projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net
[deliverable/linux.git] / net / ipv6 / route.c
CommitLineData
1da177e4
LT		 1/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4 7 *
1da177e4
LT		 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13
14/* Changes:
15 *
16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection.
18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list.
c0bece9f
YH		 23 * Ville Nuorvala
24 * Fixed routing subtrees.
1da177e4
LT		 25 */
26
f3213831
JP		 27#define pr_fmt(fmt) "IPv6: " fmt
28
4fc268d2 29#include <linux/capability.h>
1da177e4 30#include <linux/errno.h>
bc3b2d7f 31#include <linux/export.h>
1da177e4
LT		 32#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
7bc570c8 40#include <linux/mroute6.h>
1da177e4 41#include <linux/init.h>
1da177e4 42#include <linux/if_arp.h>
1da177e4
LT		 43#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
5b7c931d 45#include <linux/nsproxy.h>
5a0e3ad6 46#include <linux/slab.h>
457c4cbc 47#include <net/net_namespace.h>
1da177e4
LT		 48#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
57#include <net/xfrm.h>
8d71740c 58#include <net/netevent.h>
21713ebc 59#include <net/netlink.h>
51ebd318 60#include <net/nexthop.h>
1da177e4
LT		 61
62#include <asm/uaccess.h>
63
64#ifdef CONFIG_SYSCTL
65#include <linux/sysctl.h>
66#endif
67
afc154e9
HFS		 68enum rt6_nud_state {
69 RT6_NUD_FAIL_HARD = -2,
70 RT6_NUD_FAIL_SOFT = -1,
71 RT6_NUD_SUCCEED = 1
72};
73
1716a961 74static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 75 const struct in6_addr *dest);
1da177e4 76static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 77static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 78static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 79static struct dst_entry *ip6_negative_advice(struct dst_entry *);
80static void ip6_dst_destroy(struct dst_entry *);
81static void ip6_dst_ifdown(struct dst_entry *,
82 struct net_device *dev, int how);
569d3645 83static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 84
85static int ip6_pkt_discard(struct sk_buff *skb);
86static int ip6_pkt_discard_out(struct sk_buff *skb);
7150aede
K		 87static int ip6_pkt_prohibit(struct sk_buff *skb);
88static int ip6_pkt_prohibit_out(struct sk_buff *skb);
1da177e4 89static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 90static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
91 struct sk_buff *skb, u32 mtu);
92static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
93 struct sk_buff *skb);
52bd4c0c 94static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 95
70ceb4f5 96#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 97static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 98 const struct in6_addr *prefix, int prefixlen,
99 const struct in6_addr *gwaddr, int ifindex,
95c96174 100 unsigned int pref);
efa2cea0 101static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 102 const struct in6_addr *prefix, int prefixlen,
103 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 104#endif
105
06582540
DM		 106static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
107{
108 struct rt6_info *rt = (struct rt6_info *) dst;
109 struct inet_peer *peer;
110 u32 *p = NULL;
111
8e2ec639
YZ		 112 if (!(rt->dst.flags & DST_HOST))
113 return NULL;
114
fbfe95a4 115 peer = rt6_get_peer_create(rt);
06582540
DM		 116 if (peer) {
117 u32 *old_p = __DST_METRICS_PTR(old);
118 unsigned long prev, new;
119
120 p = peer->metrics;
121 if (inet_metrics_new(peer))
122 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
123
124 new = (unsigned long) p;
125 prev = cmpxchg(&dst->_metrics, old, new);
126
127 if (prev != old) {
128 p = __DST_METRICS_PTR(prev);
129 if (prev & DST_METRICS_READ_ONLY)
130 p = NULL;
131 }
132 }
133 return p;
134}
135
f894cbf8
DM		 136static inline const void *choose_neigh_daddr(struct rt6_info *rt,
137 struct sk_buff *skb,
138 const void *daddr)
39232973
DM		 139{
140 struct in6_addr *p = &rt->rt6i_gateway;
141
a7563f34 142 if (!ipv6_addr_any(p))
39232973 143 return (const void *) p;
f894cbf8
DM		 144 else if (skb)
145 return &ipv6_hdr(skb)->daddr;
39232973
DM		 146 return daddr;
147}
148
f894cbf8
DM		 149static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
150 struct sk_buff *skb,
151 const void *daddr)
d3aaeb38 152{
39232973
DM		 153 struct rt6_info *rt = (struct rt6_info *) dst;
154 struct neighbour *n;
155
f894cbf8 156 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 157 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 158 if (n)
159 return n;
160 return neigh_create(&nd_tbl, daddr, dst->dev);
161}
162
9a7ec3a9 163static struct dst_ops ip6_dst_ops_template = {
1da177e4 164 .family = AF_INET6,
09640e63 165 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT		 166 .gc = ip6_dst_gc,
167 .gc_thresh = 1024,
168 .check = ip6_dst_check,
0dbaee3b 169 .default_advmss = ip6_default_advmss,
ebb762f2 170 .mtu = ip6_mtu,
06582540 171 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 172 .destroy = ip6_dst_destroy,
173 .ifdown = ip6_dst_ifdown,
174 .negative_advice = ip6_negative_advice,
175 .link_failure = ip6_link_failure,
176 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 177 .redirect = rt6_do_redirect,
1ac06e03 178 .local_out = __ip6_local_out,
d3aaeb38 179 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 180};
181
ebb762f2 182static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 183{
618f9bc7
SK		 184 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
185
186 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 187}
188
6700c270
DM		 189static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
190 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 191{
192}
193
6700c270
DM		 194static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
195 struct sk_buff *skb)
b587ee3b
DM		 196{
197}
198
0972ddb2
HB		 199static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
200 unsigned long old)
201{
202 return NULL;
203}
204
14e50e57
DM		 205static struct dst_ops ip6_dst_blackhole_ops = {
206 .family = AF_INET6,
09640e63 207 .protocol = cpu_to_be16(ETH_P_IPV6),
14e50e57
DM		 208 .destroy = ip6_dst_destroy,
209 .check = ip6_dst_check,
ebb762f2 210 .mtu = ip6_blackhole_mtu,
214f45c9 211 .default_advmss = ip6_default_advmss,
14e50e57 212 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 213 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 214 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 215 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 216};
217
62fa8a84 218static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 219 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 220};
221
fb0af4c7 222static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 223 .dst = {
224 .__refcnt = ATOMIC_INIT(1),
225 .__use = 1,
2c20cbd7 226 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 227 .error = -ENETUNREACH,
d8d1f30b
CG		 228 .input = ip6_pkt_discard,
229 .output = ip6_pkt_discard_out,
1da177e4
LT		 230 },
231 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 232 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 233 .rt6i_metric = ~(u32) 0,
234 .rt6i_ref = ATOMIC_INIT(1),
235};
236
101367c2
TG		 237#ifdef CONFIG_IPV6_MULTIPLE_TABLES
238
fb0af4c7 239static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 240 .dst = {
241 .__refcnt = ATOMIC_INIT(1),
242 .__use = 1,
2c20cbd7 243 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 244 .error = -EACCES,
d8d1f30b
CG		 245 .input = ip6_pkt_prohibit,
246 .output = ip6_pkt_prohibit_out,
101367c2
TG		 247 },
248 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 249 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 250 .rt6i_metric = ~(u32) 0,
251 .rt6i_ref = ATOMIC_INIT(1),
252};
253
fb0af4c7 254static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 255 .dst = {
256 .__refcnt = ATOMIC_INIT(1),
257 .__use = 1,
2c20cbd7 258 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 259 .error = -EINVAL,
d8d1f30b
CG		 260 .input = dst_discard,
261 .output = dst_discard,
101367c2
TG		 262 },
263 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 264 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 265 .rt6i_metric = ~(u32) 0,
266 .rt6i_ref = ATOMIC_INIT(1),
267};
268
269#endif
270
1da177e4 271/* allocate dst with ip6_dst_ops */
97bab73f 272static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 273 struct net_device *dev,
8b96d22d
DM		 274 int flags,
275 struct fib6_table *table)
1da177e4 276{
97bab73f 277 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 278 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 279
97bab73f 280 if (rt) {
8104891b
SK		 281 struct dst_entry *dst = &rt->dst;
282
283 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
8b96d22d 284 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
ca4c3fc2 285 rt->rt6i_genid = rt_genid_ipv6(net);
51ebd318 286 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 287 }
cf911662 288 return rt;
1da177e4
LT		 289}
290
291static void ip6_dst_destroy(struct dst_entry *dst)
292{
293 struct rt6_info *rt = (struct rt6_info *)dst;
294 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 295 struct dst_entry *from = dst->from;
1da177e4 296
8e2ec639
YZ		 297 if (!(rt->dst.flags & DST_HOST))
298 dst_destroy_metrics_generic(dst);
299
38308473 300 if (idev) {
1da177e4
LT		 301 rt->rt6i_idev = NULL;
302 in6_dev_put(idev);
1ab1457c 303 }
1716a961 304
ecd98837
YH		 305 dst->from = NULL;
306 dst_release(from);
1716a961 307
97bab73f
DM		 308 if (rt6_has_peer(rt)) {
309 struct inet_peer *peer = rt6_peer_ptr(rt);
b3419363
DM		 310 inet_putpeer(peer);
311 }
312}
313
314void rt6_bind_peer(struct rt6_info *rt, int create)
315{
97bab73f 316 struct inet_peer_base *base;
b3419363
DM		 317 struct inet_peer *peer;
318
97bab73f
DM		 319 base = inetpeer_base_ptr(rt->_rt6i_peer);
320 if (!base)
321 return;
322
323 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create);
7b34ca2a
DM		 324 if (peer) {
325 if (!rt6_set_peer(rt, peer))
326 inet_putpeer(peer);
7b34ca2a 327 }
1da177e4
LT		 328}
329
330static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
331 int how)
332{
333 struct rt6_info *rt = (struct rt6_info *)dst;
334 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 335 struct net_device *loopback_dev =
c346dca1 336 dev_net(dev)->loopback_dev;
1da177e4 337
97cac082
DM		 338 if (dev != loopback_dev) {
339 if (idev && idev->dev == dev) {
340 struct inet6_dev *loopback_idev =
341 in6_dev_get(loopback_dev);
342 if (loopback_idev) {
343 rt->rt6i_idev = loopback_idev;
344 in6_dev_put(idev);
345 }
346 }
1da177e4
LT		 347 }
348}
349
a50feda5 350static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 351{
1716a961
G		 352 if (rt->rt6i_flags & RTF_EXPIRES) {
353 if (time_after(jiffies, rt->dst.expires))
a50feda5 354 return true;
1716a961 355 } else if (rt->dst.from) {
3fd91fb3 356 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 357 }
a50feda5 358 return false;
1da177e4
LT		 359}
360
a50feda5 361static bool rt6_need_strict(const struct in6_addr *daddr)
c71099ac 362{
a02cec21
ED		 363 return ipv6_addr_type(daddr) &
364 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
c71099ac
TG		 365}
366
51ebd318
ND		 367/* Multipath route selection:
368 * Hash based function using packet header and flowlabel.
