ip6_tunnel.c 34.9 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 *	IPv6 tunneling device
L
Linus Torvalds 已提交
3 4 5
 *	Linux INET6 implementation
 *
 *	Authors:
6
 *	Ville Nuorvala		<vnuorval@tcs.hut.fi>
7
 *	Yasuyuki Kozakai	<kozakai@linux-ipv6.org>
L
Linus Torvalds 已提交
8 9
 *
 *      Based on:
10
 *      linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
L
Linus Torvalds 已提交
11 12 13 14 15 16 17 18 19 20 21
 *
 *      RFC 2473
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 *
 */

#include <linux/module.h>
22
#include <linux/capability.h>
L
Linus Torvalds 已提交
23 24 25
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/sockios.h>
26
#include <linux/icmp.h>
L
Linus Torvalds 已提交
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
#include <linux/if.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/if_tunnel.h>
#include <linux/net.h>
#include <linux/in6.h>
#include <linux/netdevice.h>
#include <linux/if_arp.h>
#include <linux/icmpv6.h>
#include <linux/init.h>
#include <linux/route.h>
#include <linux/rtnetlink.h>
#include <linux/netfilter_ipv6.h>

#include <asm/uaccess.h>
#include <asm/atomic.h>

44
#include <net/icmp.h>
L
Linus Torvalds 已提交
45 46 47 48 49 50 51 52
#include <net/ip.h>
#include <net/ipv6.h>
#include <net/ip6_route.h>
#include <net/addrconf.h>
#include <net/ip6_tunnel.h>
#include <net/xfrm.h>
#include <net/dsfield.h>
#include <net/inet_ecn.h>
53 54
#include <net/net_namespace.h>
#include <net/netns/generic.h>
L
Linus Torvalds 已提交
55 56

MODULE_AUTHOR("Ville Nuorvala");
57
MODULE_DESCRIPTION("IPv6 tunneling device");
L
Linus Torvalds 已提交
58 59 60 61 62
MODULE_LICENSE("GPL");

#define IPV6_TLV_TEL_DST_SIZE 8

#ifdef IP6_TNL_DEBUG
63
#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__)
L
Linus Torvalds 已提交
64 65 66 67 68
#else
#define IP6_TNL_TRACE(x...) do {;} while(0)
#endif

#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
69
#define IPV6_TCLASS_SHIFT 20
L
Linus Torvalds 已提交
70 71 72

#define HASH_SIZE  32

A
Al Viro 已提交
73
#define HASH(addr) ((__force u32)((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
74 75
		     (addr)->s6_addr32[2] ^ (addr)->s6_addr32[3]) & \
		    (HASH_SIZE - 1))
L
Linus Torvalds 已提交
76

77
static void ip6_tnl_dev_init(struct net_device *dev);
78
static void ip6_tnl_dev_setup(struct net_device *dev);
L
Linus Torvalds 已提交
79

80
static int ip6_tnl_net_id __read_mostly;
81
struct ip6_tnl_net {
82 83
	/* the IPv6 tunnel fallback device */
	struct net_device *fb_tnl_dev;
84 85 86 87
	/* lists for storing tunnels in use */
	struct ip6_tnl *tnls_r_l[HASH_SIZE];
	struct ip6_tnl *tnls_wc[1];
	struct ip6_tnl **tnls[2];
88 89
};

90 91 92 93
/*
 * Locking : hash tables are protected by RCU and a spinlock
 */
static DEFINE_SPINLOCK(ip6_tnl_lock);
L
Linus Torvalds 已提交
94 95 96 97 98

static inline struct dst_entry *ip6_tnl_dst_check(struct ip6_tnl *t)
{
	struct dst_entry *dst = t->dst_cache;

99
	if (dst && dst->obsolete &&
L
Linus Torvalds 已提交
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
	    dst->ops->check(dst, t->dst_cookie) == NULL) {
		t->dst_cache = NULL;
		dst_release(dst);
		return NULL;
	}

	return dst;
}

static inline void ip6_tnl_dst_reset(struct ip6_tnl *t)
{
	dst_release(t->dst_cache);
	t->dst_cache = NULL;
}

static inline void ip6_tnl_dst_store(struct ip6_tnl *t, struct dst_entry *dst)
{
	struct rt6_info *rt = (struct rt6_info *) dst;
	t->dst_cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
	dst_release(t->dst_cache);
	t->dst_cache = dst;
}

/**
124
 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
125 126
 *   @remote: the address of the tunnel exit-point
 *   @local: the address of the tunnel entry-point
L
Linus Torvalds 已提交
127
 *
128
 * Return:
L
Linus Torvalds 已提交
129
 *   tunnel matching given end-points if found,
130
 *   else fallback tunnel if its device is up,
L
Linus Torvalds 已提交
131 132 133
 *   else %NULL
 **/

134 135 136
#define for_each_ip6_tunnel_rcu(start) \
	for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))

L
Linus Torvalds 已提交
137
static struct ip6_tnl *
138
ip6_tnl_lookup(struct net *net, struct in6_addr *remote, struct in6_addr *local)
L
Linus Torvalds 已提交
139 140 141 142
{
	unsigned h0 = HASH(remote);
	unsigned h1 = HASH(local);
	struct ip6_tnl *t;
143
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
L
Linus Torvalds 已提交
144

145
	for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[h0 ^ h1]) {
L
Linus Torvalds 已提交
146 147 148 149 150
		if (ipv6_addr_equal(local, &t->parms.laddr) &&
		    ipv6_addr_equal(remote, &t->parms.raddr) &&
		    (t->dev->flags & IFF_UP))
			return t;
	}
151 152
	t = rcu_dereference(ip6n->tnls_wc[0]);
	if (t && (t->dev->flags & IFF_UP))
L
Linus Torvalds 已提交
153 154 155 156 157 158
		return t;

	return NULL;
}

/**
159
 * ip6_tnl_bucket - get head of list matching given tunnel parameters
160
 *   @p: parameters containing tunnel end-points
L
Linus Torvalds 已提交
161 162
 *
 * Description:
163
 *   ip6_tnl_bucket() returns the head of the list matching the
L
Linus Torvalds 已提交
164 165
 *   &struct in6_addr entries laddr and raddr in @p.
 *
166
 * Return: head of IPv6 tunnel list
L
Linus Torvalds 已提交
167 168 169
 **/

static struct ip6_tnl **
170
ip6_tnl_bucket(struct ip6_tnl_net *ip6n, struct ip6_tnl_parm *p)
L
Linus Torvalds 已提交
171 172 173 174 175 176 177 178 179 180
{
	struct in6_addr *remote = &p->raddr;
	struct in6_addr *local = &p->laddr;
	unsigned h = 0;
	int prio = 0;

	if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
		prio = 1;
		h = HASH(remote) ^ HASH(local);
	}
181
	return &ip6n->tnls[prio][h];
L
Linus Torvalds 已提交
182 183 184
}

/**
185
 * ip6_tnl_link - add tunnel to hash table
L
Linus Torvalds 已提交
186 187 188 189
 *   @t: tunnel to be added
 **/

static void
190
ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
L
Linus Torvalds 已提交
191
{
192
	struct ip6_tnl **tp = ip6_tnl_bucket(ip6n, &t->parms);
L
Linus Torvalds 已提交
193

