iwlwifi: don't include iwl-dev.h from iwl-devtrace.h
[GitHub/LineageOS/android_kernel_motorola_exynos9610.git] / net / ipv6 / ip6_tunnel.c
CommitLineData
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 8 *
1da177e4 9 * Based on:
c4d3efaf 10 * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
1da177e4
LT
11 *
12 * RFC 2473
13 *
14 * This program is free software; you can redistribute it and/or
15 * modify it under the terms of the GNU General Public License
16 * as published by the Free Software Foundation; either version
17 * 2 of the License, or (at your option) any later version.
18 *
19 */
20
1da177e4 21#include <linux/module.h>
4fc268d2 22#include <linux/capability.h>
1da177e4
LT
23#include <linux/errno.h>
24#include <linux/types.h>
25#include <linux/sockios.h>
c4d3efaf 26#include <linux/icmp.h>
1da177e4
LT
27#include <linux/if.h>
28#include <linux/in.h>
29#include <linux/ip.h>
30#include <linux/if_tunnel.h>
31#include <linux/net.h>
32#include <linux/in6.h>
33#include <linux/netdevice.h>
34#include <linux/if_arp.h>
35#include <linux/icmpv6.h>
36#include <linux/init.h>
37#include <linux/route.h>
38#include <linux/rtnetlink.h>
39#include <linux/netfilter_ipv6.h>
40
41#include <asm/uaccess.h>
42#include <asm/atomic.h>
43
c4d3efaf 44#include <net/icmp.h>
1da177e4
LT
45#include <net/ip.h>
46#include <net/ipv6.h>
1da177e4
LT
47#include <net/ip6_route.h>
48#include <net/addrconf.h>
49#include <net/ip6_tunnel.h>
50#include <net/xfrm.h>
51#include <net/dsfield.h>
52#include <net/inet_ecn.h>
13eeb8e9
PE
53#include <net/net_namespace.h>
54#include <net/netns/generic.h>
1da177e4
LT
55
56MODULE_AUTHOR("Ville Nuorvala");
c4d3efaf 57MODULE_DESCRIPTION("IPv6 tunneling device");
1da177e4
LT
58MODULE_LICENSE("GPL");
59
60#define IPV6_TLV_TEL_DST_SIZE 8
61
62#ifdef IP6_TNL_DEBUG
0dc47877 63#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__)
1da177e4
LT
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
1326c3d5 77static void ip6_tnl_dev_init(struct net_device *dev);
3144581c 78static void ip6_tnl_dev_setup(struct net_device *dev);
1da177e4 79
f99189b1 80static int ip6_tnl_net_id __read_mostly;
13eeb8e9 81struct ip6_tnl_net {
15820e12
PE
82 /* the IPv6 tunnel fallback device */
83 struct net_device *fb_tnl_dev;
3e6c9fb5
PE
84 /* lists for storing tunnels in use */
85 struct ip6_tnl *tnls_r_l[HASH_SIZE];
86 struct ip6_tnl *tnls_wc[1];
87 struct ip6_tnl **tnls[2];
13eeb8e9
PE
88};
89
2922bc8a
ED
90/*
91 * Locking : hash tables are protected by RCU and a spinlock
92 */
93static DEFINE_SPINLOCK(ip6_tnl_lock);
1da177e4
LT
94
95static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
96{
97 struct dst_entry *dst = t->dst_cache;
98
1ab1457c 99 if (dst && dst->obsolete &&
1da177e4
LT
100 dst->ops->check(dst, t->dst_cookie) == NULL) {
101 t->dst_cache = NULL;
102 dst_release(dst);
103 return NULL;
104 }
105
106 return dst;
107}
108
109static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
110{
111 dst_release(t->dst_cache);
112 t->dst_cache = NULL;
113}
114
115static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
116{
117 struct rt6_info *rt = (struct rt6_info *) dst;
118 t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
119 dst_release(t->dst_cache);
120 t->dst_cache = dst;
121}
122
123/**
3144581c 124 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
1ab1457c
YH
125 * @remote: the address of the tunnel exit-point
126 * @local: the address of the tunnel entry-point
1da177e4 127 *
1ab1457c 128 * Return:
1da177e4 129 * tunnel matching given end-points if found,
1ab1457c 130 * else fallback tunnel if its device is up,
1da177e4
LT
131 * else %NULL
132 **/
133
2922bc8a
ED
134#define for_each_ip6_tunnel_rcu(start) \
135 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
136
1da177e4 137static struct ip6_tnl *
2dd02c89 138ip6_tnl_lookup(struct net *net, struct in6_addr *remote, struct in6_addr *local)
1da177e4
LT
139{
140 unsigned h0 = HASH(remote);
141 unsigned h1 = HASH(local);
142 struct ip6_tnl *t;
3e6c9fb5 143 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 144
2922bc8a 145 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[h0 ^ h1]) {
1da177e4
LT
146 if (ipv6_addr_equal(local, &t->parms.laddr) &&
147 ipv6_addr_equal(remote, &t->parms.raddr) &&
148 (t->dev->flags & IFF_UP))
149 return t;
150 }
2922bc8a
ED
151 t = rcu_dereference(ip6n->tnls_wc[0]);
152 if (t && (t->dev->flags & IFF_UP))
1da177e4
LT
153 return t;
154
155 return NULL;
156}
157
158/**
3144581c 159 * ip6_tnl_bucket - get head of list matching given tunnel parameters
1ab1457c 160 * @p: parameters containing tunnel end-points
1da177e4
LT
161 *
162 * Description:
3144581c 163 * ip6_tnl_bucket() returns the head of the list matching the
1da177e4
LT
164 * &struct in6_addr entries laddr and raddr in @p.
165 *
1ab1457c 166 * Return: head of IPv6 tunnel list
1da177e4
LT
167 **/
168
169static struct ip6_tnl **
2dd02c89 170ip6_tnl_bucket(struct ip6_tnl_net *ip6n, struct ip6_tnl_parm *p)
1da177e4
LT
171{
172 struct in6_addr *remote = &p->raddr;
173 struct in6_addr *local = &p->laddr;
174 unsigned h = 0;
175 int prio = 0;
176
177 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
178 prio = 1;
179 h = HASH(remote) ^ HASH(local);
180 }
3e6c9fb5 181 return &ip6n->tnls[prio][h];
1da177e4
LT
182}
183
184/**
3144581c 185 * ip6_tnl_link - add tunnel to hash table
1da177e4
LT
186 * @t: tunnel to be added
187 **/
188
189static void
2dd02c89 190ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 191{
2dd02c89 192 struct ip6_tnl **tp = ip6_tnl_bucket(ip6n, &t->parms);
1da177e4 193
2922bc8a 194 spin_lock_bh(&ip6_tnl_lock);
1da177e4 195 t->next = *tp;
2922bc8a
ED
196 rcu_assign_pointer(*tp, t);
197 spin_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
198}
199
200/**
3144581c 201 * ip6_tnl_unlink - remove tunnel from hash table
1da177e4
LT
202 * @t: tunnel to be removed
203 **/
204
205static void
2dd02c89 206ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4
LT
207{
208 struct ip6_tnl **tp;
209
2dd02c89 210 for (tp = ip6_tnl_bucket(ip6n, &t->parms); *tp; tp = &(*tp)->next) {
1da177e4 211 if (t == *tp) {
2922bc8a 212 spin_lock_bh(&ip6_tnl_lock);
1da177e4 213 *tp = t->next;
2922bc8a 214 spin_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
215 break;
216 }
217 }
218}
219
220/**
221 * ip6_tnl_create() - create a new tunnel
222 * @p: tunnel parameters
223 * @pt: pointer to new tunnel
224 *
225 * Description:
226 * Create tunnel matching given parameters.
