projects / linux-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
tipc: make the code look more readable
[linux-block.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);
87static int ip6_pkt_discard_out(struct sk_buff *skb);
7150aede
K		 88static int ip6_pkt_prohibit(struct sk_buff *skb);
89static int ip6_pkt_prohibit_out(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;
152 if (inet_metrics_new(peer))
153 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
154
155 new = (unsigned long) p;
156 prev = cmpxchg(&dst->_metrics, old, new);
157
158 if (prev != old) {
159 p = __DST_METRICS_PTR(prev);
160 if (prev & DST_METRICS_READ_ONLY)
161 p = NULL;
162 }
163 }
164 return p;
165}
166
f894cbf8
DM		 167static inline const void *choose_neigh_daddr(struct rt6_info *rt,
168 struct sk_buff *skb,
169 const void *daddr)
39232973
DM		 170{
171 struct in6_addr *p = &rt->rt6i_gateway;
172
a7563f34 173 if (!ipv6_addr_any(p))
39232973 174 return (const void *) p;
f894cbf8
DM		 175 else if (skb)
176 return &ipv6_hdr(skb)->daddr;
39232973
DM		 177 return daddr;
178}
179
f894cbf8
DM		 180static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
181 struct sk_buff *skb,
182 const void *daddr)
d3aaeb38 183{
39232973
DM		 184 struct rt6_info *rt = (struct rt6_info *) dst;
185 struct neighbour *n;
186
f894cbf8 187 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 188 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 189 if (n)
190 return n;
191 return neigh_create(&nd_tbl, daddr, dst->dev);
192}
193
9a7ec3a9 194static struct dst_ops ip6_dst_ops_template = {
1da177e4 195 .family = AF_INET6,
09640e63 196 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT		 197 .gc = ip6_dst_gc,
198 .gc_thresh = 1024,
199 .check = ip6_dst_check,
0dbaee3b 200 .default_advmss = ip6_default_advmss,
ebb762f2 201 .mtu = ip6_mtu,
06582540 202 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 203 .destroy = ip6_dst_destroy,
204 .ifdown = ip6_dst_ifdown,
205 .negative_advice = ip6_negative_advice,
206 .link_failure = ip6_link_failure,
207 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 208 .redirect = rt6_do_redirect,
1ac06e03 209 .local_out = __ip6_local_out,
d3aaeb38 210 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 211};
212
ebb762f2 213static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 214{
618f9bc7
SK		 215 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
216
217 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 218}
219
6700c270
DM		 220static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
221 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 222{
223}
224
6700c270
DM		 225static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
226 struct sk_buff *skb)
b587ee3b
DM		 227{
228}
229
0972ddb2
HB		 230static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
231 unsigned long old)
232{
233 return NULL;
234}
235
14e50e57
DM		 236static struct dst_ops ip6_dst_blackhole_ops = {
237 .family = AF_INET6,
09640e63 238 .protocol = cpu_to_be16(ETH_P_IPV6),
14e50e57
DM		 239 .destroy = ip6_dst_destroy,
240 .check = ip6_dst_check,
ebb762f2 241 .mtu = ip6_blackhole_mtu,
214f45c9 242 .default_advmss = ip6_default_advmss,
14e50e57 243 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 244 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 245 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 246 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 247};
248
62fa8a84 249static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 250 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 251};
252
fb0af4c7 253static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 254 .dst = {
255 .__refcnt = ATOMIC_INIT(1),
256 .__use = 1,
2c20cbd7 257 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 258 .error = -ENETUNREACH,
d8d1f30b
CG		 259 .input = ip6_pkt_discard,
260 .output = ip6_pkt_discard_out,
1da177e4
LT		 261 },
262 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 263 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 264 .rt6i_metric = ~(u32) 0,
265 .rt6i_ref = ATOMIC_INIT(1),
266};
267
101367c2
TG		 268#ifdef CONFIG_IPV6_MULTIPLE_TABLES
269
fb0af4c7 270static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 271 .dst = {
272 .__refcnt = ATOMIC_INIT(1),
273 .__use = 1,
2c20cbd7 274 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 275 .error = -EACCES,
d8d1f30b
CG		 276 .input = ip6_pkt_prohibit,
277 .output = ip6_pkt_prohibit_out,
101367c2
TG		 278 },
279 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 280 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 281 .rt6i_metric = ~(u32) 0,
282 .rt6i_ref = ATOMIC_INIT(1),
283};
284
fb0af4c7 285static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 286 .dst = {
287 .__refcnt = ATOMIC_INIT(1),
288 .__use = 1,
2c20cbd7 289 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 290 .error = -EINVAL,
d8d1f30b
CG		 291 .input = dst_discard,
292 .output = dst_discard,
101367c2
TG		 293 },
294 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 295 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 296 .rt6i_metric = ~(u32) 0,
297 .rt6i_ref = ATOMIC_INIT(1),
298};
299
300#endif
301
1da177e4 302/* allocate dst with ip6_dst_ops */
97bab73f 303static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 304 struct net_device *dev,
8b96d22d
DM		 305 int flags,
306 struct fib6_table *table)
1da177e4 307{
97bab73f 308 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 309 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 310
97bab73f 311 if (rt) {
8104891b
SK		 312 struct dst_entry *dst = &rt->dst;
313
314 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
8b96d22d 315 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
ca4c3fc2 316 rt->rt6i_genid = rt_genid_ipv6(net);
51ebd318 317 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 318 }
cf911662 319 return rt;
1da177e4
LT		 320}
321
322static void ip6_dst_destroy(struct dst_entry *dst)
323{
324 struct rt6_info *rt = (struct rt6_info *)dst;
325 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 326 struct dst_entry *from = dst->from;
1da177e4 327
8e2ec639
YZ		 328 if (!(rt->dst.flags & DST_HOST))
329 dst_destroy_metrics_generic(dst);
330
38308473 331 if (idev) {
1da177e4
LT		 332 rt->rt6i_idev = NULL;
333 in6_dev_put(idev);
1ab1457c 334 }
1716a961 335
ecd98837
YH		 336 dst->from = NULL;
337 dst_release(from);
1716a961 338
97bab73f
DM		 339 if (rt6_has_peer(rt)) {
340 struct inet_peer *peer = rt6_peer_ptr(rt);
b3419363
DM		 341 inet_putpeer(peer);
342 }
343}
344
1da177e4
LT		 345static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
346 int how)
347{
348 struct rt6_info *rt = (struct rt6_info *)dst;
349 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 350 struct net_device *loopback_dev =
c346dca1 351 dev_net(dev)->loopback_dev;
1da177e4 352
97cac082
DM		 353 if (dev != loopback_dev) {
354 if (idev && idev->dev == dev) {
355 struct inet6_dev *loopback_idev =
356 in6_dev_get(loopback_dev);
357 if (loopback_idev) {
358 rt->rt6i_idev = loopback_idev;
359 in6_dev_put(idev);
360 }
361 }
1da177e4
LT		 362 }
363}
364
a50feda5 365static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 366{
1716a961
G		 367 if (rt->rt6i_flags & RTF_EXPIRES) {
368 if (time_after(jiffies, rt->dst.expires))
a50feda5 369 return true;
1716a961 370 } else if (rt->dst.from) {
3fd91fb3 371 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 372 }
a50feda5 373 return false;
1da177e4
LT		 374}
375
a50feda5 376static bool rt6_need_strict(const struct in6_addr *daddr)
c71099ac 377{
a02cec21
ED		 378 return ipv6_addr_type(daddr) &
379 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
c71099ac
TG		 380}
381
51ebd318
ND		 382/* Multipath route selection:
383 * Hash based function using packet header and flowlabel.
384 * Adapted from fib_info_hashfn()
385 */
386static int rt6_info_hash_nhsfn(unsigned int candidate_count,
387 const struct flowi6 *fl6)
388{
389 unsigned int val = fl6->flowi6_proto;
390
c08977bb
YH		 391 val ^= ipv6_addr_hash(&fl6->daddr);
392 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 393
394 /* Work only if this not encapsulated */
395 switch (fl6->flowi6_proto) {
396 case IPPROTO_UDP:
397 case IPPROTO_TCP:
398 case IPPROTO_SCTP:
b3ce5ae1
ND		 399 val ^= (__force u16)fl6->fl6_sport;
400 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 401 break;
402
403 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 404 val ^= (__force u16)fl6->fl6_icmp_type;
405 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 406 break;
407 }
408 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 409 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 410
411 /* Perhaps, we need to tune, this function? */
412 val = val ^ (val >> 7) ^ (val >> 12);
413 return val % candidate_count;
414}
415
416static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 417 struct flowi6 *fl6, int oif,
418 int strict)
51ebd318
ND		 419{
420 struct rt6_info *sibling, *next_sibling;
421 int route_choosen;
422
423 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
424 /* Don't change the route, if route_choosen == 0
425 * (siblings does not include ourself)
426 */
427 if (route_choosen)
428 list_for_each_entry_safe(sibling, next_sibling,
429 &match->rt6i_siblings, rt6i_siblings) {
430 route_choosen--;
431 if (route_choosen == 0) {
52bd4c0c
ND		 432 if (rt6_score_route(sibling, oif, strict) < 0)
433 break;
51ebd318
ND		 434 match = sibling;
435 break;
436 }
437 }
438 return match;
439}
440
1da177e4 441/*
c71099ac 442 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 443 */
444
8ed67789
DL		 445static inline struct rt6_info *rt6_device_match(struct net *net,
446 struct rt6_info *rt,
b71d1d42 447 const struct in6_addr *saddr,
1da177e4 448 int oif,
d420895e 449 int flags)
1da177e4
LT		 450{
451 struct rt6_info *local = NULL;
452 struct rt6_info *sprt;
453
dd3abc4e
YH		 454 if (!oif && ipv6_addr_any(saddr))
455 goto out;
456
d8d1f30b 457 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 458 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 459
460 if (oif) {
1da177e4
LT		 461 if (dev->ifindex == oif)
462 return sprt;
463 if (dev->flags & IFF_LOOPBACK) {
38308473 464 if (!sprt->rt6i_idev ||
1da177e4 465 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 466 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 467 continue;
1ab1457c 468 if (local && (!oif ||
1da177e4
LT		 469 local->rt6i_idev->dev->ifindex == oif))
470 continue;
471 }
472 local = sprt;
473 }
dd3abc4e
YH		 474 } else {
475 if (ipv6_chk_addr(net, saddr, dev,
476 flags & RT6_LOOKUP_F_IFACE))
477 return sprt;
1da177e4 478 }
dd3abc4e 479 }
1da177e4 480
dd3abc4e 481 if (oif) {
1da177e4
LT		 482 if (local)
483 return local;
484
d420895e 485 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 486 return net->ipv6.ip6_null_entry;
1da177e4 487 }
dd3abc4e 488out:
1da177e4
LT		 489 return rt;
490}
491
27097255 492#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 493struct __rt6_probe_work {
494 struct work_struct work;
495 struct in6_addr target;
496 struct net_device *dev;
497};
498
499static void rt6_probe_deferred(struct work_struct *w)
500{
501 struct in6_addr mcaddr;
502 struct __rt6_probe_work *work =
503 container_of(w, struct __rt6_probe_work, work);
504
505 addrconf_addr_solict_mult(&work->target, &mcaddr);
506 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
507 dev_put(work->dev);
508 kfree(w);
509}
510
27097255
YH		 511static void rt6_probe(struct rt6_info *rt)
512{
f2c31e32 513 struct neighbour *neigh;
27097255
YH		 514 /*
515 * Okay, this does not seem to be appropriate
516 * for now, however, we need to check if it
517 * is really so; aka Router Reachability Probing.
518 *
519 * Router Reachability Probe MUST be rate-limited
520 * to no more than one per minute.
