[IP6TUNNEL]: Make the fallback tunnel device per-net.
[GitHub/mt8127/android_kernel_alcatel_ttab.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 {
15820e12
PE
85 /* the IPv6 tunnel fallback device */
86 struct net_device *fb_tnl_dev;
13eeb8e9
PE
87};
88
1da177e4
LT
89/* lists for storing tunnels in use */
90static struct ip6_tnl *tnls_r_l[HASH_SIZE];
91static struct ip6_tnl *tnls_wc[1];
92static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l };
93
94/* lock for the tunnel lists */
3144581c 95static DEFINE_RWLOCK(ip6_tnl_lock);
1da177e4
LT
96
97static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
98{
99 struct dst_entry *dst = t->dst_cache;
100
1ab1457c 101 if (dst && dst->obsolete &&
1da177e4
LT
102 dst->ops->check(dst, t->dst_cookie) == NULL) {
103 t->dst_cache = NULL;
104 dst_release(dst);
105 return NULL;
106 }
107
108 return dst;
109}
110
111static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
112{
113 dst_release(t->dst_cache);
114 t->dst_cache = NULL;
115}
116
117static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
118{
119 struct rt6_info *rt = (struct rt6_info *) dst;
120 t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
121 dst_release(t->dst_cache);
122 t->dst_cache = dst;
123}
124
125/**
3144581c 126 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
1ab1457c
YH
127 * @remote: the address of the tunnel exit-point
128 * @local: the address of the tunnel entry-point
1da177e4 129 *
1ab1457c 130 * Return:
1da177e4 131 * tunnel matching given end-points if found,
1ab1457c 132 * else fallback tunnel if its device is up,
1da177e4
LT
133 * else %NULL
134 **/
135
136static struct ip6_tnl *
2dd02c89 137ip6_tnl_lookup(struct net *net, struct in6_addr *remote, struct in6_addr *local)
1da177e4
LT
138{
139 unsigned h0 = HASH(remote);
140 unsigned h1 = HASH(local);
141 struct ip6_tnl *t;
142
143 for (t = tnls_r_l[h0 ^ h1]; t; t = t->next) {
144 if (ipv6_addr_equal(local, &t->parms.laddr) &&
145 ipv6_addr_equal(remote, &t->parms.raddr) &&
146 (t->dev->flags & IFF_UP))
147 return t;
148 }
149 if ((t = tnls_wc[0]) != NULL && (t->dev->flags & IFF_UP))
150 return t;
151
152 return NULL;
153}
154
155/**
3144581c 156 * ip6_tnl_bucket - get head of list matching given tunnel parameters
1ab1457c 157 * @p: parameters containing tunnel end-points
1da177e4
LT
158 *
159 * Description:
3144581c 160 * ip6_tnl_bucket() returns the head of the list matching the
1da177e4
LT
161 * &struct in6_addr entries laddr and raddr in @p.
162 *
1ab1457c 163 * Return: head of IPv6 tunnel list
1da177e4
LT
164 **/
165
166static struct ip6_tnl **
2dd02c89 167ip6_tnl_bucket(struct ip6_tnl_net *ip6n, struct ip6_tnl_parm *p)
1da177e4
LT
168{
169 struct in6_addr *remote = &p->raddr;
170 struct in6_addr *local = &p->laddr;
171 unsigned h = 0;
172 int prio = 0;
173
174 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
175 prio = 1;
176 h = HASH(remote) ^ HASH(local);
177 }
178 return &tnls[prio][h];
179}
180
181/**
3144581c 182 * ip6_tnl_link - add tunnel to hash table
1da177e4
LT
183 * @t: tunnel to be added
184 **/
185
186static void
2dd02c89 187ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 188{
2dd02c89 189 struct ip6_tnl **tp = ip6_tnl_bucket(ip6n, &t->parms);
1da177e4
LT
190
191 t->next = *tp;
3144581c 192 write_lock_bh(&ip6_tnl_lock);
1da177e4 193 *tp = t;
3144581c 194 write_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
195}
196
197/**
3144581c 198 * ip6_tnl_unlink - remove tunnel from hash table
1da177e4
LT
199 * @t: tunnel to be removed
200 **/
201
202static void
2dd02c89 203ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4
LT
204{
205 struct ip6_tnl **tp;
206
2dd02c89 207 for (tp = ip6_tnl_bucket(ip6n, &t->parms); *tp; tp = &(*tp)->next) {
1da177e4 208 if (t == *tp) {
3144581c 209 write_lock_bh(&ip6_tnl_lock);
1da177e4 210 *tp = t->next;
3144581c 211 write_unlock_bh(&ip6_tnl_lock);
1da177e4
LT
212 break;
213 }
214 }
215}
216
217/**
218 * ip6_tnl_create() - create a new tunnel
219 * @p: tunnel parameters
220 * @pt: pointer to new tunnel
221 *
222 * Description:
223 * Create tunnel matching given parameters.
1ab1457c
YH
224 *
225 * Return:
567131a7 226 * created tunnel or NULL
1da177e4
LT
227 **/
228
2dd02c89 229static struct ip6_tnl *ip6_tnl_create(struct net *net, struct ip6_tnl_parm *p)
1da177e4
LT
230{
231 struct net_device *dev;
232 struct ip6_tnl *t;
233 char name[IFNAMSIZ];
234 int err;
2dd02c89 235 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 236
34cc7ba6 237 if (p->name[0])
1da177e4 238 strlcpy(name, p->name, IFNAMSIZ);
34cc7ba6
PE
239 else
240 sprintf(name, "ip6tnl%%d");
241
3144581c 242 dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
1da177e4 243 if (dev == NULL)
567131a7 244 goto failed;
1da177e4 245
b37d428b
PE
246 if (strchr(name, '%')) {
247 if (dev_alloc_name(dev, name) < 0)
248 goto failed_free;
249 }
250
2941a486 251 t = netdev_priv(dev);
3144581c 252 dev->init = ip6_tnl_dev_init;
1da177e4
LT
253 t->parms = *p;
254
b37d428b
PE
255 if ((err = register_netdevice(dev)) < 0)
256 goto failed_free;
257
1da177e4 258 dev_hold(dev);
2dd02c89 259 ip6_tnl_link(ip6n, t);
567131a7 260 return t;
b37d428b
PE
261
262failed_free:
263 free_netdev(dev);
567131a7
VN
264failed:
265 return NULL;
1da177e4
LT
266}
267
268/**
3144581c 269 * ip6_tnl_locate - find or create tunnel matching given parameters
1ab1457c 270 * @p: tunnel parameters
1da177e4
LT
271 * @create: != 0 if allowed to create new tunnel if no match found
272 *
273 * Description:
3144581c 274 * ip6_tnl_locate() first tries to locate an existing tunnel
1da177e4
LT
275 * based on @parms. If this is unsuccessful, but @create is set a new
276 * tunnel device is created and registered for use.
