]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - net/ipv6/ip6_tunnel.c
[IP6TUNNEL]: Use proper net in hash-lookup functions.
[mirror_ubuntu-artful-kernel.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
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>
13eeb8e9
PE
55#include <net/net_namespace.h>
56#include <net/netns/generic.h>
1da177e4
LT
57
58MODULE_AUTHOR("Ville Nuorvala");
c4d3efaf 59MODULE_DESCRIPTION("IPv6 tunneling device");
1da177e4
LT
60MODULE_LICENSE("GPL");
61
62#define IPV6_TLV_TEL_DST_SIZE 8
63
64#ifdef IP6_TNL_DEBUG
0dc47877 65#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__)
1da177e4
LT
66#else
67#define IP6_TNL_TRACE(x...) do {;} while(0)
68#endif
69
70#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
c4d3efaf 71#define IPV6_TCLASS_SHIFT 20
1da177e4
LT
72
73#define HASH_SIZE 32
74
e69a4adc 75#define HASH(addr) ((__force u32)((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
1ab1457c
YH
76 (addr)->s6_addr32[2] ^ (addr)->s6_addr32[3]) & \
77 (HASH_SIZE - 1))
1da177e4 78
3144581c
YK
79static int ip6_fb_tnl_dev_init(struct net_device *dev);
80static int ip6_tnl_dev_init(struct net_device *dev);
81static void ip6_tnl_dev_setup(struct net_device *dev);
1da177e4 82
13eeb8e9
PE
83static int ip6_tnl_net_id;
84struct ip6_tnl_net {
85};
86
1da177e4 87/* the IPv6 tunnel fallback device */
3144581c 88static struct net_device *ip6_fb_tnl_dev;
1da177e4
LT
89
90
91/* lists for storing tunnels in use */
92static struct ip6_tnl *tnls_r_l[HASH_SIZE];
93static struct ip6_tnl *tnls_wc[1];
94static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l };
95
96/* lock for the tunnel lists */
3144581c 97static DEFINE_RWLOCK(ip6_tnl_lock);
1da177e4
LT
98
99static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
100{
101 struct dst_entry *dst = t->dst_cache;
102
1ab1457c 103 if (dst && dst->obsolete &&
1da177e4
LT
104 dst->ops->check(dst, t->dst_cookie) == NULL) {
105 t->dst_cache = NULL;
106 dst_release(dst);
107 return NULL;
108 }
109
110 return dst;
111}
112
113static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
114{
115 dst_release(t->dst_cache);
116 t->dst_cache = NULL;
117}
118
119static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
120{
121 struct rt6_info *rt = (struct rt6_info *) dst;
122 t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
123 dst_release(t->dst_cache);
124 t->dst_cache = dst;
125}
126
127/**
3144581c 128 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
1ab1457c
YH
129 * @remote: the address of the tunnel exit-point
130 * @local: the address of the tunnel entry-point
1da177e4 131 *
1ab1457c 132 * Return:
1da177e4 133 * tunnel matching given end-points if found,
1ab1457c 134 * else fallback tunnel if its device is up,
1da177e4
LT
135 * else %NULL
136 **/
137
138static struct ip6_tnl *
2dd02c89 139ip6_tnl_lookup(struct net *net, struct in6_addr *remote, struct in6_addr *local)
1da177e4
LT
140{
141 unsigned h0 = HASH(remote);
142 unsigned h1 = HASH(local);
143 struct ip6_tnl *t;
144
145 for (t = tnls_r_l[h0 ^ h1]; t; t = t->next) {
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 }
151 if ((t = tnls_wc[0]) != NULL && (t->dev->flags & IFF_UP))
152 return t;
153
154 return NULL;
155}
156
157/**
3144581c 158 * ip6_tnl_bucket - get head of list matching given tunnel parameters
1ab1457c 159 * @p: parameters containing tunnel end-points
1da177e4
LT
160 *
161 * Description:
3144581c 162 * ip6_tnl_bucket() returns the head of the list matching the
1da177e4
LT
163 * &struct in6_addr entries laddr and raddr in @p.
164 *
1ab1457c 165 * Return: head of IPv6 tunnel list
1da177e4
LT
166 **/
167
168static struct ip6_tnl **
2dd02c89 169ip6_tnl_bucket(struct ip6_tnl_net *ip6n, struct ip6_tnl_parm *p)
1da177e4
LT
170{
171 struct in6_addr *remote = &p->raddr;
172 struct in6_addr *local = &p->laddr;
173 unsigned h = 0;
174 int prio = 0;
175
176 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
177 prio = 1;
178 h = HASH(remote) ^ HASH(local);
179 }
180 return &tnls[prio][h];
181}
182
183/**
3144581c 184 * ip6_tnl_link - add tunnel to hash table
1da177e4
LT
185 * @t: tunnel to be added
186 **/
187
188static void
2dd02c89 189ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 190{
2dd02c89 191 struct ip6_tnl **tp = ip6_tnl_bucket(ip6n, &t->parms);
1da177e4
LT
192
193 t->next = *tp;
3144581c 194 write_lock_bh(&ip6_tnl_lock);
1da177e4 195 *tp = t;
3144581c 196 write_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
197}
198
199/**
3144581c 200 * ip6_tnl_unlink - remove tunnel from hash table
1da177e4
LT
201 * @t: tunnel to be removed
202 **/
203
204static void
2dd02c89 205ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4
LT
206{
207 struct ip6_tnl **tp;
208
2dd02c89 209 for (tp = ip6_tnl_bucket(ip6n, &t->parms); *tp; tp = &(*tp)->next) {
1da177e4 210 if (t == *tp) {
3144581c 211 write_lock_bh(&ip6_tnl_lock);
1da177e4 212 *tp = t->next;
3144581c 213 write_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
214 break;
215 }
216 }
217}
218
219/**
220 * ip6_tnl_create() - create a new tunnel
221 * @p: tunnel parameters
222 * @pt: pointer to new tunnel
223 *
224 * Description:
225 * Create tunnel matching given parameters.