521 */
2152caea 522 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 523 return;
2152caea
YH		 524 rcu_read_lock_bh();
525 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
526 if (neigh) {
527 write_lock(&neigh->lock);
528 if (neigh->nud_state & NUD_VALID)
529 goto out;
7ff74a59 530 }
2152caea
YH		 531
532 if (!neigh ||
52e16356 533 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
c2f17e82 534 struct __rt6_probe_work *work;
27097255 535
c2f17e82
HFS		 536 work = kmalloc(sizeof(*work), GFP_ATOMIC);
537
538 if (neigh && work)
7e980569 539 __neigh_set_probe_once(neigh);
c2f17e82
HFS		 540
541 if (neigh)
2152caea
YH		 542 write_unlock(&neigh->lock);
543
c2f17e82
HFS		 544 if (work) {
545 INIT_WORK(&work->work, rt6_probe_deferred);
546 work->target = rt->rt6i_gateway;
547 dev_hold(rt->dst.dev);
548 work->dev = rt->dst.dev;
549 schedule_work(&work->work);
550 }
f2c31e32 551 } else {
2152caea
YH		 552out:
553 write_unlock(&neigh->lock);
f2c31e32 554 }
2152caea 555 rcu_read_unlock_bh();
27097255
YH		 556}
557#else
558static inline void rt6_probe(struct rt6_info *rt)
559{
27097255
YH		 560}
561#endif
562
1da177e4 563/*
554cfb7e 564 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 565 */
b6f99a21 566static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 567{
d1918542 568 struct net_device *dev = rt->dst.dev;
161980f4 569 if (!oif || dev->ifindex == oif)
554cfb7e 570 return 2;
161980f4
DM		 571 if ((dev->flags & IFF_LOOPBACK) &&
572 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
573 return 1;
574 return 0;
554cfb7e 575}
1da177e4 576
afc154e9 577static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 578{
f2c31e32 579 struct neighbour *neigh;
afc154e9 580 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 581
4d0c5911
YH		 582 if (rt->rt6i_flags & RTF_NONEXTHOP ||
583 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 584 return RT6_NUD_SUCCEED;
145a3621
YH		 585
586 rcu_read_lock_bh();
587 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
588 if (neigh) {
589 read_lock(&neigh->lock);
554cfb7e 590 if (neigh->nud_state & NUD_VALID)
afc154e9 591 ret = RT6_NUD_SUCCEED;
398bcbeb 592#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 593 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 594 ret = RT6_NUD_SUCCEED;
7e980569
JB		 595 else
596 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 597#endif
145a3621 598 read_unlock(&neigh->lock);
afc154e9
HFS		 599 } else {
600 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 601 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 602 }
145a3621
YH		 603 rcu_read_unlock_bh();
604
a5a81f0b 605 return ret;
1da177e4
LT		 606}
607
554cfb7e
YH		 608static int rt6_score_route(struct rt6_info *rt, int oif,
609 int strict)
1da177e4 610{
a5a81f0b 611 int m;
1ab1457c 612
4d0c5911 613 m = rt6_check_dev(rt, oif);
77d16f45 614 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 615 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 616#ifdef CONFIG_IPV6_ROUTER_PREF
617 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
618#endif
afc154e9
HFS		 619 if (strict & RT6_LOOKUP_F_REACHABLE) {
620 int n = rt6_check_neigh(rt);
621 if (n < 0)
622 return n;
623 }
554cfb7e
YH		 624 return m;
625}
626
f11e6659 627static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 628 int *mpri, struct rt6_info *match,
629 bool *do_rr)
554cfb7e 630{
f11e6659 631 int m;
afc154e9 632 bool match_do_rr = false;
f11e6659
DM		 633
634 if (rt6_check_expired(rt))
635 goto out;
636
637 m = rt6_score_route(rt, oif, strict);
7e980569 638 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 639 match_do_rr = true;
640 m = 0; /* lowest valid score */
7e980569 641 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 642 goto out;
afc154e9
HFS		 643 }
644
645 if (strict & RT6_LOOKUP_F_REACHABLE)
646 rt6_probe(rt);
f11e6659 647
7e980569 648 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 649 if (m > *mpri) {
afc154e9 650 *do_rr = match_do_rr;
f11e6659
DM		 651 *mpri = m;
652 match = rt;
f11e6659 653 }
f11e6659
DM		 654out:
655 return match;
656}
657
658static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
659 struct rt6_info *rr_head,
afc154e9
HFS		 660 u32 metric, int oif, int strict,
661 bool *do_rr)
f11e6659
DM		 662{
663 struct rt6_info *rt, *match;
554cfb7e 664 int mpri = -1;
1da177e4 665
f11e6659
DM		 666 match = NULL;
667 for (rt = rr_head; rt && rt->rt6i_metric == metric;
d8d1f30b 668 rt = rt->dst.rt6_next)
afc154e9 669 match = find_match(rt, oif, strict, &mpri, match, do_rr);
f11e6659 670 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
d8d1f30b 671 rt = rt->dst.rt6_next)
afc154e9 672 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 673
f11e6659
DM		 674 return match;
675}
1da177e4 676
f11e6659
DM		 677static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
678{
679 struct rt6_info *match, *rt0;
8ed67789 680 struct net *net;
afc154e9 681 bool do_rr = false;
1da177e4 682
f11e6659
DM		 683 rt0 = fn->rr_ptr;
684 if (!rt0)
685 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 686
afc154e9
HFS		 687 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
688 &do_rr);
1da177e4 689
afc154e9 690 if (do_rr) {
d8d1f30b 691 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 692
554cfb7e 693 /* no entries matched; do round-robin */
f11e6659
DM		 694 if (!next || next->rt6i_metric != rt0->rt6i_metric)
695 next = fn->leaf;
696
697 if (next != rt0)
698 fn->rr_ptr = next;
1da177e4 699 }
1da177e4 700
d1918542 701 net = dev_net(rt0->dst.dev);
a02cec21 702 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 703}
704
70ceb4f5
YH		 705#ifdef CONFIG_IPV6_ROUTE_INFO
706int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 707 const struct in6_addr *gwaddr)
70ceb4f5 708{
c346dca1 709 struct net *net = dev_net(dev);
70ceb4f5
YH		 710 struct route_info *rinfo = (struct route_info *) opt;
711 struct in6_addr prefix_buf, *prefix;
712 unsigned int pref;
4bed72e4 713 unsigned long lifetime;
70ceb4f5
YH		 714 struct rt6_info *rt;
715
716 if (len < sizeof(struct route_info)) {
717 return -EINVAL;
718 }
719
720 /* Sanity check for prefix_len and length */
721 if (rinfo->length > 3) {
722 return -EINVAL;
723 } else if (rinfo->prefix_len > 128) {
724 return -EINVAL;
725 } else if (rinfo->prefix_len > 64) {
726 if (rinfo->length < 2) {
727 return -EINVAL;
728 }
729 } else if (rinfo->prefix_len > 0) {
730 if (rinfo->length < 1) {
731 return -EINVAL;
732 }
733 }
734
735 pref = rinfo->route_pref;
736 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 737 return -EINVAL;
70ceb4f5 738
4bed72e4 739 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 740
741 if (rinfo->length == 3)
742 prefix = (struct in6_addr *)rinfo->prefix;
743 else {
744 /* this function is safe */
745 ipv6_addr_prefix(&prefix_buf,
746 (struct in6_addr *)rinfo->prefix,
747 rinfo->prefix_len);
748 prefix = &prefix_buf;
749 }
750
f104a567
DJ		 751 if (rinfo->prefix_len == 0)
752 rt = rt6_get_dflt_router(gwaddr, dev);
753 else
754 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
755 gwaddr, dev->ifindex);
70ceb4f5
YH		 756
757 if (rt && !lifetime) {
e0a1ad73 758 ip6_del_rt(rt);
70ceb4f5
YH		 759 rt = NULL;
760 }
761
762 if (!rt && lifetime)
efa2cea0 763 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 764 pref);
765 else if (rt)
766 rt->rt6i_flags = RTF_ROUTEINFO |
767 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
768
769 if (rt) {
1716a961
G		 770 if (!addrconf_finite_timeout(lifetime))
771 rt6_clean_expires(rt);
772 else
773 rt6_set_expires(rt, jiffies + HZ * lifetime);
774
94e187c0 775 ip6_rt_put(rt);
70ceb4f5
YH		 776 }
777 return 0;
778}
779#endif
780
8ed67789 781#define BACKTRACK(__net, saddr) \
982f56f3 782do { \
8ed67789 783 if (rt == __net->ipv6.ip6_null_entry) { \
982f56f3 784 struct fib6_node *pn; \
e0eda7bb 785 while (1) { \
982f56f3
YH		 786 if (fn->fn_flags & RTN_TL_ROOT) \
787 goto out; \
788 pn = fn->parent; \
789 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
8bce65b9 790 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
982f56f3
YH		 791 else \
792 fn = pn; \
793 if (fn->fn_flags & RTN_RTINFO) \
794 goto restart; \
c71099ac 795 } \
c71099ac 796 } \
38308473 797} while (0)
c71099ac 798
8ed67789
DL		 799static struct rt6_info *ip6_pol_route_lookup(struct net *net,
800 struct fib6_table *table,
4c9483b2 801 struct flowi6 *fl6, int flags)
1da177e4
LT		 802{
803 struct fib6_node *fn;
804 struct rt6_info *rt;
805
c71099ac 806 read_lock_bh(&table->tb6_lock);
4c9483b2 807 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 808restart:
809 rt = fn->leaf;
4c9483b2 810 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 811 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 812 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
4c9483b2 813 BACKTRACK(net, &fl6->saddr);
c71099ac 814out:
d8d1f30b 815 dst_use(&rt->dst, jiffies);
c71099ac 816 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 817 return rt;
818
819}
820
ea6e574e
FW		 821struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6,
822 int flags)
823{
824 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
825}
826EXPORT_SYMBOL_GPL(ip6_route_lookup);
827
9acd9f3a
YH		 828struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
829 const struct in6_addr *saddr, int oif, int strict)
c71099ac 830{
4c9483b2
DM		 831 struct flowi6 fl6 = {
832 .flowi6_oif = oif,
833 .daddr = *daddr,
c71099ac
TG		 834 };
835 struct dst_entry *dst;
77d16f45 836 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 837
adaa70bb 838 if (saddr) {
4c9483b2 839 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 840 flags |= RT6_LOOKUP_F_HAS_SADDR;
841 }
842
4c9483b2 843 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 844 if (dst->error == 0)
845 return (struct rt6_info *) dst;
846
847 dst_release(dst);
848
1da177e4
LT		 849 return NULL;
850}
851
7159039a
YH		 852EXPORT_SYMBOL(rt6_lookup);
853
c71099ac 854/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 855 It takes new route entry, the addition fails by any reason the
856 route is freed. In any case, if caller does not hold it, it may
857 be destroyed.
858 */
859
86872cb5 860static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 861{
862 int err;
c71099ac 863 struct fib6_table *table;
1da177e4 864
c71099ac
TG		 865 table = rt->rt6i_table;
866 write_lock_bh(&table->tb6_lock);
86872cb5 867 err = fib6_add(&table->tb6_root, rt, info);
c71099ac 868 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 869
870 return err;
871}
872
40e22e8f
TG		 873int ip6_ins_rt(struct rt6_info *rt)
874{
4d1169c1 875 struct nl_info info = {
d1918542 876 .nl_net = dev_net(rt->dst.dev),
4d1169c1 877 };
528c4ceb 878 return __ip6_ins_rt(rt, &info);
40e22e8f
TG		 879}
880
1716a961 881static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
21efcfa0 882 const struct in6_addr *daddr,
b71d1d42 883 const struct in6_addr *saddr)
1da177e4 884{
1da177e4
LT		 885 struct rt6_info *rt;
886
887 /*
888 * Clone the route.