277 *
278 * Return:
567131a7 279 * matching tunnel or NULL
1da177e4
LT
280 **/
281
2dd02c89
PE
282static struct ip6_tnl *ip6_tnl_locate(struct net *net,
283 struct ip6_tnl_parm *p, int create)
1da177e4
LT
284{
285 struct in6_addr *remote = &p->raddr;
286 struct in6_addr *local = &p->laddr;
287 struct ip6_tnl *t;
2dd02c89 288 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 289
2dd02c89 290 for (t = *ip6_tnl_bucket(ip6n, p); t; t = t->next) {
1da177e4 291 if (ipv6_addr_equal(local, &t->parms.laddr) &&
567131a7
VN
292 ipv6_addr_equal(remote, &t->parms.raddr))
293 return t;
1da177e4
LT
294 }
295 if (!create)
567131a7 296 return NULL;
2dd02c89 297 return ip6_tnl_create(net, p);
1da177e4
LT
298}
299
300/**
3144581c 301 * ip6_tnl_dev_uninit - tunnel device uninitializer
1da177e4 302 * @dev: the device to be destroyed
1ab1457c 303 *
1da177e4 304 * Description:
3144581c 305 * ip6_tnl_dev_uninit() removes tunnel from its list
1da177e4
LT
306 **/
307
308static void
3144581c 309ip6_tnl_dev_uninit(struct net_device *dev)
1da177e4 310{
2941a486 311 struct ip6_tnl *t = netdev_priv(dev);
2dd02c89
PE
312 struct net *net = dev_net(dev);
313 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 314
15820e12 315 if (dev == ip6n->fb_tnl_dev) {
3144581c 316 write_lock_bh(&ip6_tnl_lock);
1da177e4 317 tnls_wc[0] = NULL;
3144581c 318 write_unlock_bh(&ip6_tnl_lock);
1da177e4 319 } else {
2dd02c89 320 ip6_tnl_unlink(ip6n, t);
1da177e4
LT
321 }
322 ip6_tnl_dst_reset(t);
323 dev_put(dev);
324}
325
326/**
327 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
328 * @skb: received socket buffer
329 *
1ab1457c
YH
330 * Return:
331 * 0 if none was found,
1da177e4
LT
332 * else index to encapsulation limit
333 **/
334
335static __u16
336parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
337{
338 struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
339 __u8 nexthdr = ipv6h->nexthdr;
340 __u16 off = sizeof (*ipv6h);
341
342 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
343 __u16 optlen = 0;
344 struct ipv6_opt_hdr *hdr;
345 if (raw + off + sizeof (*hdr) > skb->data &&
346 !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
347 break;
348
349 hdr = (struct ipv6_opt_hdr *) (raw + off);
350 if (nexthdr == NEXTHDR_FRAGMENT) {
351 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
352 if (frag_hdr->frag_off)
353 break;
354 optlen = 8;
355 } else if (nexthdr == NEXTHDR_AUTH) {
356 optlen = (hdr->hdrlen + 2) << 2;
357 } else {
358 optlen = ipv6_optlen(hdr);
359 }
360 if (nexthdr == NEXTHDR_DEST) {
361 __u16 i = off + 2;
362 while (1) {
363 struct ipv6_tlv_tnl_enc_lim *tel;
364
365 /* No more room for encapsulation limit */
366 if (i + sizeof (*tel) > off + optlen)
367 break;
368
369 tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
370 /* return index of option if found and valid */
371 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
372 tel->length == 1)
373 return i;
374 /* else jump to next option */
375 if (tel->type)
376 i += tel->length + 2;
377 else
378 i++;
379 }
380 }
381 nexthdr = hdr->nexthdr;
382 off += optlen;
383 }
384 return 0;
385}
386
387/**
e490d1d8 388 * ip6_tnl_err - tunnel error handler
1da177e4
LT
389 *
390 * Description:
e490d1d8 391 * ip6_tnl_err() should handle errors in the tunnel according
1da177e4
LT
392 * to the specifications in RFC 2473.
