]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blame - net/ipv6/route.c
projects / mirror_ubuntu-zesty-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
tcp: replace dst_confirm with sk_dst_confirm
[mirror_ubuntu-zesty-kernel.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 66
7c0f6ba6 67#include <linux/uaccess.h>
1da177e4
LT		 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);
adc176c5 530 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL, 0);
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
19bda36c
XL		 1367 if (dst_metric_locked(dst, RTAX_MTU))
1368 return;
1369
45e4fd26
MKL		 1370 dst_confirm(dst);
1371 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1372 if (mtu >= dst_mtu(dst))
1373 return;
9d289715 1374
0d3f6d29 1375 if (!rt6_cache_allowed_for_pmtu(rt6)) {
45e4fd26
MKL		 1376 rt6_do_update_pmtu(rt6, mtu);
1377 } else {
1378 const struct in6_addr *daddr, *saddr;
1379 struct rt6_info *nrt6;
1380
1381 if (iph) {
1382 daddr = &iph->daddr;
1383 saddr = &iph->saddr;
1384 } else if (sk) {
1385 daddr = &sk->sk_v6_daddr;
1386 saddr = &inet6_sk(sk)->saddr;
1387 } else {
1388 return;
1389 }
1390 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1391 if (nrt6) {
1392 rt6_do_update_pmtu(nrt6, mtu);
1393
1394 /* ip6_ins_rt(nrt6) will bump the
1395 * rt6->rt6i_node->fn_sernum
1396 * which will fail the next rt6_check() and
1397 * invalidate the sk->sk_dst_cache.
1398 */
1399 ip6_ins_rt(nrt6);
1400 }
1da177e4
LT		 1401 }
1402}
1403
45e4fd26
MKL		 1404static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1405 struct sk_buff *skb, u32 mtu)
1406{
1407 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1408}
1409
42ae66c8 1410void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
e2d118a1 1411 int oif, u32 mark, kuid_t uid)
81aded24
DM		 1412{
1413 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1414 struct dst_entry *dst;
1415 struct flowi6 fl6;
1416
1417 memset(&fl6, 0, sizeof(fl6));
1418 fl6.flowi6_oif = oif;
1b3c61dc 1419 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1420 fl6.daddr = iph->daddr;
1421 fl6.saddr = iph->saddr;
6502ca52 1422 fl6.flowlabel = ip6_flowinfo(iph);
e2d118a1 1423 fl6.flowi6_uid = uid;
81aded24
DM		 1424
1425 dst = ip6_route_output(net, NULL, &fl6);
1426 if (!dst->error)
45e4fd26 1427 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
81aded24
DM		 1428 dst_release(dst);
1429}
1430EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1431
1432void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1433{
33c162a9
MKL		 1434 struct dst_entry *dst;
1435
81aded24 1436 ip6_update_pmtu(skb, sock_net(sk), mtu,
e2d118a1 1437 sk->sk_bound_dev_if, sk->sk_mark, sk->sk_uid);
33c162a9
MKL		 1438
1439 dst = __sk_dst_get(sk);
1440 if (!dst || !dst->obsolete ||
1441 dst->ops->check(dst, inet6_sk(sk)->dst_cookie))
1442 return;
1443
1444 bh_lock_sock(sk);
1445 if (!sock_owned_by_user(sk) && !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
1446 ip6_datagram_dst_update(sk, false);
1447 bh_unlock_sock(sk);
81aded24
DM		 1448}
1449EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1450
b55b76b2
DJ		 1451/* Handle redirects */
1452struct ip6rd_flowi {
1453 struct flowi6 fl6;
1454 struct in6_addr gateway;
1455};
1456
1457static struct rt6_info *__ip6_route_redirect(struct net *net,
1458 struct fib6_table *table,
1459 struct flowi6 *fl6,
1460 int flags)
1461{
1462 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1463 struct rt6_info *rt;
1464 struct fib6_node *fn;
1465
1466 /* Get the "current" route for this destination and
67c408cf 1467 * check if the redirect has come from appropriate router.
b55b76b2
DJ		 1468 *
1469 * RFC 4861 specifies that redirects should only be
1470 * accepted if they come from the nexthop to the target.
1471 * Due to the way the routes are chosen, this notion
1472 * is a bit fuzzy and one might need to check all possible
1473 * routes.
1474 */
1475
1476 read_lock_bh(&table->tb6_lock);
1477 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1478restart:
1479 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1480 if (rt6_check_expired(rt))
1481 continue;
1482 if (rt->dst.error)
1483 break;
1484 if (!(rt->rt6i_flags & RTF_GATEWAY))
1485 continue;
1486 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1487 continue;
1488 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1489 continue;
1490 break;
1491 }
1492
1493 if (!rt)
1494 rt = net->ipv6.ip6_null_entry;
1495 else if (rt->dst.error) {
1496 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1497 goto out;
1498 }
1499
1500 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1501 fn = fib6_backtrack(fn, &fl6->saddr);
1502 if (fn)
1503 goto restart;
b55b76b2 1504 }
a3c00e46 1505
b0a1ba59 1506out:
b55b76b2
DJ		 1507 dst_hold(&rt->dst);
1508
1509 read_unlock_bh(&table->tb6_lock);
1510
b811580d 1511 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
b55b76b2
DJ		 1512 return rt;
1513};
1514
1515static struct dst_entry *ip6_route_redirect(struct net *net,
1516 const struct flowi6 *fl6,
1517 const struct in6_addr *gateway)
1518{
1519 int flags = RT6_LOOKUP_F_HAS_SADDR;
1520 struct ip6rd_flowi rdfl;
1521
1522 rdfl.fl6 = *fl6;
1523 rdfl.gateway = *gateway;
1524
1525 return fib6_rule_lookup(net, &rdfl.fl6,
1526 flags, __ip6_route_redirect);
1527}
1528
e2d118a1
LC		 1529void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
1530 kuid_t uid)
3a5ad2ee
DM		 1531{
1532 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1533 struct dst_entry *dst;
1534 struct flowi6 fl6;
1535
1536 memset(&fl6, 0, sizeof(fl6));
e374c618 1537 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1538 fl6.flowi6_oif = oif;
1539 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1540 fl6.daddr = iph->daddr;
1541 fl6.saddr = iph->saddr;
6502ca52 1542 fl6.flowlabel = ip6_flowinfo(iph);
e2d118a1 1543 fl6.flowi6_uid = uid;
3a5ad2ee 1544
b55b76b2
DJ		 1545 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1546 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1547 dst_release(dst);
1548}
1549EXPORT_SYMBOL_GPL(ip6_redirect);
1550
c92a59ec
DJ		 1551void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1552 u32 mark)
1553{
1554 const struct ipv6hdr *iph = ipv6_hdr(skb);
1555 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1556 struct dst_entry *dst;
1557 struct flowi6 fl6;
1558
1559 memset(&fl6, 0, sizeof(fl6));
e374c618 1560 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1561 fl6.flowi6_oif = oif;
1562 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1563 fl6.daddr = msg->dest;
1564 fl6.saddr = iph->daddr;
e2d118a1 1565 fl6.flowi6_uid = sock_net_uid(net, NULL);
c92a59ec 1566
b55b76b2
DJ		 1567 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1568 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1569 dst_release(dst);
1570}
1571
3a5ad2ee
DM		 1572void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1573{
e2d118a1
LC		 1574 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark,
1575 sk->sk_uid);
3a5ad2ee
DM		 1576}
1577EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1578
0dbaee3b 1579static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1580{
0dbaee3b
DM		 1581 struct net_device *dev = dst->dev;
1582 unsigned int mtu = dst_mtu(dst);
1583 struct net *net = dev_net(dev);
1584
1da177e4
LT		 1585 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1586
5578689a
DL		 1587 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1588 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1589
1590 /*
1ab1457c
YH		 1591 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1592 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1593 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1594 * rely only on pmtu discovery"
1595 */
1596 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1597 mtu = IPV6_MAXPLEN;
1598 return mtu;
1599}
1600
ebb762f2 1601static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1602{
4b32b5ad
MKL		 1603 const struct rt6_info *rt = (const struct rt6_info *)dst;
1604 unsigned int mtu = rt->rt6i_pmtu;
d33e4553 1605 struct inet6_dev *idev;
618f9bc7 1606
4b32b5ad
MKL		 1607 if (mtu)
1608 goto out;
1609
1610 mtu = dst_metric_raw(dst, RTAX_MTU);
618f9bc7 1611 if (mtu)
30f78d8e 1612 goto out;
618f9bc7
SK		 1613
1614 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1615
1616 rcu_read_lock();
1617 idev = __in6_dev_get(dst->dev);
1618 if (idev)
1619 mtu = idev->cnf.mtu6;
1620 rcu_read_unlock();
1621
30f78d8e 1622out:
14972cbd
RP		 1623 mtu = min_t(unsigned int, mtu, IP6_MAX_MTU);
1624
1625 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
d33e4553
DM		 1626}
1627
3b00944c
YH		 1628static struct dst_entry *icmp6_dst_gc_list;
1629static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1630
3b00944c 1631struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1632 struct flowi6 *fl6)
1da177e4 1633{
87a11578 1634 struct dst_entry *dst;
1da177e4
LT		 1635 struct rt6_info *rt;
1636 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1637 struct net *net = dev_net(dev);
1da177e4 1638
38308473 1639 if (unlikely(!idev))
122bdf67 1640 return ERR_PTR(-ENODEV);
1da177e4 1641
ad706862 1642 rt = ip6_dst_alloc(net, dev, 0);
38308473 1643 if (unlikely(!rt)) {
1da177e4 1644 in6_dev_put(idev);
87a11578 1645 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1646 goto out;
1647 }
1648
8e2ec639
YZ		 1649 rt->dst.flags |= DST_HOST;
1650 rt->dst.output = ip6_output;
d8d1f30b 1651 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1652 rt->rt6i_gateway = fl6->daddr;
87a11578 1653 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1654 rt->rt6i_dst.plen = 128;
1655 rt->rt6i_idev = idev;
14edd87d 1656 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1657
3b00944c 1658 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1659 rt->dst.next = icmp6_dst_gc_list;
1660 icmp6_dst_gc_list = &rt->dst;
3b00944c 1661 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1662
5578689a 1663 fib6_force_start_gc(net);
1da177e4 1664
87a11578
DM		 1665 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1666
1da177e4 1667out:
87a11578 1668 return dst;
1da177e4
LT		 1669}
1670
3d0f24a7 1671int icmp6_dst_gc(void)
1da177e4 1672{
e9476e95 1673 struct dst_entry *dst, **pprev;
