projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
ipv6: Check expire on DST_NOCACHE route
[linux-2.6-block.git] / net / ipv6 / route.c
CommitLineData
1da177e4
LT		 1/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4 7 *
1da177e4
LT		 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13
14/* Changes:
15 *
16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection.
18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list.
c0bece9f
YH		 23 * Ville Nuorvala
24 * Fixed routing subtrees.
1da177e4
LT		 25 */
26
f3213831
JP		 27#define pr_fmt(fmt) "IPv6: " fmt
28
4fc268d2 29#include <linux/capability.h>
1da177e4 30#include <linux/errno.h>
bc3b2d7f 31#include <linux/export.h>
1da177e4
LT		 32#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
7bc570c8 40#include <linux/mroute6.h>
1da177e4 41#include <linux/init.h>
1da177e4 42#include <linux/if_arp.h>
1da177e4
LT		 43#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
5b7c931d 45#include <linux/nsproxy.h>
5a0e3ad6 46#include <linux/slab.h>
457c4cbc 47#include <net/net_namespace.h>
1da177e4
LT		 48#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
904af04d 57#include <net/dst_metadata.h>
1da177e4 58#include <net/xfrm.h>
8d71740c 59#include <net/netevent.h>
21713ebc 60#include <net/netlink.h>
51ebd318 61#include <net/nexthop.h>
19e42e45 62#include <net/lwtunnel.h>
904af04d 63#include <net/ip_tunnels.h>
ca254490 64#include <net/l3mdev.h>
1da177e4
LT		 65
66#include <asm/uaccess.h>
67
68#ifdef CONFIG_SYSCTL
69#include <linux/sysctl.h>
70#endif
71
afc154e9 72enum rt6_nud_state {
7e980569
JB		 73 RT6_NUD_FAIL_HARD = -3,
74 RT6_NUD_FAIL_PROBE = -2,
75 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 76 RT6_NUD_SUCCEED = 1
77};
78
83a09abd 79static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort);
1da177e4 80static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 81static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 82static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 83static struct dst_entry *ip6_negative_advice(struct dst_entry *);
84static void ip6_dst_destroy(struct dst_entry *);
85static void ip6_dst_ifdown(struct dst_entry *,
86 struct net_device *dev, int how);
569d3645 87static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 88
89static int ip6_pkt_discard(struct sk_buff *skb);
ede2059d 90static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb);
7150aede 91static int ip6_pkt_prohibit(struct sk_buff *skb);
ede2059d 92static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1da177e4 93static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 94static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
95 struct sk_buff *skb, u32 mtu);
96static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
97 struct sk_buff *skb);
4b32b5ad 98static void rt6_dst_from_metrics_check(struct rt6_info *rt);
52bd4c0c 99static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 100
70ceb4f5 101#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 102static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 103 const struct in6_addr *prefix, int prefixlen,
104 const struct in6_addr *gwaddr, int ifindex,
95c96174 105 unsigned int pref);
efa2cea0 106static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 107 const struct in6_addr *prefix, int prefixlen,
108 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 109#endif
110
8d0b94af
MKL		 111struct uncached_list {
112 spinlock_t lock;
113 struct list_head head;
114};
115
116static DEFINE_PER_CPU_ALIGNED(struct uncached_list, rt6_uncached_list);
117
118static void rt6_uncached_list_add(struct rt6_info *rt)
119{
120 struct uncached_list *ul = raw_cpu_ptr(&rt6_uncached_list);
121
122 rt->dst.flags |= DST_NOCACHE;
123 rt->rt6i_uncached_list = ul;
124
125 spin_lock_bh(&ul->lock);
126 list_add_tail(&rt->rt6i_uncached, &ul->head);
127 spin_unlock_bh(&ul->lock);
128}
129
130static void rt6_uncached_list_del(struct rt6_info *rt)
131{
132 if (!list_empty(&rt->rt6i_uncached)) {
133 struct uncached_list *ul = rt->rt6i_uncached_list;
134
135 spin_lock_bh(&ul->lock);
136 list_del(&rt->rt6i_uncached);
137 spin_unlock_bh(&ul->lock);
138 }
139}
140
141static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
142{
143 struct net_device *loopback_dev = net->loopback_dev;
144 int cpu;
145
e332bc67
EB		 146 if (dev == loopback_dev)
147 return;
148
8d0b94af
MKL		 149 for_each_possible_cpu(cpu) {
150 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
151 struct rt6_info *rt;
152
153 spin_lock_bh(&ul->lock);
154 list_for_each_entry(rt, &ul->head, rt6i_uncached) {
155 struct inet6_dev *rt_idev = rt->rt6i_idev;
156 struct net_device *rt_dev = rt->dst.dev;
157
e332bc67 158 if (rt_idev->dev == dev) {
8d0b94af
MKL		 159 rt->rt6i_idev = in6_dev_get(loopback_dev);
160 in6_dev_put(rt_idev);
161 }
162
e332bc67 163 if (rt_dev == dev) {
8d0b94af
MKL		 164 rt->dst.dev = loopback_dev;
165 dev_hold(rt->dst.dev);
166 dev_put(rt_dev);
167 }
168 }
169 spin_unlock_bh(&ul->lock);
170 }
171}
172
d52d3997
MKL		 173static u32 *rt6_pcpu_cow_metrics(struct rt6_info *rt)
174{
175 return dst_metrics_write_ptr(rt->dst.from);
176}
177
06582540
DM		 178static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
179{
4b32b5ad 180 struct rt6_info *rt = (struct rt6_info *)dst;
06582540 181
d52d3997
MKL		 182 if (rt->rt6i_flags & RTF_PCPU)
183 return rt6_pcpu_cow_metrics(rt);
184 else if (rt->rt6i_flags & RTF_CACHE)
4b32b5ad
MKL		 185 return NULL;
186 else
3b471175 187 return dst_cow_metrics_generic(dst, old);
06582540
DM		 188}
189
f894cbf8
DM		 190static inline const void *choose_neigh_daddr(struct rt6_info *rt,
191 struct sk_buff *skb,
192 const void *daddr)
39232973
DM		 193{
194 struct in6_addr *p = &rt->rt6i_gateway;
195
a7563f34 196 if (!ipv6_addr_any(p))
39232973 197 return (const void *) p;
f894cbf8
DM		 198 else if (skb)
199 return &ipv6_hdr(skb)->daddr;
39232973
DM		 200 return daddr;
201}
202
f894cbf8
DM		 203static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
204 struct sk_buff *skb,
205 const void *daddr)
d3aaeb38 206{
39232973
DM		 207 struct rt6_info *rt = (struct rt6_info *) dst;
208 struct neighbour *n;
209
f894cbf8 210 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 211 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 212 if (n)
213 return n;
214 return neigh_create(&nd_tbl, daddr, dst->dev);
215}
216
9a7ec3a9 217static struct dst_ops ip6_dst_ops_template = {
1da177e4 218 .family = AF_INET6,
1da177e4
LT		 219 .gc = ip6_dst_gc,
220 .gc_thresh = 1024,
221 .check = ip6_dst_check,
0dbaee3b 222 .default_advmss = ip6_default_advmss,
ebb762f2 223 .mtu = ip6_mtu,
06582540 224 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 225 .destroy = ip6_dst_destroy,
226 .ifdown = ip6_dst_ifdown,
227 .negative_advice = ip6_negative_advice,
228 .link_failure = ip6_link_failure,
229 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 230 .redirect = rt6_do_redirect,
9f8955cc 231 .local_out = __ip6_local_out,
d3aaeb38 232 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 233};
234
ebb762f2 235static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 236{
618f9bc7
SK		 237 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
238
239 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 240}
241
6700c270
DM		 242static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
243 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 244{
245}
246
6700c270
DM		 247static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
248 struct sk_buff *skb)
b587ee3b
DM		 249{
250}
251
14e50e57
DM		 252static struct dst_ops ip6_dst_blackhole_ops = {
253 .family = AF_INET6,
14e50e57
DM		 254 .destroy = ip6_dst_destroy,
255 .check = ip6_dst_check,
ebb762f2 256 .mtu = ip6_blackhole_mtu,
214f45c9 257 .default_advmss = ip6_default_advmss,
14e50e57 258 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 259 .redirect = ip6_rt_blackhole_redirect,
0a1f5962 260 .cow_metrics = dst_cow_metrics_generic,
d3aaeb38 261 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 262};
263
62fa8a84 264static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 265 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 266};
267
fb0af4c7 268static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 269 .dst = {
270 .__refcnt = ATOMIC_INIT(1),
271 .__use = 1,
2c20cbd7 272 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 273 .error = -ENETUNREACH,
d8d1f30b
CG		 274 .input = ip6_pkt_discard,
275 .output = ip6_pkt_discard_out,
1da177e4
LT		 276 },
277 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 278 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 279 .rt6i_metric = ~(u32) 0,
280 .rt6i_ref = ATOMIC_INIT(1),
281};
282
101367c2
TG		 283#ifdef CONFIG_IPV6_MULTIPLE_TABLES
284
fb0af4c7 285static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 286 .dst = {
287 .__refcnt = ATOMIC_INIT(1),
288 .__use = 1,
2c20cbd7 289 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 290 .error = -EACCES,
d8d1f30b
CG		 291 .input = ip6_pkt_prohibit,
292 .output = ip6_pkt_prohibit_out,
101367c2
TG		 293 },
294 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 295 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 296 .rt6i_metric = ~(u32) 0,
297 .rt6i_ref = ATOMIC_INIT(1),
298};
299
fb0af4c7 300static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 301 .dst = {
302 .__refcnt = ATOMIC_INIT(1),
303 .__use = 1,
2c20cbd7 304 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 305 .error = -EINVAL,
d8d1f30b 306 .input = dst_discard,
ede2059d 307 .output = dst_discard_out,
101367c2
TG		 308 },
309 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 310 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 311 .rt6i_metric = ~(u32) 0,
312 .rt6i_ref = ATOMIC_INIT(1),
313};
314
315#endif
316
ebfa45f0
MKL		 317static void rt6_info_init(struct rt6_info *rt)
318{
319 struct dst_entry *dst = &rt->dst;
320
321 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
322 INIT_LIST_HEAD(&rt->rt6i_siblings);
323 INIT_LIST_HEAD(&rt->rt6i_uncached);
324}
325
1da177e4 326/* allocate dst with ip6_dst_ops */
d52d3997
MKL		 327static struct rt6_info *__ip6_dst_alloc(struct net *net,
328 struct net_device *dev,
ad706862 329 int flags)
1da177e4 330{
97bab73f 331 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 332 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 333
ebfa45f0
MKL		 334 if (rt)
335 rt6_info_init(rt);
8104891b 336
cf911662 337 return rt;
1da177e4
LT		 338}
339
d52d3997
MKL		 340static struct rt6_info *ip6_dst_alloc(struct net *net,
341 struct net_device *dev,
ad706862 342 int flags)
d52d3997 343{
ad706862 344 struct rt6_info *rt = __ip6_dst_alloc(net, dev, flags);
d52d3997
MKL		 345
346 if (rt) {
347 rt->rt6i_pcpu = alloc_percpu_gfp(struct rt6_info *, GFP_ATOMIC);
348 if (rt->rt6i_pcpu) {
349 int cpu;
350
351 for_each_possible_cpu(cpu) {
352 struct rt6_info **p;
353
354 p = per_cpu_ptr(rt->rt6i_pcpu, cpu);
355 /* no one shares rt */
356 *p = NULL;
357 }
358 } else {
359 dst_destroy((struct dst_entry *)rt);
360 return NULL;
361 }
362 }
363
364 return rt;
365}
366
1da177e4
LT		 367static void ip6_dst_destroy(struct dst_entry *dst)
368{
369 struct rt6_info *rt = (struct rt6_info *)dst;
ecd98837 370 struct dst_entry *from = dst->from;
8d0b94af 371 struct inet6_dev *idev;
1da177e4 372
4b32b5ad 373 dst_destroy_metrics_generic(dst);
87775312 374 free_percpu(rt->rt6i_pcpu);
8d0b94af
MKL		 375 rt6_uncached_list_del(rt);
376
377 idev = rt->rt6i_idev;
38308473 378 if (idev) {
1da177e4
LT		 379 rt->rt6i_idev = NULL;
380 in6_dev_put(idev);
1ab1457c 381 }
1716a961 382
ecd98837
YH		 383 dst->from = NULL;
384 dst_release(from);
b3419363
DM		 385}
386
1da177e4
LT		 387static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
388 int how)
389{
390 struct rt6_info *rt = (struct rt6_info *)dst;
391 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 392 struct net_device *loopback_dev =
c346dca1 393 dev_net(dev)->loopback_dev;
1da177e4 394
97cac082
DM		 395 if (dev != loopback_dev) {
396 if (idev && idev->dev == dev) {
397 struct inet6_dev *loopback_idev =
398 in6_dev_get(loopback_dev);
399 if (loopback_idev) {
400 rt->rt6i_idev = loopback_idev;
401 in6_dev_put(idev);
402 }
403 }
1da177e4
LT		 404 }
405}
406
5973fb1e
MKL		 407static bool __rt6_check_expired(const struct rt6_info *rt)
408{
409 if (rt->rt6i_flags & RTF_EXPIRES)
410 return time_after(jiffies, rt->dst.expires);
411 else
412 return false;
413}
414
a50feda5 415static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 416{
1716a961
G		 417 if (rt->rt6i_flags & RTF_EXPIRES) {
418 if (time_after(jiffies, rt->dst.expires))
a50feda5 419 return true;
1716a961 420 } else if (rt->dst.from) {
3fd91fb3 421 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 422 }
a50feda5 423 return false;
1da177e4
LT		 424}
425
51ebd318
ND		 426/* Multipath route selection:
427 * Hash based function using packet header and flowlabel.