393 **/
394
d2acc347 395static int
502b0935 396ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
704eae1f 397 int *type, int *code, int *msg, __u32 *info, int offset)
1da177e4
LT
398{
399 struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
400 struct ip6_tnl *t;
401 int rel_msg = 0;
402 int rel_type = ICMPV6_DEST_UNREACH;
403 int rel_code = ICMPV6_ADDR_UNREACH;
404 __u32 rel_info = 0;
405 __u16 len;
d2acc347 406 int err = -ENOENT;
1da177e4 407
1ab1457c
YH
408 /* If the packet doesn't contain the original IPv6 header we are
409 in trouble since we might need the source address for further
1da177e4
LT
410 processing of the error. */
411
3144581c 412 read_lock(&ip6_tnl_lock);
8704ca7e 413 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
2dd02c89 414 &ipv6h->saddr)) == NULL)
1da177e4
LT
415 goto out;
416
502b0935
YK
417 if (t->parms.proto != ipproto && t->parms.proto != 0)
418 goto out;
419
d2acc347
HX
420 err = 0;
421
e490d1d8 422 switch (*type) {
1da177e4
LT
423 __u32 teli;
424 struct ipv6_tlv_tnl_enc_lim *tel;
425 __u32 mtu;
426 case ICMPV6_DEST_UNREACH:
427 if (net_ratelimit())
428 printk(KERN_WARNING
429 "%s: Path to destination invalid "
430 "or inactive!\n", t->parms.name);
431 rel_msg = 1;
432 break;
433 case ICMPV6_TIME_EXCEED:
e490d1d8 434 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
1da177e4
LT
435 if (net_ratelimit())
436 printk(KERN_WARNING
437 "%s: Too small hop limit or "
1ab1457c 438 "routing loop in tunnel!\n",
1da177e4
LT
439 t->parms.name);
440 rel_msg = 1;
441 }
442 break;
443 case ICMPV6_PARAMPROB:
107a5fe6 444 teli = 0;
e490d1d8 445 if ((*code) == ICMPV6_HDR_FIELD)
107a5fe6 446 teli = parse_tlv_tnl_enc_lim(skb, skb->data);
1da177e4 447
704eae1f 448 if (teli && teli == *info - 2) {
1da177e4
LT
449 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
450 if (tel->encap_limit == 0) {
451 if (net_ratelimit())
452 printk(KERN_WARNING
453 "%s: Too small encapsulation "
454 "limit or routing loop in "
455 "tunnel!\n", t->parms.name);
456 rel_msg = 1;
457 }
107a5fe6
VN
458 } else if (net_ratelimit()) {
459 printk(KERN_WARNING
460 "%s: Recipient unable to parse tunneled "
461 "packet!\n ", t->parms.name);
1da177e4
LT
462 }
463 break;
464 case ICMPV6_PKT_TOOBIG:
704eae1f 465 mtu = *info - offset;
1da177e4
LT
466 if (mtu < IPV6_MIN_MTU)
467 mtu = IPV6_MIN_MTU;
468 t->dev->mtu = mtu;
469
cc6cdac0 470 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
1da177e4
LT
471 rel_type = ICMPV6_PKT_TOOBIG;
472 rel_code = 0;
473 rel_info = mtu;
474 rel_msg = 1;
475 }
476 break;
477 }
e490d1d8
YK
478
479 *type = rel_type;
480 *code = rel_code;
481 *info = rel_info;
482 *msg = rel_msg;
483
484out:
3144581c 485 read_unlock(&ip6_tnl_lock);
e490d1d8
YK
486 return err;
487}
488
c4d3efaf
YK
489static int
490ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
704eae1f 491 int type, int code, int offset, __be32 info)
c4d3efaf
YK
492{
493 int rel_msg = 0;
494 int rel_type = type;
495 int rel_code = code;
704eae1f 496 __u32 rel_info = ntohl(info);
c4d3efaf
YK
497 int err;
498 struct sk_buff *skb2;
499 struct iphdr *eiph;
500 struct flowi fl;
501 struct rtable *rt;
502
502b0935
YK
503 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
504 &rel_msg, &rel_info, offset);
c4d3efaf
YK
505 if (err < 0)
506 return err;
507
508 if (rel_msg == 0)
509 return 0;
510
511 switch (rel_type) {
512 case ICMPV6_DEST_UNREACH:
513 if (rel_code != ICMPV6_ADDR_UNREACH)
514 return 0;
515 rel_type = ICMP_DEST_UNREACH;
516 rel_code = ICMP_HOST_UNREACH;
517 break;
518 case ICMPV6_PKT_TOOBIG:
519 if (rel_code != 0)
520 return 0;
521 rel_type = ICMP_DEST_UNREACH;
522 rel_code = ICMP_FRAG_NEEDED;
523 break;
524 default:
525 return 0;
526 }
527
528 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
529 return 0;
530
531 skb2 = skb_clone(skb, GFP_ATOMIC);
532 if (!skb2)
533 return 0;
534
535 dst_release(skb2->dst);
536 skb2->dst = NULL;
537 skb_pull(skb2, offset);
c1d2bbe1 538 skb_reset_network_header(skb2);
eddc9ec5 539 eiph = ip_hdr(skb2);
c4d3efaf
YK
540
541 /* Try to guess incoming interface */
542 memset(&fl, 0, sizeof(fl));
543 fl.fl4_dst = eiph->saddr;
544 fl.fl4_tos = RT_TOS(eiph->tos);
545 fl.proto = IPPROTO_IPIP;
f206351a 546 if (ip_route_output_key(&init_net, &rt, &fl))
c4d3efaf
YK
547 goto out;
548
549 skb2->dev = rt->u.dst.dev;
550
551 /* route "incoming" packet */
552 if (rt->rt_flags & RTCF_LOCAL) {
553 ip_rt_put(rt);
554 rt = NULL;
555 fl.fl4_dst = eiph->daddr;
556 fl.fl4_src = eiph->saddr;
557 fl.fl4_tos = eiph->tos;
f206351a 558 if (ip_route_output_key(&init_net, &rt, &fl) ||
c4d3efaf
YK
559 rt->u.dst.dev->type != ARPHRD_TUNNEL) {
560 ip_rt_put(rt);
561 goto out;
562 }
9937ded8 563 skb2->dst = (struct dst_entry *)rt;
c4d3efaf
YK
564 } else {
565 ip_rt_put(rt);
566 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
567 skb2->dev) ||
568 skb2->dst->dev->type != ARPHRD_TUNNEL)
569 goto out;
570 }
571
572 /* change mtu on this route */
573 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
574 if (rel_info > dst_mtu(skb2->dst))
575 goto out;
576
577 skb2->dst->ops->update_pmtu(skb2->dst, rel_info);
c4d3efaf
YK
578 }
579
704eae1f 580 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
c4d3efaf
YK
581
582out:
583 kfree_skb(skb2);
584 return 0;
585}
586
e490d1d8
YK
587static int
588ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
704eae1f 589 int type, int code, int offset, __be32 info)
e490d1d8
YK
590{
591 int rel_msg = 0;
592 int rel_type = type;
593 int rel_code = code;
704eae1f 594 __u32 rel_info = ntohl(info);
e490d1d8
YK
595 int err;
596
502b0935
YK
597 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
