projects / GitHub / moto-9609 / android_kernel_motorola_exynos9610.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
remove libdss from Makefile
[GitHub/moto-9609/android_kernel_motorola_exynos9610.git] / net / ipv6 / ip6_tunnel.c
CommitLineData
1da177e4 1/*
c4d3efaf 2 * IPv6 tunneling device
1da177e4
LT		 3 * Linux INET6 implementation
4 *
5 * Authors:
1ab1457c 6 * Ville Nuorvala <vnuorval@tcs.hut.fi>
c4d3efaf 7 * Yasuyuki Kozakai <kozakai@linux-ipv6.org>
1da177e4 8 *
1da177e4 9 * Based on:
c4d3efaf 10 * linux/net/ipv6/sit.c and linux/net/ipv4/ipip.c
1da177e4
LT		 11 *
12 * RFC 2473
13 *
14 * This program is free software; you can redistribute it and/or
15 * modify it under the terms of the GNU General Public License
16 * as published by the Free Software Foundation; either version
17 * 2 of the License, or (at your option) any later version.
18 *
19 */
20
f3213831
JP		 21#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22
1da177e4 23#include <linux/module.h>
4fc268d2 24#include <linux/capability.h>
1da177e4
LT		 25#include <linux/errno.h>
26#include <linux/types.h>
27#include <linux/sockios.h>
c4d3efaf 28#include <linux/icmp.h>
1da177e4
LT		 29#include <linux/if.h>
30#include <linux/in.h>
31#include <linux/ip.h>
1da177e4
LT		 32#include <linux/net.h>
33#include <linux/in6.h>
34#include <linux/netdevice.h>
35#include <linux/if_arp.h>
36#include <linux/icmpv6.h>
37#include <linux/init.h>
38#include <linux/route.h>
39#include <linux/rtnetlink.h>
40#include <linux/netfilter_ipv6.h>
5a0e3ad6 41#include <linux/slab.h>
ddbe5032 42#include <linux/hash.h>
e837735e 43#include <linux/etherdevice.h>
1da177e4 44
7c0f6ba6 45#include <linux/uaccess.h>
60063497 46#include <linux/atomic.h>
1da177e4 47
c4d3efaf 48#include <net/icmp.h>
1da177e4 49#include <net/ip.h>
c5441932 50#include <net/ip_tunnels.h>
1da177e4 51#include <net/ipv6.h>
1da177e4
LT		 52#include <net/ip6_route.h>
53#include <net/addrconf.h>
54#include <net/ip6_tunnel.h>
55#include <net/xfrm.h>
56#include <net/dsfield.h>
57#include <net/inet_ecn.h>
13eeb8e9
PE		 58#include <net/net_namespace.h>
59#include <net/netns/generic.h>
8d79266b 60#include <net/dst_metadata.h>
1da177e4
LT		 61
62MODULE_AUTHOR("Ville Nuorvala");
c4d3efaf 63MODULE_DESCRIPTION("IPv6 tunneling device");
1da177e4 64MODULE_LICENSE("GPL");
f98f89a0 65MODULE_ALIAS_RTNL_LINK("ip6tnl");
6dfbd87a 66MODULE_ALIAS_NETDEV("ip6tnl0");
1da177e4 67
e87a8f24
JK		 68#define IP6_TUNNEL_HASH_SIZE_SHIFT 5
69#define IP6_TUNNEL_HASH_SIZE (1 << IP6_TUNNEL_HASH_SIZE_SHIFT)
1da177e4 70
f4e0b4c5
ND		 71static bool log_ecn_error = true;
72module_param(log_ecn_error, bool, 0644);
73MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
74
ddbe5032
ED		 75static u32 HASH(const struct in6_addr *addr1, const struct in6_addr *addr2)
76{
77 u32 hash = ipv6_addr_hash(addr1) ^ ipv6_addr_hash(addr2);
78
e87a8f24 79 return hash_32(hash, IP6_TUNNEL_HASH_SIZE_SHIFT);
ddbe5032 80}
1da177e4 81
8560f226 82static int ip6_tnl_dev_init(struct net_device *dev);
3144581c 83static void ip6_tnl_dev_setup(struct net_device *dev);
c075b130 84static struct rtnl_link_ops ip6_link_ops __read_mostly;
1da177e4 85
c7d03a00 86static unsigned int ip6_tnl_net_id __read_mostly;
13eeb8e9 87struct ip6_tnl_net {
15820e12
PE		 88 /* the IPv6 tunnel fallback device */
89 struct net_device *fb_tnl_dev;
3e6c9fb5 90 /* lists for storing tunnels in use */
e87a8f24 91 struct ip6_tnl __rcu *tnls_r_l[IP6_TUNNEL_HASH_SIZE];
94767632
ED		 92 struct ip6_tnl __rcu *tnls_wc[1];
93 struct ip6_tnl __rcu **tnls[2];
8d79266b 94 struct ip6_tnl __rcu *collect_md_tun;
13eeb8e9
PE		 95};
96
8560f226
ED		 97static struct net_device_stats *ip6_get_stats(struct net_device *dev)
98{
56a4342d 99 struct pcpu_sw_netstats tmp, sum = { 0 };
8560f226
ED		 100 int i;
101
102 for_each_possible_cpu(i) {
abb6013c 103 unsigned int start;
8f84985f
LR		 104 const struct pcpu_sw_netstats *tstats =
105 per_cpu_ptr(dev->tstats, i);
8560f226 106
abb6013c 107 do {
57a7744e 108 start = u64_stats_fetch_begin_irq(&tstats->syncp);
abb6013c
LR		 109 tmp.rx_packets = tstats->rx_packets;
110 tmp.rx_bytes = tstats->rx_bytes;
111 tmp.tx_packets = tstats->tx_packets;
112 tmp.tx_bytes = tstats->tx_bytes;
57a7744e 113 } while (u64_stats_fetch_retry_irq(&tstats->syncp, start));
abb6013c
LR		 114
115 sum.rx_packets += tmp.rx_packets;
116 sum.rx_bytes += tmp.rx_bytes;
117 sum.tx_packets += tmp.tx_packets;
118 sum.tx_bytes += tmp.tx_bytes;
8560f226
ED		 119 }
120 dev->stats.rx_packets = sum.rx_packets;
121 dev->stats.rx_bytes = sum.rx_bytes;
122 dev->stats.tx_packets = sum.tx_packets;
123 dev->stats.tx_bytes = sum.tx_bytes;
124 return &dev->stats;
125}
126
1da177e4 127/**
3144581c 128 * ip6_tnl_lookup - fetch tunnel matching the end-point addresses
1ab1457c
YH		 129 * @remote: the address of the tunnel exit-point
130 * @local: the address of the tunnel entry-point
1da177e4 131 *
1ab1457c 132 * Return:
1da177e4 133 * tunnel matching given end-points if found,
1ab1457c 134 * else fallback tunnel if its device is up,
1da177e4
LT		 135 * else %NULL
136 **/
137
2922bc8a
ED		 138#define for_each_ip6_tunnel_rcu(start) \
139 for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
140
1da177e4 141static struct ip6_tnl *
b71d1d42 142ip6_tnl_lookup(struct net *net, const struct in6_addr *remote, const struct in6_addr *local)
1da177e4 143{
ddbe5032 144 unsigned int hash = HASH(remote, local);
1da177e4 145 struct ip6_tnl *t;
3e6c9fb5 146 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
ea3dc960 147 struct in6_addr any;
1da177e4 148
ddbe5032 149 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
1da177e4
LT		 150 if (ipv6_addr_equal(local, &t->parms.laddr) &&
151 ipv6_addr_equal(remote, &t->parms.raddr) &&
152 (t->dev->flags & IFF_UP))
153 return t;
154 }
ea3dc960
SK		 155
156 memset(&any, 0, sizeof(any));
157 hash = HASH(&any, local);
158 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
159 if (ipv6_addr_equal(local, &t->parms.laddr) &&
68d00f33 160 ipv6_addr_any(&t->parms.raddr) &&
ea3dc960
SK		 161 (t->dev->flags & IFF_UP))
162 return t;
163 }
164
165 hash = HASH(remote, &any);
166 for_each_ip6_tunnel_rcu(ip6n->tnls_r_l[hash]) {
167 if (ipv6_addr_equal(remote, &t->parms.raddr) &&
68d00f33 168 ipv6_addr_any(&t->parms.laddr) &&
ea3dc960
SK		 169 (t->dev->flags & IFF_UP))
170 return t;
171 }
172
8d79266b 173 t = rcu_dereference(ip6n->collect_md_tun);
6c1cb439 174 if (t && t->dev->flags & IFF_UP)
8d79266b
AS		 175 return t;
176
2922bc8a
ED		 177 t = rcu_dereference(ip6n->tnls_wc[0]);
178 if (t && (t->dev->flags & IFF_UP))
1da177e4
LT		 179 return t;
180
181 return NULL;
182}
183
184/**
3144581c 185 * ip6_tnl_bucket - get head of list matching given tunnel parameters
1ab1457c 186 * @p: parameters containing tunnel end-points
1da177e4
LT		 187 *
188 * Description:
3144581c 189 * ip6_tnl_bucket() returns the head of the list matching the
1da177e4
LT		 190 * &struct in6_addr entries laddr and raddr in @p.
191 *
1ab1457c 192 * Return: head of IPv6 tunnel list
1da177e4
LT		 193 **/
194
94767632 195static struct ip6_tnl __rcu **
c12b395a 196ip6_tnl_bucket(struct ip6_tnl_net *ip6n, const struct __ip6_tnl_parm *p)
1da177e4 197{
b71d1d42
ED		 198 const struct in6_addr *remote = &p->raddr;
199 const struct in6_addr *local = &p->laddr;
95c96174 200 unsigned int h = 0;
1da177e4
LT		 201 int prio = 0;
202
203 if (!ipv6_addr_any(remote) || !ipv6_addr_any(local)) {
204 prio = 1;
ddbe5032 205 h = HASH(remote, local);
1da177e4 206 }
3e6c9fb5 207 return &ip6n->tnls[prio][h];
1da177e4
LT		 208}
209
210/**
3144581c 211 * ip6_tnl_link - add tunnel to hash table
1da177e4
LT		 212 * @t: tunnel to be added
213 **/
214
215static void
2dd02c89 216ip6_tnl_link(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 217{
94767632 218 struct ip6_tnl __rcu **tp = ip6_tnl_bucket(ip6n, &t->parms);
1da177e4 219
8d79266b
AS		 220 if (t->parms.collect_md)
221 rcu_assign_pointer(ip6n->collect_md_tun, t);
cf778b00
ED		 222 rcu_assign_pointer(t->next , rtnl_dereference(*tp));
223 rcu_assign_pointer(*tp, t);
1da177e4
LT		 224}
225
226/**
3144581c 227 * ip6_tnl_unlink - remove tunnel from hash table
1da177e4
LT		 228 * @t: tunnel to be removed
229 **/
230
231static void
2dd02c89 232ip6_tnl_unlink(struct ip6_tnl_net *ip6n, struct ip6_tnl *t)
1da177e4 233{
94767632
ED		 234 struct ip6_tnl __rcu **tp;
235 struct ip6_tnl *iter;
236
8d79266b
AS		 237 if (t->parms.collect_md)
238 rcu_assign_pointer(ip6n->collect_md_tun, NULL);
239
94767632
ED		 240 for (tp = ip6_tnl_bucket(ip6n, &t->parms);
241 (iter = rtnl_dereference(*tp)) != NULL;
242 tp = &iter->next) {
243 if (t == iter) {
cf778b00 244 rcu_assign_pointer(*tp, t->next);
1da177e4
LT		 245 break;
246 }
247 }
248}
249
8560f226
ED		 250static void ip6_dev_free(struct net_device *dev)
251{
cdf3464e
MKL		 252 struct ip6_tnl *t = netdev_priv(dev);
253
0d3c703a 254 gro_cells_destroy(&t->gro_cells);
607f725f 255 dst_cache_destroy(&t->dst_cache);
8560f226 256 free_percpu(dev->tstats);
8560f226
ED		 257}
258
0b112457
ND		 259static int ip6_tnl_create2(struct net_device *dev)
260{
261 struct ip6_tnl *t = netdev_priv(dev);
262 struct net *net = dev_net(dev);
263 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
264 int err;
265
266 t = netdev_priv(dev);
0b112457 267
b6ee376c 268 dev->rtnl_link_ops = &ip6_link_ops;
0b112457
ND		 269 err = register_netdevice(dev);
270 if (err < 0)
271 goto out;
272
273 strcpy(t->parms.name, dev->name);
0b112457
ND		 274
275 dev_hold(dev);
276 ip6_tnl_link(ip6n, t);
277 return 0;
278
279out:
280 return err;
281}
282
1da177e4 283/**
2c53040f 284 * ip6_tnl_create - create a new tunnel
1da177e4
LT		 285 * @p: tunnel parameters
286 * @pt: pointer to new tunnel
287 *
288 * Description:
289 * Create tunnel matching given parameters.