369 * Adapted from fib_info_hashfn()
370 */
371static int rt6_info_hash_nhsfn(unsigned int candidate_count,
372 const struct flowi6 *fl6)
373{
374 unsigned int val = fl6->flowi6_proto;
375
c08977bb
YH		 376 val ^= ipv6_addr_hash(&fl6->daddr);
377 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 378
379 /* Work only if this not encapsulated */
380 switch (fl6->flowi6_proto) {
381 case IPPROTO_UDP:
382 case IPPROTO_TCP:
383 case IPPROTO_SCTP:
b3ce5ae1
ND		 384 val ^= (__force u16)fl6->fl6_sport;
385 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 386 break;
387
388 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 389 val ^= (__force u16)fl6->fl6_icmp_type;
390 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 391 break;
392 }
393 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 394 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 395
396 /* Perhaps, we need to tune, this function? */
397 val = val ^ (val >> 7) ^ (val >> 12);
398 return val % candidate_count;
399}
400
401static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 402 struct flowi6 *fl6, int oif,
403 int strict)
51ebd318
ND		 404{
405 struct rt6_info *sibling, *next_sibling;
406 int route_choosen;
407
408 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
409 /* Don't change the route, if route_choosen == 0
410 * (siblings does not include ourself)
411 */
412 if (route_choosen)
413 list_for_each_entry_safe(sibling, next_sibling,
414 &match->rt6i_siblings, rt6i_siblings) {
415 route_choosen--;
416 if (route_choosen == 0) {
52bd4c0c
ND		 417 if (rt6_score_route(sibling, oif, strict) < 0)
418 break;
51ebd318
ND		 419 match = sibling;
420 break;
421 }
422 }
423 return match;
424}
425
1da177e4 426/*
c71099ac 427 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 428 */
429
8ed67789
DL		 430static inline struct rt6_info *rt6_device_match(struct net *net,
431 struct rt6_info *rt,
b71d1d42 432 const struct in6_addr *saddr,
1da177e4 433 int oif,
d420895e 434 int flags)
1da177e4
LT		 435{
436 struct rt6_info *local = NULL;
437 struct rt6_info *sprt;
438
dd3abc4e
YH		 439 if (!oif && ipv6_addr_any(saddr))
440 goto out;
441
d8d1f30b 442 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 443 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 444
445 if (oif) {
1da177e4
LT		 446 if (dev->ifindex == oif)
447 return sprt;
448 if (dev->flags & IFF_LOOPBACK) {
38308473 449 if (!sprt->rt6i_idev ||
1da177e4 450 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 451 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 452 continue;
1ab1457c 453 if (local && (!oif ||
1da177e4
LT		 454 local->rt6i_idev->dev->ifindex == oif))
455 continue;
456 }
457 local = sprt;
458 }
dd3abc4e
YH		 459 } else {
460 if (ipv6_chk_addr(net, saddr, dev,
461 flags & RT6_LOOKUP_F_IFACE))
462 return sprt;
1da177e4 463 }
dd3abc4e 464 }
1da177e4 465
dd3abc4e 466 if (oif) {
1da177e4
LT		 467 if (local)
468 return local;
469
d420895e 470 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 471 return net->ipv6.ip6_null_entry;
1da177e4 472 }
dd3abc4e 473out:
1da177e4
LT		 474 return rt;
475}
476
27097255 477#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 478struct __rt6_probe_work {
479 struct work_struct work;
480 struct in6_addr target;
481 struct net_device *dev;
482};
483
484static void rt6_probe_deferred(struct work_struct *w)
485{
486 struct in6_addr mcaddr;
487 struct __rt6_probe_work *work =
488 container_of(w, struct __rt6_probe_work, work);
489
490 addrconf_addr_solict_mult(&work->target, &mcaddr);
491 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
492 dev_put(work->dev);
493 kfree(w);
494}
495
27097255
YH		 496static void rt6_probe(struct rt6_info *rt)
497{
f2c31e32 498 struct neighbour *neigh;
27097255
YH		 499 /*
500 * Okay, this does not seem to be appropriate
501 * for now, however, we need to check if it
502 * is really so; aka Router Reachability Probing.
503 *
504 * Router Reachability Probe MUST be rate-limited
505 * to no more than one per minute.
506 */
2152caea 507 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 508 return;
2152caea
YH		 509 rcu_read_lock_bh();
510 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
511 if (neigh) {
512 write_lock(&neigh->lock);
513 if (neigh->nud_state & NUD_VALID)
514 goto out;
7ff74a59 515 }
2152caea
YH		 516
517 if (!neigh ||
52e16356 518 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
c2f17e82 519 struct __rt6_probe_work *work;
27097255 520
c2f17e82
HFS		 521 work = kmalloc(sizeof(*work), GFP_ATOMIC);
522
523 if (neigh && work)
b820bb6b 524 neigh->updated = jiffies;
c2f17e82
HFS		 525
526 if (neigh)
2152caea
YH		 527 write_unlock(&neigh->lock);
528
c2f17e82
HFS		 529 if (work) {
530 INIT_WORK(&work->work, rt6_probe_deferred);
531 work->target = rt->rt6i_gateway;
532 dev_hold(rt->dst.dev);
533 work->dev = rt->dst.dev;
534 schedule_work(&work->work);
535 }
f2c31e32 536 } else {
2152caea
YH		 537out:
538 write_unlock(&neigh->lock);
f2c31e32 539 }
2152caea 540 rcu_read_unlock_bh();
27097255
YH		 541}
542#else
543static inline void rt6_probe(struct rt6_info *rt)
544{
27097255
YH		 545}
546#endif
547
1da177e4 548/*
554cfb7e 549 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 550 */
b6f99a21 551static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 552{
d1918542 553 struct net_device *dev = rt->dst.dev;
161980f4 554 if (!oif || dev->ifindex == oif)
554cfb7e 555 return 2;
161980f4
DM		 556 if ((dev->flags & IFF_LOOPBACK) &&
557 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
558 return 1;
559 return 0;
554cfb7e 560}
1da177e4 561
afc154e9 562static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 563{
f2c31e32 564 struct neighbour *neigh;
afc154e9 565 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 566
4d0c5911
YH		 567 if (rt->rt6i_flags & RTF_NONEXTHOP ||
568 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 569 return RT6_NUD_SUCCEED;
145a3621
YH		 570
571 rcu_read_lock_bh();
572 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
573 if (neigh) {
574 read_lock(&neigh->lock);
554cfb7e 575 if (neigh->nud_state & NUD_VALID)
afc154e9 576 ret = RT6_NUD_SUCCEED;
398bcbeb 577#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 578 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 579 ret = RT6_NUD_SUCCEED;
398bcbeb 580#endif
145a3621 581 read_unlock(&neigh->lock);
afc154e9
HFS		 582 } else {
583 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
584 RT6_NUD_SUCCEED : RT6_NUD_FAIL_SOFT;
a5a81f0b 585 }
145a3621
YH		 586 rcu_read_unlock_bh();
587
a5a81f0b 588 return ret;
1da177e4
LT		 589}
590
554cfb7e
YH		 591static int rt6_score_route(struct rt6_info *rt, int oif,
592 int strict)
1da177e4 593{
a5a81f0b 594 int m;
1ab1457c 595
4d0c5911 596 m = rt6_check_dev(rt, oif);
77d16f45 597 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 598 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 599#ifdef CONFIG_IPV6_ROUTER_PREF
600 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
601#endif
afc154e9
HFS		 602 if (strict & RT6_LOOKUP_F_REACHABLE) {
603 int n = rt6_check_neigh(rt);
604 if (n < 0)
605 return n;
606 }
554cfb7e
YH		 607 return m;
608}
609
f11e6659 610static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 611 int *mpri, struct rt6_info *match,
612 bool *do_rr)
554cfb7e 613{
f11e6659 614 int m;
afc154e9 615 bool match_do_rr = false;
f11e6659
DM		 616
617 if (rt6_check_expired(rt))
618 goto out;
619
620 m = rt6_score_route(rt, oif, strict);
ba486502 621 if (m == RT6_NUD_FAIL_SOFT) {
afc154e9
HFS		 622 match_do_rr = true;
623 m = 0; /* lowest valid score */
624 } else if (m < 0) {
f11e6659 625 goto out;
afc154e9
HFS		 626 }
627
628 if (strict & RT6_LOOKUP_F_REACHABLE)
629 rt6_probe(rt);
f11e6659
DM		 630
631 if (m > *mpri) {
afc154e9 632 *do_rr = match_do_rr;
f11e6659
DM		 633 *mpri = m;
634 match = rt;
f11e6659 635 }
f11e6659
DM		 636out:
637 return match;
638}
639
640static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
641 struct rt6_info *rr_head,
afc154e9
HFS		 642 u32 metric, int oif, int strict,
643 bool *do_rr)
f11e6659
DM		 644{
645 struct rt6_info *rt, *match;
554cfb7e 646 int mpri = -1;
1da177e4 647
f11e6659
DM		 648 match = NULL;
649 for (rt = rr_head; rt && rt->rt6i_metric == metric;
d8d1f30b 650 rt = rt->dst.rt6_next)
afc154e9 651 match = find_match(rt, oif, strict, &mpri, match, do_rr);
f11e6659 652 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
d8d1f30b 653 rt = rt->dst.rt6_next)
afc154e9 654 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 655
f11e6659
DM		 656 return match;
657}
1da177e4 658
f11e6659
DM		 659static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
660{
661 struct rt6_info *match, *rt0;
8ed67789 662 struct net *net;
afc154e9 663 bool do_rr = false;
1da177e4 664
f11e6659
DM		 665 rt0 = fn->rr_ptr;
666 if (!rt0)
667 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 668
afc154e9
HFS		 669 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
670 &do_rr);
1da177e4 671
afc154e9 672 if (do_rr) {
d8d1f30b 673 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 674
554cfb7e 675 /* no entries matched; do round-robin */
f11e6659
DM		 676 if (!next || next->rt6i_metric != rt0->rt6i_metric)
677 next = fn->leaf;
678
679 if (next != rt0)
680 fn->rr_ptr = next;
1da177e4 681 }
1da177e4 682
d1918542 683 net = dev_net(rt0->dst.dev);
a02cec21 684 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 685}
686
70ceb4f5
YH		 687#ifdef CONFIG_IPV6_ROUTE_INFO
688int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 689 const struct in6_addr *gwaddr)
70ceb4f5 690{
c346dca1 691 struct net *net = dev_net(dev);
70ceb4f5
YH		 692 struct route_info *rinfo = (struct route_info *) opt;
693 struct in6_addr prefix_buf, *prefix;
694 unsigned int pref;
4bed72e4 695 unsigned long lifetime;
70ceb4f5
YH		 696 struct rt6_info *rt;
697
698 if (len < sizeof(struct route_info)) {
699 return -EINVAL;
700 }
701
702 /* Sanity check for prefix_len and length */
703 if (rinfo->length > 3) {
704 return -EINVAL;
705 } else if (rinfo->prefix_len > 128) {
706 return -EINVAL;
707 } else if (rinfo->prefix_len > 64) {
708 if (rinfo->length < 2) {
709 return -EINVAL;
710 }
711 } else if (rinfo->prefix_len > 0) {
712 if (rinfo->length < 1) {
713 return -EINVAL;
714 }
715 }
716
717 pref = rinfo->route_pref;
718 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 719 return -EINVAL;
70ceb4f5 720
4bed72e4 721 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 722
723 if (rinfo->length == 3)
724 prefix = (struct in6_addr *)rinfo->prefix;
725 else {
726 /* this function is safe */
727 ipv6_addr_prefix(&prefix_buf,
728 (struct in6_addr *)rinfo->prefix,
729 rinfo->prefix_len);
730 prefix = &prefix_buf;
731 }
732
f104a567
DJ		 733 if (rinfo->prefix_len == 0)
734 rt = rt6_get_dflt_router(gwaddr, dev);
735 else
736 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
737 gwaddr, dev->ifindex);
70ceb4f5
YH		 738
739 if (rt && !lifetime) {
e0a1ad73 740 ip6_del_rt(rt);
70ceb4f5
YH		 741 rt = NULL;
742 }
743
744 if (!rt && lifetime)
efa2cea0 745 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 746 pref);
747 else if (rt)
748 rt->rt6i_flags = RTF_ROUTEINFO |
749 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
750
751 if (rt) {
1716a961
G		 752 if (!addrconf_finite_timeout(lifetime))
753 rt6_clean_expires(rt);
754 else
755 rt6_set_expires(rt, jiffies + HZ * lifetime);
756
94e187c0 757 ip6_rt_put(rt);
70ceb4f5
YH		 758 }
759 return 0;
760}
761#endif
762
8ed67789 763#define BACKTRACK(__net, saddr) \
982f56f3 764do { \
8ed67789 765 if (rt == __net->ipv6.ip6_null_entry) { \
982f56f3 766 struct fib6_node *pn; \
e0eda7bb 767 while (1) { \
982f56f3
YH		 768 if (fn->fn_flags & RTN_TL_ROOT) \
769 goto out; \
770 pn = fn->parent; \
771 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
8bce65b9 772 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
982f56f3
YH		 773 else \
774 fn = pn; \
775 if (fn->fn_flags & RTN_RTINFO) \
776 goto restart; \
c71099ac 777 } \
c71099ac 778 } \
38308473 779} while (0)
c71099ac 780
8ed67789
DL		 781static struct rt6_info *ip6_pol_route_lookup(struct net *net,
782 struct fib6_table *table,
4c9483b2 783 struct flowi6 *fl6, int flags)
1da177e4
LT		 784{
785 struct fib6_node *fn;
786 struct rt6_info *rt;
787
c71099ac 788 read_lock_bh(&table->tb6_lock);
4c9483b2 789 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 790restart:
791 rt = fn->leaf;
4c9483b2 792 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 793 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 794 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
4c9483b2 795 BACKTRACK(net, &fl6->saddr);
c71099ac 796out:
d8d1f30b 797 dst_use(&rt->dst, jiffies);
c71099ac 798 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 799 return rt;
800
801}
802
ea6e574e
FW		 803struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6,
804 int flags)
805{
806 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
807}
808EXPORT_SYMBOL_GPL(ip6_route_lookup);
809
9acd9f3a
YH		 810struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
811 const struct in6_addr *saddr, int oif, int strict)
c71099ac 812{
4c9483b2
DM		 813 struct flowi6 fl6 = {
814 .flowi6_oif = oif,
815 .daddr = *daddr,
c71099ac
TG		 816 };
817 struct dst_entry *dst;
77d16f45 818 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 819
adaa70bb 820 if (saddr) {
4c9483b2 821 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 822 flags |= RT6_LOOKUP_F_HAS_SADDR;
823 }
824
4c9483b2 825 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 826 if (dst->error == 0)
827 return (struct rt6_info *) dst;
828
829 dst_release(dst);
830
1da177e4
LT		 831 return NULL;
832}
833
7159039a
YH		 834EXPORT_SYMBOL(rt6_lookup);
835
c71099ac 836/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 837 It takes new route entry, the addition fails by any reason the
838 route is freed. In any case, if caller does not hold it, it may
839 be destroyed.
