]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - net/ipv6/route.c
projects / mirror_ubuntu-artful-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Merge branch 'mv88e6390-batch-three'
[mirror_ubuntu-artful-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
LT		 66
67#include <asm/uaccess.h>
68
69#ifdef CONFIG_SYSCTL
70#include <linux/sysctl.h>
71#endif
72
afc154e9 73enum rt6_nud_state {
7e980569
JB		 74 RT6_NUD_FAIL_HARD = -3,
75 RT6_NUD_FAIL_PROBE = -2,
76 RT6_NUD_FAIL_DO_RR = -1,
afc154e9
HFS		 77 RT6_NUD_SUCCEED = 1
78};
79
83a09abd 80static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort);
1da177e4 81static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
0dbaee3b 82static unsigned int ip6_default_advmss(const struct dst_entry *dst);
ebb762f2 83static unsigned int ip6_mtu(const struct dst_entry *dst);
1da177e4
LT		 84static struct dst_entry *ip6_negative_advice(struct dst_entry *);
85static void ip6_dst_destroy(struct dst_entry *);
86static void ip6_dst_ifdown(struct dst_entry *,
87 struct net_device *dev, int how);
569d3645 88static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 89
90static int ip6_pkt_discard(struct sk_buff *skb);
ede2059d 91static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb);
7150aede 92static int ip6_pkt_prohibit(struct sk_buff *skb);
ede2059d 93static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1da177e4 94static void ip6_link_failure(struct sk_buff *skb);
6700c270
DM		 95static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
96 struct sk_buff *skb, u32 mtu);
97static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
98 struct sk_buff *skb);
4b32b5ad 99static void rt6_dst_from_metrics_check(struct rt6_info *rt);
52bd4c0c 100static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
1da177e4 101
70ceb4f5 102#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 103static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42 104 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 105 const struct in6_addr *gwaddr,
106 struct net_device *dev,
95c96174 107 unsigned int pref);
efa2cea0 108static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42 109 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 110 const struct in6_addr *gwaddr,
111 struct net_device *dev);
70ceb4f5
YH		 112#endif
113
8d0b94af
MKL		 114struct uncached_list {
115 spinlock_t lock;
116 struct list_head head;
117};
118
119static DEFINE_PER_CPU_ALIGNED(struct uncached_list, rt6_uncached_list);
120
121static void rt6_uncached_list_add(struct rt6_info *rt)
122{
123 struct uncached_list *ul = raw_cpu_ptr(&rt6_uncached_list);
124
125 rt->dst.flags |= DST_NOCACHE;
126 rt->rt6i_uncached_list = ul;
127
128 spin_lock_bh(&ul->lock);
129 list_add_tail(&rt->rt6i_uncached, &ul->head);
130 spin_unlock_bh(&ul->lock);
131}
132
133static void rt6_uncached_list_del(struct rt6_info *rt)
134{
135 if (!list_empty(&rt->rt6i_uncached)) {
136 struct uncached_list *ul = rt->rt6i_uncached_list;
137
138 spin_lock_bh(&ul->lock);
139 list_del(&rt->rt6i_uncached);
140 spin_unlock_bh(&ul->lock);
141 }
142}
143
144static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
145{
146 struct net_device *loopback_dev = net->loopback_dev;
147 int cpu;
148
e332bc67
EB		 149 if (dev == loopback_dev)
150 return;
151
8d0b94af
MKL		 152 for_each_possible_cpu(cpu) {
153 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
154 struct rt6_info *rt;
155
156 spin_lock_bh(&ul->lock);
157 list_for_each_entry(rt, &ul->head, rt6i_uncached) {
158 struct inet6_dev *rt_idev = rt->rt6i_idev;
159 struct net_device *rt_dev = rt->dst.dev;
160
e332bc67 161 if (rt_idev->dev == dev) {
8d0b94af
MKL		 162 rt->rt6i_idev = in6_dev_get(loopback_dev);
163 in6_dev_put(rt_idev);
164 }
165
e332bc67 166 if (rt_dev == dev) {
8d0b94af
MKL		 167 rt->dst.dev = loopback_dev;
168 dev_hold(rt->dst.dev);
169 dev_put(rt_dev);
170 }
171 }
172 spin_unlock_bh(&ul->lock);
173 }
174}
175
d52d3997
MKL		 176static u32 *rt6_pcpu_cow_metrics(struct rt6_info *rt)
177{
178 return dst_metrics_write_ptr(rt->dst.from);
179}
180
06582540
DM		 181static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
182{
4b32b5ad 183 struct rt6_info *rt = (struct rt6_info *)dst;
06582540 184
d52d3997
MKL		 185 if (rt->rt6i_flags & RTF_PCPU)
186 return rt6_pcpu_cow_metrics(rt);
187 else if (rt->rt6i_flags & RTF_CACHE)
4b32b5ad
MKL		 188 return NULL;
189 else
3b471175 190 return dst_cow_metrics_generic(dst, old);
06582540
DM		 191}
192
f894cbf8
DM		 193static inline const void *choose_neigh_daddr(struct rt6_info *rt,
194 struct sk_buff *skb,
195 const void *daddr)
39232973
DM		 196{
197 struct in6_addr *p = &rt->rt6i_gateway;
198
a7563f34 199 if (!ipv6_addr_any(p))
39232973 200 return (const void *) p;
f894cbf8
DM		 201 else if (skb)
202 return &ipv6_hdr(skb)->daddr;
39232973
DM		 203 return daddr;
204}
205
f894cbf8
DM		 206static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
207 struct sk_buff *skb,
208 const void *daddr)
d3aaeb38 209{
39232973
DM		 210 struct rt6_info *rt = (struct rt6_info *) dst;
211 struct neighbour *n;
212
f894cbf8 213 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 214 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 215 if (n)
216 return n;
217 return neigh_create(&nd_tbl, daddr, dst->dev);
218}
219
9a7ec3a9 220static struct dst_ops ip6_dst_ops_template = {
1da177e4 221 .family = AF_INET6,
1da177e4
LT		 222 .gc = ip6_dst_gc,
223 .gc_thresh = 1024,
224 .check = ip6_dst_check,
0dbaee3b 225 .default_advmss = ip6_default_advmss,
ebb762f2 226 .mtu = ip6_mtu,
06582540 227 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 228 .destroy = ip6_dst_destroy,
229 .ifdown = ip6_dst_ifdown,
230 .negative_advice = ip6_negative_advice,
231 .link_failure = ip6_link_failure,
232 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 233 .redirect = rt6_do_redirect,
9f8955cc 234 .local_out = __ip6_local_out,
d3aaeb38 235 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 236};
237
ebb762f2 238static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 239{
618f9bc7
SK		 240 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
241
242 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 243}
244
6700c270
DM		 245static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
246 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 247{
248}
249
6700c270
DM		 250static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
251 struct sk_buff *skb)
b587ee3b
DM		 252{
253}
254
14e50e57
DM		 255static struct dst_ops ip6_dst_blackhole_ops = {
256 .family = AF_INET6,
14e50e57
DM		 257 .destroy = ip6_dst_destroy,
258 .check = ip6_dst_check,
ebb762f2 259 .mtu = ip6_blackhole_mtu,
214f45c9 260 .default_advmss = ip6_default_advmss,
14e50e57 261 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 262 .redirect = ip6_rt_blackhole_redirect,
0a1f5962 263 .cow_metrics = dst_cow_metrics_generic,
d3aaeb38 264 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 265};
266
62fa8a84 267static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 268 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 269};
270
fb0af4c7 271static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 272 .dst = {
273 .__refcnt = ATOMIC_INIT(1),
274 .__use = 1,
2c20cbd7 275 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 276 .error = -ENETUNREACH,
d8d1f30b
CG		 277 .input = ip6_pkt_discard,
278 .output = ip6_pkt_discard_out,
1da177e4
LT		 279 },
280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 281 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 282 .rt6i_metric = ~(u32) 0,
283 .rt6i_ref = ATOMIC_INIT(1),
284};
285
101367c2
TG		 286#ifdef CONFIG_IPV6_MULTIPLE_TABLES
287
fb0af4c7 288static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 289 .dst = {
290 .__refcnt = ATOMIC_INIT(1),
291 .__use = 1,
2c20cbd7 292 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 293 .error = -EACCES,
d8d1f30b
CG		 294 .input = ip6_pkt_prohibit,
295 .output = ip6_pkt_prohibit_out,
101367c2
TG		 296 },
297 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 298 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 299 .rt6i_metric = ~(u32) 0,
300 .rt6i_ref = ATOMIC_INIT(1),
301};
302
fb0af4c7 303static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 304 .dst = {
305 .__refcnt = ATOMIC_INIT(1),
306 .__use = 1,
2c20cbd7 307 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 308 .error = -EINVAL,
d8d1f30b 309 .input = dst_discard,
ede2059d 310 .output = dst_discard_out,
101367c2
TG		 311 },
312 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 313 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 314 .rt6i_metric = ~(u32) 0,
315 .rt6i_ref = ATOMIC_INIT(1),
316};
317
318#endif
319
ebfa45f0
MKL		 320static void rt6_info_init(struct rt6_info *rt)
321{
322 struct dst_entry *dst = &rt->dst;
323
324 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
325 INIT_LIST_HEAD(&rt->rt6i_siblings);
326 INIT_LIST_HEAD(&rt->rt6i_uncached);
327}
328
1da177e4 329/* allocate dst with ip6_dst_ops */
d52d3997
MKL		 330static struct rt6_info *__ip6_dst_alloc(struct net *net,
331 struct net_device *dev,
ad706862 332 int flags)
1da177e4 333{
97bab73f 334 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 335 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 336
ebfa45f0
MKL		 337 if (rt)
338 rt6_info_init(rt);
8104891b 339
cf911662 340 return rt;
1da177e4
LT		 341}
342
9ab179d8
DA		 343struct rt6_info *ip6_dst_alloc(struct net *net,
344 struct net_device *dev,
345 int flags)
d52d3997 346{
ad706862 347 struct rt6_info *rt = __ip6_dst_alloc(net, dev, flags);
d52d3997
MKL		 348
349 if (rt) {
350 rt->rt6i_pcpu = alloc_percpu_gfp(struct rt6_info *, GFP_ATOMIC);
351 if (rt->rt6i_pcpu) {
352 int cpu;
353
354 for_each_possible_cpu(cpu) {
355 struct rt6_info **p;
356
357 p = per_cpu_ptr(rt->rt6i_pcpu, cpu);
358 /* no one shares rt */
359 *p = NULL;
360 }
361 } else {
362 dst_destroy((struct dst_entry *)rt);
363 return NULL;
364 }
365 }
366
367 return rt;
368}
9ab179d8 369EXPORT_SYMBOL(ip6_dst_alloc);
d52d3997 370
1da177e4
LT		 371static void ip6_dst_destroy(struct dst_entry *dst)
372{
373 struct rt6_info *rt = (struct rt6_info *)dst;
ecd98837 374 struct dst_entry *from = dst->from;
8d0b94af 375 struct inet6_dev *idev;
1da177e4 376
4b32b5ad 377 dst_destroy_metrics_generic(dst);
87775312 378 free_percpu(rt->rt6i_pcpu);
8d0b94af
MKL		 379 rt6_uncached_list_del(rt);
380
381 idev = rt->rt6i_idev;
38308473 382 if (idev) {
1da177e4
LT		 383 rt->rt6i_idev = NULL;
384 in6_dev_put(idev);
1ab1457c 385 }
1716a961 386
ecd98837
YH		 387 dst->from = NULL;
388 dst_release(from);
b3419363
DM		 389}
390
1da177e4
LT		 391static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
392 int how)
393{
394 struct rt6_info *rt = (struct rt6_info *)dst;
395 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 396 struct net_device *loopback_dev =
c346dca1 397 dev_net(dev)->loopback_dev;
1da177e4 398
97cac082
DM		 399 if (dev != loopback_dev) {
400 if (idev && idev->dev == dev) {
401 struct inet6_dev *loopback_idev =
402 in6_dev_get(loopback_dev);
403 if (loopback_idev) {
404 rt->rt6i_idev = loopback_idev;
405 in6_dev_put(idev);
406 }
407 }
1da177e4
LT		 408 }
409}
410
5973fb1e
MKL		 411static bool __rt6_check_expired(const struct rt6_info *rt)
412{
413 if (rt->rt6i_flags & RTF_EXPIRES)
414 return time_after(jiffies, rt->dst.expires);
415 else
416 return false;
417}
418
a50feda5 419static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 420{
1716a961
G		 421 if (rt->rt6i_flags & RTF_EXPIRES) {
422 if (time_after(jiffies, rt->dst.expires))
a50feda5 423 return true;
1716a961 424 } else if (rt->dst.from) {
3fd91fb3 425 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 426 }
a50feda5 427 return false;
1da177e4
LT		 428}
429
51ebd318
ND		 430/* Multipath route selection:
431 * Hash based function using packet header and flowlabel.