194
	spin_lock_bh(&ip6_tnl_lock);
L
Linus Torvalds 已提交
195
	t->next = *tp;
196 197
	rcu_assign_pointer(*tp, t);
	spin_unlock_bh(&ip6_tnl_lock);
L
Linus Torvalds 已提交
198 199 200
}

/**
201
 * ip6_tnl_unlink - remove tunnel from hash table
L
Linus Torvalds 已提交
202 203 204 205
 *   @t: tunnel to be removed
 **/

static void
206
ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
L
Linus Torvalds 已提交
207 208 209
{
	struct ip6_tnl **tp;

210
	for (tp = ip6_tnl_bucket(ip6n, &t->parms); *tp; tp = &(*tp)->next) {
L
Linus Torvalds 已提交
211
		if (t == *tp) {
212
			spin_lock_bh(&ip6_tnl_lock);
L
Linus Torvalds 已提交
213
			*tp = t->next;
214
			spin_unlock_bh(&ip6_tnl_lock);
L
Linus Torvalds 已提交
215 216 217 218 219 220 221 222 223 224 225 226
			break;
		}
	}
}

/**
 * ip6_tnl_create() - create a new tunnel
 *   @p: tunnel parameters
 *   @pt: pointer to new tunnel
 *
 * Description:
 *   Create tunnel matching given parameters.
227 228
 *
 * Return:
229
 *   created tunnel or NULL
L
Linus Torvalds 已提交
230 231
 **/

232
static struct ip6_tnl *ip6_tnl_create(struct net *net, struct ip6_tnl_parm *p)
L
Linus Torvalds 已提交
233 234 235 236 237
{
	struct net_device *dev;
	struct ip6_tnl *t;
	char name[IFNAMSIZ];
	int err;
238
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
L
Linus Torvalds 已提交
239

240
	if (p->name[0])
L
Linus Torvalds 已提交
241
		strlcpy(name, p->name, IFNAMSIZ);
242 243 244
	else
		sprintf(name, "ip6tnl%%d");

245
	dev = alloc_netdev(sizeof (*t), name, ip6_tnl_dev_setup);
L
Linus Torvalds 已提交
246
	if (dev == NULL)
247
		goto failed;
L
Linus Torvalds 已提交
248

249 250
	dev_net_set(dev, net);

251 252 253 254 255
	if (strchr(name, '%')) {
		if (dev_alloc_name(dev, name) < 0)
			goto failed_free;
	}

256
	t = netdev_priv(dev);
L
Linus Torvalds 已提交
257
	t->parms = *p;
258
	ip6_tnl_dev_init(dev);
L
Linus Torvalds 已提交
259

260 261 262
	if ((err = register_netdevice(dev)) < 0)
		goto failed_free;

L
Linus Torvalds 已提交
263
	dev_hold(dev);
264
	ip6_tnl_link(ip6n, t);
265
	return t;
266 267 268

failed_free:
	free_netdev(dev);
269 270
failed:
	return NULL;
L
Linus Torvalds 已提交
271 272 273
}

/**
274
 * ip6_tnl_locate - find or create tunnel matching given parameters
275
 *   @p: tunnel parameters
L
Linus Torvalds 已提交
276 277 278
 *   @create: != 0 if allowed to create new tunnel if no match found
 *
 * Description:
279
 *   ip6_tnl_locate() first tries to locate an existing tunnel
L
Linus Torvalds 已提交
280 281 282 283
 *   based on @parms. If this is unsuccessful, but @create is set a new
 *   tunnel device is created and registered for use.
 *
 * Return:
284
 *   matching tunnel or NULL
L
Linus Torvalds 已提交
285 286
 **/

287 288
static struct ip6_tnl *ip6_tnl_locate(struct net *net,
		struct ip6_tnl_parm *p, int create)
L
Linus Torvalds 已提交
289 290 291 292
{
	struct in6_addr *remote = &p->raddr;
	struct in6_addr *local = &p->laddr;
	struct ip6_tnl *t;
293
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
L
Linus Torvalds 已提交
294

295
	for (t = *ip6_tnl_bucket(ip6n, p); t; t = t->next) {
L
Linus Torvalds 已提交
296
		if (ipv6_addr_equal(local, &t->parms.laddr) &&
297 298
		    ipv6_addr_equal(remote, &t->parms.raddr))
			return t;
L
Linus Torvalds 已提交
299 300
	}
	if (!create)
301
		return NULL;
302
	return ip6_tnl_create(net, p);
L
Linus Torvalds 已提交
303 304 305
}

/**
306
 * ip6_tnl_dev_uninit - tunnel device uninitializer
L
Linus Torvalds 已提交
307
 *   @dev: the device to be destroyed
308
 *
L
Linus Torvalds 已提交
309
 * Description:
310
 *   ip6_tnl_dev_uninit() removes tunnel from its list
L
Linus Torvalds 已提交
311 312 313
 **/

static void
314
ip6_tnl_dev_uninit(struct net_device *dev)
L
Linus Torvalds 已提交
315
{
316
	struct ip6_tnl *t = netdev_priv(dev);
317 318
	struct net *net = dev_net(dev);
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
L
Linus Torvalds 已提交
319

320
	if (dev == ip6n->fb_tnl_dev) {
321
		spin_lock_bh(&ip6_tnl_lock);
322
		ip6n->tnls_wc[0] = NULL;
323
		spin_unlock_bh(&ip6_tnl_lock);
L
Linus Torvalds 已提交
324
	} else {
325
		ip6_tnl_unlink(ip6n, t);
L
Linus Torvalds 已提交
326 327 328 329 330 331 332 333 334
	}
	ip6_tnl_dst_reset(t);
	dev_put(dev);
}

/**
 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
 *   @skb: received socket buffer
 *
335 336
 * Return:
 *   0 if none was found,
L
Linus Torvalds 已提交
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
 *   else index to encapsulation limit
 **/

static __u16
parse_tlv_tnl_enc_lim(struct sk_buff *skb, __u8 * raw)
{
	struct ipv6hdr *ipv6h = (struct ipv6hdr *) raw;
	__u8 nexthdr = ipv6h->nexthdr;
	__u16 off = sizeof (*ipv6h);

	while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
		__u16 optlen = 0;
		struct ipv6_opt_hdr *hdr;
		if (raw + off + sizeof (*hdr) > skb->data &&
		    !pskb_may_pull(skb, raw - skb->data + off + sizeof (*hdr)))
			break;

		hdr = (struct ipv6_opt_hdr *) (raw + off);
		if (nexthdr == NEXTHDR_FRAGMENT) {
			struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
			if (frag_hdr->frag_off)
				break;
			optlen = 8;
		} else if (nexthdr == NEXTHDR_AUTH) {
			optlen = (hdr->hdrlen + 2) << 2;
		} else {
			optlen = ipv6_optlen(hdr);
		}
		if (nexthdr == NEXTHDR_DEST) {
			__u16 i = off + 2;
			while (1) {
				struct ipv6_tlv_tnl_enc_lim *tel;

				/* No more room for encapsulation limit */
				if (i + sizeof (*tel) > off + optlen)
					break;

				tel = (struct ipv6_tlv_tnl_enc_lim *) &raw[i];
				/* return index of option if found and valid */
				if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
				    tel->length == 1)
					return i;
				/* else jump to next option */
				if (tel->type)
					i += tel->length + 2;
				else
					i++;
			}
		}
		nexthdr = hdr->nexthdr;
		off += optlen;
	}
	return 0;
}

