projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
drm/nouveau/core: Move event index check from critical section
[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);
87static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 88static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
89 struct sk_buff *skb, u32 mtu);
90static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
91 struct sk_buff *skb);
52bd4c0c 92static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 93
70ceb4f5 94#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 95static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 96 const struct in6_addr *prefix, int prefixlen,
97 const struct in6_addr *gwaddr, int ifindex,
95c96174 98 unsigned int pref);
efa2cea0 99static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 100 const struct in6_addr *prefix, int prefixlen,
101 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 102#endif
103
06582540
DM		 104static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
105{
106 struct rt6_info *rt = (struct rt6_info *) dst;
107 struct inet_peer *peer;
108 u32 *p = NULL;
109
8e2ec639
YZ		 110 if (!(rt->dst.flags & DST_HOST))
111 return NULL;
112
fbfe95a4 113 peer = rt6_get_peer_create(rt);
06582540
DM		 114 if (peer) {
115 u32 *old_p = __DST_METRICS_PTR(old);
116 unsigned long prev, new;
117
118 p = peer->metrics;
119 if (inet_metrics_new(peer))
120 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
121
122 new = (unsigned long) p;
123 prev = cmpxchg(&dst->_metrics, old, new);
124
125 if (prev != old) {
126 p = __DST_METRICS_PTR(prev);
127 if (prev & DST_METRICS_READ_ONLY)
128 p = NULL;
129 }
130 }
131 return p;
132}
133
f894cbf8
DM		 134static inline const void *choose_neigh_daddr(struct rt6_info *rt,
135 struct sk_buff *skb,
136 const void *daddr)
39232973
DM		 137{
138 struct in6_addr *p = &rt->rt6i_gateway;
139
a7563f34 140 if (!ipv6_addr_any(p))
39232973 141 return (const void *) p;
f894cbf8
DM		 142 else if (skb)
143 return &ipv6_hdr(skb)->daddr;
39232973
DM		 144 return daddr;
145}
146
f894cbf8
DM		 147static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
148 struct sk_buff *skb,
149 const void *daddr)
d3aaeb38 150{
39232973
DM		 151 struct rt6_info *rt = (struct rt6_info *) dst;
152 struct neighbour *n;
153
f894cbf8 154 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 155 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 156 if (n)
157 return n;
158 return neigh_create(&nd_tbl, daddr, dst->dev);
159}
160
9a7ec3a9 161static struct dst_ops ip6_dst_ops_template = {
1da177e4 162 .family = AF_INET6,
09640e63 163 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT		 164 .gc = ip6_dst_gc,
165 .gc_thresh = 1024,
166 .check = ip6_dst_check,
0dbaee3b 167 .default_advmss = ip6_default_advmss,
ebb762f2 168 .mtu = ip6_mtu,
06582540 169 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 170 .destroy = ip6_dst_destroy,
171 .ifdown = ip6_dst_ifdown,
172 .negative_advice = ip6_negative_advice,
173 .link_failure = ip6_link_failure,
174 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 175 .redirect = rt6_do_redirect,
1ac06e03 176 .local_out = __ip6_local_out,
d3aaeb38 177 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 178};
179
ebb762f2 180static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 181{
618f9bc7
SK		 182 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
183
184 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 185}
186
6700c270
DM		 187static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
188 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 189{
190}
191
6700c270
DM		 192static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
193 struct sk_buff *skb)
b587ee3b
DM		 194{
195}
196
0972ddb2
HB		 197static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
198 unsigned long old)
199{
200 return NULL;
201}
202
14e50e57
DM		 203static struct dst_ops ip6_dst_blackhole_ops = {
204 .family = AF_INET6,
09640e63 205 .protocol = cpu_to_be16(ETH_P_IPV6),
14e50e57
DM		 206 .destroy = ip6_dst_destroy,
207 .check = ip6_dst_check,
ebb762f2 208 .mtu = ip6_blackhole_mtu,
214f45c9 209 .default_advmss = ip6_default_advmss,
14e50e57 210 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 211 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 212 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 213 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 214};
215
62fa8a84 216static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 217 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 218};
219
fb0af4c7 220static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 221 .dst = {
222 .__refcnt = ATOMIC_INIT(1),
223 .__use = 1,
2c20cbd7 224 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 225 .error = -ENETUNREACH,
d8d1f30b
CG		 226 .input = ip6_pkt_discard,
227 .output = ip6_pkt_discard_out,
1da177e4
LT		 228 },
229 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 230 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 231 .rt6i_metric = ~(u32) 0,
232 .rt6i_ref = ATOMIC_INIT(1),
233};
234
101367c2
TG		 235#ifdef CONFIG_IPV6_MULTIPLE_TABLES
236
6723ab54
DM		 237static int ip6_pkt_prohibit(struct sk_buff *skb);
238static int ip6_pkt_prohibit_out(struct sk_buff *skb);
6723ab54 239
fb0af4c7 240static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 241 .dst = {
242 .__refcnt = ATOMIC_INIT(1),
243 .__use = 1,
2c20cbd7 244 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 245 .error = -EACCES,
d8d1f30b
CG		 246 .input = ip6_pkt_prohibit,
247 .output = ip6_pkt_prohibit_out,
101367c2
TG		 248 },
249 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 250 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 251 .rt6i_metric = ~(u32) 0,
252 .rt6i_ref = ATOMIC_INIT(1),
253};
254
fb0af4c7 255static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 256 .dst = {
257 .__refcnt = ATOMIC_INIT(1),
258 .__use = 1,
2c20cbd7 259 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 260 .error = -EINVAL,
d8d1f30b
CG		 261 .input = dst_discard,
262 .output = dst_discard,
101367c2
TG		 263 },
264 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 265 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 266 .rt6i_metric = ~(u32) 0,
267 .rt6i_ref = ATOMIC_INIT(1),
268};
269
270#endif
271
1da177e4 272/* allocate dst with ip6_dst_ops */
97bab73f 273static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 274 struct net_device *dev,
8b96d22d
DM		 275 int flags,
276 struct fib6_table *table)
1da177e4 277{
97bab73f 278 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 279 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 280
97bab73f 281 if (rt) {
8104891b
SK		 282 struct dst_entry *dst = &rt->dst;
283
284 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
8b96d22d 285 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
ca4c3fc2 286 rt->rt6i_genid = rt_genid_ipv6(net);
51ebd318 287 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 288 }
cf911662 289 return rt;
1da177e4
LT		 290}
291
292static void ip6_dst_destroy(struct dst_entry *dst)
293{
294 struct rt6_info *rt = (struct rt6_info *)dst;
295 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 296 struct dst_entry *from = dst->from;
1da177e4 297
8e2ec639
YZ		 298 if (!(rt->dst.flags & DST_HOST))
299 dst_destroy_metrics_generic(dst);
300
38308473 301 if (idev) {
1da177e4
LT		 302 rt->rt6i_idev = NULL;
303 in6_dev_put(idev);
1ab1457c 304 }
1716a961 305
ecd98837
YH		 306 dst->from = NULL;
307 dst_release(from);
1716a961 308
97bab73f
DM		 309 if (rt6_has_peer(rt)) {
310 struct inet_peer *peer = rt6_peer_ptr(rt);
b3419363
DM		 311 inet_putpeer(peer);
312 }
313}
314
315void rt6_bind_peer(struct rt6_info *rt, int create)
316{
97bab73f 317 struct inet_peer_base *base;
b3419363
DM		 318 struct inet_peer *peer;
319
97bab73f
DM		 320 base = inetpeer_base_ptr(rt->_rt6i_peer);
321 if (!base)
322 return;
323
324 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create);
7b34ca2a
DM		 325 if (peer) {
326 if (!rt6_set_peer(rt, peer))
327 inet_putpeer(peer);
7b34ca2a 328 }
1da177e4
LT		 329}
330
331static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
332 int how)
333{
334 struct rt6_info *rt = (struct rt6_info *)dst;
335 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 336 struct net_device *loopback_dev =
c346dca1 337 dev_net(dev)->loopback_dev;
1da177e4 338
97cac082
DM		 339 if (dev != loopback_dev) {
340 if (idev && idev->dev == dev) {
341 struct inet6_dev *loopback_idev =
342 in6_dev_get(loopback_dev);
343 if (loopback_idev) {
344 rt->rt6i_idev = loopback_idev;
345 in6_dev_put(idev);
346 }
347 }
1da177e4
LT		 348 }
349}
350
a50feda5 351static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 352{
1716a961
G		 353 if (rt->rt6i_flags & RTF_EXPIRES) {
354 if (time_after(jiffies, rt->dst.expires))
a50feda5 355 return true;
1716a961 356 } else if (rt->dst.from) {
3fd91fb3 357 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 358 }
a50feda5 359 return false;
1da177e4
LT		 360}
361
a50feda5 362static bool rt6_need_strict(const struct in6_addr *daddr)
c71099ac 363{
a02cec21
ED		 364 return ipv6_addr_type(daddr) &
365 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
c71099ac
TG		 366}
367
51ebd318
ND		 368/* Multipath route selection:
369 * Hash based function using packet header and flowlabel.
370 * Adapted from fib_info_hashfn()
371 */
372static int rt6_info_hash_nhsfn(unsigned int candidate_count,
373 const struct flowi6 *fl6)
374{
375 unsigned int val = fl6->flowi6_proto;
376
c08977bb
YH		 377 val ^= ipv6_addr_hash(&fl6->daddr);
378 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 379
380 /* Work only if this not encapsulated */
381 switch (fl6->flowi6_proto) {
382 case IPPROTO_UDP:
383 case IPPROTO_TCP:
384 case IPPROTO_SCTP:
b3ce5ae1
ND		 385 val ^= (__force u16)fl6->fl6_sport;
386 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 387 break;
388
389 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 390 val ^= (__force u16)fl6->fl6_icmp_type;
391 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 392 break;
393 }
394 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 395 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 396
397 /* Perhaps, we need to tune, this function? */
398 val = val ^ (val >> 7) ^ (val >> 12);
399 return val % candidate_count;
400}
401
402static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 403 struct flowi6 *fl6, int oif,
404 int strict)
51ebd318
ND		 405{
406 struct rt6_info *sibling, *next_sibling;
407 int route_choosen;
408
409 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
410 /* Don't change the route, if route_choosen == 0
411 * (siblings does not include ourself)
412 */
413 if (route_choosen)
414 list_for_each_entry_safe(sibling, next_sibling,
415 &match->rt6i_siblings, rt6i_siblings) {
416 route_choosen--;
417 if (route_choosen == 0) {
52bd4c0c
ND		 418 if (rt6_score_route(sibling, oif, strict) < 0)
419 break;
51ebd318
ND		 420 match = sibling;
421 break;
422 }
423 }
424 return match;
425}
426
1da177e4 427/*
c71099ac 428 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 429 */
430
8ed67789
DL		 431static inline struct rt6_info *rt6_device_match(struct net *net,
432 struct rt6_info *rt,
b71d1d42 433 const struct in6_addr *saddr,
1da177e4 434 int oif,
d420895e 435 int flags)
1da177e4
LT		 436{
437 struct rt6_info *local = NULL;
438 struct rt6_info *sprt;
439
dd3abc4e
YH		 440 if (!oif && ipv6_addr_any(saddr))
441 goto out;
442
d8d1f30b 443 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 444 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 445
446 if (oif) {
1da177e4
LT		 447 if (dev->ifindex == oif)
448 return sprt;
449 if (dev->flags & IFF_LOOPBACK) {
38308473 450 if (!sprt->rt6i_idev ||
1da177e4 451 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 452 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 453 continue;
1ab1457c 454 if (local && (!oif ||
1da177e4
LT		 455 local->rt6i_idev->dev->ifindex == oif))
456 continue;
457 }
458 local = sprt;
459 }
dd3abc4e
YH		 460 } else {
461 if (ipv6_chk_addr(net, saddr, dev,
462 flags & RT6_LOOKUP_F_IFACE))
463 return sprt;
1da177e4 464 }
dd3abc4e 465 }
1da177e4 466
dd3abc4e 467 if (oif) {
1da177e4
LT		 468 if (local)
469 return local;
470
d420895e 471 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 472 return net->ipv6.ip6_null_entry;
1da177e4 473 }
dd3abc4e 474out:
1da177e4
LT		 475 return rt;
476}
477
27097255
YH		 478#ifdef CONFIG_IPV6_ROUTER_PREF
479static void rt6_probe(struct rt6_info *rt)
480{
f2c31e32 481 struct neighbour *neigh;
27097255
YH		 482 /*
483 * Okay, this does not seem to be appropriate
484 * for now, however, we need to check if it
485 * is really so; aka Router Reachability Probing.