840 */
841
86872cb5 842static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 843{
844 int err;
c71099ac 845 struct fib6_table *table;
1da177e4 846
c71099ac
TG		 847 table = rt->rt6i_table;
848 write_lock_bh(&table->tb6_lock);
86872cb5 849 err = fib6_add(&table->tb6_root, rt, info);
c71099ac 850 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 851
852 return err;
853}
854
40e22e8f
TG		 855int ip6_ins_rt(struct rt6_info *rt)
856{
4d1169c1 857 struct nl_info info = {
d1918542 858 .nl_net = dev_net(rt->dst.dev),
4d1169c1 859 };
528c4ceb 860 return __ip6_ins_rt(rt, &info);
40e22e8f
TG		 861}
862
1716a961 863static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
21efcfa0 864 const struct in6_addr *daddr,
b71d1d42 865 const struct in6_addr *saddr)
1da177e4 866{
1da177e4
LT		 867 struct rt6_info *rt;
868
869 /*
870 * Clone the route.
871 */
872
21efcfa0 873 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 874
875 if (rt) {
249a3630
DJ		 876 if (ort->rt6i_dst.plen != 128 &&
877 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
878 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 879
1da177e4 880 rt->rt6i_flags |= RTF_CACHE;
1da177e4
LT		 881
882#ifdef CONFIG_IPV6_SUBTREES
883 if (rt->rt6i_src.plen && saddr) {
4e3fd7a0 884 rt->rt6i_src.addr = *saddr;
1da177e4
LT		 885 rt->rt6i_src.plen = 128;
886 }
887#endif
95a9a5ba 888 }
1da177e4 889
95a9a5ba
YH		 890 return rt;
891}
1da177e4 892
21efcfa0
ED		 893static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
894 const struct in6_addr *daddr)
299d9939 895{
21efcfa0
ED		 896 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
897
887c95cc 898 if (rt)
299d9939 899 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 900 return rt;
901}
902
8ed67789 903static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 904 struct flowi6 *fl6, int flags)
1da177e4
LT		 905{
906 struct fib6_node *fn;
519fbd87 907 struct rt6_info *rt, *nrt;
c71099ac 908 int strict = 0;
1da177e4 909 int attempts = 3;
519fbd87 910 int err;
53b7997f 911 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
1da177e4 912
77d16f45 913 strict |= flags & RT6_LOOKUP_F_IFACE;
1da177e4
LT		 914
915relookup:
c71099ac 916 read_lock_bh(&table->tb6_lock);
1da177e4 917
8238dd06 918restart_2:
4c9483b2 919 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1da177e4
LT		 920
921restart:
4acad72d 922 rt = rt6_select(fn, oif, strict | reachable);
52bd4c0c
ND		 923 if (rt->rt6i_nsiblings)
924 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable);
4c9483b2 925 BACKTRACK(net, &fl6->saddr);
8ed67789 926 if (rt == net->ipv6.ip6_null_entry ||
8238dd06 927 rt->rt6i_flags & RTF_CACHE)
1ddef044 928 goto out;
1da177e4 929
d8d1f30b 930 dst_hold(&rt->dst);
c71099ac 931 read_unlock_bh(&table->tb6_lock);
fb9de91e 932
c440f160 933 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
4c9483b2 934 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
7343ff31 935 else if (!(rt->dst.flags & DST_HOST))
4c9483b2 936 nrt = rt6_alloc_clone(rt, &fl6->daddr);
7343ff31
DM		 937 else
938 goto out2;
e40cf353 939
94e187c0 940 ip6_rt_put(rt);
8ed67789 941 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 942
d8d1f30b 943 dst_hold(&rt->dst);
519fbd87 944 if (nrt) {
40e22e8f 945 err = ip6_ins_rt(nrt);
519fbd87 946 if (!err)
1da177e4 947 goto out2;
1da177e4 948 }
1da177e4 949
519fbd87
YH		 950 if (--attempts <= 0)
951 goto out2;
952
953 /*
c71099ac 954 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 955 * released someone could insert this route. Relookup.
956 */
94e187c0 957 ip6_rt_put(rt);
519fbd87
YH		 958 goto relookup;
959
960out:
8238dd06
YH		 961 if (reachable) {
962 reachable = 0;
963 goto restart_2;
964 }
d8d1f30b 965 dst_hold(&rt->dst);
c71099ac 966 read_unlock_bh(&table->tb6_lock);
1da177e4 967out2:
d8d1f30b
CG		 968 rt->dst.lastuse = jiffies;
969 rt->dst.__use++;
c71099ac
TG		 970
971 return rt;
1da177e4
LT		 972}
973
8ed67789 974static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 975 struct flowi6 *fl6, int flags)
4acad72d 976{
4c9483b2 977 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 978}
979
72331bc0
SL		 980static struct dst_entry *ip6_route_input_lookup(struct net *net,
981 struct net_device *dev,
982 struct flowi6 *fl6, int flags)
983{
984 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
985 flags |= RT6_LOOKUP_F_IFACE;
986
987 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
988}
989
c71099ac
TG		 990void ip6_route_input(struct sk_buff *skb)
991{
b71d1d42 992 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 993 struct net *net = dev_net(skb->dev);
adaa70bb 994 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 995 struct flowi6 fl6 = {
996 .flowi6_iif = skb->dev->ifindex,
997 .daddr = iph->daddr,
998 .saddr = iph->saddr,
6502ca52 999 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1000 .flowi6_mark = skb->mark,
1001 .flowi6_proto = iph->nexthdr,
c71099ac 1002 };
adaa70bb 1003
72331bc0 1004 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1005}
1006
8ed67789 1007static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1008 struct flowi6 *fl6, int flags)
1da177e4 1009{
4c9483b2 1010 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1011}
1012
9c7a4f9c 1013struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 1014 struct flowi6 *fl6)
c71099ac
TG		 1015{
1016 int flags = 0;
1017
1fb9489b 1018 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1019
4c9483b2 1020 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 1021 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1022
4c9483b2 1023 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 1024 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1025 else if (sk)
1026 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1027
4c9483b2 1028 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4
LT		 1029}
1030
7159039a 1031EXPORT_SYMBOL(ip6_route_output);
1da177e4 1032
2774c131 1033struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1034{
5c1e6aa3 1035 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1036 struct dst_entry *new = NULL;
1037
f5b0a874 1038 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1039 if (rt) {
d8d1f30b 1040 new = &rt->dst;
14e50e57 1041
8104891b
SK		 1042 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1043 rt6_init_peer(rt, net->ipv6.peers);
1044
14e50e57 1045 new->__use = 1;
352e512c
HX		 1046 new->input = dst_discard;
1047 new->output = dst_discard;
14e50e57 1048
21efcfa0
ED		 1049 if (dst_metrics_read_only(&ort->dst))
1050 new->_metrics = ort->dst._metrics;
1051 else
1052 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1053 rt->rt6i_idev = ort->rt6i_idev;
1054 if (rt->rt6i_idev)
1055 in6_dev_hold(rt->rt6i_idev);
14e50e57 1056
4e3fd7a0 1057 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1058 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1059 rt->rt6i_metric = 0;
1060
1061 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1062#ifdef CONFIG_IPV6_SUBTREES
1063 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1064#endif
1065
1066 dst_free(new);
1067 }
1068
69ead7af
DM		 1069 dst_release(dst_orig);
1070 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1071}
14e50e57 1072
1da177e4
LT		 1073/*
1074 * Destination cache support functions
1075 */
1076
1077static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1078{
1079 struct rt6_info *rt;
1080
1081 rt = (struct rt6_info *) dst;
1082
6f3118b5
ND		 1083 /* All IPV6 dsts are created with ->obsolete set to the value
1084 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1085 * into this function always.
1086 */
ca4c3fc2 1087 if (rt->rt6i_genid != rt_genid_ipv6(dev_net(rt->dst.dev)))
6f3118b5
ND		 1088 return NULL;
1089
e3bc10bd
HFS		 1090 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1091 return NULL;
a4477c4d 1092
e3bc10bd
HFS		 1093 if (rt6_check_expired(rt))
1094 return NULL;
1095
1096 return dst;
1da177e4
LT		 1097}
1098
1099static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1100{
1101 struct rt6_info *rt = (struct rt6_info *) dst;
1102
1103 if (rt) {
54c1a859
YH		 1104 if (rt->rt6i_flags & RTF_CACHE) {
1105 if (rt6_check_expired(rt)) {
1106 ip6_del_rt(rt);
1107 dst = NULL;
1108 }
1109 } else {
1da177e4 1110 dst_release(dst);
54c1a859
YH		 1111 dst = NULL;
1112 }
1da177e4 1113 }
54c1a859 1114 return dst;
1da177e4
LT		 1115}
1116
1117static void ip6_link_failure(struct sk_buff *skb)
1118{
1119 struct rt6_info *rt;
1120
3ffe533c 1121 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1122
adf30907 1123 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1124 if (rt) {
1eb4f758
HFS		 1125 if (rt->rt6i_flags & RTF_CACHE) {
1126 dst_hold(&rt->dst);
1127 if (ip6_del_rt(rt))
1128 dst_free(&rt->dst);
1129 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1130 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1131 }
1da177e4
LT		 1132 }
1133}
1134
6700c270
DM		 1135static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1136 struct sk_buff *skb, u32 mtu)
1da177e4
LT		 1137{
1138 struct rt6_info *rt6 = (struct rt6_info*)dst;
1139
81aded24 1140 dst_confirm(dst);
1da177e4 1141 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
81aded24
DM		 1142 struct net *net = dev_net(dst->dev);
1143
1da177e4
LT		 1144 rt6->rt6i_flags |= RTF_MODIFIED;
1145 if (mtu < IPV6_MIN_MTU) {
defb3519 1146 u32 features = dst_metric(dst, RTAX_FEATURES);
1da177e4 1147 mtu = IPV6_MIN_MTU;
defb3519
DM		 1148 features |= RTAX_FEATURE_ALLFRAG;
1149 dst_metric_set(dst, RTAX_FEATURES, features);
1da177e4 1150 }
defb3519 1151 dst_metric_set(dst, RTAX_MTU, mtu);
81aded24 1152 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1153 }
1154}
1155
42ae66c8
DM		 1156void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1157 int oif, u32 mark)
81aded24
DM		 1158{
1159 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1160 struct dst_entry *dst;
1161 struct flowi6 fl6;
1162
1163 memset(&fl6, 0, sizeof(fl6));
1164 fl6.flowi6_oif = oif;
1165 fl6.flowi6_mark = mark;
81aded24
DM		 1166 fl6.daddr = iph->daddr;
1167 fl6.saddr = iph->saddr;
6502ca52 1168 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1169
1170 dst = ip6_route_output(net, NULL, &fl6);
1171 if (!dst->error)
6700c270 1172 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
81aded24
DM		 1173 dst_release(dst);
1174}
1175EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1176
1177void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1178{
1179 ip6_update_pmtu(skb, sock_net(sk), mtu,
1180 sk->sk_bound_dev_if, sk->sk_mark);
1181}
1182EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1183
b55b76b2
DJ		 1184/* Handle redirects */
1185struct ip6rd_flowi {
1186 struct flowi6 fl6;
1187 struct in6_addr gateway;
1188};
1189
1190static struct rt6_info *__ip6_route_redirect(struct net *net,
1191 struct fib6_table *table,
1192 struct flowi6 *fl6,
1193 int flags)
1194{
1195 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1196 struct rt6_info *rt;
1197 struct fib6_node *fn;
1198
1199 /* Get the "current" route for this destination and
1200 * check if the redirect has come from approriate router.
1201 *
1202 * RFC 4861 specifies that redirects should only be
1203 * accepted if they come from the nexthop to the target.
