projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[IPV6]: Don't allocate memory for Tunnel Encapsulation Limit Option
[deliverable/linux.git] / net / ipv6 / ip6_tunnel.c
CommitLineData
1da177e4
LT		 1/*
2 * IPv6 over IPv6 tunnel device
3 * Linux INET6 implementation
4 *
5 * Authors:
6 * Ville Nuorvala <vnuorval@tcs.hut.fi>
7 *
8 * $Id$
9 *
10 * Based on:
11 * linux/net/ipv6/sit.c
12 *
13 * RFC 2473
14 *
15 * This program is free software; you can redistribute it and/or
16 * modify it under the terms of the GNU General Public License
17 * as published by the Free Software Foundation; either version
18 * 2 of the License, or (at your option) any later version.
19 *
20 */
21
1da177e4 22#include <linux/module.h>
4fc268d2 23#include <linux/capability.h>
1da177e4
LT		 24#include <linux/errno.h>
25#include <linux/types.h>
26#include <linux/sockios.h>
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
44#include <net/ip.h>
45#include <net/ipv6.h>
1da177e4
LT		 46#include <net/ip6_route.h>
47#include <net/addrconf.h>
48#include <net/ip6_tunnel.h>
49#include <net/xfrm.h>
50#include <net/dsfield.h>
51#include <net/inet_ecn.h>
52
53MODULE_AUTHOR("Ville Nuorvala");
54MODULE_DESCRIPTION("IPv6-in-IPv6 tunnel");
55MODULE_LICENSE("GPL");
56
57#define IPV6_TLV_TEL_DST_SIZE 8
58
59#ifdef IP6_TNL_DEBUG
60#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __FUNCTION__)
61#else
62#define IP6_TNL_TRACE(x...) do {;} while(0)
63#endif
64
65#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
66
67#define HASH_SIZE 32
68
e69a4adc 69#define HASH(addr) ((__force u32)((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
1da177e4
LT		 70 (addr)->s6_addr32[2] ^ (addr)->s6_addr32[3]) & \
71 (HASH_SIZE - 1))
72
73static int ip6ip6_fb_tnl_dev_init(struct net_device *dev);
74static int ip6ip6_tnl_dev_init(struct net_device *dev);
75static void ip6ip6_tnl_dev_setup(struct net_device *dev);
76
77/* the IPv6 tunnel fallback device */
78static struct net_device *ip6ip6_fb_tnl_dev;
79
80
81/* lists for storing tunnels in use */
82static struct ip6_tnl *tnls_r_l[HASH_SIZE];
83static struct ip6_tnl *tnls_wc[1];
84static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l };
85
86/* lock for the tunnel lists */
87static DEFINE_RWLOCK(ip6ip6_lock);
88
89static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
90{
91 struct dst_entry *dst = t->dst_cache;
92
93 if (dst && dst->obsolete &&
94 dst->ops->check(dst, t->dst_cookie) == NULL) {
95 t->dst_cache = NULL;
96 dst_release(dst);
97 return NULL;
98 }
99
100 return dst;
101}
102
103static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
104{
105 dst_release(t->dst_cache);
106 t->dst_cache = NULL;
107}
108
109static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
110{
111 struct rt6_info *rt = (struct rt6_info *) dst;
112 t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
113 dst_release(t->dst_cache);
114 t->dst_cache = dst;
115}
116
117/**
118 * ip6ip6_tnl_lookup - fetch tunnel matching the end-point addresses
119 * @remote: the address of the tunnel exit-point
120 * @local: the address of the tunnel entry-point
121 *
122 * Return:
123 * tunnel matching given end-points if found,
124 * else fallback tunnel if its device is up,
125 * else %NULL
126 **/
127
128static struct ip6_tnl *
129ip6ip6_tnl_lookup(struct in6_addr *remote, struct in6_addr *local)
130{
131 unsigned h0 = HASH(remote);
132 unsigned h1 = HASH(local);
133 struct ip6_tnl *t;
134
135 for (t = tnls_r_l[h0 ^ h1]; t; t = t->next) {
136 if (ipv6_addr_equal(local, &t->parms.laddr) &&
137 ipv6_addr_equal(remote, &t->parms.raddr) &&
138 (t->dev->flags & IFF_UP))
139 return t;
140 }
141 if ((t = tnls_wc[0]) != NULL && (t->dev->flags & IFF_UP))
142 return t;
143
144 return NULL;
145}
146
147/**
148 * ip6ip6_bucket - get head of list matching given tunnel parameters
149 * @p: parameters containing tunnel end-points
150 *
151 * Description:
152 * ip6ip6_bucket() returns the head of the list matching the
153 * &struct in6_addr entries laddr and raddr in @p.