486 *
487 * Router Reachability Probe MUST be rate-limited
488 * to no more than one per minute.
489 */
2152caea 490 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 491 return;
2152caea
YH		 492 rcu_read_lock_bh();
493 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
494 if (neigh) {
495 write_lock(&neigh->lock);
496 if (neigh->nud_state & NUD_VALID)
497 goto out;
7ff74a59 498 }
2152caea
YH		 499
500 if (!neigh ||
52e16356 501 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
27097255
YH		 502 struct in6_addr mcaddr;
503 struct in6_addr *target;
504
b820bb6b
YH		 505 if (neigh) {
506 neigh->updated = jiffies;
2152caea 507 write_unlock(&neigh->lock);
b820bb6b 508 }
2152caea
YH		 509
510 target = (struct in6_addr *)&rt->rt6i_gateway;
27097255 511 addrconf_addr_solict_mult(target, &mcaddr);
d1918542 512 ndisc_send_ns(rt->dst.dev, NULL, target, &mcaddr, NULL);
f2c31e32 513 } else {
2152caea
YH		 514out:
515 write_unlock(&neigh->lock);
f2c31e32 516 }
2152caea 517 rcu_read_unlock_bh();
27097255
YH		 518}
519#else
520static inline void rt6_probe(struct rt6_info *rt)
521{
27097255
YH		 522}
523#endif
524
1da177e4 525/*
554cfb7e 526 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 527 */
b6f99a21 528static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 529{
d1918542 530 struct net_device *dev = rt->dst.dev;
161980f4 531 if (!oif || dev->ifindex == oif)
554cfb7e 532 return 2;
161980f4
DM		 533 if ((dev->flags & IFF_LOOPBACK) &&
534 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
535 return 1;
536 return 0;
554cfb7e 537}
1da177e4 538
afc154e9 539static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 540{
f2c31e32 541 struct neighbour *neigh;
afc154e9 542 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 543
4d0c5911
YH		 544 if (rt->rt6i_flags & RTF_NONEXTHOP ||
545 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 546 return RT6_NUD_SUCCEED;
145a3621
YH		 547
548 rcu_read_lock_bh();
549 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
550 if (neigh) {
551 read_lock(&neigh->lock);
554cfb7e 552 if (neigh->nud_state & NUD_VALID)
afc154e9 553 ret = RT6_NUD_SUCCEED;
398bcbeb 554#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 555 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 556 ret = RT6_NUD_SUCCEED;
398bcbeb 557#endif
145a3621 558 read_unlock(&neigh->lock);
afc154e9
HFS		 559 } else {
560 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
561 RT6_NUD_SUCCEED : RT6_NUD_FAIL_SOFT;
a5a81f0b 562 }
145a3621
YH		 563 rcu_read_unlock_bh();
564
a5a81f0b 565 return ret;
1da177e4
LT		 566}
567
554cfb7e
YH		 568static int rt6_score_route(struct rt6_info *rt, int oif,
569 int strict)
1da177e4 570{
a5a81f0b 571 int m;
1ab1457c 572
4d0c5911 573 m = rt6_check_dev(rt, oif);
77d16f45 574 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 575 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 576#ifdef CONFIG_IPV6_ROUTER_PREF
577 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
578#endif
afc154e9
HFS		 579 if (strict & RT6_LOOKUP_F_REACHABLE) {
580 int n = rt6_check_neigh(rt);
581 if (n < 0)
582 return n;
583 }
554cfb7e
YH		 584 return m;
585}
586
f11e6659 587static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 588 int *mpri, struct rt6_info *match,
589 bool *do_rr)
554cfb7e 590{
f11e6659 591 int m;
afc154e9 592 bool match_do_rr = false;
f11e6659
DM		 593
594 if (rt6_check_expired(rt))
595 goto out;
596
597 m = rt6_score_route(rt, oif, strict);
afc154e9
HFS		 598 if (m == RT6_NUD_FAIL_SOFT && !IS_ENABLED(CONFIG_IPV6_ROUTER_PREF)) {
599 match_do_rr = true;
600 m = 0; /* lowest valid score */
601 } else if (m < 0) {
f11e6659 602 goto out;
afc154e9
HFS		 603 }
604
605 if (strict & RT6_LOOKUP_F_REACHABLE)
606 rt6_probe(rt);
f11e6659
DM		 607
608 if (m > *mpri) {
afc154e9 609 *do_rr = match_do_rr;
f11e6659
DM		 610 *mpri = m;
611 match = rt;
f11e6659 612 }
f11e6659
DM		 613out:
614 return match;
615}
616
617static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
618 struct rt6_info *rr_head,
afc154e9
HFS		 619 u32 metric, int oif, int strict,
620 bool *do_rr)
f11e6659
DM		 621{
622 struct rt6_info *rt, *match;
554cfb7e 623 int mpri = -1;
1da177e4 624
f11e6659
DM		 625 match = NULL;
626 for (rt = rr_head; rt && rt->rt6i_metric == metric;
d8d1f30b 627 rt = rt->dst.rt6_next)
afc154e9 628 match = find_match(rt, oif, strict, &mpri, match, do_rr);
f11e6659 629 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
d8d1f30b 630 rt = rt->dst.rt6_next)
afc154e9 631 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 632
f11e6659
DM		 633 return match;
634}
1da177e4 635
f11e6659
DM		 636static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
637{
638 struct rt6_info *match, *rt0;
8ed67789 639 struct net *net;
afc154e9 640 bool do_rr = false;
1da177e4 641
f11e6659
DM		 642 rt0 = fn->rr_ptr;
643 if (!rt0)
644 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 645
afc154e9
HFS		 646 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
647 &do_rr);
1da177e4 648
afc154e9 649 if (do_rr) {
d8d1f30b 650 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 651
554cfb7e 652 /* no entries matched; do round-robin */
f11e6659
DM		 653 if (!next || next->rt6i_metric != rt0->rt6i_metric)
654 next = fn->leaf;
655
656 if (next != rt0)
657 fn->rr_ptr = next;
1da177e4 658 }
1da177e4 659
d1918542 660 net = dev_net(rt0->dst.dev);
a02cec21 661 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 662}
663
70ceb4f5
YH		 664#ifdef CONFIG_IPV6_ROUTE_INFO
665int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 666 const struct in6_addr *gwaddr)
70ceb4f5 667{
c346dca1 668 struct net *net = dev_net(dev);
70ceb4f5
YH		 669 struct route_info *rinfo = (struct route_info *) opt;
670 struct in6_addr prefix_buf, *prefix;
671 unsigned int pref;
4bed72e4 672 unsigned long lifetime;
70ceb4f5
YH		 673 struct rt6_info *rt;
674
675 if (len < sizeof(struct route_info)) {
676 return -EINVAL;
677 }
678
679 /* Sanity check for prefix_len and length */
680 if (rinfo->length > 3) {
681 return -EINVAL;
682 } else if (rinfo->prefix_len > 128) {
683 return -EINVAL;
684 } else if (rinfo->prefix_len > 64) {
685 if (rinfo->length < 2) {
686 return -EINVAL;
687 }
688 } else if (rinfo->prefix_len > 0) {
689 if (rinfo->length < 1) {
690 return -EINVAL;
691 }
692 }
693
694 pref = rinfo->route_pref;
695 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 696 return -EINVAL;
70ceb4f5 697
4bed72e4 698 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 699
700 if (rinfo->length == 3)
701 prefix = (struct in6_addr *)rinfo->prefix;
702 else {
703 /* this function is safe */
704 ipv6_addr_prefix(&prefix_buf,
705 (struct in6_addr *)rinfo->prefix,
706 rinfo->prefix_len);
707 prefix = &prefix_buf;
708 }
709
efa2cea0
DL		 710 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr,
711 dev->ifindex);
70ceb4f5
YH		 712
713 if (rt && !lifetime) {
e0a1ad73 714 ip6_del_rt(rt);
70ceb4f5
YH		 715 rt = NULL;
716 }
717
718 if (!rt && lifetime)
efa2cea0 719 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 720 pref);
721 else if (rt)
722 rt->rt6i_flags = RTF_ROUTEINFO |
723 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
724
725 if (rt) {
1716a961
G		 726 if (!addrconf_finite_timeout(lifetime))
727 rt6_clean_expires(rt);
728 else
729 rt6_set_expires(rt, jiffies + HZ * lifetime);
730
94e187c0 731 ip6_rt_put(rt);
70ceb4f5
YH		 732 }
733 return 0;
734}
735#endif
736
8ed67789 737#define BACKTRACK(__net, saddr) \
982f56f3 738do { \
8ed67789 739 if (rt == __net->ipv6.ip6_null_entry) { \
982f56f3 740 struct fib6_node *pn; \
e0eda7bb 741 while (1) { \
982f56f3
YH		 742 if (fn->fn_flags & RTN_TL_ROOT) \
743 goto out; \
744 pn = fn->parent; \
745 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
8bce65b9 746 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
982f56f3
YH		 747 else \
748 fn = pn; \
749 if (fn->fn_flags & RTN_RTINFO) \
750 goto restart; \
c71099ac 751 } \
c71099ac 752 } \
38308473 753} while (0)
c71099ac 754
8ed67789
DL		 755static struct rt6_info *ip6_pol_route_lookup(struct net *net,
756 struct fib6_table *table,
4c9483b2 757 struct flowi6 *fl6, int flags)
1da177e4
LT		 758{
759 struct fib6_node *fn;
760 struct rt6_info *rt;
761
c71099ac 762 read_lock_bh(&table->tb6_lock);
4c9483b2 763 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 764restart:
765 rt = fn->leaf;
4c9483b2 766 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 767 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 768 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
4c9483b2 769 BACKTRACK(net, &fl6->saddr);
c71099ac 770out:
d8d1f30b 771 dst_use(&rt->dst, jiffies);
c71099ac 772 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 773 return rt;
774
775}
776
ea6e574e
FW		 777struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6,
778 int flags)
779{
780 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
781}
782EXPORT_SYMBOL_GPL(ip6_route_lookup);
783
9acd9f3a
YH		 784struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
785 const struct in6_addr *saddr, int oif, int strict)
c71099ac 786{
4c9483b2
DM		 787 struct flowi6 fl6 = {
788 .flowi6_oif = oif,
789 .daddr = *daddr,
c71099ac
TG		 790 };
791 struct dst_entry *dst;
77d16f45 792 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 793
adaa70bb 794 if (saddr) {
4c9483b2 795 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 796 flags |= RT6_LOOKUP_F_HAS_SADDR;
797 }
798
4c9483b2 799 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 800 if (dst->error == 0)
801 return (struct rt6_info *) dst;
802
803 dst_release(dst);
804
1da177e4
LT		 805 return NULL;
806}
807
7159039a
YH		 808EXPORT_SYMBOL(rt6_lookup);
809
c71099ac 810/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 811 It takes new route entry, the addition fails by any reason the
812 route is freed. In any case, if caller does not hold it, it may
813 be destroyed.
814 */
815
86872cb5 816static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 817{
818 int err;
c71099ac 819 struct fib6_table *table;
1da177e4 820
c71099ac
TG		 821 table = rt->rt6i_table;
822 write_lock_bh(&table->tb6_lock);
86872cb5 823 err = fib6_add(&table->tb6_root, rt, info);
c71099ac 824 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 825
826 return err;
827}
828
40e22e8f
TG		 829int ip6_ins_rt(struct rt6_info *rt)
830{
4d1169c1 831 struct nl_info info = {
d1918542 832 .nl_net = dev_net(rt->dst.dev),
4d1169c1 833 };
528c4ceb 834 return __ip6_ins_rt(rt, &info);
40e22e8f
TG		 835}
836
1716a961 837static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
21efcfa0 838 const struct in6_addr *daddr,
b71d1d42 839 const struct in6_addr *saddr)
1da177e4 840{
1da177e4
LT		 841 struct rt6_info *rt;
842
843 /*
844 * Clone the route.
