projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
ipv6: White-space cleansing : Line Layouts
[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 68enum rt6_nud_state {
7e980569
JB		 69 RT6_NUD_FAIL_HARD = -3,
70 RT6_NUD_FAIL_PROBE = -2,
71 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 72 RT6_NUD_SUCCEED = 1
73};
74
1716a961 75static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 76 const struct in6_addr *dest);
1da177e4 77static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 78static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 79static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 80static struct dst_entry *ip6_negative_advice(struct dst_entry *);
81static void ip6_dst_destroy(struct dst_entry *);
82static void ip6_dst_ifdown(struct dst_entry *,
83 struct net_device *dev, int how);
569d3645 84static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 85
86static int ip6_pkt_discard(struct sk_buff *skb);
aad88724 87static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb);
7150aede 88static int ip6_pkt_prohibit(struct sk_buff *skb);
aad88724 89static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb);
1da177e4 90static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 91static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
92 struct sk_buff *skb, u32 mtu);
93static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
94 struct sk_buff *skb);
52bd4c0c 95static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 96
70ceb4f5 97#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 98static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 99 const struct in6_addr *prefix, int prefixlen,
100 const struct in6_addr *gwaddr, int ifindex,
95c96174 101 unsigned int pref);
efa2cea0 102static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 103 const struct in6_addr *prefix, int prefixlen,
104 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 105#endif
106
e8243534 107static void rt6_bind_peer(struct rt6_info *rt, int create)
108{
109 struct inet_peer_base *base;
110 struct inet_peer *peer;
111
112 base = inetpeer_base_ptr(rt->_rt6i_peer);
113 if (!base)
114 return;
115
116 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create);
117 if (peer) {
118 if (!rt6_set_peer(rt, peer))
119 inet_putpeer(peer);
120 }
121}
122
123static struct inet_peer *__rt6_get_peer(struct rt6_info *rt, int create)
124{
125 if (rt6_has_peer(rt))
126 return rt6_peer_ptr(rt);
127
128 rt6_bind_peer(rt, create);
129 return (rt6_has_peer(rt) ? rt6_peer_ptr(rt) : NULL);
130}
131
132static struct inet_peer *rt6_get_peer_create(struct rt6_info *rt)
133{
134 return __rt6_get_peer(rt, 1);
135}
136
06582540
DM		 137static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
138{
139 struct rt6_info *rt = (struct rt6_info *) dst;
140 struct inet_peer *peer;
141 u32 *p = NULL;
142
8e2ec639
YZ		 143 if (!(rt->dst.flags & DST_HOST))
144 return NULL;
145
fbfe95a4 146 peer = rt6_get_peer_create(rt);
06582540
DM		 147 if (peer) {
148 u32 *old_p = __DST_METRICS_PTR(old);
149 unsigned long prev, new;
150
151 p = peer->metrics;
e5fd387a
MK		 152 if (inet_metrics_new(peer) ||
153 (old & DST_METRICS_FORCE_OVERWRITE))
06582540
DM		 154 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
155
156 new = (unsigned long) p;
157 prev = cmpxchg(&dst->_metrics, old, new);
158
159 if (prev != old) {
160 p = __DST_METRICS_PTR(prev);
161 if (prev & DST_METRICS_READ_ONLY)
162 p = NULL;
163 }
164 }
165 return p;
166}
167
f894cbf8
DM		 168static inline const void *choose_neigh_daddr(struct rt6_info *rt,
169 struct sk_buff *skb,
170 const void *daddr)
39232973
DM		 171{
172 struct in6_addr *p = &rt->rt6i_gateway;
173
a7563f34 174 if (!ipv6_addr_any(p))
39232973 175 return (const void *) p;
f894cbf8
DM		 176 else if (skb)
177 return &ipv6_hdr(skb)->daddr;
39232973
DM		 178 return daddr;
179}
180
f894cbf8
DM		 181static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
182 struct sk_buff *skb,
183 const void *daddr)
d3aaeb38 184{
39232973
DM		 185 struct rt6_info *rt = (struct rt6_info *) dst;
186 struct neighbour *n;
187
f894cbf8 188 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 189 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 190 if (n)
191 return n;
192 return neigh_create(&nd_tbl, daddr, dst->dev);
193}
194
9a7ec3a9 195static struct dst_ops ip6_dst_ops_template = {
1da177e4 196 .family = AF_INET6,
09640e63 197 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT		 198 .gc = ip6_dst_gc,
199 .gc_thresh = 1024,
200 .check = ip6_dst_check,
0dbaee3b 201 .default_advmss = ip6_default_advmss,
ebb762f2 202 .mtu = ip6_mtu,
06582540 203 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 204 .destroy = ip6_dst_destroy,
205 .ifdown = ip6_dst_ifdown,
206 .negative_advice = ip6_negative_advice,
207 .link_failure = ip6_link_failure,
208 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 209 .redirect = rt6_do_redirect,
1ac06e03 210 .local_out = __ip6_local_out,
d3aaeb38 211 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 212};
213
ebb762f2 214static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 215{
618f9bc7
SK		 216 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
217
218 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 219}
220
6700c270
DM		 221static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
222 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 223{
224}
225
6700c270
DM		 226static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
227 struct sk_buff *skb)
b587ee3b
DM		 228{
229}
230
0972ddb2
HB		 231static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
232 unsigned long old)
233{
234 return NULL;
235}
236
14e50e57
DM		 237static struct dst_ops ip6_dst_blackhole_ops = {
238 .family = AF_INET6,
09640e63 239 .protocol = cpu_to_be16(ETH_P_IPV6),
14e50e57
DM		 240 .destroy = ip6_dst_destroy,
241 .check = ip6_dst_check,
ebb762f2 242 .mtu = ip6_blackhole_mtu,
214f45c9 243 .default_advmss = ip6_default_advmss,
14e50e57 244 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 245 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 246 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 247 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 248};
249
62fa8a84 250static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 251 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 252};
253
fb0af4c7 254static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 255 .dst = {
256 .__refcnt = ATOMIC_INIT(1),
257 .__use = 1,
2c20cbd7 258 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 259 .error = -ENETUNREACH,
d8d1f30b
CG		 260 .input = ip6_pkt_discard,
261 .output = ip6_pkt_discard_out,
1da177e4
LT		 262 },
263 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 264 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 265 .rt6i_metric = ~(u32) 0,
266 .rt6i_ref = ATOMIC_INIT(1),
267};
268
101367c2
TG		 269#ifdef CONFIG_IPV6_MULTIPLE_TABLES
270
fb0af4c7 271static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 272 .dst = {
273 .__refcnt = ATOMIC_INIT(1),
274 .__use = 1,
2c20cbd7 275 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 276 .error = -EACCES,
d8d1f30b
CG		 277 .input = ip6_pkt_prohibit,
278 .output = ip6_pkt_prohibit_out,
101367c2
TG		 279 },
280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 281 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 282 .rt6i_metric = ~(u32) 0,
283 .rt6i_ref = ATOMIC_INIT(1),
284};
285
fb0af4c7 286static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 287 .dst = {
288 .__refcnt = ATOMIC_INIT(1),
289 .__use = 1,
2c20cbd7 290 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 291 .error = -EINVAL,
d8d1f30b 292 .input = dst_discard,
aad88724 293 .output = dst_discard_sk,
101367c2
TG		 294 },
295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 296 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 297 .rt6i_metric = ~(u32) 0,
298 .rt6i_ref = ATOMIC_INIT(1),
299};
300
301#endif
302
1da177e4 303/* allocate dst with ip6_dst_ops */
97bab73f 304static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 305 struct net_device *dev,
8b96d22d
DM		 306 int flags,
307 struct fib6_table *table)
1da177e4 308{
97bab73f 309 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 310 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 311
97bab73f 312 if (rt) {
8104891b
SK		 313 struct dst_entry *dst = &rt->dst;
314
315 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
8b96d22d 316 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
ca4c3fc2 317 rt->rt6i_genid = rt_genid_ipv6(net);
51ebd318 318 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 319 }
cf911662 320 return rt;
1da177e4
LT		 321}
322
323static void ip6_dst_destroy(struct dst_entry *dst)
324{
325 struct rt6_info *rt = (struct rt6_info *)dst;
326 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 327 struct dst_entry *from = dst->from;
1da177e4 328
8e2ec639
YZ		 329 if (!(rt->dst.flags & DST_HOST))
330 dst_destroy_metrics_generic(dst);
331
38308473 332 if (idev) {
1da177e4
LT		 333 rt->rt6i_idev = NULL;
334 in6_dev_put(idev);
1ab1457c 335 }
1716a961 336
ecd98837
YH		 337 dst->from = NULL;
338 dst_release(from);
1716a961 339
97bab73f
DM		 340 if (rt6_has_peer(rt)) {
341 struct inet_peer *peer = rt6_peer_ptr(rt);
b3419363
DM		 342 inet_putpeer(peer);
343 }
344}
345
1da177e4
LT		 346static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
347 int how)
348{
349 struct rt6_info *rt = (struct rt6_info *)dst;
350 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 351 struct net_device *loopback_dev =
c346dca1 352 dev_net(dev)->loopback_dev;
1da177e4 353
97cac082
DM		 354 if (dev != loopback_dev) {
355 if (idev && idev->dev == dev) {
356 struct inet6_dev *loopback_idev =
357 in6_dev_get(loopback_dev);
358 if (loopback_idev) {
359 rt->rt6i_idev = loopback_idev;
360 in6_dev_put(idev);
361 }
362 }
1da177e4
LT		 363 }
364}
365
a50feda5 366static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 367{
1716a961
G		 368 if (rt->rt6i_flags & RTF_EXPIRES) {
369 if (time_after(jiffies, rt->dst.expires))
a50feda5 370 return true;
1716a961 371 } else if (rt->dst.from) {
3fd91fb3 372 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 373 }
a50feda5 374 return false;
1da177e4
LT		 375}
376
51ebd318
ND		 377/* Multipath route selection:
378 * Hash based function using packet header and flowlabel.
379 * Adapted from fib_info_hashfn()
380 */
381static int rt6_info_hash_nhsfn(unsigned int candidate_count,
382 const struct flowi6 *fl6)
383{
384 unsigned int val = fl6->flowi6_proto;
385
c08977bb
YH		 386 val ^= ipv6_addr_hash(&fl6->daddr);
387 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 388
389 /* Work only if this not encapsulated */
390 switch (fl6->flowi6_proto) {
391 case IPPROTO_UDP:
392 case IPPROTO_TCP:
393 case IPPROTO_SCTP:
b3ce5ae1
ND		 394 val ^= (__force u16)fl6->fl6_sport;
395 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 396 break;
397
398 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 399 val ^= (__force u16)fl6->fl6_icmp_type;
400 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 401 break;
402 }
403 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 404 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 405
406 /* Perhaps, we need to tune, this function? */
407 val = val ^ (val >> 7) ^ (val >> 12);
408 return val % candidate_count;
409}
410
411static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 412 struct flowi6 *fl6, int oif,
413 int strict)
51ebd318
ND		 414{
415 struct rt6_info *sibling, *next_sibling;
416 int route_choosen;
417
418 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
419 /* Don't change the route, if route_choosen == 0
420 * (siblings does not include ourself)
421 */
422 if (route_choosen)
423 list_for_each_entry_safe(sibling, next_sibling,
424 &match->rt6i_siblings, rt6i_siblings) {
425 route_choosen--;
426 if (route_choosen == 0) {
52bd4c0c
ND		 427 if (rt6_score_route(sibling, oif, strict) < 0)
428 break;
51ebd318
ND		 429 match = sibling;
430 break;
431 }
432 }
433 return match;
434}
435
1da177e4 436/*
c71099ac 437 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 438 */
439
8ed67789
DL		 440static inline struct rt6_info *rt6_device_match(struct net *net,
441 struct rt6_info *rt,
b71d1d42 442 const struct in6_addr *saddr,
1da177e4 443 int oif,
d420895e 444 int flags)
1da177e4
LT		 445{
446 struct rt6_info *local = NULL;
447 struct rt6_info *sprt;
448
dd3abc4e
YH		 449 if (!oif && ipv6_addr_any(saddr))
450 goto out;
451
d8d1f30b 452 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 453 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 454
455 if (oif) {
1da177e4
LT		 456 if (dev->ifindex == oif)
457 return sprt;
458 if (dev->flags & IFF_LOOPBACK) {
38308473 459 if (!sprt->rt6i_idev ||
1da177e4 460 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 461 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 462 continue;
1ab1457c 463 if (local && (!oif ||
1da177e4
LT		 464 local->rt6i_idev->dev->ifindex == oif))
465 continue;
466 }
467 local = sprt;
468 }
dd3abc4e
YH		 469 } else {
470 if (ipv6_chk_addr(net, saddr, dev,
471 flags & RT6_LOOKUP_F_IFACE))
472 return sprt;
1da177e4 473 }
dd3abc4e 474 }
1da177e4 475
dd3abc4e 476 if (oif) {
1da177e4
LT		 477 if (local)
478 return local;
479
d420895e 480 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 481 return net->ipv6.ip6_null_entry;
1da177e4 482 }
dd3abc4e 483out:
1da177e4
LT		 484 return rt;
485}
486
27097255 487#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 488struct __rt6_probe_work {
489 struct work_struct work;
490 struct in6_addr target;
491 struct net_device *dev;
492};
493
494static void rt6_probe_deferred(struct work_struct *w)
495{
496 struct in6_addr mcaddr;
497 struct __rt6_probe_work *work =
498 container_of(w, struct __rt6_probe_work, work);
499
500 addrconf_addr_solict_mult(&work->target, &mcaddr);
501 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
502 dev_put(work->dev);
503 kfree(w);
504}
505
27097255
YH		 506static void rt6_probe(struct rt6_info *rt)
507{
f2c31e32 508 struct neighbour *neigh;
27097255
YH		 509 /*
510 * Okay, this does not seem to be appropriate
511 * for now, however, we need to check if it
512 * is really so; aka Router Reachability Probing.