1ab1457c
YH		 290 *
291 * Return:
37355565 292 * created tunnel or error pointer
1da177e4
LT		 293 **/
294
c12b395a 295static struct ip6_tnl *ip6_tnl_create(struct net *net, struct __ip6_tnl_parm *p)
1da177e4
LT		 296{
297 struct net_device *dev;
298 struct ip6_tnl *t;
299 char name[IFNAMSIZ];
109dce20 300 int err = -E2BIG;
1da177e4 301
109dce20
ED		 302 if (p->name[0]) {
303 if (!dev_valid_name(p->name))
304 goto failed;
1da177e4 305 strlcpy(name, p->name, IFNAMSIZ);
109dce20 306 } else {
34cc7ba6 307 sprintf(name, "ip6tnl%%d");
109dce20
ED		 308 }
309 err = -ENOMEM;
c835a677
TG		 310 dev = alloc_netdev(sizeof(*t), name, NET_NAME_UNKNOWN,
311 ip6_tnl_dev_setup);
63159f29 312 if (!dev)
567131a7 313 goto failed;
1da177e4 314
554eb277
PE		 315 dev_net_set(dev, net);
316
2941a486 317 t = netdev_priv(dev);
1da177e4 318 t->parms = *p;
0bd87628 319 t->net = dev_net(dev);
0b112457 320 err = ip6_tnl_create2(dev);
8560f226
ED		 321 if (err < 0)
322 goto failed_free;
1da177e4 323
567131a7 324 return t;
b37d428b
PE		 325
326failed_free:
cf124db5 327 free_netdev(dev);
567131a7 328failed:
37355565 329 return ERR_PTR(err);
1da177e4
LT		 330}
331
332/**
3144581c 333 * ip6_tnl_locate - find or create tunnel matching given parameters
1ab1457c 334 * @p: tunnel parameters
1da177e4
LT		 335 * @create: != 0 if allowed to create new tunnel if no match found
336 *
337 * Description:
3144581c 338 * ip6_tnl_locate() first tries to locate an existing tunnel
1da177e4
LT		 339 * based on @parms. If this is unsuccessful, but @create is set a new
340 * tunnel device is created and registered for use.
341 *
342 * Return:
37355565 343 * matching tunnel or error pointer
1da177e4
LT		 344 **/
345
2dd02c89 346static struct ip6_tnl *ip6_tnl_locate(struct net *net,
c12b395a 347 struct __ip6_tnl_parm *p, int create)
1da177e4 348{
b71d1d42
ED		 349 const struct in6_addr *remote = &p->raddr;
350 const struct in6_addr *local = &p->laddr;
94767632 351 struct ip6_tnl __rcu **tp;
1da177e4 352 struct ip6_tnl *t;
2dd02c89 353 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 354
94767632
ED		 355 for (tp = ip6_tnl_bucket(ip6n, p);
356 (t = rtnl_dereference(*tp)) != NULL;
357 tp = &t->next) {
1da177e4 358 if (ipv6_addr_equal(local, &t->parms.laddr) &&
2b0bb01b
SK		 359 ipv6_addr_equal(remote, &t->parms.raddr)) {
360 if (create)
37355565 361 return ERR_PTR(-EEXIST);
2b0bb01b 362
567131a7 363 return t;
2b0bb01b 364 }
1da177e4
LT		 365 }
366 if (!create)
37355565 367 return ERR_PTR(-ENODEV);
2dd02c89 368 return ip6_tnl_create(net, p);
1da177e4
LT		 369}
370
371/**
3144581c 372 * ip6_tnl_dev_uninit - tunnel device uninitializer
1da177e4 373 * @dev: the device to be destroyed
1ab1457c 374 *
1da177e4 375 * Description:
3144581c 376 * ip6_tnl_dev_uninit() removes tunnel from its list
1da177e4
LT		 377 **/
378
379static void
3144581c 380ip6_tnl_dev_uninit(struct net_device *dev)
1da177e4 381{
2941a486 382 struct ip6_tnl *t = netdev_priv(dev);
0bd87628 383 struct net *net = t->net;
2dd02c89 384 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 385
94767632 386 if (dev == ip6n->fb_tnl_dev)
a9b3cd7f 387 RCU_INIT_POINTER(ip6n->tnls_wc[0], NULL);
94767632 388 else
2dd02c89 389 ip6_tnl_unlink(ip6n, t);
607f725f 390 dst_cache_reset(&t->dst_cache);
1da177e4
LT		 391 dev_put(dev);
392}
393
394/**
395 * parse_tvl_tnl_enc_lim - handle encapsulation limit option
396 * @skb: received socket buffer
397 *
1ab1457c
YH		 398 * Return:
399 * 0 if none was found,
1da177e4
LT		 400 * else index to encapsulation limit
401 **/
402
c12b395a 403__u16 ip6_tnl_parse_tlv_enc_lim(struct sk_buff *skb, __u8 *raw)
1da177e4 404{
fbfa743a
ED		 405 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *)raw;
406 unsigned int nhoff = raw - skb->data;
407 unsigned int off = nhoff + sizeof(*ipv6h);
408 u8 next, nexthdr = ipv6h->nexthdr;
1da177e4
LT		 409
410 while (ipv6_ext_hdr(nexthdr) && nexthdr != NEXTHDR_NONE) {
1da177e4 411 struct ipv6_opt_hdr *hdr;
fbfa743a
ED		 412 u16 optlen;
413
414 if (!pskb_may_pull(skb, off + sizeof(*hdr)))
1da177e4
LT		 415 break;
416
fbfa743a 417 hdr = (struct ipv6_opt_hdr *)(skb->data + off);
1da177e4
LT		 418 if (nexthdr == NEXTHDR_FRAGMENT) {
419 struct frag_hdr *frag_hdr = (struct frag_hdr *) hdr;
420 if (frag_hdr->frag_off)
421 break;
422 optlen = 8;
423 } else if (nexthdr == NEXTHDR_AUTH) {
424 optlen = (hdr->hdrlen + 2) << 2;
425 } else {
426 optlen = ipv6_optlen(hdr);
427 }
fbfa743a
ED		 428 /* cache hdr->nexthdr, since pskb_may_pull() might
429 * invalidate hdr
430 */
431 next = hdr->nexthdr;
1da177e4 432 if (nexthdr == NEXTHDR_DEST) {
fbfa743a
ED		 433 u16 i = 2;
434
435 /* Remember : hdr is no longer valid at this point. */
436 if (!pskb_may_pull(skb, off + optlen))
437 break;
438
1da177e4
LT		 439 while (1) {
440 struct ipv6_tlv_tnl_enc_lim *tel;
441
442 /* No more room for encapsulation limit */
fbfa743a 443 if (i + sizeof(*tel) > optlen)
1da177e4
LT		 444 break;
445
63117f09 446 tel = (struct ipv6_tlv_tnl_enc_lim *)(skb->data + off + i);
1da177e4
LT		 447 /* return index of option if found and valid */
448 if (tel->type == IPV6_TLV_TNL_ENCAP_LIMIT &&
449 tel->length == 1)
fbfa743a 450 return i + off - nhoff;
1da177e4
LT		 451 /* else jump to next option */
452 if (tel->type)
453 i += tel->length + 2;
454 else
455 i++;
456 }
457 }
fbfa743a 458 nexthdr = next;
1da177e4
LT		 459 off += optlen;
460 }
461 return 0;
462}
c12b395a 463EXPORT_SYMBOL(ip6_tnl_parse_tlv_enc_lim);
1da177e4
LT		 464
465/**
e490d1d8 466 * ip6_tnl_err - tunnel error handler
1da177e4
LT		 467 *
468 * Description:
e490d1d8 469 * ip6_tnl_err() should handle errors in the tunnel according
1da177e4
LT		 470 * to the specifications in RFC 2473.
471 **/
472
d2acc347 473static int
502b0935 474ip6_tnl_err(struct sk_buff *skb, __u8 ipproto, struct inet6_skb_parm *opt,
d5fdd6ba 475 u8 *type, u8 *code, int *msg, __u32 *info, int offset)
1da177e4 476{
b71d1d42 477 const struct ipv6hdr *ipv6h = (const struct ipv6hdr *) skb->data;
1da177e4
LT		 478 struct ip6_tnl *t;
479 int rel_msg = 0;
d5fdd6ba
BH		 480 u8 rel_type = ICMPV6_DEST_UNREACH;
481 u8 rel_code = ICMPV6_ADDR_UNREACH;
acf722f7 482 u8 tproto;
1da177e4
LT		 483 __u32 rel_info = 0;
484 __u16 len;
d2acc347 485 int err = -ENOENT;
1da177e4 486
1ab1457c
YH		 487 /* If the packet doesn't contain the original IPv6 header we are
488 in trouble since we might need the source address for further
1da177e4
LT		 489 processing of the error. */
490
2922bc8a 491 rcu_read_lock();
e5d08d71 492 t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->daddr, &ipv6h->saddr);
63159f29 493 if (!t)
1da177e4
LT		 494 goto out;
495
acf722f7
AA		 496 tproto = ACCESS_ONCE(t->parms.proto);
497 if (tproto != ipproto && tproto != 0)
502b0935
YK		 498 goto out;
499
d2acc347
HX		 500 err = 0;
501
e490d1d8 502 switch (*type) {
1da177e4
LT		 503 __u32 teli;
504 struct ipv6_tlv_tnl_enc_lim *tel;
505 __u32 mtu;
506 case ICMPV6_DEST_UNREACH:
17a10c92
MB		 507 net_dbg_ratelimited("%s: Path to destination invalid or inactive!\n",
508 t->parms.name);
1da177e4
LT		 509 rel_msg = 1;
510 break;
511 case ICMPV6_TIME_EXCEED:
e490d1d8 512 if ((*code) == ICMPV6_EXC_HOPLIMIT) {
17a10c92
MB		 513 net_dbg_ratelimited("%s: Too small hop limit or routing loop in tunnel!\n",
514 t->parms.name);
1da177e4
LT		 515 rel_msg = 1;
516 }
517 break;
518 case ICMPV6_PARAMPROB:
107a5fe6 519 teli = 0;
e490d1d8 520 if ((*code) == ICMPV6_HDR_FIELD)
c12b395a 521 teli = ip6_tnl_parse_tlv_enc_lim(skb, skb->data);
1da177e4 522
704eae1f 523 if (teli && teli == *info - 2) {
1da177e4
LT		 524 tel = (struct ipv6_tlv_tnl_enc_lim *) &skb->data[teli];
525 if (tel->encap_limit == 0) {
17a10c92
MB		 526 net_dbg_ratelimited("%s: Too small encapsulation limit or routing loop in tunnel!\n",
527 t->parms.name);
1da177e4
LT		 528 rel_msg = 1;
529 }
e87cc472 530 } else {
17a10c92
MB		 531 net_dbg_ratelimited("%s: Recipient unable to parse tunneled packet!\n",
532 t->parms.name);
1da177e4
LT		 533 }
534 break;
535 case ICMPV6_PKT_TOOBIG:
704eae1f 536 mtu = *info - offset;
1da177e4
LT		 537 if (mtu < IPV6_MIN_MTU)
538 mtu = IPV6_MIN_MTU;
539 t->dev->mtu = mtu;
540
e5d08d71
IM		 541 len = sizeof(*ipv6h) + ntohs(ipv6h->payload_len);
542 if (len > mtu) {
1da177e4
LT		 543 rel_type = ICMPV6_PKT_TOOBIG;
544 rel_code = 0;
545 rel_info = mtu;
546 rel_msg = 1;
547 }
548 break;
549 }
e490d1d8
YK		 550
551 *type = rel_type;
552 *code = rel_code;
553 *info = rel_info;
554 *msg = rel_msg;
555
556out:
2922bc8a 557 rcu_read_unlock();
e490d1d8
YK		 558 return err;
559}
560
c4d3efaf
YK		 561static int
562ip4ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 563 u8 type, u8 code, int offset, __be32 info)
c4d3efaf
YK		 564{
565 int rel_msg = 0;
d5fdd6ba
BH		 566 u8 rel_type = type;
567 u8 rel_code = code;
704eae1f 568 __u32 rel_info = ntohl(info);
c4d3efaf
YK		 569 int err;
570 struct sk_buff *skb2;
b71d1d42 571 const struct iphdr *eiph;
c4d3efaf 572 struct rtable *rt;
31e4543d 573 struct flowi4 fl4;
c4d3efaf 574
502b0935
YK		 575 err = ip6_tnl_err(skb, IPPROTO_IPIP, opt, &rel_type, &rel_code,
576 &rel_msg, &rel_info, offset);
c4d3efaf
YK		 577 if (err < 0)
578 return err;
579
580 if (rel_msg == 0)
581 return 0;
582
583 switch (rel_type) {
584 case ICMPV6_DEST_UNREACH:
585 if (rel_code != ICMPV6_ADDR_UNREACH)
586 return 0;
587 rel_type = ICMP_DEST_UNREACH;
588 rel_code = ICMP_HOST_UNREACH;
589 break;
590 case ICMPV6_PKT_TOOBIG:
591 if (rel_code != 0)
592 return 0;
593 rel_type = ICMP_DEST_UNREACH;
594 rel_code = ICMP_FRAG_NEEDED;
595 break;
ec18d9a2
DM		 596 case NDISC_REDIRECT:
597 rel_type = ICMP_REDIRECT;
598 rel_code = ICMP_REDIR_HOST;
c4d3efaf
YK		 599 default:
600 return 0;
601 }
602
603 if (!pskb_may_pull(skb, offset + sizeof(struct iphdr)))
604 return 0;
605
606 skb2 = skb_clone(skb, GFP_ATOMIC);
607 if (!skb2)
608 return 0;
609
adf30907
ED		 610 skb_dst_drop(skb2);
611
c4d3efaf 612 skb_pull(skb2, offset);
c1d2bbe1 613 skb_reset_network_header(skb2);
eddc9ec5 614 eiph = ip_hdr(skb2);
c4d3efaf
YK		 615
616 /* Try to guess incoming interface */
31e4543d 617 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL,
78fbfd8a
DM		 618 eiph->saddr, 0,
619 0, 0,
620 IPPROTO_IPIP, RT_TOS(eiph->tos), 0);
b23dd4fe 621 if (IS_ERR(rt))
c4d3efaf
YK		 622 goto out;
623
d8d1f30b 624 skb2->dev = rt->dst.dev;
c4d3efaf
YK		 625
626 /* route "incoming" packet */
627 if (rt->rt_flags & RTCF_LOCAL) {
628 ip_rt_put(rt);
629 rt = NULL;
31e4543d 630 rt = ip_route_output_ports(dev_net(skb->dev), &fl4, NULL,
78fbfd8a
DM		 631 eiph->daddr, eiph->saddr,
632 0, 0,
633 IPPROTO_IPIP,
634 RT_TOS(eiph->tos), 0);
b23dd4fe 635 if (IS_ERR(rt) ||
bc280a1e 636 rt->dst.dev->type != ARPHRD_TUNNEL6) {
b23dd4fe
DM		 637 if (!IS_ERR(rt))
638 ip_rt_put(rt);
c4d3efaf