3d0f24a7 1674 int more = 0;
1da177e4 1675
3b00944c
YH		 1676 spin_lock_bh(&icmp6_dst_lock);
1677 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1678
1da177e4
LT		 1679 while ((dst = *pprev) != NULL) {
1680 if (!atomic_read(&dst->__refcnt)) {
1681 *pprev = dst->next;
1682 dst_free(dst);
1da177e4
LT		 1683 } else {
1684 pprev = &dst->next;
3d0f24a7 1685 ++more;
1da177e4
LT		 1686 }
1687 }
1688
3b00944c 1689 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1690
3d0f24a7 1691 return more;
1da177e4
LT		 1692}
1693
1e493d19
DM		 1694static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1695 void *arg)
1696{
1697 struct dst_entry *dst, **pprev;
1698
1699 spin_lock_bh(&icmp6_dst_lock);
1700 pprev = &icmp6_dst_gc_list;
1701 while ((dst = *pprev) != NULL) {
1702 struct rt6_info *rt = (struct rt6_info *) dst;
1703 if (func(rt, arg)) {
1704 *pprev = dst->next;
1705 dst_free(dst);
1706 } else {
1707 pprev = &dst->next;
1708 }
1709 }
1710 spin_unlock_bh(&icmp6_dst_lock);
1711}
1712
569d3645 1713static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1714{
86393e52 1715 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1716 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1717 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1718 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1719 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1720 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1721 int entries;
7019b78e 1722
fc66f95c 1723 entries = dst_entries_get_fast(ops);
49a18d86 1724 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1725 entries <= rt_max_size)
1da177e4
LT		 1726 goto out;
1727
6891a346 1728 net->ipv6.ip6_rt_gc_expire++;
14956643 1729 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1730 entries = dst_entries_get_slow(ops);
1731 if (entries < ops->gc_thresh)
7019b78e 1732 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1733out:
7019b78e 1734 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1735 return entries > rt_max_size;
1da177e4
LT		 1736}
1737
e715b6d3
FW		 1738static int ip6_convert_metrics(struct mx6_config *mxc,
1739 const struct fib6_config *cfg)
1740{
c3a8d947 1741 bool ecn_ca = false;
e715b6d3
FW		 1742 struct nlattr *nla;
1743 int remaining;
1744 u32 *mp;
1745
63159f29 1746 if (!cfg->fc_mx)
e715b6d3
FW		 1747 return 0;
1748
1749 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1750 if (unlikely(!mp))
1751 return -ENOMEM;
1752
1753 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1754 int type = nla_type(nla);
1bb14807 1755 u32 val;
e715b6d3 1756
1bb14807
DB		 1757 if (!type)
1758 continue;
1759 if (unlikely(type > RTAX_MAX))
1760 goto err;
ea697639 1761
1bb14807
DB		 1762 if (type == RTAX_CC_ALGO) {
1763 char tmp[TCP_CA_NAME_MAX];
e715b6d3 1764
1bb14807 1765 nla_strlcpy(tmp, nla, sizeof(tmp));
c3a8d947 1766 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
1bb14807
DB		 1767 if (val == TCP_CA_UNSPEC)
1768 goto err;
1769 } else {
1770 val = nla_get_u32(nla);
e715b6d3 1771 }
626abd59
PA		 1772 if (type == RTAX_HOPLIMIT && val > 255)
1773 val = 255;
b8d3e416
DB		 1774 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1775 goto err;
1bb14807
DB		 1776
1777 mp[type - 1] = val;
1778 __set_bit(type - 1, mxc->mx_valid);
e715b6d3
FW		 1779 }
1780
c3a8d947
DB		 1781 if (ecn_ca) {
1782 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1783 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1784 }
e715b6d3 1785
c3a8d947 1786 mxc->mx = mp;
e715b6d3
FW		 1787 return 0;
1788 err:
1789 kfree(mp);
1790 return -EINVAL;
1791}
1da177e4 1792
8c14586f
DA		 1793static struct rt6_info *ip6_nh_lookup_table(struct net *net,
1794 struct fib6_config *cfg,
1795 const struct in6_addr *gw_addr)
1796{
1797 struct flowi6 fl6 = {
1798 .flowi6_oif = cfg->fc_ifindex,
1799 .daddr = *gw_addr,
1800 .saddr = cfg->fc_prefsrc,
1801 };
1802 struct fib6_table *table;
1803 struct rt6_info *rt;
d5d32e4b 1804 int flags = RT6_LOOKUP_F_IFACE | RT6_LOOKUP_F_IGNORE_LINKSTATE;
8c14586f
DA		 1805
1806 table = fib6_get_table(net, cfg->fc_table);
1807 if (!table)
1808 return NULL;
1809
1810 if (!ipv6_addr_any(&cfg->fc_prefsrc))
1811 flags |= RT6_LOOKUP_F_HAS_SADDR;
1812
1813 rt = ip6_pol_route(net, table, cfg->fc_ifindex, &fl6, flags);
1814
1815 /* if table lookup failed, fall back to full lookup */
1816 if (rt == net->ipv6.ip6_null_entry) {
1817 ip6_rt_put(rt);
1818 rt = NULL;
1819 }
1820
1821 return rt;
1822}
1823
8c5b83f0 1824static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
1da177e4 1825{
5578689a 1826 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1827 struct rt6_info *rt = NULL;
1828 struct net_device *dev = NULL;
1829 struct inet6_dev *idev = NULL;
c71099ac 1830 struct fib6_table *table;
1da177e4 1831 int addr_type;
8c5b83f0 1832 int err = -EINVAL;
1da177e4 1833
ec462631
DA		 1834 /* RTF_PCPU is an internal flag; can not be set by userspace */
1835 if (cfg->fc_flags & RTF_PCPU)
1836 goto out;
1837
86872cb5 1838 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
8c5b83f0 1839 goto out;
1da177e4 1840#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1841 if (cfg->fc_src_len)
8c5b83f0 1842 goto out;
1da177e4 1843#endif
86872cb5 1844 if (cfg->fc_ifindex) {
1da177e4 1845 err = -ENODEV;
5578689a 1846 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1847 if (!dev)
1848 goto out;
1849 idev = in6_dev_get(dev);
1850 if (!idev)
1851 goto out;
1852 }
1853
86872cb5
TG		 1854 if (cfg->fc_metric == 0)
1855 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1856
d71314b4 1857 err = -ENOBUFS;
38308473
DM		 1858 if (cfg->fc_nlinfo.nlh &&
1859 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1860 table = fib6_get_table(net, cfg->fc_table);
38308473 1861 if (!table) {
f3213831 1862 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1863 table = fib6_new_table(net, cfg->fc_table);
1864 }
1865 } else {
1866 table = fib6_new_table(net, cfg->fc_table);
1867 }
38308473
DM		 1868
1869 if (!table)
c71099ac 1870 goto out;
c71099ac 1871
ad706862
MKL		 1872 rt = ip6_dst_alloc(net, NULL,
1873 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
1da177e4 1874
38308473 1875 if (!rt) {
1da177e4
LT		 1876 err = -ENOMEM;
1877 goto out;
1878 }
1879
1716a961
G		 1880 if (cfg->fc_flags & RTF_EXPIRES)
1881 rt6_set_expires(rt, jiffies +
1882 clock_t_to_jiffies(cfg->fc_expires));
1883 else
1884 rt6_clean_expires(rt);
1da177e4 1885
86872cb5
TG		 1886 if (cfg->fc_protocol == RTPROT_UNSPEC)
1887 cfg->fc_protocol = RTPROT_BOOT;
1888 rt->rt6i_protocol = cfg->fc_protocol;
1889
1890 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1891
1892 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1893 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1894 else if (cfg->fc_flags & RTF_LOCAL)
1895 rt->dst.input = ip6_input;
1da177e4 1896 else
d8d1f30b 1897 rt->dst.input = ip6_forward;
1da177e4 1898
d8d1f30b 1899 rt->dst.output = ip6_output;
1da177e4 1900
19e42e45
RP		 1901 if (cfg->fc_encap) {
1902 struct lwtunnel_state *lwtstate;
1903
1904 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
127eb7cd
TH		 1905 cfg->fc_encap, AF_INET6, cfg,
1906 &lwtstate);
19e42e45
RP		 1907 if (err)
1908 goto out;
61adedf3
JB		 1909 rt->dst.lwtstate = lwtstate_get(lwtstate);
1910 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1911 rt->dst.lwtstate->orig_output = rt->dst.output;
1912 rt->dst.output = lwtunnel_output;
25368623 1913 }
61adedf3
JB		 1914 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1915 rt->dst.lwtstate->orig_input = rt->dst.input;
1916 rt->dst.input = lwtunnel_input;
25368623 1917 }
19e42e45
RP		 1918 }
1919
86872cb5
TG		 1920 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1921 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1922 if (rt->rt6i_dst.plen == 128)
e5fd387a 1923 rt->dst.flags |= DST_HOST;
e5fd387a 1924
1da177e4 1925#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1926 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1927 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1928#endif
1929
86872cb5 1930 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1931
1932 /* We cannot add true routes via loopback here,
1933 they would result in kernel looping; promote them to reject routes
1934 */
86872cb5 1935 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1936 (dev && (dev->flags & IFF_LOOPBACK) &&
1937 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1938 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1939 /* hold loopback dev/idev if we haven't done so. */
5578689a 1940 if (dev != net->loopback_dev) {
1da177e4
LT		 1941 if (dev) {
1942 dev_put(dev);
1943 in6_dev_put(idev);
1944 }
5578689a 1945 dev = net->loopback_dev;
1da177e4
LT		 1946 dev_hold(dev);
1947 idev = in6_dev_get(dev);
1948 if (!idev) {
1949 err = -ENODEV;
1950 goto out;
1951 }
1952 }
1da177e4 1953 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1954 switch (cfg->fc_type) {
1955 case RTN_BLACKHOLE:
1956 rt->dst.error = -EINVAL;
ede2059d 1957 rt->dst.output = dst_discard_out;
7150aede 1958 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1959 break;
1960 case RTN_PROHIBIT:
1961 rt->dst.error = -EACCES;
7150aede
K		 1962 rt->dst.output = ip6_pkt_prohibit_out;
1963 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1964 break;
b4949ab2 1965 case RTN_THROW:
0315e382 1966 case RTN_UNREACHABLE:
ef2c7d7b 1967 default:
7150aede 1968 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
0315e382
NF		 1969 : (cfg->fc_type == RTN_UNREACHABLE)
1970 ? -EHOSTUNREACH : -ENETUNREACH;
7150aede
K		 1971 rt->dst.output = ip6_pkt_discard_out;
1972 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1973 break;
1974 }
1da177e4
LT		 1975 goto install_route;
1976 }
1977
86872cb5 1978 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1979 const struct in6_addr *gw_addr;
1da177e4
LT		 1980 int gwa_type;
1981
86872cb5 1982 gw_addr = &cfg->fc_gateway;
330567b7 1983 gwa_type = ipv6_addr_type(gw_addr);
48ed7b26
FW		 1984
1985 /* if gw_addr is local we will fail to detect this in case
1986 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1987 * will return already-added prefix route via interface that
1988 * prefix route was assigned to, which might be non-loopback.
