projects / linux-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
net: core: add UID to flows, rules, and routes
[linux-block.git] / net / ipv6 / route.c
CommitLineData
1da177e4
LT		 1/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4 7 *
1da177e4
LT		 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13
14/* Changes:
15 *
16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection.
18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list.
c0bece9f
YH		 23 * Ville Nuorvala
24 * Fixed routing subtrees.
1da177e4
LT		 25 */
26
f3213831
JP		 27#define pr_fmt(fmt) "IPv6: " fmt
28
4fc268d2 29#include <linux/capability.h>
1da177e4 30#include <linux/errno.h>
bc3b2d7f 31#include <linux/export.h>
1da177e4
LT		 32#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
7bc570c8 40#include <linux/mroute6.h>
1da177e4 41#include <linux/init.h>
1da177e4 42#include <linux/if_arp.h>
1da177e4
LT		 43#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
5b7c931d 45#include <linux/nsproxy.h>
5a0e3ad6 46#include <linux/slab.h>
457c4cbc 47#include <net/net_namespace.h>
1da177e4
LT		 48#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
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>
b811580d 65#include <trace/events/fib6.h>
1da177e4
LT		 66
67#include <asm/uaccess.h>
68
69#ifdef CONFIG_SYSCTL
70#include <linux/sysctl.h>
71#endif
72
afc154e9 73enum rt6_nud_state {
7e980569
JB		 74 RT6_NUD_FAIL_HARD = -3,
75 RT6_NUD_FAIL_PROBE = -2,
76 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 77 RT6_NUD_SUCCEED = 1
78};
79
83a09abd 80static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort);
1da177e4 81static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 82static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 83static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 84static struct dst_entry *ip6_negative_advice(struct dst_entry *);
85static void ip6_dst_destroy(struct dst_entry *);
86static void ip6_dst_ifdown(struct dst_entry *,
87 struct net_device *dev, int how);
569d3645 88static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 89
90static int ip6_pkt_discard(struct sk_buff *skb);
ede2059d 91static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb);
7150aede 92static int ip6_pkt_prohibit(struct sk_buff *skb);
ede2059d 93static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1da177e4 94static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 95static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
96 struct sk_buff *skb, u32 mtu);
97static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
98 struct sk_buff *skb);
4b32b5ad 99static void rt6_dst_from_metrics_check(struct rt6_info *rt);
52bd4c0c 100static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 101
70ceb4f5 102#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 103static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42 104 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 105 const struct in6_addr *gwaddr,
106 struct net_device *dev,
95c96174 107 unsigned int pref);
efa2cea0 108static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42 109 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 110 const struct in6_addr *gwaddr,
111 struct net_device *dev);
70ceb4f5
YH		 112#endif
113
8d0b94af
MKL		 114struct uncached_list {
115 spinlock_t lock;
116 struct list_head head;
117};
118
119static DEFINE_PER_CPU_ALIGNED(struct uncached_list, rt6_uncached_list);
120
121static void rt6_uncached_list_add(struct rt6_info *rt)
122{
123 struct uncached_list *ul = raw_cpu_ptr(&rt6_uncached_list);
124
125 rt->dst.flags |= DST_NOCACHE;
126 rt->rt6i_uncached_list = ul;
127
128 spin_lock_bh(&ul->lock);
129 list_add_tail(&rt->rt6i_uncached, &ul->head);
130 spin_unlock_bh(&ul->lock);
131}
132
133static void rt6_uncached_list_del(struct rt6_info *rt)
134{
135 if (!list_empty(&rt->rt6i_uncached)) {
136 struct uncached_list *ul = rt->rt6i_uncached_list;
137
138 spin_lock_bh(&ul->lock);
139 list_del(&rt->rt6i_uncached);
140 spin_unlock_bh(&ul->lock);
141 }
142}
143
144static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
145{
146 struct net_device *loopback_dev = net->loopback_dev;
147 int cpu;
148
e332bc67
EB		 149 if (dev == loopback_dev)
150 return;
151
8d0b94af
MKL		 152 for_each_possible_cpu(cpu) {
153 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
154 struct rt6_info *rt;
155
156 spin_lock_bh(&ul->lock);
157 list_for_each_entry(rt, &ul->head, rt6i_uncached) {
158 struct inet6_dev *rt_idev = rt->rt6i_idev;
159 struct net_device *rt_dev = rt->dst.dev;
160
e332bc67 161 if (rt_idev->dev == dev) {
8d0b94af
MKL		 162 rt->rt6i_idev = in6_dev_get(loopback_dev);
163 in6_dev_put(rt_idev);
164 }
165
e332bc67 166 if (rt_dev == dev) {
8d0b94af
MKL		 167 rt->dst.dev = loopback_dev;
168 dev_hold(rt->dst.dev);
169 dev_put(rt_dev);
170 }
171 }
172 spin_unlock_bh(&ul->lock);
173 }
174}
175
d52d3997
MKL		 176static u32 *rt6_pcpu_cow_metrics(struct rt6_info *rt)
177{
178 return dst_metrics_write_ptr(rt->dst.from);
179}
180
06582540
DM		 181static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
182{
4b32b5ad 183 struct rt6_info *rt = (struct rt6_info *)dst;
06582540 184
d52d3997
MKL		 185 if (rt->rt6i_flags & RTF_PCPU)
186 return rt6_pcpu_cow_metrics(rt);
187 else if (rt->rt6i_flags & RTF_CACHE)
4b32b5ad
MKL		 188 return NULL;
189 else
3b471175 190 return dst_cow_metrics_generic(dst, old);
06582540
DM		 191}
192
f894cbf8
DM		 193static inline const void *choose_neigh_daddr(struct rt6_info *rt,
194 struct sk_buff *skb,
195 const void *daddr)
39232973
DM		 196{
197 struct in6_addr *p = &rt->rt6i_gateway;
198
a7563f34 199 if (!ipv6_addr_any(p))
39232973 200 return (const void *) p;
f894cbf8
DM		 201 else if (skb)
202 return &ipv6_hdr(skb)->daddr;
39232973
DM		 203 return daddr;
204}
205
f894cbf8
DM		 206static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
207 struct sk_buff *skb,
208 const void *daddr)
d3aaeb38 209{
39232973
DM		 210 struct rt6_info *rt = (struct rt6_info *) dst;
211 struct neighbour *n;
212
f894cbf8 213 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 214 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 215 if (n)
216 return n;
217 return neigh_create(&nd_tbl, daddr, dst->dev);
218}
219
9a7ec3a9 220static struct dst_ops ip6_dst_ops_template = {
1da177e4 221 .family = AF_INET6,
1da177e4
LT		 222 .gc = ip6_dst_gc,
223 .gc_thresh = 1024,
224 .check = ip6_dst_check,
0dbaee3b 225 .default_advmss = ip6_default_advmss,
ebb762f2 226 .mtu = ip6_mtu,
06582540 227 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 228 .destroy = ip6_dst_destroy,
229 .ifdown = ip6_dst_ifdown,
230 .negative_advice = ip6_negative_advice,
231 .link_failure = ip6_link_failure,
232 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 233 .redirect = rt6_do_redirect,
9f8955cc 234 .local_out = __ip6_local_out,
d3aaeb38 235 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 236};
237
ebb762f2 238static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 239{
618f9bc7
SK		 240 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
241
242 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 243}
244
6700c270
DM		 245static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
246 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 247{
248}
249
6700c270
DM		 250static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
251 struct sk_buff *skb)
b587ee3b
DM		 252{
253}
254
14e50e57
DM		 255static struct dst_ops ip6_dst_blackhole_ops = {
256 .family = AF_INET6,
14e50e57
DM		 257 .destroy = ip6_dst_destroy,
258 .check = ip6_dst_check,
ebb762f2 259 .mtu = ip6_blackhole_mtu,
214f45c9 260 .default_advmss = ip6_default_advmss,
14e50e57 261 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 262 .redirect = ip6_rt_blackhole_redirect,
0a1f5962 263 .cow_metrics = dst_cow_metrics_generic,
d3aaeb38 264 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 265};
266
62fa8a84 267static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 268 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 269};
270
fb0af4c7 271static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 272 .dst = {
273 .__refcnt = ATOMIC_INIT(1),
274 .__use = 1,
2c20cbd7 275 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 276 .error = -ENETUNREACH,
d8d1f30b
CG		 277 .input = ip6_pkt_discard,
278 .output = ip6_pkt_discard_out,
1da177e4
LT		 279 },
280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 281 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 282 .rt6i_metric = ~(u32) 0,
283 .rt6i_ref = ATOMIC_INIT(1),
284};
285
101367c2
TG		 286#ifdef CONFIG_IPV6_MULTIPLE_TABLES
287
fb0af4c7 288static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 289 .dst = {
290 .__refcnt = ATOMIC_INIT(1),
291 .__use = 1,
2c20cbd7 292 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 293 .error = -EACCES,
d8d1f30b
CG		 294 .input = ip6_pkt_prohibit,
295 .output = ip6_pkt_prohibit_out,
101367c2
TG		 296 },
297 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 298 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 299 .rt6i_metric = ~(u32) 0,
300 .rt6i_ref = ATOMIC_INIT(1),
301};
302
fb0af4c7 303static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 304 .dst = {
305 .__refcnt = ATOMIC_INIT(1),
306 .__use = 1,
2c20cbd7 307 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 308 .error = -EINVAL,
d8d1f30b 309 .input = dst_discard,
ede2059d 310 .output = dst_discard_out,
101367c2
TG		 311 },
312 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 313 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 314 .rt6i_metric = ~(u32) 0,
315 .rt6i_ref = ATOMIC_INIT(1),
316};
317
318#endif
319
ebfa45f0
MKL		 320static void rt6_info_init(struct rt6_info *rt)
321{
322 struct dst_entry *dst = &rt->dst;
323
324 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
325 INIT_LIST_HEAD(&rt->rt6i_siblings);
326 INIT_LIST_HEAD(&rt->rt6i_uncached);
327}
328
1da177e4 329/* allocate dst with ip6_dst_ops */
d52d3997
MKL		 330static struct rt6_info *__ip6_dst_alloc(struct net *net,
331 struct net_device *dev,
ad706862 332 int flags)
1da177e4 333{
97bab73f 334 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 335 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 336
ebfa45f0
MKL		 337 if (rt)
338 rt6_info_init(rt);
8104891b 339
cf911662 340 return rt;
1da177e4
LT		 341}
342
9ab179d8
DA		 343struct rt6_info *ip6_dst_alloc(struct net *net,
344 struct net_device *dev,
345 int flags)
d52d3997 346{
ad706862 347 struct rt6_info *rt = __ip6_dst_alloc(net, dev, flags);
d52d3997
MKL		 348
349 if (rt) {
350 rt->rt6i_pcpu = alloc_percpu_gfp(struct rt6_info *, GFP_ATOMIC);
351 if (rt->rt6i_pcpu) {
352 int cpu;
353
354 for_each_possible_cpu(cpu) {
355 struct rt6_info **p;
356
357 p = per_cpu_ptr(rt->rt6i_pcpu, cpu);
358 /* no one shares rt */
359 *p = NULL;
360 }
361 } else {
362 dst_destroy((struct dst_entry *)rt);
363 return NULL;
364 }
365 }
366
367 return rt;
368}
9ab179d8 369EXPORT_SYMBOL(ip6_dst_alloc);
d52d3997 370
1da177e4
LT		 371static void ip6_dst_destroy(struct dst_entry *dst)
372{
373 struct rt6_info *rt = (struct rt6_info *)dst;
ecd98837 374 struct dst_entry *from = dst->from;
8d0b94af 375 struct inet6_dev *idev;
1da177e4 376
4b32b5ad 377 dst_destroy_metrics_generic(dst);
87775312 378 free_percpu(rt->rt6i_pcpu);
8d0b94af
MKL		 379 rt6_uncached_list_del(rt);
380
381 idev = rt->rt6i_idev;
38308473 382 if (idev) {
1da177e4
LT		 383 rt->rt6i_idev = NULL;
384 in6_dev_put(idev);
1ab1457c 385 }
1716a961 386
ecd98837
YH		 387 dst->from = NULL;
388 dst_release(from);
b3419363
DM		 389}
390
1da177e4
LT		 391static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
392 int how)
393{
394 struct rt6_info *rt = (struct rt6_info *)dst;
395 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 396 struct net_device *loopback_dev =
c346dca1 397 dev_net(dev)->loopback_dev;
1da177e4 398
97cac082
DM		 399 if (dev != loopback_dev) {
400 if (idev && idev->dev == dev) {
401 struct inet6_dev *loopback_idev =
402 in6_dev_get(loopback_dev);
403 if (loopback_idev) {
404 rt->rt6i_idev = loopback_idev;
405 in6_dev_put(idev);
406 }
407 }
1da177e4
LT		 408 }
409}
410
5973fb1e
MKL		 411static bool __rt6_check_expired(const struct rt6_info *rt)
412{
413 if (rt->rt6i_flags & RTF_EXPIRES)
414 return time_after(jiffies, rt->dst.expires);
415 else
416 return false;
417}
418
a50feda5 419static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 420{
1716a961
G		 421 if (rt->rt6i_flags & RTF_EXPIRES) {
422 if (time_after(jiffies, rt->dst.expires))
a50feda5 423 return true;
1716a961 424 } else if (rt->dst.from) {
3fd91fb3 425 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 426 }
a50feda5 427 return false;
1da177e4
LT		 428}
429
51ebd318
ND		 430/* Multipath route selection:
431 * Hash based function using packet header and flowlabel.
432 * Adapted from fib_info_hashfn()
433 */
434static int rt6_info_hash_nhsfn(unsigned int candidate_count,
435 const struct flowi6 *fl6)
436{
644d0e65 437 return get_hash_from_flowi6(fl6) % candidate_count;
51ebd318
ND		 438}
439
440static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 441 struct flowi6 *fl6, int oif,
442 int strict)
51ebd318
ND		 443{
444 struct rt6_info *sibling, *next_sibling;
445 int route_choosen;
446
447 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
448 /* Don't change the route, if route_choosen == 0
449 * (siblings does not include ourself)
450 */
451 if (route_choosen)
452 list_for_each_entry_safe(sibling, next_sibling,
453 &match->rt6i_siblings, rt6i_siblings) {
454 route_choosen--;
455 if (route_choosen == 0) {
52bd4c0c
ND		 456 if (rt6_score_route(sibling, oif, strict) < 0)
457 break;
51ebd318
ND		 458 match = sibling;
459 break;
460 }
461 }
462 return match;
463}
464
1da177e4 465/*
c71099ac 466 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 467 */
468
8ed67789
DL		 469static inline struct rt6_info *rt6_device_match(struct net *net,
470 struct rt6_info *rt,
b71d1d42 471 const struct in6_addr *saddr,
1da177e4 472 int oif,
d420895e 473 int flags)
1da177e4
LT		 474{
475 struct rt6_info *local = NULL;
476 struct rt6_info *sprt;
477
dd3abc4e
YH		 478 if (!oif && ipv6_addr_any(saddr))
479 goto out;
480
d8d1f30b 481 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 482 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 483
484 if (oif) {
1da177e4
LT		 485 if (dev->ifindex == oif)
486 return sprt;
487 if (dev->flags & IFF_LOOPBACK) {
38308473 488 if (!sprt->rt6i_idev ||
1da177e4 489 sprt->rt6i_idev->dev->ifindex != oif) {
17fb0b2b 490 if (flags & RT6_LOOKUP_F_IFACE)
1da177e4 491 continue;
17fb0b2b
DA		 492 if (local &&
493 local->rt6i_idev->dev->ifindex == oif)
1da177e4
LT		 494 continue;
495 }
496 local = sprt;
497 }
dd3abc4e
YH		 498 } else {
499 if (ipv6_chk_addr(net, saddr, dev,
500 flags & RT6_LOOKUP_F_IFACE))
501 return sprt;
1da177e4 502 }
dd3abc4e 503 }
1da177e4 504
dd3abc4e 505 if (oif) {
1da177e4
LT		 506 if (local)
507 return local;
508
d420895e 509 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 510 return net->ipv6.ip6_null_entry;
1da177e4 511 }
dd3abc4e 512out:
1da177e4
LT		 513 return rt;
514}
515
27097255 516#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 517struct __rt6_probe_work {
518 struct work_struct work;
519 struct in6_addr target;
520 struct net_device *dev;
521};
522
523static void rt6_probe_deferred(struct work_struct *w)
524{
525 struct in6_addr mcaddr;
526 struct __rt6_probe_work *work =
527 container_of(w, struct __rt6_probe_work, work);
528
529 addrconf_addr_solict_mult(&work->target, &mcaddr);
304d888b 530 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL);
c2f17e82 531 dev_put(work->dev);
662f5533 532 kfree(work);
c2f17e82
HFS		 533}
534
27097255
YH		 535static void rt6_probe(struct rt6_info *rt)
536{
990edb42 537 struct __rt6_probe_work *work;
f2c31e32 538 struct neighbour *neigh;
27097255
YH		 539 /*
540 * Okay, this does not seem to be appropriate
541 * for now, however, we need to check if it
542 * is really so; aka Router Reachability Probing.