YK		 639 goto out;
640 }
b23dd4fe 641 skb_dst_set(skb2, &rt->dst);
c4d3efaf
YK		 642 } else {
643 ip_rt_put(rt);
644 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos,
645 skb2->dev) ||
bc280a1e 646 skb_dst(skb2)->dev->type != ARPHRD_TUNNEL6)
c4d3efaf
YK		 647 goto out;
648 }
649
650 /* change mtu on this route */
651 if (rel_type == ICMP_DEST_UNREACH && rel_code == ICMP_FRAG_NEEDED) {
adf30907 652 if (rel_info > dst_mtu(skb_dst(skb2)))
c4d3efaf
YK		 653 goto out;
654
6a3c946b 655 skb_dst_update_pmtu(skb2, rel_info);
c4d3efaf 656 }
1ed5c48f 657 if (rel_type == ICMP_REDIRECT)
6700c270 658 skb_dst(skb2)->ops->redirect(skb_dst(skb2), NULL, skb2);
c4d3efaf 659
704eae1f 660 icmp_send(skb2, rel_type, rel_code, htonl(rel_info));
c4d3efaf
YK		 661
662out:
663 kfree_skb(skb2);
664 return 0;
665}
666
e490d1d8
YK		 667static int
668ip6ip6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
d5fdd6ba 669 u8 type, u8 code, int offset, __be32 info)
e490d1d8
YK		 670{
671 int rel_msg = 0;
d5fdd6ba
BH		 672 u8 rel_type = type;
673 u8 rel_code = code;
704eae1f 674 __u32 rel_info = ntohl(info);
e490d1d8
YK		 675 int err;
676
502b0935
YK		 677 err = ip6_tnl_err(skb, IPPROTO_IPV6, opt, &rel_type, &rel_code,
678 &rel_msg, &rel_info, offset);
e490d1d8
YK		 679 if (err < 0)
680 return err;
681
682 if (rel_msg && pskb_may_pull(skb, offset + sizeof(struct ipv6hdr))) {
1da177e4
LT		 683 struct rt6_info *rt;
684 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
305d4b3c 685
1da177e4 686 if (!skb2)
e490d1d8 687 return 0;
1da177e4 688
adf30907 689 skb_dst_drop(skb2);
1da177e4 690 skb_pull(skb2, offset);
c1d2bbe1 691 skb_reset_network_header(skb2);
1da177e4
LT		 692
693 /* Try to guess incoming interface */
2f7f54b7
PE		 694 rt = rt6_lookup(dev_net(skb->dev), &ipv6_hdr(skb2)->saddr,
695 NULL, 0, 0);
1da177e4 696
d1918542
DM		 697 if (rt && rt->dst.dev)
698 skb2->dev = rt->dst.dev;
1da177e4 699
3ffe533c 700 icmpv6_send(skb2, rel_type, rel_code, rel_info);
1da177e4 701
94e187c0 702 ip6_rt_put(rt);
1da177e4
LT		 703
704 kfree_skb(skb2);
705 }
e490d1d8
YK		 706
707 return 0;
1da177e4
LT		 708}
709
f4e0b4c5
ND		 710static int ip4ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
711 const struct ipv6hdr *ipv6h,
712 struct sk_buff *skb)
c4d3efaf
YK		 713{
714 __u8 dsfield = ipv6_get_dsfield(ipv6h) & ~INET_ECN_MASK;
715
716 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
eddc9ec5 717 ipv4_change_dsfield(ip_hdr(skb), INET_ECN_MASK, dsfield);
c4d3efaf 718
f4e0b4c5 719 return IP6_ECN_decapsulate(ipv6h, skb);
c4d3efaf
YK		 720}
721
f4e0b4c5
ND		 722static int ip6ip6_dscp_ecn_decapsulate(const struct ip6_tnl *t,
723 const struct ipv6hdr *ipv6h,
724 struct sk_buff *skb)
1da177e4 725{
8359925b 726 if (t->parms.flags & IP6_TNL_F_RCV_DSCP_COPY)
29bb43b4 727 ipv6_copy_dscp(ipv6_get_dsfield(ipv6h), ipv6_hdr(skb));
1da177e4 728
f4e0b4c5 729 return IP6_ECN_decapsulate(ipv6h, skb);
1da177e4 730}
8359925b 731
c12b395a 732__u32 ip6_tnl_get_cap(struct ip6_tnl *t,
d0087b29
VN		 733 const struct in6_addr *laddr,
734 const struct in6_addr *raddr)
735{
c12b395a 736 struct __ip6_tnl_parm *p = &t->parms;
d0087b29
VN		 737 int ltype = ipv6_addr_type(laddr);
738 int rtype = ipv6_addr_type(raddr);
739 __u32 flags = 0;
740
741 if (ltype == IPV6_ADDR_ANY || rtype == IPV6_ADDR_ANY) {
742 flags = IP6_TNL_F_CAP_PER_PACKET;
743 } else if (ltype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
744 rtype & (IPV6_ADDR_UNICAST|IPV6_ADDR_MULTICAST) &&
745 !((ltype|rtype) & IPV6_ADDR_LOOPBACK) &&
746 (!((ltype|rtype) & IPV6_ADDR_LINKLOCAL) || p->link)) {
747 if (ltype&IPV6_ADDR_UNICAST)
748 flags |= IP6_TNL_F_CAP_XMIT;
749 if (rtype&IPV6_ADDR_UNICAST)
750 flags |= IP6_TNL_F_CAP_RCV;
751 }
752 return flags;
753}
c12b395a 754EXPORT_SYMBOL(ip6_tnl_get_cap);
d0087b29 755
f1a28eab 756/* called with rcu_read_lock() */
c12b395a 757int ip6_tnl_rcv_ctl(struct ip6_tnl *t,
d0087b29
VN		 758 const struct in6_addr *laddr,
759 const struct in6_addr *raddr)
09c6bbf0 760{
c12b395a 761 struct __ip6_tnl_parm *p = &t->parms;
09c6bbf0 762 int ret = 0;
0bd87628 763 struct net *net = t->net;
09c6bbf0 764
d0087b29
VN		 765 if ((p->flags & IP6_TNL_F_CAP_RCV) ||
766 ((p->flags & IP6_TNL_F_CAP_PER_PACKET) &&
767 (ip6_tnl_get_cap(t, laddr, raddr) & IP6_TNL_F_CAP_RCV))) {
1ab1457c 768 struct net_device *ldev = NULL;
09c6bbf0
VN		 769
770 if (p->link)
f1a28eab 771 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0 772
d0087b29
VN		 773 if ((ipv6_addr_is_multicast(laddr) ||
774 likely(ipv6_chk_addr(net, laddr, ldev, 0))) &&
775 likely(!ipv6_chk_addr(net, raddr, NULL, 0)))
09c6bbf0 776 ret = 1;
09c6bbf0
VN		 777 }
778 return ret;
779}
c12b395a 780EXPORT_SYMBOL_GPL(ip6_tnl_rcv_ctl);
1da177e4 781
0d3c703a
TH		 782static int __ip6_tnl_rcv(struct ip6_tnl *tunnel, struct sk_buff *skb,
783 const struct tnl_ptk_info *tpi,
784 struct metadata_dst *tun_dst,
785 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
786 const struct ipv6hdr *ipv6h,
787 struct sk_buff *skb),
788 bool log_ecn_err)
1da177e4 789{
0d3c703a 790 struct pcpu_sw_netstats *tstats;
b71d1d42 791 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
f4e0b4c5 792 int err;
1da177e4 793
0d3c703a
TH		 794 if ((!(tpi->flags & TUNNEL_CSUM) &&
795 (tunnel->parms.i_flags & TUNNEL_CSUM)) ||
796 ((tpi->flags & TUNNEL_CSUM) &&
797 !(tunnel->parms.i_flags & TUNNEL_CSUM))) {
798 tunnel->dev->stats.rx_crc_errors++;
799 tunnel->dev->stats.rx_errors++;
800 goto drop;
801 }
8560f226 802
0d3c703a
TH		 803 if (tunnel->parms.i_flags & TUNNEL_SEQ) {
804 if (!(tpi->flags & TUNNEL_SEQ) ||
805 (tunnel->i_seqno &&
806 (s32)(ntohl(tpi->seq) - tunnel->i_seqno) < 0)) {
807 tunnel->dev->stats.rx_fifo_errors++;
808 tunnel->dev->stats.rx_errors++;
809 goto drop;
502b0935 810 }
0d3c703a
TH		 811 tunnel->i_seqno = ntohl(tpi->seq) + 1;
812 }
502b0935 813
0d3c703a 814 skb->protocol = tpi->proto;
1da177e4 815
0d3c703a
TH		 816 /* Warning: All skb pointers will be invalidated! */
817 if (tunnel->dev->type == ARPHRD_ETHER) {
818 if (!pskb_may_pull(skb, ETH_HLEN)) {
819 tunnel->dev->stats.rx_length_errors++;
820 tunnel->dev->stats.rx_errors++;
821 goto drop;
1da177e4 822 }
0d3c703a
TH		 823
824 ipv6h = ipv6_hdr(skb);
825 skb->protocol = eth_type_trans(skb, tunnel->dev);
826 skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
827 } else {
828 skb->dev = tunnel->dev;
829 }
830
831 skb_reset_network_header(skb);
832 memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
833
834 __skb_tunnel_rx(skb, tunnel->dev, tunnel->net);
835
836 err = dscp_ecn_decapsulate(tunnel, ipv6h, skb);
837 if (unlikely(err)) {
838 if (log_ecn_err)
839 net_info_ratelimited("non-ECT from %pI6 with DS=%#x\n",
840 &ipv6h->saddr,
841 ipv6_get_dsfield(ipv6h));
842 if (err > 1) {
843 ++tunnel->dev->stats.rx_frame_errors;
844 ++tunnel->dev->stats.rx_errors;
845 goto drop;
f4e0b4c5 846 }
0d3c703a 847 }
f4e0b4c5 848
0d3c703a
TH		 849 tstats = this_cpu_ptr(tunnel->dev->tstats);
850 u64_stats_update_begin(&tstats->syncp);
851 tstats->rx_packets++;
852 tstats->rx_bytes += skb->len;
853 u64_stats_update_end(&tstats->syncp);
8560f226 854
0d3c703a 855 skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(tunnel->dev)));
8990f468 856
8d79266b
AS		 857 if (tun_dst)
858 skb_dst_set(skb, (struct dst_entry *)tun_dst);
859
0d3c703a
TH		 860 gro_cells_receive(&tunnel->gro_cells, skb);
861 return 0;
862
863drop:
f1925ca5
HY		 864 if (tun_dst)
865 dst_release((struct dst_entry *)tun_dst);
0d3c703a
TH		 866 kfree_skb(skb);
867 return 0;
868}
869
870int ip6_tnl_rcv(struct ip6_tnl *t, struct sk_buff *skb,
871 const struct tnl_ptk_info *tpi,
872 struct metadata_dst *tun_dst,
873 bool log_ecn_err)
874{
875 return __ip6_tnl_rcv(t, skb, tpi, NULL, ip6ip6_dscp_ecn_decapsulate,
876 log_ecn_err);
877}
878EXPORT_SYMBOL(ip6_tnl_rcv);
879
880static const struct tnl_ptk_info tpi_v6 = {
881 /* no tunnel info required for ipxip6. */
882 .proto = htons(ETH_P_IPV6),
883};
884
885static const struct tnl_ptk_info tpi_v4 = {
886 /* no tunnel info required for ipxip6. */
887 .proto = htons(ETH_P_IP),
888};
889
890static int ipxip6_rcv(struct sk_buff *skb, u8 ipproto,
891 const struct tnl_ptk_info *tpi,
892 int (*dscp_ecn_decapsulate)(const struct ip6_tnl *t,
893 const struct ipv6hdr *ipv6h,
894 struct sk_buff *skb))
895{
896 struct ip6_tnl *t;
897 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
8d79266b 898 struct metadata_dst *tun_dst = NULL;
0d3c703a
TH		 899 int ret = -1;
900
901 rcu_read_lock();
902 t = ip6_tnl_lookup(dev_net(skb->dev), &ipv6h->saddr, &ipv6h->daddr);
903
904 if (t) {
905 u8 tproto = ACCESS_ONCE(t->parms.proto);
906
907 if (tproto != ipproto && tproto != 0)
908 goto drop;
909 if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
910 goto drop;
64d22404 911 ipv6h = ipv6_hdr(skb);
0d3c703a
TH		 912 if (!ip6_tnl_rcv_ctl(t, &ipv6h->daddr, &ipv6h->saddr))
913 goto drop;
914 if (iptunnel_pull_header(skb, 0, tpi->proto, false))
915 goto drop;
8d79266b
AS		 916 if (t->parms.collect_md) {
917 tun_dst = ipv6_tun_rx_dst(skb, 0, 0, 0);
918 if (!tun_dst)
6d1c4898 919 goto drop;
8d79266b
AS		 920 }
921 ret = __ip6_tnl_rcv(t, skb, tpi, tun_dst, dscp_ecn_decapsulate,
0d3c703a 922 log_ecn_error);
1da177e4 923 }
0d3c703a 924
2922bc8a 925 rcu_read_unlock();
50fba2aa 926
0d3c703a
TH		 927 return ret;
928
929drop:
930 rcu_read_unlock();
50fba2aa
HX		 931 kfree_skb(skb);
932 return 0;
1da177e4
LT		 933}
934
c4d3efaf
YK		 935static int ip4ip6_rcv(struct sk_buff *skb)
936{
ca4aa976 937 return ipxip6_rcv(skb, IPPROTO_IPIP, &tpi_v4,
0d3c703a 938 ip4ip6_dscp_ecn_decapsulate);
c4d3efaf
YK		 939}
940
8359925b
YK		 941static int ip6ip6_rcv(struct sk_buff *skb)
942{
0d3c703a
TH		 943 return ipxip6_rcv(skb, IPPROTO_IPV6, &tpi_v6,
944 ip6ip6_dscp_ecn_decapsulate);
8359925b
YK		 945}
946
6fb32dde
VN		 947struct ipv6_tel_txoption {
948 struct ipv6_txoptions ops;
949 __u8 dst_opt[8];
950};
1da177e4 951
6fb32dde
VN		 952static void init_tel_txopt(struct ipv6_tel_txoption *opt, __u8 encap_limit)
953{
954 memset(opt, 0, sizeof(struct ipv6_tel_txoption));
1da177e4 955
6fb32dde
VN		 956 opt->dst_opt[2] = IPV6_TLV_TNL_ENCAP_LIMIT;
957 opt->dst_opt[3] = 1;
958 opt->dst_opt[4] = encap_limit;
959 opt->dst_opt[5] = IPV6_TLV_PADN;
960 opt->dst_opt[6] = 1;
1da177e4 961
89a23c8b 962 opt->ops.dst1opt = (struct ipv6_opt_hdr *) opt->dst_opt;
6fb32dde 963 opt->ops.opt_nflen = 8;
1da177e4
LT		 964}
965
966/**
3144581c 967 * ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
1da177e4 968 * @t: the outgoing tunnel device
1ab1457c 969 * @hdr: IPv6 header from the incoming packet
1da177e4
LT		 970 *
971 * Description:
1ab1457c 972 * Avoid trivial tunneling loop by checking that tunnel exit-point
1da177e4
LT		 973 * doesn't match source of incoming packet.
