projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Merge tag 'io_uring-6.4-2023-06-15' of git://git.kernel.dk/linux
[linux-2.6-block.git] / net / ipv6 / ip6_tunnel.c
CommitLineData
2874c5fd 1// SPDX-License-Identifier: GPL-2.0-or-later
1da177e4 2/*
c4d3efaf 3 * IPv6 tunneling device
1da177e4
LT		 4 * Linux INET6 implementation
5 *
6 * Authors:
1ab1457c 7 * Ville Nuorvala <vnuorval@tcs.hut.fi>
c4d3efaf 8 * Yasuyuki Kozakai <kozakai@linux-ipv6.org>
1da177e4 9 *
1da177e4 10 * Based on:
c4d3efaf 11 * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
1da177e4
LT		 12 *
13 * RFC 2473
1da177e4
LT		 14 */
15
f3213831
JP		 16#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
17
1da177e4 18#include <linux/module.h>
4fc268d2 19#include <linux/capability.h>
1da177e4
LT		 20#include <linux/errno.h>
21#include <linux/types.h>
22#include <linux/sockios.h>
c4d3efaf 23#include <linux/icmp.h>
1da177e4
LT		 24#include <linux/if.h>
25#include <linux/in.h>
26#include <linux/ip.h>
1da177e4
LT		 27#include <linux/net.h>
28#include <linux/in6.h>
29#include <linux/netdevice.h>
30#include <linux/if_arp.h>
31#include <linux/icmpv6.h>
32#include <linux/init.h>
33#include <linux/route.h>
34#include <linux/rtnetlink.h>
35#include <linux/netfilter_ipv6.h>
5a0e3ad6 36#include <linux/slab.h>
ddbe5032 37#include <linux/hash.h>
e837735e 38#include <linux/etherdevice.h>
1da177e4 39
7c0f6ba6 40#include <linux/uaccess.h>
60063497 41#include <linux/atomic.h>
1da177e4 42
c4d3efaf 43#include <net/icmp.h>
1da177e4 44#include <net/ip.h>
c5441932 45#include <net/ip_tunnels.h>
1da177e4 46#include <net/ipv6.h>
1da177e4
LT		 47#include <net/ip6_route.h>
48#include <net/addrconf.h>
49#include <net/ip6_tunnel.h>
50#include <net/xfrm.h>
51#include <net/dsfield.h>
52#include <net/inet_ecn.h>
13eeb8e9
PE		 53#include <net/net_namespace.h>
54#include <net/netns/generic.h>
8d79266b 55#include <net/dst_metadata.h>
1da177e4
LT		 56
57MODULE_AUTHOR("Ville Nuorvala");
c4d3efaf 58MODULE_DESCRIPTION("IPv6 tunneling device");
1da177e4 59MODULE_LICENSE("GPL");
f98f89a0 60MODULE_ALIAS_RTNL_LINK("ip6tnl");
6dfbd87a 61MODULE_ALIAS_NETDEV("ip6tnl0");
1da177e4 62
e87a8f24
JK		 63#define IP6_TUNNEL_HASH_SIZE_SHIFT 5
64#define IP6_TUNNEL_HASH_SIZE (1 << IP6_TUNNEL_HASH_SIZE_SHIFT)
1da177e4 65
f4e0b4c5
ND		 66static bool log_ecn_error = true;
67module_param(log_ecn_error, bool, 0644);
68MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
69
ddbe5032
ED		 70static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2)
71{
72 u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2);
73
e87a8f24 74 return hash_32(hash, IP6_TUNNEL_HASH_SIZE_SHIFT);
ddbe5032 75}
1da177e4 76
8560f226 77static int ip6_tnl_dev_init(struct net_device *dev);
3144581c 78static void ip6_tnl_dev_setup(struct net_device *dev);
c075b130 79static struct rtnl_link_ops ip6_link_ops __read_mostly;
1da177e4 80
c7d03a00 81static unsigned int ip6_tnl_net_id __read_mostly;
13eeb8e9 82struct ip6_tnl_net {
15820e12
PE		 83 /* the IPv6 tunnel fallback device */
84 struct net_device *fb_tnl_dev;
3e6c9fb5 85 /* lists for storing tunnels in use */
e87a8f24 86 struct ip6_tnl __rcu *tnls_r_l[IP6_TUNNEL_HASH_SIZE];
94767632
ED		 87 struct ip6_tnl __rcu *tnls_wc[1];
88 struct ip6_tnl __rcu **tnls[2];
8d79266b 89 struct ip6_tnl __rcu *collect_md_tun;
13eeb8e9
PE		 90};
91
f200e98d
VF		 92static inline int ip6_tnl_mpls_supported(void)
93{
94 return IS_ENABLED(CONFIG_MPLS);
95}
96
b51cd7c8
AL		 97#define for_each_ip6_tunnel_rcu(start) \
98 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
99
1da177e4 100/**
3144581c 101 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
b51cd7c8 102 * @net: network namespace
5fdcce21 103 * @link: ifindex of underlying interface
1ab1457c
YH		 104 * @remote: the address of the tunnel exit-point
105 * @local: the address of the tunnel entry-point
1da177e4 106 *
1ab1457c 107 * Return:
1da177e4 108 * tunnel matching given end-points if found,
1ab1457c 109 * else fallback tunnel if its device is up,
1da177e4
LT		 110 * else %NULL
111 **/
112
113static struct ip6_tnl *
5fdcce21
WD		 114ip6_tnl_lookup(struct net *net, int link,
115 const struct in6_addr *remote, const struct in6_addr *local)
1da177e4 116{
ddbe5032 117 unsigned int hash = HASH(remote, local);
5fdcce21 118 struct ip6_tnl *t, *cand = NULL;
3e6c9fb5 119 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
ea3dc960 120 struct in6_addr any;
1da177e4 121
ddbe5032 122 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
5fdcce21
WD		 123 if (!ipv6_addr_equal(local, &t->parms.laddr) ||
124 !ipv6_addr_equal(remote, &t->parms.raddr) ||
125 !(t->dev->flags & IFF_UP))
126 continue;
127
128 if (link == t->parms.link)
1da177e4 129 return t;
5fdcce21
WD		 130 else
131 cand = t;
1da177e4 132 }
ea3dc960
SK		 133
134 memset(&any, 0, sizeof(any));
135 hash = HASH(&any, local);
136 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
5fdcce21
WD		 137 if (!ipv6_addr_equal(local, &t->parms.laddr) ||
138 !ipv6_addr_any(&t->parms.raddr) ||
139 !(t->dev->flags & IFF_UP))
140 continue;
141
142 if (link == t->parms.link)
ea3dc960 143 return t;
5fdcce21
WD		 144 else if (!cand)
145 cand = t;
ea3dc960
SK		 146 }
147
148 hash = HASH(remote, &any);
149 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
5fdcce21
WD		 150 if (!ipv6_addr_equal(remote, &t->parms.raddr) ||
151 !ipv6_addr_any(&t->parms.laddr) ||
152 !(t->dev->flags & IFF_UP))
153 continue;
154
155 if (link == t->parms.link)
ea3dc960 156 return t;
5fdcce21
WD		 157 else if (!cand)
158 cand = t;
ea3dc960
SK		 159 }
160
5fdcce21
WD		 161 if (cand)
162 return cand;
163
8d79266b 164 t = rcu_dereference(ip6n->collect_md_tun);
6c1cb439 165 if (t && t->dev->flags & IFF_UP)
8d79266b
AS		 166 return t;
167
2922bc8a
ED		 168 t = rcu_dereference(ip6n->tnls_wc[0]);
169 if (t && (t->dev->flags & IFF_UP))
1da177e4
LT		 170 return t;
171
172 return NULL;
173}
174
175/**
3144581c 176 * ip6_tnl_bucket - get head of list matching given tunnel parameters
2c4de211 177 * @ip6n: the private data for ip6_vti in the netns
1ab1457c 178 * @p: parameters containing tunnel end-points
1da177e4
LT		 179 *
180 * Description:
3144581c 181 * ip6_tnl_bucket() returns the head of the list matching the
1da177e4
LT		 182 * &struct in6_addr entries laddr and raddr in @p.
183 *
1ab1457c 184 * Return: head of IPv6 tunnel list
1da177e4
LT		 185 **/
186
94767632 187static struct ip6_tnl __rcu **
c12b395a 188ip6_tnl_bucket(struct ip6_tnl_net *ip6n, const struct __ip6_tnl_parm *p)
1da177e4 189{
b71d1d42
ED		 190 const struct in6_addr *remote = &p->raddr;
191 const struct in6_addr *local = &p->laddr;
95c96174 192 unsigned int h = 0;
1da177e4
LT		 193 int prio = 0;
194
195 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
196 prio = 1;
ddbe5032 197 h = HASH(remote, local);
1da177e4 198 }
3e6c9fb5 199 return &ip6n->tnls[prio][h];
1da177e4
LT		 200}
201
202/**
3144581c 203 * ip6_tnl_link - add tunnel to hash table
2c4de211 204 * @ip6n: the private data for ip6_vti in the netns
1da177e4
LT		 205 * @t: tunnel to be added
206 **/
207
208static void
2dd02c89 209ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 210{
94767632 211 struct ip6_tnl __rcu **tp = ip6_tnl_bucket(ip6n, &t->parms);
1da177e4 212
8d79266b
AS		 213 if (t->parms.collect_md)
214 rcu_assign_pointer(ip6n->collect_md_tun, t);
cf778b00
ED		 215 rcu_assign_pointer(t->next , rtnl_dereference(*tp));
216 rcu_assign_pointer(*tp, t);
1da177e4
LT		 217}
218
219/**
3144581c 220 * ip6_tnl_unlink - remove tunnel from hash table
2c4de211 221 * @ip6n: the private data for ip6_vti in the netns
1da177e4
LT		 222 * @t: tunnel to be removed
223 **/
224
225static void
2dd02c89 226ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 227{
94767632
ED		 228 struct ip6_tnl __rcu **tp;
229 struct ip6_tnl *iter;
230
8d79266b
AS		 231 if (t->parms.collect_md)
232 rcu_assign_pointer(ip6n->collect_md_tun, NULL);
233
94767632
ED		 234 for (tp = ip6_tnl_bucket(ip6n, &t->parms);
235 (iter = rtnl_dereference(*tp)) != NULL;
236 tp = &iter->next) {
237 if (t == iter) {
cf778b00 238 rcu_assign_pointer(*tp, t->next);
1da177e4
LT		 239 break;
240 }
241 }
242}
243
8560f226
ED		 244static void ip6_dev_free(struct net_device *dev)
245{
cdf3464e
MKL		 246 struct ip6_tnl *t = netdev_priv(dev);
247
0d3c703a 248 gro_cells_destroy(&t->gro_cells);
607f725f 249 dst_cache_destroy(&t->dst_cache);
8560f226 250 free_percpu(dev->tstats);
8560f226
ED		 251}
252
0b112457
ND		 253static int ip6_tnl_create2(struct net_device *dev)
254{
255 struct ip6_tnl *t = netdev_priv(dev);
256 struct net *net = dev_net(dev);
257 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
258 int err;
259
b6ee376c 260 dev->rtnl_link_ops = &ip6_link_ops;
0b112457
ND		 261 err = register_netdevice(dev);
262 if (err < 0)
263 goto out;
264
265 strcpy(t->parms.name, dev->name);
0b112457 266
0b112457
ND		 267 ip6_tnl_link(ip6n, t);
268 return 0;
269
270out:
271 return err;
272}
273
1da177e4 274/**
2c53040f 275 * ip6_tnl_create - create a new tunnel
b51cd7c8 276 * @net: network namespace
1da177e4 277 * @p: tunnel parameters
1da177e4
LT		 278 *
279 * Description:
280 * Create tunnel matching given parameters.
1ab1457c
YH		 281 *
282 * Return:
37355565 283 * created tunnel or error pointer
1da177e4
LT		 284 **/
285
c12b395a 286static struct ip6_tnl *ip6_tnl_create(struct net *net, struct __ip6_tnl_parm *p)
1da177e4
LT		 287{
288 struct net_device *dev;
289 struct ip6_tnl *t;
290 char name[IFNAMSIZ];
db7a65e3 291 int err = -E2BIG;
1da177e4 292
db7a65e3
ED		 293 if (p->name[0]) {
294 if (!dev_valid_name(p->name))
295 goto failed;
7574cc58 296 strscpy(name, p->name, IFNAMSIZ);
db7a65e3 297 } else {
34cc7ba6 298 sprintf(name, "ip6tnl%%d");
db7a65e3
ED		 299 }
300 err = -ENOMEM;
c835a677
TG		 301 dev = alloc_netdev(sizeof(*t), name, NET_NAME_UNKNOWN,
302 ip6_tnl_dev_setup);
63159f29 303 if (!dev)
567131a7 304 goto failed;
1da177e4 305
554eb277
PE		 306 dev_net_set(dev, net);
307
2941a486 308 t = netdev_priv(dev);
1da177e4 309 t->parms = *p;
0bd87628 310 t->net = dev_net(dev);
0b112457 311 err = ip6_tnl_create2(dev);
8560f226
ED		 312 if (err < 0)
313 goto failed_free;
1da177e4 314
567131a7 315 return t;
b37d428b
PE		 316
317failed_free:
cf124db5 318 free_netdev(dev);
567131a7 319failed:
37355565 320 return ERR_PTR(err);
1da177e4
LT		 321}
322
323/**
3144581c 324 * ip6_tnl_locate - find or create tunnel matching given parameters
b51cd7c8 325 * @net: network namespace
1ab1457c 326 * @p: tunnel parameters
1da177e4
LT		 327 * @create: != 0 if allowed to create new tunnel if no match found
328 *
329 * Description:
3144581c 330 * ip6_tnl_locate() first tries to locate an existing tunnel
1da177e4
LT		 331 * based on @parms. If this is unsuccessful, but @create is set a new
332 * tunnel device is created and registered for use.