/**
393
 * ip6_tnl_err - tunnel error handler
L
Linus Torvalds 已提交
394 395
 *
 * Description:
396
 *   ip6_tnl_err() should handle errors in the tunnel according
L
Linus Torvalds 已提交
397 398 399
 *   to the specifications in RFC 2473.
 **/

H
Herbert Xu 已提交
400
static int
401
ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
402
	    u8 *type, u8 *code, int *msg, __u32 *info, int offset)
L
Linus Torvalds 已提交
403 404 405 406
{
	struct ipv6hdr *ipv6h = (struct ipv6hdr *) skb->data;
	struct ip6_tnl *t;
	int rel_msg = 0;
407 408
	u8 rel_type = ICMPV6_DEST_UNREACH;
	u8 rel_code = ICMPV6_ADDR_UNREACH;
L
Linus Torvalds 已提交
409 410
	__u32 rel_info = 0;
	__u16 len;
H
Herbert Xu 已提交
411
	int err = -ENOENT;
L
Linus Torvalds 已提交
412

413 414
	/* If the packet doesn't contain the original IPv6 header we are
	   in trouble since we might need the source address for further
L
Linus Torvalds 已提交
415 416
	   processing of the error. */

417
	rcu_read_lock();
418
	if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr,
419
					&ipv6h->saddr)) == NULL)
L
Linus Torvalds 已提交
420 421
		goto out;

422 423 424
	if (t->parms.proto != ipproto && t->parms.proto != 0)
		goto out;

H
Herbert Xu 已提交
425 426
	err = 0;

427
	switch (*type) {
L
Linus Torvalds 已提交
428 429 430 431 432 433 434 435 436 437 438
		__u32 teli;
		struct ipv6_tlv_tnl_enc_lim *tel;
		__u32 mtu;
	case ICMPV6_DEST_UNREACH:
		if (net_ratelimit())
			printk(KERN_WARNING
			       "%s: Path to destination invalid "
			       "or inactive!\n", t->parms.name);
		rel_msg = 1;
		break;
	case ICMPV6_TIME_EXCEED:
439
		if ((*code) == ICMPV6_EXC_HOPLIMIT) {
L
Linus Torvalds 已提交
440 441 442
			if (net_ratelimit())
				printk(KERN_WARNING
				       "%s: Too small hop limit or "
443
				       "routing loop in tunnel!\n",
L
Linus Torvalds 已提交
444 445 446 447 448
				       t->parms.name);
			rel_msg = 1;
		}
		break;
	case ICMPV6_PARAMPROB:
449
		teli = 0;
450
		if ((*code) == ICMPV6_HDR_FIELD)
451
			teli = parse_tlv_tnl_enc_lim(skb, skb->data);
L
Linus Torvalds 已提交
452

A
Al Viro 已提交
453
		if (teli && teli == *info - 2) {
L
Linus Torvalds 已提交
454 455 456 457 458 459 460 461 462
			tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
			if (tel->encap_limit == 0) {
				if (net_ratelimit())
					printk(KERN_WARNING
					       "%s: Too small encapsulation "
					       "limit or routing loop in "
					       "tunnel!\n", t->parms.name);
				rel_msg = 1;
			}
463 464 465 466
		} else if (net_ratelimit()) {
			printk(KERN_WARNING
			       "%s: Recipient unable to parse tunneled "
			       "packet!\n ", t->parms.name);
L
Linus Torvalds 已提交
467 468 469
		}
		break;
	case ICMPV6_PKT_TOOBIG:
A
Al Viro 已提交
470
		mtu = *info - offset;
L
Linus Torvalds 已提交
471 472 473 474
		if (mtu < IPV6_MIN_MTU)
			mtu = IPV6_MIN_MTU;
		t->dev->mtu = mtu;

A
Al Viro 已提交
475
		if ((len = sizeof (*ipv6h) + ntohs(ipv6h->payload_len)) > mtu) {
L
Linus Torvalds 已提交
476 477 478 479 480 481 482
			rel_type = ICMPV6_PKT_TOOBIG;
			rel_code = 0;
			rel_info = mtu;
			rel_msg = 1;
		}
		break;
	}
483 484 485 486 487 488 489

	*type = rel_type;
	*code = rel_code;
	*info = rel_info;
	*msg = rel_msg;

out:
490
	rcu_read_unlock();
491 492 493
	return err;
}

494 495
static int
ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
496
	   u8 type, u8 code, int offset, __be32 info)
497 498
{
	int rel_msg = 0;
499 500
	u8 rel_type = type;
	u8 rel_code = code;
A
Al Viro 已提交
501
	__u32 rel_info = ntohl(info);
502 503 504 505 506 507
	int err;
	struct sk_buff *skb2;
	struct iphdr *eiph;
	struct flowi fl;
	struct rtable *rt;

508 509
	err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
			  &rel_msg, &rel_info, offset);
510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
	if (err < 0)
		return err;

	if (rel_msg == 0)
		return 0;

	switch (rel_type) {
	case ICMPV6_DEST_UNREACH:
		if (rel_code != ICMPV6_ADDR_UNREACH)
			return 0;
		rel_type = ICMP_DEST_UNREACH;
		rel_code = ICMP_HOST_UNREACH;
		break;
	case ICMPV6_PKT_TOOBIG:
		if (rel_code != 0)
			return 0;
		rel_type = ICMP_DEST_UNREACH;
		rel_code = ICMP_FRAG_NEEDED;
		break;
	default:
		return 0;
	}

	if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
		return 0;

	skb2 = skb_clone(skb, GFP_ATOMIC);
	if (!skb2)
		return 0;

E
Eric Dumazet 已提交
540 541
	skb_dst_drop(skb2);

542
	skb_pull(skb2, offset);
543
	skb_reset_network_header(skb2);
544
	eiph = ip_hdr(skb2);
545 546 547 548 549 550

	/* Try to guess incoming interface */
	memset(&fl, 0, sizeof(fl));
	fl.fl4_dst = eiph->saddr;
	fl.fl4_tos = RT_TOS(eiph->tos);
	fl.proto = IPPROTO_IPIP;
551
	if (ip_route_output_key(dev_net(skb->dev), &rt, &fl))
552 553 554 555 556 557 558 559 560 561 562
		goto out;

	skb2->dev = rt->u.dst.dev;

	/* route "incoming" packet */
	if (rt->rt_flags & RTCF_LOCAL) {
		ip_rt_put(rt);
		rt = NULL;
		fl.fl4_dst = eiph->daddr;
		fl.fl4_src = eiph->saddr;
		fl.fl4_tos = eiph->tos;
563
		if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
564 565 566 567
		    rt->u.dst.dev->type != ARPHRD_TUNNEL) {
			ip_rt_put(rt);
			goto out;
		}
E
Eric Dumazet 已提交
568
		skb_dst_set(skb2, (struct dst_entry *)rt);
569 570 571 572
	} else {
		ip_rt_put(rt);
		if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
				   skb2->dev) ||
E
Eric Dumazet 已提交
573
		    skb_dst(skb2)->dev->type != ARPHRD_TUNNEL)
574 575 576 577 578
			goto out;
	}