845 */
846
21efcfa0 847 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 848
849 if (rt) {
38308473 850 if (!(rt->rt6i_flags & RTF_GATEWAY)) {
bb3c3686 851 if (ort->rt6i_dst.plen != 128 &&
21efcfa0 852 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
58c4fb86 853 rt->rt6i_flags |= RTF_ANYCAST;
4e3fd7a0 854 rt->rt6i_gateway = *daddr;
58c4fb86 855 }
1da177e4 856
1da177e4 857 rt->rt6i_flags |= RTF_CACHE;
1da177e4
LT		 858
859#ifdef CONFIG_IPV6_SUBTREES
860 if (rt->rt6i_src.plen && saddr) {
4e3fd7a0 861 rt->rt6i_src.addr = *saddr;
1da177e4
LT		 862 rt->rt6i_src.plen = 128;
863 }
864#endif
95a9a5ba 865 }
1da177e4 866
95a9a5ba
YH		 867 return rt;
868}
1da177e4 869
21efcfa0
ED		 870static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
871 const struct in6_addr *daddr)
299d9939 872{
21efcfa0
ED		 873 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
874
887c95cc 875 if (rt)
299d9939 876 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 877 return rt;
878}
879
8ed67789 880static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 881 struct flowi6 *fl6, int flags)
1da177e4
LT		 882{
883 struct fib6_node *fn;
519fbd87 884 struct rt6_info *rt, *nrt;
c71099ac 885 int strict = 0;
1da177e4 886 int attempts = 3;
519fbd87 887 int err;
53b7997f 888 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
1da177e4 889
77d16f45 890 strict |= flags & RT6_LOOKUP_F_IFACE;
1da177e4
LT		 891
892relookup:
c71099ac 893 read_lock_bh(&table->tb6_lock);
1da177e4 894
8238dd06 895restart_2:
4c9483b2 896 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1da177e4
LT		 897
898restart:
4acad72d 899 rt = rt6_select(fn, oif, strict | reachable);
52bd4c0c
ND		 900 if (rt->rt6i_nsiblings)
901 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable);
4c9483b2 902 BACKTRACK(net, &fl6->saddr);
8ed67789 903 if (rt == net->ipv6.ip6_null_entry ||
8238dd06 904 rt->rt6i_flags & RTF_CACHE)
1ddef044 905 goto out;
1da177e4 906
d8d1f30b 907 dst_hold(&rt->dst);
c71099ac 908 read_unlock_bh(&table->tb6_lock);
fb9de91e 909
c440f160 910 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
4c9483b2 911 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
7343ff31 912 else if (!(rt->dst.flags & DST_HOST))
4c9483b2 913 nrt = rt6_alloc_clone(rt, &fl6->daddr);
7343ff31
DM		 914 else
915 goto out2;
e40cf353 916
94e187c0 917 ip6_rt_put(rt);
8ed67789 918 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 919
d8d1f30b 920 dst_hold(&rt->dst);
519fbd87 921 if (nrt) {
40e22e8f 922 err = ip6_ins_rt(nrt);
519fbd87 923 if (!err)
1da177e4 924 goto out2;
1da177e4 925 }
1da177e4 926
519fbd87
YH		 927 if (--attempts <= 0)
928 goto out2;
929
930 /*
c71099ac 931 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 932 * released someone could insert this route. Relookup.
933 */
94e187c0 934 ip6_rt_put(rt);
519fbd87
YH		 935 goto relookup;
936
937out:
8238dd06
YH		 938 if (reachable) {
939 reachable = 0;
940 goto restart_2;
941 }
d8d1f30b 942 dst_hold(&rt->dst);
c71099ac 943 read_unlock_bh(&table->tb6_lock);
1da177e4 944out2:
d8d1f30b
CG		 945 rt->dst.lastuse = jiffies;
946 rt->dst.__use++;
c71099ac
TG		 947
948 return rt;
1da177e4
LT		 949}
950
8ed67789 951static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 952 struct flowi6 *fl6, int flags)
4acad72d 953{
4c9483b2 954 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 955}
956
72331bc0
SL		 957static struct dst_entry *ip6_route_input_lookup(struct net *net,
958 struct net_device *dev,
959 struct flowi6 *fl6, int flags)
960{
961 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
962 flags |= RT6_LOOKUP_F_IFACE;
963
964 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
965}
966
c71099ac
TG		 967void ip6_route_input(struct sk_buff *skb)
968{
b71d1d42 969 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 970 struct net *net = dev_net(skb->dev);
adaa70bb 971 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 972 struct flowi6 fl6 = {
973 .flowi6_iif = skb->dev->ifindex,
974 .daddr = iph->daddr,
975 .saddr = iph->saddr,
6502ca52 976 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 977 .flowi6_mark = skb->mark,
978 .flowi6_proto = iph->nexthdr,
c71099ac 979 };
adaa70bb 980
72331bc0 981 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 982}
983
8ed67789 984static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 985 struct flowi6 *fl6, int flags)
1da177e4 986{
4c9483b2 987 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 988}
989
9c7a4f9c 990struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 991 struct flowi6 *fl6)
c71099ac
TG		 992{
993 int flags = 0;
994
1fb9489b 995 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 996
4c9483b2 997 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 998 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 999
4c9483b2 1000 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 1001 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1002 else if (sk)
1003 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1004
4c9483b2 1005 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4
LT		 1006}
1007
7159039a 1008EXPORT_SYMBOL(ip6_route_output);
1da177e4 1009
2774c131 1010struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1011{
5c1e6aa3 1012 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1013 struct dst_entry *new = NULL;
1014
f5b0a874 1015 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1016 if (rt) {
d8d1f30b 1017 new = &rt->dst;
14e50e57 1018
8104891b
SK		 1019 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1020 rt6_init_peer(rt, net->ipv6.peers);
1021
14e50e57 1022 new->__use = 1;
352e512c
HX		 1023 new->input = dst_discard;
1024 new->output = dst_discard;
14e50e57 1025
21efcfa0
ED		 1026 if (dst_metrics_read_only(&ort->dst))
1027 new->_metrics = ort->dst._metrics;
1028 else
1029 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1030 rt->rt6i_idev = ort->rt6i_idev;
1031 if (rt->rt6i_idev)
1032 in6_dev_hold(rt->rt6i_idev);
14e50e57 1033
4e3fd7a0 1034 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1035 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1036 rt->rt6i_metric = 0;
1037
1038 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1039#ifdef CONFIG_IPV6_SUBTREES
1040 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1041#endif
1042
1043 dst_free(new);
1044 }
1045
69ead7af
DM		 1046 dst_release(dst_orig);
1047 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1048}
14e50e57 1049
1da177e4
LT		 1050/*
1051 * Destination cache support functions
1052 */
1053
1054static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1055{
1056 struct rt6_info *rt;
1057
1058 rt = (struct rt6_info *) dst;
1059
6f3118b5
ND		 1060 /* All IPV6 dsts are created with ->obsolete set to the value
1061 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1062 * into this function always.
1063 */
ca4c3fc2 1064 if (rt->rt6i_genid != rt_genid_ipv6(dev_net(rt->dst.dev)))
6f3118b5
ND		 1065 return NULL;
1066
a4477c4d 1067 if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie))
1da177e4 1068 return dst;
a4477c4d 1069
1da177e4
LT		 1070 return NULL;
1071}
1072
1073static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1074{
1075 struct rt6_info *rt = (struct rt6_info *) dst;
1076
1077 if (rt) {
54c1a859
YH		 1078 if (rt->rt6i_flags & RTF_CACHE) {
1079 if (rt6_check_expired(rt)) {
1080 ip6_del_rt(rt);
1081 dst = NULL;
1082 }
1083 } else {
1da177e4 1084 dst_release(dst);
54c1a859
YH		 1085 dst = NULL;
1086 }
1da177e4 1087 }
54c1a859 1088 return dst;
1da177e4
LT		 1089}
1090
1091static void ip6_link_failure(struct sk_buff *skb)
1092{
1093 struct rt6_info *rt;
1094
3ffe533c 1095 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1096
adf30907 1097 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1098 if (rt) {
1eb4f758
HFS		 1099 if (rt->rt6i_flags & RTF_CACHE) {
1100 dst_hold(&rt->dst);
1101 if (ip6_del_rt(rt))
1102 dst_free(&rt->dst);
1103 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1104 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1105 }
1da177e4
LT		 1106 }
1107}
1108
6700c270
DM		 1109static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1110 struct sk_buff *skb, u32 mtu)
1da177e4
LT		 1111{
1112 struct rt6_info *rt6 = (struct rt6_info*)dst;
1113
81aded24 1114 dst_confirm(dst);
1da177e4 1115 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
81aded24
DM		 1116 struct net *net = dev_net(dst->dev);
1117
1da177e4
LT		 1118 rt6->rt6i_flags |= RTF_MODIFIED;
1119 if (mtu < IPV6_MIN_MTU) {
defb3519 1120 u32 features = dst_metric(dst, RTAX_FEATURES);
1da177e4 1121 mtu = IPV6_MIN_MTU;
defb3519
DM		 1122 features |= RTAX_FEATURE_ALLFRAG;
1123 dst_metric_set(dst, RTAX_FEATURES, features);
1da177e4 1124 }
defb3519 1125 dst_metric_set(dst, RTAX_MTU, mtu);
81aded24 1126 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1127 }
1128}
1129
42ae66c8
DM		 1130void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1131 int oif, u32 mark)
81aded24
DM		 1132{
1133 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1134 struct dst_entry *dst;
1135 struct flowi6 fl6;
1136
1137 memset(&fl6, 0, sizeof(fl6));
1138 fl6.flowi6_oif = oif;
1139 fl6.flowi6_mark = mark;
3e12939a 1140 fl6.flowi6_flags = 0;
81aded24
DM		 1141 fl6.daddr = iph->daddr;
1142 fl6.saddr = iph->saddr;
6502ca52 1143 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1144
1145 dst = ip6_route_output(net, NULL, &fl6);
1146 if (!dst->error)
6700c270 1147 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
81aded24
DM		 1148 dst_release(dst);
1149}
1150EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1151
1152void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1153{
1154 ip6_update_pmtu(skb, sock_net(sk), mtu,
1155 sk->sk_bound_dev_if, sk->sk_mark);
1156}
1157EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1158
b55b76b2
DJ		 1159/* Handle redirects */
1160struct ip6rd_flowi {
1161 struct flowi6 fl6;
1162 struct in6_addr gateway;
1163};
1164
1165static struct rt6_info *__ip6_route_redirect(struct net *net,
1166 struct fib6_table *table,
1167 struct flowi6 *fl6,
1168 int flags)
1169{
1170 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1171 struct rt6_info *rt;
1172 struct fib6_node *fn;
1173
1174 /* Get the "current" route for this destination and
1175 * check if the redirect has come from approriate router.
1176 *
1177 * RFC 4861 specifies that redirects should only be
1178 * accepted if they come from the nexthop to the target.
1179 * Due to the way the routes are chosen, this notion
1180 * is a bit fuzzy and one might need to check all possible
1181 * routes.