543 *
544 * Router Reachability Probe MUST be rate-limited
545 * to no more than one per minute.
546 */
2152caea 547 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 548 return;
2152caea
YH		 549 rcu_read_lock_bh();
550 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
551 if (neigh) {
8d6c31bf
MKL		 552 if (neigh->nud_state & NUD_VALID)
553 goto out;
554
990edb42 555 work = NULL;
2152caea 556 write_lock(&neigh->lock);
990edb42
MKL		 557 if (!(neigh->nud_state & NUD_VALID) &&
558 time_after(jiffies,
559 neigh->updated +
560 rt->rt6i_idev->cnf.rtr_probe_interval)) {
561 work = kmalloc(sizeof(*work), GFP_ATOMIC);
562 if (work)
563 __neigh_set_probe_once(neigh);
c2f17e82 564 }
2152caea 565 write_unlock(&neigh->lock);
990edb42
MKL		 566 } else {
567 work = kmalloc(sizeof(*work), GFP_ATOMIC);
f2c31e32 568 }
990edb42
MKL		 569
570 if (work) {
571 INIT_WORK(&work->work, rt6_probe_deferred);
572 work->target = rt->rt6i_gateway;
573 dev_hold(rt->dst.dev);
574 work->dev = rt->dst.dev;
575 schedule_work(&work->work);
576 }
577
8d6c31bf 578out:
2152caea 579 rcu_read_unlock_bh();
27097255
YH		 580}
581#else
582static inline void rt6_probe(struct rt6_info *rt)
583{
27097255
YH		 584}
585#endif
586
1da177e4 587/*
554cfb7e 588 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 589 */
b6f99a21 590static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 591{
d1918542 592 struct net_device *dev = rt->dst.dev;
161980f4 593 if (!oif || dev->ifindex == oif)
554cfb7e 594 return 2;
161980f4
DM		 595 if ((dev->flags & IFF_LOOPBACK) &&
596 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
597 return 1;
598 return 0;
554cfb7e 599}
1da177e4 600
afc154e9 601static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 602{
f2c31e32 603 struct neighbour *neigh;
afc154e9 604 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 605
4d0c5911
YH		 606 if (rt->rt6i_flags & RTF_NONEXTHOP ||
607 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 608 return RT6_NUD_SUCCEED;
145a3621
YH		 609
610 rcu_read_lock_bh();
611 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
612 if (neigh) {
613 read_lock(&neigh->lock);
554cfb7e 614 if (neigh->nud_state & NUD_VALID)
afc154e9 615 ret = RT6_NUD_SUCCEED;
398bcbeb 616#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 617 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 618 ret = RT6_NUD_SUCCEED;
7e980569
JB		 619 else
620 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 621#endif
145a3621 622 read_unlock(&neigh->lock);
afc154e9
HFS		 623 } else {
624 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 625 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 626 }
145a3621
YH		 627 rcu_read_unlock_bh();
628
a5a81f0b 629 return ret;
1da177e4
LT		 630}
631
554cfb7e
YH		 632static int rt6_score_route(struct rt6_info *rt, int oif,
633 int strict)
1da177e4 634{
a5a81f0b 635 int m;
1ab1457c 636
4d0c5911 637 m = rt6_check_dev(rt, oif);
77d16f45 638 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 639 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 640#ifdef CONFIG_IPV6_ROUTER_PREF
641 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
642#endif
afc154e9
HFS		 643 if (strict & RT6_LOOKUP_F_REACHABLE) {
644 int n = rt6_check_neigh(rt);
645 if (n < 0)
646 return n;
647 }
554cfb7e
YH		 648 return m;
649}
650
f11e6659 651static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 652 int *mpri, struct rt6_info *match,
653 bool *do_rr)
554cfb7e 654{
f11e6659 655 int m;
afc154e9 656 bool match_do_rr = false;
35103d11
AG		 657 struct inet6_dev *idev = rt->rt6i_idev;
658 struct net_device *dev = rt->dst.dev;
659
660 if (dev && !netif_carrier_ok(dev) &&
d5d32e4b
DA		 661 idev->cnf.ignore_routes_with_linkdown &&
662 !(strict & RT6_LOOKUP_F_IGNORE_LINKSTATE))
35103d11 663 goto out;
f11e6659
DM		 664
665 if (rt6_check_expired(rt))
666 goto out;
667
668 m = rt6_score_route(rt, oif, strict);
7e980569 669 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 670 match_do_rr = true;
671 m = 0; /* lowest valid score */
7e980569 672 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 673 goto out;
afc154e9
HFS		 674 }
675
676 if (strict & RT6_LOOKUP_F_REACHABLE)
677 rt6_probe(rt);
f11e6659 678
7e980569 679 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 680 if (m > *mpri) {
afc154e9 681 *do_rr = match_do_rr;
f11e6659
DM		 682 *mpri = m;
683 match = rt;
f11e6659 684 }
f11e6659
DM		 685out:
686 return match;
687}
688
689static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
690 struct rt6_info *rr_head,
afc154e9
HFS		 691 u32 metric, int oif, int strict,
692 bool *do_rr)
f11e6659 693{
9fbdcfaf 694 struct rt6_info *rt, *match, *cont;
554cfb7e 695 int mpri = -1;
1da177e4 696
f11e6659 697 match = NULL;
9fbdcfaf
SK		 698 cont = NULL;
699 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
700 if (rt->rt6i_metric != metric) {
701 cont = rt;
702 break;
703 }
704
705 match = find_match(rt, oif, strict, &mpri, match, do_rr);
706 }
707
708 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
709 if (rt->rt6i_metric != metric) {
710 cont = rt;
711 break;
712 }
713
afc154e9 714 match = find_match(rt, oif, strict, &mpri, match, do_rr);
9fbdcfaf
SK		 715 }
716
717 if (match || !cont)
718 return match;
719
720 for (rt = cont; rt; rt = rt->dst.rt6_next)
afc154e9 721 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 722
f11e6659
DM		 723 return match;
724}
1da177e4 725
f11e6659
DM		 726static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
727{
728 struct rt6_info *match, *rt0;
8ed67789 729 struct net *net;
afc154e9 730 bool do_rr = false;
1da177e4 731
f11e6659
DM		 732 rt0 = fn->rr_ptr;
733 if (!rt0)
734 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 735
afc154e9
HFS		 736 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
737 &do_rr);
1da177e4 738
afc154e9 739 if (do_rr) {
d8d1f30b 740 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 741
554cfb7e 742 /* no entries matched; do round-robin */
f11e6659
DM		 743 if (!next || next->rt6i_metric != rt0->rt6i_metric)
744 next = fn->leaf;
745
746 if (next != rt0)
747 fn->rr_ptr = next;
1da177e4 748 }
1da177e4 749
d1918542 750 net = dev_net(rt0->dst.dev);
a02cec21 751 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 752}
753
8b9df265
MKL		 754static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
755{
756 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
757}
758
70ceb4f5
YH		 759#ifdef CONFIG_IPV6_ROUTE_INFO
760int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 761 const struct in6_addr *gwaddr)
70ceb4f5 762{
c346dca1 763 struct net *net = dev_net(dev);
70ceb4f5
YH		 764 struct route_info *rinfo = (struct route_info *) opt;
765 struct in6_addr prefix_buf, *prefix;
766 unsigned int pref;
4bed72e4 767 unsigned long lifetime;
70ceb4f5
YH		 768 struct rt6_info *rt;
769
770 if (len < sizeof(struct route_info)) {
771 return -EINVAL;
772 }
773
774 /* Sanity check for prefix_len and length */
775 if (rinfo->length > 3) {
776 return -EINVAL;
777 } else if (rinfo->prefix_len > 128) {
778 return -EINVAL;
779 } else if (rinfo->prefix_len > 64) {
780 if (rinfo->length < 2) {
781 return -EINVAL;
782 }
783 } else if (rinfo->prefix_len > 0) {
784 if (rinfo->length < 1) {
785 return -EINVAL;
786 }
787 }
788
789 pref = rinfo->route_pref;
790 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 791 return -EINVAL;
70ceb4f5 792
4bed72e4 793 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 794
795 if (rinfo->length == 3)
796 prefix = (struct in6_addr *)rinfo->prefix;
797 else {
798 /* this function is safe */
799 ipv6_addr_prefix(&prefix_buf,
800 (struct in6_addr *)rinfo->prefix,
801 rinfo->prefix_len);
802 prefix = &prefix_buf;
803 }
804
f104a567
DJ		 805 if (rinfo->prefix_len == 0)
806 rt = rt6_get_dflt_router(gwaddr, dev);
807 else
808 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
830218c1 809 gwaddr, dev);
70ceb4f5
YH		 810
811 if (rt && !lifetime) {
e0a1ad73 812 ip6_del_rt(rt);
70ceb4f5
YH		 813 rt = NULL;
814 }
815
816 if (!rt && lifetime)
830218c1
DA		 817 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr,
818 dev, pref);
70ceb4f5
YH		 819 else if (rt)
820 rt->rt6i_flags = RTF_ROUTEINFO |
821 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
822
823 if (rt) {
1716a961
G		 824 if (!addrconf_finite_timeout(lifetime))
825 rt6_clean_expires(rt);
826 else
827 rt6_set_expires(rt, jiffies + HZ * lifetime);
828
94e187c0 829 ip6_rt_put(rt);
70ceb4f5
YH		 830 }
831 return 0;
832}
833#endif
834
a3c00e46
MKL		 835static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
836 struct in6_addr *saddr)
837{
838 struct fib6_node *pn;
839 while (1) {
840 if (fn->fn_flags & RTN_TL_ROOT)
841 return NULL;
842 pn = fn->parent;
843 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
844 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
845 else
846 fn = pn;
847 if (fn->fn_flags & RTN_RTINFO)
848 return fn;
849 }
850}
c71099ac 851
8ed67789
DL		 852static struct rt6_info *ip6_pol_route_lookup(struct net *net,
853 struct fib6_table *table,
4c9483b2 854 struct flowi6 *fl6, int flags)
1da177e4
LT		 855{
856 struct fib6_node *fn;
857 struct rt6_info *rt;
858
c71099ac 859 read_lock_bh(&table->tb6_lock);
4c9483b2 860 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 861restart:
862 rt = fn->leaf;
4c9483b2 863 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 864 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 865 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 866 if (rt == net->ipv6.ip6_null_entry) {
867 fn = fib6_backtrack(fn, &fl6->saddr);
868 if (fn)
869 goto restart;
870 }
d8d1f30b 871 dst_use(&rt->dst, jiffies);
c71099ac 872 read_unlock_bh(&table->tb6_lock);
b811580d
DA		 873
874 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
875
c71099ac
TG		 876 return rt;
877
878}
879
67ba4152 880struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 881 int flags)
882{
883 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
884}
885EXPORT_SYMBOL_GPL(ip6_route_lookup);
886
9acd9f3a
YH		 887struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
888 const struct in6_addr *saddr, int oif, int strict)
c71099ac 889{
4c9483b2
DM		 890 struct flowi6 fl6 = {
891 .flowi6_oif = oif,
892 .daddr = *daddr,
c71099ac
TG		 893 };
894 struct dst_entry *dst;
77d16f45 895 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 896
adaa70bb 897 if (saddr) {
4c9483b2 898 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 899 flags |= RT6_LOOKUP_F_HAS_SADDR;
900 }
901
4c9483b2 902 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 903 if (dst->error == 0)
904 return (struct rt6_info *) dst;
905
906 dst_release(dst);
907
1da177e4
LT		 908 return NULL;
909}
7159039a
YH		 910EXPORT_SYMBOL(rt6_lookup);
911
c71099ac 912/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 913 It takes new route entry, the addition fails by any reason the
914 route is freed. In any case, if caller does not hold it, it may
915 be destroyed.
916 */
917
e5fd387a 918static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 919 struct mx6_config *mxc)
1da177e4
LT		 920{
921 int err;
c71099ac 922 struct fib6_table *table;
1da177e4 923
c71099ac
TG		 924 table = rt->rt6i_table;
925 write_lock_bh(&table->tb6_lock);
e715b6d3 926 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 927 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 928
929 return err;
930}
931
40e22e8f
TG		 932int ip6_ins_rt(struct rt6_info *rt)
933{
e715b6d3
FW		 934 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
935 struct mx6_config mxc = { .mx = NULL, };
936
937 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 938}
939
8b9df265
MKL		 940static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
941 const struct in6_addr *daddr,
942 const struct in6_addr *saddr)
1da177e4 943{
1da177e4
LT		 944 struct rt6_info *rt;
945
946 /*
947 * Clone the route.
948 */
949
d52d3997 950 if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
83a09abd 951 ort = (struct rt6_info *)ort->dst.from;
1da177e4 952
ad706862 953 rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
83a09abd
MKL		 954
955 if (!rt)
956 return NULL;
957
958 ip6_rt_copy_init(rt, ort);
959 rt->rt6i_flags |= RTF_CACHE;
960 rt->rt6i_metric = 0;
961 rt->dst.flags |= DST_HOST;
962 rt->rt6i_dst.addr = *daddr;
963 rt->rt6i_dst.plen = 128;
1da177e4 964
83a09abd
MKL		 965 if (!rt6_is_gw_or_nonexthop(ort)) {
966 if (ort->rt6i_dst.plen != 128 &&
967 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
968 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 969#ifdef CONFIG_IPV6_SUBTREES
83a09abd
MKL		 970 if (rt->rt6i_src.plen && saddr) {
971 rt->rt6i_src.addr = *saddr;
972 rt->rt6i_src.plen = 128;
8b9df265 973 }
83a09abd 974#endif
95a9a5ba 975 }
1da177e4 976
95a9a5ba
YH		 977 return rt;
978}
1da177e4 979
d52d3997
MKL		 980static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
981{
982 struct rt6_info *pcpu_rt;
983
984 pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
ad706862 985 rt->dst.dev, rt->dst.flags);
d52d3997
MKL		 986
987 if (!pcpu_rt)
988 return NULL;
989 ip6_rt_copy_init(pcpu_rt, rt);
990 pcpu_rt->rt6i_protocol = rt->rt6i_protocol;
991 pcpu_rt->rt6i_flags |= RTF_PCPU;
992 return pcpu_rt;
993}
994
995/* It should be called with read_lock_bh(&tb6_lock) acquired */
996static struct rt6_info *rt6_get_pcpu_route(struct rt6_info *rt)
997{
a73e4195 998 struct rt6_info *pcpu_rt, **p;
d52d3997
MKL		 999
1000 p = this_cpu_ptr(rt->rt6i_pcpu);
1001 pcpu_rt = *p;
1002
a73e4195
MKL		 1003 if (pcpu_rt) {
1004 dst_hold(&pcpu_rt->dst);
1005 rt6_dst_from_metrics_check(pcpu_rt);
1006 }
1007 return pcpu_rt;
1008}
1009
1010static struct rt6_info *rt6_make_pcpu_route(struct rt6_info *rt)
1011{
9c7370a1 1012 struct fib6_table *table = rt->rt6i_table;
a73e4195 1013 struct rt6_info *pcpu_rt, *prev, **p;
d52d3997
MKL		 1014
1015 pcpu_rt = ip6_rt_pcpu_alloc(rt);
1016 if (!pcpu_rt) {
1017 struct net *net = dev_net(rt->dst.dev);
1018
9c7370a1
MKL		 1019 dst_hold(&net->ipv6.ip6_null_entry->dst);
1020 return net->ipv6.ip6_null_entry;
d52d3997
MKL		 1021 }
1022
9c7370a1
MKL		 1023 read_lock_bh(&table->tb6_lock);
1024 if (rt->rt6i_pcpu) {
1025 p = this_cpu_ptr(rt->rt6i_pcpu);
1026 prev = cmpxchg(p, NULL, pcpu_rt);
1027 if (prev) {
1028 /* If someone did it before us, return prev instead */
1029 dst_destroy(&pcpu_rt->dst);
1030 pcpu_rt = prev;
1031 }
1032 } else {
1033 /* rt has been removed from the fib6 tree
1034 * before we have a chance to acquire the read_lock.