1ab1457c
YH
226 *
227 * Return:
567131a7 228 * created tunnel or NULL
1da177e4
LT
229 **/
230
2dd02c89 231static struct ip6_tnl *ip6_tnl_create(struct net *net, struct ip6_tnl_parm *p)
1da177e4
LT
232{
233 struct net_device *dev;
234 struct ip6_tnl *t;
235 char name[IFNAMSIZ];
236 int err;
2dd02c89 237 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 238
34cc7ba6 239 if (p->name[0])
1da177e4 240 strlcpy(name, p->name, IFNAMSIZ);
34cc7ba6
PE
241 else
242 sprintf(name, "ip6tnl%%d");
243
3144581c 244 dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
1da177e4 245 if (dev == NULL)
567131a7 246 goto failed;
1da177e4 247
b37d428b
PE
248 if (strchr(name, '%')) {
249 if (dev_alloc_name(dev, name) < 0)
250 goto failed_free;
251 }
252
2941a486 253 t = netdev_priv(dev);
3144581c 254 dev->init = ip6_tnl_dev_init;
1da177e4
LT
255 t->parms = *p;
256
b37d428b
PE
257 if ((err = register_netdevice(dev)) < 0)
258 goto failed_free;
259
1da177e4 260 dev_hold(dev);
2dd02c89 261 ip6_tnl_link(ip6n, t);
567131a7 262 return t;
b37d428b
PE
263
264failed_free:
265 free_netdev(dev);
567131a7
VN
266failed:
267 return NULL;
1da177e4
LT
268}
269
270/**
3144581c 271 * ip6_tnl_locate - find or create tunnel matching given parameters
1ab1457c 272 * @p: tunnel parameters
1da177e4
LT
273 * @create: != 0 if allowed to create new tunnel if no match found
274 *
275 * Description:
3144581c 276 * ip6_tnl_locate() first tries to locate an existing tunnel
1da177e4
LT
277 * based on @parms. If this is unsuccessful, but @create is set a new
278 * tunnel device is created and registered for use.
279 *
280 * Return:
567131a7 281 * matching tunnel or NULL
1da177e4
LT
282 **/
283
2dd02c89
PE
284static struct ip6_tnl *ip6_tnl_locate(struct net *net,
285 struct ip6_tnl_parm *p, int create)
1da177e4
LT
286{
287 struct in6_addr *remote = &p->raddr;
288 struct in6_addr *local = &p->laddr;
289 struct ip6_tnl *t;
2dd02c89 290 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 291
2dd02c89 292 for (t = *ip6_tnl_bucket(ip6n, p); t; t = t->next) {
1da177e4 293 if (ipv6_addr_equal(local, &t->parms.laddr) &&
567131a7
VN
294 ipv6_addr_equal(remote, &t->parms.raddr))
295 return t;
1da177e4
LT
296 }
297 if (!create)
567131a7 298 return NULL;
2dd02c89 299 return ip6_tnl_create(net, p);
1da177e4
LT
300}
301
302/**
3144581c 303 * ip6_tnl_dev_uninit - tunnel device uninitializer
1da177e4 304 * @dev: the device to be destroyed
1ab1457c 305 *
1da177e4 306 * Description:
3144581c 307 * ip6_tnl_dev_uninit() removes tunnel from its list
1da177e4
LT
308 **/
309
310static void
3144581c 311ip6_tnl_dev_uninit(struct net_device *dev)
1da177e4 312{
2941a486 313 struct ip6_tnl *t = netdev_priv(dev);
2dd02c89
PE
314 struct net *net = dev_net(dev);
315 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 316
3144581c
YK
317 if (dev == ip6_fb_tnl_dev) {
318 write_lock_bh(&ip6_tnl_lock);
1da177e4 319 tnls_wc[0] = NULL;
3144581c 320 write_unlock_bh(&ip6_tnl_lock);
1da177e4 321 } else {
2dd02c89 322 ip6_tnl_unlink(ip6n, t);
1da177e4
LT
323 }
324 ip6_tnl_dst_reset(t);
325 dev_put(dev);
326}
327
328/**
329 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
330 * @skb: received socket buffer
331 *
1ab1457c
YH
332 * Return:
333 * 0 if none was found,
1da177e4
LT
334 * else index to encapsulation limit
335 **/
336
337static __u16
338parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
339{
340 struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
341 __u8 nexthdr = ipv6h->nexthdr;
342 __u16 off = sizeof (*ipv6h);
343
344 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
345 __u16 optlen = 0;
346 struct ipv6_opt_hdr *hdr;
347 if (raw + off + sizeof (*hdr) > skb->data &&
348 !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
349 break;
350
351 hdr = (struct ipv6_opt_hdr *) (raw + off);
352 if (nexthdr == NEXTHDR_FRAGMENT) {
353 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
354 if (frag_hdr->frag_off)
355 break;
356 optlen = 8;
357 } else if (nexthdr == NEXTHDR_AUTH) {
358 optlen = (hdr->hdrlen + 2) << 2;
359 } else {
360 optlen = ipv6_optlen(hdr);
361 }
362 if (nexthdr == NEXTHDR_DEST) {
363 __u16 i = off + 2;
364 while (1) {
365 struct ipv6_tlv_tnl_enc_lim *tel;
366
367 /* No more room for encapsulation limit */
368 if (i + sizeof (*tel) > off + optlen)
369 break;
370
371 tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
372 /* return index of option if found and valid */
373 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
374 tel->length == 1)
375 return i;
376 /* else jump to next option */
377 if (tel->type)
378 i += tel->length + 2;
379 else
380 i++;
381 }
382 }
383 nexthdr = hdr->nexthdr;
384 off += optlen;
385 }
386 return 0;
387}
388
389/**
e490d1d8 390 * ip6_tnl_err - tunnel error handler
1da177e4
LT
391 *
392 * Description:
e490d1d8 393 * ip6_tnl_err() should handle errors in the tunnel according
1da177e4
LT
394 * to the specifications in RFC 2473.
