lock); |
1da177e4 | 310 | } else { |
3144581c | 311 | ip6_tnl_unlink(t); |
1da177e4 LT |
312 | } |
313 | ip6_tnl_dst_reset(t); | |
314 | dev_put(dev); | |
315 | } | |
316 | ||
317 | /** | |
318 | * parse_tvl_tnl_enc_lim - handle encapsulation limit option | |
319 | * @skb: received socket buffer | |
320 | * | |
1ab1457c YH |
321 | * Return: |
322 | * 0 if none was found, | |
1da177e4 LT |
323 | * else index to encapsulation limit |
324 | **/ | |
325 | ||
326 | static __u16 | |
327 | parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw) | |
328 | { | |
329 | struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw; | |
330 | __u8 nexthdr = ipv6h->nexthdr; | |
331 | __u16 off = sizeof (*ipv6h); | |
332 | ||
333 | while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) { | |
334 | __u16 optlen = 0; | |
335 | struct ipv6_opt_hdr *hdr; | |
336 | if (raw + off + sizeof (*hdr) > skb->data && | |
337 | !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr))) | |
338 | break; | |
339 | ||
340 | hdr = (struct ipv6_opt_hdr *) (raw + off); | |
341 | if (nexthdr == NEXTHDR_FRAGMENT) { | |
342 | struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr; | |
343 | if (frag_hdr->frag_off) | |
344 | break; | |
345 | optlen = 8; | |
346 | } else if (nexthdr == NEXTHDR_AUTH) { | |
347 | optlen = (hdr->hdrlen + 2) << 2; | |
348 | } else { | |
349 | optlen = ipv6_optlen(hdr); | |
350 | } | |
351 | if (nexthdr == NEXTHDR_DEST) { | |
352 | __u16 i = off + 2; | |
353 | while (1) { | |
354 | struct ipv6_tlv_tnl_enc_lim *tel; | |
355 | ||
356 | /* No more room for encapsulation limit */ | |
357 | if (i + sizeof (*tel) > off + optlen) | |
358 | break; | |
359 | ||
360 | tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i]; | |
361 | /* return index of option if found and valid */ | |
362 | if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT && | |
363 | tel->length == 1) | |
364 | return i; | |
365 | /* else jump to next option */ | |
366 | if (tel->type) | |
367 | i += tel->length + 2; | |
368 | else | |
369 | i++; | |
370 | } | |
371 | } | |
372 | nexthdr = hdr->nexthdr; | |
373 | off += optlen; | |
374 | } | |
375 | return 0; | |
376 | } | |
377 | ||
378 | /** | |
e490d1d8 | 379 | * ip6_tnl_err - tunnel error handler |
1da177e4 LT |
380 | * |
381 | * Description: | |
e490d1d8 | 382 | * ip6_tnl_err() should handle errors in the tunnel according |
1da177e4 LT |
383 | * to the specifications in RFC 2473. |
384 | **/ | |
385 | ||
d2acc347 | 386 | static int |
502b0935 | 387 | ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt, |
e490d1d8 | 388 | int *type, int *code, int *msg, __be32 *info, int offset) |
1da177e4 LT |
389 | { |
390 | struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data; | |
391 | struct ip6_tnl *t; | |
392 | int rel_msg = 0; | |
393 | int rel_type = ICMPV6_DEST_UNREACH; | |
394 | int rel_code = ICMPV6_ADDR_UNREACH; | |
395 | __u32 rel_info = 0; | |
396 | __u16 len; | |
d2acc347 | 397 | int err = -ENOENT; |
1da177e4 | 398 | |
1ab1457c YH |
399 | /* If the packet doesn't contain the original IPv6 header we are |
400 | in trouble since we might need the source address for further | |
1da177e4 LT |
401 | processing of the error. */ |
402 | ||
3144581c YK |
403 | read_lock(&ip6_tnl_lock); |
404 | if ((t = ip6_tnl_lookup(&ipv6h->daddr, &ipv6h->saddr)) == NULL) | |
1da177e4 LT |
405 | goto out; |
406 | ||
502b0935 YK |
407 | if (t->parms.proto != ipproto && t->parms.proto != 0) |
408 | goto out; | |
409 | ||
d2acc347 HX |
410 | err = 0; |
411 | ||
e490d1d8 | 412 | switch (*type) { |
1da177e4 LT |
413 | __u32 teli; |
414 | struct ipv6_tlv_tnl_enc_lim *tel; | |
415 | __u32 mtu; | |
416 | case ICMPV6_DEST_UNREACH: | |
417 | if (net_ratelimit()) | |
418 | printk(KERN_WARNING | |
419 | "%s: Path to destination invalid " | |
420 | "or inactive!\n", t->parms.name); | |
421 | rel_msg = 1; | |
422 | break; | |
423 | case ICMPV6_TIME_EXCEED: | |
e490d1d8 | 424 | if ((*code) == ICMPV6_EXC_HOPLIMIT) { |
1da177e4 LT |
425 | if (net_ratelimit()) |
426 | printk(KERN_WARNING | |
427 | "%s: Too small hop limit or " | |
1ab1457c | 428 | "routing loop in tunnel!\n", |
1da177e4 LT |
429 | t->parms.name); |
430 | rel_msg = 1; | |
431 | } | |
432 | break; | |
433 | case ICMPV6_PARAMPROB: | |
107a5fe6 | 434 | teli = 0; |
e490d1d8 | 435 | if ((*code) == ICMPV6_HDR_FIELD) |
107a5fe6 | 436 | teli = parse_tlv_tnl_enc_lim(skb, skb->data); |
1da177e4 | 437 | |
e490d1d8 | 438 | if (teli && teli == ntohl(*info) - 2) { |
1da177e4 LT |
439 | tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli]; |
440 | if (tel->encap_limit == 0) { | |
441 | if (net_ratelimit()) | |
442 | printk(KERN_WARNING | |
443 | "%s: Too small encapsulation " | |
444 | "limit or routing loop in " | |
445 | "tunnel!\n", t->parms.name); | |
446 | rel_msg = 1; | |
447 | } | |
107a5fe6 VN |
448 | } else if (net_ratelimit()) { |
449 | printk(KERN_WARNING | |
450 | "%s: Recipient unable to parse tunneled " | |
451 | "packet!\n ", t->parms.name); | |
1da177e4 LT |
452 | } |
453 | break; | |
454 | case ICMPV6_PKT_TOOBIG: | |