154 *
155 * Return: head of IPv6 tunnel list
156 **/
157
158static struct ip6_tnl **
159ip6ip6_bucket(struct ip6_tnl_parm *p)
160{
161 struct in6_addr *remote = &p->raddr;
162 struct in6_addr *local = &p->laddr;
163 unsigned h = 0;
164 int prio = 0;
165
166 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
167 prio = 1;
168 h = HASH(remote) ^ HASH(local);
169 }
170 return &tnls[prio][h];
171}
172
173/**
174 * ip6ip6_tnl_link - add tunnel to hash table
175 * @t: tunnel to be added
176 **/
177
178static void
179ip6ip6_tnl_link(struct ip6_tnl *t)
180{
181 struct ip6_tnl **tp = ip6ip6_bucket(&t->parms);
182
183 t->next = *tp;
184 write_lock_bh(&ip6ip6_lock);
185 *tp = t;
186 write_unlock_bh(&ip6ip6_lock);
187}
188
189/**
190 * ip6ip6_tnl_unlink - remove tunnel from hash table
191 * @t: tunnel to be removed
192 **/
193
194static void
195ip6ip6_tnl_unlink(struct ip6_tnl *t)
196{
197 struct ip6_tnl **tp;
198
199 for (tp = ip6ip6_bucket(&t->parms); *tp; tp = &(*tp)->next) {
200 if (t == *tp) {
201 write_lock_bh(&ip6ip6_lock);
202 *tp = t->next;
203 write_unlock_bh(&ip6ip6_lock);
204 break;
205 }
206 }
207}
208
209/**
210 * ip6_tnl_create() - create a new tunnel
211 * @p: tunnel parameters
212 * @pt: pointer to new tunnel
213 *
214 * Description:
215 * Create tunnel matching given parameters.
216 *
217 * Return:
567131a7 218 * created tunnel or NULL
1da177e4
LT		 219 **/
220
567131a7 221static struct ip6_tnl *ip6_tnl_create(struct ip6_tnl_parm *p)
1da177e4
LT		 222{
223 struct net_device *dev;
224 struct ip6_tnl *t;
225 char name[IFNAMSIZ];
226 int err;
227
228 if (p->name[0]) {
229 strlcpy(name, p->name, IFNAMSIZ);
230 } else {
231 int i;
232 for (i = 1; i < IP6_TNL_MAX; i++) {
233 sprintf(name, "ip6tnl%d", i);
234 if (__dev_get_by_name(name) == NULL)
235 break;
236 }
237 if (i == IP6_TNL_MAX)
567131a7 238 goto failed;
1da177e4
LT		 239 }
240 dev = alloc_netdev(sizeof (*t), name, ip6ip6_tnl_dev_setup);
241 if (dev == NULL)
567131a7 242 goto failed;
1da177e4 243
2941a486 244 t = netdev_priv(dev);
1da177e4
LT		 245 dev->init = ip6ip6_tnl_dev_init;
246 t->parms = *p;
247
248 if ((err = register_netdevice(dev)) < 0) {
249 free_netdev(dev);
567131a7 250 goto failed;
1da177e4
LT		 251 }
252 dev_hold(dev);
1da177e4 253 ip6ip6_tnl_link(t);
567131a7
VN		 254 return t;
255failed:
256 return NULL;
1da177e4
LT		 257}
258
259/**
260 * ip6ip6_tnl_locate - find or create tunnel matching given parameters
261 * @p: tunnel parameters
262 * @create: != 0 if allowed to create new tunnel if no match found
263 *
264 * Description:
265 * ip6ip6_tnl_locate() first tries to locate an existing tunnel
266 * based on @parms. If this is unsuccessful, but @create is set a new
267 * tunnel device is created and registered for use.
268 *
269 * Return:
567131a7 270 * matching tunnel or NULL
1da177e4
LT		 271 **/
272
567131a7 273static struct ip6_tnl *ip6ip6_tnl_locate(struct ip6_tnl_parm *p, int create)
1da177e4
LT		 274{
275 struct in6_addr *remote = &p->raddr;
276 struct in6_addr *local = &p->laddr;
277 struct ip6_tnl *t;
278
1da177e4
LT		 279 for (t = *ip6ip6_bucket(p); t; t = t->next) {
280 if (ipv6_addr_equal(local, &t->parms.laddr) &&
567131a7
VN		 281 ipv6_addr_equal(remote, &t->parms.raddr))
282 return t;
1da177e4
LT		 283 }
284 if (!create)
567131a7
VN		 285 return NULL;
286 return ip6_tnl_create(p);
1da177e4
LT		 287}
288
289/**
290 * ip6ip6_tnl_dev_uninit - tunnel device uninitializer
291 * @dev: the device to be destroyed
292 *
293 * Description:
294 * ip6ip6_tnl_dev_uninit() removes tunnel from its list
295 **/
296
297static void
298ip6ip6_tnl_dev_uninit(struct net_device *dev)
299{
2941a486 300 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT		 301
302 if (dev == ip6ip6_fb_tnl_dev) {
303 write_lock_bh(&ip6ip6_lock);
304 tnls_wc[0] = NULL;
305 write_unlock_bh(&ip6ip6_lock);
306 } else {
307 ip6ip6_tnl_unlink(t);
308 }
309 ip6_tnl_dst_reset(t);