395 **/
396
d2acc347 397static int
502b0935 398ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
704eae1f 399 int *type, int *code, int *msg, __u32 *info, int offset)
1da177e4
LT
400{
401 struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
402 struct ip6_tnl *t;
403 int rel_msg = 0;
404 int rel_type = ICMPV6_DEST_UNREACH;
405 int rel_code = ICMPV6_ADDR_UNREACH;
406 __u32 rel_info = 0;
407 __u16 len;
d2acc347 408 int err = -ENOENT;
1da177e4 409
1ab1457c
YH
410 /* If the packet doesn't contain the original IPv6 header we are
411 in trouble since we might need the source address for further
1da177e4
LT
412 processing of the error. */
413
3144581c 414 read_lock(&ip6_tnl_lock);
8704ca7e 415 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
2dd02c89 416 &ipv6h->saddr)) == NULL)
1da177e4
LT
417 goto out;
418
502b0935
YK
419 if (t->parms.proto != ipproto && t->parms.proto != 0)
420 goto out;
421
d2acc347
HX
422 err = 0;
423
e490d1d8 424 switch (*type) {
1da177e4
LT
425 __u32 teli;
426 struct ipv6_tlv_tnl_enc_lim *tel;
427 __u32 mtu;
428 case ICMPV6_DEST_UNREACH:
429 if (net_ratelimit())
430 printk(KERN_WARNING
431 "%s: Path to destination invalid "
432 "or inactive!\n", t->parms.name);
433 rel_msg = 1;
434 break;
435 case ICMPV6_TIME_EXCEED:
e490d1d8 436 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
1da177e4
LT
437 if (net_ratelimit())
438 printk(KERN_WARNING
439 "%s: Too small hop limit or "
1ab1457c 440 "routing loop in tunnel!\n",
1da177e4
LT
441 t->parms.name);
442 rel_msg = 1;
443 }
444 break;
445 case ICMPV6_PARAMPROB:
107a5fe6 446 teli = 0;
e490d1d8 447 if ((*code) == ICMPV6_HDR_FIELD)
107a5fe6 448 teli = parse_tlv_tnl_enc_lim(skb, skb->data);
1da177e4 449
704eae1f 450 if (teli && teli == *info - 2) {
1da177e4
LT
451 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
452 if (tel->encap_limit == 0) {
453 if (net_ratelimit())
454 printk(KERN_WARNING
455 "%s: Too small encapsulation "
456 "limit or routing loop in "
457 "tunnel!\n", t->parms.name);
458 rel_msg = 1;
459 }
107a5fe6
VN
460 } else if (net_ratelimit()) {
461 printk(KERN_WARNING
462 "%s: Recipient unable to parse tunneled "
463 "packet!\n ", t->parms.name);
1da177e4
LT
464 }
465 break;
466 case ICMPV6_PKT_TOOBIG:
704eae1f 467 mtu = *info - offset;
1da177e4
LT
468 if (mtu < IPV6_MIN_MTU)
469 mtu = IPV6_MIN_MTU;
470 t->dev->mtu = mtu;
471
cc6cdac0 472 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
1da177e4
LT
473 rel_type = ICMPV6_PKT_TOOBIG;
474 rel_code = 0;
475 rel_info = mtu;
476 rel_msg = 1;
477 }
478 break;
479 }
e490d1d8
YK
480
481 *type = rel_type;
482 *code = rel_code;
483 *info = rel_info;
484 *msg = rel_msg;
485
486out:
3144581c 487 read_unlock(&ip6_tnl_lock);
e490d1d8
YK
488 return err;
489}
490
c4d3efaf
YK
491static int
492ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
704eae1f 493 int type, int code, int offset, __be32 info)
c4d3efaf
YK
494{
495 int rel_msg = 0;
496 int rel_type = type;
497 int rel_code = code;
704eae1f 498 __u32 rel_info = ntohl(info);
c4d3efaf
YK
499 int err;
500 struct sk_buff *skb2;
501 struct iphdr *eiph;
502 struct flowi fl;
503 struct rtable *rt;
504
502b0935
YK
505 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
506 &rel_msg, &rel_info, offset);
c4d3efaf
YK
507 if (err < 0)
508 return err;
509
510 if (rel_msg == 0)
511 return 0;
512
513 switch (rel_type) {
514 case ICMPV6_DEST_UNREACH:
515 if (rel_code != ICMPV6_ADDR_UNREACH)
516 return 0;
517 rel_type = ICMP_DEST_UNREACH;
518 rel_code = ICMP_HOST_UNREACH;
519 break;
520 case ICMPV6_PKT_TOOBIG:
521 if (rel_code != 0)
522 return 0;
523 rel_type = ICMP_DEST_UNREACH;
524 rel_code = ICMP_FRAG_NEEDED;
525 break;
526 default:
527 return 0;
528 }
529
530 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
531 return 0;
532
533 skb2 = skb_clone(skb, GFP_ATOMIC);
534 if (!skb2)
535 return 0;
536
537 dst_release(skb2->dst);
538 skb2->dst = NULL;
539 skb_pull(skb2, offset);
c1d2bbe1 540 skb_reset_network_header(skb2);
eddc9ec5 541 eiph = ip_hdr(skb2);
c4d3efaf
YK
542
543 /* Try to guess incoming interface */
544 memset(&fl, 0, sizeof(fl));
545 fl.fl4_dst = eiph->saddr;
546 fl.fl4_tos = RT_TOS(eiph->tos);
547 fl.proto = IPPROTO_IPIP;
f206351a 548 if (ip_route_output_key(&init_net, &rt, &fl))
c4d3efaf
YK
549 goto out;
550
551 skb2->dev = rt->u.dst.dev;
552
553 /* route "incoming" packet */
554 if (rt->rt_flags & RTCF_LOCAL) {
555 ip_rt_put(rt);
556 rt = NULL;
557 fl.fl4_dst = eiph->daddr;
558 fl.fl4_src = eiph->saddr;
559 fl.fl4_tos = eiph->tos;
f206351a 560 if (ip_route_output_key(&init_net, &rt, &fl) ||
c4d3efaf
YK
561 rt->u.dst.dev->type != ARPHRD_TUNNEL) {
562 ip_rt_put(rt);
563 goto out;
564 }
9937ded8 565 skb2->dst = (struct dst_entry *)rt;
c4d3efaf
YK
566 } else {
567 ip_rt_put(rt);
568 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
569 skb2->dev) ||
570 skb2->dst->dev->type != ARPHRD_TUNNEL)
571 goto out;
572 }
573
574 /* change mtu on this route */
575 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
576 if (rel_info > dst_mtu(skb2->dst))
577 goto out;
578
579 skb2->dst->ops->update_pmtu(skb2->dst, rel_info);
c4d3efaf
YK
580 }
581
704eae1f 582 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
c4d3efaf
YK
583
584out:
585 kfree_skb(skb2);
586 return 0;
587}
588
e490d1d8
YK
589static int
590ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
704eae1f 591 int type, int code, int offset, __be32 info)
e490d1d8
YK
592{
593 int rel_msg = 0;
594 int rel_type = type;
595 int rel_code = code;
704eae1f 596 __u32 rel_info = ntohl(info);
e490d1d8
YK
597 int err;
598
502b0935
YK
599 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
