]> git.ipfire.org Git - people/ms/linux.git/blame - net/ipv6/route.c
projects / people / ms / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
ipv6: Create RTF_CACHE clone when FLOWI_FLAG_KNOWN_NH is set
[people/ms/linux.git] / net / ipv6 / route.c
CommitLineData
1da177e4
LT		 1/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4 7 *
1da177e4
LT		 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13
14/* Changes:
15 *
16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection.
18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list.
c0bece9f
YH		 23 * Ville Nuorvala
24 * Fixed routing subtrees.
1da177e4
LT		 25 */
26
f3213831
JP		 27#define pr_fmt(fmt) "IPv6: " fmt
28
4fc268d2 29#include <linux/capability.h>
1da177e4 30#include <linux/errno.h>
bc3b2d7f 31#include <linux/export.h>
1da177e4
LT		 32#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
7bc570c8 40#include <linux/mroute6.h>
1da177e4 41#include <linux/init.h>
1da177e4 42#include <linux/if_arp.h>
1da177e4
LT		 43#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
5b7c931d 45#include <linux/nsproxy.h>
5a0e3ad6 46#include <linux/slab.h>
457c4cbc 47#include <net/net_namespace.h>
1da177e4
LT		 48#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
57#include <net/xfrm.h>
8d71740c 58#include <net/netevent.h>
21713ebc 59#include <net/netlink.h>
51ebd318 60#include <net/nexthop.h>
1da177e4
LT		 61
62#include <asm/uaccess.h>
63
64#ifdef CONFIG_SYSCTL
65#include <linux/sysctl.h>
66#endif
67
afc154e9 68enum rt6_nud_state {
7e980569
JB		 69 RT6_NUD_FAIL_HARD = -3,
70 RT6_NUD_FAIL_PROBE = -2,
71 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 72 RT6_NUD_SUCCEED = 1
73};
74
1716a961 75static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 76 const struct in6_addr *dest);
1da177e4 77static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 78static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 79static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 80static struct dst_entry *ip6_negative_advice(struct dst_entry *);
81static void ip6_dst_destroy(struct dst_entry *);
82static void ip6_dst_ifdown(struct dst_entry *,
83 struct net_device *dev, int how);
569d3645 84static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 85
86static int ip6_pkt_discard(struct sk_buff *skb);
aad88724 87static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb);
7150aede 88static int ip6_pkt_prohibit(struct sk_buff *skb);
aad88724 89static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb);
1da177e4 90static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 91static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
92 struct sk_buff *skb, u32 mtu);
93static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
94 struct sk_buff *skb);
4b32b5ad 95static void rt6_dst_from_metrics_check(struct rt6_info *rt);
52bd4c0c 96static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 97
70ceb4f5 98#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 99static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 100 const struct in6_addr *prefix, int prefixlen,
101 const struct in6_addr *gwaddr, int ifindex,
95c96174 102 unsigned int pref);
efa2cea0 103static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 104 const struct in6_addr *prefix, int prefixlen,
105 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 106#endif
107
06582540
DM		 108static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
109{
4b32b5ad 110 struct rt6_info *rt = (struct rt6_info *)dst;
06582540 111
4b32b5ad
MKL		 112 if (rt->rt6i_flags & RTF_CACHE)
113 return NULL;
114 else
3b471175 115 return dst_cow_metrics_generic(dst, old);
06582540
DM		 116}
117
f894cbf8
DM		 118static inline const void *choose_neigh_daddr(struct rt6_info *rt,
119 struct sk_buff *skb,
120 const void *daddr)
39232973
DM		 121{
122 struct in6_addr *p = &rt->rt6i_gateway;
123
a7563f34 124 if (!ipv6_addr_any(p))
39232973 125 return (const void *) p;
f894cbf8
DM		 126 else if (skb)
127 return &ipv6_hdr(skb)->daddr;
39232973
DM		 128 return daddr;
129}
130
f894cbf8
DM		 131static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
132 struct sk_buff *skb,
133 const void *daddr)
d3aaeb38 134{
39232973
DM		 135 struct rt6_info *rt = (struct rt6_info *) dst;
136 struct neighbour *n;
137
f894cbf8 138 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 139 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 140 if (n)
141 return n;
142 return neigh_create(&nd_tbl, daddr, dst->dev);
143}
144
9a7ec3a9 145static struct dst_ops ip6_dst_ops_template = {
1da177e4 146 .family = AF_INET6,
1da177e4
LT		 147 .gc = ip6_dst_gc,
148 .gc_thresh = 1024,
149 .check = ip6_dst_check,
0dbaee3b 150 .default_advmss = ip6_default_advmss,
ebb762f2 151 .mtu = ip6_mtu,
06582540 152 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 153 .destroy = ip6_dst_destroy,
154 .ifdown = ip6_dst_ifdown,
155 .negative_advice = ip6_negative_advice,
156 .link_failure = ip6_link_failure,
157 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 158 .redirect = rt6_do_redirect,
1ac06e03 159 .local_out = __ip6_local_out,
d3aaeb38 160 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 161};
162
ebb762f2 163static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 164{
618f9bc7
SK		 165 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
166
167 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 168}
169
6700c270
DM		 170static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
171 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 172{
173}
174
6700c270
DM		 175static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
176 struct sk_buff *skb)
b587ee3b
DM		 177{
178}
179
0972ddb2
HB		 180static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
181 unsigned long old)
182{
183 return NULL;
184}
185
14e50e57
DM		 186static struct dst_ops ip6_dst_blackhole_ops = {
187 .family = AF_INET6,
14e50e57
DM		 188 .destroy = ip6_dst_destroy,
189 .check = ip6_dst_check,
ebb762f2 190 .mtu = ip6_blackhole_mtu,
214f45c9 191 .default_advmss = ip6_default_advmss,
14e50e57 192 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 193 .redirect = ip6_rt_blackhole_redirect,
0972ddb2 194 .cow_metrics = ip6_rt_blackhole_cow_metrics,
d3aaeb38 195 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 196};
197
62fa8a84 198static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 199 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 200};
201
fb0af4c7 202static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 203 .dst = {
204 .__refcnt = ATOMIC_INIT(1),
205 .__use = 1,
2c20cbd7 206 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 207 .error = -ENETUNREACH,
d8d1f30b
CG		 208 .input = ip6_pkt_discard,
209 .output = ip6_pkt_discard_out,
1da177e4
LT		 210 },
211 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 212 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 213 .rt6i_metric = ~(u32) 0,
214 .rt6i_ref = ATOMIC_INIT(1),
215};
216
101367c2
TG		 217#ifdef CONFIG_IPV6_MULTIPLE_TABLES
218
fb0af4c7 219static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 220 .dst = {
221 .__refcnt = ATOMIC_INIT(1),
222 .__use = 1,
2c20cbd7 223 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 224 .error = -EACCES,
d8d1f30b
CG		 225 .input = ip6_pkt_prohibit,
226 .output = ip6_pkt_prohibit_out,
101367c2
TG		 227 },
228 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 229 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 230 .rt6i_metric = ~(u32) 0,
231 .rt6i_ref = ATOMIC_INIT(1),
232};
233
fb0af4c7 234static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 235 .dst = {
236 .__refcnt = ATOMIC_INIT(1),
237 .__use = 1,
2c20cbd7 238 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 239 .error = -EINVAL,
d8d1f30b 240 .input = dst_discard,
aad88724 241 .output = dst_discard_sk,
101367c2
TG		 242 },
243 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 244 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 245 .rt6i_metric = ~(u32) 0,
246 .rt6i_ref = ATOMIC_INIT(1),
247};
248
249#endif
250
1da177e4 251/* allocate dst with ip6_dst_ops */
97bab73f 252static inline struct rt6_info *ip6_dst_alloc(struct net *net,
957c665f 253 struct net_device *dev,
8b96d22d
DM		 254 int flags,
255 struct fib6_table *table)
1da177e4 256{
97bab73f 257 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 258 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 259
97bab73f 260 if (rt) {
8104891b
SK		 261 struct dst_entry *dst = &rt->dst;
262
263 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
51ebd318 264 INIT_LIST_HEAD(&rt->rt6i_siblings);
97bab73f 265 }
cf911662 266 return rt;
1da177e4
LT		 267}
268
269static void ip6_dst_destroy(struct dst_entry *dst)
270{
271 struct rt6_info *rt = (struct rt6_info *)dst;
272 struct inet6_dev *idev = rt->rt6i_idev;
ecd98837 273 struct dst_entry *from = dst->from;
1da177e4 274
4b32b5ad 275 dst_destroy_metrics_generic(dst);
8e2ec639 276
38308473 277 if (idev) {
1da177e4
LT		 278 rt->rt6i_idev = NULL;
279 in6_dev_put(idev);
1ab1457c 280 }
1716a961 281
ecd98837
YH		 282 dst->from = NULL;
283 dst_release(from);
b3419363
DM		 284}
285
1da177e4
LT		 286static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
287 int how)
288{
289 struct rt6_info *rt = (struct rt6_info *)dst;
290 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 291 struct net_device *loopback_dev =
c346dca1 292 dev_net(dev)->loopback_dev;
1da177e4 293
97cac082
DM		 294 if (dev != loopback_dev) {
295 if (idev && idev->dev == dev) {
296 struct inet6_dev *loopback_idev =
297 in6_dev_get(loopback_dev);
298 if (loopback_idev) {
299 rt->rt6i_idev = loopback_idev;
300 in6_dev_put(idev);
301 }
302 }
1da177e4
LT		 303 }
304}
305
a50feda5 306static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 307{
1716a961
G		 308 if (rt->rt6i_flags & RTF_EXPIRES) {
309 if (time_after(jiffies, rt->dst.expires))
a50feda5 310 return true;
1716a961 311 } else if (rt->dst.from) {
3fd91fb3 312 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 313 }
a50feda5 314 return false;
1da177e4
LT		 315}
316
51ebd318
ND		 317/* Multipath route selection:
318 * Hash based function using packet header and flowlabel.
319 * Adapted from fib_info_hashfn()
320 */
321static int rt6_info_hash_nhsfn(unsigned int candidate_count,
322 const struct flowi6 *fl6)
323{
324 unsigned int val = fl6->flowi6_proto;
325
c08977bb
YH		 326 val ^= ipv6_addr_hash(&fl6->daddr);
327 val ^= ipv6_addr_hash(&fl6->saddr);
51ebd318
ND		 328
329 /* Work only if this not encapsulated */
330 switch (fl6->flowi6_proto) {
331 case IPPROTO_UDP:
332 case IPPROTO_TCP:
333 case IPPROTO_SCTP:
b3ce5ae1
ND		 334 val ^= (__force u16)fl6->fl6_sport;
335 val ^= (__force u16)fl6->fl6_dport;
51ebd318
ND		 336 break;
337
338 case IPPROTO_ICMPV6:
b3ce5ae1
ND		 339 val ^= (__force u16)fl6->fl6_icmp_type;
340 val ^= (__force u16)fl6->fl6_icmp_code;
51ebd318
ND		 341 break;
342 }
343 /* RFC6438 recommands to use flowlabel */
b3ce5ae1 344 val ^= (__force u32)fl6->flowlabel;
51ebd318
ND		 345
346 /* Perhaps, we need to tune, this function? */
347 val = val ^ (val >> 7) ^ (val >> 12);
348 return val % candidate_count;
349}
350
351static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 352 struct flowi6 *fl6, int oif,
353 int strict)
51ebd318
ND		 354{
355 struct rt6_info *sibling, *next_sibling;
356 int route_choosen;
357
358 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
359 /* Don't change the route, if route_choosen == 0
360 * (siblings does not include ourself)
361 */
362 if (route_choosen)
363 list_for_each_entry_safe(sibling, next_sibling,
364 &match->rt6i_siblings, rt6i_siblings) {
365 route_choosen--;
366 if (route_choosen == 0) {
52bd4c0c
ND		 367 if (rt6_score_route(sibling, oif, strict) < 0)
368 break;
51ebd318
ND		 369 match = sibling;
370 break;
371 }
372 }
373 return match;
374}
375
1da177e4 376/*
c71099ac 377 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 378 */
379
8ed67789
DL		 380static inline struct rt6_info *rt6_device_match(struct net *net,
381 struct rt6_info *rt,
b71d1d42 382 const struct in6_addr *saddr,
1da177e4 383 int oif,
d420895e 384 int flags)
1da177e4
LT		 385{
386 struct rt6_info *local = NULL;
387 struct rt6_info *sprt;
388
dd3abc4e
YH		 389 if (!oif && ipv6_addr_any(saddr))
390 goto out;
391
d8d1f30b 392 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 393 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 394
395 if (oif) {
1da177e4
LT		 396 if (dev->ifindex == oif)
397 return sprt;
398 if (dev->flags & IFF_LOOPBACK) {
38308473 399 if (!sprt->rt6i_idev ||
1da177e4 400 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 401 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 402 continue;
1ab1457c 403 if (local && (!oif ||
1da177e4
LT		 404 local->rt6i_idev->dev->ifindex == oif))
405 continue;
406 }
407 local = sprt;
408 }
dd3abc4e
YH		 409 } else {
410 if (ipv6_chk_addr(net, saddr, dev,
411 flags & RT6_LOOKUP_F_IFACE))
412 return sprt;
1da177e4 413 }
dd3abc4e 414 }
1da177e4 415
dd3abc4e 416 if (oif) {
1da177e4
LT		 417 if (local)
418 return local;
419
d420895e 420 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 421 return net->ipv6.ip6_null_entry;
1da177e4 422 }
dd3abc4e 423out:
1da177e4
LT		 424 return rt;
425}
426
27097255 427#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 428struct __rt6_probe_work {
429 struct work_struct work;
430 struct in6_addr target;
431 struct net_device *dev;
432};
433
434static void rt6_probe_deferred(struct work_struct *w)
435{
436 struct in6_addr mcaddr;
437 struct __rt6_probe_work *work =
438 container_of(w, struct __rt6_probe_work, work);
439
440 addrconf_addr_solict_mult(&work->target, &mcaddr);
441 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
442 dev_put(work->dev);
662f5533 443 kfree(work);
c2f17e82
HFS		 444}
445
27097255
YH		 446static void rt6_probe(struct rt6_info *rt)
447{
f2c31e32 448 struct neighbour *neigh;
27097255
YH		 449 /*
450 * Okay, this does not seem to be appropriate
451 * for now, however, we need to check if it
452 * is really so; aka Router Reachability Probing.