310 dev_put(dev);
311}
312
313/**
314 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
315 * @skb: received socket buffer
316 *
317 * Return:
318 * 0 if none was found,
319 * else index to encapsulation limit
320 **/
321
322static __u16
323parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
324{
325 struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
326 __u8 nexthdr = ipv6h->nexthdr;
327 __u16 off = sizeof (*ipv6h);
328
329 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
330 __u16 optlen = 0;
331 struct ipv6_opt_hdr *hdr;
332 if (raw + off + sizeof (*hdr) > skb->data &&
333 !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
334 break;
335
336 hdr = (struct ipv6_opt_hdr *) (raw + off);
337 if (nexthdr == NEXTHDR_FRAGMENT) {
338 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
339 if (frag_hdr->frag_off)
340 break;
341 optlen = 8;
342 } else if (nexthdr == NEXTHDR_AUTH) {
343 optlen = (hdr->hdrlen + 2) << 2;
344 } else {
345 optlen = ipv6_optlen(hdr);
346 }
347 if (nexthdr == NEXTHDR_DEST) {
348 __u16 i = off + 2;
349 while (1) {
350 struct ipv6_tlv_tnl_enc_lim *tel;
351
352 /* No more room for encapsulation limit */
353 if (i + sizeof (*tel) > off + optlen)
354 break;
355
356 tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
357 /* return index of option if found and valid */
358 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
359 tel->length == 1)
360 return i;
361 /* else jump to next option */
362 if (tel->type)
363 i += tel->length + 2;
364 else
365 i++;
366 }
367 }
368 nexthdr = hdr->nexthdr;
369 off += optlen;
370 }
371 return 0;
372}
373
374/**
375 * ip6ip6_err - tunnel error handler
376 *
377 * Description:
378 * ip6ip6_err() should handle errors in the tunnel according
379 * to the specifications in RFC 2473.
380 **/
381
d2acc347 382static int
1da177e4 383ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
04ce6909 384 int type, int code, int offset, __be32 info)
1da177e4
LT		 385{
386 struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
387 struct ip6_tnl *t;
388 int rel_msg = 0;
389 int rel_type = ICMPV6_DEST_UNREACH;
390 int rel_code = ICMPV6_ADDR_UNREACH;
391 __u32 rel_info = 0;
392 __u16 len;
d2acc347 393 int err = -ENOENT;
1da177e4
LT		 394
395 /* If the packet doesn't contain the original IPv6 header we are
396 in trouble since we might need the source address for further
397 processing of the error. */
398
399 read_lock(&ip6ip6_lock);
400 if ((t = ip6ip6_tnl_lookup(&ipv6h->daddr, &ipv6h->saddr)) == NULL)
401 goto out;
402
d2acc347
HX		 403 err = 0;
404
1da177e4
LT		 405 switch (type) {
406 __u32 teli;
407 struct ipv6_tlv_tnl_enc_lim *tel;
408 __u32 mtu;
409 case ICMPV6_DEST_UNREACH:
410 if (net_ratelimit())
411 printk(KERN_WARNING
412 "%s: Path to destination invalid "
413 "or inactive!\n", t->parms.name);
414 rel_msg = 1;
415 break;
416 case ICMPV6_TIME_EXCEED:
417 if (code == ICMPV6_EXC_HOPLIMIT) {
418 if (net_ratelimit())
419 printk(KERN_WARNING
420 "%s: Too small hop limit or "
421 "routing loop in tunnel!\n",
422 t->parms.name);
423 rel_msg = 1;
424 }
425 break;
426 case ICMPV6_PARAMPROB:
427 /* ignore if parameter problem not caused by a tunnel
428 encapsulation limit sub-option */
429 if (code != ICMPV6_HDR_FIELD) {
430 break;
431 }
432 teli = parse_tlv_tnl_enc_lim(skb, skb->data);
433
434 if (teli && teli == ntohl(info) - 2) {
435 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
436 if (tel->encap_limit == 0) {
437 if (net_ratelimit())
438 printk(KERN_WARNING
439 "%s: Too small encapsulation "
440 "limit or routing loop in "
441 "tunnel!\n", t->parms.name);
442 rel_msg = 1;
443 }
444 }
445 break;
446 case ICMPV6_PKT_TOOBIG:
447 mtu = ntohl(info) - offset;
448 if (mtu < IPV6_MIN_MTU)
449 mtu = IPV6_MIN_MTU;
450 t->dev->mtu = mtu;
451
cc6cdac0 452 if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
1da177e4
LT		 453 rel_type = ICMPV6_PKT_TOOBIG;