889 */
890
21efcfa0 891 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 892
893 if (rt) {
249a3630
DJ		 894 if (ort->rt6i_dst.plen != 128 &&
895 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
896 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 897
1da177e4 898 rt->rt6i_flags |= RTF_CACHE;
1da177e4
LT		 899
900#ifdef CONFIG_IPV6_SUBTREES
901 if (rt->rt6i_src.plen && saddr) {
4e3fd7a0 902 rt->rt6i_src.addr = *saddr;
1da177e4
LT		 903 rt->rt6i_src.plen = 128;
904 }
905#endif
95a9a5ba 906 }
1da177e4 907
95a9a5ba
YH		 908 return rt;
909}
1da177e4 910
21efcfa0
ED		 911static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
912 const struct in6_addr *daddr)
299d9939 913{
21efcfa0
ED		 914 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
915
887c95cc 916 if (rt)
299d9939 917 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 918 return rt;
919}
920
8ed67789 921static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 922 struct flowi6 *fl6, int flags)
1da177e4
LT		 923{
924 struct fib6_node *fn;
519fbd87 925 struct rt6_info *rt, *nrt;
c71099ac 926 int strict = 0;
1da177e4 927 int attempts = 3;
519fbd87 928 int err;
53b7997f 929 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
1da177e4 930
77d16f45 931 strict |= flags & RT6_LOOKUP_F_IFACE;
1da177e4
LT		 932
933relookup:
c71099ac 934 read_lock_bh(&table->tb6_lock);
1da177e4 935
8238dd06 936restart_2:
4c9483b2 937 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1da177e4
LT		 938
939restart:
4acad72d 940 rt = rt6_select(fn, oif, strict | reachable);
52bd4c0c
ND		 941 if (rt->rt6i_nsiblings)
942 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable);
4c9483b2 943 BACKTRACK(net, &fl6->saddr);
8ed67789 944 if (rt == net->ipv6.ip6_null_entry ||
8238dd06 945 rt->rt6i_flags & RTF_CACHE)
1ddef044 946 goto out;
1da177e4 947
d8d1f30b 948 dst_hold(&rt->dst);
c71099ac 949 read_unlock_bh(&table->tb6_lock);
fb9de91e 950
c440f160 951 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
4c9483b2 952 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
7343ff31 953 else if (!(rt->dst.flags & DST_HOST))
4c9483b2 954 nrt = rt6_alloc_clone(rt, &fl6->daddr);
7343ff31
DM		 955 else
956 goto out2;
e40cf353 957
94e187c0 958 ip6_rt_put(rt);
8ed67789 959 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 960
d8d1f30b 961 dst_hold(&rt->dst);
519fbd87 962 if (nrt) {
40e22e8f 963 err = ip6_ins_rt(nrt);
519fbd87 964 if (!err)
1da177e4 965 goto out2;
1da177e4 966 }
1da177e4 967
519fbd87
YH		 968 if (--attempts <= 0)
969 goto out2;
970
971 /*
c71099ac 972 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 973 * released someone could insert this route. Relookup.
974 */
94e187c0 975 ip6_rt_put(rt);
519fbd87
YH		 976 goto relookup;
977
978out:
8238dd06
YH		 979 if (reachable) {
980 reachable = 0;
981 goto restart_2;
982 }
d8d1f30b 983 dst_hold(&rt->dst);
c71099ac 984 read_unlock_bh(&table->tb6_lock);
1da177e4 985out2:
d8d1f30b
CG		 986 rt->dst.lastuse = jiffies;
987 rt->dst.__use++;
c71099ac
TG		 988
989 return rt;
1da177e4
LT		 990}
991
8ed67789 992static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 993 struct flowi6 *fl6, int flags)
4acad72d 994{
4c9483b2 995 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 996}
997
72331bc0
SL		 998static struct dst_entry *ip6_route_input_lookup(struct net *net,
999 struct net_device *dev,
1000 struct flowi6 *fl6, int flags)
1001{
1002 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1003 flags |= RT6_LOOKUP_F_IFACE;
1004
1005 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1006}
1007
c71099ac
TG		 1008void ip6_route_input(struct sk_buff *skb)
1009{
b71d1d42 1010 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1011 struct net *net = dev_net(skb->dev);
adaa70bb 1012 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 1013 struct flowi6 fl6 = {
1014 .flowi6_iif = skb->dev->ifindex,
1015 .daddr = iph->daddr,
1016 .saddr = iph->saddr,
6502ca52 1017 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1018 .flowi6_mark = skb->mark,
1019 .flowi6_proto = iph->nexthdr,
c71099ac 1020 };
adaa70bb 1021
72331bc0 1022 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1023}
1024
8ed67789 1025static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1026 struct flowi6 *fl6, int flags)
1da177e4 1027{
4c9483b2 1028 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1029}
1030
9c7a4f9c 1031struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 1032 struct flowi6 *fl6)
c71099ac
TG		 1033{
1034 int flags = 0;
1035
1fb9489b 1036 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1037
4c9483b2 1038 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 1039 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1040
4c9483b2 1041 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 1042 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1043 else if (sk)
1044 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1045
4c9483b2 1046 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4
LT		 1047}
1048
7159039a 1049EXPORT_SYMBOL(ip6_route_output);
1da177e4 1050
2774c131 1051struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1052{
5c1e6aa3 1053 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1054 struct dst_entry *new = NULL;
1055
f5b0a874 1056 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1057 if (rt) {
d8d1f30b 1058 new = &rt->dst;
14e50e57 1059
8104891b
SK		 1060 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1061 rt6_init_peer(rt, net->ipv6.peers);
1062
14e50e57 1063 new->__use = 1;
352e512c
HX		 1064 new->input = dst_discard;
1065 new->output = dst_discard;
14e50e57 1066
21efcfa0
ED		 1067 if (dst_metrics_read_only(&ort->dst))
1068 new->_metrics = ort->dst._metrics;
1069 else
1070 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1071 rt->rt6i_idev = ort->rt6i_idev;
1072 if (rt->rt6i_idev)
1073 in6_dev_hold(rt->rt6i_idev);
14e50e57 1074
4e3fd7a0 1075 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1076 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1077 rt->rt6i_metric = 0;
1078
1079 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1080#ifdef CONFIG_IPV6_SUBTREES
1081 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1082#endif
1083
1084 dst_free(new);
1085 }
1086
69ead7af
DM		 1087 dst_release(dst_orig);
1088 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1089}
14e50e57 1090
1da177e4
LT		 1091/*
1092 * Destination cache support functions
1093 */
1094
1095static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1096{
1097 struct rt6_info *rt;
1098
1099 rt = (struct rt6_info *) dst;
1100
6f3118b5
ND		 1101 /* All IPV6 dsts are created with ->obsolete set to the value
1102 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1103 * into this function always.
1104 */
ca4c3fc2 1105 if (rt->rt6i_genid != rt_genid_ipv6(dev_net(rt->dst.dev)))
6f3118b5
ND		 1106 return NULL;
1107
e3bc10bd
HFS		 1108 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1109 return NULL;
a4477c4d 1110
e3bc10bd
HFS		 1111 if (rt6_check_expired(rt))
1112 return NULL;
1113
1114 return dst;
1da177e4
LT		 1115}
1116
1117static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1118{
1119 struct rt6_info *rt = (struct rt6_info *) dst;
1120
1121 if (rt) {
54c1a859
YH		 1122 if (rt->rt6i_flags & RTF_CACHE) {
1123 if (rt6_check_expired(rt)) {
1124 ip6_del_rt(rt);
1125 dst = NULL;
1126 }
1127 } else {
1da177e4 1128 dst_release(dst);
54c1a859
YH		 1129 dst = NULL;
1130 }
1da177e4 1131 }
54c1a859 1132 return dst;
1da177e4
LT		 1133}
1134
1135static void ip6_link_failure(struct sk_buff *skb)
1136{
1137 struct rt6_info *rt;
1138
3ffe533c 1139 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1140
adf30907 1141 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1142 if (rt) {
1eb4f758
HFS		 1143 if (rt->rt6i_flags & RTF_CACHE) {
1144 dst_hold(&rt->dst);
1145 if (ip6_del_rt(rt))
1146 dst_free(&rt->dst);
1147 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1148 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1149 }
1da177e4
LT		 1150 }
1151}
1152
6700c270
DM		 1153static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1154 struct sk_buff *skb, u32 mtu)
1da177e4
LT		 1155{
1156 struct rt6_info *rt6 = (struct rt6_info*)dst;
1157
81aded24 1158 dst_confirm(dst);
1da177e4 1159 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
81aded24
DM		 1160 struct net *net = dev_net(dst->dev);
1161
1da177e4
LT		 1162 rt6->rt6i_flags |= RTF_MODIFIED;
1163 if (mtu < IPV6_MIN_MTU) {
defb3519 1164 u32 features = dst_metric(dst, RTAX_FEATURES);
1da177e4 1165 mtu = IPV6_MIN_MTU;
defb3519
DM		 1166 features |= RTAX_FEATURE_ALLFRAG;
1167 dst_metric_set(dst, RTAX_FEATURES, features);
1da177e4 1168 }
defb3519 1169 dst_metric_set(dst, RTAX_MTU, mtu);
81aded24 1170 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1171 }
1172}
1173
42ae66c8
DM		 1174void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1175 int oif, u32 mark)
81aded24
DM		 1176{
1177 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1178 struct dst_entry *dst;
1179 struct flowi6 fl6;
1180
1181 memset(&fl6, 0, sizeof(fl6));
1182 fl6.flowi6_oif = oif;
1183 fl6.flowi6_mark = mark;
81aded24
DM		 1184 fl6.daddr = iph->daddr;
1185 fl6.saddr = iph->saddr;
6502ca52 1186 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1187
1188 dst = ip6_route_output(net, NULL, &fl6);
1189 if (!dst->error)
6700c270 1190 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
81aded24
DM		 1191 dst_release(dst);
1192}
1193EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1194
1195void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1196{
1197 ip6_update_pmtu(skb, sock_net(sk), mtu,
1198 sk->sk_bound_dev_if, sk->sk_mark);
1199}
1200EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1201
b55b76b2
DJ		 1202/* Handle redirects */
1203struct ip6rd_flowi {
1204 struct flowi6 fl6;
1205 struct in6_addr gateway;
1206};
1207
1208static struct rt6_info *__ip6_route_redirect(struct net *net,
1209 struct fib6_table *table,
1210 struct flowi6 *fl6,
1211 int flags)
1212{
1213 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1214 struct rt6_info *rt;
1215 struct fib6_node *fn;
1216
1217 /* Get the "current" route for this destination and
1218 * check if the redirect has come from approriate router.
1219 *
1220 * RFC 4861 specifies that redirects should only be
1221 * accepted if they come from the nexthop to the target.
1222 * Due to the way the routes are chosen, this notion
1223 * is a bit fuzzy and one might need to check all possible
1224 * routes.