1182 */
1183
1184 read_lock_bh(&table->tb6_lock);
1185 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1186restart:
1187 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1188 if (rt6_check_expired(rt))
1189 continue;
1190 if (rt->dst.error)
1191 break;
1192 if (!(rt->rt6i_flags & RTF_GATEWAY))
1193 continue;
1194 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1195 continue;
1196 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1197 continue;
1198 break;
1199 }
1200
1201 if (!rt)
1202 rt = net->ipv6.ip6_null_entry;
1203 else if (rt->dst.error) {
1204 rt = net->ipv6.ip6_null_entry;
1205 goto out;
1206 }
1207 BACKTRACK(net, &fl6->saddr);
1208out:
1209 dst_hold(&rt->dst);
1210
1211 read_unlock_bh(&table->tb6_lock);
1212
1213 return rt;
1214};
1215
1216static struct dst_entry *ip6_route_redirect(struct net *net,
1217 const struct flowi6 *fl6,
1218 const struct in6_addr *gateway)
1219{
1220 int flags = RT6_LOOKUP_F_HAS_SADDR;
1221 struct ip6rd_flowi rdfl;
1222
1223 rdfl.fl6 = *fl6;
1224 rdfl.gateway = *gateway;
1225
1226 return fib6_rule_lookup(net, &rdfl.fl6,
1227 flags, __ip6_route_redirect);
1228}
1229
3a5ad2ee
DM		 1230void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1231{
1232 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1233 struct dst_entry *dst;
1234 struct flowi6 fl6;
1235
1236 memset(&fl6, 0, sizeof(fl6));
1237 fl6.flowi6_oif = oif;
1238 fl6.flowi6_mark = mark;
1239 fl6.flowi6_flags = 0;
1240 fl6.daddr = iph->daddr;
1241 fl6.saddr = iph->saddr;
6502ca52 1242 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1243
b55b76b2
DJ		 1244 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1245 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1246 dst_release(dst);
1247}
1248EXPORT_SYMBOL_GPL(ip6_redirect);
1249
c92a59ec
DJ		 1250void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1251 u32 mark)
1252{
1253 const struct ipv6hdr *iph = ipv6_hdr(skb);
1254 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1255 struct dst_entry *dst;
1256 struct flowi6 fl6;
1257
1258 memset(&fl6, 0, sizeof(fl6));
1259 fl6.flowi6_oif = oif;
1260 fl6.flowi6_mark = mark;
1261 fl6.flowi6_flags = 0;
1262 fl6.daddr = msg->dest;
1263 fl6.saddr = iph->daddr;
1264
b55b76b2
DJ		 1265 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1266 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1267 dst_release(dst);
1268}
1269
3a5ad2ee
DM		 1270void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1271{
1272 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1273}
1274EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1275
0dbaee3b 1276static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1277{
0dbaee3b
DM		 1278 struct net_device *dev = dst->dev;
1279 unsigned int mtu = dst_mtu(dst);
1280 struct net *net = dev_net(dev);
1281
1da177e4
LT		 1282 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1283
5578689a
DL		 1284 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1285 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1286
1287 /*
1ab1457c
YH		 1288 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1289 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1290 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1291 * rely only on pmtu discovery"
1292 */
1293 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1294 mtu = IPV6_MAXPLEN;
1295 return mtu;
1296}
1297
ebb762f2 1298static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1299{
d33e4553 1300 struct inet6_dev *idev;
618f9bc7
SK		 1301 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1302
1303 if (mtu)
1304 return mtu;
1305
1306 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1307
1308 rcu_read_lock();
1309 idev = __in6_dev_get(dst->dev);
1310 if (idev)
1311 mtu = idev->cnf.mtu6;
1312 rcu_read_unlock();
1313
1314 return mtu;
1315}
1316
3b00944c
YH		 1317static struct dst_entry *icmp6_dst_gc_list;
1318static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1319
3b00944c 1320struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1321 struct flowi6 *fl6)
1da177e4 1322{
87a11578 1323 struct dst_entry *dst;
1da177e4
LT		 1324 struct rt6_info *rt;
1325 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1326 struct net *net = dev_net(dev);
1da177e4 1327
38308473 1328 if (unlikely(!idev))
122bdf67 1329 return ERR_PTR(-ENODEV);
1da177e4 1330
8b96d22d 1331 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1332 if (unlikely(!rt)) {
1da177e4 1333 in6_dev_put(idev);
87a11578 1334 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1335 goto out;
1336 }
1337
8e2ec639
YZ		 1338 rt->dst.flags |= DST_HOST;
1339 rt->dst.output = ip6_output;
d8d1f30b 1340 atomic_set(&rt->dst.__refcnt, 1);
87a11578 1341 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1342 rt->rt6i_dst.plen = 128;
1343 rt->rt6i_idev = idev;
14edd87d 1344 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1345
3b00944c 1346 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1347 rt->dst.next = icmp6_dst_gc_list;
1348 icmp6_dst_gc_list = &rt->dst;
3b00944c 1349 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1350
5578689a 1351 fib6_force_start_gc(net);
1da177e4 1352
87a11578
DM		 1353 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1354
1da177e4 1355out:
87a11578 1356 return dst;
1da177e4
LT		 1357}
1358
3d0f24a7 1359int icmp6_dst_gc(void)
1da177e4 1360{
e9476e95 1361 struct dst_entry *dst, **pprev;
3d0f24a7 1362 int more = 0;
1da177e4 1363
3b00944c
YH		 1364 spin_lock_bh(&icmp6_dst_lock);
1365 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1366
1da177e4
LT		 1367 while ((dst = *pprev) != NULL) {
1368 if (!atomic_read(&dst->__refcnt)) {
1369 *pprev = dst->next;
1370 dst_free(dst);
1da177e4
LT		 1371 } else {
1372 pprev = &dst->next;
3d0f24a7 1373 ++more;
1da177e4
LT		 1374 }
1375 }
1376
3b00944c 1377 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1378
3d0f24a7 1379 return more;
1da177e4
LT		 1380}
1381
1e493d19
DM		 1382static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1383 void *arg)
1384{
1385 struct dst_entry *dst, **pprev;
1386
1387 spin_lock_bh(&icmp6_dst_lock);
1388 pprev = &icmp6_dst_gc_list;
1389 while ((dst = *pprev) != NULL) {
1390 struct rt6_info *rt = (struct rt6_info *) dst;
1391 if (func(rt, arg)) {
1392 *pprev = dst->next;
1393 dst_free(dst);
1394 } else {
1395 pprev = &dst->next;
1396 }
1397 }
1398 spin_unlock_bh(&icmp6_dst_lock);
1399}
1400
569d3645 1401static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1402{
86393e52 1403 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1404 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1405 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1406 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1407 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1408 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1409 int entries;
7019b78e 1410
fc66f95c 1411 entries = dst_entries_get_fast(ops);
49a18d86 1412 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1413 entries <= rt_max_size)
1da177e4
LT		 1414 goto out;
1415
6891a346 1416 net->ipv6.ip6_rt_gc_expire++;
2ac3ac8f 1417 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size);
fc66f95c
ED		 1418 entries = dst_entries_get_slow(ops);
1419 if (entries < ops->gc_thresh)
7019b78e 1420 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1421out:
7019b78e 1422 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1423 return entries > rt_max_size;
1da177e4
LT		 1424}
1425
1da177e4
LT		 1426/*
1427 *
1428 */
1429
86872cb5 1430int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1431{
1432 int err;
5578689a 1433 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1434 struct rt6_info *rt = NULL;
1435 struct net_device *dev = NULL;
1436 struct inet6_dev *idev = NULL;
c71099ac 1437 struct fib6_table *table;
1da177e4
LT		 1438 int addr_type;
1439
86872cb5 1440 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1441 return -EINVAL;
1442#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1443 if (cfg->fc_src_len)
1da177e4
LT		 1444 return -EINVAL;
1445#endif
86872cb5 1446 if (cfg->fc_ifindex) {
1da177e4 1447 err = -ENODEV;
5578689a 1448 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1449 if (!dev)
1450 goto out;
1451 idev = in6_dev_get(dev);
1452 if (!idev)
1453 goto out;
1454 }
1455
86872cb5
TG		 1456 if (cfg->fc_metric == 0)
1457 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1458
d71314b4 1459 err = -ENOBUFS;
38308473
DM		 1460 if (cfg->fc_nlinfo.nlh &&
1461 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1462 table = fib6_get_table(net, cfg->fc_table);
38308473 1463 if (!table) {
f3213831 1464 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1465 table = fib6_new_table(net, cfg->fc_table);
1466 }
1467 } else {
1468 table = fib6_new_table(net, cfg->fc_table);
1469 }
38308473
DM		 1470
1471 if (!table)
c71099ac 1472 goto out;
c71099ac 1473
8b96d22d 1474 rt = ip6_dst_alloc(net, NULL, DST_NOCOUNT, table);
1da177e4 1475
38308473 1476 if (!rt) {
1da177e4
LT		 1477 err = -ENOMEM;
1478 goto out;
1479 }
1480
1716a961
G		 1481 if (cfg->fc_flags & RTF_EXPIRES)
1482 rt6_set_expires(rt, jiffies +
1483 clock_t_to_jiffies(cfg->fc_expires));
1484 else
1485 rt6_clean_expires(rt);
1da177e4 1486
86872cb5
TG		 1487 if (cfg->fc_protocol == RTPROT_UNSPEC)
1488 cfg->fc_protocol = RTPROT_BOOT;
1489 rt->rt6i_protocol = cfg->fc_protocol;
1490
1491 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1492
1493 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1494 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1495 else if (cfg->fc_flags & RTF_LOCAL)
1496 rt->dst.input = ip6_input;
1da177e4 1497 else
d8d1f30b 1498 rt->dst.input = ip6_forward;
1da177e4 1499
d8d1f30b 1500 rt->dst.output = ip6_output;
1da177e4 1501
86872cb5
TG		 1502 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1503 rt->rt6i_dst.plen = cfg->fc_dst_len;
1da177e4 1504 if (rt->rt6i_dst.plen == 128)
11d53b49 1505 rt->dst.flags |= DST_HOST;
1da177e4 1506
8e2ec639
YZ		 1507 if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) {
1508 u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1509 if (!metrics) {
1510 err = -ENOMEM;
1511 goto out;
1512 }
1513 dst_init_metrics(&rt->dst, metrics, 0);
1514 }
1da177e4 1515#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1516 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1517 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1518#endif
1519
86872cb5 1520 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1521
1522 /* We cannot add true routes via loopback here,
1523 they would result in kernel looping; promote them to reject routes
1524 */
86872cb5 1525 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1526 (dev && (dev->flags & IFF_LOOPBACK) &&
1527 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1528 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1529 /* hold loopback dev/idev if we haven't done so. */
5578689a 1530 if (dev != net->loopback_dev) {
1da177e4
LT		 1531 if (dev) {
1532 dev_put(dev);
1533 in6_dev_put(idev);
1534 }
5578689a 1535 dev = net->loopback_dev;
1da177e4
LT		 1536 dev_hold(dev);
1537 idev = in6_dev_get(dev);
1538 if (!idev) {
1539 err = -ENODEV;
1540 goto out;
1541 }
1542 }
d8d1f30b
CG		 1543 rt->dst.output = ip6_pkt_discard_out;
1544 rt->dst.input = ip6_pkt_discard;
1da177e4 1545 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1546 switch (cfg->fc_type) {
1547 case RTN_BLACKHOLE:
1548 rt->dst.error = -EINVAL;
1549 break;
1550 case RTN_PROHIBIT:
1551 rt->dst.error = -EACCES;
1552 break;
b4949ab2
ND		 1553 case RTN_THROW:
1554 rt->dst.error = -EAGAIN;
1555 break;
ef2c7d7b
ND		 1556 default:
1557 rt->dst.error = -ENETUNREACH;
1558 break;
1559 }
1da177e4
LT		 1560 goto install_route;
1561 }
1562
86872cb5 1563 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1564 const struct in6_addr *gw_addr;
1da177e4
LT		 1565 int gwa_type;
1566
86872cb5 1567 gw_addr = &cfg->fc_gateway;
4e3fd7a0 1568 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1569 gwa_type = ipv6_addr_type(gw_addr);
1570
1571 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1572 struct rt6_info *grt;
1573
1574 /* IPv6 strictly inhibits using not link-local
1575 addresses as nexthop address.