	/* change mtu on this route */
	if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
E
Eric Dumazet 已提交
579
		if (rel_info > dst_mtu(skb_dst(skb2)))
580 581
			goto out;

E
Eric Dumazet 已提交
582
		skb_dst(skb2)->ops->update_pmtu(skb_dst(skb2), rel_info);
583 584
	}

A
Al Viro 已提交
585
	icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
586 587 588 589 590 591

out:
	kfree_skb(skb2);
	return 0;
}

592 593
static int
ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
594
	   u8 type, u8 code, int offset, __be32 info)
595 596
{
	int rel_msg = 0;
597 598
	u8 rel_type = type;
	u8 rel_code = code;
A
Al Viro 已提交
599
	__u32 rel_info = ntohl(info);
600 601
	int err;

602 603
	err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
			  &rel_msg, &rel_info, offset);
604 605 606 607
	if (err < 0)
		return err;

	if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
L
Linus Torvalds 已提交
608 609
		struct rt6_info *rt;
		struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
610

L
Linus Torvalds 已提交
611
		if (!skb2)
612
			return 0;
L
Linus Torvalds 已提交
613

E
Eric Dumazet 已提交
614
		skb_dst_drop(skb2);
L
Linus Torvalds 已提交
615
		skb_pull(skb2, offset);
616
		skb_reset_network_header(skb2);
L
Linus Torvalds 已提交
617 618

		/* Try to guess incoming interface */
619 620
		rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
				NULL, 0, 0);
L
Linus Torvalds 已提交
621 622 623 624

		if (rt && rt->rt6i_dev)
			skb2->dev = rt->rt6i_dev;

625
		icmpv6_send(skb2, rel_type, rel_code, rel_info);
L
Linus Torvalds 已提交
626 627 628 629 630 631

		if (rt)
			dst_release(&rt->u.dst);

		kfree_skb(skb2);
	}
632 633

	return 0;
L
Linus Torvalds 已提交
634 635
}

636 637 638 639 640 641 642
static void ip4ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
					struct ipv6hdr *ipv6h,
					struct sk_buff *skb)
{
	__u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;

	if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
643
		ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
644 645

	if (INET_ECN_is_ce(dsfield))
646
		IP_ECN_set_ce(ip_hdr(skb));
647 648
}

649 650 651
static void ip6ip6_dscp_ecn_decapsulate(struct ip6_tnl *t,
					struct ipv6hdr *ipv6h,
					struct sk_buff *skb)
L
Linus Torvalds 已提交
652
{
653
	if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
654
		ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
L
Linus Torvalds 已提交
655

656
	if (INET_ECN_is_ce(ipv6_get_dsfield(ipv6h)))
657
		IP6_ECN_set_ce(ipv6_hdr(skb));
L
Linus Torvalds 已提交
658
}
659

E
Eric Dumazet 已提交
660
/* called with rcu_read_lock() */
661 662 663 664
static inline int ip6_tnl_rcv_ctl(struct ip6_tnl *t)
{
	struct ip6_tnl_parm *p = &t->parms;
	int ret = 0;
665
	struct net *net = dev_net(t->dev);
666 667

	if (p->flags & IP6_TNL_F_CAP_RCV) {
668
		struct net_device *ldev = NULL;
669 670

		if (p->link)
E
Eric Dumazet 已提交
671
			ldev = dev_get_by_index_rcu(net, p->link);
672 673

		if ((ipv6_addr_is_multicast(&p->laddr) ||
674 675
		     likely(ipv6_chk_addr(net, &p->laddr, ldev, 0))) &&
		    likely(!ipv6_chk_addr(net, &p->raddr, NULL, 0)))
676 677 678 679 680
			ret = 1;

	}
	return ret;
}
L
Linus Torvalds 已提交
681 682

/**
683
 * ip6_tnl_rcv - decapsulate IPv6 packet and retransmit it locally
L
Linus Torvalds 已提交
684
 *   @skb: received socket buffer
685 686
 *   @protocol: ethernet protocol ID
 *   @dscp_ecn_decapsulate: the function to decapsulate DSCP code and ECN
L
Linus Torvalds 已提交
687 688 689 690
 *
 * Return: 0
 **/

691
static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
692
		       __u8 ipproto,
693 694 695
		       void (*dscp_ecn_decapsulate)(struct ip6_tnl *t,
						    struct ipv6hdr *ipv6h,
						    struct sk_buff *skb))
L
Linus Torvalds 已提交
696 697
{
	struct ip6_tnl *t;
698
	struct ipv6hdr *ipv6h = ipv6_hdr(skb);
L
Linus Torvalds 已提交
699

700
	rcu_read_lock();
L
Linus Torvalds 已提交
701

702
	if ((t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr,
703
					&ipv6h->daddr)) != NULL) {
704
		if (t->parms.proto != ipproto && t->parms.proto != 0) {
705
			rcu_read_unlock();
706 707 708
			goto discard;
		}

L
Linus Torvalds 已提交
709
		if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb)) {
710
			rcu_read_unlock();
711
			goto discard;
L
Linus Torvalds 已提交
712 713
		}

714
		if (!ip6_tnl_rcv_ctl(t)) {
715
			t->dev->stats.rx_dropped++;
716
			rcu_read_unlock();
L
Linus Torvalds 已提交
717 718 719
			goto discard;
		}
		secpath_reset(skb);
720
		skb->mac_header = skb->network_header;
721
		skb_reset_network_header(skb);
722
		skb->protocol = htons(protocol);
L
Linus Torvalds 已提交
723 724 725
		skb->pkt_type = PACKET_HOST;
		memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
		skb->dev = t->dev;
E
Eric Dumazet 已提交
726
		skb_dst_drop(skb);
727
		nf_reset(skb);
728 729 730

		dscp_ecn_decapsulate(t, ipv6h, skb);

731 732
		t->dev->stats.rx_packets++;
		t->dev->stats.rx_bytes += skb->len;
L
Linus Torvalds 已提交
733
		netif_rx(skb);
734
		rcu_read_unlock();
L
Linus Torvalds 已提交
735 736
		return 0;
	}
737
	rcu_read_unlock();
L
Linus Torvalds 已提交
738
	return 1;
739 740 741 742

discard:
	kfree_skb(skb);
	return 0;
L
Linus Torvalds 已提交
743 744
}

745 746
static int ip4ip6_rcv(struct sk_buff *skb)
{
747 748
	return ip6_tnl_rcv(skb, ETH_P_IP, IPPROTO_IPIP,
			   ip4ip6_dscp_ecn_decapsulate);
749 750
}

751 752
static int ip6ip6_rcv(struct sk_buff *skb)
{
753 754
	return ip6_tnl_rcv(skb, ETH_P_IPV6, IPPROTO_IPV6,
			   ip6ip6_dscp_ecn_decapsulate);
755 756
}

757 758 759 760
struct ipv6_tel_txoption {
	struct ipv6_txoptions ops;
	__u8 dst_opt[8];
};
L
Linus Torvalds 已提交
761

762 763 764
static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
{
	memset(opt, 0, sizeof(struct ipv6_tel_txoption));
L
Linus Torvalds 已提交
765

766 767 768 769 770
	opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
	opt->dst_opt[3] = 1;
	opt->dst_opt[4] = encap_limit;
	opt->dst_opt[5] = IPV6_TLV_PADN;
	opt->dst_opt[6] = 1;
L
Linus Torvalds 已提交
771