974 *
1ab1457c 975 * Return:
1da177e4
LT		 976 * 1 if conflict,
977 * 0 else
978 **/
979
92113bfd 980static inline bool
b71d1d42 981ip6_tnl_addr_conflict(const struct ip6_tnl *t, const struct ipv6hdr *hdr)
1da177e4
LT		 982{
983 return ipv6_addr_equal(&t->parms.raddr, &hdr->saddr);
984}
985
d5005140
SK		 986int ip6_tnl_xmit_ctl(struct ip6_tnl *t,
987 const struct in6_addr *laddr,
988 const struct in6_addr *raddr)
09c6bbf0 989{
c12b395a 990 struct __ip6_tnl_parm *p = &t->parms;
09c6bbf0 991 int ret = 0;
0bd87628 992 struct net *net = t->net;
09c6bbf0 993
d5005140
SK		 994 if ((p->flags & IP6_TNL_F_CAP_XMIT) ||
995 ((p->flags & IP6_TNL_F_CAP_PER_PACKET) &&
996 (ip6_tnl_get_cap(t, laddr, raddr) & IP6_TNL_F_CAP_XMIT))) {
09c6bbf0
VN		 997 struct net_device *ldev = NULL;
998
f1a28eab 999 rcu_read_lock();
09c6bbf0 1000 if (p->link)
f1a28eab 1001 ldev = dev_get_by_index_rcu(net, p->link);
09c6bbf0 1002
d5005140 1003 if (unlikely(!ipv6_chk_addr(net, laddr, ldev, 0)))
f3213831
JP		 1004 pr_warn("%s xmit: Local address not yet configured!\n",
1005 p->name);
d5005140
SK		 1006 else if (!ipv6_addr_is_multicast(raddr) &&
1007 unlikely(ipv6_chk_addr(net, raddr, NULL, 0)))
f3213831
JP		 1008 pr_warn("%s xmit: Routing loop! Remote address found on this node!\n",
1009 p->name);
09c6bbf0
VN		 1010 else
1011 ret = 1;
f1a28eab 1012 rcu_read_unlock();
09c6bbf0
VN		 1013 }
1014 return ret;
1015}
c12b395a 1016EXPORT_SYMBOL_GPL(ip6_tnl_xmit_ctl);
1017
1da177e4 1018/**
8eb30be0 1019 * ip6_tnl_xmit - encapsulate packet and send
1da177e4 1020 * @skb: the outgoing socket buffer
1ab1457c 1021 * @dev: the outgoing tunnel device
61ec2aec 1022 * @dsfield: dscp code for outer header
8eb30be0 1023 * @fl6: flow of tunneled packet
61ec2aec
YK		 1024 * @encap_limit: encapsulation limit
1025 * @pmtu: Path MTU is stored if packet is too big
8eb30be0 1026 * @proto: next header value
1da177e4
LT		 1027 *
1028 * Description:
1029 * Build new header and do some sanity checks on the packet before sending
1030 * it.
1031 *
1ab1457c 1032 * Return:
c4d3efaf 1033 * 0 on success
61ec2aec
YK		 1034 * -1 fail
1035 * %-EMSGSIZE message too big. return mtu in this case.
1da177e4
LT		 1036 **/
1037
8eb30be0
TH		 1038int ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev, __u8 dsfield,
1039 struct flowi6 *fl6, int encap_limit, __u32 *pmtu,
1040 __u8 proto)
1da177e4 1041{
2941a486 1042 struct ip6_tnl *t = netdev_priv(dev);
0bd87628 1043 struct net *net = t->net;
3dca02af 1044 struct net_device_stats *stats = &t->dev->stats;
199ab00f 1045 struct ipv6hdr *ipv6h;
6fb32dde 1046 struct ipv6_tel_txoption opt;
d24f22f3 1047 struct dst_entry *dst = NULL, *ndst = NULL;
1da177e4
LT		 1048 struct net_device *tdev;
1049 int mtu;
d41bb33b 1050 unsigned int eth_hlen = t->dev->type == ARPHRD_ETHER ? ETH_HLEN : 0;
058214a4
TH		 1051 unsigned int psh_hlen = sizeof(struct ipv6hdr) + t->encap_hlen;
1052 unsigned int max_headroom = psh_hlen;
b5c2d495 1053 bool use_cache = false;
8d79266b 1054 u8 hop_limit;
61ec2aec 1055 int err = -1;
1da177e4 1056
8d79266b
AS		 1057 if (t->parms.collect_md) {
1058 hop_limit = skb_tunnel_info(skb)->key.ttl;
1059 goto route_lookup;
1060 } else {
1061 hop_limit = t->parms.hop_limit;
1062 }
1063
ea3dc960
SK		 1064 /* NBMA tunnel */
1065 if (ipv6_addr_any(&t->parms.raddr)) {
199ab00f
WC		 1066 if (skb->protocol == htons(ETH_P_IPV6)) {
1067 struct in6_addr *addr6;
1068 struct neighbour *neigh;
1069 int addr_type;
ea3dc960 1070
199ab00f
WC		 1071 if (!skb_dst(skb))
1072 goto tx_err_link_failure;
ea3dc960 1073
199ab00f
WC		 1074 neigh = dst_neigh_lookup(skb_dst(skb),
1075 &ipv6_hdr(skb)->daddr);
1076 if (!neigh)
1077 goto tx_err_link_failure;
ea3dc960 1078
199ab00f
WC		 1079 addr6 = (struct in6_addr *)&neigh->primary_key;
1080 addr_type = ipv6_addr_type(addr6);
ea3dc960 1081
199ab00f
WC		 1082 if (addr_type == IPV6_ADDR_ANY)
1083 addr6 = &ipv6_hdr(skb)->daddr;
ea3dc960 1084
199ab00f
WC		 1085 memcpy(&fl6->daddr, addr6, sizeof(fl6->daddr));
1086 neigh_release(neigh);
1087 }
b841bff1
EC		 1088 } else if (t->parms.proto != 0 && !(t->parms.flags &
1089 (IP6_TNL_F_USE_ORIG_TCLASS |
1090 IP6_TNL_F_USE_ORIG_FWMARK))) {
1091 /* enable the cache only if neither the outer protocol nor the
1092 * routing decision depends on the current inner header value
b5c2d495
PA		 1093 */
1094 use_cache = true;
1095 }
1096
1097 if (use_cache)
607f725f 1098 dst = dst_cache_get(&t->dst_cache);
d5005140
SK		 1099
1100 if (!ip6_tnl_xmit_ctl(t, &fl6->saddr, &fl6->daddr))
1101 goto tx_err_link_failure;
1102
89b02126 1103 if (!dst) {
8d79266b 1104route_lookup:
5f733ee6
LM		 1105 /* add dsfield to flowlabel for route lookup */
1106 fl6->flowlabel = ip6_make_flowinfo(dsfield, fl6->flowlabel);
1107
cdf3464e 1108 dst = ip6_route_output(net, NULL, fl6);
1da177e4 1109
cdf3464e 1110 if (dst->error)
a57ebc90 1111 goto tx_err_link_failure;
cdf3464e
MKL		 1112 dst = xfrm_lookup(net, dst, flowi6_to_flowi(fl6), NULL, 0);
1113 if (IS_ERR(dst)) {
1114 err = PTR_ERR(dst);
1115 dst = NULL;
452edd59
DM		 1116 goto tx_err_link_failure;
1117 }
8d79266b
AS		 1118 if (t->parms.collect_md &&
1119 ipv6_dev_get_saddr(net, ip6_dst_idev(dst)->dev,
1120 &fl6->daddr, 0, &fl6->saddr))
1121 goto tx_err_link_failure;
cdf3464e 1122 ndst = dst;
a57ebc90 1123 }
1da177e4
LT		 1124
1125 tdev = dst->dev;
1126
1127 if (tdev == dev) {
1128 stats->collisions++;
e87cc472
JP		 1129 net_warn_ratelimited("%s: Local routing loop detected!\n",
1130 t->parms.name);
1da177e4
LT		 1131 goto tx_err_dst_release;
1132 }
d41bb33b 1133 mtu = dst_mtu(dst) - eth_hlen - psh_hlen - t->tun_hlen;
6fb32dde 1134 if (encap_limit >= 0) {
1da177e4
LT		 1135 max_headroom += 8;
1136 mtu -= 8;
1137 }
b3fb9398
XL		 1138 mtu = max(mtu, skb->protocol == htons(ETH_P_IPV6) ?
1139 IPV6_MIN_MTU : IPV4_MIN_MTU);
3eeac1d1 1140
6a3c946b 1141 skb_dst_update_pmtu(skb, mtu);
d41bb33b 1142 if (skb->len - t->tun_hlen - eth_hlen > mtu && !skb_is_gso(skb)) {
61ec2aec
YK		 1143 *pmtu = mtu;
1144 err = -EMSGSIZE;
1da177e4
LT		 1145 goto tx_err_dst_release;
1146 }
1147
8eb30be0
TH		 1148 if (t->err_count > 0) {
1149 if (time_before(jiffies,
1150 t->err_time + IP6TUNNEL_ERR_TIMEO)) {
1151 t->err_count--;
1152
1153 dst_link_failure(skb);
1154 } else {
1155 t->err_count = 0;
1156 }
1157 }
1158
963a88b3 1159 skb_scrub_packet(skb, !net_eq(t->net, dev_net(dev)));
0bd87628 1160
1da177e4
LT		 1161 /*
1162 * Okay, now see if we can stuff it in the buffer as-is.
1163 */
1164 max_headroom += LL_RESERVED_SPACE(tdev);
1ab1457c 1165
cfbba49d
PM		 1166 if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
1167 (skb_cloned(skb) && !skb_clone_writable(skb, 0))) {
1da177e4 1168 struct sk_buff *new_skb;
1ab1457c 1169
e5d08d71
IM		 1170 new_skb = skb_realloc_headroom(skb, max_headroom);
1171 if (!new_skb)
1da177e4
LT		 1172 goto tx_err_dst_release;
1173
1174 if (skb->sk)
1175 skb_set_owner_w(new_skb, skb->sk);
9ff26449 1176 consume_skb(skb);
1da177e4
LT		 1177 skb = new_skb;
1178 }
cdf3464e 1179
8d79266b
AS		 1180 if (t->parms.collect_md) {
1181 if (t->encap.type != TUNNEL_ENCAP_NONE)
1182 goto tx_err_dst_release;
1183 } else {
b5c2d495 1184 if (use_cache && ndst)
8d79266b
AS		 1185 dst_cache_set_ip6(&t->dst_cache, ndst, &fl6->saddr);
1186 }
cdf3464e
MKL		 1187 skb_dst_set(skb, dst);
1188
18e1173d 1189 hop_limit = hop_limit ? : ip6_dst_hoplimit(dst);
3d483058 1190
058214a4
TH		 1191 /* Calculate max headroom for all the headers and adjust
1192 * needed_headroom if necessary.