1989 */
1990 err = -EINVAL;
330567b7
FW		 1991 if (ipv6_chk_addr_and_flags(net, gw_addr,
1992 gwa_type & IPV6_ADDR_LINKLOCAL ?
1993 dev : NULL, 0, 0))
48ed7b26
FW		 1994 goto out;
1995
4e3fd7a0 1996 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1997
1998 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
8c14586f 1999 struct rt6_info *grt = NULL;
1da177e4
LT		 2000
2001 /* IPv6 strictly inhibits using not link-local
2002 addresses as nexthop address.
2003 Otherwise, router will not able to send redirects.
2004 It is very good, but in some (rare!) circumstances
2005 (SIT, PtP, NBMA NOARP links) it is handy to allow
2006 some exceptions. --ANK
96d5822c
EN		 2007 We allow IPv4-mapped nexthops to support RFC4798-type
2008 addressing
1da177e4 2009 */
96d5822c
EN		 2010 if (!(gwa_type & (IPV6_ADDR_UNICAST |
2011 IPV6_ADDR_MAPPED)))
1da177e4
LT		 2012 goto out;
2013
a435a07f 2014 if (cfg->fc_table) {
8c14586f
DA		 2015 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
2016
a435a07f
VB		 2017 if (grt) {
2018 if (grt->rt6i_flags & RTF_GATEWAY ||
2019 (dev && dev != grt->dst.dev)) {
2020 ip6_rt_put(grt);
2021 grt = NULL;
2022 }
2023 }
2024 }
2025
8c14586f
DA		 2026 if (!grt)
2027 grt = rt6_lookup(net, gw_addr, NULL,
2028 cfg->fc_ifindex, 1);
1da177e4
LT		 2029
2030 err = -EHOSTUNREACH;
38308473 2031 if (!grt)
1da177e4
LT		 2032 goto out;
2033 if (dev) {
d1918542 2034 if (dev != grt->dst.dev) {
94e187c0 2035 ip6_rt_put(grt);
1da177e4
LT		 2036 goto out;
2037 }
2038 } else {
d1918542 2039 dev = grt->dst.dev;
1da177e4
LT		 2040 idev = grt->rt6i_idev;
2041 dev_hold(dev);
2042 in6_dev_hold(grt->rt6i_idev);
2043 }
38308473 2044 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 2045 err = 0;
94e187c0 2046 ip6_rt_put(grt);
1da177e4
LT		 2047
2048 if (err)
2049 goto out;
2050 }
2051 err = -EINVAL;
38308473 2052 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2053 goto out;
2054 }
2055
2056 err = -ENODEV;
38308473 2057 if (!dev)
1da177e4
LT		 2058 goto out;
2059
c3968a85
DW		 2060 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2061 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2062 err = -EINVAL;
2063 goto out;
2064 }
4e3fd7a0 2065 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 2066 rt->rt6i_prefsrc.plen = 128;
2067 } else
2068 rt->rt6i_prefsrc.plen = 0;
2069
86872cb5 2070 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 2071
2072install_route:
d8d1f30b 2073 rt->dst.dev = dev;
1da177e4 2074 rt->rt6i_idev = idev;
c71099ac 2075 rt->rt6i_table = table;
63152fc0 2076
c346dca1 2077 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 2078
8c5b83f0 2079 return rt;
6b9ea5a6
RP		 2080out:
2081 if (dev)
2082 dev_put(dev);
2083 if (idev)
2084 in6_dev_put(idev);
2085 if (rt)
2086 dst_free(&rt->dst);
2087
8c5b83f0 2088 return ERR_PTR(err);
6b9ea5a6
RP		 2089}
2090
2091int ip6_route_add(struct fib6_config *cfg)
2092{
2093 struct mx6_config mxc = { .mx = NULL, };
8c5b83f0 2094 struct rt6_info *rt;
6b9ea5a6
RP		 2095 int err;
2096
8c5b83f0
RP		 2097 rt = ip6_route_info_create(cfg);
2098 if (IS_ERR(rt)) {
2099 err = PTR_ERR(rt);
2100 rt = NULL;
6b9ea5a6 2101 goto out;
8c5b83f0 2102 }
6b9ea5a6 2103
e715b6d3
FW		 2104 err = ip6_convert_metrics(&mxc, cfg);
2105 if (err)
2106 goto out;
1da177e4 2107
e715b6d3
FW		 2108 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2109
2110 kfree(mxc.mx);
6b9ea5a6 2111
e715b6d3 2112 return err;
1da177e4 2113out:
1da177e4 2114 if (rt)
d8d1f30b 2115 dst_free(&rt->dst);
6b9ea5a6 2116
1da177e4
LT		 2117 return err;
2118}
2119
86872cb5 2120static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2121{
2122 int err;
c71099ac 2123 struct fib6_table *table;
d1918542 2124 struct net *net = dev_net(rt->dst.dev);
1da177e4 2125
8e3d5be7
MKL		 2126 if (rt == net->ipv6.ip6_null_entry ||
2127 rt->dst.flags & DST_NOCACHE) {
6825a26c
G		 2128 err = -ENOENT;
2129 goto out;
2130 }
6c813a72 2131
c71099ac
TG		 2132 table = rt->rt6i_table;
2133 write_lock_bh(&table->tb6_lock);
86872cb5 2134 err = fib6_del(rt, info);
c71099ac 2135 write_unlock_bh(&table->tb6_lock);
1da177e4 2136
6825a26c 2137out:
94e187c0 2138 ip6_rt_put(rt);
1da177e4
LT		 2139 return err;
2140}
2141
e0a1ad73
TG		 2142int ip6_del_rt(struct rt6_info *rt)
2143{
4d1169c1 2144 struct nl_info info = {
d1918542 2145 .nl_net = dev_net(rt->dst.dev),
4d1169c1 2146 };
528c4ceb 2147 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 2148}
2149
86872cb5 2150static int ip6_route_del(struct fib6_config *cfg)
1da177e4 2151{
c71099ac 2152 struct fib6_table *table;
1da177e4
LT		 2153 struct fib6_node *fn;
2154 struct rt6_info *rt;
2155 int err = -ESRCH;
2156
5578689a 2157 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 2158 if (!table)
c71099ac
TG		 2159 return err;
2160
2161 read_lock_bh(&table->tb6_lock);
1da177e4 2162
c71099ac 2163 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 2164 &cfg->fc_dst, cfg->fc_dst_len,
2165 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 2166
1da177e4 2167 if (fn) {
d8d1f30b 2168 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 2169 if ((rt->rt6i_flags & RTF_CACHE) &&
2170 !(cfg->fc_flags & RTF_CACHE))
2171 continue;
86872cb5 2172 if (cfg->fc_ifindex &&
d1918542
DM		 2173 (!rt->dst.dev ||
2174 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 2175 continue;
86872cb5
TG		 2176 if (cfg->fc_flags & RTF_GATEWAY &&
2177 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 2178 continue;
86872cb5 2179 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 2180 continue;
c2ed1880
M		 2181 if (cfg->fc_protocol && cfg->fc_protocol != rt->rt6i_protocol)
2182 continue;
d8d1f30b 2183 dst_hold(&rt->dst);
c71099ac 2184 read_unlock_bh(&table->tb6_lock);
1da177e4 2185
86872cb5 2186 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 2187 }
2188 }
c71099ac 2189 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2190
2191 return err;
2192}
2193
6700c270 2194static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 2195{
a6279458 2196 struct netevent_redirect netevent;
e8599ff4 2197 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 2198 struct ndisc_options ndopts;
2199 struct inet6_dev *in6_dev;
2200 struct neighbour *neigh;
71bcdba0 2201 struct rd_msg *msg;
6e157b6a
DM		 2202 int optlen, on_link;
2203 u8 *lladdr;
e8599ff4 2204
29a3cad5 2205 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 2206 optlen -= sizeof(*msg);
e8599ff4
DM		 2207
2208 if (optlen < 0) {
6e157b6a 2209 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 2210 return;
2211 }
2212
71bcdba0 2213 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 2214
71bcdba0 2215 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 2216 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 2217 return;
2218 }
2219
6e157b6a 2220 on_link = 0;
71bcdba0 2221 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 2222 on_link = 1;
71bcdba0 2223 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 2224 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 2225 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 2226 return;
2227 }
2228
2229 in6_dev = __in6_dev_get(skb->dev);
2230 if (!in6_dev)
2231 return;
2232 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2233 return;
2234
2235 /* RFC2461 8.1:
2236 * The IP source address of the Redirect MUST be the same as the current
2237 * first-hop router for the specified ICMP Destination Address.
2238 */
2239
f997c55c 2240 if (!ndisc_parse_options(skb->dev, msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 2241 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2242 return;
2243 }
6e157b6a
DM		 2244
2245 lladdr = NULL;
e8599ff4
DM		 2246 if (ndopts.nd_opts_tgt_lladdr) {
2247 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2248 skb->dev);
2249 if (!lladdr) {
2250 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2251 return;
2252 }
2253 }
2254
6e157b6a 2255 rt = (struct rt6_info *) dst;
ec13ad1d 2256 if (rt->rt6i_flags & RTF_REJECT) {
6e157b6a 2257 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 2258 return;
6e157b6a 2259 }
e8599ff4 2260
6e157b6a
DM		 2261 /* Redirect received -> path was valid.
2262 * Look, redirects are sent only in response to data packets,
2263 * so that this nexthop apparently is reachable. --ANK
2264 */
2265 dst_confirm(&rt->dst);
a6279458 2266
71bcdba0 2267 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 2268 if (!neigh)
2269 return;
a6279458 2270
1da177e4
LT		 2271 /*
2272 * We have finally decided to accept it.