598 &rel_msg, &rel_info, offset);
e490d1d8
YK
599 if (err < 0)
600 return err;
601
602 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
1da177e4
LT
603 struct rt6_info *rt;
604 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 605
1da177e4 606 if (!skb2)
e490d1d8 607 return 0;
1da177e4
LT
608
609 dst_release(skb2->dst);
610 skb2->dst = NULL;
611 skb_pull(skb2, offset);
c1d2bbe1 612 skb_reset_network_header(skb2);
1da177e4
LT
613
614 /* Try to guess incoming interface */
606a2b48 615 rt = rt6_lookup(&init_net, &ipv6_hdr(skb2)->saddr, NULL, 0, 0);
1da177e4
LT
616
617 if (rt && rt->rt6i_dev)
618 skb2->dev = rt->rt6i_dev;
619
620 icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev);
621
622 if (rt)
623 dst_release(&rt->u.dst);
624
625 kfree_skb(skb2);
626 }
e490d1d8
YK
627
628 return 0;
1da177e4
LT
629}
630
c4d3efaf
YK
631static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
632 struct ipv6hdr *ipv6h,
633 struct sk_buff *skb)
634{
635 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
636
637 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
eddc9ec5 638 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
c4d3efaf
YK
639
640 if (INET_ECN_is_ce(dsfield))
eddc9ec5 641 IP_ECN_set_ce(ip_hdr(skb));
c4d3efaf
YK
642}
643
8359925b
YK
644static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
645 struct ipv6hdr *ipv6h,
646 struct sk_buff *skb)
1da177e4 647{
8359925b 648 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
29bb43b4 649 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
1da177e4 650
8359925b 651 if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
0660e03f 652 IP6_ECN_set_ce(ipv6_hdr(skb));
1da177e4 653}
8359925b 654
09c6bbf0
VN
655static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
656{
657 struct ip6_tnl_parm *p = &t->parms;
658 int ret = 0;
659
660 if (p->flags & IP6_TNL_F_CAP_RCV) {
1ab1457c 661 struct net_device *ldev = NULL;
09c6bbf0
VN
662
663 if (p->link)
881d966b 664 ldev = dev_get_by_index(&init_net, p->link);
09c6bbf0
VN
665
666 if ((ipv6_addr_is_multicast(&p->laddr) ||
bfeade08
DL
667 likely(ipv6_chk_addr(&init_net, &p->laddr, ldev, 0))) &&
668 likely(!ipv6_chk_addr(&init_net, &p->raddr, NULL, 0)))
09c6bbf0
VN
669 ret = 1;
670
671 if (ldev)
672 dev_put(ldev);
673 }
674 return ret;
675}
1da177e4
LT
676
677/**
3144581c 678 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
1da177e4 679 * @skb: received socket buffer
8359925b
YK
680 * @protocol: ethernet protocol ID
681 * @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
1da177e4
LT
682 *
683 * Return: 0
684 **/
685
8359925b 686static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
502b0935 687 __u8 ipproto,
8359925b
YK
688 void (*dscp_ecn_decapsulate)(struct ip6_tnl *t,
689 struct ipv6hdr *ipv6h,
690 struct sk_buff *skb))
1da177e4 691{
1da177e4 692 struct ip6_tnl *t;
0660e03f 693 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
1da177e4 694
3144581c 695 read_lock(&ip6_tnl_lock);
1da177e4 696
8704ca7e 697 if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
2dd02c89 698 &ipv6h->daddr)) != NULL) {
502b0935
YK
699 if (t->parms.proto != ipproto && t->parms.proto != 0) {
700 read_unlock(&ip6_tnl_lock);
701 goto discard;
702 }
703
1da177e4 704 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
3144581c 705 read_unlock(&ip6_tnl_lock);
50fba2aa 706 goto discard;
1da177e4
LT
707 }
708
09c6bbf0 709 if (!ip6_tnl_rcv_ctl(t)) {
1da177e4 710 t->stat.rx_dropped++;
3144581c 711 read_unlock(&ip6_tnl_lock);
1da177e4
LT
712 goto discard;
713 }
714 secpath_reset(skb);
b0e380b1 715 skb->mac_header = skb->network_header;
c1d2bbe1 716 skb_reset_network_header(skb);
8359925b 717 skb->protocol = htons(protocol);
1da177e4
LT
718 skb->pkt_type = PACKET_HOST;
719 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
720 skb->dev = t->dev;
721 dst_release(skb->dst);
722 skb->dst = NULL;
53ab61c6 723 nf_reset(skb);
8359925b
YK
724
725 dscp_ecn_decapsulate(t, ipv6h, skb);
726
1da177e4
LT
727 t->stat.rx_packets++;
728 t->stat.rx_bytes += skb->len;
729 netif_rx(skb);
3144581c 730 read_unlock(&ip6_tnl_lock);
1da177e4
LT
731 return 0;
732 }
3144581c 733 read_unlock(&ip6_tnl_lock);
1da177e4 734 return 1;
50fba2aa
HX
735
736discard:
737 kfree_skb(skb);
738 return 0;
1da177e4
LT
739}
740
c4d3efaf
YK
741static int ip4ip6_rcv(struct sk_buff *skb)
742{
502b0935
YK
743 return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
744 ip4ip6_dscp_ecn_decapsulate);
c4d3efaf
YK
745}
746
8359925b
YK
747static int ip6ip6_rcv(struct sk_buff *skb)
748{
502b0935
YK
749 return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
750 ip6ip6_dscp_ecn_decapsulate);
8359925b
YK
751}
752
6fb32dde
VN
753struct ipv6_tel_txoption {
754 struct ipv6_txoptions ops;
755 __u8 dst_opt[8];
756};
1da177e4 757
6fb32dde
VN
758static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
759{
760 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 761
6fb32dde
VN
762 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
763 opt->dst_opt[3] = 1;
764 opt->dst_opt[4] = encap_limit;
765 opt->dst_opt[5] = IPV6_TLV_PADN;
766 opt->dst_opt[6] = 1;
1da177e4 767
6fb32dde
VN
768 opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
769 opt->ops.opt_nflen = 8;
1da177e4
LT
770}
771
772/**
3144581c 773 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
1da177e4 774 * @t: the outgoing tunnel device
1ab1457c 775 * @hdr: IPv6 header from the incoming packet
1da177e4
LT
776 *
777 * Description:
1ab1457c 778 * Avoid trivial tunneling loop by checking that tunnel exit-point
1da177e4
LT
779 * doesn't match source of incoming packet.