1225 */
1226
1227 read_lock_bh(&table->tb6_lock);
1228 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1229restart:
1230 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1231 if (rt6_check_expired(rt))
1232 continue;
1233 if (rt->dst.error)
1234 break;
1235 if (!(rt->rt6i_flags & RTF_GATEWAY))
1236 continue;
1237 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1238 continue;
1239 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1240 continue;
1241 break;
1242 }
1243
1244 if (!rt)
1245 rt = net->ipv6.ip6_null_entry;
1246 else if (rt->dst.error) {
1247 rt = net->ipv6.ip6_null_entry;
1248 goto out;
1249 }
1250 BACKTRACK(net, &fl6->saddr);
1251out:
1252 dst_hold(&rt->dst);
1253
1254 read_unlock_bh(&table->tb6_lock);
1255
1256 return rt;
1257};
1258
1259static struct dst_entry *ip6_route_redirect(struct net *net,
1260 const struct flowi6 *fl6,
1261 const struct in6_addr *gateway)
1262{
1263 int flags = RT6_LOOKUP_F_HAS_SADDR;
1264 struct ip6rd_flowi rdfl;
1265
1266 rdfl.fl6 = *fl6;
1267 rdfl.gateway = *gateway;
1268
1269 return fib6_rule_lookup(net, &rdfl.fl6,
1270 flags, __ip6_route_redirect);
1271}
1272
3a5ad2ee
DM		 1273void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1274{
1275 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1276 struct dst_entry *dst;
1277 struct flowi6 fl6;
1278
1279 memset(&fl6, 0, sizeof(fl6));
1280 fl6.flowi6_oif = oif;
1281 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1282 fl6.daddr = iph->daddr;
1283 fl6.saddr = iph->saddr;
6502ca52 1284 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1285
b55b76b2
DJ		 1286 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1287 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1288 dst_release(dst);
1289}
1290EXPORT_SYMBOL_GPL(ip6_redirect);
1291
c92a59ec
DJ		 1292void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1293 u32 mark)
1294{
1295 const struct ipv6hdr *iph = ipv6_hdr(skb);
1296 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1297 struct dst_entry *dst;
1298 struct flowi6 fl6;
1299
1300 memset(&fl6, 0, sizeof(fl6));
1301 fl6.flowi6_oif = oif;
1302 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1303 fl6.daddr = msg->dest;
1304 fl6.saddr = iph->daddr;
1305
b55b76b2
DJ		 1306 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1307 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1308 dst_release(dst);
1309}
1310
3a5ad2ee
DM		 1311void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1312{
1313 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1314}
1315EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1316
0dbaee3b 1317static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1318{
0dbaee3b
DM		 1319 struct net_device *dev = dst->dev;
1320 unsigned int mtu = dst_mtu(dst);
1321 struct net *net = dev_net(dev);
1322
1da177e4
LT		 1323 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1324
5578689a
DL		 1325 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1326 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1327
1328 /*
1ab1457c
YH		 1329 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1330 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1331 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1332 * rely only on pmtu discovery"
1333 */
1334 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1335 mtu = IPV6_MAXPLEN;
1336 return mtu;
1337}
1338
ebb762f2 1339static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1340{
d33e4553 1341 struct inet6_dev *idev;
618f9bc7
SK		 1342 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1343
1344 if (mtu)
1345 return mtu;
1346
1347 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1348
1349 rcu_read_lock();
1350 idev = __in6_dev_get(dst->dev);
1351 if (idev)
1352 mtu = idev->cnf.mtu6;
1353 rcu_read_unlock();
1354
1355 return mtu;
1356}
1357
3b00944c
YH		 1358static struct dst_entry *icmp6_dst_gc_list;
1359static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1360
3b00944c 1361struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1362 struct flowi6 *fl6)
1da177e4 1363{
87a11578 1364 struct dst_entry *dst;
1da177e4
LT		 1365 struct rt6_info *rt;
1366 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1367 struct net *net = dev_net(dev);
1da177e4 1368
38308473 1369 if (unlikely(!idev))
122bdf67 1370 return ERR_PTR(-ENODEV);
1da177e4 1371
8b96d22d 1372 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1373 if (unlikely(!rt)) {
1da177e4 1374 in6_dev_put(idev);
87a11578 1375 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1376 goto out;
1377 }
1378
8e2ec639
YZ		 1379 rt->dst.flags |= DST_HOST;
1380 rt->dst.output = ip6_output;
d8d1f30b 1381 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1382 rt->rt6i_gateway = fl6->daddr;
87a11578 1383 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1384 rt->rt6i_dst.plen = 128;
1385 rt->rt6i_idev = idev;
14edd87d 1386 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1387
3b00944c 1388 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1389 rt->dst.next = icmp6_dst_gc_list;
1390 icmp6_dst_gc_list = &rt->dst;
3b00944c 1391 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1392
5578689a 1393 fib6_force_start_gc(net);
1da177e4 1394
87a11578
DM		 1395 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1396
1da177e4 1397out:
87a11578 1398 return dst;
1da177e4
LT		 1399}
1400
3d0f24a7 1401int icmp6_dst_gc(void)
1da177e4 1402{
e9476e95 1403 struct dst_entry *dst, **pprev;
3d0f24a7 1404 int more = 0;
1da177e4 1405
3b00944c
YH		 1406 spin_lock_bh(&icmp6_dst_lock);
1407 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1408
1da177e4
LT		 1409 while ((dst = *pprev) != NULL) {
1410 if (!atomic_read(&dst->__refcnt)) {
1411 *pprev = dst->next;
1412 dst_free(dst);
1da177e4
LT		 1413 } else {
1414 pprev = &dst->next;
3d0f24a7 1415 ++more;
1da177e4
LT		 1416 }
1417 }
1418
3b00944c 1419 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1420
3d0f24a7 1421 return more;
1da177e4
LT		 1422}
1423
1e493d19
DM		 1424static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1425 void *arg)
1426{
1427 struct dst_entry *dst, **pprev;
1428
1429 spin_lock_bh(&icmp6_dst_lock);
1430 pprev = &icmp6_dst_gc_list;
1431 while ((dst = *pprev) != NULL) {
1432 struct rt6_info *rt = (struct rt6_info *) dst;
1433 if (func(rt, arg)) {
1434 *pprev = dst->next;
1435 dst_free(dst);
1436 } else {
1437 pprev = &dst->next;
1438 }
1439 }
1440 spin_unlock_bh(&icmp6_dst_lock);
1441}
1442
569d3645 1443static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1444{
86393e52 1445 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1446 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1447 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1448 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1449 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1450 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1451 int entries;
7019b78e 1452
fc66f95c 1453 entries = dst_entries_get_fast(ops);
49a18d86 1454 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1455 entries <= rt_max_size)
1da177e4
LT		 1456 goto out;
1457
6891a346 1458 net->ipv6.ip6_rt_gc_expire++;
2ac3ac8f 1459 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size);
fc66f95c
ED		 1460 entries = dst_entries_get_slow(ops);
1461 if (entries < ops->gc_thresh)
7019b78e 1462 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1463out:
7019b78e 1464 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1465 return entries > rt_max_size;
1da177e4
LT		 1466}
1467
1da177e4
LT		 1468/*
1469 *
1470 */
1471
86872cb5 1472int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1473{
1474 int err;
5578689a 1475 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1476 struct rt6_info *rt = NULL;
1477 struct net_device *dev = NULL;
1478 struct inet6_dev *idev = NULL;
c71099ac 1479 struct fib6_table *table;
1da177e4
LT		 1480 int addr_type;
1481
86872cb5 1482 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1483 return -EINVAL;
1484#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1485 if (cfg->fc_src_len)
1da177e4
LT		 1486 return -EINVAL;
1487#endif
86872cb5 1488 if (cfg->fc_ifindex) {
1da177e4 1489 err = -ENODEV;
5578689a 1490 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1491 if (!dev)
1492 goto out;
1493 idev = in6_dev_get(dev);
1494 if (!idev)
1495 goto out;
1496 }
1497
86872cb5
TG		 1498 if (cfg->fc_metric == 0)
1499 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1500
d71314b4 1501 err = -ENOBUFS;
38308473
DM		 1502 if (cfg->fc_nlinfo.nlh &&
1503 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1504 table = fib6_get_table(net, cfg->fc_table);
38308473 1505 if (!table) {
f3213831 1506 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1507 table = fib6_new_table(net, cfg->fc_table);
1508 }
1509 } else {
1510 table = fib6_new_table(net, cfg->fc_table);
1511 }
38308473
DM		 1512
1513 if (!table)
c71099ac 1514 goto out;
c71099ac 1515
8b96d22d 1516 rt = ip6_dst_alloc(net, NULL, DST_NOCOUNT, table);
1da177e4 1517
38308473 1518 if (!rt) {
1da177e4
LT		 1519 err = -ENOMEM;
1520 goto out;
1521 }
1522
1716a961
G		 1523 if (cfg->fc_flags & RTF_EXPIRES)
1524 rt6_set_expires(rt, jiffies +
1525 clock_t_to_jiffies(cfg->fc_expires));
1526 else
1527 rt6_clean_expires(rt);
1da177e4 1528
86872cb5
TG		 1529 if (cfg->fc_protocol == RTPROT_UNSPEC)
1530 cfg->fc_protocol = RTPROT_BOOT;
1531 rt->rt6i_protocol = cfg->fc_protocol;
1532
1533 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1534
1535 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1536 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1537 else if (cfg->fc_flags & RTF_LOCAL)
1538 rt->dst.input = ip6_input;
1da177e4 1539 else
d8d1f30b 1540 rt->dst.input = ip6_forward;
1da177e4 1541
d8d1f30b 1542 rt->dst.output = ip6_output;
1da177e4 1543
86872cb5
TG		 1544 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1545 rt->rt6i_dst.plen = cfg->fc_dst_len;
1da177e4 1546 if (rt->rt6i_dst.plen == 128)
11d53b49 1547 rt->dst.flags |= DST_HOST;
1da177e4 1548
8e2ec639
YZ		 1549 if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) {
1550 u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1551 if (!metrics) {
1552 err = -ENOMEM;
1553 goto out;
1554 }
1555 dst_init_metrics(&rt->dst, metrics, 0);
1556 }
1da177e4 1557#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1558 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1559 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1560#endif
1561
86872cb5 1562 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1563
1564 /* We cannot add true routes via loopback here,
1565 they would result in kernel looping; promote them to reject routes
1566 */
86872cb5 1567 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1568 (dev && (dev->flags & IFF_LOOPBACK) &&
1569 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1570 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1571 /* hold loopback dev/idev if we haven't done so. */
5578689a 1572 if (dev != net->loopback_dev) {
1da177e4
LT		 1573 if (dev) {
1574 dev_put(dev);
1575 in6_dev_put(idev);
1576 }
5578689a 1577 dev = net->loopback_dev;
1da177e4
LT		 1578 dev_hold(dev);
1579 idev = in6_dev_get(dev);
1580 if (!idev) {
1581 err = -ENODEV;
1582 goto out;
1583 }
1584 }
1da177e4 1585 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1586 switch (cfg->fc_type) {
1587 case RTN_BLACKHOLE:
1588 rt->dst.error = -EINVAL;
7150aede
K		 1589 rt->dst.output = dst_discard;
1590 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1591 break;
1592 case RTN_PROHIBIT:
1593 rt->dst.error = -EACCES;
7150aede
K		 1594 rt->dst.output = ip6_pkt_prohibit_out;
1595 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1596 break;
b4949ab2 1597 case RTN_THROW:
ef2c7d7b 1598 default:
7150aede
K		 1599 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1600 : -ENETUNREACH;
1601 rt->dst.output = ip6_pkt_discard_out;
1602 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1603 break;
1604 }
1da177e4
LT		 1605 goto install_route;
1606 }
1607
86872cb5 1608 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1609 const struct in6_addr *gw_addr;
1da177e4
LT		 1610 int gwa_type;
1611
86872cb5 1612 gw_addr = &cfg->fc_gateway;
4e3fd7a0 1613 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1614 gwa_type = ipv6_addr_type(gw_addr);
1615
1616 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1617 struct rt6_info *grt;
1618
1619 /* IPv6 strictly inhibits using not link-local
1620 addresses as nexthop address.