1193 */
8eb30be0 1194 max_headroom = LL_RESERVED_SPACE(dst->dev) + sizeof(struct ipv6hdr)
058214a4 1195 + dst->header_len + t->hlen;
8eb30be0
TH		 1196 if (max_headroom > dev->needed_headroom)
1197 dev->needed_headroom = max_headroom;
1198
058214a4
TH		 1199 err = ip6_tnl_encap(skb, t, &proto, fl6);
1200 if (err)
1201 return err;
1202
f288424e
SB		 1203 if (encap_limit >= 0) {
1204 init_tel_txopt(&opt, encap_limit);
1205 ipv6_push_frag_opts(skb, &opt.ops, &proto);
1206 }
1207
e2d1bca7
ACM		 1208 skb_push(skb, sizeof(struct ipv6hdr));
1209 skb_reset_network_header(skb);
0660e03f 1210 ipv6h = ipv6_hdr(skb);
0e9a7095 1211 ip6_flow_hdr(ipv6h, dsfield,
42240901 1212 ip6_make_flowlabel(net, skb, fl6->flowlabel, true, fl6));
8d79266b 1213 ipv6h->hop_limit = hop_limit;
1da177e4 1214 ipv6h->nexthdr = proto;
4e3fd7a0
AD		 1215 ipv6h->saddr = fl6->saddr;
1216 ipv6h->daddr = fl6->daddr;
79b16aad 1217 ip6tunnel_xmit(NULL, skb, dev);
1da177e4
LT		 1218 return 0;
1219tx_err_link_failure:
1220 stats->tx_carrier_errors++;
1221 dst_link_failure(skb);
1222tx_err_dst_release:
cdf3464e 1223 dst_release(dst);
61ec2aec
YK		 1224 return err;
1225}
8eb30be0 1226EXPORT_SYMBOL(ip6_tnl_xmit);
61ec2aec 1227
c4d3efaf
YK		 1228static inline int
1229ip4ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1230{
1231 struct ip6_tnl *t = netdev_priv(dev);
18bf9a72 1232 const struct iphdr *iph;
c4d3efaf 1233 int encap_limit = -1;
4c9483b2 1234 struct flowi6 fl6;
c4d3efaf
YK		 1235 __u8 dsfield;
1236 __u32 mtu;
acf722f7 1237 u8 tproto;
c4d3efaf
YK		 1238 int err;
1239
18bf9a72
PA		 1240 /* ensure we can access the full inner ip header */
1241 if (!pskb_may_pull(skb, sizeof(struct iphdr)))
1242 return -1;
1243
1244 iph = ip_hdr(skb);
5146d1f1
BH		 1245 memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
1246
acf722f7 1247 tproto = ACCESS_ONCE(t->parms.proto);
d5005140 1248 if (tproto != IPPROTO_IPIP && tproto != 0)
c4d3efaf
YK		 1249 return -1;
1250
8d79266b
AS		 1251 if (t->parms.collect_md) {
1252 struct ip_tunnel_info *tun_info;
1253 const struct ip_tunnel_key *key;
c4d3efaf 1254
8d79266b
AS		 1255 tun_info = skb_tunnel_info(skb);
1256 if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
1257 ip_tunnel_info_af(tun_info) != AF_INET6))
1258 return -1;
1259 key = &tun_info->key;
1260 memset(&fl6, 0, sizeof(fl6));
1261 fl6.flowi6_proto = IPPROTO_IPIP;
1262 fl6.daddr = key->u.ipv6.dst;
1263 fl6.flowlabel = key->label;
46f8cd9d 1264 dsfield = key->tos;
8d79266b
AS		 1265 } else {
1266 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1267 encap_limit = t->parms.encap_limit;
c4d3efaf 1268
8d79266b
AS		 1269 memcpy(&fl6, &t->fl.u.ip6, sizeof(fl6));
1270 fl6.flowi6_proto = IPPROTO_IPIP;
1271
1272 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
0e9a7095
PD		 1273 dsfield = ipv4_get_dsfield(iph);
1274 else
1275 dsfield = ip6_tclass(t->parms.flowinfo);
8d79266b
AS		 1276 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
1277 fl6.flowi6_mark = skb->mark;
0a473b82
CG		 1278 else
1279 fl6.flowi6_mark = t->parms.fwmark;
8d79266b 1280 }
c4d3efaf 1281
e2d118a1
LC		 1282 fl6.flowi6_uid = sock_net_uid(dev_net(dev), NULL);
1283
b8921ca8
TH		 1284 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6))
1285 return -1;
1286
0e9a7095
PD		 1287 dsfield = INET_ECN_encapsulate(dsfield, ipv4_get_dsfield(iph));
1288
b8921ca8
TH		 1289 skb_set_inner_ipproto(skb, IPPROTO_IPIP);
1290
8eb30be0
TH		 1291 err = ip6_tnl_xmit(skb, dev, dsfield, &fl6, encap_limit, &mtu,
1292 IPPROTO_IPIP);
c4d3efaf
YK		 1293 if (err != 0) {
1294 /* XXX: send ICMP error even if DF is not set. */
1295 if (err == -EMSGSIZE)
1296 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
1297 htonl(mtu));
1298 return -1;
1299 }
1300
1301 return 0;
1302}
1303
61ec2aec
YK		 1304static inline int
1305ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
1306{
1307 struct ip6_tnl *t = netdev_priv(dev);
18bf9a72 1308 struct ipv6hdr *ipv6h;
61ec2aec
YK		 1309 int encap_limit = -1;
1310 __u16 offset;
4c9483b2 1311 struct flowi6 fl6;
61ec2aec
YK		 1312 __u8 dsfield;
1313 __u32 mtu;
acf722f7 1314 u8 tproto;
61ec2aec
YK		 1315 int err;
1316
18bf9a72
PA		 1317 if (unlikely(!pskb_may_pull(skb, sizeof(*ipv6h))))
1318 return -1;
1319
1320 ipv6h = ipv6_hdr(skb);
acf722f7
AA		 1321 tproto = ACCESS_ONCE(t->parms.proto);
1322 if ((tproto != IPPROTO_IPV6 && tproto != 0) ||
d5005140 1323 ip6_tnl_addr_conflict(t, ipv6h))
61ec2aec
YK		 1324 return -1;
1325
8d79266b
AS		 1326 if (t->parms.collect_md) {
1327 struct ip_tunnel_info *tun_info;
1328 const struct ip_tunnel_key *key;
1329
1330 tun_info = skb_tunnel_info(skb);
1331 if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
1332 ip_tunnel_info_af(tun_info) != AF_INET6))
61ec2aec 1333 return -1;
8d79266b
AS		 1334 key = &tun_info->key;
1335 memset(&fl6, 0, sizeof(fl6));
1336 fl6.flowi6_proto = IPPROTO_IPV6;
1337 fl6.daddr = key->u.ipv6.dst;
1338 fl6.flowlabel = key->label;
46f8cd9d 1339 dsfield = key->tos;
8d79266b
AS		 1340 } else {
1341 offset = ip6_tnl_parse_tlv_enc_lim(skb, skb_network_header(skb));
21b995a9
ED		 1342 /* ip6_tnl_parse_tlv_enc_lim() might have reallocated skb->head */
1343 ipv6h = ipv6_hdr(skb);
8d79266b
AS		 1344 if (offset > 0) {
1345 struct ipv6_tlv_tnl_enc_lim *tel;
1346
1347 tel = (void *)&skb_network_header(skb)[offset];
1348 if (tel->encap_limit == 0) {
1349 icmpv6_send(skb, ICMPV6_PARAMPROB,
1350 ICMPV6_HDR_FIELD, offset + 2);
1351 return -1;
1352 }
1353 encap_limit = tel->encap_limit - 1;
1354 } else if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT)) {
1355 encap_limit = t->parms.encap_limit;
61ec2aec 1356 }
61ec2aec 1357
8d79266b
AS		 1358 memcpy(&fl6, &t->fl.u.ip6, sizeof(fl6));
1359 fl6.flowi6_proto = IPPROTO_IPV6;
61ec2aec 1360
8d79266b 1361 if (t->parms.flags & IP6_TNL_F_USE_ORIG_TCLASS)
0e9a7095
PD		 1362 dsfield = ipv6_get_dsfield(ipv6h);
1363 else
1364 dsfield = ip6_tclass(t->parms.flowinfo);
8d79266b
AS		 1365 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FLOWLABEL)
1366 fl6.flowlabel |= ip6_flowlabel(ipv6h);
1367 if (t->parms.flags & IP6_TNL_F_USE_ORIG_FWMARK)
1368 fl6.flowi6_mark = skb->mark;
0a473b82
CG		 1369 else
1370 fl6.flowi6_mark = t->parms.fwmark;
8d79266b 1371 }
61ec2aec 1372
e2d118a1
LC		 1373 fl6.flowi6_uid = sock_net_uid(dev_net(dev), NULL);
1374
815d22e5
TH		 1375 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP6))
1376 return -1;
1377
0e9a7095
PD		 1378 dsfield = INET_ECN_encapsulate(dsfield, ipv6_get_dsfield(ipv6h));
1379
815d22e5
TH		 1380 skb_set_inner_ipproto(skb, IPPROTO_IPV6);
1381
8eb30be0
TH		 1382 err = ip6_tnl_xmit(skb, dev, dsfield, &fl6, encap_limit, &mtu,
1383 IPPROTO_IPV6);
61ec2aec
YK		 1384 if (err != 0) {
1385 if (err == -EMSGSIZE)
3ffe533c 1386 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
61ec2aec
YK		 1387 return -1;
1388 }
1389
1390 return 0;
1391}
1392
6fef4c0c 1393static netdev_tx_t
8eb30be0 1394ip6_tnl_start_xmit(struct sk_buff *skb, struct net_device *dev)
61ec2aec
YK		 1395{
1396 struct ip6_tnl *t = netdev_priv(dev);
3dca02af 1397 struct net_device_stats *stats = &t->dev->stats;
61ec2aec
YK		 1398 int ret;
1399
61ec2aec 1400 switch (skb->protocol) {
60678040 1401 case htons(ETH_P_IP):
c4d3efaf
YK		 1402 ret = ip4ip6_tnl_xmit(skb, dev);
1403 break;
60678040 1404 case htons(ETH_P_IPV6):
61ec2aec
YK		 1405 ret = ip6ip6_tnl_xmit(skb, dev);
1406 break;
1407 default:
1408 goto tx_err;
1409 }
1410
1411 if (ret < 0)
1412 goto tx_err;
1413
6ed10654 1414 return NETDEV_TX_OK;
61ec2aec 1415
1da177e4
LT		 1416tx_err:
1417 stats->tx_errors++;
1418 stats->tx_dropped++;
1419 kfree_skb(skb);
6ed10654 1420 return NETDEV_TX_OK;
1da177e4
LT		 1421}
1422
3144581c 1423static void ip6_tnl_link_config(struct ip6_tnl *t)
1da177e4
LT		 1424{
1425 struct net_device *dev = t->dev;
c12b395a 1426 struct __ip6_tnl_parm *p = &t->parms;
4c9483b2 1427 struct flowi6 *fl6 = &t->fl.u.ip6;
058214a4 1428 int t_hlen;
1da177e4 1429
3a6d54c5
JP		 1430 memcpy(dev->dev_addr, &p->laddr, sizeof(struct in6_addr));
1431 memcpy(dev->broadcast, &p->raddr, sizeof(struct in6_addr));
1da177e4
LT		 1432
1433 /* Set up flowi template */
4e3fd7a0
AD		 1434 fl6->saddr = p->laddr;
1435 fl6->daddr = p->raddr;
4c9483b2
DM		 1436 fl6->flowi6_oif = p->link;
1437 fl6->flowlabel = 0;
1da177e4
LT		 1438
1439 if (!(p->flags&IP6_TNL_F_USE_ORIG_TCLASS))
4c9483b2 1440 fl6->flowlabel |= IPV6_TCLASS_MASK & p->flowinfo;
1da177e4 1441 if (!(p->flags&IP6_TNL_F_USE_ORIG_FLOWLABEL))
4c9483b2 1442 fl6->flowlabel |= IPV6_FLOWLABEL_MASK & p->flowinfo;
1da177e4 1443
d0087b29
VN		 1444 p->flags &= ~(IP6_TNL_F_CAP_XMIT|IP6_TNL_F_CAP_RCV|IP6_TNL_F_CAP_PER_PACKET);
1445 p->flags |= ip6_tnl_get_cap(t, &p->laddr, &p->raddr);
1da177e4
LT		 1446
1447 if (p->flags&IP6_TNL_F_CAP_XMIT && p->flags&IP6_TNL_F_CAP_RCV)
1448 dev->flags |= IFF_POINTOPOINT;
1449 else
1450 dev->flags &= ~IFF_POINTOPOINT;
1451
058214a4
TH		 1452 t->tun_hlen = 0;
1453 t->hlen = t->encap_hlen + t->tun_hlen;
1454 t_hlen = t->hlen + sizeof(struct ipv6hdr);
1455
1da177e4 1456 if (p->flags & IP6_TNL_F_CAP_XMIT) {
305d4b3c
VN		 1457 int strict = (ipv6_addr_type(&p->raddr) &
1458 (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL));
1459
0bd87628 1460 struct rt6_info *rt = rt6_lookup(t->net,
2f7f54b7 1461 &p->raddr, &p->laddr,
305d4b3c 1462 p->link, strict);
1da177e4 1463
63159f29 1464 if (!rt)
1da177e4
LT		 1465 return;
1466