e490d1d8 | 455 | mtu = ntohl(*info) - offset; |
1da177e4 LT |
456 | if (mtu < IPV6_MIN_MTU) |
457 | mtu = IPV6_MIN_MTU; | |
458 | t->dev->mtu = mtu; | |
459 | ||
cc6cdac0 | 460 | if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) { |
1da177e4 LT |
461 | rel_type = ICMPV6_PKT_TOOBIG; |
462 | rel_code = 0; | |
463 | rel_info = mtu; | |
464 | rel_msg = 1; | |
465 | } | |
466 | break; | |
467 | } | |
e490d1d8 YK |
468 | |
469 | *type = rel_type; | |
470 | *code = rel_code; | |
471 | *info = rel_info; | |
472 | *msg = rel_msg; | |
473 | ||
474 | out: | |
3144581c | 475 | read_unlock(&ip6_tnl_lock); |
e490d1d8 YK |
476 | return err; |
477 | } | |
478 | ||
c4d3efaf YK |
479 | static int |
480 | ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt, | |
481 | int type, int code, int offset, __u32 info) | |
482 | { | |
483 | int rel_msg = 0; | |
484 | int rel_type = type; | |
485 | int rel_code = code; | |
486 | __u32 rel_info = info; | |
487 | int err; | |
488 | struct sk_buff *skb2; | |
489 | struct iphdr *eiph; | |
490 | struct flowi fl; | |
491 | struct rtable *rt; | |
492 | ||
502b0935 YK |
493 | err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code, |
494 | &rel_msg, &rel_info, offset); | |
c4d3efaf YK |
495 | if (err < 0) |
496 | return err; | |
497 | ||
498 | if (rel_msg == 0) | |
499 | return 0; | |
500 | ||
501 | switch (rel_type) { | |
502 | case ICMPV6_DEST_UNREACH: | |
503 | if (rel_code != ICMPV6_ADDR_UNREACH) | |
504 | return 0; | |
505 | rel_type = ICMP_DEST_UNREACH; | |
506 | rel_code = ICMP_HOST_UNREACH; | |
507 | break; | |
508 | case ICMPV6_PKT_TOOBIG: | |
509 | if (rel_code != 0) | |
510 | return 0; | |
511 | rel_type = ICMP_DEST_UNREACH; | |
512 | rel_code = ICMP_FRAG_NEEDED; | |
513 | break; | |
514 | default: | |
515 | return 0; | |
516 | } | |
517 | ||
518 | if (!pskb_may_pull(skb, offset + sizeof(struct iphdr))) | |
519 | return 0; | |
520 | ||
521 | skb2 = skb_clone(skb, GFP_ATOMIC); | |
522 | if (!skb2) | |
523 | return 0; | |
524 | ||
525 | dst_release(skb2->dst); | |
526 | skb2->dst = NULL; | |
527 | skb_pull(skb2, offset); | |
c1d2bbe1 | 528 | skb_reset_network_header(skb2); |
c4d3efaf YK |
529 | eiph = skb2->nh.iph; |
530 | ||
531 | /* Try to guess incoming interface */ | |
532 | memset(&fl, 0, sizeof(fl)); | |
533 | fl.fl4_dst = eiph->saddr; | |
534 | fl.fl4_tos = RT_TOS(eiph->tos); | |
535 | fl.proto = IPPROTO_IPIP; | |
536 | if (ip_route_output_key(&rt, &fl)) | |
537 | goto out; | |
538 | ||
539 | skb2->dev = rt->u.dst.dev; | |
540 | ||
541 | /* route "incoming" packet */ | |
542 | if (rt->rt_flags & RTCF_LOCAL) { | |
543 | ip_rt_put(rt); | |
544 | rt = NULL; | |
545 | fl.fl4_dst = eiph->daddr; | |
546 | fl.fl4_src = eiph->saddr; | |
547 | fl.fl4_tos = eiph->tos; | |
548 | if (ip_route_output_key(&rt, &fl) || | |
549 | rt->u.dst.dev->type != ARPHRD_TUNNEL) { | |
550 | ip_rt_put(rt); | |
551 | goto out; | |
552 | } | |
553 | } else { | |
554 | ip_rt_put(rt); | |
555 | if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos, | |
556 | skb2->dev) || | |
557 | skb2->dst->dev->type != ARPHRD_TUNNEL) | |
558 | goto out; | |
559 | } | |
560 | ||
561 | /* change mtu on this route */ | |
562 | if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) { | |
563 | if (rel_info > dst_mtu(skb2->dst)) | |
564 | goto out; | |
565 | ||
566 | skb2->dst->ops->update_pmtu(skb2->dst, rel_info); | |
567 | rel_info = htonl(rel_info); | |
568 | } | |
569 | ||
570 | icmp_send(skb2, rel_type, rel_code, rel_info); | |
571 | ||
572 | out: | |
573 | kfree_skb(skb2); | |
574 | return 0; | |
575 | } | |
576 | ||
e490d1d8 YK |
577 | static int |
578 | ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt, | |
579 | int type, int code, int offset, __u32 info) | |
580 | { | |
581 | int rel_msg = 0; | |
582 | int rel_type = type; | |
583 | int rel_code = code; | |
584 | __u32 rel_info = info; | |
585 | int err; | |
586 | ||
502b0935 YK |
587 | err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code, |
588 | &rel_msg, &rel_info, offset); | |
e490d1d8 YK |
589 | if (err < 0) |
590 | return err; | |
591 | ||
592 | if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) { | |
1da177e4 LT |
593 | struct rt6_info *rt; |
594 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
305d4b3c | 595 | |
1da177e4 | 596 | if (!skb2) |
e490d1d8 | 597 | return 0; |
1da177e4 LT |
598 | |
599 | dst_release(skb2->dst); | |
600 | skb2->dst = NULL; | |
601 | skb_pull(skb2, offset); | |
c1d2bbe1 | 602 | skb_reset_network_header(skb2); |
1da177e4 LT |
603 | |
604 | /* Try to guess incoming interface */ | |
605 | rt = rt6_lookup(&skb2->nh.ipv6h->saddr, NULL, 0, 0); | |
606 | ||
607 | if (rt && rt->rt6i_dev) | |
608 | skb2->dev = rt->rt6i_dev; | |
609 | ||
610 | icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev); | |
611 | ||
612 | if (rt) | |
613 | dst_release(&rt->u.dst); | |
614 | ||
615 | kfree_skb(skb2); | |
616 | } | |
e490d1d8 YK |
617 | |
618 | return 0; | |
1da177e4 LT |
619 | } |
620 | ||
c4d3efaf YK |
621 | static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t, |
622 | struct ipv6hdr *ipv6h, | |