454 rel_code = 0;
455 rel_info = mtu;
456 rel_msg = 1;
457 }
458 break;
459 }
460 if (rel_msg && pskb_may_pull(skb, offset + sizeof (*ipv6h))) {
461 struct rt6_info *rt;
462 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 463
1da177e4
LT		 464 if (!skb2)
465 goto out;
466
467 dst_release(skb2->dst);
468 skb2->dst = NULL;
469 skb_pull(skb2, offset);
470 skb2->nh.raw = skb2->data;
471
472 /* Try to guess incoming interface */
473 rt = rt6_lookup(&skb2->nh.ipv6h->saddr, NULL, 0, 0);
474
475 if (rt && rt->rt6i_dev)
476 skb2->dev = rt->rt6i_dev;
477
478 icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev);
479
480 if (rt)
481 dst_release(&rt->u.dst);
482
483 kfree_skb(skb2);
484 }
485out:
486 read_unlock(&ip6ip6_lock);
d2acc347 487 return err;
1da177e4
LT		 488}
489
490static inline void ip6ip6_ecn_decapsulate(struct ipv6hdr *outer_iph,
491 struct sk_buff *skb)
492{
493 struct ipv6hdr *inner_iph = skb->nh.ipv6h;
494
495 if (INET_ECN_is_ce(ipv6_get_dsfield(outer_iph)))
496 IP6_ECN_set_ce(inner_iph);
497}
09c6bbf0
VN		 498static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
499{
500 struct ip6_tnl_parm *p = &t->parms;
501 int ret = 0;
502
503 if (p->flags & IP6_TNL_F_CAP_RCV) {
504 struct net_device *ldev = NULL;
505
506 if (p->link)
507 ldev = dev_get_by_index(p->link);
508
509 if ((ipv6_addr_is_multicast(&p->laddr) ||
510 likely(ipv6_chk_addr(&p->laddr, ldev, 0))) &&
511 likely(!ipv6_chk_addr(&p->raddr, NULL, 0)))
512 ret = 1;
513
514 if (ldev)
515 dev_put(ldev);
516 }
517 return ret;
518}
1da177e4
LT		 519
520/**
521 * ip6ip6_rcv - decapsulate IPv6 packet and retransmit it locally
522 * @skb: received socket buffer
523 *
524 * Return: 0
525 **/
526
527static int
d2acc347 528ip6ip6_rcv(struct sk_buff *skb)
1da177e4 529{
1da177e4
LT		 530 struct ipv6hdr *ipv6h;
531 struct ip6_tnl *t;
532
1da177e4
LT		 533 ipv6h = skb->nh.ipv6h;
534
535 read_lock(&ip6ip6_lock);
536
537 if ((t = ip6ip6_tnl_lookup(&ipv6h->saddr, &ipv6h->daddr)) != NULL) {
538 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
9f0ede52 539 read_unlock(&ip6ip6_lock);
50fba2aa 540 goto discard;
1da177e4
LT		 541 }
542
09c6bbf0 543 if (!ip6_tnl_rcv_ctl(t)) {
1da177e4
LT		 544 t->stat.rx_dropped++;
545 read_unlock(&ip6ip6_lock);
546 goto discard;
547 }
548 secpath_reset(skb);
549 skb->mac.raw = skb->nh.raw;
550 skb->nh.raw = skb->data;
551 skb->protocol = htons(ETH_P_IPV6);
552 skb->pkt_type = PACKET_HOST;
553 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
554 skb->dev = t->dev;
555 dst_release(skb->dst);
556 skb->dst = NULL;
53ab61c6 557 nf_reset(skb);
1da177e4
LT		 558 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
559 ipv6_copy_dscp(ipv6h, skb->nh.ipv6h);
560 ip6ip6_ecn_decapsulate(ipv6h, skb);
561 t->stat.rx_packets++;
562 t->stat.rx_bytes += skb->len;
563 netif_rx(skb);
564 read_unlock(&ip6ip6_lock);
565 return 0;
566 }
567 read_unlock(&ip6ip6_lock);
1da177e4 568 return 1;
50fba2aa
HX		 569
570discard:
571 kfree_skb(skb);
572 return 0;
1da177e4
LT		 573}
574
6fb32dde
VN		 575struct ipv6_tel_txoption {
576 struct ipv6_txoptions ops;
577 __u8 dst_opt[8];
578};
1da177e4 579
6fb32dde
VN		 580static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
581{
582 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 583
6fb32dde
VN		 584 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
585 opt->dst_opt[3] = 1;
586 opt->dst_opt[4] = encap_limit;
587 opt->dst_opt[5] = IPV6_TLV_PADN;
588 opt->dst_opt[6] = 1;
1da177e4 589
6fb32dde
VN		 590 opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
591 opt->ops.opt_nflen = 8;
1da177e4
LT		 592}
593
594/**
595 * ip6ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
596 * @t: the outgoing tunnel device
597 * @hdr: IPv6 header from the incoming packet
598 *
599 * Description:
600 * Avoid trivial tunneling loop by checking that tunnel exit-point
601 * doesn't match source of incoming packet.