1035 * In this case, don't brother to create a pcpu rt
1036 * since rt is going away anyway. The next
1037 * dst_check() will trigger a re-lookup.
1038 */
d52d3997 1039 dst_destroy(&pcpu_rt->dst);
9c7370a1 1040 pcpu_rt = rt;
d52d3997 1041 }
d52d3997
MKL		 1042 dst_hold(&pcpu_rt->dst);
1043 rt6_dst_from_metrics_check(pcpu_rt);
9c7370a1 1044 read_unlock_bh(&table->tb6_lock);
d52d3997
MKL		 1045 return pcpu_rt;
1046}
1047
9ff74384
DA		 1048struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
1049 int oif, struct flowi6 *fl6, int flags)
1da177e4 1050{
367efcb9 1051 struct fib6_node *fn, *saved_fn;
45e4fd26 1052 struct rt6_info *rt;
c71099ac 1053 int strict = 0;
1da177e4 1054
77d16f45 1055 strict |= flags & RT6_LOOKUP_F_IFACE;
d5d32e4b 1056 strict |= flags & RT6_LOOKUP_F_IGNORE_LINKSTATE;
367efcb9
MKL		 1057 if (net->ipv6.devconf_all->forwarding == 0)
1058 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 1059
c71099ac 1060 read_lock_bh(&table->tb6_lock);
1da177e4 1061
4c9483b2 1062 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 1063 saved_fn = fn;
1da177e4 1064
ca254490
DA		 1065 if (fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF)
1066 oif = 0;
1067
a3c00e46 1068redo_rt6_select:
367efcb9 1069 rt = rt6_select(fn, oif, strict);
52bd4c0c 1070 if (rt->rt6i_nsiblings)
367efcb9 1071 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 1072 if (rt == net->ipv6.ip6_null_entry) {
1073 fn = fib6_backtrack(fn, &fl6->saddr);
1074 if (fn)
1075 goto redo_rt6_select;
367efcb9
MKL		 1076 else if (strict & RT6_LOOKUP_F_REACHABLE) {
1077 /* also consider unreachable route */
1078 strict &= ~RT6_LOOKUP_F_REACHABLE;
1079 fn = saved_fn;
1080 goto redo_rt6_select;
367efcb9 1081 }
a3c00e46
MKL		 1082 }
1083
fb9de91e 1084
3da59bd9 1085 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
d52d3997
MKL		 1086 dst_use(&rt->dst, jiffies);
1087 read_unlock_bh(&table->tb6_lock);
1088
1089 rt6_dst_from_metrics_check(rt);
b811580d
DA		 1090
1091 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
d52d3997 1092 return rt;
3da59bd9
MKL		 1093 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
1094 !(rt->rt6i_flags & RTF_GATEWAY))) {
1095 /* Create a RTF_CACHE clone which will not be
1096 * owned by the fib6 tree. It is for the special case where
1097 * the daddr in the skb during the neighbor look-up is different
1098 * from the fl6->daddr used to look-up route here.
1099 */
1100
1101 struct rt6_info *uncached_rt;
1102
d52d3997
MKL		 1103 dst_use(&rt->dst, jiffies);
1104 read_unlock_bh(&table->tb6_lock);
1105
3da59bd9
MKL		 1106 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
1107 dst_release(&rt->dst);
c71099ac 1108
3da59bd9 1109 if (uncached_rt)
8d0b94af 1110 rt6_uncached_list_add(uncached_rt);
3da59bd9
MKL		 1111 else
1112 uncached_rt = net->ipv6.ip6_null_entry;
d52d3997 1113
3da59bd9 1114 dst_hold(&uncached_rt->dst);
b811580d
DA		 1115
1116 trace_fib6_table_lookup(net, uncached_rt, table->tb6_id, fl6);
3da59bd9 1117 return uncached_rt;
3da59bd9 1118
d52d3997
MKL		 1119 } else {
1120 /* Get a percpu copy */
1121
1122 struct rt6_info *pcpu_rt;
1123
1124 rt->dst.lastuse = jiffies;
1125 rt->dst.__use++;
1126 pcpu_rt = rt6_get_pcpu_route(rt);
d52d3997 1127
9c7370a1
MKL		 1128 if (pcpu_rt) {
1129 read_unlock_bh(&table->tb6_lock);
1130 } else {
1131 /* We have to do the read_unlock first
1132 * because rt6_make_pcpu_route() may trigger
1133 * ip6_dst_gc() which will take the write_lock.
1134 */
1135 dst_hold(&rt->dst);
1136 read_unlock_bh(&table->tb6_lock);
a73e4195 1137 pcpu_rt = rt6_make_pcpu_route(rt);
9c7370a1
MKL		 1138 dst_release(&rt->dst);
1139 }
d52d3997 1140
b811580d 1141 trace_fib6_table_lookup(net, pcpu_rt, table->tb6_id, fl6);
d52d3997 1142 return pcpu_rt;
9c7370a1 1143
d52d3997 1144 }
1da177e4 1145}
9ff74384 1146EXPORT_SYMBOL_GPL(ip6_pol_route);
1da177e4 1147
8ed67789 1148static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 1149 struct flowi6 *fl6, int flags)
4acad72d 1150{
4c9483b2 1151 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 1152}
1153
d409b847
MB		 1154struct dst_entry *ip6_route_input_lookup(struct net *net,
1155 struct net_device *dev,
1156 struct flowi6 *fl6, int flags)
72331bc0
SL		 1157{
1158 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1159 flags |= RT6_LOOKUP_F_IFACE;
1160
1161 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1162}
d409b847 1163EXPORT_SYMBOL_GPL(ip6_route_input_lookup);
72331bc0 1164
c71099ac
TG		 1165void ip6_route_input(struct sk_buff *skb)
1166{
b71d1d42 1167 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1168 struct net *net = dev_net(skb->dev);
adaa70bb 1169 int flags = RT6_LOOKUP_F_HAS_SADDR;
904af04d 1170 struct ip_tunnel_info *tun_info;
4c9483b2 1171 struct flowi6 fl6 = {
e0d56fdd 1172 .flowi6_iif = skb->dev->ifindex,
4c9483b2
DM		 1173 .daddr = iph->daddr,
1174 .saddr = iph->saddr,
6502ca52 1175 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1176 .flowi6_mark = skb->mark,
1177 .flowi6_proto = iph->nexthdr,
c71099ac 1178 };
adaa70bb 1179
904af04d 1180 tun_info = skb_tunnel_info(skb);
46fa062a 1181 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
904af04d 1182 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
06e9d040 1183 skb_dst_drop(skb);
72331bc0 1184 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1185}
1186
8ed67789 1187static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1188 struct flowi6 *fl6, int flags)
1da177e4 1189{
4c9483b2 1190 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1191}
1192
6f21c96a
PA		 1193struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
1194 struct flowi6 *fl6, int flags)
c71099ac 1195{
d46a9d67 1196 bool any_src;
c71099ac 1197
4c1feac5
DA		 1198 if (rt6_need_strict(&fl6->daddr)) {
1199 struct dst_entry *dst;
1200
1201 dst = l3mdev_link_scope_lookup(net, fl6);
1202 if (dst)
1203 return dst;
1204 }
ca254490 1205
1fb9489b 1206 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1207
d46a9d67 1208 any_src = ipv6_addr_any(&fl6->saddr);
741a11d9 1209 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
d46a9d67 1210 (fl6->flowi6_oif && any_src))
77d16f45 1211 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1212
d46a9d67 1213 if (!any_src)
adaa70bb 1214 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1215 else if (sk)
1216 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1217
4c9483b2 1218 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 1219}
6f21c96a 1220EXPORT_SYMBOL_GPL(ip6_route_output_flags);
1da177e4 1221
2774c131 1222struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1223{
5c1e6aa3 1224 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1225 struct dst_entry *new = NULL;
1226
f5b0a874 1227 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1228 if (rt) {
0a1f5962 1229 rt6_info_init(rt);
8104891b 1230
0a1f5962 1231 new = &rt->dst;
14e50e57 1232 new->__use = 1;
352e512c 1233 new->input = dst_discard;
ede2059d 1234 new->output = dst_discard_out;
14e50e57 1235
0a1f5962 1236 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1237 rt->rt6i_idev = ort->rt6i_idev;
1238 if (rt->rt6i_idev)
1239 in6_dev_hold(rt->rt6i_idev);
14e50e57 1240
4e3fd7a0 1241 rt->rt6i_gateway = ort->rt6i_gateway;
0a1f5962 1242 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
14e50e57
DM		 1243 rt->rt6i_metric = 0;
1244
1245 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1246#ifdef CONFIG_IPV6_SUBTREES
1247 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1248#endif
1249
1250 dst_free(new);
1251 }
1252
69ead7af
DM		 1253 dst_release(dst_orig);
1254 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1255}
14e50e57 1256
1da177e4
LT		 1257/*
1258 * Destination cache support functions
1259 */
1260
4b32b5ad
MKL		 1261static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1262{
1263 if (rt->dst.from &&
1264 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1265 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1266}
1267
3da59bd9
MKL		 1268static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1269{
1270 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1271 return NULL;
1272
1273 if (rt6_check_expired(rt))
1274 return NULL;
1275
1276 return &rt->dst;
1277}
1278
1279static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1280{
5973fb1e
MKL		 1281 if (!__rt6_check_expired(rt) &&
1282 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
3da59bd9
MKL		 1283 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1284 return &rt->dst;
1285 else
1286 return NULL;
1287}
1288
1da177e4
LT		 1289static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1290{
1291 struct rt6_info *rt;
1292
1293 rt = (struct rt6_info *) dst;
1294
6f3118b5
ND		 1295 /* All IPV6 dsts are created with ->obsolete set to the value
1296 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1297 * into this function always.
1298 */
e3bc10bd 1299
4b32b5ad
MKL		 1300 rt6_dst_from_metrics_check(rt);
1301
02bcf4e0
MKL		 1302 if (rt->rt6i_flags & RTF_PCPU ||
1303 (unlikely(dst->flags & DST_NOCACHE) && rt->dst.from))
3da59bd9
MKL		 1304 return rt6_dst_from_check(rt, cookie);
1305 else
1306 return rt6_check(rt, cookie);
1da177e4
LT		 1307}
1308
1309static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1310{
1311 struct rt6_info *rt = (struct rt6_info *) dst;
1312
1313 if (rt) {
54c1a859
YH		 1314 if (rt->rt6i_flags & RTF_CACHE) {
1315 if (rt6_check_expired(rt)) {
1316 ip6_del_rt(rt);
1317 dst = NULL;
1318 }
1319 } else {
1da177e4 1320 dst_release(dst);
54c1a859
YH		 1321 dst = NULL;
1322 }
1da177e4 1323 }
54c1a859 1324 return dst;
1da177e4
LT		 1325}
1326
1327static void ip6_link_failure(struct sk_buff *skb)
1328{
1329 struct rt6_info *rt;
1330
3ffe533c 1331 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1332
adf30907 1333 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1334 if (rt) {
1eb4f758
HFS		 1335 if (rt->rt6i_flags & RTF_CACHE) {
1336 dst_hold(&rt->dst);
8e3d5be7 1337 ip6_del_rt(rt);
1eb4f758 1338 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1339 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1340 }
1da177e4
LT		 1341 }
1342}
1343
45e4fd26
MKL		 1344static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1345{
1346 struct net *net = dev_net(rt->dst.dev);
1347
1348 rt->rt6i_flags |= RTF_MODIFIED;
1349 rt->rt6i_pmtu = mtu;
1350 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1351}
1352
0d3f6d29
MKL		 1353static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1354{
1355 return !(rt->rt6i_flags & RTF_CACHE) &&
1356 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1357}
1358
45e4fd26
MKL		 1359static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1360 const struct ipv6hdr *iph, u32 mtu)
1da177e4 1361{
67ba4152 1362 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1363
45e4fd26
MKL		 1364 if (rt6->rt6i_flags & RTF_LOCAL)
1365 return;
81aded24 1366
45e4fd26
MKL		 1367 dst_confirm(dst);
1368 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1369 if (mtu >= dst_mtu(dst))
1370 return;
9d289715 1371
0d3f6d29 1372 if (!rt6_cache_allowed_for_pmtu(rt6)) {
45e4fd26
MKL		 1373 rt6_do_update_pmtu(rt6, mtu);
1374 } else {
1375 const struct in6_addr *daddr, *saddr;
1376 struct rt6_info *nrt6;
1377
1378 if (iph) {
1379 daddr = &iph->daddr;
1380 saddr = &iph->saddr;
1381 } else if (sk) {
1382 daddr = &sk->sk_v6_daddr;
1383 saddr = &inet6_sk(sk)->saddr;
1384 } else {
1385 return;
1386 }
1387 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1388 if (nrt6) {
1389 rt6_do_update_pmtu(nrt6, mtu);
1390
1391 /* ip6_ins_rt(nrt6) will bump the
1392 * rt6->rt6i_node->fn_sernum
1393 * which will fail the next rt6_check() and
1394 * invalidate the sk->sk_dst_cache.
1395 */
1396 ip6_ins_rt(nrt6);
1397 }
1da177e4
LT		 1398 }
1399}
1400
45e4fd26
MKL		 1401static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1402 struct sk_buff *skb, u32 mtu)
1403{
1404 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1405}
1406
42ae66c8
DM		 1407void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1408 int oif, u32 mark)
81aded24
DM		 1409{
1410 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1411 struct dst_entry *dst;
1412 struct flowi6 fl6;
1413
1414 memset(&fl6, 0, sizeof(fl6));
1415 fl6.flowi6_oif = oif;
1b3c61dc 1416 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1417 fl6.daddr = iph->daddr;
1418 fl6.saddr = iph->saddr;
6502ca52 1419 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1420
1421 dst = ip6_route_output(net, NULL, &fl6);
1422 if (!dst->error)
45e4fd26 1423 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
81aded24
DM		 1424 dst_release(dst);
1425}
1426EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1427
1428void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1429{
33c162a9
MKL		 1430 struct dst_entry *dst;
1431
81aded24
DM		 1432 ip6_update_pmtu(skb, sock_net(sk), mtu,
1433 sk->sk_bound_dev_if, sk->sk_mark);
33c162a9
MKL		 1434
1435 dst = __sk_dst_get(sk);
1436 if (!dst || !dst->obsolete ||
1437 dst->ops->check(dst, inet6_sk(sk)->dst_cookie))
1438 return;
1439
1440 bh_lock_sock(sk);
1441 if (!sock_owned_by_user(sk) && !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
1442 ip6_datagram_dst_update(sk, false);
1443 bh_unlock_sock(sk);
81aded24
DM		 1444}
1445EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1446
b55b76b2
DJ		 1447/* Handle redirects */
1448struct ip6rd_flowi {
1449 struct flowi6 fl6;
1450 struct in6_addr gateway;
1451};
1452
1453static struct rt6_info *__ip6_route_redirect(struct net *net,
1454 struct fib6_table *table,
1455 struct flowi6 *fl6,
1456 int flags)
1457{
1458 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1459 struct rt6_info *rt;
1460 struct fib6_node *fn;
1461
1462 /* Get the "current" route for this destination and
1463 * check if the redirect has come from approriate router.