513 *
514 * Router Reachability Probe MUST be rate-limited
515 * to no more than one per minute.
516 */
2152caea 517 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 518 return;
2152caea
YH		 519 rcu_read_lock_bh();
520 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
521 if (neigh) {
522 write_lock(&neigh->lock);
523 if (neigh->nud_state & NUD_VALID)
524 goto out;
7ff74a59 525 }
2152caea
YH		 526
527 if (!neigh ||
52e16356 528 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
c2f17e82 529 struct __rt6_probe_work *work;
27097255 530
c2f17e82
HFS		 531 work = kmalloc(sizeof(*work), GFP_ATOMIC);
532
533 if (neigh && work)
7e980569 534 __neigh_set_probe_once(neigh);
c2f17e82
HFS		 535
536 if (neigh)
2152caea
YH		 537 write_unlock(&neigh->lock);
538
c2f17e82
HFS		 539 if (work) {
540 INIT_WORK(&work->work, rt6_probe_deferred);
541 work->target = rt->rt6i_gateway;
542 dev_hold(rt->dst.dev);
543 work->dev = rt->dst.dev;
544 schedule_work(&work->work);
545 }
f2c31e32 546 } else {
2152caea
YH		 547out:
548 write_unlock(&neigh->lock);
f2c31e32 549 }
2152caea 550 rcu_read_unlock_bh();
27097255
YH		 551}
552#else
553static inline void rt6_probe(struct rt6_info *rt)
554{
27097255
YH		 555}
556#endif
557
1da177e4 558/*
554cfb7e 559 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 560 */
b6f99a21 561static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 562{
d1918542 563 struct net_device *dev = rt->dst.dev;
161980f4 564 if (!oif || dev->ifindex == oif)
554cfb7e 565 return 2;
161980f4
DM		 566 if ((dev->flags & IFF_LOOPBACK) &&
567 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
568 return 1;
569 return 0;
554cfb7e 570}
1da177e4 571
afc154e9 572static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 573{
f2c31e32 574 struct neighbour *neigh;
afc154e9 575 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 576
4d0c5911
YH		 577 if (rt->rt6i_flags & RTF_NONEXTHOP ||
578 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 579 return RT6_NUD_SUCCEED;
145a3621
YH		 580
581 rcu_read_lock_bh();
582 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
583 if (neigh) {
584 read_lock(&neigh->lock);
554cfb7e 585 if (neigh->nud_state & NUD_VALID)
afc154e9 586 ret = RT6_NUD_SUCCEED;
398bcbeb 587#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 588 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 589 ret = RT6_NUD_SUCCEED;
7e980569
JB		 590 else
591 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 592#endif
145a3621 593 read_unlock(&neigh->lock);
afc154e9
HFS		 594 } else {
595 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 596 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 597 }
145a3621
YH		 598 rcu_read_unlock_bh();
599
a5a81f0b 600 return ret;
1da177e4
LT		 601}
602
554cfb7e
YH		 603static int rt6_score_route(struct rt6_info *rt, int oif,
604 int strict)
1da177e4 605{
a5a81f0b 606 int m;
1ab1457c 607
4d0c5911 608 m = rt6_check_dev(rt, oif);
77d16f45 609 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 610 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 611#ifdef CONFIG_IPV6_ROUTER_PREF
612 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
613#endif
afc154e9
HFS		 614 if (strict & RT6_LOOKUP_F_REACHABLE) {
615 int n = rt6_check_neigh(rt);
616 if (n < 0)
617 return n;
618 }
554cfb7e
YH		 619 return m;
620}
621
f11e6659 622static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 623 int *mpri, struct rt6_info *match,
624 bool *do_rr)
554cfb7e 625{
f11e6659 626 int m;
afc154e9 627 bool match_do_rr = false;
f11e6659
DM		 628
629 if (rt6_check_expired(rt))
630 goto out;
631
632 m = rt6_score_route(rt, oif, strict);
7e980569 633 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 634 match_do_rr = true;
635 m = 0; /* lowest valid score */
7e980569 636 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 637 goto out;
afc154e9
HFS		 638 }
639
640 if (strict & RT6_LOOKUP_F_REACHABLE)
641 rt6_probe(rt);
f11e6659 642
7e980569 643 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 644 if (m > *mpri) {
afc154e9 645 *do_rr = match_do_rr;
f11e6659
DM		 646 *mpri = m;
647 match = rt;
f11e6659 648 }
f11e6659
DM		 649out:
650 return match;
651}
652
653static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
654 struct rt6_info *rr_head,
afc154e9
HFS		 655 u32 metric, int oif, int strict,
656 bool *do_rr)
f11e6659
DM		 657{
658 struct rt6_info *rt, *match;
554cfb7e 659 int mpri = -1;
1da177e4 660
f11e6659
DM		 661 match = NULL;
662 for (rt = rr_head; rt && rt->rt6i_metric == metric;
d8d1f30b 663 rt = rt->dst.rt6_next)
afc154e9 664 match = find_match(rt, oif, strict, &mpri, match, do_rr);
f11e6659 665 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
d8d1f30b 666 rt = rt->dst.rt6_next)
afc154e9 667 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 668
f11e6659
DM		 669 return match;
670}
1da177e4 671
f11e6659
DM		 672static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
673{
674 struct rt6_info *match, *rt0;
8ed67789 675 struct net *net;
afc154e9 676 bool do_rr = false;
1da177e4 677
f11e6659
DM		 678 rt0 = fn->rr_ptr;
679 if (!rt0)
680 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 681
afc154e9
HFS		 682 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
683 &do_rr);
1da177e4 684
afc154e9 685 if (do_rr) {
d8d1f30b 686 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 687
554cfb7e 688 /* no entries matched; do round-robin */
f11e6659
DM		 689 if (!next || next->rt6i_metric != rt0->rt6i_metric)
690 next = fn->leaf;
691
692 if (next != rt0)
693 fn->rr_ptr = next;
1da177e4 694 }
1da177e4 695
d1918542 696 net = dev_net(rt0->dst.dev);
a02cec21 697 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 698}
699
70ceb4f5
YH		 700#ifdef CONFIG_IPV6_ROUTE_INFO
701int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 702 const struct in6_addr *gwaddr)
70ceb4f5 703{
c346dca1 704 struct net *net = dev_net(dev);
70ceb4f5
YH		 705 struct route_info *rinfo = (struct route_info *) opt;
706 struct in6_addr prefix_buf, *prefix;
707 unsigned int pref;
4bed72e4 708 unsigned long lifetime;
70ceb4f5
YH		 709 struct rt6_info *rt;
710
711 if (len < sizeof(struct route_info)) {
712 return -EINVAL;
713 }
714
715 /* Sanity check for prefix_len and length */
716 if (rinfo->length > 3) {
717 return -EINVAL;
718 } else if (rinfo->prefix_len > 128) {
719 return -EINVAL;
720 } else if (rinfo->prefix_len > 64) {
721 if (rinfo->length < 2) {
722 return -EINVAL;
723 }
724 } else if (rinfo->prefix_len > 0) {
725 if (rinfo->length < 1) {
726 return -EINVAL;
727 }
728 }
729
730 pref = rinfo->route_pref;
731 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 732 return -EINVAL;
70ceb4f5 733
4bed72e4 734 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 735
736 if (rinfo->length == 3)
737 prefix = (struct in6_addr *)rinfo->prefix;
738 else {
739 /* this function is safe */
740 ipv6_addr_prefix(&prefix_buf,
741 (struct in6_addr *)rinfo->prefix,
742 rinfo->prefix_len);
743 prefix = &prefix_buf;
744 }
745
f104a567
DJ		 746 if (rinfo->prefix_len == 0)
747 rt = rt6_get_dflt_router(gwaddr, dev);
748 else
749 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
750 gwaddr, dev->ifindex);
70ceb4f5
YH		 751
752 if (rt && !lifetime) {
e0a1ad73 753 ip6_del_rt(rt);
70ceb4f5
YH		 754 rt = NULL;
755 }
756
757 if (!rt && lifetime)
efa2cea0 758 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 759 pref);
760 else if (rt)
761 rt->rt6i_flags = RTF_ROUTEINFO |
762 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
763
764 if (rt) {
1716a961
G		 765 if (!addrconf_finite_timeout(lifetime))
766 rt6_clean_expires(rt);
767 else
768 rt6_set_expires(rt, jiffies + HZ * lifetime);
769
94e187c0 770 ip6_rt_put(rt);
70ceb4f5
YH		 771 }
772 return 0;
773}
774#endif
775
8ed67789 776#define BACKTRACK(__net, saddr) \
982f56f3 777do { \
8ed67789 778 if (rt == __net->ipv6.ip6_null_entry) { \
982f56f3 779 struct fib6_node *pn; \
e0eda7bb 780 while (1) { \
982f56f3
YH		 781 if (fn->fn_flags & RTN_TL_ROOT) \
782 goto out; \
783 pn = fn->parent; \
784 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
8bce65b9 785 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
982f56f3
YH		 786 else \
787 fn = pn; \
788 if (fn->fn_flags & RTN_RTINFO) \
789 goto restart; \
c71099ac 790 } \
c71099ac 791 } \
38308473 792} while (0)
c71099ac 793
8ed67789
DL		 794static struct rt6_info *ip6_pol_route_lookup(struct net *net,
795 struct fib6_table *table,
4c9483b2 796 struct flowi6 *fl6, int flags)
1da177e4
LT		 797{
798 struct fib6_node *fn;
799 struct rt6_info *rt;
800
c71099ac 801 read_lock_bh(&table->tb6_lock);
4c9483b2 802 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 803restart:
804 rt = fn->leaf;
4c9483b2 805 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 806 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 807 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
4c9483b2 808 BACKTRACK(net, &fl6->saddr);
c71099ac 809out:
d8d1f30b 810 dst_use(&rt->dst, jiffies);
c71099ac 811 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 812 return rt;
813
814}
815
67ba4152 816struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 817 int flags)
818{
819 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
820}
821EXPORT_SYMBOL_GPL(ip6_route_lookup);
822
9acd9f3a
YH		 823struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
824 const struct in6_addr *saddr, int oif, int strict)
c71099ac 825{
4c9483b2
DM		 826 struct flowi6 fl6 = {
827 .flowi6_oif = oif,
828 .daddr = *daddr,
c71099ac
TG		 829 };
830 struct dst_entry *dst;
77d16f45 831 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 832
adaa70bb 833 if (saddr) {
4c9483b2 834 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 835 flags |= RT6_LOOKUP_F_HAS_SADDR;
836 }
837
4c9483b2 838 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 839 if (dst->error == 0)
840 return (struct rt6_info *) dst;
841
842 dst_release(dst);
843
1da177e4
LT		 844 return NULL;
845}
846
7159039a
YH		 847EXPORT_SYMBOL(rt6_lookup);
848
c71099ac 849/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 850 It takes new route entry, the addition fails by any reason the
851 route is freed. In any case, if caller does not hold it, it may
852 be destroyed.