d1918542
DM		 1467 if (rt->dst.dev) {
1468 dev->hard_header_len = rt->dst.dev->hard_header_len +
058214a4 1469 t_hlen;
1da177e4 1470
058214a4 1471 dev->mtu = rt->dst.dev->mtu - t_hlen;
381601e5 1472 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
67ba4152 1473 dev->mtu -= 8;
1da177e4
LT		 1474
1475 if (dev->mtu < IPV6_MIN_MTU)
1476 dev->mtu = IPV6_MIN_MTU;
1477 }
94e187c0 1478 ip6_rt_put(rt);
1da177e4
LT		 1479 }
1480}
1481
1482/**
3144581c 1483 * ip6_tnl_change - update the tunnel parameters
1da177e4
LT		 1484 * @t: tunnel to be changed
1485 * @p: tunnel configuration parameters
1da177e4
LT		 1486 *
1487 * Description:
3144581c 1488 * ip6_tnl_change() updates the tunnel parameters
1da177e4
LT		 1489 **/
1490
1491static int
c12b395a 1492ip6_tnl_change(struct ip6_tnl *t, const struct __ip6_tnl_parm *p)
1da177e4 1493{
4e3fd7a0
AD		 1494 t->parms.laddr = p->laddr;
1495 t->parms.raddr = p->raddr;
1da177e4
LT		 1496 t->parms.flags = p->flags;
1497 t->parms.hop_limit = p->hop_limit;
1498 t->parms.encap_limit = p->encap_limit;
1499 t->parms.flowinfo = p->flowinfo;
8181b8c1 1500 t->parms.link = p->link;
502b0935 1501 t->parms.proto = p->proto;
0a473b82 1502 t->parms.fwmark = p->fwmark;
607f725f 1503 dst_cache_reset(&t->dst_cache);
3144581c 1504 ip6_tnl_link_config(t);
1da177e4
LT		 1505 return 0;
1506}
1507
0b112457
ND		 1508static int ip6_tnl_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
1509{
0bd87628 1510 struct net *net = t->net;
0b112457
ND		 1511 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1512 int err;
1513
1514 ip6_tnl_unlink(ip6n, t);
1515 synchronize_net();
1516 err = ip6_tnl_change(t, p);
1517 ip6_tnl_link(ip6n, t);
1518 netdev_state_change(t->dev);
1519 return err;
1520}
1521
acf722f7
AA		 1522static int ip6_tnl0_update(struct ip6_tnl *t, struct __ip6_tnl_parm *p)
1523{
1524 /* for default tnl0 device allow to change only the proto */
1525 t->parms.proto = p->proto;
1526 netdev_state_change(t->dev);
1527 return 0;
1528}
1529
c12b395a 1530static void
1531ip6_tnl_parm_from_user(struct __ip6_tnl_parm *p, const struct ip6_tnl_parm *u)
1532{
1533 p->laddr = u->laddr;
1534 p->raddr = u->raddr;
1535 p->flags = u->flags;
1536 p->hop_limit = u->hop_limit;
1537 p->encap_limit = u->encap_limit;
1538 p->flowinfo = u->flowinfo;
1539 p->link = u->link;
1540 p->proto = u->proto;
1541 memcpy(p->name, u->name, sizeof(u->name));
1542}
1543
1544static void
1545ip6_tnl_parm_to_user(struct ip6_tnl_parm *u, const struct __ip6_tnl_parm *p)
1546{
1547 u->laddr = p->laddr;
1548 u->raddr = p->raddr;
1549 u->flags = p->flags;
1550 u->hop_limit = p->hop_limit;
1551 u->encap_limit = p->encap_limit;
1552 u->flowinfo = p->flowinfo;
1553 u->link = p->link;
1554 u->proto = p->proto;
1555 memcpy(u->name, p->name, sizeof(u->name));
1556}
1557
1da177e4 1558/**
3144581c 1559 * ip6_tnl_ioctl - configure ipv6 tunnels from userspace
1da177e4
LT		 1560 * @dev: virtual device associated with tunnel
1561 * @ifr: parameters passed from userspace
1562 * @cmd: command to be performed
1563 *
1564 * Description:
3144581c 1565 * ip6_tnl_ioctl() is used for managing IPv6 tunnels
1ab1457c 1566 * from userspace.
1da177e4
LT		 1567 *
1568 * The possible commands are the following:
1569 * %SIOCGETTUNNEL: get tunnel parameters for device
1570 * %SIOCADDTUNNEL: add tunnel matching given tunnel parameters
1571 * %SIOCCHGTUNNEL: change tunnel parameters to those given
1572 * %SIOCDELTUNNEL: delete tunnel
1573 *
1ab1457c 1574 * The fallback device "ip6tnl0", created during module
1da177e4
LT		 1575 * initialization, can be used for creating other tunnel devices.
1576 *
1577 * Return:
1578 * 0 on success,
1579 * %-EFAULT if unable to copy data to or from userspace,
1580 * %-EPERM if current process hasn't %CAP_NET_ADMIN set
1581 * %-EINVAL if passed tunnel parameters are invalid,
1582 * %-EEXIST if changing a tunnel's parameters would cause a conflict
1583 * %-ENODEV if attempting to change or delete a nonexisting device
1584 **/
1585
1586static int
3144581c 1587ip6_tnl_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
1da177e4
LT		 1588{
1589 int err = 0;
1da177e4 1590 struct ip6_tnl_parm p;
c12b395a 1591 struct __ip6_tnl_parm p1;
74462f0d
ND		 1592 struct ip6_tnl *t = netdev_priv(dev);
1593 struct net *net = t->net;
2dd02c89 1594 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1da177e4 1595
0d3c703a
TH		 1596 memset(&p1, 0, sizeof(p1));
1597
1da177e4
LT		 1598 switch (cmd) {
1599 case SIOCGETTUNNEL:
15820e12 1600 if (dev == ip6n->fb_tnl_dev) {
67ba4152 1601 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) {
1da177e4
LT		 1602 err = -EFAULT;
1603 break;
1604 }
c12b395a 1605 ip6_tnl_parm_from_user(&p1, &p);
1606 t = ip6_tnl_locate(net, &p1, 0);
37355565 1607 if (IS_ERR(t))
74462f0d 1608 t = netdev_priv(dev);
5ef5d6c5
DC		 1609 } else {
1610 memset(&p, 0, sizeof(p));
567131a7 1611 }
c12b395a 1612 ip6_tnl_parm_to_user(&p, &t->parms);
67ba4152 1613 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p))) {
1da177e4
LT		 1614 err = -EFAULT;
1615 }
1616 break;
1617 case SIOCADDTUNNEL:
1618 case SIOCCHGTUNNEL:
1619 err = -EPERM;
af31f412 1620 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4 1621 break;
567131a7 1622 err = -EFAULT;
67ba4152 1623 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
1da177e4 1624 break;
567131a7 1625 err = -EINVAL;
502b0935
YK		 1626 if (p.proto != IPPROTO_IPV6 && p.proto != IPPROTO_IPIP &&
1627 p.proto != 0)
1da177e4 1628 break;
c12b395a 1629 ip6_tnl_parm_from_user(&p1, &p);
1630 t = ip6_tnl_locate(net, &p1, cmd == SIOCADDTUNNEL);
acf722f7 1631 if (cmd == SIOCCHGTUNNEL) {
37355565 1632 if (!IS_ERR(t)) {
567131a7
VN		 1633 if (t->dev != dev) {
1634 err = -EEXIST;
1635 break;
1636 }
1637 } else
1638 t = netdev_priv(dev);
acf722f7
AA		 1639 if (dev == ip6n->fb_tnl_dev)
1640 err = ip6_tnl0_update(t, &p1);
1641 else
1642 err = ip6_tnl_update(t, &p1);
1da177e4 1643 }
37355565 1644 if (!IS_ERR(t)) {
1da177e4 1645 err = 0;
c12b395a 1646 ip6_tnl_parm_to_user(&p, &t->parms);
1647 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
567131a7
VN		 1648 err = -EFAULT;
1649
37355565
ND		 1650 } else {
1651 err = PTR_ERR(t);
1652 }
1da177e4
LT		 1653 break;
1654 case SIOCDELTUNNEL:
1655 err = -EPERM;
af31f412 1656 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 1657 break;
1658
15820e12 1659 if (dev == ip6n->fb_tnl_dev) {
567131a7 1660 err = -EFAULT;
67ba4152 1661 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
1da177e4 1662 break;
567131a7 1663 err = -ENOENT;
c12b395a 1664 ip6_tnl_parm_from_user(&p1, &p);
1665 t = ip6_tnl_locate(net, &p1, 0);
37355565 1666 if (IS_ERR(t))
1da177e4 1667 break;
567131a7 1668 err = -EPERM;
15820e12 1669 if (t->dev == ip6n->fb_tnl_dev)
1da177e4 1670 break;
567131a7 1671 dev = t->dev;
1da177e4 1672 }
22f8cde5
SH		 1673 err = 0;
1674 unregister_netdevice(dev);
1da177e4
LT		 1675 break;
1676 default:
1677 err = -EINVAL;
1678 }
1679 return err;
1680}
1681
1da177e4 1682/**
3144581c 1683 * ip6_tnl_change_mtu - change mtu manually for tunnel device
1da177e4
LT		 1684 * @dev: virtual device associated with tunnel
1685 * @new_mtu: the new mtu
1686 *
1687 * Return:
1688 * 0 on success,
1689 * %-EINVAL if mtu too small
1690 **/
1691
79ecb90e 1692int ip6_tnl_change_mtu(struct net_device *dev, int new_mtu)
1da177e4 1693{
582442d6
OG		 1694 struct ip6_tnl *tnl = netdev_priv(dev);
1695
4132ec35
XL		 1696 if (tnl->parms.proto == IPPROTO_IPV6) {
1697 if (new_mtu < IPV6_MIN_MTU)
582442d6
OG		 1698 return -EINVAL;
1699 } else {
4132ec35 1700 if (new_mtu < ETH_MIN_MTU)
582442d6 1701 return -EINVAL;
1da177e4 1702 }
f200de49
ND		 1703 if (tnl->parms.proto == IPPROTO_IPV6 || tnl->parms.proto == 0) {
1704 if (new_mtu > IP6_MAX_MTU - dev->hard_header_len)
1705 return -EINVAL;
1706 } else {
1707 if (new_mtu > IP_MAX_MTU - dev->hard_header_len)
1708 return -EINVAL;
1709 }
1da177e4
LT		 1710 dev->mtu = new_mtu;
1711 return 0;
1712}
79ecb90e 1713EXPORT_SYMBOL(ip6_tnl_change_mtu);
1da177e4 1714
ecf2c06a
ND		 1715int ip6_tnl_get_iflink(const struct net_device *dev)
1716{
1717 struct ip6_tnl *t = netdev_priv(dev);
1718
1719 return t->parms.link;
1720}
1721EXPORT_SYMBOL(ip6_tnl_get_iflink);
1326c3d5 1722
058214a4
TH		 1723int ip6_tnl_encap_add_ops(const struct ip6_tnl_encap_ops *ops,
1724 unsigned int num)
1725{
1726 if (num >= MAX_IPTUN_ENCAP_OPS)
1727 return -ERANGE;
1728
1729 return !cmpxchg((const struct ip6_tnl_encap_ops **)
1730 &ip6tun_encaps[num],
1731 NULL, ops) ? 0 : -1;
1732}
1733EXPORT_SYMBOL(ip6_tnl_encap_add_ops);
1734
1735int ip6_tnl_encap_del_ops(const struct ip6_tnl_encap_ops *ops,
1736 unsigned int num)
1737{
1738 int ret;
1739
1740 if (num >= MAX_IPTUN_ENCAP_OPS)
1741 return -ERANGE;
1742
1743 ret = (cmpxchg((const struct ip6_tnl_encap_ops **)
1744 &ip6tun_encaps[num],
1745 ops, NULL) == ops) ? 0 : -1;
1746
1747 synchronize_net();
1748
1749 return ret;
1750}
1751EXPORT_SYMBOL(ip6_tnl_encap_del_ops);
1752
1753int ip6_tnl_encap_setup(struct ip6_tnl *t,
1754 struct ip_tunnel_encap *ipencap)
1755{
1756 int hlen;
1757
1758 memset(&t->encap, 0, sizeof(t->encap));
1759
1760 hlen = ip6_encap_hlen(ipencap);
1761 if (hlen < 0)
1762 return hlen;
1763
1764 t->encap.type = ipencap->type;
1765 t->encap.sport = ipencap->sport;
1766 t->encap.dport = ipencap->dport;
1767 t->encap.flags = ipencap->flags;
1768
1769 t->encap_hlen = hlen;
1770 t->hlen = t->encap_hlen + t->tun_hlen;