1204 * Due to the way the routes are chosen, this notion
1205 * is a bit fuzzy and one might need to check all possible
1206 * routes.
1207 */
1208
1209 read_lock_bh(&table->tb6_lock);
1210 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1211restart:
1212 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1213 if (rt6_check_expired(rt))
1214 continue;
1215 if (rt->dst.error)
1216 break;
1217 if (!(rt->rt6i_flags & RTF_GATEWAY))
1218 continue;
1219 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1220 continue;
1221 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1222 continue;
1223 break;
1224 }
1225
1226 if (!rt)
1227 rt = net->ipv6.ip6_null_entry;
1228 else if (rt->dst.error) {
1229 rt = net->ipv6.ip6_null_entry;
1230 goto out;
1231 }
1232 BACKTRACK(net, &fl6->saddr);
1233out:
1234 dst_hold(&rt->dst);
1235
1236 read_unlock_bh(&table->tb6_lock);
1237
1238 return rt;
1239};
1240
1241static struct dst_entry *ip6_route_redirect(struct net *net,
1242 const struct flowi6 *fl6,
1243 const struct in6_addr *gateway)
1244{
1245 int flags = RT6_LOOKUP_F_HAS_SADDR;
1246 struct ip6rd_flowi rdfl;
1247
1248 rdfl.fl6 = *fl6;
1249 rdfl.gateway = *gateway;
1250
1251 return fib6_rule_lookup(net, &rdfl.fl6,
1252 flags, __ip6_route_redirect);
1253}
1254
3a5ad2ee
DM		 1255void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1256{
1257 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1258 struct dst_entry *dst;
1259 struct flowi6 fl6;
1260
1261 memset(&fl6, 0, sizeof(fl6));
1262 fl6.flowi6_oif = oif;
1263 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1264 fl6.daddr = iph->daddr;
1265 fl6.saddr = iph->saddr;
6502ca52 1266 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1267
b55b76b2
DJ		 1268 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1269 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1270 dst_release(dst);
1271}
1272EXPORT_SYMBOL_GPL(ip6_redirect);
1273
c92a59ec
DJ		 1274void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1275 u32 mark)
1276{
1277 const struct ipv6hdr *iph = ipv6_hdr(skb);
1278 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1279 struct dst_entry *dst;
1280 struct flowi6 fl6;
1281
1282 memset(&fl6, 0, sizeof(fl6));
1283 fl6.flowi6_oif = oif;
1284 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1285 fl6.daddr = msg->dest;
1286 fl6.saddr = iph->daddr;
1287
b55b76b2
DJ		 1288 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1289 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1290 dst_release(dst);
1291}
1292
3a5ad2ee
DM		 1293void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1294{
1295 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1296}
1297EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1298
0dbaee3b 1299static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1300{
0dbaee3b
DM		 1301 struct net_device *dev = dst->dev;
1302 unsigned int mtu = dst_mtu(dst);
1303 struct net *net = dev_net(dev);
1304
1da177e4
LT		 1305 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1306
5578689a
DL		 1307 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1308 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1309
1310 /*
1ab1457c
YH		 1311 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1312 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1313 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1314 * rely only on pmtu discovery"
1315 */
1316 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1317 mtu = IPV6_MAXPLEN;
1318 return mtu;
1319}
1320
ebb762f2 1321static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1322{
d33e4553 1323 struct inet6_dev *idev;
618f9bc7
SK		 1324 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1325
1326 if (mtu)
1327 return mtu;
1328
1329 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1330
1331 rcu_read_lock();
1332 idev = __in6_dev_get(dst->dev);
1333 if (idev)
1334 mtu = idev->cnf.mtu6;
1335 rcu_read_unlock();
1336
1337 return mtu;
1338}
1339
3b00944c
YH		 1340static struct dst_entry *icmp6_dst_gc_list;
1341static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1342
3b00944c 1343struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1344 struct flowi6 *fl6)
1da177e4 1345{
87a11578 1346 struct dst_entry *dst;
1da177e4
LT		 1347 struct rt6_info *rt;
1348 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1349 struct net *net = dev_net(dev);
1da177e4 1350
38308473 1351 if (unlikely(!idev))
122bdf67 1352 return ERR_PTR(-ENODEV);
1da177e4 1353
8b96d22d 1354 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1355 if (unlikely(!rt)) {
1da177e4 1356 in6_dev_put(idev);
87a11578 1357 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1358 goto out;
1359 }
1360
8e2ec639
YZ		 1361 rt->dst.flags |= DST_HOST;
1362 rt->dst.output = ip6_output;
d8d1f30b 1363 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1364 rt->rt6i_gateway = fl6->daddr;
87a11578 1365 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1366 rt->rt6i_dst.plen = 128;
1367 rt->rt6i_idev = idev;
14edd87d 1368 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1369
3b00944c 1370 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1371 rt->dst.next = icmp6_dst_gc_list;
1372 icmp6_dst_gc_list = &rt->dst;
3b00944c 1373 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1374
5578689a 1375 fib6_force_start_gc(net);
1da177e4 1376
87a11578
DM		 1377 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1378
1da177e4 1379out:
87a11578 1380 return dst;
1da177e4
LT		 1381}
1382
3d0f24a7 1383int icmp6_dst_gc(void)
1da177e4 1384{
e9476e95 1385 struct dst_entry *dst, **pprev;
3d0f24a7 1386 int more = 0;
1da177e4 1387
3b00944c
YH		 1388 spin_lock_bh(&icmp6_dst_lock);
1389 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1390
1da177e4
LT		 1391 while ((dst = *pprev) != NULL) {
1392 if (!atomic_read(&dst->__refcnt)) {
1393 *pprev = dst->next;
1394 dst_free(dst);
1da177e4
LT		 1395 } else {
1396 pprev = &dst->next;
3d0f24a7 1397 ++more;
1da177e4
LT		 1398 }
1399 }
1400
3b00944c 1401 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1402
3d0f24a7 1403 return more;
1da177e4
LT		 1404}
1405
1e493d19
DM		 1406static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1407 void *arg)
1408{
1409 struct dst_entry *dst, **pprev;
1410
1411 spin_lock_bh(&icmp6_dst_lock);
1412 pprev = &icmp6_dst_gc_list;
1413 while ((dst = *pprev) != NULL) {
1414 struct rt6_info *rt = (struct rt6_info *) dst;
1415 if (func(rt, arg)) {
1416 *pprev = dst->next;
1417 dst_free(dst);
1418 } else {
1419 pprev = &dst->next;
1420 }
1421 }
1422 spin_unlock_bh(&icmp6_dst_lock);
1423}
1424
569d3645 1425static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1426{
86393e52 1427 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1428 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1429 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1430 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1431 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1432 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1433 int entries;
7019b78e 1434
fc66f95c 1435 entries = dst_entries_get_fast(ops);
49a18d86 1436 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1437 entries <= rt_max_size)
1da177e4
LT		 1438 goto out;
1439
6891a346 1440 net->ipv6.ip6_rt_gc_expire++;
2ac3ac8f 1441 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size);
fc66f95c
ED		 1442 entries = dst_entries_get_slow(ops);
1443 if (entries < ops->gc_thresh)
7019b78e 1444 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1445out:
7019b78e 1446 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1447 return entries > rt_max_size;
1da177e4
LT		 1448}
1449
1da177e4
LT		 1450/*
1451 *
1452 */
1453
86872cb5 1454int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1455{
1456 int err;
5578689a 1457 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1458 struct rt6_info *rt = NULL;
1459 struct net_device *dev = NULL;
1460 struct inet6_dev *idev = NULL;
c71099ac 1461 struct fib6_table *table;
1da177e4
LT		 1462 int addr_type;
1463
86872cb5 1464 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1465 return -EINVAL;
1466#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1467 if (cfg->fc_src_len)
1da177e4
LT		 1468 return -EINVAL;
1469#endif
86872cb5 1470 if (cfg->fc_ifindex) {
1da177e4 1471 err = -ENODEV;
5578689a 1472 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1473 if (!dev)
1474 goto out;
1475 idev = in6_dev_get(dev);
1476 if (!idev)
1477 goto out;
1478 }
1479
86872cb5
TG		 1480 if (cfg->fc_metric == 0)
1481 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1482
d71314b4 1483 err = -ENOBUFS;
38308473
DM		 1484 if (cfg->fc_nlinfo.nlh &&
1485 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1486 table = fib6_get_table(net, cfg->fc_table);
38308473 1487 if (!table) {
f3213831 1488 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1489 table = fib6_new_table(net, cfg->fc_table);
1490 }
1491 } else {
1492 table = fib6_new_table(net, cfg->fc_table);
1493 }
38308473
DM		 1494
1495 if (!table)
c71099ac 1496 goto out;
c71099ac 1497
8b96d22d 1498 rt = ip6_dst_alloc(net, NULL, DST_NOCOUNT, table);
1da177e4 1499
38308473 1500 if (!rt) {
1da177e4
LT		 1501 err = -ENOMEM;
1502 goto out;
1503 }
1504
1716a961
G		 1505 if (cfg->fc_flags & RTF_EXPIRES)
1506 rt6_set_expires(rt, jiffies +
1507 clock_t_to_jiffies(cfg->fc_expires));
1508 else
1509 rt6_clean_expires(rt);
1da177e4 1510
86872cb5
TG		 1511 if (cfg->fc_protocol == RTPROT_UNSPEC)
1512 cfg->fc_protocol = RTPROT_BOOT;
1513 rt->rt6i_protocol = cfg->fc_protocol;
1514
1515 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1516
1517 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1518 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1519 else if (cfg->fc_flags & RTF_LOCAL)
1520 rt->dst.input = ip6_input;
1da177e4 1521 else
d8d1f30b 1522 rt->dst.input = ip6_forward;
1da177e4 1523
d8d1f30b 1524 rt->dst.output = ip6_output;
1da177e4 1525
86872cb5
TG		 1526 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1527 rt->rt6i_dst.plen = cfg->fc_dst_len;
1da177e4 1528 if (rt->rt6i_dst.plen == 128)
11d53b49 1529 rt->dst.flags |= DST_HOST;
1da177e4 1530
8e2ec639
YZ		 1531 if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) {
1532 u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1533 if (!metrics) {
1534 err = -ENOMEM;
1535 goto out;
1536 }
1537 dst_init_metrics(&rt->dst, metrics, 0);
1538 }
1da177e4 1539#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1540 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1541 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1542#endif
1543
86872cb5 1544 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1545
1546 /* We cannot add true routes via loopback here,
1547 they would result in kernel looping; promote them to reject routes
1548 */
86872cb5 1549 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1550 (dev && (dev->flags & IFF_LOOPBACK) &&
1551 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1552 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1553 /* hold loopback dev/idev if we haven't done so. */
5578689a 1554 if (dev != net->loopback_dev) {
1da177e4
LT		 1555 if (dev) {
1556 dev_put(dev);
1557 in6_dev_put(idev);
1558 }
5578689a 1559 dev = net->loopback_dev;
1da177e4
LT		 1560 dev_hold(dev);
1561 idev = in6_dev_get(dev);
1562 if (!idev) {
1563 err = -ENODEV;
1564 goto out;
1565 }
1566 }
1da177e4 1567 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1568 switch (cfg->fc_type) {
1569 case RTN_BLACKHOLE:
1570 rt->dst.error = -EINVAL;
7150aede
K		 1571 rt->dst.output = dst_discard;
1572 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1573 break;
1574 case RTN_PROHIBIT:
1575 rt->dst.error = -EACCES;
7150aede
K		 1576 rt->dst.output = ip6_pkt_prohibit_out;
1577 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1578 break;
b4949ab2 1579 case RTN_THROW:
ef2c7d7b 1580 default:
7150aede
K		 1581 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1582 : -ENETUNREACH;
1583 rt->dst.output = ip6_pkt_discard_out;
1584 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1585 break;
1586 }
1da177e4
LT		 1587 goto install_route;
1588 }
1589
86872cb5 1590 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1591 const struct in6_addr *gw_addr;
1da177e4
LT		 1592 int gwa_type;
1593
86872cb5 1594 gw_addr = &cfg->fc_gateway;
4e3fd7a0 1595 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1596 gwa_type = ipv6_addr_type(gw_addr);
1597
1598 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1599 struct rt6_info *grt;
1600
1601 /* IPv6 strictly inhibits using not link-local
1602 addresses as nexthop address.