600 &rel_msg, &rel_info, offset);
e490d1d8
YK
601 if (err < 0)
602 return err;
603
604 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
1da177e4
LT
605 struct rt6_info *rt;
606 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 607
1da177e4 608 if (!skb2)
e490d1d8 609 return 0;
1da177e4
LT
610
611 dst_release(skb2->dst);
612 skb2->dst = NULL;
613 skb_pull(skb2, offset);
c1d2bbe1 614 skb_reset_network_header(skb2);
1da177e4
LT
615
616 /* Try to guess incoming interface */
606a2b48 617 rt = rt6_lookup(&init_net, &ipv6_hdr(skb2)->saddr, NULL, 0, 0);
1da177e4
LT
618
619 if (rt && rt->rt6i_dev)
620 skb2->dev = rt->rt6i_dev;
621
622 icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev);
623
624 if (rt)
625 dst_release(&rt->u.dst);
626
627 kfree_skb(skb2);
628 }
e490d1d8
YK
629
630 return 0;
1da177e4
LT
631}
632
c4d3efaf
YK
633static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
634 struct ipv6hdr *ipv6h,
635 struct sk_buff *skb)
636{
637 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
638
639 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
eddc9ec5 640 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
c4d3efaf
YK
641
642 if (INET_ECN_is_ce(dsfield))
eddc9ec5 643 IP_ECN_set_ce(ip_hdr(skb));
c4d3efaf
YK
644}
645
8359925b
YK
646static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
647 struct ipv6hdr *ipv6h,
648 struct sk_buff *skb)
1da177e4 649{
8359925b 650 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
29bb43b4 651 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
1da177e4 652
8359925b 653 if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
0660e03f 654 IP6_ECN_set_ce(ipv6_hdr(skb));
1da177e4 655}
8359925b 656
09c6bbf0
VN
657static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
658{
659 struct ip6_tnl_parm *p = &t->parms;
660 int ret = 0;
661
662 if (p->flags & IP6_TNL_F_CAP_RCV) {
1ab1457c 663 struct net_device *ldev = NULL;
09c6bbf0
VN
664
665 if (p->link)
881d966b 666 ldev = dev_get_by_index(&init_net, p->link);
09c6bbf0
VN
667
668 if ((ipv6_addr_is_multicast(&p->laddr) ||
bfeade08
DL
669 likely(ipv6_chk_addr(&init_net, &p->laddr, ldev, 0))) &&
670 likely(!ipv6_chk_addr(&init_net, &p->raddr, NULL, 0)))
09c6bbf0
VN
671 ret = 1;
672
673 if (ldev)
674 dev_put(ldev);
675 }
676 return ret;
677}
1da177e4
LT
678
679/**
3144581c 680 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
1da177e4 681 * @skb: received socket buffer
8359925b
YK
682 * @protocol: ethernet protocol ID
683 * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
1da177e4
LT
684 *
685 * Return: 0
686 **/
687
8359925b 688static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
502b0935 689 __u8 ipproto,
8359925b
YK
690 void (*dscp_ecn_decapsulate)(struct ip6_tnl *t,
691 struct ipv6hdr *ipv6h,
692 struct sk_buff *skb))
1da177e4 693{
1da177e4 694 struct ip6_tnl *t;
0660e03f 695 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
1da177e4 696
3144581c 697 read_lock(&ip6_tnl_lock);
1da177e4 698
8704ca7e 699 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
2dd02c89 700 &ipv6h->daddr)) != NULL) {
502b0935
YK
701 if (t->parms.proto != ipproto && t->parms.proto != 0) {
702 read_unlock(&ip6_tnl_lock);
703 goto discard;
704 }
705
1da177e4 706 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
3144581c 707 read_unlock(&ip6_tnl_lock);
50fba2aa 708 goto discard;
1da177e4
LT
709 }
710
09c6bbf0 711 if (!ip6_tnl_rcv_ctl(t)) {
1da177e4 712 t->stat.rx_dropped++;
3144581c 713 read_unlock(&ip6_tnl_lock);
1da177e4
LT
714 goto discard;
715 }
716 secpath_reset(skb);
b0e380b1 717 skb->mac_header = skb->network_header;
c1d2bbe1 718 skb_reset_network_header(skb);
8359925b 719 skb->protocol = htons(protocol);
1da177e4
LT
720 skb->pkt_type = PACKET_HOST;
721 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
722 skb->dev = t->dev;
723 dst_release(skb->dst);
724 skb->dst = NULL;
53ab61c6 725 nf_reset(skb);
8359925b
YK
726
727 dscp_ecn_decapsulate(t, ipv6h, skb);
728
1da177e4
LT
729 t->stat.rx_packets++;
730 t->stat.rx_bytes += skb->len;
731 netif_rx(skb);
3144581c 732 read_unlock(&ip6_tnl_lock);
1da177e4
LT
733 return 0;
734 }
3144581c 735 read_unlock(&ip6_tnl_lock);
1da177e4 736 return 1;
50fba2aa
HX
737
738discard:
739 kfree_skb(skb);
740 return 0;
1da177e4
LT
741}
742
c4d3efaf
YK
743static int ip4ip6_rcv(struct sk_buff *skb)
744{
502b0935
YK
745 return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
746 ip4ip6_dscp_ecn_decapsulate);
c4d3efaf
YK
747}
748
8359925b
YK
749static int ip6ip6_rcv(struct sk_buff *skb)
750{
502b0935
YK
751 return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
752 ip6ip6_dscp_ecn_decapsulate);
8359925b
YK
753}
754
6fb32dde
VN
755struct ipv6_tel_txoption {
756 struct ipv6_txoptions ops;
757 __u8 dst_opt[8];
758};
1da177e4 759
6fb32dde
VN
760static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
761{
762 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 763
6fb32dde
VN
764 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
765 opt->dst_opt[3] = 1;
766 opt->dst_opt[4] = encap_limit;
767 opt->dst_opt[5] = IPV6_TLV_PADN;
768 opt->dst_opt[6] = 1;
1da177e4 769
6fb32dde
VN
770 opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
771 opt->ops.opt_nflen = 8;
1da177e4
LT
772}
773
774/**
3144581c 775 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
1da177e4 776 * @t: the outgoing tunnel device
1ab1457c 777 * @hdr: IPv6 header from the incoming packet
1da177e4
LT
778 *
779 * Description:
1ab1457c 780 * Avoid trivial tunneling loop by checking that tunnel exit-point
1da177e4
LT
781 * doesn't match source of incoming packet.