2273 */
2274
f997c55c 2275 ndisc_update(skb->dev, neigh, lladdr, NUD_STALE,
1da177e4
LT		 2276 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2277 NEIGH_UPDATE_F_OVERRIDE|
2278 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
f997c55c
AA		 2279 NEIGH_UPDATE_F_ISROUTER)),
2280 NDISC_REDIRECT, &ndopts);
1da177e4 2281
83a09abd 2282 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
38308473 2283 if (!nrt)
1da177e4
LT		 2284 goto out;
2285
2286 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2287 if (on_link)
2288 nrt->rt6i_flags &= ~RTF_GATEWAY;
2289
4e3fd7a0 2290 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 2291
40e22e8f 2292 if (ip6_ins_rt(nrt))
1da177e4
LT		 2293 goto out;
2294
d8d1f30b
CG		 2295 netevent.old = &rt->dst;
2296 netevent.new = &nrt->dst;
71bcdba0 2297 netevent.daddr = &msg->dest;
60592833 2298 netevent.neigh = neigh;
8d71740c
TT		 2299 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2300
38308473 2301 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 2302 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 2303 ip6_del_rt(rt);
1da177e4
LT		 2304 }
2305
2306out:
e8599ff4 2307 neigh_release(neigh);
6e157b6a
DM		 2308}
2309
1da177e4
LT		 2310/*
2311 * Misc support functions
2312 */
2313
4b32b5ad
MKL		 2314static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2315{
2316 BUG_ON(from->dst.from);
2317
2318 rt->rt6i_flags &= ~RTF_EXPIRES;
2319 dst_hold(&from->dst);
2320 rt->dst.from = &from->dst;
2321 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2322}
2323
83a09abd
MKL		 2324static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
2325{
2326 rt->dst.input = ort->dst.input;
2327 rt->dst.output = ort->dst.output;
2328 rt->rt6i_dst = ort->rt6i_dst;
2329 rt->dst.error = ort->dst.error;
2330 rt->rt6i_idev = ort->rt6i_idev;
2331 if (rt->rt6i_idev)
2332 in6_dev_hold(rt->rt6i_idev);
2333 rt->dst.lastuse = jiffies;
2334 rt->rt6i_gateway = ort->rt6i_gateway;
2335 rt->rt6i_flags = ort->rt6i_flags;
2336 rt6_set_from(rt, ort);
2337 rt->rt6i_metric = ort->rt6i_metric;
1da177e4 2338#ifdef CONFIG_IPV6_SUBTREES
83a09abd 2339 rt->rt6i_src = ort->rt6i_src;
1da177e4 2340#endif
83a09abd
MKL		 2341 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2342 rt->rt6i_table = ort->rt6i_table;
61adedf3 2343 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
1da177e4
LT		 2344}
2345
70ceb4f5 2346#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 2347static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42 2348 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2349 const struct in6_addr *gwaddr,
2350 struct net_device *dev)
70ceb4f5 2351{
830218c1
DA		 2352 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO;
2353 int ifindex = dev->ifindex;
70ceb4f5
YH		 2354 struct fib6_node *fn;
2355 struct rt6_info *rt = NULL;
c71099ac
TG		 2356 struct fib6_table *table;
2357
830218c1 2358 table = fib6_get_table(net, tb_id);
38308473 2359 if (!table)
c71099ac 2360 return NULL;
70ceb4f5 2361
5744dd9b 2362 read_lock_bh(&table->tb6_lock);
67ba4152 2363 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2364 if (!fn)
2365 goto out;
2366
d8d1f30b 2367 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2368 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2369 continue;
2370 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2371 continue;
2372 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2373 continue;
d8d1f30b 2374 dst_hold(&rt->dst);
70ceb4f5
YH		 2375 break;
2376 }
2377out:
5744dd9b 2378 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2379 return rt;
2380}
2381
efa2cea0 2382static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42 2383 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2384 const struct in6_addr *gwaddr,
2385 struct net_device *dev,
95c96174 2386 unsigned int pref)
70ceb4f5 2387{
86872cb5 2388 struct fib6_config cfg = {
238fc7ea 2389 .fc_metric = IP6_RT_PRIO_USER,
830218c1 2390 .fc_ifindex = dev->ifindex,
86872cb5
TG		 2391 .fc_dst_len = prefixlen,
2392 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2393 RTF_UP | RTF_PREF(pref),
15e47304 2394 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2395 .fc_nlinfo.nlh = NULL,
2396 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2397 };
2398
830218c1 2399 cfg.fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO,
4e3fd7a0
AD		 2400 cfg.fc_dst = *prefix;
2401 cfg.fc_gateway = *gwaddr;
70ceb4f5 2402
e317da96
YH		 2403 /* We should treat it as a default route if prefix length is 0. */
2404 if (!prefixlen)
86872cb5 2405 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2406
86872cb5 2407 ip6_route_add(&cfg);
70ceb4f5 2408
830218c1 2409 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, dev);
70ceb4f5
YH		 2410}
2411#endif
2412
b71d1d42 2413struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2414{
830218c1 2415 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT;
1da177e4 2416 struct rt6_info *rt;
c71099ac 2417 struct fib6_table *table;
1da177e4 2418
830218c1 2419 table = fib6_get_table(dev_net(dev), tb_id);
38308473 2420 if (!table)
c71099ac 2421 return NULL;
1da177e4 2422
5744dd9b 2423 read_lock_bh(&table->tb6_lock);
67ba4152 2424 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2425 if (dev == rt->dst.dev &&
045927ff 2426 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2427 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2428 break;
2429 }
2430 if (rt)
d8d1f30b 2431 dst_hold(&rt->dst);
5744dd9b 2432 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2433 return rt;
2434}
2435
b71d1d42 2436struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2437 struct net_device *dev,
2438 unsigned int pref)
1da177e4 2439{
86872cb5 2440 struct fib6_config cfg = {
ca254490 2441 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
238fc7ea 2442 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2443 .fc_ifindex = dev->ifindex,
2444 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2445 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2446 .fc_nlinfo.portid = 0,
5578689a 2447 .fc_nlinfo.nlh = NULL,
c346dca1 2448 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2449 };
1da177e4 2450
4e3fd7a0 2451 cfg.fc_gateway = *gwaddr;
1da177e4 2452
830218c1
DA		 2453 if (!ip6_route_add(&cfg)) {
2454 struct fib6_table *table;
2455
2456 table = fib6_get_table(dev_net(dev), cfg.fc_table);
2457 if (table)
2458 table->flags |= RT6_TABLE_HAS_DFLT_ROUTER;
2459 }
1da177e4 2460
1da177e4
LT		 2461 return rt6_get_dflt_router(gwaddr, dev);
2462}
2463
830218c1 2464static void __rt6_purge_dflt_routers(struct fib6_table *table)
1da177e4
LT		 2465{
2466 struct rt6_info *rt;
2467
2468restart:
c71099ac 2469 read_lock_bh(&table->tb6_lock);
d8d1f30b 2470 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2471 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2472 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2473 dst_hold(&rt->dst);
c71099ac 2474 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2475 ip6_del_rt(rt);
1da177e4
LT		 2476 goto restart;
2477 }
2478 }
c71099ac 2479 read_unlock_bh(&table->tb6_lock);
830218c1
DA		 2480
2481 table->flags &= ~RT6_TABLE_HAS_DFLT_ROUTER;
2482}
2483
2484void rt6_purge_dflt_routers(struct net *net)
2485{
2486 struct fib6_table *table;
2487 struct hlist_head *head;
2488 unsigned int h;
2489
2490 rcu_read_lock();
2491
2492 for (h = 0; h < FIB6_TABLE_HASHSZ; h++) {
2493 head = &net->ipv6.fib_table_hash[h];
2494 hlist_for_each_entry_rcu(table, head, tb6_hlist) {
2495 if (table->flags & RT6_TABLE_HAS_DFLT_ROUTER)
2496 __rt6_purge_dflt_routers(table);
2497 }
2498 }
2499
2500 rcu_read_unlock();
1da177e4
LT		 2501}
2502
5578689a
DL		 2503static void rtmsg_to_fib6_config(struct net *net,
2504 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2505 struct fib6_config *cfg)
2506{
2507 memset(cfg, 0, sizeof(*cfg));
2508
ca254490
DA		 2509 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2510 : RT6_TABLE_MAIN;
86872cb5
TG		 2511 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2512 cfg->fc_metric = rtmsg->rtmsg_metric;
2513 cfg->fc_expires = rtmsg->rtmsg_info;
2514 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2515 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2516 cfg->fc_flags = rtmsg->rtmsg_flags;
2517
5578689a 2518 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2519
4e3fd7a0
AD		 2520 cfg->fc_dst = rtmsg->rtmsg_dst;
2521 cfg->fc_src = rtmsg->rtmsg_src;
2522 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2523}
2524
5578689a 2525int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2526{
86872cb5 2527 struct fib6_config cfg;
1da177e4
LT		 2528 struct in6_rtmsg rtmsg;
2529 int err;
2530
67ba4152 2531 switch (cmd) {
1da177e4
LT		 2532 case SIOCADDRT: /* Add a route */
2533 case SIOCDELRT: /* Delete a route */
af31f412 2534 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2535 return -EPERM;
2536 err = copy_from_user(&rtmsg, arg,
2537 sizeof(struct in6_rtmsg));
2538 if (err)
2539 return -EFAULT;
86872cb5 2540
5578689a 2541 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2542
1da177e4
LT		 2543 rtnl_lock();
2544 switch (cmd) {
2545 case SIOCADDRT:
86872cb5 2546 err = ip6_route_add(&cfg);
1da177e4
LT		 2547 break;
2548 case SIOCDELRT:
86872cb5 2549 err = ip6_route_del(&cfg);
1da177e4
LT		 2550 break;
2551 default:
2552 err = -EINVAL;
2553 }
2554 rtnl_unlock();
2555
2556 return err;
3ff50b79 2557 }
1da177e4
LT		 2558
2559 return -EINVAL;
2560}
2561
2562/*
2563 * Drop the packet on the floor
2564 */
2565
d5fdd6ba 2566static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2567{
612f09e8 2568 int type;
adf30907 2569 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2570 switch (ipstats_mib_noroutes) {
2571 case IPSTATS_MIB_INNOROUTES:
0660e03f 2572 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2573 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2574 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2575 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2576 break;
2577 }
2578 /* FALLTHROUGH */
2579 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2580 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2581 ipstats_mib_noroutes);
612f09e8
YH		 2582 break;
2583 }
3ffe533c 2584 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2585 kfree_skb(skb);
2586 return 0;
2587}
2588
9ce8ade0
TG		 2589static int ip6_pkt_discard(struct sk_buff *skb)
2590{
612f09e8 2591 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2592}
2593
ede2059d 2594static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
1da177e4 2595{
adf30907 2596 skb->dev = skb_dst(skb)->dev;
612f09e8 2597 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2598}
2599
9ce8ade0
TG		 2600static int ip6_pkt_prohibit(struct sk_buff *skb)
2601{
612f09e8 2602 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2603}
2604
ede2059d 2605static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
9ce8ade0 2606{
adf30907 2607 skb->dev = skb_dst(skb)->dev;
612f09e8 2608 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2609}
2610
1da177e4
LT		 2611/*
2612 * Allocate a dst for local (unicast / anycast) address.