853 */
854
e5fd387a
MK		 855static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
856 struct nlattr *mx, int mx_len)
1da177e4
LT		 857{
858 int err;
c71099ac 859 struct fib6_table *table;
1da177e4 860
c71099ac
TG		 861 table = rt->rt6i_table;
862 write_lock_bh(&table->tb6_lock);
e5fd387a 863 err = fib6_add(&table->tb6_root, rt, info, mx, mx_len);
c71099ac 864 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 865
866 return err;
867}
868
40e22e8f
TG		 869int ip6_ins_rt(struct rt6_info *rt)
870{
4d1169c1 871 struct nl_info info = {
d1918542 872 .nl_net = dev_net(rt->dst.dev),
4d1169c1 873 };
e5fd387a 874 return __ip6_ins_rt(rt, &info, NULL, 0);
40e22e8f
TG		 875}
876
1716a961 877static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
21efcfa0 878 const struct in6_addr *daddr,
b71d1d42 879 const struct in6_addr *saddr)
1da177e4 880{
1da177e4
LT		 881 struct rt6_info *rt;
882
883 /*
884 * Clone the route.
885 */
886
21efcfa0 887 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 888
889 if (rt) {
249a3630
DJ		 890 if (ort->rt6i_dst.plen != 128 &&
891 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
892 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 893
1da177e4 894 rt->rt6i_flags |= RTF_CACHE;
1da177e4
LT		 895
896#ifdef CONFIG_IPV6_SUBTREES
897 if (rt->rt6i_src.plen && saddr) {
4e3fd7a0 898 rt->rt6i_src.addr = *saddr;
1da177e4
LT		 899 rt->rt6i_src.plen = 128;
900 }
901#endif
95a9a5ba 902 }
1da177e4 903
95a9a5ba
YH		 904 return rt;
905}
1da177e4 906
21efcfa0
ED		 907static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
908 const struct in6_addr *daddr)
299d9939 909{
21efcfa0
ED		 910 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
911
887c95cc 912 if (rt)
299d9939 913 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 914 return rt;
915}
916
8ed67789 917static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 918 struct flowi6 *fl6, int flags)
1da177e4
LT		 919{
920 struct fib6_node *fn;
519fbd87 921 struct rt6_info *rt, *nrt;
c71099ac 922 int strict = 0;
1da177e4 923 int attempts = 3;
519fbd87 924 int err;
53b7997f 925 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
1da177e4 926
77d16f45 927 strict |= flags & RT6_LOOKUP_F_IFACE;
1da177e4
LT		 928
929relookup:
c71099ac 930 read_lock_bh(&table->tb6_lock);
1da177e4 931
8238dd06 932restart_2:
4c9483b2 933 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1da177e4
LT		 934
935restart:
4acad72d 936 rt = rt6_select(fn, oif, strict | reachable);
52bd4c0c
ND		 937 if (rt->rt6i_nsiblings)
938 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable);
4c9483b2 939 BACKTRACK(net, &fl6->saddr);
8ed67789 940 if (rt == net->ipv6.ip6_null_entry ||
8238dd06 941 rt->rt6i_flags & RTF_CACHE)
1ddef044 942 goto out;
1da177e4 943
d8d1f30b 944 dst_hold(&rt->dst);
c71099ac 945 read_unlock_bh(&table->tb6_lock);
fb9de91e 946
c440f160 947 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
4c9483b2 948 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
7343ff31 949 else if (!(rt->dst.flags & DST_HOST))
4c9483b2 950 nrt = rt6_alloc_clone(rt, &fl6->daddr);
7343ff31
DM		 951 else
952 goto out2;
e40cf353 953
94e187c0 954 ip6_rt_put(rt);
8ed67789 955 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 956
d8d1f30b 957 dst_hold(&rt->dst);
519fbd87 958 if (nrt) {
40e22e8f 959 err = ip6_ins_rt(nrt);
519fbd87 960 if (!err)
1da177e4 961 goto out2;
1da177e4 962 }
1da177e4 963
519fbd87
YH		 964 if (--attempts <= 0)
965 goto out2;
966
967 /*
c71099ac 968 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 969 * released someone could insert this route. Relookup.
970 */
94e187c0 971 ip6_rt_put(rt);
519fbd87
YH		 972 goto relookup;
973
974out:
8238dd06
YH		 975 if (reachable) {
976 reachable = 0;
977 goto restart_2;
978 }
d8d1f30b 979 dst_hold(&rt->dst);
c71099ac 980 read_unlock_bh(&table->tb6_lock);
1da177e4 981out2:
d8d1f30b
CG		 982 rt->dst.lastuse = jiffies;
983 rt->dst.__use++;
c71099ac
TG		 984
985 return rt;
1da177e4
LT		 986}
987
8ed67789 988static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 989 struct flowi6 *fl6, int flags)
4acad72d 990{
4c9483b2 991 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 992}
993
72331bc0
SL		 994static struct dst_entry *ip6_route_input_lookup(struct net *net,
995 struct net_device *dev,
996 struct flowi6 *fl6, int flags)
997{
998 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
999 flags |= RT6_LOOKUP_F_IFACE;
1000
1001 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1002}
1003
c71099ac
TG		 1004void ip6_route_input(struct sk_buff *skb)
1005{
b71d1d42 1006 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1007 struct net *net = dev_net(skb->dev);
adaa70bb 1008 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 1009 struct flowi6 fl6 = {
1010 .flowi6_iif = skb->dev->ifindex,
1011 .daddr = iph->daddr,
1012 .saddr = iph->saddr,
6502ca52 1013 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1014 .flowi6_mark = skb->mark,
1015 .flowi6_proto = iph->nexthdr,
c71099ac 1016 };
adaa70bb 1017
72331bc0 1018 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1019}
1020
8ed67789 1021static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1022 struct flowi6 *fl6, int flags)
1da177e4 1023{
4c9483b2 1024 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1025}
1026
67ba4152 1027struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 1028 struct flowi6 *fl6)
c71099ac
TG		 1029{
1030 int flags = 0;
1031
1fb9489b 1032 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1033
4c9483b2 1034 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 1035 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1036
4c9483b2 1037 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 1038 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1039 else if (sk)
1040 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1041
4c9483b2 1042 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4
LT		 1043}
1044
7159039a 1045EXPORT_SYMBOL(ip6_route_output);
1da177e4 1046
2774c131 1047struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1048{
5c1e6aa3 1049 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1050 struct dst_entry *new = NULL;
1051
f5b0a874 1052 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1053 if (rt) {
d8d1f30b 1054 new = &rt->dst;
14e50e57 1055
8104891b
SK		 1056 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1057 rt6_init_peer(rt, net->ipv6.peers);
1058
14e50e57 1059 new->__use = 1;
352e512c 1060 new->input = dst_discard;
aad88724 1061 new->output = dst_discard_sk;
14e50e57 1062
21efcfa0
ED		 1063 if (dst_metrics_read_only(&ort->dst))
1064 new->_metrics = ort->dst._metrics;
1065 else
1066 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1067 rt->rt6i_idev = ort->rt6i_idev;
1068 if (rt->rt6i_idev)
1069 in6_dev_hold(rt->rt6i_idev);
14e50e57 1070
4e3fd7a0 1071 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1072 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1073 rt->rt6i_metric = 0;
1074
1075 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1076#ifdef CONFIG_IPV6_SUBTREES
1077 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1078#endif
1079
1080 dst_free(new);
1081 }
1082
69ead7af
DM		 1083 dst_release(dst_orig);
1084 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1085}
14e50e57 1086
1da177e4
LT		 1087/*
1088 * Destination cache support functions
1089 */
1090
1091static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1092{
1093 struct rt6_info *rt;
1094
1095 rt = (struct rt6_info *) dst;
1096
6f3118b5
ND		 1097 /* All IPV6 dsts are created with ->obsolete set to the value
1098 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1099 * into this function always.
1100 */
ca4c3fc2 1101 if (rt->rt6i_genid != rt_genid_ipv6(dev_net(rt->dst.dev)))
6f3118b5
ND		 1102 return NULL;
1103
e3bc10bd
HFS		 1104 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1105 return NULL;
a4477c4d 1106
e3bc10bd
HFS		 1107 if (rt6_check_expired(rt))
1108 return NULL;
1109
1110 return dst;
1da177e4
LT		 1111}
1112
1113static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1114{
1115 struct rt6_info *rt = (struct rt6_info *) dst;
1116
1117 if (rt) {
54c1a859
YH		 1118 if (rt->rt6i_flags & RTF_CACHE) {
1119 if (rt6_check_expired(rt)) {
1120 ip6_del_rt(rt);
1121 dst = NULL;
1122 }
1123 } else {
1da177e4 1124 dst_release(dst);
54c1a859
YH		 1125 dst = NULL;
1126 }
1da177e4 1127 }
54c1a859 1128 return dst;
1da177e4
LT		 1129}
1130
1131static void ip6_link_failure(struct sk_buff *skb)
1132{
1133 struct rt6_info *rt;
1134
3ffe533c 1135 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1136
adf30907 1137 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1138 if (rt) {
1eb4f758
HFS		 1139 if (rt->rt6i_flags & RTF_CACHE) {
1140 dst_hold(&rt->dst);
1141 if (ip6_del_rt(rt))
1142 dst_free(&rt->dst);
1143 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1144 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1145 }
1da177e4
LT		 1146 }
1147}
1148
6700c270
DM		 1149static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1150 struct sk_buff *skb, u32 mtu)
1da177e4 1151{
67ba4152 1152 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1153
81aded24 1154 dst_confirm(dst);
1da177e4 1155 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
81aded24
DM		 1156 struct net *net = dev_net(dst->dev);
1157
1da177e4
LT		 1158 rt6->rt6i_flags |= RTF_MODIFIED;
1159 if (mtu < IPV6_MIN_MTU) {
defb3519 1160 u32 features = dst_metric(dst, RTAX_FEATURES);
1da177e4 1161 mtu = IPV6_MIN_MTU;
defb3519
DM		 1162 features |= RTAX_FEATURE_ALLFRAG;
1163 dst_metric_set(dst, RTAX_FEATURES, features);
1da177e4 1164 }
defb3519 1165 dst_metric_set(dst, RTAX_MTU, mtu);
81aded24 1166 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1167 }
1168}
1169
42ae66c8
DM		 1170void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1171 int oif, u32 mark)
81aded24
DM		 1172{
1173 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1174 struct dst_entry *dst;
1175 struct flowi6 fl6;
1176
1177 memset(&fl6, 0, sizeof(fl6));
1178 fl6.flowi6_oif = oif;
1b3c61dc 1179 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1180 fl6.daddr = iph->daddr;
1181 fl6.saddr = iph->saddr;
6502ca52 1182 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1183
1184 dst = ip6_route_output(net, NULL, &fl6);
1185 if (!dst->error)
6700c270 1186 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
81aded24
DM		 1187 dst_release(dst);
1188}
1189EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1190
1191void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1192{
1193 ip6_update_pmtu(skb, sock_net(sk), mtu,
1194 sk->sk_bound_dev_if, sk->sk_mark);
1195}
1196EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1197
b55b76b2
DJ		 1198/* Handle redirects */
1199struct ip6rd_flowi {
1200 struct flowi6 fl6;
1201 struct in6_addr gateway;
1202};
1203
1204static struct rt6_info *__ip6_route_redirect(struct net *net,
1205 struct fib6_table *table,
1206 struct flowi6 *fl6,
1207 int flags)
1208{
1209 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1210 struct rt6_info *rt;
1211 struct fib6_node *fn;
1212
1213 /* Get the "current" route for this destination and
1214 * check if the redirect has come from approriate router.
1215 *
1216 * RFC 4861 specifies that redirects should only be
1217 * accepted if they come from the nexthop to the target.
1218 * Due to the way the routes are chosen, this notion
1219 * is a bit fuzzy and one might need to check all possible
1220 * routes.