453 *
454 * Router Reachability Probe MUST be rate-limited
455 * to no more than one per minute.
456 */
2152caea 457 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 458 return;
2152caea
YH		 459 rcu_read_lock_bh();
460 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
461 if (neigh) {
462 write_lock(&neigh->lock);
463 if (neigh->nud_state & NUD_VALID)
464 goto out;
7ff74a59 465 }
2152caea
YH		 466
467 if (!neigh ||
52e16356 468 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
c2f17e82 469 struct __rt6_probe_work *work;
27097255 470
c2f17e82
HFS		 471 work = kmalloc(sizeof(*work), GFP_ATOMIC);
472
473 if (neigh && work)
7e980569 474 __neigh_set_probe_once(neigh);
c2f17e82
HFS		 475
476 if (neigh)
2152caea
YH		 477 write_unlock(&neigh->lock);
478
c2f17e82
HFS		 479 if (work) {
480 INIT_WORK(&work->work, rt6_probe_deferred);
481 work->target = rt->rt6i_gateway;
482 dev_hold(rt->dst.dev);
483 work->dev = rt->dst.dev;
484 schedule_work(&work->work);
485 }
f2c31e32 486 } else {
2152caea
YH		 487out:
488 write_unlock(&neigh->lock);
f2c31e32 489 }
2152caea 490 rcu_read_unlock_bh();
27097255
YH		 491}
492#else
493static inline void rt6_probe(struct rt6_info *rt)
494{
27097255
YH		 495}
496#endif
497
1da177e4 498/*
554cfb7e 499 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 500 */
b6f99a21 501static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 502{
d1918542 503 struct net_device *dev = rt->dst.dev;
161980f4 504 if (!oif || dev->ifindex == oif)
554cfb7e 505 return 2;
161980f4
DM		 506 if ((dev->flags & IFF_LOOPBACK) &&
507 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
508 return 1;
509 return 0;
554cfb7e 510}
1da177e4 511
afc154e9 512static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 513{
f2c31e32 514 struct neighbour *neigh;
afc154e9 515 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 516
4d0c5911
YH		 517 if (rt->rt6i_flags & RTF_NONEXTHOP ||
518 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 519 return RT6_NUD_SUCCEED;
145a3621
YH		 520
521 rcu_read_lock_bh();
522 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
523 if (neigh) {
524 read_lock(&neigh->lock);
554cfb7e 525 if (neigh->nud_state & NUD_VALID)
afc154e9 526 ret = RT6_NUD_SUCCEED;
398bcbeb 527#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 528 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 529 ret = RT6_NUD_SUCCEED;
7e980569
JB		 530 else
531 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 532#endif
145a3621 533 read_unlock(&neigh->lock);
afc154e9
HFS		 534 } else {
535 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 536 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 537 }
145a3621
YH		 538 rcu_read_unlock_bh();
539
a5a81f0b 540 return ret;
1da177e4
LT		 541}
542
554cfb7e
YH		 543static int rt6_score_route(struct rt6_info *rt, int oif,
544 int strict)
1da177e4 545{
a5a81f0b 546 int m;
1ab1457c 547
4d0c5911 548 m = rt6_check_dev(rt, oif);
77d16f45 549 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 550 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 551#ifdef CONFIG_IPV6_ROUTER_PREF
552 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
553#endif
afc154e9
HFS		 554 if (strict & RT6_LOOKUP_F_REACHABLE) {
555 int n = rt6_check_neigh(rt);
556 if (n < 0)
557 return n;
558 }
554cfb7e
YH		 559 return m;
560}
561
f11e6659 562static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 563 int *mpri, struct rt6_info *match,
564 bool *do_rr)
554cfb7e 565{
f11e6659 566 int m;
afc154e9 567 bool match_do_rr = false;
f11e6659
DM		 568
569 if (rt6_check_expired(rt))
570 goto out;
571
572 m = rt6_score_route(rt, oif, strict);
7e980569 573 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 574 match_do_rr = true;
575 m = 0; /* lowest valid score */
7e980569 576 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 577 goto out;
afc154e9
HFS		 578 }
579
580 if (strict & RT6_LOOKUP_F_REACHABLE)
581 rt6_probe(rt);
f11e6659 582
7e980569 583 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 584 if (m > *mpri) {
afc154e9 585 *do_rr = match_do_rr;
f11e6659
DM		 586 *mpri = m;
587 match = rt;
f11e6659 588 }
f11e6659
DM		 589out:
590 return match;
591}
592
593static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
594 struct rt6_info *rr_head,
afc154e9
HFS		 595 u32 metric, int oif, int strict,
596 bool *do_rr)
f11e6659 597{
9fbdcfaf 598 struct rt6_info *rt, *match, *cont;
554cfb7e 599 int mpri = -1;
1da177e4 600
f11e6659 601 match = NULL;
9fbdcfaf
SK		 602 cont = NULL;
603 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
604 if (rt->rt6i_metric != metric) {
605 cont = rt;
606 break;
607 }
608
609 match = find_match(rt, oif, strict, &mpri, match, do_rr);
610 }
611
612 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
613 if (rt->rt6i_metric != metric) {
614 cont = rt;
615 break;
616 }
617
afc154e9 618 match = find_match(rt, oif, strict, &mpri, match, do_rr);
9fbdcfaf
SK		 619 }
620
621 if (match || !cont)
622 return match;
623
624 for (rt = cont; rt; rt = rt->dst.rt6_next)
afc154e9 625 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 626
f11e6659
DM		 627 return match;
628}
1da177e4 629
f11e6659
DM		 630static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
631{
632 struct rt6_info *match, *rt0;
8ed67789 633 struct net *net;
afc154e9 634 bool do_rr = false;
1da177e4 635
f11e6659
DM		 636 rt0 = fn->rr_ptr;
637 if (!rt0)
638 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 639
afc154e9
HFS		 640 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
641 &do_rr);
1da177e4 642
afc154e9 643 if (do_rr) {
d8d1f30b 644 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 645
554cfb7e 646 /* no entries matched; do round-robin */
f11e6659
DM		 647 if (!next || next->rt6i_metric != rt0->rt6i_metric)
648 next = fn->leaf;
649
650 if (next != rt0)
651 fn->rr_ptr = next;
1da177e4 652 }
1da177e4 653
d1918542 654 net = dev_net(rt0->dst.dev);
a02cec21 655 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 656}
657
8b9df265
MKL		 658static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
659{
660 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
661}
662
70ceb4f5
YH		 663#ifdef CONFIG_IPV6_ROUTE_INFO
664int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 665 const struct in6_addr *gwaddr)
70ceb4f5 666{
c346dca1 667 struct net *net = dev_net(dev);
70ceb4f5
YH		 668 struct route_info *rinfo = (struct route_info *) opt;
669 struct in6_addr prefix_buf, *prefix;
670 unsigned int pref;
4bed72e4 671 unsigned long lifetime;
70ceb4f5
YH		 672 struct rt6_info *rt;
673
674 if (len < sizeof(struct route_info)) {
675 return -EINVAL;
676 }
677
678 /* Sanity check for prefix_len and length */
679 if (rinfo->length > 3) {
680 return -EINVAL;
681 } else if (rinfo->prefix_len > 128) {
682 return -EINVAL;
683 } else if (rinfo->prefix_len > 64) {
684 if (rinfo->length < 2) {
685 return -EINVAL;
686 }
687 } else if (rinfo->prefix_len > 0) {
688 if (rinfo->length < 1) {
689 return -EINVAL;
690 }
691 }
692
693 pref = rinfo->route_pref;
694 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 695 return -EINVAL;
70ceb4f5 696
4bed72e4 697 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 698
699 if (rinfo->length == 3)
700 prefix = (struct in6_addr *)rinfo->prefix;
701 else {
702 /* this function is safe */
703 ipv6_addr_prefix(&prefix_buf,
704 (struct in6_addr *)rinfo->prefix,
705 rinfo->prefix_len);
706 prefix = &prefix_buf;
707 }
708
f104a567
DJ		 709 if (rinfo->prefix_len == 0)
710 rt = rt6_get_dflt_router(gwaddr, dev);
711 else
712 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
713 gwaddr, dev->ifindex);
70ceb4f5
YH		 714
715 if (rt && !lifetime) {
e0a1ad73 716 ip6_del_rt(rt);
70ceb4f5
YH		 717 rt = NULL;
718 }
719
720 if (!rt && lifetime)
efa2cea0 721 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 722 pref);
723 else if (rt)
724 rt->rt6i_flags = RTF_ROUTEINFO |
725 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
726
727 if (rt) {
1716a961
G		 728 if (!addrconf_finite_timeout(lifetime))
729 rt6_clean_expires(rt);
730 else
731 rt6_set_expires(rt, jiffies + HZ * lifetime);
732
94e187c0 733 ip6_rt_put(rt);
70ceb4f5
YH		 734 }
735 return 0;
736}
737#endif
738
a3c00e46
MKL		 739static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
740 struct in6_addr *saddr)
741{
742 struct fib6_node *pn;
743 while (1) {
744 if (fn->fn_flags & RTN_TL_ROOT)
745 return NULL;
746 pn = fn->parent;
747 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
748 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
749 else
750 fn = pn;
751 if (fn->fn_flags & RTN_RTINFO)
752 return fn;
753 }
754}
c71099ac 755
8ed67789
DL		 756static struct rt6_info *ip6_pol_route_lookup(struct net *net,
757 struct fib6_table *table,
4c9483b2 758 struct flowi6 *fl6, int flags)
1da177e4
LT		 759{
760 struct fib6_node *fn;
761 struct rt6_info *rt;
762
c71099ac 763 read_lock_bh(&table->tb6_lock);
4c9483b2 764 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 765restart:
766 rt = fn->leaf;
4c9483b2 767 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 768 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 769 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 770 if (rt == net->ipv6.ip6_null_entry) {
771 fn = fib6_backtrack(fn, &fl6->saddr);
772 if (fn)
773 goto restart;
774 }
d8d1f30b 775 dst_use(&rt->dst, jiffies);
c71099ac 776 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 777 return rt;
778
779}
780
67ba4152 781struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 782 int flags)
783{
784 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
785}
786EXPORT_SYMBOL_GPL(ip6_route_lookup);
787
9acd9f3a
YH		 788struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
789 const struct in6_addr *saddr, int oif, int strict)
c71099ac 790{
4c9483b2
DM		 791 struct flowi6 fl6 = {
792 .flowi6_oif = oif,
793 .daddr = *daddr,
c71099ac
TG		 794 };
795 struct dst_entry *dst;
77d16f45 796 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 797
adaa70bb 798 if (saddr) {
4c9483b2 799 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 800 flags |= RT6_LOOKUP_F_HAS_SADDR;
801 }
802
4c9483b2 803 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 804 if (dst->error == 0)
805 return (struct rt6_info *) dst;
806
807 dst_release(dst);
808
1da177e4
LT		 809 return NULL;
810}
7159039a
YH		 811EXPORT_SYMBOL(rt6_lookup);
812
c71099ac 813/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 814 It takes new route entry, the addition fails by any reason the
815 route is freed. In any case, if caller does not hold it, it may
816 be destroyed.
817 */
818
e5fd387a 819static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 820 struct mx6_config *mxc)
1da177e4
LT		 821{
822 int err;
c71099ac 823 struct fib6_table *table;
1da177e4 824
c71099ac
TG		 825 table = rt->rt6i_table;
826 write_lock_bh(&table->tb6_lock);
e715b6d3 827 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 828 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 829
830 return err;
831}
832
40e22e8f
TG		 833int ip6_ins_rt(struct rt6_info *rt)
834{
e715b6d3
FW		 835 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
836 struct mx6_config mxc = { .mx = NULL, };
837
838 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 839}
840
8b9df265
MKL		 841static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
842 const struct in6_addr *daddr,
843 const struct in6_addr *saddr)
1da177e4 844{
1da177e4
LT		 845 struct rt6_info *rt;
846
847 /*
848 * Clone the route.