1576 Otherwise, router will not able to send redirects.
1577 It is very good, but in some (rare!) circumstances
1578 (SIT, PtP, NBMA NOARP links) it is handy to allow
1579 some exceptions. --ANK
1580 */
1581 err = -EINVAL;
38308473 1582 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1583 goto out;
1584
5578689a 1585 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1586
1587 err = -EHOSTUNREACH;
38308473 1588 if (!grt)
1da177e4
LT		 1589 goto out;
1590 if (dev) {
d1918542 1591 if (dev != grt->dst.dev) {
94e187c0 1592 ip6_rt_put(grt);
1da177e4
LT		 1593 goto out;
1594 }
1595 } else {
d1918542 1596 dev = grt->dst.dev;
1da177e4
LT		 1597 idev = grt->rt6i_idev;
1598 dev_hold(dev);
1599 in6_dev_hold(grt->rt6i_idev);
1600 }
38308473 1601 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1602 err = 0;
94e187c0 1603 ip6_rt_put(grt);
1da177e4
LT		 1604
1605 if (err)
1606 goto out;
1607 }
1608 err = -EINVAL;
38308473 1609 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1610 goto out;
1611 }
1612
1613 err = -ENODEV;
38308473 1614 if (!dev)
1da177e4
LT		 1615 goto out;
1616
c3968a85
DW		 1617 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1618 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1619 err = -EINVAL;
1620 goto out;
1621 }
4e3fd7a0 1622 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1623 rt->rt6i_prefsrc.plen = 128;
1624 } else
1625 rt->rt6i_prefsrc.plen = 0;
1626
86872cb5 1627 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1628
1629install_route:
86872cb5
TG		 1630 if (cfg->fc_mx) {
1631 struct nlattr *nla;
1632 int remaining;
1633
1634 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
8f4c1f9b 1635 int type = nla_type(nla);
86872cb5
TG		 1636
1637 if (type) {
1638 if (type > RTAX_MAX) {
1da177e4
LT		 1639 err = -EINVAL;
1640 goto out;
1641 }
86872cb5 1642
defb3519 1643 dst_metric_set(&rt->dst, type, nla_get_u32(nla));
1da177e4 1644 }
1da177e4
LT		 1645 }
1646 }
1647
d8d1f30b 1648 rt->dst.dev = dev;
1da177e4 1649 rt->rt6i_idev = idev;
c71099ac 1650 rt->rt6i_table = table;
63152fc0 1651
c346dca1 1652 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1653
86872cb5 1654 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1655
1656out:
1657 if (dev)
1658 dev_put(dev);
1659 if (idev)
1660 in6_dev_put(idev);
1661 if (rt)
d8d1f30b 1662 dst_free(&rt->dst);
1da177e4
LT		 1663 return err;
1664}
1665
86872cb5 1666static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1667{
1668 int err;
c71099ac 1669 struct fib6_table *table;
d1918542 1670 struct net *net = dev_net(rt->dst.dev);
1da177e4 1671
6825a26c
G		 1672 if (rt == net->ipv6.ip6_null_entry) {
1673 err = -ENOENT;
1674 goto out;
1675 }
6c813a72 1676
c71099ac
TG		 1677 table = rt->rt6i_table;
1678 write_lock_bh(&table->tb6_lock);
86872cb5 1679 err = fib6_del(rt, info);
c71099ac 1680 write_unlock_bh(&table->tb6_lock);
1da177e4 1681
6825a26c 1682out:
94e187c0 1683 ip6_rt_put(rt);
1da177e4
LT		 1684 return err;
1685}
1686
e0a1ad73
TG		 1687int ip6_del_rt(struct rt6_info *rt)
1688{
4d1169c1 1689 struct nl_info info = {
d1918542 1690 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1691 };
528c4ceb 1692 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1693}
1694
86872cb5 1695static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1696{
c71099ac 1697 struct fib6_table *table;
1da177e4
LT		 1698 struct fib6_node *fn;
1699 struct rt6_info *rt;
1700 int err = -ESRCH;
1701
5578689a 1702 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1703 if (!table)
c71099ac
TG		 1704 return err;
1705
1706 read_lock_bh(&table->tb6_lock);
1da177e4 1707
c71099ac 1708 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1709 &cfg->fc_dst, cfg->fc_dst_len,
1710 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1711
1da177e4 1712 if (fn) {
d8d1f30b 1713 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
86872cb5 1714 if (cfg->fc_ifindex &&
d1918542
DM		 1715 (!rt->dst.dev ||
1716 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1717 continue;
86872cb5
TG		 1718 if (cfg->fc_flags & RTF_GATEWAY &&
1719 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1720 continue;
86872cb5 1721 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1722 continue;
d8d1f30b 1723 dst_hold(&rt->dst);
c71099ac 1724 read_unlock_bh(&table->tb6_lock);
1da177e4 1725
86872cb5 1726 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1727 }
1728 }
c71099ac 1729 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1730
1731 return err;
1732}
1733
6700c270 1734static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1735{
e8599ff4 1736 struct net *net = dev_net(skb->dev);
a6279458 1737 struct netevent_redirect netevent;
e8599ff4 1738 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1739 struct ndisc_options ndopts;
1740 struct inet6_dev *in6_dev;
1741 struct neighbour *neigh;
71bcdba0 1742 struct rd_msg *msg;
6e157b6a
DM		 1743 int optlen, on_link;
1744 u8 *lladdr;
e8599ff4 1745
29a3cad5 1746 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1747 optlen -= sizeof(*msg);
e8599ff4
DM		 1748
1749 if (optlen < 0) {
6e157b6a 1750 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1751 return;
1752 }
1753
71bcdba0 1754 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1755
71bcdba0 1756 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1757 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1758 return;
1759 }
1760
6e157b6a 1761 on_link = 0;
71bcdba0 1762 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1763 on_link = 1;
71bcdba0 1764 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1765 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1766 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1767 return;
1768 }
1769
1770 in6_dev = __in6_dev_get(skb->dev);
1771 if (!in6_dev)
1772 return;
1773 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1774 return;
1775
1776 /* RFC2461 8.1:
1777 * The IP source address of the Redirect MUST be the same as the current
1778 * first-hop router for the specified ICMP Destination Address.
1779 */
1780
71bcdba0 1781 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1782 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1783 return;
1784 }
6e157b6a
DM		 1785
1786 lladdr = NULL;
e8599ff4
DM		 1787 if (ndopts.nd_opts_tgt_lladdr) {
1788 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1789 skb->dev);
1790 if (!lladdr) {
1791 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1792 return;
1793 }
1794 }
1795
6e157b6a
DM		 1796 rt = (struct rt6_info *) dst;
1797 if (rt == net->ipv6.ip6_null_entry) {
1798 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1799 return;
6e157b6a 1800 }
e8599ff4 1801
6e157b6a
DM		 1802 /* Redirect received -> path was valid.
1803 * Look, redirects are sent only in response to data packets,
1804 * so that this nexthop apparently is reachable. --ANK
1805 */
1806 dst_confirm(&rt->dst);
a6279458 1807
71bcdba0 1808 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1809 if (!neigh)
1810 return;
a6279458 1811
1da177e4
LT		 1812 /*
1813 * We have finally decided to accept it.
1814 */
1815
1ab1457c 1816 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1817 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1818 NEIGH_UPDATE_F_OVERRIDE|
1819 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1820 NEIGH_UPDATE_F_ISROUTER))
1821 );
1822
71bcdba0 1823 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1824 if (!nrt)
1da177e4
LT		 1825 goto out;
1826
1827 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1828 if (on_link)
1829 nrt->rt6i_flags &= ~RTF_GATEWAY;
1830
4e3fd7a0 1831 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1832
40e22e8f 1833 if (ip6_ins_rt(nrt))
1da177e4
LT		 1834 goto out;
1835
d8d1f30b
CG		 1836 netevent.old = &rt->dst;
1837 netevent.new = &nrt->dst;
71bcdba0 1838 netevent.daddr = &msg->dest;
60592833 1839 netevent.neigh = neigh;
8d71740c
TT		 1840 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1841
38308473 1842 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1843 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1844 ip6_del_rt(rt);
1da177e4
LT		 1845 }
1846
1847out:
e8599ff4 1848 neigh_release(neigh);
6e157b6a
DM		 1849}
1850
1da177e4
LT		 1851/*
1852 * Misc support functions
1853 */
1854
1716a961 1855static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1856 const struct in6_addr *dest)
1da177e4 1857{
d1918542 1858 struct net *net = dev_net(ort->dst.dev);
8b96d22d
DM		 1859 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1860 ort->rt6i_table);
1da177e4
LT		 1861
1862 if (rt) {
d8d1f30b
CG		 1863 rt->dst.input = ort->dst.input;
1864 rt->dst.output = ort->dst.output;
8e2ec639 1865 rt->dst.flags |= DST_HOST;
d8d1f30b 1866
4e3fd7a0 1867 rt->rt6i_dst.addr = *dest;
8e2ec639 1868 rt->rt6i_dst.plen = 128;
defb3519 1869 dst_copy_metrics(&rt->dst, &ort->dst);
d8d1f30b 1870 rt->dst.error = ort->dst.error;
1da177e4
LT		 1871 rt->rt6i_idev = ort->rt6i_idev;
1872 if (rt->rt6i_idev)
1873 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1874 rt->dst.lastuse = jiffies;
1da177e4 1875
4e3fd7a0 1876 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961
G		 1877 rt->rt6i_flags = ort->rt6i_flags;
1878 if ((ort->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) ==
1879 (RTF_DEFAULT | RTF_ADDRCONF))
1880 rt6_set_from(rt, ort);
1da177e4
LT		 1881 rt->rt6i_metric = 0;
1882
1da177e4
LT		 1883#ifdef CONFIG_IPV6_SUBTREES
1884 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1885#endif
0f6c6392 1886 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1887 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1888 }
1889 return rt;
1890}
1891
70ceb4f5 1892#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1893static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 1894 const struct in6_addr *prefix, int prefixlen,
1895 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 1896{
1897 struct fib6_node *fn;
1898 struct rt6_info *rt = NULL;
c71099ac
TG		 1899 struct fib6_table *table;
1900
efa2cea0 1901 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 1902 if (!table)
c71099ac 1903 return NULL;
70ceb4f5 1904
5744dd9b 1905 read_lock_bh(&table->tb6_lock);
c71099ac 1906 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
70ceb4f5
YH		 1907 if (!fn)
1908 goto out;
1909
d8d1f30b 1910 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1911 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 1912 continue;
1913 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1914 continue;
1915 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1916 continue;
d8d1f30b 1917 dst_hold(&rt->dst);
70ceb4f5
YH		 1918 break;
1919 }
1920out:
5744dd9b 1921 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1922 return rt;
1923}
1924
efa2cea0 1925static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 1926 const struct in6_addr *prefix, int prefixlen,
1927 const struct in6_addr *gwaddr, int ifindex,
95c96174 1928 unsigned int pref)
70ceb4f5 1929{
86872cb5
TG		 1930 struct fib6_config cfg = {
1931 .fc_table = RT6_TABLE_INFO,
238fc7ea 1932 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1933 .fc_ifindex = ifindex,
1934 .fc_dst_len = prefixlen,
1935 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1936 RTF_UP | RTF_PREF(pref),
15e47304 1937 .fc_nlinfo.portid = 0,
efa2cea0
DL		 1938 .fc_nlinfo.nlh = NULL,
1939 .fc_nlinfo.nl_net = net,
86872cb5
TG		 1940 };
1941
4e3fd7a0
AD		 1942 cfg.fc_dst = *prefix;
1943 cfg.fc_gateway = *gwaddr;
70ceb4f5 1944
e317da96
YH		 1945 /* We should treat it as a default route if prefix length is 0. */
1946 if (!prefixlen)
86872cb5 1947 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 1948
86872cb5 1949 ip6_route_add(&cfg);
70ceb4f5 1950
efa2cea0 1951 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 1952}
1953#endif
1954
b71d1d42 1955struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 1956{
1da177e4 1957 struct rt6_info *rt;
c71099ac 1958 struct fib6_table *table;
1da177e4 1959
c346dca1 1960 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 1961 if (!table)
c71099ac 1962 return NULL;
1da177e4 1963
5744dd9b 1964 read_lock_bh(&table->tb6_lock);
d8d1f30b 1965 for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) {
d1918542 1966 if (dev == rt->dst.dev &&
045927ff 1967 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 1968 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1969 break;
1970 }
1971 if (rt)
d8d1f30b 1972 dst_hold(&rt->dst);
5744dd9b 1973 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1974 return rt;
1975}
1976
b71d1d42 1977struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 1978 struct net_device *dev,
1979 unsigned int pref)
1da177e4 1980{
86872cb5
TG		 1981 struct fib6_config cfg = {
1982 .fc_table = RT6_TABLE_DFLT,
238fc7ea 1983 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1984 .fc_ifindex = dev->ifindex,
1985 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
1986 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 1987 .fc_nlinfo.portid = 0,
5578689a 1988 .fc_nlinfo.nlh = NULL,
c346dca1 1989 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 1990 };
1da177e4 1991
4e3fd7a0 1992 cfg.fc_gateway = *gwaddr;
1da177e4 1993
86872cb5 1994 ip6_route_add(&cfg);
1da177e4 1995
1da177e4
LT		 1996 return rt6_get_dflt_router(gwaddr, dev);
1997}
1998
7b4da532 1999void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2000{
2001 struct rt6_info *rt;
c71099ac
TG		 2002 struct fib6_table *table;
2003
2004 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2005 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2006 if (!table)
c71099ac 2007 return;
1da177e4
LT		 2008
2009restart:
c71099ac 2010 read_lock_bh(&table->tb6_lock);
d8d1f30b 2011 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2012 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2013 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2014 dst_hold(&rt->dst);
c71099ac 2015 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2016 ip6_del_rt(rt);
1da177e4
LT		 2017 goto restart;
2018 }
2019 }
c71099ac 2020 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2021}
2022
5578689a
DL		 2023static void rtmsg_to_fib6_config(struct net *net,
2024 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2025 struct fib6_config *cfg)
2026{
2027 memset(cfg, 0, sizeof(*cfg));
2028
2029 cfg->fc_table = RT6_TABLE_MAIN;
2030 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2031 cfg->fc_metric = rtmsg->rtmsg_metric;
2032 cfg->fc_expires = rtmsg->rtmsg_info;
2033 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2034 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2035 cfg->fc_flags = rtmsg->rtmsg_flags;
2036
5578689a 2037 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2038
4e3fd7a0
AD		 2039 cfg->fc_dst = rtmsg->rtmsg_dst;
2040 cfg->fc_src = rtmsg->rtmsg_src;
2041 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2042}
2043
5578689a 2044int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2045{
86872cb5 2046 struct fib6_config cfg;
1da177e4
LT		 2047 struct in6_rtmsg rtmsg;
2048 int err;
2049
2050 switch(cmd) {
2051 case SIOCADDRT: /* Add a route */
2052 case SIOCDELRT: /* Delete a route */
af31f412 2053 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2054 return -EPERM;
2055 err = copy_from_user(&rtmsg, arg,
2056 sizeof(struct in6_rtmsg));
2057 if (err)
2058 return -EFAULT;
86872cb5 2059
5578689a 2060 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2061
1da177e4
LT		 2062 rtnl_lock();
2063 switch (cmd) {
2064 case SIOCADDRT:
86872cb5 2065 err = ip6_route_add(&cfg);
1da177e4
LT		 2066 break;
2067 case SIOCDELRT:
86872cb5 2068 err = ip6_route_del(&cfg);
1da177e4
LT		 2069 break;
2070 default:
2071 err = -EINVAL;
2072 }
2073 rtnl_unlock();
2074
2075 return err;
3ff50b79 2076 }
1da177e4
LT		 2077
2078 return -EINVAL;
2079}
2080
2081/*
2082 * Drop the packet on the floor
2083 */
2084
d5fdd6ba 2085static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2086{
612f09e8 2087 int type;
adf30907 2088 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2089 switch (ipstats_mib_noroutes) {
2090 case IPSTATS_MIB_INNOROUTES:
0660e03f 2091 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2092 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2093 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2094 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2095 break;
2096 }
2097 /* FALLTHROUGH */
2098 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2099 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2100 ipstats_mib_noroutes);
612f09e8
YH		 2101 break;
2102 }
3ffe533c 2103 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2104 kfree_skb(skb);
2105 return 0;
2106}
2107
9ce8ade0
TG		 2108static int ip6_pkt_discard(struct sk_buff *skb)
2109{
612f09e8 2110 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2111}
2112
20380731 2113static int ip6_pkt_discard_out(struct sk_buff *skb)
1da177e4 2114{
adf30907 2115 skb->dev = skb_dst(skb)->dev;
612f09e8 2116 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2117}
2118
6723ab54
DM		 2119#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2120
9ce8ade0
TG		 2121static int ip6_pkt_prohibit(struct sk_buff *skb)
2122{
612f09e8 2123 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2124}
2125
2126static int ip6_pkt_prohibit_out(struct sk_buff *skb)
2127{
adf30907 2128 skb->dev = skb_dst(skb)->dev;
612f09e8 2129 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2130}
2131
6723ab54
DM		 2132#endif
2133
1da177e4
LT		 2134/*
2135 * Allocate a dst for local (unicast / anycast) address.