1221 */
1222
1223 read_lock_bh(&table->tb6_lock);
1224 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1225restart:
1226 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1227 if (rt6_check_expired(rt))
1228 continue;
1229 if (rt->dst.error)
1230 break;
1231 if (!(rt->rt6i_flags & RTF_GATEWAY))
1232 continue;
1233 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1234 continue;
1235 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1236 continue;
1237 break;
1238 }
1239
1240 if (!rt)
1241 rt = net->ipv6.ip6_null_entry;
1242 else if (rt->dst.error) {
1243 rt = net->ipv6.ip6_null_entry;
1244 goto out;
1245 }
1246 BACKTRACK(net, &fl6->saddr);
1247out:
1248 dst_hold(&rt->dst);
1249
1250 read_unlock_bh(&table->tb6_lock);
1251
1252 return rt;
1253};
1254
1255static struct dst_entry *ip6_route_redirect(struct net *net,
1256 const struct flowi6 *fl6,
1257 const struct in6_addr *gateway)
1258{
1259 int flags = RT6_LOOKUP_F_HAS_SADDR;
1260 struct ip6rd_flowi rdfl;
1261
1262 rdfl.fl6 = *fl6;
1263 rdfl.gateway = *gateway;
1264
1265 return fib6_rule_lookup(net, &rdfl.fl6,
1266 flags, __ip6_route_redirect);
1267}
1268
3a5ad2ee
DM		 1269void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1270{
1271 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1272 struct dst_entry *dst;
1273 struct flowi6 fl6;
1274
1275 memset(&fl6, 0, sizeof(fl6));
e374c618 1276 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1277 fl6.flowi6_oif = oif;
1278 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1279 fl6.daddr = iph->daddr;
1280 fl6.saddr = iph->saddr;
6502ca52 1281 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1282
b55b76b2
DJ		 1283 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1284 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1285 dst_release(dst);
1286}
1287EXPORT_SYMBOL_GPL(ip6_redirect);
1288
c92a59ec
DJ		 1289void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1290 u32 mark)
1291{
1292 const struct ipv6hdr *iph = ipv6_hdr(skb);
1293 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1294 struct dst_entry *dst;
1295 struct flowi6 fl6;
1296
1297 memset(&fl6, 0, sizeof(fl6));
e374c618 1298 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1299 fl6.flowi6_oif = oif;
1300 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1301 fl6.daddr = msg->dest;
1302 fl6.saddr = iph->daddr;
1303
b55b76b2
DJ		 1304 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1305 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1306 dst_release(dst);
1307}
1308
3a5ad2ee
DM		 1309void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1310{
1311 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1312}
1313EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1314
0dbaee3b 1315static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1316{
0dbaee3b
DM		 1317 struct net_device *dev = dst->dev;
1318 unsigned int mtu = dst_mtu(dst);
1319 struct net *net = dev_net(dev);
1320
1da177e4
LT		 1321 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1322
5578689a
DL		 1323 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1324 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1325
1326 /*
1ab1457c
YH		 1327 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1328 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1329 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1330 * rely only on pmtu discovery"
1331 */
1332 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1333 mtu = IPV6_MAXPLEN;
1334 return mtu;
1335}
1336
ebb762f2 1337static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1338{
d33e4553 1339 struct inet6_dev *idev;
618f9bc7
SK		 1340 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1341
1342 if (mtu)
30f78d8e 1343 goto out;
618f9bc7
SK		 1344
1345 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1346
1347 rcu_read_lock();
1348 idev = __in6_dev_get(dst->dev);
1349 if (idev)
1350 mtu = idev->cnf.mtu6;
1351 rcu_read_unlock();
1352
30f78d8e
ED		 1353out:
1354 return min_t(unsigned int, mtu, IP6_MAX_MTU);
d33e4553
DM		 1355}
1356
3b00944c
YH		 1357static struct dst_entry *icmp6_dst_gc_list;
1358static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1359
3b00944c 1360struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1361 struct flowi6 *fl6)
1da177e4 1362{
87a11578 1363 struct dst_entry *dst;
1da177e4
LT		 1364 struct rt6_info *rt;
1365 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1366 struct net *net = dev_net(dev);
1da177e4 1367
38308473 1368 if (unlikely(!idev))
122bdf67 1369 return ERR_PTR(-ENODEV);
1da177e4 1370
8b96d22d 1371 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1372 if (unlikely(!rt)) {
1da177e4 1373 in6_dev_put(idev);
87a11578 1374 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1375 goto out;
1376 }
1377
8e2ec639
YZ		 1378 rt->dst.flags |= DST_HOST;
1379 rt->dst.output = ip6_output;
d8d1f30b 1380 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1381 rt->rt6i_gateway = fl6->daddr;
87a11578 1382 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1383 rt->rt6i_dst.plen = 128;
1384 rt->rt6i_idev = idev;
14edd87d 1385 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1386
3b00944c 1387 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1388 rt->dst.next = icmp6_dst_gc_list;
1389 icmp6_dst_gc_list = &rt->dst;
3b00944c 1390 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1391
5578689a 1392 fib6_force_start_gc(net);
1da177e4 1393
87a11578
DM		 1394 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1395
1da177e4 1396out:
87a11578 1397 return dst;
1da177e4
LT		 1398}
1399
3d0f24a7 1400int icmp6_dst_gc(void)
1da177e4 1401{
e9476e95 1402 struct dst_entry *dst, **pprev;
3d0f24a7 1403 int more = 0;
1da177e4 1404
3b00944c
YH		 1405 spin_lock_bh(&icmp6_dst_lock);
1406 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1407
1da177e4
LT		 1408 while ((dst = *pprev) != NULL) {
1409 if (!atomic_read(&dst->__refcnt)) {
1410 *pprev = dst->next;
1411 dst_free(dst);
1da177e4
LT		 1412 } else {
1413 pprev = &dst->next;
3d0f24a7 1414 ++more;
1da177e4
LT		 1415 }
1416 }
1417
3b00944c 1418 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1419
3d0f24a7 1420 return more;
1da177e4
LT		 1421}
1422
1e493d19
DM		 1423static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1424 void *arg)
1425{
1426 struct dst_entry *dst, **pprev;
1427
1428 spin_lock_bh(&icmp6_dst_lock);
1429 pprev = &icmp6_dst_gc_list;
1430 while ((dst = *pprev) != NULL) {
1431 struct rt6_info *rt = (struct rt6_info *) dst;
1432 if (func(rt, arg)) {
1433 *pprev = dst->next;
1434 dst_free(dst);
1435 } else {
1436 pprev = &dst->next;
1437 }
1438 }
1439 spin_unlock_bh(&icmp6_dst_lock);
1440}
1441
569d3645 1442static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1443{
86393e52 1444 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1445 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1446 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1447 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1448 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1449 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1450 int entries;
7019b78e 1451
fc66f95c 1452 entries = dst_entries_get_fast(ops);
49a18d86 1453 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1454 entries <= rt_max_size)
1da177e4
LT		 1455 goto out;
1456
6891a346 1457 net->ipv6.ip6_rt_gc_expire++;
14956643 1458 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1459 entries = dst_entries_get_slow(ops);
1460 if (entries < ops->gc_thresh)
7019b78e 1461 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1462out:
7019b78e 1463 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1464 return entries > rt_max_size;
1da177e4
LT		 1465}
1466
1da177e4
LT		 1467/*
1468 *
1469 */
1470
86872cb5 1471int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1472{
1473 int err;
5578689a 1474 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1475 struct rt6_info *rt = NULL;
1476 struct net_device *dev = NULL;
1477 struct inet6_dev *idev = NULL;
c71099ac 1478 struct fib6_table *table;
1da177e4
LT		 1479 int addr_type;
1480
86872cb5 1481 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1482 return -EINVAL;
1483#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1484 if (cfg->fc_src_len)
1da177e4
LT		 1485 return -EINVAL;
1486#endif
86872cb5 1487 if (cfg->fc_ifindex) {
1da177e4 1488 err = -ENODEV;
5578689a 1489 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1490 if (!dev)
1491 goto out;
1492 idev = in6_dev_get(dev);
1493 if (!idev)
1494 goto out;
1495 }
1496
86872cb5
TG		 1497 if (cfg->fc_metric == 0)
1498 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1499
d71314b4 1500 err = -ENOBUFS;
38308473
DM		 1501 if (cfg->fc_nlinfo.nlh &&
1502 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1503 table = fib6_get_table(net, cfg->fc_table);
38308473 1504 if (!table) {
f3213831 1505 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1506 table = fib6_new_table(net, cfg->fc_table);
1507 }
1508 } else {
1509 table = fib6_new_table(net, cfg->fc_table);
1510 }
38308473
DM		 1511
1512 if (!table)
c71099ac 1513 goto out;
c71099ac 1514
c88507fb 1515 rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table);
1da177e4 1516
38308473 1517 if (!rt) {
1da177e4
LT		 1518 err = -ENOMEM;
1519 goto out;
1520 }
1521
1716a961
G		 1522 if (cfg->fc_flags & RTF_EXPIRES)
1523 rt6_set_expires(rt, jiffies +
1524 clock_t_to_jiffies(cfg->fc_expires));
1525 else
1526 rt6_clean_expires(rt);
1da177e4 1527
86872cb5
TG		 1528 if (cfg->fc_protocol == RTPROT_UNSPEC)
1529 cfg->fc_protocol = RTPROT_BOOT;
1530 rt->rt6i_protocol = cfg->fc_protocol;
1531
1532 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1533
1534 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1535 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1536 else if (cfg->fc_flags & RTF_LOCAL)
1537 rt->dst.input = ip6_input;
1da177e4 1538 else
d8d1f30b 1539 rt->dst.input = ip6_forward;
1da177e4 1540
d8d1f30b 1541 rt->dst.output = ip6_output;
1da177e4 1542
86872cb5
TG		 1543 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1544 rt->rt6i_dst.plen = cfg->fc_dst_len;
e5fd387a
MK		 1545 if (rt->rt6i_dst.plen == 128) {
1546 rt->dst.flags |= DST_HOST;
1547 dst_metrics_set_force_overwrite(&rt->dst);
8e2ec639 1548 }
e5fd387a 1549
1da177e4 1550#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1551 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1552 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1553#endif
1554
86872cb5 1555 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1556
1557 /* We cannot add true routes via loopback here,
1558 they would result in kernel looping; promote them to reject routes
1559 */
86872cb5 1560 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1561 (dev && (dev->flags & IFF_LOOPBACK) &&
1562 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1563 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1564 /* hold loopback dev/idev if we haven't done so. */
5578689a 1565 if (dev != net->loopback_dev) {
1da177e4
LT		 1566 if (dev) {
1567 dev_put(dev);
1568 in6_dev_put(idev);
1569 }
5578689a 1570 dev = net->loopback_dev;
1da177e4
LT		 1571 dev_hold(dev);
1572 idev = in6_dev_get(dev);
1573 if (!idev) {
1574 err = -ENODEV;
1575 goto out;
1576 }
1577 }
1da177e4 1578 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1579 switch (cfg->fc_type) {
1580 case RTN_BLACKHOLE:
1581 rt->dst.error = -EINVAL;
aad88724 1582 rt->dst.output = dst_discard_sk;
7150aede 1583 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1584 break;
1585 case RTN_PROHIBIT:
1586 rt->dst.error = -EACCES;
7150aede
K		 1587 rt->dst.output = ip6_pkt_prohibit_out;
1588 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1589 break;
b4949ab2 1590 case RTN_THROW:
ef2c7d7b 1591 default:
7150aede
K		 1592 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1593 : -ENETUNREACH;
1594 rt->dst.output = ip6_pkt_discard_out;
1595 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1596 break;
1597 }
1da177e4
LT		 1598 goto install_route;
1599 }
1600
86872cb5 1601 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1602 const struct in6_addr *gw_addr;
1da177e4
LT		 1603 int gwa_type;
1604
86872cb5 1605 gw_addr = &cfg->fc_gateway;
4e3fd7a0 1606 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1607 gwa_type = ipv6_addr_type(gw_addr);
1608
1609 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1610 struct rt6_info *grt;
1611
1612 /* IPv6 strictly inhibits using not link-local
1613 addresses as nexthop address.