849 */
850
21efcfa0 851 rt = ip6_rt_copy(ort, daddr);
1da177e4
LT		 852
853 if (rt) {
1da177e4 854 rt->rt6i_flags |= RTF_CACHE;
1da177e4 855
8b9df265
MKL		 856 if (!rt6_is_gw_or_nonexthop(ort)) {
857 if (ort->rt6i_dst.plen != 128 &&
858 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
859 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 860#ifdef CONFIG_IPV6_SUBTREES
8b9df265
MKL		 861 if (rt->rt6i_src.plen && saddr) {
862 rt->rt6i_src.addr = *saddr;
863 rt->rt6i_src.plen = 128;
864 }
1da177e4 865#endif
8b9df265 866 }
95a9a5ba 867 }
1da177e4 868
95a9a5ba
YH		 869 return rt;
870}
1da177e4 871
8ed67789 872static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 873 struct flowi6 *fl6, int flags)
1da177e4 874{
367efcb9 875 struct fib6_node *fn, *saved_fn;
45e4fd26 876 struct rt6_info *rt;
c71099ac 877 int strict = 0;
1da177e4 878
77d16f45 879 strict |= flags & RT6_LOOKUP_F_IFACE;
367efcb9
MKL		 880 if (net->ipv6.devconf_all->forwarding == 0)
881 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 882
c71099ac 883 read_lock_bh(&table->tb6_lock);
1da177e4 884
4c9483b2 885 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 886 saved_fn = fn;
1da177e4 887
a3c00e46 888redo_rt6_select:
367efcb9 889 rt = rt6_select(fn, oif, strict);
52bd4c0c 890 if (rt->rt6i_nsiblings)
367efcb9 891 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 892 if (rt == net->ipv6.ip6_null_entry) {
893 fn = fib6_backtrack(fn, &fl6->saddr);
894 if (fn)
895 goto redo_rt6_select;
367efcb9
MKL		 896 else if (strict & RT6_LOOKUP_F_REACHABLE) {
897 /* also consider unreachable route */
898 strict &= ~RT6_LOOKUP_F_REACHABLE;
899 fn = saved_fn;
900 goto redo_rt6_select;
367efcb9 901 }
a3c00e46
MKL		 902 }
903
3da59bd9 904 dst_use(&rt->dst, jiffies);
c71099ac 905 read_unlock_bh(&table->tb6_lock);
fb9de91e 906
3da59bd9
MKL		 907 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
908 goto done;
909 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
910 !(rt->rt6i_flags & RTF_GATEWAY))) {
911 /* Create a RTF_CACHE clone which will not be
912 * owned by the fib6 tree. It is for the special case where
913 * the daddr in the skb during the neighbor look-up is different
914 * from the fl6->daddr used to look-up route here.
915 */
916
917 struct rt6_info *uncached_rt;
918
919 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
920 dst_release(&rt->dst);
c71099ac 921
3da59bd9
MKL		 922 if (uncached_rt)
923 uncached_rt->dst.flags |= DST_NOCACHE;
924 else
925 uncached_rt = net->ipv6.ip6_null_entry;
926 dst_hold(&uncached_rt->dst);
927 return uncached_rt;
928 }
929
930done:
931 rt6_dst_from_metrics_check(rt);
c71099ac 932 return rt;
1da177e4
LT		 933}
934
8ed67789 935static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 936 struct flowi6 *fl6, int flags)
4acad72d 937{
4c9483b2 938 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 939}
940
72331bc0
SL		 941static struct dst_entry *ip6_route_input_lookup(struct net *net,
942 struct net_device *dev,
943 struct flowi6 *fl6, int flags)
944{
945 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
946 flags |= RT6_LOOKUP_F_IFACE;
947
948 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
949}
950
c71099ac
TG		 951void ip6_route_input(struct sk_buff *skb)
952{
b71d1d42 953 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 954 struct net *net = dev_net(skb->dev);
adaa70bb 955 int flags = RT6_LOOKUP_F_HAS_SADDR;
4c9483b2
DM		 956 struct flowi6 fl6 = {
957 .flowi6_iif = skb->dev->ifindex,
958 .daddr = iph->daddr,
959 .saddr = iph->saddr,
6502ca52 960 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 961 .flowi6_mark = skb->mark,
962 .flowi6_proto = iph->nexthdr,
c71099ac 963 };
adaa70bb 964
72331bc0 965 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 966}
967
8ed67789 968static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 969 struct flowi6 *fl6, int flags)
1da177e4 970{
4c9483b2 971 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 972}
973
67ba4152 974struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 975 struct flowi6 *fl6)
c71099ac
TG		 976{
977 int flags = 0;
978
1fb9489b 979 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 980
4c9483b2 981 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
77d16f45 982 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 983
4c9483b2 984 if (!ipv6_addr_any(&fl6->saddr))
adaa70bb 985 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 986 else if (sk)
987 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 988
4c9483b2 989 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 990}
7159039a 991EXPORT_SYMBOL(ip6_route_output);
1da177e4 992
2774c131 993struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 994{
5c1e6aa3 995 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 996 struct dst_entry *new = NULL;
997
f5b0a874 998 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 999 if (rt) {
d8d1f30b 1000 new = &rt->dst;
14e50e57 1001
8104891b 1002 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
8104891b 1003
14e50e57 1004 new->__use = 1;
352e512c 1005 new->input = dst_discard;
aad88724 1006 new->output = dst_discard_sk;
14e50e57 1007
21efcfa0
ED		 1008 if (dst_metrics_read_only(&ort->dst))
1009 new->_metrics = ort->dst._metrics;
1010 else
1011 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1012 rt->rt6i_idev = ort->rt6i_idev;
1013 if (rt->rt6i_idev)
1014 in6_dev_hold(rt->rt6i_idev);
14e50e57 1015
4e3fd7a0 1016 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1017 rt->rt6i_flags = ort->rt6i_flags;
14e50e57
DM		 1018 rt->rt6i_metric = 0;
1019
1020 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1021#ifdef CONFIG_IPV6_SUBTREES
1022 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1023#endif
1024
1025 dst_free(new);
1026 }
1027
69ead7af
DM		 1028 dst_release(dst_orig);
1029 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1030}
14e50e57 1031
1da177e4
LT		 1032/*
1033 * Destination cache support functions
1034 */
1035
4b32b5ad
MKL		 1036static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1037{
1038 if (rt->dst.from &&
1039 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1040 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1041}
1042
3da59bd9
MKL		 1043static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1044{
1045 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1046 return NULL;
1047
1048 if (rt6_check_expired(rt))
1049 return NULL;
1050
1051 return &rt->dst;
1052}
1053
1054static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1055{
1056 if (rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
1057 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1058 return &rt->dst;
1059 else
1060 return NULL;
1061}
1062
1da177e4
LT		 1063static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1064{
1065 struct rt6_info *rt;
1066
1067 rt = (struct rt6_info *) dst;
1068
6f3118b5
ND		 1069 /* All IPV6 dsts are created with ->obsolete set to the value
1070 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1071 * into this function always.
1072 */
e3bc10bd 1073
4b32b5ad
MKL		 1074 rt6_dst_from_metrics_check(rt);
1075
3da59bd9
MKL		 1076 if (unlikely(dst->flags & DST_NOCACHE))
1077 return rt6_dst_from_check(rt, cookie);
1078 else
1079 return rt6_check(rt, cookie);
1da177e4
LT		 1080}
1081
1082static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1083{
1084 struct rt6_info *rt = (struct rt6_info *) dst;
1085
1086 if (rt) {
54c1a859
YH		 1087 if (rt->rt6i_flags & RTF_CACHE) {
1088 if (rt6_check_expired(rt)) {
1089 ip6_del_rt(rt);
1090 dst = NULL;
1091 }
1092 } else {
1da177e4 1093 dst_release(dst);
54c1a859
YH		 1094 dst = NULL;
1095 }
1da177e4 1096 }
54c1a859 1097 return dst;
1da177e4
LT		 1098}
1099
1100static void ip6_link_failure(struct sk_buff *skb)
1101{
1102 struct rt6_info *rt;
1103
3ffe533c 1104 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1105
adf30907 1106 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1107 if (rt) {
1eb4f758
HFS		 1108 if (rt->rt6i_flags & RTF_CACHE) {
1109 dst_hold(&rt->dst);
1110 if (ip6_del_rt(rt))
1111 dst_free(&rt->dst);
1112 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1113 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1114 }
1da177e4
LT		 1115 }
1116}
1117
45e4fd26
MKL		 1118static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1119{
1120 struct net *net = dev_net(rt->dst.dev);
1121
1122 rt->rt6i_flags |= RTF_MODIFIED;
1123 rt->rt6i_pmtu = mtu;
1124 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1125}
1126
1127static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1128 const struct ipv6hdr *iph, u32 mtu)
1da177e4 1129{
67ba4152 1130 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1131
45e4fd26
MKL		 1132 if (rt6->rt6i_flags & RTF_LOCAL)
1133 return;
81aded24 1134
45e4fd26
MKL		 1135 dst_confirm(dst);
1136 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1137 if (mtu >= dst_mtu(dst))
1138 return;
9d289715 1139
45e4fd26
MKL		 1140 if (rt6->rt6i_flags & RTF_CACHE) {
1141 rt6_do_update_pmtu(rt6, mtu);
1142 } else {
1143 const struct in6_addr *daddr, *saddr;
1144 struct rt6_info *nrt6;
1145
1146 if (iph) {
1147 daddr = &iph->daddr;
1148 saddr = &iph->saddr;
1149 } else if (sk) {
1150 daddr = &sk->sk_v6_daddr;
1151 saddr = &inet6_sk(sk)->saddr;
1152 } else {
1153 return;
1154 }
1155 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1156 if (nrt6) {
1157 rt6_do_update_pmtu(nrt6, mtu);
1158
1159 /* ip6_ins_rt(nrt6) will bump the
1160 * rt6->rt6i_node->fn_sernum
1161 * which will fail the next rt6_check() and
1162 * invalidate the sk->sk_dst_cache.
1163 */
1164 ip6_ins_rt(nrt6);
1165 }
1da177e4
LT		 1166 }
1167}
1168
45e4fd26
MKL		 1169static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1170 struct sk_buff *skb, u32 mtu)
1171{
1172 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1173}
1174
42ae66c8
DM		 1175void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1176 int oif, u32 mark)
81aded24
DM		 1177{
1178 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1179 struct dst_entry *dst;
1180 struct flowi6 fl6;
1181
1182 memset(&fl6, 0, sizeof(fl6));
1183 fl6.flowi6_oif = oif;
1b3c61dc 1184 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1185 fl6.daddr = iph->daddr;
1186 fl6.saddr = iph->saddr;
6502ca52 1187 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1188
1189 dst = ip6_route_output(net, NULL, &fl6);
1190 if (!dst->error)
45e4fd26 1191 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
81aded24
DM		 1192 dst_release(dst);
1193}
1194EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1195
1196void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1197{
1198 ip6_update_pmtu(skb, sock_net(sk), mtu,
1199 sk->sk_bound_dev_if, sk->sk_mark);
1200}
1201EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1202
b55b76b2
DJ		 1203/* Handle redirects */
1204struct ip6rd_flowi {
1205 struct flowi6 fl6;
1206 struct in6_addr gateway;
1207};
1208
1209static struct rt6_info *__ip6_route_redirect(struct net *net,
1210 struct fib6_table *table,
1211 struct flowi6 *fl6,
1212 int flags)
1213{
1214 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1215 struct rt6_info *rt;
1216 struct fib6_node *fn;
1217
1218 /* Get the "current" route for this destination and
1219 * check if the redirect has come from approriate router.
1220 *
1221 * RFC 4861 specifies that redirects should only be
1222 * accepted if they come from the nexthop to the target.
1223 * Due to the way the routes are chosen, this notion
1224 * is a bit fuzzy and one might need to check all possible
1225 * routes.