782 *
1ab1457c 783 * Return:
1da177e4
LT
784 * 1 if conflict,
785 * 0 else
786 **/
787
788static inline int
3144581c 789ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
1da177e4
LT
790{
791 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
792}
793
09c6bbf0
VN
794static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
795{
796 struct ip6_tnl_parm *p = &t->parms;
797 int ret = 0;
798
1ab1457c 799 if (p->flags & IP6_TNL_F_CAP_XMIT) {
09c6bbf0
VN
800 struct net_device *ldev = NULL;
801
802 if (p->link)
881d966b 803 ldev = dev_get_by_index(&init_net, p->link);
09c6bbf0 804
bfeade08 805 if (unlikely(!ipv6_chk_addr(&init_net, &p->laddr, ldev, 0)))
09c6bbf0
VN
806 printk(KERN_WARNING
807 "%s xmit: Local address not yet configured!\n",
808 p->name);
809 else if (!ipv6_addr_is_multicast(&p->raddr) &&
bfeade08 810 unlikely(ipv6_chk_addr(&init_net, &p->raddr, NULL, 0)))
09c6bbf0
VN
811 printk(KERN_WARNING
812 "%s xmit: Routing loop! "
813 "Remote address found on this node!\n",
814 p->name);
815 else
816 ret = 1;
817 if (ldev)
818 dev_put(ldev);
819 }
820 return ret;
821}
1da177e4 822/**
61ec2aec 823 * ip6_tnl_xmit2 - encapsulate packet and send
1da177e4 824 * @skb: the outgoing socket buffer
1ab1457c 825 * @dev: the outgoing tunnel device
61ec2aec
YK
826 * @dsfield: dscp code for outer header
827 * @fl: flow of tunneled packet
828 * @encap_limit: encapsulation limit
829 * @pmtu: Path MTU is stored if packet is too big
1da177e4
LT
830 *
831 * Description:
832 * Build new header and do some sanity checks on the packet before sending
833 * it.
834 *
1ab1457c 835 * Return:
c4d3efaf 836 * 0 on success
61ec2aec
YK
837 * -1 fail
838 * %-EMSGSIZE message too big. return mtu in this case.
1da177e4
LT
839 **/
840
61ec2aec
YK
841static int ip6_tnl_xmit2(struct sk_buff *skb,
842 struct net_device *dev,
843 __u8 dsfield,
844 struct flowi *fl,
845 int encap_limit,
846 __u32 *pmtu)
1da177e4 847{
2941a486 848 struct ip6_tnl *t = netdev_priv(dev);
1da177e4 849 struct net_device_stats *stats = &t->stat;
0660e03f 850 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
6fb32dde 851 struct ipv6_tel_txoption opt;
1da177e4
LT
852 struct dst_entry *dst;
853 struct net_device *tdev;
854 int mtu;
c2636b4d 855 unsigned int max_headroom = sizeof(struct ipv6hdr);
1da177e4 856 u8 proto;
61ec2aec 857 int err = -1;
1da177e4 858 int pkt_len;
1da177e4 859
1da177e4
LT
860 if ((dst = ip6_tnl_dst_check(t)) != NULL)
861 dst_hold(dst);
a57ebc90 862 else {
4591db4f 863 dst = ip6_route_output(&init_net, NULL, fl);
1da177e4 864
61ec2aec 865 if (dst->error || xfrm_lookup(&dst, fl, NULL, 0) < 0)
a57ebc90
PM
866 goto tx_err_link_failure;
867 }
1da177e4
LT
868
869 tdev = dst->dev;
870
871 if (tdev == dev) {
872 stats->collisions++;
873 if (net_ratelimit())
1ab1457c 874 printk(KERN_WARNING
1da177e4
LT
875 "%s: Local routing loop detected!\n",
876 t->parms.name);
877 goto tx_err_dst_release;
878 }
879 mtu = dst_mtu(dst) - sizeof (*ipv6h);
6fb32dde 880 if (encap_limit >= 0) {
1da177e4
LT
881 max_headroom += 8;
882 mtu -= 8;
883 }
884 if (mtu < IPV6_MIN_MTU)
885 mtu = IPV6_MIN_MTU;
26892058
YK
886 if (skb->dst)
887 skb->dst->ops->update_pmtu(skb->dst, mtu);
1da177e4 888 if (skb->len > mtu) {
61ec2aec
YK
889 *pmtu = mtu;
890 err = -EMSGSIZE;
1da177e4
LT
891 goto tx_err_dst_release;
892 }
893
894 /*
895 * Okay, now see if we can stuff it in the buffer as-is.
896 */
897 max_headroom += LL_RESERVED_SPACE(tdev);
1ab1457c 898
cfbba49d
PM
899 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
900 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1da177e4 901 struct sk_buff *new_skb;
1ab1457c 902
1da177e4
LT
903 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
904 goto tx_err_dst_release;
905
906 if (skb->sk)
907 skb_set_owner_w(new_skb, skb->sk);
908 kfree_skb(skb);
909 skb = new_skb;
910 }
911 dst_release(skb->dst);
912 skb->dst = dst_clone(dst);
913
b0e380b1 914 skb->transport_header = skb->network_header;
1da177e4 915
61ec2aec 916 proto = fl->proto;
6fb32dde
VN
917 if (encap_limit >= 0) {
918 init_tel_txopt(&opt, encap_limit);
919 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
920 }
e2d1bca7
ACM
921 skb_push(skb, sizeof(struct ipv6hdr));
922 skb_reset_network_header(skb);
0660e03f 923 ipv6h = ipv6_hdr(skb);
61ec2aec 924 *(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000);
1da177e4
LT
925 dsfield = INET_ECN_encapsulate(0, dsfield);
926 ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
1da177e4
LT
927 ipv6h->hop_limit = t->parms.hop_limit;
928 ipv6h->nexthdr = proto;
61ec2aec
YK
929 ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src);
930 ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst);
1da177e4
LT
931 nf_reset(skb);
932 pkt_len = skb->len;
ef76bc23 933 err = ip6_local_out(skb);
1da177e4 934
b9df3cb8 935 if (net_xmit_eval(err) == 0) {
1da177e4
LT
936 stats->tx_bytes += pkt_len;
937 stats->tx_packets++;
938 } else {
939 stats->tx_errors++;
940 stats->tx_aborted_errors++;
941 }
942 ip6_tnl_dst_store(t, dst);
1da177e4
LT
943 return 0;
944tx_err_link_failure:
945 stats->tx_carrier_errors++;
946 dst_link_failure(skb);
947tx_err_dst_release:
948 dst_release(dst);
61ec2aec
YK
949 return err;
950}
951
c4d3efaf
YK
952static inline int
953ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
954{
955 struct ip6_tnl *t = netdev_priv(dev);
eddc9ec5 956 struct iphdr *iph = ip_hdr(skb);
c4d3efaf
YK
957 int encap_limit = -1;
958 struct flowi fl;
959 __u8 dsfield;
960 __u32 mtu;
961 int err;
962
502b0935
YK
963 if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
964 !ip6_tnl_xmit_ctl(t))
c4d3efaf
YK
965 return -1;
966
967 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
968 encap_limit = t->parms.encap_limit;
969
970 memcpy(&fl, &t->fl, sizeof (fl));
971 fl.proto = IPPROTO_IPIP;
972
973 dsfield = ipv4_get_dsfield(iph);
974
975 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
b77f2fa6
AV
976 fl.fl6_flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
977 & IPV6_TCLASS_MASK;
c4d3efaf
YK
978
979 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
980 if (err != 0) {
981 /* XXX: send ICMP error even if DF is not set. */