1614 Otherwise, router will not able to send redirects.
1615 It is very good, but in some (rare!) circumstances
1616 (SIT, PtP, NBMA NOARP links) it is handy to allow
1617 some exceptions. --ANK
1618 */
1619 err = -EINVAL;
38308473 1620 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1621 goto out;
1622
5578689a 1623 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1624
1625 err = -EHOSTUNREACH;
38308473 1626 if (!grt)
1da177e4
LT		 1627 goto out;
1628 if (dev) {
d1918542 1629 if (dev != grt->dst.dev) {
94e187c0 1630 ip6_rt_put(grt);
1da177e4
LT		 1631 goto out;
1632 }
1633 } else {
d1918542 1634 dev = grt->dst.dev;
1da177e4
LT		 1635 idev = grt->rt6i_idev;
1636 dev_hold(dev);
1637 in6_dev_hold(grt->rt6i_idev);
1638 }
38308473 1639 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1640 err = 0;
94e187c0 1641 ip6_rt_put(grt);
1da177e4
LT		 1642
1643 if (err)
1644 goto out;
1645 }
1646 err = -EINVAL;
38308473 1647 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1648 goto out;
1649 }
1650
1651 err = -ENODEV;
38308473 1652 if (!dev)
1da177e4
LT		 1653 goto out;
1654
c3968a85
DW		 1655 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1656 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1657 err = -EINVAL;
1658 goto out;
1659 }
4e3fd7a0 1660 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1661 rt->rt6i_prefsrc.plen = 128;
1662 } else
1663 rt->rt6i_prefsrc.plen = 0;
1664
86872cb5 1665 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1666
1667install_route:
d8d1f30b 1668 rt->dst.dev = dev;
1da177e4 1669 rt->rt6i_idev = idev;
c71099ac 1670 rt->rt6i_table = table;
63152fc0 1671
c346dca1 1672 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1673
e5fd387a 1674 return __ip6_ins_rt(rt, &cfg->fc_nlinfo, cfg->fc_mx, cfg->fc_mx_len);
1da177e4
LT		 1675
1676out:
1677 if (dev)
1678 dev_put(dev);
1679 if (idev)
1680 in6_dev_put(idev);
1681 if (rt)
d8d1f30b 1682 dst_free(&rt->dst);
1da177e4
LT		 1683 return err;
1684}
1685
86872cb5 1686static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1687{
1688 int err;
c71099ac 1689 struct fib6_table *table;
d1918542 1690 struct net *net = dev_net(rt->dst.dev);
1da177e4 1691
6825a26c
G		 1692 if (rt == net->ipv6.ip6_null_entry) {
1693 err = -ENOENT;
1694 goto out;
1695 }
6c813a72 1696
c71099ac
TG		 1697 table = rt->rt6i_table;
1698 write_lock_bh(&table->tb6_lock);
86872cb5 1699 err = fib6_del(rt, info);
c71099ac 1700 write_unlock_bh(&table->tb6_lock);
1da177e4 1701
6825a26c 1702out:
94e187c0 1703 ip6_rt_put(rt);
1da177e4
LT		 1704 return err;
1705}
1706
e0a1ad73
TG		 1707int ip6_del_rt(struct rt6_info *rt)
1708{
4d1169c1 1709 struct nl_info info = {
d1918542 1710 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1711 };
528c4ceb 1712 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1713}
1714
86872cb5 1715static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1716{
c71099ac 1717 struct fib6_table *table;
1da177e4
LT		 1718 struct fib6_node *fn;
1719 struct rt6_info *rt;
1720 int err = -ESRCH;
1721
5578689a 1722 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1723 if (!table)
c71099ac
TG		 1724 return err;
1725
1726 read_lock_bh(&table->tb6_lock);
1da177e4 1727
c71099ac 1728 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1729 &cfg->fc_dst, cfg->fc_dst_len,
1730 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1731
1da177e4 1732 if (fn) {
d8d1f30b 1733 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
86872cb5 1734 if (cfg->fc_ifindex &&
d1918542
DM		 1735 (!rt->dst.dev ||
1736 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1737 continue;
86872cb5
TG		 1738 if (cfg->fc_flags & RTF_GATEWAY &&
1739 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1740 continue;
86872cb5 1741 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1742 continue;
d8d1f30b 1743 dst_hold(&rt->dst);
c71099ac 1744 read_unlock_bh(&table->tb6_lock);
1da177e4 1745
86872cb5 1746 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1747 }
1748 }
c71099ac 1749 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1750
1751 return err;
1752}
1753
6700c270 1754static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1755{
e8599ff4 1756 struct net *net = dev_net(skb->dev);
a6279458 1757 struct netevent_redirect netevent;
e8599ff4 1758 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1759 struct ndisc_options ndopts;
1760 struct inet6_dev *in6_dev;
1761 struct neighbour *neigh;
71bcdba0 1762 struct rd_msg *msg;
6e157b6a
DM		 1763 int optlen, on_link;
1764 u8 *lladdr;
e8599ff4 1765
29a3cad5 1766 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1767 optlen -= sizeof(*msg);
e8599ff4
DM		 1768
1769 if (optlen < 0) {
6e157b6a 1770 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1771 return;
1772 }
1773
71bcdba0 1774 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1775
71bcdba0 1776 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1777 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1778 return;
1779 }
1780
6e157b6a 1781 on_link = 0;
71bcdba0 1782 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1783 on_link = 1;
71bcdba0 1784 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1785 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1786 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1787 return;
1788 }
1789
1790 in6_dev = __in6_dev_get(skb->dev);
1791 if (!in6_dev)
1792 return;
1793 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1794 return;
1795
1796 /* RFC2461 8.1:
1797 * The IP source address of the Redirect MUST be the same as the current
1798 * first-hop router for the specified ICMP Destination Address.
1799 */
1800
71bcdba0 1801 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1802 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1803 return;
1804 }
6e157b6a
DM		 1805
1806 lladdr = NULL;
e8599ff4
DM		 1807 if (ndopts.nd_opts_tgt_lladdr) {
1808 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1809 skb->dev);
1810 if (!lladdr) {
1811 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1812 return;
1813 }
1814 }
1815
6e157b6a
DM		 1816 rt = (struct rt6_info *) dst;
1817 if (rt == net->ipv6.ip6_null_entry) {
1818 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1819 return;
6e157b6a 1820 }
e8599ff4 1821
6e157b6a
DM		 1822 /* Redirect received -> path was valid.
1823 * Look, redirects are sent only in response to data packets,
1824 * so that this nexthop apparently is reachable. --ANK
1825 */
1826 dst_confirm(&rt->dst);
a6279458 1827
71bcdba0 1828 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1829 if (!neigh)
1830 return;
a6279458 1831
1da177e4
LT		 1832 /*
1833 * We have finally decided to accept it.
1834 */
1835
1ab1457c 1836 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1837 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1838 NEIGH_UPDATE_F_OVERRIDE|
1839 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1840 NEIGH_UPDATE_F_ISROUTER))
1841 );
1842
71bcdba0 1843 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1844 if (!nrt)
1da177e4
LT		 1845 goto out;
1846
1847 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1848 if (on_link)
1849 nrt->rt6i_flags &= ~RTF_GATEWAY;
1850
4e3fd7a0 1851 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1852
40e22e8f 1853 if (ip6_ins_rt(nrt))
1da177e4
LT		 1854 goto out;
1855
d8d1f30b
CG		 1856 netevent.old = &rt->dst;
1857 netevent.new = &nrt->dst;
71bcdba0 1858 netevent.daddr = &msg->dest;
60592833 1859 netevent.neigh = neigh;
8d71740c
TT		 1860 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1861
38308473 1862 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1863 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1864 ip6_del_rt(rt);
1da177e4
LT		 1865 }
1866
1867out:
e8599ff4 1868 neigh_release(neigh);
6e157b6a
DM		 1869}
1870
1da177e4
LT		 1871/*
1872 * Misc support functions
1873 */
1874
1716a961 1875static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1876 const struct in6_addr *dest)
1da177e4 1877{
d1918542 1878 struct net *net = dev_net(ort->dst.dev);
8b96d22d
DM		 1879 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1880 ort->rt6i_table);
1da177e4
LT		 1881
1882 if (rt) {
d8d1f30b
CG		 1883 rt->dst.input = ort->dst.input;
1884 rt->dst.output = ort->dst.output;
8e2ec639 1885 rt->dst.flags |= DST_HOST;
d8d1f30b 1886
4e3fd7a0 1887 rt->rt6i_dst.addr = *dest;
8e2ec639 1888 rt->rt6i_dst.plen = 128;
defb3519 1889 dst_copy_metrics(&rt->dst, &ort->dst);
d8d1f30b 1890 rt->dst.error = ort->dst.error;
1da177e4
LT		 1891 rt->rt6i_idev = ort->rt6i_idev;
1892 if (rt->rt6i_idev)
1893 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1894 rt->dst.lastuse = jiffies;
1da177e4 1895
550bab42
JA		 1896 if (ort->rt6i_flags & RTF_GATEWAY)
1897 rt->rt6i_gateway = ort->rt6i_gateway;
1898 else
1899 rt->rt6i_gateway = *dest;
1716a961 1900 rt->rt6i_flags = ort->rt6i_flags;
24f5b855 1901 rt6_set_from(rt, ort);
1da177e4
LT		 1902 rt->rt6i_metric = 0;
1903
1da177e4
LT		 1904#ifdef CONFIG_IPV6_SUBTREES
1905 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1906#endif
0f6c6392 1907 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1908 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1909 }
1910 return rt;
1911}
1912
70ceb4f5 1913#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1914static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 1915 const struct in6_addr *prefix, int prefixlen,
1916 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 1917{
1918 struct fib6_node *fn;
1919 struct rt6_info *rt = NULL;
c71099ac
TG		 1920 struct fib6_table *table;
1921
efa2cea0 1922 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 1923 if (!table)
c71099ac 1924 return NULL;
70ceb4f5 1925
5744dd9b 1926 read_lock_bh(&table->tb6_lock);
67ba4152 1927 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 1928 if (!fn)
1929 goto out;
1930
d8d1f30b 1931 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1932 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 1933 continue;
1934 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1935 continue;
1936 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1937 continue;
d8d1f30b 1938 dst_hold(&rt->dst);
70ceb4f5
YH		 1939 break;
1940 }
1941out:
5744dd9b 1942 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1943 return rt;
1944}
1945
efa2cea0 1946static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 1947 const struct in6_addr *prefix, int prefixlen,
1948 const struct in6_addr *gwaddr, int ifindex,
95c96174 1949 unsigned int pref)
70ceb4f5 1950{
86872cb5
TG		 1951 struct fib6_config cfg = {
1952 .fc_table = RT6_TABLE_INFO,
238fc7ea 1953 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1954 .fc_ifindex = ifindex,
1955 .fc_dst_len = prefixlen,
1956 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1957 RTF_UP | RTF_PREF(pref),
15e47304 1958 .fc_nlinfo.portid = 0,
efa2cea0
DL		 1959 .fc_nlinfo.nlh = NULL,
1960 .fc_nlinfo.nl_net = net,
86872cb5
TG		 1961 };
1962
4e3fd7a0
AD		 1963 cfg.fc_dst = *prefix;
1964 cfg.fc_gateway = *gwaddr;
70ceb4f5 1965
e317da96
YH		 1966 /* We should treat it as a default route if prefix length is 0. */
1967 if (!prefixlen)
86872cb5 1968 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 1969
86872cb5 1970 ip6_route_add(&cfg);
70ceb4f5 1971
efa2cea0 1972 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 1973}
1974#endif
1975
b71d1d42 1976struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 1977{
1da177e4 1978 struct rt6_info *rt;
c71099ac 1979 struct fib6_table *table;
1da177e4 1980
c346dca1 1981 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 1982 if (!table)
c71099ac 1983 return NULL;
1da177e4 1984
5744dd9b 1985 read_lock_bh(&table->tb6_lock);
67ba4152 1986 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1987 if (dev == rt->dst.dev &&
045927ff 1988 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 1989 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1990 break;
1991 }
1992 if (rt)
d8d1f30b 1993 dst_hold(&rt->dst);
5744dd9b 1994 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1995 return rt;
1996}
1997
b71d1d42 1998struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 1999 struct net_device *dev,
2000 unsigned int pref)
1da177e4 2001{
86872cb5
TG		 2002 struct fib6_config cfg = {
2003 .fc_table = RT6_TABLE_DFLT,
238fc7ea 2004 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2005 .fc_ifindex = dev->ifindex,
2006 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2007 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2008 .fc_nlinfo.portid = 0,
5578689a 2009 .fc_nlinfo.nlh = NULL,
c346dca1 2010 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2011 };
1da177e4 2012
4e3fd7a0 2013 cfg.fc_gateway = *gwaddr;
1da177e4 2014
86872cb5 2015 ip6_route_add(&cfg);
1da177e4 2016
1da177e4
LT		 2017 return rt6_get_dflt_router(gwaddr, dev);
2018}
2019
7b4da532 2020void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2021{
2022 struct rt6_info *rt;
c71099ac
TG		 2023 struct fib6_table *table;
2024
2025 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2026 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2027 if (!table)
c71099ac 2028 return;
1da177e4
LT		 2029
2030restart:
c71099ac 2031 read_lock_bh(&table->tb6_lock);
d8d1f30b 2032 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2033 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2034 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2035 dst_hold(&rt->dst);
c71099ac 2036 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2037 ip6_del_rt(rt);
1da177e4
LT		 2038 goto restart;
2039 }
2040 }
c71099ac 2041 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2042}
2043
5578689a
DL		 2044static void rtmsg_to_fib6_config(struct net *net,
2045 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2046 struct fib6_config *cfg)
2047{
2048 memset(cfg, 0, sizeof(*cfg));
2049
2050 cfg->fc_table = RT6_TABLE_MAIN;
2051 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2052 cfg->fc_metric = rtmsg->rtmsg_metric;
2053 cfg->fc_expires = rtmsg->rtmsg_info;
2054 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2055 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2056 cfg->fc_flags = rtmsg->rtmsg_flags;
2057
5578689a 2058 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2059
4e3fd7a0
AD		 2060 cfg->fc_dst = rtmsg->rtmsg_dst;
2061 cfg->fc_src = rtmsg->rtmsg_src;
2062 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2063}
2064
5578689a 2065int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2066{
86872cb5 2067 struct fib6_config cfg;
1da177e4
LT		 2068 struct in6_rtmsg rtmsg;
2069 int err;
2070
67ba4152 2071 switch (cmd) {
1da177e4
LT		 2072 case SIOCADDRT: /* Add a route */
2073 case SIOCDELRT: /* Delete a route */
af31f412 2074 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2075 return -EPERM;
2076 err = copy_from_user(&rtmsg, arg,
2077 sizeof(struct in6_rtmsg));
2078 if (err)
2079 return -EFAULT;
86872cb5 2080
5578689a 2081 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2082
1da177e4
LT		 2083 rtnl_lock();
2084 switch (cmd) {
2085 case SIOCADDRT:
86872cb5 2086 err = ip6_route_add(&cfg);
1da177e4
LT		 2087 break;
2088 case SIOCDELRT:
86872cb5 2089 err = ip6_route_del(&cfg);
1da177e4
LT		 2090 break;
2091 default:
2092 err = -EINVAL;
2093 }
2094 rtnl_unlock();
2095
2096 return err;
3ff50b79 2097 }
1da177e4
LT		 2098
2099 return -EINVAL;
2100}
2101
2102/*
2103 * Drop the packet on the floor
2104 */
2105
d5fdd6ba 2106static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2107{
612f09e8 2108 int type;
adf30907 2109 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2110 switch (ipstats_mib_noroutes) {
2111 case IPSTATS_MIB_INNOROUTES:
0660e03f 2112 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2113 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2114 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2115 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2116 break;
2117 }
2118 /* FALLTHROUGH */
2119 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2120 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2121 ipstats_mib_noroutes);
612f09e8
YH		 2122 break;
2123 }
3ffe533c 2124 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2125 kfree_skb(skb);
2126 return 0;
2127}
2128
9ce8ade0
TG		 2129static int ip6_pkt_discard(struct sk_buff *skb)
2130{
612f09e8 2131 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2132}
2133
aad88724 2134static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb)
1da177e4 2135{
adf30907 2136 skb->dev = skb_dst(skb)->dev;
612f09e8 2137 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2138}
2139
9ce8ade0
TG		 2140static int ip6_pkt_prohibit(struct sk_buff *skb)
2141{
612f09e8 2142 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2143}
2144
aad88724 2145static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb)
9ce8ade0 2146{
adf30907 2147 skb->dev = skb_dst(skb)->dev;
612f09e8 2148 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2149}
2150
1da177e4
LT		 2151/*
2152 * Allocate a dst for local (unicast / anycast) address.