2136 */
2137
2138struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2139 const struct in6_addr *addr,
8f031519 2140 bool anycast)
1da177e4 2141{
c346dca1 2142 struct net *net = dev_net(idev->dev);
8b96d22d 2143 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev, 0, NULL);
1da177e4 2144
38308473 2145 if (!rt) {
f3213831 2146 net_warn_ratelimited("Maximum number of routes reached, consider increasing route/max_size\n");
1da177e4 2147 return ERR_PTR(-ENOMEM);
40385653 2148 }
1da177e4 2149
1da177e4
LT		 2150 in6_dev_hold(idev);
2151
11d53b49 2152 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2153 rt->dst.input = ip6_input;
2154 rt->dst.output = ip6_output;
1da177e4 2155 rt->rt6i_idev = idev;
1da177e4
LT		 2156
2157 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2158 if (anycast)
2159 rt->rt6i_flags |= RTF_ANYCAST;
2160 else
1da177e4 2161 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2162
4e3fd7a0 2163 rt->rt6i_dst.addr = *addr;
1da177e4 2164 rt->rt6i_dst.plen = 128;
5578689a 2165 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2166
d8d1f30b 2167 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2168
2169 return rt;
2170}
2171
c3968a85
DW		 2172int ip6_route_get_saddr(struct net *net,
2173 struct rt6_info *rt,
b71d1d42 2174 const struct in6_addr *daddr,
c3968a85
DW		 2175 unsigned int prefs,
2176 struct in6_addr *saddr)
2177{
2178 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt);
2179 int err = 0;
2180 if (rt->rt6i_prefsrc.plen)
4e3fd7a0 2181 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2182 else
2183 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2184 daddr, prefs, saddr);
2185 return err;
2186}
2187
2188/* remove deleted ip from prefsrc entries */
2189struct arg_dev_net_ip {
2190 struct net_device *dev;
2191 struct net *net;
2192 struct in6_addr *addr;
2193};
2194
2195static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2196{
2197 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2198 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2199 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2200
d1918542 2201 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2202 rt != net->ipv6.ip6_null_entry &&
2203 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2204 /* remove prefsrc entry */
2205 rt->rt6i_prefsrc.plen = 0;
2206 }
2207 return 0;
2208}
2209
2210void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2211{
2212 struct net *net = dev_net(ifp->idev->dev);
2213 struct arg_dev_net_ip adni = {
2214 .dev = ifp->idev->dev,
2215 .net = net,
2216 .addr = &ifp->addr,
2217 };
2218 fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni);
2219}
2220
8ed67789
DL		 2221struct arg_dev_net {
2222 struct net_device *dev;
2223 struct net *net;
2224};
2225
1da177e4
LT		 2226static int fib6_ifdown(struct rt6_info *rt, void *arg)
2227{
bc3ef660 2228 const struct arg_dev_net *adn = arg;
2229 const struct net_device *dev = adn->dev;
8ed67789 2230
d1918542 2231 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2232 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2233 return -1;
c159d30c 2234
1da177e4
LT		 2235 return 0;
2236}
2237
f3db4851 2238void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2239{
8ed67789
DL		 2240 struct arg_dev_net adn = {
2241 .dev = dev,
2242 .net = net,
2243 };
2244
2245 fib6_clean_all(net, fib6_ifdown, 0, &adn);
1e493d19 2246 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2247}
2248
95c96174 2249struct rt6_mtu_change_arg {
1da177e4 2250 struct net_device *dev;
95c96174 2251 unsigned int mtu;
1da177e4
LT		 2252};
2253
2254static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2255{
2256 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2257 struct inet6_dev *idev;
2258
2259 /* In IPv6 pmtu discovery is not optional,
2260 so that RTAX_MTU lock cannot disable it.
2261 We still use this lock to block changes
2262 caused by addrconf/ndisc.
2263 */
2264
2265 idev = __in6_dev_get(arg->dev);
38308473 2266 if (!idev)
1da177e4
LT		 2267 return 0;
2268
2269 /* For administrative MTU increase, there is no way to discover
2270 IPv6 PMTU increase, so PMTU increase should be updated here.
2271 Since RFC 1981 doesn't include administrative MTU increase
2272 update PMTU increase is a MUST. (i.e. jumbo frame)
2273 */
2274 /*
2275 If new MTU is less than route PMTU, this new MTU will be the
2276 lowest MTU in the path, update the route PMTU to reflect PMTU
2277 decreases; if new MTU is greater than route PMTU, and the
2278 old MTU is the lowest MTU in the path, update the route PMTU
2279 to reflect the increase. In this case if the other nodes' MTU
2280 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2281 PMTU discouvery.
2282 */
d1918542 2283 if (rt->dst.dev == arg->dev &&
d8d1f30b
CG		 2284 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2285 (dst_mtu(&rt->dst) >= arg->mtu ||
2286 (dst_mtu(&rt->dst) < arg->mtu &&
2287 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
defb3519 2288 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
566cfd8f 2289 }
1da177e4
LT		 2290 return 0;
2291}
2292
95c96174 2293void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2294{
c71099ac
TG		 2295 struct rt6_mtu_change_arg arg = {
2296 .dev = dev,
2297 .mtu = mtu,
2298 };
1da177e4 2299
c346dca1 2300 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg);
1da177e4
LT		 2301}
2302
ef7c79ed 2303static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2304 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2305 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2306 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2307 [RTA_PRIORITY] = { .type = NLA_U32 },
2308 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2309 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
86872cb5
TG		 2310};
2311
2312static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2313 struct fib6_config *cfg)
1da177e4 2314{
86872cb5
TG		 2315 struct rtmsg *rtm;
2316 struct nlattr *tb[RTA_MAX+1];
2317 int err;
1da177e4 2318
86872cb5
TG		 2319 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2320 if (err < 0)
2321 goto errout;
1da177e4 2322
86872cb5
TG		 2323 err = -EINVAL;
2324 rtm = nlmsg_data(nlh);
2325 memset(cfg, 0, sizeof(*cfg));
2326
2327 cfg->fc_table = rtm->rtm_table;
2328 cfg->fc_dst_len = rtm->rtm_dst_len;
2329 cfg->fc_src_len = rtm->rtm_src_len;
2330 cfg->fc_flags = RTF_UP;
2331 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2332 cfg->fc_type = rtm->rtm_type;
86872cb5 2333
ef2c7d7b
ND		 2334 if (rtm->rtm_type == RTN_UNREACHABLE ||
2335 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2336 rtm->rtm_type == RTN_PROHIBIT ||
2337 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2338 cfg->fc_flags |= RTF_REJECT;
2339
ab79ad14
MŻ		 2340 if (rtm->rtm_type == RTN_LOCAL)
2341 cfg->fc_flags |= RTF_LOCAL;
2342
15e47304 2343 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2344 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2345 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2346
2347 if (tb[RTA_GATEWAY]) {
2348 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2349 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2350 }
86872cb5
TG		 2351
2352 if (tb[RTA_DST]) {
2353 int plen = (rtm->rtm_dst_len + 7) >> 3;
2354
2355 if (nla_len(tb[RTA_DST]) < plen)
2356 goto errout;
2357
2358 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2359 }
86872cb5
TG		 2360
2361 if (tb[RTA_SRC]) {
2362 int plen = (rtm->rtm_src_len + 7) >> 3;
2363
2364 if (nla_len(tb[RTA_SRC]) < plen)
2365 goto errout;
2366
2367 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2368 }
86872cb5 2369
c3968a85
DW		 2370 if (tb[RTA_PREFSRC])
2371 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2372
86872cb5
TG		 2373 if (tb[RTA_OIF])
2374 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2375
2376 if (tb[RTA_PRIORITY])
2377 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2378
2379 if (tb[RTA_METRICS]) {
2380 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2381 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2382 }
86872cb5
TG		 2383
2384 if (tb[RTA_TABLE])
2385 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2386
51ebd318
ND		 2387 if (tb[RTA_MULTIPATH]) {
2388 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2389 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2390 }
2391
86872cb5
TG		 2392 err = 0;
2393errout:
2394 return err;
1da177e4
LT		 2395}
2396
51ebd318
ND		 2397static int ip6_route_multipath(struct fib6_config *cfg, int add)
2398{
2399 struct fib6_config r_cfg;
2400 struct rtnexthop *rtnh;
2401 int remaining;
2402 int attrlen;
2403 int err = 0, last_err = 0;
2404
2405beginning:
2406 rtnh = (struct rtnexthop *)cfg->fc_mp;
2407 remaining = cfg->fc_mp_len;
2408
2409 /* Parse a Multipath Entry */
2410 while (rtnh_ok(rtnh, remaining)) {
2411 memcpy(&r_cfg, cfg, sizeof(*cfg));
2412 if (rtnh->rtnh_ifindex)
2413 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2414
2415 attrlen = rtnh_attrlen(rtnh);
2416 if (attrlen > 0) {
2417 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2418
2419 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2420 if (nla) {
2421 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2422 r_cfg.fc_flags |= RTF_GATEWAY;
2423 }
2424 }
2425 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2426 if (err) {
2427 last_err = err;
2428 /* If we are trying to remove a route, do not stop the
2429 * loop when ip6_route_del() fails (because next hop is
2430 * already gone), we should try to remove all next hops.
2431 */
2432 if (add) {
2433 /* If add fails, we should try to delete all
2434 * next hops that have been already added.
2435 */
2436 add = 0;
2437 goto beginning;
2438 }
2439 }
1a72418b
ND		 2440 /* Because each route is added like a single route we remove
2441 * this flag after the first nexthop (if there is a collision,
2442 * we have already fail to add the first nexthop:
2443 * fib6_add_rt2node() has reject it).