982 if (err == -EMSGSIZE)
983 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
984 htonl(mtu));
985 return -1;
986 }
987
988 return 0;
989}
990
61ec2aec
YK
991static inline int
992ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
993{
994 struct ip6_tnl *t = netdev_priv(dev);
0660e03f 995 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
61ec2aec
YK
996 int encap_limit = -1;
997 __u16 offset;
998 struct flowi fl;
999 __u8 dsfield;
1000 __u32 mtu;
1001 int err;
1002
502b0935
YK
1003 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
1004 !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
61ec2aec
YK
1005 return -1;
1006
d56f90a7
ACM
1007 offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb));
1008 if (offset > 0) {
61ec2aec 1009 struct ipv6_tlv_tnl_enc_lim *tel;
d56f90a7 1010 tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
61ec2aec
YK
1011 if (tel->encap_limit == 0) {
1012 icmpv6_send(skb, ICMPV6_PARAMPROB,
1013 ICMPV6_HDR_FIELD, offset + 2, skb->dev);
1014 return -1;
1015 }
1016 encap_limit = tel->encap_limit - 1;
1017 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1018 encap_limit = t->parms.encap_limit;
1019
1020 memcpy(&fl, &t->fl, sizeof (fl));
1021 fl.proto = IPPROTO_IPV6;
1022
1023 dsfield = ipv6_get_dsfield(ipv6h);
1024 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
1025 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1026 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
1027 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1028
1029 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
1030 if (err != 0) {
1031 if (err == -EMSGSIZE)
1032 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev);
1033 return -1;
1034 }
1035
1036 return 0;
1037}
1038
1039static int
1040ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1041{
1042 struct ip6_tnl *t = netdev_priv(dev);
1043 struct net_device_stats *stats = &t->stat;
1044 int ret;
1045
1046 if (t->recursion++) {
1047 t->stat.collisions++;
1048 goto tx_err;
1049 }
1050
1051 switch (skb->protocol) {
c4d3efaf
YK
1052 case __constant_htons(ETH_P_IP):
1053 ret = ip4ip6_tnl_xmit(skb, dev);
1054 break;
61ec2aec
YK
1055 case __constant_htons(ETH_P_IPV6):
1056 ret = ip6ip6_tnl_xmit(skb, dev);
1057 break;
1058 default:
1059 goto tx_err;
1060 }
1061
1062 if (ret < 0)
1063 goto tx_err;
1064
1065 t->recursion--;
1066 return 0;
1067
1da177e4
LT
1068tx_err:
1069 stats->tx_errors++;
1070 stats->tx_dropped++;
1071 kfree_skb(skb);
1072 t->recursion--;
1073 return 0;
1074}
1075
1076static void ip6_tnl_set_cap(struct ip6_tnl *t)
1077{
1078 struct ip6_tnl_parm *p = &t->parms;
09c6bbf0
VN
1079 int ltype = ipv6_addr_type(&p->laddr);
1080 int rtype = ipv6_addr_type(&p->raddr);
1da177e4
LT
1081
1082 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);
1083
09c6bbf0
VN
1084 if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1085 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1086 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
305d4b3c 1087 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
09c6bbf0
VN
1088 if (ltype&IPV6_ADDR_UNICAST)
1089 p->flags |= IP6_TNL_F_CAP_XMIT;
1090 if (rtype&IPV6_ADDR_UNICAST)
1091 p->flags |= IP6_TNL_F_CAP_RCV;
1da177e4
LT
1092 }
1093}
1094
3144581c 1095static void ip6_tnl_link_config(struct ip6_tnl *t)
1da177e4
LT
1096{
1097 struct net_device *dev = t->dev;
1098 struct ip6_tnl_parm *p = &t->parms;
1099 struct flowi *fl = &t->fl;
1100
1101 memcpy(&dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1102 memcpy(&dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1103
1104 /* Set up flowi template */
1105 ipv6_addr_copy(&fl->fl6_src, &p->laddr);
1106 ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
1107 fl->oif = p->link;
1108 fl->fl6_flowlabel = 0;
1109
1110 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
1111 fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1112 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
1113 fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1114
1115 ip6_tnl_set_cap(t);
1116
1117 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1118 dev->flags |= IFF_POINTOPOINT;
1119 else
1120 dev->flags &= ~IFF_POINTOPOINT;
1121
1122 dev->iflink = p->link;
1123
1124 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN
1125 int strict = (ipv6_addr_type(&p->raddr) &
1126 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1127
606a2b48 1128 struct rt6_info *rt = rt6_lookup(&init_net, &p->raddr, &p->laddr,
305d4b3c 1129 p->link, strict);
1da177e4
LT
1130
1131 if (rt == NULL)
1132 return;
1133
1134 if (rt->rt6i_dev) {
1135 dev->hard_header_len = rt->rt6i_dev->hard_header_len +
1136 sizeof (struct ipv6hdr);
1137
1138 dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr);
1139
1140 if (dev->mtu < IPV6_MIN_MTU)
1141 dev->mtu = IPV6_MIN_MTU;
1142 }
1143 dst_release(&rt->u.dst);
1144 }
1145}
1146
1147/**
3144581c 1148 * ip6_tnl_change - update the tunnel parameters
1da177e4
LT
1149 * @t: tunnel to be changed
1150 * @p: tunnel configuration parameters
1151 * @active: != 0 if tunnel is ready for use
1152 *
1153 * Description:
3144581c 1154 * ip6_tnl_change() updates the tunnel parameters
1da177e4
LT
1155 **/
1156
1157static int
3144581c 1158ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
1da177e4
LT
1159{
1160 ipv6_addr_copy(&t->parms.laddr, &p->laddr);
1161 ipv6_addr_copy(&t->parms.raddr, &p->raddr);
1162 t->parms.flags = p->flags;
1163 t->parms.hop_limit = p->hop_limit;
1164 t->parms.encap_limit = p->encap_limit;
1165 t->parms.flowinfo = p->flowinfo;
8181b8c1 1166 t->parms.link = p->link;
502b0935 1167 t->parms.proto = p->proto;
0c088890 1168 ip6_tnl_dst_reset(t);
3144581c 1169 ip6_tnl_link_config(t);
1da177e4
LT
1170 return 0;
1171}
1172
1173/**
3144581c 1174 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1da177e4
LT
1175 * @dev: virtual device associated with tunnel
1176 * @ifr: parameters passed from userspace
1177 * @cmd: command to be performed
1178 *
1179 * Description:
3144581c 1180 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1ab1457c 1181 * from userspace.