772 773
	opt->ops.dst0opt = (struct ipv6_opt_hdr *) opt->dst_opt;
	opt->ops.opt_nflen = 8;
L
Linus Torvalds 已提交
774 775 776
}

/**
777
 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
L
Linus Torvalds 已提交
778
 *   @t: the outgoing tunnel device
779
 *   @hdr: IPv6 header from the incoming packet
L
Linus Torvalds 已提交
780 781
 *
 * Description:
782
 *   Avoid trivial tunneling loop by checking that tunnel exit-point
L
Linus Torvalds 已提交
783 784
 *   doesn't match source of incoming packet.
 *
785
 * Return:
L
Linus Torvalds 已提交
786 787 788 789 790
 *   1 if conflict,
 *   0 else
 **/

static inline int
791
ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
L
Linus Torvalds 已提交
792 793 794 795
{
	return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
}

796 797 798 799
static inline int ip6_tnl_xmit_ctl(struct ip6_tnl *t)
{
	struct ip6_tnl_parm *p = &t->parms;
	int ret = 0;
800
	struct net *net = dev_net(t->dev);
801

802
	if (p->flags & IP6_TNL_F_CAP_XMIT) {
803 804
		struct net_device *ldev = NULL;

E
Eric Dumazet 已提交
805
		rcu_read_lock();
806
		if (p->link)
E
Eric Dumazet 已提交
807
			ldev = dev_get_by_index_rcu(net, p->link);
808

809
		if (unlikely(!ipv6_chk_addr(net, &p->laddr, ldev, 0)))
810 811 812 813
			printk(KERN_WARNING
			       "%s xmit: Local address not yet configured!\n",
			       p->name);
		else if (!ipv6_addr_is_multicast(&p->raddr) &&
814
			 unlikely(ipv6_chk_addr(net, &p->raddr, NULL, 0)))
815 816 817 818 819 820
			printk(KERN_WARNING
			       "%s xmit: Routing loop! "
			       "Remote address found on this node!\n",
			       p->name);
		else
			ret = 1;
E
Eric Dumazet 已提交
821
		rcu_read_unlock();
822 823 824
	}
	return ret;
}
L
Linus Torvalds 已提交
825
/**
826
 * ip6_tnl_xmit2 - encapsulate packet and send
L
Linus Torvalds 已提交
827
 *   @skb: the outgoing socket buffer
828
 *   @dev: the outgoing tunnel device
829 830 831 832
 *   @dsfield: dscp code for outer header
 *   @fl: flow of tunneled packet
 *   @encap_limit: encapsulation limit
 *   @pmtu: Path MTU is stored if packet is too big
L
Linus Torvalds 已提交
833 834 835 836 837
 *
 * Description:
 *   Build new header and do some sanity checks on the packet before sending
 *   it.
 *
838
 * Return:
839
 *   0 on success
840 841
 *   -1 fail
 *   %-EMSGSIZE message too big. return mtu in this case.
L
Linus Torvalds 已提交
842 843
 **/

844 845 846 847 848 849
static int ip6_tnl_xmit2(struct sk_buff *skb,
			 struct net_device *dev,
			 __u8 dsfield,
			 struct flowi *fl,
			 int encap_limit,
			 __u32 *pmtu)
L
Linus Torvalds 已提交
850
{
A
Alexey Dobriyan 已提交
851
	struct net *net = dev_net(dev);
852
	struct ip6_tnl *t = netdev_priv(dev);
853
	struct net_device_stats *stats = &t->dev->stats;
854
	struct ipv6hdr *ipv6h = ipv6_hdr(skb);
855
	struct ipv6_tel_txoption opt;
L
Linus Torvalds 已提交
856 857 858
	struct dst_entry *dst;
	struct net_device *tdev;
	int mtu;
859
	unsigned int max_headroom = sizeof(struct ipv6hdr);
L
Linus Torvalds 已提交
860
	u8 proto;
861
	int err = -1;
L
Linus Torvalds 已提交
862 863 864 865
	int pkt_len;

	if ((dst = ip6_tnl_dst_check(t)) != NULL)
		dst_hold(dst);
866
	else {
A
Alexey Dobriyan 已提交
867
		dst = ip6_route_output(net, NULL, fl);
L
Linus Torvalds 已提交
868

A
Alexey Dobriyan 已提交
869
		if (dst->error || xfrm_lookup(net, &dst, fl, NULL, 0) < 0)
870 871
			goto tx_err_link_failure;
	}
L
Linus Torvalds 已提交
872 873 874 875 876 877

	tdev = dst->dev;

	if (tdev == dev) {
		stats->collisions++;
		if (net_ratelimit())
878
			printk(KERN_WARNING
L
Linus Torvalds 已提交
879 880 881 882 883
			       "%s: Local routing loop detected!\n",
			       t->parms.name);
		goto tx_err_dst_release;
	}
	mtu = dst_mtu(dst) - sizeof (*ipv6h);
884
	if (encap_limit >= 0) {
L
Linus Torvalds 已提交
885 886 887 888 889
		max_headroom += 8;
		mtu -= 8;
	}
	if (mtu < IPV6_MIN_MTU)
		mtu = IPV6_MIN_MTU;
E
Eric Dumazet 已提交
890 891
	if (skb_dst(skb))
		skb_dst(skb)->ops->update_pmtu(skb_dst(skb), mtu);
L
Linus Torvalds 已提交
892
	if (skb->len > mtu) {
893 894
		*pmtu = mtu;
		err = -EMSGSIZE;
L
Linus Torvalds 已提交
895 896 897 898 899 900 901
		goto tx_err_dst_release;
	}

	/*
	 * Okay, now see if we can stuff it in the buffer as-is.
	 */
	max_headroom += LL_RESERVED_SPACE(tdev);
902

903 904
	if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
	    (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
L
Linus Torvalds 已提交
905
		struct sk_buff *new_skb;
906

L
Linus Torvalds 已提交
907 908 909 910 911 912 913 914
		if (!(new_skb = skb_realloc_headroom(skb, max_headroom)))
			goto tx_err_dst_release;

		if (skb->sk)
			skb_set_owner_w(new_skb, skb->sk);
		kfree_skb(skb);
		skb = new_skb;
	}
E
Eric Dumazet 已提交
915 916
	skb_dst_drop(skb);
	skb_dst_set(skb, dst_clone(dst));
L
Linus Torvalds 已提交
917

918
	skb->transport_header = skb->network_header;
L
Linus Torvalds 已提交
919

920
	proto = fl->proto;
921 922 923 924
	if (encap_limit >= 0) {
		init_tel_txopt(&opt, encap_limit);
		ipv6_push_nfrag_opts(skb, &opt.ops, &proto, NULL);
	}
925 926
	skb_push(skb, sizeof(struct ipv6hdr));
	skb_reset_network_header(skb);
927
	ipv6h = ipv6_hdr(skb);
928
	*(__be32*)ipv6h = fl->fl6_flowlabel | htonl(0x60000000);
L
Linus Torvalds 已提交
929 930 931 932
	dsfield = INET_ECN_encapsulate(0, dsfield);
	ipv6_change_dsfield(ipv6h, ~INET_ECN_MASK, dsfield);
	ipv6h->hop_limit = t->parms.hop_limit;
	ipv6h->nexthdr = proto;
933 934
	ipv6_addr_copy(&ipv6h->saddr, &fl->fl6_src);
	ipv6_addr_copy(&ipv6h->daddr, &fl->fl6_dst);
L
Linus Torvalds 已提交
935 936
	nf_reset(skb);
	pkt_len = skb->len;
H
Herbert Xu 已提交
937
	err = ip6_local_out(skb);
L
Linus Torvalds 已提交
938