1226 */
1227
1228 read_lock_bh(&table->tb6_lock);
1229 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1230restart:
1231 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1232 if (rt6_check_expired(rt))
1233 continue;
1234 if (rt->dst.error)
1235 break;
1236 if (!(rt->rt6i_flags & RTF_GATEWAY))
1237 continue;
1238 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1239 continue;
1240 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1241 continue;
1242 break;
1243 }
1244
1245 if (!rt)
1246 rt = net->ipv6.ip6_null_entry;
1247 else if (rt->dst.error) {
1248 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1249 goto out;
1250 }
1251
1252 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1253 fn = fib6_backtrack(fn, &fl6->saddr);
1254 if (fn)
1255 goto restart;
b55b76b2 1256 }
a3c00e46 1257
b0a1ba59 1258out:
b55b76b2
DJ		 1259 dst_hold(&rt->dst);
1260
1261 read_unlock_bh(&table->tb6_lock);
1262
1263 return rt;
1264};
1265
1266static struct dst_entry *ip6_route_redirect(struct net *net,
1267 const struct flowi6 *fl6,
1268 const struct in6_addr *gateway)
1269{
1270 int flags = RT6_LOOKUP_F_HAS_SADDR;
1271 struct ip6rd_flowi rdfl;
1272
1273 rdfl.fl6 = *fl6;
1274 rdfl.gateway = *gateway;
1275
1276 return fib6_rule_lookup(net, &rdfl.fl6,
1277 flags, __ip6_route_redirect);
1278}
1279
3a5ad2ee
DM		 1280void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1281{
1282 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1283 struct dst_entry *dst;
1284 struct flowi6 fl6;
1285
1286 memset(&fl6, 0, sizeof(fl6));
e374c618 1287 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1288 fl6.flowi6_oif = oif;
1289 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1290 fl6.daddr = iph->daddr;
1291 fl6.saddr = iph->saddr;
6502ca52 1292 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1293
b55b76b2
DJ		 1294 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1295 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1296 dst_release(dst);
1297}
1298EXPORT_SYMBOL_GPL(ip6_redirect);
1299
c92a59ec
DJ		 1300void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1301 u32 mark)
1302{
1303 const struct ipv6hdr *iph = ipv6_hdr(skb);
1304 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1305 struct dst_entry *dst;
1306 struct flowi6 fl6;
1307
1308 memset(&fl6, 0, sizeof(fl6));
e374c618 1309 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1310 fl6.flowi6_oif = oif;
1311 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1312 fl6.daddr = msg->dest;
1313 fl6.saddr = iph->daddr;
1314
b55b76b2
DJ		 1315 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1316 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1317 dst_release(dst);
1318}
1319
3a5ad2ee
DM		 1320void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1321{
1322 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1323}
1324EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1325
0dbaee3b 1326static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1327{
0dbaee3b
DM		 1328 struct net_device *dev = dst->dev;
1329 unsigned int mtu = dst_mtu(dst);
1330 struct net *net = dev_net(dev);
1331
1da177e4
LT		 1332 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1333
5578689a
DL		 1334 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1335 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1336
1337 /*
1ab1457c
YH		 1338 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1339 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1340 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1341 * rely only on pmtu discovery"
1342 */
1343 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1344 mtu = IPV6_MAXPLEN;
1345 return mtu;
1346}
1347
ebb762f2 1348static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1349{
4b32b5ad
MKL		 1350 const struct rt6_info *rt = (const struct rt6_info *)dst;
1351 unsigned int mtu = rt->rt6i_pmtu;
d33e4553 1352 struct inet6_dev *idev;
618f9bc7 1353
4b32b5ad
MKL		 1354 if (mtu)
1355 goto out;
1356
1357 mtu = dst_metric_raw(dst, RTAX_MTU);
618f9bc7 1358 if (mtu)
30f78d8e 1359 goto out;
618f9bc7
SK		 1360
1361 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1362
1363 rcu_read_lock();
1364 idev = __in6_dev_get(dst->dev);
1365 if (idev)
1366 mtu = idev->cnf.mtu6;
1367 rcu_read_unlock();
1368
30f78d8e
ED		 1369out:
1370 return min_t(unsigned int, mtu, IP6_MAX_MTU);
d33e4553
DM		 1371}
1372
3b00944c
YH		 1373static struct dst_entry *icmp6_dst_gc_list;
1374static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1375
3b00944c 1376struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1377 struct flowi6 *fl6)
1da177e4 1378{
87a11578 1379 struct dst_entry *dst;
1da177e4
LT		 1380 struct rt6_info *rt;
1381 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1382 struct net *net = dev_net(dev);
1da177e4 1383
38308473 1384 if (unlikely(!idev))
122bdf67 1385 return ERR_PTR(-ENODEV);
1da177e4 1386
8b96d22d 1387 rt = ip6_dst_alloc(net, dev, 0, NULL);
38308473 1388 if (unlikely(!rt)) {
1da177e4 1389 in6_dev_put(idev);
87a11578 1390 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1391 goto out;
1392 }
1393
8e2ec639
YZ		 1394 rt->dst.flags |= DST_HOST;
1395 rt->dst.output = ip6_output;
d8d1f30b 1396 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1397 rt->rt6i_gateway = fl6->daddr;
87a11578 1398 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1399 rt->rt6i_dst.plen = 128;
1400 rt->rt6i_idev = idev;
14edd87d 1401 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1402
3b00944c 1403 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1404 rt->dst.next = icmp6_dst_gc_list;
1405 icmp6_dst_gc_list = &rt->dst;
3b00944c 1406 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1407
5578689a 1408 fib6_force_start_gc(net);
1da177e4 1409
87a11578
DM		 1410 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1411
1da177e4 1412out:
87a11578 1413 return dst;
1da177e4
LT		 1414}
1415
3d0f24a7 1416int icmp6_dst_gc(void)
1da177e4 1417{
e9476e95 1418 struct dst_entry *dst, **pprev;
3d0f24a7 1419 int more = 0;
1da177e4 1420
3b00944c
YH		 1421 spin_lock_bh(&icmp6_dst_lock);
1422 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1423
1da177e4
LT		 1424 while ((dst = *pprev) != NULL) {
1425 if (!atomic_read(&dst->__refcnt)) {
1426 *pprev = dst->next;
1427 dst_free(dst);
1da177e4
LT		 1428 } else {
1429 pprev = &dst->next;
3d0f24a7 1430 ++more;
1da177e4
LT		 1431 }
1432 }
1433
3b00944c 1434 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1435
3d0f24a7 1436 return more;
1da177e4
LT		 1437}
1438
1e493d19
DM		 1439static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1440 void *arg)
1441{
1442 struct dst_entry *dst, **pprev;
1443
1444 spin_lock_bh(&icmp6_dst_lock);
1445 pprev = &icmp6_dst_gc_list;
1446 while ((dst = *pprev) != NULL) {
1447 struct rt6_info *rt = (struct rt6_info *) dst;
1448 if (func(rt, arg)) {
1449 *pprev = dst->next;
1450 dst_free(dst);
1451 } else {
1452 pprev = &dst->next;
1453 }
1454 }
1455 spin_unlock_bh(&icmp6_dst_lock);
1456}
1457
569d3645 1458static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1459{
86393e52 1460 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1461 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1462 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1463 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1464 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1465 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1466 int entries;
7019b78e 1467
fc66f95c 1468 entries = dst_entries_get_fast(ops);
49a18d86 1469 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1470 entries <= rt_max_size)
1da177e4
LT		 1471 goto out;
1472
6891a346 1473 net->ipv6.ip6_rt_gc_expire++;
14956643 1474 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1475 entries = dst_entries_get_slow(ops);
1476 if (entries < ops->gc_thresh)
7019b78e 1477 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1478out:
7019b78e 1479 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1480 return entries > rt_max_size;
1da177e4
LT		 1481}
1482
e715b6d3
FW		 1483static int ip6_convert_metrics(struct mx6_config *mxc,
1484 const struct fib6_config *cfg)
1485{
1486 struct nlattr *nla;
1487 int remaining;
1488 u32 *mp;
1489
63159f29 1490 if (!cfg->fc_mx)
e715b6d3
FW		 1491 return 0;
1492
1493 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1494 if (unlikely(!mp))
1495 return -ENOMEM;
1496
1497 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1498 int type = nla_type(nla);
1499
1500 if (type) {
ea697639
DB		 1501 u32 val;
1502
e715b6d3
FW		 1503 if (unlikely(type > RTAX_MAX))
1504 goto err;
ea697639
DB		 1505 if (type == RTAX_CC_ALGO) {
1506 char tmp[TCP_CA_NAME_MAX];
1507
1508 nla_strlcpy(tmp, nla, sizeof(tmp));
1509 val = tcp_ca_get_key_by_name(tmp);
1510 if (val == TCP_CA_UNSPEC)
1511 goto err;
1512 } else {
1513 val = nla_get_u32(nla);
1514 }
e715b6d3 1515
ea697639 1516 mp[type - 1] = val;
e715b6d3
FW		 1517 __set_bit(type - 1, mxc->mx_valid);
1518 }
1519 }
1520
1521 mxc->mx = mp;
1522
1523 return 0;
1524 err:
1525 kfree(mp);
1526 return -EINVAL;
1527}
1da177e4 1528
86872cb5 1529int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1530{
1531 int err;
5578689a 1532 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1533 struct rt6_info *rt = NULL;
1534 struct net_device *dev = NULL;
1535 struct inet6_dev *idev = NULL;
c71099ac 1536 struct fib6_table *table;
e715b6d3 1537 struct mx6_config mxc = { .mx = NULL, };
1da177e4
LT		 1538 int addr_type;
1539
86872cb5 1540 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1541 return -EINVAL;
1542#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1543 if (cfg->fc_src_len)
1da177e4
LT		 1544 return -EINVAL;
1545#endif
86872cb5 1546 if (cfg->fc_ifindex) {
1da177e4 1547 err = -ENODEV;
5578689a 1548 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1549 if (!dev)
1550 goto out;
1551 idev = in6_dev_get(dev);
1552 if (!idev)
1553 goto out;
1554 }
1555
86872cb5
TG		 1556 if (cfg->fc_metric == 0)
1557 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1558
d71314b4 1559 err = -ENOBUFS;
38308473
DM		 1560 if (cfg->fc_nlinfo.nlh &&
1561 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1562 table = fib6_get_table(net, cfg->fc_table);
38308473 1563 if (!table) {
f3213831 1564 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1565 table = fib6_new_table(net, cfg->fc_table);
1566 }
1567 } else {
1568 table = fib6_new_table(net, cfg->fc_table);
1569 }
38308473
DM		 1570
1571 if (!table)
c71099ac 1572 goto out;
c71099ac 1573
c88507fb 1574 rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table);
1da177e4 1575
38308473 1576 if (!rt) {
1da177e4
LT		 1577 err = -ENOMEM;
1578 goto out;
1579 }
1580
1716a961
G		 1581 if (cfg->fc_flags & RTF_EXPIRES)
1582 rt6_set_expires(rt, jiffies +
1583 clock_t_to_jiffies(cfg->fc_expires));
1584 else
1585 rt6_clean_expires(rt);
1da177e4 1586
86872cb5
TG		 1587 if (cfg->fc_protocol == RTPROT_UNSPEC)
1588 cfg->fc_protocol = RTPROT_BOOT;
1589 rt->rt6i_protocol = cfg->fc_protocol;
1590
1591 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1592
1593 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1594 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1595 else if (cfg->fc_flags & RTF_LOCAL)
1596 rt->dst.input = ip6_input;
1da177e4 1597 else
d8d1f30b 1598 rt->dst.input = ip6_forward;
1da177e4 1599
d8d1f30b 1600 rt->dst.output = ip6_output;
1da177e4 1601
86872cb5
TG		 1602 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1603 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1604 if (rt->rt6i_dst.plen == 128)
e5fd387a 1605 rt->dst.flags |= DST_HOST;
e5fd387a 1606
1da177e4 1607#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1608 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1609 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1610#endif
1611
86872cb5 1612 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1613
1614 /* We cannot add true routes via loopback here,
1615 they would result in kernel looping; promote them to reject routes
1616 */
86872cb5 1617 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1618 (dev && (dev->flags & IFF_LOOPBACK) &&
1619 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1620 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1621 /* hold loopback dev/idev if we haven't done so. */
5578689a 1622 if (dev != net->loopback_dev) {
1da177e4
LT		 1623 if (dev) {
1624 dev_put(dev);
1625 in6_dev_put(idev);
1626 }
5578689a 1627 dev = net->loopback_dev;
1da177e4
LT		 1628 dev_hold(dev);
1629 idev = in6_dev_get(dev);
1630 if (!idev) {
1631 err = -ENODEV;
1632 goto out;
1633 }
1634 }
1da177e4 1635 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1636 switch (cfg->fc_type) {
1637 case RTN_BLACKHOLE:
1638 rt->dst.error = -EINVAL;
aad88724 1639 rt->dst.output = dst_discard_sk;
7150aede 1640 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1641 break;
1642 case RTN_PROHIBIT:
1643 rt->dst.error = -EACCES;
7150aede
K		 1644 rt->dst.output = ip6_pkt_prohibit_out;
1645 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1646 break;
b4949ab2 1647 case RTN_THROW:
ef2c7d7b 1648 default:
7150aede
K		 1649 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1650 : -ENETUNREACH;
1651 rt->dst.output = ip6_pkt_discard_out;
1652 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1653 break;
1654 }
1da177e4
LT		 1655 goto install_route;
1656 }
1657
86872cb5 1658 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1659 const struct in6_addr *gw_addr;
1da177e4
LT		 1660 int gwa_type;
1661
86872cb5 1662 gw_addr = &cfg->fc_gateway;
48ed7b26
FW		 1663
1664 /* if gw_addr is local we will fail to detect this in case
1665 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1666 * will return already-added prefix route via interface that
1667 * prefix route was assigned to, which might be non-loopback.
1668 */
1669 err = -EINVAL;
1670 if (ipv6_chk_addr_and_flags(net, gw_addr, NULL, 0, 0))
1671 goto out;
1672
4e3fd7a0 1673 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1674 gwa_type = ipv6_addr_type(gw_addr);
1675
1676 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1677 struct rt6_info *grt;
1678
1679 /* IPv6 strictly inhibits using not link-local
1680 addresses as nexthop address.