1da177e4
LT
1182 *
1183 * The possible commands are the following:
1184 * %SIOCGETTUNNEL: get tunnel parameters for device
1185 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1186 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1187 * %SIOCDELTUNNEL: delete tunnel
1188 *
1ab1457c 1189 * The fallback device "ip6tnl0", created during module
1da177e4
LT
1190 * initialization, can be used for creating other tunnel devices.
1191 *
1192 * Return:
1193 * 0 on success,
1194 * %-EFAULT if unable to copy data to or from userspace,
1195 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1196 * %-EINVAL if passed tunnel parameters are invalid,
1197 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1198 * %-ENODEV if attempting to change or delete a nonexisting device
1199 **/
1200
1201static int
3144581c 1202ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1da177e4
LT
1203{
1204 int err = 0;
1da177e4
LT
1205 struct ip6_tnl_parm p;
1206 struct ip6_tnl *t = NULL;
2dd02c89
PE
1207 struct net *net = dev_net(dev);
1208 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4
LT
1209
1210 switch (cmd) {
1211 case SIOCGETTUNNEL:
3144581c 1212 if (dev == ip6_fb_tnl_dev) {
567131a7 1213 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1da177e4
LT
1214 err = -EFAULT;
1215 break;
1216 }
2dd02c89 1217 t = ip6_tnl_locate(net, &p, 0);
567131a7
VN
1218 }
1219 if (t == NULL)
2941a486 1220 t = netdev_priv(dev);
1da177e4
LT
1221 memcpy(&p, &t->parms, sizeof (p));
1222 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
1223 err = -EFAULT;
1224 }
1225 break;
1226 case SIOCADDTUNNEL:
1227 case SIOCCHGTUNNEL:
1228 err = -EPERM;
1da177e4
LT
1229 if (!capable(CAP_NET_ADMIN))
1230 break;
567131a7
VN
1231 err = -EFAULT;
1232 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1233 break;
567131a7 1234 err = -EINVAL;
502b0935
YK
1235 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1236 p.proto != 0)
1da177e4 1237 break;
2dd02c89 1238 t = ip6_tnl_locate(net, &p, cmd == SIOCADDTUNNEL);
3144581c 1239 if (dev != ip6_fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
567131a7
VN
1240 if (t != NULL) {
1241 if (t->dev != dev) {
1242 err = -EEXIST;
1243 break;
1244 }
1245 } else
1246 t = netdev_priv(dev);
1247
2dd02c89 1248 ip6_tnl_unlink(ip6n, t);
3144581c 1249 err = ip6_tnl_change(t, &p);
2dd02c89 1250 ip6_tnl_link(ip6n, t);
1da177e4
LT
1251 netdev_state_change(dev);
1252 }
567131a7 1253 if (t) {
1da177e4 1254 err = 0;
567131a7
VN
1255 if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
1256 err = -EFAULT;
1257
1258 } else
1259 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1da177e4
LT
1260 break;
1261 case SIOCDELTUNNEL:
1262 err = -EPERM;
1263 if (!capable(CAP_NET_ADMIN))
1264 break;
1265
3144581c 1266 if (dev == ip6_fb_tnl_dev) {
567131a7
VN
1267 err = -EFAULT;
1268 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1269 break;
567131a7 1270 err = -ENOENT;
2dd02c89 1271 if ((t = ip6_tnl_locate(net, &p, 0)) == NULL)
1da177e4 1272 break;
567131a7 1273 err = -EPERM;
3144581c 1274 if (t->dev == ip6_fb_tnl_dev)
1da177e4 1275 break;
567131a7 1276 dev = t->dev;
1da177e4 1277 }
22f8cde5
SH
1278 err = 0;
1279 unregister_netdevice(dev);
1da177e4
LT
1280 break;
1281 default:
1282 err = -EINVAL;
1283 }
1284 return err;
1285}
1286
1287/**
3144581c 1288 * ip6_tnl_get_stats - return the stats for tunnel device
1da177e4
LT
1289 * @dev: virtual device associated with tunnel
1290 *
1291 * Return: stats for device
1292 **/
1293
1294static struct net_device_stats *
3144581c 1295ip6_tnl_get_stats(struct net_device *dev)
1da177e4 1296{
2941a486 1297 return &(((struct ip6_tnl *)netdev_priv(dev))->stat);
1da177e4
LT
1298}
1299
1300/**
3144581c 1301 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1da177e4
LT
1302 * @dev: virtual device associated with tunnel
1303 * @new_mtu: the new mtu
1304 *
1305 * Return:
1306 * 0 on success,
1307 * %-EINVAL if mtu too small
1308 **/
1309
1310static int
3144581c 1311ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1da177e4
LT
1312{
1313 if (new_mtu < IPV6_MIN_MTU) {
1314 return -EINVAL;
1315 }
1316 dev->mtu = new_mtu;
1317 return 0;
1318}
1319
1320/**
3144581c 1321 * ip6_tnl_dev_setup - setup virtual tunnel device
1da177e4
LT
1322 * @dev: virtual device associated with tunnel
1323 *
1324 * Description:
1325 * Initialize function pointers and device parameters
1326 **/
1327
3144581c 1328static void ip6_tnl_dev_setup(struct net_device *dev)
1da177e4 1329{
3144581c 1330 dev->uninit = ip6_tnl_dev_uninit;
1da177e4 1331 dev->destructor = free_netdev;
61ec2aec 1332 dev->hard_start_xmit = ip6_tnl_xmit;
3144581c
YK
1333 dev->get_stats = ip6_tnl_get_stats;
1334 dev->do_ioctl = ip6_tnl_ioctl;
1335 dev->change_mtu = ip6_tnl_change_mtu;
1da177e4
LT
1336
1337 dev->type = ARPHRD_TUNNEL6;
1338 dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1339 dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1340 dev->flags |= IFF_NOARP;