623 | struct sk_buff *skb) | |
624 | { | |
625 | __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK; | |
626 | ||
627 | if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY) | |
628 | ipv4_change_dsfield(skb->nh.iph, INET_ECN_MASK, dsfield); | |
629 | ||
630 | if (INET_ECN_is_ce(dsfield)) | |
631 | IP_ECN_set_ce(skb->nh.iph); | |
632 | } | |
633 | ||
8359925b YK |
634 | static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t, |
635 | struct ipv6hdr *ipv6h, | |
636 | struct sk_buff *skb) | |
1da177e4 | 637 | { |
8359925b YK |
638 | if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY) |
639 | ipv6_copy_dscp(ipv6h, skb->nh.ipv6h); | |
1da177e4 | 640 | |
8359925b YK |
641 | if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h))) |
642 | IP6_ECN_set_ce(skb->nh.ipv6h); | |
1da177e4 | 643 | } |
8359925b | 644 | |
09c6bbf0 VN |
645 | static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t) |
646 | { | |
647 | struct ip6_tnl_parm *p = &t->parms; | |
648 | int ret = 0; | |
649 | ||
650 | if (p->flags & IP6_TNL_F_CAP_RCV) { | |
1ab1457c | 651 | struct net_device *ldev = NULL; |
09c6bbf0 VN |
652 | |
653 | if (p->link) | |
654 | ldev = dev_get_by_index(p->link); | |
655 | ||
656 | if ((ipv6_addr_is_multicast(&p->laddr) || | |
657 | likely(ipv6_chk_addr(&p->laddr, ldev, 0))) && | |
658 | likely(!ipv6_chk_addr(&p->raddr, NULL, 0))) | |
659 | ret = 1; | |
660 | ||
661 | if (ldev) | |
662 | dev_put(ldev); | |
663 | } | |
664 | return ret; | |
665 | } | |
1da177e4 LT |
666 | |
667 | /** | |
3144581c | 668 | * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally |
1da177e4 | 669 | * @skb: received socket buffer |
8359925b YK |
670 | * @protocol: ethernet protocol ID |
671 | * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN | |
1da177e4 LT |
672 | * |
673 | * Return: 0 | |
674 | **/ | |
675 | ||
8359925b | 676 | static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol, |
502b0935 | 677 | __u8 ipproto, |
8359925b YK |
678 | void (*dscp_ecn_decapsulate)(struct ip6_tnl *t, |
679 | struct ipv6hdr *ipv6h, | |
680 | struct sk_buff *skb)) | |
1da177e4 | 681 | { |
1da177e4 LT |
682 | struct ipv6hdr *ipv6h; |
683 | struct ip6_tnl *t; | |
684 | ||
1da177e4 LT |
685 | ipv6h = skb->nh.ipv6h; |
686 | ||
3144581c | 687 | read_lock(&ip6_tnl_lock); |
1da177e4 | 688 | |
3144581c | 689 | if ((t = ip6_tnl_lookup(&ipv6h->saddr, &ipv6h->daddr)) != NULL) { |
502b0935 YK |
690 | if (t->parms.proto != ipproto && t->parms.proto != 0) { |
691 | read_unlock(&ip6_tnl_lock); | |
692 | goto discard; | |
693 | } | |
694 | ||
1da177e4 | 695 | if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) { |
3144581c | 696 | read_unlock(&ip6_tnl_lock); |
50fba2aa | 697 | goto discard; |
1da177e4 LT |
698 | } |
699 | ||
09c6bbf0 | 700 | if (!ip6_tnl_rcv_ctl(t)) { |
1da177e4 | 701 | t->stat.rx_dropped++; |
3144581c | 702 | read_unlock(&ip6_tnl_lock); |
1da177e4 LT |
703 | goto discard; |
704 | } | |
705 | secpath_reset(skb); | |
706 | skb->mac.raw = skb->nh.raw; | |
c1d2bbe1 | 707 | skb_reset_network_header(skb); |
8359925b | 708 | skb->protocol = htons(protocol); |
1da177e4 LT |
709 | skb->pkt_type = PACKET_HOST; |
710 | memset(skb->cb, 0, sizeof(struct inet6_skb_parm)); | |
711 | skb->dev = t->dev; | |
712 | dst_release(skb->dst); | |
713 | skb->dst = NULL; | |
53ab61c6 | 714 | nf_reset(skb); |
8359925b YK |
715 | |
716 | dscp_ecn_decapsulate(t, ipv6h, skb); | |
717 | ||
1da177e4 LT |
718 | t->stat.rx_packets++; |
719 | t->stat.rx_bytes += skb->len; | |
720 | netif_rx(skb); | |
3144581c | 721 | read_unlock(&ip6_tnl_lock); |
1da177e4 LT |
722 | return 0; |
723 | } | |
3144581c | 724 | read_unlock(&ip6_tnl_lock); |
1da177e4 | 725 | return 1; |
50fba2aa HX |
726 | |
727 | discard: | |
728 | kfree_skb(skb); | |
729 | return 0; | |
1da177e4 LT |
730 | } |
731 | ||
c4d3efaf YK |
732 | static int ip4ip6_rcv(struct sk_buff *skb) |
733 | { | |
502b0935 YK |
734 | return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP, |
735 | ip4ip6_dscp_ecn_decapsulate); | |
c4d3efaf YK |
736 | } |
737 | ||
8359925b YK |
738 | static int ip6ip6_rcv(struct sk_buff *skb) |
739 | { | |
502b0935 YK |
740 | return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6, |
741 | ip6ip6_dscp_ecn_decapsulate); | |
8359925b YK |
742 | } |
743 | ||
6fb32dde VN |
744 | struct ipv6_tel_txoption { |
745 | struct ipv6_txoptions ops; | |
746 | __u8 dst_opt[8]; | |
747 | }; | |
1da177e4 | 748 | |
6fb32dde VN |
749 | static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit) |
750 | { | |
751 | memset(opt, 0, sizeof(struct ipv6_tel_txoption)); | |
1da177e4 | 752 | |
6fb32dde VN |
753 | opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT; |
754 | opt->dst_opt[3] = 1; | |
755 | opt->dst_opt[4] = encap_limit; | |
756 | opt->dst_opt[5] = IPV6_TLV_PADN; | |
757 | opt->dst_opt[6] = 1; | |
1da177e4 | 758 | |
6fb32dde VN |
759 | opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt; |
760 | opt->ops.opt_nflen = 8; | |
1da177e4 LT |
761 | } |
762 | ||
763 | /** | |
3144581c | 764 | * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own |