1681 Otherwise, router will not able to send redirects.
1682 It is very good, but in some (rare!) circumstances
1683 (SIT, PtP, NBMA NOARP links) it is handy to allow
1684 some exceptions. --ANK
1685 */
38308473 1686 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1687 goto out;
1688
5578689a 1689 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1690
1691 err = -EHOSTUNREACH;
38308473 1692 if (!grt)
1da177e4
LT		 1693 goto out;
1694 if (dev) {
d1918542 1695 if (dev != grt->dst.dev) {
94e187c0 1696 ip6_rt_put(grt);
1da177e4
LT		 1697 goto out;
1698 }
1699 } else {
d1918542 1700 dev = grt->dst.dev;
1da177e4
LT		 1701 idev = grt->rt6i_idev;
1702 dev_hold(dev);
1703 in6_dev_hold(grt->rt6i_idev);
1704 }
38308473 1705 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1706 err = 0;
94e187c0 1707 ip6_rt_put(grt);
1da177e4
LT		 1708
1709 if (err)
1710 goto out;
1711 }
1712 err = -EINVAL;
38308473 1713 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1714 goto out;
1715 }
1716
1717 err = -ENODEV;
38308473 1718 if (!dev)
1da177e4
LT		 1719 goto out;
1720
c3968a85
DW		 1721 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1722 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1723 err = -EINVAL;
1724 goto out;
1725 }
4e3fd7a0 1726 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1727 rt->rt6i_prefsrc.plen = 128;
1728 } else
1729 rt->rt6i_prefsrc.plen = 0;
1730
86872cb5 1731 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1732
1733install_route:
d8d1f30b 1734 rt->dst.dev = dev;
1da177e4 1735 rt->rt6i_idev = idev;
c71099ac 1736 rt->rt6i_table = table;
63152fc0 1737
c346dca1 1738 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1739
e715b6d3
FW		 1740 err = ip6_convert_metrics(&mxc, cfg);
1741 if (err)
1742 goto out;
1da177e4 1743
e715b6d3
FW		 1744 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
1745
1746 kfree(mxc.mx);
1747 return err;
1da177e4
LT		 1748out:
1749 if (dev)
1750 dev_put(dev);
1751 if (idev)
1752 in6_dev_put(idev);
1753 if (rt)
d8d1f30b 1754 dst_free(&rt->dst);
1da177e4
LT		 1755 return err;
1756}
1757
86872cb5 1758static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1759{
1760 int err;
c71099ac 1761 struct fib6_table *table;
d1918542 1762 struct net *net = dev_net(rt->dst.dev);
1da177e4 1763
6825a26c
G		 1764 if (rt == net->ipv6.ip6_null_entry) {
1765 err = -ENOENT;
1766 goto out;
1767 }
6c813a72 1768
c71099ac
TG		 1769 table = rt->rt6i_table;
1770 write_lock_bh(&table->tb6_lock);
86872cb5 1771 err = fib6_del(rt, info);
c71099ac 1772 write_unlock_bh(&table->tb6_lock);
1da177e4 1773
6825a26c 1774out:
94e187c0 1775 ip6_rt_put(rt);
1da177e4
LT		 1776 return err;
1777}
1778
e0a1ad73
TG		 1779int ip6_del_rt(struct rt6_info *rt)
1780{
4d1169c1 1781 struct nl_info info = {
d1918542 1782 .nl_net = dev_net(rt->dst.dev),
4d1169c1 1783 };
528c4ceb 1784 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1785}
1786
86872cb5 1787static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1788{
c71099ac 1789 struct fib6_table *table;
1da177e4
LT		 1790 struct fib6_node *fn;
1791 struct rt6_info *rt;
1792 int err = -ESRCH;
1793
5578689a 1794 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 1795 if (!table)
c71099ac
TG		 1796 return err;
1797
1798 read_lock_bh(&table->tb6_lock);
1da177e4 1799
c71099ac 1800 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1801 &cfg->fc_dst, cfg->fc_dst_len,
1802 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1803
1da177e4 1804 if (fn) {
d8d1f30b 1805 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 1806 if ((rt->rt6i_flags & RTF_CACHE) &&
1807 !(cfg->fc_flags & RTF_CACHE))
1808 continue;
86872cb5 1809 if (cfg->fc_ifindex &&
d1918542
DM		 1810 (!rt->dst.dev ||
1811 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 1812 continue;
86872cb5
TG		 1813 if (cfg->fc_flags & RTF_GATEWAY &&
1814 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1815 continue;
86872cb5 1816 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 1817 continue;
d8d1f30b 1818 dst_hold(&rt->dst);
c71099ac 1819 read_unlock_bh(&table->tb6_lock);
1da177e4 1820
86872cb5 1821 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1822 }
1823 }
c71099ac 1824 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1825
1826 return err;
1827}
1828
6700c270 1829static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 1830{
e8599ff4 1831 struct net *net = dev_net(skb->dev);
a6279458 1832 struct netevent_redirect netevent;
e8599ff4 1833 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 1834 struct ndisc_options ndopts;
1835 struct inet6_dev *in6_dev;
1836 struct neighbour *neigh;
71bcdba0 1837 struct rd_msg *msg;
6e157b6a
DM		 1838 int optlen, on_link;
1839 u8 *lladdr;
e8599ff4 1840
29a3cad5 1841 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 1842 optlen -= sizeof(*msg);
e8599ff4
DM		 1843
1844 if (optlen < 0) {
6e157b6a 1845 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 1846 return;
1847 }
1848
71bcdba0 1849 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 1850
71bcdba0 1851 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 1852 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 1853 return;
1854 }
1855
6e157b6a 1856 on_link = 0;
71bcdba0 1857 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 1858 on_link = 1;
71bcdba0 1859 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 1860 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 1861 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 1862 return;
1863 }
1864
1865 in6_dev = __in6_dev_get(skb->dev);
1866 if (!in6_dev)
1867 return;
1868 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1869 return;
1870
1871 /* RFC2461 8.1:
1872 * The IP source address of the Redirect MUST be the same as the current
1873 * first-hop router for the specified ICMP Destination Address.
1874 */
1875
71bcdba0 1876 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 1877 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1878 return;
1879 }
6e157b6a
DM		 1880
1881 lladdr = NULL;
e8599ff4
DM		 1882 if (ndopts.nd_opts_tgt_lladdr) {
1883 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1884 skb->dev);
1885 if (!lladdr) {
1886 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1887 return;
1888 }
1889 }
1890
6e157b6a
DM		 1891 rt = (struct rt6_info *) dst;
1892 if (rt == net->ipv6.ip6_null_entry) {
1893 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 1894 return;
6e157b6a 1895 }
e8599ff4 1896
6e157b6a
DM		 1897 /* Redirect received -> path was valid.
1898 * Look, redirects are sent only in response to data packets,
1899 * so that this nexthop apparently is reachable. --ANK
1900 */
1901 dst_confirm(&rt->dst);
a6279458 1902
71bcdba0 1903 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 1904 if (!neigh)
1905 return;
a6279458 1906
1da177e4
LT		 1907 /*
1908 * We have finally decided to accept it.
1909 */
1910
1ab1457c 1911 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1912 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1913 NEIGH_UPDATE_F_OVERRIDE|
1914 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1915 NEIGH_UPDATE_F_ISROUTER))
1916 );
1917
71bcdba0 1918 nrt = ip6_rt_copy(rt, &msg->dest);
38308473 1919 if (!nrt)
1da177e4
LT		 1920 goto out;
1921
1922 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1923 if (on_link)
1924 nrt->rt6i_flags &= ~RTF_GATEWAY;
1925
4e3fd7a0 1926 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 1927
40e22e8f 1928 if (ip6_ins_rt(nrt))
1da177e4
LT		 1929 goto out;
1930
d8d1f30b
CG		 1931 netevent.old = &rt->dst;
1932 netevent.new = &nrt->dst;
71bcdba0 1933 netevent.daddr = &msg->dest;
60592833 1934 netevent.neigh = neigh;
8d71740c
TT		 1935 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1936
38308473 1937 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 1938 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 1939 ip6_del_rt(rt);
1da177e4
LT		 1940 }
1941
1942out:
e8599ff4 1943 neigh_release(neigh);
6e157b6a
DM		 1944}
1945
1da177e4
LT		 1946/*
1947 * Misc support functions
1948 */
1949
4b32b5ad
MKL		 1950static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
1951{
1952 BUG_ON(from->dst.from);
1953
1954 rt->rt6i_flags &= ~RTF_EXPIRES;
1955 dst_hold(&from->dst);
1956 rt->dst.from = &from->dst;
1957 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
1958}
1959
1716a961 1960static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
21efcfa0 1961 const struct in6_addr *dest)
1da177e4 1962{
d1918542 1963 struct net *net = dev_net(ort->dst.dev);
4b32b5ad
MKL		 1964 struct rt6_info *rt;
1965
1966 if (ort->rt6i_flags & RTF_CACHE)
1967 ort = (struct rt6_info *)ort->dst.from;
1968
1969 rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1970 ort->rt6i_table);
1da177e4
LT		 1971
1972 if (rt) {
d8d1f30b
CG		 1973 rt->dst.input = ort->dst.input;
1974 rt->dst.output = ort->dst.output;
8e2ec639 1975 rt->dst.flags |= DST_HOST;
d8d1f30b 1976
4e3fd7a0 1977 rt->rt6i_dst.addr = *dest;
8e2ec639 1978 rt->rt6i_dst.plen = 128;
d8d1f30b 1979 rt->dst.error = ort->dst.error;
1da177e4
LT		 1980 rt->rt6i_idev = ort->rt6i_idev;
1981 if (rt->rt6i_idev)
1982 in6_dev_hold(rt->rt6i_idev);
d8d1f30b 1983 rt->dst.lastuse = jiffies;
2647a9b0 1984 rt->rt6i_gateway = ort->rt6i_gateway;
1716a961 1985 rt->rt6i_flags = ort->rt6i_flags;
24f5b855 1986 rt6_set_from(rt, ort);
1da177e4
LT		 1987 rt->rt6i_metric = 0;
1988
1da177e4
LT		 1989#ifdef CONFIG_IPV6_SUBTREES
1990 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1991#endif
0f6c6392 1992 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
c71099ac 1993 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1994 }
1995 return rt;
1996}
1997
70ceb4f5 1998#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 1999static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 2000 const struct in6_addr *prefix, int prefixlen,
2001 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 2002{
2003 struct fib6_node *fn;
2004 struct rt6_info *rt = NULL;
c71099ac
TG		 2005 struct fib6_table *table;
2006
efa2cea0 2007 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 2008 if (!table)
c71099ac 2009 return NULL;
70ceb4f5 2010
5744dd9b 2011 read_lock_bh(&table->tb6_lock);
67ba4152 2012 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2013 if (!fn)
2014 goto out;
2015
d8d1f30b 2016 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2017 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2018 continue;
2019 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2020 continue;
2021 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2022 continue;
d8d1f30b 2023 dst_hold(&rt->dst);
70ceb4f5
YH		 2024 break;
2025 }
2026out:
5744dd9b 2027 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2028 return rt;
2029}
2030
efa2cea0 2031static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 2032 const struct in6_addr *prefix, int prefixlen,
2033 const struct in6_addr *gwaddr, int ifindex,
95c96174 2034 unsigned int pref)
70ceb4f5 2035{
86872cb5
TG		 2036 struct fib6_config cfg = {
2037 .fc_table = RT6_TABLE_INFO,
238fc7ea 2038 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2039 .fc_ifindex = ifindex,
2040 .fc_dst_len = prefixlen,
2041 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2042 RTF_UP | RTF_PREF(pref),
15e47304 2043 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2044 .fc_nlinfo.nlh = NULL,
2045 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2046 };
2047
4e3fd7a0
AD		 2048 cfg.fc_dst = *prefix;
2049 cfg.fc_gateway = *gwaddr;
70ceb4f5 2050
e317da96
YH		 2051 /* We should treat it as a default route if prefix length is 0. */
2052 if (!prefixlen)
86872cb5 2053 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2054
86872cb5 2055 ip6_route_add(&cfg);
70ceb4f5 2056
efa2cea0 2057 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 2058}
2059#endif
2060
b71d1d42 2061struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2062{
1da177e4 2063 struct rt6_info *rt;
c71099ac 2064 struct fib6_table *table;
1da177e4 2065
c346dca1 2066 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 2067 if (!table)
c71099ac 2068 return NULL;
1da177e4 2069
5744dd9b 2070 read_lock_bh(&table->tb6_lock);
67ba4152 2071 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2072 if (dev == rt->dst.dev &&
045927ff 2073 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2074 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2075 break;
2076 }
2077 if (rt)
d8d1f30b 2078 dst_hold(&rt->dst);
5744dd9b 2079 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2080 return rt;
2081}
2082
b71d1d42 2083struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2084 struct net_device *dev,
2085 unsigned int pref)
1da177e4 2086{
86872cb5
TG		 2087 struct fib6_config cfg = {
2088 .fc_table = RT6_TABLE_DFLT,
238fc7ea 2089 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2090 .fc_ifindex = dev->ifindex,
2091 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2092 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2093 .fc_nlinfo.portid = 0,
5578689a 2094 .fc_nlinfo.nlh = NULL,
c346dca1 2095 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2096 };
1da177e4 2097
4e3fd7a0 2098 cfg.fc_gateway = *gwaddr;
1da177e4 2099
86872cb5 2100 ip6_route_add(&cfg);
1da177e4 2101
1da177e4
LT		 2102 return rt6_get_dflt_router(gwaddr, dev);
2103}
2104
7b4da532 2105void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2106{
2107 struct rt6_info *rt;
c71099ac
TG		 2108 struct fib6_table *table;
2109
2110 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2111 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2112 if (!table)
c71099ac 2113 return;
1da177e4
LT		 2114
2115restart:
c71099ac 2116 read_lock_bh(&table->tb6_lock);
d8d1f30b 2117 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2118 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2119 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2120 dst_hold(&rt->dst);
c71099ac 2121 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2122 ip6_del_rt(rt);
1da177e4
LT		 2123 goto restart;
2124 }
2125 }
c71099ac 2126 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2127}
2128
5578689a
DL		 2129static void rtmsg_to_fib6_config(struct net *net,
2130 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2131 struct fib6_config *cfg)
2132{