2613 */
2614
2615struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2616 const struct in6_addr *addr,
8f031519 2617 bool anycast)
1da177e4 2618{
ca254490 2619 u32 tb_id;
c346dca1 2620 struct net *net = dev_net(idev->dev);
5f02ce24
DA		 2621 struct net_device *dev = net->loopback_dev;
2622 struct rt6_info *rt;
2623
2624 /* use L3 Master device as loopback for host routes if device
2625 * is enslaved and address is not link local or multicast
2626 */
2627 if (!rt6_need_strict(addr))
2628 dev = l3mdev_master_dev_rcu(idev->dev) ? : dev;
2629
2630 rt = ip6_dst_alloc(net, dev, DST_NOCOUNT);
a3300ef4 2631 if (!rt)
1da177e4
LT		 2632 return ERR_PTR(-ENOMEM);
2633
1da177e4
LT		 2634 in6_dev_hold(idev);
2635
11d53b49 2636 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2637 rt->dst.input = ip6_input;
2638 rt->dst.output = ip6_output;
1da177e4 2639 rt->rt6i_idev = idev;
1da177e4
LT		 2640
2641 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2642 if (anycast)
2643 rt->rt6i_flags |= RTF_ANYCAST;
2644 else
1da177e4 2645 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2646
550bab42 2647 rt->rt6i_gateway = *addr;
4e3fd7a0 2648 rt->rt6i_dst.addr = *addr;
1da177e4 2649 rt->rt6i_dst.plen = 128;
ca254490
DA		 2650 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2651 rt->rt6i_table = fib6_get_table(net, tb_id);
8e3d5be7 2652 rt->dst.flags |= DST_NOCACHE;
1da177e4 2653
d8d1f30b 2654 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2655
2656 return rt;
2657}
2658
c3968a85
DW		 2659/* remove deleted ip from prefsrc entries */
2660struct arg_dev_net_ip {
2661 struct net_device *dev;
2662 struct net *net;
2663 struct in6_addr *addr;
2664};
2665
2666static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2667{
2668 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2669 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2670 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2671
d1918542 2672 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2673 rt != net->ipv6.ip6_null_entry &&
2674 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2675 /* remove prefsrc entry */
2676 rt->rt6i_prefsrc.plen = 0;
2677 }
2678 return 0;
2679}
2680
2681void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2682{
2683 struct net *net = dev_net(ifp->idev->dev);
2684 struct arg_dev_net_ip adni = {
2685 .dev = ifp->idev->dev,
2686 .net = net,
2687 .addr = &ifp->addr,
2688 };
0c3584d5 2689 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2690}
2691
be7a010d
DJ		 2692#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2693#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2694
2695/* Remove routers and update dst entries when gateway turn into host. */
2696static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2697{
2698 struct in6_addr *gateway = (struct in6_addr *)arg;
2699
2700 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2701 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2702 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2703 return -1;
2704 }
2705 return 0;
2706}
2707
2708void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2709{
2710 fib6_clean_all(net, fib6_clean_tohost, gateway);
2711}
2712
8ed67789
DL		 2713struct arg_dev_net {
2714 struct net_device *dev;
2715 struct net *net;
2716};
2717
1da177e4
LT		 2718static int fib6_ifdown(struct rt6_info *rt, void *arg)
2719{
bc3ef660 2720 const struct arg_dev_net *adn = arg;
2721 const struct net_device *dev = adn->dev;
8ed67789 2722
d1918542 2723 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2724 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2725 return -1;
c159d30c 2726
1da177e4
LT		 2727 return 0;
2728}
2729
f3db4851 2730void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2731{
8ed67789
DL		 2732 struct arg_dev_net adn = {
2733 .dev = dev,
2734 .net = net,
2735 };
2736
0c3584d5 2737 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2738 icmp6_clean_all(fib6_ifdown, &adn);
e332bc67
EB		 2739 if (dev)
2740 rt6_uncached_list_flush_dev(net, dev);
1da177e4
LT		 2741}
2742
95c96174 2743struct rt6_mtu_change_arg {
1da177e4 2744 struct net_device *dev;
95c96174 2745 unsigned int mtu;
1da177e4
LT		 2746};
2747
2748static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2749{
2750 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2751 struct inet6_dev *idev;
2752
2753 /* In IPv6 pmtu discovery is not optional,
2754 so that RTAX_MTU lock cannot disable it.
2755 We still use this lock to block changes
2756 caused by addrconf/ndisc.
2757 */
2758
2759 idev = __in6_dev_get(arg->dev);
38308473 2760 if (!idev)
1da177e4
LT		 2761 return 0;
2762
2763 /* For administrative MTU increase, there is no way to discover
2764 IPv6 PMTU increase, so PMTU increase should be updated here.
2765 Since RFC 1981 doesn't include administrative MTU increase
2766 update PMTU increase is a MUST. (i.e. jumbo frame)
2767 */
2768 /*
2769 If new MTU is less than route PMTU, this new MTU will be the
2770 lowest MTU in the path, update the route PMTU to reflect PMTU
2771 decreases; if new MTU is greater than route PMTU, and the
2772 old MTU is the lowest MTU in the path, update the route PMTU
2773 to reflect the increase. In this case if the other nodes' MTU
2774 also have the lowest MTU, TOO BIG MESSAGE will be lead to
67c408cf 2775 PMTU discovery.
1da177e4 2776 */
d1918542 2777 if (rt->dst.dev == arg->dev &&
fb56be83 2778 dst_metric_raw(&rt->dst, RTAX_MTU) &&
4b32b5ad
MKL		 2779 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2780 if (rt->rt6i_flags & RTF_CACHE) {
2781 /* For RTF_CACHE with rt6i_pmtu == 0
2782 * (i.e. a redirected route),
2783 * the metrics of its rt->dst.from has already
2784 * been updated.
2785 */
2786 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2787 rt->rt6i_pmtu = arg->mtu;
2788 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2789 (dst_mtu(&rt->dst) < arg->mtu &&
2790 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2791 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2792 }
566cfd8f 2793 }
1da177e4
LT		 2794 return 0;
2795}
2796
95c96174 2797void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2798{
c71099ac
TG		 2799 struct rt6_mtu_change_arg arg = {
2800 .dev = dev,
2801 .mtu = mtu,
2802 };
1da177e4 2803
0c3584d5 2804 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2805}
2806
ef7c79ed 2807static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2808 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2809 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2810 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2811 [RTA_PRIORITY] = { .type = NLA_U32 },
2812 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2813 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2814 [RTA_PREF] = { .type = NLA_U8 },
19e42e45
RP		 2815 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2816 [RTA_ENCAP] = { .type = NLA_NESTED },
32bc201e 2817 [RTA_EXPIRES] = { .type = NLA_U32 },
622ec2c9 2818 [RTA_UID] = { .type = NLA_U32 },
86872cb5
TG		 2819};
2820
2821static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2822 struct fib6_config *cfg)
1da177e4 2823{
86872cb5
TG		 2824 struct rtmsg *rtm;
2825 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2826 unsigned int pref;
86872cb5 2827 int err;
1da177e4 2828
86872cb5
TG		 2829 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2830 if (err < 0)
2831 goto errout;
1da177e4 2832
86872cb5
TG		 2833 err = -EINVAL;
2834 rtm = nlmsg_data(nlh);
2835 memset(cfg, 0, sizeof(*cfg));
2836
2837 cfg->fc_table = rtm->rtm_table;
2838 cfg->fc_dst_len = rtm->rtm_dst_len;
2839 cfg->fc_src_len = rtm->rtm_src_len;
2840 cfg->fc_flags = RTF_UP;
2841 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2842 cfg->fc_type = rtm->rtm_type;
86872cb5 2843
ef2c7d7b
ND		 2844 if (rtm->rtm_type == RTN_UNREACHABLE ||
2845 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2846 rtm->rtm_type == RTN_PROHIBIT ||
2847 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2848 cfg->fc_flags |= RTF_REJECT;
2849
ab79ad14
MŻ		 2850 if (rtm->rtm_type == RTN_LOCAL)
2851 cfg->fc_flags |= RTF_LOCAL;
2852
1f56a01f
MKL		 2853 if (rtm->rtm_flags & RTM_F_CLONED)
2854 cfg->fc_flags |= RTF_CACHE;
2855
15e47304 2856 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2857 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2858 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2859
2860 if (tb[RTA_GATEWAY]) {
67b61f6c 2861 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2862 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2863 }
86872cb5
TG		 2864
2865 if (tb[RTA_DST]) {
2866 int plen = (rtm->rtm_dst_len + 7) >> 3;
2867
2868 if (nla_len(tb[RTA_DST]) < plen)
2869 goto errout;
2870
2871 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2872 }
86872cb5
TG		 2873
2874 if (tb[RTA_SRC]) {
2875 int plen = (rtm->rtm_src_len + 7) >> 3;
2876
2877 if (nla_len(tb[RTA_SRC]) < plen)
2878 goto errout;
2879
2880 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2881 }
86872cb5 2882
c3968a85 2883 if (tb[RTA_PREFSRC])
67b61f6c 2884 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2885
86872cb5
TG		 2886 if (tb[RTA_OIF])
2887 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2888
2889 if (tb[RTA_PRIORITY])
2890 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2891
2892 if (tb[RTA_METRICS]) {
2893 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2894 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2895 }
86872cb5
TG		 2896
2897 if (tb[RTA_TABLE])
2898 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2899
51ebd318
ND		 2900 if (tb[RTA_MULTIPATH]) {
2901 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2902 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
9ed59592
DA		 2903
2904 err = lwtunnel_valid_encap_type_attr(cfg->fc_mp,
2905 cfg->fc_mp_len);
2906 if (err < 0)
2907 goto errout;
51ebd318
ND		 2908 }
2909
c78ba6d6
LR		 2910 if (tb[RTA_PREF]) {
2911 pref = nla_get_u8(tb[RTA_PREF]);
2912 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2913 pref != ICMPV6_ROUTER_PREF_HIGH)
2914 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2915 cfg->fc_flags |= RTF_PREF(pref);
2916 }
2917
19e42e45
RP		 2918 if (tb[RTA_ENCAP])
2919 cfg->fc_encap = tb[RTA_ENCAP];
2920
9ed59592 2921 if (tb[RTA_ENCAP_TYPE]) {
19e42e45
RP		 2922 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2923
9ed59592
DA		 2924 err = lwtunnel_valid_encap_type(cfg->fc_encap_type);
2925 if (err < 0)
2926 goto errout;
2927 }
2928
32bc201e
XL		 2929 if (tb[RTA_EXPIRES]) {
2930 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2931
2932 if (addrconf_finite_timeout(timeout)) {
2933 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2934 cfg->fc_flags |= RTF_EXPIRES;
2935 }
2936 }
2937
86872cb5
TG		 2938 err = 0;
2939errout:
2940 return err;
1da177e4
LT		 2941}
2942
6b9ea5a6
RP		 2943struct rt6_nh {
2944 struct rt6_info *rt6_info;
2945 struct fib6_config r_cfg;