1603 Otherwise, router will not able to send redirects.
1604 It is very good, but in some (rare!) circumstances
1605 (SIT, PtP, NBMA NOARP links) it is handy to allow
1606 some exceptions. --ANK
1607 */
1608 err = -EINVAL;
38308473 1609 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1610 goto out;
1611
5578689a 1612 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1613
1614 err = -EHOSTUNREACH;
38308473 1615 if (!grt)
1da177e4
LT		 1616 goto out;
1617 if (dev) {
d1918542 1618 if (dev != grt->dst.dev) {
94e187c0 1619 ip6_rt_put(grt);
1da177e4
LT		 1620 goto out;
1621 }
1622 } else {
d1918542 1623 dev = grt->dst.dev;
1da177e4
LT		 1624 idev = grt->rt6i_idev;
1625 dev_hold(dev);
1626 in6_dev_hold(grt->rt6i_idev);
1627 }
38308473 1628 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1629 err = 0;
94e187c0 1630 ip6_rt_put(grt);
1da177e4
LT		 1631
1632 if (err)
1633 goto out;
1634 }
1635 err = -EINVAL;
38308473 1636 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1637 goto out;
1638 }
1639
1640 err = -ENODEV;
38308473 1641 if (!dev)
1da177e4
LT		 1642 goto out;
1643
c3968a85
DW		 1644 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1645 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1646 err = -EINVAL;
1647 goto out;
1648 }
4e3fd7a0 1649 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1650 rt->rt6i_prefsrc.plen = 128;
1651 } else
1652 rt->rt6i_prefsrc.plen = 0;
1653
86872cb5 1654 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1655
1656install_route:
86872cb5
TG		 1657 if (cfg->fc_mx) {
1658 struct nlattr *nla;
1659 int remaining;
1660
1661 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
8f4c1f9b 1662 int type = nla_type(nla);
86872cb5
TG		 1663
1664 if (type) {
1665 if (type > RTAX_MAX) {
1da177e4
LT		 1666 err = -EINVAL;
1667 goto out;
1668 }
86872cb5 1669
defb3519 1670 dst_metric_set(&rt->dst, type, nla_get_u32(nla));
1da177e4 1671 }
1da177e4
LT		 1672 }
1673 }
1674
d8d1f30b 1675 rt->dst.dev = dev;
1da177e4 1676 rt->rt6i_idev = idev;
c71099ac 1677 rt->rt6i_table = table;
63152fc0 1678
c346dca1 1679 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1680
86872cb5 1681 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1682
1683out:
1684 if (dev)
1685 dev_put(dev);
1686 if (idev)
1687 in6_dev_put(idev);
1688 if (rt)
d8d1f30b 1689 dst_free(&rt->dst);
1da177e4
LT		 1690 return err;
1691}
1692
86872cb5 1693static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1694{
1695 int err;
c71099ac 1696 struct fib6_table *table;
d1918542 1697 struct net *net = dev_net(rt->dst.dev);
1da177e4 1698
6825a26c
G		 1699 if (rt == net->ipv6.ip6_null_entry) {
1700 err = -ENOENT;
1701 goto out;
1702 }
6c813a72 1703
c71099ac
TG		 1704 table = rt->rt6i_table;
1705 write_lock_bh(&table->tb6_lock);
86872cb5 1706 err = fib6_del(rt, info);
c71099ac 1707 write_unlock_bh(&table->tb6_lock);
1da177e4 1708
6825a26c 1709out:
94e187c0 1710 ip6_rt_put(rt);
1da177e4
LT		 1711 return err;
1712}
1713
e0a1ad73
TG		 1714int ip6_del_rt(struct rt6_info *rt)
1715{
4d1169c1 1716 struct nl_info info = {
d1918542 1717 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1718 };
528c4ceb 1719 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1720}
1721
86872cb5 1722static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1723{
c71099ac 1724 struct fib6_table *table;
1da177e4
LT		 1725 struct fib6_node *fn;
1726 struct rt6_info *rt;
1727 int err = -ESRCH;
1728
5578689a 1729 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1730 if (!table)
c71099ac
TG		 1731 return err;
1732
1733 read_lock_bh(&table->tb6_lock);
1da177e4 1734
c71099ac 1735 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1736 &cfg->fc_dst, cfg->fc_dst_len,
1737 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1738
1da177e4 1739 if (fn) {
d8d1f30b 1740 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
86872cb5 1741 if (cfg->fc_ifindex &&
d1918542
DM		 1742 (!rt->dst.dev ||
1743 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1744 continue;
86872cb5
TG		 1745 if (cfg->fc_flags & RTF_GATEWAY &&
1746 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1747 continue;
86872cb5 1748 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1749 continue;
d8d1f30b 1750 dst_hold(&rt->dst);
c71099ac 1751 read_unlock_bh(&table->tb6_lock);
1da177e4 1752
86872cb5 1753 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1754 }
1755 }
c71099ac 1756 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1757
1758 return err;
1759}
1760
6700c270 1761static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1762{
e8599ff4 1763 struct net *net = dev_net(skb->dev);
a6279458 1764 struct netevent_redirect netevent;
e8599ff4 1765 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1766 struct ndisc_options ndopts;
1767 struct inet6_dev *in6_dev;
1768 struct neighbour *neigh;
71bcdba0 1769 struct rd_msg *msg;
6e157b6a
DM		 1770 int optlen, on_link;
1771 u8 *lladdr;
e8599ff4 1772
29a3cad5 1773 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1774 optlen -= sizeof(*msg);
e8599ff4
DM		 1775
1776 if (optlen < 0) {
6e157b6a 1777 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1778 return;
1779 }
1780
71bcdba0 1781 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1782
71bcdba0 1783 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1784 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1785 return;
1786 }
1787
6e157b6a 1788 on_link = 0;
71bcdba0 1789 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1790 on_link = 1;
71bcdba0 1791 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1792 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1793 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1794 return;
1795 }
1796
1797 in6_dev = __in6_dev_get(skb->dev);
1798 if (!in6_dev)
1799 return;
1800 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1801 return;
1802
1803 /* RFC2461 8.1:
1804 * The IP source address of the Redirect MUST be the same as the current
1805 * first-hop router for the specified ICMP Destination Address.
1806 */
1807
71bcdba0 1808 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1809 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1810 return;
1811 }
6e157b6a
DM		 1812
1813 lladdr = NULL;
e8599ff4
DM		 1814 if (ndopts.nd_opts_tgt_lladdr) {
1815 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1816 skb->dev);
1817 if (!lladdr) {
1818 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1819 return;
1820 }
1821 }
1822
6e157b6a
DM		 1823 rt = (struct rt6_info *) dst;
1824 if (rt == net->ipv6.ip6_null_entry) {
1825 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1826 return;
6e157b6a 1827 }
e8599ff4 1828
6e157b6a
DM		 1829 /* Redirect received -> path was valid.
1830 * Look, redirects are sent only in response to data packets,
1831 * so that this nexthop apparently is reachable. --ANK
1832 */
1833 dst_confirm(&rt->dst);
a6279458 1834
71bcdba0 1835 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1836 if (!neigh)
1837 return;
a6279458 1838
1da177e4
LT		 1839 /*
1840 * We have finally decided to accept it.
1841 */
1842
1ab1457c 1843 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1844 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1845 NEIGH_UPDATE_F_OVERRIDE|
1846 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1847 NEIGH_UPDATE_F_ISROUTER))
1848 );
1849
71bcdba0 1850 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1851 if (!nrt)
1da177e4
LT		 1852 goto out;
1853
1854 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1855 if (on_link)
1856 nrt->rt6i_flags &= ~RTF_GATEWAY;
1857
4e3fd7a0 1858 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1859
40e22e8f 1860 if (ip6_ins_rt(nrt))
1da177e4
LT		 1861 goto out;
1862
d8d1f30b
CG		 1863 netevent.old = &rt->dst;
1864 netevent.new = &nrt->dst;
71bcdba0 1865 netevent.daddr = &msg->dest;
60592833 1866 netevent.neigh = neigh;
8d71740c
TT		 1867 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1868
38308473 1869 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1870 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1871 ip6_del_rt(rt);
1da177e4
LT		 1872 }
1873
1874out:
e8599ff4 1875 neigh_release(neigh);
6e157b6a
DM		 1876}
1877
1da177e4
LT		 1878/*
1879 * Misc support functions
1880 */
1881
1716a961 1882static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1883 const struct in6_addr *dest)
1da177e4 1884{
d1918542 1885 struct net *net = dev_net(ort->dst.dev);
8b96d22d
DM		 1886 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1887 ort->rt6i_table);
1da177e4
LT		 1888
1889 if (rt) {
d8d1f30b
CG		 1890 rt->dst.input = ort->dst.input;
1891 rt->dst.output = ort->dst.output;
8e2ec639 1892 rt->dst.flags |= DST_HOST;
d8d1f30b 1893
4e3fd7a0 1894 rt->rt6i_dst.addr = *dest;
8e2ec639 1895 rt->rt6i_dst.plen = 128;
defb3519 1896 dst_copy_metrics(&rt->dst, &ort->dst);
d8d1f30b 1897 rt->dst.error = ort->dst.error;
1da177e4
LT		 1898 rt->rt6i_idev = ort->rt6i_idev;
1899 if (rt->rt6i_idev)
1900 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1901 rt->dst.lastuse = jiffies;
1da177e4 1902
550bab42
JA		 1903 if (ort->rt6i_flags & RTF_GATEWAY)
1904 rt->rt6i_gateway = ort->rt6i_gateway;
1905 else
1906 rt->rt6i_gateway = *dest;
1716a961 1907 rt->rt6i_flags = ort->rt6i_flags;
24f5b855 1908 rt6_set_from(rt, ort);
1da177e4
LT		 1909 rt->rt6i_metric = 0;
1910
1da177e4
LT		 1911#ifdef CONFIG_IPV6_SUBTREES
1912 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1913#endif
0f6c6392 1914 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1915 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1916 }
1917 return rt;
1918}
1919
70ceb4f5 1920#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1921static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 1922 const struct in6_addr *prefix, int prefixlen,
1923 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 1924{
1925 struct fib6_node *fn;
1926 struct rt6_info *rt = NULL;
c71099ac
TG		 1927 struct fib6_table *table;
1928
efa2cea0 1929 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 1930 if (!table)
c71099ac 1931 return NULL;
70ceb4f5 1932
5744dd9b 1933 read_lock_bh(&table->tb6_lock);
c71099ac 1934 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
70ceb4f5
YH		 1935 if (!fn)
1936 goto out;
1937
d8d1f30b 1938 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1939 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 1940 continue;
1941 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1942 continue;
1943 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1944 continue;
d8d1f30b 1945 dst_hold(&rt->dst);
70ceb4f5
YH		 1946 break;
1947 }
1948out:
5744dd9b 1949 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1950 return rt;
1951}
1952
efa2cea0 1953static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 1954 const struct in6_addr *prefix, int prefixlen,
1955 const struct in6_addr *gwaddr, int ifindex,
95c96174 1956 unsigned int pref)
70ceb4f5 1957{
86872cb5
TG		 1958 struct fib6_config cfg = {
1959 .fc_table = RT6_TABLE_INFO,
238fc7ea 1960 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1961 .fc_ifindex = ifindex,
1962 .fc_dst_len = prefixlen,
1963 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1964 RTF_UP | RTF_PREF(pref),
15e47304 1965 .fc_nlinfo.portid = 0,
efa2cea0
DL		 1966 .fc_nlinfo.nlh = NULL,
1967 .fc_nlinfo.nl_net = net,
86872cb5
TG		 1968 };
1969
4e3fd7a0
AD		 1970 cfg.fc_dst = *prefix;
1971 cfg.fc_gateway = *gwaddr;
70ceb4f5 1972
e317da96
YH		 1973 /* We should treat it as a default route if prefix length is 0. */
1974 if (!prefixlen)
86872cb5 1975 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 1976
86872cb5 1977 ip6_route_add(&cfg);
70ceb4f5 1978
efa2cea0 1979 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 1980}
1981#endif
1982
b71d1d42 1983struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 1984{
1da177e4 1985 struct rt6_info *rt;
c71099ac 1986 struct fib6_table *table;
1da177e4 1987
c346dca1 1988 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 1989 if (!table)
c71099ac 1990 return NULL;
1da177e4 1991
5744dd9b 1992 read_lock_bh(&table->tb6_lock);
d8d1f30b 1993 for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) {
d1918542 1994 if (dev == rt->dst.dev &&
045927ff 1995 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 1996 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1997 break;
1998 }
1999 if (rt)
d8d1f30b 2000 dst_hold(&rt->dst);
5744dd9b 2001 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2002 return rt;
2003}
2004
b71d1d42 2005struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2006 struct net_device *dev,
2007 unsigned int pref)
1da177e4 2008{
86872cb5
TG		 2009 struct fib6_config cfg = {
2010 .fc_table = RT6_TABLE_DFLT,
238fc7ea 2011 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2012 .fc_ifindex = dev->ifindex,
2013 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2014 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2015 .fc_nlinfo.portid = 0,
5578689a 2016 .fc_nlinfo.nlh = NULL,
c346dca1 2017 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2018 };
1da177e4 2019
4e3fd7a0 2020 cfg.fc_gateway = *gwaddr;
1da177e4 2021
86872cb5 2022 ip6_route_add(&cfg);
1da177e4 2023
1da177e4
LT		 2024 return rt6_get_dflt_router(gwaddr, dev);
2025}
2026
7b4da532 2027void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2028{
2029 struct rt6_info *rt;
c71099ac
TG		 2030 struct fib6_table *table;
2031