432 * Adapted from fib_info_hashfn()
433 */
434static int rt6_info_hash_nhsfn(unsigned int candidate_count,
435 const struct flowi6 *fl6)
436{
644d0e65 437 return get_hash_from_flowi6(fl6) % candidate_count;
51ebd318
ND		 438}
439
440static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 441 struct flowi6 *fl6, int oif,
442 int strict)
51ebd318
ND		 443{
444 struct rt6_info *sibling, *next_sibling;
445 int route_choosen;
446
447 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
448 /* Don't change the route, if route_choosen == 0
449 * (siblings does not include ourself)
450 */
451 if (route_choosen)
452 list_for_each_entry_safe(sibling, next_sibling,
453 &match->rt6i_siblings, rt6i_siblings) {
454 route_choosen--;
455 if (route_choosen == 0) {
52bd4c0c
ND		 456 if (rt6_score_route(sibling, oif, strict) < 0)
457 break;
51ebd318
ND		 458 match = sibling;
459 break;
460 }
461 }
462 return match;
463}
464
1da177e4 465/*
c71099ac 466 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 467 */
468
8ed67789
DL		 469static inline struct rt6_info *rt6_device_match(struct net *net,
470 struct rt6_info *rt,
b71d1d42 471 const struct in6_addr *saddr,
1da177e4 472 int oif,
d420895e 473 int flags)
1da177e4
LT		 474{
475 struct rt6_info *local = NULL;
476 struct rt6_info *sprt;
477
dd3abc4e
YH		 478 if (!oif && ipv6_addr_any(saddr))
479 goto out;
480
d8d1f30b 481 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 482 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 483
484 if (oif) {
1da177e4
LT		 485 if (dev->ifindex == oif)
486 return sprt;
487 if (dev->flags & IFF_LOOPBACK) {
38308473 488 if (!sprt->rt6i_idev ||
1da177e4 489 sprt->rt6i_idev->dev->ifindex != oif) {
17fb0b2b 490 if (flags & RT6_LOOKUP_F_IFACE)
1da177e4 491 continue;
17fb0b2b
DA		 492 if (local &&
493 local->rt6i_idev->dev->ifindex == oif)
1da177e4
LT		 494 continue;
495 }
496 local = sprt;
497 }
dd3abc4e
YH		 498 } else {
499 if (ipv6_chk_addr(net, saddr, dev,
500 flags & RT6_LOOKUP_F_IFACE))
501 return sprt;
1da177e4 502 }
dd3abc4e 503 }
1da177e4 504
dd3abc4e 505 if (oif) {
1da177e4
LT		 506 if (local)
507 return local;
508
d420895e 509 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 510 return net->ipv6.ip6_null_entry;
1da177e4 511 }
dd3abc4e 512out:
1da177e4
LT		 513 return rt;
514}
515
27097255 516#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 517struct __rt6_probe_work {
518 struct work_struct work;
519 struct in6_addr target;
520 struct net_device *dev;
521};
522
523static void rt6_probe_deferred(struct work_struct *w)
524{
525 struct in6_addr mcaddr;
526 struct __rt6_probe_work *work =
527 container_of(w, struct __rt6_probe_work, work);
528
529 addrconf_addr_solict_mult(&work->target, &mcaddr);
304d888b 530 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL);
c2f17e82 531 dev_put(work->dev);
662f5533 532 kfree(work);
c2f17e82
HFS		 533}
534
27097255
YH		 535static void rt6_probe(struct rt6_info *rt)
536{
990edb42 537 struct __rt6_probe_work *work;
f2c31e32 538 struct neighbour *neigh;
27097255
YH		 539 /*
540 * Okay, this does not seem to be appropriate
541 * for now, however, we need to check if it
542 * is really so; aka Router Reachability Probing.
543 *
544 * Router Reachability Probe MUST be rate-limited
545 * to no more than one per minute.
546 */
2152caea 547 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 548 return;
2152caea
YH		 549 rcu_read_lock_bh();
550 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
551 if (neigh) {
8d6c31bf
MKL		 552 if (neigh->nud_state & NUD_VALID)
553 goto out;
554
990edb42 555 work = NULL;
2152caea 556 write_lock(&neigh->lock);
990edb42
MKL		 557 if (!(neigh->nud_state & NUD_VALID) &&
558 time_after(jiffies,
559 neigh->updated +
560 rt->rt6i_idev->cnf.rtr_probe_interval)) {
561 work = kmalloc(sizeof(*work), GFP_ATOMIC);
562 if (work)
563 __neigh_set_probe_once(neigh);
c2f17e82 564 }
2152caea 565 write_unlock(&neigh->lock);
990edb42
MKL		 566 } else {
567 work = kmalloc(sizeof(*work), GFP_ATOMIC);
f2c31e32 568 }
990edb42
MKL		 569
570 if (work) {
571 INIT_WORK(&work->work, rt6_probe_deferred);
572 work->target = rt->rt6i_gateway;
573 dev_hold(rt->dst.dev);
574 work->dev = rt->dst.dev;
575 schedule_work(&work->work);
576 }
577
8d6c31bf 578out:
2152caea 579 rcu_read_unlock_bh();
27097255
YH		 580}
581#else
582static inline void rt6_probe(struct rt6_info *rt)
583{
27097255
YH		 584}
585#endif
586
1da177e4 587/*
554cfb7e 588 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 589 */
b6f99a21 590static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 591{
d1918542 592 struct net_device *dev = rt->dst.dev;
161980f4 593 if (!oif || dev->ifindex == oif)
554cfb7e 594 return 2;
161980f4
DM		 595 if ((dev->flags & IFF_LOOPBACK) &&
596 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
597 return 1;
598 return 0;
554cfb7e 599}
1da177e4 600
afc154e9 601static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 602{
f2c31e32 603 struct neighbour *neigh;
afc154e9 604 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 605
4d0c5911
YH		 606 if (rt->rt6i_flags & RTF_NONEXTHOP ||
607 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 608 return RT6_NUD_SUCCEED;
145a3621
YH		 609
610 rcu_read_lock_bh();
611 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
612 if (neigh) {
613 read_lock(&neigh->lock);
554cfb7e 614 if (neigh->nud_state & NUD_VALID)
afc154e9 615 ret = RT6_NUD_SUCCEED;
398bcbeb 616#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 617 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 618 ret = RT6_NUD_SUCCEED;
7e980569
JB		 619 else
620 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 621#endif
145a3621 622 read_unlock(&neigh->lock);
afc154e9
HFS		 623 } else {
624 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 625 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 626 }
145a3621
YH		 627 rcu_read_unlock_bh();
628
a5a81f0b 629 return ret;
1da177e4
LT		 630}
631
554cfb7e
YH		 632static int rt6_score_route(struct rt6_info *rt, int oif,
633 int strict)
1da177e4 634{
a5a81f0b 635 int m;
1ab1457c 636
4d0c5911 637 m = rt6_check_dev(rt, oif);
77d16f45 638 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 639 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 640#ifdef CONFIG_IPV6_ROUTER_PREF
641 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
642#endif
afc154e9
HFS		 643 if (strict & RT6_LOOKUP_F_REACHABLE) {
644 int n = rt6_check_neigh(rt);
645 if (n < 0)
646 return n;
647 }
554cfb7e
YH		 648 return m;
649}
650
f11e6659 651static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 652 int *mpri, struct rt6_info *match,
653 bool *do_rr)
554cfb7e 654{
f11e6659 655 int m;
afc154e9 656 bool match_do_rr = false;
35103d11
AG		 657 struct inet6_dev *idev = rt->rt6i_idev;
658 struct net_device *dev = rt->dst.dev;
659
660 if (dev && !netif_carrier_ok(dev) &&
d5d32e4b
DA		 661 idev->cnf.ignore_routes_with_linkdown &&
662 !(strict & RT6_LOOKUP_F_IGNORE_LINKSTATE))
35103d11 663 goto out;
f11e6659
DM		 664
665 if (rt6_check_expired(rt))
666 goto out;
667
668 m = rt6_score_route(rt, oif, strict);
7e980569 669 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 670 match_do_rr = true;
671 m = 0; /* lowest valid score */
7e980569 672 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 673 goto out;
afc154e9
HFS		 674 }
675
676 if (strict & RT6_LOOKUP_F_REACHABLE)
677 rt6_probe(rt);
f11e6659 678
7e980569 679 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 680 if (m > *mpri) {
afc154e9 681 *do_rr = match_do_rr;
f11e6659
DM		 682 *mpri = m;
683 match = rt;
f11e6659 684 }
f11e6659
DM		 685out:
686 return match;
687}
688
689static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
690 struct rt6_info *rr_head,
afc154e9
HFS		 691 u32 metric, int oif, int strict,
692 bool *do_rr)
f11e6659 693{
9fbdcfaf 694 struct rt6_info *rt, *match, *cont;
554cfb7e 695 int mpri = -1;
1da177e4 696
f11e6659 697 match = NULL;
9fbdcfaf
SK		 698 cont = NULL;
699 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
700 if (rt->rt6i_metric != metric) {
701 cont = rt;
702 break;
703 }
704
705 match = find_match(rt, oif, strict, &mpri, match, do_rr);
706 }
707
708 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
709 if (rt->rt6i_metric != metric) {
710 cont = rt;
711 break;
712 }
713
afc154e9 714 match = find_match(rt, oif, strict, &mpri, match, do_rr);
9fbdcfaf
SK		 715 }
716
717 if (match || !cont)
718 return match;
719
720 for (rt = cont; rt; rt = rt->dst.rt6_next)
afc154e9 721 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 722
f11e6659
DM		 723 return match;
724}
1da177e4 725
f11e6659
DM		 726static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
727{
728 struct rt6_info *match, *rt0;
8ed67789 729 struct net *net;
afc154e9 730 bool do_rr = false;
1da177e4 731
f11e6659
DM		 732 rt0 = fn->rr_ptr;
733 if (!rt0)
734 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 735
afc154e9
HFS		 736 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
737 &do_rr);
1da177e4 738
afc154e9 739 if (do_rr) {
d8d1f30b 740 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 741
554cfb7e 742 /* no entries matched; do round-robin */
f11e6659
DM		 743 if (!next || next->rt6i_metric != rt0->rt6i_metric)
744 next = fn->leaf;
745
746 if (next != rt0)
747 fn->rr_ptr = next;
1da177e4 748 }
1da177e4 749
d1918542 750 net = dev_net(rt0->dst.dev);
a02cec21 751 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 752}
753
8b9df265
MKL		 754static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
755{
756 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
757}
758
70ceb4f5
YH		 759#ifdef CONFIG_IPV6_ROUTE_INFO
760int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 761 const struct in6_addr *gwaddr)
70ceb4f5 762{
c346dca1 763 struct net *net = dev_net(dev);
70ceb4f5
YH		 764 struct route_info *rinfo = (struct route_info *) opt;
765 struct in6_addr prefix_buf, *prefix;
766 unsigned int pref;
4bed72e4 767 unsigned long lifetime;
70ceb4f5
YH		 768 struct rt6_info *rt;
769
770 if (len < sizeof(struct route_info)) {
771 return -EINVAL;
772 }
773
774 /* Sanity check for prefix_len and length */
775 if (rinfo->length > 3) {
776 return -EINVAL;
777 } else if (rinfo->prefix_len > 128) {
778 return -EINVAL;
779 } else if (rinfo->prefix_len > 64) {
780 if (rinfo->length < 2) {
781 return -EINVAL;
782 }
783 } else if (rinfo->prefix_len > 0) {
784 if (rinfo->length < 1) {
785 return -EINVAL;
786 }
787 }
788
789 pref = rinfo->route_pref;
790 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 791 return -EINVAL;
70ceb4f5 792
4bed72e4 793 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 794
795 if (rinfo->length == 3)
796 prefix = (struct in6_addr *)rinfo->prefix;
797 else {
798 /* this function is safe */
799 ipv6_addr_prefix(&prefix_buf,
800 (struct in6_addr *)rinfo->prefix,
801 rinfo->prefix_len);
802 prefix = &prefix_buf;
803 }
804
f104a567
DJ		 805 if (rinfo->prefix_len == 0)
806 rt = rt6_get_dflt_router(gwaddr, dev);
807 else
808 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
830218c1 809 gwaddr, dev);
70ceb4f5
YH		 810
811 if (rt && !lifetime) {
e0a1ad73 812 ip6_del_rt(rt);
70ceb4f5
YH		 813 rt = NULL;
814 }
815
816 if (!rt && lifetime)
830218c1
DA		 817 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr,
818 dev, pref);
70ceb4f5
YH		 819 else if (rt)
820 rt->rt6i_flags = RTF_ROUTEINFO |
821 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
822
823 if (rt) {
1716a961
G		 824 if (!addrconf_finite_timeout(lifetime))
825 rt6_clean_expires(rt);
826 else
827 rt6_set_expires(rt, jiffies + HZ * lifetime);
828
94e187c0 829 ip6_rt_put(rt);
70ceb4f5
YH		 830 }
831 return 0;
832}
833#endif
834
a3c00e46
MKL		 835static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
836 struct in6_addr *saddr)
837{
838 struct fib6_node *pn;
839 while (1) {
840 if (fn->fn_flags & RTN_TL_ROOT)
841 return NULL;
842 pn = fn->parent;
843 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
844 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
845 else
846 fn = pn;
847 if (fn->fn_flags & RTN_RTINFO)
848 return fn;
849 }
850}
c71099ac 851
8ed67789
DL		 852static struct rt6_info *ip6_pol_route_lookup(struct net *net,
853 struct fib6_table *table,
4c9483b2 854 struct flowi6 *fl6, int flags)
1da177e4
LT		 855{
856 struct fib6_node *fn;
857 struct rt6_info *rt;
858
c71099ac 859 read_lock_bh(&table->tb6_lock);
4c9483b2 860 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 861restart:
862 rt = fn->leaf;
4c9483b2 863 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 864 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 865 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 866 if (rt == net->ipv6.ip6_null_entry) {
867 fn = fib6_backtrack(fn, &fl6->saddr);
868 if (fn)
869 goto restart;
870 }
d8d1f30b 871 dst_use(&rt->dst, jiffies);
c71099ac 872 read_unlock_bh(&table->tb6_lock);
b811580d
DA		 873
874 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
875
c71099ac
TG		 876 return rt;
877
878}
879
67ba4152 880struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 881 int flags)
882{
883 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
884}
885EXPORT_SYMBOL_GPL(ip6_route_lookup);
886
9acd9f3a
YH		 887struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
888 const struct in6_addr *saddr, int oif, int strict)
c71099ac 889{
4c9483b2
DM		 890 struct flowi6 fl6 = {
891 .flowi6_oif = oif,
892 .daddr = *daddr,
c71099ac
TG		 893 };
894 struct dst_entry *dst;
77d16f45 895 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 896
adaa70bb 897 if (saddr) {
4c9483b2 898 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 899 flags |= RT6_LOOKUP_F_HAS_SADDR;
900 }
901
4c9483b2 902 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 903 if (dst->error == 0)
904 return (struct rt6_info *) dst;
905
906 dst_release(dst);
907
1da177e4
LT		 908 return NULL;
909}
7159039a
YH		 910EXPORT_SYMBOL(rt6_lookup);
911
c71099ac 912/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 913 It takes new route entry, the addition fails by any reason the
914 route is freed. In any case, if caller does not hold it, it may
915 be destroyed.