1ab1457c
YH
227 *
228 * Return:
567131a7 229 * created tunnel or NULL
1da177e4
LT
230 **/
231
2dd02c89 232static struct ip6_tnl *ip6_tnl_create(struct net *net, struct ip6_tnl_parm *p)
1da177e4
LT
233{
234 struct net_device *dev;
235 struct ip6_tnl *t;
236 char name[IFNAMSIZ];
237 int err;
2dd02c89 238 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 239
34cc7ba6 240 if (p->name[0])
1da177e4 241 strlcpy(name, p->name, IFNAMSIZ);
34cc7ba6
PE
242 else
243 sprintf(name, "ip6tnl%%d");
244
3144581c 245 dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
1da177e4 246 if (dev == NULL)
567131a7 247 goto failed;
1da177e4 248
554eb277
PE
249 dev_net_set(dev, net);
250
b37d428b
PE
251 if (strchr(name, '%')) {
252 if (dev_alloc_name(dev, name) < 0)
253 goto failed_free;
254 }
255
2941a486 256 t = netdev_priv(dev);
1da177e4 257 t->parms = *p;
20461c17 258 ip6_tnl_dev_init(dev);
1da177e4 259
b37d428b
PE
260 if ((err = register_netdevice(dev)) < 0)
261 goto failed_free;
262
1da177e4 263 dev_hold(dev);
2dd02c89 264 ip6_tnl_link(ip6n, t);
567131a7 265 return t;
b37d428b
PE
266
267failed_free:
268 free_netdev(dev);
567131a7
VN
269failed:
270 return NULL;
1da177e4
LT
271}
272
273/**
3144581c 274 * ip6_tnl_locate - find or create tunnel matching given parameters
1ab1457c 275 * @p: tunnel parameters
1da177e4
LT
276 * @create: != 0 if allowed to create new tunnel if no match found
277 *
278 * Description:
3144581c 279 * ip6_tnl_locate() first tries to locate an existing tunnel
1da177e4
LT
280 * based on @parms. If this is unsuccessful, but @create is set a new
281 * tunnel device is created and registered for use.
282 *
283 * Return:
567131a7 284 * matching tunnel or NULL
1da177e4
LT
285 **/
286
2dd02c89
PE
287static struct ip6_tnl *ip6_tnl_locate(struct net *net,
288 struct ip6_tnl_parm *p, int create)
1da177e4
LT
289{
290 struct in6_addr *remote = &p->raddr;
291 struct in6_addr *local = &p->laddr;
292 struct ip6_tnl *t;
2dd02c89 293 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 294
2dd02c89 295 for (t = *ip6_tnl_bucket(ip6n, p); t; t = t->next) {
1da177e4 296 if (ipv6_addr_equal(local, &t->parms.laddr) &&
567131a7
VN
297 ipv6_addr_equal(remote, &t->parms.raddr))
298 return t;
1da177e4
LT
299 }
300 if (!create)
567131a7 301 return NULL;
2dd02c89 302 return ip6_tnl_create(net, p);
1da177e4
LT
303}
304
305/**
3144581c 306 * ip6_tnl_dev_uninit - tunnel device uninitializer
1da177e4 307 * @dev: the device to be destroyed
1ab1457c 308 *
1da177e4 309 * Description:
3144581c 310 * ip6_tnl_dev_uninit() removes tunnel from its list
1da177e4
LT
311 **/
312
313static void
3144581c 314ip6_tnl_dev_uninit(struct net_device *dev)
1da177e4 315{
2941a486 316 struct ip6_tnl *t = netdev_priv(dev);
2dd02c89
PE
317 struct net *net = dev_net(dev);
318 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 319
15820e12 320 if (dev == ip6n->fb_tnl_dev) {
2922bc8a 321 spin_lock_bh(&ip6_tnl_lock);
3e6c9fb5 322 ip6n->tnls_wc[0] = NULL;
2922bc8a 323 spin_unlock_bh(&ip6_tnl_lock);
1da177e4 324 } else {
2dd02c89 325 ip6_tnl_unlink(ip6n, t);
1da177e4
LT
326 }
327 ip6_tnl_dst_reset(t);
328 dev_put(dev);
329}
330
331/**
332 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
333 * @skb: received socket buffer
334 *
1ab1457c
YH
335 * Return:
336 * 0 if none was found,
1da177e4
LT
337 * else index to encapsulation limit
338 **/
339
340static __u16
341parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
342{
343 struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
344 __u8 nexthdr = ipv6h->nexthdr;
345 __u16 off = sizeof (*ipv6h);
346
347 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
348 __u16 optlen = 0;
349 struct ipv6_opt_hdr *hdr;
350 if (raw + off + sizeof (*hdr) > skb->data &&
351 !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
352 break;
353
354 hdr = (struct ipv6_opt_hdr *) (raw + off);
355 if (nexthdr == NEXTHDR_FRAGMENT) {
356 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
357 if (frag_hdr->frag_off)
358 break;
359 optlen = 8;
360 } else if (nexthdr == NEXTHDR_AUTH) {
361 optlen = (hdr->hdrlen + 2) << 2;
362 } else {
363 optlen = ipv6_optlen(hdr);
364 }
365 if (nexthdr == NEXTHDR_DEST) {
366 __u16 i = off + 2;
367 while (1) {
368 struct ipv6_tlv_tnl_enc_lim *tel;
369
370 /* No more room for encapsulation limit */
371 if (i + sizeof (*tel) > off + optlen)
372 break;
373
374 tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
375 /* return index of option if found and valid */
376 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
377 tel->length == 1)
378 return i;
379 /* else jump to next option */
380 if (tel->type)
381 i += tel->length + 2;
382 else
383 i++;
384 }
385 }
386 nexthdr = hdr->nexthdr;
387 off += optlen;
388 }
389 return 0;
390}
391
392/**
e490d1d8 393 * ip6_tnl_err - tunnel error handler
1da177e4
LT
394 *
395 * Description:
e490d1d8 396 * ip6_tnl_err() should handle errors in the tunnel according
1da177e4
LT
397 * to the specifications in RFC 2473.