780 *
1ab1457c 781 * Return:
1da177e4
LT
782 * 1 if conflict,
783 * 0 else
784 **/
785
786static inline int
3144581c 787ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
1da177e4
LT
788{
789 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
790}
791
09c6bbf0
VN
792static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
793{
794 struct ip6_tnl_parm *p = &t->parms;
795 int ret = 0;
796
1ab1457c 797 if (p->flags & IP6_TNL_F_CAP_XMIT) {
09c6bbf0
VN
798 struct net_device *ldev = NULL;
799
800 if (p->link)
881d966b 801 ldev = dev_get_by_index(&init_net, p->link);
09c6bbf0 802
bfeade08 803 if (unlikely(!ipv6_chk_addr(&init_net, &p->laddr, ldev, 0)))
09c6bbf0
VN
804 printk(KERN_WARNING
805 "%s xmit: Local address not yet configured!\n",
806 p->name);
807 else if (!ipv6_addr_is_multicast(&p->raddr) &&
bfeade08 808 unlikely(ipv6_chk_addr(&init_net, &p->raddr, NULL, 0)))
09c6bbf0
VN
809 printk(KERN_WARNING
810 "%s xmit: Routing loop! "
811 "Remote address found on this node!\n",
812 p->name);
813 else
814 ret = 1;
815 if (ldev)
816 dev_put(ldev);
817 }
818 return ret;
819}
1da177e4 820/**
61ec2aec 821 * ip6_tnl_xmit2 - encapsulate packet and send
1da177e4 822 * @skb: the outgoing socket buffer
1ab1457c 823 * @dev: the outgoing tunnel device
61ec2aec
YK
824 * @dsfield: dscp code for outer header
825 * @fl: flow of tunneled packet
826 * @encap_limit: encapsulation limit
827 * @pmtu: Path MTU is stored if packet is too big
1da177e4
LT
828 *
829 * Description:
830 * Build new header and do some sanity checks on the packet before sending
831 * it.
832 *
1ab1457c 833 * Return:
c4d3efaf 834 * 0 on success
61ec2aec
YK
835 * -1 fail
836 * %-EMSGSIZE message too big. return mtu in this case.
1da177e4
LT
837 **/
838
61ec2aec
YK
839static int ip6_tnl_xmit2(struct sk_buff *skb,
840 struct net_device *dev,
841 __u8 dsfield,
842 struct flowi *fl,
843 int encap_limit,
844 __u32 *pmtu)
1da177e4 845{
2941a486 846 struct ip6_tnl *t = netdev_priv(dev);
1da177e4 847 struct net_device_stats *stats = &t->stat;
0660e03f 848 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
6fb32dde 849 struct ipv6_tel_txoption opt;
1da177e4
LT
850 struct dst_entry *dst;
851 struct net_device *tdev;
852 int mtu;
c2636b4d 853 unsigned int max_headroom = sizeof(struct ipv6hdr);
1da177e4 854 u8 proto;
61ec2aec 855 int err = -1;
1da177e4 856 int pkt_len;
1da177e4 857
1da177e4
LT
858 if ((dst = ip6_tnl_dst_check(t)) != NULL)
859 dst_hold(dst);
a57ebc90 860 else {
4591db4f 861 dst = ip6_route_output(&init_net, NULL, fl);
1da177e4 862
61ec2aec 863 if (dst->error || xfrm_lookup(&dst, fl, NULL, 0) < 0)
a57ebc90
PM
864 goto tx_err_link_failure;
865 }
1da177e4
LT
866
867 tdev = dst->dev;
868
869 if (tdev == dev) {
870 stats->collisions++;
871 if (net_ratelimit())
1ab1457c 872 printk(KERN_WARNING
1da177e4
LT
873 "%s: Local routing loop detected!\n",
874 t->parms.name);
875 goto tx_err_dst_release;
876 }
877 mtu = dst_mtu(dst) - sizeof (*ipv6h);
6fb32dde 878 if (encap_limit >= 0) {
1da177e4
LT
879 max_headroom += 8;
880 mtu -= 8;
881 }
882 if (mtu < IPV6_MIN_MTU)
883 mtu = IPV6_MIN_MTU;
26892058
YK
884 if (skb->dst)
885 skb->dst->ops->update_pmtu(skb->dst, mtu);
1da177e4 886 if (skb->len > mtu) {
61ec2aec
YK
887 *pmtu = mtu;
888 err = -EMSGSIZE;
1da177e4
LT
889 goto tx_err_dst_release;
890 }
891
892 /*
893 * Okay, now see if we can stuff it in the buffer as-is.