602 *
603 * Return:
604 * 1 if conflict,
605 * 0 else
606 **/
607
608static inline int
609ip6ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
610{
611 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
612}
613
09c6bbf0
VN		 614static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
615{
616 struct ip6_tnl_parm *p = &t->parms;
617 int ret = 0;
618
619 if (p->flags & IP6_TNL_F_CAP_XMIT) {
620 struct net_device *ldev = NULL;
621
622 if (p->link)
623 ldev = dev_get_by_index(p->link);
624
625 if (unlikely(!ipv6_chk_addr(&p->laddr, ldev, 0)))
626 printk(KERN_WARNING
627 "%s xmit: Local address not yet configured!\n",
628 p->name);
629 else if (!ipv6_addr_is_multicast(&p->raddr) &&
630 unlikely(ipv6_chk_addr(&p->raddr, NULL, 0)))
631 printk(KERN_WARNING
632 "%s xmit: Routing loop! "
633 "Remote address found on this node!\n",
634 p->name);
635 else
636 ret = 1;
637 if (ldev)
638 dev_put(ldev);
639 }
640 return ret;
641}
1da177e4
LT		 642/**
643 * ip6ip6_tnl_xmit - encapsulate packet and send
644 * @skb: the outgoing socket buffer
645 * @dev: the outgoing tunnel device
646 *
647 * Description:
648 * Build new header and do some sanity checks on the packet before sending
649 * it.
650 *
651 * Return:
652 * 0
653 **/
654
655static int
656ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
657{
2941a486 658 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT		 659 struct net_device_stats *stats = &t->stat;
660 struct ipv6hdr *ipv6h = skb->nh.ipv6h;
1da177e4 661 int encap_limit = -1;
6fb32dde 662 struct ipv6_tel_txoption opt;
1da177e4
LT		 663 __u16 offset;
664 struct flowi fl;
665 struct dst_entry *dst;
666 struct net_device *tdev;
667 int mtu;
668 int max_headroom = sizeof(struct ipv6hdr);
669 u8 proto;
670 int err;
671 int pkt_len;
672 int dsfield;
673
674 if (t->recursion++) {
675 stats->collisions++;
676 goto tx_err;
677 }
678 if (skb->protocol != htons(ETH_P_IPV6) ||
09c6bbf0 679 !ip6_tnl_xmit_ctl(t) || ip6ip6_tnl_addr_conflict(t, ipv6h))
1da177e4 680 goto tx_err;
09c6bbf0 681
1da177e4
LT		 682 if ((offset = parse_tlv_tnl_enc_lim(skb, skb->nh.raw)) > 0) {
683 struct ipv6_tlv_tnl_enc_lim *tel;
684 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->nh.raw[offset];
685 if (tel->encap_limit == 0) {
686 icmpv6_send(skb, ICMPV6_PARAMPROB,
687 ICMPV6_HDR_FIELD, offset + 2, skb->dev);
688 goto tx_err;
689 }
690 encap_limit = tel->encap_limit - 1;
6fb32dde 691 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1da177e4 692 encap_limit = t->parms.encap_limit;
6fb32dde 693
1da177e4
LT		 694 memcpy(&fl, &t->fl, sizeof (fl));
695 proto = fl.proto;
696
697 dsfield = ipv6_get_dsfield(ipv6h);
698 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
90bcaf7b 699 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
1da177e4 700 if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
90bcaf7b 701 fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);
1da177e4 702
1da177e4
LT		 703 if ((dst = ip6_tnl_dst_check(t)) != NULL)
704 dst_hold(dst);
a57ebc90 705 else {
1da177e4
LT		 706 dst = ip6_route_output(NULL, &fl);
707
a57ebc90
PM		 708 if (dst->error || xfrm_lookup(&dst, &fl, NULL, 0) < 0)
709 goto tx_err_link_failure;
710 }
1da177e4
LT		 711
712 tdev = dst->dev;
713
714 if (tdev == dev) {
715 stats->collisions++;
716 if (net_ratelimit())
717 printk(KERN_WARNING
718 "%s: Local routing loop detected!\n",
719 t->parms.name);
720 goto tx_err_dst_release;
721 }
722 mtu = dst_mtu(dst) - sizeof (*ipv6h);
6fb32dde 723 if (encap_limit >= 0) {
1da177e4
LT		 724 max_headroom += 8;
725 mtu -= 8;
726 }
727 if (mtu < IPV6_MIN_MTU)
728 mtu = IPV6_MIN_MTU;
729 if (skb->dst && mtu < dst_mtu(skb->dst)) {
730 struct rt6_info *rt = (struct rt6_info *) skb->dst;
731 rt->rt6i_flags |= RTF_MODIFIED;
732 rt->u.dst.metrics[RTAX_MTU-1] = mtu;
733 }
734 if (skb->len > mtu) {
735 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu, dev);
736 goto tx_err_dst_release;
737 }
738
739 /*
740 * Okay, now see if we can stuff it in the buffer as-is.