398 **/
399
d2acc347 400static int
502b0935 401ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
d5fdd6ba 402 u8 *type, u8 *code, int *msg, __u32 *info, int offset)
1da177e4
LT
403{
404 struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
405 struct ip6_tnl *t;
406 int rel_msg = 0;
d5fdd6ba
BH
407 u8 rel_type = ICMPV6_DEST_UNREACH;
408 u8 rel_code = ICMPV6_ADDR_UNREACH;
1da177e4
LT
409 __u32 rel_info = 0;
410 __u16 len;
d2acc347 411 int err = -ENOENT;
1da177e4 412
1ab1457c
YH
413 /* If the packet doesn't contain the original IPv6 header we are
414 in trouble since we might need the source address for further
1da177e4
LT
415 processing of the error. */
416
2922bc8a 417 rcu_read_lock();
8704ca7e 418 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
2dd02c89 419 &ipv6h->saddr)) == NULL)
1da177e4
LT
420 goto out;
421
502b0935
YK
422 if (t->parms.proto != ipproto && t->parms.proto != 0)
423 goto out;
424
d2acc347
HX
425 err = 0;
426
e490d1d8 427 switch (*type) {
1da177e4
LT
428 __u32 teli;
429 struct ipv6_tlv_tnl_enc_lim *tel;
430 __u32 mtu;
431 case ICMPV6_DEST_UNREACH:
432 if (net_ratelimit())
433 printk(KERN_WARNING
434 "%s: Path to destination invalid "
435 "or inactive!\n", t->parms.name);
436 rel_msg = 1;
437 break;
438 case ICMPV6_TIME_EXCEED:
e490d1d8 439 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
1da177e4
LT
440 if (net_ratelimit())
441 printk(KERN_WARNING
442 "%s: Too small hop limit or "
1ab1457c 443 "routing loop in tunnel!\n",
1da177e4
LT
444 t->parms.name);
445 rel_msg = 1;
446 }
447 break;
448 case ICMPV6_PARAMPROB:
107a5fe6 449 teli = 0;
e490d1d8 450 if ((*code) == ICMPV6_HDR_FIELD)
107a5fe6 451 teli = parse_tlv_tnl_enc_lim(skb, skb->data);
1da177e4 452
704eae1f 453 if (teli && teli == *info - 2) {
1da177e4
LT
454 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
455 if (tel->encap_limit == 0) {
456 if (net_ratelimit())
457 printk(KERN_WARNING
458 "%s: Too small encapsulation "
459 "limit or routing loop in "
460 "tunnel!\n", t->parms.name);
461 rel_msg = 1;
462 }
107a5fe6
VN
463 } else if (net_ratelimit()) {
464 printk(KERN_WARNING
465 "%s: Recipient unable to parse tunneled "
466 "packet!\n ", t->parms.name);
1da177e4
LT
467 }
468 break;
469 case ICMPV6_PKT_TOOBIG:
704eae1f 470 mtu = *info - offset;
1da177e4
LT
471 if (mtu < IPV6_MIN_MTU)
472 mtu = IPV6_MIN_MTU;
473 t->dev->mtu = mtu;
474
cc6cdac0 475 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
1da177e4
LT
476 rel_type = ICMPV6_PKT_TOOBIG;
477 rel_code = 0;
478 rel_info = mtu;
479 rel_msg = 1;
480 }
481 break;
482 }
e490d1d8
YK
483
484 *type = rel_type;
485 *code = rel_code;
486 *info = rel_info;
487 *msg = rel_msg;
488
489out:
2922bc8a 490 rcu_read_unlock();
e490d1d8
YK
491 return err;
492}
493
c4d3efaf
YK
494static int
495ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 496 u8 type, u8 code, int offset, __be32 info)
c4d3efaf
YK
497{
498 int rel_msg = 0;
d5fdd6ba
BH
499 u8 rel_type = type;
500 u8 rel_code = code;
704eae1f 501 __u32 rel_info = ntohl(info);
c4d3efaf
YK
502 int err;
503 struct sk_buff *skb2;
504 struct iphdr *eiph;
505 struct flowi fl;
506 struct rtable *rt;
507
502b0935
YK
508 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
509 &rel_msg, &rel_info, offset);
c4d3efaf
YK
510 if (err < 0)
511 return err;
512
513 if (rel_msg == 0)
514 return 0;
515
516 switch (rel_type) {
517 case ICMPV6_DEST_UNREACH:
518 if (rel_code != ICMPV6_ADDR_UNREACH)
519 return 0;
520 rel_type = ICMP_DEST_UNREACH;
521 rel_code = ICMP_HOST_UNREACH;
522 break;
523 case ICMPV6_PKT_TOOBIG:
524 if (rel_code != 0)
525 return 0;
526 rel_type = ICMP_DEST_UNREACH;
527 rel_code = ICMP_FRAG_NEEDED;
528 break;
529 default:
530 return 0;
531 }
532
533 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
534 return 0;
535
536 skb2 = skb_clone(skb, GFP_ATOMIC);
537 if (!skb2)
538 return 0;
539
adf30907
ED
540 skb_dst_drop(skb2);
541
c4d3efaf 542 skb_pull(skb2, offset);
c1d2bbe1 543 skb_reset_network_header(skb2);
eddc9ec5 544 eiph = ip_hdr(skb2);
c4d3efaf
YK
545
546 /* Try to guess incoming interface */
547 memset(&fl, 0, sizeof(fl));
548 fl.fl4_dst = eiph->saddr;
549 fl.fl4_tos = RT_TOS(eiph->tos);
550 fl.proto = IPPROTO_IPIP;
2f7f54b7 551 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl))
c4d3efaf
YK
552 goto out;
553
554 skb2->dev = rt->u.dst.dev;
555
556 /* route "incoming" packet */
557 if (rt->rt_flags & RTCF_LOCAL) {
558 ip_rt_put(rt);
559 rt = NULL;
560 fl.fl4_dst = eiph->daddr;
561 fl.fl4_src = eiph->saddr;
562 fl.fl4_tos = eiph->tos;
2f7f54b7 563 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
c4d3efaf
YK
564 rt->u.dst.dev->type != ARPHRD_TUNNEL) {
565 ip_rt_put(rt);
566 goto out;
567 }
adf30907 568 skb_dst_set(skb2, (struct dst_entry *)rt);
c4d3efaf
YK
569 } else {
570 ip_rt_put(rt);
571 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
572 skb2->dev) ||
adf30907 573 skb_dst(skb2)->dev->type != ARPHRD_TUNNEL)
c4d3efaf
YK
574 goto out;
575 }
576
577 /* change mtu on this route */
578 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
adf30907 579 if (rel_info > dst_mtu(skb_dst(skb2)))
c4d3efaf
YK
580 goto out;
581
adf30907 582 skb_dst(skb2)->ops->update_pmtu(skb_dst(skb2), rel_info);
c4d3efaf
YK
583 }
584
704eae1f 585 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
c4d3efaf
YK
586
587out:
588 kfree_skb(skb2);
589 return 0;
590}
591
e490d1d8
YK
592static int
593ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 594 u8 type, u8 code, int offset, __be32 info)
e490d1d8
YK
595{
596 int rel_msg = 0;
d5fdd6ba
BH
597 u8 rel_type = type;
598 u8 rel_code = code;
704eae1f 599 __u32 rel_info = ntohl(info);
e490d1d8
YK
600 int err;
601
502b0935
YK