1341 dev->addr_len = sizeof(struct in6_addr);
1342}
1343
1344
1345/**
3144581c 1346 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1da177e4
LT
1347 * @dev: virtual device associated with tunnel
1348 **/
1349
1350static inline void
3144581c 1351ip6_tnl_dev_init_gen(struct net_device *dev)
1da177e4 1352{
2941a486 1353 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT
1354 t->dev = dev;
1355 strcpy(t->parms.name, dev->name);
1356}
1357
1358/**
3144581c 1359 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1da177e4
LT
1360 * @dev: virtual device associated with tunnel
1361 **/
1362
1363static int
3144581c 1364ip6_tnl_dev_init(struct net_device *dev)
1da177e4 1365{
2941a486 1366 struct ip6_tnl *t = netdev_priv(dev);
3144581c
YK
1367 ip6_tnl_dev_init_gen(dev);
1368 ip6_tnl_link_config(t);
1da177e4
LT
1369 return 0;
1370}
1371
1372/**
3144581c 1373 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1da177e4
LT
1374 * @dev: fallback device
1375 *
1376 * Return: 0
1377 **/
1378
1ab1457c 1379static int
3144581c 1380ip6_fb_tnl_dev_init(struct net_device *dev)
1da177e4 1381{
2941a486 1382 struct ip6_tnl *t = netdev_priv(dev);
3144581c 1383 ip6_tnl_dev_init_gen(dev);
502b0935 1384 t->parms.proto = IPPROTO_IPV6;
1da177e4
LT
1385 dev_hold(dev);
1386 tnls_wc[0] = t;
1387 return 0;
1388}
1389
c4d3efaf
YK
1390static struct xfrm6_tunnel ip4ip6_handler = {
1391 .handler = ip4ip6_rcv,
1392 .err_handler = ip4ip6_err,
1393 .priority = 1,
1394};
1395
1da177e4 1396static struct xfrm6_tunnel ip6ip6_handler = {
0303770d
PM
1397 .handler = ip6ip6_rcv,
1398 .err_handler = ip6ip6_err,
d2acc347 1399 .priority = 1,
1da177e4
LT
1400};
1401
13eeb8e9
PE
1402static int ip6_tnl_init_net(struct net *net)
1403{
1404 int err;
1405 struct ip6_tnl_net *ip6n;
1406
1407 err = -ENOMEM;
1408 ip6n = kmalloc(sizeof(struct ip6_tnl_net), GFP_KERNEL);
1409 if (ip6n == NULL)
1410 goto err_alloc;
1411
1412 err = net_assign_generic(net, ip6_tnl_net_id, ip6n);
1413 if (err < 0)
1414 goto err_assign;
1415
1416 return 0;
1417
1418err_assign:
1419 kfree(ip6n);
1420err_alloc:
1421 return err;
1422}
1423
1424static void ip6_tnl_exit_net(struct net *net)
1425{
1426 struct ip6_tnl_net *ip6n;
1427
1428 ip6n = net_generic(net, ip6_tnl_net_id);
1429 kfree(ip6n);
1430}
1431
1432static struct pernet_operations ip6_tnl_net_ops = {
1433 .init = ip6_tnl_init_net,
1434 .exit = ip6_tnl_exit_net,
1435};
1436
1da177e4
LT
1437/**
1438 * ip6_tunnel_init - register protocol and reserve needed resources
1439 *
1440 * Return: 0 on success
1441 **/
1442
1443static int __init ip6_tunnel_init(void)
1444{
1445 int err;
1446
c4d3efaf 1447 if (xfrm6_tunnel_register(&ip4ip6_handler, AF_INET)) {
3144581c 1448 printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n");
c4d3efaf
YK
1449 err = -EAGAIN;
1450 goto out;
1451 }
1452
73d605d1 1453 if (xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6)) {
3144581c 1454 printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n");
c4d3efaf
YK
1455 err = -EAGAIN;
1456 goto unreg_ip4ip6;
1da177e4 1457 }
3144581c
YK
1458 ip6_fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1459 ip6_tnl_dev_setup);
1da177e4 1460
3144581c 1461 if (!ip6_fb_tnl_dev) {
1da177e4
LT
1462 err = -ENOMEM;
1463 goto fail;
1464 }
3144581c 1465 ip6_fb_tnl_dev->init = ip6_fb_tnl_dev_init;
1da177e4 1466
3144581c
YK
1467 if ((err = register_netdev(ip6_fb_tnl_dev))) {
1468 free_netdev(ip6_fb_tnl_dev);
1da177e4
LT
1469 goto fail;
1470 }
13eeb8e9
PE
1471
1472 err = register_pernet_gen_device(&ip6_tnl_net_id, &ip6_tnl_net_ops);
1473 if (err < 0)
1474 goto err_pernet;
1da177e4 1475 return 0;
13eeb8e9
PE
1476err_pernet:
1477 unregister_netdevice(ip6_fb_tnl_dev);
1da177e4 1478fail:
73d605d1 1479 xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6);
c4d3efaf
YK
1480unreg_ip4ip6:
1481 xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
1482out:
1da177e4
LT
1483 return err;
1484}
1485
3144581c 1486static void __exit ip6_tnl_destroy_tunnels(void)
b3fdd9f1
YK
1487{
1488 int h;
1489 struct ip6_tnl *t;
1490
1491 for (h = 0; h < HASH_SIZE; h++) {
1492 while ((t = tnls_r_l[h]) != NULL)
1493 unregister_netdevice(t->dev);
1494 }
1495
1496 t = tnls_wc[0];
1497 unregister_netdevice(t->dev);
1498}
1499
1da177e4
LT
1500/**
1501 * ip6_tunnel_cleanup - free resources and unregister protocol
1502 **/
1503
1504static void __exit ip6_tunnel_cleanup(void)
1505{
c4d3efaf 1506 if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
3144581c 1507 printk(KERN_INFO "ip6_tunnel close: can't deregister ip4ip6\n");
c4d3efaf 1508
73d605d1 1509 if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
3144581c 1510 printk(KERN_INFO "ip6_tunnel close: can't deregister ip6ip6\n");
1da177e4 1511
b3fdd9f1 1512 rtnl_lock();
3144581c 1513 ip6_tnl_destroy_tunnels();
b3fdd9f1 1514 rtnl_unlock();
13eeb8e9
PE
1515
1516 unregister_pernet_gen_device(ip6_tnl_net_id, &ip6_tnl_net_ops);
1da177e4
LT
1517}
1518
1519module_init(ip6_tunnel_init);
1520module_exit(ip6_tunnel_cleanup);