2032 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2033 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2034 if (!table)
c71099ac 2035 return;
1da177e4
LT		 2036
2037restart:
c71099ac 2038 read_lock_bh(&table->tb6_lock);
d8d1f30b 2039 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2040 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2041 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2042 dst_hold(&rt->dst);
c71099ac 2043 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2044 ip6_del_rt(rt);
1da177e4
LT		 2045 goto restart;
2046 }
2047 }
c71099ac 2048 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2049}
2050
5578689a
DL		 2051static void rtmsg_to_fib6_config(struct net *net,
2052 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2053 struct fib6_config *cfg)
2054{
2055 memset(cfg, 0, sizeof(*cfg));
2056
2057 cfg->fc_table = RT6_TABLE_MAIN;
2058 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2059 cfg->fc_metric = rtmsg->rtmsg_metric;
2060 cfg->fc_expires = rtmsg->rtmsg_info;
2061 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2062 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2063 cfg->fc_flags = rtmsg->rtmsg_flags;
2064
5578689a 2065 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2066
4e3fd7a0
AD		 2067 cfg->fc_dst = rtmsg->rtmsg_dst;
2068 cfg->fc_src = rtmsg->rtmsg_src;
2069 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2070}
2071
5578689a 2072int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2073{
86872cb5 2074 struct fib6_config cfg;
1da177e4
LT		 2075 struct in6_rtmsg rtmsg;
2076 int err;
2077
2078 switch(cmd) {
2079 case SIOCADDRT: /* Add a route */
2080 case SIOCDELRT: /* Delete a route */
af31f412 2081 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2082 return -EPERM;
2083 err = copy_from_user(&rtmsg, arg,
2084 sizeof(struct in6_rtmsg));
2085 if (err)
2086 return -EFAULT;
86872cb5 2087
5578689a 2088 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2089
1da177e4
LT		 2090 rtnl_lock();
2091 switch (cmd) {
2092 case SIOCADDRT:
86872cb5 2093 err = ip6_route_add(&cfg);
1da177e4
LT		 2094 break;
2095 case SIOCDELRT:
86872cb5 2096 err = ip6_route_del(&cfg);
1da177e4
LT		 2097 break;
2098 default:
2099 err = -EINVAL;
2100 }
2101 rtnl_unlock();
2102
2103 return err;
3ff50b79 2104 }
1da177e4
LT		 2105
2106 return -EINVAL;
2107}
2108
2109/*
2110 * Drop the packet on the floor
2111 */
2112
d5fdd6ba 2113static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2114{
612f09e8 2115 int type;
adf30907 2116 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2117 switch (ipstats_mib_noroutes) {
2118 case IPSTATS_MIB_INNOROUTES:
0660e03f 2119 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2120 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2121 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2122 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2123 break;
2124 }
2125 /* FALLTHROUGH */
2126 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2127 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2128 ipstats_mib_noroutes);
612f09e8
YH		 2129 break;
2130 }
3ffe533c 2131 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2132 kfree_skb(skb);
2133 return 0;
2134}
2135
9ce8ade0
TG		 2136static int ip6_pkt_discard(struct sk_buff *skb)
2137{
612f09e8 2138 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2139}
2140
20380731 2141static int ip6_pkt_discard_out(struct sk_buff *skb)
1da177e4 2142{
adf30907 2143 skb->dev = skb_dst(skb)->dev;
612f09e8 2144 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2145}
2146
9ce8ade0
TG		 2147static int ip6_pkt_prohibit(struct sk_buff *skb)
2148{
612f09e8 2149 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2150}
2151
2152static int ip6_pkt_prohibit_out(struct sk_buff *skb)
2153{
adf30907 2154 skb->dev = skb_dst(skb)->dev;
612f09e8 2155 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2156}
2157
1da177e4
LT		 2158/*
2159 * Allocate a dst for local (unicast / anycast) address.
2160 */
2161
2162struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2163 const struct in6_addr *addr,
8f031519 2164 bool anycast)
1da177e4 2165{
c346dca1 2166 struct net *net = dev_net(idev->dev);
a3300ef4
HFS		 2167 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2168 DST_NOCOUNT, NULL);
2169 if (!rt)
1da177e4
LT		 2170 return ERR_PTR(-ENOMEM);
2171
1da177e4
LT		 2172 in6_dev_hold(idev);
2173
11d53b49 2174 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2175 rt->dst.input = ip6_input;
2176 rt->dst.output = ip6_output;
1da177e4 2177 rt->rt6i_idev = idev;
1da177e4
LT		 2178
2179 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2180 if (anycast)
2181 rt->rt6i_flags |= RTF_ANYCAST;
2182 else
1da177e4 2183 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2184
550bab42 2185 rt->rt6i_gateway = *addr;
4e3fd7a0 2186 rt->rt6i_dst.addr = *addr;
1da177e4 2187 rt->rt6i_dst.plen = 128;
5578689a 2188 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2189
d8d1f30b 2190 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2191
2192 return rt;
2193}
2194
c3968a85
DW		 2195int ip6_route_get_saddr(struct net *net,
2196 struct rt6_info *rt,
b71d1d42 2197 const struct in6_addr *daddr,
c3968a85
DW		 2198 unsigned int prefs,
2199 struct in6_addr *saddr)
2200{
2201 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt);
2202 int err = 0;
2203 if (rt->rt6i_prefsrc.plen)
4e3fd7a0 2204 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2205 else
2206 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2207 daddr, prefs, saddr);
2208 return err;
2209}
2210
2211/* remove deleted ip from prefsrc entries */
2212struct arg_dev_net_ip {
2213 struct net_device *dev;
2214 struct net *net;
2215 struct in6_addr *addr;
2216};
2217
2218static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2219{
2220 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2221 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2222 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2223
d1918542 2224 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2225 rt != net->ipv6.ip6_null_entry &&
2226 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2227 /* remove prefsrc entry */
2228 rt->rt6i_prefsrc.plen = 0;
2229 }
2230 return 0;
2231}
2232
2233void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2234{
2235 struct net *net = dev_net(ifp->idev->dev);
2236 struct arg_dev_net_ip adni = {
2237 .dev = ifp->idev->dev,
2238 .net = net,
2239 .addr = &ifp->addr,
2240 };
2241 fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni);
2242}
2243
8ed67789
DL		 2244struct arg_dev_net {
2245 struct net_device *dev;
2246 struct net *net;
2247};
2248
1da177e4
LT		 2249static int fib6_ifdown(struct rt6_info *rt, void *arg)
2250{
bc3ef660 2251 const struct arg_dev_net *adn = arg;
2252 const struct net_device *dev = adn->dev;
8ed67789 2253
d1918542 2254 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2255 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2256 return -1;
c159d30c 2257
1da177e4
LT		 2258 return 0;
2259}
2260
f3db4851 2261void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2262{
8ed67789
DL		 2263 struct arg_dev_net adn = {
2264 .dev = dev,
2265 .net = net,
2266 };
2267
2268 fib6_clean_all(net, fib6_ifdown, 0, &adn);
1e493d19 2269 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2270}
2271
95c96174 2272struct rt6_mtu_change_arg {
1da177e4 2273 struct net_device *dev;
95c96174 2274 unsigned int mtu;
1da177e4
LT		 2275};
2276
2277static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2278{
2279 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2280 struct inet6_dev *idev;
2281
2282 /* In IPv6 pmtu discovery is not optional,
2283 so that RTAX_MTU lock cannot disable it.
2284 We still use this lock to block changes
2285 caused by addrconf/ndisc.
2286 */
2287
2288 idev = __in6_dev_get(arg->dev);
38308473 2289 if (!idev)
1da177e4
LT		 2290 return 0;
2291
2292 /* For administrative MTU increase, there is no way to discover
2293 IPv6 PMTU increase, so PMTU increase should be updated here.
2294 Since RFC 1981 doesn't include administrative MTU increase
2295 update PMTU increase is a MUST. (i.e. jumbo frame)
2296 */
2297 /*
2298 If new MTU is less than route PMTU, this new MTU will be the
2299 lowest MTU in the path, update the route PMTU to reflect PMTU
2300 decreases; if new MTU is greater than route PMTU, and the
2301 old MTU is the lowest MTU in the path, update the route PMTU
2302 to reflect the increase. In this case if the other nodes' MTU
2303 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2304 PMTU discouvery.
2305 */
d1918542 2306 if (rt->dst.dev == arg->dev &&
d8d1f30b
CG		 2307 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2308 (dst_mtu(&rt->dst) >= arg->mtu ||
2309 (dst_mtu(&rt->dst) < arg->mtu &&
2310 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
defb3519 2311 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
566cfd8f 2312 }
1da177e4
LT		 2313 return 0;
2314}
2315
95c96174 2316void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2317{
c71099ac
TG		 2318 struct rt6_mtu_change_arg arg = {
2319 .dev = dev,
2320 .mtu = mtu,
2321 };
1da177e4 2322
c346dca1 2323 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg);
1da177e4
LT		 2324}
2325
ef7c79ed 2326static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2327 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2328 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2329 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2330 [RTA_PRIORITY] = { .type = NLA_U32 },
2331 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2332 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
86872cb5
TG		 2333};
2334
2335static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2336 struct fib6_config *cfg)
1da177e4 2337{
86872cb5
TG		 2338 struct rtmsg *rtm;
2339 struct nlattr *tb[RTA_MAX+1];
2340 int err;
1da177e4 2341
86872cb5
TG		 2342 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2343 if (err < 0)
2344 goto errout;
1da177e4 2345
86872cb5
TG		 2346 err = -EINVAL;
2347 rtm = nlmsg_data(nlh);
2348 memset(cfg, 0, sizeof(*cfg));
2349
2350 cfg->fc_table = rtm->rtm_table;
2351 cfg->fc_dst_len = rtm->rtm_dst_len;
2352 cfg->fc_src_len = rtm->rtm_src_len;
2353 cfg->fc_flags = RTF_UP;
2354 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2355 cfg->fc_type = rtm->rtm_type;
86872cb5 2356
ef2c7d7b
ND		 2357 if (rtm->rtm_type == RTN_UNREACHABLE ||
2358 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2359 rtm->rtm_type == RTN_PROHIBIT ||
2360 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2361 cfg->fc_flags |= RTF_REJECT;
2362
ab79ad14
MŻ		 2363 if (rtm->rtm_type == RTN_LOCAL)
2364 cfg->fc_flags |= RTF_LOCAL;
2365
15e47304 2366 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2367 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2368 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2369
2370 if (tb[RTA_GATEWAY]) {
2371 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2372 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2373 }
86872cb5
TG		 2374
2375 if (tb[RTA_DST]) {
2376 int plen = (rtm->rtm_dst_len + 7) >> 3;
2377
2378 if (nla_len(tb[RTA_DST]) < plen)
2379 goto errout;
2380
2381 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2382 }
86872cb5
TG		 2383
2384 if (tb[RTA_SRC]) {
2385 int plen = (rtm->rtm_src_len + 7) >> 3;
2386
2387 if (nla_len(tb[RTA_SRC]) < plen)
2388 goto errout;
2389
2390 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2391 }
86872cb5 2392
c3968a85
DW		 2393 if (tb[RTA_PREFSRC])
2394 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2395
86872cb5
TG		 2396 if (tb[RTA_OIF])
2397 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2398
2399 if (tb[RTA_PRIORITY])
2400 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2401
2402 if (tb[RTA_METRICS]) {
2403 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2404 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2405 }
86872cb5
TG		 2406
2407 if (tb[RTA_TABLE])
2408 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2409
51ebd318
ND		 2410 if (tb[RTA_MULTIPATH]) {
2411 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2412 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2413 }
2414
86872cb5
TG		 2415 err = 0;
2416errout:
2417 return err;
1da177e4
LT		 2418}
2419
51ebd318
ND		 2420static int ip6_route_multipath(struct fib6_config *cfg, int add)
2421{
2422 struct fib6_config r_cfg;
2423 struct rtnexthop *rtnh;
2424 int remaining;
2425 int attrlen;
2426 int err = 0, last_err = 0;
2427
2428beginning:
2429 rtnh = (struct rtnexthop *)cfg->fc_mp;
2430 remaining = cfg->fc_mp_len;
2431
2432 /* Parse a Multipath Entry */
2433 while (rtnh_ok(rtnh, remaining)) {
2434 memcpy(&r_cfg, cfg, sizeof(*cfg));
2435 if (rtnh->rtnh_ifindex)
2436 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2437
2438 attrlen = rtnh_attrlen(rtnh);
2439 if (attrlen > 0) {
2440 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2441
2442 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2443 if (nla) {
2444 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2445 r_cfg.fc_flags |= RTF_GATEWAY;
2446 }
2447 }
2448 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2449 if (err) {
2450 last_err = err;
2451 /* If we are trying to remove a route, do not stop the
2452 * loop when ip6_route_del() fails (because next hop is
2453 * already gone), we should try to remove all next hops.