428 * Adapted from fib_info_hashfn()
429 */
430static int rt6_info_hash_nhsfn(unsigned int candidate_count,
431 const struct flowi6 *fl6)
432{
644d0e65 433 return get_hash_from_flowi6(fl6) % candidate_count;
51ebd318
ND		 434}
435
436static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 437 struct flowi6 *fl6, int oif,
438 int strict)
51ebd318
ND		 439{
440 struct rt6_info *sibling, *next_sibling;
441 int route_choosen;
442
443 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
444 /* Don't change the route, if route_choosen == 0
445 * (siblings does not include ourself)
446 */
447 if (route_choosen)
448 list_for_each_entry_safe(sibling, next_sibling,
449 &match->rt6i_siblings, rt6i_siblings) {
450 route_choosen--;
451 if (route_choosen == 0) {
52bd4c0c
ND		 452 if (rt6_score_route(sibling, oif, strict) < 0)
453 break;
51ebd318
ND		 454 match = sibling;
455 break;
456 }
457 }
458 return match;
459}
460
1da177e4 461/*
c71099ac 462 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 463 */
464
8ed67789
DL		 465static inline struct rt6_info *rt6_device_match(struct net *net,
466 struct rt6_info *rt,
b71d1d42 467 const struct in6_addr *saddr,
1da177e4 468 int oif,
d420895e 469 int flags)
1da177e4
LT		 470{
471 struct rt6_info *local = NULL;
472 struct rt6_info *sprt;
473
dd3abc4e
YH		 474 if (!oif && ipv6_addr_any(saddr))
475 goto out;
476
d8d1f30b 477 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 478 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 479
480 if (oif) {
1da177e4
LT		 481 if (dev->ifindex == oif)
482 return sprt;
483 if (dev->flags & IFF_LOOPBACK) {
38308473 484 if (!sprt->rt6i_idev ||
1da177e4 485 sprt->rt6i_idev->dev->ifindex != oif) {
17fb0b2b 486 if (flags & RT6_LOOKUP_F_IFACE)
1da177e4 487 continue;
17fb0b2b
DA		 488 if (local &&
489 local->rt6i_idev->dev->ifindex == oif)
1da177e4
LT		 490 continue;
491 }
492 local = sprt;
493 }
dd3abc4e
YH		 494 } else {
495 if (ipv6_chk_addr(net, saddr, dev,
496 flags & RT6_LOOKUP_F_IFACE))
497 return sprt;
1da177e4 498 }
dd3abc4e 499 }
1da177e4 500
dd3abc4e 501 if (oif) {
1da177e4
LT		 502 if (local)
503 return local;
504
d420895e 505 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 506 return net->ipv6.ip6_null_entry;
1da177e4 507 }
dd3abc4e 508out:
1da177e4
LT		 509 return rt;
510}
511
27097255 512#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 513struct __rt6_probe_work {
514 struct work_struct work;
515 struct in6_addr target;
516 struct net_device *dev;
517};
518
519static void rt6_probe_deferred(struct work_struct *w)
520{
521 struct in6_addr mcaddr;
522 struct __rt6_probe_work *work =
523 container_of(w, struct __rt6_probe_work, work);
524
525 addrconf_addr_solict_mult(&work->target, &mcaddr);
38cf595b 526 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL, NULL);
c2f17e82 527 dev_put(work->dev);
662f5533 528 kfree(work);
c2f17e82
HFS		 529}
530
27097255
YH		 531static void rt6_probe(struct rt6_info *rt)
532{
990edb42 533 struct __rt6_probe_work *work;
f2c31e32 534 struct neighbour *neigh;
27097255
YH		 535 /*
536 * Okay, this does not seem to be appropriate
537 * for now, however, we need to check if it
538 * is really so; aka Router Reachability Probing.
539 *
540 * Router Reachability Probe MUST be rate-limited
541 * to no more than one per minute.
542 */
2152caea 543 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 544 return;
2152caea
YH		 545 rcu_read_lock_bh();
546 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
547 if (neigh) {
8d6c31bf
MKL		 548 if (neigh->nud_state & NUD_VALID)
549 goto out;
550
990edb42 551 work = NULL;
2152caea 552 write_lock(&neigh->lock);
990edb42
MKL		 553 if (!(neigh->nud_state & NUD_VALID) &&
554 time_after(jiffies,
555 neigh->updated +
556 rt->rt6i_idev->cnf.rtr_probe_interval)) {
557 work = kmalloc(sizeof(*work), GFP_ATOMIC);
558 if (work)
559 __neigh_set_probe_once(neigh);
c2f17e82 560 }
2152caea 561 write_unlock(&neigh->lock);
990edb42
MKL		 562 } else {
563 work = kmalloc(sizeof(*work), GFP_ATOMIC);
f2c31e32 564 }
990edb42
MKL		 565
566 if (work) {
567 INIT_WORK(&work->work, rt6_probe_deferred);
568 work->target = rt->rt6i_gateway;
569 dev_hold(rt->dst.dev);
570 work->dev = rt->dst.dev;
571 schedule_work(&work->work);
572 }
573
8d6c31bf 574out:
2152caea 575 rcu_read_unlock_bh();
27097255
YH		 576}
577#else
578static inline void rt6_probe(struct rt6_info *rt)
579{
27097255
YH		 580}
581#endif
582
1da177e4 583/*
554cfb7e 584 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 585 */
b6f99a21 586static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 587{
d1918542 588 struct net_device *dev = rt->dst.dev;
161980f4 589 if (!oif || dev->ifindex == oif)
554cfb7e 590 return 2;
161980f4
DM		 591 if ((dev->flags & IFF_LOOPBACK) &&
592 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
593 return 1;
594 return 0;
554cfb7e 595}
1da177e4 596
afc154e9 597static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 598{
f2c31e32 599 struct neighbour *neigh;
afc154e9 600 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 601
4d0c5911
YH		 602 if (rt->rt6i_flags & RTF_NONEXTHOP ||
603 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 604 return RT6_NUD_SUCCEED;
145a3621
YH		 605
606 rcu_read_lock_bh();
607 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
608 if (neigh) {
609 read_lock(&neigh->lock);
554cfb7e 610 if (neigh->nud_state & NUD_VALID)
afc154e9 611 ret = RT6_NUD_SUCCEED;
398bcbeb 612#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 613 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 614 ret = RT6_NUD_SUCCEED;
7e980569
JB		 615 else
616 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 617#endif
145a3621 618 read_unlock(&neigh->lock);
afc154e9
HFS		 619 } else {
620 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 621 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 622 }
145a3621
YH		 623 rcu_read_unlock_bh();
624
a5a81f0b 625 return ret;
1da177e4
LT		 626}
627
554cfb7e
YH		 628static int rt6_score_route(struct rt6_info *rt, int oif,
629 int strict)
1da177e4 630{
a5a81f0b 631 int m;
1ab1457c 632
4d0c5911 633 m = rt6_check_dev(rt, oif);
77d16f45 634 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 635 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 636#ifdef CONFIG_IPV6_ROUTER_PREF
637 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
638#endif
afc154e9
HFS		 639 if (strict & RT6_LOOKUP_F_REACHABLE) {
640 int n = rt6_check_neigh(rt);
641 if (n < 0)
642 return n;
643 }
554cfb7e
YH		 644 return m;
645}
646
f11e6659 647static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 648 int *mpri, struct rt6_info *match,
649 bool *do_rr)
554cfb7e 650{
f11e6659 651 int m;
afc154e9 652 bool match_do_rr = false;
35103d11
AG		 653 struct inet6_dev *idev = rt->rt6i_idev;
654 struct net_device *dev = rt->dst.dev;
655
656 if (dev && !netif_carrier_ok(dev) &&
657 idev->cnf.ignore_routes_with_linkdown)
658 goto out;
f11e6659
DM		 659
660 if (rt6_check_expired(rt))
661 goto out;
662
663 m = rt6_score_route(rt, oif, strict);
7e980569 664 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 665 match_do_rr = true;
666 m = 0; /* lowest valid score */
7e980569 667 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 668 goto out;
afc154e9
HFS		 669 }
670
671 if (strict & RT6_LOOKUP_F_REACHABLE)
672 rt6_probe(rt);
f11e6659 673
7e980569 674 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 675 if (m > *mpri) {
afc154e9 676 *do_rr = match_do_rr;
f11e6659
DM		 677 *mpri = m;
678 match = rt;
f11e6659 679 }
f11e6659
DM		 680out:
681 return match;
682}
683
684static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
685 struct rt6_info *rr_head,
afc154e9
HFS		 686 u32 metric, int oif, int strict,
687 bool *do_rr)
f11e6659 688{
9fbdcfaf 689 struct rt6_info *rt, *match, *cont;
554cfb7e 690 int mpri = -1;
1da177e4 691
f11e6659 692 match = NULL;
9fbdcfaf
SK		 693 cont = NULL;
694 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
695 if (rt->rt6i_metric != metric) {
696 cont = rt;
697 break;
698 }
699
700 match = find_match(rt, oif, strict, &mpri, match, do_rr);
701 }
702
703 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
704 if (rt->rt6i_metric != metric) {
705 cont = rt;
706 break;
707 }
708
afc154e9 709 match = find_match(rt, oif, strict, &mpri, match, do_rr);
9fbdcfaf
SK		 710 }
711
712 if (match || !cont)
713 return match;
714
715 for (rt = cont; rt; rt = rt->dst.rt6_next)
afc154e9 716 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 717
f11e6659
DM		 718 return match;
719}
1da177e4 720
f11e6659
DM		 721static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
722{
723 struct rt6_info *match, *rt0;
8ed67789 724 struct net *net;
afc154e9 725 bool do_rr = false;
1da177e4 726
f11e6659
DM		 727 rt0 = fn->rr_ptr;
728 if (!rt0)
729 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 730
afc154e9
HFS		 731 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
732 &do_rr);
1da177e4 733
afc154e9 734 if (do_rr) {
d8d1f30b 735 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 736
554cfb7e 737 /* no entries matched; do round-robin */
f11e6659
DM		 738 if (!next || next->rt6i_metric != rt0->rt6i_metric)
739 next = fn->leaf;
740
741 if (next != rt0)
742 fn->rr_ptr = next;
1da177e4 743 }
1da177e4 744
d1918542 745 net = dev_net(rt0->dst.dev);
a02cec21 746 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 747}
748
8b9df265
MKL		 749static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
750{
751 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
752}
753
70ceb4f5
YH		 754#ifdef CONFIG_IPV6_ROUTE_INFO
755int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 756 const struct in6_addr *gwaddr)
70ceb4f5 757{
c346dca1 758 struct net *net = dev_net(dev);
70ceb4f5
YH		 759 struct route_info *rinfo = (struct route_info *) opt;
760 struct in6_addr prefix_buf, *prefix;
761 unsigned int pref;
4bed72e4 762 unsigned long lifetime;
70ceb4f5
YH		 763 struct rt6_info *rt;
764
765 if (len < sizeof(struct route_info)) {
766 return -EINVAL;
767 }
768
769 /* Sanity check for prefix_len and length */
770 if (rinfo->length > 3) {
771 return -EINVAL;
772 } else if (rinfo->prefix_len > 128) {
773 return -EINVAL;
774 } else if (rinfo->prefix_len > 64) {
775 if (rinfo->length < 2) {
776 return -EINVAL;
777 }
778 } else if (rinfo->prefix_len > 0) {
779 if (rinfo->length < 1) {
780 return -EINVAL;
781 }
782 }
783
784 pref = rinfo->route_pref;
785 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 786 return -EINVAL;
70ceb4f5 787
4bed72e4 788 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 789
790 if (rinfo->length == 3)
791 prefix = (struct in6_addr *)rinfo->prefix;
792 else {
793 /* this function is safe */
794 ipv6_addr_prefix(&prefix_buf,
795 (struct in6_addr *)rinfo->prefix,
796 rinfo->prefix_len);
797 prefix = &prefix_buf;
798 }
799
f104a567
DJ		 800 if (rinfo->prefix_len == 0)
801 rt = rt6_get_dflt_router(gwaddr, dev);
802 else
803 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
804 gwaddr, dev->ifindex);
70ceb4f5
YH		 805
806 if (rt && !lifetime) {
e0a1ad73 807 ip6_del_rt(rt);
70ceb4f5
YH		 808 rt = NULL;
809 }
810
811 if (!rt && lifetime)
efa2cea0 812 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 813 pref);
814 else if (rt)
815 rt->rt6i_flags = RTF_ROUTEINFO |
816 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
817
818 if (rt) {
1716a961
G		 819 if (!addrconf_finite_timeout(lifetime))
820 rt6_clean_expires(rt);
821 else
822 rt6_set_expires(rt, jiffies + HZ * lifetime);
823
94e187c0 824 ip6_rt_put(rt);
70ceb4f5
YH		 825 }
826 return 0;
827}
828#endif
829
a3c00e46
MKL		 830static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
831 struct in6_addr *saddr)
832{
833 struct fib6_node *pn;
834 while (1) {
835 if (fn->fn_flags & RTN_TL_ROOT)
836 return NULL;
837 pn = fn->parent;
838 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
839 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
840 else
841 fn = pn;
842 if (fn->fn_flags & RTN_RTINFO)
843 return fn;
844 }
845}
c71099ac 846
8ed67789
DL		 847static struct rt6_info *ip6_pol_route_lookup(struct net *net,
848 struct fib6_table *table,
4c9483b2 849 struct flowi6 *fl6, int flags)
1da177e4
LT		 850{
851 struct fib6_node *fn;
852 struct rt6_info *rt;
853
c71099ac 854 read_lock_bh(&table->tb6_lock);
4c9483b2 855 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 856restart:
857 rt = fn->leaf;
4c9483b2 858 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 859 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 860 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 861 if (rt == net->ipv6.ip6_null_entry) {
862 fn = fib6_backtrack(fn, &fl6->saddr);
863 if (fn)
864 goto restart;
865 }
d8d1f30b 866 dst_use(&rt->dst, jiffies);
c71099ac 867 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 868 return rt;
869
870}
871
67ba4152 872struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 873 int flags)
874{
875 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
876}
877EXPORT_SYMBOL_GPL(ip6_route_lookup);
878
9acd9f3a
YH		 879struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
880 const struct in6_addr *saddr, int oif, int strict)
c71099ac 881{
4c9483b2
DM		 882 struct flowi6 fl6 = {
883 .flowi6_oif = oif,
884 .daddr = *daddr,
c71099ac
TG		 885 };
886 struct dst_entry *dst;
77d16f45 887 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 888
adaa70bb 889 if (saddr) {
4c9483b2 890 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 891 flags |= RT6_LOOKUP_F_HAS_SADDR;
892 }
893
4c9483b2 894 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 895 if (dst->error == 0)
896 return (struct rt6_info *) dst;
897
898 dst_release(dst);
899
1da177e4
LT		 900 return NULL;
901}
7159039a
YH		 902EXPORT_SYMBOL(rt6_lookup);
903
c71099ac 904/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 905 It takes new route entry, the addition fails by any reason the
906 route is freed. In any case, if caller does not hold it, it may
907 be destroyed.
908 */
909
e5fd387a 910static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 911 struct mx6_config *mxc)
1da177e4
LT		 912{
913 int err;
c71099ac 914 struct fib6_table *table;
1da177e4 915
c71099ac
TG		 916 table = rt->rt6i_table;
917 write_lock_bh(&table->tb6_lock);
e715b6d3 918 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 919 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 920
921 return err;
922}
923
40e22e8f
TG		 924int ip6_ins_rt(struct rt6_info *rt)
925{
e715b6d3
FW		 926 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
927 struct mx6_config mxc = { .mx = NULL, };
928
929 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 930}
931
8b9df265
MKL		 932static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
933 const struct in6_addr *daddr,
934 const struct in6_addr *saddr)
1da177e4 935{
1da177e4
LT		 936 struct rt6_info *rt;
937
938 /*
939 * Clone the route.