939
	if (net_xmit_eval(err) == 0) {
L
Linus Torvalds 已提交
940 941 942 943 944 945 946 947 948 949 950 951 952
		stats->tx_bytes += pkt_len;
		stats->tx_packets++;
	} else {
		stats->tx_errors++;
		stats->tx_aborted_errors++;
	}
	ip6_tnl_dst_store(t, dst);
	return 0;
tx_err_link_failure:
	stats->tx_carrier_errors++;
	dst_link_failure(skb);
tx_err_dst_release:
	dst_release(dst);
953 954 955
	return err;
}

956 957 958 959
static inline int
ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct ip6_tnl *t = netdev_priv(dev);
960
	struct iphdr  *iph = ip_hdr(skb);
961 962 963 964 965 966
	int encap_limit = -1;
	struct flowi fl;
	__u8 dsfield;
	__u32 mtu;
	int err;

967 968
	if ((t->parms.proto != IPPROTO_IPIP && t->parms.proto != 0) ||
	    !ip6_tnl_xmit_ctl(t))
969 970 971 972 973 974 975 976 977 978 979
		return -1;

	if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
		encap_limit = t->parms.encap_limit;

	memcpy(&fl, &t->fl, sizeof (fl));
	fl.proto = IPPROTO_IPIP;

	dsfield = ipv4_get_dsfield(iph);

	if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
A
Al Viro 已提交
980 981
		fl.fl6_flowlabel |= htonl((__u32)iph->tos << IPV6_TCLASS_SHIFT)
					  & IPV6_TCLASS_MASK;
982 983 984 985 986 987 988 989 990 991 992 993 994

	err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
	if (err != 0) {
		/* XXX: send ICMP error even if DF is not set. */
		if (err == -EMSGSIZE)
			icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
				  htonl(mtu));
		return -1;
	}

	return 0;
}

995 996 997 998
static inline int
ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct ip6_tnl *t = netdev_priv(dev);
999
	struct ipv6hdr *ipv6h = ipv6_hdr(skb);
1000 1001 1002 1003 1004 1005 1006
	int encap_limit = -1;
	__u16 offset;
	struct flowi fl;
	__u8 dsfield;
	__u32 mtu;
	int err;

1007 1008
	if ((t->parms.proto != IPPROTO_IPV6 && t->parms.proto != 0) ||
	    !ip6_tnl_xmit_ctl(t) || ip6_tnl_addr_conflict(t, ipv6h))
1009 1010
		return -1;

1011 1012
	offset = parse_tlv_tnl_enc_lim(skb, skb_network_header(skb));
	if (offset > 0) {
1013
		struct ipv6_tlv_tnl_enc_lim *tel;
1014
		tel = (struct ipv6_tlv_tnl_enc_lim *)&skb_network_header(skb)[offset];
1015 1016
		if (tel->encap_limit == 0) {
			icmpv6_send(skb, ICMPV6_PARAMPROB,
1017
				    ICMPV6_HDR_FIELD, offset + 2);
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
			return -1;
		}
		encap_limit = tel->encap_limit - 1;
	} else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
		encap_limit = t->parms.encap_limit;

	memcpy(&fl, &t->fl, sizeof (fl));
	fl.proto = IPPROTO_IPV6;

	dsfield = ipv6_get_dsfield(ipv6h);
	if ((t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS))
		fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_TCLASS_MASK);
	if ((t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL))
		fl.fl6_flowlabel |= (*(__be32 *) ipv6h & IPV6_FLOWLABEL_MASK);

	err = ip6_tnl_xmit2(skb, dev, dsfield, &fl, encap_limit, &mtu);
	if (err != 0) {
		if (err == -EMSGSIZE)
1036
			icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
1037 1038 1039 1040 1041 1042
		return -1;
	}

	return 0;
}

1043
static netdev_tx_t
1044 1045 1046
ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct ip6_tnl *t = netdev_priv(dev);
1047
	struct net_device_stats *stats = &t->dev->stats;
1048 1049 1050
	int ret;

	switch (skb->protocol) {
1051
	case htons(ETH_P_IP):
1052 1053
		ret = ip4ip6_tnl_xmit(skb, dev);
		break;
1054
	case htons(ETH_P_IPV6):
1055 1056 1057 1058 1059 1060 1061 1062 1063
		ret = ip6ip6_tnl_xmit(skb, dev);
		break;
	default:
		goto tx_err;
	}

	if (ret < 0)
		goto tx_err;

1064
	return NETDEV_TX_OK;
1065

L
Linus Torvalds 已提交
1066 1067 1068 1069
tx_err:
	stats->tx_errors++;
	stats->tx_dropped++;
	kfree_skb(skb);
1070
	return NETDEV_TX_OK;
L
Linus Torvalds 已提交
1071 1072 1073 1074 1075
}

static void ip6_tnl_set_cap(struct ip6_tnl *t)
{
	struct ip6_tnl_parm *p = &t->parms;
1076 1077
	int ltype = ipv6_addr_type(&p->laddr);
	int rtype = ipv6_addr_type(&p->raddr);
L
Linus Torvalds 已提交
1078 1079 1080

	p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV);

1081 1082 1083
	if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
	    rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
	    !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
1084
	    (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
1085 1086 1087 1088
		if (ltype&IPV6_ADDR_UNICAST)
			p->flags |= IP6_TNL_F_CAP_XMIT;
		if (rtype&IPV6_ADDR_UNICAST)
			p->flags |= IP6_TNL_F_CAP_RCV;
L
Linus Torvalds 已提交
1089 1090 1091
	}
}

1092
static void ip6_tnl_link_config(struct ip6_tnl *t)
L
Linus Torvalds 已提交
1093 1094 1095 1096 1097
{
	struct net_device *dev = t->dev;
	struct ip6_tnl_parm *p = &t->parms;
	struct flowi *fl = &t->fl;

1098 1099
	memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
	memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
L
Linus Torvalds 已提交
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121

	/* Set up flowi template */
	ipv6_addr_copy(&fl->fl6_src, &p->laddr);
	ipv6_addr_copy(&fl->fl6_dst, &p->raddr);
	fl->oif = p->link;
	fl->fl6_flowlabel = 0;

	if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
		fl->fl6_flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
	if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
		fl->fl6_flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;

	ip6_tnl_set_cap(t);

	if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
		dev->flags |= IFF_POINTOPOINT;
	else
		dev->flags &= ~IFF_POINTOPOINT;

	dev->iflink = p->link;

	if (p->flags & IP6_TNL_F_CAP_XMIT) {
1122 1123 1124
		int strict = (ipv6_addr_type(&p->raddr) &
			      (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));