333 *
334 * Return:
37355565 335 * matching tunnel or error pointer
1da177e4
LT		 336 **/
337
2dd02c89 338static struct ip6_tnl *ip6_tnl_locate(struct net *net,
c12b395a 339 struct __ip6_tnl_parm *p, int create)
1da177e4 340{
b71d1d42
ED		 341 const struct in6_addr *remote = &p->raddr;
342 const struct in6_addr *local = &p->laddr;
94767632 343 struct ip6_tnl __rcu **tp;
1da177e4 344 struct ip6_tnl *t;
2dd02c89 345 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 346
94767632
ED		 347 for (tp = ip6_tnl_bucket(ip6n, p);
348 (t = rtnl_dereference(*tp)) != NULL;
349 tp = &t->next) {
1da177e4 350 if (ipv6_addr_equal(local, &t->parms.laddr) &&
5fdcce21
WD		 351 ipv6_addr_equal(remote, &t->parms.raddr) &&
352 p->link == t->parms.link) {
2b0bb01b 353 if (create)
37355565 354 return ERR_PTR(-EEXIST);
2b0bb01b 355
567131a7 356 return t;
2b0bb01b 357 }
1da177e4
LT		 358 }
359 if (!create)
37355565 360 return ERR_PTR(-ENODEV);
2dd02c89 361 return ip6_tnl_create(net, p);
1da177e4
LT		 362}
363
364/**
3144581c 365 * ip6_tnl_dev_uninit - tunnel device uninitializer
1da177e4 366 * @dev: the device to be destroyed
1ab1457c 367 *
1da177e4 368 * Description:
3144581c 369 * ip6_tnl_dev_uninit() removes tunnel from its list
1da177e4
LT		 370 **/
371
372static void
3144581c 373ip6_tnl_dev_uninit(struct net_device *dev)
1da177e4 374{
2941a486 375 struct ip6_tnl *t = netdev_priv(dev);
0bd87628 376 struct net *net = t->net;
2dd02c89 377 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 378
94767632 379 if (dev == ip6n->fb_tnl_dev)
a9b3cd7f 380 RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL);
94767632 381 else
2dd02c89 382 ip6_tnl_unlink(ip6n, t);
607f725f 383 dst_cache_reset(&t->dst_cache);
d62607c3 384 netdev_put(dev, &t->dev_tracker);
1da177e4
LT		 385}
386
387/**
f7b88985 388 * ip6_tnl_parse_tlv_enc_lim - handle encapsulation limit option
1da177e4 389 * @skb: received socket buffer
2c4de211 390 * @raw: the ICMPv6 error message data
1da177e4 391 *
1ab1457c
YH		 392 * Return:
393 * 0 if none was found,
1da177e4
LT		 394 * else index to encapsulation limit
395 **/
396
c12b395a 397__u16 ip6_tnl_parse_tlv_enc_lim(struct sk_buff *skb, __u8 *raw)
1da177e4 398{
fbfa743a
ED		 399 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *)raw;
400 unsigned int nhoff = raw - skb->data;
401 unsigned int off = nhoff + sizeof(*ipv6h);
402 u8 next, nexthdr = ipv6h->nexthdr;
1da177e4
LT		 403
404 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
1da177e4 405 struct ipv6_opt_hdr *hdr;
fbfa743a
ED		 406 u16 optlen;
407
408 if (!pskb_may_pull(skb, off + sizeof(*hdr)))
1da177e4
LT		 409 break;
410
fbfa743a 411 hdr = (struct ipv6_opt_hdr *)(skb->data + off);
1da177e4
LT		 412 if (nexthdr == NEXTHDR_FRAGMENT) {
413 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
414 if (frag_hdr->frag_off)
415 break;
416 optlen = 8;
417 } else if (nexthdr == NEXTHDR_AUTH) {
416e8126 418 optlen = ipv6_authlen(hdr);
1da177e4
LT		 419 } else {
420 optlen = ipv6_optlen(hdr);
421 }
fbfa743a
ED		 422 /* cache hdr->nexthdr, since pskb_may_pull() might
423 * invalidate hdr
424 */
425 next = hdr->nexthdr;
1da177e4 426 if (nexthdr == NEXTHDR_DEST) {
fbfa743a
ED		 427 u16 i = 2;
428
429 /* Remember : hdr is no longer valid at this point. */
430 if (!pskb_may_pull(skb, off + optlen))
431 break;
432
1da177e4
LT		 433 while (1) {
434 struct ipv6_tlv_tnl_enc_lim *tel;
435
436 /* No more room for encapsulation limit */
fbfa743a 437 if (i + sizeof(*tel) > optlen)
1da177e4
LT		 438 break;
439
63117f09 440 tel = (struct ipv6_tlv_tnl_enc_lim *)(skb->data + off + i);
1da177e4
LT		 441 /* return index of option if found and valid */
442 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
443 tel->length == 1)
fbfa743a 444 return i + off - nhoff;
1da177e4
LT		 445 /* else jump to next option */
446 if (tel->type)
447 i += tel->length + 2;
448 else
449 i++;
450 }
451 }
fbfa743a 452 nexthdr = next;
1da177e4
LT		 453 off += optlen;
454 }
455 return 0;
456}
c12b395a 457EXPORT_SYMBOL(ip6_tnl_parse_tlv_enc_lim);
1da177e4 458
2c4de211
XL		 459/* ip6_tnl_err() should handle errors in the tunnel according to the
460 * specifications in RFC 2473.
461 */
d2acc347 462static int
502b0935 463ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
d5fdd6ba 464 u8 *type, u8 *code, int *msg, __u32 *info, int offset)
1da177e4 465{
383c1f88
XL		 466 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *)skb->data;
467 struct net *net = dev_net(skb->dev);
d5fdd6ba
BH		 468 u8 rel_type = ICMPV6_DEST_UNREACH;
469 u8 rel_code = ICMPV6_ADDR_UNREACH;
1da177e4 470 __u32 rel_info = 0;
383c1f88 471 struct ip6_tnl *t;
d2acc347 472 int err = -ENOENT;
383c1f88
XL		 473 int rel_msg = 0;
474 u8 tproto;
475 __u16 len;
1da177e4 476
1ab1457c
YH		 477 /* If the packet doesn't contain the original IPv6 header we are
478 in trouble since we might need the source address for further
1da177e4
LT		 479 processing of the error. */
480
2922bc8a 481 rcu_read_lock();
5fdcce21 482 t = ip6_tnl_lookup(dev_net(skb->dev), skb->dev->ifindex, &ipv6h->daddr, &ipv6h->saddr);
63159f29 483 if (!t)
1da177e4
LT		 484 goto out;
485
6aa7de05 486 tproto = READ_ONCE(t->parms.proto);
acf722f7 487 if (tproto != ipproto && tproto != 0)
502b0935
YK		 488 goto out;
489
d2acc347
HX		 490 err = 0;
491
e490d1d8 492 switch (*type) {
1da177e4 493 case ICMPV6_DEST_UNREACH:
17a10c92
MB		 494 net_dbg_ratelimited("%s: Path to destination invalid or inactive!\n",
495 t->parms.name);
1da177e4
LT		 496 rel_msg = 1;
497 break;
498 case ICMPV6_TIME_EXCEED:
e490d1d8 499 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
17a10c92
MB		 500 net_dbg_ratelimited("%s: Too small hop limit or routing loop in tunnel!\n",
501 t->parms.name);
1da177e4
LT		 502 rel_msg = 1;
503 }
504 break;
46d30cb1
KC		 505 case ICMPV6_PARAMPROB: {
506 struct ipv6_tlv_tnl_enc_lim *tel;
507 __u32 teli;
508
107a5fe6 509 teli = 0;
e490d1d8 510 if ((*code) == ICMPV6_HDR_FIELD)
c12b395a 511 teli = ip6_tnl_parse_tlv_enc_lim(skb, skb->data);
1da177e4 512
704eae1f 513 if (teli && teli == *info - 2) {
1da177e4
LT		 514 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
515 if (tel->encap_limit == 0) {
17a10c92
MB		 516 net_dbg_ratelimited("%s: Too small encapsulation limit or routing loop in tunnel!\n",
517 t->parms.name);
1da177e4
LT		 518 rel_msg = 1;
519 }
e87cc472 520 } else {
17a10c92
MB		 521 net_dbg_ratelimited("%s: Recipient unable to parse tunneled packet!\n",
522 t->parms.name);
1da177e4
LT		 523 }
524 break;
46d30cb1
KC		 525 }
526 case ICMPV6_PKT_TOOBIG: {
527 __u32 mtu;
528
b00f5432
XL		 529 ip6_update_pmtu(skb, net, htonl(*info), 0, 0,
530 sock_net_uid(net, NULL));
704eae1f 531 mtu = *info - offset;
1da177e4
LT		 532 if (mtu < IPV6_MIN_MTU)
533 mtu = IPV6_MIN_MTU;
e5d08d71
IM		 534 len = sizeof(*ipv6h) + ntohs(ipv6h->payload_len);
535 if (len > mtu) {
1da177e4
LT		 536 rel_type = ICMPV6_PKT_TOOBIG;
537 rel_code = 0;
538 rel_info = mtu;
539 rel_msg = 1;
540 }
541 break;
46d30cb1 542 }
383c1f88
XL		 543 case NDISC_REDIRECT:
544 ip6_redirect(skb, net, skb->dev->ifindex, 0,
545 sock_net_uid(net, NULL));
546 break;
1da177e4 547 }
e490d1d8
YK		 548
549 *type = rel_type;
550 *code = rel_code;
551 *info = rel_info;
552 *msg = rel_msg;
553
554out:
2922bc8a 555 rcu_read_unlock();
e490d1d8
YK		 556 return err;
557}
558
c4d3efaf
YK		 559static int
560ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 561 u8 type, u8 code, int offset, __be32 info)
c4d3efaf 562{
704eae1f 563 __u32 rel_info = ntohl(info);
b71d1d42 564 const struct iphdr *eiph;
77552cfa
XL		 565 struct sk_buff *skb2;
566 int err, rel_msg = 0;
567 u8 rel_type = type;
568 u8 rel_code = code;
c4d3efaf 569 struct rtable *rt;
31e4543d 570 struct flowi4 fl4;
c4d3efaf 571
502b0935
YK		 572 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
573 &rel_msg, &rel_info, offset);
c4d3efaf
YK		 574 if (err < 0)
575 return err;
576
577 if (rel_msg == 0)
578 return 0;
579
580 switch (rel_type) {
581 case ICMPV6_DEST_UNREACH:
582 if (rel_code != ICMPV6_ADDR_UNREACH)
583 return 0;
584 rel_type = ICMP_DEST_UNREACH;
585 rel_code = ICMP_HOST_UNREACH;
586 break;
587 case ICMPV6_PKT_TOOBIG:
588 if (rel_code != 0)
589 return 0;
590 rel_type = ICMP_DEST_UNREACH;
591 rel_code = ICMP_FRAG_NEEDED;
592 break;
593 default:
594 return 0;
595 }
596
597 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
598 return 0;
599
600 skb2 = skb_clone(skb, GFP_ATOMIC);
601 if (!skb2)
602 return 0;
603
adf30907
ED		 604 skb_dst_drop(skb2);
605
c4d3efaf 606 skb_pull(skb2, offset);
c1d2bbe1 607 skb_reset_network_header(skb2);
eddc9ec5 608 eiph = ip_hdr(skb2);
c4d3efaf
YK		 609
610 /* Try to guess incoming interface */
77552cfa
XL		 611 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL, eiph->saddr,
612 0, 0, 0, IPPROTO_IPIP, RT_TOS(eiph->tos), 0);
b23dd4fe 613 if (IS_ERR(rt))
c4d3efaf
YK		 614 goto out;
615
d8d1f30b 616 skb2->dev = rt->dst.dev;
77552cfa 617 ip_rt_put(rt);
c4d3efaf
YK		 618
619 /* route "incoming" packet */
620 if (rt->rt_flags & RTCF_LOCAL) {
31e4543d 621 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL,
77552cfa
XL		 622 eiph->daddr, eiph->saddr, 0, 0,
623 IPPROTO_IPIP, RT_TOS(eiph->tos), 0);
b2e54b09 624 if (IS_ERR(rt) || rt->dst.dev->type != ARPHRD_TUNNEL6) {
b23dd4fe
DM		 625 if (!IS_ERR(rt))
626 ip_rt_put(rt);
c4d3efaf
YK		 627 goto out;
628 }
b23dd4fe 629 skb_dst_set(skb2, &rt->dst);
c4d3efaf 630 } else {
c4d3efaf
YK		 631 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
632 skb2->dev) ||
b2e54b09 633 skb_dst(skb2)->dev->type != ARPHRD_TUNNEL6)
c4d3efaf
YK		 634 goto out;
635 }
636
637 /* change mtu on this route */
638 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
adf30907 639 if (rel_info > dst_mtu(skb_dst(skb2)))
c4d3efaf
YK		 640 goto out;
641
7a1592bc 642 skb_dst_update_pmtu_no_confirm(skb2, rel_info);
c4d3efaf
YK		 643 }
644
704eae1f 645 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
c4d3efaf
YK		 646
647out:
648 kfree_skb(skb2);
649 return 0;
650}
651
e490d1d8
YK		 652static int
653ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 654 u8 type, u8 code, int offset, __be32 info)
e490d1d8 655{
77552cfa
XL		 656 __u32 rel_info = ntohl(info);
657 int err, rel_msg = 0;
d5fdd6ba
BH		 658 u8 rel_type = type;
659 u8 rel_code = code;
e490d1d8 660
502b0935
YK		 661 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
662 &rel_msg, &rel_info, offset);
e490d1d8
YK		 663 if (err < 0)
664 return err;
665
666 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
1da177e4
LT		 667 struct rt6_info *rt;
668 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 669
1da177e4 670 if (!skb2)
e490d1d8 671 return 0;
1da177e4 672