1da177e4 | 765 | * @t: the outgoing tunnel device |
1ab1457c | 766 | * @hdr: IPv6 header from the incoming packet |
1da177e4 LT |
767 | * |
768 | * Description: | |
1ab1457c | 769 | * Avoid trivial tunneling loop by checking that tunnel exit-point |
1da177e4 LT |
770 | * doesn't match source of incoming packet. |
771 | * | |
1ab1457c | 772 | * Return: |
1da177e4 LT |
773 | * 1 if conflict, |
774 | * 0 else | |
775 | **/ | |
776 | ||
777 | static inline int | |
3144581c | 778 | ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr) |
1da177e4 LT |
779 | { |
780 | return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr); | |
781 | } | |
782 | ||
09c6bbf0 VN |
783 | static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t) |
784 | { | |
785 | struct ip6_tnl_parm *p = &t->parms; | |
786 | int ret = 0; | |
787 | ||
1ab1457c | 788 | if (p->flags & IP6_TNL_F_CAP_XMIT) { |
09c6bbf0 VN |
789 | struct net_device *ldev = NULL; |
790 | ||
791 | if (p->link) | |
792 | ldev = dev_get_by_index(p->link); | |
793 | ||
794 | if (unlikely(!ipv6_chk_addr(&p->laddr, ldev, 0))) | |
795 | printk(KERN_WARNING | |
796 | "%s xmit: Local address not yet configured!\n", | |
797 | p->name); | |
798 | else if (!ipv6_addr_is_multicast(&p->raddr) && | |
799 | unlikely(ipv6_chk_addr(&p->raddr, NULL, 0))) | |
800 | printk(KERN_WARNING | |
801 | "%s xmit: Routing loop! " | |
802 | "Remote address found on this node!\n", | |
803 | p->name); | |
804 | else | |
805 | ret = 1; | |
806 | if (ldev) | |
807 | dev_put(ldev); | |
808 | } | |
809 | return ret; | |
810 | } | |
1da177e4 | 811 | /** |
61ec2aec | 812 | * ip6_tnl_xmit2 - encapsulate packet and send |
1da177e4 | 813 | * @skb: the outgoing socket buffer |
1ab1457c | 814 | * @dev: the outgoing tunnel device |
61ec2aec YK |
815 | * @dsfield: dscp code for outer header |
816 | * @fl: flow of tunneled packet | |
817 | * @encap_limit: encapsulation limit | |
818 | * @pmtu: Path MTU is stored if packet is too big | |
1da177e4 LT |
819 | * |
820 | * Description: | |
821 | * Build new header and do some sanity checks on the packet before sending | |
822 | * it. | |
823 | * | |
1ab1457c | 824 | * Return: |
c4d3efaf | 825 | * 0 on success |
61ec2aec YK |
826 | * -1 fail |
827 | * %-EMSGSIZE message too big. return mtu in this case. | |
1da177e4 LT |
828 | **/ |
829 | ||
61ec2aec YK |
830 | static int ip6_tnl_xmit2(struct sk_buff *skb, |
831 | struct net_device *dev, | |
832 | __u8 dsfield, | |
833 | struct flowi *fl, | |
834 | int encap_limit, | |
835 | __u32 *pmtu) | |
1da177e4 | 836 | { |
2941a486 | 837 | struct ip6_tnl *t = netdev_priv(dev); |
1da177e4 LT |
838 | struct net_device_stats *stats = &t->stat; |
839 | struct ipv6hdr *ipv6h = skb->nh.ipv6h; | |
6fb32dde | 840 | struct ipv6_tel_txoption opt; |
1da177e4 LT |
841 | struct dst_entry *dst; |
842 | struct net_device *tdev; | |
843 | int mtu; | |
844 | int max_headroom = sizeof(struct ipv6hdr); | |
845 | u8 proto; | |
61ec2aec | 846 | int err = -1; |
1da177e4 | 847 | int pkt_len; |
1da177e4 | 848 | |
1da177e4 LT |
849 | if ((dst = ip6_tnl_dst_check(t)) != NULL) |
850 | dst_hold(dst); | |
a57ebc90 | 851 | else { |
61ec2aec | 852 | dst = ip6_route_output(NULL, fl); |
1da177e4 | 853 | |
61ec2aec | 854 | if (dst->error || xfrm_lookup(&dst, fl, NULL, 0) < 0) |
a57ebc90 PM |
855 | goto tx_err_link_failure; |
856 | } | |
1da177e4 LT |
857 | |
858 | tdev = dst->dev; | |
859 | ||
860 | if (tdev == dev) { | |
861 | stats->collisions++; | |
862 | if (net_ratelimit()) | |
1ab1457c | 863 | printk(KERN_WARNING |
1da177e4 LT |
864 | "%s: Local routing loop detected!\n", |
865 | t->parms.name); | |
866 | goto tx_err_dst_release; | |
867 | } | |
868 | mtu = dst_mtu(dst) - sizeof (*ipv6h); | |
6fb32dde | 869 | if (encap_limit >= 0) { |
1da177e4 LT |
870 | max_headroom += 8; |
871 | mtu -= 8; | |
872 | } | |
873 | if (mtu < IPV6_MIN_MTU) | |
874 | mtu = IPV6_MIN_MTU; | |
26892058 YK |
875 | if (skb->dst) |
876 | skb->dst->ops->update_pmtu(skb->dst, mtu); | |
1da177e4 | 877 | if (skb->len > mtu) { |
61ec2aec YK |
878 | *pmtu = mtu; |
879 | err = -EMSGSIZE; | |
1da177e4 LT |
880 | goto tx_err_dst_release; |
881 | } | |
882 | ||
883 | /* | |
884 | * Okay, now see if we can stuff it in the buffer as-is. | |
885 | */ | |
886 | max_headroom += LL_RESERVED_SPACE(tdev); | |
1ab1457c YH |
887 | |
888 | if (skb_headroom(skb) < max_headroom || | |
1da177e4 LT |
889 | skb_cloned(skb) || skb_shared(skb)) { |
890 | struct sk_buff *new_skb; | |
1ab1457c | 891 | |
1da177e4 LT |
892 | if (!(new_skb = skb_realloc_headroom(skb, max_headroom))) |
893 | goto tx_err_dst_release; | |
894 | ||
895 | if (skb->sk) | |
896 | skb_set_owner_w(new_skb, skb->sk); | |
897 | kfree_skb(skb); | |
898 | skb = new_skb; | |
899 | } | |
900 | dst_release(skb->dst); | |
901 | skb->dst = dst_clone(dst); | |
902 | ||
903 | skb->h.raw = skb->nh.raw; | |
904 | ||
61ec2aec | 905 | proto = fl->proto; |
6fb32dde VN |
906 | if (encap_limit >= 0) { |
907 | init_tel_txopt(&opt, encap_limit); | |
908 | ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL); | |
909 | } | |
e2d1bca7 ACM |