916 */
917
e5fd387a 918static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 919 struct mx6_config *mxc)
1da177e4
LT		 920{
921 int err;
c71099ac 922 struct fib6_table *table;
1da177e4 923
c71099ac
TG		 924 table = rt->rt6i_table;
925 write_lock_bh(&table->tb6_lock);
e715b6d3 926 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 927 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 928
929 return err;
930}
931
40e22e8f
TG		 932int ip6_ins_rt(struct rt6_info *rt)
933{
e715b6d3
FW		 934 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
935 struct mx6_config mxc = { .mx = NULL, };
936
937 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 938}
939
8b9df265
MKL		 940static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
941 const struct in6_addr *daddr,
942 const struct in6_addr *saddr)
1da177e4 943{
1da177e4
LT		 944 struct rt6_info *rt;
945
946 /*
947 * Clone the route.
948 */
949
d52d3997 950 if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
83a09abd 951 ort = (struct rt6_info *)ort->dst.from;
1da177e4 952
ad706862 953 rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
83a09abd
MKL		 954
955 if (!rt)
956 return NULL;
957
958 ip6_rt_copy_init(rt, ort);
959 rt->rt6i_flags |= RTF_CACHE;
960 rt->rt6i_metric = 0;
961 rt->dst.flags |= DST_HOST;
962 rt->rt6i_dst.addr = *daddr;
963 rt->rt6i_dst.plen = 128;
1da177e4 964
83a09abd
MKL		 965 if (!rt6_is_gw_or_nonexthop(ort)) {
966 if (ort->rt6i_dst.plen != 128 &&
967 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
968 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 969#ifdef CONFIG_IPV6_SUBTREES
83a09abd
MKL		 970 if (rt->rt6i_src.plen && saddr) {
971 rt->rt6i_src.addr = *saddr;
972 rt->rt6i_src.plen = 128;
8b9df265 973 }
83a09abd 974#endif
95a9a5ba 975 }
1da177e4 976
95a9a5ba
YH		 977 return rt;
978}
1da177e4 979
d52d3997
MKL		 980static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
981{
982 struct rt6_info *pcpu_rt;
983
984 pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
ad706862 985 rt->dst.dev, rt->dst.flags);
d52d3997
MKL		 986
987 if (!pcpu_rt)
988 return NULL;
989 ip6_rt_copy_init(pcpu_rt, rt);
990 pcpu_rt->rt6i_protocol = rt->rt6i_protocol;
991 pcpu_rt->rt6i_flags |= RTF_PCPU;
992 return pcpu_rt;
993}
994
995/* It should be called with read_lock_bh(&tb6_lock) acquired */
996static struct rt6_info *rt6_get_pcpu_route(struct rt6_info *rt)
997{
a73e4195 998 struct rt6_info *pcpu_rt, **p;
d52d3997
MKL		 999
1000 p = this_cpu_ptr(rt->rt6i_pcpu);
1001 pcpu_rt = *p;
1002
a73e4195
MKL		 1003 if (pcpu_rt) {
1004 dst_hold(&pcpu_rt->dst);
1005 rt6_dst_from_metrics_check(pcpu_rt);
1006 }
1007 return pcpu_rt;
1008}
1009
1010static struct rt6_info *rt6_make_pcpu_route(struct rt6_info *rt)
1011{
9c7370a1 1012 struct fib6_table *table = rt->rt6i_table;
a73e4195 1013 struct rt6_info *pcpu_rt, *prev, **p;
d52d3997
MKL		 1014
1015 pcpu_rt = ip6_rt_pcpu_alloc(rt);
1016 if (!pcpu_rt) {
1017 struct net *net = dev_net(rt->dst.dev);
1018
9c7370a1
MKL		 1019 dst_hold(&net->ipv6.ip6_null_entry->dst);
1020 return net->ipv6.ip6_null_entry;
d52d3997
MKL		 1021 }
1022
9c7370a1
MKL		 1023 read_lock_bh(&table->tb6_lock);
1024 if (rt->rt6i_pcpu) {
1025 p = this_cpu_ptr(rt->rt6i_pcpu);
1026 prev = cmpxchg(p, NULL, pcpu_rt);
1027 if (prev) {
1028 /* If someone did it before us, return prev instead */
1029 dst_destroy(&pcpu_rt->dst);
1030 pcpu_rt = prev;
1031 }
1032 } else {
1033 /* rt has been removed from the fib6 tree
1034 * before we have a chance to acquire the read_lock.
1035 * In this case, don't brother to create a pcpu rt
1036 * since rt is going away anyway. The next
1037 * dst_check() will trigger a re-lookup.
1038 */
d52d3997 1039 dst_destroy(&pcpu_rt->dst);
9c7370a1 1040 pcpu_rt = rt;
d52d3997 1041 }
d52d3997
MKL		 1042 dst_hold(&pcpu_rt->dst);
1043 rt6_dst_from_metrics_check(pcpu_rt);
9c7370a1 1044 read_unlock_bh(&table->tb6_lock);
d52d3997
MKL		 1045 return pcpu_rt;
1046}
1047
9ff74384
DA		 1048struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
1049 int oif, struct flowi6 *fl6, int flags)
1da177e4 1050{
367efcb9 1051 struct fib6_node *fn, *saved_fn;
45e4fd26 1052 struct rt6_info *rt;
c71099ac 1053 int strict = 0;
1da177e4 1054
77d16f45 1055 strict |= flags & RT6_LOOKUP_F_IFACE;
d5d32e4b 1056 strict |= flags & RT6_LOOKUP_F_IGNORE_LINKSTATE;
367efcb9
MKL		 1057 if (net->ipv6.devconf_all->forwarding == 0)
1058 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 1059
c71099ac 1060 read_lock_bh(&table->tb6_lock);
1da177e4 1061
4c9483b2 1062 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 1063 saved_fn = fn;
1da177e4 1064
ca254490
DA		 1065 if (fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF)
1066 oif = 0;
1067
a3c00e46 1068redo_rt6_select:
367efcb9 1069 rt = rt6_select(fn, oif, strict);
52bd4c0c 1070 if (rt->rt6i_nsiblings)
367efcb9 1071 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 1072 if (rt == net->ipv6.ip6_null_entry) {
1073 fn = fib6_backtrack(fn, &fl6->saddr);
1074 if (fn)
1075 goto redo_rt6_select;
367efcb9
MKL		 1076 else if (strict & RT6_LOOKUP_F_REACHABLE) {
1077 /* also consider unreachable route */
1078 strict &= ~RT6_LOOKUP_F_REACHABLE;
1079 fn = saved_fn;
1080 goto redo_rt6_select;
367efcb9 1081 }
a3c00e46
MKL		 1082 }
1083
fb9de91e 1084
3da59bd9 1085 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
d52d3997
MKL		 1086 dst_use(&rt->dst, jiffies);
1087 read_unlock_bh(&table->tb6_lock);
1088
1089 rt6_dst_from_metrics_check(rt);
b811580d
DA		 1090
1091 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
d52d3997 1092 return rt;
3da59bd9
MKL		 1093 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
1094 !(rt->rt6i_flags & RTF_GATEWAY))) {
1095 /* Create a RTF_CACHE clone which will not be
1096 * owned by the fib6 tree. It is for the special case where
1097 * the daddr in the skb during the neighbor look-up is different
1098 * from the fl6->daddr used to look-up route here.
1099 */
1100
1101 struct rt6_info *uncached_rt;
1102
d52d3997
MKL		 1103 dst_use(&rt->dst, jiffies);
1104 read_unlock_bh(&table->tb6_lock);
1105
3da59bd9
MKL		 1106 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
1107 dst_release(&rt->dst);
c71099ac 1108
3da59bd9 1109 if (uncached_rt)
8d0b94af 1110 rt6_uncached_list_add(uncached_rt);
3da59bd9
MKL		 1111 else
1112 uncached_rt = net->ipv6.ip6_null_entry;
d52d3997 1113
3da59bd9 1114 dst_hold(&uncached_rt->dst);
b811580d
DA		 1115
1116 trace_fib6_table_lookup(net, uncached_rt, table->tb6_id, fl6);
3da59bd9 1117 return uncached_rt;
3da59bd9 1118
d52d3997
MKL		 1119 } else {
1120 /* Get a percpu copy */
1121
1122 struct rt6_info *pcpu_rt;
1123
1124 rt->dst.lastuse = jiffies;
1125 rt->dst.__use++;
1126 pcpu_rt = rt6_get_pcpu_route(rt);
d52d3997 1127
9c7370a1
MKL		 1128 if (pcpu_rt) {
1129 read_unlock_bh(&table->tb6_lock);
1130 } else {
1131 /* We have to do the read_unlock first
1132 * because rt6_make_pcpu_route() may trigger
1133 * ip6_dst_gc() which will take the write_lock.
1134 */
1135 dst_hold(&rt->dst);
1136 read_unlock_bh(&table->tb6_lock);
a73e4195 1137 pcpu_rt = rt6_make_pcpu_route(rt);
9c7370a1
MKL		 1138 dst_release(&rt->dst);
1139 }
d52d3997 1140
b811580d 1141 trace_fib6_table_lookup(net, pcpu_rt, table->tb6_id, fl6);
d52d3997 1142 return pcpu_rt;
9c7370a1 1143
d52d3997 1144 }
1da177e4 1145}
9ff74384 1146EXPORT_SYMBOL_GPL(ip6_pol_route);
1da177e4 1147
8ed67789 1148static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 1149 struct flowi6 *fl6, int flags)
4acad72d 1150{
4c9483b2 1151 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 1152}
1153
d409b847
MB		 1154struct dst_entry *ip6_route_input_lookup(struct net *net,
1155 struct net_device *dev,
1156 struct flowi6 *fl6, int flags)
72331bc0
SL		 1157{
1158 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1159 flags |= RT6_LOOKUP_F_IFACE;
1160
1161 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1162}
d409b847 1163EXPORT_SYMBOL_GPL(ip6_route_input_lookup);
72331bc0 1164
c71099ac
TG		 1165void ip6_route_input(struct sk_buff *skb)
1166{
b71d1d42 1167 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1168 struct net *net = dev_net(skb->dev);
adaa70bb 1169 int flags = RT6_LOOKUP_F_HAS_SADDR;
904af04d 1170 struct ip_tunnel_info *tun_info;
4c9483b2 1171 struct flowi6 fl6 = {
e0d56fdd 1172 .flowi6_iif = skb->dev->ifindex,
4c9483b2
DM		 1173 .daddr = iph->daddr,
1174 .saddr = iph->saddr,
6502ca52 1175 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1176 .flowi6_mark = skb->mark,
1177 .flowi6_proto = iph->nexthdr,
c71099ac 1178 };
adaa70bb 1179
904af04d 1180 tun_info = skb_tunnel_info(skb);
46fa062a 1181 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
904af04d 1182 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
06e9d040 1183 skb_dst_drop(skb);
72331bc0 1184 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1185}
1186
8ed67789 1187static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1188 struct flowi6 *fl6, int flags)
1da177e4 1189{
4c9483b2 1190 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1191}
1192
6f21c96a
PA		 1193struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
1194 struct flowi6 *fl6, int flags)
c71099ac 1195{
d46a9d67 1196 bool any_src;
c71099ac 1197
4c1feac5
DA		 1198 if (rt6_need_strict(&fl6->daddr)) {
1199 struct dst_entry *dst;
1200
1201 dst = l3mdev_link_scope_lookup(net, fl6);
1202 if (dst)
1203 return dst;
1204 }
ca254490 1205
1fb9489b 1206 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1207
d46a9d67 1208 any_src = ipv6_addr_any(&fl6->saddr);
741a11d9 1209 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
d46a9d67 1210 (fl6->flowi6_oif && any_src))
77d16f45 1211 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1212
d46a9d67 1213 if (!any_src)
adaa70bb 1214 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1215 else if (sk)
1216 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1217
4c9483b2 1218 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 1219}
6f21c96a 1220EXPORT_SYMBOL_GPL(ip6_route_output_flags);
1da177e4 1221
2774c131 1222struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1223{
5c1e6aa3 1224 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1225 struct dst_entry *new = NULL;
1226
f5b0a874 1227 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1228 if (rt) {
0a1f5962 1229 rt6_info_init(rt);
8104891b 1230
0a1f5962 1231 new = &rt->dst;
14e50e57 1232 new->__use = 1;
352e512c 1233 new->input = dst_discard;
ede2059d 1234 new->output = dst_discard_out;
14e50e57 1235
0a1f5962 1236 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1237 rt->rt6i_idev = ort->rt6i_idev;
1238 if (rt->rt6i_idev)
1239 in6_dev_hold(rt->rt6i_idev);
14e50e57 1240
4e3fd7a0 1241 rt->rt6i_gateway = ort->rt6i_gateway;
0a1f5962 1242 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
14e50e57
DM		 1243 rt->rt6i_metric = 0;
1244
1245 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1246#ifdef CONFIG_IPV6_SUBTREES
1247 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1248#endif
1249
1250 dst_free(new);
1251 }
1252
69ead7af
DM		 1253 dst_release(dst_orig);
1254 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1255}
14e50e57 1256
1da177e4
LT		 1257/*
1258 * Destination cache support functions
1259 */
1260
4b32b5ad
MKL		 1261static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1262{
1263 if (rt->dst.from &&
1264 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1265 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1266}
1267
3da59bd9
MKL		 1268static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1269{
1270 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1271 return NULL;
1272
1273 if (rt6_check_expired(rt))
1274 return NULL;
1275
1276 return &rt->dst;
1277}
1278
1279static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1280{
5973fb1e
MKL		 1281 if (!__rt6_check_expired(rt) &&
1282 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
3da59bd9
MKL		 1283 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1284 return &rt->dst;
1285 else
1286 return NULL;
1287}
1288
1da177e4
LT		 1289static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1290{
1291 struct rt6_info *rt;
1292
1293 rt = (struct rt6_info *) dst;
1294
6f3118b5
ND		 1295 /* All IPV6 dsts are created with ->obsolete set to the value
1296 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1297 * into this function always.