adf30907 673 skb_dst_drop(skb2);
1da177e4 674 skb_pull(skb2, offset);
c1d2bbe1 675 skb_reset_network_header(skb2);
1da177e4
LT		 676
677 /* Try to guess incoming interface */
2f7f54b7 678 rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
b75cc8f9 679 NULL, 0, skb2, 0);
1da177e4 680
d1918542
DM		 681 if (rt && rt->dst.dev)
682 skb2->dev = rt->dst.dev;
1da177e4 683
3ffe533c 684 icmpv6_send(skb2, rel_type, rel_code, rel_info);
1da177e4 685
94e187c0 686 ip6_rt_put(rt);
1da177e4
LT		 687
688 kfree_skb(skb2);
689 }
e490d1d8
YK		 690
691 return 0;
1da177e4
LT		 692}
693
f200e98d
VF		 694static int
695mplsip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
696 u8 type, u8 code, int offset, __be32 info)
697{
698 __u32 rel_info = ntohl(info);
699 int err, rel_msg = 0;
700 u8 rel_type = type;
701 u8 rel_code = code;
702
703 err = ip6_tnl_err(skb, IPPROTO_MPLS, opt, &rel_type, &rel_code,
704 &rel_msg, &rel_info, offset);
705 return err;
706}
707
f4e0b4c5
ND		 708static int ip4ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
709 const struct ipv6hdr *ipv6h,
710 struct sk_buff *skb)
c4d3efaf
YK		 711{
712 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
713
714 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
eddc9ec5 715 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
c4d3efaf 716
f4e0b4c5 717 return IP6_ECN_decapsulate(ipv6h, skb);
c4d3efaf
YK		 718}
719
f4e0b4c5
ND		 720static int ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
721 const struct ipv6hdr *ipv6h,
722 struct sk_buff *skb)
1da177e4 723{
8359925b 724 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
29bb43b4 725 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
1da177e4 726
f4e0b4c5 727 return IP6_ECN_decapsulate(ipv6h, skb);
1da177e4 728}
8359925b 729
f200e98d
VF		 730static inline int mplsip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
731 const struct ipv6hdr *ipv6h,
732 struct sk_buff *skb)
733{
734 /* ECN is not supported in AF_MPLS */
735 return 0;
736}
737
c12b395a 738__u32 ip6_tnl_get_cap(struct ip6_tnl *t,
d0087b29
VN		 739 const struct in6_addr *laddr,
740 const struct in6_addr *raddr)
741{
c12b395a 742 struct __ip6_tnl_parm *p = &t->parms;
d0087b29
VN		 743 int ltype = ipv6_addr_type(laddr);
744 int rtype = ipv6_addr_type(raddr);
745 __u32 flags = 0;
746
747 if (ltype == IPV6_ADDR_ANY || rtype == IPV6_ADDR_ANY) {
748 flags = IP6_TNL_F_CAP_PER_PACKET;
749 } else if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
750 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
751 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
752 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
753 if (ltype&IPV6_ADDR_UNICAST)
754 flags |= IP6_TNL_F_CAP_XMIT;
755 if (rtype&IPV6_ADDR_UNICAST)
756 flags |= IP6_TNL_F_CAP_RCV;
757 }
758 return flags;
759}
c12b395a 760EXPORT_SYMBOL(ip6_tnl_get_cap);
d0087b29 761
f1a28eab 762/* called with rcu_read_lock() */
c12b395a 763int ip6_tnl_rcv_ctl(struct ip6_tnl *t,
d0087b29
VN		 764 const struct in6_addr *laddr,
765 const struct in6_addr *raddr)
09c6bbf0 766{
c12b395a 767 struct __ip6_tnl_parm *p = &t->parms;
09c6bbf0 768 int ret = 0;
0bd87628 769 struct net *net = t->net;
09c6bbf0 770
d0087b29
VN		 771 if ((p->flags & IP6_TNL_F_CAP_RCV) ||
772 ((p->flags & IP6_TNL_F_CAP_PER_PACKET) &&
773 (ip6_tnl_get_cap(t, laddr, raddr) & IP6_TNL_F_CAP_RCV))) {
1ab1457c 774 struct net_device *ldev = NULL;
09c6bbf0
VN		 775
776 if (p->link)
f1a28eab 777 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0 778
d0087b29 779 if ((ipv6_addr_is_multicast(laddr) ||
232378e8
DA		 780 likely(ipv6_chk_addr_and_flags(net, laddr, ldev, false,
781 0, IFA_F_TENTATIVE))) &&
908d140a 782 ((p->flags & IP6_TNL_F_ALLOW_LOCAL_REMOTE) ||
232378e8
DA		 783 likely(!ipv6_chk_addr_and_flags(net, raddr, ldev, true,
784 0, IFA_F_TENTATIVE))))
09c6bbf0 785 ret = 1;
09c6bbf0
VN		 786 }
787 return ret;
788}
c12b395a 789EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl);
1da177e4 790
0d3c703a
TH		 791static int __ip6_tnl_rcv(struct ip6_tnl *tunnel, struct sk_buff *skb,
792 const struct tnl_ptk_info *tpi,
793 struct metadata_dst *tun_dst,
794 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
795 const struct ipv6hdr *ipv6h,
796 struct sk_buff *skb),
797 bool log_ecn_err)
1da177e4 798{
b71d1d42 799 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
f4e0b4c5 800 int err;
1da177e4 801
0d3c703a
TH		 802 if ((!(tpi->flags & TUNNEL_CSUM) &&
803 (tunnel->parms.i_flags & TUNNEL_CSUM)) ||
804 ((tpi->flags & TUNNEL_CSUM) &&
805 !(tunnel->parms.i_flags & TUNNEL_CSUM))) {
2fad1ba3
ED		 806 DEV_STATS_INC(tunnel->dev, rx_crc_errors);
807 DEV_STATS_INC(tunnel->dev, rx_errors);
0d3c703a
TH		 808 goto drop;
809 }
8560f226 810
0d3c703a
TH		 811 if (tunnel->parms.i_flags & TUNNEL_SEQ) {
812 if (!(tpi->flags & TUNNEL_SEQ) ||
813 (tunnel->i_seqno &&
814 (s32)(ntohl(tpi->seq) - tunnel->i_seqno) < 0)) {
2fad1ba3
ED		 815 DEV_STATS_INC(tunnel->dev, rx_fifo_errors);
816 DEV_STATS_INC(tunnel->dev, rx_errors);
0d3c703a 817 goto drop;
502b0935 818 }
0d3c703a
TH		 819 tunnel->i_seqno = ntohl(tpi->seq) + 1;
820 }
502b0935 821
0d3c703a 822 skb->protocol = tpi->proto;
1da177e4 823
0d3c703a
TH		 824 /* Warning: All skb pointers will be invalidated! */
825 if (tunnel->dev->type == ARPHRD_ETHER) {
826 if (!pskb_may_pull(skb, ETH_HLEN)) {
2fad1ba3
ED		 827 DEV_STATS_INC(tunnel->dev, rx_length_errors);
828 DEV_STATS_INC(tunnel->dev, rx_errors);
0d3c703a 829 goto drop;
1da177e4 830 }
0d3c703a
TH		 831
832 ipv6h = ipv6_hdr(skb);
833 skb->protocol = eth_type_trans(skb, tunnel->dev);
834 skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
835 } else {
836 skb->dev = tunnel->dev;
da5a2e49 837 skb_reset_mac_header(skb);
0d3c703a
TH		 838 }
839
840 skb_reset_network_header(skb);
841 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
842
843 __skb_tunnel_rx(skb, tunnel->dev, tunnel->net);
844
845 err = dscp_ecn_decapsulate(tunnel, ipv6h, skb);
846 if (unlikely(err)) {
847 if (log_ecn_err)
848 net_info_ratelimited("non-ECT from %pI6 with DS=%#x\n",
849 &ipv6h->saddr,
850 ipv6_get_dsfield(ipv6h));
851 if (err > 1) {
2fad1ba3
ED		 852 DEV_STATS_INC(tunnel->dev, rx_frame_errors);
853 DEV_STATS_INC(tunnel->dev, rx_errors);
0d3c703a 854 goto drop;
f4e0b4c5 855 }
0d3c703a 856 }
f4e0b4c5 857
afd2051b 858 dev_sw_netstats_rx_add(tunnel->dev, skb->len);
8560f226 859
0d3c703a 860 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(tunnel->dev)));
8990f468 861
8d79266b
AS		 862 if (tun_dst)
863 skb_dst_set(skb, (struct dst_entry *)tun_dst);
864
0d3c703a
TH		 865 gro_cells_receive(&tunnel->gro_cells, skb);
866 return 0;
867
868drop:
f1925ca5
HY		 869 if (tun_dst)
870 dst_release((struct dst_entry *)tun_dst);
0d3c703a
TH		 871 kfree_skb(skb);
872 return 0;
873}
874
875int ip6_tnl_rcv(struct ip6_tnl *t, struct sk_buff *skb,
876 const struct tnl_ptk_info *tpi,
877 struct metadata_dst *tun_dst,
878 bool log_ecn_err)
879{
272502fc
MT		 880 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
881 const struct ipv6hdr *ipv6h,
882 struct sk_buff *skb);
883
884 dscp_ecn_decapsulate = ip6ip6_dscp_ecn_decapsulate;
885 if (tpi->proto == htons(ETH_P_IP))
886 dscp_ecn_decapsulate = ip4ip6_dscp_ecn_decapsulate;
887
888 return __ip6_tnl_rcv(t, skb, tpi, tun_dst, dscp_ecn_decapsulate,
0d3c703a
TH		 889 log_ecn_err);
890}
891EXPORT_SYMBOL(ip6_tnl_rcv);
892
893static const struct tnl_ptk_info tpi_v6 = {
894 /* no tunnel info required for ipxip6. */
895 .proto = htons(ETH_P_IPV6),
896};
897
898static const struct tnl_ptk_info tpi_v4 = {
899 /* no tunnel info required for ipxip6. */
900 .proto = htons(ETH_P_IP),
901};
902
f200e98d
VF		 903static const struct tnl_ptk_info tpi_mpls = {
904 /* no tunnel info required for mplsip6. */
905 .proto = htons(ETH_P_MPLS_UC),
906};
907
0d3c703a
TH		 908static int ipxip6_rcv(struct sk_buff *skb, u8 ipproto,
909 const struct tnl_ptk_info *tpi,
910 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
911 const struct ipv6hdr *ipv6h,
912 struct sk_buff *skb))
913{
914 struct ip6_tnl *t;
915 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
8d79266b 916 struct metadata_dst *tun_dst = NULL;
0d3c703a
TH		 917 int ret = -1;
918
919 rcu_read_lock();
5fdcce21 920 t = ip6_tnl_lookup(dev_net(skb->dev), skb->dev->ifindex, &ipv6h->saddr, &ipv6h->daddr);
0d3c703a
TH		 921
922 if (t) {
6aa7de05 923 u8 tproto = READ_ONCE(t->parms.proto);
0d3c703a
TH		 924
925 if (tproto != ipproto && tproto != 0)
926 goto drop;
927 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
928 goto drop;
cbb49697 929 ipv6h = ipv6_hdr(skb);
0d3c703a
TH		 930 if (!ip6_tnl_rcv_ctl(t, &ipv6h->daddr, &ipv6h->saddr))
931 goto drop;
932 if (iptunnel_pull_header(skb, 0, tpi->proto, false))
933 goto drop;
8d79266b
AS		 934 if (t->parms.collect_md) {
935 tun_dst = ipv6_tun_rx_dst(skb, 0, 0, 0);
936 if (!tun_dst)
74c4b656 937 goto drop;
8d79266b
AS		 938 }
939 ret = __ip6_tnl_rcv(t, skb, tpi, tun_dst, dscp_ecn_decapsulate,
0d3c703a 940 log_ecn_error);
1da177e4 941 }
0d3c703a 942
2922bc8a 943 rcu_read_unlock();
50fba2aa 944
0d3c703a
TH		 945 return ret;
946
947drop:
948 rcu_read_unlock();
50fba2aa
HX		 949 kfree_skb(skb);
950 return 0;
1da177e4
LT		 951}
952
c4d3efaf
YK		 953static int ip4ip6_rcv(struct sk_buff *skb)
954{
ca4aa976 955 return ipxip6_rcv(skb, IPPROTO_IPIP, &tpi_v4,
0d3c703a 956 ip4ip6_dscp_ecn_decapsulate);
c4d3efaf
YK		 957}
958
8359925b
YK		 959static int ip6ip6_rcv(struct sk_buff *skb)
960{
0d3c703a
TH		 961 return ipxip6_rcv(skb, IPPROTO_IPV6, &tpi_v6,
962 ip6ip6_dscp_ecn_decapsulate);
8359925b
YK		 963}
964
f200e98d
VF		 965static int mplsip6_rcv(struct sk_buff *skb)
966{
967 return ipxip6_rcv(skb, IPPROTO_MPLS, &tpi_mpls,
968 mplsip6_dscp_ecn_decapsulate);
969}
970
6fb32dde
VN		 971struct ipv6_tel_txoption {
972 struct ipv6_txoptions ops;
973 __u8 dst_opt[8];
974};
1da177e4 975
6fb32dde
VN		 976static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
977{
978 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 979
6fb32dde
VN		 980 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
981 opt->dst_opt[3] = 1;
982 opt->dst_opt[4] = encap_limit;
983 opt->dst_opt[5] = IPV6_TLV_PADN;
984 opt->dst_opt[6] = 1;
1da177e4 985
89a23c8b 986 opt->ops.dst1opt = (struct ipv6_opt_hdr *) opt->dst_opt;
6fb32dde 987 opt->ops.opt_nflen = 8;
1da177e4
LT		 988}
989
990/**
3144581c 991 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
1da177e4 992 * @t: the outgoing tunnel device
1ab1457c 993 * @hdr: IPv6 header from the incoming packet
1da177e4
LT		 994 *
995 * Description:
1ab1457c 996 * Avoid trivial tunneling loop by checking that tunnel exit-point
1da177e4
LT		 997 * doesn't match source of incoming packet.