2133 memset(cfg, 0, sizeof(*cfg));
2134
2135 cfg->fc_table = RT6_TABLE_MAIN;
2136 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2137 cfg->fc_metric = rtmsg->rtmsg_metric;
2138 cfg->fc_expires = rtmsg->rtmsg_info;
2139 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2140 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2141 cfg->fc_flags = rtmsg->rtmsg_flags;
2142
5578689a 2143 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2144
4e3fd7a0
AD		 2145 cfg->fc_dst = rtmsg->rtmsg_dst;
2146 cfg->fc_src = rtmsg->rtmsg_src;
2147 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2148}
2149
5578689a 2150int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2151{
86872cb5 2152 struct fib6_config cfg;
1da177e4
LT		 2153 struct in6_rtmsg rtmsg;
2154 int err;
2155
67ba4152 2156 switch (cmd) {
1da177e4
LT		 2157 case SIOCADDRT: /* Add a route */
2158 case SIOCDELRT: /* Delete a route */
af31f412 2159 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2160 return -EPERM;
2161 err = copy_from_user(&rtmsg, arg,
2162 sizeof(struct in6_rtmsg));
2163 if (err)
2164 return -EFAULT;
86872cb5 2165
5578689a 2166 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2167
1da177e4
LT		 2168 rtnl_lock();
2169 switch (cmd) {
2170 case SIOCADDRT:
86872cb5 2171 err = ip6_route_add(&cfg);
1da177e4
LT		 2172 break;
2173 case SIOCDELRT:
86872cb5 2174 err = ip6_route_del(&cfg);
1da177e4
LT		 2175 break;
2176 default:
2177 err = -EINVAL;
2178 }
2179 rtnl_unlock();
2180
2181 return err;
3ff50b79 2182 }
1da177e4
LT		 2183
2184 return -EINVAL;
2185}
2186
2187/*
2188 * Drop the packet on the floor
2189 */
2190
d5fdd6ba 2191static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2192{
612f09e8 2193 int type;
adf30907 2194 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2195 switch (ipstats_mib_noroutes) {
2196 case IPSTATS_MIB_INNOROUTES:
0660e03f 2197 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2198 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2199 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2200 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2201 break;
2202 }
2203 /* FALLTHROUGH */
2204 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2205 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2206 ipstats_mib_noroutes);
612f09e8
YH		 2207 break;
2208 }
3ffe533c 2209 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2210 kfree_skb(skb);
2211 return 0;
2212}
2213
9ce8ade0
TG		 2214static int ip6_pkt_discard(struct sk_buff *skb)
2215{
612f09e8 2216 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2217}
2218
aad88724 2219static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb)
1da177e4 2220{
adf30907 2221 skb->dev = skb_dst(skb)->dev;
612f09e8 2222 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2223}
2224
9ce8ade0
TG		 2225static int ip6_pkt_prohibit(struct sk_buff *skb)
2226{
612f09e8 2227 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2228}
2229
aad88724 2230static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb)
9ce8ade0 2231{
adf30907 2232 skb->dev = skb_dst(skb)->dev;
612f09e8 2233 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2234}
2235
1da177e4
LT		 2236/*
2237 * Allocate a dst for local (unicast / anycast) address.
2238 */
2239
2240struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2241 const struct in6_addr *addr,
8f031519 2242 bool anycast)
1da177e4 2243{
c346dca1 2244 struct net *net = dev_net(idev->dev);
a3300ef4
HFS		 2245 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2246 DST_NOCOUNT, NULL);
2247 if (!rt)
1da177e4
LT		 2248 return ERR_PTR(-ENOMEM);
2249
1da177e4
LT		 2250 in6_dev_hold(idev);
2251
11d53b49 2252 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2253 rt->dst.input = ip6_input;
2254 rt->dst.output = ip6_output;
1da177e4 2255 rt->rt6i_idev = idev;
1da177e4
LT		 2256
2257 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2258 if (anycast)
2259 rt->rt6i_flags |= RTF_ANYCAST;
2260 else
1da177e4 2261 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2262
550bab42 2263 rt->rt6i_gateway = *addr;
4e3fd7a0 2264 rt->rt6i_dst.addr = *addr;
1da177e4 2265 rt->rt6i_dst.plen = 128;
5578689a 2266 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4 2267
d8d1f30b 2268 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2269
2270 return rt;
2271}
2272
c3968a85
DW		 2273int ip6_route_get_saddr(struct net *net,
2274 struct rt6_info *rt,
b71d1d42 2275 const struct in6_addr *daddr,
c3968a85
DW		 2276 unsigned int prefs,
2277 struct in6_addr *saddr)
2278{
e16e888b
MS		 2279 struct inet6_dev *idev =
2280 rt ? ip6_dst_idev((struct dst_entry *)rt) : NULL;
c3968a85 2281 int err = 0;
e16e888b 2282 if (rt && rt->rt6i_prefsrc.plen)
4e3fd7a0 2283 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2284 else
2285 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2286 daddr, prefs, saddr);
2287 return err;
2288}
2289
2290/* remove deleted ip from prefsrc entries */
2291struct arg_dev_net_ip {
2292 struct net_device *dev;
2293 struct net *net;
2294 struct in6_addr *addr;
2295};
2296
2297static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2298{
2299 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2300 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2301 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2302
d1918542 2303 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2304 rt != net->ipv6.ip6_null_entry &&
2305 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2306 /* remove prefsrc entry */
2307 rt->rt6i_prefsrc.plen = 0;
2308 }
2309 return 0;
2310}
2311
2312void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2313{
2314 struct net *net = dev_net(ifp->idev->dev);
2315 struct arg_dev_net_ip adni = {
2316 .dev = ifp->idev->dev,
2317 .net = net,
2318 .addr = &ifp->addr,
2319 };
0c3584d5 2320 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2321}
2322
be7a010d
DJ		 2323#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2324#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2325
2326/* Remove routers and update dst entries when gateway turn into host. */
2327static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2328{
2329 struct in6_addr *gateway = (struct in6_addr *)arg;
2330
2331 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2332 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2333 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2334 return -1;
2335 }
2336 return 0;
2337}
2338
2339void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2340{
2341 fib6_clean_all(net, fib6_clean_tohost, gateway);
2342}
2343
8ed67789
DL		 2344struct arg_dev_net {
2345 struct net_device *dev;
2346 struct net *net;
2347};
2348
1da177e4
LT		 2349static int fib6_ifdown(struct rt6_info *rt, void *arg)
2350{
bc3ef660 2351 const struct arg_dev_net *adn = arg;
2352 const struct net_device *dev = adn->dev;
8ed67789 2353
d1918542 2354 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2355 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2356 return -1;
c159d30c 2357
1da177e4
LT		 2358 return 0;
2359}
2360
f3db4851 2361void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2362{
8ed67789
DL		 2363 struct arg_dev_net adn = {
2364 .dev = dev,
2365 .net = net,
2366 };
2367
0c3584d5 2368 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2369 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 2370}
2371
95c96174 2372struct rt6_mtu_change_arg {
1da177e4 2373 struct net_device *dev;
95c96174 2374 unsigned int mtu;
1da177e4
LT		 2375};
2376
2377static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2378{
2379 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2380 struct inet6_dev *idev;
2381
2382 /* In IPv6 pmtu discovery is not optional,
2383 so that RTAX_MTU lock cannot disable it.
2384 We still use this lock to block changes
2385 caused by addrconf/ndisc.
2386 */
2387
2388 idev = __in6_dev_get(arg->dev);
38308473 2389 if (!idev)
1da177e4
LT		 2390 return 0;
2391
2392 /* For administrative MTU increase, there is no way to discover
2393 IPv6 PMTU increase, so PMTU increase should be updated here.
2394 Since RFC 1981 doesn't include administrative MTU increase
2395 update PMTU increase is a MUST. (i.e. jumbo frame)
2396 */
2397 /*
2398 If new MTU is less than route PMTU, this new MTU will be the
2399 lowest MTU in the path, update the route PMTU to reflect PMTU
2400 decreases; if new MTU is greater than route PMTU, and the
2401 old MTU is the lowest MTU in the path, update the route PMTU
2402 to reflect the increase. In this case if the other nodes' MTU
2403 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2404 PMTU discouvery.
2405 */
d1918542 2406 if (rt->dst.dev == arg->dev &&
4b32b5ad
MKL		 2407 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2408 if (rt->rt6i_flags & RTF_CACHE) {
2409 /* For RTF_CACHE with rt6i_pmtu == 0
2410 * (i.e. a redirected route),
2411 * the metrics of its rt->dst.from has already
2412 * been updated.
2413 */
2414 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2415 rt->rt6i_pmtu = arg->mtu;
2416 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2417 (dst_mtu(&rt->dst) < arg->mtu &&
2418 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2419 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2420 }
566cfd8f 2421 }
1da177e4
LT		 2422 return 0;
2423}
2424
95c96174 2425void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2426{
c71099ac
TG		 2427 struct rt6_mtu_change_arg arg = {
2428 .dev = dev,
2429 .mtu = mtu,
2430 };
1da177e4 2431
0c3584d5 2432 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2433}
2434
ef7c79ed 2435static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2436 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2437 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2438 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2439 [RTA_PRIORITY] = { .type = NLA_U32 },
2440 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2441 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2442 [RTA_PREF] = { .type = NLA_U8 },
86872cb5
TG		 2443};
2444
2445static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2446 struct fib6_config *cfg)
1da177e4 2447{
86872cb5
TG		 2448 struct rtmsg *rtm;
2449 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2450 unsigned int pref;
86872cb5 2451 int err;
1da177e4 2452
86872cb5
TG		 2453 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2454 if (err < 0)
2455 goto errout;
1da177e4 2456
86872cb5
TG		 2457 err = -EINVAL;
2458 rtm = nlmsg_data(nlh);
2459 memset(cfg, 0, sizeof(*cfg));
2460
2461 cfg->fc_table = rtm->rtm_table;
2462 cfg->fc_dst_len = rtm->rtm_dst_len;
2463 cfg->fc_src_len = rtm->rtm_src_len;
2464 cfg->fc_flags = RTF_UP;
2465 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2466 cfg->fc_type = rtm->rtm_type;
86872cb5 2467
ef2c7d7b
ND		 2468 if (rtm->rtm_type == RTN_UNREACHABLE ||
2469 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2470 rtm->rtm_type == RTN_PROHIBIT ||
2471 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2472 cfg->fc_flags |= RTF_REJECT;
2473
ab79ad14
MŻ		 2474 if (rtm->rtm_type == RTN_LOCAL)
2475 cfg->fc_flags |= RTF_LOCAL;
2476
1f56a01f
MKL		 2477 if (rtm->rtm_flags & RTM_F_CLONED)
2478 cfg->fc_flags |= RTF_CACHE;
2479
15e47304 2480 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2481 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2482 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2483
2484 if (tb[RTA_GATEWAY]) {
67b61f6c 2485 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2486 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2487 }
86872cb5
TG		 2488
2489 if (tb[RTA_DST]) {
2490 int plen = (rtm->rtm_dst_len + 7) >> 3;
2491
2492 if (nla_len(tb[RTA_DST]) < plen)
2493 goto errout;
2494
2495 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2496 }
86872cb5
TG		 2497
2498 if (tb[RTA_SRC]) {
2499 int plen = (rtm->rtm_src_len + 7) >> 3;
2500
2501 if (nla_len(tb[RTA_SRC]) < plen)
2502 goto errout;
2503
2504 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2505 }
86872cb5 2506
c3968a85 2507 if (tb[RTA_PREFSRC])
67b61f6c 2508 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2509
86872cb5
TG		 2510 if (tb[RTA_OIF])
2511 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2512
2513 if (tb[RTA_PRIORITY])
2514 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2515
2516 if (tb[RTA_METRICS]) {
2517 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2518 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2519 }
86872cb5
TG		 2520
2521 if (tb[RTA_TABLE])
2522 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2523
51ebd318
ND		 2524 if (tb[RTA_MULTIPATH]) {
2525 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2526 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2527 }
2528
c78ba6d6
LR		 2529 if (tb[RTA_PREF]) {
2530 pref = nla_get_u8(tb[RTA_PREF]);
2531 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2532 pref != ICMPV6_ROUTER_PREF_HIGH)
2533 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2534 cfg->fc_flags |= RTF_PREF(pref);
2535 }
2536
86872cb5
TG		 2537 err = 0;
2538errout:
2539 return err;
1da177e4
LT		 2540}
2541
51ebd318
ND		 2542static int ip6_route_multipath(struct fib6_config *cfg, int add)
2543{
2544 struct fib6_config r_cfg;
2545 struct rtnexthop *rtnh;
2546 int remaining;
2547 int attrlen;
2548 int err = 0, last_err = 0;
2549
35f1b4e9 2550 remaining = cfg->fc_mp_len;
51ebd318
ND		 2551beginning:
2552 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318
ND		 2553
2554 /* Parse a Multipath Entry */
2555 while (rtnh_ok(rtnh, remaining)) {
2556 memcpy(&r_cfg, cfg, sizeof(*cfg));
2557 if (rtnh->rtnh_ifindex)
2558 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2559
2560 attrlen = rtnh_attrlen(rtnh);
2561 if (attrlen > 0) {
2562 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2563
2564 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2565 if (nla) {
67b61f6c 2566 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 2567 r_cfg.fc_flags |= RTF_GATEWAY;
2568 }
2569 }
2570 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2571 if (err) {
2572 last_err = err;
2573 /* If we are trying to remove a route, do not stop the
2574 * loop when ip6_route_del() fails (because next hop is
2575 * already gone), we should try to remove all next hops.