741 */
742 max_headroom += LL_RESERVED_SPACE(tdev);
743
744 if (skb_headroom(skb) < max_headroom ||
745 skb_cloned(skb) || skb_shared(skb)) {
746 struct sk_buff *new_skb;
747
748 if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
749 goto tx_err_dst_release;
750
751 if (skb->sk)
752 skb_set_owner_w(new_skb, skb->sk);
753 kfree_skb(skb);
754 skb = new_skb;
755 }
756 dst_release(skb->dst);
757 skb->dst = dst_clone(dst);
758
759 skb->h.raw = skb->nh.raw;
760
6fb32dde
VN		 761 if (encap_limit >= 0) {
762 init_tel_txopt(&opt, encap_limit);
763 ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
764 }
1da177e4
LT		 765 skb->nh.raw = skb_push(skb, sizeof(struct ipv6hdr));
766 ipv6h = skb->nh.ipv6h;
90bcaf7b 767 *(__be32*)ipv6h = fl.fl6_flowlabel | htonl(0x60000000);
1da177e4
LT		 768 dsfield = INET_ECN_encapsulate(0, dsfield);
769 ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
770 ipv6h->payload_len = htons(skb->len - sizeof(struct ipv6hdr));
771 ipv6h->hop_limit = t->parms.hop_limit;
772 ipv6h->nexthdr = proto;
773 ipv6_addr_copy(&ipv6h->saddr, &fl.fl6_src);
774 ipv6_addr_copy(&ipv6h->daddr, &fl.fl6_dst);
775 nf_reset(skb);
776 pkt_len = skb->len;
777 err = NF_HOOK(PF_INET6, NF_IP6_LOCAL_OUT, skb, NULL,
778 skb->dst->dev, dst_output);
779
b9df3cb8 780 if (net_xmit_eval(err) == 0) {
1da177e4
LT		 781 stats->tx_bytes += pkt_len;
782 stats->tx_packets++;
783 } else {
784 stats->tx_errors++;
785 stats->tx_aborted_errors++;
786 }
787 ip6_tnl_dst_store(t, dst);
1da177e4
LT		 788 t->recursion--;
789 return 0;
790tx_err_link_failure:
791 stats->tx_carrier_errors++;
792 dst_link_failure(skb);
793tx_err_dst_release:
794 dst_release(dst);
1da177e4
LT		 795tx_err:
796 stats->tx_errors++;
797 stats->tx_dropped++;
798 kfree_skb(skb);
799 t->recursion--;
800 return 0;
801}
802
803static void ip6_tnl_set_cap(struct ip6_tnl *t)
804{
805 struct ip6_tnl_parm *p = &t->parms;
09c6bbf0
VN		 806 int ltype = ipv6_addr_type(&p->laddr);
807 int rtype = ipv6_addr_type(&p->raddr);
1da177e4
LT		 808
809 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);
810
09c6bbf0
VN		 811 if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
812 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
813 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
305d4b3c 814 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
09c6bbf0
VN		 815 if (ltype&IPV6_ADDR_UNICAST)
816 p->flags |= IP6_TNL_F_CAP_XMIT;
817 if (rtype&IPV6_ADDR_UNICAST)
818 p->flags |= IP6_TNL_F_CAP_RCV;
1da177e4
LT		 819 }
820}
821
822static void ip6ip6_tnl_link_config(struct ip6_tnl *t)
823{
824 struct net_device *dev = t->dev;
825 struct ip6_tnl_parm *p = &t->parms;
826 struct flowi *fl = &t->fl;
827
828 memcpy(&dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
829 memcpy(&dev->broadcast, &p->raddr, sizeof(struct in6_addr));
830
831 /* Set up flowi template */
832 ipv6_addr_copy(&fl->fl6_src, &p->laddr);
833 ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
834 fl->oif = p->link;
835 fl->fl6_flowlabel = 0;
836
837 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
838 fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
839 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
840 fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
841
842 ip6_tnl_set_cap(t);
843
844 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
845 dev->flags |= IFF_POINTOPOINT;
846 else
847 dev->flags &= ~IFF_POINTOPOINT;
848
849 dev->iflink = p->link;
850
851 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN		 852 int strict = (ipv6_addr_type(&p->raddr) &
853 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
854
1da177e4 855 struct rt6_info *rt = rt6_lookup(&p->raddr, &p->laddr,
305d4b3c 856 p->link, strict);
1da177e4
LT		 857
858 if (rt == NULL)
859 return;
860
861 if (rt->rt6i_dev) {
862 dev->hard_header_len = rt->rt6i_dev->hard_header_len +
863 sizeof (struct ipv6hdr);
864
865 dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr);
866
867 if (dev->mtu < IPV6_MIN_MTU)
868 dev->mtu = IPV6_MIN_MTU;
869 }
870 dst_release(&rt->u.dst);
871 }
872}
873
874/**
875 * ip6ip6_tnl_change - update the tunnel parameters
876 * @t: tunnel to be changed
877 * @p: tunnel configuration parameters
878 * @active: != 0 if tunnel is ready for use
879 *
880 * Description:
881 * ip6ip6_tnl_change() updates the tunnel parameters
882 **/
883
884static int
885ip6ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
886{
887 ipv6_addr_copy(&t->parms.laddr, &p->laddr);
888 ipv6_addr_copy(&t->parms.raddr, &p->raddr);
889 t->parms.flags = p->flags;
890 t->parms.hop_limit = p->hop_limit;
891 t->parms.encap_limit = p->encap_limit;
892 t->parms.flowinfo = p->flowinfo;
8181b8c1 893 t->parms.link = p->link;
0c088890 894 ip6_tnl_dst_reset(t);
1da177e4
LT		 895 ip6ip6_tnl_link_config(t);
896 return 0;
897}
898
899/**
900 * ip6ip6_tnl_ioctl - configure ipv6 tunnels from userspace
901 * @dev: virtual device associated with tunnel
902 * @ifr: parameters passed from userspace
903 * @cmd: command to be performed
904 *
905 * Description:
906 * ip6ip6_tnl_ioctl() is used for managing IPv6 tunnels
907 * from userspace.