1464 *
1465 * RFC 4861 specifies that redirects should only be
1466 * accepted if they come from the nexthop to the target.
1467 * Due to the way the routes are chosen, this notion
1468 * is a bit fuzzy and one might need to check all possible
1469 * routes.
1470 */
1471
1472 read_lock_bh(&table->tb6_lock);
1473 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1474restart:
1475 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1476 if (rt6_check_expired(rt))
1477 continue;
1478 if (rt->dst.error)
1479 break;
1480 if (!(rt->rt6i_flags & RTF_GATEWAY))
1481 continue;
1482 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1483 continue;
1484 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1485 continue;
1486 break;
1487 }
1488
1489 if (!rt)
1490 rt = net->ipv6.ip6_null_entry;
1491 else if (rt->dst.error) {
1492 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1493 goto out;
1494 }
1495
1496 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1497 fn = fib6_backtrack(fn, &fl6->saddr);
1498 if (fn)
1499 goto restart;
b55b76b2 1500 }
a3c00e46 1501
b0a1ba59 1502out:
b55b76b2
DJ		 1503 dst_hold(&rt->dst);
1504
1505 read_unlock_bh(&table->tb6_lock);
1506
b811580d 1507 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
b55b76b2
DJ		 1508 return rt;
1509};
1510
1511static struct dst_entry *ip6_route_redirect(struct net *net,
1512 const struct flowi6 *fl6,
1513 const struct in6_addr *gateway)
1514{
1515 int flags = RT6_LOOKUP_F_HAS_SADDR;
1516 struct ip6rd_flowi rdfl;
1517
1518 rdfl.fl6 = *fl6;
1519 rdfl.gateway = *gateway;
1520
1521 return fib6_rule_lookup(net, &rdfl.fl6,
1522 flags, __ip6_route_redirect);
1523}
1524
3a5ad2ee
DM		 1525void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1526{
1527 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1528 struct dst_entry *dst;
1529 struct flowi6 fl6;
1530
1531 memset(&fl6, 0, sizeof(fl6));
e374c618 1532 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1533 fl6.flowi6_oif = oif;
1534 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1535 fl6.daddr = iph->daddr;
1536 fl6.saddr = iph->saddr;
6502ca52 1537 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1538
b55b76b2
DJ		 1539 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1540 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1541 dst_release(dst);
1542}
1543EXPORT_SYMBOL_GPL(ip6_redirect);
1544
c92a59ec
DJ		 1545void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1546 u32 mark)
1547{
1548 const struct ipv6hdr *iph = ipv6_hdr(skb);
1549 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1550 struct dst_entry *dst;
1551 struct flowi6 fl6;
1552
1553 memset(&fl6, 0, sizeof(fl6));
e374c618 1554 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1555 fl6.flowi6_oif = oif;
1556 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1557 fl6.daddr = msg->dest;
1558 fl6.saddr = iph->daddr;
1559
b55b76b2
DJ		 1560 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1561 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1562 dst_release(dst);
1563}
1564
3a5ad2ee
DM		 1565void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1566{
1567 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1568}
1569EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1570
0dbaee3b 1571static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1572{
0dbaee3b
DM		 1573 struct net_device *dev = dst->dev;
1574 unsigned int mtu = dst_mtu(dst);
1575 struct net *net = dev_net(dev);
1576
1da177e4
LT		 1577 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1578
5578689a
DL		 1579 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1580 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1581
1582 /*
1ab1457c
YH		 1583 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1584 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1585 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1586 * rely only on pmtu discovery"
1587 */
1588 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1589 mtu = IPV6_MAXPLEN;
1590 return mtu;
1591}
1592
ebb762f2 1593static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1594{
4b32b5ad
MKL		 1595 const struct rt6_info *rt = (const struct rt6_info *)dst;
1596 unsigned int mtu = rt->rt6i_pmtu;
d33e4553 1597 struct inet6_dev *idev;
618f9bc7 1598
4b32b5ad
MKL		 1599 if (mtu)
1600 goto out;
1601
1602 mtu = dst_metric_raw(dst, RTAX_MTU);
618f9bc7 1603 if (mtu)
30f78d8e 1604 goto out;
618f9bc7
SK		 1605
1606 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1607
1608 rcu_read_lock();
1609 idev = __in6_dev_get(dst->dev);
1610 if (idev)
1611 mtu = idev->cnf.mtu6;
1612 rcu_read_unlock();
1613
30f78d8e 1614out:
14972cbd
RP		 1615 mtu = min_t(unsigned int, mtu, IP6_MAX_MTU);
1616
1617 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
d33e4553
DM		 1618}
1619
3b00944c
YH		 1620static struct dst_entry *icmp6_dst_gc_list;
1621static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1622
3b00944c 1623struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1624 struct flowi6 *fl6)
1da177e4 1625{
87a11578 1626 struct dst_entry *dst;
1da177e4
LT		 1627 struct rt6_info *rt;
1628 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1629 struct net *net = dev_net(dev);
1da177e4 1630
38308473 1631 if (unlikely(!idev))
122bdf67 1632 return ERR_PTR(-ENODEV);
1da177e4 1633
ad706862 1634 rt = ip6_dst_alloc(net, dev, 0);
38308473 1635 if (unlikely(!rt)) {
1da177e4 1636 in6_dev_put(idev);
87a11578 1637 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1638 goto out;
1639 }
1640
8e2ec639
YZ		 1641 rt->dst.flags |= DST_HOST;
1642 rt->dst.output = ip6_output;
d8d1f30b 1643 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1644 rt->rt6i_gateway = fl6->daddr;
87a11578 1645 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1646 rt->rt6i_dst.plen = 128;
1647 rt->rt6i_idev = idev;
14edd87d 1648 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1649
3b00944c 1650 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1651 rt->dst.next = icmp6_dst_gc_list;
1652 icmp6_dst_gc_list = &rt->dst;
3b00944c 1653 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1654
5578689a 1655 fib6_force_start_gc(net);
1da177e4 1656
87a11578
DM		 1657 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1658
1da177e4 1659out:
87a11578 1660 return dst;
1da177e4
LT		 1661}
1662
3d0f24a7 1663int icmp6_dst_gc(void)
1da177e4 1664{
e9476e95 1665 struct dst_entry *dst, **pprev;
3d0f24a7 1666 int more = 0;
1da177e4 1667
3b00944c
YH		 1668 spin_lock_bh(&icmp6_dst_lock);
1669 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1670
1da177e4
LT		 1671 while ((dst = *pprev) != NULL) {
1672 if (!atomic_read(&dst->__refcnt)) {
1673 *pprev = dst->next;
1674 dst_free(dst);
1da177e4
LT		 1675 } else {
1676 pprev = &dst->next;
3d0f24a7 1677 ++more;
1da177e4
LT		 1678 }
1679 }
1680
3b00944c 1681 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1682
3d0f24a7 1683 return more;
1da177e4
LT		 1684}
1685
1e493d19
DM		 1686static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1687 void *arg)
1688{
1689 struct dst_entry *dst, **pprev;
1690
1691 spin_lock_bh(&icmp6_dst_lock);
1692 pprev = &icmp6_dst_gc_list;
1693 while ((dst = *pprev) != NULL) {
1694 struct rt6_info *rt = (struct rt6_info *) dst;
1695 if (func(rt, arg)) {
1696 *pprev = dst->next;
1697 dst_free(dst);
1698 } else {
1699 pprev = &dst->next;
1700 }
1701 }
1702 spin_unlock_bh(&icmp6_dst_lock);
1703}
1704
569d3645 1705static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1706{
86393e52 1707 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1708 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1709 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1710 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1711 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1712 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1713 int entries;
7019b78e 1714
fc66f95c 1715 entries = dst_entries_get_fast(ops);
49a18d86 1716 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1717 entries <= rt_max_size)
1da177e4
LT		 1718 goto out;
1719
6891a346 1720 net->ipv6.ip6_rt_gc_expire++;
14956643 1721 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1722 entries = dst_entries_get_slow(ops);
1723 if (entries < ops->gc_thresh)
7019b78e 1724 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1725out:
7019b78e 1726 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1727 return entries > rt_max_size;
1da177e4
LT		 1728}
1729
e715b6d3
FW		 1730static int ip6_convert_metrics(struct mx6_config *mxc,
1731 const struct fib6_config *cfg)
1732{
c3a8d947 1733 bool ecn_ca = false;
e715b6d3
FW		 1734 struct nlattr *nla;
1735 int remaining;
1736 u32 *mp;
1737
63159f29 1738 if (!cfg->fc_mx)
e715b6d3
FW		 1739 return 0;
1740
1741 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1742 if (unlikely(!mp))
1743 return -ENOMEM;
1744
1745 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1746 int type = nla_type(nla);
1bb14807 1747 u32 val;
e715b6d3 1748
1bb14807
DB		 1749 if (!type)
1750 continue;
1751 if (unlikely(type > RTAX_MAX))
1752 goto err;
ea697639 1753
1bb14807
DB		 1754 if (type == RTAX_CC_ALGO) {
1755 char tmp[TCP_CA_NAME_MAX];
e715b6d3 1756
1bb14807 1757 nla_strlcpy(tmp, nla, sizeof(tmp));
c3a8d947 1758 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
1bb14807
DB		 1759 if (val == TCP_CA_UNSPEC)
1760 goto err;
1761 } else {
1762 val = nla_get_u32(nla);
e715b6d3 1763 }
626abd59
PA		 1764 if (type == RTAX_HOPLIMIT && val > 255)
1765 val = 255;
b8d3e416
DB		 1766 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1767 goto err;
1bb14807
DB		 1768
1769 mp[type - 1] = val;
1770 __set_bit(type - 1, mxc->mx_valid);
e715b6d3
FW		 1771 }
1772
c3a8d947
DB		 1773 if (ecn_ca) {
1774 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1775 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1776 }
e715b6d3 1777
c3a8d947 1778 mxc->mx = mp;
e715b6d3
FW		 1779 return 0;
1780 err:
1781 kfree(mp);
1782 return -EINVAL;
1783}
1da177e4 1784
8c14586f
DA		 1785static struct rt6_info *ip6_nh_lookup_table(struct net *net,
1786 struct fib6_config *cfg,
1787 const struct in6_addr *gw_addr)
1788{
1789 struct flowi6 fl6 = {
1790 .flowi6_oif = cfg->fc_ifindex,
1791 .daddr = *gw_addr,
1792 .saddr = cfg->fc_prefsrc,
1793 };
1794 struct fib6_table *table;
1795 struct rt6_info *rt;
d5d32e4b 1796 int flags = RT6_LOOKUP_F_IFACE | RT6_LOOKUP_F_IGNORE_LINKSTATE;
8c14586f
DA		 1797
1798 table = fib6_get_table(net, cfg->fc_table);
1799 if (!table)
1800 return NULL;
1801
1802 if (!ipv6_addr_any(&cfg->fc_prefsrc))
1803 flags |= RT6_LOOKUP_F_HAS_SADDR;
1804
1805 rt = ip6_pol_route(net, table, cfg->fc_ifindex, &fl6, flags);
1806
1807 /* if table lookup failed, fall back to full lookup */
1808 if (rt == net->ipv6.ip6_null_entry) {
1809 ip6_rt_put(rt);
1810 rt = NULL;
1811 }
1812
1813 return rt;
1814}
1815
8c5b83f0 1816static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
1da177e4 1817{
5578689a 1818 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1819 struct rt6_info *rt = NULL;
1820 struct net_device *dev = NULL;
1821 struct inet6_dev *idev = NULL;
c71099ac 1822 struct fib6_table *table;
1da177e4 1823 int addr_type;
8c5b83f0 1824 int err = -EINVAL;
1da177e4 1825
86872cb5 1826 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
8c5b83f0 1827 goto out;
1da177e4 1828#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1829 if (cfg->fc_src_len)
8c5b83f0 1830 goto out;
1da177e4 1831#endif
86872cb5 1832 if (cfg->fc_ifindex) {
1da177e4 1833 err = -ENODEV;
5578689a 1834 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1835 if (!dev)
1836 goto out;
1837 idev = in6_dev_get(dev);
1838 if (!idev)
1839 goto out;
1840 }
1841
86872cb5
TG		 1842 if (cfg->fc_metric == 0)
1843 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1844
d71314b4 1845 err = -ENOBUFS;
38308473
DM		 1846 if (cfg->fc_nlinfo.nlh &&
1847 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1848 table = fib6_get_table(net, cfg->fc_table);
38308473 1849 if (!table) {
f3213831 1850 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1851 table = fib6_new_table(net, cfg->fc_table);
1852 }
1853 } else {
1854 table = fib6_new_table(net, cfg->fc_table);
1855 }
38308473
DM		 1856
1857 if (!table)
c71099ac 1858 goto out;
c71099ac 1859
ad706862
MKL		 1860 rt = ip6_dst_alloc(net, NULL,
1861 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
1da177e4 1862
38308473 1863 if (!rt) {
1da177e4
LT		 1864 err = -ENOMEM;
1865 goto out;
1866 }
1867
1716a961
G		 1868 if (cfg->fc_flags & RTF_EXPIRES)
1869 rt6_set_expires(rt, jiffies +
1870 clock_t_to_jiffies(cfg->fc_expires));
1871 else
1872 rt6_clean_expires(rt);
1da177e4 1873
86872cb5
TG		 1874 if (cfg->fc_protocol == RTPROT_UNSPEC)
1875 cfg->fc_protocol = RTPROT_BOOT;
1876 rt->rt6i_protocol = cfg->fc_protocol;
1877
1878 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1879
1880 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1881 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1882 else if (cfg->fc_flags & RTF_LOCAL)
1883 rt->dst.input = ip6_input;
1da177e4 1884 else
d8d1f30b 1885 rt->dst.input = ip6_forward;
1da177e4 1886
d8d1f30b 1887 rt->dst.output = ip6_output;
1da177e4 1888
19e42e45
RP		 1889 if (cfg->fc_encap) {
1890 struct lwtunnel_state *lwtstate;
1891
1892 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
127eb7cd
TH		 1893 cfg->fc_encap, AF_INET6, cfg,
1894 &lwtstate);
19e42e45
RP		 1895 if (err)
1896 goto out;
61adedf3
JB		 1897 rt->dst.lwtstate = lwtstate_get(lwtstate);
1898 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1899 rt->dst.lwtstate->orig_output = rt->dst.output;
1900 rt->dst.output = lwtunnel_output;
25368623 1901 }
61adedf3
JB		 1902 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1903 rt->dst.lwtstate->orig_input = rt->dst.input;
1904 rt->dst.input = lwtunnel_input;
25368623 1905 }
19e42e45
RP		 1906 }
1907
86872cb5
TG		 1908 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1909 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1910 if (rt->rt6i_dst.plen == 128)
e5fd387a 1911 rt->dst.flags |= DST_HOST;
e5fd387a 1912
1da177e4 1913#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1914 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1915 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1916#endif
1917
86872cb5 1918 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1919
1920 /* We cannot add true routes via loopback here,
1921 they would result in kernel looping; promote them to reject routes
1922 */
86872cb5 1923 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1924 (dev && (dev->flags & IFF_LOOPBACK) &&
1925 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1926 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1927 /* hold loopback dev/idev if we haven't done so. */
5578689a 1928 if (dev != net->loopback_dev) {
1da177e4
LT		 1929 if (dev) {
1930 dev_put(dev);
1931 in6_dev_put(idev);
1932 }
5578689a 1933 dev = net->loopback_dev;
1da177e4
LT		 1934 dev_hold(dev);
1935 idev = in6_dev_get(dev);
1936 if (!idev) {
1937 err = -ENODEV;
1938 goto out;
1939 }
1940 }
1da177e4 1941 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1942 switch (cfg->fc_type) {
1943 case RTN_BLACKHOLE:
1944 rt->dst.error = -EINVAL;
ede2059d 1945 rt->dst.output = dst_discard_out;
7150aede 1946 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1947 break;
1948 case RTN_PROHIBIT:
1949 rt->dst.error = -EACCES;
7150aede
K		 1950 rt->dst.output = ip6_pkt_prohibit_out;
1951 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1952 break;
b4949ab2 1953 case RTN_THROW:
0315e382 1954 case RTN_UNREACHABLE:
ef2c7d7b 1955 default:
7150aede 1956 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
0315e382
NF		 1957 : (cfg->fc_type == RTN_UNREACHABLE)
1958 ? -EHOSTUNREACH : -ENETUNREACH;
7150aede
K		 1959 rt->dst.output = ip6_pkt_discard_out;
1960 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1961 break;
1962 }
1da177e4
LT		 1963 goto install_route;
1964 }
1965
86872cb5 1966 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1967 const struct in6_addr *gw_addr;
1da177e4
LT		 1968 int gwa_type;
1969
86872cb5 1970 gw_addr = &cfg->fc_gateway;
330567b7 1971 gwa_type = ipv6_addr_type(gw_addr);
48ed7b26
FW		 1972
1973 /* if gw_addr is local we will fail to detect this in case
1974 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1975 * will return already-added prefix route via interface that
1976 * prefix route was assigned to, which might be non-loopback.