1125 1126
		struct rt6_info *rt = rt6_lookup(dev_net(dev),
						 &p->raddr, &p->laddr,
1127
						 p->link, strict);
L
Linus Torvalds 已提交
1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145

		if (rt == NULL)
			return;

		if (rt->rt6i_dev) {
			dev->hard_header_len = rt->rt6i_dev->hard_header_len +
				sizeof (struct ipv6hdr);

			dev->mtu = rt->rt6i_dev->mtu - sizeof (struct ipv6hdr);

			if (dev->mtu < IPV6_MIN_MTU)
				dev->mtu = IPV6_MIN_MTU;
		}
		dst_release(&rt->u.dst);
	}
}

/**
1146
 * ip6_tnl_change - update the tunnel parameters
L
Linus Torvalds 已提交
1147 1148 1149 1150
 *   @t: tunnel to be changed
 *   @p: tunnel configuration parameters
 *
 * Description:
1151
 *   ip6_tnl_change() updates the tunnel parameters
L
Linus Torvalds 已提交
1152 1153 1154
 **/

static int
1155
ip6_tnl_change(struct ip6_tnl *t, struct ip6_tnl_parm *p)
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160 1161 1162
{
	ipv6_addr_copy(&t->parms.laddr, &p->laddr);
	ipv6_addr_copy(&t->parms.raddr, &p->raddr);
	t->parms.flags = p->flags;
	t->parms.hop_limit = p->hop_limit;
	t->parms.encap_limit = p->encap_limit;
	t->parms.flowinfo = p->flowinfo;
1163
	t->parms.link = p->link;
1164
	t->parms.proto = p->proto;
1165
	ip6_tnl_dst_reset(t);
1166
	ip6_tnl_link_config(t);
L
Linus Torvalds 已提交
1167 1168 1169 1170
	return 0;
}

/**
1171
 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
L
Linus Torvalds 已提交
1172 1173 1174 1175 1176
 *   @dev: virtual device associated with tunnel
 *   @ifr: parameters passed from userspace
 *   @cmd: command to be performed
 *
 * Description:
1177
 *   ip6_tnl_ioctl() is used for managing IPv6 tunnels
1178
 *   from userspace.
L
Linus Torvalds 已提交
1179 1180 1181 1182 1183 1184 1185
 *
 *   The possible commands are the following:
 *     %SIOCGETTUNNEL: get tunnel parameters for device
 *     %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
 *     %SIOCCHGTUNNEL: change tunnel parameters to those given
 *     %SIOCDELTUNNEL: delete tunnel
 *
1186
 *   The fallback device "ip6tnl0", created during module
L
Linus Torvalds 已提交
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
 *   initialization, can be used for creating other tunnel devices.
 *
 * Return:
 *   0 on success,
 *   %-EFAULT if unable to copy data to or from userspace,
 *   %-EPERM if current process hasn't %CAP_NET_ADMIN set
 *   %-EINVAL if passed tunnel parameters are invalid,
 *   %-EEXIST if changing a tunnel's parameters would cause a conflict
 *   %-ENODEV if attempting to change or delete a nonexisting device
 **/

static int
1199
ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
L
Linus Torvalds 已提交
1200 1201 1202 1203
{
	int err = 0;
	struct ip6_tnl_parm p;
	struct ip6_tnl *t = NULL;
1204 1205
	struct net *net = dev_net(dev);
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
L
Linus Torvalds 已提交
1206 1207 1208

	switch (cmd) {
	case SIOCGETTUNNEL:
1209
		if (dev == ip6n->fb_tnl_dev) {
1210
			if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p))) {
L
Linus Torvalds 已提交
1211 1212 1213
				err = -EFAULT;
				break;
			}
1214
			t = ip6_tnl_locate(net, &p, 0);
1215 1216
		}
		if (t == NULL)
1217
			t = netdev_priv(dev);
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
		memcpy(&p, &t->parms, sizeof (p));
		if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof (p))) {
			err = -EFAULT;
		}
		break;
	case SIOCADDTUNNEL:
	case SIOCCHGTUNNEL:
		err = -EPERM;
		if (!capable(CAP_NET_ADMIN))
			break;
1228 1229
		err = -EFAULT;
		if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
L
Linus Torvalds 已提交
1230
			break;
1231
		err = -EINVAL;
1232 1233
		if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
		    p.proto != 0)
L
Linus Torvalds 已提交
1234
			break;
1235
		t = ip6_tnl_locate(net, &p, cmd == SIOCADDTUNNEL);
1236
		if (dev != ip6n->fb_tnl_dev && cmd == SIOCCHGTUNNEL) {
1237 1238 1239 1240 1241 1242 1243 1244
			if (t != NULL) {
				if (t->dev != dev) {
					err = -EEXIST;
					break;
				}
			} else
				t = netdev_priv(dev);

1245
			ip6_tnl_unlink(ip6n, t);
1246
			err = ip6_tnl_change(t, &p);
1247
			ip6_tnl_link(ip6n, t);
L
Linus Torvalds 已提交
1248 1249
			netdev_state_change(dev);
		}
1250
		if (t) {
L
Linus Torvalds 已提交
1251
			err = 0;
1252 1253 1254 1255 1256
			if (copy_to_user(ifr->ifr_ifru.ifru_data, &t->parms, sizeof (p)))
				err = -EFAULT;

		} else
			err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
L
Linus Torvalds 已提交
1257 1258 1259 1260 1261 1262
		break;
	case SIOCDELTUNNEL:
		err = -EPERM;
		if (!capable(CAP_NET_ADMIN))
			break;

1263
		if (dev == ip6n->fb_tnl_dev) {
1264 1265
			err = -EFAULT;
			if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof (p)))
L
Linus Torvalds 已提交
1266
				break;
1267
			err = -ENOENT;
1268
			if ((t = ip6_tnl_locate(net, &p, 0)) == NULL)
L
Linus Torvalds 已提交
1269
				break;
1270
			err = -EPERM;
1271
			if (t->dev == ip6n->fb_tnl_dev)
L
Linus Torvalds 已提交
1272
				break;
1273
			dev = t->dev;
L
Linus Torvalds 已提交
1274
		}
1275 1276
		err = 0;
		unregister_netdevice(dev);
L
Linus Torvalds 已提交
1277 1278 1279 1280 1281 1282 1283 1284
		break;
	default:
		err = -EINVAL;
	}
	return err;
}

/**
1285
 * ip6_tnl_change_mtu - change mtu manually for tunnel device
L
Linus Torvalds 已提交
1286 1287 1288 1289 1290 1291 1292 1293 1294
 *   @dev: virtual device associated with tunnel
 *   @new_mtu: the new mtu
 *
 * Return:
 *   0 on success,
 *   %-EINVAL if mtu too small
 **/

static int
1295
ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
L
Linus Torvalds 已提交
1296 1297 1298 1299 1300 1301 1302 1303
{
	if (new_mtu < IPV6_MIN_MTU) {
		return -EINVAL;
	}
	dev->mtu = new_mtu;
	return 0;
}

1304 1305 1306 1307 1308 1309 1310 1311

static const struct net_device_ops ip6_tnl_netdev_ops = {
	.ndo_uninit = ip6_tnl_dev_uninit,
	.ndo_start_xmit = ip6_tnl_xmit,
	.ndo_do_ioctl = ip6_tnl_ioctl,
	.ndo_change_mtu = ip6_tnl_change_mtu,
};