1298 */
e3bc10bd 1299
4b32b5ad
MKL		 1300 rt6_dst_from_metrics_check(rt);
1301
02bcf4e0
MKL		 1302 if (rt->rt6i_flags & RTF_PCPU ||
1303 (unlikely(dst->flags & DST_NOCACHE) && rt->dst.from))
3da59bd9
MKL		 1304 return rt6_dst_from_check(rt, cookie);
1305 else
1306 return rt6_check(rt, cookie);
1da177e4
LT		 1307}
1308
1309static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1310{
1311 struct rt6_info *rt = (struct rt6_info *) dst;
1312
1313 if (rt) {
54c1a859
YH		 1314 if (rt->rt6i_flags & RTF_CACHE) {
1315 if (rt6_check_expired(rt)) {
1316 ip6_del_rt(rt);
1317 dst = NULL;
1318 }
1319 } else {
1da177e4 1320 dst_release(dst);
54c1a859
YH		 1321 dst = NULL;
1322 }
1da177e4 1323 }
54c1a859 1324 return dst;
1da177e4
LT		 1325}
1326
1327static void ip6_link_failure(struct sk_buff *skb)
1328{
1329 struct rt6_info *rt;
1330
3ffe533c 1331 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1332
adf30907 1333 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1334 if (rt) {
1eb4f758
HFS		 1335 if (rt->rt6i_flags & RTF_CACHE) {
1336 dst_hold(&rt->dst);
8e3d5be7 1337 ip6_del_rt(rt);
1eb4f758 1338 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1339 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1340 }
1da177e4
LT		 1341 }
1342}
1343
45e4fd26
MKL		 1344static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1345{
1346 struct net *net = dev_net(rt->dst.dev);
1347
1348 rt->rt6i_flags |= RTF_MODIFIED;
1349 rt->rt6i_pmtu = mtu;
1350 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1351}
1352
0d3f6d29
MKL		 1353static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1354{
1355 return !(rt->rt6i_flags & RTF_CACHE) &&
1356 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1357}
1358
45e4fd26
MKL		 1359static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1360 const struct ipv6hdr *iph, u32 mtu)
1da177e4 1361{
67ba4152 1362 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1363
45e4fd26
MKL		 1364 if (rt6->rt6i_flags & RTF_LOCAL)
1365 return;
81aded24 1366
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
1467 * check if the redirect has come from approriate router.
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
86872cb5 1834 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
8c5b83f0 1835 goto out;
1da177e4 1836#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1837 if (cfg->fc_src_len)
8c5b83f0 1838 goto out;
1da177e4 1839#endif
86872cb5 1840 if (cfg->fc_ifindex) {
1da177e4 1841 err = -ENODEV;
5578689a 1842 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1843 if (!dev)
1844 goto out;
1845 idev = in6_dev_get(dev);
1846 if (!idev)
1847 goto out;
1848 }
1849
86872cb5
TG		 1850 if (cfg->fc_metric == 0)
1851 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1852
d71314b4 1853 err = -ENOBUFS;
38308473
DM		 1854 if (cfg->fc_nlinfo.nlh &&
1855 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1856 table = fib6_get_table(net, cfg->fc_table);
38308473 1857 if (!table) {
f3213831 1858 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1859 table = fib6_new_table(net, cfg->fc_table);
1860 }
1861 } else {
1862 table = fib6_new_table(net, cfg->fc_table);
1863 }
38308473
DM		 1864
1865 if (!table)
c71099ac 1866 goto out;
c71099ac 1867
ad706862
MKL		 1868 rt = ip6_dst_alloc(net, NULL,
1869 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
1da177e4 1870
38308473 1871 if (!rt) {
1da177e4
LT		 1872 err = -ENOMEM;
1873 goto out;
1874 }
1875
1716a961
G		 1876 if (cfg->fc_flags & RTF_EXPIRES)
1877 rt6_set_expires(rt, jiffies +
1878 clock_t_to_jiffies(cfg->fc_expires));
1879 else
1880 rt6_clean_expires(rt);
1da177e4 1881
86872cb5
TG		 1882 if (cfg->fc_protocol == RTPROT_UNSPEC)
1883 cfg->fc_protocol = RTPROT_BOOT;
1884 rt->rt6i_protocol = cfg->fc_protocol;
1885
1886 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1887
1888 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1889 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1890 else if (cfg->fc_flags & RTF_LOCAL)
1891 rt->dst.input = ip6_input;
1da177e4 1892 else
d8d1f30b 1893 rt->dst.input = ip6_forward;
1da177e4 1894
d8d1f30b 1895 rt->dst.output = ip6_output;
1da177e4 1896
19e42e45
RP		 1897 if (cfg->fc_encap) {
1898 struct lwtunnel_state *lwtstate;
1899
1900 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
127eb7cd
TH		 1901 cfg->fc_encap, AF_INET6, cfg,
1902 &lwtstate);
19e42e45
RP		 1903 if (err)
1904 goto out;
61adedf3
JB		 1905 rt->dst.lwtstate = lwtstate_get(lwtstate);
1906 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1907 rt->dst.lwtstate->orig_output = rt->dst.output;
1908 rt->dst.output = lwtunnel_output;
25368623 1909 }
61adedf3
JB		 1910 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1911 rt->dst.lwtstate->orig_input = rt->dst.input;
1912 rt->dst.input = lwtunnel_input;
25368623 1913 }
19e42e45
RP		 1914 }
1915
86872cb5
TG		 1916 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1917 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1918 if (rt->rt6i_dst.plen == 128)
e5fd387a 1919 rt->dst.flags |= DST_HOST;
e5fd387a 1920
1da177e4 1921#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1922 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1923 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1924#endif
1925
86872cb5 1926 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1927
1928 /* We cannot add true routes via loopback here,
1929 they would result in kernel looping; promote them to reject routes
1930 */
86872cb5 1931 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1932 (dev && (dev->flags & IFF_LOOPBACK) &&
1933 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1934 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1935 /* hold loopback dev/idev if we haven't done so. */
5578689a 1936 if (dev != net->loopback_dev) {
1da177e4
LT		 1937 if (dev) {
1938 dev_put(dev);
1939 in6_dev_put(idev);
1940 }
5578689a 1941 dev = net->loopback_dev;
1da177e4
LT		 1942 dev_hold(dev);
1943 idev = in6_dev_get(dev);
1944 if (!idev) {
1945 err = -ENODEV;
1946 goto out;
1947 }
1948 }
1da177e4 1949 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1950 switch (cfg->fc_type) {
1951 case RTN_BLACKHOLE:
1952 rt->dst.error = -EINVAL;
ede2059d 1953 rt->dst.output = dst_discard_out;
7150aede 1954 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1955 break;
1956 case RTN_PROHIBIT:
1957 rt->dst.error = -EACCES;
7150aede
K		 1958 rt->dst.output = ip6_pkt_prohibit_out;
1959 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1960 break;
b4949ab2 1961 case RTN_THROW:
0315e382 1962 case RTN_UNREACHABLE:
ef2c7d7b 1963 default:
7150aede 1964 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
0315e382
NF		 1965 : (cfg->fc_type == RTN_UNREACHABLE)
1966 ? -EHOSTUNREACH : -ENETUNREACH;
7150aede
K		 1967 rt->dst.output = ip6_pkt_discard_out;
1968 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1969 break;
1970 }
1da177e4
LT		 1971 goto install_route;
1972 }
1973
86872cb5 1974 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1975 const struct in6_addr *gw_addr;
1da177e4
LT		 1976 int gwa_type;
1977
86872cb5 1978 gw_addr = &cfg->fc_gateway;
330567b7 1979 gwa_type = ipv6_addr_type(gw_addr);
48ed7b26
FW		 1980
1981 /* if gw_addr is local we will fail to detect this in case
1982 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1983 * will return already-added prefix route via interface that
1984 * prefix route was assigned to, which might be non-loopback.
1985 */
1986 err = -EINVAL;
330567b7
FW		 1987 if (ipv6_chk_addr_and_flags(net, gw_addr,
1988 gwa_type & IPV6_ADDR_LINKLOCAL ?
1989 dev : NULL, 0, 0))
48ed7b26
FW		 1990 goto out;
1991
4e3fd7a0 1992 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1993
1994 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
8c14586f 1995 struct rt6_info *grt = NULL;
1da177e4
LT		 1996
1997 /* IPv6 strictly inhibits using not link-local
1998 addresses as nexthop address.
1999 Otherwise, router will not able to send redirects.
2000 It is very good, but in some (rare!) circumstances
2001 (SIT, PtP, NBMA NOARP links) it is handy to allow
2002 some exceptions. --ANK
2003 */
38308473 2004 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 2005 goto out;
2006
a435a07f 2007 if (cfg->fc_table) {
8c14586f
DA		 2008 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
2009
a435a07f
VB		 2010 if (grt) {
2011 if (grt->rt6i_flags & RTF_GATEWAY ||
2012 (dev && dev != grt->dst.dev)) {
2013 ip6_rt_put(grt);
2014 grt = NULL;
2015 }
2016 }
2017 }
2018
8c14586f
DA		 2019 if (!grt)
2020 grt = rt6_lookup(net, gw_addr, NULL,
2021 cfg->fc_ifindex, 1);
1da177e4
LT		 2022
2023 err = -EHOSTUNREACH;
38308473 2024 if (!grt)
1da177e4
LT		 2025 goto out;
2026 if (dev) {
d1918542 2027 if (dev != grt->dst.dev) {
94e187c0 2028 ip6_rt_put(grt);
1da177e4
LT		 2029 goto out;
2030 }
2031 } else {
d1918542 2032 dev = grt->dst.dev;
1da177e4
LT		 2033 idev = grt->rt6i_idev;
2034 dev_hold(dev);
2035 in6_dev_hold(grt->rt6i_idev);
2036 }
38308473 2037 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 2038 err = 0;
94e187c0 2039 ip6_rt_put(grt);
1da177e4
LT		 2040
2041 if (err)
2042 goto out;
2043 }
2044 err = -EINVAL;
38308473 2045 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2046 goto out;
2047 }
2048
2049 err = -ENODEV;
38308473 2050 if (!dev)
1da177e4
LT		 2051 goto out;
2052
c3968a85
DW		 2053 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2054 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2055 err = -EINVAL;
2056 goto out;
2057 }
4e3fd7a0 2058 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 2059 rt->rt6i_prefsrc.plen = 128;
2060 } else
2061 rt->rt6i_prefsrc.plen = 0;
2062
86872cb5 2063 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 2064
2065install_route:
d8d1f30b 2066 rt->dst.dev = dev;
1da177e4 2067 rt->rt6i_idev = idev;
c71099ac 2068 rt->rt6i_table = table;
63152fc0 2069
c346dca1 2070 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 2071
8c5b83f0 2072 return rt;
6b9ea5a6
RP		 2073out:
2074 if (dev)
2075 dev_put(dev);
2076 if (idev)
2077 in6_dev_put(idev);
2078 if (rt)
2079 dst_free(&rt->dst);
2080
8c5b83f0 2081 return ERR_PTR(err);
6b9ea5a6
RP		 2082}
2083
2084int ip6_route_add(struct fib6_config *cfg)
2085{
2086 struct mx6_config mxc = { .mx = NULL, };
8c5b83f0 2087 struct rt6_info *rt;
6b9ea5a6
RP		 2088 int err;
2089
8c5b83f0
RP		 2090 rt = ip6_route_info_create(cfg);
2091 if (IS_ERR(rt)) {
2092 err = PTR_ERR(rt);
2093 rt = NULL;
6b9ea5a6 2094 goto out;
8c5b83f0 2095 }
6b9ea5a6 2096
e715b6d3
FW		 2097 err = ip6_convert_metrics(&mxc, cfg);
2098 if (err)
2099 goto out;
1da177e4 2100
e715b6d3
FW		 2101 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2102
2103 kfree(mxc.mx);
6b9ea5a6 2104
e715b6d3 2105 return err;
1da177e4 2106out:
1da177e4 2107 if (rt)
d8d1f30b 2108 dst_free(&rt->dst);
6b9ea5a6 2109
1da177e4
LT		 2110 return err;
2111}
2112
86872cb5 2113static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2114{
2115 int err;
c71099ac 2116 struct fib6_table *table;
d1918542 2117 struct net *net = dev_net(rt->dst.dev);
1da177e4 2118
8e3d5be7
MKL		 2119 if (rt == net->ipv6.ip6_null_entry ||
2120 rt->dst.flags & DST_NOCACHE) {
6825a26c
G		 2121 err = -ENOENT;
2122 goto out;
2123 }
6c813a72 2124
c71099ac
TG		 2125 table = rt->rt6i_table;
2126 write_lock_bh(&table->tb6_lock);
86872cb5 2127 err = fib6_del(rt, info);
c71099ac 2128 write_unlock_bh(&table->tb6_lock);
1da177e4 2129
6825a26c 2130out:
94e187c0 2131 ip6_rt_put(rt);
1da177e4
LT		 2132 return err;
2133}
2134
e0a1ad73
TG		 2135int ip6_del_rt(struct rt6_info *rt)
2136{
4d1169c1 2137 struct nl_info info = {
d1918542 2138 .nl_net = dev_net(rt->dst.dev),
4d1169c1 2139 };
528c4ceb 2140 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 2141}
2142
86872cb5 2143static int ip6_route_del(struct fib6_config *cfg)
1da177e4 2144{
c71099ac 2145 struct fib6_table *table;
1da177e4
LT		 2146 struct fib6_node *fn;
2147 struct rt6_info *rt;
2148 int err = -ESRCH;
2149
5578689a 2150 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 2151 if (!table)
c71099ac
TG		 2152 return err;
2153
2154 read_lock_bh(&table->tb6_lock);
1da177e4 2155
c71099ac 2156 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 2157 &cfg->fc_dst, cfg->fc_dst_len,
2158 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 2159
1da177e4 2160 if (fn) {
d8d1f30b 2161 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 2162 if ((rt->rt6i_flags & RTF_CACHE) &&
2163 !(cfg->fc_flags & RTF_CACHE))
2164 continue;
86872cb5 2165 if (cfg->fc_ifindex &&
d1918542
DM		 2166 (!rt->dst.dev ||
2167 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 2168 continue;
86872cb5
TG		 2169 if (cfg->fc_flags & RTF_GATEWAY &&
2170 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 2171 continue;
86872cb5 2172 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 2173 continue;
d8d1f30b 2174 dst_hold(&rt->dst);
c71099ac 2175 read_unlock_bh(&table->tb6_lock);
1da177e4 2176
86872cb5 2177 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 2178 }
2179 }
c71099ac 2180 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2181
2182 return err;
2183}
2184
6700c270 2185static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 2186{
a6279458 2187 struct netevent_redirect netevent;
e8599ff4 2188 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 2189 struct ndisc_options ndopts;
2190 struct inet6_dev *in6_dev;
2191 struct neighbour *neigh;
71bcdba0 2192 struct rd_msg *msg;
6e157b6a
DM		 2193 int optlen, on_link;
2194 u8 *lladdr;
e8599ff4 2195
29a3cad5 2196 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 2197 optlen -= sizeof(*msg);
e8599ff4
DM		 2198
2199 if (optlen < 0) {
6e157b6a 2200 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 2201 return;
2202 }
2203
71bcdba0 2204 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 2205
71bcdba0 2206 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 2207 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 2208 return;
2209 }
2210
6e157b6a 2211 on_link = 0;
71bcdba0 2212 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 2213 on_link = 1;
71bcdba0 2214 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 2215 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 2216 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 2217 return;
2218 }
2219
2220 in6_dev = __in6_dev_get(skb->dev);
2221 if (!in6_dev)
2222 return;
2223 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2224 return;
2225
2226 /* RFC2461 8.1:
2227 * The IP source address of the Redirect MUST be the same as the current
2228 * first-hop router for the specified ICMP Destination Address.