1977 */
1978 err = -EINVAL;
330567b7
FW		 1979 if (ipv6_chk_addr_and_flags(net, gw_addr,
1980 gwa_type & IPV6_ADDR_LINKLOCAL ?
1981 dev : NULL, 0, 0))
48ed7b26
FW		 1982 goto out;
1983
4e3fd7a0 1984 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1985
1986 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
8c14586f 1987 struct rt6_info *grt = NULL;
1da177e4
LT		 1988
1989 /* IPv6 strictly inhibits using not link-local
1990 addresses as nexthop address.
1991 Otherwise, router will not able to send redirects.
1992 It is very good, but in some (rare!) circumstances
1993 (SIT, PtP, NBMA NOARP links) it is handy to allow
1994 some exceptions. --ANK
1995 */
38308473 1996 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1997 goto out;
1998
a435a07f 1999 if (cfg->fc_table) {
8c14586f
DA		 2000 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
2001
a435a07f
VB		 2002 if (grt) {
2003 if (grt->rt6i_flags & RTF_GATEWAY ||
2004 (dev && dev != grt->dst.dev)) {
2005 ip6_rt_put(grt);
2006 grt = NULL;
2007 }
2008 }
2009 }
2010
8c14586f
DA		 2011 if (!grt)
2012 grt = rt6_lookup(net, gw_addr, NULL,
2013 cfg->fc_ifindex, 1);
1da177e4
LT		 2014
2015 err = -EHOSTUNREACH;
38308473 2016 if (!grt)
1da177e4
LT		 2017 goto out;
2018 if (dev) {
d1918542 2019 if (dev != grt->dst.dev) {
94e187c0 2020 ip6_rt_put(grt);
1da177e4
LT		 2021 goto out;
2022 }
2023 } else {
d1918542 2024 dev = grt->dst.dev;
1da177e4
LT		 2025 idev = grt->rt6i_idev;
2026 dev_hold(dev);
2027 in6_dev_hold(grt->rt6i_idev);
2028 }
38308473 2029 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 2030 err = 0;
94e187c0 2031 ip6_rt_put(grt);
1da177e4
LT		 2032
2033 if (err)
2034 goto out;
2035 }
2036 err = -EINVAL;
38308473 2037 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2038 goto out;
2039 }
2040
2041 err = -ENODEV;
38308473 2042 if (!dev)
1da177e4
LT		 2043 goto out;
2044
c3968a85
DW		 2045 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2046 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2047 err = -EINVAL;
2048 goto out;
2049 }
4e3fd7a0 2050 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 2051 rt->rt6i_prefsrc.plen = 128;
2052 } else
2053 rt->rt6i_prefsrc.plen = 0;
2054
86872cb5 2055 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 2056
2057install_route:
d8d1f30b 2058 rt->dst.dev = dev;
1da177e4 2059 rt->rt6i_idev = idev;
c71099ac 2060 rt->rt6i_table = table;
63152fc0 2061
c346dca1 2062 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 2063
8c5b83f0 2064 return rt;
6b9ea5a6
RP		 2065out:
2066 if (dev)
2067 dev_put(dev);
2068 if (idev)
2069 in6_dev_put(idev);
2070 if (rt)
2071 dst_free(&rt->dst);
2072
8c5b83f0 2073 return ERR_PTR(err);
6b9ea5a6
RP		 2074}
2075
2076int ip6_route_add(struct fib6_config *cfg)
2077{
2078 struct mx6_config mxc = { .mx = NULL, };
8c5b83f0 2079 struct rt6_info *rt;
6b9ea5a6
RP		 2080 int err;
2081
8c5b83f0
RP		 2082 rt = ip6_route_info_create(cfg);
2083 if (IS_ERR(rt)) {
2084 err = PTR_ERR(rt);
2085 rt = NULL;
6b9ea5a6 2086 goto out;
8c5b83f0 2087 }
6b9ea5a6 2088
e715b6d3
FW		 2089 err = ip6_convert_metrics(&mxc, cfg);
2090 if (err)
2091 goto out;
1da177e4 2092
e715b6d3
FW		 2093 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2094
2095 kfree(mxc.mx);
6b9ea5a6 2096
e715b6d3 2097 return err;
1da177e4 2098out:
1da177e4 2099 if (rt)
d8d1f30b 2100 dst_free(&rt->dst);
6b9ea5a6 2101
1da177e4
LT		 2102 return err;
2103}
2104
86872cb5 2105static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2106{
2107 int err;
c71099ac 2108 struct fib6_table *table;
d1918542 2109 struct net *net = dev_net(rt->dst.dev);
1da177e4 2110
8e3d5be7
MKL		 2111 if (rt == net->ipv6.ip6_null_entry ||
2112 rt->dst.flags & DST_NOCACHE) {
6825a26c
G		 2113 err = -ENOENT;
2114 goto out;
2115 }
6c813a72 2116
c71099ac
TG		 2117 table = rt->rt6i_table;
2118 write_lock_bh(&table->tb6_lock);
86872cb5 2119 err = fib6_del(rt, info);
c71099ac 2120 write_unlock_bh(&table->tb6_lock);
1da177e4 2121
6825a26c 2122out:
94e187c0 2123 ip6_rt_put(rt);
1da177e4
LT		 2124 return err;
2125}
2126
e0a1ad73
TG		 2127int ip6_del_rt(struct rt6_info *rt)
2128{
4d1169c1 2129 struct nl_info info = {
d1918542 2130 .nl_net = dev_net(rt->dst.dev),
4d1169c1 2131 };
528c4ceb 2132 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 2133}
2134
86872cb5 2135static int ip6_route_del(struct fib6_config *cfg)
1da177e4 2136{
c71099ac 2137 struct fib6_table *table;
1da177e4
LT		 2138 struct fib6_node *fn;
2139 struct rt6_info *rt;
2140 int err = -ESRCH;
2141
5578689a 2142 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 2143 if (!table)
c71099ac
TG		 2144 return err;
2145
2146 read_lock_bh(&table->tb6_lock);
1da177e4 2147
c71099ac 2148 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 2149 &cfg->fc_dst, cfg->fc_dst_len,
2150 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 2151
1da177e4 2152 if (fn) {
d8d1f30b 2153 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 2154 if ((rt->rt6i_flags & RTF_CACHE) &&
2155 !(cfg->fc_flags & RTF_CACHE))
2156 continue;
86872cb5 2157 if (cfg->fc_ifindex &&
d1918542
DM		 2158 (!rt->dst.dev ||
2159 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 2160 continue;
86872cb5
TG		 2161 if (cfg->fc_flags & RTF_GATEWAY &&
2162 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 2163 continue;
86872cb5 2164 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 2165 continue;
d8d1f30b 2166 dst_hold(&rt->dst);
c71099ac 2167 read_unlock_bh(&table->tb6_lock);
1da177e4 2168
86872cb5 2169 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 2170 }
2171 }
c71099ac 2172 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2173
2174 return err;
2175}
2176
6700c270 2177static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 2178{
a6279458 2179 struct netevent_redirect netevent;
e8599ff4 2180 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 2181 struct ndisc_options ndopts;
2182 struct inet6_dev *in6_dev;
2183 struct neighbour *neigh;
71bcdba0 2184 struct rd_msg *msg;
6e157b6a
DM		 2185 int optlen, on_link;
2186 u8 *lladdr;
e8599ff4 2187
29a3cad5 2188 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 2189 optlen -= sizeof(*msg);
e8599ff4
DM		 2190
2191 if (optlen < 0) {
6e157b6a 2192 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 2193 return;
2194 }
2195
71bcdba0 2196 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 2197
71bcdba0 2198 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 2199 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 2200 return;
2201 }
2202
6e157b6a 2203 on_link = 0;
71bcdba0 2204 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 2205 on_link = 1;
71bcdba0 2206 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 2207 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 2208 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 2209 return;
2210 }
2211
2212 in6_dev = __in6_dev_get(skb->dev);
2213 if (!in6_dev)
2214 return;
2215 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2216 return;
2217
2218 /* RFC2461 8.1:
2219 * The IP source address of the Redirect MUST be the same as the current
2220 * first-hop router for the specified ICMP Destination Address.
2221 */
2222
f997c55c 2223 if (!ndisc_parse_options(skb->dev, msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 2224 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2225 return;
2226 }
6e157b6a
DM		 2227
2228 lladdr = NULL;
e8599ff4
DM		 2229 if (ndopts.nd_opts_tgt_lladdr) {
2230 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2231 skb->dev);
2232 if (!lladdr) {
2233 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2234 return;
2235 }
2236 }
2237
6e157b6a 2238 rt = (struct rt6_info *) dst;
ec13ad1d 2239 if (rt->rt6i_flags & RTF_REJECT) {
6e157b6a 2240 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 2241 return;
6e157b6a 2242 }
e8599ff4 2243
6e157b6a
DM		 2244 /* Redirect received -> path was valid.
2245 * Look, redirects are sent only in response to data packets,
2246 * so that this nexthop apparently is reachable. --ANK
2247 */
2248 dst_confirm(&rt->dst);
a6279458 2249
71bcdba0 2250 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 2251 if (!neigh)
2252 return;
a6279458 2253
1da177e4
LT		 2254 /*
2255 * We have finally decided to accept it.
2256 */
2257
f997c55c 2258 ndisc_update(skb->dev, neigh, lladdr, NUD_STALE,
1da177e4
LT		 2259 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2260 NEIGH_UPDATE_F_OVERRIDE|
2261 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
f997c55c
AA		 2262 NEIGH_UPDATE_F_ISROUTER)),
2263 NDISC_REDIRECT, &ndopts);
1da177e4 2264
83a09abd 2265 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
38308473 2266 if (!nrt)
1da177e4
LT		 2267 goto out;
2268
2269 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2270 if (on_link)
2271 nrt->rt6i_flags &= ~RTF_GATEWAY;
2272
4e3fd7a0 2273 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 2274
40e22e8f 2275 if (ip6_ins_rt(nrt))
1da177e4
LT		 2276 goto out;
2277
d8d1f30b
CG		 2278 netevent.old = &rt->dst;
2279 netevent.new = &nrt->dst;
71bcdba0 2280 netevent.daddr = &msg->dest;
60592833 2281 netevent.neigh = neigh;
8d71740c
TT		 2282 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2283
38308473 2284 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 2285 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 2286 ip6_del_rt(rt);
1da177e4
LT		 2287 }
2288
2289out:
e8599ff4 2290 neigh_release(neigh);
6e157b6a
DM		 2291}
2292
1da177e4
LT		 2293/*
2294 * Misc support functions
2295 */
2296
4b32b5ad
MKL		 2297static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2298{
2299 BUG_ON(from->dst.from);
2300
2301 rt->rt6i_flags &= ~RTF_EXPIRES;
2302 dst_hold(&from->dst);
2303 rt->dst.from = &from->dst;
2304 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2305}
2306
83a09abd
MKL		 2307static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
2308{
2309 rt->dst.input = ort->dst.input;
2310 rt->dst.output = ort->dst.output;
2311 rt->rt6i_dst = ort->rt6i_dst;
2312 rt->dst.error = ort->dst.error;
2313 rt->rt6i_idev = ort->rt6i_idev;
2314 if (rt->rt6i_idev)
2315 in6_dev_hold(rt->rt6i_idev);
2316 rt->dst.lastuse = jiffies;
2317 rt->rt6i_gateway = ort->rt6i_gateway;
2318 rt->rt6i_flags = ort->rt6i_flags;
2319 rt6_set_from(rt, ort);
2320 rt->rt6i_metric = ort->rt6i_metric;
1da177e4 2321#ifdef CONFIG_IPV6_SUBTREES
83a09abd 2322 rt->rt6i_src = ort->rt6i_src;
1da177e4 2323#endif
83a09abd
MKL		 2324 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2325 rt->rt6i_table = ort->rt6i_table;
61adedf3 2326 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
1da177e4
LT		 2327}
2328
70ceb4f5 2329#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 2330static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42 2331 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2332 const struct in6_addr *gwaddr,
2333 struct net_device *dev)
70ceb4f5 2334{
830218c1
DA		 2335 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO;
2336 int ifindex = dev->ifindex;
70ceb4f5
YH		 2337 struct fib6_node *fn;
2338 struct rt6_info *rt = NULL;
c71099ac
TG		 2339 struct fib6_table *table;
2340
830218c1 2341 table = fib6_get_table(net, tb_id);
38308473 2342 if (!table)
c71099ac 2343 return NULL;
70ceb4f5 2344
5744dd9b 2345 read_lock_bh(&table->tb6_lock);
67ba4152 2346 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2347 if (!fn)
2348 goto out;
2349
d8d1f30b 2350 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2351 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2352 continue;
2353 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2354 continue;
2355 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2356 continue;
d8d1f30b 2357 dst_hold(&rt->dst);
70ceb4f5
YH		 2358 break;
2359 }
2360out:
5744dd9b 2361 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2362 return rt;
2363}
2364
efa2cea0 2365static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42 2366 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2367 const struct in6_addr *gwaddr,
2368 struct net_device *dev,
95c96174 2369 unsigned int pref)
70ceb4f5 2370{
86872cb5 2371 struct fib6_config cfg = {
238fc7ea 2372 .fc_metric = IP6_RT_PRIO_USER,
830218c1 2373 .fc_ifindex = dev->ifindex,
86872cb5
TG		 2374 .fc_dst_len = prefixlen,
2375 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2376 RTF_UP | RTF_PREF(pref),
15e47304 2377 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2378 .fc_nlinfo.nlh = NULL,
2379 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2380 };