1621 Otherwise, router will not able to send redirects.
1622 It is very good, but in some (rare!) circumstances
1623 (SIT, PtP, NBMA NOARP links) it is handy to allow
1624 some exceptions. --ANK
1625 */
1626 err = -EINVAL;
38308473 1627 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1628 goto out;
1629
5578689a 1630 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1631
1632 err = -EHOSTUNREACH;
38308473 1633 if (!grt)
1da177e4
LT		 1634 goto out;
1635 if (dev) {
d1918542 1636 if (dev != grt->dst.dev) {
94e187c0 1637 ip6_rt_put(grt);
1da177e4
LT		 1638 goto out;
1639 }
1640 } else {
d1918542 1641 dev = grt->dst.dev;
1da177e4
LT		 1642 idev = grt->rt6i_idev;
1643 dev_hold(dev);
1644 in6_dev_hold(grt->rt6i_idev);
1645 }
38308473 1646 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1647 err = 0;
94e187c0 1648 ip6_rt_put(grt);
1da177e4
LT		 1649
1650 if (err)
1651 goto out;
1652 }
1653 err = -EINVAL;
38308473 1654 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1655 goto out;
1656 }
1657
1658 err = -ENODEV;
38308473 1659 if (!dev)
1da177e4
LT		 1660 goto out;
1661
c3968a85
DW		 1662 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1663 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1664 err = -EINVAL;
1665 goto out;
1666 }
4e3fd7a0 1667 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1668 rt->rt6i_prefsrc.plen = 128;
1669 } else
1670 rt->rt6i_prefsrc.plen = 0;
1671
86872cb5 1672 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1673
1674install_route:
86872cb5
TG		 1675 if (cfg->fc_mx) {
1676 struct nlattr *nla;
1677 int remaining;
1678
1679 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
8f4c1f9b 1680 int type = nla_type(nla);
86872cb5
TG		 1681
1682 if (type) {
1683 if (type > RTAX_MAX) {
1da177e4
LT		 1684 err = -EINVAL;
1685 goto out;
1686 }
86872cb5 1687
defb3519 1688 dst_metric_set(&rt->dst, type, nla_get_u32(nla));
1da177e4 1689 }
1da177e4
LT		 1690 }
1691 }
1692
d8d1f30b 1693 rt->dst.dev = dev;
1da177e4 1694 rt->rt6i_idev = idev;
c71099ac 1695 rt->rt6i_table = table;
63152fc0 1696
c346dca1 1697 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1698
86872cb5 1699 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1700
1701out:
1702 if (dev)
1703 dev_put(dev);
1704 if (idev)
1705 in6_dev_put(idev);
1706 if (rt)
d8d1f30b 1707 dst_free(&rt->dst);
1da177e4
LT		 1708 return err;
1709}
1710
86872cb5 1711static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1712{
1713 int err;
c71099ac 1714 struct fib6_table *table;
d1918542 1715 struct net *net = dev_net(rt->dst.dev);
1da177e4 1716
6825a26c
G		 1717 if (rt == net->ipv6.ip6_null_entry) {
1718 err = -ENOENT;
1719 goto out;
1720 }
6c813a72 1721
c71099ac
TG		 1722 table = rt->rt6i_table;
1723 write_lock_bh(&table->tb6_lock);
86872cb5 1724 err = fib6_del(rt, info);
c71099ac 1725 write_unlock_bh(&table->tb6_lock);
1da177e4 1726
6825a26c 1727out:
94e187c0 1728 ip6_rt_put(rt);
1da177e4
LT		 1729 return err;
1730}
1731
e0a1ad73
TG		 1732int ip6_del_rt(struct rt6_info *rt)
1733{
4d1169c1 1734 struct nl_info info = {
d1918542 1735 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1736 };
528c4ceb 1737 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1738}
1739
86872cb5 1740static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1741{
c71099ac 1742 struct fib6_table *table;
1da177e4
LT		 1743 struct fib6_node *fn;
1744 struct rt6_info *rt;
1745 int err = -ESRCH;
1746
5578689a 1747 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1748 if (!table)
c71099ac
TG		 1749 return err;
1750
1751 read_lock_bh(&table->tb6_lock);
1da177e4 1752
c71099ac 1753 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1754 &cfg->fc_dst, cfg->fc_dst_len,
1755 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1756
1da177e4 1757 if (fn) {
d8d1f30b 1758 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
86872cb5 1759 if (cfg->fc_ifindex &&
d1918542
DM		 1760 (!rt->dst.dev ||
1761 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1762 continue;
86872cb5
TG		 1763 if (cfg->fc_flags & RTF_GATEWAY &&
1764 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1765 continue;
86872cb5 1766 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1767 continue;
d8d1f30b 1768 dst_hold(&rt->dst);
c71099ac 1769 read_unlock_bh(&table->tb6_lock);
1da177e4 1770
86872cb5 1771 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1772 }
1773 }
c71099ac 1774 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1775
1776 return err;
1777}
1778
6700c270 1779static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1780{
e8599ff4 1781 struct net *net = dev_net(skb->dev);
a6279458 1782 struct netevent_redirect netevent;
e8599ff4 1783 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1784 struct ndisc_options ndopts;
1785 struct inet6_dev *in6_dev;
1786 struct neighbour *neigh;
71bcdba0 1787 struct rd_msg *msg;
6e157b6a
DM		 1788 int optlen, on_link;
1789 u8 *lladdr;
e8599ff4 1790
29a3cad5 1791 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1792 optlen -= sizeof(*msg);
e8599ff4
DM		 1793
1794 if (optlen < 0) {
6e157b6a 1795 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1796 return;
1797 }
1798
71bcdba0 1799 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1800
71bcdba0 1801 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1802 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1803 return;
1804 }
1805
6e157b6a 1806 on_link = 0;
71bcdba0 1807 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1808 on_link = 1;
71bcdba0 1809 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1810 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1811 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1812 return;
1813 }
1814
1815 in6_dev = __in6_dev_get(skb->dev);
1816 if (!in6_dev)
1817 return;
1818 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1819 return;
1820
1821 /* RFC2461 8.1:
1822 * The IP source address of the Redirect MUST be the same as the current
1823 * first-hop router for the specified ICMP Destination Address.
1824 */
1825
71bcdba0 1826 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1827 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1828 return;
1829 }
6e157b6a
DM		 1830
1831 lladdr = NULL;
e8599ff4
DM		 1832 if (ndopts.nd_opts_tgt_lladdr) {
1833 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1834 skb->dev);
1835 if (!lladdr) {
1836 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1837 return;
1838 }
1839 }
1840
6e157b6a
DM		 1841 rt = (struct rt6_info *) dst;
1842 if (rt == net->ipv6.ip6_null_entry) {
1843 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1844 return;
6e157b6a 1845 }
e8599ff4 1846
6e157b6a
DM		 1847 /* Redirect received -> path was valid.
1848 * Look, redirects are sent only in response to data packets,
1849 * so that this nexthop apparently is reachable. --ANK
1850 */
1851 dst_confirm(&rt->dst);
a6279458 1852
71bcdba0 1853 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1854 if (!neigh)
1855 return;
a6279458 1856
1da177e4
LT		 1857 /*
1858 * We have finally decided to accept it.
1859 */
1860
1ab1457c 1861 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1862 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1863 NEIGH_UPDATE_F_OVERRIDE|
1864 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1865 NEIGH_UPDATE_F_ISROUTER))
1866 );
1867
71bcdba0 1868 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1869 if (!nrt)
1da177e4
LT		 1870 goto out;
1871
1872 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1873 if (on_link)
1874 nrt->rt6i_flags &= ~RTF_GATEWAY;
1875
4e3fd7a0 1876 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1877
40e22e8f 1878 if (ip6_ins_rt(nrt))
1da177e4
LT		 1879 goto out;
1880
d8d1f30b
CG		 1881 netevent.old = &rt->dst;
1882 netevent.new = &nrt->dst;
71bcdba0 1883 netevent.daddr = &msg->dest;
60592833 1884 netevent.neigh = neigh;
8d71740c
TT		 1885 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1886
38308473 1887 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1888 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1889 ip6_del_rt(rt);
1da177e4
LT		 1890 }
1891
1892out:
e8599ff4 1893 neigh_release(neigh);
6e157b6a
DM		 1894}
1895
1da177e4
LT		 1896/*
1897 * Misc support functions
1898 */
1899
1716a961 1900static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1901 const struct in6_addr *dest)
1da177e4 1902{
d1918542 1903 struct net *net = dev_net(ort->dst.dev);
8b96d22d
DM		 1904 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1905 ort->rt6i_table);
1da177e4
LT		 1906
1907 if (rt) {
d8d1f30b
CG		 1908 rt->dst.input = ort->dst.input;
1909 rt->dst.output = ort->dst.output;
8e2ec639 1910 rt->dst.flags |= DST_HOST;
d8d1f30b 1911
4e3fd7a0 1912 rt->rt6i_dst.addr = *dest;
8e2ec639 1913 rt->rt6i_dst.plen = 128;
defb3519 1914 dst_copy_metrics(&rt->dst, &ort->dst);
d8d1f30b 1915 rt->dst.error = ort->dst.error;
1da177e4
LT		 1916 rt->rt6i_idev = ort->rt6i_idev;
1917 if (rt->rt6i_idev)
1918 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1919 rt->dst.lastuse = jiffies;
1da177e4 1920
550bab42
JA		 1921 if (ort->rt6i_flags & RTF_GATEWAY)
1922 rt->rt6i_gateway = ort->rt6i_gateway;
1923 else
1924 rt->rt6i_gateway = *dest;
1716a961
G		 1925 rt->rt6i_flags = ort->rt6i_flags;
1926 if ((ort->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) ==
1927 (RTF_DEFAULT | RTF_ADDRCONF))
1928 rt6_set_from(rt, ort);
1da177e4
LT		 1929 rt->rt6i_metric = 0;
1930
1da177e4
LT		 1931#ifdef CONFIG_IPV6_SUBTREES
1932 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1933#endif
0f6c6392 1934 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1935 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1936 }
1937 return rt;
1938}
1939
70ceb4f5 1940#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1941static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 1942 const struct in6_addr *prefix, int prefixlen,
1943 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 1944{
1945 struct fib6_node *fn;
1946 struct rt6_info *rt = NULL;
c71099ac
TG		 1947 struct fib6_table *table;
1948
efa2cea0 1949 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 1950 if (!table)
c71099ac 1951 return NULL;
70ceb4f5 1952
5744dd9b 1953 read_lock_bh(&table->tb6_lock);
c71099ac 1954 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
70ceb4f5
YH		 1955 if (!fn)
1956 goto out;
1957
d8d1f30b 1958 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 1959 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 1960 continue;
1961 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1962 continue;
1963 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1964 continue;
d8d1f30b 1965 dst_hold(&rt->dst);
70ceb4f5
YH		 1966 break;
1967 }
1968out:
5744dd9b 1969 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1970 return rt;
1971}
1972
efa2cea0 1973static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 1974 const struct in6_addr *prefix, int prefixlen,
1975 const struct in6_addr *gwaddr, int ifindex,
95c96174 1976 unsigned int pref)
70ceb4f5 1977{
86872cb5
TG		 1978 struct fib6_config cfg = {
1979 .fc_table = RT6_TABLE_INFO,
238fc7ea 1980 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1981 .fc_ifindex = ifindex,
1982 .fc_dst_len = prefixlen,
1983 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1984 RTF_UP | RTF_PREF(pref),
15e47304 1985 .fc_nlinfo.portid = 0,
efa2cea0
DL		 1986 .fc_nlinfo.nlh = NULL,
1987 .fc_nlinfo.nl_net = net,
86872cb5
TG		 1988 };
1989
4e3fd7a0
AD		 1990 cfg.fc_dst = *prefix;
1991 cfg.fc_gateway = *gwaddr;
70ceb4f5 1992
e317da96
YH		 1993 /* We should treat it as a default route if prefix length is 0. */
1994 if (!prefixlen)
86872cb5 1995 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 1996
86872cb5 1997 ip6_route_add(&cfg);
70ceb4f5 1998
efa2cea0 1999 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 2000}
2001#endif
2002
b71d1d42 2003struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2004{
1da177e4 2005 struct rt6_info *rt;
c71099ac 2006 struct fib6_table *table;
1da177e4 2007
c346dca1 2008 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 2009 if (!table)
c71099ac 2010 return NULL;
1da177e4 2011
5744dd9b 2012 read_lock_bh(&table->tb6_lock);
d8d1f30b 2013 for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) {
d1918542 2014 if (dev == rt->dst.dev &&
045927ff 2015 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2016 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2017 break;
2018 }
2019 if (rt)
d8d1f30b 2020 dst_hold(&rt->dst);
5744dd9b 2021 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2022 return rt;
2023}
2024
b71d1d42 2025struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2026 struct net_device *dev,
2027 unsigned int pref)
1da177e4 2028{
86872cb5
TG		 2029 struct fib6_config cfg = {
2030 .fc_table = RT6_TABLE_DFLT,
238fc7ea 2031 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2032 .fc_ifindex = dev->ifindex,
2033 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2034 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2035 .fc_nlinfo.portid = 0,
5578689a 2036 .fc_nlinfo.nlh = NULL,
c346dca1 2037 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2038 };
1da177e4 2039
4e3fd7a0 2040 cfg.fc_gateway = *gwaddr;
1da177e4 2041
86872cb5 2042 ip6_route_add(&cfg);
1da177e4 2043
1da177e4
LT		 2044 return rt6_get_dflt_router(gwaddr, dev);
2045}
2046
7b4da532 2047void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2048{
2049 struct rt6_info *rt;
c71099ac
TG		 2050 struct fib6_table *table;
2051
2052 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2053 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2054 if (!table)
c71099ac 2055 return;
1da177e4
LT		 2056
2057restart:
c71099ac 2058 read_lock_bh(&table->tb6_lock);
d8d1f30b 2059 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2060 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2061 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2062 dst_hold(&rt->dst);
c71099ac 2063 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2064 ip6_del_rt(rt);
1da177e4
LT		 2065 goto restart;
2066 }
2067 }
c71099ac 2068 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2069}
2070
5578689a
DL		 2071static void rtmsg_to_fib6_config(struct net *net,
2072 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2073 struct fib6_config *cfg)
2074{
2075 memset(cfg, 0, sizeof(*cfg));
2076
2077 cfg->fc_table = RT6_TABLE_MAIN;
2078 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2079 cfg->fc_metric = rtmsg->rtmsg_metric;
2080 cfg->fc_expires = rtmsg->rtmsg_info;
2081 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2082 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2083 cfg->fc_flags = rtmsg->rtmsg_flags;
2084
5578689a 2085 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2086
4e3fd7a0
AD		 2087 cfg->fc_dst = rtmsg->rtmsg_dst;
2088 cfg->fc_src = rtmsg->rtmsg_src;
2089 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2090}
2091
5578689a 2092int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2093{
86872cb5 2094 struct fib6_config cfg;
1da177e4
LT		 2095 struct in6_rtmsg rtmsg;
2096 int err;
2097
2098 switch(cmd) {
2099 case SIOCADDRT: /* Add a route */
2100 case SIOCDELRT: /* Delete a route */
af31f412 2101 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2102 return -EPERM;
2103 err = copy_from_user(&rtmsg, arg,
2104 sizeof(struct in6_rtmsg));
2105 if (err)
2106 return -EFAULT;
86872cb5 2107
5578689a 2108 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2109
1da177e4
LT		 2110 rtnl_lock();
2111 switch (cmd) {
2112 case SIOCADDRT:
86872cb5 2113 err = ip6_route_add(&cfg);
1da177e4
LT		 2114 break;
2115 case SIOCDELRT:
86872cb5 2116 err = ip6_route_del(&cfg);
1da177e4
LT		 2117 break;
2118 default:
2119 err = -EINVAL;
2120 }
2121 rtnl_unlock();
2122
2123 return err;
3ff50b79 2124 }
1da177e4
LT		 2125
2126 return -EINVAL;
2127}
2128
2129/*
2130 * Drop the packet on the floor
2131 */
2132
d5fdd6ba 2133static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2134{
612f09e8 2135 int type;
adf30907 2136 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2137 switch (ipstats_mib_noroutes) {
2138 case IPSTATS_MIB_INNOROUTES:
0660e03f 2139 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2140 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2141 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2142 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2143 break;
2144 }
2145 /* FALLTHROUGH */
2146 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2147 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2148 ipstats_mib_noroutes);
612f09e8
YH		 2149 break;
2150 }
3ffe533c 2151 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2152 kfree_skb(skb);
2153 return 0;
2154}
2155
9ce8ade0
TG		 2156static int ip6_pkt_discard(struct sk_buff *skb)
2157{
612f09e8 2158 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2159}
2160
20380731 2161static int ip6_pkt_discard_out(struct sk_buff *skb)
1da177e4 2162{
adf30907 2163 skb->dev = skb_dst(skb)->dev;
612f09e8 2164 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2165}
2166
9ce8ade0
TG		 2167static int ip6_pkt_prohibit(struct sk_buff *skb)
2168{
612f09e8 2169 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2170}
2171
2172static int ip6_pkt_prohibit_out(struct sk_buff *skb)
2173{
adf30907 2174 skb->dev = skb_dst(skb)->dev;
612f09e8 2175 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2176}
2177
1da177e4
LT		 2178/*
2179 * Allocate a dst for local (unicast / anycast) address.