998 *
1ab1457c 999 * Return:
1da177e4
LT		 1000 * 1 if conflict,
1001 * 0 else
1002 **/
1003
92113bfd 1004static inline bool
b71d1d42 1005ip6_tnl_addr_conflict(const struct ip6_tnl *t, const struct ipv6hdr *hdr)
1da177e4
LT		 1006{
1007 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
1008}
1009
d5005140
SK		 1010int ip6_tnl_xmit_ctl(struct ip6_tnl *t,
1011 const struct in6_addr *laddr,
1012 const struct in6_addr *raddr)
09c6bbf0 1013{
c12b395a 1014 struct __ip6_tnl_parm *p = &t->parms;
09c6bbf0 1015 int ret = 0;
0bd87628 1016 struct net *net = t->net;
09c6bbf0 1017
6712abc1
WT		 1018 if (t->parms.collect_md)
1019 return 1;
1020
d5005140
SK		 1021 if ((p->flags & IP6_TNL_F_CAP_XMIT) ||
1022 ((p->flags & IP6_TNL_F_CAP_PER_PACKET) &&
1023 (ip6_tnl_get_cap(t, laddr, raddr) & IP6_TNL_F_CAP_XMIT))) {
09c6bbf0
VN		 1024 struct net_device *ldev = NULL;
1025
f1a28eab 1026 rcu_read_lock();
09c6bbf0 1027 if (p->link)
f1a28eab 1028 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0 1029
232378e8
DA		 1030 if (unlikely(!ipv6_chk_addr_and_flags(net, laddr, ldev, false,
1031 0, IFA_F_TENTATIVE)))
6cee105e
IS		 1032 pr_warn_ratelimited("%s xmit: Local address not yet configured!\n",
1033 p->name);
908d140a
SL		 1034 else if (!(p->flags & IP6_TNL_F_ALLOW_LOCAL_REMOTE) &&
1035 !ipv6_addr_is_multicast(raddr) &&
232378e8
DA		 1036 unlikely(ipv6_chk_addr_and_flags(net, raddr, ldev,
1037 true, 0, IFA_F_TENTATIVE)))
6cee105e
IS		 1038 pr_warn_ratelimited("%s xmit: Routing loop! Remote address found on this node!\n",
1039 p->name);
09c6bbf0
VN		 1040 else
1041 ret = 1;
f1a28eab 1042 rcu_read_unlock();
09c6bbf0
VN		 1043 }
1044 return ret;
1045}
c12b395a 1046EXPORT_SYMBOL_GPL(ip6_tnl_xmit_ctl);
1047
1da177e4 1048/**
8eb30be0 1049 * ip6_tnl_xmit - encapsulate packet and send
1da177e4 1050 * @skb: the outgoing socket buffer
1ab1457c 1051 * @dev: the outgoing tunnel device
61ec2aec 1052 * @dsfield: dscp code for outer header
8eb30be0 1053 * @fl6: flow of tunneled packet
61ec2aec
YK		 1054 * @encap_limit: encapsulation limit
1055 * @pmtu: Path MTU is stored if packet is too big
8eb30be0 1056 * @proto: next header value
1da177e4
LT		 1057 *
1058 * Description:
1059 * Build new header and do some sanity checks on the packet before sending
1060 * it.
1061 *
1ab1457c 1062 * Return:
c4d3efaf 1063 * 0 on success
61ec2aec
YK		 1064 * -1 fail
1065 * %-EMSGSIZE message too big. return mtu in this case.
1da177e4
LT		 1066 **/
1067
8eb30be0
TH		 1068int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
1069 struct flowi6 *fl6, int encap_limit, __u32 *pmtu,
1070 __u8 proto)
1da177e4 1071{
2941a486 1072 struct ip6_tnl *t = netdev_priv(dev);
0bd87628 1073 struct net *net = t->net;
199ab00f 1074 struct ipv6hdr *ipv6h;
6fb32dde 1075 struct ipv6_tel_txoption opt;
d24f22f3 1076 struct dst_entry *dst = NULL, *ndst = NULL;
1da177e4
LT		 1077 struct net_device *tdev;
1078 int mtu;
d41bb33b 1079 unsigned int eth_hlen = t->dev->type == ARPHRD_ETHER ? ETH_HLEN : 0;
058214a4
TH		 1080 unsigned int psh_hlen = sizeof(struct ipv6hdr) + t->encap_hlen;
1081 unsigned int max_headroom = psh_hlen;
b09ab9c9 1082 __be16 payload_protocol;
b5c2d495 1083 bool use_cache = false;
8d79266b 1084 u8 hop_limit;
61ec2aec 1085 int err = -1;
1da177e4 1086
b09ab9c9
MM		 1087 payload_protocol = skb_protocol(skb, true);
1088
8d79266b
AS		 1089 if (t->parms.collect_md) {
1090 hop_limit = skb_tunnel_info(skb)->key.ttl;
1091 goto route_lookup;
1092 } else {
1093 hop_limit = t->parms.hop_limit;
1094 }
1095
ea3dc960
SK		 1096 /* NBMA tunnel */
1097 if (ipv6_addr_any(&t->parms.raddr)) {
b09ab9c9 1098 if (payload_protocol == htons(ETH_P_IPV6)) {
199ab00f
WC		 1099 struct in6_addr *addr6;
1100 struct neighbour *neigh;
1101 int addr_type;
ea3dc960 1102
199ab00f
WC		 1103 if (!skb_dst(skb))
1104 goto tx_err_link_failure;
ea3dc960 1105
199ab00f
WC		 1106 neigh = dst_neigh_lookup(skb_dst(skb),
1107 &ipv6_hdr(skb)->daddr);
1108 if (!neigh)
1109 goto tx_err_link_failure;
ea3dc960 1110
199ab00f
WC		 1111 addr6 = (struct in6_addr *)&neigh->primary_key;
1112 addr_type = ipv6_addr_type(addr6);
ea3dc960 1113
199ab00f
WC		 1114 if (addr_type == IPV6_ADDR_ANY)
1115 addr6 = &ipv6_hdr(skb)->daddr;
ea3dc960 1116
199ab00f
WC		 1117 memcpy(&fl6->daddr, addr6, sizeof(fl6->daddr));
1118 neigh_release(neigh);
b09ab9c9 1119 } else if (payload_protocol == htons(ETH_P_IP)) {
3a5f238f
ED		 1120 const struct rtable *rt = skb_rtable(skb);
1121
1122 if (!rt)
1123 goto tx_err_link_failure;
c1f55c5e
QD		 1124
1125 if (rt->rt_gw_family == AF_INET6)
1126 memcpy(&fl6->daddr, &rt->rt_gw6, sizeof(fl6->daddr));
199ab00f 1127 }
23263ec8
EC		 1128 } else if (t->parms.proto != 0 && !(t->parms.flags &
1129 (IP6_TNL_F_USE_ORIG_TCLASS |
1130 IP6_TNL_F_USE_ORIG_FWMARK))) {
1131 /* enable the cache only if neither the outer protocol nor the
1132 * routing decision depends on the current inner header value
b5c2d495
PA		 1133 */
1134 use_cache = true;
1135 }
1136
1137 if (use_cache)
607f725f 1138 dst = dst_cache_get(&t->dst_cache);
d5005140
SK		 1139
1140 if (!ip6_tnl_xmit_ctl(t, &fl6->saddr, &fl6->daddr))
1141 goto tx_err_link_failure;
1142
89b02126 1143 if (!dst) {
8d79266b 1144route_lookup:
5f733ee6
LM		 1145 /* add dsfield to flowlabel for route lookup */
1146 fl6->flowlabel = ip6_make_flowinfo(dsfield, fl6->flowlabel);
1147
cdf3464e 1148 dst = ip6_route_output(net, NULL, fl6);
1da177e4 1149
cdf3464e 1150 if (dst->error)
a57ebc90 1151 goto tx_err_link_failure;
cdf3464e
MKL		 1152 dst = xfrm_lookup(net, dst, flowi6_to_flowi(fl6), NULL, 0);
1153 if (IS_ERR(dst)) {
1154 err = PTR_ERR(dst);
1155 dst = NULL;
452edd59
DM		 1156 goto tx_err_link_failure;
1157 }
3789caba 1158 if (t->parms.collect_md && ipv6_addr_any(&fl6->saddr) &&
8d79266b
AS		 1159 ipv6_dev_get_saddr(net, ip6_dst_idev(dst)->dev,
1160 &fl6->daddr, 0, &fl6->saddr))
1161 goto tx_err_link_failure;
cdf3464e 1162 ndst = dst;
a57ebc90 1163 }
1da177e4
LT		 1164
1165 tdev = dst->dev;
1166
1167 if (tdev == dev) {
2fad1ba3 1168 DEV_STATS_INC(dev, collisions);
e87cc472
JP		 1169 net_warn_ratelimited("%s: Local routing loop detected!\n",
1170 t->parms.name);
1da177e4
LT		 1171 goto tx_err_dst_release;
1172 }
d41bb33b 1173 mtu = dst_mtu(dst) - eth_hlen - psh_hlen - t->tun_hlen;
6fb32dde 1174 if (encap_limit >= 0) {
1da177e4
LT		 1175 max_headroom += 8;
1176 mtu -= 8;
1177 }
82a40777
XL		 1178 mtu = max(mtu, skb->protocol == htons(ETH_P_IPV6) ?
1179 IPV6_MIN_MTU : IPV4_MIN_MTU);
c9fefa08 1180
7a1592bc 1181 skb_dst_update_pmtu_no_confirm(skb, mtu);
d41bb33b 1182 if (skb->len - t->tun_hlen - eth_hlen > mtu && !skb_is_gso(skb)) {
61ec2aec
YK		 1183 *pmtu = mtu;
1184 err = -EMSGSIZE;
1da177e4
LT		 1185 goto tx_err_dst_release;
1186 }
1187
8eb30be0
TH		 1188 if (t->err_count > 0) {
1189 if (time_before(jiffies,
1190 t->err_time + IP6TUNNEL_ERR_TIMEO)) {
1191 t->err_count--;
1192
1193 dst_link_failure(skb);
1194 } else {
1195 t->err_count = 0;
1196 }
1197 }
1198
963a88b3 1199 skb_scrub_packet(skb, !net_eq(t->net, dev_net(dev)));
0bd87628 1200
1da177e4
LT		 1201 /*
1202 * Okay, now see if we can stuff it in the buffer as-is.