602 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
603 &rel_msg, &rel_info, offset);
e490d1d8
YK
604 if (err < 0)
605 return err;
606
607 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
1da177e4
LT
608 struct rt6_info *rt;
609 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 610
1da177e4 611 if (!skb2)
e490d1d8 612 return 0;
1da177e4 613
adf30907 614 skb_dst_drop(skb2);
1da177e4 615 skb_pull(skb2, offset);
c1d2bbe1 616 skb_reset_network_header(skb2);
1da177e4
LT
617
618 /* Try to guess incoming interface */
2f7f54b7
PE
619 rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
620 NULL, 0, 0);
1da177e4
LT
621
622 if (rt && rt->rt6i_dev)
623 skb2->dev = rt->rt6i_dev;
624
3ffe533c 625 icmpv6_send(skb2, rel_type, rel_code, rel_info);
1da177e4
LT
626
627 if (rt)
628 dst_release(&rt->u.dst);
629
630 kfree_skb(skb2);
631 }
e490d1d8
YK
632
633 return 0;
1da177e4
LT
634}
635
c4d3efaf
YK
636static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
637 struct ipv6hdr *ipv6h,
638 struct sk_buff *skb)
639{
640 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
641
642 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
eddc9ec5 643 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
c4d3efaf
YK
644
645 if (INET_ECN_is_ce(dsfield))
eddc9ec5 646 IP_ECN_set_ce(ip_hdr(skb));
c4d3efaf
YK
647}
648
8359925b
YK
649static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
650 struct ipv6hdr *ipv6h,
651 struct sk_buff *skb)
1da177e4 652{
8359925b 653 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
29bb43b4 654 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
1da177e4 655
8359925b 656 if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
0660e03f 657 IP6_ECN_set_ce(ipv6_hdr(skb));
1da177e4 658}
8359925b 659
f1a28eab 660/* called with rcu_read_lock() */
09c6bbf0
VN
661static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
662{
663 struct ip6_tnl_parm *p = &t->parms;
664 int ret = 0;
2f7f54b7 665 struct net *net = dev_net(t->dev);
09c6bbf0
VN
666
667 if (p->flags & IP6_TNL_F_CAP_RCV) {
1ab1457c 668 struct net_device *ldev = NULL;
09c6bbf0
VN
669
670 if (p->link)
f1a28eab 671 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0
VN
672
673 if ((ipv6_addr_is_multicast(&p->laddr) ||
2f7f54b7
PE
674 likely(ipv6_chk_addr(net, &p->laddr, ldev, 0))) &&
675 likely(!ipv6_chk_addr(net, &p->raddr, NULL, 0)))
09c6bbf0
VN
676 ret = 1;
677
09c6bbf0
VN
678 }
679 return ret;
680}
1da177e4
LT
681
682/**
3144581c 683 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
1da177e4 684 * @skb: received socket buffer
8359925b
YK
685 * @protocol: ethernet protocol ID
686 * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
1da177e4
LT
687 *
688 * Return: 0
689 **/
690
8359925b 691static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
502b0935 692 __u8 ipproto,
8359925b
YK
693 void (*dscp_ecn_decapsulate)(struct ip6_tnl *t,
694 struct ipv6hdr *ipv6h,
695 struct sk_buff *skb))
1da177e4 696{
1da177e4 697 struct ip6_tnl *t;
0660e03f 698 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
1da177e4 699
2922bc8a 700 rcu_read_lock();
1da177e4 701
8704ca7e 702 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
2dd02c89 703 &ipv6h->daddr)) != NULL) {
502b0935 704 if (t->parms.proto != ipproto && t->parms.proto != 0) {
2922bc8a 705 rcu_read_unlock();
502b0935
YK
706 goto discard;
707 }
708
1da177e4 709 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
2922bc8a 710 rcu_read_unlock();
50fba2aa 711 goto discard;
1da177e4
LT
712 }
713
09c6bbf0 714 if (!ip6_tnl_rcv_ctl(t)) {
3dca02af 715 t->dev->stats.rx_dropped++;
2922bc8a 716 rcu_read_unlock();
1da177e4
LT
717 goto discard;
718 }
719 secpath_reset(skb);
b0e380b1 720 skb->mac_header = skb->network_header;
c1d2bbe1 721 skb_reset_network_header(skb);
8359925b 722 skb->protocol = htons(protocol);
1da177e4
LT
723 skb->pkt_type = PACKET_HOST;
724 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
725 skb->dev = t->dev;
adf30907 726 skb_dst_drop(skb);
53ab61c6 727 nf_reset(skb);
8359925b
YK
728
729 dscp_ecn_decapsulate(t, ipv6h, skb);
730
3dca02af
PE
731 t->dev->stats.rx_packets++;
732 t->dev->stats.rx_bytes += skb->len;
1da177e4 733 netif_rx(skb);
2922bc8a 734 rcu_read_unlock();
1da177e4
LT
735 return 0;
736 }
2922bc8a 737 rcu_read_unlock();
1da177e4 738 return 1;
50fba2aa
HX
739
740discard:
741 kfree_skb(skb);
742 return 0;
1da177e4
LT
743}
744
c4d3efaf
YK
745static int ip4ip6_rcv(struct sk_buff *skb)
746{
502b0935
YK
747 return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
748 ip4ip6_dscp_ecn_decapsulate);
c4d3efaf
YK
749}
750
8359925b
YK
751static int ip6ip6_rcv(struct sk_buff *skb)
752{
502b0935
YK
753 return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
754 ip6ip6_dscp_ecn_decapsulate);
8359925b
YK
755}
756
6fb32dde
VN
757struct ipv6_tel_txoption {
758 struct ipv6_txoptions ops;
759 __u8 dst_opt[8];
760};
1da177e4 761
6fb32dde
VN
762static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
763{
764 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 765
6fb32dde
VN
766 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
767 opt->dst_opt[3] = 1;
768 opt->dst_opt[4] = encap_limit;
769 opt->dst_opt[5] = IPV6_TLV_PADN;
770 opt->dst_opt[6] = 1;
1da177e4 771
6fb32dde
VN
772 opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
773 opt->ops.opt_nflen = 8;
1da177e4
LT
774}
775
776/**
3144581c 777 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
1da177e4 778 * @t: the outgoing tunnel device
1ab1457c 779 * @hdr: IPv6 header from the incoming packet
1da177e4
LT
780 *
781 * Description:
1ab1457c 782 * Avoid trivial tunneling loop by checking that tunnel exit-point
1da177e4
LT
783 * doesn't match source of incoming packet.