2576 */
2577 if (add) {
2578 /* If add fails, we should try to delete all
2579 * next hops that have been already added.
2580 */
2581 add = 0;
35f1b4e9 2582 remaining = cfg->fc_mp_len - remaining;
51ebd318
ND		 2583 goto beginning;
2584 }
2585 }
1a72418b 2586 /* Because each route is added like a single route we remove
27596472
MK		 2587 * these flags after the first nexthop: if there is a collision,
2588 * we have already failed to add the first nexthop:
2589 * fib6_add_rt2node() has rejected it; when replacing, old
2590 * nexthops have been replaced by first new, the rest should
2591 * be added to it.
1a72418b 2592 */
27596472
MK		 2593 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
2594 NLM_F_REPLACE);
51ebd318
ND		 2595 rtnh = rtnh_next(rtnh, &remaining);
2596 }
2597
2598 return last_err;
2599}
2600
67ba4152 2601static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2602{
86872cb5
TG		 2603 struct fib6_config cfg;
2604 int err;
1da177e4 2605
86872cb5
TG		 2606 err = rtm_to_fib6_config(skb, nlh, &cfg);
2607 if (err < 0)
2608 return err;
2609
51ebd318
ND		 2610 if (cfg.fc_mp)
2611 return ip6_route_multipath(&cfg, 0);
2612 else
2613 return ip6_route_del(&cfg);
1da177e4
LT		 2614}
2615
67ba4152 2616static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2617{
86872cb5
TG		 2618 struct fib6_config cfg;
2619 int err;
1da177e4 2620
86872cb5
TG		 2621 err = rtm_to_fib6_config(skb, nlh, &cfg);
2622 if (err < 0)
2623 return err;
2624
51ebd318
ND		 2625 if (cfg.fc_mp)
2626 return ip6_route_multipath(&cfg, 1);
2627 else
2628 return ip6_route_add(&cfg);
1da177e4
LT		 2629}
2630
339bf98f
TG		 2631static inline size_t rt6_nlmsg_size(void)
2632{
2633 return NLMSG_ALIGN(sizeof(struct rtmsg))
2634 + nla_total_size(16) /* RTA_SRC */
2635 + nla_total_size(16) /* RTA_DST */
2636 + nla_total_size(16) /* RTA_GATEWAY */
2637 + nla_total_size(16) /* RTA_PREFSRC */
2638 + nla_total_size(4) /* RTA_TABLE */
2639 + nla_total_size(4) /* RTA_IIF */
2640 + nla_total_size(4) /* RTA_OIF */
2641 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2642 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 2643 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6
LR		 2644 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
2645 + nla_total_size(1); /* RTA_PREF */
339bf98f
TG		 2646}
2647
191cd582
BH		 2648static int rt6_fill_node(struct net *net,
2649 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 2650 struct in6_addr *dst, struct in6_addr *src,
15e47304 2651 int iif, int type, u32 portid, u32 seq,
7bc570c8 2652 int prefix, int nowait, unsigned int flags)
1da177e4 2653{
4b32b5ad 2654 u32 metrics[RTAX_MAX];
1da177e4 2655 struct rtmsg *rtm;
2d7202bf 2656 struct nlmsghdr *nlh;
e3703b3d 2657 long expires;
9e762a4a 2658 u32 table;
1da177e4
LT		 2659
2660 if (prefix) { /* user wants prefix routes only */
2661 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2662 /* success since this is not a prefix route */
2663 return 1;
2664 }
2665 }
2666
15e47304 2667 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 2668 if (!nlh)
26932566 2669 return -EMSGSIZE;
2d7202bf
TG		 2670
2671 rtm = nlmsg_data(nlh);
1da177e4
LT		 2672 rtm->rtm_family = AF_INET6;
2673 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2674 rtm->rtm_src_len = rt->rt6i_src.plen;
2675 rtm->rtm_tos = 0;
c71099ac 2676 if (rt->rt6i_table)
9e762a4a 2677 table = rt->rt6i_table->tb6_id;
c71099ac 2678 else
9e762a4a
PM		 2679 table = RT6_TABLE_UNSPEC;
2680 rtm->rtm_table = table;
c78679e8
DM		 2681 if (nla_put_u32(skb, RTA_TABLE, table))
2682 goto nla_put_failure;
ef2c7d7b
ND		 2683 if (rt->rt6i_flags & RTF_REJECT) {
2684 switch (rt->dst.error) {
2685 case -EINVAL:
2686 rtm->rtm_type = RTN_BLACKHOLE;
2687 break;
2688 case -EACCES:
2689 rtm->rtm_type = RTN_PROHIBIT;
2690 break;
b4949ab2
ND		 2691 case -EAGAIN:
2692 rtm->rtm_type = RTN_THROW;
2693 break;
ef2c7d7b
ND		 2694 default:
2695 rtm->rtm_type = RTN_UNREACHABLE;
2696 break;
2697 }
2698 }
38308473 2699 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 2700 rtm->rtm_type = RTN_LOCAL;
d1918542 2701 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2702 rtm->rtm_type = RTN_LOCAL;
2703 else
2704 rtm->rtm_type = RTN_UNICAST;
2705 rtm->rtm_flags = 0;
2706 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2707 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 2708 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 2709 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 2710 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2711 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2712 rtm->rtm_protocol = RTPROT_RA;
2713 else
2714 rtm->rtm_protocol = RTPROT_KERNEL;
2715 }
1da177e4 2716
38308473 2717 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 2718 rtm->rtm_flags |= RTM_F_CLONED;
2719
2720 if (dst) {
930345ea 2721 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 2722 goto nla_put_failure;
1ab1457c 2723 rtm->rtm_dst_len = 128;
1da177e4 2724 } else if (rtm->rtm_dst_len)
930345ea 2725 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 2726 goto nla_put_failure;
1da177e4
LT		 2727#ifdef CONFIG_IPV6_SUBTREES
2728 if (src) {
930345ea 2729 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 2730 goto nla_put_failure;
1ab1457c 2731 rtm->rtm_src_len = 128;
c78679e8 2732 } else if (rtm->rtm_src_len &&
930345ea 2733 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 2734 goto nla_put_failure;
1da177e4 2735#endif
7bc570c8
YH		 2736 if (iif) {
2737#ifdef CONFIG_IPV6_MROUTE
2738 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 2739 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 2740 if (err <= 0) {
2741 if (!nowait) {
2742 if (err == 0)
2743 return 0;
2744 goto nla_put_failure;
2745 } else {
2746 if (err == -EMSGSIZE)
2747 goto nla_put_failure;
2748 }
2749 }
2750 } else
2751#endif
c78679e8
DM		 2752 if (nla_put_u32(skb, RTA_IIF, iif))
2753 goto nla_put_failure;
7bc570c8 2754 } else if (dst) {
1da177e4 2755 struct in6_addr saddr_buf;
c78679e8 2756 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 2757 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 2758 goto nla_put_failure;
1da177e4 2759 }
2d7202bf 2760
c3968a85
DW		 2761 if (rt->rt6i_prefsrc.plen) {
2762 struct in6_addr saddr_buf;
4e3fd7a0 2763 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 2764 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 2765 goto nla_put_failure;
c3968a85
DW		 2766 }
2767
4b32b5ad
MKL		 2768 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
2769 if (rt->rt6i_pmtu)
2770 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
2771 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 2772 goto nla_put_failure;
2773
dd0cbf29 2774 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 2775 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 2776 goto nla_put_failure;
94f826b8 2777 }
2d7202bf 2778
c78679e8
DM		 2779 if (rt->dst.dev &&
2780 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2781 goto nla_put_failure;
2782 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2783 goto nla_put_failure;
8253947e
LW		 2784
2785 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 2786
87a50699 2787 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 2788 goto nla_put_failure;
2d7202bf 2789
c78ba6d6
LR		 2790 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
2791 goto nla_put_failure;
2792
053c095a
JB		 2793 nlmsg_end(skb, nlh);
2794 return 0;
2d7202bf
TG		 2795
2796nla_put_failure:
26932566
PM		 2797 nlmsg_cancel(skb, nlh);
2798 return -EMSGSIZE;
1da177e4
LT		 2799}
2800
1b43af54 2801int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2802{
2803 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2804 int prefix;
2805
2d7202bf
TG		 2806 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2807 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2808 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2809 } else
2810 prefix = 0;
2811
191cd582
BH		 2812 return rt6_fill_node(arg->net,
2813 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 2814 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2815 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2816}
2817
67ba4152 2818static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 2819{
3b1e0a65 2820 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2821 struct nlattr *tb[RTA_MAX+1];
2822 struct rt6_info *rt;
1da177e4 2823 struct sk_buff *skb;
ab364a6f 2824 struct rtmsg *rtm;
4c9483b2 2825 struct flowi6 fl6;
72331bc0 2826 int err, iif = 0, oif = 0;
1da177e4 2827
ab364a6f
TG		 2828 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2829 if (err < 0)
2830 goto errout;
1da177e4 2831
ab364a6f 2832 err = -EINVAL;
4c9483b2 2833 memset(&fl6, 0, sizeof(fl6));
1da177e4 2834
ab364a6f
TG		 2835 if (tb[RTA_SRC]) {
2836 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2837 goto errout;
2838
4e3fd7a0 2839 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 2840 }
2841
2842 if (tb[RTA_DST]) {
2843 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2844 goto errout;
2845
4e3fd7a0 2846 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 2847 }
2848
2849 if (tb[RTA_IIF])
2850 iif = nla_get_u32(tb[RTA_IIF]);
2851
2852 if (tb[RTA_OIF])
72331bc0 2853 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 2854
2e47b291
LC		 2855 if (tb[RTA_MARK])
2856 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
2857
1da177e4
LT		 2858 if (iif) {
2859 struct net_device *dev;
72331bc0
SL		 2860 int flags = 0;
2861
5578689a 2862 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2863 if (!dev) {
2864 err = -ENODEV;
ab364a6f 2865 goto errout;
1da177e4 2866 }
72331bc0
SL		 2867
2868 fl6.flowi6_iif = iif;
2869
2870 if (!ipv6_addr_any(&fl6.saddr))
2871 flags |= RT6_LOOKUP_F_HAS_SADDR;
2872
2873 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2874 flags);
2875 } else {
2876 fl6.flowi6_oif = oif;
2877
2878 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 2879 }
2880
ab364a6f 2881 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 2882 if (!skb) {
94e187c0 2883 ip6_rt_put(rt);
ab364a6f
TG		 2884 err = -ENOBUFS;
2885 goto errout;
2886 }
1da177e4 2887
ab364a6f
TG		 2888 /* Reserve room for dummy headers, this skb can pass
2889 through good chunk of routing engine.