2153 */
2154
2155struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2156 const struct in6_addr *addr,
8f031519 2157 bool anycast)
1da177e4 2158{
c346dca1 2159 struct net *net = dev_net(idev->dev);
a3300ef4
HFS		 2160 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2161 DST_NOCOUNT, NULL);
2162 if (!rt)
1da177e4
LT		 2163 return ERR_PTR(-ENOMEM);
2164
1da177e4
LT		 2165 in6_dev_hold(idev);
2166
11d53b49 2167 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2168 rt->dst.input = ip6_input;
2169 rt->dst.output = ip6_output;
1da177e4 2170 rt->rt6i_idev = idev;
1da177e4
LT		 2171
2172 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2173 if (anycast)
2174 rt->rt6i_flags |= RTF_ANYCAST;
2175 else
1da177e4 2176 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2177
550bab42 2178 rt->rt6i_gateway = *addr;
4e3fd7a0 2179 rt->rt6i_dst.addr = *addr;
1da177e4 2180 rt->rt6i_dst.plen = 128;
5578689a 2181 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2182
d8d1f30b 2183 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2184
2185 return rt;
2186}
2187
c3968a85
DW		 2188int ip6_route_get_saddr(struct net *net,
2189 struct rt6_info *rt,
b71d1d42 2190 const struct in6_addr *daddr,
c3968a85
DW		 2191 unsigned int prefs,
2192 struct in6_addr *saddr)
2193{
67ba4152 2194 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry *)rt);
c3968a85
DW		 2195 int err = 0;
2196 if (rt->rt6i_prefsrc.plen)
4e3fd7a0 2197 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2198 else
2199 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2200 daddr, prefs, saddr);
2201 return err;
2202}
2203
2204/* remove deleted ip from prefsrc entries */
2205struct arg_dev_net_ip {
2206 struct net_device *dev;
2207 struct net *net;
2208 struct in6_addr *addr;
2209};
2210
2211static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2212{
2213 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2214 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2215 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2216
d1918542 2217 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2218 rt != net->ipv6.ip6_null_entry &&
2219 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2220 /* remove prefsrc entry */
2221 rt->rt6i_prefsrc.plen = 0;
2222 }
2223 return 0;
2224}
2225
2226void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2227{
2228 struct net *net = dev_net(ifp->idev->dev);
2229 struct arg_dev_net_ip adni = {
2230 .dev = ifp->idev->dev,
2231 .net = net,
2232 .addr = &ifp->addr,
2233 };
0c3584d5 2234 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2235}
2236
be7a010d
DJ		 2237#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2238#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2239
2240/* Remove routers and update dst entries when gateway turn into host. */
2241static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2242{
2243 struct in6_addr *gateway = (struct in6_addr *)arg;
2244
2245 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2246 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2247 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2248 return -1;
2249 }
2250 return 0;
2251}
2252
2253void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2254{
2255 fib6_clean_all(net, fib6_clean_tohost, gateway);
2256}
2257
8ed67789
DL		 2258struct arg_dev_net {
2259 struct net_device *dev;
2260 struct net *net;
2261};
2262
1da177e4
LT		 2263static int fib6_ifdown(struct rt6_info *rt, void *arg)
2264{
bc3ef660 2265 const struct arg_dev_net *adn = arg;
2266 const struct net_device *dev = adn->dev;
8ed67789 2267
d1918542 2268 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2269 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2270 return -1;
c159d30c 2271
1da177e4
LT		 2272 return 0;
2273}
2274
f3db4851 2275void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2276{
8ed67789
DL		 2277 struct arg_dev_net adn = {
2278 .dev = dev,
2279 .net = net,
2280 };
2281
0c3584d5 2282 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2283 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2284}
2285
95c96174 2286struct rt6_mtu_change_arg {
1da177e4 2287 struct net_device *dev;
95c96174 2288 unsigned int mtu;
1da177e4
LT		 2289};
2290
2291static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2292{
2293 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2294 struct inet6_dev *idev;
2295
2296 /* In IPv6 pmtu discovery is not optional,
2297 so that RTAX_MTU lock cannot disable it.
2298 We still use this lock to block changes
2299 caused by addrconf/ndisc.
2300 */
2301
2302 idev = __in6_dev_get(arg->dev);
38308473 2303 if (!idev)
1da177e4
LT		 2304 return 0;
2305
2306 /* For administrative MTU increase, there is no way to discover
2307 IPv6 PMTU increase, so PMTU increase should be updated here.
2308 Since RFC 1981 doesn't include administrative MTU increase
2309 update PMTU increase is a MUST. (i.e. jumbo frame)
2310 */
2311 /*
2312 If new MTU is less than route PMTU, this new MTU will be the
2313 lowest MTU in the path, update the route PMTU to reflect PMTU
2314 decreases; if new MTU is greater than route PMTU, and the
2315 old MTU is the lowest MTU in the path, update the route PMTU
2316 to reflect the increase. In this case if the other nodes' MTU
2317 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2318 PMTU discouvery.
2319 */
d1918542 2320 if (rt->dst.dev == arg->dev &&
d8d1f30b
CG		 2321 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2322 (dst_mtu(&rt->dst) >= arg->mtu ||
2323 (dst_mtu(&rt->dst) < arg->mtu &&
2324 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
defb3519 2325 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
566cfd8f 2326 }
1da177e4
LT		 2327 return 0;
2328}
2329
95c96174 2330void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2331{
c71099ac
TG		 2332 struct rt6_mtu_change_arg arg = {
2333 .dev = dev,
2334 .mtu = mtu,
2335 };
1da177e4 2336
0c3584d5 2337 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2338}
2339
ef7c79ed 2340static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2341 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2342 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2343 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2344 [RTA_PRIORITY] = { .type = NLA_U32 },
2345 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2346 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
86872cb5
TG		 2347};
2348
2349static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2350 struct fib6_config *cfg)
1da177e4 2351{
86872cb5
TG		 2352 struct rtmsg *rtm;
2353 struct nlattr *tb[RTA_MAX+1];
2354 int err;
1da177e4 2355
86872cb5
TG		 2356 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2357 if (err < 0)
2358 goto errout;
1da177e4 2359
86872cb5
TG		 2360 err = -EINVAL;
2361 rtm = nlmsg_data(nlh);
2362 memset(cfg, 0, sizeof(*cfg));
2363
2364 cfg->fc_table = rtm->rtm_table;
2365 cfg->fc_dst_len = rtm->rtm_dst_len;
2366 cfg->fc_src_len = rtm->rtm_src_len;
2367 cfg->fc_flags = RTF_UP;
2368 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2369 cfg->fc_type = rtm->rtm_type;
86872cb5 2370
ef2c7d7b
ND		 2371 if (rtm->rtm_type == RTN_UNREACHABLE ||
2372 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2373 rtm->rtm_type == RTN_PROHIBIT ||
2374 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2375 cfg->fc_flags |= RTF_REJECT;
2376
ab79ad14
MŻ		 2377 if (rtm->rtm_type == RTN_LOCAL)
2378 cfg->fc_flags |= RTF_LOCAL;
2379
15e47304 2380 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2381 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2382 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2383
2384 if (tb[RTA_GATEWAY]) {
2385 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2386 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2387 }
86872cb5
TG		 2388
2389 if (tb[RTA_DST]) {
2390 int plen = (rtm->rtm_dst_len + 7) >> 3;
2391
2392 if (nla_len(tb[RTA_DST]) < plen)
2393 goto errout;
2394
2395 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2396 }
86872cb5
TG		 2397
2398 if (tb[RTA_SRC]) {
2399 int plen = (rtm->rtm_src_len + 7) >> 3;
2400
2401 if (nla_len(tb[RTA_SRC]) < plen)
2402 goto errout;
2403
2404 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2405 }
86872cb5 2406
c3968a85
DW		 2407 if (tb[RTA_PREFSRC])
2408 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2409
86872cb5
TG		 2410 if (tb[RTA_OIF])
2411 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2412
2413 if (tb[RTA_PRIORITY])
2414 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2415
2416 if (tb[RTA_METRICS]) {
2417 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2418 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2419 }
86872cb5
TG		 2420
2421 if (tb[RTA_TABLE])
2422 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2423
51ebd318
ND		 2424 if (tb[RTA_MULTIPATH]) {
2425 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2426 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2427 }
2428
86872cb5
TG		 2429 err = 0;
2430errout:
2431 return err;
1da177e4
LT		 2432}
2433
51ebd318
ND		 2434static int ip6_route_multipath(struct fib6_config *cfg, int add)
2435{
2436 struct fib6_config r_cfg;
2437 struct rtnexthop *rtnh;
2438 int remaining;
2439 int attrlen;
2440 int err = 0, last_err = 0;
2441
2442beginning:
2443 rtnh = (struct rtnexthop *)cfg->fc_mp;
2444 remaining = cfg->fc_mp_len;
2445
2446 /* Parse a Multipath Entry */
2447 while (rtnh_ok(rtnh, remaining)) {
2448 memcpy(&r_cfg, cfg, sizeof(*cfg));
2449 if (rtnh->rtnh_ifindex)
2450 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2451
2452 attrlen = rtnh_attrlen(rtnh);
2453 if (attrlen > 0) {
2454 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2455
2456 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2457 if (nla) {
2458 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2459 r_cfg.fc_flags |= RTF_GATEWAY;
2460 }
2461 }
2462 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2463 if (err) {
2464 last_err = err;
2465 /* If we are trying to remove a route, do not stop the
2466 * loop when ip6_route_del() fails (because next hop is
2467 * already gone), we should try to remove all next hops.