940 */
941
d52d3997 942 if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
83a09abd 943 ort = (struct rt6_info *)ort->dst.from;
1da177e4 944
ad706862 945 rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
83a09abd
MKL		 946
947 if (!rt)
948 return NULL;
949
950 ip6_rt_copy_init(rt, ort);
951 rt->rt6i_flags |= RTF_CACHE;
952 rt->rt6i_metric = 0;
953 rt->dst.flags |= DST_HOST;
954 rt->rt6i_dst.addr = *daddr;
955 rt->rt6i_dst.plen = 128;
1da177e4 956
83a09abd
MKL		 957 if (!rt6_is_gw_or_nonexthop(ort)) {
958 if (ort->rt6i_dst.plen != 128 &&
959 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
960 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 961#ifdef CONFIG_IPV6_SUBTREES
83a09abd
MKL		 962 if (rt->rt6i_src.plen && saddr) {
963 rt->rt6i_src.addr = *saddr;
964 rt->rt6i_src.plen = 128;
8b9df265 965 }
83a09abd 966#endif
95a9a5ba 967 }
1da177e4 968
95a9a5ba
YH		 969 return rt;
970}
1da177e4 971
d52d3997
MKL		 972static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
973{
974 struct rt6_info *pcpu_rt;
975
976 pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
ad706862 977 rt->dst.dev, rt->dst.flags);
d52d3997
MKL		 978
979 if (!pcpu_rt)
980 return NULL;
981 ip6_rt_copy_init(pcpu_rt, rt);
982 pcpu_rt->rt6i_protocol = rt->rt6i_protocol;
983 pcpu_rt->rt6i_flags |= RTF_PCPU;
984 return pcpu_rt;
985}
986
987/* It should be called with read_lock_bh(&tb6_lock) acquired */
988static struct rt6_info *rt6_get_pcpu_route(struct rt6_info *rt)
989{
a73e4195 990 struct rt6_info *pcpu_rt, **p;
d52d3997
MKL		 991
992 p = this_cpu_ptr(rt->rt6i_pcpu);
993 pcpu_rt = *p;
994
a73e4195
MKL		 995 if (pcpu_rt) {
996 dst_hold(&pcpu_rt->dst);
997 rt6_dst_from_metrics_check(pcpu_rt);
998 }
999 return pcpu_rt;
1000}
1001
1002static struct rt6_info *rt6_make_pcpu_route(struct rt6_info *rt)
1003{
9c7370a1 1004 struct fib6_table *table = rt->rt6i_table;
a73e4195 1005 struct rt6_info *pcpu_rt, *prev, **p;
d52d3997
MKL		 1006
1007 pcpu_rt = ip6_rt_pcpu_alloc(rt);
1008 if (!pcpu_rt) {
1009 struct net *net = dev_net(rt->dst.dev);
1010
9c7370a1
MKL		 1011 dst_hold(&net->ipv6.ip6_null_entry->dst);
1012 return net->ipv6.ip6_null_entry;
d52d3997
MKL		 1013 }
1014
9c7370a1
MKL		 1015 read_lock_bh(&table->tb6_lock);
1016 if (rt->rt6i_pcpu) {
1017 p = this_cpu_ptr(rt->rt6i_pcpu);
1018 prev = cmpxchg(p, NULL, pcpu_rt);
1019 if (prev) {
1020 /* If someone did it before us, return prev instead */
1021 dst_destroy(&pcpu_rt->dst);
1022 pcpu_rt = prev;
1023 }
1024 } else {
1025 /* rt has been removed from the fib6 tree
1026 * before we have a chance to acquire the read_lock.
1027 * In this case, don't brother to create a pcpu rt
1028 * since rt is going away anyway. The next
1029 * dst_check() will trigger a re-lookup.
1030 */
d52d3997 1031 dst_destroy(&pcpu_rt->dst);
9c7370a1 1032 pcpu_rt = rt;
d52d3997 1033 }
d52d3997
MKL		 1034 dst_hold(&pcpu_rt->dst);
1035 rt6_dst_from_metrics_check(pcpu_rt);
9c7370a1 1036 read_unlock_bh(&table->tb6_lock);
d52d3997
MKL		 1037 return pcpu_rt;
1038}
1039
8ed67789 1040static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
4c9483b2 1041 struct flowi6 *fl6, int flags)
1da177e4 1042{
367efcb9 1043 struct fib6_node *fn, *saved_fn;
45e4fd26 1044 struct rt6_info *rt;
c71099ac 1045 int strict = 0;
1da177e4 1046
77d16f45 1047 strict |= flags & RT6_LOOKUP_F_IFACE;
367efcb9
MKL		 1048 if (net->ipv6.devconf_all->forwarding == 0)
1049 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 1050
c71099ac 1051 read_lock_bh(&table->tb6_lock);
1da177e4 1052
4c9483b2 1053 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 1054 saved_fn = fn;
1da177e4 1055
ca254490
DA		 1056 if (fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF)
1057 oif = 0;
1058
a3c00e46 1059redo_rt6_select:
367efcb9 1060 rt = rt6_select(fn, oif, strict);
52bd4c0c 1061 if (rt->rt6i_nsiblings)
367efcb9 1062 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 1063 if (rt == net->ipv6.ip6_null_entry) {
1064 fn = fib6_backtrack(fn, &fl6->saddr);
1065 if (fn)
1066 goto redo_rt6_select;
367efcb9
MKL		 1067 else if (strict & RT6_LOOKUP_F_REACHABLE) {
1068 /* also consider unreachable route */
1069 strict &= ~RT6_LOOKUP_F_REACHABLE;
1070 fn = saved_fn;
1071 goto redo_rt6_select;
367efcb9 1072 }
a3c00e46
MKL		 1073 }
1074
fb9de91e 1075
3da59bd9 1076 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
d52d3997
MKL		 1077 dst_use(&rt->dst, jiffies);
1078 read_unlock_bh(&table->tb6_lock);
1079
1080 rt6_dst_from_metrics_check(rt);
1081 return rt;
3da59bd9
MKL		 1082 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
1083 !(rt->rt6i_flags & RTF_GATEWAY))) {
1084 /* Create a RTF_CACHE clone which will not be
1085 * owned by the fib6 tree. It is for the special case where
1086 * the daddr in the skb during the neighbor look-up is different
1087 * from the fl6->daddr used to look-up route here.
1088 */
1089
1090 struct rt6_info *uncached_rt;
1091
d52d3997
MKL		 1092 dst_use(&rt->dst, jiffies);
1093 read_unlock_bh(&table->tb6_lock);
1094
3da59bd9
MKL		 1095 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
1096 dst_release(&rt->dst);
c71099ac 1097
3da59bd9 1098 if (uncached_rt)
8d0b94af 1099 rt6_uncached_list_add(uncached_rt);
3da59bd9
MKL		 1100 else
1101 uncached_rt = net->ipv6.ip6_null_entry;
d52d3997 1102
3da59bd9
MKL		 1103 dst_hold(&uncached_rt->dst);
1104 return uncached_rt;
3da59bd9 1105
d52d3997
MKL		 1106 } else {
1107 /* Get a percpu copy */
1108
1109 struct rt6_info *pcpu_rt;
1110
1111 rt->dst.lastuse = jiffies;
1112 rt->dst.__use++;
1113 pcpu_rt = rt6_get_pcpu_route(rt);
d52d3997 1114
9c7370a1
MKL		 1115 if (pcpu_rt) {
1116 read_unlock_bh(&table->tb6_lock);
1117 } else {
1118 /* We have to do the read_unlock first
1119 * because rt6_make_pcpu_route() may trigger
1120 * ip6_dst_gc() which will take the write_lock.
1121 */
1122 dst_hold(&rt->dst);
1123 read_unlock_bh(&table->tb6_lock);
a73e4195 1124 pcpu_rt = rt6_make_pcpu_route(rt);
9c7370a1
MKL		 1125 dst_release(&rt->dst);
1126 }
d52d3997
MKL		 1127
1128 return pcpu_rt;
9c7370a1 1129
d52d3997 1130 }
1da177e4
LT		 1131}
1132
8ed67789 1133static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 1134 struct flowi6 *fl6, int flags)
4acad72d 1135{
4c9483b2 1136 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 1137}
1138
72331bc0
SL		 1139static struct dst_entry *ip6_route_input_lookup(struct net *net,
1140 struct net_device *dev,
1141 struct flowi6 *fl6, int flags)
1142{
1143 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1144 flags |= RT6_LOOKUP_F_IFACE;
1145
1146 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1147}
1148
c71099ac
TG		 1149void ip6_route_input(struct sk_buff *skb)
1150{
b71d1d42 1151 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1152 struct net *net = dev_net(skb->dev);
adaa70bb 1153 int flags = RT6_LOOKUP_F_HAS_SADDR;
904af04d 1154 struct ip_tunnel_info *tun_info;
4c9483b2 1155 struct flowi6 fl6 = {
ca254490 1156 .flowi6_iif = l3mdev_fib_oif(skb->dev),
4c9483b2
DM		 1157 .daddr = iph->daddr,
1158 .saddr = iph->saddr,
6502ca52 1159 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1160 .flowi6_mark = skb->mark,
1161 .flowi6_proto = iph->nexthdr,
c71099ac 1162 };
adaa70bb 1163
904af04d 1164 tun_info = skb_tunnel_info(skb);
46fa062a 1165 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
904af04d 1166 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
06e9d040 1167 skb_dst_drop(skb);
72331bc0 1168 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1169}
1170
8ed67789 1171static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1172 struct flowi6 *fl6, int flags)
1da177e4 1173{
4c9483b2 1174 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1175}
1176
67ba4152 1177struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk,
4c9483b2 1178 struct flowi6 *fl6)
c71099ac 1179{
ca254490 1180 struct dst_entry *dst;
c71099ac 1181 int flags = 0;
d46a9d67 1182 bool any_src;
c71099ac 1183
ca254490
DA		 1184 dst = l3mdev_rt6_dst_by_oif(net, fl6);
1185 if (dst)
1186 return dst;
1187
1fb9489b 1188 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1189
d46a9d67 1190 any_src = ipv6_addr_any(&fl6->saddr);
741a11d9 1191 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
d46a9d67 1192 (fl6->flowi6_oif && any_src))
77d16f45 1193 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1194
d46a9d67 1195 if (!any_src)
adaa70bb 1196 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1197 else if (sk)
1198 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1199
4c9483b2 1200 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 1201}
7159039a 1202EXPORT_SYMBOL(ip6_route_output);
1da177e4 1203
2774c131 1204struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1205{
5c1e6aa3 1206 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1207 struct dst_entry *new = NULL;
1208
f5b0a874 1209 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1210 if (rt) {
0a1f5962 1211 rt6_info_init(rt);
8104891b 1212
0a1f5962 1213 new = &rt->dst;
14e50e57 1214 new->__use = 1;
352e512c 1215 new->input = dst_discard;
ede2059d 1216 new->output = dst_discard_out;
14e50e57 1217
0a1f5962 1218 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1219 rt->rt6i_idev = ort->rt6i_idev;
1220 if (rt->rt6i_idev)
1221 in6_dev_hold(rt->rt6i_idev);
14e50e57 1222
4e3fd7a0 1223 rt->rt6i_gateway = ort->rt6i_gateway;
0a1f5962 1224 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
14e50e57
DM		 1225 rt->rt6i_metric = 0;
1226
1227 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1228#ifdef CONFIG_IPV6_SUBTREES
1229 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1230#endif
1231
1232 dst_free(new);
1233 }
1234
69ead7af
DM		 1235 dst_release(dst_orig);
1236 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1237}
14e50e57 1238
1da177e4
LT		 1239/*
1240 * Destination cache support functions
1241 */
1242
4b32b5ad
MKL		 1243static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1244{
1245 if (rt->dst.from &&
1246 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1247 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1248}
1249
3da59bd9
MKL		 1250static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1251{
1252 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1253 return NULL;
1254
1255 if (rt6_check_expired(rt))
1256 return NULL;
1257
1258 return &rt->dst;
1259}
1260
1261static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1262{
5973fb1e
MKL		 1263 if (!__rt6_check_expired(rt) &&
1264 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
3da59bd9
MKL		 1265 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1266 return &rt->dst;
1267 else
1268 return NULL;
1269}
1270
1da177e4
LT		 1271static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1272{
1273 struct rt6_info *rt;
1274
1275 rt = (struct rt6_info *) dst;
1276
6f3118b5
ND		 1277 /* All IPV6 dsts are created with ->obsolete set to the value
1278 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1279 * into this function always.
1280 */
e3bc10bd 1281
4b32b5ad
MKL		 1282 rt6_dst_from_metrics_check(rt);
1283
d52d3997 1284 if ((rt->rt6i_flags & RTF_PCPU) || unlikely(dst->flags & DST_NOCACHE))
3da59bd9
MKL		 1285 return rt6_dst_from_check(rt, cookie);
1286 else
1287 return rt6_check(rt, cookie);
1da177e4
LT		 1288}
1289
1290static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1291{
1292 struct rt6_info *rt = (struct rt6_info *) dst;
1293
1294 if (rt) {
54c1a859
YH		 1295 if (rt->rt6i_flags & RTF_CACHE) {
1296 if (rt6_check_expired(rt)) {
1297 ip6_del_rt(rt);
1298 dst = NULL;
1299 }
1300 } else {
1da177e4 1301 dst_release(dst);
54c1a859
YH		 1302 dst = NULL;
1303 }
1da177e4 1304 }
54c1a859 1305 return dst;
1da177e4
LT		 1306}
1307
1308static void ip6_link_failure(struct sk_buff *skb)
1309{
1310 struct rt6_info *rt;
1311
3ffe533c 1312 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1313
adf30907 1314 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1315 if (rt) {
1eb4f758
HFS		 1316 if (rt->rt6i_flags & RTF_CACHE) {
1317 dst_hold(&rt->dst);
8e3d5be7 1318 ip6_del_rt(rt);
1eb4f758 1319 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1320 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1321 }
1da177e4
LT		 1322 }
1323}
1324
45e4fd26
MKL		 1325static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1326{
1327 struct net *net = dev_net(rt->dst.dev);
1328
1329 rt->rt6i_flags |= RTF_MODIFIED;
1330 rt->rt6i_pmtu = mtu;
1331 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1332}
1333
0d3f6d29
MKL		 1334static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1335{
1336 return !(rt->rt6i_flags & RTF_CACHE) &&
1337 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1338}
1339
45e4fd26
MKL		 1340static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1341 const struct ipv6hdr *iph, u32 mtu)
1da177e4 1342{
67ba4152 1343 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1344
45e4fd26
MKL		 1345 if (rt6->rt6i_flags & RTF_LOCAL)
1346 return;
81aded24 1347
45e4fd26
MKL		 1348 dst_confirm(dst);
1349 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1350 if (mtu >= dst_mtu(dst))
1351 return;
9d289715 1352
0d3f6d29 1353 if (!rt6_cache_allowed_for_pmtu(rt6)) {
45e4fd26
MKL		 1354 rt6_do_update_pmtu(rt6, mtu);
1355 } else {
1356 const struct in6_addr *daddr, *saddr;
1357 struct rt6_info *nrt6;
1358
1359 if (iph) {
1360 daddr = &iph->daddr;
1361 saddr = &iph->saddr;
1362 } else if (sk) {
1363 daddr = &sk->sk_v6_daddr;
1364 saddr = &inet6_sk(sk)->saddr;
1365 } else {
1366 return;
1367 }
1368 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1369 if (nrt6) {
1370 rt6_do_update_pmtu(nrt6, mtu);
1371
1372 /* ip6_ins_rt(nrt6) will bump the
1373 * rt6->rt6i_node->fn_sernum
1374 * which will fail the next rt6_check() and
1375 * invalidate the sk->sk_dst_cache.