2890 */
459a98ed 2891 skb_reset_mac_header(skb);
ab364a6f 2892 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2893
d8d1f30b 2894 skb_dst_set(skb, &rt->dst);
1da177e4 2895
4c9483b2 2896 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 2897 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 2898 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2899 if (err < 0) {
ab364a6f
TG		 2900 kfree_skb(skb);
2901 goto errout;
1da177e4
LT		 2902 }
2903
15e47304 2904 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 2905errout:
1da177e4 2906 return err;
1da177e4
LT		 2907}
2908
86872cb5 2909void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2910{
2911 struct sk_buff *skb;
5578689a 2912 struct net *net = info->nl_net;
528c4ceb
DL		 2913 u32 seq;
2914 int err;
2915
2916 err = -ENOBUFS;
38308473 2917 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 2918
339bf98f 2919 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
38308473 2920 if (!skb)
21713ebc
TG		 2921 goto errout;
2922
191cd582 2923 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
15e47304 2924 event, info->portid, seq, 0, 0, 0);
26932566
PM		 2925 if (err < 0) {
2926 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2927 WARN_ON(err == -EMSGSIZE);
2928 kfree_skb(skb);
2929 goto errout;
2930 }
15e47304 2931 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 2932 info->nlh, gfp_any());
2933 return;
21713ebc
TG		 2934errout:
2935 if (err < 0)
5578689a 2936 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2937}
2938
8ed67789 2939static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 2940 unsigned long event, void *ptr)
8ed67789 2941{
351638e7 2942 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 2943 struct net *net = dev_net(dev);
8ed67789
DL		 2944
2945 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 2946 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 2947 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2948#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 2949 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 2950 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 2951 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 2952 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2953#endif
2954 }
2955
2956 return NOTIFY_OK;
2957}
2958
1da177e4
LT		 2959/*
2960 * /proc
2961 */
2962
2963#ifdef CONFIG_PROC_FS
2964
33120b30
AD		 2965static const struct file_operations ipv6_route_proc_fops = {
2966 .owner = THIS_MODULE,
2967 .open = ipv6_route_open,
2968 .read = seq_read,
2969 .llseek = seq_lseek,
8d2ca1d7 2970 .release = seq_release_net,
33120b30
AD		 2971};
2972
1da177e4
LT		 2973static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2974{
69ddb805 2975 struct net *net = (struct net *)seq->private;
1da177e4 2976 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2977 net->ipv6.rt6_stats->fib_nodes,
2978 net->ipv6.rt6_stats->fib_route_nodes,
2979 net->ipv6.rt6_stats->fib_rt_alloc,
2980 net->ipv6.rt6_stats->fib_rt_entries,
2981 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 2982 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 2983 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2984
2985 return 0;
2986}
2987
2988static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2989{
de05c557 2990 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2991}
2992
9a32144e 2993static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2994 .owner = THIS_MODULE,
2995 .open = rt6_stats_seq_open,
2996 .read = seq_read,
2997 .llseek = seq_lseek,
b6fcbdb4 2998 .release = single_release_net,
1da177e4
LT		 2999};
3000#endif /* CONFIG_PROC_FS */
3001
3002#ifdef CONFIG_SYSCTL
3003
1da177e4 3004static
fe2c6338 3005int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3006 void __user *buffer, size_t *lenp, loff_t *ppos)
3007{
c486da34
LAG		 3008 struct net *net;
3009 int delay;
3010 if (!write)
1da177e4 3011 return -EINVAL;
c486da34
LAG		 3012
3013 net = (struct net *)ctl->extra1;
3014 delay = net->ipv6.sysctl.flush_delay;
3015 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3016 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3017 return 0;
1da177e4
LT		 3018}
3019
fe2c6338 3020struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3021 {
1da177e4 3022 .procname = "flush",
4990509f 3023 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3024 .maxlen = sizeof(int),
89c8b3a1 3025 .mode = 0200,
6d9f239a 3026 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3027 },
3028 {
1da177e4 3029 .procname = "gc_thresh",
9a7ec3a9 3030 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3031 .maxlen = sizeof(int),
3032 .mode = 0644,
6d9f239a 3033 .proc_handler = proc_dointvec,
1da177e4
LT		 3034 },
3035 {
1da177e4 3036 .procname = "max_size",
4990509f 3037 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3038 .maxlen = sizeof(int),
3039 .mode = 0644,
6d9f239a 3040 .proc_handler = proc_dointvec,
1da177e4
LT		 3041 },
3042 {
1da177e4 3043 .procname = "gc_min_interval",
4990509f 3044 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3045 .maxlen = sizeof(int),
3046 .mode = 0644,
6d9f239a 3047 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3048 },
3049 {
1da177e4 3050 .procname = "gc_timeout",
4990509f 3051 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3052 .maxlen = sizeof(int),
3053 .mode = 0644,
6d9f239a 3054 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3055 },
3056 {
1da177e4 3057 .procname = "gc_interval",
4990509f 3058 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3059 .maxlen = sizeof(int),
3060 .mode = 0644,
6d9f239a 3061 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3062 },
3063 {
1da177e4 3064 .procname = "gc_elasticity",
4990509f 3065 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3066 .maxlen = sizeof(int),
3067 .mode = 0644,
f3d3f616 3068 .proc_handler = proc_dointvec,
1da177e4
LT		 3069 },
3070 {
1da177e4 3071 .procname = "mtu_expires",
4990509f 3072 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3073 .maxlen = sizeof(int),
3074 .mode = 0644,
6d9f239a 3075 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3076 },
3077 {
1da177e4 3078 .procname = "min_adv_mss",
4990509f 3079 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3080 .maxlen = sizeof(int),
3081 .mode = 0644,
f3d3f616 3082 .proc_handler = proc_dointvec,
1da177e4
LT		 3083 },
3084 {
1da177e4 3085 .procname = "gc_min_interval_ms",
4990509f 3086 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3087 .maxlen = sizeof(int),
3088 .mode = 0644,
6d9f239a 3089 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3090 },
f8572d8f 3091 { }
1da177e4
LT		 3092};
3093
2c8c1e72 3094struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3095{
3096 struct ctl_table *table;
3097
3098 table = kmemdup(ipv6_route_table_template,
3099 sizeof(ipv6_route_table_template),
3100 GFP_KERNEL);
5ee09105
YH		 3101
3102 if (table) {
3103 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3104 table[0].extra1 = net;
86393e52 3105 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3106 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3107 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3108 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3109 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3110 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3111 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3112 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3113 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3114
3115 /* Don't export sysctls to unprivileged users */
3116 if (net->user_ns != &init_user_ns)
3117 table[0].procname = NULL;
5ee09105
YH		 3118 }
3119
760f2d01
DL		 3120 return table;
3121}
1da177e4
LT		 3122#endif
3123
2c8c1e72 3124static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3125{
633d424b 3126 int ret = -ENOMEM;
8ed67789 3127
86393e52
AD		 3128 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3129 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3130
fc66f95c
ED		 3131 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3132 goto out_ip6_dst_ops;
3133
8ed67789
DL		 3134 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3135 sizeof(*net->ipv6.ip6_null_entry),
3136 GFP_KERNEL);
3137 if (!net->ipv6.ip6_null_entry)
fc66f95c 3138 goto out_ip6_dst_entries;
d8d1f30b 3139 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3140 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3141 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3142 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3143 ip6_template_metrics, true);
8ed67789
DL		 3144
3145#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3146 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3147 sizeof(*net->ipv6.ip6_prohibit_entry),
3148 GFP_KERNEL);
68fffc67
PZ		 3149 if (!net->ipv6.ip6_prohibit_entry)
3150 goto out_ip6_null_entry;
d8d1f30b 3151 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3152 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3153 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3154 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3155 ip6_template_metrics, true);
8ed67789
DL		 3156
3157 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3158 sizeof(*net->ipv6.ip6_blk_hole_entry),
3159 GFP_KERNEL);
68fffc67
PZ		 3160 if (!net->ipv6.ip6_blk_hole_entry)
3161 goto out_ip6_prohibit_entry;
d8d1f30b 3162 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3163 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3164 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3165 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3166 ip6_template_metrics, true);
8ed67789
DL		 3167#endif
3168
b339a47c
PZ		 3169 net->ipv6.sysctl.flush_delay = 0;
3170 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3171 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3172 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3173 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3174 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3175 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3176 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3177
6891a346
BT		 3178 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3179
8ed67789
DL		 3180 ret = 0;
3181out:
3182 return ret;
f2fc6a54 3183
68fffc67
PZ		 3184#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3185out_ip6_prohibit_entry:
3186 kfree(net->ipv6.ip6_prohibit_entry);
3187out_ip6_null_entry:
3188 kfree(net->ipv6.ip6_null_entry);
3189#endif
fc66f95c
ED		 3190out_ip6_dst_entries:
3191 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3192out_ip6_dst_ops:
f2fc6a54 3193 goto out;
cdb18761
DL		 3194}
3195
2c8c1e72 3196static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3197{
8ed67789
DL		 3198 kfree(net->ipv6.ip6_null_entry);
3199#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3200 kfree(net->ipv6.ip6_prohibit_entry);
3201 kfree(net->ipv6.ip6_blk_hole_entry);
3202#endif
41bb78b4 3203 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3204}
3205
d189634e
TG		 3206static int __net_init ip6_route_net_init_late(struct net *net)
3207{
3208#ifdef CONFIG_PROC_FS
d4beaa66
G		 3209 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3210 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3211#endif
3212 return 0;
3213}
3214
3215static void __net_exit ip6_route_net_exit_late(struct net *net)
3216{
3217#ifdef CONFIG_PROC_FS
ece31ffd
G		 3218 remove_proc_entry("ipv6_route", net->proc_net);
3219 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3220#endif
3221}
3222
cdb18761
DL		 3223static struct pernet_operations ip6_route_net_ops = {
3224 .init = ip6_route_net_init,
3225 .exit = ip6_route_net_exit,
3226};
3227
c3426b47
DM		 3228static int __net_init ipv6_inetpeer_init(struct net *net)
3229{
3230 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3231
3232 if (!bp)
3233 return -ENOMEM;
3234 inet_peer_base_init(bp);
3235 net->ipv6.peers = bp;
3236 return 0;
3237}
3238
3239static void __net_exit ipv6_inetpeer_exit(struct net *net)
3240{
3241 struct inet_peer_base *bp = net->ipv6.peers;
3242
3243 net->ipv6.peers = NULL;
56a6b248 3244 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3245 kfree(bp);
3246}
3247
2b823f72 3248static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3249 .init = ipv6_inetpeer_init,
3250 .exit = ipv6_inetpeer_exit,
3251};
3252
d189634e
TG		 3253static struct pernet_operations ip6_route_net_late_ops = {
3254 .init = ip6_route_net_init_late,
3255 .exit = ip6_route_net_exit_late,
3256};
3257
8ed67789
DL		 3258static struct notifier_block ip6_route_dev_notifier = {
3259 .notifier_call = ip6_route_dev_notify,
3260 .priority = 0,
3261};
3262
433d49c3 3263int __init ip6_route_init(void)
1da177e4 3264{
433d49c3
DL		 3265 int ret;
3266
9a7ec3a9
DL		 3267 ret = -ENOMEM;
3268 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3269 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3270 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3271 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3272 goto out;
14e50e57 3273
fc66f95c 3274 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3275 if (ret)
bdb3289f 3276 goto out_kmem_cache;
bdb3289f 3277
c3426b47
DM		 3278 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3279 if (ret)
e8803b6c 3280 goto out_dst_entries;
2a0c451a 3281
7e52b33b
DM		 3282 ret = register_pernet_subsys(&ip6_route_net_ops);
3283 if (ret)
3284 goto out_register_inetpeer;
c3426b47 3285
5dc121e9
AE		 3286 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3287
8ed67789
DL		 3288 /* Registering of the loopback is done before this portion of code,
3289 * the loopback reference in rt6_info will not be taken, do it
3290 * manually for init_net */
d8d1f30b 3291 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3292 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3293 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3294 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3295 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3296 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3297 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3298 #endif
e8803b6c 3299 ret = fib6_init();
433d49c3 3300 if (ret)
8ed67789 3301 goto out_register_subsys;
433d49c3 3302
433d49c3
DL		 3303 ret = xfrm6_init();
3304 if (ret)
e8803b6c 3305 goto out_fib6_init;
c35b7e72 3306
433d49c3
DL		 3307 ret = fib6_rules_init();
3308 if (ret)
3309 goto xfrm6_init;
7e5449c2 3310
d189634e
TG		 3311 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3312 if (ret)
3313 goto fib6_rules_init;
3314
433d49c3 3315 ret = -ENOBUFS;
c7ac8679
GR		 3316 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3317 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3318 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3319 goto out_register_late_subsys;
c127ea2c 3320
8ed67789 3321 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3322 if (ret)
d189634e 3323 goto out_register_late_subsys;
8ed67789 3324
433d49c3
DL		 3325out:
3326 return ret;
3327
d189634e
TG		 3328out_register_late_subsys:
3329 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3330fib6_rules_init:
433d49c3
DL		 3331 fib6_rules_cleanup();
3332xfrm6_init:
433d49c3 3333 xfrm6_fini();
2a0c451a
TG		 3334out_fib6_init:
3335 fib6_gc_cleanup();
8ed67789
DL		 3336out_register_subsys:
3337 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3338out_register_inetpeer:
3339 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3340out_dst_entries:
3341 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3342out_kmem_cache:
f2fc6a54 3343 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3344 goto out;
1da177e4
LT		 3345}
3346
3347void ip6_route_cleanup(void)
3348{
8ed67789 3349 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3350 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3351 fib6_rules_cleanup();
1da177e4 3352 xfrm6_fini();
1da177e4 3353 fib6_gc_cleanup();
c3426b47 3354 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3355 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3356 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3357 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3358}
Michael's Linux tree.