910 | skb_push(skb, sizeof(struct ipv6hdr)); |
911 | skb_reset_network_header(skb); | |
1da177e4 | 912 | ipv6h = skb->nh.ipv6h; |
61ec2aec | 913 | *(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000); |
1da177e4 LT |
914 | dsfield = INET_ECN_encapsulate(0, dsfield); |
915 | ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield); | |
916 | ipv6h->payload_len = htons(skb->len - sizeof(struct ipv6hdr)); | |
917 | ipv6h->hop_limit = t->parms.hop_limit; | |
918 | ipv6h->nexthdr = proto; | |
61ec2aec YK |
919 | ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src); |
920 | ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst); | |
1da177e4 LT |
921 | nf_reset(skb); |
922 | pkt_len = skb->len; | |
1ab1457c | 923 | err = NF_HOOK(PF_INET6, NF_IP6_LOCAL_OUT, skb, NULL, |
1da177e4 LT |
924 | skb->dst->dev, dst_output); |
925 | ||
b9df3cb8 | 926 | if (net_xmit_eval(err) == 0) { |
1da177e4 LT |
927 | stats->tx_bytes += pkt_len; |
928 | stats->tx_packets++; | |
929 | } else { | |
930 | stats->tx_errors++; | |
931 | stats->tx_aborted_errors++; | |
932 | } | |
933 | ip6_tnl_dst_store(t, dst); | |
1da177e4 LT |
934 | return 0; |
935 | tx_err_link_failure: | |
936 | stats->tx_carrier_errors++; | |
937 | dst_link_failure(skb); | |
938 | tx_err_dst_release: | |
939 | dst_release(dst); | |
61ec2aec YK |
940 | return err; |
941 | } | |
942 | ||
c4d3efaf YK |
943 | static inline int |
944 | ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev) | |
945 | { | |
946 | struct ip6_tnl *t = netdev_priv(dev); | |
947 | struct iphdr *iph = skb->nh.iph; | |
948 | int encap_limit = -1; | |
949 | struct flowi fl; | |
950 | __u8 dsfield; | |
951 | __u32 mtu; | |
952 | int err; | |
953 | ||
502b0935 YK |
954 | if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) || |
955 | !ip6_tnl_xmit_ctl(t)) | |
c4d3efaf YK |
956 | return -1; |
957 | ||
958 | if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT)) | |
959 | encap_limit = t->parms.encap_limit; | |
960 | ||
961 | memcpy(&fl, &t->fl, sizeof (fl)); | |
962 | fl.proto = IPPROTO_IPIP; | |
963 | ||
964 | dsfield = ipv4_get_dsfield(iph); | |
965 | ||
966 | if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)) | |
967 | fl.fl6_flowlabel |= ntohl(((__u32)iph->tos << IPV6_TCLASS_SHIFT) | |
968 | & IPV6_TCLASS_MASK); | |
969 | ||
970 | err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu); | |
971 | if (err != 0) { | |
972 | /* XXX: send ICMP error even if DF is not set. */ | |
973 | if (err == -EMSGSIZE) | |
974 | icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, | |
975 | htonl(mtu)); | |
976 | return -1; | |
977 | } | |
978 | ||
979 | return 0; | |
980 | } | |
981 | ||
61ec2aec YK |
982 | static inline int |
983 | ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev) | |
984 | { | |
985 | struct ip6_tnl *t = netdev_priv(dev); | |
986 | struct ipv6hdr *ipv6h = skb->nh.ipv6h; | |
987 | int encap_limit = -1; | |
988 | __u16 offset; | |
989 | struct flowi fl; | |
990 | __u8 dsfield; | |
991 | __u32 mtu; | |
992 | int err; | |
993 | ||
502b0935 YK |
994 | if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) || |
995 | !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h)) | |
61ec2aec YK |
996 | return -1; |
997 | ||
d56f90a7 ACM |
998 | offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb)); |
999 | if (offset > 0) { | |
61ec2aec | 1000 | struct ipv6_tlv_tnl_enc_lim *tel; |
d56f90a7 | 1001 | tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset]; |
61ec2aec YK |
1002 | if (tel->encap_limit == 0) { |
1003 | icmpv6_send(skb, ICMPV6_PARAMPROB, | |
1004 | ICMPV6_HDR_FIELD, offset + 2, skb->dev); | |
1005 | return -1; | |
1006 | } | |
1007 | encap_limit = tel->encap_limit - 1; | |
1008 | } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT)) | |
1009 | encap_limit = t->parms.encap_limit; | |
1010 | ||
1011 | memcpy(&fl, &t->fl, sizeof (fl)); | |
1012 | fl.proto = IPPROTO_IPV6; | |
1013 | ||
1014 | dsfield = ipv6_get_dsfield(ipv6h); | |
1015 | if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)) | |
1016 | fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK); | |
1017 | if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL)) | |
1018 | fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK); | |
1019 | ||
1020 | err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu); | |
1021 | if (err != 0) { | |
1022 | if (err == -EMSGSIZE) | |
1023 | icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev); | |
1024 | return -1; | |
1025 | } | |
1026 | ||
1027 | return 0; | |
1028 | } | |
1029 | ||
1030 | static int | |
1031 | ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev) | |
1032 | { | |
1033 | struct ip6_tnl *t = netdev_priv(dev); | |
1034 | struct net_device_stats *stats = &t->stat; | |
1035 | int ret; | |
1036 | ||
1037 | if (t->recursion++) { | |
1038 | t->stat.collisions++; | |
1039 | goto tx_err; | |
1040 | } | |
1041 | ||
1042 | switch (skb->protocol) { | |
c4d3efaf YK |
1043 | case __constant_htons(ETH_P_IP): |
1044 | ret = ip4ip6_tnl_xmit(skb, dev); | |
1045 | break; | |
61ec2aec YK |
1046 | case __constant_htons(ETH_P_IPV6): |