2468 */
2469 if (add) {
2470 /* If add fails, we should try to delete all
2471 * next hops that have been already added.
2472 */
2473 add = 0;
2474 goto beginning;
2475 }
2476 }
1a72418b
ND		 2477 /* Because each route is added like a single route we remove
2478 * this flag after the first nexthop (if there is a collision,
2479 * we have already fail to add the first nexthop:
2480 * fib6_add_rt2node() has reject it).
2481 */
2482 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
51ebd318
ND		 2483 rtnh = rtnh_next(rtnh, &remaining);
2484 }
2485
2486 return last_err;
2487}
2488
67ba4152 2489static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2490{
86872cb5
TG		 2491 struct fib6_config cfg;
2492 int err;
1da177e4 2493
86872cb5
TG		 2494 err = rtm_to_fib6_config(skb, nlh, &cfg);
2495 if (err < 0)
2496 return err;
2497
51ebd318
ND		 2498 if (cfg.fc_mp)
2499 return ip6_route_multipath(&cfg, 0);
2500 else
2501 return ip6_route_del(&cfg);
1da177e4
LT		 2502}
2503
67ba4152 2504static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2505{
86872cb5
TG		 2506 struct fib6_config cfg;
2507 int err;
1da177e4 2508
86872cb5
TG		 2509 err = rtm_to_fib6_config(skb, nlh, &cfg);
2510 if (err < 0)
2511 return err;
2512
51ebd318
ND		 2513 if (cfg.fc_mp)
2514 return ip6_route_multipath(&cfg, 1);
2515 else
2516 return ip6_route_add(&cfg);
1da177e4
LT		 2517}
2518
339bf98f
TG		 2519static inline size_t rt6_nlmsg_size(void)
2520{
2521 return NLMSG_ALIGN(sizeof(struct rtmsg))
2522 + nla_total_size(16) /* RTA_SRC */
2523 + nla_total_size(16) /* RTA_DST */
2524 + nla_total_size(16) /* RTA_GATEWAY */
2525 + nla_total_size(16) /* RTA_PREFSRC */
2526 + nla_total_size(4) /* RTA_TABLE */
2527 + nla_total_size(4) /* RTA_IIF */
2528 + nla_total_size(4) /* RTA_OIF */
2529 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2530 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
339bf98f
TG		 2531 + nla_total_size(sizeof(struct rta_cacheinfo));
2532}
2533
191cd582
BH		 2534static int rt6_fill_node(struct net *net,
2535 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2536 struct in6_addr *dst, struct in6_addr *src,
15e47304 2537 int iif, int type, u32 portid, u32 seq,
7bc570c8 2538 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2539{
2540 struct rtmsg *rtm;
2d7202bf 2541 struct nlmsghdr *nlh;
e3703b3d 2542 long expires;
9e762a4a 2543 u32 table;
1da177e4
LT		 2544
2545 if (prefix) { /* user wants prefix routes only */
2546 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2547 /* success since this is not a prefix route */
2548 return 1;
2549 }
2550 }
2551
15e47304 2552 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2553 if (!nlh)
26932566 2554 return -EMSGSIZE;
2d7202bf
TG		 2555
2556 rtm = nlmsg_data(nlh);
1da177e4
LT		 2557 rtm->rtm_family = AF_INET6;
2558 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2559 rtm->rtm_src_len = rt->rt6i_src.plen;
2560 rtm->rtm_tos = 0;
c71099ac 2561 if (rt->rt6i_table)
9e762a4a 2562 table = rt->rt6i_table->tb6_id;
c71099ac 2563 else
9e762a4a
PM		 2564 table = RT6_TABLE_UNSPEC;
2565 rtm->rtm_table = table;
c78679e8
DM		 2566 if (nla_put_u32(skb, RTA_TABLE, table))
2567 goto nla_put_failure;
ef2c7d7b
ND		 2568 if (rt->rt6i_flags & RTF_REJECT) {
2569 switch (rt->dst.error) {
2570 case -EINVAL:
2571 rtm->rtm_type = RTN_BLACKHOLE;
2572 break;
2573 case -EACCES:
2574 rtm->rtm_type = RTN_PROHIBIT;
2575 break;
b4949ab2
ND		 2576 case -EAGAIN:
2577 rtm->rtm_type = RTN_THROW;
2578 break;
ef2c7d7b
ND		 2579 default:
2580 rtm->rtm_type = RTN_UNREACHABLE;
2581 break;
2582 }
2583 }
38308473 2584 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2585 rtm->rtm_type = RTN_LOCAL;
d1918542 2586 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2587 rtm->rtm_type = RTN_LOCAL;
2588 else
2589 rtm->rtm_type = RTN_UNICAST;
2590 rtm->rtm_flags = 0;
2591 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2592 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2593 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2594 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2595 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2596 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2597 rtm->rtm_protocol = RTPROT_RA;
2598 else
2599 rtm->rtm_protocol = RTPROT_KERNEL;
2600 }
1da177e4 2601
38308473 2602 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2603 rtm->rtm_flags |= RTM_F_CLONED;
2604
2605 if (dst) {
c78679e8
DM		 2606 if (nla_put(skb, RTA_DST, 16, dst))
2607 goto nla_put_failure;
1ab1457c 2608 rtm->rtm_dst_len = 128;
1da177e4 2609 } else if (rtm->rtm_dst_len)
c78679e8
DM		 2610 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2611 goto nla_put_failure;
1da177e4
LT		 2612#ifdef CONFIG_IPV6_SUBTREES
2613 if (src) {
c78679e8
DM		 2614 if (nla_put(skb, RTA_SRC, 16, src))
2615 goto nla_put_failure;
1ab1457c 2616 rtm->rtm_src_len = 128;
c78679e8
DM		 2617 } else if (rtm->rtm_src_len &&
2618 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2619 goto nla_put_failure;
1da177e4 2620#endif
7bc570c8
YH		 2621 if (iif) {
2622#ifdef CONFIG_IPV6_MROUTE
2623 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2624 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2625 if (err <= 0) {
2626 if (!nowait) {
2627 if (err == 0)
2628 return 0;
2629 goto nla_put_failure;
2630 } else {
2631 if (err == -EMSGSIZE)
2632 goto nla_put_failure;
2633 }
2634 }
2635 } else
2636#endif
c78679e8
DM		 2637 if (nla_put_u32(skb, RTA_IIF, iif))
2638 goto nla_put_failure;
7bc570c8 2639 } else if (dst) {
1da177e4 2640 struct in6_addr saddr_buf;
c78679e8
DM		 2641 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2642 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2643 goto nla_put_failure;
1da177e4 2644 }
2d7202bf 2645
c3968a85
DW		 2646 if (rt->rt6i_prefsrc.plen) {
2647 struct in6_addr saddr_buf;
4e3fd7a0 2648 saddr_buf = rt->rt6i_prefsrc.addr;
c78679e8
DM		 2649 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2650 goto nla_put_failure;
c3968a85
DW		 2651 }
2652
defb3519 2653 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2d7202bf
TG		 2654 goto nla_put_failure;
2655
dd0cbf29
YH		 2656 if (rt->rt6i_flags & RTF_GATEWAY) {
2657 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
94f826b8 2658 goto nla_put_failure;
94f826b8 2659 }
2d7202bf 2660
c78679e8
DM		 2661 if (rt->dst.dev &&
2662 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2663 goto nla_put_failure;
2664 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2665 goto nla_put_failure;
8253947e
LW		 2666
2667 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2668
87a50699 2669 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2670 goto nla_put_failure;
2d7202bf
TG		 2671
2672 return nlmsg_end(skb, nlh);
2673
2674nla_put_failure:
26932566
PM		 2675 nlmsg_cancel(skb, nlh);
2676 return -EMSGSIZE;
1da177e4
LT		 2677}
2678
1b43af54 2679int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2680{
2681 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2682 int prefix;
2683
2d7202bf
TG		 2684 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2685 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2686 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2687 } else
2688 prefix = 0;
2689
191cd582
BH		 2690 return rt6_fill_node(arg->net,
2691 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2692 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2693 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2694}
2695
67ba4152 2696static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 2697{
3b1e0a65 2698 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2699 struct nlattr *tb[RTA_MAX+1];
2700 struct rt6_info *rt;
1da177e4 2701 struct sk_buff *skb;
ab364a6f 2702 struct rtmsg *rtm;
4c9483b2 2703 struct flowi6 fl6;
72331bc0 2704 int err, iif = 0, oif = 0;
1da177e4 2705
ab364a6f
TG		 2706 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2707 if (err < 0)
2708 goto errout;
1da177e4 2709
ab364a6f 2710 err = -EINVAL;
4c9483b2 2711 memset(&fl6, 0, sizeof(fl6));
1da177e4 2712
ab364a6f
TG		 2713 if (tb[RTA_SRC]) {
2714 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2715 goto errout;
2716
4e3fd7a0 2717 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2718 }
2719
2720 if (tb[RTA_DST]) {
2721 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2722 goto errout;
2723
4e3fd7a0 2724 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2725 }
2726
2727 if (tb[RTA_IIF])
2728 iif = nla_get_u32(tb[RTA_IIF]);
2729
2730 if (tb[RTA_OIF])
72331bc0 2731 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 2732
2e47b291
LC		 2733 if (tb[RTA_MARK])
2734 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
2735
1da177e4
LT		 2736 if (iif) {
2737 struct net_device *dev;
72331bc0
SL		 2738 int flags = 0;
2739
5578689a 2740 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2741 if (!dev) {
2742 err = -ENODEV;
ab364a6f 2743 goto errout;
1da177e4 2744 }
72331bc0
SL		 2745
2746 fl6.flowi6_iif = iif;
2747
2748 if (!ipv6_addr_any(&fl6.saddr))
2749 flags |= RT6_LOOKUP_F_HAS_SADDR;
2750
2751 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2752 flags);
2753 } else {
2754 fl6.flowi6_oif = oif;
2755
2756 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2757 }
2758
ab364a6f 2759 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2760 if (!skb) {
94e187c0 2761 ip6_rt_put(rt);
ab364a6f
TG		 2762 err = -ENOBUFS;
2763 goto errout;
2764 }
1da177e4 2765
ab364a6f
TG		 2766 /* Reserve room for dummy headers, this skb can pass
2767 through good chunk of routing engine.