1771
1772 return 0;
1773}
1774EXPORT_SYMBOL_GPL(ip6_tnl_encap_setup);
1775
1326c3d5 1776static const struct net_device_ops ip6_tnl_netdev_ops = {
6c6151da 1777 .ndo_init = ip6_tnl_dev_init,
8560f226 1778 .ndo_uninit = ip6_tnl_dev_uninit,
8eb30be0 1779 .ndo_start_xmit = ip6_tnl_start_xmit,
8560f226 1780 .ndo_do_ioctl = ip6_tnl_ioctl,
1326c3d5 1781 .ndo_change_mtu = ip6_tnl_change_mtu,
8560f226 1782 .ndo_get_stats = ip6_get_stats,
ecf2c06a 1783 .ndo_get_iflink = ip6_tnl_get_iflink,
1326c3d5
SH		 1784};
1785
51c052d4
TH		 1786#define IPXIPX_FEATURES (NETIF_F_SG | \
1787 NETIF_F_FRAGLIST | \
1788 NETIF_F_HIGHDMA | \
1789 NETIF_F_GSO_SOFTWARE | \
1790 NETIF_F_HW_CSUM)
8560f226 1791
1da177e4 1792/**
3144581c 1793 * ip6_tnl_dev_setup - setup virtual tunnel device
1da177e4
LT		 1794 * @dev: virtual device associated with tunnel
1795 *
1796 * Description:
1797 * Initialize function pointers and device parameters
1798 **/
1799
3144581c 1800static void ip6_tnl_dev_setup(struct net_device *dev)
1da177e4 1801{
1326c3d5 1802 dev->netdev_ops = &ip6_tnl_netdev_ops;
cf124db5
DM		 1803 dev->needs_free_netdev = true;
1804 dev->priv_destructor = ip6_dev_free;
1da177e4
LT		 1805
1806 dev->type = ARPHRD_TUNNEL6;
1da177e4
LT		 1807 dev->flags |= IFF_NOARP;
1808 dev->addr_len = sizeof(struct in6_addr);
058214a4 1809 dev->features |= NETIF_F_LLTX;
02875878 1810 netif_keep_dst(dev);
51c052d4
TH		 1811
1812 dev->features |= IPXIPX_FEATURES;
1813 dev->hw_features |= IPXIPX_FEATURES;
1814
e837735e
ND		 1815 /* This perm addr will be used as interface identifier by IPv6 */
1816 dev->addr_assign_type = NET_ADDR_RANDOM;
1817 eth_random_addr(dev->perm_addr);
1da177e4
LT		 1818}
1819
1820
1821/**
3144581c 1822 * ip6_tnl_dev_init_gen - general initializer for all tunnel devices
1da177e4
LT		 1823 * @dev: virtual device associated with tunnel
1824 **/
1825
8560f226 1826static inline int
3144581c 1827ip6_tnl_dev_init_gen(struct net_device *dev)
1da177e4 1828{
2941a486 1829 struct ip6_tnl *t = netdev_priv(dev);
cdf3464e 1830 int ret;
058214a4 1831 int t_hlen;
8560f226 1832
1da177e4 1833 t->dev = dev;
0bd87628 1834 t->net = dev_net(dev);
1c213bd2 1835 dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
8560f226
ED		 1836 if (!dev->tstats)
1837 return -ENOMEM;
cdf3464e 1838
607f725f 1839 ret = dst_cache_init(&t->dst_cache, GFP_KERNEL);
0d3c703a
TH		 1840 if (ret)
1841 goto free_stats;
1842
1843 ret = gro_cells_init(&t->gro_cells, dev);
1844 if (ret)
1845 goto destroy_dst;
cdf3464e 1846
79ecb90e 1847 t->tun_hlen = 0;
058214a4
TH		 1848 t->hlen = t->encap_hlen + t->tun_hlen;
1849 t_hlen = t->hlen + sizeof(struct ipv6hdr);
1850
1851 dev->type = ARPHRD_TUNNEL6;
1852 dev->hard_header_len = LL_MAX_HEADER + t_hlen;
1853 dev->mtu = ETH_DATA_LEN - t_hlen;
1854 if (!(t->parms.flags & IP6_TNL_F_IGN_ENCAP_LIMIT))
1855 dev->mtu -= 8;
b96f9afe 1856 dev->min_mtu = ETH_MIN_MTU;
f200de49 1857 dev->max_mtu = IP6_MAX_MTU - dev->hard_header_len;
79ecb90e 1858
8560f226 1859 return 0;
0d3c703a
TH		 1860
1861destroy_dst:
1862 dst_cache_destroy(&t->dst_cache);
1863free_stats:
1864 free_percpu(dev->tstats);
1865 dev->tstats = NULL;
1866
1867 return ret;
1da177e4
LT		 1868}
1869
1870/**
3144581c 1871 * ip6_tnl_dev_init - initializer for all non fallback tunnel devices
1da177e4
LT		 1872 * @dev: virtual device associated with tunnel
1873 **/
1874
8560f226 1875static int ip6_tnl_dev_init(struct net_device *dev)
1da177e4 1876{
2941a486 1877 struct ip6_tnl *t = netdev_priv(dev);
8560f226
ED		 1878 int err = ip6_tnl_dev_init_gen(dev);
1879
1880 if (err)
1881 return err;
3144581c 1882 ip6_tnl_link_config(t);
8d79266b
AS		 1883 if (t->parms.collect_md) {
1884 dev->features |= NETIF_F_NETNS_LOCAL;
1885 netif_keep_dst(dev);
1886 }
8560f226 1887 return 0;
1da177e4
LT		 1888}
1889
1890/**
3144581c 1891 * ip6_fb_tnl_dev_init - initializer for fallback tunnel device
1da177e4
LT		 1892 * @dev: fallback device
1893 *
1894 * Return: 0
1895 **/
1896
8560f226 1897static int __net_init ip6_fb_tnl_dev_init(struct net_device *dev)
1da177e4 1898{
2941a486 1899 struct ip6_tnl *t = netdev_priv(dev);
3e6c9fb5
PE		 1900 struct net *net = dev_net(dev);
1901 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
1902
502b0935 1903 t->parms.proto = IPPROTO_IPV6;
1da177e4 1904 dev_hold(dev);
d0087b29 1905
cf778b00 1906 rcu_assign_pointer(ip6n->tnls_wc[0], t);
8560f226 1907 return 0;
1da177e4
LT		 1908}
1909
a8b8a889
MS		 1910static int ip6_tnl_validate(struct nlattr *tb[], struct nlattr *data[],
1911 struct netlink_ext_ack *extack)
0b112457
ND		 1912{
1913 u8 proto;
1914
c8965932 1915 if (!data || !data[IFLA_IPTUN_PROTO])
0b112457
ND		 1916 return 0;
1917
1918 proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
1919 if (proto != IPPROTO_IPV6 &&
1920 proto != IPPROTO_IPIP &&
1921 proto != 0)
1922 return -EINVAL;
1923
1924 return 0;
1925}
1926
1927static void ip6_tnl_netlink_parms(struct nlattr *data[],
1928 struct __ip6_tnl_parm *parms)
1929{
1930 memset(parms, 0, sizeof(*parms));
1931
1932 if (!data)
1933 return;
1934
1935 if (data[IFLA_IPTUN_LINK])
1936 parms->link = nla_get_u32(data[IFLA_IPTUN_LINK]);
1937
1938 if (data[IFLA_IPTUN_LOCAL])
67b61f6c 1939 parms->laddr = nla_get_in6_addr(data[IFLA_IPTUN_LOCAL]);
0b112457
ND		 1940
1941 if (data[IFLA_IPTUN_REMOTE])
67b61f6c 1942 parms->raddr = nla_get_in6_addr(data[IFLA_IPTUN_REMOTE]);
0b112457
ND		 1943
1944 if (data[IFLA_IPTUN_TTL])
1945 parms->hop_limit = nla_get_u8(data[IFLA_IPTUN_TTL]);
1946
1947 if (data[IFLA_IPTUN_ENCAP_LIMIT])
1948 parms->encap_limit = nla_get_u8(data[IFLA_IPTUN_ENCAP_LIMIT]);
1949
1950 if (data[IFLA_IPTUN_FLOWINFO])
1ff05fb7 1951 parms->flowinfo = nla_get_be32(data[IFLA_IPTUN_FLOWINFO]);
0b112457
ND		 1952
1953 if (data[IFLA_IPTUN_FLAGS])
1954 parms->flags = nla_get_u32(data[IFLA_IPTUN_FLAGS]);
1955
1956 if (data[IFLA_IPTUN_PROTO])
1957 parms->proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
8d79266b
AS		 1958
1959 if (data[IFLA_IPTUN_COLLECT_METADATA])
1960 parms->collect_md = true;
0a473b82
CG		 1961
1962 if (data[IFLA_IPTUN_FWMARK])
1963 parms->fwmark = nla_get_u32(data[IFLA_IPTUN_FWMARK]);
0b112457
ND		 1964}
1965
b3a27b51
TH		 1966static bool ip6_tnl_netlink_encap_parms(struct nlattr *data[],
1967 struct ip_tunnel_encap *ipencap)
1968{
1969 bool ret = false;
1970
1971 memset(ipencap, 0, sizeof(*ipencap));
1972
1973 if (!data)
1974 return ret;
1975
1976 if (data[IFLA_IPTUN_ENCAP_TYPE]) {
1977 ret = true;
1978 ipencap->type = nla_get_u16(data[IFLA_IPTUN_ENCAP_TYPE]);
1979 }
1980
1981 if (data[IFLA_IPTUN_ENCAP_FLAGS]) {
1982 ret = true;
1983 ipencap->flags = nla_get_u16(data[IFLA_IPTUN_ENCAP_FLAGS]);
1984 }
1985
1986 if (data[IFLA_IPTUN_ENCAP_SPORT]) {
1987 ret = true;
1988 ipencap->sport = nla_get_be16(data[IFLA_IPTUN_ENCAP_SPORT]);
1989 }
1990
1991 if (data[IFLA_IPTUN_ENCAP_DPORT]) {
1992 ret = true;
1993 ipencap->dport = nla_get_be16(data[IFLA_IPTUN_ENCAP_DPORT]);
1994 }
1995
1996 return ret;
1997}
1998
0b112457 1999static int ip6_tnl_newlink(struct net *src_net, struct net_device *dev,
7a3f4a18
MS		 2000 struct nlattr *tb[], struct nlattr *data[],
2001 struct netlink_ext_ack *extack)
0b112457
ND		 2002{
2003 struct net *net = dev_net(dev);
8d79266b 2004 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
b3a27b51 2005 struct ip_tunnel_encap ipencap;
2b0fbc2f
XL		 2006 struct ip6_tnl *nt, *t;
2007 int err;
0b112457
ND		 2008
2009 nt = netdev_priv(dev);
b3a27b51
TH		 2010
2011 if (ip6_tnl_netlink_encap_parms(data, &ipencap)) {
2b0fbc2f 2012 err = ip6_tnl_encap_setup(nt, &ipencap);
b3a27b51
TH		 2013 if (err < 0)
2014 return err;
2015 }
2016
0b112457
ND		 2017 ip6_tnl_netlink_parms(data, &nt->parms);
2018
8d79266b
AS		 2019 if (nt->parms.collect_md) {
2020 if (rtnl_dereference(ip6n->collect_md_tun))
2021 return -EEXIST;
2022 } else {
2023 t = ip6_tnl_locate(net, &nt->parms, 0);
2024 if (!IS_ERR(t))
2025 return -EEXIST;
2026 }
0b112457 2027
2b0fbc2f
XL		 2028 err = ip6_tnl_create2(dev);
2029 if (!err && tb[IFLA_MTU])
2030 ip6_tnl_change_mtu(dev, nla_get_u32(tb[IFLA_MTU]));
2031
2032 return err;
0b112457
ND		 2033}
2034
2035static int ip6_tnl_changelink(struct net_device *dev, struct nlattr *tb[],
ad744b22
MS		 2036 struct nlattr *data[],
2037 struct netlink_ext_ack *extack)
0b112457 2038{
0bd87628 2039 struct ip6_tnl *t = netdev_priv(dev);
0b112457 2040 struct __ip6_tnl_parm p;
0bd87628 2041 struct net *net = t->net;
0b112457 2042 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
b3a27b51 2043 struct ip_tunnel_encap ipencap;
0b112457
ND		 2044
2045 if (dev == ip6n->fb_tnl_dev)
2046 return -EINVAL;
2047
b3a27b51
TH		 2048 if (ip6_tnl_netlink_encap_parms(data, &ipencap)) {
2049 int err = ip6_tnl_encap_setup(t, &ipencap);
2050
2051 if (err < 0)
2052 return err;
2053 }
0b112457 2054 ip6_tnl_netlink_parms(data, &p);
8d79266b
AS		 2055 if (p.collect_md)
2056 return -EINVAL;
0b112457
ND		 2057
2058 t = ip6_tnl_locate(net, &p, 0);
37355565 2059 if (!IS_ERR(t)) {
0b112457
ND		 2060 if (t->dev != dev)
2061 return -EEXIST;
2062 } else
2063 t = netdev_priv(dev);
2064
2065 return ip6_tnl_update(t, &p);
2066}
2067
1e9f3d6f
ND		 2068static void ip6_tnl_dellink(struct net_device *dev, struct list_head *head)
2069{
2070 struct net *net = dev_net(dev);
2071 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
2072
2073 if (dev != ip6n->fb_tnl_dev)
2074 unregister_netdevice_queue(dev, head);