2946 struct mx6_config mxc;
2947 struct list_head next;
2948};
2949
2950static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2951{
2952 struct rt6_nh *nh;
2953
2954 list_for_each_entry(nh, rt6_nh_list, next) {
2955 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2956 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2957 nh->r_cfg.fc_ifindex);
2958 }
2959}
2960
2961static int ip6_route_info_append(struct list_head *rt6_nh_list,
2962 struct rt6_info *rt, struct fib6_config *r_cfg)
2963{
2964 struct rt6_nh *nh;
2965 struct rt6_info *rtnh;
2966 int err = -EEXIST;
2967
2968 list_for_each_entry(nh, rt6_nh_list, next) {
2969 /* check if rt6_info already exists */
2970 rtnh = nh->rt6_info;
2971
2972 if (rtnh->dst.dev == rt->dst.dev &&
2973 rtnh->rt6i_idev == rt->rt6i_idev &&
2974 ipv6_addr_equal(&rtnh->rt6i_gateway,
2975 &rt->rt6i_gateway))
2976 return err;
2977 }
2978
2979 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2980 if (!nh)
2981 return -ENOMEM;
2982 nh->rt6_info = rt;
2983 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2984 if (err) {
2985 kfree(nh);
2986 return err;
2987 }
2988 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2989 list_add_tail(&nh->next, rt6_nh_list);
2990
2991 return 0;
2992}
2993
2994static int ip6_route_multipath_add(struct fib6_config *cfg)
51ebd318
ND		 2995{
2996 struct fib6_config r_cfg;
2997 struct rtnexthop *rtnh;
6b9ea5a6
RP		 2998 struct rt6_info *rt;
2999 struct rt6_nh *err_nh;
3000 struct rt6_nh *nh, *nh_safe;
51ebd318
ND		 3001 int remaining;
3002 int attrlen;
6b9ea5a6
RP		 3003 int err = 1;
3004 int nhn = 0;
3005 int replace = (cfg->fc_nlinfo.nlh &&
3006 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
3007 LIST_HEAD(rt6_nh_list);
51ebd318 3008
35f1b4e9 3009 remaining = cfg->fc_mp_len;
51ebd318 3010 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318 3011
6b9ea5a6
RP		 3012 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
3013 * rt6_info structs per nexthop
3014 */
51ebd318
ND		 3015 while (rtnh_ok(rtnh, remaining)) {
3016 memcpy(&r_cfg, cfg, sizeof(*cfg));
3017 if (rtnh->rtnh_ifindex)
3018 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3019
3020 attrlen = rtnh_attrlen(rtnh);
3021 if (attrlen > 0) {
3022 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3023
3024 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3025 if (nla) {
67b61f6c 3026 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 3027 r_cfg.fc_flags |= RTF_GATEWAY;
3028 }
19e42e45
RP		 3029 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
3030 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
3031 if (nla)
3032 r_cfg.fc_encap_type = nla_get_u16(nla);
51ebd318 3033 }
6b9ea5a6 3034
8c5b83f0
RP		 3035 rt = ip6_route_info_create(&r_cfg);
3036 if (IS_ERR(rt)) {
3037 err = PTR_ERR(rt);
3038 rt = NULL;
6b9ea5a6 3039 goto cleanup;
8c5b83f0 3040 }
6b9ea5a6
RP		 3041
3042 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
51ebd318 3043 if (err) {
6b9ea5a6
RP		 3044 dst_free(&rt->dst);
3045 goto cleanup;
3046 }
3047
3048 rtnh = rtnh_next(rtnh, &remaining);
3049 }
3050
3051 err_nh = NULL;
3052 list_for_each_entry(nh, &rt6_nh_list, next) {
3053 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
3054 /* nh->rt6_info is used or freed at this point, reset to NULL*/
3055 nh->rt6_info = NULL;
3056 if (err) {
3057 if (replace && nhn)
3058 ip6_print_replace_route_err(&rt6_nh_list);
3059 err_nh = nh;
3060 goto add_errout;
51ebd318 3061 }
6b9ea5a6 3062
1a72418b 3063 /* Because each route is added like a single route we remove
27596472
MK		 3064 * these flags after the first nexthop: if there is a collision,
3065 * we have already failed to add the first nexthop:
3066 * fib6_add_rt2node() has rejected it; when replacing, old
3067 * nexthops have been replaced by first new, the rest should
3068 * be added to it.
1a72418b 3069 */
27596472
MK		 3070 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
3071 NLM_F_REPLACE);
6b9ea5a6
RP		 3072 nhn++;
3073 }
3074
3075 goto cleanup;
3076
3077add_errout:
3078 /* Delete routes that were already added */
3079 list_for_each_entry(nh, &rt6_nh_list, next) {
3080 if (err_nh == nh)
3081 break;
3082 ip6_route_del(&nh->r_cfg);
3083 }
3084
3085cleanup:
3086 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3087 if (nh->rt6_info)
3088 dst_free(&nh->rt6_info->dst);
52fe51f8 3089 kfree(nh->mxc.mx);
6b9ea5a6
RP		 3090 list_del(&nh->next);
3091 kfree(nh);
3092 }
3093
3094 return err;
3095}
3096
3097static int ip6_route_multipath_del(struct fib6_config *cfg)
3098{
3099 struct fib6_config r_cfg;
3100 struct rtnexthop *rtnh;
3101 int remaining;
3102 int attrlen;
3103 int err = 1, last_err = 0;
3104
3105 remaining = cfg->fc_mp_len;
3106 rtnh = (struct rtnexthop *)cfg->fc_mp;
3107
3108 /* Parse a Multipath Entry */
3109 while (rtnh_ok(rtnh, remaining)) {
3110 memcpy(&r_cfg, cfg, sizeof(*cfg));
3111 if (rtnh->rtnh_ifindex)
3112 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3113
3114 attrlen = rtnh_attrlen(rtnh);
3115 if (attrlen > 0) {
3116 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3117
3118 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3119 if (nla) {
3120 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3121 r_cfg.fc_flags |= RTF_GATEWAY;
3122 }
3123 }
3124 err = ip6_route_del(&r_cfg);
3125 if (err)
3126 last_err = err;
3127
51ebd318
ND		 3128 rtnh = rtnh_next(rtnh, &remaining);
3129 }
3130
3131 return last_err;
3132}
3133
67ba4152 3134static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3135{
86872cb5
TG		 3136 struct fib6_config cfg;
3137 int err;
1da177e4 3138
86872cb5
TG		 3139 err = rtm_to_fib6_config(skb, nlh, &cfg);
3140 if (err < 0)
3141 return err;
3142
51ebd318 3143 if (cfg.fc_mp)
6b9ea5a6 3144 return ip6_route_multipath_del(&cfg);
51ebd318
ND		 3145 else
3146 return ip6_route_del(&cfg);
1da177e4
LT		 3147}
3148
67ba4152 3149static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3150{
86872cb5
TG		 3151 struct fib6_config cfg;
3152 int err;
1da177e4 3153
86872cb5
TG		 3154 err = rtm_to_fib6_config(skb, nlh, &cfg);
3155 if (err < 0)
3156 return err;
3157
51ebd318 3158 if (cfg.fc_mp)
6b9ea5a6 3159 return ip6_route_multipath_add(&cfg);
51ebd318
ND		 3160 else
3161 return ip6_route_add(&cfg);
1da177e4
LT		 3162}
3163
19e42e45 3164static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
339bf98f
TG		 3165{
3166 return NLMSG_ALIGN(sizeof(struct rtmsg))
3167 + nla_total_size(16) /* RTA_SRC */
3168 + nla_total_size(16) /* RTA_DST */
3169 + nla_total_size(16) /* RTA_GATEWAY */
3170 + nla_total_size(16) /* RTA_PREFSRC */
3171 + nla_total_size(4) /* RTA_TABLE */
3172 + nla_total_size(4) /* RTA_IIF */
3173 + nla_total_size(4) /* RTA_OIF */
3174 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 3175 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 3176 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6 3177 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
19e42e45 3178 + nla_total_size(1) /* RTA_PREF */
61adedf3 3179 + lwtunnel_get_encap_size(rt->dst.lwtstate);
339bf98f
TG		 3180}
3181
191cd582
BH		 3182static int rt6_fill_node(struct net *net,
3183 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 3184 struct in6_addr *dst, struct in6_addr *src,
15e47304 3185 int iif, int type, u32 portid, u32 seq,
7bc570c8 3186 int prefix, int nowait, unsigned int flags)
1da177e4 3187{
4b32b5ad 3188 u32 metrics[RTAX_MAX];
1da177e4 3189 struct rtmsg *rtm;
2d7202bf 3190 struct nlmsghdr *nlh;
e3703b3d 3191 long expires;
9e762a4a 3192 u32 table;
1da177e4
LT		 3193
3194 if (prefix) { /* user wants prefix routes only */
3195 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3196 /* success since this is not a prefix route */
3197 return 1;
3198 }
3199 }
3200
15e47304 3201 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 3202 if (!nlh)
26932566 3203 return -EMSGSIZE;
2d7202bf
TG		 3204
3205 rtm = nlmsg_data(nlh);
1da177e4
LT		 3206 rtm->rtm_family = AF_INET6;
3207 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3208 rtm->rtm_src_len = rt->rt6i_src.plen;
3209 rtm->rtm_tos = 0;
c71099ac 3210 if (rt->rt6i_table)
9e762a4a 3211 table = rt->rt6i_table->tb6_id;
c71099ac 3212 else
9e762a4a
PM		 3213 table = RT6_TABLE_UNSPEC;
3214 rtm->rtm_table = table;
c78679e8
DM		 3215 if (nla_put_u32(skb, RTA_TABLE, table))
3216 goto nla_put_failure;
ef2c7d7b
ND		 3217 if (rt->rt6i_flags & RTF_REJECT) {
3218 switch (rt->dst.error) {
3219 case -EINVAL:
3220 rtm->rtm_type = RTN_BLACKHOLE;
3221 break;
3222 case -EACCES:
3223 rtm->rtm_type = RTN_PROHIBIT;
3224 break;
b4949ab2
ND		 3225 case -EAGAIN:
3226 rtm->rtm_type = RTN_THROW;
3227 break;
ef2c7d7b
ND		 3228 default:
3229 rtm->rtm_type = RTN_UNREACHABLE;
3230 break;
3231 }
3232 }
38308473 3233 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 3234 rtm->rtm_type = RTN_LOCAL;
d1918542 3235 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 3236 rtm->rtm_type = RTN_LOCAL;
3237 else
3238 rtm->rtm_type = RTN_UNICAST;
3239 rtm->rtm_flags = 0;
35103d11 3240 if (!netif_carrier_ok(rt->dst.dev)) {
cea45e20 3241 rtm->rtm_flags |= RTNH_F_LINKDOWN;
35103d11
AG		 3242 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3243 rtm->rtm_flags |= RTNH_F_DEAD;
3244 }
1da177e4
LT		 3245 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3246 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 3247 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 3248 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 3249 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3250 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3251 rtm->rtm_protocol = RTPROT_RA;
3252 else
3253 rtm->rtm_protocol = RTPROT_KERNEL;
3254 }
1da177e4 3255
38308473 3256 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 3257 rtm->rtm_flags |= RTM_F_CLONED;
3258
3259 if (dst) {
930345ea 3260 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 3261 goto nla_put_failure;
1ab1457c 3262 rtm->rtm_dst_len = 128;
1da177e4 3263 } else if (rtm->rtm_dst_len)
930345ea 3264 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 3265 goto nla_put_failure;
1da177e4
LT		 3266#ifdef CONFIG_IPV6_SUBTREES
3267 if (src) {
930345ea 3268 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 3269 goto nla_put_failure;
1ab1457c 3270 rtm->rtm_src_len = 128;
c78679e8 3271 } else if (rtm->rtm_src_len &&
930345ea 3272 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 3273 goto nla_put_failure;
1da177e4 3274#endif
7bc570c8
YH		 3275 if (iif) {
3276#ifdef CONFIG_IPV6_MROUTE
3277 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
2cf75070
NA		 3278 int err = ip6mr_get_route(net, skb, rtm, nowait,
3279 portid);
3280
7bc570c8
YH		 3281 if (err <= 0) {
3282 if (!nowait) {
3283 if (err == 0)
3284 return 0;
3285 goto nla_put_failure;
3286 } else {
3287 if (err == -EMSGSIZE)
3288 goto nla_put_failure;
3289 }
3290 }
3291 } else
3292#endif
c78679e8
DM		 3293 if (nla_put_u32(skb, RTA_IIF, iif))