894 */
895 max_headroom += LL_RESERVED_SPACE(tdev);
1ab1457c 896
cfbba49d
PM
897 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
898 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1da177e4 899 struct sk_buff *new_skb;
1ab1457c 900
1da177e4
LT
901 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
902 goto tx_err_dst_release;
903
904 if (skb->sk)
905 skb_set_owner_w(new_skb, skb->sk);
906 kfree_skb(skb);
907 skb = new_skb;
908 }
909 dst_release(skb->dst);
910 skb->dst = dst_clone(dst);
911
b0e380b1 912 skb->transport_header = skb->network_header;
1da177e4 913
61ec2aec 914 proto = fl->proto;
6fb32dde
VN
915 if (encap_limit >= 0) {
916 init_tel_txopt(&opt, encap_limit);
917 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
918 }
e2d1bca7
ACM
919 skb_push(skb, sizeof(struct ipv6hdr));
920 skb_reset_network_header(skb);
0660e03f 921 ipv6h = ipv6_hdr(skb);
61ec2aec 922 *(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000);
1da177e4
LT
923 dsfield = INET_ECN_encapsulate(0, dsfield);
924 ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
1da177e4
LT
925 ipv6h->hop_limit = t->parms.hop_limit;
926 ipv6h->nexthdr = proto;
61ec2aec
YK
927 ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src);
928 ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst);
1da177e4
LT
929 nf_reset(skb);
930 pkt_len = skb->len;
ef76bc23 931 err = ip6_local_out(skb);
1da177e4 932
b9df3cb8 933 if (net_xmit_eval(err) == 0) {
1da177e4
LT
934 stats->tx_bytes += pkt_len;
935 stats->tx_packets++;
936 } else {
937 stats->tx_errors++;
938 stats->tx_aborted_errors++;
939 }
940 ip6_tnl_dst_store(t, dst);
1da177e4
LT
941 return 0;
942tx_err_link_failure:
943 stats->tx_carrier_errors++;
944 dst_link_failure(skb);
945tx_err_dst_release:
946 dst_release(dst);
61ec2aec
YK
947 return err;
948}
949
c4d3efaf
YK
950static inline int
951ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
952{
953 struct ip6_tnl *t = netdev_priv(dev);
eddc9ec5 954 struct iphdr *iph = ip_hdr(skb);
c4d3efaf
YK
955 int encap_limit = -1;
956 struct flowi fl;
957 __u8 dsfield;
958 __u32 mtu;
959 int err;
960
502b0935
YK
961 if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
962 !ip6_tnl_xmit_ctl(t))
c4d3efaf
YK
963 return -1;
964
965 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
966 encap_limit = t->parms.encap_limit;
967
968 memcpy(&fl, &t->fl, sizeof (fl));
969 fl.proto = IPPROTO_IPIP;
970
971 dsfield = ipv4_get_dsfield(iph);
972
973 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
b77f2fa6
AV
974 fl.fl6_flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
975 & IPV6_TCLASS_MASK;
c4d3efaf
YK
976
977 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
978 if (err != 0) {
979 /* XXX: send ICMP error even if DF is not set. */
980 if (err == -EMSGSIZE)
981 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
982 htonl(mtu));
983 return -1;
984 }
985
986 return 0;
987}
988
61ec2aec
YK
989static inline int
990ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
991{
992 struct ip6_tnl *t = netdev_priv(dev);
0660e03f 993 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
61ec2aec
YK
994 int encap_limit = -1;
995 __u16 offset;
996 struct flowi fl;
997 __u8 dsfield;
998 __u32 mtu;
999 int err;
1000
502b0935
YK
1001 if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
1002 !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
61ec2aec
YK
1003 return -1;
1004
d56f90a7
ACM
1005 offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb));
1006 if (offset > 0) {
61ec2aec 1007 struct ipv6_tlv_tnl_enc_lim *tel;
d56f90a7 1008 tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
61ec2aec
YK
1009 if (tel->encap_limit == 0) {
1010 icmpv6_send(skb, ICMPV6_PARAMPROB,
1011 ICMPV6_HDR_FIELD, offset + 2, skb->dev);
1012 return -1;
1013 }
1014 encap_limit = tel->encap_limit - 1;
1015 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1016 encap_limit = t->parms.encap_limit;
1017
1018 memcpy(&fl, &t->fl, sizeof (fl));
1019 fl.proto = IPPROTO_IPV6;
1020
1021 dsfield = ipv6_get_dsfield(ipv6h);
1022 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
1023 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1024 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
1025 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1026
1027 err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
1028 if (err != 0) {
1029 if (err == -EMSGSIZE)
1030 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev);
1031 return -1;
1032 }
1033
1034 return 0;
1035}
1036
1037static int
1038ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1039{
1040 struct ip6_tnl *t = netdev_priv(dev);
1041 struct net_device_stats *stats = &t->stat;
1042 int ret;
1043
1044 if (t->recursion++) {
1045 t->stat.collisions++;
1046 goto tx_err;
1047 }
1048
1049 switch (skb->protocol) {
c4d3efaf
YK
1050 case __constant_htons(ETH_P_IP):
1051 ret = ip4ip6_tnl_xmit(skb, dev);
1052 break;
61ec2aec
YK
1053 case __constant_htons(ETH_P_IPV6):
1054 ret = ip6ip6_tnl_xmit(skb, dev);
1055 break;
1056 default:
1057 goto tx_err;
1058 }
1059
1060 if (ret < 0)
1061 goto tx_err;
1062
1063 t->recursion--;
1064 return 0;
1065
1da177e4
LT
1066tx_err:
1067 stats->tx_errors++;
1068 stats->tx_dropped++;
1069 kfree_skb(skb);
1070 t->recursion--;
1071 return 0;
1072}
1073
1074static void ip6_tnl_set_cap(struct ip6_tnl *t)
1075{
1076 struct ip6_tnl_parm *p = &t->parms;
09c6bbf0
VN
1077 int ltype = ipv6_addr_type(&p->laddr);
1078 int rtype = ipv6_addr_type(&p->raddr);
1da177e4
LT
1079
1080 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);
1081
09c6bbf0
VN
1082 if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1083 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
1084 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
305d4b3c 1085 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
09c6bbf0
VN
1086 if (ltype&IPV6_ADDR_UNICAST)
1087 p->flags |= IP6_TNL_F_CAP_XMIT;
1088 if (rtype&IPV6_ADDR_UNICAST)
1089 p->flags |= IP6_TNL_F_CAP_RCV;
1da177e4
LT
1090 }
1091}
1092
3144581c 1093static void ip6_tnl_link_config(struct ip6_tnl *t)
1da177e4
LT
1094{
1095 struct net_device *dev = t->dev;
1096 struct ip6_tnl_parm *p = &t->parms;
1097 struct flowi *fl = &t->fl;
1098
1099 memcpy(&dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1100 memcpy(&dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1101
1102 /* Set up flowi template */
1103 ipv6_addr_copy(&fl->fl6_src, &p->laddr);
1104 ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
1105 fl->oif = p->link;
1106 fl->fl6_flowlabel = 0;
1107
1108 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
1109 fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1110 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
1111 fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1112
1113 ip6_tnl_set_cap(t);
1114
1115 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1116 dev->flags |= IFF_POINTOPOINT;
1117 else
1118 dev->flags &= ~IFF_POINTOPOINT;
1119
1120 dev->iflink = p->link;
1121
1122 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN
1123 int strict = (ipv6_addr_type(&p->raddr) &
1124 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1125
606a2b48 1126 struct rt6_info *rt = rt6_lookup(&init_net, &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
1149 * @active: != 0 if tunnel is ready for use
1150 *
1151 * Description:
3144581c 1152 * ip6_tnl_change() updates the tunnel parameters
1da177e4
LT
1153 **/
1154
1155static int
3144581c 1156ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
1da177e4
LT
1157{
1158 ipv6_addr_copy(&t->parms.laddr, &p->laddr);
1159 ipv6_addr_copy(&t->parms.raddr, &p->raddr);
1160 t->parms.flags = p->flags;
1161 t->parms.hop_limit = p->hop_limit;
1162 t->parms.encap_limit = p->encap_limit;
1163 t->parms.flowinfo = p->flowinfo;
8181b8c1 1164 t->parms.link = p->link;
502b0935 1165 t->parms.proto = p->proto;
0c088890 1166 ip6_tnl_dst_reset(t);
3144581c 1167 ip6_tnl_link_config(t);
1da177e4
LT
1168 return 0;
1169}
1170
1171/**
3144581c 1172 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1da177e4
LT
1173 * @dev: virtual device associated with tunnel
1174 * @ifr: parameters passed from userspace
1175 * @cmd: command to be performed
1176 *
1177 * Description:
3144581c 1178 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1ab1457c 1179 * from userspace.