1047 | ret = ip6ip6_tnl_xmit(skb, dev); | |
1048 | break; | |
1049 | default: | |
1050 | goto tx_err; | |
1051 | } | |
1052 | ||
1053 | if (ret < 0) | |
1054 | goto tx_err; | |
1055 | ||
1056 | t->recursion--; | |
1057 | return 0; | |
1058 | ||
1da177e4 LT |
1059 | tx_err: |
1060 | stats->tx_errors++; | |
1061 | stats->tx_dropped++; | |
1062 | kfree_skb(skb); | |
1063 | t->recursion--; | |
1064 | return 0; | |
1065 | } | |
1066 | ||
1067 | static void ip6_tnl_set_cap(struct ip6_tnl *t) | |
1068 | { | |
1069 | struct ip6_tnl_parm *p = &t->parms; | |
09c6bbf0 VN |
1070 | int ltype = ipv6_addr_type(&p->laddr); |
1071 | int rtype = ipv6_addr_type(&p->raddr); | |
1da177e4 LT |
1072 | |
1073 | p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV); | |
1074 | ||
09c6bbf0 VN |
1075 | if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) && |
1076 | rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) && | |
1077 | !((ltype|rtype) & IPV6_ADDR_LOOPBACK) && | |
305d4b3c | 1078 | (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) { |
09c6bbf0 VN |
1079 | if (ltype&IPV6_ADDR_UNICAST) |
1080 | p->flags |= IP6_TNL_F_CAP_XMIT; | |
1081 | if (rtype&IPV6_ADDR_UNICAST) | |
1082 | p->flags |= IP6_TNL_F_CAP_RCV; | |
1da177e4 LT |
1083 | } |
1084 | } | |
1085 | ||
3144581c | 1086 | static void ip6_tnl_link_config(struct ip6_tnl *t) |
1da177e4 LT |
1087 | { |
1088 | struct net_device *dev = t->dev; | |
1089 | struct ip6_tnl_parm *p = &t->parms; | |
1090 | struct flowi *fl = &t->fl; | |
1091 | ||
1092 | memcpy(&dev->dev_addr, &p->laddr, sizeof(struct in6_addr)); | |
1093 | memcpy(&dev->broadcast, &p->raddr, sizeof(struct in6_addr)); | |
1094 | ||
1095 | /* Set up flowi template */ | |
1096 | ipv6_addr_copy(&fl->fl6_src, &p->laddr); | |
1097 | ipv6_addr_copy(&fl->fl6_dst, &p->raddr); | |
1098 | fl->oif = p->link; | |
1099 | fl->fl6_flowlabel = 0; | |
1100 | ||
1101 | if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS)) | |
1102 | fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo; | |
1103 | if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL)) | |
1104 | fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo; | |
1105 | ||
1106 | ip6_tnl_set_cap(t); | |
1107 | ||
1108 | if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV) | |
1109 | dev->flags |= IFF_POINTOPOINT; | |
1110 | else | |
1111 | dev->flags &= ~IFF_POINTOPOINT; | |
1112 | ||
1113 | dev->iflink = p->link; | |
1114 | ||
1115 | if (p->flags & IP6_TNL_F_CAP_XMIT) { | |
305d4b3c VN |
1116 | int strict = (ipv6_addr_type(&p->raddr) & |
1117 | (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL)); | |
1118 | ||
1da177e4 | 1119 | struct rt6_info *rt = rt6_lookup(&p->raddr, &p->laddr, |
305d4b3c | 1120 | p->link, strict); |
1da177e4 LT |
1121 | |
1122 | if (rt == NULL) | |
1123 | return; | |
1124 | ||
1125 | if (rt->rt6i_dev) { | |
1126 | dev->hard_header_len = rt->rt6i_dev->hard_header_len + | |
1127 | sizeof (struct ipv6hdr); | |
1128 | ||
1129 | dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr); | |
1130 | ||
1131 | if (dev->mtu < IPV6_MIN_MTU) | |
1132 | dev->mtu = IPV6_MIN_MTU; | |
1133 | } | |
1134 | dst_release(&rt->u.dst); | |
1135 | } | |
1136 | } | |
1137 | ||
1138 | /** | |
3144581c | 1139 | * ip6_tnl_change - update the tunnel parameters |
1da177e4 LT |
1140 | * @t: tunnel to be changed |
1141 | * @p: tunnel configuration parameters | |
1142 | * @active: != 0 if tunnel is ready for use | |
1143 | * | |
1144 | * Description: | |
3144581c | 1145 | * ip6_tnl_change() updates the tunnel parameters |
1da177e4 LT |
1146 | **/ |
1147 | ||
1148 | static int | |
3144581c | 1149 | ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p) |
1da177e4 LT |
1150 | { |
1151 | ipv6_addr_copy(&t->parms.laddr, &p->laddr); | |
1152 | ipv6_addr_copy(&t->parms.raddr, &p->raddr); | |
1153 | t->parms.flags = p->flags; | |
1154 | t->parms.hop_limit = p->hop_limit; | |
1155 | t->parms.encap_limit = p->encap_limit; | |
1156 | t->parms.flowinfo = p->flowinfo; | |
8181b8c1 | 1157 | t->parms.link = p->link; |
502b0935 | 1158 | t->parms.proto = p->proto; |
0c088890 | 1159 | ip6_tnl_dst_reset(t); |
3144581c | 1160 | ip6_tnl_link_config(t); |
1da177e4 LT |
1161 | return 0; |
1162 | } | |
1163 | ||
1164 | /** | |
3144581c | 1165 | * ip6_tnl_ioctl - configure ipv6 tunnels from userspace |
1da177e4 LT |
1166 | * @dev: virtual device associated with tunnel |
1167 | * @ifr: parameters passed from userspace | |
1168 | * @cmd: command to be performed | |
1169 | * | |
1170 | * Description: | |
3144581c | 1171 | * ip6_tnl_ioctl() is used for managing IPv6 tunnels |
1ab1457c | 1172 | * from userspace. |
1da177e4 LT |
1173 | * |
1174 | * The possible commands are the following: | |
1175 | * %SIOCGETTUNNEL: get tunnel parameters for device | |
1176 | * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters | |
1177 | * %SIOCCHGTUNNEL: change tunnel parameters to those given | |
1178 | * %SIOCDELTUNNEL: delete tunnel | |
1179 | * | |
1ab1457c | 1180 | * The fallback device "ip6tnl0", created during module |
1da177e4 LT |