L
Linus Torvalds 已提交
1312
/**
1313
 * ip6_tnl_dev_setup - setup virtual tunnel device
L
Linus Torvalds 已提交
1314 1315 1316 1317 1318 1319
 *   @dev: virtual device associated with tunnel
 *
 * Description:
 *   Initialize function pointers and device parameters
 **/

1320
static void ip6_tnl_dev_setup(struct net_device *dev)
L
Linus Torvalds 已提交
1321
{
1322
	dev->netdev_ops = &ip6_tnl_netdev_ops;
L
Linus Torvalds 已提交
1323 1324 1325 1326 1327 1328 1329
	dev->destructor = free_netdev;

	dev->type = ARPHRD_TUNNEL6;
	dev->hard_header_len = LL_MAX_HEADER + sizeof (struct ipv6hdr);
	dev->mtu = ETH_DATA_LEN - sizeof (struct ipv6hdr);
	dev->flags |= IFF_NOARP;
	dev->addr_len = sizeof(struct in6_addr);
1330
	dev->features |= NETIF_F_NETNS_LOCAL;
L
Linus Torvalds 已提交
1331 1332 1333 1334
}


/**
1335
 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
L
Linus Torvalds 已提交
1336 1337 1338 1339
 *   @dev: virtual device associated with tunnel
 **/

static inline void
1340
ip6_tnl_dev_init_gen(struct net_device *dev)
L
Linus Torvalds 已提交
1341
{
1342
	struct ip6_tnl *t = netdev_priv(dev);
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347
	t->dev = dev;
	strcpy(t->parms.name, dev->name);
}

/**
1348
 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
L
Linus Torvalds 已提交
1349 1350 1351
 *   @dev: virtual device associated with tunnel
 **/

1352
static void ip6_tnl_dev_init(struct net_device *dev)
L
Linus Torvalds 已提交
1353
{
1354
	struct ip6_tnl *t = netdev_priv(dev);
1355 1356
	ip6_tnl_dev_init_gen(dev);
	ip6_tnl_link_config(t);
L
Linus Torvalds 已提交
1357 1358 1359
}

/**
1360
 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
L
Linus Torvalds 已提交
1361 1362 1363 1364 1365
 *   @dev: fallback device
 *
 * Return: 0
 **/

1366
static void __net_init ip6_fb_tnl_dev_init(struct net_device *dev)
L
Linus Torvalds 已提交
1367
{
1368
	struct ip6_tnl *t = netdev_priv(dev);
1369 1370 1371
	struct net *net = dev_net(dev);
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);

1372
	ip6_tnl_dev_init_gen(dev);
1373
	t->parms.proto = IPPROTO_IPV6;
L
Linus Torvalds 已提交
1374
	dev_hold(dev);
1375
	ip6n->tnls_wc[0] = t;
L
Linus Torvalds 已提交
1376 1377
}

1378 1379 1380 1381 1382 1383
static struct xfrm6_tunnel ip4ip6_handler = {
	.handler	= ip4ip6_rcv,
	.err_handler	= ip4ip6_err,
	.priority	=	1,
};

L
Linus Torvalds 已提交
1384
static struct xfrm6_tunnel ip6ip6_handler = {
1385 1386
	.handler	= ip6ip6_rcv,
	.err_handler	= ip6ip6_err,
H
Herbert Xu 已提交
1387
	.priority	=	1,
L
Linus Torvalds 已提交
1388 1389
};

1390
static void __net_exit ip6_tnl_destroy_tunnels(struct ip6_tnl_net *ip6n)
1391 1392 1393
{
	int h;
	struct ip6_tnl *t;
1394
	LIST_HEAD(list);
1395 1396

	for (h = 0; h < HASH_SIZE; h++) {
1397 1398 1399 1400 1401
		t = ip6n->tnls_r_l[h];
		while (t != NULL) {
			unregister_netdevice_queue(t->dev, &list);
			t = t->next;
		}
1402 1403 1404
	}

	t = ip6n->tnls_wc[0];
1405 1406
	unregister_netdevice_queue(t->dev, &list);
	unregister_netdevice_many(&list);
1407 1408
}

1409
static int __net_init ip6_tnl_init_net(struct net *net)
1410
{
1411
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1412 1413
	int err;

1414 1415 1416
	ip6n->tnls[0] = ip6n->tnls_wc;
	ip6n->tnls[1] = ip6n->tnls_r_l;

1417 1418 1419 1420 1421 1422
	err = -ENOMEM;
	ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
				      ip6_tnl_dev_setup);

	if (!ip6n->fb_tnl_dev)
		goto err_alloc_dev;
1423
	dev_net_set(ip6n->fb_tnl_dev, net);
1424

1425
	ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
1426 1427 1428 1429

	err = register_netdev(ip6n->fb_tnl_dev);
	if (err < 0)
		goto err_register;
1430 1431
	return 0;

1432 1433 1434
err_register:
	free_netdev(ip6n->fb_tnl_dev);
err_alloc_dev:
1435 1436 1437
	return err;
}

1438
static void __net_exit ip6_tnl_exit_net(struct net *net)
1439
{
1440
	struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1441

1442 1443 1444
	rtnl_lock();
	ip6_tnl_destroy_tunnels(ip6n);
	rtnl_unlock();
1445 1446 1447 1448 1449
}

static struct pernet_operations ip6_tnl_net_ops = {
	.init = ip6_tnl_init_net,
	.exit = ip6_tnl_exit_net,
1450 1451
	.id   = &ip6_tnl_net_id,
	.size = sizeof(struct ip6_tnl_net),
1452 1453
};

L
Linus Torvalds 已提交
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
/**
 * ip6_tunnel_init - register protocol and reserve needed resources
 *
 * Return: 0 on success
 **/

static int __init ip6_tunnel_init(void)
{
	int  err;

1464 1465 1466 1467 1468 1469
	err = register_pernet_device(&ip6_tnl_net_ops);
	if (err < 0)
		goto out_pernet;

	err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
	if (err < 0) {
1470
		printk(KERN_ERR "ip6_tunnel init: can't register ip4ip6\n");
1471
		goto out_ip4ip6;
1472 1473
	}

1474 1475
	err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
	if (err < 0) {
1476
		printk(KERN_ERR "ip6_tunnel init: can't register ip6ip6\n");
1477
		goto out_ip6ip6;
L
Linus Torvalds 已提交
1478
	}
1479

L
Linus Torvalds 已提交
1480
	return 0;
1481 1482

out_ip6ip6:
1483
	xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
1484 1485 1486
out_ip4ip6:
	unregister_pernet_device(&ip6_tnl_net_ops);
out_pernet:
L
Linus Torvalds 已提交
1487 1488 1489 1490 1491 1492 1493 1494 1495
	return err;
}

/**
 * ip6_tunnel_cleanup - free resources and unregister protocol
 **/

static void __exit ip6_tunnel_cleanup(void)
{
1496
	if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
1497
		printk(KERN_INFO "ip6_tunnel close: can't deregister ip4ip6\n");
1498

1499
	if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
1500
		printk(KERN_INFO "ip6_tunnel close: can't deregister ip6ip6\n");
L
Linus Torvalds 已提交
1501

1502
	unregister_pernet_device(&ip6_tnl_net_ops);
L
Linus Torvalds 已提交
1503 1504 1505 1506
}

module_init(ip6_tunnel_init);
module_exit(ip6_tunnel_cleanup);