2444 */
2445 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
51ebd318
ND		 2446 rtnh = rtnh_next(rtnh, &remaining);
2447 }
2448
2449 return last_err;
2450}
2451
661d2967 2452static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2453{
86872cb5
TG		 2454 struct fib6_config cfg;
2455 int err;
1da177e4 2456
86872cb5
TG		 2457 err = rtm_to_fib6_config(skb, nlh, &cfg);
2458 if (err < 0)
2459 return err;
2460
51ebd318
ND		 2461 if (cfg.fc_mp)
2462 return ip6_route_multipath(&cfg, 0);
2463 else
2464 return ip6_route_del(&cfg);
1da177e4
LT		 2465}
2466
661d2967 2467static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2468{
86872cb5
TG		 2469 struct fib6_config cfg;
2470 int err;
1da177e4 2471
86872cb5
TG		 2472 err = rtm_to_fib6_config(skb, nlh, &cfg);
2473 if (err < 0)
2474 return err;
2475
51ebd318
ND		 2476 if (cfg.fc_mp)
2477 return ip6_route_multipath(&cfg, 1);
2478 else
2479 return ip6_route_add(&cfg);
1da177e4
LT		 2480}
2481
339bf98f
TG		 2482static inline size_t rt6_nlmsg_size(void)
2483{
2484 return NLMSG_ALIGN(sizeof(struct rtmsg))
2485 + nla_total_size(16) /* RTA_SRC */
2486 + nla_total_size(16) /* RTA_DST */
2487 + nla_total_size(16) /* RTA_GATEWAY */
2488 + nla_total_size(16) /* RTA_PREFSRC */
2489 + nla_total_size(4) /* RTA_TABLE */
2490 + nla_total_size(4) /* RTA_IIF */
2491 + nla_total_size(4) /* RTA_OIF */
2492 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2493 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
339bf98f
TG		 2494 + nla_total_size(sizeof(struct rta_cacheinfo));
2495}
2496
191cd582
BH		 2497static int rt6_fill_node(struct net *net,
2498 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2499 struct in6_addr *dst, struct in6_addr *src,
15e47304 2500 int iif, int type, u32 portid, u32 seq,
7bc570c8 2501 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2502{
2503 struct rtmsg *rtm;
2d7202bf 2504 struct nlmsghdr *nlh;
e3703b3d 2505 long expires;
9e762a4a 2506 u32 table;
1da177e4
LT		 2507
2508 if (prefix) { /* user wants prefix routes only */
2509 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2510 /* success since this is not a prefix route */
2511 return 1;
2512 }
2513 }
2514
15e47304 2515 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2516 if (!nlh)
26932566 2517 return -EMSGSIZE;
2d7202bf
TG		 2518
2519 rtm = nlmsg_data(nlh);
1da177e4
LT		 2520 rtm->rtm_family = AF_INET6;
2521 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2522 rtm->rtm_src_len = rt->rt6i_src.plen;
2523 rtm->rtm_tos = 0;
c71099ac 2524 if (rt->rt6i_table)
9e762a4a 2525 table = rt->rt6i_table->tb6_id;
c71099ac 2526 else
9e762a4a
PM		 2527 table = RT6_TABLE_UNSPEC;
2528 rtm->rtm_table = table;
c78679e8
DM		 2529 if (nla_put_u32(skb, RTA_TABLE, table))
2530 goto nla_put_failure;
ef2c7d7b
ND		 2531 if (rt->rt6i_flags & RTF_REJECT) {
2532 switch (rt->dst.error) {
2533 case -EINVAL:
2534 rtm->rtm_type = RTN_BLACKHOLE;
2535 break;
2536 case -EACCES:
2537 rtm->rtm_type = RTN_PROHIBIT;
2538 break;
b4949ab2
ND		 2539 case -EAGAIN:
2540 rtm->rtm_type = RTN_THROW;
2541 break;
ef2c7d7b
ND		 2542 default:
2543 rtm->rtm_type = RTN_UNREACHABLE;
2544 break;
2545 }
2546 }
38308473 2547 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2548 rtm->rtm_type = RTN_LOCAL;
d1918542 2549 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2550 rtm->rtm_type = RTN_LOCAL;
2551 else
2552 rtm->rtm_type = RTN_UNICAST;
2553 rtm->rtm_flags = 0;
2554 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2555 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2556 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2557 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2558 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2559 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2560 rtm->rtm_protocol = RTPROT_RA;
2561 else
2562 rtm->rtm_protocol = RTPROT_KERNEL;
2563 }
1da177e4 2564
38308473 2565 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2566 rtm->rtm_flags |= RTM_F_CLONED;
2567
2568 if (dst) {
c78679e8
DM		 2569 if (nla_put(skb, RTA_DST, 16, dst))
2570 goto nla_put_failure;
1ab1457c 2571 rtm->rtm_dst_len = 128;
1da177e4 2572 } else if (rtm->rtm_dst_len)
c78679e8
DM		 2573 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2574 goto nla_put_failure;
1da177e4
LT		 2575#ifdef CONFIG_IPV6_SUBTREES
2576 if (src) {
c78679e8
DM		 2577 if (nla_put(skb, RTA_SRC, 16, src))
2578 goto nla_put_failure;
1ab1457c 2579 rtm->rtm_src_len = 128;
c78679e8
DM		 2580 } else if (rtm->rtm_src_len &&
2581 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2582 goto nla_put_failure;
1da177e4 2583#endif
7bc570c8
YH		 2584 if (iif) {
2585#ifdef CONFIG_IPV6_MROUTE
2586 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2587 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2588 if (err <= 0) {
2589 if (!nowait) {
2590 if (err == 0)
2591 return 0;
2592 goto nla_put_failure;
2593 } else {
2594 if (err == -EMSGSIZE)
2595 goto nla_put_failure;
2596 }
2597 }
2598 } else
2599#endif
c78679e8
DM		 2600 if (nla_put_u32(skb, RTA_IIF, iif))
2601 goto nla_put_failure;
7bc570c8 2602 } else if (dst) {
1da177e4 2603 struct in6_addr saddr_buf;
c78679e8
DM		 2604 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2605 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2606 goto nla_put_failure;
1da177e4 2607 }
2d7202bf 2608
c3968a85
DW		 2609 if (rt->rt6i_prefsrc.plen) {
2610 struct in6_addr saddr_buf;
4e3fd7a0 2611 saddr_buf = rt->rt6i_prefsrc.addr;
c78679e8
DM		 2612 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2613 goto nla_put_failure;
c3968a85
DW		 2614 }
2615
defb3519 2616 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2d7202bf
TG		 2617 goto nla_put_failure;
2618
dd0cbf29
YH		 2619 if (rt->rt6i_flags & RTF_GATEWAY) {
2620 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
94f826b8 2621 goto nla_put_failure;
94f826b8 2622 }
2d7202bf 2623
c78679e8
DM		 2624 if (rt->dst.dev &&
2625 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2626 goto nla_put_failure;
2627 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2628 goto nla_put_failure;
8253947e
LW		 2629
2630 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2631
87a50699 2632 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2633 goto nla_put_failure;
2d7202bf
TG		 2634
2635 return nlmsg_end(skb, nlh);
2636
2637nla_put_failure:
26932566
PM		 2638 nlmsg_cancel(skb, nlh);
2639 return -EMSGSIZE;
1da177e4
LT		 2640}
2641
1b43af54 2642int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2643{
2644 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2645 int prefix;
2646
2d7202bf
TG		 2647 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2648 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2649 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2650 } else
2651 prefix = 0;
2652
191cd582
BH		 2653 return rt6_fill_node(arg->net,
2654 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2655 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2656 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2657}
2658
661d2967 2659static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh)
1da177e4 2660{
3b1e0a65 2661 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2662 struct nlattr *tb[RTA_MAX+1];
2663 struct rt6_info *rt;
1da177e4 2664 struct sk_buff *skb;
ab364a6f 2665 struct rtmsg *rtm;
4c9483b2 2666 struct flowi6 fl6;
72331bc0 2667 int err, iif = 0, oif = 0;
1da177e4 2668
ab364a6f
TG		 2669 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2670 if (err < 0)
2671 goto errout;
1da177e4 2672
ab364a6f 2673 err = -EINVAL;
4c9483b2 2674 memset(&fl6, 0, sizeof(fl6));
1da177e4 2675
ab364a6f
TG		 2676 if (tb[RTA_SRC]) {
2677 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2678 goto errout;
2679
4e3fd7a0 2680 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2681 }
2682
2683 if (tb[RTA_DST]) {
2684 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2685 goto errout;
2686
4e3fd7a0 2687 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2688 }
2689
2690 if (tb[RTA_IIF])
2691 iif = nla_get_u32(tb[RTA_IIF]);
2692
2693 if (tb[RTA_OIF])
72331bc0 2694 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4
LT		 2695
2696 if (iif) {
2697 struct net_device *dev;
72331bc0
SL		 2698 int flags = 0;
2699
5578689a 2700 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2701 if (!dev) {
2702 err = -ENODEV;
ab364a6f 2703 goto errout;
1da177e4 2704 }
72331bc0
SL		 2705
2706 fl6.flowi6_iif = iif;
2707
2708 if (!ipv6_addr_any(&fl6.saddr))
2709 flags |= RT6_LOOKUP_F_HAS_SADDR;
2710
2711 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2712 flags);
2713 } else {
2714 fl6.flowi6_oif = oif;
2715
2716 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2717 }
2718
ab364a6f 2719 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2720 if (!skb) {
94e187c0 2721 ip6_rt_put(rt);
ab364a6f
TG		 2722 err = -ENOBUFS;
2723 goto errout;
2724 }
1da177e4 2725
ab364a6f
TG		 2726 /* Reserve room for dummy headers, this skb can pass
2727 through good chunk of routing engine.