1376 */
1377 ip6_ins_rt(nrt6);
1378 }
1da177e4
LT		 1379 }
1380}
1381
45e4fd26
MKL		 1382static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1383 struct sk_buff *skb, u32 mtu)
1384{
1385 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1386}
1387
42ae66c8
DM		 1388void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1389 int oif, u32 mark)
81aded24
DM		 1390{
1391 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1392 struct dst_entry *dst;
1393 struct flowi6 fl6;
1394
1395 memset(&fl6, 0, sizeof(fl6));
1396 fl6.flowi6_oif = oif;
1b3c61dc 1397 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1398 fl6.daddr = iph->daddr;
1399 fl6.saddr = iph->saddr;
6502ca52 1400 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1401
1402 dst = ip6_route_output(net, NULL, &fl6);
1403 if (!dst->error)
45e4fd26 1404 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
81aded24
DM		 1405 dst_release(dst);
1406}
1407EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1408
1409void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1410{
1411 ip6_update_pmtu(skb, sock_net(sk), mtu,
1412 sk->sk_bound_dev_if, sk->sk_mark);
1413}
1414EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1415
b55b76b2
DJ		 1416/* Handle redirects */
1417struct ip6rd_flowi {
1418 struct flowi6 fl6;
1419 struct in6_addr gateway;
1420};
1421
1422static struct rt6_info *__ip6_route_redirect(struct net *net,
1423 struct fib6_table *table,
1424 struct flowi6 *fl6,
1425 int flags)
1426{
1427 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1428 struct rt6_info *rt;
1429 struct fib6_node *fn;
1430
1431 /* Get the "current" route for this destination and
1432 * check if the redirect has come from approriate router.
1433 *
1434 * RFC 4861 specifies that redirects should only be
1435 * accepted if they come from the nexthop to the target.
1436 * Due to the way the routes are chosen, this notion
1437 * is a bit fuzzy and one might need to check all possible
1438 * routes.
1439 */
1440
1441 read_lock_bh(&table->tb6_lock);
1442 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1443restart:
1444 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1445 if (rt6_check_expired(rt))
1446 continue;
1447 if (rt->dst.error)
1448 break;
1449 if (!(rt->rt6i_flags & RTF_GATEWAY))
1450 continue;
1451 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1452 continue;
1453 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1454 continue;
1455 break;
1456 }
1457
1458 if (!rt)
1459 rt = net->ipv6.ip6_null_entry;
1460 else if (rt->dst.error) {
1461 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1462 goto out;
1463 }
1464
1465 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1466 fn = fib6_backtrack(fn, &fl6->saddr);
1467 if (fn)
1468 goto restart;
b55b76b2 1469 }
a3c00e46 1470
b0a1ba59 1471out:
b55b76b2
DJ		 1472 dst_hold(&rt->dst);
1473
1474 read_unlock_bh(&table->tb6_lock);
1475
1476 return rt;
1477};
1478
1479static struct dst_entry *ip6_route_redirect(struct net *net,
1480 const struct flowi6 *fl6,
1481 const struct in6_addr *gateway)
1482{
1483 int flags = RT6_LOOKUP_F_HAS_SADDR;
1484 struct ip6rd_flowi rdfl;
1485
1486 rdfl.fl6 = *fl6;
1487 rdfl.gateway = *gateway;
1488
1489 return fib6_rule_lookup(net, &rdfl.fl6,
1490 flags, __ip6_route_redirect);
1491}
1492
3a5ad2ee
DM		 1493void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1494{
1495 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1496 struct dst_entry *dst;
1497 struct flowi6 fl6;
1498
1499 memset(&fl6, 0, sizeof(fl6));
e374c618 1500 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1501 fl6.flowi6_oif = oif;
1502 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1503 fl6.daddr = iph->daddr;
1504 fl6.saddr = iph->saddr;
6502ca52 1505 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1506
b55b76b2
DJ		 1507 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1508 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1509 dst_release(dst);
1510}
1511EXPORT_SYMBOL_GPL(ip6_redirect);
1512
c92a59ec
DJ		 1513void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1514 u32 mark)
1515{
1516 const struct ipv6hdr *iph = ipv6_hdr(skb);
1517 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1518 struct dst_entry *dst;
1519 struct flowi6 fl6;
1520
1521 memset(&fl6, 0, sizeof(fl6));
e374c618 1522 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1523 fl6.flowi6_oif = oif;
1524 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1525 fl6.daddr = msg->dest;
1526 fl6.saddr = iph->daddr;
1527
b55b76b2
DJ		 1528 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1529 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1530 dst_release(dst);
1531}
1532
3a5ad2ee
DM		 1533void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1534{
1535 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1536}
1537EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1538
0dbaee3b 1539static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1540{
0dbaee3b
DM		 1541 struct net_device *dev = dst->dev;
1542 unsigned int mtu = dst_mtu(dst);
1543 struct net *net = dev_net(dev);
1544
1da177e4
LT		 1545 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1546
5578689a
DL		 1547 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1548 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1549
1550 /*
1ab1457c
YH		 1551 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1552 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1553 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1554 * rely only on pmtu discovery"
1555 */
1556 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1557 mtu = IPV6_MAXPLEN;
1558 return mtu;
1559}
1560
ebb762f2 1561static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1562{
4b32b5ad
MKL		 1563 const struct rt6_info *rt = (const struct rt6_info *)dst;
1564 unsigned int mtu = rt->rt6i_pmtu;
d33e4553 1565 struct inet6_dev *idev;
618f9bc7 1566
4b32b5ad
MKL		 1567 if (mtu)
1568 goto out;
1569
1570 mtu = dst_metric_raw(dst, RTAX_MTU);
618f9bc7 1571 if (mtu)
30f78d8e 1572 goto out;
618f9bc7
SK		 1573
1574 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1575
1576 rcu_read_lock();
1577 idev = __in6_dev_get(dst->dev);
1578 if (idev)
1579 mtu = idev->cnf.mtu6;
1580 rcu_read_unlock();
1581
30f78d8e
ED		 1582out:
1583 return min_t(unsigned int, mtu, IP6_MAX_MTU);
d33e4553
DM		 1584}
1585
3b00944c
YH		 1586static struct dst_entry *icmp6_dst_gc_list;
1587static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1588
3b00944c 1589struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1590 struct flowi6 *fl6)
1da177e4 1591{
87a11578 1592 struct dst_entry *dst;
1da177e4
LT		 1593 struct rt6_info *rt;
1594 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1595 struct net *net = dev_net(dev);
1da177e4 1596
38308473 1597 if (unlikely(!idev))
122bdf67 1598 return ERR_PTR(-ENODEV);
1da177e4 1599
ad706862 1600 rt = ip6_dst_alloc(net, dev, 0);
38308473 1601 if (unlikely(!rt)) {
1da177e4 1602 in6_dev_put(idev);
87a11578 1603 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1604 goto out;
1605 }
1606
8e2ec639
YZ		 1607 rt->dst.flags |= DST_HOST;
1608 rt->dst.output = ip6_output;
d8d1f30b 1609 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1610 rt->rt6i_gateway = fl6->daddr;
87a11578 1611 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1612 rt->rt6i_dst.plen = 128;
1613 rt->rt6i_idev = idev;
14edd87d 1614 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1615
3b00944c 1616 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1617 rt->dst.next = icmp6_dst_gc_list;
1618 icmp6_dst_gc_list = &rt->dst;
3b00944c 1619 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1620
5578689a 1621 fib6_force_start_gc(net);
1da177e4 1622
87a11578
DM		 1623 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1624
1da177e4 1625out:
87a11578 1626 return dst;
1da177e4
LT		 1627}
1628
3d0f24a7 1629int icmp6_dst_gc(void)
1da177e4 1630{
e9476e95 1631 struct dst_entry *dst, **pprev;
3d0f24a7 1632 int more = 0;
1da177e4 1633
3b00944c
YH		 1634 spin_lock_bh(&icmp6_dst_lock);
1635 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1636
1da177e4
LT		 1637 while ((dst = *pprev) != NULL) {
1638 if (!atomic_read(&dst->__refcnt)) {
1639 *pprev = dst->next;
1640 dst_free(dst);
1da177e4
LT		 1641 } else {
1642 pprev = &dst->next;
3d0f24a7 1643 ++more;
1da177e4
LT		 1644 }
1645 }
1646
3b00944c 1647 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1648
3d0f24a7 1649 return more;
1da177e4
LT		 1650}
1651
1e493d19
DM		 1652static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1653 void *arg)
1654{
1655 struct dst_entry *dst, **pprev;
1656
1657 spin_lock_bh(&icmp6_dst_lock);
1658 pprev = &icmp6_dst_gc_list;
1659 while ((dst = *pprev) != NULL) {
1660 struct rt6_info *rt = (struct rt6_info *) dst;
1661 if (func(rt, arg)) {
1662 *pprev = dst->next;
1663 dst_free(dst);
1664 } else {
1665 pprev = &dst->next;
1666 }
1667 }
1668 spin_unlock_bh(&icmp6_dst_lock);
1669}
1670
569d3645 1671static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1672{
86393e52 1673 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1674 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1675 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1676 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1677 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1678 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1679 int entries;
7019b78e 1680
fc66f95c 1681 entries = dst_entries_get_fast(ops);
49a18d86 1682 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1683 entries <= rt_max_size)
1da177e4
LT		 1684 goto out;
1685
6891a346 1686 net->ipv6.ip6_rt_gc_expire++;
14956643 1687 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1688 entries = dst_entries_get_slow(ops);
1689 if (entries < ops->gc_thresh)
7019b78e 1690 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1691out:
7019b78e 1692 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1693 return entries > rt_max_size;
1da177e4
LT		 1694}
1695
e715b6d3
FW		 1696static int ip6_convert_metrics(struct mx6_config *mxc,
1697 const struct fib6_config *cfg)
1698{
c3a8d947 1699 bool ecn_ca = false;
e715b6d3
FW		 1700 struct nlattr *nla;
1701 int remaining;
1702 u32 *mp;
1703
63159f29 1704 if (!cfg->fc_mx)
e715b6d3
FW		 1705 return 0;
1706
1707 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1708 if (unlikely(!mp))
1709 return -ENOMEM;
1710
1711 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1712 int type = nla_type(nla);
1bb14807 1713 u32 val;
e715b6d3 1714
1bb14807
DB		 1715 if (!type)
1716 continue;
1717 if (unlikely(type > RTAX_MAX))
1718 goto err;
ea697639 1719
1bb14807
DB		 1720 if (type == RTAX_CC_ALGO) {
1721 char tmp[TCP_CA_NAME_MAX];
e715b6d3 1722
1bb14807 1723 nla_strlcpy(tmp, nla, sizeof(tmp));
c3a8d947 1724 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
1bb14807
DB		 1725 if (val == TCP_CA_UNSPEC)
1726 goto err;
1727 } else {
1728 val = nla_get_u32(nla);
e715b6d3 1729 }
b8d3e416
DB		 1730 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1731 goto err;
1bb14807
DB		 1732
1733 mp[type - 1] = val;
1734 __set_bit(type - 1, mxc->mx_valid);
e715b6d3
FW		 1735 }
1736
c3a8d947
DB		 1737 if (ecn_ca) {
1738 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1739 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1740 }
e715b6d3 1741
c3a8d947 1742 mxc->mx = mp;
e715b6d3
FW		 1743 return 0;
1744 err:
1745 kfree(mp);
1746 return -EINVAL;
1747}
1da177e4 1748
8c5b83f0 1749static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
1da177e4 1750{
5578689a 1751 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1752 struct rt6_info *rt = NULL;
1753 struct net_device *dev = NULL;
1754 struct inet6_dev *idev = NULL;
c71099ac 1755 struct fib6_table *table;
1da177e4 1756 int addr_type;
8c5b83f0 1757 int err = -EINVAL;
1da177e4 1758
86872cb5 1759 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
8c5b83f0 1760 goto out;
1da177e4 1761#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1762 if (cfg->fc_src_len)
8c5b83f0 1763 goto out;
1da177e4 1764#endif
86872cb5 1765 if (cfg->fc_ifindex) {
1da177e4 1766 err = -ENODEV;
5578689a 1767 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1768 if (!dev)
1769 goto out;
1770 idev = in6_dev_get(dev);
1771 if (!idev)
1772 goto out;
1773 }
1774
86872cb5
TG		 1775 if (cfg->fc_metric == 0)
1776 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1777
d71314b4 1778 err = -ENOBUFS;
38308473
DM		 1779 if (cfg->fc_nlinfo.nlh &&
1780 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1781 table = fib6_get_table(net, cfg->fc_table);
38308473 1782 if (!table) {
f3213831 1783 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1784 table = fib6_new_table(net, cfg->fc_table);
1785 }
1786 } else {
1787 table = fib6_new_table(net, cfg->fc_table);
1788 }
38308473
DM		 1789
1790 if (!table)
c71099ac 1791 goto out;
c71099ac 1792
ad706862
MKL		 1793 rt = ip6_dst_alloc(net, NULL,
1794 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
1da177e4 1795
38308473 1796 if (!rt) {
1da177e4
LT		 1797 err = -ENOMEM;
1798 goto out;
1799 }
1800
1716a961
G		 1801 if (cfg->fc_flags & RTF_EXPIRES)
1802 rt6_set_expires(rt, jiffies +
1803 clock_t_to_jiffies(cfg->fc_expires));
1804 else
1805 rt6_clean_expires(rt);
1da177e4 1806
86872cb5
TG		 1807 if (cfg->fc_protocol == RTPROT_UNSPEC)
1808 cfg->fc_protocol = RTPROT_BOOT;
1809 rt->rt6i_protocol = cfg->fc_protocol;
1810
1811 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1812
1813 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1814 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1815 else if (cfg->fc_flags & RTF_LOCAL)
1816 rt->dst.input = ip6_input;
1da177e4 1817 else
d8d1f30b 1818 rt->dst.input = ip6_forward;
1da177e4 1819
d8d1f30b 1820 rt->dst.output = ip6_output;
1da177e4 1821
19e42e45
RP		 1822 if (cfg->fc_encap) {
1823 struct lwtunnel_state *lwtstate;
1824
1825 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
127eb7cd
TH		 1826 cfg->fc_encap, AF_INET6, cfg,
1827 &lwtstate);
19e42e45
RP		 1828 if (err)
1829 goto out;
61adedf3
JB		 1830 rt->dst.lwtstate = lwtstate_get(lwtstate);
1831 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1832 rt->dst.lwtstate->orig_output = rt->dst.output;
1833 rt->dst.output = lwtunnel_output;
25368623 1834 }
61adedf3
JB		 1835 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1836 rt->dst.lwtstate->orig_input = rt->dst.input;
1837 rt->dst.input = lwtunnel_input;
25368623 1838 }
19e42e45
RP		 1839 }
1840
86872cb5
TG		 1841 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1842 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1843 if (rt->rt6i_dst.plen == 128)
e5fd387a 1844 rt->dst.flags |= DST_HOST;
e5fd387a 1845
1da177e4 1846#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1847 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1848 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1849#endif
1850
86872cb5 1851 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1852
1853 /* We cannot add true routes via loopback here,
1854 they would result in kernel looping; promote them to reject routes
1855 */
86872cb5 1856 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1857 (dev && (dev->flags & IFF_LOOPBACK) &&
1858 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1859 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1860 /* hold loopback dev/idev if we haven't done so. */
5578689a 1861 if (dev != net->loopback_dev) {
1da177e4
LT		 1862 if (dev) {
1863 dev_put(dev);
1864 in6_dev_put(idev);
1865 }
5578689a 1866 dev = net->loopback_dev;
1da177e4
LT		 1867 dev_hold(dev);
1868 idev = in6_dev_get(dev);
1869 if (!idev) {
1870 err = -ENODEV;
1871 goto out;
1872 }
1873 }
1da177e4 1874 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1875 switch (cfg->fc_type) {
1876 case RTN_BLACKHOLE:
1877 rt->dst.error = -EINVAL;
ede2059d 1878 rt->dst.output = dst_discard_out;
7150aede 1879 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1880 break;
1881 case RTN_PROHIBIT:
1882 rt->dst.error = -EACCES;
7150aede
K		 1883 rt->dst.output = ip6_pkt_prohibit_out;
1884 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1885 break;
b4949ab2 1886 case RTN_THROW:
0315e382 1887 case RTN_UNREACHABLE:
ef2c7d7b 1888 default:
7150aede 1889 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
0315e382
NF		 1890 : (cfg->fc_type == RTN_UNREACHABLE)
1891 ? -EHOSTUNREACH : -ENETUNREACH;
7150aede
K		 1892 rt->dst.output = ip6_pkt_discard_out;
1893 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1894 break;
1895 }
1da177e4
LT		 1896 goto install_route;
1897 }
1898
86872cb5 1899 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1900 const struct in6_addr *gw_addr;
1da177e4
LT		 1901 int gwa_type;
1902
86872cb5 1903 gw_addr = &cfg->fc_gateway;
330567b7 1904 gwa_type = ipv6_addr_type(gw_addr);
48ed7b26
FW		 1905
1906 /* if gw_addr is local we will fail to detect this in case
1907 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1908 * will return already-added prefix route via interface that
1909 * prefix route was assigned to, which might be non-loopback.