2454 */
2455 if (add) {
2456 /* If add fails, we should try to delete all
2457 * next hops that have been already added.
2458 */
2459 add = 0;
2460 goto beginning;
2461 }
2462 }
1a72418b
ND		 2463 /* Because each route is added like a single route we remove
2464 * this flag after the first nexthop (if there is a collision,
2465 * we have already fail to add the first nexthop:
2466 * fib6_add_rt2node() has reject it).
2467 */
2468 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
51ebd318
ND		 2469 rtnh = rtnh_next(rtnh, &remaining);
2470 }
2471
2472 return last_err;
2473}
2474
661d2967 2475static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2476{
86872cb5
TG		 2477 struct fib6_config cfg;
2478 int err;
1da177e4 2479
86872cb5
TG		 2480 err = rtm_to_fib6_config(skb, nlh, &cfg);
2481 if (err < 0)
2482 return err;
2483
51ebd318
ND		 2484 if (cfg.fc_mp)
2485 return ip6_route_multipath(&cfg, 0);
2486 else
2487 return ip6_route_del(&cfg);
1da177e4
LT		 2488}
2489
661d2967 2490static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2491{
86872cb5
TG		 2492 struct fib6_config cfg;
2493 int err;
1da177e4 2494
86872cb5
TG		 2495 err = rtm_to_fib6_config(skb, nlh, &cfg);
2496 if (err < 0)
2497 return err;
2498
51ebd318
ND		 2499 if (cfg.fc_mp)
2500 return ip6_route_multipath(&cfg, 1);
2501 else
2502 return ip6_route_add(&cfg);
1da177e4
LT		 2503}
2504
339bf98f
TG		 2505static inline size_t rt6_nlmsg_size(void)
2506{
2507 return NLMSG_ALIGN(sizeof(struct rtmsg))
2508 + nla_total_size(16) /* RTA_SRC */
2509 + nla_total_size(16) /* RTA_DST */
2510 + nla_total_size(16) /* RTA_GATEWAY */
2511 + nla_total_size(16) /* RTA_PREFSRC */
2512 + nla_total_size(4) /* RTA_TABLE */
2513 + nla_total_size(4) /* RTA_IIF */
2514 + nla_total_size(4) /* RTA_OIF */
2515 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2516 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
339bf98f
TG		 2517 + nla_total_size(sizeof(struct rta_cacheinfo));
2518}
2519
191cd582
BH		 2520static int rt6_fill_node(struct net *net,
2521 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2522 struct in6_addr *dst, struct in6_addr *src,
15e47304 2523 int iif, int type, u32 portid, u32 seq,
7bc570c8 2524 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2525{
2526 struct rtmsg *rtm;
2d7202bf 2527 struct nlmsghdr *nlh;
e3703b3d 2528 long expires;
9e762a4a 2529 u32 table;
1da177e4
LT		 2530
2531 if (prefix) { /* user wants prefix routes only */
2532 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2533 /* success since this is not a prefix route */
2534 return 1;
2535 }
2536 }
2537
15e47304 2538 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2539 if (!nlh)
26932566 2540 return -EMSGSIZE;
2d7202bf
TG		 2541
2542 rtm = nlmsg_data(nlh);
1da177e4
LT		 2543 rtm->rtm_family = AF_INET6;
2544 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2545 rtm->rtm_src_len = rt->rt6i_src.plen;
2546 rtm->rtm_tos = 0;
c71099ac 2547 if (rt->rt6i_table)
9e762a4a 2548 table = rt->rt6i_table->tb6_id;
c71099ac 2549 else
9e762a4a
PM		 2550 table = RT6_TABLE_UNSPEC;
2551 rtm->rtm_table = table;
c78679e8
DM		 2552 if (nla_put_u32(skb, RTA_TABLE, table))
2553 goto nla_put_failure;
ef2c7d7b
ND		 2554 if (rt->rt6i_flags & RTF_REJECT) {
2555 switch (rt->dst.error) {
2556 case -EINVAL:
2557 rtm->rtm_type = RTN_BLACKHOLE;
2558 break;
2559 case -EACCES:
2560 rtm->rtm_type = RTN_PROHIBIT;
2561 break;
b4949ab2
ND		 2562 case -EAGAIN:
2563 rtm->rtm_type = RTN_THROW;
2564 break;
ef2c7d7b
ND		 2565 default:
2566 rtm->rtm_type = RTN_UNREACHABLE;
2567 break;
2568 }
2569 }
38308473 2570 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2571 rtm->rtm_type = RTN_LOCAL;
d1918542 2572 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2573 rtm->rtm_type = RTN_LOCAL;
2574 else
2575 rtm->rtm_type = RTN_UNICAST;
2576 rtm->rtm_flags = 0;
2577 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2578 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2579 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2580 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2581 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2582 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2583 rtm->rtm_protocol = RTPROT_RA;
2584 else
2585 rtm->rtm_protocol = RTPROT_KERNEL;
2586 }
1da177e4 2587
38308473 2588 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2589 rtm->rtm_flags |= RTM_F_CLONED;
2590
2591 if (dst) {
c78679e8
DM		 2592 if (nla_put(skb, RTA_DST, 16, dst))
2593 goto nla_put_failure;
1ab1457c 2594 rtm->rtm_dst_len = 128;
1da177e4 2595 } else if (rtm->rtm_dst_len)
c78679e8
DM		 2596 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2597 goto nla_put_failure;
1da177e4
LT		 2598#ifdef CONFIG_IPV6_SUBTREES
2599 if (src) {
c78679e8
DM		 2600 if (nla_put(skb, RTA_SRC, 16, src))
2601 goto nla_put_failure;
1ab1457c 2602 rtm->rtm_src_len = 128;
c78679e8
DM		 2603 } else if (rtm->rtm_src_len &&
2604 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2605 goto nla_put_failure;
1da177e4 2606#endif
7bc570c8
YH		 2607 if (iif) {
2608#ifdef CONFIG_IPV6_MROUTE
2609 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2610 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2611 if (err <= 0) {
2612 if (!nowait) {
2613 if (err == 0)
2614 return 0;
2615 goto nla_put_failure;
2616 } else {
2617 if (err == -EMSGSIZE)
2618 goto nla_put_failure;
2619 }
2620 }
2621 } else
2622#endif
c78679e8
DM		 2623 if (nla_put_u32(skb, RTA_IIF, iif))
2624 goto nla_put_failure;
7bc570c8 2625 } else if (dst) {
1da177e4 2626 struct in6_addr saddr_buf;
c78679e8
DM		 2627 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2628 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2629 goto nla_put_failure;
1da177e4 2630 }
2d7202bf 2631
c3968a85
DW		 2632 if (rt->rt6i_prefsrc.plen) {
2633 struct in6_addr saddr_buf;
4e3fd7a0 2634 saddr_buf = rt->rt6i_prefsrc.addr;
c78679e8
DM		 2635 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2636 goto nla_put_failure;
c3968a85
DW		 2637 }
2638
defb3519 2639 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2d7202bf
TG		 2640 goto nla_put_failure;
2641
dd0cbf29
YH		 2642 if (rt->rt6i_flags & RTF_GATEWAY) {
2643 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
94f826b8 2644 goto nla_put_failure;
94f826b8 2645 }
2d7202bf 2646
c78679e8
DM		 2647 if (rt->dst.dev &&
2648 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2649 goto nla_put_failure;
2650 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2651 goto nla_put_failure;
8253947e
LW		 2652
2653 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2654
87a50699 2655 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2656 goto nla_put_failure;
2d7202bf
TG		 2657
2658 return nlmsg_end(skb, nlh);
2659
2660nla_put_failure:
26932566
PM		 2661 nlmsg_cancel(skb, nlh);
2662 return -EMSGSIZE;
1da177e4
LT		 2663}
2664
1b43af54 2665int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2666{
2667 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2668 int prefix;
2669
2d7202bf
TG		 2670 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2671 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2672 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2673 } else
2674 prefix = 0;
2675
191cd582
BH		 2676 return rt6_fill_node(arg->net,
2677 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2678 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2679 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2680}
2681
661d2967 2682static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh)
1da177e4 2683{
3b1e0a65 2684 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2685 struct nlattr *tb[RTA_MAX+1];
2686 struct rt6_info *rt;
1da177e4 2687 struct sk_buff *skb;
ab364a6f 2688 struct rtmsg *rtm;
4c9483b2 2689 struct flowi6 fl6;
72331bc0 2690 int err, iif = 0, oif = 0;
1da177e4 2691
ab364a6f
TG		 2692 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2693 if (err < 0)
2694 goto errout;
1da177e4 2695
ab364a6f 2696 err = -EINVAL;
4c9483b2 2697 memset(&fl6, 0, sizeof(fl6));
1da177e4 2698
ab364a6f
TG		 2699 if (tb[RTA_SRC]) {
2700 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2701 goto errout;
2702
4e3fd7a0 2703 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2704 }
2705
2706 if (tb[RTA_DST]) {
2707 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2708 goto errout;
2709
4e3fd7a0 2710 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2711 }
2712
2713 if (tb[RTA_IIF])
2714 iif = nla_get_u32(tb[RTA_IIF]);
2715
2716 if (tb[RTA_OIF])
72331bc0 2717 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4
LT		 2718
2719 if (iif) {
2720 struct net_device *dev;
72331bc0
SL		 2721 int flags = 0;
2722
5578689a 2723 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2724 if (!dev) {
2725 err = -ENODEV;
ab364a6f 2726 goto errout;
1da177e4 2727 }
72331bc0
SL		 2728
2729 fl6.flowi6_iif = iif;
2730
2731 if (!ipv6_addr_any(&fl6.saddr))
2732 flags |= RT6_LOOKUP_F_HAS_SADDR;
2733
2734 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2735 flags);
2736 } else {
2737 fl6.flowi6_oif = oif;
2738
2739 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2740 }
2741
ab364a6f 2742 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2743 if (!skb) {
94e187c0 2744 ip6_rt_put(rt);
ab364a6f
TG		 2745 err = -ENOBUFS;
2746 goto errout;
2747 }
1da177e4 2748
ab364a6f
TG		 2749 /* Reserve room for dummy headers, this skb can pass
2750 through good chunk of routing engine.