2728 */
459a98ed 2729 skb_reset_mac_header(skb);
ab364a6f 2730 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2731
d8d1f30b 2732 skb_dst_set(skb, &rt->dst);
1da177e4 2733
4c9483b2 2734 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2735 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2736 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2737 if (err < 0) {
ab364a6f
TG		 2738 kfree_skb(skb);
2739 goto errout;
1da177e4
LT		 2740 }
2741
15e47304 2742 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2743errout:
1da177e4 2744 return err;
1da177e4
LT		 2745}
2746
86872cb5 2747void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2748{
2749 struct sk_buff *skb;
5578689a 2750 struct net *net = info->nl_net;
528c4ceb
DL		 2751 u32 seq;
2752 int err;
2753
2754 err = -ENOBUFS;
38308473 2755 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2756
339bf98f 2757 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2758 if (!skb)
21713ebc
TG		 2759 goto errout;
2760
191cd582 2761 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2762 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2763 if (err < 0) {
2764 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2765 WARN_ON(err == -EMSGSIZE);
2766 kfree_skb(skb);
2767 goto errout;
2768 }
15e47304 2769 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2770 info->nlh, gfp_any());
2771 return;
21713ebc
TG		 2772errout:
2773 if (err < 0)
5578689a 2774 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2775}
2776
8ed67789 2777static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2778 unsigned long event, void *ptr)
8ed67789 2779{
351638e7 2780 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2781 struct net *net = dev_net(dev);
8ed67789
DL		 2782
2783 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2784 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2785 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2786#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2787 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2788 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2789 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2790 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2791#endif
2792 }
2793
2794 return NOTIFY_OK;
2795}
2796
1da177e4
LT		 2797/*
2798 * /proc
2799 */
2800
2801#ifdef CONFIG_PROC_FS
2802
1da177e4
LT		 2803struct rt6_proc_arg
2804{
2805 char *buffer;
2806 int offset;
2807 int length;
2808 int skip;
2809 int len;
2810};
2811
2812static int rt6_info_route(struct rt6_info *rt, void *p_arg)
2813{
33120b30 2814 struct seq_file *m = p_arg;
1da177e4 2815
4b7a4274 2816 seq_printf(m, "%pi6 %02x ", &rt->rt6i_dst.addr, rt->rt6i_dst.plen);
1da177e4
LT		 2817
2818#ifdef CONFIG_IPV6_SUBTREES
4b7a4274 2819 seq_printf(m, "%pi6 %02x ", &rt->rt6i_src.addr, rt->rt6i_src.plen);
1da177e4 2820#else
33120b30 2821 seq_puts(m, "00000000000000000000000000000000 00 ");
1da177e4 2822#endif
dd0cbf29
YH		 2823 if (rt->rt6i_flags & RTF_GATEWAY) {
2824 seq_printf(m, "%pi6", &rt->rt6i_gateway);
1da177e4 2825 } else {
33120b30 2826 seq_puts(m, "00000000000000000000000000000000");
1da177e4 2827 }
33120b30 2828 seq_printf(m, " %08x %08x %08x %08x %8s\n",
d8d1f30b
CG		 2829 rt->rt6i_metric, atomic_read(&rt->dst.__refcnt),
2830 rt->dst.__use, rt->rt6i_flags,
d1918542 2831 rt->dst.dev ? rt->dst.dev->name : "");
1da177e4
LT		 2832 return 0;
2833}
2834
33120b30 2835static int ipv6_route_show(struct seq_file *m, void *v)
1da177e4 2836{
f3db4851 2837 struct net *net = (struct net *)m->private;
32b293a5 2838 fib6_clean_all_ro(net, rt6_info_route, 0, m);
33120b30
AD		 2839 return 0;
2840}
1da177e4 2841
33120b30
AD		 2842static int ipv6_route_open(struct inode *inode, struct file *file)
2843{
de05c557 2844 return single_open_net(inode, file, ipv6_route_show);
f3db4851
DL		 2845}
2846
33120b30
AD		 2847static const struct file_operations ipv6_route_proc_fops = {
2848 .owner = THIS_MODULE,
2849 .open = ipv6_route_open,
2850 .read = seq_read,
2851 .llseek = seq_lseek,
b6fcbdb4 2852 .release = single_release_net,
33120b30
AD		 2853};
2854
1da177e4
LT		 2855static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2856{
69ddb805 2857 struct net *net = (struct net *)seq->private;
1da177e4 2858 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2859 net->ipv6.rt6_stats->fib_nodes,
2860 net->ipv6.rt6_stats->fib_route_nodes,
2861 net->ipv6.rt6_stats->fib_rt_alloc,
2862 net->ipv6.rt6_stats->fib_rt_entries,
2863 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2864 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2865 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2866
2867 return 0;
2868}
2869
2870static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2871{
de05c557 2872 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2873}
2874
9a32144e 2875static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2876 .owner = THIS_MODULE,
2877 .open = rt6_stats_seq_open,
2878 .read = seq_read,
2879 .llseek = seq_lseek,
b6fcbdb4 2880 .release = single_release_net,
1da177e4
LT		 2881};
2882#endif /* CONFIG_PROC_FS */
2883
2884#ifdef CONFIG_SYSCTL
2885
1da177e4 2886static
fe2c6338 2887int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 2888 void __user *buffer, size_t *lenp, loff_t *ppos)
2889{
c486da34
LAG		 2890 struct net *net;
2891 int delay;
2892 if (!write)
1da177e4 2893 return -EINVAL;
c486da34
LAG		 2894
2895 net = (struct net *)ctl->extra1;
2896 delay = net->ipv6.sysctl.flush_delay;
2897 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 2898 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 2899 return 0;
1da177e4
LT		 2900}
2901
fe2c6338 2902struct ctl_table ipv6_route_table_template[] = {
1ab1457c 2903 {
1da177e4 2904 .procname = "flush",
4990509f 2905 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2906 .maxlen = sizeof(int),
89c8b3a1 2907 .mode = 0200,
6d9f239a 2908 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 2909 },
2910 {
1da177e4 2911 .procname = "gc_thresh",
9a7ec3a9 2912 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2913 .maxlen = sizeof(int),
2914 .mode = 0644,
6d9f239a 2915 .proc_handler = proc_dointvec,
1da177e4
LT		 2916 },
2917 {
1da177e4 2918 .procname = "max_size",
4990509f 2919 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2920 .maxlen = sizeof(int),
2921 .mode = 0644,
6d9f239a 2922 .proc_handler = proc_dointvec,
1da177e4
LT		 2923 },
2924 {
1da177e4 2925 .procname = "gc_min_interval",
4990509f 2926 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2927 .maxlen = sizeof(int),
2928 .mode = 0644,
6d9f239a 2929 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2930 },
2931 {
1da177e4 2932 .procname = "gc_timeout",
4990509f 2933 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2934 .maxlen = sizeof(int),
2935 .mode = 0644,
6d9f239a 2936 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2937 },
2938 {
1da177e4 2939 .procname = "gc_interval",
4990509f 2940 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2941 .maxlen = sizeof(int),
2942 .mode = 0644,
6d9f239a 2943 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2944 },
2945 {
1da177e4 2946 .procname = "gc_elasticity",
4990509f 2947 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 2948 .maxlen = sizeof(int),
2949 .mode = 0644,
f3d3f616 2950 .proc_handler = proc_dointvec,
1da177e4
LT		 2951 },
2952 {
1da177e4 2953 .procname = "mtu_expires",
4990509f 2954 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 2955 .maxlen = sizeof(int),
2956 .mode = 0644,
6d9f239a 2957 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2958 },
2959 {
1da177e4 2960 .procname = "min_adv_mss",
4990509f 2961 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 2962 .maxlen = sizeof(int),
2963 .mode = 0644,
f3d3f616 2964 .proc_handler = proc_dointvec,
1da177e4
LT		 2965 },
2966 {
1da177e4 2967 .procname = "gc_min_interval_ms",
4990509f 2968 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2969 .maxlen = sizeof(int),
2970 .mode = 0644,
6d9f239a 2971 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 2972 },
f8572d8f 2973 { }
1da177e4
LT		 2974};
2975
2c8c1e72 2976struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 2977{
2978 struct ctl_table *table;
2979
2980 table = kmemdup(ipv6_route_table_template,
2981 sizeof(ipv6_route_table_template),
2982 GFP_KERNEL);
5ee09105
YH		 2983
2984 if (table) {
2985 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 2986 table[0].extra1 = net;
86393e52 2987 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 2988 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2989 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2990 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2991 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2992 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2993 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2994 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 2995 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 2996
2997 /* Don't export sysctls to unprivileged users */
2998 if (net->user_ns != &init_user_ns)
2999 table[0].procname = NULL;
5ee09105
YH		 3000 }
3001
760f2d01
DL		 3002 return table;
3003}
1da177e4
LT		 3004#endif
3005
2c8c1e72 3006static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3007{
633d424b 3008 int ret = -ENOMEM;
8ed67789 3009
86393e52
AD		 3010 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3011 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3012
fc66f95c
ED		 3013 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3014 goto out_ip6_dst_ops;
3015
8ed67789
DL		 3016 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3017 sizeof(*net->ipv6.ip6_null_entry),
3018 GFP_KERNEL);
3019 if (!net->ipv6.ip6_null_entry)
fc66f95c 3020 goto out_ip6_dst_entries;
d8d1f30b 3021 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3022 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3023 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3024 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3025 ip6_template_metrics, true);
8ed67789
DL		 3026
3027#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3028 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3029 sizeof(*net->ipv6.ip6_prohibit_entry),
3030 GFP_KERNEL);
68fffc67
PZ		 3031 if (!net->ipv6.ip6_prohibit_entry)
3032 goto out_ip6_null_entry;
d8d1f30b 3033 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3034 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3035 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3036 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3037 ip6_template_metrics, true);
8ed67789
DL		 3038
3039 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3040 sizeof(*net->ipv6.ip6_blk_hole_entry),
3041 GFP_KERNEL);
68fffc67
PZ		 3042 if (!net->ipv6.ip6_blk_hole_entry)
3043 goto out_ip6_prohibit_entry;
d8d1f30b 3044 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3045 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3046 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3047 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3048 ip6_template_metrics, true);
8ed67789
DL		 3049#endif
3050
b339a47c
PZ		 3051 net->ipv6.sysctl.flush_delay = 0;
3052 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3053 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3054 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3055 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3056 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3057 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3058 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3059
6891a346
BT		 3060 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3061
8ed67789
DL		 3062 ret = 0;
3063out:
3064 return ret;
f2fc6a54 3065
68fffc67
PZ		 3066#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3067out_ip6_prohibit_entry:
3068 kfree(net->ipv6.ip6_prohibit_entry);
3069out_ip6_null_entry:
3070 kfree(net->ipv6.ip6_null_entry);
3071#endif
fc66f95c
ED		 3072out_ip6_dst_entries:
3073 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3074out_ip6_dst_ops:
f2fc6a54 3075 goto out;
cdb18761
DL		 3076}
3077
2c8c1e72 3078static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3079{
8ed67789
DL		 3080 kfree(net->ipv6.ip6_null_entry);
3081#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3082 kfree(net->ipv6.ip6_prohibit_entry);
3083 kfree(net->ipv6.ip6_blk_hole_entry);
3084#endif
41bb78b4 3085 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3086}
3087
d189634e
TG		 3088static int __net_init ip6_route_net_init_late(struct net *net)
3089{
3090#ifdef CONFIG_PROC_FS
d4beaa66
G		 3091 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3092 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3093#endif
3094 return 0;
3095}
3096
3097static void __net_exit ip6_route_net_exit_late(struct net *net)
3098{
3099#ifdef CONFIG_PROC_FS
ece31ffd
G		 3100 remove_proc_entry("ipv6_route", net->proc_net);
3101 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3102#endif
3103}
3104
cdb18761
DL		 3105static struct pernet_operations ip6_route_net_ops = {
3106 .init = ip6_route_net_init,
3107 .exit = ip6_route_net_exit,
3108};
3109
c3426b47
DM		 3110static int __net_init ipv6_inetpeer_init(struct net *net)
3111{
3112 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3113
3114 if (!bp)
3115 return -ENOMEM;
3116 inet_peer_base_init(bp);
3117 net->ipv6.peers = bp;
3118 return 0;
3119}
3120
3121static void __net_exit ipv6_inetpeer_exit(struct net *net)
3122{
3123 struct inet_peer_base *bp = net->ipv6.peers;
3124
3125 net->ipv6.peers = NULL;
56a6b248 3126 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3127 kfree(bp);
3128}
3129
2b823f72 3130static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3131 .init = ipv6_inetpeer_init,
3132 .exit = ipv6_inetpeer_exit,
3133};
3134
d189634e
TG		 3135static struct pernet_operations ip6_route_net_late_ops = {
3136 .init = ip6_route_net_init_late,
3137 .exit = ip6_route_net_exit_late,
3138};
3139
8ed67789
DL		 3140static struct notifier_block ip6_route_dev_notifier = {
3141 .notifier_call = ip6_route_dev_notify,
3142 .priority = 0,
3143};
3144
433d49c3 3145int __init ip6_route_init(void)
1da177e4 3146{
433d49c3
DL		 3147 int ret;
3148
9a7ec3a9
DL		 3149 ret = -ENOMEM;
3150 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3151 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3152 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3153 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3154 goto out;
14e50e57 3155
fc66f95c 3156 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3157 if (ret)
bdb3289f 3158 goto out_kmem_cache;
bdb3289f 3159
c3426b47
DM		 3160 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3161 if (ret)
e8803b6c 3162 goto out_dst_entries;
2a0c451a 3163
7e52b33b
DM		 3164 ret = register_pernet_subsys(&ip6_route_net_ops);
3165 if (ret)
3166 goto out_register_inetpeer;
c3426b47 3167
5dc121e9
AE		 3168 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3169
8ed67789
DL		 3170 /* Registering of the loopback is done before this portion of code,
3171 * the loopback reference in rt6_info will not be taken, do it
3172 * manually for init_net */
d8d1f30b 3173 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3174 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3175 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3176 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3177 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3178 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3179 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3180 #endif
e8803b6c 3181 ret = fib6_init();
433d49c3 3182 if (ret)
8ed67789 3183 goto out_register_subsys;
433d49c3 3184
433d49c3
DL		 3185 ret = xfrm6_init();
3186 if (ret)
e8803b6c 3187 goto out_fib6_init;
c35b7e72 3188
433d49c3
DL		 3189 ret = fib6_rules_init();
3190 if (ret)
3191 goto xfrm6_init;
7e5449c2 3192
d189634e
TG		 3193 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3194 if (ret)
3195 goto fib6_rules_init;
3196
433d49c3 3197 ret = -ENOBUFS;
c7ac8679
GR		 3198 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3199 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3200 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3201 goto out_register_late_subsys;
c127ea2c 3202
8ed67789 3203 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3204 if (ret)
d189634e 3205 goto out_register_late_subsys;
8ed67789 3206
433d49c3
DL		 3207out:
3208 return ret;
3209
d189634e
TG		 3210out_register_late_subsys:
3211 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3212fib6_rules_init:
433d49c3
DL		 3213 fib6_rules_cleanup();
3214xfrm6_init:
433d49c3 3215 xfrm6_fini();
2a0c451a
TG		 3216out_fib6_init:
3217 fib6_gc_cleanup();
8ed67789
DL		 3218out_register_subsys:
3219 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3220out_register_inetpeer:
3221 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3222out_dst_entries:
3223 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3224out_kmem_cache:
f2fc6a54 3225 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3226 goto out;
1da177e4
LT		 3227}
3228
3229void ip6_route_cleanup(void)
3230{
8ed67789 3231 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3232 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3233 fib6_rules_cleanup();
1da177e4 3234 xfrm6_fini();
1da177e4 3235 fib6_gc_cleanup();
c3426b47 3236 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3237 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3238 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3239 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3240}
Linux kernel - Deliverables
This page took 2.074403 seconds and 5 git commands to generate.