1910 */
1911 err = -EINVAL;
330567b7
FW		 1912 if (ipv6_chk_addr_and_flags(net, gw_addr,
1913 gwa_type & IPV6_ADDR_LINKLOCAL ?
1914 dev : NULL, 0, 0))
48ed7b26
FW		 1915 goto out;
1916
4e3fd7a0 1917 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1918
1919 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1920 struct rt6_info *grt;
1921
1922 /* IPv6 strictly inhibits using not link-local
1923 addresses as nexthop address.
1924 Otherwise, router will not able to send redirects.
1925 It is very good, but in some (rare!) circumstances
1926 (SIT, PtP, NBMA NOARP links) it is handy to allow
1927 some exceptions. --ANK
1928 */
38308473 1929 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1930 goto out;
1931
5578689a 1932 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1933
1934 err = -EHOSTUNREACH;
38308473 1935 if (!grt)
1da177e4
LT		 1936 goto out;
1937 if (dev) {
d1918542 1938 if (dev != grt->dst.dev) {
94e187c0 1939 ip6_rt_put(grt);
1da177e4
LT		 1940 goto out;
1941 }
1942 } else {
d1918542 1943 dev = grt->dst.dev;
1da177e4
LT		 1944 idev = grt->rt6i_idev;
1945 dev_hold(dev);
1946 in6_dev_hold(grt->rt6i_idev);
1947 }
38308473 1948 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 1949 err = 0;
94e187c0 1950 ip6_rt_put(grt);
1da177e4
LT		 1951
1952 if (err)
1953 goto out;
1954 }
1955 err = -EINVAL;
38308473 1956 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 1957 goto out;
1958 }
1959
1960 err = -ENODEV;
38308473 1961 if (!dev)
1da177e4
LT		 1962 goto out;
1963
c3968a85
DW		 1964 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1965 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1966 err = -EINVAL;
1967 goto out;
1968 }
4e3fd7a0 1969 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 1970 rt->rt6i_prefsrc.plen = 128;
1971 } else
1972 rt->rt6i_prefsrc.plen = 0;
1973
86872cb5 1974 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1975
1976install_route:
d8d1f30b 1977 rt->dst.dev = dev;
1da177e4 1978 rt->rt6i_idev = idev;
c71099ac 1979 rt->rt6i_table = table;
63152fc0 1980
c346dca1 1981 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1982
8c5b83f0 1983 return rt;
6b9ea5a6
RP		 1984out:
1985 if (dev)
1986 dev_put(dev);
1987 if (idev)
1988 in6_dev_put(idev);
1989 if (rt)
1990 dst_free(&rt->dst);
1991
8c5b83f0 1992 return ERR_PTR(err);
6b9ea5a6
RP		 1993}
1994
1995int ip6_route_add(struct fib6_config *cfg)
1996{
1997 struct mx6_config mxc = { .mx = NULL, };
8c5b83f0 1998 struct rt6_info *rt;
6b9ea5a6
RP		 1999 int err;
2000
8c5b83f0
RP		 2001 rt = ip6_route_info_create(cfg);
2002 if (IS_ERR(rt)) {
2003 err = PTR_ERR(rt);
2004 rt = NULL;
6b9ea5a6 2005 goto out;
8c5b83f0 2006 }
6b9ea5a6 2007
e715b6d3
FW		 2008 err = ip6_convert_metrics(&mxc, cfg);
2009 if (err)
2010 goto out;
1da177e4 2011
e715b6d3
FW		 2012 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2013
2014 kfree(mxc.mx);
6b9ea5a6 2015
e715b6d3 2016 return err;
1da177e4 2017out:
1da177e4 2018 if (rt)
d8d1f30b 2019 dst_free(&rt->dst);
6b9ea5a6 2020
1da177e4
LT		 2021 return err;
2022}
2023
86872cb5 2024static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2025{
2026 int err;
c71099ac 2027 struct fib6_table *table;
d1918542 2028 struct net *net = dev_net(rt->dst.dev);
1da177e4 2029
8e3d5be7
MKL		 2030 if (rt == net->ipv6.ip6_null_entry ||
2031 rt->dst.flags & DST_NOCACHE) {
6825a26c
G		 2032 err = -ENOENT;
2033 goto out;
2034 }
6c813a72 2035
c71099ac
TG		 2036 table = rt->rt6i_table;
2037 write_lock_bh(&table->tb6_lock);
86872cb5 2038 err = fib6_del(rt, info);
c71099ac 2039 write_unlock_bh(&table->tb6_lock);
1da177e4 2040
6825a26c 2041out:
94e187c0 2042 ip6_rt_put(rt);
1da177e4
LT		 2043 return err;
2044}
2045
e0a1ad73
TG		 2046int ip6_del_rt(struct rt6_info *rt)
2047{
4d1169c1 2048 struct nl_info info = {
d1918542 2049 .nl_net = dev_net(rt->dst.dev),
4d1169c1 2050 };
528c4ceb 2051 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 2052}
2053
86872cb5 2054static int ip6_route_del(struct fib6_config *cfg)
1da177e4 2055{
c71099ac 2056 struct fib6_table *table;
1da177e4
LT		 2057 struct fib6_node *fn;
2058 struct rt6_info *rt;
2059 int err = -ESRCH;
2060
5578689a 2061 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 2062 if (!table)
c71099ac
TG		 2063 return err;
2064
2065 read_lock_bh(&table->tb6_lock);
1da177e4 2066
c71099ac 2067 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 2068 &cfg->fc_dst, cfg->fc_dst_len,
2069 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 2070
1da177e4 2071 if (fn) {
d8d1f30b 2072 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 2073 if ((rt->rt6i_flags & RTF_CACHE) &&
2074 !(cfg->fc_flags & RTF_CACHE))
2075 continue;
86872cb5 2076 if (cfg->fc_ifindex &&
d1918542
DM		 2077 (!rt->dst.dev ||
2078 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 2079 continue;
86872cb5
TG		 2080 if (cfg->fc_flags & RTF_GATEWAY &&
2081 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 2082 continue;
86872cb5 2083 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 2084 continue;
d8d1f30b 2085 dst_hold(&rt->dst);
c71099ac 2086 read_unlock_bh(&table->tb6_lock);
1da177e4 2087
86872cb5 2088 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 2089 }
2090 }
c71099ac 2091 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2092
2093 return err;
2094}
2095
6700c270 2096static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 2097{
a6279458 2098 struct netevent_redirect netevent;
e8599ff4 2099 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 2100 struct ndisc_options ndopts;
2101 struct inet6_dev *in6_dev;
2102 struct neighbour *neigh;
71bcdba0 2103 struct rd_msg *msg;
6e157b6a
DM		 2104 int optlen, on_link;
2105 u8 *lladdr;
e8599ff4 2106
29a3cad5 2107 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 2108 optlen -= sizeof(*msg);
e8599ff4
DM		 2109
2110 if (optlen < 0) {
6e157b6a 2111 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 2112 return;
2113 }
2114
71bcdba0 2115 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 2116
71bcdba0 2117 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 2118 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 2119 return;
2120 }
2121
6e157b6a 2122 on_link = 0;
71bcdba0 2123 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 2124 on_link = 1;
71bcdba0 2125 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 2126 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 2127 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 2128 return;
2129 }
2130
2131 in6_dev = __in6_dev_get(skb->dev);
2132 if (!in6_dev)
2133 return;
2134 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2135 return;
2136
2137 /* RFC2461 8.1:
2138 * The IP source address of the Redirect MUST be the same as the current
2139 * first-hop router for the specified ICMP Destination Address.
2140 */
2141
71bcdba0 2142 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 2143 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2144 return;
2145 }
6e157b6a
DM		 2146
2147 lladdr = NULL;
e8599ff4
DM		 2148 if (ndopts.nd_opts_tgt_lladdr) {
2149 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2150 skb->dev);
2151 if (!lladdr) {
2152 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2153 return;
2154 }
2155 }
2156
6e157b6a 2157 rt = (struct rt6_info *) dst;
ec13ad1d 2158 if (rt->rt6i_flags & RTF_REJECT) {
6e157b6a 2159 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 2160 return;
6e157b6a 2161 }
e8599ff4 2162
6e157b6a
DM		 2163 /* Redirect received -> path was valid.
2164 * Look, redirects are sent only in response to data packets,
2165 * so that this nexthop apparently is reachable. --ANK
2166 */
2167 dst_confirm(&rt->dst);
a6279458 2168
71bcdba0 2169 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 2170 if (!neigh)
2171 return;
a6279458 2172
1da177e4
LT		 2173 /*
2174 * We have finally decided to accept it.
2175 */
2176
1ab1457c 2177 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 2178 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2179 NEIGH_UPDATE_F_OVERRIDE|
2180 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
2181 NEIGH_UPDATE_F_ISROUTER))
2182 );
2183
83a09abd 2184 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
38308473 2185 if (!nrt)
1da177e4
LT		 2186 goto out;
2187
2188 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2189 if (on_link)
2190 nrt->rt6i_flags &= ~RTF_GATEWAY;
2191
4e3fd7a0 2192 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 2193
40e22e8f 2194 if (ip6_ins_rt(nrt))
1da177e4
LT		 2195 goto out;
2196
d8d1f30b
CG		 2197 netevent.old = &rt->dst;
2198 netevent.new = &nrt->dst;
71bcdba0 2199 netevent.daddr = &msg->dest;
60592833 2200 netevent.neigh = neigh;
8d71740c
TT		 2201 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2202
38308473 2203 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 2204 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 2205 ip6_del_rt(rt);
1da177e4
LT		 2206 }
2207
2208out:
e8599ff4 2209 neigh_release(neigh);
6e157b6a
DM		 2210}
2211
1da177e4
LT		 2212/*
2213 * Misc support functions
2214 */
2215
4b32b5ad
MKL		 2216static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2217{
2218 BUG_ON(from->dst.from);
2219
2220 rt->rt6i_flags &= ~RTF_EXPIRES;
2221 dst_hold(&from->dst);
2222 rt->dst.from = &from->dst;
2223 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2224}
2225
83a09abd
MKL		 2226static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
2227{
2228 rt->dst.input = ort->dst.input;
2229 rt->dst.output = ort->dst.output;
2230 rt->rt6i_dst = ort->rt6i_dst;
2231 rt->dst.error = ort->dst.error;
2232 rt->rt6i_idev = ort->rt6i_idev;
2233 if (rt->rt6i_idev)
2234 in6_dev_hold(rt->rt6i_idev);
2235 rt->dst.lastuse = jiffies;
2236 rt->rt6i_gateway = ort->rt6i_gateway;
2237 rt->rt6i_flags = ort->rt6i_flags;
2238 rt6_set_from(rt, ort);
2239 rt->rt6i_metric = ort->rt6i_metric;
1da177e4 2240#ifdef CONFIG_IPV6_SUBTREES
83a09abd 2241 rt->rt6i_src = ort->rt6i_src;
1da177e4 2242#endif
83a09abd
MKL		 2243 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2244 rt->rt6i_table = ort->rt6i_table;
61adedf3 2245 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
1da177e4
LT		 2246}
2247
70ceb4f5 2248#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 2249static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 2250 const struct in6_addr *prefix, int prefixlen,
2251 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 2252{
2253 struct fib6_node *fn;
2254 struct rt6_info *rt = NULL;
c71099ac
TG		 2255 struct fib6_table *table;
2256
efa2cea0 2257 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 2258 if (!table)
c71099ac 2259 return NULL;
70ceb4f5 2260
5744dd9b 2261 read_lock_bh(&table->tb6_lock);
67ba4152 2262 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2263 if (!fn)
2264 goto out;
2265
d8d1f30b 2266 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2267 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2268 continue;
2269 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2270 continue;
2271 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2272 continue;
d8d1f30b 2273 dst_hold(&rt->dst);
70ceb4f5
YH		 2274 break;
2275 }
2276out:
5744dd9b 2277 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2278 return rt;
2279}
2280
efa2cea0 2281static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 2282 const struct in6_addr *prefix, int prefixlen,
2283 const struct in6_addr *gwaddr, int ifindex,
95c96174 2284 unsigned int pref)
70ceb4f5 2285{
86872cb5 2286 struct fib6_config cfg = {
238fc7ea 2287 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2288 .fc_ifindex = ifindex,
2289 .fc_dst_len = prefixlen,
2290 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2291 RTF_UP | RTF_PREF(pref),
15e47304 2292 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2293 .fc_nlinfo.nlh = NULL,
2294 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2295 };
2296
ca254490 2297 cfg.fc_table = l3mdev_fib_table_by_index(net, ifindex) ? : RT6_TABLE_INFO;
4e3fd7a0
AD		 2298 cfg.fc_dst = *prefix;
2299 cfg.fc_gateway = *gwaddr;
70ceb4f5 2300
e317da96
YH		 2301 /* We should treat it as a default route if prefix length is 0. */
2302 if (!prefixlen)
86872cb5 2303 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2304
86872cb5 2305 ip6_route_add(&cfg);
70ceb4f5 2306
efa2cea0 2307 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 2308}
2309#endif
2310
b71d1d42 2311struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2312{
1da177e4 2313 struct rt6_info *rt;
c71099ac 2314 struct fib6_table *table;
1da177e4 2315
c346dca1 2316 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 2317 if (!table)
c71099ac 2318 return NULL;
1da177e4 2319
5744dd9b 2320 read_lock_bh(&table->tb6_lock);
67ba4152 2321 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2322 if (dev == rt->dst.dev &&
045927ff 2323 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2324 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2325 break;
2326 }
2327 if (rt)
d8d1f30b 2328 dst_hold(&rt->dst);
5744dd9b 2329 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2330 return rt;
2331}
2332
b71d1d42 2333struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2334 struct net_device *dev,
2335 unsigned int pref)
1da177e4 2336{
86872cb5 2337 struct fib6_config cfg = {
ca254490 2338 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
238fc7ea 2339 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2340 .fc_ifindex = dev->ifindex,
2341 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2342 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2343 .fc_nlinfo.portid = 0,
5578689a 2344 .fc_nlinfo.nlh = NULL,
c346dca1 2345 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2346 };
1da177e4 2347
4e3fd7a0 2348 cfg.fc_gateway = *gwaddr;
1da177e4 2349
86872cb5 2350 ip6_route_add(&cfg);
1da177e4 2351
1da177e4
LT		 2352 return rt6_get_dflt_router(gwaddr, dev);
2353}
2354
7b4da532 2355void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2356{
2357 struct rt6_info *rt;
c71099ac
TG		 2358 struct fib6_table *table;
2359
2360 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2361 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2362 if (!table)
c71099ac 2363 return;
1da177e4
LT		 2364
2365restart:
c71099ac 2366 read_lock_bh(&table->tb6_lock);
d8d1f30b 2367 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2368 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2369 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2370 dst_hold(&rt->dst);
c71099ac 2371 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2372 ip6_del_rt(rt);
1da177e4
LT		 2373 goto restart;
2374 }
2375 }
c71099ac 2376 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2377}
2378
5578689a
DL		 2379static void rtmsg_to_fib6_config(struct net *net,
2380 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2381 struct fib6_config *cfg)
2382{
2383 memset(cfg, 0, sizeof(*cfg));
2384
ca254490
DA		 2385 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2386 : RT6_TABLE_MAIN;
86872cb5
TG		 2387 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2388 cfg->fc_metric = rtmsg->rtmsg_metric;
2389 cfg->fc_expires = rtmsg->rtmsg_info;
2390 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2391 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2392 cfg->fc_flags = rtmsg->rtmsg_flags;
2393
5578689a 2394 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2395
4e3fd7a0
AD		 2396 cfg->fc_dst = rtmsg->rtmsg_dst;
2397 cfg->fc_src = rtmsg->rtmsg_src;
2398 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2399}
2400
5578689a 2401int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2402{
86872cb5 2403 struct fib6_config cfg;
1da177e4
LT		 2404 struct in6_rtmsg rtmsg;
2405 int err;
2406
67ba4152 2407 switch (cmd) {
1da177e4
LT		 2408 case SIOCADDRT: /* Add a route */
2409 case SIOCDELRT: /* Delete a route */
af31f412 2410 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2411 return -EPERM;
2412 err = copy_from_user(&rtmsg, arg,
2413 sizeof(struct in6_rtmsg));
2414 if (err)
2415 return -EFAULT;
86872cb5 2416
5578689a 2417 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2418
1da177e4
LT		 2419 rtnl_lock();
2420 switch (cmd) {
2421 case SIOCADDRT:
86872cb5 2422 err = ip6_route_add(&cfg);