2229 */
2230
f997c55c 2231 if (!ndisc_parse_options(skb->dev, msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 2232 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2233 return;
2234 }
6e157b6a
DM		 2235
2236 lladdr = NULL;
e8599ff4
DM		 2237 if (ndopts.nd_opts_tgt_lladdr) {
2238 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2239 skb->dev);
2240 if (!lladdr) {
2241 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2242 return;
2243 }
2244 }
2245
6e157b6a 2246 rt = (struct rt6_info *) dst;
ec13ad1d 2247 if (rt->rt6i_flags & RTF_REJECT) {
6e157b6a 2248 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 2249 return;
6e157b6a 2250 }
e8599ff4 2251
6e157b6a
DM		 2252 /* Redirect received -> path was valid.
2253 * Look, redirects are sent only in response to data packets,
2254 * so that this nexthop apparently is reachable. --ANK
2255 */
2256 dst_confirm(&rt->dst);
a6279458 2257
71bcdba0 2258 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 2259 if (!neigh)
2260 return;
a6279458 2261
1da177e4
LT		 2262 /*
2263 * We have finally decided to accept it.
2264 */
2265
f997c55c 2266 ndisc_update(skb->dev, neigh, lladdr, NUD_STALE,
1da177e4
LT		 2267 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2268 NEIGH_UPDATE_F_OVERRIDE|
2269 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
f997c55c
AA		 2270 NEIGH_UPDATE_F_ISROUTER)),
2271 NDISC_REDIRECT, &ndopts);
1da177e4 2272
83a09abd 2273 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
38308473 2274 if (!nrt)
1da177e4
LT		 2275 goto out;
2276
2277 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2278 if (on_link)
2279 nrt->rt6i_flags &= ~RTF_GATEWAY;
2280
4e3fd7a0 2281 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 2282
40e22e8f 2283 if (ip6_ins_rt(nrt))
1da177e4
LT		 2284 goto out;
2285
d8d1f30b
CG		 2286 netevent.old = &rt->dst;
2287 netevent.new = &nrt->dst;
71bcdba0 2288 netevent.daddr = &msg->dest;
60592833 2289 netevent.neigh = neigh;
8d71740c
TT		 2290 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2291
38308473 2292 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 2293 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 2294 ip6_del_rt(rt);
1da177e4
LT		 2295 }
2296
2297out:
e8599ff4 2298 neigh_release(neigh);
6e157b6a
DM		 2299}
2300
1da177e4
LT		 2301/*
2302 * Misc support functions
2303 */
2304
4b32b5ad
MKL		 2305static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2306{
2307 BUG_ON(from->dst.from);
2308
2309 rt->rt6i_flags &= ~RTF_EXPIRES;
2310 dst_hold(&from->dst);
2311 rt->dst.from = &from->dst;
2312 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2313}
2314
83a09abd
MKL		 2315static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
2316{
2317 rt->dst.input = ort->dst.input;
2318 rt->dst.output = ort->dst.output;
2319 rt->rt6i_dst = ort->rt6i_dst;
2320 rt->dst.error = ort->dst.error;
2321 rt->rt6i_idev = ort->rt6i_idev;
2322 if (rt->rt6i_idev)
2323 in6_dev_hold(rt->rt6i_idev);
2324 rt->dst.lastuse = jiffies;
2325 rt->rt6i_gateway = ort->rt6i_gateway;
2326 rt->rt6i_flags = ort->rt6i_flags;
2327 rt6_set_from(rt, ort);
2328 rt->rt6i_metric = ort->rt6i_metric;
1da177e4 2329#ifdef CONFIG_IPV6_SUBTREES
83a09abd 2330 rt->rt6i_src = ort->rt6i_src;
1da177e4 2331#endif
83a09abd
MKL		 2332 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2333 rt->rt6i_table = ort->rt6i_table;
61adedf3 2334 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
1da177e4
LT		 2335}
2336
70ceb4f5 2337#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 2338static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42 2339 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2340 const struct in6_addr *gwaddr,
2341 struct net_device *dev)
70ceb4f5 2342{
830218c1
DA		 2343 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO;
2344 int ifindex = dev->ifindex;
70ceb4f5
YH		 2345 struct fib6_node *fn;
2346 struct rt6_info *rt = NULL;
c71099ac
TG		 2347 struct fib6_table *table;
2348
830218c1 2349 table = fib6_get_table(net, tb_id);
38308473 2350 if (!table)
c71099ac 2351 return NULL;
70ceb4f5 2352
5744dd9b 2353 read_lock_bh(&table->tb6_lock);
67ba4152 2354 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2355 if (!fn)
2356 goto out;
2357
d8d1f30b 2358 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2359 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2360 continue;
2361 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2362 continue;
2363 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2364 continue;
d8d1f30b 2365 dst_hold(&rt->dst);
70ceb4f5
YH		 2366 break;
2367 }
2368out:
5744dd9b 2369 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2370 return rt;
2371}
2372
efa2cea0 2373static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42 2374 const struct in6_addr *prefix, int prefixlen,
830218c1
DA		 2375 const struct in6_addr *gwaddr,
2376 struct net_device *dev,
95c96174 2377 unsigned int pref)
70ceb4f5 2378{
86872cb5 2379 struct fib6_config cfg = {
238fc7ea 2380 .fc_metric = IP6_RT_PRIO_USER,
830218c1 2381 .fc_ifindex = dev->ifindex,
86872cb5
TG		 2382 .fc_dst_len = prefixlen,
2383 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2384 RTF_UP | RTF_PREF(pref),
15e47304 2385 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2386 .fc_nlinfo.nlh = NULL,
2387 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2388 };
2389
830218c1 2390 cfg.fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_INFO,
4e3fd7a0
AD		 2391 cfg.fc_dst = *prefix;
2392 cfg.fc_gateway = *gwaddr;
70ceb4f5 2393
e317da96
YH		 2394 /* We should treat it as a default route if prefix length is 0. */
2395 if (!prefixlen)
86872cb5 2396 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2397
86872cb5 2398 ip6_route_add(&cfg);
70ceb4f5 2399
830218c1 2400 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, dev);
70ceb4f5
YH		 2401}
2402#endif
2403
b71d1d42 2404struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2405{
830218c1 2406 u32 tb_id = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT;
1da177e4 2407 struct rt6_info *rt;
c71099ac 2408 struct fib6_table *table;
1da177e4 2409
830218c1 2410 table = fib6_get_table(dev_net(dev), tb_id);
38308473 2411 if (!table)
c71099ac 2412 return NULL;
1da177e4 2413
5744dd9b 2414 read_lock_bh(&table->tb6_lock);
67ba4152 2415 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2416 if (dev == rt->dst.dev &&
045927ff 2417 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2418 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2419 break;
2420 }
2421 if (rt)
d8d1f30b 2422 dst_hold(&rt->dst);
5744dd9b 2423 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2424 return rt;
2425}
2426
b71d1d42 2427struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2428 struct net_device *dev,
2429 unsigned int pref)
1da177e4 2430{
86872cb5 2431 struct fib6_config cfg = {
ca254490 2432 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
238fc7ea 2433 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2434 .fc_ifindex = dev->ifindex,
2435 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2436 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2437 .fc_nlinfo.portid = 0,
5578689a 2438 .fc_nlinfo.nlh = NULL,
c346dca1 2439 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2440 };
1da177e4 2441
4e3fd7a0 2442 cfg.fc_gateway = *gwaddr;
1da177e4 2443
830218c1
DA		 2444 if (!ip6_route_add(&cfg)) {
2445 struct fib6_table *table;
2446
2447 table = fib6_get_table(dev_net(dev), cfg.fc_table);
2448 if (table)
2449 table->flags |= RT6_TABLE_HAS_DFLT_ROUTER;
2450 }
1da177e4 2451
1da177e4
LT		 2452 return rt6_get_dflt_router(gwaddr, dev);
2453}
2454
830218c1 2455static void __rt6_purge_dflt_routers(struct fib6_table *table)
1da177e4
LT		 2456{
2457 struct rt6_info *rt;
2458
2459restart:
c71099ac 2460 read_lock_bh(&table->tb6_lock);
d8d1f30b 2461 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2462 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2463 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2464 dst_hold(&rt->dst);
c71099ac 2465 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2466 ip6_del_rt(rt);
1da177e4
LT		 2467 goto restart;
2468 }
2469 }
c71099ac 2470 read_unlock_bh(&table->tb6_lock);
830218c1
DA		 2471
2472 table->flags &= ~RT6_TABLE_HAS_DFLT_ROUTER;
2473}
2474
2475void rt6_purge_dflt_routers(struct net *net)
2476{
2477 struct fib6_table *table;
2478 struct hlist_head *head;
2479 unsigned int h;
2480
2481 rcu_read_lock();
2482
2483 for (h = 0; h < FIB6_TABLE_HASHSZ; h++) {
2484 head = &net->ipv6.fib_table_hash[h];
2485 hlist_for_each_entry_rcu(table, head, tb6_hlist) {
2486 if (table->flags & RT6_TABLE_HAS_DFLT_ROUTER)
2487 __rt6_purge_dflt_routers(table);
2488 }
2489 }
2490
2491 rcu_read_unlock();
1da177e4
LT		 2492}
2493
5578689a
DL		 2494static void rtmsg_to_fib6_config(struct net *net,
2495 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2496 struct fib6_config *cfg)
2497{
2498 memset(cfg, 0, sizeof(*cfg));
2499
ca254490
DA		 2500 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2501 : RT6_TABLE_MAIN;
86872cb5
TG		 2502 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2503 cfg->fc_metric = rtmsg->rtmsg_metric;
2504 cfg->fc_expires = rtmsg->rtmsg_info;
2505 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2506 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2507 cfg->fc_flags = rtmsg->rtmsg_flags;
2508
5578689a 2509 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2510
4e3fd7a0
AD		 2511 cfg->fc_dst = rtmsg->rtmsg_dst;
2512 cfg->fc_src = rtmsg->rtmsg_src;
2513 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2514}
2515
5578689a 2516int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2517{
86872cb5 2518 struct fib6_config cfg;
1da177e4
LT		 2519 struct in6_rtmsg rtmsg;
2520 int err;
2521
67ba4152 2522 switch (cmd) {
1da177e4
LT		 2523 case SIOCADDRT: /* Add a route */
2524 case SIOCDELRT: /* Delete a route */
af31f412 2525 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2526 return -EPERM;
2527 err = copy_from_user(&rtmsg, arg,
2528 sizeof(struct in6_rtmsg));
2529 if (err)
2530 return -EFAULT;
86872cb5 2531
5578689a 2532 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2533
1da177e4
LT		 2534 rtnl_lock();
2535 switch (cmd) {
2536 case SIOCADDRT:
86872cb5 2537 err = ip6_route_add(&cfg);
1da177e4
LT		 2538 break;
2539 case SIOCDELRT:
86872cb5 2540 err = ip6_route_del(&cfg);
1da177e4
LT		 2541 break;
2542 default:
2543 err = -EINVAL;
2544 }
2545 rtnl_unlock();
2546
2547 return err;
3ff50b79 2548 }
1da177e4
LT		 2549
2550 return -EINVAL;
2551}
2552
2553/*
2554 * Drop the packet on the floor
2555 */
2556
d5fdd6ba 2557static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2558{
612f09e8 2559 int type;
adf30907 2560 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2561 switch (ipstats_mib_noroutes) {
2562 case IPSTATS_MIB_INNOROUTES:
0660e03f 2563 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2564 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2565 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2566 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2567 break;
2568 }
2569 /* FALLTHROUGH */
2570 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2571 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2572 ipstats_mib_noroutes);
612f09e8
YH		 2573 break;
2574 }
3ffe533c 2575 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2576 kfree_skb(skb);
2577 return 0;
2578}
2579
9ce8ade0
TG		 2580static int ip6_pkt_discard(struct sk_buff *skb)
2581{
612f09e8 2582 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2583}
2584
ede2059d 2585static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
1da177e4 2586{
adf30907 2587 skb->dev = skb_dst(skb)->dev;
612f09e8 2588 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2589}
2590
9ce8ade0
TG		 2591static int ip6_pkt_prohibit(struct sk_buff *skb)
2592{
612f09e8 2593 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2594}
2595
ede2059d 2596static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
9ce8ade0 2597{
adf30907 2598 skb->dev = skb_dst(skb)->dev;
612f09e8 2599 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2600}
2601
1da177e4
LT		 2602/*
2603 * Allocate a dst for local (unicast / anycast) address.