2180 */
2181
2182struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2183 const struct in6_addr *addr,
8f031519 2184 bool anycast)
1da177e4 2185{
c346dca1 2186 struct net *net = dev_net(idev->dev);
a3300ef4
HFS		 2187 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2188 DST_NOCOUNT, NULL);
2189 if (!rt)
1da177e4
LT		 2190 return ERR_PTR(-ENOMEM);
2191
1da177e4
LT		 2192 in6_dev_hold(idev);
2193
11d53b49 2194 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2195 rt->dst.input = ip6_input;
2196 rt->dst.output = ip6_output;
1da177e4 2197 rt->rt6i_idev = idev;
1da177e4
LT		 2198
2199 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2200 if (anycast)
2201 rt->rt6i_flags |= RTF_ANYCAST;
2202 else
1da177e4 2203 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2204
550bab42 2205 rt->rt6i_gateway = *addr;
4e3fd7a0 2206 rt->rt6i_dst.addr = *addr;
1da177e4 2207 rt->rt6i_dst.plen = 128;
5578689a 2208 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2209
d8d1f30b 2210 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2211
2212 return rt;
2213}
2214
c3968a85
DW		 2215int ip6_route_get_saddr(struct net *net,
2216 struct rt6_info *rt,
b71d1d42 2217 const struct in6_addr *daddr,
c3968a85
DW		 2218 unsigned int prefs,
2219 struct in6_addr *saddr)
2220{
2221 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt);
2222 int err = 0;
2223 if (rt->rt6i_prefsrc.plen)
4e3fd7a0 2224 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2225 else
2226 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2227 daddr, prefs, saddr);
2228 return err;
2229}
2230
2231/* remove deleted ip from prefsrc entries */
2232struct arg_dev_net_ip {
2233 struct net_device *dev;
2234 struct net *net;
2235 struct in6_addr *addr;
2236};
2237
2238static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2239{
2240 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2241 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2242 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2243
d1918542 2244 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2245 rt != net->ipv6.ip6_null_entry &&
2246 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2247 /* remove prefsrc entry */
2248 rt->rt6i_prefsrc.plen = 0;
2249 }
2250 return 0;
2251}
2252
2253void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2254{
2255 struct net *net = dev_net(ifp->idev->dev);
2256 struct arg_dev_net_ip adni = {
2257 .dev = ifp->idev->dev,
2258 .net = net,
2259 .addr = &ifp->addr,
2260 };
2261 fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni);
2262}
2263
8ed67789
DL		 2264struct arg_dev_net {
2265 struct net_device *dev;
2266 struct net *net;
2267};
2268
1da177e4
LT		 2269static int fib6_ifdown(struct rt6_info *rt, void *arg)
2270{
bc3ef660 2271 const struct arg_dev_net *adn = arg;
2272 const struct net_device *dev = adn->dev;
8ed67789 2273
d1918542 2274 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2275 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2276 return -1;
c159d30c 2277
1da177e4
LT		 2278 return 0;
2279}
2280
f3db4851 2281void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2282{
8ed67789
DL		 2283 struct arg_dev_net adn = {
2284 .dev = dev,
2285 .net = net,
2286 };
2287
2288 fib6_clean_all(net, fib6_ifdown, 0, &adn);
1e493d19 2289 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2290}
2291
95c96174 2292struct rt6_mtu_change_arg {
1da177e4 2293 struct net_device *dev;
95c96174 2294 unsigned int mtu;
1da177e4
LT		 2295};
2296
2297static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2298{
2299 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2300 struct inet6_dev *idev;
2301
2302 /* In IPv6 pmtu discovery is not optional,
2303 so that RTAX_MTU lock cannot disable it.
2304 We still use this lock to block changes
2305 caused by addrconf/ndisc.
2306 */
2307
2308 idev = __in6_dev_get(arg->dev);
38308473 2309 if (!idev)
1da177e4
LT		 2310 return 0;
2311
2312 /* For administrative MTU increase, there is no way to discover
2313 IPv6 PMTU increase, so PMTU increase should be updated here.
2314 Since RFC 1981 doesn't include administrative MTU increase
2315 update PMTU increase is a MUST. (i.e. jumbo frame)
2316 */
2317 /*
2318 If new MTU is less than route PMTU, this new MTU will be the
2319 lowest MTU in the path, update the route PMTU to reflect PMTU
2320 decreases; if new MTU is greater than route PMTU, and the
2321 old MTU is the lowest MTU in the path, update the route PMTU
2322 to reflect the increase. In this case if the other nodes' MTU
2323 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2324 PMTU discouvery.
2325 */
d1918542 2326 if (rt->dst.dev == arg->dev &&
d8d1f30b
CG		 2327 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2328 (dst_mtu(&rt->dst) >= arg->mtu ||
2329 (dst_mtu(&rt->dst) < arg->mtu &&
2330 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
defb3519 2331 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
566cfd8f 2332 }
1da177e4
LT		 2333 return 0;
2334}
2335
95c96174 2336void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2337{
c71099ac
TG		 2338 struct rt6_mtu_change_arg arg = {
2339 .dev = dev,
2340 .mtu = mtu,
2341 };
1da177e4 2342
c346dca1 2343 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg);
1da177e4
LT		 2344}
2345
ef7c79ed 2346static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2347 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2348 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2349 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2350 [RTA_PRIORITY] = { .type = NLA_U32 },
2351 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2352 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
86872cb5
TG		 2353};
2354
2355static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2356 struct fib6_config *cfg)
1da177e4 2357{
86872cb5
TG		 2358 struct rtmsg *rtm;
2359 struct nlattr *tb[RTA_MAX+1];
2360 int err;
1da177e4 2361
86872cb5
TG		 2362 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2363 if (err < 0)
2364 goto errout;
1da177e4 2365
86872cb5
TG		 2366 err = -EINVAL;
2367 rtm = nlmsg_data(nlh);
2368 memset(cfg, 0, sizeof(*cfg));
2369
2370 cfg->fc_table = rtm->rtm_table;
2371 cfg->fc_dst_len = rtm->rtm_dst_len;
2372 cfg->fc_src_len = rtm->rtm_src_len;
2373 cfg->fc_flags = RTF_UP;
2374 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2375 cfg->fc_type = rtm->rtm_type;
86872cb5 2376
ef2c7d7b
ND		 2377 if (rtm->rtm_type == RTN_UNREACHABLE ||
2378 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2379 rtm->rtm_type == RTN_PROHIBIT ||
2380 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2381 cfg->fc_flags |= RTF_REJECT;
2382
ab79ad14
MŻ		 2383 if (rtm->rtm_type == RTN_LOCAL)
2384 cfg->fc_flags |= RTF_LOCAL;
2385
15e47304 2386 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2387 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2388 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2389
2390 if (tb[RTA_GATEWAY]) {
2391 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2392 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2393 }
86872cb5
TG		 2394
2395 if (tb[RTA_DST]) {
2396 int plen = (rtm->rtm_dst_len + 7) >> 3;
2397
2398 if (nla_len(tb[RTA_DST]) < plen)
2399 goto errout;
2400
2401 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2402 }
86872cb5
TG		 2403
2404 if (tb[RTA_SRC]) {
2405 int plen = (rtm->rtm_src_len + 7) >> 3;
2406
2407 if (nla_len(tb[RTA_SRC]) < plen)
2408 goto errout;
2409
2410 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2411 }
86872cb5 2412
c3968a85
DW		 2413 if (tb[RTA_PREFSRC])
2414 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2415
86872cb5
TG		 2416 if (tb[RTA_OIF])
2417 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2418
2419 if (tb[RTA_PRIORITY])
2420 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2421
2422 if (tb[RTA_METRICS]) {
2423 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2424 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2425 }
86872cb5
TG		 2426
2427 if (tb[RTA_TABLE])
2428 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2429
51ebd318
ND		 2430 if (tb[RTA_MULTIPATH]) {
2431 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2432 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2433 }
2434
86872cb5
TG		 2435 err = 0;
2436errout:
2437 return err;
1da177e4
LT		 2438}
2439
51ebd318
ND		 2440static int ip6_route_multipath(struct fib6_config *cfg, int add)
2441{
2442 struct fib6_config r_cfg;
2443 struct rtnexthop *rtnh;
2444 int remaining;
2445 int attrlen;
2446 int err = 0, last_err = 0;
2447
2448beginning:
2449 rtnh = (struct rtnexthop *)cfg->fc_mp;
2450 remaining = cfg->fc_mp_len;
2451
2452 /* Parse a Multipath Entry */
2453 while (rtnh_ok(rtnh, remaining)) {
2454 memcpy(&r_cfg, cfg, sizeof(*cfg));
2455 if (rtnh->rtnh_ifindex)
2456 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2457
2458 attrlen = rtnh_attrlen(rtnh);
2459 if (attrlen > 0) {
2460 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2461
2462 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2463 if (nla) {
2464 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2465 r_cfg.fc_flags |= RTF_GATEWAY;
2466 }
2467 }
2468 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2469 if (err) {
2470 last_err = err;
2471 /* If we are trying to remove a route, do not stop the
2472 * loop when ip6_route_del() fails (because next hop is
2473 * already gone), we should try to remove all next hops.