2751 */
459a98ed 2752 skb_reset_mac_header(skb);
ab364a6f 2753 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2754
d8d1f30b 2755 skb_dst_set(skb, &rt->dst);
1da177e4 2756
4c9483b2 2757 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2758 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2759 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2760 if (err < 0) {
ab364a6f
TG		 2761 kfree_skb(skb);
2762 goto errout;
1da177e4
LT		 2763 }
2764
15e47304 2765 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2766errout:
1da177e4 2767 return err;
1da177e4
LT		 2768}
2769
86872cb5 2770void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2771{
2772 struct sk_buff *skb;
5578689a 2773 struct net *net = info->nl_net;
528c4ceb
DL		 2774 u32 seq;
2775 int err;
2776
2777 err = -ENOBUFS;
38308473 2778 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2779
339bf98f 2780 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2781 if (!skb)
21713ebc
TG		 2782 goto errout;
2783
191cd582 2784 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2785 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2786 if (err < 0) {
2787 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2788 WARN_ON(err == -EMSGSIZE);
2789 kfree_skb(skb);
2790 goto errout;
2791 }
15e47304 2792 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2793 info->nlh, gfp_any());
2794 return;
21713ebc
TG		 2795errout:
2796 if (err < 0)
5578689a 2797 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2798}
2799
8ed67789 2800static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2801 unsigned long event, void *ptr)
8ed67789 2802{
351638e7 2803 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2804 struct net *net = dev_net(dev);
8ed67789
DL		 2805
2806 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2807 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2808 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2809#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2810 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2811 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2812 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2813 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2814#endif
2815 }
2816
2817 return NOTIFY_OK;
2818}
2819
1da177e4
LT		 2820/*
2821 * /proc
2822 */
2823
2824#ifdef CONFIG_PROC_FS
2825
33120b30
AD		 2826static const struct file_operations ipv6_route_proc_fops = {
2827 .owner = THIS_MODULE,
2828 .open = ipv6_route_open,
2829 .read = seq_read,
2830 .llseek = seq_lseek,
8d2ca1d7 2831 .release = seq_release_net,
33120b30
AD		 2832};
2833
1da177e4
LT		 2834static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2835{
69ddb805 2836 struct net *net = (struct net *)seq->private;
1da177e4 2837 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2838 net->ipv6.rt6_stats->fib_nodes,
2839 net->ipv6.rt6_stats->fib_route_nodes,
2840 net->ipv6.rt6_stats->fib_rt_alloc,
2841 net->ipv6.rt6_stats->fib_rt_entries,
2842 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2843 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2844 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2845
2846 return 0;
2847}
2848
2849static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2850{
de05c557 2851 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2852}
2853
9a32144e 2854static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2855 .owner = THIS_MODULE,
2856 .open = rt6_stats_seq_open,
2857 .read = seq_read,
2858 .llseek = seq_lseek,
b6fcbdb4 2859 .release = single_release_net,
1da177e4
LT		 2860};
2861#endif /* CONFIG_PROC_FS */
2862
2863#ifdef CONFIG_SYSCTL
2864
1da177e4 2865static
fe2c6338 2866int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 2867 void __user *buffer, size_t *lenp, loff_t *ppos)
2868{
c486da34
LAG		 2869 struct net *net;
2870 int delay;
2871 if (!write)
1da177e4 2872 return -EINVAL;
c486da34
LAG		 2873
2874 net = (struct net *)ctl->extra1;
2875 delay = net->ipv6.sysctl.flush_delay;
2876 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 2877 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 2878 return 0;
1da177e4
LT		 2879}
2880
fe2c6338 2881struct ctl_table ipv6_route_table_template[] = {
1ab1457c 2882 {
1da177e4 2883 .procname = "flush",
4990509f 2884 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2885 .maxlen = sizeof(int),
89c8b3a1 2886 .mode = 0200,
6d9f239a 2887 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 2888 },
2889 {
1da177e4 2890 .procname = "gc_thresh",
9a7ec3a9 2891 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2892 .maxlen = sizeof(int),
2893 .mode = 0644,
6d9f239a 2894 .proc_handler = proc_dointvec,
1da177e4
LT		 2895 },
2896 {
1da177e4 2897 .procname = "max_size",
4990509f 2898 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2899 .maxlen = sizeof(int),
2900 .mode = 0644,
6d9f239a 2901 .proc_handler = proc_dointvec,
1da177e4
LT		 2902 },
2903 {
1da177e4 2904 .procname = "gc_min_interval",
4990509f 2905 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2906 .maxlen = sizeof(int),
2907 .mode = 0644,
6d9f239a 2908 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2909 },
2910 {
1da177e4 2911 .procname = "gc_timeout",
4990509f 2912 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2913 .maxlen = sizeof(int),
2914 .mode = 0644,
6d9f239a 2915 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2916 },
2917 {
1da177e4 2918 .procname = "gc_interval",
4990509f 2919 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2920 .maxlen = sizeof(int),
2921 .mode = 0644,
6d9f239a 2922 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2923 },
2924 {
1da177e4 2925 .procname = "gc_elasticity",
4990509f 2926 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 2927 .maxlen = sizeof(int),
2928 .mode = 0644,
f3d3f616 2929 .proc_handler = proc_dointvec,
1da177e4
LT		 2930 },
2931 {
1da177e4 2932 .procname = "mtu_expires",
4990509f 2933 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 2934 .maxlen = sizeof(int),
2935 .mode = 0644,
6d9f239a 2936 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2937 },
2938 {
1da177e4 2939 .procname = "min_adv_mss",
4990509f 2940 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 2941 .maxlen = sizeof(int),
2942 .mode = 0644,
f3d3f616 2943 .proc_handler = proc_dointvec,
1da177e4
LT		 2944 },
2945 {
1da177e4 2946 .procname = "gc_min_interval_ms",
4990509f 2947 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2948 .maxlen = sizeof(int),
2949 .mode = 0644,
6d9f239a 2950 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 2951 },
f8572d8f 2952 { }
1da177e4
LT		 2953};
2954
2c8c1e72 2955struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 2956{
2957 struct ctl_table *table;
2958
2959 table = kmemdup(ipv6_route_table_template,
2960 sizeof(ipv6_route_table_template),
2961 GFP_KERNEL);
5ee09105
YH		 2962
2963 if (table) {
2964 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 2965 table[0].extra1 = net;
86393e52 2966 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 2967 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2968 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2969 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2970 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2971 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2972 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2973 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 2974 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 2975
2976 /* Don't export sysctls to unprivileged users */
2977 if (net->user_ns != &init_user_ns)
2978 table[0].procname = NULL;
5ee09105
YH		 2979 }
2980
760f2d01
DL		 2981 return table;
2982}
1da177e4
LT		 2983#endif
2984
2c8c1e72 2985static int __net_init ip6_route_net_init(struct net *net)
cdb18761 2986{
633d424b 2987 int ret = -ENOMEM;
8ed67789 2988
86393e52
AD		 2989 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
2990 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 2991
fc66f95c
ED		 2992 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
2993 goto out_ip6_dst_ops;
2994
8ed67789
DL		 2995 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
2996 sizeof(*net->ipv6.ip6_null_entry),
2997 GFP_KERNEL);
2998 if (!net->ipv6.ip6_null_entry)
fc66f95c 2999 goto out_ip6_dst_entries;
d8d1f30b 3000 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3001 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3002 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3003 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3004 ip6_template_metrics, true);
8ed67789
DL		 3005
3006#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3007 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3008 sizeof(*net->ipv6.ip6_prohibit_entry),
3009 GFP_KERNEL);
68fffc67
PZ		 3010 if (!net->ipv6.ip6_prohibit_entry)
3011 goto out_ip6_null_entry;
d8d1f30b 3012 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3013 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3014 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3015 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3016 ip6_template_metrics, true);
8ed67789
DL		 3017
3018 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3019 sizeof(*net->ipv6.ip6_blk_hole_entry),
3020 GFP_KERNEL);
68fffc67
PZ		 3021 if (!net->ipv6.ip6_blk_hole_entry)
3022 goto out_ip6_prohibit_entry;
d8d1f30b 3023 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3024 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3025 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3026 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3027 ip6_template_metrics, true);
8ed67789
DL		 3028#endif
3029
b339a47c
PZ		 3030 net->ipv6.sysctl.flush_delay = 0;
3031 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3032 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3033 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3034 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3035 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3036 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3037 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3038
6891a346
BT		 3039 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3040
8ed67789
DL		 3041 ret = 0;
3042out:
3043 return ret;
f2fc6a54 3044
68fffc67
PZ		 3045#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3046out_ip6_prohibit_entry:
3047 kfree(net->ipv6.ip6_prohibit_entry);
3048out_ip6_null_entry:
3049 kfree(net->ipv6.ip6_null_entry);
3050#endif
fc66f95c
ED		 3051out_ip6_dst_entries:
3052 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3053out_ip6_dst_ops:
f2fc6a54 3054 goto out;
cdb18761
DL		 3055}
3056
2c8c1e72 3057static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3058{
8ed67789
DL		 3059 kfree(net->ipv6.ip6_null_entry);
3060#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3061 kfree(net->ipv6.ip6_prohibit_entry);
3062 kfree(net->ipv6.ip6_blk_hole_entry);
3063#endif
41bb78b4 3064 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3065}
3066
d189634e
TG		 3067static int __net_init ip6_route_net_init_late(struct net *net)
3068{
3069#ifdef CONFIG_PROC_FS
d4beaa66
G		 3070 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3071 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3072#endif
3073 return 0;
3074}
3075
3076static void __net_exit ip6_route_net_exit_late(struct net *net)
3077{
3078#ifdef CONFIG_PROC_FS
ece31ffd
G		 3079 remove_proc_entry("ipv6_route", net->proc_net);
3080 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3081#endif
3082}
3083
cdb18761
DL		 3084static struct pernet_operations ip6_route_net_ops = {
3085 .init = ip6_route_net_init,
3086 .exit = ip6_route_net_exit,
3087};
3088
c3426b47
DM		 3089static int __net_init ipv6_inetpeer_init(struct net *net)
3090{
3091 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3092
3093 if (!bp)
3094 return -ENOMEM;
3095 inet_peer_base_init(bp);
3096 net->ipv6.peers = bp;
3097 return 0;
3098}
3099
3100static void __net_exit ipv6_inetpeer_exit(struct net *net)
3101{
3102 struct inet_peer_base *bp = net->ipv6.peers;
3103
3104 net->ipv6.peers = NULL;
56a6b248 3105 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3106 kfree(bp);
3107}
3108
2b823f72 3109static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3110 .init = ipv6_inetpeer_init,
3111 .exit = ipv6_inetpeer_exit,
3112};
3113
d189634e
TG		 3114static struct pernet_operations ip6_route_net_late_ops = {
3115 .init = ip6_route_net_init_late,
3116 .exit = ip6_route_net_exit_late,
3117};
3118
8ed67789
DL		 3119static struct notifier_block ip6_route_dev_notifier = {
3120 .notifier_call = ip6_route_dev_notify,
3121 .priority = 0,
3122};
3123
433d49c3 3124int __init ip6_route_init(void)
1da177e4 3125{
433d49c3
DL		 3126 int ret;
3127
9a7ec3a9
DL		 3128 ret = -ENOMEM;
3129 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3130 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3131 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3132 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3133 goto out;
14e50e57 3134
fc66f95c 3135 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3136 if (ret)
bdb3289f 3137 goto out_kmem_cache;
bdb3289f 3138
c3426b47
DM		 3139 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3140 if (ret)
e8803b6c 3141 goto out_dst_entries;
2a0c451a 3142
7e52b33b
DM		 3143 ret = register_pernet_subsys(&ip6_route_net_ops);
3144 if (ret)
3145 goto out_register_inetpeer;
c3426b47 3146
5dc121e9
AE		 3147 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3148
8ed67789
DL		 3149 /* Registering of the loopback is done before this portion of code,
3150 * the loopback reference in rt6_info will not be taken, do it
3151 * manually for init_net */
d8d1f30b 3152 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3153 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3154 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3155 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3156 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3157 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3158 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3159 #endif
e8803b6c 3160 ret = fib6_init();
433d49c3 3161 if (ret)
8ed67789 3162 goto out_register_subsys;
433d49c3 3163
433d49c3
DL		 3164 ret = xfrm6_init();
3165 if (ret)
e8803b6c 3166 goto out_fib6_init;
c35b7e72 3167
433d49c3
DL		 3168 ret = fib6_rules_init();
3169 if (ret)
3170 goto xfrm6_init;
7e5449c2 3171
d189634e
TG		 3172 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3173 if (ret)
3174 goto fib6_rules_init;
3175
433d49c3 3176 ret = -ENOBUFS;
c7ac8679
GR		 3177 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3178 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3179 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3180 goto out_register_late_subsys;
c127ea2c 3181
8ed67789 3182 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3183 if (ret)
d189634e 3184 goto out_register_late_subsys;
8ed67789 3185
433d49c3
DL		 3186out:
3187 return ret;
3188
d189634e
TG		 3189out_register_late_subsys:
3190 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3191fib6_rules_init:
433d49c3
DL		 3192 fib6_rules_cleanup();
3193xfrm6_init:
433d49c3 3194 xfrm6_fini();
2a0c451a
TG		 3195out_fib6_init:
3196 fib6_gc_cleanup();
8ed67789
DL		 3197out_register_subsys:
3198 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3199out_register_inetpeer:
3200 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3201out_dst_entries:
3202 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3203out_kmem_cache:
f2fc6a54 3204 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3205 goto out;
1da177e4
LT		 3206}
3207
3208void ip6_route_cleanup(void)
3209{
8ed67789 3210 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3211 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3212 fib6_rules_cleanup();
1da177e4 3213 xfrm6_fini();
1da177e4 3214 fib6_gc_cleanup();
c3426b47 3215 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3216 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3217 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3218 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3219}
Linux kernel - Deliverables
This page took 1.748566 seconds and 5 git commands to generate.