2768 */
459a98ed 2769 skb_reset_mac_header(skb);
ab364a6f 2770 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2771
d8d1f30b 2772 skb_dst_set(skb, &rt->dst);
1da177e4 2773
4c9483b2 2774 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2775 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2776 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2777 if (err < 0) {
ab364a6f
TG		 2778 kfree_skb(skb);
2779 goto errout;
1da177e4
LT		 2780 }
2781
15e47304 2782 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2783errout:
1da177e4 2784 return err;
1da177e4
LT		 2785}
2786
86872cb5 2787void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2788{
2789 struct sk_buff *skb;
5578689a 2790 struct net *net = info->nl_net;
528c4ceb
DL		 2791 u32 seq;
2792 int err;
2793
2794 err = -ENOBUFS;
38308473 2795 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2796
339bf98f 2797 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2798 if (!skb)
21713ebc
TG		 2799 goto errout;
2800
191cd582 2801 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2802 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2803 if (err < 0) {
2804 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2805 WARN_ON(err == -EMSGSIZE);
2806 kfree_skb(skb);
2807 goto errout;
2808 }
15e47304 2809 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2810 info->nlh, gfp_any());
2811 return;
21713ebc
TG		 2812errout:
2813 if (err < 0)
5578689a 2814 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2815}
2816
8ed67789 2817static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2818 unsigned long event, void *ptr)
8ed67789 2819{
351638e7 2820 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2821 struct net *net = dev_net(dev);
8ed67789
DL		 2822
2823 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2824 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2825 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2826#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2827 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2828 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2829 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2830 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2831#endif
2832 }
2833
2834 return NOTIFY_OK;
2835}
2836
1da177e4
LT		 2837/*
2838 * /proc
2839 */
2840
2841#ifdef CONFIG_PROC_FS
2842
33120b30
AD		 2843static const struct file_operations ipv6_route_proc_fops = {
2844 .owner = THIS_MODULE,
2845 .open = ipv6_route_open,
2846 .read = seq_read,
2847 .llseek = seq_lseek,
8d2ca1d7 2848 .release = seq_release_net,
33120b30
AD		 2849};
2850
1da177e4
LT		 2851static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2852{
69ddb805 2853 struct net *net = (struct net *)seq->private;
1da177e4 2854 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2855 net->ipv6.rt6_stats->fib_nodes,
2856 net->ipv6.rt6_stats->fib_route_nodes,
2857 net->ipv6.rt6_stats->fib_rt_alloc,
2858 net->ipv6.rt6_stats->fib_rt_entries,
2859 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2860 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2861 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2862
2863 return 0;
2864}
2865
2866static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2867{
de05c557 2868 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2869}
2870
9a32144e 2871static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2872 .owner = THIS_MODULE,
2873 .open = rt6_stats_seq_open,
2874 .read = seq_read,
2875 .llseek = seq_lseek,
b6fcbdb4 2876 .release = single_release_net,
1da177e4
LT		 2877};
2878#endif /* CONFIG_PROC_FS */
2879
2880#ifdef CONFIG_SYSCTL
2881
1da177e4 2882static
fe2c6338 2883int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 2884 void __user *buffer, size_t *lenp, loff_t *ppos)
2885{
c486da34
LAG		 2886 struct net *net;
2887 int delay;
2888 if (!write)
1da177e4 2889 return -EINVAL;
c486da34
LAG		 2890
2891 net = (struct net *)ctl->extra1;
2892 delay = net->ipv6.sysctl.flush_delay;
2893 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 2894 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 2895 return 0;
1da177e4
LT		 2896}
2897
fe2c6338 2898struct ctl_table ipv6_route_table_template[] = {
1ab1457c 2899 {
1da177e4 2900 .procname = "flush",
4990509f 2901 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2902 .maxlen = sizeof(int),
89c8b3a1 2903 .mode = 0200,
6d9f239a 2904 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 2905 },
2906 {
1da177e4 2907 .procname = "gc_thresh",
9a7ec3a9 2908 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2909 .maxlen = sizeof(int),
2910 .mode = 0644,
6d9f239a 2911 .proc_handler = proc_dointvec,
1da177e4
LT		 2912 },
2913 {
1da177e4 2914 .procname = "max_size",
4990509f 2915 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2916 .maxlen = sizeof(int),
2917 .mode = 0644,
6d9f239a 2918 .proc_handler = proc_dointvec,
1da177e4
LT		 2919 },
2920 {
1da177e4 2921 .procname = "gc_min_interval",
4990509f 2922 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2923 .maxlen = sizeof(int),
2924 .mode = 0644,
6d9f239a 2925 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2926 },
2927 {
1da177e4 2928 .procname = "gc_timeout",
4990509f 2929 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2930 .maxlen = sizeof(int),
2931 .mode = 0644,
6d9f239a 2932 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2933 },
2934 {
1da177e4 2935 .procname = "gc_interval",
4990509f 2936 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2937 .maxlen = sizeof(int),
2938 .mode = 0644,
6d9f239a 2939 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2940 },
2941 {
1da177e4 2942 .procname = "gc_elasticity",
4990509f 2943 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 2944 .maxlen = sizeof(int),
2945 .mode = 0644,
f3d3f616 2946 .proc_handler = proc_dointvec,
1da177e4
LT		 2947 },
2948 {
1da177e4 2949 .procname = "mtu_expires",
4990509f 2950 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 2951 .maxlen = sizeof(int),
2952 .mode = 0644,
6d9f239a 2953 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2954 },
2955 {
1da177e4 2956 .procname = "min_adv_mss",
4990509f 2957 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 2958 .maxlen = sizeof(int),
2959 .mode = 0644,
f3d3f616 2960 .proc_handler = proc_dointvec,
1da177e4
LT		 2961 },
2962 {
1da177e4 2963 .procname = "gc_min_interval_ms",
4990509f 2964 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2965 .maxlen = sizeof(int),
2966 .mode = 0644,
6d9f239a 2967 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 2968 },
f8572d8f 2969 { }
1da177e4
LT		 2970};
2971
2c8c1e72 2972struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 2973{
2974 struct ctl_table *table;
2975
2976 table = kmemdup(ipv6_route_table_template,
2977 sizeof(ipv6_route_table_template),
2978 GFP_KERNEL);
5ee09105
YH		 2979
2980 if (table) {
2981 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 2982 table[0].extra1 = net;
86393e52 2983 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 2984 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2985 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2986 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2987 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2988 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2989 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2990 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 2991 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 2992
2993 /* Don't export sysctls to unprivileged users */
2994 if (net->user_ns != &init_user_ns)
2995 table[0].procname = NULL;
5ee09105
YH		 2996 }
2997
760f2d01
DL		 2998 return table;
2999}
1da177e4
LT		 3000#endif
3001
2c8c1e72 3002static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3003{
633d424b 3004 int ret = -ENOMEM;
8ed67789 3005
86393e52
AD		 3006 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3007 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3008
fc66f95c
ED		 3009 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3010 goto out_ip6_dst_ops;
3011
8ed67789
DL		 3012 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3013 sizeof(*net->ipv6.ip6_null_entry),
3014 GFP_KERNEL);
3015 if (!net->ipv6.ip6_null_entry)
fc66f95c 3016 goto out_ip6_dst_entries;
d8d1f30b 3017 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3018 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3019 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3020 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3021 ip6_template_metrics, true);
8ed67789
DL		 3022
3023#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3024 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3025 sizeof(*net->ipv6.ip6_prohibit_entry),
3026 GFP_KERNEL);
68fffc67
PZ		 3027 if (!net->ipv6.ip6_prohibit_entry)
3028 goto out_ip6_null_entry;
d8d1f30b 3029 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3030 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3031 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3032 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3033 ip6_template_metrics, true);
8ed67789
DL		 3034
3035 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3036 sizeof(*net->ipv6.ip6_blk_hole_entry),
3037 GFP_KERNEL);
68fffc67
PZ		 3038 if (!net->ipv6.ip6_blk_hole_entry)
3039 goto out_ip6_prohibit_entry;
d8d1f30b 3040 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3041 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3042 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3043 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3044 ip6_template_metrics, true);
8ed67789
DL		 3045#endif
3046
b339a47c
PZ		 3047 net->ipv6.sysctl.flush_delay = 0;
3048 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3049 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3050 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3051 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3052 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3053 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3054 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3055
6891a346
BT		 3056 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3057
8ed67789
DL		 3058 ret = 0;
3059out:
3060 return ret;
f2fc6a54 3061
68fffc67
PZ		 3062#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3063out_ip6_prohibit_entry:
3064 kfree(net->ipv6.ip6_prohibit_entry);
3065out_ip6_null_entry:
3066 kfree(net->ipv6.ip6_null_entry);
3067#endif
fc66f95c
ED		 3068out_ip6_dst_entries:
3069 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3070out_ip6_dst_ops:
f2fc6a54 3071 goto out;
cdb18761
DL		 3072}
3073
2c8c1e72 3074static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3075{
8ed67789
DL		 3076 kfree(net->ipv6.ip6_null_entry);
3077#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3078 kfree(net->ipv6.ip6_prohibit_entry);
3079 kfree(net->ipv6.ip6_blk_hole_entry);
3080#endif
41bb78b4 3081 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3082}
3083
d189634e
TG		 3084static int __net_init ip6_route_net_init_late(struct net *net)
3085{
3086#ifdef CONFIG_PROC_FS
d4beaa66
G		 3087 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3088 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3089#endif
3090 return 0;
3091}
3092
3093static void __net_exit ip6_route_net_exit_late(struct net *net)
3094{
3095#ifdef CONFIG_PROC_FS
ece31ffd
G		 3096 remove_proc_entry("ipv6_route", net->proc_net);
3097 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3098#endif
3099}
3100
cdb18761
DL		 3101static struct pernet_operations ip6_route_net_ops = {
3102 .init = ip6_route_net_init,
3103 .exit = ip6_route_net_exit,
3104};
3105
c3426b47
DM		 3106static int __net_init ipv6_inetpeer_init(struct net *net)
3107{
3108 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3109
3110 if (!bp)
3111 return -ENOMEM;
3112 inet_peer_base_init(bp);
3113 net->ipv6.peers = bp;
3114 return 0;
3115}
3116
3117static void __net_exit ipv6_inetpeer_exit(struct net *net)
3118{
3119 struct inet_peer_base *bp = net->ipv6.peers;
3120
3121 net->ipv6.peers = NULL;
56a6b248 3122 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3123 kfree(bp);
3124}
3125
2b823f72 3126static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3127 .init = ipv6_inetpeer_init,
3128 .exit = ipv6_inetpeer_exit,
3129};
3130
d189634e
TG		 3131static struct pernet_operations ip6_route_net_late_ops = {
3132 .init = ip6_route_net_init_late,
3133 .exit = ip6_route_net_exit_late,
3134};
3135
8ed67789
DL		 3136static struct notifier_block ip6_route_dev_notifier = {
3137 .notifier_call = ip6_route_dev_notify,
3138 .priority = 0,
3139};
3140
433d49c3 3141int __init ip6_route_init(void)
1da177e4 3142{
433d49c3
DL		 3143 int ret;
3144
9a7ec3a9
DL		 3145 ret = -ENOMEM;
3146 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3147 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3148 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3149 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3150 goto out;
14e50e57 3151
fc66f95c 3152 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3153 if (ret)
bdb3289f 3154 goto out_kmem_cache;
bdb3289f 3155
c3426b47
DM		 3156 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3157 if (ret)
e8803b6c 3158 goto out_dst_entries;
2a0c451a 3159
7e52b33b
DM		 3160 ret = register_pernet_subsys(&ip6_route_net_ops);
3161 if (ret)
3162 goto out_register_inetpeer;
c3426b47 3163
5dc121e9
AE		 3164 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3165
8ed67789
DL		 3166 /* Registering of the loopback is done before this portion of code,
3167 * the loopback reference in rt6_info will not be taken, do it
3168 * manually for init_net */
d8d1f30b 3169 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3170 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3171 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3172 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3173 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3174 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3175 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3176 #endif
e8803b6c 3177 ret = fib6_init();
433d49c3 3178 if (ret)
8ed67789 3179 goto out_register_subsys;
433d49c3 3180
433d49c3
DL		 3181 ret = xfrm6_init();
3182 if (ret)
e8803b6c 3183 goto out_fib6_init;
c35b7e72 3184
433d49c3
DL		 3185 ret = fib6_rules_init();
3186 if (ret)
3187 goto xfrm6_init;
7e5449c2 3188
d189634e
TG		 3189 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3190 if (ret)
3191 goto fib6_rules_init;
3192
433d49c3 3193 ret = -ENOBUFS;
c7ac8679
GR		 3194 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3195 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3196 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3197 goto out_register_late_subsys;
c127ea2c 3198
8ed67789 3199 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3200 if (ret)
d189634e 3201 goto out_register_late_subsys;
8ed67789 3202
433d49c3
DL		 3203out:
3204 return ret;
3205
d189634e
TG		 3206out_register_late_subsys:
3207 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3208fib6_rules_init:
433d49c3
DL		 3209 fib6_rules_cleanup();
3210xfrm6_init:
433d49c3 3211 xfrm6_fini();
2a0c451a
TG		 3212out_fib6_init:
3213 fib6_gc_cleanup();
8ed67789
DL		 3214out_register_subsys:
3215 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3216out_register_inetpeer:
3217 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3218out_dst_entries:
3219 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3220out_kmem_cache:
f2fc6a54 3221 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3222 goto out;
1da177e4
LT		 3223}
3224
3225void ip6_route_cleanup(void)
3226{
8ed67789 3227 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3228 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3229 fib6_rules_cleanup();
1da177e4 3230 xfrm6_fini();
1da177e4 3231 fib6_gc_cleanup();
c3426b47 3232 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3233 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3234 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3235 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3236}
Linux kernel - Deliverables
This page took 1.271646 seconds and 5 git commands to generate.