1da177e4
LT
1180 *
1181 * The possible commands are the following:
1182 * %SIOCGETTUNNEL: get tunnel parameters for device
1183 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1184 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1185 * %SIOCDELTUNNEL: delete tunnel
1186 *
1ab1457c 1187 * The fallback device "ip6tnl0", created during module
1da177e4
LT
1188 * initialization, can be used for creating other tunnel devices.
1189 *
1190 * Return:
1191 * 0 on success,
1192 * %-EFAULT if unable to copy data to or from userspace,
1193 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1194 * %-EINVAL if passed tunnel parameters are invalid,
1195 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1196 * %-ENODEV if attempting to change or delete a nonexisting device
1197 **/
1198
1199static int
3144581c 1200ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1da177e4
LT
1201{
1202 int err = 0;
1da177e4
LT
1203 struct ip6_tnl_parm p;
1204 struct ip6_tnl *t = NULL;
2dd02c89
PE
1205 struct net *net = dev_net(dev);
1206 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4
LT
1207
1208 switch (cmd) {
1209 case SIOCGETTUNNEL:
15820e12 1210 if (dev == ip6n->fb_tnl_dev) {
567131a7 1211 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1da177e4
LT
1212 err = -EFAULT;
1213 break;
1214 }
2dd02c89 1215 t = ip6_tnl_locate(net, &p, 0);
567131a7
VN
1216 }
1217 if (t == NULL)
2941a486 1218 t = netdev_priv(dev);
1da177e4
LT
1219 memcpy(&p, &t->parms, sizeof (p));
1220 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
1221 err = -EFAULT;
1222 }
1223 break;
1224 case SIOCADDTUNNEL:
1225 case SIOCCHGTUNNEL:
1226 err = -EPERM;
1da177e4
LT
1227 if (!capable(CAP_NET_ADMIN))
1228 break;
567131a7
VN
1229 err = -EFAULT;
1230 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1231 break;
567131a7 1232 err = -EINVAL;
502b0935
YK
1233 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1234 p.proto != 0)
1da177e4 1235 break;
2dd02c89 1236 t = ip6_tnl_locate(net, &p, cmd == SIOCADDTUNNEL);
15820e12 1237 if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
567131a7
VN
1238 if (t != NULL) {
1239 if (t->dev != dev) {
1240 err = -EEXIST;
1241 break;
1242 }
1243 } else
1244 t = netdev_priv(dev);
1245
2dd02c89 1246 ip6_tnl_unlink(ip6n, t);
3144581c 1247 err = ip6_tnl_change(t, &p);
2dd02c89 1248 ip6_tnl_link(ip6n, t);
1da177e4
LT
1249 netdev_state_change(dev);
1250 }
567131a7 1251 if (t) {
1da177e4 1252 err = 0;
567131a7
VN
1253 if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
1254 err = -EFAULT;
1255
1256 } else
1257 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1da177e4
LT
1258 break;
1259 case SIOCDELTUNNEL:
1260 err = -EPERM;
1261 if (!capable(CAP_NET_ADMIN))
1262 break;
1263
15820e12 1264 if (dev == ip6n->fb_tnl_dev) {
567131a7
VN
1265 err = -EFAULT;
1266 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 1267 break;
567131a7 1268 err = -ENOENT;
2dd02c89 1269 if ((t = ip6_tnl_locate(net, &p, 0)) == NULL)
1da177e4 1270 break;
567131a7 1271 err = -EPERM;
15820e12 1272 if (t->dev == ip6n->fb_tnl_dev)
1da177e4 1273 break;
567131a7 1274 dev = t->dev;
1da177e4 1275 }
22f8cde5
SH
1276 err = 0;
1277 unregister_netdevice(dev);
1da177e4
LT
1278 break;
1279 default:
1280 err = -EINVAL;
1281 }
1282 return err;
1283}
1284
1285/**
3144581c 1286 * ip6_tnl_get_stats - return the stats for tunnel device
1da177e4
LT
1287 * @dev: virtual device associated with tunnel
1288 *
1289 * Return: stats for device
1290 **/
1291
1292static struct net_device_stats *
3144581c 1293ip6_tnl_get_stats(struct net_device *dev)
1da177e4 1294{
2941a486 1295 return &(((struct ip6_tnl *)netdev_priv(dev))->stat);
1da177e4
LT
1296}
1297
1298/**
3144581c 1299 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1da177e4
LT
1300 * @dev: virtual device associated with tunnel
1301 * @new_mtu: the new mtu
1302 *
1303 * Return:
1304 * 0 on success,
1305 * %-EINVAL if mtu too small
1306 **/
1307
1308static int
3144581c 1309ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1da177e4
LT
1310{
1311 if (new_mtu < IPV6_MIN_MTU) {
1312 return -EINVAL;
1313 }
1314 dev->mtu = new_mtu;
1315 return 0;
1316}
1317
1318/**
3144581c 1319 * ip6_tnl_dev_setup - setup virtual tunnel device
1da177e4
LT
1320 * @dev: virtual device associated with tunnel
1321 *
1322 * Description:
1323 * Initialize function pointers and device parameters