784 *
1ab1457c 785 * Return:
1da177e4
LT
786 * 1 if conflict,
787 * 0 else
788 **/
789
790static inline int
3144581c 791ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
1da177e4
LT
792{
793 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
794}
795
09c6bbf0
VN
796static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
797{
798 struct ip6_tnl_parm *p = &t->parms;
799 int ret = 0;
2f7f54b7 800 struct net *net = dev_net(t->dev);
09c6bbf0 801
1ab1457c 802 if (p->flags & IP6_TNL_F_CAP_XMIT) {
09c6bbf0
VN
803 struct net_device *ldev = NULL;
804
f1a28eab 805 rcu_read_lock();
09c6bbf0 806 if (p->link)
f1a28eab 807 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0 808
2f7f54b7 809 if (unlikely(!ipv6_chk_addr(net, &p->laddr, ldev, 0)))
09c6bbf0
VN
810 printk(KERN_WARNING
811 "%s xmit: Local address not yet configured!\n",
812 p->name);
813 else if (!ipv6_addr_is_multicast(&p->raddr) &&
2f7f54b7 814 unlikely(ipv6_chk_addr(net, &p->raddr, NULL, 0)))
09c6bbf0
VN
815 printk(KERN_WARNING
816 "%s xmit: Routing loop! "
817 "Remote address found on this node!\n",
818 p->name);
819 else
820 ret = 1;
f1a28eab 821 rcu_read_unlock();
09c6bbf0
VN
822 }
823 return ret;
824}
1da177e4 825/**
61ec2aec 826 * ip6_tnl_xmit2 - encapsulate packet and send
1da177e4 827 * @skb: the outgoing socket buffer
1ab1457c 828 * @dev: the outgoing tunnel device
61ec2aec
YK
829 * @dsfield: dscp code for outer header
830 * @fl: flow of tunneled packet
831 * @encap_limit: encapsulation limit
832 * @pmtu: Path MTU is stored if packet is too big
1da177e4
LT
833 *
834 * Description:
835 * Build new header and do some sanity checks on the packet before sending
836 * it.
837 *
1ab1457c 838 * Return:
c4d3efaf 839 * 0 on success
61ec2aec
YK
840 * -1 fail
841 * %-EMSGSIZE message too big. return mtu in this case.
1da177e4
LT
842 **/
843
61ec2aec
YK
844static int ip6_tnl_xmit2(struct sk_buff *skb,
845 struct net_device *dev,
846 __u8 dsfield,
847 struct flowi *fl,
848 int encap_limit,
849 __u32 *pmtu)
1da177e4 850{
52479b62 851 struct net *net = dev_net(dev);
2941a486 852 struct ip6_tnl *t = netdev_priv(dev);
3dca02af 853 struct net_device_stats *stats = &t->dev->stats;
0660e03f 854 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
6fb32dde 855 struct ipv6_tel_txoption opt;
1da177e4
LT
856 struct dst_entry *dst;
857 struct net_device *tdev;
858 int mtu;
c2636b4d 859 unsigned int max_headroom = sizeof(struct ipv6hdr);
1da177e4 860 u8 proto;
61ec2aec 861 int err = -1;
1da177e4 862 int pkt_len;
1da177e4 863
1da177e4
LT
864 if ((dst = ip6_tnl_dst_check(t)) != NULL)
865 dst_hold(dst);
a57ebc90 866 else {
52479b62 867 dst = ip6_route_output(net, NULL, fl);
1da177e4 868
52479b62 869 if (dst->error || xfrm_lookup(net, &dst, fl, NULL, 0) < 0)
a57ebc90
PM
870 goto tx_err_link_failure;
871 }
1da177e4
LT
872
873 tdev = dst->dev;
874
875 if (tdev == dev) {
876 stats->collisions++;
877 if (net_ratelimit())
1ab1457c 878 printk(KERN_WARNING
1da177e4
LT
879 "%s: Local routing loop detected!\n",
880 t->parms.name);
881 goto tx_err_dst_release;
882 }
883 mtu = dst_mtu(dst) - sizeof (*ipv6h);
6fb32dde 884 if (encap_limit >= 0) {
1da177e4
LT
885 max_headroom += 8;
886 mtu -= 8;
887 }
888 if (mtu < IPV6_MIN_MTU)
889 mtu = IPV6_MIN_MTU;
adf30907
ED
890 if (skb_dst(skb))
891 skb_dst(skb)->ops->update_pmtu(skb_dst(skb), mtu);
1da177e4 892 if (skb->len > mtu) {
61ec2aec
YK
893 *pmtu = mtu;
894 err = -EMSGSIZE;
1da177e4
LT
895 goto tx_err_dst_release;
896 }
897
898 /*
899 * Okay, now see if we can stuff it in the buffer as-is.