2381
830218c1 2382 cfg.fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO,
4e3fd7a0
AD		 2383 cfg.fc_dst = *prefix;
2384 cfg.fc_gateway = *gwaddr;
70ceb4f5 2385
e317da96
YH		 2386 /* We should treat it as a default route if prefix length is 0. */
2387 if (!prefixlen)
86872cb5 2388 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2389
86872cb5 2390 ip6_route_add(&cfg);
70ceb4f5 2391
830218c1 2392 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, dev);
70ceb4f5
YH		 2393}
2394#endif
2395
b71d1d42 2396struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2397{
830218c1 2398 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT;
1da177e4 2399 struct rt6_info *rt;
c71099ac 2400 struct fib6_table *table;
1da177e4 2401
830218c1 2402 table = fib6_get_table(dev_net(dev), tb_id);
38308473 2403 if (!table)
c71099ac 2404 return NULL;
1da177e4 2405
5744dd9b 2406 read_lock_bh(&table->tb6_lock);
67ba4152 2407 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2408 if (dev == rt->dst.dev &&
045927ff 2409 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2410 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2411 break;
2412 }
2413 if (rt)
d8d1f30b 2414 dst_hold(&rt->dst);
5744dd9b 2415 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2416 return rt;
2417}
2418
b71d1d42 2419struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2420 struct net_device *dev,
2421 unsigned int pref)
1da177e4 2422{
86872cb5 2423 struct fib6_config cfg = {
ca254490 2424 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
238fc7ea 2425 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2426 .fc_ifindex = dev->ifindex,
2427 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2428 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2429 .fc_nlinfo.portid = 0,
5578689a 2430 .fc_nlinfo.nlh = NULL,
c346dca1 2431 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2432 };
1da177e4 2433
4e3fd7a0 2434 cfg.fc_gateway = *gwaddr;
1da177e4 2435
830218c1
DA		 2436 if (!ip6_route_add(&cfg)) {
2437 struct fib6_table *table;
2438
2439 table = fib6_get_table(dev_net(dev), cfg.fc_table);
2440 if (table)
2441 table->flags |= RT6_TABLE_HAS_DFLT_ROUTER;
2442 }
1da177e4 2443
1da177e4
LT		 2444 return rt6_get_dflt_router(gwaddr, dev);
2445}
2446
830218c1 2447static void __rt6_purge_dflt_routers(struct fib6_table *table)
1da177e4
LT		 2448{
2449 struct rt6_info *rt;
2450
2451restart:
c71099ac 2452 read_lock_bh(&table->tb6_lock);
d8d1f30b 2453 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2454 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2455 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2456 dst_hold(&rt->dst);
c71099ac 2457 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2458 ip6_del_rt(rt);
1da177e4
LT		 2459 goto restart;
2460 }
2461 }
c71099ac 2462 read_unlock_bh(&table->tb6_lock);
830218c1
DA		 2463
2464 table->flags &= ~RT6_TABLE_HAS_DFLT_ROUTER;
2465}
2466
2467void rt6_purge_dflt_routers(struct net *net)
2468{
2469 struct fib6_table *table;
2470 struct hlist_head *head;
2471 unsigned int h;
2472
2473 rcu_read_lock();
2474
2475 for (h = 0; h < FIB6_TABLE_HASHSZ; h++) {
2476 head = &net->ipv6.fib_table_hash[h];
2477 hlist_for_each_entry_rcu(table, head, tb6_hlist) {
2478 if (table->flags & RT6_TABLE_HAS_DFLT_ROUTER)
2479 __rt6_purge_dflt_routers(table);
2480 }
2481 }
2482
2483 rcu_read_unlock();
1da177e4
LT		 2484}
2485
5578689a
DL		 2486static void rtmsg_to_fib6_config(struct net *net,
2487 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2488 struct fib6_config *cfg)
2489{
2490 memset(cfg, 0, sizeof(*cfg));
2491
ca254490
DA		 2492 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2493 : RT6_TABLE_MAIN;
86872cb5
TG		 2494 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2495 cfg->fc_metric = rtmsg->rtmsg_metric;
2496 cfg->fc_expires = rtmsg->rtmsg_info;
2497 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2498 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2499 cfg->fc_flags = rtmsg->rtmsg_flags;
2500
5578689a 2501 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2502
4e3fd7a0
AD		 2503 cfg->fc_dst = rtmsg->rtmsg_dst;
2504 cfg->fc_src = rtmsg->rtmsg_src;
2505 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2506}
2507
5578689a 2508int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2509{
86872cb5 2510 struct fib6_config cfg;
1da177e4
LT		 2511 struct in6_rtmsg rtmsg;
2512 int err;
2513
67ba4152 2514 switch (cmd) {
1da177e4
LT		 2515 case SIOCADDRT: /* Add a route */
2516 case SIOCDELRT: /* Delete a route */
af31f412 2517 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2518 return -EPERM;
2519 err = copy_from_user(&rtmsg, arg,
2520 sizeof(struct in6_rtmsg));
2521 if (err)
2522 return -EFAULT;
86872cb5 2523
5578689a 2524 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2525
1da177e4
LT		 2526 rtnl_lock();
2527 switch (cmd) {
2528 case SIOCADDRT:
86872cb5 2529 err = ip6_route_add(&cfg);
1da177e4
LT		 2530 break;
2531 case SIOCDELRT:
86872cb5 2532 err = ip6_route_del(&cfg);
1da177e4
LT		 2533 break;
2534 default:
2535 err = -EINVAL;
2536 }
2537 rtnl_unlock();
2538
2539 return err;
3ff50b79 2540 }
1da177e4
LT		 2541
2542 return -EINVAL;
2543}
2544
2545/*
2546 * Drop the packet on the floor
2547 */
2548
d5fdd6ba 2549static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2550{
612f09e8 2551 int type;
adf30907 2552 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2553 switch (ipstats_mib_noroutes) {
2554 case IPSTATS_MIB_INNOROUTES:
0660e03f 2555 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2556 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2557 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2558 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2559 break;
2560 }
2561 /* FALLTHROUGH */
2562 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2563 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2564 ipstats_mib_noroutes);
612f09e8
YH		 2565 break;
2566 }
3ffe533c 2567 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2568 kfree_skb(skb);
2569 return 0;
2570}
2571
9ce8ade0
TG		 2572static int ip6_pkt_discard(struct sk_buff *skb)
2573{
612f09e8 2574 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2575}
2576
ede2059d 2577static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
1da177e4 2578{
adf30907 2579 skb->dev = skb_dst(skb)->dev;
612f09e8 2580 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2581}
2582
9ce8ade0
TG		 2583static int ip6_pkt_prohibit(struct sk_buff *skb)
2584{
612f09e8 2585 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2586}
2587
ede2059d 2588static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
9ce8ade0 2589{
adf30907 2590 skb->dev = skb_dst(skb)->dev;
612f09e8 2591 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2592}
2593
1da177e4
LT		 2594/*
2595 * Allocate a dst for local (unicast / anycast) address.
2596 */
2597
2598struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2599 const struct in6_addr *addr,
8f031519 2600 bool anycast)
1da177e4 2601{
ca254490 2602 u32 tb_id;
c346dca1 2603 struct net *net = dev_net(idev->dev);
5f02ce24
DA		 2604 struct net_device *dev = net->loopback_dev;
2605 struct rt6_info *rt;
2606
2607 /* use L3 Master device as loopback for host routes if device
2608 * is enslaved and address is not link local or multicast
2609 */
2610 if (!rt6_need_strict(addr))
2611 dev = l3mdev_master_dev_rcu(idev->dev) ? : dev;
2612
2613 rt = ip6_dst_alloc(net, dev, DST_NOCOUNT);
a3300ef4 2614 if (!rt)
1da177e4
LT		 2615 return ERR_PTR(-ENOMEM);
2616
1da177e4
LT		 2617 in6_dev_hold(idev);
2618
11d53b49 2619 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2620 rt->dst.input = ip6_input;
2621 rt->dst.output = ip6_output;
1da177e4 2622 rt->rt6i_idev = idev;
1da177e4
LT		 2623
2624 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2625 if (anycast)
2626 rt->rt6i_flags |= RTF_ANYCAST;
2627 else
1da177e4 2628 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2629
550bab42 2630 rt->rt6i_gateway = *addr;
4e3fd7a0 2631 rt->rt6i_dst.addr = *addr;
1da177e4 2632 rt->rt6i_dst.plen = 128;
ca254490
DA		 2633 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2634 rt->rt6i_table = fib6_get_table(net, tb_id);
8e3d5be7 2635 rt->dst.flags |= DST_NOCACHE;
1da177e4 2636
d8d1f30b 2637 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2638
2639 return rt;
2640}
2641
c3968a85
DW		 2642/* remove deleted ip from prefsrc entries */
2643struct arg_dev_net_ip {
2644 struct net_device *dev;
2645 struct net *net;
2646 struct in6_addr *addr;
2647};
2648
2649static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2650{
2651 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2652 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2653 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2654
d1918542 2655 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2656 rt != net->ipv6.ip6_null_entry &&
2657 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2658 /* remove prefsrc entry */
2659 rt->rt6i_prefsrc.plen = 0;
2660 }
2661 return 0;
2662}
2663
2664void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2665{
2666 struct net *net = dev_net(ifp->idev->dev);
2667 struct arg_dev_net_ip adni = {
2668 .dev = ifp->idev->dev,
2669 .net = net,
2670 .addr = &ifp->addr,
2671 };
0c3584d5 2672 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2673}
2674
be7a010d
DJ		 2675#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2676#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2677
2678/* Remove routers and update dst entries when gateway turn into host. */
2679static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2680{
2681 struct in6_addr *gateway = (struct in6_addr *)arg;
2682
2683 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2684 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2685 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2686 return -1;
2687 }
2688 return 0;
2689}
2690
2691void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2692{
2693 fib6_clean_all(net, fib6_clean_tohost, gateway);
2694}
2695
8ed67789
DL		 2696struct arg_dev_net {
2697 struct net_device *dev;
2698 struct net *net;
2699};
2700
1da177e4
LT		 2701static int fib6_ifdown(struct rt6_info *rt, void *arg)
2702{
bc3ef660 2703 const struct arg_dev_net *adn = arg;
2704 const struct net_device *dev = adn->dev;
8ed67789 2705
d1918542 2706 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2707 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2708 return -1;
c159d30c 2709
1da177e4
LT		 2710 return 0;
2711}
2712
f3db4851 2713void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2714{
8ed67789
DL		 2715 struct arg_dev_net adn = {
2716 .dev = dev,
2717 .net = net,
2718 };
2719
0c3584d5 2720 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2721 icmp6_clean_all(fib6_ifdown, &adn);
e332bc67
EB		 2722 if (dev)
2723 rt6_uncached_list_flush_dev(net, dev);
1da177e4
LT		 2724}
2725
95c96174 2726struct rt6_mtu_change_arg {
1da177e4 2727 struct net_device *dev;
95c96174 2728 unsigned int mtu;
1da177e4
LT		 2729};
2730
2731static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2732{
2733 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2734 struct inet6_dev *idev;
2735
2736 /* In IPv6 pmtu discovery is not optional,
2737 so that RTAX_MTU lock cannot disable it.
2738 We still use this lock to block changes
2739 caused by addrconf/ndisc.
2740 */
2741
2742 idev = __in6_dev_get(arg->dev);
38308473 2743 if (!idev)
1da177e4
LT		 2744 return 0;
2745
2746 /* For administrative MTU increase, there is no way to discover
2747 IPv6 PMTU increase, so PMTU increase should be updated here.
2748 Since RFC 1981 doesn't include administrative MTU increase
2749 update PMTU increase is a MUST. (i.e. jumbo frame)
2750 */
2751 /*
2752 If new MTU is less than route PMTU, this new MTU will be the
2753 lowest MTU in the path, update the route PMTU to reflect PMTU
2754 decreases; if new MTU is greater than route PMTU, and the
2755 old MTU is the lowest MTU in the path, update the route PMTU
2756 to reflect the increase. In this case if the other nodes' MTU
2757 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2758 PMTU discouvery.
2759 */
d1918542 2760 if (rt->dst.dev == arg->dev &&
4b32b5ad
MKL		 2761 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2762 if (rt->rt6i_flags & RTF_CACHE) {
2763 /* For RTF_CACHE with rt6i_pmtu == 0
2764 * (i.e. a redirected route),
2765 * the metrics of its rt->dst.from has already
2766 * been updated.