1203 */
1204 max_headroom += LL_RESERVED_SPACE(tdev);
1ab1457c 1205
cfbba49d
PM		 1206 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
1207 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1da177e4 1208 struct sk_buff *new_skb;
1ab1457c 1209
e5d08d71
IM		 1210 new_skb = skb_realloc_headroom(skb, max_headroom);
1211 if (!new_skb)
1da177e4
LT		 1212 goto tx_err_dst_release;
1213
1214 if (skb->sk)
1215 skb_set_owner_w(new_skb, skb->sk);
9ff26449 1216 consume_skb(skb);
1da177e4
LT		 1217 skb = new_skb;
1218 }
cdf3464e 1219
8d79266b
AS		 1220 if (t->parms.collect_md) {
1221 if (t->encap.type != TUNNEL_ENCAP_NONE)
1222 goto tx_err_dst_release;
1223 } else {
b5c2d495 1224 if (use_cache && ndst)
8d79266b
AS		 1225 dst_cache_set_ip6(&t->dst_cache, ndst, &fl6->saddr);
1226 }
cdf3464e
MKL		 1227 skb_dst_set(skb, dst);
1228
36feaac3 1229 if (hop_limit == 0) {
b09ab9c9 1230 if (payload_protocol == htons(ETH_P_IP))
36feaac3 1231 hop_limit = ip_hdr(skb)->ttl;
b09ab9c9 1232 else if (payload_protocol == htons(ETH_P_IPV6))
36feaac3
HL		 1233 hop_limit = ipv6_hdr(skb)->hop_limit;
1234 else
1235 hop_limit = ip6_dst_hoplimit(dst);
1236 }
3d483058 1237
058214a4
TH		 1238 /* Calculate max headroom for all the headers and adjust
1239 * needed_headroom if necessary.
1240 */
8eb30be0 1241 max_headroom = LL_RESERVED_SPACE(dst->dev) + sizeof(struct ipv6hdr)
058214a4 1242 + dst->header_len + t->hlen;
4b397c06
ED		 1243 if (max_headroom > READ_ONCE(dev->needed_headroom))
1244 WRITE_ONCE(dev->needed_headroom, max_headroom);
8eb30be0 1245
058214a4
TH		 1246 err = ip6_tnl_encap(skb, t, &proto, fl6);
1247 if (err)
1248 return err;
1249
d4d576f5
SB		 1250 if (encap_limit >= 0) {
1251 init_tel_txopt(&opt, encap_limit);
1252 ipv6_push_frag_opts(skb, &opt.ops, &proto);
1253 }
1254
e2d1bca7
ACM		 1255 skb_push(skb, sizeof(struct ipv6hdr));
1256 skb_reset_network_header(skb);
0660e03f 1257 ipv6h = ipv6_hdr(skb);
0e9a7095 1258 ip6_flow_hdr(ipv6h, dsfield,
42240901 1259 ip6_make_flowlabel(net, skb, fl6->flowlabel, true, fl6));
8d79266b 1260 ipv6h->hop_limit = hop_limit;
1da177e4 1261 ipv6h->nexthdr = proto;
4e3fd7a0
AD		 1262 ipv6h->saddr = fl6->saddr;
1263 ipv6h->daddr = fl6->daddr;
79b16aad 1264 ip6tunnel_xmit(NULL, skb, dev);
1da177e4
LT		 1265 return 0;
1266tx_err_link_failure:
2fad1ba3 1267 DEV_STATS_INC(dev, tx_carrier_errors);
1da177e4
LT		 1268 dst_link_failure(skb);
1269tx_err_dst_release:
cdf3464e 1270 dst_release(dst);
61ec2aec
YK		 1271 return err;
1272}
8eb30be0 1273EXPORT_SYMBOL(ip6_tnl_xmit);
61ec2aec 1274
c4d3efaf 1275static inline int
e7bb18e6
VF		 1276ipxip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev,
1277 u8 protocol)
c4d3efaf
YK		 1278{
1279 struct ip6_tnl *t = netdev_priv(dev);
e7bb18e6 1280 struct ipv6hdr *ipv6h;
76c0ddd8 1281 const struct iphdr *iph;
c4d3efaf 1282 int encap_limit = -1;
e7bb18e6 1283 __u16 offset;
4c9483b2 1284 struct flowi6 fl6;
e7bb18e6 1285 __u8 dsfield, orig_dsfield;
c4d3efaf 1286 __u32 mtu;
acf722f7 1287 u8 tproto;
c4d3efaf
YK		 1288 int err;
1289
6aa7de05 1290 tproto = READ_ONCE(t->parms.proto);
e7bb18e6 1291 if (tproto != protocol && tproto != 0)
c4d3efaf
YK		 1292 return -1;
1293
8d79266b
AS		 1294 if (t->parms.collect_md) {
1295 struct ip_tunnel_info *tun_info;
1296 const struct ip_tunnel_key *key;
c4d3efaf 1297
8d79266b
AS		 1298 tun_info = skb_tunnel_info(skb);
1299 if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
1300 ip_tunnel_info_af(tun_info) != AF_INET6))
1301 return -1;
1302 key = &tun_info->key;
1303 memset(&fl6, 0, sizeof(fl6));
e7bb18e6 1304 fl6.flowi6_proto = protocol;
3789caba 1305 fl6.saddr = key->u.ipv6.src;
8d79266b
AS		 1306 fl6.daddr = key->u.ipv6.dst;
1307 fl6.flowlabel = key->label;
46f8cd9d 1308 dsfield = key->tos;
e7bb18e6
VF		 1309 switch (protocol) {
1310 case IPPROTO_IPIP:
1311 iph = ip_hdr(skb);
1312 orig_dsfield = ipv4_get_dsfield(iph);
1313 break;
1314 case IPPROTO_IPV6:
1315 ipv6h = ipv6_hdr(skb);
1316 orig_dsfield = ipv6_get_dsfield(ipv6h);
1317 break;
1318 default:
1319 orig_dsfield = dsfield;
1320 break;
1321 }
8d79266b
AS		 1322 } else {
1323 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1324 encap_limit = t->parms.encap_limit;
e7bb18e6
VF		 1325 if (protocol == IPPROTO_IPV6) {
1326 offset = ip6_tnl_parse_tlv_enc_lim(skb,
1327 skb_network_header(skb));
1328 /* ip6_tnl_parse_tlv_enc_lim() might have
1329 * reallocated skb->head
1330 */
1331 if (offset > 0) {
1332 struct ipv6_tlv_tnl_enc_lim *tel;
c4d3efaf 1333
e7bb18e6
VF		 1334 tel = (void *)&skb_network_header(skb)[offset];
1335 if (tel->encap_limit == 0) {
4372339e
JD		 1336 icmpv6_ndo_send(skb, ICMPV6_PARAMPROB,
1337 ICMPV6_HDR_FIELD, offset + 2);
e7bb18e6
VF		 1338 return -1;
1339 }
1340 encap_limit = tel->encap_limit - 1;
8d79266b 1341 }
61ec2aec 1342 }
61ec2aec 1343
8d79266b 1344 memcpy(&fl6, &t->fl.u.ip6, sizeof(fl6));
e7bb18e6 1345 fl6.flowi6_proto = protocol;
61ec2aec 1346
8d79266b
AS		 1347 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
1348 fl6.flowi6_mark = skb->mark;
0a473b82
CG		 1349 else
1350 fl6.flowi6_mark = t->parms.fwmark;
e7bb18e6
VF		 1351 switch (protocol) {
1352 case IPPROTO_IPIP:
1353 iph = ip_hdr(skb);
1354 orig_dsfield = ipv4_get_dsfield(iph);
1355 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
1356 dsfield = orig_dsfield;
1357 else
1358 dsfield = ip6_tclass(t->parms.flowinfo);
1359 break;
1360 case IPPROTO_IPV6:
1361 ipv6h = ipv6_hdr(skb);
1362 orig_dsfield = ipv6_get_dsfield(ipv6h);
1363 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
1364 dsfield = orig_dsfield;
1365 else
1366 dsfield = ip6_tclass(t->parms.flowinfo);
1367 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL)
1368 fl6.flowlabel |= ip6_flowlabel(ipv6h);
1369 break;
1370 default:
6c11fbf9 1371 orig_dsfield = dsfield = ip6_tclass(t->parms.flowinfo);
e7bb18e6
VF		 1372 break;
1373 }
8d79266b 1374 }
61ec2aec 1375
e2d118a1 1376 fl6.flowi6_uid = sock_net_uid(dev_net(dev), NULL);
e7bb18e6 1377 dsfield = INET_ECN_encapsulate(dsfield, orig_dsfield);
e2d118a1 1378
815d22e5
TH		 1379 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6))
1380 return -1;
1381
a6e3f298
JK		 1382 skb_set_inner_ipproto(skb, protocol);
1383
8eb30be0 1384 err = ip6_tnl_xmit(skb, dev, dsfield, &fl6, encap_limit, &mtu,
e7bb18e6 1385 protocol);
61ec2aec 1386 if (err != 0) {
e7bb18e6 1387 /* XXX: send ICMP error even if DF is not set. */
61ec2aec 1388 if (err == -EMSGSIZE)
e7bb18e6
VF		 1389 switch (protocol) {
1390 case IPPROTO_IPIP:
4372339e
JD		 1391 icmp_ndo_send(skb, ICMP_DEST_UNREACH,
1392 ICMP_FRAG_NEEDED, htonl(mtu));
e7bb18e6
VF		 1393 break;
1394 case IPPROTO_IPV6:
4372339e 1395 icmpv6_ndo_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
e7bb18e6
VF		 1396 break;
1397 default:
1398 break;
1399 }
61ec2aec
YK		 1400 return -1;
1401 }
1402
1403 return 0;
1404}
1405
6fef4c0c 1406static netdev_tx_t
8eb30be0 1407ip6_tnl_start_xmit(struct sk_buff *skb, struct net_device *dev)
61ec2aec
YK		 1408{
1409 struct ip6_tnl *t = netdev_priv(dev);
e7bb18e6 1410 u8 ipproto;
61ec2aec
YK		 1411 int ret;
1412
cb9f1b78
WB		 1413 if (!pskb_inet_may_pull(skb))
1414 goto tx_err;
1415
61ec2aec 1416 switch (skb->protocol) {
60678040 1417 case htons(ETH_P_IP):
e7bb18e6 1418 ipproto = IPPROTO_IPIP;
c4d3efaf 1419 break;
60678040 1420 case htons(ETH_P_IPV6):
e7bb18e6
VF		 1421 if (ip6_tnl_addr_conflict(t, ipv6_hdr(skb)))
1422 goto tx_err;
1423 ipproto = IPPROTO_IPV6;
61ec2aec 1424 break;
6c11fbf9
VF		 1425 case htons(ETH_P_MPLS_UC):
1426 ipproto = IPPROTO_MPLS;
1427 break;
61ec2aec
YK		 1428 default:
1429 goto tx_err;
1430 }
1431
e7bb18e6 1432 ret = ipxip6_tnl_xmit(skb, dev, ipproto);
61ec2aec
YK		 1433 if (ret < 0)
1434 goto tx_err;
1435
6ed10654 1436 return NETDEV_TX_OK;
61ec2aec 1437
1da177e4 1438tx_err:
2fad1ba3
ED		 1439 DEV_STATS_INC(dev, tx_errors);
1440 DEV_STATS_INC(dev, tx_dropped);
1da177e4 1441 kfree_skb(skb);
6ed10654 1442 return NETDEV_TX_OK;
1da177e4
LT		 1443}
1444
3144581c 1445static void ip6_tnl_link_config(struct ip6_tnl *t)
1da177e4
LT		 1446{
1447 struct net_device *dev = t->dev;
5fdcce21 1448 struct net_device *tdev = NULL;
c12b395a 1449 struct __ip6_tnl_parm *p = &t->parms;
4c9483b2 1450 struct flowi6 *fl6 = &t->fl.u.ip6;
058214a4 1451 int t_hlen;
d89d7ff0 1452 int mtu;
1da177e4 1453
5a1b7e1a 1454 __dev_addr_set(dev, &p->laddr, sizeof(struct in6_addr));
3a6d54c5 1455 memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1da177e4
LT		 1456
1457 /* Set up flowi template */
4e3fd7a0
AD		 1458 fl6->saddr = p->laddr;
1459 fl6->daddr = p->raddr;
4c9483b2
DM		 1460 fl6->flowi6_oif = p->link;
1461 fl6->flowlabel = 0;
1da177e4
LT		 1462
1463 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
4c9483b2 1464 fl6->flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1da177e4 1465 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
4c9483b2 1466 fl6->flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1da177e4 1467
d0087b29
VN		 1468 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV|IP6_TNL_F_CAP_PER_PACKET);
1469 p->flags |= ip6_tnl_get_cap(t, &p->laddr, &p->raddr);
1da177e4
LT		 1470
1471 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1472 dev->flags |= IFF_POINTOPOINT;
1473 else
1474 dev->flags &= ~IFF_POINTOPOINT;
1475
058214a4
TH		 1476 t->tun_hlen = 0;
1477 t->hlen = t->encap_hlen + t->tun_hlen;
1478 t_hlen = t->hlen + sizeof(struct ipv6hdr);
1479
1da177e4 1480 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN		 1481 int strict = (ipv6_addr_type(&p->raddr) &
1482 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1483
0bd87628 1484 struct rt6_info *rt = rt6_lookup(t->net,
2f7f54b7 1485 &p->raddr, &p->laddr,