900 */
901 max_headroom += LL_RESERVED_SPACE(tdev);
1ab1457c 902
cfbba49d
PM
903 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
904 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1da177e4 905 struct sk_buff *new_skb;
1ab1457c 906
1da177e4
LT
907 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
908 goto tx_err_dst_release;
909
910 if (skb->sk)
911 skb_set_owner_w(new_skb, skb->sk);
912 kfree_skb(skb);
913 skb = new_skb;
914 }
adf30907
ED
915 skb_dst_drop(skb);
916 skb_dst_set(skb, dst_clone(dst));
1da177e4 917
b0e380b1 918 skb->transport_header = skb->network_header;
1da177e4 919
61ec2aec 920 proto = fl->proto;
6fb32dde
VN
921 if (encap_limit >= 0) {
922 init_tel_txopt(&opt, encap_limit);
923 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
924 }
e2d1bca7
ACM
925 skb_push(skb, sizeof(struct ipv6hdr));
926 skb_reset_network_header(skb);
0660e03f 927 ipv6h = ipv6_hdr(skb);
61ec2aec 928 *(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000);
1da177e4
LT
929 dsfield = INET_ECN_encapsulate(0, dsfield);
930 ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
1da177e4
LT
931 ipv6h->hop_limit = t->parms.hop_limit;
932 ipv6h->nexthdr = proto;
61ec2aec
YK
933 ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src);
934 ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst);
1da177e4
LT
935 nf_reset(skb);
936 pkt_len = skb->len;
ef76bc23 937 err = ip6_local_out(skb);
1da177e4 938
b9df3cb8 939 if (net_xmit_eval(err) == 0) {
1da177e4
LT
940 stats->tx_bytes += pkt_len;
941 stats->tx_packets++;
942 } else {
943 stats->tx_errors++;
944 stats->tx_aborted_errors++;
945 }
946 ip6_tnl_dst_store(t, dst);
1da177e4
LT
947 return 0;
948tx_err_link_failure:
949 stats->tx_carrier_errors++;
950 dst_link_failure(skb);
951tx_err_dst_release:
952 dst_release(dst);
61ec2aec
YK
953 return err;
954}
955
c4d3efaf
YK
956static inline int
957ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
958{
959 struct ip6_tnl *t = netdev_priv(dev);
eddc9ec5 960 struct iphdr *iph = ip_hdr(skb);
c4d3efaf
YK
961 int encap_limit = -1;
962 struct flowi fl;
963 __u8 dsfield;
964 __u32 mtu;
965 int err;
966
502b0935
YK
967 if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
968 !ip6_tnl_xmit_ctl(t))
c4d3efaf
YK
969 return -1;
970
971 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
972 encap_limit = t->parms.encap_limit;
973
974 memcpy(&fl, &t->fl, sizeof (fl));
975 fl.proto = IPPROTO_IPIP;
976
977 dsfield = ipv4_get_dsfield(iph);
978
979 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
b77f2fa6
AV
980 fl.fl6_flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
981 & IPV6_TCLASS_MASK;
c4d3efaf
YK
982
983 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
984 if (err != 0) {
985 /* XXX: send ICMP error even if DF is not set. */
986 if (err == -EMSGSIZE)
987 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
988 htonl(mtu));
989 return -1;
990 }
991
992 return 0;
993}
994
61ec2aec
YK
995static inline int
996ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
997{
998 struct ip6_tnl *t = netdev_priv(dev);
0660e03f 999 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
61ec2aec
YK
1000 int encap_limit = -1;
1001 __u16 offset;
1002 struct flowi fl;
1003 __u8 dsfield;
1004 __u32 mtu;
1005 int err;
1006
502b0935
YK
1007 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
1008 !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
61ec2aec
YK
1009 return -1;
1010
d56f90a7
ACM
1011 offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb));
1012 if (offset > 0) {
61ec2aec 1013 struct ipv6_tlv_tnl_enc_lim *tel;
d56f90a7 1014 tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
61ec2aec
YK
1015 if (tel->encap_limit == 0) {
1016 icmpv6_send(skb, ICMPV6_PARAMPROB,
3ffe533c 1017 ICMPV6_HDR_FIELD, offset + 2);
61ec2aec
YK
1018 return -1;
1019 }
1020 encap_limit = tel->encap_limit - 1;
1021 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1022 encap_limit = t->parms.encap_limit;
1023
1024 memcpy(&fl, &t->fl, sizeof (fl));
1025 fl.proto = IPPROTO_IPV6;
1026
1027 dsfield = ipv6_get_dsfield(ipv6h);
1028 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
1029 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1030 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
1031 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1032
1033 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
1034 if (err != 0) {
1035 if (err == -EMSGSIZE)
3ffe533c 1036 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
61ec2aec
YK
1037 return -1;
1038 }
1039
1040 return 0;
1041}
1042
6fef4c0c 1043static netdev_tx_t
61ec2aec
YK
1044ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1045{
1046 struct ip6_tnl *t = netdev_priv(dev);
3dca02af 1047 struct net_device_stats *stats = &t->dev->stats;
61ec2aec
YK
1048 int ret;
1049
61ec2aec 1050 switch (skb->protocol) {
60678040 1051 case htons(ETH_P_IP):
c4d3efaf
YK
1052 ret = ip4ip6_tnl_xmit(skb, dev);
1053 break;
60678040 1054 case htons(ETH_P_IPV6):
61ec2aec
YK
1055 ret = ip6ip6_tnl_xmit(skb, dev);
1056 break;
1057 default:
1058 goto tx_err;
1059 }
1060
1061 if (ret < 0)
1062 goto tx_err;
1063
6ed10654 1064 return NETDEV_TX_OK;
61ec2aec 1065
1da177e4
LT
1066tx_err:
1067 stats->tx_errors++;
1068 stats->tx_dropped++;
1069 kfree_skb(skb);
6ed10654 1070 return NETDEV_TX_OK;
1da177e4
LT
1071}
1072
1073static void ip6_tnl_set_cap(struct ip6_tnl *t)
1074{
1075 struct ip6_tnl_parm *p = &t->parms;
09c6bbf0
VN
1076 int ltype = ipv6_addr_type(&p->laddr);
1077 int rtype = ipv6_addr_type(&p->raddr);
1da177e4
LT
1078
1079 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);
1080
09c6bbf0
VN
1081 if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1082 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1083 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
305d4b3c 1084 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
09c6bbf0
VN
1085 if (ltype&IPV6_ADDR_UNICAST)
1086 p->flags |= IP6_TNL_F_CAP_XMIT;
1087 if (rtype&IPV6_ADDR_UNICAST)
1088 p->flags |= IP6_TNL_F_CAP_RCV;
1da177e4
LT
1089 }
1090}
1091
3144581c 1092static void ip6_tnl_link_config(struct ip6_tnl *t)
1da177e4
LT
1093{
1094 struct net_device *dev = t->dev;
1095 struct ip6_tnl_parm *p = &t->parms;
1096 struct flowi *fl = &t->fl;
1097
3a6d54c5
JP
1098 memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1099 memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1da177e4
LT
1100
1101 /* Set up flowi template */
1102 ipv6_addr_copy(&fl->fl6_src, &p->laddr);
1103 ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
1104 fl->oif = p->link;
1105 fl->fl6_flowlabel = 0;
1106
1107 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
1108 fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1109 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
1110 fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1111
1112 ip6_tnl_set_cap(t);
1113
1114 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1115 dev->flags |= IFF_POINTOPOINT;
1116 else
1117 dev->flags &= ~IFF_POINTOPOINT;
1118
1119 dev->iflink = p->link;
1120
1121 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN
1122 int strict = (ipv6_addr_type(&p->raddr) &
1123 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1124
2f7f54b7
PE
1125 struct rt6_info *rt = rt6_lookup(dev_net(dev),
1126 &p->raddr, &p->laddr,
305d4b3c 1127 p->link, strict);
1da177e4
LT
1128
1129 if (rt == NULL)
1130 return;
1131
1132 if (rt->rt6i_dev) {
1133 dev->hard_header_len = rt->rt6i_dev->hard_header_len +
1134 sizeof (struct ipv6hdr);
1135
1136 dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr);
1137
1138 if (dev->mtu < IPV6_MIN_MTU)
1139 dev->mtu = IPV6_MIN_MTU;
1140 }
1141 dst_release(&rt->u.dst);
1142 }
1143}
1144
1145/**
3144581c 1146 * ip6_tnl_change - update the tunnel parameters
1da177e4
LT
1147 * @t: tunnel to be changed
1148 * @p: tunnel configuration parameters
1da177e4
LT
1149 *
1150 * Description:
3144581c 1151 * ip6_tnl_change() updates the tunnel parameters
1da177e4
LT
1152 **/
1153
1154static int
3144581c 1155ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
1da177e4
LT
1156{
1157 ipv6_addr_copy(&t->parms.laddr, &p->laddr);
1158 ipv6_addr_copy(&t->parms.raddr, &p->raddr);
1159 t->parms.flags = p->flags;
1160 t->parms.hop_limit = p->hop_limit;
1161 t->parms.encap_limit = p->encap_limit;
1162 t->parms.flowinfo = p->flowinfo;
8181b8c1 1163 t->parms.link = p->link;
502b0935 1164 t->parms.proto = p->proto;
0c088890 1165 ip6_tnl_dst_reset(t);
3144581c 1166 ip6_tnl_link_config(t);
1da177e4
LT
1167 return 0;
1168}
1169
1170/**
3144581c 1171 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1da177e4
LT
1172 * @dev: virtual device associated with tunnel
1173 * @ifr: parameters passed from userspace
1174 * @cmd: command to be performed
1175 *
1176 * Description:
3144581c 1177 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1ab1457c 1178 * from userspace.