1324 **/
1325
3144581c 1326static void ip6_tnl_dev_setup(struct net_device *dev)
1da177e4 1327{
3144581c 1328 dev->uninit = ip6_tnl_dev_uninit;
1da177e4 1329 dev->destructor = free_netdev;
61ec2aec 1330 dev->hard_start_xmit = ip6_tnl_xmit;
3144581c
YK
1331 dev->get_stats = ip6_tnl_get_stats;
1332 dev->do_ioctl = ip6_tnl_ioctl;
1333 dev->change_mtu = ip6_tnl_change_mtu;
1da177e4
LT
1334
1335 dev->type = ARPHRD_TUNNEL6;
1336 dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1337 dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1338 dev->flags |= IFF_NOARP;
1339 dev->addr_len = sizeof(struct in6_addr);
1340}
1341
1342
1343/**
3144581c 1344 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1da177e4
LT
1345 * @dev: virtual device associated with tunnel
1346 **/
1347
1348static inline void
3144581c 1349ip6_tnl_dev_init_gen(struct net_device *dev)
1da177e4 1350{
2941a486 1351 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT
1352 t->dev = dev;
1353 strcpy(t->parms.name, dev->name);
1354}
1355
1356/**
3144581c 1357 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1da177e4
LT
1358 * @dev: virtual device associated with tunnel
1359 **/
1360
1361static int
3144581c 1362ip6_tnl_dev_init(struct net_device *dev)
1da177e4 1363{
2941a486 1364 struct ip6_tnl *t = netdev_priv(dev);
3144581c
YK
1365 ip6_tnl_dev_init_gen(dev);
1366 ip6_tnl_link_config(t);
1da177e4
LT
1367 return 0;
1368}
1369
1370/**
3144581c 1371 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1da177e4
LT
1372 * @dev: fallback device
1373 *
1374 * Return: 0
1375 **/
1376
1ab1457c 1377static int
3144581c 1378ip6_fb_tnl_dev_init(struct net_device *dev)
1da177e4 1379{
2941a486 1380 struct ip6_tnl *t = netdev_priv(dev);
3144581c 1381 ip6_tnl_dev_init_gen(dev);
502b0935 1382 t->parms.proto = IPPROTO_IPV6;
1da177e4
LT
1383 dev_hold(dev);
1384 tnls_wc[0] = t;
1385 return 0;
1386}
1387
c4d3efaf
YK
1388static struct xfrm6_tunnel ip4ip6_handler = {
1389 .handler = ip4ip6_rcv,
1390 .err_handler = ip4ip6_err,
1391 .priority = 1,
1392};
1393
1da177e4 1394static struct xfrm6_tunnel ip6ip6_handler = {
0303770d
PM
1395 .handler = ip6ip6_rcv,
1396 .err_handler = ip6ip6_err,
d2acc347 1397 .priority = 1,
1da177e4
LT
1398};
1399
13eeb8e9
PE
1400static int ip6_tnl_init_net(struct net *net)
1401{
1402 int err;
1403 struct ip6_tnl_net *ip6n;
1404
1405 err = -ENOMEM;
1406 ip6n = kmalloc(sizeof(struct ip6_tnl_net), GFP_KERNEL);
1407 if (ip6n == NULL)
1408 goto err_alloc;
1409
1410 err = net_assign_generic(net, ip6_tnl_net_id, ip6n);
1411 if (err < 0)
1412 goto err_assign;
1413
15820e12
PE
1414 err = -ENOMEM;
1415 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1416 ip6_tnl_dev_setup);
1417
1418 if (!ip6n->fb_tnl_dev)
1419 goto err_alloc_dev;
1420
1421 ip6n->fb_tnl_dev->init = ip6_fb_tnl_dev_init;
1422 dev_net_set(ip6n->fb_tnl_dev, net);
1423
1424 err = register_netdev(ip6n->fb_tnl_dev);
1425 if (err < 0)
1426 goto err_register;
13eeb8e9
PE
1427 return 0;
1428
15820e12
PE
1429err_register:
1430 free_netdev(ip6n->fb_tnl_dev);
1431err_alloc_dev:
1432 /* nothing */
13eeb8e9
PE
1433err_assign:
1434 kfree(ip6n);
1435err_alloc:
1436 return err;
1437}
1438
1439static void ip6_tnl_exit_net(struct net *net)
1440{
1441 struct ip6_tnl_net *ip6n;
1442
1443 ip6n = net_generic(net, ip6_tnl_net_id);
1444 kfree(ip6n);
1445}
1446
1447static struct pernet_operations ip6_tnl_net_ops = {
1448 .init = ip6_tnl_init_net,
1449 .exit = ip6_tnl_exit_net,
1450};
1451
1da177e4
LT
1452/**
1453 * ip6_tunnel_init - register protocol and reserve needed resources
1454 *
1455 * Return: 0 on success
1456 **/
1457
1458static int __init ip6_tunnel_init(void)
1459{
1460 int err;
1461
c4d3efaf 1462 if (xfrm6_tunnel_register(&ip4ip6_handler, AF_INET)) {
3144581c 1463 printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n");
c4d3efaf
YK
1464 err = -EAGAIN;
1465 goto out;
1466 }
1467
73d605d1 1468 if (xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6)) {
3144581c 1469 printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n");
c4d3efaf
YK
1470 err = -EAGAIN;
1471 goto unreg_ip4ip6;
1da177e4 1472 }
13eeb8e9
PE
1473
1474 err = register_pernet_gen_device(&ip6_tnl_net_id, &ip6_tnl_net_ops);
1475 if (err < 0)
1476 goto err_pernet;
1da177e4 1477 return 0;
13eeb8e9 1478err_pernet:
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);