908 *
909 * The possible commands are the following:
910 * %SIOCGETTUNNEL: get tunnel parameters for device
911 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
912 * %SIOCCHGTUNNEL: change tunnel parameters to those given
913 * %SIOCDELTUNNEL: delete tunnel
914 *
915 * The fallback device "ip6tnl0", created during module
916 * initialization, can be used for creating other tunnel devices.
917 *
918 * Return:
919 * 0 on success,
920 * %-EFAULT if unable to copy data to or from userspace,
921 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
922 * %-EINVAL if passed tunnel parameters are invalid,
923 * %-EEXIST if changing a tunnel's parameters would cause a conflict
924 * %-ENODEV if attempting to change or delete a nonexisting device
925 **/
926
927static int
928ip6ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
929{
930 int err = 0;
1da177e4
LT		 931 struct ip6_tnl_parm p;
932 struct ip6_tnl *t = NULL;
933
934 switch (cmd) {
935 case SIOCGETTUNNEL:
936 if (dev == ip6ip6_fb_tnl_dev) {
567131a7 937 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
1da177e4
LT		 938 err = -EFAULT;
939 break;
940 }
567131a7
VN		 941 t = ip6ip6_tnl_locate(&p, 0);
942 }
943 if (t == NULL)
2941a486 944 t = netdev_priv(dev);
1da177e4
LT		 945 memcpy(&p, &t->parms, sizeof (p));
946 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
947 err = -EFAULT;
948 }
949 break;
950 case SIOCADDTUNNEL:
951 case SIOCCHGTUNNEL:
952 err = -EPERM;
1da177e4
LT		 953 if (!capable(CAP_NET_ADMIN))
954 break;
567131a7
VN		 955 err = -EFAULT;
956 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 957 break;
567131a7
VN		 958 err = -EINVAL;
959 if (p.proto != IPPROTO_IPV6)
1da177e4 960 break;
567131a7
VN		 961 t = ip6ip6_tnl_locate(&p, cmd == SIOCADDTUNNEL);
962 if (dev != ip6ip6_fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
963 if (t != NULL) {
964 if (t->dev != dev) {
965 err = -EEXIST;
966 break;
967 }
968 } else
969 t = netdev_priv(dev);
970
1da177e4
LT		 971 ip6ip6_tnl_unlink(t);
972 err = ip6ip6_tnl_change(t, &p);
973 ip6ip6_tnl_link(t);
974 netdev_state_change(dev);
975 }
567131a7 976 if (t) {
1da177e4 977 err = 0;
567131a7
VN		 978 if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
979 err = -EFAULT;
980
981 } else
982 err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
1da177e4
LT		 983 break;
984 case SIOCDELTUNNEL:
985 err = -EPERM;
986 if (!capable(CAP_NET_ADMIN))
987 break;
988
989 if (dev == ip6ip6_fb_tnl_dev) {
567131a7
VN		 990 err = -EFAULT;
991 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
1da177e4 992 break;
567131a7
VN		 993 err = -ENOENT;
994 if ((t = ip6ip6_tnl_locate(&p, 0)) == NULL)
1da177e4 995 break;
567131a7
VN		 996 err = -EPERM;
997 if (t->dev == ip6ip6_fb_tnl_dev)
1da177e4 998 break;
567131a7 999 dev = t->dev;
1da177e4 1000 }
567131a7 1001 err = unregister_netdevice(dev);
1da177e4
LT		 1002 break;
1003 default:
1004 err = -EINVAL;
1005 }
1006 return err;
1007}
1008
1009/**
1010 * ip6ip6_tnl_get_stats - return the stats for tunnel device
1011 * @dev: virtual device associated with tunnel
1012 *
1013 * Return: stats for device
1014 **/
1015
1016static struct net_device_stats *
1017ip6ip6_tnl_get_stats(struct net_device *dev)
1018{
2941a486 1019 return &(((struct ip6_tnl *)netdev_priv(dev))->stat);
1da177e4
LT		 1020}
1021
1022/**
1023 * ip6ip6_tnl_change_mtu - change mtu manually for tunnel device
1024 * @dev: virtual device associated with tunnel
1025 * @new_mtu: the new mtu
1026 *
1027 * Return:
1028 * 0 on success,
1029 * %-EINVAL if mtu too small
1030 **/
1031
1032static int
1033ip6ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1034{
1035 if (new_mtu < IPV6_MIN_MTU) {
1036 return -EINVAL;
1037 }
1038 dev->mtu = new_mtu;
1039 return 0;
1040}
1041
1042/**
1043 * ip6ip6_tnl_dev_setup - setup virtual tunnel device