b75cc8f9 1486 p->link, NULL, strict);
5fdcce21
WD		 1487 if (rt) {
1488 tdev = rt->dst.dev;
1489 ip6_rt_put(rt);
1490 }
1da177e4 1491
5fdcce21
WD		 1492 if (!tdev && p->link)
1493 tdev = __dev_get_by_index(t->net, p->link);
1da177e4 1494
5fdcce21
WD		 1495 if (tdev) {
1496 dev->hard_header_len = tdev->hard_header_len + t_hlen;
1497 mtu = min_t(unsigned int, tdev->mtu, IP6_MAX_MTU);
1da177e4 1498
d89d7ff0 1499 mtu = mtu - t_hlen;
381601e5 1500 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
d89d7ff0 1501 mtu -= 8;
1da177e4 1502
d89d7ff0
ED		 1503 if (mtu < IPV6_MIN_MTU)
1504 mtu = IPV6_MIN_MTU;
1505 WRITE_ONCE(dev->mtu, mtu);
1da177e4 1506 }
1da177e4
LT		 1507 }
1508}
1509
1510/**
3144581c 1511 * ip6_tnl_change - update the tunnel parameters
1da177e4
LT		 1512 * @t: tunnel to be changed
1513 * @p: tunnel configuration parameters
1da177e4
LT		 1514 *
1515 * Description:
3144581c 1516 * ip6_tnl_change() updates the tunnel parameters
1da177e4
LT		 1517 **/
1518
77788567 1519static void
c12b395a 1520ip6_tnl_change(struct ip6_tnl *t, const struct __ip6_tnl_parm *p)
1da177e4 1521{
4e3fd7a0
AD		 1522 t->parms.laddr = p->laddr;
1523 t->parms.raddr = p->raddr;
1da177e4
LT		 1524 t->parms.flags = p->flags;
1525 t->parms.hop_limit = p->hop_limit;
1526 t->parms.encap_limit = p->encap_limit;
1527 t->parms.flowinfo = p->flowinfo;
8181b8c1 1528 t->parms.link = p->link;
502b0935 1529 t->parms.proto = p->proto;
0a473b82 1530 t->parms.fwmark = p->fwmark;
607f725f 1531 dst_cache_reset(&t->dst_cache);
3144581c 1532 ip6_tnl_link_config(t);
1da177e4
LT		 1533}
1534
77788567 1535static void ip6_tnl_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
0b112457 1536{
0bd87628 1537 struct net *net = t->net;
0b112457 1538 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
0b112457
ND		 1539
1540 ip6_tnl_unlink(ip6n, t);
1541 synchronize_net();
77788567 1542 ip6_tnl_change(t, p);
0b112457
ND		 1543 ip6_tnl_link(ip6n, t);
1544 netdev_state_change(t->dev);
0b112457
ND		 1545}
1546
77788567 1547static void ip6_tnl0_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
acf722f7
AA		 1548{
1549 /* for default tnl0 device allow to change only the proto */
1550 t->parms.proto = p->proto;
1551 netdev_state_change(t->dev);
acf722f7
AA		 1552}
1553
c12b395a 1554static void
1555ip6_tnl_parm_from_user(struct __ip6_tnl_parm *p, const struct ip6_tnl_parm *u)
1556{
1557 p->laddr = u->laddr;
1558 p->raddr = u->raddr;
1559 p->flags = u->flags;
1560 p->hop_limit = u->hop_limit;
1561 p->encap_limit = u->encap_limit;
1562 p->flowinfo = u->flowinfo;
1563 p->link = u->link;
1564 p->proto = u->proto;
1565 memcpy(p->name, u->name, sizeof(u->name));
1566}
1567
1568static void
1569ip6_tnl_parm_to_user(struct ip6_tnl_parm *u, const struct __ip6_tnl_parm *p)
1570{
1571 u->laddr = p->laddr;
1572 u->raddr = p->raddr;
1573 u->flags = p->flags;
1574 u->hop_limit = p->hop_limit;
1575 u->encap_limit = p->encap_limit;
1576 u->flowinfo = p->flowinfo;
1577 u->link = p->link;
1578 u->proto = p->proto;
1579 memcpy(u->name, p->name, sizeof(u->name));
1580}
1581
1da177e4 1582/**
3e7a1c7c 1583 * ip6_tnl_siocdevprivate - configure ipv6 tunnels from userspace
1da177e4 1584 * @dev: virtual device associated with tunnel
3e7a1c7c
AB		 1585 * @ifr: unused
1586 * @data: parameters passed from userspace
1da177e4
LT		 1587 * @cmd: command to be performed
1588 *
1589 * Description:
3144581c 1590 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1ab1457c 1591 * from userspace.
1da177e4
LT		 1592 *
1593 * The possible commands are the following:
1594 * %SIOCGETTUNNEL: get tunnel parameters for device
1595 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1596 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1597 * %SIOCDELTUNNEL: delete tunnel
1598 *
1ab1457c 1599 * The fallback device "ip6tnl0", created during module
1da177e4
LT		 1600 * initialization, can be used for creating other tunnel devices.
1601 *
1602 * Return:
1603 * 0 on success,
1604 * %-EFAULT if unable to copy data to or from userspace,
1605 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1606 * %-EINVAL if passed tunnel parameters are invalid,
1607 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1608 * %-ENODEV if attempting to change or delete a nonexisting device
1609 **/
1610
1611static int
3e7a1c7c
AB		 1612ip6_tnl_siocdevprivate(struct net_device *dev, struct ifreq *ifr,
1613 void __user *data, int cmd)
1da177e4
LT		 1614{
1615 int err = 0;
1da177e4 1616 struct ip6_tnl_parm p;
c12b395a 1617 struct __ip6_tnl_parm p1;
74462f0d
ND		 1618 struct ip6_tnl *t = netdev_priv(dev);
1619 struct net *net = t->net;
2dd02c89 1620 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 1621
0d3c703a
TH		 1622 memset(&p1, 0, sizeof(p1));
1623
1da177e4
LT		 1624 switch (cmd) {
1625 case SIOCGETTUNNEL:
15820e12 1626 if (dev == ip6n->fb_tnl_dev) {
3e7a1c7c 1627 if (copy_from_user(&p, data, sizeof(p))) {
1da177e4
LT		 1628 err = -EFAULT;
1629 break;
1630 }
c12b395a 1631 ip6_tnl_parm_from_user(&p1, &p);
1632 t = ip6_tnl_locate(net, &p1, 0);
37355565 1633 if (IS_ERR(t))
74462f0d 1634 t = netdev_priv(dev);
5ef5d6c5
DC		 1635 } else {
1636 memset(&p, 0, sizeof(p));
567131a7 1637 }
c12b395a 1638 ip6_tnl_parm_to_user(&p, &t->parms);
3e7a1c7c 1639 if (copy_to_user(data, &p, sizeof(p)))
1da177e4 1640 err = -EFAULT;
1da177e4
LT		 1641 break;
1642 case SIOCADDTUNNEL:
1643 case SIOCCHGTUNNEL:
1644 err = -EPERM;
af31f412 1645 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4 1646 break;
567131a7 1647 err = -EFAULT;
3e7a1c7c 1648 if (copy_from_user(&p, data, sizeof(p)))
1da177e4 1649 break;
567131a7 1650 err = -EINVAL;
502b0935
YK		 1651 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1652 p.proto != 0)
1da177e4 1653 break;
c12b395a 1654 ip6_tnl_parm_from_user(&p1, &p);
1655 t = ip6_tnl_locate(net, &p1, cmd == SIOCADDTUNNEL);
acf722f7 1656 if (cmd == SIOCCHGTUNNEL) {
37355565 1657 if (!IS_ERR(t)) {
567131a7
VN		 1658 if (t->dev != dev) {
1659 err = -EEXIST;
1660 break;
1661 }
1662 } else
1663 t = netdev_priv(dev);
acf722f7 1664 if (dev == ip6n->fb_tnl_dev)
77788567 1665 ip6_tnl0_update(t, &p1);
acf722f7 1666 else
77788567 1667 ip6_tnl_update(t, &p1);
1da177e4 1668 }
37355565 1669 if (!IS_ERR(t)) {
1da177e4 1670 err = 0;
c12b395a 1671 ip6_tnl_parm_to_user(&p, &t->parms);
3e7a1c7c 1672 if (copy_to_user(data, &p, sizeof(p)))
567131a7
VN		 1673 err = -EFAULT;
1674
37355565
ND		 1675 } else {
1676 err = PTR_ERR(t);
1677 }
1da177e4
LT		 1678 break;
1679 case SIOCDELTUNNEL:
1680 err = -EPERM;
af31f412 1681 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 1682 break;
1683
15820e12 1684 if (dev == ip6n->fb_tnl_dev) {
567131a7 1685 err = -EFAULT;
3e7a1c7c 1686 if (copy_from_user(&p, data, sizeof(p)))
1da177e4 1687 break;
567131a7 1688 err = -ENOENT;
c12b395a 1689 ip6_tnl_parm_from_user(&p1, &p);
1690 t = ip6_tnl_locate(net, &p1, 0);
37355565 1691 if (IS_ERR(t))
1da177e4 1692 break;
567131a7 1693 err = -EPERM;
15820e12 1694 if (t->dev == ip6n->fb_tnl_dev)
1da177e4 1695 break;
567131a7 1696 dev = t->dev;
1da177e4 1697 }
22f8cde5
SH		 1698 err = 0;
1699 unregister_netdevice(dev);
1da177e4
LT		 1700 break;
1701 default:
1702 err = -EINVAL;
1703 }
1704 return err;
1705}
1706
1da177e4 1707/**
3144581c 1708 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1da177e4
LT		 1709 * @dev: virtual device associated with tunnel
1710 * @new_mtu: the new mtu
1711 *
1712 * Return:
1713 * 0 on success,
1714 * %-EINVAL if mtu too small
1715 **/
1716
79ecb90e 1717int ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1da177e4 1718{
582442d6
OG		 1719 struct ip6_tnl *tnl = netdev_priv(dev);
1720
2fa771be
XL		 1721 if (tnl->parms.proto == IPPROTO_IPV6) {
1722 if (new_mtu < IPV6_MIN_MTU)
582442d6
OG		 1723 return -EINVAL;
1724 } else {
2fa771be 1725 if (new_mtu < ETH_MIN_MTU)
582442d6 1726 return -EINVAL;
1da177e4 1727 }
f7ff1fde
ND		 1728 if (tnl->parms.proto == IPPROTO_IPV6 || tnl->parms.proto == 0) {
1729 if (new_mtu > IP6_MAX_MTU - dev->hard_header_len)
1730 return -EINVAL;
1731 } else {
1732 if (new_mtu > IP_MAX_MTU - dev->hard_header_len)
1733 return -EINVAL;
1734 }
1da177e4
LT		 1735 dev->mtu = new_mtu;
1736 return 0;
1737}
79ecb90e 1738EXPORT_SYMBOL(ip6_tnl_change_mtu);
1da177e4 1739
ecf2c06a
ND		 1740int ip6_tnl_get_iflink(const struct net_device *dev)
1741{
1742 struct ip6_tnl *t = netdev_priv(dev);
1743
1744 return t->parms.link;
1745}
1746EXPORT_SYMBOL(ip6_tnl_get_iflink);
1326c3d5 1747
058214a4
TH		 1748int ip6_tnl_encap_add_ops(const struct ip6_tnl_encap_ops *ops,
1749 unsigned int num)
1750{
1751 if (num >= MAX_IPTUN_ENCAP_OPS)
1752 return -ERANGE;
1753
1754 return !cmpxchg((const struct ip6_tnl_encap_ops **)
1755 &ip6tun_encaps[num],
1756 NULL, ops) ? 0 : -1;
1757}
1758EXPORT_SYMBOL(ip6_tnl_encap_add_ops);
1759
1760int ip6_tnl_encap_del_ops(const struct ip6_tnl_encap_ops *ops,
1761 unsigned int num)
1762{
1763 int ret;
1764
1765 if (num >= MAX_IPTUN_ENCAP_OPS)
1766 return -ERANGE;
1767
1768 ret = (cmpxchg((const struct ip6_tnl_encap_ops **)
1769 &ip6tun_encaps[num],
1770 ops, NULL) == ops) ? 0 : -1;
1771
1772 synchronize_net();
1773
1774 return ret;
1775}
1776EXPORT_SYMBOL(ip6_tnl_encap_del_ops);
1777
1778int ip6_tnl_encap_setup(struct ip6_tnl *t,
1779 struct ip_tunnel_encap *ipencap)
1780{
1781 int hlen;
1782
1783 memset(&t->encap, 0, sizeof(t->encap));
1784
1785 hlen = ip6_encap_hlen(ipencap);
1786 if (hlen < 0)
1787 return hlen;
1788
1789 t->encap.type = ipencap->type;
1790 t->encap.sport = ipencap->sport;
1791 t->encap.dport = ipencap->dport;
1792 t->encap.flags = ipencap->flags;
1793
1794 t->encap_hlen = hlen;
1795 t->hlen = t->encap_hlen + t->tun_hlen;