1da177e4
LT		 2423 break;
2424 case SIOCDELRT:
86872cb5 2425 err = ip6_route_del(&cfg);
1da177e4
LT		 2426 break;
2427 default:
2428 err = -EINVAL;
2429 }
2430 rtnl_unlock();
2431
2432 return err;
3ff50b79 2433 }
1da177e4
LT		 2434
2435 return -EINVAL;
2436}
2437
2438/*
2439 * Drop the packet on the floor
2440 */
2441
d5fdd6ba 2442static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2443{
612f09e8 2444 int type;
adf30907 2445 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2446 switch (ipstats_mib_noroutes) {
2447 case IPSTATS_MIB_INNOROUTES:
0660e03f 2448 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2449 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2450 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2451 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2452 break;
2453 }
2454 /* FALLTHROUGH */
2455 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2456 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2457 ipstats_mib_noroutes);
612f09e8
YH		 2458 break;
2459 }
3ffe533c 2460 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2461 kfree_skb(skb);
2462 return 0;
2463}
2464
9ce8ade0
TG		 2465static int ip6_pkt_discard(struct sk_buff *skb)
2466{
612f09e8 2467 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2468}
2469
ede2059d 2470static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
1da177e4 2471{
adf30907 2472 skb->dev = skb_dst(skb)->dev;
612f09e8 2473 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2474}
2475
9ce8ade0
TG		 2476static int ip6_pkt_prohibit(struct sk_buff *skb)
2477{
612f09e8 2478 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2479}
2480
ede2059d 2481static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
9ce8ade0 2482{
adf30907 2483 skb->dev = skb_dst(skb)->dev;
612f09e8 2484 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2485}
2486
1da177e4
LT		 2487/*
2488 * Allocate a dst for local (unicast / anycast) address.
2489 */
2490
2491struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2492 const struct in6_addr *addr,
8f031519 2493 bool anycast)
1da177e4 2494{
ca254490 2495 u32 tb_id;
c346dca1 2496 struct net *net = dev_net(idev->dev);
a3300ef4 2497 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
ad706862 2498 DST_NOCOUNT);
a3300ef4 2499 if (!rt)
1da177e4
LT		 2500 return ERR_PTR(-ENOMEM);
2501
1da177e4
LT		 2502 in6_dev_hold(idev);
2503
11d53b49 2504 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2505 rt->dst.input = ip6_input;
2506 rt->dst.output = ip6_output;
1da177e4 2507 rt->rt6i_idev = idev;
1da177e4
LT		 2508
2509 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2510 if (anycast)
2511 rt->rt6i_flags |= RTF_ANYCAST;
2512 else
1da177e4 2513 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2514
550bab42 2515 rt->rt6i_gateway = *addr;
4e3fd7a0 2516 rt->rt6i_dst.addr = *addr;
1da177e4 2517 rt->rt6i_dst.plen = 128;
ca254490
DA		 2518 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2519 rt->rt6i_table = fib6_get_table(net, tb_id);
8e3d5be7 2520 rt->dst.flags |= DST_NOCACHE;
1da177e4 2521
d8d1f30b 2522 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2523
2524 return rt;
2525}
2526
c3968a85
DW		 2527int ip6_route_get_saddr(struct net *net,
2528 struct rt6_info *rt,
b71d1d42 2529 const struct in6_addr *daddr,
c3968a85
DW		 2530 unsigned int prefs,
2531 struct in6_addr *saddr)
2532{
e16e888b
MS		 2533 struct inet6_dev *idev =
2534 rt ? ip6_dst_idev((struct dst_entry *)rt) : NULL;
c3968a85 2535 int err = 0;
e16e888b 2536 if (rt && rt->rt6i_prefsrc.plen)
4e3fd7a0 2537 *saddr = rt->rt6i_prefsrc.addr;
c3968a85
DW		 2538 else
2539 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2540 daddr, prefs, saddr);
2541 return err;
2542}
2543
2544/* remove deleted ip from prefsrc entries */
2545struct arg_dev_net_ip {
2546 struct net_device *dev;
2547 struct net *net;
2548 struct in6_addr *addr;
2549};
2550
2551static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2552{
2553 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2554 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2555 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2556
d1918542 2557 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2558 rt != net->ipv6.ip6_null_entry &&
2559 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2560 /* remove prefsrc entry */
2561 rt->rt6i_prefsrc.plen = 0;
2562 }
2563 return 0;
2564}
2565
2566void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2567{
2568 struct net *net = dev_net(ifp->idev->dev);
2569 struct arg_dev_net_ip adni = {
2570 .dev = ifp->idev->dev,
2571 .net = net,
2572 .addr = &ifp->addr,
2573 };
0c3584d5 2574 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2575}
2576
be7a010d
DJ		 2577#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2578#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2579
2580/* Remove routers and update dst entries when gateway turn into host. */
2581static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2582{
2583 struct in6_addr *gateway = (struct in6_addr *)arg;
2584
2585 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2586 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2587 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2588 return -1;
2589 }
2590 return 0;
2591}
2592
2593void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2594{
2595 fib6_clean_all(net, fib6_clean_tohost, gateway);
2596}
2597
8ed67789
DL		 2598struct arg_dev_net {
2599 struct net_device *dev;
2600 struct net *net;
2601};
2602
1da177e4
LT		 2603static int fib6_ifdown(struct rt6_info *rt, void *arg)
2604{
bc3ef660 2605 const struct arg_dev_net *adn = arg;
2606 const struct net_device *dev = adn->dev;
8ed67789 2607
d1918542 2608 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2609 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2610 return -1;
c159d30c 2611
1da177e4
LT		 2612 return 0;
2613}
2614
f3db4851 2615void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2616{
8ed67789
DL		 2617 struct arg_dev_net adn = {
2618 .dev = dev,
2619 .net = net,
2620 };
2621
0c3584d5 2622 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2623 icmp6_clean_all(fib6_ifdown, &adn);
e332bc67
EB		 2624 if (dev)
2625 rt6_uncached_list_flush_dev(net, dev);
1da177e4
LT		 2626}
2627
95c96174 2628struct rt6_mtu_change_arg {
1da177e4 2629 struct net_device *dev;
95c96174 2630 unsigned int mtu;
1da177e4
LT		 2631};
2632
2633static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2634{
2635 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2636 struct inet6_dev *idev;
2637
2638 /* In IPv6 pmtu discovery is not optional,
2639 so that RTAX_MTU lock cannot disable it.
2640 We still use this lock to block changes
2641 caused by addrconf/ndisc.
2642 */
2643
2644 idev = __in6_dev_get(arg->dev);
38308473 2645 if (!idev)
1da177e4
LT		 2646 return 0;
2647
2648 /* For administrative MTU increase, there is no way to discover
2649 IPv6 PMTU increase, so PMTU increase should be updated here.
2650 Since RFC 1981 doesn't include administrative MTU increase
2651 update PMTU increase is a MUST. (i.e. jumbo frame)
2652 */
2653 /*
2654 If new MTU is less than route PMTU, this new MTU will be the
2655 lowest MTU in the path, update the route PMTU to reflect PMTU
2656 decreases; if new MTU is greater than route PMTU, and the
2657 old MTU is the lowest MTU in the path, update the route PMTU
2658 to reflect the increase. In this case if the other nodes' MTU
2659 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2660 PMTU discouvery.
2661 */
d1918542 2662 if (rt->dst.dev == arg->dev &&
4b32b5ad
MKL		 2663 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2664 if (rt->rt6i_flags & RTF_CACHE) {
2665 /* For RTF_CACHE with rt6i_pmtu == 0
2666 * (i.e. a redirected route),
2667 * the metrics of its rt->dst.from has already
2668 * been updated.
2669 */
2670 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2671 rt->rt6i_pmtu = arg->mtu;
2672 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2673 (dst_mtu(&rt->dst) < arg->mtu &&
2674 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2675 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2676 }
566cfd8f 2677 }
1da177e4
LT		 2678 return 0;
2679}
2680
95c96174 2681void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2682{
c71099ac
TG		 2683 struct rt6_mtu_change_arg arg = {
2684 .dev = dev,
2685 .mtu = mtu,
2686 };
1da177e4 2687
0c3584d5 2688 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2689}
2690
ef7c79ed 2691static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2692 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2693 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2694 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2695 [RTA_PRIORITY] = { .type = NLA_U32 },
2696 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2697 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2698 [RTA_PREF] = { .type = NLA_U8 },
19e42e45
RP		 2699 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2700 [RTA_ENCAP] = { .type = NLA_NESTED },
86872cb5
TG		 2701};
2702
2703static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2704 struct fib6_config *cfg)
1da177e4 2705{
86872cb5
TG		 2706 struct rtmsg *rtm;
2707 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2708 unsigned int pref;
86872cb5 2709 int err;
1da177e4 2710
86872cb5
TG		 2711 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2712 if (err < 0)
2713 goto errout;
1da177e4 2714
86872cb5
TG		 2715 err = -EINVAL;
2716 rtm = nlmsg_data(nlh);
2717 memset(cfg, 0, sizeof(*cfg));
2718
2719 cfg->fc_table = rtm->rtm_table;
2720 cfg->fc_dst_len = rtm->rtm_dst_len;
2721 cfg->fc_src_len = rtm->rtm_src_len;
2722 cfg->fc_flags = RTF_UP;
2723 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2724 cfg->fc_type = rtm->rtm_type;
86872cb5 2725
ef2c7d7b
ND		 2726 if (rtm->rtm_type == RTN_UNREACHABLE ||
2727 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2728 rtm->rtm_type == RTN_PROHIBIT ||
2729 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2730 cfg->fc_flags |= RTF_REJECT;
2731
ab79ad14
MŻ		 2732 if (rtm->rtm_type == RTN_LOCAL)
2733 cfg->fc_flags |= RTF_LOCAL;
2734
1f56a01f
MKL		 2735 if (rtm->rtm_flags & RTM_F_CLONED)
2736 cfg->fc_flags |= RTF_CACHE;
2737
15e47304 2738 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2739 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2740 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2741
2742 if (tb[RTA_GATEWAY]) {
67b61f6c 2743 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2744 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2745 }
86872cb5
TG		 2746
2747 if (tb[RTA_DST]) {
2748 int plen = (rtm->rtm_dst_len + 7) >> 3;
2749
2750 if (nla_len(tb[RTA_DST]) < plen)
2751 goto errout;
2752
2753 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2754 }
86872cb5
TG		 2755
2756 if (tb[RTA_SRC]) {
2757 int plen = (rtm->rtm_src_len + 7) >> 3;
2758
2759 if (nla_len(tb[RTA_SRC]) < plen)
2760 goto errout;
2761
2762 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2763 }
86872cb5 2764
c3968a85 2765 if (tb[RTA_PREFSRC])
67b61f6c 2766 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2767
86872cb5
TG		 2768 if (tb[RTA_OIF])
2769 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2770
2771 if (tb[RTA_PRIORITY])
2772 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2773
2774 if (tb[RTA_METRICS]) {
2775 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2776 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2777 }
86872cb5
TG		 2778
2779 if (tb[RTA_TABLE])
2780 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2781
51ebd318
ND		 2782 if (tb[RTA_MULTIPATH]) {
2783 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2784 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2785 }
2786
c78ba6d6
LR		 2787 if (tb[RTA_PREF]) {
2788 pref = nla_get_u8(tb[RTA_PREF]);
2789 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2790 pref != ICMPV6_ROUTER_PREF_HIGH)
2791 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2792 cfg->fc_flags |= RTF_PREF(pref);
2793 }
2794
19e42e45
RP		 2795 if (tb[RTA_ENCAP])
2796 cfg->fc_encap = tb[RTA_ENCAP];
2797
2798 if (tb[RTA_ENCAP_TYPE])
2799 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2800
86872cb5
TG		 2801 err = 0;
2802errout:
2803 return err;
1da177e4
LT		 2804}
2805
6b9ea5a6
RP		 2806struct rt6_nh {
2807 struct rt6_info *rt6_info;
2808 struct fib6_config r_cfg;
2809 struct mx6_config mxc;
2810 struct list_head next;
2811};
2812
2813static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2814{
2815 struct rt6_nh *nh;
2816
2817 list_for_each_entry(nh, rt6_nh_list, next) {
2818 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2819 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2820 nh->r_cfg.fc_ifindex);
2821 }
2822}
2823
2824static int ip6_route_info_append(struct list_head *rt6_nh_list,
2825 struct rt6_info *rt, struct fib6_config *r_cfg)
2826{
2827 struct rt6_nh *nh;
2828 struct rt6_info *rtnh;
2829 int err = -EEXIST;
2830
2831 list_for_each_entry(nh, rt6_nh_list, next) {
2832 /* check if rt6_info already exists */
2833 rtnh = nh->rt6_info;
2834
2835 if (rtnh->dst.dev == rt->dst.dev &&
2836 rtnh->rt6i_idev == rt->rt6i_idev &&
2837 ipv6_addr_equal(&rtnh->rt6i_gateway,
2838 &rt->rt6i_gateway))
2839 return err;
2840 }
2841
2842 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2843 if (!nh)
2844 return -ENOMEM;
2845 nh->rt6_info = rt;
2846 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2847 if (err) {
2848 kfree(nh);
2849 return err;
2850 }
2851 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2852 list_add_tail(&nh->next, rt6_nh_list);
2853
2854 return 0;
2855}
2856
2857static int ip6_route_multipath_add(struct fib6_config *cfg)
51ebd318
ND		 2858{
2859 struct fib6_config r_cfg;
2860 struct rtnexthop *rtnh;
6b9ea5a6
RP		 2861 struct rt6_info *rt;
2862 struct rt6_nh *err_nh;
2863 struct rt6_nh *nh, *nh_safe;
51ebd318
ND		 2864 int remaining;
2865 int attrlen;
6b9ea5a6
RP		 2866 int err = 1;
2867 int nhn = 0;
2868 int replace = (cfg->fc_nlinfo.nlh &&
2869 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2870 LIST_HEAD(rt6_nh_list);
51ebd318 2871
35f1b4e9 2872 remaining = cfg->fc_mp_len;
51ebd318 2873 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318 2874
6b9ea5a6
RP		 2875 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2876 * rt6_info structs per nexthop
2877 */
51ebd318
ND		 2878 while (rtnh_ok(rtnh, remaining)) {
2879 memcpy(&r_cfg, cfg, sizeof(*cfg));
2880 if (rtnh->rtnh_ifindex)
2881 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2882
2883 attrlen = rtnh_attrlen(rtnh);
2884 if (attrlen > 0) {
2885 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2886
2887 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2888 if (nla) {
67b61f6c 2889 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 2890 r_cfg.fc_flags |= RTF_GATEWAY;
2891 }
19e42e45
RP		 2892 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
2893 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
2894 if (nla)
2895 r_cfg.fc_encap_type = nla_get_u16(nla);
51ebd318 2896 }
6b9ea5a6 2897
8c5b83f0
RP		 2898 rt = ip6_route_info_create(&r_cfg);
2899 if (IS_ERR(rt)) {
2900 err = PTR_ERR(rt);
2901 rt = NULL;
6b9ea5a6 2902 goto cleanup;
8c5b83f0 2903 }
6b9ea5a6
RP		 2904
2905 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
51ebd318 2906 if (err) {
6b9ea5a6
RP		 2907 dst_free(&rt->dst);
2908 goto cleanup;
2909 }
2910
2911 rtnh = rtnh_next(rtnh, &remaining);
2912 }
2913
2914 err_nh = NULL;
2915 list_for_each_entry(nh, &rt6_nh_list, next) {
2916 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
2917 /* nh->rt6_info is used or freed at this point, reset to NULL*/
2918 nh->rt6_info = NULL;
2919 if (err) {
2920 if (replace && nhn)
2921 ip6_print_replace_route_err(&rt6_nh_list);
2922 err_nh = nh;
2923 goto add_errout;
51ebd318 2924 }
6b9ea5a6 2925
1a72418b 2926 /* Because each route is added like a single route we remove
27596472
MK		 2927 * these flags after the first nexthop: if there is a collision,
2928 * we have already failed to add the first nexthop:
2929 * fib6_add_rt2node() has rejected it; when replacing, old
2930 * nexthops have been replaced by first new, the rest should
2931 * be added to it.