2767 */
2768 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2769 rt->rt6i_pmtu = arg->mtu;
2770 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2771 (dst_mtu(&rt->dst) < arg->mtu &&
2772 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2773 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2774 }
566cfd8f 2775 }
1da177e4
LT		 2776 return 0;
2777}
2778
95c96174 2779void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2780{
c71099ac
TG		 2781 struct rt6_mtu_change_arg arg = {
2782 .dev = dev,
2783 .mtu = mtu,
2784 };
1da177e4 2785
0c3584d5 2786 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2787}
2788
ef7c79ed 2789static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2790 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2791 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2792 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2793 [RTA_PRIORITY] = { .type = NLA_U32 },
2794 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2795 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2796 [RTA_PREF] = { .type = NLA_U8 },
19e42e45
RP		 2797 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2798 [RTA_ENCAP] = { .type = NLA_NESTED },
32bc201e 2799 [RTA_EXPIRES] = { .type = NLA_U32 },
622ec2c9 2800 [RTA_UID] = { .type = NLA_U32 },
86872cb5
TG		 2801};
2802
2803static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2804 struct fib6_config *cfg)
1da177e4 2805{
86872cb5
TG		 2806 struct rtmsg *rtm;
2807 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2808 unsigned int pref;
86872cb5 2809 int err;
1da177e4 2810
86872cb5
TG		 2811 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2812 if (err < 0)
2813 goto errout;
1da177e4 2814
86872cb5
TG		 2815 err = -EINVAL;
2816 rtm = nlmsg_data(nlh);
2817 memset(cfg, 0, sizeof(*cfg));
2818
2819 cfg->fc_table = rtm->rtm_table;
2820 cfg->fc_dst_len = rtm->rtm_dst_len;
2821 cfg->fc_src_len = rtm->rtm_src_len;
2822 cfg->fc_flags = RTF_UP;
2823 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2824 cfg->fc_type = rtm->rtm_type;
86872cb5 2825
ef2c7d7b
ND		 2826 if (rtm->rtm_type == RTN_UNREACHABLE ||
2827 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2828 rtm->rtm_type == RTN_PROHIBIT ||
2829 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2830 cfg->fc_flags |= RTF_REJECT;
2831
ab79ad14
MŻ		 2832 if (rtm->rtm_type == RTN_LOCAL)
2833 cfg->fc_flags |= RTF_LOCAL;
2834
1f56a01f
MKL		 2835 if (rtm->rtm_flags & RTM_F_CLONED)
2836 cfg->fc_flags |= RTF_CACHE;
2837
15e47304 2838 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2839 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2840 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2841
2842 if (tb[RTA_GATEWAY]) {
67b61f6c 2843 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2844 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2845 }
86872cb5
TG		 2846
2847 if (tb[RTA_DST]) {
2848 int plen = (rtm->rtm_dst_len + 7) >> 3;
2849
2850 if (nla_len(tb[RTA_DST]) < plen)
2851 goto errout;
2852
2853 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2854 }
86872cb5
TG		 2855
2856 if (tb[RTA_SRC]) {
2857 int plen = (rtm->rtm_src_len + 7) >> 3;
2858
2859 if (nla_len(tb[RTA_SRC]) < plen)
2860 goto errout;
2861
2862 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2863 }
86872cb5 2864
c3968a85 2865 if (tb[RTA_PREFSRC])
67b61f6c 2866 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2867
86872cb5
TG		 2868 if (tb[RTA_OIF])
2869 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2870
2871 if (tb[RTA_PRIORITY])
2872 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2873
2874 if (tb[RTA_METRICS]) {
2875 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2876 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2877 }
86872cb5
TG		 2878
2879 if (tb[RTA_TABLE])
2880 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2881
51ebd318
ND		 2882 if (tb[RTA_MULTIPATH]) {
2883 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2884 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2885 }
2886
c78ba6d6
LR		 2887 if (tb[RTA_PREF]) {
2888 pref = nla_get_u8(tb[RTA_PREF]);
2889 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2890 pref != ICMPV6_ROUTER_PREF_HIGH)
2891 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2892 cfg->fc_flags |= RTF_PREF(pref);
2893 }
2894
19e42e45
RP		 2895 if (tb[RTA_ENCAP])
2896 cfg->fc_encap = tb[RTA_ENCAP];
2897
2898 if (tb[RTA_ENCAP_TYPE])
2899 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2900
32bc201e
XL		 2901 if (tb[RTA_EXPIRES]) {
2902 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2903
2904 if (addrconf_finite_timeout(timeout)) {
2905 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2906 cfg->fc_flags |= RTF_EXPIRES;
2907 }
2908 }
2909
86872cb5
TG		 2910 err = 0;
2911errout:
2912 return err;
1da177e4
LT		 2913}
2914
6b9ea5a6
RP		 2915struct rt6_nh {
2916 struct rt6_info *rt6_info;
2917 struct fib6_config r_cfg;
2918 struct mx6_config mxc;
2919 struct list_head next;
2920};
2921
2922static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2923{
2924 struct rt6_nh *nh;
2925
2926 list_for_each_entry(nh, rt6_nh_list, next) {
2927 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2928 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2929 nh->r_cfg.fc_ifindex);
2930 }
2931}
2932
2933static int ip6_route_info_append(struct list_head *rt6_nh_list,
2934 struct rt6_info *rt, struct fib6_config *r_cfg)
2935{
2936 struct rt6_nh *nh;
2937 struct rt6_info *rtnh;
2938 int err = -EEXIST;
2939
2940 list_for_each_entry(nh, rt6_nh_list, next) {
2941 /* check if rt6_info already exists */
2942 rtnh = nh->rt6_info;
2943
2944 if (rtnh->dst.dev == rt->dst.dev &&
2945 rtnh->rt6i_idev == rt->rt6i_idev &&
2946 ipv6_addr_equal(&rtnh->rt6i_gateway,
2947 &rt->rt6i_gateway))
2948 return err;
2949 }
2950
2951 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2952 if (!nh)
2953 return -ENOMEM;
2954 nh->rt6_info = rt;
2955 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2956 if (err) {
2957 kfree(nh);
2958 return err;
2959 }
2960 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2961 list_add_tail(&nh->next, rt6_nh_list);
2962
2963 return 0;
2964}
2965
2966static int ip6_route_multipath_add(struct fib6_config *cfg)
51ebd318
ND		 2967{
2968 struct fib6_config r_cfg;
2969 struct rtnexthop *rtnh;
6b9ea5a6
RP		 2970 struct rt6_info *rt;
2971 struct rt6_nh *err_nh;
2972 struct rt6_nh *nh, *nh_safe;
51ebd318
ND		 2973 int remaining;
2974 int attrlen;
6b9ea5a6
RP		 2975 int err = 1;
2976 int nhn = 0;
2977 int replace = (cfg->fc_nlinfo.nlh &&
2978 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2979 LIST_HEAD(rt6_nh_list);
51ebd318 2980
35f1b4e9 2981 remaining = cfg->fc_mp_len;
51ebd318 2982 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318 2983
6b9ea5a6
RP		 2984 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2985 * rt6_info structs per nexthop
2986 */
51ebd318
ND		 2987 while (rtnh_ok(rtnh, remaining)) {
2988 memcpy(&r_cfg, cfg, sizeof(*cfg));
2989 if (rtnh->rtnh_ifindex)
2990 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2991
2992 attrlen = rtnh_attrlen(rtnh);
2993 if (attrlen > 0) {
2994 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2995
2996 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2997 if (nla) {
67b61f6c 2998 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 2999 r_cfg.fc_flags |= RTF_GATEWAY;
3000 }
19e42e45
RP		 3001 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
3002 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
3003 if (nla)
3004 r_cfg.fc_encap_type = nla_get_u16(nla);
51ebd318 3005 }
6b9ea5a6 3006
8c5b83f0
RP		 3007 rt = ip6_route_info_create(&r_cfg);
3008 if (IS_ERR(rt)) {
3009 err = PTR_ERR(rt);
3010 rt = NULL;
6b9ea5a6 3011 goto cleanup;
8c5b83f0 3012 }
6b9ea5a6
RP		 3013
3014 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
51ebd318 3015 if (err) {
6b9ea5a6
RP		 3016 dst_free(&rt->dst);
3017 goto cleanup;
3018 }
3019
3020 rtnh = rtnh_next(rtnh, &remaining);
3021 }
3022
3023 err_nh = NULL;
3024 list_for_each_entry(nh, &rt6_nh_list, next) {
3025 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
3026 /* nh->rt6_info is used or freed at this point, reset to NULL*/
3027 nh->rt6_info = NULL;
3028 if (err) {
3029 if (replace && nhn)
3030 ip6_print_replace_route_err(&rt6_nh_list);
3031 err_nh = nh;
3032 goto add_errout;
51ebd318 3033 }
6b9ea5a6 3034
1a72418b 3035 /* Because each route is added like a single route we remove
27596472
MK		 3036 * these flags after the first nexthop: if there is a collision,
3037 * we have already failed to add the first nexthop:
3038 * fib6_add_rt2node() has rejected it; when replacing, old
3039 * nexthops have been replaced by first new, the rest should
3040 * be added to it.
1a72418b 3041 */
27596472
MK		 3042 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
3043 NLM_F_REPLACE);
6b9ea5a6
RP		 3044 nhn++;
3045 }
3046
3047 goto cleanup;
3048
3049add_errout:
3050 /* Delete routes that were already added */
3051 list_for_each_entry(nh, &rt6_nh_list, next) {
3052 if (err_nh == nh)
3053 break;
3054 ip6_route_del(&nh->r_cfg);
3055 }
3056
3057cleanup:
3058 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3059 if (nh->rt6_info)
3060 dst_free(&nh->rt6_info->dst);
52fe51f8 3061 kfree(nh->mxc.mx);
6b9ea5a6
RP		 3062 list_del(&nh->next);
3063 kfree(nh);
3064 }
3065
3066 return err;
3067}
3068
3069static int ip6_route_multipath_del(struct fib6_config *cfg)
3070{
3071 struct fib6_config r_cfg;
3072 struct rtnexthop *rtnh;
3073 int remaining;
3074 int attrlen;
3075 int err = 1, last_err = 0;
3076
3077 remaining = cfg->fc_mp_len;
3078 rtnh = (struct rtnexthop *)cfg->fc_mp;
3079
3080 /* Parse a Multipath Entry */
3081 while (rtnh_ok(rtnh, remaining)) {
3082 memcpy(&r_cfg, cfg, sizeof(*cfg));
3083 if (rtnh->rtnh_ifindex)
3084 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3085
3086 attrlen = rtnh_attrlen(rtnh);
3087 if (attrlen > 0) {
3088 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3089
3090 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3091 if (nla) {
3092 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3093 r_cfg.fc_flags |= RTF_GATEWAY;
3094 }
3095 }
3096 err = ip6_route_del(&r_cfg);
3097 if (err)
3098 last_err = err;
3099
51ebd318
ND		 3100 rtnh = rtnh_next(rtnh, &remaining);
3101 }
3102
3103 return last_err;
3104}
3105
67ba4152 3106static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3107{
86872cb5
TG		 3108 struct fib6_config cfg;
3109 int err;
1da177e4 3110
86872cb5
TG		 3111 err = rtm_to_fib6_config(skb, nlh, &cfg);
3112 if (err < 0)
3113 return err;
3114
51ebd318 3115 if (cfg.fc_mp)
6b9ea5a6 3116 return ip6_route_multipath_del(&cfg);
51ebd318
ND		 3117 else
3118 return ip6_route_del(&cfg);
1da177e4
LT		 3119}
3120
67ba4152 3121static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3122{
86872cb5
TG		 3123 struct fib6_config cfg;
3124 int err;
1da177e4 3125
86872cb5
TG		 3126 err = rtm_to_fib6_config(skb, nlh, &cfg);
3127 if (err < 0)
3128 return err;
3129
51ebd318 3130 if (cfg.fc_mp)
6b9ea5a6 3131 return ip6_route_multipath_add(&cfg);
51ebd318
ND		 3132 else
3133 return ip6_route_add(&cfg);
1da177e4
LT		 3134}
3135
19e42e45 3136static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
339bf98f
TG		 3137{
3138 return NLMSG_ALIGN(sizeof(struct rtmsg))
3139 + nla_total_size(16) /* RTA_SRC */
3140 + nla_total_size(16) /* RTA_DST */
3141 + nla_total_size(16) /* RTA_GATEWAY */
3142 + nla_total_size(16) /* RTA_PREFSRC */
3143 + nla_total_size(4) /* RTA_TABLE */
3144 + nla_total_size(4) /* RTA_IIF */
3145 + nla_total_size(4) /* RTA_OIF */
3146 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 3147 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 3148 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6 3149 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
19e42e45 3150 + nla_total_size(1) /* RTA_PREF */
61adedf3 3151 + lwtunnel_get_encap_size(rt->dst.lwtstate);
339bf98f
TG		 3152}
3153
191cd582
BH		 3154static int rt6_fill_node(struct net *net,
3155 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 3156 struct in6_addr *dst, struct in6_addr *src,
15e47304 3157 int iif, int type, u32 portid, u32 seq,
7bc570c8 3158 int prefix, int nowait, unsigned int flags)
1da177e4 3159{
4b32b5ad 3160 u32 metrics[RTAX_MAX];
1da177e4 3161 struct rtmsg *rtm;
2d7202bf 3162 struct nlmsghdr *nlh;
e3703b3d 3163 long expires;
9e762a4a 3164 u32 table;
1da177e4
LT		 3165
3166 if (prefix) { /* user wants prefix routes only */
3167 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3168 /* success since this is not a prefix route */
3169 return 1;
3170 }
3171 }
3172
15e47304 3173 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 3174 if (!nlh)
26932566 3175 return -EMSGSIZE;
2d7202bf
TG		 3176
3177 rtm = nlmsg_data(nlh);
1da177e4
LT		 3178 rtm->rtm_family = AF_INET6;
3179 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3180 rtm->rtm_src_len = rt->rt6i_src.plen;
3181 rtm->rtm_tos = 0;
c71099ac 3182 if (rt->rt6i_table)
9e762a4a 3183 table = rt->rt6i_table->tb6_id;
c71099ac 3184 else
9e762a4a
PM		 3185 table = RT6_TABLE_UNSPEC;
3186 rtm->rtm_table = table;
c78679e8
DM		 3187 if (nla_put_u32(skb, RTA_TABLE, table))
3188 goto nla_put_failure;
ef2c7d7b
ND		 3189 if (rt->rt6i_flags & RTF_REJECT) {
3190 switch (rt->dst.error) {
3191 case -EINVAL:
3192 rtm->rtm_type = RTN_BLACKHOLE;
3193 break;
3194 case -EACCES:
3195 rtm->rtm_type = RTN_PROHIBIT;
3196 break;
b4949ab2
ND		 3197 case -EAGAIN:
3198 rtm->rtm_type = RTN_THROW;
3199 break;
ef2c7d7b
ND		 3200 default:
3201 rtm->rtm_type = RTN_UNREACHABLE;
3202 break;
3203 }
3204 }
38308473 3205 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 3206 rtm->rtm_type = RTN_LOCAL;
d1918542 3207 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 3208 rtm->rtm_type = RTN_LOCAL;
3209 else
3210 rtm->rtm_type = RTN_UNICAST;
3211 rtm->rtm_flags = 0;
35103d11 3212 if (!netif_carrier_ok(rt->dst.dev)) {
cea45e20 3213 rtm->rtm_flags |= RTNH_F_LINKDOWN;
35103d11
AG		 3214 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3215 rtm->rtm_flags |= RTNH_F_DEAD;
3216 }
1da177e4
LT		 3217 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3218 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 3219 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 3220 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 3221 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3222 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3223 rtm->rtm_protocol = RTPROT_RA;
3224 else
3225 rtm->rtm_protocol = RTPROT_KERNEL;
3226 }
1da177e4 3227
38308473 3228 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 3229 rtm->rtm_flags |= RTM_F_CLONED;
3230
3231 if (dst) {
930345ea 3232 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 3233 goto nla_put_failure;
1ab1457c 3234 rtm->rtm_dst_len = 128;
1da177e4 3235 } else if (rtm->rtm_dst_len)
930345ea 3236 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 3237 goto nla_put_failure;
1da177e4
LT		 3238#ifdef CONFIG_IPV6_SUBTREES
3239 if (src) {
930345ea 3240 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 3241 goto nla_put_failure;
1ab1457c 3242 rtm->rtm_src_len = 128;
c78679e8 3243 } else if (rtm->rtm_src_len &&
930345ea 3244 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 3245 goto nla_put_failure;
1da177e4 3246#endif
7bc570c8
YH		 3247 if (iif) {
3248#ifdef CONFIG_IPV6_MROUTE
3249 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
2cf75070
NA		 3250 int err = ip6mr_get_route(net, skb, rtm, nowait,
3251 portid);
3252
7bc570c8
YH		 3253 if (err <= 0) {
3254 if (!nowait) {
3255 if (err == 0)
3256 return 0;
3257 goto nla_put_failure;
3258 } else {
3259 if (err == -EMSGSIZE)
3260 goto nla_put_failure;
3261 }
3262 }
3263 } else
3264#endif
c78679e8
DM		 3265 if (nla_put_u32(skb, RTA_IIF, iif))
3266 goto nla_put_failure;
7bc570c8 3267 } else if (dst) {
1da177e4 3268 struct in6_addr saddr_buf;
c78679e8 3269 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 3270 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3271 goto nla_put_failure;
1da177e4 3272 }
2d7202bf 3273
c3968a85
DW		 3274 if (rt->rt6i_prefsrc.plen) {
3275 struct in6_addr saddr_buf;
4e3fd7a0 3276 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 3277 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3278 goto nla_put_failure;
c3968a85
DW		 3279 }
3280
4b32b5ad
MKL		 3281 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3282 if (rt->rt6i_pmtu)
3283 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3284 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 3285 goto nla_put_failure;
3286
dd0cbf29 3287 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 3288 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 3289 goto nla_put_failure;
94f826b8 3290 }
2d7202bf 3291
c78679e8
DM		 3292 if (rt->dst.dev &&
3293 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3294 goto nla_put_failure;
3295 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3296 goto nla_put_failure;
8253947e
LW		 3297
3298 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 3299
87a50699 3300 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 3301 goto nla_put_failure;
2d7202bf 3302
c78ba6d6
LR		 3303 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3304 goto nla_put_failure;
3305
61adedf3 3306 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
19e42e45 3307
053c095a
JB		 3308 nlmsg_end(skb, nlh);
3309 return 0;
2d7202bf
TG		 3310
3311nla_put_failure:
26932566
PM		 3312 nlmsg_cancel(skb, nlh);
3313 return -EMSGSIZE;
1da177e4
LT		 3314}
3315
1b43af54 3316int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 3317{
3318 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3319 int prefix;
3320
2d7202bf
TG		 3321 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3322 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 3323 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3324 } else
3325 prefix = 0;
3326
191cd582
BH		 3327 return rt6_fill_node(arg->net,
3328 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 3329 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 3330 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 3331}
3332
67ba4152 3333static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 3334{
3b1e0a65 3335 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 3336 struct nlattr *tb[RTA_MAX+1];
3337 struct rt6_info *rt;
1da177e4 3338 struct sk_buff *skb;
ab364a6f 3339 struct rtmsg *rtm;
4c9483b2 3340 struct flowi6 fl6;
72331bc0 3341 int err, iif = 0, oif = 0;
1da177e4 3342
ab364a6f
TG		 3343 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3344 if (err < 0)
3345 goto errout;
1da177e4 3346
ab364a6f 3347 err = -EINVAL;
4c9483b2 3348 memset(&fl6, 0, sizeof(fl6));
38b7097b
HFS		 3349 rtm = nlmsg_data(nlh);
3350 fl6.flowlabel = ip6_make_flowinfo(rtm->rtm_tos, 0);
1da177e4 3351
ab364a6f
TG		 3352 if (tb[RTA_SRC]) {
3353 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3354 goto errout;
3355
4e3fd7a0 3356 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 3357 }
3358
3359 if (tb[RTA_DST]) {
3360 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3361 goto errout;
3362
4e3fd7a0 3363 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 3364 }
3365
3366 if (tb[RTA_IIF])
3367 iif = nla_get_u32(tb[RTA_IIF]);
3368
3369 if (tb[RTA_OIF])
72331bc0 3370 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 3371
2e47b291
LC		 3372 if (tb[RTA_MARK])
3373 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3374
622ec2c9
LC		 3375 if (tb[RTA_UID])
3376 fl6.flowi6_uid = make_kuid(current_user_ns(),
3377 nla_get_u32(tb[RTA_UID]));
3378 else
3379 fl6.flowi6_uid = iif ? INVALID_UID : current_uid();
3380
1da177e4
LT		 3381 if (iif) {
3382 struct net_device *dev;
72331bc0
SL		 3383 int flags = 0;
3384
5578689a 3385 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 3386 if (!dev) {
3387 err = -ENODEV;
ab364a6f 3388 goto errout;
1da177e4 3389 }
72331bc0
SL		 3390
3391 fl6.flowi6_iif = iif;
3392
3393 if (!ipv6_addr_any(&fl6.saddr))
3394 flags |= RT6_LOOKUP_F_HAS_SADDR;
3395
3396 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3397 flags);
3398 } else {
3399 fl6.flowi6_oif = oif;
3400
3401 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 3402 }
3403
ab364a6f 3404 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 3405 if (!skb) {
94e187c0 3406 ip6_rt_put(rt);
ab364a6f
TG		 3407 err = -ENOBUFS;
3408 goto errout;
3409 }
1da177e4 3410
ab364a6f
TG		 3411 /* Reserve room for dummy headers, this skb can pass
3412 through good chunk of routing engine.