3294 goto nla_put_failure;
7bc570c8 3295 } else if (dst) {
1da177e4 3296 struct in6_addr saddr_buf;
c78679e8 3297 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 3298 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3299 goto nla_put_failure;
1da177e4 3300 }
2d7202bf 3301
c3968a85
DW		 3302 if (rt->rt6i_prefsrc.plen) {
3303 struct in6_addr saddr_buf;
4e3fd7a0 3304 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 3305 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3306 goto nla_put_failure;
c3968a85
DW		 3307 }
3308
4b32b5ad
MKL		 3309 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3310 if (rt->rt6i_pmtu)
3311 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3312 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 3313 goto nla_put_failure;
3314
dd0cbf29 3315 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 3316 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 3317 goto nla_put_failure;
94f826b8 3318 }
2d7202bf 3319
c78679e8
DM		 3320 if (rt->dst.dev &&
3321 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3322 goto nla_put_failure;
3323 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3324 goto nla_put_failure;
8253947e
LW		 3325
3326 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 3327
87a50699 3328 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 3329 goto nla_put_failure;
2d7202bf 3330
c78ba6d6
LR		 3331 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3332 goto nla_put_failure;
3333
ea7a8085
DA		 3334 if (lwtunnel_fill_encap(skb, rt->dst.lwtstate) < 0)
3335 goto nla_put_failure;
19e42e45 3336
053c095a
JB		 3337 nlmsg_end(skb, nlh);
3338 return 0;
2d7202bf
TG		 3339
3340nla_put_failure:
26932566
PM		 3341 nlmsg_cancel(skb, nlh);
3342 return -EMSGSIZE;
1da177e4
LT		 3343}
3344
1b43af54 3345int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 3346{
3347 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3348 int prefix;
3349
2d7202bf
TG		 3350 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3351 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 3352 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3353 } else
3354 prefix = 0;
3355
191cd582
BH		 3356 return rt6_fill_node(arg->net,
3357 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 3358 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 3359 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 3360}
3361
67ba4152 3362static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 3363{
3b1e0a65 3364 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 3365 struct nlattr *tb[RTA_MAX+1];
3366 struct rt6_info *rt;
1da177e4 3367 struct sk_buff *skb;
ab364a6f 3368 struct rtmsg *rtm;
4c9483b2 3369 struct flowi6 fl6;
72331bc0 3370 int err, iif = 0, oif = 0;
1da177e4 3371
ab364a6f
TG		 3372 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3373 if (err < 0)
3374 goto errout;
1da177e4 3375
ab364a6f 3376 err = -EINVAL;
4c9483b2 3377 memset(&fl6, 0, sizeof(fl6));
38b7097b
HFS		 3378 rtm = nlmsg_data(nlh);
3379 fl6.flowlabel = ip6_make_flowinfo(rtm->rtm_tos, 0);
1da177e4 3380
ab364a6f
TG		 3381 if (tb[RTA_SRC]) {
3382 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3383 goto errout;
3384
4e3fd7a0 3385 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 3386 }
3387
3388 if (tb[RTA_DST]) {
3389 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3390 goto errout;
3391
4e3fd7a0 3392 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 3393 }
3394
3395 if (tb[RTA_IIF])
3396 iif = nla_get_u32(tb[RTA_IIF]);
3397
3398 if (tb[RTA_OIF])
72331bc0 3399 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 3400
2e47b291
LC		 3401 if (tb[RTA_MARK])
3402 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3403
622ec2c9
LC		 3404 if (tb[RTA_UID])
3405 fl6.flowi6_uid = make_kuid(current_user_ns(),
3406 nla_get_u32(tb[RTA_UID]));
3407 else
3408 fl6.flowi6_uid = iif ? INVALID_UID : current_uid();
3409
1da177e4
LT		 3410 if (iif) {
3411 struct net_device *dev;
72331bc0
SL		 3412 int flags = 0;
3413
5578689a 3414 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 3415 if (!dev) {
3416 err = -ENODEV;
ab364a6f 3417 goto errout;
1da177e4 3418 }
72331bc0
SL		 3419
3420 fl6.flowi6_iif = iif;
3421
3422 if (!ipv6_addr_any(&fl6.saddr))
3423 flags |= RT6_LOOKUP_F_HAS_SADDR;
3424
3425 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3426 flags);
3427 } else {
3428 fl6.flowi6_oif = oif;
3429
3430 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 3431 }
3432
ab364a6f 3433 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 3434 if (!skb) {
94e187c0 3435 ip6_rt_put(rt);
ab364a6f
TG		 3436 err = -ENOBUFS;
3437 goto errout;
3438 }
1da177e4 3439
ab364a6f
TG		 3440 /* Reserve room for dummy headers, this skb can pass
3441 through good chunk of routing engine.
3442 */
459a98ed 3443 skb_reset_mac_header(skb);
ab364a6f 3444 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 3445
d8d1f30b 3446 skb_dst_set(skb, &rt->dst);
1da177e4 3447
4c9483b2 3448 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 3449 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 3450 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 3451 if (err < 0) {
ab364a6f
TG		 3452 kfree_skb(skb);
3453 goto errout;
1da177e4
LT		 3454 }
3455
15e47304 3456 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 3457errout:
1da177e4 3458 return err;
1da177e4
LT		 3459}
3460
37a1d361
RP		 3461void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3462 unsigned int nlm_flags)
1da177e4
LT		 3463{
3464 struct sk_buff *skb;
5578689a 3465 struct net *net = info->nl_net;
528c4ceb
DL		 3466 u32 seq;
3467 int err;
3468
3469 err = -ENOBUFS;
38308473 3470 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 3471
19e42e45 3472 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
38308473 3473 if (!skb)
21713ebc
TG		 3474 goto errout;
3475
191cd582 3476 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
37a1d361 3477 event, info->portid, seq, 0, 0, nlm_flags);
26932566
PM		 3478 if (err < 0) {
3479 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3480 WARN_ON(err == -EMSGSIZE);
3481 kfree_skb(skb);
3482 goto errout;
3483 }
15e47304 3484 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 3485 info->nlh, gfp_any());
3486 return;
21713ebc
TG		 3487errout:
3488 if (err < 0)
5578689a 3489 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 3490}
3491
8ed67789 3492static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 3493 unsigned long event, void *ptr)
8ed67789 3494{
351638e7 3495 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 3496 struct net *net = dev_net(dev);
8ed67789 3497
e12b8e2f
WC		 3498 if (!(dev->flags & IFF_LOOPBACK))
3499 return NOTIFY_OK;
3500
3501 if (event == NETDEV_REGISTER) {
d8d1f30b 3502 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 3503 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3504#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3505 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 3506 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 3507 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789 3508 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
e12b8e2f
WC		 3509#endif
3510 } else if (event == NETDEV_UNREGISTER) {
3511 in6_dev_put(net->ipv6.ip6_null_entry->rt6i_idev);
3512#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3513 in6_dev_put(net->ipv6.ip6_prohibit_entry->rt6i_idev);
3514 in6_dev_put(net->ipv6.ip6_blk_hole_entry->rt6i_idev);
8ed67789
DL		 3515#endif
3516 }
3517
3518 return NOTIFY_OK;
3519}
3520
1da177e4
LT		 3521/*
3522 * /proc
3523 */
3524
3525#ifdef CONFIG_PROC_FS
3526
33120b30
AD		 3527static const struct file_operations ipv6_route_proc_fops = {
3528 .owner = THIS_MODULE,
3529 .open = ipv6_route_open,
3530 .read = seq_read,
3531 .llseek = seq_lseek,
8d2ca1d7 3532 .release = seq_release_net,
33120b30
AD		 3533};
3534
1da177e4
LT		 3535static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3536{
69ddb805 3537 struct net *net = (struct net *)seq->private;
1da177e4 3538 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 3539 net->ipv6.rt6_stats->fib_nodes,
3540 net->ipv6.rt6_stats->fib_route_nodes,
3541 net->ipv6.rt6_stats->fib_rt_alloc,
3542 net->ipv6.rt6_stats->fib_rt_entries,
3543 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 3544 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 3545 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 3546
3547 return 0;
3548}
3549
3550static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3551{
de05c557 3552 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 3553}
3554
9a32144e 3555static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 3556 .owner = THIS_MODULE,
3557 .open = rt6_stats_seq_open,
3558 .read = seq_read,
3559 .llseek = seq_lseek,
b6fcbdb4 3560 .release = single_release_net,
1da177e4
LT		 3561};
3562#endif /* CONFIG_PROC_FS */
3563
3564#ifdef CONFIG_SYSCTL
3565
1da177e4 3566static
fe2c6338 3567int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3568 void __user *buffer, size_t *lenp, loff_t *ppos)
3569{
c486da34
LAG		 3570 struct net *net;
3571 int delay;
3572 if (!write)
1da177e4 3573 return -EINVAL;
c486da34
LAG		 3574
3575 net = (struct net *)ctl->extra1;
3576 delay = net->ipv6.sysctl.flush_delay;
3577 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3578 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3579 return 0;
1da177e4
LT		 3580}
3581
fe2c6338 3582struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3583 {
1da177e4 3584 .procname = "flush",
4990509f 3585 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3586 .maxlen = sizeof(int),
89c8b3a1 3587 .mode = 0200,
6d9f239a 3588 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3589 },
3590 {
1da177e4 3591 .procname = "gc_thresh",
9a7ec3a9 3592 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3593 .maxlen = sizeof(int),
3594 .mode = 0644,
6d9f239a 3595 .proc_handler = proc_dointvec,
1da177e4
LT		 3596 },
3597 {
1da177e4 3598 .procname = "max_size",
4990509f 3599 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3600 .maxlen = sizeof(int),
3601 .mode = 0644,
6d9f239a 3602 .proc_handler = proc_dointvec,
1da177e4
LT		 3603 },
3604 {
1da177e4 3605 .procname = "gc_min_interval",
4990509f 3606 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3607 .maxlen = sizeof(int),
3608 .mode = 0644,
6d9f239a 3609 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3610 },
3611 {
1da177e4 3612 .procname = "gc_timeout",
4990509f 3613 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3614 .maxlen = sizeof(int),
3615 .mode = 0644,
6d9f239a 3616 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3617 },
3618 {
1da177e4 3619 .procname = "gc_interval",
4990509f 3620 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3621 .maxlen = sizeof(int),
3622 .mode = 0644,
6d9f239a 3623 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3624 },