2474 */
2475 if (add) {
2476 /* If add fails, we should try to delete all
2477 * next hops that have been already added.
2478 */
2479 add = 0;
2480 goto beginning;
2481 }
2482 }
1a72418b
ND		 2483 /* Because each route is added like a single route we remove
2484 * this flag after the first nexthop (if there is a collision,
2485 * we have already fail to add the first nexthop:
2486 * fib6_add_rt2node() has reject it).
2487 */
2488 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
51ebd318
ND		 2489 rtnh = rtnh_next(rtnh, &remaining);
2490 }
2491
2492 return last_err;
2493}
2494
661d2967 2495static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2496{
86872cb5
TG		 2497 struct fib6_config cfg;
2498 int err;
1da177e4 2499
86872cb5
TG		 2500 err = rtm_to_fib6_config(skb, nlh, &cfg);
2501 if (err < 0)
2502 return err;
2503
51ebd318
ND		 2504 if (cfg.fc_mp)
2505 return ip6_route_multipath(&cfg, 0);
2506 else
2507 return ip6_route_del(&cfg);
1da177e4
LT		 2508}
2509
661d2967 2510static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh)
1da177e4 2511{
86872cb5
TG		 2512 struct fib6_config cfg;
2513 int err;
1da177e4 2514
86872cb5
TG		 2515 err = rtm_to_fib6_config(skb, nlh, &cfg);
2516 if (err < 0)
2517 return err;
2518
51ebd318
ND		 2519 if (cfg.fc_mp)
2520 return ip6_route_multipath(&cfg, 1);
2521 else
2522 return ip6_route_add(&cfg);
1da177e4
LT		 2523}
2524
339bf98f
TG		 2525static inline size_t rt6_nlmsg_size(void)
2526{
2527 return NLMSG_ALIGN(sizeof(struct rtmsg))
2528 + nla_total_size(16) /* RTA_SRC */
2529 + nla_total_size(16) /* RTA_DST */
2530 + nla_total_size(16) /* RTA_GATEWAY */
2531 + nla_total_size(16) /* RTA_PREFSRC */
2532 + nla_total_size(4) /* RTA_TABLE */
2533 + nla_total_size(4) /* RTA_IIF */
2534 + nla_total_size(4) /* RTA_OIF */
2535 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2536 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
339bf98f
TG		 2537 + nla_total_size(sizeof(struct rta_cacheinfo));
2538}
2539
191cd582
BH		 2540static int rt6_fill_node(struct net *net,
2541 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2542 struct in6_addr *dst, struct in6_addr *src,
15e47304 2543 int iif, int type, u32 portid, u32 seq,
7bc570c8 2544 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2545{
2546 struct rtmsg *rtm;
2d7202bf 2547 struct nlmsghdr *nlh;
e3703b3d 2548 long expires;
9e762a4a 2549 u32 table;
1da177e4
LT		 2550
2551 if (prefix) { /* user wants prefix routes only */
2552 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2553 /* success since this is not a prefix route */
2554 return 1;
2555 }
2556 }
2557
15e47304 2558 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2559 if (!nlh)
26932566 2560 return -EMSGSIZE;
2d7202bf
TG		 2561
2562 rtm = nlmsg_data(nlh);
1da177e4
LT		 2563 rtm->rtm_family = AF_INET6;
2564 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2565 rtm->rtm_src_len = rt->rt6i_src.plen;
2566 rtm->rtm_tos = 0;
c71099ac 2567 if (rt->rt6i_table)
9e762a4a 2568 table = rt->rt6i_table->tb6_id;
c71099ac 2569 else
9e762a4a
PM		 2570 table = RT6_TABLE_UNSPEC;
2571 rtm->rtm_table = table;
c78679e8
DM		 2572 if (nla_put_u32(skb, RTA_TABLE, table))
2573 goto nla_put_failure;
ef2c7d7b
ND		 2574 if (rt->rt6i_flags & RTF_REJECT) {
2575 switch (rt->dst.error) {
2576 case -EINVAL:
2577 rtm->rtm_type = RTN_BLACKHOLE;
2578 break;
2579 case -EACCES:
2580 rtm->rtm_type = RTN_PROHIBIT;
2581 break;
b4949ab2
ND		 2582 case -EAGAIN:
2583 rtm->rtm_type = RTN_THROW;
2584 break;
ef2c7d7b
ND		 2585 default:
2586 rtm->rtm_type = RTN_UNREACHABLE;
2587 break;
2588 }
2589 }
38308473 2590 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2591 rtm->rtm_type = RTN_LOCAL;
d1918542 2592 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2593 rtm->rtm_type = RTN_LOCAL;
2594 else
2595 rtm->rtm_type = RTN_UNICAST;
2596 rtm->rtm_flags = 0;
2597 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2598 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2599 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2600 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2601 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2602 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2603 rtm->rtm_protocol = RTPROT_RA;
2604 else
2605 rtm->rtm_protocol = RTPROT_KERNEL;
2606 }
1da177e4 2607
38308473 2608 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2609 rtm->rtm_flags |= RTM_F_CLONED;
2610
2611 if (dst) {
c78679e8
DM		 2612 if (nla_put(skb, RTA_DST, 16, dst))
2613 goto nla_put_failure;
1ab1457c 2614 rtm->rtm_dst_len = 128;
1da177e4 2615 } else if (rtm->rtm_dst_len)
c78679e8
DM		 2616 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2617 goto nla_put_failure;
1da177e4
LT		 2618#ifdef CONFIG_IPV6_SUBTREES
2619 if (src) {
c78679e8
DM		 2620 if (nla_put(skb, RTA_SRC, 16, src))
2621 goto nla_put_failure;
1ab1457c 2622 rtm->rtm_src_len = 128;
c78679e8
DM		 2623 } else if (rtm->rtm_src_len &&
2624 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2625 goto nla_put_failure;
1da177e4 2626#endif
7bc570c8
YH		 2627 if (iif) {
2628#ifdef CONFIG_IPV6_MROUTE
2629 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2630 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2631 if (err <= 0) {
2632 if (!nowait) {
2633 if (err == 0)
2634 return 0;
2635 goto nla_put_failure;
2636 } else {
2637 if (err == -EMSGSIZE)
2638 goto nla_put_failure;
2639 }
2640 }
2641 } else
2642#endif
c78679e8
DM		 2643 if (nla_put_u32(skb, RTA_IIF, iif))
2644 goto nla_put_failure;
7bc570c8 2645 } else if (dst) {
1da177e4 2646 struct in6_addr saddr_buf;
c78679e8
DM		 2647 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2648 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2649 goto nla_put_failure;
1da177e4 2650 }
2d7202bf 2651
c3968a85
DW		 2652 if (rt->rt6i_prefsrc.plen) {
2653 struct in6_addr saddr_buf;
4e3fd7a0 2654 saddr_buf = rt->rt6i_prefsrc.addr;
c78679e8
DM		 2655 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2656 goto nla_put_failure;
c3968a85
DW		 2657 }
2658
defb3519 2659 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2d7202bf
TG		 2660 goto nla_put_failure;
2661
dd0cbf29
YH		 2662 if (rt->rt6i_flags & RTF_GATEWAY) {
2663 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
94f826b8 2664 goto nla_put_failure;
94f826b8 2665 }
2d7202bf 2666
c78679e8
DM		 2667 if (rt->dst.dev &&
2668 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2669 goto nla_put_failure;
2670 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2671 goto nla_put_failure;
8253947e
LW		 2672
2673 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2674
87a50699 2675 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2676 goto nla_put_failure;
2d7202bf
TG		 2677
2678 return nlmsg_end(skb, nlh);
2679
2680nla_put_failure:
26932566
PM		 2681 nlmsg_cancel(skb, nlh);
2682 return -EMSGSIZE;
1da177e4
LT		 2683}
2684
1b43af54 2685int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2686{
2687 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2688 int prefix;
2689
2d7202bf
TG		 2690 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2691 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2692 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2693 } else
2694 prefix = 0;
2695
191cd582
BH		 2696 return rt6_fill_node(arg->net,
2697 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2698 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2699 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2700}
2701
661d2967 2702static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh)
1da177e4 2703{
3b1e0a65 2704 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2705 struct nlattr *tb[RTA_MAX+1];
2706 struct rt6_info *rt;
1da177e4 2707 struct sk_buff *skb;
ab364a6f 2708 struct rtmsg *rtm;
4c9483b2 2709 struct flowi6 fl6;
72331bc0 2710 int err, iif = 0, oif = 0;
1da177e4 2711
ab364a6f
TG		 2712 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2713 if (err < 0)
2714 goto errout;
1da177e4 2715
ab364a6f 2716 err = -EINVAL;
4c9483b2 2717 memset(&fl6, 0, sizeof(fl6));
1da177e4 2718
ab364a6f
TG		 2719 if (tb[RTA_SRC]) {
2720 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2721 goto errout;
2722
4e3fd7a0 2723 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2724 }
2725
2726 if (tb[RTA_DST]) {
2727 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2728 goto errout;
2729
4e3fd7a0 2730 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2731 }
2732
2733 if (tb[RTA_IIF])
2734 iif = nla_get_u32(tb[RTA_IIF]);
2735
2736 if (tb[RTA_OIF])
72331bc0 2737 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4
LT		 2738
2739 if (iif) {
2740 struct net_device *dev;
72331bc0
SL		 2741 int flags = 0;
2742
5578689a 2743 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2744 if (!dev) {
2745 err = -ENODEV;
ab364a6f 2746 goto errout;
1da177e4 2747 }
72331bc0
SL		 2748
2749 fl6.flowi6_iif = iif;
2750
2751 if (!ipv6_addr_any(&fl6.saddr))
2752 flags |= RT6_LOOKUP_F_HAS_SADDR;
2753
2754 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2755 flags);
2756 } else {
2757 fl6.flowi6_oif = oif;
2758
2759 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2760 }
2761
ab364a6f 2762 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2763 if (!skb) {
94e187c0 2764 ip6_rt_put(rt);
ab364a6f
TG		 2765 err = -ENOBUFS;
2766 goto errout;
2767 }
1da177e4 2768
ab364a6f
TG		 2769 /* Reserve room for dummy headers, this skb can pass
2770 through good chunk of routing engine.