1da177e4
LT
1179 *
1180 * The possible commands are the following:
1181 * %SIOCGETTUNNEL: get tunnel parameters for device
1182 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1183 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1184 * %SIOCDELTUNNEL: delete tunnel
1185 *
1ab1457c 1186 * The fallback device "ip6tnl0", created during module
1da177e4
LT
1187 * initialization, can be used for creating other tunnel devices.
1188 *
1189 * Return:
1190 * 0 on success,
1191 * %-EFAULT if unable to copy data to or from userspace,
1192 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1193 * %-EINVAL if passed tunnel parameters are invalid,
1194 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1195 * %-ENODEV if attempting to change or delete a nonexisting device
1196 **/
1197
1198static int
3144581c 1199ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1da177e4
LT
1200{
1201 int err = 0;
1da177e4
LT
1202 struct ip6_tnl_parm p;
1203 struct ip6_tnl *t = NULL;
2dd02c89
PE
1204 struct net *net = dev_net(dev);
1205 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4
LT
1206
1207 switch (cmd) {
1208 case SIOCGETTUNNEL:
15820e12 1209 if (dev == ip6n->fb_tnl_dev) {
567131a7 1210 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1da177e4
LT
1211 err = -EFAULT;
1212 break;
1213 }
2dd02c89 1214 t = ip6_tnl_locate(net, &p, 0);
567131a7
VN
1215 }
1216 if (t == NULL)
2941a486 1217 t = netdev_priv(dev);
1da177e4
LT
1218 memcpy(&p, &t->parms, sizeof (p));
1219 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
1220 err = -EFAULT;
1221 }
1222 break;
1223 case SIOCADDTUNNEL:
1224 case SIOCCHGTUNNEL:
1225 err = -EPERM;
1da177e4
LT
1226 if (!capable(CAP_NET_ADMIN))
1227 break;
567131a7
VN
1228 err = -EFAULT;
1229 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1230 break;
567131a7 1231 err = -EINVAL;
502b0935
YK
1232 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1233 p.proto != 0)
1da177e4 1234 break;
2dd02c89 1235 t = ip6_tnl_locate(net, &p, cmd == SIOCADDTUNNEL);
15820e12 1236 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
567131a7
VN
1237 if (t != NULL) {
1238 if (t->dev != dev) {
1239 err = -EEXIST;
1240 break;
1241 }
1242 } else
1243 t = netdev_priv(dev);
1244
2dd02c89 1245 ip6_tnl_unlink(ip6n, t);
3144581c 1246 err = ip6_tnl_change(t, &p);
2dd02c89 1247 ip6_tnl_link(ip6n, t);
1da177e4
LT
1248 netdev_state_change(dev);
1249 }
567131a7 1250 if (t) {
1da177e4 1251 err = 0;
567131a7
VN
1252 if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
1253 err = -EFAULT;
1254
1255 } else
1256 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1da177e4
LT
1257 break;
1258 case SIOCDELTUNNEL:
1259 err = -EPERM;
1260 if (!capable(CAP_NET_ADMIN))
1261 break;
1262
15820e12 1263 if (dev == ip6n->fb_tnl_dev) {
567131a7
VN
1264 err = -EFAULT;
1265 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1266 break;
567131a7 1267 err = -ENOENT;
2dd02c89 1268 if ((t = ip6_tnl_locate(net, &p, 0)) == NULL)
1da177e4 1269 break;
567131a7 1270 err = -EPERM;
15820e12 1271 if (t->dev == ip6n->fb_tnl_dev)
1da177e4 1272 break;
567131a7 1273 dev = t->dev;
1da177e4 1274 }
22f8cde5
SH
1275 err = 0;
1276 unregister_netdevice(dev);
1da177e4
LT
1277 break;
1278 default:
1279 err = -EINVAL;
1280 }
1281 return err;
1282}
1283
1da177e4 1284/**
3144581c 1285 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1da177e4
LT
1286 * @dev: virtual device associated with tunnel
1287 * @new_mtu: the new mtu
1288 *
1289 * Return:
1290 * 0 on success,
1291 * %-EINVAL if mtu too small
1292 **/
1293
1294static int
3144581c 1295ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1da177e4
LT
1296{
1297 if (new_mtu < IPV6_MIN_MTU) {
1298 return -EINVAL;
1299 }
1300 dev->mtu = new_mtu;
1301 return 0;
1302}
1303
1326c3d5
SH
1304
1305static const struct net_device_ops ip6_tnl_netdev_ops = {
1306 .ndo_uninit = ip6_tnl_dev_uninit,
1307 .ndo_start_xmit = ip6_tnl_xmit,
1308 .ndo_do_ioctl = ip6_tnl_ioctl,
1309 .ndo_change_mtu = ip6_tnl_change_mtu,
1310};
1311
1da177e4 1312/**
3144581c 1313 * ip6_tnl_dev_setup - setup virtual tunnel device
1da177e4
LT
1314 * @dev: virtual device associated with tunnel
1315 *
1316 * Description:
1317 * Initialize function pointers and device parameters
1318 **/
1319
3144581c 1320static void ip6_tnl_dev_setup(struct net_device *dev)
1da177e4 1321{
1326c3d5 1322 dev->netdev_ops = &ip6_tnl_netdev_ops;
1da177e4 1323 dev->destructor = free_netdev;
1da177e4
LT
1324
1325 dev->type = ARPHRD_TUNNEL6;
1326 dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1327 dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1328 dev->flags |= IFF_NOARP;
1329 dev->addr_len = sizeof(struct in6_addr);
554eb277 1330 dev->features |= NETIF_F_NETNS_LOCAL;
1da177e4
LT
1331}
1332
1333
1334/**
3144581c 1335 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1da177e4
LT
1336 * @dev: virtual device associated with tunnel
1337 **/
1338
1339static inline void
3144581c 1340ip6_tnl_dev_init_gen(struct net_device *dev)