3625 {
1da177e4 3626 .procname = "gc_elasticity",
4990509f 3627 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3628 .maxlen = sizeof(int),
3629 .mode = 0644,
f3d3f616 3630 .proc_handler = proc_dointvec,
1da177e4
LT		 3631 },
3632 {
1da177e4 3633 .procname = "mtu_expires",
4990509f 3634 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3635 .maxlen = sizeof(int),
3636 .mode = 0644,
6d9f239a 3637 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3638 },
3639 {
1da177e4 3640 .procname = "min_adv_mss",
4990509f 3641 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3642 .maxlen = sizeof(int),
3643 .mode = 0644,
f3d3f616 3644 .proc_handler = proc_dointvec,
1da177e4
LT		 3645 },
3646 {
1da177e4 3647 .procname = "gc_min_interval_ms",
4990509f 3648 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3649 .maxlen = sizeof(int),
3650 .mode = 0644,
6d9f239a 3651 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3652 },
f8572d8f 3653 { }
1da177e4
LT		 3654};
3655
2c8c1e72 3656struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3657{
3658 struct ctl_table *table;
3659
3660 table = kmemdup(ipv6_route_table_template,
3661 sizeof(ipv6_route_table_template),
3662 GFP_KERNEL);
5ee09105
YH		 3663
3664 if (table) {
3665 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3666 table[0].extra1 = net;
86393e52 3667 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3668 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3669 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3670 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3671 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3672 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3673 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3674 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3675 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3676
3677 /* Don't export sysctls to unprivileged users */
3678 if (net->user_ns != &init_user_ns)
3679 table[0].procname = NULL;
5ee09105
YH		 3680 }
3681
760f2d01
DL		 3682 return table;
3683}
1da177e4
LT		 3684#endif
3685
2c8c1e72 3686static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3687{
633d424b 3688 int ret = -ENOMEM;
8ed67789 3689
86393e52
AD		 3690 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3691 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3692
fc66f95c
ED		 3693 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3694 goto out_ip6_dst_ops;
3695
8ed67789
DL		 3696 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3697 sizeof(*net->ipv6.ip6_null_entry),
3698 GFP_KERNEL);
3699 if (!net->ipv6.ip6_null_entry)
fc66f95c 3700 goto out_ip6_dst_entries;
d8d1f30b 3701 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3702 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3703 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3704 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3705 ip6_template_metrics, true);
8ed67789
DL		 3706
3707#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3708 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3709 sizeof(*net->ipv6.ip6_prohibit_entry),
3710 GFP_KERNEL);
68fffc67
PZ		 3711 if (!net->ipv6.ip6_prohibit_entry)
3712 goto out_ip6_null_entry;
d8d1f30b 3713 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3714 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3715 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3716 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3717 ip6_template_metrics, true);
8ed67789
DL		 3718
3719 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3720 sizeof(*net->ipv6.ip6_blk_hole_entry),
3721 GFP_KERNEL);
68fffc67
PZ		 3722 if (!net->ipv6.ip6_blk_hole_entry)
3723 goto out_ip6_prohibit_entry;
d8d1f30b 3724 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3725 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3726 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3727 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3728 ip6_template_metrics, true);
8ed67789
DL		 3729#endif
3730
b339a47c
PZ		 3731 net->ipv6.sysctl.flush_delay = 0;
3732 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3733 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3734 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3735 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3736 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3737 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3738 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3739
6891a346
BT		 3740 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3741
8ed67789
DL		 3742 ret = 0;
3743out:
3744 return ret;
f2fc6a54 3745
68fffc67
PZ		 3746#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3747out_ip6_prohibit_entry:
3748 kfree(net->ipv6.ip6_prohibit_entry);
3749out_ip6_null_entry:
3750 kfree(net->ipv6.ip6_null_entry);
3751#endif
fc66f95c
ED		 3752out_ip6_dst_entries:
3753 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3754out_ip6_dst_ops:
f2fc6a54 3755 goto out;
cdb18761
DL		 3756}
3757
2c8c1e72 3758static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3759{
8ed67789
DL		 3760 kfree(net->ipv6.ip6_null_entry);
3761#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3762 kfree(net->ipv6.ip6_prohibit_entry);
3763 kfree(net->ipv6.ip6_blk_hole_entry);
3764#endif
41bb78b4 3765 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3766}
3767
d189634e
TG		 3768static int __net_init ip6_route_net_init_late(struct net *net)
3769{
3770#ifdef CONFIG_PROC_FS
d4beaa66
G		 3771 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3772 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3773#endif
3774 return 0;
3775}
3776
3777static void __net_exit ip6_route_net_exit_late(struct net *net)
3778{
3779#ifdef CONFIG_PROC_FS
ece31ffd
G		 3780 remove_proc_entry("ipv6_route", net->proc_net);
3781 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3782#endif
3783}
3784
cdb18761
DL		 3785static struct pernet_operations ip6_route_net_ops = {
3786 .init = ip6_route_net_init,
3787 .exit = ip6_route_net_exit,
3788};
3789
c3426b47
DM		 3790static int __net_init ipv6_inetpeer_init(struct net *net)
3791{
3792 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3793
3794 if (!bp)
3795 return -ENOMEM;
3796 inet_peer_base_init(bp);
3797 net->ipv6.peers = bp;
3798 return 0;
3799}
3800
3801static void __net_exit ipv6_inetpeer_exit(struct net *net)
3802{
3803 struct inet_peer_base *bp = net->ipv6.peers;
3804
3805 net->ipv6.peers = NULL;
56a6b248 3806 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3807 kfree(bp);
3808}
3809
2b823f72 3810static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3811 .init = ipv6_inetpeer_init,
3812 .exit = ipv6_inetpeer_exit,
3813};
3814
d189634e
TG		 3815static struct pernet_operations ip6_route_net_late_ops = {
3816 .init = ip6_route_net_init_late,
3817 .exit = ip6_route_net_exit_late,
3818};
3819
8ed67789
DL		 3820static struct notifier_block ip6_route_dev_notifier = {
3821 .notifier_call = ip6_route_dev_notify,
e12b8e2f 3822 .priority = ADDRCONF_NOTIFY_PRIORITY - 10,
8ed67789
DL		 3823};
3824
3f0bbd5f
WC		 3825void __init ip6_route_init_special_entries(void)
3826{
3827 /* Registering of the loopback is done before this portion of code,
3828 * the loopback reference in rt6_info will not be taken, do it
3829 * manually for init_net */
3830 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
3831 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3832 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
3833 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
3834 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3835 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
3836 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3837 #endif
3838}
3839
433d49c3 3840int __init ip6_route_init(void)
1da177e4 3841{
433d49c3 3842 int ret;
8d0b94af 3843 int cpu;
433d49c3 3844
9a7ec3a9
DL		 3845 ret = -ENOMEM;
3846 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3847 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3848 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3849 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3850 goto out;
14e50e57 3851
fc66f95c 3852 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3853 if (ret)
bdb3289f 3854 goto out_kmem_cache;
bdb3289f 3855
c3426b47
DM		 3856 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3857 if (ret)
e8803b6c 3858 goto out_dst_entries;
2a0c451a 3859
7e52b33b
DM		 3860 ret = register_pernet_subsys(&ip6_route_net_ops);
3861 if (ret)
3862 goto out_register_inetpeer;
c3426b47 3863
5dc121e9
AE		 3864 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3865
e8803b6c 3866 ret = fib6_init();
433d49c3 3867 if (ret)
8ed67789 3868 goto out_register_subsys;
433d49c3 3869
433d49c3
DL		 3870 ret = xfrm6_init();
3871 if (ret)
e8803b6c 3872 goto out_fib6_init;
c35b7e72 3873
433d49c3
DL		 3874 ret = fib6_rules_init();
3875 if (ret)
3876 goto xfrm6_init;
7e5449c2 3877
d189634e
TG		 3878 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3879 if (ret)
3880 goto fib6_rules_init;
3881
433d49c3 3882 ret = -ENOBUFS;
c7ac8679
GR		 3883 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3884 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3885 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3886 goto out_register_late_subsys;
c127ea2c 3887
8ed67789 3888 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3889 if (ret)
d189634e 3890 goto out_register_late_subsys;
8ed67789 3891
8d0b94af
MKL		 3892 for_each_possible_cpu(cpu) {
3893 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3894
3895 INIT_LIST_HEAD(&ul->head);
3896 spin_lock_init(&ul->lock);
3897 }
3898
433d49c3
DL		 3899out:
3900 return ret;
3901
d189634e
TG		 3902out_register_late_subsys:
3903 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3904fib6_rules_init:
433d49c3
DL		 3905 fib6_rules_cleanup();
3906xfrm6_init:
433d49c3 3907 xfrm6_fini();
2a0c451a
TG		 3908out_fib6_init:
3909 fib6_gc_cleanup();
8ed67789
DL		 3910out_register_subsys:
3911 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3912out_register_inetpeer:
3913 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3914out_dst_entries:
3915 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3916out_kmem_cache:
f2fc6a54 3917 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3918 goto out;
1da177e4
LT		 3919}
3920
3921void ip6_route_cleanup(void)
3922{
8ed67789 3923 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3924 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3925 fib6_rules_cleanup();
1da177e4 3926 xfrm6_fini();
1da177e4 3927 fib6_gc_cleanup();
c3426b47 3928 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3929 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3930 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3931 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3932}
ubuntu-zesty-kernel mirror