2604 */
2605
2606struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2607 const struct in6_addr *addr,
8f031519 2608 bool anycast)
1da177e4 2609{
ca254490 2610 u32 tb_id;
c346dca1 2611 struct net *net = dev_net(idev->dev);
5f02ce24
DA		 2612 struct net_device *dev = net->loopback_dev;
2613 struct rt6_info *rt;
2614
2615 /* use L3 Master device as loopback for host routes if device
2616 * is enslaved and address is not link local or multicast
2617 */
2618 if (!rt6_need_strict(addr))
2619 dev = l3mdev_master_dev_rcu(idev->dev) ? : dev;
2620
2621 rt = ip6_dst_alloc(net, dev, DST_NOCOUNT);
a3300ef4 2622 if (!rt)
1da177e4
LT		 2623 return ERR_PTR(-ENOMEM);
2624
1da177e4
LT		 2625 in6_dev_hold(idev);
2626
11d53b49 2627 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2628 rt->dst.input = ip6_input;
2629 rt->dst.output = ip6_output;
1da177e4 2630 rt->rt6i_idev = idev;
1da177e4
LT		 2631
2632 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2633 if (anycast)
2634 rt->rt6i_flags |= RTF_ANYCAST;
2635 else
1da177e4 2636 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2637
550bab42 2638 rt->rt6i_gateway = *addr;
4e3fd7a0 2639 rt->rt6i_dst.addr = *addr;
1da177e4 2640 rt->rt6i_dst.plen = 128;
ca254490
DA		 2641 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2642 rt->rt6i_table = fib6_get_table(net, tb_id);
8e3d5be7 2643 rt->dst.flags |= DST_NOCACHE;
1da177e4 2644
d8d1f30b 2645 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2646
2647 return rt;
2648}
2649
c3968a85
DW		 2650/* remove deleted ip from prefsrc entries */
2651struct arg_dev_net_ip {
2652 struct net_device *dev;
2653 struct net *net;
2654 struct in6_addr *addr;
2655};
2656
2657static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2658{
2659 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2660 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2661 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2662
d1918542 2663 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2664 rt != net->ipv6.ip6_null_entry &&
2665 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2666 /* remove prefsrc entry */
2667 rt->rt6i_prefsrc.plen = 0;
2668 }
2669 return 0;
2670}
2671
2672void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2673{
2674 struct net *net = dev_net(ifp->idev->dev);
2675 struct arg_dev_net_ip adni = {
2676 .dev = ifp->idev->dev,
2677 .net = net,
2678 .addr = &ifp->addr,
2679 };
0c3584d5 2680 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2681}
2682
be7a010d
DJ		 2683#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2684#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2685
2686/* Remove routers and update dst entries when gateway turn into host. */
2687static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2688{
2689 struct in6_addr *gateway = (struct in6_addr *)arg;
2690
2691 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2692 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2693 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2694 return -1;
2695 }
2696 return 0;
2697}
2698
2699void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2700{
2701 fib6_clean_all(net, fib6_clean_tohost, gateway);
2702}
2703
8ed67789
DL		 2704struct arg_dev_net {
2705 struct net_device *dev;
2706 struct net *net;
2707};
2708
1da177e4
LT		 2709static int fib6_ifdown(struct rt6_info *rt, void *arg)
2710{
bc3ef660 2711 const struct arg_dev_net *adn = arg;
2712 const struct net_device *dev = adn->dev;
8ed67789 2713
d1918542 2714 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2715 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2716 return -1;
c159d30c 2717
1da177e4
LT		 2718 return 0;
2719}
2720
f3db4851 2721void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2722{
8ed67789
DL		 2723 struct arg_dev_net adn = {
2724 .dev = dev,
2725 .net = net,
2726 };
2727
0c3584d5 2728 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2729 icmp6_clean_all(fib6_ifdown, &adn);
e332bc67
EB		 2730 if (dev)
2731 rt6_uncached_list_flush_dev(net, dev);
1da177e4
LT		 2732}
2733
95c96174 2734struct rt6_mtu_change_arg {
1da177e4 2735 struct net_device *dev;
95c96174 2736 unsigned int mtu;
1da177e4
LT		 2737};
2738
2739static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2740{
2741 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2742 struct inet6_dev *idev;
2743
2744 /* In IPv6 pmtu discovery is not optional,
2745 so that RTAX_MTU lock cannot disable it.
2746 We still use this lock to block changes
2747 caused by addrconf/ndisc.
2748 */
2749
2750 idev = __in6_dev_get(arg->dev);
38308473 2751 if (!idev)
1da177e4
LT		 2752 return 0;
2753
2754 /* For administrative MTU increase, there is no way to discover
2755 IPv6 PMTU increase, so PMTU increase should be updated here.
2756 Since RFC 1981 doesn't include administrative MTU increase
2757 update PMTU increase is a MUST. (i.e. jumbo frame)
2758 */
2759 /*
2760 If new MTU is less than route PMTU, this new MTU will be the
2761 lowest MTU in the path, update the route PMTU to reflect PMTU
2762 decreases; if new MTU is greater than route PMTU, and the
2763 old MTU is the lowest MTU in the path, update the route PMTU
2764 to reflect the increase. In this case if the other nodes' MTU
2765 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2766 PMTU discouvery.
2767 */
d1918542 2768 if (rt->dst.dev == arg->dev &&
fb56be83 2769 dst_metric_raw(&rt->dst, RTAX_MTU) &&
4b32b5ad
MKL		 2770 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2771 if (rt->rt6i_flags & RTF_CACHE) {
2772 /* For RTF_CACHE with rt6i_pmtu == 0
2773 * (i.e. a redirected route),
2774 * the metrics of its rt->dst.from has already
2775 * been updated.
2776 */
2777 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2778 rt->rt6i_pmtu = arg->mtu;
2779 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2780 (dst_mtu(&rt->dst) < arg->mtu &&
2781 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2782 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2783 }
566cfd8f 2784 }
1da177e4
LT		 2785 return 0;
2786}
2787
95c96174 2788void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2789{
c71099ac
TG		 2790 struct rt6_mtu_change_arg arg = {
2791 .dev = dev,
2792 .mtu = mtu,
2793 };
1da177e4 2794
0c3584d5 2795 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2796}
2797
ef7c79ed 2798static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2799 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2800 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2801 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2802 [RTA_PRIORITY] = { .type = NLA_U32 },
2803 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2804 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2805 [RTA_PREF] = { .type = NLA_U8 },
19e42e45
RP		 2806 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2807 [RTA_ENCAP] = { .type = NLA_NESTED },
32bc201e 2808 [RTA_EXPIRES] = { .type = NLA_U32 },
622ec2c9 2809 [RTA_UID] = { .type = NLA_U32 },
86872cb5
TG		 2810};
2811
2812static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2813 struct fib6_config *cfg)
1da177e4 2814{
86872cb5
TG		 2815 struct rtmsg *rtm;
2816 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2817 unsigned int pref;
86872cb5 2818 int err;
1da177e4 2819
86872cb5
TG		 2820 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2821 if (err < 0)
2822 goto errout;
1da177e4 2823
86872cb5
TG		 2824 err = -EINVAL;
2825 rtm = nlmsg_data(nlh);
2826 memset(cfg, 0, sizeof(*cfg));
2827
2828 cfg->fc_table = rtm->rtm_table;
2829 cfg->fc_dst_len = rtm->rtm_dst_len;
2830 cfg->fc_src_len = rtm->rtm_src_len;
2831 cfg->fc_flags = RTF_UP;
2832 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2833 cfg->fc_type = rtm->rtm_type;
86872cb5 2834
ef2c7d7b
ND		 2835 if (rtm->rtm_type == RTN_UNREACHABLE ||
2836 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2837 rtm->rtm_type == RTN_PROHIBIT ||
2838 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2839 cfg->fc_flags |= RTF_REJECT;
2840
ab79ad14
MŻ		 2841 if (rtm->rtm_type == RTN_LOCAL)
2842 cfg->fc_flags |= RTF_LOCAL;
2843
1f56a01f
MKL		 2844 if (rtm->rtm_flags & RTM_F_CLONED)
2845 cfg->fc_flags |= RTF_CACHE;
2846
15e47304 2847 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2848 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2849 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2850
2851 if (tb[RTA_GATEWAY]) {
67b61f6c 2852 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2853 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2854 }
86872cb5
TG		 2855
2856 if (tb[RTA_DST]) {
2857 int plen = (rtm->rtm_dst_len + 7) >> 3;
2858
2859 if (nla_len(tb[RTA_DST]) < plen)
2860 goto errout;
2861
2862 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2863 }
86872cb5
TG		 2864
2865 if (tb[RTA_SRC]) {
2866 int plen = (rtm->rtm_src_len + 7) >> 3;
2867
2868 if (nla_len(tb[RTA_SRC]) < plen)
2869 goto errout;
2870
2871 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2872 }
86872cb5 2873
c3968a85 2874 if (tb[RTA_PREFSRC])
67b61f6c 2875 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2876
86872cb5
TG		 2877 if (tb[RTA_OIF])
2878 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2879
2880 if (tb[RTA_PRIORITY])
2881 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2882
2883 if (tb[RTA_METRICS]) {
2884 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2885 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2886 }
86872cb5
TG		 2887
2888 if (tb[RTA_TABLE])
2889 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2890
51ebd318
ND		 2891 if (tb[RTA_MULTIPATH]) {
2892 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2893 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2894 }
2895
c78ba6d6
LR		 2896 if (tb[RTA_PREF]) {
2897 pref = nla_get_u8(tb[RTA_PREF]);
2898 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2899 pref != ICMPV6_ROUTER_PREF_HIGH)
2900 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2901 cfg->fc_flags |= RTF_PREF(pref);
2902 }
2903
19e42e45
RP		 2904 if (tb[RTA_ENCAP])
2905 cfg->fc_encap = tb[RTA_ENCAP];
2906
2907 if (tb[RTA_ENCAP_TYPE])
2908 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2909
32bc201e
XL		 2910 if (tb[RTA_EXPIRES]) {
2911 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2912
2913 if (addrconf_finite_timeout(timeout)) {
2914 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2915 cfg->fc_flags |= RTF_EXPIRES;
2916 }
2917 }
2918
86872cb5
TG		 2919 err = 0;
2920errout:
2921 return err;
1da177e4
LT		 2922}
2923
6b9ea5a6
RP		 2924struct rt6_nh {
2925 struct rt6_info *rt6_info;
2926 struct fib6_config r_cfg;
2927 struct mx6_config mxc;
2928 struct list_head next;
2929};
2930
2931static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2932{
2933 struct rt6_nh *nh;
2934
2935 list_for_each_entry(nh, rt6_nh_list, next) {
2936 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2937 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2938 nh->r_cfg.fc_ifindex);
2939 }
2940}
2941
2942static int ip6_route_info_append(struct list_head *rt6_nh_list,
2943 struct rt6_info *rt, struct fib6_config *r_cfg)
2944{
2945 struct rt6_nh *nh;
2946 struct rt6_info *rtnh;
2947 int err = -EEXIST;
2948
2949 list_for_each_entry(nh, rt6_nh_list, next) {
2950 /* check if rt6_info already exists */
2951 rtnh = nh->rt6_info;
2952
2953 if (rtnh->dst.dev == rt->dst.dev &&
2954 rtnh->rt6i_idev == rt->rt6i_idev &&
2955 ipv6_addr_equal(&rtnh->rt6i_gateway,
2956 &rt->rt6i_gateway))
2957 return err;
2958 }
2959
2960 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2961 if (!nh)
2962 return -ENOMEM;
2963 nh->rt6_info = rt;
2964 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2965 if (err) {
2966 kfree(nh);
2967 return err;
2968 }
2969 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2970 list_add_tail(&nh->next, rt6_nh_list);
2971
2972 return 0;
2973}
2974
2975static int ip6_route_multipath_add(struct fib6_config *cfg)
51ebd318
ND		 2976{
2977 struct fib6_config r_cfg;
2978 struct rtnexthop *rtnh;
6b9ea5a6
RP		 2979 struct rt6_info *rt;
2980 struct rt6_nh *err_nh;
2981 struct rt6_nh *nh, *nh_safe;
51ebd318
ND		 2982 int remaining;
2983 int attrlen;
6b9ea5a6
RP		 2984 int err = 1;
2985 int nhn = 0;
2986 int replace = (cfg->fc_nlinfo.nlh &&
2987 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2988 LIST_HEAD(rt6_nh_list);
51ebd318 2989
35f1b4e9 2990 remaining = cfg->fc_mp_len;
51ebd318 2991 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318 2992
6b9ea5a6
RP		 2993 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2994 * rt6_info structs per nexthop
2995 */
51ebd318
ND		 2996 while (rtnh_ok(rtnh, remaining)) {
2997 memcpy(&r_cfg, cfg, sizeof(*cfg));
2998 if (rtnh->rtnh_ifindex)
2999 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3000
3001 attrlen = rtnh_attrlen(rtnh);
3002 if (attrlen > 0) {
3003 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3004
3005 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3006 if (nla) {
67b61f6c 3007 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 3008 r_cfg.fc_flags |= RTF_GATEWAY;
3009 }
19e42e45
RP		 3010 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
3011 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
3012 if (nla)
3013 r_cfg.fc_encap_type = nla_get_u16(nla);
51ebd318 3014 }
6b9ea5a6 3015
8c5b83f0
RP		 3016 rt = ip6_route_info_create(&r_cfg);
3017 if (IS_ERR(rt)) {
3018 err = PTR_ERR(rt);
3019 rt = NULL;
6b9ea5a6 3020 goto cleanup;
8c5b83f0 3021 }
6b9ea5a6
RP		 3022
3023 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
51ebd318 3024 if (err) {
6b9ea5a6
RP		 3025 dst_free(&rt->dst);
3026 goto cleanup;
3027 }
3028
3029 rtnh = rtnh_next(rtnh, &remaining);
3030 }
3031
3032 err_nh = NULL;
3033 list_for_each_entry(nh, &rt6_nh_list, next) {
3034 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
3035 /* nh->rt6_info is used or freed at this point, reset to NULL*/
3036 nh->rt6_info = NULL;
3037 if (err) {
3038 if (replace && nhn)
3039 ip6_print_replace_route_err(&rt6_nh_list);
3040 err_nh = nh;
3041 goto add_errout;
51ebd318 3042 }
6b9ea5a6 3043
1a72418b 3044 /* Because each route is added like a single route we remove
27596472
MK		 3045 * these flags after the first nexthop: if there is a collision,
3046 * we have already failed to add the first nexthop:
3047 * fib6_add_rt2node() has rejected it; when replacing, old
3048 * nexthops have been replaced by first new, the rest should
3049 * be added to it.