1a72418b 2932 */
27596472
MK		 2933 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
2934 NLM_F_REPLACE);
6b9ea5a6
RP		 2935 nhn++;
2936 }
2937
2938 goto cleanup;
2939
2940add_errout:
2941 /* Delete routes that were already added */
2942 list_for_each_entry(nh, &rt6_nh_list, next) {
2943 if (err_nh == nh)
2944 break;
2945 ip6_route_del(&nh->r_cfg);
2946 }
2947
2948cleanup:
2949 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
2950 if (nh->rt6_info)
2951 dst_free(&nh->rt6_info->dst);
52fe51f8 2952 kfree(nh->mxc.mx);
6b9ea5a6
RP		 2953 list_del(&nh->next);
2954 kfree(nh);
2955 }
2956
2957 return err;
2958}
2959
2960static int ip6_route_multipath_del(struct fib6_config *cfg)
2961{
2962 struct fib6_config r_cfg;
2963 struct rtnexthop *rtnh;
2964 int remaining;
2965 int attrlen;
2966 int err = 1, last_err = 0;
2967
2968 remaining = cfg->fc_mp_len;
2969 rtnh = (struct rtnexthop *)cfg->fc_mp;
2970
2971 /* Parse a Multipath Entry */
2972 while (rtnh_ok(rtnh, remaining)) {
2973 memcpy(&r_cfg, cfg, sizeof(*cfg));
2974 if (rtnh->rtnh_ifindex)
2975 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2976
2977 attrlen = rtnh_attrlen(rtnh);
2978 if (attrlen > 0) {
2979 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2980
2981 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2982 if (nla) {
2983 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2984 r_cfg.fc_flags |= RTF_GATEWAY;
2985 }
2986 }
2987 err = ip6_route_del(&r_cfg);
2988 if (err)
2989 last_err = err;
2990
51ebd318
ND		 2991 rtnh = rtnh_next(rtnh, &remaining);
2992 }
2993
2994 return last_err;
2995}
2996
67ba4152 2997static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 2998{
86872cb5
TG		 2999 struct fib6_config cfg;
3000 int err;
1da177e4 3001
86872cb5
TG		 3002 err = rtm_to_fib6_config(skb, nlh, &cfg);
3003 if (err < 0)
3004 return err;
3005
51ebd318 3006 if (cfg.fc_mp)
6b9ea5a6 3007 return ip6_route_multipath_del(&cfg);
51ebd318
ND		 3008 else
3009 return ip6_route_del(&cfg);
1da177e4
LT		 3010}
3011
67ba4152 3012static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3013{
86872cb5
TG		 3014 struct fib6_config cfg;
3015 int err;
1da177e4 3016
86872cb5
TG		 3017 err = rtm_to_fib6_config(skb, nlh, &cfg);
3018 if (err < 0)
3019 return err;
3020
51ebd318 3021 if (cfg.fc_mp)
6b9ea5a6 3022 return ip6_route_multipath_add(&cfg);
51ebd318
ND		 3023 else
3024 return ip6_route_add(&cfg);
1da177e4
LT		 3025}
3026
19e42e45 3027static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
339bf98f
TG		 3028{
3029 return NLMSG_ALIGN(sizeof(struct rtmsg))
3030 + nla_total_size(16) /* RTA_SRC */
3031 + nla_total_size(16) /* RTA_DST */
3032 + nla_total_size(16) /* RTA_GATEWAY */
3033 + nla_total_size(16) /* RTA_PREFSRC */
3034 + nla_total_size(4) /* RTA_TABLE */
3035 + nla_total_size(4) /* RTA_IIF */
3036 + nla_total_size(4) /* RTA_OIF */
3037 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 3038 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 3039 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6 3040 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
19e42e45 3041 + nla_total_size(1) /* RTA_PREF */
61adedf3 3042 + lwtunnel_get_encap_size(rt->dst.lwtstate);
339bf98f
TG		 3043}
3044
191cd582
BH		 3045static int rt6_fill_node(struct net *net,
3046 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 3047 struct in6_addr *dst, struct in6_addr *src,
15e47304 3048 int iif, int type, u32 portid, u32 seq,
7bc570c8 3049 int prefix, int nowait, unsigned int flags)
1da177e4 3050{
4b32b5ad 3051 u32 metrics[RTAX_MAX];
1da177e4 3052 struct rtmsg *rtm;
2d7202bf 3053 struct nlmsghdr *nlh;
e3703b3d 3054 long expires;
9e762a4a 3055 u32 table;
1da177e4
LT		 3056
3057 if (prefix) { /* user wants prefix routes only */
3058 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3059 /* success since this is not a prefix route */
3060 return 1;
3061 }
3062 }
3063
15e47304 3064 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 3065 if (!nlh)
26932566 3066 return -EMSGSIZE;
2d7202bf
TG		 3067
3068 rtm = nlmsg_data(nlh);
1da177e4
LT		 3069 rtm->rtm_family = AF_INET6;
3070 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3071 rtm->rtm_src_len = rt->rt6i_src.plen;
3072 rtm->rtm_tos = 0;
c71099ac 3073 if (rt->rt6i_table)
9e762a4a 3074 table = rt->rt6i_table->tb6_id;
c71099ac 3075 else
9e762a4a
PM		 3076 table = RT6_TABLE_UNSPEC;
3077 rtm->rtm_table = table;
c78679e8
DM		 3078 if (nla_put_u32(skb, RTA_TABLE, table))
3079 goto nla_put_failure;
ef2c7d7b
ND		 3080 if (rt->rt6i_flags & RTF_REJECT) {
3081 switch (rt->dst.error) {
3082 case -EINVAL:
3083 rtm->rtm_type = RTN_BLACKHOLE;
3084 break;
3085 case -EACCES:
3086 rtm->rtm_type = RTN_PROHIBIT;
3087 break;
b4949ab2
ND		 3088 case -EAGAIN:
3089 rtm->rtm_type = RTN_THROW;
3090 break;
ef2c7d7b
ND		 3091 default:
3092 rtm->rtm_type = RTN_UNREACHABLE;
3093 break;
3094 }
3095 }
38308473 3096 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 3097 rtm->rtm_type = RTN_LOCAL;
d1918542 3098 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 3099 rtm->rtm_type = RTN_LOCAL;
3100 else
3101 rtm->rtm_type = RTN_UNICAST;
3102 rtm->rtm_flags = 0;
35103d11 3103 if (!netif_carrier_ok(rt->dst.dev)) {
cea45e20 3104 rtm->rtm_flags |= RTNH_F_LINKDOWN;
35103d11
AG		 3105 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3106 rtm->rtm_flags |= RTNH_F_DEAD;
3107 }
1da177e4
LT		 3108 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3109 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 3110 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 3111 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 3112 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3113 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3114 rtm->rtm_protocol = RTPROT_RA;
3115 else
3116 rtm->rtm_protocol = RTPROT_KERNEL;
3117 }
1da177e4 3118
38308473 3119 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 3120 rtm->rtm_flags |= RTM_F_CLONED;
3121
3122 if (dst) {
930345ea 3123 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 3124 goto nla_put_failure;
1ab1457c 3125 rtm->rtm_dst_len = 128;
1da177e4 3126 } else if (rtm->rtm_dst_len)
930345ea 3127 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 3128 goto nla_put_failure;
1da177e4
LT		 3129#ifdef CONFIG_IPV6_SUBTREES
3130 if (src) {
930345ea 3131 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 3132 goto nla_put_failure;
1ab1457c 3133 rtm->rtm_src_len = 128;
c78679e8 3134 } else if (rtm->rtm_src_len &&
930345ea 3135 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 3136 goto nla_put_failure;
1da177e4 3137#endif
7bc570c8
YH		 3138 if (iif) {
3139#ifdef CONFIG_IPV6_MROUTE
3140 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 3141 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 3142 if (err <= 0) {
3143 if (!nowait) {
3144 if (err == 0)
3145 return 0;
3146 goto nla_put_failure;
3147 } else {
3148 if (err == -EMSGSIZE)
3149 goto nla_put_failure;
3150 }
3151 }
3152 } else
3153#endif
c78679e8
DM		 3154 if (nla_put_u32(skb, RTA_IIF, iif))
3155 goto nla_put_failure;
7bc570c8 3156 } else if (dst) {
1da177e4 3157 struct in6_addr saddr_buf;
c78679e8 3158 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 3159 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3160 goto nla_put_failure;
1da177e4 3161 }
2d7202bf 3162
c3968a85
DW		 3163 if (rt->rt6i_prefsrc.plen) {
3164 struct in6_addr saddr_buf;
4e3fd7a0 3165 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 3166 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3167 goto nla_put_failure;
c3968a85
DW		 3168 }
3169
4b32b5ad
MKL		 3170 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3171 if (rt->rt6i_pmtu)
3172 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3173 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 3174 goto nla_put_failure;
3175
dd0cbf29 3176 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 3177 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 3178 goto nla_put_failure;
94f826b8 3179 }
2d7202bf 3180
c78679e8
DM		 3181 if (rt->dst.dev &&
3182 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3183 goto nla_put_failure;
3184 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3185 goto nla_put_failure;
8253947e
LW		 3186
3187 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 3188
87a50699 3189 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 3190 goto nla_put_failure;
2d7202bf 3191
c78ba6d6
LR		 3192 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3193 goto nla_put_failure;
3194
61adedf3 3195 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
19e42e45 3196
053c095a
JB		 3197 nlmsg_end(skb, nlh);
3198 return 0;
2d7202bf
TG		 3199
3200nla_put_failure:
26932566
PM		 3201 nlmsg_cancel(skb, nlh);
3202 return -EMSGSIZE;
1da177e4
LT		 3203}
3204
1b43af54 3205int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 3206{
3207 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3208 int prefix;
3209
2d7202bf
TG		 3210 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3211 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 3212 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3213 } else
3214 prefix = 0;
3215
191cd582
BH		 3216 return rt6_fill_node(arg->net,
3217 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 3218 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 3219 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 3220}
3221
67ba4152 3222static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 3223{
3b1e0a65 3224 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 3225 struct nlattr *tb[RTA_MAX+1];
3226 struct rt6_info *rt;
1da177e4 3227 struct sk_buff *skb;
ab364a6f 3228 struct rtmsg *rtm;
4c9483b2 3229 struct flowi6 fl6;
72331bc0 3230 int err, iif = 0, oif = 0;
1da177e4 3231
ab364a6f
TG		 3232 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3233 if (err < 0)
3234 goto errout;
1da177e4 3235
ab364a6f 3236 err = -EINVAL;
4c9483b2 3237 memset(&fl6, 0, sizeof(fl6));
1da177e4 3238
ab364a6f
TG		 3239 if (tb[RTA_SRC]) {
3240 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3241 goto errout;
3242
4e3fd7a0 3243 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 3244 }
3245
3246 if (tb[RTA_DST]) {
3247 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3248 goto errout;
3249
4e3fd7a0 3250 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 3251 }
3252
3253 if (tb[RTA_IIF])
3254 iif = nla_get_u32(tb[RTA_IIF]);
3255
3256 if (tb[RTA_OIF])
72331bc0 3257 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 3258
2e47b291
LC		 3259 if (tb[RTA_MARK])
3260 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3261
1da177e4
LT		 3262 if (iif) {
3263 struct net_device *dev;
72331bc0
SL		 3264 int flags = 0;
3265
5578689a 3266 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 3267 if (!dev) {
3268 err = -ENODEV;
ab364a6f 3269 goto errout;
1da177e4 3270 }
72331bc0
SL		 3271
3272 fl6.flowi6_iif = iif;
3273
3274 if (!ipv6_addr_any(&fl6.saddr))
3275 flags |= RT6_LOOKUP_F_HAS_SADDR;
3276
3277 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3278 flags);
3279 } else {
3280 fl6.flowi6_oif = oif;
3281
ca254490
DA		 3282 if (netif_index_is_l3_master(net, oif)) {
3283 fl6.flowi6_flags = FLOWI_FLAG_L3MDEV_SRC |
3284 FLOWI_FLAG_SKIP_NH_OIF;
3285 }
3286
72331bc0 3287 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 3288 }
3289
ab364a6f 3290 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 3291 if (!skb) {
94e187c0 3292 ip6_rt_put(rt);
ab364a6f
TG		 3293 err = -ENOBUFS;
3294 goto errout;
3295 }
1da177e4 3296
ab364a6f
TG		 3297 /* Reserve room for dummy headers, this skb can pass
3298 through good chunk of routing engine.