1da177e4 1341{
2941a486 1342 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT
1343 t->dev = dev;
1344 strcpy(t->parms.name, dev->name);
1345}
1346
1347/**
3144581c 1348 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1da177e4
LT
1349 * @dev: virtual device associated with tunnel
1350 **/
1351
1326c3d5 1352static void ip6_tnl_dev_init(struct net_device *dev)
1da177e4 1353{
2941a486 1354 struct ip6_tnl *t = netdev_priv(dev);
3144581c
YK
1355 ip6_tnl_dev_init_gen(dev);
1356 ip6_tnl_link_config(t);
1da177e4
LT
1357}
1358
1359/**
3144581c 1360 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1da177e4
LT
1361 * @dev: fallback device
1362 *
1363 * Return: 0
1364 **/
1365
2c8c1e72 1366static void __net_init ip6_fb_tnl_dev_init(struct net_device *dev)
1da177e4 1367{
2941a486 1368 struct ip6_tnl *t = netdev_priv(dev);
3e6c9fb5
PE
1369 struct net *net = dev_net(dev);
1370 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1371
3144581c 1372 ip6_tnl_dev_init_gen(dev);
502b0935 1373 t->parms.proto = IPPROTO_IPV6;
1da177e4 1374 dev_hold(dev);
3e6c9fb5 1375 ip6n->tnls_wc[0] = t;
1da177e4
LT
1376}
1377
c4d3efaf
YK
1378static struct xfrm6_tunnel ip4ip6_handler = {
1379 .handler = ip4ip6_rcv,
1380 .err_handler = ip4ip6_err,
1381 .priority = 1,
1382};
1383
1da177e4 1384static struct xfrm6_tunnel ip6ip6_handler = {
0303770d
PM
1385 .handler = ip6ip6_rcv,
1386 .err_handler = ip6ip6_err,
d2acc347 1387 .priority = 1,
1da177e4
LT
1388};
1389
2c8c1e72 1390static void __net_exit ip6_tnl_destroy_tunnels(struct ip6_tnl_net *ip6n)
3e6c9fb5
PE
1391{
1392 int h;
1393 struct ip6_tnl *t;
cf4432f5 1394 LIST_HEAD(list);
3e6c9fb5
PE
1395
1396 for (h = 0; h < HASH_SIZE; h++) {
cf4432f5
ED
1397 t = ip6n->tnls_r_l[h];
1398 while (t != NULL) {
1399 unregister_netdevice_queue(t->dev, &list);
1400 t = t->next;
1401 }
3e6c9fb5
PE
1402 }
1403
1404 t = ip6n->tnls_wc[0];
cf4432f5
ED
1405 unregister_netdevice_queue(t->dev, &list);
1406 unregister_netdevice_many(&list);
3e6c9fb5
PE
1407}
1408
2c8c1e72 1409static int __net_init ip6_tnl_init_net(struct net *net)
13eeb8e9 1410{
ac31cd3c 1411 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
13eeb8e9 1412 int err;
13eeb8e9 1413
3e6c9fb5
PE
1414 ip6n->tnls[0] = ip6n->tnls_wc;
1415 ip6n->tnls[1] = ip6n->tnls_r_l;
1416
15820e12
PE
1417 err = -ENOMEM;
1418 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1419 ip6_tnl_dev_setup);
1420
1421 if (!ip6n->fb_tnl_dev)
1422 goto err_alloc_dev;
be77e593 1423 dev_net_set(ip6n->fb_tnl_dev, net);
15820e12 1424
1326c3d5 1425 ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
15820e12
PE
1426
1427 err = register_netdev(ip6n->fb_tnl_dev);
1428 if (err < 0)
1429 goto err_register;
13eeb8e9
PE
1430 return 0;
1431
15820e12
PE
1432err_register:
1433 free_netdev(ip6n->fb_tnl_dev);
1434err_alloc_dev:
13eeb8e9
PE
1435 return err;
1436}
1437
2c8c1e72 1438static void __net_exit ip6_tnl_exit_net(struct net *net)
13eeb8e9 1439{
ac31cd3c 1440 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
13eeb8e9 1441
3e6c9fb5
PE
1442 rtnl_lock();
1443 ip6_tnl_destroy_tunnels(ip6n);
1444 rtnl_unlock();
13eeb8e9
PE
1445}
1446
1447static struct pernet_operations ip6_tnl_net_ops = {
1448 .init = ip6_tnl_init_net,
1449 .exit = ip6_tnl_exit_net,
ac31cd3c
EB
1450 .id = &ip6_tnl_net_id,
1451 .size = sizeof(struct ip6_tnl_net),
13eeb8e9
PE
1452};
1453
1da177e4
LT
1454/**
1455 * ip6_tunnel_init - register protocol and reserve needed resources
1456 *
1457 * Return: 0 on success
1458 **/
1459
1460static int __init ip6_tunnel_init(void)
1461{
1462 int err;
1463
d5aa407f
AD
1464 err = register_pernet_device(&ip6_tnl_net_ops);
1465 if (err < 0)
1466 goto out_pernet;
1467
1468 err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
1469 if (err < 0) {
3144581c 1470 printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n");
d5aa407f 1471 goto out_ip4ip6;
c4d3efaf
YK
1472 }
1473
d5aa407f
AD
1474 err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
1475 if (err < 0) {
3144581c 1476 printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n");
d5aa407f 1477 goto out_ip6ip6;
1da177e4 1478 }
13eeb8e9 1479
1da177e4 1480 return 0;
d5aa407f
AD
1481
1482out_ip6ip6:
c4d3efaf 1483 xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
d5aa407f
AD
1484out_ip4ip6:
1485 unregister_pernet_device(&ip6_tnl_net_ops);
1486out_pernet:
1da177e4
LT
1487 return err;
1488}
1489
1490/**
1491 * ip6_tunnel_cleanup - free resources and unregister protocol
1492 **/
1493
1494static void __exit ip6_tunnel_cleanup(void)
1495{
c4d3efaf 1496 if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
3144581c 1497 printk(KERN_INFO "ip6_tunnel close: can't deregister ip4ip6\n");
c4d3efaf 1498
73d605d1 1499 if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
3144581c 1500 printk(KERN_INFO "ip6_tunnel close: can't deregister ip6ip6\n");
1da177e4 1501
ac31cd3c 1502 unregister_pernet_device(&ip6_tnl_net_ops);
1da177e4
LT
1503}
1504
1505module_init(ip6_tunnel_init);
1506module_exit(ip6_tunnel_cleanup);