1044 * @dev: virtual device associated with tunnel
1045 *
1046 * Description:
1047 * Initialize function pointers and device parameters
1048 **/
1049
1050static void ip6ip6_tnl_dev_setup(struct net_device *dev)
1051{
1052 SET_MODULE_OWNER(dev);
1053 dev->uninit = ip6ip6_tnl_dev_uninit;
1054 dev->destructor = free_netdev;
1055 dev->hard_start_xmit = ip6ip6_tnl_xmit;
1056 dev->get_stats = ip6ip6_tnl_get_stats;
1057 dev->do_ioctl = ip6ip6_tnl_ioctl;
1058 dev->change_mtu = ip6ip6_tnl_change_mtu;
1059
1060 dev->type = ARPHRD_TUNNEL6;
1061 dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
1062 dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
1063 dev->flags |= IFF_NOARP;
1064 dev->addr_len = sizeof(struct in6_addr);
1065}
1066
1067
1068/**
1069 * ip6ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1070 * @dev: virtual device associated with tunnel
1071 **/
1072
1073static inline void
1074ip6ip6_tnl_dev_init_gen(struct net_device *dev)
1075{
2941a486 1076 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT		 1077 t->fl.proto = IPPROTO_IPV6;
1078 t->dev = dev;
1079 strcpy(t->parms.name, dev->name);
1080}
1081
1082/**
1083 * ip6ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1084 * @dev: virtual device associated with tunnel
1085 **/
1086
1087static int
1088ip6ip6_tnl_dev_init(struct net_device *dev)
1089{
2941a486 1090 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT		 1091 ip6ip6_tnl_dev_init_gen(dev);
1092 ip6ip6_tnl_link_config(t);
1093 return 0;
1094}
1095
1096/**
1097 * ip6ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1098 * @dev: fallback device
1099 *
1100 * Return: 0
1101 **/
1102
1103static int
1104ip6ip6_fb_tnl_dev_init(struct net_device *dev)
1105{
2941a486 1106 struct ip6_tnl *t = netdev_priv(dev);
1da177e4
LT		 1107 ip6ip6_tnl_dev_init_gen(dev);
1108 dev_hold(dev);
1109 tnls_wc[0] = t;
1110 return 0;
1111}
1112
1113static struct xfrm6_tunnel ip6ip6_handler = {
0303770d
PM		 1114 .handler = ip6ip6_rcv,
1115 .err_handler = ip6ip6_err,
d2acc347 1116 .priority = 1,
1da177e4
LT		 1117};
1118
1119/**
1120 * ip6_tunnel_init - register protocol and reserve needed resources
1121 *
1122 * Return: 0 on success
1123 **/
1124
1125static int __init ip6_tunnel_init(void)
1126{
1127 int err;
1128
d2acc347 1129 if (xfrm6_tunnel_register(&ip6ip6_handler)) {
1da177e4
LT		 1130 printk(KERN_ERR "ip6ip6 init: can't register tunnel\n");
1131 return -EAGAIN;
1132 }
1133 ip6ip6_fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
1134 ip6ip6_tnl_dev_setup);
1135
1136 if (!ip6ip6_fb_tnl_dev) {
1137 err = -ENOMEM;
1138 goto fail;
1139 }
1140 ip6ip6_fb_tnl_dev->init = ip6ip6_fb_tnl_dev_init;
1141
1142 if ((err = register_netdev(ip6ip6_fb_tnl_dev))) {
1143 free_netdev(ip6ip6_fb_tnl_dev);
1144 goto fail;
1145 }
1146 return 0;
1147fail:
d2acc347 1148 xfrm6_tunnel_deregister(&ip6ip6_handler);
1da177e4
LT		 1149 return err;
1150}
1151
b3fdd9f1
YK		 1152static void __exit ip6ip6_destroy_tunnels(void)
1153{
1154 int h;
1155 struct ip6_tnl *t;
1156
1157 for (h = 0; h < HASH_SIZE; h++) {
1158 while ((t = tnls_r_l[h]) != NULL)
1159 unregister_netdevice(t->dev);
1160 }
1161
1162 t = tnls_wc[0];
1163 unregister_netdevice(t->dev);
1164}
1165
1da177e4
LT		 1166/**
1167 * ip6_tunnel_cleanup - free resources and unregister protocol
1168 **/
1169
1170static void __exit ip6_tunnel_cleanup(void)
1171{
d2acc347 1172 if (xfrm6_tunnel_deregister(&ip6ip6_handler))
1da177e4
LT		 1173 printk(KERN_INFO "ip6ip6 close: can't deregister tunnel\n");
1174
b3fdd9f1
YK		 1175 rtnl_lock();
1176 ip6ip6_destroy_tunnels();
1177 rtnl_unlock();
1da177e4
LT		 1178}
1179
1180module_init(ip6_tunnel_init);
1181module_exit(ip6_tunnel_cleanup);
Linux kernel - Deliverables
This page took 0.224763 seconds and 5 git commands to generate.