2075}
2076
b58d731a 2077static size_t ip6_tnl_get_size(const struct net_device *dev)
c075b130
ND		 2078{
2079 return
2080 /* IFLA_IPTUN_LINK */
2081 nla_total_size(4) +
2082 /* IFLA_IPTUN_LOCAL */
2083 nla_total_size(sizeof(struct in6_addr)) +
2084 /* IFLA_IPTUN_REMOTE */
2085 nla_total_size(sizeof(struct in6_addr)) +
2086 /* IFLA_IPTUN_TTL */
2087 nla_total_size(1) +
2088 /* IFLA_IPTUN_ENCAP_LIMIT */
2089 nla_total_size(1) +
2090 /* IFLA_IPTUN_FLOWINFO */
2091 nla_total_size(4) +
2092 /* IFLA_IPTUN_FLAGS */
2093 nla_total_size(4) +
cfa323b6
ND		 2094 /* IFLA_IPTUN_PROTO */
2095 nla_total_size(1) +
b3a27b51
TH		 2096 /* IFLA_IPTUN_ENCAP_TYPE */
2097 nla_total_size(2) +
2098 /* IFLA_IPTUN_ENCAP_FLAGS */
2099 nla_total_size(2) +
2100 /* IFLA_IPTUN_ENCAP_SPORT */
2101 nla_total_size(2) +
2102 /* IFLA_IPTUN_ENCAP_DPORT */
2103 nla_total_size(2) +
8d79266b
AS		 2104 /* IFLA_IPTUN_COLLECT_METADATA */
2105 nla_total_size(0) +
0a473b82
CG		 2106 /* IFLA_IPTUN_FWMARK */
2107 nla_total_size(4) +
c075b130
ND		 2108 0;
2109}
2110
b58d731a 2111static int ip6_tnl_fill_info(struct sk_buff *skb, const struct net_device *dev)
c075b130
ND		 2112{
2113 struct ip6_tnl *tunnel = netdev_priv(dev);
2114 struct __ip6_tnl_parm *parm = &tunnel->parms;
2115
2116 if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) ||
930345ea
JB		 2117 nla_put_in6_addr(skb, IFLA_IPTUN_LOCAL, &parm->laddr) ||
2118 nla_put_in6_addr(skb, IFLA_IPTUN_REMOTE, &parm->raddr) ||
c075b130
ND		 2119 nla_put_u8(skb, IFLA_IPTUN_TTL, parm->hop_limit) ||
2120 nla_put_u8(skb, IFLA_IPTUN_ENCAP_LIMIT, parm->encap_limit) ||
2121 nla_put_be32(skb, IFLA_IPTUN_FLOWINFO, parm->flowinfo) ||
cfa323b6 2122 nla_put_u32(skb, IFLA_IPTUN_FLAGS, parm->flags) ||
0a473b82
CG		 2123 nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->proto) ||
2124 nla_put_u32(skb, IFLA_IPTUN_FWMARK, parm->fwmark))
c075b130 2125 goto nla_put_failure;
b3a27b51 2126
8d79266b
AS		 2127 if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE, tunnel->encap.type) ||
2128 nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT, tunnel->encap.sport) ||
2129 nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT, tunnel->encap.dport) ||
2130 nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS, tunnel->encap.flags))
b3a27b51
TH		 2131 goto nla_put_failure;
2132
8d79266b
AS		 2133 if (parm->collect_md)
2134 if (nla_put_flag(skb, IFLA_IPTUN_COLLECT_METADATA))
2135 goto nla_put_failure;
0a473b82 2136
c075b130
ND		 2137 return 0;
2138
2139nla_put_failure:
2140 return -EMSGSIZE;
2141}
2142
1728d4fa
ND		 2143struct net *ip6_tnl_get_link_net(const struct net_device *dev)
2144{
2145 struct ip6_tnl *tunnel = netdev_priv(dev);
2146
2147 return tunnel->net;
2148}
2149EXPORT_SYMBOL(ip6_tnl_get_link_net);
2150
0b112457
ND		 2151static const struct nla_policy ip6_tnl_policy[IFLA_IPTUN_MAX + 1] = {
2152 [IFLA_IPTUN_LINK] = { .type = NLA_U32 },
2153 [IFLA_IPTUN_LOCAL] = { .len = sizeof(struct in6_addr) },
2154 [IFLA_IPTUN_REMOTE] = { .len = sizeof(struct in6_addr) },
2155 [IFLA_IPTUN_TTL] = { .type = NLA_U8 },
2156 [IFLA_IPTUN_ENCAP_LIMIT] = { .type = NLA_U8 },
2157 [IFLA_IPTUN_FLOWINFO] = { .type = NLA_U32 },
2158 [IFLA_IPTUN_FLAGS] = { .type = NLA_U32 },
2159 [IFLA_IPTUN_PROTO] = { .type = NLA_U8 },
b3a27b51
TH		 2160 [IFLA_IPTUN_ENCAP_TYPE] = { .type = NLA_U16 },
2161 [IFLA_IPTUN_ENCAP_FLAGS] = { .type = NLA_U16 },
2162 [IFLA_IPTUN_ENCAP_SPORT] = { .type = NLA_U16 },
2163 [IFLA_IPTUN_ENCAP_DPORT] = { .type = NLA_U16 },
8d79266b 2164 [IFLA_IPTUN_COLLECT_METADATA] = { .type = NLA_FLAG },
0a473b82 2165 [IFLA_IPTUN_FWMARK] = { .type = NLA_U32 },
0b112457
ND		 2166};
2167
c075b130
ND		 2168static struct rtnl_link_ops ip6_link_ops __read_mostly = {
2169 .kind = "ip6tnl",
2170 .maxtype = IFLA_IPTUN_MAX,
0b112457 2171 .policy = ip6_tnl_policy,
c075b130 2172 .priv_size = sizeof(struct ip6_tnl),
0b112457
ND		 2173 .setup = ip6_tnl_dev_setup,
2174 .validate = ip6_tnl_validate,
2175 .newlink = ip6_tnl_newlink,
2176 .changelink = ip6_tnl_changelink,
1e9f3d6f 2177 .dellink = ip6_tnl_dellink,
b58d731a
ND		 2178 .get_size = ip6_tnl_get_size,
2179 .fill_info = ip6_tnl_fill_info,
1728d4fa 2180 .get_link_net = ip6_tnl_get_link_net,
c075b130
ND		 2181};
2182
3ff2cfa5 2183static struct xfrm6_tunnel ip4ip6_handler __read_mostly = {
c4d3efaf
YK		 2184 .handler = ip4ip6_rcv,
2185 .err_handler = ip4ip6_err,
2186 .priority = 1,
2187};
2188
3ff2cfa5 2189static struct xfrm6_tunnel ip6ip6_handler __read_mostly = {
0303770d
PM		 2190 .handler = ip6ip6_rcv,
2191 .err_handler = ip6ip6_err,
d2acc347 2192 .priority = 1,
1da177e4
LT		 2193};
2194
1e9f3d6f 2195static void __net_exit ip6_tnl_destroy_tunnels(struct net *net)
3e6c9fb5 2196{
1e9f3d6f 2197 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
0bd87628 2198 struct net_device *dev, *aux;
3e6c9fb5
PE		 2199 int h;
2200 struct ip6_tnl *t;
cf4432f5 2201 LIST_HEAD(list);
3e6c9fb5 2202
0bd87628
ND		 2203 for_each_netdev_safe(net, dev, aux)
2204 if (dev->rtnl_link_ops == &ip6_link_ops)
2205 unregister_netdevice_queue(dev, &list);
2206
e87a8f24 2207 for (h = 0; h < IP6_TUNNEL_HASH_SIZE; h++) {
94767632 2208 t = rtnl_dereference(ip6n->tnls_r_l[h]);
53b24b8f 2209 while (t) {
0bd87628
ND		 2210 /* If dev is in the same netns, it has already
2211 * been added to the list by the previous loop.
2212 */
2213 if (!net_eq(dev_net(t->dev), net))
2214 unregister_netdevice_queue(t->dev, &list);
94767632 2215 t = rtnl_dereference(t->next);
cf4432f5 2216 }
3e6c9fb5
PE		 2217 }
2218
cf4432f5 2219 unregister_netdevice_many(&list);
3e6c9fb5
PE		 2220}
2221
2c8c1e72 2222static int __net_init ip6_tnl_init_net(struct net *net)
13eeb8e9 2223{
ac31cd3c 2224 struct ip6_tnl_net *ip6n = net_generic(net, ip6_tnl_net_id);
731abb9c 2225 struct ip6_tnl *t = NULL;
13eeb8e9 2226 int err;
13eeb8e9 2227
3e6c9fb5
PE		 2228 ip6n->tnls[0] = ip6n->tnls_wc;
2229 ip6n->tnls[1] = ip6n->tnls_r_l;
2230
15820e12
PE		 2231 err = -ENOMEM;
2232 ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
c835a677 2233 NET_NAME_UNKNOWN, ip6_tnl_dev_setup);
15820e12
PE		 2234
2235 if (!ip6n->fb_tnl_dev)
2236 goto err_alloc_dev;
be77e593 2237 dev_net_set(ip6n->fb_tnl_dev, net);
bb814094 2238 ip6n->fb_tnl_dev->rtnl_link_ops = &ip6_link_ops;
0bd87628
ND		 2239 /* FB netdevice is special: we have one, and only one per netns.
2240 * Allowing to move it to another netns is clearly unsafe.
2241 */
2242 ip6n->fb_tnl_dev->features |= NETIF_F_NETNS_LOCAL;
15820e12 2243
8560f226
ED		 2244 err = ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
2245 if (err < 0)
2246 goto err_register;
15820e12
PE		 2247
2248 err = register_netdev(ip6n->fb_tnl_dev);
2249 if (err < 0)
2250 goto err_register;
731abb9c
JB		 2251
2252 t = netdev_priv(ip6n->fb_tnl_dev);
2253
2254 strcpy(t->parms.name, ip6n->fb_tnl_dev->name);
13eeb8e9
PE		 2255 return 0;
2256
15820e12 2257err_register:
cf124db5 2258 free_netdev(ip6n->fb_tnl_dev);
15820e12 2259err_alloc_dev:
13eeb8e9
PE		 2260 return err;
2261}
2262
2c8c1e72 2263static void __net_exit ip6_tnl_exit_net(struct net *net)
13eeb8e9 2264{
3e6c9fb5 2265 rtnl_lock();
1e9f3d6f 2266 ip6_tnl_destroy_tunnels(net);
3e6c9fb5 2267 rtnl_unlock();
13eeb8e9
PE		 2268}
2269
2270static struct pernet_operations ip6_tnl_net_ops = {
2271 .init = ip6_tnl_init_net,
2272 .exit = ip6_tnl_exit_net,
ac31cd3c
EB		 2273 .id = &ip6_tnl_net_id,
2274 .size = sizeof(struct ip6_tnl_net),
13eeb8e9
PE		 2275};
2276
1da177e4
LT		 2277/**
2278 * ip6_tunnel_init - register protocol and reserve needed resources
2279 *
2280 * Return: 0 on success
2281 **/
2282
2283static int __init ip6_tunnel_init(void)
2284{
2285 int err;
2286
8c22dab0
XL		 2287 if (!ipv6_mod_enabled())
2288 return -EOPNOTSUPP;
2289
d5aa407f
AD		 2290 err = register_pernet_device(&ip6_tnl_net_ops);
2291 if (err < 0)
2292 goto out_pernet;
2293
2294 err = xfrm6_tunnel_register(&ip4ip6_handler, AF_INET);
2295 if (err < 0) {
f3213831 2296 pr_err("%s: can't register ip4ip6\n", __func__);
d5aa407f 2297 goto out_ip4ip6;
c4d3efaf
YK		 2298 }
2299
d5aa407f
AD		 2300 err = xfrm6_tunnel_register(&ip6ip6_handler, AF_INET6);
2301 if (err < 0) {
f3213831 2302 pr_err("%s: can't register ip6ip6\n", __func__);
d5aa407f 2303 goto out_ip6ip6;
1da177e4 2304 }
c075b130
ND		 2305 err = rtnl_link_register(&ip6_link_ops);
2306 if (err < 0)
2307 goto rtnl_link_failed;
13eeb8e9 2308
1da177e4 2309 return 0;
d5aa407f 2310
c075b130
ND		 2311rtnl_link_failed:
2312 xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6);
d5aa407f 2313out_ip6ip6:
c4d3efaf 2314 xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET);
d5aa407f
AD		 2315out_ip4ip6:
2316 unregister_pernet_device(&ip6_tnl_net_ops);
2317out_pernet:
1da177e4
LT		 2318 return err;
2319}
2320
2321/**
2322 * ip6_tunnel_cleanup - free resources and unregister protocol
2323 **/
2324
2325static void __exit ip6_tunnel_cleanup(void)
2326{
c075b130 2327 rtnl_link_unregister(&ip6_link_ops);
c4d3efaf 2328 if (xfrm6_tunnel_deregister(&ip4ip6_handler, AF_INET))
f3213831 2329 pr_info("%s: can't deregister ip4ip6\n", __func__);
c4d3efaf 2330
73d605d1 2331 if (xfrm6_tunnel_deregister(&ip6ip6_handler, AF_INET6))
f3213831 2332 pr_info("%s: can't deregister ip6ip6\n", __func__);
1da177e4 2333
ac31cd3c 2334 unregister_pernet_device(&ip6_tnl_net_ops);
1da177e4
LT		 2335}
2336
2337module_init(ip6_tunnel_init);
2338module_exit(ip6_tunnel_cleanup);