1796
1797 return 0;
1798}
1799EXPORT_SYMBOL_GPL(ip6_tnl_encap_setup);
1800
1326c3d5 1801static const struct net_device_ops ip6_tnl_netdev_ops = {
6c6151da 1802 .ndo_init = ip6_tnl_dev_init,
8560f226 1803 .ndo_uninit = ip6_tnl_dev_uninit,
8eb30be0 1804 .ndo_start_xmit = ip6_tnl_start_xmit,
3e7a1c7c 1805 .ndo_siocdevprivate = ip6_tnl_siocdevprivate,
1326c3d5 1806 .ndo_change_mtu = ip6_tnl_change_mtu,
6b840a04 1807 .ndo_get_stats64 = dev_get_tstats64,
ecf2c06a 1808 .ndo_get_iflink = ip6_tnl_get_iflink,
1326c3d5
SH		 1809};
1810
51c052d4
TH		 1811#define IPXIPX_FEATURES (NETIF_F_SG | \
1812 NETIF_F_FRAGLIST | \
1813 NETIF_F_HIGHDMA | \
1814 NETIF_F_GSO_SOFTWARE | \
1815 NETIF_F_HW_CSUM)
8560f226 1816
1da177e4 1817/**
3144581c 1818 * ip6_tnl_dev_setup - setup virtual tunnel device
1da177e4
LT		 1819 * @dev: virtual device associated with tunnel
1820 *
1821 * Description:
1822 * Initialize function pointers and device parameters
1823 **/
1824
3144581c 1825static void ip6_tnl_dev_setup(struct net_device *dev)
1da177e4 1826{
1326c3d5 1827 dev->netdev_ops = &ip6_tnl_netdev_ops;
e53ac932 1828 dev->header_ops = &ip_tunnel_header_ops;
cf124db5
DM		 1829 dev->needs_free_netdev = true;
1830 dev->priv_destructor = ip6_dev_free;
1da177e4
LT		 1831
1832 dev->type = ARPHRD_TUNNEL6;
1da177e4
LT		 1833 dev->flags |= IFF_NOARP;
1834 dev->addr_len = sizeof(struct in6_addr);
058214a4 1835 dev->features |= NETIF_F_LLTX;
02875878 1836 netif_keep_dst(dev);
51c052d4
TH		 1837
1838 dev->features |= IPXIPX_FEATURES;
1839 dev->hw_features |= IPXIPX_FEATURES;
1840
e837735e
ND		 1841 /* This perm addr will be used as interface identifier by IPv6 */
1842 dev->addr_assign_type = NET_ADDR_RANDOM;
1843 eth_random_addr(dev->perm_addr);
1da177e4
LT		 1844}
1845
1846
1847/**
3144581c 1848 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1da177e4
LT		 1849 * @dev: virtual device associated with tunnel
1850 **/
1851
8560f226 1852static inline int
3144581c 1853ip6_tnl_dev_init_gen(struct net_device *dev)
1da177e4 1854{
2941a486 1855 struct ip6_tnl *t = netdev_priv(dev);
cdf3464e 1856 int ret;
058214a4 1857 int t_hlen;
8560f226 1858
1da177e4 1859 t->dev = dev;
0bd87628 1860 t->net = dev_net(dev);
1c213bd2 1861 dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
8560f226
ED		 1862 if (!dev->tstats)
1863 return -ENOMEM;
cdf3464e 1864
607f725f 1865 ret = dst_cache_init(&t->dst_cache, GFP_KERNEL);
0d3c703a
TH		 1866 if (ret)
1867 goto free_stats;
1868
1869 ret = gro_cells_init(&t->gro_cells, dev);
1870 if (ret)
1871 goto destroy_dst;
cdf3464e 1872
79ecb90e 1873 t->tun_hlen = 0;
058214a4
TH		 1874 t->hlen = t->encap_hlen + t->tun_hlen;
1875 t_hlen = t->hlen + sizeof(struct ipv6hdr);
1876
1877 dev->type = ARPHRD_TUNNEL6;
1878 dev->hard_header_len = LL_MAX_HEADER + t_hlen;
1879 dev->mtu = ETH_DATA_LEN - t_hlen;
1880 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1881 dev->mtu -= 8;
b96f9afe 1882 dev->min_mtu = ETH_MIN_MTU;
f7ff1fde 1883 dev->max_mtu = IP6_MAX_MTU - dev->hard_header_len;
79ecb90e 1884
d62607c3 1885 netdev_hold(dev, &t->dev_tracker, GFP_KERNEL);
8560f226 1886 return 0;
0d3c703a
TH		 1887
1888destroy_dst:
1889 dst_cache_destroy(&t->dst_cache);
1890free_stats:
1891 free_percpu(dev->tstats);
1892 dev->tstats = NULL;
1893
1894 return ret;
1da177e4
LT		 1895}
1896
1897/**
3144581c 1898 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1da177e4
LT		 1899 * @dev: virtual device associated with tunnel
1900 **/
1901
8560f226 1902static int ip6_tnl_dev_init(struct net_device *dev)
1da177e4 1903{
2941a486 1904 struct ip6_tnl *t = netdev_priv(dev);
8560f226
ED		 1905 int err = ip6_tnl_dev_init_gen(dev);
1906
1907 if (err)
1908 return err;
3144581c 1909 ip6_tnl_link_config(t);
5311a69a 1910 if (t->parms.collect_md)
8d79266b 1911 netif_keep_dst(dev);
8560f226 1912 return 0;
1da177e4
LT		 1913}
1914
1915/**
3144581c 1916 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1da177e4
LT		 1917 * @dev: fallback device
1918 *
1919 * Return: 0
1920 **/
1921
8560f226 1922static int __net_init ip6_fb_tnl_dev_init(struct net_device *dev)
1da177e4 1923{
2941a486 1924 struct ip6_tnl *t = netdev_priv(dev);
3e6c9fb5
PE		 1925 struct net *net = dev_net(dev);
1926 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1927
502b0935 1928 t->parms.proto = IPPROTO_IPV6;
d0087b29 1929
cf778b00 1930 rcu_assign_pointer(ip6n->tnls_wc[0], t);
8560f226 1931 return 0;
1da177e4
LT		 1932}
1933
a8b8a889
MS		 1934static int ip6_tnl_validate(struct nlattr *tb[], struct nlattr *data[],
1935 struct netlink_ext_ack *extack)
0b112457
ND		 1936{
1937 u8 proto;
1938
c8965932 1939 if (!data || !data[IFLA_IPTUN_PROTO])
0b112457
ND		 1940 return 0;
1941
1942 proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1943 if (proto != IPPROTO_IPV6 &&
1944 proto != IPPROTO_IPIP &&
1945 proto != 0)
1946 return -EINVAL;
1947
1948 return 0;
1949}
1950
1951static void ip6_tnl_netlink_parms(struct nlattr *data[],
1952 struct __ip6_tnl_parm *parms)
1953{
1954 memset(parms, 0, sizeof(*parms));
1955
1956 if (!data)
1957 return;
1958
1959 if (data[IFLA_IPTUN_LINK])
1960 parms->link = nla_get_u32(data[IFLA_IPTUN_LINK]);
1961
1962 if (data[IFLA_IPTUN_LOCAL])
67b61f6c 1963 parms->laddr = nla_get_in6_addr(data[IFLA_IPTUN_LOCAL]);
0b112457
ND		 1964
1965 if (data[IFLA_IPTUN_REMOTE])
67b61f6c 1966 parms->raddr = nla_get_in6_addr(data[IFLA_IPTUN_REMOTE]);
0b112457
ND		 1967
1968 if (data[IFLA_IPTUN_TTL])
1969 parms->hop_limit = nla_get_u8(data[IFLA_IPTUN_TTL]);
1970
1971 if (data[IFLA_IPTUN_ENCAP_LIMIT])
1972 parms->encap_limit = nla_get_u8(data[IFLA_IPTUN_ENCAP_LIMIT]);
1973
1974 if (data[IFLA_IPTUN_FLOWINFO])
1ff05fb7 1975 parms->flowinfo = nla_get_be32(data[IFLA_IPTUN_FLOWINFO]);
0b112457
ND		 1976
1977 if (data[IFLA_IPTUN_FLAGS])
1978 parms->flags = nla_get_u32(data[IFLA_IPTUN_FLAGS]);
1979
1980 if (data[IFLA_IPTUN_PROTO])
1981 parms->proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
8d79266b
AS		 1982
1983 if (data[IFLA_IPTUN_COLLECT_METADATA])
1984 parms->collect_md = true;
0a473b82
CG		 1985
1986 if (data[IFLA_IPTUN_FWMARK])
1987 parms->fwmark = nla_get_u32(data[IFLA_IPTUN_FWMARK]);
0b112457
ND		 1988}
1989
1990static int ip6_tnl_newlink(struct net *src_net, struct net_device *dev,
7a3f4a18
MS		 1991 struct nlattr *tb[], struct nlattr *data[],
1992 struct netlink_ext_ack *extack)
0b112457
ND		 1993{
1994 struct net *net = dev_net(dev);
8d79266b 1995 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
b3a27b51 1996 struct ip_tunnel_encap ipencap;
a6aa8044
XL		 1997 struct ip6_tnl *nt, *t;
1998 int err;
0b112457
ND		 1999
2000 nt = netdev_priv(dev);
b3a27b51 2001
537dd2d9 2002 if (ip_tunnel_netlink_encap_parms(data, &ipencap)) {
a6aa8044 2003 err = ip6_tnl_encap_setup(nt, &ipencap);
b3a27b51
TH		 2004 if (err < 0)
2005 return err;
2006 }
2007
0b112457
ND		 2008 ip6_tnl_netlink_parms(data, &nt->parms);
2009
8d79266b
AS		 2010 if (nt->parms.collect_md) {
2011 if (rtnl_dereference(ip6n->collect_md_tun))
2012 return -EEXIST;
2013 } else {
2014 t = ip6_tnl_locate(net, &nt->parms, 0);
2015 if (!IS_ERR(t))
2016 return -EEXIST;
2017 }
0b112457 2018
a6aa8044
XL		 2019 err = ip6_tnl_create2(dev);
2020 if (!err && tb[IFLA_MTU])
2021 ip6_tnl_change_mtu(dev, nla_get_u32(tb[IFLA_MTU]));
2022
2023 return err;
0b112457
ND		 2024}
2025
2026static int ip6_tnl_changelink(struct net_device *dev, struct nlattr *tb[],
ad744b22
MS		 2027 struct nlattr *data[],
2028 struct netlink_ext_ack *extack)
0b112457 2029{
0bd87628 2030 struct ip6_tnl *t = netdev_priv(dev);
0b112457 2031 struct __ip6_tnl_parm p;
0bd87628 2032 struct net *net = t->net;
0b112457 2033 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
b3a27b51 2034 struct ip_tunnel_encap ipencap;
0b112457
ND		 2035
2036 if (dev == ip6n->fb_tnl_dev)
2037 return -EINVAL;
2038
537dd2d9 2039 if (ip_tunnel_netlink_encap_parms(data, &ipencap)) {
b3a27b51
TH		 2040 int err = ip6_tnl_encap_setup(t, &ipencap);
2041
2042 if (err < 0)
2043 return err;
2044 }
0b112457 2045 ip6_tnl_netlink_parms(data, &p);
8d79266b
AS		 2046 if (p.collect_md)
2047 return -EINVAL;
0b112457
ND		 2048
2049 t = ip6_tnl_locate(net, &p, 0);
37355565 2050 if (!IS_ERR(t)) {
0b112457
ND		 2051 if (t->dev != dev)
2052 return -EEXIST;
2053 } else
2054 t = netdev_priv(dev);
2055
77788567
HW		 2056 ip6_tnl_update(t, &p);
2057 return 0;
0b112457
ND		 2058}
2059
1e9f3d6f
ND		 2060static void ip6_tnl_dellink(struct net_device *dev, struct list_head *head)
2061{
2062 struct net *net = dev_net(dev);
2063 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
2064
2065 if (dev != ip6n->fb_tnl_dev)
2066 unregister_netdevice_queue(dev, head);
2067}
2068
b58d731a 2069static size_t ip6_tnl_get_size(const struct net_device *dev)
c075b130
ND		 2070{
2071 return
2072 /* IFLA_IPTUN_LINK */
2073 nla_total_size(4) +
2074 /* IFLA_IPTUN_LOCAL */
2075 nla_total_size(sizeof(struct in6_addr)) +
2076 /* IFLA_IPTUN_REMOTE */
2077 nla_total_size(sizeof(struct in6_addr)) +
2078 /* IFLA_IPTUN_TTL */
2079 nla_total_size(1) +
2080 /* IFLA_IPTUN_ENCAP_LIMIT */
2081 nla_total_size(1) +
2082 /* IFLA_IPTUN_FLOWINFO */
2083 nla_total_size(4) +
2084 /* IFLA_IPTUN_FLAGS */
2085 nla_total_size(4) +
cfa323b6
ND		 2086 /* IFLA_IPTUN_PROTO */
2087 nla_total_size(1) +
b3a27b51
TH		 2088 /* IFLA_IPTUN_ENCAP_TYPE */
2089 nla_total_size(2) +
2090 /* IFLA_IPTUN_ENCAP_FLAGS */
2091 nla_total_size(2) +
2092 /* IFLA_IPTUN_ENCAP_SPORT */
2093 nla_total_size(2) +
2094 /* IFLA_IPTUN_ENCAP_DPORT */
2095 nla_total_size(2) +
8d79266b
AS		 2096 /* IFLA_IPTUN_COLLECT_METADATA */
2097 nla_total_size(0) +
0a473b82
CG		 2098 /* IFLA_IPTUN_FWMARK */
2099 nla_total_size(4) +