1a72418b 3050 */
27596472
MK		 3051 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
3052 NLM_F_REPLACE);
6b9ea5a6
RP		 3053 nhn++;
3054 }
3055
3056 goto cleanup;
3057
3058add_errout:
3059 /* Delete routes that were already added */
3060 list_for_each_entry(nh, &rt6_nh_list, next) {
3061 if (err_nh == nh)
3062 break;
3063 ip6_route_del(&nh->r_cfg);
3064 }
3065
3066cleanup:
3067 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3068 if (nh->rt6_info)
3069 dst_free(&nh->rt6_info->dst);
52fe51f8 3070 kfree(nh->mxc.mx);
6b9ea5a6
RP		 3071 list_del(&nh->next);
3072 kfree(nh);
3073 }
3074
3075 return err;
3076}
3077
3078static int ip6_route_multipath_del(struct fib6_config *cfg)
3079{
3080 struct fib6_config r_cfg;
3081 struct rtnexthop *rtnh;
3082 int remaining;
3083 int attrlen;
3084 int err = 1, last_err = 0;
3085
3086 remaining = cfg->fc_mp_len;
3087 rtnh = (struct rtnexthop *)cfg->fc_mp;
3088
3089 /* Parse a Multipath Entry */
3090 while (rtnh_ok(rtnh, remaining)) {
3091 memcpy(&r_cfg, cfg, sizeof(*cfg));
3092 if (rtnh->rtnh_ifindex)
3093 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3094
3095 attrlen = rtnh_attrlen(rtnh);
3096 if (attrlen > 0) {
3097 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3098
3099 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3100 if (nla) {
3101 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3102 r_cfg.fc_flags |= RTF_GATEWAY;
3103 }
3104 }
3105 err = ip6_route_del(&r_cfg);
3106 if (err)
3107 last_err = err;
3108
51ebd318
ND		 3109 rtnh = rtnh_next(rtnh, &remaining);
3110 }
3111
3112 return last_err;
3113}
3114
67ba4152 3115static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3116{
86872cb5
TG		 3117 struct fib6_config cfg;
3118 int err;
1da177e4 3119
86872cb5
TG		 3120 err = rtm_to_fib6_config(skb, nlh, &cfg);
3121 if (err < 0)
3122 return err;
3123
51ebd318 3124 if (cfg.fc_mp)
6b9ea5a6 3125 return ip6_route_multipath_del(&cfg);
51ebd318
ND		 3126 else
3127 return ip6_route_del(&cfg);
1da177e4
LT		 3128}
3129
67ba4152 3130static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3131{
86872cb5
TG		 3132 struct fib6_config cfg;
3133 int err;
1da177e4 3134
86872cb5
TG		 3135 err = rtm_to_fib6_config(skb, nlh, &cfg);
3136 if (err < 0)
3137 return err;
3138
51ebd318 3139 if (cfg.fc_mp)
6b9ea5a6 3140 return ip6_route_multipath_add(&cfg);
51ebd318
ND		 3141 else
3142 return ip6_route_add(&cfg);
1da177e4
LT		 3143}
3144
19e42e45 3145static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
339bf98f
TG		 3146{
3147 return NLMSG_ALIGN(sizeof(struct rtmsg))
3148 + nla_total_size(16) /* RTA_SRC */
3149 + nla_total_size(16) /* RTA_DST */
3150 + nla_total_size(16) /* RTA_GATEWAY */
3151 + nla_total_size(16) /* RTA_PREFSRC */
3152 + nla_total_size(4) /* RTA_TABLE */
3153 + nla_total_size(4) /* RTA_IIF */
3154 + nla_total_size(4) /* RTA_OIF */
3155 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 3156 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 3157 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6 3158 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
19e42e45 3159 + nla_total_size(1) /* RTA_PREF */
61adedf3 3160 + lwtunnel_get_encap_size(rt->dst.lwtstate);
339bf98f
TG		 3161}
3162
191cd582
BH		 3163static int rt6_fill_node(struct net *net,
3164 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 3165 struct in6_addr *dst, struct in6_addr *src,
15e47304 3166 int iif, int type, u32 portid, u32 seq,
7bc570c8 3167 int prefix, int nowait, unsigned int flags)
1da177e4 3168{
4b32b5ad 3169 u32 metrics[RTAX_MAX];
1da177e4 3170 struct rtmsg *rtm;
2d7202bf 3171 struct nlmsghdr *nlh;
e3703b3d 3172 long expires;
9e762a4a 3173 u32 table;
1da177e4
LT		 3174
3175 if (prefix) { /* user wants prefix routes only */
3176 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3177 /* success since this is not a prefix route */
3178 return 1;
3179 }
3180 }
3181
15e47304 3182 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 3183 if (!nlh)
26932566 3184 return -EMSGSIZE;
2d7202bf
TG		 3185
3186 rtm = nlmsg_data(nlh);
1da177e4
LT		 3187 rtm->rtm_family = AF_INET6;
3188 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3189 rtm->rtm_src_len = rt->rt6i_src.plen;
3190 rtm->rtm_tos = 0;
c71099ac 3191 if (rt->rt6i_table)
9e762a4a 3192 table = rt->rt6i_table->tb6_id;
c71099ac 3193 else
9e762a4a
PM		 3194 table = RT6_TABLE_UNSPEC;
3195 rtm->rtm_table = table;
c78679e8
DM		 3196 if (nla_put_u32(skb, RTA_TABLE, table))
3197 goto nla_put_failure;
ef2c7d7b
ND		 3198 if (rt->rt6i_flags & RTF_REJECT) {
3199 switch (rt->dst.error) {
3200 case -EINVAL:
3201 rtm->rtm_type = RTN_BLACKHOLE;
3202 break;
3203 case -EACCES:
3204 rtm->rtm_type = RTN_PROHIBIT;
3205 break;
b4949ab2
ND		 3206 case -EAGAIN:
3207 rtm->rtm_type = RTN_THROW;
3208 break;
ef2c7d7b
ND		 3209 default:
3210 rtm->rtm_type = RTN_UNREACHABLE;
3211 break;
3212 }
3213 }
38308473 3214 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 3215 rtm->rtm_type = RTN_LOCAL;
d1918542 3216 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 3217 rtm->rtm_type = RTN_LOCAL;
3218 else
3219 rtm->rtm_type = RTN_UNICAST;
3220 rtm->rtm_flags = 0;
35103d11 3221 if (!netif_carrier_ok(rt->dst.dev)) {
cea45e20 3222 rtm->rtm_flags |= RTNH_F_LINKDOWN;
35103d11
AG		 3223 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3224 rtm->rtm_flags |= RTNH_F_DEAD;
3225 }
1da177e4
LT		 3226 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3227 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 3228 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 3229 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 3230 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3231 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3232 rtm->rtm_protocol = RTPROT_RA;
3233 else
3234 rtm->rtm_protocol = RTPROT_KERNEL;
3235 }
1da177e4 3236
38308473 3237 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 3238 rtm->rtm_flags |= RTM_F_CLONED;
3239
3240 if (dst) {
930345ea 3241 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 3242 goto nla_put_failure;
1ab1457c 3243 rtm->rtm_dst_len = 128;
1da177e4 3244 } else if (rtm->rtm_dst_len)
930345ea 3245 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 3246 goto nla_put_failure;
1da177e4
LT		 3247#ifdef CONFIG_IPV6_SUBTREES
3248 if (src) {
930345ea 3249 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 3250 goto nla_put_failure;
1ab1457c 3251 rtm->rtm_src_len = 128;
c78679e8 3252 } else if (rtm->rtm_src_len &&
930345ea 3253 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 3254 goto nla_put_failure;
1da177e4 3255#endif
7bc570c8
YH		 3256 if (iif) {
3257#ifdef CONFIG_IPV6_MROUTE
3258 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
2cf75070
NA		 3259 int err = ip6mr_get_route(net, skb, rtm, nowait,
3260 portid);
3261
7bc570c8
YH		 3262 if (err <= 0) {
3263 if (!nowait) {
3264 if (err == 0)
3265 return 0;
3266 goto nla_put_failure;
3267 } else {
3268 if (err == -EMSGSIZE)
3269 goto nla_put_failure;
3270 }
3271 }
3272 } else
3273#endif
c78679e8
DM		 3274 if (nla_put_u32(skb, RTA_IIF, iif))
3275 goto nla_put_failure;
7bc570c8 3276 } else if (dst) {
1da177e4 3277 struct in6_addr saddr_buf;
c78679e8 3278 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 3279 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3280 goto nla_put_failure;
1da177e4 3281 }
2d7202bf 3282
c3968a85
DW		 3283 if (rt->rt6i_prefsrc.plen) {
3284 struct in6_addr saddr_buf;
4e3fd7a0 3285 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 3286 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3287 goto nla_put_failure;
c3968a85
DW		 3288 }
3289
4b32b5ad
MKL		 3290 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3291 if (rt->rt6i_pmtu)
3292 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3293 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 3294 goto nla_put_failure;
3295
dd0cbf29 3296 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 3297 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 3298 goto nla_put_failure;
94f826b8 3299 }
2d7202bf 3300
c78679e8
DM		 3301 if (rt->dst.dev &&
3302 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3303 goto nla_put_failure;
3304 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3305 goto nla_put_failure;
8253947e
LW		 3306
3307 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 3308
87a50699 3309 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 3310 goto nla_put_failure;
2d7202bf 3311
c78ba6d6
LR		 3312 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3313 goto nla_put_failure;
3314
61adedf3 3315 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
19e42e45 3316
053c095a
JB		 3317 nlmsg_end(skb, nlh);
3318 return 0;
2d7202bf
TG		 3319
3320nla_put_failure:
26932566
PM		 3321 nlmsg_cancel(skb, nlh);
3322 return -EMSGSIZE;
1da177e4
LT		 3323}
3324
1b43af54 3325int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 3326{
3327 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3328 int prefix;
3329
2d7202bf
TG		 3330 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3331 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 3332 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3333 } else
3334 prefix = 0;
3335
191cd582
BH		 3336 return rt6_fill_node(arg->net,
3337 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 3338 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 3339 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 3340}
3341
67ba4152 3342static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 3343{
3b1e0a65 3344 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 3345 struct nlattr *tb[RTA_MAX+1];
3346 struct rt6_info *rt;
1da177e4 3347 struct sk_buff *skb;
ab364a6f 3348 struct rtmsg *rtm;
4c9483b2 3349 struct flowi6 fl6;
72331bc0 3350 int err, iif = 0, oif = 0;
1da177e4 3351
ab364a6f
TG		 3352 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3353 if (err < 0)
3354 goto errout;
1da177e4 3355
ab364a6f 3356 err = -EINVAL;
4c9483b2 3357 memset(&fl6, 0, sizeof(fl6));
38b7097b
HFS		 3358 rtm = nlmsg_data(nlh);
3359 fl6.flowlabel = ip6_make_flowinfo(rtm->rtm_tos, 0);
1da177e4 3360
ab364a6f
TG		 3361 if (tb[RTA_SRC]) {
3362 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3363 goto errout;
3364
4e3fd7a0 3365 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 3366 }
3367
3368 if (tb[RTA_DST]) {
3369 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3370 goto errout;
3371
4e3fd7a0 3372 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 3373 }
3374
3375 if (tb[RTA_IIF])
3376 iif = nla_get_u32(tb[RTA_IIF]);
3377
3378 if (tb[RTA_OIF])
72331bc0 3379 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 3380
2e47b291
LC		 3381 if (tb[RTA_MARK])
3382 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3383
622ec2c9
LC		 3384 if (tb[RTA_UID])
3385 fl6.flowi6_uid = make_kuid(current_user_ns(),
3386 nla_get_u32(tb[RTA_UID]));
3387 else
3388 fl6.flowi6_uid = iif ? INVALID_UID : current_uid();
3389
1da177e4
LT		 3390 if (iif) {
3391 struct net_device *dev;
72331bc0
SL		 3392 int flags = 0;
3393
5578689a 3394 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 3395 if (!dev) {
3396 err = -ENODEV;
ab364a6f 3397 goto errout;
1da177e4 3398 }
72331bc0
SL		 3399
3400 fl6.flowi6_iif = iif;
3401
3402 if (!ipv6_addr_any(&fl6.saddr))
3403 flags |= RT6_LOOKUP_F_HAS_SADDR;
3404
3405 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3406 flags);
3407 } else {
3408 fl6.flowi6_oif = oif;
3409
3410 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 3411 }
3412
ab364a6f 3413 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 3414 if (!skb) {
94e187c0 3415 ip6_rt_put(rt);
ab364a6f
TG		 3416 err = -ENOBUFS;
3417 goto errout;
3418 }
1da177e4 3419
ab364a6f
TG		 3420 /* Reserve room for dummy headers, this skb can pass
3421 through good chunk of routing engine.