3413 */
459a98ed 3414 skb_reset_mac_header(skb);
ab364a6f 3415 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 3416
d8d1f30b 3417 skb_dst_set(skb, &rt->dst);
1da177e4 3418
4c9483b2 3419 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 3420 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 3421 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 3422 if (err < 0) {
ab364a6f
TG		 3423 kfree_skb(skb);
3424 goto errout;
1da177e4
LT		 3425 }
3426
15e47304 3427 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 3428errout:
1da177e4 3429 return err;
1da177e4
LT		 3430}
3431
37a1d361
RP		 3432void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3433 unsigned int nlm_flags)
1da177e4
LT		 3434{
3435 struct sk_buff *skb;
5578689a 3436 struct net *net = info->nl_net;
528c4ceb
DL		 3437 u32 seq;
3438 int err;
3439
3440 err = -ENOBUFS;
38308473 3441 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 3442
19e42e45 3443 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
38308473 3444 if (!skb)
21713ebc
TG		 3445 goto errout;
3446
191cd582 3447 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
37a1d361 3448 event, info->portid, seq, 0, 0, nlm_flags);
26932566
PM		 3449 if (err < 0) {
3450 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3451 WARN_ON(err == -EMSGSIZE);
3452 kfree_skb(skb);
3453 goto errout;
3454 }
15e47304 3455 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 3456 info->nlh, gfp_any());
3457 return;
21713ebc
TG		 3458errout:
3459 if (err < 0)
5578689a 3460 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 3461}
3462
8ed67789 3463static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 3464 unsigned long event, void *ptr)
8ed67789 3465{
351638e7 3466 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 3467 struct net *net = dev_net(dev);
8ed67789
DL		 3468
3469 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 3470 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 3471 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3472#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3473 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 3474 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 3475 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 3476 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3477#endif
3478 }
3479
3480 return NOTIFY_OK;
3481}
3482
1da177e4
LT		 3483/*
3484 * /proc
3485 */
3486
3487#ifdef CONFIG_PROC_FS
3488
33120b30
AD		 3489static const struct file_operations ipv6_route_proc_fops = {
3490 .owner = THIS_MODULE,
3491 .open = ipv6_route_open,
3492 .read = seq_read,
3493 .llseek = seq_lseek,
8d2ca1d7 3494 .release = seq_release_net,
33120b30
AD		 3495};
3496
1da177e4
LT		 3497static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3498{
69ddb805 3499 struct net *net = (struct net *)seq->private;
1da177e4 3500 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 3501 net->ipv6.rt6_stats->fib_nodes,
3502 net->ipv6.rt6_stats->fib_route_nodes,
3503 net->ipv6.rt6_stats->fib_rt_alloc,
3504 net->ipv6.rt6_stats->fib_rt_entries,
3505 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 3506 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 3507 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 3508
3509 return 0;
3510}
3511
3512static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3513{
de05c557 3514 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 3515}
3516
9a32144e 3517static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 3518 .owner = THIS_MODULE,
3519 .open = rt6_stats_seq_open,
3520 .read = seq_read,
3521 .llseek = seq_lseek,
b6fcbdb4 3522 .release = single_release_net,
1da177e4
LT		 3523};
3524#endif /* CONFIG_PROC_FS */
3525
3526#ifdef CONFIG_SYSCTL
3527
1da177e4 3528static
fe2c6338 3529int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3530 void __user *buffer, size_t *lenp, loff_t *ppos)
3531{
c486da34
LAG		 3532 struct net *net;
3533 int delay;
3534 if (!write)
1da177e4 3535 return -EINVAL;
c486da34
LAG		 3536
3537 net = (struct net *)ctl->extra1;
3538 delay = net->ipv6.sysctl.flush_delay;
3539 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3540 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3541 return 0;
1da177e4
LT		 3542}
3543
fe2c6338 3544struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3545 {
1da177e4 3546 .procname = "flush",
4990509f 3547 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3548 .maxlen = sizeof(int),
89c8b3a1 3549 .mode = 0200,
6d9f239a 3550 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3551 },
3552 {
1da177e4 3553 .procname = "gc_thresh",
9a7ec3a9 3554 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3555 .maxlen = sizeof(int),
3556 .mode = 0644,
6d9f239a 3557 .proc_handler = proc_dointvec,
1da177e4
LT		 3558 },
3559 {
1da177e4 3560 .procname = "max_size",
4990509f 3561 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3562 .maxlen = sizeof(int),
3563 .mode = 0644,
6d9f239a 3564 .proc_handler = proc_dointvec,
1da177e4
LT		 3565 },
3566 {
1da177e4 3567 .procname = "gc_min_interval",
4990509f 3568 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3569 .maxlen = sizeof(int),
3570 .mode = 0644,
6d9f239a 3571 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3572 },
3573 {
1da177e4 3574 .procname = "gc_timeout",
4990509f 3575 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3576 .maxlen = sizeof(int),
3577 .mode = 0644,
6d9f239a 3578 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3579 },
3580 {
1da177e4 3581 .procname = "gc_interval",
4990509f 3582 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3583 .maxlen = sizeof(int),
3584 .mode = 0644,
6d9f239a 3585 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3586 },
3587 {
1da177e4 3588 .procname = "gc_elasticity",
4990509f 3589 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3590 .maxlen = sizeof(int),
3591 .mode = 0644,
f3d3f616 3592 .proc_handler = proc_dointvec,
1da177e4
LT		 3593 },
3594 {
1da177e4 3595 .procname = "mtu_expires",
4990509f 3596 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3597 .maxlen = sizeof(int),
3598 .mode = 0644,
6d9f239a 3599 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3600 },
3601 {
1da177e4 3602 .procname = "min_adv_mss",
4990509f 3603 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3604 .maxlen = sizeof(int),
3605 .mode = 0644,
f3d3f616 3606 .proc_handler = proc_dointvec,
1da177e4
LT		 3607 },
3608 {
1da177e4 3609 .procname = "gc_min_interval_ms",
4990509f 3610 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3611 .maxlen = sizeof(int),
3612 .mode = 0644,
6d9f239a 3613 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3614 },
f8572d8f 3615 { }
1da177e4
LT		 3616};
3617
2c8c1e72 3618struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3619{
3620 struct ctl_table *table;
3621
3622 table = kmemdup(ipv6_route_table_template,
3623 sizeof(ipv6_route_table_template),
3624 GFP_KERNEL);
5ee09105
YH		 3625
3626 if (table) {
3627 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3628 table[0].extra1 = net;
86393e52 3629 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3630 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3631 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3632 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3633 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3634 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3635 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3636 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3637 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3638
3639 /* Don't export sysctls to unprivileged users */
3640 if (net->user_ns != &init_user_ns)
3641 table[0].procname = NULL;
5ee09105
YH		 3642 }
3643
760f2d01
DL		 3644 return table;
3645}
1da177e4
LT		 3646#endif
3647
2c8c1e72 3648static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3649{
633d424b 3650 int ret = -ENOMEM;
8ed67789 3651
86393e52
AD		 3652 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3653 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3654
fc66f95c
ED		 3655 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3656 goto out_ip6_dst_ops;
3657
8ed67789
DL		 3658 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3659 sizeof(*net->ipv6.ip6_null_entry),
3660 GFP_KERNEL);
3661 if (!net->ipv6.ip6_null_entry)
fc66f95c 3662 goto out_ip6_dst_entries;
d8d1f30b 3663 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3664 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3665 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3666 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3667 ip6_template_metrics, true);
8ed67789
DL		 3668
3669#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3670 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3671 sizeof(*net->ipv6.ip6_prohibit_entry),
3672 GFP_KERNEL);
68fffc67
PZ		 3673 if (!net->ipv6.ip6_prohibit_entry)
3674 goto out_ip6_null_entry;
d8d1f30b 3675 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3676 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3677 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3678 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3679 ip6_template_metrics, true);
8ed67789
DL		 3680
3681 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3682 sizeof(*net->ipv6.ip6_blk_hole_entry),
3683 GFP_KERNEL);
68fffc67
PZ		 3684 if (!net->ipv6.ip6_blk_hole_entry)
3685 goto out_ip6_prohibit_entry;
d8d1f30b 3686 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3687 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3688 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3689 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3690 ip6_template_metrics, true);
8ed67789
DL		 3691#endif
3692
b339a47c
PZ		 3693 net->ipv6.sysctl.flush_delay = 0;
3694 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3695 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3696 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3697 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3698 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3699 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3700 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3701
6891a346
BT		 3702 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3703
8ed67789
DL		 3704 ret = 0;
3705out:
3706 return ret;
f2fc6a54 3707
68fffc67
PZ		 3708#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3709out_ip6_prohibit_entry:
3710 kfree(net->ipv6.ip6_prohibit_entry);
3711out_ip6_null_entry:
3712 kfree(net->ipv6.ip6_null_entry);
3713#endif
fc66f95c
ED		 3714out_ip6_dst_entries:
3715 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3716out_ip6_dst_ops:
f2fc6a54 3717 goto out;
cdb18761
DL		 3718}
3719
2c8c1e72 3720static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3721{
8ed67789
DL		 3722 kfree(net->ipv6.ip6_null_entry);
3723#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3724 kfree(net->ipv6.ip6_prohibit_entry);
3725 kfree(net->ipv6.ip6_blk_hole_entry);
3726#endif
41bb78b4 3727 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3728}
3729
d189634e
TG		 3730static int __net_init ip6_route_net_init_late(struct net *net)
3731{
3732#ifdef CONFIG_PROC_FS
d4beaa66
G		 3733 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3734 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3735#endif
3736 return 0;
3737}
3738
3739static void __net_exit ip6_route_net_exit_late(struct net *net)
3740{
3741#ifdef CONFIG_PROC_FS
ece31ffd
G		 3742 remove_proc_entry("ipv6_route", net->proc_net);
3743 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3744#endif
3745}
3746
cdb18761
DL		 3747static struct pernet_operations ip6_route_net_ops = {
3748 .init = ip6_route_net_init,
3749 .exit = ip6_route_net_exit,
3750};
3751
c3426b47
DM		 3752static int __net_init ipv6_inetpeer_init(struct net *net)
3753{
3754 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3755
3756 if (!bp)
3757 return -ENOMEM;
3758 inet_peer_base_init(bp);
3759 net->ipv6.peers = bp;
3760 return 0;
3761}
3762
3763static void __net_exit ipv6_inetpeer_exit(struct net *net)
3764{
3765 struct inet_peer_base *bp = net->ipv6.peers;
3766
3767 net->ipv6.peers = NULL;
56a6b248 3768 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3769 kfree(bp);
3770}
3771
2b823f72 3772static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3773 .init = ipv6_inetpeer_init,
3774 .exit = ipv6_inetpeer_exit,
3775};
3776
d189634e
TG		 3777static struct pernet_operations ip6_route_net_late_ops = {
3778 .init = ip6_route_net_init_late,
3779 .exit = ip6_route_net_exit_late,
3780};
3781
8ed67789
DL		 3782static struct notifier_block ip6_route_dev_notifier = {
3783 .notifier_call = ip6_route_dev_notify,
3784 .priority = 0,
3785};
3786
433d49c3 3787int __init ip6_route_init(void)
1da177e4 3788{
433d49c3 3789 int ret;
8d0b94af 3790 int cpu;
433d49c3 3791
9a7ec3a9
DL		 3792 ret = -ENOMEM;
3793 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3794 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3795 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3796 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3797 goto out;
14e50e57 3798
fc66f95c 3799 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3800 if (ret)
bdb3289f 3801 goto out_kmem_cache;
bdb3289f 3802
c3426b47
DM		 3803 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3804 if (ret)
e8803b6c 3805 goto out_dst_entries;
2a0c451a 3806
7e52b33b
DM		 3807 ret = register_pernet_subsys(&ip6_route_net_ops);
3808 if (ret)
3809 goto out_register_inetpeer;
c3426b47 3810
5dc121e9
AE		 3811 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3812
8ed67789
DL		 3813 /* Registering of the loopback is done before this portion of code,
3814 * the loopback reference in rt6_info will not be taken, do it
3815 * manually for init_net */
d8d1f30b 3816 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3817 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3818 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3819 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3820 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3821 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3822 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3823 #endif
e8803b6c 3824 ret = fib6_init();
433d49c3 3825 if (ret)
8ed67789 3826 goto out_register_subsys;
433d49c3 3827
433d49c3
DL		 3828 ret = xfrm6_init();
3829 if (ret)
e8803b6c 3830 goto out_fib6_init;
c35b7e72 3831
433d49c3
DL		 3832 ret = fib6_rules_init();
3833 if (ret)
3834 goto xfrm6_init;
7e5449c2 3835
d189634e
TG		 3836 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3837 if (ret)
3838 goto fib6_rules_init;
3839
433d49c3 3840 ret = -ENOBUFS;
c7ac8679
GR		 3841 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3842 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3843 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3844 goto out_register_late_subsys;
c127ea2c 3845
8ed67789 3846 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3847 if (ret)
d189634e 3848 goto out_register_late_subsys;
8ed67789 3849
8d0b94af
MKL		 3850 for_each_possible_cpu(cpu) {
3851 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3852
3853 INIT_LIST_HEAD(&ul->head);
3854 spin_lock_init(&ul->lock);
3855 }
3856
433d49c3
DL		 3857out:
3858 return ret;
3859
d189634e
TG		 3860out_register_late_subsys:
3861 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3862fib6_rules_init:
433d49c3
DL		 3863 fib6_rules_cleanup();
3864xfrm6_init:
433d49c3 3865 xfrm6_fini();
2a0c451a
TG		 3866out_fib6_init:
3867 fib6_gc_cleanup();
8ed67789
DL		 3868out_register_subsys:
3869 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3870out_register_inetpeer:
3871 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3872out_dst_entries:
3873 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3874out_kmem_cache:
f2fc6a54 3875 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3876 goto out;
1da177e4
LT		 3877}
3878
3879void ip6_route_cleanup(void)
3880{
8ed67789 3881 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3882 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3883 fib6_rules_cleanup();
1da177e4 3884 xfrm6_fini();
1da177e4 3885 fib6_gc_cleanup();
c3426b47 3886 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3887 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3888 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3889 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3890}
Linux 6.x block layer and io_uring tree(s)