c075b130
ND		 2100 0;
2101}
2102
b58d731a 2103static int ip6_tnl_fill_info(struct sk_buff *skb, const struct net_device *dev)
c075b130
ND		 2104{
2105 struct ip6_tnl *tunnel = netdev_priv(dev);
2106 struct __ip6_tnl_parm *parm = &tunnel->parms;
2107
2108 if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) ||
930345ea
JB		 2109 nla_put_in6_addr(skb, IFLA_IPTUN_LOCAL, &parm->laddr) ||
2110 nla_put_in6_addr(skb, IFLA_IPTUN_REMOTE, &parm->raddr) ||
c075b130
ND		 2111 nla_put_u8(skb, IFLA_IPTUN_TTL, parm->hop_limit) ||
2112 nla_put_u8(skb, IFLA_IPTUN_ENCAP_LIMIT, parm->encap_limit) ||
2113 nla_put_be32(skb, IFLA_IPTUN_FLOWINFO, parm->flowinfo) ||
cfa323b6 2114 nla_put_u32(skb, IFLA_IPTUN_FLAGS, parm->flags) ||
0a473b82
CG		 2115 nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->proto) ||
2116 nla_put_u32(skb, IFLA_IPTUN_FWMARK, parm->fwmark))
c075b130 2117 goto nla_put_failure;
b3a27b51 2118
8d79266b
AS		 2119 if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE, tunnel->encap.type) ||
2120 nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT, tunnel->encap.sport) ||
2121 nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT, tunnel->encap.dport) ||
2122 nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS, tunnel->encap.flags))
b3a27b51
TH		 2123 goto nla_put_failure;
2124
8d79266b
AS		 2125 if (parm->collect_md)
2126 if (nla_put_flag(skb, IFLA_IPTUN_COLLECT_METADATA))
2127 goto nla_put_failure;
0a473b82 2128
c075b130
ND		 2129 return 0;
2130
2131nla_put_failure:
2132 return -EMSGSIZE;
2133}
2134
1728d4fa
ND		 2135struct net *ip6_tnl_get_link_net(const struct net_device *dev)
2136{
2137 struct ip6_tnl *tunnel = netdev_priv(dev);
2138
2139 return tunnel->net;
2140}
2141EXPORT_SYMBOL(ip6_tnl_get_link_net);
2142
0b112457
ND		 2143static const struct nla_policy ip6_tnl_policy[IFLA_IPTUN_MAX + 1] = {
2144 [IFLA_IPTUN_LINK] = { .type = NLA_U32 },
2145 [IFLA_IPTUN_LOCAL] = { .len = sizeof(struct in6_addr) },
2146 [IFLA_IPTUN_REMOTE] = { .len = sizeof(struct in6_addr) },
2147 [IFLA_IPTUN_TTL] = { .type = NLA_U8 },
2148 [IFLA_IPTUN_ENCAP_LIMIT] = { .type = NLA_U8 },
2149 [IFLA_IPTUN_FLOWINFO] = { .type = NLA_U32 },
2150 [IFLA_IPTUN_FLAGS] = { .type = NLA_U32 },
2151 [IFLA_IPTUN_PROTO] = { .type = NLA_U8 },
b3a27b51
TH		 2152 [IFLA_IPTUN_ENCAP_TYPE] = { .type = NLA_U16 },
2153 [IFLA_IPTUN_ENCAP_FLAGS] = { .type = NLA_U16 },
2154 [IFLA_IPTUN_ENCAP_SPORT] = { .type = NLA_U16 },
2155 [IFLA_IPTUN_ENCAP_DPORT] = { .type = NLA_U16 },
8d79266b 2156 [IFLA_IPTUN_COLLECT_METADATA] = { .type = NLA_FLAG },
0a473b82 2157 [IFLA_IPTUN_FWMARK] = { .type = NLA_U32 },
0b112457
ND		 2158};
2159
c075b130
ND		 2160static struct rtnl_link_ops ip6_link_ops __read_mostly = {
2161 .kind = "ip6tnl",
2162 .maxtype = IFLA_IPTUN_MAX,
0b112457 2163 .policy = ip6_tnl_policy,
c075b130 2164 .priv_size = sizeof(struct ip6_tnl),
0b112457
ND		 2165 .setup = ip6_tnl_dev_setup,
2166 .validate = ip6_tnl_validate,
2167 .newlink = ip6_tnl_newlink,
2168 .changelink = ip6_tnl_changelink,
1e9f3d6f 2169 .dellink = ip6_tnl_dellink,
b58d731a
ND		 2170 .get_size = ip6_tnl_get_size,
2171 .fill_info = ip6_tnl_fill_info,
1728d4fa 2172 .get_link_net = ip6_tnl_get_link_net,
c075b130
ND		 2173};
2174
3ff2cfa5 2175static struct xfrm6_tunnel ip4ip6_handler __read_mostly = {
c4d3efaf
YK		 2176 .handler = ip4ip6_rcv,
2177 .err_handler = ip4ip6_err,
2178 .priority = 1,
2179};
2180
3ff2cfa5 2181static struct xfrm6_tunnel ip6ip6_handler __read_mostly = {
0303770d
PM		 2182 .handler = ip6ip6_rcv,
2183 .err_handler = ip6ip6_err,
d2acc347 2184 .priority = 1,
1da177e4
LT		 2185};
2186
f200e98d
VF		 2187static struct xfrm6_tunnel mplsip6_handler __read_mostly = {
2188 .handler = mplsip6_rcv,
2189 .err_handler = mplsip6_err,
2190 .priority = 1,
2191};
2192
bb401cae 2193static void __net_exit ip6_tnl_destroy_tunnels(struct net *net, struct list_head *list)
3e6c9fb5 2194{
1e9f3d6f 2195 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
0bd87628 2196 struct net_device *dev, *aux;
3e6c9fb5
PE		 2197 int h;
2198 struct ip6_tnl *t;
2199
0bd87628
ND		 2200 for_each_netdev_safe(net, dev, aux)
2201 if (dev->rtnl_link_ops == &ip6_link_ops)
bb401cae 2202 unregister_netdevice_queue(dev, list);
0bd87628 2203
e87a8f24 2204 for (h = 0; h < IP6_TUNNEL_HASH_SIZE; h++) {
94767632 2205 t = rtnl_dereference(ip6n->tnls_r_l[h]);
53b24b8f 2206 while (t) {
0bd87628
ND		 2207 /* If dev is in the same netns, it has already
2208 * been added to the list by the previous loop.
2209 */
2210 if (!net_eq(dev_net(t->dev), net))
bb401cae 2211 unregister_netdevice_queue(t->dev, list);
94767632 2212 t = rtnl_dereference(t->next);
cf4432f5 2213 }
3e6c9fb5 2214 }
941ea91e
HV		 2215
2216 t = rtnl_dereference(ip6n->tnls_wc[0]);
2217 while (t) {
2218 /* If dev is in the same netns, it has already
2219 * been added to the list by the previous loop.
2220 */
2221 if (!net_eq(dev_net(t->dev), net))
2222 unregister_netdevice_queue(t->dev, list);
2223 t = rtnl_dereference(t->next);
2224 }
3e6c9fb5
PE		 2225}
2226
2c8c1e72 2227static int __net_init ip6_tnl_init_net(struct net *net)
13eeb8e9 2228{
ac31cd3c 2229 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
731abb9c 2230 struct ip6_tnl *t = NULL;
13eeb8e9 2231 int err;
13eeb8e9 2232
3e6c9fb5
PE		 2233 ip6n->tnls[0] = ip6n->tnls_wc;
2234 ip6n->tnls[1] = ip6n->tnls_r_l;
2235
79134e6c
ED		 2236 if (!net_has_fallback_tunnels(net))
2237 return 0;
15820e12
PE		 2238 err = -ENOMEM;
2239 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
c835a677 2240 NET_NAME_UNKNOWN, ip6_tnl_dev_setup);
15820e12
PE		 2241
2242 if (!ip6n->fb_tnl_dev)
2243 goto err_alloc_dev;
be77e593 2244 dev_net_set(ip6n->fb_tnl_dev, net);
bb814094 2245 ip6n->fb_tnl_dev->rtnl_link_ops = &ip6_link_ops;
0bd87628
ND		 2246 /* FB netdevice is special: we have one, and only one per netns.
2247 * Allowing to move it to another netns is clearly unsafe.
2248 */
2249 ip6n->fb_tnl_dev->features |= NETIF_F_NETNS_LOCAL;
15820e12 2250
8560f226
ED		 2251 err = ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
2252 if (err < 0)
2253 goto err_register;
15820e12
PE		 2254
2255 err = register_netdev(ip6n->fb_tnl_dev);
2256 if (err < 0)
2257 goto err_register;
731abb9c
JB		 2258
2259 t = netdev_priv(ip6n->fb_tnl_dev);
2260
2261 strcpy(t->parms.name, ip6n->fb_tnl_dev->name);
13eeb8e9
PE		 2262 return 0;
2263
15820e12 2264err_register:
cf124db5 2265 free_netdev(ip6n->fb_tnl_dev);
15820e12 2266err_alloc_dev:
13eeb8e9
PE		 2267 return err;
2268}
2269
bb401cae 2270static void __net_exit ip6_tnl_exit_batch_net(struct list_head *net_list)
13eeb8e9 2271{
bb401cae
ED		 2272 struct net *net;
2273 LIST_HEAD(list);
2274
3e6c9fb5 2275 rtnl_lock();
bb401cae
ED		 2276 list_for_each_entry(net, net_list, exit_list)
2277 ip6_tnl_destroy_tunnels(net, &list);
2278 unregister_netdevice_many(&list);
3e6c9fb5 2279 rtnl_unlock();
13eeb8e9
PE		 2280}
2281
2282static struct pernet_operations ip6_tnl_net_ops = {
2283 .init = ip6_tnl_init_net,
bb401cae 2284 .exit_batch = ip6_tnl_exit_batch_net,
ac31cd3c
EB		 2285 .id = &ip6_tnl_net_id,
2286 .size = sizeof(struct ip6_tnl_net),
13eeb8e9
PE		 2287};
2288
1da177e4
LT		 2289/**
2290 * ip6_tunnel_init - register protocol and reserve needed resources
2291 *
2292 * Return: 0 on success
2293 **/
2294
2295static int __init ip6_tunnel_init(void)
2296{
2297 int err;
2298
8c22dab0
XL		 2299 if (!ipv6_mod_enabled())
2300 return -EOPNOTSUPP;
2301
d5aa407f
AD		 2302 err = register_pernet_device(&ip6_tnl_net_ops);
2303 if (err < 0)
2304 goto out_pernet;
2305
2306 err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
2307 if (err < 0) {
f3213831 2308 pr_err("%s: can't register ip4ip6\n", __func__);
d5aa407f 2309 goto out_ip4ip6;
c4d3efaf
YK		 2310 }
2311
d5aa407f
AD		 2312 err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
2313 if (err < 0) {
f3213831 2314 pr_err("%s: can't register ip6ip6\n", __func__);
d5aa407f 2315 goto out_ip6ip6;
1da177e4 2316 }
f200e98d
VF		 2317
2318 if (ip6_tnl_mpls_supported()) {
2319 err = xfrm6_tunnel_register(&mplsip6_handler, AF_MPLS);
2320 if (err < 0) {
2321 pr_err("%s: can't register mplsip6\n", __func__);
2322 goto out_mplsip6;
2323 }
2324 }
2325
c075b130
ND		 2326 err = rtnl_link_register(&ip6_link_ops);
2327 if (err < 0)
2328 goto rtnl_link_failed;
13eeb8e9 2329
1da177e4 2330 return 0;
d5aa407f 2331
c075b130 2332rtnl_link_failed:
f200e98d
VF		 2333 if (ip6_tnl_mpls_supported())
2334 xfrm6_tunnel_deregister(&mplsip6_handler, AF_MPLS);
2335out_mplsip6:
c075b130 2336 xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6);
d5aa407f 2337out_ip6ip6:
c4d3efaf 2338 xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
d5aa407f
AD		 2339out_ip4ip6:
2340 unregister_pernet_device(&ip6_tnl_net_ops);
2341out_pernet:
1da177e4
LT		 2342 return err;
2343}
2344
2345/**
2346 * ip6_tunnel_cleanup - free resources and unregister protocol
2347 **/
2348
2349static void __exit ip6_tunnel_cleanup(void)
2350{
c075b130 2351 rtnl_link_unregister(&ip6_link_ops);
c4d3efaf 2352 if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
f3213831 2353 pr_info("%s: can't deregister ip4ip6\n", __func__);
c4d3efaf 2354
73d605d1 2355 if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
f3213831 2356 pr_info("%s: can't deregister ip6ip6\n", __func__);
1da177e4 2357
f200e98d
VF		 2358 if (ip6_tnl_mpls_supported() &&
2359 xfrm6_tunnel_deregister(&mplsip6_handler, AF_MPLS))
2360 pr_info("%s: can't deregister mplsip6\n", __func__);
ac31cd3c 2361 unregister_pernet_device(&ip6_tnl_net_ops);
1da177e4
LT		 2362}
2363
2364module_init(ip6_tunnel_init);
2365module_exit(ip6_tunnel_cleanup);
Linux 6.x block layer and io_uring tree(s)