3422 */
459a98ed 3423 skb_reset_mac_header(skb);
ab364a6f 3424 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 3425
d8d1f30b 3426 skb_dst_set(skb, &rt->dst);
1da177e4 3427
4c9483b2 3428 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 3429 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 3430 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 3431 if (err < 0) {
ab364a6f
TG		 3432 kfree_skb(skb);
3433 goto errout;
1da177e4
LT		 3434 }
3435
15e47304 3436 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 3437errout:
1da177e4 3438 return err;
1da177e4
LT		 3439}
3440
37a1d361
RP		 3441void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3442 unsigned int nlm_flags)
1da177e4
LT		 3443{
3444 struct sk_buff *skb;
5578689a 3445 struct net *net = info->nl_net;
528c4ceb
DL		 3446 u32 seq;
3447 int err;
3448
3449 err = -ENOBUFS;
38308473 3450 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 3451
19e42e45 3452 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
38308473 3453 if (!skb)
21713ebc
TG		 3454 goto errout;
3455
191cd582 3456 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
37a1d361 3457 event, info->portid, seq, 0, 0, nlm_flags);
26932566
PM		 3458 if (err < 0) {
3459 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3460 WARN_ON(err == -EMSGSIZE);
3461 kfree_skb(skb);
3462 goto errout;
3463 }
15e47304 3464 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 3465 info->nlh, gfp_any());
3466 return;
21713ebc
TG		 3467errout:
3468 if (err < 0)
5578689a 3469 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 3470}
3471
8ed67789 3472static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 3473 unsigned long event, void *ptr)
8ed67789 3474{
351638e7 3475 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 3476 struct net *net = dev_net(dev);
8ed67789
DL		 3477
3478 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 3479 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 3480 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3481#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3482 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 3483 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 3484 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 3485 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3486#endif
3487 }
3488
3489 return NOTIFY_OK;
3490}
3491
1da177e4
LT		 3492/*
3493 * /proc
3494 */
3495
3496#ifdef CONFIG_PROC_FS
3497
33120b30
AD		 3498static const struct file_operations ipv6_route_proc_fops = {
3499 .owner = THIS_MODULE,
3500 .open = ipv6_route_open,
3501 .read = seq_read,
3502 .llseek = seq_lseek,
8d2ca1d7 3503 .release = seq_release_net,
33120b30
AD		 3504};
3505
1da177e4
LT		 3506static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3507{
69ddb805 3508 struct net *net = (struct net *)seq->private;
1da177e4 3509 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 3510 net->ipv6.rt6_stats->fib_nodes,
3511 net->ipv6.rt6_stats->fib_route_nodes,
3512 net->ipv6.rt6_stats->fib_rt_alloc,
3513 net->ipv6.rt6_stats->fib_rt_entries,
3514 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 3515 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 3516 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 3517
3518 return 0;
3519}
3520
3521static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3522{
de05c557 3523 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 3524}
3525
9a32144e 3526static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 3527 .owner = THIS_MODULE,
3528 .open = rt6_stats_seq_open,
3529 .read = seq_read,
3530 .llseek = seq_lseek,
b6fcbdb4 3531 .release = single_release_net,
1da177e4
LT		 3532};
3533#endif /* CONFIG_PROC_FS */
3534
3535#ifdef CONFIG_SYSCTL
3536
1da177e4 3537static
fe2c6338 3538int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3539 void __user *buffer, size_t *lenp, loff_t *ppos)
3540{
c486da34
LAG		 3541 struct net *net;
3542 int delay;
3543 if (!write)
1da177e4 3544 return -EINVAL;
c486da34
LAG		 3545
3546 net = (struct net *)ctl->extra1;
3547 delay = net->ipv6.sysctl.flush_delay;
3548 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3549 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3550 return 0;
1da177e4
LT		 3551}
3552
fe2c6338 3553struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3554 {
1da177e4 3555 .procname = "flush",
4990509f 3556 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3557 .maxlen = sizeof(int),
89c8b3a1 3558 .mode = 0200,
6d9f239a 3559 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3560 },
3561 {
1da177e4 3562 .procname = "gc_thresh",
9a7ec3a9 3563 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3564 .maxlen = sizeof(int),
3565 .mode = 0644,
6d9f239a 3566 .proc_handler = proc_dointvec,
1da177e4
LT		 3567 },
3568 {
1da177e4 3569 .procname = "max_size",
4990509f 3570 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3571 .maxlen = sizeof(int),
3572 .mode = 0644,
6d9f239a 3573 .proc_handler = proc_dointvec,
1da177e4
LT		 3574 },
3575 {
1da177e4 3576 .procname = "gc_min_interval",
4990509f 3577 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3578 .maxlen = sizeof(int),
3579 .mode = 0644,
6d9f239a 3580 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3581 },
3582 {
1da177e4 3583 .procname = "gc_timeout",
4990509f 3584 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3585 .maxlen = sizeof(int),
3586 .mode = 0644,
6d9f239a 3587 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3588 },
3589 {
1da177e4 3590 .procname = "gc_interval",
4990509f 3591 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3592 .maxlen = sizeof(int),
3593 .mode = 0644,
6d9f239a 3594 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3595 },
3596 {
1da177e4 3597 .procname = "gc_elasticity",
4990509f 3598 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3599 .maxlen = sizeof(int),
3600 .mode = 0644,
f3d3f616 3601 .proc_handler = proc_dointvec,
1da177e4
LT		 3602 },
3603 {
1da177e4 3604 .procname = "mtu_expires",
4990509f 3605 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3606 .maxlen = sizeof(int),
3607 .mode = 0644,
6d9f239a 3608 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3609 },
3610 {
1da177e4 3611 .procname = "min_adv_mss",
4990509f 3612 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3613 .maxlen = sizeof(int),
3614 .mode = 0644,
f3d3f616 3615 .proc_handler = proc_dointvec,
1da177e4
LT		 3616 },
3617 {
1da177e4 3618 .procname = "gc_min_interval_ms",
4990509f 3619 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3620 .maxlen = sizeof(int),
3621 .mode = 0644,
6d9f239a 3622 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3623 },
f8572d8f 3624 { }
1da177e4
LT		 3625};
3626
2c8c1e72 3627struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3628{
3629 struct ctl_table *table;
3630
3631 table = kmemdup(ipv6_route_table_template,
3632 sizeof(ipv6_route_table_template),
3633 GFP_KERNEL);
5ee09105
YH		 3634
3635 if (table) {
3636 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3637 table[0].extra1 = net;
86393e52 3638 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3639 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3640 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3641 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3642 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3643 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3644 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3645 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3646 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3647
3648 /* Don't export sysctls to unprivileged users */
3649 if (net->user_ns != &init_user_ns)
3650 table[0].procname = NULL;
5ee09105
YH		 3651 }
3652
760f2d01
DL		 3653 return table;
3654}
1da177e4
LT		 3655#endif
3656
2c8c1e72 3657static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3658{
633d424b 3659 int ret = -ENOMEM;
8ed67789 3660
86393e52
AD		 3661 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3662 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3663
fc66f95c
ED		 3664 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3665 goto out_ip6_dst_ops;
3666
8ed67789
DL		 3667 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3668 sizeof(*net->ipv6.ip6_null_entry),
3669 GFP_KERNEL);
3670 if (!net->ipv6.ip6_null_entry)
fc66f95c 3671 goto out_ip6_dst_entries;
d8d1f30b 3672 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3673 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3674 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3675 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3676 ip6_template_metrics, true);
8ed67789
DL		 3677
3678#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3679 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3680 sizeof(*net->ipv6.ip6_prohibit_entry),
3681 GFP_KERNEL);
68fffc67
PZ		 3682 if (!net->ipv6.ip6_prohibit_entry)
3683 goto out_ip6_null_entry;
d8d1f30b 3684 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3685 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3686 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3687 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3688 ip6_template_metrics, true);
8ed67789
DL		 3689
3690 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3691 sizeof(*net->ipv6.ip6_blk_hole_entry),
3692 GFP_KERNEL);
68fffc67
PZ		 3693 if (!net->ipv6.ip6_blk_hole_entry)
3694 goto out_ip6_prohibit_entry;
d8d1f30b 3695 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3696 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3697 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3698 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3699 ip6_template_metrics, true);
8ed67789
DL		 3700#endif
3701
b339a47c
PZ		 3702 net->ipv6.sysctl.flush_delay = 0;
3703 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3704 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3705 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3706 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3707 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3708 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3709 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3710
6891a346
BT		 3711 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3712
8ed67789
DL		 3713 ret = 0;
3714out:
3715 return ret;
f2fc6a54 3716
68fffc67
PZ		 3717#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3718out_ip6_prohibit_entry:
3719 kfree(net->ipv6.ip6_prohibit_entry);
3720out_ip6_null_entry:
3721 kfree(net->ipv6.ip6_null_entry);
3722#endif
fc66f95c
ED		 3723out_ip6_dst_entries:
3724 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3725out_ip6_dst_ops:
f2fc6a54 3726 goto out;
cdb18761
DL		 3727}
3728
2c8c1e72 3729static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3730{
8ed67789
DL		 3731 kfree(net->ipv6.ip6_null_entry);
3732#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3733 kfree(net->ipv6.ip6_prohibit_entry);
3734 kfree(net->ipv6.ip6_blk_hole_entry);
3735#endif
41bb78b4 3736 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3737}
3738
d189634e
TG		 3739static int __net_init ip6_route_net_init_late(struct net *net)
3740{
3741#ifdef CONFIG_PROC_FS
d4beaa66
G		 3742 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3743 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3744#endif
3745 return 0;
3746}
3747
3748static void __net_exit ip6_route_net_exit_late(struct net *net)
3749{
3750#ifdef CONFIG_PROC_FS
ece31ffd
G		 3751 remove_proc_entry("ipv6_route", net->proc_net);
3752 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3753#endif
3754}
3755
cdb18761
DL		 3756static struct pernet_operations ip6_route_net_ops = {
3757 .init = ip6_route_net_init,
3758 .exit = ip6_route_net_exit,
3759};
3760
c3426b47
DM		 3761static int __net_init ipv6_inetpeer_init(struct net *net)
3762{
3763 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3764
3765 if (!bp)
3766 return -ENOMEM;
3767 inet_peer_base_init(bp);
3768 net->ipv6.peers = bp;
3769 return 0;
3770}
3771
3772static void __net_exit ipv6_inetpeer_exit(struct net *net)
3773{
3774 struct inet_peer_base *bp = net->ipv6.peers;
3775
3776 net->ipv6.peers = NULL;
56a6b248 3777 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3778 kfree(bp);
3779}
3780
2b823f72 3781static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3782 .init = ipv6_inetpeer_init,
3783 .exit = ipv6_inetpeer_exit,
3784};
3785
d189634e
TG		 3786static struct pernet_operations ip6_route_net_late_ops = {
3787 .init = ip6_route_net_init_late,
3788 .exit = ip6_route_net_exit_late,
3789};
3790
8ed67789
DL		 3791static struct notifier_block ip6_route_dev_notifier = {
3792 .notifier_call = ip6_route_dev_notify,
3793 .priority = 0,
3794};
3795
433d49c3 3796int __init ip6_route_init(void)
1da177e4 3797{
433d49c3 3798 int ret;
8d0b94af 3799 int cpu;
433d49c3 3800
9a7ec3a9
DL		 3801 ret = -ENOMEM;
3802 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3803 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3804 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3805 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3806 goto out;
14e50e57 3807
fc66f95c 3808 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3809 if (ret)
bdb3289f 3810 goto out_kmem_cache;
bdb3289f 3811
c3426b47
DM		 3812 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3813 if (ret)
e8803b6c 3814 goto out_dst_entries;
2a0c451a 3815
7e52b33b
DM		 3816 ret = register_pernet_subsys(&ip6_route_net_ops);
3817 if (ret)
3818 goto out_register_inetpeer;
c3426b47 3819
5dc121e9
AE		 3820 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3821
8ed67789
DL		 3822 /* Registering of the loopback is done before this portion of code,
3823 * the loopback reference in rt6_info will not be taken, do it
3824 * manually for init_net */
d8d1f30b 3825 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3826 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3827 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3828 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3829 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3830 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3831 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3832 #endif
e8803b6c 3833 ret = fib6_init();
433d49c3 3834 if (ret)
8ed67789 3835 goto out_register_subsys;
433d49c3 3836
433d49c3
DL		 3837 ret = xfrm6_init();
3838 if (ret)
e8803b6c 3839 goto out_fib6_init;
c35b7e72 3840
433d49c3
DL		 3841 ret = fib6_rules_init();
3842 if (ret)
3843 goto xfrm6_init;
7e5449c2 3844
d189634e
TG		 3845 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3846 if (ret)
3847 goto fib6_rules_init;
3848
433d49c3 3849 ret = -ENOBUFS;
c7ac8679
GR		 3850 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3851 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3852 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3853 goto out_register_late_subsys;
c127ea2c 3854
8ed67789 3855 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3856 if (ret)
d189634e 3857 goto out_register_late_subsys;
8ed67789 3858
8d0b94af
MKL		 3859 for_each_possible_cpu(cpu) {
3860 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3861
3862 INIT_LIST_HEAD(&ul->head);
3863 spin_lock_init(&ul->lock);
3864 }
3865
433d49c3
DL		 3866out:
3867 return ret;
3868
d189634e
TG		 3869out_register_late_subsys:
3870 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3871fib6_rules_init:
433d49c3
DL		 3872 fib6_rules_cleanup();
3873xfrm6_init:
433d49c3 3874 xfrm6_fini();
2a0c451a
TG		 3875out_fib6_init:
3876 fib6_gc_cleanup();
8ed67789
DL		 3877out_register_subsys:
3878 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3879out_register_inetpeer:
3880 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3881out_dst_entries:
3882 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3883out_kmem_cache:
f2fc6a54 3884 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3885 goto out;
1da177e4
LT		 3886}
3887
3888void ip6_route_cleanup(void)
3889{
8ed67789 3890 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3891 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3892 fib6_rules_cleanup();
1da177e4 3893 xfrm6_fini();
1da177e4 3894 fib6_gc_cleanup();
c3426b47 3895 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3896 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3897 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3898 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3899}
Ubuntu Artful kernel mirror