1181 | * initialization, can be used for creating other tunnel devices. |
1182 | * | |
1183 | * Return: | |
1184 | * 0 on success, | |
1185 | * %-EFAULT if unable to copy data to or from userspace, | |
1186 | * %-EPERM if current process hasn't %CAP_NET_ADMIN set | |
1187 | * %-EINVAL if passed tunnel parameters are invalid, | |
1188 | * %-EEXIST if changing a tunnel's parameters would cause a conflict | |
1189 | * %-ENODEV if attempting to change or delete a nonexisting device | |
1190 | **/ | |
1191 | ||
1192 | static int | |
3144581c | 1193 | ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd) |
1da177e4 LT |
1194 | { |
1195 | int err = 0; | |
1da177e4 LT |
1196 | struct ip6_tnl_parm p; |
1197 | struct ip6_tnl *t = NULL; | |
1198 | ||
1199 | switch (cmd) { | |
1200 | case SIOCGETTUNNEL: | |
3144581c | 1201 | if (dev == ip6_fb_tnl_dev) { |
567131a7 | 1202 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) { |
1da177e4 LT |
1203 | err = -EFAULT; |
1204 | break; | |
1205 | } | |
3144581c | 1206 | t = ip6_tnl_locate(&p, 0); |
567131a7 VN |
1207 | } |
1208 | if (t == NULL) | |
2941a486 | 1209 | t = netdev_priv(dev); |
1da177e4 LT |
1210 | memcpy(&p, &t->parms, sizeof (p)); |
1211 | if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) { | |
1212 | err = -EFAULT; | |
1213 | } | |
1214 | break; | |
1215 | case SIOCADDTUNNEL: | |
1216 | case SIOCCHGTUNNEL: | |
1217 | err = -EPERM; | |
1da177e4 LT |
1218 | if (!capable(CAP_NET_ADMIN)) |
1219 | break; | |
567131a7 VN |
1220 | err = -EFAULT; |
1221 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) | |
1da177e4 | 1222 | break; |
567131a7 | 1223 | err = -EINVAL; |
502b0935 YK |
1224 | if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP && |
1225 | p.proto != 0) | |
1da177e4 | 1226 | break; |
3144581c YK |
1227 | t = ip6_tnl_locate(&p, cmd == SIOCADDTUNNEL); |
1228 | if (dev != ip6_fb_tnl_dev && cmd == SIOCCHGTUNNEL) { | |
567131a7 VN |
1229 | if (t != NULL) { |
1230 | if (t->dev != dev) { | |
1231 | err = -EEXIST; | |
1232 | break; | |
1233 | } | |
1234 | } else | |
1235 | t = netdev_priv(dev); | |
1236 | ||
3144581c YK |
1237 | ip6_tnl_unlink(t); |
1238 | err = ip6_tnl_change(t, &p); | |
1239 | ip6_tnl_link(t); | |
1da177e4 LT |
1240 | netdev_state_change(dev); |
1241 | } | |
567131a7 | 1242 | if (t) { |
1da177e4 | 1243 | err = 0; |
567131a7 VN |
1244 | if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p))) |
1245 | err = -EFAULT; | |
1246 | ||
1247 | } else | |
1248 | err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT); | |
1da177e4 LT |
1249 | break; |
1250 | case SIOCDELTUNNEL: | |
1251 | err = -EPERM; | |
1252 | if (!capable(CAP_NET_ADMIN)) | |
1253 | break; | |
1254 | ||
3144581c | 1255 | if (dev == ip6_fb_tnl_dev) { |
567131a7 VN |
1256 | err = -EFAULT; |
1257 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) | |
1da177e4 | 1258 | break; |
567131a7 | 1259 | err = -ENOENT; |
3144581c | 1260 | if ((t = ip6_tnl_locate(&p, 0)) == NULL) |
1da177e4 | 1261 | break; |
567131a7 | 1262 | err = -EPERM; |
3144581c | 1263 | if (t->dev == ip6_fb_tnl_dev) |
1da177e4 | 1264 | break; |
567131a7 | 1265 | dev = t->dev; |
1da177e4 | 1266 | } |
22f8cde5 SH |
1267 | err = 0; |
1268 | unregister_netdevice(dev); | |
1da177e4 LT |
1269 | break; |
1270 | default: | |
1271 | err = -EINVAL; | |
1272 | } | |
1273 | return err; | |
1274 | } | |
1275 | ||
1276 | /** | |
3144581c | 1277 | * ip6_tnl_get_stats - return the stats for tunnel device |
1da177e4 LT |
1278 | * @dev: virtual device associated with tunnel |
1279 | * | |
1280 | * Return: stats for device | |
1281 | **/ | |
1282 | ||
1283 | static struct net_device_stats * | |
3144581c | 1284 | ip6_tnl_get_stats(struct net_device *dev) |
1da177e4 | 1285 | { |
2941a486 | 1286 | return &(((struct ip6_tnl *)netdev_priv(dev))->stat); |
1da177e4 LT |
1287 | } |
1288 | ||
1289 | /** | |
3144581c | 1290 | * ip6_tnl_change_mtu - change mtu manually for tunnel device |
1da177e4 LT |
1291 | * @dev: virtual device associated with tunnel |
1292 | * @new_mtu: the new mtu | |
1293 | * | |
1294 | * Return: | |
1295 | * 0 on success, | |
1296 | * %-EINVAL if mtu too small | |
1297 | **/ | |
1298 | ||
1299 | static int | |
3144581c | 1300 | ip6_tnl_change_mtu(struct net_device *dev, int new_mtu) |
1da177e4 LT |
1301 | { |
1302 | if (new_mtu < IPV6_MIN_MTU) { | |
1303 | return -EINVAL; | |
1304 | } | |
1305 | dev->mtu = new_mtu; | |
1306 | return 0; | |
1307 | } | |
1308 | ||
1309 | /** | |
3144581c | 1310 | * ip6_tnl_dev_setup - setup virtual tunnel device |
1da177e4 LT |
1311 | * @dev: virtual device associated with tunnel |
1312 | * | |
1313 | * Description: | |
1314 | * Initialize function pointers and device parameters | |
1315 | **/ | |
1316 | ||
3144581c | 1317 | static void ip6_tnl_dev_setup(struct net_device *dev) |
1da177e4 LT |
1318 | { |
1319 | SET_MODULE_OWNER(dev); | |
3144581c | 1320 | dev->uninit = ip6_tnl_dev_uninit; |
1da177e4 | 1321 | dev->destructor = free_netdev; |
61ec2aec | 1322 | dev->hard_start_xmit = ip6_tnl_xmit; |
3144581c YK |
1323 | dev->get_stats = ip6_tnl_get_stats; |
1324 | dev->do_ioctl = ip6_tnl_ioctl; | |