3299 */
459a98ed 3300 skb_reset_mac_header(skb);
ab364a6f 3301 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 3302
d8d1f30b 3303 skb_dst_set(skb, &rt->dst);
1da177e4 3304
4c9483b2 3305 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 3306 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 3307 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 3308 if (err < 0) {
ab364a6f
TG		 3309 kfree_skb(skb);
3310 goto errout;
1da177e4
LT		 3311 }
3312
15e47304 3313 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 3314errout:
1da177e4 3315 return err;
1da177e4
LT		 3316}
3317
37a1d361
RP		 3318void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3319 unsigned int nlm_flags)
1da177e4
LT		 3320{
3321 struct sk_buff *skb;
5578689a 3322 struct net *net = info->nl_net;
528c4ceb
DL		 3323 u32 seq;
3324 int err;
3325
3326 err = -ENOBUFS;
38308473 3327 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 3328
19e42e45 3329 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
38308473 3330 if (!skb)
21713ebc
TG		 3331 goto errout;
3332
191cd582 3333 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
37a1d361 3334 event, info->portid, seq, 0, 0, nlm_flags);
26932566
PM		 3335 if (err < 0) {
3336 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3337 WARN_ON(err == -EMSGSIZE);
3338 kfree_skb(skb);
3339 goto errout;
3340 }
15e47304 3341 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 3342 info->nlh, gfp_any());
3343 return;
21713ebc
TG		 3344errout:
3345 if (err < 0)
5578689a 3346 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 3347}
3348
8ed67789 3349static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 3350 unsigned long event, void *ptr)
8ed67789 3351{
351638e7 3352 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 3353 struct net *net = dev_net(dev);
8ed67789
DL		 3354
3355 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 3356 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 3357 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3358#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3359 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 3360 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 3361 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 3362 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3363#endif
3364 }
3365
3366 return NOTIFY_OK;
3367}
3368
1da177e4
LT		 3369/*
3370 * /proc
3371 */
3372
3373#ifdef CONFIG_PROC_FS
3374
33120b30
AD		 3375static const struct file_operations ipv6_route_proc_fops = {
3376 .owner = THIS_MODULE,
3377 .open = ipv6_route_open,
3378 .read = seq_read,
3379 .llseek = seq_lseek,
8d2ca1d7 3380 .release = seq_release_net,
33120b30
AD		 3381};
3382
1da177e4
LT		 3383static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3384{
69ddb805 3385 struct net *net = (struct net *)seq->private;
1da177e4 3386 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 3387 net->ipv6.rt6_stats->fib_nodes,
3388 net->ipv6.rt6_stats->fib_route_nodes,
3389 net->ipv6.rt6_stats->fib_rt_alloc,
3390 net->ipv6.rt6_stats->fib_rt_entries,
3391 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 3392 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 3393 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 3394
3395 return 0;
3396}
3397
3398static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3399{
de05c557 3400 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 3401}
3402
9a32144e 3403static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 3404 .owner = THIS_MODULE,
3405 .open = rt6_stats_seq_open,
3406 .read = seq_read,
3407 .llseek = seq_lseek,
b6fcbdb4 3408 .release = single_release_net,
1da177e4
LT		 3409};
3410#endif /* CONFIG_PROC_FS */
3411
3412#ifdef CONFIG_SYSCTL
3413
1da177e4 3414static
fe2c6338 3415int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3416 void __user *buffer, size_t *lenp, loff_t *ppos)
3417{
c486da34
LAG		 3418 struct net *net;
3419 int delay;
3420 if (!write)
1da177e4 3421 return -EINVAL;
c486da34
LAG		 3422
3423 net = (struct net *)ctl->extra1;
3424 delay = net->ipv6.sysctl.flush_delay;
3425 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3426 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3427 return 0;
1da177e4
LT		 3428}
3429
fe2c6338 3430struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3431 {
1da177e4 3432 .procname = "flush",
4990509f 3433 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3434 .maxlen = sizeof(int),
89c8b3a1 3435 .mode = 0200,
6d9f239a 3436 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3437 },
3438 {
1da177e4 3439 .procname = "gc_thresh",
9a7ec3a9 3440 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3441 .maxlen = sizeof(int),
3442 .mode = 0644,
6d9f239a 3443 .proc_handler = proc_dointvec,
1da177e4
LT		 3444 },
3445 {
1da177e4 3446 .procname = "max_size",
4990509f 3447 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3448 .maxlen = sizeof(int),
3449 .mode = 0644,
6d9f239a 3450 .proc_handler = proc_dointvec,
1da177e4
LT		 3451 },
3452 {
1da177e4 3453 .procname = "gc_min_interval",
4990509f 3454 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3455 .maxlen = sizeof(int),
3456 .mode = 0644,
6d9f239a 3457 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3458 },
3459 {
1da177e4 3460 .procname = "gc_timeout",
4990509f 3461 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3462 .maxlen = sizeof(int),
3463 .mode = 0644,
6d9f239a 3464 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3465 },
3466 {
1da177e4 3467 .procname = "gc_interval",
4990509f 3468 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3469 .maxlen = sizeof(int),
3470 .mode = 0644,
6d9f239a 3471 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3472 },
3473 {
1da177e4 3474 .procname = "gc_elasticity",
4990509f 3475 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3476 .maxlen = sizeof(int),
3477 .mode = 0644,
f3d3f616 3478 .proc_handler = proc_dointvec,
1da177e4
LT		 3479 },
3480 {
1da177e4 3481 .procname = "mtu_expires",
4990509f 3482 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3483 .maxlen = sizeof(int),
3484 .mode = 0644,
6d9f239a 3485 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3486 },
3487 {
1da177e4 3488 .procname = "min_adv_mss",
4990509f 3489 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3490 .maxlen = sizeof(int),
3491 .mode = 0644,
f3d3f616 3492 .proc_handler = proc_dointvec,
1da177e4
LT		 3493 },
3494 {
1da177e4 3495 .procname = "gc_min_interval_ms",
4990509f 3496 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3497 .maxlen = sizeof(int),
3498 .mode = 0644,
6d9f239a 3499 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3500 },
f8572d8f 3501 { }
1da177e4
LT		 3502};
3503
2c8c1e72 3504struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3505{
3506 struct ctl_table *table;
3507
3508 table = kmemdup(ipv6_route_table_template,
3509 sizeof(ipv6_route_table_template),
3510 GFP_KERNEL);
5ee09105
YH		 3511
3512 if (table) {
3513 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3514 table[0].extra1 = net;
86393e52 3515 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3516 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3517 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3518 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3519 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3520 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3521 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3522 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3523 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3524
3525 /* Don't export sysctls to unprivileged users */
3526 if (net->user_ns != &init_user_ns)
3527 table[0].procname = NULL;
5ee09105
YH		 3528 }
3529
760f2d01
DL		 3530 return table;
3531}
1da177e4
LT		 3532#endif
3533
2c8c1e72 3534static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3535{
633d424b 3536 int ret = -ENOMEM;
8ed67789 3537
86393e52
AD		 3538 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3539 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3540
fc66f95c
ED		 3541 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3542 goto out_ip6_dst_ops;
3543
8ed67789
DL		 3544 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3545 sizeof(*net->ipv6.ip6_null_entry),
3546 GFP_KERNEL);
3547 if (!net->ipv6.ip6_null_entry)
fc66f95c 3548 goto out_ip6_dst_entries;
d8d1f30b 3549 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3550 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3551 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3552 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3553 ip6_template_metrics, true);
8ed67789
DL		 3554
3555#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3556 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3557 sizeof(*net->ipv6.ip6_prohibit_entry),
3558 GFP_KERNEL);
68fffc67
PZ		 3559 if (!net->ipv6.ip6_prohibit_entry)
3560 goto out_ip6_null_entry;
d8d1f30b 3561 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3562 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3563 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3564 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3565 ip6_template_metrics, true);
8ed67789
DL		 3566
3567 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3568 sizeof(*net->ipv6.ip6_blk_hole_entry),
3569 GFP_KERNEL);
68fffc67
PZ		 3570 if (!net->ipv6.ip6_blk_hole_entry)
3571 goto out_ip6_prohibit_entry;
d8d1f30b 3572 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3573 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3574 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3575 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3576 ip6_template_metrics, true);
8ed67789
DL		 3577#endif
3578
b339a47c
PZ		 3579 net->ipv6.sysctl.flush_delay = 0;
3580 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3581 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3582 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3583 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3584 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3585 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3586 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3587
6891a346
BT		 3588 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3589
8ed67789
DL		 3590 ret = 0;
3591out:
3592 return ret;
f2fc6a54 3593
68fffc67
PZ		 3594#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3595out_ip6_prohibit_entry:
3596 kfree(net->ipv6.ip6_prohibit_entry);
3597out_ip6_null_entry:
3598 kfree(net->ipv6.ip6_null_entry);
3599#endif
fc66f95c
ED		 3600out_ip6_dst_entries:
3601 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3602out_ip6_dst_ops:
f2fc6a54 3603 goto out;
cdb18761
DL		 3604}
3605
2c8c1e72 3606static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3607{
8ed67789
DL		 3608 kfree(net->ipv6.ip6_null_entry);
3609#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3610 kfree(net->ipv6.ip6_prohibit_entry);
3611 kfree(net->ipv6.ip6_blk_hole_entry);
3612#endif
41bb78b4 3613 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3614}
3615
d189634e
TG		 3616static int __net_init ip6_route_net_init_late(struct net *net)
3617{
3618#ifdef CONFIG_PROC_FS
d4beaa66
G		 3619 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3620 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3621#endif
3622 return 0;
3623}
3624
3625static void __net_exit ip6_route_net_exit_late(struct net *net)
3626{
3627#ifdef CONFIG_PROC_FS
ece31ffd
G		 3628 remove_proc_entry("ipv6_route", net->proc_net);
3629 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3630#endif
3631}
3632
cdb18761
DL		 3633static struct pernet_operations ip6_route_net_ops = {
3634 .init = ip6_route_net_init,
3635 .exit = ip6_route_net_exit,
3636};
3637
c3426b47
DM		 3638static int __net_init ipv6_inetpeer_init(struct net *net)
3639{
3640 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3641
3642 if (!bp)
3643 return -ENOMEM;
3644 inet_peer_base_init(bp);
3645 net->ipv6.peers = bp;
3646 return 0;
3647}
3648
3649static void __net_exit ipv6_inetpeer_exit(struct net *net)
3650{
3651 struct inet_peer_base *bp = net->ipv6.peers;
3652
3653 net->ipv6.peers = NULL;
56a6b248 3654 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3655 kfree(bp);
3656}
3657
2b823f72 3658static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3659 .init = ipv6_inetpeer_init,
3660 .exit = ipv6_inetpeer_exit,
3661};
3662
d189634e
TG		 3663static struct pernet_operations ip6_route_net_late_ops = {
3664 .init = ip6_route_net_init_late,
3665 .exit = ip6_route_net_exit_late,
3666};
3667
8ed67789
DL		 3668static struct notifier_block ip6_route_dev_notifier = {
3669 .notifier_call = ip6_route_dev_notify,
3670 .priority = 0,
3671};
3672
433d49c3 3673int __init ip6_route_init(void)
1da177e4 3674{
433d49c3 3675 int ret;
8d0b94af 3676 int cpu;
433d49c3 3677
9a7ec3a9
DL		 3678 ret = -ENOMEM;
3679 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3680 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3681 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3682 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3683 goto out;
14e50e57 3684
fc66f95c 3685 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3686 if (ret)
bdb3289f 3687 goto out_kmem_cache;
bdb3289f 3688
c3426b47
DM		 3689 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3690 if (ret)
e8803b6c 3691 goto out_dst_entries;
2a0c451a 3692
7e52b33b
DM		 3693 ret = register_pernet_subsys(&ip6_route_net_ops);
3694 if (ret)
3695 goto out_register_inetpeer;
c3426b47 3696
5dc121e9
AE		 3697 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3698
8ed67789
DL		 3699 /* Registering of the loopback is done before this portion of code,
3700 * the loopback reference in rt6_info will not be taken, do it
3701 * manually for init_net */
d8d1f30b 3702 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3703 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3704 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3705 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3706 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3707 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3708 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3709 #endif
e8803b6c 3710 ret = fib6_init();
433d49c3 3711 if (ret)
8ed67789 3712 goto out_register_subsys;
433d49c3 3713
433d49c3
DL		 3714 ret = xfrm6_init();
3715 if (ret)
e8803b6c 3716 goto out_fib6_init;
c35b7e72 3717
433d49c3
DL		 3718 ret = fib6_rules_init();
3719 if (ret)
3720 goto xfrm6_init;
7e5449c2 3721
d189634e
TG		 3722 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3723 if (ret)
3724 goto fib6_rules_init;
3725
433d49c3 3726 ret = -ENOBUFS;
c7ac8679
GR		 3727 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3728 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3729 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3730 goto out_register_late_subsys;
c127ea2c 3731
8ed67789 3732 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3733 if (ret)
d189634e 3734 goto out_register_late_subsys;
8ed67789 3735
8d0b94af
MKL		 3736 for_each_possible_cpu(cpu) {
3737 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3738
3739 INIT_LIST_HEAD(&ul->head);
3740 spin_lock_init(&ul->lock);
3741 }
3742
433d49c3
DL		 3743out:
3744 return ret;
3745
d189634e
TG		 3746out_register_late_subsys:
3747 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3748fib6_rules_init:
433d49c3
DL		 3749 fib6_rules_cleanup();
3750xfrm6_init:
433d49c3 3751 xfrm6_fini();
2a0c451a
TG		 3752out_fib6_init:
3753 fib6_gc_cleanup();
8ed67789
DL		 3754out_register_subsys:
3755 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3756out_register_inetpeer:
3757 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3758out_dst_entries:
3759 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3760out_kmem_cache:
f2fc6a54 3761 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3762 goto out;
1da177e4
LT		 3763}
3764
3765void ip6_route_cleanup(void)
3766{
8ed67789 3767 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3768 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3769 fib6_rules_cleanup();
1da177e4 3770 xfrm6_fini();
1da177e4 3771 fib6_gc_cleanup();
c3426b47 3772 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3773 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3774 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3775 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3776}
Linux 6.x block layer and io_uring tree(s)