2771 */
459a98ed 2772 skb_reset_mac_header(skb);
ab364a6f 2773 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2774
d8d1f30b 2775 skb_dst_set(skb, &rt->dst);
1da177e4 2776
4c9483b2 2777 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2778 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2779 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2780 if (err < 0) {
ab364a6f
TG		 2781 kfree_skb(skb);
2782 goto errout;
1da177e4
LT		 2783 }
2784
15e47304 2785 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2786errout:
1da177e4 2787 return err;
1da177e4
LT		 2788}
2789
86872cb5 2790void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2791{
2792 struct sk_buff *skb;
5578689a 2793 struct net *net = info->nl_net;
528c4ceb
DL		 2794 u32 seq;
2795 int err;
2796
2797 err = -ENOBUFS;
38308473 2798 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2799
339bf98f 2800 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2801 if (!skb)
21713ebc
TG		 2802 goto errout;
2803
191cd582 2804 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2805 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2806 if (err < 0) {
2807 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2808 WARN_ON(err == -EMSGSIZE);
2809 kfree_skb(skb);
2810 goto errout;
2811 }
15e47304 2812 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2813 info->nlh, gfp_any());
2814 return;
21713ebc
TG		 2815errout:
2816 if (err < 0)
5578689a 2817 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2818}
2819
8ed67789 2820static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2821 unsigned long event, void *ptr)
8ed67789 2822{
351638e7 2823 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2824 struct net *net = dev_net(dev);
8ed67789
DL		 2825
2826 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2827 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2828 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2829#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2830 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2831 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2832 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2833 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2834#endif
2835 }
2836
2837 return NOTIFY_OK;
2838}
2839
1da177e4
LT		 2840/*
2841 * /proc
2842 */
2843
2844#ifdef CONFIG_PROC_FS
2845
33120b30
AD		 2846static const struct file_operations ipv6_route_proc_fops = {
2847 .owner = THIS_MODULE,
2848 .open = ipv6_route_open,
2849 .read = seq_read,
2850 .llseek = seq_lseek,
8d2ca1d7 2851 .release = seq_release_net,
33120b30
AD		 2852};
2853
1da177e4
LT		 2854static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2855{
69ddb805 2856 struct net *net = (struct net *)seq->private;
1da177e4 2857 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2858 net->ipv6.rt6_stats->fib_nodes,
2859 net->ipv6.rt6_stats->fib_route_nodes,
2860 net->ipv6.rt6_stats->fib_rt_alloc,
2861 net->ipv6.rt6_stats->fib_rt_entries,
2862 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2863 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2864 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2865
2866 return 0;
2867}
2868
2869static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2870{
de05c557 2871 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2872}
2873
9a32144e 2874static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2875 .owner = THIS_MODULE,
2876 .open = rt6_stats_seq_open,
2877 .read = seq_read,
2878 .llseek = seq_lseek,
b6fcbdb4 2879 .release = single_release_net,
1da177e4
LT		 2880};
2881#endif /* CONFIG_PROC_FS */
2882
2883#ifdef CONFIG_SYSCTL
2884
1da177e4 2885static
fe2c6338 2886int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 2887 void __user *buffer, size_t *lenp, loff_t *ppos)
2888{
c486da34
LAG		 2889 struct net *net;
2890 int delay;
2891 if (!write)
1da177e4 2892 return -EINVAL;
c486da34
LAG		 2893
2894 net = (struct net *)ctl->extra1;
2895 delay = net->ipv6.sysctl.flush_delay;
2896 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 2897 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 2898 return 0;
1da177e4
LT		 2899}
2900
fe2c6338 2901struct ctl_table ipv6_route_table_template[] = {
1ab1457c 2902 {
1da177e4 2903 .procname = "flush",
4990509f 2904 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2905 .maxlen = sizeof(int),
89c8b3a1 2906 .mode = 0200,
6d9f239a 2907 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 2908 },
2909 {
1da177e4 2910 .procname = "gc_thresh",
9a7ec3a9 2911 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2912 .maxlen = sizeof(int),
2913 .mode = 0644,
6d9f239a 2914 .proc_handler = proc_dointvec,
1da177e4
LT		 2915 },
2916 {
1da177e4 2917 .procname = "max_size",
4990509f 2918 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2919 .maxlen = sizeof(int),
2920 .mode = 0644,
6d9f239a 2921 .proc_handler = proc_dointvec,
1da177e4
LT		 2922 },
2923 {
1da177e4 2924 .procname = "gc_min_interval",
4990509f 2925 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2926 .maxlen = sizeof(int),
2927 .mode = 0644,
6d9f239a 2928 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2929 },
2930 {
1da177e4 2931 .procname = "gc_timeout",
4990509f 2932 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2933 .maxlen = sizeof(int),
2934 .mode = 0644,
6d9f239a 2935 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2936 },
2937 {
1da177e4 2938 .procname = "gc_interval",
4990509f 2939 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2940 .maxlen = sizeof(int),
2941 .mode = 0644,
6d9f239a 2942 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2943 },
2944 {
1da177e4 2945 .procname = "gc_elasticity",
4990509f 2946 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 2947 .maxlen = sizeof(int),
2948 .mode = 0644,
f3d3f616 2949 .proc_handler = proc_dointvec,
1da177e4
LT		 2950 },
2951 {
1da177e4 2952 .procname = "mtu_expires",
4990509f 2953 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 2954 .maxlen = sizeof(int),
2955 .mode = 0644,
6d9f239a 2956 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 2957 },
2958 {
1da177e4 2959 .procname = "min_adv_mss",
4990509f 2960 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 2961 .maxlen = sizeof(int),
2962 .mode = 0644,
f3d3f616 2963 .proc_handler = proc_dointvec,
1da177e4
LT		 2964 },
2965 {
1da177e4 2966 .procname = "gc_min_interval_ms",
4990509f 2967 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2968 .maxlen = sizeof(int),
2969 .mode = 0644,
6d9f239a 2970 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 2971 },
f8572d8f 2972 { }
1da177e4
LT		 2973};
2974
2c8c1e72 2975struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 2976{
2977 struct ctl_table *table;
2978
2979 table = kmemdup(ipv6_route_table_template,
2980 sizeof(ipv6_route_table_template),
2981 GFP_KERNEL);
5ee09105
YH		 2982
2983 if (table) {
2984 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 2985 table[0].extra1 = net;
86393e52 2986 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 2987 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2988 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2989 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2990 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2991 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2992 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2993 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 2994 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 2995
2996 /* Don't export sysctls to unprivileged users */
2997 if (net->user_ns != &init_user_ns)
2998 table[0].procname = NULL;
5ee09105
YH		 2999 }
3000
760f2d01
DL		 3001 return table;
3002}
1da177e4
LT		 3003#endif
3004
2c8c1e72 3005static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3006{
633d424b 3007 int ret = -ENOMEM;
8ed67789 3008
86393e52
AD		 3009 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3010 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3011
fc66f95c
ED		 3012 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3013 goto out_ip6_dst_ops;
3014
8ed67789
DL		 3015 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3016 sizeof(*net->ipv6.ip6_null_entry),
3017 GFP_KERNEL);
3018 if (!net->ipv6.ip6_null_entry)
fc66f95c 3019 goto out_ip6_dst_entries;
d8d1f30b 3020 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3021 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3022 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3023 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3024 ip6_template_metrics, true);
8ed67789
DL		 3025
3026#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3027 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3028 sizeof(*net->ipv6.ip6_prohibit_entry),
3029 GFP_KERNEL);
68fffc67
PZ		 3030 if (!net->ipv6.ip6_prohibit_entry)
3031 goto out_ip6_null_entry;
d8d1f30b 3032 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3033 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3034 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3035 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3036 ip6_template_metrics, true);
8ed67789
DL		 3037
3038 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3039 sizeof(*net->ipv6.ip6_blk_hole_entry),
3040 GFP_KERNEL);
68fffc67
PZ		 3041 if (!net->ipv6.ip6_blk_hole_entry)
3042 goto out_ip6_prohibit_entry;
d8d1f30b 3043 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3044 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3045 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3046 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3047 ip6_template_metrics, true);
8ed67789
DL		 3048#endif
3049
b339a47c
PZ		 3050 net->ipv6.sysctl.flush_delay = 0;
3051 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3052 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3053 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3054 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3055 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3056 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3057 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3058
6891a346
BT		 3059 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3060
8ed67789
DL		 3061 ret = 0;
3062out:
3063 return ret;
f2fc6a54 3064
68fffc67
PZ		 3065#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3066out_ip6_prohibit_entry:
3067 kfree(net->ipv6.ip6_prohibit_entry);
3068out_ip6_null_entry:
3069 kfree(net->ipv6.ip6_null_entry);
3070#endif
fc66f95c
ED		 3071out_ip6_dst_entries:
3072 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3073out_ip6_dst_ops:
f2fc6a54 3074 goto out;
cdb18761
DL		 3075}
3076
2c8c1e72 3077static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3078{
8ed67789
DL		 3079 kfree(net->ipv6.ip6_null_entry);
3080#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3081 kfree(net->ipv6.ip6_prohibit_entry);
3082 kfree(net->ipv6.ip6_blk_hole_entry);
3083#endif
41bb78b4 3084 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3085}
3086
d189634e
TG		 3087static int __net_init ip6_route_net_init_late(struct net *net)
3088{
3089#ifdef CONFIG_PROC_FS
d4beaa66
G		 3090 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3091 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3092#endif
3093 return 0;
3094}
3095
3096static void __net_exit ip6_route_net_exit_late(struct net *net)
3097{
3098#ifdef CONFIG_PROC_FS
ece31ffd
G		 3099 remove_proc_entry("ipv6_route", net->proc_net);
3100 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3101#endif
3102}
3103
cdb18761
DL		 3104static struct pernet_operations ip6_route_net_ops = {
3105 .init = ip6_route_net_init,
3106 .exit = ip6_route_net_exit,
3107};
3108
c3426b47
DM		 3109static int __net_init ipv6_inetpeer_init(struct net *net)
3110{
3111 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3112
3113 if (!bp)
3114 return -ENOMEM;
3115 inet_peer_base_init(bp);
3116 net->ipv6.peers = bp;
3117 return 0;
3118}
3119
3120static void __net_exit ipv6_inetpeer_exit(struct net *net)
3121{
3122 struct inet_peer_base *bp = net->ipv6.peers;
3123
3124 net->ipv6.peers = NULL;
56a6b248 3125 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3126 kfree(bp);
3127}
3128
2b823f72 3129static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3130 .init = ipv6_inetpeer_init,
3131 .exit = ipv6_inetpeer_exit,
3132};
3133
d189634e
TG		 3134static struct pernet_operations ip6_route_net_late_ops = {
3135 .init = ip6_route_net_init_late,
3136 .exit = ip6_route_net_exit_late,
3137};
3138
8ed67789
DL		 3139static struct notifier_block ip6_route_dev_notifier = {
3140 .notifier_call = ip6_route_dev_notify,
3141 .priority = 0,
3142};
3143
433d49c3 3144int __init ip6_route_init(void)
1da177e4 3145{
433d49c3
DL		 3146 int ret;
3147
9a7ec3a9
DL		 3148 ret = -ENOMEM;
3149 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3150 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3151 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3152 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3153 goto out;
14e50e57 3154
fc66f95c 3155 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3156 if (ret)
bdb3289f 3157 goto out_kmem_cache;
bdb3289f 3158
c3426b47
DM		 3159 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3160 if (ret)
e8803b6c 3161 goto out_dst_entries;
2a0c451a 3162
7e52b33b
DM		 3163 ret = register_pernet_subsys(&ip6_route_net_ops);
3164 if (ret)
3165 goto out_register_inetpeer;
c3426b47 3166
5dc121e9
AE		 3167 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3168
8ed67789
DL		 3169 /* Registering of the loopback is done before this portion of code,
3170 * the loopback reference in rt6_info will not be taken, do it
3171 * manually for init_net */
d8d1f30b 3172 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3173 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3174 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3175 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3176 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3177 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3178 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3179 #endif
e8803b6c 3180 ret = fib6_init();
433d49c3 3181 if (ret)
8ed67789 3182 goto out_register_subsys;
433d49c3 3183
433d49c3
DL		 3184 ret = xfrm6_init();
3185 if (ret)
e8803b6c 3186 goto out_fib6_init;
c35b7e72 3187
433d49c3
DL		 3188 ret = fib6_rules_init();
3189 if (ret)
3190 goto xfrm6_init;
7e5449c2 3191
d189634e
TG		 3192 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3193 if (ret)
3194 goto fib6_rules_init;
3195
433d49c3 3196 ret = -ENOBUFS;
c7ac8679
GR		 3197 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3198 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3199 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3200 goto out_register_late_subsys;
c127ea2c 3201
8ed67789 3202 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3203 if (ret)
d189634e 3204 goto out_register_late_subsys;
8ed67789 3205
433d49c3
DL		 3206out:
3207 return ret;
3208
d189634e
TG		 3209out_register_late_subsys:
3210 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3211fib6_rules_init:
433d49c3
DL		 3212 fib6_rules_cleanup();
3213xfrm6_init:
433d49c3 3214 xfrm6_fini();
2a0c451a
TG		 3215out_fib6_init:
3216 fib6_gc_cleanup();
8ed67789
DL		 3217out_register_subsys:
3218 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3219out_register_inetpeer:
3220 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3221out_dst_entries:
3222 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3223out_kmem_cache:
f2fc6a54 3224 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3225 goto out;
1da177e4
LT		 3226}
3227
3228void ip6_route_cleanup(void)
3229{
8ed67789 3230 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3231 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3232 fib6_rules_cleanup();
1da177e4 3233 xfrm6_fini();
1da177e4 3234 fib6_gc_cleanup();
c3426b47 3235 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3236 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3237 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3238 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3239}
Linux 6.x block layer and io_uring tree(s)