1325 | dev->change_mtu = ip6_tnl_change_mtu; | |
1da177e4 LT |
1326 | |
1327 | dev->type = ARPHRD_TUNNEL6; | |
1328 | dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr); | |
1329 | dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr); | |
1330 | dev->flags |= IFF_NOARP; | |
1331 | dev->addr_len = sizeof(struct in6_addr); | |
1332 | } | |
1333 | ||
1334 | ||
1335 | /** | |
3144581c | 1336 | * ip6_tnl_dev_init_gen - general initializer for all tunnel devices |
1da177e4 LT |
1337 | * @dev: virtual device associated with tunnel |
1338 | **/ | |
1339 | ||
1340 | static inline void | |
3144581c | 1341 | ip6_tnl_dev_init_gen(struct net_device *dev) |
1da177e4 | 1342 | { |
2941a486 | 1343 | struct ip6_tnl *t = netdev_priv(dev); |
1da177e4 LT |
1344 | t->dev = dev; |
1345 | strcpy(t->parms.name, dev->name); | |
1346 | } | |
1347 | ||
1348 | /** | |
3144581c | 1349 | * ip6_tnl_dev_init - initializer for all non fallback tunnel devices |
1da177e4 LT |
1350 | * @dev: virtual device associated with tunnel |
1351 | **/ | |
1352 | ||
1353 | static int | |
3144581c | 1354 | ip6_tnl_dev_init(struct net_device *dev) |
1da177e4 | 1355 | { |
2941a486 | 1356 | struct ip6_tnl *t = netdev_priv(dev); |
3144581c YK |
1357 | ip6_tnl_dev_init_gen(dev); |
1358 | ip6_tnl_link_config(t); | |
1da177e4 LT |
1359 | return 0; |
1360 | } | |
1361 | ||
1362 | /** | |
3144581c | 1363 | * ip6_fb_tnl_dev_init - initializer for fallback tunnel device |
1da177e4 LT |
1364 | * @dev: fallback device |
1365 | * | |
1366 | * Return: 0 | |
1367 | **/ | |
1368 | ||
1ab1457c | 1369 | static int |
3144581c | 1370 | ip6_fb_tnl_dev_init(struct net_device *dev) |
1da177e4 | 1371 | { |
2941a486 | 1372 | struct ip6_tnl *t = netdev_priv(dev); |
3144581c | 1373 | ip6_tnl_dev_init_gen(dev); |
502b0935 | 1374 | t->parms.proto = IPPROTO_IPV6; |
1da177e4 LT |
1375 | dev_hold(dev); |
1376 | tnls_wc[0] = t; | |
1377 | return 0; | |
1378 | } | |
1379 | ||
c4d3efaf YK |
1380 | static struct xfrm6_tunnel ip4ip6_handler = { |
1381 | .handler = ip4ip6_rcv, | |
1382 | .err_handler = ip4ip6_err, | |
1383 | .priority = 1, | |
1384 | }; | |
1385 | ||
1da177e4 | 1386 | static struct xfrm6_tunnel ip6ip6_handler = { |
0303770d PM |
1387 | .handler = ip6ip6_rcv, |
1388 | .err_handler = ip6ip6_err, | |
d2acc347 | 1389 | .priority = 1, |
1da177e4 LT |
1390 | }; |
1391 | ||
1392 | /** | |
1393 | * ip6_tunnel_init - register protocol and reserve needed resources | |
1394 | * | |
1395 | * Return: 0 on success | |
1396 | **/ | |
1397 | ||
1398 | static int __init ip6_tunnel_init(void) | |
1399 | { | |
1400 | int err; | |
1401 | ||
c4d3efaf | 1402 | if (xfrm6_tunnel_register(&ip4ip6_handler, AF_INET)) { |
3144581c | 1403 | printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n"); |
c4d3efaf YK |
1404 | err = -EAGAIN; |
1405 | goto out; | |
1406 | } | |
1407 | ||
73d605d1 | 1408 | if (xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6)) { |
3144581c | 1409 | printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n"); |
c4d3efaf YK |
1410 | err = -EAGAIN; |
1411 | goto unreg_ip4ip6; | |
1da177e4 | 1412 | } |
3144581c YK |
1413 | ip6_fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0", |
1414 | ip6_tnl_dev_setup); | |
1da177e4 | 1415 | |
3144581c | 1416 | if (!ip6_fb_tnl_dev) { |
1da177e4 LT |
1417 | err = -ENOMEM; |
1418 | goto fail; | |
1419 | } | |
3144581c | 1420 | ip6_fb_tnl_dev->init = ip6_fb_tnl_dev_init; |
1da177e4 | 1421 | |
3144581c YK |
1422 | if ((err = register_netdev(ip6_fb_tnl_dev))) { |
1423 | free_netdev(ip6_fb_tnl_dev); | |
1da177e4 LT |
1424 | goto fail; |
1425 | } | |
1426 | return 0; | |
1427 | fail: | |
73d605d1 | 1428 | xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6); |
c4d3efaf YK |
1429 | unreg_ip4ip6: |
1430 | xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET); | |
1431 | out: | |
1da177e4 LT |
1432 | return err; |
1433 | } | |
1434 | ||
3144581c | 1435 | static void __exit ip6_tnl_destroy_tunnels(void) |
b3fdd9f1 YK |
1436 | { |
1437 | int h; | |
1438 | struct ip6_tnl *t; | |
1439 | ||
1440 | for (h = 0; h < HASH_SIZE; h++) { | |
1441 | while ((t = tnls_r_l[h]) != NULL) | |
1442 | unregister_netdevice(t->dev); | |
1443 | } | |
1444 | ||
1445 | t = tnls_wc[0]; | |
1446 | unregister_netdevice(t->dev); | |
1447 | } | |
1448 | ||
1da177e4 LT |
1449 | /** |
1450 | * ip6_tunnel_cleanup - free resources and unregister protocol | |
1451 | **/ | |
1452 | ||
1453 | static void __exit ip6_tunnel_cleanup(void) | |
1454 | { | |
c4d3efaf | 1455 | if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET)) |
3144581c | 1456 | printk(KERN_INFO "ip6_tunnel close: can't deregister ip4ip6\n"); |
c4d3efaf | 1457 | |
73d605d1 | 1458 | if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6)) |
3144581c | 1459 | printk(KERN_INFO "ip6_tunnel close: can't deregister ip6ip6\n"); |
1da177e4 | 1460 | |
b3fdd9f1 | 1461 | rtnl_lock(); |
3144581c | 1462 | ip6_tnl_destroy_tunnels(); |
b3fdd9f1 | 1463 | rtnl_unlock(); |
1da177e4 LT |
1464 | } |
1465 | ||
1466 | module_init(ip6_tunnel_init); | |
1467 | module_exit(ip6_tunnel_cleanup); |